High-yield synthesis of “oriented attachment” TiO₂ nanorods as superior building blocks of photoanodes in quantum dot sensitized solar cells

Abstract

In this paper, a high-yield hydrothermal synthesis of “oriented attachment” TiO₂ nanorods (TiO₂-NRs) and their application as a superior photoanode material in a quantum-dot (QD) sensitized solar cell have been reported. The NRs' morphology could be finely tuned by the concentrations of titanium isopropoxide, and tetramethylammonium hydroxide, and their relative molar ratios. The production yield of TiO₂-NRs with the fine size of 18 nm × 150 nm could be as high as 20 grams per one reaction from a 500 mL autoclave. These NRs, as building blocks of photoanodes in CdSe QD sensitized solar cells, are found to be superior to traditionally used TiO₂ nanoparticles, which is reflected in the following aspects: (1) the resultant TiO₂ mesoporous film possesses a wider pore size distribution, which allows a higher loading content of QDs and fluent electrolyte diffusion; (2) the as-built quasi-one dimensional network with limited grain boundaries possesses a higher electron diffusion coefficient and longer electron lifetime. As a consequence of these, synchronized improvements on photocurrent, photovoltage and fill factor have been achieved, leading to a dramatic elevation of the overall solar conversion efficiency. Finally, in combination with a highly efficient catalytic counter electrode made of Cu₂ZnSnS(Se)₄ nanocrystals, the best cell efficiency reached 5.96%, which is comparable to the best-in-class devices reported to date.

---

1. Introduction

The growing need for low-cost and renewable energy has aroused intensive research in the types of next generation solar cells.¹ Among them, quantum-dot sensitized solar cells (QDSSCs) have attracted increasing attention in recent years owing to the following features: (1) the narrow bandgap semiconductor quantum-dots (QDs) possess high extinction coefficients and inherent dipole moments, leading to fast charge separation properties;² (2) the band gaps of semiconductor QDs can be controlled by particle size, shape and composition, allowing their light absorption with easy tunability and excellent matching with the solar spectrum;^3–7 (3) the possible existence of hot electron injection, multiple exciton generation and other special physical effects that make it possible to achieve more than 100% quantum efficiency.^8,9 Up to now, the best QDSSCs were reported with the solar conversion efficiencies of 9%, lagging much behind the optimized dye-sensitized solar cells (DSSCs) which share the similar device structure and working mechanism.^10–15 At the current stage, to explore new materials as QDs,^16–20 photoanodes,^21–27 catalytic counter electrodes (CEs)^28–32 as well as electrolytes^33–36 in the QDSSC system and systematically study the inherent reasons hold the key for the further performance improvement.

Nanostructured photoanodes, normally made of TiO₂ nanoparticles (TiO₂-NPs), undertake the following functions in QDSSC system: (1) provide a huge inner surface area for QDs adsorption; (2) accept and collect the photo-injected electrons from QDs by successful kinetic competitions with the interfacial recombinations; (3) allow electrolyte penetration through the pore channels to regenerate the QDs. These several functions sometimes are correlated with each other; optimizations between them should be well-balanced. For example, TiO₂ or ZnO one-dimensional nanoarrays were thought to be perfect for charge collection because of elimination of grain boundaries along their perpendicular charge transport paths. However, most of them were suffering from controlling pore size and surface area.^16–20 Therefore, implementations of those nanoarrays have not resulted in a performance breakthrough. Besides, the nanoarrays were generally prepared by hydrothermal methods via seeds induced growth; their scaling-up fabrication and technical reproducibility have essential limitations.

In this work, a high-yield hydrothermal synthesis of “oriented attachment” TiO₂-nanorods (TiO₂-NRs) with tunable sizes has been reported. The TiO₂-NRs with the average size of ∼18 nm × 150 nm has been selected as the alternative photoanode material to traditionally used TiO₂-NPs in QDSSC system. The notable features for these NRs-based photoanode include: (1) the grain boundaries density inside the 3-D electron transport network are largely decreased; (2) the pore size distribution can be largely widened without sacrificing the mesoporous film's structural stability because of the rod-like morphology; (3) the high-yield and highly dispersed TiO₂-NRs are compatible for the fast screen-printing technique, allowing film fabrication be uniform and reproducible. How the structural features effecting on the solar cell performance has been elucidated in this work, when a TiO₂-NPs photoanode was used as the reference for comparison. The inherent optical and electrical reasons responsible for the performance improvements have been investigated. Although, some similar oriented attachment TiO₂ NRs have been examined in dye-sensitized solar cells, showing good electron transport properties,^37,38 the porous structure of the mesoporous films to load molecular dyes and inorganic QDs should be optimized differently.

2. Experimental

Preparation of TiO₂ nanocrystals

TiO₂-NPs with the average size of 18 nm was prepared according to the literature.³⁹ TiO₂-NRs were prepared as follows: first, a certain amount of tetramethylammonium hydroxide (TMAH) was added to a 500 mL round bottom flask with 150 mL deionized water inside. Then, a mixture of 0.125–0.625 mol L⁻¹ titanium(IV) isopropoxide (TTIP) and 150 mL ethylene glycol was slowly added to the above solution with vigorous stirring at 80 °C within 30 minutes. The molar ratio of TMAH : TTIP was controlled ranging between 1 : 3 and 3 : 2, in order to tune the product morphology. After stirring for 4–8 hours until a transparent sol was obtained, the sol precursor was transferred into a 500 mL teflon-lined autoclave and treated at 230 °C for 24 h. At last, the white TiO₂ product was collected by centrifuge and washed thoroughly by excessive ethanol for several times.

Solar cells fabrication

The TiO₂-NPs and TiO₂-NRs pastes and the corresponding films were prepared according to the literature.³⁹ The film thicknesses were controlled the same at 10 μm. SILAR deposition of CdSe QDs on the TiO₂ photoanodes was according to the literature³⁶ After SILAR deposition of CdSe, a shell layer of ZnS was deposited by twice immersion the films alternately into 0.1 M Zn(CH₃COO)₂ and 0.1 M Na₂S aqueous solution for 1 min per dip.⁴⁰ The standard Pt-courter electrode was prepared by thermal decomposition of H₂PtCl₆ on FTO glass at 400 °C. The highly efficient Cu₂ZnSnS(Se)₄ CE was prepared according to our previous work.³² The QDs-sensitized photoanodes and the CEs were assembly into sandwich type cells and sealed with 25 μm thick Surlyn spacers. Polysulfide electrolyte⁴¹ consisting of 2 M Na₂S, 2 M S, and 0.2 M KCl in a mixture of methanol and water (3 : 7 by volume) was injected by vacuum backfilling. The active area of the cell was 0.16 cm².

Characterizations

The morphologies of TiO₂-NPs and TiO₂-NRs films were observed using a scanning electron microscope (SEM, FEI-Sirion 200, FEI). The crystal structures were identified by X-ray powder diffraction (XRD, X'Pert PRO, Cu Kα radiation, Panalytical B. V.). The Brunauer–Emmett–Teller (BET) of the surface areas and the pore size distributions were analysis using an ASAP 2020 accelerated surface area and porosimetry system. The film thickness was measured using a profilometer (Dektak 150, Veeco Instruments Inc.). The UV-Vis-NIR spectra of the films were recorded on a Perkin-Elmer UV/Vis spectrophotometer (model Lambda 950). The solar cells were tested using a 450 W xenon light source solar simulator (Oriel, model 9119) with AM 1.5G filter (Oriel, model 91192). Light intensity was calibrated by using a standard silicon reference cell. The current–voltage characteristics were obtained by applying external potential bias to the cell and measuring the generated photocurrent with a Keithley model 2400 digital source meter. A similar data acquisition system was used to control photon-to-current conversion efficiency (IPCE) measurement. A white light bias (1% sunlight intensity) was applied onto the sample during the IPCE measurements with AC model (10 Hz). The transient photovoltage/photocurrent decays were measured on a home-made photoelectrical system. A white light bias on the cell sample was generated from an array of diodes. The cell sample was excited by a red light pulse (0.05 s square pulse width, 100 ns rise and fall time) from an LED controlled by a fast solid-state switch. The voltage dynamics were recorded on a PC-interfaced Keithley 2602A source meter with a 100 μs response time. By fitting the voltage or current decay signal, one could get the electron lifetime (τ_n), the electron diffusion time (τ_t) and the effective diffusion lengths (L_n).

3. Results and discussion

Fig. 1a shows the morphologies of TiO₂ products dependent upon the concentrations of TTIP and the molar ratios between TMAH : TTIP. It is certain that there is a critical TTIP concentration below and above which will lead to totally different crystal growth mechanisms. As the inset SEM images in Fig. 1a depict, at a low TTIP concentration of 0.125 mol L⁻¹, the product presents uniform morphology of spindle-like particles with relatively larger diameter sizes (55–260 nm), the crystal growth of which is thought to be governed by conventional “Ostwald ripening”;^42–44 while at relatively higher TTIP concentrations (e.g., 0.375 mol L⁻¹ and 0.625 mol L⁻¹), the crystal morphology turns to be smaller NRs which was governed by non-classic “oriented attachment” mechanism.^43–47 It is thought that above the critical concentration, the supersaturation level becomes very high which promotes the fast nucleation process and leads to high concentration of crystal nucleuses in the precursor. These nucleuses have no chance to grow very big because of their mutual competition for the Ti source. The surface energy of those nano-sized nucleuses are very high; the system tends to minimize the free energy through spontaneous elimination of high energy crystal planes; “oriented attachment” just provides a way to realize the goal. According to the literature,^45–49 such “oriented attachment” describes the process occurring preferably between the highest energy {001} planes of adjacent anatase octahedral primary particles. Therefore, as the high resolution TEM image depicts (Fig. 1b), the resultant anatase NRs are with 〈001〉 oriented growth direction and zigzag side exposed surfaces, all of which are composed of low surface energy {101} planes. In the contrast, below the critical concentration, the driving force for the nucleation becomes smaller resulting in lower concentration of nucleuses in the precursor; some nucleuses have the chance to grow even larger; the primary bigger particles will further grow by exhausting other smaller unstable nucleuses to decrease the system's free energy; these processes, called “Ostwald ripening”,^42–44 result in the spindle-like shape of the as-synthesized particles. Such a shape is close to the equilibrium geometric shape of anatase determined by the Wulff rule.^50,51 Furthermore, from Fig. 1a, it is also known that the relative ratio between TMAH : TTIP has no direct impact on the crystal growth mechanism but can tune the diameter sizes largely. The higher concentration of TMAH, the larger diameter of the resultant materials; this could be explained by the faster crystal growth at the higher TMAH concentration, because the crystal growth of TiO₂ is resulted from TTIP hydrolysis and condensation which is proportional to the free hydroxyls provided by TMAH.

Fig. 1 (a) SEM images depicting the morphologies of different TiO₂ products, affected by the concentration of TTIP and the relative ratios between TMAH : TTIP, (b) high resolution TEM image depicting TiO₂-NRs made of partially fused anatase bipyramid primary crystal seeds, the anisotropic growth along 〈001〉 direction and side exposed surfaces consisting of {101} crystal planes, (c) the relationship between TTIP concentration and product yields (left y axis) as well as diameters of TiO₂-spindles or NRs (right y axis), the ratios of TMAH : TTIP were all set at 1 : 2, (d) a photograph highlighting scale-up production of the TiO₂-NRs paste made from the NRs shown in the top-right SEM inset (denoted by the dash line arrow).

From Fig. 1c, the critical TTIP concentration corresponding to crystal growth mechanism change is about 0.1875 mol L⁻¹, which is reflected by the inflection point of the products' diameter size. Besides, it is important that at a very high TTIP concentration of 0.625 mol L⁻¹, the NRs' morphology including suitable lengths (∼150 nm) and diameters (∼18 nm) could be maintained if the relative ratio of TMAH : TTIP is high enough. The production yield for one pot (500 mL volume) reaction could be as high as ∼20 grams at TTIP concentration of 0.625 mol L⁻¹, which is higher than the widely used acidic hydrothermal method (about 15 g per reaction).³⁹ Furthermore, because ethylene glycol was used as the solvent, it can help to protect the NRs from aggregation. The highly dispersive NRs could be processed easily to be high quality paste compatible for screen printing. Fig. 1d is a photo showing the large quantity of TiO₂ paste (>100 gram) made from one pot synthesized “oriented attachment” NRs.

Fig. 2a and b are SEM images of the photoanode films made of TiO₂-NPs with the size of ∼18 nm and NRs with the size of ∼18 nm × 150 nm. The TiO₂-NRs film is observed with more hollow structure than the TiO₂-NPs film due to their different packing density of nano-building blocks. Fig. 2c shows the XRD patterns of the TiO₂-NPs and TiO₂-NRs films after sintering. All of the diffraction peaks could be indexed to anatase TiO₂ with tetragonal crystal structure (JCPDS #21-1272) and no impurity phase could be detected. The crystal size of TiO₂-NPs is calculated to be 17.8 nm in average by applying the Scherrer equation to the broadened diffraction peaks from Fig. 2c, which is in consistent with SEM observation. The average diameter of TiO₂-NRs is calculated to be 18.3 nm by applying the Scherrer equation to (101) peak of TiO₂-NRs film, the crystal direction of which is nearly perpendicular to long axis of NRs. Another noteworthy feature of the XRD patterns is that, the intensity ratio between the diffraction peaks of (200)/(101) for the TiO₂-NRs film is much higher than that for the TiO₂-NPs film, which is derived from and further confirms the anisotropic growth of TiO₂-NRs along the 〈001〉 direction. The BET surface areas of the TiO₂-NPs and TiO₂-NRs films are tested to be 107.0 m² g⁻¹ and 69.8 m² g⁻¹, respectively. Such distinct surface area difference is mostly associated with the pore size broadening from NPs to NRs as reflected in Fig. 2d. The pore size distribution for the TiO₂-NPs film is centered at 16.2 nm while that for the TiO₂-NRs film is centered at 27.1 nm.

Fig. 2 SEM images of (a) TiO₂-NPs film, (b) TiO₂-NRs film, (c) XRD patterns, and (d) pore size distributions of the corresponding films.

Fig. 3a and b are J–V curves of two series QDSSCs based on the two kind photoanodes sensitized with CdSe QDs deposited by different SILAR cycles. The best QDSSC based on TiO₂-NRs gave 3.1% efficiency, with J_sc of 14.3 mA cm⁻², V_oc 514 mV of and FF of 0.442. While the best QDSSC based on TiO₂-NPs generated 2.1% efficiency, with J_sc of 13.5 mA cm⁻², V_oc 487 mV of and FF of 0.327. From Fig. 3a and b, it can be seen that, J_sc of the TiO₂-NPs series QDSSCs keeps on increasing until 8 SILAR cycles and then begins to decrease, while that for the TiO₂-NRs series QDSSCs continues increasing even after 15 SILAR cycles. The variation of J_sc of these two series QDSSCs can be well explained by their corresponding IPCEs shown in Fig. 3c and d. From the IPCEs shown in Fig. 3c, the photo-response thresholds of the TiO₂-NPs series QDSSCs red shift from ∼600 nm to ∼800 nm as the SILAR cycles increase from 1 to 15. However, for this series QDSSCs, the IPCE peak value increases until 8 cycles and then begin to decrease. From Fig. 3d, the IPCE spectra of the TiO₂-NRs series QDSSCs also red shift from about 600 nm to 800 nm as the increasing SILAR cycles, and their IPCE peak value keeps on increasing up to 15 cycles. It can be found two notable differences on IPCEs between these two series QDSSCs: (1) the TiO₂-NPs series QDSSCs get saturated at less SILAR cycles than the TiO₂-NRs series QDSSCs; (2) the IPCEs spectra of the TiO₂-NPs series QDSSCs have relatively lower shoulders near the longer wavelength region than that of the TiO₂-NRs series QDSSCs. The former difference should be mostly ascribed to the pore size limited electrolyte diffusion, which will be discussed with detailed evidences in the following. The latter difference is possibly ascribed to the quantity of larger size QDs inside the TiO₂-NRs film is more than the TiO₂-NPs film, owing to their difference on pore structure limited QDs' growth; furthermore, stronger light scattering of the TiO₂-NRs film than that of the TiO₂-NPs film may be another reason to enhance the IPCE at relative longer wavelength range, which can be confirmed from their different diffuse reflectance spectra (ESI, Fig. S1†).

Fig. 3 (a) and (b) J–V characteristic curves of QDSSCs based on TiO₂-NPs and TiO₂-NRs films with different SILAR cycles of CdSe deposition. (c) and (d) Corresponding IPCE spectra of the solar cells. All of the devices employed the standard Pt CEs.

UV-Vis absorption spectra measurements were carried out to further investigate the films' optical properties prepared with different SILAR deposition times. As shown in Fig. 4a and b, under five SILAR cycles, the absorbance of the TiO₂-NPs film is relatively higher than that of the TiO₂-NRs film because of higher surface area providing more absorbing sites for QDs. But it is obvious that, their small absorbance difference is not proportional to their large BET surface area difference of 107.0/69.8. This phenomenon reflects that there should be partial inner surface area of the TiO₂-NPs film cannot accept QDs adsorption, possibly due to its narrower pore channels than that of the TiO₂-NRs film limiting ionic diffusion/reaction. Such circumstance becomes more evident, when the cycle number increases. For example, at 8 SILAR cycles (Fig. 4c), the light absorbance of the TiO₂-NRs film becomes slightly higher than that of the TiO₂-NPs film. Over 10 SILAR cycles (Fig. 4d and e), the light absorbance of the TiO₂-NRs film becomes significant higher than that of the TiO₂-NPs film, which is accompanied with the evident red-shift of absorption thresholds. This result reflects that the larger pore size of the TiO₂-NRs film not only allows continuous growth of QDs inside (the QDs quantity can be higher), but also the QDs' size can be larger. The CdSe QDs' size could be correlated to their light absorption thresholds which is associated with quantum size effect and well documented in the literature.^52–54 Accordingly, the dependences of QDs' sizes on SILAR cycles for the two kind films are given in Fig. 4f. It is obvious that, at the first several cycles, the QD's size increases proportionally by about 2.0 nm per cycle, which basically obeys the linear rule for the both kind films. However, as the SILAR cycle increases, the growth rate of QDs for the TiO₂-NPs film slows down much earlier than that of the TiO₂-NRs film. As a consequence, the finally obtained QDs inside the TiO₂-NRs film after 8 SILAR cycles become increasingly larger than the QDs grown inside the TiO₂-NPs film. The results clearly draw the importance of the pore structure of TiO₂ films effecting on the QDs' growth, which indeed has large impact on the light harvesting of the QDs sensitized films.

Fig. 4 The dependences of UV-Vis absorption spectra of QDs sensitized TiO₂-NPs and TiO₂-NRs films on SILAR cycles (a) 2 cycles, (b) 5 cycles, (c) 8 cycles, (d) 10 cycles, (e) 15 cycles, (f) the calculated QDs' sizes dependent on SILAR cycles.

In order to evaluate the pore size limited electrolyte diffusion. Photocurrent response of the two series QDSSCs to on–off circles of illumination are shown in Fig. 5. From Fig. 5a, it can be found that within 2–4 SILAR cycles, the photocurrent densities of the TiO₂-NPs series QDSSCs remain constant after light on. However, when the SILAR cycles increase to more than 8, the time resolved photocurrent density firstly quick reaches a peak value after light on and then continuously decreases under illumination. In the beginning, peak photocurrent is generated because of higher concentration of S²⁻ already present surrounding the CdSe QDs, which is enough to regenerate the oxidized QDs after their photo-excitation and electron injection to TiO₂. Subsequently, the concentration of S²⁻ inside the TiO₂ mesopores begins to diminish due to slow mass transfer of S_x²⁻/S²⁻ redox couples in the electrolyte. Therefore, the photocurrents decrease continuously. This phenomenon has been commonly reported in kinds of sensitized solar cells when the mass diffusion of liquid electrolyte becomes a hindering factor, which is generally associated with high viscosity of electrolyte and/or narrow pore channels.⁵⁶ For the case of the TiO₂-NPs series QDSSCs after 8 SILAR cycles, it is obviously due to the remaining size of pore channels after excessive QDs deposition is too small for ionic diffusion of S²⁻. For example, at 8 SILAR cycles, the residue pore channels inside the TiO₂-NPs film is expected to be only 16.2 (average pore size) − 15.2 (QDs' size) = 1 nm, while that for the TiO₂-NRs film is expected to be about 27.1 (average pore size) − 16.1 (QDs' size) = 11 nm. That is why in Fig. 6b there is no photocurrent variation after light on during the on–off cycles of illumination for the TiO₂-NRs series QDSSCs, confirming that they have no pore size limited electrolyte diffusion problems. Obviously, the remaining pore size of the TiO₂-NRs film is large enough.

Fig. 5 (a) Photocurrent response of the TiO₂ NPs series QDSSCs and (b) the TiO₂ NRs series QDSSCs to on–off circles of illumination, with respect to different SILAR cycles.

Fig. 6 The dependences of (a) electron diffusion coefficient (D_n), (b) the electron lifetime before recombination (τ_r) and (c) the calculated effective electron diffusion length (L_n) on the offsets between quasi-Fermi level of TiO₂ and electrolyte redox potential (“E_Fn − E_redox”) under different incident light intensities. The SILAR cycles were set at 8.

In order to evaluate how the TiO₂ morphology affects the electron collection kinetics, we resorted to the photocurrent/photovoltage transient decays techniques.^55,57–60 By varying the incident light intensity, the voltages of the solar cells, or more accurately the quasi-Fermi level or the electron density of the films are accordingly tuned. Electron diffusion coefficient (D_n) and electron lifetime (τ_r) at different voltages can be obtained by fitting the photocurrent and photovoltage decay curves, respectively. The voltage is equal to the offset of the electron quasi-Fermi level of TiO₂ with respect to the redox potential of electrolyte, denoted as “E_Fn − E_redox”. The fitted D_n, τ_r and the calculated effective electron diffusion length (L_n = (D_nτ_r)^−1/2)⁶¹ for the two kind photoanodes are compared in Fig. 7.

Fig. 7 Schematic illustration of different electronic behaviors between the TiO₂-NPs and TiO₂-NRs based photoanodes in CdSe QDs sensitized solar cells.

It is known that the TiO₂-NRs are formed by oriented attachment of primary crystallites, which have been sufficiently inter-fused. Hence, the grain boundary density in the TiO₂-NRs based film is considerably lower, leading to a higher D_n. Whilst for the TiO₂-NPs based film, and no oriented attachment is involved. The nanoparticles are randomly assembled, the resultant high density of grain boundaries inside the film will involve electron scattering and trapping events.⁶² Therefore, it is reasonable to detect its lower D_n. This comparison has demonstrated the significance of removing grain boundaries in photoanode films of QDSSCs in the improvement of electron transport, by our strategy of using “oriented attachment” NRs in place of NPs. The apparent τ_r as function of “E_Fn − E_redox” is showed in Fig. 7b. τ_r of the TiO₂-NRs based film is longer than that of the TiO₂-NPs based film. Longer τ_r could explain the higher V_oc and FF of the corresponding solar cell. It is suggested to be derived from the pore size limited electrolyte diffusion in the TiO₂-NPs based film, leading to a slower regeneration of the QDs, and resulting in faster interfacial recombination and lower electron lifetime. For the TiO₂-NRs case, larger pore size will eliminate the above adverse effect and resulting in longer electron lifetime. Combining D_n together with τ_r allows us to weigh the charge collection efficiency (η_cc) of the two kind films on the basis of their L_n, by using the equation L_n = (D_nτ_r)^−1/2 for calculation. From Fig. 4d, for L_n of the two kind films, or to say their η_cc, it is the TiO₂-NPs based film greater than the TiO₂-NPs based film. The better charge collection property of the TiO₂-NRs based film is consistent with the observed higher J_sc and IPCE of the corresponding solar cell in Fig. 3.

In short, as a schematic image (Fig. 7) summarized, the two kind photoanodes represent the following three different features notable for solar cell applications: (1) pore size distribution difference will impact on the QDs growth due to different pore channel controlled mass diffusion, therefore, their optimizing of light harvesting property will be different; the wider porous structure of the TiO₂-NRs film allows higher loading content of QDs and different QDs size distribution, leading to improved light harvesting efficiency; (2) the size of pore channels after QDs deposition will be differently narrowed which impacts on ionic diffusion of electrolyte, leading to different QDs regeneration issues. The wider remaining pore channels of the TiO₂-NRs film can facilitate electrolyte diffusion; (3) the density of grain boundaries in the NRs-based film is largely decreased because of a big fraction of adjacent nanoparticles have been accurately assembled aforehand during the “oriented attachment” process, which is thought to be benefit for fast electron transport. In all of these aspects, the TiO₂-NRs based photoanode are superior to the traditional TiO₂-NPs reference, which can explain their performance difference in Fig. 3.

At last, by combination with a highly efficient Cu₂ZnSnS(Se)₄ CE and the TiO₂-NRs based photoanode, the as-fabricated QDSSC presented much improved performance. The J–V curve, IPCE, and the resultant photovoltaic parameters are shown in Fig. 8. In comparison to the Pt CE based device shown in Fig. 3b, the device gave a slightly higher V_oc of 563 mV, a slightly higher J_sc of 17.5 mA cm⁻², a much improved FF of 0.605, and a high efficiency up to 5.96%. Such a performance is among the best in class QDSSCs,⁶³ and is also largely higher than the device reported in our previous work,³⁰ which was also with the Cu₂ZnSnS(Se)₄ CE but with traditional TiO₂-NPs based photoanode.

Fig. 8 (a) J–V characteristic curve of QDSSC based on TiO₂-NRs film with Cu₂ZnSnS(Se)₄ CE, (b) corresponding IPCE spectrum of the solar cell.

4. Conclusions

In summary, a high yield synthesis method of the special “oriented attachment” TiO₂-NRs has been reported. The NRs based photoanode has been critically compared with respect to the traditionally used TiO₂-NPs in QDSSC system. The notable structural advantages of TiO₂-NRs over TiO₂-NPs and their impacts on the corresponding devices' performance have been recognized as: (1) its wider porous structure allows higher loading content of QDs and different QDs size distribution, leading to improved light harvesting efficiency; (2) the wider remaining pore channels of the TiO₂-NRs film after QDs deposition can facilitate electrolyte diffusion; (3) the TiO₂-NRs film with limited grain boundaries possesses higher electron diffusion coefficient, in combination with the longer electron lifetime leading to a better charge collection efficiency. All of these led to the TiO₂-NRs QDSSC achieve J_sc improvement from 13.5 mA cm⁻² to 14.3 mA cm⁻², V_oc improvement from 486.8 mV to 513.6 mV, FF improvement from 0.33 to 0.44 and a significant improvement on the overall efficiency from 2.1% to 3.1%, in comparison to the traditional TiO₂-NPs reference. This work also highlights the general and critical importance of balanced optimizations between the multiple functions undertaken by photoanodes in QDSSCs. At last, a remarkable efficiency of 5.96% has been achieved by combination of the TiO₂-NRs based photoanode and a highly efficient Cu₂ZnSnS(Se)₄ CE.

Acknowledgements

This work was supported by 973 Program of China (2011CBA00703), We thank Analytical and Testing Center of Huazhong University Science & Technology for the sample measurements.

References

1. N. Tetreault and M. Grätzel, Energy Environ. Sci., 2012, 5, 8506–8516.
2. D. F. Underwood, T. Kippeny and S. J. Rosenthal, Eur. Phys. J. D, 2001, 16, 241–244.
3. P. Brown and P. V. Kamat, J. Am. Chem. Soc., 2008, 130, 8890–8891.
4. A. J. Nozik, Physica E: Low-dimensional Systems and Nanostructures, 2002, 14, 115–120.
5. I. Robel, V. Subramanian, M. Kuno and P. V. Q. Kamat, J. Am. Chem. Soc., 2006, 128, 2385–2393.
6. P. V. Kamat, J. Phys. Chem. C, 2007, 11, 2834–2860.
7. P. Yu, K. Zhu, A. G. Norman, S. Ferrere, A. J. Frank and A. J. Nozik, J. Phys. Chem. B, 2006, 110, 25451–25454.
8. O. E. Semonin, J. M. Luther, S. Choi, H. Y. Chen and J. Gao, Science, 2011, 334, 1530–1533.
9. W. A. Tisdale, K. J. Williams, B. A. Timp, D. J. Norris, E. S. Aydil and X. Y. Zhu, Science, 2010, 328, 1543–1547.
10. P. K. Santra and P. V. Kamat, J. Am. Chem. Soc., 2012, 134, 2508–2511.
11. X. Y. Yu, J. Y. Liao, K. Q. Qiu, D. B. Kuang and C. Y. Su, ACS Nano, 2011, 5, 9494–9500.
12. S. Cheng, W. Fu, H. Yang, L. Zhang, J. Ma, H. Zhao, M. Sun and L. Yang, J. Phys. Chem. C, 2012, 116, 2615–2621.
13. Z. Pan, H. Zhang, K. Cheng, Y. Hou, J. Hua and X. Zhong, ACS Nano, 2012, 6, 3982–3991.
14. H. Chen, W. Fu, H. Yang, P. Sun, Y. Zhang, L. Wang, W. Zhao, X. Zhou, H. Zhao, Q. Jing, X. Qi and Y. Li, Electrochim. Acta, 2010, 56, 919–924.
15. Z. W. Ren, J. Wang, Z. X. Pan, K. Zhao, H. Zhang, Y. Li, Y. X. Zhao, I. Mora-Seró, J. Bisquert and X. H. Zhong, Chem. Mater., 2015, 27, 8398.
16. P. K. Santra and P. V. Kamat, J. Am. Chem. Soc., 2012, 134, 2508–2511.
17. Z. Pan, K. Zhao, J. Wang, H. Zhang, Y. Feng and X. Zhong, ACS Nano, 2013, 7, 5215–5222.
18. J. W. Lee, D. Y. Son, T. K. Ahn, H. W. Shin, I. Y. Kim, S.-J. Hwang, M. J. Ko, S. Sul, H. Han and N. G. Park, Sci. Rep., 2013, 3, 1050.
19. J. H. Im, C. R. Lee, J. W. Lee, S. W. Park and N. G. Park, Nanoscale, 2011, 3, 4088–4093.
20. J. A. Chang, S. H. Im, Y. H. Lee, H. Kim, C. S. Lim, J. H. Heo and S. Seok, Nano Lett., 2012, 12, 1863–1867.
21. W. Zhang, R. Zhu, L. Ke, X. Liu, B. Liu and S. Ramakrishna, Small, 2010, 6, 2176–2182.
22. B. Liu and E. S. Aydil, J. Am. Chem. Soc., 2009, 131, 3985–3990.
23. E. Enache-Pommer, B. Liu and E. S. Aydil, Phys. Chem. Chem. Phys., 2009, 11, 9648–9652.
24. D. Chen, F. Huang, Y.-B. Cheng and R. A. Caruso, Adv. Mater., 2009, 21, 2206–2210.
25. F. Sauvage, D. Chen, P. Comte, F. Huang, L. P. Heiniger, Y.-B. Cheng, R. A. Caruso and M. Graetzel, ACS Nano, 2010, 4, 4420–4425.
26. Y. Shi, C. Zhu, L. Wang, C. Zhao, W. Li, K. K. Fung, T. Ma, A. Hagfeldt and N. Wang, Chem. Mater., 2013, 25, 1000–1012.
27. Y. Shi, C. Zhao, L. Wang, B. Ma, R. Gao, Y. Zhu and Y. Qiu, Adv. Funct. Mater., 2010, 20, 437–444.
28. J. G. Radich, R. Dwyer and P. V. Kamat, J. Phys. Chem. Lett., 2011, 2, 2453–2460.
29. Z. Yang, C. Y. Chen, C. W. Liu, C. L. Li and H. T. Chang, Adv. Energy Mater., 2011, 1, 259–264.
30. X. Zeng, W. Zhang, Y. Xie, D. Xiong, W. Chen, X. Xu, M. Wanga and Y.-B. Cheng, J. Power Sources, 2013, 226, 359–362.
31. H. Yuan, J. Lu, X. Xu, D. Huang, W. Chen, Y. Shen and M. Wang, J. Electrochem. Soc., 2013, 160, H624–H629.
32. X. Zeng, D. Xiong, W. Zhang, L. Ming, Z. Xu, Z. Huang, M. Wang, W. Chen and Y.-B. Cheng, Nanoscale, 2013, 5, 6992–6998.
33. L. Li, X. Yang, J. Gao, H. Tian, J. Zhao, A. Hagfeldt and L. Sun, Highly efficient CdS quantum dot-sensitized solar cells Based on a modified polysulfide electrolyte, J. Am. Chem. Soc., 2011, 133, 8458–8460.
34. X. Y. Yu, J. Y. Liao, K. Q. Qiu, D. B. Kuang and C. Y. Su, ACS Nano, 2011, 5, 9494–9500.
35. G. Larramona, C. Choné, A. Jacob, D. Sakakura, B. Delatouche, D. Péré, X. Cieren, M. Nagino and R. Bayón, Chem. Mater., 2006, 18, 1688–1696.
36. H. J. Lee, M. Wang, P. Chen, D. R. Gamelin, S. M. Zakeeruddin, M. Grätzel and M. K. Nazeeruddin, Nano Lett., 2009, 9, 4221–4227.
37. Y. Rui, Y. Li, H. Wang and Q. Zhang, Chem.–Asian J., 2012, 7, 2313–2320.
38. H. Kang, H. Choi, S. Kang, Y. Kim, T. Hyeon and Y.-E. Sung, Adv. Mater., 2008, 20, 54–58.
39. S. Ito, T. N. Murakami, P. Comte, P. Liska, C. Grätzel, M. K. Nazeeruddinl and M. Grätzel, Thin Solid Films, 2007, 516, 4613–4619.
40. Q. Shen, J. Kobayashi, L. J. Diguna and T. Toyoda, J. Appl. Phys., 2008, 133, 084304–084305.
41. G. Hodes, A. Albu-Yaron, F. Decker and P. Motisuke, Phys. Rev. B: Condens. Matter Mater. Phys., 1987, 36, 4215–4221.
42. H. Yang and H. Zeng, J. Phys. Chem. B, 2004, 108, 3492–3495.
43. G. J. Wilson, A. S. Matijasevich, D. R. G. Mitchell, J. C. Schulz and G. D. Will, Langmuir, 2006, 22, 2016–2027.
44. J. Li and H. Zeng, J. Am. Chem. Soc., 2007, 129, 15839–15847.
45. R. L. Penn and J. F. Banfield, Science, 1998, 14, 969–971.
46. M. Adachi, Y. Murata, J. Takao, J. Jiu, M. Sakamoto and F. Wang, J. Am. Chem. Soc., 2004, 126, 14943–14949.
47. D. Li, M. H. Nielsen, J. R. I. Lee, C. Frandsen, J. F. Banfield, D. Yoreo and J. James, Science, 2012, 336, 1014–1018.
48. J. Park, J. Joo, S. G. Kwon, Y. Jang and T. Hyeon, Angew. Chem., Int. Ed., 2007, 46, 4630–4660.
49. R. L. Penn and J. F. Banfield, Science, 1998, 281, 969–971.
50. V. P. Zhdanov and B. Kasemo, Phys. Rev. Lett., 1998, 81, 2482.
51. M. Lazzeri, A. Vittadini and A. Selloni, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 155409.
52. P. Ardalan, T. P. Brennan, H. B. R. Lee, J. R. Bakke, I. K. Ding, M. D. S. McGehee and F. Bent, ACS Nano, 2011, 5, 1495–1504.
53. D. R. Baker and P. V. Kamat, Adv. Funct. Mater., 2009, 19, 805–811.
54. I. Robel, M. Kuno and P. V. Kamat, J. Am. Chem. Soc., 2007, 129, 4136–4137.
55. F. Cao, G. Oskam, G. J. Meyer and P. C. Searson, J. Phys. Chem., 1996, 100, 17021–17027.
56. W. Chen, Y. Qiu, K. Yan and S. Yang, J. Power Sources, 2011, 196, 10806–10816.
57. L. Ming, H. Yang, W. Zhang, X. Zeng, D. Xiong, Z. Xu, H. Wang, W. Chen, X. Xu, M. Wang, J. Duan, Y.-B. Cheng, J. Zhang, Q. Bao, Z. Wei and S. Yang, J. Mater. Chem. A, 2014, 2, 4566–4573.
58. T. W. Holcombe, J. H. Yum, Y. Kim, K. Rakstys and M. Grätzel, J. Mater. Chem. A, 2013, 1, 13978–13983.
59. Z. Xu, D. Xiong, H. Wang, W. Zhang, X. Zeng, L. Ming, W. Chen, X. Xu, J. Cui, M. Wang, S. Powar, U. Bach and Y.-B. Cheng, J. Mater. Chem. A, 2014, 2, 2968–2976.
60. A. K. Chandiran, M. Abdi-Jalebi, M. K. Nazeeruddin and M. Grätzel, ACS Nano, 2014, 8, 2261–2268.
61. W. H. Howie, F. Claeyssens, H. Miura and L. M. Peter, J. Am. Chem. Soc., 2008, 4, 1367–1375.
62. J. Qu, G. R. Li and X. P. Gao, Energy Environ. Sci., 2010, 3, 2003–2009.
63. S. Jiao, Q. Shen, I. Mora-Seró, J. Wang, Z. Pan, K. Zhao, Y. Kuga, X. Zhong and J. Bisquert, ACS Nano, 2015, 9, 908–915.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra04209k

---