Directed Self-Assembly of Colloidal Particles onto Nematic Liquid

Jun 12, 2017 - Bacteria use expanded genetic code. The genome of every cell on Earth uses four DNA bases—adenine, thymine, cytosine, and guanine—t...
0 downloads 13 Views 8MB Size
Directed Self-Assembly of Colloidal Particles onto Nematic Liquid Crystalline Defects Engineered by Chemically Patterned Surfaces Xiao Li,†,‡,# Julio C. Armas-Pérez,†,§,# Juan P. Hernández-Ortiz,†,∥ Christopher G. Arges,†,⊥ Xiaoying Liu,† José A. Martínez-González,† Leonidas E. Ocola,‡ Camille Bishop,† Helou Xie,† Juan J. de Pablo,*,†,‡ and Paul F. Nealey*,†,‡ †

Institute for Molecular Engineering, University of Chicago, Chicago, Illinois 60637, United States Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439, United States § División de Ciencias e Ingenierías, Campus León, Universidad de Guanajuato, Loma del Bosque 103, León (Gto.) 37150, Mexico ∥ Departamento Materiales y Minerales, Universidad Nacional de Colombia−Medellin, Calle 75 # 79A-51, Bloque M17, Medellin, Colombia ⊥ Cain Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803, United States ‡

S Supporting Information *

ABSTRACT: In exploiting topological defects of liquid crystals as the targeting sites for trapping colloidal objects, previous work has relied on topographic features with uniform anchoring to create defects, achieving limited density and spacing of particles. We report a generalizable strategy to create topological defects on chemically patterned surfaces to assemble particles in precisely defined locations with a tunable interparticle distance at nanoscale dimensions. Informed by experimental observations and numerical simulations that indicate that liquid crystals, confined between a homeotropic-anchoring surface and a surface with lithographically defined planar-anchoring stripes in a homeotropicanchoring background, display splay-bend deformation, we successfully create pairs of defects and subsequently trap particles with controlled spacing by designing patterns of intersecting stripes aligned at 45° with homeotropic-anchoring gaps at the intersections. Application of electric fields allows for dynamic control of trapped particles. The tunability, responsiveness, and adaptability of this platform provide the opportunities for assembly of colloidal structures toward functional materials. KEYWORDS: nematic liquid crystal, topological defect, colloidal particle, directed self-assembly, chemical pattern, electric field

T

dimensional (2D) or three-dimensional (3D) arrays of colloidal particles can be assembled employing such topological defect mediated interactions.9−13 Optical tweezers have been utilized to enable accurate particle positioning for their organization;2,5 however, constant intervention and manipulation is required during the assembly process, thereby limiting the application of these techniques for high-resolution and large-area systems. Topographically patterned surfaces can be engineered to introduce LC defects as the targeting sites for the assembly of colloidal particles. Both surface anchoring and topography are leveraged to endow control over the orientation of LCs at the boundary of the confining surfaces. Frustrated boundary conditions foster topological incompatibility with the existing

he directed assembly of colloidal particles by exploiting elastic deformations associated with liquid crystal (LC) systems is of great interest from both fundamental and technological points of view.1−4 Colloidal particles dispersed in nematic LC systems disturb the spatial uniformity of the director field, penalize the free energy, and generate highenergy defects in the vicinity of particles. These defects attract each other and therefore can be used for the manipulation and assembly of colloidal particles.5 The type of LC defects depends on the surface anchoring of the particles. Particles that induce homeotropic (perpendicular) LC alignment, for example, generate point or Saturn-ring defects.6 On the other hand, boojum defects are located at the particle poles for planar (parallel) anchoring.7 When multiple particles come close in a nematic LC, straight or zigzag chains are formed from dipolar or quadrupolar nematic colloids, respectively, in order to minimize the energetic penalizations.8 Furthermore, two© 2017 American Chemical Society

Received: May 24, 2017 Accepted: June 12, 2017 Published: June 12, 2017 6492

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

www.acsnano.org

Article

ACS Nano

Figure 1. |Illustration of materials and patterning process. (a) Chemical structure of PMMAZO containing azobenzene mesogen and OTS selfassembled monolayer. After forming a brush or SAM layer, both the PMMAZO mesogen and OTS induced homeotropic anchoring of 5CB. (b) Schematic representation of the PMMAZO/OTS layer substrates achieved on nanoscale, disjoint stripes with 45° angle patterns generated by e-beam lithography. The patterned areas allowed different anchoring behavior areas. The 5CB was aligned by the planar anchoring inside the stripe area, while homeotropic anchoring occurred outside the stripe area. Sw, Mw, and Gw are the width of the side stripes, the width of the middle stripe, and the width of the gap space, respectively. (c) SEM image of disjointed stripe pattern with Sw = Mw = 500 nm and Gw = 50 nm; 3D AFM images of PMMAZO brush pattern. (d) The homeotropic anchoring glass substrate was placed face-to-face with the patterned substrate as a cell, and 5CB was injected into the gap by capillary action; the splay-bend deformation from the simulation is shown.18

director field, resulting in distorted alignments or defect structures that increase the total free energy of the system.14 Importantly, the energy barrier of the confined defects is greater than the thermal energy of the system at room temperature,15 making the defect structures a highly stable system for the engineered assembly of colloidal particles. For instance, colloidal particles, with the same anchoring as the confining surface, are attracted to the surface exhibiting a certain shape (concave) complementary to that of the colloidal particle.16 Similarly, particles are immobilized onto the tip of a pyramid-like protrusion (convex) by sharing their topological defects with the topographic pattern.16,17 The separation of the colloidal particles trapped with the pyramid template is more than 5 μm, determined by the photolithographic resolution. Another concave system was implemented to organize particles in a zigzag chain by confining nematic LCs in a sinusoidal microwrinkle.14 A minimum of 150 nm is required for the amplitude of the groove in order to maintain stable particle immobilization, and the distance between trapped silica beads is limited to be larger than 15 μm. Despite the accurate positioning of individual colloidal particles, the interparticle distance is limited to the range of a few micrometers and above, hindering the practical application of using topological patterns for particle assembly. From a fundamental point of view, topological features of a size similar to that of colloidal particles generate defects comparable in size to them, behaving as pseudoparticles in surface−particle interactions. In this work, we have developed an approach for the directed assembly of colloidal particles near surfaces through judiciously designed LC topological defects induced by chemical patterns.

Regions of different anchoring and orientation characteristics in various geometries are fabricated with nanoscale precision to control both the location and distance between particles. We recently explored the behavior of 4′-pentyl-4-biphenylcarbonitrile (5CB) LC molecules in a hybrid cell with a top surface eliciting uniform homeotropic anchoring and a bottom surface that is chemically patterned with straight stripes of planar anchoring in a background of homeotropic anchoring.18 We observed a morphological transition from a uniform undistorted alignment to a dual uniform/splay-bend morphology and clearly identified the regime of geometrical and anchoring conditions for each LC alignment, i.e., cell thickness, pattern width, and surface anchoring strength. Numerical simulations revealed that LC molecules in the proximity of the homeotropic/planar boundary undergo significant elastic distortion that propagates throughout the stripe width. On the basis of this fundamental understanding, LC morphologies with custom-designed defect structures, e.g., multiple director orientations, are created through rational design of chemical patterns on 2D surfaces, thereby providing the opportunities for the assembly of colloidal particles through defect−particle interactions over large areas. As a proof of principle, we have designed a chemical pattern that creates equally spaced LC defects that serve as preferential targets to achieve controlled localization of colloidal particles, despite their tendency to form agglomerates through the interparticle defect interaction. On the basis of the Landau−de Gennes framework, we construct a detailed picture of the LC thermodynamic states, providing insights into the competition between surface anchoring frustrations (substrates and particles) and bulk elastic 6493

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

ACS Nano

Figure 2. |Ground-state nematic defects morphology. (a) PMMAZO brush as the weak homeotropic anchoring background: (a1, a3) reflective light microscopy images of the 5CB cell under the crossed polarizers at two rotation angles relative to the crossed polarizer (the crossed polarizer is marked); (the inner images of a1, a3) simulated crossed polarizers images of the 5CB cell at the same position relative to the experimental results; (a2) simulated top-down view of the director field of 5CB orientating on the pattern substrate (x−y plane); (a4) simulated cross-sectional director field of the 5CB at the gap space. (b) OTS layer as the strong homeotropic anchoring background showing comparable information with b1−b4.

chain/nm2) obtained under these conditions rendered the backbone of the PMMAZO macromolecule oriented parallel to the substrate (on average), thereby inducing a perpendicular orientation of the side-chain azobenzene mesogens (on average). This configuration enabled a weak homeotropic anchoring toward the LC molecules through “interdigitation” with the brush mesogens. On the other hand, strong homeotropic anchoring of 5CB was achieved with a surface coated with the OTS SAM (Figure 1a). The removal of PMMAZO or OTS, by exposing them to oxygen plasma, changed the LC anchoring from homeotropic to planar, the inherent anchoring imposed by the underlying silicon wafer. The anchoring strength of 5CB on the PMMAZO brush and the silicon substrate after removal of the brush was measured to be 3.2 × 10−5 J/m2 (homeotropic) and 4.2 × 10−4 J/m2 (planar), respectively. On the other hand, the anchoring strength of 5CB on the homeotropic OTS is 2 × 10−2 J/m2, typically categorized as strong anchoring. A schematic of the fabrication process is shown in Figure 1b. After the uniform PMMAZO brush was deposited on the silicon substrate, a photoresist was patterned on top of the brush using electron beam lithography (EBL), opening arrays of trenches. The sample was then exposed to oxygen plasma to remove the brushes in areas that were not protected by the overlying photoresist. After removal of the photoresist, the chemical pattern was obtained. As a proof of principle, we designed a pattern consisting of a continuous middle stripe and two groups of disjointed diagonal stripes that were positioned at an angle of 45° with respect to the middle stripe. A

distortions in depicting defect−particle interactions. Furthermore, we show that chemically engineered LC morphology and the defect-directed particle assembly can be controlled through the application of an external electric field. The nematic LC is forced to align uniformly along the direction of the field, and the deprivation of defects results in particles releasing from the designated locations and diffusing freely. The removal of the field allows full recovery of the particle assembly, demonstrating the reproducibility of these defect structures and the specificity of the defect−particle interactions. Additionally, specifically orientated external fields have also been used to facilitate LC alignment for remediation of particle misalignment. The controllability, responsiveness, and adaptability of this platform to assemble particles on chemically patterned surfaces provide the opportunities for the construction of functional colloidal materials and devices.

RESULTS AND DISCUSSION LC Morphology Control with Chemically Patterned Surfaces. To create surfaces with controlled anchoring orientation and strength toward LCs there, we deposited a polymer brush or a self-assembled monolayer (SAM) on a silicon substrate. Hydroxyl-terminated poly(6-(4-methoxyazobenzene-4′-oxy)hexyl methacrylate) (PMMAZO) formed the brush, while octadecyltrichlorosilane (OTS) was used to make the SAM (Figure 1a). The polymer brush is formed by coating a 4 to 5 nm thick film of PMMAZO (56 000 g mol−1) onto a clean silicon wafer that was annealed at 250 °C for 5 min. The grafting density of the PMMAZO brush (∼0.051 6494

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

ACS Nano

experimental observations. Recall the LC must create a twist in order to change orientation of the anchored molecules, from homeotropic to planar, at the borders of the mixed conditions in the chemical patterns. Further, the cross-sectional view of the director field (Figure 2a4, section AA) reveals that the LC orientation at the gap (Gw) is not strictly perpendicular and the splay-bend regions of the middle and side stripes merge over the 50 nm gap. The splay-bend deformation is driven by the necessity of the LC molecules to accommodate the strong homeotropic anchoring from the top OTS-modified glass surface and the planar anchoring from the chemical patterns at the bottom surface. A detailed analysis of the brightness intensity in the crossregion is included in the Supporting Information, where four distinct areas are taken into account: middle stripe, side stripe, crossed area, and gap space (Figure S1). The brightness intensity of different planar anchoring areas as a function of the angle between the middle stripe and the crossed polarizers shows a similar profile for the middle stripe and gap space, clearly indicating that the LC has a similar alignment at these two regions. LC molecules at the crossed area and the side stripe exhibit similar orientation, further confirming the origin of the bright features (Figure 2a1) on the middle stripe observed from PLM with the polarizers parallel to the middle stripe. The abrupt shift in the director field of the 5CB in the middle stripe fosters the formation of defect sites in the LC near the substrate surface. Guided by knowledge established through our previous study,18 we designed the width of the middle stripe and side stripes to be 500 nm to ensure that the planar-anchored molecules were oriented along the direction of the stripes. To introduce precisely controlled defects along the middle stripe, the two 45° stripes were placed in close proximity (Gw = 50 nm) to the middle stripe. The frustration of the 5CB orientation along the middle stripe was altered by the side stripes only if the homeotropic anchoring induced by adjacent polymer brush-grafted gap regions was weak. To conclude, the weak-anchoring PMMAZO brush at the gap allows the two side stripes to disrupt the orientation of the LC molecules that are aligned planar along the middle stripe. Conversely, both the middle and side stripes disturb the weak homeotropic anchoring at the small gap (see section AA in Figure 2a4), making it a target site for the polystyrene colloidal particles. A surface with strong homeotropic anchoring, however, mitigated the frustration of the 5CB alignment along the middle stripe even if the two side stripes nearly intersected the middle stripe (Figure 2b). The polarized optical micrographs (POMs) in Figure 2b1 and b3 show the birefringence contrast for 5CB confined on stripes against a background of OTS SAMs, which impose strong homeotropic anchoring in a chemical pattern with the same geometric features. In stark contrast to the brightness map in Figure 2a1, the middle stripe remains continuously dark along the entire stripe, including the regions where it interfaces with the two side stripes. After rotating the polarizer by 45° (Figure 2b3), the middle stripe shows a uniform brightness along the entire stripe, suggesting a consistent alignment of 5CB molecules along the middle stripe. It is clear that the surface chemistry of the background (PMMAZO brushes or OTS SAMs) plays a pivotal role in generating the 5CB defect sites. The experimental observations are confirmed by numerical simulations (Figure 2b2 and insets of Figure 2b1 and b3). Different from a PMMAZO brush that

topographic step of 4 nm, the thickness of the PMMAZO brush, was created at the borders of each stripe and the homeotropic background (Figure 1c). The width of the patterned middle stripes (Mw) was varied between 300 and 900 nm, and the length was designed to be 80 μm to avoid end effects. The width (Sw) and length of the diagonal side stripes were kept at 500 nm and 40 μm, respectively. The separation of the side stripes was 5 μm for all samples to ensure that the LC behavior at each stripe was independent of the neighboring features in the pattern. A 50 nm gap (Gw), grafted with the same PMMAZO brush, separates the middle stripe and the side stripes. As will be discussed later, the anchoring strength of the gap region plays a critical role in defect generation. The LC was confined in a hybrid cell with a thickness (H) between 100 nm and 1.5 μm, where the patterned silicon substrate and an OTSmodified glass substrate were used as the bottom and top plates, respectively (Figure 1d). The director field of 5CB on the chemical patterns was analyzed by observing the birefringence phenomenon of the LC cell between crossed polarizers using polarized light microscopy (PLM). To interpret these images, we note that 5CB molecules with homeotropic anchoring at the top and bottom substrates display no bright features. On the other hand, 5CB molecules with planar orientation give rise to the brightest features when oriented 45° with respect to the crossed polarizers and appear dark when oriented parallel (0°) to the crossed polarizers. These observations are a reflection of the splay-bend deformation that the LC undergoes to respond to the change in anchoring conditions between the planar substrate at the bottom and the homeotropic direction at the top (Figure 1d). Ground-State Nematic Defects Morphology. The PMMAZO chemical brush layer introduced weak homeotropic anchoring for 5CB in the background region, rendering the planar 5CB in the patterned stripes susceptible to frustration where the diagonal stripes nearly intersect the middle stripe. Figure 2a1 shows the birefringence contrast for the 5CB over the chemical pattern. The crossed polarizers are oriented at a 45° angle with respect to the continuous stripe. A set of discontinuous bright spots along the continuous stripe are observed that are located near the gaps between stripes. The simulated birefringence is included in the inset, and the nematic orientation along the stripes, obtained from simulations, is shown in Figure 2a2. The LC alignment of the middle and side stripes at regions far away from the gaps follows the direction of each stripe, as shown by the color scheme of the director in Figure 2a2. At the disjointed intersections, the azimuthal alignment of the LC molecules on the middle stripe is modified due to the presence of the two 45° stripes despite the existence of 50 nm homeotropic anchoring gaps, as illustrated with red arrows, confirming the weak homeotropic anchoring imposed by PMMAZO brushes. At 45° rotation of the crossed polarizers (Figure 2a3), the discontinuous side stripes become dark and the continuous middle stripe appears bright. The birefringence profile along the middle stripe shows discontinuous spots at the crossing regions, which can be attributed to the twist deformation of LCs at the borders of the stripes (left inset, high-twist regions are in yellow isosurfaces), where two regions of competing anchoring properties are in close proximity in the chemical patterns. According to the inset of Figure 2a3, for the PMMAZO chemical substrates, the twist isosurfaces are highly deformed at these crossing areas, thereby generating the dark spots in the birefringence, as shown in the simulated brightness intensity map (right inset), which is consistent with 6495

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

ACS Nano

Figure 3. |Directed self-assembly of colloidal particles onto target defect sites. The polystyrene colloidal particles were trapped at the gap spaces on patterned PMMAZO brush layer substrates. The reflective light microscopy images of the 5CB mixture with a colloidal particle cell (cell thickness = 1.5 μm) under the crossed polarizers with different middle stripe width of the disjointed stripe patterns on the PMMAZO brush substrate and the corresponding optical images of a pair of trapped PS colloidal particles at the distorted gap spaces: (a1−a3) 500 nm and (b1−b3) 900 nm. The polarizer is marked. (c1, c2) Schematic representation of the polystyrene colloidal particles trapped at the gap spaces obtained by simulation in the Landau−de Gennes framework (the director fields affected by the two colloidal particles are shown as the topdown and cross-sectional images).

Directed Self-Assembly of Colloidal Particles. When polystyrene (PS) colloidal particles (800 nm in diameter), with the outer surface planar anchoring toward 5CB, are introduced into the nematic cell, they are immediately attracted to the topological defects created by the chemically patterned surface. The cell thickness was increased from 100 nm to 1.5 μm to accommodate the PS particles. Increasing the thickness of the hybrid cell did not alter the defect structures, as confirmed by the POM images in Figure S2 for Mw = 300, 500, and 900 nm. The two arrays of side stripes, separated by a 50 nm gap at each side from the continuous middle stripe, lead to the formation of an ordered array of two rows of equally spaced defects that subsequently dictate the arrangement of the PS particles assembled on the surface, as shown in the POM images with the middle stripe at 500 and 900 nm in width. Figure 3a1, a2, b1, and b2 show dark and bright features along the stripes of the cell with the colloidal particles under crossed polarizers, while the optical micrograph images under parallel-oriented polarizers (Figure 3a3 and b3) exhibit a blue LC background decorated with black spots that represent the colloidal particles. For Mw = 500 and 900 nm, the colloidal particles are assembled perfectly along the middle stripe near the intersection gaps with the side

allows the side stripes to disrupt the orientation of the planaranchored LC molecules along the middle stipe, a stronger homeotropic-anchoring OTS SAM in the 50 nm gap regions prevents the planar-anchored 5CB in the middle stripe from altering its orientation in the presence of side stripes. The 50 nm gap acts as a wall that blocks the interconnection between the molecules on the middle stripe and side stripes, as illustrated in section BB of Figure 2b4, with the splay-bend dome at the middle stripe being completely formed and independent from the domes of the side stripes. In summary, strong homeotropic anchoring in the gap areas mitigates the deformation of the LC alignment at the intersection between the stripes, while weak homeotropic anchoring allows the deformation of the LC molecules in the gap. The latter incurs mixed x−y orientations of the planaranchored LCs in the middle stripe and drives the formation of topological defects at the gap regions that can serve as anchoring spots for the assembly of colloidal particles. The LC morphology and defect distribution at the ground state in the hybrid nematic cell can be rationally designed through the type of LC, the geometry of chemical patterns, and the anchoring strength. 6496

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

ACS Nano

were carried out in order to predict and understand the minimum energy state: (i) the particles are distributed in a series of different configurations with respect to the gap, and (ii) the particles are separated from the bottom substrate at various distance Δz. We tested eight different scenarios and found that the minimum energy state is when the particles are located above the homeotropic gap, which is in good agreement with experimental observations. The findings from four scenarios are summarized in Figure 4. For the first scenario

stripes. The presence of the particles results in slightly different birefringence contrast (Figure 3a1, a2, b1, and b2): the discontinued bright stripe exhibits a round contour instead of straight edges (Figure 2a3) when the middle stripe is aligned at the brightest position (45° to the crossed polarizers). This change can be attributed to the deformation of LC molecules in order to follow the curvature of the colloids whose surface is planar anchoring. In a systematic study, we gradually rotated the hybrid cell with respect to the cross polarizers using the middle stripe as a reference (Figure S3), both patterns, with Mw = 500 and 900 nm, show the similar wavy shape for the middle stripe. Further, a similar disjointed stripe pattern with a round end for the side stripes was also designed (Figure S3c) that demonstrates the same wavy shape for the middle stripe and the same round discontinuous dots when the middle stripe is placed at 45° to the crossed polarizers. Numerical simulations aid to provide the detailed thermodynamic and morphological states of the LC/colloid system. Figure 3c shows the arrangement of defects at the particle surface (magenta isosurfaces) and the nematic director for two particles that are located above the homeotropic gap regions between stripes. Planar-anchored particles in a nematic LC results in two boojum defects that are located at opposite poles, which typically drives the self-assembly of particles into aggregates to minimize the energy penalty. The interparticle interaction in our system, however, is mitigated (Figure 3c) in the specifically designed LC ground state. Once the particles are located at the chemically induced defects, the boojum near the substrate diffuses to avoid the surface; as a result, the boojums are no longer located at opposite poles. The width of the middle stripe is a key design parameter to control the separation between the particles: ∼20 nm for Mw = 500 nm and ∼500 nm for Mw = 900 nm. On the other hand, the particles induce a disturbance of the LC’s director field at the intersectional area, as illustrated in the top and side views of the director field (Figure 3c). It is important to note that once the colloidal particles are immobilized above the gap regions, the LC alignments at the middle and side stripes are “disconnected” and follow their own direction along the stripe, respectively. In addition, the LC alignment at the middle stripe near the intersection areas is affected by colloidal particles and forced to follow the curvature of the particles, forming a “wavelike” nematic, which is consistent with the experimental observation discussed above. Moreover, the splay-bend domes are expanded and merged with the alignment at the particle surface. These behaviors of the nematic direction provide an entire rationale behind the tendency of the particles to “relax” the defects; the particles alleviate many of the nematic distortions that are induced by our designed ground state. Experiments and simulations with a narrower strip, Mw = 300 nm, are shown in Figure S4. In this situation, the distance between defects is smaller than the particle size and some of colloidal particles are located above one of the defects, creating a single particle line. The remaining particles are now distributed randomly in the LC cell (Figure S4a). Consistent results are observed once PS particles of 1 μm in diameter are introduced in the Mw = 300 nm continuous stripe (Figure S4b); that is, a single line of particles is located above the defects. The minimum energy location of the particles with respect to the homeotropic gap and the chemically patterned substrate is revealed with full-scale numerical simulations that take into account different anchoring conditions and full geometrical details used in experiments. Two sets of numerical experiments

Figure 4. |Numerical simulations provide thermodynamic and morphological states of the LC/colloids system. Energy comparison of different colloidal particle positions located in the Mw = 500 nm system: (a) Potential of mean force F(case) − F(4) as a function of the four cases proposed. The insets schematically represent the location of two nanoparticles for each case. The colored area correspond to areas where the orientation of the LC molecules anchored in a random planar way; otherwise, a perpendicular alignment was imposed. (b) Elastic energy of a liquid crystal mixture with a colloidal particle system as a function of the height of colloids suspended in the bottom surface.

(N = 1), both particles are located far from the defect regions and the stripes. For N = 2 and 3, both colloidal particles are positioned at the edge of the middle stripe. Specifically, for N = 2, the particles are close to the side stripe, while for N = 3 they are far away from it. Finally, N = 4 corresponds to the arrangement of particles observed in the experiments (above the homeotropic gap). Figure 4a plots the free energy difference between N = 1, 2, and 3 and the minimum N = 4, F − F4, for particles that are Δz = 50 nm from the bottom substrate. The free energy difference (∼600 kBT) is the largest for N = 1 because of the additional energy penalization due to the presence of the two particles in the homeotropic background. Locating the particles at the edge of the middle stripe relaxes the distortions around the particles that are merged with the splay-bend dome. Consequently, for N = 2 the energy excess decreases to ∼300 kBT. On the other hand, for N = 3, there is only the interaction between particles and the splay-bend dome of the middle stripe, thereby relaxing the energy to a greater extent for a total excess of ∼50 kBT. Finally, when the particles are localized at the defect regions (N = 4), 6497

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

ACS Nano

Figure 5. |Manipulating the arrangement of colloids with electric fields. (a) Schematic of trapped PS nanoparticles at the distorted gap areas and applied electric field in the z- and x-direction, which was obtained by simulation in the Landau−de Gennes framework (the director fields effected by the two colloidal particles are shown as the top-down images). Optical images of trapped PS colloidal particles following the electric field first in the z-direction from the top surface to the bottom surface and then in the x-direction under two middle stripe widths: (b) Mw = 500 nm and (c) Mw = 900 nm.

particles are located on every defect region with the desirable separation between them. For instance, for 800 nm particles, we demonstrated that particle separation could be varied from tens to hundreds of nanometers by changing Mw. As long as the ratio between particle size and Mw is on the order of 1, pairs of particles will be attracted to every single defect spot, as shown in Figure 3. Notice that the defects driven by the chemical pattern inhibit the particles’ tendency to form clusters. Therefore, our approach is different from previous works, where topographically patterned surfaces were used to form concave or convex pseudoparticles, where the localization of the particles was then done following the rationale of particle− particle agglomeration.14,16,17 These valid schemes do not allow total control on particle separation at the nanoscale, and particle localization in every defect is also hard to achieve. Manipulating the Arrangement of Colloidal Particles with Electric Fields. Electric fields have the ability to align LCs and therefore can be used to examine the reversibility of the LC ground states and further the mechanism for the directed assembly of colloidal particles. The applied electric field dominates the morphology of the nematic LC that is otherwise controlled by the chemical pattern. As a consequence, it eliminates the defect regions and disturbs the assembly of the colloidal particles. Figure 5a shows the simulated influence of applied fields on the LC director field and the boojums on the particles (magenta isosurfaces). With an applied vertical field, Ez, the nematic aligns perpendicular to the walls and the boojums are generated in opposite poles that point to the walls. A field parallel to the wall and along the middle stripe, Ex, induces a nematic that is parallel to the walls

the defects in the ground state of LC and on the surface of the particles interact to minimize the distortions and therefore free energy excess. In this configuration, the particles act like a wall that breaks the preferential alignment imposed by the side stripe. The optimal substrate-to-particle separation is predicted by calculating the free energy, as a function of the distance Δz between the particle surface and substrate in reference to F(Δz = 50 nm), i.e., F − F(Δz = 50 nm), as shown in Figure 4b. The minimum energy is found at the particle−substrate separation of 150 nm, although intuition may suggest that the particles are in contact with the bottom surface of the substrate. The physical mechanism behind this separation can be understood as follows: if the particles are too close to the bottom substrate, the director field around them would affect that near the homeotropic gap region, which drives an energy penalization of ∼20 kBT. On the other hand, if the particles move further toward the top substrate, the free energy increases monotonically because of the lack of interaction with the topological defects over chemical patterns to relax the ground-state energy. In particular, at Δz = 700 nm, the free energy difference is around ∼300 kBT, which is similar to that at Δz = 50 nm with the particles located far from the defect regions (Figure 4a). There are two major features to highlight from our directed particle assembly platform. First, the design and fabrication of the chemical patterns at the nanoscale allow for control over the particle separation at narrow and precisely designed distances. Second, the characteristic size of the LC defect region depends on the geometry and anchoring conditions imposed on the chemical patterns. Consequently, all the 6498

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

ACS Nano

LC molecules to align planar along the orientation of the middle stripe and therefore facilitates the reorientation of LCs following the planar-anchoring stripes. The responsiveness of the LC topological defect structures to external electric fields provides opportunities for the manipulation and reconfiguration of the attained particle assemblies to achieve adaptive functional materials.

and boojums at opposite poles that are orientated along the middle stripe. For both fields, the nematic spreads throughout the entire cell. Previous studies have shown that planaranchoring particles in a nematic may aggregate or disperse, depending on the particle alignment with respect to the nematic and the distance between particles.19 Therefore, we expect the delocalization of particles upon application of the electric field will allow the particles to aggregate or diffuse. The colloidal particles resume their ordered arrangement upon removal of the electric field, demonstrating the reproducibility and responsiveness of our LC-based colloid assembly platform. The observations are summarized in Figure 5b and c for Mw = 500 and 900 nm, respectively. Prior to the electric field, the hybrid LC cell is in its stable ground state, where the colloidal particles are trapped in the defect regions, forming assembled particles in an organized array. The electric field in the zdirection, Ez, is achieved by sandwiching the LC cell between two electrodes with one connected to the indium tin oxide (ITO) glass surface and the other to the bottom silicon substrate. A 90 mA current was used to generate the field. Figure 5b and c show the progression of the nematic field along the z-direction from t = 0 to t = 67 s. At t = 2 s, the lines and particles appear blurry. This effect becomes more predominant until at t = 67 s the colloidal particles and patterned lines are completely invisible, indicating that the LC molecules are entirely oriented along the electric field in the z-direction. Immediately following the removal of the Ez field, the stripes and particle arrays are restored (t′ = 9 s), clearly demonstrating the reversibility of the system, which has important implications for manipulation and reconfiguration of the colloid arrangement. For Mw = 500 nm, the particles recover the wellorganized array, whereas for Mw = 900 nm, some of the particles appear aggregated and meanwhile unoccupied defect sites occur. Both systems were tested in repeated on−off cycles for the external electric field. Particles in the Mw = 500 nm cell show a perfectly reproducible arrangement. The Mw = 900 nm system, however, appears less reliable, showing particle aggregation and missing anchor spots in some of the experiments. The distance between particles is a few tens of nanometers for Mw = 500 nm, and a nematic drives the particle separation; for Mw = 900 nm the original interparticle distance is hundreds of nanometers, and a nematic pushes the particles to aggregation, depending on the particle diffusion, thereby resulting in the “random” character of the observations. Importantly, the misalignment in the restored particle array can be readily eliminated by use of an electric field applied across the x−y plane, Ex. It takes t″ = 5 s for the nematic to align parallel to the walls (Figure 5b and c) (Ex was generated with a 250 mA current). Likewise, boojums are generated at opposite poles on the surface of the particles that are also parallel to the walls (Figure 5a, right panel). Once the field is turned off and the ground state recovered, all particles are organized instantly into the original array structure for both Mw = 500 and 900 nm. The rapid restoration of the colloidal array is an indication of fast kinetics, which is an important feature for adaptive material systems. On the basis of our previous work,18 LC on the 900 nm wide middle stripe assumes multiple director orientations, which can cause the shifting or joining of defects at the gap regions when the LC reorients from homeotropic alignment to follow the planar-anchoring stripes upon removal of the Ez electric field, thereby resulting in misalignment of colloidal particles. In comparison, applying an electric field along the middle stripe in the x−y plane forces all

CONCLUSIONS In summary, we have developed an approach for the directed assembly of colloidal particles through interactions with rationally designed LC topological defects induced by chemically patterned surfaces with nanoscale features. On a chemical pattern that contains a homeotropic background and a continuous planar anchoring middle stripe and two arrays of disjointed side stripes, the 50 nm wide gap regions impose a nematic distortion of the LC alignment, thereby creating equally spaced topological defects, which serve as preferential anchor spots for the localization of colloidal particles to form well-organized arrays. We use numerical simulations based on the Landau−de Gennes framework to construct a detailed picture of the LC thermodynamic states and to provide insight into the competition between surface-anchoring frustrations for the substrate and particles and the bulk elastic distortions in depicting defect−particle interactions. Furthermore, we show that a chemically engineered LC morphology and the defectdirected particle assembly can be controlled through the application of an external electric field, demonstrating the stability of the defect structures generated with 2D chemical patterns and the specificity of the defect−particle interactions. Additionally, oriented external fields oriented along specific direction have also been used to facilitate LC alignment for remediation of particle misalignment. The controllability, responsiveness, and adaptability of the LC-based particle assembly platform provide the opportunities to organize colloidal particles into functional materials and devices. MATERIALS AND METHODS Materials. The LC used in our study was 4-pentyl-4cyanobiphenyl, a thermotropic LC that exhibits a nematic phase between 24 and 35 °C, which was purchases from Sigma-Aldrich and used without further purification. Octadecyltrichlorosilane, heptane, toluene, chlorobenzene, tetrahydrofuran (THF), petroleum ether, anisole, and dichloromethane (DCM) were purchased from SigmaAldrich and used without further purification. Glass microscope slides were obtained from Fisher Scientific. Synthesis of the PMMAZO Brush. The azobenzene monomer 6(4-methoxyazobenzene-4′-oxy)hexyl methacrylate (MMAZO) was synthesized according to the procedure described by Stewart and Imrie,20 Then the PMMAZAO brush was synthesized by nitroxidemediated controlled radical polymerization (NMP). A mixture of initiator NMP (0.0244 g, 0.075 mmol) and MMAZAO (2.3160 g, 6 mmol) was degassed by three freeze−thaw cycles. The mixture was heated at 120 °C for 24 h under a nitrogen atmosphere. The resulting polymers were dissolved in THF and then precipitated in petroleum ether. The average molecular weight was around 56 000 g/mol. Preparation of Chemical Patterns with a PMMAZO Brush. The process to create chemical patterns is shown in Figure 1. A ∼4 nm thick PMMAZO film was deposited on an oxygen-plasma-cleaned silicon substrate by spin-coating from a 0.05 wt % toluene solution and annealed at 250 °C for 5 min under vacuum. Un-cross-linked PMMAZO was removed by sonication in toluene, and the remaining PMMAZO brush was ∼4 nm. A 40 nm thick GL2000 photoresist film was deposited onto the PMMAZO brush from a 1.0 wt % anisole solution and baked at 160 °C for 5 min. Stripe patterns were created 6499

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

ACS Nano

about the modeling and simulations are provided in the Supporting Information.

on the resists using electron beam lithography. EBL was performed using the JEOL 9300FS electron-beam writer at a 30 keV acceleration voltage and 1 nA current at the Center for Nanoscale Materials, Argonne National Laboratory. Exposed substrates were developed with n-amyl acetate for 15 s and rinsed with isopropyl alcohol. The resulting resist pattern was transformed into a chemical pattern on the PMMAZO brush layer by exposing the sample to an oxygen plasma followed by stripping the GL2000 photoresist in chlorobenzene. Assembly of Optical Cells. The glass microscope slides were boiled in a piranha solution (7:3 (v/v) of 98% H2SO4/30% H2O2) for 30 min to remove any stains on the surface, then washed with deionized water, and dried with a nitrogen flow. Then the cleaned glass slides were immersed in a mixture of 13.8 μL of OTS and 120 mL of heptane. After 30 min they were removed from the OTS solution, washed with DCM several times, and quickly dried under a nitrogen flow. Finally, the two substrates, an OTS-glass and a chemically patterned Si substrate, were placed face-to-face to create a cell. The optical cell and 5CB were both warmed to above the clearing point of 5CB, which was then injected into the optical cell and dispersed by capillary action and slowly cooled to room temperature. Colloidal Particle Localization at Defect Regions and under an Electric Field. Polystyrene particles with a diameter of 800 nm (Life Technologies, 4 wt %, water dispersed) were mixed with 5CB to make a 0.1 wt % suspension. After water was removed using a vacuum pump, the mixture was sonicated to disperse the particles in LC. The mixture was then introduced into the cell for further characterization. The surface anchoring (planar) strength of the particles was determined to be 1.06 × 10−5 J/m2. For studies with external electric fields (Gamry 600Ref potentiostat/galvanostat), the top substrate was changed to ITO-glass and the bottom substrate was conductive Si. The electric field was applied between the electrodes with a constant current of 90 mA/5.5 V (two electrodes at the top and bottom substrates, respectively), and 250 mA/3.6 V (two electrodes both at the Si bottom substrate along the middle stripe direction) was applied. Characterization. The thickness of the PMMAZO brush thin film was measured by a Woollam VUV-VASE32 variable-angle spectroscopic ellipsometer. The stripe patterns were examined with atomic force microscopy (AFM). Scanning electron microscopy images were obtained using a Zeiss Merlin FE-SEM with an accelerating voltage of 1 kV and working distance of 3 mm. The optical images of LCs in cells were measured by a polarized light microscope (BX 60, Olympus). The intensity of the patterns was determined with the software ImageJ. The color images were converted to 8-bit grayscale, in which pixel values range from 0 (complete black) to 255 (complete white). One single bright line was selected in each image, and the intensity value was measured. Intensity values of three separate lines were then averaged for each image. Numerical Simulations and Thermodynamic Description. Numerical simulations were carried out based on the mean-field Landau−de Gennes approach, and the free energy minimization was achieved using a theoretically informed Monte Carlo relaxation method.21,22 The free energy functional (or Hamiltonian) was constructed utilizing the Q-tensor representation that included shortrange contributions to the energy and long-range contributions and surface contributions to the free energy that control the local LC alignment near the substrate. The free energy functional used in this work is described by F = F(A, U, k11, k22, k33, WH,OTS, WH,PMMAZO, WP,SiO, WP,Nano), where U controls the isotropic−nematic transition, A sets the energy scale, k11, k22, and k33 are nonvanishing elastic moduli corresponding to independent splay, twist, and bend deformation modes,23 and WH,OTS, WH,PMMAZO, WP,SiO, and WP,Nano represent the anchoring strength for perpendicular (homeotropic) and parallel (random planar) alignment for OTS, the PMMAZO polymer brush, SiOx, and polystyrene particles, respectively. The parameters for 5CB are the same as the values previously reported: A ≈ 1 × 105, k11 = 6 pN, k33 = 10 pN, and k22 = 3 pN. The isotropic−nematic transition parameter is set to be U = 3.165, which corresponds to a bulk value of the scalar order parameter of S ≈ 0.54.24 Cross-polarizer images were obtained based on the Jones 2 × 2 formalism.25 Additional details

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.7b03641. Reflective light microscope images of the 5CB cell under crossed polarizers: under different rotation angle; changing the cell thickness (H = 1.5 μm); 5CB mixture with colloids when Mw = 500 and 900 nm of different rotation angle; trapped colloidal particles at the gap spaces on patterned substrates when Mw = 300 nm (PDF)

AUTHOR INFORMATION Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Xiao Li: 0000-0003-3456-1933 Leonidas E. Ocola: 0000-0003-4990-1064 Camille Bishop: 0000-0002-2889-1752 Juan J. de Pablo: 0000-0002-3526-516X Paul F. Nealey: 0000-0003-3889-142X Author Contributions

X.L. and P.F.N. conceived and designed the experiments. J.C.A.-P., J.P.H.-O, and J.J.d.P. conceived and performed the numerical simulations and theoretical calculations. X.L., J.C.A.P., J.P.H.-O., J.J.d.P., and P.F.N. wrote the manuscript. J.J.d.P. and P.F.N. guided the work. All authors discussed the results and contributed to data analysis and manuscript revision. Author Contributions #

X. Li and J. C. Armas-Pérez contributed equally to this work.

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS We gratefully acknowledge support by the U.S. Department of Energy, Basic Energy Sciences, Materials Sciences and Engineering Division. We acknowledge the use of the facility resources for the experimental part from the Center for Nanoscale Materials, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357, and the University of Chicago MRSEC Shared User Facilities (NSF DMR-1420709). We acknowledge the use of the computational facility for the simulation part from University of Chicago Research Computing Center; Argonne Laboratory Resource Computing Center (LCRC); and the Innovation and Novel Computational Impact on Theory and Experiment (INCITE) program of the Argonne Leadership Computing Facility at Argonne National Laboratory. REFERENCES (1) Rahimi, M.; Roberts, T. F.; Armas-Perez, J. C.; Wang, X.; Bukusoglu, E.; Abbott, N. L.; de Pablo, J. J. Nanoparticle Self-Assembly at the Interface of Liquid Crystal Droplets. Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 5297−5302. 6500

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501

Article

ACS Nano

(24) Londono-Hurtado, A.; Armas-Perez, J. C.; Hernandez-Ortiz, J. P.; de Pablo, J. J. Homeotropic Nano-Particle Assembly on Degenerate Planar Nematic Interfaces Films and Droplets. Soft Matter 2015, 11, 5067−5076. (25) Ondris-Crawford, R.; Boyko, E. P.; Wagner, B. G.; Erdmann, J. H.; Ž umer, S.; Doane, J. W. Microscope Textures of Nematic Droplets in Polymer Dispersed Liquid Crystals. J. Appl. Phys. 1991, 69, 6380.

(2) Martinez, A.; Mireles, H. C.; Smalyukh, I. I. Large Area Optoelastic Manipulation of Colloidal Particles in Liquid Crystals Using Photoresponsive Molecular Surface Monolayers. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 20891−20896. (3) Moreno-Razo, J. A.; Sambriski, E. J.; Abbott, N. L.; HernandezOrtiz, J. P.; de Pablo, J. J. Liquid-Crystal-Mediated Self-Assembly at Nanodroplet Interfaces. Nature 2012, 485, 86−89. (4) Araki, T.; Buscaglia, M.; Bellini, T.; Tanaka, H. Memory and Topological Frustration in Nematic Liquid Crystals Confined in Porous Materials. Nat. Mater. 2011, 10, 303−309. (5) Nych, A.; Ognysta, U.; Skarabot, M.; Ravnik, M.; Zumer, S.; Musevic, I. Assembly and Control of 3D Nematic Dipolar Colloidal Crystals. Nat. Commun. 2013, 4, 1489. (6) Hung, F. R.; Guzmán, O.; Gettelfinger, B. T.; Abbott, N. L.; de Pablo, J. J. Anisotropic Nanoparticles Immersed in a Nematic Liquid Crystal: Defect Structures and Potentials of Mean Force. Phys. Rev. E 2006, 74, 011711. (7) Tasinkevych, M.; Silvestre, N. M.; Telo da Gama, M. M. Liquid Crystal Boojum-Colloids. New J. Phys. 2012, 14, 073030. (8) Škarabot, M.; Muševič, I. Direct Observation of Interaction of Nanoparticles in a Nematic Liquid Crystal. Soft Matter 2010, 6, 5476. (9) Gharbi, M. A.; Nobili, M.; Blanc, C. Use of Topological Defects as Templates to Direct Assembly of Colloidal Particles at Nematic Interfaces. J. Colloid Interface Sci. 2014, 417, 250−255. (10) Yoshida, H.; Asakura, K.; Fukuda, J.; Ozaki, M. ThreeDimensional Positioning and Control of Colloidal Objects Utilizing Engineered Liquid Crystalline Defect Networks. Nat. Commun. 2015, 6, 7180. (11) Muševič, I.; Škarabot, M. Self-Assembly of Nematic Colloids. Soft Matter 2008, 4, 195−199. (12) Musevic, I. S. M.; Tkalec, U.; Ravnik, M.; Zumer, S. TwoDemensional Nematic Colloidal Crystals Self-Assembled by Topological Defects. Science 2006, 18, 954−958. (13) Murray, B. S.; Pelcovits, R. A.; Rosenblatt, C. Creating Arbitrary Arrays of Two-Dimensional Topological Defects. Phys. Rev. E 2014, 90, 052501. (14) Ohzono, T.; Fukuda, J. Zigzag Line Defects and Manipulation of Colloids in a Nematic Liquid Crystal in Microwrinkle Grooves. Nat. Commun. 2012, 3, 701. (15) Miller, D. S.; Carlton, R. J.; Mushenheim, P. C.; Abbott, N. L. Introduction to Optical Methods for Characterizing Liquid Crystals at Interfaces. Langmuir 2013, 29, 3154−3169. (16) Eskandari, Z.; Telo da Gama, M. M.; Ejtehadi, M. R.; Silvestre, N. M. Particle Selection through Topographic Templates in Nematic Colloids. Soft Matter 2014, 10, 9681−9687. (17) Silvestre, N. M.; Liu, Q.; Senyuk, B.; Smalyukh, I. I.; Tasinkevych, M. Towards Template-Assisted Assembly of Nematic Colloids. Phys. Rev. Lett. 2014, 112, 225501. (18) Li, X.; Armas-Perez, J. C.; Martinez-Gonzalez, J. A.; Liu, X.; Xie, H.; Bishop, C.; Hernandez-Ortiz, J. P.; de Pablo, J. J.; Nealey, P. F. Directed Self-Assembly of Nematic Liquid Crystals on Chemically Patterned Surfaces: Morphological States and Transitions. Soft Matter 2016, 12, 8595−8605. (19) Tomar, V.; Roberts, T. F.; Abbott, N. L.; Hernandez-Ortiz, J. P.; de Pablo, J. J. Liquid Crystal Mediated Interactions between Nanoparticles in a Nematic Phase. Langmuir 2012, 28, 6124−6131. (20) Stewart, D.; Imrie, C. Synthiesis and Characterization of SpinLabelled and Spin-Probed Side-Chain Liquid Crystal Polymers. Polymer 1996, 37, 3419−3425. (21) Armas-Perez, J. C.; Hernandez-Ortiz, J. P.; de Pablo, J. J. Liquid Crystal Free Energy Relaxation by a Theoretically Informed Monte Carlo Method Using a Finite Element Quadrature Approach. J. Chem. Phys. 2015, 143, 243157. (22) Armas-Perez, J. C.; Londono-Hurtado, A.; Guzman, O.; Hernandez-Ortiz, J. P.; de Pablo, J. J. Theoretically Informed Monte Carlo Simulation of Liquid Crystals by Sampling of Alignment-Tensor Fields. J. Chem. Phys. 2015, 143, 044107. (23) Lubensky, T. C. Molecular Description of Nematic Liquid Crystals. Phys. Rev. A: At., Mol., Opt. Phys. 1970, 2, 2497−2514. 6501

DOI: 10.1021/acsnano.7b03641 ACS Nano 2017, 11, 6492−6501