Membranes for industrial microfiltration and ultrafiltration
In this chapter a general overview of industrial microfiltration (MF) and ultrafiltration (UF) processes is given. Basic principles, membrane materials, membrane preparation technologies, membrane module configurations and process designs involved in these operations are described. Membrane fouling and concentration polarization phenomena, as well as methods to control and reduce their effects on membrane performance, are also discussed. Finally the main applications of UF and MF membranes and their integration on a large technical scale are reviewed.
Microfiltration (MF) and ultrafiltration (UF) are typical pressure-driven membrane operations that can be considered, together with reverse osmosis (RO), to be well-established industrial separation technologies. These processes are based on the use of polymeric membranes and represent a valid approach to solving separations problems involving particulate material and macromolecules. They are a valid alternative to competitive separation processes like distillation, extraction, fractionation, and adsorption, since the separation process is athermal and involves no phase change or chemical agents. Additional advantages are the relative simplicity of operation, low energy consumption, easy scale-up, low weight and space requirements, modularity and the possibility of carrying out the separation continuously (Bilstad, 1997).
The utilization of MF and UF processes on an industrial scale is relatively recent, spanning only 40 years. Up until 1945, membrane filters were used primarily for removal of microorganisms and particles from liquid and gaseous streams, for sizing of macromolecules and for culturing bacteria cells. The development of a RO cellulose acetate membrane with an asymmetric structure in the early 1960s (Loeb and Sourirajan, 1962) is considered to be a milestone in membrane science and technology. Soon, other synthetic polymers were used as basic materials for synthetic membranes. UF was an offshoot of RO and really came into use only in the 1960s when MF on a smaller scale was already well established. The first commercially successful industrial UF system was installed in 1969 by Abcor to recover electrocoated paint from automobile paint shop rinsewater (Goldsmith et al., 1971).
In both MF and UF processes the separation mechanism is mainly based on molecular sieving using porous membranes with increasingly fine pores (size exclusion). In general, MF is used to separate suspended particles with diameters between 0.1 and 10 μm from a fluid mixture; this size range includes a wide variety of natural and industrial particles. Typically solutes separated by MF are larger than those separated by UF and RO; consequently, the osmotic pressure for MF is negligible. Also, the membrane pore size and permeate flux are typically larger for MF than for UF and RO (Ho and Sirkar, 1992).
Membranes used in UF are characterized by pore sizes in the range 1–100 nm capable of retaining species in the molecular weight range 300–500 000 Da. Typical rejected species include dissolved macromolecules (such as proteins, sugars, polymers, biomolecules) and colloidal particles, whilst solvent and salts will pass through the membrane.
The driving force for transport across MF and UF membranes is a pressure differential which forces the suspending fluid and small solutes to pass through the membrane where they are collected as permeate; particles retained by the filter medium are collected as retentate. Because MF and UF membranes do not typically reject salts, osmotic pressure differentials are small compared to RO; consequently, the operative pressure required is relatively small (typically 1–10 bar).
The rejection characteristics of MF and UF membranes are usually expressed as nominal molecular weight cut-off (MWCO), defined as the smallest molecular weight species for which the membrane has more than 90% rejection. The permeate flux through the filter medium is affected by the applied pressure difference across the membrane, the resistance of the membrane and the viscosity of the fluid being filtered. The volumetric flux is given by:
where P is a permeability coefficient (mol m− 1 s− 1 Pa− 1), Δp the hydrostatic pressure difference (Pa), Δπ the osmotic pressure difference (Pa) between feed and permeate phases and l the membrane thickness (m).
MF and UF membranes can be used both in dead-end and cross-flow mode. In the former, the feed solution is forced perpendicularly through the membrane under pressure; rejected particles accumulate on the membrane surface and the pressure needed to maintain the required flow increases, until at some point the membrane must be replaced. In the cross-flow configuration, feed material sweeps tangentially across the membrane surface producing a clean, particle-free permeate and a retentate containing the particles. The equipment is more complex but flux rates and membrane lifetime are maximized if compared to the dead-end filtration.
Materials for fabrication of commercial MF and UF membranes include synthetic polymers (polypropylene, perfluoropolymers, polyamides, polysulphones, etc.), cellulose derivatives, ceramics, inorganics and metals (Table 20.1). Membrane materials must be chemically resistant to both feed and cleaning solutions, mechanically and thermally stable and characterized by high selectivity and permeability.
Hydrophilic materials are not suitable for MF and UF membranes since water molecules act as plasticizers affecting mechanical strength and thermal stability. On the other hand, crystalline polymers show high chemical resistance and thermal stability since the crystalline domain contributes to the effect of cross-linking between amorphous domains and hinders the free rotation of polymer segments (Osada and Nakagawa, 1992). MF membranes generally consist of crystalline polymers. Polytetrafluoroethylene (PTFE), poly(vinylidene fluoride) (PVDF) and polypropylene (PP) are typical hydrophobic materials commonly used as MF membranes. Hydrophobic membranes are also widely used to minimize adsorption phenomena which reduce permeate fluxes and create difficulties in membrane cleaning.
Aliphatic polyamides such as nylon 4-6, nylon-6 and nylon 6-6, also widely used as MF membranes, are characterized by good chemical, thermal and mechanical stability. Amorphous polymers are generally used to produce UF membranes owing to their convenient regulation and control of small pores size. Polymeric materials which have a high glass transition temperature are generally employed. In particular, polyacrylonitrile (PAN), widely used for general aqueous systems because of its resistance to solvents and chemicals, also exhibits relatively low protein binding owing to its hydrophilic properties. Polysulphone (PS), also widely used in the production of UF membranes, has good mechanical strength and good resistance to heat and pH; in contrast it exhibits poor resistance to solvents.
In terms of structure, MF and UF membranes can be either symmetric or asymmetric. Symmetric membranes are characterized by identical structure and transport properties over the entire cross-section of the membrane. In asymmetric membranes structural and transport properties vary over the membrane cross-section; typically, a relatively dense thin selective layer (0.1–1 μm) is supported by a much thicker (100–200 μm) porous substrate. The mass flux is determined mainly by the thickness of the selective layer, whereas the porous sub-layer provides mechanical strength and has little effect on the separation characteristics of the membrane. MF membranes can be prepared by several methods: slip coating-sintering, stretching, phase inversion and track-etching.
Thermoplastic polymers can be melted and extruded through a die producing a macroporous membrane whose pore structure is induced by stretching the material. The partially crystalline polymeric material is stretched perpendicularly to the direction of extrusion so that crystalline regions are located parallel to the direction of extrusion. Under mechanical stress, small slit-like ruptures (generally 0.2 μm in length and 0.02 μm in width) occur in the membrane (Kesting, 1985). Pore diameters of the produced membrane are in the range 0.1–3 μm, while the porosity can approach 90%. PP and PTFE are materials typically used in the production of these membranes.
Ceramic membranes for UF and MF made from aluminium, titanium or silica oxides are generally realised by a slip coating-sintering process (Strathmann et al., 2006). In this process a porous ceramic support tube is made by pressing a dispersion of a fine-grain ceramic material and then heating at high temperature. By using the correct sintering temperature, the interface between the particles disappears producing a porous structure. One surface of the tube is then coated with a suspension of finer particles in a solution of a cellulosic polymer or polyvinyl alcohol which acts as a binder and viscosity enhancer to hold the particles in suspension. When this mixture, called a slip suspension, is dried and sintered at high temperature, a finely porous surface layer remains. Several slip-coated layers formed by suspensions of progressively finer particles can be applied in series producing an asymmetric structure. Pore diameters in ceramic membranes for MF and UF are in the range from 0.01–10 μm.
An ideal form of porous membrane is a dense polymer with cylindrical pores. Such membranes can be produced by track-etching (Fleischer et al., 1969). This procedure involves irradiation of a thin polymer film (about 10 μm thick) with fission particles from a nuclear reactor or other radiation source: the highly energetic ions pierce the polymeric film and break the polymer chains leaving ‘tracks’ in the membrane material (tracking). In the etching step, the tracked film is immersed in an acid (or alkaline) solution in which tracks are converted into cylindrical pores with a uniform diameter and a narrow pore size distribution. The exposure time of the film to radiation controls the number of membrane pores (pore density), while the etching time determines the pore diameter which can range between 0.2 μm and 10 μm. Membranes prepared by the track-etching procedure are symmetrical and the permeate flux is proportional to the membrane thickness; consequently, they have to be thinner than asymmetric microporous membranes in order to have a comparable flux. Polycarbonate or polyester films are the usual materials used for track-etched membranes. The porosities of these membranes are of the order of 10%.
Most polymeric MF and UF membranes are prepared by the phase inversion process (Baker, 2000). In this process a colloidal dispersion (sol) is converted, by removal of solvent, into a swollen three-dimensional macromolecular network (gel) which ultimately forms a solid matrix, the membrane. Thus the phase inversion process can be considered a sol–gel transition. The dope mixture (i.e. casting solution) contains the polymer, the solvent (which is the most volatile component), and may include a swelling agent and/or a non-solvent as well as ingredients, such as lithium or magnesium salts, that influence the kinetics of the phase separation process. After a specific period of maturation, the casting solution is cast as a film either as a flat sheet on a smooth surface (i.e. a glass plate) or inside a porous tube or extruded through a spinnerette as a hollow fibre. Following a period of evaporation of the solvent, the cast film is transferred to a non-solvent gelation bath (usually a water bath) where exchange occurs between solvent and non-solvent leading to polymer precipitation. The procedure produces a liquid phase forming the membrane pores and a solid phase forming the membrane structure, which may be either symmetric or asymmetric (Strathmann, 1985).
The performance of phase inversion membranes, in terms of flux and selectivity, is affected by different parameters such as polymer concentration, evaporation time before immersion, humidity, temperature, composition of the casting solution and the coagulation bath composition.
In the Accurel process developed by Enka (Schneider, 1982) PP is mixed with N,N-bis(2-hydroxyethyl)hexadecylamine to form a binary system, which is miscible at 150 °C but immiscible at lower temperatures (50 °C). Phase separation is obtained by extruding the note dope, as films, tubes or fibres into a cooled bath of the hexadecylamine. This phase inversion technique, named thermal phase inversion, produces a symmetric microporous membrane with a relatively narrow distribution of effective pore size.
MF and UF membranes used at the industrial level are installed in a device generally called a membrane module. The choice of the membrane module depends on different parameters such as the production cost, packing density, energy consumption and especially the control of concentration polarization and membrane fouling. On a large industrial scale, membrane modules are available in five basic designs: hollow fibre, spiral wound, tubular, plate and frame and capillary (Fig. 20.1). They are quite different in their design, mode of operation, production costs and energy requirement for pumping the feed solution through the module.
In the following the basic concepts of the principal module types used in cross flow MF and UF are discussed. The design of the plate-and-frame module (Fig. 20.1(a)) has its origin in the conventional filter press concept. Membranes, feed flow spacers and porous permeate support plates are layered together between two endplates and placed in a housing. The sheets are either in the form of circular discs, elliptical sheets or rectangular plates. The feed mixture is pressurized in the housing and forced across the membrane surface. A portion passes through the membrane, enters the permeate channel and makes its way to a central permeate collection manifold. Plate-and-frame units are mainly used for small-scale applications (production of pharmaceuticals, bioproducts or fine chemicals); these units, however, are quite expensive and the membrane replacement is labour intensive. They are used in a limited number of UF applications with highly fouling feeds. The feed channels are often less than 1 mm thick and, although more sensitive to fouling, are easier to clean as no mesh support is used.
The tubular membrane module (Fig. 20.1(b)) consists of membrane tubes surrounded by porous paper or fibreglass supports. The pressurized feed solution flows internally along the tubes and the permeate is collected on the outer side of the porous support. Tube diameters are in the range 1–2.5 cm and a number of tubes are placed in one pressure housing in order to increase the module productivity. Tubular modules can also be made by welding flat sheet membranes that are cast on a mechanically strong porous polyester support. These tubes, which have a diameter of 0.5–1 cm, do not require an additional support if operated at pressures of less than 4 bar. They are typically used in UF at low hydrostatic pressure.
Concentration polarization and membrane fouling phenomena can be easily controlled in tubular membrane modules; moreover, plugging of the membrane module can be avoided even with feed solutions containing a very high concentration of solid matter. On the other hand the low surface area that can be installed in a given unit volume and the high costs are the main drawbacks of the tubular design. Consequently, tubular membranes are generally used when feed solutions containing high solid content and with high viscosity have to be treated (i.e. industrial effluents or treatment of solutions coming from the food and pharmaceutical industries).
The capillary membrane module (Fig. 20.1(c)) is essentially constituted by a large number of membrane capillaries with an inner diameter of 0.5–3 mm arranged in parallel as a bundle in a shell tube. The feed solution is pumped into the lumen of the membranes and the permeate is collected in the shell tube. Fibres are open at both ends with the tube-fibre bank sealed in epoxy blocks. UF capillary membranes are available in a wide range of polymers including PS, PAN and chlorinated polyolefins.
The capillary membrane module is characterized by a high membrane area per module volume and low production costs; another advantage is the possibility of controlling concentration polarization and membrane fouling through a physical process called back flushing in which the permeate flow is reversed allowing the fouling material to be dislodged from the membrane surface. The main drawbacks of the capillary membrane module are the required low operating pressure which cannot exceed values of 4–6 bar owing to the limited stability of the membranes.
Hollow fibre membrane modules (Fig. 20.1(d)) consist of a bundle of several membrane fibres with the free ends potted with an epoxy resin into a cylindrical housing or shell. Fibres with diameters in the range 50–200 μm are usually called fine hollow fibres. The selective layer is on the outside of the fibres where the feed fluid is applied while the permeate is removed down the fibre bore. Such fibres can withstand very high hydrostatic pressures applied from the outside and they are typically used in RO desalination of sea water or in high-pressure gas separation. When the fibre diameter is between 200 μm and 500 μm, the feed solution is commonly applied to the inside bore of the fibre and the permeate is removed from the shell side. These fibres are used for low-pressure gas separation and for UF applications. Fibres with diameters greater than 500 μm are called capillary fibres.
The hollow fibre membrane module is characterized by the highest packing density compared with other configurations and its production is very cost effective. Its drawbacks are related to the difficulty in controlling concentration polarization and membrane fouling. Consequently an extensive pretreatment of the solution is needed in order to remove particles, macromolecules or other materials which can precipitate at the membrane surface.
The spiral wound membrane module (Fig. 20.1(e)) is characterized by a simple design consisting of an envelope containing the feed flow channel spacer, the membrane and a porous membrane support rolled around a perforated central collection tube; the module is placed inside a tubular pressure housing made from stainless steel or polyvinyl chloride (PVC). The feed solution passes axially through the feed channel across the membrane surface. The permeate moves along the permeate channel and is collected in the collection tube. Small spiral wound modules for laboratory applications consist of a single envelope wrapped around the collection tube. These units have a membrane area in the range 0.2–1 m2. Industrial scale modules are realised according to a multi-leaf arrangement with several membrane envelopes, each one with an area of 1–2 m2, wrapped around the central collection tube. This design permits the pressure drop encountered by the permeate fluid moving down the permeate channel to the central pipe to be minimized. In particular, in a single membrane envelope the path taken by the permeate to reach the collection tube would be several metres long resulting in a large pressure drop in the permeate collection channel. UF spiral wound modules are produced in standard module sizes and have a diameter of 0.05, 0.1 and 0.2 m and lengths between 0.15 and 1.2 m corresponding to membrane areas up to 15 m2 depending upon channel width. The spiral wound configuration is compact and provides a relatively large area per unit volume that is relatively inexpensive but prone to particulate fouling (thus pre-filtration is needed). Commercially available membrane modules for UF and MF applications are summarized in Table 20.2.
Selection of the correct membrane in MF and UF applications is one of the most important factors for achieving a desired separation. However, the process design is equally important since it determines: (1) the technical feasibility and the economics of the process, (2) the achievable recovery rate, (3) the maximum concentration of a component in the retentate and (4) the lifetime of the membranes (the process design is an important factor for the control of concentration polarization and membrane fouling). In these processes the driving force in transporting specific compounds through a semi-permeable membrane is the difference in hydrostatic pressure applied between the two sides of the membrane. As a result the feed solution is converted into a permeate and a retentate stream according to the scheme reported in Fig 20.2. The performance of a MF or UF membrane is expressed in terms of permeate flux and separation properties. They are a function of the membrane permeability for different compounds in the feed solution and the applied hydrostatic pressure. In addition, they are also a function of the process design.
The separation capability of a MF or UF membrane is expressed as membrane rejection according to Equation [20.1]. However, the concentrations in the retentate and permeate streams depend not only by the membrane rejection but also by the recovery rate (Δ) which is given by:
where Vp = Vp(t) and V0 are the permeate volume and the initial feed volume (m3), respectively, and t is the time. The recovery rate ranges between 0 and 1. Sometimes data are also presented as volume concentration ratio (VCR):
The concentration of a solute at any time in the process is a function of both volume concentration ratio and rejection according to Equation [20.5]:
According to Equation [20.7], the yield of a specific compound is an exponential function of the decreasing volume of the feed in the system (Cheryan, 1998). The most common process configurations in MF and UF are:
A schematic representation of these operation modes is reported in Fig. 20.3.
20.3 Schematic diagram of different operation modes in MF and UF processes: (a) total recycle, (b) batch concentration with full recycle of retentate, (c) batch concentration with partial recycle of retentate, (d) single-pass continuous filtration, (e) feed-and-bleed operation, (f) diafiltration, (g) multistage filtration.
In the total recycle configuration (Fig. 20.3(a)), the permeate and retentate streams are returned to the feed reservoir and no net concentration of particles occurs. This configuration is mainly used in order to study the effect of different operating parameters on the permeate flux.
In the batch concentration configuration, the retentate is returned to the feed reservoir and the permeate is collected separately (full recycle of retentate) (Fig. 20.2(b)). This approach requires the minimum membrane area and one pump can be used for both feed and recirculation. The batch concentration with partial recycle of retentate (Fig. 20.3(c)) configuration is used when a continuous feedstream has to be processed. Batch operations are used when the permeate is the product of interest such as in fruit juice clarification or treatment of effluents in which the retentate has to be discharged. In this approach the residence time of particles within the system is longer than in other configurations.
In single-pass processing (Fig. 20.3(d)), the feed solution is continuously fed into the filtration device: in most cases the recirculation flow rate is higher than the permeate flux. This approach results in a very low permeate volume and low recovery rate unless a very large membrane area is used. In this case the residence time of particles in the system is minimized. Higher recovery rates can be obtained by applying the feed-and-bleed operation mode in which the permeate is removed from the system together with a small part of the retentate (Fig. 20.3(e)). Most of the retentate is recycled and mixed with the feed solution to maintain a high tangential velocity in the membrane module. A feed pump and a recirculation pump are required to provide the transmembrane pressure and the cross-flow, respectively. If a component is not completely retained by the membrane, its concentration is proportional to the retentate concentration and is increased by increasing the recovery rate. Since the process loop is operated continuously at a concentration factor equivalent to the final concentration of a batch system, the permeate flux is lower than the average flux in a batch mode and the membrane area required is correspondingly higher.
In diafiltration the retentate is recycled to the feed reservoir and the permeate is replaced by an equal volume of pure solvent (Fig. 20.3(f)). This configuration is often used when more complete separation of micro- and macrosolutes is required (Porter and Michaels, 1971). Diafiltration is used, for example, to remove salts from a mixture containing macromolecular compounds such as proteins. The variation of the permeate and retentate concentration with time during the diafiltration process is given by:
where Cip and Cir are the concentration of the component i in the permeate and in the feed tank, and V0 and Vp the volumes of the feed tank and the permeate, respectively. Introducing Vp = Vw and Cip = (1 - R) Cir into Equation [20.8], and integrating over a certain time period, the concentration in the feed tank will be given by:
where Cir and are the concentrations in the feed tank at time t (s) and at the beginning of the filtration, V0 is the volume of the feed tank, Vw is the volume of the solvent added during time t and R the rejection of the membrane towards the component i. The ratio Vw/V0, named the diavolume coefficient, is a parameter used to evaluate the diafiltration process.
Finally, the multistage recycle operation (Fig. 20.3(g)) permits the low flux disadvantage of the feed-and-bleed operation to be overcome. In this case, a multitude of filtration devices is arranged in series and the individual devices are referred to as stages. Only the final stage operates at the highest concentration and lowest flux, while the other stages operate at lower concentrations with higher fluxes. This permits the total membrane area to be reduced in comparison with a single stage operated in a feed-and-bleed configuration.
Concentration polarization and membrane fouling are typical phenomena affecting the performance of MF and UF processes. The general effect of these phenomena is a reduction in the permeate flux through the formation of an additional barrier caused by the retention of feed solution compounds with a consequent increase in the mass transfer resistance.
When a feed solution containing a solvent and a solute or suspended solids is filtered through a porous membrane, some components permeate the membrane under a given driving force while others are retained. This results in a higher local concentration of the rejected solute at the membrane surface, compared to the bulk, with formation of a viscous or gelatinous cake layer. This phenomenon is referred to as concentration polarization (Aimar et al., 1989; Cheryan, 1998; Mulder, 1991). It is not to be confused with the membrane fouling phenomenon that is essentially caused by deposition of retained particles onto the membrane surface or in the membrane pores.
Two mechanisms have been proposed to explain the reduction in flux in polarization-limited systems. In the former, the increased solute concentration at the membrane surface leads to an increase in the osmotic pressure and thus to a decrease in the driving force (PT-Δπ) and flux at constant applied pressure. This mechanism can be considered valid in RO systems in which the feed solutions have a considerable osmotic pressure. In UF and MF, only macromolecules are retained by the membrane: consequently, the osmotic pressure of the feed solution is generally small. Owing to the high molecular weight of the retained compounds, their diffusion back from the membrane surface to the bulk is relatively low. According to the second mechanism, the retained components are precipitated to form a solid layer at the membrane surface. This consolidated layer affects the performance of the process by both reducing the membrane flux and modifying the rejection towards lower molecular weight compounds.
A mathematical model generally accepted to describe the concentration polarization in MF and UF processes is the well-known film theory (Fig. 20.4). This relates the solute concentration at the membrane surface to that in the bulk and permeate solutions, the membrane flux and fluid flow conditions in the boundary layer between membrane surface and bulk solution. The film model assumes that the solute is brought to the membrane surface by convective transport at a rate Js (mol m− 2 s− 1) defined as:
20.4 Schematic diagram of the concentration polarization phenomenon according to the film model (Cg is the solute concentration at the membrane surface, CP is the solute concentration in the permeate and Cb is the solute concentration in the bulk solution).
According to Equation [20.13], the concentration polarization is expressed as the ratio of the solute concentration at the membrane surface and the solute concentration in the bulk. It is a function of the membrane flux Jv, the boundary layer thickness δ(m), the solute diffusion coefficient in the boundary layer D and the membrane solute rejection R.
Introducing k into Equation [20.13] leads to:
The mass transfer coefficient can be estimated by using the following general equation (Sherwood et al., 1965):
where Sh, Re and Sc are the Sherwood, Reynolds and Schmidt numbers, respectively, L is the length (m) of the feed flow channel (or tube), dh is the hydraulic diameter (m) of the feed channel and a, b, c and d are characteristic constants for different geometries to be experimentally determined.
In general, Re values less than 1800 are considered to be laminar flow and Re greater than 4000 is turbulent flow. The film model can be used to describe the concentration polarization in both the turbulent and laminar flow regimes. In the turbulent flow regime entrance effects can be neglected: in this case the exponent d in Equation [20.16] is zero. In the laminar flow regime entrance effects cannot be neglected and the Sherwood number is expressed as (Bird et al., 1965):
The film model is not always applicable in MF and UF because the feed solutions contain macromolecular compounds or suspended particles with low diffusion coefficients; consequently, the diffusive mass transport of retained compounds from the membrane surface back into the bulk solution is slower than the convective mass transport towards the membrane. When the solute concentration at membrane surface exceeds the solubility of the feed constituents it can precipitate forming a solid layer on the membrane surface. This layer results in an additional hydrodynamic resistance to the membrane flux. In these conditions, if the transmembrane pressure is increased, more solutes are transported towards the membrane increasing the layer thickness at the membrane surface without a corresponding increase in flux. The experimental evidence of this concept is illustrated in Fig. 20.5 referred to the UF of a macromolecular solution, such as a kiwifruit juice (Cassano et al., 2008). Here permeate fluxes in the steady state are reported as a function of the applied transmembrane pressure at different feed flow rates. At low pressures, the permeate flux is proportional to the applied pressure. When the particles start to deposit on the membrane surface the rate of increase of flux decreases. Further increases in pressure determine an increase in the thickness of the particle layer and when the concentration polarization layer reaches a limiting concentration, the flux becomes independent of the pressure. The gel-polarized layer is assumed to be dynamic. Higher flow rates tend to remove the deposited material reducing the hydraulic resistance through the membrane: consequently, higher permeate fluxes can be obtained.
Membrane fouling is the major limiting step in MF and UF processes. The term fouling is referred to a long term flux decline caused by the deposition of retained particles (colloids, suspended particles, macromolecules, etc.) onto the membrane surface and/or within the pores of the membrane. The fouling behaviour is specific to the system of interest and is strongly affected by the physicochemical nature of the membrane, the solutes and the fluid dynamic system design.
Membrane fouling can also be a consequence of concentration polarization phenomena. Gel or cake layer formation may be caused by inorganic precipitates (such as CaSO4 and metal hydroxides), organic materials (such as proteins, lipids, humic acids and other macromolecular materials) and biological components (such as microorganisms and products of their metabolism). However, membrane fouling can be also caused by adsorption of specific substances onto the membrane surface owing to hydrophobic interactions, van der Waals force attractions or electrostatic forces (Strathmann et al., 2006).
Both polarization concentration and fouling determine a reduction of permeate flux in MF and UF processes. However, they can have an opposite effect on the observed rejection: the concentration polarization determines a reduction in the rejection. In the case of fouling, if the build-up of solids on the membrane is significant enough, it may act as a secondary membrane and change the effective sieving and transport properties of the system: consequently, the rejection can be increased or maintained constant. While the concentration polarization is a reversible process based on diffusion taking place over a few seconds, fouling is generally irreversible and the flux decline is a long-term process. A constant flux is not generally reached at all. Finally, the concentration polarization can be described by a simple mathematical model and minimized by means of hydrodynamics, such as the feed flow velocity and the membrane module design; in contrast, the control and description of membrane fouling are more difficult.
The attachment of biological materials, such as proteins, organic acids, polysaccharides and microorganisms to membrane surfaces is recognized as biofouling. A general model accepted for bacterial fouling in water systems includes the attachment of organisms to a specific surface, followed by the adsorption of simple organic molecules that fuel further metabolic actions and a final step in which a polysaccharide, named glycocalyx, is generated stabilizing the microorganism–surface interaction. The consequences of biofouling are in terms of loss of performance, incresase in energy input and costly cleaning. Methods generally accepted for controlling and minimizing fouling phenomena can be summarized as: feed pretreatment, modification of membrane properties, modification of operating conditions, flow manipulation and membrane cleaning with the correct chemical agents.
Pretreatment methods for feed solutions include coagulation, sedimentation, precipitation, prefiltration, pH adjustment, chlorination and carbon adsorption. The modification of membrane properties is another approach for reducing the extent of fouling. For example, hydrophobic membranes, which strongly adsorb proteins, can be modified by introducing hydrophilic characteristics. This can be achieved by mixing hydrophilic and hydrophobic polymers, or by pretreatment of hydrophobic membranes with hydrophilic surfactants or enzymes or by surface modification (Strathmann et al., 2006). Surface modifications include plasma treatment, polymerization or grafting of the surface initiated by heat, chemicals or UV light, interfacial polymerization and introduction of ionic or polar groups by chemical reaction.
Process parameters such as temperature, flow rate, pressure and feed concentration have great influence on membrane fouling. Generally, an increasing temperature results in an increasing permeate flux. However in some cases, such as in the UF of cheese whey, a flux decreasing at a temperature lower than 30 °C has been observed owing to a decrease in the solubility of calcium phosphate (Cheryan, 1998). In addition, in most UF applications concerning biological systems, the adsorption of proteins generally increases with temperature. In many cases, an increase in the shear rate generated at the membrane surface tends to shear off deposited material and thus reduces the hydraulic resistance of the fouling layer. When the polarization layer reaches a limiting concentration, an increase in operating pressure does not improve the performance of the system and the flux remains practically unchanged. High pressures may also cause severe fouling in MF and UF processes owing to compaction of the fouling layer.
Removal of foreign material from both the membrane surface and membrane pores is defined as membrane cleaning. Typical systems used for membrane cleaning are based on mechanical, hydraulic, chemical or electrical methods. Hydraulic cleaning is generally performed by backflushing conducted by pumping permeate back into the feed channel to lift deposited material off the membrane surface. In most cases, backflushing is done with permeate for 1–5 s at a frequency of 1–10 times/min at 1–10 bar. Backflushing is effective with many kinds of foulants, in particular with materials that are not linked to the membrane as adherent films.
Mechanical cleaning is generally obtained by forcing foam balls down tubular modules at high velocities. The balls create turbulence at the membrane surface dislodging fouling materials. Electrical cleaning methods are based on the use of electrical pulsing which is able to remove charged species from the membrane surface. Special modules are required to introduce the charge to the membrane surface which is generally realised in metallic form.
The most important method of controlling and minimizing membrane fouling is chemical cleaning. It aims to solubilize or disperse the foulant or soil. The choice of cleaning agent depends upon the type of foulant, type of membrane and the severity of fouling. Typical cleaning agents are acids, alkalis, detergents, enzymes, complexing agents and disinfectants. Alkaline solutions containing NaOH or KOH, sometimes supplemented with hypochlorite, are particularly effective for solubilizing fats and proteins. Sodium carbonate, soda ash or phosphates are used as buffering compounds to control the pH during the cleaning. Mineral acids, polyacrylates and ethylenediaminetetraacetic acid (EDTA) are used essentially to remove salt precipitates and mineral scalants. Phosphoric acid is largely used because it is not very aggressive; in addition, it also has a detergent action because of the phosphate groups. Citric acid combines acidity with detergency and chelating ability and is preferred, especially for iron removal. Blends of acids may be also particularly effective. HCl and sulphuric acid are very corrosive and should be avoided.
Enzymatic cleaners based on either proteases, amylases, glucanases, and so on, are used at neutral pH for specific instances followed by sanitization with an oxidizing solution such as sodium hypochlorite or hydrogen peroxide. Surfactants possess both hydrophobic and hydrophilic functional groups. Their main purpose in a cleaning solution is to enhance wettability and rinsability, improve contact between the foulants and cleaning agents and reducing water usage and cleaning time. Surfactants can be anionic (soaps and alkylsulphonates), cationic (quaternary ammonium compounds) or nonionic (phenol compounds and ethylene oxide).
Cleaning solutions should be recirculated through membrane modules at low pressure and at high velocity to prevent deeper penetration of foulants within the porous substructure of the membrane. Several companies sell chemical cleaners, as powders or liquid, specifically for membranes and for the most usual foulants. For new process applications the most effective cleaning protocol must be established experimentally.
At the industrial level MF is widely used in both dead-end and cross-flow configurations. Applications of dead-end MF are generally in the areas of purification, clarification, sterilization and analysis. A wide range of applications in the pharmaceutical industry including antibiotic processing, production of therapeutic proteins, ophthalmic preparations, fermentation, injectable drugs (parenterals), and so on, have already been consolidated (Scott, 1995). Antibiotic processing includes a variety of treatments such as the bulk clarification after the fermentation step, filtration of solvents and processing the bulk final solution in a solvent-based process involving the use of MF membranes. In the production of biologicals from mammalian cells, applications of microporous systems include cell-harvesting, perfusion of continuous cell culture vessels, clarification of biologicals and precolumn clarification. In the fermentation process used in the production of antibiotics, hormones, aminoacids and enzymes, the inlet and outlet air is filtered to remove airborne microorganic contaminants from the fermentor. MF applications also include the sterile filtration of the nutrient feed and liquid product and 0.22-μm-rated filters are usually used. These filters are also used to sterilize heat-sensitive ophthalmic components.
In processing parenterals, MF is primarily used for final sterilization from heat labile products, reduction of the bacterial burden before the final sterilization by autoclave and removal of particles from parenteral solutions and aerosols generated during their processing. The use of MF in the food industry can be considered the most consolidated application of membrane technology at the industrial level. A wide range of materials can be processed, including milk, vinegar, edible oils, drinking water, alcoholic beverages (beer, wine, whiskies) and soft drinks.
In the dairy industry MF is used to remove bacteria and spores from milk in order to extend its shelf-life as an alternative to the ultrapasteurization. In the Bactocatch system developed by Alfa Laval (Holm et al., 1986) raw milk is separated into skimmed milk and cream. The resulting skimmed milk is treated by MF obtaining a permeate with a bacterial concentration less than 0.5% of the original value. The retentate containing nearly all the bacteria and spores is submitted to a conventional high heat treatment (130 °C, 4 s) and then mixed with the permeate before the pasteurization step. In this process only a small volume of milk is treated at high temperatures, preserving its sensory qualities. Another promising application is the fractionation of milk proteins: the separation of micellar casein from whey proteins can be achieved by ceramic MF membranes producing a retentate enriched in native casein that can be used for cheese making. In whey processing MF can be used to remove bacteria and spores from the initial feed, obtaining a permeate that can be submitted to the UF step for the production of high quality whey protein concentrate and isolates.
Cross-flow MF is a valid approach for removing dirt, coagulated proteins, fats and other particles from raw gelatine derived from selective hydrolysis of collagen. Suspended solids can be also removed by MF from glucose solutions employed in syrups and sweeteners production. Several applications of both dead-end and cross-flow MF can be found in wine production. In particular, the use of cross-flow MF in wine production offers significant advantages over conventional processing steps in terms of reduced use of SO2, improved quality and brilliance of wine, reduced costs for the disposal of diatomaceous earths with no changes in the wine composition. Cross-flow MF membranes are used for the clarification of both white and red wines before storage as an alternative to the use of fining substances and filter materials. Other applications include cold sterilisation of unfermented must, the last cellar filtration after ageing in a storage tank, cold sterilization of wine before bottling and filtration of water used for bottle cleaning. Sterile venting of unfermented must tanks, sterile filtration of compressed air to remove unfermented must and filtration of steam and air using disposable filter capsules and cartridges are typical dead-end MF applications. Plate-and-frame MF modules are commonly used to remove yeast, microorganisms and haze in the clarification step of beer without affecting its taste.
In the electronics industry, MF membranes are used to remove microparticles from a variety of chemicals such us sulphuric acid, nitric acid, ammonia and hydrogen peroxide, in the production of ultrapure water. These membranes are typically made of polyethylene, PVDF and PTFE with a 0.1 μm pore diameter or less. Teflon MF membranes are also used to remove particles from a variety of reactive gases and solvents employed in the electronics industry.
Tubular MF membranes are used in treating the sludge from the treatment of sewage and chemical coagulation of surface water, as alternative to the common use of polyelectrolytes, centrifugation and plate and frame filter presses. Typical applications also include the removal of oil and fine metals from industrial wastewaters, the removal of precipitated metals (metal hydroxide flakes) from industrial effluents and the processing of fruit juices. Table 20.3 summarizes main applications of the MF process at the industrial level.
|Pharmaceutical||Sterile filtration: parenterals, ophthalmic solutions, fermentation products (antibiotics, vaccines and bioengineered proteins), tissue culture media; removal of microorganisms and particulates from air and other gases (vent filters, filtration of air or nitrogen used in fermentors)|
|Electronic||Sterile filtration: final point-of-use filters for ultrapure water|
|Nuclear||Removal of corrosion products in the primary coolant loop of boiling water reactors|
|Food/beverage||Sterile filtration: wine, beer, bottled water, gases, sugar solutions, edible oil, syrups, vinegar, whisky, brandy; clarification of cheese whey, defatting and reduction of microbial load in milk; clarification of wine, beer, fruit juices, vinegar, bottled water, beet and sugar cane solutions, purification of dextrose stream|
|Medical||Haemodialysis, biohybrid organs, analytical and diagnostic devices|
|Biotechnology||Downstream processing (concentration, clarification); marine biotechnology; biological conversions by membrane bioreactors (membrane recycle bioreactors, plug flow bioreactors)|
|Water treatment||Process water, treatment of grinding and polishing waters, finely dispersed suspension; retention of activated carbons, recovery of process water, filtration of process chemicals, solvents; removal of heavy metals as hydroxides, removal of lignin, removal of oil water effluents|
|Bulk chemicals/petrochemical||Sterile filtration: solvents and reagents, inorganic solutions, fatty acids, waxes, polymer fibers and films|
|Paints/coatings||Painting solutions, hydraulic fluid inks, foam, plating solutions, wastewaters|
|Sugar refining||Removal of colloidal and macromolecular impurities from raw juice, clarification of thin juice|
|Vegetable oils||Removal of waxes from sunflower oil|
|Corns and other grains||Dextrose clarification|
The first large application of UF was the recovery of electrocoat paint in an automobile plant (Cardew and Le, 1998). The UF system removes ionic impurities from the paint tank, producing a clean permeate, which is sent to the countercurrent rinsing operation, and a concentrated paint that is returned to the paint tank. Tubular modules were used in the first plants; recently, capillary and spiral wound modules have also been used.
Significant applications of UF exist in the food industry especially in the dairy industry and in the production of fruit juices, beer and wine. During cheese production, milk proteins are precipitated and the solid fraction (curd) formed is sent to the cheese fermentation plant. The supernatant liquor (whey) containing most of the dissolved solids and sugars present in the original milk and about 25% of the original protein content create remarkable disposal problems for the dairy industry. The protein fraction is removed from the whey by UF in order to obtain a concentrated protein, which can be successively dehydrated by evaporation, and a permeate enriched in lactose and salts. Since whey proteins are characterized by both nutritional and functional properties (gelling, emulsifying and foaming properties), they can be exploited in a wide range of applications (nutritional foods, beverages, processed meats, etc.).
Typically, a first UF step is used to achieve a 5- to 10-fold volume reduction and remove most of the lactose; the feed is then diluted with water and reconcentrated in a second UF step in order to remove the remaining lactose. The permeate can be discharged to a biological wastewater treatment plant or submitted to a concentration step by RO. UF can be also used to treat whole or skimmed milk in order to produce a pre-cheese concentrate (MMV process) that can be used directly to produce soft cheese (Camembert, mozzarella and feta) and yogurt (Maubois et al., 1980). Another application of UF is the standardization of proteins and total solids in milk for use in fermented dairy products (yogurt, cream and cottage cheeses) (Rosemberg, 1995).
UF is successfully applied in the clarification of fruit juices. In particular, UF represents a valid alternative to the use of traditional fining agents such as gelatine, bentonite and silica sol, which cause problems of environmental impact caused by their disposal (Eykamp, 1995). This process can be used to separate juices into a fibrous concentrated pulp and a clarified fraction free of spoilage microorganisms. Then, the clarified fraction can undergo non-thermal membrane concentration, such as membrane or osmotic distillation and, in the case of whole juice reconstitution, by combination with pasteurised pulp to obtain a product with improved sensorial properties (De Barros et al., 2003). Other advantages of UF over conventional methods include the production of clarified juices by a continuous simplified process, elimination of filter aids and filter presses, reduction in the filtration time, increased juice yield, better product quality and reduced enzyme usage (Fukumoto et al., 1998). UF polymeric and ceramic membranes with a MWCO of 10–50 kDa, packaged as tubular or capillary hollow-fibre modules, are generally used.
In winemaking, UF can be used before the fermentation in order to remove colloids, high molecular weight tannins, haze proteins, suspended solids and microorganisms from musts. After fermentation, UF can be used as alternative to fining agents (i.e. bentonite) to remove haze proteins from the wine (especially white wine) to improve the stability of the finished wine. Flavour compounds, generally removed by fining agents, can be preserved during the UF treatment. Additionally, UF membranes are able to remove enzymes such as polyphenol oxidase that causes the formation of undesirable brown compounds, increasing the colour/flavour stability of the wine. A wide range of vinegar types can be clarified by UF as a substitute for conventional methodologies. The process allows the removal of undesirable compounds such as proteins, pectins, yeast, fungi, bacteria and colloids and a reduction of the storage time.
Other well-established UF applications in the food industry are related to the concentration and purification of animal blood plasma, concentration of whole-egg and the egg-white and concentration of agar, agarose, carrageenan, gelatine, apple and citrus pectin (Lipinzki, 2010). Spent oil-water emulsions used in automated machining operations (automobile plants, steel rolling mills, etc.) can be treated by UF to recover the oil component for reuse and to obtain a dilute permeate that can be discharged or reused. The removal of microparticles, ions and microorganisms from water in the production of ultrapure water used in the manufacture of semiconductor devices is another consolidate application of the UF process (Cheryan, 1998). Finally, UF membranes can be used in surface water treatment (such as natural organic matter and pathogen removal), in membrane bioreactors (MBRs), in pretreatment processes for RO and nanofiltration (NF), and in many other industrial applications (Aoustin et al., 2001; Van der Bruggen et al., 2003, Durham et al., 2001; Jarusutthirak and Amy, 2001). Main applications of industrial UF are summarized in Table 20.4.
|Pharmaceutical||Sterile filtration: parenterals, ophthalmic solutions, fermentation products (antibiotics, vaccines and bioengineered proteins), tissue culture media|
|Food/beverage||Clarification of fruit and vegetable juices, must, wine, beer and vinegar; fractionation of cheese whey, preconcentration of milk for cheese manufacture|
|Biotechnology||Downstream processing: separation and harvesting of microbial cells from the suspending medium, recovery of enzymes from plant and microbial sources; biological conversions by membrane bioreactors (membrane recycle bioreactors, plug flow bioreactors)|
|Water treatment||Process water, water treatment in drinking water production; treatment of metalworking oily waste, recovery of brine (NaCl) in cheese manufacture, treatment of wash water from printing process, final polishing of laundry wastewaters, removal of heavy metals by micellar-enhanced UF|
|Textile||Recovery of synthetic warp sizing agents, treatment of wool scouring effluents, dyes recovery|
|Chemicals||Treatment of latex emulsions|
|Pulp and paper||Colour removal from kraft mill bleaching effluents, concentration of dilute spent sulphite liquor, recovery of lignin from kraft black liquor, recovery of paper coatings|
|Tanning and leather||Sulphide recovery from spent dehairing bath, recovery of vegetable tanning baths, chromium recovery from spent chromium liquors|
|Electrocoat paint||Recovery of paint from electrocoat painting solutions|
|Sugar refining||Removal of colloidal and macromolecular impurities from raw juice, clarification of thin juice|
|Soybean and vegetable proteins||Removal of oligosaccharides, phytic acid and trypsin inhibitors from soybean; removal of undesirable compounds from vegetable proteins|
|Vegetable oils||Removal of reverse micelles from oil–hexane mixtures; removal of fatty acids from oil|
|Corns and other grains||Separation of corn proteins|
|Animal products||Treatment of wastewaters from meat processing plants, reduction in low molecular weight components from gelatine, removal of glucose from egg white, recovery of proteins from effluents of fish processing|
MF and UF can be integrated with conventional processes (centrifugation, evaporation, liquid–liquid extraction and adsorption) or other membrane operations in different productive sectors according to the process intensification strategy. Significant advantages can be obtained in terms of improvement of product quality, reduction of energy consumption and environmental impact, recovery and reuse of water and chemicals.
Traditionally, precipitation, filtration and ion-exchange are used in the production of industrial water. These processes can be replaced by MF, RO and electrodialysis (ED). Integrated processes including water softening, MF, RO, UF, UV-sterilization and mixed bed ion-exchange have been proposed for purifying well or surface waters (Strathmann et al., 2006).
The combination of bioreactors and UF membranes (MBRs) permits an innovative and effective cleaning process for both municipal and industrial wastewaters to be obtained. The bioreactor is an oxygenated feed tank in which the dissolved organic substances are decomposed by microorganisms into nitrogen, water and carbon dioxide. The biomass is separated from the effluent by the UF membrane and recycled in the feed tank. Using this treatment high molecular weight substances are rejected by the membrane and can be further decomposed. The biomass concentration in the system can reach values 10 times higher than those obtained in conventional plants (Zenon Environmental b.v., 1995). A highly concentrated activated sludge process has also been combined with rotary disk type UF membranes in the treatment of fermentation wastewaters (Lu et al., 1999).
The recovery of water, toxic and valuable compounds from water and wastewaters in the chemical industry, to avoid pollution of the environment, is another interesting field of application (Reynolds, 1996). The application of ED with MF, UF and ion-exchange processes provides a good solution for recovering and recycling valuable compounds (nickel, cadmium, copper) from rinse solutions in the metal processing industry. An integrated membrane process based on the use of UF and NF has been proposed by Cassano et al. (2001) for the recovery of chromium from spent tanning liquors in the leather industry. Osmota GmbH (Germany) proposed a treatment scheme for the recovery and recycling of water and sulphuric acid from the rinse of a lead battery production line including MF, NF, ED and RO (Strathmann, 2004). Kim et al. (2006) reported a pilot-scale wastewater treatment and reuse system by integrating MF and ED. MF removes suspending solids and heavy metal ions (immobilized by separating materials) from sewage, while the following ED operation further purifies and desalinates the outlet. UF and RO can be also integrated to recover dyes from wastewaters produced by the textile industry (Short, 1993).
In the food and beverage industry, integrated membrane operations including MF and UF can be successfully applied for clarification, concentration, fractionation, desalting, recycling, recovery and purification. In this field, the use of membrane technology is dominated mainly by the application in the treatment of whey and milk followed by beverages, wine, beer and fruit juices.
A general scheme of integrated membrane processes in milk processing is depicted in Fig. 20.6. The UF concentrated milk is used to formulate speciality milk-based beverages (such as beverages with a high calcium content and relatively low content of fat and cholesterol) and for a pre-cheese mixture production which is then processed into hard or soft cheese (Cheddar, Camembert, Roquefort, etc.). Milk protein concentrates, containing 50–58% protein, can be produced by an integrated MF/UF process. These concentrates are used in the production of food additives in which the functionality of proteins has to be guaranteed.
20.6 Integrated membrane process in milk processing. adapted from Cheryan, 1998
Whey protein concentrates, lactose and waste streams with reduced biological oxygen demand (BOD) can be produced by integrating UF and RO processes. Integrated processes involving MF and UF are also efficient in whey fractionation. In particular, MF membranes retain fats, precipitated salts, bacteria and casein as particulate. UF can be used to fractionate β-lactoglobulin and α-lactoglobulin from other proteins (lactoferrin, albumin, immunoglobuline, etc.) (Zydney, 1998).
High-strength citrus juice concentrates can be produced by using UF or MF as a pretreatment to separate the pulp from the serum, followed by a combination of high- and low-retention RO membranes (Cheryan and Alvarez, 1995). A pilot scale system investigating an integrated process for the clarification (by UF), concentration (by thermal evaporation) and aroma compounds recovery (by pervaporation, PV) from apple juice has been exploited by Alvarez et al. (2000). Integrated membrane processes for the clarification (by UF), preconcentration (by RO) and concentration (by osmotic distillation, OD) of different fruit and vegetable juices (orange, carrot, lemon, kiwifruit and cactus-pear) have been also proposed by Cassano et al. (2003, 2004, 2007). In wine production, the UF step can be used before fermentation to remove microorganisms, colloids and high molecular weight compounds from the must before its fermentation. This process can be integrated to a second membrane step (MF) in order to remove yeasts from the wine. A final step before bottling could be a sterile filtration by using MF membranes.
In a typical recovery process for intracellular products made by fermentation, MF and UF processes are typically used for the medium purification (MF), cell harvesting (UF), concentration of products (UF) and pyrogen removal (UF).
As already mentioned, in most applications MF and UF processes are characterized by several advantages compared to conventional procedures. Typically, these processes are energy efficient, do not require the addition of chemical compounds or phase changes, are modular and easy to scale up and down, yield a higher quality product and offer the possibility of carrying out the separation continuously. In addition they operate at ambient temperature avoiding any change or degradation of products. This is a very important feature in separation, concentration and purification processes involving thermosensitive products such as food and pharmaceutical products. However, in many MF and UF applications extensive pretreatments are needed in order to limit concentration polarization and membrane fouling phenomena. The low mechanical resistance of the membrane, in some cases, is an additional drawback which can lead to the membrane breaking when there are uncorrected operating procedures.
The market for dead-end cartridge MF membranes used in microelectronics and pharmaceutical applications is well developed. A potential increase of the cross-flow MF market is expected in some areas of interest such as drinking water production, tertiary treatment of sewage and replacement of conventional depth filtration in food industries. Ceramic MF membranes with long lifetimes are available for this purpose but their high cost is the main drawback for large scale application. The development of cheaper MF membranes characterized by long lifetimes could offer new perspectives in market growth related to the areas mentioned.
Further membrane fouling and gel layer formation are inherent characteristics of MF and UF processes, which affect membrane fluxes and membrane lifetimes and consequently increasing operating costs. Significant progresses have been made in controlling and limiting these phenomena in the last few years, and thus the development of fouling-resistant membranes and membrane modules is expected. Investigations concerning the modification of membrane surface absorption properties and the reduction of the deposited layer-membrane surface link are promising approaches for this purpose.
Similar to the MF process, ceramic UF membranes could replace polymeric membranes in many applications if their cost were to be competitive. The development of new applications of MF and UF processes will be driven by economic and environmental targets. An additional driver for these processes is the high growth rate of the market for functional foods, a segment in which membrane operations have a high potential. Finally, MF and UF membranes have great potential as systems to improve the design of chemical transformations. In particular, the properties of catalysts immobilized in polymeric membranes and membrane bioreactors offer new perspectives in the production and purification of specific bioactive compounds.
Alvarez, S., Riera, F.A., Alvarez, R., Coca, J., Cuperus, F.P., Bouwer, S., Boswinkel, G., van Gemert, R.W., Veldsink, J.W., Giorno, L., Donato, L., Todisco, S., Drioli, E., Olsson, J., Trägårdh, G., Gaeta, S.N., Panyor, L. A new integrated membrane process for producing clarified apple juice and apple juice aroma concentrate. J Food Eng. 2000; 46:109–125.
Cassano, A., Drioli, E., Galaverna, G., Marchelli, R., Di Silvestro, G., Cagnasso, P. Clarification and concentration of citrus and carrot juices by integrated membrane processes. J Food Eng. 2003; 57:153–163.
Kim, J.O., Jeong, J.T., Kim, S.K., Kim, R.H., Lee, Y.J. Development of novel wastewater reclamation system using microfiltration with advanced new membrane material and electrodialysis. Mater Sci Forum. 2006; 510–511:586–589.
Lipinzki, F., Cross-flow membrane applications in the food industryPeinemann K.V., Nunes S.P., Giorno L., eds. Membrane Technology, Volume 3: Membranes for Food Applications. Membranes for Food Applications; Volume 3. Wiley-VCH, Weinheim, 2010:1–24.
Loeb, S., Sourirajan, S. Seawater demineralization by means of a semipermeable membrane. In: Gould R., ed. Advances in Chemistry, ACS Series No. 38. Washington: American Chemical Society; 1962:117–132.
Strathmann, H., Production of microporous media by phase inversion processes ACS Symposium Series No. 269Lloyd D.R., ed. Material Science of Synthetic Membranes. American Chemical Society: Washington, 1985:165.
Van der Bruggen, B., Vandecasteele, C., Van Gestel, T., Doyen, W., Leysen, R. A review of pressure-driven membrane processes in wastewater treatment and drinking water production. Environ Prog. 2003; 22:46–56.