N, La Co-Doped TiO 2 for Use in Low-Temperature-Based Dye-Sensitized Solar Cells Dye-sensitized solar

In this survey, N, La co-doped TiO 2 nanocrystals are synthesized through simple sol–gel method. Results reveal that both elements of N and La are introduced successfully into the structure of TiO 2 and exist as the form of La (Ti–O–La) bonds, N (N–Ti–O) bonds and N (Ti–O–N) bonds by several characterization analysis, thereby resulting in the formation of new impurity energy level between the forbidden bands, and great enhancement of light absorption ability in visible light region. These resulting samples are fabricated for low-temperature-based dye-sensitized solar cells (DSSCs), and the highest temperature of the whole process is 120 ◦ C. EIS, Bode plot and I-V analysis are employed to measure photovoltaic performance of DSSCs. The photoelectric conversion efﬁciency ( η ) of DSSCs with co-doped TiO 2 materials as the photoanode is 5.33%, and when compared to pure TiO 2 -based cell (4.02%), the η is increased by 32.6%. 6 octahedron higher absorption coefﬁcient in the visible-light the optical properties of N, La-codoped TiO . we synthesize N, La co-doped TiO 2 powder using and lanthanum as the dopants. The resulting doped samples are applied in the fabrication of the photoanodes of low temperature-based DSSCs. The highest temperature of the whole process was only 120 ◦ C. To the best of our knowledge, this is also the ﬁrst report for the fabrication of low temperature-based DSSCs employing N, La co-doped TiO 2 nanocrystals as photoanodes, whose performance was compared with DSSCs fabricated with pure TiO 2 , La-doped TiO 2 and N-doped TiO 2 samples. As expected, N, La co-doped TiO 2 -based solar cell exhibited higher Jsc than pure and single element-doped TiO 2 -based DSSCs.

Dye-sensitized solar cells (DSSCs) have attracted a lot of research interest due to the low-cost, easy fabrication, good stability, and high photoelectric conversion efficiency. 1 However, the traditional fabrication procedure of DSSCs requires not only the heavy FTO substrate, but also a high temperature calcination process at about 500 • C. These will limit the large area production and commercial application of DSSCs. In order to overcome these problems, flexible DSSCs with their incomparable advantages have attracted more interest of scientists, 2 compared with the traditional high temperaturebased DSSCs. By using the flexible substrate, the high temperature calcination process is not needed, which can save energy and reduce the production cost. More importantly, a wide range of flexible substrates can be applied in the fabrication of flexible DSSCs. Generally, the transparency of the solar cell substrate need to be more than 80%, and thus only FTO can be employed in the high temperature-based DSSCs. For the low temperature-based DSSCs, the substrate can involve FTO, ITO-PET (ITO-coated pol(ethylene terephthalate), ITO-PEN (ITO-coated polyethylene naphthalate), Ti foil and Ti meshes, [3][4][5] ZnO/Al thin film with high transmittance, 6 etc. It is believed that the cell based on the flexible substrate should possess the merits of being light weight, foldable and easy to carry, and thus can be used in portable electronic products. Furthermore, the cell preparation process is simple and suitable for industrial roll-to-roll mass production. It is known that the fabrication of photoanode at low temperature is essential for these flexible DSSCs. The paste employed for the generation of photoanode does not have to undergo a high temperature (about 500 • C) annealing process, since no organic materials are introduced. After a slight agitation, the paste could be dried at low temperature(<150 • C). Unfortunately, it is reported that the short-circuit current (J sc ) is low without the high temperature calcination process, which is the main reason for the poorer efficiency realized for the low temperature-based DSSCs, compared to that of the traditional high temperature-based DSSCs. To enhance the J sc of low temperature-based DSSCs, more efforts have been devoted to the strategies of fabricating the photoanode of DSSCs, such as high pressure method, 7,8 hydrothermal procedure, 9,10 and low temperature sintering process, 11 ultraviolet irradiation technology, 12 and microwave heating process 13 and so on. By employing these strategies, the photoelectric conversion efficiency of the low temperature-based soalr cells is increased. In particular, by using the compression method which greatly increases the connection between the TiO 2 photoanode particles, the short circuit current is significantly improved. So far, the highest efficiency 7.6% for the low temperature-based solar cell is z E-mail: baozhang@tju.edu.cn; yqfeng@tju.edu.cn realized by this method. 8 However, the efficiency of the cell prepared by this method is still far from the practical application. Meanwhile, it is a pity that the most efforts of scientists are only devoted to the strategies of fabricating the photoanode of DSSCs, few is attempted to improve the Jsc values by employing other strategies in the field of low temperature-based DSSCs. It is well known that doping TiO 2 with nonmetal elements (such as nitrogen, carbon, sulfur, boron and phosphorus) could improve the light absorption in the visible region and greatly enhance the short circuit current of DSSCs. 10 And among those doping nonmetal elements, N is found to be more promising. 14 It was reported that TiO 2 nano-material doped with rare-earth metal element could also improve the short circuit current of DSSCs by reducing the recombination rate of photogenerated electron-hole pairs and enhancing separation efficiency of charge carriers. 15 As a rareearth metal element investigated widely, lanthanum is effective to improve the photoelectric activity of TiO 2 . 16 Actually, there are few reports about doped TiO 2 for use in low temperature-based DSSCs. Ting et al. 17 synthesized TiO 2 mesoporous beads doped with N by using a microwave-assisted method and fabricated low temperaturebased DSSCs, the η was found to be 4.81%, which was affected by Jsc. Chen et al. 18 synthesized N-doped TiO 2 samples by wet method with urea as nitrogen source and applied them to low-temperature DSSCs, and the maximum value of η reached 5.18%. Qin et al. 19 prepared N-doped TiO 2 photoelectrodes on titanium sheet at low-temperature by micro-plasma oxidation method in electrolyte with (NH 4 ) 2 SO 4 and NH 3 · H 2 O as nitrogen sources, respectively. However, the efficiency was found to be only 0.104%. An'amt, et al. 20 synthesized Bismuth-TiO 2 nanocubes via a facile sol-gel hydrothermal method at low temperature, the η was found to be 2.11%. Salavati-Niasari et al. 21 deposited CdS on TiO 2 using successive ion layer adsorption and reaction, the η was 4.50%.
Herein, in our survey, we aim at enhancing the Jsc values significantly by combining rare-earth metal elements of La and N doped in TiO 2 materials to improve the efficiency of the low temperaturebased DSSCs. Recently, Wang et al. 22 studied the optical properties of N, La co-doped anatase TiO 2 via calculation, which is based on the density functional theory. They reported that after N and La atoms are codoped into anatase TiO 2 , there are some significant changes in the local electronic density. When an O atom is replaced by an N or La atom, the charge distribution of the system is correspondingly changed. Thus, the electrons of higher electronic density can easily obtain smaller energy and enter into a lower energy region when the holes are left in their original positions. which are beneficial for the separation of photoexcited electrons and holes and useful for enhancing photo activity. 23 Simultaneously, the dipole moment of a distorted octahedron can induce the local internal fields after doping N,La element and a larger dipole moment of TiO 6 octahedron causes higher absorption coefficient in the visible-light region. 22 All of these could be helpful for the optical properties of N, La-codoped TiO 2 . Therefore, we synthesize N, La co-doped TiO 2 powder using urea and lanthanum nitrate as the dopants. The resulting doped samples are applied in the fabrication of the photoanodes of low temperature-based DSSCs. The highest temperature of the whole process was only 120 • C. To the best of our knowledge, this is also the first report for the fabrication of low temperature-based DSSCs employing N, La co-doped TiO 2 nanocrystals as photoanodes, whose performance was compared with DSSCs fabricated with pure TiO 2 , La-doped TiO 2 and N-doped TiO 2 samples. As expected, N, La co-doped TiO 2 -based solar cell exhibited higher Jsc than pure and single element-doped TiO 2 -based DSSCs.

Fabrication of different TiO 2 nano-crystalline particles.-Typ-
ically, N, La co-doped TiO 2 nano-crystalline particles (NLa/TiO 2 ) were successfully fabricated by the simple wet method. In detail, the desired amount of La(HNO 3 ) 3• 6H 2 O (0.17 g), urea (0.12 g) and 10 mL glacial acetic acid were dissolved in 90 mL deionized water to form solution A at room temperature; 34.03 g tetrabutyltitanate was gradually added into absolute ethanol with constant stirring for 1 h to obtain a pale yellow solution B. Then the solution B was added dropwise to solution A in 60 min with vigorous stirring and after addition the resultant mixture was kept stirring for another 2 h, which was followed by aging for 48 h. The precipitates were centrifuged, washed with water and ethanol 3 times respectively, dried, grinded, and calcined at 500 • C for 3 h in oven with a heating rate of 5 • C /min. Finally, the white powders were obtained.
For comparison, La-doped TiO 2 and N-doped TiO 2 systems were also fabricated according to the same procedure without CO(NH 2 ) 2 and La(HNO 3 ) 3• 6H 2 O, respectively. Meanwhile, La-doped, N-doped and N, La-codoped systems are optimized, in which the optimal doping amount of La and N is 0.4% and 4% (the mole percent of atoms (La:Ti, N:Ti), respectively. Finally, these optimized samples were designated as La/TiO 2 and N/TiO 2 .The pure TiO 2 nano-crystalline particle was also synthesized as the same process without adding dopants.

Characterization techniques of different TiO 2 nano-crystalline
particles.-FE-SEM (S-4800, Japan Hitachi Ltd) was used to measure the microstructure and size of particles. The crystalline phase of the fabricated samples was identified by X-ray diffraction analysis (XRD, D/MAX-2500, Japan, Rigaku) using Cu Ka radiation (λ = 0.15406 nm), a 40 mA tube current, 40 kV voltage and a 2θ range from 10 to 90. The BET specific surface area analysis of the samples was determined by the Brunauer-Emmett-Teller (BET) method (Tristar 3000, Micromeritics, USA). The doping method and the amount of dopants were performed using X-ray photo emission spectroscopy (XPS, PHI-1600, PE, USA) equipped with Mg Ka (1253.6 eV) as the excitation light source with the power at 300.0 W and and the spot size was 0.8 mm 2 .

Fabrication of different TiO 2 based photoelectrodes at lowtemperature and DSSCs assembly.-Low temperature TiO 2 pastes
were fabricated as the following method: 0.8 g of the above fabricated samples, 0.2 g 200 nm TiO 2 , 2 g hydrothermal cement 24 and 1 g acetic acid were dispersed in 3 g isopropanol and stirred for 2 h in the ice bath, then NH 4 OH was added dropwise and pH was adjusted to about 4.7. The resultant mixture was stirred and treated with ultrasound method for 2 h, respectively. Finally, the slurry was fabricated after slightly milling, which was applied to FTO substrate with a doctor-blade technique to obtain the photoanode. Afterwards, the photoanode was dried to eliminate the volatile compounds at 120 • C for 30 min and the stable low-temperature photoanode was obtained.
Next, the above photoanode was immersed in a dye solution N719 for 18 h to absorb the dye molecules adequately. The DSSCs were assembled to form a sandwich-type cell using the obtained photoanodes, an electrolyte and a Pt counter-electrode. 0.6 M DMPII, 0.03 M I 2 , 0.5 M 4-TBP, 0.1 M GuSCN in acetonitrile and valeronitrile (at a volume ratio of 85/15).

Results and Discussion
SEM and TEM analysis.-The morphology of the as-prepared N, La co-doped TiO 2 samples observed by SEM and TEM images are given in Figs. 1a and 1b. As can be seen from the SEM photograph, many spherical nanoparticles are uniformly distributed and the particle size is about 20 nm. Meanwhile, there are a few agglomeration particles, because those spherical nano-crystallite TiO 2 are connected compactly. Fig. 1b shows that although there are some aggregation of the prepared N, La co-doped TiO 2 sample, most nano-crystallite TiO 2 particles are separated from each other and the dispersity of the particles is good. Fig. 2 shows the X-ray diffractogram of pure TiO 2 , La/TiO 2 , N/TiO 2 , N, La/TiO 2 samples in the range of 20-80 (2θ). It can be found that the crystal phase of pure TiO 2 -based sample contained both anatase and rutile. However, N-doped TiO 2 , La-doped TiO 2 and N, La co-doped TiO 2 -based samples do not exhibit any additional phase except for anatase. Thus, we can see that doping N or La can greatly favors the formation of anatase and effectively suppresses the formation of rutile, because the introduction of dopant ions can change the pH of the hydrolysis of TiO 2 sol and promote the formation of anatase phase. Furthermore, it is noteworthy that we do not see the diffraction peak of the dopants in the XRD pattern, which is caused by the small amount of the dopants and their incorporation into TiO 2 lattice or adhesion to the interstitial site. 15 And according to Pauling's principle, 25 it is not possible for La 3+ to replace Ti 4+ to enter into the titania lattice because of its larger ion radius (0.108 nm) compared with that of Ti 4+ (0.068 nm). The characteristic peaks of Ti-N and La 2 O 3 phases are not observed in the X-ray patterns of doped   Laser Raman spectroscopy analysis.-Laser Raman spectroscopy (LRS) is one of the most effective methods to characterize the near surface structure of inorganic oxides, which is very sensitive to the response of the crystal structure. 27,28 Fig. 3 is the Laser Raman spectroscopy (LRS) of N, La co-doped TiO 2 and N-doped TiO 2 sample, and N-doped TiO 2 sample is taken as a reference. As shown in Fig. 3, several Raman bands of the two samples are both observed at about 144, 196, 396, 516, and 638 cm −1 , which shows that N, La co-doped TiO 2 and N-doped TiO 2 samples are both anatase phase. At the same time, we can also see that the LRS band of N, La co-doped TiO 2 material, in comparison with N-doped material, occurred a slight redshift, and the peak pattern was significantly reduced, which shows that the LRS band of N, La co-doped TiO 2 samples is smaller and disper-  sion is higher. Therefore, the Laser Raman spectroscopy exhibits that not only N, La co-doped TiO 2 samples is much smaller, but also the dispersion is better. The results agree well with those obtained from SEM, TEM and XRD images.

XRD analysis.-
BET analysis.-Meanwhile, the specific surface area of the four samples was measured by the BET analysis. As shown in Table I, there are 51.04, 106.83, 117.79 and 134.56 m 2 g −1 for pure TiO 2 , La-doped TiO 2 , N-doped TiO 2 and N, La co-doped TiO 2 powders, respectively. N, La co-doped TiO 2 sample possesses the biggest specific surface area. The reason is that N, La co-doped TiO 2 samples has not only the smallest particle size, but also the best dispersion.

UV-vis diffuse reflectance spectrum analysis.-UV-vis
diffuse reflectance spectrum of pure TiO 2 , La/TiO 2 , N/TiO 2 , and N, La/TiO 2 particles are shown in Fig. 4. Obviously, the light absorption ability of N, La co-doped TiO 2 nano-particles is the strongest at the wavelength range from 300 to 650 nm. Especially, at 400-600 nm, the results demonstrate that co-doping with elements of La and N plays a synergistic effect in the enhancement of the light absorption, which is caused by the narrowing bandgaps after doping La and N elements and  increasing light absorption in visible light region. Generally, TiO 2 is a material with large bandgap and only absorbs ultraviolet light, however, the ultraviolet light only accounts for 2-4% of the solar light that can reach the earth's surface, while visible light has accounted for 48%. Therefore, narrow bandgaps can greatly increase the light harvesting yield of the solar light that can reach the earth's surface. To confirm that the bandgaps are narrowed after co-doping, the bandgap energy of the samples was calculated through the Kubelka-Munk equation, 29 which are 3.16, 3.01, 2.90 and 2.83 eV for pure TiO 2 , La-TiO 2 , N-TiO 2 , and N, La co-doped TiO 2 powders, respectively. It is true that the highest light absorption of N, La co-doped TiO 2 sample is due to the smallest bandgap energy. Namely, above the valence band for the substitutional nitrogen and below the conduction band for La 3+ doping can form a dopant level, which could decrease the bandgap of TiO 2 . 30 The doping of La 3+ will cause lattice deformation and form vacancies, which will probably results an impurity state of titania. 31 The existing of impurity state can narrow the bandgap and improves the light absorption.
XPS analysis.-X-ray photoelectron spectroscopy (XPS) is performed to determine the surface component and chemical state of the samples. Fig. 5a is the full scanned spectra of N, La co-doped TiO 2 powder in the range of 0-1100 eV. Obviously, N and La element are introduced successfully into the sample. Fig. 5b is the binding energy peaks for N1s of N, La co-doped TiO 2 sample at about 399.46 eV and 401.93 eV. The first peak at 399.46 eV is attributed to form the N-Ti-O bond in TiO 2 lattice, 32 which suggests that some N atoms are doped in TiO 2 lattice by substituting the oxygen atom. Another strong peak at 401.93 eV is assigned to the host in an interstitial site directly linked to lattice O and forms Ti-O-N, Ti-O-N-O. 33 Therefore, we can conclude that two forms of N elements (substituted N and interstitial N) coexisted in the N, La co-doped TiO 2 sample. Fig. 5c, compared with pure TiO 2 powder, the Ti2p peak of N, La co-doped TiO 2 sample showed a slight shift toward the higher binding energy. Meanwhile, Ti2p3/2 peak of N, La co-doped TiO 2 sample is at 458.67 ev, and that of pure TiO 2 samples is at 458.45 ev, the slight increase in Ti2p3/2 binding energy shows that the electronic interaction of cation with anions in N, La co-doped TiO 2 was considerably different from that in pure TiO 2 . A plausible explanation 34 is that N has a lower electron negativity compared to O, La has a much stronger electropositivity than Ti. As a result, the electron density around the Ti cation is changed by doping N and La. Therefore, this also proved that La is doped into TiO 2 lattice. Fig. 5d shows the La3d spectrum of the N, La co-doped TiO 2 sample, two peaks are presented at 853.89 eV and 837.4 eV, which can be ascribed to La3d3/2 and La3d5/2 photoelectrons, respectively. However, the standard peak is normally positioned at 851.8 eV and 834.9 eV. 35 The increase of the peaks suggests that La is incorporated into the lattice of TiO 2 and Ti-O-La bond is formed.
Photovoltaic performance studies. -Fig. 6 shows the photovoltaic performances of the current-voltage characteristics and the corresponding photovoltaic characteristics such as the short circuit current (J sc ), open-circuit photovoltage (V oc ), fill factor (FF) and photoelectric conversion efficiency (η) are summarized in Table III. It can be seen In order to find out the reasons for the increase of the short circuit current, the amount of dye adsorption of the four types of cells are measured and shown in Fig. 7. Meanwhile, the values are collected in Table III. N, La co-doped TiO 2 -based cell absorbs the most dye (7.11 × 10 −7 mol•cm −2 ), while the pure TiO 2 -based cell absorbs the least (3.21 × 10 −7 mol•cm −2 ). The results are ascribed to the crystalline sizes of the samples. The size of N, La co-doped TiO 2 sample is the smallest, while that of pure TiO 2 is the biggest. The smaller crystalline size, the larger the surface areas, and the more dye loading amount. N, La co-doped TiO 2 -based DSSCs has greatest amount of dye molecules absorbed, which led to the highest short circuit current.
The charge transport and the electrons lifetime were investigated by EIS analysis to determine another reason for the enhanced Jsc value when treated with N, La co-doping, La-doping and N-doping, respectively. Typically, the EIS Nyquist plot of DSSCs has three semicircles with increasing frequency. The three semicircles correspond to the diffusion resistance within the electrolyte at low frequency, electron transport resistance at the oxide/electrolyte interface at middle frequency, and the resistance of redox reaction at the platinum counter electrode at high frequency. However, in our survey, only two semicircles can be observed, and the resistance of diffusion within the electrolyte at low frequency do not show up, which is explained by the limited frequency range of our instrument (from 1 Hz to 1000 Hz under the same conditions with J-V measurement and the open circuit voltage is set as the open applied bias voltage); however, the diffusion process is normally observed over the frequency range 0.01 Hz and 0.1 Hz. 36 Fig. 8a is the Nyquist image of the four cells. The detailed EIS parameters are as shown in Table IV. Rs represents the external circuit resistance, R 1 represents the resistance of redox reaction at the platinum counter electrode, R 2 represents the resistance of electron transport at the oxide/electrolyte interface. From Table IV, we can see that the values of Rs, R 1 and R 2 all decrease after doping N, La element, because the sizes become smaller after doping, thus TiO 2 particles and FTO substrate become closer, which increase the inter-connection between the TiO 2 particles or contact with FTO substrate. R 2 has a great influence on DSSCs for different photoanodes. From Table IV, it can be seen that R 2 value of the N, La co-doped TiO 2 -based solar cell is the smallest, which means that N, La codoped TiO 2 -based cell has the smallest electron transport resistance, indicating that the electron transport of N, La co-doped TiO 2 -based DSSCs is faster than that of others. Therefore, N, La co-doped TiO 2 based cell has the highest Jsc. Fig. 8b shows the Bode plots of the four type DSSCs, the first peak at low frequency stands for the lifetime of electrons at the TiO 2 /dye/electrolyte interface, and the value of the electron lifetime are also summarized in Table III. Obviously, N, La co-doped TiO 2 -based DSSC has the longest electron lifetime (19.29 ms), while the pure TiO 2 -based DSSCs has the shortest one (6.10 ms), which also shows that N, La co-doped TiO 2 -based solar cell has higher Jsc.

Conclusions
We synthesized N, La co-doped TiO 2 samples fabricated for lowtemperature DSSCs. The photoelectric conversion efficiency reached 5.33%, with a V oc of 728 mV, J sc of 10.518 mA•cm −2 , and FF of 69.58%. When compared with that of pure TiO 2 -based, La and Ndoped TiO 2 -based cell, the η of N, La co-doped TiO 2 -based cell increased by 32.6%, 16.4% and 12.9%, respectively. The enhanced performance mainly depended on the improvement of J sc , which is attributed to the increase of the dye absorption amount, electron lifetimes, and the decrease of the charge transport resistance after doping non-metal element N and rare-earth metal element La into TiO 2 photoanode. Simultaneously, increasing the light-harvesting was realized by narrowing bandgaps and increasing visible light absorption, which led to high photo current density.