The TiO₂ nanorod arrays were grown directly on fluorine-doped tin oxide (FTO)-coated glass using the following hydrothermal methods: 50 mL of deionized water was mixed with 40 mL of concentrated hydrochloric acid. After stirring at ambient temperature for 5 min, 400 μL of titanium tetrachloride was added to the mixture. The mixture was injected into a stainless steel autoclave with a Teflon container cartridge. The FTO substrates were ultrasonically cleaned for 10 min in a mixed solution of deionized water, acetone, and 2-propanol with volume ratios of 1:1:1 and were placed at an angle against the Teflon container wall with the conducting side facing down. The hydrothermal synthesis was conducted at 180°C for 2 h.After synthesis, the autoclave was cooled to room temperature under flowing water, and the FTO substrates were taken out, rinsed thoroughly with deionized water, and dried in the open air.

In a typical SILAR cycle for the deposition of PbS nanparticles, the FTO conductive glass, pre-grown with TiO₂ nanorod arrays, was dipped into the 0.02 M Pb(NO₃)₂ methanol solution for 2 min then dipped into 0.02 M Na₂S solution (obtained by dissolving Na₂S in methanol/water with volume ratios of 1:1) for another 5 min. This entire SILAR process was repeated from 1 to 10 cycles to achieve the desired thickness of PbS nanoparticle layer. Similarly, for the CdS nanoparticle layer, Cd²⁺ ions were deposited from a 0.05 M Cd(NO₃)₂ ethanol solution, and the sulfide sources were 0.05 M Na₂S in methanol/water (50/50 v/v). For the hybrid PbS/CdS co-sensitized samples, the CdS deposition was carried out immediately after PbS deposition. The samples are labeled as PbS(X)/CdS(Y)-TiO₂, where X and Y refer to the number of PbS and CdS SILAR cycles, respectively.

The crystal structure of the CdS-TiO₂ and PbS-TiO₂ samples were examined by X-ray diffraction (XRD; XD-3, PG Instruments Ltd., Beijing, China) with Cu Kα radiation (λ = 0.154 nm) at a scan rate of 2°/min. X-ray tube voltage and current were set at 40 kV and 30 mA, respectively. The surface morphology and the cross section of the CdS-TiO₂, PbS-TiO₂, and PbS/CdS-TiO₂ nanostructures were examined by a field-emission scanning electron microscopy (FESEM; FEI Sirion, FEI Company, Hillsboro, OR, USA).

The solar cells were assembled using the CdS-TiO₂, PbS-TiO₂, and PbS/CdS-TiO₂ nanostructures as the photoanodes, respectively. Pt counter electrodes were prepared by depositing 20-nm Pt film on FTO glass using a magnetron sputtering. A 60-μm-thick sealing material (SX-1170-60, Solaronix SA, Aubonne, Switzerland) was pasted onto the Pt counter electrodes. The Pt counter electrode and a nanostructure photoanode were sandwiched and sealed with the conductive sides facing inward. A polysulfide electrolyte was injected into the space between two electrodes. The polysulfide electrolyte was composed of 0.1 M sulfur, 1 M Na₂S, and 0.1 M NaOH, which were dissolved in methanol/water (7:3 v/v) and stirred at 60°C for 1 h.

A solar simulator (model 94022A, Newport, OH, USA) with an AM1.5 filter was used to illuminate the working solar cell at light intensity of 1 sun (100 mW/cm²). A sourcemeter (2400, Keithley Instruments Inc., Cleveland, OH, USA) was used for electrical characterization during the measurements. The measurements were carried out with respect to a calibrated OSI standard silicon solar photodiode.

Figure 1a shows the typical FESEM images of TiO₂ nanorod arrays on an FTO-coated glass substrate, confirming that the FTO-coated glass substrate was uniformly covered with ordered TiO₂ nanorods. The density of nanorods was approximately 20 nanorods/μm² with suitable space for deposition of PbS and CdS nanoparticles. Figure 1b,c,d shows TiO₂ nanorods coated by PbS nanoparticles after 1, 3, 5 SILAR cycles, respectively. With the increase of SILAR cycles, the thickness of the PbS nanoparticles increased correspondingly. For the sample coated with 5 SILAR cycles, the space between the TiO₂ nanorods was filled with PbS nanoparticles, and a porous PbS nanoparticle layer was formed on the surface of the TiO₂ nanorods. As discussed later, this porous PbS layer can cause a dramatic decrease in photocurrent and efficiency for the solar cells.

Figure 2 shows the cross-sectional SEM images of PbS(3)/CdS(0)-TiO₂ and PbS(3)/CdS(10)-TiO₂ nanostructures. Compared with Figure 2a, a uniform protective layer of CdS was successfully deposited on the top of PbS nanoparticles. As we will discuss later, after the CdS coating, a remarkable enhancement of the cell performance and the photochemical stabilization of PbS sensitizer was observed. XRD patterns of the bare TiO₂ nanorod array, the PbS(3)/CdS(0)-TiO₂ nanostructure, and PbS(0)/CdS(10)-TiO₂ nanostructure were shown in Figure 3. As shown in Figure 3a, besides the diffraction peaks from cassiterite on structured SnO₂, all the other peaks could be indexed as the (101), (211), (002), (310), and (112) planes of tetragonal rutile structure TiO₂ (JCPDS no.02-0494). The formation of rutile TiO₂ nanorod arrays could be attributed to the small lattice mismatch between FTO and rutile TiO₂ 25 . Both rutile and SnO₂ have near identical lattice parameters with a = 0.4594, c = 0.2958, and a = 0.4737, c = 0.3185 nm for TiO₂ and SnO₂, respectively, making the epitaxial growth of rutile TiO₂ on FTO film possible. On the other hand, anatase and brookite have lattice parameters of a = 0.3784, c = 0.9514 and a = 0.5455, c = 0.5142 nm, respectively. The production of these phases is unfavorable due to a very high activation energy barrier which cannot be overcome at the low temperatures used in this hydrothermal reaction. As noted in Figure 3b,c, the as-synthesized CdS-TiO₂ nanostructure exhibited weak diffraction peaks of CdS at 2θ = 26.5°, 43.9°, 54.6°, and 70.1°, corresponding to the (111), (220), (222), and (331) planes of cubic CdS with the lattice constant a = 0.583 nm (JCPDS no. 89–0440). The diffraction peaks of as-synthesized PbS-TiO₂ nanostructure could be indexed as (111), (200), (220), (222), (400), (331), (420), and (422) planes, correspondingly, of cubic PbS with the lattice constant a = 0.593 nm (JCPDS no. 78–1901).

Figure 4 showed the photocurrent-voltage (I-V) performance of the sensitized solar cells assembled using PbS/CdS-TiO₂ nanostructured photoanodes. All the photocurrent-voltage performance parameters were summarized in Table 1. Solar cell sensitized by only CdS exhibits a short-circuit photocurrent density (J _SC) of 5.7 mA/cm² and an open-circuit voltage (V _OC) of 0.39 V. On the other hand, solar cell sensitized by only PdS present a poor photovoltaic performance with very low J _SC and V _OC. Optimal PbS SILAR cycles on this photoanode were investigated. As we can see from Figure 4b, with the increase of PbS SILAR cycles, a non-monotonic change of both J _SC and V _OC is recorded. Both J _SC and V _OC of the PbS-sensitized solar cells increase with the SILAR cycles first, and a maximum J _SC of 2.5 mA/cm² and V _OC of 0.3 V are obtained for the sample with 3 SILAR cycles. With further increasing PbS SILAR cycles, J _SC and V _OC decrease simultaneously, which demonstrates that a thick Pbs nanoparticles layer may hinder PbS regeneration by the electrolyte and enhance the recombination reaction. During the measurement, a continuous decrease of the current was observed, indicating the progressive degradation of PbS, which can be reasonably attributing to PbS oxidative processes. To protect the PbS nanoparticles from the chemical attack by polysulfide electrolytes, a uniform CdS layer was capped on the PbS-TiO₂ photoanode to avoid the direct contact of PbS with the polysulfide electrolyte. As shown in Figure 4c, under the same PbS deposition cycles, the cell with CdS capping layer presents both increased J _SC and V _OC, indicating that CdS QDs is indispensable to highly efficient PbS-sensitized solar cells. With the appearance of CdS layer, J _SC of the cell with 3 PbS SILAR cycles was improved from about 2.5 to 10.4 mA/cm², and the V _oc was increased from 0.3 to 0.47 V. The cell efficiency reached a promising 1.3%, indicating a five times increase, which is beyond the arithmetic addition of the efficiencies of single constituents (PbS and CdS). In addition to the increase of the cell performance for the co-sensitized configurations, a significant increase of the photochemical stability of PbS takes place with the presence of the CdS coating.

V _oc, open-circuit voltage; J _sc, short-circuit photocurrent density; FF, fill factor; η, energy conversion efficiency.

With further improvement of their performance, this kind of PbS/CdS co-sensitized TiO₂ nanorod solar cells may play a promising role in the future due to the following reasons: (1) The bandgap of PbS nanoparticles is quite small and extends the absorption band towards the NIR part of the solar spectrum, which will result in a high current density. (2) TiO₂ nanorod arrays grown directly on FTO conductive glass avoid the particle-to-particle hopping that occurs in polycrystalline mesoscopic TiO₂ films, which can also contribute to a higher efficiency. (3) TiO₂ nanorods form a relatively open structure, which is advantageous over the diffusion problems associated with the redox couples in porous TiO₂ network.

In our present work, the cell efficiency was still not high enough for practical application. The drawback limiting the energy conversion efficiency of this type of solar cells was the rather poor fill factor. This low fill factor may be ascribed to the lower hole-recovery rate of the polysulfide electrolyte, leading to a higher probability for charge recombination 26 . To further improve the efficiencies of these PbS/CdS-TiO₂ nanostructured solar cells, a new hole transport medium with suitable redox potential and low electron recombination at the semiconductor-electrolyte interface should be developed. Counter electrode was another important factor influencing the energy conversion efficiency. Recently, Sixto et al. 27 and Seol et al. 28 reported that the fill factor was clearly influenced by counter electrode materials where Au, CuS₂, and carbon counter electrode show better performance than Pt ones. Moreover, deposition of a ZnS passivation layer on the photoanode after the PbS/CdS sensitization would greatly eliminate interfacial charge recombination and improve the photovoltaic performance of PbS/CdS-TiO₂ nanostructured solar cells 29 . Further work to improve the photovoltaic performance of these solar cells is currently under investigation.