Hierarchically assembled microspheres consisting of nanosheets of highly exposed (001)-facets TiO₂ for dye-sensitized solar cells

Abstract

In this work, the mono-dispersed TiO₂ microspheres with highly exposed (001)-facets (ca. 82%), high surface area (112.2 m² g⁻¹), and self-ordered 3D porous network have been rapidly synthesized by an in situ facet-controlling approach without any organic templates and employed as photoanodes of dye-sensitized solar cells (DSSCs). Owing to the stacking of 3D nanosheets of microspheres, the self-ordered internal porous network (ca. 30 nm) provides an efficient pathway for electrolyte and dye molecules penetrating into the interior part of a microsphere. Besides, the large voids among microspheres (ca. 300 nm) establish a highway for electrolyte diffusion. The formation processes of the prepared microspheres have been discussed, in which oriented self-assembly is involved. A reasonable mechanism is proposed to explain its high dye loading capacity (dye loading per gram/surface area of TiO₂). Furthermore, according to the difference of normalized photocurrents of the cells with (001)-facet TiO₂ and Ref-TiO₂, the long wavelength region (600 to 800 nm) contributes 67.5% of the integrated photocurrent density; this result confirms its superior light scattering property. Finally, under 100 mW cm⁻² light irradiation, a high photoelectric conversion efficiency of about 11.13% was achieved, as compared to that of the cell with a Ref-TiO₂ film (8.11%). In light of their successful application in high-performing I₃⁻/I⁻-based DSSCs, it is envisaged that these TiO₂ microspheres with highly exposed (001)-facets can be used as excellent semiconductor materials in mass transfer limited cobalt-based DSSCs, and also in other fields, such as photocatalysis, water splitting, and lithium ion battery.

---

1 Introduction

Nanostructures of metal oxides with (001)-facets have been widely studied for their fundamental properties and also for technological applications in the areas of catalysis,^1–3 photo-catalytic degradation of organic pollutants,^4–10 and Li-ion batteries.¹¹ Much of the research in this direction was focused on tetragonal TiO₂.^12–14 Titanium dioxides with the facets of (101) and (001) were found to have surface energies of 0.44 J m⁻² and 0.90 J m⁻², respectively.¹⁵ Although (001)-facet TiO₂ is usually the prime product in the initial stage of crystal formation, its content reduces rapidly owing to these highly energetic facets.¹⁶ Therefore, it is desirable to obtain truncated octahedral bi-pyramidal nanocrystalline TiO₂ having eight equivalent (101)-facets and two equivalent (001)-facets on the top and bottom of its crystalline structure, respectively. Obtainment of stable film surfaces of anatase TiO₂ with highly exposed (001)-facets, under ambient conditions, becomes one of the challenging tasks in the research on tailor-made synthesis of TiO₂.¹⁷

In the recent past, the role of fluorine (or its effect) has been a subject of intensive research in the area of photocatalysis; this can be attributed to its capability for controlling the particle morphology of TiO₂ and thereby improve its photocatalytic ability.^10,18 Yang et al., made the first report on the synthesis of anatase TiO₂ microcrystals having highly reactive (001)-facets (ca. 47%);¹⁹ they used hydrofluoric acid for controlling the growth direction of TiO₂ nanoparticles. Following this work, a number of attempts have been made to synthesize sheet-like single-crystalline anatase TiO₂ with a high content of (001)-facets (∼89%).^5,15,20–22

Highly exposed (001)-facets TiO₂ single crystals have been studied focusing on the performance of dye-sensitized solar cell (DSSC).^23–28 It is believed that the percentage of exposed (001)-facets in an anatase TiO₂ plays a very important role in deciding the power conversion efficiency of its DSSC.²⁹ The (001)-facets TiO₂ (i) enhances electron transfer;³⁰ (ii) enables easier adsorption of dye;³¹ and (iii) improves scattering ability.³² Therefore, synthesizable high-energy anatase (001)-facets structures as semiconductor layers in DSSCs are very promising; however, because of their vulnerability to overlap on each other, in other words, due to the relatively large crystal thickness in the [001] direction, these structures are not expected to show a high surface area. It has been reported that the anatase TiO₂ nanosheets with 64% of (001)-facets, an average width of 1.09 μm, and an average thickness of 260 nm show an extremely low surface area of only 1.6 m² g⁻¹.²² Prevention of the aggregation of TiO₂ nanostructures, having a large amount of exposed (001)-facets, is rather a difficult task. Reduction of thickness of the nanosheets in [001] direction and increase of 2D lateral size of (001) planes, while simultaneously allowing the sheets to assemble in a hierarchical 3D architecture is one of the optimal solutions for preventing this aggregation. In this way, not only the specific surface area of the sheet-like anatase TiO₂, but also the percentage of its exposed (001)-facets can be increased. However, ordered structures of this type are yet to be obtained for anatase TiO₂.

Recently, sea urchin-like structures, nanorod-microspheres, and dandelion-like microspheres of TiO₂ have been reported to show open structures for the penetration of electrolytes.^33–36 It is important to note that the hitherto mentioned structures correspond to rutile TiO₂. In terms of lattice-packing, anatase TiO₂ is highly preferred to rutile for the application in DSSCs owing to its facile electron-transfer ability in a TiO₂ film.³⁷ Obtainment of an anatase TiO₂ with open structure is still a challenging task.^38,39 Very recently, Yu et al. synthesized a nitrogen- and fluorine-doped flower-like TiO₂ with (001)-dominated facets, applied it in a DSSC as the semiconductor film, and obtained a power conversion efficiency of 8.20%;²⁴ the percentage increases in the short-circuit current density and power conversion efficiency of the DSSC with higher content of (001)-TiO₂, with reference to the values of these parameters of the DSSC with undoped flower-like TiO₂, are 52% and 22%, respectively. Sun and co-workers synthesized hierarchical fastener-like spheres, consisting of anatase TiO₂ nanosheets with exposed (001)-facets;²⁵ they applied the pertinent TiO₂ photoanode in a dye-sensitized solar cell and achieved an improved conversion efficiency of 7.01%, compared to that of the DSSC with commercial P25 (5.78%).

In this work, the mono-dispersed TiO₂ microspheres with highly exposed (001)-facets (ca. 82%), high surface area (112.2 m² g⁻¹), good scattering ability, self-ordered 3D porous network and suitable mesopores in microspheres had been successfully synthesized by an in situ facet-controlling approach and employed as the semiconductor layer for the DSSCs. Dye loading measurement, Kubelka–Munk plots, and incident photon-to-current conversion efficiency (IPCE) spectra were used to substantiate the photovoltaic parameters of the DSSCs. Besides, we used electrochemical impedance spectra (EIS) to determine the steady-state electron density on the TiO₂ conduction band (n_s), electron diffusion coefficient (D_eff), and electron lifetime (τ_n). Time-dependent experiments were carried out to understand the formation process of the TiO₂ spheres.

2 Experimental

2.1 Materials

We obtained synthetic grade lithium iodide and iodine from Merck. Across has supplied 4-tert-butylpyridine (tBP, 96%), tert-butyl alcohol (tBA, 96%), and acetone (≥99%). Titanium(IV) tetraisopropoxide (TTIP, >98%), nitric acid (HNO₃, 70%), hydrofluoric acid (HF, ≥48%), anhydrous ethanol (≥99.5%), hexane (≥99%), Triton® X-100, and 2-methoxyethanol (≥99.5%) were procured from Aldrich. Fluke were the suppliers of 3-methoxypropionitrile (99%) and acetonitrile (ACN, ≥99.99%). The dye, coded as N719 and 1,2-dimethyl-3-propylimidazolium iodide (DMPII) were bought from Solaronix (S.A., Aubonne, Switzerland). Hexane was acquired from Avantor Performance Materials (Center Valley, PA). All solvents and reagents were supplied by commercial suppliers and used without further purification.

2.2 Preparation of spheres of highly exposed (001)-facets anatase TiO₂

Spheres of highly exposed (001)-facets anatase TiO₂ were synthesized by a facile hydrothermal route. In a typical synthesis, 1.5 mL of Triton® X-100 was added to a solution composed of 20 mL of hexane and 1 mL of anhydrous ethanol. The solution was stirred for about 10 min, and 0.9 mL of HF (≥48%) was added to it. The stirring was continued for another 30 min; during this period TTIP was introduced into it, and the related pH value of the solution is 3.2 ± 0.1. The contents were taken into an autoclave, lined with Teflon (40 mL capacity). The temperature of the autoclave was raised to 160 °C, maintained for 24 h, and then cooled to normal. A white precipitate was obtained at the bottom. The white precipitate was isolated from the solution by centrifugation at 10 000 rpm, and was subsequently washed with deionized water and anhydrous ethanol for 10 times to remove any impurities; the precipitate was then dried overnight at 80 °C. It is to be noted here that the (001)-facets anatase TiO₂ powder was also obtained through the hydrothermal route, without adding HF and used for comparative study. All the discussions below pertain to the (001)-facets anatase TiO₂ microspheres formed by the addition of HF, unless and otherwise specifically mentioned.

2.3 Characterization of spheres of highly exposed (001)-facets anatase TiO₂

Morphologies of the photoanodes were characterised by scanning electron microscopy (SEM, Hitachi S-4700). High resolution transmission electron microscope (HR-TEM) was operated at 200 keV, and the images were obtained by using a PhilipTecnai G2 LaB6 Gun TEM, having a Gatan Dual Vision CCD Camera. The compositions of TiO₂ spheres were verified by X-ray diffraction (XRD) (Rigaku, Tokyo, Japan) patterns with Cu Kα radiation. Brunauer–Emmett–Teller (BET) surface areas were determined by using “Micromeritics, ASAP2010”. A Micromeritics AutoPore IV 9520 instrument was used to obtain mercury porosimetry data. To estimate a dye loading on a TiO₂ film, the dye was desorbed form the film using a 0.01 M NaOH solution. The solution was then subjected to an UV-vis spectroscopy analysis (UV-4802, Unico, USA) for finding out the amount of dye on the film. To measure the dye loading capacity of a TiO₂ film, 10 g of TiO₂ powder was heated up to 80 °C for 20 min and then immersed into the dye solution (3 × 10⁻⁴ M solution of N719 dye in a 1 : 1 (v/v) mixture of ACN : tBA) at room temperature for 30 h. The dye loading per gram was then measured; this value was divided by the surface area of TiO₂.

2.4 Preparation of electrodes and solar cell assembly

The TiO₂ paste was prepared by adding (20 wt%) of the powder to ethanol, containing 0.5 wt% of acetyl acetate and 0.2 wt% of acetic acid. Thereafter, the paste was gently mixed under magnetic stirring at 200 rpm for 24 h. For the reference TiO₂ (Ref-TiO₂) colloid solution, the TiO₂ precursor was synthesized by the sol–gel method, described below. The TTIP was added to 100 mM nitric acid (430 mL, 70%, J. T. Baker) aqueous solution with constant stirring at 400 rpm and then heated to 85 °C for 8 h. Then the resultant colloid was cooled to room temperature and placed in the autoclave (PARR4540, USA), heated for 12 h to 240 °C; here, TiO₂ particles assume good shapes (∼20 nm). At this stage, the solution possesses about 8.5 wt% of the TiO₂. To this solution, 25% of PEG was added to obtain a paste of TiO₂ for preparing a homemade transparent layer, HTL; PEG serves here for preventing the film of TiO₂ from being cracked at later stage, and also for controlling the pore size of the film. For the second type of TiO₂ paste (homemade scattering layer, HSL), the same was incorporated with 50 wt% of light scattering TiO₂ powder (PT-501A, 15 m² g⁻¹, 100 nm, 99.74%, Ya Chung Industrial, Taiwan; with respect to the 20 nm TiO₂) for reducing the light loss by back scattering. Fluorine-doped SnO₂ conducting glass (FTO, TEC-7, 7 Ω sq⁻¹, NSG America, Inc., New Jersey, USA) was washed using a cleaning solution and rinsed with demineralized water, acetone and isopropanol, in this order. The FTO glass was then spin-coated with TTIP in 2-methoxyethanol (v/v = 1/3); this is necessary for a strong adherence of the TiO₂ film on the substrate and also for an insulation of the substrate from the electrolyte. Afterwards, a mesoscopic TiO₂ film of 16, 20, and 24 μm thickness was obtained on the substrate using a doctor-blade technique. The reference TiO₂ paste was also coated on an FTO glass and used for the comparative study, composed of 5 μm of HSL on the top of different thicknesses of HTL were prepared. For comparison, commercial TiO₂ electrodes (hereafter, COM-TiO₂), composed of 5 μm of HSL on the top of different thicknesses of commercial transparent layers (commercial Dyesol 18NR-T-TiO₂ transparent layer) were prepared.

The TiO₂ film was then trimmed so that it had an active area of 0.4 × 0.4 cm² (side portions were scrapped). The film was then slowly annealed by raising the temperature upto 500 °C at the rate of 10 °C min⁻¹, and was allowed to remain at this temperature for another 30 min. Post treatment with TiCl₄ was applied using 40 mM aqueous TiCl₄ solution at 70 °C for 60 min. After reheating the film at 500 °C for 30 min, it was cooled down to about 80 °C. The TiO₂ film was then dipped for 24 h in N719-dye (5 × 10⁻⁴ M in ACN and tBA at the volume ratio of 1/1). The photoanode of the DSSC was thus obtained. Following the immersion procedure the sensitized electrodes were shortly rinsed with acetonitrile and dried in air. The counter electrode (CE) was prepared by sputtering a layer of Pt on another FTO glass (film thickness = 50 nm, which was obtained from the calibration curve of the sputtering).

To fabricate the DSSC, the above prepared TiO₂ photoanode was paired with the counter electrode. A 25 μm-thick Surlyn® tape was used to separate them and to seal them latter by heating. The electrolyte of the DSSC consisted of a solution of 0.05 M I₂, 0.6 M DMPII, 0.1 M LiI and 0.5 M tBP in the cosolvent of MPN and ACN at the volume ratio of 1/1. After introducing the electrolyte through the gap between the two electrodes by means of a capillary technique, the gap was closed using a hot-melt glue.

2.5 Photovoltaic properties analysis

A solar simulator (XES-301S, AM1.5G, San-Ei Electric Co., Ltd., Osaka, Japan) was used to illuminate the DSSC. The light intensity (100 mW cm⁻²) was calibrated using a standard Si cell (PECSI01, Peccell Technologies, Inc., Kanagawa, Japan). Photovoltaic parameters of the device were measured by using a potentiostat/galvanostat (PGSTAT 30, Autolab, Eco-Chemie, Utrecht, the Netherlands). The same apparatus equipped with an FRA2 module was utilized to obtain the electrochemical impedance spectra (EIS) at 100 mW cm⁻², in the frequency range of 10 mHz to 65 kHz. Starting from the open-circuit condition, a bias voltage (same as the V_OC of the cell) was applied to the DSSC, at an ac amplitude of 10 mV. The EIS data were obtained and optimized by fitting to an equivalent circuit, using Z-View software.^40,41

Short-circuit conditions were used to acquire the incident photon-to-current conversion efficiency (IPCE) spectra. Another solar simulator (PEC-L11, AM1.5G, Peccell Technologies, Inc., Kanagawa, Japan) was used for illuminating the device. A monochromator (model 74100, Oriel Instrument, California, USA) controlled the light on the cell. It was moved in steps in the region of visible light to obtain the IPCE, defined as follows,

IPCE(λ) = 1240(J_SC/λφ) (1)

where λ is the wavelength, J_SC is the short-circuit photocurrent density (mA cm⁻²), and φ is the incident radiative flux (W m⁻²). We measured φ with an optical detector (model 71580, Oriel Instrument, California, USA) and a power meter (model 70310, Oriel Instrument, California, USA). Energy band gaps of the TiO₂ films were determined by using their diffuse reflection spectra; these spectra were gained by using an UV-vis-NIR spectrophotometer, fitted with an integrating sphere.

2.6 Determine energy bandgap and flat-band potential values of (001)-facets TiO₂ and Ref-TiO₂

In order to construct energy band diagram for semiconductors, their energy band gaps (E_g) and flat-band potentials (E_fb) must be determined; these can be done by using UV-vis-NIR spectrophotometry (Kubelka–Munk plots, Fig. 6a) and electrochemical impedance spectroscopy (Mott–Schottky plots, Fig. 6b), respectively. A three-electrode system was used to obtain these Mott–Schottky plots through EIS. The TiO₂ film (1 cm²), a Pt foil, and an Ag/AgCl electrode were the working, counter, and reference electrodes, respectively. The supporting electrolyte contained 0.5 M KCl in water. The Mott–Schottky plots were obtained by the same potentiostat/galvanostat as mentioned before.

A plot of (Khν)^0.5 versus hν can be used to determine the energy band gap (E_g), where K is the Kubelka–Munk transformed coefficient and hν is the incident photon energy. The value of K can be obtained through diffuse reflection spectra (Fig. 4b), by using the Kubelka–Munk function given in eqn (3).⁴²

(Khν)^0.5 ∝ hν − E_g (2)

K = (1 − R)²/(2R) (3)

where R is the diffuse reflection value. According to the plot of (Khν)^0.5 versus hν as shown in Fig. 6a, the value of E_g of a semiconductor can be estimated through an extrapolated tangent line to the plot of (Khν)^0.5 = 0 versus hν. Table 3 gives that the E_g values of (001)-facets TiO₂ and Ref-TiO₂ films are 3.15 eV and 3.08 eV, respectively.

Secondly, the Mott–Schottky plot of a semiconductor gives the E_fb of the semiconductor. A curve obtained by plotting reciprocals of root of capacitance (C⁻²) of a semiconductor against its applied voltages (E) is known as a Mott–Schottky plot. At a frequency of 80 Hz with a small amplitude of 10 mV using a computer-controlled electrochemical workstation (CHI660B, CH Instruments) A resistor–capacitor (RC) circuit can simply represent the equivalent circuit for a semiconductor/electrolyte interface. The E_fb of a semiconductor can be obtained from the intercept and slope of the fitting line of its Mott–Schottky plot. The following Mott–Schottky equation is used for the purpose.⁴³

1/(C²) = [2/(qεε₀N_DA²)][E − E_fb − (kT/q)] (4)

in which C symbolizes the capacitance of the space-charge of the film, q is its electric charge, ε means the dielectric constant of the film, ε₀ gives the permittivity of the vacuum, N_D refers to the charge density of the film, A stands for the surface area of the film (1 cm²), E is the symbol for the applied voltage, k represents Boltzmann constant, T denotes the absolute temperature, E_fb means the flat-band potential of the film.

3 Results and discussion

3.1 Synthesis and physical characterization of (001)-facets TiO₂

Mesoporous TiO₂ microspheres (thereafter, (001)-facets TiO₂) were synthesized according to a one-step hydrothermal procedure. The “Experimental section” gives the details of the procedure. A sol–gel method was applied to prepare a solution of TiO₂ microspheres, in which titanium(IV) tetraisopropoxide (TTIP) was used as the precursor and hydrofluoric acid (HF) as the shape-directing agent. The solution was autoclaved for 24 h at 160 °C to obtain a slurry of the TiO₂ spheres. After a purification process, the slurry was mixed with acetyl acetate to obtain a paste of the TiO₂ spheres. The paste was then coated on an FTO glass using a doctor blade technique.

Mono-dispersed TiO₂ microspheres have diameter ca. 1.52 μm (Fig. 1a), obtained from 330 microspheres, and several pores are obvious on the surface of a microsphere (Fig. 1b). This porous structure is compatible for the penetration of the adjacent electrolyte. Besides, it consists of TiO₂ nanosheets with lateral sizes of 30 to 50 nm and thickness of 3 to 5 nm (Fig. 1c). These sheets are stacked together with 3D porous network. The selected area electron diffraction (SAED) pattern of a microsphere (Fig. 1d) indexed to (101), (004), and (200), indicating an anatase phase for the TiO₂. Fig. 1e shows TEM image of a single microsphere, and Fig. 1f shows magnified TEM image of the single microsphere, exhibiting nanosheets at its edge. Fig. 1g is the enlarged TEM image of Fig. 1e. Fig. 1h and i are the cross-sectional and planar HR-TEM images of nanosheets, obtained from the portion marked with a square at the left and right side in Fig. 1g, respectively. These images clearly reveal that the lattice spacing parallel to the top and bottom facets as well as the lattice spacing parallel to the side facets are about 0.237 nm and 0.35 nm which correspond to the planes of (001) and (101), respectively. The synthesized TiO₂ is thus dominated by (001)-facets. Ball milling method is used to crack the microsphere to confirm that in the interior part of a microsphere is also composed of nanosheets (as seen in Fig. S1a and b of “ESI†”).

Fig. 1 (a) SEM image of the TiO₂ microspheres, (b) SEM image of an individual microsphere, (c) highly magnified SEM image of the microsphere, (d) selected area electron diffraction (SAED) pattern of the microsphere, (e) TEM image of a single TiO₂ microsphere, (f) magnified TEM images of the single TiO₂ microsphere, showing nanosheets at its edge, (g) enlarged TEM image of (e), where the enlargement is made for the lower left corner of the image in (e), (h) cross-sectional HR-TEM image of the microsphere, obtained from the portion marked with a square at the left side in (g), and (i) planar HR-TEM image of the microsphere, obtained from the portion marked with a square at the right side in (g).

Fig. 2a shows the typical XRD patterns of as-prepared TiO₂ powder and the TiO₂ powder obtained after annealing at 500 °C. The diffraction peaks at 2θ = 25.3°, 37.8°, and 48.1° can be indexed to (101), (004), and (200) crystal planes of anatase TiO₂ (JCPDS no. 21-1272), respectively. This is consistent with the SAED result (Fig. 1d). Furthermore, after the annealing, the peaks become sharper, thereby indicating a higher degree of crystallinity of the TiO₂ powder. This high degree of crystallinity is required for good electron transfer through its film, and superior performance of its DSSC. Moreover, a change in the ratio of intensity of the peaks after an annealing is a common phenomenon;^26,44 this indicates a higher degree of crystallinity for the TiO₂ powder. Included in Fig. 2a is also the XRD pattern of a homemade 20 nm-sized nanocrystalline TiO₂ powder (HTL, as seen in Fig. S1c of “ESI†”), it shows the same positions of diffraction peaks as the annealed (001)-facets TiO₂ powder.

Fig. 2 (a) XRD patterns of as-prepared TiO₂ powder and TiO₂ powder obtained after the annealing treatment at 500 °C. (b) Nitrogen sorption isotherms of the (001)-facets TiO₂ powder (c) bimodal pore size distribution of the (001)-facets TiO₂ powder. Included in (a–c) are also the curves of HTL TiO₂ powder.

Moreover, the crystal phase of TiOF₂ (JCPDS no. 08-0060), formed due to the reaction between TiO₂ and HF, is also identified in the case of as-prepared sample. The peak of the TiOF₂ disappears completely after sintering, indicating the removal of the fluoride. The percentage of exposed (001)-facets was calculated to be ca. 82%, specific details are given in the “ESI”, Fig. S2.†

Nitrogen sorption experiments and mercury porosimetry data were collected to find the surface area and pore size distribution of the TiO₂ powder. Fig. 2b shows nitrogen sorption isotherms of the (001)-facets TiO₂ powder and HTL, respectively. These type IV isotherms, with a sharp capillary condensation step at high relative pressures (P/P₀ = 0.8–0.9), imply the presence of mesoporous structures.⁴⁵ Fig. 2c shows a bimodal pore size distribution with 2 peaks, where the first peak is located at the value of 291.4 nm, corresponding to the macropores among microspheres, formed a freeway to facilitate the penetration of the electrolyte solution through macropores. For the second peak, it is located at the value of 31.2 nm and belongs to the mesopores in a microsphere formed by 3D stacking of nanosheets allowing the electrolyte solution to penetrate into the interior part of a microsphere, while dye molecules adsorbed onto the surface of interior part nanosheets of a microsphere, causing enhanced cell efficiency. As shown in Scheme 1, it is expected that this hierarchically assembled spherical structure can provide both adsorption of the dye molecules and efficient electrolyte diffusion simultaneously. Finally, Brunauer–Emmett–Teller (BET) measurement of the (001)-facets TiO₂ shows an extreme high surface area of 112.2 m² g⁻¹. Included in Fig. 2c is also the pore size distribution of HTL, it shows only one peak located at the value of 15.1 nm among nanocrystalline particles. Further, it has a smaller surface area of 88.1 m² g⁻¹. The surface areas we obtained were through experiments, and there is no reason to doubt their values. Even for calculating the geometric area, the dimensions of different nanosheets or nanoparticles in the film at random are to be taken into account and the average values only gives some rough value. Besides, the surface of a nanosheet is not smooth but rough; this may be the reason for the larger surface area (112.2 m² g⁻¹) of the microspheres, as compared to those of the nanoparticles (88.1 m² g⁻¹).

Scheme 1 Schematic representation of (left) a DSSC with macroporous (001)-facets TiO₂ film, (centre) facile penetration of electrolyte through the spheres and light scattering effect by the spheres, and (right) a single TiO₂ sphere showing penetration of electrolyte through it; a single nanosheet of the sphere adsorbed with the dye is also shown at the right side.

To estimate a dye loading on a TiO₂ film, the dye was desorbed form the film using 0.01 M NaOH solution. The solution was then subjected to UV-vis spectroscopy analysis to find the amount of dye on the film. The amount of dye loaded on each TiO₂ film is given in Table 1. In spite of the fact that all the TiO₂ films have the same thickness (∼12 μm), the amount of dye on (001)-facets TiO₂ is 70% higher than that on HTL film. This may be due to the following two reasons. First, a higher BET specific surface area of (001)-facets TiO₂ (112.2 m² g⁻¹) compared to that of HTL (88.1 m² g⁻¹). Second, the highly exposed (001)-facets (ca. 82%) results in a higher dye loading capacity (dye loading per gram/surface area of TiO₂). Here, we demonstrate an enhancement dye loading capacity of the (001)-facets TiO₂ by using the TiO₂ powder, based on the same weight of 10 g, composed of (001)-facets TiO₂ or HTL. By calculating the ratio of amount of dye loading to surface area, dye loading capacities can be obtained. According to the experimental data, the amount of dye loadings of the two types of TiO₂ were found to be quite different, i.e., 2.48 × 10⁻⁴ mol g⁻¹ for (001)-facets TiO₂ electrode and 1.52 × 10⁻⁴ mol g⁻¹ for HTL. The results show that (001)-facets TiO₂ provides higher dye loading capacity (2.21 × 10⁻⁶ mol m⁻²) than that of HTL (1.72 × 10⁻⁶ mol m⁻²); in other words, there is a 28.1% enhancement in the dye loading capacity in favor of (001)-facets TiO₂, with reference to HTL.

Table 1 Physical properties of the calcined mesoporous TiO₂ spheres (MS) and a homemade 20 nm-sized nanocrystalline TiO₂ (HTL)

Photoanode	External pore size (nm)	Internal pore size (nm)	Surface area (m^2 g^−1)	Dye loading^a (×10^−7 mol cm^−2)	Dye loading capacity^b (×10^−6 mol m^−2)
a The dye loadings are for the films with a thickness of 12 μm.b Dye loading capacity = dye loading per gram/surface area of TiO2.
(001)-facets TiO2	291.4 ± 10.4	31.2 ± 0.8	112.2 ± 0.8	2.06 ± 0.03	2.21 ± 0.03
HTL	15.1 ± 0.5	̶	88.1 ± 0.7	1.21 ± 0.02	1.72 ± 0.02

---

According to previous studies, the high surface energy of (001)-facets facilitates a strong attachment of (COOH) groups of the dye to the (001)-facets TiO₂.^19,46 In our previous studies, in order to clarify the effect of amount of (001)-facets in TiO₂ on its dye-adsorption, we calculated the dye loading capacity of nanosheets with 71%, 51%, and 28% of exposed (001)-facets, and obtained the relative values of 2.18 × 10⁻⁶, 1.98 × 10⁻⁶, and 1.81 × 10⁻⁶ mol m⁻², respectively, which are higher than that of (101)-facets TiO₂ (1.74 × 10⁻⁶ mol m⁻²).³⁰ The result indicates that with increasing amount of (001)-facets, the value of dye loading capacity also increases. Furthermore, the densities of Ti_5c atoms were calculated to be 5.1 × 10⁻² and 7.0 × 10⁻² Å⁻² on the facets of (101) and (001), respectively, based on “ab initio calculation”;⁴⁷ in other words the (001)-facets TiO₂ has a higher dye loading capacity than the (101)-facets TiO₂.

3.2 Propose growth mechanism for the formation of the anatase (001)-facets TiO₂

The polarities of the solvents used in the synthesis of the TiO₂ are reported to play crucial roles in controlling the micro- and nanostructures of the corresponding TiO₂ film. In this study, water and hexane were introduced as the co-solvents in the system. Two solvents with different polar characteristics are known to be immiscible. It has been verified that the interface of such solvents is favourable for the formation of hierarchical structures.⁴⁸ The interface of water and hexane, two immiscible solvents, has been thus utilized to develop the nanosheet assembles in the reaction system. Nucleation sites of TiO₂ are expected to develop at the interface of water and hexane; the nucleations lead to the formation of nanoparticles, which in turn self-assemble into self-ordered spheres. Fig. S3a in the “ESI†” shows such spheres and single sphere is displayed in Fig. S3b;† it is important to note that these spheres are formed in the absence of HF. The spheres are apparently composed of tiny nanoparticles. Further, the surface of the spheres formed in the absence of HF is smooth and dense, and has no obvious mesopores structure (inset of Fig. S3b†), hindering electrolyte penetration into a microsphere. Fig. S4† shows schematically the formation of a single TiO₂ microsphere with TiO₂ nanoparticles (formation in the absence of HF).

In order to obtain mesoporous spheres of (001)-facets TiO₂, we added HF in the reactant solution before hydrothermal process. Titanium(IV) tetraisopropoxide (TTIP) can be dissolved in an organic solvent, but is hardly soluble in an aqueous media, while HF behaves reversely. It is therefore assumed that the nucleation and growth of TiO₂ nanocrystals occur at the boundary of aqueous and organic phases; the self-assembly can also occur at this interface. In the present case, only aqueous media can dissolve the HF. Fluorides in a strong acidic condition favour the formation of anatase TiO₂.²⁰ Both proton and fluoride are expected to be present at very high concentrations at the water–hexane interface, as the water present in the system is small. In addition to highly acidic condition, selective adsorption of fluorides to pentacoordinate Ti (Ti_5C) atoms on (001) and (101) surfaces also promotes the growth of nanosheets along the orientations of [100] and [010].¹⁹ Besides, the nanosheets aggregate into microspheres to lower their total free energy. Therefore, it is assumed that the HF aqueous solution has three main functions in the co-solvent of water and hexane: (1) gives water molecules and thereby facilitates the formation of a water/hexane interface, (2) controls the pH of the system and thereby reduces the rate of hydrolysis of the titania, and (3) promotes the anisotropic growth of nanosheets along [100] and [010] orientations. The formation of the nanosheets and a microsphere of the TiO₂ are schematically represented in Fig. 3; unlike in the scheme shown in Fig. S4,† the TiO₂ spheres in this scheme are mesoporous, owing to addition of HF in the hydrothermal process. The morphology evolution can be proposed as follows: (1) formation of random nanosheets, (2) oriented self-assembly, and (3) further growth and densification. The first two stages might overlap. At the very beginning, titania nuclei forms at the water–hexane interface. High acidic condition and selective adsorption of fluorides on the titania nuclei are favourable for the anisotropic growth of nanosheets. It was indicated that fluorides can interact strongly with (001)-facets; thus reduces the surface energy and kinetically inhibits crystal growth which are favourable for the formation of (001)-facets. The loosely packed nanosheets assemble together to form microspheres. After a while, the nanosheets combine strongly to form densely packed microspheres; these microspheres are, however, mesoporous, as is shown in Fig. 1c. Time-dependant experiments were carried out to understand the formation process of nanosheets and microspheres of the TiO₂ (Fig. S5 and S6†).

Fig. 3 Schematic representation of the synthesis of (001)-facets TiO₂ nanosheets and microspheres (HF addition).

3.3 The photovoltaic properties and electrochemical impedance behavior of (001)-facets TiO₂ based DSSCs

To determine the photovoltaic properties of the spheres of nanosheets, DSSCs were prepared using TiO₂ microsphere electrodes, N719 sensitizer, and an iodine-based redox electrolyte in MPN/ACN (v/v = 1/1). Prior to the dye-loading, the TiO₂ film was decorated with a very thin layer of TiO₂ by a chemical bath deposition process using TiCl₄.

For all further experiments a 40 mM TiCl₄ post-treatment was applied to all photoanodes. Results of the current–voltage characterization under AM1.5G solar irradiation (100 mW cm⁻²) are summarized in Table 2 and Fig. 4a. For the sake of the comparison to the state of the art film, included in Table 2 and Fig. 4a are also the photovoltaic properties of the reference TiO₂, composed of a 5 μm scattering layer (HSL) on top of a 15 μm transparent layer (HTL), hereafter Ref-TiO₂. Fig. S7† shows the relative cross-sectional SEM images of the films of (001)-facets TiO₂ and Ref-TiO₂. In order to clarify the effect of concentration of TiCl₄ on the dye loading of the TiO₂, we measured the dye loadings of films of (001)-facets TiO₂ and Ref-TiO₂ at different concentrations of TiCl₄, i.e., at 0, 10, 20, 30, 40, and 50 mM of TiCl₄; the relative values are shown in Table S1 of “ESI†”. It can be seen in Table S1† that, both in the cases of (001)-facets TiO₂ and Ref-TiO₂, the dye loading increases steadily with the increase of concentration of TiCl₄. The (001)-facets TiO₂ film with 50 mM TiCl₄ treatment shows a 8.56% higher dye loading (3.345 × 10⁻⁶ mol m⁻²) compared to that of untreated one (3.081 × 10⁻⁶ mol m⁻²). As for Ref-TiO₂, after the post-treatment with 50 mM of TiCl₄, the dye loading is 8.22% higher for treated TiO₂ (1.946 × 10⁻⁶ mol m⁻²) compared to that of untreated Ref-TiO₂ film (1.798 × 10⁻⁶ mol m⁻²). Besides, for comparing the photovoltaic parameters of the DSSCs with the films of TiO₂ and Ref-TiO₂ under optimum condition of each cell, J–V curves of the cells were obtained at different thicknesses of the TiO₂ films. The photovoltaic parameters are given in Table S2 of the “ESI†”. The results demonstrate that for both the cells the optimum film thickness of (001)-facets TiO₂ and Ref-TiO₂ is ca. 20 um. Actually, we have obtained the photovoltaic parameters of a DSSC with a commercial TiO₂ (Com-TiO₂), and the power conversion efficient was calculated to be 7.96%, which is just close to that of the reported value in this manuscript (8.11%). The photovoltaic parameters are given in Table S2 of the “ESI†”.

Table 2 Photovoltaic parameters of the DSSCs with the films of (001)-facets TiO₂ and Ref-TiO₂, obtained for the same thickness, measured under 100 mW cm⁻² (AM1.5G); the table also shows the corresponding amounts of dye loading

Photoanode	L^a (μm)	η (%)	JSC (mA cm^−2)	JIPCE^b (mA cm^−2)	VOC (mV)	FF	Dye loading^c (×10^−7 mol cm^−2)	RS^d (Ω)	RSh^e (Ω)
a TiO2 film thickness.b The integrated photocurrent density calculated from IPCE spectrum with AM1.5G solar spectrum (JIPCE).c The dye loadings are for the films with a thickness of 20 μm.d Series resistance (RS).e Shunt resistance (RSh).
(001)-facets TiO2	20	11.13 ± 0.1	19.15 ± 0.2	17.29	784.9 ± 2.2	0.741 ± 0.003	3.31 ± 0.04	40.3	2080.5
Ref-TiO2	20	8.11 ± 0.2	15.96 ± 0.2	14.39	711.5 ± 3.1	0.712 ± 0.003	1.92.±0.03	57.6	1337.2

---

Fig. 4 (a) Photocurrent density–voltage curves of the DSSCs with the films of (001)-facets TiO₂ and Ref-TiO₂, obtained for the same thickness, measured at 100 mW cm⁻² (AM1.5G); the figure also shows the corresponding dark currents. (b) Diffuse reflection spectra of the films of (001)-facets TiO₂ and Ref-TiO₂. (c) The spherical diameter distribution of (001)-facets TiO₂, obtained from 330 microspheres; the average spherical diameter of a microsphere is 1.52 μm. (d) Incident-photon-to-current conversion efficiency (IPCE) spectra of the DSSCs with (001)-facets TiO₂ and Ref-TiO₂. (e) Plots of normalized photocurrent versus wavelength for the DSSCs with the films of (001)-facets TiO₂ and Ref-TiO₂, obtained by IPCE integration.

The photovoltaic parameters of the cells are given in Table 2. The DSSC based on (001)-facets TiO₂ yields a J_SC of 19.15 ± 0.2 mA cm⁻², a V_OC of 784.9 ± 2.2 mV, a fill factor (FF) of 0.741 ± 0.003, and a cell efficiency (η) of 11.13 ± 0.1%, while the DSSC based on Ref-TiO₂ yields a J_SC of 15.96 ± 0.2 mA cm⁻², a V_OC of 711.5 ± 3.1 mV, an FF of 0.712 ± 0.003, and a η of 8.11 ± 0.2%. The DSSC with (001)-facets TiO₂ outperforms in all the photovoltaic parameters than that of the DSSC with the Ref-TiO₂ film. It can be seen in Table 2 that the (001)-facets TiO₂ adsorbs more dye molecules than the Ref-TiO₂ owing to its high surface area value (112.2 m² g⁻¹). This is clearly the reason for the higher J_SC in favour of the cell with (001)-facets TiO₂. Besides, the dark current is higher for the DSSC with the Ref-TiO₂ film than that of the DSSC with the (001)-facets TiO₂ film (at the bottom in Fig. 4a); this result is in confirmation with the V_OC values of the two cells.

In addition to J_SC and V_OC, FF is another important parameter to determine the cell performance. FF value is dependent on the series resistance (R_S) and shunt resistance (R_Sh). R_S of the cells with the films obtained through different photoanodes are in general consistent with the corresponding FF values of the cells (Table 2). Besides, R_Sh of a photovoltaic cell indicates the resistance of leakage current of the cell, i.e., the resistance of current due to the recombination reactions. The R_Sh of a cell can be determined from the slope of the reverse bias J–V curve, i.e., from the linear region of the curve. The R_Sh can be calculated by using the following relationship:

R_Sh = (ΔV_{reverse bias}/ΔI_{reverse bias})_{at short-circuit condition} (5)

The R_Sh of the cells were estimated from the J–V curves in Fig. 4a. Table 2 gives the values of R_Sh of the DSSCs with different films. The more the R_Sh is, the more the FF value becomes (see Table 2). As a consequence, the decrease of the R_S and the increase of the R_Sh and can explain the ascension of the FF, and vice versa.

It is well-known that an increase in the light-scattering ability of a TiO₂ film leads to an increase in the light-harvesting efficiency of the dyed film, and thereby to an increase in the incident photon-to-current conversion efficiency (IPCE) of the pertinent DSSC.⁴⁹ The amount of diffusely scattered light, as a result of a beam of irradiation on a semiconductor film can be quantified by the corresponding diffuse-reflection spectrum. As a comparison, the diffuse reflection spectra of the films made of (001)-TiO₂ and Ref-TiO₂ were measured (Fig. 4b). In the full wavelength range (λ = 400 to 800 nm), the diffuse reflection (R) value of (001)-facets TiO₂ film is higher than that of Ref-TiO₂ film, especially in the invisible region (λ = 600 to 800 nm). Even in the infrared region above 700 nm, the (001)-facets TiO₂ maintains a high R value of about 55%, whereas the Ref-TiO₂ shows a value lower than 40% at this region. A possible explanation can be that the (001)-facets of TiO₂, owing to their mirror-like planes-structure, scatter the incident light to all directions within the bulk of the TiO₂, and thereby utilize the incident light in the whole wavelength region to a higher degree, than that is possible in the case of the Ref-TiO₂.³² This result suggests the superior light scattering property of (001)-TiO₂, with reference to that of Ref-TiO₂. Fig. 4c shows the spherical diameter distribution. From this distribution, the average diameter of the microspheres was estimated to be about 1.52 μm. The average diameter clearly justifies the higher scattering ability of the (001)-facets TiO₂ compared to that of the Ref-TiO₂. As shown in Scheme 1, it is expected that this superior scattering capacity can provide bright chances for the dye molecules on the TiO₂ surface area to catch the photons, and then may turn into electrons. We also measured the UV-vis absorption spectra of the dye-loaded TiO₂ films (Fig. S8 in “ESI†”). The (001)-faceted TiO₂ shows a higher absorbance, compared to the Ref-TiO₂. This result is in agreement with the corresponding J_SC values.

The incident-photon-to-current conversion efficiency (IPCE) spectra of the DSSCs with (001)-facets TiO₂ and Ref-TiO₂ in Fig. 4d and e provide more evidence for the scattering effect of (001)-facets TiO₂. The following equation is known for IPCE:

IPCE(λ) = LHE(λ)φ_inj(λ)φ_regη_col(λ) (6)

in which LHE(λ) represents the light-harvesting efficiency at a given wavelength, φ_inj(λ) stands for the quantum yield of electron injection, φ_reg is the regeneration efficiency of the dye, and η_col(λ) symbolizes the charge collection efficiency. Fig. 4d shows the cell with the (001)-facets TiO₂ exhibits higher IPCE values in the short wavelength range (λ = 400 to 600 nm), compared to the cell with the Ref-TiO₂ film. This higher IPCE value of the cell with the (001)-facets TiO₂ film could be mainly attributed to the higher dye loading of this film (3.31 × 10⁻⁷ mol cm⁻²), compared to that of Ref-TiO₂ film (1.92 × 10⁻⁷ mol cm⁻²). In the long wavelength region (λ = 600 to 800 nm), the N719 dye is expected to show only a poor absorption-ability for incident light. The high IPCE values of the DSSC in this region is therefore to be ascribed to the superior light scattering ability of (001)-facets of the TiO₂ film;^49–52 this phenomenon is also consistent with the results of the diffuse reflection spectra, which confirmed the excellent light scattering ability of the (001)-facets-TiO₂ in the longer wavelength region. The integrated photocurrent density calculated from IPCE spectrum (J_IPCE) can be estimated by using the following equation,⁵³

J_IPCE = ∫IPCE(λ)eϕ_AM1.5G(λ)dλ (7)

where the parameter e is the elementary charge and ϕ_AM1.5G is the photon flux at AM 1.5G (100 mW cm⁻²). Table 2 gives the values of J_IPCE obtained using this equation. It can be seen that the J_IPCE values follow the same trend as that of the J_SC values obtained by actual measurements with in less than 10%. Fig. 4e shows the normalized photocurrent values, normalized to the J_SC value at the wavelength of 520 nm, obtained from eqn (1), of the DSSCs with (001)-facets TiO₂ and Ref-TiO₂. It can be noticed surprisingly in Fig. 4e that the difference of normalized photocurrents of the cells with (001)-facets TiO₂ and Ref-TiO₂ is 32.5% in the short wavelength region (400 to 600 nm), and this difference is 67.5% in the long wavelength region (600 to 800 nm). We can visualize the strong impact of this highly exposed (001)-facets morphology on the conversion of lower photon energies in the long wavelength region.

This study used electrochemical impedance spectroscopy (EIS) to characterize the electron transfer behaviors in these photovoltaic devices. The pertinent equivalent circuit is used to extract some electron transfer parameters of the cells. Fig. 5a shows Nyquist plots of the DSSCs with the films of (001)-facets TiO₂ and Ref-TiO₂. Actually, the first semicircle is over-lapped with the second semicircle, owing to the low resistance of Pt counter electrode. The frequencies associated with the second semicircle in Fig. 5b indicate that the semicircle refers to the resistance at the interface of semiconductor/electrolyte. The third semicircle belongs to the Warburg impedance. So the EIS analyses are proper through entire paper. Using the pertinent fitting curves and following equations, the steady-state electron density in the TiO₂ conduction band (n_s) and electron diffusion coefficient (D_eff) at the photoanode can be determined:^{40,41,54–60}

n_s = k_BT/(q²AR_kLk_eff) (8)

D_eff = (R_w/R_k)L²k_eff (9)

in which k_B stands for Boltzmann constant, T represents absolute temperature, q means proton charge, A denotes working area of the photoanode, R_k is the symbol for charge-recombination resistance, L symbolizes the film thickness, k_eff indicates the effective rate constant for recombination, and R_w refers to the electron transport resistance in the TiO₂ film. Though R_w and R_k are not indicated in the equivalent circuit, their values were obtained through fitting curves according to previous reports and given in Table 3.^40,41,60

Fig. 5 EIS spectra of DSSCs with the films of (001)-facets TiO₂ and Ref-TiO₂ (cell gap: 25 μm). (a) Nyquist plots, the inset shows the corresponding equivalent circuit. (b) Bode phase plots.

Table 3 Physical and electrochemical parameters of the DSSCs containing the films of (001)-facets TiO₂ and Ref-TiO₂^a

Photoanode	Rk (Ω)	Rw (Ω)	τn (s)	ns (×10^−18 cm^−3)	Deff (×10^5 cm^2 s^−1)	E^NHEfb (V)	Eg (eV)
a The table shows the values of charge-transfer resistance (Rk), electron transport resistance (Rw), electron lifetime (τn), steady-state electron density on the TiO2 conduction band (ns), and electron diffusion coefficient (Deff).
(001)-facets TiO2	11.25	0.93	0.28	8.05	16.26	−0.66	3.15
Ref-TiO2	26.91	6.72	0.19	2.33	7.76	−0.58	3.08

---

The characteristic parameters of n_s and D_eff are listed in Table 3. The V_OC of a DSSC is the difference between the Fermi level of its semiconductor (here TiO₂) and the redox potential of its redox couple (here I₃⁻/I⁻). The Fermi level of a semiconductor shifts to a more negative potential, if the conduction band of the semiconductor has a higher density of steady-state electrons (n_s), and leads to a higher V_OC for the pertinent DSSC; this is true in the other way round. The steady-state electron density (n_s) of the DSSC with (001)-facets TiO₂ is about 2.5 times than that of the DSSC with the Ref-TiO₂; the respective values are 8.05 × 10⁻¹⁸ cm⁻³ and 2.33 × 10⁻¹⁸ cm⁻³ (Table 3). This higher n_s value of the cell with (001)-facets TiO₂ can be ascribed to a good electronic coupling between the sensitizer and the (001)-facets TiO₂; this better electronic coupling could increase electron injection from the sensitizer into the conduction band of the TiO₂, and thereby increase the electron concentration in the TiO₂ conduction band; these explanations are substantiated by theoretical and experimental results.^47,55,56 Further, the D_eff of the cell with (001)-facets TiO₂ (1.63 × 10⁻⁴ cm² s⁻¹) is found to be larger than that of cell with Ref-TiO₂ (7.76 × 10⁻⁵ cm² s⁻¹), which results in a more efficient electron transport ability in (001)-facets TiO₂ film and is consistent with the enhanced J_SC values. Meanwhile, it has been reported that the anatase TiO₂ crystals with exposed (001)-facets could enhance the electron transport.^30,57–59

The maximum frequencies of charge transfer (f_max) in (001)-facets TiO₂ and Ref-TiO₂ are depicted by the Bode plots of the corresponding cells in Fig. 5b. The characteristic frequency peaks of (001)-facets TiO₂ and Ref-TiO₂ are 5.75 Hz and 8.55 Hz, respectively. The recombination lifetime for electrons (τ_n) is given as follows:^40,41,56,60

τ_n = 1/(2πf_max) (10)

The values of τ_n for (001)-facets TiO₂ and Ref-TiO₂ were estimated to be 0.28 s and 0.19 s, respectively (Table 3). A long recombination lifetime for electrons implies a slow recombination rate for the electrons or a long time for their transport in the TiO₂ film, and thereby a high V_OC for the pertinent DSSC, which in turn may end up with a high conversion efficiency (η) for the cell.

3.4 Energy band diagram and light intensity analysis of (001)-facets TiO₂

In order to construct energy band diagram for semiconductors, their energy band gaps (E_g) and flat-band potentials (E_fb) must be determined; these can be done by using UV-vis-NIR spectrophotometry (Fig. 6a) and electrochemical impedance spectroscopy (EIS), respectively. Specific details are given in the “Experimental section”. Fig. 6b shows positive slopes for Mott–Schottky plots of (001)-facets TiO₂ and Ref-TiO₂; these positive slopes indicate that these are n-type semiconductors. As summarized in Table 3, the E_fb values of (001)-facets TiO₂ and Ref-TiO₂ are −0.66 eV and −0.58 eV, respectively, vs. normal hydrogen electrode (NHE). The (001)-facets TiO₂ thus shows a more negative E_fb value. If the E_fb of a TiO₂ is more negative, the difference between its Fermi level and the redox potential of the redox couple (here I⁻/I₃⁻) of its DSSC will be larger, and the corresponding V_OC will be higher. It is noted to mention that, meanwhile the conduction band of (001)-facets TiO₂ is also moving upward, which means that charge injection from LUMO of N719 dye, (−1.5 eV versus NHE), to the conduction band of (001)-facets TiO₂ will become to be limited, thus reducing photocurrent. Since the observed J_SC of (001)-facets TiO₂ is higher than that of Ref-TiO₂, we conclude that the limited charge injection is compensated by a better electronic coupling between the sensitizer and the (001)-facets TiO₂, increasing electron injection from the sensitizer into the conduction band of the TiO₂, and thereby a higher electron concentration on the TiO₂ conduction band (n_s), which is supported by theoretical and experimental results.^47,55,56 In other words the (001)-facets TiO₂ has a higher steady-state electron density in the conduction band (n_s) than that of (101)-facets TiO₂. On the other hand, a pronounced reduction in recombination rate, namely the longer electron lifetime for recombination (τ_n), increases the V_OC of (001)-facets TiO₂. By using the E_g and E_fb values of each film, their energy band constructions can be depicted in Fig. 6c.

Fig. 6 (a) Plots of Kubelka–Munk function versus energy of light for the films of (001)-facets TiO₂ and Ref-TiO₂. (b) Mott–Schottky plots for the films of (001)-facets TiO₂ and Ref-TiO₂ measured in a neutral 0.5 M KCl aqueous electrolyte (pH = 7). (c) Energy band construction of (001)-facets TiO₂ and Ref-TiO₂.

Mass transport in the electrolyte of the solar cell with (001)-facets TiO₂ is expected to be better than that in the electrolyte of the solar cell with Ref-TiO₂, owing to the presence of large pores among the microspheres of TiO₂ in the former case. To prove this, short-circuit photocurrent density (J_SC) and plateau current density (the steady-state current density at the end of a light pulse) were plotted as functions of light intensity for the cells with (001)-facets TiO₂ and Ref-TiO₂ in Fig. 9Sa and b,† respectively. Actually, it is a facile method to observe the mass transfer behaviour, especially for a DSSC with high viscosity or low mass transport ability for redox couple, and this method has been reported in literatures.^61–63 Besides, we do not want to discuss the mass transfer behavior in the bulk solution of the electrolyte. So we did not measure the EIS at very low frequency (in Warburg impedance region). The pertinent solar cells were prepared using a high-viscous MPN-based electrolyte. A pulse of light was used for 5 s at each intensity of light. At the first instance it can be seen that the J_SC increases for both the cells with increase of light. At the intensities of 1 Sun and 1.25 Suns, the J_SC is steady for 5 s of light illumination for both the cells; this indicates that the rate of mass transport is sufficient at these intensities for both the cells. With the increase of light intensity, the J_SC is still steady in the case of the cell with (001)-facets TiO₂ for 5 s of illumination, while the J_SC decreases rapidly within the 5 s at higher light intensities in the case of the cell with Ref-TiO₂ (Fig. 9Sa†). In the case of Ref-TiO₂, under 2 sun illumination, the photocurrent decreases sharply by about 24%, with reference to its initial value (Fig. 9Sb†). At higher intensity of light, the magnitude of light will be higher, the injected electrons will be higher and the mass transport should cope with this higher amount of injected charge. The decreased J_SC and plateau current density in the case of the cell with Ref-TiO₂ at higher light intensities of light clearly indicate that the mass transport is insufficient in this cell; this is to be attributed to the limitations of penetration of electrolyte into the structure of Ref-TiO₂. The plateau current density shows a linear behavior for the cell with (001)-facets TiO₂ at all the intensities (Fig. S9b†), thereby indicating an excellent mass transport in (001)-facets TiO₂, as against that in Ref-TiO₂; this excellent mass transport is to be attributed to the presence of large pores among the microspheres of TiO₂ in the former case.

4 Conclusions

Microspheres of highly exposed (001)-facets TiO₂ were synthesized. Scanning electron microscope (SEM) images show several pores in individual microsphere and also reveal that each microsphere consists of TiO₂ nanosheets with lateral sizes of 30 to 50 nm and thickness of 3 to 5 nm. Analysis of HR-TEM images suggests the dominance of (001)-facets of the synthesized TiO₂. X-ray diffraction (XRD) patterns suggest the formation of a crystalline anatase TiO₂. Nitrogen sorption isotherms imply the presence of a bimodal pore size distribution. Brunauer–Emmett–Teller (BET) surface area measurement shows the specific surface area of the TiO₂ film as 112.2 m² g⁻¹. In combination with time-dependent observation, the proposed mechanism during the formation processes involve oriented self-assembly. The DSSC with the (001)-facets TiO₂ exhibits a power conversion efficiency (η) of 11.13% under 1 sun illumination, which represents a 37.2% improvement over that obtained for a DSSC containing the common Ref-TiO₂ (η = 8.11%). We believe that this power conversion efficiency (11.13%) is the highest ever reported for a dye-sensitized solar cell using a liquid electrolyte and a TiO₂ film with highly exposed (001)-facets.

Superior light scattering ability of the (001)-facets TiO₂ film (diffuse reflection spectra), enhanced dye loading, and higher diffusion coefficient (D_eff), compared with that of the Ref-TiO₂ film, are prominent reasons for the elevated J_SC in favor of the cell with (001)-facets TiO₂. Further, larger band gap energy of the (001)-facets TiO₂ compared to that of Ref-TiO₂ (Kubelka–Munk plots), negative shift Fermi level, higher steady-state electron density in the TiO₂ conduction band (n_s), longer electron lifetime (τ_n) lead to higher V_OC in favor of the cell with (001)-facets TiO₂. Further improvement in the structure of (001)-facets TiO₂ aiming for a better cell performance is in progress.

In summary, we demonstrate that macroporous TiO₂ film, consisting of nanosheets-assembled microspheres, is an excellent semiconductor film for a dye-sensitized solar cell. Experiments related to the dynamics of short-circuit photocurrent density (J_SC) and plateau current density under different light intensities clearly reveal that the mass transport is excellent in (001)-facets TiO₂, as against in Ref-TiO₂; this excellent mass transport is to be attributed to the presence of large pores among the microspheres of TiO₂. This beneficial property of facile mass transport in (001)-facets TiO₂ greatly reduces the detrimental effects of using high viscous electrolytes or electrolytes with bulky redox couples in DSSCs. It has been established through this work that a film of TiO₂ microspheres, with highly exposed (001)-facets, can be used as an excellent semiconductor in mass transfer limited cobalt-based dye-sensitized solar cells. Further research on (001)-facets TiO₂ aiming at its application in cobalt-based DSSCs is in progress.

Acknowledgements

HR-TEM support received from NanoCore, the Core Facilities for Nanoscience and Nanotechnology at Academia Sinica in Taiwan is greatly appreciated. This work was supported in part by the Ministry of Science and Technology (MOST) of Taiwan under grant number MOST 103-2119-M-007-012.

Notes and references

1. M. Schrinner, M. Ballauff, Y. Talmon, Y. Kauffmann, J. Thun, M. Moller and J. Breu, Science, 2009, 323, 617.
2. N. Tian, Z. Y. Zhou, S. G. Sun, Y. Ding and Z. L. Wang, Science, 2007, 316, 732.
3. B. Lim, M. J. Jiang, P. H. C. Camargo, E. C. Cho, J. Tao, X. M. Lu, Y. M. Zhu and Y. N. Xia, Science, 2009, 324, 1302.
4. N. Roy, Y. K. Sohn and D. Pradhan, ACS Nano, 2013, 7, 2532.
5. Y. Dai, C. M. Cobley, J. Zeng, Y. Sun and Y. Xia, Nano Lett., 2009, 9, 2455.
6. S. W. Liu, J. G. Yu and M. Jaroniec, J. Am. Chem. Soc., 2010, 132, 11914.
7. N. Roy, Y. H. Park, Y. K. Sohn, K. T. Leung and D. Pradhan, ACS Appl. Mater. Interfaces, 2014, 5, 16498.
8. L. A. Gu, J. Y. Wang, H. Cheng, Y. Z. Zhao and X. J. Han, ACS Appl. Mater. Interfaces, 2013, 5, 3085.
9. Q. J. Xiang and J. G. Yu, Chin. J. Catal., 2011, 32, 525.
10. J. C. Yu, J. G. Yu, W. K. Ho, Z. T. Jiang and L. Z. Zhang, Chem. Mater., 2002, 14, 3808.
11. X. L. Cheng, M. Hu, R. Huang and J. S. Jiang, ACS Appl. Mater. Interfaces, 2014, 6, 19176.
12. M. Grätzel, Nature, 2001, 414, 338.
13. W. G. Yang, F. R. Wan, Q. W. Chen, J. J. Li and D. S. Xu, J. Mater. Chem., 2010, 20, 2870.
14. D. K. Roh, J. A. Seo, W. S. Chi, J. K. Koh and J. H. Kim, J. Mater. Chem., 2012, 22, 11079.
15. D. Q. Zhang, G. S. Li, X. F. Yang and J. C. Yu, Chem. Commun., 2009, 4381.
16. Y. Yin and A. P. Alivisatos, Nature, 2005, 437, 664.
17. Y. W. Jun, M. F. Casula, J. H. Sim, S. Y. Kim, J. Cheon and A. P. Alivisatos, J. Am. Chem. Soc., 2003, 125, 15981.
18. H. Park and W. Choi, J. Phys. Chem. B, 2004, 108, 4086.
19. H. G. Yang, C. H. Sun, S. Z. Qiao, J. Zou, G. Liu, S. C. Smith, H. M. Cheng and G. Q. Lu, Nature, 2008, 453, 638.
20. X. Han, Q. Kuang, M. Jin, Z. Xie and L. Zheng, J. Am. Chem. Soc., 2009, 131, 3152.
21. G. Liu, H. G. Yang, X. Wang, L. Cheng, J. Pan, G. Q. Lu and H. M. Cheng, J. Am. Chem. Soc., 2009, 131, 12868.
22. H. G. Yang, G. Liu, S. Z. Qiao, C. H. Sun, Y. G. Jin, S. C. Smith, J. Zou, H. M. Cheng and G. Q. Lu, J. Am. Chem. Soc., 2009, 131, 4078.
23. D. P. Wu, S. Zhang, S. W. Jiang, J. J. He and K. Jiang, J. Alloys Compd., 2015, 624, 94.
24. J. Yu, Y. L. Yang, R. Q. Fan, L. Li and X. Y. Li, J. Phys. Chem. C, 2014, 118, 8795.
25. W. W. Sun, K. Sun, T. Peng, S. J. You, H. M. Liu, L. L. Liang, S. S. Guo and X. Z. Zhao, J. Power Sources, 2014, 262, 86.
26. X. Wu, Z. G. Chen, Q. M. Lu and L. Wang, Adv. Funct. Mater., 2011, 21, 4167.
27. D. P. Wu, J. J. He, S. Zhang, K. Cao, Z. Y. Gao, F. Xu and K. Jiang, J. Power Sources, 2015, 282, 202.
28. J. Xu, G. X. Wang, J. J. Fan, B. S. Liu, S. W. Cao and J. G. Yu, J. Power Sources, 2015, 274, 77.
29. J. G. Yu, J. J. Fan and K. L. Lv, Nanoscale, 2010, 2, 2144.
30. J. D. Peng, P. C. Shih, H. H. Lin, C. M. Tseng, R. Vittal, V. Suryanarayanan and K. C. Ho, Nano Energy, 2014, 10, 212.
31. J. J. Fan, W. Q. Cai and J. G. Yu, Chem.–Asian J., 2011, 6, 2481.
32. H. M. Zhang, Y. H. Han, X. L. Liu, P. R. Liu, H. Yu, S. Q. Zhang, X. D. Yao and H. J. Zhao, Chem. Commun., 2010, 46, 8395.
33. Y. Rui, Y. Li, Q. Zhang and H. Wang, CrystEngComm, 2013, 15, 1651.
34. E. H. Kong, Y. J. Chang, Y. C. Park, Y. H. Yoon, H. J. Park and H. M. Jang, Phys. Chem. Chem. Phys., 2012, 14, 4620.
35. C. M. Lan, S. E. Liu, J. W. Shiu, J. Y. Hu, M. H. Lin and W. G. Diau, RSC Adv., 2013, 3, 559.
36. E. Hosono, S. Fujihara, H. Imai, I. Honma, I. Masaki and H. S. Zhou, ACS Nano, 2007, 1, 273.
37. N. G. Park, J. van de Lagemaat and A. J. Frank, J. Phys. Chem. B, 2000, 104, 8989.
38. B. X. Lei, P. Zhang, M. L. Xie, Y. Li, S. N. Wang, Y. Y. Yu, W. Sun and Z. F. Sun, Electrochim. Acta, 2015, 173, 497.
39. B. T. Liu, C. H. Jin, Y. Ju, L. L. Peng, L. L. Tian, J. B. Wang and T. J. Zhang, Appl. Surf. Sci., 2014, 311, 147.
40. F. T. Li, X. J. Wang, Y. Zhao, J. X. Liu, Y. J. Hao, R. H. Liu and D. S. Zhao, Appl. Catal., B, 2014, 144, 442.
41. D. H. Chen, F. Z. Huang, Y. B. Cheng and R. A. Caruso, Adv. Mater., 2009, 21, 2206.
42. J. H. Huang, P. Y. Hung, S. F. Hu and R. S. Liu, J. Mater. Chem., 2010, 20, 6505.
43. M. Lazzeri, A. Vittadini and A. Selloni, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 155409.
44. J. D. Debad, J. C. Morris, V. Jynch, P. Magnus and A. J. Bard, J. Am. Chem. Soc., 1996, 118, 2374.
45. S. Hore, C. Vetter, R. Kern, H. Smit and A. Hinsch, Sol. Energy Mater. Sol. Cells, 2006, 90, 1176.
46. M. Grätzel, Inorg. Chem., 2005, 44, 6841.
47. S. Ito, P. Chen, P. Comte, M. K. Nazeeruddin, P. Liska, P. Péchy and M. Grätzel, Progress in Photovoltaics: Research and Applications, 2007, 15, 603.
48. Y. Chiba, A. Islam, R. Komiya, N. Koide and L. Han, Appl. Phys. Lett., 2006, 88, 223505.
49. A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo and H. Pettersson, Chem. Rev., 2010, 110, 6595.
50. Q. Wang, S. Ito, M. Grätzel, F. Fabregat-Santiago, I. Mora-Sero, J. Bisquert, T. H. Bessho and H. Imai, J. Phys. Chem. B, 2006, 110, 25210.
51. F. de Angelis, S. Fantacci, A. Selloni, M. K. Nazeeruddin and M. Grätzel, J. Am. Chem. Soc., 2007, 129, 14156.
52. Q. Wang, J. E. Moser and M. Grätzel, J. Phys. Chem. B, 2005, 109, 14945.
53. M. M. Maitani, K. Tanaka, D. Mochizuki and Y. Wada, J. Phys. Chem. Lett., 2011, 2, 2655.
54. C. Hu, X. Zhang, W. T. Li, Y. Yan, G. C. Xi, H. F. Yang, J. F. Li and H. Bai, J. Mater. Chem. A, 2014, 2, 2040.
55. W. J. Ong, L. L. Tan, S. P. Chai, S. T. Yong and A. R. Mohamed, Nanoscale, 2014, 6, 1946.
56. M. Adachi, M. Sakamoto, J. Jiu, Y. Ogata and S. Isoda, J. Phys. Chem. B, 2006, 110, 13872.
57. L. Han, N. Koide, Y. Chiba, A. Islam and T. Mitate, C. R. Chim., 2006, 9, 645.
58. L. Han, N. Koide, Y. Chiba and T. Mitate, Appl. Phys. Lett., 2004, 84, 2433.
59. L. P. Heiniger, F. Giordano, T. Moehl and M. Grätzel, Adv. Energy Mater., 2014, 4, 1400168.
60. M. Pazoki, N. Taghavinia, A. Hagfeldt and G. Boschloo, J. Phys. Chem. C, 2014, 118, 16472.
61. Y. Chen, F. Z. Huang, W. C. Xiang, D. H. Chen, L. Cao, L. Spiccia, R. A. Caruso and Y. B. Cheng, Nanoscale, 2014, 6, 13787.
62. J. Tauc, R. Grigorovici and A. Vancu, Phys. Status Solidi, 1966, 15, 627.
63. A. J. Nozik and R. Memming, J. Phys. Chem., 1996, 100, 13061.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra26307g

---