Synthesis and Characterization of Cerium, Nitrogen, and co-doped TiO2 Photoanodes for Dye-Sensitized Solar Cells

INTRODUCTION
The rapid and extensive use of fossil fuels, coupled with the negligent implementation of environmental regulations, has brought our planet to the brink of destruction [1]. This has compelled humanity to seek environmentally benign alternatives [2]. Renewable energy sources are widely recognized as the most effective means of mitigating the large-scale exploitation of fossil fuels, a primary driver of flooding, global warming, and air pollution [3]. To address these pressing issues, there is an urgent need for cost-effective and proficient photovoltaic devices that can harness solar energy and convert it into electricity. The third generation of photovoltaic technology, i.e., Dye-sensitized solar cell (DSSC), has emerged as a promising solution. The first DSSC was developed in 1991 through the pioneering work of Grätzel and O’Regan [4]. Following silicon-based solar cells, DSSCs have gained significant prominence in photovoltaic technology because of their inherent advantages including environmental friendliness and low cost. Fundamentally, a DSSC comprises four key components: the photoanode, the sensitizer, the redox couple, and the counter electrode. A major obstacle hindering the attainment of higher power conversion efficiency (PCE) is interfacial recombination between the fluorine-doped tin oxide (FTO), the sensitizer, and the photoanode [5]. Furthermore, while dye absorption spans a wide wavelength range, leading to the generation of electrons with varying energy levels, only those electrons that possess energies equal to or less than the bandgap of the photoanode can participate in the charge transfer process [6]. 
Anatase titanium dioxide (TiO2) serves as a widely utilized photoanode material for manufacturing DSSCs, primarily attributed to its simultaneous visible light and ultraviolet absorption capability. In addition, its energy bandgap is 3.2 eV with an acceptable electron mobility of 0.4 cm²/Vs, contributing to its chemical stability, high conversion efficiency, non-toxicity, and low cost [6, 7].
When sunlight photons strike the solar cell, electrons located in the dye’s highest occupied molecular orbital (HOMO) become excited and are promoted to its lowest unoccupied molecular orbital (LUMO). Following this excitation, the electrons are transferred to the conduction band of TiO₂, a process driven by the energy alignment, as the conduction band of TiO₂ is positioned at a lower energy level than the dye’s LUMO. This electron transfer process highly influences the efficiency of the solar cell, as photocurrent exhibits a direct proportionality to the number of electrons transferred. One of the key factors that diminishes the DSSC efficiency is the charge carrier recombination. Specifically, the elevated electron from the dye to the conduction band of the semiconductor recombines with the holes in the electrolyte. This phenomenon is commonly referred to as dark current. In this context, the photoanode assumes a crucial role in facilitating the efficient separation and transport of electrons towards the external contact (i.e., the FTO) [9]. Consequently, the overall PCE is adversely affected by factors such as electron-hole recombination, limited light harvesting efficiency, sluggish electron transfer rates within the photoanode, and dark current [10]. To mitigate these electron losses, researchers have investigated a diverse range of strategies, encompassing the synthesis of semiconductors with distinct morphologies [8], the utilization of semiconductor composites [7]–[9], and the doping of semiconductors with metallic or non-metallic atoms [14], [15], [16], [17].
Doping, a relatively straightforward technique, offers a means for modifying the physical and optical properties of materials. Even with the incorporation of minute amounts of dopant, significant enhancements in PCE can be achieved. To ensure optimal dopant selection for TiO2, it is crucial to minimize the disparity between the ionic radius of the Ti4+ ion and that of the dopant, thereby preventing lattice distortion [5]. 
Doping TiO2 with lanthanide elements, such as Pr, Eu, Fe, Ce, Nd, and Yb, and their corresponding oxides, has been shown to increase the separation of electron-hole and its visible light absorption [18]. This visible light activity is crucial from an energy perspective [19]. In addition, previous studies have reported a delay in the recombination of electron-hole in TiO2 nanostructures. From the lanthanides group, cerium oxides show appropriate catalytic and optical properties, along with redox potential for Ce4+/Ce3+ couple [20]. Ce doping in TiO2 not only red-shifts the absorption spectra but inhibits the phase transformation from anatase to rutile [20]. Cerium ions (Ce4+/Ce3+) are unique p-type semiconductors characterized by a wide band gap, multiple electron energy levels, and variable valence states [21]. The partially filled f-orbitals of cerium impart distinctive properties to TiO2. They can reduce the band gap energy of TiO2 by introducing impurity states below the conduction band, even at low doping concentrations (<1%) [22]. 
Furthermore, doping TiO2 with non-metals, for example N [23, 24], C [25], B [26], S [27] and F [28, 29], has been demonstrated to narrow the bandgap and enhance light absorption in the visible region. While numerous studies have investigated TiO2 doping with either transition metals or non-metals individually, there are limited research studies exploring the combined effects of these dopants in TiO2 for photovoltaic applications [30]. In this study, we doped TiO2 with Ce as a transition metal and N as a non-metal for improving photovoltaic performance. The influence of the Ce/N co-doped TiO2 photoanodes on the photovoltaic properties of DSSCs has been also examined. 
Numerous approaches such as electrochemical, hydrothermal, sonochemical, microwave, and sol-gel methods can be applied for synthesizing and modifying TiO2. The sol-gel technique is particularly notable among these methods owing to its straightforward approach and its capacity to achieve precise regulation of crystallite size, surface area, phase composition, and morphology [31].
This research investigates doping of cerium into TiO2 nanopowder and its effect on the physical characteristics of the material and the electrical behavior of DSSCs. For this purpose, Ce-doped TiO2 (TC) nanopowders were synthesized via sol-gel method. Their optical properties, structure, electrical properties, and dynamics were analyzed through various analytical techniques. After that, photovoltaic cells were assembled using Ce-doped, N-doped, and their co-doped TiO2 to evaluate the photovoltaic properties of the cells.

MATERIALS AND METHODS
Cerium-doped TiO2 (CexTi1-xO2) nanopowders were prepared by the sol-gel technique. The starting reagent materials were cerium (III) nitrate hexahydrate (Ce(NO3)3.6H2O, Aldrich), titanium tetra-isopropoxide (TTIP, ≥97.0%, Sigma Aldrich), acetic acid (99%, Sigma Aldrich), ethyl alcohol (95.0%, Sigma Aldrich), isopropyl alcohol (IPA, ≥98%, Sigma Aldrich), PEG (Poly ethylene glycol, Aldrich), and distilled water (DI).

Synthesis of pure TiO2 and Ce-doped TiO2 nanopowders by sol-gel process
In a typical procedure, a mixture of 3.2 mL of TTIP and 9.5 mL of IPA was prepared under stirring for 1 hour, followed by the addition of 24 mL ethyl alcohol and 10 mL acetic acid. The mixture was stirred for another 1.5 hours to ensure homogeneity. The color of mixture changed from transparent to white after the addition of acetic acid. After drying the obtained products in an oven at 90°C for 12 hours, calcination was performed at 450 °C for 4 hours to achieve a homogenous particle size [32], [33]. Ce-doped TiO2 powders with varying Ce concentrations of 3-10 mol% (TC3-10) were synthesized and labeled according to Table 1.

TiO2 paste preparation
TiO2 pastes were prepared by using the TC0 and TC3-10 nanopowders. First, the synthesized TiO2 powder was heated up to 400oC for 30 min to remove the absorbed moisture and organic impurities. Then, 1g of the powder was mixed with 2ml of PEG (Poly ethylene glycol) and 4ml of ethanol using magnetic stirring at 300 rpm to obtain a homogenous and viscous mixture. This paste was then used for TiO2 doctor blade printing on FTO substrates.

Characterization
Optical properties of the TC0 and the TC3-10 nanoparticles were obtained using diffuse reflection spectroscopy (DRS) with an S-4100 SCINCO spectrophotometer, while photoluminescence (PL) spectra were recorded on a Perkin-Elmer spectrophotometer. The crystalline structure of the synthesized TiO2 and TC3-10 nanoparticles were studied using an X’Pert PRO MPD X-ray powder diffractometer (XRD). The bonding environment of all samples was evaluated through Fourier-transform infrared spectroscopy (FTIR) using a Thermo Avatar FTIR system covering a wide range from 4000 to 500 cm-1. Raman shift of the samples was measured in the 100-3500 cm-1 range using a confocal Raman spectrometer (Horiba Jobin Yvon HR800) which uses a 632.8 nm laser. The morphology and particle size of the samples were determined using a field emission scanning electron microscope (FESEM, Zeiss Sigma HV 300-S) and an energy dispersive X-ray spectroscopy (EDS, Aztec Oxford) system. The electrical properties were measured under AM1.5 irradiation intensity, which corresponds to a standard solar spectrum. The electron dynamics in the DSSC was examined using electrochemical impedance spectroscopy (EIS, ZVIE 5MP).

RESULTS AND DISCUSSION 
XRD patterns 
Fig. 1 presents XRD patterns obtained for the TC0 and TC3-10 samples. The peaks at 25.68°, 38.28°, 48.43°, 54.38°, 55.39°, 63.10°, 69.33°, 70.62° and 75.61° correspond to (110), (400), (020), (510), (121), (321), (611), (022) and (710) diffraction planes in anatase TiO2 phase according to JCPDS card No. 96-101- 0943. The introduction of Ce creates extra peaks at 29.17°, 33.44° and 44.95° corresponding to (010), (011) and (110) planes of anatase TiO2 according to JCPDS card No. 96-900- 8492. Ce-doped TiO2 exhibited similar peaks but with lower intensity, which have also shifted slightly to higher angles mainly due to the interstitial replacement of Ce ions for Ti ions [34]. This doping may suppress the transformation of anatase phase to rutile or brookite [35]. The crystalline size and the lattice parameter are also affected by doping of Ce. Since the ionic radius of Ce³⁺ (1.03 Å) lies between that of Ti⁴⁺ (0.68 Å) and O²⁻ (1.32 Å), it is more likely that Ce³⁺ occupies interstitial positions [36]. A mixture of phases can be seen for higher Ce contents (TC7 and TC10) [37]. The observed slight shift in the diffraction peaks after Ce doping indicates lattice distortions induced by Ce doping. The structural parameters measured from XRD patterns are summarized in Table 2, confirming the successful synthesis of both TC0 and TC3-10 photoanodes.

FTIR spectra
According to Fig. 2, FTIR spectra confirm the high purity of the samples. The spectra exhibit weak peaks at 2920 cm-1, 2850 cm-1, and 2340 cm-1 associated with the vibration of organic groups such as carboxylate, hydroxyl, and alkane. The peaks corresponding to the stretching vibration of C-H in alkane groups were not completely removed even after washing with ethanol and DI water. The strong peak at 537-561 cm-1 represents Ti-O and Ce-O bonds [38]. This study indicates that samples are of good quality and are suitable for further characterization.

Raman Spectroscopy

Characteristic peaks of anatase TiO2 can be observed at 148, 400, 521, and 641 cm-1 according to Raman spectra shown in Fig. 3 [39]. In doped TiO2 samples, these vibrational modes are shifted to higher wavenumbers compared to TC0. Lack of additional Raman bands relating to the dopants, combined with the observed peak shifts, strongly suggests uniform doping and distribution of dopant ions within the titanium sites [40]. The observed shifts in Raman peak positions, reductions in peak intensity, as well as the broadening of the symmetric vibration peak of O-Ti-O are indicative of several events: reduced crystallite size, grain boundary formation, and potential oxygen deficiencies within the TiO2 lattice [41], [42]. 

FE-SEM and EDS analyses 
The morphology of the particles was almost the same irrespective of the Ce content, as shown in FE-SEM images (Fig. 4). Since the particles were agglomerated, mechanical grinding was used for preparing TiO2 paste. The utilization of nanocrystalline particles in the solar cell enhances electron/hole transfer, thereby avoiding the need for any additives. In addition, the EDS spectra (Fig. 5) confirmed the stoichiometric ratio of Ce, Ti, and O, with no unwanted impurities, confirming the high purity of the doped samples. 

Optical properties
DRS analysis 
DRS spectra of TC0 and TC3-10 samples at room temperature in the wavelength range of 200-800 nm are shown in Fig. 6. The reduction of diffuse reflectance intensity suggests that the light absorption in the visible spectrum is increased. In the case of TC0, electron transitions between Ti 3d and O 2p states result in optical absorption [43]. It is evident that the absorption bands shift towards higher wavelengths (red shift) after the addition of the Ce dopant. The band gap energy (Eg) of TC0 and TC3-10 samples was determined using Kubelka-Munk plots, as illustrated in Fig. 7. The energy relating to the indirect allowed transition from the valence band (V.B.) maximum to the conduction band (C.B.) minimum was determined by linearly extrapolating the plot of (F(Rα)hυ)1/2 as a function of photon energy (hυ) [44]. The calculated Eg for TC0, TC3, TC5, TC7, and TC10 samples are 3.61 eV, 3.57 eV, 3.48 eV, 3.89 eV, and 3.97 eV, respectively.
The incorporation of impurities or the presence of vacancy defects within the band gap (between the V.B. and C.B.) of TiO2 can lead to modifications in its bandgap energy [45]. These changes provide sites for cerium doping, leading to the creation of intermediate energy levels associated with the Ce 4f states within the TiO2 bandgap. This phenomenon can significantly influence the photocatalytic and electrocatalytic activities of the material [46], [47]. The Eg of TC0 is 3.61 eV, which reduces to 3.48 eV upon increasing the Ce content to 5 mol% (TC5 sample). Researchers have reported that the creation of electron-hole pairs under the exposure of visible light is significantly facilitated in the presence of Ce 4f levels [48], [49]. 

PL analysis
Photoluminescence (PL) test was employed to investigate electron-hole pair recombination in TC0 and TC3-10 samples, as depicted in Fig. 8. All samples exhibited a similar PL behavior. This similarity in PL signals is often observed because of the half- or fully-filled outer electron configuration of the dopant in its stable chemical state [50]. In the case of TC3-10, the electron configuration of the filled outer shell of Ce4+ ([Xe] 4f0 5d0 6s0) does not significantly alter the spectral shape. The PL spectra of all samples displayed a broad and intense emission band in the 350-450 nm range upon excitation at 254 nm, which can be attributed to excitonic luminescence [51]. A prominent peak observed around 400 nm is associated with radiative transitions between bands involving photoexcited electrons [52]. Additionally, the band at 464 nm is characteristic of rutile TiO2 [53]. Despite the similar spectral shapes, the PL intensity changed considerably, mainly due to the existence of oxygen vacancies and surface defects, which in turn are dependent on the ability of the dopant to capture electrons [50]. With increasing the Ce content, the formation of Ti-O-Ce bonds reduces the concentration of oxygen vacancies, hence decreasing the PL intensity [52].
Studies have shown that surface defects and oxygen vacancies are not capable of binding electrons for the formation of excitons. Instead, they are captured by Ce⁴⁺ ions, resulting in the reduction of Ce⁴⁺ to Ce³⁺ [50], and hence a decrease in PL intensity. The lower PL intensity is often associated with reduced recombination of photogenerated charges [54]. This observation suggests that increasing the Ce content can decrease the recombination rate of electron-hole pairs, thereby enhancing the potential for improved photovoltaic properties. 

Photovoltaic properties
Current density (J) as a function of voltage (V) and the power measurements of the DSSCs fabricated based on TC0 and TC3-10 samples are shown in Fig. 9 and Fig. 10, respectively, to clarify the effect of Ce doping on the photovoltaic performance of DSSCs. The tests were conducted by using a sun simulator under AM1.5 standard condition. Table 3 lists the associated photovoltaic parameters for these measurements. According to Fig. 9, the incident photoelectric conversion efficiency (IPCE) is as follows: TC5 > TC3 > TC0 > TC7 > TC10. Because of the impact of Ce doping, TC5 has 29% more efficiency than TC0 with short circuit current density (JSC) of 10.3 mA/cm2, open circuit voltage (VOC) of 545 mv, energy conversion efficiency (η) of 5.613%, and fill factor (FF) of 47.4%. The greater electron density is a result of improved photo-injection and effective photoexcitation in TiO2. Table 4 compares the results of this investigation with previously reported studies. 
The electrical properties of nitrogen-doped TiO2 samples (TN3, TN5, and TN7) and Ce and N co-doped TiO2 samples (TN3C5, TN5C5, and TN7C5) were also investigated and compared with those of the TC5 sample, as shown in Fig. 11 and Fig. 12. The corresponding photovoltaic parameters of these samples are listed in Table 5. DSSCs employing the co-doped TiO2 exhibit significantly enhanced energy conversion efficiency (6.51) and fill factor (67.4) compared to TC0 (η =4.45, FF=43) and TC5 (η=5.613, FF=47.4). While nitrogen doping also enhanced efficiency by 22%, cerium doping proved to be more beneficial. Remarkably, the co-doped cell exhibited a 46% increase in efficiency. The fill factor followed a similar trend, reaching 47.4% for the optimized cerium-doped cell (TC5), 64.5% for the optimized nitrogen-doped cell (TN5), and an impressive 67.4% for the co-doped cell (TN3C5), confirming the positive interplay between cerium and nitrogen doping. The increased electric current towards the cathode is likely attributed to the introduction of intermediate levels by both dopants.

EIS analysis
In DSSCs, electron-hole recombination poses a significant challenge to device performance. Excited dye molecules contain holes which can recombine with electrons, either with the ionized species of electrolyte or in the semiconductor matrix [60]. These charge transfer processes within the DSSC can be detected by using electrochemical impedance spectroscopy (EIS). Higher the resistance to charge recombination, higher is the cell efficiency. The radius of the semicircle in the impedance spectrum is indicative of the resistance to charge recombination, with a larger radius signifying lower recombination resistance [61]. EIS was employed to investigate the electron transport kinetics and internal resistance of TiO2 films in DSSCs. The charge recombination resistance at the interface of TiO2 and electrolyte (R3) can be determined according to impedance response. The larger semicircle observed in the low-frequency region reflects the effects of electron accumulation and transport within the TiO2 photoanodes. In contrast, charge transfer resistance at both electrolyte/Pt counter electrode and FTO/TiO2 interfaces (R2) can be detected by the smaller semicircle in the high-frequency region [62].
Nyquist plots, developed under short-circuit conditions at light intensities below 100 mW/cm², are presented in Fig.13. Within the frequency range of 100 kHz to 1 Hz, two distinct semicircles are observed in the complex plane plot, representing three-time constants. The corresponding parameters are summarized in Table 6. According to Fig. 13, the second semicircle of TC0 is larger than that of TC5 which indicates a decrease in R2 resistance. This reduction in interfacial resistance leads to improved efficiency and an overall decrease in internal resistance. Equation 1 is used to estimate the photoelectrons lifetime (τe) [63]:

τe =1/2π fmax                                                         

where fmax is the maximum frequency of semicircle at low frequencies. 
Fig. 14 reveals that the peak frequency increases by increasing the dopant concentration, except for TC5. This increase in frequency peak directly translates to a decrease in electron lifetime. Conversely, TC5 exhibits a significantly longer lifetime (15.23 ms), facilitating electron diffusion and transfer due to an increase in diffusion length.
The extended electron lifetime observed in TC5 indicates a reduced recombination rate, consequently leading to higher VOC. Improved electron transport in the TiO2 after doping can be caused by a combination of factors; bandgap reduction and lattice distortions. These factors may lead to the accumulation of oxygen vacancies, thereby effectively trapping photogenerated holes. This mechanism contributes to the prolonged lifespan of both photo-excited electrons and holes [64], a finding which is consistent with the observed photovoltaic characteristics.

CONCLUSION 
This study focused on the sol-gel synthesis of TiO2 nanoparticles with different Ce concentrations and their characterization using various techniques. These nanoparticles were used for the fabrication of photoanodes and their performance in DSSCs was compared with that of N-doped and Ce,N co-doped TiO2 photoanodes. All the synthesized samples were predominantly consisted of anatase phase, while the morphology changed from nanosheets to nanopowders upon cerium doping. The bandgap of the TiO2 nanomaterial fell within the reported range (3.48-3.76 eV), confirming the successful bonding between Ti and Ce. Raman spectroscopy further corroborated the presence of these elemental bonds. According to PL spectra, the peak intensity increased initially at low Ce concentrations, followed by a decrease in the PL intensity by increasing the dopant concentration. Photovoltaic test of DSSC showed a 26% increase in performance for the TC5 and 46% increase for the TN3C5 compared to the TC0 sample. The electrical properties obtained for TC5, TN3, and TN3C5 samples were as follows: VOC of 545 mv, 640 mv, and 650 mv, Jsc of 10.3 mA/cm2, 10.97 mA/cm2, and 10.23 mA/cm2, FF of 47.4%, 64.5%, and 67.4%, and η of 5.613%, 5.44%, and 6.51%, respectively. EIS analysis was utilized to investigate the electron dynamics within the DSSC. The Nyquist plot revealed that the TC5 sample exhibits lower resistance, characterized by Rs of 23.14 Ω, R1 of 4.61 Ω, and R2 of 13.84 Ω, along with the longest electron lifetime recorded at 15.23 ms. This study demonstrates that co-doping of Ce and N is an effective approach for enhancing the DSSCs photovoltaic performance. The observed synergistic effect between Ce and N doping highlights the potential of this approach for developing high-efficiency, low-cost solar cells. Future studies should focus on the optimization of doping levels, examination of various co-doping strategies, and evaluation of the long-term stability of the devices.

CONFLICT OF INTEREST
The authors declare that there is no conflict of interests regarding the publication of this manuscript.