fullerene bilayer and blend photovoltaic devices

Organic Electronics 10 (2009) 501–505

Contents lists available at ScienceDirect

Organic Electronics journal homepage: www.elsevier.com/locate/orgel

Letter

Photogenerated charge carrier transport and recombination in polyfluorene/fullerene bilayer and blend photovoltaic devices Bekele Homa, Mattias Andersson, Olle Inganäs * Biomolecular and Organic Electronics, IFM, Center of Organic Electronics, Linköping University, S-5813 Linkoping, Sweden

a r t i c l e

i n f o

Article history: Received 30 May 2008 Received in revised form 14 November 2008 Accepted 22 November 2008 Available online 25 December 2008

PACS: 72.20.Jv 72.80.Le 72.80.Rj

a b s t r a c t Using extraction of photogenerated charge carriers by linearly increasing voltage (photoCELIV), we investigated two key transport parameters in photovoltaic materials based on the donor APFO-3 and acceptor PCBM: the mobility and lifetime of photogenerated charge carriers, in bilayers of varying geometry and in blends with various acceptors loading. We find that mobility depends strongly on delay time for shorter delay time in all devices. The observed recombination kinetics is found to be monomolecular. The mean lifetime of charge carriers is 2–3 ls in blends and is slightly greater than 4 ls in bilayer devices. In addition, the implications of mobility and lifetime values on the collection efficiency of the devices are presented. Ó 2008 Elsevier B.V. All rights reserved.

Keywords: Polymer photovoltaics Electrical transport Photogeneration

1. Introduction Organic solar cells have become a center of attention because they are considered to be a potential low cost renewable energy sources [1–3]. There are several challenges to be overcome to realize this potential in practical devices. Among the challenges, increasing power conversion efficiency is paramount. In general the overall efficiency of organic solar cells is influenced by four main processes; absorption (creation of bound electron-hole pairs (excitons)), charge generation (dissociation of exciton to free carriers), recombination and/or collection of carriers to their respective electrodes [1]. Recently conversion efficiency above 6% was reported [4]. This remarkable improvement of polymer solar cells was reached by using conjugated polymer–fullerene heterojunctions (HJ) and improved device architecture in a * Corresponding author. E-mail address: [email protected] (O. Inganäs). 1566-1199/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.orgel.2008.11.010

tandem cell. A doubling of the efficiency of organic solar cells was also achieved by folding two planar cells, similar or spectrally different, towards each other [5]. In HJ solar cells the photogenerated excitons dissociate at the donor/ acceptor interface via an ultrafast electron transfer from the donor (polymer) to the acceptor [6]. In order to dissociate, excitons must be created within the exciton diffusion length from the interface. There are two categories of HJ solar cells, bilayer HJ and bulk HJs. Bulk HJ solar cells have interpenetrating donor/acceptor interface, which provides large interface area for exciton dissociation and hence increase free carrier generation compared to bilayer HJ. On the other hand, numerical simulations show that collection efficiency of charge carriers is higher in bilayers [7], while still the overall efficiency is better in bulk heterojunctions. In studies using transient optical spectroscopy, the different neutral and charged species formed by photoexcitation of bulk heterojunctions of APFO3 and PCBM have been followed during the ps–ls time interval after excitation. A full kinetic model has been proposed, where geminate

502

B. Homa et al. / Organic Electronics 10 (2009) 501–505

recombination is an important loss mechanism at long times [8]. It is desirable to follow the fate of these charge carriers as we enter the time slot where electrical transport can be directly followed. This is our ambition in the present study. Collection efficiency and loss through recombination of free carriers are controlled by the mobility of free carriers. Therefore, better knowledge of mobility and carrier lifetime will certainly help further improvement of efficiency of polymer solar cells. In this work we report the mobility and lifetime of photogenerated carriers in donor/acceptor blends with various compositions, and in donor/acceptor bilayer photovoltaic devices and their significance for conversion efficiency. We used the photo-CELIV technique [9], which enables determination of mobility and recombination rate of free carriers simultaneously. The donor polymer used was APFO-3 (poly [2,7-(9,9-dioctyl-fluorene)-alt-5, 5(40 ,70 -di-2 thienyl-20 ,10 ,3-benzo-thiadiazole)]). The acceptors are PCBM ([6,6]-phenyl-C61-butyric acid methyl ester) in blends and C60 in bilayer.

storage oscilloscope (Tektronix TDS 340A) for different delay times using a 50 X load resistance. From the transients, the mobility and carrier concentration with varied delay time were calculated. The mobility of carriers is related to the time at which the extraction current reaches its maximum by Eq. (1) [9]: 2

l¼

Devices studied in this work are bilayers and blends with a configuration of ITO/APFO-3/C60/Al and ITO/ (APFO-3:PCBM) blend/Al, respectively. Solutions of APFO3:PCBM blend in chloroform with a PCBM content of 80%, 60%, 40% and 0% (pure APFO-3) with concentrations 75 mg/ml, 40 mg/ml, 25 mg/ml and 15 mg/ml were spin coated on pre cleaned glass substrate coated with ITO. Then aluminum was thermally evaporated on the film in a vacuum chamber at a pressure of 2.6  106 mbar, to complete the device fabrication. Bilayer devices were made the same way as that of pure polymer devices, except C60 was thermally evaporated on top of the polymer in a vacuum chamber at 5.2  106 mbar prior to aluminum deposition. The thicknesses of the active layers vary from sub 100 nm to 200 nm. The areas of the active layers were 4– 5 mm2. The technique used in this study was photo-CELIV, which employs excitation of charge carriers by absorbed laser light followed by extraction of carriers by linearly increasing voltage [9]. Generated carriers either undergo recombination or are collected by the built in electric field at the junction between electrode and active material. If the built in electric field is fully compensated, the only means through which the generated carrier concentration decays is recombination. These charges, which are created and forced to stay in the layer by an offset field, can then be extracted by linearly increasing voltage. The devices were illuminated from the ITO side by a 3–5 ns pulse of a Nd: YAG laser at 530 nm excitation wavelength. After a given delay time the reverse bias, triangular shaped voltage pulse, with a slope of A = 2  105 V/s was applied by a digital function generator (Stanford Research DS 345) by connecting aluminum to the positive terminal and ITO to negative terminal. An offset voltage of 0.4– 0.7 V was used. The delay time between laser light and onset of extraction was varied by a TTi TGP110 10 MHz Pulse generator. The transients were recorded by a digital

2d

 Dj 3At 2max 1 þ 0:36 jð0Þ

ð1Þ

where A is the increasing speed of extraction voltage, d is the thickness of the active layer, j(0) the displacement current density between the electrodes, Dj is the maximum conductive current density (Dj = jmax  j(0)), and tmax the time at which the transient reaches maximum. The displacement current density is jð0Þ ¼ eed0 A, where ee0 is permittivity of the material. The photo generated charge carrier concentration for the different delay times were calculated by integrating the conduction current over the extraction time given by Eq. (2).

n¼ 2. Experimental



1 ed

Z

T

ðjðtÞ  jð0ÞÞdt

ð2Þ

0

3. Results Fig. 1 shows transient curves obtained from photoCELIV measurements; (a) bilayer and (b) blend of 1:4 weight ratios APFO-3:PCBM. The extraction current decreases as delay time is increased for all devices. The time at which the maximum of extraction is reached also increased with increasing delay time. The mobility calculated using Eq. (1) for various delay times are displayed in Fig. 2a for bilayer and b for blends of different compositions of APFO-3 and PCBM. Calculated concentration of carriers versus delay time is plotted and shown in Fig. 3a and b. The curves are fitted to exponential decay of the form given by Eq. (3): t

nðtÞ ¼ nð0Þe s þ n1

ð3Þ

where n1 is the concentration of carriers that remain in the layer in the given extraction time, n(0) is the concentration of carrier with delay time close to zero. s is the mean lifetime of photogenerated carriers. Table 1 shows the average mean lifetime of several devices and, by a rough estimation, the minimum thickness a carrier can traverse with the slowest mobility driven by the weakest built in field, where the field is estimated by the applied voltage. Accurate measurements of both mobility and concentration are affected by offset voltage and thickness measurement. The offset voltage, applied to compensate the built in potential, was not exactly equal to the built in potential. The difference between the offset voltage and built in potential will extract carriers, which will result in significant photo current prior to the application of the triangular pulse. The thicknesses of devices were calculated from displacement current given by taking the approximate value 3 for the dielectric constant of the material. They were also measured mechanically by profilometer (DekTak 3030) and the values of both measurements agree well. However, inevitable error in measuring thickness due to

503

B. Homa et al. / Organic Electronics 10 (2009) 501–505

Mobility (cm2/Vs)

a

1.2x10 -4

Total Thickness 147 nm 150 nm 168 nm 200 nm 210 nm

1.0x10 -4

8.0x10 -5

6.0x10 -5

4.0x10 -5 0.0

3.0x10 -6 6.0x10 -6 9.0x10 -6 1.2x10 -5

Delay Time (s)

b

PCBM content of blends 80% pcbm 60% pcbm 40% pcbm 0% pcbm

Mobility (cm2/Vs)

1.2x10 -4 1.0x10 -4 8.0x10 -5 6.0x10 -5 4.0x10 -5 2.0x10 -5

Fig. 1. Photo-CELIV transients of (a) APFO-3/C60 bilayer device (b) APFO3:PCBM (1:4) ratio blend at different delay times. The extraction current decreases from shorter delay time to longer delay time.

roughness affects the absolute concentration and mobility values, but does not affect the variation of mobility and concentration as a function of delay time. 4. Discussion The charge transport mechanism in disordered organic systems depends upon the morphology, order, and molecular structure. In such systems the energy distribution of the HOMO (highest occupied molecular orbital) and the LUMO (lowest unoccupied molecular orbital) levels can be well approximated by a Gaussian distribution [10] as in Eq. (4) where r is the standard deviation energy levels.

1 E2 gðEÞ ¼ pffiffiffiffiffiffiffiffiffiffi exp  2 2r 2pr

! ð4Þ

Charge carriers transport in such system is via hopping transport between localized states, in which carriers have to hop from one localized state to another following the Miller-Abraham hopping rate given in Eq. (5),

( vij ¼ v0 expð2cRij Þ

  e e exp  jkT i ; 1

ej > ei ej < ei

ð5Þ

0.0

3.0x10 -6 6.0x10 -6 9.0x10 -6 1.2x10 -5

Delay Time (s) Fig. 2. (a) Photo-CELIV mobility versus delay time of (a) bilayer devices with different thickness of polymer films and (b) for blend devices. The thickness C60 in bilayer is 45 nm for all devices.

where vij is the hopping rate between sites i and j with energies ei and ej. The first exponential in Eq. (5) describes the electronic wave function overlap and the second describes a Boltzmann factor for sites upward in energy. Carriers need to absorb thermal energy, which amounts to energy difference of the two sites j and i for an upward jump. The rate for downward movement is only limited by the first exponential term, i.e. only the spatial disorder controls the downward jump. The decrease in mobility as indicated in Fig. 2a is attributed to the relaxation of carriers toward the tail of the density of states (DOS). Carriers that are excited to higher energy have higher mobility or higher probability to move, compared to those in the lower energy states in the distribution. The disordered nature of the material gives a high density of traps to which carriers continue to relax, until they eventually recombine with opposite carriers. The extraction current is negligible in the dark, implying the density of intrinsic charge carrier density is so small that it cannot be detected with this experiment. All the carriers extracted are photogenerated which means that both carrier types are extracted. In all devices

504

B. Homa et al. / Organic Electronics 10 (2009) 501–505

a Carrier Concentration (#/cm3)

5x1015 Total thickness 147 nm 150 nm 168 nm 200 nm 210 nm

4x1015

3x1015

2x1015

1x1015 0.0

3.0x10-6

6.0x10-6

9.0x10-6 1.2x10-5

Delay Time (s)

Carrier Concentration (#/cm3)

b

6x1015

PCBM content of blends 80% pcbm 60% pcbm 40% pcbm 0% pcbm

5x1015 4x1015 3x1015

which means that the actual window for observing both mobilities is rather limited. More often than not, it is thus the case that only one extraction peak is observed. It is not possible to ascertain if it is one or both mobilities, neither which one, that should be associated with the peak. By examining mobility values obtained through other means, and based on some general assumptions regarding different measurement techniques [11,12], we infer that it should be possible to make an APFO-3:PCBM blend sample of suitable thickness and suitable stoichiometry so as to observe two resolved mobility peaks in CELIV. Since this has proved difficult, another possibility is to use a bilayer to determine which peak it is that is observed. If a suitable excitation wavelength is selected so that almost all carrier generation in a photo-CELIV experiments occurs in the polymer layer, a systematic variation of the layer thicknesses could give the desired information. Unfortunately, the reported electron mobilities [13] in C60 are so high that manageable thicknesses have almost no effect on the peak position. Moreover, it can be argued that density of electrons and holes are equal. For equal number of both carriers, the decay of carrier density may follow the Eq. (6), which depends linearly and nonlinearly on concentration n [14]:

dn ¼ ðAn þ Bn2 Þ dt

15

2x10

1x1015 0.0

3.0x10-6

6.0x10-6

9.0x10-6 1.2x10-5

Delay Time (s) Fig. 3. Charge carrier concentration plotted versus delay time (a) bilayer devices (b) different composition of APFO-3:PCBM blends. The solid lines in both (a) and (b) are the curve fits using Eq. (2).

Table 1 Mean lifetime of charge carriers and extraction depth by built in field. Devices Pure Bilayer Blend (80% PCBM) Blend (60% PCBM) Blend (40% PCBM)

Mean life time (s) (s) 6

(2.3 ± 0.7)  10 (4.1 ± 0.6)  106 (2.7 ± 0.4)  106 (2.3 ± 0.5)  106 (2.2 ± 0.6)  106

l  s  E (nm) 61 91 110 94 90

investigated only one peak is observed, which indicates that the mobility of electron and hole are comparable or else very different. From mathematical considerations, there needs to be a factor  3 in difference between the electron and hole mobility, in order to observe two resolved peaks in geometries and under measurement conditions pertinent to samples resembling well functioning solar cells. In reality the peaks usually are less well defined and so the required difference is probably larger still. Furthermore, the mobility range possible to observe with a CELIV pulse is about three orders of magnitude in the relevant samples, due to RC- and signal amplitude effects,

ð6Þ

where A is the monomolecular recombination coefficient, describing recombination through defects (traps). The recombination rate is directly proportional to the density of carriers. B is the coefficient of describing the direct radiative or non-radiative recombination of an electron in LUMO to holes in HOMO, which is therefore a bimolecular process and hence proportional to the square of the carrier density. Clearly, both types of recombination can happen at the same time, but depending on the intrinsic properties of the material, one may be the dominant process. For instance, if a material has low trap density and high charge density bimolecular outweighs the monomolecular recombination. On the other hand, if a material has high trap density and low concentration of carriers, monomolecular recombination dominates. In line with the above argument, the results displayed in Fig. 3 show that the recombination type is predominantly monomolecular type, which is the result of the disordered nature of the material i.e., high trap density. The mean lifetime in bilayer structures is longer compared to pure and blend materials (see Table 1). This is because the photogenerated carriers generated at the interface, stay apart for a longer time in the different materials (polymer and C60) by internal electric field, which is oriented in one direction, whereas in blends the orientation of junctions is quite random and therefore the internal electric field does not induce cumulative barriers to carriers. The result in Table 1 shows the minimum distance that carriers can traverse during their lifetime driven by internal field. It can therefore be concluded that in sub 100 nm thick devices carrier collection is not a limiting factor.

B. Homa et al. / Organic Electronics 10 (2009) 501–505

5. Conclusion Photo-CELIV technique has been used to investigate mobility and lifetime of photo generated charge carriers in blend and bilayer structures of photovoltaic devices. It is found that the mobility decreases very fast for shorter delay time and slower for longer delay time in all blends and bilayer devices. This decrease in mobility is ascribed to the relaxation of carriers to lower energy states in the distribution of energy states. Monomolecular recombination is found to be dominant in all APFO-3 based devices with acceptors used here, which is attributed to intrinsic energetic and spatial disorder of the material. The mean lifetime of photogenerated carriers is longer in bilayer structures as a result of a built in barriers at bilayer interface. Finally, the overall efficiency of sub 100 nm devices is not significantly limited by the collection of charge carriers. Acknowledgments We would like to thank Fengling Zhang for her assistance in device fabrication. One of the authors (Bekele Homa) acknowledges the financial support from the International Program in the Physical Sciences (IPPS) of Uppsala University, Sweden. These investigations were financially

505

supported by the Center of Organic Electronics (COE) at Linköping University, Sweden, financed by the Strategic Research Foundation SSF. References [1] Privikas, N.S. Sarciftci, G. Juska, R. Osterbacka, Progress in photovoltaic, Res. Appl. 15 (2007) 677–696. [2] C.J. Brabec, N.S. Sariciftci, J.C. Hummelen, Adv. Funct. Mater. 11 (2001) 15. [3] J. Nelson, Science 293 (2001) 1059–1060. [4] J.Y. Kim, K. Lee, N.E. Coates, D. Moses, Th.Q. Nguyen, M. Dante, A.J. Heeger, Science 317 (2007) 222–225. [5] K. Tvingstedt, V. Andersson, F.L. Zhang, O. Inganäs, Appl. Phys. Lett. 91 (2007) 123514. [6] D. Moses, J. Wang, G. Yu, A.J. Heeger, Phys. Rev. Lett. 80 (12) (1998). [7] A. Marsh, C. Groves, N.C. Greenham, J. App. Phys. 101 (2007) 083509. [8] D. Swati, T. Pascher, M. Maiti, K.G. Jespersen, T. Kesti, F.L. Zhang, O. Inganäs, A. Yartsev, V. Sundström, J. Am. Chem. Soc. 129 (27) (2007) 8466–8472. [9] A.J. Mozer, N.S. Sarciftci, L. Lutsen, D. Vanderzande, R. Osterbacka, M. Westerling, G. Juska, Appl. Phys. Lett. 86 (2005) 112104. [10] H. Bassler, Phys. Status Solidi B 175 (1993) 15. [11] L.M. Andersson, F.L. Zhang, O. Inganäs, Appl. Phys. Lett. 89 (2006) 142111. [12] L. Mattias Andersson, Fengling Zhang, Olle Inganäs, Appl. Phys. Lett. 91 (2007) 071108. [13] T. Singh, H. Yang, B. Plochberger, L. Yang, H. Sitter, H. Neugebauer, N. Sariciftci, Phys. Status Solidi B 244 (2007) 3845–3848. [14] J. Szmytkowski, J. Phys. D: Appl. Phys. 40 (2007) 3352– 3357.