The invention relates to the process for preparing an inorganic porous membrane, to be supported on a suitable carrier, according to the preamble of claim 1. Such a process is known from EP 1 089 806 B1.

Although this known process results in membranes which are suitable for dehydration processes, the drawback of this known process is an unacceptably low yield, a bad reproducibility, and thus very-high costs of usable membranes.

The invention aims to obviate these drawbacks. This aim is obtained by a process as defined in claim 1.

Surprisingly the first application step in which one thus uses, according to the invention, a strongly diluted sol without surfactant results into a well adhering and very homogenously first layer on the support with well distributed nano-openings, while the subsequent layers, applied by using the thus modified diluted sol, to which the surface active agent has been added, builds-up a composite layer with the required final pores.

It is understood that, in the known process, one will never obtain in the first application step a satisfactory coating layer. When one tries to remedy this by applying a consecutive second step, it is found that this second, still wet, coating layer will only locally adhere to the surface of the first layer, because of local repeiience of the wet sol by the already coated and sintered first layer, caused by large differences in surface tension, leading to a non-wetting behaviour, which results in small island formation onto the first coated and sintered layer. Cracks as^ formed during the sintering step at the edge of those formed islands, so that no satisfactory result is obtained. However when one uses in the first coating step a very diluted version of the stock sol as such and without a surface active agent and applies, after the usual sintering thereof, in a consecutive step or steps this highly diluted version with added thereto a surface active agent, such as e.g. Triton 100, one finds very surprisingly that each subsequent layer adheres perfectly to the proceeding applied and sintered layer. The final resulting membrane shows very good performance despite the use of surfactants. This is very surprisingly, because adding surface active agents into ceramic precursors is known to be used for promotion of pore formation.

Preferred embodiments of the process are described in the subclaims. It has been demonstrated that, by executing the process as defined in the claims, it was possible to manufacture ceramic membranes for pervaporation with extremely good performance as to selectivity and with extremely high production yield. The methodology according to the invention can be applied to any suitable ceramic porous carrier, onto tubular carriers, either on the outside or on the inside of such tubular carriers, or onto any ceramic porous carrier in any shape, provided that the surface of the carrier around the pores thereof, there where the sol must adhere, so as to bridge the pores, is smooth and without defects

It is observed that the favourable influence of the addition of a surface active agent to the sol used in a process according to the prior art is described by S. Giessler et al in the Article < Hydrothermal Stability of modified Membranes for Gas Separation > , 6 ^th World Congress of Chemical Engineering, Australia 23-27 September 2001. This Article reports on Laboratory Experiments only and is thus not representative for Industrial Application of this proposal. Moreover, this Article does not disclose the dilution of the sol in the strength as proposed by the present Invention and also does not disclose not to use the surfactant in the first coating step. This latter measure is, however, thought by the inventor to be critical in ensuring a commercially acceptable yield.

Test methodology In order to judge the performance of the functionality of pervaporation membranes, each individual membrane tube is tested under identical conditions and the flux and selectivity is measured. Good functioning ceramic pervaporation membranes show a performance of about 2 to 3 kg/m ² /h throughput with selectivity of > 200, which equals to a water concentration of > 91% in the permeate, when measured under standard conditions as specified below. However in Quality Control a selectivity of 100 is used as threshold for approval.

Standard measurement conditions are: 95% lso Propyl Alcohol, 5% water at 7O ⁰ C, linear velocity to diminish concentration polarisation of 2 m/sec in tube 7 mm duct, giving a Reynolds number of at least 5000 and a vacuum was applied of about 10 m bar for applying sufficient driving force, where the permeate was condensed with liquid nitrogen in a glass cold trap.

Selectivity is defined as the ratio of water over organic in the permeate, divided by the ratio of water over organic in the feed.

Example

A stock sol with a concentration of 1 mol/litre silica material, according to the prior art, such as known from EP 1 089 806 B1 as mentioned above, was prepared as the starting material. A) Coat solution concentration of the diluted Methyl-silica sol: coating the first layer with 0.05 mol/litre to be used for the first coating.

Added to this: Triton 100 in 5% weight fraction, based on the weight of the Methyl-silica compound to be used in the second and subsequent coatings. Number of subsequent coatings with sintering after every coating: 2x

Number of batches and tubes: 3 batches, in total 8 ceramic membrane tubes Yield of 100%, average selectivity of 1395 Worst batch: yield 100%, average-selectivity 817 Best batch: yield 100%, average selectivity 1652

B) Coat solution concentration of the diluted Methyl-silica sol: coating the first layer with 0.05 mol/litre to be used for the first coating.

Added to this: Triton 100 in 5% weight fraction, based on the weight of the Methyl-silica compound, plus different amounts of hexane, from 5% up to 20%, to be used in the second and subsequent coatings.

Number of subsequent coatings with sintering after every coating: 2x Number of batches and tubes: 4 batches, in total 6 ceramic membrane tubes Yield of 100%, average selectivity of 2237 Worst batch: yield 100%, average selectivity 821 Best patch: yield 100%, average selectivity 3416

C) Coat solution concentration of the diluted Methyl-silica sol: coating the first layer with 0.01 mol/litre to be used for the first coating.

For the second coating a concentration of 0.025 mol/litre methyl-silica sol was used, in which 50% of the solvent ethanol was replaced by IPA. Added to this for the second coating: Triton

100 in 5% weight fraction, based on the weight of the Methyl-silica compound to be used in the second coating.

The third coating was identical as the second coating.

Number of subsequent coatings with sintering after every coating: 2x Number of batches and tubes: 1 batch, 4 ceramic membrane tubes

Yield of 100%, average selectivity of 2078

Worst tube: selectivity 506

Best tube: selectivity 3863 D) Coat solution concentration of the diluted Methyl-silica sol: coating the first layer with 0.15 mol/litre to be used for the first coating.

For the second coating a concentration of 0.025 mol/litre methyl-silica sol was used, in which 50% of the solvent ethanol was replaced by IPA. Added to this for the second coating: 2% weight fraction Triton 100.

The third coating was identical as the second coating except 10% Triton 100 was used instead of 2% Triton 100.

Number of subsequent coatings with sintering after every coating: 2x Number of batches and tubes: 8 batches, 76 ceramic membrane tubes Yield of 100%, average selectivity of 436

Worst batch: yield 100%, average selectivity 267 Best batch: yield 100%, average selectivity 1120