Author’s Accepted Manuscript Preparation and pervaporation performance of vinyl-functionalized silica membranes Sadao Araki, Ami Okabe, Akira Ogawa, Daisuke Gondo, Satoshi Imasaka, Yasuhisa Hasegawa, Koichi Sato, Kang Li, Hideki Yamamoto www.elsevier.com/locate/memsci
PII: DOI: Reference:
S0376-7388(17)31471-0 https://doi.org/10.1016/j.memsci.2017.10.066 MEMSCI15693
To appear in: Journal of Membrane Science Received date: 29 May 2017 Revised date: 28 October 2017 Accepted date: 31 October 2017 Cite this article as: Sadao Araki, Ami Okabe, Akira Ogawa, Daisuke Gondo, Satoshi Imasaka, Yasuhisa Hasegawa, Koichi Sato, Kang Li and Hideki Yamamoto, Preparation and pervaporation performance of vinyl-functionalized silica membranes, Journal of Membrane Science, https://doi.org/10.1016/j.memsci.2017.10.066 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Preparation and pervaporation performance of vinyl-functionalized silica
membranes
Sadao Araki 1 * , Ami Okabe 1 , Akira Ogawa 1 , Daisuke Gondo 1 , Satoshi Imasaka 1 ,
Yasuhisa Hasegawa 2 , Koichi Sato 2 , Kang Li 3 , Hideki Yamamoto 1
1. Department of Chemical Engineering, Kansai University, 3 -35 Yamatecho
3-chome, Suita 564-8680, Japan
2. National Institute of Advanced Industrial Science and Technology (AIST),
Research Institute for Chemical Process Techno logy, Nigatake 4-2-1, Sendai
983-8551, Japan
3. Department of Chemical Engineering, Imperial College London, London
SW7 2AZ, UK
*
[email protected] 㻌
Research highlights
࣭ A vinyl-functionalized silica (VFS) membrane was prepared on an alumina
hollow fiber.
࣭ The ethyl acetate flux of this membrane is three times higher than that of a
membrane prepared on a tubular substrate.
࣭ This membrane exhibited a separation factor of 85 and an EA flux of 3.5 kgm –
2 –1
h for EA removal.
Abstract
Pervaporation using hydrophobic membranes to separate azeotropic and
thermally degradable mixtures, and organics from aqueous solutions with low
concentrations has attracted attention. On the other hand, distillati on, which is
the conventional and commercialized separation me thod, is unsuitable for the
separation of these mixtures. In this study, an D-alumina substrate coated with a
J-alumina layer in both hollow fibe r and tubular geometries was used for
preparation of vinyl-functionalized silica (VFS) membrane s by the sol-gel
method using vinyltrimethoxysilane as the silica precursor. Silica functionalized
with vinyl groups shows excellent affinity for ethyl acetate (EA) based on
Hansen solubility parameters [1]. The VF S membrane prepared on a tubular
alumina substrate at 453 K showed the EA flux (0.85 kg m – 2 h – 1 ) and separation
factor (77.3) for pervaporation of a 5 % (mass fraction) EA aqueous solution at
295 K. An D-alumina hollow fiber was further used instead of the conventional
tubular substrate to improve the EA flux. The hollow fiber-supported VFS
membrane has a flux of 2.53 kg m – 2 h – 1 , which is three times higher than that
prepared on the tubular-substrate, and a separation factor of 81.5 for the
pervaporation of a 5 % (mass fraction) EA from aqueous solution at 295 K . The
fluxes of methyl acetate and methyl ethyl ketone for the hollow fiber -supported
VFS membrane are also about three times higher than th ose obtained for
tubular-supported VFS membranes.
Keywords: Pervaporation; Hydrophobic; Vinyl; Silica; Hollow fiber
1. Introduction
Separation by pervaporation (PV) uses hydrophobic membranes to separate
azeotropic or thermally degradable mixtures, and has attracted attention [2].
Recovery or removal of volatile organic compounds (VOCs) with low
concentration
in
aqueous
solutions, such
as
wastewater,
by
PV using
hydrophobic membranes has also been reported [3, 4]. VOCs are used in the
manufacture of various products, including paints, pharmaceuticals, plasticizers,
varnishes, and adhesives [3]. Factory wastewater containing VOCs can be
harmful to the environment and to human health; it can also lead to air pollution
without proper treatment [4]. PV can be used to remove or recover VOCs from
water with high efficiency, and in eco -friendly manner compared to conventioanl
distillation [5].
Recently,
Zhou
et
al.
reported
using
an
octyl -functionalized
polydimethylsiloxane membrane to remove VOCs from water, and achieved a
separation factor of 592 and total flux of 0.19 kg m – 2 h – 1 for PV of 1 % (mass
fraction) ethyl acetate (EA) in aqueous sol ution at 313 K [4]. Penkova et al.
reported
the
physicochemical
poly(2,6-dimethyl-1,4-phenylene
features
oxide)
and
of
nanocomposites
fullerene
(C 6 0 )
of
composite
membranes for EA production [6]. This membrane concentrated EA from 17.9 %
to 67.4% in a solution containing ethanol, water, and acetic acid. In their PV
process, the EA flux was about 0.06 kg m – 2 h – 1 at 293 K. Kujawa et al. reported
the use of a hydrophobic ceramic membrane to remove VOCs from a waste
solution [7]. They modified the surfaces of a lumina (pore size 5 nm), titania
(pore size 2–4 nm), and zirconia (pore size 2 –4 nm) substrates with
1H,1H,2H,2H-perfluorooctyltriethoxysilane (C6). The C6 -modifed titania gave a
separation factor of about 30 and EA flux of about 1.8 kg m – 2 h – 1 in PV
experiments for 2 % (mass fraction) EA aqueous solution at 308 K. In our
previous study, we separated EA and methyl ethyl ketone (MEK) from water
using PV with hydrophobic silica membranes modified with methyl, propyl,
isobutyl, n-hexyl, and phenyl groups [1]. In addition, the effects of the
concentrations of organic compounds in the feed solution and temperature on PV
performance for the propyl-functionalized silica membrane were also evaluated.
A separation factor of 78 and EA flux of 1.75 kg m – 2 h – 1 were achieved by PV
using the membrane for a 5 % (mass fraction) EA aqueous solution at 315 K. Our
results suggested that the organic flux improved when the Hansen solubility
parameters (HSPs) of the membrane were close to that of the organic compound
[1]. The HSP consists of the dispersion force (δ d MPa 1 /2 ), the dipole interaction
force factor (δ p MPa 1 /2 ), and the hydrogen-bonding force factor (δ h MPa 1 / 2 ). The
compatibility between substances A and B can generally be evaluated using the
HSP distance between A and B, conventionally called R a , as follows: R a = [4(δ d A − δ d B ) 2 + (δ p A – δ p B ) 2 + (δ h A – δ h B ) 2 ] 1 /2 (1)
Based on the HSP values [1], the vinyl group for modified silica was selected for
the separation EA, which is widely used for the production of perfumes,
plasticizers, varnishes, synthetic resins and adhesive agents, from water [ 8].
Compared with silica functionalized in previous wo rk [1], the HSP of
vinyl-functionalized silica (VFS) is more similar to that of EA. An earlier study
found that for a hydrophobic silica membrane prepared by the sol -gel method
using vinyltrimethoxysilane (VTMS) and t e t r a e t h y l o r t h o s i l i c a t e ( T E O S ) ,
the contact angle of water droplets on the membrane increased with the ratio of
VTMS/TEOS, and reached 144° at a VTMS/TEOS ratio of 0.97 [9]. Therefore,
the VFS membrane is expected to show good performance for the separation of
EA from water by PV.
In addition to the type of membrane used in PV, the properties of the substrate
also affect the permeation flux [10]. Since hollow fibers generally show a lower
transport resistance than conventional tubular substrates, the use of hollow
fibers could improve the permeation flux for the same membrane [11, 12]. Wan g
et al. reported the water flux of a zeolite LTA membrane prepared on a hollow
fiber substrate is 6.2 kg m – 2 h – 1 and the separation factor (D H 2 O / e th an o l ) is 125,000.
This separation factor is about three times higher than that of the membrane
prepared on a tubular substrate for a 90 % (mass fraction) eth anol aqueous
solution at 348 K [13]. Shao et al. found the water flux increased linearly with
increasing porosity of the hollow fiber substrate [14]. The effect of changes in
the substrate properties should be investigated further to improve the flux.
Although many researchers have reported hollow fiber supported membranes,
there are little studies of the impact of support properties and feed concentration
on PV performance [15-20].
In this study, the effect of the calcination temperature on PV performance for
a 5 % (mass fraction) EA aqueous solution at 295 K a of a VFS membrane
prepared on a conventional tubular alumina substrate (outer diameter = 10mm)
was investigated for the first time. An alumina hollow fiber (outer diameter =
2mm) with porosity and average pore diameter similar to that of the tubular
substrate was used to improve the EA flux. In addition, the effect of support
resistance was investigated. The thickness of the J-alumina layer was controlled by the withdrawal rate during dip-coating. The influence of the J-alumina layer
and type of substrate on the PV properties of the VFS membrane for separation
of EA from water were investigated in detail. In addition, we evaluat e the effect
of the concentration of EA in the feed solution on the EA flux and se paration
factor for VFS membranes prepared on the conventional tubular substrates as
well as the hollow fiber substrates.
2. Experimental
2.1 D-alumina supports Porous D-alumina tubes (outer diameter = 10 mm, inner diameter = 6 mm, porosity = 60 %, average pore size = 150 nm) and porous D-alumina hollow
fibers (outer diameter = 2 mm, inner diameter = 1.6 mm, porosity = 60 %,
average pore size = 150 nm) were used as the support for the silica membrane
(Fig. 1). One end of the supports was capped with a dense D-alumina plate and the other end was connected to a dense D-alumina tube, in both cases using a
glass sealant.
2.2 Preparation of J-alumina coating
A boehmite sol was prepared by the hydrolysis of aluminum sec-butoxide
(TCI, Japan) using the following procedure. Aluminum sec-butoxide was added
to approximately 200 mL of water at 353 K and then 10 mL of 1 mol/L HNO 3
was
㻔㼍㻕
㻔㼎㻕
Fig. 1 Photograph of the D-alumina (a) tubular support and (b) hollow fiber. 㻌
added to this mixture under stirring. The mixture was heated to 363 K for 3 h,
and then refluxed for 24 h. The concentration of the boehmite sol was adjusted
to 1.0 mol/L by evaporation or by dilution with deionized water. A polyvinyl
alcohol (PVA, MW = 2,000, Wako Pure Chemical Industries, Japan) solution was
prepared by adding 1.75 g of PVA to 50 mL of 0.05 mol/L HNO 3 aqueous
solution at 363 K. The boehmite solutions used for dip -coating were prepared by
mixing 25 mL of boehmite sol and 10 mL of the PVA solution, and mixing
ultrasonically for 20 min. The α-alumina substrate was dip-coated in the
boehmite solution for 20 s. The thickness of the J-alumina layer was controlled
by the withdrawal rate of the alumina support from the solution (0.5, 1, 3 5 mm
s – 1 ). The J-alumina layer which was prepared using a withdrawal rate of 1 mm min – 1 was used to confirm the optimal calcination temperature of the VTMS
membrane. The support was dried at room temperature for 3 h, and then calcined
at 873 K for 24 h.
2.3 Preparation of VFS membranes
The VFS layer was formed on the D-alumina supports coated with J-alumina
layer. Cetyl trimethylammonium bromide (0.008 mol), VTMS (0.1 mol), and 7.5
mL of 1 mol L – 3 HNO 3 were dissolved into 25 mL of ethanol, and then the
mixture was stirred at room temperature for 3 h. The J-alumina coated support
was immersed in the sol for 60 s and then removed at a rate of 1 mm s – 1 . The
membrane was dried at room temperature and then calcined in air at 453 K.
Finally, the membrane was washed with a large volume of ethanol to remove
cetyl trimethylammonium bromide.
2.4 Characterization
The membrane morphology was observed by a scanning electron microscopy
(-4800,
Hitachi
High-Technologies
Corp).
The
hydrophobicity
of
VFS
membranes was evaluated by measuring the contact angle of water (DMs-400,
Kyowa Interface Science Co., Ltd.), and the effective pore diameter was
determined using permporometry with EA [1]. In addition, the thermal properties
of the VTMS was analyzed by thermogravimetric analysis (Thermo plus EV O2,
Rigaku, Japan).
2.5 Gas permeation and pervaporation
Single gas permeation of helium, nitrogen and methane were determined from
a permeation pressure drop across the membrane at room temperature. The pure
gas was fed at the outer surface of membrane (feed side), and the permeated gas
flow rate at the inside of membrane (permeate side) was measured using a
bubble flowmeter. The pressure difference between feed and permeate side s was
kept at 100kPa.
The PV properties of the VFS membrane for EA, methyl acetate (MA) and
MEK in aqueous solutions were measured at 295 K. The membrane was
immersed in the aqueous solution, and vacuum was applied at the inner surface
of the membrane using a vacuum pump. The permeated vapors were collected
using a condenser with liquid nitrogen for 1 h, and the mass and composition of
the liquid after condensation for 1 h on liquid nitrogen were measured using a
scale and a Karl Fischer water meter (AQV-2100, Hiranuma Industry),
respectively. When the permeated mixture had separa ted into two phases,
ethanol was added to obtain a homogeneous mixture. The mass fraction of EA in
the feed was changed from 1 % to 7 % to confirm the effect of feed concentration
on the PV performance of the VFS membrane. The flux and separation factor of
the organic compounds with respect to water were calculated as follows:
J=W/At D=(y o rg an i c /x o rg an i c )/(y w a t e r /x w at e r )
(2)
(3)
where W is the permeate mass (kg), A is membrane area (m 2 ), t is the
measurement time (h), and x and y are the r mass fractions of water and an
organic compound in the feed ( x) and permeate (y).
3. Results and Discussion
3.1 Effect of calcination temperature
First, the VFS membranes were prepared on a conventional alumina tube
to optimize the calcination temperature. Figure 2 shows the effect of the
calcination temperature for the VFS membrane on the PV properties for 5 %
(mass fraction) EA aqueous solution at 290 K. The EA flux gradually increased
to
100
2
Flux / kg m-2 h-1
80 60 1 40
Separation factor / -
EA Water
20 0
0 380
430
480
530
Calcination temperature / K
Fig. 2 Effect of calcination temperature on the EA and water fluxes and
separation factor.
0.85 kg m – 2 h – 1 from 0.64 kg m – 2 h – 1 for calcination temperatures of up to 453 K,
and then drastically decreased at higher calcination temperatures. On the other
hand, the water flux was almost constant for calcination temperatures of up to
473 K, and became much higher for higher calcination temperatures . As a result,
the separation factor of the membrane calcined at 453 K was the highest among
all the membranes.
Previously, we suggested that the organic fl ux can be predicted using the
following equation [1]. J = a u e (b
u Ra)
M
(4) 㻌
where J is the flux of the organic compound [kg m – 2 h – 1 ], M is molecular weight [g mol - 1 ] of organic compound, and a (0.0193 kmol m - 2 h -1 ) and b (– 0.0245 MPa - 1 /2 ) are constants. Since R a [MPa 1 /2 ] between EA and VFS is 2.67, the EA flux of the VFS membrane is calculated to be 0.87 kg m - 2 h -1 . The EA
flux at 453 K agreed with the predicted value. This suggests that the calcination
temperature influences details of the membrane structure such as the surface
modification group, pore diameter. In addition, the EA flux of the VFS
membrane calcined at 453 K was higher than that of the hydrophobic silica
membrane with n-butyl or n-propyl groups [1].
Fig. 3 shows the contact angles
of the membranes calcined at temperatures between 383 K and 493 K. The
contact angle increased slightly to 86 r for the membranes calcined at
temperatures up to 453 K and was much less for membranes made with a higher
calcination temperatures. This behavior with the variation of the EA flux and
separation factor with the calcination temperature, shown in Fig. 2. The
thermogravimetric analysis curve of VTMS is shown in Fig. 4. A gradual
decrease was observed at around 430 K, and then a sharp decr ease occurred at
around 470 K. This behavior was similar to that observed for the effect of
calcination temperatures of VFS membranes on their contact angle .
It is
expected that the vinyl groups started to decompose at around 470 K and are
almost completely absent at around 530 K. These results also support the
postulate that the vinyl group plays an important role in the permeation of the
organic solvent.
The results of permporometry for VFS membranes calcined at 393 K, 453 K,
and 493 K are shown in Fig. 5. The He permeances of the VFS membrane
prepared at 393 K, at 453 K and 493 K were 7. 1 u 10 – 8 mol m – 2 s – 1 Pa – 1 8.2 u 10 – 8
mol m – 2 s – 1 Pa – 1 2.4 u 10 – 7 mol m – 2 s – 1 Pa – 1 at 293 K, respectively. The
dimensionless permeability of He of the VFS membrane calcined at 393 K was
broad in the Kelvin diameter range of 0 -4.5 nm, and the apparent average pore
diameter, which is the value at half the dimensionless permeability of He [21],
was 2.67 nm. By comparison, the VFS membrane calcined at 453 K has an
apparent average pore diameter of 0.96 nm and pores
Fig. 3 Effect of calcination temperature on the contact angles of water droplets
on the VFS membranes
1
TG / %
0
-1
-2 380
430
480
530
Temperature / K
Dimensionless Permieability of He / -
Fig. 4 Thermogravimetric analysis curve of VTMS
1 393 K 453 K 493 K
0.8
0.6
0.4
0.2
0 0
1
2
3
4
5
Kelvin diameter / nm
Fig. 5 Dimensionless permeability of He as a function of the Kelvin diameter for
VFS membranes prepared at 393 K, 453 K, and 493 K.
over 5.0 nm were not observed. The apparent average diameter of the VFS
membrane calcined at 493 K (1.12 nm) was slightly bigger than that at 453 K,
and some pores were larger than 4 nm. The affinity between the membrane and
condensable gas is very important to measure the pore size accurately by
permporometry. Tsuru et al. reported that the N 2 permeation of the hydrophobic
methylated SiO 2 was decreased by only 10% when water was used as
condensable vapor [22] even at Kelvin diameter of 10 nm. Since the VFS
membranes calcined at 495 K are expected to be hydrophilic nature, the pore
distribution derived from permporometry may not match the actual pore
distribution.
The EA flux of the VFS membrane calcined at 393 K was slightly lower than
that of the membrane calcined at 453 K (Fig. 2). There are two possible causes
for this lower flux, even with the pore size of the membrane at 393 K being
bigger than that of the membrane at 453 K. One is based on the lower contact
angle of the VFS membrane calcined at 393 K compared with that calcined at
453 K. A temperature of 393 K is lik ely to be insufficient for the condensation
reaction, and hydroxyl groups will remain in the pores. The other reason is based
on a lower porosity. The He permeance of the VFS membrane prepared at 393 K
was slightly lower than that of membrane prepared at 4 53 K. Even though the
VFS membrane prepared at 393 K has big pores, its porosity was similar to or
less than that of the membrane prepared at 453 K. Therefore, the difference in
hydrophobicity or porosity might affect the EA flux. The VFS membrane
calcined at 493 K has a low contact angle, and it showed separation factor (D=
5.5) due to the decomposition of vinyl groups. We concluded that the drastic
decrease in EA flux at around 480 K was caused by a decrease in the affinity
between EA and the VFS membrane.
3.2 Influence of the support
The effect of the substrate on the PV properties of VFS membrane was
investigated using an D-alumina hollow fiber instead of a tubular substrate, and
(a)
(b)
(c)
(d)
(e)
Fig. 6 Field emission scanning electron microscopy images of cro ss-sections of
VFS membranes on D-alumina hollow fibers with a J-alumina interlayer prepared with withdrawal rates of 0.5 (a), 1 (b), 3 (c), or 5 (d) mm s – 1 , and on the
conventional alumina substrate (f).
by changing the thickness of the J-alumina layer. The thickness of the J-alumina
layer was controlled by the withdrawal rate from the boehmite sol during
dip-coating. Cross-sectional images were obtained for these VFS membranes
(Fig. 6 (a)–(b)). In addition, the cross-sectional image of the VFS membrane
prepared on the D-alumina tube is shown in Fig .6 (e). The withdrawal rate of the substrate during dip-coating was used to control the thickness of the J-alumina layer between 1.0 Pm and 1.7 Pm. In addition, the J-alumina layer on the hollow
fiber was thinner than that on the tube despite the fact that the same withdrawal
rate was used. The thickness of the VFS layer on the tubular support was about
0.8 Pm. By comparison, the thickness of the VFS layer on the hollo w fiber coated with J-alumina (1.0–1.5 Pm) was about 0.1 Pm, and on that with a
J-alumina layer of 1.7 Pm was about 0.2 Pm. It is not clear why the J-alumina
layer thickness affected the VFS layer because the procedure for VSF layer
preparation was consistent. This difference might be within the normal
variability of VFS membrane thickness
120
5 EA
100
Water
80 3 60 2 40 1
Separation factor / -
Flux / kg m-2 h-1
4
20 0
0 0.9
1.1
1.3
1.5
1.7
Intermediate layer thickness / vm Fig. 7 Effect of intermediate layer thickness on the fluxes and separation factors.
Table 1 Comparison of fluxes and separation factors for VFS membranes on
tubular substrate and hollow fiber.
for this process. For 5 % (mass fraction) EA aqueous solution at 290 K, the PV
performances of VSF membranes on hollow fibers with different J-alumina layer
thicknesses are shown in (Fig. 7). It was found that the PV performance was not
affected by the thickness of the J-alumina layer. However, the EA and water
fluxes for the hollow fiber-supported VFS membrane were three times higher
than those obtained for the tube-supported VFS membrane. The separation facto r
of the hollow fiber-supported VFS membrane was almost the same as that of the
tube-supported VFS membrane.
Table 1 compares the PV performances for the VFS membranes prepared
using the hollow fiber and tubular support for EA, MA, and MEK solutions ( 5 %
mass fraction) at 295 K. The J-alumina layer was prepared with a withdrawal rate of 1 mm s – 1 . MA and MEK fluxes for the tube -supported VFS membrane
were estimated using eq. 4 [1]. The estimated fluxes of MA and MEK were 0.61
kg m – 2 h – 1 and 0.44 kg m – 2 h – 1 . Although the estimated and measured values for
the MA flux were slightly different, the estimated MEK flux was very close to
the measured value. From these results , it was concluded that eq. 4 can be used
to determine the most suitable membrane struc ture and the flux for the
membrane on a conventional tubular support. However, modification of eq. 4 is
needed to estimate fluxes for hollow fiber-supported membranes. To do this, the
contribution of the substrate to the flux needs to be investigated through further
experiments. The EA and MEK fluxes for the hollow fiber -supported VFS
membrane
were
also
three
times
higher
than
th ose
obtained
for
the
tube-supported VFS membrane, and the separation factors were similar. These
results show that the affinity properties of the hollow fiber-supported VSF
membrane were similar to those of the tube -supported VFS membrane under the
same conditions.
Table 2 Comparison of gas permeance properties for tubular and hollow fiber
substrates
6XSSRUWW\SH
7KLFNQHVV
3 HUP HDQFH> P ROP
V 3 D @
>P P @
+H
1
&+
7XEXODOU
+ ROORZ ILEHU
Table 2 shows the single gas permeation results for helium, nitrogen, and
methane for the hollow fiber and tubular support. The helium, nitrogen and
methane permeances of the hollow fiber was about 9 times higher than those of a
tubular support. On the other hand, the helium permea nce of the hollow fiber
with the J-alumina layer decreased slightly with the thickness of the J-alumina layer increased as shown in Fig.8. For a J-alumina thickness of 1.0 Pm and 1.7 Pm, the helium permeance was 2.3 u 10 – 5 mol m – 2 s – 1 Pa – 1 and 2.0 u 10 – 5 mol m – 2
s – 1 Pa – 1 ,respectively. The relationship between the permeance and the
reciprocal of square root of the quantity of the absolute temperature times
molecular weight is shown in Fig.9. The permeance vs conditions of all supports
follows closely predictions based on Knudsen mechanism [23].
The He permeance results for the hollow fiber-supported VFS membrane
with J-alumina layers of different thicknesses are also shown in Fig. 8. There was only a small decrease in He permeance with the thickness of J-alumina layer.
Interestingly, the He permeance of VFS membranes on the hollow fibre was
similar to that obtained for the tube -supported membrane, despite the presence
of a thicker VFS layer. This result also suggested that the variation in
thicknesses of the D-alumina, J-alumina, and VFS layers is too small to affect
the He permeance. In any case, for gas permeation of the VFS membrane,
differences in the support did not affect the He permeance. This means that mass
transfer through the D-alumina substrate would not be the rate determining
process. However, the EA flux of the VSF membrane did improve when the
hollow fiber substrate was used instead of the tubular substrate (Table 1).
Therefore, mass transfer through the D-alumina substrate greatly affects the EA
flux. The transfer of EA molecules
Fig. 8 Effect of J-alumina thickness on He permeance of the alumina substrate
and VFS membrane
Fig.9 Relation between the reciprocal of square root of the quantity of the
temperature times molecular weight and gas permeances.
through the VFS membrane involves adsorption of EA molecules on the VFS
layer, diffusion of EA as a liquid, evaporation of the EA molecule, and then
diffusion as a vapor. Because the thicknesses of the J-alumina and VFS layers on
the hollow fiber did not affect the EA flux, EA molecules might adsorb on the
VFS layer and occupy pores of the J-alumina. Therefore, diffusibilit y of organic compounds through D-alumina is improved using the hollow fiber compared with
the tubular support.
The effect of the EA mass fraction in the feed solution on the PV is shown
in Fig. 10. Is was found that the EA fluxes with a 1 % mass fraction of EA in t he
feed solution were similar for VFS membranes prepared on hollow fiber and
tubular support. Although the EA fluxes increased with the EA mass fraction in
the feed solution, the rate of increase of the EA flux for the hollow
fiber-supported VFS membrane was larger than that for the tube -supported
membranes. The EA flux of the hollow fiber-supported VFS membrane was
about 3.5 kg m – 2 h – 1 , which was three times that of the tube-supported membrane,
and the separation factor was 85. These values are higher than the upper bound
of the relationship between flux and separation factor of hydrophobic ceramic
membranes reported in the past three years (Fig. 11). By contrast, the water
fluxes of the VFS membranes on both the hollow fiber and tubular support
6
120 EA Water Open: Tube Close: Hollow fiber
100
4
80
3
60
2
40
1
20
0
Separation factor / -
Flux / kg m-2 h-1
5
0 0
2
4
6
8
EA concentration / wt%
Fig. 10 Effect of EA mass fraction on the PV performance of VFS membranes
prepared on tubular and hollow fiber substrates.
Separation factor / -
100
Tube Hollow fiber
80 60 40 20 0 0
2
4 Flux / kg m-2 h-1
6
8
Fig 11 Relationship flux and separation factor of ceramic hydrophobic
membranes for EA separation from aqueous solution reported from 2015 to 2017.
Open: Temp.= 308 K, EA conc. 1-2 % (mass fraction) [7, 23-25], Close: Temp.=
295 K, EA conc. =3-7 % (mass fraction) [1].
were constant at about 0.6 kg m – 2 h – 1 and 0.2 kg m – 2 h – 1 , respectively. The
hydrophobic silica membranes prepared on tubular support s have water fluxes
independent of the nature of the functional groups, and the type and
concentration of organic compound in the feed solut ion [1]. The water flux for
the hollow fiber-supported VFS membrane was higher than that of the
tube-supported VFS membrane because of the reduced thickness of the substrate
or VFS layer. If silica and J-alumina layers does not depend on EA flux
regardless of types of support as shown in Fig.7, the ratio of fluxes equals the
ratio of reciprocals of support resistances. Although we expected that the EA
flux for the hollow fiber-supported VFS membrane would increase in proportion
to the EA flux for the tube-supported VFS membrane, this did not occur.
Therefore, the EA flux must be affected by different processes. In other words,
EA flux was determined by the adsorption process with an EA mass fraction of
1 % and 3% in the feed solution, and then the influence of the substrate
increased as the EA mass fraction increased. For EA mass fraction of 5 % and
7% in the feed solution, the EA fluxes of hollow fiber-supported VFS membrane
were three times higher than the EA fluxes of tube -supported VFS membrane,
which was approximately same with the ratio of water fluxes of VFS membranes
on hollow fiber to tube-supported. Therefore, it is suggested that the resistance
of hollow fiber is three times lower than that of tube support.
Conclusions
The PV properties of VFS membranes for the separation of EA from water were
investigated. A VFS membrane prepared on a tubular alumina support at 453 K
gave the highest EA flux and separation factor. The vinyl group decomposes at
temperatures above 453 K, and this greatly increases the water flux and
decreases the EA flux. A VFS membrane prepared on an alumina hollow fiber at
the optimal temperature was used for PV tests of EA/water, MA/water, and
MEK/water mixtures with organic mass fractions of 5 %. The fluxes with
obtained with this VFS membrane are three times those obtained with a VFS
membrane prepared on a conventional tubular support. However, for PV of a 1 %
(mass fraction) EA solution, the EA fluxes of VFS membranes on hollow fiber
and tubular support are similar. The EA flux of a hollow fiber-supported VFS
membrane (3.5 kg m – 2 h – 1 ) was three times that of a tube-supported VFS
membrane, and the separation factor was 85 for PV of a 7 % (mass fraction) EA
solution. These results suggest that the adsorption rate of EA on the VFS has a
major influence on the EA flux in the 1 -3% (mass fraction) solution, and the
diffusivity of the EA molecule through the alumina support has a major
influence on the EA flux in the 5-7% (mass fraction) solution.
Acknowledgments
This work was partially supported by MEXT-Supported Program for the
Strategic Research Foundation at Private Universities, 2012 –2016 and 㻌 Kansai
University Overseas Research Program, 2016.
Literature Cited [1] S. Araki, D. Gondo, S. Imasaka, H. Yamamoto, Permeation properties of organic compounds from aqueous solutions through hydrophobic silica membranes with different functional groups by pervaporation, J. Membr. Sci. 514 (2016) 458 -466. [2] D.H. Park, N. Nishiyama, Y. Egashira, K. Ueyama, Separation of organic/water mixtures with silylated MCM -48 silica membranes, Microporous and Mesoporous Materials, 66 (2003) 69 -76. [3] K. Liu, Z. Tong, L. Liu, X. Feng, Separation of organic compounds from water by pervaporation in the production of n -butyl acetate via esterification by reactive distillation J. Membr. Sci. 256 (22005) 193-201 [4] H. Zhou, Y. Su, X. Chen, J. Luo, Y. Wan, High -performance PDMS membranes for pervaporative removal of VOCs from water: The role of alkyl
grafting,
J.
Appl.
Polym.
Sci.
2016,
in
press
(DOI:
10.1002/APP.43700) [5] P. Shao. R.Y.M. Huang, Polymeric membra ne pervaporation, J. Membr. Sci. 287 (2007) 162-179. [6]
Anastasia
Penkova,
Galina
Polotskaya,
Alexander
Toikka,
Pervaporation composite membranes for ethyl acetate production,
Chemical Engineering and Processing 87 (2015) 81 –87 [7] J. Kujawa, S. Cerneaux, W. Kujawski, Highly hydrophobic ceramic membranes applied to the removal of volatile organic compounds in pervaporation, Chemical Engineering Journal 260 (2015) 43 –54 [8] H. Yuan, J. Ren, X. Ma, Z. Xu, Dehydration of ethyl acetate aqueous solution by pervaporation using PVA/PAN hollow fiber composite membrane, Desalination 280 (2011) 252 –258 [9]
S.
Latthe,
H.
Imai,
V.
Ganesan,
A.
Venkateswara
Rao,
Ultrahydrophobic silica films by sol –gel process., J Porous Mater 1 7 (2010) 565-571. [10] S. Xia, Y. Peng, H. Lu, Z. Wang, The influence of nanoseeds on the pervaporation performance of MFI -type zeolite membranes on hollow fibers, Microporous and Mesoporous Materials222 (2016) 128 -137 [11] X. Wang, Y. Chen, C. Zhang, X. Gu, N. Xu, Preparation and characterization of high-flux T-type zeolite membrane supported on YSZ hollow fibers, J. Membr. Sci. 455 (2014) 294 -304. [12] X. Shu, X. Wang, Q. Kong, X. Gu, N. Xu, High -Flux MFI Zeolite Membrane
Supported
on
YSZ
Hollow
Fiber
for
Separation
of
Ethanol/Water, Ind. Eng. Chem. Res. 51 (2012) 12073 -12080. [13] Z. Wang, Q. Ge, J. Shao, Y. Yan, High Performance Zeolite LTA Pervaporation
Membranes
on
Ceramic
Hollow
Fibers
by
Dipcoating-Wiping Seed Deposition, J. Am. Chem. Soc. 131 (2009) 6910–6911. [14] J. Shao, Z. Zhan, J. Li, Z. Wang, K Li, Y. Yan, Zeolite NaA membranes supported on alumina hollow fibers: Effect of support resistances on pervaporation performance, J . Membr. Sci. 451 (2014) 10–17. [15] H. Yan, N. Ma, Z. Zhan, Z. Wang, Fabrication of zeolite NaA membranes on hollow fibers using nano -sized seeds exfoliated from mesoporous zeolite crystals, Microporous and Mesoporous Materials 215 (2015) 244-248. [16] Z. Dong, G. Liu, S. Liu, Z. Liu, W. Jin, High Performance ceramic hollow fiber supported PDMS composite pervaporation membrane for bio-butanol recovery, J. Membr. Sci. 450 (2014) 38 –47. [17] S. Xia, Y. Peng, Z. Wang, Microstructure manipulation of MFI -type zeolite membranes on hollow fibers for ethanol –water separation, J. Membr. Sci. 498 (2016) 324-335. [18] Y. Liu, X. Wang, Y. Zhang, Y. He, X. Gu, Scale -up of NaA zeolite membranes on α-Al 2 O 3 hollow fibers by a secondary growth method with vacuum seeding, Chinese Journal of Chemical Engineering 23 (2015) 1114–1122. [19] S. Fan, J. Liu, F. Zhang, S. Zhou, F. Sun, Fabrication of zeolite MFI
membranes supported by a-Al2O3 hollow ceramic fibers for CO 2 separation, J. Mater. Res. 28 (2013) 1870 -1876 [20] J. Shao, Q. Ge, L. Shan, Z. Wang, Y. Yan, Influences of Seeds on the Properties of Zeolite NaA Membran es on Alumina Hollow Fibers, Ind. Eng. Chem. Res. 50 (2011) 9718–9726. [21] T Tsuru, T. Hino, T. Yoshioka, M. Asaeda, Permporometry characterization of microporous ceramic membranes, J. Membr. Sci., 186 (2001) 257-265. [22] T. Tsuru, T. Nakasuji, M. Oka, M. Kanezashi, T. Yoshioka, Preparation of hydrophobic nanoporous methylated SiO 2 membranes and application to nanofiltration of hexane solutions , J. Membr. Sci. 384 (2011) 149-156. [23] J. Fan, H. Ohya, T. Suga, H. Ohashi, K. Yamashita, S. Tsuchiya, M. Aihara,
T.
Takeuchi,
Y.
Negishi,
High
flux
zirconia
composite
membraene for hydrogen separation at elevated temperature, J. Membr. Sci. 170 (2000) 113-125. [23] W. Kujawski, J. Kujawa, E. Wierzbowska, S. Cerneaux, M. Bryjak, J. Kujawski, Influence of hydrophob ization conditions and ceramic membranes pore size on their properties in vacuum membrane distillation of water –organic solvent mixtures, J. Membr. Sci. 499 ( 2016) 442 -451 [24] J. Kujawa, S. Cerneaux, W. Kujawski, Removal of hazardous volatile organic
compounds from water by vacuum pervaporation with hydrophobic ceramic membranes, J. Membr. Sci. 474 (2015) 11 -19 [25] J. Kujawa, S. Al-Gharabli, W. Kujawski, K. Knozowska, Molecular Grafting of
Fluorinated
and
Nonfluorinated
Alkylsiloxanes
on
Various
Ceramic
Membrane Surfaces for the Removal of Volatile Organic Compounds Applying Vacuum Membrane Distillation , ACS Appl. Mater. Interfaces, 9 (2017) 6571– 6590
㻌
Table and figure legends
Fig.1 Photograph of D-alumina (a) tubular support and (b) hollow fiber
Fig.2 Effect of calcination temperature on the EA and water fluxes and
separation factor
Fig. 3 Effect of calcination temperature on contact angle of water droplets on the
VFS membranes
Fig.4 TG curve of VTMS
Fig.5 Dimensionless permeability of He as a function of the Kelvin diameter for
VFS membranes prepared at 393 K, 453 K and 493 K.
Fig.6 FE-SEM images of cross-sections of VFS membranes on a-alumina hollow
fibre with J-alumina interlayer prepared by pulling rate of 0.5 (a), 1 (b), 3 (c)
and 5 (d) mm s-1, and on the conventional alumina substrate (f).
Fig.7 Effect of intermediate layer thickness on the fluxes and separation factor.
Fig.8 Effect of intermediate layer thickness on He permeance of alumina
substrates and VFS membranes.
Fig.9 Relationship between the reciprocal of square root of the quantity of the
temperature times molecular weight and gas permeances.
Fig. 10 Effect of EA mass fraction on the PV performance of VFS membranes
prepared on tubular and hollow fiber substrates.
Fig 11 Relationship flux and separation factor of ceramic hydrophobic
membranes for EA separation from aqueous solution reported from 2015 to 2017.
Open: Temp.= 308 K, EA conc. 1-2 % (mass fraction) [7, 23-25], Close: Temp.=
295 K, EA conc. =3-7 % (mass fraction) [1].
Table 1 Comparison of fluxes and separation factors for VFS membranes on
tubular substrate and hollow fiber.
Table 2 Comparison of gas permeances for tubular substrate and hollow fibre
Tubular support
Hollow fiber
*UDSKLFDODEVWUDFW
EA concentration / wt%
0.0
8
0
6
20.0
1
4
40.0
2
2
60.0
3
100.0
120.0
80.0
0
Substrate type Open: Tube Close: hollow fiber
4
5
6
Separation factor / -
EA flux / kg m-2 h-1