Home

Search

Collections

Journals

About

Contact us

My IOPscience

Poly(3-hexylthiophene):TiO2 nanocomposites for solar cell applications

This article has been downloaded from IOPscience. Please scroll down to see the full text article. 2004 Nanotechnology 15 1156 (http://iopscience.iop.org/0957-4484/15/9/008) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 147.8.31.43 The article was downloaded on 23/09/2011 at 15:41

Please note that terms and conditions apply.

INSTITUTE OF PHYSICS PUBLISHING Nanotechnology 15 (2004) 11561161

NANOTECHNOLOGY PII: S0957-4484(04)74605-X

Poly(3-hexylthiophene):TiO2 nanocomposites for solar cell applications

Department of Electrical and Electronic Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong 2 Department of Physics, University of Hong Kong, Pokfulam Road, Hong Kong 3 Department of Chemistry, University of Hong Kong, Pokfulam Road, Hong Kong E-mail: [email protected]

Received 28 December 2003 Published 1 July 2004 Online at stacks.iop.org/Nano/15/1156

Abstract The properties of organic/inorganic poly(3-hexylthiophene) (P3HT):TiO2 nanocomposite lms and nanocomposite based solar cells as a function of TiO2 concentration and the solvent used for the lm fabrication were studied. For low nanoparticle concentration (2030%) the device performance was worse compared to pure P3HT, while for nanoparticle concentration of 50% and 60% signicant improvements were obtained. P3HT photoluminescence quenching in 600800 nm spectral region changes by a factor of two for the increase in TiO2 concentration from 20% to 60%, while the AM1 power conversion efciency increases 35 times. Photoluminescence quenching and solar cell efciency were found to be strongly dependent not only on nanoparticle concentration but also on the solvent used for spin-coating. The changes in the lm and device properties were explained by the change in the lm morphology. For optimal fabrication conditions, external quantum efciency up to 15% and AM1 power conversion efciency of 0.42% were obtained. (Some gures in this article are in colour only in the electronic version)

1. Introduction
Organic/inorganic nanocomposites are very promising for applications in devices such as light emitting diodes, photodiodes, photovoltaic cells, and gas sensors [1]. Since the properties of a nanocomposite lm can be easily changed by varying its composition, such materials are highly versatile, while their fabrication shares the same advantages with traditional organic device technology, such as low cost production and the possibility of device fabrication on large area and exible substrates. Applications of different organicinorganic nanocomposites to organic solar cells [2 9] have been reported. Typically, the performance of a nanocomposite solar cell is improved compared to a pristine polymer one. External quantum efciency (EQE) over 54% and AM 1.5 power conversion efciency of 1.7% were achieved in poly(3-hexylthiophene) (P3HT):CdSe
0957-4484/04/091156+06$30.00 2004 IOP Publishing Ltd

nanocomposite solar cells [6]. An EQE up to 45% and power conversion efciency of 0.49% has been reported for a PbS quantum dot sensitized organic solar cell with nanocrystalline TiO2 electrode. However, the majority of the work on nanocomposite solar cells still reports efciencies below 0.1% [2, 5, 7]. Huisman et al [2] reported 0.06% efciency in TiO2 /poly(3-octyl)thiophene solar cells. Breeze et al [8] reported an EQE of 6.4% and cell parameters corresponding to a power conversion efciency of 0.002% for a cell consisting of a junction between a TiO2 nanoparticle layer and poly(2methoxy,5-(2 -ethyl-hexyloxy)- p-phenylenevinylene) (MEHPPV) sandwiched between indium tin oxide (ITO) and Au electrodes. In order to improve the efciency of TiO2 /polymer solar cells, it is necessary that the photogenerated excitons in the polymer reach the interface with TiO2 and successfully dissociate. One of the common methods to improve the exciton dissociation in organic photovoltaics is the use of bulk 1156

Printed in the UK

Poly(3-hexylthiophene):TiO2 nanocomposites for solar cell applications

heterojunction structures. Bulk heterojunction TiO2 -poly(2methoxy-5-(3 ,7 -dimethyl-octyloxy)- p-phenylene vinylene) (MDMO-PPV) photovoltaic cells have been reported, with EQE reaching 11% and power conversion efciency of 0.18% for 20% TiO2 [9]. While this result is an improvement over the results obtained with two layer architecture, further work is needed to improve the performance of these devices and increase the exciton dissociation and charge collection efciency. The performance of polymerinorganic nanoparticle bulk heterojunction solar cells will be strongly dependent on the charge transport properties of the composite as well as its morphology which will determine the exciton dissociation efciency. There have been several studies of the charge transport processes in organicinorganic nanocomposites [10 13]. However, both increase [13] and decrease [10] in mobility with increased nanoparticle concentration have been reported. The contradictory results are most likely due to differences in fabrication techniques, as well as the different polymers and nanoparticles used. The fabrication conditions can signicantly affect phenomena such as local electrical eld enhancement and chain alignment and thus affect the carrier mobility within the layer and the device performance. It is well known that the properties of organic donoracceptor blends and bulk heterojunction devices are strongly dependent on the lm morphology [14, 15]. However, studies on the inuence of the lm morphology on the electrical and optical properties of the organicinorganic nanocomposite lms and the performance of nanocomposite based devices have been scarce [16, 17]. In this work, we investigate the inuence of the nanoparticle concentration and the solvent used for the lm preparation on the optical properties of P3HT:TiO2 nanocomposite lms and the performance of P3HT:TiO2 based solar cells. The lms were characterized using photoluminescence (PL) and atomic force microscopy (AFM). The solar cells were characterized by measuring I V characteristics in the dark and under AM1 solar illumination and by determining the external quantum efciency (EQE) as a function of wavelength. The solar cell structure (ITO/PEDOT:PSS (poly(ethylenedioxythiophene) blended with polystyrene sulfonic acid)/P3HT:TiO2 /Al) and the energy levels of the materials used according to [1820] are shown in gure 1. The paper is organized as follows. In the following section, experimental details are given. In section 3, obtained results are presented and discussed. Finally, conclusions are drawn.

Al

P3HT:TiO2 PEDOT:PSS ITO

Glass substrate

3.2eV P3HT 4.2eV 4.7eV ITO 5.0eV 5.2eV TiO2 PEDOT:PSS 7.4eV
Figure 1. The structure of P3HT:TiO2 solar cell and the energy level diagram.

4.3eV Al

After the baking of PEDOT:PSS, 100 nm thick pure P3HT or P3HT:TiO2 lms were spin-coated. In order to achieve sufcient concentration of P3HT to fabricate 100 nm thick lms, the solutions were prepared in a heated ultrasonic bath (at 50 C) to improve solubility. The lms were then baked in a vacuum oven for 24 h at 110 C. The Al electrode was then evaporated in high vacuum. The pressure during evaporation was of the order of 104 Pa. The currentvoltage characteristics were measured using a Keithley 2400 sourcemeter. For white light efciency measurements, an Oriel 66002 solar simulator with AM1 lter was used. The light intensity was 100 mW cm2 . For the external quantum efciency measurement, an Oriel Cornerstone monochromator was used. Atomic force microscopy (topography and phase contrast) measurements were performed using a Digital Instruments Nanoscope III operating in dynamic mode. The room temperature photoluminescence was measured using an HeCd laser excitation source (325 nm) and Ar ion laser excitation source (514 nm). For both excitation wavelengths similar results were 2. Experimental details obtained for the characteristic two peak P3HT emission in the ITO glass substrates with surface sheet resistance 10 /square 550800 nm spectral range. were supplied by China Southern Glass Holding Co. Ltd, Shenzhen, China. ITO glass substrates were cleaned in an ultrasonic 3. Results and discussion bath using toluene, acetone, ethanol, and deionized water. PEDOT:PSS was obtained from Bayer AG; regioregular P3HT Photoluminescence quenching in a bulk heterojunction is was purchased from Aldrich. TiO2 nanoparticles with par- a useful indication of the degree of success of exciton ticle size 2040 nm (solid spherical particles, anatase/rutile dissociation. Figure 2 shows the PL spectra of 100 nm structures) were purchased from Nanopowder Enterprises. A thick lms for different concentrations of TiO2 nanoparticles. 30 nm layer of PEDOT:PSS was spin-coated on cleaned ITO It can be observed that even for low TiO2 concentrations, glass, followed by 24 h baking at 110 C in a vacuum oven. such as 20%, signicant PL quenching is obtained. With the 1157

C Y Kwong et al

0.02
0%
P3HT 20% 30% 40% 50% 60% 70%

10

Intensity (a.u)

60

EQE ( %)

Jdark (mA/cm )

40 20 0

40% 20% 70%

50% 30%

P3HT:TiO2

-0.5

0.0

0.5 Voltage (V)

1.0

1.5

0.1
60%

0% 40% 50% 60%

0.00 600 700 800

400

450

500

550

600

650

Wavelength (nm)
Figure 2. The photoluminescence spectra from P3HT:TiO2 layers for different TiO2 concentrations.

Wavelength (nm)
Figure 3. The comparison between external quantum efciencies (EQEs) of nanocomposite (40%, 50%, and 60% TiO2 ) and pure P3HT cells. The inset shows the corresponding dark current densities versus voltage. Table 1. Comparison of P3HT:TiO2 based solar cells for different TiO2 concentrations. TiO2 conc (%) 0 20 30 40 50 60 70 VOC (V) 0.8 0.3 0.24 0.38 0.46 0.44 0.34 ISC (mA cm2 ) 0.325 0.129 0.191 0.531 1.896 2.759 0.505 FF 0.230 0.274 0.269 0.344 0.321 0.356 0.396 (%) 0.059 0.012 0.013 0.076 0.274 0.424 0.067

increase of TiO2 nanoparticle concentration, the PL is further quenched. The intensities of the PL spectra for lms with 20% and 60% TiO2 differ by a factor two. The conductivity in the dark is also substantially improved for the composite lms, but it does not show signicant dependence on the TiO2 concentration. However, more signicant difference can be observed in the performance of solar cells with different TiO2 concentration. The obtained results are summarized in table 1, while gure 3 shows a comparison between the EQEs for different nanoparticle concentrations. For low nanoparticle concentration (20% and 30%), the cell performance is inferior to that of the pure P3HT, in spite of the signicant PL quenching. For 40% nanoparticle concentration, the cell performance becomes comparable to that of pure P3HT, while for 50% and 60% TiO2 signicant improvement is obtained. For nanoparticle concentrations higher than 70%, the cell performance deteriorates, since it is not possible to produce good quality uniform lms. The quenching of P3HT photoluminescence indicates that the excitons successfully dissociate at the P3HT:TiO2 interface. This is in agreement with the results obtained for TiO2 :PPV poly( p-phenylene vinylene) composites [11]. When the TiO2 concentration was above 20 wt%, most excitons dissociated at the interface of the polymer and nanocrystals, but the charge collection efciency was limited by recombination [11]. Therefore, successful exciton dissociation is not sufcient to ensure good photovoltaic cell performance. It is necessary to prevent the recombination of separated charges in order to achieve good photovoltaic performance. For low TiO2 concentrations, it is possible that electrons will be trapped at the dead end conduction paths and recombine with the holes in the polymer. With increased nanoparticle concentration, it can be concluded from the decreased PL intensity that the exciton dissociation is further improved, but more signicantly electron transport through the nanocrystal network without recombination with the holes in the polymer becomes more likely. Exciton dissociation and charge transport in nanocomposite solar cells have been modelled for several material systems [11, 12, 21]. In nanocomposite solar cells containing either a nanoporous electrode/polymer junction or a nanoparticle/polymer bulk heterojunction, the electrons are injected from the polymer into the conduction band of the semiconductor, so that the polymer acts as a sensitizer and hole transport 1158

layer [22]. Electron transfer from poly( p-phenylenevinylene) and polythiophene derivatives to TiO2 has been conrmed by photoinduced absorption and photoinduced electron spin resonance (ESR) measurements [23]. The separated carriers (electrons in TiO2 and holes in the polymer) can either be collected at the electrodes or recombine at the polymer/nanoparticle interface [22]. It is likely that more than one recombination process or reaction species contributes to the recombination dynamics [21]. Recombination of the separated charges was found to be a serious problem limiting the efciency of polymerTiO2 nanoparticle composites [11]. The charge collection efciency is dependent on the connectivity of the nanoparticle network and the distribution of the available paths [21]. Therefore, it is expected that the charge collection and thus overall device efciency will be strongly dependent on the nanocomposite lm morphology. In order to investigate the inuence of the lm morphology on the PL quenching and the photovoltaic cell performance, we prepared P3HT:TiO2 lms by spincoating using different solvents. It was shown that the performance of poly(2-methoxy-5-(2 -ethylhexyloxy)1,4-phenylenevinylene) (MEH-PPV)-C60 bulk heterojunction solar cells exhibits strong dependence on the solvent used [15]. In general, the performance of devices based on polymer blends is known to be strongly dependent on the lm morphology, and the highest efciencies are observed for devices with nanoscale phase separation [24]. Figure 4 shows the PL spectra of 100 nm P3HT:TiO2 thin lms with 60% TiO2 concentration fabricated by spin-coating from different solvents. The performance of solar cells for 60% TiO2

Poly(3-hexylthiophene):TiO2 nanocomposites for solar cell applications

0.08
Chlorobenzene CHCl3 THF Xylene

Table 2. Comparison of solar cells with 100 nm P3HT:TiO2 layer prepared from different solvents. Solvent THF CHCl3 Chlorobenzene Xylene VOC (V) 0.34 0.38 0.44 0.44 ISC (mA cm2 ) 0.246 74 0.646 84 1.192 8 2.758 6 FF 0.396 0.358 0.341 0.356 (%) 0.033 0.086 0.175 0.424

0.06

Intensity ( a.u.)

0.04

0.02

0.00 600 700 800

Wavelength (nm)
Figure 4. The photoluminescence spectra from P3HT:TiO2 layers spin-coated from different solvents.
16 14 12

EQE (%)

10 8 6 4 2 0 400
THF CHCl3 Chlorobenzene Xylene

450

500

550

600

650

Wavelength (nm)

Figure 5. The external quantum efciency (EQE) of P3HT:TiO2 cells with P3HT:TiO2 lms spin-coated from different solvents.

fabricated using different solvents is summarized in table 2, while gure 5 shows obtained EQE versus wavelength. The effects of solvent used for spin-coating on the solar cell performance were also examined for 50% TiO2 concentration, as well as for pure P3HT. For 50% nanoparticle concentration, similar trends can be observed as for 60% TiO2 . The best AM1 power conversion efciency (0.27% for 50% TiO2 ) was also obtained for lms spin-coated from xylene. For pure P3HT lms, however, no signicant effect of solvent on the cell performance was observed, and for all solvents used an efciency of 0.060.08% was obtained. It can be observed that the strongest PL quenching and the highest cell efciency is obtained for xylene. However, the PL of lms spin-coated from THF is more strongly quenched compared to the PL of lms spin-coated from chlorobenzene, and yet cells fabricated from chlorobenzene solutions exhibit 5 times higher power conversion efciency and 2 times higher EQE. Therefore, it can be concluded that the solvent used affects both the PL quenching and the recombination of separated charges, which is likely due to the different lm morphologies caused by the different solvents. It is known that the solvent and spin-coating conditions can affect the aggregation of

the polymer chains [25], as well as the quality of polymer metal contacts [26]. In order to examine the inuence of the solvent used on the lm morphology, lms spin-coated from different solvents were examined using dynamic mode AFM (topography and phase contrast). Phase contrast imaging enables mapping of the variations in the composition of the lm surface [27]. Phase contrast imaging is based on the principle that the phase lag of the tip depends on the material [27]. The measurement of the phase lag between the tip and the excitation signal while the amplitude is kept constant enables detection of the variation in the material properties [27]. Thus, phase contrast imaging enables us to investigate the degree of mixing and the domain sizes for the two components of the nanocomposite lms. The results obtained (for 60% TiO2 concentration) are shown in gure 6. The measurements were performed after baking the layers spin-coated on the top of baked PEDOT-PSS, i.e. the same preparation conditions used for devices. The solvent affects the surface roughness of the lms, but no direct correlation is observed between the surface roughness and the solar cell performance. Excessive surface roughness which would result in poor electrode contact, pinholes, and defects (such as in the case of the lm with 70% TiO2 ) should be avoided, but small variations in the surface roughness do not appear to have signicant effect on the device performance. The rms surface roughness values are similar for all lms, and vary from 20.5 nm for THF to 29.6 nm for xylene. No signicant differences can be observed from the topography images. However, phase contrast images reveal that the mixing between P3HT and TiO2 is strongly dependent on the solvent used. For the lms prepared from chlorobenzene and xylene, better mixing between TiO2 and P3HT was obtained compared to the lms spin-coated from chloroform and THF. The large number of small domains obtained in the former case yield a large interface and improved exciton dissociation. The results obtained are similar to those reported for MEH-PPV-C60 cells [15], where the difference between the performance of the cells fabricated from xylene, dichlorobenzene, chlorobenzene, THF, and chloroform was explained by preferential solvation of -electron conjugated segments in aromatic solvents. Unlike the results reported by Liu et al [15], our results did not show any signicant variation of the performance of pure P3HT cells on the solvent used for lm fabrication. Therefore, the mechanism proposed by Liu et al [15] is probably not the most signicant factor in determining the performance of P3HT:TiO2 solar cells. The main difference between the lms prepared from different solvents is the degree of mixing between P3HT and TiO2 , which is dependent on the solvating power of the solvent used and on the solvent evaporation rate. It is known that the solvent evaporation rates inuence the surface morphology of polymer lms [28]. THF and chloroform have one order of 1159

C Y Kwong et al

1 60

140nm

0.75

0.75

70 nm

0.5

0.5

30

0.25 0 nm 0 1 0

0.25

a)
0.25 0.5 0.75 1 m 0 0.25 0.5 0.75 0 1 m 1

60 50 nm 0.75 0.75

25 nm

0.5

0.5

30

0.25 0m

0.25

b)
0 1 0 0.25 0.5 0.75 1 m 0 0.25 0.5 0.75 0 1 m 1

50 nm 0.75

165 0.75

25 nm

0.5

0.5

82.5

0.25 0 nm 0 1

0.25

c)
0 0.25 0.5 0.75 1 m 0 0.25 0.5 0.75 0 1 m 1

150 50 nm 0.75 0.75

25 nm

0.5

0.5

75

0.25 0 nm 0 0 0.25 0.5 0.75

0.25

d)
1 m 0 0.25 0.5 0.75 0 1 m

Figure 6. AFM topography (left) and phase contrast (right) images of P3HT:TiO2 lms spin-coated from different solvents: (a) chloroform, (b) THF, (c) chlorobenzene, and (d) xylene.

1160

Poly(3-hexylthiophene):TiO2 nanocomposites for solar cell applications

magnitude higher vapour pressure compared to xylene [28], and hence evaporate signicantly faster than xylene and chlorobenzene. Also, since the solvating power of aromatic solvents like xylene and chlorobenzene for P3HT is higher, this could lead to a more extended polymer chain in the solid state and thus to different lm morphology. It should be noted that, while the best efciencies are generally obtained for the nanoscale phase separation, the solvent resulting in the nanoscale phase separation may be different for different material systems. Arias et al [24] obtained improved performance for their polymer bulk heterojunction cell when chloroform was used as a solvent compared to the cell prepared from xylene solution. However, in their case, the lms prepared from chloroform exhibited smaller scale phase separation compared to the lms prepared from xylene [24]. Therefore, the general guideline for improving the efciency of composite or blend solar cells is to achieve the phase separation on the nanoscale in order to improve the exciton dissociation efciency which would otherwise be poor due to the short exciton diffusion length. The morphology of the composite or blend layer is strongly solvent dependent [15, 24], but the optimal solvent for the small scale phase separation may differ for different materials used. Nanoscale phase separation as a rule results in efcient exciton dissociation, but this is not sufcient for the fabrication of efcient photovoltaic cells due to charge collection limitations [11]. Thus, the cells with 20% TiO2 show efcient photoluminescence quenching, but poor photovoltaic performance, in agreement with the previously reported results [11]. The solar cell performance can be signicantly improved by optimizing the TiO2 concentration to improve the charge collection by improving the connectivity of the nanoparticle network. However, further work is needed in order to improve the performance of these devices, mainly by inhibiting the recombination of the separated charges. This can be achieved either by changing the polymer or nanoparticles used, or by surface modications of the nanoparticles or the use of core-shell nanoparticles. It has been shown that the performance of nanostructured TiO2 cells can be improved with a suitably chosen recombination barrier coatings [29, 30].

performance was obtained for the cell with 60% TiO2 fabricated from xylene solution, with AM1 power conversion efciency of 0.42% and EQE reaching 15%.

Acknowledgments
This work is partly supported by Research Grants Council of the Hong Kong Special Administrative Region, China (Project Nos. HKU 7096/00P, 7075/01P, and 7056/02E).

References
[1] Godovsky D Y 2000 Adv. Polym. Sci. 153 163 [2] Huisman C L, Goosens A and Schoonman J 2003 Synth. Met. 138 237 [3] Breeze A J, Schlessinger Z, Carter S A, H rhold H H, o Tillman H, Ginley D S and Brock P J 2001 Proc. SPIE 4108 57 [4] Plass R, Pelet S, Krueger J, Gr tzel M and Bach U 2002 a J. Phys. Chem. B 106 7578 [5] Kymakis E, Alexandrou I and Amaratunga G A J 2003 J. Appl. Phys. 93 1764 [6] Huynh W U, Dittmer J A and Alivisatos A P 2002 Science 295 2425 [7] Arici E, Serdar Sariciftci N and Meissner D 2003 Adv. Funct. Mater. 13 165 [8] Breeze A J, Schlesinger Z, Carter S A and Brock P J 2001 Phys. Rev. B 64 125205 [9] van Hal P A, Wienk M M, Kroon J M, Verhees W J H, Slooff L H, van Gennip W J H, Jonkheijm P and Janssen R A J 2003 Adv. Mater. 15 118 [10] Ho P K H and Friend R H 2002 J. Chem. Phys. 116 6782 [11] Salafsky J S 1999 Phys. Rev. B 59 10885 [12] Huynh W U, Dittmer J J, Teclemariam N, Milliron D J, Alivisatos A P and Barnham K W 2003 Phys. Rev. B 67 115326 [13] Choudhury K R, Winiarz J G, Samoc M and Prasad P N 2003 Appl. Phys. Lett. 82 406 [14] Moons E 2002 J. Phys.: Condens. Matter 14 12235 [15] Liu J, Shi Y and Yang Y 2001 Adv. Funct. Mater. 11 420 [16] Huynh W U, Dittmer J J, Libby W C, Whiting G L and Alivisatos A P 2003 Adv. Funct. Mater. 13 73 [17] Kwong C Y, Djuri i A B, Chui P C, Cheng K W and sc Chan W K 2004 Chem. Phys. Lett. 384 372 [18] Fung M K, Lai S L, Tong S W, Chan M Y, Lee C S, Lee S T, Wu W W, Inbasekaran M and OBrien J J 2002 Appl. Phys. Lett. 81 1497 [19] Arango A C, Carter S A and Brock P J 1999 Appl. Phys. Lett. 74 1698 [20] Valaski R, Moreira L M, Micaroni L and H mmelgen I A 2002 u J. Appl. Phys. 92 2035 [21] Salafsky J S, Lubberhuizen W H and Schropp R E I 1998 Chem. Phys. Lett. 290 297 [22] Savenije T J, Vermeulen M J W, de Haas M P and Warman J M 2000 Sol. Energy Mater. Sol. Cells 61 9 [23] van Hal P A, Christiaans M P T, Wienk M M, Kroon J M and Janssen R A J 1999 J. Phys. Chem. B 103 4352 [24] Arias A C, MacKenzie J D, Stevenson R, Halls J J M, Inbasekaran M, Woo E P, Richards D and Friend R H 2001 Macromolecules 34 6005 [25] Shi Y, Liu J and Yang Y 2000 J. Appl. Phys. 87 4254 [26] Liu J, Guo T-F, Shi Y and Yang Y 2001 J. Appl. Phys. 89 3668 [27] Garcia R and P rez R 2002 Surf. Sci. Rep. 47 197 e [28] Strawhecker K E, Kumar S K, Douglas J F and Karim A 2001 Macromolecules 34 4669 [29] Lenzmann F, Nanum M, Kijatkina O and Belaidi A 2004 Thin Solid Films 451/452 639 [30] Zhang X-T, Liu H-W, Taguchi T, Meng Q-B, Sato O and Fujishima A 2004 Sol. Energy Mater. Sol. Cells 81 197

4. Conclusions
We have studied the properties of P3HT:TiO2 nanocomposite lms and the performance of the P3HT:TiO2 nanocomposite solar cells as a function of TiO2 concentration and the solvent used for the lm preparation. While signicant quenching of the P3HT emission is obtained for all nanoparticle concentrations, the performance of cells with low ( 40%) and high ( 70%) TiO2 concentration is comparable to or worse than that of cells with pure P3HT. This was attributed to the recombination of the separated charges for lower nanoparticle concentrations and poor lm quality for higher nanoparticle concentrations. For 50% and 60% TiO2 concentrations, however, signicant improvements are obtained. We also found that the PL quenching and the cell performance exhibit strong dependence on the solvent used, which was attributed to differences in the lm morphology, i.e. differences in the degree of mixing between P3HT and TiO2 . The best

1161