Journal of Applied Science and Engineering, Vol. 21, No. 1, pp. 89-98 (2018) DOI: 10.6180/jase.201803_21(1).

0011

Effect of Inner Non-solvent Condition on Morphology

of PVDF Hollow Fiber Membranes Prepared by NIPS
Spinning Process
Hsu-Hsien Chang1,2, Sheng-Chang Chen1,2 and Liao-Ping Cheng1,3*
1
Department of Chemical and Materials Engineering, Tamkang University,
Tamsui, Taiwan 251, R.O.C.
2
Bright Sheland International Co., New Taipei City, Taiwan 221, R.O.C.
3
Energy and Opto-Electronic Materials Research Center, Tamkang University,
Tamsui, Taiwan 251, R.O.C.

Abstract
Hollow fiber membranes of PVDF were prepared by isothermal wet spinning process from the
water/DMF/PVDF system at 25 °C. The dope solution comprised 21 wt% PVDF in DMF, and pure
water was employed both as the inner and outer coagulants. SEM imaging of the membrane indicated
an unexpected interesting morphology: columnar macrovoids extended from the outer surface to the
central region, while cellular pores constituted the inner half of the membrane. As water was a harsh
nonsolvent, one tended to think that macrovoids shall form towards the inner surface of the hollow
fiber. Absence of macrovoids in this region was explained based on the fact that DMF would
accumulate rapidly in the inner extruding stream as it underwent mass-exchange with water in the
inner bath. Such activity rendered effectively the coagulant became effectively a soft one. As a result,
macrovoids was eliminated and the inner surface became porous.

Key Words: PVDF, Hollow Fiber, Membrane, Morphology, NIPS

1. Introduction the formed membranes are collected by means of a take-

Porous hollow fiber membranes are used over a wide
range of applications, such as reverse osmosis, and ul-
tra-filtration, membrane distillation, etc. [1-5]. They are
generally prepared by the non-solvent induced phase se-
paration (NIPS) spinning process. In such process, a
polymer solution (dope) consisting of polymer, solvent,
and sometimes additives is pumped through the annulus,
and a bore injection fluidis pumped through the inner
needle tube of the spinneret. Together these two streams
are extruded into a non-solvent bath where exchange of
solvent and non-solvent across the dope/bath interfaces
induces instability and subsequent phase separation of
the dope solution. After appropriate washing and drying,
*Corresponding author. E-mail: lpcheng@mail.tku.edu.tw
up device [1-10]. By varying the spinning parameters,
e.g. the inner and outer diameters of the spinneret, the
compositions of the polymer solution, non-solvent st-
rength of the inner and outer baths, etc., porous hollow
fiber membranes can be produced encompassing a wide
range of morphologies.
NIPS process typically results in asymmetric or sym-
metric membranes (flat sheet or hollow fiber). Asymmet-
ric membrane has a skin layer on the surface which ser-
ves as a selective barrier to differentiate incoming feeds.
Underneath the skin is a porous bulk, which functions as
the mechanical support to sustain the transmembrane
pressure during filtration operations [11-17]. The po-
rous bulk may be sponge-like or contain teardrop- or fin-
ger-like macrovoids, depending on the preparation con-
ditions. Asymmetric structure is typically formed by pre-
90 Hsu-Hsien Chang et al.
cipitation of the dope solution in a so-called harsh non-
solvent bath, e.g. water, such that polymer can accumu-
late rapidly near the surface region (contacting the bath)
to give a continuous gel layer that eventually develops
into a skin. The pores in the bulk are derived from poly-
mer-lean phase formed during phase separation induced
by the influx of non-solvent. In contrast to asymmetric
membrane, symmetric membrane is skinless and has a
relatively uniform pore distribution across the mem-
brane [18-24]. The membrane may be formed by carry-
ing out polymer precipitation in a soft nonsolvent bath
(e.g., a bath containing significant amount of solvent)
such that exchange of solvent and nonsolvent is slow and
polymer concentration would not build-up sharply dur-
ing the phase separation process [18-24].
In the present research, porous PVDF hollow fiber
membranes were prepared by isothermal NIPS process
from the water/DMF/PVDF system. Morphologies of
the membranes were observed by means of SEM. It is in-
teresting to find that finger-like macrovoids on the lumen
(inner) side can be eliminated simply by using spinneret
with a sufficiently small inner tube. Such is the case even
when a harsh nonsolvent, e.g., water, is employed as the
inner coagulant. The reason for this unusual new finding
is explained based on mass transfer and phase separation
considerations. With a macrovoid-free lumen side, the
membrane is expected to be stronger than ordinary ultra-
or nano-filtration membranes that have macrovoids on
both inner and outer sides of the membrane.
2. Experimental
2.1 Material
Poly(vinylidene fluoride) (PVDF) polymer (Kynar
740, Elf Atochem; intrinsic viscosity = 0.881 dL/g, Mn =
254,000 g/mole) and a terpolymer (Kynar 9301) of vin-
ylidene fluoride (VDF), tetrafluoro ethylene (TFE), and
hexafluoropropylene (HFP) with VDF/HFP/TFE = 60/
20/20, intrinsic viscosity = 0.4 dl/g, and Mn = 79200 g/
mol was supplied by Elf Atochem Inc. N,N-dimethyl-
formamide (DMF, Baker Analyzed, reagent grade) was
used as the solvent for preparation of dopes for hollow fi-
ber spinning. Distilled-deionized water was used as the
non-solvent for precipitation of PVDF from the dope.
All materials were used as received.
2.2 Phase Diagram Determination
The gelation phase boundary at 25 °C for the ternary
system, water/DMF/PVDF, was determined by the widely
used cloud point method [20,23]. Briefly, a specific
amount of PVDF (dried in an oven at 60 °C) was mixed
with DMF and sealed in a glass bottle with Teflon-lined
cap. The mixture was blended at 60 °C until the polymer
was completely dissolved. A known quantity of water
was then added to the polymer solution. Local precipita-
tion may be observed at the contact of water with the so-
lution. The mixture (termed dope) was further blended at
60 °C until it became clear and homogeneous again.
Then it was put in a thermostat maintained at 25 °C over
the period of 30 days during which gelation might occur
and the dope became a gel. The gelation points in the
phase diagram were identified as the compositions at
which gelation first occurred in a series of samples with
increasing non-solvent/solvent ratios.
The terpolymer (Kynar 9301) is largely amorphous,
and hence can be used as a model to access the liquid-liq-
uid (L-L) demixing boundary (often termed binodal in
the literature). Such is based on the fact that Kynar 9301
is chemical structurally similar to PVDF (Kynar 740)
and that these two polymers have close solubility para-
meters, 7.2 and 7.74 (cal/cm3)1/2 respectively for the ter-
polymer and PVDF. To experimentally access the bino-
dal, a series of solutions composed of water/DMF/Kynar
9301were prepared and then put in a thermostat main-
tained at 25 °C, as in the procedure described above. Un-
stable solutions underwent L-L demixing into two clear
homogeneous liquid layers. The binodal points were de-
fined as the compositions of the separated liquid phases.
It is also noted that the dope would not gel (still flows)
even at the concentration of 50 wt.%, which may be as-
sociated with the low molecular weight of the polymer.
2.3 Membrane Preparation and SEM Observation
A schematic representation of the employed NIPS
spinning process is illustrated in Figure 1. The dope so-
lution for spinning was prepared by dissolving 21 wt%
of PVDF in DMF at 60 °C. It was cooled to 25 °C and
then pumped through the annulus of the spinneret. On
the other hand, water was pumped through the central
tube of the (ID/OD = 0.16/0.4 mm). Together these two
streams entered the coagulation bath where phase sepa-
 Effect of Inner Non-solvent Condition on Morphology of PVDF Hollow Fiber Membranes Prepared by NIPS Spinning Process 91

Figure 1. Schematic representation of the hollow fiber spinning process. 1: inner non-solvent fluid; 2: hollow fiber spinning solu-
tion; 3: outer coagulation bath; P: pumps; (a): fluids in spinneret and inner coagulation bath; (b): cross section of the in-
ner tube and the spinneret.

ration occurred and polymer precipitated to form a po-
rous hollow fiber membrane. The membrane was wa-
shed repeatedly first with distilled water, then with iso-
propanol, followed by hexane. Afterwards, it was dried
at 50 °C in a convective oven. Flat sheet membranes
were also prepared for comparison of porous structures.
The membrane was prepared by immersion the 21 wt%
dope cast on a glass plate into a water bath at 25 °C. The
nascent membrane, as being peeled off from the glass
plate, was washed and dried with the same procedure for
processing hollow fiber membranes.
Morphologies of the porous membranes (both flat
sheet and hollow fiber) were observed using a field emis-
sion scanning electron microscope (FESEM, Leo 1530,
Carl Zeiss, Oberkochen, Germany). A piece of mem-
brane sample was vacuum-dried and then attached to a
sample holder by means of conductive copper tapes. The
cross section of the membrane was obtained by fractur-
ing in liquid nitrogen. Silver paste was applied at the
edges of the sample to enhance electronic conductivity.
Then, the sample was sputtered with a thin layer (~2 nm)
of Pt-Pd alloy and observed under a low acceleration
voltage, 2 kV, by means of an in-lens detector.
3. Results and Discussion
3.1 Phase Diagram of the Water/DMF/PVDF System
In Figure 2, the experimental phase equilibrium cur-
ves at 25 °C are presented both for the amorphous ter-
polymer and the crystalline PVDF in water-DMF sys-
tems. The solid circles (·) represent the cloud point com-
positions measured for the terpolymer in water-DMF so-
lutions. These points mark the L-L demixing phase
boundary; inside which an originally homogeneous solu-
tion will separate into two liquid phases in equilibrium.
The filled triangle points (s) in Figure 2 depict the crys-
tallization-induced gelation equilibrium points observed
for crystalline PVDF in water-DMF solutions. These
data define a gelation phase boundary. Above the gela-
tion line is a single-phase region, in which a homoge-
neous dope can be prepared with a long-term stability.
Below this line is a metastable zone, in which an origi-
nally homogeneous dope solution will gel upon standing
at 25 °C over an extended period of time. The fact that
gelation was induced by crystallization has been proven
by various techniques, such as differential scanning cal-
orimetry (DSC), polarized optical microscopy (POM),
92 Hsu-Hsien Chang et al.
SEM, etc. [11,25]. As the gelation region encloses the
binodal, a dope inside the binodal can undergo both L-L
demixing and crystallization types of phase separation.
3.2 Morphology of PVDF Hollow Fiber and Flat
Membranes
The casting dope for preparing the PVDF flat mem-
brane consisted of 21 wt% PVDF dissolved in DMF (cf.,
the solid square point in Figure 2). It was a clear uniform
viscous solution. When immersed in pure water (a typi-
cal harsh nonsolvent) it underwent a very rapid precipi-
tation process to form a membrane with morphology that
is characteristic of L-L demixing dominant precipitation
process. The total cross section of the membrane, as
shown in Figure 3(a), comprises finger-like macrovoids
and cellular pores resembling those seen in common am-
orphous membranes. That is, L-L demixing took place
first and set-in the overall porous morphology before
crystallization occurred in the polymer-rich phase that
surrounded the polymer-poor liquid droplets. Figure 3(b)
shows the high magnification image of the upper part of
Figure 3(a). A thin skin followed by parallel columnar
macrovoids extending into the central regionis seen.
Roughened and dendritic features on the pore walls sug-
gest occurrence of crystallization during the late stage of
the phase separation process. DSC and XRD analyses of
similar PVDF membranes can be found in the literature
[25]. Figure 3(c) shows the enlarged view of the cross
section near the bottom surface of the membrane. Cellu-
lar pores of about 3-5 mm in diameter are clearly ob-
Figure 2. Phase diagram of the water/DMF/PVDF ternary
system at 25 °C. Binodal was determined for the
water/DMF/Kynar 9301 system. Gelation line was
determined for the water/DMF/Kynar 740 system.
served. On the pore walls, spherulitic entities (e.g., cen-
tral region, pointed out by an arrow) are in evidence,
which were derived from spherulitic growth of PVDF
crystallites. Top surface of the membrane, as shown in
Figure 3(d), is dense without appreciable pores under the
resolution of ca. 40 nm. Such is typical of immersing a
good dope solution in a harsh non-solvent bath, such as
water. Upon contact of dope and bath, a stiff gellayer
forms due to a sharp upsurge of polymer concentration at
the bath-dope interface, which subsequently develops
into the dense skin of the membrane. The bottom surface,
as shown in Figure 3(e), consists of truncated PVDF glo-
bular objects along with cellular pores. Figure 3(f) shows
the high magnification of Figure 3(e), nano-granular and
dendritic features are clearly demonstrated, and because
the globules grow pressing-against the glass plate during
precipitation, they appear flattened [23-25].
Asymmetric polymer membranes consisting of a thin
Figure 3. Morphologies of the cross section, top, and bottom
surfaces of the flat sheet membrane prepared by im-
mersing a 21 wt% PVDF/DMF dope in pure water.
(a) total cross section; (b) high magnification of the
upper part of (a); (c) high magnification of the
lower part of (a); (d) top surface; (e) bottom sur-
face; (f) high magnification of (e).
 Effect of Inner Non-solvent Condition on Morphology of PVDF Hollow Fiber Membranes Prepared by NIPS Spinning Process 93

top layer supported by a porous sublayer composed of outer non-solvent bath and an inner stream of non-sol-
macrovoids and cellular pores are often observed for vent coagulant that flows through the center region of the
membranes prepared by NIPS process in a harsh non- system. That is, there exist two membrane/bath inter-
solvent bath [1,11-17,26-31]. It is well accepted that faces, across which solvent and non-solvent exchange
cellular pores are resulted from radial growth of poly- occurs. As harsh non-solvent was employed for both the
mer-poor phase during L-L demixing. As to the macro- inner and outer baths in the present work, one would
voids, the causes of their formation have been exten- speculate that the morphology of the membrane shall be
sively studied in the literature [26-31]. A number of similar to the combination of two flat-sheet membranes
mechanisms have been proposed, e.g., surface breakage arranged in the Top/Bottom/Bottom/Top order, as shown
due to shrinkage stress, liquid-liquid demixing and dif- in Figure 6. That is, finger-like macrovoids dominate the
fusional growth of cellular pores, osmosis effect during regimes near the inner and outer surfaces, while cellular
growth of cellular pores, instability caused by perturba- pores are inserted in-between the macrovoids. The real
tion of the interface in steep concentration profile, etc. situation is quite different, however, as shown below.
Based on Mulder’s proposition [1], formation of macro- SEM photo micrographs of the PVDF hollow fiber
voids is initiated by L-L demixing. The liquid micro- membrane prepared by spinning of 21 wt% PVDF/DMF
drops (micelles) of the polymer-poor phase are the em- dope solution in pure water are shown in Figure 7. Pre-
bryos for macrovoids developments. A schematic repre-
sentation is illustrated in Figure 4 for such case. L-L
demixing occurs rapidly after the binodal is crossed due
to exchange of nonsolvent and solvent at the mem-
brane-bath interface. This causes formation of first mi-
crodrops at t = t1 near the top surface. Continuing in-
fluxes of solvent and non-solvent into the microdrops, as
driven by osmosis forces, lead to the downward-growth
of macrovoids (upward growth is unlikely due to higher
polymer concentration there is) cf., t = t2. This activity
proceeds until the polymer concentration of the poly-
mer-rich phase enclosing the macrovoids becomes so
high that gelation occurs, and further macrovoids growth
becomes inhibited, cf., t = t3. Finally, asymmetric mem-
braneis formed with macrovoids extending from the top
surface to the central or even bottom region of the mem-
brane.
A schematic representation of the mass transfer ac-
ross the polymer solution/bath interface during immer-
sion of a casting dope in the form of flat sheet is shown in
Figure 5(a). The solvent diffuses into the non-solvent
bath with flux J2, whereas the non-solvent diffuse into
the casting dope with flux J1. As is well recognized that
such mass transfer gives rise to phase separations (the
dope crosses the binodal/crystallization boundaries) and
lead to porous structure formation in the membrane. For Figure 4. Schematic representation of the growth of macro-
the spinning of hollow fibers, a somewhat different cir- voids at different times (t1, t2, and t3) during precipi-
tation. t1: initial formation of microdrops near the
cumstance is encountered. As shown in Figure 5(b), the top surface. t2: downward-growth of macrovoids.
polymer solution is sandwiched between a reservoir of t3: completeness of macrovoids formation.
94 Hsu-Hsien Chang et al.
Figure 5. Schematic representation of a polymer solution/
bath interface. (a) formation of a flat membrane by
NIPS method; (b) formation of a hollow fiber mem-
brane by NIPS spinning process. J1: flux of non-
solvent diffusing into the dope. J2: flux of solvent
diffusing into the bath.
Figure 6. Schematic representation of the formation of PVDF follow fiber membrane by NIPS spinning process.
cipitation is observed to occur very rapidly to yield a
morphology dictated by L-L demixing. The cross section
of the hollow fiber membrane, cf., Figure 7(a), exhibits
finger-like macrovoids from outer surface to the central
part of the membrane followed by cellular pores all the
way to the inner surface, which is similar to those shown
in Figure 3(a). The outer surface, cf., Figure 7(b), forms a
relatively dense skin with sporadic nm-scale dints evenly
distributed on the surface. The magnified view of the in-
ner surface can be seen from Figure 7(c). It is corrugate
with identifiable pores of ~2 mm; namely, the surface is
not a skin. Surprisingly, finger-like macrovoids are ab-
sent towards the lumen side of the hollow fiber, even
with water being used as the inner coagulant, as opposed
to what is illustrated in Figure 6. The reason can be ex-
plained based on mass transfer events that occur across
the inner-interface during immersion. As DMF in the
dope out flows into the inner tube, water and DMF mixed
quickly to make the inner bath a soft coagulant (i.e., con-
taining significant amount of solvent) for the polymer.
Such is possible as long as the inner tube is thin enough
to allow a quick build-up of DMF concentration in the
inner bath. In this case, the inner gel layer would be soft
 Effect of Inner Non-solvent Condition on Morphology of PVDF Hollow Fiber Membranes Prepared by NIPS Spinning Process 95

(as oppose to the rigid one on the outer surface) and eas-
ily broken into pores during subsequent precipitation
process [20-23]. Shieh et al. has demonstrated a similar
immersion situation for the water/formic acid/Nylon-66
system. The calculated diffusion trajectory indicates clearly
accumulation of solvent in the inner bath, and hence the
polymer concentration at the inner-interface is much
lower than that at the outer-interface. By contrast, if a
large inner tube is used, the solvent-concentration built-
up in the inner bath may become ineffective. In such
case, the inner bath remains harsh and the formed mem-
brane has macrovoids from both inner and outer sides of
the hollow fiber, as is illustrated in Figure 8. The mem-
brane was prepared by spinning 21% PVDF/DMF dope
in water by means of a spinneret with larger inner tube
(ID/OD = 0.33/0.6 mm/mm, comparing with the small
tube, 0.16/0.4, shown in Figure 1). Obviously, the cross
section has a macrovoids/cellular pore/macrovoids sand-
wich structure, being consistent with the immersion situ-
ation shown in Figure 6 where dope solution contacts
harsh bath on both sides.

4. Conclusions

PVDF membranes in the forms of hollow fibers and

flat sheets were prepared by nonsolvent induced phase
separation at 25 °C from the water/DMF/PVDF system.
SEM micrographs of the flat sheet membrane indicate
an asymmetric structure featuring finger-like macro-

Figure 7. SEM micrographs of the hollow fiber membrane

prepared by spinning of a 21 wt% solution in water Figure 8. SEM micrograph of the PVDF hollow fiber mem-
bath. (a) outer surface/cross section; (b) outer sur- brane prepared by using a spinneret with larger in-
face (c) inner surface/cross section. ner tube (ID/OD = 0.33/0.6 mm/mm).
96 Hsu-Hsien Chang et al.
voids and cellular pores, typical of L-L demixing domi-
nated phase separation process during precipitation.
For the PVDF hollow fiber membrane spun by means of
a spinneret with small inner tube, a similar asymmetric
structure was demonstrated, in which finger-like ma-
crovoids were absent from the inner (lumen) side con-
tacting the harsh water bath. Such unexpected result
was explained based on the fact that the out-fluxed
DMF can quickly soften the small inner bath and delay
the demixing process. In case that a larger inner tube
was employed in the spinneret, macrovoids formed on
the lumen side as usual simply because the amount of
out-fluxed DMF was too small to effectively soften the
inner bath.
Acknowledgement
The authors would like to thank the Ministry of Sci-
ence and Technology of Taiwan for financial support.
(NSC 99-2221-E-032-002-MY3).
References
[1] Strathmann, H., Introduction to Membrane Science
and Technology, Wiley-VCH Verlag & Co., Germany
(2011).
[2] Mulder, M., Basic Principles of Membrane Technol-
ogy, 2nd ed., Kluwer Academic Publisher, Dordrecht
(1996).
[3] Culfaz, P. Z., Rolevink, E., Rijn, C. V., Lammertink, R.
G. H. and Wessling, M., “Microstructured Hollow Fi-
bers for Ultrafiltration,” Journal of Membrane Sci-
ence, Vol. 347, No. 1-2, pp. 32–41 (2010). doi: 10.
1016/j.memsci.2009.10.003
[4] Peng, N., Widjojo, N., Sukitpaneenit, M. M., Teoh, P.,
Lipscomb, G. G., Chung, T. S. and Lai, J. Y., “Evolu-
tion of Polymeric Hollow Fibers as Sustainable Tech-
nologies: Past, Present, and Future,” Progress in Poly-
mer Science, Vol. 37, No. 10, pp. 1401–1424 (2012).
doi: 10.1016/j.progpolymsci.2012.01.001
[5] Liu, F., Hashim, N. A., Liu, Y., Abed, M. R. M. and Li,
K., “Review-progress in the Production and Modifica-
tion of PVDF Membranes,” Journal of Membrane
Science, Vol. 375, No. 1-2, pp. 1-27 (2011). doi:
10.1016/j.memsci.2011.03.014
[6] Kang, G. D. and Cao, Y. M., “Application and Modifi-
cation of Poly(vinylidene fluoride) (PVDF) Mem-
branes – A Review,” Journal of Membrane Science,
Vol. 463, No. 1, pp. 145-165 (2014). doi: 10.1016/
j.memsci.2014.03.055
[7] Drioli, E., Ali, A., Simone, S., Macedonio, F., Al-Jiil,
S. A., Shabonah, F. S., Al-Romaih, H. S., Al-Harbi, O.,
Figoli, A. and Criscuoli, A., “Novel PVDF Hollow Fi-
ber Membranes for Vacuum and Direct Contact Mem-
brane Distillation Applications,” Separation and Puri-
fication Technology, Vol. 115, No. 30, pp. 27-38
(2013). doi: 10.1016/j.seppur.2013.04.040
[8] Bottino, A., Camera, G. R., Capannelli, G. and Munari,
S., “The Formation of Microporous Polyvinylidene
Difluoride Membranes by Phase Separation,” Journal
of Membrane Science, Vol. 577, No. 1, pp. 1-20
(1991). doi: 10.1016/S0376-7388(00)81159-X
[9] Lin, D. J., Lin, C. L. and Guo, S. Y., “Network
Nano-porous Poly(vinylidene fluoride-co-hexafluoro-
propene) Membranes by Nano-gelation Assisted Phase
Separation of Poly(vinylidene fluoride-co-hexafluoro-
propene)/Poly(methyl methacrylate) Blend Precursor
in Toluene,” Macromolecules, Vol. 45, No. 21, pp.
8824-8832 (2012). doi: 10.1021/ma301075e
[10] Xu, J. and Xu, Z. L., “Poly(vinyl chloride) (PVC) Hol-
low Fiber Ultrafiltration Membranes Prepared from
PVC/additives/solvent,” Journal of Membrane Sci-
ence, Vol. 208, No. 1-2, pp. 203-212 (2002). doi:
10.1016/S0376-7388(02)00261-2
[11] Cheng, L. P., “Effect of Temperature on the Formation
of Microporous PVDF Membranes by Precipitation
from 1-Octanol/DMF/PVDF and Water/DMF/PVDF
Systems,” Macromolecules, Vol. 32, No. 20, pp.
6668-6674 (1999). doi: 10.1021/ma990418l
[12] Sukitpaneenit, P. and Chung, T. S., “Molecular Eluci-
dation of Morphology and Mechanical Properties of
PVDF Hollow Fiber Membranes from Aspects of Phase
Inversion, Crystallization and Rheology,” Journal of
Membrane Science, Vol. 340, No. 1-2, pp. 192-205
(2009). doi: 10.1016/j.memsci.2009.05.029
[13] Ansorge, W., Schuster, O., Wechs, F. and Dombrowski,
K., U.S. Patent 8,727,136 B2 (2014).
[14] Setiawan, L., Wang, R., Li, K. and Fane, A. G., “Fabri-
cation and Characterization of Forward Osmosis Hol-
low Fiber Membranes with Antifouling NF-like Selec-
 Effect of Inner Non-solvent Condition on Morphology of PVDF Hollow Fiber Membranes Prepared by NIPS Spinning Process
tive Layer,” Journal of Membrane Science, Vol. 394-
395, No. 15, pp. 80-88 (2012). doi: 10.1016/j.memsci.
2011.12.026
[15] Gryta, M. and Barancewicz, M., “Influence of Mor-
phology of PVDF Capillary Membranes on the Per-
formance of Direct Contact Membrane Distillation,”
Journal of Membrane Science, Vol. 358, No. 1-2,
pp. 158-167 (2010). doi: 10.1016/j.memsci.2010.
04.044
[16] Shih, C. H., Gryte, C. C. and Cheng, L. P., “Precipita-
tion Dynamics for the Formation of Nylon-6 Poly-
amide Membranes by Isothermal Precipitation in Wa-
ter/Formic Acid Solutions,” Journal of Applied Sci-
ence and Engineering, Vol, 15, No. 4, pp. 357-366
(2012). doi: 10.6180/jase.2012.15.4.06
[17] Lin, D. J., Chang, C. L., Chen, T. C. and Cheng, L. P.,
“On the Structure of Porous Poly(vinylidene fluoride)
Membrane Prepared by Phase Inversion from Water-
NMP-PVDF System,” Tamkang Journal of Science
and Engineering, Vol. 5, No. 2, pp. 95-98 (2002). doi:
10.6180/jase.2002.5.2.04
[18] Lin, D. J., Beltsios, K., Chang, C. L. and Cheng, L. P.,
“Fine Structure and Formation Mechanism of Particu-
late Phase-inversion Poly(vinylidene fluoride) Mem-
branes,” Journal of Polymer Science Part B Polymer
Physics, Vol. 41, No. 13, pp. 1578-1588 (2003). doi:
10.1002/polb.10513
[19] Kuo, C. Y., Lin, H. N., Tsai, H. A., Wang, D. M. and
Lai, J. Y., “Fabrication of a High Hydrophobic PVDF
Membrane via Nonsolvent Induced Phase Separa-
tion,” Desalination, Vol. 233, No. 1-3, pp. 40-47
(2008). doi: 10.1016/j.desal.2007.09.025
[20] Chang, H. H., Chen, S. C., Lin, D. J. and Cheng, L. P.,
“Preparation of Bi-continuous Nylon-66 Porous
Membranes by Coagulation of Incipient Dopes in
Soft Non-solvent Baths,” Desalination, Vol. 313, No.
15, pp. 77-86 (2013). doi: 10.1016/j.desal.2012.12.
009
[21] Buonomenna, M. G., Macchi, P., Davoli, M. and
Drioli, E., “Poly(vinylidene fluoride) Membranes by
Phase Inversion: the Role the Casting and Coagulation
Conditions Play in Their Morphology, Crystalline St-
ructure and Properties,” European Polymer Journal,
Vol. 43, No. 4, pp. 1557-1572 (2007). doi: 10.1016/
j.eurpolymj.2006.12.033
97
[22] Young, T. H., Lin, C. W., Cheng, L. P. and Hsieh, C. C.,
“Preparation of EVAL Membranes with Smooth and
Particulate Morphologies for Neuronal Culture,” Bio-
materials, Vol. 22, No. 13, pp. 1771-1777 (2001). doi:
10.1016/S0142-9612(00)00337-9
[23] Chang, H. H., Chang, L. K., Yang, C. D., Lin, D. J.
and Cheng, L. P., “Effect of Polar Rotation on the
Formation of Porous Poly(vinylidene fluoride) Mem-
branes by Immersion Precipitation in an Alcohol
Bath,” Journal of Membrane Science, Vol. 513, No.
1, pp. 186-196 (2016). doi: 10.1016/j.memsci.2016.
04.052
[24] Chang, H. H., Chen, S. C., Lin, D. J. and Cheng, L. P.,
“The Effect of Tween-20 Additive on the Morphology
and Performance of PVDF Membranes,” Journal of
Membrane Science, Vol. 466, No. 15, pp. 302-312
(2014). doi: 10.1016/j.memsci.2014.05.011
[25] Lin, D. J., Chang, C. L., Huang, F. M. and Cheng, L. P.,
“Effect of Salt Additive on the Formation of Micro-
porous Poly(vinylidene fluoride) Membranes by Phase
Inversion From LiClO4/Water/DMF/PVDF System,”
Polymer, Vol. 44, No. 2, pp. 413-422 (2003). doi:
10.1016/S0032-3861(02)00731-0
[26] Smolders, C. A., Reuvers, A. J., Boom, R. M. and
Wienk, I. M., “Microstructures in Phase-inversion
Membranes. Part 1: Formation of Macrovoids,” Jour-
nal of Membrane Science, Vol. 73, No. 2-3, pp. 259-
275 (1992). doi: 10.1016/0376-7388(92)80134-6
[27] Strathmann, H., in Lloyd DR (editor), “Material Sci-
ence of Synthetic Membrane,” ACS Symposium Se-
ries, Washington, DC: American Chemical Society,
pp. 165-195 (1985).
[28] McKelvey, S. A. and Koros, W. J., “Phase Separation,
Vitrification, and the Manifestation of Macrovoids in
Polymeric Asymmetric Membranes,” Journal of Mem-
brane Science, Vol. 112, No. 1, pp. 29-39 (1996). doi:
10.1016/0376-7388(95)00197-2
[29] Ray, R. J., Krantz, W. B. and Sani, R. L., “Linear Sta-
bility Theory Model for Finger Formation in Asym-
metric Membranes,” Journal of Membrane Science,
Vol. 23, No. 2, pp. 155-182 (1985). doi: 10.1016/
S0376-7388(00)82216-4
[30] Termonia, Y., “Fundamentals of Polymer Coagula-
tion,” Journal of Polymer Science Part B Polymer
Physics, Vol. 33, No. 2, pp. 279-288 (1995). doi:
98 Hsu-Hsien Chang et al.

10.1002/polb.1995.090330213 31-43 (1999). doi: 10.1016/S0376-7388(98)00292-0

[31] Lai, J. Y., Lin, F. C., Wu, T. T. and Wang, D. M., “On
the Formation of Macrovoids in PMMA Membranes,” Manuscript Received: Mar. 9, 2017
Journal of Membrane Science, Vol. 155, No. 1, pp. Accepted: Jul. 30, 2017