Enhanced Power Conversion Efficiency of Dye-Sensitized Solar Cells by Band Edge Shift of TiO2 Photoanode.

By simple soaking titanium dioxide (TiO2) films in an aqueous Na2S solution, we could prepare surface-modified photoanodes for application to dye-sensitized solar cells (DSSCs). An improvement in both the open-circuit voltage (Voc) and the fill factor (FF) was observed in the DSSC with the 5 min-soaked photoanode, compared with those of the control cell without any modification. The UV-visible absorbance spectra, UPS valence band spectra, and dark current measurements revealed that the Na2S modification led to the formation of anions on the TiO2 surface, and thereby shifted the conduction band edge of TiO2 in the negative (upward) direction, inducing an increase of 29 mV in the Voc. It was also found that the increased FF value in the surface-treated device was attributed to an elevation in the shunt resistance.

1. Introduction

Titanium dioxide (TiO2, titanium (IV) oxide, or titania) has a wide range of applications such as paint, sunscreen, food coloring, photocatalysts, perovskite solar cells, and dye-sensitized solar cells (DSSCs) [1,2]. In particular, TiO2 photoanode is one of the important factors determining the photovoltaic properties of DSSCs. The mesoporous TiO2 layer provides both an extended surface area for dye adsorption and a sufficient soaking of the hole-carrying liquid electrolytes, leading to the enhanced light harvesting efficiency (LHE) and dye regeneration efficiency of DSSCs, respectively. In addition, the mesoporous TiO2 is an excellent recipient of photo-injected electrons from the adsorbed dyes and provides highly efficient electron transport from the injection site to the transparent electrode layer. Thus, a high short-circuit current (J) can be realized in DSSCs [3]. Meanwhile, it has been reported that a surface modification of the TiO2 layer can enhance the open-circuit voltage (V) due to the shift of the conduction band edge (CBE) of the TiO2 arising from the formation of a surface dipole. The negative shift of the TiO2 CBE can increase the V value, since it is directly proportional to the energy difference between the TiO2 CBE and the redox potential of the electrolyte [4]. It was reported in previous work that grafting of SrTiO3 [4], ZnO [5], chenodeoxycholate [6], BaCO3 [7], and MgO [8] onto the TiO2 surface shifted the CBE of TiO2 in the negative direction owing to the formation of a surface dipole, leading to an increase in the V.

However, in the previous works, the surface modification process using inorganic compounds was time and energy consuming because additional thermal annealing at an elevated temperature should be conducted after dipping TiO2 films in respective solutions of their precursors. In this study, we selected a strong base, i.e., sodium sulfide (Na2S), as a surface modifier for TiO2. TiO2 films formed on F-doped SnO2 (FTO)/glass substrate were immersed in an aqueous Na2S solution to modify their surfaces, and the resultant films were used as DSSC photoanodes. Solar cells with the Na2S-modified TiO2 layer were fabricated, and the effects of the surface modification on the photovoltaic performance were studied. We expected the TiO2 surface modification using strong base to alter surface charges in a short processing time, which could increase photovoltaic properties of DSSCs.

2. Results and Discussion

2.1. Modification of the TiO2 Surface Using an Aqueous Na2S Solution

In an aqueous solution, Na2S exists as Na+ and SH− because sulfide (S2−) is more basic than hydroxide, and thus deprotonates water, as described by Equations (1) and (2). The bisulfide (SH−) can also react with water to produce hydroxide (OH−) ions (Equation (3) [9]. Two different strong bases (SH− and OH−) exist in an aqueous Na2S solution (pH = 12.6), and will generate a negative charge on the TiO2 surface by deprotonating the surface hydroxyl groups as described by Equations (4) and (5) [10,11]. Thus, when the pristine TiO2/FTO was soaked in the Na2S solution, the TiO2 surface was negatively charged. Na S SH TiO TiO

Four different photoanodes were prepared by immersing the pristine TiO2/FTO in the solution for 5–20 min, thereby producing Na2S(5, 10, 15, and 20)-TiO2/FTO, where “(5)” means that the soaking time was 5 min. To characterize the TiO2 surface, Na2S(360)-TiO2/FTO immersed in the solution for 6 h was also prepared, and XPS measurements were performed for both the Na2S(5)-TiO2/FTO and the Na2S(360)-TiO2/FTO photoanodes. As expected from Equations (4) and (5), a sulfur peak at approximately 165 eV attributed to S 2p3/2 was not detected in both photoanodes, as displayed in Figure 1a,b. As a reference, it is well known that sulfonates (−SO3−) can easily adsorb on the TiO2 surface [12,13]. However, even with extended dipping time (6 h), sulfur was not detected on the TiO2 surface, indicating that bisulfide (SH−) was not adsorbed, and rather deprotonated the surface hydroxyl groups of TiO2 as shown in Equation (4). To further confirm the nonexistence of sulfur on TiO2 surface, the ATR-FTIR spectra were recorded for the pristine TiO2/FTO, the Na2S(5)-TiO2/FTO, and the Na2S(360)-TiO2/FTO photoanodes, as displayed in Figure 2. As expected, we could not observe an absorption band at 2600–2550 cm−1, attributed to the stretching of the S-H bond. Meanwhile, we could also observe peaks detected at 1072.9eV corresponding to the binding energy of 1s in Na in both photoanodes (Figure 1c).

Figure 1

XPS spectra of the Na2S-TiO2/F-doped SnO2 (FTO) photoanodes; (a) survey scan, (b) S 2p3/2, and (c) Na 1s spectra.

Figure 2

Attenuated total reflectance (ATR)-FTIR spectra of the pristine TiO2/FTO, the Na2S(5)-TiO2/FTO and the Na2S(360)-TiO2/FTO photoanodes.

2.2. Photovoltaic Properties of DSSCs with Na2S-TiO2/FTO

The surface-modified photoanodes were applied in the fabrication of the DSSCs, and then their photovoltaic properties were characterized. To investigate effects of the surface modification time on the cell performance, four DSSCs were fabricated in each condition. Figure 3 shows the variations in the averaged device performance as a function of the surface modification time, namely the time of soaking in the aqueous Na2S solution. Detailed photovoltaic parameters are compared in Table S1 in the Supplementary Information. For all of the surface-treated DSSCs, an improved V was achieved compared to that of the reference cell with the pristine TiO2/FTO (Figure 3a). However, the J values decreased with increasing soaking time, as shown in Figure 3b. The fill factor (FF) values of the surface-treated cells were improved compared to that of the control device. Overall, the averaged power conversion efficiency (PCE) values were increased by the Na2S modification, because the improvement in both the V and the FF had a stronger effect on the PCE than the reduction in the J.

Figure 3

Variations in performance parameters with the soaking time in the Na2S solution: (a) V, (b) J, (c) fill factor (FF), and (d) PCE of the dye-sensitized solar cells (DSSCs) measured under AM 1.5 (100 mW/cm2) illumination.

Since the photoanode modified with the Na2S solution for 5 min showed the highest PCE, we focused on this device (Na2S(5)-TiO2/FTO) in our investigations of the origin of the efficiency enhancement. Among the four cells, one that showed a PCE similar to the averaged value was chosen. Photovoltaic properties of the selected cells are compared in Figure S1 and Table S2 in the Supplementary Information. The current density (J) and voltage (V) curves of the selected devices with the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO photoanodes were compared in Figure 4, and their cell performance characteristics are listed in Table 1. The control cell with the pristine TiO2/FTO showed a PCE of 7.70%, while the DSSCs with the Na2S(5)-TiO2/FTO photoanode exhibited a PCE of 8.08% due to the increase in both the V and the FF. Even though the J of the Na2S(5)-TiO2/FTO device was lower than that of the control cell, the PCE was improved due to an increase in the V and the FF. It is important to reveal the origin of the enhancement in the V and the FF, in addition to the decrease in the J.

Figure 4

Current density–voltage (J–V) characteristics of the selected DSSCs with the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO photoanodes.

Table 1

Photovoltaic performance characteristics of the selected cells with the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO photoanodes.

Applied Photoanodes	Voc(mV)	Jsc(mA/cm2)	FF(%)	PCE(%)	Loaded Dye (μmol/cm3)	Rse(Ωcm2)	Rsh(Ωcm2)
Pristine TiO2/FTO	671	17.35	66.16	7.70	17.39	6.96	1418
Na2S(5)-TiO2/FTO	700	16.53	69.81	8.08	17.09	6.05	2347

2.3. Effects of Surface Modification on Voc

The DSSC with the Na2S(5)-TiO2/FTO exhibited 700 mV of the V, which was increased from that of the control cell (671 mV). This increase in the V is most likely due to the transformation in the energy band structure of TiO2 by the Na2S modification. To determine the electronic band structures of the photoanodes, it is necessary to measure the band gap energy (Eg) and the valence band edge (VBE) energy, by which the CBE energy can be estimated using the equation: CBE energy = VBE energy − E

We thus measured the UV–visible absorption and the UPS valence band spectra to estimate the Eg and the VBE energy, respectively. The absorption edge of both the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO was observed at 366 nm, which is corresponding to an Eg of 3.39 eV (Figure 5a). We can also see in Figure S2a in the Supplementary Information section that the Eg of the Na2S(365)-TiO2/FTO was also maintained as 3.39 eV. The VBE of the pristine TiO2/FTO was positioned at about 4.00 eVNHE (NHE = normal hydrogen electrode), while it moved to 3.81 eVNHE for the Na2S(5)-TiO2/FTO, as shown in Figure 5b. The VBE of the Na2S(360)-TiO2/FTO was further shifted to a lower binding energy (i.e., 3.44 eVNHE), as shown in Supplementary Figure S2b. The band edge positions in NHE potential (ENHE) were converted into the energy of the absolute vacuum scale (EAVS) using the equation:E where E0 (approximately 4.50 eV) is the energy of the free electrons on the hydrogen scale [14,15,16]. Accordingly, the VBE energies in the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO were calculated to be −8.50 and −8.31 eVAVS, respectively. Based on the identical Eg (3.39 eV), the CBE energies of the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO were determined as −5.11 and −4.92 eV, respectively. As a result, the CBE in the Na2S(5)-TiO2/FTO was shifted by 19 meV in the negative direction compared to that of the pristine TiO2/FTO, as illustrated in Figure 6. Thus, the Na2S modification led to a negative shift of the TiO2 CBE that resulted from the presence of surface negative charges. Thus, this negative shift of the TiO2 CBE explains the increase in the V of the DSSC with the Na2S(5)-TiO2/FTO, since it induces a wider potential gap (ΔVpristine < ΔVNa2S) between the TiO2 CBE and the redox potential of the electrolyte, as shown in Figure 6. Considering that the potential difference between the TiO2 CBE and the redox potential of the electrolyte is proportional to the V [17,18], it can be concluded that the increased V originates from the negative shift of the TiO2 CBE. Similar results have been reported in several previous studies that showed that the TiO2 surface modification using basic substances shifted the TiO2 CBE to a negative direction, resulting in an enhancement in V [10,11,19]. It is interesting that the Na2S modification can lead to an increase in V even the very short modification time of 5 min, compared with other surface modifiers studied in previous reports.

Figure 5

(a) UV–vis absorbance and (b) ultraviolet photoelectron spectroscopy (UPS) valence band spectra for the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO.

Figure 6

Schematic energy band diagram of (a) the pristine TiO2/FTO and (b) the Na2S(5)-TiO2/FTO, indicating a negative shift of the TiO2 conduction band edge (CBE).

The dark current values of DSSCs can be used to determine a shift of the CBE [4,5]. We measured dark currents of the DSSCs with the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO, as displayed in Figure 7. The onset potential of the dark current for the pristine TiO2/FTO photoanode was measured to be 0.560 V, whereas it was shifted to 0.578 V for the Na2S(5)-TiO2/FTO photoanode. Due to the Na2S modification, a higher onset potential was recorded in the 5 min-treated cell, implying that the potential difference between the FTO work function and the TiO2 CBE in the Na2S(5)-TiO2/FTO was higher than that for the pristine TiO2/FTO (ΔE1pristine < ΔE1Na2S). This result suggests that the TiO2 CBE is shifted in the negative direction (upward direction), which is in agreement with the results of the energy band structure.

Figure 7

Dark current–voltage profiles of the DSSCs with the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO photoanodes.

Based on the UV–visible absorbance, UPS valence band, and dark current measurements, it was concluded that the V was increased by the negative shift in the TiO2 CBE attributed to the surface negative charges arising from the Na2S modification.

2.4. Effects of Surface Modification on Jsc

As shown in Table 1, the Na2S modification induced a decrease in the J from 17.35 mA/cm2 for pristine TiO2/FTO to 16.53 mA/cm2 for Na2S(5)-TiO2/FTO. The J value can be expressed as J where LHE(λ), ηinj(λ), ηcoll(λ), and ηreg are the light harvesting, electron injection, electron collection, and dye regeneration efficiencies, respectively; e is the elementary charge and Φph,AM1.5G is the photon flux in AM 1.5 G conditions of 100 mW/cm2 [20,21].

The LHE is mostly affected by the light absorbance (A) of the adsorbed dyes, that is LHE = 1 − 10−A [20,21]. To estimate the effects of LHE on the J reduction, we measured the amount of dyes adsorbed on the surface of the TiO2 photoanode using the Beer–Lambert equation [22,23,24,25]. The average dye-loading amounts, as measured using five photoanodes, for the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO were 17.39 ± 1.79 and 17.09 ± 2.61 μmol/cm3, respectively. The adsorbed amounts of dyes were almost identical in both photoanodes, indicating that LHE did not affect J. It is believed that the very short soaking time of 5 min did not influence dye-loading amounts. However, when the photoelectrodes were soaked in the Na2S solution more than 5 min, the J values were further decreased, while the V values were almost maintained (Figure 3). The additional decrease in the J is probably due to a lowered dye-loading amount in DSSCs with the Na2S(10, 15, and 20)-TiO2/FTO. At the extended soaking time, surface negative charges (i.e., TiO2-O−) would be increased, and thus they would disturb dye adsorption on TiO2 surface due to the repulsion between TiO2-O− and carboxylate anions of N719 dye molecules. This could lead to an additional decrease in the J.

The ηinj value is influenced by the potential difference between the lowest unoccupied molecular orbital (LUMO) level of the dye and the TiO2 CBE. The measurements of the UV–visible spectra, UPS valence band spectra, and dark current revealed that the CBE of the TiO2 photoanode was shifted to a more negative potential by the Na2S modification, resulting in a smaller potential difference (ΔE2pristine > ΔE2Na2S) between the LUMO level of the N719 dye and the TiO2 CBE, as illustrated in Figure 6. This can lead to a less efficient electron injection from the excited N719 dye into the conduction band of the surface-modified TiO2 [4,5,6,7,8]. Thus, the reduced Jsc can be attributed to the lowered ηinj arising from the negative (or upward) shift of the TiO2 CBE.

Meanwhile, it has been reported that adsorption of various metal cations leads to the positive shift (away from the vacuum level) of the TiO2 CBE and thereby enhances the ηinj value [26,27]. However, in our study, it is believed that the positive shift of the TiO2 CBE by the adsorption of the sodium cations is negligible, judging from the measurements of the UV–visible spectra, UPS valence band spectra, and dark current. In view of the relatively poor signal-to-noise of the Na 1s peak (Figure 1c), it seems that a small amount of sodium was adsorbed regardless of the soaking time. It is probably because most of adsorbed sodium ions (i.e., TiO2-O−·Na+) were removed during rinsing with water.

The ηcoll value can be conjectured from the lifetime of the electrons photo-injected into the TiO2 valence band; that is, a prolonged electron lifetime induces an increase in ηcoll. We conducted EIS analysis to estimate the electron lifetime and charge transport of the devices with the pristine and surface-modified photoanodes [28,29,30]. As shown in the Bode phase plots of the EIS spectra (Figure 8) measured under illumination, the peak frequency (fmax) of the device with the Na2S(5)-TiO2/FTO was obtained as 26.94 Hz, which was identical to that of the reference device. The electron lifetime (τn) estimated using the equation [29,30]τ was calculated to be 5.91 ms in both devices. Thus, the unchanged electron lifetime means that ηcol did not vary with the Na2S modification. This suggests that ηcol did not affect J.

Figure 8

Bode plots of EIS spectra for the DSSCs with the pristine TiO2/FTO and the Na2S(5)-TiO2/FTO, measured under illumination (100 mW/cm2).

Overall, it was concluded that the reduced J value of the DSSC with the Na2S(5)-TiO2/FTO was due to the decrease in ηinj resulting from the negative shift of the TiO2 CBE by the Na2S modification.

In our previous work [31], by incorporating polyacrylonitrile, a surface dipole was formed on the TiO2 surface, which induced a positive shift of the TiO2 CBE. The positive shift resulted in an enhancement in the ηinj, and hence the J. On the other hand, in this study, the TiO2 CBE was shifted to a negative direction by the reactions between the surface hydroxyl groups and the strong bases, leading to a decrease in the J. Effects of the surface modification in the two studies are totally opposite. In summary, the Na2S and the polyacrylonitrile act as a V-improver by a negative shift and as a J-improver by a positive shift of the TiO2 CBE, respectively. We can conclude that, in order to improve the J, surface modifiers play a role of shifting the TiO2 CBE to a positive direction.

2.5. Effects of Surface Modification on FF

The FF value of the device with the 5 min-treated photoanode (69.81%) was increased compared to that of the control cell (66.16%). It is well known that the series (R) and shunt (R) resistances of cells affect the FF value; lower R and/or higher R are favorable to improve the FF [32,33,34]. We estimated the R and R values from the slope of the J-V curves at V and J, respectively [32,33]. The DSSC with the Na2S(5)-TiO2/FTO had a R of 6.05 Ωcm2, which was similar to that of the reference device (6.96 Ωcm2) with the non-treated photoanode (Table 1). The R value of the Na2S-treated DSSC was increased by 2347 Ωcm2 from that of the reference device (1418 Ωcm2). We therefore believe that the increased R value results in an enhancement in the FF.

3. Experimental Details

3.1. Materials

For DSSC fabrication, we obtained commercial products of Solaronix (Aubonne, Switzerland) such as F-doped SnO2 (FTO) glass (TCO22-7) with sheet resistance of ~7 Ω/square, TiO2 paste for the photoanode (Ti-nanoxide T/SP), TiO2 paste for the scattering layer (Ti-nanoxide R/SP), N719 dye (Ruthenizer 535-bisTBA), hot-melt adhesive (SX1170-60PF, Surlyn), and iodide-based electrolytes (AN-50). Platinum paste (PT-1, acquired from Dyesol-Timo JV, Seoul, Korea) was selected as the source for the Pt counter electrode. Sodium sulfide was purchased from Sigma-Aldrich (St. Louis, MO, USA). All reagents and materials were used as received without further purification.

3.2. Fabrication of the DSSCs

Except for the modification process of the TiO2 surface, the same procedures presented in our earlier work [31] were applied to prepare DSSCs. The modification process using Na2S is as follows: the TiO2 films formed on FTO glass were soaked in an aqueous Na2S solution (80 mM) for 0–20 min to modify the TiO2 surfaces. Next, the resultant electrodes were rinsed with deionized water and ethanol, and then dried at 65 °C for 10 min to produce the modified photoanodes (Na2S-TiO2/FTO). Detailed fabrication conditions are provided in the Supplementary Information.

3.3. Measurements

X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS) analyses were conducted using a VG Multilab ESCA 2000 apparatus (Thermo VG Scientific, Needham, MA, USA), with Al Kα radiation (hv = 1486.6 eV) used for the XPS measurements and with He I radiation (hv = 21.22 eV) used for the UPS measurements. The energy of the C1s photoelectron peak (binding energy of 284.6 eV) was used as the reference energy. The Fourier transform infrared (FT-IR) spectra were recorded using an FT/IR 4100 spectrometer (Jasco, Tokyo, Japan) equipped with an attenuated total reflectance (ATR) accessory (PRO450-S, Jasco). The photocurrent and voltage were measured using a CompactStat potentiostat (Ivium Technologies B.V., Eindhoven, The Netherlands) and a PEC-L01 solar simulator system equipped with a 150 W xenon arc lamp (Peccell Technologies, Inc., Yokohama, Japan). To adjust light intensity to 1 sun (100 mW/cm2), we used a silicon photodiode (PEC-SI01, Peccell Technologies, Inc., Yokohama, Japan). The UV–Vis absorption spectra were obtained using a spectrophotometer (NEOSYS-2000, SINCO Co., Ltd., Seoul, Korea). An electrochemical analyzer (CompactStat, Ivium Technologies B.V., Eindhoven, The Netherlands) was used for electrochemical impedance spectroscopy (EIS) analysis. The active areas of the dye-sensitized TiO2 layers were estimated using a digital microscope camera (SZ61, OLUNPUS Corporation, Tokyo, Japan) equipped with image analysis software.

4. Conclusions

Na2S-modified TiO2 films were prepared by immersing TiO2 films into an aqueous Na2S solution and were applied as DSSC photoanodes. The reference device without any modification demonstrated a PCE of 7.70% (J = 17.35 mA/cm2, V = 671 mV, and FF = 66.16%), while the PCE of the cell with the Na2S-modified TiO2 layer was increased to 8.08% (J = 16.53 mA/cm2, V = 700 mV, and FF = 69.81%), due to an improvement in both the V and the FF. By the Na2S modification, an improvement in the V was achieved due to a negative shift of the TiO2 CBE, and the FF value was increased by an elevated R. This result indicates that Na2S appears to be a promising material for enhancing both the V and the FF of DSSCs.