Preparation of PDMS - Al O Composite Hollow Fibre Membranes For VOC Recovery From Waste Gas Streams
Preparation of PDMS - Al O Composite Hollow Fibre Membranes For VOC Recovery From Waste Gas Streams
Preparation of PDMS - Al O Composite Hollow Fibre Membranes For VOC Recovery From Waste Gas Streams
Preparation of PDMS
vi
Al
2
O
3
composite hollow bre
membranes for VOC recovery from waste gas streams
Shaomin Liu
a
, W.K. Teo
a
, Xiaoyao Tan
b,c
, K. Li
b,
a
Department of Chemical and Environmental Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore
b
Department of Chemical Engineering and Chemical Technology, Imperial College London, University of London, London SW7 2AZ, UK
c
Department of Chemical Engineering, Shandong University of Technology, Zibo 255049, PR China
Received 17 November 2004; received in revised form 20 February 2005; accepted 30 April 2005
Abstract
A new PDMS
vi
Al
2
O
3
composite hollow bre membrane has been developed by coating a PDMS
vi
-oligo lm on the outer surface of
an Al
2
O
3
hollow bre porous substrate prepared by a dry/wet phase inversion method. Microstructures of the composite membranes were
examined using scanning electron microscopy (SEM), displaying a uniform PDMS
vi
-oligo coating layer, free of defects with a thickness of
around 15 m. Such composite hollow bre membranes are suitable for VOC recovery from waste gas streams because of their high chemical
stability. The separation performances of the PDMS
vi
Al
2
O
3
composite hollow bre membrane were studied using chloroform in N
2
as a test
vapour mixture. By analyzing the experimental data obtained at various conditions with a mathematical model derived, the permeabilities of
chloroform and nitrogen in the PDMS
vi
polymer membrane were obtained. The effects of temperature, feed ow rate, vacuum pressure on
the permeate side and feed concentration on the chloroform recovery were also investigated both experimentally and theoretically.
2005 Elsevier B.V. All rights reserved.
Keywords: Ceramic hollow bre; Composite membrane; VOC removal
1. Introduction
Waste gas streams emitted from many industrial pro-
cesses such as printing, metal cleaning, painting and other
sources using solvent and paint contain low concentra-
tions (10800 g m
3
(STP)) of volatile organic compounds
(VOCs). Control of emission of these VOC laden waste gas
streams has become an issue of increasing attention due to
many adverse effects of these pollutants. Incineration was a
usual method in past years to eliminate VOCs from waste
gas streams [13], but it also means a signicant wastage of
valuable materials and energy resources. Recently, several
procedures, such as carbon adsorption, absorption and con-
densation, have been developed to recover the valuable VOCs
fromplant efuents. The main disadvantage of these methods
is the high cost and therefore not economically applicable.
m
thickness of PDMS
vi
coating (m)
A
,
B
,
p
viscosity of chloroform, nitrogen gas and
the permeate gas (Pa s)
Recently, PDMS
vi
cross-linked by a novel oligomer,
oligosilylstyrene-polydimethylsiloxane
vi
was prepared and
studied [13,14]:
Membranes cast fromthis newsilicone rubber were found
to be capable of separating chlorinated and aromatic hydro-
carbons from water efciently [13]. The sorption equilibria
and kinetics of some VOCs (i.e., benzene, chloroform and
acetone) in this new polymer lms were studied by Zhen
et al. [14]. The sorption and diffusion data revealed that the
newly developed PDMS
vi
-oligo polymer is an excellent VOC
sorbent. The polymer appears tobe a goodcandidate for fabri-
cation of composite membranes for use in VOCseparation by
vapour permeation [14]. Although this kind of polymer has a
good processing ability for making ultra-thin membranes on
polymeric hollowbre supports [15], the feasibility of apply-
ing the same material on the ceramic surface, particularly in
hollow bre conguration, to form a ultra-thin membrane,
has not been well examined.
In this study, the possibility of turning PDMS
vi
and
Al
2
O
3
ceramic into a composite hollow bre membrane was
explored by coating the PDMS
vi
-oligo lm on tailor-made
porous alumina ceramic hollow bres. To demonstrate the
application of this new membrane in VOCs recovery, this
study focused on removal of chloroform from nitrogen
streams using the hollow bre membranes developed. The
effects of operating conditions, such as temperature, feed
ow rate, vacuum pressure in the permeate side and feed
concentration on the chloroform recovery, were investi-
gated.
2. Experimental
2.1. Materials
Divinyl terminated polydimethylsiloxane (PDMS
vi
)
(1000 cs, UCT, analytical grade), platinum-divinyltetra-
methyldisiloxane complex (UCT, analytical grade), oligosi-
lylstyrene and hexane (analytical grade, J.T. Baker) were
used for preparing the PDMS
vi
coating lms. Aluminium
oxide powders with particle size of 0.3 m in diameter
(gamma/alpha, surface area 15 m
2
g
1
) [purchased fromAlfa
AESAR, a Johnson Matthey company] was used as the mem-
brane substrate material. Polyethersulfone (PESf) [Radel A-
300, Ameco Performance, USA], N-methyl-2-pyrrolidone
(NMP) [synthesis grade, Merck] and polyvinylpyrrolidone
(PVP, K90) [GAF
C for at least
2 h. Three repetitions of coating and post-cross-linking pro-
cedure were carried out to obtain a uniform PDMS
vi
layer of
about 15 m. The microstructures of the hollowbres before
and after coating were examined using a scanning electron
microscopy (JEOL JSM-5600).
2.3. Chloroform recovery in the PDMS
vi
/Al
2
O
3
composite hollow bre membrane module
Amodule consisting of six PDMS
vi
/Al
2
O
3
composite hol-
low bre membranes was manufactured to perform the VOC
recovery. The properties of the hollow bre membrane mod-
ule are summarized in Table 1. A chloroform/nitrogen gas
mixture was used as a testing gas. Fig. 1 shows schematically
Fig. 1. Schematic diagram for the experimental system for chloroform recovery.
S. Liu et al. / Separation and Purication Technology 46 (2005) 110117 113
the experimental system for chloroform recovery. Feed gas
containing chloroformwas obtained by bubbling the nitrogen
gas through the chloroform liquid. Feed gas was introduced
into the shell side of the module. The whole module was
housed in a constant temperature control box where heat-
ing element and circulating fan and temperature sensor were
used to maintain the isothermal condition. A vacuum pump
was applied to the bre lumen, providing the necessary driv-
ing force for the vapour permeation. The ow rate of residue
gas stream was measured using a soap bubble ow meter. In
order to keep the membranes clean, the system was purged
with nitrogen both before and after each run. Perkin-Elmer
8700 gas chromatography with a FID detector was used to
analyze the concentrations of samples taken from the feed,
permeate and residual gas streams. The recovery of chloro-
form was calculated by:
R
ev
= 1
L
w
x
w
L
f
x
f
(1)
where L
w
and L
f
are the ow rates of the residual and feed
gas streams, x
w
and x
f
are the mole fractions of chloroform
in the residue stream and in the feed stream, respectively.
3. Theory
3.1. Formation of model equations
The mathematical model developed in the following is
basedonthe actual operatingmode: chloroform/nitrogenfeed
gas is fed into the shell side while a vacuum is applied to the
lumen of the membrane module at the feed end, and thus,
the permeate streams ows counter-currently with that of the
feed stream. The residue end of the bre lumen is closed.
Fig. 2 shows the feed and permeate ows in the hollow bre
membrane module, in which only one bre is depicted. Based
on the operating arrangement, the following assumptions are
adopted for the model development:
(1) The resistance of Al
2
O
3
support to gas permeation is
negligible. Also, no mass transfer resistance in the bulk
gas phases should be in consideration.
(2) Gas streams in both the shell and lumen sides are plug
ow, and the concentration gradients in the radial direc-
tion are negligible.
(3) The pressure inthe shell side is constant alongthe module
while HagenPoiseuille law governs the pressure drop
through the bre lumen.
(4) The end effects inside the module as well as the defor-
mation of the hollow bres under external pressure are
negligible.
(5) The system is operated isothermally and in steady state.
Based on the above assumptions, the general mass conserva-
tion equations for both chloroform and nitrogen, represented
by A and B respectively, can then be written as:
dN
A
dz
= 2nR
m
_
P
A
m
_
(p
s
x p
l
y)
=
d(Lx)
dz
, for chloroform (2)
dN
B
dz
= 2nR
m
_
P
B
m
_
[(p
s
(1 x) p
l
(1 y)]
=
d[L(1 x)]
dz
, for N
2
(3)
where N
A
and N
B
are the molar ow rates of chloroform
and nitrogen in the bre lumen (mol s
1
), L the feed-side
gas ow rate (mol s
1
), n the number of the hollow bres,
R
m
the logarithmic mean radius of the coated hollow bres
_
R
m
=
m
ln(1+
m
/R
o
)
_
, in which R
o
and
m
are the outer radius
of Al
2
O
3
hollow bre and the thickness of PDMS
vi
coating,
respectively (m), P
A
and P
B
are the gas permeabilities of
VOC (chloroform) and of nitrogen (mmol m
2
s
1
Pa
1
), x
and y are the feed-side and permeate-side mole fractions of
VOC and p
s
and p
l
are the gas pressures on the shell side and
in the bre lumen, respectively (Pa).
The equation governing the pressure drop in the bre
lumen is given by:
dp
l
dz
=
8T
p
(N
A
+N
B
)
nR
4
in
p
l
(4)
where is the gas constant (8.314 J mol
1
K
1
), T the oper-
ating temperature (K) and
p
is the viscosity of the permeate
gas which can be expressed as:
p
=
A
y +
B
(1 y) (5)
Fig. 2. Schematic operation diagram of the hollow bre membrane module for VOC recovery.
114 S. Liu et al. / Separation and Purication Technology 46 (2005) 110117
The boundary conditions for the above governing equations
are given by:
z = 0, N
A
= 0, N
B
= 0 (6a)
z = Z, L = L
f
, x = x
f
, p
l
= p
f
(6b)
3.2. Numerical solution
The governing equations are a group of ODEs and can be
solved numerically with the boundary conditions. However,
in the vacuumoperating mode, Eqs. (2) and (3) are indetermi-
nate with the boundary condition of Eq. (6a) at the closed end
of the module, z =0. This problem however may be solved as
follows:
Applying the LHospital rule, the governing equations
may be changed to:
dN
A
dz
z=0
= 2nR
m
_
P
A
m
_
d[p
s
x(N
A
+N
B
)p
l
N
A
]
dz
z=0
d(N
A
+N
B
)
dz
z=0
(7a)
dN
B
dz
z=0
= 2nR
m
_
P
B
m
_
d[p
s
(1x)(N
A
+N
B
)p
l
N
B
]
dz
z=0
d(N
A
+N
B
)
dz
z=0
(7b)
Rearranging the above equations gives:
dN
A
dz
z=0
= 2nR
m
_
P
A
m
_
p
s
x (1 +F) p
l
F
1 +F
(8a)
dN
B
dz
z=0
= 2nR
m
_
P
B
m
_
p
s
(1 x)(1 +F) p
l
1 +F
(8b)
in which
F =
dN
A
dN
B
z=0
=
[1 (S
e
+1)x +(S
e
1)p
l
/p
s
] +
_
[1 (S
e
+1)x +(S
e
1)p
l
/p
s
]
2
+4S
e
x(1 x)
2(1 x)
(9)
where S
e
is the VOC-N
2
selectivity, S
e
=P
A
/P
B
.
Shooting method was employed in this work since the
initial values of the pressure, the residual VOCconcentration
and the ow rate for integration are unknown. The tolerance
function is dened as:
Er =
_
_
p
l
z=L
p
f
1
_
2
+
_
x|
z=L
x
f
1
_
2
+
_
L|
z=L
L
f
1
_
2
(10)
The value of Er for convergence was set at 10
4
in this work.
4. Results and discussion
4.1. PDMS
vi
/Al
2
O
3
composite hollow bre membranes
The porous Al
2
O
3
hollowbres were preparedthroughthe
dry/wet phase inversion method [16]. Using this method, the
particle size of the starting Al
2
O
3
powders plays an important
role in the properties, such as surface roughness, pore size and
mechanical strength of the resulting ceramic hollow bres
[17]. Our experimental result reveals that the hollow bres
prepared from 0.3 m particles exhibit the appropriate sur-
face roughness for lm formation. Therefore, all the porous
Al
2
O
3
hollow bre substrates for the further preparation
of PDMS
vi
/Al
2
O
3
composite membranes in this work were
made fromthe Al
2
O
3
powders with particle size of 0.3 min
diameter. The characteristic parameters of the porous Al
2
O
3
hollow bres are reported in Table 1. Nitrogen gas per-
meance through the Al
2
O
3
hollow bres was measured to
be 2.54 10
5
mol m
2
Pa
1
s
1
. The microstructure of the
prepared Al
2
O
3
hollow bres is shown in Fig. 3. The appear-
ance of the asymmetric structure of the Al
2
O
3
hollow bres,
as shown in Fig. 3AI and BI, can be attributed to the differ-
ent local precipitation rates of the polymer/Al
2
O
3
solution
within the hollow bre wall, which is the special characteris-
tic of the phase inversion process. Fig. 3CI shows the surface
of the Al
2
O
3
hollow bres, indicating that the hollow bres
are highly porous.
Preparation of the PDMS
vi
Al
2
O
3
composite membrane
involves two steps: vacuumcoating and post-cross-linking in
succession. In order to obtain a desired composite membrane,
each step must be carried out carefully. The thickness and
uniformity of PDMS
vi
coating layer depend critically on the
physical properties of the substrate (i.e., surface roughness
and pore size distribution), the vacuum duration and also
on the viscosity of coating solution. Heat treatment and
post-cross-linking are to enhance the strength of the bondage
between the pre-coated PDMS
vi
layer and the Al
2
O
3
substrate. Our experimental results indicate that three rep-
etitions of the 1-min vacuumcoating and drying operation are
necessary to obtain a uniform PDMS
vi
coating layer without
defects. The microstructure of the nal PDMS
vi
Al
2
O
3
com-
posite membranes is also shown in Fig. 3. As can be seen from
Fig. 3AII and BII, which illustrates the cross-sectional wall
of the composite hollow bre membrane, the PDMS
vi
coat-
ing layer scarcely penetrate into the ceramic surface pores
and a perfect uniform layer of PDMS
vi
with a thickness of
15.4 mhas been formed on the Al
2
O
3
hollowbre substrate
surrounding the outer surface. Fig. 3CII shows the surface of
the PDMS
vi
lm. It can be seen that the membrane surface
is perfectly smooth. No micropores, defects or peeling-off
was observed throughout the whole deposited PDMS
vi
lm.
S. Liu et al. / Separation and Purication Technology 46 (2005) 110117 115
Fig. 3. SEM micrograph of the Al
2
O
3
hollow bre substrate (I) and the PDMS
vi
Al
2
O
3
composite membrane (II); (A and B) cross-sectional and (C) outside
membrane surface.
Due to the formation of the dense PDMS
vi
layer, the result-
ing PDMS
vi
/Al
2
O
3
composite membrane exhibits a much
lower N
2
gas permeance than the original Al
2
O
3
hollowbre
substrate, i.e., 1.68 10
9
mol m
2
Pa
1
s
1
at room tem-
perature.
4.2. Recovery of chloroform from nitrogen streams
Chloroform recovery from nitrogen gas streams was car-
ried out in the PDMS
vi
/Al
2
O
3
composite hollow bre mem-
brane module with properties as summarized in Table 1.
A vacuum of 8 mbar was applied to the bre lumen with
counter-current ow, while the feed-side pressure was kept
at atmospheric pressure. The chloroform concentration in
the feed stream was kept constant, i.e., 20,800 ppmv in all
runs. Effects of temperature and feed ow rate on the chlo-
roform recovery were investigated experimentally. Based on
the experimental results, the studies were also extended the-
oretically to other factors affecting the performances of the
membrane module for VOCrecovery including the operating
pressure and the feed concentration.
Fig. 4 plots the chloroformrecovery against feed owrate
at four different operating temperatures. It can be seen that the
chloroform recovery decreases with increasing the feed ow
rate at the same operating temperature. Note that the lower the
ow rate, the longer the retention time of chloroform in the
membrane module, and therefore, the higher the possibility
for the chloroform molecules in the feed stream permeating
through the membrane into the permeate side. When the feed
ow rate is as low as 15 ml min
1
, the chloroform recovery
approaches to 1 at any temperatures. This suggests that the
chloroformcan be completely removed fromthe nitrogen gas
stream using the hollow bre membrane module. The effect
of temperature on the chloroform recovery may also be seen
116 S. Liu et al. / Separation and Purication Technology 46 (2005) 110117
Fig. 4. Chloroformrecovery against feed owrate at different temperatures,
() simulation; experimental data: () at 40
C, () at 50
C, () at 60
C
and () at 70
C.
from Fig. 4. It indicates that as temperature increases, the
chloroform recovery is decreased unless the feed ow rate is
lower than 15 ml min
1
, under which the chloroform recov-
ery reaches the maximum value. This may be explained as
follows. According to the solution-diffusion model, gas per-
meability in rubbery polymers is related not only to the gas
mobility but also to the gas solubility in the membrane, which
is a measure of the energy required for the permeant to be
sorbed by the polymer. For organic vapours, however, sorp-
tion of the polymer is usually the dominant term. As a result,
the permeability of VOCs increases with decreasing temper-
ature due to the exothermic nature of the sorption process.
Subsequently, it follows that the lower temperature favors
the recovery of chloroform in the PDMS
vi
/Al
2
O
3
composite
hollow bre membrane module.
The theoretical results of chloroform recovery under var-
ious temperatures and feed ow rates are also illustrated in
Fig. 4 in solid lines where permeability values of chloroform
and nitrogen were obtained by regression with the experi-
mental data and presented in Fig. 5. These values have the
same order of magnitude as those reported for vapour per-
meation in the normal PDMS membrane [18]. As described
above, the gas permeability is related to the gas mobility and
the solubility in the membrane, and may be expressed in the
form of Arrhenius equation as:
P = DS = P
0
exp
_
E +H
T
_
(11)
where D is the gas diffusion coefcient (m
2
s
1
), S the gas
solubility in the polymer membrane (mol m
3
Pa
1
) and E
Fig. 5. Relationship between the gas permeance and temperature.
and H are the activation energy for diffusion and the heat
of solution of the gas, respectively (J mol
1
).
The plot of gas permeance (P/
m
) against the reciprocal
of temperature is shown in Fig. 5. It can be seen that the per-
meances of both chloroform and nitrogen can be perfectly
illustrated against temperature in Arrhenius equation form.
Fromthe intercept and slope of the lines, values of the Arrhe-
nius parameters are given in Table 2. Since the activation
energy for diffusion of gases is always positive, the negative
value of E+H for chloroform implies that the permeability
of chloroformis dominated by its solubility in the membrane.
However, the positive value of E+H for nitrogen suggests
that the permeation of nitrogen in the PDMS
vi
lm is domi-
nated by the diffusion term, i.e., the same as that in the normal
rubbery membranes. In addition, the selectivity for chloro-
form and nitrogen is further given by:
S
e
= 6.159 10
3
exp
_
22638.19
T
_
(12)
This suggests that the selectivity decreases with increas-
ing temperature. Therefore, the recovery of chloroform from
nitrogen gas streams is not appropriate to be operated at ele-
vated temperatures in the PDMS
vi
/Al
2
O
3
composite hollow
bre membrane module.
It should be mentioned that the gas permeances through
the PDMS
vi
/Al
2
O
3
composite membrane are much lower
than those through the Al
2
O
3
substrate, e.g., 2.54 10
5
mol m
2
Pa
1
s
1
for N
2
, as described above. Accordingly,
the assumption of negligible mass transfer resistance of the
substrate to gas permeation is, therefore, justied.
Using the parameters obtained above, the effect of vac-
uum pressure and feed concentration on the chloroform
recovery has been studied theoretically, as shown in Fig. 6.
As can be seen, the chloroform recovery decreases with
Table 2
Arrhenius parameters for the permeances (P/
m
) of chloroform and nitrogen in PDMS
vi
polymer
Gas Pre-exponential factor (mol m
2
s
1
Pa
1
) E+H (J mol
1
) Temperature range (K) Correlation coefcient (R
2
)
Chloroform 7.218 10
11
18284.98 313343 0.997
Nitrogen 1.272 10
8
4353.21 313343 0.956
S. Liu et al. / Separation and Purication Technology 46 (2005) 110117 117
Fig. 6. Effects of vacuumpressure and feed concentration on the chloroform
recovery (temperature: 40