SCIENCE ADVANCES | RESEARCH ARTICLE COLLOIDS

Opto-thermophoretic assembly of colloidal matter Linhan Lin,1,2 Jianli Zhang,3 Xiaolei Peng,2 Zilong Wu,2 Anna C. H. Coughlan,3 Zhangming Mao,4 Michael A. Bevan,3 Yuebing Zheng1,2* Colloidal matter exhibits unique collective behaviors beyond what occurs at single-nanoparticle and atomic scales. Treating colloidal particles as building blocks, researchers are exploiting new strategies to rationally organize colloidal particles into complex structures for new functions and devices. Despite tremendous progress in directed assembly and self-assembly, a truly versatile assembly technique without specific functionalization of the colloidal particles remains elusive. We develop a new strategy to assemble colloidal matter under a lightcontrolled temperature field, which can solve challenges in the existing assembly techniques. By adding an anionic surfactant (that is, cetyltrimethylammonium chloride), which serves as a surface charge source, a macro ion, and a micellar depletant, we generate a light-controlled thermoelectric field to manipulate colloidal atoms and a depletion attraction force to assemble the colloidal atoms into two-dimensional (2D) colloidal matter. The general applicability of this opto-thermophoretic assembly (OTA) strategy allows us to build colloidal matter of diverse colloidal sizes (from subwavelength scale to micrometer scale) and materials (polymeric, dielectric, and metallic colloids) with versatile configurations and tunable bonding strengths and lengths. We further demonstrate that the incorporation of the thermoelectric field into the optical radiation force can achieve 3D reconfiguration of the colloidal matter. The OTA strategy releases the rigorous design rules required in the existing assembly techniques and enriches the structural complexity in colloidal matter, which will open a new window of opportunities for basic research on matter organization, advanced material design, and applications. INTRODUCTION

Colloidal particles exhibit collective behaviors beyond their individual properties when the particles are built into colloidal matter (1). For example, colloidal matter, which consists of particles with sizes comparable to or below optical wavelength, has been explored as photonic crystals and meta-materials with unique properties such as optical bandgap and negative refractive index. Rationally designed assemblies of colloidal atoms, such as plasmonic superstructures (2, 3), Janus particle clusters (4–7), and chiral clusters (8), feature various emerging physical and chemical performances. However, challenges in the precise engineering of attractive interactions to induce bonding between colloidal atoms have limited the construction of colloidal matter of any desired properties. Both intrinsic interparticle attractive force and external attractive force have been exploited in current techniques for colloidal bonding. The former includes electrostatic attraction (9) and hydrophobic attraction (4, 10, 11). The latter includes depletion attractive interaction by mixing colloidal atoms with nonadsorbed polymers or small particles (12). These techniques feature high throughput but with limited structural configurations. Much effort has been exerted to enrich the configurations of colloidal matter. For example, patchy colloids can achieve directional bonding to break symmetry in colloidal superstructures. Patchy colloids include colloidal atoms regionselectively modified with organic or biological molecules (for example, DNA) (13–20) and those with modified surface geometry (21, 22). Directed motion of active colloidal atoms under external fields, including electric fields (23, 24), magnetic fields (5, 8, 25), and electromagnetic fields (26, 27), has been further explored to achieve colloidal assembly in a controllable and programmable manner. However, manipulation 1 Department of Mechanical Engineering, University of Texas at Austin, Austin, TX 78712, USA. 2Materials Science & Engineering Program and Texas Materials Institute, University of Texas at Austin, Austin, TX 78712, USA. 3Department of Chemical and Biomolecular Engineering, Johns Hopkins University, Baltimore, MD 21218, USA. 4 Department of Engineering Science and Mechanics, Pennsylvania State University, University Park, PA 16802, USA. *Corresponding author. Email: [email protected]

Lin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

Copyright © 2017 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution NonCommercial License 4.0 (CC BY-NC).

via external fields is highly material-dependent and only works for colloidal particles with specific electric, magnetic, or optical properties, limiting its broader applications to general colloidal particles. With an ever-increasing interest in enriching the structures of colloidal matter for new properties and in understanding the factors that determine its structures and dynamics, a versatile on-demand technique that can construct colloidal matter of arbitrary structures with colloidal atoms of a wide range of sizes, materials, and surface chemistry is highly desired. Developing such a technique will require progress in both colloidal manipulation and colloidal bonding. Optical tweezers have proved effective in colloidal manipulation (28). However, trapping and manipulation of nanoscale colloids and anisotropic particles with optical tweezers are challenging. Specifically, trapping of metal nanoparticles via optical tweezers requires even higher optical power because of the optical heating of trapped particles and possible light-particle coupling. Plasmonic tweezers can capture nanoscale particles using the plasmon-enhanced optical force but lack the capability to perform dynamic manipulation (29). We have also reported the reversible assembly of plasmonic nanoparticles using plasmon-enhanced thermophoresis, whereas the configuration of the assemblies was not controllable in building colloidal superstructures. To overcome the limitations of current methods, we developed a new strategy, termed opto-thermophoretic assembly (OTA), to manipulate and construct colloidal matter based on ionic depletants under a light-controlled temperature field. Because of its low-power operation, wide range of colloidal sizes and materials, and flexibility in orientation control of anisotropic particles, the OTA strategy allows a versatile assembly of colloidal matter with diverse compositions, configurations, and bonding lengths and strengths.

RESULTS

General concept of OTA The general concept of OTA strategy is shown in Fig. 1. Colloidal atoms of different sizes and materials are dispersed in water (solvent), serving 1 of 9

SCIENCE ADVANCES | RESEARCH ARTICLE

Fig. 1. Schematic illustration of the concept of OTA. (A) Dispersion of colloidal atoms in solvents. (B) CTAC surfactants adsorbed on the surface of the colloidal particles form molecular double layers, generating a positive and hydrophilic surface. Meanwhile, the CTAC molecules self-assemble into CTAC micelles with high surface charge density. (C) Manipulation of the colloidal atoms with the light-directed dynamic temperature field, and bonding of the closely positioned colloidal atoms with depletion attraction. (D) Construction of colloidal atoms into one-dimensional (1D) and 2D assemblies.

as building blocks (Fig. 1A). The ionic surfactant cetyltrimethylammonium chloride (CTAC), whose concentration is above its critical micelle concentration (ccmc ; 0.13 to 0.16 mM), was added to the colloidal dispersion to enable controllable optothermal response of the colloidal atoms. The CTAC molecules adsorb onto the colloidal particles through electrostatic interaction and/or hydrophobic interaction to obtain positively charged surfaces (fig. S1), with the positive and hydrophilic head groups pointing to the solvent. Meanwhile, above ccmc, surfactant molecules form positively charged micelles with a hard sphere radius (without consideration of the electrostatic interaction) of several nanometers (Fig. 1B). Although the micelles are much larger than common ions, their high charge density makes it possible for them to serve as “macro ions.” On the other hand, the micelles are significantly smaller than the colloidal atoms studied here, making them a natural choice as “micellar depletants.” To optically manipulate the colloidal atoms, we used a continuous-wave laser and an Au thin film (porous or continuous) as the optothermally responsive substrate to convert photon energy to thermal energy (Fig. 1C). Spatial separation of CTAC micelles and counterions creates a localized electric field to confine the colloidal atoms at the thermal hot spot. The colloidal atoms are further driven dynamically by steering the laser beam. The optical intensity that we typically use to manipulate colloidal atoms of different sizes is 0.08 mW/mm2, which is two to three orders of magnitude lower than that of optical tweezers. At the laser beam location, micellar depletion occurs in closely positioned colloidal atoms, leading to their bonding for directed assembly. Using dynamic control of the laser beam, we can arrange and assemble the colloidal atoms into different structures, as shown in Fig. 1D. Optical trapping and dynamic manipulation of colloidal atoms under a light-controlled thermoelectric field Under an externally controlled temperature gradient, colloidal particles (or molecules) will drift from hot to cold regions or from cold to hot regions, a process that is usually described as thermophoresis or the Soret effect. The drift velocity is given by v T = − D T∇T, where D T is the thermophoretic mobility and ∇T is the temperature gradient. At steady state, the concentration gradient of the colloidal particles is written as ∇c = − cS T∇T, where c is the particle concentration and S T = D T/D is the Soret coefficient (D is the diffusion coefficient). One of the underlying physical mechanisms of thermophoresis is the thermoelectric effect (30–32). The different S T values of the positive and negative ions arising from their different sizes and solvation Lin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

energies lead to the spatial separation of these ions under the temperature gradient, generating a thermoelectric field to drive the charged colloids along the temperature gradient. Here, we demonstrate that the CTAC micelles can be used as positive macro ions to create a thermoelectric field for optical trapping of colloidal atoms. Figure 2A shows the working principle of optical trapping under a thermoelectric field. A green laser is directed from the bottom of a porous Au film for optical heating. The simulated temperature gradient along the z axis irradiated by a laser beam, which is 2 mm in diameter and 0.253 mW in optical power, is embedded in the scheme of Fig. 2A. A maximum temperature gradient of −1.8 × 107 K/m is obtained. Under the temperature gradient, both the CTAC micelles and Cl − ions are driven to the cold region. By treating the CTAC micelles as macro ions, we calculate the thermoelectric field using the equation (32)

ET ¼

kB T∇T e

∑ Zi ni STi i ∑ Zi2 ni

ð1Þ

i

where kB is the Boltzmann constant, e is the elemental charge, i is the ionic species (that is, CTAC micellar ions or Cl− ions), and ni and STi are the corresponding ionic concentration and Soret coefficient. The extremely small ST value of Cl− ions (7.8 × 10−4 1/K) gives a positive sign of ∑ Zi ni STi . Thus, the electric field has the same i

direction as ∇T and points from cold to hot regions, which leads to the trapping of the positively charged colloidal atoms at the thermal hot spot. The trapping velocity of the colloidal particles can be calculated using the Helmholtz-Smoluchowski electrophoretic mobility ez/h (33) v ¼ eET z=h

ð2Þ

where e is the dielectric constant of the solvent, h is the viscosity, and z is the z potential of the particles. To illustrate the role of the thermoelectric field in optical trapping, we carried out a control experiment using both anionic surfactant (that is, SDS) and nonionic surfactant (that is, Triton X-100). We observed the repelling of PS beads, instead of trapping by the laser beam, using 10 mM SDS. The use of Triton X-100 led to neither trapping nor repelling of the particles. Instead, 2 of 9

SCIENCE ADVANCES | RESEARCH ARTICLE

Fig. 2. Colloidal manipulation with a light-directed thermoelectric field. (A) Working principle of colloidal trapping. The black arrows indicate the trapping force. The temperature gradient field along the z axis is calculated when a laser beam with a diameter of 2 mm and an optical power of 0.253 mW is incident onto the substrate. (B) Top: Optical images of the parallel trapping of polystyrene (PS) beads of different sizes. Scale bar, 10 mm. Bottom: Schematic of colloidal trapping with a light-controlled temperature field. (C to E) Measured escape velocities of the trapped colloidal atoms as a function of optical intensity at variable CTAC concentrations. The diameters of the PS colloidal atoms are 2 mm (C), 0.96 mm (D), and 500 nm (E), respectively. The purple lines indicate the maximum temperature gradient along the r axis [as shown in (A)] as a function of optical intensity.

the particles illuminated by the laser beam went upward following the thermal convective flow, indicating the absence of the thermoelectric field (movie S1). To demonstrate the parallel manipulation of colloidal atoms, we used a digital micromirror device (DMD) to generate and project multiple laser beams on the substrate to simultaneously capture and arrange PS beads with diameters of 2 and 0.96 mm in designed patterns, as shown in Fig. 2A. With the light-controlled temperature field, dynamic manipulation of the PS beads is achieved (see fig. S2 and movies S2 and S3). This provides the possibility of manipulating colloidal atoms in a programmable manner. Furthermore, we demonstrate the orientational control of nonspherical particles. We fabricated the anisotropic PS particles using microsphere lithography (see Materials and Methods and fig. S3 for details). We used the DMD to create a 1D optical image to trap and align the anisotropic PS particles (fig. S4). By rotating the 1D optical image, we have achieved the rotation of the anisotropic particles. To further evaluate trapping capability, we trapped single colloidal particles using a laser beam and introduced a Stokes drag force to release the particles by setting a certain value for the moving velocity of the motorized sample stage. The critical velocity at which the maxiLin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

mum trapping force of the particle is balanced by the drag force is defined as escape velocity (see also Materials and Methods for details). It can be estimated using the equation qETm ¼ 6phave fðhÞ

ð3Þ

where q is the surface charge of the particles, E Tm is the thermoelectric field at the maximum in-plane temperature gradient, h is the viscosity of the solvent, a is the radius of the particle, ve is the escape velocity, and f(h) is the correction factor that accounts for the hydrodynamic boundary effect, which is a function of the particle-substrate distance h. The left term is the maximum trapping force in the thermoelectric field, and the right term is the fluidic drag force. Figure 2 (C to E) summarizes the measured escape velocities of the trapped PS beads as a function of optical intensity and CTAC concentration. An increase in optical intensity increases the escape velocity because of the increased temperature gradient ∇T (Fig. 2, C to E, purple lines). An increase in CTAC concentration increases the z potential of the PS beads but decreases the Soret coefficient of the CTAC micelles. The former increases the electrophoretic mobility of the colloidal atoms, whereas 3 of 9

SCIENCE ADVANCES | RESEARCH ARTICLE the latter decreases the intensity of the electric field, leading to a maximum escape velocity of ~1 mM CTAC. Surprisingly, the escape velocity and trapping stability significantly increase when the size of the trapped particles is decreased, which is attributed to the increased temperature gradient around the smaller particles (fig. S5). Therefore, a light-controlled thermoelectric field can manipulate nanoscale particles at a much lower optical power than optical tweezers. Assembly of colloidal matter using OTA Accompanied by the manipulation of colloidal atoms, an attraction force between the colloidal atoms is induced to bond them together.

The bonding mechanism is shown in Fig. 3A. Without any external energy, both the CTAC micelles and the colloidal atoms are dispersed in the solvent in their free forms. Spontaneous self-assembly of the colloidal atoms may occur if the depletants are driven outside the depletion region by thermal perturbations and entropy. However, the process of bonding and assembly is slow. When the laser beam illuminates two colloidal atoms (Fig. 3A, middle), laser heating causes thermophoresis. Under the temperature field, the CTAC micelles migrate to the cold region, and the colloidal atoms migrate to the hot region with a trapping force FT. As a result, thermophoresis depletes the micelles at the particle-particle spacing with a depletion volume

Fig. 3. Physical principle of bonding in OTA. (A) Schematics of the depletion attraction between two colloidal atoms, which is enhanced by the light-controlled temperature field. Middle: The green arrows indicate that thermophoresis drives the CTAC micelles from hot to cold regions. FT is the trapping force, Fvdw is the van der Waals pp ðrÞ is the depletion volume, which is indicated with dashed circles, and DP is the osmotic pressure difference force, and Fe is the electrostatic repulsive force. Right: VEV between the bulk region (cold region) and the depletion region (hot region). In all panels, the gray rings around the micelles indicate an effective depletion size. (B to D) Calculated interaction potential between two PS beads and optical images of colloidal superstructures of beads of different diameters: 2 mm (B), 0.96 mm (C), and 500 nm (D). In (C), the optical images (from left to right) are recorded, where the distances between the focus plane of the 100× objective [numerical aperture (NA), 0.9] and the substrate are 12.275, 13.125, and 13.950 mm, respectively. Scale bars, 5 mm (B) and 2 mm (C and D). Lin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

4 of 9

SCIENCE ADVANCES | RESEARCH ARTICLE pp

VEV ðrÞ and an osmotic pressure difference DP, which binds the colloids together with the depletion attraction force. When the laser is turned off (Fig. 3A, right), the thermoelectric field disappears. The bonding between the colloidal atoms can be maintained if the balance pp pp among depletion potential [UD ðrÞ ¼ DPVEV ðrÞ], van der Waals interaction, and electrostatic repulsive interaction produces an attractive potential well exceeding several kBT. To elucidate the bonding between the colloidal atoms, we calculated the total interaction potential between two PS beads at various CTAC concentrations, including electrostatic interaction pp pp potential UE ðrÞ, van der Waals interaction potential UV ðrÞ, and pp depletion attraction potential UD ðrÞ (34–39). To predict accurate interaction potentials, we directly measured interaction potentials between silica particles and silica wall by total internal reflection microscopy (TIRM; see Materials and Methods for details) and fitted theoretical models to the measurement results. The particle-particle interaction of silica particles was obtained using the particle-particle excluded volume term. The fitted parameters were used to calculate the particle-particle interaction potential of PS beads, which have Hamaker constants (40) and z potentials similar to those of the silica particles. The theoretical model has been proven to work for the micelle depletion interaction measured by TIRM (37). The measurements and models are detailed in note S1. We introduce the micelle-particle electrostatic interaction with an effective micellar radius L and the compressibility factor of the micelles with a depletion fraction b to calpp culate UD ðrÞ. The accuracy of the theoretical models was further verified by Monte Carlo simulations of the TIRM experiments, which produced a quantitative agreement of both the interaction potentials and 2D crystallization behavior with the experimental results (note S1). This model only works when the size of the micellar depletants is significantly smaller than that of the colloidal atoms. In addition, we have also excluded other possible bonding mechanisms in CTAC surfactants, as discussed in note S2. The calculated results for PS beads of different sizes are summarized in Fig. 3 (B to D). A decreased minimum interaction potential occurs at a shorter bonding length when the CTAC concentration is increased, which increases the osmotic pressure difference DP and the ionic strength. This provides opportunities to tune the bonding length and bonding strength by changing the CTAC concentration. The thresholds of CTAC concentration for stable colloidal assemblies are 8, 15, and 40 mM for colloidal sizes of 2 mm, 0.96 mm, and 500 nm, respectively. For 2-mm PS beads (Fig. 3B), stable colloidal bonding occurs when the CTAC concentration is ~10 mM. The bonding becomes stronger with a shorter bonding length when the CTAC concentration is further increased. Using the light-controlled colloidal manipulation and bonding, we have demonstrated the construction of the thermodynamically stable, close-packed colloidal assembly of 2-mm PS beads in a 10 mM CTAC solution, as shown in the left panel of Fig. 3B. The assembly process via a single laser beam is demonstrated in movie S4. The colloidal structure remained after the laser was turned off. Because the micellar depletants cannot diffuse into the particle-particle gap, which is smaller than the effective sizes of the CTAC micelles, the depletion attraction interaction can be maintained when the temperature gradient disappears. When balance is achieved, the interaction potential after optical heating is the same as that without optical heating. The substrateparticle interaction potential, which is about two times higher than the particle-particle interaction potential, should also be considered for the stability of the colloidal matter. At 10 mM, the 2D assembly can be further manipulated by the light-controlled thermoelectric field. Lin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

For example, we used a single laser beam for translation of the 2D assembly and a 1D optical image to rotate the 2D assembly (movie S5). It is also possible to assemble a thermodynamically unstable colloidal structure. To improve the stability of the colloidal structures, we increased the CTAC concentration for stronger particle-particle and particle-substrate interactions. In the middle and right panels of Fig. 3B, we demonstrate the assembly of 2-mm PS beads into a 1D chain using a 1D optical image (see also fig. S6 and movie S6) and into a 2D chiral structure using a single beam (fig. S7), respectively. As discussed in note S1, the interaction potential is highly dependent on the size of the colloidal atoms. For smaller PS beads (Fig. 3, C and D), the critical CTAC concentration for stable colloidal bonding is increased. We built a heptamer with 0.96-mm PS beads at a CTAC concentration of 20 mM. As shown in the right panels of Fig. 3C, the heptamer of colloidal atoms exhibits light transmission and focusing properties, where the observed color of transmitted light depends on the objective (or detector) location. The optical transmission characteristics of the heptamer arise from the interaction between the spherical PS beads, which can be treated as microlenses. In note S3, we decompose the recorded optical images with different color channels (blue, green, and red). When the substrate-objective distance is 12.275 mm, an enhanced transmission is observed in the blue channel, whereas no obvious enhanced transmission in the green channel and a supressed transmission in the red channel are observed, which leads to a blue color in the optical image that is composed of all the three channels. When the substrate-objective distance is increased, the wavelength of the extracted light makes a red shift with an enhanced transmission in the green and red regions, leading to intense green and orange colors through the PS beads. The improved light extraction efficiency will find potential applications in photovoltaics and flat optics. To demonstrate the bonding between nanoscale colloidal atoms, we assembled 500-nm PS beads into various thermodynamically unstable colloidal superstructures in a 50 mM CTAC solution, as shown in the right panels of Fig. 3D. To better understand the stability of colloidal superstructures, we built a heptamer with 2-mm PS beads in different CTAC concentrations and examined the time response after the laser was turned off, as shown in fig. S8. At a low CTAC concentration (below 5 mM), the colloidal structure changed at 9 s because the interaction potential was too shallow and the thermal fluctuation could easily change the configuration of the colloidal superstructure. An increase in CTAC concentration improves the bonding strength between the colloidal atoms, as shown in Fig. 3B, and the colloidal superstructure becomes more stable. At 8 mM, the heptamer can be maintained for more than 2 hours without reconfiguration. We further show the time response of a thermodynamically unstable colloidal superstructure (fig. S9), which was assembled in a 20 mM CTAC solution. Because of the strong particle-particle and particle-substrate interactions, the superstructure was very stable (that is, without significant structural change after 2 hours). Furthermore, we demonstrate that the construction of colloidal superstructures is highly repeatable because of the precise control of colloidal atoms under the light-controlled thermoelectric field. As an example, we carried out multiple experiments during which we built the same type of thermodynamically unstable colloidal superstructures six times. As shown in fig. S10, the superstructures show excellent repeatability. Diversity of colloidal matter built via OTA To examine the general applicability of OTA, we built diverse colloidal superstructures of complex configurations using colloidal atoms of a 5 of 9

SCIENCE ADVANCES | RESEARCH ARTICLE wide range of materials and sizes. Figure 4 (A to C) shows a 1D hybrid chain, a 2D hybrid lattice, and a 2D double-layer “Saturn ring” built from PS beads that are 2 and 0.96 mm in diameter (see also fig. S11). Beyond spherical particles, we also assembled nonspherical particles into 2D colloidal matter. One of the challenges in the assembly of anisotropic particles is the precise orientational control. We overcome this challenge by using the 1D optical image (fig. S4). In Fig. 4D, we used a single 2-mm Gaussian laser beam to assemble a heptamer with 2-mm PS beads, and a 1D optical image to trap, align, and assemble

the anisotropic PS particles. The tips of the anisotropic particles were controlled to point toward the center of the heptamer. A CTAC concentration of 15 mM was chosen for stable bonding among the particles. Another challenge in the assembly of anisotropic particles arises from the different curvature radii at different locations of the anisotropic particles. As shown in Fig. 3 (B to D), the interaction potential highly depends on the radius of the colloidal atoms, indicating that the critical CTAC concentration depends on the orientation of the anisotropic particles when bonding. In fig. S12, we built the

Fig. 4. General applicability of the OTA strategy. (A) 1D hybrid assembly of 2- and 0.96-mm PS beads. The assembly was achieved by adding the 2- and 0.96-mm PS beads alternatively. (B) 2D hybrid square assembly of 2- and 0.96-mm PS beads. Each 0.96-mm PS bead was added at the interface between two 2-mm PS beads and then confined by two other 2-mm PS beads. (C) 2D hybrid assembly of a double-layer Saturn-ring structure with 2- and 0.96-mm PS beads. The structure was built layer by layer from the inside out. (D) 2D hybrid assembly of 2-mm PS beads and anisotropic PS particles. The anisotropic particles were aligned with a 1D optical image and inserted into the heptamer. (E) 3D assembly and manipulation of four 500-nm PS beads. (F) 2D hybrid superlattice of 2-mm PS, 0.96-mm PS, 2-mm silica, and 1-mm silica beads. The assembly process is similar to (B), whereas half of the PS beads were replaced with silica beads. (G) 2D assembly of a star pattern of 2-mm silica beads. The structure was built layer by layer from top to bottom. (H) A heterogeneous dimer consisted of a 500-nm PS bead and a 200-nm Au nanosphere. (I) 2D assembly of a Saturn-ring structure with 2- and 0.96-mm PS beads. The 0.96-mm PS beads were added around the 2-mm PS bead to obtain the structure. (J) 2D assembly of a 0.96-mm PS with three 500-nm PS beads as satellites. The three 500-nm PS beads were added around the 0.96-mm PS at the specific location with a rotation angle of 120°. (K) An Au dimer consisted of two 200-nm Au nanospheres. (A to F, I, and J) Schematics and bright-field optical images of colloidal superstructures. (G) Bright-field and darkfield optical images of the superstructure. (H and K) Schematics and dark-field optical images of the colloidal atoms and dimers. Arrows in the optical images of (K) indicate the use of different incident light polarizations. Scale bars, 5 mm (A to D, F, and G) and 2 mm (E and H to K). Lin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

6 of 9

SCIENCE ADVANCES | RESEARCH ARTICLE hybrid assemblies of spherical and anisotropic particles, where the tips of anisotropic particles point away from the assembly of 2-mm PS beads (see also movie S7). The assembly can be stable at a lower CTAC concentration (that is, 10 mM). Thus, the CTAC concentration should be carefully chosen according to the minimum curvature radius at the bonding sites. To achieve 3D manipulation of the colloidal structures, we incorporated optical scattering force into the thermoelectric field to provide both out-of-plane and in-plane manipulations of the colloidal atoms using a 100× objective (NA, 0.9). As shown in movie S8 and note S4, we aligned the colloidal atoms into an out-of-plane 1D chain using the optical scattering force and steered the location of the laser beam to drag individual colloidal atoms from the chain to modify the configuration. The different configurations of four 500-nm PS beads are summarized in Fig. 4E. Increasing the number of colloidal atoms can improve the challenge of manipulation because multiple colloidal atoms around the laser beam have similar responses to the thermoelectric field. In fig. S13, we demonstrated the 3D manipulation of six 500-nm PS beads. These 3D assemblies can find applications in 3D plasmonic and chiral meta-materials. Because the CTAC surfactants can be widely applied to different colloidal suspensions to modify the surface charges of the colloidal atoms, the OTA strategy can be used to construct colloidal superstructures of various materials. Considering the material-dependent parameters, including the z potential of the colloidal atoms and the Hamaker constant, we slightly modify the CTAC concentration to obtain stable colloidal bonding. In Fig. 4F, we selectively picked up 2-mm PS, 0.96-mm PS, 2-mm silica, and 1-mm silica from the colloidal suspensions to build a 2D hybrid superlattice, where the different beads are arranged in a precisely controlled fashion. Figure 4G shows both bright-field and dark-field optical images of a large-area star pattern of 2-mm SiO2 colloids assembled in a 10 mM CTAC solution. We have also demonstrated the assembly of an Au nanoparticle and a PS bead into a heterogeneous dimer (Fig. 4H), which is also known as an optoplasmonic molecule, where photonic-plasmonic coupling can lead to interesting optical phenomena (41). Furthermore, we study the applicability of the OTA strategy in different colloidal sizes. In Fig. 4 (I and J), we show the Saturn-ring structures with PS beads of different sizes by adding the smaller PS beads surrounding the bigger ones (fig. S14). Finally, we demonstrate the assembly of a metallic dimer with plasmonic coupling using OTA. The precise control of particle-particle gap from several nanometers to tens of nanometers is critical to control the resonant peak location and local electric field intensity in the plasmonic dimers. Figure 4K shows the dark-field optical images of an Au nanosphere dimer assembled in a 50 mM CTAC solution and a single Au nanosphere when the polarization of the incident light is vertical and parallel to the dimer axis, respectively. Enhanced light scattering by the dimer is observed for the parallel incidence due to the near-field plasmonic coupling. Because of tunable bonding length and hence interparticle distance, we expect to precisely control the plasmonic coupling in the colloidal assemblies, which underpins an emerging field of metasurfaces and meta-materials (42). The assembly of sub–100-nm colloidal atoms using OTA remains challenging. Although the optical manipulation of sub–100-nm colloidal atoms can still be achieved using a low-power laser beam, depletion attraction cannot provide sufficient bonding strength for stable assembly because the size of the micellar depletants is not significantly smaller than that of the sub–100-nm colloidal atoms. Lin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

DISCUSSION

In conclusion, we mimic atomic systems to develop the OTA strategy to build diverse colloidal matter based on the opto-thermo-fluidic platform. Because of the thermal transport of ions (including micellar macro ions) under the temperature field, we generate a light-controlled thermoelectric field for versatile manipulation of colloidal atoms and interlocking of the arranged colloidal atoms with depletion attraction. This OTA strategy features a wide range of colloids, leading to complex colloidal superstructures with tunable bonding strength and length between the colloidal atoms. The advantages of the OTA strategy over optical tweezers include its manipulation capability for a wider range of colloidal materials, sizes, and shapes; low-power and invasive operation; tunable bonding length and strength; and flexible control of anisotropic colloids, which markedly enrich the geometry and functions of colloidal superstructures. Specifically, the capability to manipulate and assemble subwavelength colloids with controllable interparticle coupling is extremely important for material applications. Because of their strong interaction with light and unique optical response, the subwavelength colloids provide tremendous opportunities to build colloidal photonic materials with ondemand tunable optical properties. Therefore, OTA will overcome the limitation of optical tweezers in the assembly of functional colloidal nanomaterials. Although current results are demonstrated based on PS, SiO2, and Au nanospheres, the CTAC surfactants can be widely applied to different colloidal suspensions with good stability. Thus, the OTA concept can be easily extended to a bundle of many other colloids such as Si nanoparticles, metal oxide nanoparticles, and nanowires to build diverse functional materials and devices. The throughput of the OTA strategy is lower than that of some self-assembly techniques, which can create large-scale ordered colloidal assemblies. However, OTA has the advantage of providing precise control of colloidal assembly configuration and interparticle interaction, which is critical in tuning coupling between colloidal atoms and thus the functions of the assemblies. In addition, the flexibility in light management and the low-power requirement for the thermophoretic manipulation will allow parallel operation of the colloidal construction with improved throughput and lead to a programmable platform for the construction of designer colloidal matter. On the basis of our current understanding and demonstration, the OTA strategy can be widely applied to colloidal atoms with sizes ranging from the micrometer scale to ~100 nm. Assembly of sub–100-nm colloidal atoms remains challenging because of their relevance to the size of the micellar depletants. Although we have demonstrated that reconfigurable 3D assembly is achievable by incorporating the optical scattering force into the OTA strategy, the configuration of the 3D assemblies and the number of colloidal atoms in the 3D assemblies are still limited. One of the potential strategies to improve the upper limit is to use multiple laser beams to manipulate and assemble colloidal atoms in parallel. The understanding of colloid science and opto-thermo-fluidics in the OTA strategy paves the way toward precisely tuning and reconfiguring the structures and properties of colloidal matter. The versatile construction of optical matter with enriched structures and reconfigurability, in combination with real-space imaging of the motions of individual particles by optical microscopy, will also advance the understanding of the general principles of self-assembly. Eventually, the scale-up of the directed assembly for a large quantity of designer structures and materials will lead to a wide range of technical applications of colloidal matter. 7 of 9

SCIENCE ADVANCES | RESEARCH ARTICLE MATERIALS AND METHODS

Materials and sample preparation The substrates were prepared by depositing 4.5-nm Au thin films on glass slides under thermal deposition (Denton thermal evaporator; base pressure, 9 × 10−6 torr) followed by thermal annealing at 550°C for 2 hours. The PS beads and silica beads (purchased from Bangs Laboratories Inc.) were dispersed in the CTAC solutions at a volume ratio of 1:20,000. The 200-nm Au nanospheres in phosphate-buffered saline (0.1 mM; purchased from Sigma-Aldrich) were centrifuged for 10 min (4500 rpm) and redispersed in the CTAC solutions at a volume ratio of 1:1000. The particle concentration in the colloidal suspensions was diluted to 5 × 105 to 5 × 106 particles/ml to avoid spontaneous aggregation. For the assembly of hybrid colloidal structures, the PS beads, silica beads, or centrifuged Au nanospheres were codispersed in the CTAC solutions. The colloidal dispersions were added to the substrates in a 1-mm-thick chamber. The anisotropic PS microparticles were fabricated by reactive ion etching of close-packed PS microsphere bilayers. The colloidal solution of PS microspheres (9.6%) with a diameter of 5.31 mm was purchased from Bangs Laboratories Inc. The PS microspheres were first concentrated by centrifugation and then dispersed in ethanol at a volume fraction of 33%. The PS microspheres were then spin-coated (1500g, 5 min) on a clean glass substrate to form close-packed bilayers. Before spin coating, the glass substrate was cleaned by ultrasonication in acetone, methanol, and deionized (DI) water. The substrate was further cleaned for 10 min in an ultraviolet ozone system to obtain a hydrophilic surface. Reactive ion etching was performed in a plasma etching system (March Plasma CS170IF) with O2 plasma (100 W, 30 sccm, 600 s). The etched PS microparticles were collected by ultrasonication of the glass substrate in DI water for 2 min. Optical setup A 532-nm diode-pumped solid-state laser (Genesis MX STM-1 W; Coherent) was expanded with a 5× beam expander and projected onto the DMD. The optical images reflected off the DMD were focused onto the substrate, which was placed on a motorized stage. The size of the beam was reduced by 200 times after being relayed by a 1000-mm doublet lens, a 200-mm doublet lens, an infinity-corrected tube lens, and a 40× objective (Nikon; NA, 0.75) in an inverted microscope. For dark-field imaging, an air condenser was used to focus the incident white light onto the sample from the top, and the scattering light was collected from the bottom with a 40× objective. A linear visible polarizer (Thorlabs Inc.) was inserted between the white light source and the condenser to control the incident light polarization. Either the DMD or the motorized stage was used to manipulate the trapped colloidal particles. Measurement of escape velocity The escape velocities of the trapped PS beads were measured with a motorized sample stage with precise velocity control. A single PS bead was trapped with a laser beam irradiated on the Au thin film. A certain value for the moving velocity of the sample stage was set to introduce a drag force on the trapped particle. Escape velocity was defined as the critical velocity at which the maximum trapping force of the particle was balanced by the drag force. Measurement of interaction potentials TIRM was used to directly measure the evolution of colloid-surface depletion attraction at different CTAC concentrations. The experiLin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

mental details and data analysis approaches were reported previously (37). SiO2 colloids with a diameter of 2.1 mm (Bangs Laboratories Inc.) were used in the TIRM measurements. Dispersions of SiO2, CTAC, and DI water were prepared to yield a ~1% SiO2 area fraction, and the sample was sealed by vacuum grease in 5–mm–inside diameter Vinton O-rings (McMaster-Carr) between two cover slips. A 12-bit charge-coupled device camera (ORCA-ER; Hamamatsu) on an upright optical microscope (Axio Imager A1m; Zeiss) was used. The camera was operated in four-binning mode with a 40× objective to yield 28 frames/s and 607 nm/pixel. A 632.8-nm Helium Neon laser (Melles Griot) with a power of 15 mW and a 68° dovetail prism (Reynard Corporation) was used to generate an evanescent wave decay length of 114 nm. Image analysis algorithms coded in FORTRAN were used to track lateral motion in TIRM experiments and to integrate evanescent wave scattering intensity from each colloid. Monte Carlo simulations with and without noise were used to estimate the noise-free potentials corresponding to the direct TIRM measurements, which were previously demonstrated to produce an accurate quantitative agreement between measured potentials and phase behavior in depletion systems (38). Computational fluid dynamics simulations Finite element method numerical simulations (COMSOL Multiphysics) were used to analyze the temperature gradient distribution around the laser beam focused on the substrate-water interface. For simplicity, a 2D axisymmetric model composed of a glass substrate, an Au film, and water was established. Because the physics involved heat transfer in solids and fluids, the predefined conjugate heat transfer and laminar flow models were used. A Gaussian distribution heat source was placed at the substrate-fluid interface to model the heating from the laser beam. Room temperature was set at all other boundaries. SUPPLEMENTARY MATERIALS Supplementary material for this article is available at http://advances.sciencemag.org/cgi/ content/full/3/9/e1700458/DC1 fig. S1. z potential of 2-mm PS beads dispersed in a 10 mM CTAC solution. fig. S2. Parallel trapping and dynamic manipulation of PS beads by the light-controlled temperature field. fig. S3. Scanning electron micrograph of the anisotropic PS particles fabricated with microsphere lithography. fig. S4. Rotation of a single anisotropic PS particle using a 1D optical image. fig. S5. Size dependency of the escape velocity of the trapped PS beads. fig. S6. Trapping, rotation, and bonding of a 1D chain of colloidal atoms. fig. S7. Assembly of a chiral structure of 2-mm PS beads. fig. S8. Time response of a heptamer of 2-mm PS beads at different CTAC concentrations. fig. S9. Time response of a plus pattern of 2-mm PS beads at a CTAC concentration of 20 mM. fig. S10. Multiple experiments on the assembly of the plus pattern with 2-mm PS beads at a CTAC concentration of 20 mM. fig. S11. Assembly of heterogeneous 1D chains of 2- and 0.96-mm PS beads. fig. S12. Hybrid assembly of 2-mm PS beads and anisotropic PS particles. fig. S13. 3D reconfiguration and manipulation of colloidal assemblies of six 500-nm PS beads. fig. S14. Assembly of the Saturn-ring structure with 0.96-mm and 500-nm PS beads. fig. S15. Interaction potential of silica particles. fig. S16. Electrostatic interaction potential between the CTAC micelle and the PS bead as a function of the micelle-particle distance. fig. S17. Light manipulation by a heptamer assembled from seven 0.96-mm PS beads. fig. S18. 3D reconfiguration and manipulation of colloidal assemblies. note S1. Calculation of the interaction potential. note S2. Exclusion of other bonding mechanisms in CTAC surfactants. note S3. Light manipulation with designed colloidal assembly. note S4. 3D reconfiguration and manipulation of colloidal assemblies. movie S1. Repelling of PS beads in SDS solution and thermal convection of PS beads in Triton X-100 solution. movie S2. Dynamic rotation of eight 0.96-mm PS beads. movie S3. Dynamic rotation of a hybrid pattern of 2- and 0.96-mm PS beads.

8 of 9

SCIENCE ADVANCES | RESEARCH ARTICLE movie S4. Assembly process of a 2D thermodynamically stable structure with 2-mm PS beads. movie S5. Manipulation of the 2D assembly of 2-mm PS beads. movie S6. Trapping, rotation, and bonding of the 1D assembly of 2-mm PS beads. movie S7. Assembly of anisotropic PS particles. movie S8. Manipulation of the 3D assembly of 500-nm PS beads.

REFERENCES AND NOTES 1. V. N. Manoharan, Colloidal matter: Packing, geometry, and entropy. Science 349, 1253751 (2015). 2. J. A. Fan, C. Wu, K. Bao, J. Bao, R. Bardhan, N. J. Halas, V. N. Manoharan, P. Nordlander, G. Shvets, F. Capasso, Self-assembled plasmonic nanoparticle clusters. Science 328, 1135–1138 (2010). 3. J. Gong, R. S. Newman, M. Engel, M. Zhao, F. Bian, S. C. Glotzer, Z. Tang, Shape-dependent ordering of gold nanocrystals into large-scale superlattices. Nat. Commun. 8, 14038 (2017). 4. Q. Chen, J. K. Whitmer, S. Jiang, S. C. Bae, E. Luijten, S. Granick, Supracolloidal reaction kinetics of Janus spheres. Science 331, 199–202 (2011). 5. J. Yan, S. C. Bae, S. Granick, Colloidal superstructures programmed into magnetic Janus particles. Adv. Mater. 27, 874–879 (2015). 6. J. Zhang, E. Luijten, S. Granick, Toward design rules of directional Janus colloidal assembly. Annu. Rev. Phys. Chem. 66, 581–600 (2015). 7. L. C. Bradley, K. J. Stebe, D. Lee, Clickable Janus particles. J. Am. Chem. Soc. 138, 11437–11440 (2016). 8. D. Zerrouki, J. Baudry, D. Pine, P. Chaikin, J. Bibette, Chiral colloidal clusters. Nature 455, 380–382 (2008). 9. M. E. Leunissen, C. G. Christova, A.-P. Hynninen, C. P. Royall, A. I. Campbell, A. Imhof, M. Dijkstra, R. van Roij, A. van Blaaderen, Ionic colloidal crystals of oppositely charged particles. Nature 437, 235–240 (2005). 10. Q. Chen, S. C. Bae, S. Granick, Directed self-assembly of a colloidal kagome lattice. Nature 469, 381–384 (2011). 11. A. Sánchez-Iglesias, M. Grzelczak, T. Altantzis, B. Goris, J. Pérez-Juste, S. Bals, G. Van Tendeloo, S. H. Donaldson Jr., B. F. Chmelka, J. N. Israelachvili, L. M. Liz-Marzán, Hydrophobic interactions modulate self-assembly of nanoparticles. ACS Nano 6, 11059–11065 (2012). 12. L. Feng, B. Laderman, S. Sacanna, P. Chaikin, Re-entrant solidification in polymer–colloid mixtures as a consequence of competing entropic and enthalpic attractions. Nat. Mater. 14, 61–65 (2015). 13. R. J. Macfarlane, B. Lee, M. R. Jones, N. Harris, G. C. Schatz, C. A. Mirkin, Nanoparticle superlattice engineering with DNA. Science 334, 204–208 (2011). 14. Y. Wang, Y. Wang, D. R. Breed, V. N. Manoharan, L. Feng, A. D. Hollingsworth, M. Weck, D. J. Pine, Colloids with valence and specific directional bonding. Nature 491, 51–55 (2012). 15. L. Y. T. Chou, K. Zagorovsky, W. C. W. Chan, DNA assembly of nanoparticle superstructures for controlled biological delivery and elimination. Nat. Nanotechnol. 9, 148–155 (2014). 16. M. R. Jones, R. J. Macfarlane, B. Lee, J. Zhang, K. L. Young, A. J. Senesi, C. A. Mirkin, DNA-nanoparticle superlattices formed from anisotropic building blocks. Nat. Mater. 9, 913–917 (2010). 17. M. N. O’Brien, H.-X. Lin, M. Girard, M. Olvera de la Cruz, C. A. Mirkin, Programming colloidal crystal habit with anisotropic nanoparticle building blocks and DNA bonds. J. Am. Chem. Soc. 138 14562–14565 (2016). 18. W. Liu, M. Tagawa, H. L. Xin, T. Wang, H. Emamy, H. Li, K. G. Yager, F. W. Starr, A. V. Tkachenko, O. Gang, Diamond family of nanoparticle superlattices. Science 351, 582–586 (2016). 19. Y. Kim, R. J. Macfarlane, M. R. Jones, C. A. Mirkin, Transmutable nanoparticles with reconfigurable surface ligands. Science 351, 579–582 (2016). 20. J. Guo, B. L. Tardy, A. J. Christofferson, Y. Dai, J. J. Richardson, W. Zhu, M. Hu, Y. Ju, J. Cui, R. R. Dagastine, I. Yarovsky, F. Caruso, Modular assembly of superstructures from polyphenol-functionalized building blocks. Nat. Nanotechnol. 11, 1105–1111 (2016). 21. D. J. Kraft, R. Ni, F. Smallenburg, M. Hermes, K. Yoon, D. A. Weitz, A. van Blaaderen, J. Groenewold, M. Dijkstra, W. K. Kegel, Surface roughness directed self-assembly of patchy particles into colloidal micelles. Proc. Natl. Acad. Sci. U.S.A. 109, 10787–10792 (2012). 22. S. Sacanna, M. Korpics, K. Rodriguez, L. Colón-Meléndez, S.-H. Kim, D. J. Pine, G.-R. Yi, Shaping colloids for self-assembly. Nat. Commun. 4, 1688 (2013).

Lin et al., Sci. Adv. 2017; 3 : e1700458

8 September 2017

23. J. Yan, M. Han, J. Zhang, C. Xu, E. Luijten, S. Granick, Reconfiguring active particles by electrostatic imbalance. Nat. Mater. 15, 1095–1099 (2016). 24. J. Zhang, J. Yan, S. Granick, Directed self-assembly pathways of active colloidal clusters. Angew. Chem. Int. Ed. 55, 5166–5169 (2016). 25. R. M. Erb, H. S. Son, B. Samanta, V. M. Rotello, B. B. Yellen, Magnetic assembly of colloidal superstructures with multipole symmetry. Nature 457, 999–1002 (2009). 26. J. Yeom, B. Yeom, H. Chan, K. W. Smith, S. Dominguez-Medina, J. H. Bahng, G. Zhao, W.-S. Chang, S.-J. Chang, A. Chuvilin, D. Melnikau, A. L. Rogach, P. Zhang, S. Link, P. Král, N. A. Kotov, Chiral templating of self-assembling nanostructures by circularly polarized light. Nat. Mater. 14, 66–72 (2015). 27. S. Srivastava, A. Santos, K. Critchley, K.-S. Kim, P. Podsiadlo, K. Sun, J. Lee, C. Xu, G. D. Lilly, S. C. Glotzer, N. A. Kotov, Light-controlled self-assembly of semiconductor nanoparticles into twisted ribbons. Science 327, 1355–1359 (2010). 28. D. G. Grier, A revolution in optical manipulation. Nature 424, 810–816 (2003). 29. M. L. Juan, M. Righini, R. Quidant, Plasmon nano-optical tweezers. Nat. Photonics 5, 349–356 (2011). 30. S. Duhr, D. Braun, Why molecules move along a temperature gradient. Proc. Natl. Acad. Sci. U.S.A. 103, 19678–19682 (2006). 31. A. Würger, Transport in charged colloids driven by thermoelectricity. Phys. Rev. Lett. 101, 108302 (2008). 32. M. Reichl, M. Herzog, A. Götz, D. Braun, Why charged molecules move across a temperature gradient: The role of electric fields. Phys. Rev. Lett. 112, 198101 (2014). 33. L. Lin, X. Peng, M. Wang, L. Scarabelli, Z. Mao, L. M. Liz-Marzán, M. F. Becker, Y. Zheng, Light-directed reversible assembly of plasmonic nanoparticles using plasmon-enhanced thermophoresis. ACS Nano 10, 9659–9668 (2016). 34. W. B. Russel, D. A. Saville, W. R. Schowalter, Colloidal Dispersions (Cambridge Univ. Press, 1989). 35. A. Tulpar, V. Subramanian, W. Ducker, Decay lengths of double-layer forces in solutions of partly associated ions. Langmuir 17, 8451–8454 (2001). 36. B. V. Derjaguin, N. V. Churaev, V. M. Muller, Surface Force (Springer, 1987). 37. T. D. Iracki, D. J. Beltran-Villegas, S. L. Eichmann, M. A. Bevan, Charged micelle depletion attraction and interfacial colloidal phase behavior. Langmuir 26, 18710–18717 (2010). 38. T. D. Edwards, M. A. Bevan, Depletion-mediated potentials and phase behavior for micelles, macromolecules, nanoparticles, and hydrogel particles. Langmuir 28, 13816–13823 (2012). 39. E. Roelants, F. C. De Schryver, Parameters affecting aqueous micelles of CTAC, TTAC, and DTAC probed by fluorescence quenching. Langmuir 3, 209–214 (1987). 40. M. A. Bevan, D. C. Prieve, Direct measurement of retarded van der Waals attraction. Langmuir 15, 7925–7936 (1999). 41. W. Ahn, S. V. Boriskina, Y. Hong, B. M. Reinhard, Photonic–plasmonic mode coupling in on-chip integrated optoplasmonic molecules. ACS Nano 6, 951–960 (2012). 42. P. Banzer, P. Woźniak, U. Mick, I. De Leon, R. W. Boyd, Chiral optical response of planar and symmetric nanotrimers enabled by heteromaterial selection. Nat. Commun. 7, 13117 (2016). Acknowledgments: We acknowledge the Texas Advanced Computing Center at the University of Texas at Austin for providing high-performance computing resources that have contributed to the research results reported in this paper (http://www.tacc.utexas.edu). Funding: This work was supported by the Beckman Young Investigator Program. Author contributions: L.L. and Y.Z. conceived the idea for this study. L.L., X.P., and Z.W. prepared the materials. L.L. worked on the experiments and collected the data. J.Z. and M.A.B. measured the interaction potentials. L.L., J.Z., and M.A.B. calculated the interaction potentials. A.C.H.C. and M.A.B. conducted the Monte Carlo simulations. Z.M. conducted the computational fluid dynamics simulations. All authors discussed the results and wrote the manuscript. Competing interests: Y.Z., L.L., and X.P. are authors on a patent application related to this paper [Provisional Patent Application 62/462,581 (2017)]. The authors declare no other competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors. Submitted 10 February 2017 Accepted 9 August 2017 Published 8 September 2017 10.1126/sciadv.1700458 Citation: L. Lin, J. Zhang, X. Peng, Z. Wu, A. C. H. Coughlan, Z. Mao, M. A. Bevan, Y. Zheng, Opto-thermophoretic assembly of colloidal matter. Sci. Adv. 3, e1700458 (2017).

9 of 9