16.11.2013 Views

Raman Anisotropy Measurements: An Effective Probe of Molecular ...

Raman Anisotropy Measurements: An Effective Probe of Molecular ...

Raman Anisotropy Measurements: An Effective Probe of Molecular ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

FULL PAPER<br />

<strong>Raman</strong> <strong><strong>An</strong>isotropy</strong> <strong>Measurements</strong>: <strong>An</strong> <strong>Effective</strong> <strong>Probe</strong> <strong>of</strong> <strong>Molecular</strong><br />

Orientation in Conjugated Polymer Thin Films**<br />

By Hai-Ming Liem, Pablo Etchegoin,* Katherine S. Whitehead, and Donal D. C. Bradley*<br />

Understanding the molecular alignment <strong>of</strong> conjugated polymers within thin-film samples is essential for a complete picture <strong>of</strong><br />

their optical and transport properties, and hence for the continued development <strong>of</strong> optoelectronic device applications. We<br />

report here on the efficacy <strong>of</strong> <strong>Raman</strong> anisotropy measurements as a probe <strong>of</strong> molecular orientation, presenting results for<br />

aligned polyfluorene nematic glass films. Comparison is made with the results <strong>of</strong> optical dichroism measurements performed<br />

on the same samples. We show that in many cases molecular orientation can be more directly characterized by <strong>Raman</strong> anisotropy,<br />

and that it can have a greater sensitivity to the degree <strong>of</strong> molecular orientation than conventional optical dichroism. The fact<br />

that the <strong>Raman</strong> measurements can be readily performed on the same thin films (~ 100 nm thickness) that are required for<br />

optical dichroism means that there is no ambiguity in a direct comparison <strong>of</strong> results. This situation differs from that for standard<br />

X-ray diffraction measurements (these require film thicknesses <strong>of</strong> several lm) and electron diffraction or electron energy loss<br />

spectroscopy measurements (these require film thicknesses <strong>of</strong> 10 nm or less). The <strong>Raman</strong> data allow the angle (relative to the<br />

chain axis) for the optical dipole transition moment to be deduced from the dichroic ratio, and confirm the role that its <strong>of</strong>f-axis<br />

component plays in limiting this ratio. The added fact that <strong>Raman</strong> anisotropy data can be collected in situ, in reflection geometry<br />

for standard device structures, and with microscopic resolution and chemical specificity makes the technique even more<br />

attractive as a non-invasive device probe.<br />

1. Introduction<br />

The intrinsic anisotropy associated with strong intra- and<br />

weak inter-chain p-bonding within conjugated polymer films<br />

opens up the possibility <strong>of</strong> constructing semiconductor devices<br />

with novel electrical and optical properties, including a highly<br />

polarized emission [1±6] that is <strong>of</strong> interest for liquid-crystal display<br />

backlights. Orientation can also facilitate enhanced charge<br />

transport [7,8] and optical confinement, [9±11] leading to improved<br />

optoelectronic and photonic devices. Enhanced mobility <strong>of</strong>fers<br />

advantages for light emitting diode (LED) displays, since it allows<br />

higher current densities and thus <strong>of</strong>fers access to the high<br />

peak brightness needed for passive matrix addressed displays. [7]<br />

Thin-film transistors benefit from higher mobility through faster<br />

switching. [8] In photovoltaic diodes, mobility limits efficiency<br />

by controlling the film thickness from which carriers can be<br />

efficiently extracted. For typical organic materials this is below<br />

the optimum for full light absorption. [12] Mobility is also expected<br />

to be important for achieving electrically pumped<br />

amplifiers and lasers in which high current densities are<br />

required. Optical confinement is assisted both by the orientation<br />

<strong>of</strong> the emission dipoles [9] and by the large accompanying<br />

±<br />

[*] Pr<strong>of</strong>. D. D. C. Bradley, Dr. P. Etchegoin, Mr. H. M. Liem,<br />

Dr. K. S. Whitehead<br />

Imperial College <strong>of</strong> Science, Technology, and Medicine<br />

Prince Consort Road, SW7 2BZ London (UK)<br />

E-mail: d.bradley@imperial.ac.uk, p.etchegoin@imperial.ac.uk<br />

[**] This work was supported by the United Kingdom Engineering and Physical<br />

Sciences Research Council (GR/M21201) and the Dow Chemical Company<br />

(ªPolymer Semiconductorsº). We thank Mark Bernius, Mike Inbasekaran<br />

and Jim O'Brien <strong>of</strong> the Dow Chemical Company for providing the polymers<br />

used in this study.<br />

birefringence, [10] and leads to reduced gain thresholds as a<br />

result <strong>of</strong> reduced mode volumes. [11] In order to optimize these<br />

benefits for device application, relatively high degrees <strong>of</strong><br />

molecular orientation are required, somewhat limiting the<br />

choice <strong>of</strong> materials systems and orientation methods. Grell and<br />

Bradley [4] have recently reviewed the methods for alignment <strong>of</strong><br />

molecular electronic materials. Uniaxial alignment is relatively<br />

easy to achieve via tensile drawing, [13,14] mechanical rubbing,<br />

[15±17] and via lyotropic [3] or thermotropic [1,18] mesophase<br />

formation. In particular, the availability <strong>of</strong> a thermotropic liquid-crystal<br />

phase allows a straightforward approach to the fabrication<br />

<strong>of</strong> highly oriented thin films on device substrates. [3,5,8,10]<br />

Spin coating a film onto a rubbed [1,3,5] or photoactivated [19]<br />

alignment layer is followed by a heat-treatment protocol to<br />

initiate the orientation process. The film is then either<br />

quenched to room temperature to freeze in the alignment within<br />

a liquid crystalline glass, or slowly cooled to generate a crystalline<br />

film. Mono-domain alignment has been demonstrated<br />

for a range <strong>of</strong> fluorene homopolymers [1,3,20] and copolymers. [18]<br />

Photoluminescence (PL) and electroluminescence (EL) anisotropy<br />

and optical dichroism measurements have been used<br />

to directly characterize the emission properties and to probe<br />

the underlying degree <strong>of</strong> molecular alignment [1,2,5,7,14,17,18,21] .<br />

Typically, the molecular order parameters are estimated from<br />

the ratio at the long wavelength absorption peak for light<br />

polarized parallel and perpendicular to the alignment direction,<br />

as defined for example by the rubbing direction <strong>of</strong> the<br />

alignment layer. PL and EL measurements use the intensity<br />

ratio at the peak <strong>of</strong> the emission spectrum, or the ratio <strong>of</strong> the<br />

integrated areas <strong>of</strong> the emission spectra for parallel and perpendicular<br />

polarizations. Each <strong>of</strong> the resulting anisotropies is<br />

expected to be influenced by the details <strong>of</strong> the polymer elec-<br />

66 Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1616-301X/03/0101-0066 $ 17.50+.50/0 Adv. Funct. Mater. 2003, 13, No. 1, January


H.-M. Liem et al./<strong>Molecular</strong> Orientation in Conjugated Polymer Films<br />

tronic structure, by the nature and effectiveness <strong>of</strong> the interactions<br />

with the alignment substrate, by the alignment protocols<br />

used, and by the sample's thermal history. The largest anisotropies<br />

obtained so far for polyfluorene homo- and co-polymers<br />

aligned on rubbed substrates are ~ 25 in EL, [5] ~ 10 in PL [5] and<br />

~ 11 in UV-vis absorption. [3] Two questions that arise are:<br />

i) How does the anisotropy relate to the structural order<br />

parameter, S, describing the average microscopic orientation <strong>of</strong><br />

molecular chain axes relative to the macroscopic alignment<br />

direction?<br />

ii) Why is the anisotropy found to be different for different<br />

experimental techniques?<br />

The optical dichroism, D, is expected to be related to S by: [22]<br />

D £ (1 + 2S)/(1 ± S) (1)<br />

Note that the equality applies only for a transition dipole<br />

moment that is aligned along the molecular chain axis, and for<br />

a mono-domain sample. Mono-domain formation has been<br />

clearly demonstrated for the thermotropic polyfluorene samples<br />

that we have studied to date. [1,18] The alignment <strong>of</strong> the<br />

transition dipole is, however, unknown. Polyphenylenes and<br />

polyarylene ethynylenes are expected to have axial transition<br />

dipole moments on account <strong>of</strong> their extended linear structure.<br />

[2] Polyfluorenes are not. Theoretical considerations <strong>of</strong> the<br />

polyfluorene geometry and experimental light scattering studies<br />

[23] point to an angle <strong>of</strong> some 19 between the r-bonds at<br />

opposite ends <strong>of</strong> each fluorene moiety. This distortion <strong>of</strong> shape<br />

compared with a linear biphenyl moiety is forced by the C9 carbon<br />

link, and can be expected to induce a non-axial transition<br />

dipole. As a consequence, even for perfect chain alignment<br />

(i.e., S = 1), the optical dichroism would be limited to a finite<br />

value. This situation can be described by rewriting Equation 1<br />

with an effective order parameter S¢ = S ” S op with:<br />

S op = 1/2 (3 cos 2 b ±1) (2)<br />

Here b is the angle that the transition dipole makes with the<br />

polymer chain axis. Interestingly, the value <strong>of</strong> b for a single<br />

molecular repeat unit does not necessarily dictate the value for<br />

the effective chromophore as a whole (the polymer chain segment<br />

involved in the transition). The chain conformation/configuration<br />

may lead to cancellation <strong>of</strong> <strong>of</strong>f-axis transition dipole<br />

contributions from neighboring repeat units within an alternating<br />

higher order structure. This is the case for the 2 1 helical<br />

conformation <strong>of</strong> poly(9,9-dioctylfluorene) (PFO), for which<br />

the optical dichroism is substantially enhanced. [23] A trans±cisoid<br />

configuration for polyarylene vinylenes [14] similarly leads to<br />

an enhancement <strong>of</strong> the photoluminescence anisotropy.<br />

We note also that the effects <strong>of</strong> heterogeneity need to be<br />

considered when comparing EL and PL anisotropy and optical<br />

dichroism measurements. In the EL experiments the location<br />

<strong>of</strong> the recombination zone, carrier trapping effects, and energy<br />

migration processes lead to selection <strong>of</strong> a subset <strong>of</strong> chain segments<br />

from which the emission occurs. In PL it is simply energy<br />

migration that matters but the effect is the same. For absorption<br />

dichroism, however, the whole <strong>of</strong> the heterogeneous ensemble<br />

<strong>of</strong> chain segments is probed simultaneously. This measurement-dependent<br />

selectivity <strong>of</strong> emission sites has been cited<br />

as an explanation for the differences seen between absorption,<br />

PL and EL anisotropy measurements in PFO. [5]<br />

There have been attempts to compare anisotropies obtained<br />

by optical methods with truly structural probes like X-ray diffraction.<br />

However, as noted above, these measurements cannot<br />

be readily taken from the same films. X-ray experiments are<br />

typically conducted on fibers or thick films, and the degree <strong>of</strong><br />

molecular alignment may not then be directly comparable with<br />

that for the thin films needed for optical dichroism and used in<br />

EL devices. Indeed, materials such as PFO are known to show<br />

strong morphology dependence as a function <strong>of</strong> processing<br />

conditions. [24] Access to a structural probe that would allow<br />

measurements to be undertaken on the same thin film samples<br />

used for optical anisotropy measurements, would ensure direct<br />

comparability <strong>of</strong> results. Removing the ambiguity would then<br />

allow determination <strong>of</strong> b, thus facilitating a greater understanding<br />

<strong>of</strong> the differences between absorption, PL and EL anisotropy.<br />

We report here on the use <strong>of</strong> <strong>Raman</strong> scattering as just such a<br />

thin-film structural probe, and show that it is very effective in<br />

this regard. The use <strong>of</strong> <strong>Raman</strong> scattering data requires knowledge<br />

<strong>of</strong> the <strong>Raman</strong> tensor for one or more specific vibrations<br />

so that the angular variation <strong>of</strong> the <strong>Raman</strong> intensity can be<br />

mapped onto a molecular-orientation distribution function.<br />

Such information is in general either unavailable or difficult to<br />

infer, but we have previously investigated the tensor properties<br />

<strong>of</strong> the ~ 1600 cm ±1 in-plane stretching mode characteristic <strong>of</strong><br />

phenyl/phenylene rings. [25] This mode has a highly uniaxial<br />

<strong>Raman</strong> tensor that is little dependent on the details <strong>of</strong> chemical<br />

structure. It is therefore suitable for use as a monitor <strong>of</strong> molecular<br />

structure. We note further that this mode is expected to be<br />

found in the <strong>Raman</strong> spectra <strong>of</strong> a wide range <strong>of</strong> conjugated<br />

polymers that are currently under development for device<br />

applications, and hence that the following discussions have reasonable<br />

generality.<br />

2. Results and Discussion<br />

2.1. Theoretical Description <strong>of</strong> Optical Dichroism and <strong>Raman</strong><br />

<strong><strong>An</strong>isotropy</strong> and Their Dependence on <strong>Molecular</strong> Orientation<br />

We consider a distribution <strong>of</strong> rod-shaped molecules (representing<br />

individual chain segments) in the coordinate system<br />

defined in Figure 1. We restrict ourselves to the case <strong>of</strong> uniaxial<br />

polymer-chain orientation, i.e., to systems in which only the<br />

angles h and b (see Fig. 1) take non-random values. This is expected<br />

to be entirely appropriate for the oriented nematic-glass<br />

samples that we have studied here. We also introduce the simplest<br />

description <strong>of</strong> the orientation distribution function due to<br />

Fraser. [26]<br />

We do this to gain insight into the problem, and for lack <strong>of</strong> a<br />

justifiable alternative. The assumption is that a fraction, f, <strong>of</strong><br />

the molecules are perfectly oriented along z, whilst the remaining<br />

(1 ± f) are isotropically (randomly) oriented. The perfectly<br />

FULL PAPER<br />

Adv. Funct. Mater. 2003, 13, No. 1, January Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1616-301X/03/0101-0067 $ 17.50+.50/0 67


H.-M. Liem et al./<strong>Molecular</strong> Orientation in Conjugated Polymer Films<br />

FULL PAPER<br />

θ<br />

φ<br />

β<br />

µ<br />

ψ<br />

H 17 C 8 C 8 H 17<br />

H 17 C 8 C 8 H 17<br />

N N<br />

S<br />

Fig. 1. The general coordinate system used to describe the molecular orientation<br />

<strong>of</strong> a polymer chain axis with respect to the macroscopic sample axes x, y, andz.<br />

The angles h and u define the orientation <strong>of</strong> the chain axis with respect to the<br />

axes x, y, andz. The dipole transition moment l lies at an angle b with respect to<br />

this axis, with an azimuthal angle w. The chemical structures <strong>of</strong> PFO (upper) and<br />

F8BT (lower) are also shown.<br />

oriented fraction, f, corresponds to the Hermans' orientation<br />

function that can be derived from X-ray diffraction data. [27]<br />

Note also that if the Fraser distribution function is expanded in<br />

terms <strong>of</strong> spherical harmonics, f can be shown to be equal to the<br />

average <strong>of</strong> the second-order term in the expansion, namely<br />

áP 2 cos bñ. [28] We next assume that we are considering molecular<br />

vibrations with a highly uniaxial <strong>Raman</strong> tensor (as is the<br />

case for the ~ 1600 cm ±1 in-plane stretching mode [25] ) and that<br />

their principal axis is coincident with the polymer-chain axis.<br />

This might seem an inadvisable restriction, but the experimental<br />

evidence indicates that there are many families <strong>of</strong> polymers<br />

where this condition applies (and sometimes for more than one<br />

mode (see below)). In accordance with a highly uniaxial <strong>Raman</strong><br />

tensor, we assume that only the a zz component <strong>of</strong> the<br />

polarizability makes a significant contribution to the scattering<br />

efficiency for optical polarizations that are accessible in backscattering<br />

from the plane <strong>of</strong> a thin film in the experimental<br />

geometry shown in Figure 2.<br />

Backscattered<br />

light<br />

e →<br />

S<br />

Aligned film is<br />

rotated to measure<br />

Incoming<br />

laser<br />

e → L<br />

Alignment<br />

direction<br />

Fig. 2. Schematic scattering geometry: <strong>Measurements</strong> were made with the polarization<br />

direction <strong>of</strong> the laser e L initially perpendicular to the alignment direction<br />

<strong>of</strong> the film. The scattered polarization e S was kept parallel to e L for all measurements<br />

and the film was rotated.<br />

A word must be said at this stage about the nature <strong>of</strong> the<br />

<strong>Raman</strong> tensor because this is crucial for the applicability <strong>of</strong> the<br />

technique. In previous studies, we have found different depolarization<br />

ratios for different experimental conditions, a situation<br />

that can only be rationalized if a certain degree <strong>of</strong> alignment<br />

<strong>of</strong> the chains occurs. When the polymer was dissolved in<br />

a solvent (i.e., the orientation <strong>of</strong> the rings is completely random),<br />

the depolarization ratio <strong>of</strong> the strongest in-plane breathing<br />

(stretching) mode <strong>of</strong> the phenylene rings was 0.74. For the<br />

same reason, this value was also found in films that were spin<br />

coated under conditions where solvent evaporation occurs<br />

rapidly, leading to a freezing-in <strong>of</strong> the isotropic structure found<br />

in solution. [25] This suggests a traceless tensor as proposed by<br />

Liem et al. [25] The tensor must have two diagonal components<br />

different from zero: Otherwise it could not explain both the<br />

isotropic depolarization ratio seen in solution and the depolarization<br />

ratio <strong>of</strong> one seen, under special circumstances, at the<br />

polymer surface. [25] This is in good agreement with the known<br />

properties <strong>of</strong> the <strong>Raman</strong> tensor for the equivalent mode in<br />

benzene, and with the fact that benzene-related modes do not<br />

drastically change their symmetry even when the environment<br />

<strong>of</strong> the benzene ring is altered. There is plenty <strong>of</strong> experimental<br />

evidence to the latter effect from detailed <strong>Raman</strong> studies on<br />

mono- and di-substituted benzene molecules. In the case <strong>of</strong> oriented<br />

films, our previous work [29] suggests that the phenylene<br />

rings <strong>of</strong> the backbone lie perpendicular to the surface. Accordingly,<br />

the polarized <strong>Raman</strong> measurements have access to a projection<br />

<strong>of</strong> the <strong>Raman</strong> tensor where only one <strong>of</strong> the components<br />

is different from zero. This is what produces the large anisotropy<br />

that we observe experimentally (see below).<br />

In this paper we consider how a departure from perfect<br />

alignment along z will influence the observed anisotropy and<br />

therefore how we can use <strong>Raman</strong> anisotropy to probe the chain<br />

alignment. It is interesting, and perhaps somewhat surprising,<br />

to note that straightforward <strong>Raman</strong> anisotropy measurements<br />

seem not to have previously been used to study the extent <strong>of</strong><br />

molecular alignment in conjugated polymer films. This is despite<br />

the very large number <strong>of</strong> publications in the literature<br />

that have been published concerning the optical and structural<br />

properties <strong>of</strong> these materials. The only related studies concern<br />

<strong>Raman</strong> measurements on oriented films <strong>of</strong> trans- [30] and cispolyacetylene.<br />

[31] All <strong>of</strong> these studies were performed under<br />

resonance-excitation conditions, with the primary interest<br />

focused on the role that the electronic excitation plays in determining<br />

the observed features <strong>of</strong> the <strong>Raman</strong> spectra. A particular<br />

focus was on understanding the nature <strong>of</strong> the vibrational±<br />

electronic mode coupling effects that arise under resonance<br />

excitation, especially for trans-polyacetylene. The variation in<br />

depolarization across the carbon±carbon <strong>Raman</strong> modes and<br />

the variations in lineshape, parametric in resonant-excitation<br />

wavelength, were interpreted within a conjugation-length distribution<br />

model on the assumption that axial alignment <strong>of</strong> the<br />

<strong>Raman</strong> tensor occurs along the molecular chain. [30] Experiments<br />

on cis-polyacetylene had a similar focus, with an additional<br />

interest in understanding the changes that occur during<br />

the isomerization process from cis- totrans-polyacetylene. [31]<br />

We emphasize that resonance excited <strong>Raman</strong> depolarization<br />

measurements were used in all <strong>of</strong> these studies. Furthermore,<br />

the measurements were performed on stretch-aligned films<br />

several micrometers thick, for which X-ray diffraction was used<br />

to characterize the state <strong>of</strong> order, and for which significant,<br />

model based, corrections <strong>of</strong> the scattering intensities were<br />

required to account for the effect <strong>of</strong> the anisotropic optical<br />

constants. In contrast, our measurements are performed with<br />

non-resonant excitation and, because <strong>of</strong> the experimental ar-<br />

68 Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1616-301X/03/0101-0068 $ 17.50+.50/0 Adv. Funct. Mater. 2003, 13, No. 1, January


H.-M. Liem et al./<strong>Molecular</strong> Orientation in Conjugated Polymer Films<br />

rangement that we use (see below for further details), we do<br />

not need to calculate correction coefficients for the scattering<br />

intensities. Our focus is on using <strong>Raman</strong> anisotropy to extract<br />

information on the state <strong>of</strong> order, not on using resonance enhancements<br />

to understand the electronic±vibrational mode<br />

coupling. Moreover, we study thin film samples, as required for<br />

device structures, and for which X-ray scattering measurements<br />

<strong>of</strong> the state <strong>of</strong> order are difficult to perform.<br />

In our experimental scattering geometry (see Fig. 2), when<br />

the polymer is highly oriented, one <strong>of</strong> the components <strong>of</strong> the<br />

tensor (in the direction <strong>of</strong> the beam wave vector k) cannot be<br />

measured because the polarization is always perpendicular to<br />

k. We define the direction perpendicular to z within the plane<br />

<strong>of</strong> the film to be x. The light beam then impinges onto the sample<br />

at normal incidence along y. Under these conditions, the<br />

<strong>Raman</strong> intensity due to the perfectly oriented fraction I(perf)<br />

for incoming and outgoing beams both polarized along z, and<br />

for a normalized incident laser intensity, I = 1, is given by:<br />

I zz (perf) = fa 2 zz (3)<br />

and components I zx (perf) and I xx (perf) are clearly zero.<br />

For the random (1 ± f) fraction we have to evaluate the average<br />

<strong>of</strong> the <strong>Raman</strong> tensor. After Margulies and Stockburger, [32]<br />

we can write this average using tensor invariants:<br />

I zz (rand) = I xx (rand) = 1 45 (1 ± f)(45a2 +4c 2 ±5d 2 ) (4)<br />

Here a 2 , c 2 , and d 2 are the usual isotropic, anisotropic, and<br />

asymmetric invariants <strong>of</strong> the <strong>Raman</strong> tensor. Since a zz is the<br />

only non-zero component in the plane, these invariants take<br />

simple forms, namely a 2 = 1/9a 2 zz, c 2 = a 2 zz, and d 2 = 0. The expected<br />

<strong>Raman</strong> anisotropy, defined as R = I zz /I xx , then becomes:<br />

R = I zz<br />

ˆ Izz …perf†‡I zz …rand† ˆ …1‡4 f †<br />

I xx I xx …rand† …1 f †<br />

and hence,<br />

…R 1†<br />

f = (6)<br />

…R‡4†<br />

We therefore have a simple relationship linking the perfectly<br />

oriented fraction f <strong>of</strong> Fraser's model with the observed <strong>Raman</strong><br />

anisotropy. Note that, as expected for random (isotropic) alignment,<br />

R = 1 and f = 0, whereas for perfect uniaxial alignment<br />

R = ¥ and f =1.<br />

We can now calculate the corresponding relationship expected<br />

between the optical absorption dichroism and the perfectly<br />

oriented fraction f <strong>of</strong> Fraser's model. The dichroism, D,<br />

is defined as D = A z /A x , where A z and A x are the absorption<br />

intensities at the long wavelength absorption peak <strong>of</strong> the first<br />

dipole-allowed transition for polarization parallel to z and x,<br />

respectively. Assuming that there are N absorption dipoles per<br />

unit volume in the sample, each with a transition moment l,<br />

and that this transition moment is, like the <strong>Raman</strong> tensor,<br />

aligned with the polymer-chain axis, the observed intensities<br />

would be given by:<br />

(5)<br />

A z =kNl 2 f+ 1 3 kNl2 (1 ± f) (7)<br />

A x = 1 3 kNl2 (1 ± f) (8)<br />

with k a constant. The kNl 2 f term in Equation 7 is the absorbance<br />

due to oscillating dipoles within the perfectly oriented<br />

fraction, whereas the 1/3kNl 2 (1 ± f) term is the projected contribution<br />

along z from the isotropically distributed fraction.<br />

The x component has only contributions from the latter. The<br />

dichroic ratio would then be given by:<br />

D = 1‡2 f<br />

1 f<br />

and hence,<br />

f = D 1<br />

D‡2<br />

(9)<br />

(10)<br />

It would then be possible to determine a value for f by measuring<br />

D. As expected, an absence <strong>of</strong> dichroism (D = 1) corresponds<br />

to an isotropic sample (f = 0) and infinite dichroism corresponds<br />

to a perfectly aligned sample (f = 1).<br />

In general, however, the transition dipole will not lie along<br />

the polymer-chain axis, but will rather be oriented at an angle<br />

b, as already shown in Figure 1 and discussed above. The effect<br />

that the angle b has on the expected relationship between f and<br />

D is, from Fraser, [28] as follows:<br />

…D 1† …D<br />

f = 0<br />

‡2†<br />

…D‡2† …D 0<br />

1†<br />

(11)<br />

Here D is the dichroic ratio that would be observed if the all<br />

chains were parallel to z. It can be shown [28] that D 0 = 2 cot 2 b,<br />

such that:<br />

<br />

2 …D 1†<br />

f =<br />

3 cos 2 b 1 …D‡2†<br />

(12)<br />

Equation 12 can then be rearranged to yield an expression<br />

for D in terms <strong>of</strong> f:<br />

D = 1‡2 f hP cos bi<br />

2<br />

1 f hP 2<br />

cos bi<br />

(13)<br />

Here áP 2 cos bñ is the second order Legendre polynomial in<br />

cos b. Its value varies from ±1/2 for a transition dipole oriented<br />

perpendicular to the polymer chain axis, through zero for random<br />

orientation (where ácos 2 bñ = 1/3), to unity for parallel orientation.<br />

Correspondingly, D varies from (2 ±2f)/(2 + f) for a<br />

perpendicularly oriented transition dipole to (1 + 2f)/(1 ± f) for<br />

a parallel dipole. A comparison with Equation 5 immediately<br />

reveals that R should always be greater than D for a given value<br />

<strong>of</strong> f. Alternatively, R and D can be related to the second and<br />

fourth order parameters, P 2 and P 4 , as discussed in detail in<br />

LagugnØ Labarthet et al. [33]<br />

FULL PAPER<br />

Adv. Funct. Mater. 2003, 13, No. 1, January Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1616-301X/03/0101-0069 $ 17.50+.50/0 69


0<br />

H.-M. Liem et al./<strong>Molecular</strong> Orientation in Conjugated Polymer Films<br />

FULL PAPER<br />

2.2. Experimental Studies on Oriented PFO and F8BT<br />

Nematic Glass Films<br />

<strong>Raman</strong> scattering and UV-vis absorption dichroism measurements<br />

were performed on oriented nematic-glass films <strong>of</strong> PFO<br />

and poly[(9,9-dioctylfluorene)-co-benzothiadiazole] (F8BT),<br />

and the results are now directly compared within the theoretical<br />

framework discussed above. <strong>Raman</strong> spectra were measured<br />

with the incident laser light polarized parallel (i.e., I zz ) and perpendicular<br />

(i.e., I xx ) to the alignment direction <strong>of</strong> the film. The<br />

resulting spectra are plotted in Figure 3 across the spectral<br />

region <strong>of</strong> the dominant 1605 cm ±1 phenyl/phenylene in-plane<br />

stretching mode in PFO and F8BT, and BT stretching mode at<br />

a)<br />

270<br />

b)<br />

300<br />

240<br />

300<br />

330<br />

210<br />

330<br />

0<br />

180<br />

0<br />

30<br />

150<br />

30<br />

60<br />

90<br />

120<br />

60<br />

a)<br />

270<br />

90<br />

PF0 Intensity [arb. units]<br />

⊥<br />

1550 1580 1610 1640<br />

-1<br />

c)<br />

240<br />

210<br />

330<br />

180<br />

0<br />

150<br />

30<br />

120<br />

b)<br />

300<br />

60<br />

F8BT Intensity [arb. units]<br />

⊥<br />

1510 1540 1570 1600 1630<br />

-1<br />

Fig. 3. <strong>Raman</strong> spectra <strong>of</strong> a) an aligned PFO film and b) an aligned F8BT film,<br />

collected using 633 nm laser excitation. The spectra were taken for incoming and<br />

outgoing polarizations either both parallel (i), or both perpendicular (^) tothe<br />

alignment direction. <strong>Raman</strong> anisotropies <strong>of</strong> 16.67 and 43.33 were obtained for<br />

PFO and F8BT, respectively, from the ratio <strong>of</strong> these intensities.<br />

Fig. 4. Polar plots <strong>of</strong> parallel <strong>Raman</strong> intensities as a function <strong>of</strong> angle for the<br />

a) PFO 1605 cm ±1 (open diamonds), b) the F8BT 1544.9 cm ±1 (open triangles),<br />

and c) the F8BT 1608.5 cm ±1 (open squares) modes.<br />

~ 1545 cm ±1 . As already discussed, this mode has the highly uniaxial<br />

<strong>Raman</strong> tensor that is assumed in the above derivations.<br />

We also note that the <strong>Raman</strong> intensities for polarization precisely<br />

parallel and perpendicular to the orientation direction<br />

are expected to be independent <strong>of</strong> the strong birefringence that<br />

arises in aligned samples. [10,34] Thus by selecting data from measurements<br />

along the main optical axes <strong>of</strong> the film we can minimize<br />

the polarization scrambling effects expected to accompany<br />

the film birefringence. Differences in reflectivity arising<br />

from the birefringence along the main axes are negligible for<br />

the <strong>Raman</strong> measurements. We obtain <strong>Raman</strong> anisotropy values<br />

<strong>of</strong> 16.67 for the 1605 cm ±1 <strong>Raman</strong> active mode <strong>of</strong> PFO, and<br />

43.33 for both the 1544.9 cm ±1 and 1608.5 cm ±1 modes <strong>of</strong> F8BT.<br />

The anisotropy for the 1544.9 cm ±1 mode <strong>of</strong> F8BT implies that<br />

it too is characterized by a highly uniaxial <strong>Raman</strong> tensor with<br />

its principal axis along the chain direction. For completeness,<br />

the parallel <strong>Raman</strong> intensities as a function <strong>of</strong> angle are plotted<br />

in Figure 4 for the 1605 cm ±1 mode <strong>of</strong> PFO and the two modes<br />

<strong>of</strong> F8BT. The <strong>Raman</strong> intensity reaches a maximum (minimum)<br />

when the polarization direction <strong>of</strong> the laser (e L ) is parallel (perpendicular)<br />

to the alignment direction. At intermediate angles,<br />

the polarized <strong>Raman</strong> signal is affected by both the relative orientation<br />

<strong>of</strong> the <strong>Raman</strong> tensor with respect to e L and e S , and the<br />

intrinsic birefringence <strong>of</strong> the film. If e L is not parallel to the<br />

main dielectric axes <strong>of</strong> the sample, the polarized <strong>Raman</strong> intensities<br />

will be altered by polarization scrambling due to birefringence.<br />

Figure 5 shows the polarized absorption (dichroism) spectra<br />

in the region <strong>of</strong> the first dipole-allowed UV absorption <strong>of</strong><br />

a) PFO and b) F8BT. The observed dichroic ratios obtained as<br />

the absorbance ratio at the maxima <strong>of</strong> the absorption spectra<br />

for PFO and F8BT are 4.4 and 8.2, respectively. The <strong>Raman</strong> anisotropies<br />

are therefore four (in PFO) and five (in F8BT) times<br />

larger than the dichroic ratios obtained from polarized absorption.<br />

Note that larger dichroism values have been observed for<br />

more highly aligned PFO samples than studied here. Even<br />

then, however, the maximum dichroic ratio measured was only<br />

~ 7. That is a factor <strong>of</strong> more than two less than the results presented<br />

here for <strong>Raman</strong> scattering from a more poorly aligned<br />

sample. It is thus clear that, as anticipated theoretically above,<br />

higher anisotropy can be measured via <strong>Raman</strong> scattering than<br />

via dichroism.<br />

270<br />

240<br />

210<br />

180<br />

150<br />

120<br />

90<br />

70 Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1616-301X/03/0101-0070 $ 17.50+.50/0 Adv. Funct. Mater. 2003, 13, No. 1, January


H.-M. Liem et al./<strong>Molecular</strong> Orientation in Conjugated Polymer Films<br />

a)<br />

b)<br />

⊥<br />

340 390 440<br />

and dD/df for different values <strong>of</strong> f. It is straightforward to show<br />

that dR/df is smaller than dD/df at small f values, but that the<br />

opposite applies above a threshold value <strong>of</strong> f. The <strong>of</strong>f-axis dipole<br />

moment angle b plays a significant role in determining this<br />

threshold. Above the threshold, the sensitivity dR/df overtakes<br />

dD/df. As an example, the threshold f value is 0.07 for<br />

b = 26.5, but increases to 0.46 for b = 19. These thresholds are<br />

relatively low compared with the required orientations for<br />

practical application and it is therefore anticipated that, in<br />

many cases, <strong>Raman</strong> anisotropy measurements will be more<br />

sensitive than polarized absorption measurements as a probe<br />

<strong>of</strong> small changes in molecular alignment.<br />

FULL PAPER<br />

⊥<br />

400 450 500 550<br />

Fig. 5. Polarized optical absorption <strong>of</strong> a) PFO aligned film and b) F8BT aligned<br />

film, with light polarized parallel (//) and perpendicular (^) to the alignment direction.<br />

In the simplest model <strong>of</strong> molecular orientation treated here,<br />

and for a uniaxial <strong>Raman</strong> tensor (entirely consistent with the<br />

data in Fig. 4), the only parameter that determines the <strong>Raman</strong><br />

anisotropy is the fraction, f. For the PFO sample studied here,<br />

the value <strong>of</strong> f calculated from R using Equation 6 is 0.76. If we<br />

combine this with the measured value <strong>of</strong> D, we obtain b = 26.5<br />

(using Eq. 12). Similarly, for the F8BT aligned film, f is 0.89<br />

from the <strong>Raman</strong> measurement, and when combined with the<br />

optical dichroism measurement a value b = 22.0 can be deduced.<br />

It would clearly be interesting to compare these values<br />

with quantum chemical calculations <strong>of</strong> the transition dipole<br />

moment orientation with respect to the repeat unit geometry.<br />

Such a comparison would assist in assessing the relative importance<br />

<strong>of</strong> any polymer-chain superstructure on the effective orientation<br />

<strong>of</strong> the transition dipole in these materials. A study <strong>of</strong><br />

the angle b as a function <strong>of</strong> the degree <strong>of</strong> alignment, characterized<br />

for instance by the value <strong>of</strong> f deduced from <strong>Raman</strong> anisotropy<br />

data, would also be interesting in this respect, as would a<br />

study <strong>of</strong> the influence <strong>of</strong> the film structure (crystalline vs glassy<br />

vs the extended-chain phase [24] ).<br />

The specific b values deduced for PFO and F8BT seem reasonable.<br />

If, as discussed above, the transition dipole aligns with<br />

the para-para axis <strong>of</strong> a phenylene subunit <strong>of</strong> the fluorene moiety<br />

for these polymers, then b should be ~ 19. Note also that<br />

irrespective <strong>of</strong> the precise value <strong>of</strong> b, as previously discussed,<br />

its finite value is the main limitation on the magnitude <strong>of</strong> the<br />

dichroic ratio. Even for f = 1 (perfect orientation), with the<br />

deduced values <strong>of</strong> b, the largest possible dichroic ratios would<br />

be D = 8 and D = 12.3 for PFO and F8BT, respectively.<br />

We can further discuss the relative merits <strong>of</strong> dichroism and<br />

<strong>Raman</strong> anisotropy, this time in terms <strong>of</strong> their relative sensitivity<br />

to increasing degrees <strong>of</strong> orientation. This sensitivity is<br />

important in the context <strong>of</strong> the optimization <strong>of</strong> devices: The<br />

ability to reliably monitor small changes in orientation is critical<br />

to generate the largest anisotropy. We assess the different<br />

sensitivities <strong>of</strong> the techniques by considering the changes dR/df<br />

3. Conclusion<br />

In closing, we have reported polarized <strong>Raman</strong> and absorption<br />

measurements on oriented nematic-glass films <strong>of</strong> the two<br />

conjugated polyfluorenes PFO and F8BT. These data were<br />

used to demonstrate the efficacy <strong>of</strong> <strong>Raman</strong> anisotropy measurements<br />

as a probe <strong>of</strong> molecular orientation, and comparison<br />

was made with the results <strong>of</strong> more conventional optical dichroism<br />

measurements performed on the same samples. We demonstrated<br />

that the molecular orientation can be more directly<br />

characterized by <strong>Raman</strong> anisotropy, and that <strong>Raman</strong> anisotropy<br />

can also have a greater sensitivity to the degree <strong>of</strong> molecular<br />

orientation for highly oriented films. The relation between<br />

<strong>Raman</strong> anisotropy and optical dichroism was analyzed on the<br />

assumption that a highly uniaxial <strong>Raman</strong> tensor exists in the<br />

plane <strong>of</strong> the film. This condition is met for the in-plane ring<br />

stretching modes at ~ 1605 cm ±1 in PFO and F8BT, and BT<br />

stretching mode at ~ 1545 cm ±1 in F8BT. Using <strong>Raman</strong> anisotropy<br />

to determine the Hermans' orientation function, f, it was<br />

then possible to analyze the optical dichroism data to yield a<br />

value for b, the average angle that the optical transition dipole<br />

moment makes with the polymer chain axis. Values <strong>of</strong> b = 26.5<br />

and 22 were deduced for PFO and F8BT, respectively. The<br />

<strong>of</strong>f-axis orientation <strong>of</strong> the transition dipole moment is in accord<br />

with previous inferences [1,5,23] and limits the maximum dichroic<br />

ratio that can be expected. This result confirms the need for<br />

alternative approaches to maximize emission anisotropy. [5,9,23]<br />

Finally, the fact that <strong>Raman</strong> anisotropy data can be collected<br />

in-situ, in reflection geometry for standard device structures,<br />

and with microscopic resolution and chemical specificity makes<br />

the technique very attractive as a non-invasive device probe.<br />

4. Experimental<br />

The films <strong>of</strong> PFO and F8BT were prepared on rubbed polyimide alignment<br />

layers on Spectrosil substrates as previously described [1,20]. The conjugated<br />

polymers were spin coated onto the polyimide, and then annealed at a temperature<br />

in their nematic liquid-crystal (LC) phases to induce orientation. The polyimide<br />

layers were prepared by spin coating a polyamic acid precursor (Merck<br />

Liquicoat PI ZLI 2650) from a 30 g L ±1 solution. The precursor was baked to<br />

form an insoluble polyimide layer, and then rubbed using a machine with a rotating<br />

velvet-coated drum. This mechanical rubbing induces the requisite preferential<br />

orientation direction. PFO and F8BT were subsequently spin-coated on top<br />

<strong>of</strong> the alignment layer and heated into their nematic phase on a Linkham Microscope<br />

Hotstage purged with nitrogen (to avoid oxidation and degradation). PFO<br />

Adv. Funct. Mater. 2003, 13, No. 1, January Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1616-301X/03/0101-0071 $ 17.50+.50/0 71


H.-M. Liem et al./<strong>Molecular</strong> Orientation in Conjugated Polymer Films<br />

FULL PAPER<br />

was heated to 200 C and slowly cooled to 170 C at a rate <strong>of</strong> 1 C min ±1 . F8BT<br />

was heated to 265 C and slowly cooled to 235 C at a rate <strong>of</strong> 1 C min ±1 . Both<br />

were then quenched to room temperature to produce aligned nematic-glass films.<br />

<strong>Raman</strong> microscopy was performed using a Renishaw 2000 CCD spectrometer<br />

coupled to an Olympus BH-2 confocal microscope. The 633 nm line <strong>of</strong> a He±Ne<br />

laser was used as the excitation source to avoid absorption and resonance effects.<br />

A ”50 objective was used with a laser power <strong>of</strong> approximately 1 mW on the sample.<br />

Integration times <strong>of</strong> 60 s were used for each measurement. Polarized absorption<br />

spectra were obtained using a Unicam UV-vis spectrophotometer equipped<br />

with a Glan-Thompson polarizer oriented either parallel or perpendicular to the<br />

alignment direction.<br />

Received: May 17, 2002<br />

Final version: September 2, 2002<br />

±<br />

[1] M. Grell, D. D. C. Bradley, M. Inbasekaran E. P. Woo, Adv. Mater. 1997, 9,<br />

798.<br />

[2] C. Weder, C. Sarwa, A. Motali, C. Bastiaansen, P. Smith, Science 1998, 279,<br />

835.<br />

[3] M. Grell, W. Knoll, D. Lupo, A. Mesisel, T. Miteva, D. Neher, H. G. Noth<strong>of</strong>er,<br />

U. Scherf, A. Yasuda, Adv. Mater. 1999, 11, 671.<br />

[4] M. Grell, D. D. C. Bradley. Adv. Mater. 1999, 11, 895.<br />

[5] K. S. Whitehead, M. Grell, D. D. C. Bradley, M Jandke, P. Strohriegl, Appl.<br />

Phys. Lett. 2000, 76, 2946.<br />

[6] T. Miteva, A. Meisel, W. Knoll, H. G. Noth<strong>of</strong>er, U. Scherf, D. C. Müler,<br />

K. Meerholz, A. Yasuda, D. Neher, Adv. Mater. 2001, 13, 565.<br />

[7] M. Redecker, D. D. C. Bradley, M. Inbasekaran, E. P. Woo, Appl. Phys.<br />

Lett. 1999, 74, 1400.<br />

[8] H. Sirringhaus, R. J. Wilson, R. H. Friend, M. Inbasekaran, E. P. Woo,<br />

M. Grell, D. D. C. Bradley, Appl. Phys. Lett. 2000, 77, 406.<br />

[9] X. Long, M. Grell, A. Malinowski, D. D. C. Bradley, M. Inbasekaran, E. P.<br />

Woo, Opt. Mater. 1998, 9,70.<br />

[10] T. Virgili, D. G. Lidzey, M. Grell, S. Walker, A. Asimakis, D. D. C. Bradley,<br />

Chem. Phys. Lett. 2001, 341,219.<br />

[11] T. Virgili, D. G. Lidzey, M. Grell, D. D. C. Bradley, S. Stagira, M. Zavelani-<br />

Rossi, S. De Silvestri, Appl. Phys. Lett. 2002, 80, 4088.<br />

[12] J. J. M. Halls, K. Pichler, R. H. Friend, S. C. Moratti, A. B. Holmes, Appl.<br />

Phys. Lett. 1996, 68, 3120.<br />

[13] D. D. C. Bradley, R. H. Friend, H. Lindenberger, S. Roth, Polymer 1986,<br />

27, 1709.<br />

[14] T. W. Hagler, K. Pakbaz, K. F. Voss, A. J. Heeger, Phys. Rev. B 1991, 44,<br />

8652.<br />

[15] M. Jandke, P. Strohriegl, J. Gmeiner, W. Brütting, M. Schwoerer, Adv.<br />

Mater. 1999, 11, 1518.<br />

[16] M. Hamaguchi, K. Yoshino Polym. Adv. Technol. 1997, 8, 399.<br />

[17] A. Bolognesi, C. Botta, M. Martinelli, W. Porzio, Org. Electron. 2000, 1, 27.<br />

[18] M. Grell, M. Redecker, K. Whitehead, D. D. C. Bradley, M. Inbasekaran,<br />

E. P. Woo, Liq. Cryst. 1999, 26, 1403.<br />

[19] D. Sainova, A. Zen, H. G. Noth<strong>of</strong>er, U. Asawapirom, U. Scherf, R. Hagen,<br />

T. Bieringer, S. Kostromine, D. Neher, Adv. Funct. Mater. 2002, 12,49.<br />

[20] D. Neher, Macromol. Rapid Commun. 2001, 22, 1366.<br />

[21] B. Schartel, V. Wachtendorf, M. Grell, D. D. C. Bradley, M. Hennecke,<br />

Phys. Rev. B 1999, 60, 277.<br />

[22] J. Michl, E. W. Thulstrup, Spectroscopy with Polarized Light, VCH, New<br />

York 1996.<br />

[23] M. Grell, D. D. C. Bradley, X. Long, T. Chamberlain, M. Inbasekaran, E. P.<br />

Woo, M. Soliman, Acta Polym. 1998, 49, 439.<br />

[24] A. J. Cadby, P. A. Lane, H. Mellor, S. J. Martin, M. Grell, C. Giebeler,<br />

D. D. C. Bradley, M. Wohlgenannt, C. <strong>An</strong>, Z. V. Vardeny, Phys. Rev. B<br />

2000, 62, 15 604.<br />

[25] H. Liem, P. Etchegoin, D. D. C. Bradley, Phys. Rev. B 2001, 64, 144 209.<br />

[26] R. D. B. Fraser, J. Chem. Phys. 1953, 21, 1511.<br />

[27] J. J. Hermans, P. H. Hermans, D. Vermaas, A. Weidinger, Recl. Trav. Chim.<br />

Pays-Bas 1946, 65, 427.<br />

[28] B. Jasse, J. L. Koenig, J. Macromol. Sci., Rev. Macromol. Chem. 1979, C17,<br />

61.<br />

[29] H. Liem, P. Etchegoin, K. S. Whitehead, D. D. C. Bradley, J. Appl. Phys.<br />

2002, 92, 1154.<br />

[30] a) E. Faulques, E. Rzepka, S. Lefrant, E. Mulazzi, G. P. Brivio, G. Leising,<br />

Phys. Rev. B. 1986, 33, 8622. b) G. Masetti, E. Campani, G. Gorini, R. Tubino,<br />

P. Paiggio, G. Dellepiane, Chem. Phys. 1986, 108, 141.<br />

[31] a) E. Perrin, E. Faulques, S. Lefrant, E. Mulazzi, G. Leising, Phys. Rev. B.<br />

1988, 38, 10 645. b) G. Lanzani, S. Luzzati, R. Tubino, G. Dellepiane,<br />

J. Chem. Phys. 1989, 91, 732.<br />

[32] L. Margulies, M. Stockburger, J. <strong>Raman</strong> Spectrosc. 1979, 8, 26.<br />

[33] F. LagugnØ Labarthet, T. Buffeteau, C. Sourisseau, J. Phys. Chem. B 1998,<br />

102, 5754.<br />

[34] a) Y. Liao, N. A. Clark, P. S. Pershan, Phys. Rev. Lett. 1973, 30, 639. b) S. Jen,<br />

N. A. Clark, P. S. Pershan, E. B. Priestly, J. Chem. Phys. 1977, 66, 4635.<br />

______________________<br />

72 Ó 2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 1616-301X/03/0101-0072 $ 17.50+.50/0 Adv. Funct. Mater. 2003, 13, No. 1, January

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!