Efficient dye-sensitized solar cells using mesoporous submicrometer TiO₂ beads

Abstract

This work reports on mesoporous submicrometer TiO₂ beads as “building blocks” for loading multiple dyes. The multiple dyes separately adsorbed on individual beads can reduce the complications between different dye pairs in close proximity. When combination with zinc porphyrin LW4 and ruthenium complex KW1, the sensitized solar cells give an enhanced light absorption, and hence a high power conversion efficiency of 10.5% is achieved by the judicious consideration of processing parameters, including film nanostructure, proportion of pre-dyed TiO₂ beads, and film thickness.

---

Introduction

Dye-sensitized solar cells (DSSCs) have been intensely investigated as a promising candidate for low-cost, lightweight, and scalable solar cells.^1–4 A typical DSSC device consists of a dye-adsorbed photoanode, a Pt counter electrode, and an iodide/triiodide redox electrolyte. Photo-sensitizers such as ruthenium complexes, zinc porphyrins and metal free organic dyes have been developed as light harvesters and attained remarkable power conversion efficiencies (PCEs) of 10–13%.^5–10 However, there have been difficulties for single panchromatic dye that absorbs all the light from visible to infrared (IR) domain.^11,12 Therefore, co-sensitization has been proposed as an efficient method to obtain complementary light to extend the absorption spectrum, thus improving light utilization and the overall device efficiency.^4,13–16 The cocktail of sensitizers is known to be highly dependent on the nature of dyes used. Different dye pairs present complications, such as unfavourable adsorption and de-sorption as well as unfavourable interactions between the adsorbed dyes.^17,18 One solution is to separate the adsorption sites on the TiO₂ for various sensitizers, for example sensitization using two separated layers,^19,20 but they generally require complicated manufacturing processes.^21–23

Mesoporous TiO₂ beads possess dual functions of high dye loading and scattering effect because of their relatively high refractive index and comparative particle sizes to optical wavelengths, which make them as ideal candidates for DSSCs.^24–28 The mesoporous beads display a high diffuse reflection capacity in the entire visible and near-infrared regions from 400 nm to 800 nm, allowing it possible to achieve a high efficiency even by employing a thinner photoactive layer, which reduces consumption of TiO₂ nanocrystals and photoactive dye materials. Moreover, a “building block” concept is introduced based on mono-dispersed mesoporous TiO₂ beads for construction of photoelectrodes for DSSCs application.²⁹ The “building blocks” can be subjected to different pre-treatments such as pre-sintered and pre-sensitized prior to be fabricated as photoelectrodes. A favorable chemical connection between nanoparticles within the beads can be obtained by the pre-sinter processing, which decrease the negative effects of electron grain boundary crossing and is expected to be responsible for the extended electron lifetime and reduced recombination.³⁰ Furthermore, compared to the nanoparticles, the build blocks can be easily separated from the dye solution after sensitization process for the reason of mesoporous TiO₂ beads possessing diameters typically on the order of micrometer. Huang et al. firstly represented mesoporous TiO₂ beads photoelectrode for flexible DSSCs with multiple dyes in separate layers and achieved a PCE of 4.9%.³¹ The adsorption of multiple dyes in the separate multi-layer of nanocrystalline TiO₂ would be ideal for extending the absorption range of the solar spectrum with negligible adverse effects such as the unfavorable dye–dye interactions.³¹ However, in the case of nontransparent photoanodes made of sub-micrometer sized mesoporous spheres, the low energy photons, which are not absorbed by the dye in the first compartment, may be lost before being harvested in the second one due to the high diffuse reflection capacity of beads. Therefore, development of new methods for multiple dyes loading in the TiO₂ beads photoanodes is highly desirable.

Herein, an alternative method of fabrication sensitized photoelectrodes by using pre-sensitized TiO₂ beads aiming to improve the absorption of light over a broader wavelength region and eventually to improve the deice efficiency. Therefore, an overall PCE of 10.5% can be achieved by using zinc-porphyrin LW4 and ruthenium(II) complex KW1.^32,33 This paper also reports on the link between J_SC, spectral response and device efficiency for DSSCs based on nontransparent TiO₂ beads photoelectrodes.

Experimental

Materials

All solvents and reagents were of puriss grade quality and were used as received. Zinc-porphyrin LW4 and ruthenium(II) complex KW1 were synthesized as described in our earlier reports.^32,33

Synthesis of mesoporous TiO₂ beads

Mesoporous TiO₂ beads were prepared from a combined precipitation and solvothermal process.²⁸ In a typical procedure, 5.296 g HAD was dissolved in 800 mL ethanol, followed by the addition of 3.20 mL KCl aqueous solution (0.1 M). To this solution, 17.6 mL titanium(IV) isopropoxide (TIP, 97%, Sigma-Aldrich) was added under vigorous stirring at ambient temperature. The resulting white TiO₂ suspension was kept static at the same temperature for 18 h, and then centrifuged. The amorphous precursor TiO₂ beads were washed with ethanol three times and dried in air at room temperature. Secondly, to prepare mesoporous TiO₂ beads with a highly crystalline framework, a solvothermal treatment of the air-dried precursor beads was performed. Specifically, a 1.6 g portion of air-dried precursor beads underwent solvothermal treatment in an ethanol/water mixture (2 : 1 by volume) containing 1.0 mL of 25 wt% ammonia solution, at 160 °C for 16 h. The solid products were collected by filtration, washed with ethanol, and dried in air at room temperature. Such solvothermally treated TiO₂ beads were used to prepare the screen-printing slurry for fabricating the TiO₂ films or the pre-sensitized mesoporous TiO₂ beads.

Fabrication of DSSCs

The calcined TiO₂ beads powder was first dispersed in an absolute ethanol solution for 1 h under ultrasonication. A 0.3 mM dye solution was prepared by dissolving ruthenium(II) complex KW1 dye in a mixture of acetonitrile and tert-butanol solvent (1 : 1, volume ratio). To prepare the pre-sensitized mesoporous TiO₂ beads, these two solutions were mixed together and stirred overnight at room temperature in order to adsorb the dye molecules onto the TiO₂ beads surface. The mixed solution was centrifuged (Combi 514R) at 1000 rpm for 10 min for the separation of solvent and KW1 dye coated particles. The excess dye molecules were washed with ethanol. The same procedures were repeated to make the LW4 dye coated TiO₂ beads by mixing with 0.2 mM zinc-porphyrin LW4 dye solution in toluene/ethanol (1 : 1, volume ratio). Diverse proportions of pre-sensitized TiO₂ beads were mixed in the ethanol solvent to make slurry for the next sensitized photoanode fabrication. This slurry with a final solid loading of around 25 wt% and further used to make photoanode films by doctor-blading method. After mechanical pressing, the films were rinsed with ethanol and dried in air before making into devices. In this study, cold isostatic compression was carried out at a pressure of 100 MPa for 5 min. The cold isostatic press method was firstly reported by Cheng et al. and has been well reproduced in our laboratory.^34,35 The film thickness of the TiO₂ beads film was measured by a Profile-system (DEKTAK, VECCO, Bruker). Finally, the dye soaked film were stacked with Pt-based counter electrode and filled with iodine/iodide-based electrolyte. To prepare the sensitized photoanodes using the conventional sequential soaking method, single layer film of mesoporous beads were fabricated on fluorine doped tin-oxide (FTO) coated glass substrates by screen-printed. And then the TiO₂ bead films were sintered at 500 °C for 30 min. For dye loading, the electrodes were firstly immersed into 0.2 mM zinc-porphyrin LW4 dye bath in toluene/ethanol (1 : 1, volume ratio) solution for 4 h. Afterwards, the electrodes were transferred into 0.3 mM ruthenium(II) complex KW1 dye bath in acetonitrile/tert-butyl alcohol (1 : 1, volume ratio) solution for different time.

DSSC device characterization

The current–voltage (J–V) characteristics of the solar cells were measured using a digital source meter (Keithley model) under standard air mass (AM1.5) with a xenon lamp (Oriel, model 9119) coupled with optical filter (Oriel, model 91192), at illumination intensity of 100 mW cm⁻². A similar data acquisition system was used to control the incident photon-to-current conversion efficiency (IPCE) measurement. A white light bias (1% sunlight intensity) was used to bring the total light intensity on the device under test closer to operating conditions. The electronic impedance spectra (IS) of DSSCs were carried out with an Autolab Frequency Analyzer set-up, which consists of an Auto lab PGSTAT 30 (Eco Chemie B. V., Utrecht, The Netherlands) producing a small-amplitude harmonic voltage, and a frequency response analyzer module.

Results and discussion

Fig. 1a–c show the scanning electron microscopy (SEM) images of as-prepared precursor material and the calcined mesoporous TiO₂ beads prepared via the solvothermal process with ammonia solution. The precursor materials with spherical structure were highly mono-dispersed, which possess a uniform size of approximately 1000 nm and a smooth surface without any pores (Fig. 1a). By a subsequent hydrothermal reaction at 160 °C for 16 h, the amorphous spherical structures were converted to a relatively rough surface with a little shrinkage in the bead diameter of ∼850 nm (Fig. 1b and c). Fig. 2a shows the XRD patterns of the precursor material and the calcined TiO₂ beads after solvothermal treatment. The precursor material is amorphous. After the solvothermal and calcination treatment, the XRD patterns of these beads shows well-resolved diffraction peaks, indicating highly crystallized anatase structures without any impurity phase.³⁶ The BET analysis indicates a surface area of 75 m² g⁻¹ and a mean pore size of 23 nm calculated from N₂ adsorption data (Fig. 2b). By contrast, the N₂ adsorption–desorption isotherms and the corresponding pore size distribution curves of commercially available 20 nm TiO₂ (Dyesol, Australia) are shown in Fig. S1.† The surface area was determined to be 95.17 m² g⁻¹ and the pore size with a narrow distribution of 5–50 nm. These results indicate that the as-prepared TiO₂ beads show a mesoporous structure from the surface to the core and therefore the titania nanoparticles within remain highly accessible to both chemisorption of dye molecules and electrolyte diffusion.

Fig. 1 SEM images of (a) the as-prepared precursor material, (b) calcined mesoporous TiO₂ beads obtained after a solvothermal process, and (c) the monodispersed mesoporous TiO₂ beads with a diameter of 800–900 nm.

Fig. 2 (a) XRD patterns of the precursor material and the calcined mesoporous TiO₂ beads, and (b) nitrogen sorption isotherms and the corresponding pore diameter distributions of the calcined mesoporous TiO₂ beads.

Fig. 3a compares the diffuse reflectance spectra of the mesoporous TiO₂ bead films by varying their thickness. The film provides high diffuse reflection capabilities in the visible and near-infrared regions (from 450 to 800 nm), indicating that the incident light is significantly scattered within the film of mesoporous TiO₂ beads due to the comparable size to the wavelength of visible light. For comparison purposes, a ∼10 μm thick TiO₂ layer was formed utilizing nanoparticles. The measured diffused reflectance of this film is lower than 10% as shown in Fig. S2.† This is mainly attributed to the particle size (∼20 nm), which is much smaller than the wavelength of visible light. These results clearly demonstrated that the reflectance of the films composed of mesoporous TiO₂ beads show higher reflectance than the TiO₂ nanoparticle films in the range of 400–800 nm. The transmittance spectra of the thin films made from sub-micrometer mesoporous TiO₂ beads possess low transmittance in the visible and near-infrared regions, even for a few microns, further suggesting that the incident light is significantly blocked due to the strong scattering or reflection within the films as illustrated in Fig. 3b. These results should take into account for the construction of sensitized photoanode that multiple dyes with complementary light absorption are deposited in separate layers. For instance, as schematic illustration in Fig. 3c, the diffuse reflection significantly extends the traveling distance of light within the photoanode film and eventually enhances the interaction between TiO₂ and dye molecules. The incident photons that not absorbed by the dye in the first compartment may be lost before traveling to the second one due to the high diffuse reflection capacity of beads in the entire visible and near-infrared regions. Therefore, as illustrated in the schematic of Fig. 3d, we propose an alternative sensitized photoanode with multiple dyes adsorbed in discrete beads. The remarkable diffuse reflection in the beads layer can be significantly extend the traveling distance of light within the photoanode film and thus increases the opportunities for incident photons to be captured by the multiple sensitizers simultaneously. What's more, compared with the dye molecules randomly adsorbed on the TiO₂ films, the multiple dyes separately loaded in individual beads could reduce the opportunities of electron/hole recombination between different dyes in close proximity.

Fig. 3 (a) Diffuse reflectance spectra, (b) transmittance spectra of the photoanode consisting of mesoporous TiO₂ beads, (c) schematic of the light scattering effect, and (d) schematic of the sensitized photoanode with multiple dyes on the discrete beads.

Continuing our studies on DSSCs based on porphyrin and ruthenium dyes, herein we report on the utilization of zinc porphyrin (LW4) in combination with ruthenium complex (KW1) as sensitizers for the construction of DSSCs. The synthesis and characterization of the LW4 and KW1 dyes were reported earlier,^32,33 while their chemical structures and electronic absorption spectra measured in THF or DMF are displayed in Fig. 4. The KW1 dye dissolved in DMF solution displays two intense absorption bands centered at 392 nm and 538 nm with absorption coefficients of 4.17 × 10⁴ M⁻¹ cm⁻¹ and 1.84 × 10⁴ M⁻¹ cm⁻¹, respectively. By contrast, the push–pull zinc porphyrin LW4 dye shows typical absorption peak associated with Soret and Q bands, locating at 459 nm and 667 nm with absorption coefficients of 25.4 × 10⁴ M⁻¹ cm⁻¹ and 6.4 × 10⁴ M⁻¹ cm⁻¹. It is apparent that these two light absorption bands are complementary to the KW1 dye molecule, which with strong absorption in the range of 500–600 nm.

Fig. 4 The chemical structures of KW1 and LW4 and their absorption spectra recorded in DMF (red) and THF (green) respectively.

Fig. 5 shows a schematic representation of the method used in the fabrication of sensitized photoanode using the pre-dyed beads. The calcined beads are separately soaked in a 0.3 mM KW1 dye solution in acetonitrile and tert-butanol at room temperature for 12 h or a 0.2 mM LW4 solution in ethanol and toluene for 4 h, which both are the optimized conditions for the typical ruthenium and zinc porphyrin sensitizers.^32,33 And then these pre-dyed beads are centrifuged and re-dispersed in the ethanol solution with a final solid loading of around 25 wt%, which is used to make sensitized photoanode films. We firstly mixed the pre-dyed beads of LW4 (PDB-LW4) and KW1 (PDB-KW1) with different mass ratios of 1/0, 1/1, 1/2, 1/3 and 0/1, and then coated on substrates by a doctor blading method. Fig. 6 shows the appearances and colors of the TiO₂ photoanodes under different sensitizing conditions. The colors for the photoanodes consisting of individual PDB-KW1 and PDB-LW4 are red and green, respectively. The sensitized photoanode shows a tendency of gradual color change when the proportion of the PDB-KW1 and PDB-LW4 changes gradually. Fig. 7 shows the absorption spectra of the solution of multiple dyes after desorption from the mixed PDB-LW4 and PDB-KW1 films with different mass ratios. Obviously, with increasing the ratio of PDB-KW1 in the photoanodes, the absorption of light in the range of 500–600 nm increases while the absorption in the range of 650–700 nm remains unchanged, which is consistent with the unchanged content of PDB-LW4. On the whole, a broader range of light absorption can be noticed for the co-sensitized photoanodes with reference to those of the individual dyes. From this result, it is clear that the absorption spectra of sensitized photoanodes can be controlled by the mass ratio of PDB-KW1 and PDB-LW4.

Fig. 5 Schematic preparation of the sensitized photoanode with the pre-dyed beads of KW1 and LW4.

Fig. 6 The photograph of mixed PDB-LW4 and PDB-KW1 films with the mass ratios of 1 : 0, 1 : 1, 1 : 2, 1 : 3, and 0 : 1.

Fig. 7 Absorption spectra of the solutions of multiple dyes desorbed from mixed PDB-LW4 and PDB-KW1 films with different mass ratios.

In order to estimate the possibility of enhancing the light harvesting efficiency for the sensitized device, photovoltaic experiments were conducted employing TiO₂ beads photoanodes with different thicknesses. Initially, relatively thin TiO₂ beads films (with the thickness of ∼5 μm) were used to fabricate DSSCs, taking the advantage of high optical absorption coefficients of the KW1 and LW4 dyes combined with the high diffuse reflection capabilities of the beads. These cells employed a volatile acetonitrile-based electrolyte, containing 1.0 M 1,3-dimethylimidazolium iodide (DMII), 50 mM LiI, 30 mM I₂, 0.5 M tert-butylpyridine, and 0.1 M guanidinium thiocyanate (GNCS) in the mixed solvent of acetonitrile and valeronitrile (v/v, 85/15).^6,37 Typical values of the photovoltaic parameters open circuit voltage (V_OC), short circuit photocurrent density (J_SC) and fill factor (FF) of all three devices measured under 1 sun irradiation conditions (the PDB-KW1 for device A, the PDB-LW4 for device B and the mixed PDB-LW4 and PDB-KW1 with a mass ratio of 1 : 1 for device C) are summarized in Table 1. Encouragingly, device A (with PDB-KW1) exhibits a V_OC of 0.817 V, a J_SC of 8.14 mA cm⁻², and a FF of 0.75, giving an overall PCE of 5.0%. Under the same conditions, the J_SC of device B (with PDB-LW4) is about 8.99 mA cm⁻² with a PCE of ∼5.3%. As shown in Fig. 8a, device C (with mixed PDB-LW4 and PDB-KW1) exhibits a significant increase in the short circuit current density (J_SC = 10.84 mA cm⁻²), yielding an overall PCE of 6.5%. These results clearly revealed the advantages of the high light-harvesting absorbers in cooperation with non-transparent TiO₂ beads films for DSSC application. This observation most probably stems from the complementary light harvest relative to the individual KW1 and LW4 dye, which can be further verified from the corresponding IPCE spectra as exhibited in Fig. 8b. The IPCEs of device C (with mixed PDB-LW4 and PDB-KW1) is boosted in a broad spectrum and exhibits two humps in 400–500 nm and 600–700 nm when compared with the device A (with PDB-KW1). Obviously, these two bands are corresponding to the strong absorption of LW4, which exhibits two intense absorption bands around at 459 and 667 nm. On the other hand, compared with the device B (with PDB-LW4), the photocurrent generation of the device C in the range of 500–600 nm is greatly enhanced. As observed in Fig. 8b, the IPCE spectra of LW4 cells show typical characteristics of porphyrin-sensitized cells, exhibiting strong S-band signals in the higher energy region, intense Q-band responses in the near-IR region, and a spectral valley around 530–630 nm. In our previous study, we reported that KW1 and LW4 sensitized TiO₂ nanocrystal film (20 nm size, ∼3 μm thick) devices showed PCE of ∼5%.^32,33 They showed lower IPCE values than that of devices using TiO₂ beads, especially in the long wave length region. This could be due to the scattering property of the electrode film based on TiO₂ beads, which is important for the improvement of the light harvesting efficiency of the sensitized solar cells.³⁸ Therefore, the increased J_SC values for device C stemmed from the higher IPCE response, which covered almost the entire visible spectrum from 450 nm to 700 nm. The V_OC of device C (0.806 V) falls in between those of the devices sensitized with solely LW4 (0.787 V) and KW1 (0.817 V). Such phenomenon has been commonly observed in a co-sensitizing configuration.¹⁵ As evidently demonstrated by the dark current–voltage curves in Fig. 8a, device A based on PDB-KW1 showed much higher dark current onset as compared to the device B fabricated with PDB-LW4. This is indicative of an effectively inhibited charge recombination in the former. The reduction in the dark-current in device C relative to device B is consistent with the increased PDB-KW1 content.

Table 1 Detailed photovoltaic parameters of DSSC devices using the PDB-LW4, PDB-KW1, and mixed PDB-LW4/PDB-KW1 with film thicknesses of ∼6 μm. The devices were measured using a mask (area = 0.158 cm²)

Device	JSC^d [mA cm^−2]	VOC^d [V]	FF^d	PCE^d [%]
a PDB-KW1.b PDB-LW4.c PDB-mixed with KW1 and LW4 (1 : 1).d Average device performance parameters were calculated with a standard deviation from 10 piece of devices.
A^a	8.14	0.817	0.76	5.0
B^b	8.99	0.787	0.74	5.3
C^c	10.84	0.806	0.74	6.5

---

Fig. 8 (a) J–V characterization for device A (PDB-KW1), device B (PDB-LW4), and device C (mixed PDB-LW4/PDB-KW1) under AM 1.5G full-sunlight intensity and in the dark, (b) the photocurrent action spectra of devices A, B, and C, respectively.

To gain insight into the recombination dynamics of sensitized devices, we performed electronic impedance spectroscopy measurement (IS) in dark. As shown in Fig. 9a, the Nyquist plots at forward bias (0.7 V) exhibits two semicircles which corresponded to the electron exchange at the counter electrode/electrolyte interface at high frequencies (R₁, smaller semicircle), and the electron transfer at the TiO₂/dyes/electrolyte interface and transport in the TiO₂ film (R₂, bigger semicircle), respectively.³⁹ Apparently, R₂ increased from device B to C, and A, which suggested the reduced interfacial recombination in device C and A, which stem from better surface protection of the KW1 dyed TiO₂ beads.³³ Clearly, the device C has an moderate surface protection between the device A (PDB-KW1) and device B (PDB-LW4). Electron lifetimes (τ_e) were used to evaluate the interfacial charge recombination at the TiO₂ films sensitized with KW1 and LW4 dyes. A longer τ_e value of 21.02 ms is found in device A than that of device B (13.09 ms). This result is in consistency with the higher V_OC of the DSSC sensitized with KW1 dye. As for the device C, a τ_e value of 20.59 ms is found, which is consistent with the trends in photovoltage of the three devices.

Fig. 9 (a) Nyquist plots and (b) Bode plots of impedance spectra for DSSC devices A, B, and C in the dark with bias at −0.7 V.

In addition, the DSSCs incorporate PDB-KW1 and PDB-LW4 with different mass ratios of 1/1, 1/2 and 1/3 (labeled as device D, device E and device F respectively) were subsequently examined. It is notable that these devices possess an increased film thickness of about 10 μm for better light harvesting. Fig. 10a shows the current density–voltage (J–V) curves of these devices under the illumination of simulated AM 1.5G solar light (100 W cm⁻²) and the corresponding photovoltaic characteristics are presented in Table 2. Device D generates a PCE of 9.6% with short-circuit photocurrent density (J_SC) of 15.65 mA cm⁻², an open-circuit voltage (V_OC) of 0.804 V and a fill factor (FF) of 0.76. Comparison of the photovoltaic values of device C shows that J_SC is increased from 10.84 mA cm⁻² to 15.65 mA cm⁻² by changing the thicknesses of sensitized TiO₂ beads films from ∼5 to ∼10 μm. Optimizations show that device E consisting of PDB-KW1 and PDB-LW4 with a mass ratio of 1/2 give rise to the highest PCE of 10.5% with short-circuit photocurrent density (J_SC) of 17.3 mA cm⁻². However, when further increases the proportion of PDB-LW4 in the sensitized photoanode, device F generates PCE of 10.2% with a reduced short-circuit photocurrent density (J_SC) of 16.57 mA cm⁻². Fig. S3† shows the J–V curve of a DSSC using a ∼10 μm thick TiO₂ film prepared from 20 nm sized nanoparticles. This device shows an overall PCE of about 7.4% with short-circuit photocurrent density (J_SC) of 14.2 mA cm⁻². Apart from the V_OC and FF, the improvement in photovoltaic performance for the porous TiO₂ beads-based devices arises from a larger J_SC value or, in other words, due to the improved light harvest efficiency. The IPCE spectra of devices D (with PDB-LW4/PDB-KW1 (1 : 1)), E (with PDB-LW4/PDB-KW1 (1 : 2)) and F (with PDB-LW4/PDB-KW1 (1 : 3)) are illustrated in Fig. S4.† The broad bands covered almost the entire visible spectrum from 450 nm to 700 nm with the maxima of around 90% for these devices. These results indicate that the increased J_SC for device E stems from the higher IPCE response, which attains a broad plateau of over 80% in the region around 550 nm. These results clearly show that, upon co-sensitization, the PCEs of the devices are significantly influenced by the proportions of pre-dyed beads, which is consistent with effects of ratios of multiple dyes used in the co-sensitization system. Therefore, the significantly improved performance in device E is thus mainly due to the improved light harvesting with an optimized sensitization of PDB-KW1 and PDB-LW4 with optimized proportions.

Fig. 10 (a) J–V characterization for DSSC devices D, device E, and device F based on mixed PDB-LW4/PDB-KW1 with the mass ratios of 1 : 1, 1 : 2, and 1 : 3 under AM 1.5G full-sunlight intensity. (b) Absorption spectra of the multiple dyes using the first-dye coated films further soaked in second dye for 2, 4, 6 and 10 h.

Table 2 Detailed photovoltaic parameters of DSSC devices using the mixed PDB-KW1/PDB-LW4 with different portions of 1 : 1, 1 : 2, and 1 : 3 with film thicknesses of ∼10 μm

Device	JSC^d [mA cm^−2]	VOC^d [V]	FF^d	PCE^d [%]
a PDB-mixed with KW1 and LW4 (1 : 1).b PDB-mixed with KW1 and LW4 (1 : 2).c PDB-mixed with KW1 and LW4 (1 : 3).d Average device performance parameters were calculated with a standard deviation from 10 piece of devices.
D^a	15.65	0.804	0.76	9.6
E^b	17.3	0.796	0.76	10.5
F^c	16.57	0.798	0.77	10.2

---

Moreover, another conventional method to fabricate the co-sensitized photoanode by sequential soaking of anode film in dye solutions was also performed for comparison (see the Experimental section). As shown in Fig. 10b, the absorption spectra of the multiple dyes desorbed from co-sensitized films after dipping in the second dye solution for 2 h, 4 h, 6 h and 10 h demonstrate that unfavourable adsorption and desorption interactions were happened. Along with the dipping time changed from 2 h to 10 h in the second dye solution (KW1 dye), the characteristic absorption of KW1 dye molecule in the range of 500–600 nm is gradually increased, which indicates the increased KW1 dye adsorption. On the other hand, the light absorption bands around at 459 nm and 667 nm associated with Soret and Q bands of the LW4 dye are simultaneously decreased, which demonstrates strongly desorption of the LW4 dye happened at the same time. The photovoltaic performances of the DSSC devices prepared from this sequential soaking method are shown in Fig. S5.† These devices possess average PCEs ranged from 7 to 8%, depending on the dipping time for the second dye solution. It is noted that to obtain an efficient working electrode, the first dye adsorbed on TiO₂ should not come off while the second dye coating is processed. Otherwise, a simple replacement of the first dye occurs during the second dye adsorption process, which is not an ideal way for the application of two different dyes.¹¹ By contrast, for the sensitized solar cells using two types of pre-dyed beads (PDB-KW1 and PDB-LW4), which are prepared in the separated dye solutions individually, the unfavourable adsorption and desorption interactions are can be effectively avoided.

Conclusions

Mesoporous TiO₂ beads possessing dual functions of high dye loading and scattering effect have been prepared and employed as electrode for sensitized solar cells. Due to the unique characteristics of the mesoporous TiO₂ beads, the sensitized photoanode with mixed pre-dyed beads was demonstrated. The multiple dyes adsorbed in discrete beads reduce the opportunities of interaction between different dye molecules in close proximity. When combined with two important classes of molecules, zinc porphyrin LW4 and ruthenium complex KW1, the sensitized solar cells give an enhanced absorption than the individually sensitized electrode by LW4 or KW1, and hence resulted in a much improved PCE of 10.5%. This type of sensitization is anticipated to arouse broad interest in further boosting the efficiency of dye-sensitized solar cells by using these two important classes of molecules.

Acknowledgements

Financial support from the 973 Program of China (2011CBA00703, 2013CB922104, and 2014CB643506), and the NSFC (21173091, 21103057, and 21161160445) is gratefully acknowledged. We thank the Analytical and Testing Centre at the HUST and for characterizing various samples.

Notes and references

1. B. O'Regan and M. Grätzel, Nature, 1991, 353, 737.
2. U. Bach, D. Lupo, P. Comte, J. Moser, F. Weissortel, J. Salbeck, H. Spreitzer and M. Grätzel, Nature, 1998, 395, 583.
3. L. Han, A. Islam, H. Chen, C. Malapaka, B. Chiranjeevi, S. Zhang, X. Yang and M. Yanagida, Energy Environ. Sci., 2012, 5, 6057.
4. J. Shiu, Y. Chang, C. Chan, H. Wu, H. Hsu, C. Wang, C. Lin and E. Diau, J. Mater. Chem. A, 2015, 3, 1417.
5. W. Campbell, A. Burrell, D. Officer and K. Jolley, Coord. Chem. Rev., 2004, 248, 1363.
6. C. Chen, M. Wang, J. Li, N. Pootrakulchote, L. Alibabaei, C. Ngoc-le, J. Decoppet, J. Tsai, C. Grätzel, C. Wu, S. Zakeeruddin and M. Grätzel, ACS Nano, 2009, 3, 3103.
7. A. Mishra, M. Fischer and P. Bäuerle, Angew. Chem., Int. Ed., 2009, 48, 2474.
8. S. Mathew, A. Yella, P. Gao, R. Humphry-Baker, F. CurchodBasile, N. Ashari-Astani, I. Tavernelli, U. Rothlisberger, K. Nazeeruddin and M. Grätzel, Nat. Chem., 2014, 6, 242.
9. K. Kakiage, Y. Aoyama, T. Yano, K. Oya, T. Kyomen and M. Hanaya, Chem. Commun., 2015, 51, 6315.
10. Z. Yao, M. Zhang, H. Wu, L. Yang, R. Li and P. Wang, J. Am. Chem. Soc., 2015, 137, 3799.
11. S. Balasingam, M. Lee, M. Kang and Y. Jun, Chem. Commun., 2013, 49, 1471.
12. J. Yum, S. Jang, P. Walter, T. Geiger, F. Nuesch, S. Kim, J. Ko, M. Gratzel and M. Nazeeruddin, Chem. Commun., 2007, 44, 4680.
13. A. Ehret, L. Stuhl and M. Spitler, J. Phys. Chem. B, 2001, 105, 9960.
14. K. Sayama, S. Tsukagoshi, T. Mori, K. Hara, Y. Ohga, A. Shinpou, Y. Abe, S. Suga and H. Arakawa, Sol. Energy Mater. Sol. Cells, 2003, 80, 47.
15. G. Sharma, D. Daphnomili, K. Gupta, T. Gayathri, S. Singh, P. Angaridis, T. Kitsopoulos, D. Tasis and A. Coutsolelos, RSC Adv., 2013, 3, 22412.
16. S. Chang, H. Wang, L. Lee, S. Zheng, Q. Li, K. Wong, W. Wong, X. Zhu, W. Wong and X. Xiao, J. Mater. Chem. C, 2014, 2, 3521.
17. Z. Xue, L. Wang and B. Liu, Nanoscale, 2013, 5, 2269.
18. J. Lim, M. Lee, S. Balasingam, J. Kim, D. Kim and Y. Jun, RSC Adv., 2013, 3, 4801.
19. K. Lee, S. Park, M. Ko, K. Kim and N. Park, Nat. Mater., 2009, 8, 665.
20. F. Inakazu, Y. Noma, Y. Ogomi and S. Hayase, Appl. Phys. Lett., 2008, 93, 093304.
21. H. Choi, S. Kim, S. O. Kang, J. Ko, M. Kang, J. Clifford, A. Forneli, E. Palomares, M. Nazeeruddin and M. Grätzel, Angew. Chem., Int. Ed., 2008, 47, 8259.
22. P. Holliman, K. Al-Salihi, A. Connell, M. Davies, E. Jones and D. Worsley, RSC Adv., 2014, 4, 2515.
23. Y. Ooyama, K. Uenaka, T. Sato, N. Shibayama and J. Ohshita, RSC Adv., 2015, 5, 2531–2535.
24. F. Huang, D. Chen, X. Zhang, R. Caruso and Y. Cheng, Adv. Funct. Mater., 2010, 20, 1301.
25. D. Chen, L. Cao, F. Huang, P. Imperia, Y. Cheng and R. Caruso, J. Am. Chem. Soc., 2010, 132, 4438.
26. D. Hwang, H. Lee, S. Jang, S. Jo, D. Kim, Y. Seo and D. Kim, ACS Appl. Mater. Interfaces, 2011, 3, 2719.
27. F. Huang, D. Chen, Y. Chen, R. A. Caruso and Y. Cheng, J Mater. Chem. C, 2014, 2, 1284.
28. L. Heiniger, F. Giordano, T. Moehl and M. Grätzel, Adv. Energy Mater., 2014, 4, 1400168.
29. F. Huang, D. Chen, Q. Li, R. Caruso and Y. Cheng, Appl. Phys. Lett., 2012, 100, 123102.
30. A. Pascoe, D. Chen, F. Huang, N. Duffy, R. Caruso and Y. Cheng, J. Phys. Chem. C, 2014, 118, 16635.
31. F. Huang, D. Chen, L. Cao, R. Caruso and Y. Cheng, Energy Environ. Sci., 2011, 4, 2803.
32. J. Lu, X. Xu, K. Cao, J. Cui, Y. Zhang, Y. Shen, X. Shi, L. Liao, Y. Cheng and M. Wang, J. Mater. Chem. A, 2013, 1, 10008.
33. K. Cao, J. Lu, J. Cui, Y. Shen, W. Chen, G. Alemu, Z. Wang, H. Yuan, J. Xu and M. Wang, J. Mater. Chem. A, 2014, 2, 4945.
34. H. Weerasinghe, F. Huang and Y. Cheng, Nano Energy, 2013, 2, 174.
35. L. Ming, H. Yang, W. Zhang, X. Zeng, D. Xiong, Z. Xu, H. Wang, W. Chen, X. Xu and M. Wang, J. Mater. Chem. A, 2014, 2, 4566.
36. D. Chen, F. Huang, Y. Cheng and R. Caruso, Adv. Mater., 2009, 21, 2206.
37. M. Wang, P. Chen, R. Humphry-Baker, S. Zakeeruddin and M. Grätzel, ChemPhysChem, 2009, 10, 290.
38. J. Kim, H. Lee, D. Kim and Y. Seo, Adv. Mater., 2014, 26, 5192.
39. R. Zhou, Q. Zhang, E. Uchaker, J. Lan, M. Yin and G. Cao, J. Mater. Chem. A, 2014, 2, 2517.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra11281h

---