Membranes defect-free

The following are some of the reasons that microreactors can be be used (i) reduced mass and heat transfer limitations, (ii) high area to volume ratio, (iii) safer operation, and (iv) ease of seating up by numbering out. The advantages of scaling down zeolite membranes are that it could be easier to create defect-free membranes and... 

Several factors contribute to the successful fabrication of a high-performance membrane module. First, membrane materials with the appropriate chemical, mechanical and permeation properties must be selected this choice is very process-specific. However, once the membrane material has been selected, the technology required to fabricate this material into a robust, thin, defect-free membrane and then to package the membrane into an efficient, economical, high-surface-area module is similar for all membrane processes. Therefore, this chapter focuses on methods of forming membranes and membrane modules. The criteria used to select membrane materials for specific processes are described in the chapters covering each application. 

Not much research was performed on coating of tubes in this project. Results on the coating of commercial tubes show however that it is rather difficult to coat sufficiently defect-free membranes on these supports. As was already stated in chapter 4, the surface roughness of the used tubes was possibly too high to coat high quality silica layers. Another possibility is that here the same problem occurs as was encountered with the sealed flat membranes. Some defects in the membrane layer might result from a bad adherence of the coated layer at the enamel/membrane interface. 

Scaling up of CMRs will require the ability to prepare large areas of defect-free membranes at a reasonable cost. The critical cost will depend on the advantages afforded by the membrane reactor and will therefore be closely related to the type of process con-... 

There are several models that describe the transport of mass through RO membranes. These are based on different assumptions and have varying degrees of complexity. The solution-diffusion model best describes the performance of "perfect," defect-free membranes and is considered the leading theory on membrane transport.2 Three other theories are presented here for completeness. 

Alumina is much more adapted to the particulate sol route while silica and titania are more flexible in that they can lead to both particulate and polymeric gels as the membrane precursors. This is particularly true with the silica system. Hydrolysis and condensation reactions in silica systems are known to be slower than most other alkoxides. Their reactions, therefore, are easier to control. The fact that it is much easier to prepare alumina sol-gel membranes than their titania equivalents may suggest that the plate-shaped particles like alumina can be more readily processed to form defect-free membranes than spherical particles such as titania [van Praag et al., 1989]. 

Heinzelmann [49] and Briston [52] have given details of these methods. The main problem with direct coating is that incompatibility of the porous support with polymer solution can result in the formation of pinholes in the membranes. However, this problem can be overcome if care is taken to ensure sufficient wetting of the support by the polymer solution. Parameters that can be varied in optimizing the process for a defect free membrane are... 

Choice of appropriate coating method along with drying/curing conditions can yield a stable defect-free membrane. 

Defect-free zeolite membranes have so far only been produced for membranes of the MFI (silicalite type) with thicknesses of about 50 im on stainless steel supports and 3-10 pm on alumina and carbon supports. They are produced by in situ methods of zeolite crystals grown directly on the support system. There are some reports of formation of defective membranes with, e.g., zeolite A. Much more research is needed to widen the range of available zeolite membrane types especially small and wide pore systems. The permeance values of the defect-free membranes is lower than that of the amorphous membranes (see Chapter 6) and to improve this the layer thickness must be decreased together with improving the crystal quality (no impurities, no surface layers, high crystallinity, crystal orientation) and microstructure (grain boundary engineering). 

The mean pore size and pore size distribution can be evaluated by performing this measurement by a stepwise increase of the pressure. In this case the gas flow across the wet defect-free membrane is recorded (Fig. 4.18) as a function of the applied pressure difference across the sample ("wet curve"). The point of first flow is identified as the "bubble point". This continues until the smallest detectable pore is reached. Then the flow rate response corresponds to the situation in a completely dry sample. The measurement of gas flow through the same membrane in a dry state gives a linear function of the applied pressure difference ("dry curve"). The pressure at which the "half-dry" curve intersects with the "wet" curve can be used to calculate an average pore diameter. Pore number distributions can also be derived from flow distribution curves. 

A minimum roughness of the support surface is also required to produce defect-free membrane layers. In the present context, surface roughness is defined as the average perpendicular (to the surface) distance between peaks and dips in the support surface. As discussed in Chapter 6, several other definitions of roughness can be given and used. The roughness of the support may limit the minimum achievable layer thickness. From a fracture mechanics point of view, surface roughness determines the maximum size and sharpness of flaws which can act as crack initiators (via the stress intensity factor). 

It is well documented that the addition of organic polymeric additives to precursor sols promotes the formation of defect-free membranes [4,12,33]. To investigate this effect, stress measurements were performed on boehmite membranes obtained from standard sols (1 mole AlOOH/1) mixed with different amounts of PVA solutions (containing 35 ml PVA/1, molecular weight 72000). 

As shown in Fig. 8.19 [13,31] the stress in the constant rate region decreases with increasing amount of PVA to zero at a weight ratio PVA/Y-AI2O3 > 0.25 (0.7 ml PVA solution per ml AlOOH sol). More recently the effect of additives on the formation of zirconia and alumina-zirconia membranes was further investigated by Ziiter [32]. Ziiter showed that the molecular weight had no important influence but that the pore size showed a minimum at certain concentration of the added PVA. The addition of PVA strongly promoted the formation of defect-free membranes. 

It is appropriate here to treat structural characteristics of zeolites. A brief summary has been given in the Chapter 9, Section 9.4.1 in which a typical structure is illustrated by Fig. 9.19 for MFI type (ZSM5, siHcalite) zeolites, being almost exclusively used for defect-free membrane s)mthesis. For a more detailed discussion the reader is referred to the works of Breck [92] and of van Bekkum et al. [93]. 

Jia/Noble and coworkers [87,88] reported the successful synthesis of silicalite membranes on y-alumina composite supports using an interesting modification of the in situ crystallisation method. The support consisted of a short a-alumina tube coated on the inside with a 5 pm thick y-alumina film with an average pore diameter of 5 nm, commercially available from US Filter. The precursor solution was put into the support tube after plugging both ends with teflon and the filled tube was then placed in a teflon-lined autoclave. Hydrothermal treatment was carried out at 180°C for 12 h. After removal from the autoclave and washing the formed zeolite layer with water, the procedure was repeated with the tube inverted from its previous orientation to obtain a uniform coating. As reported by Vroon et al. [82,84,98], Jia/Noble [88] also concluded that at least two synthesis steps are necessary to obtain defect-free membranes. 

A plot of Ft vs P for defect-free membranes which are definitely in the transition region )delds a curve with a certain slope, which intersects the permeation axis. This is shown in Fig. 9.5. 

The advantage of the preferential sorption-capillary flow approach to reverse osmosis lies in its emphasis on the mechanism of separation at a molecular level. This knowledge is useful when it becomes necessary to predict membrane performance for unknown systems. Also, the approach is not restricted to the so-called "perfect", defect-free membranes, but encompasses the whole range of membrane pore size. Until recently, the application of a quantitative model to the case of solute preferential sorption has been missing. Attempts to change this situation have been made by Matsuura and Sourirajan (21) by using a modified finely porous model. In addition to the usual features of this model (9-12), a Lennard-Jones type of potential function is Incorporated to describe the membrane-solute interaction. This model is discussed elsewhere in this book. 

A variety of polymers and copolymers are used for gas separation membranes. To be suitable for gas separation, the polymer must have good permeability and selectivity and the material must be capable of forming a strong, thin, defect-free membrane with good chemical and thermal stability. Commercial gas separation membranes are based on modified cellulose, treated polysulfone or a substituted polycarbonate polymer. Membranes... 

Parameters that can be varied in optimizing the process for a defect-free membrane are... 

A battery of single-gas permeation experiments using molecules with different kinetic diameters is used to gauge the effective pore size in defect-free membranes... 

Concerning the preparation of thin membranes directly on porous supports, a lower thickness limit seemingly exists for which a dense metal layer can be obtained. This thickness limit increases with increasing surfaee roughness and pore size in the support s top layer." " Clearly, this relation puts strong demands on the support quality in terms of narrow pore size distribution, and the amount of surface defects. Therefore both pore size and roughness of the support surface are often reduced by the application of meso-porous intermediate layers prior to deposition of the permselective metal layer. This procedure facilitates the preparation of thin defect-free membranes beeause it is relatively easier to cover small pores by filling them with metal. It is therefore conceivable that for a certain low Pd-alloy thickness and support pore size, the H2 flux becomes limited by the support resistance. ... 

A two-step membrane manufacturing process has been reported where a defect free Pd-alloy membrane is first prepared by sputtering deposition onto the perfect surface of a silicon wafer, for example. In a second step the membrane is removed from the wafer and transferred to a porous stainless steel support (see Figure 11.1). This allows the preparation of very thin ( 1-2 pm) defect-free membranes supported on macroporous substrates (pore size equals 2 pm). By this technique, the ratio of the membrane thickness over the pore size of the support may become less than 1, which is two orders of magnitude smaller than obtained by more conventional membrane preparation techniques. Tubular-supported palladium membranes prepared by the two-step method show a H2/N2 permselectivity equal to 2600 at 26 bars and hydrogen flux of 2477 mL(STP) min cm . Since the method enables the combination of macro-porous stainless steel supports and thin membrane layers, the support resistance is negligible. ... 

Defect-free membranes comprising zeolites and amorphous glassy perfluoropolymers can be prepared by modifying the surface of the filler. The pure gas permeation experiments of a series of Teflon AF 1600 membranes with various amounts of 80 and 350nm silicalite-1 crystals cannot be interpreted on the basis of the Maxwell model, but are compatible with a model in which a barrier to transport exists on the zeolite surface and a lower density polymer layer surrounds the crystals. With a small zeolite size (80nm) the low density layers around the crystals may coalesce and form percolation paths of lesser resistance and less selectivity. Silicalite-1 crystals improve the CO2/CH4 selectivity of Hyflon AD60X, and drive the N2/CH4 selectivity beyond the Robeson s upper bound. It also turns out that the presence of silicaUte-l crystals, like fumed silica, promote the inversion of the methane/butane selectivity of Teflon AF2400 in mixed gas experiments. 

---