A sol-hydrothermal route to truncated tetragonal bipyramid nanocrystals and hierarchical hollow microspheres of anatase TiO₂ for application in dye-sensitized solar cells

Abstract

In this work, a series of nano- and microcrystals of anatase TiO₂ including well-defined truncated tetragonal bipyramids (TTBs) (15–20 nm) and hierarchical hollow microspheres (HHMs) (1.5 μm) were successfully synthesized using a facile sol-hydrothermal approach free from surfactants and templates. The latter HHMs also consisted of similar TTBs but bigger in size. The conversion pathway to TiO₂ mainly depended on the volume ratio of H₂O₂ to NH₃. In the presence of NH₄F, small TTBs were generated as a result of inhibited growth along both [100] and [001] directions. The formation of HHMs through gradual depletion of the yolk followed a typical Ostwald ripening process. Their photovoltaic activities were investigated on dye-sensitized solar cells (DSSCs) using bilayer photoanodes made of typical TTBs and HHMs. The cells delivered a power conversion efficiency (PCE) of 9.06%, corresponding to a 51.5% increment over those fabricated in the same way using P25 and 400 nm TiO₂ (CCIC). Meanwhile, the short-circuit current density (J_SC) reached up to 18.38 mA cm⁻², which was 1.46 times that of the P25 + CCIC cells. They were also superior to previously reported DSSCs based on highly crystalline nanooctahedra and agglutinated mesoporous microspheres. The small particle size of TTBs and micrometer size and hierarchical structure of HHMs endowed the bilayer photoanodes with a synergetic effect of dye loading and light scattering. Moreover, their exposed {001} and {101} facets, which could facilitate spontaneous charge separation through preferential carrier flow towards specific facets, and highly crystalline anatase phase rendered the cells efficient in charge collection.

---

1. Introduction

Ever-increasing global energy consumption, depleting fossil fuels and overwhelming environmental issues have made it urgent to develop new types of efficient, renewable and clean energy-related devices. Of particular interest are the photovoltaic (PV) technologies that offer a number of strategic benefits by directly converting abundant solar energy into electricity. Among these PV devices, dye-sensitized solar cells (DSSCs) stand out as a rising star owing to their intrinsic superiority in separating the function of light absorption from electron transport during the photoelectric conversion process, allowing for a large choice for each part. The low production cost, facile fabrication process and high power conversion efficiency (PCE) make DSSCs one of the most promising alternatives to silicon solar cells.^1–7 After more than two decades' intensive exploration, a certified PCE as high as 13%⁸ has been demonstrated, which almost doubled the original value¹ by O'Regan and Grätzel in 1991. Among the components of DSSCs, photoanode is of pivotal importance to the PCE due to its multiple roles such as supporting dyes, scattering light, transporting electrons and providing channels for electrolytes. Since the advent of the DSSC technology,¹ TiO₂ has been widely used as the photoanode material in light of its photo- and chemical stability, appropriate Fermi level, fast electron mobility, ease and versatility of preparation and diversity in morphology and size. Therefore, tailoring the composition and configuration of TiO₂ photoanodes is supposed to be an effective strategy to improve PCE.

Conventionally, TiO₂ photoanode films are made from random 20 nm nanoparticles.^1,3 The advantage of such films is that they can accommodate abundant dye molecules due to the large specific surface area. However, the small particle size and the presence of numerous surface trapping sites render them inefficient in light scattering and charge collection. To enhance the light scattering, larger particles have to be introduced. They could form a monolayer alone or could be mixed with small particles as scattering centers in a single layer.^4,9 However, this usually boosts light scattering at the cost of sacrificing dye loading. To address this problem, various multifunctional TiO₂ architectures have been developed. For instance, porous solid submicrospheres, multi-shell hollow nanospheres and hierarchical microspheres consisting of nanorods and nanoparticles have been used to fabricate monolayer photoanode based DSSCs and the resultant PCEs ranged from 7.20% to 10.34%.^10–13 Alternatively, large particles could form independent scattering layers on top of active layers made from small particles, giving rise to bilayer photoanode films bearing a synergetic effect of dye loading and light scattering.^{4–6,14–21} Usually, P25 or 20 nm TiO₂ nanoparticles were chosen as the active layers. When P25 was used, PCEs of 7.05%¹⁴ and 8.27%¹⁵ were obtained with hollow microspheres of anatase TiO₂ containing V-shaped channels and having shell-in-shell structures as the scattering layers, respectively. Small TiO₂ nanoparticles gave better PCEs, varying from 8.84% to 9.43%, once they were combined with multifunctional TiO₂ structures such as porous submicro- and microspheres in solid or hollow configurations for making DSSCs.^5,16–18 Although three- or four-layered structures of DSSC photoanodes^4,6,22 have also been proposed and demonstrated to be superior to the mono- and bilayer ones, their fabrications were complicated and could be tedious to integrate all different particles into a whole.

To improve the charge collection efficiency, 1 D TiO₂ nanostructures^23–26 or thin blocking layers²⁷ have been devised to facilitate electron transport or retard charge recombination. It was also revealed that anatase TiO₂ nanoparticles with high crystallinity could facilitate electron transport and thus offer an efficient charge collection.^23,28 Recently, a spontaneous charge separation through preferential carrier flow towards specific facets was observed on highly crystalline faceted TiO₂,^29,30 which might inhibit charge recombination, promote electron transport and enhance charge collection in DSSCs. Some efforts have been made to deploy highly crystalline or faceted TiO₂ nanostructures as different layers of photoanodes, however, the PCEs of these DSSCs are not competitive.^19–21

Previously, most of the TiO₂ particles constructing bi- or even multilayer photoanodes were obtained from different synthetic methods or commercial sources. In this work, using one facile sol-hydrothermal approach, we were able to synthesize a series of nano- and microcrystals of anatase TiO₂ ranging from well-faceted truncated tetragonal bipyramids (TTBs) (15–20 nm) to hierarchical hollow microspheres (HHMs) (1.5 μm) consisting of bigger similar TTBs. To prepare bilayer photoanodes of DSSCs, typical TTBs and HHMs were chosen as the active and scattering layers respectively. Effects of H₂O₂, NH₃ and NH₄F on their shape and size were scrutinized. The involved growth mechanisms were proposed. Their photovoltaic activities were investigated by fabricating N719 dye-sensitized solar cells. For comparison, the other two cells were also fabricated with P25 as the active layers and typical HHMs or 400 nm TiO₂ particles (CCIC) as the scattering layers. The cells were subjected to current–voltage (J–V) curve measurements and the dynamics of electron transport and charge recombination were analyzed by electrical impedance spectra (EIS). Their differences in the photovoltaic performances were elucidated.

2. Experimental section

All chemicals were used as received. Detailed information of the chemicals is available in the ESI†.

2.1 Synthesis of nano- and microcrystals of anatase TiO₂

Nano- and microcrystals of anatase TiO₂ were synthesized via a facile sol-hydrothermal approach similar to the previous report.³¹ Briefly, an aqueous solution containing z mmol of NH₄F and 5.0 mL of ultrapure water was added in the sol formed by the oxidation of Ti powder (20.0 mg) in the presence of both H₂O₂ (x mL) and NH₃ (y mL) aqueous solutions. The mixtures then underwent a hydrothermal process to produce TiO₂. Here the sum of x and y is 10.0 and z ranges from 2.0 to 10.0. Samples prepared with x mL of H₂O₂, y mL of NH₃ and z mmol of NH₄F are defined as T_x/y/z and the ratio of the three variables, namely x/y/z, is defined as R_HNF. It should be noted that the units of these three variables are different. Typical truncated tetragonal bipyramids (TTBs) and hierarchical hollow microspheres (HHMs) are termed as T_2/8/5 and T_9/1/9, respectively.

2.2 Investigations on the experimental conditions

Effects of H₂O₂ (2.0–9.0 mL), NH₃ (1.0–8.0 mL), NH₄F (2.0–10.0 mmol) and reaction temperature (140–200 °C) and duration (2–48 h) on the morphology and size of the anatase TiO₂ crystals were investigated with similar experimental procedures. Control experiments in which NH₄F was substituted by NaF or NH₄Cl were carried out. Besides, dry powders of TiO₂ nanorods (T_2/8 and T_9/1) reported in our previous work³¹ were hydrothermally treated with aqueous solutions containing 2.0–10.0 mmol of NH₄F and 15.0 mL of ultrapure water. In addition, NH₄F solid (6.0, 4.0 and 9.0 mmol) was directly added into the hydrothermally treated reaction solutions of T_2/8 and T_9/1 followed by another cycle of hydrothermal treatment. Here, it should be pointed out that the hydrothermal processes of these control experiments all proceeded for 24 h at 200 °C.

2.3 Fabrication of DSSCs

2.3.1 Pretreatment of substrates. FTO glass substrates (1.1 mm in thickness, 80% transmittance in the visible light region, 5% haze, 10 Ω sq⁻¹) were washed with detergent, ultrapure water and anhydrous ethanol in sequence and then immersed in a TiCl₄ aqueous solution (40 mM) at 70 °C for 30 min followed by washing with ultrapure water and anhydrous ethanol and sintering at 450 °C for 30 min.

2.3.2 Preparation of electrodes. Photoanodes were prepared in a bilayer configuration, namely, a scattering layer on top of an active layer. First, dry powders of T_2/8/5 were mixed with ethyl cellulose, terpineol, acetic acid and ethanol to form a viscous paste. Then, the paste was spread onto the pretreated substrates by the doctor-blading technique with adhesive tape (Scotch brand, 3M) as frame and spacer. After removing the tape, the substrate was preheated at 70 °C for 30 min on a hot plate and then calcined under an air flow at 325 °C for 5 min, at 375 °C for 5 min, at 450 °C for 15 min, and at 500 °C for 15 min. Subsequently, a scattering layer made of T_9/1/9 was coated on the preformed semitransparent layer of T_2/8/5 and heated in the same way. Then, the as-prepared TiO₂ bilayer photoanode films were immersed in a TiCl₄ aqueous solution (40 mM) at 70 °C for 30 min and washed with ultrapure water and anhydrous ethanol and then sintered at 450 °C for 30 min. When cooled down to 80 °C, the prepared photoanode (T_2/8/5 + T_9/1/9) was immersed in an ethanol solution of N719 dye (5.0 mM) in the dark for 20 h to guarantee sufficient dye loading. Afterwards, the films were rinsed with anhydrous ethanol to remove loosely bound dye molecules and then dried under an N₂ stream. For comparison, photoanodes based on TiO₂ bilayer films of P25 + T_9/1/9 and P25 + CCIC were prepared in the same way as described above, in which commercial P25 served as the active layers and T_9/1/9 and CCIC served as the scattering layers, respectively. Pt counter electrode was prepared by drop-casting an ethanol solution of H₂PtCl₆ (10.0 mM) onto the FTO glass, followed by sintering at 450 °C for 30 min. It should be mentioned here that the ramp rate of all the heating profiles was 2 °C min⁻¹.

2.3.3 Cell assembly. The as-prepared dye-sensitized TiO₂ photoanode and Pt counter electrode were assembled into a sandwich type cell using a hot-melt gasket (60 μm, Surlyn 1702, DuPont) as the spacer. The empty cell was tightly held and sealed by heating (130 °C). Then, a thin layer of electrolyte was introduced into the inter-electrode space from the counter electrode through a predrilled hole via the vacuum backfilling technique. Finally, the hole was sealed with a microscope cover slide and Surlyn film to avoid leakage of the electrolyte solution. The liquid electrolyte used here was composed of I₂ (0.03 M), 1-butyl-3-methylimidazolium iodide (BMII) (0.60 M), guanidine thiocyanate (GuSCN) (0.10 M) and tert-butylpyridine (TBP) (0.50 M) in acetonitrile.

The active area of the solar cells was 0.16 cm². Each cell was covered with a black mask with an aperture to screen any additional illumination through the lateral space. To make the measurement more reliable, at least three identical cells based on each combination were assembled and characterized. The average standard deviation was ±0.2%.

2.4 Characterization

X-ray diffraction (XRD) patterns were collected on a Bruker D8 focus diffractometer with Cu Kα radiation and a Lynx Eye detector at a scanning rate of 0.5° min⁻¹. Field emission scanning electron microscopy (FESEM) was performed with a Hitachi S-4800 scanning electron microscope operated at an acceleration voltage of 10.0 kV. Transmission electron microscopy (TEM) images were obtained on an FEI Tecnai F20 transmission electron microscope operated at an accelerating voltage of 200 kV. N₂ adsorption–desorption isotherm curves and Brunauer–Emmett–Teller (BET) surface areas were determined using a Micromeritics ASAP 2020 surface-area and pore-size analyzer. All the samples were degassed at 120 °C prior to BET measurements.

The diffuse reflectance spectra of the bilayer TiO₂ films were measured on an UV-vis-NIR spectrophotometer (UV-3100) equipped with an integrating sphere. The amount of dye adsorbed by each TiO₂ electrode was determined by immersing the bilayer photoanode in a 0.1 M NaOH solution in water and ethanol mixture (1 : 1, v/v) for sufficient desorption and then measuring the absorption spectrum of the resulting solution using a UV/vis spectrophotometer (U-4100). The photocurrent–voltage (J–V) tests of DSSCs were performed on a 94023A Oriel sol3A solar simulator (Newport Oriel) under AM 1.5 illumination (100 mW cm⁻²) in ambient conditions. The light source was a 450 W Xenon lamp. The incident light intensity was calibrated with a standard Si solar cell. Electrochemical impedance spectra (EIS) were measured with a Zennium electrochemical workstation (Zahner Co., Germany) in dark with an AC modulation signal of 10 mV and bias DC voltage of 0 V. The frequency range was between 10 mHz and 100 kHz. The obtained impedance spectra were analyzed by the ZView software with an appropriate equivalent circuit model.

3. Results and discussion

In our previous work,³¹ a series of tapered tetragonal nanorods of anatase TiO₂ enclosed by {001}, {100} and {101} facets were prepared in aqueous solution using Ti powder as the precursor via a sol-hydrothermal process with the presence of both H₂O₂ and NH₃. We found that H₂O₂ and NH₃ had opposite effects on the growth of the nanorods. NH₃ facilitated the oriented growth of the nanorod along its [001] axis, while H₂O₂ suppressed such anisotropic growth. To further tune the morphology and size of the TiO₂ products, here NH₄F was introduced into the reaction system. Compared with the widely used HF, which is highly efficient but more toxic and corrosive in tailoring the crystal facets of TiO₂,^32,33 NH₄F appears to be a promising alternative as it could mildly decompose into NH₃ and HF in water and effectively facilitate the crystallization process.³⁴ Actually, it has been widely utilized in preparing TiO₂ nanotubes because of its dissolving and etching capability in the electrochemical anodization processes.^35–38

3.1 Effects of H₂O₂, NH₃ and NH₄F

Here, three cases with typical volume ratios of H₂O₂ to NH₃ (R_HN), namely, 2/8, 5/5 and 9/1, which yielded the representative nanorods corresponding to samples of T_2/8, T_5/5 and T_9/1 in our previous work, were subjected to further investigation by adding NH₄F into the reaction system (Fig. 1). In each case, the amount of NH₄F was varied from 2.0 mmol to 10.0 mmol. For R_HN = 2/8, without NH₄F, rod-like nanoparticles were obtained.³¹ Similar morphology could be retained when small quantities of NH₄F, e.g., 2.0 mmol, were used (Fig. 1a). With the increase of the amount of NH₄F, nanorods gradually vanished and turned into well-defined TTBs (Fig. 1b and c). When the amount of NH₄F was larger than 6.0 mmol, e.g. 10.0 mmol, some larger tetragonal bipyramids and irregular nanoparticles were observed (Fig. 1c). Such shape evolution trends also hold for the case of R_HN = 5/5 (Fig. 1d–f). Therefore, NH₄F was believed to play a similar role to H₂O₂ in suppressing the oriented growth of TiO₂. For R_HN = 9/1, diverse particles were obtained at varied amounts of NH₄F. At 2.0 mmol NH₄F, the products were short tetragonal nanorods bearing smooth surfaces and small etched holes at the vertices (Fig. 1g). At 4.0 mmol NH₄F, well-defined TTBs with larger and deeper holes were observed (Fig. S1a†). Clearly, the etching effect of NH₄F was greatly enhanced in this case. Then, increasing the amount of NH₄F to 6.0 mmol afforded nanoparticles with similar morphology but smaller in size (Fig. 1h). At even higher amount of NH₄F, e.g., 8.0 mmol (Fig. S1b†) or 10.0 mmol (Fig. 1i), the products turned into HHMs constructed by TTBs. One difference between these two HHMs was that the shell of T_9/1/8 was slightly thicker than that of T_9/1/10. All X-ray diffraction (XRD) patterns of these products indicate that they were pure anatase TiO₂ (Fig. S2†).

Fig. 1 SEM images of samples T_2/8/2 (a), T_2/8/6 (b), T_2/8/10 (c), T_5/5/2 (d), T_5/5/6 (e), T_5/5/10 (f), T_9/1/2 (g), T_9/1/6 (h) and T_9/1/10 (i).

To identify the role of NH₄F in the growth of TiO₂ crystals, control experiments were carried out. Fig. S3 and S4† show the products obtained from the recipes of samples T_2/8/z (Fig. 1a–c) and T_9/1/z (Fig. 1g–i) with the replacement of NH₄F by NaF or NH₄Cl, respectively. However, they were nothing like their counterparts presented in Fig. 1. Therefore, both NH₄⁺ and F⁻ were indispensable for the growth of TTBs and HHMs. Single NH₄⁺ cation or F⁻ anion was not sufficient to cause the etching phenomenon or promote the formation of hollow microspheres. This might be rationalized by that the three additives have different solubility in water and NH₄⁺ and F⁻ preferentially bind onto specific facets of anatase TiO₂.^31,32,39,40 Actually, in the presence of NH₄F, a fluoride mediated Ostwald ripening process has been proposed for the formation of hollow microspheres of anatase TiO₂.⁴¹ To give further insight into the effect of NH₄F on the growth of TiO₂, we hydrothermally treated pre-synthesized dry powders of TiO₂ nanorods (T_2/8 and T_9/1) with different amounts of NH₄F (Fig. S5†). When small amounts of NH₄F, e.g., 2.0 or 6.0 mmol, were used, the resultant products were still anatase TiO₂ (Fig. S6†). Meanwhile, nanorod-like particles were changed to well-faceted TTBs (Fig. S5a, b, d and e†). However, when the amount of NH₄F was increased to 10.0 mmol, other species appeared and coexisted with anatase TiO₂ (Fig. S5c, f and S6†). These observations suggested that the TiO₂ nanorods (T_2/8 and T_9/1) underwent a dissolution, at least partially, and recrystallization process during our hydrothermal process in the presence of NH₄F. In the above, dry powders of T_2/8 and T_9/1 were already separated from their reaction solutions and hydrothermal treatment might cause the phase change. We also carried out experiments by adding different amounts of NH₄F solid directly into the hydrothermally treated reaction solutions of T_2/8 and T_9/1 followed by another cycle of hydrothermal treatment (Fig. S7†). Here, no phase change occurred as shown in the XRD patterns (Fig. S8†). In addition, compared with those products shown in Fig. 1, in which one-pot synthesis was deployed, the etching phenomenon and the morphology of the final products were dramatically altered (Fig. S7†). Thus, it is fairly safe to make a conclusion that NH₄F exerts its influence on the morphology of the final product in the stages prior to the generation of TiO₂.

3.2 Effects of reaction temperature and duration

The systematic tuning of R_HNF in our hydrothermal processes resulted in various TiO₂ products including small TTBs and large HHMs. For TTBs in T_2/8/z (z = 2–10) samples, to avoid the formation of rod-like particles at low amount of NH₄F or larger particles at high amount of NH₄F, we chose T_2/8/5 as active layers in future fabrication of bilayer photoanodes of DSSCs. The same strategy was applied for choosing T_9/1/9 as scattering layers to balance the mechanical stability and surface area. Accordingly, we investigated the effects of reaction temperature and duration on products prepared with R_HNF = 2/8/5 (Fig. S9†) and 9/1/9 (Fig. S10†). At R_HNF = 2/8/5, amorphous flurry spheres (Fig. S9m, n and i†) were obtained at a low temperature for a short hydrothermal duration. With the increase of reaction temperature or duration, such amorphous particles gradually vanished and evolved into highly crystalline and well-defined uniform TTBs of anatase TiO₂ (Fig. S9c, d, g and h†). At R_HNF = 9/1/9, the initial yolk-shell spheres (Fig. S10m–o†) gradually turned into completely hollow microspheres (Fig. S10c and d†) with prolonging reaction duration or increasing reaction temperature. Concurrently, the primary crystallites comprising their outer and inner shells evolved respectively from spindle-like nanoparticles with rough surfaces and small irregular nanoparticles (Fig. S10e, i, j, m and n†) to well-defined TTBs (Fig. S10c and d†).

3.3 Growth mechanisms

The formation of all the TiO₂ products (Fig. 1) in our hydrothermal processes could be interpreted using our previously proposed growth mechanisms.³¹ Generally, the conversion pathway from the sol to TiO₂ depends on the value of R_HN in the reaction system. At low R_HN, e.g., 2/8, a dehydration process from titanium hydroxide complex tended to occur. At high R_HN, e.g., 9/1, TiO₂ was formed from the decomposition of peroxotitanium anions in the sol. In between these two cases, e.g., R_HN = 5/5, the compositions of the sol species were quite complicated and both pathways might be involved, leading to multiply dispersed products.

In the previous work, TiO₂ nanorods prepared with R_HN = 2/8 had a length of about 240 nm.³¹ However, after NH₄F was introduced into the reaction system, extremely small (15–20 nm) TTBs were produced. Such a phenomenon might be due to the selective binding of NH₄⁺ (or NH₃ and OH⁻) to {100} facets and F⁻ to {001} facets of anatase TiO₂, which would suppress the growth along both [100] and [001] directions.^{31,32,40,42,43} This could be further confirmed by the results using NH₄Cl to replace NH₄F (Fig. S3†).

The preferential adsorption of F⁻ on {001} facets of anatase TiO₂ can also cause the gradual dissolution of the {001} facets via forming soluble hexafluorotitanium complex [TiF₆]²⁻, which is a well-known etching process of TiO₂.^35,36,44 With the increase of the amount of NH₄F, the etching degree became more significant and the particle size was gradually reduced, and eventually HHMs were formed. In Fig. S10† where R_HNF = 9/1/9, at the early stages of reaction, the products mainly took on the yolk-shell configurations and the primary crystallites constructing the internal shell and the yolk part had very small particle sizes. These small nanocrystallites had high surface energies and tended to dissolve into the solution through the highly porous crystalline shells. The gradual depletion of the yolk led to the formation of hollow microsphere, which is consistent with traditional Ostwald ripening process.^41,45–49 We noticed that the etching phenomenon only occurred for R_HN = 9/1. In this case, much less NH₃ was used and the etching effect of F⁻ would be substantially enhanced in such relatively acidic reaction media.^50–52 As a result, holes were observed at the vertices of the TiO₂ nanoparticles. While in the cases of R_HN = 5/5 or 2/8, more ammonia was added and higher amounts of OH⁻ were present in the reaction solution, which might suppress the surface fluorination. In this way, high energy {001} facets of anatase TiO₂ could be retained in the final products.

3.4 Photovoltaic performances of DSSCs

Conventionally, P25 are employed as the active layer to adsorb dye molecules, while 400 nm TiO₂ particles (CCIC) are introduced as the overlayer to scatter light in the bilayer photoanodes of DSSCs. In this work, TTBs of sample T_2/8/5 and HHMs of sample T_9/1/9 were selected as the active and scattering layer of the bilayer TiO₂ film photoanode, respectively.

3.4.1 Characterizations of typical samples of T_2/8/5 and T_9/1/9. Both samples of T_2/8/5 and T_9/1/9 were pure anatase TiO₂ (tetragonal, I4₁/amd, JCPDS 21-1272) (Fig. 2b and 3d). The sharp and strong XRD peaks reveal their high crystallinity. SEM (Fig. 2a) and TEM images (Fig. 2c and d) indicate that sample T_2/8/5 were monodispersed and well-defined TTBs about 15–20 nm in size. Sample T_9/1/9 were uniform spheres with an average diameter of about 1.5 μm as shown by the SEM image (Fig. 3a). Their hollow interiors were unambiguously revealed by the cracked spheres (Fig. 3a) and the contrast difference between the central part and the boundary in the TEM images (Fig. 4a and b). The white slits in the central part and the width of the boundary imply the shells of the hollow spheres were highly porous and had a thickness of 100 nm. High-magnification SEM (Fig. 3b and c) and TEM (Fig. 4b and c) images further indicate that the shells consisted of well-faceted TTBs. Their average long and short edge lengths were about 80 nm and 30 nm, respectively.

Fig. 2 SEM image (a), XRD patterns (b), low magnification TEM (c), HRTEM (d), enlarged HRTEM images (e) indicated by the red square in (d) and the corresponding fast-Fourier transform (FFT) patterns (f) of sample T_2/8/5.

Fig. 3 Low- (a) and high-magnification (b and c) SEM images and X-ray diffraction patterns (d) of the typical sample T_9/1/9.

Fig. 4 Low- (a), high-magnification (b and c) TEM and HRTEM ((d) and left inset) images and the corresponding fast-Fourier transform (FFT) patterns (right inset in (d)) of the typical sample T_9/1/9.

The enlarged HRTEM images of both samples T_2/8/5 (Fig. 2e) and T_9/1/9 (left inset in Fig. 4d) show two clear lattice fringes with spacings of 0.35 and 0.48 nm or 0.237 nm, which correspond to (101) and (002) or (004) atomic planes of anatase TiO₂, respectively. The angles labeled in the corresponding fast-Fourier transform (FFT) images of both samples (Fig. 2f and right inset in Fig. 4d) are 68.3°, identical to the theoretical value for the angle between {101} and {001} facets in an anatase crystal. Therefore, sample T_2/8/5 were TTBs enclosed by {101} and {001} facets and sample T_9/1/9 were HHMs constructed by similar TTBs but bigger in particle size. The Brunauer–Emmett–Teller (BET) surface areas, calculated from N₂ adsorption–desorption isotherm (Fig. S11†), were 52.55 and 37.28 m² g⁻¹ for sample T_2/8/5 and T_9/1/9, respectively.

3.4.2 Characterizations of the bilayer TiO₂ photoanode films. The bilayer TiO₂ photoanode T_2/8/5 + T_9/1/9 consisted of a compact layer with a thickness of about 10 μm and a hollow sphere overlayer about 15 μm in thickness as revealed from the cross-section SEM image (Fig. 5a). The films were uniform and almost crack-free (Fig. 5b and c). The well-connected TiO₂ particles and large voids formed among them imply fast electron transport and efficient electrolyte diffusion in DSSCs. For comparison, the other two bilayer TiO₂ films including P25 + T_9/1/9 and P25 + CCIC with similar layer thickness were also prepared by the same doctor-blade technique and used as photoanodes in DSSCs.

Fig. 5 Cross-section SEM images of T_2/8/5 + T_9/1/9 bilayer TiO₂ films (a) and the corresponding surface SEM images of scattering overlayer (b) and (c) active underlayer. Diffuse reflectance spectra (d) of bilayer TiO₂ photoanode films of T_2/8/5 + T_9/1/9, P25 + T_9/1/9 and P25 + CCIC.

Fig. 5d shows the diffuse reflectance spectra of the three bilayer TiO₂ photoanode films. Compared with the P25 + CCIC films, T_2/8/5 + T_9/1/9 and P25 + T_9/1/9 films exhibited higher diffuse reflectance and thus better light scattering ability. This could be attributed to the larger size and rougher surface of the hollow spheres of T_9/1/9 than those of CCIC. Between T_2/8/5 + T_9/1/9 and P25 + T_9/1/9 films, the former showed slightly poorer diffuse reflectance, which is consistent with the smaller size of T_2/8/5 as compared with P25. The dye loading amounts of these three bilayer TiO₂ films followed a decreasing order from T_2/8/5 + T_9/1/9 to P25 + T_9/1/9 and to P25 + CCIC (Table 1). Considering T_9/1/9 had a hierarchical structure that consisted of primary crystallites much smaller than CCIC, such order is fairly reasonable and could be expected from their specific surface areas.

Table 1 Photovoltaic characterization results of DSSCs based on bilayer TiO₂ photoanode films of T_2/8/5 + T_9/1/9, P25 + T_9/1/9 and P25 + CCIC

Photoanode	Jsc (mA cm^−2)	Voc (V)	PCE (%)	FF	Dye loading (nmol cm^−2)
T2/8/5 + T9/1/9	18.38	0.764	9.06	0.65	398.94
P25 + T9/1/9	15.23	0.750	7.42	0.65	381.21
P25 + CCIC	12.85	0.736	5.98	0.63	363.48

---

3.4.3 Photovoltaic activity of the DSSCs. After loading dye, the three bilayer photoanodes were fabricated into DSSCs together with N719 dye, Pt counter electrode and iodine based electrolyte. Fig. 6a shows the J–V curves of the three cells. All photovoltaic parameters are summarized in Table 1. The T_2/8/5 + T_9/1/9 cell delivered a short-circuit current density (J_SC) of 18.38 mA cm⁻² and open-circuit voltage (V_OC) of 0.764 V, which led to a power conversion efficiency (PCE) of 9.06%. Replacing T_2/8/5 with P25 or replacing both T_2/8/5 and T_9/1/9 with P25 and CCIC in the other two cells caused a progressive decrease of J_SC and V_OC to 15.23 mA cm⁻² and 0.750 V for P25 + T_9/1/9 cell and to 12.85 mA cm⁻² and 0.736 V for P25 + CCIC cell. Their corresponding PCEs also decreased to 7.42% and 5.98%. A 51.5% increment in PCE of T_2/8/5 + T_9/1/9 cell was realized compared with P25 + CCIC cell. These three cells showed similar fill factor (FF). All other parameters were consistent with the variation trend of the dye loading of their corresponding photoanode films. As demonstrated above, P25 + T_9/1/9 bilayer films exhibited both higher dye loading and light scattering than P25 + CCIC films, which could be ascribed to the larger surface area originated from hierarchical structure and bigger micrometer scale size of T_9/1/9 than those of CCIC. Hence the J_sc of P25 + T_9/1/9 cell was higher than that of P25 + CCIC cell. Since particle size of T_2/8/5 is smaller than P25, T_2/8/5 + T_9/1/9 bilayer films adsorbed more dye but showed slightly poorer light scattering than that of P25 + T_9/1/9. Even though, the T_2/8/5 + T_9/1/9 cell still had the highest J_sc among the three fabricated cells, implying that the amount of dye anchored in the TiO₂ films is more important than the light scattering effect in converting solar energy into electricity in this system.

Fig. 6 Photocurrent density–voltage (J–V) curves (a) and electrical impedance spectra (EIS) (b and c) of three DSSCs fabricated with bilayer TiO₂ photoanode films of T_2/8/5 + T_9/1/9, P25 + T_9/1/9 and P25 + CCIC, respectively. (b) Nyquist plots together with the fitted results (line) according to the equivalent circuit model shown in the inset; (c) Bode phase plots.

The kinetics of electron transport and recombination as well as electrolyte diffusion, which determine the charge collection efficiency of DSSCs, were investigated with electrical impedance spectra (EIS). Fig. 6b shows the corresponding Nyquist plots together with the fitted lines based on the equivalent circuit model in the inset.^53–56 For each cell, three semicircles extending from the total series resistance (R_S) could be observed. The semicircles in high, intermediate and low frequency regions, respectively, represent the charge transfer resistance at the Pt counter electrode (R₁), resistance related to the charge recombination at the interface of dyed-TiO₂/electrolyte (R₂) and the Nernst diffusion resistance of I₃⁻/I⁻ in the electrolyte (R₃).^11,21 The specific EIS parameters are listed in Table 2. Considering the three cells using similar FTO glass substrates, the values of the three R_s are close. Similar phenomenon could be found in R₁ for the Pt counter electrodes. The slight discrepancy originated mainly from the difference among the three Pt counter electrode as well as the fabrication of each cell.

Table 2 EIS parameters of DSSCs based on bilayer TiO₂ photoanode films of T_2/8/5 + T_9/1/9, P25 + T_9/1/9 and P25 + CCIC

Photoanode	Rs (Ω)	R1 (Ω)	R2 (Ω)	R3 (Ω)	fmax (Hz)	τr (ms)
T2/8/5 + T9/1/9	23.37	5.46	12.30	4.68	25.96	6.14
P25 + T9/1/9	21.17	7.08	11.54	3.92	35.90	4.44
P25 + CCIC	21.95	6.34	8.62	4.88	48.89	3.26

---

R₂ values for the T_2/8/5 + T_9/1/9, P25 + T_9/1/9 and P25 + CCIC cells were 12.30, 11.54 and 8.56 Ω in sequence (Table 2). In Bode phase plots, their characteristic middle-frequency peaks (f_max) were located at 25.96, 35.90 and 48.89 Hz (Fig. 6c), corresponding to electron lifetimes (τ_r) of 6.14, 4.44 and 3.26 ms calculated from τ_r = 1/2πf_max.^55,57,58 Larger recombination resistance and longer electron lifetime suggest less recombination and higher charge collection efficiency and eventually larger V_OC. A gradual decrease of V_OC from T_2/8/5 + T_9/1/9 to P25 + T_9/1/9 and to P25 + CCIC cells can be rationally derived from the EIS results. The lower V_OC of the latter two cells might be associated with the utilization of P25 nanoparticles, which had plenty defects or traps serving as charge recombination centers and thus reduced the charge collection efficiency. By contrast, T_9/1/9 and T_2/8/5 were of high crystallinity and there were fewer surface trapping sites presented in the corresponding bilayer films. Besides, their pure anatase phase could serve as an additional contributing factor for the large V_OC of the T_2/8/5 + T_9/1/9 cell as the electron mobility of anatase TiO₂ has been reported to be larger than its rutile counterpart or mixed phases such as P25.^10,14 The coexistence of oxidative {001} and reductive {101} facets in our highly crystalline TiO₂ could also improve charge collection by facilitating the spontaneous charge separation through preferential carrier flow towards specific facets.^29,30

From the values of R₃ in Table 2, it can be seen that T_2/8/5 + T_9/1/9 and P25 + T_9/1/9 cells had smaller Nernst diffusion resistance of I₃⁻/I⁻ in the electrolyte than P25 + CCIC cell. The porous surface of T_9/1/9 as well as the large voids among particles in the films allowed for efficient diffusion of I₃⁻/I⁻. Meanwhile, that the particles in T_2/8/5 + T_9/1/9 are in good contact also benefits electron transport (Fig. 5a). Therefore, an overall high charge collection efficiency could be achieved in the solar cells.

In all, by taking advantage of the small particle size of T_2/8/5 and big particle size of T_9/1/9 with hierarchical structure, the light harvesting efficiency of our fabricated bilayer photoanode of DSSCs could be improved through loading more dye in the underlayer and enhancing light scattering in the overlayer. The anatase crystalline phase and the coexistence of {001} and {101} facets of these two synthesized particles, along with channels provided in the bilayer films for electrolyte diffusion, endowed the cell with high charge collection efficiency. With these improvements in light harvesting and charge collection, a high PCE of 9.06% was achieved for the T_2/8/5 + T_9/1/9 cell, corresponding to a 51.5% increment over the P25 + CCIC cell fabricated in the same way. This was also superior to previously reported DSSCs based on highly crystalline nanooctahedra and agglutinated mesoporous microspheres of anatase TiO₂ and comparable to those fabricated with 20 nm nanocrystalline TiO₂ and porous micro- and submicrospheres.^19,21 The employment of our synthesized well-faceted TTBs and HHMs consisting of similar TTBs as bilayer photoanode materials renders the DSSCs efficient in dye loading, light scattering and charge collection and therefore competitive in converting light into electricity.

4. Conclusions

In conclusion, we have synthesized a series of nano- and microcrystals of anatase TiO₂ ranging from TTBs to HHMs via one facile sol-hydrothermal approach that is free from templates and surfactants. Their morphology and size could be finely tuned by adjusting the ratio of H₂O₂, NH₃ and NH₄F. Moreover, the growth mechanisms involving fluoride mediated etching and Ostwald ripening were proposed. Bilayer photoanodes were fabricated with typical TTBs (T_2/8/5) and HHMs (T_9/1/9) as the active and scattering layers, respectively. DSSCs equipped with such photoanodes, together with N719 dye, Pt counter electrode and iodine based electrolyte, exhibited a PCE of 9.06%, corresponding to a 51.5% increment compared with the P25 + CCIC cell fabricated in the same way. Meanwhile, the short-circuit current density reached up to 18.38 mA cm⁻², which is 1.46 times of the P25 + CCIC cell. Our T_2/8/5 + T_9/1/9 cell was superior to previously reported DSSCs based on highly crystalline nanooctahedra and agglutinated mesoporous microspheres of anatase TiO₂ and comparable to those fabricated with 20 nm nanocrystalline TiO₂ and porous micro- or submicrospheres. Such improved performance could be ascribed to the enhanced efficiency of both light harvesting and charge collection. The former might be originated from increased dye loading amount in the underlayer of T_2/8/5 with small particle size and boosted light scattering in the overlayer of T_9/1/9 with the micrometer size and hierarchical structure, while the latter was mainly derived from their exposed {001} and {101} facets, high crystalline anatase phase and channels in the bilayer film for efficient electrolyte diffusion.

Acknowledgements

The authors are grateful for the financial support from the National Natural Science Foundation of China (NSFC) through the NSFC (21571170 and 21501168) and Innovative Research Groups (21521092), and Tianjin Key Subject for Materials Science and Engineering.

Notes and references

1. B. O'Regan and M. Grätzel, Nature, 1991, 353, 737.
2. A. Yella, H.-W. Lee, H. N. Tsao, C. Yi, A. K. Chandiran, M. K. Nazeeruddin, E. W.-G. Diau, C.-Y. Yeh, S. M. Zakeeruddin and M. Grätzel, Science, 2011, 334, 629.
3. M. K. Nazeeruddin, A. Kay, I. Rodicio, R. Humpbry-Baker, E. Müller, P. Liska, N. Vlachopoulos and M. Grätzel, J. Am. Chem. Soc., 1993, 115, 6382.
4. Z.-S. Wang, H. Kawauchi, T. Kashima and H. Arakawa, Coord. Chem. Rev., 2004, 248, 1381.
5. H.-J. Koo, Y. J. Kim, Y. H. Lee, W. I. Lee, K. Kim and N.-G. Park, Adv. Mater., 2008, 20, 195.
6. W.-Q. Wu, Y.-F. Xu, H.-S. Rao, C.-Y. Su and D.-B. Kuang, J. Am. Chem. Soc., 2014, 136, 6437.
7. Z. Yao, M. Zhang, H. Wu, L. Yang, R. Li and P. Wang, J. Am. Chem. Soc., 2015, 137, 3799.
8. S. Mathew, A. Yella, P. Gao, R. Humphry-Baker, B. F. E. Curchod, N. Ashari-Astani, I. Tavernelli, U. Rothlisberger, M. K. Nazeeruddin and M. Grätzel, Nat. Chem., 2014, 6, 242.
9. S. Hore, P. Nitz, C. Vetter, C. Prahl, M. Niggemann and R. Kern, Chem. Commun., 2005, 2011.
10. D. Chen, F. Huang, Y.-B. Cheng and R. A. Caruso, Adv. Mater., 2009, 21, 2206.
11. Y. J. Kim, M. H. Lee, H. J. Kim, G. Lim, Y. S. Choi, N.-G. Park, K. Kim and W. I. Lee, Adv. Mater., 2009, 21, 3668.
12. S. H. Hwang, J. Yun and J. Jang, Adv. Funct. Mater., 2014, 24, 7619.
13. J.-Y. Liao, B.-X. Lei, D.-B. Kuang and C.-Y. Su, Energy Environ. Sci., 2011, 4, 4079.
14. H. Pan, J. Qian, Y. Cui, H. Xie and X. Zhou, J. Mater. Chem., 2012, 22, 6002.
15. X. Wu, G. Q. Lu and L. Wang, Energy Environ. Sci., 2011, 4, 3565.
16. F. Huang, D. Chen, X. L. Zhang, R. A. Caruso and Y.-B. Cheng, Adv. Funct. Mater., 2010, 20, 1301.
17. I. G. Yu, Y. J. Kim, H. J. Kim, C. Lee and W. I. Lee, J. Mater. Chem., 2011, 21, 532.
18. Y.-C. Park, Y.-J. Chang, B.-G. Kum, E.-H. Kong, J. Y. Son, Y. S. Kwon, T. Park and H. M. Jang, J. Mater. Chem., 2011, 21, 9582.
19. H. Zhang, Y. Han, X. Liu, P. Liu, H. Yu, S. Zhang, X. Yao and H. Zhao, Chem. Commun., 2010, 46, 8395.
20. Y. Qiu, W. Chen and S. Yang, Angew. Chem., Int. Ed., 2010, 49, 3675.
21. K. Yan, Y. Qiu, W. Chen, M. Zhang and S. Yang, Energy Environ. Sci., 2011, 4, 2168.
22. K. Mahmood, B. S. Swain and A. Amassian, Adv. Mater., 2015, 27, 2859.
23. J. Chen, H. B. Yang, J. Miao, H.-Y. Wang and B. Liu, J. Am. Chem. Soc., 2014, 136, 15310.
24. M. Adachi, Y. Murata, J. Takao, J. Jiu, M. Sakamoto and F. Wang, J. Am. Chem. Soc., 2004, 126, 14943.
25. K. Zhu, N. R. Neale, A. Miedaner and A. J. Frank, Nano Lett., 2007, 7, 69.
26. J. Zhang, S. Li, P. Yang, W. Que and W. Liu, Sci. China Mater., 2015, 58, 785.
27. E. Palomares, J. N. Clifford, S. A. Haque, T. Lutz and J. R. Durrant, J. Am. Chem. Soc., 2003, 125, 475.
28. N.-G. Park, J. van de Lagemaat and A. J. Frank, J. Phys. Chem. B, 2000, 104, 8989.
29. N. Roy, Y. Sohn and D. Pradhan, ACS Nano, 2013, 7, 2532.
30. J. G. Yu, J. X. Low, W. Xiao, P. Zhou and M. Jaroniec, J. Am. Chem. Soc., 2014, 136, 8839.
31. L. Zhang, L. Tian, Y. Liu, T. Tan, D. Liu and C. Wang, J. Mater. Chem. A, 2015, 3, 15265.
32. 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.
33. G. Liu, H. G. Yang, J. Pan, Y. Q. Yang, G. Q. Lu and H.-M. Cheng, Chem. Rev., 2014, 114, 9559.
34. D. Li, H. Haneda, S. Hishita and N. Ohashi, Chem. Mater., 2005, 17, 2588.
35. J. M. Macák, H. Tsuchiya and P. Schmuki, Angew. Chem., Int. Ed., 2005, 44, 2100.
36. D. Wang, B. Yu, C. Wang, F. Zhou and W. Liu, Adv. Mater., 2009, 21, 1964.
37. J. M. Macák, H. Tsuchiya, L. Taveira, S. Aldabergerova and P. Schmuki, Angew. Chem., Int. Ed., 2005, 44, 7463.
38. H. E. Prakasam, K. Shankar, M. Paulose, O. K. Varghese and C. A. Grimes, J. Phys. Chem. C, 2007, 111, 7235.
39. W. M. Haynes, CRC Handbook of Chemistry and Physics, CRC Press, Boca Raton, 2011.
40. T. Sugimoto, K. Okada and H. Itoh, J. Colloid Interface Sci., 1997, 193, 140.
41. J. Yu, H. Guo, S. A. Davis and S. Mann, Adv. Funct. Mater., 2006, 16, 2035.
42. A. S. Barnard and L. A. Curtiss, Nano Lett., 2005, 5, 1261.
43. J. M. Li and D. S. Xu, Chem. Commun., 2010, 46, 2301.
44. Q. Xiang, J. Yu and M. Jaroniec, Chem. Commun., 2011, 47, 4532.
45. H. G. Yang and H. C. Zeng, J. Phys. Chem. B, 2004, 108, 3492.
46. S. Liu, J. Yu and M. Jaroniec, J. Am. Chem. Soc., 2010, 132, 11914.
47. H. Li, Z. Bian, J. Zhu, D. Zhang, G. Li, Y. Huo, H. Li and Y. Lu, J. Am. Chem. Soc., 2007, 129, 8406.
48. Y. Wang, Q. Zhu and H. Zhang, Chem. Commun., 2005, 5231.
49. X. Yang, J. Fu, C. Jin, J. Chen, C. Liang, M. Wu and W. Zhou, J. Am. Chem. Soc., 2010, 132, 14279.
50. Y. Wang, H. Zhang, Y. Han, P. Liu, X. Yao and H. Zhao, Chem. Commun., 2011, 47, 2829.
51. J. H. Pan, X. Zhang, A. J. Du, D. D. Sun and J. O. Leckie, J. Am. Chem. Soc., 2008, 130, 11256.
52. H. Park and W. Choi, J. Phys. Chem. B, 2004, 108, 4086.
53. J. van de Lagemaat, N.-G. Park and A. J. Frank, J. Phys. Chem. B, 2000, 104, 2044.
54. T.-S. Kang, K.-H. Chun, J. S. Hong, S.-H. Moon and K.-J. Kim, J. Electrochem. Soc., 2000, 147, 3049.
55. Q. Wang, J. E. Moser and M. Grätzel, J. Phys. Chem. B, 2005, 109, 14945.
56. T. Hoshikawa, M. Yamada, R. Kikuchi and K. Eguchi, J. Electrochem. Soc., 2005, 152, E68.
57. R. Kern, R. Sastrawan, J. Ferber, R. Stangl and J. Luther, Electrochim. Acta, 2002, 47, 4213.
58. Y.-Z. Zheng, X. Tao, L.-X. Wang, H. Xu, Q. Hou, W.-L. Zhou and J.-F. Chen, Chem. Mater., 2010, 22, 928.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra14857c

---