Crosslinking Strategies for Microfluidic Production of Micro- gels

This article provides a systematic review of the crosslinking strategies used to produce microgel particles in microfluidic chips. Various ionic crosslinking methods for gelation of charged polymers are discussed, including external gelation via crosslinkers dissolved or dispersed in the oil phase, internal gelation methods using crosslinkers added to the dispersed phase in their nonactive forms, such as chelating agents, photo-acid generators, sparingly soluble or slowly hydrolyzing compounds, and methods involving competitive ligand exchange, rapid mixing of polymer and crosslinking streams, and merging polymer and crosslinker droplets. Covalent crosslinking methods using enzymatic oxidation of modified biopolymers, photo-polymerization of crosslinkable monomers or polymers, and thiol-ene “click” reactions are also discussed, as well as the methods based on sol-gel transitions of stimuli responsive polymers triggered by pH or temperature change. In addition to homogeneous microgel particles, the production of structurally heterogeneous particles such as composite hydrogel particles entrapping droplet interface bilayers, core-shell particles, organoids, and Janus particles are also discussed. Microfluidics offers the ability to precisely tune chemical composition, size, shape, surface morphology, and internal structure of microgels by bringing in contact multiple fluid streams in a highly controlled fashion using versatile channel geometries and flow configurations and allowing controlled crosslinking.


Introduction
Hydrogels are three-dimensional networks of hydrophilic crosslinked polymers that can hold large amount of water in their intermolecular space, but they are not soluble in water in their crosslinked form. Hydrogels are widely used as excipients for drug delivery systems [1,2], scaffolds in tissue engineering [3,4], wound dressings [5], absorbents in hygiene products (diapers, napkins, hospital bed sheets, sanitary towels) [6], gelling agents, thickeners, and packaging materials in food products [7], and irritation-free, transparent materials for contact lenses [8]. Their high-water retention capacity and soft, porous structure mimic the in vivo extracellular matrix (ECM) microenvironment. Hydrogels can undergo large and reversible volume changes by expelling or absorbing water in response to external stimuli, such as light, temperature, pH, ionic strength, and the presence of chemical triggers, which can dramatically change their physical properties and permeability to small molecules [9].
In 'physical' hydrogels, molecular entanglements and/or secondary forces such as ionic, H-bonding or hydrophobic forces play the main role in the network formation [10]. Physical gels are reversible and can be disintegrated by changing environmental conditions, such as pH, temperature, and ionic strength of the solution. Typical physical hydrogels, such as alginate, carboxymethyl cellulose and chitosan, are prepared by ionotropic gelation with oppositely charged divalent ions [11,12]. In 'chemical' gels, polymer chains are permanently connected by covalent bonds. Chemical gels are prepared in two different ways: (i) Free radical polymerization of low molecular weight hydrophilic monomers and (ii) Polymerization of polymers. Free-radical polymerization often results in a significant level of residual monomers, and therefore, hydrogels must be purified to remove unreacted monomers, which are often harmful.
Microgels are micron-sized microparticles composed of hydrogels. They can be generated using a variety of fabrication methods including emulsification using ultrasonication, mechanical agitation or high-pressure homogenization [13], atomization [14], extrusion through a syringe or nozzle [15], micromolding [16], and molecular self-association [17]. Compared to most of these techniques, microfluidic platforms offer superior control over the size and morphology of microgels, cell co-culture with a precise control over the number of cells of each type per single bead [18], high encapsulation efficiency due to low shear forces during droplet generation, integration of particle generation and on-chip manipulation, continuous processing, and operation under sterile conditions. Monodisperse microgels produced in microfluidic devices have been widely used in molecular and synthetic biology [19], biotechnology, and tissue engineering [20].
Microgels can have a homogeneous (matrix-type) or core-shell structure. Matrix-type microgels are commonly synthesized via water-in-oil (W/O) emulsions by crosslinking polymers within aqueous template droplets. The incorporation of additional phase within the dispersed phase results in the formation of multiple emulsion droplets, such as oil-inwater-in-oil (O1/W/O2), water-in-water-in-oil (W1/W2/O) or even water-in-water-in-water (W1/W2/W3), which can be transformed into core/shell particles by the gelation of middle phase or both inner and middle phase. The objective of this paper is to review recent developments in crosslinking polymers or monomers in microfluidic chips for the purpose of producing both matrix-type and core-shell microgels. The main emphasis was placed on the production of microgels with complex morphology and chemical structure.

Microfluidic production of spherical matrix-type microgels
Spherical matrix-type microgels are commonly prepared using a two-step process consisting of the formation of W/O emulsion droplets containing a gel-forming polymer solution (step 1) and the crosslinking of polymer chains within the droplets (step 2). These two steps must be spatially and temporally separated, i.e., droplets must be pinched off prior to the crosslinking reaction, to avoid blockage of the droplet forming channel by the gellified polymer. The crosslinking step can be completed on the chip downstream of the droplet generator or in a collection vial (off-chip).
Physical microgels are usually formed by ionic crosslinking of polymers within droplets, while chemical microgels are formed either by covalent crosslinking of polymers or by polymerizing monomers within droplets.

Ionic crosslinking of droplets in microfluidic channels
Ionic crosslinking occurs between charged polymer chains and oppositely charged divalent or multivalent ions. The typical example of ionic crosslinking is electrostatic interaction between negatively charged alginate chains and positively charged Ca 2+ ions. Ionic crosslinking in microfluidic channels can be achieved using five different strategies [21]: internal gelation, external gelation, rapid in-droplet mixing (chaotic advection), droplet merging (coalescence) and competitive ligand exchange.
2.1.1. Internal (in-situ) gelation Figure 1. Formation of alginate beads by internal gelation. The organic acid dissolved in the oil phase dissociates at the droplet interface into acetate anions and hydrogen cations and the pH drops below 6.5. Released H + ions react with CaCO3 or Ca-EDTA dispersed or dissolved within the droplet, thereby releasing Ca 2+ ions that crosslink the polymer.
Here, the dispersed phase is an aqueous solution of charged polymer containing an undissociated calcium or barium compound (shown as solid black circles in Figure 1), such as CaCO3, BaCO3 [22], tricalcium citrate [23], and Ca-EDTA [24]. The continuous phase is a mixture of hydrophobic surfactant and organic acid, usually acetic acid (CH3COOH), dissolved in an inert oil. Typical inert oils used in this process are vegetable oils [25][26][27][28], dimethyl carbonate [29], fluorocarbon oils [30], and hexadecane [31]. After droplet generation, the organic acid diffuses through the oil phase and dissociates at the droplet surface into acetate anions (CH 3 COO − ) and protons (H + ). H + reacts with undissociated Ca or Ba compound inside droplets and releases Ca 2+ or Ba 2+ ions, which crosslink the polymer. In the case of CaCO3 / alginate / acetic acid system, the following reactions occur: Although the method offers a homogenous gelation throughout the entire droplet volume, it results in the reduction of pH below physiological pH, which can be detrimental to cell viability [32]. To minimize the exposure of cells to low pH, a stoichiometric amount of acetic acid can be added off-chip after droplet collection [33].
Figs. 2 and 3 show microfluidic internal gelation strategies implemented to produce alginate and composite alginate/pectin Janus beads, respectively.

Figure 2.
The formation of alginate beads by internal gelation. Droplets of sodium alginate solution loaded with CaCO3 are formed at the upstream T-junction. The resulting emulsion is mixed with acid-saturated oil delivered through the side channel and forced to pass through the wavy channel to allow enough time for acetic acid to diffuse to the droplets and trigger Ca 2+ release [26]. The formation of alginate/pectin Janus beads by internal gelation in the microfluidic chip consisting of upstream Y-junction and two downstream  junctions. Aqueous solutions of alginate and pectin are introduced through separate inlet channels of the Y junction to form a co-axial biphasic flow. Janus droplets are formed at the upstream  junction and the crosslinking reaction occurs at the downstream  junction. Due to laminar flow conditions, the two immiscible polymer solutions remain segregated between the two  junctions enabling a Janus droplet morphology to be preserved after crosslinking. The Janus beads are collected in a CaCl2 solution to strengthen the gel network structure [27].
Internal gelation can also be achieved by adding a slow hydrolyzing acid or a photoacid generator (PAG) into polymer solution rather than supplying acid from the oil phase across the droplet boundary. Morimoto et al. [34] added glucono-1,5-lactone (GDL), a slow release acidifier to the dispersed phase, which slowly hydrolyzes to gluconic acid to release Ca 2+ from CaCO3 and initiate alginate crosslinking without any external trigger. Photo-acid generators (PAGs) are compounds that release protons irreversibly upon illumination. Liu et al. [35] added diphenyliodonium nitrate (DPIN) to the aqueous phase to initiate the gelation of alginate solution from the middle phase of O/W/O emulsion droplets upon UV irradiation. The photolysis of DPIN results in the release of protons (H + ) and hydrophobic uncharged by-products [36]: DPIN → hydrophobic by-products + H + To prevent these by-products from precipitating within droplets and causing channel clogging, a macrocyclic compound can be added to the dispersed phase that can bind these hydrophobic species [36]. The emulsion formulations used to produce microgels by internal gelation are summarized in Table S1 in the supplementary material.

External gelation
The external gelation in microfluidic systems can be achieved using several methods: (i) An oil-soluble crosslinking agent, such as calcium acetate, can be dissolved in the oil phase and used for on-chip crosslinking; (ii) A crosslinking agent such as CaCl2 can be added to the gelation bath where the droplets are collected and crosslinked [37]; (iii) An aqueous crosslinker solution can be emulsified in a carrier oil and this emulsion can be used for on-chip crosslinking [38]; (iv) The same emulsion can be used as a shell liquid in core/shell droplets and used for on-chip crosslinking of aqueous cores [39]; (v) An aqueous CaCl2 solution can be emulsified in a carrier oil to produce a W/O emulsion, which can be dehydrated to form surfactant-coated CaCl2 nanoparticles dispersed in the oil phase. This nano-dispersion can be used for on-chip crosslinking [40,41]; (vi) An alcoholic CaCl2 solution can be dissolved in oleic acid and used for on-chip crosslinking after alcohol evaporation [42][43][44]; or (vii) A powdered crosslinking medium, such as dehydrated cell culture medium, can be dispersed in a carrier oil and used for on-chip crosslinking [45]. Table S2 provides the examples of microfluidic methods used for external gelation.
In the method (i), the dispersed phase is usually an aqueous polyanion solution, while the continuous phase is an oil-soluble salt of divalent or multivalent cations, e.g., Ca 2+ , Zn 2+ , and Fe 3+ , dissolved in an oil-surfactant mixture (Fig. 4). Depending of the type of oil and polymer, the oil-soluble salt can be calcium acetate [25], Fe (NO3)3 [46], CaJ2 [46], BaCl2 [47], SrCl2 [48], or ZnCl2 [49]. In the case of alginate crosslinking with calcium acetate, Ca-acetate diffuses through the oil phase to the droplet interface and dissociates in water to form Ca 2+ ions, which trigger polymer crosslinking within the droplets, according to the following reactions:  . Formation of alginate beads by droplet generation in a T-junction and subsequent external gelation. The gelation is triggered by Ca-acetate, an oil soluble salt, which diffuses to the droplet interface and releases Ca 2+ ions by hydrolysis, according to Eq. (5) [25]. The droplet gelation occurs first in the interfacial layer resulting in a core-shell morphology and then progresses towards the droplet interior.
For on-chip gelation, the combined diffusion and reaction time, + , must be shorter than the residence time of droplets in the chip, but greater than the droplet formation time, 1/f, where f is the frequency of droplet generation. The diffusive flux of salt across the oil/water interface depends on the salt concentration in the oil phase and its diffusivity in the oil phase. Another limiting factor is the salt solubility in the aqueous phase which must be sufficiently high to trigger polymer crosslinking. The solubility of calcium salts of fatty acids in water decreases with increasing the number of carbon atoms in a molecule [25]. For example, Ca-butanoate and Ca-2-ethylhexanoate cannot be used for external gelation, because no gelation occurs within 7 days when the concentration of either salt in the oil phase is 0.5 wt% [25]. The salt solubility in the oil phase could also be a limiting factor. A mixture of CaCO3 and acetic acid can be used instead of Ca-acetate to increase its solubility in oil [50]. The external gelation of pH sensitive Eudragit polymers can be triggered by the exchange of H + ions between an organic acid dissolved in the carrier oil and the polymer dissolved in aqueous droplets. Above pH 7.4, Eudragit S100 is soluble in water due to dissociation of carboxylic acid of methacrylic acid monomer units. A sol-gel transition occurs at pH < 7.4 due to charge neutralization (Fig. 5c). Monodispersed gel beads were produced when the sol-gel transition was triggered by p-aminobenzoic acid (PABA), a weak acid with a pKa of 2.38 (Fig. 5a). However, when the gelation was triggered by ptoluenesulfonic acid, a strong acid with a pKa of −2.8, the beads were large and polydisperse, due to premature polymer gelation at the oil/water interface (Fig. 5b) [51]. Eudragit microgels can be loaded with Clostridium difficile bacteriophages at low pH and used for the treatment of C. difficile infections as an alternative to conventional antibiotic therapies. The beads are stable in the acidic environment and release the phages at the infection site in the colon due to weakly alkaline conditions (Fig. 5c). Similarly, chitosan beads can be produced by exposing droplets of acidified chitosan solution to OHions [52]. The gelation of chitosan chains occurs due to deprotonation of amine groups (NH3 + ) above pH 6.2-6.5. Figure 5. Formation of Eudragit S100 beads by external gelation of aqueous Eudragit droplets in acidified mineral oil: (a) Sol-gel transition within droplets triggered by 1 wt% p-aminobenzoic acid (PABA) dissolved in mineral oil; (b) Premature sol-gel transition triggered by 1 wt% p-toluenesulfonic acid (PTSA) dissolved in mineral oil; (c) Site-specific release of bacteriophage from gel beads. The disentanglement of polymer chains occurs due to electrostatic repulsion between negatively charged carboxyl groups of methacrylic acid monomer units at a pH value above the polymer pKa [51].
In method (ii), microgels are formed by off-chip polymer crosslinking in the gelation bath. Capretto et al. [22] produced Ba-alginate beads in a microfluidic Y-junction by collecting generated W/O emulsion in a gelation bath containing 1.5 wt% BaCl2 (Fig. 6). Potential issues with this technique are accumulation of droplets at the oil/water interface due to small density difference between alginate droplets and BaCl2 solution and formation of tail-shaped beads due to droplet deformation during settling in the oil phase. The problems can be minimized by pouring a low viscosity oil above the BaCl2 solution in the gelation bath to decrease shear force from the oil phase and adding glycerol to the Na-alginate solution to increase density of the dispersed phase [22]. Two serial Y-junctions are useful for the preparation of mixed gel beads composed of two different polymers, e.g., Matrigel and alginate. Matrigel is a mixture of extracellular matrix proteins extracted from mouse sarcoma, composed of 60% laminin, 30% collagen IV, and 8% entactin. Matrigel is liquid at 4°C, but gels at 24-37°C by self-assembly of laminin and collagen IV into crosslinked networks via entactin bridges. In Figure 7, a mixture of tumor cells and Matrigel in a cell culture medium at 4 C was mixed with alginate solution in the upstream junction and this mixture was emulsified in mineral oil in the downstream junction. Since Matrigel and alginate are delivered through separate inlet channels, a mixing ratio between the two polymers can be tuned on-chip. The droplets are crosslinked in a 4 wt% CaCl2 solution to form cell-laden composite beads. On-line tuning of the hydrogel composition was provided by supplying the two polymers via separate inlet channels which allows a flexible Matrigel to alginate mixing ratio [37].
In method (iii), a fine W/O emulsion containing aqueous CaCl2 droplets dispersed in corn oil is introduced through the downstream cross junction to crosslink alginate droplets formed in the upstream cross junction (Fig. 8). Using two consecutive cross junctions, droplet generation is spatially separated from the crosslinking reaction to avoid clogging of droplet-forming channel by the gel. The channel downstream of the second junction has an increased width to decrease oil velocity and prevent shear-induced deformation of droplets during gelation [38]. The concentration of lipophilic surfactant used to stabilize CaCl2 droplets should be sufficiently low to allow their merging with alginate drops. The optimum concentration of Y-Glyster CRS-75 was found to be 0.1 wt% [38]. The similar method can be used for oxidative covalent crosslinking of modified natural polymers. Figure 8. Formation of Ca-alginate beads by external gelation in a chip consisting of two serial crossjunctions using nanodroplets of aqueous CaCl2 solution as the crosslinking agent [38].
A slow gelation by CaCl2 nanodroplets may cause coalescence of alginate droplets and clogging of the outlet channel. A possible solution could be to increase the length of the downstream channel, which may lead to excessive pressure buildup in the chip, particularly due to high viscosity of the nanoemulsion. To prevent droplet coalescence, a single emulsion method can be replaced by the double emulsion method shown in Fig. 9. Figure 9. Formation of cell-laden calcium alginate beads using a two-step external gelation process within a double W/O/W emulsion. CaCl2 nano-droplets in the shell penetrate inside core droplets and cause their surface gelation followed by complete gelation in a CaCl2 gelation bath [39].
In the method shown in Fig. 9, individual alginate droplets are coated by a thin layer of CaCl2 nanoemulsion to form core/shell droplets dispersed in the outer aqueous phase. The outer aqueous phase plays an important role: (i) prevents fusion of partially gelled alginate droplets; (ii) decreases viscosity in the downstream channel, which improves the chip functionality; (iii) facilitates the separation of beads from the oil phase and minimizes cell exposure to oil and surfactant, which increases biocompatibility of the process.
In the method (vii), self-assembling peptide (SAP) solution containing mammalian cells was emulsified in the oil phase composed of a powdered cell culture medium dispersed in mineral oil (Fig. 10a). In the downstream channel, particles of cell culture medium collide with the droplets and become dissolved in the SAP solution. The dissolution of low molecular weight compounds, such as inorganic salts and amino acids, leads to an increase of the ionic strength within droplets and triggers gelation of SAPs, as shown in Fig. 10b. Over 93% of the cells survived the microfluidic process and the fabricated microgels allowed diffusion of nutrients, as well as cell growth and differentiation [45].

Rapid mixing of fluid streams within droplets
In this method, gelation is achieved by rapidly mixing two aqueous streams, usually polymer solution and crosslinking solution (Fig. 11a), or three aqueous streams, e.g., polymer solution, crosslinking solution, and a cell suspension (Fig. 11b), immediately before droplet formation [53]. The number of cells per bead and the crosslinking density can be controlled by adjusting the flow rate ratio and composition of the inlet aqueous streams.

Diffusion into aqueous phase
Endothelial cell The examples of fluid compositions and channel geometries used in this method are provided in Table S3 in the Supplementary material. Figure 11. Formation of alginate beads by in-drop mixing [54]: (a) A microfluidic T-junction consisting of two converging inlet channels for supplying alginate and CaCl2 solutions; (b) A microfluidic T-junction for cell encapsulation withing microgel consisting of three converging inlet channels [55].
In Fig. 11(a), aqueous droplets comprising alginate and CaCl2 are formed in an immiscible continuous phase, and the gelation is achieved by chaotic advection within the droplets. Chaotic advection is caused by hydrodynamic interactions between the droplets and the microchannel walls and winding collection channels provide more efficient internal circulations than straight channels. The T-junction shown in Fig. 11(b) is composed of three converging inlet channels and can be used for encapsulation of cells within alginate microgels [55]. The time required for crosslinking (gel formation time) should be longer than the droplet formation time to prevent premature gelation and channel clogging. For constant geometry of microfluidic channels, the gel formation time primarily depends on the concentration of the reagents, while the droplet formation time mainly depends on the fluid flowrates and channel size [54]. The gelation of alginate was also achieved by in situ mixing a solution of slow hydrolyzing acid and alginate/CaCO3 solution [54].
To prevent premature alginate crosslinking inside a droplet, a stream of water can be injected between CaCl2 and alginate solutions, as shown in Figure 12. At the upstream junction, the two side streams, sodium alginate and CaCl2 solutions, are separated by injecting deionized water through the middle channel to prevent crosslinking reaction before the droplets are formed in the downstream junction. The droplets are solidified in the wavy reaction channel. To prevent diffusive transport of Ca 2+ ions through the thin layer of middle stream, the thickness of the middle stream must be greater than the diffusional distance, , given by = √2 , where is the diffusion constant of Ca 2+ and is the contact time of the two laminar streams [56]. The concentration of CaCl2 and alginate solutions plays an important role. The optimum CaCl2 concentration was found to be 0.5-1.0 wt% [56]. At the CaCl2 concentration below 0.5 wt%, droplets cannot polymerize, and the jetting regime occurs at the CaCl2 concentration of 2 wt% or above. Here, microgels are produced by injecting a crosslinking solution into the polymer droplets or by merging polymer droplets and crosslinking solution droplets. In Fig. 13, the chip is consisted of two double T-junctions for encapsulation of glucose oxidase (GOx), horseradish peroxidase (HRP), and Amplex Red within alginate beads (T1), and for colorimetric detection of glucose (T2). Aqueous alginic acid droplets loaded with GOx, HRP and Amplex Red are generated at the cross junction J1 and merged with a stream of CaCl2 solution injected from the inlet A3. The fused droplets generated at the junction T1 are transformed into hydrogel beads as they pass through the serpentine channel between T1 and T2. A glucose-containing sample supplied from the inlet A4 is injected into the beads at the junction T2. In the presence of hydrogen peroxide released upon enzymatic oxidation of glucose within the beads, Amplex Red reagent is transformed into highly fluorescent resorufin. The merging efficiency of double T-junctions was over 90% under optimal conditions and higher than the merging efficiency of single T-junctions [58]. used for alginate droplet generation and two double T-junctions used for injection of CaCl2 solution (T1) and glucose solution (T2) from the inlets A3 and A4, respectively [58].
The crosslinking strategy shown in Fig. 14 is based on fusion of alginate and CaCl2 droplets in a cylindrical fusion chamber. The cell-laden alginate solution and the alginate solution loaded with magnetic nanoparticles are injected through two separate inlets of the head-on junction to form a biphasic flow. Further downstream, this bicolored stream is split into Janus droplets by flow focusing with an oil phase. To inhibit the exchange of the content between the two solutions within a droplet, the diameter of the Janus droplets was limited to 80% of the channel width [59]. CaCl2 droplets were delivered to the main channel from the side channel and their production was synchronised with the production of Janus droplets to place CaCl2 droplets between neighbouring Janus droplets. Janus droplets were merged with CaCl2 droplets in a cylindrical chamber to form Janus particles with magnetic anisotropy. Magnetic halves allow particle manipulation by a magnetic field, while alginate halves provide a good microenvironment for cell encapsulation. Figure 14. Formation of alginate/magnetic Janus beads by droplet merging in the chip consisting of head-on junction / flow focusing junction for generation of Janus droplets, a side channel for introduction of CaCl2 droplets and a fusion chamber for merging Janus and CaCl2 droplets [59].
A similar strategy was applied in the chip shown in Fig. 15. Here, alginate and CaCl2 droplets formed in separate cross junctions were alternately fed to the expansion chambers where they merge and form microgels. Figure 15. Formation of disk-like, plug-like and spherical alginate beads in the chip consisting of two flow-focusing junctions for alternating generation of alginate and CaCl2 droplets and a fusion channel with two circular expansion chambers for droplet merging. The droplet shape was modified by their confinement in one direction (disks) or two directions (plugs) [60].
As shown by others [60][61][62], in situ gelling of droplets with dimensions beyond the height and/or width of a microfluidic channel can be utilized to form non-spherical microgels, such as disks, rods and threads.
By increasing the flow rate of the continuous phase or decreasing the flow rate of the dispersed phase, the shape of the beads can be changed from threads to rods to disks to spheres reflecting different droplet volumes [60]. The resultant droplet volume depends on the competition between viscous forces, tending to stretch the dispersed phase into a long jet, and interfacial tension, acting in the opposite direction. Spherical beads are formed if the droplet volume, , is smaller than ℎ 3 /6, where ℎ is the channel height which is smaller than the channel width. Non-spherical beads are generated from confined droplets [61]. Plugs (rods) and threads are formed from droplets, which are confined in two directions and have the roundness, (= 4 / 2 ) greater than unity, where is the projected surface area of the beads and is the projected bead perimeter [63]. For spherical and discoidal beads, = 2 /4 and = , and the roundness is = 1.
The production of disk-like magnetic alginate beads loaded with cells is shown in Fig. 16. The size of the beads and the number of cells encapsulated per each bead can be adjusted by controlling the fluid flow rates. Disk-like beads with flat top and bottom surfaces are formed due to vertical droplet confinement in the serpentine channel since the equivalent droplet diameter is greater than the channel height, and the channel width is sufficiently large to prevent formation of plugs. A disk-like shape allows for the cell division process to be monitored without image distortion [62]. Other strategies used for production of microgels by droplet merging are summarized in Table S4.

Competitive ligand exchange crosslinking (CLEX)
This crosslinking method is based on the competition between a gelling ion (Ca 2+ ) and an exchange ion (Zn 2+ ) for binding sites on chelating agents (EDDA and EDTA) and a charged polymer [64]. At pH 6.7, a mixture of Zn-EDDA and alginate will not gel, nor will a mixture of Ca-EDTA and alginate, because both complexes do not dissociate at this pH (Figs. 17 a-b). However, upon mixing, Zn 2+ will be exchanged between EDDA and EDTA due to their higher affinity to EDTA, compared to Ca 2+ . It will result in the release of Ca 2+ ions (Fig. 17 c), which will then crosslink the alginate, due to a higher binding affinity of Ca 2+ to alginate than EDDA (Fig. 17 d). At pH > 7.2, the amount of Ca 2+ is insufficient for gelling since the competition between alginate and EDDA for Ca 2+ is shifted towards Ca-EDDA at such pH. On the other hand, at pH < 6.7, the gelation is too fast due to high concentration of crosslinking Ca 2+ ions, which leads to clogging at the junction where the aqueous phases meet.
The chip for producing alginate beads by CLEX is comprised of three inlet channels: one inlet is for the carrier oil phase and two inlets are for alginate solutions containing Ca-EDTA and Zn-EDDA, respectively (Fig. 17 e). Alginate should be added to both aqueous streams to balance the hydrodynamic resistance in the two inlets as well as to avoid the polymer dilution after mixing. The kinetics of the ion exchange process and the amount of released Ca 2+ ions depend on the pH and the type of chelators used, which can be used to control the gelation kinetics and gel strength [32]. The gelation time can be controlled in the range from seconds to minutes while maintaining the pH within the physiological range [32]. It ensures enhanced cell survival rate compared to internal gelation approach where Ca 2+ ions are released from Ca-EDTA or solid CaCO3 using an acidified oil phase, which inevitably results in a pH drop well below the physiological range.

Enzymatic crosslinking
The mechanical properties of ionically crosslinked natural polymers, such as elastic modulus, and swelling ratio, are unstable due to potential loss of crosslinking ions. However, functional groups (e.g., -OH, -COOH, and -NH2) of natural polymers can be chemically modified to allow for their covalent crosslinking. For example, phenol containing molecules such as tyrosine and tyramine can be conjugated to alginate via carbodiimide chemistry [65] or periodate chemistry [66]. The alginate-tyramine conjugates can be crosslinked via horseradish peroxidase (HRP)-catalyzed oxidative coupling of phenol moieties in the presence of hydrogen peroxide (H2O2). Gel networks composed of covalently crosslinked polymer chains have better mechanical properties and greater chemical and thermal stability compared to ionically crosslinked polymer networks [67]. In addition, enzymatic crosslinking offers high reaction rates under physiological conditions and a 'green' approach to hydrogel synthesis, including the mildness of the reaction and biocompatible catalysts [67]. Typical microgels prepared by enzymatic crosslinking of modified polysaccharides in microfluidic chips are shown in Table S5. Most of the research has been done using alginate-tyramine, dextran-tyramine, and hyaluronic acid-tyramine conjugates and HRP/H2O2 catalysts. The same approach can be used for crosslinking the polymers functionalised with the resorcinol and catechol groups.
A strategy used for encapsulation of mammalian cells within alginate-tyramine microgel is shown in Fig. 18. Alginate-tyramine (Alg-Tyr) is produced by conjugating tyramine to alginate in the presence of N-hydroxy sulfosuccinimide (NHS) and 1-ethyl-(dimethylaminopropyl) carbodiimide (EDC), Fig. 18(a). Cell-laden microgel beads can be formed by injecting aqueous solution composed of Alg-Tyr, HRP, and cells into a co-flowing stream of liquid paraffin saturated with molecularly dissolved H2O2, Fig. 18(b). H2O2 penetrates inside droplets and triggers HRP-catalyzed crosslinking of Alg-Tyr via C-C and C-O bonding of phenol moieties [65]. H2O2 can also be supplied from W/O nanoemulsion composed of nanodroplets of H2O2 solution dispersed in oil [68]. Dextran-tyramine (Dex-Tyr) conjugates can be prepared by activating hydroxyl groups of dextran with p-nitrophenyl chloroformate (PNC) and reacting the obtained Dex-PNC with tyramine, as shown in Fig. 19(b) [69]. Microfluidic encapsulation of single cells within Dex-Tyr microgels achieved by mixing the polymer, cells, and crosslinkers in situ just before droplet pinch-off is shown in Fig. 19(a) [70,71]. One problem with this approach is in off-centred cell encapsulation, which may cause cell escape during subsequent manipulation. Namely, immediately after droplet pinch-off, cells take positions close to the aqueous/oil interface due to temporary inertial and hydrodynamic effects. Since gelation occurs within milliseconds, cells become trapped in their unwanted offcentre positions [70]. The problem can be overcome by delaying on-chip gelation of droplets, which can be achieved by slow diffusion of H2O2 through a PDMS wall [70], rather than by direct mixing of H2O2 with polymer solution within a droplet. Tyramine-conjugated polymer can also be crosslinked within droplets through a silicone tubing submerged in a H2O2 bath [68]. Cell centring via delayed droplet crosslinking was also applied to the crosslinking of hyaluronic acid-tyramine (Hy-Tyr) conjugate and co-crosslinking of Dex-Tyr and Hy-Tyr (Table S5).

Polymer-polymer crosslinking
Hyperbranched polyglycerol (hPG) and polyethyleneglycol (PEG) can be functionalized with acrylate groups and undergo free radical co-polymerisation within cell-laden droplets upon UV irradiation in the presence of a photoinitiator [72]. Since photoinitiators and UV irradiation are detrimental for cell viability, further work was focused on the use of UV-and initiator-free, thiol-ene "click" reactions between dithiolated PEG macro-crosslinkers and acrylated hPG (hPG-Ac) building blocks. Microfluidic emulsification of aqueous solutions containing PEG-dithiol, hPG-Ac, and cells is shown in Fig. 20. The two polymer solutions and the cell-containing medium are injected into three separate inlets of the first cross junction where they meet and form a coflowing stream in the microchannel. In the second junction, this stream breaks up into droplets by flow focusing with a paraffin oil. Subsequent mixing of the three liquids inside the droplets leads to homogenisation of droplet content and gelation reaction [73]. Another example of polymer-polymer crosslinking by click chemistry is the reaction between azide-functionalized poly(N-(2-hydroxypropyl)-methacrylamide) (PHPMA) chains and cyclooctyne-functionalized poly(N-isopropylacrylamide) (PNIPAAm) and poly(ethylene glycol) (PEG) chains [74]. Figure 20. Formation of cell-laden polyethyleneglycol (PEG)-hyperbranched polyglycerol (hPG) microgel particles by thiol-ene "click" reaction. The gelation is achieved by nucleophilic Michael addition of PEG-dithiol to acrylated hPG building blocks without any initiator or UV light [73,75].
Thiol-terminated PEG (PEG-dithiol) can be synthesized by the reaction between PEG-diamine and 2-iminothiolane at room temperature, as shown in Fig. 21. The chemical structure of microgel particle after crosslinking is shown in the same figure.

Photopolymerisation
In this approach, droplets composed of a mixture of functional monomers and photoinitiators are exposed to UV or visible light to initiate free-radical polymerisation, as shown in Table S6. Photopolymerization offers several advantages compared to thermal or redox initiation, including short crosslinking times, ambient reaction temperatures, and high spatial and temporal reaction control [76]. Formation of biocompatible hydrogels requires the use of cytocompatible photoinitiators, such as Irgacure 2959, 1173, 819, and 651, riboflavin phosphate, camphorquinone, and eosin Y. Visible light photoinitiation is advantageous for encapsulation of biological materials since UV radiation can cause DNA damage and accelerate tissue aging and cancer onset. Blue light photo-initiators that can be used are camphorquinone [77], eosin Y [78], and riboflavin [79,80].
Linear pNIPAAm chains with pendant dimethylmaleimide (DMMI) side groups can be crosslinked by dimerization of DMMI moieties upon UV exposure in the presence of a triplet sensitizer, Fig. 23. The concentration of the poly(NIPAAm-DMMI) precursor inside droplets should be in the semi dilute non-entangled concentration regime, therefore above the overlap concentration of polymer coils, * to ensure that the polymer chains are close enough to undergo crosslinking, but below the entanglement concentration, to avoid excessive viscosity of the dispersed phase [88]. The device shown in Fig. 23(a) is used to generate structurally homogeneous microgels. Structural microgels (core-shell microgels with Janus shells or hollow Janus shells) can be produced using the same the poly(NIPAAm-DMMI) polymer labelled with different fluorescent dyes and a modified microfluidic chip with two sequential cross junctions [92]. Biocompatible poly(ethylene glycol diacrylate) (PEGDA) and poly(ethylene glycol methyl ether acrylate) (PEGMA) microgels can be generated via UV light-induced free radical polymerization of single emulsion drops generated using aqueous PEGDA or PEGMA solutions containing a suitable photoinitiator [93], as shown in Table S6 and Fig. 24. PEGDA microgels can be also generated using W/O/W double emulsion droplets containing PEGDA and photoinitiator in the inner-most drop to minimise the use of oil phase and simplify washing steps [94]. The continuous phase can be mineral oil [90], fluorocarbon oils [93] or hydrocarbon oils [83] containing Span 80, ABIL EM 90 or fluorinated surfactants. Droplets are polymerized in a wavy downstream channel within 5 s at the UV light intensity of 70 mW/cm 2 [90]. A jacketed wavy channel can be used to supply nitrogen under pressure through PDMS wall into the main channel to prevent scavenging of radicals by oxygen [93].
Natural polymer conjugated with photopolymerizable groups are attractive alternatives to synthetic hydrogels, because they can combine light polymerizable groups with inherent cell adhesion properties, due to the presence of natural cell-binding motifs, and excellent biodegradability, due to the presence of enzyme-sensitive links [95]. The example of such modified natural polymers is gelatin methacryloyl (GelMA), which is synthesized through the reaction between gelatin and methacrylic anhydride, Figure 25(a) [86,95]. The conjugation of the methacryloyl moieties occurs mainly on primary amine groups of lysine and hydroxylysine residues, but hydroxyl groups of serine, threonine, hydroxyproline, and hydroxylysine residues are also affected [96]. GelMA microgels can be fabricated in a flow focusing microfluidic device using the dispersed phase composed of 8 wt% GelMA and 0.2-0.5 wt% Irgacure 2959 dissolved in PBS and exposing droplets to UV light, Figure 25(b) [97,98]. Generated MelMA beads can be coated with silica hydrogel to protect encapsulated cells from oxidative stresses [86].

Gelation of droplets by temperature trigerred sol-gel transition
The gelation of stimuli-responsive polymer solutions can be achieved via a sol-gel transition triggered by a temperature or pH change. Thermo-responsive hydrogels can be divided into two groups: (1) upper critical solution temperature (UCST) hydrogels and (2) lower critical solution temperature (LCST) hydrogels [99]. UCST hydrogels such as gelatin and agarose are formed by cooling polymer solution to below a UCST. Cell-loaded agarose beads can be produced by flow focusing a cell-laden agarose solution at elevated temperature [100][101], Fig. 26. Agarose beads can also be generated by in situ mixing of cell suspension and molten agarose just before droplet generation [102]. Furthermore, agarose beads can be produced by droplet generation at room temperature using ultra-low gelling agarose with a UCST < 17 °C [103,104]. Once solidified, the beads remain solid up to 56 °C and keep their integrity during handling at room temperature. Agarose droplets can be solidified off-chip in an ice bath [100] or by trapping droplets on-chip and exposing the chip to low temperatures [105].
The number of cells entrapped per droplet is dictated by Poisson distribution [106]. Single-cell encapsulation can be achieved using a diluted cell suspension in the dispersed phase so that ~10% of agarose droplets will contain single cells and most of the remaining droplets will be empty [101]. The entrapment of single cells within monodispersed beads ensures monoclonal cultivation and provides identical growth conditions [107]. After incubation period, the beads are washed with a buffer solution to remove the oil phase and sorted by Fluorescence Activated Cell Sorting (FACS) based on their fluorescence. The cells with improved properties can be recovered from the beads by enzymatic degradation of agarose with agarase and can be subjected to the next round of random mutagenesis. Figure 26. Encapsulation of single E. coli cells within agarose beads by temperature-controlled emulsification of agarose solution in a flow-focusing device followed by cooling. The entrapped cells grow into monoclonal microcolonies inside the beads and can be sorted by FACS [100].
Gelatin microgels were fabricated by injecting 5 wt% gelatin solution at 40 C through silicon microchannel arrays into isooctane [108]. Droplets were generated by step microfluidic emulsification and solidified by cooling the emulsion to 25 C. After a slow gelation at 25 C overnight, the droplet solidification was completed at 5 C.

External gelation of charged polymers
Core-shell microgel particles with an oil core and hydrogel shell were produced in a 4-phase glass capillary device by on-chip crosslinking of alginate solution in core/shell double emulsion droplets [109]. In this process, an O/W1 emulsion composed of soybean oil droplets dispersed in Na-alginate solution was prepared in the emulsification tube I and further emulsified in soybean oil in the emulsification tube II to form O1/W/O2 emulsion with core-shell droplet morphology (Fig. 27). The alginate solution in the shell was solidified in the collection tube by a CaCl2 solution. Since double emulsion droplets are formed via a two-step process using two dripping instabilities, the diameters of inner and outer droplets can be controlled independently by adjusting the diameters of both emulsification tubes and fluid flow rates [109]. Figure 27. Co-flow glass capillary device for fabrication of oil-core alginate-shell microgels via onchip gelation of shell solution within O1/W/O2 emulsion droplets. The droplets sink to the bottom of the collection tube due to density difference between the formed beads and the oil phase [109].
Core-shell microgels with an oil core and a gellan shell were fabricated using O/W/O double emulsion templates. On-chip crosslinking of gellan polymer chains in the middle phase was induced by calcium acetate dissolved in the outer oil phase, Figure 28 [110].

Internal gelation of charged polymers
Core-shell microgels can be generated by internal gelation of core-shell droplets in W/W/O or O/W/O double emulsion. The crosslinking can be achieved by adding a noncharged calcium compound such as Ca-EDTA [111,112] CaCO3 [113] or CaCO3/PAG mixture [35] to the aqueous polymer solution in the middle phase. The crosslinking reaction can be trigerred either by acetic acid added to the outer oil phase or by UV irradiation of template droplets. Typical emulsion formulations are symmmarised in Table S7.
The fabrication of water-core alginate-shell microgels using W/W/O core/shell template droplets is shown in Fig. 28. The alginate chains in the shell can be crosslinked by the release of Ca 2+ from Ca-EDTA [112] or CaCO3 [114], trigerred by the diffusion of acetic acid dissolved in the outer oil phase. By encapsulating different types of liver cells in the core and shell region, it is possible to mimic the structure of human liver and create a portable artificial liver in each hydrogel particle. In a similar way, core-shell beads consisting of a mixture of Matrigel, collagen and alginate in the core (solidified by temperature control) and alginate in the shell (solidfied by internal gelation using CaCO3 and acetic acid) were produced via a W/W/O template emulsion [113]. Figure 28. Fabrication of core-shell microgels consisting of an aqueous core and an alginate shell via W/W/O double emulsion template droplets. Alginate in the shell is crosslinked by in situ triggered release of Ca 2+ from Ca-EDTA. Different types of liver cells can be encapsulated in the core and shell regions to create artificial liver on a microgel [112].
Hollow microgel capsules can be generated by internal gelation of core/shell droplets within O/W/O double emulsion [111,115]. In this approach, usually Ca-EDTA or CaCO3 is added to the alginate solution in the middle phase and crosslinking is initiated on-chip by organic acid dissolved in the outer oil phase [115] or off-chip by collecting the formed double emulsion droplets in an acidified oil phase [111]. Alternatively, a photo-acid generator (PAG) can be added to the middle phase to trigger crosslinking by UV light. The fabrication of capsules by UV irradiation of template droplets containing a mixture of alginate, PAG and CaCO3 in the middle phase in shown in Fig. 29 (Table S8). Figure 29. Formation of alginate capsules with an oil core by internal gelation of the middle phase of O/W/O emulsion droplets containing CaCO3 and photo-acid generator (PAG). The crosslinking was triggered by UV irradiation which releases acid from PAG and dissolves CaCO3 [35].

Photocrosslinking
The oil-free, organic-solvent-free, and surfactant-free synthesis of core-shell microgels using W/W/W emulsion droplets as templates is shown in Fig. 30. Aqueous solutions of dextran (DEX) and polyethylene glycol (PEG) at sufficiently high concentrations form an aqueous two-phase system (ATPS) composed of DEX-rich and PEG-rich solutions, which can be exploited in "all-aqueous" multiphase microfluidics [116]. To form droplets composed of a PEG core and a DEX shell (Fig. 30), a stream of aqueous PEG solution was focused by a stream of aqueous DEX solution at the upstream junction, and the resulting thread was broken-up into droplets by another stream of PEG solution at the downstream junction. The formed core-shell droplets were exposed to UV light to crosslink DEX in the shell via the thiol-ene reaction between alkyne-functionalized dextran (DEX-GPE) and thiol-functionalized dextran (DEX-SH) chains (Fig. 30b). To prevent loss of photoinitiator from the middle phase, it was added to all phases at equal concentration (Table S9). Core-shell microgel particles consisting of a polyacrylamide (PAAm) core and a poly(N-isopropylacrylamide) (PNIPAAm) shell were fabricated using solid-in-water-inoil (S/W/O) template emulsion [118]. In this strategy, pre-synthesized PAAm microgel particles were encapsulated in a shell liquid composed of a semi dilute solution of photocrosslinkable P(NIPAAm-DMMI) chains. The shell was crosslinked by UV light in the presence of a triplet sensitizer, Fig. 23(b). Core-shell gelatin methacryloyl (GelMA) microgels were fabricated from core-shell template droplets consisting of a photocurable GelMA solution in the shell and a methyl cellulose solution in the core [119].

Microfluidic production of structured microgels
Microfluidics can be used to fabricate structurally heterogeneous microgels, such as multi-core, multi-shell, Janus, and composite microgels [120]. Janus particles shown in Fig. 31(a) consist of three distinguishable PNIPAm polymers (colorless and tagged with different fluorescent dyes) [92]. The central PNIPAm stream forms a colorless core of the droplets, while the two side PNIPAm streams tagged with different fluorescent dyes form a Janus shaped shell. The hollow microspheres shown in Fig. 31(b) are formed from a O/W/O emulsion with a Janus-shaped middle phase. In both cases, template droplets are crosslinked by UV-induced dimerization of dimethylmaleimide (DMMI) side groups on a polymer backbone. Janus microgels can also be generated by photopolymerization of phase-separated and dewetted crosslinkable polymer droplets [121]. Janus microgels produced from three distinguishable PNIPAm streams; (b) Hollow Janus microgels produced from two distinguishable PNIPAm streams and one inert oil stream [92].
Microfluidics can be used to produce composite microgels with heterogeneous chemical structure. The examples of such composite microgel particles are pH-responsive Eudragit S100 microgel encapsulated in a poly(lactic acid) (PLA) shell, Fig. 32(a) [122], droplet interface bilayers (DIBs) encapsulated in an alginate shell, Fig. 32(b) [123], distinct hydrogel beads encapsulated in a water droplet, Fig. 32(c), and microgel scaffolded oil-inwater-in-oil emulsion droplets, Fig. 32(d). Droplet interface bilayers (DIBs) can be formed via a W/O/W/O triple emulsion through contact of aqueous droplets in an oil phase in the presence of dissolved lipids) [123]. A semipermeable microgel shell allows a structural stability of DIBs, while allowing communication with the environment. Figure 32. Composite microgels prepared in microfluidic devices: (a) pH-responsive Eudragit S100 microgel encapsulated within a Nile Red-labelled poly(lactic acid) (PLA) shell. At pH > 7, the core is a colorless liquid due to hydrophilic character of charged Eudragit chains. At pH < 7, the core is a red-colored solid gel due to diffusion of Nile red from the shell. A small amount of water was expelled from the gel after sol/gel transition [122]; (b) Droplet interface bilayers (DIBs) formed in an oil phase and encapsulated within an alginate shell [123]; (c) Gel beads from two distinct polymers encapsulated in a single water drop using a microfluidic particle zipper [124]; (d) A single gel bead and oil droplet encapsulated within a larger water droplet [125].

Conclusions
Microfluidic devices can be used to produce monodispersed microgel particles of versatile chemical composition, physical properties and morphology including composite microgels with complex internal structure composed of different solid and liquid phases. Microfluidic flow configurations traditionally used for microfluidic emulsification, such as cross, T-, Y-, and -junctions, can be combined in different ways in a manifold to generate template droplets and crosslink them into spherical and non-spherical monodispersed particles. Microgels can be loaded with cells and their spatial arrangement and number can be controlled with high precision. Versatile internal and external ionic crosslinking methods have been developed to crosslink charged polymers without channel clogging. Covalently crosslinked microgels with mechanically improved properties compared to their ionically crosslinked counterparts were produced by introducing moieties that can allow enzymatic, photo-induced, and click chemistry crosslinking. The choice of the most appropriate continuous and disperse phase formulations, channel geometry, and crosslinking strategy is ultimately dictated by the fluid dynamics of the selected droplet generation process, the diffusion kinetics and solubility of the crosslinking agents, the kinetics of the crosslinking reaction, and the targeted applications for which the microgels are designed. For cell culture applications, the selected crosslinking process should not compromise cell viability.

Supplementary Materials:
The following are available online at www.mdpi.com/xxx/s1: Tables from S1 to S9 containing dispersed and continuous phase formulations used to form different types of microgels in microfluidic devices.