Liquid crystal–based open surface microfluidics manipulate liquid mobility and chemical composition on demand

Liquid crystal surfaces enable manipulation of liquid mobility and cargo release via temperature, electrolytes, and light.

Abstract

The ability to control both the mobility and chemical compositions of microliter-scale aqueous droplets is an essential prerequisite for next-generation open surface microfluidics. Independently manipulating the chemical compositions of aqueous droplets without altering their mobility, however, remains challenging. In this work, we address this challenge by designing a class of open surface microfluidic platforms based on thermotropic liquid crystals (LCs). We demonstrate, both experimentally and theoretically, that the unique positional and orientational order of LC molecules intrinsically decouple cargo release functionality from droplet mobility via selective phase transitions. Furthermore, we build sodium sulfide–loaded LC surfaces that can efficiently precipitate heavy metal ions in sliding water droplets to final concentration less than 1 part per million for more than 500 cycles without causing droplets to become pinned. Overall, our results reveal that LC surfaces offer unique possibilities for the design of novel open surface fluidic systems with orthogonal functionalities.

INTRODUCTION

The design of open surface microfluidics that enable orthogonal control of liquid mobility and chemical compositions is critical for devising the next generation of microfluidic platforms that will find use in applications across chemical (1–5), environmental (6, 7), and biomedical fields (8–10). To achieve these desirable functionalities, extensive studies have demonstrated stimuli-responsive liquid mobility on open fluidic platforms based on either micro/nanoscale topographical surfaces (8, 11–15) or water-immiscible liquid-coated surfaces (16–19). However, methods of manipulating droplets’ chemical compositions tend to rely upon chemical adsorption directly from the underlying surface, which has been shown to subsequently pin droplets to the surface and render them immobile (7, 20, 21), as illustrated in Fig. 1A. This intrinsic coupling of droplets’ chemical composition and mobility greatly weakens the robustness of conventional open surface microfluidic systems and hinders their use in real-world applications, presenting a substantial challenge that must be overcome to realize the full potential of open surface microfluidics.

Fig. 1. Manipulation of liquid mobility and chemical composition on open surface microfluidic systems.

(A) Chemicals released from conventional functional surfaces causing water droplets to become pinned. (B) Manipulation of liquid mobility and chemical composition at liquid crystal (LC) surfaces on demand.

Thermotropic liquid crystals (LCs) are a particularly promising class of anisotropic fluids that produce a remarkable diversity of colloidal and interfacial phenomena with unprecedented complexities and functionalities (22, 23). Because of their intrinsic properties, including positional order and orientational order in various mesophases (24), immobilized LCs have been exploited for a variety of applications, including sensing chemicals (25, 26) and activating the release of cargo (27, 28). We herein propose that these intrinsic properties may enable the design of LC-based open surface microfluidics that can manipulate both the mobility and chemical compositions of resting droplets on demand, as illustrated in Fig. 1B. Although promising, the design of LC-based open surface microfluidics has not yet been achieved because of water droplet–induced dewetting of LC films coated on conventional hydrophobically modified substrates, including silane-functionalized surfaces (see movie S1) (29), azlactone-functionalized surfaces (30), and porous polystyrene–coated surfaces (31).

In this work, we report the design of an LC-based open surface microfluidic platform that enables the manipulation of the mobility and chemical compositions of droplets on demand. Specifically, we use porous LC polymeric networks to stabilize thermotropic LC mesogens to overcome the aforementioned issue of water-induced LC dewetting. We find that the mobility of water droplets on LC-based surfaces depends only on the positional order of the LC: Water droplets become highly pinned at LC surfaces in the smectic A phase, whereas droplets can freely slide without pinning at LC surfaces in both the nematic and isotropic phases. Moreover, we experimentally and theoretically demonstrate that the mesogenic orientational order of the LC surface plays a pivotal role in the release of chemicals from the LC surface to droplets. Last, we demonstrate that as a consequence of the inherent decoupling between a droplet’s mobility and the release of cargo from the LC, LC-based open surface microfluidic platforms can capture and precipitate heavy metal ions in droplets of water at LC surfaces without hindering droplets’ mobilities. Our work provides novel design principles for fabricating anisotropic liquid–based open surface microfluidics that enable promising applications including liquid droplet–based chemical synthesis and medical diagnostics.

RESULTS

Stabilization of LC film against water droplet–induced dewetting

Inspired by the pitcher plant Nepenthes, previous studies reported the design of slippery liquid–infused porous surfaces (SLIPS), which exhibit excellent liquid repellency (17, 18, 32, 33). In this work, we applied concepts from SLIPS to engineer stable LC surfaces that will not be dewetted by water. To achieve this behavior, the total interfacial energy of the porous substrate without LC wetted by water (E_A) needs to be higher than the interfacial energy of an LC surface with (E₁) or without (E₂) water floating on top of the LC (32)

$$\mathrm{\Delta }{E}_1=E_{\mathrm{A}}–E_1=r({\mathrm{\gamma }}_{\text{LC}}{\text{cos}\mathrm{\theta }}_{\text{LC}}–{\mathrm{\gamma }}_{\mathrm{w}}{\text{cos}\mathrm{\theta }}_{\mathrm{w}})–{\mathrm{\gamma }}_{\mathrm{w}–\text{LC}}>0$$ (1)

$$\mathrm{\Delta }{E}_2=E_{\mathrm{A}}–E_2=r({\mathrm{\gamma }}_{\text{LC}}{\text{cos}\mathrm{\theta }}_{\text{LC}}–{\mathrm{\gamma }}_{\mathrm{w}}{\text{cos}\mathrm{\theta }}_{\mathrm{w}})+{\mathrm{\gamma }}_{\mathrm{w}}–{\mathrm{\gamma }}_{\text{LC}}>0$$ (2)

in which r is the roughness factor (defined as the ratio of the porous surface’s true surface area to its projected surface area), γ_w–LC is the interfacial tension between water and the LC, and γ_w and γ_LC represent the surface tensions of water and the LC, respectively. θ_w and θ_LC are the equilibrium contact angles of water and the LC on the solid surface, respectively. In this work, we used 4′-octyl-4-biphenylcarbonitrile (8CB) as the LC because its intrinsic mesophases (described below) offer decoupled functionalities for LC-based open microfluidic platforms. To satisfy Eqs. 1 and 2, we used LC polymer 1,4-bis-[4-(3-acryloyloxypropyloxy)benzoyloxy]-2-methylbenzene (RM257) to form the porous substrate, as its strong dipole-dipole interactions with 8CB stabilized the lubricating LC film against dewetting by water (Fig. 2A). On the basis of experimental measurements, we calculated both ΔE₁ and ΔE₂ > 0 for a combination of the LC polymer polyRM257 and 8CB (see the Supplementary Materials).

Fig. 2. LC-based open surface microfluidics.

(A) Schematic illustration of pitcher plant Nepenthes-inspired stabilization of LC film against dewetting by water droplets. A film of 8CB is coated on a porous polyRM257 network, and a wrapping layer of 8CB is formed on the water droplet. (B) Polarized light micrograph (top image, top view) and a schematic illustration demonstrating the LC orientation (bottom image, side view; not scaled to actual size) of a nematic 8CB film. The bright appearance in the center of the droplet occurs because of the degenerate planar anchoring of LC underneath the water droplets. Scale bar, 200 μm. Inset is a conoscopic image confirming homeotropic alignment of 8CB at the interface between air and 8CB surface and the interface between 8CB and a polyRM257-coated substrate. Crossed double-headed arrows indicate the orientations of crossed polarizers. (C) Sliding angle of a 3-μl water droplet on the surface of 8CB and silicone oil at different temperatures. Error bars represent SDs and n = 3 for each data point. (D) Representative force diagram (F_d versus time) of a 3-μl water droplet moving on smectic A and nematic 8CB surfaces; see movie S1. Inset in (D) shows data in the range of 0.4 to 0.8 μN.

To verify that a textured polyRM257 surface can stabilize a lubricating layer of 8CB, we photopolymerized a mixture of reactive RM257 [10 weight % (wt %)] and 8CB (90 wt %) to form 8CB-swelled nanoporous polyRM257 structures on dimethyloctadecyl[3-(trimethixysilyl) propyl]ammonium chloride (DMOAP)–functionalized glass substrates. 8CB and RM257 adopted a perpendicular anchoring at the DMOAP-functionalized glass surface (Fig. 2B). Subsequently, we dropcast a ~130-μm-thick layer of pure 8CB on these structures to obtain stable 8CB-infused porous polyRM257 surfaces, which we refer to as “LC surfaces.” Next, we characterized the behavior of water droplets deposited on a nematic LC surface where the constituent molecules had long-range orientational order but no positional order (i.e., self-aligned with centers of mass randomly distributed). When observed under a polarized light microscope, the nematic 8CB film was dark in air and turned bright when it came in contact with a water droplet, as shown in Fig. 2B and movie S1. This transition is consistent with different surface anchorings of nematic 8CB at the air-LC interface (perpendicular to the interface) and water-LC interface (parallel to the interface) (24). Using transmission electron microscopy (TEM), we found that the sizes of the nanopores in the polyRM257 substrate ranged between 10 and 40 nm, and using nitrogen sorption isotherms, we measured the surface area and the volume average pore sizes of the nanoporous polyRM257 to be 419 m²/g and 12 ± 5 nm, respectively (see Materials and Methods and fig. S1 for details). We also note here that there was a thin 8CB wrapping layer around the water droplet, which may slow the evaporation of water droplets at LC surfaces (see fig. S2). In addition, we note that two concentric rings appear around the water droplet border (Fig. 2B) due to the wrapping layer and the wetting ridge of 8CB.

Water droplets need to overcome a static pinning force (F_static) to start moving against dynamic friction (F_dynamic) at a surface (14). As shown in Fig. 2C, the sliding angle of a 3-μl water droplet was ~2° when 8CB was in the nematic phase. To further characterize droplet mobility on LC surfaces, we measured the dissipative force acting on a droplet sliding across an LC surface (F_d) using a customized cantilever force sensor (see fig. S8 and movie S2). As shown in Fig. 2D, water droplets at the nematic LC surface exhibited a small F_static of 0.6 μN with a smooth transition from the static to the dynamic regime (F_static ≈ F_dynamic), which is consistent with the pinning-free slippery mode (34, 35) (i.e., where the droplet rests upon the lubricating layer without displacing it and without contacting the underlying substrate). In addition, we observed that F_dynamic increases as the speed at which the droplet moves across the surface (U) increases, which agrees with the Landau-Levich-Derjaguin theory (see fig. S9) (34)

$$F_{\text{dynamic}}\approx 2\mathrm{\pi }{\mathrm{\gamma }}_{\mathrm{w}–\mathrm{l}}R {\mathit{Ca}}^{2/3}$$ (3)

in which γ_w–l is the interfacial tension between water and the lubricant, R is the base radius of the droplet, and Ca is the capillary number. We attribute the observed increase to F_dynamic to an increase in the thickness of the lubricating nematic 8CB layer, which, in turn, increases the viscous dissipation that impedes the water droplet’s motion. The experimentally measured F_dynamic yielded a linear relationship with F_dynamic calculated using Eq. 3 (see fig. S9B). The line best fits the data with a nematic LC dynamic viscosity of ~20 centipoise, indicating degenerate planar anchoring of nematic 8CB at the water droplet’s surface (36).

The above results lead us to conclude that the porous polyRM257 substrate stabilized the nematic 8CB film, thereby preventing water droplets from becoming pinned to the surface. We note here that the design principles that we used to stabilize LC surface against dewetting by water droplets (summarized in Eqs. 1 and 2) are applicable to a range of thermotropic LCs, such as 4′-pentyl-cyanobiphenyl (5CB) and E7. In addition, this combination of polyRM257 and 8CB can be used to coat a variety of substrates (see fig. S11). Furthermore, our LC surfaces exhibit good physical stability, remaining slippery after the substrate had been damaged by scratching it with a razor (fig. S12). Last, the sliding behavior of droplets does not notably change even when the pH of the droplet was varied from 1 to 13, as shown in fig. S13.

Molecular positional order–dependent static friction of water droplets at LC surface

As temperature increases, 8CB undergoes phase transitions from the crystal to the smectic A, to the nematic, and lastly to the isotropic phase, each characterized by different types and degrees of molecular order (24). To understand how phase transitions could provide functionality to our LC surfaces, we studied the wetting behavior and mobility of water droplets on LC surfaces where the LC’s phase was tuned by controlling the substrate’s temperature. As summarized in fig. S4A, the apparent advancing contact angles (θ_adv) of a sessile water droplet on the 8CB surface remained almost constant between 70° and 80° for droplets residing upon LC surfaces regardless of whether the LC was in the smectic A, nematic, or isotropic phases (temperature range of 25° to 50°C). However, the sliding angle of 3-μl water droplets exhibited strong dependence on the LC phases, which abruptly changed from ~2° in the nematic or isotropic phases to ~35° in the smectic A phase, suggesting that water droplets became severely pinned to the smectic A surface (see movie S1). We note here that the sliding angle of water droplets on LC surfaces was measured by placing water droplets directly on 8CB surfaces with specific 8CB mesophases. In addition, we observed that focal conic domain arrays (37–40) formed at the LC-water interface when the LC was in the smectic A phase (fig. S3), which is consistent with different surface anchorings of smectic A 8CB at the air-LC interface (perpendicular anchoring) and water-LC interface (parallel anchoring) (24). We note here that only one sliding angle value was observed for each LC mesophase, even across multiple temperatures within the same mesophase (Fig. 2C). We also observed that silicone oil–infused porous substrate exhibits temperature-independent droplet mobility, as shown in Fig. 2C. These two results lead us to conclude that observed mesophase-dependent mobilities cannot merely be attributed to changes of surface energies/contact line lengths from varying the substrate’s temperature.

We further measure F_d of droplets sliding across LC surfaces in various phases to provide insights into the mechanism by which the LC mesophase affects the mobility of droplets on LC surfaces. As shown in Fig. 2D, water droplets were highly pinned at the smectic A 8CB surface (F_static = 16.2 μN) with F_static > F_dynamic, which is characteristic of the stick-slip mode (i.e., where the water droplet is repeatedly pinned and unpinned as it slides across the LC surface) (14). In addition, we observed signatures of the slippery and stick-slip modes when the 8CB surface was isotropic (no intrinsic order) and crystalline (three-dimensional positional order and long-range orientational order), respectively, which is strongly analogous to the behavior of isotropic materials (e.g., the liquid and amorphous solid states of wax, respectively) (41). Considering the fact that the fluidity of smectic A 8CB is much lower than that of nematic or isotropic 8CB, we hypothesize that the viscous smectic A 8CB wetting ridge and wrapping layer can affect the F_static of water droplets on LC surface. This phenomenon is shown in fig. S5, where we observe that a fully developed wetting ridge and wrapping layer of smectic A 8CB increase F_static compared to water droplets with partial wetting ridges and wrapping layers of smectic A 8CB.

The above results, in combination with degenerate planar anchoring of 8CB at the interface between water and 8CB in both the nematic and smectic A phases, lead us to conclude that the frictional behavior (Fig. 2D) of water droplets at LC surfaces is substantially affected by the contact angle hysteresis of water droplets at the surface (fig. S4B). We also conclude from the results in fig. S5 that the frictional behavior of water droplets is affected by both the wrapping layer and wetting ridge of LCs around water droplets, which strongly depend on the positional order of LCs, despite them having no noticeable impact on droplets’ wetting behavior (i.e., contact angles). It is well established that surface roughness substantially affects the pinning of the contact lines at the surface and thus the contact angle hysteresis (15, 42–44). Previous research has reported that the surface roughness of nematic and isotropic 8CB is on the order of angstroms (45). Meanwhile, the depressions caused by focal conic domains (37–40) increase the surface roughness of smectic A 8CB to the order of nanometers and above. These results lead us to propose that the high surface roughness of smectic A 8CB induces larger contact angle hysteresis and more severe pinning of water droplets at its surface relative to those observed in the nematic phase. Therefore, we attribute the LC mesophase–dependent frictional behavior to the intrinsic surface roughness of the LC in different phases.

In addition to heat, we sought to use light to toggle the LC surface’s phase, thereby allowing us to remotely manipulate the mobility of water droplets on LC surfaces. Past studies have shown that the ultraviolet (UV)–induced trans-to-cis isomerization of doped azobenzene disrupts the LC’s molecular order (46). To take advantage of this behavior, we added azobenzene to 8CB at 1 wt % concentration, which resulted in a −2°C shift in the smectic A–nematic (SmA-N) phase transition temperature upon exposure to UV radiation (Fig. 3A). When the LC surface was periodically exposed to UV light (where one cycle consisted of 20 s of exposure followed by 10 s of darkness), the azobenzene-doped 8CB phase switched repeatedly between smectic A (stick-slip mode) and nematic (slippery mode) phases causing droplets to repeatedly slide and become pinned on an LC surface inclined at 3° (Fig. 3B and movie S3). We note here that the temperature of the LC surface was fixed at 30.5°C, which was halfway between 8CB’s SmA-N phase transition temperature with UV exposure (29.8°C) and without UV exposure (31.7°C). Next, we sought to pattern the UV radiation to program the positioning of water droplets at LC surfaces. As shown in Fig. 3C, water droplets at portions of the LC surface covered with an aluminum foil photomask (which was not exposed to UV and hence remained in the smectic A phase) remained stationary, whereas water droplets on the UV-exposed LC surface (which transitioned to nematic phase) slid to the edge of the region covered by the photomask. Last, we demonstrate that the trajectory of water droplets could be controlled by selectively exposing only certain portions of the LC surface to UV radiation. Figure 3D and movie S4 reveal that the UV radiation pattern could be reversibly written and erased to guide the path of water droplets sliding down an inclined 8CB surface, which deviated from the trajectory droplets that would normally follow because of gravity. Here, we note that from Fig. 3 (A and C), using a tilting angle of only 3° causes droplets to become pinned at SmA-N interfaces. Consequently, to ensure that the droplet would slide along the SmA-N boundary (Fig. 3D), we increased the tilting angle of the LC surface to 30°.

Fig. 3. LC positional order–dependent static friction of water droplets at LC surfaces.

(A) SmA-N transition temperature of azobenzene-doped 8CB and the corresponding sliding angles of 3-μl water droplets with and without 365-nm-wavelength UV exposure. The horizontal red dashed line indicates the substrate temperature set in other experiments. Azobenzene is doped at 1 wt % based on 8CB. Error bars represent SDs and n = 3 for each data point. (B) Plot and sequential photographs showing the displacements of 3-μl water droplets at an inclined 8CB surface upon periodic UV radiation; see movie S3. Blue and orange markers indicate the data with and without UV radiation, respectively. (C) Photographs showing manipulation of 3-μl water droplets’ positions at an inclined 8CB surface through patterned UV exposure. The water droplets were dyed for better imaging. (D) Photographs showing reversible erasing and writing of UV radiation patterns to program the path of 3-μl water droplet sliding down an inclined 8CB surface; see movie S4. White dashed arrows indicate the trajectory of water droplets at the inclined 8CB surface. The tilting angle of the 8CB surface is 3° (A to C) and 30° (D). Scale bars, 2 mm (B) and 5 mm (C and D). Photo credit: Yang Xu, Jen-Chun Fang, and Rajdeep S. Mamtani, The Ohio State University.

Orientational order–mediated cargo release at LC surfaces

Besides tunable droplet mobility, another important feature for the design of open surface microfluidics is the ability to manipulate the chemical compositions of resting droplets, which, to the best of our knowledge, has not yet been achieved using isotropic lubricant–based SLIPS. Here, we sought to achieve this functionality by taking advantage of elastic force barriers arising from the inherent orientational order of the LC, which can prevent contact between guest cargo (e.g., colloids and droplets in bulk LCs) and the boundaries of the nematic LC (24, 47–50). First, we coated an inverse emulsion of well-dispersed ethyl orange (EO) aqueous microdroplets in 8CB onto the 8CB-swelled porous polyRM257 substrate. When a millimeter-sized water droplet was placed at the surface, the EO microdroplets (used as “cargo” for release) concentrated in the wetting ridge around the water droplet but did not enter the droplet. However, upon the nematic-isotropic (N-I) transition, we observed that more than 90% of the encapsulated EO microdroplets continuously released into the water droplet, as shown in Fig. 4A and movie S5. Besides EO, we observed that this phenomenon is generalizable to a range of water-soluble dyes such as rhodamine B, methyl orange, and methylene blue (see fig. S14, G to I). We also observed that release of EO microdroplets to water droplets could be activated by UV radiation, as shown in Fig. 4B and fig. S14F. In addition, compared with the N-I transition, the SmA-N phase transition does not cause notable cargo release (Fig. 4C), suggesting that cargo release is decoupled from the manipulation of droplet mobility at LC surfaces.

Fig. 4. LC orientational order–dependent cargo release at LC surfaces.

(A) Scheme and sequential photographs showing the thermally triggered release of encapsulated aqueous EO microdroplets (referred to as cargo and represented by orange spheres) to a 10-μl water droplet at an 8CB surface; see movie S5. Black arrows indicate cargo microdroplets becoming concentrated in the droplet’s wetting ridge. Scale bars, 5 mm. (B) Plots showing the release of EO microdroplets triggered thermally (crosses), in the presence of 10 mM CaCl₂ (circles), and by exposure to UV (triangles). In the charge-triggered release experiment, a 5-μl water droplet was placed on the LC surface at 0 min. At 60 min, another 5-μl aqueous droplet containing 20 mM CaCl₂ was added to obtain a 10-μl aqueous droplet with 10 mM CaCl₂ on the LC surface. In the UV-triggered release experiment, UV radiation caused an N-I transition of 1 wt % azobenzene-doped 8CB at 38°C. One square centimeter of the 8CB surface contains ~2.6 mg of loaded EO aqueous microdroplets. (C) Plot showing decoupled cargo release and sliding angle of water droplets at an 8CB surface. Error bars represent SDs and n = 3 for each data point. (D) Calculated net force (F_net) acting on a 10-μm-diameter cargo microdroplet encapsulated in an LC surface as a function of surface-to-surface distance between the cargo microdroplet and the millimeter-sized water droplet or bulk water (see fig. S15). Photo credit: Yang Xu and Jen-Chun Fang, The Ohio State University.

Next, we present a thermodynamic model to provide additional insight to the role of orientational order in the process of cargo release in our system. We reason here that the capillary force induced by the curved capillary menisci (i.e., the wetting ridges) surrounding the water droplets concentrates the cargo microdroplets in the wetting ridge and the capillary force alone is unable to overcome the LC elastic barrier in the nematic phase (Fig. 4A). Therefore, we modified the classic Derjaguin-Landau-Verwey-Overbeek model by combining the van der Waals force (F_vdW), capillary force (F_cap), electric double layer force (F_edl), and LC orientational order–originated elastic force (F_el) into a thermodynamic model to describe the essential behaviors of millimeter-sized water droplets interacting with LC surfaces during the process of activated cargo release. The net force (F_net) acting on the cargo microdroplet can be written as (see the Supplementary Materials for details)

$$\begin{array}{c}{F}_{\text{net}}=F_{\text{vdW}}+F_{\text{cap}}+F_{\text{edl}}+F_{\text{el}}=-\frac{A_{\text{water-LC-water}}{R}_{\text{cargo}}}{6x^2}\\ -\frac{2{\mathrm{\gamma }}_{\text{LC}}\mathrm{\pi }\mathrm{\xi }{R}_{\text{cargo}}^3}{{(R+x)}^2}+64{\mathrm{\pi }\mathrm{\epsilon }}_0{\mathrm{\epsilon }}_{\text{LC}}{(k_{\mathrm{B}}T/e)}^2\text{tanh}(\mathit{ze}{\mathrm{\psi }}_{\text{cargo}}/4k_{\mathrm{B}}T)\\ \text{tanh}(\mathit{ze}{\mathrm{\psi }}_{\text{drop}}/4k_{\mathrm{B}}T)\mathrm{\kappa }{R}_{\text{cargo}}\text{exp}(‐\mathrm{\kappa }x)+\frac{{\mathrm{\alpha }}^2\mathrm{\beta }\mathrm{\pi }K{R}_{\text{cargo}}^4}{{(R_{\text{cargo}}+x)}^4}\end{array}$$ (4)

in which A_{water-LC-water} is the Hamaker constant for the interaction between two water phases across the LC, R_cargo is the radius of the cargo microdroplet, x is the surface-to-surface distance between the cargo microdroplet and millimeter-sized water droplet, ξ is a coefficient to estimate the average hydrostatic pressure acting on each hemisphere, ε₀ is the vacuum permittivity, ε_LC is the relative permittivity of the LC, k_B is the Boltzmann constant, T is the temperature, z is the valence number of the dominant aqueous ionic species, e is the elementary charge, κ⁻¹ is the Debye length, and ψ_cargo and ψ_drop are the zeta potentials of a cargo microdroplet and a millimeter-sized water droplet, respectively. α and β denote material constants of the LC, and K denotes the Frank elastic constant of the LC.

To further understand the characteristic behaviors of our LC surfaces, we plot F_net as a function of the surface-to-surface distance between the cargo microdroplet and millimeter-sized water droplet in Fig. 4D. From this result, we make three key comments. First, the red curve in Fig. 4D (and fig. S15C) shows that at a nematic LC surface with a millimeter-sized droplet of pure water deposited on the surface, the repulsive F_el and F_edl lead to a kinetic barrier (F_net ~ 25 pN) that prevents the ejection of cargo microdroplets, which is consistent with our experimental observation that cargo microdroplets do not release into water droplets at a nematic LC surface.

Second, our thermodynamic model predicts that when the LC film is immersed in pure water, the absence of the curved wetting ridge (F_cap = 0) leads to an entirely repulsive force barrier (black curve in Fig. 4D and fig. S15F). Consequently, our model predicts that no cargo release will occur from a noncurved isotropic LC film to water. This prediction agrees with a recent study showing that cargo release from flat LC films into water is driven by a propagating N-I interface across the LC as the LC undergoes the N-I phase transition, leading to continuous cargo release during the phase transition, followed by no cargo release once the LC is entirely isotropic (27). Our model hints that the long-distance F_cap caused by the curved capillary meniscus of the LC wetting ridge provides an additional driving force to transport the cargo microdroplets to the wetting ridge and activate the subsequent cargo release, which is consistent with a previous report (51) and our own experimental observations.

Last, at an isotropic LC surface (F_el = 0; as shown in the blue curve in Fig. 4D and fig. S15D) with a millimeter-sized droplet of pure water deposited on the surface, the absence of the repulsive F_el allows F_net to become negative (attractive), thus overcoming the repulsive F_edl and releasing the cargo into the water droplet. This result confirms that the N-I transition (induced by either heat or UV) can activate the release of cargo microdroplets at the isotropic LC surface. These results lead us to conclude that, rather than LC’s positional order, the LC’s orientational order plays an essential role in cargo release processes, which enables a new class of LC-based open surface microfluidic platforms that can control droplet mobility and chemical composition on demand.

In addition to accurately modeling cargo release induced by phase transitions, our thermodynamic model predicts that the presence of cations (e.g., Ca²⁺) in millimeter-sized droplets would create an attractive F_edl acting on the cargo microdroplet (fig. S15E), which could, in principle, activate the release of cargo (see movie S5). As shown in Fig. 4B, we observed that this was the case with the presence of Ca²⁺ providing similar cargo release behavior as UV light and temperature changes. Next, we investigated the effect of surface anchoring of the LC at the aqueous-LC interface on cargo release. When an aqueous droplet containing 3 mM SDS was placed on a LC surface encapsulated with SDS-stabilized cargos, no cargo release was observed because of a repulsive electric double layer force, as shown in fig. S16. In contrast, when a dodecyltrimethylammonium bromide (DTAB) aqueous droplet was placed on the same LC surface, we observed release of SDS-stabilized cargo to the DTAB aqueous droplet caused by an attractive electric double layer force. These results lead us to conclude that (i) the LC surface can be applied to water droplets containing surface active materials without hindering cargo release processes and (ii) the effect of LC anchoring on the cargo release can be negligible compared with electric double layer forces. As the presence of cations does not promote a phase transition in LCs (and hence should not affect the mobility of droplets), this observation suggests that the presence of cations could be exploited to create a novel orthogonal control over cargo release and droplet mobility, which we explore in the next section.

Removal of heavy metal ions from water droplets at LC surface

Over the past few decades, the monitoring and removal of heavy metal ions that have been shown to cause physical, muscular, and neurological diseases from water have become increasingly important for social and environmental sustainability (52). Although a variety of polymeric surfaces with coordinating groups (e.g., amino, thiol, and carboxyl) have been designed to capture heavy metal ions in water (20, 21, 53), the precipitation of heavy metal ions causes droplets to become severely pinned to surfaces due to the intrinsic coupling between droplet mobility and chemical composition at these surfaces, which prevents further heavy metal ion capture and thus lowers the removal efficiency of the polymeric surfaces. In our final set of experiments, we sought to address this limitation of prior open microfluidic systems by creating LC surfaces to detect and capture heavy metal ions without affecting the mobilities of contaminated droplets at the LC surface. As shown in Fig. 5 (A and B) and movie S6, when 5 mM Pb²⁺ droplets were in contact with an 8CB film loaded with aqueous cargo droplets of sulfide anions (S²⁻), S²⁻ was released into the droplet where it reacted with Pb²⁺, forming black PbS precipitates. Concurrently, the aqueous droplets carried the PbS precipitates off the LC surface without any measurable decrease to the droplets’ sliding velocities. We note here that a concentration of Pb²⁺ of ≥10 parts per million (ppm) can trigger the release of S²⁻ from the LC surface. Our LC platform can efficiently remove Pb²⁺ with initial concentrations up to 50 mM while still ensuring that the mobility of droplets remained unchanged. As shown in Fig. 5C, LC surfaces can be used to repeatedly precipitate 5 mM Pb²⁺ from water droplets for more than 500 cycles while still maintaining a removal efficiency greater than 95% (where removal efficiency is defined in Eq. 7).

Fig. 5. Removal of heavy metal ions using LC surfaces with decoupled liquid mobility and cargo release.

(A) Scheme showing the removal of heavy metal ions at an Na₂S-loaded nematic 8CB surface. (B) Plot showing sliding velocities and removal efficiencies of Pb²⁺ in 5-μl aqueous droplets as a function of position on the nematic 8CB surface. Inset shows the corresponding photograph of Pb²⁺ water droplets sliding at an inclined Na₂S-loaded nematic 8CB surface. The time that droplet initially touched the surface was set to 0 s. Scale bar, 5 mm. (C) Plot showing the reusability of nematic 8CB surfaces for removal of Pb²⁺ in 5-μl aqueous droplets. The Na₂S-loaded 8CB surface is 7.5 cm by 5 cm, and the surface tilting angle is 12°. The initial Pb²⁺ concentration is 5 mM. (D) Concentration of leftover heavy metal ions in water droplets after removal by Na₂S-loaded nematic 8CB surfaces. Inset shows the corresponding removal efficiency of heavy metal ions. The initial heavy metal ion concentration in water droplets is 5 mM. Error bars represent SDs and n = 3 for each data point. The temperature was set to 35°C.

Besides Pb²⁺, we found that this approach could be generalized to precipitate a wide range of heavy metal ions from aqueous droplets, including Cd²⁺, Fe³⁺, Ag⁺, and Cu²⁺. In all cases, the concentration of leftover heavy metal ions in aqueous droplets was measured to be less than 1 ppm (<5 μM; initial concentrations were 5 mM), as shown in Fig. 5D. To the best of our knowledge, this is the first demonstration of an open surface microfluidic platform capable of efficient removal of heavy metal ions in water without causing droplets to become severely pinned. Overall, these results unmask new principles by which the droplet mobility and chemical composition can be manipulated at interfaces on demand and hint at principles by which novel functional open surface fluidic systems can be designed for efficient detection and treatment of toxic substances in water.

DISCUSSION

In conclusion, our results demonstrate that the functionalities of LC surfaces can be decoupled as a result of the unique molecular orders intrinsically offered by LCs. Specifically, the positional order of LCs determines the force of static friction acting on droplets on LC surfaces. Moreover, we have experimentally and theoretically shown that cargo release from an LC surface to a droplet is strongly affected by both mesogenic orientational order and the droplet’s wetting ridge. More broadly, we believe that these results hint at new designs of LC surfaces that will expand the potential utility of conventional isotropic material–based open surface microfluidics into the realm of smart surface reactors and water treatment. Recent research related to active LCs (54, 55) suggests another promising direction that explores how dynamics of active matter encapsulated inside thermotropic LC surfaces influence the behavior of water droplets. Recent studies have also reported the use of biocompatible LC materials for tissue regeneration (56, 57) and the alignment of human cells (58) and bacteria (59), which suggests that our LC surfaces could be used for in vivo drug delivery. Future efforts will seek to explore droplet mobility and activated release at the surface of other LC mesophases, including the cholesteric and blue phases. Last, on the basis of recent study of dynamics of multiple droplets on isotropic surfaces (60), we anticipate that intrinsic molecular order embedded in LC surfaces will offer new avenues to manipulate multidroplet motion and interactions.

MATERIALS AND METHODS

Materials

The following LC monomers were purchased from Jiangsu Hecheng Advanced Materials Co., Ltd.: 5CB, 8CB, E7, and RM257. The following chemical compounds were purchased from Sigma-Aldrich: water-soluble dyes (EO, rhodamine B, methyl orange, and methylene blue), SDS, DTAB, DMOAP (42 wt % in methanol), photoinitiator 2,2-dimethoxy-2-phenylacetophenone (DMPAP), salts (CaCl₂, Na₂S, AgNO₃, PbNO₃, CuSO₄, CdCl₂, and FeCl₃), poly(pyromellitic dianhydride-co-4,4′-oxydianiline) amic acid (PAA), and 1-methyl-2-pyrrolidinone. Azobenzene was purchased from Fisher Scientific. Anhydrous ethanol was purchased from Decon Labs Inc. TEM grids (G75-Au; 3.05 mm in diameter) were purchased from Electron Microscopy Sciences. Water used in all experiments was purified using a Milli-Q water purification system (Simplicity, C9210). Plain microscope slides (25 mm by 75 mm by 1 mm) were purchased from Thermo Fisher Scientific. Unless stated otherwise, purchased chemicals were used as received without further modification or purification.

Preparation of DMOAP-functionalized glass slides

First, glass slides were rinsed with water and ethanol and dried under a stream of nitrogen gas. Then, the cleaned glass slides were placed in a 1% (v/v) DMOAP water solution (120 ml) for 15 min. Next, the glass slides were washed first with water and then with ethanol to remove unreacted DMOAP molecules. Afterward, the DMOAP-functionalized glass slides were dried using nitrogen gas. These slides were stored in a dark room at ambient pressure and temperature to prevent light from damaging the DMOAP coating.

Preparation of polyimide-coated glass slides

We used a Laurell WS-650Mz-23NPPB spin processor to spin-coat a mixture of PAA (10%, v/v) and 1-methyl-2-pyrrolidinone (90%, v/v) on a cleaned glass slide at 4000 rpm for 2 min. The glass slide was then heated to 350°C for 3 hours to coat the glass with polyimide. Next, we used a velvet cloth to rub the surface of the polyimide unidirectionally 60 times. Last, optical cells were fabricated by pairing two rubbed polyimide glass slides (oriented to induce a 0° twist) with 100-μm-thick spacers.

Preparation of LC-filled TEM grids

A 20-μm-thick gold TEM grid was placed on a DMOAP-functionalized glass slide. Next, each mesh of the TEM grid was filled with 5CB using a syringe to obtain an approximately flat 5CB interface.

Preparation of LC surfaces

To make the LC surfaces, we prepared an LC mixture containing nonreactive LC mesogens (8CB, 90 wt %) and a reactive LC monomer (RM257, 10 wt %). DMPAP was added to the mixture as a photoinitiator at 1 wt % based on the total mass of the LC. Next, 100 μl of the homogenous LC mixture was uniformly spread across a 2.5 cm by 2.5 cm DMOAP-functionalized glass slide. Subsequently, the LC mixture–coated glass was placed under a UV lamp (Spectroline, EA-140; 365 nm) and exposed at 2.0 mW/cm² for 20 min at 35°C to prepare a uniform ~160-μm-thick 8CB-swelled porous structure. Last, we dropcast the same nonreactive LC mesogen (8CB; 80 μl) onto the 8CB-swelled polyRM257 nanoporous structure to form a 130-μm-thick 8CB lubricating film. We fabricated the azobenzene-doped LC surface using the same protocol as reported above for 8CB-infused porous surfaces but using a mixture of azobenzene-doped 8CB (1 wt % azobenzene) instead of pure 8CB. The optical appearance of the 8CB surface with deposited water droplets was recorded using an Olympus BX53 microscope equipped with polarizers.

Characterization of morphology of porous polyRM257 substrates

The porous polyRM257 substrate was synthesized from photopolymerization of a mixture of 10 wt % RM257 in 8CB, followed by extraction of 8CB using ethanol. The sample was cut into 100-nm-thick slices using a Leica Ultramicrotome for TEM imaging. The morphologies of porous polyRM257 substrates were measured using both TEM and scanning electron microscopy (SEM). TEM imaging was performed on a FEI Tecnai 12 TEM, and the images were collected on a Gatan UltraScan charge-coupled device camera. The SEM imaging was performed on an FEI Quanta 200 SEM equipment with an acceleration voltage of 5 kV. The working distance is around 9 mm.

Pore size measurements of porous polyRM257 substrates

We used a Micromeritics ASAP (Accelerated Surface Area And Porosimetry) 2020 system to measure the pressure-dependent nitrogen sorption isotherms of a porous polyRM257 substrate. The sample was degassed at 423 K, and nitrogen sorption isotherms were acquired by equilibrating a series of known doses of nitrogen gas at 77 K (from 0 to saturation pressure) against 0.1 g of porous substrate, which was synthesized from mixtures of 10 wt % RM257 in 8CB, followed by extraction of the nonreactive mesogens (fig. S1C). The size distribution of the pores within polyRM257 substrate calculated by nitrogen sorption isotherm is consistent with that measured using TEM (fig. S1B).

Imaging of LC wrapping layers on water droplets at LC surfaces

We imaged the LC wrapping layer that encapsulated water droplets at nematic 8CB surfaces using a custom-made color interferometer. We illuminated a 3-μl water droplet using diffuse white light-emitting diodes, and the interference patterns were captured using a Canon digital single lens reflex camera.

Goniometer measurement

A KRÜSS DSA 100 goniometer was used to measure contact angles (both advancing and receding) and sliding angles using the sessile drop method, as well as interfacial tensions using the pendant drop method, which calculates interfacial tensions or forces from the shape of a suspended droplet. During these measurements, the subject liquid was pushed through the needle slowly (5 μl/min) to minimize the effect of dynamic forces on the shape of the droplet. Images of the pendant droplet near departure were captured and analyzed using a drop shape analyzer to estimate interfacial forces. A Linkam PE120 Peltier hot stage was used to control the temperature of the LC surface during these measurements. Contact angles and sliding angles were measured using 3-μl water droplets that were deposited on the surface using the liquid dispensing feature of the goniometer. For sliding angle measurements, we placed 3-μl water droplets on the LC surface and tilted the surface at a rate of 1°/min using a built-in command on the goniometer, where the angle of inclination at which a droplet began to slide was its sliding angle. The air-LC surface and water-LC interfacial tensions were measured using the pendant droplet method, and average values were calculated from approximately 10 measurements taken from three different droplets. For water-LC interfacial tension measurements, water was placed inside a quartz cell. The LC was placed in a syringe with a needle tip that was held under the surface of the water for the pendant drop method. A high-resolution camera captured images of these droplets, which were then used to calculate the surface/interfacial tensions using a built-in software.

F_d measurements using a cantilever force sensor

F_d acting on moving aqueous droplets was measured using a custom-made cantilever consisting of an acrylate needle (whose tip was sealed with epoxy) with inner and outer diameters of 300 and 700 μm, respectively. The cantilever was fixed on the goniometer and a Thorlabs PRM1Z8 brushless DC motor rotated the substrate with an angular velocity of 0.1° to 5.0°/s. Droplets were adhered to the side of the acrylic needle 1 to 2 cm away from the center of rotation. This caused the needle to deflect from the equilibrium position when the stage rotated. We recorded this process using a camera on the goniometer, and the deflection of the capillary tube Δx was measured using the open source software Tracker. F_d was calculated using Hooke’s law as

$$F_{\mathrm{d}}=k\mathrm{\Delta }x$$ (5)

where k is the spring constant of the needle, which was measured by placing droplets of different volumes on the horizontally positioned capillary tube and measuring the deflection of the tip position in the z direction (see fig. S8). For F_d measurements on nematic and smectic A phases of 8CB, we used capillary tubes with k values of 17.3 and 23.3 mN/m, respectively. This approach was sensitive enough to resolve changes in F_d as small as 0.1 μN.

Tuning mobility of water droplets on azobenzene-doped 8CB surfaces via UV radiation

We first determined the phase transition temperature of azobenzene-doped 8CB using polarized light microscopy. The temperature was controlled by a Linkam PE120 Peltier hot stage, which allowed us to change the temperature at a rate of 0.5°C/min. To examine the photo-responsive droplet mobility on azobenzene-doped LC surfaces, we kept the azobenzene-doped 8CB surface at 30.5°C and tilted the surface to a 3° inclination. Then, we placed a 3-μl water droplet on the 8CB surface, which was initially pinned to the surface. Next, the 8CB surface was periodically exposed to UV to tune the mobility of water droplets on 8CB surface (see movie S3).

Tunable positioning and trajectory of water droplets on azobenzene-doped 8CB surfaces through patterned UV radiation

We kept the azobenzene-doped 8CB surface at 30.5°C. Dark blue droplets (60 mM methylene blue) were placed at the top of the surface and acted as “rain,” while red droplets (60 mM rhodamine B) were placed at the bottom of the surface in the pattern of an umbrella. The volume of each droplet was 3 μl. Then, we covered the red droplets with an aluminum foil photomask and exposed the entire LC surface to UV light. After 3 min of continuous exposure, we found that the uncovered blue water droplets (rain) had slid down the surface, while red droplets (which were masked with aluminum foil) had not moved.

Tunable trajectory of water droplets on azobenzene-doped 8CB surfaces through patterned UV radiation

We kept an azobenzene-doped 8CB surface at 30.5°C and placed a 3-μl droplet of water upon it, which was initially pinned to the surface. Next, we illuminated the entire LC surface with UV light. The water droplet slid following a vertical trajectory due to the force of gravity. Second, we stopped UV radiation for 10 min and placed a new water droplet (3 μl) on the surface, which was initially pinned. Next, we covered a region of the LC surface with an aluminum foil photomask, which was positioned such that the interface between the exposed and unexposed regions followed a diagonal line relative to the direction of gravity (the dashed arrow in Fig. 3D). We then illuminated the surface with UV light for 3 min, after which we tilted the surface at a 30° inclination. We observed that the water droplet slid down the surface following the diagonal interface between the exposed and unexposed region (see movie S4).

Methods of loading cargo into 8CB surfaces

We mixed 20 wt % of a solution containing 5 mM SDS and 1 mM EO and 80 wt % of 8CB at 35°C to make a water-in-LC inverse emulsion. Next, we dropcast this inverse emulsion onto an LC surface. In addition, the inverse emulsion of EO aqueous microdroplets in 8CB was introduced into rubbed polyimide-coated glass–based optical cells via capillary force, and we used polarized light microscopy to image the ordering of nematic 8CB around the cargos in the optical cell (see fig. S14, A and B).

Cargo release determined by gravimetry

Cargo release at LC surfaces was quantified using gravimetry. Specifically, we first measured the mass of the 8CB-swelled polyRM257 nanoporous structures on a DMOAP-functionalized glass substrate (m₁) using a Mettler Toledo analytical balance. Next, we dropcast cargo-loaded 8CB onto the obtained 8CB/polyRM257 surface and recorded its mass (m₂). The mass of loaded cargo was calculated as (m₂ − m₁) y, in which y is the mass fraction of loaded cargo based on the mass of cargo-loaded 8CB. After activated release of cargos to the water droplets placed at the 8CB surface, we removed the water droplets and measured the mass of the 8CB surface (m₃). The mass of the released cargo was calculated as m₂ − m₃, and the percentage of release of cargo loaded in 8CB surface was calculated as

$$\text{Percentage of cargo release}=[(m_2–m_3)/[(m_2–m_1)y]\times 100\%$$ (6)

UV–visible spectrophotometry measurements

The concentration of dyes in water droplets was determined using a PerkinElmer Lambda 950 UV–visible spectrophotometer. After releasing the cargos into water droplets that had been placed on the 8CB surface, we withdrew 5 μl of aliquots from the droplet, placed them into UV cuvettes, and diluted them with 995 μl of water. Then, we measured the UV–visible absorption spectra ranging from 200 to 800 nm.

Stimuli-activated release of cargo from 8CB surfaces

We applied external stimuli (i.e., thermal, charge, and UV) to trigger the release of loaded cargo. For these experiments, all 8CB surfaces were 2.5 cm by 2.5 cm and doped using the inverse emulsion approach described above.

Thermally triggered release

A 10-μl water droplet was deposited on a nematic 8CB surface loaded with EO microdroplets at 35°C. Upon increasing the temperature to 45°C (isotropic), the EO microdroplets, which were initially concentrated in the wetting rim of the water droplet, released into the water droplet (see movie S5).

Charge-triggered release

We deposited 10-μl aqueous droplets of Ca²⁺ (10 mM) onto a nematic 8CB surface at 35°C. As a control, we also placed a droplet of pure water on the LC surface. The Ca²⁺ aqueous droplets turned orange, while the droplet of pure water did not (see movie S5).

UV-triggered release

We mixed a 5 mM SDS solution consisting of 1 mM EO with 1 wt % azobenzene-doped 8CB at 38°C following the previously described protocol. Then, we placed a droplet of water onto this 8CB surface while maintaining the temperature at 38°C. The EO microdroplets released to the droplet after being exposed to UV light (see movie S5).

Removal of heavy metal ions from water at 8CB surfaces

We mixed a 3 mM SDS solution consisting of Na₂S (130 mM) with 8CB (80 wt %). Afterward, we applied the obtained mixture (500 μl) onto an 8CB-swelled porous structure (7.5 cm by 5.0 cm). Next, we prepared a 5 mM aqueous solution of Pb²⁺ and placed 5-μl Pb²⁺–containing aqueous droplets onto the Na₂S-loaded 8CB surface, causing the heavy metal ions to react with S²⁻ anions and precipitate (see movie S6). The removal of other different heavy metal ions, such as Cd²⁺, Cu²⁺, Fe³⁺, and Ag⁺, followed the same procedure. To measure the remaining concentration of aqueous heavy metal ions, we withdraw 3 μl from the aqueous droplet, which was then placed in 997 μl of water and centrifuged at 10,000 rpm for 3 min using an accuSpin Micro 17R microcentrifuge. After centrifugation, the aliquot was collected, and the final concentration of heavy metal ions (c_f) was determined using inductively coupled plasma–optical emission spectrometry (ICP-OES; Agilent, 5800 ICP-OES) at selected wavelengths for different heavy metal ion solutions (328, 214, 223, 238, and 220 nm for Ag⁺, Cd²⁺, Cu²⁺, Fe³⁺, and Pb²⁺, respectively). We then calculated the remaining concentration of heavy metal ions and used the corresponding change in concentration to estimate the removal efficiency

$$\text{Removal efficiency}=[(c_{\mathrm{i}}–c_{\mathrm{f}})/c_{\mathrm{i}}]\times 100\%$$ (7)

where c_i represents the initial concentration of heavy metal ions in the water droplet.

Supplementary Materials

This PDF file includes:

Supplementary Results

Figs. S1 to S19

Table S1

Legends for movies S1 to S6

References

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S6