Clarifications of stevia extract using cross flow ultrafiltration and concentration by nanofiltration

Clarifications of stevia extract using cross flow ultrafiltration and concentration by nanofiltration

Separation and Purification Technology 89 (2012) 125–134 Contents lists available at SciVerse ScienceDirect Separation and Purification Technology jou...

616KB Sizes 1 Downloads 64 Views

Separation and Purification Technology 89 (2012) 125–134

Contents lists available at SciVerse ScienceDirect

Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur

Clarifications of stevia extract using cross flow ultrafiltration and concentration by nanofiltration Chhaya a, Sourav Mondal b, G.C. Majumdar a, Sirshendu De b,⇑ a b

Department of Agriculture and Food Engineering, Indian Institute of Technology, Kharagpur 721302, India Department of Chemical Engineering, Indian Institute of Technology, Kharagpur 721302, India

a r t i c l e

i n f o

Article history: Received 9 December 2011 Received in revised form 8 January 2012 Accepted 9 January 2012 Available online 16 January 2012 Keywords: Stevioside Clarification Ultrafiltration Permeate flux Nanofiltration

a b s t r a c t Cross flow ultrafiltration was employed to clarify the pretreated stevia extract. Two practical modes of cross flow ultrafiltration, namely, steady state under total recycle mode and batch concentration mode, were used. A detailed investigation of effects of the operating conditions on the permeate flux and permeate quality was undertaken. It was observed that the significant flux enhancement was achieved with transmembrane pressure drop and cross flow rate. Maximum 200% flux enhancement with cross flow rate and 140% with transmembrane pressure drop were attained in the range of operating conditions studied herein. Effects of cross flow rate on the permeate properties were marginal but that of the transmembrane pressure drop was significant. Recovery of stevioside in the permeate was in the range of 30– 56% for various transmembrane pressure drop and it was maximum for lower operating pressure, 276 kPa. However, the recovery of stevioside decreased to 38% at 276 kPa pressure after 10 h of operation. Nanofiltration was employed to concentrate the ultrafiltered liquor. During nanofiltration, the ultrafiltered feed was concentrated maximum twice at 1241 kPa and 1500 rpm of stirrer speed within 1 h of operation. Maximum overall (ultrafiltration followed by nanofiltration) purity and recovery of 60% is obtained for a particular set of operating conditions. Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction Stevia is an herbaceous plant and it is mainly found in South America. Chemicals belonging to glycoside family, for example, stevioside, rebaudioside A, B, D, E, dulcoside A and B are responsible for sweet taste of the leaves of this plant [1]. Out of these, stevioside occurs in maximum amount in stevia leaves and other sweeteners comprise about 4–20% of the dried leaves [2]. Stevioside is 300 times sweeter than sugar, thermally stable at high temperature (about 100–120 0C), non-calorific and safer for diabetic patients [3,4]. Thus, the use of stevioside as a natural sweetener in place of sugar is having high demand. Therefore, extraction of this natural sweetener from stevia leaves is an important technical challenge. Traditional processes for extraction of stevioside include addition of chelating agents, like calcium hydroxide, followed by crystallization [5]; centrifugation followed by treatment with calcium hydroxide and further treatment by ion exchange resin [6,7]; solvent extraction followed by ion exchange [8]; adsorption with zeolite CaX in fixed bed columns [9,10]. These processes are time consuming, expensive and sometimes are not suitable for edible purposes when the solvent like methanol is used for extraction. An additional challenge is removal of the added chemicals and ⇑ Corresponding author. Tel.: +91 3222 283926; fax: +91 3222 255303. E-mail address: [email protected] (S. De). 1383-5866/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.seppur.2012.01.016

traces of organic solvent. Membrane based technologies can be a viable alternative in this regard. Various advantages of membrane based processes include, no phase change; addition of no chemicals and ease of scaling up. Membrane based processes have already been used for processing of various juice and beverages. Some of these are, apple juice [11], mosambi juice [12], pomegranate juice [13], carrot juice [14], pineapple juice [15], kiwi fruit juice [16], water melon juice [17], green coconut water [18], black tea [19], selective extraction of ()epigallocatechin gallate from green tea leaves [20], etc. Few uses of membrane based technologies for extraction of stevioside from stevia leaves have also been attempted. In most of the cases, hybrid membrane separation was reported. Fuh and Chiang, carried out extraction of stevioside using two routes [21]. These were, (i) precipitation by inorganic salts; (ii) clarification by ultrafiltration in multi-effect diafiltration mode followed by concentration by reverse osmosis. 25,000 and 100,000 Da molecular weight cut off membrane were used for clarification at 12 and 8.5 bar transmembrane pressure, respectively and at 25 l/min flow rate. Reverse osmosis was carried out at 45 bar and 25 l/min flow rate and the solution was concentrated 10 folds. Stevioside with higher purity was obtained in membrane based processes. Zhang et al. adopted a membrane based process for processing of stevia extract [22]. Microfiltration (using 0.35 m ceramic membrane at 104 kPa transmembrane pressure) was used as pre-treatment followed by

126

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134

ultrafiltration (using 2500 Da cut off membrane at 440 kPa pressure) for subsequent clarification. Finally, nanofiltration was used for further purification and concentration of stevia extract at 510 kPa. However, they have not reported any data regarding permeate flux and recovery or final concentration of nanofiltration. Modified Zeolite (NaX) was used for pre-treatment of crude stevia extract followed by microfiltration [23]. Approximately 80–100% clarification of stevia extract was achieved with microfiltration membranes of varying pore sizes and at different operating pressures. A comparison based on the use of commercial and taylormade PES membrane for stevioside purification has been reported [24]. Using microfiltration, ultrafiltration and nanofiltration, 37% purity and 30% yield were achieved. However, the above studies lack the analysis of the effect of operating conditions on system performance. Decline of permeate flux during the course of filtration is a major problem [25,26]. Membrane fouling is responsible for this. Two types of fouling normally occur, namely, reversible and irreversible. Reversible fouling is due to a phenomenon, known as concentration polarization, i.e. accumulation of solutes over the membrane surface resulting to additional resistance against the solvent flux. This fouling can be removed by adopting a suitable washing protocol. On the other hand, during filtration, solute may adsorb on the pore mouth or inside the pores, thereby, blocking the pores partially and/or completely. This fouling cannot be removed completely by washing the membrane and a fraction of membrane permeability is lost permanently. This is known as irreversible fouling. It is well-established that the fouling phenomena involved and the consequent membrane flux decline are inseparably associated with the operating conditions. Therefore, by selecting a suitable set of operating conditions and the mode of operation, one can reduce membrane fouling. Hence, the role of operating conditions is extremely important in membrane separation processes. Cross flow mode of operation is quite effective in arresting the growth of polarized layer of solutes over the membrane surface due to the shearing action of the retentate flow. The works related to application of membrane processes for treatment of stevia extract as described earlier do not provide any details of the effects of the operating conditions on the permeate flux decline as well as the permeate quality during ultrafiltration/nanofiltration/reverse osmosis of stevia extract. These aspects are extremely important for an efficient design of industrial-scale processing unit. The present study is therefore, taken up to bridge this gap. In this work, cross flow ultrafiltration was conducted in two modes, namely, total recycle and batch concentration mode with a detailed investigation of the effects of operating conditions. Ultrafiltered solution was concentrated using nanofiltration in a stirred cell and the effects of transmembrane pressure drop and the stirrer speed were investigated in detail. 2. Materials and methods 2.1. Materials Dry stevia leaves powder was obtained from M/s, RAS Agro Associates, Maharashtra, India. Distilled water was used as solvent for extraction process. M/s, Merck India Limited, Mumbai, India, supplied high performance liquid chromatography (HPLC) grade acetonitrile and water. Standard stevioside of 98% purity was obtained from M/s, Sigma–Aldrich, USA. 2.2. Membranes For ultrafiltration, a 30 kDa membrane of polyethersulfone with permeability 4.4  1011 m/Pa s has been used. This membrane

was selected after investigating the performance of 5, 10, 30 and 100 kDa membranes for extraction of stevia. 30 kDa membrane was found to have the highest permeate flux and maximum recovery of stevioside in the permeate [27]. These membranes were supplied by M/s, Permionics Membranes Pvt. Ltd., Gorwa, Vadodara, India. Nanofiltration was conducted using 400 molecular weight cut off membrane consisting of a polyamide skin over a polysulphone support is supplied by M/s, Genesis Membrane Sepratech Pvt. Ltd., Mumbai, India. The permeability of the nanofiltration membrane was 1.44  1011 m/Pa s. 2.3. Experimental set up Two different experimental set ups have been used. A cross flow ultrafiltration set up for clarification of stevia extract using ultrafiltration. A stirred experimental set up was used for conduction of nanofiltration experiments. 2.3.1. Cross flow ultrafiltration set up A rectangular cross-flow cell, made of stainless steel, was designed and fabricated. Two neoprene rubber gaskets were placed over the membrane forming the flow channel. The channel height after tightening the two flanges was found to be 3.0  103 m. The effective dimension of the membrane was 14.5 cm in length and 5.5 cm in width. The cell consisted of two rectangular matching flanges. The inner surface of the top flange was mirror polished. The bottom flange was grooved, forming the channels for the permeate flow. A porous stainless steel plate was placed on the lower flange that provides mechanical support to the membrane. Two flanges were tightened to create a leak proof channel for conducting experiments in cross flow mode. The centrifuged extract was pumped by a high pressure reciprocating pump from the stainless steel feed tank to the cross flow cell. The retentate stream was recycled to the feed tank routed through a rotameter. The pressure and the cross flow rate inside the membrane channel were independently set by operating the valves in the bypass line and that at the outlet of the membrane cell. Permeate samples were collected from the bottom of the cell and were analyzed for color, clarity, total solids and stevioside concentration. The membrane module assembly is presented in Fig. 1. 2.3.2. Stirred batch cell (for nanofiltration) The experimental set-up consisted of a stirred batch cell of 650 ml capacity made of stainless steel and it was pressurised using a nitrogen cylinder. For a typical run, about 300 ml of ultrafiltered feed was charged into the batch cell. The stirrer speed was set using a variac (variable AC transformer for smooth control of voltage and thereby controlling the stirring speed) and was measured by a hand held digital tachometer (Agronic, India). Inside the cell, a circular membrane was placed over a base support. The membrane diameter was 6.6 cm and the effective membrane area was 34.2 cm2. The permeating solution from the bottom of the cell was used for further analysis. The schematic diagram of the experimental set-up is shown in Fig. 2. 2.4. Methods 2.4.1. Extraction process Dry stevia leaves powder was mixed with hot distilled water at a ratio of 1:14(g:ml). Temperature was fixed at 78 ± 1 °C for 56 min. The above operating conditions were selected based on the optimization experiments carried out for maximum extraction of stevioside in the liquor [28]. Constant temperature water bath was used for hot water extraction. Next, the aqueous stevia extract was cooled to room temperature and cloth filtered. The filtered

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134

127

Fig. 1. Schematic diagram of the cross flow set up: (a) two flanges; (b) the top view of grooved bottom flange; (c) top view of bottom flange after putting the membrane.

were selected based on the related optimization study so that maximum clarity and recovery of stevioside were obtained.

1 4

3 2

1: Filtration cell. 2: Electronic balance. 3: Beaker for collection of permeate 4: Nitrogen cylinder Fig. 2. Schematic diagram of the stirred batch cell.

extract was analyzed for their color, clarity, stevioside concentration and total solid content.

2.4.2. Primary clarification Primary clarification of the extract was carried out in a laboratory centrifuge (Model number R-24, supplied by M/s, Remi International Ltd., Mumbai, India). The centrifugation capacity was 200 ml per batch. The operating conditions were stirrer speed 5334 g and centrifugation time 26 min. These operating conditions

2.4.3. Cross flow ultrafiltration A fresh membrane was compacted at a pressure higher than the maximum operating pressure for 3 h using distilled water and then its permeability was measured. The extract was placed in a stainless steel feed tank of 3 l capacity. A high pressure reciprocating pump was used to feed the effluent into the cross-flow membrane cell. Cumulative volumes of permeate were collected during the experiment. Permeate samples were collected at different time intervals for analysis. A bypass line was provided from the pump delivery to the feed tank. Retentate and bypass control valves were used to vary the pressure and flow rate accordingly. Values of permeate flux were determined from the slopes of cumulative volume versus time plot. The precision of flux measurement was in the order of ±5%. The permeate stream after collecting required amount of sample was recycled to the feed tank to maintain a constant concentration in the feed tank under total recycle mode. The permeate was not recycled under batch concentration mode of operation. Duration of the cross-flow experiments was 45 min for total recycle mode of operation and it was 10 h for batch concentration mode. The feed volumes were 2 l for total recycle mode and the initial feed volume was 1.8 l for batch concentration mode. The schematic of cross flow experimental set-up is given in Fig. 3. Once an experimental run was over, the membrane was thoroughly washed, in situ, with distilled water for 30 min applying a

128

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134

Fig. 3. Schematic diagram of cross flow set up.

maximum pressure of 200 kPa. The cell was dismantled and the membrane was rinsed with distilled water and was dipped in 2% sodium dodecyl sulphate solution overnight. Next, the membrane was washed carefully with distilled water to remove traces of surfactant. The cell was reassembled and the membrane permeability was again measured using distilled water. After that, the set up was ready for the next experiment with centrifuged stevia extract. All the experiments were conducted at a room temperature of 32 ± 2 °C. 2.4.4. Nanofiltration experiments in stirred batch cell First, the membrane was compacted for 2 h at 800 kPa pressure using distilled water. Water flux was measured at five different transmembrane pressure drop values. From the slope of the permeate flux and pressure drop curve, the membrane permeability was determined. Next, the cell was filled up by 300 ml of ultrafiltered stevia extract (collected at 550 kPa and 100 l/h) and the operating pressure was set using the nitrogen cylinder through a regulator. The stirring speed in the cell was set at an appropriate rpm by using a variac. Each experiment was conducted for 1 h at the room temperature of 30 ± 2 °C. Clarified stevia extract (permeate) was collected in a measuring cylinder. Cumulative volume of permeate as a function of time was measured. From the slope of the cumulative volume–time plot, the permeate flux as a function of time was obtained. At the end of the experiment, the permeate samples were collected and analyzed for total solids, color, clarity and stevioside concentration. After the experiment, the set up was dismantled and the membrane was rinsed with distilled water carefully and was kept in 2% surfactant, sodium dodecyl sulfate solution overnight. The cleaned membrane was rinsed carefully so that the traces of surfactant were removed and again its permeability was measured using distilled water.

ing pressure drops were 827, 965, 1103 and 1241 kPa. At each pressure drop, three stirrer sppeds, namely, 500, 1000 and 1500 were used. 2.6. Analysis The original, centrifuged and ultrafiltered stevia extract were analyzed for their color, clarity, total solid content and stevioside concentration. Color of the extract was measured in terms of optical absorbance (A) at a wavelength of 420 nm using a spectrophotometer (BIORAD Smart Spec 3000, USA). Clarity of the extract was measured in terms of percentage of transmittance (%T) using a spectrophotometer. This is given by the equation, %T=100  10A, where, A is optical absorbance at a wavelength of 660 nm. Total solid of the sample was measured gravimetrically by heating the extract in hot air oven at 104 ± 2 °C until the difference in the weight of the extract becomes constant at successive intervals [29]. Total solid was represented in terms of gram per 100 ml of stevia extract. Amount of stevioside present in the stevia extract (before and after clarification) was analyzed by HPLC (M/s, Perkin Elmer Co., Shelton, Connecticut, USA). Initially a calibration curve was prepared using standard stevioside of 98% purity. Then, for each analysis, a sample volume of 30 ll was injected into an Agilent Zorbax SB C-18 reverse phase HPLC column (4.6 mm ID, 250 mm length and 5 lm particle size). Mobile phase was a mixture of acetonitrile and water at a ratio of 80:20 (volume basis) and the flow rate was 1 ml/min. UV detector (Perkin Elmer series 200) was used for detection of stevioside present in the sample at a wavelength of 210 nm. Stevioside concentration was represented in terms of percentage (%) of stevioside recovered in the clarified extract as:

%Stevioside ¼ ðstevioside retained in clarified 2.5. Experimental design Cross flow ultrafiltration experiments under total recycle mode were performed using the operating pressure difference as 276, 414, 552 and 690 kPa. The cross flow rates were 60, 80, 100 and 120 l/h. These experiments under batch concentration mode were undertaken at 276, 414 and 552 kPa pressure and 100 l/h cross flow rate. The operating variables for nanofiltration experiments were transmembrane pressure drop and the stirring speed. The operat-

extract=stevioside present in feedÞ  100:

2.7. Statistical Analysis All membrane filtration experiments were carried out in triplicate. All the property data presented in the tables are reported with standard deviation. The permeate flux data were within ±3% variation and these were presented as error bars in the figures. The permeability values had ±5% variation.

129

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134 Table 1 Various properties of the ultrafiltered liquor at different operating conditions under total recycle mode of operation. Operating pressure (kPa)

Flow rate in l/h (cross flow velocity in m/s)

Permeate color (A)

Permeate clarity (%T)

Permeate total solid (g/l)

Permeate stevioside (%)

Purity of stevioside   Stev per TSper

Selectivity of   Stev per stevioside LMW per

276

60 80 100 120 60 80 100 120 60 80 100 120 60 80 100 120 _ _

1.5 ± 0.13 1.4 ± 0.12 1.1 ± 0.13 1.1 ± 0.15 0.9 ± 0.14 0.9 ± 0.13 0.9 ± 0.12 0.8 ± 0.14 0.8 ± 0.15 0.7 ± 0.16 0.7 ± 0.12 0.6 ± 0.13 0.6 ± 0.14 0.6 ± 0.12 0.6 ± 0.15 0.4 ± 0.12 10.31 ± 1.13 _

53.1 ± 1.1 62.5 ± 1.3 69.2 ± 2.3 69.5 ± 1.3 80.9 ± 1.5 81.3 ± 2.1 81.8 ± 2.3 81.5 ± 2.4 83.4 ± 1.5 84.9 ± 2.6 85.5 ± 2.4 87.3 ± 1.8 85.3 ± 1.7 86.1 ± 1.6 86.9 ± 2.2 87.5 ± 2.3 1.3 ± 0.5 _

17 15 12 11 14 14 13 11 13 11 10 10 11 11 10 10 27 ± 1.3 _

71.8 ± 3.1 58.0 ± 2.1 49.0 ± 2.4 49.0±3.2 49.0 ± 3.5 45.0 ± 2.8 43.3 ± 2.6 40.4 ± 2.9 45.0 ± 2.4 40.5 ± 3.2 39.7 ± 2.6 37.3 ± 3.5 37.2 ± 2.6 29.7 ± 2.8 31.1 ± 2.5 27.5 ± 2.2 15753 ± 5.2 (mg/l) 17167 ± 8.4 (mg/l)

0.67 0.61 0.64 0.70 0.55 0.51 0.52 0.58 0.55 0.58 0.63 0.59 0.53 0.43 0.49 0.43

2.0 1.6 1.8 2.4 1.2 1.0 1.1 1.4 1.2 1.4 1.7 1.4 1.1 0.7 1.0 0.8

414

552

690

Centrifuged extract (feed) Crude extract

(0.10) (0.14) (0.17) (0.20) (0.10) (0.14) (0.17) (0.20) (0.10) (0.14) (0.17) (0.20) (0.10) (0.14) (0.17) (0.20)

3. Results and discussions 3.1. Cross flow ultrafiltration As mentioned earlier, two modes were used for conducting cross flow ultrafiltration experiments. First total recycle mode, where, the permeate was recycled back and the feed concentration was maintained constant. Second, the batch concentration mode, where the feed concentration was not recycled and the volume of the feed tank was continued to reduce and the feed concentration increased. The ultrafiltration feed essentially contains various components, of which only Stevioside is desirable. The mixture of components can be clubbed together as high molecular weight components (HMW). Components with lower molecular weight than Stevioside are grouped in low molecular weight solutes (LMW). It is considered that the HMW is completely rejected, LMW is freely permeable, and Stevioside is partially retained by gel layer and membrane. Following this categorization, one can essentially get an estimate of the amount of LMW in the permeate (which is equivalent to that present in feed). LMW in permeate can be estimated by TSperStevper. With this definition, one can determine the purity and selectivity of Stevioside in permeate by,

purity ¼

stev ioside concentration in permeate and; concentration of total solids in permeate

selectiv ity ¼

stev ioside concentration in permeate : concentration of LMW in permeate

These values are presented in Table 1 for different operating conditions in total recycle mode and in Fig. 7 in case of batch mode. 3.1.1. Total recycle mode In this mode of operation, variation of permeate flux as a function of time for various transmembrane pressure drop and the cross flow rates are shown in Fig. 4. Three general trends are observed from these figures. First, the permeate flux declines over time of operation and finally, a steady state is reached. Second, at any point of time, the permeate flux increases with cross flow rates at a fixed transmembrane pressure drop. Third, at any point of time, the permeate flux increases with transmembrane pressure drop at a fixed cross flow rate. The first observation is due to con-

centration polarization. As time of filtration progresses, more solutes are convected towards the membrane and a cake type of layer starts growing over the membrane surface. Thickness of this layer increases with the time of filtration. This layer offers a resistance against the solvent flux. Since, with time of filtration, thickness of this layer increases, the permeate flux declines. For example, at 276 kPa operating pressure and 120 l/h cross flow rate, the permeate flux decreases from about 18 to 14 l/m2 h (a decrease of 22%) after 45 min (Fig. 4a). At the same transmembrane pressure drop and 80 l/h cross flow rate, the decrease in flux is about 35% over 45 min (Fig. 4a). Similar, decline in flux over the filtration duration for 414 kPa at 120 and 80 l/h of cross flow rates is 20% and 38% (Fig. 4b). These values of flux decline for 552 kPa are 20% and 22% at these flow rates (Fig. 4c). At 690 kPa, these values are 17% and 16% (Fig. 4d). Therefore, the permeate flux declines over the filtration time in between 16% and 38% for different transmembrane pressure drop and the cross flow rates. Thus, at higher cross flow rate, the flux decline is less. This is due to the fact that at higher cross flow rate, thickness of the cake type layer decreases due to increased forced convection. It is also observed from these figures that a steady state is attained in all the cases. Initially, the convective flux of solutes towards the membrane due to pressure gradient is more and more solutes are deposited over the membrane surface, forming a cake layer. This layer keeps on growing as more solutes are convected towards the membrane. After sometime, the growth of this layer is arrested by the forced convection imposed by the cross flow rate in the flow channel and a steady state is attained. As observed from Fig. 4a, that at 276 kPa pressure, steady state is attained after about 22 min. This time is reduced 10–20 min at higher operating pressure drops. At a fixed transmembrane pressure drop, the permeate flux increases with the cross flow rates. At higher cross flow rate, the shearing action of the convective flow on the cake layer is more and its growth is restricted. Therefore, the resistance against the solvent flow offered by the cake layer is less, leading to an increase in permeate flux. For example, at 276 kPa pressure drop, the steady state flux increases from 5 to 15 l/m2h as the cross flow rate increases from 60 to 120 l/h, affecting an increase of 200%. This increase is 125% at 414 kPa, 100% for 551 kPa and 83% at 690 kPa. At a fixed cross flow rate, the permeate flux also increases with the transmembrane pressure drop. Increase in pressure drop has two opposing effects. First, it increases the driving force leading to flux enhancement. Second, more solutes are convected towards

130

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134

24 22

Permeate flux ( L/m2h)

20 18

35

b

276 kPa 60 l/h 80 l/h 100 l/h 120 l/h Error bar: ± 3%

414 kPa 60 l/h 80 l/h 100 l/h 120 l/h Error bar:± 3%

30

Permeate flux (l/m2h)

a

16 14 12 10

25

20

15

8 10

6 0

10

20

30

0

40

10

20

Time (min)

552 kPa 60 l/h 80 l/h 100 l/h 120 l/h Error bar:±3%

Permeate flux (l/m2h)

25

40

20

15

35

d

690 kPa 60 l/h 80 l/h 100 l/h 120 l/h Error bar:± 3%

30 Permeate flux (l/m2h)

c

30

Time (min)

25

20

15

0

10 0

10

20

30

10

20

40

30

40

Time (min)

Time (min) Fig. 4. Flux decline profiles during ultrafiltration in total recycle mode: (a) 276 kPa; (b) 414 kPa; (c) 552 kPa; (d) 690 kPa.

1.40

30

12

21

10

2

24

11

Permeate flux (l/m .h)

Steady state permeate flux (l/m2.h)

27

Flow rate = 100 l/h 276 kPa 414 kPa 552 kPa Error bar: ±3%

18 15 12 9

9 8

1.35 1.30 1.25

7 1.20

6 5

1.15

4 1.10

3

6

2

3

1

1.05

0

0 0

100

200

300

400

500

600

700

Volume Concentration Factor (VCF)

60 l/h 80 l/h 100 l/h 120 l/h Error bar: ±3%

1.00 0

1

2

3

4

5

6

7

8

9

10

11

Time (h)

Operating pressure (kPa) Fig. 5. Variation of steady state permeate flux with transmembrane pressure drop and cross flow rate.

Fig. 6. Flux decline profile and variation of volume concentration factor with transmembrane pressure drop in batch concentration mode of cross flow ultrafiltration.

the membrane surface, thereby increasing the thickness of the cake layer, resulting to increase in resistance against the solvent flow and consequently decline in permeate flux. However, by observing the trends of flux decline profiles in Fig. 4a–d, it is clear that the first effect dominates the second one and the flux increases with

pressure. For example, at 60 l/h cross flow rate, the steady state permeate flux increases from 5 to 12 l/m2 h (140% flux enhancement) as the pressure increases from 276 to 690 kPa. This enhancement over this pressure range is about 100% for 80 l/h, 81% for 100 l/h and 57% for 120 l/h cross flow rate.

131

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134

a

1.8

b

100 l/h 276 kPa 414 kPa 552 kPa

1.5

120

100

Clarity (%T)

1.2

Color (A)

100 l/h 276 kPa 414 kPa 552 kPa

0.9 0.6

80

0.3 60

0.0 0

2

4

6

8

10

0

2

4

Time (h)

c

10

8

10

100

100 l/h 276 kPa 414 kPa 552 kPa

80 Stevioside Recovery (%)

Total solid (g/100 mL)

d

100 l/h 276 kPa 414 kPa 552 kPa

2.4

8

Time (h)

3.0 2.7

6

2.1 1.8 1.5 1.2 0.9

60

40

20

0.6 0

2

4

6

8

0

10

0

2

4

Time (h)

6 Time (h)

e

0.70 276 kPa 414 kPa 552 kPa

0.65

Purity

0.60

0.55

0.50

0.45

0.40

0

2

4

6

8

10

Time (h) Fig. 7. Profiles of permeate properties for various operating conditions in batch concentration mode of ultrafiltration: (a) color; (b) clarity; (c) total solids; (d) recovery of stevioside; (e) purity of stevioside.

The steady state flux values at different operating pressure drop and cross flow rates are presented in Fig. 5. The trends are as expected and the reasons are already discussed earlier. The properties of the permeate at the steady state with different operating conditions are presented in Table 1. Some general trends are observed from this table. As the operating pressure drop increases, the stevioside recovery in the permeate decreases. The selectivity

and purity of stevioside in the permeate are almost independent of flow rate. Both selectivity and purity decrease with transmembrane pressure drop. At higher pressure drop, the cake layer becomes compact (associated with increasing porosity) and it acts as a dynamic membrane. Therefore, this layer retains some of the stevioside and recovery of stevioside in the permeate becomes less. For example, average (over various cross flow rates) recovery of

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134

2 Permeate flux (l/m .h)

1.5

1.4

1.3

25

1.2 20 1.1

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.7

45

Operating pressure = 965 kPa 500 rpm 1000 rpm 1500 rpm Error bar: ± 3%

40

35

1.4 1.3 25

1.2 20

1.0 1.2

15 0.0

1.1 1.0 0.1

0.2

0.3

0.4

2

1.6

35 1.4

30 25

1.2 20 15 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.6

0.7

0.8

0.9

1.0

1.1

0.7

0.8

0.9

1.0

1.0 1.1

Time (Hours)

60

Operating pressure = 1241 kPa 500 rpm 1000 rpm 1500 rpm Error bar:± 3%

55 50

2.0

1.8

45 1.6 40 35

1.4

30 1.2 25 20 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Volume Concentration Factor (VCF)

40

1.8

Volume Concentration Factor (VCF)

Permeate flux (l/m .h)

d

Operating pressure = 1103 kPa 500 rpm 1000 rpm 1500 rpm Error bars: ± 3%

45

0.5

Time (h)

2 Permeate flux (l/m .h)

50

1.5

30

Time (h)

c

1.6

Volume Concentration Factor (VCF)

Operating pressure = 827 kPa 500 rpm 1000 rpm 1500 rpm Error bar: ± 3%

30

15 0.0

b

1.6

35

Volume Concentration Factor (VCF)

a

2 Permeate flux (l/m .h)

132

1.0 1.1

Time (Hours)

Fig. 8. Flux decline profiles and variation of volume concentration factor with operating conditions during stirred batch nanofiltration: (a) 827 kPa; (b) 965 kPa; (c) 1103 kPa; (d) 1241 kPa.

stevioside at 276 kPa is 56% and it is 44% at 414 kPa, 40% for 552 kPa and 31% for 690 kPa. This dynamic cake type layer retains other solids at higher pressure drop, thereby, increasing the clarity of the permeate remarkably at higher operating pressure. Clarity is about 87% at 690 kPa and 120 l/h cross flow rate whereas, that in the feed of ultrafiltration is only 1.26%. Thus, the total solids in the permeate also decreases at higher operating pressure. It is also noted from this table that the stevioside recovery decreases marginally with cross flow rates. Except the first two experiments, at 276 kPa, 60 and 80 l/h, the variation of permeate recovery for different cross rate at a fixed pressure value is insignificant. This is due to the fact that the membrane for the first experiment is fresh and after one experiment, there occurs some irreversible fouling that reduces the stevioside recovery drastically. However, this fouling is present for subsequent experiments, but it is marginal and the stevioside recovery shows a declining trend (although extremely small) with the cross flow rates at a fixed transmembrane pressure drop. 3.1.2. Batch concentration mode As mentioned earlier, in this mode of operation, the permeate flux is not recycled back. In fact, for clarification of the stevia extract, the permeate is the product and this is the most favorable mode of operation. Three experiments were conducted in this case, at 100 l/h cross flow rate, the operating pressure difference were varied as 276, 414 and 552 kPa. The permeate flux profile along with the volume concentration ratio is presented in Fig. 6, as a function of time. Two general trends are observed from this figure.

First, the permeate flux decline is more in this case compared to the total recycle mode (in Fig. 4a–d) and second, there exists no steady state. Flux decline is more at higher operating pressure. In this mode of operation, the permeate is not recycled to the feed tank; as a result, the volume of the feed tank reduces leading to an increase in feed concentration. As the feed concentration increases, the concentration polarization becomes more severe. More solutes are convected towards the membrane surface, resulting to a thicker cake layer. This increases the resistance against the solvent flux and the permeate flux declines. At higher operating pressure, solute deposition on the membrane surface is augmented by forced convection, leading to a further decline in permeate flux. As the above phenomena increases as a time of filtration increases, a steady state is never attained. Over a period of 10 h of operation, the flux decline is 6 to 2 l/m2 h at 276 kPa. It is 8.2 to 2.2 l/m2 h at 414 kPa and 11.5 to 3 l/m2 h at 552 kPa. Volume concentration factor (VCF) is defined as V0/V, where, V0 is the initial volume of feed and V is the volume at ant time. Variation of VCF with time is presented in Fig. 6. It is observed from this figure that VCF is more at higher pressure as more volume of permeate is filtrated. After 10 h of operation, VCF reaches a value of 1.35 at 552 kPa pressure and 100 l/h cross flow rate. The properties of the permeate were also monitored over the filtration period. Profiles of color, clarity, total solids, stevioside recovery and purity in permeate are presented in Fig. 7a–e. It is observed from these figures that color, total solids and the stevioside recovery in permeate decrease with time and clarity increases with time. As discussed earlier, with progress in filtration time, the cake

133

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134 Table 2 Various properties of permeate of nanofiltration at the end of the experiment (feed is ultrafiltration permeate at 552 kPa and 100 l/h). Operating pressure (kPa)

Stirring speed (rpm)

Permeate color (A)

827

500 1000 1500 500 1000 1500 500 1000 1500 500 1000 1500 552 kPa, 100 l/h

0.02 ± 0.003 0.02 ± 0.002 0.02 ± 0.003 0.02 ± 0.003 0.02 ± 0.002 0.02 ± 0.004 0.01 ± 0.005 0.02 ± 0.003 0.02 ± 0.002 0.02 ± 0.004 0.02 ± 0.002 0.02 ± 0.005 0.62 ± 0.003

Optimum operating condition Optimum operating condition

10.6 ± 1.4

_

_

14129 ± 6.2

12.3 ± 1.6

0.006 ± 0.003

2.9 ± 0.4

15699 ± 4.4

965

1103

1241

UF extract (feed for NF) Centrifuge extract Crude extract

Permeate clarity (%T)

Permeate total solid (g/100 ml)

99.5 ± 0.4 99.9 ± 0.3 99.1 ± 0.4 99.0 ± 0.5 99.5 ± 0.4 99.6 ± 0.3 99.3 ± 0.2 98.6 ± 0.3 98.1 ± 0.2 99.3 ± 0.4 99.3 ± 0.4 99.1 ± 0.3 92.8 ± 0.3

0.2 ± 0.05 0.2 ± 0.04 0.2 ± 0.02 0.2 ± 0.06 0.2 ± 0.05 0.2 ± 0.04 0.2 ± 0.05 0.2 ± 0.04 0.2 ± 0.03 0.1 ± 0.04 0.1 ± 0.03 0.2 ± 0.04 0.9 ± 0.05

type of layer grows on membrane surface that acts as dynamic membrane and retains the solutes. Thus, total solids and stevioside recovery decreases. Although, color decreases with time, its variation is marginal. An interesting observation is made from Fig. 7d. Stevioside recovery at the end of 10 h for all three operating pressures is in between 30% (at higher pressure, i.e. 552 kPa) and 38% (at lower pressure, i.e. 276 kPa). This is due to the enhanced retention of the dynamic membrane at higher pressure by making it more compact. From Fig. 6, it is also observed that the permeate flux after 10 h is 2 l/m2 h at 276 kPa that is marginally less than that at 552 kPa (3 l/m2 h). Since, stevioside recovery is our main concern, a lower transmembrane pressure drop must be selected with a reasonable permeate flux. On the other hand, the cross flow rate should be maximum to obtain a higher permeate flux (refer Fig. 4). Thus, among the operating conditions studied herein, 276 kPa pressure and 120 l/h cross flow rate are suitable operating conditions for ultrafiltration of stevia extract with 30 kDa membrane. Another interesting feature that can be observed from Fig. 7e is that the purity in case of lower pressure decreases on increasing time of operation, which is not the case for higher pressure. The compactness of the dynamic cake layer over the membrane is not significant enough to screen other solids at lower pressure. This results to increase in total solids in the permeate (refer to Fig. 7c), thereby, decreasing purity. It must be noted here, that purity does not exclusively depend on the amount of Stevioside present in the permeate, rather, it the relative ratio of the amount of Stevioside to total solids in the permeate. So, even for the same amount of Stevioside content, purity can be increases if the total solid content is decreased. So, ideally, the batch operation at lower pressure and high flowrate should be limited up to 5 h (as observed from the present study) for maintaining high purity of stevioside in the permeate. 3.2. Concentration by nanofiltration Profiles of permeate flux and volume concentration factor with transmembrane pressure drop are shown in Fig. 8a–d at various stirring speeds. It is observed from these figures that the permeate flux declines with time and the flux is higher at higher operating pressure, as expected. At 1241 kPa pressure and 1500 rpm, flux is the highest and hence, volume concentration factor is the maxi-

Permeate stevioside (mg/l) 127.9 ± 1.4 143.1 ± 1.4 196.8 ± 2.4 173.0 ± 3.4 107.7 ± 2.4 208.5 ± 1.4 348.0 ± 3.4 179.2 ± 2.4 349.5 ± 3.4 246.5 ± 2.4 276.7 ± 1.4 201.3 ± 2.4 4945 ± 5.4

Product (retentate) total solid (g/100 ml)

Product (retentate) stevioside (mg/l)

Overall Recovery (%) [UF + NF]

Overall Purity (%)[UF + NF]

1.1 ± 0.05 1.2 ± 0.04 1.3 ± 0.02 1.2 ± 0.06 1.3 ± 0.05 1.3 ± 0.04 1.3 ± 0.05 1.4 ± 0.04 1.4 ± 0.03 1.4 ± 0.04 1.5 ± 0.03 1.6 ± 0.04

6486 ± 5 7250 ± 3 7462 ± 6 7283 ± 4 7606 ± 8 7976 ± 3 7565 ± 5 8233 ± 2 8529 ± 7 7952 ± 6 8726 ± 8 9594 ± 10

41.3 46.2 47.5 46.4 48.4 50.8 48.2 52.4 54.3 50.6 55.6 61.1

59.0 60.4 57.4 60.7 58.5 61.4 58.2 58.8 60.9 56.8 58.2 60.0

mum to about 2 in the test cell. Various properties of the permeate are reported in Table 2. It is observed from this table that the clarity of permeate is more than 99% in most of the cases. Color and total solids in the permeate are quite low. Retention of color is in between 96% and 98% for different operating conditions studied herein. Retention of stevioside is in the range of 93% to 98%. The concentration factor (ratio of feed concentration of stevioside to its initial concentration in feed) of stevioside is also presented in Table 2. It is observed from this table that at 1241 kPa pressure and 1500 rpm, the feed is concentrated about two times in 1 h of operation. It can be concluded that the purity of overall process (UF + NF) is constant around 60%. However, the overall recovery of Stevioside increases with stirring and transmembrane pressure drop. Maximum recovery is obtained at 1241 kPa and 1500 rpm. 4. Conclusion Clarification of centrifuged stevia extract by cross flow ultrafiltration and subsequent concentration by nanofiltration was studied in this work. In continuous cross flow under total recycle mode, permeate flux declined in between 16% and 38% for different operating conditions. However, steady state flux increased upto 200% when the cross flow rate increased from 60 to 120 l/h at 276 kPa. Flux enhancement upto 140% was attained when transmembrane pressure drop increased from 276 to 690 kPa. At higher operating pressure, recovery of stevioside in the permeate was less. Average 56% recovery of stevioside was obtained in the permeate at 276 kPa. However, in batch concentration mode of cross flow ultrafiltration, about 38% stevioside recovery was attained after 10 h. During nanofiltration, stevioside in the feed was concentrated twice in 1 h at 1241 kPa pressure drop and 1500 rpm. Maximum recovery is attained by ultrafiltration at 552 kPa, 100 l/h followed by nanofiltration at 1241 kPa and 1500 rpm. References [1] A.Y. Leung, S. Foster, Encyclopedia of Common Natural Ingredients — Used in Food, Drugs and Cosmetics, second ed., John Wiley and Sons, New York, 1996. pp.478. [2] J.M.C. Geuns, Molecules of interest stevioside, Phytochemistry 64 (2003) 913– 921.

134

Chhaya et al. / Separation and Purification Technology 89 (2012) 125–134

[3] S.S. Chang, J.M. Cook, Stability studies of stevioside and rebaudioside A in carbonated beverages, J. Agri. Food Chem. 31 (1983) 409–412. [4] G.T. Kroyer, The low calorie sweetener stevioside: stability and interaction with food ingredients, Lebensm. Wiss. Technol. (1999) 509–512. [5] S. Kumar, Method for recovery of stevioside, US Patent 4599403, 1986. [6] R.H. Giovanetto, Method for the recovery of stevioside from plant raw material, US Patent 4892938, 1990. [7] J.D. Payzant, J.K. Laidler, R.M. Ippolito, Method of extracting selected sweet glycosides from the Stevia rebaudiana plant, US Patent 5962678, 1999. [8] R. Shi, M. Xu, Z. Shi, Y. Fan, X. Guo, Y. Liu, C. Wang, B. He, Synthesis of bifunctional polymeric adsorbent and its application in purification of stevia glycosides, React. Funct. Polym. 50 (2002) 107–116. [9] E.P. Moraes, N.R.C.F. Machado, Purification of the extract of Stevia rebaudiana Bertoni through adsorption in modified zeolites, in: Proceeding of VI Seminar on Enzymatic Hydrolysis of Biomass, (SHEB), Maringá, Brazil, 1999. [10] I.C.C. Mantovaneli, E.C. Ferretti, M.R. Simoes, C.F. da Silva, The effect of temperature and flow rate on the clarification of the aqueous stevia-extract in a fixed-bed column with zeolites, Braz. J. Chem. Eng. 21 (2004) 449–458. [11] O. Cetinkaya, V. Gokmen, Assessment of an exponential model for ultrafiltration of apple juice, J. Food Process Eng. 29 (2006) 508–518. [12] P. Rai, G.C. Majumdar, S. DasGupta, S. De, Effect of various pretreatment methods on permeate flux and quality during ultrafiltration of mosambi juice, J. Food Eng. 78 (2007) 561–568. [13] M. Neifar, R. Ellouze-Ghorbel, A. Kamouin, S. Baklouti, A. Mokni, A. Jaouani, S. Ellouze-Chaabouni, Effective clarification of pomegranate juice using laccase treatment optimized by response surface methodology followed by ultrafiltration, J. Food Process Eng. 34 (2009) 1199–1219. [14] A. Cassano, E. Drioli, G. Galaverna, R. Marchelli, G. Di Silvestro, P. Cagnasso, Clarification and concentration of citrus and carrot juices by integrated membrane processes, J. Food Eng. 57 (2003) 153–163. [15] S.T.D. de Barros, C.M.G. Andrade, E.S. Mendes, L. Peres, Study of fouling mechanism in pineapple juice clarification by ultrafiltration, J. Membr. Sci. 215 (2003) 213–224. [16] F. Tasselli, A. Cassano, E. Drioli, Ultrafiltration of kiwifruit juice using modified poly(ether ether ketone) hollow fibre membranes, Sep. Purif. Technol. 57 (2007) 94–102.

[17] C. Rai, P. Rai, G.C. Majumdar, S. DasGupta, S. De, Mechanism of permeate flux decline during microfiltration of water melon (Citrullus lanatus) juice, Food Bioprocess Technol. 3 (2010) 545–553. [18] V.K. Jayanti, P. Rai, S. DasGupta, S. De, Quantification of flux decline and design of ultrafiltration system for clarification of tender coconut water, J. Food Process Eng. 33 (2010) 128–143. [19] P.J. Evans, M.R. Bird, A. Pihlajamäki, M. Nyström, The influence of hydrophobicity, roughness and charge upon ultrafiltration membranes for black tea liquor clarification, J. Membr. Sci. 313 (2008) 250–262. [20] A. Kumar, B.K. Thakur, S. De, Selective extraction of ()epigallocatechin gallate from green tea leaves using two-stage infusion coupled with membrane separation, Food Bioprocess Technol. (2011), doi:10.1007/s11947-011-0580-0. [21] W.S. Fuh, B.H. Chiang, Purification of stevioside by membrane and ion exchange processes, J. Food Sci. 55 (1990) 1454–1457. [22] S.Q. Zhang, A. Kumar, O. Kutowy, Membrane based separation scheme for processing sweeteners from Stevia leaves, Food Res. Int. 33 (2000) 617–620. [23] F.V. Silva, R. Bergamasco, C.M.G. Andrade, N. Pinheiro, N.R.C.F. Machado, M.H.M. Reis, A.A. Araujo, S.L. Rezende, Purification process of stevioside using zeolites and membranes, Int. J. Chem. Reactor Eng. 5 (2007) A40. [24] J. Vanneste, A. Sotto, C.M. Courtin, V. Van Craeyveld, K. Bernaerts, J. Van Impe, J. Vandeur, S. Taes, B. Van der Bruggen, Application of tailor-made membranes in a multi-stage process for the purification of sweeteners from Stevia rebaudiana, J. Food Eng. 103 (2011) 285–293. [25] R. Rautenbach, R. Albrecht, Membrane Processes, John Wiley, New York, 1986. [26] M.C. Porter, Handbook of Industrial Membrane Technology, Crest Publishing House, New Delhi, 2005. [27] Chhaya, C. Sharma, S. Mondal, G.C. Majumdar, S. De, Clarification of Stevia extract by ultrafiltration: selection criteria of the membrane and effects of operating conditions, Food Bioprod. Process. (2011), doi:10.1016/ j.fbp.2011.10.002. [28] Chhaya, G.C. Majumdar, S. De, Optimization of process parameters for water extraction of stevioside using response surface methodology, Sep. Sci. Technol. 2011.accepted. [29] S. Ranganna, Hand Book of Analysis and Quality Control for Fruit and Vegetable Products, Tata McGraw Hill Publishing Company Limited, New Delhi, 1986.