Porous membranes comprising sorbent particles for improved urea capture

Abstract

The present invention relates to a method for preparing a membrane comprising sorbent particles that bind urea. The invention also relates to the sorbent-comprising membranes per se, and to methods of using the membranes. The membranes are useful for undergoing subsequent reactions with small molecules such as urea, for instance to remove urea from a solution.

Claims

1. Membrane comprising embedded particulate material, wherein the membrane is a porous polymeric membrane, wherein the particulate material comprises a urea sorbent, and wherein the particle size of the particulate material is at most 250 μm along the largest diameter.

2. Membrane according to claim 1, wherein the particulate material is present in the membrane in a range of from 5 wt % to 80 wt %, based on the total dry weight of the membrane and the particulate material.

3. Membrane according to claim 1, wherein the particle size of the particulate material is at most 150 μm along the largest diameter.

4. Membrane according to claim 1, wherein the membrane is in the form of a hollow fibre, a full fibre, or a flat sheet.

5. Membrane according to claim 1, wherein the membrane comprises at least one polymer selected from polysulfone, polyethersulfone, polyphenylenesulfone, polyarylethersulfone, polyamide, polyetherimide, polyimide, polyethylene-co-vinyl alcohol, polyethylene-co-vinyl acetate, cellulose acetate, cellulose triacetate, polyvinylidene fluoride, polyvinylchloride, polyacrylonitrile, polyurethane, polyether ether ketone, and/or polyacrylic acid.

6. Membrane according to claim 1, wherein the membrane has a water permeability of at least 1 L/(m.sup.2.Math.h.Math.Bar).

7. Membrane according to claim 1, wherein the urea sorbent is a sorbent that covalently captures urea, wherein at least 10% of urea binding is covalent capture of urea.

8. Membrane according to claim 7, wherein the urea sorbent is a macromolecular composition comprising a polymeric backbone grafted with moieties that can covalently capture urea.

9. Membrane according to claim 1, wherein the urea sorbent is selected from a ninhydrin-type sorbent, a phenylglyoxaldehyde-type sorbent, and/or a triformylmethyl-type sorbent.

10. Membrane according to claim 1, wherein the particulate material further comprises activated carbon particles, ion exchange particles such as ion exchange resin particles or ion exchange silica particles, silica particles such as unmodified silica particles or alkylated silica particles, zeolite particles, ceramic particles, polymeric particles such as porous polymeric particles or non-porous polymeric particles, and/or molecularly imprinted particles, and/or wherein the membrane further comprises an additive such as a hydrophilic additive, preferably polyvinylpyrrolidone, chitosan, polyethylene glycol, dextran, glycerol, diethylene glycol, octanol, oxalic acid, maleic acid, tartaric acid, fumaric acid, lithium chloride, and/or calcium chloride.

11. (canceled)

12. Method for the preparation of a membrane as defined claim 1, comprising the steps of i. providing urea sorbent particles having a size of at most 250 μm along the largest diameter; ii. mixing the urea sorbent particles with polymeric material in a solvent for the polymeric material to obtain a mixture; iii. extruding or casting the mixture to form a membrane; and iv. optionally solidifying said membrane, preferably by phase inversion.

13. Method for removing nucleophilic waste solutes from a fluid, comprising the steps of: i) providing a fluid comprising nucleophilic waste solutes, and iia) contacting said fluid with a membrane as defined in claim 1, or alternately iib) contacting said fluid with a dialysis fluid through a membrane, wherein the dialysis fluid is in contact with a membrane as defined in claim 1, and iii) optionally, recovering the fluid.

14.-15. (canceled)

Description

DESCRIPTION OF DRAWINGS

[0151] FIG. 1—scanning electron microscopy images of the mixed matrix membrane, prepared as described in Example 2. FIG. 1A) cross-section image of the MMM corrugated lumen morphology of the membrane. FIG. 1B) magnified image of the wall of the MMM displaying a finger-like macrovoids structure. The sorbent particles are well dispersed, without aggregation and they are well surrounded by the polymer solution (white arrows). FIGS. 1C and 1D show, respectively, the lumen (inner layer) and the outer layer of the MMM which present very thin dense layers with no visible pores at the lumen surface (FIG. 1E) or at the outer surface (FIG. 1F). The outer dense layer (FIG. 1D) is slightly thicker (0.5 μm) compared to the inner dense layer (0.2 μm).

[0152] FIG. 2—water permeability of the MMM, prepared as described in Example 2, at varying pressure. Data expressed as mean±SE (n=6). Linear approximation yielded a water permeability equal to 238±9 L/(m.sup.2.Math.h.Math.Bar).

[0153] FIG. 3—kinetics of urea binding measured at 37° C., 50° C., and 70° C. using ninhydrin-type sorbent particles in suspension (containing about 2.5 mmol ninhydrin per g of sorbent), having an average diameter ±SD of 483±282 μm (n=30).

[0154] FIG. 4—urea binding results taken over time on ninhydrin-type sorbent particles in suspension (containing about 2.5 mmol/g ninhydrin) before (diameter ±standard deviation of 483±282 μm) and after (diameter <63 μm) grinding and sieving (n=3).

[0155] FIG. 5—urea binding results taken over time on ninhydrin-type sorbent (diameter <63 μm) in suspension, and on the MMM comprising such sorbent particles (average ±standard deviation).

[0156] FIG. 6—urea isotherm binding on the sorbent-comprising MMM, prepared as described in Example 2, and on a control hollow fibre (HF, PES/PVP), prepared as described in Example 2.2, without sorbent particles (n=3).

[0157] FIG. 7—dynamic urea binding results using the MMM comprising sorbent particles, prepared as described in Example 2. In this experiment, urea solution is continuously recirculated through the membrane for 4 hours (n=5). Urea binding at 4 hours (indicated with an asterisk) has been normalized with the amount of urea eluted from the MMM.

[0158] FIG. 8—kinetics of urea binding using the MMM comprising PGA-type sorbent particles, prepared in the form of a hollow fiber as described in Example 2. Here, urea solution is continuously recirculated through the MMM (n=2) or continuously stirred with ground sorbent particles (n=3). PGA-type sorbent in the membrane shows faster urea binding behavior than free PGA-type sorbent that was ground to obtain small particles.

EXAMPLES

Example 1—Provision of Sorbents

1.1 Provision of Ninhydrin-Type Sorbent

[0159] This type of sorbent is known in the art and can be prepared as for example described in EP121275A1, U.S. Pat. No. 4,897,200A, or WO2019110557.

1.2 Provision of PGA-Type Sorbent

[0160] This type of sorbent is known in the art and can be prepared as for example described in U.S. Pat. No. 3,933,753A or WO2004078797A1. Alternately, it can be produced as described below.

1.2.1 Provision of a precursor monomer

##STR00014##

[0161] In a 3-neck round bottom flask p-(ethynylphenyl)ethanone (10.0 g, 69.4 mmol) was suspended in EtOH (350 mL) and Lindlar's catalyst (300 mg, 3 w %) was added. Air was replaced by H.sub.2 and the suspension was stirred at RT for 2-16 hours. To monitor the conversion (and thus preventing over-reduction of VPE into the alkane), samples were frequently taken from the reaction mixture and, after evaporation of EtOH under reduced pressure, the conversion was determined by .sup.1H-NMR (CDCl.sub.3). After the conversion was >90%, the H.sub.2-filled balloon was removed and the reaction mixture was concentrated under reduced pressure. The crude product was re-dissolved in CH.sub.2Cl.sub.2 and purified by filtration over Hyflo. The filtrate was concentrated under reduced pressure, giving crude pVPE (p-(vinylphenyl)ethanone) as a yellow liquid in a 99% yield (10.1 g, 69.0 mmol). Melting point 29° C., melt enthalpy 90.6 J/g. .sup.1H-NMR (CDCl.sub.3, 600 MHz) δ 7.92 (d, J=8.3 Hz, 2H), 7.48 (d, J=8.2 Hz, 2H), 6.75 (dd, J=17.6 Hz, 10.9 Hz, 1H), 5.87 (d, J=17.6 Hz, 1H), 5.39 (d, J=10.9 Hz, 1H), 2.59 (s, 3H).

1.2.2 Polymerisation of Precursor Monomer

[0162] A two-phase suspension polymerization was used. In brief, the aqueous phase was prepared by addition of NaCl (11 mg), polymethacrylic acid sodium salt solution (452 mg of a 10% gel in water) and CaHPO.sub.4 (84 mg) to water (15 mL). The organic phase was composed of VPE (2.1 g, 14.4 mmol, 2 mL), porogen (2.9 mL, heptane/toluene), 80% technical grade divinylbenzene (3-6 mol %) and a 50% benzoylperoxide blend with dicyclohexyl phthalate (174 mg, 0.36 mmol, 2.5 mol %). After mixing and polymerization (heated at 73° C. in an oil bath for 16 hours under mechanical stirring), the resulting suspension was allowed to cool to RT and poured over a filter (cut-off 200 μm, Veco B.V.). The residue was washed with acetone and water, and finally dried over P.sub.2O.sub.5 under vacuum, resulting in pVPE (1.1-1.9 grams, yield 52-90%).

1.2.3 Preparation of PGA-Type Sorbent Based on Poly-VPE

[0163] The obtained crude pVPE mixtures were then further converted into PGA-type sorbents. The acetyl aromatic groups in pVPE beads were halogenated and subsequently converted into PGAH-groups by a Kornblum oxidation in a one-pot procedure. In a glass reactor equipped with a teflon blade stirrer, pVPE beads (60.0 g) were swollen in DMSO (600 mL, 8.45 mol) for 30 minutes under continuous stirring, after which an aqueous solution of 48% HBr (175 mL, 1.55 mol) was slowly added. One of the outlets of the reactor was capped with a septum containing a needle allowing escape of the formed Me.sub.2S. The suspension was stirred at 80° C. for 8 hours, after which the reaction mixture was filtered (cut-off 200 μm, Veco B.V.). The residue was washed with water until the pH of the filtrate was >5. The residue was dried over P.sub.2O.sub.5 under vacuum, resulting in PGA-type sorbent (55.2 grams).

Example 2—Preparation of Sorbent-Comprising Membranes

2.1 General Method for Sorbent-Comprising Membrane Preparation

[0164] A desired amount of polymers is weighed and thoroughly dissolved in ultrapure NMP. After stirring for three days, dry sorbent particles with a longest diameter of 120 μm are added, after which the suspension is agitated on a roller bank for four hours. The resulting suspension is then cast on water to form a sheet-like membrane, or extruded through a spinneret to form fibres.

2.2 Preparation of a PES/PVP Membrane Comprising Ninhydrin-Type Sorbent

[0165] Prior to membrane preparation, particles of ninhydrin-type sorbent or PGA type sorbent (e.g. about 2.5 mmol of ninhydrin per g of particles) having an average diameter ±standard deviation of 483±282 μm were ground using a mortar and pestle. Afterwards, the ground sorbent particles were sieved through a 63 μm sieve. A hollow-fibre (HF) Mixed Matrix Membrane (MMM) was prepared with sorbent particles embedded in a PES/PVP polymer matrix. The HF MMM was prepared via a dry-wet spinning technique. The polymer dope solution was prepared by dissolving Ultrason E6020 PES (BASF, Ludwigshafen, Germany) and PVP K90 (molecular weight ≈360 kDa, Sigma-Aldrich Chemie GmbH, Munchen, Germany) in ultrapure N-methylpyrrolidone (NMP) (Acros Organics, Geel, Belgium). The particles were added to the dope solution to have a final weight of particles equal to 55% of the dried weight of the membrane. The PES/PVP/sorbent polymer solution was stirred for two days at 60° C. to ensure proper dispersion of the particles in the polymer solution. Afterwards it was transferred in stainless-steel syringes and left to degas for 24 hours. The concentrations of PES, PVP and sorbent and the spinning parameters used in the study are specified in Table 1. After degassing, the syringe was connected to a high-pressure syringe pump and to a spinneret for preparing the HF (specifications in Table 1). Ultrapure water was used as bore forming solution. The air-gap between the spinneret and the coagulation bath was adjusted to 5.5 cm. The HF was left to free-fall in the water coagulation bath. The fabricated membrane was washed with demineralized-water and stored in demineralized-water for further use.

TABLE-US-00002 TABLE 1 Spinning parameters for the Mixed Matrix Membrane Dope composition 6.6/3.1/11.9/ (PES/PVP/Particles/NMP) 78.4 wt. % Dope pumping speed 1 mL/min Spinneret - Thickness dope orifice 0.6 mm Bore liquid Ultra-pure water Bore liquid pumping speed 1 mL/min Spinneret - Diameter bore needle 1.35 mm Air gap 5.5 cm Coagulation bath composition Ultra-pure water fibres collection Free falling

Example 3—Characterization of Sorbent-Comprising Membranes

3.1 Scanning Electron Microscopy (SEM)

[0166] The morphology of the HF MMM (prepared as described in Example 2.2) was analyzed by SEM (JEOL JSM-IT 100, Tokyo, Japan). Samples of the membrane were dried in air and fractured in liquid nitrogen for the imaging of the cross-sections. Prior to SEM imaging, the samples were gold sputtered (Cressington 108 auto sputter, (Cressington Scientific Instruments, Watford, UK).

[0167] A cross-section image of the MMM (FIG. 1A) shows corrugated lumen morphology of the membrane. Although the corrugated morphology might indicate a certain instability of the membrane forming system, the grooves are axially well-aligned with the flow direction. Such corrugation generally does not disturb mass transfer, flow rate and transmembrane pressure along the fibre. In addition, the presence of grooves in the lumen side of the fibre increases the active surface area, with advantages regarding filtration and diffusion performance of the membrane. Also, the thick wall of the membrane allows for more particles to be embedded in the polymer matrix per unit of length, thus enhancing capturing properties of the MMM.

[0168] From the magnification of the wall of the MMM (FIG. 1B), a finger-like macrovoids structure, typical of SEM membranes, is visible at the lumen and outer sides of the membrane, while it vanishes along the center of the wall cross-section, where more particles are hosted in the polymer matrix. The sorbent particles are well dispersed, without aggregation and they are well surrounded by the polymer solution (white arrows in FIG. 1B).

[0169] Both the lumen and the outer layer of the MMM (FIGS. 1C and 1D, respectively) present very thin dense layers with no visible pores at the lumen surface (FIG. 1E) or at the outer surface (FIG. 1F). However, the outer dense layer (FIG. 1D) is slightly thicker (0.5 μm) compared to the inner dense layer (0.2 μm). For this reason, it is believed that the selective layer of the membrane, determining the molecular-weight cut-off and the filtration properties of the membrane, is the outer dense layer.

3.2 Water Transport Experiment

[0170] The MMM was also characterized in terms of water transport properties. Membrane modules composed of 3 HF with a total surface area of 13.9 cm.sup.2 were used. A 2-component epoxy glue (Griffon Combi Snel-Rapide, Bison International, Goes, The Netherlands) was used for the preparation of the modules. Before water transport experiments, the HF modules (n=6) were pre-wetted with EtOH for 30 minutes at a trans-membrane pressure (TMP) of 0.2 Bar and pre-compacted with ultra-pure water at a TMP of 0.6 Bar for 30 minutes. Afterwards, the amount of permeated water was measured over time at TMP of 0.2, 0.4 and 0.6 Bar. The resulting water permeability was calculated as the slope of the linear fit of the flux (L/(m.sup.2.Math.h)) versus the TMP (Bar). The MMM is an ultrafiltration membrane with water permeability equal to 238±9 L/(m.sup.2.Math.h.Math.Bar) (FIG. 2). The ultrafiltration coefficient (K.sub.uf) of the MMM is 309 mL/(h-mmHg-m.sup.2). The water flux trough the membrane increases linearly over pressure, without compaction or breakage of the HF MMM. Overall, these results suggest that the MMM has excellent morphology characteristics and filtration properties for use in ultrafiltration.

Example 4—Capturing Urea Using Sorbent-Comprising Membranes

4.1 Static Capture of Urea

4.1.1 Effect of Temperature on Urea Binding Kinetics Using Ninhydrin-Type Sorbent

[0171] A urea kinetic binding experiment was performed at 37° C., 50° C. and 70° C. to study the effect of temperature on the binding reaction of urea with ninhydrin groups (about 2.5 mmol/g of sorbent) in ninhydrin particles prepared from crosslinked polystyrene (PS-Nin). The free particles used for this experiment have an average diameter ±standard deviation of 483±282 μm (N=30). The sorbent beads (15 mg) were incubated with a urea solution (1.5 mL, 30 mM) in PBS. The samples (n=3 for each time point) were placed in an oven on a rotating device at 37, 50 or 70° C. After 1, 2, 4, 8, 16 and 24 hours, the urea concentration in the supernatants was analyzed with an AU 5800 routine chemistry analyzer (Beckman Coulter, Brea, Calif.) using a coupled enzyme reaction, which results in a colorimetric (570 nm) product proportional to the urea concentration. Via the mass balance, the amount of urea bound was calculated from the depleted amount of urea in the solution. The results in FIG. 3 are shown as average ±standard deviation. At higher temperatures the urea binding kinetic is faster. Indeed, at 70° C. the particles bind 1.4±0.0 mmol/g of urea over a period of 24 hours, which is much higher compared to the binding at 50° C. and 37° C. Saturation of the ninhydrin moieties with urea is not reached at 24 hours even at 70° C. Here, the maximum achievable saturation was established to be 1.6-1.7 mmol/g for the particles. Given the faster binding kinetic at 70° C., later urea binding experiments were performed at 70° C.

4.1.2 Particle Size Effect on Urea Binding Kinetics

[0172] To study the effect of particle size on urea binding, a urea kinetic binding experiment was performed at 70° C. on PS-Nin particles (about 2.5 mmol/g ninhydrin) with average diameter of 483±282 μm and on the same particles after grinding and sieving (example 2.2) having a diameter of less than 63 μm. The bigger sorbent beads (15 mg) were incubated with urea solution (1.5 mL, 30 mM) in PBS (pH 7.4); the grinded and sieved particles were incubated in urea solution (1.5 mL, 30 mM) in dialysate (pH 7.4). The samples (n=3 for each time point) were shaken at 70° C. and after 1, 2, 4, 8, 16 and 24 hours the supernatants were collected via filtration. Urea concentration was determined by the enzymatic assay Urea FS* (Diasys, Holzheim, Germany) or by the AU 5800 routine chemistry analyzer (Beckman Coulter, Brea, Calif.). Both methods are based on a coupled enzyme reaction, which results in a colorimetric product proportional to the urea concentration. Via the mass balance, the amount of urea bound was calculated from the depleted amount of urea in the solution. The results in FIG. 4 are shown as average ±standard deviation. Urea binding kinetics are much faster on the smaller particles compared to the bigger particles. At 4 hours urea binding on the smaller particles is over double the binding of urea on the bigger particles. Urea binding kinetics on the small particles are very fast during the first 4 hours and then it slowly decreases. Over a period of 24 hours, binding of urea is equal to 1.7±0.1 mmol/g of urea on the smaller particles and 1.4±0.0 mmol/g on the bigger particles. Despite the higher surface area, maximum binding capacity (2.5 mmol/g) was not reached. Particle size thus has a strong effect on binding kinetics, but has a much smaller effect on binding capacity.

4.1.3 Urea Binding Kinetics for the MMM Comprising Sorbent Particles

[0173] To study the effect of the incorporation of the particles in the polymer matrix of MMM, a urea kinetic binding experiment was performed at 70° C. on ground PS-Nin particles (about 2.5 mmol/g ninhydrin) with a diameter of less than 63 μm and on the MMM prepared as described in example 2.2. The sorbent beads (15 mg) and the MMM (27 mg, containing 15 mg of embedded particles) were incubated with urea solution (1.5 mL, 30 mM) in dialysate (pH 7.4) and shaken at 70° C. After 1, 2, 4, 8, 16 and 24 hours, the supernatants were collected (via filtration for the particles) and the urea concentration in the supernatants was determined using an enzyme assay (Urea FS*, Diasys, Holzheim, Germany). Via the mass balance, the amount of urea bound was calculated from the depleted amount of urea in the solution. Each timepoint was represented by three different samples (n=3) (FIG. 5). The results in FIG. 5 are shown as average ±standard deviation. Urea binding kinetics are very similar for the particles in suspension and for the particles in the MMM over a period of 8 hours. Interestingly, at 8 hours urea binding the particles in suspension reach a plateau, while the particles embedded in the MMM keep binding. At 24 hours, urea binding to the particles in suspension is equal to 1.7±0.1 mmol/g, while on the particles embedded in the MMM it is 2.1±0.1 mmol/g. After a longer period of time urea binding to the particles in the MMM can reach saturation (maximum binding capacity is about 2.5 mmol/g). This demonstrates how the polymer matrix does not limit the ability of urea to reach the particles; apparently binding is even enhanced compared to free sorbent particles.

[0174] Further improved urea binding was demonstrated using PGA-type sorbent in a hollow fiber MMM. In a comparative experiment, an amount of MMM comprising 88 mg of a PGA-type sorbent was contacted with a PBS buffer comprising 30 mM urea, using a recirculating setup with a flow rate of 5 mL/min (total volume 25 mL) at a temperature of 70° C. Urea binding was determined at various time points, as shown in FIG. 8 (n=2; yet n=1 for t=6-8 h). As comparison, a static batch experiment was performed using 5 mg of the same PGA-type sorbent, in a total volume of 1 mL of the same urea solution that was continuously stirred at 70° C., and urea binding was determined at various time points (n=3). The results in FIG. 8 demonstrate that the binding kinetics of the PGA-type sorbent in the MMM is faster than the kinetics of ground sorbent under agitation.

4.1.4 Urea Isotherm Binding to the MMM and to a Control Membrane

[0175] Urea binding at various concentrations was measured with the MMM (prepared as described in Example 2.2) and with a PES/PVP control HF. PES/PVP control HF has been prepared via dry-wet spinning technique as described in Acta Biomaterialia 90 (2019) 100-111. Briefly, the polymer dope solution was prepared by dissolving Ultrason E6020 PES (15 wt %) (BASF, Ludwigshafen, Germany) and PVP K90 (7 wt %) (molecular weight ≈360 kDa, Sigma-Aldrich Chemie GmbH, Munchen, Germany) in ultrapure NMP (78 wt %) (Acros Organics, Geel, Belgium). The polymer solution was mixed on a roller bench for 3 days, then it was transferred into a stainless-steel syringe and left to degas for 24 hours. Afterwards, the syringe was connected to a high-pressure syringe pump and to a designed spinneret. The dope solution pumping speed was set at 0.4 mL/min. Ultrapure water was used as bore forming solution and the bore solution pumping speed was set at 1.2 mL/min. The air-gap between the spinneret and the coagulation bath was adjusted to 10 cm. A collecting wheel (speed 8.3 m/min) was used for the collection of the produced HF. The fabricated membrane was washed with demineralized-water and stored for further use.

[0176] 27 mg of the MMM (15 mg of particles in the MMM) and 27 mg of the control HF were incubated with 1.5 mL of urea solution in dialysate (pH 7.4) at different concentrations. The samples were placed in a horizontally shaking water bath at 70° C. After 24 hours, the supernatants were collected and the urea concentration in the supernatants was determined using an enzyme assay (Urea FS*, Diasys, Holzheim, Germany). Via the mass balance, the amount of urea bound was calculated from the depleted amount of urea in the solution. The results in FIG. 6 are shown as average ±standard deviation (N=3). The graph presents binding capacity (mmol/g) versus equilibrium urea concentration (mM). The PES/PVP control HF does not bind urea, suggesting that only the sorbent particles inside the MMM are responsible for urea binding.

[0177] Interestingly, at an equilibrium concentration of 90.1 mmol/g, urea binding to the MMM is equal to 3.4±0.1 mmol/g, surpassing expected values for 1-on-1 binding to ninhydrin. This higher binding of urea might be explained via multi-layered binding of urea, where urea molecules adsorb via hydrogen bonding (physisorption) on urea covalently bound (chemisorption) to the ninhydrin moieties of the particles.

4.1.5 Dynamic Binding of Urea

[0178] Membrane modules composed of 3 HF (prepared as described in Example 2) with a total surface area of 23.3 cm.sup.2 and with a total amount of particles embedded in the PES/PVP matrix equal to 77 mg were used to study urea binding in dynamic conditions. A 2-component epoxy glue (Griffon Combi Snel-Rapide, Bison International, Goes, The Netherlands) was used for the preparation of the modules. Before dynamic urea binding experiments, the HF modules (n=5) were kept in demi-water. Urea dynamic experiments were performed in filtration mode (TMP=0.15 Bar) with urea solution 30 mM in dialysate continuously recirculated through the fibres at a flow rate of 20 mL/min using a dedicated setup (Convergence, Enschede, The Netherlands). The binding experiments (n=5) were performed at 70° C. for 4 hours. To maintain a temperature of the urea solution of 70° C. inside the recirculation system, the feed solution was heated at 70° C., the tubing was insulated and the HF module was immerged in a water bath heated at 90° C. Samples of the urea solution were collected every hour for quantification. At the end of the binding experiment, the HF module was removed from the 90° C. water bath and was emptied. 24.5 mL of MilliQ water at room temperature were recirculated through the module at a flow rate of 20 mL/min for 1 hour to collect possibly eluted urea from the MMM. The urea concentration was quantified using an enzymatic assay (Urea FS*, Diasys, Holzheim, Germany).

[0179] Binding results are shown in FIG. 7. Over a period of 4 hours, the membrane is able to bind 3.4±0.3 mmol/g and does not appear to reach saturation. When a binding experiment is performed in dynamic mode, the binding kinetic is much faster compared to static conditions. Not only is the kinetic faster, but also the total amount of urea bound to the sorbent particles is higher compared to the static condition. The amount of urea bound per gram of particles in the MMM is higher than the total amount of ninhydrin moieties in the particles (about 2.5 mmol/g). It is believed that 1) all ninhydrin moieties in the particles are saturated and 2) that urea molecules adsorb via hydrogen bonding (physisorption) on urea covalently bound (chemisorption) to the ninhydrin moieties of the particles, thus having multi-layered binding which is possibly enabled by the formed surface saturated with covalently bound urea.

[0180] When the MMM modules (n=2) were rinsed with MilliQ water at the end of the experiment, 0.05 millimoles were detected after 1 hour. Urea binding at 4 hours normalized with the amount of urea eluted from the MMM is indicated with an asterisk in FIG. 7. It was expected that, for a longer rinsing period, all urea physisorbed in the system would have been eluted out. This experiment confirms that in dynamic conditions urea binding to the MMM is a combination of chemisorption and physisorption. Because urea did not bind to control membranes with no sorbent particles, the combined binding is an effect of the membranes according to the invention.

[0181] In conclusion, the recirculation and filtration of urea solution through the membrane improves binding kinetic and total binding. Moreover, when maximum chemisorption capacity is reached, adsorption of urea can continue, possibly due to the hydrogen bonding of urea molecules to the newly formed surface of covalently bound urea.