Preparation of smooth surface TiO 2 photoanode for high energy conversion efficiency in dye-sensitized solar cells

Dye-sensitized solar cells (DSSCs) based on a TiO2 photoanode have been considered as an alternative source in the field of renewable energy resources. In DSSCs, photoanode plays a key role to achieve excellent photo-to-electric conversion efficiency. The surface morphology, surface area, TiO2 crystal phase, and the dispersion of TiO2 nanoparticles are the most important factors influencing the properties of a photoanode. The smooth TiO2 surface morphology of the photoanode indicates closely packed arrangement of TiO2 particles which enhance the light harvesting efficiency of the cell. In this paper, a smooth TiO2 photoanode has been successfully prepared using a well-dispersed anatase TiO2 nanosol via a simple hydrothermal process. The above TiO2 photoanode was then compared with the photoanode made from commercial TiO2 nanoparticle pastes. The morphological and structural analyses of both the aforementioned photoanodes were comprehensively characterized by scanning electron microscopy and X-ray diffraction analysis. The DSSC fabricated by using a-TiO2 nanosol-based photoelectrode exhibited an overall light conversion efficiency of 7.20% and a short-circuit current density of 13.34mA cm-2, which was significantly higher than those of the DSSCs with the TiO2 nanoparticles-based electrodes.


Introduction
Much interest has been focused on the development of dye-sensitized solar cell (DSSC) technology, due to its low cost and easy fabrication with excellent photo-to-electric conversion efficiency [1][2][3][4][5].Recently, the maximum phototo-electric conversion efficiency of DSSCs was reported to exceed 13% at 509 W/m 2 simulated solar intensity [5].Even though DSSCs have their unique advantages, further improving the efficiency is still a key challenge [6][7][8].The DSSCs were composed of a photoanode, a sensitizer, a redoxcoupled electrolyte, and a counter electrode [9].Among them, the photoanode is the main component in DSSC and it significantly influences the photo-to-electric conversion efficiency of the cell, due to its dye loading, electron transportation, and electron collection characteristics [10,11].The photogenerated electron transport and collection are slow in the nanoparticles-based photoanode due to the recombination of electrons [7,12].Hence, a well performing photoanode with desirable properties, such as high surface area, smooth surface morphology with less grain boundaries for fast electron transport, is essential for the development of a high performance solar cell [13][14][15].
The nanocrystalline TiO 2 film is one of the most commonly employed photoanode materials in DSSCs due to its excellent optoelectronic properties [16].TiO 2 nanomaterials mainly exist as anatase, rutile, and brookite crystalline phase and anatase TiO 2 phase has been mostly utilized in DSSCs application [17].Even though the rutile TiO 2 is thermodynamically stable, it possesses a smaller surface area, larger crystallite size, and lower Fermi energy level, compared to anatase TiO 2 [18,19].The surface morphology, particle size, surface area, porosity, crystalline phase, and dispersion of TiO 2 nanoparticles are the various influencing factors which determine the performance of a photoanode [20][21][22].For instance, Park et al. investigated the effect of both base-treated and acid-treated TiO 2 to analyze the effect of particle size, shape, film porosity, and surface structure on the performance of DSSCs [23].For a typical photoanode, TiO 2 nanoparticle film with a thickness of 12-14 m was utilized for highly efficient DSSCs [9].So far, TiO 2 nanoparticle films were fabricated using the pastes prepared from TiO 2 nanopowders.Ito et al. have studied the homogenization effect of a different TiO 2 pastes by dispersing TiO 2 powders via both ball-milling and mortar-grinding route, in which the monodispersed particles were obtained by the former process [24].Lee et al. studied the influence of the surface morphology of the TiO 2 film on the performance of DSSC, in which the TiO 2 paste was prepared using a poly(ethylene glycol) binder [25].Dhungel and Park fabricated DSSCs using different TiO 2 pastes by varying the proportion of nanocrystalline particles of TiO 2 with particle size distribution in a wide range [26].The TiO 2 film morphology, interparticle interaction, and the connection between the TiO 2 film and conductive substrate are essential for the better dye adsorption and transportation of electrons to the counter electrode with reduced charge recombination.Hence, it is very important to prepare the TiO 2 film with highly dispersed TiO 2 nanoparticles.The preparation of unagglomerated TiO 2 paste from the homogeneous TiO 2 suspension is one of the approaches to obtain a desired TiO 2 film.Jeong et al. used a TiO 2 colloidal suspension to prepare an -terpineolbased TiO 2 paste and compared with the commercial dyesol TiO 2 paste [27].Based on the above considerations, in order to obtain an effective TiO 2 photoanode, the preparation of different types of TiO 2 pastes using the TiO 2 nanosol and TiO 2 nanopowders was carried out.
In the present investigation, the TiO 2 nanosol was prepared by hydrothermal process using titanium (Ti) precursor and its effect on the performance of DSSCs was evaluated.The TiO 2 nanosol-based pastes were compared with the TiO 2 nanopowders-based pastes.The photovoltaic performance indicated that the anatase TiO 2 nanosol-based photoelectrode exhibited higher photocurrent density and higher efficiency than that of the photoelectrode using TiO 2 nanoparticles.This result may be attributed to the large amount of dye adsorption on the TiO 2 nanosol which has higher surface area with highly dispersed uniform a-TiO 2 particles.

Preparation of TiO 2
Nanosol.Titanium (IV) n-butoxide was added slowly into the acetic acid (2 M) solution.The solution was stirred for 4-5 days until obtaining a translucent white solution.Then, the mixture was hydrothermally treated at 200 ∘ C for 5 h.The obtained white precipitate solution was cooled to room temperature, centrifuged, and washed once with distilled (DI) water and twice with ethanol.The centrifuged white precipitate was dispersed into 40 mL of anhydrous ethanol and kept stirring for one day.Subsequently, the centrifuged white precipitate was dispersed into 40 mL of anhydrous ethanol and kept stirring for one day.This sample is hereafter named as anatase TiO 2 (a-TiO 2 ) nanosol (NS) because of its anatase crystal phase.

Preparation of a-TiO 2 Nanosol-Based Paste.
To prepare a-TiO 2 nanosol-based paste, the obtained precipitate after hydrothermal treatment was centrifuged/washed and dispersed into anhydrous ethanol under stirring for 24 h and sonicated further for proper homogeneous dispersion.Then, 25.96 g of -terpineol and the mixture solution of two viscosities of ethyl cellulose (10 cp-1.8 g and 45 cp-1.4 g) in anhydrous ethanol (29 g) were added into the above solution, followed by repeated sonication for three times.The above solution was evaporated using the rotary evaporator at 40 ∘ C until to obtain a viscous paste.Finally, the paste was rolled using three-roller miller.

Preparation of TiO 2
Nanoparticles-Based Paste.Little modifications were carried out for the preparation of TiO 2 paste using TiO 2 nanoparticles, namely, P-25, ST-01, and ST-21.Briefly, 6 gram of TiO 2 nanoparticles was dispersed into the mixture of DI water, acetic acid, and anhydrous ethanol.Then the mixture was kept stirring for 24 h and sonicated for 5 min.Furthermore, the solution was sonicated 5 min followed by the addition of -terpineol and again sonicated for 5 min after adding ethyl cellulose.The remaining procedure is the same as mentioned above in the preparation of TiO 2 nanosol paste.

Preparation of TiO 2
Photoanode and Solar Cell Assembly.Fluorine doped tin oxide (FTO, Solaronix, 8 Ωcm −2 ) conducting glass was cleaned in detergent liquid, distilled water, acetone and methanol using an ultrasonic bath each for 30 min.Then the TiO 2 paste was screen printed (Screen Printing Machine, Weger, WE-400F, Guger Industries Co., Ltd.) on cleaned FTO substrate and dried at 110 ∘ C. The process was repeated until to achieve 6 m thick TiO 2 layer.The TiO 2 films were annealed with a programmable heating process: at 110 ∘ C for 30 min, 125 ∘ C for 15 min, 325 ∘ C for 5 min, 375 ∘ C for 5 min, 450 ∘ C for 15 min, and 500 ∘ C for 15 min.The working area of the TiO 2 films was 0.16 cm 2 .The TiO 2 electrodes were cooled to 80 ∘ C and then immersed in 0.3 mM N719 dye in acetonitrile/tert-butyl alcohol (1 : 1) for 36 h.The sensitized photoanodes and the platinum (Pt, 20 nm, ion beam sputtering E-105) sputtered FTO substrates were sandwiched together using 60 m surlyn spacers.The ionic electrolyte consisting lithium iodide, iodide, and 4-tertbutylpyridine in acetonitrile was introduced into the hole predrilled on the counter electrode.Finally the hole was sealed using 30 m surlyn and cover glass.

Characterization of Samples and Device.
The particle size and the morphology of the TiO 2 nanoparticles were analyzed by transmission electron microscopy (TEM, Hitachi, H-7100).The surface morphology of TiO 2 films was studied by scanning electron microscopy (SEM, Hitachi, S-4800).X-ray diffraction (XRD, Rigaku PANalytical X'Pert PRO) measurement was carried out with Cu K radiation.The surface area analysis of the TiO 2 nanoparticles was obtained using Brunauer-Emmett-Teller (BET, Micromeritics, Gemini V, ASAP-2010) surface area analyzer, in which all the samples were degassed prior to analyzes.The reflectance UV-visible spectrum was obtained using a Shimadzu UV-3600 spectrometer with an integrating sphere.The amount of adsorbed dye was determined by desorbing the dye from a-TiO 2 surface into a solution of 0.1 M NaOH.The concentration of the adsorbed dye was analyzed by UV-visible spectrophotometer (JASCO, V-630).The current-voltage (I-V) characteristics of the cell under one sun irradiation (AM 1.5 filter-Oriel, #81094) using solar simulator with 300 W Xenon lamp (Oriel #91160) were obtained by applying external bias to the cell, and the generated photocurrent was measured by a Keithley model 2400 digital source meter.The incident photon-tocurrent efficiency (IPCE) was obtained by Model SR830 DSP Lock-In Amplifier and a Model SR540 Optical Chopper (Stanford Research Corporation, USA), a 150 W Xenon lamp and power supply (Oriel, #66902), and a monochromator (Oriel CornerstoneTM 130).It can be further confirmed by the higher resolution SEM image (Figure 1(e), ×50,000).Comparatively, all the nanoparticle-based TiO 2 films show rough surface morphology which is evident in the higher magnification images (Figures 1(f)-1(h)).For P-25 film (Figures 1(b) and 1(f)), some voids between the closely packed TiO 2 nanoparticles were observed on the porous TiO 2 layers.The ST-21 film (Figures 1(d) and 1(h)) surface is rough with TiO 2 aggregates of about 300 nm.In the case of ST-01 film, a completely clumpy and nonuniform surface morphology was observed.The clumpy structure is formed by aggregation of TiO 2 particles of size from submicron to several microns.The observation of individual TiO 2 nanoparticles is hardly distinguishable.

Results and Discussions
This result is consistent with the surface-profiling measurement (results not shown here) in which the screen printed TiO 2 film made from nanosol-based TiO 2 paste is relatively flat compared with that made from nanoparticlebased TiO 2 pastes.For comparison, the thickness of TiO 2 films from nanosol-based TiO 2 paste and nanoparticle-based TiO 2 pastes were all controlled at about 6 m.
It is concluded from the aforementioned SEM analysis that the nanosol-based TiO 2 electrode reveals smoother surface morphology characteristics in comparison with the nanoparticle-based TiO 2 electrodes.This can be insight by investigation of the size and surface morphology of TiO 2 nanoparticles.Figure 2 shows the TEM micrographs of (a) TiO 2 nanosol, (b) P-25, (c) ST-01, and (d) ST-21 samples.The TiO 2 particles in TiO 2 nanosol are well-dispersed irregular polyhedron which is typical for anatase phase TiO 2 .Similar TEM morphologies are observed for commercial TiO 2 nanoparticles except for ST-01, where severe agglomeration occurs presumably due to the extremely small TiO 2 particle size.The average particle size of P-25, ST-01, and ST-21 samples is ∼35 nm, 5 nm, and 27 nm, respectively.The size of TiO 2 nanoparticles (TiO 2 -NP, TEM not shown) worked out (just drying) from TiO 2 nanosol is ∼20 nm.The particle size of TiO 2 can also be estimated from BET surface area measurement.The TiO 2 -NP, P-25, ST-01, and ST-21 nanoparticles possess specific surface area of 108.5, 51.6, 247.9, and 65.1 m 2 /g, respectively, which is consistent with the TEM size measurement.The results of TiO 2 particle size calculated from TEM and BET are also summarized in Table 1 and compared with the crystalline size obtained from XRD.
The TEM particle size analysis was also performed for TiO 2 samples after sintering at 500 ∘ C in order to imitate the annealing process after screen-printing the TiO 2 pastes onto FTO substrate.The particle size is marginally enlarged.The average particle size of TiO 2 -NP, P-25, ST-01, and ST-21 samples after sintering is ∼35 nm, 42 nm, 27 nm, and 39 nm, respectively (Table 1).
Consistent with TEM analysis, the surface area from BET measurement decreased to 82.3, 51.3, 99.5, and 58.6 m 2 /g.The ST-01 TiO 2 is more dispersed.It is expected that the growth of TiO 2 nanoparticles upon high temperature heat treatment may lead to the clumpy surface morphology of TiO 2 electrodes as seen from SEM images (Figure 1) for P-25 film, ST-01 film, and ST-21 film.As for the NS-film, the as-synthesized TiO 2 nanoparticles in a-TiO 2 nanosol were evenly well dispersed with unique morphological configuration on FTO.Although the heat-induced aggregation may occur to some extent after annealing, it does not significantly affect the surface morphology.Both TEM characterization and BET analysis reveal that the utilization of TiO 2 nanosol is more desirable than the TiO 2 nanoparticles for the fabrication of TiO 2 films.The crystal phase of TiO 2 films was determined by XRD. Figure 3 illustrates the XRD patterns of (a) NS films, (b) P-25 film, (c) ST-01 film, and (d) ST-21 film deposited on FTO substrate.The XRD patterns mainly indexed to anatase phase (A: anatase) with diffraction angle of 2 at 25.23 ∘ , 37.67 ∘ , 48.05 ∘ , 53.90 ∘ , 55.03 ∘ , and 62.68 ∘ corresponding to the characteristic anatase peaks.Several peaks appearing at 26.6 ∘ , 33.9 ∘ , and 51.8 ∘ in Figure 3 are due to the SnO 2 from FTO substrate.These results indicate that the TiO 2 films exhibit a stable anatase phase even after annealing at higher temperature of 500 ∘ C. No phase transformation was detected.Only P-25 film shows some rutile (R: rutile) characteristics diffraction peaks corresponding to about 20% of rutile TiO 2 .This is consistent with the crystal phase of original P-25 TiO 2 nanoparticles (XRD not shown).The average crystallite size of NS film, P-25 film, ST-01 film, and ST-21 film was estimated from the (101) peak according to the Scherrer equation,  = 0.89/ cos , where , , and  refer to X-ray wavelength (nm), the full width at half-maximum, and the diffraction  angle, respectively.The crystallite sizes listed in Table 1 are in good agreement with the TEM data.

Photovoltaic Properties.
The photovoltaic performance of DSSCs (NS cell, P-25 cell, ST-01 cell, and ST-21 cell) fabricated using the above prepared TiO 2 photoelectrodes (NS film, P-25 film, ST-01 film, and ST-21 film) was characterized by evaluating the current-voltage behaviour under one sun AM 1.5 irradiation from a solar simulator.Figure 4(a) shows the photocurrent-voltage characteristics (I-V) curves.Table 2 summarizes the TiO 2 film thickness, the amount of dye adsorption (N719 ads ), and the photoelectric data of the DSSCs in Figure 4(a), including, the short-circuit photocurrent density ( sc ), an open circuit voltage ( oc ), fill factor (FF), and photo-to-electric conversion efficiency ().It is apparent that the a-TiO 2 nanosol based photoelectrode achieved the highest photo-to-electric conversion efficiency of 7.20% with a short circuit current density of 13.3 mA/cm 2 , N719 ads of 0.132 mol/cm 2 ,  oc of 0.77 V, and FF of 0.70.In contrast to the DSSCs made from TiO 2 nanoparticlebased photoelectrode, the photo-to-electric conversion performance was significantly enhanced by using the TiO 2 nanosol-based photoelectrode.The efficiency and shortcircuit current density of DSSCs are mainly affected by the dye adsorption on the TiO 2 electrode.As expected, from the SEM analysis, the amount of dye adsorbed on TiO 2 nanosol based-photoelectrode was higher than that on the TiO 2 nanoparticles-based photoelectrodes (Table 2).Even though, the dye adsorption is higher for ST-01 nanoparticle based-photoelectrode than P-25 and ST-21 photoelectrode, the  sc (mA/cm 2 ) and  (%) are lower.This might be attributed to the large aggregation of the TiO 2 nanoparticles.Inhomogeneous film configuration creates electron traps that hinders the electron transportation and leads to the lower DSSCs efficiency.The TiO 2 nanosol-based photoelectrode exhibits 19%, 30%, and 16% increase in the photo-to-electric conversion efficiency compared to the P-25, ST-01, and ST-21 photoelectrode, respectively.
Higher photo-to-electric conversion efficiency and shortcircuit density of a-TiO 2 nanosol-based photoelectrode are attributed to higher amount of dye adsorption owing to larger surface area and more compact smooth surface morphology of TiO 2 photoanode.The higher dye adsorption value reveals that a-TiO 2 nanosol-based paste is well interconnected and the electrons are efficiently transported through the film which is consistent with the work reported by Jeong et al. [27].The incident photon-to-current conversion efficiency (IPCE) spectra were further measured for the above prepared DSSCs.From Figure 4(b), we can observe that the IPCE% efficiency for all the samples was maximum at the wavelength of 550 nm.Again, the NS cell shows a significant increase in IPCE percentage value over the long-wavelength range (530-700 nm) compared with other cells made from TiO 2 nanoparticles.This improvement in the long-wavelength range could be attributed to the enhancement of dye adsorption of the smooth TiO 2 film surface, leading to higher  sc .Further studies are required to analyze the properties of charge transport and electron recombination process in DSSCs.

Conclusion
The present work has demonstrated the effect of a-TiO 2 nanosol on the performance of DSSCs in comparison with the TiO 2 nanoparticles.The well dispersed oval shaped a-TiO 2 nanosol-based TiO 2 film (6 m thickness) with larger surface area results in higher dye adsorption, compared to the commercial TiO 2 nanoparticles film.Hence, a-TiO 2 nanosol-based photoanode exhibits a short-circuit current density ( sc ) of 13.3 mA cm −2 , an open-circuit voltage ( oc ) of 0.77 V, and a fill factor (FF) of 0.70 and achieves an overall light conversion efficiency () of 7.2%.The performance of the above photoelectrode was significantly higher than the commercial P-25, ST-01, and ST-21 nanoparticle-based electrodes, owing to the higher dye loading.In addition, the NS cell shows higher IPCE efficiency than that of TiO 2 nanoparticles based DSSCs.

3. 1 .
Structural Characterization of TiO 2 Nanoparticles and TiO 2 Films.The anatase TiO 2 (a-TiO 2 ) nanosol prepared by hydrothermal process was used to prepare TiO 2 paste (nanosol-based TiO 2 paste).The commercially available TiO 2 nanoparticles (P-25, ST-01, and ST-21) were concomitantly utilized to prepare TiO 2 pastes (nanoparticle-based TiO 2 pastes) by similar procedure (Section 2.4).Both types of TiO 2 pastes were screen-printed on FTO substrate and annealed with stepwise heating process in order to compare the effect of TiO 2 photoanodes prepared from nanosol-based TiO 2 paste and nanoparticle-based TiO 2 pastes in DSSCs.Figure1shows the typical SEM photographs of nanosol-based TiO 2 film (NS film) and nanoparticle-based TiO 2 films (P-25 film, ST-01 film, and ST-21 film) with two kinds of magnifications.The noteworthy difference in the morphology of NS film (Figure1(a)) and nanoparticle-based TiO 2 films (Figures1(b)-1(d)) could be observed instantaneously from the SEM images with lower magnifications (×5,000).The NS film (Figure1(a)) is extremely smooth and uniform.This result indicates that the highly dense TiO 2 particles formed are homogeneously dispersed throughout the NS film.

Table 1 :
Crystallite size, particle size, and specific surface area of various TiO 2 nanoparticles.
* Represents for film, NS indicates for TiO 2 nanosol.