Influence of the Sol-Gel pH Process and Compact Film on the Efficiency of TiO 2-Based Dye-Sensitized Solar Cells

1 Laboratory of Photochemistry and Energy Conversion (USP), Instituto de Quı́mica, Universidade de São Paulo, Avenue Prof. Lineu Prestes 748, 05508-900 São Paulo, SP, Brazil 2 Instituto de Quı́mica, Universidade Federal de Uberlândia 38400-902 Uberlândia, MG, Brazil 3 Departamento de Quı́mica, Universidade Federal de Minas Gerais 31270-010 Belo Horizonte, MG, Brazil 4 Departamento de Fı́sica, Universidade Federal de Minas Gerais 31270-010 Belo Horizonte, MG, Brazil


Introduction
Dye-sensitized solar cells (DSCs) are one of the new photovoltaic technologies capable of directly and efficiently converting solar light to electrical current at reasonable costs [1][2][3][4][5][6].Inspired on natural photosynthesis, sensitization of wide bang-gap semiconductors by selected dyes allows the separation of light harvesting and electron transport functions, which reduces the necessity of high-purity materials to reach reasonable conversion efficiencies [7][8][9].
Among oxide semiconductors, titanium dioxide, TiO 2 , is by far the most commonly used semiconductor in DSCs.It is an inert, nontoxic, cheap, and readily available material with wide band gap (E bg = 3.2 eV) and high refractive index, which are very suitable properties for solar cell applications [10][11][12].Several studies have shown that the methodology and the experimental conditions to prepare TiO 2 nanoparticles influence their morphologic and electronic properties and, consequently, the overall performance of DSCs [13][14][15][16][17][18][19][20][21].
Hore and coworkers [17] reported the influence of hydrolysis pH of titanium(IV) isopropoxide on the interfacial electron transfer kinetics of TiO 2 thin films sensitized by Ru(II) polypyridyl complexes.Their results showed that DSCs with TiO 2 nanoparticles synthesized under alkaline conditions exhibit slower interfacial recombination losses and higher open-circuit voltages than solar cells with oxide particles prepared by acid peptization.
In this paper, we provide new insights on the influence of the hydrolysis pH on the overall efficiency of DSCs, based on morphological characterization of TiO 2 nanoparticles synthesized under acid or alkaline conditions.Bare and sensitized TiO 2 films were analyzed by N 2 adsorption-desorption isotherms, scanning electron microscopy (SEM), and X-ray photoelectron spectroscopy International Journal of Photoenergy (XPS).Photoelectrochemical parameters determined for DSCs with nanoparticles obtained in different conditions were rationalized based on their morphological characterization.Additionally, the influence of TiCl 4 after treatment on mesoporous oxide surface is investigated, as well as the characteristics of DSCs using, simultaneously, acid and alkaline TiO 2 nanoparticles.

Experimental
All chemicals were used as received with the exception of 3-methyl-2-oxazolidinone (Aldrich), which was purified by distillation under reduced pressure.The N3 dye, cis-[Ru(dcbH 2 ) 2 (NCS) 2 ], dcbH 2 = 4,4 -dicarboxylic acid-2,2bipyridine, was synthesized as previously reported [22].TiO 2 nanoparticles were prepared by the sol-gel method [23] employing acid or alkaline hydrolysis conditions.In acid route, 12 mL of titanium(IV) isopropoxide (Strem, 98%) was slowly added to 70 mL of 0.1 mol L −1 HNO 3 aqueous solution under vigorous stirring.In alkaline hydrolysis, the same volume of titanium(IV) isopropoxide was slowly dropped into 70 mL of 0.1 mol L −1 ammonia solution.Both, acid and alkaline, mixtures were left under stirring and heating (80 • C) for 8 hours.Aliquots of the resulted sols, named acid and alkaline TiO 2 , respectively, were used for deposition of layer-by-layer (LbL) compact films.The remaining aliquots were autoclaved for 8 hours at 200 • C in a nonstirred titanium pressure vessel (Parr, 4750 Series), concentrated to ∼200 mg mL −1 , and stabilized with Carbowax 20 M (Supelco) to yield a paste employed for mesoporous TiO 2 layer.TiO 2 compact films, used as blocking and contact layers, were deposited onto cleaned FTO substrates (Pilkington, TEC-15, 15 Ω per square) using the LbL technique, as described elsewhere [24].The substrate was immersed alternately for 5 min in a 10 mg mL −1 suspension of the nonautoclaved acid TiO 2 at pH 2 and in a 10 mg mL −1 suspension of the nonautoclaved alkaline TiO 2 at pH 8.In this approach, acid and alkaline TiO 2 nanoparticles were used as cations and anions, respectively.
Mesoporous TiO 2 films were deposited directly over FTO substrates or on the compact film by the painting technique [23].The films were dried at room temperature and sintered at 450 • C for 30 minutes.The sensitization was achieved by immersion of electrodes in an N3 saturated ethanolic solution, and the amount of adsorbed dye was determined spectrophotometrically by desorption with NaOH 10 −4 mol L −1 aqueous solution.The surfaces of some TiO 2 films were treated by 0.05 mol L −1 TiCl 4 solution prior to sensitization, as described elsewhere [25].Sintered electrodes were immersed into the solution and kept at 70 • C for 30 minutes.After treatment, they were washed with water and resintered at 450 • C for 30 minutes.
Dye-sensitized solar cells with 0.25 cm 2 active area were assembled in a sandwich-type arrangement [23] using the sensitized TiO 2 photoanode and a transparent Pt-covered FTO as a counterelectrode.A solution of 0.05 mol L −1 I 2 /0.5 mol L −1 LiI/0.5 mol L −1 pyridine in 90 : 10 mixture of acetonitrile (Aldrich) and 3-methyl-2-oxazolidinone was used as an electrolyte mediator.
UV-Vis absorption spectra were obtained by using an 8453 UV-Vis spectrophotometer (Hewlett Packard).The size of TiO 2 nanoparticles and the morphology of the films were evaluated by field-emission scanning electron microscopy (FESEM) using a JSM 7401F (JEOL) microscope.Morphological parameters such as surface area and porosity were obtained by using nitrogen adsorption-desorption data acquired in an NOVA 2200e surface area analyzer (Quantachrome Instruments).The data were modeled by the Brunauer-Emmett-Teller (BET) equation [26].XPS analyses of TiO 2 films were carried out in an ESCALAB 220ixL spectrometer (VG Scientific) equipped with a hemispherical electron energy analyzer using Mg-Kα radiation (hν = 1487 eV).Photoelectron spectra were recorded in constant analyzer energy (CAE) mode.The binding energies were measured in reference to the C-1s peak at 284.6 eV.
Photoelectrochemical characterization of DSCs was carried out by current-potential measurements using a PAR 270 galvanostat/potentiostat (EG&G Instruments) system at simulated AM 1.5 solar radiation (100 mW cm −2 ) provided by a solar simulator (Newport/Oriel), as previously described [24,27].Photocurrent action spectra were obtained by using an Oriel system comprised by a 400 W Xe lamp coupled to a 0.25 m Czerny-Turner monochromator as described earlier [28].All parameters were determined from the average values measured with at least three individual cells of each type of photoanode.

Results and Discussion
3.1.Characterization of TiO 2 Nanoparticles.TiO 2 nanoparticles synthesized by acid and alkaline routes were analyzed by N 2 adsorption-desorption isotherms using the BET method.The main morphological parameters are shown in Table 1.
TiO 2 nanoparticles obtained by acid route showed a larger surface area and slightly higher porosity than the ones obtained at alkaline medium.In fact, acid pH hydrolysis results in a very transparent sol, different from the film obtained by alkaline route, which is opaque due to the presence of larger particles and aggregates.Scanning electron micrographs of TiO 2 films deposited over FTO substrates confirm that the alkaline route yields larger nanoparticles, Figure 1.Similar effect was reported by Barbé et al. by changing the pH of hydrothermal treatment of nanoparticles produced after an acid hydrolysis [13].
Typically, acid or alkaline-promoted hydrolysis of Ti(IV) alkoxides occurs as follows [29][30][31]: Hydrolysis is followed by a condensation step to yield Ti-O-Ti chains as follows:  Continuous hydrolysis/condensation of alkoxyl groups will lead to three-dimensional polymeric Ti-O-Ti chains, which yield TiO 2 particles with different shapes and sizes, depending on experimental conditions (temperature, condensing agents, templates, etc.) [19,29,32].
In acid promoted hydrolysis, it is expected that the superficial Ti-OH sites will be constantly protonated during peptization, which decreases the condensation rate and results in smaller particles with positive surface charges that avoid aggregation and increase the porosity of the resulted film.On the other hand, in alkaline catalysis, the superficial Ti-OH Ti-O − + H + equilibrium is shifted to the dissociated form, which facilitates the condensation with other Ti-OH or Ti-OR sites to yield larger nanoparticles.Moreover, during the hydrothermal step in alkaline conditions, the Ostwald ripening effect or coarsening is more effective than in acid medium enhancing the particle growth as discussed by Barbé and coworkers [13].In the coarsening growth mechanism, larger particles are formed at expense of dissolution of small ones [33].The growth rate is proportional to the particle solubility and solid-liquid interfacial tension [34].In alkaline medium, the solubility of TiO 2 is increased, likely due to the complexation of generated anionic species by NH 4 + ions [13], consequently, the coarsening rate constant is enhanced.The resulted TiO 2 nanoparticles are larger than the particles obtained in acid medium, as can be seen in Figure 1.
Both acidic or alkaline processed, TiO 2 pastes result in 6 to 8 μm thickness films, with considerable different optical properties.The acid TiO 2 films are highly transparent with transmittance of 58% at 500 nm, while the alkaline TiO 2 films exhibit only 40% transmittance at 500 nm.
Adsorption of N3 on both TiO 2 films was carried out in ethanolic solutions without any other reagent.In this condition, the N3 surface concentration on films with nanoparticles obtained by the alkaline pathway (2.6 × 10 −8 mol cm −2 ) is c.a. 3 times lower than that determined for acid TiO 2 nanoparticles (8.6 × 10 −8 mol cm −2 ).Higher surface area as well as concentration of hydroxyl groups on the surface of TiO 2 nanoparticles synthesized under acid conditions justifies higher dye loadings.
XPS analyses of sensitized films have shown similar adsorption modes for both acid and alkaline TiO 2 films.O-1s and Ru-3d 5/2 signals, Figures 2(a) and 2(b), respectively, are very similar in both films, which indicates that the chemical environments around Ru(II) and O 2− ions are the same.In Figure 2(a), one can also observe a broadening of the O-1s peak at the higher binding energy side, which is caused by the presence of free and bounded carboxylate groups from N3 species on TiO 2 surface, as previously discussed [23].

Photoelectrochemical Behavior of DSCs Employing Acid-
or Alkaline-Hydrolyzed TiO 2 Nanoparticles.Photoelectrochemical parameters determined for DSCs prepared with acid or alkaline TiO 2 nanoparticles are shown in Table 2. Some solar cells were prepared with films further treated with TiCl 4 solutions in order to increase the quantum yield of charge separation at TiO 2 /dye interface, enhancing the photocurrent [25,35].
One can observe that DSCs prepared with TiO 2 nanoparticles obtained by the acid pathway exhibited conversion efficiencies c.a. 40% higher than the ones using the particles obtained by the alkaline route.Despite their better opencircuit voltages (V oc ) and fill factor (ff), the solar cells using alkaline TiO 2 have shown short-circuit current densities (J sc ) 50% lower than the acid TiO 2 -based DSCs, which is the main reason for the difference observed in the overall efficiency.In DSCs, the short-circuit current (I sc ) can be defined by (3) [35], in which I 0 is the incident photon flux in Coulomb s −1 , Φ lh is the light-harvesting efficiency, Φ cs is the quantum yield for charge separation at TiO 2 /dye interface, and Φ col is the charge collection efficiency.Φ lh is dependent on the nature of the sensitizer and its concentration on the surface, whereas Φ cs and Φ col depend basically on the morphological and electronic properties of the semiconductor metal oxide film The reduction of photocurrent observed for the DSCs employing alkaline TiO 2 nanoparticles can be justified by the decrease of Φ lh , since the amount of N3 adsorbed in this film is three times lower than adsorbed on acid TiO 2 .In fact, Φ lh is only dependent on the optical and excited state properties of the selected sensitizer.Thus, it can be expected that the observed photocurrent will be proportional to the N3 concentration in each TiO 2 film.However, while the photocurrent of DSCs using the alkaline TiO 2 decreases 40% in relation to the ones using acid TiO 2 films, the N3 concentration in the different photoanodes varies 70%, that is, the photocurrent is not proportional to the N3 concentration.
These results evidence that TiO 2 nanoparticles synthesized under alkaline hydrolysis conditions exhibit better charge separation and charge collection efficiencies.From photoelectrochemical parameters determined in this work, it can be estimated that the product Φ cs Φ col is c.a. 27% higher for alkaline TiO 2 in relation to the nanoparticles synthesized in acid conditions.This analysis agrees with previous discussion carried out by Hore and coworkers based on transient absorption data, in which the different dependence of charge recombination rates on applied bias for acid and alkaline TiO 2 films evidences that the alkaline hydrolysis results in a decrease of charge recombination rate [17].Additional evidence for the better electronic properties of TiO 2 nanoparticles synthesized by alkaline route is given by the lower dark current observed for the DSCs prepared with their films, as in Figure 3.
The surface posttreatment of alkaline TiO 2 -based films with acidified TiCl 4 solutions resulted in an increase of c.a. 7% of adsorbed N3, similarly to what is observed for acid TiO 2 -based films and to earlier reported works [25,35,36].As expected, the photocurrent observed for DSCs with both treated films increases linearly with the N3 concentration, as in Table 2. Photoaction spectra of DSCs show that the TiCl 4 treatment improves the light-to-current conversion efficiency in all visible region, as in Figure 4.
Photoanodes comprised by an acid TiO 2 film deposited over an alkaline TiO 2 layer were prepared in an attempt to integrate the better electronic properties of TiO 2 synthesized by the alkaline route with the higher dye loading of the acidhydrolyzed particles.The resulting solar cells have shown intermediate properties in relation to those with individual acid and alkaline TiO 2 films and, in this new setup, the maximum overall efficiency achieved 6.3%, as shown in Figure 5. Another effective strategy is the use of a compact oxide layer on the FTO surface beneath the mesoporous semiconductor layer to aggregate the desired morphological and electronic properties of the distinct TiO 2 nanoparticles.Positive charged nonautoclaved acid TiO 2 are attracted by negative charged alkaline ones to yield a compact film over FTO.This procedure leads to self-assembled TiO 2 films more stable to the sintering temperature than those obtained using organic polyanions [37].Thus, the produced films address the two main requisites to be an efficient compact blocking layer in DSCs [37]: low porosity and temperature stability.This compact film is an elegant solution to prevent the physical contact between the electrolyte and the FTO surface, decreasing the charge recombination at this interface and, consequently, improving the conversion efficiency [38][39][40].A mesoporous layer of acid TiO 2 was deposited over the FTO substrate modified by 15 or 30 bilayers of the LbL TiO 2 compact film.The use of resulted electrodes as photoanodes in DSCs increases the solar cell overall efficiency by 6% and 12%, respectively.Thus, the compact film produced with acid and alkaline TiO 2 nanoparticles effectively avoids the charge recombination at FTO/electrolyte interface and improves the electrical contact of FTO with the mesoporous TiO 2 layer [24,37].However, these compact films have also led to a decrease of 16% and 23%, respectively, on the transmittance of the FTO substrate.Therefore, when photoelectrochemical data are corrected, the increase on the overall efficiency is, respectively, 26% and 46%.

Conclusions
In this work, the influence of pH during hydrolysis of titanium(IV) isopropoxide on the morphologic and electronic properties of TiO 2 thin films is discussed and correlated to the solar light-to-current conversion efficiency of DSCs based on such films.TiO 2 particles hydrolyzed under alkaline conditions exhibit higher diameters, lower porosity, and lower dye loading than the nanoparticles synthesized by the acid route.On the other hand, when N3-sensitized thin films based on alkaline TiO 2 are applied as photoanode in DSCs, one can observe a better charge separation as well as higher charge collection efficiencies, which evidence that electron transport among alkaline TiO 2 nanoparticles is more efficient than in acid TiO 2 -based films.
Based on these experimental observations, two new strategies were proposed in order to improve the DSC efficiency.In a first attempt, DSCs were prepared with a photoanode comprised by acid TiO 2 nanoparticles over an alkaline TiO 2 -based film, and higher dye loadings are achieved as well as high charge separation and collection efficiencies, which result in a better overall conversion efficiency than DSCs using the nanoparticles separately.

International Journal of Photoenergy
This synergic effect is also achieved in compact layers prepared by the LbL technique using acid TiO 2 as cations and alkaline TiO 2 as anions before the mesoporous layer.The resulted compact film physically prevents the electron recombination at FTO/electrolyte interface and also improves the electrical contact between the conductive substrate and the mesoporous layer.Such a strategy has resulted in an improvement of 12% in the conversion efficiency of solar cells.

Figure 1 :
Figure 1: Scanning electron micrographs of TiO 2 nanoparticles obtained by acid (a) or alkaline (b) routes on FTO.

Figure 5 :
Figure 5: Current-voltage curve for a DSC with a photoanode having an acid TiO 2 film deposited over an alkaline TiO 2 layer (AM 1.5 illumination; 100 mW cm −2 ).

Table 1 :
Morphological parameters determined by the BET method for TiO 2 nanoparticles synthesized in different hydrolysis conditions.
a Average diameter considering spherical particles.