Microporous polyvinylidene fluoride membrane

Abstract

Hydrophobic hollow-fiber membrane made from a vinylidene fluoride polymer with a wall and a wall thickness, an outer surface on its outer side, an inner surface on its inner side and facing its lumen and adjacent to the inner surface a supporting layer having a structure that is substantially isotropic across the wall thickness, the supporting layer extending over at least 80% of the wall thickness and comprising pores having an average diameter of less than 1 m, and wherein the hollow-fiber membrane has pores on its outer surface and on its inner surface, characterized in that the vinylidene fluoride polymer has a weight-average molecular weight M.sub.W in the range from 550 000 to 700 000 daltons and a polydispersivity greater than 3.0; the pores in the outer and in the inner surface are formed like islands and have a maximum ratio of their longitudinal extension to the transverse extension of 10; the porosity lies in the range from 50 to 90 vol. %, the wall thickness in the range from 50 to 300 m, and the diameter of the lumen in the range from 100 to 500 m; and the hollow-fiber membrane has a maximum separating pore diameter d.sub.max in the range from 0.3 to 0.7 m, determined according to the bubble point method.

Claims

1. Hydrophobic hollow-fiber membrane made from a vinylidene fluoride polymer, wherein the hollow-fiber membrane has a wall with a wall thickness, an outer surface on its outer side and an inner surface on its inner side and facing its lumen, wherein the hollow-fiber membrane has a continuous skin on its inner surface and a continuous skin on its outer surface, wherein pores are formed in the skin of the inner surface and in the skin of the outer surface, and wherein the hollow-fiber membrane, adjacent to the skin of the inner surface, has a supporting layer having a microporous, sponge-like, open-pored structure that is substantially isotropic across the wall thickness, the supporting layer extending over at least 80% of the wall thickness and comprising pores having an average diameter of less than 1 m, and wherein the vinylidene fluoride polymer forming the structure of the hollow-fiber membrane has a weight-average molecular weight Mw in the range from 550 000 to 700 000 daltons and a polydispersivity, given by the ratio of the weight average M.sub.W and the number average M.sub.N of the molecular weight, from 3 to 7, the pores in the skin of the outer surface and the pores in the skin of the inner surface have a closed perimeter in a plane of the skin, and the pores in the skin of the outer or the inner surface have a ratio of their longitudinal extension in the direction of the longitudinal axis of the hollow-fiber membrane to their transverse extension in the circumferential direction of the hollow-fiber membrane of a maximum of 10, the porosity of the hollow-fiber membrane lies in the range from 50 to 90 vol. %, the wall thickness in the range from 50 to 300 m, and the diameter of the lumen in the range from 100 to 500 m, a transmembrane flow for isopropyl alcohol in the range from 2 to 8 ml/(cm.sup.2.Math.min.Math.bar), the hollow-fiber membrane has a maximum separating pore diameter d.sub.max in the range from 0.3 to 0.7 m, determined according to a bubble point test method DE-A-36 17 724; and the pores in the skin of the inner surface are smaller than the pores of the skin of the outer surface and the pores in the skin of the outer surface are larger than the pores in the supporting layer.

2. Hollow-fiber membrane according to claim 1 comprising a transmembrane flow for water vapor of at least 35 l/(m.sup.2.Math.h), determined by means of a module of the hollow-fiber membrane with a membrane area of 40 cm.sup.2 at a salt water circuit temperature of 80 C. and a distillate circuit temperature of 30 C., a volume flow in the circuits of 200 l/h, a pressure level in the circuits of 500 mbar at the inlet to the hollow-fiber membrane module, and a salt concentration in the salt circuit of 36 g/l.

3. Hollow-fiber membrane according to claim 1 wherein the hollow fiber membrane has an elongation at break of at least 50%.

4. Hollow-fiber membrane according to claim 1 wherein the hollow fiber membrane has a breaking strength of at least 400 cN/mm.sup.2.

5. Hollow-fiber membrane according to claim 1 wherein the hollow fiber membrane has a volume porosity in the range from 70 to 85 vol. %.

6. Hollow-fiber membrane according to claim 1 wherein the hollow fiber membrane has a maximum pore tortuosity of 1.5.

Description

(1) The invention is to be explained in more detail on the basis of the following examples and figures. The content of the figures is as follows:

(2) FIG. 1: Scanning electron microscopic (SEM) image of a cross section of the membrane according to Example 1 at 500 magnification.

(3) FIG. 2: SEM image of a portion of the cross section of the membrane according to Example 1 at 2000 magnification.

(4) FIG. 3: SEM image of the outer surface of the membrane according to Example 1 at 100 magnification.

(5) FIG. 4: SEM image of the outer surface of the membrane according to Example 1 at 500 magnification.

(6) FIG. 5: SEM image of the inner surface of the membrane according to Example 1 at 500 magnification.

(7) FIG. 6: SEM image of the inner surface of the membrane according to Example 1 at 2000 magnification.

(8) FIG. 7: SEM image of a cross section of the membrane according to Example 2 at 500 magnification.

(9) FIG. 8: SEM image of a portion of the cross section of the membrane according to Example 2 at 2000 magnification.

(10) FIG. 9: SEM image of the outer surface of the membrane according to Example 2 at 100 magnification.

(11) FIG. 10: SEM image of the outer surface of the membrane according to Example 2 at 500 magnification.

(12) FIG. 11: SEM image of the inner surface of the membrane according to Example 2 at 500 magnification.

(13) FIG. 12: SEM image of the inner surface of the membrane according to Example 2 at 2000 magnification.

(14) To determine the properties of the inventive hollow-fiber membrane, the following methods were used:

(15) Maximum Separating Pore:

(16) The diameter of the maximum separating pore is determined by means of the bubble point method (ASTM No. 128-99 and F 316-03), for which the method described in DE-A-36 17 724 is suitable. Thereby, d.sub.max results from the vapor pressure P.sub.B associated with the bubble point according to the equation
d.sub.max=.sub.B/P.sub.B

(17) where .sub.B is a constant that is primarily dependent on the wetting liquid used during the measurement. For IPA, .sub.B is 0.61 m.Math.bar at 25 C.

(18) Transmembrane Flow for Isopropyl Alcohol (Permeability for IPA):

(19) The hollow-fiber membranes to be tested are used to produce a test cell with a defined hollow-fiber quantity and length. Both ends of the hollow fibers are embedded in a polyurethane resin for this. After curing of the resin, the embeddings are cut to a length of approx. 30 mm, wherein the lumina of the hollow-fiber membranes are opened by the cut. The hollow-fiber lumina in the embeddings must be verified as open. The free length of the hollow-fiber membranes between the embeddings is usually 180+/10 mm. The number of hollow-fiber membranes is to be selected such that, taking into account the free length and the inside diameter of the hollow-fiber membranes, a filtration surface of approximately 20 cm.sup.2 in the test cell is provided.

(20) The test cell is incorporated in a testing apparatus and subjected to a flow of isopropyl alcohol (IPA) ultrapure, conditioned to 25 C. at a defined test pressure (approx. 0.2 bar). The filtered IPA volume obtained during a measuring time of 2 min, i.e. the permeate generated during measurement, is determined gravimetrically or volumetrically. Before measurement is begun, the system must be purged of air. In order to determine the TMF, the input and output pressure is measured on the test cell in the testing apparatus. The measurement is performed at 25 C.

(21) The transmembrane flow, TMF, is determined according to formula (III)

(22) $\begin{array}{cc}TMF=\frac{\mathrm{Vw}}{t.Math.A_M.Math.p}\left[\frac{\mathrm{ml}}{{\mathrm{cm}}^2.Math.\min .Math.\mathrm{bar}}\right]& \left(\mathrm{III}\right)\end{array}$

(23) where:

(24) V.sub.W=volume of IPA [ml] flowing through the membrane sample during the measuring period

(25) t=measuring time [min]

(26) A.sub.M=area of the membrane sample penetrated (normally 20 cm.sup.2)

(27) =pressure set during the measurement [bar]

(28) Transmembrane Flow for Water Vapor:

(29) The measurement of the transmembrane flow for water vapor is performed on a hollow-fiber membrane module with a test surface of 40 cm.sup.2. Two liquid circuits are connected to this hollow-fiber membrane module, wherein the feed stream (salt water circuit) is connected to the hollow-fiber membrane module in such a way that it flows through the hollow-fiber membranes on the lumen side. The distillate circuit absorbing the permeate flows through the hollow-fiber membrane module in the extracapillary area. The starting volumes of the circuits were each 1.6 l. During measurement, the distillate circuit continuously increases as a result of the added permeate. The salt water circuit has a salt content of 360.5 g/l, which is kept constant by conductivity measurement by adding deionized water.

(30) The volume flow in both circuits is set to 200 l/h5%, wherein the circuits flow counter to the processing direction. The distillate circuit is conditioned to a temperature of 30 C. and the salt water circuit to a temperature of 80 C. Both circuits are kept at the same pressure level, with a pressure of 500 mbar set at the respective inlet to the hollow-fiber membrane module.

(31) In order to determine the transmembrane flow for water vapor, the increase in weight in the distillate circuit is determined gravimetrically over time. The minimum measurement time is at least 15 min.

(32) The transmembrane flow for water vapor in l/(m.sup.2.Math.h) is determined as the increase in weight, or the resulting increase in volume per time unit, referred to the current membrane surface (inner surface) of the hollow-fiber membrane module used.

(33) Force and Elongation at Break:

(34) Measuring the breaking force of the hollow-fiber membranes takes place using a standard, universal testing machine from Zwick in Ulm.

(35) The hollow-fiber membrane sample is stretched at a constant speed in the longitudinal direction until it breaks. The force required for this is measured as a function of the change in length and retained in a force/elongation curve. The measurement takes place as multiple determinations on multiple hollow-fiber membrane samples at 100 mm clamping length and at a traction speed of 500 mm/min. The pretension weight is 2.5 cN. The force BK required for break is given as an average numeric value in cN.

(36) The breaking strength .sub.B of the hollow-fiber membrane sample is obtained by normalizing the breaking force BK to the cross-sectional area A.sub.Q of the membrane wall.

(37) Molecular Weight, Polydispersivity:

(38) The determination of the molecular weight and the mole mass distribution (polydispersivity) takes place using gel permeation chromatography (GPC; columns: PSS GRAM: 10 m, G, 30, 100, 3000 ) on polystyrene standards with N-methyl-2-pyrrolidone (NMP)/0.1M LiCl as the eluent, and at a flow rate of 1 ml/min. The sample concentration is 3 mg/ml, and the injection volume is 100 l (injection system TSP AS 3000). The oven temperature is set to 70 C., and the detection takes place with the Shodex RI 71 differential refractometer. The number average M.sub.N and the weight average M.sub.W of the molar mass distribution are determined from the molar mass distribution using conventional methods. The dispersivity results from the ratio of the weight average M.sub.W to the number average M.sub.N, thus M.sub.W/M.sub.N.

(39) Pore Size in the Surfaces:

(40) The determination of the average diameter of the pores in the surfaces takes place using image analysis methods based on scanning electron microscope images of the surfaces at 500 magnification (outer surface) or 2000 magnification (inner surface). The scanning electron microscope images of the surfaces were also used to assess the ratio of the longitudinal extension of the pores to the transverse extension thereof.

(41) Volume Porosity:

(42) Samples of at least 0.5 g of the membrane to be examined are dry weighed. In the case of hollow-fiber membranes, 5 samples of the hollow-fiber membrane, each with a length of approximately 20 cm, can be used. The membrane samples are first wetted twice with a liquid that wets but does not cause swelling of the membrane material for 10 min, wherein for hollow-fiber membranes the liquid is also injected into the lumen of the hollow-fiber membranes using a syringe. For the present PVDF membranes, a silicone oil with a viscosity of 200 mPa s at 25 C. (Merck) is used. The samples are subsequently placed in the liquid for 24 hours such that the liquid penetrates into all pores. This is visually discernible in that the membrane samples change from an opaque to a glassy, transparent state. The membrane samples are subsequently removed from the liquid. Liquid adhering to the membrane samples is removed by centrifuging at approx. 1800 g and carefully blown out of the lumen of the hollow-fiber membranes with a weak air stream. The mass of the thus pretreated wet membrane samples, i.e. having liquid-filled pores, is subsequently determined by weighing.

(43) The volume porosity is determined according to the following formula:

(44) $\mathrm{Volume}\mathrm{porosity}\mathrm{.Math.}=\frac{\left(m_{\mathrm{wet}}-m_{\mathrm{dry}}\right)/{}_{\mathrm{liquid}}}{\left(m_{\mathrm{wet}}-m_{\mathrm{dry}}\right)/{}_{\mathrm{liquid}}+m_{\mathrm{dry}}/{}_{\mathrm{polymer}}}$

(45) where: m.sub.dry=weight of the dry membrane sample after wetting and drying [g] m.sub.wet=weight of the wet, liquid-filled membrane sample [g] .sub.liquid=density of the liquid used [g/cm.sup.3] .sub.polymer=density of the membrane polymer [g/cm.sup.3]

(46) Average Pore Radius (r.sub.P)

(47) The determination of the average pore radius of the membranes takes place via the permporometry commonly used for membranes with microporous structure, as described in ASTM F 316-03. For this purpose, the pores of a porous membrane sample are first filled with a liquid of a known interfacial tension. The membrane sample is subsequently impinged with a gas on one side, the pressure of which is increased in stages. With increasing pressure, the liquid is forced out of the pores until the gas above the opening pressure begins to flow through the pores of the membrane sample. The opening pressure P for a pore with a defined size depends on the surface tension of the liquid and the pore radius according to Laplace's equation:

(48) $P=\frac{2.Math..Math.\cos }{r_p},$

(49) where

(50) =Surface tension of the wetting liquid

(51) =Contact angle of the liquid

(52) r.sub.p=Pore radius

(53) During measurement, the gas flow through a wetted sample is continuously measured as a function of the applied measuring pressure. Increasing the gas pressure in stages first opens the largest pore and then the smaller pores until all pores in the sample are dry. The total gas flow is determined continuously in the process. In this way, the wet curve is obtained. The measurement is subsequently repeated on the dry sample to obtain the dry curve as a reference curve. Comparing the gas flow values of the wet and dry curves determines the pore size distribution curve.

(54) The average pore radius is determined from the test series for flow measurements, which is performed using a permporometer (capillary flow porometer, PMI, Porous Materials Inc.). A chlorofluorocarbon (Porewick) having a surface tension of 16 mN/m serves as the wetting reagent. The ends of 6 20 cm long hollow-fiber membrane samples are glued air tight into the holes of a holder such that the lumina of the hollow-fiber samples remain open and can be perfused with gas. After hardening of the glue, the hollow fibers are cut flush where they exit the holder. The free measuring length of the sample pieces is 3 cm in each case. The thus prepared samples are installed in the test chamber of the permporometer with the holder.

(55) The determination of the average pore radius r.sub.P takes place according to the method specified in ASTM F 316-03.

(56) Tortuosity:

(57) The tortuosity of the porous capillary membranes is determined via the gas permeability method according to M. Khayet et al., Polymer, 2002, 43, 3879-3890, Elsevier.

(58) The gas flows for the porous membranes can be described by the pore flow model. Taking into account contributions from diffusion and convection, the permeability (B) of a porous membrane is determined as a function of the measuring pressure, as represented in the following formula (1):

(59) $\begin{array}{cc}B=\frac{3}{4}{\left(\frac{2}{\mathrm{MRT}}\right)}^{0.5}.Math.\frac{r_p\mathrm{.Math.}}{L_p}+\frac{P_m}{8\mathrm{RT}}.Math.\frac{r_p^2\mathrm{.Math.}}{L_p}=I_0+S_0.Math.P_m& \left(1\right)\end{array}$

(60) where B=Permeability of the measuring gas through the membrane in mol/(s.Math.m.sup.2.Math.Pa) M=Molecular weight of the measuring gas in (kg/mol) R=Gas constant =8.314 J/(mol.Math.K) T=Absolute temperature in (K) r.sub.p=Pore radius of the membrane in (m) =Porosity of the membrane as a fractional amount L.sub.p=Effective pore length of the membrane in (m) P.sub.m=Average measuring pressure in the membrane pores in (Pa) =Dynamic viscosity of the measuring gas in (Pa.Math.s) /L.sub.p=Effective porosity of the membrane in (m.sup.1)

(61) If, for different transmembrane pressures, the measured gas permeability B is plotted against the measuring pressures P.sub.m, a linear correlation results. The slope S.sub.0 and intercept I.sub.0 on the B axis can be calculated from the curve. In this way, the effective porosity /L.sub.p can be represented according to the following formula:

(62) $\begin{array}{cc}\frac{\mathrm{.Math.}}{L_p}=\frac{8\mathrm{RT}}{r_p^2}.Math.S_0& \left(2\right)\end{array}$

(63) The measurement of the gas permeability takes place according to the previously described measuring method for pore size distribution and average pore radius and is realized on dry membrane samples by means of a permporometer (capillary flow porometer, PMI). The membrane samples are inserted in the intended holders. The permeability flows of nitrogen through the membrane samples are measured at different transmembrane pressures at room temperature. With hollow-fiber membranes, nitrogen is applied to the lumen side.

(64) With the porosity determined for membrane samples as described previously and the average pore radius r.sub.P determined via the previously described permporometry, the effective porosity /.sub.P can be calculated. For the dynamic viscosity of the measuring gas nitrogen, a value of 17.84 Pa.Math.s is taken as a basis.

(65) The tortuosity can be calculated as =L.sub.P/L.sub.min, wherein the thickness of the membrane, which corresponds to the minimum pore length L.sub.min, is incorporated into the calculation.

EXAMPLE 1

(66) A mixture of PVDF granules of type Hylar 461 and Solef 6020 from Solvay Solexis (mixture ratio 50:50) was melted in an extruder at 235-245 C. The polymer melt was mixed in a mixer with a solvent system consisting of 40 wt. % glyceryl triacetate (component A) and 60 wt. % dioctyl adipate (component B) at 230-245 C. and subsequently processed to form a homogeneous solution. The polymer proportion was set at 26.5 wt. %.

(67) This solution was fed into a hollow-fiber die conditioned to 210 C. and extruded above the phase separation temperature into a hollow fiber. Nitrogen was used as the inner filling. After passing through an air gap, the hollow fiber was fed through an approximately 2 m long spinning tube, which was perfused with a cooling medium conditioned to room temperature. A mixture of glyceryl triacetate and dioctyl adipate in a ratio of 65:35 was used as the cooling medium.

(68) The hollow fiber, solidified as a result of the cooling in the spinning tube, was drawn at a speed of 75 m/min from the spinning tube, wound on a drum, subsequently first extracted using isopropyl alcohol heated to approx. 60 C., and then dried online in a convection oven at approx. 125 C.

(69) The hollow-fiber membranes produced in this way had an outside diameter of 608 m and a wall thickness of 158 m. The transmembrane flow for isopropyl alcohol was 3.24 ml/(cm.sup.2.Math.min.Math.bar). The breaking strength of the hollow-fiber membrane was 522 cN/mm.sup.2, and the elongation at break was 80%. The membrane had a bubble point of 1.31 bar, determined by means of the bubble point method with isopropyl alcohol, corresponding to a maximum separating pore of 0.47 m, and had a transmembrane flow for water vapor of 40 l/(m.sup.2.Math.h). The porosity of the hollow-fiber membrane was 78 vol. %, and the average pore diameter was 0.247 m. The PVDF polymer component forming the membrane structure had a weight-average M.sub.W of the molar mass distribution of 61 800 daltons and a polydispersivity M.sub.W/M.sub.N of 4.43.

(70) With the previously listed data, a tortuosity of 1.43 results for the membrane of this example.

(71) As proven by the scanning electronic microscopic (SEM) examination of the fracture plane of the hollow-fiber membrane, this hollow-fiber membrane had a very finely pored structure across its wall (FIG. 1). The SEM image of the entire cross section of the membrane wall at 500 magnification clearly shows a microporous supporting layer, free of finger pores, extending across approximately 85% of the cross-section, with a sponge-like, open-pored pore structure that is substantially isotropic across the wall thickness, wherein the pores in this supporting layer have on average a size less than 1 m (FIG. 2).

(72) SEM images of the outer surface of the membrane at 100 magnification show a uniform and relatively even structure of the surface, which has island-shaped pores with a slightly elongated shape (FIGS. 3, 4). A comparison of FIG. 2 with FIG. 4 shows that the pores of the hollow-fiber membrane in the outer surface are larger than the pores in the area of the supporting layer with an isotropic pore structure. In comparison with the outer surface, the pores in the inner surface are significantly smaller (FIGS. 5, 6).

EXAMPLE 2

(73) The procedure was the same as in Example 1. Deviating from Example 1, the polymer proportion was set at 26.3 wt. %. Nitrogen was used as the inner filling.

(74) The hollow-fiber membrane had an outside diameter of 654 m and a wall thickness of 141 m. The transmembrane flow for isopropyl alcohol was 5.87 ml/(cm.sup.2.Math.min.Math.bar). The breaking strength of the hollow-fiber membrane was 471 N/mm.sup.2, and the elongation at break was 97%. The membrane had a bubble point of 1.31 bar, determined by means of the bubble point method with isopropyl alcohol, corresponding to a maximum separating pore of 0.47 m, and had a transmembrane flow for water vapor of 56 l/(m.sup.2.Math.h). The porosity of the hollow-fiber membrane was 81 vol. %, and the average pore diameter was 0.274 m. The PVDF polymer component forming the membrane structure had, as in Example 1, a weight-average M.sub.W of the molar mass distribution of 61 800 daltons and a polydispersivity M.sub.W/M.sub.N of 4.43.

(75) With the previously listed data, a tortuosity of 1.27 results for the membrane of this example.

(76) As proven by the scanning electronic microscopic (SEM) examination of the fracture plane of the hollow-fiber membrane, this hollow-fiber membrane had a very finely pored structure across its wall (FIG. 7). The SEM image of the entire cross section of the membrane wall at 500 magnification clearly shows a microporous supporting layer, free of finger pores, extending across approximately 85% of the cross-section, with a sponge-like, open-pored pore structure that is substantially isotropic across the wall thickness, wherein the pores in this supporting layer have on average a size less than 1 m (FIG. 8).

(77) SEM images of the outer surface of the membrane at 100 magnification show a uniform and relatively even structure of the surface, which has island-shaped pores with a slightly elongated shape (FIGS. 9, 10). A comparison of FIG. 7 with FIG. 10 shows that the pores of the hollow-fiber membrane in the outer surface are larger than the pores in the area of the supporting layer with an isotropic pore structure. In comparison with the outer surface, the pores in the inner surface are significantly smaller (FIGS. 11, 12).

COMPARATIVE EXAMPLE 1

(78) The procedure was the same as in Example 1. Deviating from Example 1, the PVDF granules of type Solef 1015 from Solvay Solexis with a weight-average M.sub.W of the molar mass distribution of 513 000 daltons and a polydispersivity M.sub.W/M.sub.N of 5.52 were used as the polymer component. The polymer melt was mixed in a mixer with a solvent system consisting of 35 wt. % glyceryl triacetate (component A) and 65 wt. % dioctyl adipate (component B) at 230-245 C. and subsequently processed to form a homogeneous solution. The polymer proportion was set at 27.1 wt. %. Nitrogen was used as the inner filling. The polymer solution was spun at 235 C.

(79) The resulting hollow-fiber membrane had an outside diameter of 619 m and a wall thickness of 136 m. The transmembrane flow for isopropyl alcohol was 1.70 ml/(cm.sup.2.Math.min.Math.bar). The breaking strength of the hollow-fiber membrane was 358 cN/mm.sup.2, and the elongation at break was 45%. The membrane had a bubble point of 1.10 bar, determined by means of the bubble point method with isopropyl alcohol, corresponding to a maximum separating pore of 0.55 m, and had a transmembrane flow for water vapor of 32 l/(m.sup.2.Math.h). The porosity of the hollow-fiber membrane was 81 vol. %, and the average pore diameter was 0.199 m.

(80) With the previously listed data, a tortuosity of 1.62 results for the membrane of this comparative example.

COMPARATIVE EXAMPLE 2

(81) The procedure was the same as in Comparative example 1. Deviating from Comparative example 1, a solvent system consisting of 26 wt. % glyceryl triacetate, 67.5 wt. % dioctyl adipate, and 6.5 wt. % -caprolactam was used, with which the polymer component was mixed and subsequently a homogenous solution produced at 230-245 C. The polymer proportion was set at 30.6 wt. %. Nitrogen served as the inner filling. The polymer solution was spun at 200 C. A mixture of glyceryl triacetate and castor oil in a ratio of 70:30 was used as the cooling medium.

(82) The obtained hollow-fiber membrane had an outside diameter of 560 m and a wall thickness of 133 m. The transmembrane flow for isopropyl alcohol was 2.62 ml/(cm.sup.2.Math.min.Math.bar). The breaking strength of the hollow-fiber membrane was 415 cN/mm.sup.2, the elongation at break was 45.3%. The membrane had a bubble point of 1.30 bar, determined by means of the bubble point method with isopropyl alcohol, corresponding to a maximum separating pore of 0.47 m, and had a transmembrane flow for water vapor of 30 l/(m.sup.2.Math.h).

COMPARATIVE EXAMPLE 3

(83) The procedure was the same as in Comparative example 1. Deviating from Comparative example 1, a solvent system consisting of glyceryl triacetate and dioctyl adipate in a ratio of 50:50 was used. The polymer proportion was set at 27 wt. %. Nitrogen served as the inner filling. The polymer solution was spun at 225 C. As in Comparative example 2, a mixture of glyceryl triacetate and castor oil in a ratio of 70:30 was used as the cooling medium.

(84) The obtained hollow-fiber membrane had an outside diameter of 549 m and a wall thickness of 132 m. The transmembrane flow for isopropyl alcohol was 0.31 ml/(cm.sup.2.Math.min.Math.bar). The membrane had a bubble point of 1.55 bar, determined by means of the bubble point method with isopropyl alcohol, corresponding to a maximum separating pore of 0.395 m, and had a transmembrane flow for water vapor of 8 l/(m.sup.2.Math.h).