1 Microfiltration for casein and serum protein separation

Kang Hu1, James M. Dickson1 and Sandra E. Kentish2

1Department of Chemical Engineering, McMaster University, Hamilton, Ontario, Canada, L8S 4L7

2ARC Dairy Innovation Hub, Department of Chemical and Biomolecular Engineering, The University of Melbourne, Parkville, Victoria 3010, Australia

1.1 INTRODUCTION OF MICROFILTRATION

Microfiltration(MF), probably the oldest membrane separation technology, was developed between the First and the Second World Wars in Germany for the purpose of bacteria removal (Zsigmondy and Bachmann, 1922). Generally, MF membranes have a pore size ranging from 0.1 to 10 µm. This size range encompasses a wide variety of natural and industrial particles, such as colloids, bacteria, and red blood cells.

MF is a pressure-driven separation process, which is similar to other widely used membrane processes such as ultrafiltration, nanofiltration, and reverse osmosis. Compared to these processes, MF is typically operated at a relatively lower pressure and is mainly applied for larger particle separation and fractionation.

In this section, the principle of MF is introduced. This includes the introduction of MF membranes and processes, the mechanism of cross-flow MF, and membrane fouling.

1.1.1 Microfiltration membranes and processes

MF membranes can be synthesized from a wide variety of materials, normally categorized as either organic, such as polymers, or inorganic, such as ceramic materials. Polymeric materials can be hydrophobic, including polytetrafluoroethylene (PTFE), poly(vinylidene fluoride) (PVDF) and polypropylene (PP), or hydrophilic, including polycarbonate (PC) and poly(ethersulfone) (PES) (Mulder, 1996a). Ceramic membranes are typically made from alumina (Al2O3), zirconia (ZrO2), and titania (TiO2). During membrane fabrication, some hydrophobic materials can be surface-modified to become hydrophilic, as required by specific applications. Gekas and Hallstrom (1990) reviewed these materials and summarized both the advantages and disadvantages of them. They suggested that comparing the two materials, polymer membranes are generally less expensive and have a higher area/volume ratio than ceramic membranes. On the other hand, polymer membranes bind protein more readily and have a wider pore size distribution. Ceramic membranes have exceptional thermal and chemical resistance and a much longer membrane life.

Various techniques have been employed to fabricate microporous membranes for MF. For example, ceramic membranes could be prepared by sintering layers on supporting materials. Polymeric membranes, benefiting from current polymer processing technology, could be prepared by various methods including: melt stretching, track etching, phase inversion, and casting (Glimenius, 1985; Mulder, 1996b).

Depending upon the materials and techniques used to prepare the membranes, MF membrane pore structure varies significantly. Figure 1.1 illustrates some typical examples of membrane porous surface structures obtained with different fabrication methods and materials. From the images, membrane pores created by stretching (a) are not circular, but the manufacturing process is relatively simple. Pores created by track etching (b) are cylindrically shaped with uniform dimensions but with lower porosity, while pores created by phase inversion (c) have a much higher porosity (or pore density). For ceramic membranes, sintering results typically in a nodular structure (d).

Figure 1.1 Microfiltration membrane surface images. (a) Polymeric membranes fabricated by melt-stretch (from Barbe, Hogan, and Johnson, 2000. Reproduced with permission of Elsevier). (b) Polymeric membranes fabricated by track-etching (Millipore Product Catalogue, 2013). (c): Polymeric membranes fabricated by phase inversion (Ying, Kang, and Neoh, 2002. Reproduced with permission of Elsevier). (d) Ceramic membranes fabricated by sintering (Zhang, Zhong, and Xing, 2013. Reproduced with permission of Elsevier).

The MF process is pressure-driven and generally is carried out in two modes: dead-end filtration or cross-flow filtration. As illustrated in Figure 1.2a, in the dead-end MF, an applied pressure pushes the feed stream toward the membrane perpendicularly. The solvent and some smaller particles permeate through the membrane due to the pressure difference across the membrane, but larger particles are retained and then accumulate and deposit on top of the membrane. As a consequence, a fouling layer is formed, resulting in a decrease in filtration permeate flux. As the fouling increases with time, the permeate flux decreases drastically. When the fouling is sufficiently large, no more permeate can be collected and the membrane needs to be replaced. Due to the nature of the process, dead-end MF is economically feasible only when applied to rather dilute and/or valuable solutions, since the fouled membranes are difficult to clean and are normally just discarded. Dead-end MF is typically conducted in batch mode and mostly applied on a laboratory scale, using cylindrical vessels such as syringes as the filtration cell.

Figure 1.2 (a) Dead-end flow and (b) cross-flow MF.

In cross-flow MF (Figure 1.2b), the feed stream flows tangentially to the membrane surface on the feed side. Similar to the dead-end mode, the solvent and some small particles transport through the membrane and are collected in the permeate side due to the pressure difference across the membrane. The remainder of the feed continuously flows along the membrane surface and is removed separately as a concentrated solution called the retentate.

On the retentate side, similar to the dead-end mode, the retained particles accumulate at the membrane surface forming a fouling layer, resulting in a decrease of permeate flux over time. However, a shearing effect, induced by the tangential flow of feed, limits the build-up and the growth of the fouling layer. Thus, during cross-flow MF, after an initial rapid formation, the growth of the layer slows down or even stops. At this point, the accumulation of particles becomes equivalent to the amount of particles lifted back to the bulk stream by cross-flow. The hydraulic resistance generated by the fouling layer becomes relatively constant and the permeate flux becomes stable. Under such conditions, cross-flow MF can operate continuously at a steady-state filtration flow for a long operating time, which substantially improves the separation performance and broadens the industrial application.

1.1.2 Cross-flow MF mechanisms

In MF, the separation mechanism is primarily size exclusion or the so-called "sieve effect." Particles smaller than the pore size may pass through the membrane but particles larger than the pore size are rejected. A great number of parameters influence this separation performance, including the cross-flow velocity, applied pressure, physical and chemical properties of the membranes and particles, interactions between particles and particles, and interactions between particles and membrane materials. Considering all these parameters, many mathematical models have been developed to describe the transport mechanisms of cross-flow MF. All these models have their limitations owing to the fact that the phenomena occurring on the membrane surface and interface are far from clear, especially when various types of particles are present in the feed solution, adding to the complexity. However, utilizing these models can improve understanding of the process and provide guidance on the process design of such operations.

Earlier attempts to modeling cross-flow MF started with the concentration polarization model, since concentration polarization affects membrane performance for almost all types of membrane separations. As illustrated in Figure 1.3, during MF the particles are rejected by the membrane and then accumulate near the membrane surface. Thus, the concentration of particle i at the membrane surface is always higher than that in the bulk solution . This phenomenon, just outside the membrane surface, is called "concentration polarization". The formed polarized layer (boundary layer), with thickness z, on the membrane feed side has some undesirable effects on the membrane separation. The higher concentration of particles at the membrane surface can cause increased fouling (binding of particles to the membrane surface). In addition, the concentration polarization invariably leads to reduced flux since the layer has a relatively large resistance to permeation (Zeman and Zydney, 1996) and can cause either increased rejection or decreased rejection as the fouled layer can be more restricting to solute permeation (hence increasing rejection) and the higher concentration in the boundary layer translates into a higher concentration in the permeate layer (hence decreasing rejection).

Figure 1.3 Concentration polarization effects of a microfiltration membrane with particle concentration profiles.

On the retentate side, the concentration build-up on the membrane surface generates a diffusive flow back of the particles from the membrane to the bulk of the feed solution. After a period of time, a steady state is achieved and the flow of particles i to the membrane surface is balanced by the particle flow through the membrane plus the diffusive flow from the membrane surface back to the bulk (D dCi/dz). The mass balance is given by

where JV is the...