WO2022254018A1

WO2022254018A1 - An annular tubular phase separator

Info

Publication number: WO2022254018A1
Application number: PCT/EP2022/065213
Authority: WO
Inventors: Steven Ferguson; Matthew Harding
Original assignee: University College Dublin
Priority date: 2021-06-04
Filing date: 2022-06-03
Publication date: 2022-12-08
Also published as: GB202108007D0

Abstract

A continuous liquid-liquid phase separator (100) comprising a porous insert (101) housed concentrically within a non-permeable outer tube (102), the outer tube (102) having an inside surface (103), and wherein the configuration of the outer tube (102) and porous insert (101) form an annular gap (104) therebetween.

Description

Title

An annular tubular phase separator

Field of the Invention

The invention relates to a device that separates an immiscible or partially miscible mixture of liquid phases. Specifically, the invention relates to the device that separates an immiscible mixture of solvents due to their differing capillary pressure for a tubular membrane.

Background to the Invention

Liquid-liquid extraction (phase separation) is a standard unit operation in the chemical and pharmaceutical manufacturing industries. The density difference between immiscible liquids is typically the driving force for phase separation, which is performed in separating funnels for standard laboratory scale batch synthesis and settler tanks, centrifugal separators or plate coalescers for larger scale batch manufacturing. At microscales (typically mm and below), more readily encountered in microfluidics and continuous flow synthesis, gravity is not sufficient to overcome surface tension forces and may not provide complete phase separation in practical timescales, particularly where liquid densities are similar.

While examples can be found for continuous processing steps using gravimetric settling vessels, far more rapid coalescence and phase separation can be achieved at microscales using differential surface wetting. In these separators, the organic phase preferentially wets a hydrophobic surface such as poly(tetrafluoroethylene) (PTFE), while the aqueous phase preferentially wets a hydrophilic surface such as glass or stainless steel. Separators of the surface-wetting type were first designed in the late 1970’s for use in analytical chemistry and have been shown to function at flow rates as fast as 50 mL/min (3 L/hr) for a water/n-heptane mixture in an array of glass and PTFE.

Thin, flat, microporous PTFE membrane for phase separation have been used previously in flow injection analysis (FIA), comparing PTFE and cellulose acetate as membrane materials. While commercial versions of this separator type have been shown to be highly effective, they can be expensive, even for laboratory scale variants, and are substantially more costly for units capable of processing up to 3 L/min total flow. This has limited their wider uptake. The use of tubular membranes as a surface-wetting based phase separator followed shortly after the development of the flat microporous membrane separator, with porous PTFE tubing being employed to allow permeation of gaseous analytes into a liquid stream for analysis. A porous tubular membrane was then shown to be capable of acting as a liquid-liquid phase separator in 1988, where a range of membrane porosities from 20-70 % were investigated to separate mixtures of carbon tetrachloride and aqueous solutions prior to analysis with inductively coupled plasma optical emission spectroscopy (ICP-OES).

In recent years it has been shown that tubular PTFE membranes could provide excellent separation efficiencies at micro and meso scale flow rates, and for a wide range of solvent mixtures, including mixtures of perfluorinated and organic solvents with smaller interfacial differences than those typical in flow injection analysis (FIA) work, such as chloroform and aqueous. Flow rates achieved with tubular membranes so far range from 0.5 - 12 mL/min, which is already suitable for small scale continuous synthesis but insufficient for large scale manufacturing.

Previously reported tubular membrane phase separators have functioned by pumping the liquid mixture into the centre of the tubular membrane and forcing the preferentially wetting phase to permeate out, leaving the non-preferentially wetting phase flowing within the channel. While effective for small capillary diameters, as the internal diameter (I.D.) of the membrane increases, density driven separation can begin to occur within the channel, leading to reduced permeation rates and lowered throughput as up to half of the available membrane surface area is lost. In addition, the flux of the preferentially wetting liquid may be reduced further due to effects like annular liquid- liquid flow or suspended bubbles within the membrane capillary preventing efficient contact with the membrane surface.

The use of tubular membranes for separator construction and scale-up, with linear improvements in membrane surface area as either membrane length or diameter increase have been developed but with the relatively low throughputs currently being reported.

Continuous flow phase separators utilising flat microporous membranes are commercially available (Zaiput, Syriss FLLEX). Capillary phase separators operating on a similar basis to flat membrane separators have been published in the literature but are not commercially available. These separators can be applied to telescoped continuous manufacturing. Centrifugal separators which magnify the separation of immiscible liquids due to differing densities are also available and used in large scale manufacturing. Settling tanks, large batch vessels that allow the solvents to separate based on density, enhanced by tapering walls, are also commonplace.

WO 201/069770 describes an elongated fluid degassing apparatus comprising one more tubular separation membranes and an outer flexible liquid-permeable jacket having an aspect ratio of at least two. The apparatus is for gas-liquid separation.

EP2361660 describes an elongated flow-through degassing apparatus comprising a gas and liquid impermeable outer tube and one or more gas-permeable, liquid- permeable elongated inner tubes. The apparatus is for gas-liquid separation.

GB2530282 describes a condensate drain comprising a body defining a drain chamber having a liquid-gas inlet and a liquid outlet, a porous membrane having a pore size of about 0.2mGh or less is disposed within the body between the inlet and outlet. The apparatus is for gas-liquid separation.

However, larger scale flat sheet phase separators require custom built membrane equipment and as such can be extremely expensive. The thin flat sheet can require a specialized often integrated differential pressure controller as well as an external back pressure regulator for the successful phase separation to occur due to the small differences in transmembrane pressure associated with breakthrough of the preferentially wetting and non-preferentially wetting liquid phases. The Syrris FLLEX module is limited in its operating total flow rate to ca. 1 mL/min. Gravity/density-based separators suffer from long hold up times and are not well suited to being used as part of a telescoped synthetic route.

It is an object of the present invention to overcome at least one of the above-mentioned problems.

Summary of the Invention

Continuous flow synthesis in the pharmaceutical industry and in fine chemical production often encounters bottlenecks in production due to purification and isolation of intermediates and end products. Continuous liquid-liquid extraction is one method of purification used, but requires subsequent separation of the immiscible solvents, which if performed by relying on density differences can be slow to occur for some solvent systems where liquid densities are similar or when emulsions are formed and can simply take a long time due to the large size of the batch vessels.

The invention relates to a low-cost, modular, robust, and easily customisable continuous liquid-liquid phase separator that uses a tubular membrane and annular channels to allow high fluidic throughputs while maintaining rapid, surface wetting dominated, phase separation. The claimed system is constructed from standard fluidic tube fittings and allows leak tight connections to be made without the need for adhesives, or O-rings. Accurate control of the annular gap promotes separation via surface tension instead of gravity and results in faster and more compact separators. This can be maintained because as the tube diameter is increased, the increase in throughput is quite dramatic. The relatively high differential pressure for separation with, for example, a tubular expanded PTFE (ePTFE) membranes as utilised in this study, compared to flat sheet membranes, should allow the use of standard back pressure regulators (BPRs) to control phase separation, which will be of significant utility in scale up and manufacturing. However, any tubular microporous membrane material could be used, for example, a hydrophilic membrane.

An additively manufactured diaphragm based BPR was developed and printed in poly(etheretherketone) (PEEK) via fused filament fabrication (FFF), allowing highly accurate and adjustable pressure control with excellent chemical compatibility.

The units of the claimed system have been shown to operate at flow rates of 0.1 - 300 mL/min in the laboratory setting, with equivalent residence times from 80 to 4 seconds, demonstrating the simplicity of scale-up with these units. Further scale-up to litre per minute scales of operation for single units, and tens of litres/minute through limited numbering up, should allow these low-cost concentric annular tubular membrane separators to be used at continuous production scales for pharmaceutical applications for many solvent systems. Scale up can be achieved in three ways: increase the diameter of the tube; at a fixed tube diameter increase the length; and increase the number of units used in tandem.

The invention described herein would greatly increase the capacity of current batch vessels as no space would be required for the extraction solvent, which could be introduced inline during recirculation and removed before returning solution to the batch vessel, with for example impurities or un-wanted reagents removed within extraction solvent removed by the membrane separator. In such a scenario higher yield may be achieved as the entirety of the retentate solvent volume can be removed cycled through and returned to the batch vessel as it will be quantitatively retained by the membrane. In contrast some yield is often lost in separation under gravity in batch vessels as part of the light phase is retained within the vessel for example during bottom removal, particularly when an ill-defined separation layer or stable emulsion is to be separated. In such scenarios enhanced coalescence by surface wetting and retention of desired retentate phase by the membrane will typically allow complete resolution of the liquid phases without loss. Multiple separators and inline mixers together could be utilized to conduct counter current or other multistage separations in batch or continuous manufacturing improving yield and lowering solvent consumption compared to batch extraction. The invention described herein has further applications in continuous manufacturing as it generates a constant flow of reagent or synthetic step product in solvent. The invention described herein also tolerates being operated with downstream pressure, so it can telescope directly into subsequent continuous manufacturing steps that may have a large pressure drop. In the event that downstream pressure is higher than breakthrough pressure, the user can put a BPR on the permeate side, or a differential pressure regulator and an additional back pressure regulator on the retentate side for pressure control.

There is provided, as set out in the appended claims, a continuous liquid-liquid or liquid-gas phase separator (100) comprising a porous insert (101) housed concentrically within a non-permeable outer tube (102), the outer tube (102) having an inner channel (105) and an inner surface (103), and wherein the configuration of the outer tube (102) and porous insert (101) form an annular gap (104) therebetween.

In one aspect, the annular gap (104) is less than or equal to about 15 mm. Preferably, less than or equal to about 14 mm. Preferably, less than or equal to about 13 mm. Preferably, less than or equal to about 12 mm. Preferably, less than or equal to about 11 mm. Preferably, less than or equal to about 10 mm. Preferably, less than or equal to about 9 mm. Preferably, less than or equal to about 8 mm. Preferably, less than or equal to about 7 mm. Preferably, less than or equal to about 6 mm. Preferably, less than or equal to about 5 mm. Preferably, less than or equal to about 4 mm. Preferably, less than or equal to about 3.5 mm. Preferably, less than or equal to about 3 mm. Preferably, less than or equal to about 2.5 mm. Preferably, less than or equal to about 2 mm. Ideally, the annular gap (104) is less than or equal to about 5 mm, such as less than or equal to about 5mm, 4.5mm, 4mm, 3.5mm, 3mm, 2.5mm or less than or equal to about 2 mm.

In one aspect, the inner surface (103) of the outer tube (102) has a hydrophobicity opposite to that of the porous insert (101).

In one aspect, the inner surface (103) of the outer tube (102) and the porous insert (101) have the same hydrophobic or hydrophilic properties.

In one aspect, the porous insert (101) is composed of a hydrophilic material selected from the group comprising a ceramic, alumina, glass, a thermoplastic polymer, a copolymer, or a thermoset polymer.

In one aspect, the outer tube (102) is composed of a material selected from the group comprising stainless steel, PFA, glass, glass lined metal, an inert metal, a polymer, a ceramic, and silicon carbide.

In one aspect, the porous insert (101) is configured to be tubular, flat, square or having triangular cross-sectional channels.

In one aspect, the separator (100) is in the form of a linear configuration, a coiled configuration, a circular configuration, a curved configuration, or a combination of such configurations put together in a series, in parallel or counter current arrangements.

In one aspect, the inner channel (105) accommodates a central membrane channel (106) within the porous insert (101). Preferably, the central membrane channel (106) is a cylindrical rod.

In one aspect, the porous insert (101) is a first porous membrane.

In one aspect, the inner surface (103) of the outer tube (102) comprises a second porous membrane. The use of an annulus confers several advantages. Flowing the immiscible solvent mixture into an outer annulus of controlled width of less than 15mm, 14 mm, 13 mm, 12 mm, 11 mm, 10 mm, 9 mm, 8 mm, 7 mm, 6 mm, 5.5 mm, 5 mm, 4.5 mm, 4 mm, 3.5 mm, 3 mm, 2.9 mm, 2.8 mm, 2.7 mm, 2.6 mm, 2.5 mm, 2.4 mm, 2.3 mm, 2.2 mm, 2.1 mm, 2.0 mm, 1.9 mm, 1.8 mm, 1.7 mm, 1.6 mm, 1.5 mm, 1.4 mm, 1.3 mm, 1.2 mm, 1.1 mm, 1.0 mm, 0.9 mm, 0.8 mm, 0.7 mm, 0.6 mm, 0.5 mm, 0.4 mm, 0.3 mm, 0.2 mm, 0.1 mm, allows fast flow rates with low pressure drops and low Reynolds numbers, enhances coalescence and maximises available membrane surface. The use of an annulus also increases the available membrane area compared to a flat sheet with the same footprint and allows easy scale-up by maintaining the annulus gap while the membrane diameter is increased - maintaining the annular gap is simple due to the wide availability of tubing in varying diameters and wall thicknesses. For emulsions, the annular gap would need to be narrower, for example, 6 mm, 5 mm, 4 mm or 3 mm, while for some systems that separate well and form separated slugs of immiscible fluid spontaneously 10mm, 11 mm, 12 mm, 13 mm, 14 mm, or 15 mm will function continue to function efficiently.

The device of the claimed invention could also be used to introduce reagents in a uniform manner over the course of the length of a tube from either the outside in or from the inside out for both gas and liquids, a gas-liquid separator, and a liquid-liquid mixer.

Definitions

In the specification, the term “fluid” should be understood to mean a substance that continually deforms (flows) under an applied shear stress, or external force, and should be understood to include both a liquid and a gas.

In the specification, the term “liquid” should be understood to mean a nearly incompressible fluid that conforms to the shape of its container but retains a (nearly) constant volume independent of pressure.

In the specification, the term “gas” should be understood to mean substance or matter in a state in which it will expand freely to fill the whole of a container, having no fixed shape (unlike a solid) and no fixed volume (unlike a liquid). In the specification, the term “opposite hydrophobicity” should be understood to mean opposite characteristics. For example, if the outside tube is hydrophilic, the inner membrane is hydrophobic, and vice versa.

In the specification, the term “inert metal” should be understood to mean a metal selected from ruthenium (Ru), rhodium (Rh), palladium (Pd), osmium (Os), iridium (Ir), platinum (Pt), gold (Au), silver (Ag), copper (Cu), titanium (Ti), rhenium (Re), stainless steel, iron, and brass, or alloys thereof. Preferably, the inert metal is selected from silver, gold, copper, platinum, and titanium. Alloys can be, for example, a nickel- molybdenum alloy (Hastelloy®), a nickel-chromium-molybdenum alloy, a nickel- chromium-iron-molybdenum alloy, a nitrogen-strengthened stainless-steel alloy (such as Nitronic 60), an austenitic nickel-chromium-based superalloy (such as Inconel®), and other chemically-resistant alloys and oxidation-corrosion-resistant alloys or glass (Si02) lined metal tubing.

In the specification, the terms “plastic” and “polymer” should be understood to mean a material consisting of any of a wide range of synthetic or semi-synthetic organic compound that is malleable and can be moulded into solid objects. The terms can be used interchangeably. Plastics or polymers can be grouped into thermoplastics, thermosetting polymers, amorphous plastics, crystalline plastics, conductive polymers, biodegradable polymers, bioplastics, and the like. Preferably, the polymer is a thermoplastic, a thermoset or a copolymer.

When the polymer is a thermoplastic it may be selected from, but not limited to, the group comprising acrylonitrile butadiene styrene, polypropylene (PP), polyethylene, low-density polyethylene (LDPE), polyethylene terephthalate (PET), polyvinylchloride, polyamide, polyetherimide (PEI), polyester, acrylic, polyacrylic, polyacrylonitrile, polycarbonate, ethylene-vinyl acetate, ethylene vinyl alcohol, polytetrafluoroethylene (PTFE), expanded polytetrafluoroethylene, fluorinated ethylene propylene (FEP), PEEK, polyetherketoneketone (PEKK), ethylene chlorotrifluoroethylene, ethylene tetrafluoroethylene, liquid crystal polymer, polybutadiene, polychlorotrifluoroehtylene, polystyrene, polyurethane, nylon, and polyvinyl acetate.

When the polymer is a thermoset, it may be selected from, but not limited to, the group comprising vulcanised rubber, bakelite (polyoxybenzylmethylenglycolanhydride), urea- formaldehyde foam, melamine resin, polyester resin, epoxy resin, polyimides, cyanate esters or polycyanurates, silicone, and the like known to the skilled person.

When the polymer is a copolymer, it may be selected from, but not limited to, the group comprising copolymers of propylene and ethylene, copolymers of tetrafluoroethylene (C2F4) and perfluoroethers (such as perfluoroalkoxy alkanes (PFA)), acetal copolymers (Polyoxymethylenes), polymethylpentene copolymer (PMP), amorphous copolyester, polyethylene terephthalate glycol (PETG), acrylic and acrylate copolymers, polycarbonate (PC) copolymer, styrene block copolymers (SBCs) to include poly(styrene-butadiene-styrene) (SBS), poly(styrene-isoprene-styrene) (SIS), poly(styrene-ethylene/butylene-styrene) (SEBS), ethylene vinyl acetate (EVA) and ethylene vinyl alcohol copolymer (EVOH) amongst others.

Brief Description of the Drawings

The invention will be more clearly understood from the following description of an embodiment thereof, given by way of example only, with reference to the accompanying drawings, in which:-

Figure 1 illustrates (a) an exploded CAD image showing a two-part BPR construction and (b) a transparent view showing the internal channels (highlighted) of the BPR of Figure 1(a). It is noted that the BPR illustrated in Figure 1 does not form part of the invention but is an example of a BPR that can be used with the claimed invention.

Figure 2 illustrates one half of an asymmetrical tubular arrangement of the claimed invention showing the resulting annular gap. The separator is connected to one side of a T-union connector 202, with a reducing union 204 on the other side of the T- union connector 202. Tubular ePTFE membranes with outer/inner diameters (mm) of 3.00/2.10, 6.35/3.00 and 12.7/9.53 mm were used for all separators used to demonstrate the invention. Swagelok® parts used for the construction of typical lab scale and larger scale separators were also used. The amount of housing tubing required will vary with the desired length of separator.

Figure 3 illustrates (a) an overview of annular operating mode, where Pper = permeate channel back pressure, Pret = retentate channel back pressure; and (b) an assembled annular phase separator of the claimed invention, with an effective separator region in this formation indicated with a dashed line. As per Figure 2, there is an annular section (external to the membrane tube, which is inside the steel tube where the separation takes place). The T-junction is a standard tube fitting and is used to feed a two-phase feed into the annular gap from the left as per Figure 3(a). The T-piece on the right removes the retentate from outlet Pret., while the outlet Pper. perpendicular to this takes out the permeate stream that has passed through the membrane. The membrane runs the length of the whole tube fitting.

Figure 4 shows (a) the retentate output volume ratio (annulus residence time becomes insufficient at total flow rates of between 6 and 8 mL/min); and (b) the permeate output volume ratio (slight aqueous breakthrough is clear at flow rates < 0.5 mL/min). Insufficient membrane surface area, or residence time, in the annular channel was provided at the operating transmembrane pressure and at flow rates above 8 ml/min to allow the entirety of the preferentially wetting hydrophobic (“organic”) phase permeating through the hydrophobic ePTFE, thus resulting in organic phase passing out the retentate outlet from the annulus. This result shows what the failure modes mapped out in the later graphs look like in real operation.

Figure 5 illustrates a scatter plot showing the operating region for a 1.03 g/cm³ annular phase separator with 1.1 mm annular gap. Regions are illustrated only approximately with shading. This plot illustrates a point of maximum productivity for the configuration and is indicative of large-scale performance.

Figure 6 shows a graph identifying operating regions for a 1.15 g/cm³ membrane with a 0.7 mm annular gap. This plot illustrates a point of maximum productivity for the configuration and is indicative of large-scale performance.

Detailed Description of the Drawings

The invention works by separating an immiscible mixture of liquids due to their differing capillary pressure for a tubular membrane. The device described herein allows fast flow rates (> 100 mL/min) to be achieved while maintaining laminar regimes because of the annular channel profile, while also maintaining rapid phase separation based on surface wetting rather than solvent density. This reduces the holdup time to the residence time of the device, which has been found to be as low as 2-3 seconds.

The system is also extremely cheap to construct compared to commercial units using standard fittings and tubing rather than a system with a designed and manufactured flat membrane housing. Because of this the membrane is sealed as part of the tube fittings without the need for O-rings. The annular strategy allows the system to be scalable in such configurations without the need for custom built flat sheet configurations. Referring now to the figures, where Figure 1 illustrates a general embodiment of a back pressure regulator used with the present invention. Specifically, Figure 1(a) and Figure 1(b) illustrate a back pressure regulator 1 comprising a body 2 having a first body part 2a and a second body part 2b, the second part 2b further comprising an inlet 3 and an outlet 4. The first part 2b further comprises a threaded port 5 and a plurality of apertures 6. The threaded port 5 defines a central core 7a, which is continuous with a central core 7b in the second body part 2b, and within which a pressurisation line is accommodated. The inlet 3 and the outlet 4 of the second body part 2b are typically threaded to accommodate tubing.

The second body part 2b further comprises an outer indentation 8 and an inner indentation 9, both configured to accommodate an O-ring 10,11 of a particular diameter. The central core 7a, 7b is inside the inner indentation 9. Sandwiched between the O-ring 10,11 is a polymer film 12. The polymer film 12 fully covers the O-rings 10,11 with a small excess. The same arrangement of outer and inner indentations 8,9 is arranged on the underside 14 of the first body part 2a. The O-rings 10,11 are BS020 standard size (1.78 mm CS, 21.92 mm I.D., 25.51 mm O.D.). Perfluorinated/FFKM O- rings can be used to give the widest chemical compatibility.

Figure 2 and Figure 3(b) illustrate a general embodiment of a continuous liquid-liquid phase separator of the present invention. Specifically, Figure 2 illustrates one aspect of the present invention and is generally referred to by reference numeral 100. The continuous liquid-liquid phase separator 100 comprises a porous insert 101 housed concentrically within a larger diameter non-permeable outer tube 102. The outer tube 102 comprising an inner channel 105 and inner surface 103. The porous insert 101 is placed within the inner channel 105. The porous insert 101, when in a three- dimensional form (such as a tube, a square, a triangle etc.), further comprises a central insert channel 106. The central insert channel 106 is the internal channel inside the porous insert 101 where the permeate (preferentially-wetting phase) passes through the membrane. The inner surface 103 of the outer tube 102 can be of opposite hydrophobicity to the porous insert 101 to enhance separation operation based on preferential wetting and would be preferred for highest efficiency. The outer tube 102 can be constructed out of hydrophobic material too, as the non-preferentially wetting solvent membrane is prevented from getting through the porous insert 101, by operating below a transmembrane pressure between the annular gap 104 and internal channel 106 where permeate flows. In one aspect, if the porous insert 101 is a tubular membrane (a first porous membrane), it is possible to configure an annular gap 104 concentrically by putting a tube around the tubular membrane making a concentric annular gap or insert a cylindrical rod down the centre of the porous insert (tubular membrane) 101 (a central membrane channel 106) creating an annular gap between the rod and the first porous membrane (porous insert 101).

The outer diameter of the porous insert 101 is such that when housed within the outer tube 102, the resulting annular gap 104 is below 6 mm for systems that need significant surface wetting to operate and below 15 mm for systems that can form stable two slug flow sufficiently rapidly. In the working examples provided herein, the annular gap 104 is 3 mm or less. When an immiscible solvent flow is pumped into the annular gap 104, the controlled annular gap 104 can enhances coalescence of the different solvent phases and also allows surface wetting to be the dominant separation force rather than gravity. When the annular gap 104 is used for the separation of the immiscible solvent flow, the small gap can be maintained as the overall diameter of the separator outer tube 102 is made bigger, by selecting a wider tubular membrane 101 diameter also. Hence, this allows scalable separation while staying within the surface wetting, rather than gravity driven, separation regime.

The annular gap 104 also allows for faster flow rates while maintaining a laminar flow regime compared to a circular channel of the same profile area, thereby improving throughput. As the onset of turbulence will undo the process of coalescence and preferential wetting, ultimately this is the limiting factor for scale at a given pipe diameter. The central insert channel 106 transports the permeate flow to the device outlet following permeation, the inner channel could also be made annular via an inserted cylindrical rod or tube, with separation of components occurring in the reverse direction. However, for maximum available membrane surface area, and because onset of turbulence within the flow will occur at lower flow rates within an annular gap created within the central insert channel 106 within the porous insert 101. As such, an outside annular space for multiphase liquid flow with flow inside through the tubular membrane of permeate flow is preferred. The separator 100 used in the examples below was uses a hydrophobic porous insert 101 of expanded PTFE (ePTFE) and a hydrophilic stainless-steel outer tube 102 to house it. Connections are made using Swagelok®, a proprietary fluidic/gas connection system that is suitable for high pressures and used in numerous manufacturing plants. The porous insert 101 has a tube inserted a short length into each end before the Swagelok® connections are tightened to provide the leak tight seal to the central insert channel 106 and to allow downstream connection and capping as necessary. It should be noted that the separator 100 can be constructed from any pipe material or tube fittings.

Materials and Methods

Solvents were pumped using peristaltic pumps (PLP-380, behr Labor Technik, Dusseldorf, Germany) where applied pressures were approximately 30 psi (0.21 MPa) or lower. Silicone tubing was used to deliver ethyl acetate and water. Viton tubing was used to deliver toluene and heptane. Where applied pressures were greater than approximately 30 psi, HPLC pumps were used (LD class, Teledyne SSI, State College, PA, USA). Back pressure was provided by using spring-based cartridges or short lengths of PTFE tubing (0.3 - 0.8 mm I.D.), except for where numerous pressure values were required at various flow rates. For these experiments, an additively manufactured (AM) BPR printed in PEEK, was used, printed using a Funmat HT fused filament fabrication (FFF) printer (Intamsys, Shanghai, China) using recommended manufacturer conditions and annealed according to manufacturer recommendations. The BPR operates by sealing an inlet with a 50 pm thick fluorinated ethylene polymer (FEP) or PEEK film which is pressurised, upstream liquid must overcome this pressure to exit through a second recessed outlet on the same face as the film sealed inlet (see Figure 1). In-line pressure measurement upstream and for the BPR was performed using analogue pressure gauges and used to accurately set retentate pressures by controlling the film pressure, the great advantage being that pressures anywhere from 0 - 60 psi can be varied in small increments with no disconnections.

Output from the retentate and permeate channels was collected into measuring cylinders and the time taken for the volume to be collected recorded so that accurate flow rates could be determined and the effect of changing pressure on pump performance be accounted for. The collected outputs were left to stand briefly, and the volume ratio of each phase recorded. Separator construction

To improve upon the limitations of tubular membrane systems described above and allow for surface-wetting driven separation at high throughputs, a new approach to the use of such membranes was envisaged that can utilize tubular membranes of larger diameters, hence allowing higher permeation flowrates, while continuing to have narrow channel diameters within length scales (approximately 3 mm) that allow for surface tension rather than gravity driven separation of the immiscible liquid phases. Flowing the immiscible liquid mixture into the annular gap 104 between the porous insert 101 and the inner surface 103 of the tube 102 and driving permeation via differential transmembrane pressure of the preferentially wetting solvent to the central insert channel 106 (see Figure 2) maximises the available membrane surface area compared to flowing into the central membrane channel. When compared flowing the mixture through a membrane with a 1 mm I.D. and 2 mm outside diameter (O.D.), simply switching to contacting the outer membrane surface with the liquid mixture immediately doubles the available surface area. It would also be expected that faster flow rates can be achieved within an annulus gap compared to a circular profile channel of the same O.D., while remaining in a laminar flow regime, so phase separation by preferential membrane wetting will not be disrupted by turbulent mixing within the separator.

The use of a tube-in-tube configuration allows control over the annular gap width to maintain surface tension as the dominant driving force in the separation. It is estimated that annular gaps greater than 3 mm would start to allow density driven separation within the annulus, reducing effective throughput. The minimum annular gap should be any distance that continues to allow two liquid phases to exist independently. Selecting a hydrophilic outer tubing material confers an additional benefit to separation, as stainless steel or glass will be preferentially wetted by aqueous solvents compared to the hydrophobic membrane surface, helping to drive coalescence. This arrangement could be easily reversed if desired to have a hydrophilic ceramic or alumina membrane on the inside and a hydrophobic PTFE or PFA housing tube. Please note that the device will still function when the device does not have a wall of opposite hydrophilicity. It is helpful there is a wide range of tubing material and tubing diameter combinations commercially available, making selection of the appropriate tubing material and sizing easy. The membrane material chosen in this study was ePTFE purchased from Zeus Industrial Products (Orangeburg, NJ, USA). The density of the ePTFE tubing is inversely related to the porosity (the percentage volume of the membrane due to pores), so a higher density membrane would be expected to have smaller and fewer pores, and result in higher breakthrough pressures for the aqueous phase. The ePTFE used in this work ranged from 1.03 - 1.52 g/cm³, or approximately 50-30 % porosity, respectively. The operating principle for separation is the same as for other membrane phase separators, the pressure on the retentate channel is set to be above breakthrough of the preferentially wetting phase but below breakthrough of the non- preferentially wetting phase.

Table 1. ePTFE dimensions

Membrane No. 1 (3.0 mm O.D., 2.1 mm I.D.) was purchased from stock material at Zeus and required the insertion of a small piece of 2.5 mm O.D. PTFE tube to flare the ePTFE membrane and allow 3.175 mm (1/8 in.) I.D. PTFE ferrules to seal fully. Downstream connections were also made to this tubing, which being non-standard in flow chemistry, required a sheath of 3.175 mm O.D. PTFE tubing to allow proper sealing. This approach avoids the use of adhesives entirely and meant that connections were not at risk of degrading with prolonged solvent contact. All other dimensions (#2-5) were custom extrusions and were chosen with dimensions that would allow direct use of Swagelok® tube fittings. While stainless steel ferrules can be successfully utilised the use of PTFE ferrules to seal the membrane meant that the construction was not permanent, unlike the use of metal ferrules, and allowed the membrane to be removed and re-used in a different housing without cutting. In addition, the dimensions specified result in very robust membrane material, capable of being handled as if it were regular tubing. Separator construction was entirely from standard Swagelok® fittings, generally consisting of two Tee-unions, two reducing unions, and whatever unions were required to connect to the feed pump and outlet reservoirs (see Figure 3). There is some deviation from the housing tube diameter, typically being greater, when the membrane is within the T-union depending on the housing tubing being used. For the larger membrane dimension (Membrane No. #5) run tees were used, which have a female NPT thread opposite from a tube fitting and help to reduce the dead volume of the system.

Solvent Systems

The pressure difference between controlled breakthrough of the preferential wetting phase and unwanted breakthrough of non-preferentially wetting phase will typically decrease as the interfacial tension of the solvent system also decreases. For a system like n-heptane and water this value is high at approximately 50 mN/m allowing for a high applied pressure before aqueous permeation occurs. Whereas for a common system like ethyl acetate and water this value drops to 6.8 mN/m and the resulting acceptable pressure range will drop significantly, making control of the process more difficult. For liquid-liquid extraction applications in flow chemistry it would be reasonable to expect that there will be organic solvents present that are miscible with both phases, such as ethanol, acetonitrile, and tetrahydrofuran. All of these will lead to a reduction in the interfacial tension compared to a binary system and reduce the pressure window that can be applied for successful separation.

Results and Discussion Initial Testing

Initial experiments investigated flow rate only by separating ethyl acetate and water using Membrane No. 1 (1.09 g/cm³, 3.0 mm O.D., 17 cm effective separator region) inside PFA tubing with an I.D. of 4 mm resulting in an annular gap of 0.5 mm. The applied back pressure was nominally set close to the aqueous breakthrough pressure. The fastest achievable total flow rate with this configuration to give full separation was 6 mL/min, with 3 mL/min flow rates of ethyl acetate and water (see Figure 4). This flow rate corresponds to a maximum annulus residence time of around 19 s for this configuration.

This is indicative of the design choices faced in the development of membrane-based liquid-liquid separations. Increasing the flowrate in a fixed device geometry, membrane composition and differential pressure, decreases the residence time available for coalescence and the available membrane area for separation per unit mass flow of feed. The Reynolds number is increased also, which at higher values or disruptions in the flow (inlets and outlets) may impact preferential wetting of the desired (organic) permeate phase onto the membrane. Increasing the diameter of the annular gap 104 at fixed channel length has no effect on the available membrane area per unit mass flow of feed, it does however reduce potential mixing of the phases at the inlet and increases the residence time to allow separation through preferential wetting and then selective permeation of the organic phase. It also increases the range over which localised preferential wetting of the membrane must act, which may lead to increased residence time requirements for membrane wetting and ultimately permeation.

Changing the housing to a 9.525 mm (3/8 in.) stainless steel tube allowed an increase in the annulus volume due to the increase in the annular gap 104 to 2.4 mm, even with a shorter 9 cm effective length of membrane. This combination was able to reach a total flow rate of 29 mL/min with equal flows of heptane and water, corresponding to an effective annulus residence time of approximately 15 s. Housing Membrane No. 2 (1.03 g/cm³, 6.35 mm O.D.) within a 9.525 mm O.D. tube gave an annular gap 104 of 0.7 mm due to the 0.89 mm wall thickness. With an effective tubing length of 31.5 cm, the annulus volume was 4.9 ml_. The highest total flow rate that could be fully separated was 44 mL/min with equal flow rates of heptane and water, equating to a residence time of approximately 9 seconds.

Operating region determination - heptane water separation

A more in-depth investigation was performed by systematically varying both the applied retentate pressure and the total heptane and water flow rate for Membrane No. 2 (1.03 g/cm³, 6.35 mm O.D.) within a fixed 1.1 mm annular gap. This revealed an operating region bounded by expected failure regions (see Figure 5). When the applied pressure is excessive, permeation of the non-preferentially wetting phase occurs (approximately marked as a grey region). For a fixed membrane geometry as the flow rate increases the annulus residence time eventually becomes insufficient for full permeation of the preferentially wetting phase, and organic solvent is seen in the retentate (blue region). Finally, when the applied retentate pressure is excessive, combined with insufficient residence time, retention of the preferentially wetting phase and permeation of the non- preferentially wetting phase are observed (red region). The organic in retentate region (blue) would be expected to extend below the successful operating region, below differential pressures where the preferentially wetting phases can permeate the membrane although this was not mapped out in this experiment because a minimum value of 2 psi was always set. Under ideal conditions these four regions will intersect at a single point, marking the maximum productivity/minimum residence time for this membrane separator and two- phase liquid system combination corresponding to complete separation. This operating region scatter plot in Figure 5 is represented in terms of residence time here, however it could also be presented in terms of flowrate as in Figure 6. The point of maximum productivity with complete separation found at the minimum residence time and maximum tolerated retentate pressure will be a system dependant combination of the timescales for coalescence/membrane wetting, and of the organic phase flux through the available membrane area. If mixing effects can be ignored due to maintenance of laminar flow and annular gap on scale-up through increased unit diameter and length, minimum residence time for complete separation may provide a simple first approximation for scale-up of these systems.

The minimum residence time for complete separation with this membrane configuration is 2-3 seconds, corresponding to a total flow rate of 96 mL/min with 10 cm of effective membrane. Similar testing for Membrane No. 3 (1.15 g/cm³, 6.35 mm O.D.) found a minimum residence time of approximately 16 seconds. Given identical geometry and materials of construction, this increase in observed minimum residence time is due to reduced flux through a membrane of increased density, with the area available then being insufficient to allow complete permeation organic phase at flowrates lower than the equivalent minimum residence time near the breakthrough pressure for the aqueous phase. Given the highly divergent affinity for the membrane and binary heptane water systems, available membrane area will remain the limiting factor in the productivity. Table 2. Comparison of the effect of ePTFE density on the separation of heptane-water and ethyl acetate-water systems with equal volumetric flowrates in 6.35 mm O.D with a fixed 1.1 mm annular gap of length 10cm.

For ePTFE membranes this reduction in throughput is to be expected as the number of pores and their size reduces with increased density. For example, the lowest residence time achieved with membrane 4, the densest material used in this work (1.52 g/cm3, 6.35 mm O.D.), was found to be approximately 83 seconds. Membranes of higher density would be required for systems with a low interfacial tension and so a balancing of suitable membrane density with maximal throughput is required.

Permeation pressures for different density membranes

An HPLC pump (LD class, Teledyne SSI, State College, PA, USA) was used to deliver water or heptane into a 20 cm length of 6.35 mm O.D. membrane with one end sealed. The pressure generated due to permeation at varying flow rates was recorded via an in-line pressure gauge or by the pump’s built in pressure transducer.

Table 3. Permeation pressures for water through ePTFE membrane

Table 4. Permeation pressures for n-heptane through ePTFE membrane

Effect on operating region with changing interfacial tension - 6.35 mm O.D., 1.15 g/cm³ membrane.

The 1.03 g/cm³ density membrane allows for rapid separation of high interfacial tension systems, however when a low interfacial tension system like ethyl acetate/water (6.8 mN/m) is used, a high porosity ePTFE membrane offers a low breakthrough pressure to water which can make stable operation difficult. In the case of an ethyl acetate water system the maximum transmembrane pressure for the ethyl acetate water separation was found to be 0 psi (/.e., separation could not be achieved) for 1.03 g/cm³ (membrane 1), 2 psi for the 1.15 g/cm³ (membrane 3) and 12 psi for the 1.5 g/cm³ (membrane 4).

The effect on interfacial tension on the operating region of the membrane was investigated by using a 1.15 g/cm³ membrane of 17.5 cm effective length in a 3/8 in. O.D. stainless steel housing with combinations of heptane, toluene and ethyl acetate with water (see Figure 6).

The operating range for higher interfacial tension systems is far wider than for the lower tension examples, as can be seen in Figure 6. The tapering effect of the successful separation region at higher flow rates is again to be expected as the point of maximum productivity/ minimum residence time for complete separation is approached, robust operation would be achieved by running further within the operating regions identified in Figure 6.

It was further found that increasing the annular gap size above 0.7 mm allowed for higher retentate pressures to be applied, giving higher throughputs. This may be indicative of increased homogenization of aqueous and organic phases as flow accelerates into the annular gap. In the case of heptane and water, an increase in the annular gap to 1.1 and 1.9 mm resulted in minimum residence times approximately half that of the 0.7 mm gap width. The large increase in throughput from the change in annular gap is due to the increase in annular volume and hence residence time, and the increase in maximum applied pressure - in the case of 1.1 mm and 1.9 mm gaps, the retentate pressure could be set at up to 15 psi with no permeation of water visible when the 0.7 mm gap operating pressure was limited to 10 psi.

Operation of the same unit with multiphasic feed pumped into the centre of the membrane tube rather than the annulus, and flow from inside to outside of permeate would still be expected to work for the 6.35 mm O.D. membrane used for this experiment which has an I.D. of 3.0 mm, and successful operation at up to 21.6 mL/min could be achieved before retention of the organic phase was observed. This is more than double what could be processed through an annular gap of 0.7 mm with the same length of membrane but dwarfed by the 1.1 mm and 1.9 mm annular gap sizes where the increase in membrane surface area could be fully realised and total flow rates of 41 and 63 mL/min could be processed respectively. Extended operation of an annular separator, recycling the system output back, showed that it was also extremely stable. Settings of 1.15 g/cm³, 1.1 mm annular gap and 9 psi of retentate pressure applied with a diaphragm based BPR allowed continued operation for 53 hours with no change to the solvent reservoir composition, processing the equivalent of 33 L of biphasic solvent mixture.

Scale-up of concentric tubular annular separators

The scaled-up version of the annular separator was larger in both length and diameter than previous versions, with Membrane No. 5, 12.7 mm (0.5 in.) O.D. and 1.15 g/cm³, being used. The housing tube was stainless steel with an O.D. of 19.05 mm (0.75 in.) and a wall thickness of 1.65 mm (0.065 in.) for an annular gap of 1.5 mm. Based on the residence time values achieved previously in smaller units, it was anticipated that this separator would be able to process in excess of 250 mL/min at maximum capacity, and with an applied pressure of 15 psi, this unit was able to fully separate heptane and water at a total flow rate of 316 mL/min, or approximately 19 L/hour. Based on the results obtained for lower density, 1.03 g/cm³ membranes, an estimated flow rate of over 1.5 L/min or 90 L/hour could be processed using this size of annular separator. Therefore, the demonstrated configuration would meet the throughput requirements of many continuous pharmaceutical manufacturing applications with a single compact unit.

For the membranes used in this work even a 2 mm annular gap gives a larger channel profile area in all cases than operating the membrane via inside to outside permeate flow, allowing for increased residence time per unit length and improving throughput. This trend continues to hold until a membrane I.D. of approximately 5 mm is used, by which point the separation will be outside surface-wetting dominated scales with negative impacts on separation efficiency for systems that do not spontaneously form segregated flows of slugs under prevailing conditions. Table 5. Example Reynolds number for equivalent area annular channels compared to inside to out flow of permeate configurations with tubular membranes.

Furthermore, Reynolds numbers for annular channels of the same area as the membranes used here are lower in all cases for equivalent flow rates (see Table 5), affording a larger process operating window where laminar flow is maintained. The increased range of flowrates that operate under laminar conditions using annular configurations suggests that there is scope for further scale up with annular Membrane No. 5 based configurations through increase in length of tubular section before limitations around the onset of turbulence would require numbering up of units. With a 3 mm gap size, where interfacial tension dominated separation to density-based separation transition is thought to occur, the 12.7 mm (0.5 inch) O.D., Membrane No. 5 annular flow configuration, has a flowrate of 2.96 L/min at transition to transitional flow regimes (Re=2,000), and 5.94 L/min (Re = 4,000) at the transition to fully turbulent flow regimes within the annulus, assuming the physical properties of water in both cases. Given that differential pressure control can be implemented across the membrane, onset of significant turbulent mixing within the separator should set the throughput limit for the proposed annular membrane separator configuration, and thus a limit in terms of tube length for a single unit. The modular nature of the parts consisting of tubes and standard tube fittings of 1.91 cm or 2.54 cm (0.75 or 1 inch) diameter used to construct such a configuration, further numbering up of units and utilization within standard heat exchanger configurations for temperature control is trivial, and relatively inexpensive.

As such, further scaling through increasing the diameter of the porous insert 101, longer tubes 102, and increasing the number of separators aligned together, up to tens of litres per minute should be possible for many liquid-liquid systems. This should be suitable to meet production scale requirements for continuous pharmaceutical synthesis applications in many cases. In principle, this approach may be sufficiently scalable to be utilized in batch pharmaceutical manufacturing with significant numbering up, where in-line two-phase liquid extraction with a static mixer followed by membrane-based extraction could improve productivity significantly by allowing a complete fill of vessels in batch production by eliminating the need to leave space for extractions solvent to be added post synthesis. Furthermore, given the rapid and complete separations that membrane separations can achieve, cycle time and yield of batch manufacturing could be positively impacted by their application if sufficient scale- up can be realized.

This invention relates to a new type of continuous phase separator using an annular gap to promote rapid phase separation of immiscible liquids. The separator utilizes a preferential wetting of the porous insert (101) in an enclosing non-permeable outer tube (102) creating a narrow annular channel or gap (104). Liquid phases enter this annular gap (104) to drive separation between immiscible or partially miscible liquid phases, that present two liquid phases (for example, oil and water, non-polar solvents and water, salt water and water-soluble polymers). The phase with the highest affinity for the porous insert (101) will preferentially contact with the inner surface (103) of the outer tube (102) where the separation will occur. Once in contact the liquid with the higher affinity will pass through the porous insert (101), with the lower affinity liquid unable to pass through the porous insert (101) as the operating pressure difference across the porous insert (101) is set so that it will be retained by the porous insert (101). In the examples given herein, the separation of water and organic solvents is allowed with a hydrophobic membrane (porous insert (101)), but the opposite is the case with a hydrophilic membrane (porous insert (101)).

The shortest achieved residence time of approximately 3 seconds for a heptane-water system equates to a flow rate of approximately 0.1 L/min in a separator less than 20 cm long, cheaply constructed from off-the-shelf parts. It is hoped the reduced cost combined with the ease of scale-up, compatibility with standard process equipment such as heat exchangers, wide degree of customisation, and the inherent robustness in operation of the unit will lower the barrier to integration significantly where a liquid-liquid phase separation step needs to be incorporated into a continuous manufacturing route. In principle, this approach may be sufficiently scalable to be utilized in-line in batch pharmaceutical manufacturing also through further scale-up and numbering up of units. The increase of surface area achieved by scaling up is done by increasing the length of the outer tube (102) or increasing the diameter of the porous insert (101) and the outer tube (102). The annular gap (104) allows the gap diameter to be kept constant as one goes up in scale.

In the specification the terms "comprise, comprises, comprised and comprising" or any variation thereof and the terms “include, includes, included and including" or any variation thereof are considered to be totally interchangeable and they should all be afforded the widest possible interpretation and vice versa.

The invention is not limited to the embodiments hereinbefore described but may be varied in both construction and detail.

Claims

1. A continuous liquid-liquid separator (100) comprising a porous insert (101) housed concentrically within a non-permeable outer tube (102), the outer tube (102) having an inner channel (105) and an inner surface (103), and wherein the configuration of the outer tube (102) and porous insert (101) form an annular gap (104) therebetween.

2. The continuous liquid-liquid separator (100) according to Claim 1, wherein the annular gap (104) is less than or equal to about 15 mm.

3. The continuous liquid-liquid separator (100) according to Claim 1 or Claim 2, wherein the annular gap (104) is less than or equal to about 10 mm.

4. The continuous liquid-liquid separator (100) according to any one of Claims 1 to 3, wherein the annular gap (104) is less than or equal to about 6 mm.

5. The continuous liquid-liquid separator (100) according to any one of Claims 1 to 4, wherein the annular gap (104) is less than or equal to about 5 mm.

6. The continuous liquid-liquid separator (100) according to any one of Claims 1 to 5, wherein the annular gap (104) is less than or equal to about 4 mm.

7. The continuous liquid-liquid separator (100) according to any one of Claims 1 to 6, wherein the annular gap (104) is less than or equal to about 3 mm.

8. The continuous liquid-liquid separator (100) of any one of the preceding claims, wherein the inner surface (103) of the outer tube (102) has a hydrophobicity opposite to that of the porous insert (101).

9. The continuous liquid-liquid separator (100) of any one of Claims 1 to 7, wherein the inner surface (103) of the outer tube (102) and the porous insert (101) have the same hydrophobic or hydrophilic properties.

10. The continuous liquid-liquid separator (100) according to any one of the preceding Claims, wherein the porous insert (101) is composed of a hydrophilic material selected from the group comprising a ceramic, alumina, glass, a thermoplastic polymer, a copolymer, or a thermoset polymer.

11. The continuous liquid-liquid separator (100) according to any one of the preceding claims, wherein the outer tube (102) is composed of a material selected from the group comprising stainless steel, PFA, glass, glass lined metal, an inert metal, a polymer, a ceramic, and silicon carbide.

12. The continuous liquid-liquid separator (100) according to any one of the preceding claims, wherein the porous insert (101) is configured to be tubular, flat, square or having triangular cross-sectional channels.

13. The continuous liquid-liquid separator (100) according to any one of the preceding claims, wherein the separator (100) is in the form of a linear configuration, a coiled configuration, a circular configuration, a curved configuration, or a combination of such configurations put together in a series, in parallel or counter current arrangements.

14. The continuous liquid-liquid separator (100) according to any one of the preceding claims, wherein the inner channel (105) accommodates a central membrane channel (106) within the porous insert (101).

15. The continuous liquid-liquid separator (100) according to Claim 12, wherein the central membrane channel (106) is a cylindrical rod.

16. The continuous liquid-liquid separator (100) according to any one of the preceding claims, wherein the porous insert (101) is a first porous membrane.

17. The continuous liquid-liquid separator (100) according to any one of the preceding claims, wherein the inner surface (103) of the outer tube (102) comprises a second porous membrane.