You are on page 1of 6

Separation and Purification Technology 22-23 (2001) 361 366

www.elsevier.com/locate/seppur

Dewatering of organics by pervaporation with silica


membranes
H.M. van Veen *, Y.C. van Delft, C.W.R. Engelen, P.P.A.C. Pex
Department of Energy Efficiency,Netherlands Energy Research Foundation, ECN, Westerduinweg 3,
1755 ZG Petten, The Netherlands

Abstract
A major drawback of polymeric membranes for pervaporation is their limited solvent and temperature stability.
This means that for several potential applications the membrane lifetime in combination with a relatively low
performance is the limiting factor for introducing them into the market. More stable membranes are therefore needed.
ECN has developed a new tubular microporous membrane based on hydrophilic silica for the dewatering of organic
solvents. The membranes can be made on a large scale, with lengths of up to 1 meter and have a pore size of about
0.4 nm. The performance of these ceramic membranes for the dewatering of several organic streams has been tested
as a function of feed temperature, feed flow, feed concentration, permeate pressure and time-on-stream. Under the
same conditions the silica membranes give much higher fluxes and selectivities than commercially available dewatering
membranes made of polyvinylalcohol. Up to periods of several weeks the performance of the silica membranes
remains constant. In contrast to the polymeric membranes the ECN silica membranes can be used above 100C, even
up to 300C. Due to an increase in driving force, the water flux in dewatering by pervaporation increases
exponentially with the temperature whereas the organics flux remains small. This means that the membrane surface
area needed for silica membranes can be decreased even further due to the use at higher temperatures. Experiments
learn that at high temperatures the required membrane area for a case study (dewatering of 30 000 l/day 95% ethanol
to 99,9% ethanol) decreases strongly from about 1000 m2 at 80C for polymeric membranes and about 100 m2 for
silica membranes at the same temperature to only a few square meters for silica membranes at 200C. Thus, due to
the outstanding performance at high temperatures the higher price of the ceramic membranes is no longer a
drawback. Furthermore the acid stability of the membrane is much better than zeolite A pervaporation membranes.
The results of the dewatering of several organic solvents are shown as examples of the dewatering capability of the
silica membranes. 2001 Elsevier Science B.V. All rights reserved.
Keywords: Pervaporation; Dewatering; Silica; Ceramic; Membrane

1. Introduction
* Corresponding author. Tel.: +31-224-564606; fax: + 31224-563615.
E-mail address: vanveen@ecn.nl (H.M. van Veen).

Chemical feedstocks, solvents and products of


(chemical) reactions in the liquid phase are very
often a complex mixture of organic components

1383-5866/01/$ - see front matter 2001 Elsevier Science B.V. All rights reserved.
PII: S 1 3 8 3 - 5 8 6 6 ( 0 0 ) 0 0 1 1 9 - 2

362

H.M. 6an Veen et al. / Separation/Purification Technology 22-23 (2001) 361366

and in many cases water. These mixtures are


difficult to separate. The separation is mostly
performed by various distillation techniques
sometimes combined with the use of entrainers
which are hazardous to the environment. Distillation is a very energy intensive process mainly
due to the need to evaporate liquid for generating reflux in addition to the liquid taken off at
the column head as top product. This disadvantage becomes most pronounced in the separation
of close boiling mixtures, where the reflux
stream is far larger than the tops product
stream and in the separation of azeotropes
where multiple distillation steps are required.
Pervaporation is the selective evaporation of
one component of a liquid stream by a membrane, which is in direct contact with the
liquid mixture. Because of this, a much more
energy efficient process can be obtained. Even
difficult separation of azeotropic mixtures can
be performed in one step. Thus, the (partial)
replacement of distillation by pervaporation
or the combination of pervaporation and reaction in one unit operation will have important
benefits, with respect to energy consumption,
yield, product quality and size of down-stream
process equipment. Polymeric membranes are
commercially being used as pervaporation
membranes. A major drawback of these membranes is their limited solvent and temperature stability. This means that for several applications the membrane lifetime in combination with a relatively low performance
is the limiting factor for introducing them into
the market. More stable membranes are therefore needed.
A new microporous membrane based on hydrophilic silica for the dewatering of organic
solvents has been developed by ECN. The performance of these ceramic membranes for the
dewatering of several organic streams has been
tested as a function of feed temperature, feed
flow, feed concentration, permeate pressure and
time-on-stream. In contrast to the polymeric
membranes the ECN silica membranes can be
used above 100C, even up to 300C. Furthermore this membrane is stable in all solvents
tested.

2. Silica membrane development and performance

2.1. Membrane manufacturing


The support system for the silica membrane
consists of 4 layers and is basically made in the
following way. The a-alumina macroporous support tubes which are used as structural carrier
for the actual membrane are made by ceramic
paste extrusion followed by a sintering procedure. The standard diameter of the support tube
is ID/OD =8/14 mm whereas the tubes can be
manufactured in a length up to 1 m. Before the
final membrane layers can be applied two intermediate layers are applied to the support. These
layers accommodate predominantly the surface
roughness and pore size in order to obtain a
nearly defect free support system for the pervaporation membrane layer. The intermediate
layer is coated onto the support tube by means
of a filmcoat technique using an a-alumina colloidal suspension. After drying a sintering step
is involved ensuring consolidation. The so called
gamma layer is applied onto the second intermediate layer by slipcoating of a boehmite sol.
After drying and during a heat treatment this
boehmite will transform to gamma-alumina. The
silica membrane, which is the final separation
layer, is made by means of sol-gel processing. A
silicon alkoxide is hydrolysed from which a
polymeric inorganic silica sol is obtained. This
sol is coated onto the support followed by drying and calcination. All layers are applied on
the outside of the tube. The final structure of
the membrane can be seen in Fig. 1. The thickness of the silica membrane as measured with
SEM is in the range 150 200 nm. From gas
permeancy measurements using several gasses
with different kinetic diameters, the pore size
has been estimated to be about 0.4 nm. The
silica membrane layer is currently calcined
at 400C which limits the operating temperature to 350C. Due to the very hydrophilic
nature of the silica and the small pores this
membrane can be used for dehydration. Details
on the membrane preparation can be found in
[1,2].

H.M. 6an Veen et al. / Separation/Purification Technology 22-23 (2001) 361366

363

2.2. Per6aporation results


The silica membranes have been used for long
term testing in the dewatering of isopropylalcohol. The feed contained approximately 4.5 wt.%
water, 95.5 wt.% IPA and some heavy hydrocarbons, which are present in the process mixture
used. Polymeric membranes for pervaporation are
not stable in (exactly) the same mixture as used
here for long term testing of the silica membranes,
amongst others due to the presence of these heavies [3,4]. The process has been operated at 80C
and 25 mbar pressure at the permeate side, with a

Fig. 2. Dewatering of 95.5 wt.% IPA, 80C, 25 mbar.

total time on stream of 73 days, see Fig. 2. In this


figure the process selectivity is defined as:
XH2O
Xi perm
process selectivity=
XH2O
Xi feed
In which,

 
 

XH20
XH20

Fig. 1. SEM micrograph of a high selective silica membrane.

Xi,

perm

Xi,

feed

= waterconcentration in the permeate


(wt.%)
= waterconcentration in the feed
(wt.%)
= concentration other/organic component in the permeate (wt.%)
= concentration other/organic component in the feed (wt.%)

After an initial decrease of both the water and


IPA flux, both fluxes stabilise and a process selectivity of about 1100 is obtained. This means that
the permeate contains 98.1 wt.% water. The
change of fluxes is probably due to a rearrangement of the hydroxyl groups on the pore surface
of the silica membrane. No deterioration of the
membrane has been found.
The dehydration of ethanol has been used to
study the influence of temperature on the membrane performance. In Fig. 3 it can be seen that
due to an increase in driving force, the water flux
in dewatering by pervaporation increases exponentially with the temperature whereas the organics flux remains small: the selectivity increases
significantly with temperature. This means that

364

H.M. 6an Veen et al. / Separation/Purification Technology 22-23 (2001) 361366

Fig. 3. Dehydration of 96 wt.% EtOH vs. temperature.

the membrane surface area needed for silica membranes can be decreased by applying higher feed
temperatures.
Experimentally supported model calculations
(for the model used, see [5]) show that at high
temperatures the required membrane area for a
case study (batchwise dewatering of 30 000 l/day
95% ethanol to 99,9% ethanol) decreases strongly
from about 1000 m2 at 80C for commercially
available polymeric membranes and about 100 m2
for silica membranes at the same temperature to
only a few square meters for silica membranes at
200C, see Fig. 4.
Thus, due to the outstanding performance at
high temperatures the higher price of the ceramic
membranes is no longer a drawback. Measurements at higher temperatures are being performed
and show that the silica membranes can operate
up to temperatures of 240C without problems.
Detailed results will be reported in future.

Fig. 4. Membrane area needed for polymeric and silica membranes as a function of temperature.

Fig. 5. Water fluxes for several available membranes as a


function of the feed temperature [6].

In Fig. 5 an overview of the performance of


several available pervaporation membranes is
given as a function of the feed temperature. It can
be seen that in general inorganic membranes have
much higher fluxes than polymeric pervaporation
membranes. Compared to zeolite A membranes
the silica membrane had a flux which was somewhat lower than these membranes, however, due
to changes in the production of the silica membrane, fluxes are now comparable to the zeolite A
membranes. A very important advantage of the
silica membranes compared to the zeolite A membranes is the better stability in acid environments.
However, in several dewatering applications a
feed stream with an extreme acidity or alkaline
solutions e.g. for cleaning are used. For these
applications even more stable membranes, by using other types of ceramics, are under
development.
The microporous silica membranes have been
tested in a wide variety of organic solvents, see
Table 1. In almost all cases high fluxes and high
selectivities are obtained. Even the dewatering of
very low water concentrations is possible. For
example in the dewatering of 0.24% water in
dichloorethaan almost 99% water in the permeate
is obtained with a waterflux of almost 1 kg/m2h at
70C. In general dewatering of such low concentrations is not possible with polymeric membranes
as these membranes need a certain minimum water concentration. The silica membranes have
shown to have a stable performance in the wide
variety of solvents tested.

Measurements by partners of ECN.

THF

6
8
10
10
11

10
2
5
5
5

Acetone
Ethyl Acetate
DMF

Feed permeate pressure


(mbar)
6
5
25
10
10
8
10
10
10

Water concentration
(%)

3.6
4.5
Isopropylalcohol
4.5
n-Butanol
5
1,2-dichlorethanea 0.24
Triethyleenglycol
9
Ethyleen diamine 30
Acetonitrila
10
Methylethylketone 2.5

Ethanol

Organic
component

Table 1
Dewatering results of silica membranes for several organic solventsa

50
70
75
100
60

70
71
80
75
70
80
75
70
66

Feed temperature
(C)

752
2936
189
1007
5819

1485
1220
1855
4500
964
184
28
2630
2280

Waterflux
(g/m2h)

78.8
95.9
56.1
84.8
87.9

92.8
94.2
98.1
97
98.9
99.5
99.7
96.4
97.7

Wt.% water in
permeate

33
1118
24
102
147

350
208
1150
600
39645
2054
210
100
1458

Process selectivity

H.M. 6an Veen et al. / Separation/Purification Technology 22-23 (2001) 361366


365

366

H.M. 6an Veen et al. / Separation/Purification Technology 22-23 (2001) 361366

In all tests described above tubular silica membranes with a length of 10 40 cm have been used.
These membranes have been cut out of 1 meter
tubes, which are made in batches of 20 tubes at a
time. The membranes used are selected at random
and all measurements have been performed in
duplo.

tion for energy and environment (NOVEM) for


financially supporting part of the work described
here. We would like to thank Stefan Sommer of
the IVT-RWTH in Aachen for performing the
measurements on ACN dehydration and Ine Bos
and Wridzer Bakker of Akzo Nobel in Arnhem
for the 1,2 DCE measurements.

3. Conclusions

References

Ceramic pervaporation membranes have much


higher fluxes and selectivities than commercially
available polymeric membranes. This means that
the higher prices of these membranes is no limiting factor any more for commercial use. The
manufacturing process of the silica membranes is
relatively easy upscalable. The performance of the
silica membranes is comparable to commercially
available zeolite A membranes. Due to the better
stability and the development of even more stable
ceramic membranes the range of applications for
these membranes is much wider. The silica membranes have shown good performances in the
dewatering of solvents like methylethylketone,
DMF, THF and ethylacetate.

[1] B.C. Bonekamp, Preparation of asymmetric ceramic membrane supports by dipcoating, in: A.J. Burggraaf, L. Cot
(Eds.), Fundamentals of Inorganic Membrane Science
and Technology, vol. 4, Elsevier, Amsterdam, 1996 Chapter 6.
[2] B.C. Bonekamp, P.P.A.C. Pex, Suspensions and Sol Processing for the Manufacturing of High Performance Ceramic Pervaporation and Gas Separation Membranes, to
be published in Industrial Ceramics, 2000.
[3] G.W. Meindersma, Membrane Experiences from the Petrochemical Industry, presentation at the Membrane Technology information day, Antwerp, Belgium 1996.
[4] G.W. Meindersma, M. Kuczynski, Implementing membrane technology in the process industry: problems and
opportunities, J. Membr. Sci. 113 (1996) 285.
[5] M. Mulder; Basic principles of Membrane Technology-second edition, Kluwer Academic Publishers, 1998.
[6] (a) Brochure GFT-Carbone Lorraine, Applications of
Pervaporation Processes, 1995. (b) K. Okamoto, H.
Kita, Membrane for liquid mixture separation, European Patent Application 1995, no. 659 469 A2. (c)
S.A.I. Barri, G.J. Bratton, T. de V. Naylor, Membranes, European Patent Application 1992, no. 481 660
A1.

Acknowledgements
We gratefully acknowledge the Dutch organisa-

You might also like