Spray-deposited zinc titanate ﬁ lms obtained via sol – gel synthesis for application in dye-sensitized solar cells †

ac Foam-like zinc orthotitanate (Zn 2 TiO 4 ) is successfully synthesized via the wet chemical sol – gel route assisted with a structure-directing diblock copolymer template. The wet chemical route enables spray deposition of Zn 2 TiO 4 ﬁ lms. Calcination temperature of the spray-deposited ﬁ lms is shown to be crucial for the synthesis of the compound phase, Zn 2 TiO 4 . Surface composition and optical properties of the ﬁ lms are also studied. Finally, Zn 2 TiO 4 ﬁ lms are shown to o ﬀ er a reasonable functioning as an electron acceptor in dye-sensitized solar cells, with the best preliminary performance reported so far.


Introduction
Evolution of highly ordered inorganic metal oxide nanostructures has gained substantial momentum in recent years, owing to their extraordinary structural and electronic properties. The tremendous increase in the demand for nanoscale electronic devices 1,2 has motivated researchers to produce complex functional nanomaterials consisting of binary and/or ternary composites in addition to pure phases. In the present investigation, we focus on the synthesis of a ternary compound, namely, zinc orthotitanate in a solution-based sol-gel approach. As reported in literature, the simultaneous synthesis of TiO 2 -ZnO nanocomposites gives rise to three different compounds and some minor impurities. These three compounds are: zinc orthotitanate (Zn 2 TiO 4 ) with a cubic spinel structure, Zn 2 Ti 3 O 8 with a cubic defect spinel structure and zinc metatitanate (ZnTiO 3 ) with a rhombohedral ilmenite structure. The minor impurity phases consist of rutile TiO 2 or ZnO. Zn 2 Ti 3 O 8 is a metastable form of ZnTiO 3 and is known to occur at temperatures lower than 800 C, whereas, ZnTiO 3 is well-known to decompose to Zn 2 TiO 4 and rutile TiO 2 at temperatures above 945 C. 3 Hence, the production of a pure composite is a challenge resulting from the temperature requirements for the synthesis of a particular compound.
Among the three reported common compounds, Zn 2 TiO 4 is the center of interest in the present work due to its high potential for several applications. Zn 2 TiO 4 has been established as one of the most important regenerable photocatalysts, additionally showing wide-scale applications in the removal of sulfur during coal gasication, 3 photocatalytic splitting of water and degradation of organic compounds, as an active anode material in Li-ion batteries, microwave dielectrics, pigments and many more. 4 A major eld of application, where inorganic metal oxides have largely contributed, is in energy conversion and storage. In particular, dye-sensitized solar cells (DSSCs) is one of the most successful technologies for the conversion of solar energy into electricity. 5,6 Conventionally, TiO 2 has been used for DSSCs, 7 however, more recent studies have shown promising results for ZnO-based DSSCs as well. 8 In addition, several attempts have been made in order to combine these two materials together to harvest their synergetic structural and electronic properties. [8][9][10] These initiatives have inspired us to prove the competence of the titania (TiO 2 )-zinc oxide (ZnO) compound, Zn 2 TiO 4 in the eld of DSSCs. The present article is focused on the synthesis procedure to obtain a pure compound phase of Zn 2 TiO 4 via a wet chemical route and its characterization rather than device optimization. Nevertheless, to our knowledge, this is the rst contribution so far, reporting a reasonable device efficiency using Zn 2 TiO 4 .
The most common technique to synthesize Zn 2 TiO 4 is the direct solid-state route employing a high temperature of 1000 C. 3 Other methods producing pure and mixed phases of Zn 2 TiO 4 involve metal-organic chemical vapor deposition, 11 ball milling, 12,13 sol-electrospinning 14 and synthesis using inorganic metal oxide templates via sputtering. 15 Although a few morphologies of the compound are proclaimed, such as nanorods, twinned nanowires, 15 bres 14,16 and nanocrystalline powders, 17 a thorough investigation about the possible nanomorphologies is still lacking, which is crucial for nal device applications. Moreover, all the above-mentioned synthesis routes suffer the common disadvantage of nontunability of length scale of the nanostructures obtained.
In the present work, we use a diblock copolymer templateassisted sol-gel technique to produce Zn 2 TiO 4 nanostructures. This technique has been already well-established for structuring inorganic metal oxides such as TiO 2 (ref. 18) and ZnO. 19 The major advantages provided by this approach involve the ability to obtain multiple morphologies using the same copolymer template and the exibility to tune the length scales of the nanostructures at the same time for a specic need. Moreover, being a solution-based procedure, sol-gel synthesis allows for different deposition methods that can be applied in order to obtain thin lms. This is particularly benecial for fabrication of solar cells, where large-scale techniques such as printing, slot-die coating or spraying can be employed for solutions. Furthermore, different methods of deposition of the solution subject the lms to different drying times, which in turn provide additional tuneability to the nal length scales evolved in the lms. Fig. 1 shows a schematic representation of the multi-step process involved in producing Zn 2 TiO 4 nanostructures via a template-assisted sol-gel route. The preparation routine begins with two sols obtained individually for ZnO and TiO 2 . This is represented in Fig. 1(a) and (b), respectively. For both cases, a required amount of the amphiphilic diblock copolymer, poly(styrene-block-ethylene oxide), P(S-b-EO) is dissolved in a solvent pair, one with good and the other with weak interactions with the blocks, in order to induce micro-phase separation leading to formation of micelles. The micelles in the solution are represented with red hydrophobic (PS) core and blue hydrophilic (PEO) corona in the insets of Fig. 1(a) and (b). The respective metal oxide precursors are then added to the individual solutions (depicted as green spheres for the ZnO sol and gray spheres for TiO 2 sol), which selectively get incorporated in the hydrophilic part of the template and thereby react in a limited volume to initiate the formation of nanostructures in the system. In this fashion, the morphologies to be produced in the end are controlled and tuned as a function of the weight fraction of the solvents and the precursors added to the system. Finally, the two sols are mixed with a specic volume ratio to obtain a nal sol as indicated in Fig. 1(c), which controls the morphology in the nal system. To realize practical applications, the nanostructures are deposited in the form of solid lms on the substrate by the spraying technique as shown in Fig. 1(d). Spray deposition is an industry-oriented approach towards preparation of lms with an easy upscaling of the lm thicknesses and nal production volume. 20 Combinations of different volumes of the ZnO and the TiO 2 sols are sprayed in this study and are noted in Section S1 of the ESI † along with details about the spray protocol. In the course of this synthesis routine, it has been shown that the mixing ratio plays a crucial role to obtain the pure Zn 2 TiO 4 phase which can only be synthesized in a narrow mixing regime for ZnO and TiO 2 precursor concentrations (see Section S1 and S2 †). Aer the spray deposition of the lms, a nal calcination step is required to impart crystallinity to the system. These steps are sketched in Fig. 1(e) and (f) (the SEM images of the calcined lms produced from sols containing varied molar concentrations of the ZnO and the TiO 2 precursors are given in Section S1 †). Amongst the combinations of different volumes of ZnO and TiO 2 sols, only the nal sol containing molar ratio of ZnO precursor : TiO 2 precursor of 1.05 : 1, gives the pure Zn 2 TiO 4 phase as characterized in the following.

Materials and sol preparation
The amphiphilic diblock copolymer, poly(styrene-block-ethylene oxide), abbreviated as P(S-b-EO) [molar masses -23 kg mol À1 for PS block; 7 kg mol À1 for PEO block; polydispersity index -1.07] was used as received from Polymer Source Inc., Canada. Zinc acetate dihydrate [Zn(CH 3 COO) 2 $2H 2 O] 99.999% trace metals basis, with a density of 1.84 g mL À1 was purchased from Sigma Aldrich. Ethylene glycol-modied titanate [Ti(O 2 C 2 H 4 ) 2 ], with a density of 0.343 g mL À1 was synthesized using a procedure described in literature. 21  was chosen as the good solvent, hence dissolving both the blocks of the diblock copolymer template. Water (density ¼ 0.997 g mL À1 ) on the other hand only selectively dissolved the PEO block (for ZnO solw C 3 H 7 NO : w H 2 O : w Zn(CH 3 COO) 2 $2H 2 O ¼ 0.92 : 0.005 : 0.075). For the TiO 2 sol, C 3 H 7 NO was again used as the good solvent, whereas, hydrochloric acid (37% HCl), with a density of 1.2 g mL À1 served as the selective solvent (for TiO 2 sol w C 3 H 7 NO : w HCl : w Ti(O 2 C 2 H 4 ) 2 ¼ 0.905 : 0.08125 : 0.01375).
Detailed synthesis of the zinc orthotitanate (Zn 2 TiO 4 ) sol To synthesize the principle compound, Zn 2 TiO 4 , ZnO and TiO 2 sols were rst prepared separately. For the preparation of 10 mL of the nal (Zn 2 TiO 4 ) sol, 2 mL of ZnO sol was mixed with 8 mL TiO 2 sol.
In order to prepare the ZnO sol, rstly, 30 mg of P(S-b-EO) was dissolved in 1 mL of C 3 H 7 NO by stirring at room temperature for 30 minutes. Secondly, 155 mg of Zn(CH 3 COO) 2 $2H 2 O was dissolved in 1 mL of C 3 H 7 NO in a separate glass vial also by stirring at room temperature for 30 minutes. Aerwards the polymer and the precursor solutions were ltered using Teon lters with pore size of 0.45 mm into two separate glass vials, respectively. Next, 10.4 mL of deionized water was added to the polymer solution which was then again allowed to stir for 30 minutes at room temperature. Finally, the polymer solution (polymer + C 3 H 7 NO + deionized water) and the precursor solution (Zn(CH 3 COO) 2 $2H 2 O + C 3 H 7 NO) were mixed together with the aid of a syringe pump with a controlled mixing rate of the two solutions, as described in Section S1.1 of the ESI, † in order to obtain the nal ZnO sol.
For preparation of the TiO 2 sol, 120 mg of P(S-b-EO) was dissolved in 8 mL of C 3 H 7 NO by stirring the solution at room temperature for 30 minutes. Once a clear polymer solution was obtained, it was ltered into another glass vial using a Teon lter (pore size -0.45 mm). To this solution, 568.6 mL of HCl was added drop-wise followed by 115.45 mg of Ti(O 2 C 2 H 4 ) 2 . The TiO 2 precursor (Ti(O 2 C 2 H 4 ) 2 ) was added within 30 s aer the addition of HCl to the polymer solution. This solution was then allowed to stir for 30 minutes at room temperature at the end of which a pale yellow turbid solution was obtained. Aer 30 minutes, the solution was heated to 90 C and was stirred at this high temperature for 15 minutes until a pale yellow clear solution was obtained. This was then the nal TiO 2 sol, which stayed clear also at room temperature.
2 mL of the ZnO sol was then added to 8 mL of the TiO 2 sol in a drop-wise manner (without a syringe pump) at room temperature, within 1 minute time. This solution was referred to as the nal Zn 2 TiO 4 sol.

Thin lm and device preparation
The nal zinc orthotitanate sol was then spray-deposited on a substrate at 80 C for lm preparation. The spray deposition as well as the post-treatment parameters are given in Section S1.2 of the ESI. † Aer the spray deposition, the lms were sintered in a tube furnace, RETTH 230/3 provided by GERO Hochtemperaturofen GmbH at a heating ramp of 150 C h À1 to 600 C for 30 minutes.
Only aer this high temperature treatment at 600 C of the spray-deposited lm prepared using the nal sol, the zinc orthotitanate (Zn 2 TiO 4 ) compound was formed. The calcined lms were then used for the fabrication of the nal dye-sensitized solar cells (DSSCs). Details about the solar cell assembly and the materials involved are given in the ESI (Section S3 †).

Characterization
The scanning electron microscopy (SEM) measurements on the samples were performed using a Zeiss Gemini NVision 40 apparatus. A constant accelerating voltage of 5 kV, a working distance of 3.5 mm and an aperture of 10 mm were maintained for all the measurements.
X-ray diffraction measurements were performed at a Bruker D8 ADVANCE powder diffractometer. The data obtained were background subtracted using the program included in the DIFFRAC.SUITE provided by Bruker. A 2Q range from 25-60 was probed using a copper anode X-ray source with wavelength l ¼ 1.54Å operated at 40 kV and 40 mA. The theoretical peak positions for zinc oxide and zinc orthotitanate along with their relative intensities were obtained from Inorganic Crystal Structure Database (ICSD).
The UV/vis measurements were carried out in a Lambda 650S spectroscope, which provided UV (from deuterium lamp) and visible (from halogen lamp) wavelengths covering a range from 190 to 900 nm. The lms were prepared on glass and were measured in transmission geometry. The 150 mm integrating sphere was used to register the scattered light from the sample before the signal was directed to the detector.
The current of the DSSC was measured for an applied voltage with a Keithly 2400 source meter in dark and under illumination. A solar simulator Solar Constant by K. H. Steuernagel Lichttechnik GmbH was used. The intensity was calibrated and set to 1000 W m À2 using a silicon-based calibration solar cell (WPVS Reference Solar Cell Typ RS-ID-3 by Fraunhofer ISE).  obtained at the above-mentioned molar ratio of the ZnO and TiO 2 precursors in the sol. Homogeneous foam-like morphology with inter-connected network is clearly observed. It has been successfully shown by Perlich et al. and Sacco et al. that spongelike network morphologies for TiO 2 and ZnO are highly benecial for increased surface area, leading to high dye uptake and enhanced charge transport in photovoltaic devices. 23,24 In the present work, a foam-like Zn 2 TiO 4 nanostructured lm is successfully obtained via a wet chemical synthesis route based on a template-assisted sol-gel approach. The porosity of this foamlike Zn 2 TiO 4 lm is extracted from the scattering length density prole of the lm in the vertical direction (see Section S2.3 of the ESI †). A promising porosity of (52.5 AE 4.2)% is obtained for the foam-like lm. The porosity of Zn 2 TiO 4 lm is slightly lower than that reported for pure ZnO foam-like lms obtained from the same diblock copolymer template. 25 This can be explained by the presence of two metal oxide precursors and the nal composition of the sols used to prepare the lm. The formation of the homogeneous foam-like morphology is also supported by the similar molar concentrations of Zn(CH 3 COO) 2 $2H 2 O and Ti(O 2 C 2 H 4 ) 2 in the respective sols. This leads to uniform mixing of the constituents resulting in the formation of a single compound phase. For dissimilar molar concentrations of Zn(CH 3 COO) 2 $2H 2 O and Ti(O 2 C 2 H 4 ) 2 , nal morphologies with large aggregates are produced and the formation of the desired foam-like morphology is hampered (see Section S1.3 of the ESI †). Hence, the importance of the molar concentrations of the metal oxide precursors in their individual sols is crucial in order to obtain a single phase with uniform lm morphology, when the sols are mixed. This promising morphology, obtained aer the calcination of the sprayed lm, is further characterized in order to determine its crystallinity and optical properties. The nanostructured lm is nally manifested as the active layer in a functional DSSC as described in the following.

Crystallinity
Together with the composition of the sol, the impact of the calcination temperature on the formation of pure compound is also investigated. The spray-deposited lm obtained from the sol containing molar concentration of Zn(CH 3 COO) 2 $2H 2 : Ti(O 2 C 2 H 4 ) 2 ¼ 1.05 : 1 is calcined at 400 and 600 C. The X-ray diffraction (XRD) spectra of these lms are plotted in Fig. 3. The XRD spectra are measured over a 2Q range of 25-60 . The peaks obtained show a clear signature of the presence of a pure Zn 2 TiO 4 phase 3 when compared to the theoretical peak positions depicted by the green vertical lines, for the lm calcined at 600 C (green curve).
The crystallite size for the lm is calculated using the Scherrer's equation: 26 where l denotes the wavelength of the X-rays, B the full width at half maximum in radian and Q the Bragg angle. The crystallite size calculated for the lm calcined at 600 C, from the most intense (311) reection is approximately 11 nm.
On the other hand, for the diffraction spectrum of the lm calcined at 400 C (black curve in Fig. 3), there is a clear absence of any signal corresponding to the Zn 2 TiO 4 phase. The peaks however, match with the theoretical positions for the wurtzite phase of ZnO (indicated by the red vertical lines in Fig. 3). Therefore, it can be concluded that a high temperature of at least 600 C is essential for the formation of Zn 2 TiO 4 phase, in addition to the correct composition of the sol. This nding is in good agreement with the formation behavior of Zn 2 TiO 4 phase, as established by Dulin and Rase via the melt-mixing method. 27

Film composition
Energy dispersive X-ray spectroscopy (EDX) is performed on the samples synthesized from the sol with molar concentration of Zn(CH 3 COO) 2 $2H 2 : Ti(O 2 C 2 H 4 ) 2 ¼ 1.05 : 1. The spray-deposited lms are again calcined at 400 and 600 C. The EDX spectra of the lms are shown in Fig. 4(a) and (b).
It is observed from the spectra that characteristic peaks of Ti, Zn and O are available for both samples. In addition, a strong Si peak contributed by the substrate is also present. This is a substantial evidence of the purity of the sample because other elements could not be detected. The composition of both samples obtained in terms of atomic percent is listed in Table 1. It is clearly stated from the results that a calcination temperature of 600 C is essential in order to produce Zn 2 TiO 4 as for this particular sample, the atomic % of Zn is nearly twice of that of Ti, which matches with the stoichiometric composition of the compound. The XRD patterns are shifted along the intensity axis for clarity. The high intensity peak observed at 2Q ¼ 33 in both graphs corresponds to the silicon (100) peak originating from the substrate underneath. In order to concentrate more on the peaks obtained from the films, this high intensity peak is cut off at higher intensities. The Miller indices are indicated for the zinc orthotitanate peaks, corresponding to the green curve.

Optical properties
The optical properties of the Zn 2 TiO 4 calcined (at 600 C) lm are tested by UV/vis spectroscopy. The absorbance data plotted in Fig. 5(a) shows that Zn 2 TiO 4 also absorbs in the UV region like TiO 2 and ZnO. In the literature, different band gaps for this compound have been recorded depending on the synthesis route, composition and theoretical techniques such as density functional theory used for the calculation of the same. 17,[28][29][30] In order to evaluate the band gap of the sample, Tauc's equation for direct band gap semiconductors 31 is used as shown in eqn (2): where a is the absorption coefficient, E the incident photon energy, A is a constant and E g the band gap energy. The corresponding Tauc plot for Zn 2 TiO 4 foam-like lm is shown in Fig. 5(b) giving a band gap energy, E g of 4.01 eV which is within the estimated range as listed in literature for crystalline Zn 2 TiO 4 phase. 32,33 Solar cell performance Using the optimized foam-like lm morphology produced by spray deposition technique, preliminary tests have been performed with two different thicknesses of 4 and 10 mm of the active layer for the DSSC (details about the device assembly are given in Section S3 †). The fabricated DSSC is measured under AM1.5 solar spectrum conditions providing a light intensity set to 1000 W m À2 . The typical current density-voltage plots for the solar cells are shown in Fig. 6. The results from current-voltage characterization of active layers having different thicknesses are listed in Table 2. The increase in the short-circuit current density and open-circuit voltage with increasing lm thickness of the active layer is clearly visible from the J-V curves. This can be related to extra charge carrier pathways generated in the system with increasing amount of material. On the other hand, the increase in the ll factor is really minor. The power conversion efficiency of 1.5% is however very promising as a preliminary result and is the best announced so far with Zn 2 TiO 4 lms synthesized via sol-gel technique. As a comparison, the device efficiency is comparable to that obtained for DSSCs fabricated from ZnO nanowires as the electron-     conducting layer. 34 Although optimizing the device efficiency was not the main focus in the present study, we believe, with certain modications in the device fabrication protocol, such as incorporation of a blocking layer and a scattering layer 5 will largely enhance the overall device performance and will be addressed in future work.

Conclusions
In summary, a suitable solution-based approach is shown to successfully synthesize pure Zn 2 TiO 4 phase. In addition, the so called sol-gel route is used in combination with a structuredirecting diblock copolymer which allows for modifying and tuning the morphology and length scale of the nanostructures produced. The key to obtain a pure composite phase is therefore, individual synthesis of TiO 2 and ZnO nanostructures followed by mixing them in a specic volume ratio rather than simultaneous synthesis of the Zn 2 TiO 4 nanocomposite. Based on spray deposition method, homogeneous Zn 2 TiO 4 lms have been synthesized for applications in DSSCs. The requirement for high surface area of the active layer for intensied dye adsorption for applications in DSSCs 35 is met by synthesizing the sponge-like network morphology of the lm. Spray deposition is also shown to allow for an upscaling in the lm thickness and thereby improving the nal device performance. To the best of our knowledge, Zn 2 TiO 4 DSSCs are reported for the rst time in this article showing reasonable preliminary device performances. Hence, the present research sets a new landmark in the area of inorganic metal oxides and their compounds which have been increasingly sought for to develop new functional nanoscale devices. We postulate improvement in power conversion efficiency of Zn 2 TiO 4 based DSSCs in near future with more optimized device fabrication procedure. Hence, the development of Zn 2 TiO 4 as a new functional material parallel to conventional inorganic metal oxides in large-scale electrooptical applications is foreseen.