phys. stat. sol. (a) 202, No. 11, R125 R127 (2005) / DOI 10.1002/pssa.

200521149

Impedance spectroscopy on organic bulk-heterojunction solar cells

pss
www.pss-rapid.com

M. Glatthaar*, 1, N. Mingirulli2, B. Zimmermann1, T. Ziegler2, R. Kern2, M. Niggemann2, A. Hinsch2, and A. Gombert2

Freiburger Materialforschungszentrum (FMF), Stefan-Meier-Str. 21, 79104 Freiburg, Germany Fraunhofer Institut fr Solare Energiesysteme ISE, Heidenhofstr. 2, 79110 Freiburg, Germany

Received 13 July 2005, revised 5 August 2005, accepted 11 August 2005 Published online 16 August 2005 PACS 73.40.Ei, 73.61.Ph, 73.61.Wp, 84.60.Jt
*

Corresponding author: e-mail markus.glatthaar@fmf.uni-freiburg.de, Phone: +49 761 203 4782, Fax: +49 761 203 4801

We measured the electrical impedance spectra of organic bulk-heterojunction solar cells based on an absorber blend of poly(3-hexylthiophene) and [6,6]-phenyl C61-butyric acid methyl ester. Comparing the spectra of the non-treated device and after two consecutive treatments with applied forward

bias voltage at 110 C, we observed a region in the semiconductor with a low conductivity, which was expanding after the treatments. We concluded that this region is a depletion region at the aluminium contact. This was confirmed by the bias dependence of the impedance spectra.

2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction Organic bulk-heterojunction solar cells are based on an organic blend of two different materials, which form two interpenetrating networks. The one network usually consists of an optically active polymer, which can donate excited electrons to the other network, which consists of electron accepting molecules. The remaining vacancy (hole) on the donor network and the electron on the acceptor network are able to move like free charge carriers via hopping transport. The respective transport levels for electrons and holes are comparable to the valence band edge and the conduction band edge in an inorganic semiconductor. Hence, the organic blend is regarded as one effective semiconductor in the following. The simplest approach to realize a solar cell is to sandwich a semiconductor layer between two metal layers. Under illumination one metal should match the quasi-Fermi potential of the holes, the other the quasi-Fermi potential of the electrons. In the dark and without bias voltage there is an electrical potential drop across the semiconductor which compensates the work function difference of the two metal electrodes. For an undoped semiconductor the situation is shown in Fig. 1a. The electrical field is constant across the semiconductor because it contains very few charges. Thus the electrical potential drops linearly and the charge transport levels are straight tilted lines. For a p-doped semiconductor the situation is shown in Fig. 1b. At the region close to the

hole contact there is no electric field when the work function of the contacting metal matches the chemical potential of the holes in the semiconductor. This is a conductive region, as there are free holes. Close to the electron contact a space charge of the negatively charged acceptors leads to an approximately linearly increasing electric field towards the contact. This region has very few free charge carriers and is much less conductive. Such a device is called a Schottky diode. For organic bulk-heterojunction solar cells it is commonly assumed that the doping level is small enough, so that the former case with a constant field (Fig. 1a) should be valid [1]. Poly(3-hexylthiophene) (P3HT, donor) and [6,6]-phenyl C61-butyric acid methyl ester (PCBM, acceptor) are among the most promising materials for organic solar cells [2]. Our investigations with impedance spectroscopy have shown that the assumption of a low doping level and therefore an approximately constant field does not hold for devices based on a blend of P3HT and PCBM. Further we were able to observe a decrease of the doping level and an increase of the rectification behaviour when the devices were annealed under forward bias. This post-treatment is known to have a great effect on the photovoltaic performance [3, 4]. 2 Experimental details The devices were fabricated on indium tin oxide (ITO) covered glass substrates. First
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

physica

R126

status

solidi

M. Glatthaar et al.: Impedance spectroscopy on organic bulk-heterojunction solar cells

a) Figure 1 Diagrams for a) an undoped thin film diode and b) a p-doped Schottky diode at 0 V bias.

b)

an approximately 40 nm thick layer of poly(3,4 ethylenedioxythiophene) doped with poly(styrenesulfonate) (PEDOT:PSS) was spin coated from aqueous solution (Baytron P AI 4083, H.C. Starck). PEDOT:PSS is highly doped, so it is comparable to a metallic contact. On top the absorber blend (350 nm) was spin coated from chlorobenzene solution (~1 wt%) with a ratio of 1:1.5 wt. P3HT:PCBM (P3HT synthesized by Rieke Metals, PCBM synthesized by Nano-C). The top electrode is made of a 0.6 nm thick layer of lithium fluoride (LiF) followed by a 100 nm thick layer of aluminium (Al). Both layers were thermally evaporated under vacuum. The impedance spectra of the devices were recorded from 1 Hz to 1 MHz with a Zahner IM6 electrochemical workstation. The current voltage characteristic was measured in dark or at low illumination intensity with a halogen lamp. The low illumination intensity was used to avoid unintended heating of the samples. After preparation the cells were post-treated with two consecutive annealing steps and a simultaneously applied voltage. 3 Results and discussion The impedance spectra in this paper are represented by the modulus plot. In this plot an ideal RC element has the shape of a semicircle, where
(a) dark:
10

the diameter is the reciprocal capacitance 1/C. Two serially connected RC elements, R1C1 and R2C2, are still represented by a semicircle when R1C1 = R2C2. The diameter is then the sum of the reciprocal capacitances 1/C1 + 1/C2. But the more the products RC of the two elements differ, the stronger a neck occurs in the semicircle ending up in two separate semicircles. The diameter of each semicircle then is the reciprocal capacitance of the respective RC element. The semicircle of the RC element with the higher resistance R, will occur at the left in the spectrum. In Fig. 2 the currentvoltage characteristics of a nontreated device and the same device after two consecutive post-treatment steps are shown. Almost no rectification of the diode can be observed for the non-treated device. After a first treatment for 30 s at 110 C and 1 V forward bias, the rectification behaviour is established and a certain photovoltaic behaviour can be observed. After a second treatment for 7 min 30 s at 110 C and 2 V forward bias, the photovoltaic performance is drastically improved. In Fig. 3 the related impedance spectra in the dark are shown. We observe a strong deviation from the semicircle shape for the non-treated device. This indicates a continuous non-constant distribution of the charge carrier density across the semiconductor. After the first treatment an acute
8

Current Density [mA/cm ]

Current Density [mA/cm ]

untreated after 1st treatment after 2nd treatment

(b) Illuminated:
6

after 1st treatment after 2nd treatment

0,1

-2 0,01 -4

-1,0

-0,8

-0,6

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

1,0

-0,4

-0,2

0,0

0,2

0,4

0,6

0,8

Voltage [V]

Voltage [V]

Figure 2 (a) Current voltage characteristics in the dark of the untreated device, after the first post-treatment (30 s at 110 C and 1 V forward bias) and after the second post-treatment (450 s at 110 C and 2 V forward bias). (b) Current voltage characteristics under illumination after the first and after the second treatment.
2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Rapid Research Letter

0,3
2

Re(Z) [10 cm /F]

Re(Z) [10 cm /F]

untreated after 1st treatment after 2nd treatment

-0.9 V
0,2

0,1

0.3 V
0 0 1 2
7

Bias-Variation from -0.9V to 0.3V

0,0
3
2

Im(Z) [10 cm /F]

Im(Z) [10 cm /F]

Figure 3 Impedance spectra in the dark of the non-treated device, after the first post-treatment (30 s at 110 C and 1 V forward bias) and after the second post-treatment (450 s at 110 C and 2 V forward bias).

Figure 4 Impedance spectra after the two consecutive posttreatment steps at different bias voltages.

neck becomes visible, indicating that the semiconductor is divided into a small region with a low conductivity and a region with a high conductivity. After the second treatment the region with the low conductivity is much larger. Our interpretation for this effect is, that after fabrication the semiconductor is p-doped by acceptor impurities as described in literature [5]. Hence we would expect the device to be a Schottky-like diode. On the other hand it is noted for organic semiconductors, that injection from a metal even at high barriers becomes possible, when the semiconductor is highly doped [6]. This explains why there is no rectification for the non-treated device. The missing neck at the impedance spectrum, might be due to the fact, that the depletion region is much more conductive than expected for a Schottky contact [7], which is also indicated by the currentvoltage characteristic around 0 V. We interpret the effect of the post-treatment as a temperature enhanced vaporization of the acceptor impurities. There might also be a drift of the remaining impurities towards the PEDOT:PSS anode due to the applied bias voltage. Hence, the reduced doping level leads to a larger depletion region. From the reciprocal capacitance 1/C at the position of the neck the width d of the depletion region can be calculated: d = 0 A/C , (1) where 0 is the dielectric constant of the vacuum, the dielectric constant of the absorber blend, and A the electrode area. Assuming an of 3.9 we can estimate a width of 50 nm after the first and 100 nm after the second treatment. To prove that the low conductivity region is a depletion region, we measured the impedance spectra for different bias voltages. The results are shown in Fig. 4. The low conductivity region expands at increasing reverse bias voltages as expected for a depletion region.

4 Conclusions We conclude that the effect of posttreatment on P3HT:PCBM bulk-heterojunction solar cells cannot only be explained by an increased crystallinity of P3HT and therefore a changed absorption and improved hole mobility [3, 4]. In addition it was already assumed that the post-treatment causes a burning of shunts in the device [3]. We were able to render this effect more precisely by the observation of a reduced doping level at the Al interface, which leads to an improved rectification behaviour at contacts between organic semiconductors and metals. The photovoltaic efficiency might additionally be increased by the larger space charge region and a positive doping gradient towards the hole contact after the post-treatment, which leads to an improved charge separation and reduced electron recombination at the hole contact, as PEDOT:PSS probably does not perfectly block electrons.
Acknowledgement This work was supported by the European Union Project Molycell, project-Nr. SES6-CT-3003502783.

References
[1] C. Waldauff, P. Schilinsky, J. Hauch, and C. J. Brabec, Thin Solid Films 451 452, 503 (2004). [2] V. Dyakonov, Appl. Phys. A 79, 21 (2004). [3] F. Padinger, R. S. Rittberger, and N. S. Sariciftci, Adv. Funct. Mater. 13, 85 (2003). [4] Y. Kim, S. A. Choulis, J. Nelson, D. D. C. Bradley, S. Cook, and J. R. Durrant, Appl. Phys. Lett. 86, 063502 (2005). [5] M. M. Erwin, J. McBride, A. V. Kadavanich, and S. J. Rosenthal, Thin Solid Films 409, 198 205 (2002). [6] A. R. Hosseini, M. H. Wong, Y. Shen, and G. G. Malliaras, J. Appl. Phys. 97, 023705 (2005). [7] J. Singh, Semiconductor Devices Basic Principles (Wiley & Sons, New York, 2001), p. 238.

2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim