1  

A step forward towards the design of a continuous process to produce 1

hybrid liposome/protein microcapsules 2

3

Laura G. Gómez-Mascaraque1, Caroline Casagrande Sipoli2, Lucimara Gaziola de La 4

Torre2, Amparo López-Rubio1,* 5

6

1Food Safety and Preservation Department, IATA-CSIC, Avda. Agustin Escardino 7, 7

46980 Paterna, Valencia, Spain 8

2 School of Chemical Engineering, University of Campinas, UNICAMP, P.O. Box 9

6066, 13083-970, Campinas-SP, Brazil 10

11

*Corresponding author: Tel.: +34 963900022; fax: +34 963636301 12

E-mail addresses: lggm@iata.csic.es (L. G. Gómez-Mascaraque), 13

latorre@feq.unicamp.br (L. Gaziola de La Torre), carolinesipoli@gmail.com (C. 14

Casagrande Sipoli), amparo.lopez@iata.csic.es (A. López-Rubio) 15

       

2  

ABSTRACT 16

Microfluidics and electrospraying, two revolutionary technologies with industrial 17

potential for the microencapsulation of lipophilic bioactive ingredients, have been 18

combined to produce hybrid liposome/protein microencapsulation structures in a semi-19

continuous process, reducing the number of steps required for their manufacture. Three 20

different microfluidic mixing devices, one of them consisting of a simple straight 21

microchannel (cross junction design) and the other two exhibiting patterned 22

microchannels with different geometries (Tesla and ‘splitting and recombination’ 23

designs), were used to mix a liposome suspension with a whey protein concentrate 24

dispersion. The Tesla design showed the best mixing performance, as observed by 25

fluorescence microscopy, so it was selected to be assembled to an electrospraying 26

apparatus. The proposed in-line setup was successfully used to produce the micron-27

sized encapsulation structures, as observed by scanning electron microscopy. 28

29

KEYWORDS 30

Microfluidics; electrospraying; liposome; encapsulation; functional food 31

32

33

       

3  

1. Introduction 34

Liposomes can be used as delivery vehicles for the incorporation of lipophilic bioactive 35

compounds into food products, increasing their stability and bioavailability (Frenzel et 36

al., 2015). However, the high semi-permeability of their membranes and low physical 37

stability (Frenzel and Steffen-Heins, 2015) limit their practical application. 38

Microencapsulation of the liposomes within dry biopolymeric matrices, obtaining 39

convenient powdery ingredients, has been proposed as a strategy to overcome these 40

limitations (Gültekin-Özgüven et al., 2016; Tan et al., 2016; Van Den Hoven et al., 41

2012; Wang et al., 2015). 42

Specifically, the electrospraying technique, a drying technology based on the 43

electrohydrodynamic processing or atomization of polymeric fluids, has been very 44

recently used to produce dry liposome/protein microencapsulation structures, which 45

proved to successfully stabilize and improve the bioaccessibility of curcumin (Gómez-46

Mascaraque et al., 2017). The main advantage of electrospraying over other more 47

commonly used drying techniques such as spray-drying is its operation under mild 48

temperature conditions avoiding the thermal degradation of thermosensitive bioactive 49

ingredients (Gómez-Mascaraque and López-Rubio, 2016). 50

On the other hand, conventional technologies used for the manufacturing of liposomes 51

usually require numerous pre- and post-processing steps (Tien Sing Young and 52

Tabrizian, 2015). Microfluidics, a technology which allows manipulation of tiny 53

volumes of fluids inside micron-sized channels, has been proposed as an alternative for 54

the continuous production of liposomes by flow focusing (Balbino et al., 2013a). 55

The combination of microfluidics and electrospraying would allow the design of an 56

industrially attractive continuous process for the production of liposome/protein 57

       

4  

microencapsulation structures similar to those previously developed in batch (Gómez-58

Mascaraque et al., 2017). For this purpose, an intermediate mixing step would be 59

necessary in order to blend the liposomes with the polymer prior to electrospraying the 60

mixture. Due to the low flow rates used in both techniques, this mixing step should be 61

also accomplished by means of microfluidics. 62

63

Several designs for microfluidic mixing devices (‘micromixers’) have been proposed. 64

The most simple design is based on straight channels (Bothe et al., 2006). In these 65

devices, mixing occurs through a diffusion mechanism due to the low Reynolds 66

numbers (Re) obtained for the fluids, which flow in the laminar regime as a 67

consequence of the micro scale (Yang et al., 2015). However, diffusion-driven mixing 68

is time-consuming and inefficient, and thus passive micromixers based on the 69

modification of the flow channel geometry by adding obstacles in the flow path have 70

been also designed in order to enhance fluid mixing by increasing contact between the 71

two fluids. Several patterned microchannels with different geometries have already been 72

developed (Afzal and Kim, 2012; Bhagat and Papautsky, 2008; Hong et al., 2004; Tran-73

Minh et al., 2014). 74

75

In this work, we report on the study of three different micromixer designs based on the 76

cross flow junction, Tesla and SAR (‘splitting and recombination’) designs (Bothe et 77

al., 2006; Hong et al., 2004; Tran-Minh et al., 2014), to achieve an effective mixing of a 78

liposome suspension with a whey protein concentrate (WPC) dispersion. The selected 79

micromixer was then assembled to an electrospraying apparatus to produce 80

liposome/protein microencapsulation structures in a one-step process. 81

82

       

5  

2. Materials and Methods 83

2.1. Materials 84

A whey protein concentrate (WPC), under the commercial name of Lacprodan® DI-85

8090 was kindly donated by ARLA (ARLA Food Ingredients, Denmark). Pure 86

phosphatidylcholine (98% ± 4%) stabilized with 0.1% ascorbyl palmitate, was obtained 87

from Lipoid GmbH (Germany). Rhodamine B was obtained from Sigma-Aldrich 88

(Germany). Absolute ethanol (>99.5%) was purchased from Synth (Brasil). Deionized 89

water (>18 M cm-1 resistivity) was purified using Milli-Q® SP Reagent water system 90

plus from Millipore Corp. (USA). Sylgard 184 Silicone Elastomer Kit (Dow Corning, 91

Midland, MI, USA) was used as material precursor of PDMS layers for the microfluidic 92

devices, and NANOTM SU-8 photoresist formulation (MicroChem Corp., USA) was 93

used to prepare the masks. 94

95

2.2. Production of the microfluidic mixing devices (micromixers) 96

Three different designs for the micromixers were projected using AutoCAD (Autodesk). 97

The geometry of their channels is depicted in Figure 1. The polydimethylsiloxane 98

(PDMS)/glass microfluidic devices were then produced following the procedure 99

described in (Balbino et al., 2013b), which is based on the conventional UV 100

photolithographic and soft-lithography methods. The cross-section of the microchannels 101

was rectangular in all cases, with a depth of 50 µm. 102

1

111

   

 

04

Figur06 mixin07 07

re 1. Schemeng devices use

es and corresed in this wor

sponding micrk. The micro

 

crographs of ochannels are

f the three die 200 µm wid

ifferent desigde and 50 µm

gns of microfm deep.

 

6

fluidic

       

7  

2.3. Preparation of protein and liposome dispersions 107

WPC (25% w/v) was dispersed in milliQ water at room temperature under vigorous 108

magnetic stirring. Liposome dispersions (80 g/L) were prepared using the ethanol 109

injection method followed by ultrasonication as described previously (Gómez-110

Mascaraque et al., 2017). Both dispersions were filtered prior to their pumping into the 111

micromixers, to avoid clotting. The liposome dispersions were filtered using 0.45 µm, 112

nylon, Whatman® syringe filters, while the WPC dispersions were filtered using filter 113

paper under vacuum due to its high viscosity. 114

115

2.4. Mixing of protein and liposome dispersions using micromixers 116

The dispersions were introduced in glass syringes and were pumped through the 117

selected microfluidic mixer devices with digitally controlled syringe pumps model 118

KDS-100 (KDScientific, Massachusetts, USA). The protein dispersion was infused 119

through the lateral inlets of the micromixers, each stream flowing at 0.95 µL/min. The 120

liposomes were introduced through the central inlet at 0.6 µL/min. These rates were 121

selected by carrying out a mass balance taking into account the prerequisites of the 122

subsequent electrospraying process, which was optimized in a previous work (Gómez-123

Mascaraque et al., 2017) and required a total flow rate of 2.5 µL/min and a final WPC 124

concentration in the mixture of 20% (w/w). 125

126

2.5. Assessment of the quality of mixing 127

To assess whether the different microfluidic devices were effective in mixing both 128

suspensions, 0.1mM rhodamine B was added to the liposome stream as a fluorescent 129

       

8  

probe and the flow of the fluids through the microchannels was observed under an 130

inverted research microscope Eclipse Ti-U, (Nikon, Tokyo, Japan) equipped with a 131

digital imaging head which integrates an epifluorescence illuminator. A digital camera 132

head (Nikon 1.5MP DS-Qi1Mc, Japan) was attached to the microscope. Nis Elements 133

software (Nikon, Japan) was used for image capturing, and FIJI (Schindelin et al., 2012) 134

was used to process the images. The mixing index in the outlet of each micromixer was 135

calculated according to the method and equations described in (Shiroma et al., 2017). 136

137

2.6. In-line production of hybrid microencapsulation structures 138

The outlet of the micromixer was connected to an electrospraying apparatus, equipped 139

with a variable high-voltage (0-30 kV) power supply and assembled in-house. 140

Specifically, the blend dispersion of protein and liposomes coming out from the 141

micromixer was pumped through a polytetrafluoroethylene (PTFE) thin tube connected 142

to a stainless steel needle (0.84 mm inner diameter). Figure 2 shows a schematic 143

representation of the set-up. The dispersions were introduced in 5 mL plastic syringes 144

and were pumped through the selected micromixer with digitally controlled syringe 145

pumps model KDS-100 (KDScientific, Massachusetts, USA) at the same flow-rates 146

used in Section 2.4. A grounded collector was placed facing the needle in a horizontal 147

configuration, at a distance of 10 cm from its tip, and a voltage of 10 kV was applied. 148

1

11

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

   

 

50

51 52

2.7. M53

Scan57

S-4858

Samp59

exam60

58

3. R59

3.1. M60

In or68

their 69

as m70

cross71

200672

comm73

Masc74

with 75

Figur

Morpholog

nning electro

00) at an

ples were

mination.

Results and

Microfluid

rder to achie

electrospra

micromixers

s junction d

6). Due to

monly used

caraque and

a second f

re 2. Basic sch

gical charac

on microsco

accelerating

sputter-coat

d discussion

dic mixing d

eve an effec

aying, the p

was studie

design cons

the relative

d to produ

d López-Ru

fluid in a sim

heme of the p

cterization

opy (SEM)

g voltage o

ted with a

n

devices

ctive mixin

erformance

d. Figure 1

isting of 3

ely high vi

uce electros

ubio, 2016;

mple cross

 

proposed mic

of the hyb

was condu

of 10 kV

gold-pallad

ng of the pro

e of three di

a) shows th

inlets and a

scosity of

sprayed mi

; Gómez-M

junction de

cromixer-elec

brid microe

ucted on a H

and a work

dium mixtu

otein and lip

ifferent desi

he simplest

a single str

the polyme

icroencapsu

Mascaraque

esign may r

ctrospraying

ncapsulatio

Hitachi micr

king distan

ure under v

posome disp

igns of micr

microfluid

aight chann

er solutions

ulation stru

et al., 201

require long

setup.

on structur

roscope (H

nce of ~10

vacuum pri

spersions pr

rofluidic de

dic device w

nel (Bothe e

s or suspen

uctures (Gó

6), their m

g microchan

 

9

res

itachi

mm.

ior to

rior to

evices

with a

et al.,

nsions

ómez-

mixing

nnels,

       

10  

as vortex formation was not expected given the low flow rates, and thus low Reynolds 168

numbers, required for subsequent electrospraying (Orsi et al., 2013). Therefore, two 169

other geometries were also explored. Figures 1b) and 1c) show different geometries 170

based on the Tesla (Hong et al., 2004) and SAR (splitting and recombination) (Tran-171

Minh et al., 2014) designs, respectively. These geometries were aimed at creating 172

swirling sections to accelerate mixing. Figure 1 also shows micrographs of the produced 173

micromixers, where the detailed geometry can be observed. 174

175

3.2. Mixing performance of the micromixers 176

The protein and liposome dispersions were prepared in batch as described in a previous 177

work, where their characterization was reported (Gómez-Mascaraque et al., 2017). 178

0.1mM rhodamine B was added to the liposome dispersion as a fluorescent probe. The 179

mixture of both fluids when infused into the microfluidic device was observed under the 180

microscope. Figure 3 shows fluorescence micrographs of different sections of the 181

microfluidic devices filled with the flowing dispersions, together with a reconstruction 182

of the whole central channel obtained from different pictures and assembled using the 183

software ImageJ. 184

1

111

   

 

86

Fig89 desig90

91

gure 3. Fluid gns. Relevant

mixing usingt sections are

g microfluidicshown in detby the lipos

 

c devices withtail. The arrosomes in the S

h T (left), Tesow indicates tSAR design.

sla (central), the preferent

and SAR (rigtial pathway t

 

11

ght) taken

       

12  

From the images in Figure 3, the software Fiji was used to calculate the fluorescence 189

profile (i.e. intensity of the pixels) along the main channel of the micromixers. Figure 4 190

shows, for each micromixer design, the fluorescence profile of different sections of the 191

microchannel, where the baseline corresponded to black pixels. The profiles after the 192

confluence of the three inlets, which are highlighted in black in Figure 4, showed a 193

narrow peak corresponding to the rhodamine-containing liposomes stream. At that 194

point, the protein and liposomes dispersions could be clearly distinguished, both 195

visually in Figure 3 and quantitatively in Figure 4. The following sections along the 196

channels, in all cases, exhibited a gradual widening of the fluorescence signal, due to 197

the progressive mixing of both streams throughout the micromixer, up to the outlet 198

(section 19000 µm), which is highlighted in red in Figure 4. However, the shape of 199

these profiles was different for each micromixer. In the SAR design, a preferential 200

pathway was taken by the liposomes dispersion (cf. arrow in Figure 3), so the ‘splitting 201

and recombination’ concept did not apply. Consequently, the fluorescence increased 202

preferentially in one of the ends of the microchannel. Conversely, the fluorescence 203

distribution in the cross junction and Tesla designs was more symmetric. In the outlet, 204

the fluorescence profile was flatter for the cross junction and Tesla designs than for the 205

SAR design, due to the aforementioned preferential pathway in the latter, indicating that 206

the former achieved a better mixing of both fluids. 207

2

2

2

2

2

2

2

2

2

2

2

2

   

 

209

Fig210

211

To a217

(outl218

Shiro219

junct220

cross221

and i222

218

Altho220

mixin221

gure 4. Fluore

assess the qu

let) was ca

oma et al.

tion, Tesla a

s junction d

its performa

ough the el

ng human b

escence profi

uality of m

alculated fo

(2017). Th

and SAR d

design was m

ance was sim

llipse-like g

blood (Tran

les along the

mixing quant

or each mic

he mixing i

esigns, resp

more effecti

milar to that

geometry o

n-Minh et a

 

length of the

titatively, th

cromixer a

indexes we

pectively. S

tive for mix

t of the Tes

of the SAR

al., 2014), t

e main chann

he mixing i

ccording to

ere 0.77, 0.

urprisingly,

xing both str

la design.

R design had

the flow ra

el of the micr

index at sec

o the meth

.74 and 0.5

, the simple

reams than

d proved to

ates used in

rofluidic devi

ction 19000

hod describ

58 for the

e diffusion i

the SAR de

o be efficie

n that work

 

13

ices.

0 µm,

ed in

cross

in the

esign,

ent in

were

       

14  

considerably higher than the ones required for electrospraying, and the viscosity 220

substantially lower, so the Reynolds number was expected to be significantly higher. 221

The Reynolds number (Re) in our microfluidic devices was calculated according to Eq. 222

(1-3) (assuming channels with rectangular shape), where ρ is the density of the fluids, µ 223

is their viscosity, W is the the width of the microchannels, L is their depth, Dh is their 224

hydraulic diameter, A is the area of their section, Q is the flow rate and V is the flow 225

velocity. And, indeed, it was estimated to be 0.03 when both dispersions were mixed 226

(cf. Table 1). Increasing the flow rate 10-fold in the SAR design avoided the 227

aforementioned preferential pathway taken by the liposomes stream (cf. Figure S1 of 228

the Supplementary Material). However, the flow rates in this work were fixed as a 229

requirement for the subsequent electrospraying process. 230

231

Eq. (1) 232

2 Eq. (2) 233

Eq. (3) 234

Table 1. Calculation of Reynolds number (Re) in the inlets and outlet of the micromixers, according 235 to Eq. (1-3). For the outlet, the viscosity and density of a mixture of the polymer and the liposomes 236

was assumed. 237

    Inlets 1,3  Inlet 2  Outlet 

Fluid Density (ρ, kg/m3)  1023  980  1013 

Viscosity (µ, Pa∙s)  0.023  0.00163  0.011 

Geometry of the 

channel 

Depth (L, m)  5.0∙10‐5  5.0∙10‐5  5.0∙10‐5 

Width (W, m)  2.0∙10‐4  2.0∙10‐4  2.0∙10‐4 

Hydraulic diameter (Dh, m)  8.0∙10‐5  8.0∙10‐5  8.0∙10‐5 

Area of section (A, m2)  1.0∙10‐8  1.0∙10‐8  1.0∙10‐8 

Flow Flow rate (Q, m3/s)  1.6∙10‐11  1.0∙10‐11  4.2∙10‐11 

Flow velocity (V, m/s)  1.6∙10‐3  1.0∙10‐3  4.2∙10‐3 

  Re  0.006  0.05  0.03 

       

15  

238

Low flow rates involve low Peclet numbers, which favour mixing by diffusion 239

(Ismagilov et al., 2000). This may be the reason why, for this particular system, the 240

cross junction design provided a better mixing. However, the Tesla micromixer 241

achieved an equally good mixing index. Previous studies showed similar performances 242

for Tesla and cross junction designs at low flow rates (low Peclet and Reynolds 243

numbers, since diffusion is favoured), although the performance of the Tesla design was 244

considerably better at higher flow rates (Hong et al., 2004). Since the flow rates used in 245

this work were extremely low, and mixing by diffusion was then favoured, both 246

micromixer designs exhibited similar mixing indexes. However, given the greater 247

potential of the Tesla geometry in terms of flow-rate flexibility, the latter was selected 248

to assemble the in-line mixing and electrospraying set-up. 249

250

3.3. In-line production of hybrid microencapsulation structures 251

An in-line mixing and electrospraying set-up was assembled using the Tesla 252

micromixer. For this purpose, a PTFE tube was used to connect the outlet of the 253

microfluidic device to the stainless steel needle, combining the microfluidic and 254

electrohydrodynamic technologies in a quasi-continuous process. A stereoscope was 255

used to ensure the correct performance of the micromixer and the absence of clogging. 256

Figure 5 a) illustrates the aforementioned set-up, which was used to produce the novel 257

hybrid encapsulation structures recently proposed in a previous work (Gómez-258

Mascaraque et al., 2017). Figure 5 b) shows the morphology of the obtained 259

microcapsules, where populations in the micron and sub-micron range can be observed. 260

This morphology was very similar to those previously obtained (Gómez-Mascaraque et 261

2

2

2

222

2

2

2

2

2

2

2

   

 

al., 2264

conti265

265

F268 obta269

270

269

4. C270

In th275

produ276

manu277

were278

Thes279

2017), thus

inuous prod

Figure 5. a) Inained hybrid m

Conclusions

his work, t

uce hybrid

ufacturing p

e used to mi

se designs w

s confirmin

duction of hy

n-line micrommicroencapsu

s

the microflu

liposome/p

process. For

ix a stream o

were based

ng the pote

ybrid encap

mixer-electrosulation struct

bar

uidics and

protein mic

r this purpos

of liposome

d on the T,

 

ential of th

psulation str

spraying assetures togethe

r indicates 10

electrospra

croencapsul

se, three dif

es with a dis

, Tesla and

he develope

ructures.

embled setup.er with their p

µm.

aying techn

lation struc

fferent desig

spersion of

d SAR (‘sp

ed methodo

. b) SEM micparticle size d

nologies we

tures in a

gns for micr

a whey pro

litting and

ology for

crograph of tdistribution.

ere combin

semi-contin

rofluidic de

otein concen

recombina

 

16

semi-

he Scale

ed to

nuous

evices

ntrate.

tion’)

       

17  

designs previously proposed in the literature. Due to the high viscosity of the protein 275

dispersion and the low flow rates required for the subsequent electrospraying process, 276

the SAR design was not effective in mixing both fluids, as a preferential pathway was 277

taken by the liposomes instead of splitting and recombining. Conversely, Tesla and 278

cross junction designs exhibited adequate mixing performances. An assembled 279

micromixer-electrospraying setup was built using the Tesla design, and this 280

arrangement was successfully used to produce the dry liposome/protein microcapsules, 281

which exhibited a similar morphology as those produced in batch in a previous work. 282

The next step towards the design of a fully continuous manufacturing process would 283

involve the incorporation of additional microchannels in the microfluidic devices to 284

continuously produce the liposomes by flow focusing before their mixture with the 285

protein. 286

287

Acknowledgements 288

Laura G. Gómez-Mascaraque is recipient of a predoctoral contract from the Spanish 289

Ministry of Economy, Industry and Competitiveness (MINECO), Call 2013. The 290

authors would like to thank the Spanish MINECO projects AGL2012-30647 and 291

AGL2015-63855-C2-1-R for financial support. 292

293

References 294

Afzal,  A.,  Kim,  K.Y.,  (2012).  Passive  split  and  recombination  micromixer  with  convergent‐295 divergent walls. Chemical Engineering Journal 203, 182‐192. 296 Balbino, T.A., Aoki, N.T., Gasperini, A.A.M., Oliveira, C.L.P., Azzoni, A.R., Cavalcanti, L.P., de  la 297 Torre,  L.G.,  (2013a).  Continuous  flow  production  of  cationic  liposomes  at  high  lipid 298 concentration  in  microfluidic  devices  for  gene  delivery  applications.  Chemical  Engineering 299 Journal 226(0), 423‐433. 300

       

18  

Balbino,  T.A.,  Azzoni,  A.R.,  de  la  Torre,  L.G.,  (2013b).  Microfluidic  devices  for  continuous 301 production  of  pDNA/cationic  liposome  complexes  for  gene  delivery  and  vaccine  therapy. 302 Colloids and Surfaces B: Biointerfaces 111(0), 203‐210. 303 Bhagat,  A.A.S.,  Papautsky,  I.,  (2008).  Enhancing  particle  dispersion  in  a  passive  planar 304 micromixer  using  rectangular  obstacles.  Journal  of  micromechanics  and  microengineering 305 18(8), 085005. 306 Bothe,  D.,  Stemich,  C.,  Warnecke,  H.‐J.,  (2006).  Fluid  mixing  in  a  T‐shaped  micro‐mixer. 307 Chemical Engineering Science 61(9), 2950‐2958. 308 Frenzel, M., Krolak, E., Wagner, A.E., Steffen‐Heins, A.,  (2015). Physicochemical properties of 309 WPI coated liposomes serving as stable transporters in a real food matrix. LWT ‐ Food Science 310 and Technology 63(1), 527‐534. 311 Frenzel,  M.,  Steffen‐Heins,  A.,  (2015). Whey  protein  coating  increases  bilayer  rigidity  and 312 stability of liposomes in food‐like matrices. Food Chemistry 173, 1090‐1099. 313 Gómez‐Mascaraque,  L.G.,  Casagrande  Sipoli,  C.,  de  La  Torre,  L.G.,  López‐Rubio,  A.,  (2017). 314 Microencapsulation  structures  based  on  protein‐coated  liposomes  obtained  through 315 electrospraying for the stabilization and improved bioaccessibility of curcumin. Food Chemistry 316 233, 343‐350. 317 Gómez‐Mascaraque,  L.G.,  López‐Rubio,  A.,  (2016).  Protein‐based  emulsion  electrosprayed 318 micro‐  and  submicroparticles  for  the  encapsulation  and  stabilization  of  thermosensitive 319 hydrophobic bioactives. Journal of Colloid and Interface Science 465, 259‐270. 320 Gómez‐Mascaraque, L.G., Sanchez, G., López‐Rubio, A., (2016). Impact of molecular weight on 321 the  formation  of  electrosprayed  chitosan  microcapsules  as  delivery  vehicles  for  bioactive 322 compounds. Carbohydrate Polymers 150, 121‐130. 323 Gültekin‐Özgüven, M., Karadaʇ, A., Duman, Ş., Özkal, B., Özçelik, B., (2016). Fortification of dark 324 chocolate  with  spray  dried  black  mulberry  (Morus  nigra)  waste  extract  encapsulated  in 325 chitosan‐coated liposomes and bioaccessability studies. Food Chemistry 201, 205‐212. 326 Hong,  C.‐C.,  Choi,  J.‐W., Ahn,  C.H.,  (2004). A  novel  in‐plane  passive microfluidic mixer with 327 modified Tesla structures. Lab on a Chip 4(2), 109‐113. 328 Ismagilov, R.F., Stroock, A.D., Kenis, P.J., Whitesides, G., Stone, H.A., (2000). Experimental and 329 theoretical  scaling  laws  for  transverse  diffusive  broadening  in  two‐phase  laminar  flows  in 330 microchannels. Applied Physics Letters 76(17), 2376‐2378. 331 Orsi, G., Roudgar, M., Brunazzi, E., Galletti, C., Mauri, R.,  (2013). Water–ethanol mixing  in T‐332 shaped microdevices. Chemical Engineering Science 95, 174‐183. 333 Schindelin, J., Arganda‐Carreras,  I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., 334 Rueden,  C.,  Saalfeld,  S.,  Schmid,  B.,  Tinevez,  J.‐Y., White,  D.J.,  Hartenstein,  V.,  Eliceiri,  K., 335 Tomancak, P., Cardona, A., (2012). Fiji: an open‐source platform for biological‐image analysis. 336 Nat Meth 9(7), 676‐682. 337 Shiroma,  L.S., Oliveira, A.F.,  Lobo‐Júnior,  E.O.,  Coltro, W.K., Gobbi, A.L.,  Lucimara, G.,  Lima, 338 R.S.,  (2017). High adhesion strength and hybrid  irreversible/reversible  full‐PDMS microfluidic 339 chips. Analytica Chimica Acta 951, 116‐123. 340 Tan,  C.,  Feng,  B.,  Zhang,  X.,  Xia,  W.,  Xia,  S.,  (2016).  Biopolymer‐coated  liposomes  by 341 electrostatic  adsorption of  chitosan  (chitosomes)  as novel delivery  systems  for  carotenoids. 342 Food Hydrocolloids 52, 774‐784. 343 Tien  Sing  Young,  R.V.,  Tabrizian,  M.,  (2015).  Rapid,  one‐step  fabrication  and  loading  of 344 nanoscale  1,  2‐distearoyl‐sn‐glycero‐3‐phosphocholine  liposomes  in  a  simple,  double  flow‐345 focusing microfluidic device. Biomicrofluidics 9(4), 046501. 346 Tran‐Minh, N., Dong, T., Karlsen, F., (2014). An efficient passive planar micromixer with ellipse‐347 like micropillars  for continuous mixing of human blood. Computer methods and programs  in 348 biomedicine 117(1), 20‐29. 349 Van Den Hoven, J.M., Metselaar, J.M., Storm, G., Beijnen, J.H., Nuijen, B., (2012). Cyclodextrin 350 as  membrane  protectant  in  spray‐drying  and  freeze‐drying  of  PEGylated  liposomes. 351 International Journal of Pharmaceutics 438(1‐2), 209‐216. 352

       

19  

Wang, L., Hu, X., Shen, B., Xie, Y., Shen, C., Lu, Y., Qi,  J., Yuan, H., Wu, W.,  (2015). Enhanced 353 stability  of  liposomes  against  solidification  stress  during  freeze‐drying  and  spray‐drying  by 354 coating with calcium alginate. Journal of Drug Delivery Science and Technology 30, 163‐170. 355 Yang, A.‐S., Chuang, F.‐C., Chen, C.‐K., Lee, M.‐H., Chen, S.‐W., Su, T.‐L., Yang, Y.‐C., (2015). A 356 high‐performance micromixer  using  three‐dimensional  Tesla  structures  for  bio‐applications. 357 Chemical Engineering Journal 263, 444‐451. 358

359

360

3

3

333

3

3

3

   

 

Su362

363

Figur366 fol367

368

367

368

369

upple

re S1. Mixturld higher tha

ment

re of the liposn the one req

liposome d

taryM

somes and Wquired for eledispersion wh

 

Mater

WPC dispersioectrosprayinghen the flow r

rial

ons in the SARg. No preferenrate is increas

R micromixential pathwaysed 10-fold.

er at a flow ray is taken by

 

20

ate 10-the