Applied Surface Science 258 (2012) 7872–7881

Contents lists available at SciVerse ScienceDirect

Applied Surface Science

journal homepage: www.elsevier.com/locate/apsusc

Effects of exposure time on variations in the structure and hydrophobicity of

polyvinylidene fluoride membranes prepared via vapor-induced phase
separation
Yuelian Peng ∗ , Hongwei Fan, Yajun Dong, Yanna Song, Hua Han
College of Environmental and Energy Engineering, Beijing University of Technology, Beijing 100124, PR China

a r t i c l e i n f o a b s t r a c t

Article history: The present investigation revealed how the surface morphology and hydrophobicity of polyvinylidene
Received 26 October 2011 fluoride (PVDF) membranes, which were prepared via a vapor-induced phase separation (VIPS) method,
Received in revised form 15 February 2012 were affected by the exposure time. The mass variation of the cast film was recorded. Membrane mor-
Accepted 16 April 2012
phologies were observed by scanning electron microscopy (SEM) and thermal behaviors of membranes
Available online 23 April 2012
were examined by differential scanning calorimetry (DSC). Wide angle X-ray diffraction (WAXD) was
employed to analyze the crystalline structures of the overall membranes and the surface layers. The
Keywords:
results showed that different membrane morphologies and hydrophobicities could be obtained by chang-
Surface morphology
Hydrophobicity
ing the exposure time. A long exposure time facilitated the crystallization process, resulting in the
Polyvinylidene fluoride membranes formation of a porous skin and particle morphology, which increased the hydrophobicity of the sur-
Vapor-induced phase separation face. A short exposure time favored the formation of a digitate macrovoid and dense skin resulting from
Vacuum membrane distillation liquid–liquid phase separation in the immersion process, which reduced surface hydrophobicity. The
water permeate flux in vacuum membrane distillation was greatly affected by the membrane porosity
and surface hydrophobicity.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction
Phase separation is a widely used method for making microp-
orous polymer membranes. There are two main phase separations
involved in this process, i.e., non-solvent induced phase separation
(NIPS) and thermally induced phase separation (TIPS), depending
on the driving force for the phase separation. NIPS is the most
important method in the preparation of reverse osmosis, nanofil-
tration, ultrafiltration and microfiltration membranes. Typically, a
clear solution (dope) consisting of a polymer, a solvent and usu-
ally a non-solvent is formed into the desired shape (e.g., a flat
sheet or hollow fiber) and is then contacted with a non-solvent
that is miscible with the solvent. The ensuing transfer of solvent
and non-solvent leads to phase inversion and the formation of a
three-dimensional porous polymer network. Wet casting is char-
acterized by rapid mass-transfer rates and typically, the formation
of a dense skin at the top surface of the membrane.
The non-solvent vapor-induced phase-separation (VIPS) pro-
cess, which is a NIPS process, has attracted an increasing amount
of attention in recent years. In VIPS, the cast film is first exposed
to humid air prior to immersion in the coagulation bath. Typically,
∗ Corresponding author. Tel.: +86 10 67391090; fax: +86 10 67391983.
E-mail address: pyl@bjut.edu.cn (Y. Peng).
for these systems, water is used as the non-solvent, so that the
water transfer associated with a humid air exposure step leads
to phase separation. The extent and rate of water transfer can be
controlled by adjusting the air velocity, relative humidity (RH) and
temperature of the air, as well as the exposure time. By a judicious
combination of these variables, symmetric membranes with large
pores at the top surface can be obtained [1–7].
This unique hierarchical micro- and nanostructuring on the
leg’s surface of water striders therefore appears to be responsible
for its water resistance and the strong supporting force [8]. VIPS
has also been used to construct a superhydrophobic surface. Xie
et al. [9] used water-vapor-induced phase separation in an ambi-
ent environment to prepare a fluorine-end-capped polyurethane
(FPU)/poly (methyl methacrylate) (PMMA) film that possessed a
micro- and nanoscale binary surface structure similar to the sur-
face of a lotus leaf and a sponge-like bulk structure. The water
contact angle (CA) of the film was 166◦ , whereas the oil CA was
140◦ . This film is thus classified as a superamphiphobic surface.
Peng et al. [10] prepared a porous and highly hydrophobic sur-
face of polyvinylidene fluoride (PVDF) by the gelation of a PVDF/N,
N-dimethylacetamide (DMAc) casting solution in open air instead
of its immersion into a precipitation bath. It was found that the
membranes exhibited certain micro- and nanoscale hierarchical
surface roughnesses, and the water CA of the samples was as high
as 150◦ .

0169-4332/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.apsusc.2012.04.108
 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881
Membrane distillation (MD) processes or membrane contactors
have many industrial applications, mostly in water desalination
[11–16] and wastewater treatment [17–20]. In the MD process,
the liquid entry pressure (LEP), defined as the minimum trans-
membrane pressure that is required for process liquid to enter the
pores of the membrane, should be as high as possible. A high LEP
reduces the occurrence of pore wetting that can lead to the deteri-
oration of the quality and rate of permeate production. The LEP is
a characteristic of each membrane, which is calculated as follows:
−2BL cos 
LEP =
rmax
where B is a membrane tortuosity factor with a value that is
typically assumed to be 2,  L is the surface tension of the solu-
tion with a value of 69.63 mN/m for water at 40 ◦ C,  is the CA
between the solution and the membrane surface, and rmax is the
largest membrane pore radius. Eq. (1) indicates that a high LEP
can be achieved by using materials of high hydrophobicity (i.e.,
large contact angles with feed solutions) and small maximum pore
radius. If the maximum pore radius of a membrane is 1 ␮m [2],
the LEPs of such a membrane with contact angles of 95◦ and 150◦
calculated from Eq. (1) are 24 and 241 kPa, respectively, which
would be the upper application pressure limits. These calculations
demonstrate the large effect of the CA on the LEP and subse-
quently on the maximum operating pressures. Therefore, to avoid
wetting, the hydrophobicity of a membrane is very important in
commercial design [15]. Thus, the preparation of a hydrophobic
microporous membrane surface is an actively pursued goal in the
current development of membrane science. El-Bourawi et al. [21]
and Khayet [22] have reviewed the characteristics of most of the
commercially available membranes used in MD processes. These
membranes are mainly made of polypropylene (PP), polyethy-
lene (PE), PVDF and polytetrafluoroethylene (PTFE). PVDF is a
commercially available fluoropolymer with a low surface energy
(25 dynes/cm), and microporous hydrophobic PVDF membranes
prepared by the phase inversion method have been used success-
fully for many years in MD and membrane contactors. An increasing
number of researchers are focusing on how to prepare microporous
and hydrophobic PVDF membranes [23–30]. Superhydrophobic
surfaces reported in the literature are mainly found in materials in
which both hydrophobicity and surface roughness have been com-
bined. Xie et al. [9] and Peng et al. [10] prepared porous and highly
hydrophobic PVDF surfaces with certain micro- and nanoscale hier-
archical roughnesses via VIPS. Unfortunately, the ambient RH was
not controlled in these studies, and the formation mechanism of the
hydrophobic surfaces was not further discussed. Therefore, more
research is needed to determine how the RH of the environment
and other conditions influence the micro- and nanostructures and
hydrophobicity.
To date, most studies on VIPS have focused on the phase sepa-
ration mechanism and the mass transformation during membrane
formation. To the best of the authors’ knowledge, few studies
have concentrated on the membrane surface structure changes
occurring in VIPS, especially the hydrophobicity changes in PVDF.
This study therefore investigated the membrane surface hydropho-
bicity changes in VIPS, in which a wide range of membrane
morphologies was produced by controlling the phase-separation
process. The major factor affecting the morphology of the PVDF
membrane, namely, exposure time and the mechanisms that gov-
ern their formation are discussed. The membrane morphology,
hydrophobicity and crystallinity were investigated. The membrane
performance was investigated in vacuum membrane distillation
(VMD).
7873
(1)
Fig. 1. Diagram of the saturated air generator.
2. Experimental
2.1. Materials
PVDF was obtained from ShangHai New Materials Company
Limited (3FNM, FR-904, intrinsic viscosity1.4–1.9 dL/g). DMAc
(analytical reagent) was purchased from Tianjin Yongda Chemical
Reagent Co. (China). A commercial PVDF membrane with pore size
0.1 ␮m was provided by Beijng Institute of Plastic Research.
2.2. Membrane preparation
PVDF membranes were prepared via a VIPS method. 15 wt.%
PVDF (predried in an oven at 120 ◦ C) was dissolved in DMAc at
room temperatures. After the solution became homogeneous, it
was degassed overnight at room temperature. The degassed poly-
mer solution was cast onto a glass plate to a thickness of 250 ␮m,
and the cast film was exposed to humid air at 25 ◦ C over the planned
time interval; the cast film was then immersed into a water bath to
allow for solidification of the resulting morphology. Thirty minutes
later, the solidified film was transferred to another water bath for
24 h to remove residual DMAc. The membrane was then dried in
air at room temperature. The time that the cast films were exposed
to the humid air was varied between 0 min and 10 min. The bath
temperature was kept constant at 25 ◦ C.
In VIPS, the RH of the humid air is very important; in this study,
it was controlled at 100% in the device shown in Fig. 1 during mem-
brane formation. Briefly, a Pyrex© glass chamber (custom fabricated
in our laboratory) was placed in a thermostatic water bath (HH-I,
RongHua Apparatus Manufacturer), and a grille with open area of
60% was placed between the chamber and the water bath. There
was a movable drawer in the chamber that was used to support the
glass plate and the cast film. Before the glass plate with the film
was placed into the drawer, the temperature in the water bath was
set to 40 ◦ C. Water continuously evaporated, and thus, the RH in
the chamber increased. The temperature in the chamber was mon-
itored with the temperature probe and was shown on the panel,
which was 25 ◦ C. When water droplets were observed on the roof,
suggesting that the air in the chamber was saturated with water,
the RH value was assumed to be 100%. The glass plate with the cast
film was then placed in the drawer, and the drawer was immedi-
ately pushed back into the chamber. After a planned exposure time
period, the glass plate was removed and placed in the water bath.
7874 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881
2.3. Mass transfer
The glass plate with the cast film was placed on a balance scale
(AR5120, Ohaus, USA) in the Pyrex© glass chamber described above
with a 100% RH value. The mass variation of the cast film was
recorded in situ every 10 min, and the data were saved automati-
cally by a computer connected to the balance.
2.4. Membrane characterization
2.4.1. Morphological characterization
The morphology of the top surface (defined as the surface in con-
tact with the coagulation bath or humid air), the bottom surface (the
surface in contact with the glass plate), and the cross-section of the
sample films was characterized by scanning electron microscopy
(SEM, JSM-5510LV, JEOL, Japan; FEI Quanta 200, Holland; Hitachi,
S-4300, Japan) operated at 5–20 kV. Prior to the observations, all
samples were coated with gold in vacuum.
2.4.2. Static contact angle
The static CA of the PVDF membrane was measured using the
sessile drop method with a contact angle meter (DSA100, Krüss,
Germany). The water droplets used for the measurements were
5 ␮L in volume. The CA values of each sample were measured at
five different positions of one sample and then averaged.
2.4.3. Mechanical properties
Wet porous membranes were cut into the standard dumbbell
shape for the measurements of tensile strength and modulus. The
tensile strengths and elastic moduli at the breaking point were
measured using an Instron 4302 universal tensile testing machine
under ambient conditions for at least three samples, and the aver-
age values were recorded.
2.4.4. Differential scanning calorimeter (DSC)
Thermal behaviors of the membranes were determined by a
differential scanning calorimeter (PerkinElmer Pyris-1 DSC, USA)
calibrated with indium. The membrane crystallization was char-
acterized by DSC at 10 ◦ C/min increasing temperature rate. About
10 mg sample was used. Melting heat Hm was determined from
the melting peak area. Crystallinity was evaluated by:
H
DSC =
H100
H100 = 104.7 J/g is the melting heat for a 100% crystalline sample
of PVDF [31].
2.4.5. Wide-angle X-ray diffraction (WXRD)
Crystalline structures of the overall membranes were investi-
gated by WXRD measurements. WAXD patterns with 2 angles
ranging from 10◦ to 50◦ were obtained on a rotating-anode X-ray
diffractometer (Bruker D8 Advance, Germany). All the X-ray diffrac-
tometer data given below were recorded by with voltage 40 kV and
current 40 mA applied to the X-ray rotating-anode.
2.4.6. X-ray photoelectron spectroscopy (XPS)
The XPS analyses were carried out with a PHI Quantera SXM
(ULVAC-PH INC) spectrometer using a monochromatic Al K␣ source
(20 mA, 2.0 kV). Binding energy accuracy is ±0.5 eV and an analysis
area of 1 mm × 1 mm.
2.4.7. Mean pore diameter and pore size distribution
The pore size distribution of all membrane samples was deter-
mined by a modified bubble point method as described elsewhere
[32].
2.4.8. Porosity
The total porosity (Ak , %) was calculated as a function of the
membrane weight as follows:
W1 − W2
Ak = × 100% (3)
A × ı × H2 O
where W1 and W2 are the equilibrium weights of a membrane in the
swollen and dry states, respectively, A is the area of the membrane,
ı is the average thickness of the membrane, and H2 O is the density
of water.
2.4.9. VMD process
A flat-sheet VMD configuration with an area of 26.4 × 10−4 m2
was used to evaluate the performance of membranes prepared in
this experiment and a purchased PTFE membrane. Fig. 2 shows a
schematic diagram of the VMD experimental apparatus. The hot
feed flow was controlled by a peristaltic pump (BT300-1J, Baod-
ing Longer Precision Pump Co, Ltd.). The temperature of the feed
was controlled by a thermostat water bath. The hot water temper-
ature inside the configuration was measured, after steady state was
reached, by a sensor connected to a digital temperature gauge with
an accuracy of ±0.1 ◦ C. A vacuum pump (ZXZ-2, Linhai Tan Vac-
uum Equipment Co, Ltd.) was connected to the permeate side of the
configuration to remove the vapor. The vacuum in the downstream
was measured with a vacuum gauge with an accuracy of ±0.1 kPa.
The feed temperature, feed flow and vacuum were maintained at
50 ◦ C, 48 L/h, and 40 kPa, respectively. A filter flask connected to a
chiller was used to collect the permeate water. The VMD flux was
calculated in every case by weighing the permeate water collected
in the filter flask for a predetermined period. Practically all VMD
experiments were repeated twice to ensure reproducibility of the
measurements.
A 3.5 wt.% NaCl aqueous solution was employed as feed, to assess
the membrane flux and NaCl rejection rate. A conductivity indica-
tor was used to measure NaCl concentration in the feed solution
and permeate solutions. Because of the good linear relationship
between concentration and conductivity, the permeate concentra-
tion can be calculated. The NaCl rejection rate (R) was determined
as follows
Cf − Cp
R= × 100 (4)
Cf
where Cf , Cp are the NaCl concentration in the feed and permeate
(2)
solution respectively.
3. Results
3.1. Membrane structure
The cast films were exposed to humid air (25 ◦ C, 100% RH) over
the planned time interval and then immersed in a water bath. The
exposure time was varied between 0 min and 10 min. The depen-
dence of the final membrane cross-section, top and bottom surface
morphologies on the exposure time are shown in Figs. 3–5, respec-
tively. To clearly show the structures formed by VIPS, the walls
of the microvoids in the cross-section shown on the left side of
Fig. 3 are magnified and displayed on the right side of Fig. 3; great
differences in the structures of the films prepared with different
exposure times were readily observed by SEM.
When the exposure time was close to 0 min, the typical large,
digitate macrovoids were formed (Fig. 3a-1), the top surface was
dense and smooth, and no micropores were observed by SEM
(Fig. 4a), which is the characteristic structure obtained with the wet
process (i.e., direct immersion of the cast film into a water bath).
Furthermore, the walls of the microvoids showed a cellular struc-
ture (Fig. 3a-2). As the exposure time was increased to 3 min, the
 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881
Fig. 2. Schematics of the experimental setup used for the VMD experiments.
digitate macrovoid structures remained but became shorter; the
bottom of the macrovoids still had a cellular structure, although the
cells were slightly larger with the exposure time of 3 min than those
with 0 min. With an exposure time of 5 min, the digitate macrovoids
completely disappeared and the cross-section showed a spherulitic
morphology (Fig. 3c-1), but a large amount of cellular structure
remained among the particles (Fig. 3c-2). The aggregate spheres
were mostly 1–2 ␮m in diameter. With an exposure time of 6 min,
the cellular structure again decreased but still persisted (Fig. 3d-2).
When the exposure time was further increased to 10 min, the cellu-
lar structure completely disappeared (Fig. 3e-2). No obvious change
was observed in the cross-sectional membrane structure when the
exposure time was increased from 5 min to 10 min, as shown in
Figs. 3c-1, d-1 and e-1, except that the particles were larger at the
longer exposure time.
It was clearly observed that the large crystals were composed
of many smaller crystals (Fig. 4f). On the surface of the microscale
aggregate spheres, we observed nanoscale multimolecular micellar
spheres. The microstructured surface of these micellar aggregates
of PVDF is very similar to the micro-nanoscale binary structure of
the lotus leaf [33]. This endowed the PVDF surfaces produced via
VIPS with a bionic “lotus effect”. All of the bottom surfaces showed
the particulate structure illustrated in Fig. 5, which was beneficial
to hydrophobicity. The CAs were between 113◦ and 132◦ .
3.2. Performance in membrane distillation
The membranes described in Section 3.1 were used in VMD
experiments. The salt rejection rate was between 99.6% and 99.8%
for all of the tested membranes. Fig. 6 shows how the trends of
changing permeate flux depended on the CA and the porosity of the
membranes, indicating that the permeate flux generally increased
with increasing membrane hydrophobicity and porosity. When
the CA was increased from 84.5◦ to 142.5◦ and the porosity was
increased from 73.4% to 87.3%, the flux increased from 1800 g/m2 h
to 7000 g/m2 h.
The permeate flux was not only affected by the hydrophobicity
and porosity of the membranes but also by the pore diameter of the
membranes. Table 1 presents the pore diameters, porosities and
CAs of membranes when the cast film was exposed to the humid
air for different period time. The average pore diameter did not
7875
change when the exposure time was increased, but the porosity
increased greatly. Membranes Nos. 1, 2, 3 and 4 had similar CAs,
but the permeate flux increased with increasing porosity. For the
three membranes Nos. 4–6, the flux increased greatly; for these
membranes, the porosity only increased from 83.4% to 85.4%, a
net increase of 2.0%, so the hydrophobicity increase played the
major role. Because membranes Nos. 7 and 8 possessed similar
morphologies (Fig. 3d and e), porosities and CA values (Table 1),
they had approximately the same permeate fluxes. The permeate
flux increases were due to the combined porosity increases and
hydrophobicity improvements.
Membrane No. 8 possesses the best performance; the highest
permeate flux and about 100% NaCl rejection rate, and which was
chosen to undergo long time VMD test. It was found that the per-
meate flux and NaCl rejection rate nearly did not change in 4 h,
in other words, no wetting happened in membrane No. 8. But the
NaCl rejection rate decreased sharply for the commercial PVDF
membrane in the second hour, which is the result of wetting. The
difference of membrane NO. 8 and commercial membrane NO. 9 is
no hydrophilic pore-forming agents like LiCl and PEG were added
in the casting solution when membrane No. 8 was prepared, and
the pores in membrane were formed in VIPS process via PVDF crys-
tallization.
4. Discussions
Hydrophobic PVDF membranes with water CAs greater than
140◦ were produced by controlling the exposure time in VIPS. The
micro-nano bionic spheres and the low surface energy of the nano-
structure together contributed to the high CA on the PVDF surfaces.
Detail discussions were presented in this section.
4.1. Mass transfer and phase separation in exposure stage
The morphology change can attribute to the slow membrane
formation process in VIPS. For such a membrane formation process,
the membrane structure could form either in the humid air by
VIPS or in the coagulation bath via a phase separation induced by
liquid water. During the exposure to humid air, at a constant RH,
the chemical potential of water in the gaseous phase is constant;
therefore, the water intake rates were similar for all casting
7876 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881

Fig. 3. Dependence of the membrane cross-sectional structure on the exposure time that the PVDF/DMAC (15 wt.% PVDF) cast film was exposed to humid air (room
temperature, 100% RH). Exposure time: (a) 0 min, (b) 3 min, (c) 5 min, (d) 6 min, (e) 10 min. Magnification: 1k× (left column); 10k× (a-2 to d-2), 30k× (e-2) (right column).
 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881 7877

Fig. 4. Dependence of the membrane top surface structure on the exposure time that the PVDF/DMAc (15 wt.% PVDF) cast film was exposed to humid air (room temperature,
100% RH). Exposure time: (a) 0 min, (b) 3 min, (c) 5 min, (d) 6 min, and (e) 10 min. Magnification: 10k× (a). f is the magnification of the particle in e (100k×) The up-right
insert in each photos is the profile of a water droplet on the surface with a CA of (a) 84.5◦ ; (b)88.3◦ ; (c)107.3◦ ; (d)138.2◦ ; (e)142.5◦ .

solutions at the various exposure time. The content of absorbed
water in the cast film thus depends only on the exposure time,
which controls the phase separation mechanism.
During the exposure stage, there are two types of mass transfer:
water absorption into the film by the hygroscopic solvent (DMAc)
and the loss of DMAc from the casting film due to evaporation.
These two mass fluxes are evidenced by the differing mass varia-
tions in the cast films with time (Fig. 7). The mass variations are
reported as percentages of the initial mass of the casting solution.
At the beginning of the exposure step, it is clear that the predom-
inant phenomenon was water absorption, leading to the observed
mass increase. After the film reached the maximum mass, solvent
and water evaporation prevailed over the water sorption, thereby
causing a decrease in weight. Finally, when all of the DMAc and
water that could be evaporated under the applied drying condi-
tions were removed, the film mass was constant. We observed that

Table 1
Pore diameters and porosities of membranes when the film was exposed to humid air for different times (with a casting solution of 15% PVDF and 85% DMAc).

Membrane Exposure time to Water contact Average pore Porosity

number the humid air (min) angle (◦ ) diameter (␮m) (%)

1 0 84.5 0.11 73.4

2 1 86.2 0.11 74.1
3 2 88.3 0.11 82.7
4 3 90.8 0.11 83.4
5 4 107.3 0.11 84.2
6 5 138.2 0.11 85.4
7 6 140.1 0.11 87.1
8 10 142.5 0.11 87.3
9a 140.3 0.13 80.5
a
Commercial PVDF membrane.
7878 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881

Fig. 5. Dependence of the membrane bottom surface structure on the exposure time that the PVDF/DMAC (15 wt.% PVDF) cast film was exposed to humid air (room
temperature, 100%RH). Exposure time: (a) 0 min, (b) 3 min, (c) 5 min, (d) 6 min, and (e) 10 min. Magnification: 10k×. The up-right insert in each photos is the profile of a
water droplet on the surface with a CA of 115.8◦ (a); 113.3◦ (b); 115.7◦ (c); 126.9◦ (d); 132.1◦ (e).

20
10
0
-10
-20
Mass variation(%)

-30
-40
-50
-60
-70
-80
-90
-100
0 5 10 15 20 25
time(h)

Fig. 7. Mass variation of the PVDF forming system as a function of exposure time.
The inset shows the mass variation during the first five hours.
Fig. 6. Influence of contact angle and porosity on the permeate flux in VMD.
 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881 7879

the final mass of the cast film was approximately the same when
the polymer content in the casting solution was constant.
The measured value for this absorption rate at 100% RH was
approximately 4.66 × 10−6 g/cm2 s. This value represents the dif-
ferences between the water intake and solvent evaporation fluxes.
During the mass decrease, the rate of mass variation became nearly
constant. The measured value for this rate 100% RH was approxi-
mately 1.20 × 10−6 g/cm2 s. Because the water content in the films
was low, the mass loss due to water evaporation is negligible com-
pared with that due to DMAc evaporation. The above absorption
rate plus the evaporation rate is the water intake rate, which was
5.86 × 10−6 g/cm2 s at 100% RH. From these values, we calculated
that the water intake was approximately 1.41 × 10−3 g/cm2 after
4 min of contact with the humid air at a RH of 100%. We know
that less than 5 wt.% water is required to induce phase separation
when the PVDF content is 10% in a casting solution at 25 ◦ C. A water Fig. 8. The dependence of the tensile strengths of the membranes on exposure time.
intake of 1.41 × 10−3 g/cm2 can thus cause phase separation in
2.82 × 10−2 g of solution per square centimeter of film, correspond-
ing to a region with a thickness of approximately 262 ␮m, based on
a casting solution density (PVDF 15%, DMAc 85%) of 1.085 g/cm3 . On
the basis of the above calculations, we concluded that with a 4 min
exposure and a RH of 100%, the water intake from the humid air
was sufficient to generate a phase separation over the entire film
cross-section.
When the exposure time was not more than 4 min, for example,
3 min, the absorbed water was insufficient to cause phase sepa-
ration in the entire film; the composition was in the immiscible
region. Once the film was immersed into the water bath, NG was the
predominant mechanism, resulting in a dense skin (Fig. 4b) and the
typical digitate bulk and cellular macrovoid walls (Fig. 3b-1). Once
the exposure time was more than 4 min, for example, 5–10 min,
the absorbed water was enough to cause phase separation in the
entire film. We did observe the film turned opaque when the expo-
sure time was longer than 4 min. Because mass transfer was slow
in exposure stage, with the water intake from air, PVDF had suf-
ficient time to crystallize. Crystallization demixing dominated in
VIPS, and the phase separation (crystallization) occurred due to the
Fig. 9. Effect of exposure time on melting curve pattern.
water absorbed from the humid air. This led to a porous skin and
small crystalline particle morphology (Fig. 3c-1, d-1 and e-1). With
an exposure time of 5 min, the PVDF crystals did not have suffi- for each membrane. It can be clearly seen that the crystallinity of
cient time to grow before the film was immersed in the water bath, membranes with longer exposure time show higher crystallinity
so they remained small (Figs. 3c-2 and 4c). When the exposure except the membrane with 3 min exposure time. This indicates that
time was increased, the crystal size also increased. Besides VIPS, long exposure time is favorable to crystallization in VIPS. Because
PVDF can also be crystallized in a “soft” coagulation bath, such as slower mass exchange between solvent and non-solvent will delay
water-DMAc [10], 1-octanol [34], or ethanol [35].
The residual cellular structure was formed of those PVDF
domains which had not crystallized by NG phase separation in the
immersion stage (Fig. 3c-2). Thus, a longer exposure time results
in less uncrystallized PVDF and therefore a less cellular structure
(Fig. 3d-2). A porous skin and small crystallite particle morphology
is highly detrimental to the mechanical strength of a membrane.
The tensile strengths of the membranes decreased sharply with
increasing exposure time, as shown in Fig. 8.

4.2. Crystallization and micro-nano bionic spheres

Crystallization plays a very important role in the membrane

morphology formation. Because the casting solution was prepared
at room temperature and the casting environment was also main-
tained at room temperature, crystallization by quenching was
impossible; in other words, TIPS did not occur. The membrane crys-
tallization behaviors were demonstrated by DSC (Fig. 9). It can be
seen that the membranes exhibit a similar melting behavior with
one broad melting peak at 155–158 ◦ C.
The total crystallinity was calculated by melt heat (Fig. 10). The
obtained data (Fig. 10) were average values of triplicate samples Fig. 10. Total crystallinities of PVDF membranes.
7880 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881

Fig. 11. Effect of exposure time on WXRD.

liquid–liquid demixing, which might provide more time for crys- separation mechanism and the fast L–L demixing existed, the crys-
tallization process to yield the membrane, and crystallization is an tallinity was low.
important phase separation mechanism in VIPS.
WAXD measurements were performed to examine the crys- 4.3. Low surface energy
talline structures of the overall membranes. As shown in Fig. 11, the
membranes formed with different exposure time exhibit a similar During the exposure stage, with the evaporation of DMAc and
diffraction pattern except the membrane with 3 min exposure time. the adsorption of water, phase separation begins, and a lot of micro-
The predominant presence of ␣-phase is conformed by the occur- nano bionic spheres formed. Because of the low surface energy of
rence of two most distinctive diffraction peaks at around 18.3◦ and fluorocarbons, the fluorine component in PVDF molecules tends to
19.9◦ [36]. self-enrich on the outer surface of the micro-nano bionic spheres.
The total crystallinity was determined by deconvolution of Because the area of air–liquid intersurface on the top surface of
diffraction peaks into contributions from amorphous and crys- membrane is larger than that in the bulk, more fluorine com-
talline regions, following a method described in the literature ponent self-enrich on the top of the polymer film, resulting in
[37]. It is noted that the total crystallinity values measured by the low surface energy of the bionic polymeric film. We mea-
WAXD are lower than those from DSC measurements. The dis- sured the chemical composition near the PVDF membrane surface
crepancy between WAXD and DSC results may be associated with with X-ray photoelectron spectroscopy (XPS) at different depths
the presence of some imperfect crystals and part of the crystalline (Fig. 12), and the results showed that the ratio of the number of
amorphous interphase, which contribute to the DSC estimate but fluorine atoms ( CF2 ) to the total number of fluorocarbon atoms
are not probed by WAXD. This discrepancy may also because WXRD
is a tool analyzing the crystalline structure at the membrane sur-
face, while DSC is for the whole membrane; it may be possible that
for the membranes the crystallinities are different at the surface
and in the bulk; and the crystallinity at the surface is smaller than
in the bulk. Nevertheless, the trends of WAXD and DSC crystallinity
are in good agreement.
In addition, we can find that the membrane with 3 min exposure
time is quite different from the others in DSC and WXRD patterns
and the crystallinity determined from DSC and WXRD results. The
following may be the reason. For the membrane with 3 min expo-
sure time, the membrane forming mechanism is different from the
membranes without exposure time (0 min, NG mechanism) and
long exposure time (5–10 min, crystallization). When the film was
exposed to the humid air, at 3 min exposure time, its composition
was in the immiscible region, no phase separation happened. But
the water percentage in the film was higher than that of the film
without exposure time, that is to say the composition was closer to
phase separation line, and less time was needed to start NG phase
separation in coagulation bath. Fast L–L demixing led to a dense top
surface (Fig. 4b) and a typical digitate bulk and cellular macrovoid Fig. 12. Plot of the ratio of the number of fluorine atoms to the total number of
walls (Fig. 3b-1). Because crystallization was not the main phase fluorocarbon atoms against XPS detection angle.
 Y. Peng et al. / Applied Surface Science 258 (2012) 7872–7881
( CF2 and CH2 ) increased near top surface when the exposure
time was 10 min, which showed that during exposure stage, fluo-
rine atoms had enough time to move to the top surface to low the
surface energy of membrane.
5. Conclusions
The porous and spherulitic morphology resulting from PVDF
crystallization favored a hydrophobic surface in VIPS. With the
increase in exposure time, the phase separation process varied
from L–L demixing to crystallization. The membrane morphology
changed from typical large digitate macrovoids and dense and
smooth skin with a CA of approximately 90◦ (at less than 3 min of
exposure time) to a spherulitic morphology (at longer than 5 min
of exposure time). The CA reached 142.5◦ when the exposure time
was 10 min. The micro-nano bionic spheres and the low surface
energy of the nanostructure together contributed to the high CA
on the PVDF surfaces. The permeate flux increase in VMD was due
to the combined effects of the increased porosity and improved
hydrophobicity
Acknowledgments
The authors would like to thank Professor Liu Zhongzhou of
the Research Center for Eco-Environmental Science, the Chinese
Academy of Sciences for his valuable discussions and suggestions.
The authors also thank National Natural Science Foundation of
China for financial support (21176008).
References
[1] H.C. Park, Y.P. Kim, H.Y. Kim, Y.S. Kang, Membrane formation by water vapor
induced phase inversion, Journal of Membrane Science 156 (1999) 169.
[2] H.J. Lee, B. Jung, Y.S. Kang, H. Lee, Phase separation of polymer casting solution
by non-solvent vapor, Journal of Membrane Science 245 (2004) 103.
[3] Y.S. Su, C.Y. Kuo, D.M. Wang, J.Y. Lai, A. Deratani, C. Pochat, D. Bouyer, Interplay
of mass transfer, phase separation, and membrane morphology in vapor-
induced phase separation, Journal of Membrane Science 338 (2009) 17.
[4] C.L. Li, D.M. Wang, A. Deratani, D. Quémener, D. Bouyer, J.Y. Lai, Insight into the
preparation of poly(vinylidene fluoride) membranes by vapor-induced phase
separation, Journal of Membrane Science 361 (2010) 154.
[5] J.T. Tsai, Y.S. Su, D.M. Wang, J.L. Kuo, J.Y. Lai, A. Deratani, Retainment of pore
connectivity in membranes prepared with vapor-induced phase separation,
Journal of Membrane Science 362 (2010) 360.
[6] A. Ripoche, P. Menut, C. Dupuy, H. Caquineau, A. Deratant, Poly(ether imide)
membrane formation by water vapour induced phase inversion, Macromolec-
ular Symposium 188 (2002) 37.
[7] H. Matsuyama, M. Teramoto, R. Nakatani, T. Maki, Membrane formation via
phase separation induced by penetration of non-solvent from vapor phase. I.
Phase diagram and mass transfer process, Journal of Applied Polymer Science
74 (1999) 159.
[8] X. Gao, L. Jiang, Water-repellent legs of water striders, Nature 432 (2004) 36.
[9] Q. Xie, J. Xu, L. Feng, L. Jiang, W. Tang, X. Luo, C. Han, Facile creation of a super-
amphiphobic coating surface with bionic microstructure, Advanced Materials
16 (2004) 302.
[10] M. Peng, H. Li, L. Wu, Q. Zheng, Y. Chen, W. Gu, Porous poly(vinylidene fluo-
ride) membrane with highly hydrophobic surface, Journal of Applied Polymer
Science 98 (2005) 1358.
[11] J. Zhang, N. Dow, M. Duke, E. Ostarcevic, J.D. Li, S. Gray, Identification of material
and physical features of membrane distillation membranes for high perfor-
mance desalination, Journal of Membrane Science 349 (2010) 295.
[12] S. Al-Obaidani, E. Curcio, F. Macedonio, G.D. Profio, H. Al-Hinai, E. Drioli, Poten-
tial of membrane distillation in seawater desalination: Thermal efficiency,
sensitivity study and cost estimation, Journal of Membrane Science 323 (2008)
85.
7881
[13] M. Khayet, J.I. Mengual, T. Matsuura, Porous hydrophobic/hydrophilic com-
posite membranes Application in desalination using direct contact membrane
distillation, Journal of Membrane Science 252 (2005) 101.
[14] Z. Wang, F. Zheng, S. Wang, Experimental study of membrane distillation with
brine circulated in the cold side, Journal of Membrane Science 183 (2001)
171.
[15] J. Zhang, J.D. Li, M. Duke, Z. Xie, S. Graya, Performance of asymmetric hollow
fibre membranes in membrane distillation under various configurations and
vacuum enhancement, Journal of Membrane Science 362 (2010) 517.
[16] G.W. Meindersma, C.M. Guijt, A.B. de Haan, Water Recycling, Desalination
by air gap membrane distillation, Environmental Progress 24 (4) (2005)
434.
[17] J. Phattaranawik, A.G. Fane, A.C.S. Pasquier, W. Bing, F.S. Wong, Experimental
Study and design of a submerged membrane distillation bioreactor, Chemical
Engineering and Technology 32 (1) (2009) 38.
[18] Y. Wu, Y. Kong, J. Liu, J. Zhang, J. Xu, An experimental study on membrane
distillation -crystallization for treating waste water in taurine production,
Desalination 80 (1991) 235.
[19] Z. Wang, Z. Gu, S. Feng, Y. Li, Application of vacuum membrane distillation
to lithium bromide absorption refrigeration system, International Journal of
Refrigeration 32 (2009) 1587.
[20] P.P. Zolotarev, V.V. Ugrozov, I.B. Volkina, V.M. Nikulin, Treatment of waste water
for removing heavy metals by membrane distillation, Journal of Hazardous
Materials 37 (1994) 77.
[21] M.S. El-Bourawi, Z. Ding, R. Ma, M. Khayet, A framework for better understand-
ing membrane distillation separation process, Journal of Membrane Science
285 (2006) 4.
[22] M. Khayet, Membranes and theoretical modeling of membrane distillation: a
review, Advances in Colloid Interface Science, in press.
[23] B. Li, K.K. Sirkar, Novel membrane and device for direct contact membrane
distillation-based desalination process, Industrial and Engineering Chemistry
Research 43 (2004) 5300.
[24] B. Li, K.K. Sirkar, Novel membrane and device for vacuum membrane
distillation-based desalination process, Journal of Membrane Science 257
(2005) 60.
[25] C. Feng, R. Wang, B. Shi, G. Li, Y. Wu, Factors affecting pore structure and per-
formance of poly(vinylidene fluoride-co-hexafluoro propylene) asymmetric
porous membrane, Journal of Membrane Science 277 (2006) 55.
[26] N. Tang, Q. Jia, H. Zhang, J. Li, S. Cao, Preparation and morphological characteri-
zation of narrow pore size distributed polypropylene hydrophobic membranes
for vacuum membrane distillation via thermally induced phase separation,
Desalination 256 (2010) 27.
[27] M. Khayet, T. Matsuura, J.I. Mengual, Porous hydrophobic/hydrophilic com-
posite membranes: estimation of the hydrophobic layer thickness, Journal of
Membrane Science 266 (2005) 68.
[28] A. Gugliuzza, F. Ricca, E. Drioli, Controlled pore size, thickness and surface free
energy of super-hydrophobic PVDF and Hyflon AD membranes, Desalination
200 (2006) 26.
[29] M. Khayet, Membrane surface modification and characterization by X-ray
photoelectron spectroscopy, atomic force microscopy and contact angle mea-
surements, Applied Surface Science 238 (2004) 269.
[30] S. Bonyadi, T.S. Chung, Flux enhancement in membrane distillation by fabrica-
tion of dual layer hydrophilic–hydrophobic hollow fiber membranes, Journal
of Membrane Science 306 (2007) 134.
[31] C. Marega, A. Marigo, Influence of annealing and chain defects on the melting
behaviour of poly(vinylidene fluoride), European Polymer Journal 39 (2003)
1713.
[32] G. Capannelli, F. Vigo, S. Munari, Ultrafiltration membranes: characterization
methods, Journal of Membrane Science 15 (1983) 289.
[33] L. Feng, S. Li, Y. Li, H. Li, L. Zhang, J. Zhai, Y. Song, B. Liu, L. Jiang, D. Zhu, Super-
hydrophobic surfaces from natural to artificial, Advanced Materials 14 (2002)
1857.
[34] T.H. Young, L.P. Cheng, D.J. Lin, L. Fane, W.Y. Chuang, Mechanisms of PVDF
membrane formation by immersion-precipitation in soft (1-octanol) and harsh
(water) non-solvents, Polymer 40 (1999) 5315.
[35] L. Qian, Z.L. Xu, J. Lu, Effect of coagulation bath composition and temperature
on morphology and property of microporous PVDF hydrophobic membrane,
Journal of Chemical Engineering of Chinese Universities (Chinese) 24 (2) (2010)
336.
[36] M. Zhang, A.Q. Zhang, B.K. Zhu, C.H. Du, Y.Y. Xu, Polymorphism in porous
poly(vinylidene fluoride) membranes formed via immersion precipitation pro-
cess, Journal of Membrane Science 319 (2008) 169.
[37] Y.S. Park, T. Hatae, H. Itoh, M.Y. Jang, Y. Yamazaki, High proton-conducting
Nafion/calcium hydroxyphosphate composite membranes for fuel cells, Elec-
trochimica Acta 50 (2004) 595.