J. MATER. CHEM., 1991, 1(3), 331-337 33 1 Dielectric Relaxation Spectroscopy and Molecular Dynamics of a Liquid-crystalline Polyacrylate containing Spiropyran Groups Ewen J. C. Kellar," Graham Williams,*" Valeri Krongauzb and Shlomo Yitzchaikb a Department of Chemistry, University College of Swansea, Singleton Park, Swansea SA2 8PP, UK Department of Structural Chemistry, The Weizmann Institute of Science, Rehovot 761000, Israel The dielectric properties of a liquid-crystalline (LC) copolymer containing photochromic spiropyran groups have been determined over wide ranges of frequency and temperature. The observed loss curves contain a component from the dipole reorientations of the mesogenic groups and a further component from a conductivity-related process. An electrical cleaning method was used to reduce the contribution from the latter process and a subtraction method was devised to separate the overall loss curves into the two components.Using electrical/ thermal treatments it was found possible to align the copolymer homeotropically but not planarly. The dielectric loss spectra for unaligned and homeotropically aligned samples were very different. The data were analysed to yield two relaxation parameters for unaligned and aligned samples and were interpreted in terms of the anisotropic motions of dipolar mesogenic groups as have been discussed previously. In addition, the temporal stability of alignment of homeotropic samples at temperatures close to the clearing temperature was determined using the dielectric method.Keywords: Liquid crystal; Dielectric relaxation; Photochromism; A.C. alignment Since the late 1970s when liquid-crystalline (LC) side-chain polymers were first synthesized, much work has been done in assessing and varying their properties with a view toward their potential applications in the field of molecular elec- tronics, e.g. as media for optical information storage, optical waveguides and non-linear optical devices. '-'One such way in which the optical and spectroscopic properties of an LC polymer can be modified is by the incorporation of dye molecules alongside the mesogenic units either physically, as in mixtures, or chemically, by including the dye within the chemical structure to form a copolymer. In the particular case studied here, the LC polymer is a random copolymer in which 20% of the mesogenic groups have been substituted by a spiropyran derivative (see Scheme 1 and Experimental later).Note that the representation of the dimer below is schematic spiropyran yellow red blue merocyanine -* merocyanine dime rised isolated 0-0-0-Scheme 1 only. A proper description would take into account its sand- wich structure in which the rings ~verlap.~ Spiropyran copolymers have been the centre of much attention for several years owing to their photochromic and thermochromic properties."-" The polymer material con- taining the spiropyran group is yellow but irradiation by UV light at low temperatures, below the polymer glass-transition temperature (Q, causes the colour of the polymer to change to deep blue.The change is indicative of a photochemical conversion of the spiropyran to merocyanine, see Scheme 1. The reaction can be reversed by irradiation with visible light. If the copolymer film containing the spiropyran group is irradiated with UV radiation at higher temperatures the compound changes colour to red, as a result of complete or partial merocyanine dimerisation. This reaction is reversible thermally or in visible light.l4-'' Merocyanine groups will also be formed owing to the thermochromic reaction and dimerise if a sample is melted, and so a colour change from yellow to red is also possible. From previous work by Wis- montski-Knittel and Krongauz" and Goldburt and Kron- gauz,13 it has been demonstrated that the merocyanine tends to aggregate further to form higher aggregates and even a semicrystalline phase.It was possible to incorporate up to 40mol% of the photochrome units and still observe the liquid-crystalline properties of the material, though the clear- ing temperature (T,) declined with increasing photochrome concentration. Using UV spectrophotometry," the local order parameter, S, for the dye was found to be ca. 0.1 even when the polymer was aligned in an electric field. Yitzchaik et al." have proposed therefore that the photochromic side groups are in some way separate from the rest of the liquid-crystalline mesophase. Apart from the work of Yitzchaik et al.,18 who demonstrated alignment of the copolymer by virtue of the sample becoming optically transparent in the presence of a d.c. electric field between interdigitated electrodes mounted on a glass plate, no work has been done to date on the precise nature and extent of alignment within the sample and the effect, if any, of the addition of the spiropyran into the system.Dielectric relaxation spectroscopy is an ideal technique for such a and is used extensively in this paper to explore further the properties of this photochromic copolymer. CH, CH, CH,=CHCONH(CH,)SCONH -No2 1 2 Experimental The copolymer studied was produced by the free-radical copolymerisation of the two monomers 1 and 2 and was synthesized by The content of the photochromic como- nomer in the polymer is ca.20%. The polymer was purified by reprecipitation from a tetrahydrofuran solution with meth- anol and drying in a vacuum.17 In order to study both the electrical properties and the optical transparency variations occurring within the copoly- mer, the LC sample was melted and sandwiched between two conducting indium tin oxide glass electrodes.A PTFE spacer was used to maintain constant thickness (80 pm) and to insulate one electrode from the other. The sample area was ca.0.5 cm2. The glass cell was placed in an insulated mount inside a watertight metal jacket and each end was connected to a metal contact so that electrical measurements across the cell could be made. Also contained within the metal housing was an optical-fibre arrangement which enabled the optical transmission characteristics of the sample to be monitored via photo diode^.^^ The output was recorded on an XY chart recorder.The whole apparatus was immersed in a water bath so that the temperature of the sample could be controlled accurately, provided sufficient time (10 min) was left for the system to equilibrate after each temperature change. The electrical connections to the cell allowed the sample to be subjected to various d.c. and a.c. fields. All dielectric experiments were made using a Genrad 1693 RLC Digibridge over the range 12-10' Hz coupled to a Hewlett-Packard computer system, to collect, store and manipulate the data thus acquired. The equivalent parallel capacitance, C,, and conductance, G,, of the sample were taken at 20 spot frequencies by a substitution method involv- ing first zeroing the Digibridge without the sample and then measuring with the sample present.A scan across the fre- quency range required ca.8 min. The values of C, and G, are related to the relative permittivity E' and loss factor E" according to the equations,28 c, =ElC, +c1 G,/w =E"C, where C, is the geometrical capacitance of the interelectrode space, C1is the fringing capacitance and w =2nfis the angular frequency of the measurement. Thus plots of C, and G,/o against frequency are sufficient to indicate the dielectric relaxation behaviour of the liquid-crystalline polymer. The use of an Olympus BHSP polarizing microscope coupled to a Linkam hot stage (THM 600) enabled accurate measurements (within 0.1 "C) to be made of the clearing point for the LC copolymer, which was found to lie between 95.7 and 95.9 "C.Results and Discussion The copolymer was studied primarily using dielectric relax- ation spectroscopy which gave useful information about the J. MATER. CHEM., 1991, VOL. I dynamic molecular motion of the molecules, especially the mesogenic side groups. It is therefore possible to assess the degree of alignment of the liquid crystal from the size and shape of the loss peak. Further information concerning opti- mum conditions for sample alignment, alignment stability and T, can also be obtained. Initial Studies and D.C. Cleaning Fig. l(a) shows the loss spectra obtained for an unaligned sample at different temperatures.Consider first the data at the lower temperatures. As the frequency decreases, the loss increases with no evidence of a maximum. As the temperature is increased, the loss values increase markedly and there is clear evidence that a loss peak is emerging from the low- frequency conductivity 'tail', until at the highest temperatures studied, the two processes can be resolved. An alternative, and informative, representation of these data is shown in Fig. 2(a), where again, the emergence of a dielectric relaxation process from the dominant conductivity tail is observed as the sample temperature is increased. The sample was subjected to repeated pumping at high temperatures in the LC state in an attempt to remove impurities, which would contribute to the low-frequency conductivity process, but this process was unaffected by such treatment.In order to reduce the conductivity process, and hence enhance the dielectric loss process at higher frequencies, we used the technique of 'electrical cleaning' as described by Osaki et ~21.~' and Cebe and Gr~bb.~' This required the sample in the melt (T>T,) to be subjected to a large d.c. voltage (ca.300 V) for a period of several hours. During this time, extraneous ions which led to the conductivity process seen in Fig. l(a) are swept to the electrodes. As time progressed the effective d.c. conductivity fell, so it was possible to raise the applied voltage to assist the cleaning process.On cooling the sample back into the LC state, with the voltage still maintained, the ions are trapped at the electrodes and their greatly reduced mobility in the LC state, compared with that in the isotropic state, means that their contribution to the low-frequency conductivity tail is reduced. In practice, as the current traversing the sample, during the cleaning process, approached very small values irrespective of the value of the applied d.c. voltage, the electrical cleaning process was deemed to be complete, so the sample was cooled to just below < (ca.45 "C) and subsequently its dielectric properties were measured at successively higher temperatures. Fig. l(b) shows data for a sample that was treated in the way just described and the change in behaviour is evident.The loss peak is now seen over a wide range of sample temperatures since the magnitude of the low-frequency con- ductivity process has been reduced significantly. Fig. 2(b) shows the three-dimensional representation of these data and comparison with Fig. 2(a) shows how the low-frequency con- ductivity process has been reduced and the loss peak enhanced. Clearly, the cleaning process is successful if the sample tem- perature is below ca. 80 "C. However, the effect becomes short- lived at higher temperatures as T, is approached, because the ions which had been swept to the electrodes diffuse back into the bulk of the material and restore the original levels of low-frequency conductivity. The other effect of the d.c.voltage is to cause partial alignment of the polymer film as the LC phase reforms on cooling from the melt. This gives rise to an increase in the height of the emerging ion peak [compare Fig. l(a), l(b) and also 2(a), 2(b)]. J. MATER. CHEM., 1991, VOL. 1 4050/ LLa.30: I3 L-I . . ..'* 1-. . .__ .~-.. 1 2 3 4 5 2 3 4 5 log ( f/Hz) log ( f/Hz) log ( f/Hz) 50 (f t 2 3 4 5 log ( f/Hz) log ( f/Hz) log ( f/Hz) Fig. 1 G/w us. log f over a range of temperatures (50-99.5 "C) for (a)uncleaned and unaligned sample, (b)d.c.-cleaned sample (300 V, 3 h, 110 "C),(c) a.c.-aligned sample (200 V, 300 Hz, 2 h, 85 "C), (d) uncleaned and unaligned sample with conductivity tail subtracted, (e)d.c.-cleaned sample (300 V, 3 h, 110 "C) with conductivity tail subtracted, (f) a.c.-aligned sample (200 V, 300 Hz, 2 h, 85 "C) with conductivity tail subtracted A.C.Alignment The sample was subjected to a field of 300 Hz, 200 V for 2 h at 85 "Cfollowed by cooling (with the field removed) to room temperature. The resulting loss spectra are shown in Fig. l(c) and 2(c). The almost two-fold increase in the height of the loss peak compared with that for the unaligned sample indicates that the sample is almost totally homeotropically (h-) aligned i.e. the long axes of the liquid crystal side groups are aligned normal to the electrode surfaces. Conductivity is still dominant but the intensity of the loss peak is such that it is not so fully obscured at low temperatures as was the case in the unaligned sample.Conductivity Subtraction A third technique to remove the low-frequency conductivity feature was employed and in many respects was the most successful for all the observed loss curves. It took advantage of the fact that at high temperatures the conductivity process dominates at low frequencies, so its functional behaviour can be estimated over the extra frequency range. Once calculated, the theoretical curve can be subtracted from the original data to reveal a residual (or 'cleaned') loss feature. Such a method required a plot of log(G/w) us. log f to be made. For a true d.c. conductivity process, the loss factor should be proportional to f-'. However, it is usually found for organic solids that an f-" law applies, and this is also the case for the present work. For the low-frequency conductivity regime log(G/u)=A -nlog f (3) We have determined values (A, n) for each sample at each temperature from the best fits to the limiting low-frequency data where the conductivity process was dominant.An esti- mate of the accuracy of fit was made by examining the residuals in the low-frequency range where the dipole loss contribution was negligible. A good fit was assumed if the first low-frequency points gave differences of <1% of the averaged value. This method allowed the conductivity tail to be subtracted from ca.40% of the loss spectra, but for the remainder it was necessary to extrapolate (A, n) values with respect to the temperature, and to see if the dielectric relax- ation peak obtained subsequently was well behaved before accepting the (A, n) values in a particular case. Fig.3 and 4 show the values of n(T)and A(T)thus determined. The values of n range from ca. 0.5 at 50 "C to ca. 1.0 at 100 "C,whereas A changes by two orders of magnitude in the same temperature range. The plot of log A us. T-' curved upwards, showing that conductivity increases more strongly with increasing temperature than expected by a simple Arrhenius relation. At higher temperatures n4 1 (a simple conductivity process), whereas n<l implies dispersive transport of ions. The loss curves for dipole relaxation were obtained by subtracting the calculated loss due to a conductivity process from the observed loss curves; the resulting data are shown in Fig.l(d)-(f) and 2(d)-( f). Comparing Fig. l(d)-( f),the preferred alignment (homeotropic) leads to an increase in the height of the loss peak. For the range 95-97.5 "C we see a marked decrease in peak height, showing that the material is transformed from h-aligned liquid crystal to isotropic liquid over a narrow temperature range. Discussion of Dielectric Relaxation Spectroscopy Results The data of Fig. l(a)-l(c) show that the dipole relaxation process is partly obscured by the low-frequency conductivity process, but they may be separated with the aid of an electrical cleaning operation and numerical subtraction of the conduc- tivity process using eqn.(3). The subtraction process is well illustrated in the comparison of the curves shown in Fig. 5 J. MATER. CHEM., 1991, VOL. 1 10 Fig. 2 Three-dimensional plot of G/o us. log f us. T for (a) uncleaned and unaligned sample, (b) d.c.-cleaned sample (300 V, 3 h, 110 "C), (c) a.c.-aligned sample (200 V, 300 Hz,2 h, 85 "C),(d) uncleaned and unaligned sample with conductivity tail subtracted, (e) d.c.-cleaned sample (300 V, 3 h, 110 "C)with conductivity tail subtracted, (f)a.c.-aligned sample (200 V, 300 Hz, 2 h, 85 "C)with conductivity tail subtracted and 6 for different samples at 85 "C. Note the d.c.-cleaned and unaligned loss curves reflect different weighted sums of sample gives a loss peak greater than the untreated unaligned the underlying dipole relaxation modes23 (see also Fig.8 later). sample, since the cleaning process induces some h-alignment. Note that an isosbestic point occurs at log f=3.9, at which The shift of the frequency of maximum loss to lower values point the loss curves all have the same value. This is predicted on partial alignment (for d.c.-cleaned sample) and fuller align- from a semi-macroscopic theory for the complex permittivity ment (for a.c.-aligned sample) is expected since the h-aligned E(o), where c0=2n$ for a uniaxial sample whose macroscopic J. MATER. CHEM., 1991, VOL. 1 c3ao 0.7 40.01" 0.61 g 0 O "p 30.0 0.5 0.4 0.0 1°.OL0.01 2 3 4 log( f1Hz) Fig. 6 Plot of G/w us. log f for uncleaned and unaligned (U),d.c.-cleaned (A)and a.c.-aligned (0)sample at 85 "C with conductivity tail removed specimens, respectively.E(O) =&'(a)-id'(a), where &''(a)is the dielectric loss factor (= G/oC,, where C, is the inter-electrode 00 8 capacitance). Thus, for the isosbestic frequency for loss factor, E",,(O)=E"~(O) and hence &"(a)at this frequency is the same for any value of sd. A similar condition applies to &'(a)but 0 the value of the isosbestic frequency at which ~',,(a)=E'&) O O is different from that for the loss factor. Fig. 7 shows the plots 0 of C, against logf, and the isosbestic frequency (cross-over -0.5 1 0 0 0 0 0 frequency) is seen to occur at ca. 700 Hz and is, coincidentally, close to the frequency of maximum loss factor f, (Fig.6). Thus, the frequency of 300Hz, chosen to align the sample homeotropically at 85 "C,was conveniently placed away from the dielectric loss maximum, which causes dielectric heating, -1 .oIand from the conductivity tail, which also causes dielectric 2.6 2.7 2.8 2.9 3.0 3.1 103~1~ Fig. 4 log A us. 1/Tfor uncleaned and unaligned 0)and a.c.-aligned sample (0) 40.0 -30.0-Q.h -20.0 10.0 -i0.0 L AL-12-Lld 1 2 3 4 5 log ( f/W Fig. 5 Plot of G/ous. log f for uncleaned and unaligned (0),d.c.-cleaned (A)and a.c.-aligned (0)sample at 85 "C heating. According to Fig. 7, it should be possible to align the sample using a directing a.c. field the frequency of which exceeds the cross-over frequency, but we did not succeed in doing so in our experiments.The plots of logf, us. T-' for the unaligned and h-aligned samples are shown in Fig. 8, and are seen to be approximately linear giving apparent activation energies of 184 kJ mol-' and 209 kJ mol-' for unaligned and h-aligned samples, respectively. In contrast, the apparent activation energy for the conductivity process, as obtained from the slope of log A 170 I 160 --150 -140 130 -120 -U ,n 110 -O" 100 -90 -80 -70 --60 alignment is intermediate between the h- and planarly (p-) 501 I "1111" I "111"1 I'I""" ' iI1'llllaligned conditions. We obtained23 log ( f/Hz) E(W)=(1 +2Sd)&li(o)/3+2(1 -Sd)&L(a)/3 (4) Fig. 7 C, us. logffor uncleaned and unaligned (U),d.c.-cleaned (A)where Sd is equal to 1.0 and -0.5 for fully h- and p-aligned and a.c.-aligned (e)sample at 85 "C 336 5 0.4 0 0 h 0' \2 3 0 .-E cn- 0 2 1 103~1T Fig. 8 logfus. 1/T for uncleaned and unaligned (0)and a.c.-aligned (0)sample us. T-' in Fig. 4, is only ca. 0.33 of that for the dipole relaxation process (Fig. 8) and this differenceenables the latter process to emerge from the conductivity tail as is shown in Fig. 1 and 2. In amorphous solid polymers the apparent activation ener-gies for the principal (a)relaxation process (segmentalmotion) and conductivity process are often found to be similar, implying that the diffusion of ions is effected via segmental motions. For the LC polymer, the dielectric relaxation process observed in Fig.1 and 2 is to be assigned to a weighted sum of relaxation modes, 00, 01, 10 and 11 (as has been described earlier in ref. 22-25, 31) in which the longitudinal dipole moment pl and transverse dipole moment pt associated with both the mesogenic groups and the photochromic groups are relaxed,z3and it is evident that these anisotropic motions and the motions of extraneous ions through the material have very different mechanisms, and hence different activation energies. We note that the half-width of the loss curve for the h-aligned sample (Fig. 6) is ca. 1.5, which is a little broader than that (1.14) for a single relaxation time process. Also the curve has a long high-frequency tail, which is indicative of the 01 process.The loss curve for the unaligned sample is very broad and asymmetrical. If we take values for ~",,(f)and &""naljgned(f) from Fig. 6, then using eqn. (4) it is possible to calculate the loss curve for a p-aligned sample. The calculated curve shows a rising loss with frequency, and passes through the isosbestic frequencyz3at ca. 9 kHz. The molecular origin of the losses for a p-aligned sample is the 10 and 11 relaxation modes as we have discussed earlier (ref. 22-25, 31). The unaligned sample gives a dielectric loss spectrum which is a weighted sum of all four relaxation modes [see eqn. (4), for sd=o]. Stability of Homeotropic Alignment Dielectric relaxation spectroscopy provides a simple and direct method for monitoring the stability of the field-induced align-ment of the polymer.The h-aligned material was prepared using 200 V at 300 Hz applied for 2 h at 85 "C, and samples were studied at different temperatures by cooling the sample quickly to the desired temperature, removing the voltage and monitoring the dielectric loss factor at a frequency close to fm at that temperature at 1 min intervals over a period of 3 h. In order to standardise the experiments, a delay of 3 min from field removal was imposed at the start of each run to take into account the time required to achieve a constant J. MATER. CHEM., 1991, VOL. 1 43 42 .. 41 40 39 38 LL a 37 \236 30 -'...... '...... ...... -..... ......... 29 -.. ........ ".'.. ........,, ...... . ......... .... ....... ......... 0 1800 3600 5400 7200 9000 10 800 tls Fig.9 Plot of G/o peak decay at different temperatures us. t: top (log f=2), 75 "C;middle (log f=2.25), 80 "C;bottom (log f= 2.Q 85 "C temperature in the cell. Fig. 9 shows our data for 85, 80 and 75 "C. As the temperature is decreased the rate of disalignment decreases. The observation that disalignment occurs in the LC phase is in contrast with our earlier observation for siloxane-chain homopolymers where no disalignment occurred below z.22-z5We have also observed disalignment in siloxane-chain LC copolymers where the longitudinally attached mesogenic groups are diluted by the presence of transversely attached mesogenic groups.32 It is evident from Fig.9 that disalignment occurs even at temperatures 20 "C below T,. Optical Studies Some optical studies were made with a sample contained between conducting glass electrodes as the temperature was varied and the directing voltage was applied. Optical trans-mission of the whole sample area was detected by a fibre-optic bundle linked to a photodetector. The optical trans-mission of the sample increased steadily for T> with a final rapid increase to transparency in the range 95-95.7 "C. On application of a directing a.c. voltage, the optical transmission increased with time but with this method it was not possible to estimate the degree of homeotropic alignment achieved as time increased.Conclusions Dielectric loss spectra for unaligned, partially h-aligned and h-aligned samples of a photochromic polymer have been obtained over a range of frequencies for different sample temperatures. The spectra have been resolved into conduc-tivity and relaxation component processes and their phenom-enological parameters have been determined over a range of temperatures and sample conditions. The resolved dipole-relaxation process is mainly due to the 6 (or 00) motional process of the mesogenic groups, but clear evidence is obtained for the presence of 01, 10 and 11 relaxation modes, which result from the anisotropic motions of the dipolar mesogenic groups. A feature of the work is that it was found possible to 'electrically clean' the samples, allowing the conductivity process to be suppressed for T< T, and the dipole relaxation process to be made more prominent.Disalignment of h-aligned samples was observed at temperatures well below T, after removing the directing voltage and monitoring the dielectric loss factor at a frequency chosen to be close to J. MATER. CHEM., 1991, VOL. 1 337 logf, at that temperature. The rate of decay increases with increasing temperature. 14 15 I. Cabrera, V. Krongauz and H. Ringsdorf, Mol. Cryst. Liq. Cryst., 1988, 155, 221. I. Cabrera and V. Krongauz, Macromolecules, 1987, 20, 2713. 16 I. Cabrera and V. Krongauz, Nature (London), 1987, 326, 582. The authors would like to thank SERC for the postdoctoral award to E.J.C.K.,which enabled much of this work to take place.Thanks are also made to Dr. M. Garley for his help in developing the program required for the three-dimensional 17 18 19 I. Cabrera, V. Krongauz and H. Ringsdorf, Angew. Chem. Znt. Ed. Engl., 1987, 26, 1178. S. Yitzchaik, I. Cabrera, F. Buchholtz and V. Krongauz, Macro-molecules, 1990, 23, 707. W. Haase and H. Pranoto, Polymeric Liquid Crystals, ed. A. plots in Fig. 2. 20 Blumstein, Plenum, New York, 1985, p. 313. W. Haase, H. Pranoto and F. J. Bormuth, Ber. Bunsenges. Phys. Chem., 1985,89, 1229. 21 F. J. Bormuth and W. Haase, Liq. Cryst., 1988, 3, 881. References 22 23 G. S. Attard and G. Williams, Liq. Cryst., 1986, 1, 253. G. S. Attard, K. Araki and G. Williams, Br. Polym. J., 1987, 19, ~1 2 H.Finkelmann and G. Rehage, Ado. Polym. Sci., 1984, 60/61, 99. V. P. Shibaev and N. A. Plate, Adv. Polym. Sci., 1984, 60/61, 24 25 119. G. S. Attard, Mol. Phys., 1986, 58, 1087. G. S. Attard, J. J. Moura-Ramos and G. Williams, J. Polym. 3 173. G. S. Attard and G. Williams, Chem. Br., 1986, 22, 919. 26 Sci., Polym. Phys. Ed., 1987, 25, 1099. G. S. Attard, K. Araki, J. J. Moura-Ramos and G. Williams, 4 5 G. S. Attard and G. Williams, Nature (London), 1987, 326, 544. M. Engel, B. Hisgen, G. Keller, W. Kreuder, B. Reck, H. Ringsdorf, H. W. Schmidt and P. Tschirner, Pure Appl. Chem., 1987, 57, 1009. 27 28 Liq. Cryst., 1988, 3, 861. K. Araki, A. Kozak, G. Williams, G. W. Gray, D. Lacey and G. Nestor, J. Chem. Soc., Faraday Trans. 2, 1988,84, 1067.N. G. McCrum, B. E. Read and G. Williams, Anelastic and 6 H. J. Coles, in Developments in Crystalline Polymers, ed. D. C. Dielectric Efects in Polymeric Solids, Wiley, London-New York, 7 Bassett, Elsevier Applied Science, Barking, 1988, vol. 2, p. 297. Side Chain Liquid Crystal Polymers, ed. C. B. McArdle, Blackie, 29 1967. S. Osaki, S. Uemura and Y. Ishida, J. Polym. Sci., Polym. Phys. Glasgow, 1989. Ed., 1971, 9, 585. 8 I. Cabrera, M. Engel and H. Ringsdorf, Extended abstract of a 30 P. Cebe and D. T. Grubb, Macromolecules, 1984, 17, 1374. paper presented at the 3rd EEC Workshop on Photochemical 31 K. Araki, G. S. Attard, A. Kozak, G. Williams, G. W. Gray, D. and Photobiological Processes for the Production of Energy Lacey and G. Nestor, J. Chem. Soc., Faraday Trans. 2, 1988,84, 9 Rich Compounds, London, April 18-2 1, 1989. The Theory of the Photographic Process, ed. H. James, McMillan, 32 1067. G. Williams, A. Nazemi, F. E. Karasz, J. S. Hill, D. Lacey and New York, 1977, ch. 7, 8. G. W. Gray, Macromolecules, in the press. 10 Y. Kalisky and D. J. Williams, Macromolecules, 1984, 17, 292. 33 G. S. Attard, K. Araki, J. J. Moura-Ramos, G. Williams, A. C. 11 F. Shvartsman and V. Krongauz, Nature (London), 1984, 309, Griffin, K. Bhatti and R. S. L. Hung, in Polymer Association 12 609. T. Wismontski-Knittel and V. Krongauz, Macromolecules, 1984, Structures, Microemulsions and Liquid Crystals, ed. M. A. El Nokaly, ACS Symp. Ser. 1989, 3484,255. 18, 2124. 13 E. Goldburt and V. Krongauz, Macromolecules, 1986, 19, 247. Paper 0/04402D; Received 1st October, 1990