Chemical bath deposition of CdS quantum dots onto mesoscopic TiO2 films for application in quantum-dot-sensitized solar cells

(1)

Chemical bath deposition of CdS quantum dots onto mesoscopic TiO

₂

films for application in quantum-dot-sensitized solar cells

Chi-Hsiu Chang and Yuh-Lang Leea兲

Department of Chemical Engineering, National Cheng Kung University, Tainan 70101, Taiwan 共Received 25 May 2007; accepted 12 July 2007; published online 31 July 2007兲

Alcohol, instead of water, was used as a solvent in a chemical bath deposition process for the in situ synthesis of CdS quantum dots onto mesoporous TiO2films. Due to low surface tension, the alcohol solutions have high wettability and superior penetration ability on the mesoscopic TiO2film, leading to a well-covered CdS on the surface of mesopores. The CdS-sensitized TiO2 electrode prepared using the alcohol system not only has a higher incorporated amount of CdS but also greatly inhibits the recombination of injected electrons. The efficiency of a CdS quantum-dots-sensitized solar cell prepared using the present method is as high as 1.84% under the illumination of one sun共AM1.5, 100 mW/ cm2兲. © 2007 American Institute of Physics. 关DOI:10.1063/1.2768311兴

Recently, dye-sensitized solar cells 共DSSCs兲 have re-ceived significant attention because of their high efficiency and relatively inexpensive fabrication procedure compared with conventional inorganic solar cells. DSSCs are based on the photosensitization of nanocrystalline TiO2semiconductor electrodes by absorbed dyes.1–4One of the key factors deter-mining the efficiency of a DSSC is the light-harvesting prop-erty of the dye attached to the surface of the TiO2 nanocrys-tallite. Energy conversion efficiencies of up to 11% have been achieved using ruthenium complex as the sensitizer.4As an alternative, semiconductor quantum dots共QDs兲 which ad-sorb light in the visible region, such as CdS,5–7 CdSe,8 PbS,9,10PbSe,11and InP,12,13have been used as sensitizers of DSSCs. Due to the quantum confinement effect, the optical properties and the band gap of a QD can be adjusted by changing the size of the QDs.14–16It is also possible to utilize hot electrons to generate multiple electron-hole pairs per photon through the impact ionization effect.17 Another ad-vantage of the QD sensitizers over conventional dyes is their high extinction coefficient, which is known to reduce the dark current and increase the overall efficiency of a solar cell.

Although there have been studies on the synthesis and application of QDs for light harvesting in a DSSC cell, they are relatively few compared to those using organic dyes. Therefore, the photophysics and photochemistry of QDs are currently poorly understood and the energy conversion effi-ciency of a QD-sensitized DSSC is still low 共less than 0.5%兲.9

One of the reasons for the poor efficiency of a QD-sensitized DSSC is the difficulty of assembling the QDs into the mesoporous TiO₂matrix to obtain a well-covered mono-layer of QDs on the TiO2crystalline surface. Chemical bath deposition 共CBD兲 is the most common method used for in situ synthesis of the QDs in the mesoporous TiO₂ substrate. The other method, called the self-assembled monolayer 共SAM兲 technique, links the preprepared QDs to the TiO2 surface using a bifunctional molecule.8,18Although the SAM technique has the advantage of being able to control the size of the QDs, the coverage ratio of the QDs on the TiO2 sur-face is always poor. In a previous paper, a method combining the SAM and CBD techniques was proposed for assembling

CdS QDs, obtaining an energy conversion efficiency of 1.35%.7

In the CBD process used to synthesize the semiconduc-tor QDs onto mesoporous TiO2 films, the TiO2 films were dipped, in sequence, into aqueous solutions of the reactants 关e.g., Cd共NO3兲2and Na2S for CdS QDs兴.19The ionic state of the reactants 共Cd2+ _{and S}2− _ions_{兲 could thus penetrate the} TiO2 film and incorporate into the inner region of a meso-pore. However, a solution with a high surface tension, such as aqueous solution, is known to have a poor wetting ability on a solid surface, which also leads to poor penetrating of the solution in a porous matrix. Based on this knowledge, non-aqueous solutions with a lower surface tension are used in a CBD process to improve the penetration of the reacting so-lution and the assembling of the CdS QDs onto a mesoscopic TiO2 film. In the present study, this strategy is proven to greatly enhance the QDs assembly and the energy conversion efficiency of a QD-sensitized DSSC.

Commercially available indium tin oxide 共ITO兲共about 13⍀/sq兲 and F-doped tin oxide substrate 共FTO兲共about 8⍀/sq, Solaronix SA兲 were used as transparent conducting substrates to prepare the TiO2 photoelectrode. Mesoscopic TiO₂films were prepared by spin coating of TiO₂paste 共De-gussa P25兲 on the substrate, followed by sintering at 450 °C for 30 min. The thickness of the TiO2 film, measured from the cross sectional image from a scanning electron micro-scope, was about 5.5␮m. Instead of water, ethanol and methanol were used as solvents to dissolve Cd共NO3兲2 and Na₂S, respectively. The CBD process involved dipping the TiO2 film in a 0.5M Cd共NO3兲2 ethanol solution for 5 min, rinsing it with ethanol, and then dipping it for another 5 min in a 0.5M Na₂S methanol solution, and rinsing it again with methanol. The two-step dipping procedure is considered one CBD cycle. The incorporated amount of CdS can be in-creased by repeating the assembly cycle.

The CdS-modified TiO2electrode and a Pt-coated coun-terelectrode were sandwiched using 60␮m thick sealing ma-terial 共SX-1170-60, Solaronix SA兲. 3-methoxypropionitrile 共MPN兲 solution consisting of 0.1M lithium iodide, 0.05M iodine, 0.6M 1-propyl-2,3-dimethylimidazolium iodide 共DMPII兲, and 0.5M 4-tert-butylpyridine 共TBP兲 was used as the redox electrolyte of the DSSC cells. The active area of the cell was 0.16 cm2_{. The photocurrent-voltage}_{共I-V兲 curves} a兲_{FAX: 886-6-2344496; electronic mail: [email protected]}

APPLIED PHYSICS LETTERS 91, 053503共2007兲

0003-6951/2007/91共5兲/053503/3/$23.00 91, 053503-1 © 2007 American Institute of Physics Downloaded 19 Oct 2009 to 140.116.208.56. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp

(2)

were measured under an illumination of a solar simulator 共Newport, Oriel class A, 91160A兲 at one sun 共AM1.5, 100 mW/ cm2兲. An Eco Chemie Autolab potentiostat/ galvanostat was used to record the current-voltage 共I-V兲 characteristics. An Oriel 500 W xenon arc lamp and Keithley 2400 electrometer were used during the measurements of incident photon to current conversion efficiency共IPCE兲.

The amount of the CdS QDs incorporated in a TiO2film can be evaluated using the absorbance of the UV-vis spec-trum. The variation of the spectra with the CBD cycles for the alcohol solution is shown in Fig.1. The absorbance of the spectra increases with an increase of CBD cycles, indi-cating an increased adsorption amount of CdS. Furthermore, the redshifts of the absorption shoulder and onset position with increasing CBD cycles imply the growth of the CdS QDs. For the CBD process performed in aqueous solutions, the variation of the absorption spectra with CBD cycles has a similar tendency to that of the process performed in alcohol solutions. To compare the difference between the two sys-tems, the absorbance of the excitonic peaks after various CBD cycles for the two processes is shown in the inset of Fig.1. For the aqueous system, the absorbance moderately increases in the early cycles and then slows down gradually; a small increment can also be observed at the tenth cycle. For the alcohol system, the absorbance increases very quickly in the first two cycles and then slows down gradually, ap-proaching a constant value after four to five cycles. It is noteworthy that, the absorbance of the CdS assembled in the alcohol system after one CBD cycle is more than twice as higher as that assembled in the aqueous solution.

Because the adsorption amount of CdS QDs is propor-tional to the solid/liquid contact area, the large increment of absorbance in the early cycles of the alcohol system indicates a higher surface area available for the adsorption, attributable to the better penetration and wetting of the alcohol solution in the mesoporous TiO₂ matrix. The CdS QDs can adsorb into the inner region of the mesoporous film in the alcohol system, leading to a better-covered QD layer on the TiO2 surface. The TiO2 surface cannot be covered completely in one or two assembly cycles. The later CBD cycles, thus, play a role in replenishing the uncovered surface area, as well as increasing the QD thickness. The uncovered surface area of the TiO2 film decreases as CdS is absorbed, with the absor-bance decreasing in later CBD cycles. In the alcohol system, the approach of the equilibrium adsorption amount at earlier cycles共four to five cycles兲 indicates that a well-covered CdS QDs can be obtained efficiently using the present method.

The photocurrent-voltage共I-V兲 curves of the DSSCs pre-pared by the CdS-sensitized TiO2photoelectrodes were mea-sured under the illumination of one sun 共AM1.5, 100 mW/ cm2_{兲. The open circuit potential 共V}

OC兲, short circuit current共I_SC兲, fill factor 共ff兲, and the total energy conversion efficiency共␩兲 of these cells prepared in alcohol and aqueous systems are listed in TableI. After one CBD cycle, the en-ergy conversion efficiency␩is only 0.001% for the aqueous system, but a much higher value,␩= 0.31%, is obtained for the alcohol system. The higher ISCand fill factor values mea-sured for the alcohol system indicate a better coverage and higher incorporated amount of CdS on the TiO2 film, which can be attributed to the better penetration of the alcohol so-lutions in the mesopores. With an increase of CBD cycles, the overall efficiency first increases, and then decreases after reaching a maximum value. For the aqueous system, the maximum efficiency of␩= 0.70% was obtained at five CBD cycles. For the alcohol system, a much higher maximum efficiency of ␩= 1.26% was obtained at four CBD cycles, which is consistent to the rapid approach of the UV-vis ab-sorbance shown in Fig. 1. After reaching a maximum, the energy conversion efficiency decreases with additional CBD cycles. According to the data shown in TableI, the decrease in efficiency is mainly caused by the reduction of the fill factor due to the blocking of the mesopores by the additional loading of the CdS.

When a FTO substrate was used instead of ITO in the best performing device共four CBD cycles in alcohol system兲, the energy conversion efficiency was increased from 1.26% FIG. 1.共Color online兲 UV-vis absorption spectra of TiO2共0兲, and the TiO2

electrode after one cycle共1兲 to ten cycles 共10兲 of the CBD process in alcohol solutions. The inset shows the absorbance of excitonic peaks after various cycles of the CBD process for alcohol and aqueous systems.

TABLE I. Parameters obtained from the photocurrent-voltage共I-V兲 measurements of the DSSCs constructed using various electrodes.

Architecture UV-vis absorption ISC共mA/cm2兲 VOC共mV兲 ff ␩共%兲

ITO/ TiO2/ CBD1-aqueous 0.25 0.02 471.5 0.26 0.001

ITO/ TiO2/ CBD1-alcohol 0.56 0.99 573.0 0.55 0.31

ITO/ TiO₂/ CBD3-alcohol 0.82 2.52 643.5 0.64 1.04 ITO/ TiO2/ CBD4-alcohol 0.84 2.74 706.5 0.65 1.26

ITO/ TiO₂/ CBD5-alcohol 0.87 2.95 598.6 0.12 0.22 FTO/ TiO2/ CBD4-alcohol 0.84 4.30 680.8 0.63 1.84

053503-2 C.-H. Chang and Y.-L. Lee Appl. Phys. Lett. 91, 053503共2007兲

(3)

to 1.84%. The I-V characteristic of this device as well as the IPCEs measured from the ISCmonitored at different excita-tion wavelengths are shown in Fig.2. An IPCE value as high as 65%共at 400 nm兲 can be obtained for the alcohol system, which is much higher than the value obtained for the aque-ous system共also shown in the inset of Fig.2兲. It is worthy to

note that the performance of the device prepared by the present method 共␩= 1.84%兲 is better than the device pre-pared by coupling the SAM and CBD techniques 共␩ = 1.35%兲.7

In that work, the CBD was performed in aqueous solution, with the preassembled CdS QDs, self-assembled in pyridine, acting as a seed layer to induce the nucleation and growth of CdS in the CBD process. The seed layer was re-quired because the wetting ability of the aqueous solution was inadequate. The present results demonstrate that the seed layer is not necessary when an alcohol solution is employed because of the higher wetting ability of alcohol solution on the TiO2 mesoporous matrix.

The effect of the two assembly systems on the CdS / TiO2interfacial structure was studied by measuring the

I-V characteristic under dark conditions. The results are shown in Fig. 3. Comparing the best devices prepared in aqueous 共CBD5-aqueous兲 and alcohol 共CBD4-alcohol兲 sys-tems, the applied voltage required to drive the electrons across the photoelectrodes is higher for the alcohol system, attributed to the better-covered CdS on the TiO2 surface. This result shows that the CBD4-alcohol device has a supe-rior interfacial structure to inhibit the interfacial recombina-tion of the injected electrons from TiO₂ to the electrolyte,20,21 which is also responsible for its higher en-ergy conversion efficiency.

In summary, alcohol is an efficient solvent in the CBD process for in situ synthesis of CdS QDs onto the surface of a mesoporous TiO2 film. Due to the high wetting ability of an alcohol solution on the TiO2 surface, the solution can

deeply penetrate into the TiO2mesopores, leading to a well-coved CdS film. A CdS QD-sensitized DSSC prepared using the CBD in alcohol system has an efficiency as high as ␩ = 1.84%.

The support of this research by the National Science Council of Taiwan through Grant Nos. NSC-96-ET-7-006-001-ET and NSC-95-2120-M-006-001 is gratefully acknowl-edged.

1_{B. O’Regan and M. Grätzel, Nature}_{共London兲 353, 737 共1991兲.} 2_{A. Hagfeldt and M. Grätzel, Acc. Chem. Res. 33, 269}_共2000兲. 3_{M. Grätzel, Inorg. Chem. 44, 6841}_共2005兲.

4_{K. N. Mohammad, D. A. Filippo, F. Simona, S. Annabella, V. Guido, L.}

Paul, I. Seigo, T. Bessho, and M. Grätzel, J. Am. Chem. Soc. 127, 16853 共2005兲.

5_{L. M. Peter, D. J. Riley, E. J. Tull, and K. G. U. Wijayantha, Chem.}

Commun.共Cambridge兲 2002, 1030.

6_{K. G. U. Wijayantha, L. M. Peter, and L. C. Otley, Sol. Energy Mater. Sol.}

Cells 83, 263共2004兲.

7_{S. C. Lin, Y. L. Lee, C. H. Chang, Y. J. Shen, and Y. M. Yang, Appl. Phys.}

Lett. 90, 143517共2007兲.

8_{I. Robel, V. Subramanian, M. Kuno, and P. V. Kamat, J. Am. Chem. Soc.} 128, 2385共2006兲.

9_{R. Plass, P. Serge, J. Krüger, and M. Grätzel, J. Phys. Chem. B 106, 7578}

共2002兲.

10_{P. Hoyer and R. Könenkamp, Appl. Phys. Lett. 66, 349}_共1995兲. 11_{R. D. Schaller and V. I. Klimov, Phys. Rev. Lett. 92, 186601}_共2004兲. 12_{A. Zaban, O. I. Micici, B. A. Gregg, and A. J. Nozik, Langmuir 14, 3153}

共1998兲.

13_{J. L. Blackburn, D. C. Selmarten, R. J. Ellingson, M. Jones, O. Micic, and}

A. J. Nozik, J. Phys. Chem. B 109, 2625共2005兲.

14_{W. W. Yu and X. Peng, Angew. Chem., Int. Ed. 41, 2368}_共2002兲. 15_{X. Michalet, F. F. Pinaud, L. A. Bentolila, J. M. Tsay, S. Doose, J. J. Li, G.}

Sundaresan, A. M. Wu, S. S. Gambhir, and S. Weiss, Science 28, 538 共2005兲.

16_{W. W. Yu, L. Qu, W. Guo, and X. Peng, Chem. Mater. 15, 2854}_共2003兲. 17_{A. J. Nozik, Inorg. Chem. 44, 6893}_共2005兲.

18_{Y. J. Shen, Y. L. Lee, and Y. M. Yang, J. Phys. Chem. B 110, 9556}_共2006兲. 19_{R. Vogel, P. Hoyer, and H. Weller, J. Phys. Chem. 98, 3183}_共1994兲. 20_{J. Krüger, R. Plass, and M. Grätzel, Appl. Phys. Lett. 81, 367}_共2002兲. 21_{J. Krüger, R. Plass, L. Cevey, M. Piccirelli, and M. Grätzel, Appl. Phys.}

Lett. 79, 2085共2001兲. FIG. 2. I-V characteristics of the CdS-sensitized TiO2electrode measured

under the illumination of one sun共AM1.5, 100 mW/cm2_{兲. FTO was used as}

a substrate and the electrode was prepared using four CBD cycles in the alcohol system. The inset shows the incident photon to current conversion efficiencies共IPCEs兲 of the best performing devices prepared in alcohol 共four CBD cycles兲 and aqueous 共five CBD cycles兲 systems.

FIG. 3. I-V characteristics of the cells measured under dark conditions. The CdS-sensitized TiO2electrodes were prepared using four CBD cycles in

alcohol solution and five CBD cycles in aqueous solution, respectively.

053503-3 C.-H. Chang and Y.-L. Lee Appl. Phys. Lett. 91, 053503共2007兲