Assembly of CdS nanoparticles on boron and ﬂ uoride co-doped TiO 2 nano ﬁ lm for solar energy conversion applications †

A highly crystalline mesoporous boron and ﬂ uoride (B/F) co-doped TiO 2 nanomaterial is successfully synthetized using a facile process, followed by chemical bath deposition (CBD) in an organic solution to prepare the QD-cell to ensure high wettability and superior penetration ability of the B/F co-doped TiO 2 ﬁ lms. A modi ﬁ ed polysul ﬁ de redox couple, ((CH 3 ) 4 N) 2 S/((CH 3 ) 4 N) 2 S n , was employed in a cadmium sul ﬁ de (CdS) quantum dot (QD)-sensitized B/F co-doped TiO 2 solar cell covered with ZnS passivation layers with cobalt sul ﬁ de (CoS) as a counter electrode; then, an open-circuit photovoltage of 1.223 V, a high FF of 85.9%, a short-circuit photocurrent ( J sc ) of 4.52 mA cm (cid:1) 2 , and a high overall energy conversion of 4.74% were obtained.


Introduction
Quantum-dot-sensitized solar cells (QDSSCs), as a derivative of dye-sensitized solar cells (DSCs), have attracted extensive attention in recent years due to their higher theoretical conversion efficiency and lower production costs.Recently, a series of QDs (CdS, CdSe, PbS, CdSe x Te 1Àx , and CuInS 2 ) and their derivatives [1][2][3][4][5][6][7][8][9][10] were employed as light-absorber sensitizers.8][19] More importantly, QDs can harvest hot electrons, generate multiple electron-hole pairs and be designed with intermediate bands, which offer the opportunity to achieve considerable high performance solar cells. 20The theoretical conversion efficiency of QDSSCs is as high as 44%, 21,22 which makes the QDs prominent candidates for application in solar cells.However, the photoelectric conversion efficiency of QDSSCs is still lower than that of the DSSCs because of the severe recombination of electrons of the quantum dot conduction band.Recent advances in quantum dot solar cells (QDSCs) have propelled them to the forefront of the photovoltaic research, for power conversion efficiencies (PCEs) in excess of 11% that have been achieved by Zhong's group. 23This is a new record efficiency for QD solar cells in any conguration and also a new record for QDSSCs with any QD sensitizer.To date, the highest reported efficiency of CdS (QDSSCs) using I À /I 3 À based electrolyte and Pt counter electrode is 1.84%. 24n the past few years, there has been a considerable effort to enhance the energy conversion efficiencies of QDSSCs using various strategies.6][37] Chen et al. fabricated a novel double-sided CdS quantum-dot-sensitized TiO 2 nanotube (TNT)/ITO photoelectrode, and the experimental results show that the double-sided CdS quantum-dot-sensitized NT/ITO photoelectrodes show enhanced light absorption and achieve an optimum conversion efficiency of 7.5%, which is an enhancement of about 120% when compared with the single-sided CdS/TNT/Ti photoelectrode. 38In either DSCs or QDSSCs, the nanoparticle porous lm electrode plays a key role in the improvement of power conversion efficiency.However, TiO 2 is a large band gap semiconductor (E g ¼ 3.2 eV), which can only be excited by UV radiation at a wavelength below 390 nm.To make better use of the full light spectrum, doping TiO 2 with metal and nonmetal elements has been considered as a promising way to tailor the electronic properties of TiO 2 photoanodes in QDSSCs and has succeeded in improving photovoltaic performance of QDSSCs.In particular, non-metallic doping of TiO 2 electrodes has attracted much attention due to the excellent performance of the resulting doped material.Moreover, the technology of co-doping has been proven to be an effective solution to improve the performance of the TiO 2 electrodes due to its charge compensation on the basis of different ions through an internal charge transfer, with a large stabilization effect resulting in a less defective co-doped TiO 2 system. 39,40o the best of our knowledge, research into non-metallic B/F co-doping in QDSSCs has not been systematically carried out.Therefore, it would be highly desirable to research the codoping effect on QDSSC systems.Herein, we investigated the effect of B/F co-doping on the QDSSC's performance and demonstrated that the application of a B/F co-doped TiO 2 electrode in QDSSCs can enhance the photovoltaic performance of QDSSCs.In this study, we report the rst example of CdS quantum dot (QD)-sensitized B/F co-doped TiO 2 solar cells with ZnS passivation layers using CoS as a counter electrode.The conguration of the cells is illustrated in Scheme 1.As a result, a CdS QD-sensitized B/F co-doped TiO 2 solar cell using the same ((CH 3 ) 4 N) 2 S/((CH 3 ) 4 N) 2 S n electrolyte shows a promising photovoltaic performance, with an efficiency of 4.74%, a signicantly high V oc of 1.223 V, and a high FF of 85.9% with a short-circuit photocurrent (J sc ) of 4.52 mA cm À2 .

Preparation of B-F co-doped TiO 2 nanoparticles
The doped and modied nanocrystalline powder was synthesized using the sol-gel route, followed by pyrolysis and calcination treatments.Titanium tetraisopropoxide (Ti(Oi-Pr) 4 ) (Alfa Aesar), acetylacetone (acac, Aldrich), boron uoride acetic acid complex (BF 3 $2CH 3 CO 2 , 97%), and tetrauoroboric acid (HBF 4 , 48%, Strem Chemicals, Newburyport, MA) were used as received.An alcoholic complex was prepared by slowly adding BF 3 $2CH 3 CO 2 (0.03 mol) to 10 mL of EtOH at 0 C under ambient conditions, which formed a homogeneous and viscous solution.Once this solution cooled to room temperature, it was slowly added to Ti(OiPr) 4 (0.07 mol) in 10 mL iPrOH, producing a vivid yellow homogeneous solution.Aer the mixture was stirred for 15 min, deionized water (500 mL) was added, which brought no discernible change; hence 100 mL of HBF 4 was added, and aer a few minutes, the yellow solution became turbid.It was then reuxed at 80 C for 30 min, which effectively rehomogenized the mixture (pH: 1-2).Aer the mixture was reuxed for 6 h at 70 C, Ti(OiPr) 4 (0.069 mol) was added to the reaction vessel to facilitate particle formation, and reuxing was continued for 18 h at 70-80 C while stirring.The volume of the solution was reduced to approximately one-h, yielding a nely dispersed white powder.The powder was separated by ltration through a medium porosity ceramic frit, repeatedly washed with water until the pH of the ltrate was neutral and nally dried at 100 C in an oven.
The as-synthesized material was subjected to different thermal treatments.Specically, the puried powder was placed in a borosilicate test tube, which was attached to a glass manifold employing Cajon connectors and stainless steel tubing, and was heated to 400 C (4 C min À1 ).A short ceramic oven was employed to allow manipulation of the sample tube while it was still connected to the manifold.The sample's appearance was monitored visually over the course of the heating procedure, and the temperature was increased to 500 C while under vacuum or under gas ow for periods of up to 3 h.The heavily reduced powder, on the basis of its coloration (gray), was exposed briey to air by opening the vessel to the atmosphere or to pure oxygen gas that was delivered directly to the sample container.The nal stage of the thermal treatment consisted of the evacuation of the tube's contents or rapidly cooling it to 100 C while in air.

Deposition of CdS QD-sensitized B-F co-doped TiO 2 solar cells
The B-F co-doped TiO 2 (B-F-TiO 2 ) oxide used in the front subcell electrode was individually mixed with a polymer binder, and dissolved in terpineol, resulting in a paste.The paste was then coated on the uorine-doped (FTO) S n O 2 conductive glass (TEC15, 15 U per square, Pilkington, USA) by screen printing and then dried for 5 min at 125 C.This procedure was repeated 4 times (to give ca.8 mm lm).The electrodes were gradually heated at 500 C for 30 min, and the sintered lm was further treated with 40 mM TiCl 4 aqueous solution at 70 C for 30 min, washed with ethanol and water, and annealed at 480 C for 30 min.SILAR was used to assemble CdS-QDs into the B-F-TiO 2 photoelectrode.B-F-TiO 2 lms were surface-modied by immersing in 0.1 M mercaptoacetic acid (TGA) for 1 min; the TGA-modied B-F-TiO 2 substrate electrode was dipped rst into an ethanol solution containing Cd(NO 3 ) 2 (0.5 M) for 1 min, rinsed with ethanol, and dried with an air gun.The electrodes were then dipped for another 1 min into a Na 2 S methanol solution (0.5 M) and rinsed again with methanol.The process was repeated up to N cycles (N ¼ 1 to 5).These as-prepared electrodes are represented herein as "B-F-TiO 2 / TGA/CdS-N" electrodes. 41The CdS-sensitized B-F-TiO 2 lms were coated with ZnS for 3 cycles by dipping alternately into 0.1 M Zn(OAc) 2 methanol solution and 0.1 M Na 2 S solutions for 1 min per dip.As the working electrode, CoS was separated with a hotmelt Surlyn 1702 lm (25 mm, DuPont).Cleaned FTO glass was dipped into 0.5 M Co(NO 3 ) 2 ethanolic solution for 30 s, rinsed with ethanol, and dried with an air gun.The sample was then dipped for 30 s into 0.5 M Na 2 S methanolic solution, followed by rinsing with methanol and drying with an air gun.The process was repeated for up to ve cycles to obtain the CoS electrode. 42The electrolyte solution consisting of 0.01 M tetramethyl ammonium sulde ((CH 3 ) 4 N) 2 S, 0.002 M S, 0.02 M LiClO 4 , 0.02 M 4-tert-butylpyridine (TBP), 3-methoxypropionitrile (MPN) was used as a solvent for the polysulde electrolyte.The ((CH 3 ) 4 N) 2 S was prepared by heating tetramethyl ammonium hydroxide ((CH 3 ) 4 N) OH and ammonium sulde (NH 4 ) 2 S at 100 C. 43 We investigated three experiment samples with the same conditions ve times to test the repeatability and stability of the experiment, which gave excellent results.

Optical and photovoltaic measurements
Crystal structures of the samples were analyzed by X-ray diffraction (XRD) (Japan, XD-3A) using Cu Ka radiation and a scintillation counter detector.The patterns were recorded in the 2q range of 20-80 .The elemental distribution and concentration were analyzed by an energy dispersive spectrometer (EDS) (Oxford Inca) attached to the eld emission scanning electron microscope (FESEM).UVvis absorption spectra were recorded on a UV-visible spectrometer (U-4100, HITACHI, Japan).The electronic structure and B-Fdoping amount was determined by X-ray photoelectron spectroscopy (XPS) (AXIS-165 Shimadzu, Japan).Transmission electron microscopy (TEM) images were obtained using a JEOL JEM-2100 instrument.The at-band potential (V  ) of the nanostructured TiO 2 and B-F-TiO 2 electrode was determined by measuring the absorbance at 780 nm as a function of the applied potential.For the spectroscopic electrochemistry measurement, an 8 mm thick TiO 2 lm formed the working electrode (1 cm 2 surface area) of a three-electrode photoelectrochemical cell employing a platinum wire counter electrode and an Ag/AgCl reference electrode.Potential control was carried out on a CHI 660E potentiostat, and the applied potential was scanned at 5 mV s À1 .A 780 nm monochromatic light source was obtained from the UV-vis spectrophotometer (U-4100, HITACHI, Japan).For each determination of V  , a new working electrode and freshly prepared electrolyte solution were used. 44he irradiation source for the photocurrent density-voltage (J-V) measurement was an AM 1.5G solar simulator (16S-002, SolarLight Co. Ltd., USA).The incident light intensity was 100 mW cm À2 , calibrated with a standard Si solar cell.The novel solar cells were masked to a working area of 0.159 cm 2 .The J-V curves were obtained using a linear sweep voltammetry (LSV) method on an electrochemical workstation (LK9805, Lanlike Co. Ltd., China).The measurement of the incident photon-tocurrent conversion efficiency (IPCE) was performed using a Hypermonolight (SM-25, Jasco Co. Ltd., Japan) system.The short current density was calibrated by integrating the product of the IPCE value and solar photon density against the wavelength. 45Electrochemical impedance spectroscopy (EIS) for QDSSCs under dark, with a bias of À0.75 V, was measured using an impedance/gain-phase analyzer (ZENIUM, ZAHNER, Germany).The spectral regions were scanned using a frequency range of 10 À1 to 10 5 Hz at room temperature.The alternate current (AC) amplitude was set at 10 mV.

Optical properties
Fig. 1 shows that the absorbance intensity of these lms increases with noticeable shis of the absorbance edge towards longer wavelength, and a shoulder to the visible light region was observed with doped B and F, where the lm doped with B and F shows yellow colour while the undoped lm shows white colour.The UVvis absorption spectra of the B-F-doped TiO 2 /TGA/CdS-3 lm and the TiO 2 /TGA/CdS-3 lm in the ((CH 3 ) 4 N) 2 S/((CH 3 ) 4 N) 2 S n electrolyte also shied about 30 nm to longer wavelength, which is benecial for for the B-F-doped TiO 2 /TGA/CdS-3 lm to capture more light and to obtain higher light to electrical efficiency.

Spectro-electrochemical properties
In order to further investigate the cause of the enhancement of V oc in QDSSCs, the spectro-electrochemical measurements for undoped and B/F co-doped TiO 2 were carried out, and the results are shown in Fig. 2. The at-band potential V  of TiO 2 samples with and without B/F co-doping were determined as À2.2 V and À2.097 V (vs.Ag/AgCl), respectively.Therefore, a higher quasi-Fermi level closer to the conduction band was denitely achieved, which could lead to the larger V oc for liquidstate QDSSCs employing a B/F co-doped TiO 2 electrode.
Under Fermi level pinning, the two parameters (V oc and V  ) are linked by eqn (1). 46 where V red is the standard reduction potential of a redox couple, assuming that V red does not vary with the addition of boron and nitrogen.Therefore, a higher quasi-Fermi level closer to the conduction band is denitely achieved, which could lead to larger V oc for liquid-state QDSSCs employing B/F co-doped TiO 2 electrodes.With the co-doping of B/F in TiO 2 , its band gap was narrowed and the light capture range was signicantly extended; the excellent photoelectrochemical performance could be attributed to the ideal combination of retarded electron recombination and superior energy band structure from the unique B/F co-doped TiO 2 particle structure.

Device properties
New CdS-functionalized QDSSCs were fabricated in this study in which tetramethyl ammonium sulphide/polysulde, ((CH 3 ) 4 N) 2 S/((CH 3 ) 4 N) 2 S n , was used as the redox couple in an organic solvent (3-methoxy propionitrile (MPN)).The conguration of the cells is illustrated in Scheme 1.
The ll factor (FF) and the overall light-to-electrical energy conversion efficiency (h) of the DSSC were calculated according to the following equations: 47 where P in is the incident light power, and J max (mA cm À2 ) and V max (V) are the current density and voltage at the point of the maximum power output in the J-V curves, respectively.The cell based on B-F-TiO 2 /TGA/CdS-3 photoelectrode was found to give the best performance in this series.Under one sun illumination (AM 1.5G, 100 mW cm À2 ), this cell shows an ideal energy conversion efficiency of 4.74%, a large open-circuit photovoltage of 1.223 V and an extremely high FF value up to 85.9% with a short-circuit photocurrent (J sc ) of 4.52 mA cm À2 ; see Fig. 3. Compared with the cell based on TiO 2 /TGA/CdS-3, the J sc was improved from 3.54 mA cm À2 to 4.52 mA cm À2 , leading to an improvement in efficiencies from 3.45% to 4.74%, and the V oc improved from 1.124 V to 1.223 V; the open-circuit potential of the B-F-TiO 2 electrode is negatively shied by 99 mV compared to the undoped TiO 2 electrode, which is consistent with the at-band potential result.Moreover, using V oc as a reference is a standard way to describe the J-V characteristics of an electrochemical cell.V oc is the light intensity dependent rest potential of the cell.Connecting the cell to an electric load causes the electric current to ow in the cell, which gives rise to voltage losses, V k , that are also called overpotentials or overvoltages, due to the internal cell resistances R k .This decreases the cell voltage from its open-circuit value and determines the shape of the solar cell J-V curve and hence its ll factor. 48he IPCE (incident photon-to-current conversion efficiency, also referred to as quantum efficiency) of the QDSSCs was determined by measuring the short-circuit current under incident monochromatic light irradiated through a monochromator with a grating of 1200 grooves per mm (Shimadzu Corporation, Kyoto, Japan); light from a xenon lamp passing through a monochromator (PXJ43B11-6W, Japan) was focused onto the cell.The IPCE is calculated according to the following equations: 49,50 where J sc is the short-circuit current density (mA cm À2 ), V oc is the open-circuit voltage (V), P in is the incident light power, and J max (mA cm À2 ) and V max (V) are the current density and voltage in the J-V curves, respectively, at the point of the maximum power output.As shown in Fig. 4, the incident photon-to-electron conversion efficiency spectrum of a B-F-TiO 2 /TGA/CdS-N based cell is about 80% in comparison with the TiO 2 /TGA/CdS-3 based cell, which is over 60% at 400 nm; therefore, the higher IPCE for the QDSC with the B-F-TiO 2 /TGA/CdS-N electrode than that with the TiO 2 /TGA/CdS-3 electrode is due to the B-F doping in the former, and the IPCE response in the visible region is in agreement with the trend for the J sc .The active range in the visible light region is consistent with the corresponding UV-vis spectrum.The absorption edge of the B-F-TiO 2 /TGA/CdS-3based cell shis towards longer wavelength of about 30 nm compared to that of TiO 2 /TGA/CdS-3 based cell.

Electrochemical impedance spectra
Fig. 5 shows the electrochemical impedance spectra [51][52][53][54] for CoS and Pt counter electrodes with the same B-F-TiO 2 /TGA/CdS-3 lm under forward bias (À0.75 V) in the dark.Using the identical electrolyte, the reduction rate is primarily determined by the catalytic activity of the counter electrodes.The cells for EIS measurements are symmetric thin-layer sandwich cells.At high frequency (>10 5 Hz), the ohmic series resistance (R s ) of the FTO layer, the CoS (or Pt) layer and the electrolyte can be determined in the middle frequency range of 10 À1 to 10 5 Hz.In the low-frequency range of 0.1-10 Hz, the Warburg diffusion impedance, within the electrolyte, will be estimated.The R s value being 3.325 Ohm for the CoS electrode is lower than the 7.583 Ohm for the Pt electrode.

Conclusions
In summary, we have successfully developed the rst example of an efficient CdS QD-sensitized B/F co-doped TiO 2 solar cell with a modied polysulde electrolyte in a pure organic system.B/F doping into the TiO 2 lattice results in a red shi of the electronic absorption and enhanced photocurrent response.As a result, the performance of the cell is superb with an efficiency of 4.74%, a signicantly high V oc of 1.223 V, and a high FF of 85.9% with a short-circuit photocurrent (J sc ) of 4.52 mA cm À2 .The results show that the combination of B/F co-doped and CdS QD sensitization of the TiO 2 thin lms is an effective way to enhance the photoresponse, which is promising for photovoltaic and photoelectrochemical applications.

Scheme 1
Scheme 1 Schematic of the CdS QD-sensitized B/F co-doped TiO 2 solar cell structure.

Fig. 2
Fig. 2 Absorbance measured at 780 nm as a function of applied potential for nanostructured undoped and B/F co-doped TiO 2 electrodes at 298 K in 3 M KCl solution (Ag/AgCl reference electrode).

Fig. 5
Fig. 5 Electrochemical impedance spectra for CoS and Pt counter electrodes with the same B-F-TiO 2 /TGA/CdS-3 film, scanned from 10 À1 to 10 5 Hz at room temperature.The cells were measured at À0.75 V in the dark.The alternate current (AC) amplitude was set at 10 mV.