2 mL aqueous solution of acetic acid (pH 3) was added to 6 g of anatase-type TiO₂ nano powder (P25; Nippon Aerosil Co., Ltd.), and the mixture was rigorously pulverized using a mortar and pestle for 15 min until obtaining a uniform suspension. Then, the same acetic acid solution was added 1 mL each, and further mixed until adding the total amount of 8 mL. About 30 min after starting the pulverization, 4 g of polyethylene glycol (molecular weight 20,000) was added and further continued mixing. The resultant suspension was coated on a TCO glass (1 × 1 inch, 1.8 mm thickness, type-VU, AGC Inc., Japan) substrate by a screen-printing machine (Mitani Micronics Co., Ltd., Japan) using a screen with 50 µm thickness. After drying, the coated glasses were fired in an electric furnace at 100˚C for 20 min, at 350˚C for 40 min, and at 450˚C for 60 min. Figure 1(a) and Figure 1(b) show the surface and cross section, respectively, of TiO₂ thin film thus prepared confirming that it had no cracks, and that its film thickness was about 10 μm.

Three kinds of 200 ml aqueous solutions were prepared by dissolving 0.1 M Na₂SO₄ as a supporting electrolyte using no buffer or 0.01 M Tris(hydroxymethyl)aminomethane (THAM) or 0.01 M 2-(N-morpholino)ethanesulfonic acid (MES) in distilled water. The pH of aqueous solutions was adjusted to 4.0, 6.0, 7.0 or 8.5 by adding dilute NaOH or H₂SO₄ solutions. After bubbling N₂ for 1 hr, each aqueous solution was divided into two glass containers connected with each

other by a glass filter. To one container, Dopamine hydrochloride was dissolved to give 0.01 M DA solution, and N₂ bubbling was continued for 30 min. Electro-oxidative polymerization of DA was carried out using a potentiostat (HAB-151A; HOKUTO DENKO Co., Ltd., Japan) linked to a three-electrode system consisting of a counter electrode, a reference electrode and a working electrode. A square Pt plate with dimensions of 1 × 1 inch was immersed as a counter electrode in a glass container without DA, and a TiO₂ film coated TCO glass substrate was immersed as a working electrode in another glass container with DA, in which a Luggin capillary connected to a Ag/AgCl reference (+0.199 V vs. SHE at 25˚C) electrode by a KCl salt bridge was also immersed. The PDA dye was coated on the TiO₂ electrode using cyclic process between 1 V and −1 V vs. Ag/AgCl with a sweep speed of 10 mV/sec while bubbling N₂ for 30 min. All electrochemical measurements were repeated three times to ensure the reproducibility.

A few drops of 10 mm H₂PtCl₆ 2-propanol solution were put onto a TCO glass substrate. After spreading evenly, the glass substrate was immediately set in a pre-heated furnace at 450˚C, and was fired for 30 min. A plastic film with 25 μm thickness was sandwiched between TiO₂ coated and Pt coated TCO glass substrates as a spacer, and a few drops of I − / I 3 − electrolyte containing 0.3 M KI and 30 mm I₂ was filled between them to assemble the DSSC. The irradiation area was limited to 1 cm². The DSSC assembly was connected to a potentiostat (HAB-151A; HOKUTO DENKO Co., Ltd., Japan), and its photocurrent density-photovoltage characteristics (I-V curve) were measured under the AM1.5 irradiation using a solar simulator (XES-40S1, SAN-EI ELECTRIC Co., Ltd., Japan). All measurements were repeated three times to ensure the reproducibility.

Figure 2(a) and Figure 2(b) show the cyclic voltammograms (CV curves) of DA obtained at pH 4.0 using THAM and MES, respectively, as a buffer. Both anodic and cathodic currents are very small, and no clear oxidation peak was observed in these figures. Figure 3(a) and Figure 3(b) show the CV curves of DA obtained at pH 6.0 using THAM and MES, respectively, as a buffer solution. When THAM was used as a buffer, large oxidation peaks were observed at about +0.55 V and +0.75 V vs. Ag/AgCl, the latter which was not observed in the first cycle but appeared in the second cycle onward. The corresponding reduction peak was only observed for the oxidation peak at about +0.7 V vs. Ag/AgCl, suggesting that the electrochemical reaction proceeding at about +0.55 V vs. Ag/AgCl would be an irreversible process, and that at +0.75 V vs. Ag/AgCl would be a reversible process. When MES was used as a buffer, both anodic and cathodic currents were much smaller, and a small oxidation peak was observed only at around +0.55 V vs. Ag/AgCl. Figure 4(a) and Figure 4(b) show the CV curves of DA

obtained at pH 7.0 using THAM and MES, respectively, as a buffer solution. When THAM was used as a buffer, the largest anodic current flowed in this experiment, and an oxidation peak appeared at about +0.75 V vs. Ag/AgCl, and that at about +0.55 V vs. Ag/AgCl was observed only in the first cycle and was not clearly observed in the second cycle onward. When MES was used as a buffer, an oxidation peak at about +0.5 V vs. Ag/AgCl was observed, but was much smaller than that using THAM buffer, and decreased with cycles. In contrast, the reduction peaks at around −0.7 V vs. Ag/AgCl were clearly observed in all cycles. Figure 5(a) and Figure 5(b) show the CV curves of DA obtained at pH 8.5 using THAM and MES, respectively, as a buffer solution. When THAM was used as a buffer, large oxidation peaks were observed only at about +0.4 V and the corresponding

reduction peak was not observed. The reduction peaks at around −0.75 V vs. Ag/AgCl were clearly observed in the first and second cycles, which decreased with cycles. When MES was used as a buffer, the oxidation peaks were smaller than that using THAM, but the CV curves were similar to those using THAM. The reason why almost no oxidation of DA occurred in low pH would be due to the positive charge of an amine group in the DA molecule resulting in repelling of DA molecules against the TiO₂ electrode to inhibit their adsorption on TiO₂ surface when the potential of working electrode was positively polarized. Thus, when pH approached neutral, pH 6.0 and 7.0, the number of positively charged DA molecules should decrease, and the neutral DA molecules were able to be adsorbed on the TiO₂ surface to be oxidized. At pH 8.5, the self-polymerization of DA molecules would proceed [10], and hence, the anodic current of DA would decrease. The reason why the oxidation peaks using MES are much smaller than those using THAM would be explained by the different charging states of these buffers. Because MES with a sulfonyl group having low pK_a should be negatively charged in aqueous solutions in the current experimental condition, MES ions would be strongly attracted to the TiO₂ electrode to inhibit the adsorption of DA molecules on TiO₂ surface when its potential was positively polarized.

PDA adsorption at Au electrodes during CV oxidation of DA in aqueous solutions was investigated as functions of DA concentration, solution pH and potential-sweep rate with the use of electrochemical quartz crystal microbalance technique [14], in which two anodic peaks were observed at 0.19 and −0.25 V vs. SCE, designated as P_a1 and P_a2 respectively, and two cathodic peaks were observed at 0.09 and −0.31 V vs. SCE, designated as P_c1 and P_c2 respectively, in the phosphate buffer solution (pH = 7.4). Following reversible electrochemical reactions were attributed to these peaks;

DA ⇄ P c1 P a1 DAQ

LDAC ⇄ P c2 P a2 DAC

where DA, DAQ, LDAC and DAC represent dopamine, dopamine quinone, leucodopaminechrome and dopaminechrome, respectively. It was concluded that the intramolecular cyclization of the first-step oxidation product of DA occurred significantly and further isomerization and oxidation of the cyclization product led to polymer growth at an Au electrode [14]. In the current work, the potentials of anodic peaks were much higher than those reported in Ref. [14], which would be due to the difference in the working electrode. The working electrode used in this work, TiO₂ coated on TCO, would require a much higher over-potential than Au electrode used in Ref. [14]. However, the fact that the PDA polymerization proceeded not at pH less than 5.0 but at pH higher than 7.0 was consistent with the current experimental results. It was reported that redox peaks shifted negatively with the increase of solution pH [14], which was also confirmed in the current experimental results. Consequently, the ECECEE (“E” denotes the electrochemical reactions while the ‘‘C’’ denotes the chemical reactions) mechanism for DA oxidation and subsequent polymerization for PDA formation [14] would be also applicable to the results obtained in this work. It was suggested that the presence of protons of relatively high concentration inhibited the coupled intramolecular cyclization of DAQ [14], which might be the cause for much less PDA formation at low pH in this work.

Table 1 summarizes the appearances of TiO₂ films and weight gain after electrochemical adsorption of PDA in an aqueous solution of pH 4.0, 6.0, 7.0 and 8.0 using THAM or MES as a buffer. It was found that the color of TiO₂ films after PDA adsorption became darker with increasing pH for both buffers. At low pH, brawn color was deeper for THAM than for MES, and vice versa at high pH.

Weight gains after PDA adsorption increased with increasing pH, using THAM as buffer gave larger weight gains as expected from the CV curves. It is noteworthy that weight gains were smaller for MES than THAM, even though the color was deeper for MES, especially at pH 8.5. As mentioned earlier, the self-polymerization of DA molecules at high pH would give PDA a higher molecular weight that would result in longer π-conjugated systems with a high molar absorption coefficient.

Figure 6(a) and Figure 6(b) compare the I-V curves of PDA-coated DSSC using THAM and MES, respectively. At pH 4.0, using THAM and MES as buffer both produced certain levels of photocurrents closed to that produced by TiO₂ film without any dyes, which would be expected by the fact that almost no anodic currents in CV curves were observed and the colors of TiO₂ films were very pale at pH 4.0 for both buffers. At pH 6.0 and 7.0, using THAM as buffer gave higher photocurrent than that at pH 4.0 suggesting that PDA dye would possess the sensitized effect. However, using MES gave much fewer photocurrents than using THAM at pH 6.0 and 7.0, even though TiO₂ films were colored brawn. At pH 8.5, almost no photocurrent was produced for both buffers, even though the colors of TiO₂ films were the darkest. As above discussed, negatively charged MES molecules would be attracted to the TiO₂ electrode to inhibit the adsorption of DA molecules on TiO₂ surface when the potential of working electrode was positively polarized, and the self-polymerized PDA in aqueous solution would be physically, not chemically, adsorbed on the TiO₂ surface. If PDA were not chemically adsorbed on the TiO₂ surface, exited electrons in PDA dye during irradiation would not be transferred to the conduction band of TiO₂ though the chemical bonding between them. This speculation would also be applied for the cases at pH 8.5 in both buffers. Physically adsorbed PDA on the TiO₂ surface would merely absorb incident light, which decreased the light intensity to decrease the number of photoelectrons in the TiO₂ conduction band resulting in much lower photocurrents without any dyes. The maximum η obtained in these experiments was calculated to be 0.049% at pH 7.0 using THAM, which would be extremely low as DSSC performance. It was reported that the zeta potential of DA-adsorbed TiO₂ particle was highly positive at pH higher than 7 and increased with DA concentration [12]. Since DA molecules in the aqueous solution were also positively charged at pH lower than 7, DA molecules

would repel against the surface of DA-adsorbed TiO₂ particles.

Experimental results obtained in the above section suggest that MES molecules would strongly adsorb on the surface of TiO₂ particles to inhibit the adsorption of DA molecules. Since a THAM molecule possesses one amino group and three hydroxyl groups, it might also inhibit the adsorption of DA molecules. Thus, electrochemical oxidation of DA without any buffer was investigated. Figure 7(a) shows the CV curves of DA obtained at pH 7.0 without buffer, which was found to be significantly different from those obtained with THAM and MES buffers at pH 7.0 as shown in Figure 4(a) and Figure 4(b), respectively. Two cathodic peaks were clearly observed at about −0.05 and −0.35 V vs. Ag/AgCl, which were close to those observed at 0.09 and −0.31 V vs. SCE in the previous work [14]. Two anodic peaks at about 0.55 and 0.73 V vs. Ag/AgCl were observed from the second cycle onward, although only one anodic peak at about 0.55 V vs. Ag/AgCl was observed in the first cycle. These anodic peaks were also observed in Figure 4 and Figure 5(a). As discussed in the above section, the working electrode used in this work, TiO₂ coated on TCO, would require a much higher over-potential than Au for oxidation. Considering this influence, it would be concluded that over all CV-curves obtained in this work without buffer are quite similar to those reported in Ref. 14. Figure 7(b) shows the I-V curves of PDA-coated DSSC fabricated at pH 6.0, 7.0 and 8.5 without buffer. Compared with those shown in Figure 6(a), both V_oc and I_sc are quite similar, however, the fill factor (FF) is much higher without buffer than that with THAM buffer, and η is calculated to be about 0.065%, which is much higher than the maximum η about 0.049% with THAM buffer. This suggests that THAM buffer adsorbed on

the TiO₂ surface might inhibit the electrochemical oxidation of DA although THAM buffer was used for fabricating PDA (DC)-DSSC [10].

Because the adsorption time was only 30 min. in the preceding experiments, it was prolonged up to 8 hrs. Table 2 summarizes the appearances of TiO₂ films and weight gain after electrochemical adsorption of PDA for 0.5, 1, 2, 4 and 8 hrs in aqueous solution at pH 7.0 without buffer. The brawn color became darker with increasing adsorption time, and after 4 and 8 hrs, the film colors were dark brawn. The weight gain after adsorption increased with increasing adsorption time, however, the amount of adsorbed PDA was only limited to 0.3% - 0.5% of the total amount of DA dissolved in aqueous solution.

Figure 8 shows the I-V curves of DSSCs after electrochemical adsorption of PDA in aqueous solution at pH 7.0 without buffer for 0.5, 1, 2, 4 and 8 hrs. It was found that I_sc did not change significantly up to 2 hrs and decreased after 4 hrs, however, V_oc monotonically decreased with increasing adsorption time yielding η = 0.065%, 0.049%, 0.049%, 0.038% and 0.026%, respectively. This apparent contradiction would be due to the increased resistivity of PDA with a larger degree of polymerization. The addition of dopants that would decrease the resistivity of PDA might improve η.

As mentioned in the previous section, two cathodic peaks were clearly observed at about −0.05 and −0.35 V vs. Ag/AgCl in Fig. 7 (a), which may be attributed to the reduction of DAQ and DAC, the intramolecular cyclization reaction product of DAQ, respectively, according to the ECECEE mechanism [14]. Because PDA formation might be improved by suppressing these reverse reactions, electrode potential was fixed at +1.0 V vs. Ag/AgCl or potential sweep range was changed to −0.5 - +1.0 V vs. Ag/AgCl instead of −1.0 - +1.0 V vs. Ag/AgCl. Figure 9 compares the I-V curves of DSSCs after electrochemical adsorption of PDA in aqueous solution at pH 7.0 without buffer for 30 min using the electrode potentials above mentioned. When the electrode potential was fixed at +1.0 V vs. Ag/AgCl, both V_oc and I_sc significantly decreased, which suggests that the reversal of electrode potential from positive to negative would be essential for the electrochemical adsorption of PDA. Since DA exists as a neutral molecule or DA⁺ ion in an aqueous solution at pH 7.0, DA⁺ ions would be attracted to a working electrode when the potential was negative, and then they would be oxidized when the electrode potential returned to positive. When the potential sweep range was changed to −0.5 ~ +1.0 V vs. Ag/AgCl instead of −1.0 ~ +1.0 V vs. Ag/AgCl, V_oc slightly decreased but I_sc slightly increased that improved η from 0.065% to 0.073%. This change of negative electrode potential would suppress the reverse reduction reactions of oxidized DA at positive electrode potential. This experiment would suggest the importance of optimization not only in electrode potential but also sweep rate. Further investigation would be necessary to improve η using the electrochemical adsorption of PDA.