Contents lists available at ScienceDirect [622676]

Contents lists available at ScienceDirect
Journal of Membrane Science
journal homepage: www.elsevier.com/locate/memsci
The role of pre-evaporation in the preparation process of EVOH
ultrafiltration membranes via TIPS
Zheng Sun, Zhensheng Yang, Zhiying Wang, Chunli Li
National-Local Joint Engineering Laboratory for Energy Conservation of Chemical Process Integration and Resources Utilization, School of Chemical Engineering and
Technology, Hebei University of Technology, Tianjin 300130, China
ARTICLE INFO
Keywords:Poly(ethylene- co-vinyl alcohol) (EVOH)
Ultrafiltration membrane
Pre-evaporationThermally induced phase separation (TIPS)Co-solventABSTRACT
The poly(ethylene- co-vinyl alcohol) (EVOH) ultrafi ltration membrane is prepared via pre-evaporation combined
thermally induced phase separation (TIPS), using a novel co-solvent that is a mixture of sulfolane and 1,3-
propanediol. The porous surface is mainly caused by the low cooling rate of pre-evaporation stage. Based on theshorter pre-evaporation time, the evaporation of 1,3-propanediol provides a foundation for asymmetric struc-
ture. Subsequently, the quenching eff ect of water bath forms a dense layer below the porous surface. The
compact degree of dense layer increases first and then decreases with the increase of pre-evaporation time ( τ)o r
pre-evaporation temperature (T) or the mass fraction of 1,3-propanediol in co-solvent ( β). Therefore, the re-
jection of bovine serum albumin (BSA) increases first and then decreases as τor T or βincreases. Meanwhile, the
connectivity of membrane structure can also be adjusted by changing β. Finally, the EVOH ultra filtration
membrane prepared from τ= 20 s, T = 25 °C and β= 15% achieves relatively high BSA rejection of 98% and
relatively large pure water flux of 302 L m
−2h−1bar−1.
1. Introduction
Poly(ethylene- co-vinyl alcohol) (EVOH) is a crystalline random co-
polymer with vinyl alcohol and ethylene segments. EVOH membrane is
employed successfully in several fields; e.g. as a filtration membrane in
fine separation processes, as electrolyte separator in fuel cells, etc.,
because of its outstanding anti-pollution properties, superior chemical
resistance and good permeability, and that it can be applied to plas-
mapheresis because of its good blood compatibility and wettability
[1–7].
EVOH porous membranes (ultra filtration membrane, micro filtration
membrane) have mainly been prepared by non-solvent induced phase
separation (NIPS) method [8–13]. In the NIPS process, a homogeneous
polymer solution is immersed in a non-solvent bath, the penetration ofthe non-solvent into the polymer solution induces phase separation. An
alternative way to produce EVOH membranes is TIPS process [14–18].
In the TIPS process, because the phase separation is thermally inducedrather than non-solvent exchange induced, there are fewer variables to
be controlled [19]. In addition, TIPS process also has many unique
advantages such as high reproducibility, low tendency to form defects,high porosity, and the ability to form interesting microstructures with
narrow pore size distribution [19,20] .
The permeation and rejection properties of the EVOH porous
membranes prepared via TIPS are summarized in Table 1. As shown inTable 1, the molecular weight cuto ffof the EVOH membranes is rela-
tively high. However, the molecular weight cuto ffof the commercia-
lized ultra filtration membrane is usually less than 100,000 (the
equivalent diameter is about 11 nm) [23]. Therefore ，the rejection
performance of the EVOH ultra filtration membranes prepared via TIPS
need to be further improved. In addition, Table 1 also summarizes the
recent studies on other polymeric ultra filtration membranes prepared
via TIPS as well as their permeation and rejection properties. Ob-
viously, their permeation and rejection properties also need to be im-
proved. Finally, a conclusion that TIPS method may be di fficult to
prepare ultra filtration membranes with high flux and high rejection
performance is obtained.
In order to obtain the ultra filtration membrane with high flux and
high rejection performance, it is necessary to prepare an asymmetric
membrane with a thinner dense layer and a loose sponge structure [6].
Pre-evaporation is consistent with the above idea. Although a limitedamount of e fforts have been made to obtain asymmetric membrane via
pre-evaporation combined TIPS method [28–30],
the study of asym-
metrical ultra filtration membrane is lesser [31]. The reason may be that
the solvent evaporation stage (the cooling rate is lower) is always ac-companied by the occurrence of slow solid-liquid phase separation for
TIPS process, which is disadvantageous for the formation of dense
layer.
In the present study, the phase diagrams of EVOH/sulfolane/1,3-
https://doi.org/10.1016/j.memsci.2018.06.003
Received 9 February 2018; Received in revised form 5 May 2018; Accepted 4 June 2018E-mail address: zsyang211@163.com (Z. Yang).Journal of Membrane Science 563 (2018) 238–246
Available online 04 June 2018
0376-7388/ © 2018 Elsevier B.V. All rights reserved.
T

propanediol system were measured, which provided crucial informa-
tion about the phase behavior and the basis to control membrane
morphology. EVOH ultra filtration membrane was prepared via pre-
evaporation combined TIPS method ( Scheme 1), by using a novel co-
solvent that is a mixture of sulfolane and 1,3-propanediol. Here, the
evaporation of 1,3-propanediol could not only increase the polymer
concentration but also change the composition of co-solvent. In addi-
tion, the e ffects of pre-evaporation time ( τ), pre-evaporation tempera-
ture (T) and the mass fraction of 1,3-propanediol in co-solvent ( β)o n
the morphology and performance of the resulting EVOH ultra filtration
membranes were also investigated.
2. Experimental
2.1. Materials
EVOH containing 38 mol% ethylene (EVOH38) was purchased fromKuraray Co. (the degree of polymerization is 960). BSA (MW =
67,000 g/mol), sulfolane (reagent purity) and 1,3-propanediol (reagent
purity) were purchased from Shanghai Chemical Reagents Co. Non-
woven fabric was purchased from Shanghai Tianlue Textile New
Materials Co. All chemicals were used without further puri fication.
2.2. Determination of phase diagrams
The mass fraction of EVOH in the casting solution was 23%, the
mass fraction of 1,3-propanediol in co-solvent ( β) varied from 0% to
20%. The mixture of EVOH/sulfolane/1,3-propanediol was fed into a
flask, heated at 170 °C, and stirred for about 5 h until a homogeneous
melted solution was obtained. The above operations were carried outunder a nitrogen atmosphere. The melted solution was poured into li-
quid nitrogen to solidify. The specimens were collected and kept in
refrigerator for further use.
An inverted microscope (Ningbo ShunYu instrument co., XD30,
China) with a hot stage (Linkam, THMS600, UK) and a temperature
controller (Linkam, TMS92 UK) was used to measure liquid-liquid
phase separation temperature of resulting specimens. Specimen covered
with cover glass was mounted on the hot stage, and the edge of the
bottom cover slip was sealed with Tefl on tape and vacuum grease to
minimize diluent loss by evaporation during the heating or coolingprocess. Liquid-liquid phase separation equilibrium temperature was
measured according to Kim's method [32]. The specimen was heated
rapidly to 170 °C, annealed at a constant temperature in the vicinity ofthe expected phase boundary for 10 min. The temperature at which the
optical image first started to change during annealing was taken as the
liquid-liquid phase separation temperature. With the annealing tem-perature rising, the induction time would be longer.
A Perkin –Elmer Diamond DSC was used to determine dynamic
crystallization temperature of resulting specimens. The sealed alu-
minum DSC pan containing 3 –5 mg specimen was heated to 170 °C and
maintained at this temperature for 10 min. It was then cooled down atthe rate of 20 °C/min. The onset of the exothermic peak during the
cooling was taken as the dynamic crystallization temperature.Table 1
The published studies on EVOH and other polymeric membranes prepared via TIPS.
Authors Polymer MF/UF Rejection PWFs (L m−2h−1bar−1)
Zhou [7] EVOH MF 21.7% ( γ-globulin, Mw= 160,000) 170.44
Wu[21] EVOH MF 94% (Dextran, Mw= 2000,000) 300
Matsuyama [22] EVOH UF 90% (Ferritin, Mw= 440,000) 6.3
Matsuyama [18] EVOH UF 90% ( Latex particle, Diameter, 100 nm) 1.8
Shang [14] EVOH UF 92% ( Polystyrene, Diameter, 20 nm) 50
Li[24] PVDF UF 91.7% (BSA, Mw= 67,000) 182.6
Jin[25] PVC UF 95% (BSA, Mw= 67,000) 5
Pang [26] CA UF 96.1% (Dextran, 20 kDa) 0.13
Xu[27] PVDF UF 90.8% (BSA, Mw= 67,000) 164.5
Scheme 1. Schematic for the preparation process of EVOH ultra filtration membrane.
Fig. 1. Relationship between phase behavior and co-solvent composition.
(Polymer concentration = 23 wt%).Z. Sun et al. Journal of Membrane Science 563 (2018) 238–246
239

2.3. Membrane preparation
The homogeneous melted solution of EVOH/sulfolane/1,3-propa-
nediol system was obtained, similar to that described in 2.2.
Subsequently, the homogeneous solution was cooled to 160 °C and
vacuum degassed.
The non-woven fabric was fastened on a stainless steel plate heated
to 140 °C, the homogeneous solution was casted on the non-woven
fabric by a self-made casting knife heated to 160 °C. Here, as variables,
τvaried from 0 s to 25 s, and T varied from 25 °C to 70 °C. After pre-
evaporation, both the liquid membrane and the stainless steel platewere quickly immersed into coagulation bath (deionized water, 40 °C)
for 40 min. The nascent membrane was soaked in deionized water
(30 °C) for about 48 h, in which the diluents, 1,3-propanediol and sul-
folane, were extracted from the obtained membrane. The wet mem-
brane was dried at room temperature.
2.4. Characterization of EVOH ultra filtration membrane
2.4.1. Membrane morphology
The EVOH membrane was freeze-dried with a freeze dryer. The dry
membrane was fractured in liquid nitrogen and treated with Au/
Pdsputtering. The cross section and outer layer of the membrane were
observed by a Philips XL30 scanning electron microscope (SEM) underan accelerating voltage of 20 kV.
2.4.2. Pure water flux
The pure water flux was measured via a self-made device using a
cross-flow mode. Each membrane sample was initially pressurized at
0.15 MPa for approximately 15 min. The pressure then decreased to
0.1 MPa, and began to collect water. The pure water flux was calculated
using the following equation:
=JV
AΔtw
whereJw(L m−2h−1) is the pure water flux,V(L) represents the volume
of permeated water, A(m2) is the surface area of the membrane and
Δt(h) represents filtration time.
2.4.3. Rejection performance
The rejection performance of EVOH ultra filtration membrane was
measured using BSA (1 g/L, pH7.0). The rejection of BSA was calculated
by the following equation:
=− ×RCC (1 / ) 100 % pf
whereRis the rejection of BSA, Cp(g/L) and Cf(g/L) are the BSA con-
centrations of permeation and feed solutions, respectively. They weremeasured with an ultraviolet-spectrophotometer (UV-9200) at a
Fig. 2. SEM images of EVOH membranes, β= 15%, T = 25 °C. ((1) Surface; (2) cross Section; (3) enlarged cross section; (A) τ= 0 s; (B) τ= 10 s; (C) τ= 20 s; (D)
τ= 25 s).Z. Sun et al. Journal of Membrane Science 563 (2018) 238–246
240

wavelength of 280 nm. Each sample was measured for three times.
2.4.4. XRD analysis
The crystal structure of EVOH membrane was determined by a wide
angle X-ray di ffractometer (XRD, D8 Advance, Bruker, Germany). The
diffraction patterns were obtained at 40 kV and 40 mA, with a scanning
rate of 0.2°/s.
3. Results and discussion
3.1. Determination of ternary phase diagram
It can be seen from Fig. 1, when β≤10%, the cloud point tem-
perature ( Tcloud) of EVOH38/sulfolane/1,3-propanediol system is higher
than the polymer crystallization temperature ( Ton), which means that
L–L phase separation will occur prior to polymer crystallization. Fur-
thermore, both Tcloudand Tondecrease as βincreases, and the decrease
ofTcloudis more obvious than that of Ton. As a result, with the increase
ofβ,Tcloud is getting closer to Ton. When β= 15% or 20%, the cloud
point of the ternary system cannot be successfully observed. This is
because Tcloudhas now dropped below Ton, the polymer crystallization
will happen prior to L –L phase separation, which interferes with the
observation of the cloud point. The results show that the interaction
between polymer and co-solvent can be adjusted and the phase beha-
vior of the system can also be controlled by changing β, which provide
basis to control membrane morphology.
3.2. Membrane morphology
For constant β= 15% and T = 25 °C, the e ffects of τon membrane
morphology were shown in Fig. 2 . Both the contribution of particle
structure to membrane morphology and the size of the particles in-
crease as τincreases. The saturated vapor pressures of sulfolane and
1,3-propanediol at 150 °C are 1.93 KPa and 13.33 KPa, respectively[33]. The evaporation of solvent, especially 1,3-propanediol, not only
increases the polymer concentration, which is conducive to the occur-rence of polymer crystallization [19,20] , but also decreases β, which
can reduce the contribution of polymer crystallization to phase se-paration (see Fig. 1). On the other hand, the low cooling rate of pre-
evaporation stage is also advantageous for the occurrence of polymercrystallization [34,35] . As a result, due to the combined e ffect of in-
creased polymer concentration, decreased β, and lower cooling rate
during pre-evaporation stage, the contribution of particle structure
originating from polymer crystallization to membrane morphology in-
creases with increasing τ. Meanwhile, the increase of τmeans that the
coarsening time of particles is prolonged, so that the size of the particlesalso increases.
As can be seen from Fig. 3, with the increase of pre-evaporation
temperature (T), the contribution of particle structure to membrane
morphology decreases first and then increases. The saturated vapor
pressures of 1,3-propanediol at 25 °C, 50 °C and 70 °C are 0.11 KPa,0.125 KPa and 0.336 KPa, respectively [33]. When T changes from
25 °C to 50 °C, there is no signi ficant change in the saturated vapor
pressure of 1,3-propanediol, that is, increased T only slightly promotes
the evaporation of 1,3-propanediol. According to the initial composi-
tion of casting solution, it can be concluded that the slight increase in
evaporation of 1,3-propanediol has a greater e ffect on the composition
of co-solvent than it does on the polymer concentration. That is to say,the polymer concentration will increase slightly and the βwill decrease
significantly, which implies that the in fluence of βon polymer crys-
tallization may be dominant. Also, the decrease in βis detrimental to
the occurrence of polymer crystallization (see Fig. 1). Therefore, the
contribution of particle structure originating from polymer crystal-lization to the membrane morphology decreases. Meanwhile, based on
the fact that polymer crystallization is inhibited, a slight increase in
polymer concentration makes the surface more dense (see Fig. 3B1).
When T changes from 50 °C to 70 °C, saturated vapor pressures of 1,3-propanediol increases to three times that of the original, which means
Fig. 3. SEM images of EVOH membranes, τ=1 0s , β= 15%. ((1) Surface; (2) cross Section; (3) enlarged cross section; (A) T = 25 °C; (B) T = 50 °C; (C) T = 70 °C).Z. Sun et al. Journal of Membrane Science 563 (2018) 238–246
241

that the polymer concentration will increase signi ficantly. Also ，no
matter how the co-solvent composition changes, polymer crystallization
still dominates in phase separation of the system as long as the polymer
concentration is high enough. Therefore, when T = 70 °C, the increase
in the contribution of particle structure to membrane morphology maybe due to the signi ficant increase in polymer concentration. Meanwhile,
the higher polymer concentration brings about a larger nucleation
driving force. As a result, the nucleus density increases, which reduces
the particle size (see Fig. 3C1).
It can be seen from Fig. 4A3-D3, L –L phase separation occurs prior
to polymer crystallization in all the samples. As shown in Fig. 1, when
β= 5% or 10%, the Tcloud is higher than the Tonand the range between
Tcloud and Tonis narrow. As a result, L –L phase separation will occur
prior to polymer crystallization, after the system has been cooled to Ton,
the polymer crystallization can also occur until the system is fully so-lidifi ed. However, when β= 15% or 20%, the T
cloudhas dropped below
theTon(see Fig. 1), that is, polymer crystallization will happen prior to
L–L phase separation, which is contrary to the fact. The reason is that
the cooling rate where the membranes were prepared is much greaterthan that showed in the phase diagrams. Furthermore, dynamic crys-
tallization temperature is signi ficantly decreased as the cooling rate
increases [34,35] . Therefore, as shown in Fig. 5, the actual crystal-
lization temperature is less than the measured one. When the phase
separation of the system occurs according to the cooling path (1), the
actual phase separation way will change, that is, although the measured
T
onis above the Tcloud,L–L phase separation will still happen prior to
polymer crystallization.
For constant τ= 20 s and T = 25 °C, the e ffects of βon membrane
morphology were shown in Fig. 4. The contribution of particle structure
Fig. 4. SEM images of EVOH membranes, τ= 20 s, T = 25 °C. ((1) Surface; (2) cross Section; (3) enlarged cross section; (A) β= 5%; (B) β= 10%; (C) β= 15%; (D)
β= 20%).
Fig. 5. Schematic of TIPS phase behavior.Z. Sun et al. Journal of Membrane Science 563 (2018) 238–246
242

to membrane morphology increases with the increase of β, as the same
as the size of the particles. According to Fig. 1, with the increase of β,
the decline of Tcloud is more obvious than that of Ton, which means that
the contribution of polymer crystallization to phase separation in-
creases. Therefore, the contribution of particle structure to membrane
morphology increases. As we all know, the gelation process starting
from the L –L phase separation is a faster process while polymer crys-
tallization process is a relatively slow one [36]. Furthermore, with the
increase of β, the contribution of polymer crystallization to phase se-
paration increases (see Fig. 1). Therefore, the coarsening time of the
system increases as βincreases. As a result, the size of the particles
increases as βincreases.3.3. Rejection performance
It can be seen from Fig. 6 b, when β= 15%, τ= 10 s and T = 50 °C,
the EVOH membrane presents the highest BSA rejection of 99%, whichindicates that the molecular weight cuto ffof the resulting EVOH ul-
trafiltration membrane is less than 67,000. Also, in Fig. 6 b, the
minimum rejection rate of BSA is also as high as 85%. However, as
shown in Fig. 3A1-C1, the surfaces of the resulting asymmetric mem-
branes show a porous structure or a particle accumulation structure,which cannot e ffectively reject the BSA. Combining with Fig. 3A2-C2,
we can conclude that a dense layer with a depth of about 10 µm existedbelow the porous surface is responsible for the rejection of BSA.
The porous surface is mainly caused by the slow solid-liquid phase
Fig. 6. BSA rejection of EVOH membranes. (a) β= 15%, T = 25 °C; (b)
τ=1 0s , β= 15%; (c) τ= 20 s, T = 25 °C.
Fig. 7. The pure water flux of EVOH membranes. (a) β= 15%, T = 25 °C; (b)
τ=1 0s , β= 15%; (c) τ=2 0s ,T=2 5° C .Z. Sun et al. Journal of Membrane Science 563 (2018) 238–246
243

separation in the pre-evaporation stage. Meanwhile, the porous surface
layer is so thin that it is di fficult to be observed by SEM because the τis
very small. The dense layer is caused by the higher polymer con-centration originating from the evaporation of 1,3-propanediol and the
quenching e ffect of water bath [17,18] .
As shown in Fig. 6a, the rejection of BSA first increases and then
decreases with the increase of τ, and attains its maximum value (98%)
atτ= 20 s. On the one hand, the evaporation of 1,3-propanediol in-
creases as τincreases. Based on the shorter pre-evaporation time, the
polymer concentration in the top layer of the liquid membrane willincrease as τincreases, which is advantageous for the formation of
dense layer. On the other hand, due to the combined e ffect of increased
polymer concentration, decreased β, and lower cooling rate during pre-
evaporation stage, the contribution of polymer crystallization to phase
separation increases with increasing τ(see Fig. 2 ). Also, the particle
structure originating from polymer crystallization is disadvantageousfor the formation of dense layer. As a result, the compact degree of
dense layer may be best when τ= 20 s. Therefore, the rejection of BSA
first increases and then decreases as τincreases, and attains its max-
imum value at τ=2 0s .
When T changes from 25 °C to 50 °C ( τ=1 0s , β= 15%), the re-
jection of BSA increases from 84% to 99% (see Fig. 6b). Based on the
fact that polymer crystallization is inhibited, a slight increase inpolymer concentration makes membrane surface and dense layer more
dense (see Fig. 3), which brings about an increase in the rejection of
BSA. When T changes from 50 °C to 70 °C, the signi ficant increase in
polymer concentration increases the contribution of polymercrystallization to phase separation. Also, the particle structure origi-nating from polymer crystallization is disadvantageous for the forma-
tion of dense layer. Therefore, the rejection of BSA drops again.
When τ= 20 s and T = 25 °C, the increase of βalso enhances the
evaporation of 1,3-propanediol. Similarly, the polymer concentration in
the top layer of the liquid membrane increases as βincreases, which is
advantageous for the formation of dense layer. On the other hand, thecontribution of polymer crystallization to phase separation increases
with the increase of β(see Fig. 1), which is disadvantageous for the
formation of dense layer. Finally, the compact degree of dense layermay be best at β= 15%. Accordingly, the rejection of BSA first in-
creases and then decreases with the increase of β, and attains its
maximum value at β=
15% (see Fig. 6c).
3.4. Pure water flux
It is well known that the existence of dense layer will seriously a ffect
the permeability of the membrane. As mentioned before, the compact
degree of dense layer first increases and then decreases with the in-
crease of τor T. Accordingly, the pure water flux should first decrease
and then increase with the increase of τor T, which coincides with the
Fig. 7a-b.
However, according to Fig. 7c, the pure water flux increases first
and then decreases as βincreases, which cannot be exclusively ex-
plained from the viewpoint of dense layer. The reason is that the per-
meability of membrane is not only related to the compact degree of
dense layer, but also closely related to the connectivity of membrane
structure. When βincreases from 5% to 15%, the pure water flux of
ultrafiltration membranes increases from 256 L m−2h−1to 302 L m−2
h−1. According to Fig. 4, the contribution of particle structure to
membrane morphology increases as βincreases. Therefore, there are
enough voids among particles in the cellular walls to connect the cells
into open cellular structure, which can improve the connectivity of the
membrane structure. In addition, the coarsening time of the system
increases with the increase of β, the body of membrane becomes looser,
which can also improve the permeability of the resulting membrane. Asa result, the pure water flux of the resulting membrane increases as β
increases. When β= 20%, the pure water flux goes down again. As
shown in Fig. 4D3, the further growth of cellular pores makes the cel-
lular walls denser, and the accumulation of particles is also closer,which signi ficantly reduce the connectivity of membrane structure.
Therefore, the pure water flux of the resulting membrane goes down
again.
3.5. XRD analysis of the membranes
The crystallization behaviors during the pre-evaporation combined
TIPS process is veri fied by XRD. As shown in Fig. 8, the di ffraction
patterns of all membranes are similar. The di ffraction peaks at 2 θequal
Fig. 8. XRD patterns of EVOH membranes prepared under di fferent conditions.
(a)β= 15%, T = 25 °C, τ= 0 s; (b) β= 15%, T = 25 °C, τ= 10 s; (c) β= 15%,
T = 25 °C, τ= 20 s; (d) β= 20%, T = 25 °C, τ= 20 s; (e) β= 15%, T = 70 °C,
τ=1 0s .
Fig. 9. SEM images at τ= 0 s (a) the temperature of coagulation bath is 5 °C; (b) the temperature of coagulation bath is 40 °C.Z. Sun et al. Journal of Membrane Science 563 (2018) 238–246
244

to 20° and near 22° correspond to the re flection of (110) and (200)
planes of α-type EVOH crystals [21]. The di ffraction patterns could be
decomposed into amorphous and crystalline regions by a curve fitting
technique, from which the crystallinity of membrane can be calculated.
As can be seen from Fig. 8, the crystallinity values fall in the range of
34–41%, which is close to that of pure EVOH membranes reported in
the literature [37]. Such phenomenon has also been observed for other
kinds of crystalline membranes formed by means of NIPS or TIPS pro-cess [2,21,38,39] .
3.6. Discussion
The formation process of EVOH ultra filtration membrane is divided
into two stages, one is pre-evaporation stage and the other is water bath
stage. In the pre-evaporation stage, there is a double di ffusion between
solvent and water vapor. Because of the lower water vapor content inthe air, the di ffusion of water vapor into the liquid membrane can be
ignored. Therefore, the evaporation of solvent induces the phase se-paration, and the polymer concentration of the liquid membrane is
increased, which is advantageous for the formation of dense layer.
Meanwhile, the evaporation of solvent, especially 1,3-propanediol, is
beneficial to the occurrence of L-L phase separation (see Fig. 1), which
can minimize the damage of polymer crystallization to dense layer. Inaddition, the decrease of temperature also induces phase separation,
that is, TIPS. Due to the lower cooling rate of the air bath and the higher
polymer concentration originating from the evaporation of 1,3-propa-
nediol, the TIPS of the system tends to produce particle structure ori-
ginating from polymer crystallization, which is disadvantageous for the
formation of dense layer [19,20] . As we all know, polymer crystal-
lization does not occur until the temperature of the system has droppedto the dynamic crystallization temperature and it is a relatively slow
process [36]. However, solvent evaporation can occur immediately.
Therefore, in order to minimize the damage of polymer crystallization
to dense layer, the control of pre-evaporation time ( τ) is crucial. In the
present study, the maximum value of τis 25 s.
Short pre-evaporation time allows the solvent on the top layer of the
membrane evaporate merely. Therefore, the polymer concentration oftop layer increases and that of body is hardly a ffected by the pre-eva-
poration, which provides a basis for the preparation of asymmetricultrafiltration membrane. On the basis of pre-evaporation, the liquid
membrane is immersed in water bath (40 °C). The quenching e ffect
causes the top layer with high polymer concentration to become dense
[17,18] .
In order to further investigate the role of pre-evaporation in the
preparation process of EVOH ultra filtration membranes via TIPS, we
prepared EVOH membrane using conventional TIPS method, that is, thepre-evaporation time ( τ)i s0s .
As shown in Fig. 9a, when the coagulation bath temperature is 5 °C,
the thickness of dense layer and porous one are about 94 µm and 28 µm,
respectively. At this point, the cooling rate of the liquid membrane is
too fast, polymer poor phase and polymer rich phase are too late to
grow, so the non-porous dense structure is extremely thick. The thermal
resistances existed on both sides of the membrane bottom reduce the
cooling rate of membrane bottom, which leads to its porosity. As a
result, although the resulting EVOH membrane achieves ultra filtration
performance
(the BSA rejection is 93%), the pure water flux at 0.4 MPa
is only 43 L m−2h−1.
As shown in Fig. 9b, when the coagulation bath temperature is
40 °C, although the top layer is denser than the membrane body, thecompact degree is still poor. As a result, the pure water flux at 0.1 MPa
can reach 680 L m
−2h−1, however, the BSA rejection is only 61%.
In summary, it is di fficult to prepare EVOH ultra filtration membrane
with high flux and high rejection performance via the conventional
TIPS method. This also proves that pre-evaporation is crucial for the
preparation of asymmetric ultra filtration membrane via TIPS.4. Conclusions
The EVOH ultra filtration membrane with a molecular weight cuto ff
of less than 67,000 is successfully prepared via pre-evaporation com-bined thermally induced phase separation (TIPS) method. The surface
of the resulting asymmetric ultra filtration membrane is porous. The
dense layer with a depth of about 10 µm exists below the porous sur-face, it is responsible for the rejection of solutes.
Based on the shorter pre-evaporation time, the evaporation of 1,3-
propanediol provides a foundation for asymmetric structure.
Subsequently, the quenching e ffect of water bath forms a dense layer
below the porous surface. The compact degree of dense layer increasesfirst and then decreases with the increase of pre-evaporation time ( τ)o r
pre-evaporation temperature (T) or the mass fraction of 1,3-propane-diol in co-solvent ( β). Accordingly, the rejection of bovine serum al-
bumin (BSA) increases first and then decreases as τor T or βincreases.
In addition, the connectivity of membrane structure increases first and
then decreases as βincreases. Accordingly, the pure water flux in-
creases first and then decreases as βincreases. As a result, when
β= 15%, τ= 10 s and T = 50 °C, the resulting EVOH ultra filtration
membrane presents the highest BSA rejection of 99% and relatively
small pure water flux of 137.1 L m
−2h−1bar−1. However, the EVOH
ultrafiltration membrane prepared from β= 15%, τ= 20 s and
T = 25 °C achieves relatively high BSA rejection of 98% and relatively
large pure water flux of 302 L m−2h−1bar−1.
When the rejection performance of ultra filtration membrane is im-
proved by increasing polymer concentration or reducing cooling rate, it
is always accompanied by a serious decrease in permeability. However,
in this experiment, the rejection performance of ultra filtration mem-
brane is improved via pre-evaporation method, which can minimize thedamage to permeability. As a result, an ultra filtration membrane with
highflux and high rejection performance is obtained via TIPS.
Acknowledgements
This work was supported by the Key Project of Scienti fic and
Technological Research of Hebei Provincial University [No.ZD2015107].
References
[1]M.X. Shang, H. Matsuyama, T. Maki, M. Tearamoto, D.R. Lloyd, Preparation and
characterization of poly(ethylene-co-vinyl alcohol) membranes via thermally in-
duced liquid –liquid phase separation, J. Appl. Polym. Sci. 87 (2003) 853 –860.
[2]H.H. Chang, K. Beltsios, Y.H. Chen, D.J. Lin, L.P. Cheng, E ffects of cooling tem-
perature and aging treatment on the morphology of nano-and micro-porous poly(ethylene-co-vinyl alcohol) membranes by thermal induced phase separation
method, J. Appl. Polym. Sci. 131 (2014) 383 –390.
[3]H. Ye, L. Huang, W. Li, Y.Z. Zhang, L. Zhao, Q. Xin, S. Wang, L. Lin, X. Ding, Proteinadsorption and desorption behavior of a pH-responsive membrane based on ethy-
lene vinyl alcohol copolymer, RSC Adv. 7 (2017) 21398 –21405.
[4]S.M. Sau fi, C.J. Fee, Recovery of lactoferrin from whey using cross- flow cation
exchange mixed matrix membrane chromatography, Sep. Purif. Technol. 77 (2011)68–75.
[5]S.M. Sau fi, C.J. Fee, Simultaneous anion and cation exchange chromatography of
whey proteins using a customizable mixed matrix membrane, J. Chromatogr. A
1218 (2011) 9003 –9009 .
[6]T.H. Young, L.P. Cheng, H.Y. Lin, Interesting behavior for filtration of macro-
molecules through EVAL membranes, Polymer 41 (2000) 377 –383.
[7]J. Zhou, S. Meng, Z. Guo, Q.G. Du, W. Zhong, Phosphorylcholine-modi fied poly
(ethylene-co-vinyl alcohol) microporous membranes with improved protein-ad-sorption-resistance property, J. Membr. Sci. 305 (2007) 279 –286.
[8] H. Ye, L. Chen, A. Li, L. Huang, Y. Zhang, Y. Li, H. Li, Alkali-responsive membrane
prepared by grafting dimethyllaminoethyl methacrylate onto ethylene vinyl alcoholcopolymer membrane, J. Appl. Polym. Sci. 132 (2015), http://dx.doi.org/10.1002/
app.41775 .
[9]N. Riyasudheen, A. Sujith, Formation behavior and performance studies of poly
(ethylene-co-vinyl alcohol)/poly(vinyl pyrrolidone) blend membranes prepared by
non-solvent induced phase inversion method, Desalination 294 (2012) 17 –24.
[10] T.H. Young, Y.H. Huang, L.Y. Chen, E ffect of solvent evaporation on the formation
of asymmetric and symmetric membranes with crystallizable EVAL polymer, J.Membr. Sci. 164 (2000) 111 –120.
[11] L.P. Cheng, H.Y. Lin, L.W. Chen, T.H. Young, Solute rejection of dextran by EVALZ. Sun et al.
Journal of Membrane Science 563 (2018) 238–246
245

membranes with asymmetric and particulate morphologies, Polymer 39 (1998)
2135 –2142 .
[12] N. Riyasudheen, M.J. Paul, A. Sujith, E ffect of poly(vinyl pyrrolidone) on anti-
fouling properties of asymmetric poly(ethylene-co-vinyl alcohol) membranes,
Chem. Eng. Technol. 37 (2014) 1021 –1029 .
[13] T.H. Young, J.Y. Lai, W.M. You, L.P. Cheng, Equilibrium phase behavior of themembrane forming water-DMSO-EVAL copolymer system, J. Membr. Sci. 128(1997) 55 –65.
[14] M.X. Shang, H. Matsuyama, M. Teramoto, Preparation and membrane performanceof poly(ethylene-co-vinyl alcohol) hollow fiber membrane via thermally induced
phase separation, Polymer 44 (2003) 7441 –7447 .
[15] S.S. Fu, H. Matsuyama, M. Teramoto, D.R. Lloyd, Preparation of microporouspolypropylene membrane via thermally induced phase separation as support of li-
quid membranes used for metal ion recovery, J. Chem. Eng. Jpn. 36 (2003)
1397 –1404 .
[16] M.X. Shang, H. Matsuyama, T. Maki, M. Teramoto, D.R. Lloyd, E ffect of crystal-
lization and liquid –liquid phase separation on phase-separation kinetics in poly
(ethylene-co-vinyl alcohol)/glycerol solution, J. Polym. Sci. Polym. Phys. 41 (2003)194–201.
[17] H. Matsuyama, T. Iwatani, Y. Kiitamura, M. Tearamoto, N. Sugoh, Formation of
porous poly(ethylene-co-vinyl alcohol) membrane via thermally induced phase
separation, J. Appl. Polym. Sci. 79 (2001) 2449 –2455 .
[18] H. Matsuyama, T. Iwatani, Y. Kitamura, Solute rejection by poly(ethylene-co-vinylalcohol) membrane prepared by thermally induced phase separation, J. Appl.Polym. Sci. 79 (2001) 2456 –2463 .
[19] J.F. Kim, J.H. Kim, Y.M. Lee, Thermally induced phase separation and electro-spinning methods for emerging membrane applications: a review, AIChE J. 62
(2016) 461 –490.
[20] R. Lv, J. Zhou, Q.G. Du, Preparation and characterization of EVOH/PVP membranesvia thermally induced phase separation, J. Membr. Sci. 281 (2006) 700 –706.
[21] Y.H. Wu, K.G. Beltsios, J.D. Lin, P.Y. Huang, D.J. Lin, L.P. Cheng, E ffects of bath
temperature on the morphology and performance of EVOH membranes prepared bythe cold-solvent induced phase separation (CIPS) method, J. Appl. Polym. Sci. 134
(2017), http://dx.doi.org/10.1002/app.44553 .
[22] H. Matsuyama, K. Kobayashi, T. Maki, E ffect of the ethylene content of poly
(ethylene-co-vinyl alcohol) on the formation of microporous membranes via ther-mally induced phase separation, J. Appl. Polym. Sci. 82 (2001) 2583 –2589 .
[23] J. Shi, Q. Yuan, C.J. Gao, Handbook of Membrane Technology, Chemical IndustryPress, Beijing, 2001 .
[24] Z.K. Li, W.Z. Lang, W. Miao, X. Yan, Y.J. Guo, Preparation and properties of PVDF/SiO
2@GO nanohybrid membranes via thermally induced phase separation method,
J. Membr. Sci. 511 (2016) 151 –161.
[25] T.T. Jin, Z.P. Zhao, K.C. Chen, Preparation of a poly(vinyl chloride) ultra filtration
membrane through the combination of thermally induced phase separation andnon-solvent-induced phase separation, J. Appl. Polym. Sci. 133 (2016), http://dx.
doi.org/10.1002/app.42953 .
[26] B. Pang, Q. Li, Y.H. Tang, B. Zhou, T.Y. Liu, Y.K. Lin, X.L. Wang, Fabrication of
cellulose acetate ultra filtration membrane with diphenyl ketone via thermally in-
duced phase separation, J. Appl. Polym. Sci. 132 (2015), http://dx.doi.org/10.
1002/app.42669 .
[27] H.P. Xu, W.Z. Lang, X. Yan, X. Zhang, Y.J. Guo, Preparation and characterizations ofpoly(vinylidene fluoride)/oxidized multi-wall carbon nanotube membranes with bi-
continuous structure by thermally induced phase separation method, J. Membr. Sci.467 (2014) 142 –152.
[28] B.Z. Luo, Z.H. Li, J. Zhang, X.L. Wang, Formation of anisotropic microporous iso-tactic polypropylene (iPP) membrane via thermally induced phase separation,Desalination 233 (2008) 19 –31.
[29] L.S. Wu, J.F. Sun, Structure and properties of PVDF membrane with PES-C additionvia thermally induced phase separation process, Appl. Surf. Sci. 322 (2014)
101–110.
[30] D.J. Hellman, A.R. Greenberg, W.B. Krantz, A novel process for membrane fabri-cation: thermally assisted evaporative phase separation (TAEPS), J. Membr. Sci.
230 (2004) 99 –109.
[31] H. Matsuyama, M. Yuasa, Y. Kitamura, M. Teramoto, D.R. Lloyd, Structure control
of anisotropic and asymmetric polypropylene membrane prepared by thermally
induced phase separation, J. Membr. Sci. 179 (2000) 91 –100.
[32] M.Y. Jeon, C.K. Kim, Phase behaviour of polymer/diluent/diluent mixtures andtheir application to control microporous membrane structure, J. Membr. Sci. 300(2007) 172 –181.
[33] N.L. Cheng, Handbook of Solvents, Chemical Industry Press, Beijing, 2007 .
[34] Z.S. Yang, P.L. Li, L.X. Xie, Z. Wang, S.C. Wang, Preparation of iPP hollow- fiber
microporous membranes via thermally induced phase separation with co-solventsof DBP and DOP, Desalination 192 (2006) 168 –181.
[35] D.R. Lloyd, S.S. Kim, K.E. Kinzer, Microporous membrane formation via thermallyinduced phase separation, J. Membr. Sci. 64 (1991) 1 –11.
[36] P.V.D. Witte, P.J. Dijkstra, J.W.A.V.D. Berg, J. Feijen, Phase separation processes in
polymer solutions in relation to membrane formation, J. Membr. Sci. 117 (1996)
1–31.
[37] J.A. Lima, M.I. Felisberti, Poly(ethylene-co-vinyl alcohol) and poly(methyl metha-crylate) blends: phase behavior and morphology, Eur. Polym. J. 44 (2008)1140 –1148 .
[38] H.H. Chang, L.K. Chang, C.D. Yang, D.J. Lin, L.P. Cheng, E ffect of polar rotation on
the formation of porous poly(vinylidene fluoride) membranes by immersion pre-
cipitation
in an alcohol bath, J. Membr. Sci. 513 (2016) 186 –196.
[39] H.H. Chang, S.C. Chen, D.J. Lin, L.P. Cheng, Preparation of bi-continuous Nylon-66porous membranes by coagulation of incipient dopes in soft non-solvent baths,
Desalination 313 (2013) 77 –86.Z. Sun et al.
Journal of Membrane Science 563 (2018) 238–246
246