Improving Interfacial Charge-Transfer Transitions in Nb-Doped TiO 2 Electrodes with 7 , 7 , 8 , 8-Tetracyanoquinodimethane

Interfacial charge-transfer (ICT) transitions involved in charge-separation mechanisms are expected to enable efficient photovoltaic conversions through one-step charge-separation processes. With this in mind, the charge-transfer complex fabricated from TiO2 nanoparticles and 7,7,8,8-tetracyanoquinodimethane (TCNQ) has been applied to dye-sensitized solar cells. However, rapid carrier recombination from the conduction band of TiO2 to the highest occupied molecular orbital (HOMO) of TCNQ remains a major issue for this complex. In this study, to inhibit surfacecomplex recombinations, we prepared Nb-doped TiO2 nanoparticles with different atomic ratios for enhanced electron transport. To investigate the effects of doping on electron injection through ICT transitions, these materials were examined as photoelectrodes. When TiO2 was doped with 1.5 mol % Nb, the Fermi level of the TiO2 electrode shifted toward the conduction band minimum, which improved electron back-contact toward the HOMO of TCNQ. The enhancement in electron transport led to increases in both short circuit current and open circuit voltage, resulting in a slight (1.1% to 1.3%) improvement in photovoltaic conversion efficiency compared to undoped TiO2. Such control of electron transport within the photoelectrode is attributed to improvements in electron injection through ICT transitions.


Introduction
Interfacial charge-transfer (ICT) transitions between inorganic semiconductors and π-conjugated organic compounds are characteristic electronic transitions that enable direct photoinduced charge separation.Due to this feature, ICT transitions are applicable to photovoltaic conversions [1].To date, dicyanomethylene-based compounds, such as tetracyanoethylene (TCNE) and 7,7,8,8tetracyanoquinodimethane (TCNQ), form surface complexes with TiO 2 that have been reported to absorb visible light due to ICT transitions from the π-conjugated system to the conduction band of TiO 2 [2][3][4].Although photovoltaic effects due to direct electron injection have been investigated using these surface complexes as photoanodes in photoelectrochemical cells, the photovoltaic conversion efficiencies under AM1.5 illumination are quite low (under 2%) when compared to those of dye-sensitized solar cells (DSSCs) [5].
To date, photovoltaic conversions and electron injections between surface complexes, such as bis(dicyanomethylene) compounds (TCNX) [TCNE, TCNQ, and 11,11,12,1 2-tetracyanonaphtho-2, 6-quinodimethane (TCNAQ)] and TiO 2 , have been studied theoretically by Fujisawa et al. using density functional theory (DFT) on the basis of Marcus theory, which revealed that the structure and formation mechanism of the surface complex need to be considered to control interfacial electronic transitions and carrier recombinations by adjusting the electron affinity of TCNX [6][7][8].In addition, they also demonstrated that carrier recombinations from the conduction band of TiO 2 to the highest occupied molecular orbital (HOMO) levels of these compounds occur quite rapidly, which, as geminate recombinations, are more severe than in DSSCs.Hence, the rapid carrier recombinations of surface complexes hinder the use of ICT transitions in photovoltaic conversion [9] and is the most significant problem faced.
To overcome rapid electron recombination, the properties of surface complexes require further investigation through different approaches that include investigating the interactions between TCNX and modified-TiO 2 photoanodes.Zaban et al. studied the suppression of electron recombination in surface complexes aided by a thin SrTiO 3 coating layer on TiO 2 that improved electron injection and electron transport [10].On the other hand, many researchers reported that the TiO 2 photoanode in a DSSC is one of the most important components affecting photovoltaic performance because it acts as the support for dye molecules as well as the electron-transport region.Furthermore, TiO 2 can, in principal, be n-type-doped to enhance charge collection and electron-transport efficiency within the TiO 2 layer [11].
Among n-type-doped TiO 2 systems, Nb-doped TiO 2 photoanodes have been studied to improve electron conductance and injection because Nb has one more electron than Ti (IV) [12][13][14][15][16][17][18].Recently, Lin et al. reported the influence of TiO 2 doped with Group V-b metal atoms on the photovoltaic performance of dye-sensitized solar cells [18].Although V, Nb, and Ta belong to the same group (V-b) and have one more electron than Ti, DSSCs based on Nb-doped TiO 2 showed the best photovoltaic performance as a result of the creation of donor levels, which increased the concentration of the carriers.According to this study, the charge transport and conductivity for Nb-doped TiO 2 were superior to those of V-and Ta-doped TiO 2 .Therefore, Nb is a superior doping element.
In this study, to inhibit rapid electron recombinations in surface complexes, we examined the effects of Nb-doped TiO 2 electrodes with TCNQ on photovoltaic performance.The effects of different amounts of Nb-doping on the photovoltaic properties of surface complexes were evaluated by means of incident photon-to-current efficiency (IPCE) spectroscopy and by acquiring current-density-voltage (J-V) curves.TiO 2 doped with 1.5 mol % Nb exhibited improved Jsc and Voc values, resulting in a 17% improvement in photoconversion efficiency compared to undoped TiO 2 .

Structural Characterization of Nb-Doped TiO 2
Nb-doped TiO 2 nanoparticles were formed by a hydrothermal method.Figure 1 displays the X-ray diffraction (XRD) patterns of undoped and Nb-doped TiO 2 samples with varying Nb contents.The anatase and rutile phases of TiO 2 are distinct, and the intensity of the peak corresponding to the anatase phase increased with increasing Nb content in the sample, while that of the rutile phase decreased (Figure 1a).Hence, doping the TiO 2 lattice with Nb was observed to enhance the growth of the anatase TiO 2 phase while hindering the formation of the rutile TiO 2 phase [19].The average crystallite size of each phase was calculated by the Scherrer-equation [20]: where D is the crystallite size, k is a dimensionless shape factor that has a typical value of 0.94, λ is the Cu-Kα X-ray wavelength (1.5406 Å), θ is the Bragg angle in degrees, and B is the full-width-at-half-maximum (FWHM) of the peak.The anatase and rutile crystallinities were determined from their (101) and (110) peak heights, respectively; crystallite size are listed in Table S1.Incorporation of the Nb dopant into the TiO 2 structure typically decreased crystallite growth; indeed, the crystallite size of the anatase phase slowly decreased from 10.1 nm to 8.8 nm with increased doping, led by the effect of Nb on nucleation following doping into the TiO 2 lattice.A similar effect of different oxide additives on the TiO 2 phase transformation was reported by Yanagida et al. [19].In addition, the diffraction peaks shifted to lower θ values with increasing Nb content as a result of the larger radius of Nb 5+ (0.64 Å) compared to that of Ti 4+ (0.61 Å) [15], in accordance with the Bragg equation: λ = 2dsinθ (Figure 1b).
Catalysts 2018, 8, x FOR PEER REVIEW 3 of 10 half-maximum (FWHM) of the peak.The anatase and rutile crystallinities were determined from their (101) and (110) peak heights, respectively; crystallite size are listed in Table S1.Incorporation of the Nb dopant into the TiO2 structure typically decreased crystallite growth; indeed, the crystallite size of the anatase phase slowly decreased from 10.1 nm to 8.8 nm with increased doping, led by the effect of Nb on nucleation following doping into the TiO2 lattice.A similar effect of different oxide additives on the TiO2 phase transformation was reported by Yanagida et al. [19].In addition, the diffraction peaks shifted to lower θ values with increasing Nb content as a result of the larger radius of Nb 5+ (0.64 Å) compared to that of Ti 4+ (0.61 Å) [15], in accordance with the Bragg equation: λ = 2dsinθ (Figure 1b). Figure 2 displays the X-ray photoelectron spectroscopy (XPS) spectra of the undoped and Nbdoped TiO2 samples with varying Nb contents.Nb 3d3/2 and 3d5/2 peaks are evident in the spectra of the Nb-doped TiO2 samples, and their intensities increased with increasing Nb content (Figure 2a).The Nb 3d3/2 and 3d5/2 peaks are located at binding energies of 209 and 206 eV, respectively.Nbdoping also caused the Ti 2p1/2 and 2p3/2 peaks to shift slightly toward higher binding energies, which is attributable to the higher electronegativity of Nb (1.6) compared to that of Ti (1.54) (Figure 2b) [15].The peaks in the O 1s spectra correspond to Ti 4+ -O bonds; the positions of these peaks show similar trends to those observed for Ti due to increases in both lattice oxygen and Nb 5+ -O bonds in these samples.Figure 2d displays a double-band structure, with the main peak (29.4 eV) related to O 1s electron binding in TiO2 and the other peak, at a higher binding energy (531 eV), attributed to OH groups on the surfaces of these samples.Typically, OH groups are beneficial for TCNQ anchoring [21].Figure 2 displays the X-ray photoelectron spectroscopy (XPS) spectra of the undoped and Nb-doped TiO 2 samples with varying Nb contents.Nb 3d 3/2 and 3d 5/2 peaks are evident in the spectra of the Nb-doped TiO 2 samples, and their intensities increased with increasing Nb content (Figure 2a).The Nb 3d 3/2 and 3d 5/2 peaks are located at binding energies of 209 and 206 eV, respectively.Nb-doping also caused the Ti 2p 1/2 and 2p 3/2 peaks to shift slightly toward higher binding energies, which is attributable to the higher electronegativity of Nb (1.6) compared to that of Ti (1.54) (Figure 2b) [15].The peaks in the O 1s spectra correspond to Ti 4+ -O bonds; the positions of these peaks show similar trends to those observed for Ti due to increases in both lattice oxygen and Nb 5+ -O bonds in these samples.Figure 2d displays a double-band structure, with the main peak (29.4 eV) related to O 1s electron binding in TiO 2 and the other peak, at a higher binding energy (531 eV), attributed to OH groups on the surfaces of these samples.Typically, OH groups are beneficial for TCNQ anchoring [21].

Optical Properties of Nb-Doped TiO2
Examination of the electronic structures of the Nb-doped TiO2 electrodes is essential to understand the ICT transition from the highest occupied molecular orbital (HOMO) of TCNQ to the conduction band of TiO2.XPS and calculated Eg values for the valence band can provide electronicpotential information.The reflectance spectra displayed in Figure S1a enable the bandgaps of the samples to be determined.Eg values were calculated after converting the reflectance data into the equivalent absorption coefficients using Equation (1), as shown in Figure S1b and listed in Table 1.Eg increased from 3.06 eV to 3.14 eV as the Nb content was increased from 0 mol % to 5.0 mol %, which is ascribable to strong hybridization between the Ti 3d and Nb 3d states that forms a d-type conduction band [22].The observed increase in bandgap with increasing Nb content can be also explained by the decrease in the amount of rutile phase TiO2 formed (Table S1); indeed, the bandgap of anatase is 3.2 eV, which is larger than that of rutile (3.0 eV). Figure 3 shows the photoemission spectra of TiO2 samples with different Nb contents, in which the binding energies are referenced against the Fermi energy level (EF).Two peaks are evident in each photoemission spectrum, one centered at 6.4 eV and the other at 8.2 eV.These peaks are attributed to emissions from πand σ-type O 2p orbitals, respectively.The positions of the valence band maxima (VBMs) were directly determined from the photoemission spectra by linear extrapolation of the

Optical Properties of Nb-Doped TiO 2
Examination of the electronic structures of the Nb-doped TiO 2 electrodes is essential to understand the ICT transition from the highest occupied molecular orbital (HOMO) of TCNQ to the conduction band of TiO 2 .XPS and calculated Eg values for the valence band can provide electronic-potential information.The reflectance spectra displayed in Figure S1a enable the bandgaps of the samples to be determined.Eg values were calculated after converting the reflectance data into the equivalent absorption coefficients using Equation (1), as shown in Figure S1b and listed in Table 1.Eg increased from 3.06 eV to 3.14 eV as the Nb content was increased from 0 mol % to 5.0 mol %, which is ascribable to strong hybridization between the Ti 3d and Nb 3d states that forms a d-type conduction band [22].The observed increase in bandgap with increasing Nb content can be also explained by the decrease in the amount of rutile phase TiO 2 formed (Table S1); indeed, the bandgap of anatase is 3.2 eV, which is larger than that of rutile (3.0 eV). Figure 3 shows the photoemission spectra of TiO 2 samples with different Nb contents, in which the binding energies are referenced against the Fermi energy level (E F ). Two peaks are evident in each photoemission spectrum, one centered at 6.4 eV and the other at 8.2 eV.These peaks are attributed to emissions from πand σ-type O 2p orbitals, respectively.The positions of the valence band maxima (VBMs) were directly determined from the photoemission spectra by linear extrapolation of the onsets of the valence-band emissions [12].In the case of the undoped TiO 2 sample, the VBM was found to be located 2.6 eV below the Fermi level.With the optical bandgap determined to be 3.06 eV, we deduce that E F is 0.46 eV lower than the energy of the conduction band minimum (E CBM ).The calculated positions of the conduction band minimum (CBM) with respect to the Fermi level are summarized in Table 1.The energy of the CBM of the 1.5 mol % Nb-doped TiO 2 was found to be 0.37 eV higher than that of the Fermi level, while that of the undoped TiO 2 was 0.46 eV higher (Table 1).The CBM of the sample containing 1.5 mol % Nb was therefore 0.09 eV lower in energy than that of undoped TiO 2 .The observed lowering of the CBM of the sample containing 1.5 mol % and 3.0 mol % Nb due to donor levels induced by these dopants [18] is attributable to higher electron transitions between the Nb-doped TiO 2 electrode and the anchoring TCNQ.On the other hand, the CBM for the sample containing 5.0 mol % Nb exhibited a large offset between its CBM energy and E F , compared to the undoped TiO 2 , due to intra-band transport [15].onsets of the valence-band emissions [12].In the case of the undoped TiO2 sample, the VBM was found to be located 2.6 eV below the Fermi level.With the optical bandgap determined to be 3.06 eV, we deduce that EF is 0.46 eV lower than the energy of the conduction band minimum (ECBM).The calculated positions of the conduction band minimum (CBM) with respect to the Fermi level are summarized in Table 1.The energy of the CBM of the 1.5 mol % Nb-doped TiO2 was found to be 0.37 eV higher than that of the Fermi level, while that of the undoped TiO2 was 0.46 eV higher (Table 1).The CBM of the sample containing 1.5 mol % Nb was therefore 0.09 eV lower in energy than that of undoped TiO2.The observed lowering of the CBM of the sample containing 1.5 mol % and 3.0 mol % Nb due to donor levels induced by these dopants [18] is attributable to higher electron transitions between the Nb-doped TiO2 electrode and the anchoring TCNQ.On the other hand, the CBM for the sample containing 5.0 mol % Nb exhibited a large offset between its CBM energy and EF, compared to the undoped TiO2, due to intra-band transport [15].

Photovoltaic Performance of Interfacial Charge-Transfer (ICT) Photoconversion Devices
The photovoltaic performance of ICT-transition devices based on the undoped and Nb-doped TiO2 electrodes with varying Nb contents under solar illumination (AM 1.5 G, 100 mW/cm 2 ) are displayed in Figure 4a, with the photovoltaic performance parameters summarized in Table 2. Jsc was observed to increase from 4.5 mA/cm 2 to 5.7 mA/cm 2 at an Nb content of 3.0 mol %, which is attributable to improved electronic transitions between the nanoparticles and the anchoring TCNQ associated with the negative CBM-energy shift (Table 1).On the other hand, Voc increased slightly, from 0.40 V to 0.41 V, as the dopant content was increased to 1.5 mol % due to the increased gap between EF and the redox potential of the I − /I 3− couple.This is also attributable to Nb-doped TiO2surface passivation, according to electrochemical impedance spectroscopy (EIS) (Nyquist plots,

Photovoltaic Performance of Interfacial Charge-Transfer (ICT) Photoconversion Devices
The photovoltaic performance of ICT-transition devices based on the undoped and Nb-doped TiO 2 electrodes with varying Nb contents under solar illumination (AM 1.5 G, 100 mW/cm 2 ) are displayed in Figure 4a, with the photovoltaic performance parameters summarized in Table 2. Jsc was observed to increase from 4.5 mA/cm 2 to 5.7 mA/cm 2 at an Nb content of 3.0 mol %, which is attributable to improved electronic transitions between the nanoparticles and the anchoring TCNQ associated with the negative CBM-energy shift (Table 1).On the other hand, Voc increased slightly, from 0.40 V to 0.41 V, as the dopant content was increased to 1.5 mol % due to the increased gap between E F and the redox potential of the I − /I 3− couple.This is also attributable to Nb-doped TiO 2 -surface passivation, according to electrochemical impedance spectroscopy (EIS) (Nyquist plots, Figure S2) [23,24].However, increased Nb doping has a negative influence on photovoltaic behavior at levels over 5.0 mol %; indeed, Voc decreased from 0.40 V to 0.36 V, which is ascribable to a detrimental conduction-band-edge effect.Indeed, the XRD result shows the decrease in rutile crystallization in 5.0 mol % Nb-doped TiO 2 (Table S1) led to defects in crystallization, and affected electron recombination such the degradation of V oc parameter.
Catalysts 2018, 8, x FOR PEER REVIEW 6 of 10 Figure S2) [23,24].However, increased Nb doping has a negative influence on photovoltaic behavior at levels over 5.0 mol %; indeed, Voc decreased from 0.40 V to 0.36 V, which is ascribable to a detrimental conduction-band-edge effect.Indeed, the XRD result shows the decrease in rutile crystallization in 5.0 mol % Nb-doped TiO2 (Table S1) led to defects in crystallization, and affected electron recombination such the degradation of Voc parameter.Figure 4b reveals that the device containing the 1.5 mol % Nb-doped TiO2 electrode exhibits enhanced IPCE compared to that of the undoped electrode.Indeed, the absorption edge was observed to shift from 730 nm to 750 nm.Nb has been reported to create oxygen vacancies in TiO2 that act as active sites, resulting in a photoresponse red shift [25].However, the lowest IPCE was obtained in the device containing the 5.0 mol % Nb-doped TiO2 electrode, as revealed by the J-V curves.
According to previous studies on the fabrication of DSSCs based on TiO2 doped with Nb at concentrations up to 5.0 mol %, Feng et al. and Huang et al. reported that the highest photovoltaic efficiency was observed for Nb concentrations of 2.0 mol % and 5.0 mol %, respectively [15,16].However, our results demonstrated that Nb-doping of 1.5 mol % produced the highest efficiency.Considering the behavior of Jsc with doping, we can conclude that the optimum Nb concentration for reaching the best efficiency is in the range of 1.5-3.0mol %.In addition, the error bar of photovoltaic efficiency in this study is shown in Figure S3.  Figure 4b reveals that the device containing the 1.5 mol % Nb-doped TiO 2 electrode exhibits enhanced IPCE compared to that of the undoped electrode.Indeed, the absorption edge was observed to shift from 730 nm to 750 nm.Nb has been reported to create oxygen vacancies in TiO 2 that act as active sites, resulting in a photoresponse red shift [25].However, the lowest IPCE was obtained in the device containing the 5.0 mol % Nb-doped TiO 2 electrode, as revealed by the J-V curves.
According to previous studies on the fabrication of DSSCs based on TiO 2 doped with Nb at concentrations up to 5.0 mol %, Feng et al. and Huang et al. reported that the highest photovoltaic efficiency was observed for Nb concentrations of 2.0 mol % and 5.0 mol %, respectively [15,16].However, our results demonstrated that Nb-doping of 1.5 mol % produced the highest efficiency.Considering the behavior of Jsc with doping, we can conclude that the optimum Nb concentration for reaching the best efficiency is in the range of 1.5-3.0mol %.In addition, the error bar of photovoltaic efficiency in this study is shown in Figure S3.

Preparation of Nb-Doped TiO 2
The Nb-doped TiO 2 nanoparticles were synthesized following the procedure described by Nikolay et al. [12].These syntheses used Ti and Nb precursors and hydrothermal processes.To prepare Nb-doped TiO 2 (0, 1.5, 3.0, or 5.0 mol %) nanoparticles, 16.4 mL of titanium tetraisopropoxide (Kanto Chemical Co., Tokyo, Japan) was mixed with 0, 60, 120, or 200 µL of niobium ethoxide (Wako Co., Tokyo, Japan), respectively, after which 2.64 mL of acetic acid (Kanto Chemical Co., Tokyo, Japan) was added under stirring condition with a Teflon stirrer blade for 15 min at room temperature.The mixture was dropped into 68.4mL of deionized water while stirred at about 800 rpm.After stirring for 1 h, 2.35 mL of 65% nitric acid (Kanto Chemical Co., Tokyo, Japan) was added to the solution.The solution temperature was increased to 80 • C over 40 min and then held at 80 • C for 80 min under reflux conditions with intensive stirring.The nanoparticles were hydrothermally grown using the prepared colloidal solution in a Teflon-lined mini-autoclave at 180 • C for 12 h, after which 0.52 mL of 65% nitric acid was added to the colloidal solution, followed by ultrasonication with stirring for 1 h.The prepared mixture was finally washed three times with ethanol by centrifugation.

Cell Fabrication
The Nb-doped TiO 2 pastes were synthesized following the procedure reported by Ito et al. [26].The final screen-printing pastes correspond to 18 wt % TiO 2 , 9 wt % ethyl cellulose and 73 wt % terpineol.Two kinds of pure ethyl cellulose (EC) powders, i.e., EC (10 mPas, Kanto Chemical Co., Tokyo, Japan) and EC (45 mPas, Kanto Chemical Co., Tokyo, Japan) were dissolved before usage in ethanol to yield 10 wt % solutions.Then, 0.325 g of EC (10 mPas) and 0.175 g of EC (45 mPas) of these 10 wt % ethanolic mixtures were added to a round-bottomed rotovap flask containing 1 g pure TiO 2 (obtained from a previously prepared precipitate) and 4.05 g of terpineol (Kanto Chemical Co., Tokyo, Japan), and diluted with approximately 100 ml of ethanol.This mixture was then ultrasonicated for 48 h.Ethanol and water were removed from these TiO 2 /ethyl cellulose solutions using a rotary-evaporator.The final formulations of the pastes were made with a three-roll mill (Exakt, Nagase Screen Printing Research Co., Aichi, Japan).The Nb-doped TiO 2 electrodes were fabricated by screen-printing pastes onto glass substrates coated with transparent conducting F-doped SnO 2 (FTO), Nippon Sheet Glass Co., Tokyo, Japan) with a sheet resistance of 10 Ω sq −1 , followed by sintering at 500 • C for 30 min.The thickness of the TiO 2 electrode was set to 8 µm.

Figure 2 .
Figure 2. High-resolution (a) Nb 3d, (b) Ti 2p, and (c) O 1s X-ray photoelectron spectroscopy (XPS) spectra of TiO2 with varying Nb contents.(d) O 1s spectral fitting, revealing the presence of peaks related to OH groups.

Figure 2 .
Figure 2. High-resolution (a) Nb 3d, (b) Ti 2p, and (c) O 1s X-ray photoelectron spectroscopy (XPS) spectra of TiO 2 with varying Nb contents.(d) O 1s spectral fitting, revealing the presence of peaks related to OH groups.

Figure 4 .Table 2 .
Figure 4. (a) Photocurrent-density-voltage curves and (b) incident photon-to-current conversion efficiency spectra of Interfacial charge-transfer (ICT) photoconversion devices based on the undoped and Nb-doped TiO2 electrodes prepared in this study.

Figure 4 .
Figure 4. (a) Photocurrent-density-voltage curves and (b) incident photon-to-current conversion efficiency spectra of Interfacial charge-transfer (ICT) photoconversion devices based on the undoped and Nb-doped TiO 2 electrodes prepared in this study.

Table 1 .
Optical bandgap values for TiO2 samples with different Nb contents.

Table 1 .
Optical bandgap values for TiO 2 samples with different Nb contents.

Table 2 .
Photovoltaic-performance parameters of Interfacial charge-transfer (ICT) photoconversion devices based on the undoped and Nb-doped TiO 2 electrodes prepared in this study.