Membrane processes cover a group of separation processes in which the characteristics of a membrane (porosity, selectivity, electric charge) are used to separate the components of a solution or a suspension. In these processes the feed stream is separated into two: the fraction that permeates through the membrane, called the permeate, and the fraction containing the components that have not been transported through the membrane, usually called the retentate The size of the components to be separated and the nature and magnitude of the driving force provide criteria for a classification of the membrane separation processes, as shown in the table. It should be noted that the boundaries between some of the processes, such as reverse osmosis and ultrafiltration, are arbitrary.
Of the processes listed in Table 1, reverse osmosis and ultafiltration are the most widely used industrially. Abundant examples of their application can be found in water Desalinization, the dairy industries and the separation of organic solutes from aqueous solutions. Membrane processes do not require heating, which makes the process suitable for the treatment of thermolabile products. In addition the relatively low capital and operating costs involved make membrane processes an appealing alternative to more conventional separation processes, particularly when dealing with dilute solutions. In the next sections the principles and characteristics of the processes listed in Table 1 will be described briefly.
Osmosis is the process of transfer of the solvent of a solution across a membrane separating two liquid solutions of different concentrations. The transfer takes place from the phase in which the chemical potential solvent is higher. A classical example is the transfer of water to a sugar solution. Figure 1(a) shows two compartments separated by a water permeable membrane. One contains the solution, while the other holds pure water. Initially the level of the solution is placed at the same level as that of the water. Since the chemical potential of pure water is higher than that of water in the solution, water transfers across the membrane to the solution, increasing the level on the solution side and thus creating a hydrostatic head (Figure 1b). When the process is at equilibrium the difference in the hydrostatic pressure between the two phases is that required to make the chemical potential of water in both phases the same. In the special case of this example, when the pure solvent is on one side of the membrane this pressure difference is the osmotic pressure, P. If solutions are at both sides of the membrane, the pressure difference gives the osmotic pressure difference ΔП. For very dilute solutions the osmotic pressure is given by
where β is a constant that takes into account the dissociation of the solute, ci is the molar concentration of solute i, R the gas constant and T the absolute temperature. The constant (β is in general equal to the number of ions produced by the dissolved solute; thus for electrolytes that dissociate fully such as Nalco and CAC2 it is equal to 2 and 3, respectively, while β = 1 for molecules that do not dissociate. When the solute has a large molecular mass, the osmotic pressure of the solvent ceases to depend linearly on the concentration of the solute and a polynomial approximation must be used. For concentrated solutions as well as nonideal solutions the osmotic pressure must be determined experimentally or estimated using activities instead of concentrations.
The term reverse osmosis is applied to the separation of water from solutions by transfer though a semipermeable membrane that ideally should only be permeable to water. If as shown in Figure 1(c) a mechanical pressure is applied to the compartment where the level is higher, the chemical potential of the solvent on the solution side will increase, forcing the transfer of the solvent through the membrane into the pure solvent phase. This pressure difference between the phases, ΔP, creates the driving force for the transfer and is the principle on which the reverse osmosis and ultrafltration processes are based. In order to achieve the transfer, ΔP must be greater than the difference between the osmotic pressures ΔП. The rate of solvent permeation across the membrane is measured as its linear velocity normal to the membrane, J1. Ideally the membrane should be only permeable to the solvent. However this is very difficult to achieve and membranes have some permeability to the solute. The rate of permeation of the solvent is given by:
where K1 is the permeability coefficient of the membrane and ΔP is the transmembrane pressure. The permeability coefficient depends on the membrane resistance, the resistance to transfer caused by deposits on the membrane and the diffusion resistance in the boundary layer at the retentate side of the membrane. Depending on the thickness of the solid deposits and the hydrodynamic conditions of the retentate the two latter resistances may be negligible. In fact membrane equipment are operated under conditions that tend to minimize these two resistances in order to increase J.
The difference in solute concentration between the retentate and the permeate (in practice membranes have some permeability to the solute) creates a driving force for solute transfer across the membrane. The flux of solute, J2, is given by:
where K2 is a constant that includes the characteristics of the membrane, and cR and cP are the concentration of the solute in the retentate and permeate respectively.
Equations (2) and (3) indicate that the driving forces for solvent and solute transport are independent. Therefore by increasing transmembrane pressure it is possible to increase the rate of solvent transfer without increasing the concentration of solute in the retentate. This in fact leads to an increase in the solute partition coefficient between retentate and permeate. However the increase in solute concentration in the retentate, and in particular at the membrane wall, may lead to an increase in osmotic pressure important enough to reduce J1. This effect is called membrane polarization and will be discussed in connection with ultrafiltration.
Ultrafiltration is also a membrane separation process driven by pressure. Its main difference from reverse osmosis is the type of membrane used. While reverse osmosis membranes are nonporous and therefore, permeable only to very small molecules, ultrafiltration membranes are microporous and permeable to small solute species in addition to the solvent. Ultrafiltration is therefore used to keep large solute species in the retentate. The effect of osmotic pressure is therefore less important in this process so that, the transmembrane pressures required are not as high as for reverse osmosis. The solvent flux obtained with uitrafiltration is also given by Eq. (2), but the membrane permeability K1 now depends on the characteristics of the ultrafiltration membrane and the diffusional resistances specific to this process. As a result of solvent transfer the retentate becomes more concentrated and a concentration gradient of solute builds up in the direction of transfer with the higher concentration at the membrane wall. This phenomenon, called membrane polarization, creates a concentration driving force for solute diffusion in the direction opposite to solvent transfer. At equilibrium the diffusional flux of solute back to the retentate phase equals the convective flux towards the membrane:
where J is the velocity of the permeate phase through the membrane, cp is the concentration of solute in the permeate, D is the diffusivity of the solute in the retentate and y the distance in the direction normal to the membrane. As the separation proceeds the concentration gradient becomes steeper due to solute accumulation next to the membrane. As the solutes retained in ultrafiltration are larger than in reverse osmosis, their diffusivities are lower and thus accumulation near the membrane wall is more likely in ultrafiltration than in reverse osmosis.
Equation (2) allows comparison of the response of reverse osmosis and ultrafiltration to transmembrane pressure. As ΔP is increased, solute accumulation at the membrane wall on the retentate side is higher in ultrafiltration than in reverse osmosis due to the lower diffusivity of the solute. This accumulation of solute leads to membrane polarization as well as solute precipitation on the membrane (gelification), which in turn decrease K1 and increase ΔП. The process reaches a region in which the permeate flow rate becomes independent of pressure.
The separation of particles of micron and submicron levels can effectively be performed using membrane filters. The suspended particles for which the process is industrially used include colloids, microorganisms and emulsion droplets. There are two different configurations for the microfiltration operation: (1) dead-end microfiltration, and (2) crossflow microfiltration. In the first, the membrane plane is normal to the feed flux, while in the second, it is tangential. The advantages and disadvantages of the type of configuration depend on the characteristics of the feed. Highly concentrated suspensions are not suitable for dead-end treatment since the separated particles rapidly accumulate on the membrane, increasing the resistance to filtration by forming a cake and/or clogging the membrane. In crossflow filtration, the flow is parallel to the membrane and the drag forces close to the membrane wall reduce the amounts of particulate material deposited on the membrane. Therefore the first is used for dilute suspensions while the latter can deal with concentrated ones, such as slurries.
Microfiltration is a pressure driven process and as such the flux of filtrate is given by Equation (2). In this case ΔП can be neglected since it is not relevant to this process.
Dialysis is a process in which the solute, usually an electrolyte, transfers across the membrane driven by the difference in concentration between the two sides of the membrane. Two conditions have to be fulfilled for the process to be effective: (1) the concentration on the permeate side has to be kept low so that the driving force remains as high as possible, and (2) the osmotic pressure must be low, and remain low during the process, so that a counterflux of solvent does not result in feed dilution. Since the combination of these two conditions is not likely to occur, the use of selective membranes has become general practise so that the partition coefficient of the solute between the two phases can be substantially increased by using the Donnen effect. This is achieved by adding to the feed a salt with one of its ions in common with the solute and the other one rejected by the membrane. If, for example, NaCl is the solute to be separated, the addition of NaX (where X is the ion rejected by the membrane) will lead to the following expression for the NaCl partition coefficient at equilibrium:
where the square brackets indicate concentrations. This equation show that the larger the concentration of NaX, the greater the separation enhancement. (See also Dialysis.)
Electrodialysis is a process in which ion-selective membranes are used together with an electric field normal to the membrane phases. The basic principles of electrodialysis are better explained with reference to Figure 2. An electrodialysis stack consists of parallel compartments separated alternately by cation-exchange and anion-exchange membranes. The feed is introduced into each compartment, so that aided by the electric field and selected by the membranes, the cations and the anions are transferred in opposite directions to the neighboring compartments. In this way a demineralized solution leaves compartments 2 and 4 and the concentrated one compartments 1, 3 and 5.
Depending on the type of membrane used, ions can be selected according to their valence. Therefore this process can be used for fractionating ions of different valence. The main application of electrodialysis is in demineralization in general and in the desalinization of brackish water. (See also Electrodialysis.)
Pervaporation is the combination of the selective separation and transfer of a component across the membrane and its evaporation on the permeate side. In order to achieve evaporation the pressure on the permeate side must be such that the partial pressure of this component is lower than its saturation vapor pressure. Therefore although the driving force for transfer is the difference in activity of the transferred species, this is the result of applying a vacuum on the permeate side.
Pervaporation is more expensive than other membrane processes due to the required supply of heat to produce the evaporation. It is, therefore, used in the separation of components from mixtures that are difficult to treat, such as azeotopic solutions and isomer mixtures, for which conventional processes would be more costly.
A liquid membrane is a liquid phase that separates two fluid phases of different composition. In most applications the liquid membrane is an organic phase placed between two aqueous phases. The principles involved in liquid membrane separations are those of solvent extraction, rather than membrane separation. The solute is first selectively extracted by the liquid membrane, it then transfers across the liquid film driven by its concentration gradient and when it reaches the other side of the liquid membrane it is stripped by the third phase.
There are two types of mechanisms of transfer to and from the membrane: (1) Physical transfer, and (2) carrier mediated transfer. The first is based on the solubility of the solute in the membrane, while the second requires the presence of a selective reactant in the organic phase called the carrier. In this case the process is based on the reversibility of the chemical reaction. The extraction of Zn2+ from an aqueous feed with a carrier RH dissolved in the organic membrane, will be used as an example of carrier mediated extraction. The overall reaction between the cation and the carrier can be represented by:
where bars indicate species in the organic phase. Reaction 6 is reversible. At pH 3 the forward reaction takes place while at low pH values the reverse reaction predominates. A suitable liquid membrane system for the separation of a membrane containing RH should have a high pH value in the feed phase and a low one in the stripping phase. Figure 3 shows a schematic drawing of the concentration gradients in the feed, the membrane and the stripping phase with the corresponding concentration profiles.
The main advantage of liquid membranes over conventional solvent extraction is that by contacting the three phases simultaneously, the solute partition coefficient between the two aqueous phases may be several orders of magnitude greater than the one that can be obtained with conventional solvent extraction.
There are two ways in which a liquid membrane can be formed leading to two configurations: (1) emulsion liquid membranes and (2) supported liquid membranes. These are illustrated in Figures 4a and 4b. The emulsion liquid membrane is obtained by adding a surfactant into the organic phase in order to stabilize a water in oil emulsion. The emulsion is then dispersed in the feed so that the organic phase, which is the continuous phase of the dispersed globules, becomes the membrane. Supported liquid membranes are obtained by impregnation of porous fibers of hollow fiber modules.
Bowen, W. R. (1991) Membrane separation processes, in Chemical Engineering, Vol. 2, 4th edn., Coulson, J. M. and Richardson, J. R, Pergamon Press, Oxford.
Rautenbach, R. and Albrecht, R. (1989) Membrane Processes, John Wiley and Sons, Chichester.
Way, J. D., Noble, R. D., Flynn, T. M., and Sloan, E. D. (1982) Liquid membrane transport: a survey, J. Memb. Sci., 12, 239-259. DOI: 10.1016/S0376-7388(00)80185-4