Microfluidic production of aqueous suspensions of gellan-based microcapsules containing hydrophobic compounds

Chemical Engineering Science 211 (2020) 115314

Contents lists available at ScienceDirect

Chemical Engineering Science journal homepage: www.elsevier.com/locate/ces

Microfluidic production of aqueous suspensions of gellan-based microcapsules containing hydrophobic compounds Mariano Michelon, Bruna C. Leopércio, Marcio S. Carvalho ⇑ Laboratory of Microhydrodynamics and Flow in Porous Media, Department of Mechanical Engineering, Pontificia Universidade Catolica Rio de Janeiro, Rio de Janeiro 22451900, RJ, Brazil

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


Gellan-based microcapsules produced

from O-W-O double emulsion templates.
System for encapsulating hydrophobic active agents in a suspending aqueous medium.
Operability window of the microfluidic process for double emulsion template.

a r t i c l e

i n f o

Article history: Received 3 June 2019 Received in revised form 16 October 2019 Accepted 23 October 2019 Available online 24 October 2019 Keywords: Microfluidics Gellan gum Biopolymer Microcapsules

a b s t r a c t We discuss the production of microcapsules with hydrogel-based shells by ionotropic gelation of gellan gum from monodispersed oil-in-water-in-oil (O/W/O) double emulsion templates obtained using glasscapillary microfluidic devices. An oil extraction step was added after the shell gelation process to enable the dispersion of the microcapsules in an aqueous medium. The proposed approach for producing biopolymer-based microcapsules for hydrophobic agents dispersed in a water phase has a broad range of applications in delivery of hydrophobic compounds dispersed in an aqueous medium. We report the operability window for the production of monodispersed microcapsules as a function of the flow rate of each fluid phase and the dimensions of the device. Microcapsules with mean diameters ranging from 95 to 260 mm and a maximum coefficient of variation of 5% were formed. The results show how to independently control the capsule diameter and shell thickness by varying the outer and middle phase flow rates. Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction

⇑ Corresponding author. E-mail addresses: [email protected] (M. Michelon), bruna@lmmp. mec.puc-rio.br (B.C. Leopércio), [email protected] (M.S. Carvalho). https://doi.org/10.1016/j.ces.2019.115314 0009-2509/Ó 2019 Elsevier Ltd. All rights reserved.

Biopolymer-based microcapsules are highly hierarchized structures with submillimeter dimensions containing an inner phase and a biopolymer shell. In general, they are obtained after a solidification step of the polymer-based middle phase of a double

2

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

emulsion template. Microcapsules are promising candidates for encapsulating, delivering, and controllably releasing active ingredients, including surfactants for enhanced oil recovery, agricultural substances, cosmetic components, building materials, food additives and pharmaceuticals (Amstad, 2017; Datta et al., 2014; Abbaspourrad et al., 2013; Lapidot et al., 2003; Kong et al., 2014; Xu et al., 2018; Zhao et al., 2011; Werner et al., 2018; Michelon et al., 2019). These applications often require a shell that effectively retains the inner phase and releases its content only after exposure to specific external conditions, such as heat, pH, or physical stresses (Abbaspourrad et al., 2013; Abate et al., 2011). The use of biocompatible or biodegradable polymers as the shell material broadens the range of applications to areas at which interaction with living organisms and natural degradation of the capsules are required. The preparation of double emulsion templates by conventional methods is not trivial and, in general, two emulsification steps are needed (Balcaen et al., 2016; Akamatsu et al., 2015). First, an emulsion is formed from high shear mixing of two immiscible phases. In a second step, a double emulsion is formed using milder shear conditions to avoid the disruption of the innermost phase. The highly variable shear inherent in these conventional emulsification routes results in microcapsules with a broad size distribution, making it difficult to control the shell thickness, and consequently, generating poorly controlled encapsulation and release characteristics (Zhao et al., 2011; Abbaspourrad et al., 2013; Balcaen et al., 2016; Chen et al., 2012). Microfluidic technology offers a route to overcome these drawbacks, emerging as a promising alternative technique for the fabrication of monodispersed microcapsules with well-controlled release characteristics. The formation of double emulsion templates in capillary microfluidic devices enables a fine control of the template dimensions and offers high flexibility with regard to the materials that can be used to form the microcapsule shell (Zhao et al., 2011; Chen et al., 2012). Utada et al. (2005) proposed a microfluidic process involving co-flow and flow-focusing of three fluids inside an array of glasscapillaries that promotes the generation of double emulsion droplets at the exit of the injection tube (Utada et al., 2005). The liquid properties and flow rate of each phase need to be adjusted in order to achieve the desired continuous dripping flow regime. Using this microfluidic approach, a variety of hydrophilic or hydrophobic active compounds can be truly encapsulated as the inner phase of water-in-oil-in-water (W/O/W) or oil-in-water-in-oil (O/W/O) double emulsion templates, respectively. Then, the middle phase of these double emulsions is solidified to form stable and robust microcapsules, ensuring that a solid layer separates the inner phase from the suspending liquid (Zhao et al., 2011; Kim et al., 2011). W/O/W double emulsions have been widely used as templates for microcapsules, because its aqueous core provides an ideal environment for the dissolution of hydrophilic compounds, whereas the microfluidic formation of O/W/O templates for the encapsulation of hydrophobic compounds has been less reported. This fact can be associated to: (i) the shear-thinning behavior of aqueous polymer dispersions used to form O/W/O templates, that may hinder the flow stabilization; (ii) the difficulty on achieving the continuous dripping flow regime due to the high viscosity of the inner and outer oil phases; (iii) the difficulty to choose the ideal surfactant to stabilize each interface, as well as the oil-types to compose the inner and outer phases; (iv) the scarce or insipient alternatives to solidify water-based polymers, as well as to control their crosslinking rate and (v) the continuous phase of the final microcapsules dispersions being an oil phase, which may be a limiting factor to some applications. We report the production of biocompatible microcapsules from O/W/O double emulsion templates using a glass-capillary microfluidic device. The capsules were collected in a vial containing hexane and acetate buffer solution was added in order to disperse the pro-

duced microcapsules in an aqueous phase. We have used gellan gum as the biocompatible polymer aqueous middle phase, as an alternative to calcium alginate, commonly used for the hydrogelbased capsules production (Xu et al., 2018; Liu et al., 2013; Silva et al., 2018). Gellan gum is a linear, anionic, and high molecular weight exopolysaccharide secreted by Pseudomonas elodea. It is biocompatible and approved by the USFDA and European Union acting as a stabilizer, thickening and gelling agent in a wide variety of products. The gellan gum is capable to form stable hydrogel structures by ionotropic process, i.e., physical crosslink induced by the presence of cations (Agnello et al., 2017; Bacelar et al., 2016). These structures are stronger and less permeable then alginate-based gels and moreover, are thermo-responsive (Graham et al., 2019). These characteristics broaden the application range of water-based microcapsules. The proposed method enables the production of monodispersed microcapsules containing hydrophobic compounds and dispersed in an aqueous phase. We define the operability window based on the flow rate of each phase at which the continuous dripping flow regime occurs and monodispersed double emulsions are formed with a single inner phase core. Moreover, we show how the microcapsule diameter and shell thickness can be controlled by changing the flow parameters.

2. Materials and methods 2.1. Microcapsules production The O/W/O templates were produced by microfluidics. The three-dimensional coaxial microfluidic devices used were based on the configuration proposed by Utada et al. (2005) and sketched in Fig. 1 (Utada et al., 2005). They were built on a glass slide, and consisted of two cylindrical glass-capillaries (World Precision Instruments Inc., USA), with inner and outer diameters of 0.58 mm and 1 mm, inserted into the opposite ends of a square capillary with inner dimension of 1.05 mm (Atlantic International Technology Inc., USA). The cylindrical glass-capillaries were tapered to an inner diameter of approximately 20 mm with a micropipette puller (model P-1000, Sutter Instrument Co., USA). Then, the tips were carefully sanded to the desired final inner diameter. The collection capillary (right capillary in Fig. 1b) has a larger inner tip diameter (/c ) and was treated with a commercial rain repellent Glass Shield (Inove Pack do Brasil, Brazil) to render a hydrophobic surface. The injection capillary (left capillary in Fig. 1b) has a smaller inner tip diameter (/i ) and was treated with a polyelectrolytes solution composed of 1 wt% poly(acrylamide-co-diallydimethyl-a mmonium chloride) (Sigma-Aldrich, USA) and 2 mol/L NaCl to render a hydrophilic surface. Both treatments were performed over 60 min. For assembling the microfluidic device, the square capillary was fixed to the slide with EpoxyÒ resin (Devcon Corp., USA). Afterwards, the treated capillaries with sanded tips were inserted into the square capillary at both ends, which enabled the alignment of the axes of the capillaries, maintaining a desired separation distance (l) between them. Finally, stainless steel dispensing needles of inner and outer diameters 0.66 mm and 0.91 mm (model 304, McMaster-Carr, USA) were placed at the junctions between capillaries or at their ends, and fixed to the slide with EpoxyÒ resin. To determine the effect of the device geometry on the range of flow rate at which the continuous dripping flow regime is observed and on the diameter and shell thickness of the capsules, we have used two microfluidic designs, with different values of /i , /c and l, as shown in Fig. 1d. In the three-dimensional coaxial microfluidic devices, the inner phase flows through the injection capillary, while the middle and continuous phase flow in opposite directions through the interstices

3

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

Fig. 1. (a) Perspective view of the coaxial glass-capillary device; (b) detail of the droplet generation region. Q i , Q m and Q o are the flow rate of inner, middle and continuous phases, respectively; (c) image of the O/W/O template formation in the ideal intermittent dripping regime and (d) geometric parameters of the two designs used. /i and /c are the inner diameter of the injection and collection capillary tubes, respectively, and l denotes the separation distance between them.

between the cylindrical and square capillaries, as sketched in Fig. 1b. The flow rates of each of the three phases (Q i , Q m and Q o denotes the flow rate of the inner, middle and outer/continuous phases, respectively) were controlled by three syringe-pumps (model Pump 11, Harvard Apparatus, USA). The O/W/O template formation was monitored within the microfluidic device using an inverted microscope (model DMi8, Leica Microsystem, Germany) equipped with a highspeed camera (model Fastcam SA-3, Photron, USA). The inner phase was a refined commercial sunflower oil (Liza, Cargill Agricola S.A., Brazil) labeled with an orange food-grade dye. The biopolymer-based middle phase consisted of a mixture of 0.5 wt% low-acyl gellan gum KelcogelÒ CG-LA (CP Kelco Brasil S/A, Brazil) and 2 wt% polyoxyethylene sorbitan monolaurate, TweenÒ 20 (Sigma-Aldrich, USA), in ultrapure water with resistivity 18.2 MX/cm (Direct-Q3 UV System, Millipore Co., USA). The middle phase was previously mixed at 80 °C for 10 min under magnetic stirring. The continuous phase was a sunflower oil dispersion containing 1 wt% calcium acetate (Sigma-Aldrich, USA) and 5 wt% polyglycerol-polyricinoleate commercially named GrinsteadÒ PGPR super (Danisco Brasil, Brazil). The density (q) of each phase was measured in a digital densimeter (model DMA 4200M, Anton Paar, Austria) at 25 °C and is presented in Table 1. Rheological measurements for all fluids used were performed in a stress-controlled rheometer (model DHR-3, TA Instruments, USA) using a stainless steel Couette geometry. The shear rate ranged from 1 to 1000 s1 and the flow curves were obtained in sequential three flow steps: up-down-up cycles. The viscosity as a function of shear rate for the three phases is presented in Fig. 2. The inner and continuous oil phases show Newtonian behavior, gi = 55.3 mPa s and go = 73.9 mPa s, which is a typical flow behavior of vegetable oils (Franco and Nguyen,

Table 1 Density (q), viscosity (g) and interfacial tension (r) of the phases at 25 °C. Phase

Composition

q (kg/m3)

g (mPa s)

r (mN/m)

Inner (Oi) Middle (W)

Sunflower oil Ultrapure water Gellan gum (0.5 wt%) TweenÒ 20 (2 wt%) Sunflower oil PGPR (5 wt%) Calcium acetate (1 wt%)

752.8 ± 0.1b 1019.1 ± 0.9a

55.3 21.6–291.4

5.4 ± 0.3a

755.2 ± 0.2b

73.9

Continuous (Oo)

2.7 ± 0.2b

Different lowercase letters (a-b) represent significant differences (p < 0.05) between the phases.

2011). The viscosity difference between the oil phases is associated with the presence of calcium acetate in the continuous phase. The biopolymer-based phase showed shear thinning behavior, with the viscosity varying from gm = 21.6 mPa s to gm = 291.4 mPa s in the range of shear rate explored. Therefore, the viscosity value of the middle phase changes with flow rate and may affect the flow regime in the droplet formation region. We estimated a deformation rate value of approximately 103 s1 at the droplet breakup region, which agrees with the shear rate range of 102–103 s1 estimated by Steegmans et al. (2009). The interfacial tensions (r) were measured by Wilhelmy plate method, using a digital tensiometer (model DCAT 25, DataPhysicsÒ, Germany) and are also presented in Table 1. We observed a reduction of the interfacial tension between the middle and continuous phases, ro = 2.7 mN/m, in comparison with interfacial tension between middle and inner

4

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

The mechanism of O/W/O templates formation in coaxial glasscapillary devices is governed by the interactions between viscous, inertial and interfacial forces. Therefore, the flow dynamics is a function of the capillary number and Weber number of each phase

Fig. 2. Flow curves at 25 °C: ( ) inner phase, (j) middle phase and ( ) continuous phase.

phases, ri = 5.4 mN/m. PGPR and TweenÒ 20 act as surfactants in the system favoring the formation and stabilization of the O/W/O templates. Apart from that, calcium acetate can also act as a co-surfactant, which can promote an additional stability for the O/W/O droplet before its full gelation (Costa et al., 2017). 2.2. Recovery and characterization The microcapsules were collected in a glass vial containing a small volume of hexane. Then, to obtain an oil-free capsular dispersion, a small volume of acetate buffer (0.074 mol/L, pH 4.5) was added to the vial. The hexane excess containing the oil from the continuous phase was removed and the residual hexane was evaporated at room temperature during 24 h. The micrographs of the microcapsules in the acetate buffer were obtained using an inverted microscope (model DMi8, Leica Microsystems, Germany). The micrographs were processed using the software Leica Application Suite X (Leica Microsystems, Germany) to determine the outer diameter (D) and the shell thickness (t) of the microcapsules. The polydispersity of the capsular systems was expressed in terms of coefficient of variation (CV) relating standard deviation to mean diameter. 3. Results and discussion 3.1. Microfluidic production In our microfluidic process, as shown in Fig. 1c and Video 1, due to the hydrophilic nature of the treated injection capillary, the inner oil phase (Oi) flows through the tip of the injection capillary without wetting its outer surface, while the aqueous biopolymer middle phase (W) flows along the outer wall of the capillary, coaxially shielding the innermost stream. In addition, the hydrophobic treatment of the collection capillary ensured that the middle phase flows through the center of the collection capillary without contacting the glass-capillary wall.

Video 1.

Fig. 3. (a) Operability window as a function of the flow rates (Q m and Q o ) at a fixed inner phase flow rate, Q i = 200 mL/h using Design #1 (/i = 80 mm; /c = 350 mm and l = 120 mm). (b) Operability window represented as a function of the outer and middle phases capillary numbers: (h) monodispersed O/W/O templates, (▲) jetting regime, formation of double emulsion with multiple inner compartments (d) formation of droplets containing only the middle phase without the inner core together with the formation of polydispersed microcapsules, (j) back flow regime, the outer phase invades the injection channels of the inner and middle phases. (c) example of a condition visualized at high Q m values, which led to formation of polydispersed microcapsules with multiple inner compartments.

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

(Anna, 2016). The desired dripping regime is the result of an absolute instability, wherein the viscous and capillary force balance is such that both interfaces, one between the inner and middle phases (Oi/W) and the other between the middle and continuous phases (W/Oo), break simultaneously at the same spatial location and frequency, which depends on the flow parameters, leading to the formation of monodispersed droplets (Nabavi et al., 2017). This regime only occurs at a small range of flow rates and the flow rates of the inner (Q i ), middle (Q m ), and continuous (Q o ) phases have a strong effect on the breakup dynamics of the two coaxial interfaces. The O/W/O templates were transformed into monodispersed biopolymer-based microcapsules through gelation of the middle phase induced by the calcium ions present in the continuous phase. We investigated the effect of the flow rates of each phase on the O/W/O template generation for each of the two designs explored using the same device, to avoid any small variation in flow velocity associated with small differences in the geometry from device-todevice of the same design. We determined the range of parameters at which the dripping flow regime is observed and monodispersed microcapsules with a single inner core are produced and how the microcapsule diameter and shell thickness vary with the flow parameters inside the operability window. At each condition, we measured 10 microcapsules and report the average value and the standard deviation (error bars in the plots). 3.1.1. Microfluidic design #1 (/i = 80 mm, /c = 350 mm and l = 120 mm) We first fixed the value of the inner phase flow rate to Q i = 200 mL/h and varied the flow rates of the middle and outer

5

phases over a wide range. The observed flow regimes are summarized in Fig. 3a as a function of the outer and middle phase flow rates and in Fig. 3b, as a function of the outer Cao and middle Cam phase capillary numbers, defined as lm ðQ m þQ i Þ o . Cao ¼ r l/o Q2 / 2 and Cam ¼ rmo /2i mo ð c i Þ

o The Reynolds number of each phase, defined as Reo ¼ l qð/ocQþ/ iÞ o iÞ and Rem ¼ qm ðlQ m/þQ , varied from 0:026 < Re and o < 0:4 i o

0:18 < Rem < 0:8. The open symbols represent flow conditions at which the breakup of the inner and middle phases occurs simultaneously and at the same position, near the entrance of the collection capillary, as shown in Fig. 1c. Double emulsion is generated with a single inner phase drop. This is the desired continuous dripping regime at which monodispersed microcapsules are produced. The filled symbols represent flow conditions that did not lead to the desired dripping regime. At a fixed value of the inner and outer flow rates, the formation of monodisperse O/W/O templates with a single inner core only occurs inside a certain range of the middle phase flow rate Q m . If the middle phase flow rate (capillary number) is above a critical value, the inertial force of the middle and inner phases becomes comparable to the capillary forces, a long jet is formed and the drop breakup occurs far from the entrance of the collection capillary, in what is called the jetting regime (Utada et al., 2005; Vladisavljevic´ et al., 2012). At these conditions, the drop breakup is quite irregular leading to polydisperse capsules. Moreover, because of the relatively high middle phase flow rate, the breakup

Fig. 4. (a) Microcapsule diameter (D) as a function of middle and outer phase flow rates (Q m and Q o ) at a fixed Q i = 200 mL/h with Design #1 (/i = 80 mm; /c = 350 mm and l = 120 mm) and (b) power-law fit of the diameter in units of the collecting capillary diameter (continuous line) as a function of Q o =ðQ i þ Q m Þ; Q m = ( ) 25 mL/h, ( ) 50 mL/h, (▲) 100 mL/h, ( ) 150 mL/h, ( ) 200 mL/h, ( ) 250 mL/h and (r) 300 mL/h; (c) Examples of micrographs and particle size distributions of the microcapsules produced at different Q o : left (Q i = 200 mL/h, Q m = 200 mL/h and Q o = 1000 mL/h) and right (Q i = 200 mL/h, Q m = 200 mL/h and Q o = 4500 mL/h).

6

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

time of the inner phase is lower than that of the middle phase, leading to multiple inner phase drops inside a single middle phase drop, as shown in Fig. 3c. The flow conditions that correspond to the jetting regime are marked with the filled triangles in Fig. 3a and b. On the other hand, if the middle phase flow rate is below a critical value, the shear force exerted by the outer phase along the interface is high, because of the large velocity difference between the outer and middle phases (Kim et al., 2013). The high shear force associated with a very thin middle phase filament leads to breakup prior to the complete growth of the inner core and the formation of droplets containing only the middle phase without the inner core together with the formation of polydispersed microcapsules. These flow states are marked by filled circles in Fig. 3a and b. The production of monodispersed O/W/O double emulsion does not occur if the outer phase flow rate is very high, flow states marked by filled squares in the plots. At this condition, the inner and middle phases flow back into the injection capillary and through the interstices between the square and cylindrical capillaries, respectively, disrupting the desired periodic dripping flow. In general, the range of the flow rates that led to formation of O/ W/O templates was approximately 10x smaller than the one

observed in the formation of the W/O/W templates (Chen et al., 2012; Kim et al., 2011; Vladisavljevic´ et al., 2012; Kim et al., 2013; Vilanova et al., 2013). This fact is probably associated to the differences in viscosity ratio between the phases, interfacial tension of the fluids and tip diameter of capillary tube used, which are affecting synergistically on the force balance. Inside the operability window that led to the ideal dripping regime and the formation of the monodispersed O/W/O templates, it was possible to accurately control microcapsules diameter by changing the outer phase flow rate Q o . Regardless the value of the middle phase flow rate Q m , the microcapsule diameter falls as Q o rises, as shown in Fig. 4a. As Q o increases, the viscous drag force between the continuous and middle phase increases, causing the droplets to detach quickly from the injection capillary tube. Thus, smaller microcapsules are produced at higher frequencies. Similar behavior has been reported in the production of W/O/W double emulsion templates (Utada et al., 2005; Kim et al., 2011; Nabavi et al., 2017; Vladisavljevic´ et al., 2012; Lee and Weitz, 2008). For the range of parameters explored, microcapsules were produced with a diameter ranging from 130 to 260 mm with a maximum coefficient of variation of 5% (Fig. 4c). The data is

Fig. 5. (a) Shell thickness in units of droplet radius, tðD=2Þ, as a function of Q m and Q o , at a fixed Q i = 200 mL/h using Design #1 (/i = 80 mm; /c = 350 mm and l = 120 mm). Q m : ( ) 25 mL/h, ( ) 50 mL/h, (▲) 100 mL/h, ( ) 150 mL/h, ( ) 200 mL/h, ( ) 250 mL/h and (r) 300 mL/h; (b) Shell thickness evaluated by Equation 2 (continuous line) as a function of the flow rate ratio Q m =Q i . Q o : ( ) 500 mL/h, ( ) 1000 mL/h, (▲) 1500 mL/h, ( ) 2000 mL/h, ( ) 2500 mL/h, ( ) 3000 mL/h, (r) 3500 mL/h, ( ) 4000 mL/h, ( ) 4500 mL/h and ( ) 5000 mL/h; (c) Examples of microcapsules with different shell thickness (paired red arrows) produced at different Q m : left (Q i = 200 mL/h, Q m = 250 mL/h and Q o = 1500 mL/h) and right (Q i = 200 mL/h, Q m = 50 mL/h and Q o = 1500 mL/h).

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

represented in dimensionless form in Fig. 4b, which shows the drop to collecting capillary diameter ratio as a function of the ratio of the outer phase flow rate to the sum of the middle and inner phase flow rates. The drop diameter, in units of the collecting capillary diameter, obtained in our setup was approximately half of the values reported in the production of W/O/W double emulsion templates (Kong et al., 2014; Vladisavljevic´ et al., 2012). The data for all values of Q m collapse in a single power-law curve (continuous line in

Fig. 4b). Vladisavljevic´ et al. (2012) have reported similar powerlaw dependence, with an exponent of 1/3.

 0:3 D Qo ¼ 0:85 Qm þ Qi /c

ð1Þ

The shell thickness of the produced microcapsules was calculated by the difference between the outer and inner drop radii. As expected, at a constant outer phase flow rate, the shell thickness (t) rises as the middle phase flow rate is increased; there is more material available to form the shell. Because the droplet diameter changes with Q o at a constant Q m , it is instructive to plot the shell thickness in units of microcapsule radius. Fig. 5a shows that the ratio thickness-radius, t=ðD=2Þ, remains almost constant as the outer phase flow rate varies, except at the very high middle phase flow rate, near the boundary of the operability window. The same behavior was reported by Nabavi et al. (2015) for W/O/W double emulsion template. Fig. 5b presents the shell thickness to microcapsule radius ratio as a function of Q m =Q i at different Q o . Lee and Weitz (2008), Hennequin et al. (2009), Vilanova et al. (2013) and Chen et al. (2012) have used mass balance to estimate the shell thickness as a function of Q m =Q i for W/O/W templates (Chen et al., 2012; Hennequin et al., 2009; Vilanova et al., 2013; Nascimento et al., 2017; Lee and Weitz, 2008). The proposed equation assumes that the volume of the middle phase does not change during the solidification process and was able to accurately describe the measured shell thickness for their experiments. In the production of microcapsules with hydrogel-based shell from O/W/O double emulsion templates, the volume of the middle phase is not constant during solidification process. This fact can be associated to the loss of aqueous mass from the shell during the gelation and recovery steps. Hydrogels based on gellan gum and divalent cations form bridges between the polymer chains, forming a porous structure that leaves significant amounts of free water in the gel network (Picone and Cunha, 2010; Santos and Cunha, 2018). Therefore, the thickness of the microcapsule shell is thinner than the thickness of the middle phase of the double emulsion template before solidification. We have incorporated the water loss phenomenon, quantified by the ratioa of the volume after and before solidification, in the mass balance to predict the shell thickness, according to Eq. (2):

  -1=3 t Qm ¼ 1 1 þ ð1  aÞ ðD=2Þ Qi

Fig. 6. (a) Operability window as a function of the flow rates (Q i and Q o ) at Q m = 100 mL/h with Design #1 (/i = 80 mm; /c = 350 mm and l = 120 mm). (b) Operability window represented as a function of the outer and inner phases capillary numbers: (h) monodispersed O/W/O templates, (.) jetting regime, formation of polydispersed microcapsules, (j) back flow regime, the outer phase invades the injection channels of the inner and middle phases. (c) Example of a condition visualized at high Q i values, which led to the appearance of a jetting flow with the droplet detachment far from the injection capillary tube.

7

ð2Þ

The amount of water loss from a hydrogel based on gellan gum is directly associated to the mechanism of gelation and polymer concentration. In general, gellan gum hydrogels at 0.5 wt% have shown variable amounts of water loss, ranging from 22% to 65% according to acid pH condition applied (Yamamoto and Cunha, 2007). To fit the data with Eq. (2), as shown in Fig. 5b, we consider a water loss equal to a = 0.5, which matches what was observed in a previous work that used calcium ions to gelify the gellan gum at 0.5 wt% (Vilela et al., 2012). For the specific conditions used in this study, we were able to produce microcapsules with minimum shell thickness of approximately 3 lm. The microcapsules with smaller shell thickness are more challenging to produce, as the thinner stream of the middle phase makes them more susceptible to break. The micrographs of Fig. 5c clearly show how the shell becomes thicker as the middle phase flow rate is increased. The large-shell thickness capsules, produced at high middle phase flow rate are eccentric and the shell thickness is not uniform.

8

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

In summary, the results show that at a fixed inner phase flow rate (Q i = 200 mL/h), the capsule diameter is mainly a function of the outer phase flow rate Q o and that the shell thickness can be tuned by changing the flow rate of middle phase Q m . Fig. 6a presents the operability window in a different plane of the parameter space. The middle phase flow rate was fixed at Q m = 100 mL/h, and the diagram is shown as a function of the inner and outer phase flow rates. The flow states map as a function of the outer Cao and inner Cai phase capillary numbers is shown if Fig. 6b. The inner phase capillary number is defined as

Cai ¼

li Q i rmi /2i

Monodispersed O/W/O double emulsion was only produced inside a narrow range of parameters, marked as open symbols in Fig. 6a and b. For inner phase capillary number higher than Cai  2, a jetting flow regime occurs, and droplet detachment occurs far from the tip of the injection capillary tube, as shown

in Fig. 6c. The jetting regime is marked as filled triangles in the maps and the produced microcapsules were polydisperse (Utada et al., 2005). At each value of Q i (Cai Þ;there was a maximum value of the outer phase flow rate (Cao Þabove which the inner and middle phases flow back into the injection capillary and through the interstices between the square and cylindrical capillaries. These conditions are marked as filled squares in the plots. Inside the operability window, the double emulsion diameter falls as the outer phase flow rate rises, as shown in Fig. 7a. All the data collapses into a single curve when plotted in dimensionless form, as presented in Fig. 7b. The drop diameter in units of the collecting capillary diameter is plotted as a function of Q o =ðQ i þ Q m Þ. The data is well fitted the same power-law relation (Eq. (1)), which is shown as a continuous line in the plot. The variation of the shell thickness, in units of capsule radius, as a function of the middle and outer phases flow rate is shown in Fig. 7c and d. Except for the very low inner phase flow rate (Q m =Q i = 4), the data is well described by Eq. (2) (continuous line

Fig. 7. (a) Influence of the flow rates (Q i and Q o ) at a fixedQ m = 100 mL/h using Design #1 (/i = 80 mm; /c = 350 mm and l = 120 mm) on the experimental microcapsule’s diameter (D); (b) Power-law fit of the diameter in units of collecting capillary diameter (continuous line) as a function of Q o =ðQ i þ Q m Þ and (c) shell thickness in units of droplet radius, tðD=2Þ, Q i : ( ) 25 mL/h, ( ) 50 mL/h, (▲) 100 mL/h, ( ) 200 mL/h and ( ) 300 mL/h; (d) predicted relative shell thickness by Equation 2 (continuous line) in function of the flow rate ratio Q m =Q i , Q o : ( ) 500 mL/h, ( ) 1000 mL/h, (▲) 1500 mL/h, ( ) 2000 mL/h, ( ) 2500 mL/h, ( ) 3000 mL/h, (r) 3500 mL/h, ( ) 4000 mL/h.

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

9

at which the monodispersed O/W/O double emulsions are produced is wider than the one presented in Fig. 3b for Design #1, with larger capillary tips. The maximum middle phase capillary number above which the jetting regime starts increased from Cam  0:75 to Cam  3, and the maximum outer phase capillary number above which back flow occurs increased from Cao  1:2 to Cao  2. Besides widening the operability window, the capillary device using Design #2 produced microcapsules with smaller diameter. As shown in Fig. 9a, the diameter of the microcapsules varied from 95 to 200 mm. In contrast, Design #1 produced microcapsules ranging from 130 to 260 mm. The reduction of the microcapsule diameter range as a function of the characteristic dimensions of the microfluidic device is mainly associated to an increase in the shear rate that pulls the middle and inner phase leading to a faster droplet detachment and consequently a reduction on the microcapsule diameter (Nabavi et al., 2017; Nabavi et al., 2017; Lee and Weitz, 2009). As in the previous case, the diameter in units of collecting capillary diameter is represented by a single power-law function (Eq. (3)) when plotted as a function of Q 0 =ðQ m þ Q i Þ, as shown in Fig. 9b. The relation has the same exponent of Eq. (1) with a slightly higher coefficient.

 0:3 D Q0 ¼ 1:05 /c Qm þ Qi

ð3Þ

The range of shell thickness in units of microcapsule radius was similar to the values obtained with Design #1. The dependence on flow rate is well described by Eq. (2), except at very low middle phase flow rate, close to the boundaries of the operability window.

Fig. 8. (a) Operability window as a function of the flow rates (Q m and Q o ) at a fixed Q i = 100 mL/h with Design #2 (/i = 50 mm, /c = 250 mm and l = 75 mm). (b) Operability window represented as a function of the outer and middle phases capillary numbers: (h) monodispersed O/W/O templates, (▲) jetting regime, with the formation of double emulsion with multiple inner compartments (d) formation of droplets containing only the middle phase without the inner core together with the formation of polydispersed microcapsules, (j) back flow regime, the outer phase invades the injection channels of the inner and middle phases.

in Fig. 7d). The dimensionless shell thickness rises as the inner phase flow rate falls. 3.1.2. Microfluidic Design #2 (/i = 50 mm, /c = 250 mm and l = 75 mm) In order to study the effect of the device geometry on the microcapsule production process and its dimensions, we ran experiments using smaller injection and collection capillary tips, called Design #2. The injection capillary tip /i and the distance between the injection and collection capillaries l were reduced by a factor of 0.625. The collection capillary tip was reduced by a factor of 0.7. Fig. 8a presents the operability window as a function of middle and outer phase flow rates at a constant inner phase flow rate Q i = 100 mL/h. The dimensionless flow map as a function of the outer and middle phases capillary numbers is shown in Fig. 8b. The range of dimensionless parameters that defines the conditions

4. Conclusion We successfully demonstrated the formation of O/W/O double emulsion templates aiming biopolymer-based microcapsules production through a glass-capillary microfluidic device using entirely FDA-approved materials. The O/W/O templates were transformed into solid microcapsules through an external ionotropic gelation of the middle phase induced by the calcium acetate. After gelation, the capsules were dispersed in an aqueous phase, enabling the encapsulation of hydrophobic actives in a water-based dispersion. The operability window of the process was determined as a function of the flow rate of each phase. We observed that fixing the flow rate of the inner phase and varying the flow of the middle phase is the best choice to create microcapsules with an accurate control of the shell thickness, while a systematic variation in the capsules diameter can be achieved by changing the flow rate of the continuous phase. The range of the microcapsule diameter scaled with the diameter of the collection capillary. Therefore, smaller capsules can be obtained using devices with smaller tip diameters of the injection and collection capillaries. The proposed methodology can potentially be used in a broad range of applications for encapsulating, delivering, and controllably releasing hydrophobic active compounds in a water based medium.

10

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314

Fig. 9. (a) Effect of the flow rates on the microcapsule’s diameter (D); (b) Power-law fit of the diameter in units of the collecting capillary diameter (continuous line) as a function of Q o =ðQ i þ Q m Þ and (c) shell thickness in units of droplet radius, t ðD=2Þ, Q m : (d) 12.5 mL/h, ( ) 25 mL/h, ( ) 50 mL/h, ( ) 75 mL/h, (◄) 100 mL/h, ( ) 125 mL/h, ( ) 150 mL/h, ( ) 200 mL/h and (j) 250 mL/h; (d) predicted relative shell thickness by Equation 2 (continuous line) in function of the flow rate ratio Q m =Q i , Q o : ( ) 500 mL/h, ( ) 1000 mL/h, (▲) 1500 mL/h, ( ) 2000 mL/h, ( ) 2500 mL/h, ( ) 3000 mL/h, (r) 3500 mL/h, ( ) 4000 mL/h.

Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments This work was performed in association with the ongoing project entitled ‘‘Complex dispersion flow through porous media” funded by Shell Brasil under the ANP R&D levy. The authors also would like to thank CP Kelco and DuPont Nutrition & Health for the donation of materials. References Abate, A.R., Thiele, J., Weitz, D.A., 2011. One-step formation of multiple emulsions in microfluidics. Lab Chip 11, 253–258. https://pubs.rsc.org/en/content/ articlehtml/2011/lc/c0lc00236d. Abbaspourrad, A., Carroll, N.J., Kim, S.H., Weitz, D.A., 2013. Polymer microcapsules with programmable active release. J. Am. Chem. Soc. 135, 7744–7750. https:// pubs.acs.org/doi/abs/10.1021/ja401960f. Abbaspourrad, A., Datta, S.S., Weitz, D.A., 2013. Controlling release from pHresponsive microcapsules. Langmuir 29, 12697–12702. https://pubs.acs. org/doi/abs/10.1021/la403064f. Agnello, S., Gasperini, L., Mano, J.F., Pitarresi, G., Palumbo, T.S., Reis, R.L., Giammona, G., 2017. Synthesis, mechanical and thermal rheological properties of new gellan gum derivatives. Int. J. Biol. Macromol. 98, 646–653. https://doi.org/ 10.1016/j.ijbiomac.2017.02.029.

Akamatsu, K., Kanasugi, S., Nakao, S., Weitz, D.A., 2015. Membrane-integrated glass capillary device for preparing small-sized water-in-oil-in-water emulsion droplets. Langmuir 31, 7166–7172. https://pubs.acs.org/doi/abs/10.1021/acs. langmuir.5b01514. Amstad, E., 2017. Capsules: their past and opportunities for their future. ACS Macro Lett. 6, 841–847. https://pubs.acs.org/doi/abs/10.1021/acsmacrolett.7b00472. Anna, S.L., 2016. Droplets and bubbles in microfluidic devices. Ann. Rev. Fluid Mech. 48–1, 285–309. https://doi.org/10.1146/annurev-fluid-122414-034425. Bacelar, A.H., Silva-Correia, J., Oliveira, J.M., Reis, R.L., 2016. Recent progress in gellan gum hydrogels provided by functionalization strategies. J. Mater. Chem. B 4, 6164–6174. https://pubs.rsc.org/en/content/articlepdf/2016/tb/c6tb01488g. Balcaen, M., Vermeir, L., Declerck, A., Van der Meeren, P., 2016. Influence of internal water phase gelation on the shear- and osmotic sensitivity of W/O/W-type double emulsions. Food Hydrocoll. 58, 356–363. https://doi.org/10.1016/ j.foodhyd.2016.03.011. Chen, P.W., Erb, R.M., Studart, A.R., 2012. Designer polymer-based microcapsules made using microfluidics. Langmuir 28, 144–152. https://pubs.acs.org/doi/abs/ 10.1021/la203088u. Costa, A.L.R., Gomes, A., Ushikubo, F.Y., Cunha, R.L., 2017. Gellan microgels produced in planar microfluidic devices. J. Food Eng. 209, 18–25. https://doi.org/10.1016/ j.jfoodeng.2017.04.007. Datta, S.S., Abbaspourrad, A., Amstad, E., Fan, J., Kim, S.H., Romanowsky, M., Shum, H.C., Sun, B., Utada, A.S., Windbergs, M., Zhou, S., Weitz, D.A., 2014. 25th Anniversary Article: Double wmulsion templated solid microcapsules: mechanics and controlled release. Adv. Mater. 26, 2205–2218. https:// onlinelibrary.wiley.com/doi/full/10.1002/adma.201305119. Franco, Z., Nguyen, Q.D., 2011. Flow properties of vegetable oil-diesel fuel blends. Fuel 90, 838–843. https://doi.org/10.1016/j.fuel.2010.09.044. Graham, S., Marina, P.F., Blencowe, A., 2019. Thermoresponsive polysaccharides and their thermoreversible physical hydrogel networks. Carbohydr. Polym. 207, 143–149. https://doi.org/10.1016/j.carbpol.2018.11.053. Hennequin, Y., Pannacci, N., Torres, C.P., Tetradis-Meris, G., Chapuliot, S., Bouchaud, E., Tabling, P., 2009. Synthesizing microcapsules with controlled geometrical and mechanical properties with microfluidic double emulsion technology. Langmuir 25, 7857–7861. https://pubs.acs.org/doi/10.1021/la9004449.

M. Michelon et al. / Chemical Engineering Science 211 (2020) 115314 Kim, S.H., Kim, J.W., Cho, J.C., Weitz, D.A., 2011. Double-emulsion drops with ultrathin shells for capsule templates. Lab Chip 11, 3162–3166. https://pubs.rsc.org/ en/content/articlepdf/2011/lc/c1lc20434c. Kim, S.H., Kim, J.W., Kim, D.H., Han, S.H., Weitz, D.A., 2013. Enhanced-throughput production of polymersomes using a parallelized capillary microfluidic device. Microfluid. Nanofluid. 14, 509–514. https://link.springer.com/article/10.1007/ s10404-012-1069-5. Kong, F., Zhang, X., Hai, M., 2014. Microfluidics fabrication of monodisperse biocompatible phospholipid vesicles for encapsulation and delivery of hydrophilic drug or active compound. Langmuir 30, 3905–3912. https://pubs. acs.org/doi/abs/10.1021/la404201m. Lapidot, N., Gans, O., Biagini, F., Sosonkin, L., Rottman, C., 2003. Advanced sunscreens: UV absorbers encapsulated in sol-gel glass microcapsules. J. SolGel Sci. Technol. 26, 67–72. https://link.springer.com/article/10.1023/ A:1020785217895. Lee, D., Weitz, D.A., 2008. Double emulsion-templated nanoparticle colloidosomes with selective permeability. Adv. Mater. 20, 3498–3503. https://onlinelibrary. wiley.com/doi/abs/10.1002/adma.200800918. Lee, D., Weitz, D.A., 2009. Nonspherical colloidosomes with multiple compartments from double emulsions. Small 5, 1932–1935. https://onlinelibrary. wiley.com/doi/full/10.1002/smll.200900357. Liu, L., Wu, F., Ju, X.J., Xie, R., Wang, W., Niu, C.H., Chu, L.Y., 2013. Preparation of monodisperse calcium alginate microcapsules via internal gelation in microfluidic-generated double emulsions. J. Coll. Interface Sci. 404, 85–90. https://doi.org/10.1016/j.jcis.2013.04.044. Michelon, M., Huang, Y., de la Torre, L.G., Weitz, D.A., Cunha, R.L., 2019. Single-step microfluidic production of W/O/W double emulsions as templates for bcarotene-loaded giant liposomes formation. Chem. Eng. J. 366, 27–32. https:// doi.org/10.1016/j.cej.2019.02.021. Nabavi, S.A., Vladisavljevic´, G.T., Gu, S., Ekanem, E.E., 2015. Double emulsion production in glass capillary microfluidic device: parametric investigation of droplet generation behavior. Chem. Eng. Sci. 130, 183–196. https://doi.org// 10.1016/j.ces.2015.03.004. Nabavi, S.A., Vladisavljevic´, G.T., Bandulasena, M.V., Arjmandi-Tash, O., Manovic´, V., 2017. Prediction and control of drop formation modes in microfluidic generation of double emulsions by single-step emulsification. J. Coll. Interface Sci. 505, 315–324. https://doi.org/10.1016/j.jcis.2017.05.115. Nabavi, S.A., Vladisavljevic´, G.T., Manovic´, V., 2017. Mechanisms and control of single-step microfluidic generation of multi-core double emulsion droplets. Chem. Eng. J. 322, 140–148. https://doi.org/10.1016/j.cej.2017.04.008. do Nascimento, D.F., Avendaño, J.A., Mehl, A., Moura, M.J.B., Carvalho, M.S., Duncanson, W.J., 2017. Flow of tunable elastic microcapsules through

11

constrictions, Sci. Rep., 7:11898 . Picone, C.S.F., Cunha, R.L., 2010. Interactions between milk proteins and gellan gum in acidified gels. Food Hydrocoll. 24, 502–511. https://doi.org/10.1016/ j.foodhyd.2009.12.007. Santos, T.P., Cunha, R.L., 2018. Role of process variables on the formation and in vitro digestion of gellan gels. Carbohydr. Polym. 192, 111–117. https://doi.org/ 10.1016/j.carbpol.2018.03.062. Silva, K.C.G., Cezarino, E.C., Michelon, M., Sato, A.C.K., 2018. Symbiotic microencapsulation to enhance Lactobacillus acidophilus survival. LWT – Food Sci. Technol. 89, 503–509. https://doi.org/10.1016/j.lwt.2017.11.026. Steegmans, M.L.J., Warmerdam, A., Schroen, K.G.P.H., Boom, R.M., 2009. Dynamic interfacial tension measurements with microfluidic Y-junctions. Langmuir 25, 9751–9758. https://pubs.acs.org/doi/10.1021/la901103r. Utada, A.S., Lorenceau, E., Link, D.R., Kaplan, P.D., Stone, H.A., Weitz, D.A., 2005. Monodisperse double emulsions generated from a microcapillary device, Sci., 308 (2005) 537–541. https://science.sciencemag.org/content/308/5721/537. Vilanova, N., Rodríguez-Abreu, C., Fernández-Nieves, A., Solans, C., 2013. Fabrication of novel silicone capsules with tunable mechanical properties by microfluidic techniques. ACS Appl. Mater. Interfaces 5, 5247–5252. https://pubs.acs.org/doi/ abs/10.1021/am4010896. Vilela, J.A.P., Cavallieri, A.L.F., Cunha, R.L., 2012. The influence of gelation rate on the physical properties/structure of salt-induced gels of soy protein isolate-gellan gum. Food Hydrocoll. 25, 1710–1718. https://doi.org/10.1016/j.foodhyd.2011.03.012. Vladisavljevic´, G.T., Shum, H.C., Weitz, D.A., 2012. Control over the shell thickness of core/shell drops in three-phase glass capillary devices. Prog. Coll. Polym. Sci. 139, 115–118. https://link.springer.com/chapter/10.1007%2F978-3-642-289743_20. Werner, J.G., Nawar, S., Solovev, A.A., Weitz, D.A., 2018. Hydrogel microcapsules with dynamic pH-responsive properties from methacrylic anhydride. Macromol. 51, 5798–5805. https://pubs.acs.org/doi/abs/10.1021/ acs.macromol.8b00843?journalCode=mamobx. Xu, S., Tabakovic´, A., Liu, X., Schlangen, E., 2018. Calcium alginate capsules encapsulating rejuvenator as healing system for asphalt mastic. Constr. Build. Mater. 169, 379–387. https://doi.org/10.1016/j.conbuildmat.2018.01.046. Yamamoto, F., Cunha, R.L., 2007. Acid gelation of gellan: effect of final pH and heat treatment conditions. Carbohydr. Polym. 68, 517–527. https://doi.org/10.1016/ j.carbpol.2006.11.009. Zhao, Y., Shum, H.C., Adams, L.L.A., Sun, B., Holtze, C., Gu, Z., Weitz, D.A., 2011. Enhanced encapsulation of actives in self-sealing microcapsules by precipitation in capsule shells. Langmuir 27, 13988–13991. https://pubs.acs. org/doi/10.1021/la2034774.