Interfacial Engineering at Quantum Dot-Sensitized TiO2 Photoelectrodes for Ultrahigh Photocurrent Generation

Metal oxide semiconductor/chalcogenide quantum dot (QD) heterostructured photoanodes show photocurrent densities >30 mA/cm2 with ZnO, approaching the theoretical limits in photovoltaic (PV) cells. However, comparative performance has not been achieved with TiO2. Here, we applied a TiO2(B) surface passivation layer (SPL) on TiO2/QD (PbS and CdS) and achieved a photocurrent density of 34.59 mA/cm2 under AM 1.5G illumination for PV cells, the highest recorded to date. The SPL improves electron conductivity by increasing the density of surface states, facilitating multiple trapping/detrapping transport, and increasing the coordination number of TiO2 nanoparticles. This, along with impeded electron recombination, led to enhanced collection efficiency, which is a major factor for performance. Furthermore, SPL-treated TiO2/QD photoanodes were successfully exploited in photoelectrochemical water splitting cells, showing an excellent photocurrent density of 14.43 mA/cm2 at 0.82 V versus the Reversible Hydrogen Electrode (RHE). These results suggest a new promising strategy for the development of high-performance photoelectrochemical devices.


INTRODUCTION
Chalcogenide quantum dots (QDs) have attracted much attention as building blocks for next-generation light-harvesting devices due to their outstanding optical characteristics such as a wide light absorption range over the near-IR regions and high extinction coefficient. 1−5 Typical examples of such lightharvesting devices include photovoltaic (PV) cells and photoelectrochemical (PEC) water splitting cells, consisting of light-harvesting materials deposited on a mesoporous n-type semiconductor layer of TiO 2 , ZnO, or SnO 2 ; an electrolyte; and a counter electrode. 6−8 Over the last 5 years, high photocurrent densities greater than 30 mA/cm 2 have been reported for PV cells using metal oxide/chalcogenide QD heterostructured photoanodes having high light-harvesting ability ( Figure 1). For instance, ZnO/ chalcogenide QDs have led to ultrahigh photocurrent densities in PV cells, with a record value of 39 mA/cm 2 , closely approaching the theoretical photocurrent density (44 mA/ cm 2 ) for the 1.1 eV band gap of such QDs. 9 However, TiO 2 / chalcogenide QD heterostructures have been struggling to reach photocurrent densities greater than 30 mA/cm 2 , mostly due to the lower electron conductivity of TiO 2 compared to ZnO. 10 In this work, we achieved a photocurrent density close to 35 mA/cm 2 , the highest recorded to date (Figure 1).
Chalcogenide QDs have been actively exploited for the generation of solar hydrogen in PEC water splitting cells due to their outstanding water durability compared to other light-harvesting materials such as metalorganic dyes and halide perovskites. 18,19 In this regard, TiO 2 /chalcogenide QD photoanodes are much more attractive for PEC hydrogen generation compared to ZnO/chalcogenide QDs due to the excellent stability of TiO 2 in strong-base electrolyte conditions. 20 Therefore, the development of TiO 2 /chalcogenide QD photoanodes showing a high photocurrent density greater than 30 mA/cm 2 in PV cells is extremely attractive for PEC applications.
To design the optimal architecture of photoanodes with TiO 2 /chalcogenide QDs for high photocurrents, enhancement of both charge transfer kinetics at the TiO 2 /chalcogenide QD interface and charge transport via the TiO 2 films should be simultaneously considered. 21 As a general strategy, the introduction of a surface passivation layer (SPL) on TiO 2 film has led to significant improvements in sensitized PV cells. 22,23 Particularly, TiCl 4 treatment to form a TiO 2 SPL at the interface of TiO 2 /light-harvesting materials has led to an about 10−30% photocurrent density increase. 24 In dyesensitized solar cells (DSSCs) with a SPL, performance enhancement has been attributed to the improved electron injection, mostly because of driving force enhancement through the conduction band (CB) shift of TiO 2 and suppression of electron recombination at the TiO 2 /dye interface. 24 However, changes in electron transport induced by SPLs have been reported to be minimal due to the high diffusion length of electrons through the mesoporous TiO 2 layer (20−25 μm with I 3 − /I − redox couples and 10−12.5 μm for water oxidation 25,26 ) compared to the film thickness (∼13 μm). 24,27,28 Conversely, multiple electron recombination pathways have been reported for TiO 2 /chalcogenide QD photoanodes due to the complex distribution of energy states as well as the ultrafast electron injection rate to TiO 2 (<10 −9 s). 3,18,29 Therefore, it is expected that the effects of the SPL on the functional performance of TiO 2 /QD photoanodes will be significantly different from those of conventional TiO 2 /dye photoanodes, although explicit effects on transport properties have not been completely elucidated.
Electron transport in mesoporous TiO 2 films is commonly interpreted with the trap state-mediated transport model with multiple trapping/detrapping events. 30 In semiconductors, shallow traps in a band gap significantly affect electron transport and are typically located 0.5−1.0 eV below the CB. 31 Therefore, the deposition of the SPL impacts the population and distribution of trap states of TiO 2 with a concomitant effect on the TiO 2 /chalcogenide QD interface. During SPL formation on the TiO 2 film, the surface states, which are localized and generated by chemical surface treatment, can affect electron transport. 32 The surface states of TiO 2 are formed mainly at the trap state area 0.3−0.4 eV below the CB and originate mainly from undercoordinated Ti 4+ atoms on the anatase TiO 2 surface. 33 It is well-known that charge transfer at the interface of the mesoporous semiconductor/solution is affected by surface states. 34−36 However, even though the increase in the density of surface trap states in relation to deposition of the TiO 2 SPL has been extensively reported, its effects on electron transport through mesoporous TiO 2 films are uncertain. 24,28,37 Additionally, changes in film morphology during the formation of the SPL have a significant impact on the electron transport of TiO 2 films. 30 Therefore, the effects of the SPL on electron transport, as well as on energetics and the morphology of the photoelectrodes, should be precisely tuned to improve the photocurrent density.
Herein, we achieved a photocurrent density of 34.59 mA/ cm 2 in PV cells with a 0.18 cm 2 active area and 14.43 mA/cm 2 in PEC water splitting cells with an active area of 1.33 cm 2 at 1 sun condition with about 20% enhancement (compared to the reference samples) by hydrothermal treatment of TiCl 4 to form an SPL on the TiO 2 /PbS-CdS QD photoanodes. Based on the results of photocurrent densities on PV cells and PEC water splitting cells, the SPL facilitates electron conductivity of mesoporous TiO 2 films by increasing the density of surface states and the coordination number of TiO 2 nanoparticles (NPs). The increased electron conductivity is mostly due to an increase in surface states for multiple trapping/detrapping transport through TiO 2 films, which was quantified in terms of the chemical diffusion coefficient. The suppression of electron recombination was also observed for the SPL. Therefore, we conclude that the main effects of the SPL on the photocurrent density of TiO 2 /PbS-CdS QD photoanodes are related to improved charge collection efficiency (η cc ), driven by enhancement of electron transport and suppression of electron recombination.

Preparation of Mesoporous TiO 2 /PbS-CdS QD Photoanodes and Counter
Electrodes. An FTO glass substrate was cleaned in a two-step sonication process with an aqueous detergent solution for 1 h (2% Hellmanex III in DI water, v/v) and subsequently with an organic solvent mixture (EtOH and ACT, 1:1, v/v). The cleaned FTO substrates were used for all working-and counter electrodes. For the purification of TiCl 4 , a 0.05 mol of TiCl 4 solution was slowly added dropwise to an HCl solution in a 1:5 ratio at −20°C. 38 A TiO 2 blocking layer (∼20 nm) on the FTO substrate of the working electrode was treated using the hydrothermal method with a 40 mM TiCl 4 aqueous solution at 70°C for 30 min. After rinsing with DI water and EtOH, the working electrode was sintered at 450°C for 30 min. This process was repeated to form double blocking layers. Next, two layers of TiO 2 NPs and one layer of TiO 2 light-scattering particles (∼400 nm, CCIC) were doctor-bladed onto the blocking layer-coated FTO glass and sintered at 450°C for 30 min. The thickness of the photoanode was around 12 μm. For the PT20 and PT30 samples, TiCl 4 treatment was conducted on neat TiO 2 films with 20 and 30 nm NPs (termed T20 and T30, hereafter) using the hydrothermal method with a 40 mM TiCl 4 aqueous solution at 70°C for 18 min and then sintered at 450°C for 30 min before rinsing with DI water and EtOH.
For the photoanode used in both PV cells and PEC water splitting cells, the co-sensitization of PbS and CdS QDs was applied in this work using the successive ionic layer adsorption and reaction (SILAR) method for the direct growth of QDs on the surface of TiO 2 films. The detailed synthetic method of QDs was followed according to our previous study with slight modifications. 39   deposition. For the ZnS passivation layer, TiO 2 /QD heterostructured electrodes were additionally immersed in a Zn 2+ (0.1 M Zn-(CH 3 COO) 2 ·2H 2 O in MeOH) and a S 2− solution (0.1 M Na 2 S· 9H 2 O in MeOH/water (1:1, v/v)) twice repeatedly. Between each step, a cleaning process should be conducted for the production of high-quality QD films by rinsing the films with MeOH and DI water to remove impurities during the successive deposition. The immersion period for each step was 1 min per solution.
The fabrication method used to create the Cu x S counter electrode (CE) for PV cells was adopted from previous studies. 40,41 Copper sulfides were deposited on FTO glass by the chemical bath deposition (CBD) method. A mixture of 0.1 M CuSO 4 and Na 2 S 2 O 3 aqueous solutions was used for the deposition of Cu x S films. Then, FTO was dipped in the solution at 70°C for 3 h at pH 3, rinsed with DI water and EtOH, and dried using N 2 gas. The as-prepared film was annealed at 130°C for 30 min under ambient conditions. For PEC water splitting cells, a 2.5 cm × 2.5 cm commercial Pt plate was used.
2.3. Fabrication of PV Cells and PEC Water Splitting Cells. For PV cells, we employed the sandwich fabrication method using the Cu x S CE with two holes predrilled for electrolyte injection and the QD-photoanode. A 25 μm thick Surlyn film was placed between the two electrodes as a spacer, and the electrolyte composed of 2.0 M S, Na 2 S, and KCl in aqueous solution was injected into this space. The holes were then covered up using a Surlyn film and cover glass for finishing.
In PEC water splitting cells, a Pt plate (2.5 cm × 2.5 cm) and a Ag/ AgCl (3 M KCl) electrode were used as the counter and the reference electrodes, respectively. The same photoanodes were used as PV cells. In the aqueous electrolyte, 0.25 M Na 2 S and 0.35 M Na 2 SO 3 (pH 13) were utilized as hole scavengers. Sufficient N 2 bubbling was conducted for over 30 min before every measurement. The characteristics of PEC water splitting cells for hydrogen generation were measured using a custom-built cell.
2.4. Characterization. 2.4.1. Morphology and Energy Level of TiO 2 . A transmission electron microscope (TEM) JEM 2100 F (JEOL, Japan) was used with the ZrO/W(100) electron gun. The specific surface areas and pore size distributions of TiO 2 NPs were measured using Brunauer−Emmett−Teller (BET) and Barret−Joyner−Halenda (BJH) analyses with 3Flex (Micromeritics) using a N 2 carrier gas. The crystal phase of TiO 2 films coated on FTO substrates was measured by X-ray diffraction (XRD) using SmartLab (Rigaku) with Cu Kα radiation. Scanning electron microscope (SEM) images were measured using an FE-SEM (NOVA NANO SEM 450, FEI) at a 10.0 kV acceleration voltage and a field-free lens mode. To trace the energy level of TiO 2 films, UV−vis absorption spectroscopy (V-670 UV/vis spectrophotometer, Jasco) was used for TiO 2 without the scattering layer/FTO samples, and ultraviolet photoelectron spectroscopy (UPS, Thermo Fisher Scientific Co.) data were collected from a He I excitation source (21.2 eV). The UPS data were calibrated using the Fermi edge of Au. All samples for UPS were coated on a Si substrate.

Measurement of Time-Resolved Photoluminescence
(TRPL). Time-resolved photoluminescence (TRPL) imaging was measured using an inverted-type scanning confocal microscope (MicroTime-200, Picoquant) with a 40× (air) objective at the Korea Basic Science Institute (KBSI) in Daegu Center (Republic of Korea). For an excitation source, a single-mode pulsed diode laser (470 nm with ∼30 ps pulse width and ∼10 μW laser power) was employed. The emission of each sample was collected using a singlephoton avalanche diode (SPAD; PDM series, MPD). Using the timetagged time-resolved (TTTR) data acquisition route, TRPL images (200 × 200 pixels) were recorded with 5 ms of the acquisition time for each pixel. Fittings of the PL decays were conducted by an exponential decay model using the following equation: is the time-dependent PL intensity, A is the amplitude, and τ is the PL lifetime. 42 2.4.3. Measurement of the Current−Voltage (J−V) Curve and Incident Photon-to-Current Conversion Efficiencies (IPCEs). A Keithley (Model 2400) source meter and a solar simulator with a 300 W Xenon arc-lamp (Newport) were used for the measurement of the J−V characteristics of photoelectrochemical devices. The IPCEs of PV cells were measured using a QEX7 system (PV Measurements, Inc.). To avoid overestimating the power conversion efficiency (PCE) and the photocurrent density, masks (0.18 and 1.33 cm 2 active areas) were placed on the front side of the photoanodes.

Electrochemical Analyses.
To estimate the density of surface states in the TiO 2 films with and without SPL, electrochemical methods of cyclic voltammetry (CV) and impedance spectroscopy (IS) were used with a three-electrode system and an aqueous electrolyte consisting of 0.25 M Na 2 S and 0.35 M Na 2 SO 3 , which was the same composition used for PEC water splitting cells under dark conditions. Because of the basic pH of the electrolyte in the PEC water splitting cells (pH 13), a reversible hydrogen electrode (RHE) should be used as the reference electrode via correction of the Ag/ AgCl reference electrode potential using the equation V RHE = V Ag/AgCl + 0.197 + pH × 0.059. 43 The results of chemical capacitance (C μ ) and recombination resistance (R rec ) with the TiO 2 /PbS-CdS QD photoanodes were extracted from IS measurements through the simplified equivalent circuit in the dark for PEC water splitting cells. 34 Different frequency ranges were applied in PV cells (1 MHz−10 mHz) and PEC water splitting cells (400 kHz−10 mHz) based on previous reports. 34,39 The film conductivity (σ) was directly calculated from the charge transport resistance (R t ) of the films, which was also analyzed by the IS measurements with the simplified equivalent circuit under dark conditions using the equation σ = L/R t A(1 − p o ), where L is the film thickness, A is the film area, and p o is the film porosity. 44 The same quasi-Fermi levels in the TiO 2 films for both R t and R rec were used to obtain a correct η cc at 1 sun illumination. 45 The equivalent conduction band potential (V ecb ) was used for the IS parameters (R t and R rec ) and calculation of η cc . Here, V ecb = V F − ΔE c /q, where V F is the corrected Fermi voltage, which is the actual potential applied to the photoanode without the voltage drops from the counter electrode and the series resistance from the applied potential, and ΔE c is the shift in the CB with respect to the reference 46 For PV cells, IS has been frequently used to investigate the interfacial characteristics of charge accumulation, transport, and recombination using a diffusion−recombination equivalent circuit model. 47 3. RESULTS AND DISCUSSION 3.1. Morphological and Structural Characteristics. Deposition of a surface passivation layer (SPL) by thermal hydrolysis of TiCl 4 led to clear morphological changes in mesoporous TiO 2 films, as shown in Figure S1a. Hereafter, mesoporous TiO 2 layers created with NPs having 20 and 30 nm particle diameters are referred to as T20 and T30, and the corresponding ones with SPLs are denoted as PT20 and PT30, respectively. An SPL whose thickness ranged from 2 to 5 nm was identified at the surface of the mesoporous TiO 2 films by TEM, high-resolution TEM (HR-TEM), and X-ray photoelectron spectroscopy (XPS) (Figures 2a and S1). In Figure 2a, the HR-TEM image of PT30 clearly indicates the effective growth of the SPL on the surface of TiO 2 NPs. The highly textured atomic structure represents the (101) plane of the anatase phase of the TiO 2 NP, and the corresponding dspacing is 0.35 nm. On the top of the anatase surface, a distinct crystalline layer is formed with a comparable narrow d-spacing of 0.21−0.22 nm, attributed to the (003) plane of TiO 2 (B). Interestingly, XRD patterns (Figures 2b and S2) of PT20 and PT30 showed a dual-phase (anatase and TiO 2 (B)) nanocrystalline structure with polygonal plate-and needle-shaped crystallites, respectively, whereas only anatase crystals were obtained for T20 and T30 without the SPL ( Figure S2). Here, a peak at 2θ = 48.0°corresponds to the anatase in the (200) direction, and another peak at 2θ = 43.5° (Figure 2b) is identified as the (003) plane of the polymorph TiO 2 (B) structure. 48,49 Figure 2a also clearly shows the interplanar distance of the TiO 2 (B) structure (0.21−0.22 nm) at the (003) plane, as confirmed by JCPDS no. 46-1237 in the SPL of PT30. 49 TEM and XRD data suggested that the SPL created a distinct crystal structure of the TiO 2 (B) phase on the TiO 2 films, which resulted in morphological and structural changes and a corresponding change in electron transport properties. However, further analyses were not carried out to reveal the characteristics of TiO 2 (B), such as a detailed growth mechanism and physical properties, since it was out of the scope of the current work.
Concerning the effects of the morphological changes on the SPL, BET and BJH data were found to correlate well with the SEM top-view images, indicating that there is a decrease in the pore size, pore volume, and surface area of the SPL (see Table  1 and Figures S3 and S4). For instance, the porosities (p o ) of 58 and 66% for T20 and T30 were reduced to 43 and 65% for PT20 and PT30, respectively.
The average coordination number of TiO 2 NPs (N), representing the average number of interconnections between TiO 2 per particle, is readily calculated using the equation N = 3.08/p o − 1.13, with p o being the porosity, 50 as listed in Table  1. N increased as the NP size decreased. Interestingly, the formation of the SPL notably resulted in an increase in N, particularly for PT20, suggesting a better architecture for electron transport.
PbS and CdS QDs were deposited on mesoporous TiO 2 films to prepare heterostructured photoanodes for TiO 2 /PbS-CdS QDs by a SILAR method (see XRD and HR-TEM images of Figures S5 and S6). The deposition method is detailed in the Experimental Section. PbS and CdS QDs featuring particles 2−3 and 5−7 nm in diameter, respectively ( Figure  2c and TEM and energy-dispersive X-ray spectrometry (EDX) images of Figure S7), were evenly distributed on the surface of TiO 2 films ( Figure S8). The surface area of TiO 2 mesoporous films showed little difference after SPL deposition (Table 1), which would have a negligible influence on the surface coverage of QDs on TiO 2 films. Atomic ratios of PbS and CdS QDs on the TiO 2 films measured by XPS support the surface area results (Figure S9), suggesting that the surface fraction of deposited QDs on TiO 2 films with the SPL would be slightly lower than that of neat samples without the SPL.
3.2. Photoelectrochemical Performances of PV Cells and PEC Water Splitting Cells. The mesoporous TiO 2 /PbS-CdS QD heterostructured photoanodes with/without the SPL were applied for both PV cells and PEC water splitting cells to evaluate their performances. A ZnS passivation layer was additionally deposited on the photoanodes to prevent photocorrosion and reduce recombination. Figure 3a depicts the device architectures of PV and PEC cells with mesoporous TiO 2 /PbS-CdS QD heterostructured photoanodes.
The J−V curves of PV cells with a 0.18 cm 2 active area of the photoanodes using the two-electrode sandwich devices at 1 sun conditions are shown in Figure 3b, and their corresponding photovoltaic parameters are summarized in Table 2. The IPCE and the integrated photocurrent density from the IPCE spectra support the J−V results of PV cells (Figure 3c). The shortcircuit current density (J sc ) significantly increased in the presence of the SPL (about 20%) compared to those of the references (T20 and T30), showing a J sc higher than 25 mA/ cm 2 ( Table 2). In particular, PT20 showed a J sc of 34.59 mA/ cm 2 at 1 sun condition ( Figure S11), which is the highest value reported in PV cells with TiO 2 /QD photoanodes to date (Figure 1), whereas PT30 achieved the best PCE of PV cells (6.85%). The pore sizes of mesoporous TiO 2 films can account for the different PCEs observed in the samples. As shown in Table 1, the pore size of PT20 decreased to 9.27 nm after SPL deposition from the 15.41 nm of T20, increasing the likelihood of pore blockage when the QDs are deposited on mesoporous TiO 2 films. However, PT30 probably could maintain an adequate pore size to allow for electrolyte penetration even after the formation of the SPL. The small pore sizes of PT20 are also related to the large series resistance (R s ) in the film, directly resulting in a decreased FF and a consequent decrease in the PCE.
where V p is the cumulative specific pore volume and 1/ρ is the reciprocal of the density of TiO 2 (0.257 cm 3 / g). 30 The outstanding photocurrents obtained for PV cells motivated us to use the photoanode in PEC water splitting cells for solar H 2 production. In comparison with PV cells, the performance of PEC water splitting photoanodes with a larger geometrical area (1.33 cm 2 ) showed similar trends (e.g., an increase in J sc upon introduction of a SPL). Additionally, the photocurrent densities of PEC water splitting cells (J ph ) were higher than 10 mA/cm 2 at 0.82 V RHE for all samples, tested in a three-electrode arrangement (Figure 3d). Impressively, the highest J ph of 14.4 mA/cm 2 at 0.82 V RHE was achieved with PT20. Note that the potential of 0.82 V RHE is based on all samples at a steady state. This is remarkable compared to previously reported values (see Table S1 for comparison). The steady-state, long-term stability of the photoanodes was tested (Figure 3e) at the same applied voltage (0.82 V RHE), showing relatively stable operation above 9 mA/cm 2 for more than 15 h under a strong alkaline condition (pH 13). H 2 generation was monitored by gas chromatography equipped with gas-enclosed PEC water splitting cells, and the corresponding current density profile was recorded at 1.4 V RHE under the same irradiation condition (1 sun) (a and b Figure S12b). The trend of H 2 generation was in excellent agreement with that of the photocurrent density in both PV cells and PEC water splitting cells. The faradic efficiency was also enhanced when the TiO 2 films were treated with a SPL ( Figure S12c).

Effects of the SPL on Charge Injection Efficiency.
To gain further mechanistic insights into the operation of the photoelectrodes used, light-harvesting efficiency (η lh ), charge separation efficiency (η sp ), and charge collection efficiency (η cc ) were determined. 24 η sp must be divided into charge injection and regeneration efficiencies (η inj and η reg , respectively), and η reg is assumed to be unity for all cases since the regeneration rate of QDs by excess S 2− in the electrolyte is  The numbers indicate the average photovoltaic performances (J sc and PEC) and standard deviations for five or six different cells (see Figure S10 for J−V curves). nearly 3−4 orders of magnitude faster than the electron transfer rate from TiO 2 to oxidized QDs. 21,51 Consequently, the effects of the SPL on η inj were examined following the Marcus theory. 52 We considered the energy levels of TiO 2 films determining the reaction free energy for the injection rate (ΔG 0 ) from the CB of QDs (E c_QD ) to the CB of TiO 2 (E c_TiO2 ) films: ΔG 0 = E c_QD − E c_TiO 2 . Based on the results of XRD and XPS for QDs ( Figures S5 and S9), we assume that E c_QD is the same in all samples. In other words, the relative ΔG 0 value is solely determined by E c_ TiO 2 . In this regard, we determined E c_TiO 2 first, as shown in Figure 4a, using the Tauc plots from UV−vis absorption spectra and UPS valence band (VB) spectra ( Figure S13). These results suggest that the SPL slightly decreases the optical band gap (E g ) of TiO 2 films along with the CB energy level. PT20 (E g = 3.40 eV) showed little difference from T20 (3.41 eV) in terms of band gap, while PT30 showed a larger change in the CB. The results also showed that ΔG 0 increased in the following order: T30 < PT30 < T20 ∼ PT20. Figure 4b shows the emission decay spectra of TRPL for QDs on TiO 2 and QDs on Al 2 O 3 as a reference, from which all electron injection parameters can be obtained. 53 Table 3 lists the electron injection parameters of PL lifetime (τ PL ), electron injection rate constant (k inj ), and electron injection efficiency (η inj ). The results confirm that both k inj and η inj are consistent with ΔG 0 in all TiO 2 /QD heterostructured photoanodes, suggesting their strong correlation. Even though a change in η inj was observed, it was very small. This is different from the large increase in η inj typically observed when an SPL was introduced by TiCl 4 treatment in TiO 2 /dye photoanodes. 24 Therefore, the reason for the above results is more likely due to electron injection from PbS QDs to the TiO 2 film in the ultrafast range of femtoseconds. 54 3.4. Increases in the Surface State Concentration and the Charge Collection Efficiency. It is well-known that surface states modified by the presence of an SPL play a key role in the energetic and kinetic properties for charge transport and transfer reactions, which are directly related to η cc . 34 To estimate the density of surface states in the TiO 2 films, chemical capacitance (C μ ) was measured by cyclic voltammetry (CV) and electrochemical impedance spectroscopy (IS) in a PEC cell ( Figure S14). Figure 5a shows the density of trap states (DOS) estimated from the C μ of TiO 2 films as a function of applied potential. The DOS peaks at 0−0.2 V RHE showed Gaussian behavior (marked as the yellow area in Figure 5a), which has been interpreted as a reversible filling of surface states located in traps below the CB of TiO 2 . 34,35,55 The DOS increased after the deposition of the SPL (PT20 and PT30), suggesting that this layer increased the density of surface states on the TiO 2 films.
The impact of the SPL on the energy distribution of TiO 2 films directly correlates to the change of the chemical diffusion coefficient (D n ). When a diffusion coefficient depends on E F , it is referred to as the "chemical" diffusion coefficient. 31,53,56 Mesoporous nanostructured semiconductors featured charge transport within a broad distribution of localized states. Consequently, the change in electrochemical potential (Fermi energy level, E F ) of charges modified the occupation of localized states, impacting their diffusion. Therefore, D n can be affected by both the energy distributions of nanostructured semiconductors and morphology. Figure 5b shows D n as a function of applied potential for TiO 2 films. D n of all samples increased with negative potential, indicating that the transport mechanism follows the multiple trapping/detrapping transport model. 56 Interestingly, D n exhibited the same trends as DOS and increased as T30 < T20 < PT30 < PT20 in the potential range of 0−0.1 V RHE. These results provide strong evidence that the surface states in the TiO 2 film introduced by the SPL have a direct correlation with D n . Since the energy level of the surface states of the TiO 2 films is located at the effective range of conduction (about 0.5−1.0 eV below the CB), 56 we conclude that D n increases due to the increase in the density of surface states.
On the other hand, morphological changes such as porosity (p o ), coordination number (N), and particle size induced by  the SPL can also affect the intrinsic diffusion properties of TiO 2 films. As expected, D n increases with an increase in N and a decrease in p o during SPL formation. 57 Regarding the particle size effect, we expected that the small TiO 2 NPs would have a low D n primarily due to the increased grain boundary area. 30 However, Figure 5b shows that the small TiO 2 NPs have a high D n instead. Therefore, we conclude that both p o and N have a stronger influence on D n compared to the particle size.
In any case, surface states and particle sizes may be related. 58 Based on the multiple trapping/detrapping transport model for charge transport in mesoporous nanostructured semiconductors and the generalized Einstein relation, conductivity (σ) can be expressed by the following equation Here, C μ trap is the trap state capacitance, and D n (E F ) is the electron diffusion coefficient. 53 If the density of trap states can produce localized electrons, the DOS in the band gap is associated with the trap state capacitance. Therefore, the energetic distribution of the density of trap states (DOS = g(E F )) can be calculated by the equation C μ trap (E) = qg(E F ), where q is the elemental charge. 35,47 Clearly, the SPL increased the conductivity of TiO 2 films through an increase in the density of surface states and morphological changes in the TiO 2 film (Figure 5c).
Electron recombination is another important factor affected by the SPL via the density of surface states, significantly influencing charge collection. Figure 5d shows the evolution of recombination resistance (R rec ) of heterostructured TiO 2 /PbS-CdS photoanodes with the applied potential. This resistance increased with the anodic (positive) potential. The valley at 0− 0.2 eV is related to recombination through surface states. 34 Interestingly, although the density of surface states was increased by the SPL, it efficiently blocked electron recombination. Furthermore, R rec and σ followed similar trends and were increased by the SPL. The decreased dark currents in Figure 3d treated with the SPL supported well the recombination blocking effect of the SPL.
Both conductivity and recombination through the TiO 2 films ultimately influence the charge collection efficiency, η cc , which in turn directly controlled the photocurrent densities (J sc and J ph ) of the photoelectrochemical devices. η cc can be estimated by the equation where R t is the transport resistance. 45,59 Figure S15 shows the results of R t and R rec as a function of the equivalent conduction band potential, V ecb , in PV cells extracted from the IS measurements. Here, V ecb can be The IS measurements were used to extract the parameters with the simplified equivalent circuit for transport, chemical capacitance, and charge transfer using a three-electrode system. T20 (blue empty circles), PT20 (blue filled circles), T30 (red empty squares), and PT30 nm (red filled squares). Figure 6. Charge collection efficiency (η cc ) of PV cells with TiO 2 / PbS-CdS QD heterostructured photoanodes with T20 (blue empty circles), PT20 (blue filled circles), T30 (red empty squares), and PT30 nm (red filled squares) at 1 sun condition. Values were calculated using the results of R t and R rec as a function of V ecb in PV cells ( Figure S15). obtained from V ecb = V F − ΔE c /q to compare the recombination resistance (R rec ), where V F is the corrected Fermi voltage excluding the influences of series resistance and the CE in the device and ΔE c is the shift in the CB with respect to the reference E c,ref ; ΔE c = E c − E c,ref . 46 The calculated η cc shown in Figure 6 increases with nearly the same trend as the conductivity and recombination resistance of the TiO 2 films and the photocurrent densities of TiO 2 /PbS-CdS QD heterostructured photoanodes in both PV cells and PEC water splitting cells. Comparing the parameters for photocurrent densities such as the η lh calculated via IPCE in Table  S2, η cc , and η inj values, we demonstrated that η cc controls the photocurrent densities of PV cells with the TiO 2 films. Consequently, high photocurrent densities greater than 30 mA/cm 2 for PV cells with SPL samples (PT20 and PT30) were obtained primarily due to the enhancement of film conductivity, as well as the decrease in recombination kinetics induced by the SPL. Figure 7 summarizes the effects of the SPL on the photocurrent density of the TiO 2 /QD heterostructured photoanode.
• After light absorption of QDs, neither charge injection nor regeneration of oxidized QDs was significantly affected by the SPL. This may be partly related to the ultrafast electron injection rate of QDs caused by the direct deposition method. 54 (1) and (2), further experiments are required. Since both η inj and η lh are almost unity at the TiO 2 /QD heterostructured photoanode, maximizing the photocurrent density by improving η cc can be an effective strategy. A higher photocurrent density can be obtained by incorporating an SPL despite the slight loss of η lh due to the decreased surface area (Table S2). As a result, the introduction of a SPL is a viable strategy to maximize η cc and to optimize the photocurrent density by enhancing conductivity and blocking recombination at the TiO 2 /QD heterostructured photoanode.

CONCLUSIONS
We investigated the effect of a TiO 2 SPL deposited on a mesoporous TiO 2 film surface for TiO 2 /PbS-CdS QD heterostructured photoanodes for PV and water splitting PEC devices. Such a TiO 2 SPL increased the density of surface states and also the coordination number of the TiO 2 NPs, both leading to an increase in electron conductivity through the TiO 2 film. The increase in electron conductivity in the TiO 2 film is mostly due to the improvement of the chemical diffusion coefficient according to the multiple trapping/ detrapping transport model. Concomitantly, the reduction of back electron transfers, known to be the conventional role of SPLs, also helps to increase η cc in the TiO 2 films. Therefore, the TiO 2 /PbS-CdS QD photoanodes showed a J sc of 34.59 mA/cm 2 in PV cells and a photocurrent density of 14.43 mA/ cm 2 at 0.82 V RHE in PEC water splitting cells. On the other hand, η inj , which mainly affects the photocurrent density in photoanodes, was not significantly different for devices with/ without SPLs. The results could provide new directions and important milestones for the development of high-performance PEC devices with TiO 2 /QD heterostructured photoanodes.