Chapter 5 Microfiltration A.S. GRANDISON, Department of Food Science and Technology, The University ot Reading, Reading RG6 6AP and T J. A FINNIGAN, Marlow Foods, Middlesbrough 5.1 INTRODUCTION Microfiltration (MF) is the oldest membrane technology, having been used several decades before the first industrial use of reverse osmosis(Glimenius, 1985). However, subsequent development of the technology has been slow. Until recently microfilters were operated in the dead-end mode and were exclusively of the depth-filter type which particles become trapped within the filter structure, but recent developments have led to membrane-type microfilters, with a narrow pore size distribution, which can be operated in the cross-flow mode. This has led to an increase in possible applications, ncluding clarification of fluids in the food and beverage industries, recovery of cells and cell debris in the biotechnology industries, and the treatment of wastes 5.2 THEORY, MATERIALS AND EQUIPMENT Like ultrafiltration (UF), MF is a pressure-driven process employing pressures consider- ably lower than reverse osmosis. In fact the distinction between U MF is Somewhe arbitrary and there is no distinction on purely theoretical grounds. The distinction lies in the size ranges of the materials which are separated. UF is considered to involve the processing of dissolved macromolecules, while MF involves separation of dispersed articles such as colloids, fat globules or cells. MF can be considered to fall between UF and conventional filtration, although there is overlap at both ends of the spectrum. A guide to the pore sizes used for MF could be 0.01-10 um 2. For many years, MF has been applied as a dead-end operation using highly microporous symmetric membranes of the depth-filter type. Such membranes retain particles and consequently result in the build-up of a filter cake. This reduces flow, and when the pressure drop has reached a certain level the membrane must be replaced or removed and regenerated. In addition, the presence of a filter cake radically alters the filtering characteristics, effectively acting as a prefilter which removes particles which
Chapter 5 Microfiltration A. S. GRANDISON, Departrnent of Food Science and Technology, The University of Reading, Reading RG6 6AP and T. J. A. FINNIGAN, Marlow Foods, Middlesbrough. 5.1 INTRODUCTION Microfiltration (MF) is the oldest membrane technology, having been used several decades before the first industrial use of reverse osmosis (Glimenius, 1985). However, subsequent development of the technology has been slow. Until recently microfilters were operated in the dead-end mode and were exclusively of the depth-filter type in which particles become trapped within the filter structure, but recent developments have led to membrane-type microfilters, with a narrow pore size distribution, which can be operated in the cross-flow mode. This has led to an increase in possible applications, including clarification of fluids in the food and beverage industries, recovery of cells and cell debris in the biotechnology industries, and the treatment of wastes. 5.2 THEORY, MATERIALS AND EQUIPMENT Like ultrafiltration (UF), MF is a pressure-driven process employing pressures considerably lower than reverse osmosis. In fact the distinction between UF and MF is somewhat arbitrary and there is no distinction on purely theoretical grounds. The distinction lies in the size ranges of the materials which are separated. UF is considered to involve the processing of dissolved macromolecules, while MF involves separation of dispersed particles such as colloids, fat globules or cells. MF can be considered to fall between UF and conventional filtration, although there is overlap at both ends of the spectrum. A guide to the pore sizes used for MF could be 0.01-10 pm. For many years, MF has been applied as a dead-end operation using highly microporous symmetric membranes of the depth-filter type. Such membranes retain particles and consequently result in the build-up of a filter cake. This reduces flow, and when the pressure drop has reached a certain level the membrane must be replaced or removed and regenerated. In addition, the presence of a filter cake radically alters the filtering characteristics, effectively acting as a prefilter which removes particles which
142 A.S. Grandison and T, J. A. Finnigan could otherwise pass through the membrane itself. On a large scale, therefore, it is only this technique when small amounts of particles are present Cross-flow MF (CMF) is a development which combines the cross-flow technique, as applied to UF and reverse osmosis, with MF CMF can be used to minimise (although it should be emphasised, not completely eradicate) the problems encountered in dead-ene MF, and thus permit the processing of fluids containing quite large amounts of suspended solids on a large scale. The advantages result from the fact that the build-up of filter cake is avoided due to the shearing effect of the feed stream flowing parallel to the membrane (Fig. 5. 1). CMf plants can be operated in the same batch or continuous modes described in Chapter 3 PERMEATE Membrane Feed FIG. 5.1. Principles of (a)'dead-end and (b)cross-tlow'filtration 5.2. 1 Membrane configurations and characteristics The geometric designs of MF membranes are the same as for UF as described in Chapter 3. Hence the module housings and ancillary equipment are also similar. Also the mem brane types are the same as for UF, i.e. cellulose and synthetic polymers(described in Chapter 3)or inorganic. It is notable that the development of inorganic membranes has been towards applications in MF rather than UF and reverse osmosis. In fact some type of inorganic membrane are only available with pore sizes in the MF range Various inorganic materials have been employed for membrane manufacture including glass, metals and compounds of aluminium, zirconium and titanium, and the geometries can vary radically from conventional membrane design The structures and methods of manufacture of inorganic membranes are described in greater detail by Rios et aL. (1989). Inorganic membranes consist of two parts-a macro porous support and the active membrane coated onto the surface. The supporting
142 could otherwise pass through the membrane itself. On a large scale, therefore, it is only practicable to use this technique when small amounts of particles are present. Cross-flow MF (CMF) is a development which combines the cross-flow technique, as applied to UF and reverse osmosis, with MF. CMF can be used to minimise (although it should be emphasised, not completely eradicate) the problems encountered in dead-end MF, and thus permit the processing of fluids containing quite large amounts of suspended solids on a large scale. The advantages result from the fact that the build-up of filter cake is avoided due to the shearing effect of the feed stream flowing parallel to the membrane (Fig, 5.1). CMF plants can be operated in the same batch or continuous modes as described in Chapter 3. A, S. Grandison and T. J. A. Finnigan FEED Filter cake 0 OOOc$ O O O AP{ lo+:0 **;," ' ' ' ' L +A Membrane PERMEATE (a) 0 000 0 - 0 O O+ 0 oo -0 Ooo0 0 C+SCC++++ Concentrate 0- Feed 0 00 Ap-- - __---_ - Membrane Permeate (b) FIG. 5. I. Principles of (a) 'dead-end' and (b) 'cross-tlow' filtration. 5.2.1 Membrane configurations and characteristics The geometric designs of MF membranes are the same as for UF as described in Chapter 3. Hence the module housings and ancillary equipment are also similar. Also the membrane types are the same as for UF, i.e. cellulose and synthetic polymers (described in Chapter 3) or inorganic. It is notable that the development of inorganic membranes has been towards applications in MF rather than UF and reverse osmosis. In fact some types of inorganic membrane are only available with pore sizes in the MF range. Various inorganic materials have been employed for membrane manufacture including glass, metals and compounds of aluminium, zirconium and titanium, and the geometries can vary radically from conventional membrane design. The structures and methods of manufacture of inorganic membranes are described in greater detail by Rios et al. (1989). Inorganic membranes consist of two parts-a macroporous support and the active membrane coated onto the surface. The supporting
Microfiltration 143 materials must drain away the permeate without any hydrodynamic resistance, and thus have a pore diameter of about 10 um or more. They are produced from sintered fine owdered materials including alumina, carbon, stainless steel and nickel. The tubular or multichannel geometries of modules(e.g. Fig. 5.2(a-c) are produced by extrusion of the Filtrate flows in conduits to downstream d of module Plugged passageways which isolate filtrate conduits from feed flow channels Opposite end has exactly ame appearance Passageways coated FEED
Microfiltration 143 materials must drain away the permeate without any hydrodynamic resistance, and thus have a pore diameter of about 1Opm or more. They are produced from sintered fine powdered materials including alumina, carbon, stainless steel and nickel. The tubular or multichannel geometries of modules (e.g. Fig. 5.2 (a-c)) are produced by extrusion of the Filtrate flows in conduits to downstream end of module
144 A. S. Grandison and T.J. A. finnigan Surface view Zirconia mesh support membranes sy of SFEC):(b) different designs of Ceraver alumina membranes (re n of Membralox):(c)alumina membrane module (reproduced by permission of CeraMem) ;(d)ceramic/metal-mesh composite membrane (reproduced by permission of Nww Acumem Ltd powder and binder in aqueous media. The membrane layer may be coated directly onto the macroporous support where pore size is quite large, but for UF and the smaller pore size MF membranes an intermediate sintered, ceramic layer is necessary due to the surface rugosity of the support. The membrane layer (usually composed of alumina, titania or zirconia) is formed by coating the support with a colloidal suspension and firing at a lower temperature than the firing temperature of the support. To prevent rapid flux decline, the membrane thickness must not exceed a few microns- titanium and zirco- nium membranes of thickness 3-5 Am have been achieved. Accurate control of the col lodal particle size allows the possibility of producing membranes with an extremely narrow pore size distribution compared to conventional membranes. The final pore size is also related to the sintering temperature. An example of the structure of a sintered ce- ramic membrane is shown in Fig 5.3
144 A. S. Grandison and T. J. A. Finnigan Surface view Cross-section I 1 I Nickellbased Zirconia meshsupport membranes Fig. 5.2. Some designs of inorganic membranes: (a) CARBOSEP membrane composed of zirconia on a carbon support (courtesy of SFEC); (b) different designs of Ceraver alumina membranes (reproduced by permission of Membralox@); (c) alumina membrane module (reproduced by permission of CeraMem); (d) ceramic/metal-mesh composite membrane (reproduced by permission of NWW Acumem Ltd). powder and binder in aqueous media. The membrane layer may be coated directly onto the macroporous support where pore size is quite large, but for UF and the smaller poresize MF membranes an intermediate sintered, ceramic layer is necessary due to the surface rugosity of the support. The membrane layer (usually composed of alumina, titania or zirconia) is formed by coating the support with a colloidal suspension and firing at a lower temperature than the firing temperature of the support. To prevent rapid flux decline, the membrane thickness must not exceed a few microns - titanium and zirconium membranes of thickness 3-5 pn have been achieved. Accurate control of the colloidal particle size allows the possibility of producing membranes with an extremely narrow pore size distribution compared to conventional membranes. The final pore size is also related to the sintering temperature. An example of the structure of a sintered ceramic membrane is shown in Fig. 5.3
Microfiltration 145 Microporous→ membrane Macroporous→一 Fig. 5.3. Electron micrograph showing structure of Ceraver sintered embrane with 0. 2 Am pore size(reproduced by permission of Mem Early work on inorganic membranes used glass, but the first reliable cross-flow system was the CARBOSEP membrane made of a microporous layer of zirconia coated onto a macroporous carbon support(Fig. 5. 2(a). New products have subsequently appeared including several designs of alumina membrane (e. g. Ceraver and CeraMem designs Figs. 5.2(b)and 5. 2(c). A novel design is the composite membrane produced by Acumen omposed of a zirconia ceramic membrane with nickel-based superalloy mesh support (Fig.5.2(d) The potential advantages of inorganic membranes result from their greater structural strength and resistance to abrasive degradation, as well as improved chemical and temperature properties. Their rigidity and strength allow the processing of feed materials, such as cheese curd(Mahaut et al., 1982)or particulate materials which would not be possible using conventional designs. Another possibility is that they could be used in conjunction with fluidised turbulence promoters to increase permeate flux, which would be out of the question with the more fragile surfaces of organic membranes. The wide ph ranges of inorganic membranes (e.g. alumina membranes are resistant to pHs ranging from 0.5 to 13.5, although phosphoric and hydrofluoric acids should be avoided) are a great advantage during cleaning and sterilisation. In-place cleaning regimes using high concentrations of caustic soda (3%), nitric acid (2%) and sodium hypochlorite are possible. The modules can withstand temperatures of several hundred oC, which is far beyond the temperatures used for food processing. However, this is an advantage during eaning cycles and the modules can be sterilised by steam. In practice, operating temperatures are limited by other components such as gaskets, but it could be feasible to
Microfiltration 145 Microporous - membrane Macroporous - support Fig. 5.3. Electron micrograph showing structure of Ceraver sintered alumina membrane with 0.2 pm pore size (reproduced by permission of Membralox@). Early work on inorganic membranes used glass, but the first reliable cross-flow system was the CARBOSEP membrane made of a microporous layer of zirconia coated onto a macroporous carbon support (Fig. 5.2(a)). New products have subsequently appeared including several designs of alumina membrane (e.g. Ceraver and CeraMem designs - Figs. 5.2(b) and 5.2(c)). A novel design is the composite membrane produced by Acumem composed of a zirconia ceramic membrane with nickel-based superalloy mesh support (Fig. 5.2(d)). The potential advantages of inorganic membranes result from their greater structural strength and resistance to abrasive degradation, as well as improved chemical and temperature properties. Their rigidity and strength allow the processing of feed materials, such as cheese curd (Mahaut et al., 1982) or particulate materials which would not be possible using conventional designs. Another possibility is that they could be used in conjunction with fluidised turbulence promoters to increase permeate flux, which would be out of the question with the more fragile surfaces of organic membranes. The wide pH ranges of inorganic membranes (e.g. alumina membranes are resistant to pHs ranging from 0.5 to 13.5, although phosphoric and hydrofluoric acids should be avoided) are a great advantage during cleaning and sterilisation. In-place cleaning regimes using high concentrations of caustic soda (3%), nitric acid (2%) and sodium hypochlorite are possible. The modules can withstand temperatures of several hundred "C, which is far beyond the temperatures used for food processing. However, this is an advantage during cleaning cycles and the modules can be sterilised by steam. In practice, operating temperatures are limited by other components such as gaskets, but it could be feasible to
146 A.S. Grandison and T.J. A. finnigan employ high temperatures to defoul individual inorganic membrane modules. The major isadvantage of inorganic membranes is that they are considerably more expensive than polymeric membranes, and while they have some potential advantages, their performance is still limited by concentration polarisation and fouling as with conventional systems MF has also been described using membranes produced from woven fabrics or metals 5.2.2 Performance of microfiltration systems and membrane fouling The description of flux rate and factors affecting flux rate in UF systems, presented in Chapter 4, applies equally to CMF systems. Engineering aspects and detailed modelling of MF systems are beyond the scope of this chapter, but have been dealt with elsewhere (e. g. Davis and Birdsell, 1987; Ripperger, 1988). In general the larger pore sizes of Mf membranes allow very high initial permeate fluxes, although these can decline very rapidly during processing. The flux decline progressively lessens until a near equilibrium filtration rate is observed in most cases(see Fig. 5. 4). At this point the shearing action of the cross-flow stream prevents further deposition of material. It is generally assumed that this is actually a state of dynamic equilibrium where particles are leaving and joining the concentration polarisation/fouling layer at the same rate. For any given application it is necessary to carry out experimental trials to establish the most suitable membrane typ and to optimise the pressure and flow characteristics Time Fig. 5.4. Permeate flux and the effect of backflushing As with all membrane processes the permeate flux during MF is limited by concentration polarisation and fouling(see Chapter 4), even with CMF where the formation of filtration cake is prevented or limited. Deep bed clogging of the pores is a particular problem during MF, due to the larger pore sizes. A paradoxical situation occur in CMF whereby flux may be maintained at a higher level when a tighter membrane (i.e. with lower pore size)is used. This arises when the pores are smaller than the suspended solids in the feed, so that the particles cannot penetrate the pores. An xample of this was described by Gatenholm et al.(1988), who compared MF with UF membranes for harvesting Escherichia coli cells from fermentation broths. The MF membranes showed such a high initial flux decline that the steady state flux was actually igher for the UF than MF. Also Le and Atkinson(1985) noted that permeate flux during CMF of cell suspensions using membranes with a range of pore sizes followed the order 0.45 um>0.6 um>0.2 um with respect to pore size. Clearly there is a need to choose
+j E & a i *- With backflush x\ zi i. \. \. \. <. +.- .-.-.-.-.-._._. Without backflush
Microfiltration 147 membranes of the appropriate pore size in relation to the characteristics of the particles upended in the feed. Unfortunately it is very difficult to predict the optimal pore size from first principles, and experimentation is required. However, a recent study of Tarleton and Wakeman(1993)has attempted to understand the relationship between pore size and particle size in flux decline. Clearly the particle size distribution is an important factor specially when even a small proportion of fines are present in the feed. Many anomalies remain, and further systematic studies are required A large amount of effort has been put into devising means of avoiding or counter- acting the effects of fouling in MF, and several promising techniques have evolved. The first requirement is to maintain high shear in the retentate and thus reduce the thickness of the boundary layer at the membrane surface and remove deposits from the membrane It is desirable that turbulent flow be maintained on the retentate side of the membrane Murkes(1989)has described a new high shear cross-flow method for Mf, and claimed an order of magnitude increase in flux rates over conventional CMF Forcing a quantity of permeate back through the membrane (i.e. backflushing )can ive deposited particles away from the membrane surface, as well as breaking up con centration polarisation and or gel layers in the bulk flow. This can lead to regeneration of permeate flux, but it is necessary to ensure that a higher volume of permeate is not used for the backflushing than is gained from the resulting increased permeate flux (Milisic and Bersillon, 1986). However, assuming the correct sequencing and volume of pulsations are used, the technique can be most beneficial(Fig. 5.4). An alternative method of backwashing is to use cleaning sequences consisting of blowing air through the mem branes-i.e. gas backflushing(Dietrich et al., 1988; Peters, 1989). This may prove beneficial with problem feed streams. Another approach to maintaining flux is to use the uniform transmembrane pressure' mode of operation. This requires simultaneous operation of a retentate pumping loop and a permeate pumping loop, adjusted so that the pressure drop across the membrane is small, and is uniform along the length of the mbrane. This system has allowed very high fluxes(500-7001 m-2h-)to be mair ained during processing of skimmed milk using ceramic membranes in the Bactocatcl system(see Section 5.3.1) The application of an electric field has been shown to reduce fouling due to colloids and particles during CMF (Visvanathan and Ben Aim, 1989)and thus prevent flux decline(Tarleton and Wakeman, 1988), although the mechanism of action is not full understood. Bowen et al. (1989)reported that electrical pulses were effective in both improving flux rates and restoring flux rates in fouled systems, and could be used as an alternative to back washing and conventional cleaning techniques Other approaches to reducing fouling during Mf have included the use of abrasives to break down the fouling layers, and the application of pulsations to the feed stream (Milisic and Bersillon, 1986). Chemical pretreatments of the feed streams can also sed to reduce the problems of fouling(Bedwell et aL., 1988), although this is frequently not a suitable option. Pretreatment of the membrane may reduce the flux decline during harvesting of yeast cells by conditioning the MF membranes with respect to pl x during fouling. Taddei and Howell (1989)reported a 70% improvement of flt The initial challenge to the membrane has been found to affect the subsequent flux during MF. Care must be taken during start-up procedures to prevent too-rapid flux
Microfiltration 147 membranes of the appropriate pore size in relation to the characteristics of the particles suspended in the feed. Unfortunately it is very difficult to predict the optimal pore size from first principles, and experimentation is required. However, a recent study of Tarleton and Wakeman (1993) has attempted to understand the relationship between pore size and particle size in flux decline. Clearly the particle size distribution is an important factor, especially when even a small proportion of fines are present in the feed. Many anomalies remain, and further systematic studies are required. A large amount of effort has been put into devising means of avoiding or counteracting the effects of fouling in MF, and several promising techniques have evolved. The first requirement is to maintain high shear in the retentate and thus reduce the thickness of the boundary layer at the membrane surface and remove deposits from the membrane. It is desirable that turbulent flow be maintained on the retentate side of the membrane. Murkes (1989) has described a new high shear cross-flow method for MF, and claimed an order of magnitude increase in flux rates over conventional CMF. Forcing a quantity of permeate back through the membrane (i.e. backflushing) can drive deposited particles away from the membrane surface, as well as breaking up concentration polarisation and/or gel layers in the bulk flow. This can lead to regeneration of permeate flux, but it is necessary to ensure that a higher volume of permeate is not used for the backflushing than is gained from the resulting increased permeate flux (Milisic and Bersillon, 1986). However, assuming the correct sequencing and volume of pulsations are used, the technique can be most beneficial (Fig. 5.4). An alternative method of backwashing is to use cleaning sequences consisting of blowing air through the membranes - i.e. gas backflushing (Dietrich et al., 1988; Peters, 1989). This may prove beneficial with problem feed streams. Another approach to maintaining flux is to use the ‘uniform transmembrane pressure’ mode of operation. This requires simultaneous operation of a retentate pumping loop and a permeate pumping loop, adjusted so that the pressure drop across the membrane is small, and is uniform along the length of the membrane. This system has allowed very high fluxes (500-700 1 m-* h-’) to be maintained during processing of skimmed milk using ceramic membranes in the ‘Bactocatch’ system (see Section 5.3.1). The application of an electric field has been shown to reduce fouling due to colloids and particles during CMF (Visvanathan and Ben Aim, 1989) and thus prevent flux decline (Tarleton and Wakeman, 1988), although the mechanism of action is not fully understood. Bowen et al. (1989) reported that electrical pulses were effective in both improving flux rates and restoring flux rates in fouled systems, and could be used as an alternative to backwashing and conventional cleaning techniques. Other approaches to reducing fouling during MF have included the use of abrasives to break down the fouling layers, and the application of pulsations to the feed stream (Milisic and Bersillon, 1986). Chemical pretreatments of the feed streams can also be used to reduce the problems of fouling (Bedwell et al., 1988), although this is frequently not a suitable option. Pretreatment of the membrane may reduce the flux decline during fouling. Taddei and Howell (1989) reported a 70% improvement of flux during harvesting of yeast cells by conditioning the MF membranes with respect to pH. The initial challenge to the membrane has been found to affect the subsequent flux during MF. Care must be taken during start-up procedures to prevent too-rapid flux
148 A S. Grandison and T.J. A. Finnigan cline. It is frequently an advantage to start the run in the dynamic mode (i.e. with water circulating prior to introduction of the feed), or with the permeate side filled with solvent While the abov ve methods of combating fouling can be ve effective. it is essential mbine them with efficient chemical cleaning procedures A characteristic of MF is that the separation may alter during processing due to fouling, such that the MF operation, in reality, becomes UF through a secondary membrane'formed at the surface of the MF membrane, consisting of macromolecules or lloids. This generally constitutes a problem. However, this property has been used to advantage in the French dairy industry for the concentration of milk. Bennassar et al (1982)described the use of the selective UF properties of the secondary membrane formed while using 0.2 um membranes. Almost total retention of milk protein was sible with this membrane 5. 3 APPLICATIONS IN THE FOOD AND BIOTECHNOLOGY INDUSTRIES MF is generally used to separate particles suspended in liquid media, and may frequently be considered as an alternative to conventional filtration or centrifugation. For industrial use the aim is usually to obtain either a clear permeate or the concentrate. Therefore most applications are either clarification, or the recovery of suspended particles such as cells or colloids, or the concentration of slurries. MF is a useful process for the treatment of fermentation broths, frequently as an alternative to centrifugation. During recovery of intracellular components, the cells can be enriched and washed by diafiltration prior to disintegration of the cells, and cell debris can be removed from the products of lysis MF membranes can be incorporated in water polishing systems as they are effective in removing suspended solids and bacteria. They may also be applied to a variety of liquid food streams, and may give the accompanying benefit of chemical and/or micro- biological stabilisation. Clarification of biochemical or microbiological reaction products or effluents is also possible. MF is a well-established laboratory technique for the production of sterile fluids without the application of heat 5.3. 1 Food industry Perhaps the majority of applications in the food industry have been in the treatment juices and beverages. As MF is a purely physical process, it can have advantages over traditional methods involving chemical additives, in terms of the quality of the product as ell as the costs of processing Finnigan and Skudder(1989)discussed the application of ceramic microfilters to the processing of beer and cider. Very good quality clear permeate was found for both products with high flux rates. Fluxes of 200-400 I m h were obtained with beer at low temperatures with no rejection of essential components. Recovery of beer from tank bottoms is also possible using ceramic MF, but this is less important in the UK now that duty is not paid at the wort stage of production Clarification and biological stabilisation of wine musts and unprocessed wine have been described. This avoids the requirement for fining and, possibly, pasteurisation. Dau et al.(1988)used membrane filters with pore size 0. 1-5 um to clarify wine preheated
148 decline, It is frequently an advantage to start the run in the dynamic mode @e. with water circulating prior to introduction of the feed), or with the permeate side filled with solvent. While the above methods of combating fouling can be very effective, it is essential to combine them with efficient chemical cleaning procedures. A characteristic of MF is that the separation may alter during processing due to fouling, such that the MF operation, in reality, becomes UF through a ‘secondary membrane’ formed at the surface of the MF membrane, consisting of macromolecules or colloids. This generally constitutes a problem. However, this property has been used to advantage in the French dairy industry for the concentration of milk. Bennassar et al. (1982) described the use of the selective UF properties of the secondary membrane formed while using 0.2 pm membranes. Almost total retention of milk protein was possible with this membrane. A. S. Grandison and T. J. A. Finnigan 5.3 APPLICATIONS IN THE FOOD AND BIOTECHNOLOGY INDUSTRIES MF is generally used to separate particles suspended in liquid media, and may frequently be considered as an alternative to conventional filtration or centrifugation. For industrial use the aim is usually to obtain either a clear permeate or the concentrate. Therefore most applications are either clarification, or the recovery of suspended particles such as cells or colloids, or the concentration of slurries. MF is a useful process for the treatment of fermentation broths, frequently as an alternative to centrifugation. During recovery of intracellular components, the cells can be enriched and washed by diafiltration prior to disintegration of the cells, and cell debris can be removed from the products of lysis. MF membranes can be incorporated in water polishing systems as they are effective in removing suspended solids and bacteria. They may also be applied to a variety of liquid food streams, and may give the accompanying benefit of chemical and/or microbiological stabilisation. Clarification of biochemical or microbiological reaction products or effluents is also possible. MF is a well-established laboratory technique for the production of sterile fluids without the application of heat. 5.3.1 Food industry Perhaps the majority of applications in the food industry have been in the treatment of juices and beverages. As MF is a purely physical process, it can have advantages over traditional methods involving chemical additives, in terms of the quality of the product as well as the costs of processing. Finnigan and Skudder (1989) discussed the application of ceramic microfilters to the processing of beer and cider. Very good quality clear permeate was found for both products with high flux rates. Fluxes of 200-400 1 m-2 h-’ were obtained with beer at low temperatures with no rejection of essential components. Recovery of beer from tank bottoms IS also possible using ceramic MF, but this is less important in the UK now that duty is not paid at the wort stage of production. Clarification and biological stabilisation of wine musts and unprocessed wine have been described. This avoids the requirement for fining and, possibly, pasteurisation. Dau et al. (1988) used membrane filters with pore size 0.1-5 pm to clarify wine preheated to
Microfiltration 149 30-40C to increase permeate flux. However, Rios et al. (1989)selected ceramic me branes with pore diameter 0. 2 um at temperatures <20.C to avoid sensory changes Conventional methods of clarification of fruit juices incorporate the use of diatomaceous earth and enzymes to remove a variety of materials including suspended solids, colloidal particles, proteins and condensed polyphenols. Also they are quite time consuming in that the juice must be stored in settling tanks, Juice clarification using MF is becoming an established commercial practice as it is a continuous process which can result in significant savings in time, materials and labour as well as improved yiel (Note: The use of diatomaceous earth results in loss of yield. Short (1988)described the commercial application of CMf to the processing of a range of fruit juices. A plant producing more than 5 x 10 litres per day of apple juice exists in the USA. This has the major benefit of increasing juice yield from 95 to 97%o as well as eliminating the require ment for diatomaceous earth as a pretreatment, and the use of enzymes and fining agents A saving of $350 000 per annum was estimated. With highly coloured juices, a further benefit of MF is that the colour is retained very well in the permeate. Commercial systems are also known to exist for grape, pear, kiwifruit, pineapple, cranberry and citrus juices. Clarifications of vinegar, vodka, pickling brines and liqueurs by Mf have also been described (Le and Billigheimer, 1985; Short, 1988) It seems likely that CMF could replace carbonatation in the treatment of raw suga eet juice, although this is not carried out commercially to the author's knowle Another section of the industry with several applications is dairy processing. Piot et al (1984)and Merin (1986) have clarified sweet cheese whey using CMF. This has the dual benefit of removing fat and reducing the bacterial population and could eliminate the need for fat separation and heat treatment in the production of whey protein powders orior to UF. The former authors reported that a decimal reduction of microorganisms of 5 ould be obtained in the microfiltrate compared to the whey, although some loss of whey protein was observed. Hanemaaijer(1985)described a scheme for whey treatment incor porating MF and UF to produce'tailor-made'whey products with specific properties for specific applications. The products include whey protein concentrates which are rich in whey lipids, as well as highly purified protein Bacterial removal from whole milk by MF is a problem because the size range of the bacteria overlaps with the fat globules, and to a lesser extent with the casein micelles However, some success has been achieved with skim milk. The 'Bactocatch'system can remove 99.6% of the bacteria from skim milk using ceramic membranes on commercial scale(Malmbert and Holm, 1988). The retentate(approximately 10% of the feed)can then be sterilised by a UHT process, mixed with the permeate and the mixture asteurised,to give a product with 50% longer shelf-life but no deterioration in organoleptic properties compared to milk that has only been pasteurised,, The combination of MF and heat reduces the bacterial numbers by 99.99%(4D) Alternatively the permeate could be used for cheesemaking, or the production of low-heat milk powder(Hansen, 1988). Piot et al.(1987)described the use of membranes of pore diameter 1.8 um to produce skim milk of low bacterial content. Recovery of fat from buttermilk has also been described (rios et al., 1989) Membranes have been used to concentrate milk prior to the manufacture of many cheese types. This results in improved yields and other associated benefits such as
Microfiltration 149 30-40”C to increase permeate flux. However, Rios et al. (1989) selected ceramic membranes with pore diameter 0.2 pm at temperatures <20”C to avoid sensory changes. Conventional methods of clarification of fruit juices incorporate the use of diatomaceous earth and enzymes to remove a variety of materials including suspended solids, colloidal particles, proteins and condensed polyphenols. Also they are quite time consuming in that the juice must be stored in settling tanks. Juice clarification using MF is becoming an established commercial practice as it is a continuous process which can result in significant savings in time, materials and labour as well as improved yields. (Note: The use of diatomaceous earth results in loss of yield.) Short (1988) described the commercial application of CMF to the processing of a range of fruit juices. A plant producing more than 5 x lo5 litres per day of apple juice exists in the USA. This has the major benefit of increasing juice yield from 95 to 97% as well as eliminating the requirement for diatomaceous earth as a pretreatment, and the use of enzymes and fining agents. A saving of $350 000 per annum was estimated. With highly coloured juices, a further benefit of MF is that the colour is retained very well in the permeate. Commercial systems are also known to exist for grape, pear, kiwifruit, pineapple, cranberry and citrus juices. Clarifications of vinegar, vodka, pickling brines and liqueurs by MF have also been described (Le and Billigheimer, 1985; Short, 1988). It seems likely that CMF could replace carbonatation in the treatment of raw sugar beet juice, although this is not carried out commercially to the author’s knowledge. Another section of the industry with several applications is dairy processing. Piot et al. (1984) and Merin (1986) have clarified sweet cheese whey using CMF. This has the dual benefit of removing fat and reducing the bacterial population and could eliminate the need for fat separation and heat treatment in the production of whey protein powders prior to UF. The former authors reported that a decimal reduction of microorganisms of 5 could be obtained in the microfiltrate compared to the whey, although some loss of whey protein was observed. Hanemaaijer (1985) described a scheme for whey treatment incorporating MF and UF to produce ‘tailor-made’ whey products with specific properties for specific applications. The products include whey protein concentrates which are rich in whey lipids, as well as highly purified protein. Bacterial removal from whole milk by MF is a problem because the size range of the bacteria overlaps with the fat globules, and to a lesser extent with the casein micelles. However, some success has been achieved with skim milk. The ‘Bactocatch’ system can remove 99.6% of the bacteria from skim milk using ceramic membranes on commercial scale (Malmbert and Holm, 1988). The retentate (approximately 10% of the feed) can then be sterilised by a UHT process, mixed with the permeate and the mixture pasteurised, to give a product with 50% longer shelf-life but no deterioration in organoleptic properties compared to milk that has only been pasteurised. The combination of MF and heat reduces the bacterial numbers by 99.99% (4D). Alternatively the permeate could be used for cheesemaking, or the production of low-heat milk powder (Hansen, 1988). Piot et al. (1987) described the use of membranes of pore diameter 1.8 pm to produce skim milk of low bacterial content. Recovery of fat from buttermilk has also been described (Rios et al., 1989). Membranes have been used to concentrate milk prior to the manufacture of many cheese types. This results in improved yields and other associated benefits such as
150 A.S. Grandison and T J. A. Finnigan reduced requirement for rennet and starter, and the ability to produce much more cheese per vat( Grandison and Glover, 1994). In most cases UF membranes have been used as it is necessary to retain all the protein in the concentrate, An alternative approach is to oncentrate the curd after coagulation of the milk in which case a solution of lactose and tion in the manufacture of some soft cheese type el. This can be done using centrifuga minerals is removed from the semi-solid protein However, the use of MF is an attractive alternative Rios et al. (1989) have carried out extensive trials on this application and concluded that the use of 0. 2 um pore diameter membranes gave a product with better texture and yield than with centrifugation. The choice of ceramic membranes allowed the curd to be contacted directly with the membrane Other food applications have been reported with meat and vegetable products includ ing the following Devereux and Hoare (1986) described the use of Mf to recover precipitated so protein. This could have advantages over recovery of the dissolved protein using UF. Gelatin is a proteinaceous material derived by hydrolysis of collagen. This is purified y filtration incorporating diatomaceous earth. The latter process can be replaced by CMF which effectively removes dirt, coagulated proteins, fats and other particulate materials from the feed. Again the CMF method gives higher yields of high quality product on a continuous basis. Short (1988)calculated that incorporating CMF plants for gelatin would have a payback time of 3 years for a capacity of 30 tonnes/h 5.3. 2 Applications for biotechnology Applications of MF in the biotechnology industry are very promising although the process has not yet made the breakthroughs that may have been expected. The major problem with MF of suspensions of cells or cell debris is the exponential flux decay resulting from adhesion of cells or cell fragments to the membrane -i.e. biofouling. This has limited the application of CMf into biotechnological downstream processing Solutions to this problem, other than general methods of avoiding fouling, are discussed by Defrise and Gekas(1988)and include choice of biocompatible membrane materials and the use of surfactants and polyelectrolytes. Most biotechnological applications of MF are as a competitor to centrifugation, and mF is becoming recognised as a viable alterna ive in many cases. CMF may have advantages over centrifugation when containment is required(e. g. when handling pathogenic organisms or in recombinant DNA technology) as aerosols are not produced during the operation( Kroner et al., 1984). The capital and maintenance costs of MF are lower than for centrifugation, although membrane replace ment is expensive. Tutunjian(1985) compared the costs of process-scale harvesting and washing of Escherichia coli cells using MF with a hollow fibre system to centrifugation The MF system had about 70% lower capital, and 25%o lower running costs than centrifugation. Also virtually 100%o recovery of solids is possible with MF, whereas it is often less than 90% with centrifugation CMF is a versatile technique for cell harvesting and cell debris removal during recovery of intracellular products. The viability and enzyme content of the cells is unaffected by MF, and it is possible to concentrate cell suspensions to 20-25% dr weight, limited by pumping considerations (Le and Atkinson, 1985). A 0 I um
150 reduced requirement for rennet and starter, and the ability to produce much more cheese per vat (Grandison and Glover, 1994). In most cases UF membranes have been used as it is necessary to retain all the protein in the concentrate. An alternative approach is to concentrate the curd after coagulation of the milk in which case a solution of lactose and minerals is removed from the semi-solid protein gel. This can be done using centrifugation in the manufacture of some soft cheese types. However, the use of MF is an attractive alternative. Rios et al. (1989) have carried out extensive trials on this application and concluded that the use of 0.2 pm pore diameter membranes gave a product with better texture and yield than with centrifugation. The choice of ceramic membranes allowed the curd to be contacted directly with the membrane. Other food applications have been reported with meat and vegetable products including the following. Devereux and Hoare (1986) described the use of MF to recover precipitated soya protein. This could have advantages over recovery of the dissolved protein using UF. Gelatin is a proteinaceous material derived by hydrolysis of collagen. This is purified by filtration incorporating diatomaceous earth. The latter process can be replaced by CMF which effectively removes dirt, coagulated proteins, fats and other particulate materials from the feed. Again the CMF method gives higher yields of high quality product on a continuous basis. Short (1988) calculated that incorporating CMF plants for gelatin would have a payback time of 3 years for a capacity of 30 tonnes/h. 5.3.2 Applications for biotechnology Applications of MF in the biotechnology industry are very promising although the process has not yet made the breakthroughs that may have been expected. The major problem with MF of suspensions of cells or cell debris is the exponential flux decay resulting from adhesion of cells or cell fragments to the membrane - i.e. biofouling. This has limited the application of CMF into biotechnological downstream processing. Solutions to this problem, other than general methods of avoiding fouling, are discussed by Defrise and Gekas (1988) and include choice of biocompatible membrane materials and the use of surfactants and polyelectrolytes. Most biotechnological applications of MF are as a competitor to centrifugation, and MF is becoming recognised as a viable alternative in many cases. CMF may have advantages over centrifugation when containment is required (e.g. when handling pathogenic organisms or in recombinant DNA technology) as aerosols are not produced during the operation (Kroner et d., 1984). The capital and maintenance costs of MF are lower than for centrifugation, although membrane replacement is expensive. Tutunjian ( 1985) compared the costs of process-scale harvesting and washing of Escherichia coli cells using MF with a hollow fibre system to centrifugation. The MF system had about 70% lower capital, and 25% lower running costs than Centrifugation. Also virtually 100% recovery of solids is possible with MF, whereas it is often less than 90% with centrifugation. CMF is a versatile technique for cell harvesting and cell debris removal during recovery of intracellular products. The viability and enzyme content of the cells is unaffected by MF, and it is possible to concentrate cell suspensions to 20-25% dry weight, limited by pumping considerations (Le and Atkinson, 1985). A 0.1 pm A. S. Grandison and T. J. A. Finnigan