Fabrication of Bijels via Solvent Transfer‐Induced Phase Separation using Liquid‐in‐Liquid Printing

Bicontinuous interfacially jammed emulsion gels or bijels are a new class of soft materials containing both surfactants and particles. They are, however, distinguished from conventional surfactant‐stabilized and particle‐stabilized systems. Due to the partitioning of percolating sheets of particles in a bicontinuous morphology, the interpenetrating phases are stabilized by kinetically arresting the phase separation process. This leads to the generation of rheological features in bijels with a range of viscoelastic properties based on types, size, anisotropy, and volume fractions of particles. Research on bijels has been around for only about 15 years, and the need for substantial research is recognized in the fabrication of bijels. Herein, the focus is on fabricating on fabricating bijels using solvent transfer‐induced phase separation (STRIPS) and from polar oils, cationic/anionic surfactants, and various nanoparticles to investigate their microstructure–properties relation. Using the newly emerged liquid‐in‐liquid 3D printing approach, the fabrication of these bijels via STRIPS into arbitrary complex designs which are not reported before is shown. The rheological properties of the printed bijels and the baths used for printing are characterized. This work paves the way for the production of bijels using a broader range of materials and hence their application in various fields.


Introduction
Bicontinuous interfacially jammed emulsion gels (bijels) are a new class of soft materials mainly composed of two interpenetrating continuous phases of immiscible liquids arrested by a layer of colloidal components such as nanoparticles that are jammed at the liquid-liquid interface. [1,2]Bijels were first theoretically predicted by Cates and co-workers [1] and then experimentally observed by Clegg et al. [3] The physicochemical and rheological properties of bijels significantly differ from surfactant-stabilized systems and particle-stabilized emulsions decomposition and diffusion of the solvent out of the mixture into the surrounding aqueous phase (Figure 1, insets i,ii). [20]uring the phase separation process by spinodal decomposition, both phases undergo coarsening over time.As they approach their equilibrium compositions, the interface length between the two phases decreases, consequently resulting in an increase in the interfacial tension between them. [3,25]This increase will, in turn, induce the trapping of nanoparticles (with neutral wetting) at the interface to the extent that they form a rigid nanoparticlejammed interface.This would then result in arresting the phase separation process, suppressing the nucleation growth of these two immiscible phases and finally leaving the system in a thermodynamically metastable state. [25]STRIPS method offers advantages over other methods (conventional quenching and direct mixing), such as the freedom to continuously create bijels (rather than bulk production). [16]espite such efforts, fabricating bijels is still limited to specific pairs of surfactant-nanoparticle and binary liquid mixtures with restrictive chemical or physical properties such as high viscosity. [25]In this study, the STRIPS technique is employed to produce bijels using two different oils.The first oil is oleic acid (OA), a polar fatty acid oil (lipid) with carboxylic acid head groups that have never been explored for fabricating bijels.][28][29] While DEP or any other oils in previous studies have been used with negatively charged nanoparticles, this study aims to broaden the use of both positive and negative nanoparticles for bijel fabrication in each oil system.
Expanding bijel fabrications to unexplored materials, especially the possibility of having nanoparticle-surfactant pairs with complementary functionality (i.e., electrostatic charges), opens up new opportunities for creating functional bijels, such as for applications involving diverse biological entities that carry charges.Using these two nanoparticle-surfactant pairs for creating bijels further improves the flexibility of our fabrication platform in terms of component selection.
Apart from material selection, to the best of our knowledge, all the previous studies on the fabrication of bijels have been limited to traditional random bulk fabrication methods or, at best, the generation of structures such as films, microparticles, microropes, or fibers. [20,22,30,31]The current work aims to create bijels (with their bicontinuous microstructures) in predetermined shapes using 3D printing approaches.As the bijel STRIPS method relies on injecting a liquid into another liquid, the newly emerged liquid-in-liquid 3D printing can be considered the most appropriate 3D printing approach.5][37][38][39][40] However, there has not been any report on creating relatively complex bijel constructs using either liquid-in-liquid 3D printing or any other type of freeform fabrication methods.
Figure 1.Schematic illustration of printing constructs with bijel microstructures.The initial homogeneous ternary mixture is composed of water, oil, and solvent (containing oppositely charged surfactants and nanoparticles) with a composition close to the plait point (shown in light blue, inset (i)).Once this mixture is injected into the bath composed of aqueous surfactant solution (containing poly acrylic acid), it undergoes spinodal decomposition (inset (ii)), creating a bicontinuous internal structure with domains enriched in either water or oil.The compositions of these water-and oil-rich phases are illustrated in green and blue colors, respectively.While the bijels are being extruded from the nozzle into the bath, the print head moves according to paths defined in the model, yielding the final print.
The printing approach presented in this study is of an associative 3D printing type since the phase separation and hence, bijel formation are dependent on the interaction of the bath and printing phase (inherent in the STRIPS method). [32]In most prints in this work, the print bath also contains thickener which results in favorable rheology (such as a large enough yield stress and shear-thinning behavior) in order to maintain the position of printed structures (similar to any other embedded extrusion 3D printing techniques), making the current technique to fall under the category of embedded associative 3D printing. [32]This platform can meet the growing demand for computer-aided and freeform fabrication methods (such as 3D printing) to enable bijel production in a controlled manner and into predetermined patterns.Moreover, compared to other soft matter systems explored in (liquid-in-liquid) 3D printing literature, bijel prints offer several unique attributes, such as interconnected phases of hydrophobic and hydrophilic liquids confined into a defined shape with attractive rheological and structural properties. [32]pecifically, the rheological properties of the created bijel constructs in this work possess shear-thinning behavior with solid-like viscoelastic properties while remaining stable over a range of temperatures.

Bijel Production
In this study, two oils are used, OA, which is a naturally abundant monounsaturated fatty acid (lipid oil), and diethyl phthalate (DEP), which is a biodegradable aromatic hydrocarbon oil.For each oil system, two pairs of surfactant and nanoparticle are used that have opposite charges: one with a positively charged nanoparticle (Ludox CL, 12 nm) paired with a negatively charged surfactant (sodium dodecyl sulfate, SDS) and the other containing a negatively charged nanoparticle (Ludox TMA, 22 nm) paired with a positively charged surfactant (cetyltrimethylammonium bromide, CTAB), creating four systems of study in total (hereafter referred to as DEP-Ludox CL, DEP-Ludox TMA, OA-Ludox CL, OA Ludox TMA).The physical properties of each nanoparticle are described in detail in Table 1.For creating bijels in STRIPS, previous studies have shown that the composition of the initial ternary mixture has to be selected near the plait point in order to phase separate into spinodal decomposition (Figure 1, insets i,ii).Otherwise, the mixture does not result in bijel but rather nucleation and coalescence. [20,23]To determine what composition to use for the ternary solutions for each system of study, we have used the underlying phase diagrams for ternary components of water, ethanol, and oil from the literature (water-ethanol-DEP [41] and water-ethanol-OA, [42] Figure S1a,b, Supporting Information).In the case of DEP as the oil phase, the composition that results in the formation of bijel has already been studied and determined in the literature, and we have selected the same point. [20]o locate the appropriate composition in the OA system, we have moved along the constant oil and water lines near the plait point and tested several points on each line (both in monophasic and two-phasic regions) in the ternary phase diagram (Figure S1 and Table S1 and S2, Supporting Information).The prepared ternary mixtures were then injected into the continuous phase composed of the aqueous solution of the surfactant.Similar to DEP, only the ternary mixture close to the plait point (i.e., point OII in Figure S1a, Supporting Information) resulted in bijel structure formation, as evidenced by light microscopy (Figure S1c, Supporting Information); compositions not close to the plait point or located in the two-phasic region did not yield an interconnected bicontinuous network of water and oil.To further confirm the formation of bicontinous structures, confocal microscopy is performed, where Nile red (a lipophilic dye) is added to the oil phase of the ternary mixture.The confocal microscopy images reveal the formation of a well-defined, relatively uniform, micrometer-sized bicontinuous arrangement of oil and water liquid phases (Figure 2).The light microscopy images (Figure S2, Supporting Information) are also obtained, confirming the formation of bicontinuous morphology.Note that the bijel in our system is made in situ and is in uncured and soft gel conditions, where the gravity and contact with the glass bed cause them to flatten.This flattening is a physical phenomenon and occurs over a couple of minutes depending on the mechanical properties of the created bijel fiber and its thickness.Limited by the use of needles compatible with the 3D printer, the bijel filaments created here have a higher thickness compared to the literature on STRIPS (with less than 150 μm-thick fibers), which makes them more susceptible to flattening out.Thus, the confocal images visualize the flattened-out surface of bijels from the bottom of the chamber.Comparing the two nanoparticle systems in this study, one can observe relatively larger domains made in bijels with larger nanoparticles (Ludox TMA) (Figure 2b,d).These larger domains could be due to the reduced ability of the larger nanoparticles to pack closely together, resulting in larger domains.Alternatively, this difference in domain sizes could be attributed to more effective barriers for smaller nanoparticles at the interface compared to the larger particles, which may allow for easier transfer of fluid and hence greater connectivity between the domains, leading to the formation of larger domains.These gels were softer and more difficult to handle than those made by smaller nanoparticles (Ludox CL).Regarding the domain spacings, all bijel systems follow high uniformity in the domain structure.However, looking closely at bijels made from Ludox CL, one can observe a slight decrease in the domains from the middle of the surface toward the sides (Figure 2a,c).In contrast, the trends are opposite in other bijels made with larger Ludox TMA (Figure 2b,d).This opposite trend is likely due to the weaker nature of the gels produced by the larger Ludox TMA nanoparticles and their tendency to get swollen in the bath over time.For DEP-Ludox CL system, the confocal scanning Z-stack of a fiber with a thickness of 27 μm confirmed the formation of water columns along the Z axis in the bijel fiber (Video S1, Supporting Information).It is worth noting that the confocal microscopy for the OA-Ludox TMA (Figure 2d) was challenging, presumably due to the high refractive index of the oil in this system and light scattering.Moreover, due to the dynamic nature of both STRIPS process and LL3DP platform, mass transfer occurring from the extruded ternary mixture toward the bath will be nonuniform, thereby resulting in various phase separation mechanisms (such as spinodal decomposition or nucleation and growth). [20]The variety in phase separation mechanism will lead to a slight degree of asymmetry in the structure, as observed in confocal images with the formation of aqueous droplets with various sizes at the center or the edge of DEP bijel fibers.In fact, the elapsed time due to a long confocal imaging process combined with the fact that created bijels are in soft and uncured leads to a high chance of coalescence events, which can explain the presence of larger droplets in the Z-stack analysis (Video S1, Supporting Information).In some methods, bicontinuous structure can be achieved by the partial coalescence of the nucleated droplets.By reducing the surface area of the coalescing droplets, particles jam at the interface, thereby arresting the coalescence process.As part of the process, many water-rich droplets can be arrested from coalescing and form a percolating structure as a result of nucleation and growth. [25,43]o further shed light on the microstructure of the bijel printed using this platform, scanning electron microscopy (SEM) was performed.Bijels, in their gel state, presented challenges for SEM and, therefore, need to be solidified.Inspired by a method commonly used for STRIPS bijels [22] and following the printing process, the print bath is replaced with a 3 wt% tetraethyl orthosilicate (TEOS) solution in the mineral oil.The immersed prints in TEOS solution were then placed in an oven and heated to 45 °C for at least 48 h prior to SEM analysis.The printed DEP bijels were successfully solidified such that they could be easily handled with a tweezer.The SEM images of the DEP bijels prints (Figure 3) illustrate a sponge-like surface structure comprising a vast network of channels with a size of a few micrometers, comparable to the size of the liquid domains captured in confocal microscopy (Figure 2).Such hierarchical pore structures have also been observed for other STRIPS bijels in the literature. [20,28,29,44]The results from the SEM characterization were in agreement with those from the confocal microscopy results, suggesting that bijels made with the smaller nanoparticles (Ludox CL) exhibit smaller domain sizes, while the bijels fabricated using the larger nanoparticles (Ludox TMA) produce bijels with larger domains.However, it should be noted that the fibers visualized in SEM analysis were placed in TEOS solution immediately after printing whereas for the confocal microscopy, the bijels aged at least 20 min due to the lengthy confocal imaging process, which in turn can increase the likelihood of coalescence.It is also worth noting that the bijels made from OA did not withstand this solidification process and disintegrated into smaller fragments inside the TEOS solution and, therefore, could not be analyzed using SEM.

Printing of Bijels
After bijel microstructures are confirmed in the resulting gels through both confocal and electron microscopy techniques, the same ternary mixtures are used for liquid-in-liquid 3D printing, as shown schematically in Figure 1.In this process, the print head moves according to paths defined in the print model, positioning the bijels filament at specified locations by extruding the homogeneous ternary mixture (at a constant extrusion rate).However, certain challenges remain to be solved, which primarily stem from under extrusion, movement of prints due to structurally weak print bath, or nozzle clogging (Figure S3, Supporting Information).To address these issues, parameters impacting these phenomena and the print success are investigated and optimized.All the printing parameters used in successful or failed prints (along with the reason for their failure) are listed in Table S3, Supporting Information.Furthermore, the rheological properties of the print bath are modified, which will be discussed thoroughly in Section 2.4.Using selected printing parameters, various planar models are printed from all ternary systems, as shown in Figure 4.The printed constructs are complex in shape with various geometrical/structural features (intersecting segments, curved features, bulgy or smooth surfaces, etc.).The complexity of prints was higher in the systems with DEP as the oil phase because its density, unlike OA, is higher than water which makes the created bijels heavier than the bath, resulting in gentle sedimentation of extruded bijel filaments.Regardless, considering the used models, final prints for all systems show high shape fidelity, printing accuracy, and dimensional adherence (Figure 4 and S4, Supporting Information), features not easily found in soft matter printing.
For instance, with the DEP-Ludox CL system, two designs of "bijel" text and the letter "B" are printed (Figure 4, bottom-left and Video S2, Supporting Information).The "bijel" text design not only possesses complexity in shape, but also features overlapping parts between letters where the extruded filament is deposited on the top of an already printed part and merged into the print.
In the designs containing merging structures such as the "bijel" text, the first printed filament already formed a bijel structure because the formation of bijel in this method is upon contacting the ternary mixture with the bath solution and a bicontinuous structure is formed before the printing of the second filament.However, since bijels are gel-based soft structures, they can merge when stacked on top of each other, unlike UVcured solid systems.The merging at the contact areas of these two filaments was suggested through confocal microscopy where a smooth seamless surface can be seen along the first filament length and on both sides of the second fiber (Figure S5 and Video S3, Supporting Information).
The print of the letter "B" also confirmed that a print can consist of discontinued parts and it does not necessarily need to be a continuous cohort part while maintaining the relative distance between the inner part and the outer part (Figure 4).Due to the robust mechanical properties of the bijels, the printed filament is not dragged around and deposited somewhere other than where it was supposed to, which otherwise would compromise the overall shape fidelity.Using the system of OA-Ludox CL, a flower-shaped and a 2D planar spiral construct were printed (Figure 4, bottom-right).The disconnection at the top of the flower print is in accordance with the used model (Figure S4, Supporting Information) and shows the start and end points of the print.In the planar spiral print, by lowering the print head height (closer to the bottom of the Petri dish), the extrusion will follow an alternating pattern, resulting in periodic thickness variations (rather than an even profile) in the deposited filament.
With the DEP-Ludox TMA system, the Statue of Liberty model was successfully printed with excellent details (Figure 4, top-left).In this structure, the printing of some segments is conducted in separate and discrete steps, which demonstrates the possibility of printing discontinuous structures, speaking to the fact this bijel fabrication technique is not limited to continuous printing.In the artery design print, the ability to revisit the print and add segments to already printed constructs is showcased (Figure 4, top-left).The main channel of the artery is printed first followed by the printing of other branches, consequently, from left to right which merges into the main channel to form a single cohesive part that keeps its integrity once the bath phase is drawn from the dish.For the OA-Ludox TMA system, two prints including a star and a meandering channel are demonstrated (Figure 4, top-right).In the star shape, the possibility of acute angles is illustrated for such soft materials specifically in the corners of the print.The starting point and the endpoint at the bottom part of this print did not connect, leaving a small gap, which is most likely due to the delay in the extrusion that commonly occurs at the beginning of the prints.With the meandering channel, the 180°curve within the short distances is printed.For the bijels printed with OA, simpler designs with no merging segments are used since the much lower density of OA (compared to the aqueous bath phase) causes the printed bijel to rise to the bath surface upon extrusion, which makes it challenging to print a fiber onto an already deposited fiber.For all the prints discussed here, a few minutes after printing, the gel bath could be drained from the container and the printed bijel constructs were robust enough to hold onto the bottom of the dish.
Besides the prints shown so far (with planar designs, i.e., parallel to the XY-plane), we have also showcased the printing of a 3D model comprising overhanging parts outside the XY plane and with movement in all three directions (i.e., in freeform manner).For this purpose, a 3D helical spiral with a diameter of 1 cm is selected and printed using a 27 G tapered nozzle for DEP-Ludox TMA system (Figure 5, and Video S4, Supporting Information).Due to the higher density of DEP oil phase in the bijel (compared to the surrounding bath phase), the created 3D structure starts to sink to the bottom of the bath within a minute and loses its 3D spatial configuration.However, the time scale for this deformation is high enough that with a crosslinking strategy such as the incorporation of a photocurable polymer into the initial ternary mixture in future studies, one could lock in the shape of these 3D structures.Furthermore, future work could benefit from minimizing the density difference between two phases as well as using yield-stress granular print baths where Figure 4. Printing of various constructs with planar designs including a bijel text and letter "B" with DEP-Ludox CL system (top-left), a flower-shaped and a 2D planar spiral with OA-Ludox CL system (top-right), a statue of liberty and an artery design with DEP-Ludox TMA system (bottom-left), and a star and a meandering channel with OA-Ludox TMA system (bottom-right).Statute of Liberty and "bijel" text are printed in a 10 cm-diameter Petri dish, while the rest was printed in 5 cm-diameter Petri dishes.Except for the DEP-Ludox CL system, the rest of the baths contains poly(acrylic acid, PAA) as a thickener to enhance the rheological behavior of the bath and improve print quality.the printed filament will be entrapped and locked in place.The omnidirectional printing in 3D space is not possible for OA bijels since the much lower density of OA compared to the surrounding bath phase makes the extruded filament come to the surface and float immediately.
Given the dynamic nature of a liquid-in-liquid printing process like the one presented in this work, it is necessary to investigate the consistency of the bijel microstructures across the prints.Therefore, a simple 2D fiber with the largest length possible (given the μ-slide dimensions) is created from the OA-Ludox CL system, as an example, inside the wells of μ-slides (containing bath solutions).The confocal microscopy images along the edge of this fiber (at locations at least 3 mm apart from each other) confirm the consistency in bicontinuous structures throughout the print (Figure S6, Supporting Information).The provided confocal images are from the different locations of the top fiber.

Rheological Properties of Printed Bijels
The printed bijels are mechanically robust, thereby can be once the bath solution is removed from the Petri dish (by tilting it to one side).The tilted Petri dish (covered with a layer of Parafilm to avoid drying the gel) was rested for 10 min to allow excess bath solutions to flow out of the printed constructs.Then, the resulting gel was carefully placed on the bottom plate of the rheometer.Four oscillatory tests, including amplitude, frequency, time, and temperature sweeps, are performed for rheological characterization.Following amplitude sweep and determining the linear viscoelastic region (LVR) for each bijel, a frequency sweep test was carried out at a strain of 0.1% (within the LVR) to characterize their viscoelastic properties.The frequency sweep results (shown in Figure 6a,c) indicate higher storage shear modulus (G 0 ) compared to loss modulus (G 00 ), suggesting a more elastic behavior than viscous.This higher elastic behavior is also observed for bijels made using quenching approaches [5,45] as well as those made with direct mixing. [18]In comparison with another set of work on bijels rheology where frequency sweep measurements reveal G 0 ~10 0 -10 1 Pa, [46,47] the bijel prints in this study exhibit G 0 ~10 2 -10 4 Pa.Furthermore, the storage moduli for all four bijels were shown to be nearly independent of angular frequency (ω), that combined with the dominant elastic component suggests a solid-like viscoelastic behavior. [48,49]uch behavior has also been reported for other bicontinuous structures such as the ones made from a ternary blend of two polymers and silica particles. [50]Moreover, a local minimum in G 00 curve is observed in frequency sweep results (Figure 6), similar to what is reported in another bijel study, supposedly due to various relaxation processes at short and long time scales, corresponding to high-and low-frequency range, respectively. [47]hese relaxation processes correspond respectively to lubrication phenomena for particles at the jammed interface and escape/ exchange of particles out of the jammed interface. [47,50]The combination of such rheological behavior is typically called "softglassy" [46] and has also been observed for bijels formed by quenching. [47]In the rheology of bijels, two mechanisms could contribute to the elastic behavior (G 0 ) at their equilibrium state: elasticity of the constituent two liquid phases which are typically small (below 1 Pa) and the one arising from the networks of particles at the interface. [45]It has been reported in the literature that bijels with smaller domain size and therefore larger interfacial area exhibit larger elastic modulus, [45] similar to what was observed in this work with Ludox CL bijels showing a higher G 0 in all oscillatory experiments.However, such a behavior might have been mitigated by the slightly higher volume fraction of Ludox TMA nanoparticles in the suspension compared to Ludox CL nanoparticles (Table 1) since the higher particle loading stiffens the interface to to a greater extent, resulting in an increase in G 0 .
To investigate whether artifacts such as drying occurs for the bijels throughout the rheological measurements, an oscillatory time sweep test is carried out as the first test for another set of bijel samples, immediately after loading and trimming of the sample.The results (Figure S7, Supporting information) show constant values for both G 0 and G 00 , illustrating no change in viscoelastic properties over the measured time period and hence, robust stability for bijel prints.Such behavior rules out the chance of any instability like drying during the rheological measurement.To further inspect the possibility of drying during rheological measurements, we have put together the shear moduli curves obtained from two separate frequency sweep tests carried out on the same sample with at least 7 min elapsing between the two and one can see that the curves are almost identical (Figure S8, Supporting Information).This observation further ensures that the bijels do not exhibit any instability throughout the rheological measurements.
In the next test, the variations of G 0 and G 00 with temperature were also investigated using temperature sweep tests.Results (Figure 6b,d) reveal that the printed bijels are stable at various temperatures ranging from 15 to 40 °C (close to physiologically relevant [human body] temperatures).Across the measured temperature range, the storage modulus remains higher than the loss modulus, maintaining the predominantly solid-like behavior. [51]Such lowtemperature sensitivity throughout the entire temperature sweep implies that the bijel microstructures have not been disrupted upon heating to higher temperatures, unlike SeedGels which are another class of soft materials with bicontinuous channels but high sensitivity to temperature changes. [52,53]low sweep tests were carried out for flow behavior characterization and investigating the shear rate dependency of viscosity.The viscosity versus shear rates curves shown in Figure 7a,b indicate shear-thinning behavior, which has also been reported for other bijels. [18]Such shear-thinning behavior spans across the entire shear rate range by over four orders of magnitude decrease in the viscosity.According to previous studies based on microscopy techniques, such shear thinning is associated first with uneven distribution of particles across the interface.With further increases in shear rates, the complete separation of large water/ oil domains (accompanied by unjamming of the particle monolayer) is most likely to be responsible. [13,18]This is consistent with the fact that harsh mixing procedures in bijel preparation cannot yield the formation of bijels, potentially due to destroying the bicontinuous microstructure by high shear rates. [16]Using experimental setups [13] and simulation techniques, [54] it was even shown that the orientation of such domains is in the direction of the force being applied at the bijel interface.At high shear rates, the printed bijels containing Ludox CL nanoparticles have a comparable viscosity to bijels made by direct mixing (and using relatively same size nanoparticles) in another study. [18]At low-shear rate range, however, the Ludox CL bijel prints in this work exhibit at least one order of magnitude higher viscosities. [18]Comparing the flow curves for all four bijel systems in this work (Figure 7a,b) suggests that the OA-Ludox CL has the highest viscosity at almost all shear rates.In contrast, the system with DEP but with Ludox TMA-CTAB is measured to have the lowest viscosity across the entire shear rate range.A consistent order in shear storage moduli of all samples can also be observed (from Figure 6a,c) once averaged across the entire frequency range and over three measurements, as depicted in Figure 7c.The data suggests that the DEP-Ludox CL bijels have the highest value, reflecting a more robust microstructure, potentially due to changes in the size and chemistry of nanoparticles.In contrast, the Ludox TMA-CTAB system shows the lowest shear storage modulus.

Optimization and Rheological Characterization of Printing Baths
When attempting to print bijels in aqueous surfactant solutions, certain difficulties have arisen.In addition to under extrusion or nozzle clogging, discussed in Section 2.2, these include issues with the bijels floating or moving on the surface of the bath and sticking to the dispensing nozzle (Figure S3, Supporting Information).The challenges are related to the insufficient mechanical strength of the aqueous surfactant solution as the printing bath and will be exacerbated when using OA as the oil component in the ternary mixture because of its lower density compared to water (and the bath itself ), which causes the prints to flow on the bath surface during printing.Therefore, it is necessary to enhance the mechanical properties of the printing bath such that it acts as a support for keeping the prints in place, and this can be accomplished by incorporating a thickening agent.To this end, two common thickeners for aqueous systems, namely xanthan gum and poly(acrylic acid, PAA), were used, and the printing of bijels within the resulting baths was repeated.Although attempts to print in xanthan gum baths were unsuccessful, PAA was found to be a more effective alternative, providing greater precision and control over the printing process.Furthermore, through light microscopy, it was confirmed that the presence of PAA in the aqueous solution has no significant impact on the formation of the bicontinuous structure.Also, no fundamental change was observed in the bicontinuous structure except for a slight increase in the domain sizes of the system containing PAA, as demonstrated in the case of the DEP-Ludox TMA bijel (Figure S9, Supporting Information).
Flow sweep measurements were carried out to investigate the origin of the observed difference in the behavior of bath solutions.The flow sweep curves (as shown in Figure S10, Supporting Information) first show a viscosity for the xanthan gum bath that is over three orders of magnitude higher at very low shear rates (close to the equilibrium structure at rest), and that could be the primary reason for the ineffectiveness of the xanthan gum bath.Such high viscosity and, consequently, high resistance to flow prevents the solvent (ethanol) from diffusing adequately from the extruded jet into the bath, thereby compromising the formation of bijel and leading to unsatisfactory printing results.Moreover, by further comparing the curves, a significant difference in their shear-thinning behavior can be seen where the xanthan gum bath (at a low concentration of 1.2 wt%) demonstrates almost four orders of magnitude decrease in viscosity across the measured shear rate range.In contrast, the PAA bath at a concentration of 9 wt% shows no more than one order of magnitude decrease in the viscosity over the measured shear rate range.Overall, the flow sweep measurements of the baths reveal why the behavior of the PAA bath is superior to that of the xanthan gum bath; In addition to providing a robust enough polymeric network to act as support for keeping the prints in place, it exhibits only a mild shear-thinning behavior.
In the next step, the PAA concentration was adjusted and tuned separately for each material system as it was found that a single PAA concentration does not guarantee printability for all the ternary mixtures (due to different oil components or surfactant-nanoparticle pairs).Our choice of these concentrations is dictated by the robustness of the created bijel and the speed at which printed filament sinks to the bottom of the dish or stays afloat, which collectively determine whether the filament gets dragged around and attached to the dispensing nozzle (common reasons for print failure).In some cases, like the DEP-Ludox CL system, the addition of PAA was not necessary since the bijels formed upon injection of the ternary mixture are robust enough, making them settle rapidly to the bottom of the dish.In contrast, for other systems, the addition of PAA was crucial as it yielded a bath with favorable rheological behavior, as discussed earlier, resulting in superior prints in terms of robustness, gentle sedimentation of filaments, and shape fidelity.The concentration of PAA that proved to give the best print results (Figure 4) in the DEP-Ludux TMA system is 6 wt%, in the OA-Ludux TMA system 9 wt%, and for the OA-Ludux CL system 9 wt%.To further shed light on the reason for the favorable behavior of PAA baths for printing and the effect of PAA concentration on the bath behavior, another set of rheological measurements is carried out.
Similar to bijel prints, the oscillatory tests (amplitude and frequency sweep) as well as flow sweep tests were carried out for the rheological characterization of bath solutions.Amplitude sweep tests are performed to determine the LVR.Results, as shown in Figure 8a, indicate that shear moduli are independent of strain amplitude for all bath solutions since both shear moduli remained constant across the measured strain range, and hence, no significant critical strain was detected.Furthermore, they show a predominantly viscous response, that is, G 0 < G 00 at all strain amplitudes.Frequency sweep results measured at a strain range of 0.1-0.5% are also demonstrated for all aqueous baths (Figure 8b).With the neat surfactant aqueous solutions, it was not possible to measure any significant shear moduli (due to a very low viscous/elastic nature and dominant instrument inertia) in both frequency and amplitude sweep tests, but with the addition of PAA, both storage and loss moduli were recorded for all the baths.The loss modulus, indicative of viscous-like behavior, was higher than the storage modulus (solid-like behavior) for all bath solutions with added PAA, suggesting the dominant viscous nature.In the flow sweep curves (Figure 8c), it can be seen that with the addition of PAA into the neat surfactant aqueous solutions, a transition from almost Newtonian behavior (constant viscosity) to a shear-thinning liquid occurs, which is favorable in the context of printing in the bath; the higher viscosity and hence higher resistance against deformation are needed for the bath when there is no stress applied to keep the printed bijels in place and prevent the shape deformation.With the stress imposed by the translating nozzle, the bath yields and shows lower viscosity which facilitates the deposition of the printing ternary mixture into the bath.Moreover, the significant increase in the viscosity of the bah is evident with the addition of PAA, which further helps achieve optimum printing.
Additionally, shear moduli recovery and rheogram tests are conducted, and results indicate complete structural recovery and lack of any time-dependent (thixotropic) behavior (Figure 9).Specifically, as depicted in Figure 9a for the recovery test, the complete recovery of shear moduli is evident for both aqueous bath solutions containing 6 or 9 wt% PAA with CTAB as a surfactant (Figure 9a-top) and 9 wt% PAA when SDS is dissolved as the surfactant (Figure 9a-bottom).For the CTAB solution with 6 wt% PAA, a significant storage modulus is only captured at step 2 with a high amplitude strain of 50%, not at steps 1 and 3 (with a low strain).The fact that samples show excellent recovery to their initial shear moduli immediately after step 2 implies that the initial PAA polymeric network and hence bath properties are not disrupted due to the exposure to the high shear rates experienced throughout the printing.This behavior contributes to the optimum rheological behavior of baths in the bijel printing since the extruded bijel will be kept in place immediately after deposition, excluding deformation in shape or being dragged around with the movement of translating nozzle.Rheogram tests were also conducted to investigate the thixotropic behavior, and curves are depicted in Figure 9b.A complete overlap of up-curve and down-curve, that is, lack of hysteresis loops between these two curves, for all aqueous bath solutions indicates there is no thixotropic behavior for these baths.This behavior is favorable in terms of planar (2D) printing but more importantly for 3D printing within a bath phase as it indicates the structure of the bath can recover itself rapidly, avoiding the collapse of a print that otherwise could occur due to time-dependent recovery in thixotropic baths.Overall, the rheological measurements outlined and discussed above showed the means for assessing whether or not an aqueous bath can be used for liquid-in-liquid 3D printing of bijels.

Conclusion
In this study, bijel production from more versatile materials (including different oil components and nanoparticle/surfactants with varying charges) has been presented.We have also shown for the first time the printing of relatively complex bijel constructs through STRIPS and using the embedded associative liquid-in-liquid 3D printing.The microstructures of printed gel constructs, that is, two interwoven continuous liquid phases, were examined by both confocal and electron microscopy techniques.Rheological measurements showed that printed bijels have solid-like features with shear-thinning behavior as well as minimal temperature sensitivity.Furthermore, it was shown how rheological measurements can be used to characterize the bath behavior and its optimization for ensuring printing success.In the bijel constructs printed in this work, both aqueous and oil-based phases remain liquid after print and are not cured, unlike the bijels created in many STRIPS approaches where the oil phase usually undergoes photopolymerization, eliminating its fluidity. [20,44]This enables the mobility and rapid transport of mass or ion both within the phases and across the liquidliquid interface.The bijels created in this work are suitable for applications that require excellent thermal robustness over a wide range of (physiologically relevant) temperatures while maintaining the liquid state of the two constituent phases.Furthermore, compared to most of the other associative LL3DP platforms where the prints inside the bath are in the pure liquid state and prone to deformation upon rapid mechanical perturbation, [32] the bijel prints in this study exhibit relatively higher mechanical stability due to their bicontinuous microstructures.Potential applications for such bijels include chemical separation, catalysis, [25] tissue engineering, [6,55] drug delivery, or functional emulsions (with complex materials at the interface).The work presented here opens up opportunities for bijels in various fields where oils with different physiochemical properties are in demand.Moreover, the freedom for creating various patterns out of bijels combined with their morphological attributes and the contrasting properties of the constituent liquid phases (oil and water) is a significant leap forward in materials design.

Experimental Section
Materials: Silica (SiO 2 ) nanoparticles (Ludox TMA (22 nm, negative) or Ludox CL (12 nm, positive)), cetyltrimethylammonium bromide (CTAB), diethyl phthalate (DEP, 99.5%), sodium dodecyl sulfate (SDS), poly (acrylic acid, PAA) with average M w 1800, xanthan gum, tetraethyl orthosilicate (TEOS, >99%), and Nile red fluorescent dye (technical grade) were purchased from Sigma-Aldrich and used as received.For preparation of the 3 wt% TEOS solution, a commercial mineral oil was used.The physical properties of the nanoparticle dispersions are provided in Table 1.Based on numerous studies, [23,56] the acidic condition of nanoparticle dispersions helps them to be readily dispersed in the initial ternary mixture at high weight fractions as it was the case for our systems.The underlying mechanism is that pH has a direct impact on the surface charge of nanoparticles (and hence the adsorption of oppositely charged surfactants) since the stability of the nanoparticles is dominantly controlled by electrostatic interactions. [23]OA (technical grade 90%) was obtained from Alfa Aesar.Pure ethanol (200 proof ) was purchased from Decon Lab and deionized ultrapure water with a pH of 5.5 and a resistivity of 18.2 MΩ cm (PURELAB Chorus 1, ELGA Veolia) was used for preparing samples.
Ternary Mixture Preparation: The ternary mixtures were composed of five main components; oil, water, solvent (ethanol), nanoparticles, and surfactant.Four different ternary solutions were prepared using two oils (OA and diethyl phthalate) and two nanoparticles (Ludox TMA and Ludox CL, Table 1).Since surfactant was used in the form of a solution in ethanol and the nanoparticles were suspended in water, the summation volume of pure ethanol and the dissolved surfactant should be the volume of the ethanol needed for the desired composition in the ternary diagram and  the total volume of water and nanoparticles should be equal to the volume of water.The compositions of these ternary solutions are provided in Table 2.As common in previous studies on bijels, [20][21][22][23] compositions were chosen close enough to the plait point of the corresponding ternary phase diagram to ensure the formation of the bicontinuous structure in the resultant gels (Figure S1, Supporting Information).For preparing these mixtures, solutions of surfactants (SDS or CTAB) in pure ethanol should be prepared first at the concentration of 200 mM.Then all the components were mixed together in a vial based on the volumes outlined in Table 2. Using an ultrasound bath, prepared ternary mixtures were sonicated for 5 min prior to printing to ensure their uniformity.Since the Ludox CL nanoparticles have different physical properties (with 12 nm size), the ternary mixture composition which yields bijel formation was detected for both systems.(See Note S1 and Table S1 and S2, Supporting Information) Confocal Microscopy: For confocal microscopy, a small amount (0.1-0.2 mg mL À1 ) of the fluorescent dye Nile red was added to the oil phase for preparing the ternary mixture.The ternary mixture was then printed in glass μ-chamber slides with the dimension of 21.6 mm Â 23.8 mm Â 9.3 mm (W Â L Â H, μ-Slide, 2 well, ibidi) into a single bijel fiber manually using a syringe and a 27 G tapered needle within the continuous aqueous phase contained in the wells of μ-chambers.It is worth mentioning that in order to set up the microscope, a few minutes elapsed between when the time fibers were printed and the time microscopy images were taken.For imaging, a laser scanning confocal microscope (Leica STELLARIS 8 confocal microscope, Leica Microsystems) on an inverted microscope platform with laser excitation of 480 nm (Nile red emission 600-700 nm) was used.The Petri dishes and any other surfaces used in this work for creating and imaging bijels like the chamber slides were made out of glass and used without any coating.
Scanning Electron Microscopy (SEM): A field-emission scanning electron microscope (Philips-XL30, ESEM-FEG) was used for imaging the microstructures of the solidified bijel prints.A small cut of bijel prints which were solidified using the TEOS solution were placed onto a stub covered with an adhesive conductive carbon tape.Upon drying the samples, they were coated with a thin conductive layer of gold-palladium in a sputter coater (Leica EM SCD050).
Generation of Gcodes for Printing: Gcode is a series of commands for controlling various hardware mechanisms of a 3D printer, such as print head trajectories and its speed, extrusion rate of materials, and adjustment of other parameters such as the print bed or head temperature.For printing of bijel constructs in this work, gcodes for various planar (2D) architectures were developed, including a spiral, the Statue of Liberty, a coronary artery tree model with branches, the word "bijel" in a script font, a flower-shaped design, a five-pointed star, outline of the letter "B," and a meandering channel.Coordinates of the 2D spiral design and curved sections of the meandering channel or the artery tree were obtained using an in-house MATLAB code.For the rest of the designs, a free open-source vector graphics editor called Inkscape was used for converting the designs/images into a bitmap for obtaining the outlines and consequent generation of the gcode.
Printing of Bijel Constructs: For printing, a CELLINK BIO X6 3D bioprinter (Gothenburg, Sweden), the syringe printhead (with the mechanically driven extrusion mechanism), and 3 mL syringes (Luer-Lok Tip, BD) were used.The reason for choosing this printhead was the control it offered over ink injection rate, dispensed volume, and the diameter of injected filament, compared to the pneumatic-driven module.Unless otherwise noted, a tapered plastic needle with a size of 20 G was used due to lower incidence of clogging and smoother extrusion of material compared to other types of needles, such as general-purpose stainless steel needles.The extrusion rate for all the prints was 30 μL s À1 , and the printhead speed was 5 mm s À1 .All the printing parameters tested in this work, including those resulting in successful prints and those failing (along with the reason for their failure) are listed in Table S3, Supporting Information.
Rheological Characterization of Printed Bijel Constructs: A stresscontrolled Discovery-HR3 rheometer (TA Instruments, USA) was used for all rheological measurements.A 20 mm-diameter parallel plate geometry was used, and tests were carried out at a gap of 500 μm and a controlled temperature of 25 AE 0.5 °C.Following the printing and prior to rheometry, the bath material was drawn from the Petri dish using a syringe.The Petri dish was then covered with a layer of parafilm to avoid drying of the bijels.Upon holding the Petri dish in a tilted position for 10 min, excess bath solutions flowed out of the printed constructs and could be then drawn.The elapsed time between when the bath solution was drained from the Petri dish and the beginning of rheology measurements did not exceed 15 min and for almost all of this time, the Petri dish containing the bijels was covered with Parafilm, minimizing the chance of drying.To minimize the drying of bijels throughout the rheological measurement, the solvent trap sealed with water was used that maintained a water-saturated atmosphere around samples.As discussed in Results and Discussion and described below, additional oscillatory tests were carried out to ensure that the bijels did not exhibit any instability such as drying throughout the rheological measurements.
After removing the bath material from the container, the printed bijel constructs which remained in the container were then transferred and loaded onto the bottom plate of the rheometer.As the first test, an oscillatory time sweep test was carried out to study the evolution of shear moduli and determine whether the samples underwent any deformation over time following sample loading and trimming.In these tests, the sample was exposed to a constant (small) strain amplitude at an angular frequency of 1 Hz, and shear moduli were obtained as a function of time (5 min).Next, an amplitude sweep was carried out to determine the LVR for further rheological tests.At a constant angular frequency (ω) of 1 rad s À1 (6.22 Hz), the amplitude strain was swept from 0.001% up to 100%, and the test was stopped once a noticeable drop in storage modulus (G 0 ) curve was observed.Then, a strain within the LVR of each bijel was selected for frequency sweep measurements where the viscoelastic properties were characterized over a frequency range of 0.1-100 rad s À1 .For oscillatory measurements (i.e., amplitude and frequency sweep tests), the data might be affected by instrument inertia, specifically at high frequencies, and it manifests itself in the raw phase difference between the oscillating displacement and torque signals where values higher than 175 are considered instrument inertia artifacts and hence, omitted from plots.Furthermore, data obtained with a torque lower than 1 mN m À1 were not reported as they were close to the limit of the instrument.For all samples, both amplitude and frequency sweep tests were repeated three times to ensure the consistency of data.
Then flow sweep test was performed for the characterization of the flow behavior of samples with an increasing shear rate from 0.001 to 100 s À1 .To maximize the data reliability in the flow test, the accurate steady-state sensing mode was used, where samples were subjected to the desired shear rate for a sufficiently long time to achieve a steady state.Finally, in the temperature sweep experiments, the effect of temperature on the rheological behavior of bijels was investigated where shear moduli were measured at a frequency of 1 Hz within the LVR while the temperature was increased from 15 to 40 °C at a constant rate of 3 °C min À1 .For these temperature sweep measurements, new bijel samples were collected and mounted on the rheometer plate since their internal structures were likely to be destroyed after large shearing deformations in flow sweep tests.
Preparation and Rheological Characterization of Printing Baths: Printing support baths were prepared by dissolving either poly(acrylic acid, PAA) or xanthan gum in the aqueous solution of surfactants (either 1 mM CTAB or SDS) at desired concentrations.For the flow behavior comparison between PAA and xanthan baths, PAA and Xanthan gum were dissolved in the CTAB aqueous solution (1 mM) at concentrations of 9 and 1.2 wt%, respectively.After the PAA or xanthan gum was dissolved, bath solutions were put into a vacuum desiccator (Bel-Art, SP Scienceware) with a piston pump (Welch model number 2546B-01) to remove entrapped air bubbles.For these rheological characterizations, concentric cylinder geometry with Bob rotor was used due to the low viscosity and easy flow of bath solutions.In addition to the oscillatory and flow sweep tests outlined earlier, the oscillatory recovery test was performed on the bath samples to characterize their yielding and recovery behavior.This test consisted of three steps where shear moduli (G 0 and G 00 ) were measured in an oscillatory manner when experiencing different strain amplitudes at a constant speed (ω = 1 rad s À1 ).In the first step, a strain amplitude of 0.5% (within LVR) was applied for 1 min to obtain the steady-state measurement for both storage (G 0 ) and loss (G 00 ) moduli.In the second step, a strain outside of LVR, that is, 50% was applied, which was high enough to represent the deformation experienced during printing and upon extrusion of the ternary mixture into the bath.This step lasted for 10 s, which was long enough for perturbing the PAA network and was followed by the third and final step, where the same strain of 0.5% was used to assess the recovery of shear moduli measured in the first step.As the last test and to investigate the thixotropic characteristics, the upward and downward flow curves (rheograms) were measured by increasing the shear rate from 0 to 60 s À1 over a period of 120 s.

Figure 2 .
Figure 2. Confocal microscopy images of bijels formed from different ternary mixtures at various magnifications, illustrating their bicontinuous microstructures.The appearance of red color in these images originates from the Nile red dissolved in the oil phase (OA or DEP) of the initial ternary mixtures.The images are captured for a) DEP-Ludox CL, b) DEP-Ludox TMA, c) OA-Ludox CL, and d) OA-Ludox TMA systems.

Figure 3 .
Figure 3. Scanning electron micrographs of the surface of the DEP bijel prints and two nanoparticle systems (top: Ludox Cl, bottom: Ludox TMA) at various magnifications.

Figure 5 .
Figure 5.A 3D spiral helix structure printed with DEP-Ludox TMA system inside the aqueous bath containing 6 wt% PAA.

Figure 6 .
Figure 6.Rheological characterization of printed bijels, illustrating the frequency sweep and temperature sweep results for systems with a,b) DEP and c,d) OA as the oil phase.An amplitude strain within the LVR (determined by amplitude sweep curves) was chosen for each bijel in these measurements.

Figure 7 .
Figure 7. Rheological characterization of printed bijels.Flow sweep behavior of printed bijels for systems a) with DEP and b) OA as the oil.c) The averaged shear storage modulus (G 0 ) for four systems illustrates the strongest network for the DEP-Ludox CL system while the DEP-Ludox TMA shows the lowest storage modulus and hence, weaker mechanical property.

Figure 8 .
Figure 8. Rheological characterization of aqueous bath solutions used for printing that contain a surfactant (CTAB or SDS) at 1 mM and PAA as the thickener.a) Amplitude sweep measurements were carried out at a frequency of 1 Hz, and b) strain within LVR is selected for the frequency sweep tests.c) The flow sweep curves for neat surfactant aqueous solutions (of SDS and CTAB) are shown for comparison, but their amplitude and frequency sweep results could not be obtained due to dominant instrument inertia (see Experimental Section).

Figure 9 .
Figure 9. Structural recovery properties of baths measured through a) shear moduli recovery test and b) rheogram.

Table 1 .
Physical properties of the nanoparticle (Ludox TMA and CL) dispersions used in this work.

Table 2 .
Summary of compositions used for each system in creating or printing bijels.