- Email: [email protected]
Mechanism of carrier generation and recombination in conjugated polymers
Synthetic Metals 116 (2001) 19±22
Mechanism of carrier generation and recombination in conjugated polymers Daniel Moses*, Arthur Dogariu, Alan J. Heeger Institute for Polymers and Organic Solids, University of California, Santa Barbara, CA 93106, USA
Abstract Transient excited-state absorption measurements in the spectral region spanning the infrared active vibrational active (IRAV) modes in prototypical luminescent polymers, poly(phenylene vinylene) (PPV), and poly[2-methoxy-5-(2-ethyl-hexyloxy)-(phenylene vinylene)] (MEH-PPV), reveal charge carrier generation within 100 fs after photoexcitation. The photocarrier quantum ef®ciency in MEH-PPV is f0 0:1 in zero applied electric ®eld. There is no correlation between the temporal behavior of the photoinduced IRAV signals and the exciton lifetime. Thus, carriers are photoexcited directly and not generated via a secondary process from exciton annihilation. Our data indicate that the carrier recombination rate is sensitive to the strength of the interchain interaction. # 2001 Elsevier Science B.V. All rights reserved. Keywords: Conjugated polymers; Carrier generation; Carrier recombination
While considerable progress has been made toward understanding the properties of luminescent conjugated polymers, there is a long-standing controversy on the mechanism of photogeneration of charge carriers: are charged photocarriers (polarons) primary excitations, or do they result from secondary processes that involve exciton annihilation [1]? Here we address the carrier photogeneration mechanism and determine the initial quantum ef®ciency (f0) by means of transient excited-state absorption (photoinduced absorption (PIA)) measurements, pumped in the visible and probed with 100 fs temporal resolution in the 6± 10 mm spectral region which spans the infrared active vibrational (IRAV) modes. The method is based on the one-to-one correspondence between the steady-state [2±4] and transient [5] photoinduced IRAV absorption in pristine conjugated polymers and the linear IRAV absorption of chemically doped polymers; the strength of the IRAV modes is proportional to the density of carriers on the polymer chain [2]. Thus, monitoring the strength of the IRAV absorption provides a direct ultrafast probe to the charge carrier density at times typical of carrier thermalization in disordered semiconductors [6]. Photoinduced IRAV absorption in MEH-PPV/C60 con®rms charge transfer in less than 100 fs. Ultrafast electron transfer from the photogenerated excited state of PPV (and
* Corresponding author. Fax: 1-805-893-4755. E-mail address: [email protected] (D. Moses).
its soluble derivatives) to C60 occurs because the lowest energy unoccupied state in C60 lies within the energy gap, and because energy can be conserved in the charge transfer process by promoting the hole left behind to a higher energy state within the relatively broad p-band of the semiconducting polymer [7]. Because of the ultrafast charge transfer, the quantum ef®ciency (QE) for charge separation and charge carrier generation in MEH-PPV/C60 approaches unity, consistent with the quenching of the photoluminescence and the enhancement of the photoconductivity in blends containing C60 [7±10]. Thus, the photocarrier QE in the pristine polymer can be determined from the ratio of the photoinduced IRAV signals from MEH-PPV and MEH-PPV/C60. In zero external ®eld, f0 0:1 in MEH-PPV, comparable to the QE for photoluminescence (PL). Finally, by comparing the photoinduced IRAV signals in PPV (where there are no side chains) and MEH-PPV (where the side-chains introduced for improved solubility reduce the strength of the interchain hopping interaction), we demonstrate the sensitivity of the rate of carrier recombination to the strength of interchain interactions. The ``traditional'' theoretical approach to photogeneration of carriers in low mobility materials involves the initial creation of bound geminate electron±hole (e±h) pairs (Frenkel excitons) [11±13]. The exciton model focuses on the role of temperature (T) and external electric ®eld (E) in the dissociation of geminate e±h pairs into ``free'' carriers and, therefore, predicts a strong dependence of j0 on T and on E [11±13]. The T-independence of the transient
0379-6779/01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 9 - 6 7 7 9 ( 0 0 ) 0 0 5 0 6 - 3
20
D. Moses et al. / Synthetic Metals 116 (2001) 19±22
photoconductivity in semiconducting polymers [14] (and other low mobility systems [15±17]), con®rmed through steady-state photoconductivity in thin ®lms [18], is not consistent with the exciton model. Field-induced quenching of the PL [19] and excited-state absorption in high external ®elds (typically, E > 106 V cmÿ1) [20] have been interpreted in terms of exciton dissociation. However, an increase in the carrier yield at high ®elds does not preclude the possibility of direct photoexcitation of carriers [21]. Moreover, assuming that both excitons and charged carriers are directly produced, there is no information available on the branching ratio. The ®rst utilization of photoinduced IRAV modes for measuring photocarrier density was reported by Mizrahi et al. for MEH-PPV/C60 [5]. With 60 ps temporal resolution, they demonstrated PIA with the spectral signatures of the IRAV modes. We have extended this initial work by probing the photocarrier density via PIA measurements in the IRAV spectral region with signi®cantly higher temporal resolution (100 fs), by measuring the photoinduced IRAV signals in pristine samples of PPV and MEH-PPV, and by determining j0 from the ratio of the signals obtained in MEH-PPV and MEH-PPV/C60. Measurements were taken on free-standing (15 mm thick) stretched-aligned PPV (draw ratio of l=l0 4), on ®lms of MEH-PPV (40 mm thick) and on two MEH-PPV/C60 blends (10 and 50% C60 by weight with thicknesses of 2 and 12 mm, respectively). The MEH-PPV and MEH-PPV/C60 ®lms were prepared by drop-casting on to KBr substrates. In order to produce more homogeneous MEH-PPV/C60 blends, we used the soluble derivative of C60, 1-(3-methoxycarbonyl)propyl1-phenyl[6,6]C61. An ampli®ed Ti-sapphire laser system equipped with an optical parametric ampli®er (OPA) was used to produce 100 fs pulses at a repetition rate of 1 KHz. The OPA, pumped by the ampli®ed oscillator at opump generates idler and signal beams with photon frequencies oi and os, where opump oi os . These two beams were mixed in a nonlinear crystal (AgGaSe2) to generate a beam at oprobe os ÿ oi [22], tuned to the desired probe beam frequency, oIRAV, between 6 and 10 mm. The probe pulse bandwidth was approximately 0.02 eV, consistent with that expected from the Fourier transform of the short pulses. For the pump beam, we used either the fundamental laser beam at 800 nm, or its second harmonic at 400 nm. The probe pulse was tuned to the IRAV frequencies as determined by Mizrahi et al. for MEH-PPV/C60 [5]. Although the intrinsic bandwidth of the 100 fs pulses (Dl 1 mm, for l between 6 and 10 mm) prevented accurate determination of the IRAV spectrum (resolution of the sharp features in the IRAV spectrum would require pulse widths >1 ps), the PIA signals resulted from photoinduced enhancement of the IRAV bands (see [4]). Moreover, in our experiments, there was no detectable signal when probed in the IR but well away from the IRAV modes
Fig. 1. The measured PIA waveform in PPV when pumped at 800 nm via a two photon absorption and probed at 7 mm; the inset shows the quadratic PIA dependence on light intensity when pumped at 800 nm.
(e.g. at wavelengths shorter than 5 mm); PIA was detected only when oprobe was tuned to the IRAV modes (around 7 and 9 mm) [5]. When pumped at 800 nm, the dependence of the PIA signal strength on light intensity (I) is quadratic, as shown in the inset of Fig. 1. Since the excitation is via two-photon absorption, the I2 dependence of the PIA indicates a linear dependence of the PIA signal on the carrier density. Note that the small two-photon absorption coef®cient results in a small density of excitations, but the relatively long samples used facilitated an adequate cross-section for the IRAV absorption. Fig. 1 shows the PIA waveform as obtained from PPV when excited at 800 nm and probed at 7 mm. The PIA response is limited by the system temporal resolution, 100 fs (see inset in Fig. 3). A fast initial PIA decay due to mono and bimolecular recombination (identi®ed from the variation of the PIA waveform at various I) is followed by a longer-lived exponential decay with t 250 ps. Fig. 2 compares the PIA obtained from MEH-PPV to that obtained from MEH-PPV/C60 (50% C60, by weight) when pumped with identical intensity at 800 nm and probed at 7 mm. Similarly, Fig. 3 compares the PIA responses in MEHPPV and MEH-PPV/C60 (10% C60, by weight), when pumped at 400 nm, and probed at 9 mm. The short rise time sets an upper bound of 100 fs for the onset of the carrier generation process in PPV and MEHPPV and for electron transfer from MEH-PPV to C60 [7±9]. Since the strengths of the IRAV mode signals are proportional to the density of carriers, these short rise times imply ultrafast carrier generation and polaron formation at times smaller by more than three orders of magnitude than the exciton lifetime (t 300 ps). Thus, the branching of the photoexcitations into the carrier and exciton channels occurs
D. Moses et al. / Synthetic Metals 116 (2001) 19±22
Fig. 2. Comparison of the PIA waveform in MEH-PPV (*); and MEHPPV/C60 (50% C60, by weight) (~), when pumped at the same light intensity at 800 nm and probed at 7 mm; the inset shows the data on an expanded scale near t 0.
at t < 100 fs (most likely before the completion of the thermalization process)1 [23]. There is no correlation between the temporal behavior of the photoinduced IRAV signals from MEH-PPV (see Figs. 2 and 3) and the exciton lifetime (as determined by the decay of the photoluminescence). Thus, carriers are photoexcited directly and not generated through a secondary process from exciton annihilation (e.g. from interaction with impurities, defects, etc.). Moreover, carrier generation via exciton± exciton interaction is not consistent with the linear dependence of the carrier density on light intensity; moreover, this process is not likely when pumped at 800 nm, when the density of excitation is relatively small. The strength of the prompt PIA signal, PIA(0), is proportional to the charge carrier density at t 0; PIA 0 ÿDT=T sN 0d for small a (relevant to pumping at 800 nm), or PIA 0 ÿDT=T sN 0=a for large a (relevant to pumping at 400 nm), where N(0) is the carrier density at t 0, s the cross-section of the IRAV absorption bands, a the absorption coef®cient of the pump beam, and d the sample thickness. Thus PIA 0MEH-PPV =PIA 0MEH-PPV =C60 / N 0MEH-PPV = N 0MEH-PPV =C60
(1)
From the data in Fig. 2, N(0)MEH-PPV/N(0)MEH-PPV/ 50%C60 0:3, and from the data in Fig. 3, N(0)MEH-PPV/ N(0)MEH-PPV/10%C60 0:32. Thus, in pristine MEH-PPV, the initial charge carrier density, photogenerated in t < 100 fs in zero field, is smaller by only a factor of three than in MEH-PPV/C60 where the quantum efficiency approaches unity [7±9]. 1 Prompt carrier photogeneration has been proposed earlier in [1,14± 18,21,23].
21
Fig. 3. Comparison of the PIA waveform in MEH-PPV (*), magnified by a factor of 3.2; and MEH-PPV/C60 (10% C60 by weight) (~), when pumped at the same light intensity at 400 nm and probed at 9 mm; the inset shows the fast risetime of the PIA in PPV (the solid curve represents a step function convoluted with a Gaussian with 100 fs full width at half maximum).
To determine f0 from the measured ratio R N(0)MEHone must correct for the following: (1) the volume fraction g, of MEH-PPV which was replaced by C60 in the MEH-PPV/C60 composite reduces the number of photons absorbed by the polymer chain and thus reduces the carrier density N(0); (2) the probability f of electron transfer from the polymer to C60 changes N(0) in the MEH-PPV/C60 composite. Including these corrections, we obtain 2f0 ; thus; R f 1 ÿ gf0 f0 1 ÿ f f 1 ÿ f0 g R 1 ÿ gf =2 : (2) f0 1 ÿ R 1 ÿ g 1 ÿ f PPV/N(0)MEH-PPV/C60
Eq. (2) accounts for the contribution to the PIA signal in the MEH-PPV/C60 composite of fraction f0 of holes directly excited along the polymer chain, the remaining fraction f0 1 ÿ f electrons directly excited on the polymer chain which, in blends with relatively small concentration of C60, did not undergo charge transfer reaction, and the fraction f 1 ÿ f0 of holes left on the MEH-PPV chain following the charge transfer reaction involving excitons. In MEHPPV/50%C60, f 1, and g 0:26. Thus, from the data in Fig. 2, f0 0:1. For the 10% blend, f 0:5 as inferred from the concentration dependence of the photoconductivity [8], and g 0 since nearly all of the 400 nm pump light is absorbed in the polymer. Thus, from the data in Fig. 3, f0 0:1. The two values obtained from samples with different concentrations of C60 and using two widely different pump frequencies (one involving a direct p±p transition and the other involving two-photon absorption) are in agreement. The relatively large value of f0 implies signi®cant delocalization of the excited state wavefunctions. Covalent
22
D. Moses et al. / Synthetic Metals 116 (2001) 19±22
bonding and the associated short intra-chain bond lengths lead to relatively large p-electron overlap integrals and broad p and p bands. As a result, even in disordered ®lms cast from solution, there is a relatively high probability that the electron±hole separation in a geminate pair will be suf®cient to prevent geminate recombination. Comparison of the temporal evolution of the photoinduced IRAV signals in PPV (Fig. 1) and MEH-PPV (Fig. 2) reveals that the initial decay depends on the strength of the interchain interaction (there are no functional groups to separate the main chains in PPV). The t 250 ps decay time obtained for oriented PPV is similar to that deduced from transient photoconductivity measurements [21]. In disordered ®lms of MEH-PPV where the side chains reduce the interchain interaction, f0 is relatively large, but the photocarrier density decays within 10±15 ps (see Figs. 2 and 3). Thus, interchain hopping/tunneling enhances the spatial extent of the wavefunctions and reduces the probability of early time recombination. Note that ``early time'' recombination and ``geminate'' recombination are not the same; e±h pairs which undergo geminate recombination do not contribute to the current (Figs. 2 and 3 indicate a relatively large f0, consistent with the fast transient photoconductivity of PPV and MEH-PPV) [8,21]. These observations are consistent with the photoconductivity data. The longer carrier lifetimes in PPV and MEHPPV/C60 enhance the photoconductive response [8,18,21]; while the fast carrier recombination in MEH-PPV reduces the magnitude of the transient photoconductivity measured at times (100 ps), when most of the photocarriers have already recombined [8,18]. In summary, ultrafast photoinduced IRAV absorption measurements enable, for the ®rst time in any semiconductor, direct determination of the initial QE for photogeneration of charge carriers. The data reveal photogeneration and thermalization of charged carriers and con®rm the ultrafast (<100 fs) electron transfer in MEH-PPV/C60 blends. The initial QE for charged photocarriers in MEH-PPV, f0 0:1, is comparable to the QE for photoluminescence. The absence of correlation between the fast photogeneration of charge carriers (<100 fs) and the exciton lifetime (tex 300 ps) indicates that carriers are directly photogenerated. Comparison of the recombination dynamics in MEHPPV and PPV demonstrates the sensitivity of the initial carrier recombination processes to the magnitude of interchain interactions.
Acknowledgements We are grateful to Prof. E.W. van Stryland and Dr. D. McBranch for providing us with AgGaSe2 crystal, and to Jian Wang for assistance with sample preparation. Dr. D. McBranch pointed out the importance of the different time scales observed for carrier decay and exciton decay. This research was supported by the Air Force Of®ce of Scienti®c Research (Charles Lee, Program Of®cer) under F49620-991-0031. References [1] N.S. Sariciftci (Ed.), The Nature of Photoexcitations in Conjugated Polymers, World Scientific, Singapore, 1997. [2] B. Horovitz, Solid State Commun. 41 (1982) 729. [3] E. Ehrenfreund, Z.V. Vardeny, SPIE Proc. 3145 (1997) 324. [4] Z.G. Soos, G.W. Hayden, A. Girlando, A. Painelli, J. Chem. Phys. 100 (1994) 7114. [5] U. Mizrahi, I. Shtrichman, D. Gershoni, E. Ehrenfreund, Z.V. Vardeny, in: Paper presented at the Meeting of the International Conference on Science and Technology of Synthetic Metals, Montpellier, France, 12 July 1998. [6] Z.V. Vardeny, J. Tauc, J. Phys. Coll. 42 (C-7) (1981) 477. [7] N.S. Sariciftci, L. Smilowitz, A.J. Heeger, F. Wudl, Science 258 (1992) 1474. [8] C.H. Lee, G. Yu, D. Moses, K. Pakbaz, C. Zhang, N.S. Sariciftci, A.J. Heeger, F. Wudl, Phys. Rev. B 48 (1993) 15425. [9] B. Kraabel, J.C. Hummelen, D. Vacar, D. Moses, A.J. Heeger, J. Chem. Phys. 104 (1996) 4267. [10] D.W. McBranch, E.S. Maniloff, D. Vacar, A.J. Heeger, J. Nonlinear Opt. Phys. Mater. 7 (3) (1998) 313. [11] M. Pope, C.E. Swenberg, Electronic Processes in Organic Crystals, Oxford University Press, New York, 1982. [12] R.C. Enck, G. Pfister, in: J. Mort, D.M. Pai (Eds.), Photoconductivity and Related Phenomena, Elsevier Scientific Publications, New York, 1976. [13] H. Scher, S. Rackovsky, J. Chem. Phys. 81 (1984) 1994. [14] D. Moses, A.J. Heeger, in: T. Kobayashi (Ed.), Relaxation in Polymers, World Scientific, Singapore, 1993. [15] D. Moses, Phys. Rev. B 53 (1996) 4462±4470. [16] D. Moses, Solid State Commun. 69 (1989) 721. [17] C.H. Lee, G. Yu, B. Kraabel, D. Moses, V.I. Srdanov, Phys. Rev. B 49 (1994) 10572. [18] D. Moses, J. Wang, G. Yu, A.J. Heeger, Phys. Rev. Lett. 80 (1998) 2685. [19] R. Kersting, et al., Phys. Rev. Lett. 73 (1994) 1440. [20] W. Graupner, et al., Phys. Rev. Lett. 81 (15) (1998) 3259. [21] D. Moses, et al., Phys. Rev. B 54 (1996) 4748. [22] M.K. Reed, M.K. Steiner-Shepard, in: X.P.F. Barbara, J.G. Fujimoto, W. Knox, W. Zinth (Eds.), Ultrafast Phenomena, Vol. X, Springer, Berlin, 1996, p. 40. [23] L.J. Rothberg, M. Yan, A.W.P. Fung, T.M. Jedju, E.W. Kwock, M.E. Galvin, Synth. Met. 84 (1997) 537.
Mechanism of carrier generation and recombination in conjugated polymers Daniel Moses*, Arthur Dogariu, Alan J. Heeger Institute for Polymers and Organic Solids, University of California, Santa Barbara, CA 93106, USA
Abstract Transient excited-state absorption measurements in the spectral region spanning the infrared active vibrational active (IRAV) modes in prototypical luminescent polymers, poly(phenylene vinylene) (PPV), and poly[2-methoxy-5-(2-ethyl-hexyloxy)-(phenylene vinylene)] (MEH-PPV), reveal charge carrier generation within 100 fs after photoexcitation. The photocarrier quantum ef®ciency in MEH-PPV is f0 0:1 in zero applied electric ®eld. There is no correlation between the temporal behavior of the photoinduced IRAV signals and the exciton lifetime. Thus, carriers are photoexcited directly and not generated via a secondary process from exciton annihilation. Our data indicate that the carrier recombination rate is sensitive to the strength of the interchain interaction. # 2001 Elsevier Science B.V. All rights reserved. Keywords: Conjugated polymers; Carrier generation; Carrier recombination
While considerable progress has been made toward understanding the properties of luminescent conjugated polymers, there is a long-standing controversy on the mechanism of photogeneration of charge carriers: are charged photocarriers (polarons) primary excitations, or do they result from secondary processes that involve exciton annihilation [1]? Here we address the carrier photogeneration mechanism and determine the initial quantum ef®ciency (f0) by means of transient excited-state absorption (photoinduced absorption (PIA)) measurements, pumped in the visible and probed with 100 fs temporal resolution in the 6± 10 mm spectral region which spans the infrared active vibrational (IRAV) modes. The method is based on the one-to-one correspondence between the steady-state [2±4] and transient [5] photoinduced IRAV absorption in pristine conjugated polymers and the linear IRAV absorption of chemically doped polymers; the strength of the IRAV modes is proportional to the density of carriers on the polymer chain [2]. Thus, monitoring the strength of the IRAV absorption provides a direct ultrafast probe to the charge carrier density at times typical of carrier thermalization in disordered semiconductors [6]. Photoinduced IRAV absorption in MEH-PPV/C60 con®rms charge transfer in less than 100 fs. Ultrafast electron transfer from the photogenerated excited state of PPV (and
* Corresponding author. Fax: 1-805-893-4755. E-mail address: [email protected] (D. Moses).
its soluble derivatives) to C60 occurs because the lowest energy unoccupied state in C60 lies within the energy gap, and because energy can be conserved in the charge transfer process by promoting the hole left behind to a higher energy state within the relatively broad p-band of the semiconducting polymer [7]. Because of the ultrafast charge transfer, the quantum ef®ciency (QE) for charge separation and charge carrier generation in MEH-PPV/C60 approaches unity, consistent with the quenching of the photoluminescence and the enhancement of the photoconductivity in blends containing C60 [7±10]. Thus, the photocarrier QE in the pristine polymer can be determined from the ratio of the photoinduced IRAV signals from MEH-PPV and MEH-PPV/C60. In zero external ®eld, f0 0:1 in MEH-PPV, comparable to the QE for photoluminescence (PL). Finally, by comparing the photoinduced IRAV signals in PPV (where there are no side chains) and MEH-PPV (where the side-chains introduced for improved solubility reduce the strength of the interchain hopping interaction), we demonstrate the sensitivity of the rate of carrier recombination to the strength of interchain interactions. The ``traditional'' theoretical approach to photogeneration of carriers in low mobility materials involves the initial creation of bound geminate electron±hole (e±h) pairs (Frenkel excitons) [11±13]. The exciton model focuses on the role of temperature (T) and external electric ®eld (E) in the dissociation of geminate e±h pairs into ``free'' carriers and, therefore, predicts a strong dependence of j0 on T and on E [11±13]. The T-independence of the transient
0379-6779/01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 9 - 6 7 7 9 ( 0 0 ) 0 0 5 0 6 - 3
20
D. Moses et al. / Synthetic Metals 116 (2001) 19±22
photoconductivity in semiconducting polymers [14] (and other low mobility systems [15±17]), con®rmed through steady-state photoconductivity in thin ®lms [18], is not consistent with the exciton model. Field-induced quenching of the PL [19] and excited-state absorption in high external ®elds (typically, E > 106 V cmÿ1) [20] have been interpreted in terms of exciton dissociation. However, an increase in the carrier yield at high ®elds does not preclude the possibility of direct photoexcitation of carriers [21]. Moreover, assuming that both excitons and charged carriers are directly produced, there is no information available on the branching ratio. The ®rst utilization of photoinduced IRAV modes for measuring photocarrier density was reported by Mizrahi et al. for MEH-PPV/C60 [5]. With 60 ps temporal resolution, they demonstrated PIA with the spectral signatures of the IRAV modes. We have extended this initial work by probing the photocarrier density via PIA measurements in the IRAV spectral region with signi®cantly higher temporal resolution (100 fs), by measuring the photoinduced IRAV signals in pristine samples of PPV and MEH-PPV, and by determining j0 from the ratio of the signals obtained in MEH-PPV and MEH-PPV/C60. Measurements were taken on free-standing (15 mm thick) stretched-aligned PPV (draw ratio of l=l0 4), on ®lms of MEH-PPV (40 mm thick) and on two MEH-PPV/C60 blends (10 and 50% C60 by weight with thicknesses of 2 and 12 mm, respectively). The MEH-PPV and MEH-PPV/C60 ®lms were prepared by drop-casting on to KBr substrates. In order to produce more homogeneous MEH-PPV/C60 blends, we used the soluble derivative of C60, 1-(3-methoxycarbonyl)propyl1-phenyl[6,6]C61. An ampli®ed Ti-sapphire laser system equipped with an optical parametric ampli®er (OPA) was used to produce 100 fs pulses at a repetition rate of 1 KHz. The OPA, pumped by the ampli®ed oscillator at opump generates idler and signal beams with photon frequencies oi and os, where opump oi os . These two beams were mixed in a nonlinear crystal (AgGaSe2) to generate a beam at oprobe os ÿ oi [22], tuned to the desired probe beam frequency, oIRAV, between 6 and 10 mm. The probe pulse bandwidth was approximately 0.02 eV, consistent with that expected from the Fourier transform of the short pulses. For the pump beam, we used either the fundamental laser beam at 800 nm, or its second harmonic at 400 nm. The probe pulse was tuned to the IRAV frequencies as determined by Mizrahi et al. for MEH-PPV/C60 [5]. Although the intrinsic bandwidth of the 100 fs pulses (Dl 1 mm, for l between 6 and 10 mm) prevented accurate determination of the IRAV spectrum (resolution of the sharp features in the IRAV spectrum would require pulse widths >1 ps), the PIA signals resulted from photoinduced enhancement of the IRAV bands (see [4]). Moreover, in our experiments, there was no detectable signal when probed in the IR but well away from the IRAV modes
Fig. 1. The measured PIA waveform in PPV when pumped at 800 nm via a two photon absorption and probed at 7 mm; the inset shows the quadratic PIA dependence on light intensity when pumped at 800 nm.
(e.g. at wavelengths shorter than 5 mm); PIA was detected only when oprobe was tuned to the IRAV modes (around 7 and 9 mm) [5]. When pumped at 800 nm, the dependence of the PIA signal strength on light intensity (I) is quadratic, as shown in the inset of Fig. 1. Since the excitation is via two-photon absorption, the I2 dependence of the PIA indicates a linear dependence of the PIA signal on the carrier density. Note that the small two-photon absorption coef®cient results in a small density of excitations, but the relatively long samples used facilitated an adequate cross-section for the IRAV absorption. Fig. 1 shows the PIA waveform as obtained from PPV when excited at 800 nm and probed at 7 mm. The PIA response is limited by the system temporal resolution, 100 fs (see inset in Fig. 3). A fast initial PIA decay due to mono and bimolecular recombination (identi®ed from the variation of the PIA waveform at various I) is followed by a longer-lived exponential decay with t 250 ps. Fig. 2 compares the PIA obtained from MEH-PPV to that obtained from MEH-PPV/C60 (50% C60, by weight) when pumped with identical intensity at 800 nm and probed at 7 mm. Similarly, Fig. 3 compares the PIA responses in MEHPPV and MEH-PPV/C60 (10% C60, by weight), when pumped at 400 nm, and probed at 9 mm. The short rise time sets an upper bound of 100 fs for the onset of the carrier generation process in PPV and MEHPPV and for electron transfer from MEH-PPV to C60 [7±9]. Since the strengths of the IRAV mode signals are proportional to the density of carriers, these short rise times imply ultrafast carrier generation and polaron formation at times smaller by more than three orders of magnitude than the exciton lifetime (t 300 ps). Thus, the branching of the photoexcitations into the carrier and exciton channels occurs
D. Moses et al. / Synthetic Metals 116 (2001) 19±22
Fig. 2. Comparison of the PIA waveform in MEH-PPV (*); and MEHPPV/C60 (50% C60, by weight) (~), when pumped at the same light intensity at 800 nm and probed at 7 mm; the inset shows the data on an expanded scale near t 0.
at t < 100 fs (most likely before the completion of the thermalization process)1 [23]. There is no correlation between the temporal behavior of the photoinduced IRAV signals from MEH-PPV (see Figs. 2 and 3) and the exciton lifetime (as determined by the decay of the photoluminescence). Thus, carriers are photoexcited directly and not generated through a secondary process from exciton annihilation (e.g. from interaction with impurities, defects, etc.). Moreover, carrier generation via exciton± exciton interaction is not consistent with the linear dependence of the carrier density on light intensity; moreover, this process is not likely when pumped at 800 nm, when the density of excitation is relatively small. The strength of the prompt PIA signal, PIA(0), is proportional to the charge carrier density at t 0; PIA 0 ÿDT=T sN 0d for small a (relevant to pumping at 800 nm), or PIA 0 ÿDT=T sN 0=a for large a (relevant to pumping at 400 nm), where N(0) is the carrier density at t 0, s the cross-section of the IRAV absorption bands, a the absorption coef®cient of the pump beam, and d the sample thickness. Thus PIA 0MEH-PPV =PIA 0MEH-PPV =C60 / N 0MEH-PPV = N 0MEH-PPV =C60
(1)
From the data in Fig. 2, N(0)MEH-PPV/N(0)MEH-PPV/ 50%C60 0:3, and from the data in Fig. 3, N(0)MEH-PPV/ N(0)MEH-PPV/10%C60 0:32. Thus, in pristine MEH-PPV, the initial charge carrier density, photogenerated in t < 100 fs in zero field, is smaller by only a factor of three than in MEH-PPV/C60 where the quantum efficiency approaches unity [7±9]. 1 Prompt carrier photogeneration has been proposed earlier in [1,14± 18,21,23].
21
Fig. 3. Comparison of the PIA waveform in MEH-PPV (*), magnified by a factor of 3.2; and MEH-PPV/C60 (10% C60 by weight) (~), when pumped at the same light intensity at 400 nm and probed at 9 mm; the inset shows the fast risetime of the PIA in PPV (the solid curve represents a step function convoluted with a Gaussian with 100 fs full width at half maximum).
To determine f0 from the measured ratio R N(0)MEHone must correct for the following: (1) the volume fraction g, of MEH-PPV which was replaced by C60 in the MEH-PPV/C60 composite reduces the number of photons absorbed by the polymer chain and thus reduces the carrier density N(0); (2) the probability f of electron transfer from the polymer to C60 changes N(0) in the MEH-PPV/C60 composite. Including these corrections, we obtain 2f0 ; thus; R f 1 ÿ gf0 f0 1 ÿ f f 1 ÿ f0 g R 1 ÿ gf =2 : (2) f0 1 ÿ R 1 ÿ g 1 ÿ f PPV/N(0)MEH-PPV/C60
Eq. (2) accounts for the contribution to the PIA signal in the MEH-PPV/C60 composite of fraction f0 of holes directly excited along the polymer chain, the remaining fraction f0 1 ÿ f electrons directly excited on the polymer chain which, in blends with relatively small concentration of C60, did not undergo charge transfer reaction, and the fraction f 1 ÿ f0 of holes left on the MEH-PPV chain following the charge transfer reaction involving excitons. In MEHPPV/50%C60, f 1, and g 0:26. Thus, from the data in Fig. 2, f0 0:1. For the 10% blend, f 0:5 as inferred from the concentration dependence of the photoconductivity [8], and g 0 since nearly all of the 400 nm pump light is absorbed in the polymer. Thus, from the data in Fig. 3, f0 0:1. The two values obtained from samples with different concentrations of C60 and using two widely different pump frequencies (one involving a direct p±p transition and the other involving two-photon absorption) are in agreement. The relatively large value of f0 implies signi®cant delocalization of the excited state wavefunctions. Covalent
22
D. Moses et al. / Synthetic Metals 116 (2001) 19±22
bonding and the associated short intra-chain bond lengths lead to relatively large p-electron overlap integrals and broad p and p bands. As a result, even in disordered ®lms cast from solution, there is a relatively high probability that the electron±hole separation in a geminate pair will be suf®cient to prevent geminate recombination. Comparison of the temporal evolution of the photoinduced IRAV signals in PPV (Fig. 1) and MEH-PPV (Fig. 2) reveals that the initial decay depends on the strength of the interchain interaction (there are no functional groups to separate the main chains in PPV). The t 250 ps decay time obtained for oriented PPV is similar to that deduced from transient photoconductivity measurements [21]. In disordered ®lms of MEH-PPV where the side chains reduce the interchain interaction, f0 is relatively large, but the photocarrier density decays within 10±15 ps (see Figs. 2 and 3). Thus, interchain hopping/tunneling enhances the spatial extent of the wavefunctions and reduces the probability of early time recombination. Note that ``early time'' recombination and ``geminate'' recombination are not the same; e±h pairs which undergo geminate recombination do not contribute to the current (Figs. 2 and 3 indicate a relatively large f0, consistent with the fast transient photoconductivity of PPV and MEH-PPV) [8,21]. These observations are consistent with the photoconductivity data. The longer carrier lifetimes in PPV and MEHPPV/C60 enhance the photoconductive response [8,18,21]; while the fast carrier recombination in MEH-PPV reduces the magnitude of the transient photoconductivity measured at times (100 ps), when most of the photocarriers have already recombined [8,18]. In summary, ultrafast photoinduced IRAV absorption measurements enable, for the ®rst time in any semiconductor, direct determination of the initial QE for photogeneration of charge carriers. The data reveal photogeneration and thermalization of charged carriers and con®rm the ultrafast (<100 fs) electron transfer in MEH-PPV/C60 blends. The initial QE for charged photocarriers in MEH-PPV, f0 0:1, is comparable to the QE for photoluminescence. The absence of correlation between the fast photogeneration of charge carriers (<100 fs) and the exciton lifetime (tex 300 ps) indicates that carriers are directly photogenerated. Comparison of the recombination dynamics in MEHPPV and PPV demonstrates the sensitivity of the initial carrier recombination processes to the magnitude of interchain interactions.
Acknowledgements We are grateful to Prof. E.W. van Stryland and Dr. D. McBranch for providing us with AgGaSe2 crystal, and to Jian Wang for assistance with sample preparation. Dr. D. McBranch pointed out the importance of the different time scales observed for carrier decay and exciton decay. This research was supported by the Air Force Of®ce of Scienti®c Research (Charles Lee, Program Of®cer) under F49620-991-0031. References [1] N.S. Sariciftci (Ed.), The Nature of Photoexcitations in Conjugated Polymers, World Scientific, Singapore, 1997. [2] B. Horovitz, Solid State Commun. 41 (1982) 729. [3] E. Ehrenfreund, Z.V. Vardeny, SPIE Proc. 3145 (1997) 324. [4] Z.G. Soos, G.W. Hayden, A. Girlando, A. Painelli, J. Chem. Phys. 100 (1994) 7114. [5] U. Mizrahi, I. Shtrichman, D. Gershoni, E. Ehrenfreund, Z.V. Vardeny, in: Paper presented at the Meeting of the International Conference on Science and Technology of Synthetic Metals, Montpellier, France, 12 July 1998. [6] Z.V. Vardeny, J. Tauc, J. Phys. Coll. 42 (C-7) (1981) 477. [7] N.S. Sariciftci, L. Smilowitz, A.J. Heeger, F. Wudl, Science 258 (1992) 1474. [8] C.H. Lee, G. Yu, D. Moses, K. Pakbaz, C. Zhang, N.S. Sariciftci, A.J. Heeger, F. Wudl, Phys. Rev. B 48 (1993) 15425. [9] B. Kraabel, J.C. Hummelen, D. Vacar, D. Moses, A.J. Heeger, J. Chem. Phys. 104 (1996) 4267. [10] D.W. McBranch, E.S. Maniloff, D. Vacar, A.J. Heeger, J. Nonlinear Opt. Phys. Mater. 7 (3) (1998) 313. [11] M. Pope, C.E. Swenberg, Electronic Processes in Organic Crystals, Oxford University Press, New York, 1982. [12] R.C. Enck, G. Pfister, in: J. Mort, D.M. Pai (Eds.), Photoconductivity and Related Phenomena, Elsevier Scientific Publications, New York, 1976. [13] H. Scher, S. Rackovsky, J. Chem. Phys. 81 (1984) 1994. [14] D. Moses, A.J. Heeger, in: T. Kobayashi (Ed.), Relaxation in Polymers, World Scientific, Singapore, 1993. [15] D. Moses, Phys. Rev. B 53 (1996) 4462±4470. [16] D. Moses, Solid State Commun. 69 (1989) 721. [17] C.H. Lee, G. Yu, B. Kraabel, D. Moses, V.I. Srdanov, Phys. Rev. B 49 (1994) 10572. [18] D. Moses, J. Wang, G. Yu, A.J. Heeger, Phys. Rev. Lett. 80 (1998) 2685. [19] R. Kersting, et al., Phys. Rev. Lett. 73 (1994) 1440. [20] W. Graupner, et al., Phys. Rev. Lett. 81 (15) (1998) 3259. [21] D. Moses, et al., Phys. Rev. B 54 (1996) 4748. [22] M.K. Reed, M.K. Steiner-Shepard, in: X.P.F. Barbara, J.G. Fujimoto, W. Knox, W. Zinth (Eds.), Ultrafast Phenomena, Vol. X, Springer, Berlin, 1996, p. 40. [23] L.J. Rothberg, M. Yan, A.W.P. Fung, T.M. Jedju, E.W. Kwock, M.E. Galvin, Synth. Met. 84 (1997) 537.