Ultrasound assisted extraction and nanofiltration of phenolic compounds from artichoke solid wastes

Accepted Manuscript Ultrasound assisted extraction and nanofiltration of phenolic compounds from artichoke solid wastes Renata S. Rabelo, Mariana T.C. Machado, Julian Martínez, Miriam D. Hubinger PII:

S0260-8774(16)30018-8

DOI:

10.1016/j.jfoodeng.2016.01.018

Reference:

JFOE 8456

To appear in:

Journal of Food Engineering

Received Date: 9 September 2015 Revised Date:

15 December 2015

Accepted Date: 19 January 2016

Please cite this article as: Rabelo, R.S., Machado, M.T.C., Martínez, J., Hubinger, M.D., Ultrasound assisted extraction and nanofiltration of phenolic compounds from artichoke solid wastes, Journal of Food Engineering (2016), doi: 10.1016/j.jfoodeng.2016.01.018. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

ULTRASOUND ASSISTED EXTRACTION AND NANOFILTRATION OF PHENOLIC

2

COMPOUNDS FROM ARTICHOKE SOLID WASTES

3 4

Renata S. Rabelo*, Mariana T.C. Machado, Julian Martínez and Miriam D. Hubinger

RI PT

5 6

Department of Food Engineering, School of Food Engineering, University of Campinas

7

(UNICAMP), 80, Monteiro Lobato Street, P.O. Box 6121, ZIP 13083-862, Campinas/SP, Brazil.

SC

8 9

M AN U

10 11 12 13 14

TE D

15 16 17

EP

18 19

AC C

20 21 22 23 24

*

25

Telephone number: +55(19)3521-4036

26

Fax number: +55(19)3521-4027

27

E-mail address: [email protected]

Corresponding author.

ACCEPTED MANUSCRIPT ABSTRACT

29

Artichoke wastes from the canning industry are rich in phenolic compounds, which can be used as

30

food additives. This his study aims to evaluate the potential of sequential process based on the use

31

of ultrasound extraction and membrane technology for phenolic recovery. In the extraction step,

32

solvent composition and ultrasound power were evaluated to understand their impact on phenolic

33

content and antioxidant capacity. The highest yields were observed for extracts with higher ethanol

34

content (50 and 75%). Therefore, these extracts were selected for concentration by nanofiltration in

35

a tangential module, using differents membranes (NF270, DK and DL). The flux decrease and the

36

phenolic retention were evaluated. Hermia’s models and membrane surface characterization were

37

used to investigate the fouling. The highest flux was observed for extracts with 50% of ethanol, and

38

a retention higher than 95% of chlorogenic acid was observed in this process. DK membrane

39

showed to be less susceptible to fouling, although the cake formation occurred in all evaluated

40

membranes. The most suitable process condition to obtain the highest phenolic yields was

41

extraction with 50% ethanol and an ultrasound power of 240 W and extracts nanofiltration using

42

DK membrane.

43

Keywords: Cynara scolymus, phenolic compounds, ultrasound, nanofiltration, fouling mechanism.

44

1. INTRODUCTION

TE D

M AN U

SC

RI PT

28

The artichoke (Cynara scolymus L.) is a perennial herb belonging to the family Asteraceae.

46

This plant is native from subtropical regions and its inflorescences are consumed widely in

47

Mediterranean countries (PANDINO et al., 2012; PISTÓN et al., 2014).

AC C

48

EP

45

Besides being a good source of inulin, fiber and minerals, artichoke is a rich source of

49

phenolic compounds (RUIZ-CANO et al., 2014). These compounds have been associated with

50

scavenging capacities of artichoke extracts against ROS (reactive oxygen species) and RNS

51

(reactive nitrogen species) (PISTÓN et al., 2014), and anti-obesity effects (CHO et al., 2010), that

52

are some of the reasons for the high popularity of artichoke products around the world. A recent

53

paper showed that extracts obtained from artichoke waste have a remarkable delaying effect against

54

canola oil oxidation (CLAUS et al., 2015).

2

ACCEPTED MANUSCRIPT The artichoke processing main residues are the inner and outer bracts. The phenolic

56

composition of this waste is similar to the edible parts of the plant (FRATIANNI et al., 2007;

57

LOMBARDO et al., 2012; PANDINO et al., 2011; SIHEM et al., 2015), and according to Zuorro

58

et al. (2015) it is higher than that those found on grape pomace (LOULI et al., 2004), carrot peels

59

(CHANTARO et al., 2008) and spent coffee grounds (RANIC et al., 2014).

RI PT

55

The canning industry discards around 70% of the harvested inflorescence (LÓPEZ-

61

MOLINA et al., 2005). The world production of this vegetable grows an average of 6% annually,

62

and in 2012 was around 1.6 million tonnes (FAOSTAT, 2012); the amount of waste currently

63

generated exceeds one million tons. Therefore, it is important to search for alternatives to recover

64

phenolic compounds from artichoke wastes.

M AN U

SC

60

65

Among the extraction processes, the one assisted by ultrasound emerged as being the

66

simplest and the most inexpensive technique; it allows the use of various solvents, low

67

temperatures, does not have restrictions on the polarity of the compound of interest nor to the

68

moisture of the matrix; it

69

(GHITESCU et al., 2015; RASTOGI, 2011). The efficiency of the ultrasound, relative to

70

conventional extraction methods, is attributed to the cavitational effect which facilitates the release

71

of extractable compounds and enhances the mass transport by diffusion or by disrupting the plant

72

cell walls (CHEMAT et al., 2011; ESCLAPEZ et al., 2011; GAETE-GARRETÓN et al., 2011;

73

LUQUE DE CASTRO et al., 2011). But like most conventional methods of extraction, one of the

74

disadvantages of this method is that does not obtain solvent-free extracts, and a concentration step

75

is required after extraction.

TE D

EP

AC C

76

has also good reproducibility and a high potential for scale-up

To concentrate phenolics, the use of nanofiltration membranes shows higher fluxes than the

77

reverse osmosis and higher retentions than ultrafiltration (CONIDI et al., 2012), since artichoke

78

phenolic acids have a molar mass between 300 and 600 g/mol (ABU-REIDAH et al., 2013). This

79

technology allows concentration at mild temperatures, not involving a phase change, preserves the

80

biological activity of the compounds and enables the solvent recycling by capturing the permeate,

3

ACCEPTED MANUSCRIPT 81

which can be reused in further extractions enabling a sustainable process of extraction and

82

concentration. The use of different fractions of GRAS (Generally Regarded as Safe) solvents such as

84

water and ethanol in the extraction and concentration processes allow the application of the final

85

product in foods, drugs and cosmetics, contributes to selective extraction of bioactive compounds

86

(GHITESCU et al., 2015) and, due to the membrane-solvent interactions, significantly change the

87

permeation features of the membranes (FIRMAN et al., 2013; LABANDA et al., 2013). Thus, the

88

importance to investigate different ratio of these solvents in sequential processes.

SC

RI PT

83

The concentration of extracts obtained from different ratios of solvent can lead to different

90

interactions between the extract compounds. In hydroalcoholic systems these interactions have not

91

yet been deeply analysed, but they can influence membrane permeation. The flux decrease caused

92

by concentration polarization and membrane fouling is a major limitation for the industrial

93

application of membrane technology. In previous investigations (GRENIER et al., 2008; HWANG

94

et al., 2007; NG et al., 2014), a number of mathematical models were utilized to explain fouling

95

mechanisms. When particle size present in extract is smaller than or comparable to the membrane

96

pore size, the membrane blocking model is commonly used to explain how and when the particle

97

blocking occurs (HWANG AND LIAO, 2012).

98

mechanisms (complete blocking, intermediate blocking, standard blocking, and cake formation)

99

were proposed by Hermia (1982) to characterize the membrane fouling. These models can be a

100

useful tool for selecting membranes less susceptible to fouling, especially when it comes to the

101

filtration of vegetable hydroalcoholic extracts, due to the sample complexity.

TE D

EP

Based on this model, four different fouling

AC C

102

M AN U

89

The aim of this work was to (1) evaluate the effect of solvent composition and ultrasound

103

power in the extraction of phenolic compounds and (2) to concentrate the best extracts by three

104

different nanofiltration membranes. Flux decrease, coefficient retention and membrane

105

characterization were carried out. At last, we select the most suitable sequential system of

106

extraction and concentration of phenolic compounds from artichoke solid waste.

4

ACCEPTED MANUSCRIPT 107 108

2. METHODS 2.1. Extraction process Artichoke wastes (external and internal bracts) were donated by Bonsucesso factory (São

110

Roque/SP, Brazil). This material was stored in a freezing chamber at -18 °C and thawed according

111

to the quantity required for each extraction. All trials were carried out with the same lot of raw

112

material.

RI PT

109

Initially, the wastes were crushed and placed in a jacketed reactor (6.5 cm of internal

114

diameter and maximum volume of 250 mL). For extraction, an ultrasonic cell disruptor at 20 kHz,

115

equipped with a titanium alloy 13 mm diameter flat tip probe (UNIQUE, São Paulo, Brazil) was

116

fitted into the reactors. The w/v ratio was fixed at 1:10 (g/mL). The ultrasonic power evaluated was

117

0, 240, 480 and 720 W. The solvent compositions studied were 0, 25, 50 and 75% ethanol in water

118

(v/v). After 60 minutes of extraction, the extracts were filtered with filter paper, and characterized.

119

A total of 16 experiments were carried out, each done in duplicate. The experimental conditions

120

that showed the best yields of phenolic compounds were selected, and kinetic studies (5, 10, 20, 30,

121

40 50 and 60 min) were performed in order to reduce extraction time. Finally, a new extraction

122

time was fixed, and the experimental conditions selected were used as feed in the extract

123

concentration step. All extractions were performed in batches of 100 mL at 25 ± 1 °C.

M AN U

TE D

2.2. Concentration process

EP

124

SC

113

The tests were performed in a tangential filtration system (INVICT, MENTEST, Brazil)

126

using flat sheet membranes (NF270, DK and DL), presented in Table 1. This system had 2 L

127

capacity, an effective permeation area of 7.7x10-3 m2, and the maximum liquid recirculation flow

128

rate was 60 L/h.

129

AC C

125

A new membrane was used in each test to assure the same initial conditions. Conditioning

130

of the membranes consisted of their immersion during 12 hours in solutions with the same

131

alcoholic concentrations of the extracts. During concentration, the system pressure was maintained

132

at 20 bar and temperature was 25 °C. The initial feed volume was 900 mL and the concentration

133

process was carried out to a volume reduction factor of 2.5.

5

ACCEPTED MANUSCRIPT 134

2.2.1. Evaluated parameters The volumetric flux of permeate (Jv) was calculated by Equation 1, and the average

136

permeate flux (Js) was obtained from the straight line inclination described by the function
 = f

137

( × ), according to Tylkowski et al. (2010).

Jv =

RI PT

135

mp

ρ .t . A p

[1]

Where Jv is the volumetric flux of the permeate (L/m²h); mp is the permeate collected mass (g); ρ is

139

the density (g/L), t is the time (h), and Ap is the membrane permeation area (m²).

140

The flux decrease (Df) was calculated according to Equation 2.

142

(J i − J s )

M AN U

Df = 141

Ji

The retention coefficient (R) was calculated by Equation 3:

146 147

[3]

TE D

CP CA

Where Cp and CA are the concentration of solute in the permeate and in feed, respectively. The volumetric concentration factor (VCF) was determined by Equation 4, where Vi and Vf are the initial and the final feed volume (L), respectively.

EP

145

[2]


 =







[4]

2.2.2. Fouling mechanisms

AC C

144

.100

Where Ji is the initial and Js is the average permeate flux (L/m²h).

R(%) = 1 − 143

SC

138

The models proposed by Hermia (1982) were used to identify the predominant fouling

148

mechanism in the evaluated processes; these models incorporate four mechanisms of fouling (cake

149

formation, partial and complete blocking of the pores, and the absorption of solute particles in the

150

inner walls of the pores), that are described in Table 2 according to NG et al. (2014).

151

2.2.3. Surface membranes characterization

152

All the membranes were maintained for 24 hours in a desiccator before characterization.

153

Unconditioned (or virgin membranes), conditioned and fouled membranes were characterized

154

according to the purpose of the analysis. 6

ACCEPTED MANUSCRIPT 155

a) Atomic force microscopy (AFM) The analysis were performed according to Carvalho et al. (2011). The membranes were

157

analysed in an atomic force microscopy (Park Scientific Instruments, Alto Probe CP, Korea).

158

Images with a length of 1.0, 3.0, 5.0 and 10.0 µm in the same sample area, and three different

159

regions of each membrane were analysed. Quantitative analysis of surface roughness was

160

performed using the software Gwyddion 2.39 (Czech Metrology Institute, Czech Republic).

161

b) Infrared Fourier transformed spectroscopy

RI PT

156

The attenuated total reflectance (ATR) was used to investigate the functional groups and

163

molecular structures on the membranes surfaces, before and after the filtration process, in an

164

infrared spectrometer (JASCO Model FT/IR-6100, Japan). For data acquisition, a zinc selenide

165

crystal (ZnSe) was used, and the readings were performed with 4 cm-1 resolution.

M AN U

166

SC

162

c) Contact angle

The sessile drop method was applied on static contact mode for determining the contact

168

angle (θ) in a tensiometer (Teclis, Tracker, France). The initial drop volume of water deposited on

169

each sample was 5 µL, and all tests were performed at 25 °C and at 53% of relative moisture. The

170

results were obtained after 10 seconds of contact between the droplet and the membrane surface,

171

and the results were presented as average ± standard deviation from five samples of each

172

membrane.

EP

TE D

167

2.3. Characterization of the extracts

174

a) Total phenolic content (TPC)

175 176 177

AC C

173

The TPC was determined according to Singleton et al. (1999). A standard curve with gallic

acid was plotted to quantify the concentration. Assays were performed in triplicate. b) Antioxidant capacity by DPPH (2,2-diphenyl-1-picryl-hidrazil)

178

The antioxidant capacity was determined by inhibition percentage the DPPH using the

179

method described by Choi et al. (2002). To determine the reading time, a previous evaluation was

180

performed, and a reaction time of 60 min was found. The result was expressed as mg Trolox/g of

181

dry matter, and all assays were performed in triplicate.

7

ACCEPTED MANUSCRIPT 182

c) Antioxidant capacity by FRAP (Ferric reducing antioxidant power)

183

The antioxidant capacity was evaluated using FRAP by the method described by Vieira et

184

al. (2013). For quantification, a Trolox standard curve was plotted. All assays were performed in

185

triplicate. d) Chlorogenic acid content

RI PT

186

The extracts were analysed according to Chisté and Mercadante (2012). The analysis was

188

performed on a Dionex Chromatography (LC Standard Ultimate 3000, USA), equipped with a

189

diode array detector (Dionex, DAD–3000, USA). The wavelength of 324 nm was monitored, and

190

the separation was performed on a reverse phase Poroshel 120 EC-C18 column (100 mm x 4.6 mm

191

and 2.7 µm) at 29 °C. All analysis were made in duplicate. e) Particle size distribution

M AN U

192

SC

187

The size and particle size distribution of the feed and concentrate were determined by laser

194

diffraction equipment Mastersizer 2000 (Malvern Instruments, UK), and the diameter was

195

determined based on the average diameter of a sphere of the same volume (Brouckere Diameter –

196

D4.3). In the permeate, the particle size distribution was determined in Zetasizer Nano-ZS (Malvern

197

Instruments, USA).

198

f)

TE D

193

Proximal composition and physico-chemical properties The proximal composition and some physico-chemical properties of the selected extracts

200

for the nanofiltration step were determined as follows: moisture (AOAC, 2002; Method n° 925.09),

201

ash (AOAC, 2002; Method n° 923.03), protein (AOAC 2002; Method n° 920.152), lipids (BLIGH

202

and DYER, 1959), dietary fiber (by difference), carbohydrates (HODGE and HOFREITER, 1962),

203

specific mass (Picnometry), apparent viscosity (Flow curves using rheometer AR 1500ex TA

204

Instruments, England) and pH using a phmeter (Kasvi, k39-2014B, Brazil).

205

AC C

EP

199

2.4. Results analysis

206

The normality of the variables was confirmed by the Shapiro-Wilk test and the

207

homogeneity of variance by the Levene test. Two-way analysis of variance (ANOVA) was used to

208

indicate significant differences among the evaluated conditions in the extraction. The correlations

8

ACCEPTED MANUSCRIPT between antioxidant capacity and phenolic contents were assessed using Pearson’s correlation

210

coefficients. Tukey test was used to compare the means obtained in the membrane concentration

211

processes. All analysis were performed using the PROC GLM (procedure general linear model)

212

option of SAS 9.4 (SAS Institute Inc, USA). For evaluation of fouling mechanisms, the

213

mathematical modeling was performed with Statistica 8.0 (Statsoft, USA).

214

3. RESULTS AND DISCUSSION

216 217

3.1. Extraction process

The effects of extraction variables on phenolic compounds, antioxidant capacity and

SC

215

RI PT

209

chlorogenic acid are shown in Figure 1.

The highest extraction yields were obtained using 50 and 75% EtOH. Ethanol reduces the

219

dielectric constant of the solvent, thus enhancing the solubility and diffusion of phenolic

220

compounds (GARCIA-CASTELLO et al., 2015).

M AN U

218

The analysis of variance (R2 ≥0.98) confirmed that this variable was the most significant

222

(p<0.001) in the extraction process. The use of ultrasound was also significant for the extraction

223

efficiency (p<0.05), but ultrasound power higher than 240 W did not contribute to increasing

224

phenolic yields (p<0.05). This condition was used to continue this study in the nanofiltration step.

TE D

221

The interaction between the solvent composition and ultrasound power had no significant

226

effect on the phenolic compound content, whereas for chlorogenic acid content, DPPH and FRAP

227

had. All responses evaluated had high correlation by Pearson test, where the highest correlation

228

was observed for chlorogenic acid and DPPH (r = 0.99; p<0.001), and the lowest for TPC and

229

DPPH (r = 0.94; p<0.001). This result may indicate that most of the antioxidant capacity of

230

artichoke extracts is related to the phenolic acid content present.

AC C

231

EP

225

The yield of chlorogenic acid ranged from 0.02 to 16.47 mg of chlorogenic acid/g dry

232

matter (Figure 1D). For Pandino et al. (2011) and Sihem et al. (2015), the chlorogenic acid content

233

determined in artichoke bracts were 4.16 and 3.18 mg of chlorogenic acid/g of dry matter,

234

respectively. Lombardo et al. (2010) evaluated the influence of genotype and harvest time in

235

artichoke bracts and the levels of this compound ranged from 0.02 to 14.84 mg of chlorogenic

9

ACCEPTED MANUSCRIPT 236

acid/g dry matter. These studies used conventional extraction methods, and the extractions were

237

performed with 70, 95 and 60% of methanol in water (v/v), respectively. We can conclude that the use of EtOH for phenolics extraction using ultrasound is an

239

efficient alternative. The extracts with 50 and 75% EtOH were better than 0 and 25% EtOH to

240

recover phenolic compounds. Kinetic curves (Figure 2) were carried out for these extracts and 10

241

min of sonication were sufficient to obtain a good yield of chlorogenic acid. This time was much

242

lower than the time usually required for extractions by conventional extraction methods. Therefore,

243

these extracts were used in the concentration processes.

3.2.1. Retention of phenolic compounds

SC

245

3.2. Concentration process

M AN U

244

RI PT

238

246

Differences among feed, concentrate and permeate were significant (p<0.001) to all

247

evaluated membranes. Evaluating the retention coefficients (Figure 3), differences among the

248

membranes were observed to TPC, FRAP and chlorogenic acid content. Comparing the extracts (50 and 75% EtOH), for DL membrane, a significant difference

250

was observed for chlorogenic acid content. For DK, the difference was observed to TPC, and for

251

NF270, to TPC, FRAP and chlorogenic acid content. For DPPH, no significant differences were

252

observed between feed and membranes.

TE D

249

For all the membranes, the highest retention was obtained in processes that were conducted

254

with extracts containing 50% EtOH. The major retention of chlorogenic acid was 97.86% for DK

255

(50% EtOH) and the smaller retention for this compound was observed for NF270 in the extract

256

with 75% EtOH (88.67%). This result indicated that EtOH may be promoting the transport of

257

phenolic compounds through the membrane pores. This behaviour was also verified by Pinto et al.

258

(2014) in hydroalcoholic extract (52% and 80% EtOH) obtained from eucalyptus bark. They had

259

also worked with polyamide membranes and claimed that these membranes had been more

260

effectively solvated by water than EtOH. Thus the increase on ethanol percentage leads to a larger

261

pore effective diameter, and hence a smaller hurdle to phenolic acids transport (in the case of

AC C

EP

253

10

ACCEPTED MANUSCRIPT 262

authors, tannic acid), which reflects a greater retention of these compounds in 52% EtOH than in

263

80% EtOH. Despite a smaller retention observed in the 75% EtOH extracts, the retention coefficients to

265

chlorogenic acid were high for all treatments, close to those reported by Conidi et al. (2013), which

266

found values from 82 to 96% of chlorogenic acid, cynarin, and apigenin-7-O-glucoside, in

267

nanofiltration, using wastewater from artichoke bleaching process.

RI PT

264

Even with a different retention, at the end of the concentration processes, for both extracts,

269

the chlorogenic acid content increased around 30 % in the concentrate fraction (22.24 ± 0.75 mg/g

270

dry matter for 75% EtOH and 13.80 ± 0.89 mg/g dry matter for 50% EtOH) in relation to feed

271

(17.14 ± 0.87 mg /g dry matter for 75% EtOH and 11.06 ± 0.10 mg/g dry matter for 50% EtOH). 3.2.2. Permeate flux evaluation

M AN U

272

SC

268

Flux decline can be observed in Figure 4(A) and 4(B), for the extracts obtained with 50%

274

and 75% EtOH, respectively. When comparing filtrations performed with 50% and 75% EtOH, the

275

smallest permeate flux and, consequently, the highest concentration time were observed for the

276

extract obtained with the biggest EtOH percentage. This behaviour can be explained by the

277

hydrophilic nature of the membranes, where the permeation of EtOH/water systems is hindered by

278

restricted hydrogen bonding formation of EtOH compared to aqueous system. Thus, the bigger the

279

solvent polarity, the higher the flux in hydrophilic membranes and the reverse situation is observed

280

for hydrophobic membranes (YANG et al., 2001; BHANUSHALI et al., 2001).

EP

Fluxes showed lower values than those reported for aqueous systems using the same

AC C

281

TE D

273

282

membranes. Cissé et al. (2011) using the NF270, DK and DL for the aqueous extracts

283

concentration of Hibiscus sabdariffa reported flows of 38, 36 and 39 L/m²h, respectively. The

284

permeate fluxes shown in this work, that ranged from 8.42 to 25.27 L/m2h (Table 3) are similar

285

to those reported by Tsibranska and Tylkowski (2013) and Tsui and Cheryan (2007), which

286

obtained fluxes from 8.08 to 15.58 L/m²h (ethanolic extract to concentrate phenolic compounds

287

from Sideritis ssp. using polyamide membranes at 20 bar and 25 °C), and 10,2 L/m²h (extract with

11

ACCEPTED MANUSCRIPT 288

85% EtOH to concentrate xanthophylls from corn, using DK membranes at 50 °C and 27 bar),

289

respectively. For the fluxes obtained with NF270, during the nanofiltration of 50% EtOH extract, the

291

reducing volume factor was reached around three hours of process (Figure 4), whereas for the

292

concentration process with 75% EtOH extract, the time required to reach the same factor was more

293

than twice. This extract has more clusters in suspension that the extract with 50% EtOH (Section

294

3.2.3.1), and this may have influenced on permeate flux. DK and DL membranes showed the same

295

behaviour regarding the two evaluated feeds, confirming that the pore sizes of NF270 (Table 1) are

296

bigger and consequently have a greater available surface for particles deposition (HWANG et al.,

297

2008).

M AN U

SC

RI PT

290

298

Comparing DK and DL membranes (Table 3), the average flux of DL was higher than that

299

of DK. Figure 4 shows that the DK membrane has a good flux stability in both processes, despite

300

its bigger initial flux decay. DL presented a continuously flux decay, indicating that this membrane

301

was more susceptible to fouling along the process, comparing to the DK membrane.

303

3.2.3. Fouling evaluation

TE D

302

3.2.3.1. Fouling mechanisms

Hermia’s models were applied to identify the predominant fouling mechanisms during the

305

processes. Based on the initial stage of the filtration, the flux decline occured due to the

306

concentration polarization (MARSHALL and DAUFIN, 1996). Thus, the modeling was made after

307

a 4.5 minute processing time. This point was established because it represents the transition to the

308

filtration stage where the fouling began.

AC C

309

EP

304

Table 4 shows the estimated constants (ɛc, ɛs, ɛi and ɛCF) and the determination coefficients

310

(R2). The parameters ɛc and ɛi are linked to the resistance caused by the fouling; ɛs, is related to the

311

resistance from the solute particles absorption within the pore, and ɛCF is associated with the

312

resistance due to the cake formation (VELA et al., 2008).

313

Evaluating the determination coefficients, the most predominant fouling mechanism is the

314

cake formation followed by the intermediary, the standard and the complete blocking of the pores.

12

ACCEPTED MANUSCRIPT For NF270, the determination coefficient of cake formation mechanism and intermediary fouling

316

are only slightly different, which is in agreement with the discussion in the last section with respect

317

to particles deposition in the pores of the NF270. Still for NF270, in the extract with 50% EtOH,

318

the difference among standard fouling, intermediary and cake formation of this membrane was in

319

the third decimal place, also showing that solutes adsorption had occurred in membrane pores,

320

which is consistent with the smaller clusters observed in this extract (Section 3.2.3.1).

RI PT

315

At last, for the ɛCF parameter, the NF270 and DL membranes showed the highest values,

322

indicating that these membranes were more sensitive to fouling. Furthermore, these membranes

323

showed the same continuously flux decay along the processes (Figure 4).

325 326

3.2.3.1. Particles size distribution

M AN U

324

SC

321

In the concentration processes, the mass transfer through the membrane was directly affected by the solute particles size (Figure 5).

For the extracts with 50% EtOH, the amount of particles in the range of 1 µm present in the

328

concentrate reduced in relation to feed. Therefore, these particles may have permeated or fouled the

329

membrane. This result is in accordance with the average particle size in permeate, that was less

330

than 0.89 ± 0.01 nm in all evaluated conditions.

TE D

327

The average particle size was bigger for the extract with 75% EtOH (≈ 300 µm) than for

332

the 50% EtOH (≈ 100 µm). These particles may be formed by the complexation of phenolic

333

compounds with carbohydrates and proteins present in the extract (EAGLES AND WAKEMAN,

334

2002; MARROQUIN et al., 2014; ZATOR et al., 2009). Table 5 shows the proximal composition

335

of these extracts. It can be observed that the extract with 75% EtOH contained a higher amount of

336

carbohydrates and proteins than the one with 50% EtOH, which may explain the bigger particles in

337

the first extract.

AC C

EP

331

338

Increasing the percentage of ethanol may have favored particles agglomeration (clusters

339

formation), since in polar solvents, the macromolecules suspended in the extract are generally

340

electrostatically more stable and less susceptible to agglomeration or precipitation, suggesting that

341

the extract with 75% EtOH in water (v/v) is the most likely to promote the fouling.

13

ACCEPTED MANUSCRIPT 342 343

3.2.3.2. Membrane surfaces characterization a) Force atomic microscopy (AFM) From AFM analysis the mean square roughness (Figure 6) was determined. It is important

345

to highlight that this measurement is dependent on the scale property (BOUSSU et al., 2005;

346

WONG et al., 2009), i.e, the higher the scanned area, the higher the roughness values. Therefore, 1,

347

3, 5 and 10 µm² areas were analysed.

RI PT

344

In Figure 6, we can see that for virgin membranes the roughness increased in this order

349

NF270-V < DL-V < DK-V. This result explained the flux decay in the beginning of DK membrane

350

processes. According to Vrijenhoek et al. (2001), in the initial stages of filtration, particles

351

accumulated in the “valleys” of rough membranes, resulting in “valley clogging” and hence in a

352

flux decline.

M AN U

SC

348

The NF270 virgin membrane (NF270-V), has a less rough surface, compared to the other

354

membranes. According to Balcıoğlu and Gönder (2014), this was observed because this membrane

355

is a semi-aromatic polyamide based on piperazine, showing higher hydrophilic properties [due to

356

higher carboxylic acid content (-COOH)], and a smooth surface.

TE D

353

After the concentration processes, a significant increase in surface roughness in relation to

358

virgin membranes was observed, due to feed solute deposition on membranes surface. This result

359

was another indicator of the cake formation during concentration. Furthermore, the roughness for

360

evaluated membranes (Figure 6) after concentration was positively correlated to the decay on

361

permeate flux (Table 3), and both parameters (roughness and flux drecrease) followed the same

362

order: NF270-50% < DL-50% < DK-50% < DL-75% < DK-75% < NF270-75%.

364

AC C

363

EP

357

b) Infrared Fourier transformed spectroscopy In the FTIR analysis, Figure 7 shows the spectrum of virgin and fouled membranes. Some

365

characteristic bands were identified in the virgin membranes indicating the presence of polysulfone

366

(1585, 1504, 1488, 1387, 1364, 1324, 1241, 1150 and 874 cm-1) (PURO et al., 2006; SARI and

367

CHELLAM, 2013; TANG et al., 2009), and polyamide (≈ 1637 cm-1) (TANG et al., 2009), that are

368

respectively, the support and the selective layer of the membranes.

14

ACCEPTED MANUSCRIPT Comparing the fouled membranes with the virgin ones, the absorption intensity of the

370

characteristic bands in virgin membranes was reduced due to the organic matter overlapping in the

371

membrane surface. In fouled membranes, bands with marked intensity in the region of 950-1110

372

cm-1 (mainly 1035 cm-1) were attributed to C-O-C and C-O-P vibrations, which corresponds to the

373

presence of polysaccharides (SARI and CHELLAM, 2013; THYGESEN et al., 2014). Thus, we

374

can infer that most of the fouling in these membranes is due to presence of carbohydrates, which

375

agrees with the proximal composition of the extracts (Table 5). c) Contact angle

SC

376

RI PT

369

The contact angle (θ) measurements allow the evaluation of the membrane surface

378

hydrophobicity. According to this, if 0° < θ < 90°, the membrane surface is more hydrophobic than

379

if 90° < θ < 180° (WANG et al., 2006). The contact angles of the membrane surfaces NF270, DK

380

and DL both conditioned and unconditioned are shown in Figure 8.

M AN U

377

For condictioned membranes, the hydrophobicity increased toward the membranes NF270

382

< DL < DK (the same order was observed for roughness of virgin membranes), and the molecular

383

affinity for nonpolar species of these membranes increased with the conditioning process, i.e,

384

higher contact angles were observed in conditioned membranes than in unconditioned ones. As

385

hydrophobicity depends on the presence of polar and nonpolar groups at membrane surface, it is

386

possible to conclude that exposing the membranes for several hours to the solvents, the adsorption

387

of alcoholic functional group (C-OH) at the membrane structure makes it more hydrophobic. This

388

result shows that conditioning facilitates mass transport of nonpolar species through the membrane.

389

Besides that, Jansen et al. (2013) and Van der Bruggen et al. (2002) showed that conditioning can

390

prevent pore collapse during permeation of nonpolar solvents, ensuring the structural integrity of

391

the membrane.

AC C

EP

TE D

381

392

In conditioned membranes, the contact angles were higher for conditioning processes in

393

solutions with 75% EtOH than 50%, suggesting a bigger flux for membranes conditioned with 50%

394

EtOH, as observed in this study (Figure 4). This behaviour has already been mentioned by other

395

authors in respect to hydrophilic membranes (BHANUSHALI et al., 2001; YANG et al., 2001),

15

ACCEPTED MANUSCRIPT and it is linked to the restricted hydrogen bonding ability of EtOH, which renders difficult an

397

effective membrane solvation. Furthermore, the contact angle is also related to the effective pore

398

size and impacts directly the retention of phenolic compounds, because hydrophilic membranes

399

were not satisfactorily solvated by organic solvent as they are by water (PINTO et al., 2014).

400

4. CONCLUSION

RI PT

396

The extracts with higher ethanol concentration (50 and 75% EtOH) showed the best

402

extraction yields . The use of ultrasound also favored the extraction of phenolic compounds, but

403

ultrasound power higher than 240 W had no influence on process efficiency. In addition, 10 min of

404

sonication were enough to obtain good extraction yields. Membranes NF270, DK and DL can be

405

used to concentrate hydroalcoholic extracts from artichoke waste due to almost total retention of

406

phenolic compounds. The susceptibility to fouling in the membranes follows the order NF270 >

407

DL > DK, and the predominant compound found on membranes surface was carbohydrates. As a

408

retention of more than 95% of chlorogenic acid was observed for extracts with 50% of ethanol and

409

the DK membrane showed less susceptible to fouling, with a more stable flux and a short

410

concentration time, the combination of the most suitable process to obtain the highest chlorogenic

411

acid should be ultrasound extraction (20 kHz and 240 W), with 50% EtOH, and concentration by

412

nanofiltration at 20 bar in a tangential filtration using DK membrane.

413

5. ACKNOWLEDGMENTS

EP

TE D

M AN U

SC

401

The authors thank FAPESP (2009/54137-1 and 2009/50593-2), CNPq (143885/2011-1,

415

130647/2013-6 and 304475/2013-0) and CAPES for the financial support. The authors are also

416

very grateful to Bonsucesso factory for the raw material donation and to LAMULT (IFGW-

417

UNICAMP) for AFM and FTIR analysis.

418

6. REFERENCES

419 420 421

Abu-reidah, I.M., Arráez-Román, D., Segura-Carretero, A., Fernández-Gutiérrez, A., 2013. Extensive characterisation of bioactive phenolic constituents from globe artichoke (Cynara scolymus L .) by HPLC-DAD-ESI-QTOF-MS. Food Chem. 141, 2269–2277.

422 423 424

Almazán, J.E., Romero-Dondiz, E.M., Rajal, V.B., Castro-Vidaurre, E.F., 2015. Nanofiltration of glucose: Analysis of parameters and membrane characterization. Chem. Eng. Res. Des. 94, 485–493.

AC C

414

16

ACCEPTED MANUSCRIPT AOAC - Official Methods of Analysis of the Association of Official Analytical Chemists, 17th ed, 2002. Maryland, USA.

427 428 429

Aspé, E., Fernández, K., 2011. The effect of different extraction techniques on extraction yield, total phenolic, and anti-radical capacity of extracts from Pinus radiata Bark. Ind. Crops Prod. 34, 838–844.

430 431

Balcıoğlu, G., Gönder, Z.B., 2014. Recovery of baker’s yeast wastewater with membrane processes for agricultural irrigation purpose: Fouling characterization. Chem. Eng. J. 255, 630–640.

432 433 434

Bhanushali, D., Kloos, S., Kurth, C., Bhattacharyya, D., 2001. Performance of solvent-resistant membranes for non-aqueous systems: solvent permeation results and modeling. J. Memb. Sci. 189, 1–21.

435 436

Bligh, E.G., Dyer, W.J., 1959. A rapid method of total lipid extraction and purification. Can. J. Biochem. Physiol. 37.

437 438 439

Boussu, K., Van der Bruggen, B., Volodin, A., Snauwaert, J., Van Haesendonck, C., Vandecasteele, C., 2005. Roughness and hydrophobicity studies of nanofiltration membranes using different modes of AFM. J. Colloid Interface Sci. 286, 632–638.

440 441 442

Carvalho, A. L., Maugeri, F., Silva, V., Hernández, A., Palacio, L., Pradanos, P., 2011. AFM analysis of the surface of nanoporous membranes: application to the nanofiltration of potassium clavulanate. J. Mater. Sci. 46, 3356–3369.

443 444

Chantaro, P., Devahastin, S., Chiewchan, N., 2008. Production of antioxidant high dietary fiber powder from carrot peels. LWT - Food Sci. Technol. 41, 1987–1994.

445 446

Chemat, F., Zill-e-Huma, Khan, M.K., 2011. Applications of ultrasound in food technology: Processing, preservation and extraction. Ultrason. Sonochem. 18, 813–835.

447 448 449

Chisté, R.C., Mercadante, A.Z., 2012. Identification and quantification, by HPLC-DAD-MS/MS, of carotenoids and phenolic compounds from the amazonian fruit Caryocar villosum. J. Agric. Food Chem. 60, 5884–5892.

450 451 452

Cho, A.-S., Jeon, S.-M., Kim, M.-J., Yeo, J., Seo, K.-I., Choi, M.-S., Lee, M.-K., 2010. Chlorogenic acid exhibits anti-obesity property and improves lipid metabolism in high-fat diet-induced-obese mice. Food Chem. Toxicol. 48, 937–943.

453 454 455

Choi, C.W., Kim, S.C., Hwang, S.S., Choi, B.K., Ahn, H.J., Lee, M.Y., Park, S.H., Kim, S.K., 2002. Antioxidant activity and free radical scavenging capacity between Korean medicinal plants and flavonoids by assay-guided comparison. Plant Sci. 163, 1161–1168.

456 457 458

Cissé, M., Vaillant, F., Pallet, D., Dornier, M., 2011. Selecting ultrafiltration and nanofiltration membranes to concentrate anthocyanins from roselle extract (Hibiscus sabdariffa L.). Food Res. Int. 44, 2607–2614.

459 460 461 462

Claus, T., Maruyama, S.A., Palombini, S.V., Montanher, P.F., Bonafé, E.G., de Oliveira Santos Junior, O., Matsushita, M., Visentainer, J. V., 2015. Chemical characterization and use of artichoke parts for protection from oxidative stress in canola oil. LWT - Food Sci. Technol. 61, 346–351.

463 464

Coinu, R., Carta, S., Urgeghe, P.P., Mulinacci, N., Pinelli, P., Franconi, F., Romani, A., 2007. Dose-effect study on the antioxidant properties of leaves and outer bracts of extracts obtained

AC C

EP

TE D

M AN U

SC

RI PT

425 426

17

ACCEPTED MANUSCRIPT 465

from Violetto di Toscana artichoke. Food Chem. 101, 524–531. Conidi, C., Cassano, A., Drioli, E., 2012. Recovery of phenolic compounds from orange press liquor by nanofiltration. Food Bioprod. Process. 90, 867–874.

468 469

Conidi, C., Cassano, A., Garcia-Castello, E., 2013. Valorization of artichoke wastewaters by integrated membrane process. Water Res. 39, 1–12.

470 471

Eagles, W., Wakeman, R., 2002. Interactions between dissolved material and the fouling layer during microfiltration of a model beer solution. J. Memb. Sci. 206, 253–264.

472 473

Esclapez, M.D., García-Pérez, J. V., Mulet, A., Ca, J.A., 2011. Ultrasound-Assisted Extraction of Natural Products. Food Eng. Rev. 3, 108–120.

474 475 476

Fratianni, F., Tucci, M., Palma, M. De, Pepe, R., 2007. Polyphenolic composition in different parts of some cultivars of globe artichoke (Cynara cardunculus L . var . scolymus (L.) Fiori). Food Chem. 104, 1282–1286.

477 478 479

Gaete-Garretón, L., Vargas-Hernández, Y., Cares-Pacheco, M.G., Sainz, J., Alarcón, J., 2011. Ultrasonically enhanced extraction of bioactive principles from Quillaja Saponaria Molina. Ultrasonics 51, 581–585.

480 481 482

Garcia-Castello, E.M., Rodriguez-Lopez, A.D., Mayor, L., Ballesteros, R., Conidi, C., Cassano, A., 2015. Optimization of conventional and ultrasound assisted extraction of flavonoids from grapefruit (Citrus paradisi L.) solid wastes. LWT - Food Sci. Technol. 64, 1114–1122.

483 484 485

Ghafoor, K., Hui, T., Choi, Y.H., 2011. Optimization of ultrasonic-assisted extraction of total anthocyanins from grape peel using response surface methodology. J. Food Biochem. 35, 735–746.

486 487 488

Ghitescu, R.-E., Volf, I., Carausu, C., Bühlmann, A.-M., Gilca, I.A., Popa, V.I., 2015. Optimization of ultrasound-assisted extraction of polyphenols from spruce wood bark. Ultrason. Sonochem. 22, 535–541.

489 490

Grenier, A., Meireles, M., Aimar, P., Carvin, P., 2008. Analysing flux decline in dead-end filtration. Chem. Eng. Res. Des. 86, 1281–1293.

491 492

Hermia, J., 1982. Constant pressure blocking filtration laws - Apliccation to power-law nonnewtonian fluids. Chem. Eng. Res. Des. 60, 183–187.

493 494

Hodge, J.E., Hofreiter, B.R., 1962. Determination of reducing sugar and carbohydrates. Academic Press, New York.

495 496

Hwang, K.-J., Liao, C.-Y., 2012. Effects of membrane morphology and operating conditions on microfiltration particle fouling. J. Taiwan Inst. Chem. Eng. 43, 46–52.

497 498

Hwang, K.-J., Liao, C.-Y., Tung, K.-L., 2007. Analysis of particle fouling during microfiltration by use of blocking models. J. Memb. Sci. 287, 287–293.

499 500

Hwang, K.-J., Liao, C.-Y., Tung, K.-L., 2008. Effect of membrane pore size on the particle fouling in membrane filtration. Desalination 234, 16–23.

501 502 503

Jansen, J.C., Darvishmanesh, S., Tasselli, F., Bazzarelli, F., Bernardo, P., Tocci, E., Friess, K., Randova, A., Drioli, E., Van der Bruggen, B., 2013. Influence of the blend composition on the properties and separation performance of novel solvent resistant

AC C

EP

TE D

M AN U

SC

RI PT

466 467

18

ACCEPTED MANUSCRIPT 504

polyphenylsulfone/polyimide nanofiltration membranes. J. Memb. Sci. 447, 107–118. Llorach, R., Espín, J.C., Tomás-Barberán, F.A., Ferreres, F., 2002. Artichoke (Cynara scolymus L.) byproducts as a potential source of health-promoting antioxidant phenolics. J. Agric. Food Chem. 50, 3458–3464.

508 509

Lombardo, S., Pandino, G., Ierna, A., Mauromicale, G., 2012. Variation of polyphenols in a germplasm collection of globe artichoke. Food Res. Int. 46, 544–551.

510 511 512

Lombardo, S., Pandino, G., Mauromicale, G., Knödler, M., Carle, R., Schieber, A., 2010. Influence of genotype , harvest time and plant part on polyphenolic composition of globe artichoke [Cynara cardunculus L . var . scolymus (L.) Fiori]. Food Chem. 119, 1175–1181.

513 514 515

López-Molina, D., Navarro-Martínez, M.D., Rojas Melgarejo, F., Hiner, A.N.P., Chazarra, S., Rodríguez-López, J.N., 2005. Molecular properties and prebiotic effect of inulin obtained from artichoke (Cynara scolymus L.). Phytochemistry 66, 1476–1484.

516 517

Louli, V., Ragoussis, N., Magoulas, K., 2004. Recovery of phenolic antioxidants from wine industry by-products. Bioresour. Technol. 92, 201–208.

518 519 520

Luque de Castro, M.D., Priego-Capote, F., Peralbo-Molina, A., 2011. The role of ultrasound in analytical derivatizations. J. Chromatogr. B. Analyt. Technol. Biomed. Life Sci. 879, 1189– 1195.

521 522 523

Machado, M.T.C., Mello, B.C.B.S., Hubinger, M.D., 2014. Evaluation of pequi (Caryocar brasiliense Camb.) aqueous extract quality processed by membranes. Food Bioprod. Process. doi: 10.1016/j.fbp.2014.10.013

524 525 526

Marroquin, M., Vu, A., Bruce, T., Ranil Wickramasinghe, S., Zhao, L., Husson, S.M., 2014. Evaluation of fouling mechanisms in asymmetric microfiltration membranes using advanced imaging. J. Memb. Sci. 465, 1–13.

527 528

Marshall, A.D., Daufin, G., 1996. Physico-chemical aspects of membrane fouling by dairy fluids. Cat./[S.I.] 311, 39.

529 530 531

Mohammad, A.W., Teow, Y.H., Ang, W.L., Chung, Y.T., Oatley-Radcliffe, D.L., Hilal, N., 2015. Nanofiltration membranes review: Recent advances and future prospects. Desalination 356, 226–254.

532 533

Ng, C.Y., Mohammad, A.W., Ng, L.Y., Jahim, J.M., 2014. Membrane fouling mechanisms during ultrafiltration of skimmed coconut milk. J. Food Eng. 142, 190–200.

534 535 536

Nghiem, L.D., Hawkes, S., 2007. Effects of membrane fouling on the nanofiltration of pharmaceutically active compounds (PhACs): Mechanisms and role of membrane pore size. Sep. Purif. Technol. 57, 176–184.

537 538

Pan, B., Yan, P., Zhu, L., Li, X., 2013. Concentration of coffee extract using nanofiltration membranes. Desalination 317, 127–131.

539 540 541

Pandino, G., Lombardo, S., Mauromicale, G., Williamson, G., 2011. Profile of polyphenols and phenolic acids in bracts and receptacles of globe artichoke (Cynara cardunculus var . scolymus) germplasm. J. Food Compos. Anal. 24, 148–153.

542

Pandino, G., Lombardo, S., Williamson, G., Mauromicale, G., 2012. Polyphenol profile and

AC C

EP

TE D

M AN U

SC

RI PT

505 506 507

19

ACCEPTED MANUSCRIPT 543

content in wild and cultivated Cynara cardunculus L. Ital. J. Agron. 7, 254-261. Pinto, P.C.R., Mota, I.F., Loureiro, J.M., Rodrigues, A.E., 2014. Membrane performance and application of ultrafiltration and nanofiltration to ethanol/water extract of Eucalyptus bark. Sep. Purif. Technol. 132, 234–243.

547 548 549 550

Pistón, M., Machado, I., Branco, C.S., Cesio, V., Heinzen, H., Ribeiro, D., Fernandes, E., Chisté, R.C., Freitas, M., 2014. Infusion, decoction and hydroalcoholic extracts of leaves from artichoke (Cynara cardunculus L. subsp. cardunculus) are effective scavengers of physiologically relevant ROS and RNS. Food Res. Int. 64, 150–156.

551 552 553

Puro, L., Mänttäri, M., Pihlajamäki, A., Nyström, M., 2006. Characterization of modified nanofiltration membranes by octanoic acid permeation and FTIR analysis. Chem. Eng. Res. Des. 84, 87–96.

554 555 556

Ranic, M., Nikolic, M., Pavlovic, M., Buntic, A., Siler-Marinkovic, S., Dimitrijevic-Brankovic, S., 2014. Optimization of microwave-assisted extraction of natural antioxidants from spent espresso coffee grounds by response surface methodology. J. Clean. Prod. 80, 69–79.

557 558

Rastogi, N.K., 2011. Opportunities and challenges in application of ultrasound in food processing. Crit. Rev. Food Sci. Nutr. 51, 705–722.

559 560 561 562

Ruiz-Cano, D., Pérez-Llamas, F., Frutos, M.J., Arnao, M.B., Espinosa, C., López-Jiménez, J.Á., Castillo, J., Zamora, S., 2014. Chemical and functional properties of the different by-products of artichoke (Cynara scolymus L.) from industrial canning processing. Food Chem. 160, 134– 140.

563 564 565

Sari, M.A., Chellam, S., 2013. Surface water nanofiltration incorporating (electro) coagulation– microfiltration pretreatment: Fouling control and membrane characterization. J. Memb. Sci. 437, 249–256.

566 567 568

Sihem, D., Samia, D., Gaetano, P., Sara, L., Giovanni, M., Hassiba, C., Laura, G., Ahmed Noureddine, H., 2015. In vitro antioxidant activities and phenolic content in crop residues of Tunisian globe artichoke. Sci. Hortic. (Amsterdam). 190, 128–136.

569 570 571

Singleton, V.L., Orthofer, R., Lamuela-Raventós, R.M., 1999. Analysis of total phenols and others oxidation substrates and antioxidants by means of Folin-Ciocauteau reagent. Methods Enzymol. 299, 152–178.

572 573 574

Tang, C.Y., Kwon, Y.-N., Leckie, J.O., 2009. Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes. Desalination 242, 149–167.

575 576

Thygesen, O., Hedegaard, M.A.B., Zarebska, A., Beleites, C., Krafft, C., 2014. Membrane fouling from ammonia recovery analyzed by ATR-FTIR imaging. Vib. Spectrosc. 72, 119–123.

577 578 579

Tsibranska, I.H., Tylkowski, B., 2013. Concentration of ethanolic extracts from Sideritis ssp. L. by nanofiltration: Comparison of dead-end and cross-flow modes. Food Bioprod. Process. 91, 169–174.

580 581

Tsui, E.M., Cheryan, M., 2007. Membrane processing of xanthophylls in ethanol extracts of corn. J. Food Eng. 83, 590–595.

582

Tylkowski, B., Trusheva, B., Bankova, V., Giamberini, M., Peev, G., Nikolova, A., 2010.

AC C

EP

TE D

M AN U

SC

RI PT

544 545 546

20

ACCEPTED MANUSCRIPT Extraction of biologically active compounds from propolis and concentration of extract by nanofiltration. J. Memb. Sci. 348, 124–130.

585 586

Van der Bruggen, B., Geens, J., Vandecasteele, C., 2002. Influence of organic solvents on the performance of polymeric nanofiltration membranes. Sep. Sci. Technol. 37, 783–797.

587 588

Vela, M.C.V., Blanco, S.Á., García, J.L., Rodríguez, E.B., 2008. Fouling dynamics modelling in the ultrafiltration of PEGs. Desalination 222, 451–456.

589 590 591

Vieira, G.S., Cavalcanti, R.N., Meireles, M.A.A., Hubinger, M.D., 2013. Chemical and economic evaluation of natural antioxidant extracts obtained by ultrasound-assisted and agitated bed extraction from jussara pulp (Euterpe edulis). J. Food Eng. 119, 196–204.

592 593 594

Vrijenhoek, E.M., Hong, S., Elimelech, M., 2001. Influence of membrane surface properties on initial rate of colloidal fouling of reverse osmosis and nanofiltration membranes. J. Memb. Sci. 188, 115–128.

595 596

Wang, X., Zhang, H., Zhang, J., Xu, H., Zhu, X., Chen, J., Yi, B., 2006. A bi-functional microporous layer with composite carbon black for PEM fuel cells. J. Power Sources 162, 474–479.

597 598 599

Wong, P.C.Y., Kwon, Y.-N., Criddle, C.S., 2009. Use of atomic force microscopy and fractal geometry to characterize the roughness of nano-, micro-, and ultrafiltration membranes. J. Memb. Sci. 340, 117–132.

600 601

Yang, X.J., Livingston, A.G., Santos, L.F. dos, 2001. Experimental observations of nanofiltration with organic solvents. J. Memb. Sci. 190, 45–55.

602 603 604

Zator, M., Ferrando, M., López, F., Güell, C., 2009. Microfiltration of protein/dextran/polyphenol solutions: Characterization of fouling and chemical cleaning efficiency using confocal microscopy. J. Memb. Sci. 344, 82–91.

605 606 607

Zuorro, A., Maffei, G., Lavecchia, R., 2015. Reuse potential of artichoke (Cynara scolimus L.) waste for the recovery of phenolic compounds and bioenergy. J. Clean. Prod. doi: 10.1016/j.jclepro.2015.06.011

AC C

EP

TE D

M AN U

SC

RI PT

583 584

21

ACCEPTED MANUSCRIPT Table 1. Characteristics of the nanofiltration membranes Membrane

NF270

DK

DL

Producer

GE Osmonics

Molecular weight cut-off (g/mol)

Dow FilmTec Polyamide thin-film composite(1) 200-400(1)

200(1)

GE Osmonics Cross-linked aromatic polyamide(1) 150-300(1)

Pore size (nm)

0.84(2)

0.43(3)

0.46(4)

Maximum operating temperature (°C)

45

45

45

Maximum operating pressure (bar)

41

40

40

pH range

3-10

2-11

2-11

(1)

96(1)

4

>97

(1)

98

6(5)

7(5)

SC

17(5)

M AN U

Salt retention (%) Permeability to pure water (kg.h-1.m-2.bar-1) 1 Mohammad et al. (2015) 2 Nghiem and Hawkes (2007) 3 Almazán et al. (2015) 4 Retention of MgSO4, 2000 mg/L at 25 ⁰C 5 Cissé et al. (2011)

Polyamide(1)

RI PT

Composition on top layer

Table 2. Summary of characteristics of Hermia’s models Fouling mechanism Filtration cake formation

n 0

Equation of the fouling mechanism

Jv =

J0

(1+ ε CF .t )

12

Linearized Equation of the fouling mechanism

1

(J v )

2

=

1

(J o )

2

+ ε CF .t

Constant of the fouling mechanism

ε CF =

(2Rr ) K CF v0 J0

2

Intermediate pore blocking

1

Jv =

J0 (1+ εit )

1 1 = + εit J Jo

εi =

K c v0 J0

Standard pore blocking

1.5

Jv =

J0 (1+ ε st )2

1 1 = 1 2 + ε st 12 J Jo

εs =

Ks v0

TE D

Representation

12

J0

AC C

EP

Complete ε c = K c v0 Jv = Jo exp(−ε c .t) ln Jv = ln Jo − ε c .t 2 pore blocking Where Jv is the volumetric flux of the permeate (L/m²h), J0 is the initial permeate flux (L/m²h), t is the time (h), Rr is the ratio of resistance of the cake over the resistance of the clean membrane, KCF is the area of cake formed per unit of permeate volume (m-1), v0 the initial velocity per surface unit area of the membrane surface (m/h), Kc is the membrane blocked surface area by the total number of permeate thought the membrane unit (m-1), Ks is the reduction in the cross sectional area of the pores per total unit permeate flux (m-1).

Table 3. Measured flux during nanofiltration of 50% and 75% EtOH in water (v/v) extracts Process Initial flux (L/m2h) Average flux (L/m2h) Flux decrease (%)

NF270-50%EtOH NF270-75%EtOH DK-50%EtOH DK-75%EtOH DL-50%EtOH DL-75%EtOH

81.00±5.54 76.16±0.93 78.71±3.06 61.58±3.58 88.85±5.54 77.17±1.97

25.27±1.35 8.42±0.22 19.26±2.44 11.77±0.15 24.95±7.14 18.14±2.44

68.80±0.04 88.90±0.01 75.53±0.02 80.88±0.01 71.92±0.10 76.49±0.03

RI PT

ACCEPTED MANUSCRIPT

0.1787 0.1413 0.0648 0.2025

0.989 0.932 0.905 0.654

0.0152 0.0162 0.0068 0.0027

0.995 0.971 0.926 0.680

0.0054 0.0090 0.0029 0.0015

0.997 0.986 0.934 0.690

0.0004 0.0015 0.0003 0.0002

0.998 0.999 0.945 0.710

0.2161 0.1719

0.955 0.960

0.0169 0.0165

0.977 0.979

0.0058 0.0067

0.986 0.987

0.0004 0.0006

0.996 0.993

M AN U

NF270-50%EtOH NF270-75%EtOH DK-50%EtOH DK-75%EtOH DL-50%EtOH DL-75%EtOH

SC

Table 4. Values of the parameters adjusted by Hermia’s model and the respective coefficients of determination (R2) related to the fouling process in membranes NF270, DK and DL Complete blocking (n=2) Standard blocking (n=1.5) Intermediate blocking (n=1) Cake formation (n=0) Process εc(s-1) R2 εs(s-1/2m-1/2) R2 εi(m-1) R2 εCF(s.m-2) R2

AC C

EP

TE D

Where R2 is the coefficient of determination, ɛc is the constant of the complete blocking model, ɛs is the constant of the standard blocking model, ɛi is the constant of the intermediate blocking model, ɛCF is the constant of the cake formation model, and n is the exponent in Hermia’s models that characterizes the type of filtration mechanism

ACCEPTED MANUSCRIPT Table 5. Proximal composition and physico-chemical properties of artichoke extracts obtained with 50% and 75% of EtOH in water (v/v). 75% EtOH 99.40 ± 0.01 13.44 ± 0.97 22.19 ± 1.46 13.80 ± 1.99 27.49 ± 0.04 51.56 ± 1.66 0.86 ± 0.01 2.03 ± 0.01 6.40 ± 0.04

RI PT

Moisture (g/100 g, w.b.) Ash (g/100 g, d.b.) Protein (g/100 g, d.b.) Lipids (g/100 g, d.b.) Dietary fiber (g/100 g, d.b.) Carbohydrates (g/100 g, d.b.) Specific mass (g/mL) Apparent viscosity (mPa.s) pH

50% EtOH 99.51 ± 0.01 9.27 ± 0.97 16.59 ± 1.46 32.18 ± 1.99 7.66 ± 0.71 43.11 ± 2.05 0.92 ± 0.01 2.40 ± 0.01 6.35 ± 0.04

AC C

EP

TE D

M AN U

SC

Data presented corresponds to an average of five determinations, with standard deviations. w.b. = wet basis, d.w. = dry weight

ACCEPTED MANUSCRIPT Figure 1. TPC (A), DPPH (B), FRAP (C) and Chlorogenic acid content (D) in function of ultrasound power and solvent composition Figure 2. Extraction kinetics of TPC (A), DPPH (B), FRAP (C) and Chlorogenic acid

RI PT

content (D) from artichoke waste Figure 3. Retention coefficient of the responses TPC (A), DPPH (B), FRAP (C) and Chlorogenic acid content (D) - (Capital letters indicate a statistically significant

SC

difference (p≤0.05) for the same feed in different membranes. Lowercase letters indicate a statistically significant difference (p≤0.05) for the same membrane in

M AN U

different feeds)

Figure 4. Permeate flux decline for membranes NF270, DK and DL refers to the extracts with 50% (A) and 75% (B) of EtOH in water (v/v), with a volume reduction

TE D

factor of 2.5

Figure 5. Particle size distribution for feed and concentrated, obtained from the extracts obtained with 50% (A) and 75% (B) of ethanol in water (v/v)

EP

Figure 6. Mean square roughness (nm), in function of the length scale L (side of the

AC C

scanned area, µm) for membranes NF270, DK and DL before and after the filtration process

Figure 7. Fourier transform infrared spectra of virgin and fouled NF270, DK and DL membranes

Figure 8. Contact angle to the conditioned and unconditioned membranes (Capital letters indicate a statistically significant difference (p≤0.05) for the same membrane in

ACCEPTED MANUSCRIPT different conditions evaluated. Lowercase letters indicate a statistically significant

AC C

EP

TE D

M AN U

SC

RI PT

difference (p≤0.05) for the same condition evaluated in different membranes)

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Highlights Ultrasound of 20 kHz was used to extract phenolics from artichoke waste (bracts). Total phenols and chlorogenic acid were concentrated by nanofiltration. Permeate fluxes and fouling were influenced by alcoholic content of extract.

RI PT

Hermia's models and surface analyses identified the membrane more susceptible to fouling

AC C

EP

TE D

M AN U

SC

Membrane selection was done in terms of flux, fouling behavior and phenols retention.