Femtosecond dynamics of excitons in π-conjugated oligomers: the role of intrachain two-exciton states in the formation of interchain species

3 October 1997

CHEMICAL PHYSICS LETTERS FJ SEVIER

Chemical Physics Letters 277 (1997) 109-117

Femtosecond dynamics of excitons in 7r-conjugated oligomers: the role of intrachain two-exciton states in the formation of interchain species Victor I. Klimov a, Duncan W. McBranch a Nikolay N. Barashkov b, John P. Ferraris b a Chemical Science and Technology Division, CST-6, MS J567, Los Alamos National Laboratory, Los Alamos, NM 87545, USA b University of Texas, Dallas, TX, USA

Received 1 May 1997

Abstract

We report femtosecond transient absorption results for solutions and thin films of a substituted oligomer of poly(paraphenylene vinylene) performed over wide spectral and pomp-intensity ranges. Solutions and films exhibit a photoinduced absorption (PA) band with dynamics matching those of the stimulated emission, demonstrating unambiguously that these features originate from intrachain singlet excitons. Thin films exhibit an additional short-wavelength PA band with pump-independent dynamics, indicating the formation of non-emissive interchain excitons. Correlations in the dynamics of the two PA features, as well as the intensity-dependence, provide strong evidence that the formation of interchain excitons is mediated by intrachain two-exciton states. © 1997 Elsevier Science B.V.

Despite numerous studies on the ultrafast photophysics of ~r-conjugated polymer systems such as poly(para-phenylene vinylene) (PPV) and its derivafives, great controversy persists regarding the nature and mechanisms of formation of the primary photoexcitations in these materials [1-13]. It is widely accepted that in dilute solutions with weakly interacting molecules, the dominant type of excitations are intrachain singlet excitons which have been described either as Coulombically bound electron-hole pairs [14,15] or self-localized "polaron excitons." [ 16,17] The situation is more complex in solid-state samples, where strong inter-molecular interactions can lead to the formation of non-emissive interchain species, also referred to as "indirect excitons," or

bound polaron pairs [5,6,8,9]. These species have been cited by some authors as primary excitations in solid-state samples [6-8], although this point of view is not universally accepted [3,10]. The mechanism of formation of indirect excitons remains unclear, and there is at present no agreement regarding the proportion of intra- and inter-chain excitons generated by absorbed photons [6-8,10]. As the formation and branching of initial photoexcitations occur on subpicosecond timescales, these processes are usually studied using methods of ultrafast spectroscopy [ 1 13]. In the present paper we apply the methods of femtosecond transient absorption (TA) to address two important issues regarding the nature of primary

0009-2614/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PH S0009-2614(97)00923-8

110

V.L Klimov et aL / Chemical Physics Letters 277 (1997) 109-117

photoexcitations in 7r-conjugated systems: (1) the mechanism for the generation of interchain species and (2) the role of intra- and inter-chain excitons in the formation of the nonlinear optical response. To evaluate the role of intermolecular interactions, we study two different forms of the same material: dilute solutions and solid-state films. We demonstrate that in solutions, both the stimulated emission (SE) and the photoinduced absorption (PAL) can be explained in terms of generation of a single species: intra-chain singlet excitons. In pristine films, in addition to the singlet exciton PA feature, we observe a new PA band which we attribute to non-emissive interchaln excitons. The correlations in dynamics of the two PA features, as well as the pump intensitydependence, indicate that the generation of interchain excitons is a quadratic nonlinear process mediated by intra-chain two-exciton states. Additionally, we see a strong effect of photodegradation on the SE dynamics in solid-state samples. As a model system for our studies we have se-

0.25 -I

I

I

lected a five-ring PPV oligomer 2-methoxy-5-(2'ethylhexyloxy)-distyryl benzene (MEH-DSB) (see a molecular structure in Fig. 1). MEH-DSB can be prepared as high quality thin films by vacuum sublimation, and is also soluble [18]. Investigation of a model oligomer system allows us to avoid the effects of inhomogeneous broadening resulting from a distribution of conjugation lengths which is always present in amorphous polymer films. These distributions can lead to associated ultrafast "random-walk" energy relaxation dynamics [2] which complicate the interpretation of experimental data on subpicosecond timescales. MEH-DSB was prepared and purified as described in [18]. Solutions were prepared at a concentration of 5 × 10 -4 g / m l in p-xylenes. Amorphous thin films ( ~ 100rim thickness) were deposited by vacuum sublimation onto sapphire substrates and transferred into an optical cryostat in a dry box to minimize the effects of photo-oxidation. As several authors have discussed [19-21,13], the similarities in the optical and electronic properties of

I

I

I

/ r

"I

i

O.2O

_

'|

Solution ]

( - ~ 0.15-

_/(3

MEH-DSB

)

sE:.

PAl

',

,oo/

; o.,o_ PA2

|

o o5-

.

:

,"

"-" ~

/

~

'

/i

_I ',21,"~|

/

t,'

1

/

0.00 -0.05 1.4

1.6

1.8

2.0

2.2

2.4

2.6

2.8

P h o t o n e n e r g y (eV) Fig. I. Compmisonof A T / T o spectra of MEH-DSB in solution (dashed line) and thin film (solid line) measured at Wp= 3mJcm-2. The time delay At is approximately 2 ps, so that the spectra are undistorted by the effects of chirp on the continuum probe. The dotted line shows the linear absorption in solution. Inset: molecularstructure of MEH-DSB.

V.L Klimov et al. / Chemical Physics Letters 277 (1997) 109-117

I

I

I

I

I

111

I

I

0.1

o

0.01 e 5 4 .q 2

0.001

• .A

O

(a)

I 0

I 1 O0

I 200

I 300

I 400

I 500

I 600

Delay time (ps)

0.1

[-I I

0

[]

1.o

[]

~--"

O.Ol

o.8

•

i

o.e

~:~

0.4 0.2 0.0

0.001

-0.2

0.0

0.2 0.4 0.6 o.e Delay t i m e (ps)

1.0

(b)

I

I

I

I

I

0

10

20

30

40

Delay time (ps) Fig. 2. Time transients of the photoinduced absorption IAadL measured at Wp- - 3mJcm -2 in (a) MEH-DSB solution and in (b) thin film sample. (a) In solution, the dynamics of the SE (open squares, 2.43 eV) and PA features (open circles, 1.81 eV and solid triangles, 1.46eV) are nearly identical. (b) In the film, the dynamics of the PA I (solid triangles, 1.46eV) and SE features (open squares, 2.32eV) still match, while those of the PA 2 feature (open circles, 1.81 eV) are clearly different. Inset: Complementary subpicosecond dynamics recorded for the PA I and PA 2 features in the film, plotted together with the pump-probe cross-corrclation, which defines the temporal resolution of the experiment.

112

V.L Klimov et aL / Chemical Physics Letters 277 (1997) 109-117

conjugated oligomers and polymers suggest that the results reported below are of general importance also for conjugated polymers. TA studies were performed using a femtosecond p u m p - p r o b e experiment. The samples were excited at 3.1 eV using 100fs frequency-doubled pulses from a regeneratively amplified mode-locked Ti-sapphire laser (Clark-MXR CPA-1000). The pump-photon energy slightly exceeds the energy of the ~'-rr * transition in MEH-DSB ( ~ 2.9eV), corresponding to near-resonant band-edge excitation. The transmission of the excited sample is probed by delayed pulses of a femtosecond continuum generated in a 1 mm thick sapphire plate. As a measure of transmission changes we use the differential transmission (DT), defined as follows: DT ffi ( T - T o ) / T o = A T / T o, where TO and T are transmissions in the absence and in the presence of the pump, respectively. In the small signal limit (DT << 1), DT is proportional to an absorption change A a : D T = - A a d ( d is the sample thickness). Time-resolved DT data were measured in two different experimental configurations. DT spectra at a fixed delay time A t between pump and probe pulses were recorded with a 0 . 1 5 m spectrometer coupled to a liquid-nitrogen-cooled CCD camera by averaging the signal over 1000-2000 pulses. This type of measurement provides information on the spectral distribution of the nonlinear optical response over a broad spectral range (0.45-1.1 I~m) with an accuracy up to 10 -3 in DT. The single-wavelength DT dynamics were monitored with much higher accuracy (up to 10 -5 in DT) using phase-sensitive detection with a lock-in amplifier synchronized to a chopped pump beam [22]. To avoid photodegradation, in all measurements reported in this paper the fluence Wp of individual pump pulses was kept below 5 m J c m -2. Furthermore, care was taken at each sample spot to keep the integrated pump fluence for

the entire scan below ~ 0 . 8 k J c m -2, a threshold above which irreversible photochemical changes could be detected in our measurements (see discussion below). Fig. 1 shows the comparison of DT spectra recorded at 2ps after excitation for the solution (dashed line) and the film (solid line) of MEH-DSB. The DT spectrum of the solution shows an intense positive band with a maximum at 2.6eV which occurs due to pump-induced bleaching of the ground state absorption (see the absorption curve shown by the dotted line in Fig. 1). A pronounced shoulder at ~ 2.4 eV, at spectral energies below the absorption edge can be assigned to SE, in agreement with TA data from [6] and the demonstration of efficient lasing in MEH-PPV solutions in [23]. The third pronounced TA feature in solutions is a broad PA band extending from 2 eV to below 1.3 eV. The DT dynamics recorded at different spectral energies (see Fig. 2(a)) show nearly identical exponential decay, with a time constant of around 650 ps for both SE and PA, indicating that they originate from the same species. As these species are associated with the efficient SE, they can be assigned to intrachain singlet excitons, in agreement with previous measurements in MEH-PPV solutions [8]. The TA relaxation dynamics give a measure of a radiative lifetime of excitons in MEH-DSB: ~'r = 650ps, which is close to the reported values for MEH-PPV [23]. The DT spectrum in amorphous films (solid line in Fig. 1) is distinctly different from that in solutions. We clearly see a reduction in the relative intensity of the SE in comparison with solutions, as was previously reported for polymeric samples [8]. Another important feature of TA in MEH-DSB films is the increase in the PA signal at ~ 1.8 eV, which is indicative of the formation of a new short-wavelength PA band. The fact that two distinct spectral bands are present in solid-state samples is confirmed

Fig. 3. (a) Time transients of the differential transmission IAT/To[, for two different pump fiuences, demonstrating matching dynamicsfor the SE (2.32eV) and PA I features (1.46eV) in a pristine film. The upper curves show data taken at wp ffi 3mJcm -2, for SE (solid squares) and PA 1 multiplied by - 0.3 (solid line). The lower curves show data taken at Wp= 0.6 mJ cm -2, for SE (open circles) and PAi multiplied by -0.3 (dashed line). Inset: SE and PA t dynamics at wp=3mJcm -2 in a photodegraded film (see text). (b) Time transients demonstrating pump-independent dynamics of the PA 2 feature, for Wp= 6mJcm -2 (open triangles), 3mJcm -2 (solid squares), and 0.6mJcm -2 (open circles). Solid and dashed lines are data taken at 3 and 0.6mlcm -2, respectively, scaled to the 6mJcm -2 data. Inset: PA pump dependence indicating a quadratic correlation (PA 2 ~ (PAt) 2) in the peak magnitudes of the PAI and PA 2 features.

113

V.L Klimov et al. / Chemical Physics Letters 277 (1997) 109-117

by an analysis of the DT time-transients at different probe wavelengths. In solid-state samples, different dynamics are observed for both the decay (Fig. 2(b)) and the build-up (inset to Fig. 2(b)) of the PA signal

I

at 1.46eV (PA l) and 1.81 eV (PA2), indicating contributions from at least two different species. The existence of two species is also clearly manifested in the pump-intensity dependence of the DT dynamics

I

I

I

I

Pristine film 0.01

otodo d ,, film

_

0.0111 .

_,.,

o.o

j :

-0.5

1

_.._______

,

0.0

,"t,~~,%,~

0.5

1.0 1.S 2.0 Delay time (ps)

2.5

(a)

t

i

i

i

,

0.0

0.5

1.0

1.5

2.0

2.5

I

I

3.0

Delay time (ps)

0.01

.

I

I

I

I

oOO°°°°°°°°°oOooOooooooOoooOOOoooooOOOo

8 6

°

~,o

|

|

,

,

,

| ,|

•

.~DA1

0.1

|

•

|

|

,

J

|

a I i

. . 41"""

2

|

!

0.001 8

2

6

J

0

--

2

O.Oal

_ ~ __~9

6

PA2 . i

I 0

3

4

5

6 "/89

2 1

0.0001

a

I

I

s

3

4

5

6 7119 10

2

CO)

Pump fluence (mJ era" ) I

I

I

I

I

I

0.0

0.5

1.0

1.5

2.0

2.5

Delay time (ps)

3.0

V.L Klimov et aL / Chemical Physics Letters 277 (1997) 109-117

114

Fig. 2(b) and Fig. 3(a)). This clearly shows that, as in solution, the SE and PA l features arise from the same species, namely from the emissive intrachaln singlet excitons. A comparison of the TA dynamics in films and solutions demonstrates that the decay of these excitons in films is predominantly nonradiative and occurs on a much faster timescale than in solutions. We attribute this enhancement in the relaxation rate to the opening of new relaxation channels due to strong intermolecular interactions in solid-sate samples (see discussion published elsewhere) [13]. The interpretation of the ultrafast PA features in solid-state samples of PPV derivatives has been highly controversial [1,3-13]. In films of PPV [6,7], methoxy-PPV [5] MEH-PPV [8], BuEH-PPV [12] and a ladder-type poly(para-pbenylene) derivative [9]

shown in Fig. 3(a) (PA l) and Fig. 3(b) (PA 2). On the 3ps timescale shown in Fig. 3(b), the PA 2 exhibits pump-independent and almost exponential decay with a time constant of about 6 ps (notice the overlap of the scaled time-transients of the lower intensity data (solid and dashed lines) and the higher intensity data (open triangles) in Fig. 3(b)). Over longer delay times ( ~ 50ps), the PA 2 decay is nonexponential. In stark contrast, the dynamics of PA l are clearly nonexponential even at short delay times, and show a pronounced pump intensity-dependence (see Fig. 3(a) and Fig. 4), which will be discussed below. Although the PAl decay dynamics are intensity-dependent, in pristine films they match very closely the SE dynamics on all timescales studied, and over a wide pump-intensity range [13] (see

I

I

I

I

|

I

V@

V 10 .2

•0 VE1 @

~

- %

_

'

•

I []

'

•

0.01

789

2

3

4

5

0.0

m

o.s

....................... 1 .............

6 789

0.1 I

.

I ..... .... ,-'...-..,-r-4 ........................ • ............. r-------,--'--.--",---,---r--,--~ 8

10 s

As

2

3

12 ,

P u m p Fluence (nd cm" ) ,

i

,

2.0

2.5

a.o

Delay Time (ps) Fig. 4. Pump-intensitydependenttime transientsof PAi featureat 1.46eV (symbols)plotted togetherwith double-exponentialfits (lines). Dam are shown for wp= 2.1 mlcrn-2 (open triangles), 1mJcm-2 (solid diamonds), 0.44mlorn-2 (solid triangles), 0.22mlcm -2 (open squares), and 0.06mJcm-2 (solid circles).Inset: quadraticcorrelationin the pump-dependenceof the magnitudesderivedfor the fast (A r) and the slow (,4,) componentsfrom the double-exponentialfits (,4f ~ (,4,)2).

V.L Klimov et al. / Chemical Physics Letters 277 (1997) 109-117

the SE and PA dynamics were observed to be different. Hence, for these materials it was concluded that the PA is entirely or partially due to species which are different from those responsible for the SE. However, correlations between the SE and PA decays have been reported for films of PPV [4], P3HT [24] and poly(p-pyridyl vinylene) [11]. We believe that the reason for many of the inconsistencies in previously published experimental data is the uncontrolled photo-oxidation of the samples, which has a very strong effect on the TA (and especially the SE) dynamics [ 10,13,25,26]. Our measurements show that irreversible photochemical changes take place in MEH-DSB, even in an evacuated cryostat, under moderate pulsed excitation levels. Photochemical degradation is characterized by an overall reduction in the magnitude of DT (by at least of a factor of 2), and a dramatic change in the dynamics of the SE emission, which no longer match the dynamics of the PA~ band, as shown in the inset to Fig. 3(a). These effects were observed at per pulse excitation levels above ~ 5 mJ cm-2, or when irradiating the films with a cumulative fluence greater than ~ 0.8 kJ cm-2 (e.g., during a single scan of the decay dynamics at a particular spot on the sample). In photodegraded (photo-oxidized) samples, we see a rapid switch from SE (positive DT) to PA (negative DT), similar to the crossover previously observed in PPV films [7], where the difference in PA and SE dynamics was cited as evidence of a distinct species contributing to the near-IR PA band. It has been established that the photo-oxidation results in the cleaving of the vinylene double bonds to form terminal carbonyl species, with associated loss of conjugation and quenching of the photoluminescence [27,28]. We now examine in more detail the short-wavelength PA 2 band, which is only observed in solidstate samples. The fact that we do not see any associated SE features indicates that the PA 2 band might be related to the non-emissive interchaln excitons predicted in [5-8]. This explanation is consistent with the observed pump-independent dynamics which are expected for strongly correlated geminate electron-hole pairs located on closely separated chains [6]. To explore the possible mechanisms for the generation of interchain species, we performed a careful study of the short-term PA ~ and PA 2 build-up dynamics (inset to Fig. 2(b)), and decay dynamics

115

(Fig. 3(b) and Fig. 4). We find first that the PA 1 and PA 2 features display complementary dynamics: the build-up of the PA 2 band (rise time of ~ 500fs) matches the fast initial decay of the PA 1 band (relaxation constant of ~ 450fs) (see the inset to Fig. 2(b)). We can derive further correlations related to the formation of the PA 2 band by fitting the decay of PA l at early times ( < 3ps), for different pump intensities, to a double exponential decay as shown in Fig. 4. Of course, since the PA decay is nonexponential over timescales from I ps-1 ns, such a double exponential fit cannot be used for a quantitative description of the decay over this whole range. However, at early times such a fit gives a good description of the dominant decay time constants and their relative magnitudes. It was recently demonstrated that the intensity-dependence of the PA 1 decay can be modelled semi-quantitatively as a bimolecular annihilation process [13]. At the shortest timescales, however, this model predicts that the initial effective time constant of the decay will grow continually shorter (as the inverse of the pump intensity), limited only by the pump pulsewidth of ~ 120fs. In contrast to this prediction, we find that both the initial and subsequent decay time constants in our double exponential fit at early times are essentially independent of the pump fluence, and are close to 450 fs and 6 ps, respectively. The apparent intensity-dependent decay on the shortest timescales results from an evolution of the relative magnitudes of these two decay components. The pump intensity-dependence of the amplitudes of the fast (A t) and the slow (A s) components is displayed in the inset to Fig. 4. These data reveal the dependence Af a A~, which is indicative of two correlated components, contributing to the PA 1 feature. Given the observed quadratic pump-dependence of Af, and the ultrafast PA~ build-up dynamics (see inset to Fig. 2(b)), it is reasonable to infer that the fast PA l component originates from doubly excited chains (intrachain two-exciton states) which are generated during the pump laser pulse via sequential re-excitation of singly excited molecules. Furthermore, the observed correlation in the initial fast decay of PA 1 ( A f ~ 450fs) and the build-up dynamics of the PA 2 band shown in the inset to Fig. 2(b) indicates that the generation of interchain species occurs at the expense of the doubly excited

116

V.L Klimov et al. / ChemicalPhysicsLetters 277 (1997) 109-117

molecules. This generation mechanism is confirmed by the quadratic correlation in the pump dependence of the PAj and PA 2 features shown in inset to Fig. 3(b), and is in agreement with a reduction of the PL yield at high pump fiuences observed in PPV films [4,25,29]. The above picture suggests that at high pump fluences the PA l absorption has two different components - a short-lived one, arising from doubly excited chains, and a longer-lived component, associated with singly excited intrachain species. The quadratic intensity-dependence of the magnitude of the two-exciton contribution leads to the experimentally observed pump-dependent decay of the PA 1 band. A central conclusion of this and other recent studies on MEH-DSB is that the interchain nonemissive species are likely generated as a byproduct of the decay of intrachain excitons, preferentially at high fluences, and primarily as the result of bimolecular or two-exciton interactions [13,1]. We emphasize here the difference between the intensity-dependent decay predicted for bimolecular exciton-exciton annihilation, which has been suggested to explain the relaxation dynamics in conjugated polymers by several authors [13,30,25,31], and the more complex correlated decay of singly and doubly excited excitons as suggested by the present results. In the former case, the nonlinearity occurs via the interaction and annihilation of excitons: a nonlinear decay process. In the latter case, the nonlinearity occurs during the generation process, within the pump pulsewidth, and results from sequential absorption o f pump photons into one- and two-exciton states. A

simple bimolecular annihilation picture fails in two respects to predict the correct dynamics at the shortest times for correlated intrachain and interchain excitons. First, as described above, the time constant for the fast decay in a bimolecular process should be inversely related to the intensity, limited only by the pulsewidth. Equally important, if the secondary species is generated as a byproduct of the nonlinear decay of excitons, then the formation time of the secondary species should also become progressively faster, remaining complementary to the fastest time constant of the annihilation decay. In fact we observe such complementary behavior between PAj and PA 2, but the initial time constant never falls below ~ 500 fs, well above our experimental resolution. To

explain this behavior, one must either postulate the existence of a relaxation bottleneck which limits the early time annihilation to 500 fs, or one may infer, as in the present paper, that the nonlinearity is observed primarily in the generation process. In conclusion, we have performed femtosecond TA studies in both solutions and thin films of a five-ring PPV oligomer-MEH-DSB. In both types of samples, we see the signatures of singlet excitons in both PA and SE. However, the TA spectra of solidstate samples are strongly affected by intermolecular interactions resulting in an additional short-wavelength PA feature attributed to non-emissive interchain excitons. A close match of dynamics for the long-wavelength PA and SE in pristine films over a wide pump-intensity range demonstrates unambiguously that these features originate from the same species identified as intrachain singlet excitons. Photochemical degradation of the solid-state samples is demonstrated to dramatically shorten the SE dynamics above a moderate incident pump fluence, whereupon the dynamics of the SE and the long-wavelength PA no longer coincide. We observe clear correlations between the subpicosecond relaxation of the long-wavelength PA and the build-up dynamics of the short-wavelength PA component. These data, along with the observed quadratic correlation in pump intensity-dependence of the magnitude of the two PA features, indicate that the formation of the interchain species is mediated by two-exciton states (doubly excited molecules) generated during the laser pulse sequentially via intra-chain excitons.

Acknowledgements We thank I. Campbell for the preparation of MEH-DSB samples, and I. Campbell, N. Kirova, S. Brazovski, and R. Kohlman for helpful discussions. This work was supported by Los Alamos Directed Research and Development funds, under the auspices of the US Department of Energy.

References [1] D. McBranch.M. Sinclair,in: N. Sariciftci(Ed.), The Nature of the Photoexcitationsin ConjugatedPolymers,World Scientific Publishing, Singapore, 1997.

V.L Klimov et al. / Chemical Physics Letters 277 (1997) 109-117

[2] R. Kersting, U. Lemmer, R.F. Mahrt, K. Leo, H. Kurz, H. Biissler, E.O. GSbel, Phys. Rev. Lett. 70 (1993) 3820. [3] I.D.W. Samuel, F. Raksi, D.D.C. Bradley, R.H. Friend, P.L. Bum, A.B. Holmes, H. Murata, T. Tsutsui, S. Saito, Synthetic Metals 55 (1993) 15. [4] J. Leng, S. Jeglinski, X. Wei, R. Benner, Z. Vardeny, Phys. Rev. Lett. 72 (1994) 156. [5] J. Hsu, M. Yah, T. Jedju, L. Rothberg, B. Hseih, Phys. Rev. B 49 (1994) 712. [6] M. Yan, L. Rothberg, F. Papadimitrakopouios, M. Galvin, T. Miller, Phys. Rev. Lett. 72 (1994) 1104. [7] M. Yan, L. Rothberg, B. Hsieh, R. Alfano, Phys. Rev. B 49 (1994) 9419. [8] M. Yah, L. Rothberg, E. Kwoek, T. Miller, Phys. Rev. Lett. 75 (1995) 1992. [9] W. Graulmer, G. Leising, G. Lanzani, M. Nisofi, S. DeSilvestri, U. Scberf, Phys. Rev. Lett. 76 (1996) 847. [I0] N. Harrison, G. Hayes, R. Phillips, R. Friend, Phys. Rev. Lett. 77 (1996) 1881. [1 I] J. Blatehford, S. Jessen, L. Lih, T. Gustafson, A. Epstein, D. Fu, T. Swagler, A. Maediarmid, S. Yamaguchi, H. Hamaguchi, Phys. Rev. Lett. 76 (1996) 1513. [12] B. Schwartz, F. Hide, M. Andersson, A. Heeger, Chem. Phys. Lett. 265 (1997) 327. [13] E. Maniloff, V. Klimov, D. McBranch, Phys. Rev. B 56 (1997) 1876. [14] S. Mukamel, H.X. Wang, Phys. Rev. Lett. 69 (1992) 65. [15] S. Abe, M. Schreiber, W.P. Su, J. Yu, Phys. Rev. B 45 (1992) 9432. [16] A.J. Heeger, S. Kiveison, J.R. Schrieffer, W.-P. Su, Rev. Mud. Phys. 60 (1988) 781. [17] R. Friend, D. Bradley, P. Townsend, J. Phys. D 20 (1987) 1367.

117

[18] M. Joswick, I. Campbell, N. Barashkov, J. Ferraris, J. Appl. Phys. 80 0996) 2883. [19] M. Fahlman, M. Logdlund, S. Stafstrom, W. Salaneck, R. Friend, P. Bum, A. Holmes, Macromolecules 28 0995) 1959. [20] B. Kohler, I. Samuel, J. Chem. Phys. 103 0995) 6248. [21] J. Cornil, D. Beljonne, Z. Shuai, T. Hagler, I. Campbell, D. Bradley, J. Bredas, C. Spangler, K. Mullen, Chem. Phys. Lett. 247 (1995) 425. [22] V. Klimov, D. McBranch, Opt. Lett., submitted. [23] L. Smilowitz, A. Hays, A.J. Heeger, G. Wang, J. Bowers, J. Chem. Phys. 98 (1993) 6504. [24] G. Kanner, X. Wei, B. Hess, L. Cben, Z. Vardeny, Phys. Rev. Lett. 69 (1992) 538. [25] G. Denton, N. Harrison, N. Tessler, R. Friend, Phys. Rev. Lett. 78 (1997) 733. [26] L. Rothberg, F. Papadimitrakopoulos, M. Galvin, Synthetic Metals 80 (1996) 41. [27] M. Yah, L. Rothberg, F. Papadimitrakopoulos, M. Galvin, T. Miller, Phys. Rev. Lett. 73 (1994) 744. [28] D. Sutherland, J.A. Carlisle, P. Elliker, G. Fox, T. Hagler, I. Jimenez, H. Lee, K. Pakbaz, L. TermineUo, S. Williams, F. Himpsel, D. Shuh, W. Wong, J. Jia, T. Calcott, D. Ederer, Appl. Phys. Lett. 68 (1996) 2046. [29] J.W. Yu, J.H. Hsu, K.R. Chuang, C.L. Chan, S.A. Cben, F.J. Kao, W.S. Farm, H. Lin, Synthetic Metals 74 (1995) 7. [30] R. Kepler, V. Valencia, S. Jacobs, J. McNamara, Synthetic Metals 78 (1996) 227. [31] K. Wong, D. Bradley, W. Hayes, J. Ryan, R. Friend, H. Lindenberger, S. Roth, J. Phys. C: Solid State Phys. 20 (1987) L187.