Vous êtes sur la page 1sur 5

Synthetic Metals 160 (2010) 445449

Contents lists available at ScienceDirect

Synthetic Metals
journal homepage: www.elsevier.com/locate/synmet

Investigation of annealing effects on microstructure of hybrid nanocrystalpolymer solar cells by impedance spectroscopy
WenBo Huang , JunBiao Peng, Li Wang
Institute of Polymer Optoelectronic Materials and Devices, Key Laboratory of Special Functional Materials, South China University of Technology, Guangzhou 510641, Peoples Republic of China

a r t i c l e

i n f o

a b s t r a c t
We have fabricated hybrid bulk heterojunction solar cells with blends of MEH-PPV (poly[2-methoxy,5(2-ethylhexoxy)-1,4-phenylene vinylene]) and nano-CdS as electron donor and electron acceptor, respectively. The effect of thermal annealing was investigated with impedance spectroscopy (IS) and capacitancevoltage (CV) characteristic measurements on devices. The IS and CV characteristic manifested very different before and after thermal annealing. We performed equivalent circuit to explain the effect of thermal annealing. The thermal treatment of nanocrystalpolymer lms is seen to aid in the formation of a continuous network for electron transport between nanorods, and hence improves devices performance. The method based on the IS is available to probe the microstructure of hybrid bulk heterojunction solar cells before and after thermal annealing, and therefore detect the mechanism for the annealing improvements. 2009 Elsevier B.V. All rights reserved.

Article history: Received 2 September 2008 Received in revised form 27 September 2009 Accepted 21 November 2009 Available online 22 December 2009 Keywords: Impedance spectroscopy Hybrid bulk heterojunction solar cells Nanocrystalpolymer lms Thermal annealing

1. Introduction In recent decade of research in polymer-based photovoltaic devices [1,2], the so-called bulk heterojunction structure where intimate mixing of electron-donating (p-type) and electronaccepting (n-type) molecules have generated much interest in developing photovoltaic cells because of the formation of microscopic heterojunctions between domains of the two materials throughout the bulk lm. However, the efciency of organic photovoltaic device is limited by the low carrier mobility of polymer donor and organic acceptors as compared to inorganic semiconductors. Inorganic nanocrystals as electron acceptor have been introduced into conjugated polymers to create hybrid bulk heterojunctions [3]. IIVI semiconductor nanoparticles have been researched for their extraordinary properties due to the possibility of their application in nanocrystal-conjugated polymer composite photovoltaic devices, such as CdS [4]. The band gaps and offsets of conjugated polymers and inorganic nanocrystals are such that charges will separate across an interface between them [5]. In nanocrystalpolymer systems not only requires good dispersion of the nanocrystals in the polymer donor phase which could create a large interfacial surface area for charge transfer between polymer and nanocrystals but also formation of a continuous network for electron transport between nanocrystals which could approach

high mobility removal of the electron through the nanophase, so that they may be extracted from the device without recombination. It is well known that processing parameters such as solvent and thermal annealing play important role in determination of performance of the hybrid lms. In recent years, it was found out that thermal annealing could further improve device performance [611]. However, no detailed information has been given on the effect of thermal treatment. In order to try and explain the detailed mechanism for the annealing improvements, the associated frequency response of the electrical properties was reported in this work. We report the effect of thermal annealing on the microstructure of hybrid bulk heterojunction solar cells made from blend lms of MEH-PPV and CdS nanocrystals by impedance spectroscopy (IS), which has proved to be a useful tool to analyze the microstructure in the lm of polymer doped with inorganic nanocrystals [12]. Frequency-dependent impedance data are obtained in the range from 10 Hz to 10 MHz and the devices are modeled with the equivalent circuits for the before and after thermal annealing device. The IS and capacitancevoltage (CV) characteristic are very different before and after thermal annealing result from the altering in microstructure of devices. From the equivalent circuit analysis, conducting mechanisms and dielectric behaviors of the unannealed and annealed device are discussed. 2. Experimental

Corresponding author. Tel.: +86 020 87112321x806; fax: +86 020 87110606. E-mail address: pswbh@scut.edu.cn (W. Huang). 0379-6779/$ see front matter 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.synthmet.2009.11.028

MEH-PPV was synthesized in our lab [13]. The CdS nanorods typically 40 nm long and 4 nm in diameter were synthesized in

446

W. Huang et al. / Synthetic Metals 160 (2010) 445449

our lab, too [14]. The hexadecylamine (HDA) on the surface of preprepared CdS was removed by reuxing with pyridine, as reported previously [14]. The samples were sandwich structure with Indium tin oxide (ITO)/poly (3,4-ethylenedioxythiopene) (PEDOT): poly (styrenesulfonate) (PSS)/active layer/Al. The ITO coated glass substrates are cleaned by ultrasonic treatment in deionized water, acetone, detergent and isopropyl alcohol sequentially. On top of the cleaned ITO glass, a 50 nm thin layer of PEDOT:PSS (Bayer Baytron P 4083) was spin-coated. Film thickness was measured by the surface prolometer (Alpha-Step 500, Tencor). After baking at 80 C for 10 h, the substrates are then transferred into a nitrogen-lled glove box. A 0.5 mL amount of 5 mg/mL MEH-PPV dissolved in pyridine and 0.5 mL of 30 mg/mL nano-CdS also dissolved in pyridine was mixed and then samples were fabricated by spin-coating on the top of PEDOT layer from mixed solutions containing CdS nanocrystals and conjugated polymers. Typically lm thicknesses were 100 nm. Finally, 150 nm-thick Al layer was thermally evaporated at the pressure less than 106 Torr on the MEH-PPV doped lm. The Al deposition rate (12 nm/s) and the lm thickness were monitored by a thickness/rate meter (Sycon). Once the electrodes were deposited, the devices were encapsulated in an inert atmosphere (dry nitrogen glovebox). The active area of the device is 1.2 cm2 . Devices annealing were carried out in ambient air at 150 C for 20 min. The IS was recorded by using HP4192A impedance analyzer with a four-terminal pair conguration in a frequency range from 10 Hz to 10 MHz and with an oscillating voltage of 50 mV under dc voltage 3 V. The CV characteristic measurement was also carried out by using HP4192A impedance analyzer but in a xed frequency of 1 kHz and with an oscillating voltage of 50 mV during dc voltage scanning from 5 V to 5 V with voltage steps of 0.1 V. All device measurements were performed at room temperature and in ambient air. 3. Results and discussion The frequency-dependent real (Re(Z)) and the negative of the imaginary (Im(Z)) parts of the measured IS biased at 3 V of CdS/MEH-PPV lms before and after thermal treatment at 150 C, respectively, are shown in Figs. 1(a) and 2(a). The ColeCole impedance plots based on the same data are replicated in Figs. 1(b) and 2(b), in which the implicit variable is the frequency increasing from right to left of the X-axis (from 10 Hz to 10 MHz). For the unannealed sample, the sign of Im(Z) turned into negative in the low-frequency region (10 Hz1 kHz) and the corresponding ColeCole impedance plot arcs back toward the origin, the corresponding arc elongating into the fourth quadrant, thus exhibiting an apparent negative capacitance (NC) effect or an inductance. The NC effect in the IS of hybrid bulk heterojunction solar cells could provide useful information on the carrier recombination and transportation [12,15]. In the Ref. [15], the NC effect in the low-frequency range was observed from four different samples, and it was assumed that the effect was associated with an additional recombination pathway which opened up at forward bias and reduced the charge accumulation of the solar cell. The NC phenomenon could be described by an equivalent circuit consisting of a resistance and an inductor (RL) according to the Ref. [12]. These conclusions can be extended to hybrid bulk heterojunction solar cells discussed in this work. As a result, here we t the experimenTable 1 Fitting parameters based on Eqs. (1) and (3) for unannealed device, respectively. Rs ( ) 10.8 10.8 R1 ( ) 12.9 12.9 C1 108 (F) 1.29 1.29 R2 ( ) 119 119 C2 108 (F) 2.46 A0 108 (F)/n / 10.2/0.89 Rrec ( ) 668.7 668.7 L1 (H) 0.9 0.9

Fig. 1. The frequency-dependent (a) real (open squares) and negative of imaginary (open circles) part of the impedance and (b) ColeCole plot of the unannealed device at the bias voltage of 3 V. Symbols are measured data. The solid and dashed lines are the tting data using the equivalent circuit (1) and (2) in the inset of (b), respectively. The equivalent circuit (2) ts the data well.

tal data by equivalent circuit to investigate the effect of thermal annealing. The physical signicance of equivalent circuit elements is similar to that discussed in Ref. [12] because that they are similar hybrid nanocrystalpolymer solar cells system. First, we t the data of unannealed device using the equivalent circuit (1) which is shown in the inset of Fig. 1(b) and its corresponding model equation can be expressed as: Z = R + iX = Rs + + 1 + iC2 R2 1 + iC1 R1
1

+ (Rrec + iL)

(1)

The tting parameters based on Eq. (1) for unannealed device is listed in Table 1. However, this circuit does not t the data well,

W. Huang et al. / Synthetic Metals 160 (2010) 445449

447

and orientations present. Since the response is a three-dimensional average of the response of many interacting grains, it can often be described by a distribution of relaxation times [16]. In order to more accurately model the IS of unannealed device, a CPE was introduced into equivalent circuit to better agree with the experimental data. The equivalent circuit (2) shown in the inset of Fig. 1(b) is used to t the measured data and its corresponding model equation can be expressed as: Z = R + iX = Rs + + 1 + iC1 R1
1

1 n 1 + (i) A0 + (Rrec + iL) R2

(3)

Fig. 2. The frequency-dependent (a) real (open squares) and negative of imaginary (open circles) part of the impedance and (b) ColeCole plot of the annealed device at the bias voltage of 3 V. Symbols are measured data and the solid lines are the tting data using the equivalent circuit in the inset of (b).

which is shown in Fig. 1(a) and (b) as solid line. It is evident that experimental data in the low and moderate frequency region do not yield a full semicircular arc in the ColeCole plot. It shows typical depressed circular arc compared to the t from the equivalent circuit model, which can be modeled as a constant phase element (CPE) [16]. The impedance of a CPE has the form: Z= 1 [A0 (i) ]
n

The tting parameters based on Eq. (3) for unannealed device is listed in Table 1. It should be noted that the tting curve is perfectly matched to the experimental data except the measured in low-frequency ranges (10 Hz1 kHz), which is shown in Fig. 1(a) and (b) as dashed line. The experimental data in the low-frequency region show typical depressed circular arc, but we cannot use a CPE to better t the experimental data because that CPEs only can exemplify distributions that arises from distributed of RC relaxation times. Although there have been dispersive behavior in the lowfrequency region for RL relaxation times, we are more concerned about the emergence of negative capacitance in this frequency region because that NC effect is an important feature to explain poor transport pathways in a blends system. We attributed NC effect to the additional recombination pathway which opened up at forward bias as mentioned in Ref. [15]. For the unannealed sample, interfacial pyridine of nanocrystalpolymer systems results in inadequacy interpenetrating network of nanocrystals CdS. Therefore, electrons do not have connected pathways to the appropriate electrode so that they are likely to recombine with holes before being extracted from the device. That is to say, there are additional recombination pathways in unannealed sample. This is the reason why the presence of pyridine helps in appearance of NC effect. For the annealed sample, the overall IS looks different compared to the untreated case, i.e., the Im(Z) values in the impedance spectroscopy are always positive even in the low-frequency region, as shown in Fig. 2(a). Three well-resolved semicircles were detected in ColeCole plot (Fig. 2 (b)). It should be noted that the uncompleted right-hand semicircle in the plot (in low-frequency ranges) is due to the instrument limitation. Here we t the data of annealed device using the equivalent circuit which is shown in the inset of Fig. 2(b) and its corresponding model equation can be expressed as: Z = R + iX = Rs + + 1 + iC1 R1
1

(2) 1 1 + iC2 + R3 + R2 iC3

1 1

when this equation describes a capacitor, the constant A0 = C (the capacitance) and the exponent n = 1. It describes a resistor for n = 0. For a constant phase element, the exponent n is less than one. CPEs are distributed elements and can exemplify distributions that arises because microscopic material properties are themselves distributed, such as interface defects, distribution and nonuniform nature of the space charge in three dimensions [16]. In particular, in the polycrystalline devices, there is a distribution of grain sizes

(4)

The tting parameters based on Eq. (4) for annealed device is listed in Table 2. In Fig. 2(b), the open circles are the measured points and the solid line is t that result from the above equivalent circuit model in the frequency range from 10 Hz to 10 MHz. It is noticed that the t is good agreement of the experimental data over all the measured frequency range.

Table 2 Fitting parameters based on Eq. (4) for annealed device. Rs ( ) 12.4 R1 ( ) 10 C1 109 (F) 9.33 R2 ( ) 112.7 C2 108 (F) 1.55 R3 ( ) 146 C2 104 (F) 3.1

448

W. Huang et al. / Synthetic Metals 160 (2010) 445449

The apparent modication of the IS after annealing suggests that the electrical relaxation mechanism of annealed sample is different from that of unannealed one. Huynh et al. [7] suggested that the thermal treatment of nanocrystalpolymer lms was seen to aid both in the removal of interfacial pyridine and in bringing nanorods closer together. Our measurement of IS supports these earlier ndings, since the closer neighboring nanorods will be benecial to form a continuous network for electron transport between nanorods, and hence result in the disappearance of NC effect [12]. Meanwhile, the thermal treatment of nanocrystalpolymer lms also aid in the removal of polymernanocrystals interfacial pyridine, hence being benecial to create larger donoracceptor interface area which could improve light harvesting. The reason for the disappearance of CPE element in the equivalent circuit for the annealed device may be due to the reorganization of the inorganic nanocrystals/polymer interface [11] and to reduce the free volume and the density of defects at the interface during evaporation of the solvent [6]. On the other hand, the glass transition temperature of MEH-PPV is around 70 C, and the annealing is being done at 150 C. Annealing above the glass transition temperature of polymer causes reorganization of the polymer and increases its crystallinity, which is also responsible for enhancement of hole conductivity in the polymer. This is also seen to aid in the disappearance of NC effect. In order to further study the effect of thermal annealing on the CdS-doped MEH-PPV blend lms, the dark and illuminated capacitancevoltage CV measurements were carried out on both before and after thermal annealing CdS-doped MEH-PPV samples. Fig. 3 shows the CV characteristics for the two devices under dc bias voltage range from 5 V to +5 V with voltage steps of 0.1 V, superimposed on an ac voltage of 50 mV and a constant frequency of 1 kHz. It is apparent that the thermal annealing strongly inuences the voltage dependence of the cell capacitance in forward bias. At reverse bias, the dark capacitances of the two devices remain constant. The constant capacitance at reverse bias suggests that there exists a depleted region across the blend lms of MEH-PPV and CdS nanocrystals in the device. From Fig. 3 it can be observed that the dark and illuminated capacitances of unannealed sample increase in forward bias until a point where it reaches a maximum value, and decrease after that. This is similar to the previously reported results by Shrotriya and Yang [17]. Shrotriya and Yang suggested that the increase in the capacitance value associated with charge injection from electrodes into the polymer in forward bias and the decrease is due to recombination of electrons and holes in the polymer. This result is in accordance with the NC effect, which

is associated with an additional recombination pathway, in the IS of hybrid bulk heterojunction solar cells [15]. On the contrary, the dark and illuminated capacitances of annealed sample remain almost constant in forward bias. This result suggests that there is a decrease of recombination pathway in annealed device. Additionally, the dark and illuminated capacitances of annealed sample are signicantly higher than that of unannealed sample. From Fig. 3 also it can be observed that the dark and illuminated capacitances are about 45 nF and 140 nF for the unannealed sample in forward bias, respectively, and increases to about 135 nF and 350 nF for annealed sample. We speculate that the decrease of recombination pathway in annealed device results in this increase in forward bias capacitance. It is because that insufcient recombination pathway is benecial to the charge accumulation of the solar cell. On the other hand, it is apparent that the capacitance increases with illumination, especially for annealed sample as shown in Fig. 3. The increase of illuminated capacitance of annealed sample also conrms that the thermal treatment of nanocrystalpolymer lms is benecial to accumulate charge, and hence assist light harvesting. This result was supported by the performances of the photovoltaic cells. Under simulated AM1.5 global solar conditions at an intensity of 100 mW cm2 the Isc varies from 1.25 mA cm2 (for the device without heat treatment) to 1.46 mA cm2 (for the device annealing at 150 C for 20 min), and FF show an increase from 37% to 45% before and after annealing. 4. Conclusion The effect of device annealing was studied for bulk heterojunction solar cells made from blends of MEH-PPV and CdS. Both frequency-dependent IS and CV characteristic measurements were analyzed before and after thermal annealing. For the device without heat treatment, the NC effect was observed, because the channels for electron transport were inadequate, leading to low efciency in solar cells. Otherwise, for the device annealing at 150 C for 20 min, the adequate electron transport channels were formed due to the removal of interfacial pyridine of nanocrystalpolymer systems and to bringing nanorods closer together in the polymer matrix, and hence the NC effect was eliminated, producing higher energy conversion efciency in the cells. We conclude that IS technique can be applied to inspect the mechanism for the annealing improvements of hybrid bulk heterojunction solar cells. Acknowledgements The authors acknowledge the nancial support from the National Basic Research Program of China (Grant No. 2009CB623602 and 2009CB623604), the National Natural Science Foundation of China (Grant No. 50573024 and 50433030), the Key Project of Chinese Ministry of Education (Grant No.104208), and the Doctorate Foundation of South China University of Technology. References
[1] N.S. Sariciftci, L. Smilowitz, A.J. Heeger, F. Wudl, Science 258 (1992) 1474. [2] J.J.M. Halls, C.A. Walsh, N.C. Greenham, E.A. Maeseglia, R.H. Friend, S.C. Moratti, A.B. Holmes, Nature 376 (1995) 498. [3] G. Yu, J. Gao, J.C. Hummelen, F. Wudl, A.J. Heeger, Science 270 (1995) 1789. [4] S. Tiwari, S. Tiwari, Cryst. Res. Technol. 41 (2006) 78. [5] D.S. Ginger, N.C. Greenham, Synth. Met. 124 (2001) 117. [6] T. Ahn, H. Lee, S.H. Han, Appl. Phys. Lett. 80 (2002) 392. [7] W.U. Huynh, J.J. Dittmer, W.C. Libby, G.L. Whiting, A.P. Alivisatos, Adv. Funct. Mater. 13 (2003) 73. [8] Y. Kim, S.A. Choulis, J. Nelson, D.D.C. Bradley, Appl. Phys. Lett. 86 (2005) 063502. [9] M. Reyes-Reyes, K. Kim, D.L. Carroll, Appl. Phys. Lett. 87 (2005) 083506. [10] G. Li, V. Shrotriya, Y. Yao, Y. Yang, J. Appl. Phys. 98 (2005) 043704. [11] L. Wang, Y.S. Liu, X. Jiang, D.H. Qin, Y. Cao, J. Phys. Chem. 111 (2007) 9538.

Fig. 3. The dark and illuminated capacitancevoltage curve for both before and after thermal annealing CdS-doped MEH-PPV samples.

W. Huang et al. / Synthetic Metals 160 (2010) 445449 [12] W.B. Huang, J.B. Peng, L. Wang, J. Wang, Y. Cao, Appl. Phys. Lett. 92 (2008) 013308. [13] Y.Q. Mo, J. Huang, J.X. Jiang, X.Y. Deng, Y.H. Niu, Y. Cao, Chin. J. Polym. Sci. 20 (2002) 461. [14] Y.S. Liu, L. Wang, D.H. Qin, Y. Cao, Chin. Phys. Lett. 23 (2006) 3345.

449

[15] I. Mora-Sero, J. Bisquert, F. Fabregat-Santiago, G. Garcia-Belmonte, Nano lett. 6 (2006) 640. [16] J.R. Macdonald, Impedance Spectroscopy, Wiley, New York, 1987. [17] V. Shrotriya, Y. Yang, J. Appl. Phys. 97 (2005) 054504.

Vous aimerez peut-être aussi