All the reagents and solvents were purchased from Sigma-Aldrich (St. Louis, MO, USA). Cis-diisothiocyanato–bis(2,2′-bipyridyl-4,4′-dicarboxylato) ruthenium (II) bis(tetrabutylammonium) (N719) dye was purchased from Solaronix, Switzerland. ITO/PEN (13 Ω/sq.) and glass/FTO (8 Ω/sq.) were purchased from Peccell Technologies Inc (Japan) and Pilkington (USA), respectively.

Before preparing the TiO₂ pastes, the TiO₂ powder (Degussa P25) was pre-treated with acetylacetone to minimize NPs aggregation [9]. Then, 1 g of pretreated TiO₂ powder was mixed with two compositions of 1-octanol and CCl₄ (1:1 and 1:0 (v/v)) and stirred at room temperature (RT) for 6 h. Subsequently, the mixture was homogenized for another 30 min to obtain homogenous dispersion of TiO₂ NPs by using an ultrasonic horn. The mixture was heated again at 70°C with stirring (300 rpm/s) for 24 h to obtain the high viscous pastes. The pastes were denoted hereafter as P_BL and P_O, where the composition of 1-octanol and CCl₄ were 1:1 and 1:0 (v/v), respectively.

The as-prepared TiO₂ pastes were coated on flexible ITO/PEN and glass/FTO substrates by the doctor blade method and dried at RT. Then, the TiO₂ films were sintered at 150°C for 60 min in an electric oven in the presence of air. Subsequently, the films were treated with UV-O₃ for 30 min. For comparison, the P_BL and a commercial binder containing TiO₂ paste (P_C) (TTPH-20N, ENBKOREA Co., Ltd., Korea, particle size 20 nm) were used to prepare 500°C sintered TiO₂ films on glass/FTO substrate according to a previously reported method [28]. The thickness of the P_BL and P_O-based TiO₂ films was ca. 15.8 and 15.3 μm, respectively, while it was ca. 15.4 μm for P_C-based TiO₂ photoanode. All the TiO₂ electrodes were dipped into the N719 dye solution (0.3 mM) in ethanol for 18 h. For the preparation of f-DSSCs CE, Pt was sputtered on ITO/PEN substrate for about 10 s. While for the preparation of g-DSSCs CEs, 5 mM chloroplatinic acid hexahydrate (H₂PtCl₆·6H₂O) solution in ethanol was drop cast onto the glass/FTO substrate and sintered in an electric muffle furnace at 400°C for 20 min in air. The dye-loaded photoelectrodes (active area ca. 0.25 cm²) and Pt-CEs were sandwiched using 60-μm thick Surlyn film (Solaronix SA, Switzerland) as a spacer and sealing agent using a hot press. The electrolyte solution with the composition of 0.6 M 1,2-dimethyl-3-propylimidazolium iodide (DMPII), 0.1 M LiI, 0.1 M I₂, and 0.5 M 4-tert-butylpyridine (tBP) and 1 mM 4′,6-diamidino-2-phenylindole (DAPI) as an efficient energy relay dye (ERD) [29] were dissolved in 3-methoxypropionitrile (MPN) and injected into the cells through the drilled holes on the CEs. The holes were sealed by using surlyn film with cover glass.

The viscosity of the TiO₂ pastes was measured with a viscometer (μVisc, Rheosense, Inc., USA). The thickness of the TiO₂ films was measured by a surface profilometer (Accretech, Japan), and the adhesion of TiO₂ film with FTO was evaluated by a scotch-tape scratch test. The strength of adhesion of the films was evaluated using a peel-off tester (Top-Tac 2000, Korea). The crystallographic pattern of TiO₂ was characterized with an X-ray diffractometer (XRD, Philips, X’pert, Netherland) using Cu K_a radiation. A field-emission scanning electron microscope (FE-SEM, JSM-6700F, JEOL) equipped with energy-dispersive X-ray spectroscopy (EDS) was used (EDS, INCAx-sight7421, Oxford Instruments) to characterize the surface morphology and elemental analysis of TiO₂ films. The amount of adsorbed dye onto the TiO₂ film was determined by desorbing the dye in a 0.1 M NaOH aqueous solution, and the absorption spectra of the desorbed dye solutions were measured using a UV–Vis absorption spectrophotometer (Perkin-Elmer Lamda 35). A solar simulator (Polaronix^® K201, McScience, Korea) with a 200 W Xenon lamp with the incident light intensity of 100 mW/cm² (AM 1.5 G) was used for the photovoltaic (PV) measurements of the DSSCs. A PV power meter (Polaronix^® K101 LAB20, McScience, Korea) was used to measure the current density-voltage (J-V) characteristics of DSSCs. Incident photon-to-electron conversion efficiency (IPCE) spectra were obtained by an IPCE measurement system (Polaronix, K3100 Spectral IPCE, McScience) with a 300 W xenon lamp source. Electrochemical impedance spectra (EIS) were measured using an impedance analyzer (IM6ex, Zahner Zahner-Elektrik GmbH & Co. KG, Germany) in the frequency range from 10⁵ to 0.1 Hz with the ac amplitude of 5 mV under open-circuit potential and dark conditions. The EIS spectra were analyzed with an appropriate equivalent circuit using Z-view software (Scribner Associates Inc., version 3.1). Thermogravimetric (TGA) analyses were performed with a Scinco TGA-N 1000 analyzer (Seoul, Korea) under ambient atmospheric conditions.

Fig. S1 presents the bar diagram of the T_b variation of the CCl₄ and 1-octanol mixture against their volume fraction. The T_b of the binary mixture was increased by increasing the volume fraction of 1-octanol that is in accordance with Raoult’s law [27]. Considering the preparation of binder-free, low-temperature chemically sintered TiO₂ paste, and the composition of CCl₄ and 1-octanol complex formation [30], a 1:1 volume ratio is selected as the optimized composition, where the T_b of 1-octanol:CCl₄ (1:1 v/v) was 97°C. The schematic illustration of the TiO₂ NPs inter-connections in the mp-TiO₂ film based on P_BL is presented in Scheme 1. In the mixture of 1-octanol, CCl₄, and TiO₂ NPs, 1-octanol and CCl₄ could easily form a complex with a 1:1 (v/v) ratio via hydrogen bonding (O-H·····C1-C) [30]. Subsequently, this complex could interact with the -OH groups on the TiO₂ NPs surface and form Ti-O-C-O-C₈H₁₇ and Ti-O-C-O-Ti bonds by the elimination of HCl. This enabled to decrease in the TiO₂ interparticle distance and formed a viscous paste. In the subsequent heat treatment of the as-prepared TiO₂ NPs film at 150°C, CO₂ and H₂O can be eliminated by the decomposition of hydrocarbon chains and the dehydration of -OH groups on the surfaces of TiO₂ NPs, respectively. This induces to interconnect the TiO₂ NPs through the Ti-O-C-O-Ti bonds. Finally, upon the UV-O₃ treatment, Ti-O-C-O-Ti bonds can be broken and the residual carbon can be eliminated in the form of CO₂ and O₂ [31], resulting in the further reduction of TiO₂ interparticle distance and forming a highly stable mp-TiO₂ network through Ti-O-Ti bonds. This induces the decoloring of the film, which can be attributed to the complete removal of the remaining organic residues, as shown in Fig. S2. Meanwhile, for the case of P_O, TiO₂ NPs are homogeneously distributed into the 1-octanol due to the physical adsorption as well as the hydrogen bonding interaction between the -OH groups of TiO₂ and 1-octanol. Upon heat treatment of the as-prepared P_O-based TiO₂ film at 150°C, 1-octanol can effectively bridge the gap between TiO₂ NPs by dehydration process. Finally, in the UV-O₃ treatment, the residual 1-octanol and carbon can be eliminated as CO₂ and O₂ (Fig. S2), thus, facilitating the formation of TiO₂ film with moderate stability.

Fig. 1a shows the viscosity of the TiO₂ pastes (P_BL, P_O, and P_C) and the mechanical strength of their corresponding TiO₂ films prepared onto the glass/FTO substrates by the doctor blade method. The corresponding photo-image of P_BL and P_O pastes is presented in Fig. 1b. The rheological behavior of the TiO₂ pastes revealed that the addition of CCl₄ with 1-octanol significantly increased the viscosity of the TiO₂ paste from 38,214 (P_O) to 53,158 (P_BL) cPs. This can be attributed to the hydrogen bonding between the TiO₂ NPs and the complex of CCl₄ and 1-octanol [30,32], which concurrently increase the maximum strength of adhesion of the 150°C sintered TiO₂ film from 192 (P_O) to 229 (P_BL) gf/mm². These values were 98245 cPs and 282 gf/mm² for P_C under 500°C sintering conditions. These results are consistent with the tape test results for the mechanical adhesion of P_BL compared to P_O (Fig. S3) on a glass/FTO substrate upon sintering at 150°C followed by subsequent UV-O₃ treatment. P_BL showed the highest level of adhesion without affecting the cross-cut areas (~0%), while P_O exhibited a much lower degree of adhesion by detaching the cross-cut areas of ~10%.

The morphology of TiO₂ films prepared from the P_BL and P_O was studied with FE-SEM. The TiO₂ film of P_BL, before heat treatment at 150°C, exhibited crack-free homogenous distribution of TiO₂ NPs with a relatively homogenous distribution of particle aggregation (Fig. S4a). In contrast, the TiO₂ film of P_O showed some cracks over the TiO₂ film with an enhanced particle aggregation by forming bigger local clusters over the film (Fig. S4b). After sintering at 150°C and subsequent UV-O₃ treatment, the TiO₂ film of P_BL showed high porosity without the presence of NPs aggregation (Fig. 2a), while the TiO₂ film of P_O exhibited lower porosity with the presence of bigger local clusters over the film (Fig. 2b). The carbon content in the TiO₂ film of P_BL was analyzed by EDS under different treatment conditions (e.g., drying at 25°C, 150°C, 150°C followed by UV-O₃ treatment, and 500°C), as shown in Fig. 2c. The corresponding EDS spectra are presented in Fig. S5. It was observed that the carbon content in P_BL decreased significantly after both sintering at 150°C and subsequent UV-O₃ treatment with a carbon content of 1.76%. This value for the RT dried and 150°C sintered P_BL-based TiO₂ films was 3.0 and 2.48%, respectively. Further, a small reduction of the carbon content at 500°C sintering condition (1.47 wt%) evidently showed that the additional UV-O₃ treatment is very critical to eliminating the carbon residues after sintering at relatively low temperatures. The thermogram in Fig. 2d exhibited a continuous weight loss up to 62 and 51.75%, respectively, for P_BL and P_O at 138°C, while no further significant weight loss was observed until 500°C (63.50 and 53.15%, respectively, for P_BL and P_O), consistent with the previous weight loss data. The little high residual percentage of P_O compared to P_BL can be ascribed to the incomplete removal of 1-octanol from P_O induced by its high T_b (194.5°C). These results demonstrated that both P_BL and P_O pastes are suitable for the preparation of binder-free f-DSSCs photoanodes under low-temperature sintering conditions with subsequent UV-O₃ treatment. Additionally, the crystallographic pattern of the TiO₂ NPs in the P_BL and P_O-based TiO₂ films exhibited that anatase is the major crystal phase (JCPDS card no. 21-1272) (Fig. S6) [33]. This is advantageous for improved dye-loading and the development of high-performance DSSCs [34].

Fig. 3a presents the photo-current density-voltage (J-V) plots of the g-DSSCs based on the TiO₂ photoanodes of P_BL and P_O, sintered at 150°C. The resulting PV parameters are summarized in Table 1. It was observed that the P_O cell showed the lowest PV performance with the short-circuit current density (J_sc), open-circuit potential (V_oc), and PCE of 8.45 mA/ cm², 0.73 V, and 4.2%, respectively. Meanwhile, the P_BL cell exhibited the PCE enhancement of 30% compared to the P_O cell with the J_sc, V_oc, and PCE of 12.07 mA/cm², 0.73 V, and 6.0%, respectively. The increase in the PCE of P_BL cell compared to the P_O device is mostly due to the increase in J_sc. This can be attributed to the improved dye loading in P_BL (2.51 × 10⁻⁸ mol/cm²) compared to the P_O (ca. 1.8 × 10⁻⁸ mol/ cm²) (Fig. S7a). The high visible transparency of the TiO₂ film of P_BL compared to the P_O film (Fig. S7b) is another possible reason for increased J_sc. Meanwhile, the 500°C sintered P_BL-based g-DSSC exhibited a PCE of 6.3%, which is comparable to the PCE of P_C cell (7.10%) (Fig. 3b and Table 1). The improved PV performance of the P_C-based g-DSSCs is due to the enhanced TiO₂ interparticle connection compared to P_BL, which corroborates a 24.32% higher J_sc. This result suggests that P_BL is promising to develop conventional high-temperature sintered g-DSSCs.

Finally, f-DSSCs were fabricated by utilizing P_BL and P_O on ITO/PEN substrate, and the resulting J-V plots and PV parameters are summarized in Fig. 3c and Table 1, respectively. Similar to the g-DSSCs, the fully flexible cell with P_BL exhibited a PCE of 3.5% with the J_sc, V_oc, and FF of 9.42 mA/cm², 0.70 V, and 52.3%, respectively, while the PCE was 2.0% for P_O with the J_sc, V_oc, and FF of 6.42 mA/cm², 0.68 V, and 46.8%, respectively. By inspection, the 75% PCE enhancement for P_BL-based f-DSSC compared to the P_O device is mainly due to the significant increase of J_sc (46.72% enhancement). This can be attributed to the improved TiO₂ interparticle connection and dye loading of the P_BL-based TiO₂ film compared to the TiO₂ film based on P_O. Nevertheless, The PCE of the P_BL-based f-DSSC is 71.40% lower than the g-DSSC prepared under similar conditions. This can be ascribed to the high ohmic resistance of the ITO/PEN substrate as well as the internal resistance of the devices, which corroborated a significantly low FF. The PCE of the P_BL DSSC is higher or comparable to the previously reported g-DSSCs and f-DSSCs prepared under low-temperature sintering conditions (Table S1), suggesting that P_BL can be effectively utilized for developing high-performance DSSCs with additional electrode engineering. Fig. 3d shows the IPCE spectra of the corresponding f-DSSCs based on P_BL and P_O. It can be observed that the external quantum efficiency (EQE) of the P_BL cell exhibited an upward shift over the wide wavelength range (400–800 nm) with the maximum EQE value of 54% at 540 nm, while this value for the P_O device was 41% at 540 nm. This result is consistent with the J_sc values of the corresponding f-DSSCs devices.

EIS spectra of the P_BL and P_O-based f-DSSCs were measured to investigate the electron transport mechanism and the interfacial properties. Fig. 4 shows the EIS spectra of the devices in the form of Nyquist plots, which exhibited three semicircles. The semicircle at the high-, mid-, and low-frequency regions corresponds to the charge transfer resistance at the CE/ electrolyte interface (R_ct1), charge transfer resistance at the TiO₂/dye/electrolyte interface (R_ct2), and the diffusion resistance (Z_diffusion) within the electrolyte, respectively [35]. The intercept with the real axis at the highfrequency region corresponds to the contact resistance (R_s) between the TiO₂/substrate or the ohmic resistance of the cells [35,36]. Generally, the R_s values do not alter much in a DSSC with similar substrates and paste. However, a little lower R_s for the P_BL cell (13.2 W) than the P_O-based f-DSSC (14.4 W) suggests the lower contact resistance between the P_BL TiO₂ film and ITO/PEN substrates. This can be attributed to the strong adhesion of the P_BL TiO₂ film with the substrate [36], consistent with the scotch tape and mechanical adhesion tests. The R_ct2 values for P_BL and P_O devices are 21.36 and 27.49 W, respectively. The lower R_ct2 designates the faster interfacial charge transfer process and lower recombination in the P_BL device compared to the P_O device [35], which is consistent with the PV parameters of their corresponding f-DSSC.