• colloidal inclusions
• liquid crystals
• photoisomerization
• elasticity
• topological defects

Liquid crystals (LCs) are self-organized mesomorphic materials that exhibit various symmetries and structures (1). They are widely used in flat panel displays for their exceptional electrooptical properties and a combination of orientational elasticity and fluidity. For example, nematic LCs (NLCs) are distinguished by their long-range orientational order, which favors alignment of the molecules (mesogens) in a preferred direction denoted as the director n. An exceptional feature of NLCs is that, despite their fluidity, they exhibit anisotropic optical and mechanical properties, and thus can transmit mechanical torque because of directional elasticity (1). Such torque occurs in response to deformations away from a uniform equilibrium state.

Such unique features of LCs can be exploited for designing smart multifunctional materials. Among these materials, colloidal dispersions of microparticles and nanoparticles in LCs have been actively studied in research on soft-matter physics (2⇓⇓⇓⇓–7). Tunable anisotropic interactions between microparticles dispersed in LCs give rise to self-assembled 1D and 2D colloidal structures (6, 8, 9). Such colloidal dispersions are interesting not only from a fundamental point of view but also from a technological one. A wide range of self-assembled structures of particles and topological defects stabilized by LC-mediated interactions find numerous applications in designing metamaterials (10), photonic devices (5, 11), sensors (12), and microrheology (5, 10⇓–12). Here, we demonstrate a phenomenon that can be used in intelligent devices using colloidal dispersions: controlled light-driven translational and rotational motions of microrods in a NLC matrix.

The orientation of LC molecules at an interface is governed by anchoring conditions, i.e., whether the director is perpendicular (homeotropic) or parallel (planar) to the interface. The orientation of the director at surfaces can be controlled through interfacial energy, anisotropy of the surface tension, and surface topography, by the pretreatment of the surfaces using surface agents, such as polymers and surfactants, together with mechanical or optical treatments. In most of the previous experiments, the solid interface was fixed. Here we use a so-called command surface (13) for LC alignment and its real-time control; the director manipulation is achieved by manipulating the anchoring condition through a light-induced isomerization of a photoactive azo-dendrimer adsorbed at the surface of microrods. As the light-induced isomerization takes place, the anchoring conditions provided by the cis isomer are different from those of the trans isomer. The initial equilibrium state of the director is lost. An important point here is that the bending direction in the cis form is not random but determined by the distorted local director near the surfaces. The torque on the particle exerted by the liquid crystal director reorientation results in specific particle motions toward a new equilibrium state. Such molecular-assisted manipulation of particles provides a tool for studying interfacial effects in their interplay with the topology of the nematic director field, which is a key concept for smart microdevices. Development of such devices requires better understanding of the photoisomerization. Some studies of azobenzenes bound to a dendrimer core have been conducted and reported in the literature (14, 15). The photoisomerization mechanism, the dependence of the quantum yield on the phenyl ring substituents, the solvent properties, and the irradiation wavelength, are still not fully understood (16). The molecular groups attached to the dendrimer core with highly regular branching do not entangle as in the case of conventional macromolecules and seem to express similar properties to the original azobenzene chromophores. Thus, this provides another way to study isomerization and photochemical properties of azobenzene molecules.

Active steering of particles in LC hosts is usually achieved by convection mechanisms, such as electrophoresis, or using high-power laser radiation in optical tweezers (17). Manipulation of the mesogen orientations is another way to control colloidal particles in an LC host. This approach mimics molecular motors, which use conformational changes of molecules (18). Several types of artificial molecular motors and actuators were designed whereby the energy of light is transformed into the mechanical energy (19⇓–21). In the case of LCs, a change of the anchoring conditions can be achieved by photosensitive functionalizations of the surface with mesogen-like moieties connected via light-sensitive azo linkages. Yamamoto et al. used a photosensitive surfactant and succeeded in manipulating the anchoring condition of colloidal particles in NLCs (22). There is, of course, a large number of papers dealing with light-driven motion and deformation in (soft) solids, e.g., liquid crystal elastomers that make use of cis−trans isomerization (for instance, refs. 23⇓⇓⇓⇓⇓–29).

Yonetake et al. synthesized a poly (propyleneimine) liquid crystalline dendrimer, which spontaneously adsorbs at LC−glass interfaces and favors homeotropic alignment of NLC molecules (30). This material was used to fabricate in-plane-switching-mode LC displays without pretreatment of substrate surfaces (31). Li et al. recently synthesized a photosensitive dendrimer with azo linkages (azo-dendrimer) in the mesogenic end chains (Fig. S1) and demonstrated the adsorption of the dendrimer at a glass interface (32). By UV irradiation, an orientational change of the mesogenic moieties occurs associated with the trans−cis photoisomerization, as illustrated in Fig. 1 A and B. Such asymmetric adsorption of dendrimers at surfaces has already been proved with a surface second-harmonic generation experiment (33). Hence, the director field was distorted under suitable illumination. The azo-dendrimer has already been used for controlling ordering transitions in mobile LC microdroplets in a polymer matrix (34) and defect structures in microparticles in LCs (35). In contrast, the situation is different in our study: Because the embedded particles are mobile and anisometric (rod-shaped), light irradiation brings about a dynamic motion of the enclosed colloids. We exploit the spontaneous adsorption of the photoactive dendrimer on various interfaces to functionalize the surfaces of rod-shaped microparticles suspended in a nematic host. As a result, we not only can control the molecular orientation at the immobile rods by light but can also mechanically rotate and translate them.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Conformations of the photoactive dendrimer: (A) trans-conformation favoring homeotropic alignment of the mesogens and (B) cis-conformation favoring planar alignment obtained under UV light irradiation.

Now we describe our present experimental results: All of the experiments were performed using 4′-n-pentyl-4-cyanobiphenyl liquid crystal (5CB) mixed with 0.1 wt% azo-dendrimer and glass rods of 10- to 20-µm length and 1.5-µm diameter. For details, the reader is referred to Experimental Procedures. First, we describe the photo-induced director field change around immobile microrods. The changes of the director configuration can be easily studied when the rods are immobile, i.e., attached to the glass substrate. The case of cells with homeotropic anchoring is described in Fig. S2. The situation is more complicated in cells with a planar anchoring condition. Several configurations are possible: rods aligned parallel (i), perpendicular (ii), and diagonal (iii) to the rubbing direction (Fig. S3). Experimentally, many rods are initially aligned at an angle Θ of 60°–70° to the director n, although normal anchoring conditions seem to favor Θ = 90°. This discrepancy may be attributed to an asymmetry of the director field at the ends of the rod and in the vertical dimension, and the different energies of the associated defect lines. Two kinds of director field configurations were found: dipolar with a single hyperbolic defect (Fig. 2A) in case i and quadrupolar with a disclination loop surrounding the rod (Fig. 2C) in case iii. These configurations have been established earlier in thin 2-μm cells by Tkalec et al. (6). In our case, to study mobile rods, thicker cells are preferred, which, however, makes the optical characterization difficult. Without UV irradiation, the director is normal to the surface of the rod (Fig. 2 A and C). This corresponds to a radial hedgehog configuration of the director field with the effective topological strength +1. To comply with the homogeneous far-field director outside the rod, a disclination loop is expected to encircle the rod in case iii. Evidence of that is shown in Fig. 2 E and F, and the structure of the disclination loop is schematically sketched in Fig. 2C. The loop is attracted to the diagonal edges of the rod to relive the mechanical strain on the director field.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Immobilized microrods attached to a surface in a planarly aligned LC cell. The rubbing direction is the horizontal direction of the images. (A) A rod lying along the rubbing direction without UV irradiation. The director map with dipolar point defects is shown, together with a microscope image in the Inset. With the inserted λ wave plate diagonal to the crossed polarizers, different birefringence colors are seen at both sides along the rod, and they are consistent with the director map shown. (B) The same rod under UV irradiation. In the vicinity of the particle, each of the director fields in A and B is mirror symmetric in the cell plane and, in first approximation, axially symmetric about the rod axis. (C) A rod lying approximately diagonal to the rubbing direction without UV irradiation. The director map with quadrupolar point defects is shown, together with a microscope image in the Inset. (D) The same rod under UV irradiation. Both director fields in C and D have C₂ symmetry about the cell normal in the cell midplane. The disclination loop shown in C, that is present at normal anchoring, vanishes under UV irradiation when the director anchors tangentially, as shown in D. It leaves two point defects. The director orientation and its change by UV irradiation are clearly visible in the images with a wave plate (see Insets). (E) Image of a rod between crossed polarizers without UV irradiation. (F) A bright field image of the same rod without polarizers. The arrows indicate the position of topological defects at the ends of the rod. The cell thickness is 10 μm.

Under UV irradiation, the director configuration changes: The normal configuration of the director transforms into a tangential one. This is evident from the changes of the interference colors to the complementary ones observed with a wave plate (see schematics in Fig. 2 A−D, and microscope images in Fig. 2 A−D, Insets). The loop collapses and transforms into two separated surface boojums with1/2 topological charge on the surface of the rod to satisfy planar anchoring conditions. This motion results in a pair of defects attached to the corners (see Fig. 2F). Disclination loops (Saturn ring) were confirmed by Tkalec et al. (6) for rods aligned perpendicular to the rubbing direction (case ii). Those loops were also tilted with respect to the rods axes.

Free rods exhibit opto-mechanical responses: (i) rotation of rods in the cell plane, (ii) rotation in the vertical plane, and (iii) translation in the cell plane. In the first case, the rods initially appear at an angle of ∼70° to the nematic director (Fig. 3 A and B). Under applied UV irradiation, the rods align nearly along the director (Fig. 3 A and C). In the second case (Fig. S4), rotation occurs about an axis perpendicular to the rubbing direction and the cell normal. In both cases, the rotations are fully reversible. The rods return to the original state when UV irradiation is removed (Fig. 3A). The angular variation and the switching rate depend on the intensity of the UV irradiation (Fig. 3F). The rotational motion involves several stages. In the first stage, the reorientation of the dendrimer moieties takes place resulting in a change of the anchoring condition from nearly orthogonal to nearly planar. The next (fast, characteristic time scale is 0.01–0.05 s) stage is accompanied by a rearrangement of the topological defects and a continuous reorientation of the director field. This triggers the last (slow, characteristic time scale is 0.1–2 s) stage: motion of the rod. The time dependence of the rod reorientation is shown in Fig. 3F. The solid curve is a best fit to one of the experimental data based on theoretical consideration (SI Text). Even at low light intensities, the rotation occurs, but the rotation angle is reduced (Fig. 3G). Both angular variations and the switching rates show a saturation-type dependence on the UV intensity (Fig. 3 G and H), which is also explained theoretically (Fig. S5B).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Dynamic motion of rods under the action of UV irradiation. (A) Time dependence of the angle between the rod and the rubbing direction. (B) Image of the initial state of a rod without UV, and (C) the final state under UV irradiation. The length of the rod is 20 μm. The rubbing direction is marked by a white arrow (parallel to n). B and C reveal the rotational motion. (D) Image of a rod without UV irradiation. (E) Image of a rod under UV irradiation at t_exp = 43 s. D and E reveal the translational motion. A white arrow marks the rubbing direction. A dashed line indicates the initial position of the rod. Images D and E were taken between polarizers parallel to the sides of the frame. The length of the rods in D and E was 15 μm. (F) Time dependence of the rod inclination at various intensities of the UV light. A solid line is the theoretical fit of the experimental data (see SI Text for details). (G and H) UV intensity dependences of the deflection angle and the switching rates, respectively. The saturation behaviors are theoretically explained (compare H and Fig. S5B).

Translational motion was observed in some rods, and the direction was more or less parallel to the long axis of the rods. Fig. 3 D and E shows the initial and intermediate stages during UV irradiation. The translational movement was extended to displacements of the order of one rod length. We could observe a slight backlash motion when the UV irradiation was stopped.

Since UV irradiation affects the conformational state of the dendrimer adsorbed at the rod surface, it is reasonable to consider that the director reorientation direction is directly related to the cis conformation. Straight (trans) to bent (cis) conformational change occurs upon photoisomerization and results in a change of the anchoring condition. The rotational motion of the rods originates from the torques exerted by the director. How do those torques develop? The initial state depicted in Fig. 3B is stabilized by the director deformation and the configuration of the topological defects around the rod as well as the anchoring on the rod surface. The torque by the director in the volume is counteracted by the (anchoring) torque at the rod/LC boundary. Under UV light irradiation, the orthogonal anchoring condition changes to the planar one and the torque in the volume is not in balance with the boundary any more. This leads to the rotation of the rod to the new equilibrium state. In the first approximation, one may describe this anchoring by a potential as Wasin2(θ), where θ is the polar anchoring angle with respect to the surface, and Wa is the strength of the anchoring energy (1). The anchoring energy is determined by the ratio of cis and trans isomers at a given light intensity and spectral composition. Planar anchoring corresponds to positive Wa, while the orthogonal anchoring is achieved with Wa<0. Strong anchoring is given if the penetration length ξ=|K/Wa| is short respective to typical geometrical sizes of the experiment; for weak anchoring, ξ is comparable or larger than system sizes. Obviously, strong illumination (large fraction of cis molecules) will correspond to large positive Wa, while no illumination (mainly trans molecules) corresponds to large negative Wa. The experiments suggest a monotonous functional form of the transition between both states with increasing/decreasing illumination intensity. One can demonstrate that intermediate stationary Wa situations can be reached at certain moderate UV intensities. Then, the director field is not influenced by the rod; the microscopy image has a uniform color around the inclusions. This corresponds to an infinite ξ. These arguments enable us to roughly estimate the switching rates and qualitatively describe their dependence on the light intensity, as described in SI Text. At strong anchoring, the torque acting on the director is proportional to KL, where L is a characteristic length (of the order of magnitude of the rod length) and K is the mean Frank elastic constant. The viscous torque is proportional to ηL3, where η is a mean viscosity of the liquid crystal. This results in a switching time τ=ηL2/K≈0.5  s. In case of weak anchoring, the switching rate Γ depends on the light intensity through the penetration length ξ : Γ=K/[ηL(L+ξ)]. Since short ξ correspond to high UV intensities, whereas low UV intensities yield long ξ, the switching rate increases with UV intensity and is expected to reach some saturation Γ = 1/τ. A qualitative agreement of the estimated switching rate with the experiment can be found by comparing Fig. 3H and Fig. S5B.

How is the reorientation direction chosen? For rods with a surrounding director field similar to that shown in Fig. 2A, the director has an opposite sense of bending on two sides of the rod, as recognized by different birefringence colors under a wave plate. UV light irradiation under this condition induces the opposite director rotation at both sides of the rod (Fig. 4A), leading to lateral translational motion. In contrast, for rods similar to those shown in Fig. 2C, the director bending direction is the same at both sides (Fig. 4B), as shown by the same birefringence color. Hence, the same director rotation at both sides of the rod upon UV light irradiation exerts torques with the same sign, causing the rod to reorient (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Schematic illustrations of the microrod motions and the acting torques. (A) Translational and (B) rotational motions. Azo-dendrimers and their bending direction are marked by brown lines and arrows, respectively. Red arrows show the director rotation from 1 (normal to the rod) to 4 (parallel to the rod) and the torque direction. Green arrows designate the rod motion.

Interactions between microparticles dispersed in an LC matrix lead to a formation of complex ordered structures. Such agglomerates of particles may have a form of chains or even more intricate structures. In this case, optomechanical effect manifests in the motion of either the whole agglomerate or its separate parts. Different examples of such structures are shown in Fig. 5.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Switchable agglomerates of particles stabilized by the director-mediated interactions: doublets of rods (crossed polarizers with a wave-plate) without UV (A) and under UV irradiation (B). There is no rotation of the doublet since the mechanical torques on two arms cancel out. The switching of the director is clearly seen from the interference colors. A dumbbell of a rod with two beads without UV (C) and under UV irradiation (D) in unpolarized light. The director orients along the vertical direction in the images.

In conclusion, we demonstrated the optomechanical effect of light-induced rotation and translation of micrometer-sized rod particles in a nematic host. This system represents an optically driven molecular microactuator, which exploits molecular reorientation on a particle surface and transforms it into a mechanical torque.