Small scale selective filtration

The use of selective filtration to separate material of different sizes is a fundamental technique applicable at all scales in industry, as Lina Christopoulou, from Imperial College, London, explains

Nanofiltration refers to filtration at the nanoscale, dealing with molecules no greater than 1000 Da.In aqueous systems it developed in the 1980s and 1990s from membrane manufacturers making 'loose' reverse osmosis membranes, which had low rejections of mono-valent ions and moderate rejections of divalent (and higher) ions. Consequently the materials these membranes are made from tend to be similar to reverse osmosis membranes, e.g. polymers such polyamide, celluslosics and ceramics. However, finding materials suitable for membrane manufacture that give tight molecular weight cut-offs (MWCO) on the nanoscale and are also resistant to organic solvents has been a bigger challenge.

Such membranes were recently developed1 by WR Grace, the speciality chemicals and materials manufacturer, for use in their Max-Dewax process, which recovers solvent from oil refinery lube oil dewaxing units. These membranes, known as Starmem, offer four notional MWCOs, at 200, 220, 280 and 400 Da and are resistant to a range of apolar aprotic solvents such as toluene, methanol, ethanol, methyl ethyl ketone and xylene. A scanning electron micrograph of these membranes can be seen in figure 1.

The solvent resistance of these membranes opens up their use in fine chemical and pharmaceutical applications, in addition to the refinery process, and this article discusses some of the novel potential uses relevant to manufacturing chemists in industry.

application technology

Complex synthetic routes in the pharmaceutical industry are likely to involve one or more steps where the molecule of interest must be taken from one reaction solvent to another.

Typically this is undertaken using 'put and take' distillation, where part of the first solvent is boiled off, a volume of the second solvent is added, and the mixture is boiled again to remove more of the first solvent.

This sequence is repeated until the exchange is complete. However this process is not without its limitations. Temperature constraints often apply due to the thermal instability of the product molecules, which in some cases might make this process inappropriate. In systems where azeotropes are formed, it is likely that the first solvent will not be completely removed. Most importantly, carrying out solvent exchange using distillation is not effective when the second solvent has a lower boiling point than the first solvent, for example when swapping toluene to methanol.

These limitations can be overcome by the application of membrane technology. The principle is to use a membrane with an appropriate MWCO to retain the organic solute (OS) while swapping from one solvent to another.

This is illustrated in Figure 2. Filtration removes 70-90% of the high-boiling-point solvent (HBS); then the desired low-boiling-point solvent (LBS) is added. A second round of filtration removes the majority of the LBS/HBS mixture, resulting, after a number of filtrations, in a final solution where the organic solute is dissolved in 99.7% LBS. The advantages of this method over conventional distillation go beyond the relative simplicity and economy of nanofiltration. Organic solvent nanofiltration (OSN) also avoids the problems that occur when encountering azeotropes in solvent exchange; avoids drying out of the solute, which can damage thermally-sensitive components; and can be carried out at ambient temperatures. Moreover, the amount of total solvent used in OSN can be as little as one fifth of that required in distillation. Solvents which have been tested and found to be compatible with the Starmem membranes to date are listed in Table 1.

Even more 'difficult' solvents, such as DMF, can be tolerated by the membranes to effect certain applications. For example, solvent exchange allowed a toluene stream to be purified from 5 vol% DMF to 0.04 vol% DMF (99.2% DMF removal) with three filtrations taking in total less than one hour from start to finish and giving >95% yield of the high molecular weight solute dissolved in the toluene stream.

catalysis applications

Catalysis is a key enabling technology used in pharmaceutical and fine chemical synthesis. Nevertheless, although it is widely used, a number of technical issues still need to be resolved. Firstly, when using homogeneous and phase transfer catalysts, separating the catalyst from the product and solvent in the post-reaction mixture can be inefficient and hard to achieve.

Secondly, the cost of many of these catalysts can be high and so there is scope for a technology which will allow for improved recovery and potential reuse of these catalysts. OSN offers a solution to both these problems in several catalyst applications.

Phase transfer catalysts are used to shuttle reactants from an aqueous phase to an organic phase. At the end of the reaction, there is typically a saline aqueous phase, with the product and the catalyst partitioned into the organic phase (see figure 3).

Separating the product and catalyst by conventional means such as water or acid washes, or distillation, can generate toxic wastes and lead to decomposition of the catalyst. In an OSN system, by contrast, the catalyst can be separated from the product and recycled, while avoiding emulsions and the release of toxic catalyst components in the wastewater.

This was demonstrated experimentally2 for a phase transfer catalyst using the model reaction of converting bromoheptane to iodoheptane, in toluene, with tetraoctylammonium bromide (TOABr) as the catalyst.

After the biphasic reaction was completed, the organic phase, containing the iodoheptane product and the catalyst was transferred to a stirred filtration cell (see figure 4) loaded with a Grace Starmem membrane with a 220 Da cut-off. 30 bar nitrogen pressure was applied to filter 90% of the mixture through the membrane, while the remaining 10% retentate was re-introduced to the reactor and shown to retain the catalytic activity without further addition of fresh catalyst. This demonstrated that the membranes were able to separate the product from the catalyst without affecting catalytic activity. Moreover, a comparison of the mass intensity showed that removal of the phase transfer catalyst by OSN was twice as efficient as removing the catalyst using a citric acid wash.

Organic reactions such as hydrogenations and C-C couplings are catalysed by organometallic complexes of transition metals such as palladium, ruthenium, and rhodium. These metals are expensive and are also often attached to valuable organic ligands. Some of the value in the catalyst can be recovered after the reaction, by passing the post-reaction mixture through beds of activated carbon and/or silica, and then recovering the metal content.

Nevertheless this process is not trivial, and leaves no scope for reuse of the catalyst as it does not recover the metal complex intact.

As with the phase transfer catalysts above, nanofiltration can be used to retain organometallic catalysts so that they can be re-used. In a C-C Heck coupling reaction used to demonstrate the principle³, 95% of the catalyst was retained in each filtration. This catalyst was re-used in four reactions, without adding more fresh catalyst, before showing a marked reduction in reaction rate. Furthermore, the application of OSN has a further advantage because it reduces the level of metal impurities in the post-reaction product. This is particularly valuable in manufacturing pharmaceuticals, where levels of residual heavy metals in the final product are tightly regulated.

A further category of catalyst operations where OSN has applicability is in chiral hydrogenation reactions. Here the membranes can be used to retain the catalyst, allowing its reuse in sequential batch reactions, while the product-rich permeate filters through the membrane. As illustrated in figure 5, the hydrogenation reaction proceeds inside the filtration cell in the presence of the membrane. The essentially catalyst-free permeate filters through the membrane, while the catalyst-rich retentate remains in the reactor. Sufficient catalytic activity remains to allow the reaction to proceed again upon fresh addition of the substrates.

assessing technology

In summary, OSN has shown successful catalyst retention, as high as 99+% for the following catalysts:

Heck (Pd triphenylphosphine derivatives and Pd-imidazoles)

Chiral hydrogenation catalysts (Ru-BINAP, FerroTANE)

Transfer hydrogenation catalysts (metallocenes)

Chiral epoxidation catalysts (Jacobsen)

Phase-transfer catalysts (quaternary ammonium/ phosphonium halides, Crown ethers).

In the reaction-separation systems that have been developed, it has been demonstrated that recovering a batch of catalyst from a post-reaction mixture using the membrane, and recycling it multiple times, while the reaction product is completely transmitted through the membrane, is a robust process. Typically, excellent catalyst separation is observed, with high selectivity between the catalyst and product, and low energy usage is required compared with distillation and chromatography. Furthermore, gains in operating costs can be made through catalyst cost savings by reusing the catalyst - these cost savings may make processes economic which might otherwise have been prohibitively expensive to carry out at pilot/process scale.

As well as retaining the catalyst, nanofiltration also acts to produce a product with fewer impurities, thus reducing the treatment costs associated with product purification. The Starmem membranes which have been used in these experiments show high membrane stability, ensuring that separation performance is repeatable for multiple successive filtrations.

simple solutions

The principle of using molecular weight differences to separate molecules is clearly applicable to a wide variety of processes. In terms of product purification, the nanofiltration membranes can be used either to retain a large target molecule while allowing a smaller impurity to permeate, or, as appropriate, allowing the target molecule to permeate while retaining the impurity. Similarly, it can also be used to concentrate solutes in solvents, by retaining the molecule of interest while filtering through the solvent.

This application is particularly relevant in solvent-intensive unit operations such as distillation and chromatography, where, for example, the energy required to achieve a given concentration using OSN can be as little as 10% of that required for distillation.

Using OSN to purify products has been demonstrated for several processes, including: separation of desired Heck reaction products from byproducts in ethyl acetate (EA), methyl-tert-butyl-ether, tetrahydrofuran and acetonitrile; recovery of amine products and glutarate salts from industrial methylated spirit; recovery of solvent and by-product removal (both halogenated and non-halogenated) from post-sulfonation reaction toluene streams; recovery of several pharmaceutical intermediates with molecular weight greater than 650 Da in EA and toluene with byproduct removal to meet target product purity; and separation of unreacted monomer species from product polymer.

Aside from the energy- and cost-saving benefits offered by nanofiltration, the technique has additional advantages as it can avoid heat treatment or going to dryness in cases where thermally-sensitive products are involved.

The beauty of nanofiltration is its simplicity. The concepts are easy to grasp Ð as it says, it is filtering on the nanoscale. How then can manufacturing chemists access the technology in order to develop processes around it?

Professor Andrew Livingston, whose laboratories at Imperial College, London have pioneered several OSN applications, commented: 'To ultimately apply OSN at the production scale the processes must first be tried and tested on the bench scale. To this end we sought to develop and design equipment (figure 4) with manufacturing chemists in mind, allowing them to rapidly test new applications using membranes, whether they be product purifications or catalyst reactions.'

The equipment is a pressurised stirred cell known as the METcell, featuring an active membrane area of 51cm², a processable liquid volume of between 25 and 270ml, at a maximum pressure of 69 bar and maximum temperature of 50C.

The ancillary gas unit includes a regulator, gauges, and isolation, vent and pressure relief valves, ensuring safe and simple operation under pressure. The cell also includes a port for connection to an HPLC pump, allowing charging of solutions into the METcell without opening it, which is particularly useful when solutes are oxygen/moisture sensitive. In catalyst applications, the body of the cell acts as the reactor, and can be attached to a thermocouple in order to regulate the temperature.

Such a cell is useful for scoping and proof-of-concept work. The next stage of operation at the 0.5-5l scale, to test longer-term membrane stability and simulate flux and flowrate at process scale, would be carried out using dedicated cross-flow equipment.

At pilot and production scale, clearly a greater membrane surface area would be required, and this is provided using spiral wound modules, pictured. In the Max-Dewax process for which the Starmem membranes were originally developed, 8in x 40in modules are used, giving a surface area of 24m². For smaller applications, 2m² (figure 6) and 6m² modules are also available.

It is always exciting to be among the forerunners of a new technology. Currently, the industrial-scale applications of nanofiltration are limited. However, the possibilities are vast, and OSN has already attracted interest from companies both in Europe and in the US. Applications are continuously being developed based on industry needs.

Currently, this is on a case-by-case basis, but as the technology becomes more widely adopted, it is anticipated that many of these applications will become generic unit operations.