Integrated membrane process for purification and concentration of aqueous Syzygium cumini (L.) seed extract

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

Contents lists available at ScienceDirect

Food and Bioproducts Processing journal homepage: www.elsevier.com/locate/fbp

Integrated membrane process for purification and concentration of aqueous Syzygium cumini (L.) seed extract Upasna Balyan, Biswajit Sarkar ∗ University School of Chemical Technology, GGS Indraprastha University, Delhi 110078, India

a r t i c l e

i n f o

a b s t r a c t

Article history:

The aim of the present study was to investigate the potentiality of an integrated membrane

Received 21 May 2015

process for the purification and the concentration of phenolic compounds from aqueous

Received in revised form 23

jamun (Syzygium cumini L.) seed extract. The aqueous seed extract obtained at optimal

November 2015

condition (temperature: 49.2 ◦ C, time: 89.4 min, and liquid to solid ratio: 51.6:1 mL/g) was

Accepted 14 December 2015

submitted to cross flow ultrafiltration for initial clarification, followed by concentration

Available online 23 December 2015

using nanofiltration under batch concentration mode. A detailed parametric study was carried out to investigate the effect of various process parameters such as transmembrane

Keywords:

pressure, cross-flow velocity (or stirrer speed) on the permeate flux and permeate quality.

Syzygium cumini (L.)

Using classical film theory, a steady state gel polarization model incorporating the effect

Polyphenol

of transmembrane pressure difference and viscosity variation was proposed for the predic-

Membrane processes

tion of permeate flux during cross flow ultrafiltration of aqueous seed extract. The predicted

Concentration polarization

flux values were successfully compared with the experimental results. Experimental results showed that the operating conditions had significant effect on permeate flux, recovery of polyphenols, purity and antioxidant activity of phenolic extract. Ultrafiltration experiments at lower operating pressures (276 and 414 kPa) using 100 kDa membrane resulted in the recovery of 59–66.7% of total polyphenol content in the clarified extract with the purity of 49–58.3% starting from an extract purity of 39.2%. The clarified extract could be successfully concentrated about three times higher using 250 Da nanofiltration membrane at volume concentration ratio of three. The present study revealed that the UF/NF integrated membranes process was successful in clarifying and concentrating phenolic extract obtained from jamun seed with enhanced purity and antioxidant activity. © 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1.

Introduction

Jamun (Syzygium cumini L.), is one of the nutritious fruit of the Myrtaceae family. Jamun fruit, native to India, is widely cultivated in the various parts of India (ICAR Report, 2014). It is also being grown in most of the tropical and subtropical regions of the world such as Thailand, Philippines, Madagascar, West Indies, California, and Algeria. The domestic and industrial use of these large quantities of jamun fruit,

∗

especially for the production of juice and wine, results in the accumulation of large amounts of seed as a by-product which account about 20% of the fruit weight. This represents a serious disposal problem from an economical and environmental point of view (Patil et al., 2012). Previous studies have reported that jamun seed extract is a rich source of phenolic compounds with antioxidant, anti-inflammatory capacity, such as ellagitannins, gallic acid, ␤-sitosterol and ellagic acid (Kaneria and Chandra, 2013; Swami et al., 2012). The

Corresponding author. Tel.: +91 11 25302474. E-mail address: [email protected] (B. Sarkar). http://dx.doi.org/10.1016/j.fbp.2015.12.005 0960-3085/© 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

30

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

Nomenclature a1 , a2 , a3 a, b C Co Cb Cg C∗g CP de D k0 k1 L N Pew Re Rg Rm Rg∗ Rt Rrev Rirr RCP ∗ RCP R0 Sc Sh Shm t0 tf VP VR Vw 0 Vw f Vw exp Vw,ss cal Vw,ss P 0 m ıc ˇ

coefficients in Eq. (A1) model parameter concentration of solute, kg/m3 solute concentration in feed, kg/m3 solute concentration at bulk, kg/m3 solute concentration in gel layer, kg/m3 dimensionless solute concentration in gel layer defined as, C∗g = Cg /C0 solute concentration in permeate, kg/m3 equivalent diameter, m effective diffusivity of solute, m2 /s mass transfer coefficient, m/s modified mass transfer coefficient, m/s channel length, m number of experimental data dimensionless permeate flux defined as, Pew = Vw de /D Reynolds number, dimensionless gel layer resistance, m−1 membrane resistance, m−1 dimensionless gel layer resistance defined as, Rg∗ = Rg /Rm total resistance, m−1 reversible fouling resistance, m−1 irreversible fouling resistance, m−1 concentration polarization resistance, m−1 dimensionless concentration polarization ∗ = R /R resistance defined as, RCP m CP observed retention, dimensionless Schmidt number, dimensionless Sherwood number, dimensionless modified Sherwood number, dimensionless initial time, s final time, s volume of permeate, mL volume of retentate, mL permeate flux, L/m2 h initial permeate flux, L/m2 h final permeate flux, L/m2 h experimental steady state permeate flux, L/m2 h calculated steady state permeate flux, L/m2 h transmembrane pressure difference, kPa viscosity of bulk solution, Pa s viscosity at membrane surface, Pa s thickness of concentration boundary layer, m parameter in Eq. (A7)

recovery of any target compound from food by-products can be accomplished with the so-called “5-Stages Universal Recovery Processing” including: (a) macroscopic pre-treatment, (b) separation of high-molecular from low-molecular compounds, (c) extraction, (d) purification/isolation and (e) encapsulation or product formation (Galanakis, 2012, 2014). In the recovery downstream processing, microfiltration and ultrafiltration are able to remove macro-molecules (i.e. proteins, pectin, etc.) while nanofiltration isolate, purify and concentrate target compounds prior their encapsulation with conventional spray or freeze drying (Galanakis, 2015). The energy efficient

membrane technologies have been used in different sectors of food industry for several decades. Nowadays, membrane technology has become one of the most important industrial separation technique and has been a topic of growing interest for purifying and concentrating bioactive phenolic compounds from extract of various plant sources, such as fruits, seed, peels, leaves, roots, and barks (Conidi et al., 2011, 2012; Chhaya et al., 2011; Yu et al., 2007; Prudêncioa et al., 2012) although some emerging technologies are explored recently in research level and in some cases applied in the food industry (Galanakis, 2013). During extraction of phenolic compounds, higher molecular weight polysaccharides (cellulose, pectins, etc.) are also co-extracted leading to decrease in purity of the phenolic extract. The aim of the membrane-based clarification of plant extract is the removal of polysaccharides and maximum permeation of phenolic and flavonoid compounds resulting to an increase in purity of the extract. Microfiltration (MF) or ultrafiltration (UF) represent well-established technologies in the clarification of plant extracts, while concentration by nanofiltration (NF) or reverse osmosis (RO) has been reported in several studies (Díaz-Reinoso et al., 2009; Murakami et al., 2011; Mello et al., 2010). Recently, the potential applications of integrated membrane process have been proposed for concentration of plant extracts (Cissé et al., 2011; Chhaya et al., 2012; Torun et al., 2014) as well as fruit juices (Cassano et al., 2003, 2006; Alvarez et al., 2000). Very few studies with limited information are available on the antioxidant potential of the jamun seed (Swami et al., 2012). However, there are almost no scientific references dealing with the treatment of jamun seed extract by membrane processes. For this reason the present work is focused on the aqueous extract from jamun seed and in membrane processing for clarification/concentration of polyphenols. Although there are many advantages of membrane-based processes, a common phenomenon in this process is the decline in permeate flux with time. This occurs due to concentration polarization and membrane fouling. In the jamun seed extract clarification, the process is mainly limited by the accumulation of pectin, starch, hemicelluloses, cellulose, etc. over the membrane surface. Membrane fouling can be reduced by selecting a suitable membrane and set of operating conditions. Hence, the role of operating conditions is extremely important in membrane separation processes. The aim of this work is to evaluate the potential of an integrated membrane process, by selecting suitable membrane and operating conditions, for purifying and concentrating phenolic compounds from jamun seed extract using water as a solvent. In particular, aqueous jamun seed extract is submitted to a preliminary purification process using UF membranes, followed by a concentration by nanofiltration. Furthermore, a suitable model has been formulated to predict steady state permeate flux during cross flow ultrafiltration of seed extract. Aqueous seed extract being a complex mixture of several solutes, the unknown transport properties like gel layer concentration, solute diffusivity etc. are difficult to estimate. However, these parameters are system specific and their estimation is warranted for efficient design and subsequent scaling. Moreover, during ultrafiltration there is a significant variation of solution viscosity and solute diffusivity within thin concentration boundary layer near the membrane surface due to sharp variation in the solute concentration. The ultrafiltration of seed extract is assumed to be a gel controlling and the proposed model is based on classical film theory. The

31

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

effect of transmembrane pressure on mass transfer coefficient and viscosity variation effect within the thin concentration boundary layer are incorporated in the proposed model.

2.

Pew = Sh ln C∗g

(1)

where C∗g was the dimensionless gel concentration of the gel forming solute. The pressure independent mass transfer coefficient under laminar flow condition could be obtained from the following expression of Sherwood number (Sarkar et al., 2008):



k0 de de = 2.1 ReSc D L

 Pew = Sh[1 − a e

Theory

During cross-flow ultrafiltration of aqueous jamun seed extract flowed tangentially over the membrane surface, and the permeate flowed normally through the membrane because of the transmembrane pressure difference. Both retentate and permeate streams were recycled to the feed tank. Thus, the solute concentration in feed tank was kept constant and a steady state was attained for a given set of operating condition. High molecular weight components present in the seed extract such as, polysaccharides, were assumed to be fully retained by the membrane and formed a gel type layer over the membrane surface whereas, lower molecular weight phenolic compounds permeated through UF membrane. Due to accumulation of solutes near the membrane surface, a concentration boundary layer developed from the bulk of the solution up to the gel layer which facilitated diffusion of gel forming solutes from the gel layer toward the bulk of the solution. The solute mass balance within thin stagnant concentration boundary layer over the gel layer could be obtained based on the equal rates of solute transport to the membrane surface by convection and back diffusive solute flux. The expression for steady state dimensionless permeate flux could be written as:

Sh =

factor (Pritchard et al., 1995) was included in the expression of Sherwood number (Eq. (4)) as:

1/3 (2)

where k0 was the mass transfer coefficient, de was the equivalent diameter of the flow channel and D was the effective diffusivity of solute. In this study, the permeate flux was shown to increase linearly at lower operating pressure and at higher operating pressure it was gradual. The mass transfer coefficient was assumed to be a function of transmembrane pressure drop within the range of operating pressure studied herein as, k1 = k0 (1 − a exp(− bP)), where ‘a’ and ‘b’ were the model parameters which were taken into account the effect of pressure on permeate flux in gel layer controlling ultrafiltration. Hence, including the effect of pressure, Sherwood number relation could be modified as: Shm = Sh(1 − a e−bP )

(3)

Therefore, Eq. (1) could be written as: Pew = Sh[1 − a e−bP ] ln C∗g

(4)

Since, the gel layer concentration was several order of magnitude higher than the bulk concentration it was obvious that the viscosity variation within concentration boundary layer was significant. Therefore, a Sieder–Tate viscosity correction

−bP

] ln

C∗g

m g

0.14 (5)

where m was the viscosity at the mean concentration of the solution within thin concentration boundary layer and g was the viscosity at the gel concentration (Cg ). The estimation of viscosity ratio was shown in Appendix A. An optimization method was employed to determine the values of these five unknown parameters (Cg , D, a, b, and ˇ) by minimizing the following error function as:

S=

 N  exp  vw,ss − vcal w,ss

2

exp

i=1

vw,ss

(6)

BCPOL of IMSL Math library using direct complex search algorithm was used for optimization. The results obtained were discussed in the subsequent sections.

3.

Experimental

3.1.

Materials

Dry jamun seed powder was obtained from M/s. Shree Enterprises, Maharashtra, India. Distilled water was used as the solvent for extraction process. The chemicals, 1,1-diphenyl2-picrlthydrazyl (DPPH), 2,4,6- tripyridyl-s-triazine (TPTZ) and Folin–Ciocalteu phenol reagent were purchased from Sigma–Aldrich Chemical Co. (St. Louis, MO). Gallic acid, ascorbic acid, glacial acetic acid, sodium acetate trihydrate, butylatedhydroxytoluene (BHT), potassium acetate, methanol, ferrous sulphate, sodium hydroxide were obtained from SRL (M/s. Sisco Research Laboratories Pvt. Ltd., India). All other chemicals used were of analytical grade. Six commercial flat sheet polymeric membranes were used in this work. Three of them were ultrafiltration membranes with molecular weight cut-off (MWCO) of 25, 50, and 100 kDa, which were procured from M/s Alfa Laval Pvt. Ltd., India. The permeability of the 25, 50, and 100 kDa membranes was measured using distilled water and found to be (3.5 ± 0.2) × 10−11 , (8.5 ± 0.3) × 10−11 , and (20 ± 1) × 10−11 m/Pa s respectively. The other three were nanofiltration membranes characterized by MWCO of 1000, 400, and 250 Da, which were procured from M/s. Permionics Membranes Pvt. Ltd., India. The permeability of the 1000, 400, and 250 Da membranes was found to be (2.3 ± 0.1) × 10−11 , (1.6 ± 0.08) × 10−11 , and (0.9 ± 0.05) × 10−11 m/Pa s, respectively. According to the manufacturer, all the membranes were made of polyethersulphone.

3.2.

Procedure

3.2.1. Extraction process and response surface methodology Extractions were carried out in simple laboratory Quickfit apparatus under stirring with a Teflon-coated magnetic stirrer, at 700 rpm, for predetermined temperature and time. The apparatus consisted of 2 L distillation flask and condenser. After the selected temperature was attained, the appropriate amount of dried seed powder of jamun was mixed with required volume of distilled water taken in distillation flask and the predefined extraction time started. A full factorial

32

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

central composite design with a quadratic model of Montgomery (2001) was employed to study the combined effect of three independent variables i.e., extraction temperature; extraction time; liquid to solid ratio on the four response functions such as extraction yield (Y1 , %), extract purity (Y2 , %), and the antioxidant capacity measured by DPPH radical scavenging activity (Y3 , % inhibition) and ferric reducing antioxidant power (Y4 , ␮M Fe(II)/L of extract). The range and the levels of the independent variables for extraction process were selected based on the data available in the literature. The independent variables were the incubation temperature (34.8–85.2 ◦ C), incubation time (49.8–100.2 min) and liquid to solid ratio (9.8–60.2 mL/g). These variables were considered with five levels and two star points for each of them. A total of 20 experiments with six replicates of the center points were required for this procedure. The experimental design and statistical analysis of the data were carried out using ‘Design Expert’ software (Version 8.0.7.1, Stat-Ease, Inc., Minneapolis, USA). At the end of extraction, the extract was cooled to room temperature (25 ± 1 ◦ C), and then filtered through Whatman No. 1 filter paper (M/s. Whatman International Ltd., England). Then the extract was used for estimation of yield, purity, and assessment of antioxidant activities etc. through various chemical assays. All the analyses were carried out in triplicate. Extraction yield and extract purity were defined as, Extraction yield (%) =

Weight of total solid in the extract Weight of dried seed taken for extration × 100

(7)

Concentration of total phenolic content in the extract Purity (%) = Concentration of total solid in the extract × 100

(8)

3.2.2.

Clarification and concentration of seed extract

Fig. 1a illustrates the schematic flow-chart for the concentration of jamun seed extract by using the suggested integrated membrane process. Clarification of seed extract was performed in cross flow module (Fig. 1b) using ultrafiltration membrane, followed by concentration in stirred cell (Fig. 1c) using nanofiltration membrane under various operating conditions. A stirred cell was used for selection of suitable UF and NF membranes. For selection of appropriate UF membrane, the stirred experiments were conducted in a filtration cell (capacity: 500 mL; diameter: 67 mm; effective membrane area: 32.15 cm2 ) under batch concentration mode with seed extract as a feed at operating transmembrane pressures of 207 and 345 kPa, stirring speed of 1000 rpm, for 25, 50 and 100 kDa membranes. However, for selection of suitable NF membrane, the stirred experiments were performed in a batch concentration mode with ultrafiltered extract as a feed at an operating transmembrane pressure of 828 kPa and stirring speed of 1000 rpm for 1000, 400 and 250 Da membranes. For a typical experiment, about 300 mL of jamun seed extract was charged into the batch cell (Fig. 1c). Nitrogen gas was used to generate required pressure in the test cell. The stirring speed was fixed using a variac and it was measured by a hand held digital tachometer (Agronic, India). For clarification of seed extract, cross flow experiments were performed to observe the effects of various operating conditions such as transmembrane pressure (P) and cross-flow velocity (u) on the permeate flux and permeate quality. One parameter was varied keeping the other parameter constant to get the exact picture of parameter dependence. The cross flow experiments were performed using selected ultrafiltration membrane at different transmembrane pressure (276, 414, 552 and 690 kPa) and cross flow

Fig. 1 – (a) Schematic flow-chart of the suggested integrated membrane process (UF-NF) for concentration jamun seed extract. (1) Aqueous seed extract, (2) filtrate, (3) residue, (4) retentate, (5) clarified extract, (6) concentrated extract, (7) diluted extract; (b) schematic diagram of cross flow ultrafiltration system. (1) Feed tank, (2) feed pump, (3) by pass control valve, (4) feed inlet, (5) retentate, (6) cross flow ultrafiltration cell, (7) permeate, (8) flow control valve, (9) pressure gauge, (10) rotameter; and (c) schematic of batch nanofiltration setup.

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

velocity (0.07, 0.10, 0.14, and 0.18 m/s). The corresponding Reynolds numbers were 780, 1230, 1566, and 2014. In actual experiments, aqueous seed extract was pumped from the feed tank of 8 L capacity and allowed to flow tangentially over the membrane surface through a thin channel (length: 17.4 cm; width: 7.4 cm; height: 0.63 cm) (Fig. 1b). The channel height was determined by the neoprene rubber gasket thickness. The membrane was placed on a stainless steel porous support. The retentate was recycled to the feed tank. The flow rate was measured by a rotameter in the retentate line. Pressure inside the ultrafiltration cell was maintained by operating valves at the bypass and retentate lines simultaneously and measured by a pressure gauge. Detailed parametric study were also performed for concentration of ultrafiltered seed extract in a stirred cell under batch concentration mode using selected nanofiltration membrane at different transmembrane pressure (828, 966 and 1104 kPa) and stirring speed (600, 1000 and 1400 rpm). Each ultrafiltration and nanofiltration experiment was performed at the room temperature of 25 ± 1 ◦ C for 40 min and 90 min, respectively. In the actual experiments, samples were collected from the bottom of the cell in a measuring cylinder during the experiment. Cumulative volume of permeate was plotted as a function of operation time. Values of permeate flux were determined from the slopes of cumulative volume versus time plot. The permeate flux decline (FD) as a result of fouling was expressed as:

 FD (%) =

0 − Vf Vw w 0 Vw

 × 100

(9)

0 was the initial permeate flux measured at the where Vw f was the final permeate flux meainitial moment (t0 ), and Vw sured at the end of experiment (tf ). The higher value of flux decline indicated more intense membrane fouling. Therefore, FD appeared to be an important parameter for selection of appropriate membrane. At the end of the experiment, both the permeate samples were analyzed for various physicochemical properties. The separation capability of a membrane could be established in terms of observed solute retention (R0 ) which was defined as, R0 = (1 − (CP /C0 )) × 100; where C0 and CP were the solute concentration of feed and permeate, respectively. R0 measured the fraction of solute retained by the membrane.

3.2.3.

Membrane cleaning protocol

In both ultrafiltration and nanofiltration experiments, the membranes were regenerated between experiments using the following procedure. After completion of each experiment, the cell was dismantled and the membrane was rinsed with distilled water and was submitted to chemical cleaning procedure: 30 min of acid (pH 3.0) washing using HCl, 30 min of washing using distilled water, 30 min of alkaline (pH 10) washing using 2% NaOH and finally membrane was washed with distilled water. The cell was reassembled and the membrane permeability was again measured using distilled water before the next experiment.

3.3.

33

equation, T = 10−A (Bruijn et al., 2003). The pH value was measured by a pH Meter (Mettler Toledo, AG-8603, Switzerland). For titratable acidity (TA) measurement, extract sample was titrated with 0.1 N NaOH to pH 8.3 and expressed as percentage of malic acid equivalent (Bruijn et al., 2003). Viscosity and conductivity of the extracts were measured at 25 ± 1 ◦ C by Ostwald viscometer and autoranging conductivity meter (Chemiline, CL-220, India), respectively. The content of pectinious materials present in the extracts was measured in terms of alcohol-insoluble-solids (AIS). AIS was determined by boiling 5 mL aqueous extract with 150 mL 80% ethanol, simmering for 30 min, and filtering through Whatman No. 1 filter paper. The filtered residue was washed again with 80% ethanol. The residue was dried at 100 ◦ C for 2 h and expressed in percentage by weight (Hart and Fisher, 1971). The total solid (TS) content of the sample was measured gravimetrically by heating the extract in a hot air oven at 85 ± 1 ◦ C until the difference in the weight of the extract became constant at successive intervals (Ranganna, 1986) and was used for calculation of extraction yield and extract purity.

3.3.1.

Determination of total polyphenol content (TPC)

Total polyphenol content (TPC) in the extract was measured using a modified Folin and Ciocalteu method (Singleton and Rossi, 1965; Vasco et al., 2008) based on colorimetric oxidation/reduction reaction. Folin–Ciocalteu reagent was used as oxidizing agent. Briefly, 0.5 mL aliquot (extract) was mixed with 0.5 mL of the Folin–Ciocalteu reagent in a 25 mL flask. The solutions were mixed thoroughly and incubated at room temperature (25 ± 1 ◦ C) for 3 min. After incubation, 10 mL sodium carbonate (7.5 wt%) solution was added and mixed well. The volume was then adjusted up to 25 mL with distilled water and kept at room temperature for 60 min. The absorbance was then measured at 750 nm using a UV spectrophotometer. Then the absorbance value was compared with a Gallic acid standard curve (from 50 to 500 ␮g/mL) for estimation of concentration of TPC in the extract and results were expressed as milligrams of gallic acid equivalents (GAE) per gram of seed powder on dry weight basis.

3.3.2.

Determination of total flavonoid contents (TFC)

The amount of flavonoid content in the extract was determined using aluminum chloride colorimetric method (Chang et al., 2002). The reaction mixture 3.0 mL consisted of 1.0 mL of sample (1 mg/mL), 1.0 mL methanol, 0.5 mL of (1.2 wt%) aluminum chloride and 0.5 mL (120 mM) potassium acetate and incubated at room temperature for 30 min. The absorbance of all samples was measured at 415 nm using a UV spectrophotometer. The absorbance was then compared with a Quercetin standard curve (from 12.5 to 100 ␮g/mL) for determining the concentration of TFC in the samples. The results were expressed in mg quercetin per gram of dried extract.

3.3.3.

Determination of antioxidant activities (AA)

The antioxidant capacity of seed extract was determined by two methods, DPPH free radical-scavenging activity, ferric reducing antioxidant power (FRAP) assay.

Physico-chemical parameters determination 3.3.3.1. DPPH free radical scavenging activity. The free radi-

The color of the extract was measured in terms of optical absorbance (A) at a wavelength of 420 nm using a U-2900, UV/VIS Spectrophotometer (Hitachi, Japan). Clarity of the extract was measured in terms of transmittance (T) spectrophotometrically at 625 nm which was given by the

cal scavenging activity of jamun seed extract was measured by using 2,2-diphenyl-1-picrylhydrazyl (DPPH) by the modified method of McCune and Johns (2002). In the DPPH test, the reaction mixture 3.0 mL, which consisted of aliquot (1 mL) of methanol solution of extract with a concentration of 50 ␮g/mL,

34

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

1.0 mL of DPPH (0.3 mM), and 1.0 mL of methanol, was incubated for 10 min in dark at 37 ± 1 ◦ C, after incubation the absorbance of the mixture was measured at 517 nm using a UV spectrophotometer. Ascorbic acid and BHT were used as positive control. Percentage of inhibition was calculated using the following formula: % Inhibition =

B − A B

× 100

(10)

where B was the absorbance of blank (DPPH and methanol); A was the absorbance of extract sample (DPPH, methanol and sample). The IC50 value was calculated from the linear regression of the % inhibition curves obtained for all extract concentrations. The assay was carried out in triplicate and values were presented as mean ± standard deviation.

3.3.3.2. Ferric reducing antioxidant power (FRAP). FRAP assay measured the ability of the antioxidants in the extract which was based on the reduction of ferric-tripiridyl-triazine complex [Fe+3 -TPTZ] to the ferrous form [Fe+2 -TPTZ] at an acidic condition. [Fe+2 -TPTZ] was an intense blue color complex with a maximum absorption at 593 nm (Gohari et al., 2011). The FRAP reagent was prepared by mixing 2.5 mL of a 10 mM/L TPTZ solution in 40 mM/L hydrochloric acid with 2.5 mL of a 20 mM/L FeCl3 solution and 25 mL of a 0.3 M/L acetate buffer (pH 3.6). A total of 5 mL of extract with a concentration of 50 ␮g/mL was mixed with 4.5 mL of FRAP reagent. The absorbance was measured after 4 min of incubation in the dark at room temperature (25 ± 1◦ C) at 593 nm by UV spectrophotometer using water as a blank. The absorbance was then compared with a calibration curve of aqueous solution of ferrous sulphate (FeSO4 ·7H2 O; 25–800 ␮M). Results were of triplicate analyses and the FRAP values were expressed in ␮M Fe(II)/L.

4.

Results

Optimization and experimental validation of 4.1. optimized condition The optimization of extraction process parameters was carried out in Design Expert Software using numerical optimization technique by selecting the desired goals (maximum extraction yield, maximum purity and maximum antioxidant activities). The goals were combined into an overall desirability function. The optimization started at a random point in the design space and proceeds up the steepest slope to a maximum to find a point that maximized the desirability function. The optimal extraction conditions were found to be: extraction temperature, 49.2 ◦ C; extraction time, 89.4 min; liquid to solid ratio 51.6:1 mL/g with desirability value of 0.96. Under the optimized conditions, the predicted values were as follows: Y1 (15.3 ± 0.72)%, Y2 (43.5% ± 2.06)%, Y3 (77.7 ± 5)% and Y4 (665 ± 39) ␮M Fe(II)/L. To confirm the values obtained from the numerical optimization, triplicate experiments were performed under the optimized conditions. The measured values of the Y1 (18.9 ± 0.5)%, Y2 (39.2 ± 0.78)%, Y3 (75.5 ± 1.8)% and Y4 (720 ± 42) ␮M Fe(II)/L were in good agreement with the predicted values. The analysis of variance of four responses Y1 , Y2 , Y3 and Y4 showed that the model F values of 10.31, 12.29, 14.17, and 14.01, respectively were highly significant (p < 0.0001). The values of coefficient of determination (R2 ) of Y1 , Y2 , Y3 and Y4 were found to be 0.903, 0.917, 0.927, and 0.926 respectively. The

coefficient of variation (CV) of Y1 , Y2 , Y3 and Y4 were found to be 5.1%, 5.82%, 9.07% and 6.88% respectively.

4.2.

Clarification of seed extract

4.2.1.

Selection of ultrafiltration membrane

Selection of suitable ultrafiltration membrane was very important for clarification of aqueous jamun seed extract. The technical performance of three membranes with different MWCO (100, 50 and 25 kDa) was characterized in terms of permeate flux, flux decline, recovery of total polyphenol content, and extract clarity. At the end of 80 min operation, the obtained values of permeate flux and flux decline at transmembrane pressure of 207 kPa and 345 kPa, for 25, 50 and 100 kDa membrane were shown in Fig. 2, respectively. This indicated that about 2.2–2.4-fold enhancement in permeate flux was observed in case of 100 kDa membrane with slightly higher value of FD than 50 kDa membrane. Concerning the clarity, no significant differences were observed among the products obtained with the three membranes, and were found to be 98–99% compared to 84.3% in the feed. Photographs of seeds extract obtained from different UF membrane were shown in the inset of Fig. 2a.

4.2.2. Cross flow ultrafiltration 4.2.2.1. Estimation of model parameters and prediction of steady state permeate flux. As discussed in Section 2, the steady state permeate flux of ultrafiltration of seed extract was calculated using the modified film theory equation (Eq. (5)) including the effects of transmembrane pressure and viscosity. This calculation involved estimation of five unknown parameters, namely, effective diffusivity of gel forming material, gel concentration and the model parameters (a, b, and ˇ). These five parameters were estimated by the optimization technique as outlined in Section 2. The estimated values of effective diffusivity, gel concentration were found to be (9.78 ± 0.4) × 10−11 m2 /s; (56.9 ± 1.0) kg/m3 respectively. The model parameters a, b and ˇ were found to be (7.66 ± 0.4), (1.08 ± 0.1) × 10−2 kPa−1 , and (5.37 ± 1.0) × 10−3 , respectively. The estimated values of effective solute diffusivity and gel concentration were in the same order of magnitude with the results reported in literature (Mondal et al., 2012; Sarkar et al., 2008). The comparison between experimental and predicted steady state permeate flux values obtained under various operating conditions of cross-flow velocity and transmembrane pressure was presented in Fig. 3. A good agreement is obtained. The maximum percent error, defined in Eq. (B1), was found to be ±6.5% whereas, the standard deviation (SD), defined in Eq. (B2), was within ±3.54%.

4.2.2.2. Estimation of various resistances during ultrafiltration under steady state. The effects of various operating condition on various resistance during ultrafiltration of seed extract were shown in Table 1. The calculation procedure for estimation of various resistances during ultrafiltration of seed extract under steady state operation was discussed in detail by Sarkar (2014). It was observed from the table that both gel layer resistance (Rg ) and concentration polarization resistance (RCP ) decreased with increase in Reynolds number and decrease in transmembrane pressure difference. The contribution of concentration polarization resistance to the total resistance (Rcp /Rt ) was about 30% and 36% at 276 kPa and 690 kPa, respectively, for a fixed Reynolds number of 780. However, with the same increment in transmembrane

35

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

Fig. 2 – Treatment of phenolic extracts with three different UF membranes under batch concentration mode. (a) Variation of permeate flux with time for different pressure at a fixed stirrer speed of 1000 rpm (inset: 1, 2, 3 and 4 are for feed and permeate of 100 kDa, 50 kDa and 25 kDa membranes, respectively). (b) Variation of permeate flux, FD, TPC recovery and permeate clarity with MWCO (ω = 1000 rpm and P = 207 kPa), (c) variation of permeate flux, FD, TPC recovery and permeate clarity with MWCO (ω = 1000 rpm and P = 345 kPa). Error bars represent standard deviation for n = 3. pressure, the contribution of reversible fouling resistance (Rrev ) to the total resistance (Rt ) increased from 92.46% to 95.17%. Further, at 414 kPa, with increasing Reynolds number from 780 to 2014, Rcp /Rt decreased by only 3%. The irreversible

component of the fouling resistance (Rirr ) was found to be only (0.34–0.6)% of the total resistance indicating that apart from freely permeable solutes such as small polyphenols, and acid, there were some particles with smaller size than the membrane pore size in the extract (small fibers, large polyphenols, etc.) which entered into membrane pores causing irreversible fouling during ultrafiltration. However, change in irreversible fouling resistance with the operating condition was found to be insignificant. The dimensionless gel layer resistance and concentration polarization resistance were correlated with the operating conditions as: ∗ RCP = 2.395(P)

Rg∗ = 6.44(P)

Fig. 3 – Comparison between predicted and experimental steady state permeate flux during ultrafiltration of seed extract using optimized values of effective solute diffusivity, gel concentration and model parameters involved in pressure and viscosity correction.

0.614

0.389

 1 0.43 Re

 1 0.283 Re

(R2 = 0.96)

(R2 = 0.91)

(11)

(12)

where P is in kPa. High correlation coefficient (R2 ) for the above equations indicated that the above expressions effectively described the effects of operating conditions on both gel layer resistance and polarization resistance.

4.2.2.3. Effect of various operating conditions on permeate flux and properties of clarified extract. Fig. 4 illustrates the variation of steady state permeate flux for different Reynolds number at

36

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

Table 1 – Estimation of various resistances during ultrafiltration of S. cumini (L.) seed extract at various operating conditions (Rm = 5.8 × 1012 m−1 ). P (kPa)

Re

Rirr (×1012 ) (m−1 )

Rrev (×1012 ) (m−1 )

RCP (×1012 ) (m−1 )

Rg (×1012 ) (m−1 )

Rm /Rt (%)

Rirr /Rt (%)

RCP /Rt (%)

Rrev /Rt (%)

276

780 1230 1566 2014

0.38 0.38 0.38 0.38

76.10 68.45 63.58 56.75

24.70 21.64 19.53 16.98

51.40 46.81 44.05 39.77

7.04 7.77 8.31 9.22

0.46 0.50 0.54 0.60

30.01 28.99 27.99 26.99

92.46 91.71 91.12 90.22

414

780 1230 1566 2014

0.40 0.40 0.40 0.41

84.63 76.80 69.01 61.60

29.06 25.73 22.56 19.66

55.57 51.90 47.21 40.59

6.38 6.98 7.71 8.55

0.44 0.48 0.53 0.60

31.99 31.01 30.01 28.99

93.17 92.53 91.75 90.85

552

780 2014

0.43 0.42

104.4 76.07

38.73 24.70

65.67 51.37

5.23 7.04

0.38 0.51

34.98 30.01

94.37 92.43

690

780 2014

0.45 0.44

123.1 89.27

46.58 30.44

76.52 58.83

4.48 6.09

0.34 0.46

35.99 31.99

95.17 93.82

276 kPa and 414 kPa. From the figure it was seen that permeate flux increased with increase in Reynolds number. Similar trends were observed at both the operating pressures. Experimental flux values showed repeatability to within ±3% and were shown in this figure with error bar. It was evident from the figure that an increase in Reynolds number from 780 to 2014 led to an increase in permeate flux from 14 to 19 L/m2 h (35.7%) at 276 kPa and 19 to 25.6 L/m2 h (34.7%) at 414 kPa, respectively. The variations of the steady state permeate flux with transmembrane pressure drop for various Reynolds numbers were shown in Fig. 5. It was observed from the figure that permeate flux increased linearly at lower operating pressure and was gradual at higher operating pressure. For example, with an increase in transmembrane pressure from 276 to 690 kPa, the permeate flux increased from 14 to 22.3 L/m2 h and 18.4 to 30.2 L/m2 h at Reynolds number 780 and 2014, respectively. This result was in accordance with the study of Chhaya et al. (2012), where they reported a similar flux enhancement behavior with increase in transmembrane pressure during cross flow ultrafiltration of stevia extract. The effects of various operating conditions (transmembrane pressure and Reynolds number) on extract purity and polyphenol recovery were presented in Table 2. Experimental results shown that with increase in transmembrane pressure from 276 to

690 kPa, the extract purity and polyphenol recovery (averaged over all Reynolds numbers studied herein) decreased from 58.3 to 37.6% and 63 to 50.3%, respectively.

Fig. 4 – Variations of steady state permeate flux with Reynolds number for different transmembrane pressure drop during ultrafiltration of seed extract (error bars represent standard deviation for n = 3).

Fig. 5 – Variations of steady state permeate flux with transmembrane pressure drop for different Reynolds number during ultrafiltration of seed extract (error bars represent standard deviation for n = 3).

4.3.

Concentration of clarified seed extract

4.3.1.

Selection of nanofiltration membrane

It was apparent that most appropriate nanofiltration membrane was required to be selected before detailed parametric study of concentration of UF clarified seed extract. The technical performance of three NF membranes with different MWCO (100,400 and 250 Da) was characterized in terms of permeate flux, flux decline, rejection of polyphenols in retentate, and extract clarity. As seen in Fig. 6a, at the end of 80 min operation, the permeate flux values were 23.10 L/m2 h, 20.5 L/m2 h and 19.3 L/m2 h at 828 kPa for 1000, 400 and 250 Da membrane, respectively. The NF membrane with the lowest MWCO showed the highest rejection toward total polyphenol content. As observed from Fig. 6b that the rejection of total polyphenols in retentate decreased in the following order: 250 Da > 400 Da >1000 Da. The 250 Da membrane showed the observed rejection toward TPC of 78% while the rejection for the 1000 Da membrane was about 61%. Tylkowski et al. (2011) found similar trends in the evaluation of NF membranes for the retention of polyphenols and flavonoids during concentration of ethanolic extracts from Sideritis ssp. L. The flux decline was observed

37

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

Table 2 – Various properties of phenolic extract at different operating conditions during ultrafiltration (average values of three samples ± standard deviation). P (kPa)

Re

Color A420nm

Clarity %T660nm

Total solid (mg/L)

Purity of polyphenol (%)

Recovery of polyphenol (%)

276

780 1230 1566 2014

0.527 0.459 0.444 0.411

± ± ± ±

0.02 0.01 0.01 0.01

90.2 96.5 95.1 94.1

± ± ± ±

1.80 1.93 1.90 1.88

1567 1678 1745 1835

± ± ± ±

31.3 33.6 35.0 36.8

58.3 55.3 54.0 52.7

± ± ± ±

1.20 1.10 1.08 1.05

63.0 64.0 65.0 66.7

± ± ± ±

1.26 1.28 1.30 1.33

414

780 1230 1566 2014

0.454 0.419 0.393 0.408

± ± ± ±

0.01 0.01 0.01 0.02

96.4 96.1 96.4 96.4

± ± ± ±

1.92 1.92 1.92 1.93

1579 1676 1645 1775

± ± ± ±

31.6 33.5 32.9 35.5

56.0 54.5 52.0 49.0

± ± ± ±

1.12 1.09 1.04 0.98

61.0 63.0 59.0 60.0

± ± ± ±

1.22 1.26 1.18 1.20

552

780 1230 1566 2014

0.295 0.295 0.288 0.302

± ± ± ±

0.01 0.01 0.01 0.01

97.0 96.8 96.7 96.3

± ± ± ±

1.94 1.93 1.93 1.92

1817 1772 1702 1635

± ± ± ±

36.3 35.4 34.1 32.7

45.5 45.0 46.0 47.0

± ± ± ±

0.91 0.90 0.92 0.94

57.0 55.0 54.0 53.0

± ± ± ±

1.14 1.10 1.08 1.06

690

780 1230 1566 2014

0.298 0.296 0.302 0.302

± ± ± ±

0.01 0.01 0.02 0.02

96.0 97.0 96.8 96.8

± ± ± ±

1.92 1.94 1.93 1.93

1940 1762 1662 1483

± ± ± ±

38.8 35.2 33.2 29.6

37.6 39.5 41.0 43.0

± ± ± ±

0.75 0.79 0.82 0.86

50.3 48.0 47.0 44.0

± ± ± ±

1.01 0.96 0.94 0.88

S. cumini (L.) seed extract

1.150 ± 0.10

85.6 ± 1.71

3700 ± 74.0

39.2 ± 0.78

Fig. 6 – Treatment of phenolic extracts with three different NF membranes under batch concentration mode. (a) Variation of permeate flux with time (inset: 1 and 1 , are for retentate and permeate, respectively, for 1 kDa membrane; 2 and 2 are for retentate and permeate, respectively, for 400 Da membrane; 3 and 3 are for retentate and permeate, respectively, for 250 Da membrane). (b) Variation of permeate flux, FD, TPC rejection and permeate clarity with MWCO (ω = 1000 rpm and P = 828 kPa). Error bars represent standard deviation for n = 3. as 48.5%, 37% and 33% for 1000, 400 and 250 Da membrane, respectively (Fig. 6b). All the NF membranes showed high retentate clarity of about 95–99%.

4.3.2. Effect of various operating conditions on permeate flux and properties of concentrated extract Fig. 7 illustrates the transient behavior of permeate flux and volume concentration ratio (VCR) with time at different operating transmembrane pressure for fixed value of stirrer speed. VCR was defined as the ratio of initial feed volume to the retentate volume at any time of operation. Since the retentate volume decreased continuously with the progress of the operation as permeate was taken out without recycle, VCR continued to increase with time of operation. It could be observed from Fig. 7 that at the end of 90 min operation, with increase in transmembrane pressure from 828 to 966 kPa, permeate flux increased from 16.6 to 24.15 L/m2 h with increase in VCR from 1.61 to 2.1. In addition the flux decline values were found to be 38%, 41.2% and 44% at 828 kPa, 966 kPa, and

Fig. 7 – Time course of permeate flux and VCR for different transmembrane pressure drops during nanofiltration of seed extract under batch concentration mode.

38

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

Fig. 8 – Time course of permeate flux and VCR for different stirrer speeds during nanofiltration of seed extract under batch concentration mode.

Fig. 9 – TPC in NF retentate and required operation time as a function of VCR (error bars represent standard deviation for n = 3). 1104 kPa, respectively. The profiles of permeate flux and VCR with time at various stirrer speeds for fixed value of transmembrane pressure were shown in Fig. 8. After 90 min of processing, the permeate flux declined from 32 to 18.3 L/m2 h (about 43%) corresponding to a final VCR value of 1.61, for fixed pressure (966 kPa) and stirrer speed (600 rpm). Furthermore, at the end of the operation, at 966 kPa, with an increase in stirrer speed from 600 to 1400 rpm, permeate flux increased from 18.3 to 21 L/m2 h with increase in VCR from 1.61 to 1.8. TPC in the NF retentate obtained at different VCRs were significantly different from that of feed extract (Fig. 9). The concentration of phenolic compounds in the NF retentate increased with increase in VCR. In particular, TPC increased by 3.2-fold with observed retention of about 0.8 (data not shown) after 160 min

of NF processing for fixed pressure (966 kPa) and stirrer speed (1000 rpm), when VCR was 3. These results were better than those observed by Murakami et al. (2011) in the concentration of phenolic compounds through nanofiltration of aqueous mate, achieving 2.5-fold increase in retentate concentration at same VCR. This study indicated that the seed extract could be concentrated by nanofiltration in order to increase the phenolic contents accompanied by reduction in volume and the concentrate could be diluted to the desired phenolic concentration (antioxidant activity), if required. It was further observed from Table 3 that the TPC in NF permeate decreased with increase in operating pressure. For example, with increase in transmembrane pressure from 866 to 1104 kPa, TPC in permeate decreased from 190 to 145 mg GAE/L, indicating that the loss of TPC decreases from 20.16 to 15.4%. However, no significant effect of operating pressure on retentate purity was observed. Also, no significant change in total polyphenol concentration in permeate as well as purity of the retentate were noticed with change in stirrer speed. The average value of purity in the NF retentate for the range of operating condition studied herein was about 54.2% with average polyphenols retention of about 83%.

5.

Discussion

5.1.

Clarification of seed extract

Fig. 2a presents the transient behavior of the permeate flux obtained in the UF treatment of the aqueous seed extract for three different MWCO. Similar trends were observed for all UF membranes at both the operating pressure. The sharp decline in permeate flux observed at the beginning of filtration could be attributed to the rapid pore blocking and increase in concentration polarization. As the operation time progressed, there was more accumulation of solutes over the membrane surface leading to severe concentration polarization. The gradual flux decline at later stages could be attributed to the formation of gel type layer on the membrane surface. It was further noticed that at any point of time, the permeate flux increased with increase in both transmembrane pressure and membrane MWCO keeping other operating conditions unchanged. The recovery of UF membranes toward the polyphenols increased by increasing the MWCO of the membranes. The highest recovery of TPC obtained from 100 kDa membrane was about 56–60%, while lowest recovery of about 30–32% was achieved for 25 kDa membrane. This was attributed to the fact that lower cut-off membrane (25 kDa) rejected most of the higher molecular weight solutes due to its smaller pore size and form the dynamic cake layer over it which retained some more polyphenols, resulting in low

Table 3 – Various properties of phenolic extract during nanofiltration under batch concentration mode at different operating conditions at VCR = 2 (average values of three samples ± standard deviation). P (kPa)

966 828 966 1104 NF feed

Stirring speed (rpm)

600 1400 1000 –

Total solid in permeate (mg/L)

310 350 320 440 270 1778

± ± ± ± ± ±

6.2 7.0 6.4 8.8 5.4 35.6

Total polyphenols in permeate (mg GAE/L)

Total solid in retentate (mg/L)

Total polyphenols in retentate (mg GAE/L)

± ± ± ± ± ±

2641 ± 52.82 2545 ± 50.90 2830 ± 56.60 2486 ± 49.72 2425 ± 48.50 ND

1400 ± 28.0 1400 ± 28.0 1500 ± 30.0 1400 ± 28.0 1300 ± 26.0 ND

164 155 190 170 145 942.5

3.3 3.1 3.8 3.4 2.9 18.9

Observed deviation of TPC (mg) 42.6 ± 0.8 41.2 ± 0.8 24.2 ± 0.5 39.4 ± 0.6 55.0 ± 1.1 ND

Observed deviation of TS (mg)

85.9 ± 1.7 82.9 ± 1.6 51.0 ± 1.1 79.0 ± 1.6 107 ± 2.1 ND

Overall Polyphenols polyphenols retention in purity (UF + NF) retentate (%) (%) 53.0 55.0 53.0 56.3 53.6 53.0

± ± ± ± ± ±

1.06 1.10 1.06 1.12 1.07 1.06

82.6 ± 1.65 83.5 ± 1.67 79.8 ± 1.59 81.9 ± 1.64 84.6 ± 1.69 ND

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

recovery of polyphenols in permeate. This result was in accordance with experimental data reported by Galanakis et al. (2013) where they reported a similar trends of total polyphenol recovery in the permeate (36% and 19% with 100 kDa and 20 kDa polysulfone membrane, respectively) during fractionation of phenolic compounds recovered from winery sludge by UF. From the experimental results it was clear that 25 kDa membrane showed worst performance for the clarification of seed extract while 100 kDa membrane exhibited better performance with highest values of permeate flux and polyphenols recovery compared to 50 kDa membrane. Considering both the recovery of total polyphenols of interest in the permeate and absolute values of permeate flux, 100 kDa membrane was preferred among the UF membranes and hence selected for detailed parametric study of clarification of seed extract. The increase of permeate flux with Reynolds number (Fig. 4) could be explained by the fact that concentration polarization was less at higher Reynolds number due to forced convection imposed by cross-flow velocity. This was a direct consequence ∗ with increasing Reynolds numof decreasing both Rg∗ and RCP ber (Table 1). Similar behavior was observed by Chhaya et al. (2012) in the clarification of stevia extract by cross flow ultrafiltration when using cross flow rate between 60 and 120 L/h. In an ideal gel controlled filtration, permeate flux exhibited a pressure independence indicating that the mass transfer coefficient was independent of transmembrane pressure difference (Cheryan, 1998). However, in several studies during filtration of gel forming solute, permeate flux was reported to be pressure dependent as also observed in the present study indicating the pressure dependence of mass transfer coefficient (Gekas and Hallstrom, 1987; Van Den Berg et al., 1989). Recently, modeling of the ultrafiltration process was performed considering mass transfer coefficient as a linear function of transmembrane pressure drop (Mondal et al., 2012). However, in the present study, the permeate flux was shown to increase linearly at lower operating pressure and at higher operating pressure it was gradual (Fig. 5). At lower operating pressure, concentration polarization was less. Hence, with increase in operating pressure (at a fixed Reynolds number), the driving force across the membrane increased leading to a linear enhancement in permeate flux. Nonetheless, at higher operating pressure, accumulation of solutes near the membrane surface increased due to increase of solute convective flux toward the membrane. This led to an increase in concentration polarization followed by the formation of thicker gel type layers over the membrane surface which in turn increased the resistance to solvent flow. This was a direct ∗ with increasing consequence of increasing both Rg∗ and RCP transmembrane pressure (Table 1). Nevertheless, permeate flux was found to increase gradually at higher operating pressure since the enhanced driving force due to pressure overcame the resistance of the developed gel type layer. The effect of transmembrane pressure was much more pronounced on extract purity and polyphenol recovery compared to cross flow velocity as observed in Table 2. Experimental results showed that with increase in transmembrane pressure, both extract purity and polyphenol recovery (averaged over all the cross flow velocity) decreased. This might be explained by the fact that at higher operating pressure, the gel type foulant layer formed by the rejected higher molecular weight (mostly polysaccharides) over the membrane surface got compressed and acted as an additional membrane which in turn retained some more polyphenols. This led to a decrease in recovery of polyphenols in the permeate side accompanied

39

by lower values of color. The lower extract purity obtained at higher operating pressure could be explained by increasing the permeation of smaller molecular weight solutes other than polyphenols with increase in pressure. This resulted to a decrease in purity as the purity depended on relative ratio of phenolic and non-phenolic compounds in the extract. Hence, purity of the extract was found to decrease with increase in transmembrane pressure. Similar types of observations were available in the literature (Chhaya et al., 2012; Vanneste et al., 2011).

5.2.

Concentration of clarified seed extract

Experimental results shown in Fig. 6b indicates that 250 Da membrane performed better for the concentration of phenolic extract with minimum loss of TPC in the permeate side and minimum flux decline. Although 1000 Da membrane possessed higher permeate flux but it exhibited worst performance with respect to loss of polyphenols and flux decline. Considering both the maximum rejection toward polyphenols in the retentate and the minimum flux decline, 250 Da membrane was found most promising and hence was selected for detailed parametric study for concentration of UF clarified seed extract. During nanofiltration of clarified extract a significant flux decline was noticed. As observed in Figs. 7 and 8, the sharp decline in permeate flux at the beginning of the operation could be attributed to the rapid buildup of concentration polarization at the membrane–liquid interface. This similar behavior was observed by Conidi et al. (2011) in the recovery of TPC bergamot juice using nanofiltration process where they reported about 55% decline in permeate flux after 80 min operation. The smoother and slower flux decline at later stages could be attributed to the solute deposition on the membrane surface. As evident from Fig. 7 that both permeate flux and VCR increased with increase in transmembrane pressure. This might be explained by the fact that the enhanced driving force for solvent flux due to increase in pressure was more than the resistance offered by the concentration polarization and membrane fouling. Hence, at higher operating pressure, higher permeate flux was obtained with higher value of VCR. In the batch concentration mode of operation, the permeate stream was continuously taken out. This resulted to an increase in feed concentration accompanied by a reduction in feed volume with the progress of operation. As the feed concentration increased, the concentration polarization became more severe which increased the resistance to solvent flow and hence permeate flux declined. It was further noticed that both permeate flux and VCR increased with increase in stirrer speed (Fig. 8). This was due to the fact that at higher stirrer speed, growth of the concentration boundary layer was arrested due to enhanced turbulence close to the membrane surface, leading to an increase in mass transfer coefficient. Hence, with increase in stirrer speed, more permeate flux was obtained with higher value of VCR. The obtained flux values were in accordance with the results reported earlier (Xu and Wang, 2005; Mello et al., 2010). Though the data obtained in this study (Table 3) revealed that the concentration of total polyphenols in the NF retentate decreased with increasing operating pressure. This could be explained by the fact that increasing operating pressure had two effects. Firstly, increased pressure (at a fixed Reynolds number) enhanced the driving force across the membrane leading to an enhancement of solvent flux. Secondly, increased pressure also facilitated solute convective

40

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

flux toward the membrane resulting in higher rates of solute accumulation near the membrane surface leading to more severe concentration polarization and consequently membrane surface concentration was higher for higher pressure. Lower solute concentration in permeate side induced a large concentration difference across the membrane resulting to an increase in solute flux through the membrane and therefore permeate concentration increased. In this study within the range of operating pressure, the first effect dominated. Therefore, the solvent flux was more than the solute flux resulting in decrease in polyphenol concentration in permeate with increase in pressure. The deviation from the mass balance with respect to TPC and TS was also reported in Table 3. The solute mass balance in the batch cell was expressed as: C0 V0 = CR VR + CP VP

(13)

where C0 , CP and CR were the solute concentration at feed, permeate and retentate, respectively. The mass balance of TPC and TS for various operating conditions was carried out using Eq. (13) and it was observed that they were nor equilibrate. The observed deviation of TPC and TS estimated from mass balance equation (Eq. (13)) were found to be transmembrane pressure dependent. As observed from Table 3, at 1000 rpm, with increase in transmembrane pressure from 828 to 1104 kPa, the observed deviation of TPC and TS increased from 10.3 to 23.3% and 11.4 to 24.2%, respectively. This results could be correlated to the solute–membrane interaction as well as severe concentration polarization leading to adsorption on or in the membrane pore walls and deposition over the membrane surface. However, no significant effect of stirrer speed on the extent of deposition was observed.

5.3.

Physico-chemical properties of extract

Table 4 represents the results of physico-chemical properties of samples of fresh, clarified and concentrated seed extract. The TS, extract purity, TPC, TFC, TA, AIS and AA levels in the extract, as the most important parameters of this study, were significantly affected by the membrane application. The absence of AIS content in the UF permeate confirmed that the 100 kDa membrane retained all the pectin substances that could improve the clarity of the extract from 84.3 to 96.2%. A significant decrease of about 35% and 36% in TPC and TFC,

respectively, were observed in the UF clarified extract. This might occur because of adsorption of polyphenols on or in the membrane pore walls as well as its retention within the gel type layer formed by the rejected solutes over the membrane surface. UF clarification significantly enhanced the purity of the extract from (39.2 ± 0.78) to (53 ± 1.06)%. This enhancement in purity might have occurred because of removal of AIS which was co extracted along with polyphenols extraction. In the UF permeate, a reduction of 32% toward the total acidity was observed in comparison with the feed. The conductivity of the extract was related to the concentration of mineral salts, organic acids, pectins, etc. As observed in Table 4, conductivity of the extract decreased about 30–38% after membrane treatment. Concentration by nanofiltration also significantly affected the TPC and TFC in the retentate. After NF concentration (at VCR 2), about 96.55% of TPC and 78.5% of TFC were restored in comparison with the original extract while TA increased by 1.2-fold. The pH and viscosity of original extract were found to be 4.98 and 0.85 × 10−3 Pa s, respectively, which remained almost unchanged after UF clarification as well as NF concentration. During NF concentration, the extract purity increased from (53 ± 1.06) to (56.3 ± 0.78)%. The total polyphenol content in the NF retentate (72.3 mg GAE/g seed or 563 mg GAE/g dry extract) was comparable with the values reported for other plant seeds (Soong and Barlow, 2004; Al-Farsi and Lee, 2008). The antioxidant activity of the seed extract could be attributed to the presence of various polyphenols with antioxidant properties and its purity. Fig. 10a shows the antioxidant activity (DPPH radicalscavenging activities and ferric reducing power) as a function of the extract concentration and Fig. 10b represented antioxidant activity (DPPH radical-scavenging activities) as a function of standard compounds concentration. Linear regression analyses showed a positive correlations between phenolic content and the ferric reducing activities (R2 = 0.98); phenolic content and DPPH radical-scavenging capacities (R2 = 0.91–0.99). This suggested that the observed antioxidant properties of extract were mainly contributed by the phenolics in the plant extracts (Kaur and Kapoor, 2000). The antioxidant activity was also presented in terms of IC50 shown in Table 4. The difference in antioxidant capacity values for NF feed and retentate was in agreement with the high rejection of phenolic compounds observed through nanofiltration membrane. The NF retentate showed higher antioxidant activity (94.6 ± 1.6% of

Table 4 – Physico-chemical properties of S. cumini (L.) seed extract (feed), UF permeate, NF permeate and NF retentate. Values are expressed in mean ± standard error of the mean (n = 3). Properties TPC (mg/L) TFC (mg/L) AIS (mg/L) Total solid (mg/L) Purity (%) DPPH (%I)a IC50 (␮g/mL) FRAP (␮M FeSO4 )a Color (A420nm ) Clarity (%T625nm ) pH Conductivity (mS) Viscosity (×103 ) (Pa s) Total acidity (mg/L) a

S. cumini (L.) seed extract

UF permeate

NF retentate

NF permeate

1450 ± 29.0 153 ± 3 697.5 ± 3.95 3700 ± 74.0 39.20 ± 0.78 75.5 ± 1.8 35.44 ± 0.72 430 ± 12 1.204 + 0.03 84.30 ± 1.70 4.98 ± 0.09 1.3 ± 0.03 0.843 ± 0.05 530 ± 10.6

942.5 ± 18.9 98 ± 2 ND 1778 ± 35.6 53.00 ± 1.06 90.2 ± 1.5 25.70 ± 0.51 510 ± 15 0.424 ± 0.01 98.20 ± 1.94 4.98 ± 0.09 0.7 ± 0.01 0.861 ± 0.05 360 ± 7.20

1400 ± 28.0 120 ± 3 ND 2487 ± 49.74 56.30 ± 1.13 94.6 ± 1.6 20.64 ± 0.47 530 ± 10 0.881 ± 0.02 96.0 ± 1.7 4.73 ± 0.08 0.90 ± 0.02 0.867 ± 0.02 630 ± 12.6

190 ± 3.80 ND ND 300 ± 6.0 ND ND ND ND 0.021 ± 0.01 98.9 ± 1.9 6.10 ± 0.12 0.2 ± 0.01 0.85 ± 0.02 330 ± 6.6

DPPH and FRAP are measured at an extract concentration of 50 ␮g/mL. ND: not detected.

41

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

Fig. 10 – Variation of antioxidant activity (DPPH radical-scavenging activities and ferric reducing power) as a function of concentration (a) S. cumini (L.) seed extract [(A) DPPH of feed (y = −12.06 + 1.752x, R2 = 0.98), (D) FRAP of feed (y = −13.15 + 8.954x, R2 = 0.98); (B) DPPH of UF permeate (y = 7.481 + 1.654x, R2 = 0.91); (C) DPPH of NF retentate (y = 18.61 + 1.52x, R2 = 0.99)]. (b) Standard compounds (ascorbic acid and BHT) [(F) DPPH of BHT (y = 15.657 + 1.131x, R2 = 0.97); (G) DPPH of ascorbic acid (y = 16.232 + 2.373x, R2 = 0.98)]. The error bars represent standard deviation for n = 3. inhibition by DPPH method and 530 ± 10 ␮M Fe(II)/L by FRAP method) than the UF permeate (75.5 ± 1.8% of inhibition by DPPH method and 430 ± 10 ␮M Fe(II)/L by FRAP method), as a consequence of the higher TPC with increasing extract purity. IC50 values of the initial feed extract, UF clarified extract and NF concentrate were estimated as (35.44 ± 0.72), (25.70 ± 0.51), (20.64 ± 0.47) ␮g/mL, respectively. The IC50 value of standard, ascorbic acid was found to be (12.23 ± 1.11) ␮g/mL while that of standard, BHT is (23.63 ± 1.65) ␮g/mL. The aqueous feed extract presented lower antioxidant activity when compared with the standards. However, the radical scavenging activities of NF concentrate exhibited better than that of standard, BHT. The higher antioxidant activity, thereby the lower IC50 value of NF concentrate demonstrated well that the NF concentrated seed extract possessed outstanding antioxidant activity.

6.

Conclusions

The investigation suggested that by integration of ultrafiltration and nanofiltration membrane processes, concentrated aqueous jamun seed extract with high antioxidant activity and purity was possibly produced which could be used as potential sources of natural antioxidants for application in food and pharmaceutical industry. Experimental results revealed that purity of the UF clarified phenolic extract was increased from 39.2 to 53% with increase in DPPH free radical scavenging capacity from 75.5 to 90.2%, in comparison with the feed. The steady state performance during of seed extract was successfully modeled by modified film theory. Furthermore, the UF clarified extract, with an initial total polyphenol content of 942 mg GAE/L, was concentrated by nanofiltration up to a factor of 3.2 with IC50 = 20.6 ␮g/mL and purity (about 56%) within 160 min of operation at pressure of 966 kPa and stirrer speed of 1000 rpm. The parametric study also confirmed the significant enhancement in permeate flux with increase in transmembrane pressure and cross-flow velocity (or stirrer speed). The trends were consistent with the physical understanding of the process. According to the results obtained, integrated membrane process offered a good preservation of phenolic compounds in the jamun seed extract in which thermal and mechanical stress on the extract is sturdily reduced in comparison with conventional techniques. Nonetheless,

further studies are required to make the process economically feasible at industrial scale.

Acknowledgement The authors sincerely thanked the Council of Scientific and Industrial Research, India (under grant – CSIR38/(1392)/14/EMR-II) for financial support of this research project.

Appendix A. An exponential relationship was used to account for the variation of viscosity with concentration in the boundary layer as,  = 0 exp(ˇC), where, ˇ was a constant and 0 was the viscosity of solvent. Therefore, the viscosity at the gel layer became g = 0 exp(ˇCg ). The following quadratic concentration profile within the thin concentration boundary layer over the membrane surface was assumed: C = a1 + a2

y ıc

+ a3

 y 2

(A1)

ıc

Eq. (A1) satisfied the following boundary conditions: at y = ıc ,

C = Cb

(A2)

at y = ıc ,

∂C =0 ∂y

(A3)

at y = 0,

C = Cg

(A4)

With the help of the above boundary conditions Eq. (A1) could be written as: C = Cg − 2(Cg − Cb )

y ı

 y 2

+ (Cg − Cb )

ı

(A5)

In evaluating solution viscosity at mean concentration (m ), the mean concentration (Cm ) in the concentration boundary layer was required to be calculated. The mean concentration (Cm ) could be evaluated as: Cm =

1 ı



ı

C dy = 0

Cg − 2Cb 3

(A6)

42

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

Hence, the expression of viscosity at mean concentration (m ) was

 m = 0 exp

ˇ(Cg − 2Cb ) 3

 (A7)

The ratio of viscosity at mean concentration to the viscosity at gel concentration could be written as: m = exp g

  2  −

3

ˇ(Cg − Cb

 (A8)

Appendix B. Percent error and standard deviation for permeate flux was calculated using the following expressions:

 Error (%) =

exp

cal Vw,ss − Vw,ss exp

Vw,ss

 × 100

  N exp exp 2 cal  [(Vw,ss − Vw,ss )/Vw,ss ] 1 SD (%) = 100 × N−1

(B1)

(B2)

where N = number of experimental data.

References Al-Farsi, M.A., Lee, C.Y., 2008. Optimization of phenolics and dietary fibre extraction from date seeds. Food Chem. 108, 977–985. Alvarez, S., Riera, F.A., Alvarez, R., Coca, J., Cuperus, F.P., Bouwer, S.T., Boswinkel, G., Gemert, R.W.V., Veldsink, J.W., Giorno, L., Donato, L., Todisco, E.D.S., Olsson, J., Trägårdh, G., Gaeta, S.N., Panyor, l., 2000. A new integrated membrane process for producing clarified apple juice and apple juice aroma concentrate. J. Food Eng. 46, 109–125. Bruijn, J.D., Venegas, A., Martinez, J.A., Borquez, R., 2003. Ultrafiltration performance of carbosep membranes for the clarification of apple juice. Lebensm. Wiss. Technol. 36, 397–406. Cassano, A., Figoli, A., Tagarelli, A., Sindona, G., Drioli, E., 2006. Integrated membrane process for the production of highly nutritional kiwifruit juice. Desalination 189 (1–3), 21–30. Cassano, A., Drioli, E., Galaverna, G., Marchelli, R., Silvestro, G.D., Cagnasso, P., 2003. Clarification and concentration of citrus and carrot juices by integrated membrane processes. J. Food Eng. 57, 153–163. Chang, C., Yang, M., Wen, H., Chern, J., 2002. Estimation of total flavonoid content in propolis by two complementary colorimetric methods. J. Food Drug Anal. 10, 178–182. Cheryan, M., 1998. Ultrafiltration and Microfiltration Handbook. Technomic Publishing Company Inc., Lancaster, PA, USA. Chhaya, Mondal, S., Majumdar, G.C., De, S., 2012. Clarifications of stevia extract using cross flow ultrafiltration and concentration by nanofiltration. Sep. Purif. Technol. 89, 125–134. Chhaya, Sharma, C., Mondal, S., Majumdar, G.C., De, S., 2011. Clarification of stevia extract by ultrafiltration: selection criteria of the membrane and effects of operating conditions. Food Bioprod. Process., http://dx.doi.org/10.1016/ j.fbp.2011.10.002. 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.

Conidi, C., Cassano, A., Drioli, E., 2011. A membrane-based study for the recovery of polyphenols from bergamot juice. J. Membr. Sci. 375, 182–190. Conidi, C., Cassano, A., Drioli, E., 2012. Recovery of phenolic compounds from orange press liquor by nanofiltration. Food Bioprod. Process. 90, 867–874. Díaz-Reinoso, B., Moure, A., Domínguez, H., Parajó, J.C., 2009. Ultra- and nanofiltration of aqueous extracts from distilled fermented grape pomace. J. Food Eng. 91, 587–593. Final Report (2014) on World Bank funded NAIP sub-project on “Value chain on production of food-grade nutraceuticals for use as natural antioxidants and food colorants”. Indian Council of Agricultural Research, New Delhi. Galanakis, C.M., 2012. Recovery of high-added value components from fruit wastes: conventional, emerging technologies and commercialized applications. Trends Food Sci. Techol. 26, 68–87. Galanakis, C.M., 2013. Emerging technologies for the production of nutraceuticals from agricultural by-products: a viewpoint of opportunities and challenges. Food Bioprod. Process. 91, 575–579. Galanakis, C.M., 2014. Editorial. Food Res. Int. 65, 299–300. Galanakis, C.M., 2015. Separation of functional macromolecules and micromolecules: from ultrafiltration to the border of nanofiltration. Trends Food Sci. Technol. 42, 44–63. Galanakis, C.M., Markouli, E., Gekas, V., 2013. Recovery and fractionation of different phenolic classes from winery sludge using ultrafiltration. Sep. Purif. Technol. 107, 245–251. Gekas, V., Hallstrom, B., 1987. Mass transfer in the membrane concentration polarization layer under turbulent cross flow. I. Critical literature review and adaptation of existing Sherwood correlations to membrane operations. J. Membr. Sci. 80, 153–170. Gohari, A.R., Hajimehdipoor, H., Saeidnia, S., Ajani, Y., Hadjiakhoondi, A., 2011. Antioxidant activity of some medicinal species using FRAP assay. J. Med. Plants 10 (37). Hart, F.L., Fisher, H.J., 1971. Methods of Analysis. Modern Food Analysis. Springer, Berlin, pp. 43. Kaneria, M., Chandra, S., 2013. Evaluation of antioxidant and antimicrobial capacity of Syzygium cumini L. leaves extracted sequentially in different solvents. J. Food Biochem. 37, 168–176. Kaur, C., Kapoor, H.C., 2000. Antioxidant activity and total phenolic content of some, Asian vegetables. Int. J. Food Sci. Technol. 37, 153–161. McCune, L.M., Johns, T., 2002. Antioxidant activity in medicinal plants associated with the symptoms of diabetes mellitus used by the indigenous peoples of the North American boreal forest. J. Ethnopharmacol. 82, 197–205. Mello, B.C.B.S., Petrus, J.C.C., Hubinger, M.D., 2010. Concentration of flavonoids and phenolic compounds in aqueous and ethanolic propolis extracts through nanofiltration. J. Food Eng. 96, 533–539. Mondal, S., Chhaya, De, S., 2012. Modeling of cross flow ultrafiltration of stevia extract in a rectangular cell. J. Food Eng. 112, 326–337. Montgomery, D.C., 2001. Design and Analysis of Experiments, fifth ed. John Wiley and Sons, New York. Murakami, A.N.N., Amboni, R.D.D.M.C., Prudêncio, E.S., Amante, E.R., Zanotta, L.D.M., Maraschin, M., Petrus, J.C.C., Teófilo, R.F., 2011. Concentration of phenolic compounds in aqueous mate (Ilex paraguariensis A. St. Hil) extract through nanofiltration. LWT – Food Sci. Technol. 44, 2211–2216. Patil, S.S., Thorat, R.M., Rajasekaran, P., 2012. Utilization of jamun fruit (Syzygium cumini) for production of red wine. J. Adv. Lab. Res. Biol. 2 (2), 200–203. Pritchard, M., Howell, J.A., Field, R.W., 1995. The ultrafiltration of viscous fluids. J. Membr. Sci. 102, 223–235. Prudêncioa, A.P.A., Prudêncioa, E.S., Ambonia, R.D.M.C., Murakamia, A.N.N., Maraschinb, M., Petrusc, J.C.C., Ogliarid, P.J., Leite, R.S., 2012. Phenolic composition and antioxidant

food and bioproducts processing 9 8 ( 2 0 1 6 ) 29–43

activity of the aqueous extract of bark from residues from mate tree (Ilex paraguariensis St. Hil.) bark harvesting concentrated by nanofiltration. Food Bioprod. Process. 90, 399–405. Ranganna, S., 1986. Hand Book of Analysis and Quality Control for Fruit and Vegetable Products. Tata McGraw Hill Publishing Company Limited, New Delhi, India. Sarkar, B., 2014. Enhanced cross-flow ultrafiltration of apple juice using electric field. J. Food Process. Preserv., 1–13. Sarkar, B., Dasgupta, S., De, S., 2008. Cross-flow electro-ultrafiltration of mosambi (Citrus sinensis (L.) Osbeck) juice. J. Food Eng. 89, 241–245. Singleton, V., Rossi, J.A., 1965. Colorimetry of total phenolics with phosphomolybdic–phosphotungstic acid reagents. Am. J. Enol. Vitic. 16, 144–158. Soong, Y., Barlow, P.J., 2004. Antioxidant activity and phenolic content of selected fruit seeds. Food Chem. 88, 411–417. Swami, S.B., Singh, N., Thakor, J., Patil, M.M., Haldankar, P.M., 2012. Jamun (Syzygium cumini (L.)): a review of its food and medicinal uses. Food Nutr. Sci. 3, 1100–1117. Torun, M., Rácz, G., Fogarassy, E., Vatai, G., Dinc¸er, C., Topuz, A., Ozdemir, F., 2014. Concentration of sage (Salvia fruticosa Miller)

43

extract by using integrated membrane process. Sep. Purif. Technol. 132, 244–251. Tylkowski, B., Tsibranska, I., Kochanov, R., Peev, G., Giamberini, M., 2011. Concentration of biologically active compounds extracted from Sideritis ssp. L. by nanofiltration. Food Bioprod. Process. 89, 307–314. Van Den Berg, G.B., Racz, I.G., Smolders, C.A., 1989. Mass transfer coefficients in cross flow ultrafiltration. J. Membr. Sci. 47, 25–51. Vanneste, J., Sotto, B.A., Courtin, C.M., Craeyveld, V.V., Bernaerts, K., Van Impe, J., Vandeur, J., Taes, S., Bruggen, B.V.D., 2011. Application of tailor-made membranes in a multi-stage process for the purification of sweeteners from Stevia rebaudiana. J. Food Eng. 103, 285–293. Vasco, C., Ruales, J., Eldin, A.K., 2008. Total phenolic compounds and antioxidant capacities of major fruits from Ecuado. Food Chem. 111, 816–823. Xu, L., Wang, S., 2005. The Ginkgo biloba extract concentrated by nanofiltration. Desalination 184 (1–3), 305–313. Yu, Z.R., Hung, C.C., Weng, Y.M., Su, C.L., Wang, B.J., 2007. Physicochemical antioxidant and whitening properties of extract from root cortices of mulberry as affected by membrane process. LWT – Food Sci. Technol. 40, 900–907.