FUNCTIONAL SOFT MATERIALS DESIGNED USING COLLOIDS AND INTERFACES AT AND BEYOND EQUILIBRIUM A Dissertation Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy (Chemical Engineering) by Xin Wang December 2022 © 2022 Xin Wang FUNCTIONAL SOFT MATERIALS DESIGNED USING COLLOIDS AND INTERFACES AT AND BEYOND EQUILIBRIUM Xin Wang, Ph. D. Cornell University 2022 ABSTRACT Biological systems rely on complex, dynamic and reconfigurable structures to achieve functions that are essential for life, such as directed motion, transport of energy, synthesis and release of hormones to coordinate metabolism. They also operate beyond equilibrium to exchange information and matter with their environment. Inspired by nature, synthetic materials have been used to recapitulate functions of living systems, including the design of adaptive films/emulsions/vesicles that can reorganize and respond to external stimuli, and active matter that can convert energy from their surrounding environments into mechanical work. The research described in this dissertation explores physical principles for the design of various functional soft matter systems. First, I demonstrate the design of adaptive materials as sensors for amphiphiles and dynamical reaction pathways based on liquid crystals that output optical signals, through a tightly coordinated experimental and theoretical approach. The research is enabled by engineering dynamic and reconfigurable hierarchical emulsions and films via a combination of dynamic interfacial processes, including adsorption and desorption of analytes and the generation of interfacial tension gradients (Marangoni stresses). Second, I describe a progression of dynamic self-propelled motion of emulsion systems dictated by phase separation inside the emulsion droplets. Third, adaptive and active emulsion systems that respond to external analytes or chemical gradients and eject cargo are described. In particular, miniature soft machines that respond to external chemical gradients and autonomously eject cargo at a remote location are discussed. Fourth, I describe how advective flows and interfacial tensions can be designed to collect and sort microplastics. The materials systems developed in this dissertation leverage processes both at and beyond equilibrium in the context of adaptive and active matter. To achieve autonomously functioning soft machines, such as droplets and particles, we move beyond stimuli-responsive strategies, which still require intervention by the user or tight control of the environment, to explore the use of dissipative processes to create functional active matter. The strategy for microplastics removal and recycling presented here also develops foundational knowledge for a set of innovative technologies that have the potential to address societal needs on environmental issues facing our planet. BIOGRAPHICAL SKETCH Xin was born in Datong, Shanxi province, China and finished her primary and middle school education in Datong. Then she moved to Taiyuan and received her high school diploma at the Affiliated High School of Shanxi University in 2012, where she earned First Prize of China National Biology Olympiad (CNBO, 2011), National First Prize of Chinese Mathematical Olympiad (CMO, 2011), National Second Prize of Chinese Physics Olympiad (CPhO, 2011) and National Third Prize of Chinese Chemistry Olympiad (CChO, 2011). Afterwards, she took her BEng in Chemical Engineering at Zhejiang University in 2016, minor in finance, with an Honor from Chu Kochen Honors College (CKC), (Science and Engineering) Mixed Class, and was awarded National Scholar. Xin also led two teams in the two chemical engineering competitions of Zhejiang University in 2015, Unit Operation Design Competition and Chemical Industry Design Competition. Her teams ranked first in both competitions. She wrote a thesis on the synthesis of stimuli-responsive star polymers, supervised by Professor Wenjun Wang and Dr. Yuchen Zhang. In the summer of her junior year, Xin attended the Research Experiences for Undergraduates (REU) program at the University of Wisconsin-Madison (UW-Madison), supervised by Professor Nicholas L. Abbott. She then returned to UW-Madison and began her Ph.D. study with Professor Abbott. After successfully completing her prelim exam during the second year, she moved with Professor Abbott to continue her Ph.D. research on the design of functional colloids at Cornell University, where she completed the remainder of her degree. Besides, Xin is a USA Archery Level 3 NTS Coach. She served in the Archery Club at Cornell (Head Coach) and Ithaca Archery Club (Volunteer Coach). Xin also served as the Secretary of the CBE Women group in the year of 2020-2021. v All humans are related to me. vi ACKNOWLEDGMENTS I would like to express my most sincere gratitude and thanks to my advisor, Professor Nicholas L. Abbott. His guidance, inspiration, passion, responsibility and kindness were greatly appreciated. His inputs were irreplaceable for my graduate study. I thank him for his support in my exploration of new but seemingly unrealistic systems during my graduate research, which turned out to be successful. I developed my understanding of research and Chemical Engineering from our numerous discussions and his comments. This was the most valuable experience, and will continue to inspire me in my future work and study. I would also thank my committee members Professor Susan Daniel for her help and her great encouragement, and Professor Itai Cohen, for his guidance on research communication. I also thank the Abbott group for their collaboration and support. In particular, I learned a lot from Young-Ki Kim, Xiaoguang Wang, Emre Bukusoglu, Chenxuan Wang, Yu Yang, Huaizhe Yu, Hector Fuster, Michael Tsuei, Nanqi Bao, Sangchul Roh and Junghyun Noh. I had the privilege to collaborate with many wonderful groups. I thank Professor Juan J. de Pablo, Ye Zhou, Rui Zhang, Gustavo Pérez and Ali Mozaffari who introduced continuous and atomistic simulation to me. I thank Professor Nathan C. Gianneschi and Hao Sun for their support in polymer science. I appreciate Professor S. Thai Thayumanavan, Ann Fernandez and Jithu Krishna from whom I benefited greatly in organic chemistry. I thank Professor Sarah Hormozi for her patience at our meetings for hydrodynamics. They are ideal collaborators, always responsible and constructive. I wish to thank my parents, grandparents, great grandparents, uncles and aunts who are from “non-STEM” backgrounds, but strongly supported me when I took my chance to learn Chemical Engineering that they seldom knew about, and my other crazy ideas. They are also my exemplars of courage as they never complain but always move forward. And I thank Zhejiang University for having the special program of Chu Kochen vii College that allows students to choose their major(s) freely in the sophomore year, as almost all Universities and programs in China require major confirmation before entering the University. I thank my archery coach Charles Rendleman as an exemplar of focus, commitment, humor and inspiration. I also thank my nice and knowledgeable friends who I always turned to. I thank other people; I benefitted and learned a lot from their dedication even though I don’t know them personally. viii TABLE OF CONTENTS Chapter 1 Introduction and Overview ............................................................................ 1 1.1 Introduction ...................................................................................................................... 1 1.2 Thesis Overview ............................................................................................................... 4 1.3 References ........................................................................................................................ 9 Chapter 2 Literature Review ........................................................................................ 13 2.1 Introduction to Thermotropic Liquid Crystals (LCs) ..................................................... 13 2.2 Optical Analysis of Orientational Ordering of LCs ....................................................... 19 2.3 Balance of Interfacial Tensions between Multiple Phases ............................................. 26 2.4 Marangoni Effect ............................................................................................................ 30 2.5 References ...................................................................................................................... 32 Chapter 3 ...................................................................................................................... 39 Thermally Reconfigurable Janus Droplets with Nematic Liquid Crystalline and Isotropic Perfluorocarbon Oil Compartments .............................................................. 39 1. Introduction ...................................................................................................................... 39 2. Experiments and Simulations ........................................................................................... 44 3. Results and Discussion ..................................................................................................... 48 4. Conclusions ...................................................................................................................... 65 5. Supporting Information .................................................................................................... 66 6. References ........................................................................................................................ 75 Chapter 4 ...................................................................................................................... 83 Reconfigurable Multicompartment Emulsion Drops Formed by Nematic Liquid Crystals and Immiscible Perfluorocarbon Oils ............................................................. 83 1. Introduction ...................................................................................................................... 83 2. Experimental Methods ..................................................................................................... 86 3. Results and Discussion ..................................................................................................... 90 4. Conclusions .................................................................................................................... 109 5. Supporting Information .................................................................................................. 110 6. References ...................................................................................................................... 116 Chapter 5 .................................................................................................................... 124 Active Motion of Multiphase Oil Droplets: Emergent Dynamics of Squirmers with Evolving Internal Structure ........................................................................................ 124 1. Introduction .................................................................................................................... 124 ix 2. Experiments .................................................................................................................... 126 3. Results and Discussion ................................................................................................... 128 4. Conclusions .................................................................................................................... 142 5. Supporting Information .................................................................................................. 143 6. References ...................................................................................................................... 152 Chapter 6 .................................................................................................................... 158 Stimuli-Responsive Liquid Crystal Printheads for Spatial and Temporal Control of Polymerization ............................................................................................................ 158 1. Introduction .................................................................................................................... 158 2. Results and Discussion ................................................................................................... 159 3. Conclusions .................................................................................................................... 169 4. Supporting Information .................................................................................................. 170 5. References ...................................................................................................................... 188 Chapter 7 .................................................................................................................... 194 Self-Timed and Spatially-Targeted Delivery of Chemical Cargo by Chemotactic Self- Propelled Droplets ...................................................................................................... 194 1. Introduction .................................................................................................................... 194 2. Results and Discussion ................................................................................................... 196 3. Conclusions .................................................................................................................... 210 4. Supporting Information .................................................................................................. 211 5. References ...................................................................................................................... 224 Chapter 8 .................................................................................................................... 231 Substitution of Molecular Structure Alters Surface Anchoring of Liquid Crystals ... 231 1. Introduction .................................................................................................................... 231 2. Results and Discussion ................................................................................................... 233 3. Conclusions .................................................................................................................... 244 4. Supporting Information .................................................................................................. 245 5. References ...................................................................................................................... 266 Chapter 9 .................................................................................................................... 271 Optical Fingerprinting of Dynamic Interfacial Reaction Pathways using Liquid Crystals ....................................................................................................................... 271 1. Introduction .................................................................................................................... 271 2. Results and Discussion ................................................................................................... 273 3. Conclusions .................................................................................................................... 291 4. Supporting Information .................................................................................................. 292 x 5. References ...................................................................................................................... 323 Chapter 10 .................................................................................................................. 328 Active Capturing and Sorting of Microplastics Using Liquid Interfaces ................... 328 1. Introduction .................................................................................................................... 328 2. Results and Discussion ................................................................................................... 329 3. Conclusions .................................................................................................................... 340 4. Supporting Information .................................................................................................. 341 5. References ...................................................................................................................... 351 Chapter 11 Summary and Future Directions .............................................................. 355 1. Summary ........................................................................................................................ 355 2. Future Directions ............................................................................................................ 359 xi LIST OF FIGURES Figure 2.1 Nematic liquid crystal as an intermediate phase of matter. ........................ 13 Figure 2.2 Molecular structure of 4′-pentyl-4-biphenylcarbonitrile (5CB) as a typical LC compound. .............................................................................................................. 14 Figure 2.3 Physical properties of nematic LC. ............................................................. 16 Figure 2.4 LC organizations in droplet geometry. ....................................................... 19 Figure 2.5 Birefringence in nematic LCs. .................................................................... 20 Figure 2.6 Michel-Lévy interference color chart. ........................................................ 21 Figure 2.7 Examples of micrographs of LC droplets. .................................................. 23 Figure 2.8 Examples of micrographs of LC films. ....................................................... 24 Figure 2.9 Balance of interfacial tensions gives rise to multiple morphologies of complex emulsions. ...................................................................................................... 26 Figure 2.10 Contact angle of a solid particle at a fluid interface. ................................ 28 Figure 2.11 Contact angle of an immiscible oil droplet placed on the water-air interface. ....................................................................................................................... 29 Figure 2.12 Spontaneous motion of an oil droplet dispersed in micellar solution due to Marangoni effect. ......................................................................................................... 32 Figure 3.1 (A) Sketch of interfacial tensions of a Janus droplet that define (B) a Neumann’s triangle and satisfy γ13 ~ γ23 >> γ12. .......................................................... 40 Figure 3.2 Molecular structures of (A) the mixture of nematogens called E7, (B) perfluorobenzene (FB), (C) perfluorooctanoic acid (PFOA) and (D) sodium dodecyl sulfate (SDS). ............................................................................................................... 43 Figure 3.3 Phase diagram of the FB-E7 mixture .......................................................... 49 Figure 3.4 LC ordering at N-I interfaces of a FB-E7 mixture. ..................................... 51 Figure 3.5 Thermal reconfiguration of one LC droplet comprised of FB-E7 mixture via nematic-isotropic (N-I) phase transition. ...................................................................... 54 Figure 3.6 Optical micrographs showing the influence of surfactant type and concentration on the internal configurations ................................................................ 56 Figure 3.7 Simulation for single-phase nematic droplets with increasing tilt angles at the interface. ................................................................................................................. 59 Figure 3.8 Influence of surfactant type on the morphologies of biphasic E7-FB droplets. ........................................................................................................................ 61 Figure 3.9 Interfacial tensions between nematic-aqueous and isotropic-aqueous interfaces ....................................................................................................................... 64 Figure S3.1 Curvature of the N-I interface. .................................................................. 68 xii Figure S3.2 FB-E7 (CFB=5 % v/v) droplet in 100 µM SDS aqueous solution. ............ 69 Figure S3.3 FB-E7 (CFB=5 % v/v) droplet in 1 mM SDS aqueous solution. ............... 70 Figure S3.4 FB-E7 (CFB=5 % v/v) droplet in 2 mM SDS aqueous solution. ............... 71 Figure S3.5 FB-E7 (CFB=5 % v/v) droplet in 0.5 mM PFOA aqueous solution. ......... 72 Figure S3.6 FB-E7 (CFB=5 % v/v) droplet in 1 mM PFOA aqueous solution. ............ 73 Figure S3.7 A comparison of FB-E7 droplets at (A-F) pH = 3 and (G-L) pH = 7. ...... 74 Figure S3.8 Simulated schematic illustrations (row 2) and polarized light micrographs (row 3) for multi-compartment droplets under the influence of strong anchoring strength at N-I interfaces. ............................................................................................. 75 Figure 4.1 Molecular structures of the nematogens (A) 4′-pentyl-4- biphenylcarbonitrile (5CB), (B) 4-(trans-4-pentylcyclohexyl) benzonitrile (PCH5); (C) perfluorononane (F9) and (D) perfluoroheptane (F7). ................................................. 84 Figure 4.2 Micrographs between parallel polars (left), crossed polars (middle), and schematic illustrations (right) of (A-C) 5CB droplets in a continuous phase of F9 and (D) an F9 droplet dispersed in 5CB at room temperature. (B,C) size-dependence of LC ordering in droplets with radii of 1.4 μm and 0.9 μm, respectively. (E) Single-polar, crossed-polar micrographs and schematic illustration of a PCH5 droplet in F9. The defect denoted by the blue arrow is slightly tilted out of the plane. Scale bars, 10 μm. The double-headed arrows indicate the orientation of the two polarizers. The dash lines show the ordering of 5CB (director). The red dots indicate the defects. ............. 92 Figure 4.3 (A-C) Polarized-light micrographs (left: parallel-polars; middle: crossed- polars) and schematic illustrations (right) showing internal organizations of double emulsions with nematic 5CB shells and F9 cores of various sizes. Droplets are dispersed in glycerol. Scale bars, 10 μm. The dash lines show the ordering of 5CB (director). The red dots indicate the defects. (D) Estimated elastic free energy Fe as a function of the ratio of core radius to droplet radius 10 μm) for droplets with dipolar and quadrupolar symmetries. (E) Schematic illustration shows the internal organization with strong anchoring at both interfaces (perpendicular inner, planar outer). ............................................................................................................................ 97 Figure 4.4 Effect of strength of LC anchoring on the organization of core-shell emulsions with F9 core and 5CB shells (quadrupolar symmetry). ............................ 100 Figure 4.5 Anchoring strength effect on the organizations of a double emulsion with dipolar symmetry. ....................................................................................................... 101 Figure 4.6 Influence of Surfactant Type and Concentration on Morphologies and Internal Configuration of F9-Nematic Emulsion Droplets. ........................................ 103 Figure 4.7 Thermal reconfiguration of multiphase LC emulsion droplets with a perfluorocarbon (F7) core and nematic (HTW) shell. ................................................ 107 Figure S4.1 Time-lapse micrographs of a 5CB droplet (radius 0.9 μm) dispersed in F9. .................................................................................................................................... 111 Figure S4.2 F9 droplets suspended in a continuous nematic 5CB phase forming (A) xiii dimers (denoted by blue arrows) with F9:5CB = 1:100 (volume ratio) and (B) chains of droplet with F9:5CB = 10:100. The observations were made at room temperature. Scale bars: 10 μm. ...................................................................................................... 112 Figure S4.3 Schematic illustrations showing the volumes (blue) of LC .................... 115 Figure S4.4 Parallel-polar, crossed-polar micrographs and schematic illustrations showing the defects (denoted by the blue arrows and the red dots) in emulsions with nematic HTW shells and (A) F9 liquid cores or (B) F9 vapor cores, respectively. Scale bars, 10 μm. ................................................................................................................ 116 Figure 5.1 Morphogenesis and dynamical motion of droplets. .................................. 129 Figure 5.2 Characterization of droplet trajectories. .................................................... 132 Figure 5.3 Dynamics of droplets can be described by the squirmer model. .............. 136 Figure 5.4 Dynamical transitions of droplets. ............................................................ 138 Figure 5.5 Time-lapse micrographs showing the reorganization of the morphology of a biphasic Janus droplet ................................................................................................. 141 Figure S5.1 Preferential extraction of perfluorobenzene (FB) from emulsion droplets prepared from mixtures of FB and E7. ....................................................................... 144 Figure S5.2 Trajectory during which the handedness of the spiralling motion switched sign (denoted by blue arrows). ................................................................................... 145 Figure S5.3 Reorganization of the internal morphology of droplet during the transition from State IV to V dynamical states. .......................................................................... 146 Figure S5.4 (a) Schematic illustration of the phase diagram of the LC+FB oil phase. (b) Velocity of a nematic single phase droplet that comprised of the LC-FB mixture deceases as a function of time (i.e., FB concentration). ............................................. 147 Figure S5.5 Schematic illustration of the phase diagram of the LC+FB oil phase. ... 148 Figure S5.6 Velocity of droplet motion during State III ............................................ 149 Figure S5.7 Reconfiguration of droplet morphology with external concentration gradient of TTAB. ...................................................................................................... 150 Figure 6.1 Molecular structures .................................................................................. 160 Figure 6.2 Design principles for LC printheads that trigger polymerization. ............ 162 Figure 6.3 LC printhead for polymerization-induced self-assembly (PISA). ............ 165 Figure 6.4 Stimulus-triggered dynamic control of polymerization using LC printheads. .................................................................................................................................... 167 Figure S6.1 Zeta potentials for aqueous-dispersed LC droplets used in the experiments reported in this paper. ................................................................................................. 181 Figure S6.2 Snapshots of the release of microdroplets in a mini-well. ...................... 182 Figure S6.3 Permanent surface-confined polymeric film in the presence of cross-linker .................................................................................................................................... 183 xiv Figure S6.4 1H NMR (500 MHz, CDCl3) spectrum of PNIPAm synthesized with the setup in Figure 6.1g-k. ................................................................................................ 184 Figure S6.5 GPC trace of PNIPAm synthesized with the setup in Figure 6.1g-k. ..... 185 Figure S6.6 1H NMR (500 MHz, CHCl3-d6, δ) spectrum of PDMA47 macroCTA. ... 186 Figure S6.7 1H NMR (500 MHz, CHCl3-d6, δ) spectra of PDMA47-b-PDAAmX. ..... 187 Figure 7.1 Design of chemotactic self-propelled delivery system and molecular constituents ................................................................................................................. 195 Figure 7.2 Chemotactic droplet self-propels along a gradient and autonomously releases microcargo at target location to trigger polymerization ............................... 198 Figure 7.3 Stage II: Clustering and coalescence of microcargo inside the LC phase with and without flow. ................................................................................................ 201 Figure 7.4 Polymeric microparticles by release of initiator-containing microcargo from LC carrier droplets during Stage III. .......................................................................... 205 Figure 7.5 Clusters of microcargo within LC film that gain sufficient momentum from the shear flow can overcome the repulsive barrier at the LC interface to escape. ..... 206 Figure 7.6 Chemotactic LC carrier droplet functioning in channels. ......................... 209 Extended Data Figure 7.1 Chemical gradient generated by contact of two aqueous phases ......................................................................................................................... 217 Extended Data Figure 7.2 Flow fields and measurements of microcargo sizes. ........ 219 Extended Data Figure 7.4 ATR-FTIR spectra of (a) NIPAm monomer and (b) PNIPAm synthesized from chemotactic self-propelled LC carrier droplets, respectively. ................................................................................................................ 221 Extended Data Figure 7.5 Flow velocity increased by two folds at the time of release. .................................................................................................................................... 222 Extended Data Figure 7.6 Colloidal forces as a function of cluster size. ................... 223 Extended Data Figure 7.7 Evaluation of stress of inertia. .......................................... 224 Figure 8.1 Distinct alignment of two nematic liquid crystals at interfaces. ............... 235 Figure 8.2 Alignment of 5CB and PCH5 molecules at F9 interfaces obtained from atomistic molecular dynamics simulations. ................................................................ 237 Figure 8.3 Origin of the orientational ordering. ......................................................... 241 Figure 8.4 Temperature dependence of molecular alignment. ................................... 243 Figure S8.1 a,c Polarized light and b,d polscope micrographs using 632 nm light source show retardance for 5CB-F9 and PCH5-F9 interfaces, respectively. ............. 248 Figure S8.2 United Atom representation of 5CB with atoms enumerated. ................ 250 Figure S8.3 United Atom representation of PCH5 with atoms enumerated. ............. 254 Figure S8.4 United Atom representation of F9 with atoms enumerated. ................... 258 Figure S8.5 Densities of 5CB, PCH5 and F9, respectively, as a function of xv temperature. ................................................................................................................ 262 Figure S8.6 Interfacial tensions for 5CB-F9 and PCH5-F9. ...................................... 264 Figure S8.7 Free Energy maps of 5CB-F9 in the isotropic phase. ............................. 265 Figure S8.8 Rapinni-Papoular coefficients (a) and orientational distribution functions (b) for PCH5-F9 mixture at different temperatures. ................................................... 266 Figure 9.1 Molecular structures of (a) tetraalkylammonium-12 (TEA12) and (b) 4- cyano-4’-pentylbiphenyl (5CB). (c) Schematic illustrations and (d) polarized-light micrographs showing that adsorbed TEA12 alters 5CB alignment at the aqueous interface, leading to a switch of optical texture from bright to dark. Red arrows indicate interfacial alignment of 5CB molecules. Scale bars, 100 m. ...................... 273 Scheme 9.1 (a) TEA12 undergoes an amine-triggered cleavage of its charged head group. Molecular structures of (b) N,N,N′-trimethylethylenediamine (TMEN), (c) N,N- dimethylethylenediamine (DMEN), (d) N,N′-dimethylethylenediamine (DMEDA), (e) N-methylethylenediamine (MEDA) and (f) ethylenediamine (EDA). ....................... 275 Table 9.1 Products and their molecular weight formed from TEA12 with an excess amount of different amines, respectively. .................................................................. 276 Figure 9.2 ESI mass spectrum of products from reaction of TEA12 with EDA (1.1 equiv., 4.4 equiv. -NH) in methanol. .......................................................................... 277 Spectra from several scans were added to generate the spectrum shown. Typically, ions carrying 1 positive charge (+H+ or +Na+) were detected. The peaks in the spectrum are consistent with formation of all possible products of the reaction (5a-5e). The measured mass is calculated from the m/z of these ions after assigning the charge states. .......................................................................................................................... 277 Figure 9.3 Micrographs obtained with crossed polarizers for transmitted light show distinct optical responses of LC film to reactions with different amine nucleophiles in the first column (22 mM TMEN, 11 mM DMEN, 5 mM EDA, 11 mM DMEDA and 7.33 mM MEDA, respectively). ................................................................................. 279 Scale bars, 200 m. The inset reveals the two distinct optical patterns with the white lines denoting the key features. .................................................................................. 279 Figure 9.4 Origin of the “type-one” response. ........................................................... 285 Figure 9.5 Dynamic LC-Aqueous Interfacial Tensions in the Presence of Interfacial Reactions. ................................................................................................................... 287 Figure 9.6 Relaxation time of aqueous-LC interfaces in different product suspensions with the dominant product species shown on the bottom axis, indicating that the relaxation processes are associated with the adsorption of different products. .......... 289 Figure S9.1 1H-NMR spectrum of precursor 1. .......................................................... 299 Figure S9.2 13C-NMR spectrum of precursor 1. ......................................................... 300 Figure S9.3 1H-NMR spectrum of precursor 2. .......................................................... 301 Figure S9.4 13C-NMR spectrum of precursor 2. ......................................................... 302 xvi Figure S9.5 1H-NMR spectrum of surfactant 3 (TEA12). .......................................... 303 Figure S9.6 13C-NMR spectrum of surfactant 3 (TEA12). ........................................ 304 Figure S9.7 ESI mass spectrum of products of the reaction of TEA12 with TMEN (4.4 equiv., 4.4 equiv. -NH) in methanol. .......................................................................... 305 Figure S9.8 ESI mass spectrum of products of the reaction of TEA12 with DMEN (2.2 equiv., 4.4 equiv. -NH) in methanol. .......................................................................... 306 Figure S9.9 ESI mass spectrum of products of the reaction of TEA12 with DMEDA (2.2 equiv., 4.4 equiv. -NH) in methanol. ................................................................... 307 Figure S9.10 ESI mass spectrum of products of the reaction of TEA12 with MEDA (1.47 equiv., 4.4 equiv. -NH) in methanol. ................................................................. 308 Figure S9.11 (a-e) ESI mass spectrum of products of the reaction of TEA12 with nucleophiles (1 equiv. -NH) in methanol. .................................................................. 310 Figure S9.12 (a) Schematic illustration of experimental setup. ................................. 311 Figure S9.13 Micrographs under crossed polarizers show optical textures of 5CB films as a function of TEA12 concentration before introducing nucleophiles. ................... 312 Figure S9.14 Polarized light micrographs of three replicated fingerprinting experiments with TMEN. ........................................................................................... 313 Figure S9.15 Polarized light micrographs of three replicated fingerprinting experiments with DMEN. ........................................................................................... 314 Figure S9.16 (a-c) Polarized light micrographs of three replicated fingerprinting experiments with DMEDA. (d) Micrographs obtained using 0.5 equiv. DMEDA (1 equiv. -NH) and surfactant. ........................................................................................ 315 Figure S9.17 Polarized light micrographs of three replicated fingerprinting experiments with MEDA. ........................................................................................... 316 Figure S9.18 (a-c) Polarized light micrographs of three replicated fingerprinting experiments with EDA. (d) Micrographs obtained using 0.25 equiv. EDA (1 equiv. - NH) and surfactant. .................................................................................................... 317 Figure S9.19 Interfacial tensions of 5CB-aqueous interfaces using product suspensions prepared by reacting 5 mM TEA12 with either 22 mM TMEN, 11 mM DMEN, 11 mM DMEDA, 7.33 mM MEDA, or 5.5 mM EDA (4.4 equiv. -NH to surfactant), respectively. ................................................................................................................ 318 Figure S9.20 Polarized light micrographs show adsorption of monomer (1, products from 5 mM TEA12 reacted with 5 mM TMEN). ....................................................... 319 Figure S9.21 Polarized light micrographs show adsorption of dimer (2b, products from 5 mM TEA12 reacted with 2.5 mM DMEN). ............................................................ 319 Figure S9.22 Polarized light micrographs show adsorption of NN’-dimer (3b, products from 5 mM TEA12 reacted with 2.5 mM DMEDA). ................................................. 320 Figure S9.23 Polarized light micrographs show adsorption of trimer (4e, products from 5 mM TEA12 reacted with 1.67 mM MEDA). .......................................................... 320 xvii Figure S9.24 Polarized light micrographs show adsorption of tetramer (5e, products from 5 mM TEA12 reacted with 1.25 mM EDA). ..................................................... 321 Figure S9.25 Polarized light micrographs of LC films during adsorption of products from TEA12-TMEN (4.4 equiv., 4.4 equiv. -NH) reaction ....................................... 322 Figure S9.26 Polarized light micrographs of LC films during adsorption of products from TEA12-DMEN (2.2 equiv., 4.4 equiv. -NH) reaction ....................................... 322 Figure S9.27 Polarized light micrographs of LC films during adsorption of products from TEA12-EDA (1.1 equiv., 4.4 equiv. -NH) reaction ........................................... 322 Figure 10.1 Design of this work: surface liquid extraction to collect microplastics. 331 Figure 10.2 Factors that influence transport of microparticles. .................................. 333 Figure 10.3 Particle sorting based on their interfacial property. ................................ 338 Figure 10.4 Pre-treatment for microplastics in bulk aqueous phases and post- procedure. ................................................................................................................... 340 Extended Data Figure 10.1 Spreading of the oil droplet after decanol was fully absorbed. ..................................................................................................................... 346 Extended Data Figure 10.2 PPG, which spreads across the water surface to form a film, was also used as the carrier. ............................................................................... 347 Extended Data Figure 10.3 Surface extraction process of PPG. ................................ 348 Extended Data Figure 10.4 Particle velocity with decanol as the carrier. .................. 349 Extended Data Figure 10.5 Contact angles and interfacial tensions. ......................... 350 Figure 11.1 Proposed process for collecting and sorting microplastics on surface of a large water body. ........................................................................................................ 360 Figure 11.2 Proposed separation mechanism of surface liquid extraction using a set of fluids with interfacial tension “gradient”. .................................................................. 361 Figure 11.3 Examples of amines found in organisms. ............................................... 363 xviii Chapter 1 Introduction and Overview 1.1 Introduction Intermolecular interactions[1] that are dictated by the chemical nature of molecules underlie mesoscale to macroscale behaviors of materials in and out of equilibrium[2], such as phase separation[3], adhesion[2], balance of interfacial tensions [4,5], electrostatic interaction[6] and molecular self-assembly[7]. Hierarchical organizations that arise from a combination of local equilibrium and global, out-of- equilibrium phenomena involving diffusive and advective transport processes[8–13] are vital in a large range of technologies, including pharmaceuticals and medical diagnostics[14], semiconductors[15], oil recovery[4], renewable energy[16], food processing[17], cosmetics[18], robotics[19] and optical devices[10]. To facilitate the design of hierarchical materials with engineered behaviors, relevant physical principles at the mesoscale that involve colloid physics, organic chemistry, hydrodynamics and polymer science need to be addressed through a combination of experimental and theoretical approaches. Two concepts applicable to soft matter systems are of particular interest, namely, (i) “adaptive materials”, which use physical and chemical stimuli as inputs, perform on-board computation using interactions programmed by the material, and then output complex responses[20,21], and (ii) “active matter”, which can convert energy from surrounding environments into mechanical work[13,22–24]. “Adaptive” soft materials have received considerable attention because complex yet predictable intermolecular and colloidal interactions can be programmed into soft materials systems via control of chemical composition and mesoscale organizations, which function in response to external stimuli. Synthetic structured oils, such as thermotropic liquid crystals (LCs), have been used to prepare nematic films and emulsion[6,11,25–27], leading to a range of rich physics controlled by elastic strain and 1 conservation of topological defect charge[28]. Moreover, LC interfaces formed with an immiscible aqueous phase have been used as sensors for interfacial adsorbates because encoded interactions between adsorbates and LCs define alignments of LCs at boundaries that propagate through the entire LC organization[10]. Thus, local molecular-level events can be amplified into the macroscale through the hierarchical structure of LC, and generate optical/mechanical outputs due to anisotropic optics and mechanics of LCs. In order to apply these principles, I have designed functional systems based on the responsiveness of LCs: (i) Multiple sensing systems based on multi-compartment LC emulsions and films were designed to explore the coupling of temperature, interfacial adsorbates and interfacial chemical reactions of amphiphiles with LC systems through a combination of experiments and continuum simulations. The systems eventually output complex optical signals. (Chapter 3, 4, 9) (ii) I used a combination of experimental methods to investigate how multiple chemical stimuli, including small organic compounds, biomolecules, multivalent ions, at and beyond equilibrium, can trigger liquid crystals to eject colloidal microcargo. (Chapter 6, 7) (iii) How sub-molecular structure of LC mesogens dictates the orientational ordering at immiscible fluorocarbon interface was explored by both experiments and atomistic simulations (Chapter 8). Living systems rely on processes that are out of thermal equilibrium to carry out functions essential for life, such as directed transport of nutrients, response to stimuli, energy storage and reproduction[29]. More recently, and reminiscent of the dynamics of biological systems that function out of equilibrium (e.g., bacteria, flagellum, kinesin), 2 dissipative processes have been used to design “active matter”[13,30]. For example, spontaneous symmetry breaking and self-propelled “swimming” behaviors of emulsion droplets were found in micellar solutions, which arise from interfacial tension gradients (Marangoni stress) generated by solubilization of oils[30]. Leveraging the advances and insights previously reported, I designed soft systems possessing behaviors that capture functions and essential physics of the far more complicated naturally occurring systems: (i) I explored how compartmentalization influences self-propelled motion of active emulsion droplets was explored. (Chapter 5) (ii) I used non-equilibrium processes associated with sustained, dynamic flows generated by Marangoni stress drive the internal organization of droplets inside self-propelled double emulsions. The fluid flow can eventually lead to delivery of cargo (release of droplets) to a remote location. (Chapter 7) (iii) Combining the knowledge of interfacial tensions and transport via Marangoni stresses, I designed interfacial flows that sweep, concentrate and sort microplastics (colloidal polymeric particles) at liquid interfaces. (Chapter 10) Overall, I integrated theory, simulations from atomistic scale to mesoscale, and experiment in colloid science, hydrodynamics, organic chemistry and polymer science to design hierarchical materials assemblies with engineered behaviors. 3 1.2 Thesis Overview Chapter 2 introduces the reader to (i) liquid crystals (LCs) and relevant characterization methods, (ii) thermodynamics of interfaces between multiple phases and (iii) Marangoni stresses at fluid interfaces. Chapters 3-10 were originally prepared as separate manuscripts for publication and may be read independently. Chapter 11 concludes the dissertation with a summary of key results and future directions. Brief descriptions of Chapters 3-10 are detailed below. LC oils offer the basis of stimuli-responsive LC-in-water emulsions. Although past studies have explored the properties of single-compartment LC emulsions, few studies have focused on complex multicompartment emulsions containing co-existing isotropic and LC domains. Chapter 3 reports that mixtures of perfluorocarbon oils and hydrocarbon mesogens can be used to prepare multi-compartment (Janus) emulsion drops comprising coexisting nematic LC and isotropic oil phases. The droplets exhibit stable spherical shapes with internal Janus-type morphologies that can be tuned widely through changes in temperature or adsorbates. In particular, I provide evidence of preferential adsorption of hydrocarbon or fluorocarbon surfactants on the interfaces of nematic versus isotropic domains, respectively, providing added control over the droplet structure. Comparisons of experiments and numerical simulations using a Landau–de Gennes continuum model provide insight into the relative importance of the LC elasticity and orientational-dependent interfacial energies on droplet morphologies and properties. I show that the hierarchical organization of the LC compartments generates optical properties and responsiveness not found in emulsions of isotropic oils. Chapter 4 also describes a study of multiphase emulsions using LCs and immiscible perfluoroalkanes dispersed in water or glycerol (the latter continuous phase is used to enable characterization). I found that the nematogen 4′-pentyl-4- biphenylcarbonitrile (5CB) anchors homeotropically (perpendicularly) and weakly at 4 liquid perfluorononane (F9) interfaces, consistent with the smectic layering of 5CB molecules. The proposed role of smectic layering is supported by experiments performed with 4-(trans-4-pentylcyclohexyl)benzonitrile, a nematogen that possesses a cyclohexyl group that frustrates the smectic packing and leads to tilted orientations at the F9 interface. By employing perfluorocarbon and hydrocarbon surfactants in combination with multiphase 5CB and F9 emulsion droplets dispersed in a continuous water or glycerol phase, I observe a range of emulsion droplet morphologies to form, including core-shell and Janus structures, with internal organizations that reflect an interplay of interfacial (anchoring energies; F9 and glycerol) and elastic energies within the confines of the geometry of the emulsion droplet. By comparing experimental observations to simulations of the LC–perfluorocarbon droplets based on a Landau-de Gennes model of the free energy, we place bounds on the orientation-dependent interfacial energies that underlie the internal ordering of these complex emulsions. Additionally, by forming core–shells emulsion droplets from 5CB (shell) and perfluoroheptane (cores), I demonstrate how a liquid-to-vapor phase transition in the perfluorocarbon core can be used to actuate the droplet and rapidly thin the nematic shell. Overall, the results reported in this chapter demonstrate that multiphase LC emulsions formed from mixtures of perfluoroalkanes and LCs provide new opportunities to engineer hierarchical and stimuli-responsive emulsion systems. Synthetic soft matter systems, when driven beyond equilibrium by active processes, offer the potential to achieve dynamical states and functions of a complexity found in living matter. Emulsions offer the basis of a simple yet versatile system for identification of the physicochemical principles underlying active soft matter, but how multiple internal phases within emulsion droplets (e.g., Janus morphologies) organize to impact emergent dynamics is not understood. In Chapter 5, I create multiphase oil droplets with ultralow interfacial tensions but distinct viscosities, and drive them into 5 motion in aqueous micellar solutions. Preferential solubilization of select components of the oil both drives the droplet motion and yields a progression of internal phase morphological states with distinct symmetries. I find the active droplets to exhibit five dynamical states during morphogenesis. By quantifying microscopic flow fields, we show that it is possible to map the diverse droplet behaviors to squirmer models of spherical microswimmers in Stokes flow, thus showing that multiphase droplets offer the basis of a versatile platform with which to study and engineer the hydrodynamics of microswimmers. Polymerization reactions triggered by stimuli play a pivotal role in materials science, with applications ranging from lithography to biomedicine to adaptive materials. However, the development of chemically triggered, stimuli-responsive systems that can confer spatial and temporal control on polymerization remains a challenge. In Chapter 6, chemical-stimuli-induced polymerization based on a LC printhead is presented. The LC responds to a local chemical stimulus at its aqueous interface, resulting in the ejection of initiator into the solution to trigger polymerization. Various LC printhead geometries are designed, allowing programming of: i) bulk solution polymerization, ii) synthesis of a thin surface-confined polymeric coating, iii) polymerization-induced self-assembly of block copolymers to form various nanostructures (sphere, worm-like, and vesicles), and iv) 3D polymeric structures printed according to local solution conditions. The approach is demonstrated using amphiphiles, multivalent ions, and biomolecules as stimuli. A key challenge underlying the design of miniature machines is to develop autonomous time- and space-specific functional behaviors that require little human intervention. Dissipative processes beyond equilibrium that are evolving continuously over time and space provide a complex yet promising strategy. In Chapter 7, I report that a self-propelled LC double emulsion droplet can be programmed to spontaneously 6 swim to a targeted location and autonomously trigger a delayed release of chemical cargo from the droplet that initiates polymerization. The programmed release of cargo relies on Marangoni flows that generate an internal circulation to interact with LC elasticity, which leads to coarsening and clustering of cargo inside the droplet in a time- dependent manner. Above a threshold cargo cluster size, inertial forces exceed interfacial colloidal forces that initially trap cargo within the droplets during swimming, leading to the site-specific release of cargo. Our findings provide a platform that utilizes non-equilibrium strategies to achieve spatial and temporal functioning of active soft matter. Encoding macroscopic behaviors from nanoscale structural transformations is a promising and challenging approach for designing functional and responsive materials. In Chapter 8, I demonstrate experimentally that replacement of a phenyl ring by a cyclohexyl ring in a mesogen can dramatically alter macroscopic orientational ordering of a LC at a fluorocarbon interface. We reconstruct the LC interfaces using atomistic molecular dynamics simulations of LC films in contact with perfluoroalkane oils, which recapitulate the intermolecular interactions between mesogens and fluorocarbon molecule and their macroscopic behaviors. Compartmentalization of mesogen and calculation of potential of mean force between each compartment and fluorocarbon molecule provides an explicit explanation of the formation of molecular ordering where affinity between perfluorocarbon and cyclohexyl group is favored over phenyl group. Reactions at interfaces between fluid phases are widely used to synthesize small molecules, polymers and nanoparticles. In situ monitoring of the underlying dynamic reaction pathways remains challenging. Liquid crystals (LCs) have been used to detect simple chemical transformations at interfaces in situations where interface-bound reactants and products trigger distinct equilibrium orientations of LCs. However, whether or not LCs can be used to report complex reaction pathways via non- 7 equilibrium states generated by reactions has not been explored. In Chapter 9, I explore this question using SN2′ nucleophilic substitution reactions that involve a synthetic amphiphile and a series of amine-based nucleophiles with one to four reaction sites. Although all reactants and products generate the same equilibrium LC orientation, we find that each nucleophile defines a dis-tinct set of possible reaction pathways with a characteristic spatial and temporal LC optical response unique to the nucleophile. Additional experiments reveal that the non-equilibrium orientational states of the LCs arise from a combination of dynamic interfacial processes that include adsorption/desorption of reactants, the presence of reaction intermediates on the LC interface, as well as the generation of interfacial tension gradients (Marangoni stresses). Overall, our results reveal that the spatiotemporal optical outputs of LCs (“optical fingerprints”) can be a rich source of information regarding interfacial reactions. Microplastics (small plastic pieces < 5 mm) in the environment are one of the biggest global issue requiring urgent intervention. The large surface-area-to-volume ratio of microplastics provides an opportunity to collect them using interfaces. As water surfaces are a major location for microplastic accumulation, in Chapter 10, I propose a separations principle called “surface liquid extraction”, and employ it to collect microparticles on water surfaces using oil interfaces. A carrier species, e.g., decanol, polypropylene glycol or naturally occurring fatty acids (oleic acid, linoleic acid), is dispensed on the water surface with a drop of vegetable oil as a sink that continuously absorbs the carrier. The absorption of the carrier by the sink generates sustained surface fluxes that “actively” skim the water surface due to Marangoni stresses, which carry and collect polymeric microparticles presented on the surface. This method provides a transport velocity on the order of 100-1000 m/s for 10-700 m microparticle aggregates, which is at least 109 times faster than diffusion, with similar or promoted efficacy for small particles. This method also simultaneously sorts polymeric particles 8 into distinct locations of the oil sink based on their interfacial properties. These findings provide a platform that uses interfaces to address the microplastics problem. 1.3 References [1] J. N. Israelachvili, Intermolecular and Surface Forces (Academic press, 2011). [2] P. C. Hiemenz and R. Rajagopalan, Principles of Colloid and Surface Chemistry, Revised and Expanded (CRC press, 2016). [3] L. D. Zarzar, V. Sresht, E. M. Sletten, J. A. Kalow, D. Blankschtein, and T. M. Swager, Dynamically Reconfigurable Complex Emulsions via Tunable Interfacial Tensions, Nature 518, 520 (2015). [4] D. Gupta, B. Sarker, K. Thadikaran, V. John, C. Maldarelli, and G. John, Sacrificial Amphiphiles: Eco-Friendly Chemical Herders as Oil Spill Mitigation Chemicals, Sci. Adv. 1, e1400265 (2015). [5] A. Nikolov and D. Wasan, Oil Lenses on the Air–Water Surface and the Validity of Neumann’s Rule, Adv. Colloid Interface Sci. 244, 174 (2017). [6] J. K. Gupta, S. Sivakumar, F. Caruso, and N. L. Abbott, Size-Dependent Ordering of Liquid Crystals Observed in Polymeric Capsules with Micrometer and Smaller Diameters, Angew. Chemie, Int. Ed. 48, 1652 (2009). [7] X. Wang, D. S. Miller, E. Bukusoglu, J. J. De Pablo, and N. L. Abbott, Topological Defects in Liquid Crystals as Templates for Molecular Self-Assembly, Nat. Mater. 15, 106 (2016). [8] E. Bukusoglu, X. Wang, J. A. Martinez-Gonzalez, J. J. De Pablo, and N. L. Abbott, Stimuli-Responsive Cubosomes Formed from Blue Phase Liquid Crystals, Adv. Mater. 27, 6892 (2015). [9] D. Seč, T. Porenta, M. Ravnik, and S. Žumer, Geometrical Frustration of Chiral Ordering in Cholesteric Droplets, Soft Matter 8, 11982 (2012). 9 [10] S. Roh, M. Tsuei, and N. L. Abbott, Using Liquid Crystals for in Situ Optical Mapping of Interfacial Mobility and Surfactant Concentrations at Flowing Aqueous- Oil Interfaces, Langmuir 37, 5810 (2021). [11] I. H. Lin, D. S. Miller, P. J. Bertics, C. J. Murphy, J. J. De Pablo, and N. L. Abbott, Endotoxin-Induced Structural Transformations in Liquid Crystalline Droplets, Science (80-. ). 332, 1297 (2011). [12] M. Roché, Z. Li, I. M. Griffiths, S. Le Roux, I. Cantat, A. Saint-Jalmes, and H. A. Stone, Marangoni Flow of Soluble Amphiphiles, Phys. Rev. Lett. 112, 208302 (2014). [13] X. Wang, R. Zhang, A. Mozaffari, J. J. de Pablo, and N. L. Abbott, Active Motion of Multiphase Oil Droplets: Emergent Dynamics of Squirmers with Evolving Internal Structure, Soft Matter 17, 2985 (2021). [14] E. Hermans, M. Saad Bhamla, P. Kao, G. G. Fuller, and J. Vermant, Lung Surfactants and Different Contributions to Thin Film Stability, Soft Matter 11, 8048 (2015). [15] J. Kosco, M. Bidwell, H. Cha, T. Martin, C. T. Howells, M. Sachs, D. H. Anjum, S. Gonzalez Lopez, L. Zou, A. Wadsworth, W. Zhang, L. Zhang, J. Tellam, R. Sougrat, F. Laquai, et al., Enhanced Photocatalytic Hydrogen Evolution from Organic Semiconductor Heterojunction Nanoparticles, Nat. Mater. 19, 559 (2020). [16] H. Chen, R. Zhang, X. Chen, G. Zeng, L. Kobera, S. Abbrent, B. Zhang, W. Chen, G. Xu, J. Oh, S. H. Kang, S. Chen, C. Yang, J. Brus, J. Hou, et al., A Guest- Assisted Molecular-Organization Approach for >17% Efficiency Organic Solar Cells Using Environmentally Friendly Solvents, Nat. Energy 6, 1045 (2021). [17] M. A. Augustin and Y. Hemar, Nano- and Micro-Structured Assemblies for Encapsulation of Food Ingredients, Chem. Soc. Rev. 38, 902 (2009). [18] J. M. Gutiérrez, C. González, A. Maestro, I. Solè, C. M. Pey, and J. Nolla, 10 Nano-Emulsions: New Applications and Optimization of Their Preparation, Curr. Opin. Colloid Interface Sci. 13, 245 (2008). [19] H. Shahsavan, A. Aghakhani, H. Zeng, Y. Guo, Z. S. Davidson, A. Priimagi, and M. Sitti, Bioinspired Underwater Locomotion of Light-Driven Liquid Crystal Gels, Proc. Natl. Acad. Sci. U. S. A. 117, 5125 (2020). [20] Z. Wang, J. Wang, J. Ayarza, T. Steeves, Z. Hu, S. Manna, and A. P. Esser‐ Kahn, Bio-Inspired Mechanically Adaptive Materials through Vibration-Induced Crosslinking, Nat. Mater. 20, 869 (2021). [21] X. Wang, H. Sun, Y.-K. Kim, D. B. Wright, M. Tsuei, N. C. Gianneschi, and N. L. Abbott, Stimuli-Responsive Liquid Crystal Printheads for Spatial and Temporal Control of Polymerization, Adv. Mater. 34, 2106535 (2022). [22] R. Seemann, J.-B. Fleury, and C. C. Maass, Self-Propelled Droplets, Eur. Phys. J. Spec. Top. 225, 2227 (2016). [23] J. Jeong, A. Gross, W.-S. Wei, F. Tu, D. Lee, P. J. Collings, and A. G. Yodh, Liquid Crystal Janus Emulsion Droplets: Preparation, Tumbling, and Swimming, Soft Matter 11, 6747 (2015). [24] M. Li, M. Brinkmann, I. Pagonabarraga, R. Seemann, and J.-B. Fleury, Spatiotemporal Control of Cargo Delivery Performed by Programmable Self- Propelled Janus Droplets, Commun. Phys. 1, 23 (2018). [25] C. Krüger, G. Klös, C. Bahr, and C. C. Maass, Curling Liquid Crystal Microswimmers: A Cascade of Spontaneous Symmetry Breaking, Phys. Rev. Lett. 117, 1 (2016). [26] A. Fernández-Nieves, D. R. Link, M. Márquez, and D. A. Weitz, Topological Changes in Bipolar Nematic Droplets under Flow, Phys. Rev. Lett. 98, 1 (2007). [27] Y. Zhou, A. Guo, R. Zhang, J. C. Armas-Perez, J. A. Martínez-González, M. Rahimi, M. Sadati, and J. J. de Pablo, Mesoscale Structure of Chiral Nematic Shells, 11 Soft Matter 12, 8983 (2016). [28] D. S. Miller, X. Wang, and N. L. Abbott, Design of Functional Materials Based on Liquid Crystalline Droplets, Chem. Mater. 26, 496 (2014). [29] E. P. Solomon, L. R. Berg, and D. W. Martin, Biology (Boorks/Cole Cengage Learning, 2011). [30] C. C. Maass, C. Krüger, S. Herminghaus, and C. Bahr, Swimming Droplets, Annu. Rev. Condens. Matter Phys. 7, 171 (2016). 12 Chapter 2 Literature Review 2.1 Introduction to Thermotropic Liquid Crystals (LCs) The three classical states of matter are solid, liquid and gas. However, many intermediate states also exist (Figure 2.1). In crystalline solids, the components (molecules or groups of molecules) are arranged in a highly ordered state, with both long-range positional order and long-range orientational order[1,2]. As temperature increases, some materials can go through a so-called crystal-to-glass transition into amorphous solids. The physical properties of amorphous solids are the same along all directions (no long-range orientational order but maintaining the long-range positional Figure 2.1 Nematic liquid crystal as an intermediate phase of matter. Schematic illustrations of the molecular organization of crystalline solid, amorphous solid, liquid crystalline, and isotropic liquid states. 13 order). As temperature further increases, materials can lose both long-range orientational and positional characteristics, and form isotropic liquids. However, during this transition, some organic materials maintain the long-range orientational order but lose the long-range positional order. One example is the rod-like molecule 4′-pentyl-4- biphenylcarbonitrile (5CB), which forms a so-called nematic liquid crystalline (LC) phase at ambient conditions. Macroscopically, long-range positional order is characterized by mechanical stiffness, so nematic 5CB forms a fluid phase due to the lack of positional order, and thus can flow like a liquid. In addition, 5CB molecules in the nematic phase tend to be parallel to some common axis (director) because of the orientational order. Macroscopically, nematic 5CB exhibits anisotropy similar to crystalline solids (e.g., optical, electrical and magnetical), which will be discussed later in this Chapter. LCs are classified as thermotropic when the temperature of the system alone determines the phase behavior; in contrast, lyotropic systems consist of compositions that require the addition of solvent to form a LC phase. Chemistry of Mesogen Among many types of LCs (nematic, smectic, cholesteric and Blue Phases), nematic LCs are the simplest phase that is uniaxial and achiral. For purpose of this Figure 2.2 Molecular structure of 4′-pentyl-4-biphenylcarbonitrile (5CB) as a typical LC compound. 14 dissertation, we introduce the chemistry and physics of nematic LCs. Although what chemical composition of a compound can lead to a LC phase is not fully understood, empirically LC materials share some common chemical characteristics, such as elongated shape and sufficient rigidity[3]. 5CB is considered a typical single-component species that possesses a nematic phase over a large temperature range (20 °C – 35.5 °C) at ambient conditions (Figure 2.2). This class of so-called cyano-biphenyl compounds shares structures of a polar head group, a rigid core and a flexible long aliphatic tail with an overall elongated shape. The structure dictates the orientational order of the material from both entropy and enthalpy (e.g., π-π stacking) perspectives. Mathematical Description of Orientational Order of LC The orientational order present in a liquid crystal is described mathematically by the order parameter, S, which is defined as the average of second Legendre polynomials[1]: 3 cos2 𝜃 − 1 𝑆 = 𝑃2(cos 𝜃) = 〈 〉 (2.1) 2 where 𝜃 is the angle between director and the long axis of any given mesogen. Perfect orientational order (all 𝜃 equals to 0°) lead to 𝑆 = 1 , whereas 𝑆 = 0 indicates no orientational order. Typical values for the order parameter are between 0.3 and 0.9[4]. Surface Anchoring In reality, the organizations of a LC entity are defined by their boundary conditions, which are called surface anchoring. Surface anchoring refers to the surface- induced orientational ordering of LCs, dictated by the intermolecular interactions between LCs and a confining medium (surface chemistry). In the absence of external fields, the lowest free energy orientation of the LC director defines an “easy axis” 15 Figure 2.3 Physical properties of nematic LC. Schematic illustrations of the three key properties of (i) surface anchoring, which defines the orientational ordering of LCs at interfaces, (ii) elasticity associated with three typical modes of strain in LCs and (iii) three types of topological defects. (Figure 2.3). Any deviation of the director from the easy axis causes an orientation- dependent increase in the free energy of the interface, which is described by the Rapini- Papoular expression: 1 𝜎 = 𝜎 2o + 𝑊sin (𝜃d − 𝜃o) (2.2) 2 where 𝜎 is the interfacial free energy density, 𝜎o is the orientation-independent component of interfacial free energy density, 𝑊 is the anchoring energy, and 𝜃d and 𝜃o are the orientations of the director at the surface and the easy axis, respectively. The 16 anchoring energy is typically on the order of 10−6 – 10−5 N/m2, which is small in comparison to the interfacial tension of isotropic liquids (10−2 N/m2). This indicates that surface anchoring can be easily affected by external fields or interfacial absorbates[5]. A director of LCs parallel to an interface is called planar anchoring. In contrast, perpendicular alignment to the interface is homeotropic anchoring; otherwise, it is called tilted anchoring. Elasticity The long-range orientational ordering of mesogens dictates a propagation of LC alignment from interfaces into bulk over distances of hundreds of micrometers and on time scales of milliseconds. LCs in the bulk become strained to accommodate their boundary conditions (Figure 2.3 Elasticity). Free energy density that is stored in typical modes of LC deformations is described by the Frank-Oseen expression[1]: 1 𝐹elastic = [𝐾11(∇ ∙ 𝒏) 2 + 𝐾 222(𝒏 ∙ ∇ × 𝒏) + 𝐾33(𝒏 × ∇ × 𝒏) 2 + 𝐾24∇ ∙ (𝒏 × ∇ × 𝒏 + 𝒏(∇ ∙ 𝒏))] (2.3) 2 where 𝐾11 , 𝐾22 , 𝐾33 and 𝐾24 are the splay, twist, bend and saddle-splay elastic constants, respectively. 𝐾11, 𝐾22 and 𝐾33 are typically in the order of 10 -11 – 10-12 N. 𝐾24 is usually neglected, but it can play an important role in specific geometries, such as spherical LC droplets. In a so-called “one constant approximation”, 𝐾11 = 𝐾22 = 𝐾33 = 𝐾, 𝐾24 = 0. Application of this approximation reduces the expression for elastic free energy density in LCs (equation 2.3) into 1 𝐹elastic = 𝐾[(∇ ∙ 𝒏) 2 + (∇ × 𝒏)2] (2.4) 2 where the elastic energy is often estimated by the elastic constant multiplies a length scale ~ Kl. A competition between anchoring energy (Wl2) and elastic energy (Kl) exists. When 𝐾𝑙 > 𝑊𝑙2, that is 𝑙 < 𝐾/𝑊, elastic force is large enough to bend the surface anchoring, and the critical length 𝐾/𝑊 is referred as extrapolation length. 17 Topological Defects When the LC ordering changes abruptly that continuous deformation cannot accommodate, a defect forms because of the high stress in the vicinity (Figure 2.3). Thus the defect possesses a large free energy density compared to bulk LC. The free energy of a defect core can be expressed by: 4 𝐹d = ∫ 𝑓d𝑑𝑉d ~ 𝜋𝑟 3 d 𝜀d (2.5) 𝑉 3d where 𝑟d is the radius of the core of a defect, and 𝜀d is the free energy density associated with local “melting” of the nematic phase to an isotropic phase. Based on the origin of the defect, the free energy of core melting is provided by elastic free energy (𝐾𝑟d), therefore 𝐾 𝑟d ~ √ ~ 5 nm (2.6) 𝜀d And the magnitude of the free energy of the core of a point defect is typically around 10-19 J. LCs exhibit multiple configurations dependent on types of LCs, boundary conditions (anchoring), geometry of the confinements and sizes of the LC droplet[6– 28]. In droplet geometry, for example, homeotropic anchoring dictates a so-called radial configuration that directors of LCs radiate from a center point defect (hedgehog), while planar anchoring gives rise to a bipolar configuration that the directors are roughly parallel but converge into two surface defects (boojum) at the poles (Figure 2.4). The transition of anchoring also leads to a series of configurations resulting from tilted anchoring, such as preradial and axial configurations. LC droplets have been explored as templates for the synthesis of spherical and non-spherical polymer particles[9,16,29], 18 sensors for biomolecules[24], and templates of molecular self-assembly into defects[30]. Figure 2.4 LC organizations in droplet geometry. Schematic illustrations of typical LC organizations in the confinement of droplet geometry, which is defined by the boundary conditions. Radial configuration with directors radiating from a center defect (hedgehog) corresponds to homeotropic anchoring at the LC interface. Bipolar configuration with directors aligning approximately in the same direction throughout the droplet, but converges into two surface defects (boojum). Tilted anchoring dictates preradial and axial configurations. Additionally, when the elastic energy surpasses anchoring energy in small droplets with dimensions less than the extrapolation length, anchoring is bent into uniform alignment. 2.2 Optical Analysis of Orientational Ordering of LCs The optical properties of LCs are one of the most important characteristics of the phase, which also form the basis for the most widely used applications of LCs – liquid crystal display (LCD). While it is possible to observe and characterize bulk samples of LCs macroscopically (which appear to be cloudy because of the crystallinity that scatters light), one of the most powerful ways is to observe a thin sample of LC under a polarized light microscope. Birefringence in Nematic Liquid Crystals 19 Light as an electromagnetic wave interacts with matter as it propagates. The interaction between light and matter is characterized by a refractive index (the ratio of the speed of light in vacuum to that in a second medium of greater density). The anisotropy of mesogen (Figure 2.5) causes light polarized along the director traveling at a different speed than light polarized perpendicular to the director. Nematic LCs are thus birefringent, with 𝑛e and 𝑛o the two refractive indices parallel to the director (extraordinary axis) and perpendicular to the director (ordinary axis), respectively. The Figure 2.5 Birefringence in nematic LCs. Bright and dark (extinguished) regions in a LC viewed with a sample between a crossed polarizer and analyzer. Dark region appears when the director is parallel to either axes of polarizer or analyzer. Abrupt change of brightness indicates an abrupt change of the orientation of the director. polarized light from the polarizer can be considered as composed light oscillating along or perpendicular to the director with zero phase difference. The light in the two directions emerges as elliptically polarized light after traveling through the LCs because of a phase shift due to the distinct speed of light in the two directions[4]. The part of the elliptically polarized light that is parallel to the analyzer can be observed. When the director is parallel or perpendicular to the polarizer, the polarized light still oscillates in the direction of the polarizer when it leaves the LC (linearly polarized), so the light from the analyzer is distinguished (Figure 2.5). In contrast, the brightness is maximized when the director is 45° to the two polarizers. In addition, because the brightness indicates the 20 orientation of the director, any abrupt change in brightness also suggests an abrupt change of the director, that is the location of defects. Optical Retardance The retardance of LC materials, ∆𝑟, describes the difference in the optical path between two characteristic polarizations of light parallel or perpendicular to the director upon propagation through LCs. The retardance of LC organizations is dependent on the optical properties of the LCs (birefringence) and the thickness. Because light with different wavelengths propagates at slightly different speeds in LCs, a thin LC sample can exhibit different interference colors indicative of its thickness. For thin samples of LCs, the retardance can be read out via a comparison of observed color to the Michel- Lévy interference color chart (Figure 2.6). For example, given a color of the first-order pink and the birefringence of 5CB (𝑛e = 1.71, 𝑛o = 1.53, ∆𝑛 = 𝑛e − 𝑛o = 0.18), we can find a tilted line ends at -0.180 on the right axis, find the point where this line across Figure 2.6 Michel-Lévy interference color chart. 21 the first-order pink on the chart, and read out the retardance on the bottom axis at 551 and the effective thickness of the sample at 3 m on the left axis. However, as the retardance increases (the thickness of the sample increases), the interference color fades out into pale pink and green or even pale yellow. Therefore this method is imprecise for high optical retardance unless there is a continuous transition of retardance from 0 to the high value. In this case, we can use a microscope that is equipped with a compensator to obtain optical retardance. In a thin film of LC with a homeotropic anchoring at the bottom substrate and a continuous transition of the orientational order throughout the film, the LC tilt angle at the top interface can be calculated from the retardance. Assuming that minimization of the LC elastic energy leads to LC tilt that varies linearly across the film, the optical retardance of a LC film as a function of the thickness (d) and the tilt angle at the top interface (𝜃) is described by: 𝑑 𝑛𝑒𝑛𝑜 ∆𝑟~∫ − 𝑛 𝑜 𝑑𝑧 (2.7) 0 2 𝑧 𝑧√𝑛𝑜sin 2 ( 𝜃) + 𝑛2cos2 ( 𝜃) ( 𝑑 𝑒 𝑑 ) where 𝑛𝑜 and 𝑛𝑒 are the refractive indices perpendicular (ordinary axis) and parallel (extraordinary axis) to the director. Chemical and Biomolecular Sensing via LC Droplets LC droplets have been used as sensitive reporters of interfacial absorbates (examples of LC droplets are shown in Figure 2.7), because the small confinement reinforces the competition between surface anchoring, bulk elastic deformations and topological defects to reach the free energy minimum of the system[29]. For example, past study has investigated the ordering transitions of LC droplets dispersed in aqueous phases, resulting from the adsorption of amphiphiles at the aqueous-LC droplet interface[23]. LC droplets were found to exhibit six sequential ordered states as a 22 function of increasing concentration of sodium dodecylsulfate (SDS), which corresponds to a continuous transition of LC anchoring at the aqueous interface. Amphiphiles can also self-assemble at the sites of topological defects, as opposed to “surface-driven ordering transitions” that involve changes in surface energetics, to trigger ordering transitions. Bipolar-to-radial configurational transitions have been observed in LC droplets dispersed in an aqueous solution in the presence of bacterial endotoxin (lipid A) at picogram-per-milliliter concentrations[24], which is at least five orders of magnitude lower in concentration than that required to trigger surface-driven anchoring transitions in LC droplets. Confocal fluorescence microscopy confirmed that endotoxin was localized at the defect formed at the center of each LC droplet in a radial configuration. Figure 2.7 Examples of micrographs of LC droplets. Schematic illustrations, bright-field and polarized light micrographs of LC droplets exhibiting bipolar and radial configurations, respectively[24]. 23 Detection of Physical and Chemical Transformations via LC Films Recently, amphiphiles interactions at aqueous-LC interfaces in and out of equilibrium have been increasingly explored due to the key role of water in preserving the structure and function of many chemicals and biomolecules[31,32]. Three key features of aqueous-LC interfaces facilitate the sensing applications of LCs: (i) aqueous- LC interfaces exhibit high accessibility of amphiphiles from the aqueous bulk solution to the LC interface. (ii) LC molecules at aqueous interfaces possess high mobility, which allows them to rapidly reorganize in milliseconds, and consequently report the interactions of a molecular stimulus at their interfaces. (iii) Molecule-level events at aqueous-LC interfaces can be amplified through the orientational ordering of LCs into a few hundred micrometers, which can be easily identified under polarized light microscope (Figure 2.8). These properties lead to interfacial phenomena of LCs that are Figure 2.8 Examples of micrographs of LC films. Polarized light micrographs (left) and corresponding schematic illustrations of LC films with homeotropic anchoring at the bottom substrates and (a) planar or (b) homeotropic anchoring at the top aqueous-LC interfaces due to the presence of interfacial adsorbates, respectively. 24 rich in information on the interactions and dynamics of the molecules at the LC interface. In particular, LC interfaces provide fresh approaches for exploring the dynamic interactions of molecular assemblies of amphiphiles with fluid interfaces at the single-event level, which is not easy to characterize through other methods, such as surface plasmon resonance, mass spectroscopy or nuclear magnetic resonance. Past study has reported that incubation of aqueous dispersions of phospholipid 1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC) assemblies against interfaces of 5CB films triggers spatially localized (micrometer-scale) and transient (subsecond) flashes of light, so-called “blinking”, to be transmitted through the LC[31]. Analysis of the spatiotemporal response of the LC supports the proposal that each optical “blinking” event results from collision of a single DLPC assembly with the LC interface. The collision and subsequent spreading of amphiphiles at the interface generates a surface pressure-driven interfacial flow (Marangoni flow) that causes transient reorientation of LC and generation of a bright optical flash between crossed polarizers. The spatial and temporal optical responses of LC–aqueous interfaces can also be used to report interfacial phenomena associated with flow-induced states and dynamics of amphiphiles[32]. Dependent on the surfactant concentration, the LC can map spatial variation in the mobility of the LC–aqueous interface at a steady state, which reflects the presence of Marangoni stresses, or map spatial variation in the surfactant concentration via the coupling between the surfactants and LCs. Thus, the dynamic response of the LC is controlled by the mass transport of the surfactant from the bulk solution onto the LC interface, the transport of the surfactant along the interface, and the kinetics of desorption of amphiphiles from the interface. Aqueous-LC interfaces can also be used to monitor the relaxation of the surfactant distribution across the interface following the cessation of the imposed flow. Overall, aqueous-LC interfaces offer the basis for mapping interfacial mobilities and distributions of surfactants at flowing 25 interfaces. 2.3 Balance of Interfacial Tensions between Multiple Phases When one type of molecule primarily attracts each other versus another type of molecule, a mixture of the two types of molecules separates into two phases of matter. An interface is a boundary between two phase. Fluid interfaces tend to shrink to minimize surface free energy. Macroscopically, the intermolecular force between the molecules at the interface of two fluids is described by interfacial tension. When it comes to a mixture of multiple phases, the balance between multiple interfacial tensions leads to morphologies that reach free energy minimum. Complex Emulsion Complex emulsions (Figure 2.9), often referring to multiple emulsions, which Figure 2.9 Balance of interfacial tensions gives rise to multiple morphologies of complex emulsions. Schematic illustrations of double emulsion (core-shell), Janus emulsion and separated drop. When the interfacial tension between phase 1 and phase 3 is larger than the sum of the other two interfacial tensions, double emulsions form. When the interfacial tension between phase 1 and phase 2 exceeds the sum of the other two interfacial tensions, phases 1 and 2 separate into discrete droplets. If none of the three interfacial tensions can overcome the sum of the other two interfacial tensions, Janus emulsions emerge. 26 are highly structured fluids consisting of emulsion drops that contain smaller shells and droplets inside, and Janus droplets, which are droplets with chemically distinct faces, are increasingly explored in medical diagnostics[33], polymer synthesis[34], chemical or biological sensing[35] and food processing[36]. The morphology of complex emulsions[21,37,38] is largely dependent on the balance of interfacial tensions 𝛾𝑖𝑗 , 𝑖, 𝑗, 𝑘 = 1,2,3, etc. In terms of multiple emulsions consisting of three phases (Figure 2.9), Janus emulsions emerge when either one of the three interfacial tensions is smaller than the sum of the other two interfacial tensions 𝛾𝑖𝑗 < 𝛾𝑘𝑖 + 𝛾𝑗𝑘 . If one of the interfacial tensions exceeds the sum of the other two interfacial tensions 𝛾𝑖𝑗 > 𝛾𝑘𝑖 + 𝛾𝑗𝑘 , phase 𝑘 tends to wet the interface between the phase 𝑖 and the phase 𝑗 . Therefore, double emulsions form when phase 𝑘 is a discontinuous phase; or phase 𝑖 and phase 𝑗 are separated into discrete droplets when phase 𝑘 is the continuous phase. This turns out to be consistent with the triangle inequality, which is known as Neumann's triangle rule. Recently, temperature-sensitive miscibility of hydrocarbon and fluorocarbon oils has been used to fabricate complex emulsions in one step in aqueous solutions[37]. Droplet geometry can switch between complex emulsions and single-phase droplets as a function of temperature. Use of hydrocarbon and fluorocarbon surfactants can also tune the droplet morphology between double emulsions and Janus droplets because of the preferential adsorption of two types of surfactants on two types of oils. Also the low interfacial tensions between hydrocarbon and fluorocarbon due to the high affinity between the two oils leads to a low interfacial tension, which forms a stable spherical shape, even in the presence of external stimuli, complex emulsions that is well-suited for sensing [35,39]. Particle at Interface (Pickering Emulsion) 27 The colloidal interactions between a micro-sized solid particle and an interface have been intensively explored in the context of Pickering emulsions[40] (emulsions that are stabilized by solid particles). Below we describe the thermodynamics of particles located at interfaces that is relevant to the sorting of microplastics (Chapter 10). For an ideal particle surface that possesses zero contact angle hysteresis (Figure 2.10), the contact angle 𝜃 is determined by three interfacial tensions, 𝛾31 , 𝛾32 , 𝛾21 , which can be estimated by Young’s equation[40]: 𝛾31 − 𝛾32 cos 𝜃 = (2.8) 𝛾21 Figure 2.10 Contact angle of a solid particle at a fluid interface. Schematic illustration of a solid particle 3 at the interface between 1 and 2. The contact angle 𝜃 is dependent on the three interfacial tensions. When one interfacial tension is larger than the sum of the other two interfacial tensions, e.g., 𝛾31 > 𝛾32 + 𝛾21, 𝜃 = 0°, phase 2 wets the surface of 3, and the particle is submerged in phase 2. When the sum of any two interfacial tension is larger than the third interfacial tension, the particle is pinned at the interface in equilibrium. This turns out to be also consistent with the triangle inequality (Neumann's triangle rule). The energy required to detach the particle from the interface can be expressed as: ∆𝐸 = 𝛾21𝜋𝑟 2(1 − |cos 𝜃|)2 (2.9) where 𝑟 is the particle radius. For a 10 mm particle at water-air interface with 𝜃 = 90°, 𝛾21 = 72.8 mN/m, ∆𝐸 = 2.3 × 10 −11 J = 5.6 × 109 𝑘B𝑇, where 𝑘B is the Boltzmann constant. Because the energy needed to remove the particle from the interface is large, 28 the attachment of the particle is considered irreversible. Fluid at Fluid Surface How a liquid drop located at a fluid interface adopt its morphology (whether or not the droplet spreads) to accommodate the balance of interfacial interactions is also important in term of collecting microplastics as described in Chapter 10. Traditionally, this topic has been explored for the purpose of cleaning up the oil spill on ocean surface[41–43]. When a drop of immiscible oil is placed on the water-air interface (Figure 2.11), the three interfacial tensions 𝛾w, of the water-air interface, 𝛾o, of the oil- air interface, and 𝛾ow, of the oil-water interface govern the behavior of the oil droplet if the influence from gravity is small compared to interfacial tensions. The state of the oil droplet also follows the Neumann's triangle rule: Figure 2.11 Contact angle of an immiscible oil droplet placed on the water-air interface. Schematic illustration of an oil droplet at the water surface. The contact angle 𝜃 is dependent on the three interfacial tensions. (i) Typically in the absence of surfactant 𝛾w > 𝛾o + 𝛾ow, the oil spreads over the water surface to form a thin film ~100 m. (ii) If 𝛾o > 𝛾w + 𝛾ow, water wets the oil-air interface to form oil in water 29 emulsions. The oil droplet tends to stay in the bulk water. (iii) If 𝛾ow > 𝛾w + 𝛾o, an air film will separate the oil droplet from water, which is not likely to happen since the high surface tension of water 𝛾w ~ 73 mN/m indicates low affinity of water and air than the affinity between water and most oil species (𝛾ow < 40 mN/m). (iv) When |𝛾o − 𝛾ow| < 𝛾w < 𝛾o + 𝛾ow, the oil phase forms a droplet (lens). A spreading coefficient of oil is also defined as 𝑆o = 𝛾w − (𝛾o + 𝛾ow), such that a positive spreading coefficient indicates spreading of the oil on the water surface, and a negative coefficient leads to the formation of an oil droplet on the water surface. In the case of cleaning up oil spills on ocean surface (herding[43]), oil spreads over the water surface initially, because of a positive spreading coefficient. The first step is to spay surfactants (herder) surrounding the oil slick. The presence of surfactants on the water surface can decrease the surface tension of water, which alters the spreading coefficient to be negative. The oil film retracts as the oil-air and oil-water tensions pull the liquid back. The thickened oil slick can be recovered mechanically or burned in situ after the oil slick reaches at least 3 mm. 2.4 Marangoni Effect Mass transport involving both diffusion and convection is crucial for living organisms to carry out functions essential for life. In particular, convective flow facilitates fast transport phenomena. For example, alveolar cells produce lung surfactant, a mixture of proteins and lipids, that emerges as a thin film lining the alveolar surface. This surfactant lining can lower the surface tension of alveoli walls, and thus increase pulmonary compliance for breathing. It can also prevent collapse of the lung at the end of exhalation[44]. The fast spreading of the surfactant on the alveolar surface relies on a non-equilibrium phenomenon called the Marangoni effect. 30 The Marangoni effect describes the mass transfer via convective flow along an interface between two fluids driven down surface pressure gradients. The origin of the surface pressure can be from gradients in temperature, surfactant concentration or electrostatic potential, leading to thermo-, soluto- or electro-capillary motion[45]. When the convective flow, for example, is generated by an interfacial tension gradient due to inhomogeneous adsorption of surfactant at the water-air interface, the flow velocity 𝑢∗ can be evaluated by stress continuity at the interface 𝜂𝑢 𝑧−1 ≈ (𝛾 − 𝛾 )𝑙−1∗ 𝜈 w s ∗ (2.10) where 𝜂 is the dynamic viscosity of the liquid (1 mPa s for water), 𝑙∗ is the distance over which velocity gradients are established, 𝑧 ≈ (𝜈𝑙 𝑢−1)1/2𝜈 ∗ ∗ is the thickness of a Blasius viscous boundary layer of surface by assuming laminar boundary layer in steady state, with 𝜈 = 𝜂/𝜌 the kinetic viscosity, 𝜌 is the density of the fluid (~1000 kg/m3 for water), 𝛾w (~73 mN/m) is the surface tension of water, 𝛾s is the surface tension of e.g., a surfactant solution that decreases 𝛾. Therefore, a fluctuation in the surface tension of water (~10 mN/m) can potentially generate a flow velocity up to meters per second. When an oil droplet is placed into an aqueous micellar solution (Figure 2.12), micelles continuously solubilize the oil to minimize the free energy of the system. The solubilization process induces inhomogeneous coverage of the interface with surfactant molecules (Marangoni stress), which leads to convective flow along the droplet interface. This flow breaks the symmetry of the droplet, and brings fresh micelles from one side, resulting in a sustained self-propelled motion of the oil droplet[46]. The convective flow was found to influence the internal organizations inside the droplets, which in turn changes the trajectory of the motion[47,48]. Curling self-propulsion was reported for droplets of tens of micrometers that comprised of nematic LCs[47], because 31 Figure 2.12 Spontaneous motion of an oil droplet dispersed in micellar solution due to Marangoni effect. Schematic illustration of an oil droplet in aqueous micellar solution. Solubilization of oil by the surrounding micelles generates an interfacial tension gradient (Marangoni stress) that leads to spontaneous symmetry breaking and motion of the droplet. of an interplay between the elasticity and the flow. Janus droplets also exhibited five distinct states of motion dependent on the relative volume of each compartment, including a spiraling motion due to preferential solubilization of select components of the oil from different compartments, as described in this dissertation[48]. On the surface of oil, a drop of a mixture of water and volatile alcohol spreads and spontaneously fragments into a myriad of minute droplets due to Marangoni flows generated by the evaporation of alcohol[49]. Overall, these studies opened up opportunities for the design of functional materials operating beyond thermal equilibrium. 2.5 References [1] P. G. de Gennes and J. Prost, The Physics of Liquid Crystals (Clarendon Press, Oxford, 1993). [2] M. A. Wahab, Solid State Physics: Structure and Properties of Materials (Alpha Science International, 2005). 32 [3] G. R. Luckhurst, G. W. Gray, and Eds, The Molecular Physics of Liquid Crystals (Academic Press, 1979). [4] P. J. Collings, Liquid Crystals: Nature’s Delicate Phase of Matter (Princeton University Press, 2002). [5] P. S. Drzaic, Liquid Crystal Dispersions (World Scientific, 1995). [6] J. H. Erdmann, S. Žumer, and J. W. Doane, Configuration Transition in a Nematic Liquid Crystal Confined to a Small Spherical Cavity, Phys. Rev. Lett. 64, 1907 (1990). [7] R. Ondris-Crawford, E. P. Boyko, B. G. Wagner, J. H. Erdmann, S. Žumer, and J. W. Doane, Microscope Textures of Nematic Droplets in Polymer Dispersed Liquid Crystals, J. Appl. Phys. 69, 6380 (1991). [8] J. K. Whitmer, X. Wang, F. Mondiot, D. S. Miller, N. L. Abbott, and J. J. De Pablo, Nematic-Field-Driven Positioning of Particles in Liquid Crystal Droplets, Phys. Rev. Lett. 111, 1 (2013). [9] F. Mondiot, X. Wang, J. J. de Pablo, and N. L. Abbott, Liquid Crystal-Based Emulsions for Synthesis of Spherical and Non- Spherical, J. Am. Chem. Soc. 135, 9972 (2013). [10] X. Wang, D. S. Miller, J. J. De Pablo, and N. L. Abbott, Organized Assemblies of Colloids Formed at the Poles of Micrometer-Sized Droplets of Liquid Crystal, Soft Matter 10, 8821 (2014). [11] E. Bukusoglu, X. Wang, J. A. Martinez-Gonzalez, J. J. De Pablo, and N. L. Abbott, Stimuli-Responsive Cubosomes Formed from Blue Phase Liquid Crystals, Adv. Mater. 27, 6892 (2015). [12] J. Jeong, A. Gross, W.-S. Wei, F. Tu, D. Lee, P. J. Collings, and A. G. Yodh, Liquid Crystal Janus Emulsion Droplets: Preparation, Tumbling, and Swimming, Soft Matter 11, 6747 (2015). 33 [13] Y. Zhou, E. Bukusoglu, J. A. Martínez-González, M. Rahimi, T. F. Roberts, R. Zhang, X. Wang, N. L. Abbott, and J. J. De Pablo, Structural Transitions in Cholesteric Liquid Crystal Droplets, ACS Nano 10, 6484 (2016). [14] J.-Y. Kwon, M. Khan, and S.-Y. Park, PH-Responsive Liquid Crystal Double Emulsion Droplets Prepared Using Microfluidics, RSC Adv. 6, 55976 (2016). [15] Y. Zhou, A. Guo, R. Zhang, J. C. Armas-Perez, J. A. Martínez-González, M. Rahimi, M. Sadati, and J. J. de Pablo, Mesoscale Structure of Chiral Nematic Shells, Soft Matter 12, 8983 (2016). [16] X. Wang, E. Bukusoglu, D. S. Miller, M. A. Bedolla Pantoja, J. Xiang, O. D. Lavrentovich, and N. L. Abbott, Synthesis of Optically Complex, Porous, and Anisometric Polymeric Microparticles by Templating from Liquid Crystalline Droplets, Adv. Funct. Mater. 26, 7343 (2016). [17] M. Urbanski, C. G. Reyes, J. Noh, A. Sharma, Y. Geng, V. S. R. Jampani, and J. P. F. Lagerwall, Liquid Crystals in Micron-Scale Droplets, Shells and Fibers, J. Phys. Condens. Matter 29, 133003 (2017). [18] A. Fernández-Nieves, D. R. Link, M. Márquez, and D. A. Weitz, Topological Changes in Bipolar Nematic Droplets under Flow, Phys. Rev. Lett. 98, 1 (2007). [19] X. Wang et al., Patterned Surface Anchoring of Nematic Droplets at Miscible Liquid–Liquid Interfaces, Soft Matter 13, 5714 (2017). [20] T. Suzuki, Y. Li, A. Gevorkian, and E. Kumacheva, Compound Droplets Derived from a Cholesteric Suspension of Cellulose Nanocrystals, Soft Matter 14, 9713 (2018). [21] X. Wang, Y. Zhou, Y.-K. Kim, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Thermally Reconfigurable Janus Droplets with Nematic Liquid Crystalline and Isotropic Perfluorocarbon Oil Compartments, Soft Matter 15, 34 2580 (2019). [22] J. K. Gupta, S. Sivakumar, F. Caruso, and N. L. Abbott, Size-Dependent Ordering of Liquid Crystals Observed in Polymeric Capsules with Micrometer and Smaller Diameters, Angew. Chemie, Int. Ed. 48, 1652 (2009). [23] J. K. Gupta, J. S. Zimmerman, J. J. De Pablo, F. Caruso, and N. L. Abbott, Characterization of Adsorbate-Induced Ordering Transitions of Liquid Crystals within Monodisperse Droplets, Langmuir 25, 9016 (2009). [24] I. H. Lin, D. S. Miller, P. J. Bertics, C. J. Murphy, J. J. De Pablo, and N. L. Abbott, Endotoxin-Induced Structural Transformations in Liquid Crystalline Droplets, Science 332, 1297 (2011). [25] H. Liang, J. Noh, R. Zentel, P. Rudquist, P. F. Jan, and J. P. Lagerwall, Tuning the Defect Configurations in Nematic and Smectic Liquid Crystalline Shells, Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 371, 20120258 (2013). [26] D. Seč, T. Porenta, M. Ravnik, and S. Žumer, Geometrical Frustration of Chiral Ordering in Cholesteric Droplets, Soft Matter 8, 11982 (2012). [27] V. Tomar, S. I. Hernández, N. L. Abbott, J. P. Hernández-Ortiz, and J. J. de Pablo, Morphological Transitions in Liquid Crystal Nanodroplets, Soft Matter 8, 8679 (2012). [28] H.-L. Liang, R. Zentel, P. Rudquist, and J. Lagerwall, Towards Tunable Defect Arrangements in Smectic Liquid Crystal Shells Utilizing the Nematic-Smectic Transition in Hybrid-Aligned Geometries, Soft Matter 8, 5443 (2012). [29] D. S. Miller, X. Wang, and N. L. Abbott, Design of Functional Materials Based on Liquid Crystalline Droplets, Chem. Mater. 26, 496 (2014). [30] X. Wang, D. S. Miller, E. Bukusoglu, J. J. De Pablo, and N. L. Abbott, Topological Defects in Liquid Crystals as Templates for Molecular Self- Assembly, Nat. Mater. 15, 106 (2016). 35 [31] M. Tsuei, M. Shivrayan, Y. K. Kim, S. Thayumanavan, and N. L. Abbott, Optical “Blinking” Triggered by Collisions of Single Supramolecular Assemblies of Amphiphilic Molecules with Interfaces of Liquid Crystals, J. Am. Chem. Soc. 142, 6139 (2020). [32] S. Roh, M. Tsuei, and N. L. Abbott, Using Liquid Crystals for in Situ Optical Mapping of Interfacial Mobility and Surfactant Concentrations at Flowing Aqueous-Oil Interfaces, Langmuir 37, 5810 (2021). [33] P. A. Gresham, M. Barnett, S. V. Smith, and R. Schneider, Use of a Sustained- Release Multiple Emulsion to Extend the Period of Radioprotection Conferred by Cysteamine [9], Nature 234, 149 (1971). [34] E. Bukusoglu, X. Wang, Y. Zhou, J. A. Martínez-González, M. Rahimi, Q. Wang, J. J. de Pablo, and N. L. Abbott, Positioning Colloids at the Surfaces of Cholesteric Liquid Crystal Droplets, Soft Matter 12, 8781 (2016). [35] Q. Zhang, S. Savagatrup, P. Kaplonek, P. H. Seeberger, and T. M. Swager, Janus Emulsions for the Detection of Bacteria, ACS Cent. Sci. 3, 309 (2017). [36] M. A. Augustin and Y. Hemar, Nano- and Micro-Structured Assemblies for Encapsulation of Food Ingredients, Chem. Soc. Rev. 38, 902 (2009). [37] L. D. Zarzar, V. Sresht, E. M. Sletten, J. A. Kalow, D. Blankschtein, and T. M. Swager, Dynamically Reconfigurable Complex Emulsions via Tunable Interfacial Tensions, Nature 518, 520 (2015). [38] X. Wang, Y. Zhou, V. Palacio-Betancur, Y.-K. Kim, L. Delalande, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Reconfigurable Multicompartment Emulsion Drops Formed by Nematic Liquid Crystals and Immiscible Perfluorocarbon Oils., Langmuir 35, 16312 (2019). [39] A. Concellón, C. A. Zentner, and T. M. Swager, Dynamic Complex Liquid Crystal Emulsions, J. Am. Chem. Soc. 141, 18246 (2019). 36 [40] L. E. Low, S. P. Siva, Y. K. Ho, E. S. Chan, and B. T. Tey, Recent Advances of Characterization Techniques for the Formation, Physical Properties and Stability of Pickering Emulsion, Adv. Colloid Interface Sci. 277, 102117 (2020). [41] W. D. Harkins and A. Feldman, Films. The Spreading of Liquids and the Spreading Coefficient., J. Am. Chem. Soc. 44, 2665 (1922). [42] A. Nikolov and D. Wasan, Oil Lenses on the Air–Water Surface and the Validity of Neumann’s Rule, Adv. Colloid Interface Sci. 244, 174 (2017). [43] D. Gupta, B. Sarker, K. Thadikaran, V. John, C. Maldarelli, and G. John, Sacrificial Amphiphiles: Eco-Friendly Chemical Herders as Oil Spill Mitigation Chemicals, Sci. Adv. 1, e1400265 (2015). [44] E. Hermans, M. Saad Bhamla, P. Kao, G. G. Fuller, and J. Vermant, Lung Surfactants and Different Contributions to Thin Film Stability, Soft Matter 11, 8048 (2015). [45] H. Manikantan and T. M. Squires, Surfactant Dynamics: Hidden Variables Controlling Fluid Flows, J. Fluid Mech. 892 (2020). [46] C. C. Maass, C. Krüger, S. Herminghaus, and C. Bahr, Swimming Droplets, Annu. Rev. Condens. Matter Phys. 7, 171 (2016). [47] C. Krüger, G. Klös, C. Bahr, and C. C. Maass, Curling Liquid Crystal Microswimmers: A Cascade of Spontaneous Symmetry Breaking, Phys. Rev. Lett. 117, 1 (2016). [48] X. Wang, R. Zhang, A. Mozaffari, J. J. de Pablo, and N. L. Abbott, Active Motion of Multiphase Oil Droplets: Emergent Dynamics of Squirmers with Evolving Internal Structure, Soft Matter 17, 2985 (2021). [49] L. Keiser, H. Bense, P. Colinet, J. Bico, and E. Reyssat, Marangoni Bursting: Evaporation-Induced Emulsification of Binary Mixtures on a Liquid Layer, 37 Phys. Rev. Lett. 118, 074504 (2017). 38 Chapter 3 Thermally Reconfigurable Janus Droplets with Nematic Liquid Crystalline and Isotropic Perfluorocarbon Oil Compartments 1. Introduction Oil in water emulsions are encountered in a wide range of contexts[1], including within foods[2], in pharmaceutical preparations[3] and as templates for materials synthesis[4,5]. More recently, emulsion droplets with increasingly complex internal organizations have been explored, including multi-compartment droplets[6–8] and droplets comprised of structured oils such as liquid crystals[9–12]. The latter, in particular, offer the basis of systems that undergo changes in organization and optical properties in response to subtle stimuli such as bacterial lipids[13,14], metabolites[15], protein and cellulosic filaments[16]. In this paper, we report an investigation of multi- compartment emulsions where one compartment is a liquid crystalline (LC) oil and the other is an isotropic oil. The work is motivated by the long-range goal of designing complex emulsion droplets that possess morphological stability, yet can reorganize to generate optical responses to a range of triggers, including the presence of amphiphiles and targeted temperatures. Multi-compartment emulsions comprised of coexisting isotropic oil domains have been widely investigated, and a thermodynamic framework has been established to describe droplet morphology[7,17,18]. In brief, the stability of Janus droplets (structured emulsions comprised of domains that define two distinct faces) is determined by the interfacial tensions of the immiscible constituent phases (isotropic phases 1 and 2) and the third immiscible surrounding phase (phase 3; Figure 3.1). Specifically, to form a Janus emulsion droplet, the three spreading coefficients Si = γjk – (γij + γik) (i ≠ j ≠ k =1, 2, 3) must all be negative. If this condition is not satisfied, 39 either double emulsions with core-shell structures or separate droplets, each comprised of a single phase form at equilibrium. The morphology of a Janus droplet is described by the three interfacial tensions that form a Neumann’s triangle, namely, γ γ γ 13 = 23 = - 12 sin θ2 sin θ1 sin (θ1 + θ2) This equation predicts that a Janus droplet comprised of two hemispheres (θ1 → π/2 and θ2 → π/2) will form if γ13 ~ γ23 >> γ12. Figure 3.1 (A) Sketch of interfacial tensions of a Janus droplet that define (B) a Neumann’s triangle and satisfy γ13 ~ γ23 >> γ12. A range of experimental systems have been demonstrated to satisfy the above- described formalism, including systems comprised of isotropic phases of polymers, hydrocarbons and fluorocarbons[7,18,19]. In particular, the phase separation of isotropic hydrocarbon and fluorocarbon oils has been utilized to fabricate Janus droplets that, when combined with hydrocarbon and fluorocarbon surfactants, enabled access to various morphologies including Janus-type and core-shell structures[20]. The interfacial tensions between hydrocarbon and fluorocarbon compartments (~1 mN m-1) were an order of magnitude smaller than interfacial tensions at the oil-aqueous interfaces (~10 40 mN m-1), which endowed droplets with spherical shapes and mechanical stability in the presence of surfactants. As noted above, single-compartment emulsion droplets comprised of liquid crystalline oils have also been reported[4,10,21–34]. These studies have revealed that the internal organization of the liquid crystal (LC) within the droplet reflects a delicate balance of contributions to the free energy that are absent in isotropic oil emulsions, including contributions arising from orientation-dependent interfacial free energies, elastic strain associated with deformation of the LC, and the presence of singular topological defects[10,26,27]. This balance of energetic contributions is strongly droplet-size dependent[12], and can yield emulsion systems that reorganize in the presence of very low concentrations of amphiphilic species such as bacterial endotoxins[13]. Whereas single-compartment LC droplets[9,10,13–16,28], and multi- compartment isotropic emulsion droplets[6–8,17–20] have been explored, only a limited range of multi-compartment LC systems have been reported, the majority of which have been metastable liquid crystalline shells surrounded by inner and outer aqueous phases[35–41]. As described above, surface-energetic and elastic contributions to the free energy, as well as contributions to the free energy arising from formation of defects, depend strongly on geometry (topology), and thus we hypothesized that multi- compartment emulsions droplets with coexisting non-spherical LC and isotropic oil domains may offer access to both thermodynamic stability and responsiveness not found in spherically symmetric LC systems (shells). We note that one past study has reported Janus droplets formed of a liquid crystalline oil (5CB) and polymer (polydimethylsiloxane, PDMS) dispersed in aqueous solution[42]. The droplet morphology was demonstrated to be sensitive to surfactant concentration in the aqueous phase and the compartment volume ratio. The interfacial tensions of the three interfaces 41 of the droplets, however, were similar in magnitude (~1 mN m-1), resulting in non- spherical droplets. In addition, in some instances, the Janus-type morphology was unstable (relative to two single-phase droplets) when surfactants were added. In the experiments and simulations reported in this paper, we explore the morphologies and internal organization of Janus droplets with nematic and isotropic oil compartments formed using mixtures of hydrocarbon mesogens and isotropic fluorocarbon oils, namely, E7, a mixture of hydrogenated cyanobiphenyl and terphenyl mesogens that forms a nematic liquid crystalline phase at room temperature, and perfluorobenzene (FB), an isotropic fluorocarbon oil, respectively (Figure 3.2). We used these two constituents to design thermally reconfigurable Janus droplets for three reasons. First, the temperature range over which nematic (N) and isotropic (I) phases coexist with typical thermotropic LCs is narrow (it is typically limited to 1-2 oC 42-44). When using the mixture of E7 and FB, however, we found N-I coexistence to persist over an unusually broad temperature range (~10 oC at concentration of FB (CFB) =10 % v/v)[45,46]. This characteristic of the system enabled precise and continuous control over the morphologies of the droplets. As discussed below, the phase behaviour of this system also contrasts to previously studied two-phase isotropic perfluorocarbon/hydrocarbon systems[6–8,27]. Second, as both nematic and isotropic compartments are comprised of FB-E7 mixtures (the concentration difference between the two phases is less than 5 % v/v FB), the interfacial tension between the N-I compartments is small compared to the two oil-aqueous interfaces, and thus stable, spherical multicompartment droplets were formed over a wide range of conditions (as internal structure varied). Third, whereas the similarity in composition of the two compartments is sufficiently small to provide stable droplets, the difference in composition between isotropic and nematic domains allowed control of the internal structures of the droplets by using either fluorocarbon or hydrocarbon surfactants. The 42 Figure 3.2 Molecular structures of (A) the mixture of nematogens called E7, (B) perfluorobenzene (FB), (C) perfluorooctanoic acid (PFOA) and (D) sodium dodecyl sulfate (SDS). choice of surfactant led to differential effects on the interfacial tensions of the hydrocarbon-rich (nematic) and perfluorocarbon-rich (isotropic) phases, resulting in surfactant-dependent changes in partial wetting of the internal phases on the aqueous interface of the droplets and thus morphology of the droplets. Additional insights regarding the interfacial energetics and internal ordering of the LC within the Janus emulsion droplets are described below by comparing experimental results and numerical simulations. Finally, we comment that an additional goal of our study was to determine if Janus droplets with nematic compartments deviate significantly from the above-described thermodynamic criteria used to describe the stability and morphology of Janus droplets comprised of isotropic oils. 43 2. Experiments and Simulations Materials The nematic liquid crystal, E7, was purchased from HCCH (Jiangsu Hecheng Display Technology Co., LTD). Hexafluorobenzene (99%, perfluorobenzene), toluene (anhydrous, 99.8%), n-hexane (≥99%), sodium dodecyl sulfate (99.0%, SDS) and perfluorooctanoic acid (96%, PFOA) were purchased from Sigma–Aldrich (St. Louis, MO, USA). Fisher Finest Premium Grade glass slides was purchased from Fisher Scientific (Pittsburgh, PA). Polyimide 2555, which was used to induce planar alignment of E7 when coated on glass substrates was purchased from HD Microsystems. Purification of water (18.2 MΩ cm resistivity at 25 oC) was performed using a Milli-Q water system (Millipore, Bedford, MA, USA). Preparation of Optical Cells Optical cells were assembled from two glass slides separated from each other by either 20 μm-diameter particles or 100 μm-thick tape. Glass slides were spin-coated with polyimide 2555 using a Laurell spin coater, baked at 250 oC, and then rubbed and assembled in an anti-parallel fashion to yield unidirectional tangential alignment of LCs. Microscopy observations Optical microscopy was performed using an Olympus BX41 microscope with a 20× objective and a Moticam 10.0 MP camera. Temperature was controlled using a Linkam TMS 94 hot stage with an accuracy of 0.1 oC. Measurement of Phase Behavior of FB-E7 Mixtures To measure the phase behavior of mixtures of E7 and FB, we prepared 5% v/v perfluorobenzene in E7 (FB-E7) by adding 20 µL of perfluorobenzene to 400 µL of 44 nematic E7 at room temperature. Next, the mixture was vortexed at 3000 rpm for 10 seconds. The phase behavior of the FB-E7 mixture was measured using 20 µm-thick optical cells and polarized light microscopy. Measurement of LC Orientation at Nematic-Isotropic (N-I) Interface To characterize the orientation of the LC at the N-I interface, we prepared 5 % v/v FB-E7 mixture as described above, heated the mixture to an isotropic state at 55 oC (to avoid flow-induced alignment), and injected the mixture into a polyimide 2555- coated optical cell with a cavity thickness of 20 µm. The sample was then cooled to 47 °C, at which temperature coexisting N and I phases were observed using optical microscopy. Preparation of Emulsions To prepare droplets of FB-E7 stabilized by SDS, we dispersed 100 µL of the 5 % v/v FB in E7 mixture described above in 1000 µL FB-saturated aqueous SDS solution by vortexing at 3000 rpm for 8 seconds. The mixture was injected into a 100 µm-thick optical cell at room temperature and sealed with silicone grease to prevent evaporation. The samples were equilibrated for 30 min prior to making observations. To prepare droplets of FB-E7 stabilized by 1 mM of fluorocarbon surfactant (PFOA), we prepared 200 mM PFOA in FB, and then mixed 400 µL E7 with 20 µL of the PFOA dissolved in perfluorobenzene. 100 µL of the resulting FB-E7 mixture (containing 10 mM PFOA) was dispersed in 1000 µL FB saturated 100 µM SDS aqueous solution. This procedure resulted in an overall concentration of PFOA in the system of 1 mM. SDS was originally added to stabilize the droplets after vortex. We also dispersed PFOA-contained droplets in FB saturated water in the absence of SDS and non-PFOA droplets in the PFOA solution. We confirmed the internal structures of 45 N-I biphasic droplets were similar. Measurement of Interfacial Tensions We first prepared an 8 % v/v FB-E7 mixture by adding 32 µL of perfluorobenzene to 400 µL of E7 at room temperature. The resulting system was nematic at room temperature, and was heated to 36 oC to reach a biphasic state. The nematic and the isotropic phases were separated using an Eppendorf 5417R centrifuge for 15 min at 12000 rpm at 36 oC. The densities of the nematic and isotropic phases of the 8 % v/v FB-E7 mixture were 1.0462 g cm3 and 1.0487 g cm3, respectively, measured with an Anton Paar DMA 5000 density meter. The interfacial tensions of the nematic- aqueous (N-A) interface, γN, the isotropic-aqueous (I-A) interface, γI, and the nematic- isotropic (N-I) interface, γN-I were measured using a Dataphysics OCA 15 plus contact angle measurement device with SCA 20 software. Pendant drops of LC were formed using Hamilton metal 22-gauge needles (blunt point). The equilibration time was 10 min for N-A and I-A interfaces, and 5 min for the N-I interface. Temperature was controlled using an environmental chamber with an accuracy of 0.1 oC. The 8 % v/v FB-E7 mixture was used in these experiments to ensure that coexistence of the N-I phases occurred within the operating temperature range of the centrifuge (-9 oC to + 40 oC). We confirmed that the 8 % v/v FB-E7 mixture did not have any qualitative difference compared to the 5 % v/v FB-E7 mixture in the experiments described in Figures 3.5 and 3.8. Simulations We used a Landau-de Gennes (LdG) continuum model for the order tensor Q, defined by Qij=S (ni nj −1/3 ij). Here, ni are the x, y, z components of the local director vector[47]. The scalar order parameter is denoted by S. The bulk free energy is given by 46 A U AU AU 2 𝐿 ∂Q ∂Qij ij Fbulk(Q)=∫ ( (1− ) Q Q − Q Q Q + (Q Q ) ) dV + ∫ dV (3.1) 2 3 ij ji 3 ij jk ki 4 ij jibulk bulk 2 ∂xk ∂xk where A and U are phenomenological parameters that depend on temperature and pressure. The one-constant representation is adopted here, where L denotes the elastic constant of the LC (the symbol K is used for the elastic constant in Frank-Oseen elastic energy and K = 2LS2). The first term in Equation (1), which corresponds to enthalpic contributions to the free energy, serves to control the equilibrium value of the order parameter 1 8 Seq= (1 + 3√1− ) 4 3U The second term represents the elastic free energy, which governs long-range director distortions[48]. The surface free energy, which represents the degenerate conic anchoring, was introduced in our previous work: 2 Fsurf(Q)=∫ W(P 2 surf ik Q̃ P kl lj − Seq cos θe Pij) dΣ (2) where W is the anchoring strength; Q̃ =Q +(1/3) Seqδij ; the preferred tilting angle ij ij between surface normal 𝜈 and the ‘easy cone’ is denoted by 𝜃𝑒; and the operator tensor Pij=νiνj. The definition of operator tensor P here is different from that in the Fournier- Galatola degenerate planar surface functional[49]. An iterative Ginzburg-Landau relaxation with finite difference method on a cubic mesh (with resolution of 7.15 nm) is adopted to minimize the free energy[50]. Polarization micrographs were calculated using the Jones matrix formalism, in which light of a single wavelength (350 nm) travels along a chosen direction and the total phase shift is accumulated[51]. The numerical parameters used in this work are A = 1.067  105 J m-3, U = 3.5, L = 6 pN, and K = 4.55 pN. The droplet diameter D is 2 μm[52]. The surface anchoring strengths are chosen according to experimental measurements. 47 3. Results and Discussion Phase behavior of E7-FB mixtures To guide the design of Janus droplets using phase-separated mixtures of E7 and FB, we measured the temperature and composition-dependent phase behavior of FB-E7 mixtures. The inset in Figure 3.3 shows that nematic domains dispersed in the isotropic continuous phase of a FB-E7 mixture (5 % v/v FB) at 44.0 oC were readily identified by optical microscopy as bright regions between crossed polars. Figure 3.3 also shows that the phase diagram of FB-E7 mixtures comprises nematic (N), coexisting nematic- isotropic (N-I), and isotropic (I) regions. In the absence of FB, consistent with the Gibbs phase rule for a nematic mixture (Figure 3.2), N and I phases coexist over a temperature range of ~2.4 oC[53]. Addition of FB to E7 decreases the N-I transition temperature (TNI) and leads to an expansion of the biphasic region. Specifically, for a mixture containing 10 % v/v FB, the temperature interval of the biphasic region expands to ~10 oC. We note that this temperature interval is wider than that for typical mixtures of hydrocarbons and E7. For example, the N-I coexistence of a mixture of E7 and hexane (10 % v/v) is ~ 4.8 oC (green triangles in Figure 3.3), approximately one half that of the FB-E7 mixture. A toluene-E7 mixture (blue diamonds in Figure 3.3) also possesses a N-I coexistence that is narrower than the FB-E7 mixture. The broad N-I coexistence of the FB-E7 mixture enables precise manipulation (via changes in temperature) of the relative volumes of the I and N compartments of Janus droplets. By separating nematic and isotropic phases of an 8% v/v mixture of FB and E7 at 36oC, we found that the isotropic phase is FB-rich (12% v/v) relative to the coexisting nematic phase (7% v/v FB). 48 We hypothesized that the unusually broad biphasic region of the FB and E7 mixture occurred because of uneven partitioning of FB between the isotropic and nematic phases (i.e., C I NFB /CFB is large). Past studies have reported that fluorine atoms P A I N Figure 3.3 Phase diagram of the FB-E7 mixture (red circles ), plotted as a function of volume fraction of FB (CFB). Mean ± s.d. (n = 6) are shown. The green triangles ( ) and the blue diamonds ( ) mark the limits of the N-I coexistence of hexane-E7 mixtures and toluene-E7 mixtures, respectively. The inset is an optical micrograph (crossed polars) of a FB-E7 mixture (CFB=5 % v/v) at 44.0 °C (N+I). Scale bar, 20 µm. Double headed arrows indicate the orientation of polarizer (P) and analyzer (A). of the FB molecule withdraw electron density from the aromatic ring due to the high electronegativity of fluorine. Thus, the charge distribution of FB is opposite to that of benzene[54]. The resulting strong electrostatic attraction between FB and benzene has been shown previously (melting points at 5.0 oC and 5.4 oC, respectively) to lead to the formation of a solid complex containing equimolar quantities of the two compounds at 23.7 oC[55]. Within FB-E7 mixtures, we hypothesize that FB has a similar attraction to the aromatic rings of the biphenyls comprising E7, leading to a large ratio of activity 49 coefficients for FB in nematic and isotropic phases, and thus a broad biphasic region (i.e., C I NFB /CFB is large; see Supporting Information for additional discussion). In the experiments reported below, we used mixtures of E7 and FB containing 5% v/v FB because it possesses a coexistence interval that starts 10 oC above room temperature (i.e., the mixture is homogenous at room temperature and the two-phase region can be conveniently accessed by heating from room temperature). We surveyed mixtures of FB and other thermotropic LCs, including 5CB (4-cyano-4’- pentylbiphenyl), 8CB (4’-n-octyl-4-cyano-biphenyl) and TL205 (a mixture of fluorinated biphenyl and terphenyl compounds that forms a nematic phase at room temperature). For each mixture containing 10% v/v FB, the temperature interval for N- I coexistence was ~20 oC below the N-I phase transition temperature (TNI) of the LC prior to addition of FB. Specifically, mixing FB with 5CB (TNI = 35.5 oC) or 8CB (TNI = 40.5 oC) results in a two-phase region below room temperature; while for TL205 (TNI ~ 80 oC), the temperature range of two-phase region is high (~60 oC). We also surveyed mixtures of perfluoroalkanes and these thermotropic LCs, but found them to be immiscible and to possess an upper consolute temperature similar to the isotropic systems reported in a past study[20]. The mixtures of perfluoroalkanes and LCs formed two-phase systems with coexisting phases that differed greatly in composition (close to 0% and 100%, perfluoroalkane, respectively) below the critical temperature, making it difficult to form stable, multicompartment emulsions in the presence of the stimuli described in later parts of this paper. Anchoring of nematic LC at interface to perfluorocarbon-rich isotropic phase To understand the internal organization of multicompartment emulsions based on FB and E7, a key underlying issue is the orientational ordering of the N phase at the interface of the I phase (orientational anchoring). To provide insight into this issue, we 50 A P, A AP I I N no I N B N N n I o no C P, A P A N I no no D P AP, A N N I I n no o no Figure 3.4 LC ordering at N-I interfaces of a FB-E7 mixture. (A) Optical micrograph (crossed polars) of N-I interface with an alignment that is parallel to the far field orientation of LCs (no) and (B) is inclined from no. (Brightness and contrast of crossed-polarized micrographs in (A) and (B) is increased to enable easy visualization.) The schematic illustrations on the right side of A and B show an interface separating N and I domains as imaged in A, when viewed from the top (left illustration) or side (right illustration). (C) Optical image of a domain wall defect, indicated by the red frame, within a nematic domain attached to one surface of an optical cell. The schematic illustration to the right of C shows the LC orientations near the wall when viewed from the side. The green ellipsoids represent molecules in the nematic phase. (D) Optical image (crossed polars) of a cylindrical nematic domain spanning two surfaces. The schematic illustrations on the right are top and side views, respectively. Scale bars, 10 µm. The anchoring of the LC at the confining surfaces is planar, and no is parallel to the rubbing direction. performed experiments (Figure 3.4) using films of the FB-E7 (5 % v/v) mixture confined between two polymer-coated glass surfaces that induced unidirectional tangential alignment (no) of LCs. The thickness of the LC film was 20 μm. Figure 3.4A and B show two N-I interfaces of a sample that have distinct orientations relative to the 51 rubbing direction of the polymer film (no) at T = 47.0 oC. When viewed between crossed polars, the interface (indicated by a blue frame) oriented parallel to the no and polarizer (P) was bright (Figure 3.4A), whereas the interface rotated from n by 40oo was dark (Figure 3.4B). These observations are consistent with a tilted alignment (50o with respect to the normal direction of N-I interface) of LC at the N-I interface. To exclude the possibility that the optical appearance in Figure 3.4A was due to curvature of the N- I interface (i.e., a meniscus; the contact angle between the N-I interface and isotropic- glass interface was 72o (Figure S3.1)), we also examined a droplet of nematic LC dispersed in an isotropic continuous phase (Figure 3.4C). The nematic domain was attached to one surface of the optical cell (did not span the thickness) and had a diameter less than 30 μm. Inspection of Figure 3.4C reveals a white band (indicated by a red frame) corresponding to a domain wall defect[56], an observation that is consistent with our conclusion of tilting of the LC at the N-I interface. Additionally, when the diameter of the nematic domain was larger than 30 μm, the LC domain spanned the two glass surfaces and formed a cylindrical shape (Figure 3.4D). The LC in the domain exhibited a uniform colour (green between parallel polars; red between crossed polars,) except at the two poles. The uniform colour indicates unidirectional alignment of the LC across the nematic domain, and thus weak surface anchoring of the LC at the N-I interface. The order of magnitude of the elastic modulus (K) of E7 is ~5 pN. To obtain an anchoring energy (~Wl2) that is comparable in magnitude to the elastic energy (~Kl), we estimate the anchoring strength (W) to be of order of 10-6-10-7 N m-1, where the length scale l is ~10 μm. This value is small compared to typical values of anchoring energies of LCs[9],[57] (typically 10-4-10-5 N m-1). Furthermore, the nematic domains we observed assumed a spherical shape (Figure 3.4C and D), hinting that the elastic energy associated with confinement of the LC is small compared to the interfacial free energy. As detailed below, we use this experimental characterization of the N-I interface to 52 provide boundary conditions for numerical simulations of the internal organization of the LC droplets. Temperature-dependent morphology and internal ordering of complex LC emulsions Next, we leveraged the above-described understanding of the phase behavior of the FB-E7 mixture and ordering at the N-I interface to prepare and provide insights into emulsions formed from E7 and FB mixtures. The FB-E7 mixture (5% v/v FB) was dispersed into aqueous SDS solution to explore the temperature-dependent droplet morphology and the ordering of LC within the nematic compartments. We prepared the dispersions by vortexing the mixture in the single-phase region of the phase diagram (room temperature) to ensure that the droplets were homogeneous in composition. As detailed below, we focus initially on the morphology of the emulsions; subsequently, we address internal ordering of LC domains. Figure 3.5 shows the morphologies of FB-E7 droplets observed as a function of increasing temperature in the presence of 1 mM SDS (the morphology and internal ordering is dependent on surfactant type and concentration; see Figures 3.6 and 3.8). At 35 °C, the entire droplet volume was a single nematic phase (Figure 3.5A). Optical images between crossed polars revealed a so-called radial configuration[10], indicating a perpendicular alignment of the LC at the aqueous interface. The optical texture is characterized by a point defect at the center of droplets and Maltese cross between crossed polars. Upon heating the droplet above TN-NI = 42 oC, we observed an isotropic domain to nucleate at the central defect (Figure 3.5B and C). We hypothesized that nucleation of the isotropic phase occurred at the defect because of the high free energy density of the defect core[9,58]. The configuration shown in Figure 3.5B was unstable and transient. The isotropic domain translated to the aqueous interface after formation. 53 A P, A A P B C D E I N F G Figure 3.5 Thermal reconfiguration of one LC droplet comprised of FB-E7 mixture via nematic-isotropic (N-I) phase transition. (A,D-G) Columns 1 and 2 are polarized light micrographs (single polarizer and crossed polars, respectively) of droplets of FB-E7 (CFB=5 % v/v) dispersed in FB- saturated aqueous 1 mM SDS solutions, upon heating across the N-I coexistence region at (A) 35.0 oC, (nematic), (D) 45.0 oC, (E) 46.0 oC, (F) 47.0 oC, and (G) 47.5 oC (isotropic), respectively. Scale bar, 10 µm. The double-headed arrows indicate the orientation of the polarizer (P) and analyzer (A). Column 3 shows corresponding director profiles obtained from simulations, and Column 4 shows the associated (simulated) polarized light micrographs (crossed polars). The micrograph and schematic illustration in (B) and (C), respectively, show that an isotropic domain formed first at the defect at the center of the droplet upon heating. 54 Figure 3.5D shows a Janus morphology observed at T= 45.0 oC. At 46.0 oC, the isotropic compartment expanded in volume, ultimately yielding a Janus droplet comprised of N and I hemispheres of equal size (Figure 3.5E). With further heating, the isotropic compartment expanded (Figure 3.5F) to eventually occupy the entire volume of the droplet at 47.5 oC (Figure 3.5G). The sequence of morphological states of the droplet described above was largely reversible and seen to occur upon reducing the temperature (Figure S3.3). In brief, during the cooling process, we observed multiple nematic domains to nucleate at the aqueous interface and then to coalesce into a single nematic domain. This coalescence process is likely driven by minimization of both the elastic energy of the nematic LC and the interfacial energy of the emulsion system. The isotropic domain was observed to adhere to the aqueous interface and shrank during cooling; with further cooling, the isotropic domain on the aqueous interface collapsed to a defect and translated back into the center to reform a single-phase nematic droplet. In comparison[20], isotropic multi- compartment droplet systems reported previously do not provide control over the volume ratio of internal phases via temperature due to the immiscibility of the fluorocarbon and hydrocarbon below the upper consolute temperature and complete miscibility above this critical temperature. At temperatures below the upper consolute temperature, the volumes of the two domains of the emulsions are determined by composition. To aid interpretation of the optical micrographs in Figure 3.5 in terms of internal ordering of LC within the compartments of the Janus droplets, we performed numerical simulations of the director profiles via a Landau-de Gennes (LdG) continuum model, simulated the corresponding optical micrographs, and then compared the simulated light micrographs to our experimental observations. In particular, by matching experimental and simulated optical micrographs, we obtained information about the surface 55 anchoring strength (W) of the LC domains at the boundaries of the compartments of the Janus droplets. We note that the radii of the simulated droplets were 1 μm, one order smaller than droplets observed in experiment. To enable comparisons between computation and experiments, we increased the anchoring strength by one order of magnitude to keep the ratio of anchoring energy (~Wl2) to elastic energy (~Kl) unchanged. We also note that the simulation was conducted with incident light of a single wavelength, and excluded optical influences of the interface (e.g. reflection and refraction). As a result, no interference patterns (denoted by a white arrow in experimental image in Figure 3.5A) were observed in simulated images. We used an elastic modulus (K) of 4.55 pN in the simulations. At the nematic- aqueous (N-A) interface, we specified perpendicular (homeotropic) LC anchoring consistent with our experimental observations (Figure 3.5A), and W of 1×10-3 N m-1. We used these boundary conditions at all temperatures, as we observed the perpendicular orientation at the N-A interface to be retained during heating and cooling, independent of temperature. At the N-I interface, we specified the LC to be tilted by 50o with respect to the surface normal (Figure 3.4); we varied W at the N-I interface to A A P C P, A A E A P P B P, A A A P D P, A F P, A A P P Figure 3.6 Optical micrographs showing the influence of surfactant type and concentration on the internal configurations (shown schematically to the right of the images) of nematic LC droplets (E7 or E7 + FB mixtures): (A,C,E) pure E7 and (B,D,F) FB-E7 (CFB=5 % v/v) mixtures dispersed in aqueous solutions at 35 oC of (A,B) 100 µM SDS, (C,D) 1 mM SDS and (E,F) 1 mM PFOA. Scale bars, 10 µm. The dashed black lines within the droplet boundaries in the illustrations represent the orientation of the LC director; the black dots in (A-D) represent defects; and the black solid loops in (E,F) represent line defects. 56 obtain the best agreement between simulated and experimental optical textures. Specifically, we found that use of boundary conditions that correspond to weak anchoring (W = 1×10-5 N m-1) in the simulations (Columns 3 and 4 in Figure 3.5) provided good agreement with experiments (Columns 1 and 2 in Figure 3.5). One key observation is the absence of defects on the N-I interfaces (Figure 3.5D-G) in both experiments and simulations, indicating the alignment of LC at the N-I interface can readily deviate from the easy axis (due to the elasticity of the bulk LC). If a higher anchoring strength was used in simulations to describe the N-I interface (e.g. W = 1×10- 4 N m-1 and W=1×10-3 N m-1), defect loops on N-I interfaces were observed, in contradiction to experiments (Figure S3.8). The absence of defects is also consistent with the minor impact of LC elasticity on the morphology of the LC compartment[31], which will be addressed below. In addition, simulated optical micrographs in Figure 3.5D and E show two extinction (dark) brushes between crossed polars (denoted by blue arrows), which is consistent with experiments. We also varied the orientation of the easy axis of LCs at the N-I interface from 30o to 60o to examine the sensitivity of the simulations to deviations from the experimental tilting angle (50o). We found the simulated optical texture was not sensitive to the exact tilting angle due to weak anchoring. Influence of surfactant type and perfluorobenzene on internal configurations of single-phase nematic droplets As noted in the Introduction, inspired by past studies of emulsions comprised of coexisting isotropic phases of hydrocarbons and perfluorocarbons[20], a key motivation for using perfluorocarbons to induce multiple domains in LC droplets was the possibility of manipulating the morphology of the droplets via preferential adsorption of fluorocarbon and hydrocarbon surfactants at the interfaces of the domains of the 57 droplets. As an initial step in this direction, we investigated single-phase nematic droplets to determine how SDS and perfluorooctanoic acid (PFOA) influence LC orientations at surfactant-decorated aqueous interfaces in the presence and absence of FB. First, we address the influence of FB on anchoring in the presence of SDS surfactant. Pure E7 dispersed in aqueous 100 µM SDS at 35 oC exhibited a “bipolar” configuration with two defects located at poles of the droplet[10,34] (Figure 3.6A). In contrast, the FB-E7 mixture (5% v/v FB) at the same temperature, and in the presence of 100 µM SDS, formed droplets with a director profile that radiated from a point defect located on the droplet surface (Figure 3.6B). This configuration is a so-called “pre- radial” configuration[10,34], consistent with a tilted alignment of LC at aqueous interface. We hypothesize that the presence of FB decreases the order of the LC at a given temperature, and thus facilitates SDS to promote the pre-radial configuration. As reported above, at higher concentrations of SDS (CSDS = 1 mM in Figure 3.6C and D and also CSDS = 2 mM), the nematic droplets comprised of E7 or FB-E7 both exhibited a radial configuration, as detailed above (Figure 3.5A). Second, in contrast to SDS, FB has only a minor impact on anchoring of E7 in the presence of PFOA surfactant at 35 oC. With aqueous 1 mM PFOA (Figure 3.6E and F), the droplets in the absence and presence of FB exhibited similar configurations, as evidenced by line defects (loops in schematic illustrations in Figure 3.6E and F) at a location between the equatorial plane and one of the poles of the droplets, revealing a tilted alignment of LC at aqueous interface. We note that the defect loop in the presence of FB was displaced towards the pole as compared to the LC free of FB. The results above lead to the conclusion that, in the presence of FB, the FB-E7 mixture assumes distinct configurations in the presence of 1 mM SDS solution (Figure 3.6D) and the 1 mM PFOA solution Figure 3.6F) due to differences in interfacial 58 anchoring. We note also that the addition of PFOA caused the pH to decrease from 7 to 3, which we confirmed to not impact our experimental observations by performing control experiment at pH = 7 (Figure S3.7). To provide boundary conditions for simulations (reported below) of the multi- phase droplets in the presence of 100 µM SDS and 1 mM PFOA, we simulated micrometer-sized single-phase nematic droplets with increasing tilt angles at the interface. We used strong anchoring conditions (W = 1×10-3 N m-1) in the simulations. Figure 3.7 Simulation for single-phase nematic droplets with increasing tilt angles at the interface. (A-C) Director fields and calculated polarized light micrographs for nematic droplets (R = 1 μm, W = 1×10-3 N m-1) with preferred surface tilt angles of (A) 30o, (B) 45o and (C) 60o, respectively. Defects are shown in red (isosurface for S=0.5). (D) Ratio of defect-to-droplet-center distance (d), normalized by droplet radius (R), as a function of 𝜃𝑒. 59 We define 𝜃𝑒 as the tilt angle of the easy axis of the LC with respect to surface normal. For 𝜃𝑒 = 0 o, the nematic droplet adopts a radial configuration with a small defect ring located in the droplet center. As 𝜃𝑒 increases, the spherical symmetry breaks and the defect migrates towards the droplet surface (escaped radial configuration in Figure 3.7A), until a pre-radial configuration forms for 𝜃𝑒 between 45 o and 55o (Figure 3.7B). As 𝜃𝑒 continues to increase, a morphological transition to an axial configuration occurs (Figure 3.7C), and, finally, it evolves to a bipolar structure for 𝜃𝑒 = 90 o. The dependence of defect position (d/R) on 𝜃𝑒 is shown in Figure 3.7D, where d is defined as the distance between defect to droplet center. The specific tilt angles of the FB-E7 mixture in the presence of surfactants were assigned as 45o, 60o, 0o and 0o in the presence of 100 µM SDS, 1 mM PFOA, 1 mM and 2 mM SDS, respectively, by comparing the droplet configurations obtained from experiments with simulations (Figure s3.6 and 7). We note one minor difference between our experimental observations and simulations: whereas the experiments with 1 mM PFOA led to ring defects that were displaced slightly from the equatorial plane (Figure 3.6E), our simulations predicted defect rings to lie on the equatorial plane (Figure 3.7C). The simulations contain a number of approximations, such as the use of one elastic constant, that likely underlie this difference. Surfactant-dependent configurations of multicompartment isotropic-LC droplets Next, we describe results obtained using biphasic droplets. We investigated the relative influence of hydrocarbon- and fluorocarbon-based surfactants on the morphogenesis of biphasic Janus droplets, as shown in Figure 3.8. We define θI, the angle between N-I and I-A interfaces, as the contact angle. When using PFOA, we observed 1 mM surfactant to lead to θI ~35o (Figure 3.8A), with a decrease in concentration of PFOA (0.5 mM) leading to an increase in θI (44o) (Figure S3.5). This trend suggests that the interfacial tension at the I-A interface decreases more than at the 60 Figure 3.8 Influence of surfactant type on the morphologies of biphasic E7-FB droplets. (A-D) Columns 1 and 2 are optical micrographs (single polarizer, and crossed polars, respectively) of Janus droplets in aqueous solutions containing (A) 1 mM PFOA or (B) 100 μM, (C) 1 mM, and (D) 2 mM SDS in N-I coexistence region. Scale bars, 10 µm. Columns 3 and 4 show corresponding simulated director profiles and calculated polarized light micrographs. (E) Contact angles θI in the presence of fluorocarbon surfactant (PFOA) versus hydrocarbon surfactant (SDS). N-A interface with increase in concentration of PFOA, a result that is consistent with preferential adsorption of fluorocarbon surfactant on the fluorocarbon-rich (isotropic) 61 domain. When SDS was used (Figure 3.8B-D), the effect of the surfactant was to reverse the trend: θI of biphasic Janus droplets increased from 54o to 125o with increase in SDS concentration. This result indicates preferential adsorption of SDS (and decrease of the interfacial tension) at the interface of the hydrocarbon-rich (nematic) domain. Overall, the opposite trend in partial wetting of the internal domains of the droplets (summarized in Figure 3.8E) is consistent with an unequal distribution of FB between isotropic and nematic compartments and thus preferential adsorption of fluorocarbon and hydrocarbon surfactants on the isotropic and nematic domains, respectively. Overall, the manipulation of temperature and surface anchoring enables access to a rich range of hierarchical structures within the Janus droplets, as shown in Figures S3.2-3.6. As discussed above, control experiments performed at both pH = 3 and pH = 7 established that the distinct droplet morphologies seen with PFOA and SDS are due to the surfactant type and not difference in pH (Figure S3.7). To understand the internal ordering of LC within the compartments, we performed numerical simulations with the LdG continuum model. We assumed that, at the N-I interface, the easy axis and the anchoring energy of LC were not affected in the presence of surfactants, that the tilt angle of the easy axis from the surface normal was 50o and that the anchoring energy was W = 1×10-5 N m-1 (as described above). The orientations of the LC at the aqueous interfaces were the same as those determined above from simulations, as shown in Figures 3.6 and 3.7. We assumed these orientations to be fixed by implementing a strong anchoring strength W = 1×10-3 N m- 1. Using the simulated internal order of the droplets (Column 3 in Figure 3.8A-D), we calculated the polarized light textures (Column 4 in Figure 3.8A-D) and compared them to the experimental observations. We found excellent agreement (see Columns 1 and 2 in Figure 3.8A-D). For example, optical textures in experiments and simulations possess no defects at N-I interfaces, and the extinction brushes and regions are 62 consistent. Additional comments on the free energies of two-phase LC-isotropic droplets Our observation that the biphasic Janus droplets with nematic and isotropic compartments assume spherical shapes suggests that γ13 ~ γ23 >> γ12 (Figure 3.1). To test this interpretation, we separated coexisting nematic and isotropic phases of a FB- E7 (8% v/v FB) mixture at 36 oC by centrifugation. We used the 8% v/v mixture to ensure that the N-I temperature interval was in the operational temperature range of the centrifuge (-10 oC to 40 oC). The density of the nematic phase (1.0462 g cm-3) was slightly lower than that of the isotropic phase (1.0487 g cm-3) at 36°C. Next we measured the interfacial tensions of the nematic-aqueous interface, γNA, the isotropic- aqueous interface, γIA, and the nematic-isotropic interface, γNI, using the pendant drop method. The results, shown in Figure 3.9, indicate that the orders of magnitude of γNA and γIA are 1 mN m-1, and the two interfacial tensions were numerically indistinguishable. In contrast, the magnitude of γNI was 10-2 mN m-1 (4 ± 2 × 10-2 mN m-1), which is two orders of magnitude smaller than γNA and γIA. The magnitude of γNI is in good agreement with that of other thermotropic LCs reported elsewhere[59]. Although the interfacial tensions of the N-I interfaces were low, the density difference between the two phases was also low, thus giving rise to accurate determination of interfacial tension based on analysis of droplet shape. Overall, the three interfacial tensions satisfy the thermodynamic criteria, γNA ~ γIA >> γNI, for spherical Janus droplets. Although the negative slopes of the plots shown in Figure 3.9 for both SDS and PFOA indicate a positive surface excess concentration of these surfactants at the aqueous interfaces, a quantitative interpretation is challenging for a number of reasons. For example, we note that PFOA may not completely dissociate at pH = 3, and that addition of surfactants may lead to redistribution of FB molecules in this FB-E7 multi- 63 component system. For droplets with size (l) of 10 μm, we use the results described above to calculate that the overall free energy of the N-I interface (~γl2) is ~10-15 J, and the anchoring energy (orientation-dependent contribution) is ~10-17 J; at the nematic- aqueous interface, the overall interfacial free energy (~γl2) is ~10-13 J, and the anchoring energy is ~10-15 J. When compared to the elastic energy (~Kl), which is ~10-17 J, the interfacial free energies dominate the shapes of nematic domains in the coexistence region for E7-FB mixtures, as evidenced by the spherical (Figure 3.4C) and cylindrical (Figure 3.4D) shape of the nematic domains. The dominating interfacial energies are Figure 3.9 Interfacial tensions between nematic-aqueous and isotropic- aqueous interfaces, γNA and γIA, respectively using the pendant drop method. Mean ± s.d. (n = 3) are shown. γNA and γIA are significantly larger than N-I interfacial tension, γN-I (4 ± 2 × 10-2 mN m-1). 64 also responsible for the spherical morphologies of the droplets comprising coexisting N-I phases dispersed in water (Figures 3.5 and 3.8). In addition, for the pendant drop measurements reported above (using droplets with l ~1 mm), the elastic energy (Kl ~10- 15 J) is unlikely to significantly influence the determination of interfacial tension (γl2 ~10-9-10-11 J). 4. Conclusions We report that mixtures of FB-E7 dispersed in an aqueous continuous phase form emulsion droplets that exhibit stable Janus morphologies comprised of coexisting nematic and isotropic oil domains. Partial miscibility of FB and E7 leads to a broad temperature range over which the nematic (N) and isotropic (I) phases coexist within the droplets, enabling facile and reversible control of the volumes of each compartment of the Janus droplets. We note that this characteristic of the system is challenging to realize in isotropic emulsion systems20. Significantly, the similarity in composition of the N and I domains within the droplets leads to N-I interfacial tensions that are lower than the aqueous interfacial tensions in the system, which results in droplets that maintain their spherical shape during changes in internal morphology induced by variations in temperature or interfacial adsorbates (surfactants). For the latter case, the droplet morphology was tuned by using hydrocarbon and fluorocarbon surfactants. As compared to isotropic multiphase droplets, the presence of the LC domain was shown to lead to morphology-dependent optical signatures. These optical signatures were largely reproduced by simulations based on a Landau-de Gennes free energy. Additional studies are needed, however, to understand why the additional of FB leads to changes in interfacial anchoring of the LC phases at aqueous interfaces. Although the majority of results presented in this paper were obtained using the nematic LC called E7, qualitatively similar results have been obtained with other nematic LCs, including 65 nematic phases of 5CB, 8CB and TL205. We anticipate that the approach demonstrated in this paper can be extended in future studies to LC phases other than achiral nematics, including cholesteric, smectic and blue phases. Ongoing experiments are also exploring the dynamics of the transitions between morphological states that are triggered by changes in surfactant concentration and type. We comment that the influence of surfactant architecture (e.g., use of bolaform surfactants) on the orientational anchoring of LCs may also yield morphologies of droplets that go beyond those reported here[60]. Overall, the results reported in this paper advance the design of reconfigurable soft matter systems based on isotropic perfluorocarbon oils and nematic LCs. 5. Supporting Information Thermodynamics of phase coexistence Our observations of coexisting nematic and isotropic phases (as shown in Figure 3.3 of the main manuscript) is consistent with the Gibb’s phase rule. The number of independent compositional degrees of freedom (C) is 2 for perfluorobenzene (FB) and E7 (E7 is a mixture of 4 components, but they are not independent because the composition of E7 is constant). The number of phases (P) is 2 (nematic and isotropic). It follows that the degrees of freedom (F) are equal to C+2-P=2, thus allowing the existence of coexisting nematic and isotropic phases. For nematic (N) and isotropic (I) phases at equilibrium, the solute (FB) chemical potential in each of the two phases must be equal[61]: γ𝑁𝐶 𝑁 𝐼 𝐼𝐹𝐵 = γ 𝐶𝐹𝐵 where γ𝑁 and γ𝐼 are activity coefficients of the solute in the nematic phase and isotropic phase, respectively, and 𝐶𝑁𝐹𝐵 and 𝐶 𝐼 𝐹𝐵 are the concentrations of solute. Accordingly, 𝐶 𝐼 𝑁𝐹𝐵 γ 𝐶 𝑁 = , 𝐹𝐵 γ 𝐼 66 γ𝑁 Inspection of Figure 3.3 of the main manuscript reveals that 𝐼 is larger for FB γ than toluene in the two-phase system based on E7. We speculate that this difference arises from electrostatic interactions between FB and the aromatic components of E7 that are absent for toluene (see main text for details). 67 Figure S3.1 Curvature of the N-I interface. Top row are micrographs of a N-I interface that expands both glass surfaces of an optical cell with changing focal plane from top to bottom of a FB-E7 (CFB=5 % v/v) mixture at 44.0 oC (Nematic-Isotropic coexistence). Scale bar, 10 µm. 68 Figure S3.2 FB-E7 (CFB=5 % v/v) droplet in 100 µM SDS aqueous solution. (A-E) Top and bottom rows are micrographs of a LC droplet upon heating and cooling across the N-I biphasic region at (A) 30.0 oC (Nematic), (B) 42.5 oC, (C) 43.5 oC, (D) 44.3 oC, (E) 45.0 oC (Isotropic), respectively. Scale bar, 10 µm. 69 Figure S3.3 FB-E7 (CFB=5 % v/v) droplet in 1 mM SDS aqueous solution. (A-E) Top and bottom rows are micrographs of a LC droplet upon heating and cooling across the N-I coexisting region at (A) 35.0 oC (Nematic), (B) 45.0 oC, (C) 46.0 oC, (D) 47.0 oC, (E) 47.5 oC (Isotropic), respectively. Scale bar, 10 µm. 70 Figure S3.4 FB-E7 (CFB=5 % v/v) droplet in 2 mM SDS aqueous solution. (A-E) Top and bottom rows are micrographs of a LC droplet upon heating and cooling across the N-I biphasic region at (A) 41.0 oC (Nematic), (B) 42.3 oC, (C) 43.0oC, (D) 43.8 oC, (E) 45.0 oC (Isotropic), respectively. Scale bar, 10 µm. 71 Figure S3.5 FB-E7 (CFB=5 % v/v) droplet in 0.5 mM PFOA aqueous solution. (A- E) Top and bottom rows are micrographs of a LC droplet upon heating and cooling across the N-I biphasic region at (A) 35.0 oC (Nematic), (B) 40.5 oC, (C) 41.5oC, (D) 42.0 oC, (E) 42.5 oC (Isotropic), respectively. Scale bar, 10 µm. 72 Figure S3.6 FB-E7 (CFB=5 % v/v) droplet in 1 mM PFOA aqueous solution. (A-E) Top and bottom rows are micrographs of a LC droplet upon heating and cooling across the N-I biphasic region at (A) 35.0 oC (Nematic), (B) 44.3 oC, (C) 45.0 oC, (D) 45.5 oC, (E) 46.8 oC (Isotropic), respectively. Scale bar, 10 µm. 73 Figure S3.7 A comparison of FB-E7 droplets at (A-F) pH = 3 and (G-L) pH = 7. Droplets both (A,C,E,G,I,K) in single phase state and (B,D,F,H,J,K) in biphasic state in the presence of (A,B,G,H) 1 mM PFOA, (C,D,I,J) 1 mM SDS and (E,F,K,L) 2 mM SDS. Scale bar, 10 µm. pH value was adjusted using sodium hydroxide and hydrochloric acid. 74 Figure S3.8 Simulated schematic illustrations (row 2) and polarized light micrographs (row 3) for multi-compartment droplets under the influence of strong anchoring strength at N-I interfaces. Defect lines (blue loops schematic illustrations) were generated. Acknowledgements The authors acknowledge support from the Department of Energy, Basic Energy Sciences, Division of Materials Research, Biomaterials Program under Grant No. DE- SC0004025. YY acknowledges partial fellowship support from the Department of Chemical and Biological Engineering at University of Wisconsin-Madison. 6. References *This chapter was prepared as a Research Article reporting original research in the journal Soft Matter. My contribution to this project was in designing and performing the experiments, designing the simulations and writing the manuscript. My co-authors contributed to the designing of the project, experiments, designing and performing the simulations and writing of the manuscript. 75 Adapted with permission from: X. Wang, Y. Zhou, Y.-K. Kim, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Thermally Reconfigurable Janus Droplets with Nematic Liquid Crystalline and Isotropic Perfluorocarbon Oil Compartments, Soft Matter 15, 2580 (2019). Copyright The Royal Society of Chemistry 2019. [1] K. J. Lissant, Emulsions and Emulsion Technology, vol. 6 (Marcel Dekker, New York, 1974). [2] M. A. Augustin and Y. Hemar, Nano- and Micro-Structured Assemblies for Encapsulation of Food Ingredients, Chem. Soc. Rev. 38, 902 (2009). [3] P. A. Gresham, M. Barnett, S. V. Smith, and R. Schneider, Use of a Sustained- Release Multiple Emulsion to Extend the Period of Radioprotection Conferred by Cysteamine [9], Nature 234, 149 (1971). [4] E. Bukusoglu, M. Bedolla Pantoja, P. C. Mushenheim, X. Wang, and N. L. Abbott, Design of Responsive and Active (Soft) Materials Using Liquid Crystals, Annu. Rev. Chem. Biomol. Eng. 7, 163 (2016). [5] X. Wang, E. Bukusoglu, D. S. Miller, M. A. Bedolla Pantoja, J. Xiang, O. D. Lavrentovich, and N. L. Abbott, Synthesis of Optically Complex, Porous, and Anisometric Polymeric Microparticles by Templating from Liquid Crystalline Droplets, Adv. Funct. Mater. 26, 7343 (2016). [6] A. S. Utada, E. L. Lorenceau, D. R. Link, P. D. Kaplan, H. A. Stone, and D. A. Weitz, Monodisperse Double Emulsions Generated from a Microcapillary Device, Science 308, 537 (2005). [7] N. Pannacci, H. Bruus, D. Bartolo, I. Etchart, T. Lockhart, Y. Hennequin, H. Willaime, and P. Tabeling, Equilibrium and Nonequilibrium States in Microfluidic Double Emulsions, Phys. Rev. Lett. 101, 1 (2008). [8] J. A. Hanson, C. B. Chang, S. M. Graves, Z. Li, T. G. Mason, and T. J. Deming, 76 Nanoscale Double Emulsions Stabilized by Single-Component Block Copolypeptides, Nature 455, 85 (2008). [9] P. S. Drzaic, Liquid Crystal Dispersions.Vol.1 (World Scientific, 1995). [10] J. K. Gupta, J. S. Zimmerman, J. J. De Pablo, F. Caruso, and N. L. Abbott, Characterization of Adsorbate-Induced Ordering Transitions of Liquid Crystals within Monodisperse Droplets, Langmuir 25, 9016 (2009). [11] J. A. Moreno-Razo, E. J. Sambriski, N. L. Abbott, J. P. Hernández-Ortiz, and J. J. De Pablo, Liquid-Crystal-Mediated Self-Assembly at Nanodroplet Interfaces, Nature 485, 86 (2012). [12] E. Bukusoglu, X. Wang, Y. Zhou, J. A. Martínez-González, M. Rahimi, Q. Wang, J. J. de Pablo, and N. L. Abbott, Positioning Colloids at the Surfaces of Cholesteric Liquid Crystal Droplets, Soft Matter 12, 8781 (2016). [13] I.-H. Lin, D. S. Miller, P. J. Bertics, C. J. Murphy, J. J. de Pablo, and N. L. Abbott, Endotoxin-Induced Structural Transformations in Liquid Crystalline Droplets, Science 332, 1297 (2011). [14] Y.-K. Kim, X. Wang, P. Mondkar, E. Bukusoglu, and N. L. Abbott, Self- Reporting and Self-Regulating Liquid Crystals, Nature 557, 539 (2018). [15] T. Bera and J. Fang, Optical Detection of Lithocholic Acid with Liquid Crystal Emulsions, Langmuir 29, 387 (2013). [16] L. E. Aguirre, A. de Oliveira, D. Seč, S. Čopar, P. L. Almeida, M. Ravnik, M. H. Godinho, and S. Žumer, Sensing Surface Morphology of Biofibers by Decorating Spider Silk and Cellulosic Filaments with Nematic Microdroplets, Proc. Natl. Acad. Sci. 113, 1174 (2016). [17] S. Torza and S. G. Mason, Coalescence of Two Immiscible Liquid Drops, Science 163, 813 (1969). [18] J. Guzowski, P. M. Korczyk, S. Jakiela, and P. Garstecki, The Structure and 77 Stability of Multiple Micro-Droplets, Soft Matter 8, 7269 (2012). [19] T. Nisisako and T. Hatsuzawa, A Microfluidic Cross-Flowing Emulsion Generator for Producing Biphasic Droplets and Anisotropically Shaped Polymer Particles, Microfluid. Nanofluidics 9, 427 (2010). [20] L. D. Zarzar, V. Sresht, E. M. Sletten, J. A. Kalow, D. Blankschtein, and T. M. Swager, Dynamically Reconfigurable Complex Emulsions via Tunable Interfacial Tensions, Nature 518, 520 (2015). [21] X. Wang et al., Patterned Surface Anchoring of Nematic Droplets at Miscible Liquid–Liquid Interfaces, Soft Matter 13, 5714 (2017). [22] F. Mondiot, X. Wang, J. J. de Pablo, and N. L. Abbott, Liquid Crystal-Based Emulsions for Synthesis of Spherical and Non- Spherical, J. Am. Chem. Soc. 135, 9972 (2013). [23] M. A. Gharbi, M. Nobili, and C. Blanc, Use of Topological Defects as Templates to Direct Assembly of Colloidal Particles at Nematic Interfaces, J. Colloid Interface Sci. 417, 250 (2014). [24] X. Wang, D. S. Miller, J. J. de Pablo, and N. L. Abbott, Organized Assemblies of Colloids Formed at the Poles of Micrometer-Sized Droplets of Liquid Crystal, Soft Matter 10, 8821 (2014). [25] X. Wang, D. S. Miller, J. J. De Pablo, and N. L. Abbott, Reversible Switching of Liquid Crystalline Order Permits Synthesis of Homogeneous Populations of Dipolar Patchy Microparticles, Adv. Funct. Mater. 24, 6219 (2014). [26] J. K. Whitmer, X. Wang, F. Mondiot, D. S. Miller, N. L. Abbott, and J. J. De Pablo, Nematic-Field-Driven Positioning of Particles in Liquid Crystal Droplets, Phys. Rev. Lett. 111, 1 (2013). [27] M. Urbanski, C. G. Reyes, J. Noh, A. Sharma, Y. Geng, V. S. R. Jampani, and J. P. F. Lagerwall, Liquid Crystals in Micron-Scale Droplets, Shells and Fibers, J. Phys. 78 Condens. Matter 29, 133003 (2017). [28] E. Bukusoglu, X. Wang, J. A. Martinez-Gonzalez, J. J. De Pablo, and N. L. Abbott, Stimuli-Responsive Cubosomes Formed from Blue Phase Liquid Crystals, Adv. Mater. 27, 6892 (2015). [29] A. Fernández-Nieves, D. R. Link, M. Márquez, and D. A. Weitz, Topological Changes in Bipolar Nematic Droplets under Flow, Phys. Rev. Lett. 98, 1 (2007). [30] D. Seč, T. Porenta, M. Ravnik, and S. Žumer, Geometrical Frustration of Chiral Ordering in Cholesteric Droplets, Soft Matter 8, 11982 (2012). [31] Y. K. Kim, S. V. Shiyanovskii, and O. D. Lavrentovich, Morphogenesis of Defects and Tactoids during Isotropic-Nematic Phase Transition in Self-Assembled Lyotropic Chromonic Liquid Crystals, J. Phys. Condens. Matter 25, 404202 (2013). [32] Y.-K. Kim, B. Senyuk, S.-T. Shin, A. Kohlmeier, G. H. Mehl, and O. D. Lavrentovich, Surface Alignment, Anchoring Transitions, Optical Properties, and Topological Defects in the Thermotropic Nematic Phase of Organo-Siloxane Tetrapodes, Soft Matter 10, 500 (2014). [33] Y. Zhou, E. Bukusoglu, J. A. Martínez-González, M. Rahimi, T. F. Roberts, R. Zhang, X. Wang, N. L. Abbott, and J. J. De Pablo, Structural Transitions in Cholesteric Liquid Crystal Droplets, ACS Nano 10, 6484 (2016). [34] G. E. Volovik and O. D. Lavrentovich, Topological Dynamics of Defects: Boojums in Nematic Drops, Sov. Phys. JETP 58, 1159 (1983). [35] P. Poulin and D. A. Weitz, Inverted and Multiple Nematic Emulsions, Phys. Rev. E 57, 626 (1998). [36] T. Lopez-Leon and A. Fernandez-Nieves, Topological Transformations in Bipolar Shells of Nematic Liquid Crystals, Phys. Rev. E 79, 021707 (2009). [37] T. Lopez-Leon, V. Koning, K. B. S. Devaiah, V. Vitelli, and A. Fernandez- Nieves, Frustrated Nematic Order in Spherical Geometries, Nat. Phys. 7, 391 (2011). 79 [38] D. Seč, T. Lopez-Leon, M. Nobili, C. Blanc, A. Fernandez-Nieves, M. Ravnik, and S. Žumer, Defect Trajectories in Nematic Shells: Role of Elastic Anisotropy and Thickness Heterogeneity, Phys. Rev. E 86, 020705 (2012). [39] H. Liang, J. Noh, R. Zentel, P. Rudquist, P. F. Jan, and J. P. Lagerwall, Tuning the Defect Configurations in Nematic and Smectic Liquid Crystalline Shells, Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 371, 20120258 (2013). [40] T. Lopez-Leon, A. Fernandez-Nieves, M. Nobili, and C. Blanc, Smectic Shells, J. Phys. Condens. Matter 24, 284122 (2012). [41] A. Darmon, M. Benzaquen, D. Seč, S. Čopar, O. Dauchot, and T. Lopez-Leon, Waltzing Route toward Double-Helix Formation in Cholesteric Shells, Proc. Natl. Acad. Sci. U.S.A. 113, 9469 (2016). [42] J. Jeong, A. Gross, W.-S. Wei, F. Tu, D. Lee, P. J. Collings, and A. G. Yodh, Liquid Crystal Janus Emulsion Droplets: Preparation, Tumbling, and Swimming, Soft Matter 11, 6747 (2015). [43] G. A. Oweimreen and M. Hasan, The Effect of Quasispherical Solutes on the Nematic to Isotropic Transition in 7CB, Mol. Cryst. Liq. Cryst. 100, 357 (1983). [44] I. M. Tang and S. Denprayoonwong, On the Nature of the Biphasic Region in Binary Nematic Liquid Crystal Mixtures, Phys. Lett. A 127, 435 (1988). [45] M. A. Gidley and D. Stubley, Phase Diagrams of Some Mixtures of (a Nematic Liquid Crystal + an Ordinary Liquid), J. Chem. Thermodyn. 14, 785 (1982). [46] M. A. Gidley and D. Stubley, Activity Solutes Coefficients at Infinite Dilution in Liquid-Crystalline Solvents, J. Chem.Thermodynamics 18, 595 (1986). [47] P. G. de Gennes and J. Prost, The Physics of Liquid Crystals (Clarendon Press, 1993). [48] M. Kleman and O. D. Lavrentovich, Soft Matter Physics : An Introduction (Springer, New York, NY, USA, 2003). 80 [49] J. B. Fournier and P. Galatola, Modeling Planar Degenerate Wetting and Anchoring in Nematic Liquid Crystals, Europhys. Lett. 72, 403 (2005). [50] M. Ravnik and S. Žumer, Landau-de Gennes Modelling of Nematic Liquid Crystal Colloids, Liq. Cryst. 36, 1201 (2009). [51] R. Ondris-Crawford, E. P. Boyko, B. G. Wagner, J. H. Erdmann, S. Žumer, and J. W. Doane, Microscope Textures of Nematic Droplets in Polymer Dispersed Liquid Crystals, J. Appl. Phys. 69, 6380 (1991). [52] D. S. Seo, K. I. Muroi, T. R. Isogami, H. Matsuda, and S. Kobayashi, Polar Anchoring Strength and the Temperature Dependence of Nematic Liquid Crystal (5CB) Aligned on Rubbed Polystyrene Films, Jpn. J. Appl. Phys. 31, 2165 (1992). [53] S. D. Kim, B. Lee, S. W. Kang, and J. K. Song, Dielectrophoretic Manipulation of the Mixture of Isotropic and Nematic Liquid, Nat. Commun. 6, 1 (2015). [54] K. Shimizu, M. F. C. Gomes, A. A. H. Padua, L. P. N. Rebelo, and J. N. C. Lopes, On the Role of the Dipole and Quadrupole Moments of Aromatic Compounds in the Solvation by Ionic Liquids, J. Phys. Chem. B 113, 9894 (2009). [55] C. R. Patrick and G. S. Prosser, A Molecular Complex of Benzene and Hexafluorobenzene, Nature. [56] Y.-K. Kim, G. Cukrov, J. Xiang, S.-T. Shin, and O. D. Lavrentovich, Domain Walls and Anchoring Transitions Mimicking Nematic Biaxiality in the Oxadiazole Bent- Core Liquid Crystal C7., Soft Matter 11, 3963 (2015). [57] J. M. Brake and N. L. Abbott, An Experimental System for Imaging the Reversible Adsorption of Amphiphiles at Aqueous-Liquid Crystal Interfaces, Langmuir 18, 6101 (2002). [58] X. Wang, D. S. Miller, E. Bukusoglu, J. J. de Pablo, and N. L. Abbott, Topological Defects in Liquid Crystals as Templates for Molecular Self-Assembly, Nat. Mater. 15, 106 (2016). 81 [59] S. Faetti and V. Palleschi, Nematic-Isotropic Interface of Some Members of the Homologous Series of 4-Cyano-4-(n-Alkyl)Biphenyl Liquid Crystals, Phys. Rev. A 30, 3241 (1984). [60] F. P. Hubbard and N. L. Abbott, Effect of Light on Self-Assembly of Aqueous Mixtures of Sodium Dodecyl Sulfate and a Cationic, Bolaform Surfactant Containing Azobenzene, Langmuir 23, 4819 (2007). [61] G. R. Luckhurst and G. W. Gray, eds., The Molecular Physics of Liquid Crystal, Academic Press, 1979. 82 Chapter 4 Reconfigurable Multicompartment Emulsion Drops Formed by Nematic Liquid Crystals and Immiscible Perfluorocarbon Oils 1. Introduction Liquid crystals (LCs) are a phase of matter within which the constituent molecules exhibit long-range orientational order and high levels of mobility[1,2]. Confinement of LCs (e.g., to thin films[3–5], emulsions[6–12] and spherical shells[13– 18]) imposes constraints on the orientational ordering of LCs, leading to mesoscale organizations that reflect a delicate balance of contributions to the free energy arising from orientation-dependent interfacial free energies, elastic strain associated with deformation of LCs, and the presence of singular topological defects [6,7,14,19–27]. The delicate balance of energetics that controls the ordering of confined LCs have led to their use in a range of contexts[1,2], including recently for chemical and biological sensing[5,28–30], in design of autonomous soft matter microsystems[31–34] and for templating molecular assemblies[21,35]. In this manuscript, we explore how immiscible perfluorocarbons can be used to confine LCs, including within domains of emulsion droplets with core-shell and Janus-type morphologies[9,10,36–38], in ways that are not accessible in conventional single-phase emulsions[6,7,18,39]. Past studies have reported that a range of organizations of nematic LCs emerge within, for example, thin shells confined between two immiscible aqueous phases as a function of surface orientational ordering (anchoring), thickness of the shells, as well as variation in shell thickness[8,13,15,17]. Topological defects form within the LC shells, depending on the boundary conditions (anchoring) imposed by the two aqueous surfaces. The resulting defects can organize into bipolar, triangular or tetrahedral arrangements, as documented by both experiments and numerical simulations[13– 83 15,17]. The exploration of LC ordering within the confines of shells also extends to cholesteric and smectic phases, leading to a rich range of shell textures and defect configurations[8,16,18]. In this paper, we reveal that anchoring conditions imposed by immiscible perfluoroalkane phases enable access to LC organizations within shells (and domains defined by Janus morphologies) that have not previously been reported. Whereas past studies have focused on LCs confined by aqueous phases, including aqueous-glycerol mixtures containing surfactants[8,13–16], we used immiscible perfluorocarbons to confine LCs for three reasons. First, in advance of this study, it was not obvious to us what orientations LCs would adopt at immiscible perfluorocarbon interfaces. Past studies have explored the ordering of LCs at the interfaces of solid perfluorocarbons[3,4,40,41]. The anchoring (easy axis), however, largely depends on the method used to prepare the surfaces. Plasma deposition of perfluorocarbon films induces perpendicular alignment[3,4] while the anchoring was Figure 4.1 Molecular structures of the nematogens (A) 4′-pentyl-4- biphenylcarbonitrile (5CB), (B) 4-(trans-4-pentylcyclohexyl) benzonitrile (PCH5); (C) perfluorononane (F9) and (D) perfluoroheptane (F7). 84 found to be planar (parallel) on a surface deposited by rubbing a Teflon bar[40,41]. In this manuscript, we explore LC alignments at liquid perfluoroalkane interfaces using the nematogens 4′-pentyl-4-biphenylcarbonitrile (5CB) and 4-(trans-4- pentylcyclohexyl) benzonitrile (PCH5) and isotropic perfluorocarbon oils perfluorononane (F9) and perfluoroheptane (F7) (Figure 4.1). We compare the behaviors of 5CB and PCH5 in contact with F9 because 5CB is known to form smectic layering at interfaces whereas the cyclohexyl ring of PCH5 frustrates smectic layering[42]. This comparison thus allows us to probe the role of smectic layering in determining the orientations of LCs at liquid perfluorocarbon interfaces. Second, we used perfluorocarbons to prepare multiphase emulsions because our initial observations revealed that they provide access to a range of LC organizations in emulsion droplets that have not been seen previously. In particular, as detailed below, we found that 5CB molecules anchor weakly at perfluorocarbon interfaces and thus can be readily reoriented by elastic strain of the LC. This yields new configurations of LCs within shells that reflect an interplay between elastic energy, geometry of confinement and anchoring condition. We explored and provide insight into this interplay by comparing our experimental observations to numerical simulations of multidomain LC droplets using a Landau-de Gennes description of the free energy. The use of perfluorocarbons also provided an opportunity to compare the influence of perfluorocarbon versus hydrocarbon surfactant type on emulsion droplet morphology and internal organization[37,43,44]. The third reason we used perfluorocarbon oils in the experiments reported in this paper was because perfluorocarbons can be readily driven through phase transitions to create vapor-filled cavities within emulsion droplets[45–47]. We sought to determine if such phase transitions could be used to drive rapid changes in the shapes of the LC domains of the multiphase droplets explored in our study. As described below, we 85 found that it was possible to rapidly thin nematic shells (formed from a high clearing temperature nematic LC called HTW) by thermally driving liquid perfluorocarbon cores (formed from F7) through liquid-to-vapor phase transitions. These results provide fresh ideas for actuation of LC emulsion droplets in ways that may provide insight into the dynamic properties of LCs in confinement. They also hint at new designs of stimuli- responsive LC systems driven far from equilibrium. 2. Experimental Methods Materials. The nematogens, 4′-pentyl-4-biphenylcarbonitrile (5CB, nematic phase from 22.0 oC to 35.5 oC) and HTW45800-000 (HTW, nematic phase from -30 oC to 101 oC), were purchased from HCCH (Jiangsu Hecheng Display Technology Co., LTD). Perfluorononane (97%, F9), the nematogen 4-(trans-4-pentylcyclohexyl) benzonitrile (99%, PCH5, nematic phase from 30 oC to 55 oC), sodium dodecyl sulfate (99.0%, SDS), perfluorooctanoic acid (96%, PFOA), trichloro(1H,1H,2H,2H- perfluorooctyl)silane (97%) and glycerol (≥99.5%) were purchased from Sigma– Aldrich (St. Louis, MO, USA). Perfluoroheptane (mixture of isomers, 98%, F7) was purchased from Alfa Aesar, Thermo Fisher Scientific (Tewksbury, MA, USA). Fisher Finest Premium Grade glass slides was purchased from Fisher Scientific (Pittsburgh, PA). Polyimide 2555, which was used to induce planar alignment of 5CB when coated on glass substrates, was purchased from HD Microsystems. Purification of water (18.2 MΩ cm resistivity at 25 °C) was performed using a Milli-Q water system (Millipore, Bedford, MA, USA). F9 and F7 were stored and degassed by freezing at -20 oC and -80 oC, respectively, prior to use. Preparation of Optical Cells. Optical cells were assembled from two glass slides separated from each other by ~100-300 μm-thick spacers. Glass slides used for observation of 5CB droplets in F9 continuous phases were coated with 86 trichloro(1H,1H,2H,2H-perfluorooctyl) silane to make them fluorophilic; glass slides were immersed into trichloro(1H,1H,2H,2H-perfluorooctyl) silane, subsequently rinsed with ethanol, F9 and dried with compressed air. Glass slides used for observations of F9 droplets in 5CB phases were spin-coated with polyimide 2555 using a Laurell spin coater, baked at 250 oC; the two glass surfaces of cells were rubbed and assembled in an anti-parallel fashion to yield unidirectional tangential alignment of LCs for F9 droplet dispersed in 5CB. Characterization of Anchoring of Nematic 5CB at interfaces of isotropic F9. We prepared 5CB (or PCH5) droplets in a F9 continuous phase by vortexing 20 μl of 5CB (PCH5) and 200 μl of F9 at 3000 rpm for 30 seconds. The resulting mixture was injected into a 100 μm-thick silane-coated optical glass cell at elevated temperatures (isotropic LCs), and cooled to room temperature for 5CB (or 35 oC for PCH5) for observation. PCH5 was preheated to isotropic state above 60 oC, and remained fluidic during the preparation process because of supercooling. To prepare F9 droplets in a 5CB-rich continuous phase, 4 μl of F9 and 200 μl of 5CB was homogenized for 10 seconds using a probe ultrasonicator (60 Sonic Dismembrator, Fisher Scientific) with a cold-water bath. The resulting mixture was injected into a 100 μm-thick optical cell, coated with polyimide 2555 on both surfaces to induce a uniform alignment of LCs. We injected the mixture in isotropic state at 40 oC and cooled it to room temperature. The samples were observed using polarized light microscopy. Preparation of Emulsions using perfluorocarbon and LC mixtures. To prepare droplets of F9-5CB, we added 50-100 μl 5CB and 50 μl F9 into glycerol in the absence of surfactants (or aqueous solutions in the presence of surfactants). The mixtures with glycerol were subsequently homogenized using the probe ultrasonicator with a cold water bath. The mixtures with aqueous solutions were vortexed at 3000 rpm 87 for 30 seconds. The resulting mixtures were injected into 100 μm-thick optical cells at room temperature. The samples were heated above the LC clearing temperature (35.5 oC) and cooled to room temperature to equilibrate prior to making observations using optical microscopy. We observed the formation of droplets with core-shell structure as described below. We focused our observations on droplets between 10 and 30 micrometers (external diameter), and 3-10 micrometers (internal). To prepare F7-HTW droplets, we dispersed 80 μl HTW and 20 μl F7 into glycerol by sonicating the mixture with a cold water bath. 30 μl of the resulting mixtures were diluted into 1000 μl glycerol, subsequently injected into 300 μm-thick optical cells at room temperature and heated to vaporize the F7 cores using a Linkam LTS350 hot stage with an accuracy of 0.1 oC. Measurements of Interfacial Tensions. The interfacial tensions were measured using the pendant drop method[48] performed with an optical tensiometer (Attension Theta T200, Biolin Scientific). Hamilton metal 22-gauge needles (blunt point) were used. The equilibration time was 5 min for each drop. The results presented in this paper were obtained using 5 independent measurements. An environmental chamber heated by a fluid bath was used to control temperature. The densities used to analyze the pendant drop measurements were 1.0258 g cm-3 (5CB), 0.9700 g cm-3 (PCH5) 1.2613 g cm-3 (glycerol, room temperature), 1.2522 g cm-3 (glycerol, 35 oC)[49], 1.7990 g cm- 3 (F9, room temperature), 1.7633 g cm-3 (F9, 35 oC)[50], and 0.9982 g cm-3 (aqueous solutions). The density of 5CB was measured to be 1.025753 ± 0.000073 g cm-3 using an Anton Paar DSA 5000 M density meter at 20 oC. Simulations. We used a Landau-de Gennes (LdG) continuum model for the order tensor Q, defined by Qij=S (ni nj -1/3 ij), where, ni are the x, y, z components of the local director vector[1]. The scalar order parameter is denoted by S. The bulk free energy is given by 88 𝐴 𝑈 𝐴𝑈 𝐴𝑈 2 Fbulk(Q)=∫ ( (1 − ) 𝑄 𝑄2 3 𝑖𝑗 𝑗𝑖 − 𝑄𝑖𝑗𝑄𝑗𝑘𝑄𝑘𝑖 + (𝑄 𝑄 ) ) 𝑑𝑉 bulk 3 4 𝑖𝑗 𝑗𝑖 𝐿 ∂Q ∂Qij ij +∫ 𝑑𝑉 (1) bulk 2 ∂xk ∂xk where A and U are phenomenological parameters that depend on temperature and pressure. The one constant representation is adopted for the results reported in this paper, where L denotes the elastic constant of the LC (the symbol K is used for the elastic constant in Frank-Oseen elastic energy, where K = 2LS2). The first term in equation (1), which corresponds to enthalpic contributions to the free energy, serves to control the equilibrium value of the order parameter 1 8 Seq= (1 + 3√1− ) 4 3U The second term represents the elastic free energy, which governs long-range director distortions[2]. The surface free energy, which represents the degenerate conic anchoring, was introduced in our previous work[37,51]: 2 Fsurf(Q)=∫ W(P Q̃ P -S cos 2 ik lj eq θe Pij) dΣ (2) surf kl where W is the anchoring strength; Q̃ =Q +(1/3) Seqδij ; the preferred tilting ij ij angle between surface normal 𝜈 and the ‘easy cone’ is denoted by 𝜃𝑒; and the operator tensor Pij=νiνj. The definition of operator tensor P here is different from that in the Fournier-Galatola degenerate planar surface functional[52]. An iterative Ginzburg- Landau relaxation with finite difference method on a cubic mesh (with resolution of 7.15 nm) is adopted to minimize the free energy[53]. Polarization micrographs were calculated using the Jones matrix formalism, in which light of a single wavelength (350 nm) travels along a chosen direction and the total phase shift is accumulated.[54] The numerical parameters used in this work are A = 1.067  105 J m-3, U = 3.5, L = 6 pN, K = 4.55 pN for 5CB[37,51]. The droplet radii, r and R, and the surface anchoring 89 strengths were chosen by comparison of predictions of the simulations to experimental measurements. 3. Results and Discussion Anchoring of 5CB and PCH5 at F9 interfaces. Many perfluoroalkanes and isotropic hydrocarbons are immiscible at low temperatures, and mix above their upper consolute temperatures[43,55]. Perfluoromethylcyclohexane and hexane, for example, form two phases at 0 oC at a 1:1 volume ratio, but form one phase at room temperature[55]. At the outset of our study, we surveyed a range of perfluoroalkanes for the design of multi-compartment LC emulsions, including perfluorohexane (F6), perfluoroheptane (F7) and perfluorononane (F9). These perfluorocarbons form emulsions with similar morphologies, as described below, but we chose to focus on F9 because it can be readily degassed by freezing at -20 oC. In the absence of degassing, we found that gas bubbles were released from the oils when heating them during studies of phase behaviors. Unlike mixtures of perfluorocarbons and hydrocarbons described above, we found that the F9-5CB mixture (1:1 volume ratio) forms two phases over the temperature range 22 oC to 120 oC, which includes the temperature range of the nematic phase of 5CB (22 oC to 34 oC). We also found that a finite amount of F9 can dissolve into 5CB, with the mixture F9:5CB = 1:100 (volume ratio) forming a single phase above 50.5 ± 3 oC. This level of miscibility, although low, has the potential to change the properties (e.g., elastic constants) of the nematic phase. It is noteworthy that the phase behavior of the F9-LC mixtures is different with the mixtures of perfluoroarenes (e.g., hexafluorobenzene, FB) and hydrocarbon liquid crystals (e.g., E7) due to the polarizable nature of perfluoroarenes[37,55]. The FB-E7 mixture (e.g., 5:100 volume ratio) forms a single nematic phase below a critical temperature and forms coexisting nematic- 90 isotropic phases at elevated temperatures. As described below, in the study reported in this paper, we used the largely immiscible mixtures of F9 and 5CB to form emulsions. Initially, we prepared emulsions using F9 and 5CB, characterized the morphologies of the F9 or 5CB emulsion drops and inferred the orientational ordering of 5CB at F9 interfaces based on observation of the internal organization of the LC domains. To this end, we examined nematic 5CB droplets dispersed in a continuous phase of F9 (Figures 4.2A-C) as well as droplets of F9 dispersed in continuous phases of nematic 5CB (Figure 4.2D). To make observations of 5CB droplets in F9, we emulsified 10% v/v 5CB into F9 at room temperature, and performed experiments using films of the F9-5CB mixture confined within optical cells with surfaces that were treated to be fluorophilic (see Experimental Methods). Our experiments revealed that nematic phases of 5CB assume a perpendicular (homeotropic) alignment (easy axis) at the interfaces of F9, as indicated by the so-called radial configuration with a point defect at the center of the droplets[28,56] (Figure 4.2A). We also examined the influence of 5CB droplet size on LC ordering. The droplet in Figure 4.2B with a 1.4 μm radius generates four brushes under crossed polars (consistent with a radial configuration), while the droplet in Figure 4.2C with a 0.9 μm radius generates a birefringent texture that disappeared periodically during droplet rotation (inconsistent with a radial configuration; Figure S4.1). The latter result suggests that the smaller droplet has adopted a configuration with a uniform LC orientation. Overall, these observations suggest that a decrease in 5CB droplet radius below a critical value of ~1 μm caused a spontaneous transition from a radial to a uniform director profile within the 5CB domain. Past studies have demonstrated that the elastic energy of a LC droplet scales with the droplet radius (∼KR) and the surface anchoring energy scales with the square of the droplet radius (∼WR2)[7,9]. These two contributions to the free energy lead to the prediction that droplets with R ≤ K/W will exhibit a director 91 Figure 4.2 Micrographs between parallel polars (left), crossed polars (middle), and schematic illustrations (right) of (A-C) 5CB droplets in a continuous phase of F9 and (D) an F9 droplet dispersed in 5CB at room temperature. (B,C) size- dependence of LC ordering in droplets with radii of 1.4 μm and 0.9 μm, respectively. (E) Single-polar, crossed-polar micrographs and schematic illustration of a PCH5 droplet in F9. The defect denoted by the blue arrow is slightly tilted out of the plane. Scale bars, 10 μm. The double-headed arrows indicate the orientation of the two polarizers. The dash lines show the ordering of 5CB (director). The red dots indicate the defects. profile dominated by elastic energy whereas larger droplets will satisfy the perpendicular anchoring and form the radial configuration. These predictions are consistent with our experimental observations, and permit us to estimate that the 92 anchoring coefficient W of 5CB at F9 interface is 10-5 N/m, by using K of 5CB ~10-11 N[7,9,56] and a critical radius of ~1 μm. The observation of uniform LC ordering of 5CB droplets in F9 below the critical radius of ~1 μm contrasts with prior reports of 5CB droplets confined within polymeric capsules, where a transition from bipolar (parallel anchoring) to radial configuration was seen below a critical radius of ~400 nm[6,7]. The transition was attributed to the contribution that K24 (saddle-splay elastic constant) makes to the elastic free energy. For the F9-5CB emulsions reported in this manuscript, as noted above, a small amount of F9 dissolves into 5CB. It is possible that the dissolved F9 changes the elastic constants of the 5CB (e.g., to increase the splay and the bend constants more than the twist and the saddle-splay constants), leading to uniform ordering of LC within droplets with radii less than 1 μm. Additionally, we examined the temperature-dependence of LC ordering within the LC droplets but observed no significant difference in director profiles between 25 oC and 34 oC (prior to the nematic to isotropic phase transition of 5CB). We also observed a perpendicular alignment of 5CB at the F9 interface when droplets of F9 were dispersed in a continuous phase of nematic 5CB (Figure 4.2D). In these experiments, we emulsified 2% v/v F9 into 5CB and confined films of the mixture between two polymer-coated glass surfaces that induced unidirectional tangential alignment (no) of nematic 5CB. The homeotropic anchoring of 5CB at the F9 droplet interface generates a hyperbolic hedgehog defect[35,57] accompanying the F9 droplets (Figure 4.2D). A twist deformation can be seen near the point defect in Figure 4.2D. Both clockwise and anticlockwise twisted structures were observed with equal probability. It is possible that the dissolution of F9 into 5CB also plays a role in the observation of the twist in the 5CB (potentially reducing the ratio between twist and splay elastic constants)[56,58,59]. As noted above, a mixture F9:5CB = 1:100 (volume ratio) forming a single phase above 50.5 ± 3 oC. Upon cooling to room temperature, 93 however, we observed F9 droplets to phase separate from the nematic 5CB and form dimers and chains (Fig. S2)[60]. The observation of homeotropic anchoring of 5CB at perfluoroalkane interfaces contrasts to observations made at interfaces of other immiscible liquids (e.g., water and glycerol, for which the orientations of nematic 5CB are parallel). At least three mechanisms have been proposed to lead to the homeotropic anchoring of 5CB at interfaces. First, at a free surface (air interface), 5CB molecules form a highly ordered smectic-like layer, revealing that an entropic loss is compensated by enthalpic gains arising from layering of the LC molecules[61]. Second, surfactants (e.g., sodium dodecyl sulfate and dodecyltrimethylammonium bromide) can give rise to homeotropic anchoring of 5CB at aqueous interfaces by interdigitating with the tails of 5CB[7,9,14,16,56]. Third, interfacial disordering can give rise to weak anchoring energies, which allows the LC elastic energy or external fields to dominate the LC orientation at the interface[1,56,62,63]. Because surfactant is absent in the experiments reported above, and because our experiments lead us to conclude that the easy axis of 5CB is perpendicular to the interface to F9, we propose that smectic layering is likely behind the observation of homeotropic ordering of 5CB at F9 interfaces. Below we explore this proposition. To probe the possible role of the smectic layering in leading to perpendicular anchoring of 5CB at the F9 interface, we examined the behavior of the nematogen PCH5 at the F9 interface. The molecular structure of PCH5 differs from 5CB in that one aromatic ring of 5CB is replaced by a cyclohexyl group on PCH5 (Figure 4.1B). PCH5 exhibits planar anchoring (parallel to the interface) at free surfaces[42] because the non- planar configuration of the cyclohexyl group of PCH5 frustrates smectic-like layering. We dispersed PCH5 droplets into F9 films at 35 oC, where PCH5 is a nematic LC (5 oC above the crystal-to-nematic transition temperature). In contrast to 5CB, we observed 94 tilted anchoring of nematic PCH5 at the F9 interface. This is indicated by a “pre-radial” configuration of PCH5, with a director profile that radiates from a point defect located near the interface of the nematic droplet (Figure 4.2E)[37,56]. The preradial configuration comprises a point hedgehog located near the surface of the droplet with a topological charge of N = 1[64,65]; weak, tilted surface anchoring (so-called weak conic anchoring) prevents formation of defects on the surface of the droplet. The configuration in Figure 4.2E is, therefore, topologically equivalent to a radial droplet. The tilted anchoring is consistent with our hypothesis that the presence of the cyclohexyl group frustrates smectic layering of PCH5. Ongoing molecular-level simulations are being performed to provide additional insight into the ordering of PCH5 and 5CB at F9 interfaces. Nematic Shells with Weak Anchoring. Next we explored how nematic LCs and isotropic perfluorocarbons can be used to design complex multi-compartment emulsions. To form multi-compartment emulsions, we emulsified F9 and 5CB (1:2 volume ratio) mixtures in glycerol at room temperature. Glycerol (more viscous than water) was used to slow the rate of diffusion of the emulsion droplets to enable microscopic observations, although similar results were observed with water. Because F9 and 5CB are largely immiscible during emulsification at room temperature, the relative volumes of the two phases varied between droplets, providing a means to rapidly screen for possible configurations formed by the two phases. The morphology of the multi-compartment emulsion droplets explored in this paper are determined by a force balance between three interfacial tensions[37,43,66], namely the interfacial tensions of the F9-glycerol interface, γF9, the 5CB-glycerol interface, γ5CB, and the F9-5CB interface, γF9-5CB. To guide our understanding of the morphologies formed by these complex emulsions, we performed measurements of interfacial tensions using a pendant drop tensiometer. In the absence of surfactants, γF9 95 was measured to be 33.0 ± 0.5 mN/m; γ5CB was 17.2 ± 0.1 mN/m; and γF9-5CB was measured to be 13.7 ± 0.1 mN/m using pendant drop method[9,36,37,43,48]. Because γF9 > γ5CB + γF9-5CB, we predicted the formation of morphologies with F9 cores and 5CB shells[37,43,66]. Consistent with these predictions, we observed the formation of core-shell droplets with 5CB shells and F9 cores. We observed three distinct internal organizations of the nematic 5CB shells. First, when the radii of the inner F9 cores (r) were larger than 40% of the droplet radii (R), r ≥ 0.4R, nematic shells with quadrupolar symmetry were observed (Figure 4.3A). Interestingly, neither Saturn-ring defects[19,67] (-1/2 disclination rings at the equator) surrounding F9 cores nor boojums[7] (+1 point defects at interface) at 5CB-glycerol interface were observed for the quadrupolar configuration. We hypothesized that the relatively small energetic contribution of surface anchoring compared to elastic strain of the LC (weak anchoring) led to the absence of defects within the 5CB shells. To obtain anchoring energies (~Wl2) that are comparable in magnitude to the elastic energy (~Kl), we estimate the anchoring strength (W) at the both F9 and glycerol interfaces to be of order of 10-5-10-6 N/m, where l , the length scale, is ~5 μm and K is ~10-11 N. This estimate of anchoring energy is supported by numerical simulations shown below. Second, for F9 cores with radius 0.4R ≥ r ≥ 0.3R, the 5CB shells exhibited a dipolar symmetry (Figure 4.3B). The isotropic F9 inner droplets were accompanied by hyperbolic hedgehogs (-1 point defects). Similar to the quadrupolar symmetry, no boojums were observed at the 5CB-glycerol interface with the dipolar symmetry, indicating a higher anchoring strength at the F9 interface compared to the glycerol interface. Based on the anchoring coefficient estimated above, we inferred at the F9 interface that Winner = 10-5 N/m and at glycerol interface that W = 10-6outer N/m. Our results contrast to nematic 5CB shells with strong anchoring at aqueous interfaces in the 96 Figure 4.3 (A-C) Polarized-light micrographs (left: parallel-polars; middle: crossed-polars) and schematic illustrations (right) showing internal organizations of double emulsions with nematic 5CB shells and F9 cores of various sizes. Droplets are dispersed in glycerol. Scale bars, 10 μm. The dash lines show the ordering of 5CB (director). The red dots indicate the defects. (D) Estimated elastic free energy Fe as a function of the ratio of core radius to droplet radius 10 μm) for droplets with dipolar and quadrupolar symmetries. (E) Schematic illustration shows the internal organization with strong anchoring at both interfaces (perpendicular inner, planar outer). presence of surfactants, where the interfacial energy (anchoring energy) dominates the 97 LC elastic energy, and the shells exhibit a dipolar symmetry with two +1 defects (boojums) at the planar interface (Figure 4.3E)[14,16]. Additionally, we note that the F9-5CB emulsions with core-shell morphologies differ from the emulsions formed by the partially miscible perfluorobenzene and LCs[37], where the mixture only creates emulsion droplets with Janus morphologies. Third, we observed the F9 inner droplet, when r < 0.3R, to be localized to the site of one of the defects of a 5CB droplet in a so-called bipolar configuration[6,14,56] (Figure 4.3C). This configuration has been seen before in emulsions with 5CB shells and homeotropic silica particles[68]. Defect regions possess a high local free energy density and thus they tend to attract inclusions. Similarly, we interpret our results to indicate that the small F9 droplet replaces the strained LC at and near the defect[21], consistent with past numerical simulation showing that the free energy is minimized when an isotropic sphere (r = 0.286R, where r and R are the inclusion and LC droplet sizes, respectively) with homeotropic anchoring is located at boojums[68]. To provide insight into the interplay of mechanical confinement and configurational stability that underlies the above-described experiments, we compared the free energies of the LC configurations with quadrupolar and dipolar symmetries[7,14,16,21]. We formulated a simple model to evaluate the free energy of a multi-compartment droplet. (see Figure S4.3 in Supplemental Material for details.) The elastic free energy Fe, evaluated, as a function of the ratio r/R is shown in Figure 4.3D. Our simple model predicts that the configuration with dipolar symmetry is stable for small inner core sizes, that is r/R < 0.406; and quadrupolar symmetry is stable for large cores, r/R > 0.406. These predictions agree closely with our experimental observation of a transition between configurations with dipolar and quadrupolar symmetries at r/R ~ 0.4, and they are validated further by numerical simulations that are presented below. Numerical Simulations and Estimates of Anchoring Strengths at Nematic- 98 Perfluorcarbon Oil Interfaces. The scaling arguments presented above provide order of magnitude estimates of the anchoring energies of 5CB at F9 interfaces. To obtain quantitative estimates, we performed numerical simulations to precisely model the detailed orientational ordering of LCs in the confines of the morphologies of the multiphase LC droplets. We used a Landau-de Gennes (LdG) continuum model, calculated optical textures and compared them with experiments[37,51]. The one elastic constant K = 4.55 pN for 5CB was used. We note that the radii of the simulated droplets were of the order of 1 μm, one order smaller than most droplets observed in experiment. To enable comparisons between computation and experiments, we increased the anchoring strength by one order of magnitude to keep the ratio of anchoring energy (~Wl2) to elastic energy (~Kl) comparable in experiment and simulation. Figures 4.4 and 4.5 reveal that the simulations predicted configurations with quadrupolar and dipolar symmetry by setting r = 1/2R and r = 1/3R, respectively, consistent with our experimental results and thermodynamic scaling arguments (Figure 4.3). To aid interpretation of the optical micrographs in Figure 4.3 in terms of the influence of anchoring strength of LC at the boundaries of the shells, we varied the anchoring coefficient W at the two interfaces, simulated the corresponding optical micrographs, and then compared the simulated light micrographs to our experimental observations. First, we discuss the simulations for the emulsions with quadrupolar symmetry. The use of strong anchoring (Wouter = 10-3 N/m and 10-4 N/m) at the outer interface led to two boojums (Figures 4.4A,B), contrary to our experimental results showing the absence of defects at the outer interface (Figure 4.3A). When W -5outer = 10 N/m, defects were absent at the outer interface, in good agreement with experiments (Figures 4.4C-E). Similarly, the use of W -3inner = 10 N/m at the inner interface lead to formation of Saturn-ring defects surrounding the inner isotropic cores (Figures 4.4A- C), in contradiction to experimental results. We found that Winner = 10-4 N/m for the F9 99 Figure 4.4 Effect of strength of LC anchoring on the organization of core-shell emulsions with F9 core and 5CB shells (quadrupolar symmetry). Simulated director profiles (left) and polarized images from side view (middle) and top-down view (right) of nematic shells with anchoring coefficients W (N/m) = (A) 10-3 (inner) and 10-3 (outer), (B) 10-3 (inner) and 10-4 (outer), (C) 10-3 (inner) and 10-5 (outer), (D) 10-4 (inner) and 10-5 (outer), and (E) 10-5 (inner) and 10-5 (outer). Droplet radii in simulations: 1 μm (outer), 0.5 μm (inner). Blue loops indicate defects. interface (inner) in Figure 4.4D leads to a defect-free configuration, and the simulated optical micrographs shows four extinction (dark) brushes between crossed polars, in 100 agreement with experiments. When W = 10-5inner N/m (Figure 4.4E), the simulated optical textures are also similar to the experiment. This anchoring strength, however, was found not sufficient to generate dipolar symmetry, as described below. As noted above, the anchoring strength was increased by one order of magnitude in simulations to capture the difference of droplet size between experiments and simulations. The Figure 4.5 Anchoring strength effect on the organizations of a double emulsion with dipolar symmetry. Simulated director profiles (left) and polarized light images (right) of nematic shells with anchoring coefficients W (N/m) = (A) 10-3 (inner) and 10-5 (outer), (B) 10-4 (inner) and 10-5 (outer), (C) 10-5 (inner) and 10-5 (outer). Droplet radii in simulations: 2.25 μm (outer), 0.75 μm (inner). anchoring coefficients of 5CB used in the simulations, Wouter = 10-5 N/m at outer phase and Winner =10-4 N/m at inner interface, are thus equivalent to the anchoring coefficients estimated from our experiments and scaling arguments (W -6outer = 10 N/m at outer glycerol phase and Winner = 10-5 N/m at inner F9 core). The director profiles and polarized images of the emulsion droplets with dipolar 101 symmetry were also simulated (Figure 4.5). We found simulations with Wouter = 10-5 N/m and W -3 -4inner = 10 -10 N/m, as shown in Figures 4.5A,B, to capture the key features of experimental results, specifically the formation of hyperbolic hedgehog defects (Figure 4.3B). In contrast, the use of weak anchoring coefficients (W =10-5 N/m) for both inner and outer interfaces led to the dipolar configuration being unstable (Figure 4.5C). The simulations also revealed that the anchoring strength of nematic 5CB at F9 interface is one order of magnitude larger than that at glycerol interface, consistent with our scaling arguments reported above. Influence of Surfactant Type and Concentration on Morphologies and Internal Configuration of F9-Nematic Emulsion Droplets. As mentioned in the Introduction, the use of perfluorocarbon oils provides the opportunity to use both perfluorocarbon and hydrocarbon surfactants to tune droplet morphologies and internal organizations via preferential adsorption at interfaces of the two domains of the droplets[37,43,44]. To explore this opportunity, similar to the fabrication of nematic shells, F9 and 5CB (1:1-2:1 volume ratio) mixtures were dispersed into aqueous surfactant solutions at room temperature to characterize droplet morphologies. To guide our understanding of the influence of surfactants on droplet morphology, we measured the interfacial tensions between aqueous phases and oils as a function of the concentration of sodium dodecyl sulfate (SDS), and perfluorooctanoic acid (PFOA) (Figure 4.6A,B). We assume that the interfacial tension at the F9-5CB interface of the emulsion droplets, γF9-5CB = 13.7 ± 0.1 mN/m, remains unchanged in the presence of the surfactants. In contrast to the surfactant-free emulsions formed by either 5CB and F9, we predicted that in the presence of surfactants (i.e., 2 mM SDS, 1 mM PFOA and 2 mM PFOA), the droplets would all form Janus morphologies, as the three 102 Figure 4.6 Influence of Surfactant Type and Concentration on Morphologies and Internal Configuration of F9-Nematic Emulsion Droplets. (A,B) Interfacial tensions as a function of concentrations of the surfactants. The insets are the molecular structures of (A) the hydrocarbon surfactant, sodium dodecyl sulfate (SDS), and (B) the fluorocarbon surfactant, perfluorooctanoic acid (PFOA). (C-F) Parallel-polar (first column), crossed-polar micrographs (second column), simulated director profiles (third column) and simulated polarized light micrographs (fourth column) show surfactant-dependent morphology and organization of F9-5CB droplet emulsions dispersed in the aqueous phases of (C) 2 mM SDS, (D) 1 mM PFOA and (E,F) 2 mM PFOA, respectively. Scale bars, 10 μm. Adopted boundary conditions in simulations: perpendicular at the 5CB-F9 interface with W 4inner =10- N/m and at the aqueous interfaces: (C) perpendicular with W =10-3outer N/m ; (D) parallel with W -4outer =10 N/m ; (E) perpendicular with W -4outer =10 N/m . Droplet radii in simulations: 1 μm. 103 interfacial tensions satisfy the thermodynamic criteria that γjk – (γij + γik) (i ≠ j ≠ k = F9, 5CB, F9-5CB) are all negative[36,37,43]. This prediction is validated by our experiments (Figures 4.6C-E). We also observed droplets with 5CB cores and F9 shells in 2 mM PFOA solution (Figure 4.6F), but we concluded that this inverted core-shell structure was a metastable morphology because of its small population (less than 5%) and the thermodynamic analyses described above. Although all surfactants induced the Janus morphology, we observed that the internal organizations depended on the surfactant type and concentration. We elucidated the influence of surfactant type and concentration on the internal organization of the 5CB compartments by comparing experiments to simulations. We assume the anchoring of 5CB at the F9 interface (the inner interface) remains homeotropic in the presence of the surfactants; the anchoring coefficient used in the simulation was unchanged, Winner = 10-4 N/m; and, as discussed above, the anchoring coefficients used in the simulations are one order of magnitude larger than the experiments to accommodate the smaller size of droplets in simulations. When we added 2 mM SDS into aqueous phase, 5CB molecules exhibited homeotropic anchoring at the aqueous interface, confirmed by the radial configuration of nematic single-phase droplet. In the Janus droplets, we observed no defect formation in the nematic compartment (Figure 4.6C). This configuration is in a good agreement with simulations, when strong anchoring W -3outer =10 N/m was used at the aqueous interface. Similar configurations have been seen for droplets formed by partially miscible perfluorobenzene-E7 mixtures in 1 mM SDS solution[37]. In contrast to 2 mM SDS, 5CB was measured to assume a planar orientation at an interface to 1 mM PFOA. The nematic 5CB compartments of the Janus droplets prepared with 1mM PFOA were found to exhibit a director profile that radiates from a point defect located at the pole (planar anchoring at the aqueous interface, Figure 4.6D). 104 The simulated textures captured the location of the defect and the brushes. Emulsions comprised of poly(dimethylsiloxane) and 5CB in poly(vinyl alcohol) solution also form similar organizations, as reported previously[9]. When using 2 mM PFOA, 5CB anchors homeotropically, similar to 2 mM SDS. However, the configuration of the 5CB within the Janus droplet formed with 2 mM PFOA comprises a director profile radiating from a point defect located at the center, which contrasts to the 5CB compartment formed using the 2 mM SDS (Figure 4.6E). We hypothesized that the SDS and the PFOA surfactants provide different anchoring strengths at the aqueous interface. Numerical simulations validated the hypothesis. When Wouter =10-4 N/m, the locations of defects and dark brushes were in good agreement with experiments performed with PFOA. We concluded that the PFOA provides an anchoring strength that is one order of magnitude smaller than the SDS surfactant. We note that the configuration of the LC domain within the Janus droplet is a sensitive reporter of changes in surface anchoring energy. We end this section by noting that we observed no visible morphological changes when heating the F9-5CB emulsions above the nematic-isotropic transition temperature of 5CB (35.5 oC), leading to the conclusion that the elasticity and anchoring energy of the LC droplets have a minor impact on droplet morphology. Liquid-to-Vapor Phase Transitions in the Perfluorocarbon Core of Complex LC Emulsions. Perfluoroalkanes and nematic LCs, as described above, provide access to multicompartment emulsions with morphologies and internal organizations that are not accessible using other systems. In addition, immiscible perfluorocarbons are interesting components of emulsions because they can be driven to undergo liquid-vapor phase transitions in emulsion systems.[45–47] In contrast, LC shells with aqueous cores are difficult drive through reversible changes because they are generally unstable[8,13,16,17]. In the experiments reported below, we used 105 perfluoroheptane (F7) and the nematogens HTW to form core-shell droplets because the upper part of the nematic temperature range of HTW (-30 oC-101 oC) is above the boiling point of F7 (boiling point (bp) = 82 oC). In our experiments, F7 and HTW was homogenized into glycerol to form droplets with core-shell morphology. Figure 4.7 shows the droplet behaviors during one heating and cooling cycle. At 50.0 oC, the F7 core was a liquid phase and the HTW shell was a nematic phase (Figure 4.7A). At 90.0 oC, which is above the boiling point of F7 (82 oC), the morphology and internal organizations of the droplet remained unchanged (Figure 4.7B), indicating the core at this temperature was a superheated liquid phase. At 130.0 oC, the nematic shell turned dark, corresponding to a nematic to isotropic phase transition in the shell while the core remained in the metastable liquid state (Figure 4.7C). Remarkably, it was not until the droplet was heated to 165 oC-170.0 oC did the F7 core rapidly vaporize. Within a period of 0.05 s, we observed the 5CB shell to rapidly expand as the F7 core transformed to a gas (dark micrograph under the crossed polars, Figure 4.7D). We confirmed that the increase in the size of the core was consistent with formation of a gas by using the ideal gas law pV = nRT, where p is atmospheric pressure (N/m2), V is volume to be calculated (m3), n is the amount of substance (mol), R is the gas constant 8.31 (J mol-1 K-1), and T is the temperature (K). We calculated that the volume of the core should expand with the liquid-to-vapor 106 transition by a factor of 133 (factor of 5 in diameter). We measured in our experiments Figure 4.7 Thermal reconfiguration of multiphase LC emulsion droplets with a perfluorocarbon (F7) core and nematic (HTW) shell. Parallel-polar (left), crossed-polar (middle) micrographs and schematic illustrations (right) of a droplet with an F7 core and an HTW shell during heating and cooling cycles; heating to (A) 50.0 oC (liquid core, nematic shell), (B) 90.0 oC (liquid core, nematic shell), (C) 130.0 oC (liquid core, isotropic shell), (D) 170.0 oC (vapor core, isotropic shell), and upon cooling to (E) 90.0 oC (vapor core, nematic shell) and (F) 50.0 oC (liquid core, nematic shell). The defect in (E) (red dot) is not within the focal plane. (G) Droplet with defect evident in the nematic HTW shell. Scale bars, 10 μm. 107 that the core diameter expanded from 9.5 μm to 49.3 μm in Figure 4.7, consistent with the above estimate. We note that the F7 core was superheated by ~85 oC prior to undergoing the liquid to vapor phase transition and that the phase transition in the nematic shell did not trigger the liquid to vapor transition in the core. Comparable levels of superheating have been reported in prior experiments performed with perfluorocarbons having shorter chain lengths (i.e., perfluoropropane and perfluorobutane)[46,47]. We subsequently cooled the HTW-F7 gas-core emulsion back to 90.0 oC, which is above the bp of F7 but below the clearing temperature of HTW, and we observed the shell to recover its liquid crystalline state as evidenced by the appearance of a bright shell when viewed under crossed polars (Figure 4.7E). We also observed defects to be present within the thin nematic shell, as shown in Figures 4.7G and S4.4 (denoted by the blue arrow). We note that the defect is not located at the focal plane for the shell shown in Figure 4.7E. Subsequently, we cooled the system to 50.0 oC. After 30 min, the F7 core was observed to condense into its liquid state and the emulsion droplet returned to its original size (Figure 4.7F). Interestingly, the time scale for expansion (~0.01 s) versus contraction (~10 min) of the gaseous core differed greatly. We hypothesize that the nematic ordering of HTW within the shell slows the contraction process as compared to the expansion during which the HTW is an isotropic phase. Although we observed one defect near the F7 core in both thick and thin shells, thinning of nematic shells can potentially actuate the transform of defects[13,15,17], including elimination of defects if the anchoring extrapolation length becomes larger than the shell thickness. Future studies will explore in more detail the dynamics of these emulsions. Overall, the experiments reported above demonstrate that LC-perfluorocarbon emulsions can be actuated reversibly via heating and cooling of the perfluorocarbon cores through liquid-to-vapor phase transitions. 108 4. Conclusions This paper reports the formation of complex emulsions from immiscible LCs and isotropic perfluoroalkane oils (F9 and F7). To provide insight into the structures of the emulsion droplets, we characterized the anchoring of two nematic LCs, 5CB and PCH5, at interfaces of isotropic F9 phases. We found that 5CB assumes a perpendicular orientation while PCH5 exhibits a tilted anchoring, consistent with the role of smectic ordering in the perpendicular orientation assumed by 5CB. In the absence of surfactant added to a continuous aqueous/glycerol phase, we found that F9 and 5CB form emulsion droplets with perfluorocarbon cores and nematic shells. Our observations reveal that the anchoring of 5CB at the F9 interface is unusually weak, giving rise to shells of 5CB with either quadrupolar (r/R > 0.4) or dipolar symmetry (r/R < 0.4), respectively, a conclusion that is supported by numerical simulations based on a Landau–de Gennes free energy. By dispersing the F9-5CB emulsions into glycerol or aqueous solutions of the surfactants SDS or PFOA, droplets with Janus morphologies were formed, with the internal organization of the nematic domains dependent on the specific surfactant type. Interestingly, under condition for which SDS and PFOA generate the same internal configurations of single phase LC droplets, we observed the LC domains of the multiphase droplets to give rise to distinguishable internal organizations. This result reveals that confinement of LCs within multiphase droplets provides new opportunities to differentiate interfacial phenomena that are not distinguishable in single phase LC droplets. We also showed that LC-perfluorocarbon emulsions provide new opportunity to create thermally actuated and dynamically reconfigurable nematic shells by vaporizing the perfluorocarbon core. More broadly, the results reported in this manuscript demonstrate that perfluoroalkane-nematic two-phase systems provide new opportunities to design stimuli-responsive emulsions The results reported in this paper also generate a range of unresolved questions. 109 For example, our results suggest that F9 dissolved in 5CB, although low in concentration, may change the elastic constants of 5CB to impact the configurations assumed by 5CB under confinement. Additionally, although our results support the proposal that 5CB assumes smectic ordering at the F9 interface, additional studies are needed to fully understand the molecular-level origins of the LC anchoring observed at liquid perfluorocarbon interfaces. Finally, our studies reveal that the expansion and contraction of the F7 cores within HTW shells differ greatly in dynamics. We do not yet fully understand the factors that control these dynamics. The results presented in this paper also revealed that the liquid F7 cores of the nematic emulsions were superheated by ~80oC prior to transforming to a vapor phase. We envisage that future studies may exploit this superheated state of the perfluorocarbon core in a range of potentially useful way, such as in microsensors, e.g., by using the selective dissolution of gases in LCs[5,29]. We also anticipate that the approach demonstrated in this paper can be extended in future studies to other achiral and chiral LC phases, including smectic, cholesteric and blue phases, as well as other perfluorocarbons. Additionally, the reconfiguration of the LC compartment of the emulsion via vaporization of the perfluorocarbon, as demonstrated in our paper, is not limited to core-shell morphologies, but can also be explored in symmetry-broken morphologies, such as Janus droplets. We predict that the interplay of phase transitions between LCs and perfluorocarbon can also be manipulated to form a range of new non- equilibrium LC organizations that reflect the history of heating and cooling of the emulsion systems. 5. Supporting Information 110 Figure S4.1 Time-lapse micrographs of a 5CB droplet (radius 0.9 μm) dispersed in F9. The brightness and contrast are increased to enable visualization. Scale bar, 10 μm. 111 Figure S4.2 F9 droplets suspended in a continuous nematic 5CB phase forming (A) dimers (denoted by blue arrows) with F9:5CB = 1:100 (volume ratio) and (B) chains of droplet with F9:5CB = 10:100. The observations were made at room temperature. Scale bars: 10 μm. 112 A simple thermodynamic model to analyze the influence of the relative volumes of the F9 and 5CB phases on structure of the core-shell emulsions. To provide insight into the interplay of energetics that underlies the experiments described in the main text, we compared the free energies of the LC configurations with quadrupolar and dipolar symmetries[7,14,16]. In particular, we evaluated the free energy F of a multi-compartment droplet as where 𝐹s is the anchoring energy and 𝐹e is the elastic free energy. We evaluated 𝐹s for the nematic shells as where Winner and Wouter are surface anchoring energy per unit area of the inner F9-5CB interface and the outer 5CB-aqueous interface, respectively, and 𝑟 and 𝑅 are radii of the inner F9 core and the outer 5CB shell, respectively. By using following parameters, W -5 -6 inner = 10 N/m, Wouter = 10 N/m, r ~ 5 m, and R ~ 10 m, we determined 𝐹~10−16s − 10 −15 J. We evaluated the elastic free energy 𝐹e for dipolar and quadrupolar symmetries, respectively, as[19,35]: 113 in which 𝐾 is the average elastic constant (one constant approximation), R is the radius of the nematic 5CB shell, r is the radius of the inner F9 core, 𝑉d and 𝑉q are the volumes of the strained LC generated by the inner F9 core within the 5CB shells for dipolar and quadrupolar symmetries, respectively, 𝑟−1 is the radius of the defect core for the dipolar symmetry, 𝜀c is the energy density of the defect core, and 𝜎o is the interfacial energy between the core of the defect and the surrounding nematic bulk LCs. The first terms on the right side of equations (3) and (4) describe the free energies of bipolar configurations of LC droplets, which is estimated for the volume of the nematic shell. For a typical elastic constant using the one constant approximation7 K ~ 10-11 N, the term is of order 10-16-10-15 N, which is comparable to 𝐹s, and thus consistent with weak anchoring described above. The second terms on the right side of equations (3) and (4) capture the additional free energy penalty caused by the inner F9 droplet. We evaluated the free energy penalty due to the F9 core for the shell with quadrupolar symmetry and for the distortion around the point defect (hyperbolic hedgehog) for the dipolar case (Figure S4.3). As shown in Figure S4.3, we estimated the volume of the strained region 𝑉d to occupy a cylindrical volume spanning the thickness of the nematic shell, where 𝜉d = 𝑅 − 𝑟 ; or 𝑉d was estimated as a toroidal volume with diameter 𝜉q. In the absence of confinement (i.e., 𝑅 → ∞, 𝑟 → 0), 𝜉d is the diameter of a sphere, 𝜉q is the diameter of a torus, and 4 𝜉 3 3 𝜋 ( d 𝜉 ) = 2𝜋2 ( q) ⇒ 𝜉q = 0.596𝜉d. The distortion volume for dipolar symmetry 3 2 2 𝑉d was evaluated by the sum of three volumes as shown in Figure S4.3: 𝑉1 = 1 4 𝜋𝑟3(cylinder) − × 𝜋𝑟3(halfsphere) , the height of 𝑉3 ℎ = 𝑅 − √𝑅 2 − 𝑟2 , 𝑉 2 3 2 = 1 𝜋𝑟2(𝜉d − ℎ) (cylinder), 𝑉3 = 𝜋ℎ(3𝑟 2 + ℎ2) (sphere cap). 6 The third term on the right side of equation (3) represent the excess free energy of the defect core. The fourth term of (3) is an energy associated with the interface between the defect core and the surrounding LC. 114 Overall, we evaluated 𝐹e as a function of the ratio r/R using the following parameters[35]: K ~ 10-11 N, 𝑅 = 10 μm, 𝑟−1 = 16.8 nm, 𝜀 = 3 × 10 4 J/m3c , 𝜎o = 1 1 10−5 J/m2 , 𝜉q = 0.596𝜉d , 𝑉d = 𝜋𝑟 3 + 𝜋𝑟2(𝜉d − ℎ) + 𝜋ℎ(3𝑟 2 + ℎ2) , ℎ = 𝑅 − 3 6 𝜉 2 𝜉 √𝑅2 − 𝑟2 , 𝑉q = 2𝜋 2 ( q) (𝑟 + q) . Our analytical results revealed that the 2 2 configuration with dipolar symmetry is more stable for small inner core sizes, that is, for r/R < 0.406; and the configuration with quadrupolar symmetry is more stable for r/R > 0.406. This result agrees closely with our experimental observation of a transition between configurations with dipolar and quadrupolar symmetries at r/R ~ 0.4. We also note that it is consistent with numerical simulations reported in the main text. We comment that the outer terms of equations (3) and (4) dominate 𝐹e, with the remaining terms contributing ~10% of the total free energy. Figure S4.3 Schematic illustrations showing the volumes (blue) of LC that are strained by the presence of the inner F9 core, for configurations with (A) dipolar and (B) quadrupolar symmetries, respectively. 115 Figure S4.4 Parallel-polar, crossed-polar micrographs and schematic illustrations showing the defects (denoted by the blue arrows and the red dots) in emulsions with nematic HTW shells and (A) F9 liquid cores or (B) F9 vapor cores, respectively. Scale bars, 10 μm. Acknowledgement The authors acknowledge support from the Department of Energy, Basic Energy Sciences, Division of Materials Research, Biomaterials Program under Grant No. DE- SC0004025 and DE-SC0019762. YY acknowledges partial fellowship support from the Department of Chemical and Biological Engineering at University of Wisconsin– Madison. 6. References *This chapter was prepared as a Research Article reporting original research in the journal Langmuir. My contribution to this project was in designing and performing the experiments, designing the simulations and writing the manuscript. My co-authors 116 contributed to the designing of the project, experiments, designing and performing the simulations and writing of the manuscript. Adapted with permission from: X. Wang, Y. Zhou, V. Palacio-Betancur, Y.-K. Kim, L. Delalande, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Reconfigurable Multicompartment Emulsion Drops Formed by Nematic Liquid Crystals and Immiscible Perfluorocarbon Oils., Langmuir 35, 16312 (2019). Copyright 2019 American Chemical Society. [1] P. G. de Gennes and J. Prost, The Physics of Liquid Crystals (Clarendon Press, 1993). [2] M. Kleman and O. D. Lavrentovich, Soft Matter Physics : An Introduction (Springer, New York, NY, USA, 2003). [3] G. J. Sprokel and R. M. Gibson, Liquid Crystal Alignment Produced by RF Plasma Deposited Films, J. Electrochem. Soc. 124, 557 (1977). [4] G. J. Sprokel, Molecular Order Induced by Cell Walls: Part I, Experimental Results, Mol. Cryst. Liq. Cryst. 42, 233 (1977). [5] R. R. Shah and N. L. Abbott, Principles for Measurement of Chemical Exposure Based on Recognition-Driven Anchoring Transitions in Liquid Crystals, Science 293, 1296 (2001). [6] J. K. Gupta, S. Sivakumar, F. Caruso, and N. L. Abbott, Size-Dependent Ordering of Liquid Crystals Observed in Polymeric Capsules with Micrometer and Smaller Diameters, Angew. Chemie, Int. Ed. 48, 1652 (2009). [7] D. S. Miller, X. Wang, and N. L. Abbott, Design of Functional Materials Based on Liquid Crystalline Droplets, Chem. Mater. 26, 496 (2014). [8] M. Urbanski, C. G. Reyes, J. Noh, A. Sharma, Y. Geng, V. S. R. Jampani, and J. P. F. Lagerwall, Liquid Crystals in Micron-Scale Droplets, Shells and Fibers, J. Phys. 117 Condens. Matter 29, 133003 (2017). [9] J. Jeong, A. Gross, W.-S. Wei, F. Tu, D. Lee, P. J. Collings, and A. G. Yodh, Liquid Crystal Janus Emulsion Droplets: Preparation, Tumbling, and Swimming, Soft Matter 11, 6747 (2015). [10] T. Suzuki, Y. Li, A. Gevorkian, and E. Kumacheva, Compound Droplets Derived from a Cholesteric Suspension of Cellulose Nanocrystals, Soft Matter 14, 9713 (2018). [11] H. Peng, W. Jiang, Q. Liu, G. Chen, M. Ni, F. Liang, Y. Liao, X. Xie, and I. I. Smalyukh, Liquid Crystals under Confinement in Submicrometer Capsules, Langmuir 34, 10955 (2018). [12] X. Chen, B. D. Hamlington, and A. Q. Shen, Isotropic-to-Nematic Phase Transition in a Liquid-Crystal Droplet, Langmuir 24, 541 (2008). [13] A. Fernández-Nieves, V. Vitelli, A. S. Utada, D. R. Link, M. Márquez, D. R. Nelson, and D. A. Weitz, Novel Defect Structures in Nematic Liquid Crystal Shells, Phys. Rev. Lett. 99, 157801 (2007). [14] T. Lopez-Leon and A. Fernandez-Nieves, Topological Transformations in Bipolar Shells of Nematic Liquid Crystals, Phys. Rev. E 79, 021707 (2009). [15] T. Lopez-Leon, V. Koning, K. B. S. Devaiah, V. Vitelli, and A. Fernandez- Nieves, Frustrated Nematic Order in Spherical Geometries, Nat. Phys. 7, 391 (2011). [16] H.-L. Liang, R. Zentel, P. Rudquist, and J. Lagerwall, Towards Tunable Defect Arrangements in Smectic Liquid Crystal Shells Utilizing the Nematic-Smectic Transition in Hybrid-Aligned Geometries, Soft Matter 8, 5443 (2012). [17] D. Seč, T. Lopez-Leon, M. Nobili, C. Blanc, A. Fernandez-Nieves, M. Ravnik, and S. Žumer, Defect Trajectories in Nematic Shells: Role of Elastic Anisotropy and Thickness Heterogeneity, Phys. Rev. E 86, 020705 (2012). [18] A. Darmon, M. Benzaquen, D. Seč, S. Čopar, O. Dauchot, and T. Lopez-Leon, 118 Waltzing Route toward Double-Helix Formation in Cholesteric Shells, Proc. Natl. Acad. Sci. U.S.A. 113, 9469 (2016). [19] H. Stark, Saturn-Ring Defects around Microspheres Suspended in Nematic Liquid Crystals: An Analogy between Confined Geometries and Magnetic Fields, Phys. Rev. E 66, 032701 (2002). [20] I. Muševič, M. Škarabot, U. Tkalec, M. Ravnik, and S. Žumer, Two- Dimensional Nematic Colloidal Crystals Self-Assembled by Topological Defects, Science 313, 954 (2006). [21] X. Wang, D. S. Miller, E. Bukusoglu, J. J. de Pablo, and N. L. Abbott, Topological Defects in Liquid Crystals as Templates for Molecular Self-Assembly, Nat. Mater. 15, 106 (2016). [22] H. Kikuchi, M. Yokota, Y. Hisakado, H. Yang, and T. Kajiyama, Polymer- Stabilized Liquid Crystal Blue Phases, Nat. Mater. 1, 64 (2002). [23] M. A. Gharbi, S. Manet, J. Lhermitte, S. Brown, J. Milette, V. Toader, M. Sutton, and L. Reven, Reversible Nanoparticle Cubic Lattices in Blue Phase Liquid Crystals, ACS Nano 10, 3410 (2016). [24] J. Zou and J. Fang, Director Configuration of Liquid-Crystal Droplets Encapsulated by Polyelectrolytes, Langmuir 26, 7025 (2009). [25] K. Slyusarenko, C. Blanc, Y. Reznikov, and M. Nobili, Quenched Disorder of a Nematic Liquid Crystal under a Magnetic Field, J. Mol. Liq. 267, 100 (2018). [26] M. Tasinkevych, M. G. Campbell, and I. I. Smalyukh, Splitting, Linking, Knotting, and Solitonic Escape of Topological Defects in Nematic Drops with Handles, Proc. Natl. Acad. Sci. U. S. A. 111, 16268 (2014). [27] M. G. Campbell, M. Tasinkevych, and I. I. Smalyukh, Topological Polymer Dispersed Liquid Crystals with Bulk Nematic Defect Lines Pinned to Handlebody Surfaces, Phys. Rev. Lett. 112, 197801 (2014). 119 [28] I.-H. Lin, D. S. Miller, P. J. Bertics, C. J. Murphy, J. J. de Pablo, and N. L. Abbott, Endotoxin-Induced Structural Transformations in Liquid Crystalline Droplets, Science 332, 1297 (2011). [29] T. Szilvási, N. Bao, K. Nayani, H. Yu, P. Rai, R. J. Twieg, M. Mavrikakis, and N. L. Abbott, Redox-Triggered Orientational Responses of Liquid Crystals to Chlorine Gas, Angew. Chemie 130, 9813 (2018). [30] Y.-K. Kim, X. Wang, P. Mondkar, E. Bukusoglu, and N. L. Abbott, Self- Reporting and Self-Regulating Liquid Crystals, Nature 557, 539 (2018). [31] C. Krüger, G. Klös, C. Bahr, and C. C. Maass, Curling Liquid Crystal Microswimmers: A Cascade of Spontaneous Symmetry Breaking, Phys. Rev. Lett. 117, 048003 (2016). [32] C. Jin, C. Krüger, and C. C. Maass, Chemotaxis and Autochemotaxis of Self- Propelling Droplet Swimmers, Proc. Natl. Acad. Sci. U.S.A 114, 5089 (2017). [33] T. Yamamoto and M. Sano, Chirality-Induced Helical Self-Propulsion of Cholesteric Liquid Crystal Droplets, Soft Matter 13, 3328 (2017). [34] M. Suga, S. Suda, M. Ichikawa, and Y. Kimura, Self-Propelled Motion Switching in Nematic Liquid Crystal Droplets in Aqueous Surfactant Solutions, Phys. Rev. E 97, 062703 (2018). [35] X. Wang, Y.-K. Kim, E. Bukusoglu, B. Zhang, D. S. Miller, and N. L. Abbott, Experimental Insights into the Nanostructure of the Cores of Topological Defects in Liquid Crystals, Phys. Rev. Lett. 116, 147801 (2016). [36] T. Nisisako, Recent Advances in Microfluidic Production of Janus Droplets and Particles, Curr. Opin. Colloid Interface Sci. 25, 1 (2016). [37] X. Wang, Y. Zhou, Y.-K. Kim, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Thermally Reconfigurable Janus Droplets with Nematic Liquid Crystalline and Isotropic Perfluorocarbon Oil Compartments, Soft Matter 15, 2580 (2019). 120 [38] M. Zhang, P. T. Corona, N. Ruocco, D. Alvarez, P. Malo De Molina, S. Mitragotri, and M. E. Helgeson, Controlling Complex Nanoemulsion Morphology Using Asymmetric Cosurfactants for the Preparation of Polymer Nanocapsules, Langmuir 34, 978 (2017). [39] E. M. Nomena, C. Remijn, F. Rogier, M. van der Vaart, P. Voudouris, and K. P. Velikov, Unravelling the Mechanism of Stabilization and Microstructure of Oil-in- Water Emulsions by Native Cellulose Microfibrils in Primary Plant Cells Dispersions, ACS Appl. Bio Mater. 1, 1440 (2018). [40] P. Hubert, H. Dreyfus, D. Guillon, and Y. Galerne, Anchoring Orientation of Nematic and Smectic A Liquid Crystals on PTFE Treated Plates, J. Phys. II 5, 1371 (1995). [41] E. Campanelli, S. Faetti, and M. Nobili, Azimuthal Anchoring Energy at the Interface between a Nematic Liquid Crystal and a PTFE Substrate, Eur. Phys. J. E 11, 199 (2003). [42] K. Nayani, P. Rai, N. Bao, H. Yu, M. Mavrikakis, R. J. Twieg, and N. L. Abbott, Liquid Crystals with Interfacial Ordering That Enhances Responsiveness to Chemical Targets, Adv. Mater. 30, 1706707 (2018). [43] L. D. Zarzar, V. Sresht, E. M. Sletten, J. A. Kalow, D. Blankschtein, and T. M. Swager, Dynamically Reconfigurable Complex Emulsions via Tunable Interfacial Tensions, Nature 518, 520 (2015). [44] A. E. Goodling, S. Nagelberg, B. Kaehr, C. H. Meredith, S. I. Cheon, A. P. Saunders, M. Kolle, and L. D. Zarzar, Colouration by Total Internal Reflection and Interference at Microscale Concave Interfaces, Nature 566, 523 (2019). [45] O. D. Kripfgans, J. B. Fowlkes, D. L. Miller, O. P. Eldevik, and P. L. Carson, Acoustic Droplet Vaporization for Therapeutic and Diagnostic Applications, Ultrasound Med. Biol. 26, 1177 (2000). 121 [46] M. L. Fabiilli, K. J. Haworth, N. H. Fakhri, O. D. Kripfgans, P. L. Carson, and J. B. Fowlkes, The Role of Inertial Cavitation in Acoustic Droplet Vaporization, IEEE Trans. Ultrason. Ferroelectr. Freq. Control 56, 1006 (2009). [47] P. A. Mountford, A. N. Thomas, and M. A. Borden, Thermal Activation of Superheated Lipid-Coated Perfluorocarbon Drops, Langmuir 31, 4627 (2015). [48] J.-W. Kim, H. Kim, M. Lee, and J. J. Magda, Interfacial Tension of a Nematic Liquid Crystal/Water Interface with Homeotropic Surface Alignment, Langmuir 20, 8110 (2004). [49] M.-L. Ge, J.-L. Ma, and B. Chu, Densities and Viscosities of Propane-1,2,3- Triol + Ethane-1,2-Diol at T = (298.15 to 338.15) K, J. Chem. Eng. Data 55, 2649 (2010). [50] A. M. A. Dias, C. M. B. Gonçalves, A. I. Caço, L. M. N. B. F. Santos, M. M. Piñeiro, L. F. Vega, J. A. P. Coutinho, and I. M. Marrucho, Densities and Vapor Pressures of Highly Fluorinated Compounds, J. Chem. Eng. Data 50, 1328 (2005). [51] Y. Zhou, B. Senyuk, R. Zhang, I. I. Smalyukh, and J. J. de Pablo, Degenerate Conic Anchoring and Colloidal Elastic Dipole-Hexadecapole Transformations, Nat. Commun. 10, 1000 (2019). [52] J. B. Fournier and P. Galatola, Modeling Planar Degenerate Wetting and Anchoring in Nematic Liquid Crystals, Europhys. Lett. 72, 403 (2005). [53] M. Ravnik and S. Žumer, Landau-de Gennes Modelling of Nematic Liquid Crystal Colloids, Liq. Cryst. 36, 1201 (2009). [54] R. Ondris-Crawford, E. P. Boyko, B. G. Wagner, J. H. Erdmann, S. Žumer, and J. W. Doane, Microscope Textures of Nematic Droplets in Polymer Dispersed Liquid Crystals, J. Appl. Phys. 69, 6380 (1991). [55] J. A. Gladysz, D. P. Curran, and I. T. Horváth, Handbook of Fluorous Chemistry (2004). 122 [56] P. S. Drzaic, Liquid Crystal Dispersions (World Scientific, 1995). [57] H. Stark, Director Field Configurations around a Spherical Particle in a Nematic Liquid Crystal, Eur. Phys. J. B 10, 311 (1999). [58] M. J. Press and A. S. Arrott, Theory and Experiments on Configurations with Cylindrical Symmetry in Liquid-Crystal Droplets, Phys. Rev. Lett. 33, 403 (1974). [59] B. ed. Bahadur, Liquid Crystals: Applications and Uses.Vol. 1 (World scientific, 1990). [60] J. Loudet, P. Barois, and P. Poulin, Colloidal Ordering from Phase Separation in a Liquid- Crystalline Continuous Phase, Nature 407, 611 (2000). [61] M. Sadati et al., Molecular Structure of Canonical Liquid Crystal Interfaces, J. Am. Chem. Soc. 139, 3841 (2017). [62] G. P. Bryan-Brown, E. L. Wood, and I. C. Sage, Weak Surface Anchoring of Liquid Crystals, Nature 399, 338 (1999). [63] X. Nie, R. Lu, H. Xianyu, T. X. Wu, and S.-T. Wu, Anchoring Energy and Cell Gap Effects on Liquid Crystal Response Time, J. Appl. Phys. 101, 103110 (2007). [64] G. E. Volovik and O. D. Lavrentovich, Topological Dynamics of Defects: Boojums in Nematic Drops, Sov. Phys. JETP 58, 1159 (1983). [65] O. D. Lavrentovich, Topological Defects in Dispersed Liquid Crystals, or Words and Worlds around Liquid Crystal Drops, Liq. Cryst. 24, 117 (1998). [66] S. Torza and S. G. Mason, Coalescence of Two Immiscible Liquid Drops, Science 163, 813 (1969). [67] Y. Gu and N. L. Abbott, Observation of Saturn-Ring Defects around Solid Microspheres in Nematic Liquid Crystals, Phys. Rev. Lett. 85, 4719 (2000). [68] M. Sadati, Y. Zhou, D. Melchert, A. Guo, J. A. Martinez-Gonzalez, T. F. Roberts, R. Zhang, and J. J. de Pablo, Spherical Nematic Shells with a Prolate Ellipsoidal Core, Soft Matter 13, 7465 (2017). 123 Chapter 5 Active Motion of Multiphase Oil Droplets: Emergent Dynamics of Squirmers with Evolving Internal Structure 1. Introduction Recent studies of active soft matter[1–5], including self-propelled colloids (particles and droplets)[6–11], active gels[12] and granular systems[13] have unmasked universal physical principles that underlie non-equilibrium states (single species and assemblies) found in both synthetic and living systems (e.g., swarming)[14–17]. In particular, artificial self-propelled microswimmers that convert energy in their environment into states of matter far from equilibrium[6–11,18] have been shown to mimic key dynamical behaviors of motile organisms such as bacteria[14,19], phytoplankton[20], and epithelial cells[21]. For example, as oil droplets dissolve into aqueous micellar solutions[22–24], they propel themselves via creation of interfacial tension gradients (Marangoni stresses) and exhibit behaviors such as chemotaxis[6,9,17]. Whereas past studies of self-propelled droplets have focused on single phase systems with spherical symmetry at rest, including single-phase droplets[6,10,18,25,26] and shells[11], multiphase emulsion droplets with internal symmetry-broken morphologies (e.g., Janus droplets[27–33]) with potential to program circulation internal to the droplets have not been explored. When Janus droplets are dispersed into micellar solutions, they generally depart from being spherical[28,33] or break into two separate droplets[31]. These changes influence the motions of the droplets[33], and thus mask efforts to understand how changes in internal domain structure impact dynamical behaviors. Furthermore, such shape changes prevent use of analytical models of microswimmers, such as the squirmer model[34,35], as they assume spherical shapes. In this paper, we report an experimental 124 approach that uses multi-phase oil droplets prepared from mixtures of hydrocarbon oils with nematic order and isotropic perfluorocarbon oils[30]. In contrast to other multiphase droplet systems, these oils droplets exhibit spherical shapes that are stable over a wide range of surfactant concentrations because the interfacial tension of the inner interface is two orders of magnitude lower than the outer two interfaces[30]. Here, we leverage this finding to explore the dynamics of self-propelled multiphase droplets that evolve in their internal morphology but maintain an overall spherical shape, and demonstrate how the rich dynamics displayed during morphogenesis can be mapped onto different squirmer types. Living systems operate beyond equilibrium[36], and morphogenesis of the structure of living systems is a strategy widely used by Nature to control emergent dynamics[37,38]. At macroscopic scales, the metamorphosis of a caterpillar into a butterfly transforms dynamics from crawling to flying[37]. Alternatively, at the microscopic scale, cellular differentiation of a stem cell into a cardiomyocyte generates a periodic contractile behavior that is responsible for the beating of the heart[38]. In contrast to the diverse examples of the impact of morphogenesis on dynamics encountered in active biological systems, the use of morphogenesis in synthetic soft matter systems as a strategy to program complex dynamics far from equilibrium has not been widely explored[11]. Here we show how multiphase oil droplets can be used as a model system to explore how morphogenesis of oil droplets impacts emergent dynamics driven by active processes. The droplets used in our study undergo phase transitions from a single isotropic oil phase with low viscosity to biphasic Janus droplets with two domains with distinct viscosities and, ultimately, to a viscous nematic single-phase, as a consequence of the preferential solubilization (by a surrounding aqueous micellar solution) of specific components of the multicomponent oils. The evolution of the internal droplet 125 morphology, as a function of time, generates a progression of five distinct droplet dynamical states that correlate strongly with droplet morphology. We find that the complex motions exhibited by the droplets can be described within the physical framework of the squirmer model for spherical microswimmers in Stokes flow[34]. Transitions between squirmer types were observed to accompany changes in droplet morphology, thus illustrating also how our experimental system provides a versatile platform for understanding and engineering microswimmers that display complex swimming patterns[35,39,40]. 2. Experiments Materials The nematogen, E7 (clearing temperature 60 oC), was purchased from HCCH (Jiangsu Hecheng Display Technology Co., LTD). Hexafluorobenzene (99%, perfluorobenzene, FB), tetradecyltrimethylammonium bromide (≥ 99.0%, TTAB) and sodium dodecyl sulfate (≥ 99.0%, SDS) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Fluorescent particles (FluoSpheresTM sulfate, 1 m diameter) were purchased from Life Technologies (Thermo Fisher Scientific). Fisher Finest Premium Grade glass slides was purchased from Fisher Scientific (Pittsburgh, PA). Purification of water (18.2 MΩ cm resistivity at 25 °C) was performed using a Milli-Q water system (Millipore, Bedford, MA, USA). Preparation of samples without macroscopic surfactant concentration gradients Optical cells were assembled from two glass slides separated from each other by ~100 mm-thick spacers. 400 ml of E7 and 32 ml of FB were mixed by vortexing at 3000 rpm for 10 seconds, and 20 ml of the resulting oil mixture was dispersed into 1000 ml of aqueous 100 mM TTAB solution by gentle shaking of the mixture. We subsequently 126 injected the mixture into optical cells and sealed the cells with silicone grease to avoid evaporation of water and unintended generation of surfactant concentration gradients. 100 mM TTAB solution was prepared using FB saturated water. Preparation of samples with macroscopic surfactant concentration gradients We performed a limited number of experiments in the presence of macroscopic surfactant concentration gradients (see Figure 5.5 and associated text). To generate the samples with macroscopic concentration gradients, we left open one end of the optical cell to enable evaporation of water. When using 100 mM sodium dodecyl sulfate (SDS) or TTAB, we observed concentration gradients to form over a distance of ~1 mm from the opening in the optical cell. Observation and Analysis The temperature of each sample was controlled using a Linkam LTS350 hot stage with an accuracy of 0.1 °C. The trajectories of the oil droplets, along with the transitions from isotropic to biphasic to nematic phases, were recorded using a Canon EOS Rebel T6i camera at 30 fps under polarized light or fluorescent microscopy. Other videos were recorded using a Moticam 10+ camera. The motion of droplets ceased after ~30 min with oil droplets still present. The droplet positions were extracted every 0.2 s using ImageJ and analyzed subsequently. Gas chromatograph 50 l FB-E7 mixture was dispersed into 10 ml TTAB (100 mM) solution and equilibrated at 85 oC overnight. The oil droplets sedimented and the top solution was removed. The oil phase was dissolved into hexane and analyzed using a JEOL gas chromatography–mass spectrometry. 127 3. Results and Discussion Morphogenesis and dynamical motion of multiphase droplets We used oil-in-water emulsions formed by mixing E7, a multicomponent nematic liquid crystal (at room temperature) comprised of hydrogenated cyanobiphenyl and terphenyl species, with isotropic hexafluorobenzene (FB), and then mechanically dispersing the mixture as droplets in an aqueous phase of the surfactant tetradecyltrimethylammonium bromide (100 mM TTAB, with a critical micelle concentration (CMC) of 4 mM). We found that a FB-E7 mixture containing a 8:100 volume ratio, respectively, exhibited a single nematic phase below 32 °C; an isotropic phase above 40 °C; and nematic-isotropic phase coexistence between the two temperatures[30]. The experiments reported below were performed at 49 °C, at which temperature the oil mixture initially within the droplets was an isotropic phase (Figure 5.1a). After dispersing the FB-E7 droplets in the aqueous surfactant solution, solubilization of select components of oil from the droplets over time caused the droplets to undergo a series of phase and morphological transitions (Figure 5.1a-e). Specifically, we observed a nematic domain to nucleate within isotropic phase droplets (Figure 5.1b) and then grow (Figure 5.1c and d) to eventually occupy the entire droplet volume (Figure 5.1e). We confirmed that the above-described progression of phase states of droplets was due to preferential extraction of FB from the oil mixture within the droplets (Supplementary Figure S5.1)[22,23]. The process of solubilization of oil from the droplets also generated interfacial tension gradients and interfacial flows that actively propelled the droplets through the surfactant solution[10,17,24]. To track the active motions of the droplets during the above-described changes in internal morphology, we confined the emulsion droplets within ~100 m-thick optical cells filled with surfactant solution, sealed them with silicon grease to prevent evaporation of water, and then heated the system to 49 oC 128 Figure 5.1 Morphogenesis and dynamical motion of droplets. (a-e) Droplet morphologies under parallel polars and (f-h) Typical trajectories showing five states of active droplet motion, in the presence of 100 mM TTAB, observed to accompany distinct droplet morphologies in (f) isotropic (I), (g) biphasic and (h) nematic (N) states. The blue, green and yellow arrows indicate directions of motion. The red arrow points to the defect of the nematic droplet. Scale bars for droplets are 50 m. (isotropic phase droplets). The diameters of the droplets were between 30 and 85 m, and typical speeds 𝑣 ranged from 10 to 40 m/s. By observing both droplet morphology 129 and dynamics, we classified droplet motion into five dynamical states (State I to V, with increasing time). During State I, the oil within the droplets comprised a single isotropic phase (Figure 5.1a). We characterized trajectories over a duration of ~100 s (Figure 5.1f) by evaluating the root mean squared displacement (RMSD) as a function of time interval t [10,41,42]:RMSD = √〈∆𝑥2(𝑡)〉 = √〈[𝒙(𝑡0 + 𝑡) − 𝒙(𝑡 )] 2 0 〉 , where x is the position vector of the droplet, t0 samples all possible times on a trajectory (Figure 5.2a), the average is taken over all displacements occurring over time interval t (Figure 5.2a). The RMSD of the isotropic oil droplets was found to be proportional to time t over the duration of our observations (~100 s), indicating so-called ballistic motion[10]. The rotational diffusion coefficients of the droplets were calculated from the Stokes– Einstein relationship[43] as 𝐷𝑟 = 𝑘B𝑇/8𝜋𝜂𝑟 3, where 𝑘B is the Boltzmann constant, 𝑇 (323 K) is the temperature, 𝜂 (10-3 Pa s) is viscosity of water, 𝑟 (~10 m) is the droplet size; we calculated 𝐷𝑟 to be ~2 × 10 −4 radians2/s, corresponding to a characteristic rotation period of 30,000 s that is much larger than the duration over which droplets were observed. Rotational diffusion, therefore, does not significantly influence droplet motion, consistent with the observation of ballistic trajectories. As the solubilization process continued, we observed nematic domains to nucleate within the isotropic phase droplets (Figure 5.1b). The nematic domain within the isotropic droplet (Janus droplet) did not perturb the ballistic trajectory for a period of ~ 30 s following the nucleation event (State II, Figure 5.1g). In State II, the nematic domain localized at the trailing edge of the droplet with the inner interface (separating the nematic and isotropic domain) oriented perpendicular to the direction of motion. Whereas the equilibrium curvature of the inner interface of the nematic domain is concave[30], we observed flows within the self-propelled droplets to deform the interface to a convex shape (Figure 5.1b). Additional and more pronounced examples 130 of flow-induced changes in internal morphology are described in Figure 5.5. As the volume of the nematic domain increased in size (Figure 5.1c), the ballistic motion of the droplet became unstable and abruptly transformed to that of a spiral (State III, Figure 5.1g and Supplementary Video 1), a dynamical state that persisted for ~ 200 s. Interestingly, accompanying this transition in dynamics, the inner interface rotated to align parallel to the direction of droplet motion (Supplementary Video 2), with the isotropic (nematic) domain consistently located on the outer (inner) side of the spiraling trajectory. The radius of the spiraling trajectory was not constant, but increased continuously over time, from 50 m to 150 m with increase in nematic domain size (State III, Figure 5.1g). Typically, the Janus droplets executed 3-7 revolutions in State III. The dynamical behavior of the system was found to be achiral, with clockwise and counterclockwise spirals observed with equal probability. We also observed the handedness of some droplet motions to switch during State III, with an accompanying reorganization of internal morphology (Supplementary Figure S5.2). The spiraling motion was evident as an oscillatory feature in the calculated RMSD, following the functional form 𝑎 sin(2𝜋𝑇 + 𝑏𝑡) (Figure 5.2a), with scalars a(t), b, time t and a rotation period T of 25 s. The transition from dynamical State III to State IV was denoted by the resumption of ballistic motion (Figure 5.1d and g, Supplementary Video 1), which persisted for ~30 s. In contrast to State II, however, the inner interface during State IV was oriented parallel to the direction of droplet motion (same as State III). State IV ended when the isotropic domain shrank to a topological defect that was then convected by internal flow within the droplet to the leading edge (Supplementary Figure S5.3), as described elsewhere[10,30]. During dynamical State V (single nematic phase), the droplet exhibited a ballistic motion (RMSD ∝ 𝑡, Figure 5.2a). We note that a past study has reported that the type of trajectory (e.g., straight or curly) of single phase nematic 131 Figure 5.2 Characterization of droplet trajectories. (a) RMSD of an isotropic (ballistic motion, blue), biphasic (spiraling motion, green) and nematic droplets (ballistic motion, yellow), respectively. The RMSDs were collected from independent experiments and thus document the behaviors of droplets with distinct sizes. (b) Autocorrelation of velocity of 3 biphasic droplets (56, 53 and 53 m diameter) exhibiting spiraling trajectories at different temperatures with distinct average droplet velocity, respectively. (brown: 47 oC, 18.4 m/s; green: 49 oC, 16.1 m/s; yellow: 50 oC, 11.5 m/s). droplets changes with droplet velocity, which in turn depends on both surfactant concentration and droplet size[10,25]. The relevant dimensionless parameter is the Ericksen number 𝐸𝑟 = 𝑣𝜂𝑅/𝐾, where 𝑣 is the velocity of droplet, 𝜂 is a characteristic viscosity of the LC, 𝑅 is the droplet radius and 𝐾 is a characteristic elastic constant of the nematic phase (one constant approximation). The Ericksen number characterizes the relative importance of viscous and elastic stresses; for small Ericksen numbers, the flow is not able to break the symmetry of the director field (including location of defect) that causes the curly motion. The crossover from ballistic to curly motion is reported in ref 25 to occur for Er of approximately 2. In our system 𝑣 ~ 10 m/s, 𝜂 ~ 10 mPa s, 𝑅 ~ 15 − 50 m, 𝐾 ~ 10 pN, and thus 𝐸𝑟~ 0.15 − 0.5 , consistent with the observation of a ballistic trajectory with a single nematic phase. From this result we also conclude that the origin of the spiraling motion observed during State III is the breaking of symmetry due to the Janus morphology (see Supporting Information)[25]. 132 To provide insight into the above behaviors, we evaluated the velocity autocorrelation function of spiraling State III trajectories as a function of time interval t [41]: 𝐯(𝑡0 + 𝑡) ∙ 𝐯(𝑡0) 〈𝐶(𝑡)〉 = ⟨ ⟩ |𝐯(𝑡0 + 𝑡)| ∙ |𝐯(𝑡0)| , where v is the velocity vector of the droplet and t0 samples all possible time values on a trajectory, the average is taken over all pairs of velocities separated by time interval t. Figure 5.2b shows 〈𝐶(𝑡)〉 calculated from spiraling trajectories measured at three different temperatures (and thus with three different compositions). The oscillating characteristic ∝ cos(2𝜋𝑇) with period T varies from 25 s to 70 s [10] (droplets at the higher temperatures exhibiting the longer periods). The increase in rotational period T arises from a lower droplet velocity (see caption of Figure 5.2b), which in turn occurs because the biphasic state at higher temperature arises at a lower FB concentration[30], and thus is accompanied by a lower rate of solubilization (Figure S5.4). In addition, we 1 calculated average curvatures 𝜅 of the spiraling trajectories as[44]:〈𝜅(𝑡0)〉 = 〈 〉, 𝑅(𝑡0) where R refers to radii of trajectories as a function of time 𝑡0. The average curvature of each trajectory is similar ~0.013 – 0.016 m-1, corresponding to trajectories with diameters of around 150 m. We also performed experiments to determine if temperature-induced changes in phase state and morphology of the droplets would induce transitions in dynamical behaviors similar to those triggered by changes in composition. By increasing temperature by 1 oC (to cause the size of the nematic domain to decrease), we triggered droplet motion to change from spiraling (State III) back to ballistic with an inner interface that was oriented perpendicular to the direction of motion (State II) (Supplementary Video 3). Furthermore, after a droplet had exhibited all five dynamical 133 states, by raising the temperature such that the droplet reentered the single isotropic phase state (Figure S5.5, the oil mixture with a lower FB concentration has a higher clearing temperature[30]), we observed the droplets to progress again through a second set of dynamical states (due to isothermal solubilization). By increasing the temperature each time, a droplet completed a life cycle (from 47 – 53 oC), we observed droplets to progress through dynamical States I to V a total of five times. Overall, these results confirm that the dynamical behaviors exhibited by the droplets are programmed by the internal morphologies of the droplets. Mechanisms of dynamical behaviors of droplets To develop an understanding of the physical mechanisms that underlie the transitions between different droplet dynamical states, we dispersed fluorescent microparticles (1 m diameter) into the aqueous surfactant phase to quantify advection near the surface of the self-propelled droplets[11,45]. We note that anisotropy associated with the liquid crystalline phase is not the origin of the spiraling behavior or the switch between ballistic and spiraling behaviors (nematic single-phase droplets exhibit ballistic motion), as described above. We used the liquid crystalline phase as a viscous oil phase. The influence of the elasticity of the liquid crystal on droplet motion is discussed further below. We first discuss the origin of the spiraling motion (State III) and the subsequent transition back to ballistic motion (State III to State IV). Our measurements with tracer particles revealed that the aqueous phase flow field near a spiraling droplet was asymmetric (Figure 5.3a and Supplementary Video 4). Tracer particles near the I domain (left, red arrow) and N domain (right, yellow arrows) moved from the leading to the trailing face of the droplet with a trajectory that was dominated by fluid advection (not Brownian motion), but the displacement of tracer particles near the I interface of 134 the droplet was fourfold greater than near the N interface (40 m versus 9 m). We formulated a scaling argument to predict the velocity of the interfaces of the I or N domains by assuming that the availability of free energy for viscous dissipation is controlled by the rate of solubilization of the oil by the micellar solution, namely: 𝛼𝜋𝑟2I/N𝑐𝑣I/N𝑓 = 𝐹drag𝑣I/N = 𝜑𝜋𝜂 2 I/N𝑟I/N𝑣I/N (5.1) where the left side describes the rate of generation of free energy by solubilization, with 𝑟I/N (m) representing the size of I or N domain, 𝑐 (mol m -3) is the micelle concentration in the bulk aqueous solution, 𝑣I/N is the domain velocity, 𝑓 (J/mol) is the effective free energy of solubilization and is a scalar quantity. The center and right side of equation (5.1) describe dissipation of energy due to viscous forces, where 𝐹drag is the drag force acting on the droplet, 𝜂I/N is effective viscosity of the I or N domain (𝜂N ≈ 3𝜂I)[10,46] and 𝜑 is a scalar, respectively. Before using this model to describe Janus droplets, we tested the predictions of equation (5.1) using single-phase droplets (𝑟I/N → 𝑟, where 𝑟 𝑐𝑓 𝑐𝑓 is the droplet radius), specifically that 𝑣 ∝ 𝑟 . As is determined by material 𝜂 𝜂 properties, equation (5.1) predicts that 𝑣 is proportional to 𝑟 for a given system. This prediction was tested against previously reported experimental data for oil droplets with radii smaller than 30 m (for which interaction with walls is negligible)[25] and found to be in close agreement (Figure 5.3b). 𝑓 To apply equation (5.1) to spiraling multi-phase droplets, we assumed I to be 𝑓N of order 1 given the similarity in composition of the I phase (~ 4% FB) and N phases (~ 3% FB) at ~50 oC[30]. Although it is possible that 𝑓I is slightly larger than 𝑓N, since the I phase has a higher FB concentration and likely faster mass transfer (lower effective viscosity), our conclusion described below is not expected to change significantly during the motion analyzed (the compositions of the two phases in the biphasic region are constant, only the relative volumes of the two phases change over time). This leads to the prediction that the relative velocity of the N and I phases within the two 135 compartments is given by: 𝑣I 𝑟I 𝜂N ∝ (5.2) 𝑣N 𝑟N 𝜂I If I and N domains are similar in size ( 𝑟I = 𝑟N ), because 𝜂N > 𝜂I (𝜂N ≈ 3𝜂I)[10,46], equation (5.2) predicts that 𝑣I > 𝑣N, thus generating a torque that leads to a spiraling motion with the I domain on the outside of the spiral. Furthermore, equation (5.2) predicts that growth of the N domain will lead to a decrease in 𝑣I/𝑣N, which in turn will result in a continuous increase in the radius of the spiraling trajectory as a function of time, as observed in experiments (Figure 5.1g). When 𝑟I/𝑟N = 0.39 (Figure 5.4b, characterized by the ratio of perimeters of I versus N), the relative sizes of the N Figure 5.3 Dynamics of droplets can be described by the squirmer model. (a) Time-lapse micrographs of tracer particles located near the surfaces of spiraling droplets. The large displacement of tracer particles near the isotropic domain is indicated by the red arrow, and the relatively smaller displacement of tracer particles near the nematic domain is shown by the yellow arrows. Scale bar, 50 m. (b) Velocity of single-phase droplets formed from 4-pentyl-4’- cyanobiphenyl in 7.5% wt TTAB solution, plotted as a function of size (data from [ref. 25]). (c,d) Schematic illustrations and 𝑢𝜃 of tracer particles near surfaces of isotropic compartments of biphasic droplets in (c) State III (neutral-type squirmer) and (d) State II (puller-type). Black and gray arrows show direction of convective flow and direction of motion of droplets, respectively. Red dots denote stagnation points. Blue and green dots are experimental data fit using equation 4 with a phase shift (yellow and red, respectively). 136 and I domains offset the relative viscosities, and thus 𝑣I ≈ 𝑣N ; droplet motion is predicted to be ballistic again. This condition corresponds to the onset of dynamical State IV, thus confirming that the spiraling behavior arises from compartmentalization of flow within the domains of the droplets. This contrasts to previous observations of helical motions of liquid crystal (LC) droplets[10,25,26,47], for which the interplay of nematic elasticity and the convective flow breaks the symmetry of the ballistic motion. Next, we discuss the origin of the transition from ballistic (State II) to spiraling motion (State III). Because the Janus droplets reported here are spherical, we describe the droplet motion using a so-called 3-dimensional (3D) squirmer model[34] for spherical microswimmers in low Reynolds number flow (Stokes flow). Three types of squirmers have been defined, namely pullers, pushers and neutral squirmers, each of which is characterized by a distinct flow field and number of stagnation points on the droplet surface, as described below. In the laboratory frame, the flow fields are described by[34]: 2 𝑢𝑟(𝜃) = 𝐵1cos𝜃 (5.3) 3 1 1 𝑢𝜃(𝜃) = 𝐵1sin𝜃 + 𝐵2 sin(2𝜃) (5.4) 3 2 𝐵2 𝛽 = (5.5) |𝐵1| where 𝑢𝑟 and 𝑢𝜃 are the flow velocities perpendicular and tangential to the droplet interface, 𝜃 is the angle with respect to the motion direction, 𝐵1 and 𝐵2 are constant coefficients. The parameter 𝛽 is used to categorize microswimmers into neutral squirmers (𝛽 = 0), with two stagnation points and a symmetric flow field on front and rear sides, and pullers (𝛽 > 0) and pushers (𝛽 < 0), each generating thrust in the front and rear, respectively, with four stagnation points[35]. 137 Our analysis of tracer particle motion near the spiraling droplets (State III, Figure 5.3c) revealed that the surface flow near the isotropic compartment exhibits two stagnation points, one at the leading edge and one at trailing edge of the droplet. This droplet behavior was observed when the two phases within the droplets were similar in Figure 5.4 Dynamical transitions of droplets. (a) Values of 𝜃N at the transition from ballistic (State II) to spiraling (State III) motion, plotted as a function of droplet radius. (b) Ratio of perimeters of I versus N at the transition from spiraling (State III) to ballistic (State IV), plotted as a function of droplet radius. Blue arrows in (a,b) show surface flow. (c) Trajectory of a Janus droplet with velocity shown by a color map with the transition from ballistic (State II) to spiraling (State III) motion marked by the yellow arrow. Red, black and blue arrows denote self-crossing of the droplet trajectory. The gray arrow shows the direction of motion. (d) Droplet velocity plotted as a function of 𝜃, a descriptor of the N domain size. Arrows correspond to (c). volume and the inner interface between the two phases was aligned towards the direction of propagation of the droplet. The fluid velocity profile near the droplet surface 138 is consistent with a neutral squirmer (Figure 5.3c, left). We fit the tracer velocity to the squirmer model (Figure 5.3c), which revealed 𝐵2 ≈ 0 and 𝛽 ≈ 0, confirming that the experiments approximate neutral squirmer behavior. We note that the biphasic droplet is not exactly axisymmetric, and thus 𝛽 is not expected to be exactly zero (consistent with the presence of spiraling)[48]. Our experiments also yielded an effective 𝐵1~ 76 3 m/s (yellow curve, Figure 5.3c), which differs from the predicted value of 𝐵1 = 𝑣 ~ 2 40 m/s, where 𝑣 is the droplet velocity, likely a consequence of asymmetric flow across the surface of the Janus droplets. For droplets in State II, four stagnation points were observed (Figure 5.3d), with the two additional stagnation points found near the three phase contact line towards the trailing edge (i.e., a puller-type squirmer). During State II, the flow field near the isotropic compartment changes as a function of the domain size. Figure 5.3d shows one example where the inner interface is located at 𝜃 = 120o, and the two additional stagnation points were found at 𝜃 = 130o. Fitting of the experimental velocity data to the squirmer model reveals 𝐵2 ~ 42 m/s > 0 and 𝛽 = 0.54 > 0, confirming again puller-type behavior; good agreement with the value of 𝛽 = 0.52, estimated from the location of the stagnation points (𝑢𝜃(130 o) = 0), is also evident. A phase shift was observed for experimental values of 𝑢𝜃 (red curve, Figure 5.3d), consistent with the effects of a density mismatch between tracer and aqueous fluid (inertia). Specifically, a theoretical framework exists for the analysis of various squirmer types, and this framework has the potential to provide insight into experimental observations such as the number of vortices within the multicompartment droplets, the reorientation of the inner interface, and the increase in velocity of the droplets associated with the State II to State III transition. The evidence provided above leads us to conclude that the transition from the ballistic (State II) to the spiraling (State III) motion is a switch from puller-type to 139 neutral-type squirmer behavior. We quantified the morphology of the two-compartment droplet at the transition between squirmer behaviors by the angle 𝜃N defining the relative size of nematic and isotropic compartments (Figure 5.4a); the value of 𝜃N at the ballistic-to-spiraling transition was found to be about 65o. We also observed that smaller droplets transitioned from puller (State II) to neutral squirmer (State III) at larger values of 𝜃N. This size-dependence likely reflects a competition between viscous and elastic stresses within the LC droplet[25], as characterized by the Ericksen number 𝑣𝜂𝑟/𝐾 (𝐾 is the elastic constant; elastic stresses dominate with decrease in 𝑟). We generated a color map of the droplet velocity along its trajectory (Figure 5.4c) as a function of 𝜃N (Figure 5.4d and Supplementary Figure S5.6). Interestingly, the droplet gained velocity, from 17 m/s to 25 m/s, during the transition from State II to State III, as denoted by the yellow arrow. The change in velocity is accompanied by a decrease in the number of vortices, from four to two, in the puller and neutral squirmer, respectively, and thus a decrease in viscous dissipation within the oil domains. In addition, in contrast to the neutral squirmer, the nematic domain within State II exhibits flow against droplet motion, likely contributing to the lower droplet velocity in State II. We also observed fluctuations in droplet velocities when droplets spiraled across their own trajectory (red, black and blue arrows in Figure 5.4c and d), a phenomenon caused by negative autochemotaxis (see Supporting Information)[47]. Morphogenesis and droplet behavior in the presence of external surfactant gradients In the presence of external gradients in surfactant concentration, we found that droplet motion can dramatically reconfigure internal morphology to increase the number of internal domains. Fig. 5 shows droplets migrating along SDS gradients (Fig. 5 and Supplementary Video 5, or TTAB gradients, Supplementary Fig. S7), moving to 140 the bottom right corner where the SDS concentration was highest. The observation in Figure 5.5 was recorded at 53°C, with the droplet motion corresponding to State II (nematic domain trailing the motion (Figure 5.5a)). With increase in speed of the droplet along the concentration gradient, we observed advection inside the droplet to deform the inner interface (Figure 5.5b-g). Ultimately, the isotropic domain broke into two lobes (Figure 5.5h). Figure 5.5 Time-lapse micrographs showing the reorganization of the morphology of a biphasic Janus droplet, with the isotropic (I) domain breaking into two lobes when moving towards the bottom right along an external surfactant concentration gradient. Scale bar, 50 m. Past studies have shown that the coupling of advection and solubilization can generate sustained interfacial tension gradients (Marangoni stresses) across droplet surfaces[17]. The rate of energy dissipation due to the surface flow[17] can be estimated as ~2πrvΔγ, where r ~10-5 m is the radius of the droplet, v ~10-5 m/s is the droplet velocity, and Δγ is the change in interfacial tension across the aqueous interface of the droplet. We estimate that the energy stored in the inner interface increases by ~πr2γ over approximately Δt = 5 s, where γ is the interfacial tension of the inner interface, measured to be of order of 10-5 N/m[30]. Because the work done over 5 s by the interfacial tension 141 gradient on the droplet outer interface (2πrvΔγΔt) must be larger than the energy stored in the extended inner interface (πr2γ), we conclude that Δγ > 10-6 N/m. We note that this value is small compared to the absolute values of the interfacial tensions of the outer interfaces (~10-3 N/m) and thus consistent with our observation that the droplets are spherical during self-propelled motion (model spherical squirmers). Overall, this result emphasizes the complex interplay between morphology and dynamics that exists in this active, multiphase emulsion system. 4. Conclusions We have used multiphase oil droplets to reveal how morphogenesis of domains internal to the droplets can impact emergent dynamics when driven by active processes. Our experimental approach embeds a number of advances in soft matter science, including the design of multiphase droplets with low internal interfacial tensions (that maintain spherical microswimmer geometry), and selective solubilization of components of the droplets by micellar extraction, leading to both morphogenesis and self-propulsion. We show that morphogenesis of the emulsion droplets and emergent changes in dynamics, including cycles of ballistic and spiraling behaviors, can be described within the physical framework of squirmer models for spherical microswimmers in Stokes flow. A range of complex behaviors were shown to be programmed via morphology, including a discontinuous transition from ballistic (puller) to spiraling (neutral squirmer) behaviors, followed by continuous transition back to ballistic (neutral squirmer), as the size of the nematic domain increased with progressive solubilization of the isotropic oil. Our results also reveal that some transitions between squirmer states are accompanied by a decrease in drag. Overall, our results establish an experimental platform that can be mathematically described by the squirmer model, opening opportunities for future experimental studies of the behaviors 142 and interactions of various types of squirmers. In the long-term, the physical framework established by such studies offer the potential to guide the engineering of functional soft matter systems that exhibit the complexity of behaviors and functions found in living systems that undergo morphogenesis (e.g., ability to change form in response to environment to modify dynamics). 5. Supporting Information Additional comments on past studies of nematic droplets. We note that past studies have reported that high TTAB concentrations (e.g., 250 mM) can lead to curly motions of single phase nematic droplets due to an interplay between convective flow and the elasticity of the internal nematic phase[10,25]. In our experiments, we used a low TTAB concentration (100 mM) [25] that, as discussed in the main text, does not generate viscous stresses sufficient to break the symmetry of single phase nematic droplets. Additional comments on negative autochemotaxis that lead to oscillation of droplet velocity. Negative autochemotaxis is an effective repulsion acting between a droplet and it previous trajectory[47]. In our study, the autochemotactic effect reflects the presence of oil-swollen micelles along the droplet trajectory, thus causing deceleration of a droplet when it approaches its past trajectory and acceleration of the droplet when moving away from the previous trajectory. Inspection of Figure 5.4d also reveals that the changes in velocity with time become less abrupt with each recrossing of the droplet trajectory, consistent with diffusive broadening of the concentration profile of oil-swollen micelles in the wake left behind the droplet. 143 Figure S5.1 Preferential extraction of perfluorobenzene (FB) from emulsion droplets prepared from mixtures of FB and E7. (a) Phase transition from isotropic to nematic phase during extraction of a droplet at 58 oC in 100 mM TTAB micellar solution. Scale bar, 50 m. Gas chromatogram showing the composition of droplets (b) before and (c) after the solubilization process in the TTAB solutions. 5CB, 7CB, 8OCB and 5CT are the four components of the E7 liquid crystalline mixture. 144 Figure S5.2 Trajectory during which the handedness of the spiralling motion switched sign (denoted by blue arrows). Scale bar, 50 m. This switch was observed more frequently with droplets with sizes (~30 - 40 m) that were small compared to the aqueous film thickness (~100 m, Z- axis), suggesting that hydrodynamic interactions with the confining surfaces suppress switching. 145 Figure S5.3 Reorganization of the internal morphology of droplet during the transition from State IV to V dynamical states. Time-lapse micrographs show that the defect (denoted by red arrows) is convected to the leading edge during the transition from State IV to State V. Scale bar, 50 m. 146 We have explored the phase behavior of perfluorobenzene (FB) and E7 in a prior publication (ref. 30). For the convenience of readers, we show a schematic illustration of the phase diagram (Figure S5.4a, temperature vs. concentration of FB in LC). To observe the spiraling motion, the oil phase in the droplet must be in the N+I coexistence region. Oil compositions that exhibit N+I coexistence at low temperature, e.g., mixture (i) in Figure S5.4a, have an overall higher FB concentration compared to oil compositions that exhibit N+I coexistence at high temperature, e.g., mixture (ii) in Figure S5.4a. The low concentration of FB at high temperature will lead to a low rate of solubilization and thus low propulsive velocity and long rotational period. We note that we have measured that the velocity does depend on FB concentration (Figure S5.4b, droplet velocity decreases as a function of time, i.e., FB concentration). Figure S5.4 (a) Schematic illustration of the phase diagram of the LC+FB oil phase. (b) Velocity of a nematic single phase droplet that comprised of the LC-FB mixture deceases as a function of time (i.e., FB concentration). 147 The droplet behaviors described in Figure 5.1 are driven by a phase transition (isotropic to nematic) that accompanies solubilization of select components of the oil mixture at constant temperature. This change in composition is depicted as process (1) in Figure S5.5. As reported in the main text, after the isothermal solubilization process had resulted in formation of a nematic droplet (ending at point A in Figure S5.5), we increased the temperature, causing the nematic oil mixture to transform back into an isotropic phase again (process (2) in Figure S5.5). After the thermal phase transition, we allowed the solubilization experiment (process (3)) to again transform the isotropic droplet into a nematic droplet. This process was repeated up to 5 times in our system. Figure S5.5 Schematic illustration of the phase diagram of the LC+FB oil phase. 148 Figure S5.6 Velocity of droplet motion during State III showing oscillations due to autochemotaxis while spiralling (re-crossing of trajectory). 149 Figure S5.7 Reconfiguration of droplet morphology with external concentration gradient of TTAB. Time-lapse micrographs showing the isotropic domain of a Janus droplet (isotropic and nematic domains) breaking up into two lobes when moving to the right following an external TTAB gradient. Scale bar, 50 m. We used SDS in Figure 5.5 of the main text for ease of visualization because SDS generates a slower surface flow enabling formation of a steady state before the droplet reached the edge of the glass optical cell. 150 Supplementary Video 1: Movie showing a Janus droplet transitioning from ballistic (State II, puller) to spiraling (State III, neutral squirmer) and back to ballistic (State IV, neutral squirmer) motion, with orientation of internal interfaced between compartments changing as indicated in Figure 5.1 and Supplementary Video 2. Supplementary Video 2: Movie showing the inner interface of a Janus droplet reorienting from perpendicular to parallel to the direction of motion during the transition from puller (State II) to neutral squirmer (State III). Supplementary Video 3: Movie showing that spiraling motion (State III) of the droplet can be reversed back to ballistic motion (State II) by increasing temperature by 1 oC. Supplementary Video 4: Movie showing how tracer particles are used to quantify the flow field near the surface of the self-propelled droplets. The droplet shown is undergoing a spiraling motion (State III), with an isotropic domain on the left and a nematic domain on the right compartment of the droplet. The displacement of the tracer particles reveals a higher flow rate near the isotropic surface (40 m) versus nematic surface (9 m) of the droplet. Supplementary Video 5: Movie showing the motion and morphological reconfiguration of droplets in an external surfactant gradient, as described in Figure 5.5. 151 Acknowledgements The authors acknowledge support from the Department of Energy, Basic Energy Sciences, Division of Materials Research, Biomaterials Program under Grant No. DE- SC0019762. 6. References *This chapter was prepared as a Research Article reporting original research in the journal Soft Matter. My contribution to this project was in designing and performing the experiments, developing the mechanism and writing the manuscript. My co-authors contributed to the designing of the project, experiments, developing the mechanism and writing of the manuscript. Adapted with permission from: X. Wang, R. Zhang, A. Mozaffari, J. J. de Pablo, and N. L. Abbott, Active Motion of Multiphase Oil Droplets: Emergent Dynamics of Squirmers with Evolving Internal Structure, Soft Matter 17, 2985 (2021). Copyright The Royal Society of Chemistry 2021. [1] L. Berthier and J. Kurchan, Non-Equilibrium Glass Transitions in Driven and Active Matter, Nat. Phys. 9, 310 (2013). [2] B. X. Li, V. Borshch, R. L. Xiao, S. Paladugu, T. Turiv, S. V. Shiyanovskii, and O. D. Lavrentovich, Electrically Driven Three-Dimensional Solitary Waves as Director Bullets in Nematic Liquid Crystals, Nat. Commun. 9, 1 (2018). [3] A. U. Oza, L. Ristroph, and M. J. Shelley, Lattices of Hydrodynamically Interacting Flapping Swimmers, Phys. Rev. X 9, 041024 (2019). [4] H. R. O. Sohn and I. I. Smalyukh, Electrically Powered Motions of Toron Crystallites in Chiral Liquid Crystals, Proc. Natl. Acad. Sci. U. S. A. 117, 6437 (2020). [5] A. Izzet, P. G. Moerman, P. Gross, J. Groenewold, A. D. Hollingsworth, J. 152 Bibette, and J. Brujic, Tunable Persistent Random Walk in Swimming Droplets, Phys. Rev. X 10, 021035 (2020). [6] T. Toyota, H. Tsuha, K. Yamada, K. Takakura, T. Ikegami, and T. Sugawara, Listeria -like Motion of Oil Droplets, Chem. Lett. 35, 708 (2006). [7] J. R. Howse, R. A. L. Jones, A. J. Ryan, T. Gough, R. Vafabakhsh, and R. Golestanian, Self-Motile Colloidal Particles: From Directed Propulsion to Random Walk, Phys. Rev. Lett. 99, 048102 (2007). [8] F. Kümmel, B. ten Hagen, R. Wittkowski, I. Buttinoni, R. Eichhorn, G. Volpe, H. Löwen, and C. Bechinger, Circular Motion of Asymmetric Self-Propelling Particles, Phys. Rev. Lett. 110, 198302 (2013). [9] Z. Izri, M. N. van der Linden, S. Michelin, and O. Dauchot, Self-Propulsion of Pure Water Droplets by Spontaneous Marangoni-Stress-Driven Motion, Phys. Rev. Lett. 113, 248302 (2014). [10] C. Krüger, G. Klös, C. Bahr, and C. C. Maass, Curling Liquid Crystal Microswimmers: A Cascade of Spontaneous Symmetry Breaking, Phys. Rev. Lett. 117, 048003 (2016). [11] B. V. Hokmabad, K. A. Baldwin, C. Krüger, C. Bahr, and C. C. Maass, Topological Stabilization and Dynamics of Self-Propelling Nematic Shells, Phys. Rev. Lett. 123, 178003 (2019). [12] A. R. Bausch and K. Kroy, A Bottom-up Approach to Cell Mechanics, Nat. Phys. 2, 231 (2006). [13] N. Kumar, H. Soni, S. Ramaswamy, and A. K. Sood, Flocking at a Distance in Active Granular Matter, Nat. Commun. 5, 1 (2014). [14] J. P. Celli et al., Helicobacter Pylori Moves through Mucus by Reducing Mucin Viscoelasticity, Proc. Natl. Acad. Sci. U. S. A. 106, 14321 (2009). [15] E. Lauga and T. R. Powers, The Hydrodynamics of Swimming Microorganisms, 153 Reports Prog. Phys. 72, 096601 (2009). [16] V. Schaller, C. Weber, C. Semmrich, E. Frey, and A. R. Bausch, Polar Patterns of Driven Filaments, Nature 467, 73 (2010). [17] S. Herminghaus, C. C. Maass, C. Krüger, S. Thutupalli, L. Goehring, and C. Bahr, Interfacial Mechanisms in Active Emulsions, Soft Matter 10, 7008 (2014). [18] F. Lancia, T. Yamamoto, A. Ryabchun, T. Yamaguchi, M. Sano, and N. Katsonis, Reorientation Behavior in the Helical Motility of Light-Responsive Spiral Droplets, Nat. Commun. 10, 1 (2019). [19] E. Lauga, W. R. DiLuzio, G. M. Whitesides, and H. A. Stone, Swimming in Circles: Motion of Bacteria near Solid Boundaries, Biophys. J. 90, 400 (2006). [20] J. S. Guasto, R. Rusconi, and R. Stocker, Fluid Mechanics of Planktonic Microorganisms, Annu. Rev. Fluid Mech. 44, 373 (2012). [21] A. S. Chin, K. E. Worley, P. Ray, G. Kaur, J. Fan, and L. Q. Wan, Epithelial Cell Chirality Revealed by Three-Dimensional Spontaneous Rotation, Proc. Natl. Acad. Sci. U. S. A. 115, 12188 (2018). [22] P. D. Todorov, P. A. Kralchevsky, N. D. Denkov, G. Broze, and A. Mehreteab, Kinetics of Solubilization of N-Decane and Benzene by Micellar Solutions of Sodium Dodecyl Sulfate, J. Colloid Interface Sci. 245, 371 (2002). [23] K. Peddireddy, P. Kumar, S. Thutupalli, S. Herminghaus, and C. Bahr, Solubilization of Thermotropic Liquid Crystal Compounds in Aqueous Surfactant Solutions, Langmuir 28, 12426 (2012). [24] C. C. Maass, C. Krüger, S. Herminghaus, and C. Bahr, Swimming Droplets, Annu. Rev. Condens. Matter Phys. 7, 171 (2016). [25] M. Suga, S. Suda, M. Ichikawa, and Y. Kimura, Self-Propelled Motion Switching in Nematic Liquid Crystal Droplets in Aqueous Surfactant Solutions, Phys. Rev. E 97, 062703 (2018). 154 [26] T. Yamamoto and M. Sano, Chirality-Induced Helical Self-Propulsion of Cholesteric Liquid Crystal Droplets, Soft Matter 13, 3328 (2017). [27] L. D. Zarzar, V. Sresht, E. M. Sletten, J. A. Kalow, D. Blankschtein, and T. M. Swager, Dynamically Reconfigurable Complex Emulsions via Tunable Interfacial Tensions, Nature 518, 520 (2015). [28] J. Jeong, A. Gross, W.-S. Wei, F. Tu, D. Lee, P. J. Collings, and A. G. Yodh, Liquid Crystal Janus Emulsion Droplets: Preparation, Tumbling, and Swimming, Soft Matter 11, 6747 (2015). [29] T. Nisisako, Recent Advances in Microfluidic Production of Janus Droplets and Particles, Curr. Opin. Colloid Interface Sci. 25, 1 (2016). [30] X. Wang, Y. Zhou, Y.-K. Kim, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Thermally Reconfigurable Janus Droplets with Nematic Liquid Crystalline and Isotropic Perfluorocarbon Oil Compartments, Soft Matter 15, 2580 (2019). [31] N. Pannacci, H. Bruus, D. Bartolo, I. Etchart, T. Lockhart, Y. Hennequin, H. Willaime, and P. Tabeling, Equilibrium and Nonequilibrium States in Microfluidic Double Emulsions, Phys. Rev. Lett. 101, 1 (2008). [32] X. Wang, Y. Zhou, V. Palacio-Betancur, Y.-K. Kim, L. Delalande, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Reconfigurable Multicompartment Emulsion Drops Formed by Nematic Liquid Crystals and Immiscible Perfluorocarbon Oils., Langmuir 35, 16312 (2019). [33] M. Li, M. Brinkmann, I. Pagonabarraga, R. Seemann, and J.-B. Fleury, Spatiotemporal Control of Cargo Delivery Performed by Programmable Self-Propelled Janus Droplets, Commun. Phys. 1, 23 (2018). [34] J. R. Blake, A Spherical Envelope Approach to Ciliary Propulsion, J. Fluid Mech. 46, 199 (1971). [35] G.-J. Li and A. M. Ardekani, Hydrodynamic Interaction of Microswimmers near 155 a Wall, Phys. Rev. E 90, 013010 (2014). [36] L. Berg, E. Solomon, and D. W. Martin, Biology (Boorks/Cole, Cengage Learning: USA, 2011). [37] J. L. Capinera, Encyclopedia of Entomology (Springer Science & Business Media, 2008). [38] N. J. Severs, The Cardiac Muscle Cell, BioEssays 22, 188 (2000). [39] S. Van Teeffelen and H. Löwen, Dynamics of a Brownian Circle Swimmer, Phys. Rev. E 78, 020101 (2008). [40] R. Wittkowski and H. Löwen, Self-Propelled Brownian Spinning Top: Dynamics of a Biaxial Swimmer at Low Reynolds Numbers, Phys. Rev. E 85, 021406 (2012). [41] N. Tarantino, J.-Y. Tinevez, E. F. Crowell, B. Boisson, R. Henriques, M. Mhlanga, F. Agou, A. Israël, and E. Laplantine, TNF and IL-1 Exhibit Distinct Ubiquitin Requirements for Inducing NEMO-IKK Supramolecular Structures, J. Cell Biol. 204, 231 (2014). [42] S. Burov, S. M. Ali Tabei, T. Huynh, M. P. Murrell, L. H. Philipson, S. A. Rice, M. L. Gardel, N. F. Scherer, and A. R. Dinner, Distribution of Directional Change as a Signature of Complex Dynamics, Proc. Natl. Acad. Sci. U. S. A. 110, 19689 (2013). [43] K. L. Ngai, Relaxation and Diffusion in Complex Systems (2011). [44] A. Mjaavatten, Curvature of a 2D or 3D Curve, (unpublished). [45] R. Seemann, J.-B. Fleury, and C. C. Maass, Self-Propelled Droplets, Eur. Phys. J. Spec. Top. 225, 2227 (2016). [46] A. G. Chmielewski, Viscosity Coefficients of Some Nematic Liquid Crystals, Molecualr Cryst. Liq. Cryst. 132, 339 (1986). [47] C. Jin, C. Krüger, and C. C. Maass, Chemotaxis and Autochemotaxis of Self- Propelling Droplet Swimmers, Proc. Natl. Acad. Sci. U.S.A 114, 5089 (2017). 156 [48] O. S. Pak and E. Lauga, Generalized Squirming Motion of a Sphere, J. Eng. Math. 88, 1 (2014). 157 Chapter 6 Stimuli-Responsive Liquid Crystal Printheads for Spatial and Temporal Control of Polymerization 1. Introduction Polymerization reactions triggered in response to chemical, optical and physical cues[1,2] are at the core of a wide range of technologies, including photolithography,[3– 5] polymeric coatings[6], nanomedicine[7,8], pollutant removal[9], shape-memory[10] and self-healing[11,12] materials. In conventional chain-growth polymerization[13], initiation of polymerization is typically achieved by thermal decomposition[2], photolysis,[2,14,15] or electrolysis[1] of molecular initiators that are tailored to respond to targeted triggers (e.g., light of a particular frequency to decompose an azo initiator)[2,16]. Another strategy is to sequester reagents for polymerization within nano- or microcapsules[17–19] that release their contents upon arrival of targeted stimuli, such as redox or mechanical cues (stress/fracture/damage/self-healing etc.)[17,19]. Here we report an approach to achieving spatially and temporally controlled polymerization in response to a range of cues by using initiator loaded stimuli-responsive LCs. Our approach is inspired by the success of stereolithographic technologies for the fabrication of 3D polymeric structures[3–5,20,21]. Conventional stereolithography involves the rastering of a light beam across a solution containing monomer and initiator to control the time and location of polymerization[20,21]. This permits additive assembly of a wide range of complex 3D microstructures but is limited to photoinitiation of the polymerization process. The principles developed herein provide a first step towards a technology that permits spatial and temporal control over the synthesis of polymeric structures in response to local chemical cues. The approach involves 158 sequestering initiators for polymerization within LCs, and uses the well-known stimuli- responsive properties of LCs to program ejection of the initiator into solution at a desired time and location to synthesize a targeted polymeric structure. 2. Results and Discussion We designed LC printheads that sense interactions at the interface of the LC and respond by ejecting initiator-filled microdroplets to trigger synthesis of polymers. The experimental set-up comprises three phases (Figures 6.1 and 6.2a): (i) a bulk aqueous phase containing monomers, cross-linker (if used) and catalyst; (ii) a water-immiscible LC phase called E7, which is a mixture of cyanobiphenyl and terphenyl mesogens; and (iii) a micrometer-sized aqueous droplet phase dispersed within the LC containing the initiator ammonium persulfate (APS) and the anionic surfactant sodium dodecyl sulfate (SDS). We used E7 in our experiments because it has a broad nematic temperature range (from below room temperature to 60 °C), and thus remains in the LC state during operation of the LC printhead (including in the presence of any heat generated by the polymerization process). As detailed below, the triggering of the LC printhead occurs via the arrival of chemical stimuli at the LC-aqueous interface, and the system responds via a combination of elastic and electrical double layer interactions that act on the micrometer-sized aqueous droplets within the LC; the aqueous microdroplets, which are filled with an initiator for polymerization, are ejected into the bulk aqueous phase to cause polymerization. The elastic and electrical double layer interactions that control the response of the LC printhead have been described previously[22–24]. In brief, the strain of the LC around the microdroplets generates a repulsive elastic interaction between the microdroplets and the interface between the LC and bulk aqueous phase, which can be evaluated as: 𝑅 4 𝐹e = −(2.04) 2𝐴𝜋𝐾 ( ) (6.1) ℎ+𝑅 159 where A is a constant that depends on the orientation of the LC at the bulk aqueous interface (3/4 and 1/2 for planar and homeotropic anchoring, respectively), K ~10-11 N is the elastic constant of the LC, R is the radius of the microdroplet, and h is the distance Figure 6.1 Molecular structures of (a) E7 (liquid crystal), (b) N- isopropylacrylamide (NIPAm, monomer), (c) 2-hydroxypropyl methacrylate (HPMA, monomer), (d) diacetone acrylamide (DAAm, monomer), (e) N,N’- methylenebis(acrylamide) (BIS, cross-linker), (f) N,N,N’,N’- tetramethylethylenediamine (TEMED, catalyst), (g) ammonium persulfate (APS, initiator) and (h) methacryloxyethyl thiocarbamoyl rhodamine B (fluorescent dye monomer). 160 separating the microdroplets from the LC-bulk aqueous interface. For a 1 m droplet near the interface (h = 0) and planar anchoring of E7 at the water interface[25], 𝐹e is evaluated to be 9.8 × 10−11 N, which is large compared to the gravitational force 𝐹g ~ 8 × 10−22 4 N (∆𝜌𝑔 ∙ 𝜋𝑅3 , where ∆𝜌 ~ 2 × 10−5 kg m-3 is the density difference 3 between aqueous droplet and E7, see Supporting Information, 𝑔 ~ 9.81 m s-2 is the gravitational acceleration). The electrical double layer interaction arises from the presence of charges (due to SDS) at the interface of the aqueous microdroplets and the bulk aqueous phase, and can be evaluated as[24,26,27]: 𝑅 𝑘 𝑇 2 ℎ 𝐹edl = −4𝜋𝜀0𝜀 ( B ) 𝑌p𝑌i𝑒 − 𝜆 (6.2) 𝜆 𝑒𝑐 where 𝜀0 is vacuum permittivity, 𝜀 is the relative permittivity of the LC (for E7 with homeotropic anchoring, 𝜀∥ = 19 ; planar anchoring, 𝜀⊥ = 5.2 )[28], 𝜆 is Debye screening length, 𝑘B is the Boltzmann constant, 𝑇 is temperature, 𝑒𝑐 is the elementary charge, 𝑌p is the effective surface potential of the microdroplet, 𝑌i the is effective surface potential of the LC-bulk aqueous interface (see Supporting Information and Figure S6.1 for calculations). Inspection of Figure 6.2b and c reveals that equations 1 and 2, when combined, predict that adsorption of a cationic amphiphile at the bulk aqueous-LC interface (input) will be processed by the LC printhead (via changes in both electrical double layer and elastic interations) to generate a force that ejects the microcargo from the LC and thus intiates polymerization (output). To test the predicted initiation of polymerization, we filled mini-wells (3.5 mm in depth) with LC containing initiator-filled microdroplets. We submerged the mini- wells into a bulk aqueous phase containing monomer N-isopropylacrylamide (NIPAm) and 1 vol% catalyst N,N,N’,N’-tetramethylethylenediamine (TEMED) (Figure 6.2d), and subsequently added 10 mM cationic surfactant dodecyltrimethylammonium bromide (DTAB). After 2 hours, we collected the bulk aqueous phase and heated it to 60 °C for 10 min (PNIPAm possesses a lower critical solution temperature (LCST) of 161 Figure 6.2 Design principles for LC printheads that trigger polymerization. a) Schematic illustrations of SDS-stabilized and initiator-filled microdroplets in LC. The microdroplets are initially repelled from the LC interface (red arrow) due to elastic and electrical double layer forces; addition of DTAB alters both forces, thus generating a net attraction (red arrow) that results in droplet ejection from the LC. Black arrows in (a, right) indicate droplet release. b,c) Calculated elastic force (Fe, yellow), electrical double layer force (Fedl, blue) and net force (Fnet, red) acting on a microdroplet (R = 1.5 m, positive indicates attraction) before and after arrival of the cationic stimulus, respectively. d) Schematic illustration a miniwell filled with LC containing microdroplets; the mini-well is submerged under an aqueous solution containing monomer (10 wt% NIPAm) and catalyst (1 vol% TEMED). e,f) Optical microg raphs of aqueous solution from (d) before (e) and after (f) exposure to DTAB (f) at 60 °C (above the LCST of PNIPAm (32 °C)), revealing formation of polymer 162 only in the presence of the chemical trigger. g-j) Optical images and k) corresponding schematic illustrations of interfacial polymer film triggered to form by release of initiators and local heating (from polymerization) in the presence of 3 vol% TEMED. containing monomer (10 wt% NIPAm) and catalyst (1 vol% TEMED). e,f) Optical micrographs of aqueous solution from (d) before (e) and after (f) exposure to DTAB (f) at 60 °C (above the LCST of PNIPAm (32 °C)), revealing formation of polymer only in the presence of the chemical trigger. g-j) Optical images and k) corresponding schematic illustrations of interfacial polymer film triggered to form by release of initiators and local heating (from polymerization) in the presence of 3 vol% TEMED. l) Intensity of reflected light (blue line) and the rate of increase of intensity (yellow line) in the region marked by red dashed lines in (j), revealing rate of formation of polymer. Scale bars, 5 mm (e,f), 2 mm (g). 32 °C)[29–31]. In the absence of added DTAB, a clear solution was observed upon heating (Figure 6.2e); in contrast, when DTAB was present, we observed heating to result in the formation of an opaque gel (Figure 6.2f), consistent with release of initiator triggered by adsorption of DTAB and subsequent polymerization (Figure S6.2 is the visualization of the release process achieved by adding ink to the aqueous microdroplets). Interestingly, using the same experimental design described above, when we increased the concentration of the catalyst TEMED to 3 vol%, we observed the in situ formation of a polymeric film at the interface of the LC and the bulk aqueous phase following addition of DTAB (Figure 6.2g-k, red frame in j). The growth of the film was quantified by the increase in the intensity of scattered light (Figure 6.2l), revealing that the maximum rate of growth occured 10 min after addition of DTAB. The subsequent slowing of growth suggests a self-limiting growth mechanism. At much longer times (> 2 hours), the film dissolved, indicating that heating during polymerization leads to a local temperature that exceeded the LCST of PNIPAm (32 °C)[31]. We confirmed formation of PNIPAm by 1H nuclear magnetic resonance (1H NMR) spectroscopy (Figure S6.4), and characterized the molecular weight of PNIPAm to be Mn = 279, 200 g mol-1, with a polydispersity index Ð = 1.4 by gel permeation chromatography (GPC, Figure S6.5). By adding cross-linker N,N’-methylenebis(acrylamide) (BIS (Figure 6.1); 163 0.05 wt% relative to monomer) to the aqueous solution, a permanently cross-linked film of PNIPAm was formed at the LC interface (Figure S6.3). We envisage locally-triggered synthesis of polymer films at interfaces to be potentially useful in various contexts, such as for the self-regulated synthesis of lubricating layers or for the entrapment of local targeted species at interfaces. A substantial challenge in the deployment of block copolymers in a range of technologies (e.g., for formulations) is precise control over their self-assembly, as they are often kinetically trapped in undesirable morphologies if added directly to solution[32,33]. One approach to overcoming this barrier is polymerization-induced self-assembly (PISA), where self-assembly occurs concurrently with synthesis, giving rise to predictable and well-defined morphologies of block copolymer nanostructures[34,35]. To determine if a LC printhead could be used to trigger PISA (Figure 6.3), we synthesized a macro chain transfer agent (macroCTA) consisting of poly(N,N-dimethylacrylamide) (PDMA) with 47 repeat units by reversible addition- fragmentation chain transfer (RAFT) polymerization (see 1H NMR spectrum in Figure S6.6, Supporting Information for synthesis). Our choice of monomers and monomer- initiator ratios were guided by prior studies of PISA[36]. PDMA serves as the macroCTA because it is water-soluble, and thus could be dispersed into a bulk aqueous phase along with a core-forming monomer (i.e., diacetone acrylamide (DAAm)). In the experiments shown in Figure 6.3, the total monomer concentration in water was 20 wt% (1 vol% TEMED, 10 mM DTAB were also in the bulk aqueous solution; the temperature was 40 °C)[36]. After triggering the LC printhead (using DTAB) with the goal of initiating PISA (Figure 6.3a-c), we employed GPC (Figure 6.3d) and 1H NMR (Figure S6.7) to characterize the polymers formed in the bulk aqueous phase. GPC confirmed the successful chain extension of PDMA47 macroCTA with DAAm monomers, as 164 Figure 6.3 LC printhead for polymerization-induced self-assembly (PISA). a-c) Schematic illustration of PISA triggered by release of initiator from a LC printhead. d) GPC traces of PDMA47 macroCTA and PDMA47-b-PDAAmX diblock copolymers synthesized by LC printhead, and a table showing degree of polymerization of DMA and DAAm blocks, number-average molecular weight (Mn) and dispersity (Ð). e) Intensity-average hydrodynamic diameters of assemblies formed by LC-triggered PISA (measured by DLS) using three ratios of monomers. f-h) TEM images of PDMA47-b-PDAAm71 (f), PDMA47-b-PDAAm118 (g) and PDMA47-b-PDAAm222 (h) based assemblies formed by LC-triggered PISA. evidenced by the the shorter elution times found for the diblock copolymers. NMR analysis further revealed that the degree of polymerization of DAAm repeating units in the block copolymers ranged from 71 to 222 (Figure 6.3d and S6.7). The intensity- average diameter obtained by dynamic light scattering (DLS) (Figure 6.3e) exhibited a modest increase in hydrodynamic size as the comonomer ratio of DMA to DAAm 165 increased from 1:1.5 (PDMA47-b-PDAAm71) to 1:2.5 (PDMA47-b-PDAAm118). A further increase in comonomer ratio to 1: 4 led to a significant increase in the size of the assemblies, suggestive of a morphological transformation. Transmission electron microscopy (TEM) confirmed this interpretation, revealing spherical nanostructures for PDMA47-b-PDAAm71 (DMA:DAAm = 1:1.5, Figure 6.3f), a mixture of spherical and short worm-like aggregates for PDMA47-b-PDAAm118 (1:2.5, Figure 6.3g) and vesicles for PDMA47-b-PDAAm222 (1:4, Figure 6.3h). This evolution of morphology is consistent with physical arguments using a so-called packing parameter model (see Supporting Information for details)[37]. Overall, these results demonstrate a LC printhead that can be used for chemically-triggered synthesis and in-situ self-assembly of block copolymers, which may provide, for example, new approaches to autonomous formulation of active pharmaceutical ingredients. A key feature of the LC printheads described in this paper is that the initiation of polymerization is regulated by local solution conditions at the interface of the LC. To exploit this attribute, we created a LC printhead by filling a glass capillary with LC- containing initiator-loaded microdroplets and then submerging the open end of the capillary into an aqueous solution (Figure 6.4a). The aqueous microdroplets were observed to be slightly denser than the LC, as evidenced by sedimentation of the microdroplets in LC after 5 hours (Figure 6.4b). This density difference provided a continuous flux of microdroplets towards LC-bulk aqueous interface (see Supporting Information for a discussion of density). When the LC printhead was immersed into a monomer solution in a cuvette (10 wt% NIPAm, 1 vol% TEMED, at 25-30 °C for observation), in the absence of surfactants, no polymer was observed to be synthesized (Figure 6.4c). The introduction of aqueous DTAB (also containing monomers and catalyst) to a concentration of 3 mM turned “on” the printing process (Figure 6.4d), and subsequent dilution to 1 mM DTAB switched “off” the process (Figure 6.4e). This 166 Figure 6.4 Stimulus-triggered dynamic control of polymerization using LC printheads. a) Schematic illustration of the operation of the LC printhead: polymerization is triggered at the end of a capillary by a local stimulus (cationic surfactant). b) Optical micrograph showing sedimentation of SDS-stabilized aqueous microdroplets (containing the initiator APS) dispersed in E7 (left is initial state and right is after 5 hours). c-h) Dynamical control of polymerization by (d,f) increasing (polymerization on, red arrows in h) and (e,g) decreasing (polymerization off, green arrows in h) DTAB concentration at the LC printhead (10 wt% NIPAm, 1 vol% TEMED). h) Intensity of light scattered from the region of solution indicated by the dashed rectangle in (g), with and without the chemical stimulus (DTAB) at the LC printhead. i) Control experiment showing that 5 mM SDS (10 wt% NIPAm, 1 vol% TEMED) does not trigger polymerization, in contrast to (d,f) DTAB. j-m) LC printheads triggered by a range of chemical stimuli, including multivalent ions (e.g., Pb2+ in j) and biomolecules (e.g., polyarginine in k), and LC printheads that initiate polymerization of various monomers, including HPMA (l, 5 wt%) and DAAm (m, 10 wt%). n-q) Polymeric structures fabricated using LC printheads: n) letter “L”, o) letter “C” and p) shape of a heart (10 wt% monomers, cross-linker BIS:NIPAm = 1:9, 1 vol% TEMED). A monomer solution containing a fluorescent dye was used in (p). q) Fluorescent micrograph showing the edge of the polymeric structure in (p). r) Schematic illustrations and s) optical images of polymeric structures fabricated as “negative” prints of networks of LC printheads (see text for details). Scale bars, 5 mm (b, n-p, s), 1 mm (g, i-m), 100 m (q). process was repeated multiple times until the microdroplets in the LC were exhausted 167 (Figure 6.4f,g). We characterized the process by measuring the intensity of light (scattered) from solution near the printhead (Figure 6.4h), revealing that dynamical control of polymer printing can be achieved by the LC system. In contrast, in the presence of anionic surfactant (5 mM SDS, Figure 6.4i) in the bulk aqueous phase (10 wt% NIPAm, 1 vol% TEMED), no release or polymer synthesis was observed. Additionally, we demonstrated that the principles are not limited to DTAB or PNIPAm. A range of stimuli, such as multivalent ions (e.g., Pb2+, Figure 6.4j) and biomolecules (e.g., polyarginine, Figure 6.4k), that change surface potentials (Figure S6.1) can also trigger the polymerization (see Supporting Information). In addition, various monomers, including 2-hydroxypropyl methacrylate (HPMA, 5 wt%, Figure 6.4l) and DAAm (10 wt%, Figure 6.4m), can also be polymerized using LC printheads. To preserve the polymeric structures formed using LC printheads, we added the cross-linker BIS (Figure 6.1) into aqueous solutions containing 10 wt% of monomer (10 wt% NIPAm + BIS, NIPAm:BIS = 9:1, 1 vol% TEMED, 5 mM DTAB). With cross- linker (BIS) included, we observed polymer fibers to be printed, leading ultimately to permanent soft structures at the bottom of the container within 20 min. Hand writing using the above-described setup (Figure 6.4n-q) revealed that LC printheads can be used to fabricate well-defined soft structures to which we were also able to integrate fluorescence (dye monomer: methacryloxyethyl thiocarbamoyl rhodamine B, 0.01 wt% of total monomers, Figure 6.4p,q). These results hint at new approaches to "additive fabrication", in which the printing is controlled by local solution conditions. Finally, we used LC droplets supported on surfaces simple networks of printheads that can be used to create polymeric arrays (Figure 6.4r,s). Specifically, we dispensed three droplets of LC containing initiator-filled microdroplets (20 wt% APS in microdroplets, droplet phase:E7 = 1:10 vol) onto a glass surface that was immersed in monomer solution to which a trigger was added (5 mM DTAB, 10 wt% NIPAm + 168 BIS, NIPAm:BIS = 9:1, 1 vol% TEMED, Figure 6.4r). After 20 min, crosslinked polymer filled the solution except at the locations of the LC printheads, which formed hollow hemispheres (Figure 6.4s), thus leading to a “negative” print. This approach is potentially useful for triggering polymerization at multiple locations where stimuli are present locally and delivery of initiator is challenging. 3. Conclusions Materials that use physical and chemical stimuli as inputs, perform on-board computation using interactions programmed by the material, and then output complex responses are often termed “adaptive” materials[38–41]. The design principles underlying the operation of the LC printhead can thus be viewed as an example of an adaptive material system. In contrast to traditional 3D printing methods such as fused filament fabrication[42] and stereolithography[4,5,15,21], we show how the arrival of chemical stimuli change the boundary conditions in a simple computation performed by the LC (using equations 1 and 2), the solution to which is the ejection of micrometer- scale inclusions (microcargo) from the LC[24]. By including initiators for polymerization as microcargo[43], we show that the locally computed response of the LC to a chemical stimulus provides the basis of fresh approaches to programmed polymerization. By using nanofluidics to generate spatially localized sources of chemical species, we anticipate that the principles reported in this paper will enable the fabrication of high spatial resolution polymeric structures. The principles reported in this paper leading to LC printheads are compatible with the broad range of stimuli to which LCs respond, many of which are "unconventional" stimuli for initiating polymerization, such as electric and magnetic fields[44], motion of bacteria[24] and adsorption of proteins[45]. The diversity of LC materials available, including achiral (e.g., nematic, smectic) and chiral LCs (e.g., 169 cholesteric, blue phases)[44], further expands the ways in which the responses of LC printheads can be programmed. We envisage the LC printheads described in this paper to be an early step towards a range of materials systems that function as autonomous devices, such as systems wherein formation of thin polymeric films are triggered by toxins (e.g., heavy metals) or bacteria[24] thus trapping the targets, new approaches to massively distributed manufacturing (MDM) where formulations based on assemblies of polymeric amphiphiles are triggered to form upon synthesis of active compounds (e.g., drugs), self-propelled LC droplets[46] that undergo chemotaxis to a desired location and then trigger polymerization, or multi-dimensional polymer printers that form complex structures in response to local solution conditions (e.g., action of an enzyme). 4. Supporting Information Experimental Methods Materials: Nematic liquid crystal E7 was purchased from HCCH (Jiangsu Hecheng Display Technology Co., Ltd). N-Isopropylacrylamide (NIPAm, 97%), N,N- dimethylacrylamide (DMA, 99%), 2-hydroxypropyl methacrylate (HPMA, 97%), diacetone acrylamide (DAAm, 99%), N,N’-methylenebis(acrylamide) (BIS), ammonium persulfate (APS), 2,2′‐azobis (isobutyronitrile) (AIBN, 98%), N,N,N’,N’- tetramethylethylenediamine (TEMED), sodium dodecyl sulfate (SDS) and dodecyltrimethylammonium bromide (DTAB) were purchased from Sigma-Aldrich. The fluorescent monomer methacryloxyethyl thiocarbamoyl rhodamine B (λexc = 548 nm/ λem = 570 nm) was purchased from Polysciences, Inc. A Sylgard 184 silicone elastomer kit for preparing polydimethylsiloxane (PDMS) was purchased from Dow Corning. Biopsy punches were purchased from Integra Miltex. NIPAm, HPMA, DMA were purified by passage through a short column of basic alumina prior to use. DAAm 170 was purified by crystallization twice from ethyl acetate before use. AIBN was recrystallized from methanol prior to use. The RAFT agent, 2- (dodecylthiocarbonothioylthio)-2-methylpropionic acid (DDMAT) was synthesized as described earlier[46,47]. Purification of water (18.2 MΩ cm resistivity at 25 °C) was performed using a Milli-Q water system (Millipore, Bedford, MA, USA). Synthesis of PDMA macroCTA (chain transfer agent): In a typical procedure, DDMAT (74 mg, 1.0 equiv.), DMA (1000 mg, 50 equiv.), and AIBN (6.7 mg, 0.2 equiv.) were dissolved in 2.0 ml of dry DMF in a 20 ml glass vial. The solution was degassed with nitrogen for 20 min, and then placed in a preheated oil bath at 65 °C. After 3 hours, polymerization was quenched by exposure to air and liquid nitrogen. The monomer conversion was determined to be 95% by 1H NMR spectroscopy. The product PDMA was recovered by precipitation into cold diethyl ether and dried under vacuum. Preparation of mini-wells: Mini-wells were made of the elastomer PDMS. Elastomer base and curing agent (Sylgard elastomer kit) were mixed at a ratio of 10:1, and subsequently cured at 60 °C for 2 hours to form a PDMS sheet. A PDMS mini-well, with a 6 mm outer diameter and a 3 mm inner hole, was obtained using 6-mm and 3- mm biopsy punches, respectively. We treated the PDMS with an oxygen plasma for 20 s and attached the PDMS to a glass substrate to create a mini-well with a depth of 3.5 mm. After fabrication, the mini-wells were stored for at least 3 days before use to obtain a hydrophobic PDMS surface. Preparation of LCs containing aqueous initiator microdroplets: First, SDS (2-5 mM) and APS (10 wt%) were dissolved into water. The density of the mixture was measured to be 1.05 g cm-3, larger than the density of E7 (measured to be 1.03 g cm-3). Second, 171 the resulting solution was emulsified into E7 at a 1:10 ratio by vortexing for 1 min at 3,000 rpm. Observation of microdroplet release and polymer formation: monomer solution was prepared by adding 10-20 wt% monomer and 1-3 vol% TEMED into water. For polymerization induced self-assembly (PISA), 20 wt% monomer (calculated based on the total amount of DMA and DAAm with PDMA47 macroCTA at the ratio indicated in the main text) and 1 vol% TEMED were added into water. Mini-wells filled with LC- microdroplet mixtures were submerged into the monomer solutions. To trigger the release of microdroplets containing initiator, DTAB solution was added into system, resulting in a final 10 mM concentration. The system was kept at room temperature for 2 hours for polymerization, or at 40 °C for 15 hours for PISA. Zeta potential measurements: 0.1 l E7 was homogenized for 30 s into 10 ml water. 240 l of LC-in-water emulsion was then diluted in 10 ml of either water, 5 mM SDS or 5 mM DTAB. When using 10 or 20 mM DTAB, polyarginine or Pb2+, 0.1 l E7 was homogenized for 30 s directly into 10 ml of solution to obtain zeta potential data. After allowing 30 min for the sample to equilibrate, zeta potentials on the aqueous side of the LC-aqueous interface were measured using a Malvern Zetasizer Nano instrument. Data were averaged over five consecutive runs. Dynamic light scattering (DLS): Product copolymer dispersions from the PISA experiments were diluted in water to 0.2 wt%. The intensity-average sphere-equivalent diameters of diblock copolymer nanoparticles were determined at room temperature by Zetasizer Nano-ZS (Malvern, 90° angle, wavelength 633 nm laser) via the Stokes−Einstein equation. Data were averaged over three consecutive runs. 172 1H Nuclear magnetic resonance (1H NMR): 1H NMR spectra were recorded on a Varian Inova spectrometer (500 MHz) in CDCl3. Chemical shifts are given in ppm downfield from tetramethylsilane TMS. Transmission electron microscopy (TEM): 20 l of block copolymer solution was applied onto a 400 mesh carbon grid (Ted Pella, INC.). Grids were then stained with 1% (w/v) uranyl acetate, washed, and allowed to dry. The grids were observed on a Hitachi HT 7700 microscope operating at 120 kV. The images were recorded with a slow-scan charge-coupled device (CCD) camera (Veleta 2k × 2k). Gel permeation chromatography (GPC): GPC measurements were performed on a Phenomenex Phenogel 5μ, 1K-75K column (300 x 7.80 mm) in series with a Phenomenex Phenogel 5μ, 10K-1000K column (300 x 7.80 mm) using HPLC grade solvents as eluents: dimethylformamide (DMF) with 0.05 M of LiBr at 60 °C. Detection was performed with a Hitachi UV-Vis Detector L-2420, a Wyatt Optilab T-rEX refractive index detector operating at 658 nm and a Wyatt DAWN® HELEOS® II light scattering detector operating at 659 nm. Absolute molecular weights and polydispersities were calculated using the Wyatt ASTRA software with dn/dc values determined by assuming 100% mass recovery during GPC analysis. Observation of microdroplet release and polymer formation from capillaries: Aqueous phases containing 10 wt% APS and 2–5 mM SDS were emulsified into E7 at 1:10 ratio. The resulting mixtures were loaded into glass or metal capillaries, and subsequently submerged into aqueous solutions that contain 5–10 wt% monomer, 1 vol% TEMED and in 0–5 mM DTAB or SDS at 25–30 °C for observation. Cross-linker 173 (BIS) was used in a few experiments at 10 wt% of monomers in total. Fluorescent monomer was used in indicated experiments at 0.01 wt% of total monomers. The hand- written structures were observed directly after fabrication using an Olympus IX71 inverted epifluorescence microscope (Center Valley, PA) equipped with a Hg lamp, and an Olympus U-MNB2 filter (470 nm ≤ λexc ≤ 490 nm; λem ≥ 520 nm) to excite the monomer. Trapping of initiator microdroplets. Initiator microdroplets (microcargo) in LC are sequestered within the LC bulk phase by the elastic energies associated with the deformation of LC around the microdroplets[43,44]. Assuming one elastic constant K, the elastic energy penalty depends on droplet size R (KR, typical K ~10-11 N), which competes with surface anchoring energy (WR2, where W ~10-5 N m-1 is anchoring energy density)[48]. A droplet suspended in LC with size R > K/W ~1 m (that is WR2 > KR) generates elastic repulsion that prevents coalescence with other droplets. Continuous release of microdroplets from capillary: To enable the continuous release of initiator from the capillary, the density of the microdroplets must be higher than that of the LC phase that controls droplet release. In our system, we measured the density of the APS aqueous phase (with SDS) to be 1.05 g cm-3, higher than the density of E7 of 1.03 g cm-3. Additionally, after 5 hrs, we observed microdroplets in LC to sediment to the bottom of a capillary (Figure 6.4b). This difference in density is insufficient to release the microdroplets as 𝐹e > 1000𝐹g (see above). Visualization of PNIPAm: Visualization of the PNIPAm (Figure 6.4d,f) is made possible due to heat generated during the polymerization process (above 32 °C locally), as described above. As the heat is dissipated, the polymer dissolves into the solution. 174 Using polyarginine as a stimulus: For polyarginine, the polymerization ceased after 5 min. We hypothesize that polyarginine has a low desorption rate from the interface between the LC and bulk aqueous phase, and that it forms long-lived complexes with the SDS at the interface. Calculation of elastic and electrical double layer interactions We evaluated the elastic repulsion between the LC-bulk aqueous interface and microdroplets as[22–24]: 𝑅 4 𝐹e = −(2.04) 2𝐴𝜋𝐾 ( ) (6.3) ℎ+𝑅 where A is a constant that is dependent on anchoring (3/4 and 1/2 for planar, where mesogens are parallel to interface, and homeotropic, where mesogens are perpendicular to interface, anchoring, respectively), K ~10-11 N is a characteristic LC elastic constant, R (> K/W) is the radius of the microdroplet, h is the distance of the microdroplet surface from the LC-bulk aqueous interface. For 1 m droplet near the interface (h = 0) and planar anchoring of E7 at water interface, 𝐹e is evaluated to be 9.8 × 10−11 N, which is large compared to gravitational forces 𝐹g ~ 8 × 10 −22 N 4 (∆𝜌𝑔 ∙ 𝜋𝑅3, where ∆𝜌 ~ 2 × 10−5 kg m-3 is the density difference between aqueous 3 droplet and E7, 𝑔 ~ 9.81 m s-2 is the gravitational acceleration). Other interactions, such as electrostatic attractions, can reinforce or overcome the elastic trapping of microdroplets[24]. Electrical double layer interactions between charges at the LC-bulk aqueous interface and the negatively charged SDS on the interface of microdroplets, thus generating either repulsions or attractions depending on the sign of the electric charges. We evaluated the forces generated by the electrical double-layer interactions as[24,26,27]: 𝑅 𝑘 𝑇 2 ℎ 𝐹edl = −4𝜋𝜀 𝜀 ( B 0 ) 𝑌 𝑌 − 𝜆 𝑒 p i 𝑒 𝜆 (6.4) 𝑐 175 where 𝜀0 is vacuum permittivity, 𝜀 is the relative permittivity of the LC (for E7 with homeotropic anchoring 𝜀∥ = 19; for planar anchoring 𝜀⊥ = 5.2)[28], 𝜆 is the Debye screening length, 𝑘B is the Boltzmann constant, 𝑇 is temperature, 𝑒𝑐 is the elementary charge, 𝑌p is the effective surface potential of the microdroplet, 𝑌i is the effective surface potential of the LC-bulk aqueous interface. 𝜆, 𝑌p and 𝑌i can be expressed as: 𝜀 𝜀𝑘 𝑇 𝜆 = √ 0 B2 (6.5) 2𝑒𝑐 𝑁A𝐼 𝐷𝜁p𝑒𝑐 8tanh( ) 𝑘B𝑇 𝑌p = (6.6) 𝑅 2( )+1 𝐷𝜁 𝑒 1+ 1− 𝜆 tanh2 p 𝑐 √ 2 ( )𝑘B𝑇𝑅 (( )+1) 𝜆 𝐷𝜁 𝑒 𝑌i = 4tanh ( i 𝑐) (6.7) 𝑘B𝑇 where 𝑁A = 6.02 × 10 23 is the Avogadro constant, 𝐼 is ionic strength of LC, 𝐷 is a constant, 𝜁p and 𝜁i are zeta potentials (Figure S6.1) at the microdroplet surface and LC- aqueous interface, respectively. In the absence of cationic surfactant DTAB, 𝐴 = 3/4 , 𝐾 = 7 pN, 𝜀0 = 8.854 × 10−12 C V-1 m-1, 𝜀 = (𝜀⊥ + 𝜀∥)/2 = 12.1, 𝑘B = 1.38 × 10 −23 J K-1, 𝑇 = 293.15 K, 𝑒𝑐 = 1.602 × 10 −19 C , 𝐼 = 1.8 × 10−5 mol m-3, 𝐷 is between 1 to 8[49], 𝜁p = −66.25 mV, 𝜁i = −43.18 mV, and 𝜆 = 0.89 m. For 𝑅 = 1.5 m, droplets at ℎ = 2 m, we calculate 𝐹e = −2.32 pN, 𝐹edl = −2.23 pN, and thus the net force 𝐹net = 𝐹e + 𝐹edl = −4.55 pN, which indicates repulsion from LC interface (Figure 6.2a,b). In contrast, we calculated that addition of 10 mM DTAB alters both the alignment of LC to homeotropic and the electrical double layer to generate attractions. Specifically, for this case we used 𝐴 = 1/2, 𝜀 = 𝜀∥ = 19, 𝜁i = −57.93 mV, and 𝜆 = 1.12 m. For 𝑅 = 1.5 m at ℎ = 2 m, 𝐹e = −1.54 pN, 𝐹edl = +4.36 pN, and 𝐹net = 𝐹e + 𝐹edl = +2.81 pN results in microdroplet release (Figure 6.2a,c). We note 𝐹net ~ 10 10𝐹g as described above, so gravity is negligible for droplets near the E7-bulk 176 aqueous phase interface. 177 Estimation of elastic constant for E7 We found that monomers can diffuse into E7, which decreases the elastic constant. To estimate the elastic constant, E7 was equilibrated with 10 wt% NIPAm solution at 30 °C for 5 hours, and then we measured the nematic to isotropic phase transition temperature of the E7 phase to be 50 °C, 10 °C lower than pure E7. Elastic constant (K = K11) of pure E7 at 30 °C was reported to be 9-10 pN; and 4-8 pN at 40 °C[50,51]. We estimated the elastic constant as 𝐾 = 7 pN. A change in the value of K from 4 pN to 10 pN does not change the conclusions of our model. 178 Explanation of morphology of block copolymer assemblies Release of APS (20 wt% in microdroplet phase, microdroplet:E7 = 1:10 vol) initiates polymerization to form a PDMA47-b-PDAAmX diblock copolymer (Figure 6.3a-c). PDAAm, when used in combination with PDMA, defines an amphiphilic block copolymer. We expected the self-assembly of the diblock in an aqueous phase, with PDAAm in the core of the assembly and the morphology to vary as a function of the degree of polymerization (DP) of the DAAm block (Figure 6.3c). TEM results revealed that DMA:DAAm = 1:1.5 yields spherical assemblies (Figure 6.3f), whereas DMA:DAAm = 1:2.5 leads to spherical and worm-like (cylindrical) assemblies (Figure 6.3g), and DMA:DAAm = 1:4 leads to vesicles (Figure 6.3h). This can be explained by the so-called packing parameter 𝑝[35], given by: 𝑣o 𝑝 = (6.8) 𝑎e𝑙o where 𝑣o is the volume of the DAAm block, 𝑙o is the length of the DAAm block, and 𝑎e is the equilibrium area per polymer chain at the aggregate surface. For a spherical aggregate of radius 𝑟 with chain number 𝑛a, 4 𝜋𝑟3 4𝜋𝑟2 𝑛a = 3 = (6.9) 𝑣o 𝑎e 𝑣 𝑟 𝑣 𝑟 1 𝑟 This implies that 𝑜 = . That is 𝑝 = o = ( ) ( ). Because ≤ 1, 𝑝 has to 𝑎𝑒 3 𝑎e𝑙o 𝑙o 3 𝑙o be less than or equal to 1/3. Similar calculations can be applied to worm-like 1 1 1 (cylindrical) aggregation to obtain ≤ 𝑝 ≤ , and vesicle for 𝑝 ≥ . 3 2 2 Quantitatively, this theory is consistent with our observations that increasing PDMA block length (𝑎e) results in spherical aggregations (a small packing parameter). We also measured the intensity-average diameter of PDMA47 by DLS to be 9.5 ± 0.3 nm. Assuming it is singly dispersed, and in assembled structures, PDMA47 occupies the same volume, then 𝑎e is estimated to be π(9.5/2) 2 = 70.88 nm2. The intensity-average diameter of PDMA47-b-PDAAm71 is 59.5 ± 1.3 nm, therefore we estimate 𝑙o = 179 59.5−9.5 1 = 25 nm. The sphere to worm transition (𝑝 = ) for PDMA47 happens at DP of 2 3 PDAAm is 92[36]. We assume changing DP only affect 𝑣o in a proportional manner, 1 9.5 2 whereas 𝑙o is unchanged; thus 𝑣o = ∗ 25 ∗ π ( ) = 590 nm 3 for DP = 92, 𝑣 3 2 o = 71 590 ∗ = 455 nm3, 𝑝 = 0.26. Similarly, we estimate that for PDMA47-b-PDAAm118, 92 𝑝 = 0.42, and for PDMA47-b-PDAAm222, 𝑝 = 0.80, consistent with the theory above. Overall, The DMA to DAAm molar ratio leads to control over assembly morphology. 180 80 60 40 20 5 mM water SDS 0 5 mM 10 mM 20 mM 1 mg/ml 5 mM DTAB DTAB DTAB polyarginine 2+ Pb -20 -40 -60 Figure S6.1 Zeta potentials for aqueous-dispersed LC droplets used in the experiments reported in this paper. 181 zeta potentials () Figure S6.2 Snapshots of the release of microdroplets in a mini-well. Ink was loaded into microdroplets to permit visualization. Scale bar, 2 mm. 182 Figure S6.3 Permanent surface-confined polymeric film in the presence of cross- linker (0.05 wt% BIS relative to monomer). Film in a) hydrated state, b) semi-dried state and c) fully-dried state. 183 Figure S6.4 1H NMR (500 MHz, CDCl3) spectrum of PNIPAm synthesized with the setup in Figure 6.1g-k. 184 Figure S6.5 GPC trace of PNIPAm synthesized with the setup in Figure 6.1g-k. The molecular weight of PNIPAm was calculated to be Mn = 279, 200 g mol-1 with the polydispersity index, Ð = 1.4. 185 Figure S6.6 1H NMR (500 MHz, CHCl3-d6, δ) spectrum of PDMA47 macroCTA. 186 Figure S6.7 1H NMR (500 MHz, CHCl3-d6, δ) spectra of PDMA47-b-PDAAmX. The degree of polymerization of DAAm (x) increased from 71 to 222 with increase in the comonomer feed ratio of DAAm to DMA. Acknowledgements X. Wang and Dr. H. Sun equally contributed to this work. Support of this research from the Army Research Office through W911NF-15-1-0568 and W911NF- 17-1-0575 is acknowledged. 187 5. References *This chapter was prepared as a Research Article reporting original research in the journal Advanced Materials. X.W., H.S., Y.K.K., N.C.G. and N.L.A. developed the concept for the research. X.W. and Y.K.K. conducted experiments involving stimuli- responsive LC printheads and imaging. H.S. and D.B.W. synthesized the MacroCTAs. H.S. measured the DLS, GPC and NMR. M.T. measured the Zeta potentials. N.C.G. and N.L.A. supervised the research. All authors contributed to the writing of the manuscript. Adapted with permission from: X. Wang, H. Sun, Y.-K. Kim, D. B. Wright, M. Tsuei, N. C. Gianneschi, and N. L. Abbott, Stimuli-Responsive Liquid Crystal Printheads for Spatial and Temporal Control of Polymerization, Adv. Mater. 34, 2106535 (2022). Copyright 2022 Wiley-VCH GmbH. [1] M. P. Stevens, Polymer Chemistry: An Introduction., 3rd ed. (Oxford University Press, New York, 1999). [2] J. M. G. Cowie and V. Arrighi, Polymers: Chemistry and Physics of Modern Materials (CRC press, Boca Raton, 2007). [3] O. Prucker, M. Schimmel, G. Tovar, W. Knoll, and J. Rühe, Microstructuring of Molecularly Thin Polymer Layers by Photolithography, Adv. Mater. 10, 1073 (1998). [4] M. Hegde, V. Meenakshisundaram, N. Chartrain, S. Sekhar, D. Tafti, C. B. Williams, and T. E. Long, 3D Printing All-Aromatic Polyimides Using Mask-Projection Stereolithography: Processing the Nonprocessable, Adv. Mater. 29, 1701240 (2017). [5] C. C. Cook et al., Highly Tunable Thiol-Ene Photoresins for Volumetric Additive Manufacturing, Adv. Mater. 32, 2003376 (2020). [6] M. F. Montemor, Functional and Smart Coatings for Corrosion Protection: A Review of Recent Advances, Surf. Coatings Technol. 258, 17 (2014). 188 [7] J. Niu, D. J. Lunn, A. Pusuluri, J. I. Yoo, M. A. O’Malley, S. Mitragotri, H. T. Soh, and C. J. Hawker, Engineering Live Cell Surfaces with Functional Polymers via Cytocompatible Controlled Radical Polymerization, Nat. Chem. 9, 537 (2017). [8] H. Sun et al., Proapoptotic Peptide Brush Polymer Nanoparticles via Photoinitiated Polymerization-Induced Self-Assembly, Angew. Chemie - Int. Ed. 59, 19136 (2020). [9] E. Kumarasamy, I. M. Manning, L. B. Collins, O. Coronell, and F. A. Leibfarth, Ionic Fluorogels for Remediation of Per-and Polyfluorinated Alkyl Substances from Water, ACS Cent. Sci. 6, 487 (2020). [10] X. Li, M. Li, Q. Shou, L. Zhou, A. Ge, D. Pei, and C. Li, Liquid Metal Initiator of Ring-Opening Polymerization: Self-Capsulation into Thermal/Photomoldable Powder for Multifunctional Composites, Adv. Mater. 32, 2003553 (2020). [11] X. Chen, M. A. Dam, K. Ono, A. Mal, H. Shen, S. R. Nutt, K. Sheran, and F. Wudl, A Thermally Re-Mendable Cross-Linked Polymeric Material, Science 295, 1698 (2002). [12] J. F. Patrick, M. J. Robb, N. R. Sottos, J. S. Moore, and S. R. White, Polymers with Autonomous Life-Cycle Control, Nature 540, 363 (2016). [13] R. J. Young and P. A. Lovell, Introduction to Polymers (CRC press, Boca Raton, 2011). [14] J. C. Theriot, C.-H. Lim, H. Yang, M. D. Ryan, C. B. Musgrave, and G. M. Miyake, Organocatalyzed Atom Transfer Radical Polymerization Driven by Visible Light, Science 352, 1082 (2016). [15] K. Jung, N. Corrigan, M. Ciftci, J. Xu, S. E. Seo, C. J. Hawker, and C. Boyer, Designing with Light: Advanced 2D, 3D, and 4D Materials, Adv. Mater. 32, 1903850 (2020). 189 [16] H. J. Hageman, Photoinitiators for Free Radical Polymerization, Prog. Org. Coatings 13, 123 (1985). [17] S. R. White, N. R. Sottos, P. H. Geubelle, J. S. Moore, M. R. Kessler, S. R. Sriram, E. N. Brown, and S. Viswanathan, Autonomic Healing of Polymer Composites, Nature 409, 794 (2001). [18] Y. Zhao, J. Fickert, K. Landfester, and D. Crespy, Encapsulation of Self-Healing Agents in Polymer Nanocapsules, Small 8, 2954 (2012). [19] L.-P. Lv, Y. Zhao, N. Vilbrandt, M. Gallei, A. Vimalanandan, M. Rohwerder, K. Landfester, and D. Crespy, Redox Responsive Release of Hydrophobic Self-Healing Agents from Polyaniline Capsules, J. Am. Chem. Soc. 135, 14198 (2013). [20] A. Sydney Gladman, E. A. Matsumoto, R. G. Nuzzo, L. Mahadevan, and J. A. Lewis, Biomimetic 4D Printing, Nat. Mater. 15, 413 (2016). [21] D. A. Walker, J. L. Hedrick, and C. A. Mirkin, Rapid, Large-Volume, Thermally Controlled 3D Printing Using a Mobile Liquid Interface, Science 366, 360 (2019). [22] S. B. Chernyshuk and B. I. Lev, Theory of Elastic Interaction of Colloidal Particles in Nematic Liquid Crystals near One Wall and in the Nematic Cell, Phys. Rev. E 84, 011707 (2011). [23] O. P. Pishnyak, S. Tang, J. R. Kelly, S. V. Shiyanovskii, and O. D. Lavrentovich, Levitation, Lift, and Bidirectional Motion of Colloidal Particles in an Electrically Driven Nematic Liquid Crystal, Phys. Rev. Lett. 99, 127802 (2007). [24] Y.-K. Kim, X. Wang, P. Mondkar, E. Bukusoglu, and N. L. Abbott, Self- Reporting and Self-Regulating Liquid Crystals, Nature 557, 539 (2018). [25] X. Wang, Y. Zhou, Y.-K. Kim, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Thermally Reconfigurable Janus Droplets with Nematic Liquid Crystalline and Isotropic Perfluorocarbon Oil Compartments, Soft Matter 15, 2580 (2019). 190 [26] P. C. Hiemenz and R. Rajagopalan, Principles of Colloid and Surface Chemistry (CRC press, Boca Raton, 1997). [27] B. Mustin and B. Stoeber, Single Layer Deposition of Polystyrene Particles onto Planar Polydimethylsiloxane Substrates, Langmuir 32, 88 (2016). [28] Y.-Y. Luk, K.-L. Yang, K. Cadwell, and N. L. Abbott, Deciphering the Interactions between Liquid Crystals and Chemically Functionalized Surfaces: Role of Hydrogen Bonding on Orientations of Liquid Crystals, Surf. Sci. 570, 43 (2004). [29] M. Heskins and J. E. Guillet, Solution Properties of Poly(N- Isopropylacrylamide), J. Macromol. Sci. Part A - Chem. 2, 1441 (1968). [30] C. Wu and X. Wang, Globule-to-Coil Transition of a Single Homopolymer Chain in Solution, Phys. Rev. Lett. 80, 4092 (1998). [31] J. F. Lutz, Ö. Akdemir, and A. Hoth, Point by Point Comparison of Two Thermosensitive Polymers Exhibiting a Similar LCST: Is the Age of Poly(NIPAM) Over?, J. Am. Chem. Soc. 128, 13046 (2006). [32] P. Alexandridis and B. Lindman, Amphiphilic Block Copolymers: Self-Assembly and Applications (Elsevier, Amsterdam, 2000). [33] J. Yeow and C. Boyer, Photoinitiated Polymerization-Induced Self-Assembly (Photo-PISA): New Insights and Opportunities, Adv. Sci. 4, 1700137 (2017). [34] M. Semsarilar, E. R. Jones, A. Blanazs, and S. P. Armes, Efficient Synthesis of Sterically-Stabilized Nano-Objects via RAFT Dispersion Polymerization of Benzyl Methacrylate in Alcoholic Media, Adv. Mater. 24, 3378 (2012). [35] J. Rieger, Guidelines for the Synthesis of Block Copolymer Particles of Various Morphologies by RAFT Dispersion Polymerization, Macromol. Rapid Commun. 36, 1458 (2015). [36] S. J. Byard, M. Williams, B. E. McKenzie, A. Blanazs, and S. P. Armes, Preparation and Cross-Linking of All-Acrylamide Diblock Copolymer Nano-Objects 191 via Polymerization-Induced Self-Assembly in Aqueous Solution, Macromolecules 50, 1482 (2017). [37] F. D’Agosto, J. Rieger, and M. Lansalot, RAFT-Mediated Polymerization- Induced Self-Assembly, Angew. Chemie - Int. Ed. 59, 8368 (2020). [38] C. A. Rogers, Intelligent Materials, Sci. Am. 273, 154 (1995). [39] M. Shahinpoor and H.-J. Schneider, editors , Intelligent Materials (RSC Publishing, Cambridge, U.K., 2008). [40] M. M. Schwartz, editor , Smart Materials (CRC Press, Boca Raton, 2009). [41] Z. Wang, J. Wang, J. Ayarza, T. Steeves, Z. Hu, S. Manna, and A. P. Esser‐ Kahn, Bio-Inspired Mechanically Adaptive Materials through Vibration-Induced Crosslinking, Nat. Mater. 20, 869 (2021). [42] E. T. Mwema, Fredrick Madaraka Akinlabi, Fused Deposition Modeling: Strategies for Quality Enhancement (Springer Nature, 2020). [43] J. Loudet, P. Barois, and P. Poulin, Colloidal Ordering from Phase Separation in a Liquid- Crystalline Continuous Phase, Nature 407, 611 (2000). [44] P. G. de Gennes and J. Prost, The Physics of Liquid Crystals (Clarendon Press, 1993). [45] J. M. Brake, M. K. Daschner, Y.-Y. Luk, and N. L. Abbott, Biomolecular Interactions at Phospholipid-Decorated Surfaces of Liquid Crystals, Science 302, 2094 (2003). [46] X. Wang, R. Zhang, A. Mozaffari, J. J. de Pablo, and N. L. Abbott, Active Motion of Multiphase Oil Droplets: Emergent Dynamics of Squirmers with Evolving Internal Structure, Soft Matter 17, 2985 (2021). [46] J. T. Lai, D. Filla, and R. Shea, Functional Polymers from Novel Carboxyl- Terminated Trithiocarbonates as Highly Efficient RAFT Agents, Macromolecules 35, 6754 (2002). 192 [47] K. Ponnusamy, R. P. Babu, and R. Dhamodharan, Synthesis of Block and Graft Copolymers of Styrene by Raft Polymerization, Using Dodecyl-Based Trithiocarbonates as Initiators and Chain Transfer Agents, J. Polym. Sci. Part A Polym. Chem. 51, 1066 (2013). [48] P. S. Drzaic, Liquid Crystal Dispersions (World Scientific, Singapore, 1995). [49] M. A. Brown, Z. Abbas, A. Kleibert, R. G. Green, A. Goel, S. May, and T. M. Squires, Determination of Surface Potential and Electrical Double-Layer Structure at the Aqueous Electrolyte-Nanoparticle Interface, Phys. Rev. X 6, 011007 (2016). [50] E. P. Raynes, R. J. A. Tough, and K. A. Davies, Voltage Dependence of the Capacitance of a Twisted Nematic Liquid Crystal Layer, Mol. Cryst. Liq. Cryst. 56, 63 (1979). [51] H. Hakemi, E. F. Jagodzinski, and D. B. Dupré, Temperature Dependence of the Anisotropy of Turbidity and Elastic Constants of Nematic Liquid Crystal Mixture E7, Mol. Cryst. Liq. Cryst. 91, 129 (1983). 193 Chapter 7 Self-Timed and Spatially-Targeted Delivery of Chemical Cargo by Chemotactic Self-Propelled Droplets 1. Introduction Biological systems make extensive use of active processes[1] in which chemical fuels (e.g., adenosine triphosphate, ATP) are used to drive cellular motility[2,3], such as leukocyte migration[4,5], and embryonic development[6]. During these active processes, chemical signals often polarize the direction of motion (chemotaxis) to spatially target a desired outcome[4,6]. Inspired by this strategy, here we report the design of a synthetic soft matter system that recapitulates key aspects of such behaviors. A liquid crystalline (LC) oil droplet that undergoes biased ballistic motion (chemotaxis) in an aqueous solution is used as a carrier for a chemical initiator of polymerization. A sustained circulatory flow causes aqueous microdroplet inclusions (microcargo) carrying initiator to coarsen and cluster within the elastic environment of the LC such that hydrodynamic inertial forces grow sufficiently to permit the escape of the microcargo through the outer interface of the carrier droplet. This microcargo reorganization inside the carrier droplet delays the release of microcargo from the onset of motion, which happens after the carrier LC droplet arrives at a targeted location (the end of the chemical gradient); this event triggers polymerization at the targeted location (Figure 7.1a). To introduce advective flow that leads to a self-propelled motion of a carrier droplet, the carrier droplet must have the ability to convert chemical potential in the system into mechanical work (i.e., functions as active matter)[7–11]. One approach is to utilize the Marangoni effect (an interfacial flow driven by a spatial gradient in interfacial energy), which has been shown previously to be sustained on the interface of an oil droplet immersed in a micellar solution by an inhomogeneous interfacial distribution of surfactant due to solubilization of oil[10–16]. This generates advective flows both inside and outside of the oil droplet, and subsequently drives droplet motion. Past studies have reported that Marangoni flows can also reorganize LC orientations 194 Figure 7.1 Design of chemotactic self-propelled delivery system and molecular constituents a, Schematic illustration of the system. The LC carrier droplet is a double emulsion. Initiator is loaded into the interior aqueous microdroplets (microcargo) that are stabilized by a surfactant within the LC oil (red dots). The LC carrier droplet can self-propel and move to a targeted location with a chemical gradient dictating its directional ballistic motion. Once the LC carrier droplet arrives at its destination, a combination of advection internal to the LC carrier droplet and LC elasticity leads to coarsening and clustering of the microcargo and autonomous release into the aqueous phase surrounding the LC droplet. Polymerization is subsequently triggered in the outer monomer solution. Molecular structures of b, E7 (LC), c, sodium dodecyl sulfate (SDS), d, ammonium persulfate (APS), e, N,N,N’,N’- tetramethylethylenediamine (TEMED), f, N-isopropylacrylamide (NIPAm) and g, N,N’- methylenebis(acrylamide) (BIS). within droplets, which in turn patterns the trajectory of the droplets[17,18]. In this manuscript, Marangoni flows are used to drive three processes: (i) the self-propelled swimming of carrier droplets, (ii) the reorganization of the microcargo (i.e., surfactant- stabilized aqueous inclusions) within the carrier droplets and (iii) the ultimate release of microcargo. Specifically, we demonstrate that an interplay between the effects of the 195 Marangoni flow and the elasticity of the LC oil controls the sizes of microcargo clusters. Additionally, we show that the shear of flow generates inertial forces of microcargo clusters that overcome colloidal repulsive forces at the carrier droplet interface to mediate the release of clusters of microcargo, and thus to achieve targeted delivery. This strategy is distinct from past examples, in which direct contact[19] led to delivery of DNA, from active droplets, and passive processes, such as diffusion[20] or stimuli- driven strategies[21,22] (e.g., pH-responsive coatings that dissolve in targeted environments) were used to release of cargo from active particles. 2. Results and Discussion We illustrate our approach by delivering microcargo that contain a chemical initiator for polymerization. The approach differs from methods such as photopolymerization, which have exquisite spatial resolution but require advance knowledge of the location to be illuminated[23]. In contrast, we seek to develop an integrated system (adaptive material system) that can sense a chemical signal, migrate in response to the signal, hold the cargo during the time of motion and trigger polymerization at the source of the signal in a self-timed manner. The capability to autonomously seek out and create a polymeric structure at the source of a chemical signal has the potential to be useful for chasing chemical leaks, for creating microenvironments (e.g., for living cellular systems) at chemically-defined locations, or for broadly initiating desired functions at locations that are not accessible to external illumination (e.g., internal regions of complex media that strongly absorb or scatter light). Overall, our results provide proof of concept for a spatially and temporally self- controlled delivery strategy (via a combination of dissipative and interfacial processes) that can be applied to the targeted delivery of drugs, dyes, anti-bacterial agents and/or nutrients. We prepared millimeter-sized LC carrier droplets as water/oil/water double emulsions (Figure 7.1a). A chemical initiator for free radical polymerization (ammonium persulfate (APS), Figure 7.1d, 10% wt/wt) was hosted within aqueous inclusions (3.3 ± 3.5 m, called microcargo) that were stabilized by an anionic 196 surfactant (sodium dodecyl sulfate, SDS, Figure 7.1c) within the LC oil droplet. The LC oil used was a nematic phase called E7 (Figure 7.1b), a eutectic mixture of four mesogens with a LC temperature range from below room temperature to 60 °C. We selected this composition because it remains as a LC phase under all conditions reported in this paper. Within the LC carrier droplet, each aqueous inclusion was surrounded by a region of strained LC. The LC strain affects microdroplet inclusions in multiple ways[24,25]: (i) it elastically traps the inclusions inside the LC carrier droplets; (ii) it mediates long-range (100 m) attractive interactions between inclusions, which promotes clustering of inclusions, and (iii) it generates topological defects in the LC, which result in short-range (1 m) repulsive interactions that serve as a barrier to coalescence of inclusions. We quantify these effects later in this manuscript. We used a concentration gradient of SDS in aqueous solutions to define the direction of motion of the carrier droplets (chemotaxis[14,26,27]). The gradient was created by confining two aqueous phases in a sample chamber fabricated from two microscope slides (separated by 1-2 mm). One aqueous phase contained the double emulsion dispersed in aqueous 5 mM SDS and the other phase contained an aqueous micellar solution of SDS (1 M), monomer (20 wt%), crosslinker (cross-linker: monomer = 1:100) and 1 v% catalyst (Figure 7.1 and Extended Data Figure 7.1). In initial experiments, we added a dye to the aqueous micellar solution to establish that approximately 50 s after contact of the two phases, a gradient in SDS concentration extended across a distance of ~7 mm within the system (Extended Data Figure 7.1 shows the time-dependent chemical gradient). Because nanometer-size species (micelles) can only diffuse micrometers within tens of seconds, the results obtained using dye indicate that convective processes associated with the contact of the two phases play a key role in generating the SDS concentration gradient in the sample chamber. When imaging LC carrier droplets in the sample chamber, each droplet was observed to be opaque against a black background because light was scattered by the aqueous inclusions inside the LC (Figure 7.2). Macroscopically, we found that the dynamic behavior of each LC carrier droplet in the sample chamber could be categorized as exhibiting four stages (Figure 7.2). Prior to arrival of the chemical 197 Figure 7.2 Chemotactic droplet self-propels along a gradient and autonomously releases microcargo at target location to trigger polymerization a, Images of LC as a function of time. The first image (-0.2 s) shows the two aqueous phases, a carrier droplet in 5 mM SDS solution (left) and a micellar/monomer solution (right, 1 M SDS, 20 wt% monomers, monomer: cross-linker = 100:1, 1 v% catalyst), before contact. A ballistic movement of carrier droplet towards the region of high chemical concentration was observed (50–110 s). Next, the LC carrier droplet still exhibited some slow motion with coarsening and reorganization of microdroplet inclusions (microcargo) (110–170 s). Finally, the LC carrier droplet released the microcargo containing initiator and triggered polymerization in the outer monomer solution (170–280 s). The appearance of the carrier droplet changed from opaque to relatively transparent as microcargo were released. Scale bars, 2 mm. b, Schematic illustrations of flow fields of carrier droplets at different stages. Green arrows indicate fluid flow. White arrow denotes droplet motion. c, Displacement and velocity of the LC carrier droplet as a function of time and stages. Droplet velocity rapidly increases and then decreases for Stage I Directional Motion. gradient at the location of a LC carrier droplet (Figure 7.2a,b, -0.2–50 s), the droplet was effectively immobile (they are too large to exhibit measurable Brownian motion (Stage O, Initial Stationary State)). When a LC carrier droplet was 5 mm from the location where the two aqueous droplets initially contacted each other, we observed the droplet to begin to exhibit motion approximately 50 s after contact of the two phases (Figure 7.2a, 50 s, Extended Data Figure 7.1). We determined (see Extended Data Figure 7.1) that this time corresponded with the arrival of the SDS concentration gradient at the location of the LC carrier droplet. The LC carrier droplet then moved 198 ballistically towards the region of high SDS concentration, consistent with the predicted effects of Marangoni stresses generated at the interface of the droplet (Stage I, Directional Motion). During Stage I, an axisymmetric circular flow internal to the droplet was observed (Figure 7.2b, Extended Data Figure 7.2), similar to previously reported flow fields[16,17]. We note elastic force from LC ordering interplays with flow on the length of tens of micrometer (see scaling analysis below). For LC droplets of tens of micrometer, the interplay of the two forces changes droplet trajectory, which is not observed in the large carrier droplet (1 mm) here, but rather leading to a reorganization inside of the carrier droplet. The self-propelled ballistic motion persisted for almost a minute (50–110s) with the velocity of the droplet rapidly increasing to 800 m/s and then decreasing to 50–0 m/s (Figure 7.2c). Overall, during Stage I, the droplet in Figure 7.2 moved approximately 10 mm. The next two stages of behavior (Stage II, Reorganization (110–170 s) and Stage III, Release (170–280 s)) were observed after the translational motion of the LC carrier droplet slowed down or had largely ceased. While the center of mass of the LC carrier droplet did not move a long distance, advection within the LC carrier droplet persisted during both Stage II and III, with multiple circulating flows evident near the carrier droplet interface (Figure 7.2b, Extended Data Figure 7.2). During Stage II, which is elucidated in more detail below, we observed the aqueous microcargo to coalesce and cluster under the influence of the flow internal to the LC carrier droplet but the microcargo were not released from the carrier droplet. In contrast, Stage III was defined by the onset of release of the microcargo from the LC carrier droplets, and subsequent triggering of polymerization in the solution surrounding the droplets (170–280 s). After triggering the formation of cross-linked polymer, the LC carrier droplet exhibited a transparent appearance, consistent with the effects of a decrease in the number of microcargo within the droplet. All motion of the LC carrier droplet ceased droplet after the release of the microcargo (~10 min). 199 To understand the processes that governed the coalescence and clustering of microcargo during Stage II, we observed the behaviors of the microcargo using a microscope. When using LC carrier droplets, we observed coarsening (growth in size) of microcargo during Stage II (Figure 7.3a). The inset in Figure 7.3a shows two microcargo coalescing into one large droplet inside a LC carrier droplet with an internal circulating flow. Our observations suggest that the coalescence process plays an important role in the release of the microcargo. Specifically, at the onset of release (start of Stage III), the average diameter of the microcargo increased from 3.2 ± 3.5 m to 33 ± 30 m within the LC carrier droplets (Extended Data Figure 7.2). After release, the microcargo remaining in the LC had an average size of 17 ± 17 m indicating that the large microcargo are preferentially released from the LC carrier droplets. We also note that the coalescence of microcargo occurred only in the presence of flow (flow acted as “attraction”). In addition to the coalescence described above in LC carrier droplets, we observed clustering of microcargo. Because the high density of microcargo in LC carrier droplets led to scattering of light that prevented unambiguous interpretation, we performed microscopic observations by confining a thin film of LC containing microcargo (vortexed to disperse the microcargo) within a glass sample chamber with a thickness of ~100 m. The LC film was then contacted with an aqueous micellar solution (0.8 M SDS and 5 wt% monomer). The experiments performed in this thin film geometry were characterized by velocities that were lower than those measurements when using LC carrier droplets (thin film confinement lowered the flow velocity below 200 m/s, which we observed to lower the rates of coalescence of microcago and generate polymer with a morphology, see below, that differed from using LC carrier droplets), but the flow fields in thin films were qualitatively similar to those observed in droplets during Stages II and III (multiple circulatory flows near the LC-aqueous 200 interface; Figure 7.3b and Extended Data Figure 7.3). In the absence of flow (no aqueous SDS phase added to the optical cell), microcargo (3.2 ± 3.5 m diameter) Figure 7.3 Stage II: Clustering and coalescence of microcargo inside the LC phase with and without flow. a, Coalescence of microcargo in LC carrier droplet in the presence of a flow circulating within the LC carrier droplet. b, Schematic illustrations of experiments in thin film geometry (left), with the microcargo depicted to form clusters over time inside the LC phase (middle 201 and right). c and d, Micrographs of microcargo in the absence of flow. The formation of large clusters (denoted by red arrows) with diameters of 200–500 m (due to LC elasticity) is evident. e-h, Micrographs with different magnifications of microcargo in LC in the a, Coalescence of microcargo in LC carrier droplet in the presence of a flow circulating within the LC carrier droplet. b, Schematic illustrations of experiments in thin film geometry (left), with the microcargo depicted to form clusters over time inside the LC phase (middle and right). c and d, Micrographs of microcargo in the absence of flow. The formation of large clusters (denoted by red arrows) with diameters of 200–500 m (due to LC elasticity) is evident. e-h, Micrographs with different magnifications of microcargo in LC in the presence of interfacially driven flows. e and g, A strong interfacial flow can disrupt clusters more effectively than f and h, a weak flow. Additionally, large microcargo (10–30 m, in red circles) in h favor a dispersed state as compared to small microcargo (1–5 m, on the right of the image). dispersed initially by vortexing were observed to spontaneously formed clusters in the LC within 10 min (Figure 7.3c,d), with a typical cluster size was 200–500 m. We interpret the origin of the clustering process to be the LC elastic energy associated with the straining of the LC around the microcargo, as described in past studies of colloidal assembly processes in LCs[28]. We note that the microcargo did not coalesce within these clusters over at least one hour, consistent with a short range repulsive interaction mediated by topological defects in the LC[24,29]. Next, we introduced an aqueous micellar/monomer solution (0.8 M SDS and 5 wt% monomer) into the optical sample cell (Figure 7.3, Extended Data Figure 7.3). Upon contact with the LC phase, we observed advective flows to be generated (Extended Data Figure 7.3), which resulted in the release of initiator-loaded microcargo into the contacting aqueous phase and the formation of polymer near the LC interface (Extended Data Figure 7.3). We measured the LC velocity by tracking the displacement of small microcargo near the clusters. When the initial flow velocity was high (e.g., 120 m/s), the clusters were disrupted by the flow, resulting in a cluster size of approximately 50 m (Figure 7.3e,g). In contrast, when the characteristic velocity of the LC decreased to 30 m/s (Figure 7.3f,h, due to formation of polymer near the interface after 4 mins), large clusters (200–500 m) of a size comparable to that observed in the absence of flow were found to form. Figure 7.3h also reveals that large microcargo (left of the image, red circles) are less likely to cluster as compared to small 202 microcargo (red arrow). Overall, these observations suggest that the microcargo cluster size is set by a competition between LC elasticity (promoting formation of clusters) and shear stresses (promoting disruption of clusters). As discussed above, the LC elasticity mediates long-range attraction and short- range repulsion. In contrast, flow can bring the microcargo together to promote coalescence (compressional fluid element that leads to the increase of local density of dispersions), or the tensile viscous stresses from flow may pull the clusters apart (extension that involves stretching along streamlines)[30]. The latter establish a microstructural distribution of the clusters. The long-range clustering reported above can be interpreted within a physical framework of competition between the tensile viscous (driving disassembly) and elastic stresses (driving assembly). We evaluated this competition using the Ericksen number 𝐸𝑟 = 𝜂𝑣𝑟/𝐾 , where 𝜂 is a characteristic viscosity of the LC (80 mPa s), 𝑣 is flow velocity (10–1000 m/s), 𝑟 is the size of the microcargo and 𝐾 is the Frank-Oseen elastic constant of the LC in the one-constant approximation (7 pN, Methods). For 𝑣 = 30 m/s and 𝑟 = 1 m, we calculated 𝐸𝑟 = 0.3 <1, indicating that elastic forces dominate (assembly); whereas for high fluid velocities (𝑣 = 300 m/s and 𝑟 = 1 m, 𝐸𝑟 = 3 >1) or large microcargo (𝑟 = 20 m and 𝑣 = 30 m/s, 𝐸𝑟 = 6 >1), viscous forces dominate the organization of microcargo (disassembly). This result is consistent with our observation that the microcargo form large clusters in the LC carrier films as the flow (droplet motion) slows. It also indicates that small microcargo will cluster before large microcargo at a given velocity. In terms of the flow-driven coalescence that happened when microcargo close to each other, we compared the magnitude of short-range colloidal repulsions (elastic and elactrical double layer forces) to the compressional viscous interaction. We evaluated the contribution of the LC elastic energy to the repulsive force acting between 𝑟 4 two microcargo as (Methods)[31,32]: 𝐹e,ss = −12𝜋𝐾𝜁 2 ( ) , where 𝜁 = 2.04 is a 𝑑+2𝑟 203 constant, K ~7 pN is the elastic constant of the LC, 𝑟 is the radius of microcargo, 𝑑 is the distance that separates the interfaces of the two microcargo. Additionally, an electrical double layer repulsion can be evaluated as (Methods)[24,33,34]: 𝐹edl,ss = 4𝜋𝜀0𝜀 𝑟 2 𝑑 − ( 1 +𝑟2 𝑘) ( B 𝑇 ) 𝑌2𝑒−𝜆, where 𝑟 , 𝑟 are the radii of the two microcargo, 𝜀 is the 𝜆 𝑟1𝑟2 𝑒 p 1 2 0 𝑐 vacuum permittivity, 𝜀 is the relative permittivity of the LC, 𝜆 is Debye screening length, 𝑘B is the Boltzmann constant, 𝑇 is the temperature, 𝑒𝑐 is the elementary charge, 𝑌p is the effective surface potential of the cluster interface. For 𝑣 = 200 m/s, 𝑟 = 𝑟1 = 𝑟2 = 10 m and 𝑑 = 1 m, (viscous forces (𝜂𝑣𝑟) ~160 pN) > (elastic force ~56 pN) + (electrical double layer force ~48 pN); whereasfor 𝑣 = 20 m/s, (viscous force (𝜂𝑣𝑟) ~16 pN) < (elastic force ~56 pN) + (electrical double layer force ~48 pN), indicating that a large flow velocity can provide viscous forces that overcome colloidal repulsions. Overall, we conclude that the advection within the LC carrier droplet generated by the Marangoni effect not only drives the motion of the LC carrier droplet (Figure 7.2), but it also plays a central role in programming size and organization of the microcargo that leads to their release (as discussed below, Figures 7.3-7.5). We observed that microcargo (as clusters) were autonomously released from the LC carrier droplets during Stage III (Figures 7.2 and 7.4). The initiator (APS) within the microcargo triggered free radical polymerization of monomer in the aqueous solution surrounding the LC carrier droplets to form PNIPAm (polymer), which we characterized by attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR, Extended Data Figure 7.4). Under a bright field microscope, we observed that the polymer product formed after release of microcargo consisted of clusters of discrete polymeric particles (Figure 7.4a). Additional observations hint that one polymeric particle was generated by the release of one microcargo droplet. Clusters of small polymeric particles (Figure 7.4b, 6.6 ± 2.5 m) tended to form before clusters of large particles (Figure 7.4c,d, 53.6 ± 17.6 m), consistent with the observation in Figure 7.3h that microcargo with small size were observed to form clusters before large ones. 204 Also, the polymeric particles generated by release of initiator were in the form of clusters (of either small or large particles, Figure 7.4a-c), similar to the microcargo clusters observed under the influence of flow within the LC phase (Figure 7.3). This correspondence led us to hypothesize that clustering plays a role in the autonomous release of the microcargo by the LC carrier droplets. Figure 7.4 Polymeric microparticles by release of initiator-containing microcargo from LC carrier droplets during Stage III. a, Micrograph shows a trail of polymeric (PNIPAm) particles of various sizes generated by release of initiator from LC carrier droplet, formed 3 min after the contact of the two phases. b, Small polymeric particles tended to form before c, large polymeric particles, both exhibiting cluster structures. d, Scanning electron microscope (SEM) image of a single polymeric bead (dry). e, Statistics of the particle sizes formed at early and late stages of release of microcargo, respectively. We further explored the process leading to the release of the microcargo by using the planar interface of a ~100 m thick LC film in contact with a micellar solution (0.8 M SDS) (Figure 7.5). We note that we used a low monomer concentration (5% wt/wt) in these experiments to slow polymerization and thus enable observation of the release process over an extended duration. The initiator-to-monomer ratio was higher than that 205 used in the LC carrier droplet experiments, and we observed this difference to cause a Figure 7.5 Clusters of microcargo within LC film that gain sufficient momentum from the shear flow can overcome the repulsive barrier at the LC interface to escape. a, Micrograph showing that singly dispersed microcargo were reflected back from the LC interface of film, whereas b, a large cluster of microcargo was largely ejected from the LC phase via the LC interface. Red arrows denote locations of a single microcargo in a. Yellow arrow indicates location of release of the cluster in red circle in b. c-e, Schematic illustrations of the proposed mechanism. Green arrows indicate fluid flow. Black arrows indicate release of microcargo. c, Small clusters and single microcargo cannot overcome the repulsive barrier from the interface, while d, clusters of mic2ro0c6a rgo gain sufficient momentum in the shear flow to overcome the repulsion. e, The released surfactant from the microcargo can accelerate the advection by reinforcing the Marangoni effect, promoting further release. f, Cluster size versus shear rate that reveals a critical cluster size, above which clusters can be ejected out arrow indicates location of release of the cluster in red circle in b. c-e, Schematic illustrations of the proposed mechanism. Green arrows indicate fluid flow. Black arrows indicate release of microcargo. c, Small clusters and single microcargo cannot overcome the repulsive barrier from the interface, while d, clusters of microcargo gain sufficient momentum in the shear flow to overcome the repulsion. e, The released surfactant from the microcargo can accelerate the advection by reinforcing the Marangoni effect, promoting further release. f, Cluster size versus shear rate that reveals a critical cluster size, above which clusters can be ejected out of the LC phase. used in the LC carrier droplet experiments, and we observed this difference to cause a polymer film rather than polymeric particles to form at the LC interface (Extended Data Figure 7.3). Our microscopic observations with this experimental geometry revealed that small clusters and singly dispersed microcargo advected by the flow internal to the LC film were reflected back from the LC interface rather than being released (Figure 7.5a), whereas large clusters of microcargo tended to be ejected out through the LC interface (Figure 7.5b). These observations led us to propose that only large clusters of microcargo possessed sufficient inertia to escape the LC carrier droplets (Figure 7.5c): the high momentum being sufficient to overcome repulsive barriers at interface (discussed below), allowing clusters above a threshold size to be autonomously ejected from the LC phase (Figure 7.5d). Accompanying the initiation of the release process, we also observed the local flow velocity to increase in strength (Extended Data Figure 7.5), which appeared to promote the release majority of the cluster (Figure 7.5e). We interpret the transient increase of flow velocity to likely arise from reinforcement of the Marangoni stress generated by the transfer of surfactant from within microcargo onto the LC interface. The repulsive forces acting on the microcargo clusters near the aqueous interface of the LC carrier droplets are generated by elastic and electrical double layer interactions[24,25]. In brief, the strain of the LC around a microcargo cluster generates a repulsive elastic force between the cluster (approximated as a sphere) and outer interface of the LC carrier droplets (Methods), which can be evaluated as[24]: 𝐹e,si = 207 4 −𝐴𝜋𝐾𝜁2 𝑅 ( ) , where A is a constant that depends on the orientation of the LC at the ℎ+𝑅 outer interface of the LC carrier droplets (1/2 for homeotropic anchoring in this work), R is the characteristic size of the cluster, and h is the distance separating the cluster surface and outer interface of the LC carrier droplet. The electrical double layer interaction arises from the presence of interfacial charges (including those due to the anionic surfactant SDS in both the aqueous microcargo and the bulk aqueous micellar phase) on the clusters and outer interface of the LC carrier droplets, which can be 𝑅 𝑘 𝑇 2 ℎ− evaluated as[24]: 𝐹edl,si = −4𝜋𝜀0𝜀 ( B ) 𝑌p𝑌i𝑒 𝜆, where 𝑌i is the effective surface 𝜆 𝑒𝑐 potential of outer interface of the LC carrier droplet (Extended Data Figure 7.6 and Methods for calculations). Whereas past studies have shown that attractive colloidal forces (electrical double layer forces) can be designed to trigger release of aqueous inclusions from LC[24,25], below the self-regulated release mechanism reported in this paper is based on non-equilibrium (inertial) forces. Evaluation of the total colloidal repulsive force acting on the microcargo clusters (𝐹R = 𝐹e,si + 𝐹edl,si and Extended Data Figure 7.6) reveals that it scales approximately with the size of the cluster 𝐹R ~ 𝛼𝑅, where 𝛼 is a scalar, as electrical double layer forces dominate for large clusters. Using this result, we formulated a scaling argument to predict release of a cluster by balancing the stress arising from the colloidal forces (𝜎R) with the inertia stress associated with the cluster motion ( 𝜎I , Methods): 𝜎 2𝑅 ~ 𝐹R⁄𝑅c ~ 𝛼⁄𝑅c = 𝜎I ~ 𝜌LC(?̇?𝑅 2 c) , where 𝑅c is a critical cluster size at which release occurs, 𝜌LC is the average density of the LC, ?̇? is an average shear rate of the flow acting on the cluster. We measured the average shear rate ?̇? using particle tracking velocimetry (Extended Data Figure 7.7). This scaling argument leads to the prediction 𝛼 1 that clusters of the order 𝑅𝑐 ~ ( 32) and larger overcome the repulsive barrier. We 𝜌LC?̇? plotted the cluster sizes 𝑅 versus the average shear rate ?̇? (Figure 7.5f). Small clusters (blue dots) are contained inside the LC phase whereas large clusters are released into the aqueous phase, with the critical cluster size being proportional to the power of minus two thirds of the average shear ?̇?, consistent with the scaling prediction. Using data from Release Stage (III) and the scaling analysis, we go back to Directed Motion Stage (II) 208 2 when 𝑣 = − 800 m/s. We estimated that 𝛾 3 > 0.40 (Extended Data Figure 7.7), and thus the critical cluster size is > 18 m. Because the average size of microcargo of 3.2 m < 18 m, and clusters are not predicted to form under a high flow velocity, it is reasonable that no release was observed at the Directed Motion Stage. Overall, these results support the proposal that release occurs when the inertial forces acting on the microcargo cluster can overcome the repulsive colloidal forces that trapped the microcargo in the LC carrier droplets. Directed motion of active colloids guided by chemotaxis[14,26,27,35–37] or electric potential[38] has been used to find the shortest path through complex structures, such as a maze[14,26,35,38], and were used to perform logic functions by passing through binary branch channel[14,36]. The latter occurs because the droplets are sensitive to the “products” of motion (e.g., oil-filled micelles), and therefore tend to avoid their own trail (negative autochemotaxis). To show that the chemotactic behavior of the LC carrier droplets can be used to direct the droplets to navigate complex Figure 7.6 Chemotactic LC carrier droplet functioning in channels. Time-lapse images show that a LC carrier droplet (white) was guided into branch 3 of the channel by a chemical gradient of surfactant and monomers, and subsequently released the initiator microcargo inside the carrier and triggered polymerization to block the channel. The LC carrier droplet was initially in 5 mM SDS aqueous solution; a concentrated solution was injected into branch 3 to build up the chemical gradient; concentrated solution contained 1 M SDS, 30 wt% monomers, monomer:cross-linker = 100:1, 1 v% catalyst. White dashed line indicates the trajectory of the carrier droplet. Scale bar, 5 mm. 209 environments to trigger polymerization, we designed a millifluidic channel system with a main channel split into three branches (Figure 7.6). By generating a chemical gradient of surfactant and monomers in branch 3, the LC carrier droplet was preferentially guided into this branch. After 4–5 min, the droplet motion ceased, initiator microcargo was autonomously released, and subsequently polymerization was triggered in the monomer solution. The carrier droplet and the polymer formed from it blocked the channel branch (Figure 7.6). Overall, this result demonstrates that the chemotactic droplets can select their path of motion and deliver microcargo to targeted locations. 3. Conclusions In summary, our results demonstrate how active droplets can be used for autonomous delivery of reactive species (initiator) to remote locations (source of chemoattractant) driven by advective flow resulted from Marangoni stress. These findings reveal that dissipative processes and colloidal phenomena can be used jointly in the design of functional materials, and such a combination enables a time-dependent targeted release that does not need additional stimuli to function. Future efforts will seek to understand how the interplay between the dynamic flow and elasticity of LC shapes the clusters of microcargo, and to design systems that function in complex three- dimensional mazes. We note that pharmaceutical compounds[39], anti-bacterial agents[24] and essential nutrients[19,40] can also be loaded into aqueous microcargo, with which our system provides a basis for targeted delivery and autonomous release of these active chemicals. Additionally, external stimuli such as magnetic or electric fields[41], thermal signaling[24], interfacial enzymatic events[42] and bacterial shear[24] have been explored to modulate LC responses in the past studies, which can also be incorporated into the active delivery system to manifest functions. 210 4. Supporting Information METHODS Materials. Nematic liquid crystal E7 was purchased from HCCH (Jiangsu Hecheng Display Technology Co., Ltd). N-Isopropylacrylamide (NIPAm, 97%, monomer), N,N’- methylenebis(acrylamide) (BIS, cross-linker), ammonium persulfate (APS, initiator), N,N,N’,N’-tetramethylethylenediamine (TEMED, catalyst) and sodium dodecyl sulfate (SDS) were purchased from Sigma-Aldrich. The dye monomer methacryloxyethyl thiocarbamoyl rhodamine B was purchased from Polysciences. Fisher Finest Premium Grade glass slides was purchased from Fisher Scientific. Purification of water (18.2 MΩ cm resistivity at 25 °C) was performed using a Milli-Q water system (Millipore, Bedford, MA, USA). Preparation of w/o/w LC double emulsions containing aqueous initiator microcargo. First, surfactant SDS (5 mM) and initiator APS (10 wt%) were dissolved into MilliQ water. We measured the density of the APS-SDS aqueous phase to be 1.05 g cm-3, slightly higher than the density of E7 of 1.03 g cm-3. Next, the resulting solution was emulsified into E7 at 1:10 ratio by vortexing for 1 min at 3,000 rpm. The last step is adding the oil mixture into SDS solution (5 mM) at 0.5-1:10 ratio, and gently shaking the solution to form double emulsions. Preparation of micellar/monomer solution. First, surfactant SDS (1 M) was dissolved into MilliQ water by vortexing and heating to 60 °C for 2 hr to facilitate the process (SDS micellar solution). Second, monomers (10–30 wt%, NIPAm: BIS = 100:1 wt) were dissolved into the SDS micellar solution. Third, catalyst TEMED (1 v%) was added into the aqueous phase to make the final micellar solution. Chemotactic LC carrier droplet triggering polymerization experiment. Glass slides were first washed using ethanol and water. Optical cells were assembled from two 25⨉25 mm glass slides separated from each other by 1-2 mm-thick spacers. We subsequently injected the double emulsion solution from one end of the optical cell, and 211 the micellar solution from the other end of the cell to induce the contacting of the two solutions. The experiments were done at an elevated temperature at 35 °C using a Linkam hot stage. The behaviors of the double emulsion droplets were recorded using a Canon EOS Rebel T6i camera at 30 fps with a Canon EF-mount 105mm f/2.8 EX DG OS HSM Macro lens or under polarized light microscopy. The droplet positions were extracted using ImageJ and analyzed. Observation of microcargo coalescence and clustering in thin films. 15⨉12 mm optical cells were assembled from two glass slides separated from each other by ~100 m-thick spacers. SDS (5 mM) and initiator APS (10 wt%) were dissolved into MilliQ water. The resulting solution was emulsified into E7 at 1:10 ratio by vortexing for 1 min at 3,000 rpm. We subsequently injected the oil mixture from one end of the optical cell, and the micellar solution (0.8-1 M SDS, 5 wt% monomers, NIPAm: BIS = 100:1 wt, 1 v% TEMED) from the other end of the cell to induce the contacting of the two fluid phases. The experiments were done at an elevated temperature at 35 °C using a Linkam hot stage. The behaviors of the initiator microcargo were recorded under polarized light microscopy, and were processed and analyzed using ImageJ. Fabrication of millifluidic channel. Millifluidic channels of thickness 1.6 mm were printed on a laser cutter (Zing 40 watt; Epilog Laser) from a piece of acrylic sheet. Acrylic covers were also printed on the laser cutter. A set of acrylic cover and channel was bound together by dropping chloroform on the surfaces, and pressing the surfaces to contact until chloroform evaporated (~2 min). The channel was then glued onto a glass substrate using epoxy resin, and coated with poly(vinyl alcohol) (PVA) to make the channel hydrophilic. The droplet experiments were done at an elevated temperature at 40-45 °C on the hot stage. Zeta potential measurements. 0.1 l E7 was homogenized for 30 s into 10 ml water. 240 l of LC-in-water emulsion was then diluted in 10 ml of either water or 5 mM SDS. After allowing 30 min for the sample to equilibrate, zeta potentials on the aqueous side 212 of the LC-aqueous interface were measured using a Malvern Zetasizer Nano instrument. Data were averaged over five consecutive runs. Attenuated total reflectance Fourier transform infrared spectroscopy (ATR- FTIR)[43,44]. The Polymer pieces (PNIPAm) from the chemotactic LC droplet experiments were taken out and washed with water and ethanol, and dried under ambient condition for more than 5 days. NIPAm monomer (dry) and dried PNIPAm samples from the chemotactic LC carrier droplet experiments were ground and then compressed into disks before testing. Infrared spectroscopy analysis was done using a Bruker Vertex V80V vacuum FTIR system, equipped with an ATR accessory, in the 4000–600 cm−1 spectral range. Scanning electron microscope (SEM). For SEM characterization of polymer beads, the polymers were immersed in ethanol for 48 hr and dried under a stream of nitrogen to extract remaining LCs, water and unreacted monomers. Following the extraction, imaging of the morphology of the polymer networks was performed by Zeiss Gemini 500 Scanning Electron Microscope. Trapping and segregation of initiator microcargo. Initiator microdroplets (microcargo) in LCs are sequestered within the LC bulk phase by the elastic energies associated with the strain of orientational ordering of LCs around the microdroplets[45,46]. Assuming one elastic constant K, the elastic energy depends on droplet size R (KR, typical K ~10-11 N), which competes with surface anchoring energy (WR2, where W ~10-5 N m-1 is anchoring energy density)[47]. A droplet suspended in LC with size R > K/W ~1 m (that is WR2 > KR) generates elastic repulsion that prevents coalescence with other droplets. Estimation of elastic constant for E7. We found that monomers can diffuse into E7, which decreases the elastic constant. To estimate the elastic constant, E7 was equilibrated with 10 wt% NIPAm solution at 35 °C for 5 hr, and then we measured the 213 nematic to isotropic phase transition temperature of the E7 phase to be 50 °C, 10 °C lower than pure E7. Elastic constant (K = K11) of pure E7 at 30 °C was reported to be 9- 10 pN; and 4-8 pN at 40 °C[48,49]. We estimated the elastic constant as 𝐾 = 7 pN. A change in the value of K from 4 pN to 10 pN does not change the conclusions of our model. Calculation of elastic and electrical double layer interactions. We evaluated the elastic repulsion between the LC-bulk aqueous interface and microcargo clusters as[24,32,50]: 𝑅 4 𝐹 = −𝐴𝜋𝐾𝜁2e,si ( ) (1) ℎ+𝑅 where A is a constant that is dependent on anchoring (3/4 and 1/2 for planar, where mesogens are parallel to interface, and homeotropic anchoring, where mesogens are perpendicular to interface (our work), respectively), 𝜁 = 2.04 is a constant, K is a characteristic LC elastic constant, R (> K/W) is the radius of the microcargo cluster, h is the distance of the cluster surface from the LC-bulk aqueous interface. For 1 m microdroplet/cluster near the interface (h = 0) and homeotropic anchoring of E7 at water interface, 𝐹e,si is evaluated to be 6.5 × 10 −11 N, which is large compared to 4 gravitational forces 𝐹g ~ 8 × 10 −22 N (∆𝜌𝑔 ∙ 𝜋𝑅3, where ∆𝜌 ~ 2 × 10−5 kg m-3 is the 3 density difference between aqueous droplet and E7, 𝑔 ~ 9.81 m s-2 is the gravitational acceleration). Similarly, we evaluated the forces generated by the LC elasticity between two microcargo spheres as[29,31,32]: 𝑟 4 𝐹e,ss = −12𝜋𝐾𝜁 2 ( ) (2) 𝑑+2𝑟 where 𝑟 is the radius of microcargo, 𝑑 is the distance that separates the interfaces of the two microcargo. Other interactions, such as electrostatic attractions, can reinforce the trapping of microcargo[24]. Electrical double layer interactions exist between negative charges at the LC-bulk aqueous interface and the negatively charged SDS on the interface of microcargo, thus generating repulsions. We evaluated the forces generated by the 214 electrical double-layer interactions between the cluster sphere and the interface as[24,33,51]: 𝑅 𝑘 𝑇 2 ℎ 𝐹edl,si = −4𝜋𝜀0𝜀 ( B ) 𝑌p𝑌 − i𝑒 𝜆 (3) 𝜆 𝑒𝑐 where 𝜀0 is vacuum permittivity, 𝜀 is the relative permittivity of the LC (for E7 with homeotropic anchoring 𝜀∥ = 19; for planar anchoring 𝜀⊥ = 5.2)[52], 𝜆 is the Debye screening length, 𝑘B is the Boltzmann constant, 𝑇 is temperature, 𝑒𝑐 is the elementary charge, 𝑌p is the effective surface potential of the microcargo, 𝑌i is the effective surface potential of the LC-bulk aqueous interface, and the negative sign − means repulsion. 𝜆, 𝑌p and 𝑌i can be expressed as: 𝜀0𝜀𝑘B𝑇 𝜆 = √ 2 (4) 2𝑒𝑐 𝑁A𝐼 𝐷𝜁p𝑒𝑐 8tanh( ) 𝑘 𝑇 𝑌p = B (5) 𝑅 2( )+1 𝜆 𝐷𝜁p𝑒𝑐1+√1− 2 tanh 2( ) 𝑘 𝑅 B 𝑇 (( )+1) 𝜆 𝐷𝜁 𝑒 𝑌i = 4tanh ( i 𝑐) (6) 𝑘B𝑇 where 𝑁A = 6.02 × 10 23 is the Avogadro constant, 𝐼 is ionic strength of LC, 𝐷 is a constant, 𝜁p and 𝜁i are zeta potentials (Figure S1) at the microcargo surface and LC- aqueous interface, respectively. 1 In the presence of anionic surfactant SDS, 𝐴 = , 𝐾 = 7 pN, 𝜀0 = 8.854 × 10 −12 C 2 V-1 m-1, 𝜀 = 𝜀 = 19, 𝑘 = 1.38 × 10−23 J K-1∥ B , 𝑇 = 308.15 K, 𝑒𝑐 = 1.602 × 10−19 C, 𝐼 = 1.8 × 10−5 mol m-3, 𝐷 is between 1 to 8,[53] 𝜁p = −66.25 mV, 𝜁i = −66.25 mV, and 𝜆 = 0.89 m. For 𝑅 = 3 m, cluster at ℎ = 2 m, we calculate 𝐹e,si = −1.54 pN, 𝐹edl,si = −6.89 pN, and thus the net force 𝐹net = 𝐹e,si + 𝐹edl,si = −8.43 pN, which indicates repulsion from LC interface. We note 𝐹net ~ 10 10𝐹g as described above, so gravity is negligible for droplets near the E7-bulk aqueous phase interface. Similarly, we evaluated the forces generated by the electrical double-layer interactions between two microcargo spheres as[24,34]: 215 4𝜋𝜀0𝜀 𝑟1+𝑟2 𝑘B𝑇 2 𝑑 𝐹 = − ( ) ( ) 𝑌2 −edl,ss p 𝑒 𝜆 (7) 𝜆 𝑟1𝑟2 𝑒𝑐 𝑟 +𝑟 where 𝑟1 , 𝑟 1 2 2 are the radii of the two spheres, 𝑌p obtained with replacing 𝑅 in 𝑟1𝑟2 Equation 5, d is the distance separating the interfaces of the two spheres, and the negative sign − means repulsion. For 𝑟1 = 𝑟2 = 1.5 m, 𝑑 = 1 m, we calculate 𝐹edl,ss = −10.60 pN; and for 𝑟1 = 𝑟2 = 15 m, 𝑑 = 1 m, 𝐹edl,ss = −70.10 pN, approximately one order of magnitude larger compared to the small microcargo. Stress of inertia of LC acting on a cluster under advective flow. We evaluated the stress of inertia using the following dimension analysis: 𝑚 ∙ ∆𝑣 MLT−2 MT−2 𝜌 3 −2𝐿𝐶L T 𝜎I ~ = = = = 𝜌 L 2T−2 ~ 𝜌 2 2 𝐴𝑟𝑒𝑎 ∙ ∆𝑡 L2 L L 𝐿𝐶 𝐿𝐶 𝑣LC ~ 𝜌𝐿𝐶(?̇?𝑅) (8) where 𝑚 is the mass of the cluster, ∆𝑣 is the velocity increase of the cluster in the presence of flow, Area is a characteristic area associated with the size of the cluster, ∆𝑡 is the time span related to the motion of the cluster, 𝜌c is the average density of the cluster, 𝑣LC is the effective scaling velocity of the cluster that generated by shear, which scales with an average shear rate of the flow acting on the cluster ?̇? times the cluster size 𝑅. The estimation of ?̇? and 𝑅 is shown in Extended Data Figure 7.7. 216 Extended Data Figure 7.1 Chemical gradient generated by contact of two aqueous phases. a, Schematic illustration of contact of an aqueous solution (5 mM SDS) containing LC carrier droplets (double emulsion droplets) with a micellar solution (1 M SDS, 20 wt% monomer and catalyst). The LC carrier droplets migrate along the resulting gradient in concentration of SDS. b and c, Images of the development of the chemical gradient as revealed by addition of a dye 217 to the solution containing 1 M SDS, monomer and catalyst (the concentration of dye was <0.1% relative to monomer) b in the absence or c in the presence of LC carrier droplet. Scale bars, 2 mm. The carrier droplet, which is white, exhibited pink appearance because of light reflection. When the SDS gradient arrives at the initial location of the LC carrier droplet, as shown in c (14s), the droplet starts its self-propelled motion which, in turn, leads to convective rolls near the gradient. d, Concentration profiles of the chemical gradient, as characterized by measurement of the intensity of light absorbed by the sample shown in b, as a function of time, using Imagej software. The two aqueous phases contacted at 0 mm. Data were fitted using sigmoid or double sigmoid functions. 218 Extended Data Figure 7.2 Flow fields and measurements of microcargo sizes. a,c Micrographs and b,d corresponding schematic illustrations of LC carrier droplets at Stage I and II, respectively. The halo around the carrier droplet formed due to solubilization of LC oil by the micellar solution. a and b In Stage I, the axisymmetric flow field left one trail of oil- filled micelle, denoted by red arrows; whereas c and d in Stage II, multiple trails were observed, indicative of multiple circular flow. e-g, Micrographs of e, LC-microcargo mixture in the absence of flow, f at the onset of release inside a LC carrier droplet (under flow) and g after release in a carrier droplet (under flow). Red circles mark several microcargo. Microcargo in f,g were tracked in multiple frames to measure the sizes. 219 Extended Data Figure 7.3 Film of LC containing microcargo (microdroplets containing aqueous initiator of polymerization) shown in contact with a micellar solution (0.8 M SDS and 5 wt% monomer). a, Schematic illustration (showing circulating flows due to Marangoni stresses) and b, micrograph of the interface region just after 100 m-thick films of the two liquid phases contacted each other. Red circles indicate cirvulating flow. c, After 30 min, polymer formed at the interface (denoted by red arrow, invisible) due to the release of initiator-containing microcargo in the LC, which morphs the interface into a wavy shape. d, Image of a piece of polymer formed near the LC-aqueous interface in a thicker film (~1 mm), which was separated from the system. 220 Extended Data Figure 7.4 ATR-FTIR spectra of (a) NIPAm monomer and (b) PNIPAm synthesized from chemotactic self-propelled LC carrier droplets, respectively. Intensity of the bands at 3072 cm−1, which was assigned to the C-H stretching vibration in the unsaturated C=C moieties, and at 1620 cm−1, which was related to the vibration of C=C stretching, was largely reduced for PNIPAm. This suggests that the free radical polymerization of NIPAm monomers had occurred in the chemotactic LC carrier droplet experiments. The double peak at 3462/3282 cm−1 was attributed to the N–H stretching vibrations. Characteristic absorption bands for both NIPAm and PNIPAm are at 1657 cm−1 (C=O stretching of amide I), at 1547 cm−1 (N–H bending of amide II), and the bands at 1387/1367 cm−1 were assigned to vibration of isopropyl groups (-CH(CH3)2). 221 Extended Data Figure 7.5 Flow velocity increased by two folds at the time of release. Control experiments were performed in the absence of initiator, monomer or catalyst. Data obtained from thin film experiments with LC-microcargo (5 mM SDS, 10 wt% APS if added) mixture contacting with 0.8 M SDS and, if added, 5 wt% monomer and cross-linker, monomer: cross-linker = 100:1, 1 v% catalyst in the aqueous phases. 7 samples under each condition, respectively. 222 Fedl 200 Fe net total force 100 R (m) 50 100 150 0 -100 -200 -300 Extended Data Figure 7.6 Colloidal forces as a function of cluster size. Colloidal forces act as repulsions between the LC-aqueous interface and microcargo. Net repulsion 𝐹𝑅 is sum of elastic force 𝐹e,si and electrical double layer force 𝐹edl,si. For clusters larger than 20 m, net repulsion is approximately proportional to cluster size. 223 F , F , F (pN) R e edl Extended Data Figure 7.7 Evaluation of stress of inertia. a, Micrograph of clusters near LC-aqueous interface under circular flow in film geometry (left). Average shear rate ?̇? , was measured by tracking both small microcargo inside the LC phase (denoted by green arrows) and tracer particles near the LC- aqueous interface (denoted by blue arrow) and calculating their velocity (𝑉) gradient normal to the flow direction, as shown in the schematic illustration (right): ?̇? ~ 𝑉max/𝐻. Typical 𝐻 is 100-200 m for 𝑉max < 200 m/s, with high flow velocity tended to create large 𝐻. For 𝑉max = 800 m/s, we estimated 𝐻 > 200 m, 2 − so ?̇? 3 > 0.4. Inside a carrier droplet, when 𝑣 ~ 800 m/s (Stage I Directed Motion), 𝐻 was comparable with droplet size (1 mm) because of the axisymmetric flow field. Cluster size 𝑅 was evaluated using the average of height and width of a cluster. Yellow arrow indicates a release event of a cluster, whereas purple arrows point at clusters that did not cause any release. Green arrows indicate Marangoni flow. b, Micrographs of another example that (i) two small clusters caused no release; and (ii) when they were close enough to form a large cluster, multiple release events were observed. The inset focuses on the site of release. 5. References *This chapter was prepared as a Research Article reporting original research close to submission. X.W., Y.Y., N.C.G. and N.L.A. developed the concept for the research. 224 X.W. conducted experiments and imaging, and analysed data on cluster formation and microdroplet coalescence. X.W. and S.H. analysed data on releasing of cargo. S.R. contributed to the characterizations. N.C.G. and N.L.A. supervised the research. All authors contributed to the writing of the manuscript. Adapted with permission from: X. Wang, Y. Yang, S. Roh, S. Hormozi, N. C. Gianneschi and N. L. Abbott, Self-Timed and Spatially-Targeted Delivery of Chemical Cargo by Chemotactic Self-Propelled Droplets. Close to submission. [1] E. P. Solomon, L. R. Berg, and D. W. Martin, Biology, Ninth Edition (Boorks/Cole, Cengage Learning, 2011). [2] Y. Rikitake and Y. Takai, Directional Cell Migration. Regulation by Small G Proteins, Nectin-like Molecule-5, and Afadin., Int. Rev. Cell Mol. Biol. 287, 97 (2011). [3] C.-M. Lo, H.-B. Wang, M. Dembo, and Y.-L. Wang, Cell Movement Is Guided by the Rigidity of the Substrate, Biophys. J. 79, 144 (2000). [4] K. Ley, C. Laudanna, M. I. Cybulsky, and S. Nourshargh, Getting to the Site of Inflammation: The Leukocyte Adhesion Cascade Updated, Nat. Rev. Immunol. 7, 678 (2007). [5] W. He et al., Circadian Expression of Migratory Factors Establishes Lineage- Specific Signatures That Guide the Homing of Leukocyte Subsets to Tissues, Immunity 49, 1175 (2018). [6] B. Bénazéraf, P. Francois, R. E. Baker, N. Denans, C. D. Little, and O. Pourquié, A Random Cell Motility Gradient Downstream of FGF Controls Elongation of an Amniote Embryo, Nature 466, 248 (2010). [7] T. Sanchez, D. T. N. Chen, S. J. Decamp, M. Heymann, and Z. Dogic, Spontaneous Motion in Hierarchically Assembled Active Matter, Nature 491, 431 (2012). [8] A. Bricard, J. B. Caussin, N. Desreumaux, O. Dauchot, and D. Bartolo, Emergence of Macroscopic Directed Motion in Populations of Motile Colloids, 225 Nature 503, 95 (2013). [9] M. C. Marchetti, J. F. Joanny, S. Ramaswamy, T. B. Liverpool, J. Prost, M. Rao, and R. A. Simha, Hydrodynamics of Soft Active Matter, Rev. Mod. Phys. 85, 1143 (2013). [10] M. M. Hanczyc, T. Toyota, T. Ikegami, N. Packard, and T. Sugawara, Fatty Acid Chemistry at the Oil-Water Interface: Self-Propelled Oil Droplets, J. Am. Chem. Soc. 129, 9386 (2007). [11] T. Toyota, N. Maru, M. M. Hanczyc, T. Ikegami, and T. Sugawara, Self- Propelled Oil Droplets Consuming “Fuel” Surfactant, J. Am. Chem. Soc. 131, 5012 (2009). [12] Z. Izri, M. N. van der Linden, S. Michelin, and O. Dauchot, Self-Propulsion of Pure Water Droplets by Spontaneous Marangoni-Stress-Driven Motion, Phys. Rev. Lett. 113, 248302 (2014). [13] S. Herminghaus, C. C. Maass, C. Krüger, S. Thutupalli, L. Goehring, and C. Bahr, Interfacial Mechanisms in Active Emulsions, Soft Matter 10, 7008 (2014). [14] C. Jin, C. Krüger, and C. C. Maass, Chemotaxis and Autochemotaxis of Self- Propelling Droplet Swimmers, Proc. Natl. Acad. Sci. U. S. A. 114, 5089 (2017). [15] X. Wang, R. Zhang, A. Mozaffari, J. J. de Pablo, and N. L. Abbott, Active Motion of Multiphase Oil Droplets: Emergent Dynamics of Squirmers with Evolving Internal Structure, Soft Matter 17, 2985 (2021). [16] C. H. Meredith, P. G. Moerman, J. Groenewold, Y. J. Chiu, W. K. Kegel, A. van Blaaderen, and L. D. Zarzar, Predator–Prey Interactions between Droplets Driven by Non-Reciprocal Oil Exchange, Nat. Chem. 12, 1136 (2020). [17] C. Krüger, G. Klös, C. Bahr, and C. C. Maass, Curling Liquid Crystal Microswimmers: A Cascade of Spontaneous Symmetry Breaking, Phys. Rev. Lett. 117, 048003 (2016). [18] B. V. Hokmabad, K. A. Baldwin, C. Krüger, C. Bahr, and C. C. Maass, Topological Stabilization and Dynamics of Self-Propelling Nematic Shells, Phys. Rev. Lett. 123, 178003 (2019). [19] M. Li, M. Brinkmann, I. Pagonabarraga, R. Seemann, and J.-B. Fleury, 226 Spatiotemporal Control of Cargo Delivery Performed by Programmable Self- Propelled Janus Droplets, Commun. Phys. 1, 23 (2018). [20] M. Wan et al., Platelet-Derived Porous Nanomotor for Thrombus Therapy, Sci. Adv. 6, eaaz9014 (2020). [21] B. Esteban-Fernández De Ávila, D. E. Ramírez-Herrera, S. Campuzano, P. Angsantikul, L. Zhang, and J. Wang, Nanomotor-Enabled PH-Responsive Intracellular Delivery of Caspase-3: Toward Rapid Cell Apoptosis, ACS Nano 11, 5367 (2017). [22] M. Hansen-Bruhn, B. E. F. de Ávila, M. Beltrán-Gastélum, J. Zhao, D. E. Ramírez-Herrera, P. Angsantikul, K. Vesterager Gothelf, L. Zhang, and J. Wang, Active Intracellular Delivery of a Cas9/SgRNA Complex Using Ultrasound- Propelled Nanomotors, Angew. Chemie Int. Ed. 57, 2657 (2018). [23] P. J. Bártolo, Stereolithography: Materials, Processes and Applications (Springer, 2011). [24] Y.-K. Kim, X. Wang, P. Mondkar, E. Bukusoglu, and N. L. Abbott, Self- Reporting and Self-Regulating Liquid Crystals, Nature 557, 539 (2018). [25] X. Wang, H. Sun, Y.-K. Kim, D. B. Wright, M. Tsuei, N. C. Gianneschi, and N. L. Abbott, Stimuli-Responsive Liquid Crystal Printheads for Spatial and Temporal Control of Polymerization, Adv. Mater. 34, 2106535 (2022). [26] I. Lagzi, S. Soh, P. J. Wesson, K. P. Browne, and B. A. Grzybowski, Maze Solving by Chemotactic Droplets, J. Am. Chem. Soc. 132, 1198 (2010). [27] S. Zhang, C. Contini, J. W. Hindley, G. Bolognesi, Y. Elani, and O. Ces, Engineering Motile Aqueous Phase-Separated Droplets via Liposome Stabilisation, Nat. Commun. 12, 1673 (2021). [28] I. Muševič, M. Škarabot, U. Tkalec, M. Ravnik, and S. Žumer, Two-Dimensional Nematic Colloidal Crystals Self-Assembled by Topological Defects, Science (80- . ). 313, 954 (2006). [29] P. Poulin and D. A. Weitz, Inverted and Multiple Nematic Emulsions, Phys. Rev. E 57, 626 (1998). [30] M. Firouznia, B. Metzger, G. Ovarlez, and S. Hormozi, The Interaction of Two 227 Spherical Particles in Simple-Shear Flows of Yield Stress Fluids, J. Nonnewton. Fluid Mech. 255, 19 (2018). [31] P. Poulin, H. Stark, T. C. Lubensky, and D. A. Weitz, Novel Colloidal Interactions in Anisotropic Fluids, Science (80-. ). 275, 1770 (1997). [32] O. P. Pishnyak, S. Tang, J. R. Kelly, S. V. Shiyanovskii, and O. D. Lavrentovich, Levitation, Lift, and Bidirectional Motion of Colloidal Particles in an Electrically Driven Nematic Liquid Crystal, Phys. Rev. Lett. 99, 127802 (2007). [33] P. C. Hiemenz and R. Rajagopalan, Principles of Colloid and Surface Chemistry (CRC press, Boca Raton, 1997). [34] J. N. Israelachvili, Intermolecular and Surface Forces (Academic press, 2011). [35] J. Čejková, M. Novák, F. Štěpánek, and M. M. Hanczyc, Dynamics of Chemotactic Droplets in Salt Concentration Gradients, Langmuir 30, 11937 (2014). [36] R. Fujita, T. Matsufuji, M. Matsuo, and S. Nakata, Alternate Route Selection of Self-Propelled Filter Papers Impregnated with Camphor for Two-Branched Water Channels, Langmuir 37, 7039 (2021). [37] B. D. Frank, S. Djalali, A. W. Baryzewska, P. Giusto, P. H. Seeberger, and L. Zeininger, Reversible Morphology-Resolved Chemotactic Actuation and Motion of Janus Emulsion Droplets, Nat. Commun. 13, 2562 (2022). [38] G. Holló, N. J. Suematsu, E. Ginder, and I. Lagzi, Electric Field Assisted Motion of a Mercury Droplet, Sci. Rep. 11, 2753 (2021). [39] H. Xie, Z.-G. She, S. Wang, G. Sharma, and J. W. Smith, One-Step Fabrication of Polymeric Janus Nanoparticles for Drug Delivery, Langmuir 28, 4459 (2012). [40] E. Acosta, Bioavailability of Nanoparticles in Nutrient and Nutraceutical Delivery, Curr. Opin. Colloid Interface Sci. 14, 3 (2009). [41] B. X. Li, V. Borshch, R. L. Xiao, S. Paladugu, T. Turiv, S. V. Shiyanovskii, and O. D. Lavrentovich, Electrically Driven Three-Dimensional Solitary Waves as Director Bullets in Nematic Liquid Crystals, Nat. Commun. 9, 2912 (2018). [42] C. D. Ma, L. Adamiak, D. S. Miller, X. Wang, N. C. Gianneschi, and N. L. Abbott, Liquid Crystal Interfaces Programmed with Enzyme-Responsive 228 Polymers and Surfactants, Small 11, 5747 (2015). [43] J. Wei, H. Yu, H. Liu, C. Du, Z. Zhou, Q. Huang, and X. Yao, Facile Synthesis of Thermo-Responsive Nanogels Less than 50 Nm in Diameter via Soap- and Heat-Free Precipitation Polymerization, J. Mater. Sci. 53, 12056 (2018). [44] X.-L. Tang, S.-M. Guo, Z.-D. Liu, R.-Z. Tang, J.-Y. Pang, and Y. Chen, Preparation of Thermo-Sensitive Poly(N-Isopropylacrylamide) Film Using KHz Alternating Current Dielectric Barrier Discharge, in 2017 3rd International Forum on Energy, Environment Science and Materials (IFEESM 2017), Vol. 120 (Atlantis Press, 2018), pp. 598–602. [45] P. G. de Gennes and J. Prost, The Physics of Liquid Crystals (Clarendon Press, Oxford, 1993). [46] J. Loudet, P. Barois, and P. Poulin, Colloidal Ordering from Phase Separation in a Liquid- Crystalline Continuous Phase, Nature 407, 611 (2000). [47] P. S. Drzaic, Liquid Crystal Dispersions (World Scientific, Singapore, 1995). [48] E. P. Raynes, R. J. A. Tough, and K. A. Davies, Voltage Dependence of the Capacitance of a Twisted Nematic Liquid Crystal Layer, Mol. Cryst. Liq. Cryst. 56, 63 (1979). [49] H. Hakemi, E. F. Jagodzinski, and D. B. Dupré, Temperature Dependence of the Anisotropy of Turbidity and Elastic Constants of Nematic Liquid Crystal Mixture E7, Mol. Cryst. Liq. Cryst. 91, 129 (1983). [50] S. B. Chernyshuk and B. I. Lev, Theory of Elastic Interaction of Colloidal Particles in Nematic Liquid Crystals near One Wall and in the Nematic Cell, Phys. Rev. E 84, 011707 (2011). [51] B. Mustin and B. Stoeber, Single Layer Deposition of Polystyrene Particles onto Planar Polydimethylsiloxane Substrates, Langmuir 32, 88 (2016). [52] Y.-Y. Luk, K.-L. Yang, K. Cadwell, and N. L. Abbott, Deciphering the Interactions between Liquid Crystals and Chemically Functionalized Surfaces: Role of Hydrogen Bonding on Orientations of Liquid Crystals, Surf. Sci. 570, 43 (2004). [53] M. A. Brown, Z. Abbas, A. Kleibert, R. G. Green, A. Goel, S. May, and T. M. 229 Squires, Determination of Surface Potential and Electrical Double-Layer Structure at the Aqueous Electrolyte-Nanoparticle Interface, Phys. Rev. X 6, 011007 (2016). 230 Chapter 8 Substitution of Molecular Structure Alters Surface Anchoring of Liquid Crystals 1. Introduction Liquid crystals (LCs) anchor at interfaces with preferential orientations, combining with their fluidity and long-range ordering, leading to a range of hierarchical organizations in bulk and confined LC phases[1,2]. The responsiveness of LC anchoring to external stimuli, such as electric and magnetic fields[3,4], shear stresses[5] and chemical and biological compounds[6–8], underlies the design of LC displays[9], detection of chemicals and biomolecules[6,7] and self-assembled structures of LC organizations[10]. While the experimental data of multiple LC interfaces is increasing over time, an explicit molecular-level explanation of anchoring of LCs, however, is still lacking. One difficulty is that the final orientation of the liquid crystalline molecules at an interface results from a combination of the chemical nature of the LC interface and the topology of the interface that may induce elastic distortion of the director[11–15]. In experiments, these two factors are often entangled[6,10,12–14], and past studies on theory attempt to evaluate the relative importance of the two factors[11,15–17]. A theoretical model by Berreman[11] attributes the polar and azimuthal surface anchoring to the elastic distortion of the liquid crystal induced by the grooves of a surface. Fukuda, et al.[15] further corrected the azimuthal distortion term in Berreman’s model. While these models emphasize geometry-induced surface anchoring, several papers leverage molecular interactions. A seminal theoretical work from Parsons[18] describes how the van der Waals interactions lead to planar anchoring for axisymmetric particles against a free surface. The inclusion of dipole-dipole interactions leads to the same planar 231 interactions or changes to homeotropic orientations[19]. On the other hand, some theoretical works have described that short attractive and repulsive interactions lead to homeotropic anchoring for cylindrical molecules[20], while entropy is known to favor planar anchoring at the nematic-isotropic interface[21] To address the thermotropic LC systems, theoretical works have been complemented with experimental and simulation data, where the surface electric polarization at interfaces has played a leading role in the homeotropic anchoring at free surface for liquid crystal molecules like 5CB and 8CB[22]. Then, highly complex simulations and advanced sampling calculations have put into account the role of dipole interaction at the interface for the observed planar anchoring of 5CB in water[23]. What has been missed in those past works, was the contribution of the intra-molecular flexibility and the dispersion interaction between different molecules at interfaces. In this work, we investigate how intra- and inter-molecular interaction contributes to the orientational ordering of thermotropic LCs by combining experiments and atomistic simulations. We chose a non-polar liquid, perfluorononane (F9), that can phase separate from the nematic liquid crystalline phases (4′-pentyl-4- biphenylcarbonitrile, 5CB and 4-(trans-4′-pentylcyclohexyl)-benzonitrile, PCH5) to form an interface[24], within which only interfacial chemistry is responsible for the formation of anchoring, and no surface topology is involved as compared to a solid substrate. Alternation of a submolecular structure in the two mesogens leads to two distinct macroscopic anchoring, which can be explained explicitly using partial distribution functions and free energy calculations. This work demonstrates that the orientational ordering can be traced back to molecule structures that transmit interactions across length scales. 232 2. Results and Discussion Anchoring of 5CB and PCH5 at F9 interfaces from experimental observations. Our investigation started with two liquid crystals 4′-pentyl-4- biphenylcarbonitrile (5CB, Figure 8.1a) and 4-(trans-4′-pentylcyclohexyl)-benzonitrile (PCH5, Figure 8.1b) at fluorocarbon (perfluorononane, F9, Figure 8.1c) interfaces in the temperature range of the nematic phases. Unlike many miscible mixtures of LCs and hydrocarbons (e.g., hexane and toluene), we found that 5CB (or PCH5) and F9 are immiscible over the temperature range of study[19], which enable the exploration of interfacial properties of LCs. 5CB and PCH5 molecules are similar as they both possess a benzonitrile head group and a pentyl tail, and differ in that one aromatic ring of 5CB is replaced by a cyclohexyl group on PCH5 (Figure 8.1). To characterize the orientational ordering of LCs at interfaces, thin films (22 - 24 μm) of liquid crystals were supported by glass substrates that were coated with polyimide to induce a strong homeotropic (perpendicular) anchoring of LC to the bottom substrates, and were in contact with F9 at the top. The samples were subsequently observed under polarized light optical microscopy. Surprisingly, although the subtle difference in molecular structure, the microscopic anchoring behaviors of 5CB and PCH5 are dramatically distinct. Whereas the 5CB film shows a dark optical appearance that indicates homeotropic anchoring at the F9 interface (Figure 8.1d,e), the PCH5 film exhibits a bright texture, which means the orientational ordering tilts from the surface normal (Figure 8.1f,g). With the one elastic constant approximation[1] of the full Frank–Oseen expression for the elastic free energy of the thin LC film, the tilt angle of the LCs at F9 interfaces can be quantified by the so-called optical retardance (i.e., optical phase shifts between two polarization directions that determine the color of the film), ∆𝑟, which is given by[25]: 233 𝑑 𝑛𝑜𝑛𝑒 ∆𝑟 ≈ ∫ − 𝑛 𝑜 𝑑𝑧 (8.1) 0 √𝑛 2𝑠𝑖𝑛2 𝑧 ( 𝜃 ) + 𝑛 2 2 𝑧 𝑜 𝑐𝑜𝑠 ( 𝜃 )( 𝑑 𝑡𝑜𝑝 𝑒 𝑑 𝑡𝑜𝑝 ) in which 𝑛𝑜 and 𝑛𝑒 are the indices of refraction perpendicular and parallel to the optical axis of the LC, respectively, 𝑑 is the thickness of the film, 22 - 24 μm. The solution of the equation from the experimentally measured value of ∆𝑟 yields 𝜃𝑡𝑜𝑝, which is the tilt angle with respect to the surface normal of the F9 interface. We measured the retardance (Supplementary Figure S8.1) across the 5CB (𝑛𝑒 = 1.716 and 𝑛𝑜 = 1.533 at 24 °C) and PCH5 (𝑛𝑒 = 1.600 and 𝑛𝑜 = 1.492 at 40 °C) films[26] using polscope to be 6.2 ± 3.5 𝑛𝑚 and 1096.8 ± 16.3 𝑛𝑚, respectively, which yield 𝜃𝑡𝑜𝑝 = 4° (5CB) and 80 − 86° (PCH5). Atomistic molecular dynamics simulations of films of 5CB and PCH5 with F9 and their characterizations. While the optical experiments offer macroscopic characterization of average alignment of LC molecules at interfaces, atomistic molecular dynamics simulations of the same system can provide three-dimensional structural and dynamic information. In our simulations, 5CB, PCH5 and F9 molecules were described at the united-atoms (UA) level of chemical detail[27,28], where the CH, CH2, CH3, CF2 and CF3 groups are treated as single isotropic interaction sites (Supplementary Tables S8.1-S8.16 and Figures S8.2-S8.4). The UA model was shown in good agreement with experiments on densities of 5CB, F9 and PCH5, respectively (Methods and Supplementary Figure S8.5). The intermolecular interactions between 5CB or PCH5 and F9 molecules were modeled with the Lennard-Jones potential, where the parameters CHx-CFy were modified based on the proposed modification for the TRAPPE-UA force field and fitted to reproduce the temperature-dependent interfacial tensions (Methods and Supplementary Figure S8.6). 234 We start quantifying the molecular orientation of each mesogen with respect to the normal to the interface, ?̂? = (0,0,1), by calculating the molecular Figure 8.1 Distinct alignment of two nematic liquid crystals at interfaces. Molecular structures (left) and corresponding schematic illustrations used in the simulations with united atom force field (right) of the nematogens a 4′-pentyl-4- biphenylcarbonitrile (5CB), b 4-(trans-4′-pentylcyclohexyl)-benzonitrile (PCH5) and c perfluorononane (F9). d-g, Cross-polarized micrographs (top view) and corresponding schematic illustrations (side view) of d,e 5CB films (dark, homeotropic anchoring, at 25 °C) and f,g PCH5 films (bright, tilted anchoring, at 40 °C) in F9, respectively. Liquid crystal films in d,f are stabilized using copper grids. Scale bars, 200 m. The inset in d shows the conoscopy micrograph in one grid. The red double-headed arrow indicates the orientational ordering of the liquid crystals at the interfaces. 235 second Legendre polynomial: 3 1 𝑃2(𝑖) = 𝑐𝑜𝑠 2𝛽(𝑖) − (8.2) 2 2 where 𝑐𝑜𝑠𝛽(𝑖) = ?̂? ∙ 𝑢?̂? defines the angle between the long axis of the mesogen i pointing toward the cyano (CN) group (thus a molecule with the CN group pointing toward, away from or parallel to the upper F9 interface has a 𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 = +1,−1 𝑜𝑟 0). It is clearly shown that majority of the mesogen orienting along the 5CB- F9 surface normal (𝑃2 = ± 0.8 − 1) across the nematic 5CB film, which appear as red- orange-yellow in the color-coded profile of the second Legendre polymonial 𝑃2 by molecule at 200 ns (Figure 8.2a). In the nematic phase of PCH5 (Figure 8.2e), however, the molecules adopt preferentially perpendicular to the surface normal (blue). In addition to 𝑃2 , two-dimensional (2D) maps of probability distribution function 𝑃 (𝑧,𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 ) of molecular orientation at any distance 𝑧 provide the assessment on the preferred orientation of molecular dipoles. For the 5CB-F9 film (Figure 8.2b), the 2D map shows that the layer near the interface consists of molecules having an antiparallel orientation of molecular dipole (𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 = ± 1), with the alkyl tail of 5CB pointing toward the F9 interface. Two to three induced layers are present near the interface, which gradually fade out and exhibit a uniform homeotropic anchoring ( 𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 = ± 1 ) when moving to the center of the film. The two symmetric narrow bands at 𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 = ± 1 indicate that mesogens assume the antiparallel orientation across the entire film. In the nematic phase of PCH5 at 40 °C, the molecules also exhibit small polarity at the interface with their alkyl tail pointing toward the F9 interface; the distribution of orientations, however, is wide with a range of 𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 = ± 0.3 − 1 , consistent with the tilt from the surface normal that is observed in the experiments. A single surface layer is formed with a much weaker 236 Figure 8.2 Alignment of 5CB and PCH5 molecules at F9 interfaces obtained from atomistic molecular dynamics simulations. a,e The second Legendre polynomial 𝑃2 by molecule and b,f two-dimensional histograms of orientational probabilities 𝑃(𝑧,𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 ), c,g scalar order parameter 𝑆 and d,h components of director obtained by construction and diagonalization of the Q tensor for a-d 5CB at 25 °C and e-h PCH5 at 40 °C. In the nematic phase of 5CB, two induced surface layers are formed with homeotropic anchoring, and the perpendicular orientation near the free surface spreads over the entire film; whereas in the nematic phase of PCH5, a tilt orientational ordering forms. Here, 𝑛𝑧 is the 𝑧 component of the director and 𝑛𝑥𝑦 = √(𝑛 2 𝑥 + 𝑛 2 𝑦). The error bars show the standard deviation of the computed values over the final 20 ns block of MD simulations. orientational preference, and the molecules assume a preferred tilt orientation across the entire PCH5 film. Overall, the atomistic molecular dynamics simulations recapitulate 237 the distinct anchoring behaviors of 5CB versus PCH5 observed in experiments. We also evaluated the scalar order parameter 𝑆 (i.e., the extent of molecular ordering[1]) with respect to the preferred alignment direction (i.e., director) 𝑛 as a function of the distance 𝑧 from the film center using the ensemble average of the tensorial orientation order parameter[22,29] 𝑄: 3 1 𝑄 = ⟨𝑎𝑎 − 𝐼⟩ (8.3) 2 3 where 𝑄 is a traceless and symmetric tensor, 𝐼 is the identity matrix, 𝑎 is a unit vector giving the orientation of the long axis of mesogen, 𝑎𝑎 = ?⃗??⃗?𝑇 represents the dyadic product, and 〈⋯ 〉 denotes ensemble average. Figure 8.2c,d shows the order parameter and the profiles of director components for 5CB at F9 interface. The scalar order parameter in the nematic phase of 5CB at 25 °C is 0.6-0.7 across the film, with, similar to the profile of 𝑃(𝑧, 𝑐𝑜𝑠𝛽), weak oscillations in 𝑆 near the interface, which indicates the formation of a few induced layers. The mesogens are oriented on average along the surface normal. In the nematic phase of PCH5 (Figure 8.2g,h), 𝑆 is around 0.75 across the film without the oscillations presented in the 5CB system and an average alignment that is neither homeotropic nor perfect planar, and no induced layering is observed, different from the nematic 5CB. Compartmentalization of mesogen and spectral sampling of free energy calculations. In order to provide an explicit explanation of LC molecular orientational ordering from atomistic chemistry, we segregate the mesogen into four compartments (Figure 8.3), namely (i) the cyano group, (ii) the phenyl ring that is adjacent to the cyano group (ring 1), (iii) the other phenyl ring that is spatially close to the alkyl tail for 5CB or the cyclohexyl ring for PCH5 (ring 2) and (iv) the alkyl chain, and characterize the 238 free energy change 𝛥𝐴 𝑖𝑗(𝑟) between each compartment and perfluorocarbon center of mass as ∂A 𝛥𝐴𝑖𝑗(𝑟) =∫ dr (8.4) ∂r 𝜕𝐴 where is the generalized force. To calculated these free energy maps, we used a new 𝜕𝑥 method based on the adaptive biasing force algorithm[30] with a speed up in convergence using spectral methods[31]. Figure 8.3a shows the comparison of the four free energy maps for 5CB and F9. The perfluorocarbon-alkyl tail interaction is favored in 5CB-F9 over other interactions with the lowest free energy of all compartments in all the studied distances in the nematic phase, which is consistent with the observed polarization results, where the polar head points to the nematic bulk. The two aromatic rings are the ones with higher barrier energy for close contact with the perfluoroalkane molecule, with the cyano group in the range between the tail and those rings. This also help us to explain the homeotropic behavior, since in order to minimize the free energy of contact between those molecules, they will interact with the head or tail, leading to perpendicular alignment of 5CB at the interface. Moreover, in Figure S8.7 we show that once 5CB reaches the isotropic phase at 313K, all free energy maps per compartment falls onto the same curve showing that there is no preferred interaction between them and the F9, and therefore there is no defined anchoring of the isotropic molecules. On the other hand, for PCH5 the free energy maps (Figure 8.3d), the molecular block with lowest free energy with F9 is the one that distinguished between 5CB and PCH5, the cyclohexyl ring, followed closely of the pentane tail. These favored interactions between those groups are exerting a force imbalance in the molecule that is solved by lying not perpendicular to interface, like 5CB, but with a tilted orientation to the surface, leading to the experimentally observed anchoring for PCH5. To get more insight into the molecular mechanism for the anchoring formation, we calculate 2D free energy maps using projected distance of the center of mass of molecules to the molecular axis 239 in a LC molecule that is defined by the nitrogen in cyano group and the carbon in ortho position in the first ring. In this way, positive values of the parallel projection correspond to regions close to the cyano head, while negative are related to the alkyl tail region. The map in Figure 8.3b corresponds to the local environment of a 5CB molecule interacting with F9. This map shows the core region of the LC molecule in yellow that has a radius around 2.5 Angstroms. Then we can see that two minima regions appears in the upper and lower right borders of the map with a barrier of ~6 kJ/mol in between those regions. This again is expected for the homeotropic anchoring of 5CB. Next, the Figure 8.3c shows the 2D free energy map between two 5CB molecules in the nematic phase (T=300K). Here the distinct minima appear around the coordinates (4.8, 2.0) and it spreads parallel to the main axis of the 5CB. This shape of the minima corresponds to the π stacking formation of dimers observed in the whole nCB family. All these maps complement to explain the homeotropic alignment; the preferred interaction of 5CB and F9 with the alkyl tail and cyano head over core region (Figure 8.3a) leads to two minima region with an energy barrier in between in the local environment of 5CB (Figure 8.3b) that has as consequence the homeotropic alignment and together with the minima shape in the 5CB-5CB map that compensate for the positional entropic loss due to homeotropic alignment and is observed as a smectic like layering near the interface (Figure 8.2b). In the case of PCH5, Figure 8.3e has just one minima region corresponding to cyclohexyl and alkyl tail interaction with F9 with higher barriers for cyano and phenyl interactions. This shape for the free energy surface does not allow homeotropic alignment leaving to configurations with tilted orientations. Figure 8.3f 240 Figure 8.3 Origin of the orientational ordering. a,d Potential of Mean Force using Spectral ABF method in PySAGES for each compartment in a 5CB or d PCH5 to F9. b,c,e,f Free energy surfaces obtained with Spectral ABF with the parallel and perpendicular projection collective variables by orienting the C-N axis (described in text) in the z direction and calculating the distance from the COM for 5CB b F9 favors tail contact, and c 5CB molecules exhibit parallel packing by π stacking; whereas for PCH5, e F9 favors in contact with tail and cyclohexyl ring, and f PCH5 molecules pack by phenyl ring, leading to a tilt alignment. has the free energy map for two PCH5 molecules interacting, here the closer minima to the central molecule corresponds to a π stacking of the phenyl rings, this leads to the minima spreading in a direction that is not parallel to the axis of the molecule, like in 5CB, but with a certain angle. This shape of the minima could be limiting the layering at the interface showed in Figure 8.2e. So now, in the case of PCH5, the cyclohexyl ring substitution favors its interaction with F9 (Figure 8.3d) by a tilted orientation (Figure 8.3e) and the self-assembly of PCH5 dimers are not compatible with a layering structure 241 necessary for homeotropic anchoring (Figure 8.3f). Temperature-dependent ordering and Rapini-Papoular anchoring coefficient. Figure 8.4a-d shows, for example, 2D maps of probability distribution function 𝑃2(𝑧,𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 ) of 5CB molecular orientation at F9 interface as a function of temperature. An ordered layer was induced near the surface for all temperatures, especially it is also present in the isotropic phase at 40 °C. In the nematic phase at 22, 25 and 30 ° C (Figures 8.2b and 8.4a-c), a similar trend was observed, where two asymmetric narrow white/red bands throughout the film demonstrate the normal ordering with antiparallel dipoles. Whereas in the isotropic phase of 5CB at 40 °C (Figure 8.4d), the mesogens do not adopt an orientational ordering beyond the induced surface layer. PCH5 shows a similar trend that the orientational ordering relaxes with the increase in temperature (Supplementary Figure S8.8). We finally evaluate the anchoring energy by calculating a Boltzmann inversion of the positional-orientational distribution function as[16,17]: 𝑁(𝑧𝑖) 𝑊(𝑧𝑖 ,𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 ) = −𝑘𝐵𝑇 𝑙𝑛 𝑙𝑛 𝑃(𝑧𝑖 ,𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 ) +𝑊0 (8.5) 𝐴 in which the subscript 𝑖 indicates that 𝑧𝑖 is a discrete variable, 𝑘𝐵 is the Boltzmann constant, 𝑇 is system temperature, 𝑁(𝑧𝑖)/𝐴 takes into account the number of molecules per unit area, constant 𝑊0 is used to shift the energy minimum to zero. The surface free energy can be fitted into the Rapini-Papoular form 1 𝑊𝑅(𝑧, 𝛽) = 𝑊 𝐴 0 (𝑧) − 𝑊 𝐴 2 (𝑧) (𝛽 − 𝛽𝑒𝑞(𝑧)) (8.6) 2 where 𝑊𝐴 𝐴0 and 𝑊2 are the fitting parameters. We note that the resulting Rapini-like anchoring coefficient 𝑊𝐴2 values from atomistic molecular dynamics simulations are of the order of 0.01 – 0.1 J m-2, typically larger than that determined from macroscopic experimental observation8 (~0.0001 J m-2). 242 The resulting Rapini-Papoular coefficients 𝑊𝐴2 (𝑧) of 5CB are plotted for different temperatures (Figure 8.4e). In the nematic phase of 5CB (at 22, 25 and 30 °C), two peaks appear at 15 and 40 Å, indicating again two induced layers near the surface. The anchoring energy start to relax near the nematic to isotropic transition at 35 °C, where the second peak decreases significantly. In the isotropic phase (40 ° C), the first peak still exists with approximately the same 𝑊𝐴2 because of the packed layer of mesogens near the interface. Figure 8.4f. The nematic phase of PCH5 exhibit a lower 𝑊𝐴2 (𝑧) against F9 compared to 5CB (isotropic) at 40 ° C (Figure 8.4g). This is consistent with a higher affinity of PCH5 with Figure 8.4 Temperature dependence of molecular alignment. a-d The second Legendre polynomial 𝑃(𝑧,𝑐𝑜𝑠 𝑐𝑜𝑠 (𝛽) ) of 5CB-F9 films as a function of temperature. From nematic to isotropic phase, profile shows that bulk 5CB loses orientational ordering, while a distinct oriented layer is maintained near the interface. e Rapini-Papoular coefficient for 5CB-F9 interface from atomistic molecular dynamic simulation as a function of temperature, and f surface free energy map at 30 °C. In the nematic phase of 5CB, two induced surface layers are formed with sinusoidal fluctuation. g Rapini-Papoular coefficient for PCH5-F9 interface at 40 °C as compared to 5CB-F9 interface, and h surface free energy map of PCH5 at F9 interface. In contrast to nematic 5CB, PCH5 at nematic phase only formed one surface layer. 243 F9, and also less anisotropy because of the cyclohexyl group in the PCH5 molecule. Similar to isotropic 5CB, one densely packed layer with high anchoring energy appears at the F9 interface, and decays rapidly beyond this surface layer (Figure 8.4h and Supplementary Figure S8.8). 3. Conclusions In the broad range of applications for liquid crystal materials, the anchoring behavior of the material plays an important role. The final orientation of the liquid crystalline molecules at an interface is a consequence of the chemical nature of the molecular components presented at that interface and the geometrical constraints imposed by the topology of the phases in contact. In this work, we selected two highly similar molecules, 5CB and PCH5, that differs only in a substitution of a phenyl ring with a cyclohexyl ring in their structure. We proposed to systematically decouple the chemical interaction and the topology of the interface by using an isotropic fluid, in this case perfluorononane, with a sharp interface with the LC molecules. Then, atomistic molecular dynamic calculations with enhanced sampling methods were used to understand the experimental results by interactions of the molecules at the interface. More broadly, this approach disentangles the complex interactions that give rise to a determined macroscopic behavior as the liquid crystalline orientation by compartalizing a mesogen and by comparing the interactions between each compartment and an outer fluid phase. A larger affinity to the interfacial fluid and flexibility of the LC molecules traduces to larger anchoring angles, while a rigid core favors the homeotropic alignment. In this way, atomistic molecular dynamics simulations could pave the way for exploration of how diversity of intermolecular interactions, including complex interactions like non-polar and polar forces and hydrogen bonding, further bias molecule orientation. 244 4. Supporting Information Methods Anchoring of liquid crystals at perfluorocarbon interfaces. The nematogens, 4′- pentyl-4-biphenylcarbonitrile (5CB, 22 °C nematic 35.3 °C isotropic) and 4-(trans-4′- pentylcyclohexyl)-benzonitrile (PCH5, 30 °C nematic 55 °C isotropic) were purchased from HCCH (Jiangsu Hecheng Display Technology Co., LTD) and used as received. Benzonitrile and pentane (Sigma-Aldrich) were also used as received. Perfluorononane (F9, Sigma-Aldrich, stored and degassed by freezing at -20 °C) was first added into a chamber incubator (Thermo Fisher Scientific). Glass slides (Thermo Fisher Scientific) were coated with polyimide (HD Microsystems) as substrates to induce a homeotropic anchoring (perpendicular). A thin film (18 μm) of liquid crystals was stabilized by Copper TEM grids (Electron Microscopy Sciences, G75-Cu, 18 μm thick) on the glass substrate and subsequently submerged into F9 for microscopy observation. Density and interfacial tension measurements. Density of 5CB, PCH5 and F9 as a function of temperature was measured using a Density Meter (DSA 5000 M, Anton Paar). The interfacial tensions were measured using the pendant drop method performed with an optical tensiometer (Attension Theta T200, Biolin Scientific). The results presented in this paper were obtained using metal 22 gauge needles (Hamilton, blunt point) from three independent measurements with an equilibration time of 5 min. An environmental chamber heated by a water bath was used to control the temperature. Force fields. Molecular dynamics simulations of 5CB-F9 and PCH5-F9 thin films were carried out over the same temperature range as that investigated in the anchoring measurements. The united-atom model of Tibero et al.[28], which has been shown to 245 predict liquid-phase behavior in good agreement with experiment, was used for 5CB. The united atom model for PCH5 was generated for this work: ESP charges were calculated using Gaussian 09 Rev A [32]. The bonded and non-bonded parameters were taken from past studies[28,33-35]. The TRAPPE united model for F9 was used in the simulations[36], with the bond parameters taken from Amat et al[27]. Detailed information of the force fields is shown below in SI. Density for single phases. Simulations were done using the HOOMD-Blue package version 2.9.7[37]. The equation of state for 5CB or PCH5 was calculated using a cubic box with 256 molecules in each case, while for F9 512 molecules were used. The isothermal-isobaric ensemble (NPT) in the MTTK algorithm was used for the calculation. A time step of 0.5 fs was employed with thermostat and barostat coefficients of 1 ps-1 and 10ps-1 respectively. For non-bonded interactions a spherical cutoff radius of 12 Å coupled to a smoothing function between 10 and 12 Å was also implemented. The summation of Coulomb interactions was handled using the particle-particle-particle mesh Ewald (PPPM) method with a cut-off of 12 Å 643 grid points and an order of 4[38]. All systems were subjected to multiple intervals of 50 ns MD simulations for each temperature (4 × 50 = 200 ns). The density was calculated using 5000 snapshots taken from the last interval. See below for detailed information. Mixture states. F9-LC films were created by exposing the lower and upper planes of a LC film to F9. A periodic simulation box of dimensions 110 × 110 × 330 Å3 was prepared with 4,096 molecules of 5CB or PCH5 molecules and 2048 F9 molecules. The FIRE algorithm was used to minimize energy with a tolerance of 10−7 kJ/mol, followed by 1 ns simulation in the NVT ensemble, then 100ns in the isotropic NPT ensemble, and finally 200 ns in the constant area-normal pressure isothermal ensemble (NPNAT). 246 Spectral ABF Free Energy Calculations. The PyAGES suite was selected to make the free energy calculations. For each compartment in the LC molecule (5CB/PCH5) free energy calculations were done based on their relative distance to the center of mass of perfluorononane. 10ns were sampled for each free energy map in the region of 4-20 Å where restraints were applied outside the region with a force constant of 10kJ/mol Å. For the 2D free energy maps, 100ns was used to calculate the surfaces. Density, orientational profiles and correlation functions in mixed states. The density, orientational profiles and correlation functions were calculated in the NPNAT ensemble. For profiles, the box was divided in 100 bins along the z dimension and 2000 snapshots were taken during 20 ns of simulation after previous equilibration described above. The results were block averaged using 10 blocks. Interfacial tension calculations. The interfacial tension was calculated using the Kirwood-Buff formula[39] in the NPNAT ensemble. For each temperature, the pressure tensor was calculated every 0.2 ps during 100ns and block averaged using 10 blocks. The long-range correction in the interfacial tension was calculated using the method described before[40]. The reader is referred to the text below in the SI for details in the calculations. Rapini-Papoular coefficient and anchoring angle. The Rapini-Papoular coefficient was calculated using the method described in the past study[22]. The anchoring angle corresponds to the one observed in the first peak of the Figure 8.4e. Extra information is given later in SI. 247 Figure S8.1 a,c Polarized light and b,d polscope micrographs using 632 nm light source show retardance for 5CB-F9 and PCH5-F9 interfaces, respectively. Scale bars, 200 μm. 248 Model force field potentials In this work we adopt the following shape for the force field potentials: 𝑉𝑡𝑜𝑡𝑎𝑙 = 𝑉𝑣𝑑𝑊 + 𝑉𝑒𝑙𝑒𝑐 + 𝑉𝑏𝑜𝑛𝑑𝑒𝑑 (8.7) 12 6 𝜎𝑖𝑗 𝜎𝑖𝑗 𝑉𝑣𝑑𝑊 = 4𝜀𝑖𝑗 [( ) − ( ) ] (8.8) 𝑟𝑖𝑗 𝑟𝑖𝑗 1 𝑞𝑖𝑞𝑗 𝑉𝑒𝑙𝑒𝑐 = (8.9) 4𝜋𝜖0 𝑟𝑖𝑗 𝑉𝑏𝑜𝑛𝑑𝑒𝑑 = 𝑉𝑏𝑜𝑛𝑑 + 𝑉𝑎𝑛𝑔𝑙𝑒 + 𝑉𝑑𝑖ℎ𝑒𝑑𝑟𝑎𝑙∗ (8.10) 1 𝑉𝑏𝑜𝑛𝑑 = 𝑘𝑏(𝑟 − 𝑅 ) 2 0 (8.11) 2 1 𝑉𝑎𝑛𝑔𝑙𝑒 = 𝑘𝑎(𝜃 − 𝐵 ) 2 0 (8.12) 2 For the 5CB and PCH5 models, the dihedral potential takes the following shape: 𝑉𝑑𝑖ℎ𝑒𝑑𝑟𝑎𝑙 =∑ 𝑘𝑛(1 +𝑐𝑜𝑠 𝑐𝑜𝑠 (𝑛𝜑 − 𝜑0) ) (8.13) While for the F9 model: 7 𝑉𝑑𝑖ℎ𝑒𝑑𝑟𝑎𝑙 =∑ 𝐶𝑖(𝑐𝑜𝑠𝜑) 𝑖 (8.14) 𝑖=0 Also, the 5CB and PCH5 models follow the amber 1-4 scaling; 0.5 reduction for van der Waals interactions and 0.8333 for electrostatic interactions, while the F9 model follows the TRAPPE-UA convention with no 1-4 non-bonded interactions. 249 1. Force Field parameters for the 5CB model. Figure S8.2 United Atom representation of 5CB with atoms enumerated. Table S8.1 Atomic properties of the elements of the 5CB model. Atom number Atom Type Mass [u.m.a] Charge [e-] 1 NY 14.010 -0.559 2 CY 12.000 0.767 3 CA 12.000 -0.804 4 CD 13.008 0.242 5 CD 13.008 0.156 6 CA 12.000 -0.211 7 CD 13.008 0.156 8 CD 13.008 0.242 9 CA 12.000 -0.330 10 CD 13.008 0.150 250 11 CD 13.008 0.196 12 CA 12.000 -0.640 13 CD 13.008 0.196 14 CD 13.008 0.150 15 C2 14.016 0.286 16 C2 14.016 0.030 17 C2 14.016 -0.056 18 C2 14.016 0.030 19 C3 15.024 -0.001 Table S8.2 Bond parameters for 5CB Bond Name K [kJ/mol Å2] R0 [Å] CY-NY 5020.8 1.16 CY-CA 802.49 1.42 CA-CD 3924.59 1.41 CD-CD 3924.59 1.41 CA-CA 3924.59 1.48 CA-C2 2652.65 1.51 C2-C2 802.49 1.54 C2-C3 802.49 1.54 Table S8.3 Angle Parameters for 5CB. Angle Name K [kJ/mol rad2] Β0 [degrees] C2-C2-C2 527.184 112.4 C2-C2-C3 519.653 114.0 251 C3-C2-C3 519.653 114.0 CA-CA-CD 711.28 120.0 CD-CA-CD 711.28 120.0 CA-CD-CD 711.28 120.0 CY-CA-CD 711.28 120.0 CA-CY-NY 665.256 180.0 CD-CA-C2 585.76 120.0 CA-C2-C2 527.184 112.4 Table S8.4 Dihedral Parameters for 5CB Dihedral Name K [kJ/mol] n Phi0 (degrees) X-CA-CD-X 11.09 2 180 X-CD-CD-X 22.18 2 180 X-C2-CA-X 2.09 6 0.0 NY-CY-CA-CD 0.0 1 180 X-C2-C2-X 2.81 1 0.0 X-C2-C2-X 0.57 2 0.0 X-C2-C2-X 5.88 3 0.0 X-CA-CA-X 0.04 1 180 X-CA-CA-X 2.90 2 180 X-CA-CA-X 0.004 3 180 X-CA-CA-X 4.45 4 0.0 Table S8.5 Non-bonded parameters for 5CB Atom Epsilon Sigma [Å] Epsilon 1-4 Sigma 1-4 [Å] 252 [kJ/mol] [kJ/mol] NY 0.711 3.25 0.3555 3.25 CY 0.3598 3.3997 0.1799 3.3997 CD 0.2962 3.4745 0.1481 3.4745 CA 0.2962 3.4745 0.1481 3.4745 C2 0.2950 3.626 0.1475 3.626 C3 0.4393 3.6527 0.2197 3.6527 253 2. Force Field parameters for the PCH5 model Figure S8.3 United Atom representation of PCH5 with atoms enumerated. Table S8.6 Atomic properties of the elements of the PCH5 model. Atom number Atom Type Mass [u.m.a.] Charge [e-] 1 N 14.010 -0.459 2 C 12.000 0.310 3 CB 12.000 0.053 4 CA1 13.008 0.029 5 CA1 13.008 -0.092 6 CB 12.000 0.131 7 CA1 13.008 -0.092 8 CA1 13.008 0.029 9 CS1 13.008 -0.007 10 CS2 14.016 -0.053 254 11 CS2 14.016 0.067 12 CS1 13.008 0.209 13 CS2 14.016 0.067 14 CS2 14.016 -0.053 15 CH2 14.016 -0.206 16 CH2 14.016 0.115 17 CH2 14.016 -0.083 18 CH2 14.016 0.167 19 CH3 15.024 -0.132 Table S8.7 Bond parameters for PCH5. Bond Name K [kJ/mol Å2] R0 [Angstrom] C-N 5020.8 1.16 C-CB 802.49 1.42 CB-CA1 3924.59 1.41 CA-CA1 3924.59 1.41 CB-CS1 2652.65 1.51 CS1-CS2 2510.4 1.53 CS2-CS2 2510.4 1.53 CS1-CH2 2510.4 1.53 CH2-CH2 802.49 1.54 CH2-CH3 802.49 1.54 Table S8.8 Angle parameters for PCH5 Angle Name K [kJ/mol rad2] B0 [degrees] 255 N-C-CB 665.256 180 C-CB-CA1 711.28 120 CB-CA1-CA1 711.28 120 CB-CS1-CS2 288.84 109.5 CS1-CS2-CS2 288.84 109.5 CS1-CH2-CH2 529.54 111.0 CH2-CH2-CH2 527.184 112.4 CH2-CH2-CH3 519.653 114.0 Table S8.9 Dihedral parameters for PCH5 Dihedral Name K [kJ/mol] n Phi0 [degrees] X-CB-CA1-X 11.0876 2 180 X-CA1-CA1-X 22.1752 2 180 X-CB-CS1-X 1.0999 2 180 X-CB-CS1-X −0.4824 4 180 X-CB-CS1-X −0.0965 6 180 X-CS1-CS2-X 2.8058 1 0 X-CS1-CS2-X 0.5711 2 0 X-CS1-CS2-X 5.8752 3 0 X-CS2-CS2-X 2.8058 1 0 X-CS2-CS2-X 0.5711 2 0 X-CS2-CS2-X 5.8752 3 0 N-C-CB-CA1 0 1 180 X-CH2-CH2-X 2.8058 1 0 X-CH2-CH2-X 0.5711 2 0 256 X-CH2-CH2-X 5.8752 3 0 Table S8.10 Non-bonded parameters for PCH5 Atom Epsilon Sigma [Å] Epsilon 1-4 Sigma 1-4 [Å] [kJ/mol] [kJ/mol] N 0.4992 2.950 0.2496 2.950 C 0.4992 3.550 0.2496 3.550 CB 0.4200 3.695 0.2100 3.695 CA1 0.4197 3.695 0.20985 3.695 CS1 0.0831 4.650 0.04155 4.650 CS2 0.3825 3.950 0.19125 3.950 CH2 0.3825 3.950 0.19125 3.950 CH3 0.8148 3.750 0.4074 3.750 257 3. Force Field parameters for the F9 model. Figure S8.4 United Atom representation of F9 with atoms enumerated. Table S8.11 Atomic properties of the elements of the F9 model Atom Number Atom type Mass [u.m.a.] Charge [e-] 1 CF3 69.01 0.0 2 CF2 50.01 0.0 3 CF2 50.01 0.0 4 CF2 50.01 0.0 5 CF2 50.01 0.0 6 CF2 50.01 0.0 7 CF2 50.01 0.0 8 CF2 50.01 0.0 9 CF3 69.01 0.0 258 Table S8.12 Bond parameters for F9 Bond name K [kJ/mol Å2] R0 [Angstrom] CFX-CFX 802.345 1.53 Table S8.13 Angle parameters for F9 Angle name K [kJ/mol rad2] B0 [degrees] CFX-CFX-CFX 519.6537 114 Table S8.14 Dihedral parameters for F9 Dihedral name C0[kJ/ C1[kJ/ C2[kJ/ C3[kJ/ C4[kJ/ C5[kJ/ C6[kJ/ C7[kJ/ mol] mol] mol] mol] mol] mol] mol] mol] CFX-CF2-CF2- 7.8189 - 11.267 56.538 - - 76.606 34.286 CFX 2.3505 8 3 65.478 117.79 9 3 9 93 Table S8.15 Non-bonded parameters for F9 Atom Epsilon [kJ/mol] Sigma [Å] CF2 0.2286 4.730 CF3 0.7236 4.360 The combination rules used in this work follow the classic Lorentz-Berthelot scheme for 5CB-5CB, PCH5-PCH5 and F9-F9 interactions, while for 5CB-F9 and PCH5-F9, these rules are modified by two scaling parameters: 𝜀𝑖𝑗 = (1 − 𝑏𝑖𝑗)√𝜀𝑖𝜀𝑗 (8.15) 1 + 𝑎𝑖𝑗 𝜎𝑖𝑗 = (𝜎𝑖 + 𝜎𝑗) (8.16) 2 259 In the Table S8.16 are described such parameters. Table S8.16 Scaling parameters for the intermolecular LJ interactions. Atoms a b Cyano-CFX 0.03 0.33 Aromatic-CFX 0.03 0.33 Aliphatic Cyclic-CFX 0.01 0.11 Aliphatic Linear-CFX 0.01 0.11 260 Density, orientational profiles and correlation functions The density profiles were calculated in the z direction of the box according to the equation: 𝑚𝑖𝑛𝑖(𝑧) 𝜌(𝑧) = 〈∑ 〉 (8.17) 𝐿𝑥𝐿𝑦∆𝑧 𝑖 where 𝑚𝑖 denotes the mass of molecule 𝑖, 𝑛𝑖(𝑧) is the molecule at position z, 𝐿𝑥,𝑦 the length of the box in the x and y direction and ∆𝑧 is the slab size in the z direction. The 〈⋯ 〉 denotes ensemble average. The orientational profiles are generated by calculating the average of the Q tensor in z slabs: 𝑁(𝑧) 3 1 𝑄(𝑧) = 〈∑ ( ?̂?𝑖(𝑧)?̂?𝑖(𝑧) − 𝐼) /𝑁(𝑧)〉 (8.18) 2 2 𝑖=1 where ?̂?𝑖(𝑧) is the molecular director of molecule 𝑖 computed as the eigenvector associated to the lowest eigenvalue of the instantaneous molecular inertia tensor whose center of mass (COM) is in z direction, I denotes the identity tensor and N(z) is the number of molecules in the bin. Once we diagonalize the tensor, we can obtain the 𝑆(𝑧), the scalar nematic order parameters and ?̂?(𝑧), the nematic director, whose components 𝑛𝑧(𝑧) = |?̂?(𝑧) ∙ ?̂?| and 𝑛𝑥𝑦(𝑧) = |?̂?(𝑧) − (?̂?(𝑧) ∙ ?̂?)?̂?| can be calculated. The Orientational distribution function 𝑃(𝑧, 𝑐𝑜𝑠𝛽) is described by: 𝑃(𝑧, 𝑐𝑜𝑠𝛽) = 〈𝛿(𝑧 − 𝑧𝑖)𝛿(𝑐𝑜𝑠𝛽 − ?̂?𝑖 ∙ ?̂?)〉 (8.19) where 𝛽 is the angle between 5CB molecule and the ?̂? director, while z is the projection of the molecular COM in the z direction. The Collective variables of projection parallel and perpendicular 𝑃∥,⟂(𝑟) is defined as: 𝑃∥(𝑧) = (?̂?0 ∙ (𝑟 − 𝑟0)) (8.20) 𝑃⟂(𝑟) = |(𝑟 − 𝑟0) − (?̂?0 ∙ (𝑟 − 𝑟0))?̂?0| (8.21) 261 where ?̂?0 is the molecular director of central molecule, 𝑟0 is the position of the central molecule and 𝑟 is the position of the second molecule. Figure S8.5 Densities of 5CB, PCH5 and F9, respectively, as a function of temperature. Comparisons between experiments (n = 3) and calculations using the United Atom force field reveal good agreements. 262 Interfacial tension calculations The interfacial tension is calculated with the Kirkwood-Buff formula: 1 𝑃𝑥𝑥 + 𝑃𝑦𝑦 𝛾 = 〈 𝐿𝑧 (𝑃𝑧𝑧 − )〉 (8.22) 2 2 where 𝐿𝑧 is the box length in the z direction, and 𝑃𝑖𝑖 are the diagonal components of the pressure tensor. The long range correction is calculated by: 3𝜋(∆𝜌)2 𝜋2 𝑑 2 𝑑 4 𝛾𝑡𝑎𝑖𝑙 = [1 − ( ) + 3.53 ( ) ] (8.23) 8𝑟2𝑐 6 𝑟𝑐 𝑟𝑐 where 𝑟𝑐 is the cutoff radius used for LJ interactions and the parameters ∆𝜌 and 𝑑 are obtained by fitting the dispersion density: 6 𝑛𝑖(𝑟)𝜌(𝑟) =∑ √𝐶 𝜌𝑖𝑖 𝑉(𝑟); 𝐶𝑖 = 4𝜀 𝜎 ; 𝜌 𝑖 𝑖 𝑖 𝑉(𝑟) = (8.24) 𝑉 𝑖 where 𝑛𝑖(𝑟) is the atomic number of type 𝑖 at position 𝑟, 𝜀𝑖 and 𝜎𝑖 are the LJ parameter of atom 𝑖 and 𝑉 is the volume. The fitting is done using an hyperbolic tangent function: 1 𝑧 𝜌(𝑧) = 𝜌0 + ∆𝜌 𝑡𝑎𝑛ℎ 𝑡𝑎𝑛ℎ ( ) (8.25) 2 𝑑 263 Figure S8.6 Interfacial tensions for 5CB-F9 and PCH5-F9. Snapshots for a experiment using pendant drop method and b simulations using the Buff-Kirkwood formula for interfacial tension measurements. Interfacial tensions from c,e experiments and d,f simulations as a function of temperature of c,d 5CB-F9 and e,f PCH5-F9 interfaces, respectively. To estimate the parameters for the Lennard-Jones potentials between the benzonitrile head or the alkyl tail of LCs and F9, we measure the interfacial tension between benzonitrile and F9 to be 14.3 ± 0.1 mN/m (𝑎𝑖𝑗 = 0.33, 𝑏𝑖𝑗 = 0.03 in simulation), whereas pentane and F9 are miscible above 21 °C (𝑎𝑖𝑗 = 0.11, 𝑏𝑖𝑗 = 0.01). 264 Rapini Papoular coefficient and anchoring angle. Using the equation 8.5 in main text 𝑁(𝑧𝑖) 𝑊(𝑧𝑖 ,𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 ) = −𝑘𝐵𝑇 𝑙𝑛 𝑙𝑛 𝑃(𝑧𝑖 ,𝑐𝑜𝑠 𝑐𝑜𝑠 𝛽 ) +𝑊0 (8.26) 𝐴 The anchoring strength 𝑊𝐴2 (𝑧) and the coefficient 𝑊 𝐴 0 (𝑧) are obtained by minimizing the weighted squared error of the difference between the surface free energy and the Rapini Papoular expression: 2 𝜒2 1 (𝑧𝑖) = ∑𝑐𝑜𝑠𝛽 𝑃(𝑧𝑖 , 𝑐𝑜𝑠𝛽) [𝑊(𝑧𝑖 , 𝑐𝑜𝑠𝛽) −𝑊 𝐴 0 (𝑧) + 𝑊 𝐴(𝑧 )𝑐𝑜𝑠22 𝑖 (𝛽 − 𝛽𝑒𝑞(𝑧𝑖))] (8.27) 2 Figure S8.7 Free Energy maps of 5CB-F9 in the isotropic phase. 265 Figure S8.8 Rapinni-Papoular coefficients (a) and orientational distribution functions (b) for PCH5-F9 mixture at different temperatures. 5. References * This chapter was prepared as a Research Article reporting original research close to submission. All authors developed the concept for the research, and contributed to the writing of the manuscript. X.W. and N.L.A. designed the experiments. X.W. conducted the experiments. G.R.P.L., X.W. and J.J.D.P. designed the simulations. G.R.P.L. conducted the simulations. N.L.A. and J.J.D.P. supervised the research. Adapted with permission from: G. R. Perez-Lemus†, X. Wang†, N. L. Abbott and J. J. de Pablo, Substitution of sub-molecular structure alter surface anchoring of liquid crystals. Close to submission. †Equally contributed. [1] P. G. de Gennes and J. Prost, The Physics of Liquid Crystals (Clarendon Press, 1993). [2] P. S. Drzaic, Liquid Crystal Dispersions.Vol.1 (World Scientific, 1995). [3] M. Peccianti, A. Dyadyusha, M. Kaczmarek, and G. Assanto, Tunable Refraction and Reflection of Self-Confined Light Beams, Nat. Phys. 2, 737 (2006). 266 [4] Y. Izdebskaya, V. Shvedov, G. Assanto, and W. Krolikowski, Magnetic Routing of Light-Induced Waveguides, Nat. Commun. 8, 14452 (2017). [5] Y.-K. Kim, X. Wang, P. Mondkar, E. Bukusoglu, and N. L. Abbott, Self- Reporting and Self-Regulating Liquid Crystals, Nature 557, 539 (2018). [6] R. R. Shah and N. L. Abbott, Principles for Measurement of Chemical Exposure Based on Recognition-Driven Anchoring Transitions in Liquid Crystals, Science (80-. ). 293, 1296 (2001). [7] Y. Liu and K. L. Yang, Applications of Metal Ions and Liquid Crystals for Multiplex Detection of DNA, J. Colloid Interface Sci. 439, 149 (2015). [8] Y. Zou, J. Namkung, Y. Lin, and R. Lindquist, Optical Monitoring of Anchoring Change in Vertically Aligned Thin Liquid Crystal Film for Chemical and Biological Sensor, Appl. Opt. 49, 1865 (2010). [9] Q. Li, Y. Li, J. Ma, D. K. Yang, T. J. White, and T. J. Bunning, Directing Dynamic Control of Red, Green, and Blue Reflection Enabled by a Light-Driven Self- Organized Helical Superstructure, Adv. Mater. 23, 5069 (2011). [10] A. Nych, J. I. Fukuda, U. Ognysta, S. Zumer, and I. Muševic, Spontaneous Formation and Dynamics of Half-Skyrmions in a Chiral Liquid-Crystal Film, Nat. Phys. 13, 1215 (2017). [11] D. W. Berreman, Solid Surface Shape and the Alignment of an Adjacent Nematic Liquid Crystal, Phys. Rev. Lett. 28, 1683 (1972). [12] J. Cheng and G. D. Boyd, The Liquid-Crystal Alignment Properties of Photolithographic Gratings, Appl. Phys. Lett. 35, 444 (1979). [13] J. M. Geary, J. W. Goodby, A. R. Kmetz, and J. S. Patel, The Mechanism of Polymer Alignment of Liquid-Crystal Materials, J. Appl. Phys. 62, 4100 (1987). [14] D. S. Seo, K. I. Muroi, T. R. Isogami, H. Matsuda, and S. Kobayashi, Polar Anchoring Strength and the Temperature Dependence of Nematic Liquid Crystal 267 (5CB) Aligned on Rubbed Polystyrene Films, Jpn. J. Appl. Phys. 31, 2165 (1992). [15] J.-I. Fukuda, M. Yoneya, and H. Yokoyama, Surface-Groove-Induced Azimuthal Anchoring of a Nematic Liquid Crystal: Berreman’s Model Reexamined, Phys. Rev. Lett. 98, 187803 (2007). [16] A. Pizzirusso, R. Berardi, L. Muccioli, M. Ricci, and C. Zannoni, Predicting Surface Anchoring: Molecular Organization across a Thin Film of 5CB Liquid Crystal on Silicon, Chem. Sci. 3, 573 (2012). [17] O. M. Roscioni, L. Muccioli, and C. Zannoni, Predicting the Conditions for Homeotropic Anchoring of Liquid Crystals at a Soft Surface. 4-n-Pentyl-4′- Cyanobiphenyl on Alkylsilane Self-Assembled Monolayers, ACS Appl. Mater. Interfaces 9, 11993 (2017). [18] J. D. Parsons, A Molecular Theory of Surface Tension in Nematic Liquid Crystals, J. Phys. 37, 1187 (1976). [19] M. A. Osipov, T. J. Sluckin, and S. J. Cox, Influence of Permanent Molecular Dipoles on Surface Anchoring of Nematic Liquid Crystals, Phys. Rev. E 55, 464 (1997). [20] B. Tjipto-Margo and D. E. Sullivan, Molecular Interactions and Interface Properties of Nematic Liquid Crystals, J. Chem. Phys. 88, 6620 (1988). [21] M. P. Allen, Molecular Simulation and Theory of the Isotropic-Nematic Interface, J. Chem. Phys. 112, 5447 (2000). [22] M. Sadati et al., Molecular Structure of Canonical Liquid Crystal Interfaces, J. Am. Chem. Soc. 139, 3841 (2017). [23] H. Ramezani-Dakhel, M. Sadati, M. Rahimi, A. Ramirez-Hernandez, B. Roux, and J. J. De Pablo, Understanding Atomic-Scale Behavior of Liquid Crystals at Aqueous Interfaces, J. Chem. Theory Comput. 13, 237 (2017). [24] X. Wang, Y. Zhou, V. Palacio-Betancur, Y.-K. Kim, L. Delalande, M. Tsuei, 268 Y. Yang, J. J. de Pablo, and N. L. Abbott, Reconfigurable Multicompartment Emulsion Drops Formed by Nematic Liquid Crystals and Immiscible Perfluorocarbon Oils., Langmuir 35, 16312 (2019). [25] D. S. Miller, R. J. Carlton, P. C. Mushenheim, and N. L. Abbott, Introduction to Optical Methods for Characterizing Liquid Crystals at Interfaces, Langmuir 29, 3154 (2013). [26] W. Martienssen and H. Warlimont, editors , Springer Handbook of Condensed Matter and Materials Data (Springer Science & Business Media, 2006). [27] M. A. Amat and G. C. Rutledge, Liquid-Vapor Equilibria and Interfacial Properties of n-Alkanes and Perfluoroalkanes by Molecular Simulation, J. Chem. Phys. 132, 114704 (2010). [28] G. Tiberio, L. Muccioli, R. Berardi, and C. Zannoni, Towards in Silico Liquid Crystals. Realistic Transition Temperatures and Physical Properties for n- Cyanobiphenyls via Molecular Dynamics Simulations, ChemPhysChem 10, 125 (2009). [29] M. Kleman and O. D. Lavrentovich, Soft Matter Physics : An Introduction (Springer, New York, NY, USA, 2003). [30] J. Comer, J. C. Gumbart, J. Hénin, T. Lelievre, A. Pohorille, and C. Chipot, The Adaptive Biasing Force Method: Everything You Always Wanted to Know but Were Afraid to Ask, J. Phys. Chem. B 119, 1129 (2015). [31] P. F. Z. Rico and J. J. de Pablo, Sobolev Sampling of Free Energy Landscapes, ArXiv Prepr. arXiv:2202.01876 (2022). [32] Frisch, M. J., Trucks, G. W., Schlegel, H. B., Scuseria, G. E., Robb, M. A., Cheeseman, J. R., Scalmani, G., Barone, V., Petersson, G. A., Nakatsuji, H., Li, X., Caricato, M., Marenich, A., Bloino, J., Janesko, B. G., Gomperts, R., Mennucci, B., Hratchian, H. P., Ortiz, J. V., et al. Gaussian 09, Revision A.02. (2016). 269 [33] Wick, C. D., Martin, M. G. & Siepmann, J. I. Transferable Potentials for Phase Equilibria. 4. United-Atom description of linear and branched alkenes and alkylbenzenes. J. Phys. Chem. B 104, 8008–8016 (2000). [34] Cheung, D. L., Clark, S. J. & Wilson, M. R. Parametrization and validation of a force field for liquid-crystal forming molecules. Phys. Rev. E 65, 051709 (2002). [35] Zhang, J., Su, J. & Guo, H. An atomistic simulation for 4-cyano-4′- pentylbiphenyl and its homologue with a reoptimized force field. J. Phys. Chem. B 115, 2214–2227 (2011). [36] Zhang, L. & Siepmann, J. I. Pressure dependence of the vapor-liquid-liquid phase behavior in ternary mixtures consisting of n-alkanes, it-perfluoroalkanes, and carbon dioxide. J. Phys. Chem. B 109, 2911–2919 (2005). [37] Anderson, J. A., Glaser, J. & Glotzer, S. C. HOOMD-blue: A Python package for high-performance molecular dynamics and hard particle Monte Carlo simulations. Comput. Mater. Sci. 173, 109363 (2020). [38] Lebard, D. N., Levine, B. G., Mertmann, P., Barr, S. A., Jusufi, A., Sanders, S., Klein, M. L. & Panagiotopoulos, A. Z. Self-assembly of coarse-grained ionic surfactants accelerated by graphics processing units. Soft Matter 8, 2385–2397 (2012). [39] Kirkwood, J. G. & Buff, F. P. The statistical mechanical theory of surface tension. J. Chem. Phys. 17, 338–343 (1949). [40] Lundberg, L. & Edholm, O. Dispersion Corrections to the Surface Tension at Planar Surfaces. J. Chem. Theory Comput. 12, 4025–4032 (2016). 270 Chapter 9 Optical Fingerprinting of Dynamic Interfacial Reaction Pathways using Liquid Crystals 1. Introduction Chemical reactions at fluid-fluid interfaces enable diverse syntheses, ranging from the simple conversion of amines and acid chlorides into amides[1] to the synthesis of conducting polymers[2,3], nanoparticle membranes[4–6] and small molecules using coupled chemical and enzymatic catalysts[7]. The products of interfacial reactions form the basis of numerous products in the pharmaceutical[8], electronics[2], food[9] and fine chemicals industries[10]. In this paper, we report that liquid crystals (LCs) can be used to detect and “fingerprint” complex interfacial reactions. We demonstrate this capability using nucleophilic substitution reactions involving a synthetic surfactant and a series of amine-based nucleophiles at aqueous interfaces. When observed using polarized-light optical microscopy, the LCs exhibit complex spatial and temporal optical responses (fingerprints) that correlate to the structures of the nucleophiles. The responses, which arise from non-equilibrium processes, result from dynamic interfacial interactions between the reactants, intermediates, products and the LC. LC phases can amplify local molecular events into macroscopic ordering transitions as a consequence of their hierarchical organization[11,12], thus providing the opportunity to detect chemical reactions at interfaces. The long-range internal organization of molecules within LCs also generates anisotropic optical properties, such that reorientation of the LC can be reported using polarized light[11]. The observation that LCs assume distinct equilibrium orientations in contact with solutions of reactants or products has led to the sensing of a range of chemical reactions and binding interactions, such as enzymatic processing of oligopeptides[13], reversible binding 271 events involving antibodies[14], and complexation of DNA and amphiphiles[15]. However, complex multi-step reaction pathways, which generate multiple intermediates and non-equilibrium interfacial states of LCs, have not been explored. Prior studies have explored LC interfaces decorated with amphiphiles, including synthetic surfactants[16,17], phospholipids[18,19], polyelectrolytes[20] and peptide– polymer amphiphiles[21]. At equilibrium, the presence of amphiphiles at the aqueous interface of the LC anchors the LC into preferred orientations that reflect the lowest energy interfacial state. These states are caused by interactions such as the interdigitation of the aliphatic chains of surfactants and LC molecules[17,19], electrostatic interactions[20] and hydrogen bonding[22]. However, amphiphiles are also capable of generating a range of non-equilibrium states at aqueous interfaces of LCs in part because LCs possess mobilities and dynamic behaviors that are characteristic of liquids[11,12]. The high mobilities lead to rapid responses to external stimuli that drive LCs beyond equilibrium.[17,19] For example, events such as the fusion and spreading of single vesicles at LC interfaces can generate transient lateral concentration gradients of amphiphiles at the LC interface which in turn generate optical flashes[19]. Here we move beyond detection of chemical transformations that involve a single reactant and product to explore reaction pathways involving an amphiphile and a series of model nucleophiles with one to four reaction sites. We sought to determine if the reactions generate dynamic interfacial LC states and corresponding spatio-temporal patterns (fingerprints) that permit reactions involving each nucleophile to be differentiated. The results of our study suggest new approaches for in situ detection of interfacial reactions and hint that LC systems are a fertile source of data for the integration of artificial intelligence (AI) for automated experimentation[23]. 272 2. Results and Discussion Introduction to Reactions and LCs Used in This Work. We used a synthetic surfactant TEA12 (Figure 9.1a, see Scheme S9.1 and Figures S9.1-S9.6 for synthesis) that is comprised of a pre-installed quaternary ammonium moiety (charged hydrophilic head group) and an alkyl-methacrylate moiety (hydrophobic tail)[24]. The LC phase used in the study reported in this paper was formed from 4-cyano-4’-pentylbiphenyl (5CB, Figure 9.1b). The compound 5CB forms a nematic liquid crystalline phase at room temperature[11,25]. Films of 5CB were formed in the pores of cop-per grids (22 m in thickness) supported on glass slides that were treated with octyltrichlorosilane (OTS) to provide a uniform homeotropic anchoring (perpendicular) of the 5CB. Next, the supported LC films were incubated against aqueous solutions of 5 mM TEA12 inside a chamber (Figure S9.12). The incubation process triggered a transition in the Figure 9.1 Molecular structures of (a) tetraalkylammonium-12 (TEA12) and (b) 4-cyano-4’-pentylbiphenyl (5CB). (c) Schematic illustrations and (d) polarized- light micrographs showing that adsorbed TEA12 alters 5CB alignment at the aqueous interface, leading to a switch of optical texture from bright to dark. Red arrows indicate interfacial alignment of 5CB molecules. Scale bars, 100 m. 273 optical appearance of the LC film (observed using polarized-light optical microscopy) from a bright texture to a uniformly dark appearance within 5 min (Figure 9.1c,d). This transition reflects the adsorption of TEA12 onto the LC-aqueous interface, which switches 5CB anchoring from planar (tangential to the water interface) to homeotropic (perpendicular to the interface). In the experiments described below, a concentration of 5 mM of TEA12 was chosen because it induced homeotropic anchoring (Figure 9.1) of the LC at the aqueous interface; lower concentrations (e.g., 1 mM) generated tilted anchoring of the LC (Figure S9.13). The homeo-tropic anchoring (dark) of 5CB set the initial condition of our experiments involving reactions of TEA12 with nucleophiles (as described below). The acrylate moiety of the TEA12 surfactant possesses Michael acceptor properties, which is further aided by the quaternary ammonium moiety as the leaving group[24,26]. The surfactant is susceptible to a formal SN2′ reaction[26], such that an amine-based nucleophile can trigger the release of the charged quaternary ammonium head group (Scheme 1a, one primary amine can react with two surfactant mole- cules[26]). This reaction can alter the amphiphilic properties of TEA12 adsorbates on an oil-aqueous interface leading, for example, to destabilization of oil-in-water emulsions[24]. Whereas past studies focused on the functional properties of TEA12 and its reaction products[24], below we show how LC interfaces can characterize dynamic processes that occur during reactions involving TEA12. We react TEA12 with a series of nucleophiles with one or two reactive amine groups (primary or secondary, Scheme 1b-f, green amine groups are reactive). Possible products of the reaction of TEA12 with each nucleophile are listed in Table 9.1, revealing that a broad range of products are possible. 274 a R NH2 + N R N + N H Br Br O O O O O O 10 10 10 R N O O O O 10 10 Scheme 9.1 (a) TEA12 undergoes an amine-triggered cleavage of its charged head group. Molecular structures of (b) N,N,N′-trimethylethylenediamine (TMEN), (c) N,N-dimethylethylenediamine (DMEN), (d) N,N′- dimethylethylenediamine (DMEDA), (e) N-methylethylenediamine (MEDA) and (f) ethylenediamine (EDA). Characterization of Products Formed from TEA12 and Amines in Bulk Solutions. Prior to investigating the optical responses of LCs to the reactions of TEA12 with the various amine-based nucleophiles in Scheme 9.1, we characterized the products of the reactions. We use the EDA nucleophile as an example with which to describe our procedure. Briefly, EDA was mixed into a methanolic solution of TEA12, resulting in a 1.1 mol equiv. of EDA versus TEA12. We note that methanol was used here because the products of reaction are insoluble in water (see Supporting Information for discussion of the products formed in different solvent environments)[24]. Because EDA has two primary amine groups and can react with four TEA12 molecules, the nucleophilic reaction site -NH to surfactant ratio is 4.4:1. We chose this ratio because, as described below, it gives rapid and distinct optical responses when using LC films (see below). The reactions were determined to occur within 10 min of mixing the 275 reactants but all solutions were incubated for 15 minutes before analysis by electrospray ionization mass spectrometry (ESI-MS) to ensure complete conversion. Table 9.1 Products and their molecular weight formed from TEA12 with an excess amount of different amines, respectively. The ESI mass spectrum of the TEA12-EDA reaction product is shown in Figure 9.2. Peaks corresponding to monomer (5a), dimer (5b,5c), trimer (5d) and tetramer (5e) were all observed in the spectrum. This result indicates that even though excess (4.4- fold) nucleophile was used to trigger the nucleophilic substitution reaction, some EDA 276 molecules were still able to react with multiple TEA12 molecules. This likely reflects both the statistical nature of products resulting from multi-site reactions and the similar or higher reactivity of secondary amines compared to primary amines[26]. Reactions triggered by other nucleophiles were also characterized in methanol (Figures S9.7- S9.10), and similar results were obtained; i.e., all possible product molecules were found in the system. Although the LC optical finger prints are generated by reactions between TEA12 and nucleophiles in an aqueous phase contacting LC and not methanol, we anticipate that the nucleophiles in the LC experiments will also react with multiple TEA12 molecules. Specifically, hydrophobically driven association of TEA12 and reaction products are likely to further promote multi-site reactions in the bulk aqueous phase and at the LC interface (see Figure S9.11 for support). Figure 9.2 ESI mass spectrum of products from reaction of TEA12 with EDA (1.1 equiv., 4.4 equiv. -NH) in methanol. Spectra from several scans were added to generate the spectrum shown. Typically, ions carrying 1 positive charge (+H+ or +Na+) were detected. The peaks in the spectrum are consistent with formation of all possible products of the reaction (5a- 5e). The measured mass is calculated from the m/z of these ions after assigning the charge states. 277 LC Responses Triggered by Reactions of TEA12 with Nucleophiles. After LC films were equilibrated against 5 mM TEA12 solutions to obtain the initial homeotropic anchoring (dark appearance), a nucleophile was added into the aqueous phase to a concentration of 22 mM based on reactive -NH sites (4.4 equiv. -NH to TEA12). The corresponding nucleophile concentrations were 22 mM for TMEN, 11 mM for DMEN and DMEDA, 7.33 mM for MEDA and 5 mM for EDA. Prior to arriving at this experimental design, we surveyed a range of nucleophile concentrations and found that use of a lower nucleophile concentration, e.g., 1.1 equiv. -NH to TEA12, led to long reaction times (1-2 hours and weak LC film optical responses that reflected small changes in the LC tilt angle). Alternatively, substantially higher concentrations, e.g., 11 equiv. -NH to TEA12, led to rapid formation of insoluble products that generated turbidity and obscured observations of the LC films. Overall, 4.4 equiv. -NH generated distinct “fingerprint” patterns of LC responses for each nucleophile, as described below, with responses starting within 15 min of addition of the nucleophile. Experiments with each nucleophile were repeated at least 3 times to ensure the reproducibility of observations (Figures S9.14-S9.18). The LC optical fingerprints generated by each nucleophile, using the conditions described above, are shown in Figure 9.3. The initial time stamp on the first micrograph shown for each nucleophile denotes the time at which the change in optical appearance was first observed (time zero is when the nucleophile was added). For example, TMEN took about 12 minutes to generate an initial optical response (Figure 9.3a) whereas EDA only needed 5 minutes (Figure 9.3c). This result reveals that the onset of reaction does not correlate with the concentration of the reactant molecules (22 mM for TMEN versus 5 mM for EDA), but rather reflects relative reaction rates. The initial response of the LC was observed to correlate with the arrival of a turbid reaction front (formed in the 278 aqueous phase) at the interface of the LC film (Figure S9.12). As discussed above, the turbidity is generated by aqueous-insoluble reaction products. Figure 9.3 Micrographs obtained with crossed polarizers for transmitted light show distinct optical responses of LC film to reactions with different amine nucleophiles in the first column (22 mM TMEN, 11 mM DMEN, 5 mM EDA, 11 mM DMEDA and 7.33 mM MEDA, respectively). Scale bars, 200 m. The inset reveals the two distinct optical patterns with the white lines denoting the key features. For all five nucleophiles, optical patterns characterized by bands of bright interference colors were observed to form subsequent to the above-described onset of the LC responses (Figure 9.3a-e, green frames, also shown in the inset). The appearance of the interference colors (under white light illumination) is indicative of a tilting of the 279 orientation of the LC away from the initial homeotropic alignment at the LC-aqueous interface. We call this a “type-one” response. As detailed below, the change in orientation of the LC is due to an interfacial flow generated by reaction-induced interfacial tension gradients (Marangoni flows)[17,19]. Prior to ad-dressing the mechanism leading to the generation of the interfacial flow (see discussion surrounding Figures 9.4 and 9.5), we first quantify the apparent tilt angle of the LC during the “type- one” response. When using white light illumination, the Michel-Lévy interference color chart can be used to determine the optical retardance ∆𝑟 (i.e., optical path difference between two light rays with orthogonal polarizations) of the LC film. Inspection of Figure 9.3 reveals that the optical retardance varies with position across the films, and here we focus on quantifying the maximum retardance (e.g., as illustrated by the red arrow in the inset shown in the top right corner of Figure 9.3, “type-one”). The interference colors in Figure 9.3, when interpreted using the Michel-Lévy interference color chart, correspond to maximum retardance values of 210±110 nm for TMEN and 1660±80 nm for other nucleophiles (DMEN, EDA, DMEDA, MEDA). To provide a physical interpretation of these retardance values, we considered a LC film confined by two surfaces, one of which causes homeotropic anchoring and the other that anchors the LC away from the surface normal with a tilt of angle 𝜃. For such a LC film, the retardance can be calculated as[17]: 𝑑 𝑛 𝑒 𝑛𝑜 ∆𝑟 ≈ ∫ − 𝑛 𝑜 𝑑𝑧 (1) 0 √𝑛2sin2 𝑧 𝑧 𝑜 ( 𝜃) + 𝑛 2cos2 ( 𝜃) ( 𝑑 𝑒 𝑑 ) where 𝑛o (1.716 for 5CB) and 𝑛e (1.533) are the indices of refraction perpendicular and parallel to the optical axis of the LC, respectively, 𝑑 is the thickness of the film (22 m). Using the values of the maximum optical retardance reported above, we used equation 280 1 to calculate the corresponding maximum values of 𝜃 to be 24±7° for TMEN and 80±3° for all other nucleophiles (90° is planar (tangential) anchoring). We considered it possible that the weak optical response of the LC to the reaction of TEA12 with TMEN may reflect the slow rate of this reaction (as indicated by the time at which the LC optical response was observed). However, the onset of the optical response to the reaction of TEA12 with DMEN is not substantially different from TMEM (12 mins versus 11 min) and DMEN generates a strong optical response (effective change in tilt angle of the LC of 80±3°). Instead, we note that TMEN only generates monomer (1) upon reaction with TEA12 whereas all other nucleophiles can form at least dimers. This result suggests that the small tilt angle (24±7°) observed with TMEN correlates to the formation of monomer (1), whereas large tilt angles (80±3°) are associated with reactions with TEA12 leading to formation of dimers or higher order oligomers (e.g., 2b, 3b). The “type-one” optical transition described above was transient, persisting for 1-2 min for all nucleophiles. With the exception of DMEDA, the end of the “type-one” optical transition was signaled by the LCs resuming a homeotropic orientation (e.g., see Figure 9.3c for EDA at +90 s). This indicates that adsorbates are present on the LC interface at the end of the “type-one” transition (in the absence of adsorbates, the LC will exhibit a bright optical appearance). The dark optical appearance of the LC did not change when the aqueous suspension formed by reacting TEA12 with TMEN or DMEN for 30 mins was diluted, indicating the presence of intermediate or final reaction products that adsorbed irreversibly to the interface (in contrast, TEA12 adsorbs reversibly to the LC interface). Further evidence presented below reveals that the products that cause the return of the homeotropic state are intermediate products of the reaction (not final products). Because we used excess nucleophile (-NH), all TEA12 was ultimately consumed by the reaction, a result that was confirmed by mass 281 spectroscopy (Figures S9.7-S9.10, the peak corresponding to TEA12 at ~354 amu was not observed)[24]. It is possible, however, that some TEA12 still remains on the LC at the end of the type-one response. We also measured the pH value of the product suspensions to be 8.5 to 9, lower than the pKa of the conjugate acids of amines (9.7- 10.7)[27], indicating that the product molecules absorb protons from the solution. With the exception of TMEN and DMEN, the homeotropic orientation observed at the end of the type-one response was not the final state of the LC system. For example, as shown in Figure 9.3c, at +170 s, the LC film responded to the reaction between TEA12 and EDA by exhibiting another tilted state – we refer to this state as a “type- two” tilted state of the LC (Figure 9.3c, red frame; +170 s, also shown in inset). The “type-two’ state was characterized by an optical pattern that was substantially different from the “type-one” pattern. Specifically, a so-called Schlieren texture with bright and dark brushes was observed to radiate from defects (denoted by white arrows in Figure 9.3c-e) present within the LC. Schlieren textures are not typically observed in the presence of interfacial flows but rather reflect the presence of adsorbates uniformly distributed across the interface[11]. For EDA, the Schlieren texture persisted for tens of minutes, after which the LC relaxed again to the homeotropic state. Dilution of the final product suspension with EDA did not change the LC anchoring (irreversible adsorption of products). The conclusion that the type-two transition with EDA is an adsorbate- driven anchoring transition provides additional support for our conclusion that the homeotropic orientation observed following the “type-one” transition is caused by the presence of an interfacial composition that is formed along the reaction pathway but does not correspond to the final composition of the system. A second non-equilibrium (tilted) state of the LC (type-two transition) was also observed when TEA12 was reacted with DMEDA (Figure 9.3d, +61 s). However, it 282 was not observed when using DMEN (Figure 9.3b). While both nucleophiles can form dimer products, a comparison of products between DMEDA and DMEN reveals that the “type-two” transition correlates with the formation of products that result from reactions of TEA12 with two different amines on the nucleophile (3b versus 2b). This is further supported by the existence of the “type-two” responses from EDA and MEDA (Figure 9.3c,e, red frames) that are also capable of forming products on two separated amines (5b, 4c). The re-entrant homeotropic state that separates the type-one and type-two responses of the LC was observed with EDA (Figure 9.3c, +90 s) and MEDA (Figure 9.3e, +49 s) but not DMEDA (Figure 9.3d). A homeotropic state following the type-one response was also observed with DMEN (Figure 9.3b, +91 s). Because DMEN can react with TEA12 on the same amine (2b) whereas DMEDA can only react on separate amine groups (3b), this result suggests that the transient homeotropic state between the type-one and type-two responses involves the presence of adsorbates that contain nucleophiles that have reacted twice on the same amine. This proposal also leads us to predict that EDA reacts first on one primary amine (5c is preferred over NN’-dimer 5b) to generate the transient homeotropic state. Additionally, the relaxation process for DMEDA involved formation of domains (Figure 9.3d, +478s), consistent with the molecular mobility of dimers being higher than trimers and tetramers on the LC interface (see Figure 9.6 and associated discussion in the text below). Finally, MEDA exhibited a series of responses (Figure 9.3e) similar to EDA (Figure 9.3c), indicating that trimeric products (4e, 5d) are not distinguishable from tetramers (5e). Similar to EDA, the optical fingerprints generated by MEDA suggest a reaction pathway that first involves two substitutions on the primary amine group followed by a third substitution on the secondary amine because the optical fingerprint 283 involves that re-entrant homeotropic state (absent for DMEDA). Overall, the spatial and temporal optical responses of LCs during the reaction of TEA12 with the family of nucleophiles in Scheme 1 hint at the potential utility of the approach as the basis of a platform to report on reaction pathways with multiple intermediates and products. Characterization of the Reaction-Induced Interfacial Flow that Generates the “Type-One” Tilted State. Above we proposed that the “type-one” tilted state was generated by an interfacial flow resulting from interfacial tension gradients (Marangoni stress). To provide additional support for this interpretation, we observed the motion of 1-3 m diameter polystyrene microparticles adsorbed at the LC-aqueous interface revealed. These observations revealed that the triangle-shaped pattern evident in Figure 9.4a was associated with an interfacial flow along the grid diagonal (in a direction away from the corner). In contrast, the band-like pattern evident in Figure 9.4b was consistent with an interfacial flow perpendicular to the top side of the grid (as shown in Figure 9.4b). To elucidate the origin of the interfacial flows, we then characterized the interfacial tensions between the LC and aqueous solutions of TEA12 (reactants) or product suspensions (following the reaction of the TEA12 with the nucleophiles). For these experiments, we prepared the product suspensions from 5 mM TEA12 and nucleophiles, respectively, in aqueous solutions at a TEA12:-NH = 1:1 molar ratio. We used this ratio of TEA12 to nucleophile because independent experiments (Figure S9.11) using mass spectroscopy established that these conditions led to fully substituted products as the major products (experiments run at TEA12:-NH = 1:4.4 molar ratio led to similar relative values of interfacial tensions to those shown in Figure 9.4c (see Figure S9.19)). The solutions of TEA12 and nucleophiles were reacted for 2 hours to ensure 284 Figure 9.4 Origin of the “type-one” response. (a,b) Time-lapse micrographs of the motion of polystyrene tracer particles adsorbed at the LC-aqueous interface. The experiments were performed with EDA using the conditions reported in Figure 7.3c. Scale bars, 100 m. (c) Interfacial tensions of 5CB-aqueous interfaces using aqueous solutions of 5 mM TEA12, or product suspensions prepared by reacting 5 mM TEA12 with either 5 mM TMEN, 2.5 mM DMEN, 2.5 mM DMEDA, 1.7 mM MEDA, or 1.25 mM EDA, respectively. The interfacial tensions were measured using the pendant drop method. The dominant species present in the product suspensions are indicated on the bottom axis of the bar chart. complete conversion. Next, a 5-7 L drop of 5CB was dispensed from a metal needle into either water (control), TEA12 or the product suspensions, and equilibrated for 5 min (pendant drop method). The cross-sectional profiles of the 5CB droplets were then analyzed using the Young-Laplace equation[28] to obtain interfacial tensions (Figure 285 9.4c). Product suspensions containing trimer (4e) and tetramer (5e) were turbid, and they were diluted 1:1 vol with water to enable imaging of the 5CB droplets. The interfacial tension between 5CB and aqueous 5 mM TEA12 was determined to be 9.0±0.4 mN/m, whereas the interfacial tensions for product suspensions ranged from 23 to 32 mN/m (Figure 9.4c). The observation that the interfacial tensions of the product suspensions were higher than TEA12 provides support for our conclusion that the type- one optical response of the LC arises from reaction-induced interfacial tension changes (∆𝛾 ) that induce surface flows (Marangoni stress-driven flows[17]) that shear and reorient the LC phase. Dynamic LC-Aqueous Interfacial Tensions in the Presence of Interfacial Reactions. We also monitored the change of interfacial tension between 5CB and an aqueous phase during reactions between TEA12 and nucleophiles using the pendant drop method. We used DMEDA because the suspension formed during the reaction possesses a turbidity that is lower than that of MEDA and EDA and permits imaging of the LC droplets (the system was imaged using a camera and light transmitted through the sample). For this experiment, an aqueous solution of 5 mM TEA12 was hosted in a cuvette, then DMEDA was added to the aqueous solution near the top of the cuvette to reach a final concentration of DMEDA of 2.5 mM (1 equiv. -NH: surfactant). Experiments attempted using 4.4 equiv. -NH: surfactant resulted in intense Marangoni flows that caused LC droplet to detach from the needle during pendant drop measurements of interfacial tension. We confirmed that 2.5 mM DMEDA also generated the type-one (Marangoni- flow driven) response on the LC interface (Figure S9.16). We observed the formation of a reaction front, identified by the reaction-induced turbidity of the aqueous phase (Figure 9.5). The turbid reaction front propagated from 286 Figure 9.5 Dynamic LC-Aqueous Interfacial Tensions in the Presence of Interfacial Reactions. (a,c) Images and (b,d) corresponding schematic illustrations show reaction fronts (a,b) in the absence and (c,d) in the presence of a pendant drop (5 mM TEA12 with 2.5 mM DMEDA). Scale bars, 2 mm. (e) Dynamic interfacial tensions during the process of the reaction front moving past the pendant droplet. the top to the bottom of the cuvette. In the absence of a 5CB interface, the reaction front was flat and moved towards the bottom of the cuvette (Figure 9.5a,b). However, in the presence of a pendant drop of LC, two vortices formed in the aqueous phase adjacent to the drop at the time that the reaction front passed along the LC interface (Figure 9.5c,d). The advective flow near the LC interface was from the bottom of the droplet to the top of the droplet, consistent with a Marangoni flow from locations on the droplet with low interfacial tension (TEA12) to locations with high interfacial tension (products)[17]. 287 Interfacial tensions were estimated by fitting droplet profiles to the Young- Laplace equations (Figure 9.5e)[28] as the reaction front approached the LC droplet. Prior to arrival of the reaction front, the interfacial tensions were measured to be 11.0±0.6 mN/m, consistent with the presence of the TEA12 surfactant (Figure 9.5e, green regime). When the reaction front contacted the interface of the LC droplet (Figure 9.5e, blue regime), the “apparent” interfacial tensions increased to 13.3±1.0 mN/m and remained constant for 10 min until the reaction front moved beyond the location of the droplet. We note that although advective flows can deform the droplet shape during passage of the reaction front (Figure 9.5e, blue regime), the fact that the “apparent” interfacial tension remained unchanged for 10 min in the presence of interfacial advection suggests that there was a continuous refreshing of adsorbates at the interface. This refreshment of the interface, which is equivalent to the “type-one” response, persisted for 2 min for LC films (Figure 9.3) or 10 min for the pendant drop (Figure 9.5). These time scales are long compared to the typical time that it takes to set up a steady- state interfacial flow in response to an imbalance of interfacial tensions (typically less than a second)[29]. This reflects the inherent reactivity and transport of nucleophiles during the reaction with TEA12, revealing that the reaction does occur on the interface (see below for more evidence). Following passage of the reaction front, the product suspension generated an interfacial tension of 25.4±2.8 mN/m (Figure 9.5e, purple regime). Characterization of Adsorbate Product Molecules that Lead to the “Type-Two” Tilted State. To provide additional insight into the origins of the “type-two” transition, product suspensions were prepared by mixing TEA12 and -NH at 1:1 molar ratio for each of the amine-containing nucleophiles, as described above for the interfacial tension 288 Figure 9.6 Relaxation time of aqueous-LC interfaces in different product suspensions with the dominant product species shown on the bottom axis, indicating that the relaxation processes are associated with the adsorption of different products. measurement shown in Figure 9.4c. Next, LC films were submerged into each aqueous product suspension and observed using polarized-light microscopy over a duration of 1 hour. Immediately following contact with the product suspension, we observed the LCs to exhibit optical appearances that were characterized by bright domains, dark brushes and defects. The optical features of the LC films were similar to the “type-two” tilted state observed in Figure 9.3 for EDA, DMEDA and MEDA. The tilted appearance evolved over time, ending with the LC assuming a dark appearance (homeotropic anchoring, Figures S9.20-S9.24). Significantly, the characteristic relaxation time leading to the homeotropic end state varied with the nucleophile (Figure 9.6). TMEN and DMEN relaxed rapidly (~2 min) to the homeotropic orientation. In contrast, DMEDA relaxed over a duration of 15 mins and MEDA and EDA took approximately 30 mins. The long relaxation time when using the final products of reaction of DMEDA, MEDA and EDA (3b, 4e, 5e), each of which have two separate reactive amine groups, correlates with the appearance of the “type-two” tilted states during reactions involving 289 these nucleophiles (Figure 9.3). The products of reaction with TMEN and DMEN, in contrast, are either monomeric (TMEN) or involve a single reactive amine (DMEN). This result provides additional support for our proposal that the “type-two” tilted states are created by the product molecules with substitutions on two separate amine groups and arise because the product molecules equilibrate slowly (possess slow dynamics) at the interface of the LC. Overall, our results reveal that the optical fingerprints generated by each nucleophile reflect a complex series of transport, adsorption and reaction processes that generate non-equilibrium states of the LC interface that are unique to each nucleophile. Additional support for this general conclusion can be found in the results of experiments that we conducted with aqueous solutions of TEA12 and the pre-formed products of reaction of TEA12 and the nucleophiles. These experiments were predicted to give rise to transient compositions of the LC interface that differ from those encountered during in situ reaction between TEA12 and the nucleophiles, and thus to generate distinct optical fingerprints. In one such experiment, we pre-equilibrated LC films with 5 mM TEA12, and then added concentrated suspensions generated by pre-reacting TEA12 and either TMEN, DMEN or EDA (TEA12:-NH at 1:4.4 molar ratio). The final product concentrations were similar to those present in experiments shown in Figure 9.3, but the optical fingerprints observed in the two experiments were distinct (compare Figure 9.3 and Figures S9.25-S9.27). Specifically, the optical response of the LC was observed to be largely independent of the nucleophile used in the experiments shown in Figures S9.25-S9.27, in contrast to the nucleophile-dependent fingerprints shown in Figure 9.3. 290 3. Conclusions The key conclusion of the study reported in this paper is that complex interfacial reactions, which are characterized by the transient presence of multiple reactive intermediate species, can generate dynamic non-equilibrium interfacial states of LC that serve as unique optical fingerprints of the reaction pathway. We determined that the optical fingerprints generated reflect a range of non-equilibrium processes occurring at the LC interface, including reaction-induced interfacial tension gradients as well as adsorbate-induced anchoring transitions triggered by transient intermediates along the reaction pathway. Additionally, we also found that interfacial kinetic properties of product molecules correlated with key features of the optical fingerprints. The interfacial reaction pathways studied in our experiments involved nucleophilic substitution reactions between TEA12 and a family of amine-based nucleophiles. We found that reactions that involved formation of dimers and higher order oligomers initially generated strong Marangoni flows and associated characteristic optical responses in the LC. At longer times, we found characteristic transient states of the LC that correlated with the presence of distinct reactive intermediates, such as dimer intermediates formed by the reaction of TEA12 with either the same amine or two separate amines of the nucleophile. Finally, the optical fingerprints also integrated differences in the dynamics of adsorption of reaction products (monomers faster than specific dimers and trimers). The observations reported in this paper generate a number of interesting questions. For example, we interpret our results to suggest that TEA12 reacts preferentially on the same amine of EDA and MEDA. However, it is likely that the interfacial environment of the reaction (aqueous-LC interface) impacts the reaction pathway and thus such measurements appear to be beyond routine approaches such as 291 nuclear magnetic resonance (NMR), gel permeation chromatography (GPC) and mass spectrometry (MS). Additionally, the results presented in this paper suggest that DMEN and DMEDA possess distinctly different dynamic interfacial tension properties, indicating the measurements of dynamic interfacial tensions provide a facile approach to screening for reaction sites on molecules. Why these two products possess such distinct interfacial dynamics, however, is not understood. Finally, although our results provide strong support for the presence of a reaction-induced interfacial flow for all nucleophiles, additional studies are required to fully understand the mechanism leading to the generation of the interfacial tension gradient. Overall, the results in this study suggest that the spatial and temporal optical response of LCs to complex reaction pathways could be combined with computer vision methods to identify additional features of the response that are not apparent to the human eye. We note that a series of recent studies have used machine-learning methods to characterize LC optical responses in the context of LC sensing[30–32]. The combination of LCs and artificial intelligence (AI) approaches could, for example, be used to explore how various stimuli which are known to impact LC ordering (e.g. electrostatic interactions[33], local magnetic fields[34], interactions with defects[35], designs of enzymatic species[36] and antibodies[37]) impact interfacial reactions. Additionally, our study is focused on nematic phases, but additional opportunities exist to explore how other LC phases impact interfacial reactions, including smectic phases[11], cholesteric phases[38] blue phases[39,40] and polymeric liquid crystalline phases[41]. 4. Supporting Information 4.1. Materials and Methods 292 Materials. All reagents were obtained from commercial sources and used as received without further purification. The nematogen, 4’-pentyl-4-biphenylcarbonitrile (5CB, nematic phase from 18.0 to 35.5 °C) was purchased from HCCH (Jiangsu Hecheng Display Technology Co., LTD). Tetrahydrofuran (THF) was purchased from Acros Organics. Dodecyl acrylate, 1,4-diazabicyclo[2.2.2]octane (DABCO), phosphorous tribromide (PBr3), triethylamine, octyltrichlorosilane (OTS), N,N,N′- trimethylethylenediamine (TMEN), N,N-dimethylethylenediamine (DMEN), N,N′- dimethylethylenediamine (DMEDA), N-methylethylenediamine (MEDA) and ethylenediamine (EDA) were purchased from Sigma-Aldrich (St. Louis, MO). Formaldehyde, methylene chloride, sodium sulfate, hexanes, ethyl acetate, diethyl ether and Lab-Tek Chamber Slide System were purchased from Thermo Fisher Scientific (Tewksbury, MA). Fisher Finest Premium grade glass slides and Fisherbrand disposable cuvettes were purchased from Fisher Scientific (Pittsburgh, PA). Transmission electron microscopy (TEM) grids were purchased from Electron Microscopy Sciences (Hatfield, PA). Purification of water (18.2 MΩ cm resistivity at 25 °C) was performed using a Milli-Q water system (Millipore, Bedford, MA, USA). Surfactant Synthesis. The procedure for synthesis of the tetraalkylammonium- 12 surfactant (TEA12) is outlined in Scheme S9.1 and Figures S9.1−S9.6. Preparation of Octyltrichlorosilane-Coated Glass Slides. Glass slides were cleaned with Alconox detergent using Milli-Q water and ethanol, and then incubated in Nochromix solution (a mixture of Nochromix and 98% sulfuric acid) for 1 day to remove any organic residue on the glass surface. The slides were rinsed with Milli-Q water, and dried under nitrogen gas. Next, the clean glass slides were immersed in 1 vol % OTS in hexane solution for 30 min. Following immersion in the OTS solution, the glass slides were rinsed with Milli-Q water followed by ethanol to remove unreacted OTS from the surface. The glass slides coated with OTS were then dried with nitrogen 293 gas. The quality of the OTS-functionalized glass slides was tested by preparing a thin film of 5CB on the OTS-functionalized glass and in contact with air. A uniform dark appearance under crossed-polarized microscopy indicated that the OTS coating aligned 5CB perpendicular to the glass substrate (homeotropic anchoring). Preparation of Thin Liquid Crystal (LC) Films. Thin films of 5CB were prepared by pipetting 1 μL of 5CB into the pores of 75 mesh (thickness 20 m; lateral pore size 285 μm) TEM grids supported on an OTS-coated glass substrate at 50 °C, which is above 5CB clearing temperature (35.5 °C), to avoid any flow-induced alignment. Excess 5CB was then removed to produce a flat LC film with a thickness of ~22 μm. Detection of Reaction Pathways Using LC Films. The experimental setup is sketched in Figure S9.12. The 5CB film stabilized in a TEM grid was first submerged into TEA12 solution (5 mM, 400 L) in a Lab-Tek chamber and equilibrate for 15 min. During the equilibration process, 5CB films gradually transitioned from a bright optical appearance to a dark optical appearance within the first 2 min when observed between crossed polarizers under microscopy, indicating adsorption of TEA12 surfactant at the LC-aqueous interface. A solution containing nucleophile (4 L, 2200 mM TMEN, 1100 mM DMEN, 1100 mM DMEDA, 733 mM MEDA or 550 mM EDA, prepared based on 2200 mM -NH reaction site, to achieve 4.4 equiv. -NH to TEA12 in the chamber reaction systems) was then added into the left side of the chamber to trigger reaction with TEA- 12 (yielding final concentrations of nucleophiles of 22 mM TMEN, 11 mM DMEN, 11 mM DMEDA, 7.33 mM MEDA or 5.5 mM EDA, respectively). The optical appearance of 5CB films during the reaction was observed using polarized optical microscopy. Measurements of Interfacial Tensions using Pendant Drop Method. For Figure 9.4, TEA12 (5 mM) solutions were incubated with nucleophile solutions at concentrations of 1 equiv. -NH reaction site (5 mM TMEN, 2.5 mM DMEN, 2.5 mM 294 DMEDA, 1.67 mM MEDA or 1.25 mM EDA, respectively) for 2 h to generate targeted product suspensions. Product suspensions from MEDA and EDA, which are opaque, were diluted with 1:1 vol Milli-Q water to allow transmitted light to pass through them. For Figure S9.19, TEA12 (5 mM) was first reacted with nucleophile solutions at concentrations of 4.4 equiv. -NH reaction site (22 mM TMEN, 11 mM DMEN, 11 mM DMEDA, 7.33 mM MEDA or 5.5 mM EDA, respectively) for 1 h to generate targeted product suspensions. No dilutions of products were performed. The interfacial tensions were measured using the pendant drop method performed with an optical tensiometer (Attension Theta T200, Biolin Scientific). Hamilton metal 22 gauge needles (blunt point) and 1.5 mL Fisherbrand disposable cuvettes were used. The equilibration time was 5 min for each drop. The results presented in this paper were obtained using 3-5 independent measurements. The densities used to analyze the pendant drop measurements were 1.0258 g/cm3 (5CB) and 0.9982 g/cm3 (aqueous solutions). The density of 5CB was measured to be 1.025753 ± 0.000073 g/cm3 using an Anton Paar DSA 5000 M density meter at 20 °C. To measure dynamic interfacial tensions (Figure 9.5), pendant drops were first equilibrated with TEA12 (5 mM, 800 L). Next, a nucleophile was added from the solution surface in a cuvette, and observed using the optical tensiometer. Measurements were taken at 2 min time intervals. Measurements of Dynamics of Product Adsorption. TEA12 (5 mM) was first reacted with nucleophile solutions at concentrations indicated in Figures S9.20-S9.24 (5 mM TMEN, 2.5 mM DMEN, 2.5 mM DMEDA, 1.67 mM MEDA or 1.25 mM EDA, based on 1 equiv. -NH) for 2 h to generate targeted product molecules. 5CB films were then submerged into 400 L of the product suspensions and observed under polarized 295 optical microscopy. The results presented in this paper were obtained using three independent measurements. For Figures S9.25-S9.27, each 5CB film was first equilibrated with TEA12 (5 mM, 300 L) to induce homeotropic anchoring. Next, 35 L of a product suspension of TEA12 (50 mM) pre-reacted with nucleophile (220 mM TMEN, 110 mM DMEN, or 55 mM EDA, based on 4.4 equiv. -NH) was added into the TEA12 solution and observed under polarized optical microscopy. Microscopy Observations. An Olympus BX41 microscope with 4× objective, two rotating polarizers, a Moticam 10.0 MP camera and a halogen lamp (Philips 6 V 30 W bulb) for light illumination was used for optical microscopy. Nuclear Magnetic Resonance (NMR). All the 1H NMR and 13C NMR spectra were recorded on a Brucker 400 MHz NMR spectrometer. Electrospray Ionization Mass Spectrometry (ESI-MS). Below we describe procedures used to collect the mass spectra using the EDA nucleophile as an example. 5 mM TEA12 and 550 mM EDA were prepared in methanol. Next, 4 L EDA was mixed into 400 L TEA12 solution, resulting in a 1.1 equiv. EDA versus TEA12 solution. Because EDA has two primary amine groups and can react with four TEA12 molecules, the nucleophile reaction site -NH to surfactant ratio is 4.4:1. We chose this ratio because it generates fast and distinct optical responses in LC films, as stated in the main text. The reactions were measured to be complete within ~2 min in methanol. The solutions were maintained for 15 minutes before further analysis to ensure complete conversion. Mass spectra were recorded on a Bruker AmaZon quadrupole ion trap mass spectrometer coupled with an electrospray ionization source or a Thermo Orbitrap 296 Fusion tribrid (quadrupole, orbitrap, and ion trap) mass spectrometer coupled with Easy nLC 1000 nanoLC system. Discussion on products formed in different environments. We characterized the products formed in methanol using ESI-MS, because both reactants and products are soluble in methanol. The results of ESI-MS reveal that all possible products are found in the methanol system. The LC response experiments were performed in aqueous solutions, where TEA12 forms micelles in the aqueous environments.26 We expect that micelles or interfaces concentrate TEA12, which is likely to promote the formation of high-order products with multiple substitutions. This is supported by results shown in Figure S9.11e,f. 297 2. Analyte TEA12 Synthesis Scheme S9.1 Synthesis of precursors and surfactant TEA12 (1-3). Synthesis of 1: n-Dodecyl acrylate (40 g, 0.166 mol) was added to a round bottom flask and dissolved in tetrahydrofuran (THF, 120 mL) under inert gas. Next, 1,4-diazabicyclo [2.2.2] octane (DABCO, 9.3 g, 0.083 mol) was added dropwise into the solution, and the reaction mixture was stirred for 10 minutes. Formaldehyde (2.5 g, 0.083 mol) was then added dropwise to the reaction mixture and stirred overnight. The solution was then concentrated under reduced pressure, partitioned using methylene chloride and water, and extracted with methylene chloride. The combined extracts were dried over anhydrous Na2SO4 and the crude product was purified by silica gel chromatography using hexane and ethyl acetate as eluents to afford 1 in 17% yield. 1H NMR (CDCl3, 400 MHz, TMS): δ (ppm) = 6.22 (s, 1H), 5.81 (s, 1H), 4.28-4.30 (d, 2H), 4.12-4.15 (t, 2H), 2.78 (broad s, 1H), 1.61-1.68 (m, 2H), 1.23-1.30 (m, 18H), 0.83-0.86 (t, 3H). 13C NMR (125 MHz, CDCl3): δ (ppm) = 166.7, 139.8, 125.7, 65.3, 62.6, 32.2, 29.9, 29.9, 29.8, 29.8, 29.6, 29.5, 28.8, 26.2, 23.0, 14.4. [M+Na]+ from ESI spectroscopy: 293.20879. Synthesis of 2: 1 (7.5 g, 0.027 mol) was added to a round bottom flask and dissolved in methylene chloride (50 mL) under inert gas. Phosphorous tribromide (PBr3, 7.48g, 0.027 mol) was added dropwise at 0 °C and reacted with stir overnight. The solution was then concentrated under reduced pressure, partitioned with methylene chloride and water, and extracted with methylene chloride. The extracts were dried over anhydrous Na2SO4, and the crude product was purified by silica gel chromatography using hexane and ethyl acetate as eluents to afford 2 in 77% yield. 1H NMR (CDCl3, 400 MHz, TMS): δ (ppm) = 6.31 (s, 1H), 5.93 (s, 1H), 4.15-4.21 (m, 4H), 3.71-3.77 (m, 1H), 1.64-1.70 (m, 2H), 1.25-1.32 (m, 18H), 0.85-0.88 (t, 3H). 13C NMR (125 MHz, CDCl3): δ (ppm) = 169.6, 164.9, 137.6, 129.0, 65.9, 65.5, 48.7, 31.9, 30.7, 29.7, 29.6, 29.5, 29.5, 29.4, 29.3, 28.5, 26.0, 22.7, 14.2. [M+Na]+ from ESI spectroscopy: 355.12525. Synthesis of 3 (TEA12): 2 (1.5 g, 0.0045 mol) was dissolved in THF (5 mL) in a round bottom flask. Triethylamine (0.5 g, 0.0049 mol) was then added dropwise, and the solution was stirred overnight. The resulting mixture was concentrated under reduced pressure, and precipitated in diethyl ether 6 times to afford 3 (TEA12) in 66 % yield. 1H NMR (D2O, 400 MHz, TMS): δ (ppm) = 6.87 (s, 1H), 6.56 (s, 1H), 4.22-4.25 (m, 4H), 3.30-3.36 (m, 6H), 1.72-1.75 (m, 2H), 1.30-1.39 (m, 27H), 0.88-0.92 (t, 3H). 13C NMR (125 MHz, D2O): δ (ppm) = 166.5, 140.3, 128.9, 66.2, 55.6, 53.3, 31.9, 29.8, 29.7, 29.6, 29.4, 29.3, 28.2, 25.8, 22.6, 13.9, 7.4. [M]+ from ESI spectroscopy: 354.33692. 298 Figure S9.1 1H-NMR spectrum of precursor 1. 299 Figure S9.2 13C-NMR spectrum of precursor 1. 300 Figure S9.3 1H-NMR spectrum of precursor 2. 301 Figure S9.4 13C-NMR spectrum of precursor 2. 302 Figure S9.5 1H-NMR spectrum of surfactant 3 (TEA12). 303 Figure S9.6 13C-NMR spectrum of surfactant 3 (TEA12). 304 3. Products formed from TEA12 and amine nucleophiles Figure S9.7 ESI mass spectrum of products of the reaction of TEA12 with TMEN (4.4 equiv., 4.4 equiv. -NH) in methanol. Typically ions carrying 1 positive charge (+H+) were detected, suggestive of formation of the monomer product (1). The measured mass is calculated from the m/z of these ions after assigning the charge states. 305 Figure S9.8 ESI mass spectrum of products of the reaction of TEA12 with DMEN (2.2 equiv., 4.4 equiv. -NH) in methanol. Typically ions carrying 1 positive charge (+H+) were detected, appearing to exhibit multiple peaks, suggestive of formation of the monomer and dimer products (2a,2b). The measured mass is calculated from the m/z of these ions after assigning the charge states. 306 Figure S9.9 ESI mass spectrum of products of the reaction of TEA12 with DMEDA (2.2 equiv., 4.4 equiv. -NH) in methanol. Typically ions carrying 1 positive charge (+H+) were detected, appearing to exhibit multiple peaks, suggestive of formation of the monomer and dimer products (3a,3b). The measured mass is calculated from the m/z of these ions after assigning the charge states. 307 Figure S9.10 ESI mass spectrum of products of the reaction of TEA12 with MEDA (1.47 equiv., 4.4 equiv. -NH) in methanol. Typically ions carrying 1 positive charge (+H+ or +Na+) were detected, appearing to exhibit multiple peaks, suggestive of formation of the products from monomer to trimer (4a-4e). The measured mass is calculated from the m/z of these ions after assigning the charge states. 308 309 Figure S9.11 (a-e) ESI mass spectrum of products of the reaction of TEA12 with nucleophiles (1 equiv. -NH) in methanol. (f) The reaction with EDA was performed in aqueous solutions for 2 h, and the products were subsequently extracted into methanol for ESI measurements. Typically ions carrying 1 positive charge (+H+ or +Na+) were detected. A dominant peak was observed on (a-d,f), suggesting that the fully substituted products are the major components in the suspensions from reactions of TEA12 with 1 equiv. nucleophiles. The measured mass is calculated from the m/z of these ions after assigning the charge states. We note that for the reaction with EDA in methanol (e), trimer products dominated, but tetramer products dominated in the aqueous phase (f). 310 4. Fingerprinting of Interfacial Reaction Pathways on Oil-Aqueous Interfaces Using Liquid Crystals Figure S9.12 (a) Schematic illustration of experimental setup. Copper grid-stabilized 5CB film is submerged into TEA12 solution. After equilibration for 15 min, an amine nucleophile is added to the solution from the left side of the copper grid. (b) Visualization of reaction via formation of insoluble products (turbid). Images were taken from the side-view of the chamber with DMEDA nucleophile as an example. Other nucleophiles showed a similar trend. The reaction took place first at the left side near the aqueous surface upon injection of the nucleophile, then expanded near the water surface, also penetrated into the bulk aqueous solution, and eventually filled the entire chamber. Scale bar, 5 mm. 311 Figure S9.13 Micrographs under crossed polarizers show optical textures of 5CB films as a function of TEA12 concentration before introducing nucleophiles. 5 mM TEA12 was used in this work to achieve the initial homeotropic (perpendicular) anchoring. Scale bar, 200 m. 312 Figure S9.14 Polarized light micrographs of three replicated fingerprinting experiments with TMEN. Scale bars, 200 m. 313 Figure S9.15 Polarized light micrographs of three replicated fingerprinting experiments with DMEN. Scale bars, 200 m. 314 Figure S9.16 (a-c) Polarized light micrographs of three replicated fingerprinting experiments with DMEDA. (d) Micrographs obtained using 0.5 equiv. DMEDA (1 equiv. -NH) and surfactant. Scale bars, 200 m. 315 Figure S9.17 Polarized light micrographs of three replicated fingerprinting experiments with MEDA. Scale bars, 200 m. 316 Figure S9.18 (a-c) Polarized light micrographs of three replicated fingerprinting experiments with EDA. (d) Micrographs obtained using 0.25 equiv. EDA (1 equiv. -NH) and surfactant. Scale bars, 200 m. 317 5. Interfacial Tensions of 5CB-Aqueous Product Suspensions Formed from Surfactant and Excess Nucleophiles. Interfacial Tensions 40 30 20 10 0 TMEN DMEN DMEDA MEDA EDA Figure S9.19 Interfacial tensions of 5CB-aqueous interfaces using product suspensions prepared by reacting 5 mM TEA12 with either 22 mM TMEN, 11 mM DMEN, 11 mM DMEDA, 7.33 mM MEDA, or 5.5 mM EDA (4.4 equiv. -NH to surfactant), respectively. The interfacial tensions were measured using the pendant drop method. 318 Interfacial Tensions (mN/m) 6. Adsorption of Products on Liquid Crystal Interface. Figure S9.20 Polarized light micrographs show adsorption of monomer (1, products from 5 mM TEA12 reacted with 5 mM TMEN). The optical retardance decreased from 2100 nm to 0 nm within ~1 min. Scale bar, 200 m. Figure S9.21 Polarized light micrographs show adsorption of dimer (2b, products from 5 mM TEA12 reacted with 2.5 mM DMEN).The optical retardance decreased from 2100 nm to 0 nm within ~1 min. Scale bar, 200 m. 319 Figure S9.22 Polarized light micrographs show adsorption of NN’-dimer (3b, products from 5 mM TEA12 reacted with 2.5 mM DMEDA). The optical retardance decreased from 2100 nm to 0 nm within ~15 min. Scale bar, 200 m. Figure S9.23 Polarized light micrographs show adsorption of trimer (4e, products from 5 mM TEA12 reacted with 1.67 mM MEDA). The optical retardance decreased from 2100 nm to 0 nm within ~30 min. Scale bar, 200 m. 320 Figure S9.24 Polarized light micrographs show adsorption of tetramer (5e, products from 5 mM TEA12 reacted with 1.25 mM EDA). The optical retardance increased (to 2100 nm) in the first 5 min and decreased for the next 25 min (from 2000 nm to 0 nm). Scale bar, 200 m. 321 Figure S9.25 Polarized light micrographs of LC films during adsorption of products from TEA12-TMEN (4.4 equiv., 4.4 equiv. -NH) reaction; the LC films were pre-equilibrated against TEA12 to yield a final product concentration close to that present in the experiments in Figure 7.3. Scale bar, 200 m. Figure S9.26 Polarized light micrographs of LC films during adsorption of products from TEA12-DMEN (2.2 equiv., 4.4 equiv. -NH) reaction; the LC films were pre-equilibrated against TEA12 to yield a final product concentration close to that used in Figure 9.3. Scale bar, 200 m. Figure S9.27 Polarized light micrographs of LC films during adsorption of products from TEA12-EDA (1.1 equiv., 4.4 equiv. -NH) reaction; the LC film was pre-equilibrated with TEA12 to yield a final product concentration close to that present in Figure 9.3. Scale bar, 200 m. 322 Acknowledgement We acknowledge the support of this research from the Army Research Office through W911NF-15-1-0568. 5. References *This chapter was prepared as a Research Article reporting original research submitted to the journal Langmuir. My contribution to this project was in designing and performing the project, experiments, characterizations and writing the manuscript. My co-authors contributed to the designing of the project, synthesis of TEA12, and writing of the manuscript. Adapted with permission from: X. Wang†, J. Krishna†, A. Fernandez, S. Thayumanavan and N. L. Abbott, Optical Fingerprinting of Dynamic Interfacial Reaction Pathways using Liquid Crystals. Submitted. Copyright 2023 American Chemical Society. †Equally contributed. [1] M. B. Smith, March’s Advanced Organic Chemistry: Reactions, Mechanisms, and Structure. (John Wiley & Sons, 2013). [2] H. D. Tran, D. Li, and R. B. Kaner, One-Dimensional Conducting Polymer Nanostructures: Bulk Synthesis and Applications, Adv. Mater. 21, 1487 (2009). [3] F. Zhang, J. Fan, and S. Wang, Interfacial Polymerization: From Chemistry to Functional Materials, Angew. Chemie, Int. Ed. 59, 21840 (2020). [4] Y. Lin, H. Skaff, A. Böker, A. D. Dinsmore, T. Emrick, and T. P. Russell, Ultrathin Cross-Linked Nanoparticle Membranes, J. Am. Chem. Soc. 125, 12690 (2003). [5] P. Arumugam, D. Patra, B. Samanta, S. S. Agasti, C. Subramani, and V. M. 323 Rotello, Self-Assembly and Cross-Linking of FePt Nanoparticles at Planar and Colloidal Liquid-Liquid Interfaces, J. Am. Chem. Soc. 130, 10046 (2008). [6] Y. Jeong, Y.-C. Chen, M. K. Turksoy, S. Rana, G. Y. Tonga, B. Creran, A. Sanyal, A. J. Crosby, and V. M. Rotello, Tunable Elastic Modulus of Nanoparticle Monolayer Films by Host-Guest Chemistry, Adv. Mater. 26, 5056 (2014). [7] C. A. Denard, H. Huang, M. J. Bartlett, L. Lu, Y. Tan, H. Zhao, and J. F. Hartwig, Cooperative Tandem Catalysis by an Organometallic Complex and a Metalloenzyme, Angew. Chemie, Int. Ed. 53, 465 (2014). [8] J. Fischer and C. R. Ganellin, Analogue-Based Drug Discovery (Weinheim: Wiley-VCH, 2006). [9] H. L. Constant, G. A. Cordell, and D. P. West, Nonivamide, a Constituent of Capsicum Oleoresin, J. Nat. Prod. 59, 425 (1996). [10] F. Rudroff, M. D. Mihovilovic, H. Gröger, R. Snajdrova, H. Iding, and U. T. Bornscheuer, Opportunities and Challenges for Combining Chemo- and Biocatalysis, Nat. Catal. 1, 12 (2018). [11] P. G. de Gennes and J. Prost, The Physics of Liquid Crystals (Clarendon Press, Oxford, 1993). [12] D. Demus, J. Goodby, G. W. Gray, H. ‐W. Spiess, and V. Vill, Handbook of Liquid Crystals (Wiley-VCH, Weinheim, Germany, 1998). [13] J.-S. Park and N. L. Abbott, Ordering Transitions in Thermotropic Liquid Crystals Induced by the Interfacial Assembly and Enzymatic Processing of Oligopeptide Amphiphiles, Adv. Mater. 20, 1185 (2008). [14] A. Concellón, D. Fong, and T. M. Swager, Complex Liquid Crystal Emulsions for Biosensing, J. Am. Chem. Soc. 143, 9177 (2021). [15] A. D. Price and D. K. Schwartz, DNA Hybridization-Induced Reorientation of Liquid Crystal Anchoring at the Nematic Liquid Crystal/Aqueous Interface, J. Am. 324 Chem. Soc. 130, 8188 (2008). [16] Y.-K. Kim, X. Wang, P. Mondkar, E. Bukusoglu, and N. L. Abbott, Self- Reporting and Self-Regulating Liquid Crystals, Nature 557, 539 (2018). [17] S. Roh, M. Tsuei, and N. L. Abbott, Using Liquid Crystals for in Situ Optical Mapping of Interfacial Mobility and Surfactant Concentrations at Flowing Aqueous- Oil Interfaces, Langmuir 37, 5810 (2021). [18] D. A. Paterson, P. Bao, R. H. Abou-Saleh, S. A. Peyman, J. C. Jones, J. A. T. Sandoe, S. D. Evans, H. F. Gleeson, and R. J. Bushby, Control of Director Fields in Phospholipid-Coated Liquid Crystal Droplets, Langmuir 36, 6436 (2020). [19] M. Tsuei, M. Shivrayan, Y. K. Kim, S. Thayumanavan, and N. L. Abbott, Optical “Blinking” Triggered by Collisions of Single Supramolecular Assemblies of Amphiphilic Molecules with Interfaces of Liquid Crystals, J. Am. Chem. Soc. 142, 6139 (2020). [20] J. Zou, T. Bera, A. A. Davis, W. Liang, and J. Fang, Director Configuration Transitions of Polyelectrolyte Coated Liquid-Crystal Droplets, J. Phys. Chem. B 115, 8970 (2011). [21] C. D. Ma, L. Adamiak, D. S. Miller, X. Wang, N. C. Gianneschi, and N. L. Abbott, Liquid Crystal Interfaces Programmed with Enzyme-Responsive Polymers and Surfactants, Small 11, 5747 (2015). [22] R. R. Shah and N. L. Abbott, Principles for Measurement of Chemical Exposure Based on Recognition-Driven Anchoring Transitions in Liquid Crystals, Science 293, 1296 (2001). [23] T. Caelli and W. F. Bischof, Machine Learning and Image Interpretation (Springer Science & Business Media, 1997). [24] A. Fernandez, C. A. Zentner, M. Shivrayan, E. Samson, S. Savagatrup, J. Zhuang, T. M. Swager, and S. Thayumanavan, Programmable Emulsions via 325 Nucleophile-Induced Covalent Surfactant Modifications, Chem. Mater. 32, 4663 (2020). [25] X. Wang, Y. Zhou, V. Palacio-Betancur, Y.-K. Kim, L. Delalande, M. Tsuei, Y. Yang, J. J. de Pablo, and N. L. Abbott, Reconfigurable Multicompartment Emulsion Drops Formed by Nematic Liquid Crystals and Immiscible Perfluorocarbon Oils, Langmuir 35, 16312 (2019). [26] J. Zhuang, B. Zhao, X. Meng, J. D. Schiffman, S. L. Perry, R. W. Vachet, and S. Thayumanavan, A Programmable Chemical Switch Based on Triggerable Michael Acceptors, Chem. Sci. 11, 2103 (2020). [27] H. K. Hall Jr, Correlation of the Base Strengths of Amines, J. Am. Chem. Soc. 79, 5441 (1957). [28] P. C. Hiemenz and R. Rajagopalan, Principles of Colloid and Surface Chemistry (CRC press, Boca Raton, 1997). [29] H. Manikantan and T. M. Squires, Surfactant Dynamics: Hidden Variables Controlling Fluid Flows, J. Fluid Mech. 892 (2020). [30] A. D. Smith, N. Abbott, and V. M. Zavala, Convolutional Network Analysis of Optical Micrographs for Liquid Crystal Sensors, J. Phys. Chem. C 124, 15152 (2020). [31] S. Jiang, J. Noh, C. Park, A. D. Smith, N. L. Abbott, and V. M. Zavala, Using Machine Learning and Liquid Crystal Droplets to Identify and Quantify Endotoxins from Different Bacterial Species, Analyst 146, 1224 (2021). [32] N. Bao, S. Jiang, A. Smith, J. J. Schauer, M. Mavrikakis, R. C. Van Lehn, V. M. Zavala, and N. L. Abbott, Sensing Gas Mixtures by Analyzing the Spatiotemporal Optical Responses of Liquid Crystals Using 3D Convolutional Neural Networks, ACS Sensors (2022). [33] P. S. Noonan, R. H. Roberts, and D. K. Schwartz, Liquid Crystal Reorientation Induced by Aptamer Conformational Changes, J. Am. Chem. Soc. 135, 5183 (2013). 326 [34] M. I. Boamfa, M. W. Kim, J. C. Maan, and T. Rasing, Observation of Surface and Bulk Phase Transitions in Nematic Liquid Crystals, Nature 421, 149 (2003). [35] I. Lin, D. S. Miller, P. J. Bertics, C. J. Murphy, J. J. de Pablo, and N. L. Abbott, Endotoxin-Induced Structural Transformations in Liquid Crystalline Droplets, Science 332, 1297 (2011). [36] I. H. Lin, L. S. Birchall, N. Hodson, R. V. Ulijn, and S. J. Webb, Interfacing Biodegradable Molecular Hydrogels with Liquid Crystals, Soft Matter 9, 1188 (2013). [37] A. Concellón, C. A. Zentner, and T. M. Swager, Dynamic Complex Liquid Crystal Emulsions, J. Am. Chem. Soc. 141, 18246 (2019). [38] N. Tamaoki, Cholesteric Liquid Crystals for Color Information Technology, Adv. Mater. 13, 1135 (2001). [39] M. A. Gharbi, S. Manet, J. Lhermitte, S. Brown, J. Milette, V. Toader, M. Sutton, and L. Reven, Reversible Nanoparticle Cubic Lattices in Blue Phase Liquid Crystals, ACS Nano 10, 3410 (2016). [40] Y. Yang, Y.-K. Kim, X. Wang, M. Tsuei, and N. L. Abbott, Structural and Optical Response of Polymer-Stabilized Blue Phase Liquid Crystal Films to Volatile Organic Compounds, ACS Appl. Mater. Interfaces 12, 42099 (2020). [41] A. Ciferri, editor, Polymer Liquid Crystals (Elsevier, 2012). 327 Chapter 10 Active Capturing and Sorting of Microplastics Using Liquid Interfaces 1. Introduction Microplastics (small plastic pieces < 5 mm) in the environment are a global issue requiring urgent intervention[1–3]. Small plastics are more likely to be uptaken by organisms[4–7], which can cause immune responses[5,8], transfer along the food chain[9,10], blockage of the gastrointestinal tract of marine animals[1,11], and difficulty in collection using methods such as filtration[12]. Nonetheless, the larger surface-area-to-volume ratio of microplastics[13] provides an opportunity to collect them using interfaces. As water surfaces are a major location for microplastic accumulation[2], here we propose a separation principle of “surface liquid extraction”, and employ it to collect microparticles on water surfaces using oil interfaces. A carrier species, e.g., decanol, polypropylene glycol (PPG) or naturally occurring fatty acids (oleic acid, linoleic acid), is dispensed on the water surface with a drop of vegetable oil as a sink that continuously absorbs the carrier. The absorption of the carrier by the sink generates sustained surface fluxes that “actively” skim the water surface due to Marangoni stress[14], which carries and collects polymeric microparticles presented on the surface. This method also simultaneously sorts polymeric particles into distinct locations of the oil sink based on their interfacial properties. These findings provide a platform that uses interfaces to address major microplastics concerns. Interfaces separate two phases of matter, which form when molecules within one phase primarily attract each other versus molecules in the other phase[13]. The phenomena are described macroscopically by interfacial/ surface free energy (interfacial/ surface tension). Adsorbates such as particles (Pickering emulsions[15]) 328 that attach onto the interfaces can decrease the interfacial free energy by 109 kBT (see Methods), which means that at equilibrium the particles are irreversibly anchored. The presence of surface tension gradient ∇𝛾, the so-called Marangoni stress[14,16–21], on the fluid-air interface spontaneously pulls fluid from a low tension region to high tension region. In steady states, the flow velocity 𝑢∗ can be evaluated by[17] 𝜂𝑢 𝑧−1∗ 𝜈 ≈ (𝛾w − 𝛾 −1 s)𝑙∗ (1) where 𝜂 is the dynamic viscosity of the liquid (1 mPa s for water), 𝑙∗ is the distance over which velocity gradients are established,, 𝑧𝜈 ≈ (𝜈𝑙 𝑢 −1)1/2∗ ∗ is the thickness of a viscous boundary layer of surface, with 𝜈 = 𝜂/𝜌 the kinetic viscosity, 𝜌 is the density of the fluid (~1000 kg/m3 for water), 𝛾w (~73 mN/m) is the surface tension of water, 𝛾s is the surface tension of e.g., a surfactant solution that decreases 𝛾. Hence, a fluctuation in the surface tension of water (~10 mN/m) can potentially generate a flow velocity up to meters per second as reported[17,19]. Here, leveraging the Marangoni effect beyond equilibrium and microparticles trapping enabled by interfaces, we establish that the creation of continuous surface flow, which is maintained by consumption of chemical energy, can “actively” transport and concentrate microplastics to a sink. To satisfy the balance between interfacial tensions between solid particles and different fluids, particles of different species were sorted to distinct locations. The physical principle contributes to the design of microplastic recycling and reuse, and can potentially apply to other scenarios that require fast collection and separation of materials from interfaces. 2. Results and Discussion To explore the feasibility of using Marangoni flow to collect microplastics on water surface, we designed a system that is comprised of a “carrier” species, decanol, polypropylene glycol (PPG) or fatty acids, and a vegetable oil sink (dyed with < 0.1% Sudan II, Figure 10.1 and Extended Data Figure 10.1-10.3). Decanol and PPG (4000 329 molecular weight) are insoluble in water, but miscible with vegetable oil even at 1:1 v ratio. We used this setup to actively transport model microplastics, namely polyethylene (PE, two sizes: 1–4 m and 10–27 m), polystyrene (PS, 14–20 m) and polytetrafluoroethylene (PTFE, 1 m), that were located at the water surface. The chemicals were added into the system with following order for easy visualization: water, carrier, microparticles and vegetable oil. Microparticles were also added as tracers after the initial particles were collected. Another order of addition, such as microparticles, sink and carrier, further promotes the transport of microplastics due to the initial fast spreading of carrier. Figure 10.1 used decanol as the carrier, a majority of which formed a droplet floating on the water surface, from which a layer of decanol spreads due to a positive spreading coefficient 𝑆 = 𝛾w − (𝛾d + 𝛾dw) with 𝛾d = 28.7 ± 0.2 mN/m, the decanol surface tension and 𝛾ow = 8.4 ± 0.2 mN/m, the decanol-water interfacial tension. In the absence of decanol, vegetable oil spreads on water surface, similarly, due to a positive 𝑆 = 𝛾w − (𝛾o + 𝛾ow), 𝛾o = 31.9 ± 0.5 mN/m, the oil surface tension, and 𝛾ow = 19.4 ± 1.6 mN/m, the oil-water interfacial tension. The presence of the decanol layer decreases the surface tension of water from[22] ~73 mN/m to ~35 mN/m, leading to a negative spreading coefficient of vegetable oil. The vegetable oil sink hence also formed a droplet (Figure 10.1a,b), similar to a so-called herding process that remediates marine crude oil spill by adding surfactants on water surface[23,24]. The vegetable oil sink extracted decanol from the water surface continuously, which created a surface flux of decanol that carried microplastics/aggregates of 1–4 m PE particles toward the vegetable oil sink (Figure 10.1a,b). We note that microparticles were slowly repelled by the decanol source before the dispense of the oil sink. All of the aggregates from 1–4 m PE particles initially on the water surface were eventually collected at oil-air and oil-water interfaces or oil-water-air three phase contact line (inset in Figure 10.1b). Microscopic observations of this process are shown in Supplementary 330 Videos 2–5 in the original paper with different polymeric particles (PE 10–27 m, PS 14–20 m, PE 1–4 m and PTFE), respectively. After ~20 min, the oil droplet expanded, indicative of an increase of water surface tension that led to wetting of the vegetable oil on the water surface (Extended Data Figure 10.1). This led us to conclude that decanol was completely removed from the water surface. We quantified the time needed to transport PE aggregates (200-1000 m) over a distance from their initial location to the oil sink (Figure 10.1c) in the macroscopic setup shown in Figure 10.1b. We note that smaller aggregates of ~10 m exhibited similar behavior to the large aggregates with decanol, as discussed later in Figure 10.2f. Our results revealed an average velocity of the order of 1 mm/s over the distance of 5 cm, 109 times faster compared to diffusion of a 10 m particle (Methods, diffusion coefficient 𝐷 = 2.2 × 10−14 m2/s based on the Stokes–Einstein equation[13]) and 1010 times faster for a 100 m particle (𝐷 = 2.2 × 10−15 m2/s). Figure 10.1 Design of this work: surface liquid extraction to collect microplastics. a, Schematic illustration of the concept of surface liquid extraction. A carrier species travels from a carrier source along the water interface toward a sink, which carries 331 microplastics by creating a surface flow. Experiments were done in a Petri dish with 9 cm diameter. b, Image of an experiment with carrier: decanol (denoted by red circle), model microplastics: 200-1000 m aggregates formed from 1–4 m PE particles that are visible in this macroscopic view (denoted by white arrows) and sink: vegetable oil (orange droplet, dyed with Sudan II). Green dashed lines indicate trajectories of particles. Inset show microplastics were collected on the vegetable oil interfaces. Brightness and contrast were adjusted for easy visualization. c, Time needed to transport microplastics as a function of distance from their initial location to the sink, indicative of an average velocity that is at least 109 folds compared to diffusion of a 10 m particle. For comparison, we used PPG as a carrier species. Unlike decanol that formed a droplet at the water surface, PPG spread on the surface and formed a ~100 nm-thick film (Extended Data Figure 10.2). During extraction of PPG by the vegetable oil droplet, the PPG film broke into discontinuous domains (Extended Data Figure 10.3). The extraction of PPG provided an overall velocity of 0.2–0.4 mm/s (Figure 10.2), slower than that with decanol (Figure 10.1). In addition, naturally occurring fatty acids, oleic acid and linoleic acid, were also used as carriers to collect the microplastics (Supplementary Videos 6,7 in the original paper), which exhibited similar behaviors as decanol. To facilitate an understanding of the extraction process of the carrier species, we compare our results to liquid-liquid extraction[25,26]. Liquid-liquid extraction is a separation method involving solute partitioning between two immiscible liquids (Figure 10.2a)[25,26]. For example, one solute can preferentially transfer from aqueous environment into oil due to a chemical potential difference (i.e., higher solubility of solute in oil versus water). Similarly, in our experiments the carrier was continuously transferred into a vegetable oil that is immiscible with water, thus providing a sustained surface flow (Figure 10.2b). However, the carrier in our experiments was initially dispersed on the water surface instead of the bulk water phase, so we refer to the separation process described in this paper as surface liquid extraction. Similar to the liquid-liquid extraction, the driving force for carrier transfer is the high affinity of the 332 carrier to the oil sink. We evaluated a so-called transfer coefficient 𝐾 based on the Nernst distribution law, a constant describing the concentration 𝐶 ratio (or estimated by solubilities) of the carrier between vegetable oil and water surface at equilibrium[25,26]. For example, because decanol was completely removed from the water surface as ⦵ 𝐶oil/𝐶 ⦵ described above, we evaluated 𝐾 = oil⦵ = ∞ , with 𝐶 3 𝐶 /𝐶 oil (1 mol/m ) and surface surface 𝐶⦵surface (1 mol/m 2) the standard concentrations of decanol in bulk oil and on water surface, respectively. Figure 10.2 Factors that influence transport of microparticles. Transport of carrier: a, Schematic illustration of liquid-liquid extraction, indicating a solute (denoted by orange dots) is partitioning from aqueous phase into oil phase, in contrast to b, our results of surface liquid extraction. b, A micrograph (top) shows particle (blue arrows) flux from far field on the water surface toward the vegetable oil sink, indicative of the extraction of the carrier species (orange dots with tails in bottom schematic illustration) that is transported into the oil sink. c, Particle velocity with different carrier species, decanol versus PPG, show distinct trends as a function of time (velocity data obtained at a distance of 10–20 mm away from the oil sink in macroscopic view). d, Interfacial tensions of oil-water and oil-air interfaces, respectively, in the absence (control) and in the presence of decanol or PPG dissolved in the vegetable oil. 333 PPG is interfacial active on the oil-water interface. Size of particle aggregates: e, Micrographs show distance that particle aggregates with different sizes being transported in 5 s (1–4 m PE particles, PPG). f, Average particle (aggregate) velocities for two groups of sizes with PPG or decanol, respectively. Species of particles: g, Velocity of singly dispersed PE (10–27 m) and PS (14–20 m) particles with PPG. h, Micrographs of PE and PS particles pinned at the water-air interface. Inspired by the theory for kinetics of liquid-liquid extraction[25,26], we describe the kinetics of our surface liquid extraction process of carriers as a function of the solvation reaction occurring in the system and the rates of transport/diffusion of the carrier species. Specifically, three steps, (i) Marangoni transport on water surface, (ii) absorption through water-oil interface and (iii) diffusion in vegetable oil, were involved (Figure 10.2b). Estimated from equation (1), a surface flow velocity of 5 mm/s reflects an interfacial tension gradient of ~0.1 mN/m, which is small[17–19] compared to water surface tension difference ~30 mN/m between in the absence and in the presence of carriers. This interfacial tension gradient is set by the absorption process of carriers into the oil. Hence, the transport of carrier toward sink under Marangoni stress is not a rate limiting step of the extraction process. In a steady state, the concentration of the carrier on the water surface near the oil sink 𝐶surface was considered a constant. Second, the absorption process involved change of solvation environment of the carrier from water surface to bulk oil. Because the absorption of decanol or PPG into oil is a simple redistribution of noncharged components (no ionization or other chemical reactions involved) with large values of the transfer coefficient 𝐾 , we treated the 𝑘 solvation as first-order irreversible reactions[25]: carrier + oil → [carrier + oil] , ∂(𝐶oil/𝐶 ⦵ ) where 𝑘 is the reaction rate constant. The flux was evaluated by 𝐽 = oil = ∂𝑡 𝑘(𝐶 ⦵surface/𝐶surface) = constant. Therefore, if the absorption process is the rate limiting step (so-called kinetic regime[26]), we predicted that the particle velocity as a function of time at one location is a constant. The particle velocity (Methods, at 10–20 mm away 334 from the oil sink in macroscopic view) was characterized as a constant over time from the dispense of the oil sink with PPG (Figure 10.2c), indicative of an extraction process for PPG that is close to absorption limited. Third, the diffusion of carrier away from the oil-water interface into bulk oil also controls the extraction kinetics. Based on Fick’s first law of diffusion[25], and assuming ∂𝐶 a linear concentration profile within the diffusion thickness 𝛿 , 𝐽 = −𝐷 = ∂𝑥 𝐷 𝐷 (𝐶∗oil−𝐶oil), where 𝐷 is the diffusion constant, is a constant in steady state, 𝐶 ∗ 𝛿 𝛿 oil is the bulk concentration of carrier near the oil-water interface, which equilibrates with 𝐶 ∗surface , 𝐶oil > 𝐶oil . As 𝐶oil builds up in the vegetable oil sink, we predict that the velocity decreases over time if the extraction is diffusion limited (so-called diffusion transport regime[26]). Our results with decanol are consistent with this prediction (Figure 10.2c), in contrast to PPG. In addition, we measured the interfacial tensions between vegetable oil and water or air, respectively, using the pendant drop method (Figure 10.2d). The influence on the oil-water interfacial tensions with decanol in oil was not observed, whereas the presence of PPG reduced the interfacial tension to a half. These results revealed that PPG exhibits affinity for the oil-water interface in contrast to decanol (desorption of PPG from the interface is slow), which is consistent with the results that the extraction of PPG was in the kinetic regime. Both PPG and decanol were not surface active for the oil-air interface. We also quantified the influence of the size of particle aggregates on particle velocity. With PPG, particle velocity of PE (1–4 m) aggregates decreased as the size of aggregates increased (Figure 10.2e,f). Figure 10.2f shows the particle velocity near the oil sink (1–2 mm away under microscope, 4–6 min from the dispense of the oil sink) was 0.38 ± 0.12 mm/s for 20–100 m aggregates, whereas 0.23 ± 0.12 mm/s for 300– 500 m aggregates. In contrast, the particle velocity with decanol did not show 335 significant difference for the two groups of aggregates (Figure 10.2f). We speculated that because the interfacial tension between decanol and water is higher than PPG (Extended Data Figure 10.5), decanol exhibits a higher affinity to PE, so that PE particles experience less influence from e.g., water. We also characterized the particle velocities for different microplastic species with the carrier PPG (Figure 10.2g and Extended Data Figure 10.4). For example, singly dispersed PE particles (10–27 m) exhibited a higher velocity (0.40 ± 0.12 mm/s, measured 1–2 mm away from the vegetable oil sink under microscope) than singly dispersed PS particles (14–20 m, 0.23 ± 0.09 mm/s). We speculated this difference is due to the drag from water. Figure 10.2h shows PE and PS particles pinned at the water surface. The two types of particles exhibited different 𝜃wa, angle defined by the water- air interface and the particle-water interface, and hence a distinct surface area of particle submerged in water. 𝜃wa is determined by three interfacial tensions between particle, water and air[15,27], 𝛾sw, 𝛾s and 𝛾w (Methods), 𝛾sa − 𝛾sw = 𝛾w cos 𝜃wa. We evaluated the drag from water acting on the particles based on the Stokes’ law for sphere[13]: 𝐹drag = 6𝜋𝜂𝑟′𝑣, where 𝜂 is the dynamic viscosity of water, 𝑟′ is the equivalent radius estimated from the surface area of the particles that were submerged in water (Methods), 𝑣 is the particle velocity. We estimated that 𝑟′PS/𝑟′PE =1.35, assuming similar driving force from extraction of PPG, 𝑣PE/𝑣PS ~ 1.35, consistent with the particle velocity results (Figure 10.2g). PS and PE particles were found to collect at different locations in the oil sink (Figure 10.3a-g). PS particles gathered at and near the oil-water-air three phase contact line (Figure 10.3a left,b,d,f), whereas PE were largely collected at the oil-air interface (Figure 10.3a right,c,e,g). These gathering locations can be understood by balancing six interfacial tensions of the particle-air interface, 𝛾s, the particle-water interface, 𝛾sw, the particle-oil interface, 𝛾so, the water-air interface, 𝛾w, the oil-air interface, 𝛾o, and the 336 oil-water interface, 𝛾ow , at the three contact points in equilibrium (Figure 10.3h, Methods and Extended Data Figure 10.5). For PS, each triad forms a triangle (every one of the interfacial tensions is smaller than the sum of the other two, Extended Data Figure 10.5). When the six interfacial tensions formed a tetrahedral (Figure 10.3h), PS particles are collected on the boundary of the three phases. In contrast, for PE, 𝛾sw (40.0 ± 1.1 mN/m) > 𝛾so (16.7 ± 1.9 mN/m) +𝛾ow (19.4 ± 1.6 mN/m), the tetrahedral collapses into a triangle. Physically, the oil therefore wets the PE-water interface completely, leading to the collecting of PE on the oil-air interface. Later in Figure 10.4, we also show that PS particles with sulfate-modified surface were collected at the water-oil interface. Overall, this method can sort microparticles based on their interfacial properties at equilibrium. 337 Figure 10.3 Particle sorting based on their interfacial property. a, Macrograph, b-e, micrographs and f-g, schematic illustrations of PS and PE particles collected at two locations, PS on the oil-water-air three phase contact line versus PE on the oil-air interface. h, Sketch of the balance of the six interfacial tensions at equilibrium. For PS, the six interfacial tensions form a tetrahedral; and for PE, the tetrahedral collapses into a triangle. 338 Although the surface of natural water sources are enriched in microplastics, microplastics are also found suspended in bulk water[2]. To demonstrate that liquid interface is capable of collecting microplastics in bulk water, we designed a simple experiment, where fluorescent PS microparticles (2 m diameter, sulfate-modified) were initially dispersed in aqueous phases. After placing a vegetable oil droplet onto the water surface, air was bubbled through the system for 40 min (Figure 10.4a). The system was left overnight to allow oil droplets to float and coalesce at surface (flotation, Figure 10.4b). We note that for the system with pure water, vegetable oil formed emulsion droplets that were stable for three days (0.0076 g/L oil in water, Methods). The stability of the emulsion can be suppressed in the presence of 3.5 wt% salt, which is the average salt concentration in ocean. Fluorescent PS particles were found to be collected at the vegetable oil droplet-water interface (Figure 10.4c). Particle counts were quantified using flow cytometer (Figure 10.4d). We found that for diluted samples in water (950 ± 23 particles/L initially), less than 10% of particles were collected, whereas for concentrated samples in water (14501 ± 60 particles/L initially), 70% particles were collected. Surprisingly, in the presence of 3.5 wt% salt, 99.9% of microparticles were removed from the bulk aqueous phase. Additionally, oil droplets at the water interface can also be transported and coalesce in the presence of carrier species (Figure 10.4e), which can facilitate the recycling process if multiple oil sinks were used or oil sinks break up into multiple droplets after the bubbling process. 339 Figure 10.4 Pre-treatment for microplastics in bulk aqueous phases and post- procedure. a, Vegetable oil droplets were stirred in aqueous solutions in the presence of PS microparticles suspension by air bubbling. b, After setting still overnight, c, PS particles were found to be collected by the oil-aqueous interfaces (schematic illustration, top, and fluorescent micrograph focusing on the edge of the vegetable oil sink, bottom). d, Particle counts in the absence and in the presence of the bubbling process. Less than 10% microplastics were collected for diluted suspension in water, 70% were collected for concentrated suspension in water, and 99.9% were collected in the presence of 3.5 wt% salt (mimicking ocean environment). e, The oil sinks can also be transported and coalesce using the surface liquid extraction method in this paper. The location of the carrier decanol droplet (red circle) and trajectories of oil droplets (green dashed lines) are shown. 3. Conclusions Interfacial phenomena at equilibrium and beyond equilibrium provide a powerful tool for microplastics removal from both water surface and bulk, also for particle sorting. This platform can be combined with microsensors that detect microplastic, natural flocculants that collect microplastics[28], enzymes and microbes that break down plastics[29], and oil absorbents, gelators[30] or solidifiers[31] that potentially provide a “skeleton” for the oil sink. The surface liquid extraction method and the physical principles presented here are general and can be broadly applied to 340 separation of macromolecules, materials and bioparticles. 4. Supporting Information METHODS Materials. All materials were used as received: vegetable oil (Good & Gather, Target Brands, Inc.); decanol (TCI America); polypropylene glycol M.W. ∼4,000 (Thermo Scientific Chemicals); clear polyethylene (PE) microspheres 1–4 m, clear polyethylene (PE) microspheres 10–27 m and polystyrene (PS) microspheres 14–20 m (Cospheric LLC); FluoSpheresTM polystyrene (PS) microsphere (2 m, sulfate- modified, 2% dispersion in water, yellow-green fluorescent), FisherbrandTM Petri dishes and FisherbrandTM glass pipets (Thermo Fisher Scientific); poly(tetrafluoroethylene) (PTFE, powder, 1 m particle size), Sudan II (90%, fat-soluble colorant) and sodium chloride (99.0%) (MilliporeSigma). Purification of water (18.2 MΩ cm resistivity at 25 °C) was performed using a Milli-Q water system (Millipore). Demonstration of surface liquid extraction. Vegetable oil was dyed with Sudan II (< 0.1 wt%). 25 mL water was added into a Petri dish. 0.5 L carrier decanol or PPG was then carefully placed onto the water surface. Next, model microplastics, PE (1–4 m or 10–27 m), PS (14–20 m) or PTFE (m), were dispersed onto the water surface using glass pipets. 60 L vegetable oil was placed onto the water surface. Experiments on particle velocity as a function of time in Figure 10.2c were done with additional microparticles dispersed onto water surface once initial particles were collected. Macroscopically, videos were recorded using a Moticam 10+ camera or a Canon EOS Rebel T6i camera at 30 fps for velocity tracking. Microscopic results were recorded using the Moticam camera under polarized light or fluorescent microscopy. Scheme 10.1 Velocity tracking under macroscopic and microscopic view, respectively. Velocities of microparticles or their aggregates were extracted when moving through the purple annulus regions as denoted. 341 Air bubbling pre-treatment for microplastics in bulk aqueous phases. 2 m PS particles were diluted into water or salt water containing 3.5 wt% sodium chloride at 1:10,000–100,000 v ratio. 120 L vegetable oil was added into the microparticle suspensions. After bubbling air for 40 min, samples were set still overnight. PS particles and remaining oil droplets (measured and calculated to be 12,829 ± 2906 /L in water) in the aqueous phase were counted by a BD AccuriTM C6 Plus flow cytometer (384873 ± 87192). Average size of remaining oil droplets was measured to be 173 ± 8 nm in water using dynamic light scattering (DLS). In 3.5 wt% salt solutions, oil droplets were not detected using DLS. Measurement of interfacial tensions and contact angles. Interfacial tensions of water-air, oil-water, oil-air, decanol-air, decanol-water, PPG-air and PPG-water were measured using the pendant drop method performed with an optical tensiometer (Attension Theta T200, Biolin Scientific). Hamilton metal 22 gauge needles (blunt point) and 4.5 mL Fisherbrand disposable cuvettes were used. The equilibration time was 5 min for each drop. The oils used in Figure 10.2d were either untreated (containing dye), or with 0.5 L decanol or PPG added into 60 L oil. The contact angles of PS and PE particles at different interfaces shown in Extended Data Figure 10.5 were measured by first mixing particles with vegetable oil (for oil surface) or water (for water surface), then injected into 100 m thick cells confined by two glass slides, and observe under microscope. Contact angles of particles on the oil-water interface were measured by mixing particles into oil, then adding water at 1:1 v ratio. The mixtures were vortexed and observed in 100 m thick glass cells under microscope. The results presented in this paper were obtained using three independent measurements. The densities used to analyze the pendant drop measurements were 0.0012 g/cm3 (air), 0.9978 g/cm3 (aqueous 342 phases), 0.9200 g/cm3 (vegetable oil), 0.8297 g/cm3 (decanol) and 1.0040 g/cm3 (PPG). Comparison of particle velocity using surface liquid extraction versus diffusion. The diffusion coefficient for microparticles can be estimated by the Stokes-Einstein 𝑘 equation[13]: 𝐷 = 𝐵 𝑇 , where 𝑘 −23B = 1.38 × 10 J/K is the Boltzmann constant, 6𝜋𝜂𝑟 𝑇~300 K is the room temperature, 𝜂~1 mPa s is the dynamic viscosity of water. For a 𝑟 = 10 m particle, 𝐷 is estimated to be 2.2 × 10−14 J/K. The time 𝑡 needed to diffuse across 5 cm can be evaluated by[13] 𝑥̅̅ 2̅ = 2𝐷𝑡, 𝑡 ~ 5.7 × 1010 s, which leads to an average velocity for diffusion of 8.8 × 10−10 mm/s. The results in Figures 10.1 and 10.2 reveal an average particle velocity of ~1 mm/s over 5 cm using the surface extraction method, 109 times faster than diffusion. For 𝑟 = 100 m, 𝐷 = 2.2 × 10−15 J/K , 𝑡 ~ 5.7 × 1011 s , and diffusion velocity is estimated to be 8.8 × 10−11 mm/s for the 5 cm distance. Equilibrium contact angle for particles at interfaces. For an ideal particle surface that possesses zero contact angle hysteresis (Scheme 2), the contact angle 𝜃 is determined by three interfacial tensions, 𝛾31 , 𝛾32 , 𝛾21 , which can be estimated by Young’s equation[15]: 𝛾31 − 𝛾32 cos 𝜃 = (10.1) 𝛾21 Scheme 10.2 Contact angle of a solid particle 3 at the interface between 1 and 2. When one interfacial tension is larger than the sum of the other two interfacial tensions, e.g., 𝛾31 > 𝛾32 + 𝛾21, 𝜃 = 0°, phase 2 wets the surface of 3. When the sum of any two interfacial tension is larger than the third interfacial tension, the particle is pinned at the interface in equilibrium. This turns out to be consistent with the triangle 343 inequality, which is known as Neumann's triangle rule[32]. The energy required to detach the particle from the interface can be expressed as: ∆𝐸 = 𝛾 2 221𝜋𝑟 (1 − |cos 𝜃|) (10.2) where 𝑟 is the particle radius. For a 10 m particle at water-air interface with 𝜃 = 90°, 𝛾21 = 72.8 mN/m, ∆𝐸 = 2.3 × 10 −11 J = 5.6 × 109 𝑘B𝑇, where 𝑘B is the Boltzmann constant. Because the energy needed to remove the particle from the interface is large, the attachment of particle is considered irreversible. Estimation of effective radius for Stokes’ drag. To evaluate the Stokes’ drag acting on the particle surface that is submerged in water, we estimated an effective radius based on the area of the particle-water interface. The surface area of a spherical cap is given by: 𝐴cap = 2𝜋𝑟 2(1 − cos 𝜃) (10.3) The area of the particle-water interface is: 𝐴 = 4𝜋𝑟2 − 𝐴 2cap = 2𝜋𝑟 (1 + cos 𝜃) (10.4) This surface area is equivalent to a sphere with radius 𝑟′ (Scheme 3): 2 𝐴′ = 4𝜋𝑟′ = 𝐴 = 2𝜋𝑟2(1 + cos 𝜃) (10.5) Scheme 10.3 Area of the particle-water interface equals to the surface area of a sphere; the radius of the latter is estimated to be the effective radius for Stokes’ drag. Then the ratio between two effective radii is: 344 𝑟′21 1 + cos 𝜃1 2 = √ (10.6) 𝑟′2 1 + cos 𝜃2 345 Extended Data Figure 10.1 Spreading of the oil droplet after decanol was fully absorbed. After 1 hour, the oil droplet became a circular film with a diameter of 8 cm, which occupied most of the water surface in the Petri dish. 346 Extended Data Figure 10.2 PPG, which spreads across the water surface to form a film, was also used as the carrier. a, Schematic illustration of the surface liquid extraction using PPG. PPG travels along the water interface toward a sink, which carries microplastics by creating a surface flow. Experiments were done in a Petri dish with 9 cm diameter. b, Image of an experiment with carrier: PPG (distributed across the interface), model microplastics: 200-1000 m aggregates formed from 1–4 m PE particles that are visible in this macroscopic view (denoted by white arrows) and sink: vegetable oil (orange droplet, dyed with Sudan II). Green dashed lines indicate motion of particles. Inset show microplastics were collected on the vegetable oil interfaces after 10 min. 347 Extended Data Figure 10.3 Surface extraction process of PPG. As PPG was absorbed into the vegetable oil droplet, the film of PPG broke up and form discontinuous domains. The film, which is visible in this macroscopic view, was removed after 30 min. White arrows indicate the discontinuous PPG film. 348 Extended Data Figure 10.4 Particle velocity with decanol as the carrier. a, Particle velocity decreased as a function of time for various microparticles with decanol as the carrier. b, With decanol as the carrier, no significant differences on the velocities of various microparticle species were found (400-700 m particle aggregates, velocity data obtained at a distance of 10–20 mm away from the oil sink, and at 4–6 min after vegetable oil dispensing), in contrast to the particle species dependent velocity with PPG. 349 Extended Data Figure 10.5 Contact angles and interfacial tensions. a-f, Micrographs of PS (14–20 m) and PE (10–27 m) particles at air-water, air-oil and oil-water interfaces, respectively. The contact angles (red) are determined by the three interfacial tensions (solid-fluids and fluid-fluid) existing in each micrograph. When the three interfacial tensions form Neumann's triangles[32] (Method) in a-e, the 350 particles were pinned at the interfaces; but when PE-water interfacial tension is larger than the sum of the PE-oil and the oil water interfacial tensions in f, oil wet the PE interface against water. g, Interfacial tensions mentioned in this manuscript. The fluid- fluid interfacial tensions were measured using the pendant drop method. PS and PE surface tensions were obtained from literature, and particle-liquid interfacial tensions were calculated from the contact angles. We note the contact angle of PS at the water surface in a was smaller than expected, so we used the contact angles in b,c for the calculations. 5. References *This chapter was prepared as a Research Article reporting original research. N.L.A. asked X.W. how to collect micropastics. X.W. conceived the initial idea and conducted the experiments. X.W. and N.L.A. developed the concept and mechanism. S.R. contributed to the measurements of contact angles. Adapted with permission from: X. Wang, S. Roh and N. L. Abbott, Active Collecting and Sorting Microplastics Using Liquid Interfaces. Close to submission. [1] X. Lim, Microplastics Are Everywhere-but Are They Harmful?, Nature 593, 22 (2021). [2] A. Stubbins, K. L. Law, S. E. Muñoz, T. S. Bianchi, and L. Zhu, Plastics in the Earth System, Science 373, 51 (2021). [3] M. MacLeod, H. P. H. Arp, M. B. Tekman, and A. Jahnke, The Global Threat from Plastic Pollution, Science 373, 61 (2021). [4] M. A. Browne, A. Dissanayake, T. S. Galloway, D. M. Lowe, and R. C. Thompson, Ingested Microscopic Plastic Translocates to the Circulatory System of the Mussel, Mytilus Edulis (L.), Environ. Sci. Technol. 42, 5026 (2008). 351 [5] Y. Deng, Y. Zhang, B. Lemos, and H. Ren, Tissue Accumulation of Microplastics in Mice and Biomarker Responses Suggest Widespread Health Risks of Exposure, Sci. Rep. 7, 1 (2017). [6] A. Ragusa et al., Plasticenta: First Evidence of Microplastics in Human Placenta, Environ. Int. 146, 106274 (2021). [7] T. Braun, L. Ehrlich, W. Henrich, S. Koeppel, I. Lomako, P. Schwabl, and B. Liebmann, Detection of Microplastic in Human Placenta and Meconium in a Clinical Setting, Pharmaceutics 13, 921 (2021). [8] C. G. Avio, S. Gorbi, M. Milan, M. Benedetti, D. Fattorini, G. D’Errico, M. Pauletto, L. Bargelloni, and F. Regoli, Pollutants Bioavailability and Toxicological Risk from Microplastics to Marine Mussels, Environ. Pollut. 198, 211 (2015). [9] M. E. Miller, M. Hamann, and F. J. Kroon, Bioaccumulation and Biomagnification of Microplastics in Marine Organisms: A Review and Meta- Analysis of Current Data, PLoS One 15, e0240792 (2020). [10] S. Krause et al., Gathering at the Top? Environmental Controls of Microplastic Uptake and Biomagnification in Freshwater Food Webs, Environ. Pollut. 268, 115750 (2021). [11] A. L. Lusher, M. McHugh, and R. C. Thompson, Occurrence of Microplastics in the Gastrointestinal Tract of Pelagic and Demersal Fish from the English Channel, Mar. Pollut. Bull. 67, 94 (2013). [12] T. Sparks and G. Chase, Filters and Filtration Handbook (Sixth Edition) (Elsevier, 2016). [13] P. C. Hiemenz and R. Rajagopalan, Principles of Colloid and Surface Chemistry, Revised and Expanded (CRC press, 2016). [14] G. J. Elfring, L. G. Leal, and T. M. Squires, Surface Viscosity and Marangoni 352 Stresses at Surfactant Laden Interfaces, J. Fluid Mech. 792, 712 (2016). [15] L. E. Low, S. P. Siva, Y. K. Ho, E. S. Chan, and B. T. Tey, Recent Advances of Characterization Techniques for the Formation, Physical Properties and Stability of Pickering Emulsion, Adv. Colloid Interface Sci. 277, 102117 (2020). [16] A. Mizev, A. Trofimenko, D. Schwabe, and A. Viviani, Instability of Marangoni Flow in the Presence of an Insoluble Surfactant. Experiments, Eur. Phys. J. Spec. Top. 219, 89 (2013). [17] M. Roché, Z. Li, I. M. Griffiths, S. Le Roux, I. Cantat, A. Saint-Jalmes, and H. A. Stone, Marangoni Flow of Soluble Amphiphiles, Phys. Rev. Lett. 112, 208302 (2014). [18] L. Keiser, H. Bense, P. Colinet, J. Bico, and E. Reyssat, Marangoni Bursting: Evaporation-Induced Emulsification of Binary Mixtures on a Liquid Layer, Phys. Rev. Lett. 118, 074504 (2017). [19] H. Kim, K. Muller, O. Shardt, S. Afkhami, and H. A. Stone, Solutal Marangoni Flows of Miscible Liquids Drive Transport without Surface Contamination, Nat. Phys. 13, 1105 (2017). [20] D. Lohse and X. Zhang, Physicochemical Hydrodynamics of Droplets out of Equilibrium, Nat. Rev. Phys. 2, 426 (2020). [21] H. Manikantan and T. M. Squires, Surfactant Dynamics: Hidden Variables Controlling Fluid Flows, J. Fluid Mech. 892, P1 (2020). [22] S.-Y. Lin, T.-L. Lu, and W.-B. Hwang, Adsorption Kinetics of Decanol at the Air-Water Interface, Langmuir 11, 555 (1995). [23] D. Gupta, B. Sarker, K. Thadikaran, V. John, C. Maldarelli, and G. John, Sacrificial Amphiphiles: Eco-Friendly Chemical Herders as Oil Spill Mitigation Chemicals, Sci. Adv. 1, e1400265 (2015). [24] J. G. Lee, L. L. Larive, K. T. Valsaraj, and B. Bharti, Binding of Lignin 353 Nanoparticles at Oil-Water Interfaces: An Ecofriendly Alternative to Oil Spill Recovery, ACS Appl. Mater. Interfaces 10, 43282 (2018). [25] J. Rydberg, Solvent Extraction Principles and Practice, Revised and Expanded (CRC press, 2004). [26] V. S. Kislik, Solvent Extraction: Classical and Novel Approaches (Elsevier, 2012). [27] D. W. van Krevelen and K. te. Nijenhuis, Properties of Polymers (Elsevier, 2009). [28] J. Michels, A. Stippkugel, M. Lenz, K. Wirtz, and A. Engel, Rapid Aggregation of Biofilm-Covered Microplastics with Marine Biogenic Particles, Proc. R. Soc. B 285, 20181203 (2018). [29] H. S. Zurier and J. M. Goddard, Biodegradation of Microplastics in Food and Agriculture, Curr. Opin. Food Sci. 37, 37 (2021). [30] A. M. Vibhute, V. Muvvala, and K. M. Sureshan, A Sugar-Based Gelator for Marine Oil-Spill Recovery, Angew. Chemie 128, 7913 (2016). [31] H. Guan, Z. Cheng, and X. Wang, Highly Compressible Wood Sponges with a Spring-like Lamellar Structure as Effective and Reusable Oil Absorbents, ACS Nano 12, 10365 (2018). [32] J. S. Rowlinson and B. Widom, Molecular Theory of Capillarity (Dover Publications, 1982). 354 Chapter 11 Summary and Future Directions 1. Summary The research described in this thesis contributes to the design of functional soft matter systems that operate both at and beyond equilibrium. Both experimental and computational approaches involving colloid science, organic chemistry, polymer synthesis, hydrodynamics, atomistic simulation and continuum simulations are used to elucidate the physical mechanisms acting in the systems. Overall, these studies have developed new design principles for adaptive materials and active matter as summarized below. Mixtures of miscible and immiscible perfluorocarbon oils and hydrocarbon mesogens have been used to prepare multi-compartment (double and Janus) emulsion drops comprising coexisting nematic LC and isotropic oil phases (Chapters 3 and 4). I designed the morphologies and internal organizations of the complex emulsion droplets at equilibrium by using delicate balance between orientation-dependent interfacial energies, elasticity and topological defects that arise from the orientational ordering of LCs. The LC organizations were further explored by combining experiments and simulations based on the Landau-de Gennes continuum model. The droplets exhibit various shapes with internal Janus-type or core-shell morphologies that can be tuned widely through changes in temperature or interfacial adsorbates. The results demonstrate that multiphase LC emulsions formed from mixtures of perfluorocarbons and LCs provide new opportunities to engineer hierarchical and stimuli-responsive emulsion systems. In particular, the Janus-type multiphase droplets with low internal interfacial tensions described above maintain stable spherical geometry in the presence of chemical stimuli. Chapter 5 goes beyond equilibrium to explore how this type of oil Janus droplets that are placed into aqueous micellar solutions can generate spontaneous 355 motion due to the Marangoni effect (microswimmers). Selective solubilization of components of the droplets by micellar extraction has led to both morphogenesis and self-propulsion. The results revealed how morphogenesis of domains internal to the droplets can impact emergent dynamics when driven by active processes, including cycles of ballistic and spiraling behaviors, which can be described within the physical framework of squirmer models for spherical microswimmers in Stokes flow. This study offers the potential to guide the engineering of functional soft matter systems that possess ability to change form in response to the environment to modify dynamics as a reminiscence of living systems that undergo morphogenesis. During our exploration of fluorocarbons and LCs, we found that two very similar mesogens 5CB and PCH5 turned out to exhibit distinct anchoring behaviors at fluorocarbon interface. This finding led us to investigate how alternation of nanoscale (sub-molecular) structure of molecules dictates intermolecular interactions, and subsequently generates distinct macroscopic behaviors (Chapter 8). 5CB and PCH5 molecules are similar as they both possess a benzonitrile head group and a pentyl tail, and differ in that one aromatic ring of 5CB is replaced by a cyclohexyl group on PCH5. LC films were submerged into immiscible fluorocarbon, and were subsequently observed under polarized light optical microscope. Surprisingly, although the subtle difference in molecular structure, the microscopic anchoring behaviors of 5CB and PCH5 are dramatically distinct. Whereas the 5CB film shows a dark optical appearance that indicates homeotropic anchoring at the F9 interface, the PCH5 film exhibits a bright texture, which means the orientational ordering tilts from the surface normal. We use atomistic simulation to reconstruct the LC-fluorocarbon interfaces. Subsequent compartmentalization of mesogen and calculation of potential of mean force between each compartment and fluorocarbon molecule provide an explicit explanation of the formation of molecular ordering where affinity between perfluorocarbon and 356 cyclohexyl group is favored over phenyl group. In the long term, deciphering macroscopic behavior from nanoscale structural information sets a promising and challenging approach for designing functional and responsive materials. Chapter 6 presents an adaptive material system for chemical-stimuli-induced polymerization based on a liquid crystal (LC) printhead. The LC responds to a local chemical stimulus at its aqueous interface, which modulates the elastic forces and the electrical double layer forces, resulting in the ejection of initiator into the solution to trigger polymerization. Various LC printhead geometries are designed, allowing programming of: i) bulk solution polymerization, ii) synthesis of a thin surface-confined polymeric coating, iii) polymerization-induced self-assembly of block copolymers to form various nanostructures (sphere, worm-like, and vesicles), and iv) 3D polymeric structures printed according to local solution conditions. The approach is demonstrated using amphiphiles, multivalent ions, and biomolecules as stimuli. Overall, we design the approach to achieving spatially and temporally controlled polymerization in response to a range of cues. Based on the knowledge obtained from Chapters 5 and 6, in Chapter 7, we explore the idea that a carrier droplet carrying cargo can autonomously move to a remote location, and spontaneously and precisely release the cargo at its destination. The idea is simple, but designing a system that can realize the idea is challenging. Since it is unlikely to use equilibrium processes in the absence of additional stimuli to precisely control the on/off of the release, we integrated dynamic flow into the functioning of the system. The timed release of the cargo relies on Marangoni flows at the surface of the LC droplets, which not only drive the motion of the droplets (swimming) but also generate an internal circulation that advects inclusions (cargo) to, together with the LC elasticity, cause a time-dependent coarsening and clustering. Above a threshold cargo cluster size, inertial forces exceed interfacial colloidal forces (elastic and electrical 357 double layer forces) that initially trap cargo within the LC droplets during swimming, which leads to polymerization at the targeted location. We show that the LC droplets can navigate their way through a simple millifluidic network of channels to form a polymer plug at the desired location. Overall, our results provide a platform that utilizes active soft matter to achieve spatial and temporal targeting of chemical cargo. Chapter 9 leverages the knowledge of LCs and dynamic reactions to explore whether LCs can sense molecular-level chemical transformations and output the signal through optical patterns. Our results revealed that the dynamic, non-equilibrium processes of interfacial reaction pathways between a surfactant and amine nucleophiles can trigger spatial and temporal optical “fingerprints” of LC films. The change in optical appearance was generated by the interactions between surfactant/products and LC interfaces. We found that the LC films exhibited one or two types of responses that happened sequentially, dependent on the added nucleophile. The “type-one” non- equilibrium state of the LC was generated by reaction-induced interfacial tension gradient, also confirmed by subsequent measurements on dynamic and equilibrium 5CB-aqueous interfacial tensions, while the “type-two” non-equilibrium state indicated adsorption of products with substitutions on two reactive amines within one molecule. Additional observation of a re-entrant behavior of a uniform dark texture between the two non-equilibrium states correlates with the primary amine group versus the secondary amine. These results open up new opportunities for developing intelligent materials for in situ, rapid and delicate detection of chemical reactions and dynamic processes, especially when combined with current AI technology. In Chapter 10, we focus on the global issue of microplastics in the environment. A separation principle “surface liquid extraction” using liquid interfaces was proposed. A carrier species is dispersed on the water surface, with a drop of vegetable oil as a sink that continuously absorbs the carrier. The absorption of the carrier by the sink generates 358 sustained surface fluxes that “actively” skims the water surface due to Marangoni stress, carrying and collecting polymeric microparticles presented on the surface. The rate- limiting step of the extraction process depends on whether the carrier is active on the oil-water interface. This method provides a transport velocity of the order of 100-1000 m/s for 10-700 m microparticle aggregates, at least 109 times faster than diffusion. Oil interfaces can also collect microplastics in bulk water by direct contact with microplastics through a bubbling process. Our method also simultaneously sorts polymeric particles into distinct locations of the oil sink, air-oil interface, oil-water interface or air-oil-water three phase boundary, based on the balance of interfacial tensions. These findings provide a platform that uses interfaces to address the microplastics problem. 2. Future Directions The design of functional materials using a combination of processes both at and beyond equilibrium to achieve functions, though challenging, can open up many opportunities. While some designs concerning important issues and concepts have been realized in this dissertation, more can be done by leveraging the knowledge that has been learned. I suggest some future directions outlined below. The plastics in the environment can be viewed as a pollution issue that needs to be resolved as we mentioned in Chapter 10, but can also be regarded as a type of “resource”. Plastics come from petroleum, which is not in infinite supply as it took millions of years to form on the earth. It takes only about 50 years to exhaust current proven oil reserves based on current technology and production (from bp Statistical Review of World Energy 2021). Even though oil drilling technology may advance, the cost will increase as the depth of oil reservoirs left underneath the earth increases. Having the ability to collect and sort microplastics has a large impact on the reuse of 359 these plastics, which is important for both the environment and energy. We have demonstrated in Chapter 10 how to use liquid interfaces to collect and sort microplastics at the scale of centimeters. This method has the potential to be scaled up to a scale of meters or even larger. For example, below I propose a process that enables the collecting and sorting of microplastics at surface of ocean (Figure 11.1). An oil sink is first dispensed on top of the water surface, which spreads out forming a film. Next, carrier species are dispensed on the water surface with a proper distance surrounding the oil film. The flux of carrier on the water surface changes the pressure acting on the oil sink, which causes the oil sink to retract into a small area. Microplastics are carried by the flux of the carrier, and are sorted spontaneously onto different interfaces of the oil sink via the balance between interfacial tensions. After the carrier is fully extracted from the water surface, microplastics are separated from the oil mixtures, and the oil mixture is subsequently collected and distilled for reuse. Figure 11.1 Proposed process for collecting and sorting microplastics on surface of a large water body. Schematic illustrations (top view) of process for collecting and sorting microplastics. (a) Oil sink dispensed on water surface spreads out into a film. (b) Carrier liquid is dispensed on water surface at a proper distance around the oil sink. (c) Flux of carrier causes oil sink to retract into small area, also carries microplastics toward the oil sink. Microplastics, oil sink and carrier can be separated using mechanical or chemical techniques for reuse. 360 A separations principle based on the balance of interfacial tensions (“tetrahedral rule”) and “surface liquid extraction” has been demonstrated in Chapter 10 using one type of oil (vegetable oil) as the sink, and two types of microparticles (polyethylene and polystyrene) as the model particles that need to be sorted. This method can be extended to other types of fluid sinks, microparticles and macromolecules that are pined at interfaces (Figure 11.2). Multiple types of fluids such as ionic liquid, water, different hydrocarbons, fluorocarbons and air can be stacked to create a “gradient” of sinks that possess different surface tensions from ~100 mN/m to ~15 mN/m. The balance of interfacial tensions needs to be examined first (see Chapter 2.3, Fluid at Fluid Surface) to design a structure of multiple fluids that can form layers or lenses on top of another fluid (Figure 11.2). Additional designs can include column devices that consist of fluid absorbents to provide a “skeleton” that holds the multiple fluid sinks against e.g., gravity. A fluid carrier can carry microparticles/macromolecules through the system. Because different microparticles/macromolecules tend to have distinct relative affinities to different sinks, which leads to separation of microparticles/macromolecules onto Figure 11.2 Proposed separation mechanism of surface liquid extraction using a set of fluids with interfacial tension “gradient”. Schematic illustration of the separation process using multiple types of fluids as sinks (stationary phases) and a fluid carrier as the mobile phase that moves through the system for separation of microparticles/macromolecules (denoted as black dots), the latter are eventually sorted onto different interfaces or into fluid bulk dependent on the relative affinity between microparticles/macromolecules and the stationary phase. 361 interfaces or into the bulk of the sinks. For example, polyethylene particles are predicted to be collected on the hydrocarbon, fluorocarbon and air three-phase contact line. In contrast, Teflon particles are predicted to be located in the fluorocarbon sink because fluorocarbon is likely to wet both the air and hydrocarbon interface in contact with the particles. Non-charged hydrophilic particles can be collected on the hydrocarbon-water interface to reduce the interfacial tensions. These processes eventually separate particle or macromolecule mixtures onto different interfaces or bulk fluids. The concepts of mobile phase and stationary phase from chromatography are applied here for the fluid carrier and fluid sink, respectively. The process also shares some common features with density gradient centrifugation, a technique in which molecules move through a density gradient due to centrifugation until they find a density equal to their own. Here, components move through an interfacial tension gradient until they find a set of interfacial tension that leads to a force balance. We have used amines as model reactants to demonstrate how molecular-level chemical transformations can be detected by LCs in Chapter 9. I envisage that LC- aqueous interfaces can be widely used in monitoring interfacial reactions, such that interfacial polymerization is likely to generate optical fingerprints of LCs that correlates with the polymer formation of different chain length and chemistry. Multi-step reactions involving transporting of intermediates and products between aqueous and organic phases can also be monitored using LC interfaces. In addition, amines are widely found in all organisms (Figure 11.3), such as monoamine neurotransmitters (dopamine, norepinephrine and serotonin) that are involved in memory and emotion, and polyamines that regulate the growth of plant cells and response to external stressors. Although these molecules perform functions in vivo at a much lower concentration as compared to our experiments, we can use LCs to detect and explore the molecular-level structure transformations of these molecules in real-time, and potentially understand 362 how such processes correlate with their functions, which is not easy to achieve with other chemistry methods such as mass spectrometry, chromatography or nuclear magnetic resonance. Figure 11.3 Examples of amines found in organisms. Micromachines attract increasing interest in the prospect of biomedical applications, such as visual diagnosis and targeted therapy. Although performing such experiments in vivo is challenging, it is interesting to explore active matter systems in the presence of proteins and lipids. Soft machines described in Chapter 7 are sensitive to the chemical gradients of monomer/micelle. This sensitivity is general to background gradients of chemicals. We propose to explore whether micromachines can navigate their way along a gradient of biomolecules in viscous fluidic environments, and release cargo at remote locations. We envisage that processes such as coupled adsorption and 363 desorption of small molecules and macromolecules, and assembly or disassembly of liposomes can be used to guide the autonomous motion and function of soft micromachines. 364