Spectroscopic and Morphological Studies of Metal-Organic and Metal-Free Dyes onto Titania Films for Dye-Sensitized Solar Cells

1 Institute for the Study of Nanostructured Materials (ISMN), CNR, Via Salaria km 29300, Monterotondo, 00015 Rome, Italy 2 Center for Biomolecular Nanotechnologies (CBN), Fondazione Istituto Italiano di Tecnologia, Energy Platform Via Barsanti, Arnesano, 73010 Lecce, Italy 3 National Nanotechnology Laboratory (NNL), CNR Istituto Nanoscienze, c/o Distretto Tecnologico, Via Arnesano km 5, 73100 Lecce, Italy 4Department of Mathematics and Physics “E. De Giorgi”, University of Salento, Via per Arnesano, 73100 Lecce, Italy 5 Institute of Structure of Matter (ISM), CNR, Via Salaria km 29300, Monterotondo, 00015 Rome, Italy


Introduction
Dye-sensitized solar cells (DSSCs) are receiving increasing attention during the last decades since they are considered a viable and cost-effective technology for the conversion of sunlight into electricity [1][2][3].In these devices, metalorganic or metal-free dyes are chemically adsorbed onto a wide band gap semiconductor oxide [1][2][3][4][5][6][7][8].The working principle of a DSSCs system is based on the photochemical and photophysical processes: upon photoexcitation of the dyes by visible light, electrons are excited from the ground state of the dye to its excited state, and then their injection can occur from the excited dye into the conduction band of the semiconductor.
To achieve a high power conversion efficiency, the electron injection process must be certain and faster than International Journal of Photoenergy the competing fluorescence and nonradiative processes.This photoinduced electron transfer leads to the production of photocurrent and consequently it is crucial to the device efficiency [1][2][3].
In DSSCs, the sensitizer is one of the critical components as it absorbs sunlight and induces the charge separation process.In order to enhance power conversion efficiencies of DSSCs, it is imperative to improve the light-harvesting ability of the dyes.It is also fundamental to investigate and to optimize the dye-substrate interaction that is involved in the electron transfer process [9].
The critical factors that influence sensitization are the excited-state redox potential, which should match the energy of the conduction band edge of the semiconductor substrate, the light-harvesting ability, and the electron transfer process that is influenced by the electronic coupling between the lowest unoccupied molecular orbital (LUMO) of the dye and the TiO 2 conduction band.One of the major factors for low conversion efficiency of many organic dyes in the DSSC is the formation of dye aggregates on the semiconductor surface.Therefore, to obtain optimal performance, aggregation of dyes needs to be avoided by promoting the chemisorption onto the substrate and by increasing the surface area of the semiconductor oxide.High surface area substrates can also promote the adsorption of a larger amount of dye molecules enhancing the light-harvesting efficiency.However, in order to improve the power conversion efficiency of photoanodes for DSSCs, it is necessary to achieve both optimal morphological properties and fast electron transport [19][20][21][22].
To gain information about the photoinduced electron transfer process from the sensitizers to the semiconductors and indications about the linkage between dye and substrate, the study of the spectroscopic features of chromophores onto the semiconductor is of crucial importance.Absorption and steady-state emission spectral features can give information on the dye interaction with the substrate, since they are influenced by the chemical environment of the sensitizers and hence, it is expected that the structural and chemical nature of the sensitizers and the structural properties of titania substrates have a key role in determining the final spectral features of the chromophore.Moreover, time-resolved fluorescence decay can provide insights into the electron injection process, since it is competitive with fluorescence relaxation.Electron injection is found to generally occur with a time much shorter than the excited-state lifetime of typical dyes, therefore time-resolved fluorescence decay studies are very useful to understand these dynamic photoinduced processes.

Preparation of Titania Thin Films
2.1.1.Transparent and Mesoporous Films.Mesoporous titania films were synthesized using the P123 block copolymer as template in order to obtain an ordered porous structure with a controlled geometry.The titania precursor solutions were prepared according to the procedures described in the literature using pluronic surfactant [24].
In a typical synthesis, 1.4 mL of concentrated HCl were slowly added to 2.1 g of titanium(IV) ethoxide at room temperature.In another flask, 0.65 g of P123 were dissolved in 6 g of 1-butanol and subsequently added to the titanium solution.
The system was kept under stirring at room temperature for 3 hours.The as-prepared solution of titania precursor was deposited on FTO conducting glass substrates (Dyesol, 15 Ω sq −1 ) by spin-coating at 2400 rpm.After aging at 25 ∘ C for 2 days, optically uniform and transparent films were obtained.These films were calcined at 350 for 30 minutes using a ramp rate of 1 ∘ C/min to remove the template and the residual organic precursors.TiO 2 electrodes were obtained by deposition of three layers with a final calcination at 350 ∘ C for 2 hours and were labeled as TiMS (Table 1).The resulting mesoscopic oxide film was around 1 m thick and transparent.

Semiopaque
Films from Mesoporous Titania.Mesoporous titania powder was prepared from the solution of titanium (IV) ethoxide and P123 described above after aging at 25 ∘ C for 5 days and subsequent calcination at 350 ∘ C for 4 hours using a ramp rate of 1 ∘ C/min, according to a recently published procedure [25].
Titania films were prepared from mesoporous titania powder by using the doctor-blade technique, according to a procedure described elsewhere [26].Dried powder was dispersed in ethanol (EtOH), treated with ultrasonic bath for 30 min, then stirred at 50 ∘ C for 1 h.This procedure was repeated three times in order to obtain a homogeneous and ).Two edges of the FTO glass plate were covered with a layer of adhesive tape (3 M Magic) to control the thickness of the film; successively the titania paste was spread uniformly on the substrate by sliding a glass rod along the tape spacer.After drying the coated plates were sintered in air for 1 h at 450 ∘ C and labelled as TiMS DB (Table 1).The resulting mesoscopic oxide film was around 1 m thick and semiopaque.

Semiopaque Films from Commercial Titania.
Reference titania films were prepared from commercial P25 titania powder by using the doctor-blade technique, according to a procedure described elsewhere [26].Briefly, acetic acid, water and ethanol were added drop by drop into a mortar containing P25 titanium dioxide nanoparticles (provided by Evonik Degussa).The TiO 2 dispersion was transferred with excess of ethanol into a beaker for stirring a sonication steps; then anhydrous terpineol and ethyl cellulose (previously dissolved in ethanol) were added.After further stirring and sonication steps the dispersion was concentrated by rotary evaporator.The final colloidal paste was characterized by the following composition (wt/wt): 5% TiO 2 , 0.5% acetic acid, 2.5% ethylcellulose, 87% terpineol, and 5% ethanol.The soobtained titania paste was deposited on the FTO conducting glass (Dyesol, 15 Ω sq −1 ).Two edges of the FTO glass plate were covered with a layer of adhesive tape (3 M Magic) to control the thickness of the film; successively the titania paste was spread uniformly on the substrate by sliding a glass rod along the tape spacer.After drying, the coated plates were sintered in air for 1 h at 450 ∘ C and labelled as P25 DB (Table 1).The resulting mesoscopic oxide film was around 1 m thick and semiopaque.

Adsorption of Dyes onto Titania Films.
The dye-sensitized films were prepared using a 0.2 mM N719 solution (ditetrabutylammonium cis-bis(isothiocyanato) bis(2,2  -bipyridyl-4,4  -dicarboxylato)ruthenium(II)), which was obtained by dissolving the dye in a mixture of acetonitrile and tert-butyl alcohol (1 : 1 volume ratio).The titania films were immersed into the N719 solution overnight at room temperature.After that all the dyed films were washed with acetonitrile and stored in the dark.
A similar procedure was used for the adsorption of D5 and ZnPc.The D5 dye was dissolved in acetonitrile to obtain a dye concentration of 0.2 mM, whereas the solution of ZnPc was obtained by dissolution in ethanol (0.89 mg of dye in 10 mL of ethanol).

Characterization.
UV-vis spectra of the films were collected in the wavelength range 300-870 nm using a double beam spectrophotometer V-660 (Jasco) equipped with an integrated sphere.The absorbance of the dye solution was measured using a quartz cell of thickness 1 cm.The absorbance of dye-TiO 2 systems was recorded using 60 mm integrated sphere.
Steady-state fluorescence spectra for dyes solution and dyes-TiO 2 films were recorded in a Jobin Yvon Fluorolog3 spectrofluorometer, using grids of 5 nm for the excitation and 5 nm for emission.Time-resolved fluorescence measurements were carried out by a time-correlated single-photon-counting (TCSPC) system (Horiba-Jobin Yvon).Time-resolved measurements are carried on exciting the samples using a 405 nm pulsed laser diode and collecting the emission decay at the corresponding maximum emission wavelength for the specific analysed dye.The fluorescence decay profiles were analysed through decay analysis software (DAS6a HORIBA Scientific) to a multiexponential decay equation.The quality of the fits was checked by examining the residual distribution and the  2 value.
Both steady-state and time-resolved fluorescence measurements of thin films TiO 2 systems were collected in rightangle (RA) configuration with a tilt angle of 30 ∘ to minimize substrate scattering effects.
Fourier transform infrared (FT-IR) spectra were collected using a Jasco 4100 FT/IR spectrophotometer equipped with an attenuated total reflectance (ATR) unit.The samples for the analysis were prepared through preliminary drying at 120 ∘ C. FE-SEM characterizations were carried out with a highbrilliance LEO 1530 field emission scanning electron microscope apparatus equipped with an INCA 450 energydispersive X-ray spectrometer (EDS) and a four-sector backscattered electron detector (BSD).Before the analysis, the surfaces were coated with a thin layer of carbon in order to avoid charging effects.The carbon coating was deposited by using an Emitech sputter coater K550 unit, a K 250 carboncoating attachment, and a carbon cord at a pressure of 1 × 10 −2 mbar in order to produce a carbon film with a thickness of a few nanometers.

Results and Discussion
The spectral properties of different dyes have been compared in order to investigate the electron transfer process from the dye to the conduction band of titania films that is responsible for the production of photocurrent in DSSCs.Metalorganic and metal-free dyes were adsorbed onto transparent and semiopaque mesoporous titania thin films and onto a semiopaque film form commercial P25 titania.In particular, we have investigated the behaviour of a ruthenium complex (N719), an organic dye (D5), and a zinc phthalocyanine system (ZnPc).The chemical structure of these dyes is shown in Figure 1.
All the dyes have been adsorbed onto titania films with different morphology: a mesoporous transparent film deposited by spin-coating (TiMS), a semiopaque film deposited by doctor-blade from mesoporous titania (TiMS DB), and a semiopaque film deposited by doctorblade form commercial P25 titania (P25 DB).The preparation of the films was properly tailored in order to get thin substrates with a comparable thickness of about 1 m.

FTIR Characterization.
The surface functional groups of titania substrates play an important role in the chemisorption of the dyes.Indeed, the sensitizers that we are considering in this work are usually bonded to titania through their carboxylic groups, which interact with hydroxyl groups at titania surface.
ATR-FTIR measurements were used to investigate the amount of hydroxyl groups in the different titania films, and the spectra are displayed in Figure 2. Furthermore, by these analyses it is also possible to get more information on the presence of carbon species in the samples.
The ATR-FTIR spectra show similar trends, but with some significant differences.The bending vibration band of the surface hydroxyl groups is observed at about 1630 cm −1 , and the corresponding stretching vibration peak is located at 3200-3600 cm −1 [27,28].From these spectra, a significantly higher amount of hydroxyl groups was noticed in transparent titania thin films (TiMS).This could be attributed to the lower sinterization temperature (350 ∘ C instead of 450 ∘ C used for DB films) that allows preserving a high amount of hydroxyl surface groups, and it seems also reasonable to assume that it could be also related to a higher surface area.On the contrary, the film from Degussa P25 has the lowest amount of surface hydroxyl groups.
Analyzing the FTIR spectra, it can also be observed that titania films contain carbon groups.Indeed, the peaks at about 1000-1100 cm −1 suggest the presence of Ti-O-C bonds [27].Furthermore, the peak at about 1400 cm −1 has been attributed to asymmetric bending vibrations of C-H bonds [27,28].The observation from FTIR that the films contain Ti-O-C and C-H bonds indicates the presence of residual organic groups.The intensity of these signals for P25 DB and TiMS is comparable, whereas the residual carbon content is apparently lower for TiMS DB film (Figure 2).

Morphological Investigation.
Further insights into the structural properties of titania films were obtained through FE-SEM analysis.The micrographs of titania surface films prepared by doctor-blade (Figure 3) reveal that when commercial P25 was used for the formulation of the paste the photoanode surface consists of small nanoparticles with voids into the structure.By using mesoporous titania, the photoanode surface is composed of small nanoparticles and a porous structure can be observed.It is worth to notice that in TiMS DB the surface morphology is more uniform with the presence of small pores that could increase the interface between dye and titania, whereas larger voids are not detected.
In TiMS prepared by spin-coating, the surface morphology is completely different.The films are furthermore compact, and can be observed the presence of domains with an ordered structure.The latter are mesoporous channels, which are responsible for a nanoporous structure and high surface area.

UV-Vis and Fluorescence Measurements.
In order to gain information on the anchorage of the chromophore onto the semiconductor and on the electron transfer process from the dyes to titania, the spectroscopic features of the dye solutions and of the dye-sensitized titania films have been investigated.
The absorption and the steady-state fluorescence spectra of the free dye solutions and the corresponding dye-sensitized TiO 2 films were recorded, and the data are summarized in Table 2.
The absorption spectra of N719 sensitized TiO 2 films along with the reference spectrum of the free dye in an acetonitrile and tert-butyl alcohol solution (1 : 1 volume ratio) are shown in Figure 4.In particular, the UV-vis absorption spectrum of N719 solution is dominated by two broad bands at around 386 and 537 nm, generally assigned to a Metal-to-Ligand Charge-Transfer (MLCT) transition in which an electron is transferred from Ru to one of the bipyridine ligands, and an intense UV absorption at around 310 nm attributed to an electronic transition between the - * orbitals of the dcbpyH ligands [29][30][31].When the N719 molecules are adsorbed onto the surface of mesoporous TiO 2 films, the MLCT band at 537 nm becomes predominant and shifts at lower wavelengths depending on the considered TiO 2 system.The measured blue shifts for all the dye-sensitized TiO 2 systems are reported in Table 2; as it can be seen it is higher for the N719 adsorbed onto TiO 2 films obtained by doctorblade (P25 and TiMS DB), whereas it is lower for the N719 adsorbed onto transparent TiO 2 obtained by spin-coating technique (TiMS).The observed shifts in the absorption spectra are probably due to chemical interactions between the dye and the semiconductor TiO 2 surface, which modify HOMO and LUMO levels of the adsorbed complexes with respect to those of the free dye.At the same time, we would like to underline the higher absorption intensity shown by the transparent N719@TiMS systems with respect to the other TiO 2 systems, which may indicate that this transparent layer can adsorb more N719 molecules compared with the other TiO 2 layers having the same thickness.This is also confirmed by the intense red coloration of the TiMS sample after the dye adsorption.These findings are in agreement with ATR-FTIR results previously discussed, since the most intense peaks in UV-vis spectra of dye-sensitized TiMS films is promoted by the presence of a high number of surface hydroxyl groups.
A similar trend has been also found in the case of the metal-free organic dye D5 adsorbed on TiO 2 mesoporous systems (Table 2).Two absorption bands at 479 nm and 300 nm are visible in the Uv-Vis spectrum of dye solution; the band in the UV region corresponds to a - * transition of the conjugated molecules, whereas the band in the Vis can be assigned to an intramolecular charge-transfer state between the diphenylaniline electron donating moiety and the cyanoacetic acceptor [15].When the dye is attached to TiO 2 surface, a blue shift of the absorption maximum at 479 nm is found (Table 2).Also in the case of D5, the transparent TiMS thin films have the highest intense absorption band and the lowst blue shift with respect to those observed for D5@TiMS DB and D5@P25 DB, thus indicating a higher dye uptaking.
The fluorescence measurements of N719 and D5 sensitized mesoporous TiO 2 systems, reported in Figures 5 and  6, show the characteristic fluorescence emission of the free dyes but shifted to lower wavelenghts.In the case of N719 (Figure 5), for example, upon excitation at  ex = 510 nm, the solution has an emission maximum at 728 nm, which shifts at 725, 719, and 720 nm for N719@P25 DB, N719@TiMS DB, and N719@TiMS, respectively.In analogy, the D5 sensitized TiO 2 systems have a blue shift in the emission spectra (Figure 6) that depend on the TiO 2 system.More interesting, if we consider the intensity of the emission band for both of the two sets of sample we observe that they decrease following  the order P25 DB > TiMS DB > TiMS.The reduction in the emission intensity is thus more evident in the case of transparent mesoporous TiO 2 layer, and we hypothesized that it can be associated in first approximation to a better and more  Figure 7.The UV-Vis absorption spectrum of ZnPc exhibits an intense Soret band (300-400 nm) due to the - * transition with charge-transfer character and a strong Q band centered at 687 nm [32].In addition, compared with the solution, the zinc phthalocyanine derivative on mesoporous TiO 2 films show broader absorption spectra extended to 800 nm, suggesting the formation of aggregate on the surface of the semiconductor [33].The absorption spectra of the phthalocyanine adsorbed onto TiO 2 systems exhibit a small red shift (Table 2) that may be due to the presence of carboxylic protons of phthalocyanine, which, upon adsorption on TiO 2 , release the proton and bind to Ti 4+ centres.The Ti 4+ sites act as electron withdrawing and produce the observed red shift in the absorption bands [34].
Figure 8 shows the steady-state fluorescence emission spectra ( ex = 400 nm) of ZnPc solution: the typical phthalocyanine emission band centred around 700 nm is easily visible [35].The ZnPc@P25 DB, ZnPc@TiMS DB, and ZnPc@TiMS exhibit a strong Q band emission peak at 695 nm, 698 nm, and 702 nm, respectively.In analogy to the other two dyes, when ZnPc is adsorbed onto the TiO 2 layers, a net decrease in the emission intensity can be observed, thus suggesting the possibility of an electron injection process.
Time-correlated single-photon counting (TCSPC) measurements are performed to obtain information about the decay times of the relaxed excited state at S 1 of the dye interacting with the semiconductor materials.As reported by several studies [18,23,36,37], TCSPC measurements can be useful to evaluate the photoelectron injection dynamics in dye-sensitized semiconductor films.In Figure 9 the timeresolved emission kinetics for the three dyes in solution and the correspondent dye sensitised TiO 2 systems are shown.In particular, the emission decay measurements are collected at  em = 730 nm (Figure 9  for N719 interacting with different TiO 2 based films, the decays have a more complex behaviour and need a three component exponential function.The results shown in Table 3 reveal that for dye solution the most significative time value ( 1 ) is about 3 ns ( 1 = 93%) and a little percent of molecules decay with a time of  2 = 27.7 ns ( 2 = 7%).The longer decay is comparable with time decay reported in the literature for N719 systems [38], while the shorter excited-state lifetime indicates an enhancing of the nonradiative deactivation channels, probably due to the presence of protons on carboxyl groups [39].The  1 = 3 ns value appears to be retained in N719-TiO 2 systems, but the corresponding amplitude ( 1 ) decreases from 93% in solution to 30% in solid state.Also the longest time decay in solution ( 2 = 27.7 ns) is preserved when the dye is adsorbed on TiO 2 surface, but this value is lower ( 2 = 15 ns in solid state) because dye molecules are now closer to each other than in solution, and this results in the quenching phenomena.In addition, a third shorter time appears (0.56 ns for N719@P25 DB, 0.41 ns for N719@TiMS DB and 0.36 ns for N719@TiMS) that can be related to the electron transfer process from the excited state of the dye to the conduction band of the semiconductor.Lifetime of D5 dye in solution is analysed by a three exponential component function, and the result of 0.14 ns, associated with the most significative amplitude ( 1 = 80%), presents a good agreement with analogous systems [40].However, for D5 sensitised TiO 2 systems the resolution of the TCSPC setup does not allow for resolving numerically the changes in lifetime.They are all sub 0.01 ns, a time close to the time response of the detector.Nevertheless, just from a qualitatively point of view, it is interesting to observe that D5@TiMS system seems to have the sharpest decay profile with respect to D5@TiMS DB and D5@P25 DB.The better quenching effect could be related to a different dye-TiO 2 interaction that could promote electron injection in semiconductor bands.
In the case of ZnPc solution, the fluorescence emission decay exhibits a monoexponential behaviour with a lifetime of 3 ns, comparable for values reported in the literature for unsubstituted phthalocyanine systems [18,41].
ZnPc sensitised TiO 2 systems fluorescence emission decays, instead, are analysed using a three component exponential function and the results are summarized in Table 3. From the fitting results, the  1 component is preserved in solid state systems, even if the molecule population associated to this decay mechanism decreases to 10-20%.The component  2 , observable only in solid state systems, can be associated to a quenching effect with a mechanism similar to the behavior of J aggregates in solution [42].The faster time decay component in dye-TiO 2 system ( 3 ) could be, in principle, indicative of an efficient injection occurring at the interface between dye molecules and semiconductor surface.Mesoporous TiO 2 systems present  3 decay very similar in value (0.13 ns for ZnPc@TiMS DB and 0.11 ns for ZnPc@TIMS), while ZnPc@P25 DB shows a quite lower  3 component, about 0.05 ns.
The two observed fast time decays  2 and  3 in dyes-TiO 2 systems could be also attributable to a contemporary presence of different injection processes; due to different possible configurations of ZnPc molecules when they are interacting with TiO 2 surface, one is more favourable than the other.

Conclusions
The steady-state measurements and fluorescence emission decay of the dye-sensitized titania films suggest that the transparent mesoporous titania film (TiMS) could be responsible for a more efficient electronic injection when N719 and D5 are used as sensitizers.Indeed, the emission peak of these dyes has a lower intensity onto TiMS, and a faster time decay was observed.In the case of ZnPc, the fluorescence emission decay is faster if the dye is adsorbed onto P25 DB.Probably the formation of dye aggregates hinders the diffusion into the mesoporous structure, and only the external surface area can interact with the sensitizer.The mesoporous structure of TiMS could not be accessible to ZnPc aggregates with a consequent lower performance.
These findings suggest that mesoporous titania films could be considered the most efficient substrates for the adsorption of N719 and D5 to realize a photoanode with high efficient electronic injection.
In fact, as spectroscopic measurements showed, the N719@TiMS presents the highest UV-vis absorbance value, compared to the other TiO 2 substrates, but the highest decrease in fluorescence emission intensity.These results suggest a strong interaction between the dye and the TiO 2 substrate, and this condition is particularly favourable to enhance the electron injection.
Figure7.The UV-Vis absorption spectrum of ZnPc exhibits an intense Soret band (300-400 nm) due to the - * transition with charge-transfer character and a strong Q band centered at 687 nm[32].In addition, compared with the solution, the zinc phthalocyanine derivative on mesoporous TiO 2 films show broader absorption spectra extended to 800 nm, suggesting the formation of aggregate on the surface of the semiconductor[33].The absorption spectra of the phthalocyanine adsorbed onto TiO 2 systems exhibit a small red shift (Table2) that may be due to the presence of carboxylic protons of phthalocyanine, which, upon adsorption on TiO 2 , release the proton and bind to Ti 4+ centres.The Ti 4+ sites act as electron withdrawing and produce the observed red shift in the absorption bands[34].Figure8shows the steady-state fluorescence emission spectra ( ex = 400 nm) of ZnPc solution: the typical phthalocyanine emission band centred around 700 nm is easily visible[35].The ZnPc@P25 DB, ZnPc@TiMS DB, and ZnPc@TiMS exhibit a strong Q band emission peak at 695 nm, 698 nm, and 702 nm, respectively.In analogy to the other two dyes, when ZnPc is adsorbed onto the TiO 2 layers, a net decrease in the emission intensity can be observed, thus suggesting the possibility of an electron injection process.Time-correlated single-photon counting (TCSPC) measurements are performed to obtain information about the decay times of the relaxed excited state at S 1 of the dye interacting with the semiconductor materials.As reported by several studies[18,23,36,37], TCSPC measurements can be useful to evaluate the photoelectron injection dynamics in dye-sensitized semiconductor films.In Figure9the timeresolved emission kinetics for the three dyes in solution and the correspondent dye sensitised TiO 2 systems are shown.In particular, the emission decay measurements are collected at  em = 730 nm (Figure9(a)) for N719, at  em = 640 nm (Figure 9(b)) for D5, and at  ex = 700 nm for ZnPc (Figure 9(c)).Fluorescence emission decay curves of the N719 solution have been deconvoluted by a biexponential function, while

Table 1 :
Details about the preparation of titania films.

Table 2 :
Peak position of UV-Vis absorbance and fluorescence emission.