Concentration of macromolecules from aqueous solutions: A new swellex process

ChemicaIEnginrering Scimcr. Vol. 47, No. 1, pp. 39,1992. Printedin Great Britain.

lloos2509/92 s5.m + 0.00 ~1991i%rgmoaRuspk

CONCENTRATION OF MACROMOLECULES AQUEOUS SOLUTIONS: A NEW SWELLEX

FROM PROCESS+

M. V. BADIGER, M. G. KULKARNI and R. A. MASHELKAR’ Polymer Science and Engineering Group, Chemical Engineering Division, National Chemical Laboratory, Pune 411 008, India

(Receivedfur publication 16 May 1991) Abatrati-A new swellex process (extraction through swelling polymers) to concentrate macromolecules from aqueous solutions has been demonstrated. Dilute solutions of biological macromolecules such as proteins have been concentrated with good recoveries and high selectivity.The selectivity was controlled by

the extent of crosslinking in the gels. Electrolyte solution when pulsed through the column, causes the collapse of the gel. The reversible volume phase transition has been exploited for the regeneration of the gels._

In this paper we report the use of a lightly crosslinked, sulphonated poly(styrene+divinyl benzene) gel for the concentration of macromolecular solutions. The gels have much higher swelling capacity (100-200 g/g) than the conventional ion-exchange resins (0.1-1.0 g/g) used in water treatment. This process, which makes use of the highly swellable ionexchange resins is named as “swellex process”. The term “swellex” signifies the absorption of the solvent by a superabsorbent polymer which undergoes swelling and excludes the macromolecular solutes depending upon their molecular size and mesh size of the polymer. For conventional processes such as solvent extraction, the capacities are high but selectivities are low. For membrane separations or adsorptive separations, the capacities are low but the selectivities are high. The swellex process, on the other hand, can successfully exploit the most desirable features of the solvent extraction and membrane separation/adsorption processes, viz. high capacity and high selectivity. An additional goal of this work was to explore the possibility of developing a semicontinuous process for the concentration of macromolecular solutes. This would necessitate a column operation. It is then obvious that the gel materials to be used have to be rigid so as to retain their mechanical integrity, minimize the attrition losses and avoid excessive pressure drops during the operation. The gels used in the past (Vartak et al., 1983, Cussler et al., 1984) lack these desirable features. During the process of concentration, the gel beads swell and selectively absorb low molecular weight solutes and water and exclude the macromolecular solutes. The process could use any polymeric gel which would possess the desired swelling characteristics and the mechanical rigidity necessary for the operation. By varying the composition of the gels we have synthesized swellex gels of specific mesh sixes (c) and examined their performance in terms of the concentration factor, percentage recovery and selectivity. The regeneration of the gel was effected by pulsing the electrolyte solution through the column.

INTRODUCTION

Significant advances have been made in biotechnology during the last two decades. However, a major problem in the biotechnological route for the manufacture of the products is the need to handle very dilute solutions of products and the need to attain highest purity levels. The separation/purification processes involved are, therefore, intensive in terms of energy and capital. Traditionally, a number of membrane separation processes, viz. reverse osmosis, ultrafiltration, microfiltration, etc. have found applications in biotechnological processes. The relative merits, limitations and commercial applications of each of these have been well documented (Lonsdale, 1982). In most cases, the economic viability of the process depends on the separation and purification costs. It is, therefore, not surprising that a major research effort is underway to develop innovative bioseparation proCeSseS.

In a communication from this laboratory, Vartak et al. (1983) reported the use of poly(acrylate) gels for the concentration of a wide range of proteins from their aqueous solutions. A distinctive feature of these materials was their ability to absorb very large quantities of water ( N 100 g/g) and low molecular weight solutes along with the simultaneous exclusion of the macromolecular solutes. Since the aim of the work reported by Vartak et al. (1983) was essentially to demonstrate the concept, no effort was made to regenerate the gels. Cussler and co-workers (1984) exploited very elegantly a wide range of gels for the concentration of a variety of macromolecules from their dilute solutions. A fundamental framework to elucidate the factors controlling the swelling and deswelling of these gels was provided. Novel strategies for the regeneration of the gels induced by pH changes (Gehrke et al., 1986) and temperature changes (Freitas and Cussler, 1987) were also proposed.

‘NCL Communication No. 4767. ‘Author to whom correspondence should be addressed. 3

M. V. BADICER etal.

4

Swelling behaviour: theoretical

predictions

A three-dimensional,polymeric network in contact with the solvent, exhibits swelling. The extent of swelling attainable in these systems can be predicted on the basis of Flory’s theory (Flory, 1953). When the polymeric network comes in contact with the solvent molecules, the polymeric chains between the crosslinks attain strained conformations, which gives rise to a retractive force. The swelling continues till an equilibrium is reached when the swelling force is balanced by the retractive force. The swelling equilibrium is dependent on the entropy of dilution, the heat of dilution and the entropy of the polymer network (Flory, 1953). The equilibrium swelling ratio (q), defined as the ratio of the equilibrium swollen volume to the original unswollen volume of the network is given by the equation,

q5’3 = (iQkfJV,)(l

- zM,/A#,)-‘(*

- x)

(1)

where 4 is the equilibrium swelling ratio, x is the polymer-solvent interaction parameter, M, is the molecular weight between crosslinks, Vi is the molar volume of the solvent, c is the specitic volume of the polymer and Mm is the number average molecular weight of the uncrosslinked polymer. The above relationship correlates the swelling ratio q with the extent of crosslinking and the quality of the solvent. Equation (1) has been extensively used to predict the swelling ratios of non-ionic networks. In most of the non-ionic network systems, the value of the maximum swelling ratio is of the order of ten. The network system used in this work contains ionizable groups. The electrostatic repulsion between the ionizable groups leads to a further expansion of the network, which results in higher swelling ratios (q N 200). However, in electrolyte solutions, the electrostatic forces are screened which causes deswelling. Equation (2) describes the swelling ratio of an ionic polymer network in an electrolyte solution (Flory, 1953). 4 ‘I3 = C(i2/4Fk&J + (t -

iWpvLl

x C~CIU - =4,/~,)1.

(2)

Here i is the degree of ionization of the polymer network, I,, is the ionic strength of the electrolyte and V, is the molar volume of the structural repeat unit of the gel. f and. pI, are specific volume and the density of the gel, respectively. An examination of these equations reveals that the degree of swelling can be controlled by the degree of ionization (i) of the polymer network, ionic strength (I,,) of the medium, the polymer-solvent interaction parameter (x) and the molecular weight between crosslinks (M,). The addition of an electrolyte reduces the extent of swelling markedly. This provides a means of controlling the swelling and deswelling of the gels. In order to predict selectivity, which is related to the ability of the gel to sieve the macromolecules, we need to know the mesh size of the network.

Bar-Howell and Peppas (1985) deduced that the mesh size of the network can be approximated by the formula c = vz- 1/3y0 where y,, denotes the end-to-end perturbed state and is given as

(3) distance in the un-

Yo = 0;

(4)

y,=IJ=

(5)

N = LZUJM~.

(6)

where

and

In the above equations y, is the end-toend distance in the freely jointed chain, 1 is the C-C bond length (1.54 A), C. is the characteristic ratio (for polystyrene, C, = lo), rZ is the number of links per repeat unit (A = 2 in our case), A4, is the molecular weight of the structural repeat unit (for polystyrene sulphonic acid, A& = 184) and iU, is the molecular weight between crosslinks. The mesh sizes were estimated from the above equations.

EXPERIMENTAL

Synthesis The gels were prepared by the copolymerization.of styrene and divinyl benzene (DVB) followed by sulphonation using concentrated sulphuric acid. The detailed procedures are described elsewhere (Badiger, 1988). The styrene monomer was supplied by SD. Fine Chemicals, India. Benzoyl peroxide was obtained from Loba Chemie Co., India, and DVB was supplied by Fluka AG, as 50% solution in ethylvinylbenzene. It was used as such. Other reagents were purified as per the standard procedures (Perrin et al, 1980). The polymerizations and sulphonations were carried out at 60°C and 98°C respectively. The products were isolated in the form of beads. The degree of sulphonation was determined by titrating 1 g of sulphonated polymer in excess water against standard 0.1 N sodium hydroxide solution.

Swelling ratio measurements The swelling measurements were carried out as follows. A measured amount of the dry gel was placed in a glass cell provided with a stopper, sintered glass disc and a dram valve. The solvent was poured in the cell and allowed to stand in the closed condition for 24 h during which equilibrium was reached. The excess solvent was drained out and the total weight of the swollen gel was determined from which the swelling ratio was estimated.

Separation of macromolecules The macromolecular solutes such as caesein, dextran, bovine serum albumin, egg albumin and insulin

5

Concentrationof macromoleculesfrom aqueous solutions were obtained from Sigma Chemical Co., U.S.A. and used as such The dry swellex gel was thoroughly washed with (70/30) mixture of acetone and water ,and dried in an oven at 60°C. The dried gel was sterilixed at 120°C for 15 min. Then the requisite quantity of the dry gel was packed in a glass column fitted with a sintered disc at the bottom. The dilute macromolacular solutions were passed through the column by a peristaltic pump. After 30 min the raffinate was collected and solute concentration was estimated at appropriate wavelength+ on a spectrophotometer (SHIMADZU W-240). RESULTS

AND DISCUSSION

Characterization of swellex gels We have determined the swelling ratios of the swellex gels as a function of electrolyte concentration and compared the results with the predictions of eq. (2) (Fig. 1). At this stage we need to explain in detail the calculations of swelling ratios from eqs (I) and (2), in particular the terms x and i. For the system x = 0.43 (Brandrup polystyrene-toluene, and Immergut, 1966). The value of x( = 0.3) for the polystyrene sulphonic acid-electrolyte solutions was obtained from the values of the second virial coefficients reported by Orolino and Flory (1967). The number average molecular weight for polystyrene was determined by measuring intrinsic viscosity, [n] in toluene at 30°C. From the relationship [n] = KM:, with K = 9.2 x 10m5 dl/g and cc = 0.75. The value of M,, obtained was 6.17 x lo4 g/mol. The ionic strength (I,) of the electrolyte (NaCl) solution was determined by using the relation ze = $r, c,z:, where ci and zI are the concentration and valency of the ion of the electrolyte, respectively. Since the degree of substitution of sulphonic acid groups was 75.0%) and the sulphonic acid groups are completely ionized under the experimental conditions, the value of i used in the equation is 0.75. The characteristic parameters of the swellex gel such as swelling ratio (q), average molecular weight between crosslinks (A4,) and mesh size (c) were estimated by using Flory’s theory. The results are summarized in Table 1. It is observed that, on sulphonation, the crosslinked hydrophobic poly(styrene-divinyl ben-

160

c

f_

.

zone) gels containing lower amounts of DVB become highly hydrophilic and exhibit superabsorption characteristics. The swelling ratios rangod from 20 to 150 depending on the degree of crosslinking and the degree of slphonation. However, beyond a speci8c degree of sulphonation (69.63 SOsH groups/100 styreno units) the swelling ratio becomes independent of the degree of sulphonation (Fig. 2). Tho average molecular weight between cross links (M,) of the gels varied from 655 to 18,116 depending on the degree of crosslinking. In order to verify these results, A4, values of the same samples were determined by measuring their glass transition temperatures (T,) on a differential scanning calorimeter (DSC-2, Perkin-Elmer). Both MC and Tp. depend upon the degree of crosslinking and are related by the following empirical equation (Nielson, 1970): 3.94 x 104 M’=(q-

Here TV0and T, are the glass transition temperatures of the uncrosslinked and crosslinked polymers. The results obtained were found to be comparable to the II& values estimated on the basis of Flory’s theory of swelling (see Table 2). Concentration of macromolecular solutes The factors which need to be considered while examining the actual effectiveness of a separation Table 1. Characterizationof polystyrene-DVB networks DVB content NO.

(%I

I 2

.

6 Ekctrolyte

’

’ 10

“2

M

e 6,

: :

z-t 1:o

0.3

0.125 0.098 0.250 0204

18,116 13,759 3979 6137

196.3 158.3 67.2 89.3

5

2.0

0.487

655

21.8

Experimental

Predicted

by eq. (2)

I

1001

(7)

GJ’

’ 14

concentration Wl1) (x

4 16

.

’

22

1 O-‘)

Fig. 1. Comparison of experimentaland theoreticalvalues of swellingratios of swellexgels in NaCl solution.

O

1

40

I

60

1

80

I

100

Degree of ~ulphonstion (SO&l groups/l00 styrene units)

Fig. 2.. Effect of the degreeof sulphonationon swellingratios of poly(styrenesulphonicacid) gels.

’

M. V.

6

BADIQBR

process based on superabsorbing polymers uis-a-vis other conventional separation processes, would be (1) percentage recovery, (2) extent of separation, (3) fluxes, (4) size exclusion, and (5) ease of regeneration. The performance of the swellex gels in terms of these parameters is discussed in subsequent sections. Macromolecular solutes having a wide range of molecular weights (6 x lo”-6 x E06)were concentrated by using these gels. Recoveries artd extent of separation The effectiveness of the swellex process can be seen from the data on concentration of macromolecular solutions (Table 3). Very good recoveries with 5-lofold concentrations were achievedThe low recoveries for polyethylene glycol (PEG) and insulin may be attributed to their small sixes, because of which they might diffuse in the gel matrix along with the solvent. Fluxes

The fluxes (in terms of grams of water removed per unit area and unit weight of the gel) have been calculated and compared with typical fluxes obtained in ultrafiltration processes. Conventionally the transfer area assumed in the calculation of the fluxes corresponds to the active membrane area across which the transfer occurs. In the case of processes based on gels it is not easy to define the exact surface area. However, by knowing the diameter ( N 1.5 mm) and the density (1.04 g/cm3)of the gel particle one can determine the area per unit weight of the gel. For the gels used in this work the area of the gel/unit weight was found to be 40.0cm"/g.It must be mentioned here that the comparison of fluxes would not be very Table 2. Comparison of M, values of crosslinked polystyrene gels estimated by swelling and T# measurements DVB content W)

0.0 0.4 0.7 1.0

et

ai.

meaningful since the fluxes can be enhanced by increasing the pressure drop in the UF module or by decreasing the particle size of the gel. However, we have tried to give a qualitative comparison by quoting the fluxes in g/cm’ min atm under the identical molecular weight cut-off values for the membrane and the gel (Fig. 3). Selectivity The most important parameter which determines the efficiency of a separation process is the selectivity. In our system the selectivity depends upon the mesh size of the gel which is governed by the degree of crosslinking in the gels (Fig. 4). We have examined the selectivity of the swellex process by synthesizing gels of different mesh sixes to concentrate proteins of different molecular weights. The results are summarixed in Table 4. It can be seen that biomolecules such as egg albumin and lysozyme could be concentrated only by gels of low mesh sixes and the recoveries were also low. 100% selectivity would be possible only if all the mesh sizes were the same and were below the size of the macromolecular solute being separated. Ideally, if the crosslinks were uniformly -placed then the mesh

t

5

0.1:

.

.E E “E

Swellrx

S!

Ultrafiltration

z 0.01

Me values (??)

100.0’ 102.5 108.7 111.2

T, method

!

z!zY

-

-

13,759 6137 3979

15,600 4482 3444

t T,, = lOO.O”Cfor uncrosslinked polystyrene.

0.001~ 100

@

I

1000 10.040 Molecular weight cutoff

Fig. 3. Dependence of flux on the molecular weight cut off. Ultrafiltration uis-d-ois swellex process.

Table 3. Macromolecular separation by the swellex process Macromolecule Casein Dextran Bovine serum albumin (BSA) Egg albumin Poly(ethylene glycol) (PEG) Insulin

Molecular weight 115,000 71,500 68,000 45,080 9cuio 6000

!

Concentration factor

4.8 1.7 4.6 4.4 4.0 1.9

‘Recovery (%) = loo(mcasured increase in concentration)/(increase volume change).

Recovery’ (%)

Flux (g/cm2 min atm)

94.6 96.8 98.6 98.5 80.8 60.8

0.048

0.024 0.048 0.050 0.032 0.016

in concentration expected due to

Concentrationof macromoleculesfrom aqueous solutions Table 4. Mcctivity

in the swellex process -1s

MUleCul&r weight

Pcotcins Casein Dextran Bovine serum albumin (=A) Egg albumin Lysozyme

7

18,116

M)

13,759

6137

3979

655

115,ooo 71,ooo

:

z

+ +

+ +

+ +

zz 14:tMo

: 0

: 0

A 0

A 0

+ + +

0: concentration nat effected. +: concentration effected.

5 x s

18-

-180

14-

-140

lo-

-100

B-

-60

2-

-20

0

0.4

0.8

1.2

1.6

2.0

3 s

2.4

WE (%I Fig. 4. The molecular weight between crosslinks and mesh sizes of the network as a function of DVB content.

size would be identical too. In practice it is difficult to achieve this since there is a distribution of crosslinks in the gel, which gives a distribution in &f, and c. For example, in the case of swellex gels, the difference in

the reactivity ratios of styrene and divinyl benzene gives rise to a network size distribution (Regas and Valkanas, 1984). Therefore, in order to get uniform distribution of crosslinks in the network, newer techniques of synthesis have to be evolved. Recently some

gel by temperature variation is possible only when the gel exhibits a lower critical solution temperature (LCST). However, if the LCST is high then one has to collapse the swollen gel at higher temperature. The time required for the regeneration of the gel by an electrical field induced collapse is very long (4-5 days) and therefore this technique is not desirable for largescale operation. In the case of the swellex process, the collapse of the gels with a change in pH is not as dramatic as compared to hydrolysed poly(acrylamide) gels (Fig. 5). Therefore one has to look for an alternative approach to regenerate these gels. From the study of the effect of an electrolyte solution on swelling behaviour of swellex gels, it is apparent that the swellex gel can be regenerated by collapsing it in an electrolyte solution and draining the desorbed liuid (Fig. 6). During the process, the salt is washed off and the gel is ready for the next cycle. Figure 7 shows the influence of regeneration on the ability of the gel beads to regain their absorption characteristics. Up to 75% of the original absorption

2oor

modified processes with new crosslinking agents have been reported (Andreopoulos, 1987). Regeneration A crucial test of the viability of the swellex process is the ability to regenerate and recycle the gel easily. A

strategy for regeneration can be worked out on the basis of phase transitions in gels. The collapse of the swollen gel can be brought by change in the solvent composition (Tanaka, 1981), variation of pH (Cussler et al., 1984). temperature (Freitas and Cussler, 1987), by the addition of the electrolytes (Ohmine and Tanaka, 1982) or by subjecting the swollen gel to an electric Aeld (Tanaka et al., 1982). In the present case the use of non-solvent for collapse and regeneration of the gel is not desirable since it adds another step, viz. solvent recovery in the process. Additionally, the activity of certain biomolecules may be lost due to the presence of the non-solvent. The regeneration of the

‘1

.-. 3

/i

5

7

8

11

13

16 ,

PH

Fig. 5. Swelling ratios as a function of pH for swellex and hydrolysedpoly(acrylamide) gels.

M. V. BADIGERet al.

\ Concentrated

protein

solution

Electrolyte solution

t Swollen beads s-l

Collapsed ,bds

Dilute protein solution

EIect&ylya> + r

watar

electrolyte can be calculated if the external mass transfer coefficient (k,) in the bed is known. We can approximately estimate t,,, to be of the order of D/k:, where D is the dilfusivity of the electrolyte in water (lo- 5 c&/s). Assuming k, = lo- 3 cm/s, we obtain telu= 10.0 s. On the other hand, L,~ depends on the cooperative diffusion coefficient of the polymeric network. Experimentally it was estimated to be of the order of 2000 s. We thus have, t,,&, = 10e2. It is therefore clear that the time of electrolyte solution. pulsing coul@ be suitably chosen to ensure an elegant regencratron facility. It should be however noted that for specific applications the equilibrium swelling ratio will have to be optimized against the friability of the gel beads. The use of macroporous resins would be desirable.

Industrial application of the swellex process

Fig. 7. Regeneration cycles by electrolyte pulsing (NaCl).

We wanted to explore the possiblity of using the swellex technique for a commercial bioseparation problem. We could explore this in collaboration with an industry. In the preparation of vaccines (such as infectious bronchitis, IB, or egg drop syndrome, EDS) the dilute virus solution (allantoic fluid), isolated from the chick or the cell culture, contains lot of water and useful proteins. Separation and removal of water is a crucial aspect of vaccine preparation. The swellex process was tried in an industrial organisation to concentrate the dilute virus solution. Prior to the use of the swellex technique, this organisation used dialysis for this purpose in a routine way. Dialysis was known to take about 12 to 15 h for concentrating 100 ml fluid. By making use of the swellex process, 100 ml dilute virus solution was concentrated to 10 ml in 30 min. The concentration and the activity in terms of haemagglutination (HA) was estimated before and after the concentration. The initial activity of 1: 256 HA and the final activity of 1: 2406 clearly indicated that the process of concentration can be effected without any loss in the activity.

capacity can be regained by such cycling. The loss of 25% capacity is presumably due to the retention of small amounts of electrolyte. We considered a strategy for preventing the loss of capacity. The electrolyte could be removed provided one was able to elute it with water. On the other hand, since the polymers are superabsorbents, there is a possibility of the polymer swelling again appreciably and thereby reducing the column capacity. One has to therefore work out a strategy to solve this problem. The time required for the elution of the adsorbed electrolytes by water (t,,,) should be compared to the rate of absorption of water in the gel (t&. The solute could be leached out without appreciable swelling, provided t,,,,/~t,, 4 1. These time scales were estimated as follows. The time required for the elution of this surface adsorbed

In the foregoing we have described a new swellex separation process. The method appears to have many advantages over some of the efforts described in the literature. The swellex gels possess good mechanical strength in the swollen state as compared to the other gels used in the past and therefore can be used in a column operation. The semicontinuous column operation is amenable to scale-up. The gels used in the experiment are size selective and their mesh sixes can be tailor made so that the desired selectivity could be achieved. It is demonstrated that biological macromolecules, having a wide range of molecular weights (103-10s), such as proteins and enzymes, can be concentrated by using this process. High fluxes with excellent recoveries could be achieved without loss in the activity of biomolecules. A new regeneration strategy based on dilute electrolyte solution pulsing

Absorption

Regeneration

Fig. 6. Regeneration of gel by electrolyte solution (NaCl).

r

+ Collapsed state 0 Swollen stat

No. of cycles

CONCLUDING REMARKS

Concentration of macromolecules from aqueous solutions

has been develowd. The swellex urocess annears to h&e a considerable potential in ironcentratiig biological macromolecules both at the laboratory scale as well as on the industrial scale. NOTATlON

characteristic ratio degree of ionization ionic strength of the electrolyte average molecular weight between crosslinks number average molecular weight swelling ratio ’ time for absorption time for elution glass transition temperature molar volume of the solvent volume fraction of the network molar volume of the structural repeat unit of the gel Greek YO

:

a!

letters

end-to-end distance in the unperturbed state specific volume of the polymer mesh size of the network, A polymer-solvent interaction parameter REFERENcEs

Andmopoulos, A. G., 1987. New crosslinking agents for vinyl polymers. J. appl. Polyp. Sci. 34,2384-2397. Bediger, M. V., 1988, Transport phenomena in polymeric

9

media.Ph.D. thesis, University of Bombay. India. Bar-Howell, B. D. and Peppas, &J.A., 1985,~Dynamics and equilibrium swelling of DVB-crosslinked polystyrene partick J. appl. Polyp Sci. 30,4583-4589. Brandrun. J. and Immeraut. E. H.. 1966. Polvmer Handbook. Indmce Publisbe