Ferroelectric liquid crystals (FLC’s) have been a subject ofintensive research because of their microsecond-order re-sponses in electro-optic switching as well as of interest intheir fundamental physical properties.1,2 In 1992 Mac-Donald et al.3 reported optically induced rotation of thedirector axis in a surface-stabilized FLC (SSFLC). Theyattributed a tilt angle change to heating of the SSFLCthat was caused by irradiation with a 514.5-nm pump la-ser. Since the tilt angle is an order parameter in thephenomenological Landau theory of the phase transition,a temperature rise of the SSFLC leads to a change in itstilt angle. However, high intensities of the order of 103
W/cm2 were required for such a change because the lightabsorption that occurred at thin indium-tin-oxide (ITO)electrodes having a thickness of about 150 nm was ineffi-cient. Despite the suggestion of MacDonald et al. thatabsorption layers or dyes be added for more efficient opti-cal switching, no experimental investigation of such ther-mal nonlinearities in dye-doped SSFLC’s has been re-ported so far, to our knowledge. In this paper we presentan experimental study of the light-induced director axisreorientation in dye-doped SSFLC’s by means of a two-wavelength pump–probe setup, which was employed pre-viously by MacDonald et al.3 A large improvement insensitivity and response speed of the light-induced direc-tor axis reorientation, as compared with undopedSSFLC’s, is obtained. In addition, self-aligned spatial fil-tering by use of the dye-doped SSFLC is demonstratedand its principle is also discussed.
2. EXPERIMENTIn the experiment a smectic-C* FLC (CS-1014, fromChisso) was used. Its phase transition temperatures areK, 221 °C; SmC* , 55.2 °C; SmA, 69.1 °C; N* , 80.8 °C; I,where K, SmC* , SmA, N* and I stand for crystalline, chi-ral smectic-C, smectic-A, chiral nematic and isotropic
0740-3224/96/0901916-05$10.00
phases, respectively. Dichroic dyes (LSR-405, from Mit-subishi Petrochemical) having an absorption peak at 518nm were used as guest dopants for a 514.5-nm pump ex-citation. They were deposited into a 2-mm gap cell con-sisting of two glass substrates coated with ITO electrodes('20 nm in thickness and 10 mm 3 10 mm in size) andunidirectionally rubbed polymer layers for a homoge-neous alignment. Four types of SSFLC sample cell con-taining either undoped or dye-doped (3, 5, and 10 wt. %)FLC’s in a planar oriented bookshelf configuration2,3 (i.e.,the preferred n-director axis of FLC’s and the smecticlayer planes are parallel to and perpendicular to the glasssubstrates, respectively) were prepared. A tilt angle U0between the director axis and the normal to the smecticlayer planes was measured to be ;17° at the sample tem-perature of 27 °C, and its sign could be reversed by an ap-plication of a small bias voltage across the ITO electrodes.Figure 1 illustrates the absorptance for the undoped andthe doped samples, which is defined as 2log10 T (where Tis the transmissivity). It is shown that although thelight absorption is quite low at the probe wavelength of632.8 nm it is enhanced at 514.5 nm by incorporation ofdichroic dyes into the SSFLC cells. The low-intensity ab-sorptance is ;5 times larger for the light polarization par-allel to the director axis than for the orthogonal polariza-tion (see the inset in Fig. 1). This indicates that thepreferred orientation of the elongated dye molecules hav-ing a strong absorption anisotropy is typically chosen par-allel to the director axis. However, the absorptance forthe parallel polarization decreases with an increase of theinput intensity. This is mainly because the dye mol-ecules follow the director reorientation caused by laserheating generated at the dye molecules. In addition, weconfirmed from a separate measurement using the samedye in a solvent that the decrease of the absorptance withthe input intensities higher than 40 W/cm2 was alsocaused by the absorption saturation of the dye molecules.The sample cell was mounted on a temperature-controllable stage and placed between a crossedpolarizer–analyzer combination for the probe transmit-
T. Inoue and Y. Tomita Vol. 13, No. 9 /September 1996 /J. Opt. Soc. Am. B 1917
tance measurement. The pump beam was polarized par-allel to the director axis for efficient light absorption asmentioned above and was focused onto the sample surfacewith a diameter of about 50 mm measured at a 1/e2 value.Temperature dependencies of U0 and a light-induced
tilt angle change DU (<0) were determined by measure-ment of the probe transmittance Tprobe given by
Tprobe 5 sin2S pDndl D sin2@2~U0 1 DU 2 f0!#, (1)
where Dn 5 n i 2 n' (50.15) is the birefringence of theFLC, l is the probe wavelength in vacuum, d is thesample thickness, and f0 is an angle between the probepolarization direction and the normal to the smectic lay-ers. It was found that U0 could be approximately scaledas (Tc 2 T)b, where Tc is the SmC* –SmA phase-transition temperature and b is the critical exponent.The estimated values for b were 0.39, 0.37, 0.39, and 0.43for the undoped and the three doped (3, 5, and 10 wt. %)samples, respectively. Although the Landau theory forthe second-order phase transition, which would be appli-cable to the SmC* –SmA phase transition in FLC’s,4 gives0.5 for b, most of the reported measurements (includingours) have given its value as between 0.3 and 0.4.3,5 Thelight-induced change in DU can be explained by consider-ing the temperature dependence of U0 described aboveand a linear dependence of a temperature rise on thepump intensity, as is shown below.Figure 2(a) shows the steady-state Tprobe as a function
of pump intensity at the sample temperature of 29.7 °C.In this measurement we set f0 5 U0 so that Tprobe waszero with the pump beam absent. It is seen that Tprobefor each doped sample starts increasing significantlywhen the pump intensity exceeds several tens of wattsper square centimeters (which corresponds to a pumppower of the order of a milliwatt). Henceforth we refer tosuch intensity levels as switch-on intensities. We find in
Fig. 1. Absorptance for unpolarized light. The inset illustratesthe absorptance for the doped (5 wt. %) sample as a function ofthe input intensity at 514.5 nm for the light polarizations paral-lel to and orthogonal to the (dark state) director axis.
Fig. 2(a) that the switch-on intensities for the dopedsamples are more than 2 orders of magnitude lower thanthat for the undoped sample. It should be noted that al-though the change of Tprobe is the result of a thermal ef-fect Tprobe is a rather nonlinear function of the pump in-tensity, as is observed in Fig. 2(a). This condition occursbecause, in addition to a sine-squared dependence ofTprobe on DU [see Eq. (1)], the light-induced tilt anglechange is also a nonlinear function of the pump intensityas confirmed from the temperature dependence of U0 .The solid curves are the least-squares fit to the data withEq. (1) and the pump-intensity dependence of DU as men-tioned above. They are in good agreement with the data,
Fig. 2. Pump intensity dependencies of (a) the steady-stateprobe transmittance, (b) the rise time, and (c) the decay time atthe sample temperature of 29.7 °C. In (b) [(c)] the rise (decay)time is defined as the time when the probe signal reaches a1 2 1/e(1/e) value with respect to the steady-state (the initial)value.
Fig. 3. Experimental setup for the self-aligned spatial filtering,where f denotes a focal length.
1918 J. Opt. Soc. Am. B/Vol. 13, No. 9 /September 1996 T. Inoue and Y. Tomita
Fig. 4. Experimental result of the self-aligned spatial filtering. The sample temperature of 37.5 °C was chosen so that U0 was ;14°.(a) Output image without an analyzer; (b) intensity inversion (cP 5 10.9°, cA 5 8.1°); (c) high-pass filtering (cP 5 2.4°, cA 5 6.0°); (d)low-pass filtering (cP 5 11.2°, cA 5 33.1°). Input powers were (b) 10 mW, (c) 3 mW, and (d) 3 mW.
indicating that the observed transmittance changes re-sult from the light-induced director axis reorientation.Departures from the fitting curves at high intensities canbe attributed to the decreased light absorption with theincrease of the pump intensity as shown in Fig. 1. Wealso found that the difference in the switch-on intensitiesbetween the undoped and the doped samples was morethan 3 orders of magnitude at the sample temperature of45.2 °C. This is because the temperature dependence ofthe tilt angle change becomes stronger as the sample tem-perature is closer to Tc .Rise and decay times of the light-induced director axis
reorientation were measured under pulsed illumination
of the cw pump beam. It is seen in Fig. 2(b) that the risetimes for the doped samples are shortened by more than 2orders of magnitude as compared with that for the un-doped sample. Apparent peaks of the rise times as afunction of the pump intensity for the doped samples mayresult from the critical slowing-down phenomenon nearTc in the heat-flow dynamics between the two light-absorbing ITO layers. Similar behavior was also ob-served in the experiment of MacDonald et al.3 However,the decay times for the doped samples, which are of theorder of submilliseconds to milliseconds, exhibit dips atthe same pump intensities as those for the peaks of therise times. These dips must also be related to the heat-
T. Inoue and Y. Tomita Vol. 13, No. 9 /September 1996 /J. Opt. Soc. Am. B 1919
flow dynamics in the SSFLC layer, since the perfect heatconduction at the two ITO layers would not cause suchanomalous dips.3 However, the observation that the de-cay times are shorter than the rise times at a given pumpintensity may be attributed to the relaxation process de-termined by better heat conduction at the two ITO layers.Further studies are necessary to explain such decay be-havior. In addition we observed a gradual buildup of theprobe signal followed by much faster signal component ofthe order of milliseconds in the undoped sample. Suchbuildup behavior caused the rise time for the undopedsample to be much slower than that for the doped sample,as seen in Fig. 2(b). Likewise, the observed decay timefor the undoped sample was approximately 1,000 timeslonger than the result of MacDonald et al.3 Although theorigin of these observations is still unknown, it must berelated to heat-conduction properties of the undopedsample including the ITO layers.
3. APPLICATION TO SELF-ALIGNEDSPATIAL FILTERINGIn this section we describe an experimental demonstra-tion of self-aligned spatial filtering using the light-induced director axis reorientation in the dye-doped (5wt. %) SSFLC sample. Both filter formation and itsreadout can be performed by the same beam in a self-aligned manner. Note that our new scheme is differentfrom the previously reported method,6 which utilizedlight-induced screening of an internal electric field in adc-biased photorefractive Bi12SiO20 , crystal. Figure 3 il-lustrates the experimental setup. An input image to befiltered was illuminated by a 514.5-nm argon-ion laserbeam having a linear polarization oriented at an angle cPwith respect to the normal to the smectic layers. Thesample was placed at the Fourier plane in a double dif-fraction imaging system. Because intensities of the zero-and lower-order spatial frequency components of the in-put image are usually much higher than those of thehigher-order components, the transmitted light polariza-tion just after the sample is spatially modulated by thelight-induced reorientation effect. An output image wasobserved at the conjugate image plane through an ana-lyzer with an angle cA . The filtering operation can bedone, in a self-aligned manner, by an appropriate combi-nation of cP and cA . This method can be qualitativelyexplained as follows: Low-pass filtering is realized withcP ' U0 and cA ' cP. These values are chosen becausethe zero- and lower-order spatial frequency componentsexperience their polarization changes after the sample be-cause of the director axis reorientation by DU, and thusthe higher-order components can be suppressed by theanalyzer. Likewise, high-pass filtering is realized withcP ' U0 1 DU and cA ' cP. In addition, intensity in-version is also realized with an appropriate combinationof cP and cA chosen so that a relative phase between thezero-order and the nonzero-order components through theanalyzer becomes 180°. Figure 4 shows the experimentalresult. All the filtering operations could be performedwith milliwatt laser powers and millisecond responsetimes. As far as the spatial resolution of the optical sys-tem used in this experiment is concerned, any noticeable
reduction in the output image was not observed with thedye-doped sample as compared with the undoped one.This is because the thickness of the sample is very thin(i.e., 2 mm), and thus the effect of lateral heat diffusion inthe sample on the spatial resolution may be neglected.To analyze the filtering results more quantitatively, we
performed the Jones matrix calculation for the complexamplitude transmittance after the analyzer as a functionof the tilt angle. In the calculation the experimentallyobtained values for cP and cA as shown in Fig. 4 wereused. The absorption anisotropy of the dye molecules,which were assumed to be aligned along the director axisof the FLC’s, was also taken into account, although therewas no qualitative difference between the calculated re-sults with and without the absorption anisotropy. Theresult is shown in Fig. 5, where U0 corresponds to thehigher-order spatial frequency region, since local tem-peratures in that region are usually much lower thanthose in the zero- and lower-order spatial frequency re-gions. A decrease of the tilt angle thus corresponds tothe zero- and lower-order spatial frequency regions. It isclear from Fig. 5(a) [Fig. 5(b)] that the amplitudes of thelower- (higher-) order components are suppressed [i.e.,high- (low-) pass filtering can be performed]. In addition,Fig. 5(c) indicates that, while the phase of the zero-ordercomponent (which may correspond to the tilt angleslightly smaller than 10° in this calculation) are 180°-phase shifted with respect to the higher-order compo-nents, their amplitudes are more or less equal. This con-dition approximately corresponds to intensity inversion,as seen in Fig. 4(b).
4. CONCLUSIONIn this paper we have experimentally investigated thesteady-state and dynamical properties of the light-induced director axis reorientation in the dye-dopedSSFLC’s. We have demonstrated several orders of mag-nitude enhancement in sensitivity and response speed ofthe light-induced director axis reorientation by incorpo-rating dichroic dyes into FLC’s. Wavelength sensitivities
Fig. 5. Calculated complex amplitude transmittance through atandem combination of the polarizer, the sample, and the ana-lyzer for (a) high-pass, (b) low-pass, and (c) intensity inversionoperations.
1920 J. Opt. Soc. Am. B/Vol. 13, No. 9 /September 1996 T. Inoue and Y. Tomita
can be adjusted by an appropriate choice of dyes. Thisenables us to use a diode laser as a pump beam. Self-aligned spatial filtering with the dye-doped SSFLC hasalso been demonstrated. A semi-quantitative explana-tion of the filtering operation has been given. Since aSSFLC cell is inherently very thin (a few micrometers)and can be fabricated to have a large area, its use inFourier-domain image processing would be most suitable.
ACKNOWLEDGMENTThe authors thank Kazuo Yoshinaga for his valuable ad-vice on sample preparation.
REFERENCES1. See, for example, I. C. Khoo and S. T. Wu, Optics and Non-
linear Optics of Liquid Crystals (World Scientific, Sin-gapore, 1993).
2. N. A. Clark and S. T. Lagerwall, ‘‘Submicrosecond bistableelectro-optic switching in liquid crystal,’’ Appl. Phys. Lett.36, 899 (1980).
3. R. MacDonald, J. Schwartz, and H. J. Eichler, ‘‘Laser-induced optical switching of a ferroelectric liquid crystal,’’Int. J. Nonlinear Opt. Phys. 1, 103 (1992).
4. R. Blinc and B. Zeks, ‘‘Dynamics of helicodal SmC* liquidcrystals,’’ Phys. Rev. A 18, 740 (1978).
5. J. Hoffmann, W. Kuczynski, and J. Malecki, ‘‘Dielectricstudy of ferroelectric properties in chiral smectic C,’’ Mol.Cryst. Liq. Cryst. 44, 287 (1978).
6. G. Mubus, B. Schmidt, and H. J. Tiziani, ‘‘Self-induced pho-torefractive spatial frequency filter,’’ J. Opt. Soc. Am. B 7,2374 (1990).
