High-Throughput Fabrication of Size-Controlled Pickering Emulsions, Colloidosomes, and Air-Coated Particles via Clog-Free Jetting of Suspensions

of 2.0 mL min − 1 . Silica capsules were collected in HMDSO bath in a plastic petri dish and were stored under ≈ 15 ° C for a few days until the shell was solidified. The capsules together with a petri dish were firmly attached to an empty ultrasonic bath with double-sided tape. Water was added to the bath and then removed to rinse the sample three times before the releasing test to remove the unencapsulated dyes.


Introduction
Microencapsulation is broadly used for the protection and controlled release of active compounds in pharmacy, [1] life microcapsule shell determines the preservation and release of the active ingredients from the core.
Encapsulating aqueous cores with tunable solid shells is still challenging since there is lack of suitable processing methods to produce controllable capsules at scale. [12] In bulk emulsification, [4] a particle suspension or polymer solution (the precursor for the shell layer), is mechanically mixed with the core liquid to form an emulsion followed by a drying step to evaporate the solvent. This method is scalable and is capable of handling suspensions with high concentrations of particles, but provides limited control over the size and morphology of the capsules. As an alternative, microcapsules with a narrow size distribution and well-controlled morphologies were achieved with microfluidic chips, for particle loadings up to 0.2% (w/v), [13] but the dripping mechanism that forms double emulsions only functions at low throughput (0.3-1.5 mL h −1 ). Jet-based ejection from coaxial nozzles that eject jets into the air [14] or inkjet printing [15] provides high throughput and precisely tunable size and morphology of the capsules. Jetting of dense particle laden fluids has been demonstrated for hard spheres such as glass beads that have no interaction with each other. [16] However, nozzle clogging is still a major problem in formulation of microcapsules, where particles that form networks or jam due to their roughness or stickiness are commonly desired, such as colloidal silica nanoparticles. [17] Clog-free jetting of suspensions, therefore, remains a major challenge on the road to scalable fabrication of controlled microcapsules with particle-based shells.
Here, we present "jetting through a liquid layer" (JetALL) to enable continuous jetting of dense suspensions with particle loading and viscosity up to 10% (w/v) and 1 Pa s −1 , respectively. By enveloping a core liquid jet with a layer of suspension, JetALL enables the production of tailored microcapsules at a per-nozzle throughput over two orders of magnitude higher than chip microfluidics. [18] Without the requirement of flowing nanoparticle suspensions through microscale channels, JetALL broadens the choice of the shell materials for functional capsules. Consequently, more concentrated slurries can be jetted to create tailored microcapsules with controlled functional properties. We highlight this capability by fabricating monodisperse microcapsules with shells composed of biocompatible silica nanoparticles and biodegradable poly(lactic acid) (PLA) polymers. Drying these capsules resulted in air-coated capsules (particle-stabilized antibubbles) for ultrasound-triggered release of a model compound. Figure 1a shows the concept of JetALL. A liquid jet was ejected through a suspension layer. This layer was maintained by capillary forces, preventing its flow from a downward-oriented opening or a reservoir as shown in Figure 1b,c. Upon starting of the jet, a gentle flow of the suspension was applied, resulting in coating of the jet by the suspension, as shown in Figure 1d and Figure S1 (Supporting Information). Controlled break-up of the resulting compound jet was achieved by vibrating the nozzle, which produced monodisperse droplets comprising an aqueous liquid core and a particle-laden shell (Figure 1e). The size of the droplets can be well controlled by the frequency of the vibration. [18] Collecting these compound droplets in a miscible liquid bath resulted in the formation of solid capsules via rapid solvent extraction (Figure 1f). Figure 1g shows the liquid containing core-shell structure in the collecting bath. SEM pictures of the dried capsules in Figure 1h,i show that the shell is assembled by the silica nanoparticles. Furthermore, hybrid shells composed of a biodegradable polymer, PLA, and silica nanoparticles were successfully achieved via JetALL (Figure 1j). To confirm that the capsules contain liquid cores they were broken by mechanical compression, revealing release of the encapsulated liquid (Figure 1k,l).

Results and Discussion
Conceptually, JetALL integrates key elements of jetting from a coaxial nozzle and submerged jetting from a liquid bath. [19] Core-shell nozzles where both the core and the shell liquids are jetted are commonly used to create liquid core-shell templates. However, the narrow orifice that typically separates the outer surface of core nozzle and the inner surface of shell nozzle by a distance ΔR ≈ R inner-nozzle is prone to clogging by particles that block the flow, thereby significantly limiting the composition of core and shell liquid composition. Furthermore, when processing ethanol or acetone with core-shell nozzles in preliminary experiments, rapid evaporation from the small wetted surface caused clogging by precipitation of shell liquid components. The other extreme is a liquid bath or liquid film, where the nozzle is placed just under the interface (ΔR ≫ R inner-nozzle ). [19,20] Here, clogging of the shell liquid is prevented, but a large suspension reservoir is challenging to stabilize. Furthermore, jetting must be directed against the direction of gravity, since the surface tension of the liquid does not contain it to the nozzle surface if ΔR > L cap , with capillary length scale L cap = / g γ ρ ( ) ≈ 2 mm for organic solvents such as ethanol, with γ the surface tension, ρ the density, and g the gravitational acceleration. JetALL eliminates all these disadvantages by forming a microscale jet through a suspension layer that is contained by an open, millimeter-scale reservoir such that R inner-nozzle ≪ ΔR < L cap . In this fashion, directional freedom is ensured, evaporation is limited, and clogging of the nozzle is prevented.
To identify the operational boundaries of clog-free jetting, ejection regimes of compound jets at varying flow rates were visualized in Figure 2a-c. Water was used as a model liquid for the inner jet, and silica suspensions were used to form the liquid layer, extending previous research that focused on jetting through pure-liquid layers without particles. [19][20][21] JetALL enabled jetting of silica suspension as dense as 10% (w/v) with a viscosity of ≈1 Pa s −1 without clogging (see Figure S2, Supporting Information for details on the viscosity measurements). In contrast, a simple nozzle with an orifice of 100 µm clogged within 1 min for particle concentrations exceeding 1% (w/v). Dripping, unstable jetting for a few seconds and then clogging were observed for concentrations of 7% (w/v) ( Figure 2d). However, this challenging 7% (w/v) silica suspension was jetted repeatedly without clogging via JetALL for 20 min (Video S1, Supporting Information), the maximum time that we could supply the liquid from a syringe, indicating an increase of almost 3 orders of magnitude in jetting time. We expanded the processing time of JetALL with a high-performance liquid chromatography (HPLC) Agilent 1100 Series Isocratic Pump that ran for 2.5 h ( Figures S3 and S4, Supporting Information), resulting in ≈600 mL of capsules ( Figure S5, Supporting Information). Here however, clogging happened occasionally in the feeding tube of particle-laden suspension rather than the nozzle. Although beyond the focus of this study, we expect that surface modification to prevent adhesion of particles on the surface of the channel or choosing wider tubing may further improve the operation time. We compare these results to an existing model that describes the number of stably jetted particles by a nozzle before clogging is observed. [22] The number of particles (N) that were stably jetted before clogging was derived (see Supporting Information) and is shown in Figure 2e for the 100 µm nozzle and JetALL nozzle. The model provides a scaling 4 N D ∼ , with D the channel diameter, as shown in Figure 2e. Based on this model, clog is expected only after 2 years for JetALL. Our experiments cannot validate this model result, as the operation time was insufficient and other causes (e.g., clogging of the tubing) were limiting. Still, they are consistent with the observation that choosing large shell nozzles to fabricate small capsules is strongly beneficial for extended operation. Further scale-up could be achieved by parallelization, as shown in Figure S1c (Supporting Information) where multiple nozzles were jetting through one liquid layer simultaneously. Figure 2f-h shows the operation boundaries of the desired jetting regime. A water jet from a 100 µm nozzle was coated by a layer of acetone with a flow rate (Q lay ) from 0 to 20 mL min −1 . Stable compound jet and monodisperse compound droplets formed at low Q lay . When Q lay increased the compound jet became unstable until only dripping was found (also see Figure S6, Supporting Information). Similar results were obtained when the flow rate of the inner jet was varied, as indicated by the phase diagram in Figure 2i. Remarkably, the inertia of the inner jet can drive a liquid layer with a flow rate that is almost 10 times higher than the inner jet, ensuring a large and robust operational window for JetALL.
The next step is to analyze the minimal flow rate of the inner jet Q in for which the transition from dripping to (desired) jetting of the compound jet is observed. As the liquid layer has  Interfacial phase separation between the core and the shell was selected to solidify the shell precursors to form capsules due to its fast and tunable solidification speed. b) Scheme and c) pictures of the nozzle design of JetALL. A 3D-printed small reservoir was used to hold the liquid layer. The liquid layer of JetALL is stabilized in the reservoir by its surface tension. In (c), (i) and (ii) show the overview and the cross-section of the nozzle while (iii) and (iv) show the bottom view of the reservoir and the inner capillary nozzle. d) Fluorescence image confirms the successful formation of the compound jet by the JetALL. The inner jet is water and outer layer is acetone with Rhodamine 6G. The jet was visualized under UV light. e) Microscopy image of a compound jet and monodisperse compound droplets generated by JetALL. The formation of menisci indicates the jet was successfully coated. f) Microscopy image of microcapsules with silica shell fabricated by JetALL. g) Microscopy picture of the detailed core-shell structure of the capsules in oil. h,i) SEM images at different magnification indicate that the shell of the capsules is composed of silica nanoparticles. j) Microscopy image of capsules with a hybrid shell composed of PLA and silica. k,i) Microscopy images of capsules before and after the shell breaks by mechanical pressing. The released water was visible after breaking, indicating that the capsules successfully encapsulated the liquid. a diameter ≈40 times larger than the diameter of the nozzle, the initial velocity of the liquid layer is negligible compared with the velocity of the inner jet (see calculation in Supporting Information). Therefore, we assume that the momentum flux of the compound jet equals that of the inner jet. The velocity of the compound jet is then calculated as where D in is the diameter of the inner jet, which equals the inner diameter of the inner nozzle (see Supporting Information for the detailed calculation). For single-liquid jets, the transition from dripping to jetting occurs for a Weber number (the ratio of inertia over surface tension) We = 4. [23] The Weber number of the compound jet is calculated as We where ρ comp is the density of the compound jet assumed to equal the density of water (see Supporting Information). As indicated by the line in Figure 2i, the threshold We comp > 4 reasonably describes the transition from dripping to jetting for our compound jets produced via JetALL.
To stabilize the droplets generated by JetALL and transform these droplets into microcapsules or colloidosomes, merging and coalescence must be prevented by rapid formation of a stabilizing layer outside the ejected droplets. Here, fast solidification of the shell precursor by interfacial phase separation between the core droplet and the shell liquid was selected. Figure 2j-m show the influence of the Q lay and the solvent composition on the capsule formation. As was illustrated in Figure 1a, JetALL was applied to generate compound droplets possessing a water core and a shell layer containing silica nanoparticles, PLA or their mixture in solvents that can partially diffuse into both the water and oil phases. The interfacial phase separation was triggered as water-soluble ingredient in the solvent diffused into the water core to pre-solidify those capsules on-the-fly. The in-air formed capsules were collected in a silicone oil, hexamethyldisiloxane (HMDSO), bath to prevent deformation and merging of capsules before the precipitation of the shell was completed. HMDSO was selected as the main collecting bath due to its less hazardous nature and broad application in the cosmetics industry. It was found that the composition of the binary solvent and the flow rate of the shell precursor liquid layer have major influences on the formation of capsules. By tuning those two parameters, we identified the optimal composition and flow rate of the shell precursor liquid layer in JetALL to fabricate the good capsules (Figure 2j). Using  The curve indicates We = 4 as an estimate for the transition from dripping to jetting. [23] j-l) Microscopy images of good capsules (j), broken capsules (k), and aggregated capsules (l). m) A phase diagram of the morphology of capsules at varying acetone fractions and Q lay . Q in was fixed as 2.0 mL min −1 . more than 85% acetone or too low Q lay , the phase separation (and thus precipitation) happened too fast, which resulted in nozzle-clogging and incomplete encapsulation (Figure 2k). Reducing the acetone concentration below 60% resulted in too slow precipitation, which caused merging of capsules upon collection (Figure 2l). The phase diagram in Figure 2m provides an overview of the capsules made under varying conditions. The corresponding capsule morphology is shown in Figure S7 (Supporting Information), revealing that a larger flow rate of the outer phase (containing PLA or silica) results in darker capsules and therefore exhibits a thicker shell. SEM images of the shell are shown in Figure S8 (Supporting Information).
JetALL is compatible with various materials to fabricate capsules with high tunability. As shown in Figure 3a-c, production of highly controllable monodisperse (coefficient of variation < 10%, Figure S9, Supporting Information) water-laden capsules with colloidal silica/PLA shells were achieved. As a showcase, small capsules possessing pure PLA shells with diameters of ≈100 µm were fabricated (Figure 3d). Similarly, a precursor containing a much higher loading of PLA (30% (w/v)) was successfully processed to make the PLA capsules ( Figure 3e). Those capsules are mechanically robust even after being taken out from the collection bath into air (Figure 3f). Furthermore, JetALL allows the self-assembly of silica particles to form Pickering emulsions or colloidosomes, as is shown in Figure 3g. The detailed microscopy image in Figure 3h indicates the silica particles inside the shell layer formed the socalled colloidosome structure. After retrieving the capsules from the bath, the silica capsules form self-standing translucent microreservoirs (Figure 3i). If surfactants are added instead of nanoparticles, simple emulsions are formed as shown in Figure S10 (Supporting Information). JetALL was adapted toward biological compatibility by preventing diffusion of a polar solvent into the aqueous core of the capsule. Figure S11 (Supporting Information) shows phase separation by diffusion of a non-polar solvent (which is immiscible with water) into a surrounding third liquid that was jetted onto the core-shell precursors. Solvents can be prevented altogether by forming a shell with hydrogel precursors such as alginate, as shown in Figure S12 (Supporting Information). Remarkably, the symmetry and thickness of the shell layer were tuned by introducing a side jet for in-air collision of the jetted compound droplets (schematic shown in Figure 3j). Compared with symmetric capsules obtained without the side jet (Figure 3k), the asymmetric shell has a much thinner side that could be leveraged, e.g., for accelerated release (Figure 3l). A high surface tension side jet will create an opening on the collided capsules (Figure 3m), which can for example be used in self-propelled micro-robotics. [24] As a final example, we demonstrate that JetALL enables the formation of hydrophobic shells composed of nanoparticles, resulting in ultrasound-triggerable capsules with an air coating that protects their core (illustrated in Figure 4a, also called particle-stabilized antibubbles [25] ). We jetted and collected concentrated suspensions of hydrophobic silica nanoparticles around an aqueous core without clogging. 3000 capsules per second were jetted, as calculated in the Supporting Information. The capsules were dried (Figure 4b) and immersed into water, resulting in an air layer outside the capsules (Figure 4c).
The capsules were stored in a freezer to slow down the evaporation, as water in the aqueous core could evaporate through the porous shell ( Figure S13 Supporting Information). The formation of the silver-color reflection indicates a stable air layer formed outside the capsules (Figure 4d). The resulting air-shell was used to fully isolate the encapsulated liquid core, as illustrated in Figure 4e. The releasing results in Figure 4f,g show this air layer successfully protected the liquid core containing red food dye for up to 72 h of contact with water. Afterward, sonication was used to trigger the release of the aqueous core encapsulated by only hydrophobic silica nanoparticles as the shell (Figure 4h). As shown in Figure 4i, during sonification, the air layer of the capsules was released, forming large visible air pockets. In the meanwhile, the silica shell was ruptured to release the encapsulated liquid core. UV-vis spectroscopy confirmed both the near-perfect containment of the cargo for 72 h, as well as the sudden release of cargo due to the rupture of air-shell upon ultrasound exposure (Figure 4j and Figure S15, Supporting Information). Alternatively, adding ethanol also triggered the rupture of the air-shell and therefore released the cargo, by reducing the surface tension of the water ( Figure S16, Supporting Information). Notably, by adding PLA, the hydrophobicity of the shell reduces (Figure 4k and Figure S14, Supporting Information) and, as a result, the air layer will collapse after 1 h (Figure 4l,m), providing the possibility to tune the release of the core.

Conclusion
By jetting nanoparticle suspensions without clogging, JetALL enables high-throughput production of well-controlled microcapsules including PLA shells, liquid marbles [26] and colloidosomes. [27] The nanoparticles that form the shell of these capsules provide them with unique properties, however, their scaled production was out of reach (Table S1, Supporting Information). JetALL fills this gap as a clog-free method to jet suspensions with high throughput and high concentrations of particles (up to 10 wt.% silica) and polymers (up to 30% (w/v) PLA). The operating conditions to form a stable compound jet by JetALL were systematically investigated. A fast solidification system based on the interfacial phase separation was developed to make capsules with an aqueous core that was robust enough to be dried in the air. As a highlight, we demonstrate that JetALL could move the promising concept of antibubble with air shells towards its applications in controlled release. Similarly, well-controlled antibubbles may enable CO 2 electrolysis for renewable fuels [28] or adsorption of volatile organic compounds. [29]

Experimental Section
Setup Design: The inner nozzle was a glass capillary glued onto a syringe needle (with an inner diameter of 340 µm), the outer holder was 3D printed (FORMIGA P101, Nylon). The needle was inserted into the holder to assemble the JetALL nozzle. The needle was connected to a syringe (core liquid via a plastic disposal syringe, and shell suspension via a glass syringe) that was pushed by syringe pumps (New Era NE-1000 and NE-300). The jet from the syringe was visualized by a camera (IDS, UI-1240ML-M-GL) with a lens (HAYEAR, HY-300XA) with bright-field illumination. To generate monodisperse droplets for producing particles, a piezo element was attached to the nozzle to generate vibration in the direction perpendicular to the jet. The piezo element was driven by a signal of the sine wave from a function generator and a high voltage amplifier. The piezoelectric element was turned on after the syringe pump was on and the frequency was tuned until a train of monodisperse droplets was visualized by the camera. Typically, frequencies within the range of 0.5 to 6 kHz were applied.
Nozzle Clogging Test: Hydrophobic colloidal silica particles (AEROSIL R972 Phama) suspended in acetone (Sigma-Aldrich, ACS reagent) were prepared as suspensions with concentrations varying from 0.5% (w/v) to 10% (w/v), to test the clogging of nozzles with varying configurations and diameters. The suspension was dispersed in an ultrasonic bath for   The air layer was clearly visualized in the high-magnification picture, confirming the formation of an air-shell that tolerates the water pressure. e) Scheme showing a perfect air-shell outside superhydrophobic silica shells, that will fully isolate the encapsulate cargo from the environment. f,g) Microscopy images of underwater silica microcapsules. Encapsulated Rhodamine B fluorescent dye was kept into the capsule after 72 h of immersing in the water, indicating that full air-shells were formed. h) Scheme of the ultrasound-triggered release system. Microcapsules with air-shells attached onto a Petri dish were immersed into an ultrasound bath. The release process of liquid core was triggered by the ultrasound and was recorded. i) Microscopy images showing the ultrasound-triggered release of encapsulated red dye from the air-shells in (g). j) UV-vis spectroscopy curves of the water media loaded with silica capsules containing Rhodamine B for varying times. There was only a very minor release after 3 days, indicating the air-shell isolated the aqueous core from the aqueous environment. k) Scheme showing air-shells containing hydrophilic defects caused by polymer aggregates. l,m) Microscopy images demonstrating the release of encapsulated dyes in PLA-silica particle shells within 1 h after immersing in the water. form a shell that appears more solid-like layer. JetALLstill operated as normal with this layer. To prevent any potential influence, this layer was removed simply by rinsing the nozzle with acetone without stopping the jet (see Figure S17 (Supporting Information) for more details).
Preparation of Capsules: A suspension or solution was pushed through 360 µm PEEK tubing connected to the 3D printed nylon holder to form the shell layer while water (with food dye) was pushed through the PEEK tubing connected to the capillary as the inner nozzle at varying flow rates to form the aqueous core. The generated capsules were collected with HMDSO with a distance of 40 cm unless otherwise noted. Following suspensions or solutions were prepared to fabricate varying types of capsules.
Silica capsules (Figure 1g-i, Figure 3g-i, Figure 3f-h,m-p): 7 g of silica (AEROSIL R 972, EVONIK) was mixed with 10 mL Tetrahydrofuran. The silica product was used as received. It was a hydrophobic fumed silica modified with dichlorodimethylsilane. These fumed silica particles typically was of mean primary particle size of 5 to 50 nm and form aggregates with size between 4 and 200 µm measured by laser diffraction method after being dispersed in the air stream. After shaking, the mixture was put into an ultrasound bath for 10 min to disperse the silica particles.
PLA capsules (Figure 3d-f,k-m): 30% (w/v) PLA (Ecosoft 608, Micro Powders Inc.) for cosmetic application was dissolved in the binary solvent of DCM and hexane with the ratio of 1:1 to prepare the capsules in Figure 3e. For all other samples, 10% PLA was dissolved in the binary solvent of acetone and xylene with the ratio of 9:1 to prepare the capsules.
Hybrid capsules (Figure 1j-l, Figure 2j-l, Figure 3a-b, Figure 4b-d, j,k): Suspension of 3% (w/v) silica together with 5% (w/v) PLA (Ingeo 4043D, NatureWorks) in a binary solvent of xylene and acetone with the ratio of 7:3 was prepared to fabricate the capsules in Figure 2b, Figure 3c,d. For all other samples, the shell layer was prepared from the suspension of 3% silica with 10% (w/v) PLA (Ecosoft 608, Micro Powders Inc.) in the binary solvent of xylene and acetone with the ratio is 7:3 unless otherwise noted.
Capsules with Air-Shell (Figure 4c,d,f,j): To prepare capsules with an air layer, the silica capsules in hexane were transferred to a petri dish. The excessive hexane was removed by wipe paper to make a concentrated capsule in hexane. The petri dish was put in a freezer set at −15 °C for 2-3 days until the solvent was evaporated. The sample was taken out from the freezer and DI water was added to the Petri dish to submerge the silica capsules to study the formation of air shells. The capsule will tend to stick onto the bottom of the Petri dish.
Capsule Release Study: To study the releasing of the liquid core from the air-shells, food dye was encapsulated into the capsules during the fabrication. The capsules were immersed into DI water in a Petri dish with a diameter of 9 mm. The dish was then fully immersed into an ultrasonic bath. A camera and a light source were put above the ultrasonic bath to visualize the process. The releasing test for the hybrid capsule lasted for ≈1 h until the cores were found released. For the silica capsules, the releasing test lasted for ≈3 days. At the end of the releasing test for silica capsules, the ultrasound was turned on to trigger the release. For the UV-vis spectroscopy measurement, silica capsules were fabricated from 7% wt/v silica in THF with a flow rate of 1.5 mL min −1 . The core liquid was from a solution of Rhodamine B with a concentration of 1 mg g −1 . The core liquid was ejected through a 100 µm nozzle with a flow rate of 2.0 mL min −1 . Silica capsules were collected in HMDSO bath in a plastic petri dish and were stored under ≈15 °C for a few days until the shell was solidified. The capsules together with a petri dish were firmly attached to an empty ultrasonic bath with double-sided tape. Water was added to the bath and then removed to rinse the sample three times before the releasing test to remove the unencapsulated dyes.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.