Journal of Membrane Science 542 (2017) 308–319

Contents lists available at ScienceDirect

Journal of Membrane Science

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

Superhydrophobic polysulfone/polydimethylsiloxane electrospun MARK

nanofibrous membranes for water desalination by direct contact membrane
distillation
⁎ ⁎⁎
X. Lia,b, M.C. García-Payob, M. Khayetb,c, , M. Wanga, X. Wanga,
a
State Key Laboratory for Modification of Chemical Fibers and Polymer Materials, Donghua University, Shanghai 201620, People's Republic of China
b
Department of Applied Physics I, Faculty of Physics, University Complutense of Madrid, Avda. Complutense s/n, 28040 Madrid, Spain
c
Madrid Institute of Advances Studies of Water (IMDEA Water Institute), Calle Punto Com no 2, 28805 Alcalá de Henares, Madrid, Spain

A R T I C L E I N F O A B S T R A C T

Keywords: Superhydrophobic electrospun nanofibrous membranes (ENMs) with a fully interconnected porous web structure
Electrospun nanofibrous membrane and high void volume fraction have been proposed for membrane distillation (MD). This study describes a novel
Superhydrophobicity and facile design strategy for the fabrication of superhydrophobic and self-cleaning polysulfone-poly-
Polysulfone dimethylsiloxane (PSU-PDMS) ENMs for direct contact membrane distillation (DCMD) via electrospinning PSU
Polydimethylsiloxane
nanofibers followed by PDMS coating and cold-press post-treatment. A comprehensive investigation of the ef-
Membrane distillation
fects of PDMS concentration and cold-press post-treatment pressure on the membrane surface morphology,
wettability, membrane characteristics and DCMD performance was carried out. The best ENM in this study, PSU-
PDM-1-4, prepared with 1 g PDMS in 40 mL hexane and 4 MPa applied cold-press post-treatment pressure,
exhibited a competitive permeate flux of about 21.5 kg/m2h and a stable low permeate concentration with an
electrical conductivity of 4.224 – 4.523 µS/cm when using 30 g/L NaCl aqueous solution as feed and a tem-
perature difference of 50 °C over a DCMD period of 12 h without detecting inter-fiber space wetting. The pro-
posed PSU-PDMS ENM type of membrane is promising in MD desalination.

1. Introduction as high void volume fraction, interconnected open structure, high sur-
face-to-mass (or volume) ratio, good resistance to the penetration of
Membrane distillation (MD), as an emerging separation process chemical and biological agents, low heat transfer by conduction, highly
with moderate temperature demand, low operation pressure and high ordered polymer chains, more controllable structure than the phase
solute rejection (100% in theory), is an attractive technology for de- inversion hollow fiber and flat-sheet membranes [10–16].
salination, environmental/waste treatment, food industry, etc. [1–5]. In MD process, the surface wettability and structural characteristics
MD is a thermally driven non-isothermal membrane separation process, of the membrane dictate the separation performance. Membrane wet-
in which only vapor and non-condensable gases are transported ting may occur during continuous MD operations. In order to improve
through a highly hydrophobic microporous membrane. The MD driving the wetting resistance, a membrane with significantly high hydro-
force is supplied by the vapor pressure difference resulting from either a phobicity and uniform distribution of small pore size is required. In
temperature difference between both membrane sides or by applying fact, it is possible to prepare ENMs designed specifically for MD with
vacuum or sweeping gas in the permeate side [1,3]. This membrane- superhydrophobic character, high void volume fraction, appropriate
based process can take full advantage of renewable energy, low-grade mean inter-fiber size, narrow size distribution of inter-fiber space, high
and industrial waste heat [4,6,7]. The simplest operation configuration, liquid entry pressure of water (LEPw) and high permeate flux with high
direct contact membrane distillation (DCMD), has been applied fre- quality water permeate during long-term DCMD operation [15–23]. It is
quently to evaluate the desalination performance using different types worth quoting that PVDF and PVDF derivatives (such as polyvinylidene
of membranes [8,9]. fluoride-co-hexafluoropropylene (PVDF-HFP)) are so far the polymers
Recently, some electrospun nano-fibrous membranes (ENMs) have utilized mostly to prepare ENMs for MD, since they can be easily dis-
been proposed for MD owing to their several attractive attributes such solved in common solvents. However, compared to the hydrophilic

⁎
Corresponding author at: Department of Applied Physics I, Faculty of Physics, University Complutense of Madrid, Avda. Complutense s/n, 28040 Madrid Spain.
⁎⁎
Corresponding author at: State Key Laboratory for Modification of Chemical Fibers and Polymer Materials, Donghua University, Shanghai 201620, People's Republic of China.
E-mail addresses: khayetm@fis.ucm.es (M. Khayet), [email protected] (X. Wang).

http://dx.doi.org/10.1016/j.memsci.2017.08.011
Received 27 April 2017; Received in revised form 25 June 2017; Accepted 5 August 2017
Available online 09 August 2017
0376-7388/ © 2017 Elsevier B.V. All rights reserved.
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

filtration membranes, the electrospinnable polymeric materials with 2.2. Electrospinning details
intrinsic hydrophobicity used for MD are extremely limited. Therefore,
much attention should be paid to the exploitation of novel electrospun Typically, 23 wt% of PSU was dissolved in DMF by mild stirring at a
membrane materials [24–27]. For instance, the hydrophilic polyamide temperature of about 60 °C for 24 h to obtain an homogeneous trans-
(PA) and polyvinylalcohol (PVA) ENMs subjected to surface modifica- parent solution. About 5 mL of the PSU/DMF solution was placed in a
tion with fluoride proved to be competent in MD [28,29], as well as the 5 mL syringe connected to a metal needle of 0.37 mm inner diameter.
well-designed aromatic fluorinated polyazole ENMs [30]. The syringe was placed in a pump that maintained a feed solution flow
Nature has provided numerous inspirations for the fabrication of rate of 1.0 mL/h. A grounded metallic plate (with a diameter of 10 cm)
novel biomimetic materials, such as the well-known superhydrophobic covered with aluminum foil was used as collector. The distance be-
surfaces with “petal effect”, “gecko effect” and “lotus effect” [31,32]. It tween the needle tip and the collector was 15 cm and the applied vol-
must be pointed out that the “petal effect” is a superhydrophobic state tage was 20 kV. The relevant temperature and humidity were 25 ± 2 °C
with high adhesive force, i.e., the Cassie impregnating wetting state and 25 ± 5%, respectively. The obtained ENMs were dried in a vacuum
[33]. While the “lotus effect” is a superhydrophobic state with a low oven at 60 °C for 6 h to remove completely the residual solvent. In this
sliding angle (less than 5°), i.e., the self-cleaning character or the Cas- study, the polymer solution with a volume of 1.25 mL was electrospun
sie's state [34]. Thus, with respect to the ENMs with “petal effect” for to obtain the resultant PSU ENMs with the same membrane thickness of
MD applications, the water droplets will be immediately adhered to the about 220 µm.
membrane surface, the air and/or vapor entrapped under the water-
membrane interface will be finally lost during the long-term MD op- 2.3. Superhydrophobic modification of PSU ENMs and cold-press post-
eration, and then the wetting of the surface nanofibrous layer can treatment
occur, even though it exhibits a superhydrophobic property. In contrast,
the ENMs with “lotus effect” or the self-cleaning character, the water The superhydrophobic PSU ENMs were prepared by a typical dip-
droplets will be immediately rolled/slid off the membrane surface coating process. First, PDMS solutions with different concentrations
during the MD process, which is more beneficial to avoid the wetting of were prepared by adding certain contents of the SYLGARD® 184
the membrane pores. (elastomer base and the curing agent) in 40 mL of hexane. The ratio of
Polysulfone (PSU) is an attractive membrane material used in dif- the base PDMS pre-polymer and the curing agent in the solution was
ferent filtration processes due to its excellent chemical resistance, good kept at 10:1 (wt./wt.). Then, the prepared PSU ENMs were dipped in
thermal stability and suitable mechanical properties [35,36]. However, the coating solutions for 5 min. Subsequently, the solvent was evapo-
due to its high propensity to biofouling, various surface modification rated and the ENMs were cured at 80 °C for 12 h to obtain the super-
approaches have been developed to improve its hydrophilicity [37–40]. hydrophobic PSU-PDMS ENMs.
PSU membranes applied in MD applications have been rarely reported. Both the PSU and the modified PSU-PDMS ENMs were cold-pressed
Tian et al.[41] reported highly inter-connective porous PSU membranes (Servosis S.A.) at room temperature under different pressures in the
modified by CF4 plasma for DCMD process. In addition, polysiloxanes range 2–12 MPa for 2 min to improve their dimensional integrity and
are well-known for their hydrophobic property. Polydimethylsiloxane reduce the wetting phenomenon in MD process. The obtained mem-
(PDMS) is a typical elastomeric material with low surface energy of branes are denoted hereafter as PSU-y and PSU-PDMS-x-y ENMs, where
about 20 mN/m. The contact angle of PDMS is reported to be in the x and y refer to the PDMS content (g) in 40 mL of hexane and the ap-
range 100–110° and does not exhibit low sliding angles [42]. Conse- plied pressure during the cold-press post-treatment, respectively.
quently, it is often used for the fabrication of superhydrophobic and/or
self-cleaning surfaces [42,43]. The common considered approaches are: 2.4. ENMs characterizations
(i) to construct a hierarchical roughness on the PDMS substrate [42,43];
(ii) to serve as coating material curing on a rough surface [44,45]. The surface and cross-section morphologies of ENMs were examined
In this study, a novel and facile design strategy for the fabrication by field emission scanning electron microscopy (FE-SEM, JEOL Model
of self-cleaning PSU-PDMS ENMs for DCMD is presented. PSU ENMs JSM-6335F). The ENMs samples were prepared by fracturing in liquid
prepared by electrospinning technique were modified by PDMS nitrogen. Subsequently, all the samples were gold coated by a rotary-
coating, followed by cold-press post-treatment. This ENM prepara- pumped sputter coater (QUORUM Q150R S) during 60 s under 20 mA.
tion method can be easily scaled up due to its simple procedure. The The average fiber diameter of ENMs was measured using the ImageJ 2X
effects of PDMS concentration and the applied pressure during the software analyzer. The diameters of a total number of 200 fibers were
cold-press post-treatment on the membrane surface morphology, measured and an average value together with its standard deviation
surface wettability and structural characteristics have been in- were determined.
vestigated. The obtained superhydrophobic PSU-PDMS ENMs with The water contact angle (θw) of the ENMs samples were measured
appropriate structural properties have been applied in desalination using a computerized CAM100 device (Sb), equipped with a CCD
by DCMD. camera, frame grabber and an image analysis software CAM200usb,
enabling to acquire images of the distilled water drop on the surface of
the sample and to calculate the contact angle value. A Hamilton
2. Experimental stainless steel needle permitted to control the volume of the drop,
which was about (2.0 ± 0.1) µL. For each drop, five images were re-
2.1. Materials corded during 4 s and for each membrane at least 10 drops were con-
sidered for each sample to determine the average θ value together with
Commercial polymer polysulfone pellets (PSU, P-1700 LCD, with its standard deviation.
average molecular weight of 72 000–82 000 g/mol) was purchased The chemical compositions of the ENMs were determined by Fourier
directly from Solvay Co., Ltd. SYLGARD® 184 Silicone Elastomer Kit, Transform Infrared (FTIR) spectroscopy with a Nicolet device (model
with components of polydimethylsiloxane (PDMS) base and curing Magna-IR 750 series II), equipped with a detector DTGS-KBr (triglicerin
agent, purchased from Dow Corning was used for membrane mod- sulfate deuterated with KBr window), a beam splitter KBr and an in-
ification. N,N’-dimethylformamide (DMF), isopropyl alcohol (IPA) and frared source (Ever-Glo) employing an attenuated total reflectance
hexane were supplied by Sigma Aldrich. All chemicals were of analy- (model H-ATR Multiple Bounce, Spectra Tech) with ZnSe crystal and 13
tical grade and were used as received without further purification. steps. The measurements were carried out with 128 scans in the wave
length range 400 − 4000 cm−1 with 8 cm−1 resolution.

309
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

Membrane thickness was measured with a digital micrometer (S.L. the salt rejection factor was determined as stated elsewhere [15].
Isocontrol) with an accuracy of 0.001 mm. At least 20 measurements
along different diameters of three samples were performed for each 3. Results and discussion
ENM prepared under the same conditions. The thickness was de-
termined as the average value with its corresponding standard devia- 3.1. Surface morphology and FTIR characterization of ENMs
tion.
Liquid entry pressure of water (LEPw) measurements of distilled Fig. 2 shows the representative FESEM images of PSU and PSU-
water were carried out using the experimental system detailed else- PDMS-x ENMs, which all exhibited randomly oriented 3D nonwoven
where [15]. The effective area of ENMs samples for the LEPw mea- structure with various surface morphologies. As can be seen in
surements was 12.56 cm2. Each ENM sample was placed in the test cell Fig. 2(A), the PSU nanofibers showed a bead-free and cylindrical
and the container was filled with distilled water. By means of a nitrogen morphology, as well as a typical roughness with micro-grooves along
gas cylinder a slight pressure was applied for 5 min and then increased the fiber axis introduced by micro-wrinkles on the nanofibers surface
stepwise by 0.01 bar. The maximum applied pressure before distilled (zoom images). This unique surface topography can be explained by the
water penetrates into the inter-fiber space is taken as the LEPw value. buckling instabilities associated with the solvent evaporation from the
These measurements were carried out three times using three different surface of the formed jet [47]. The average fiber diameter of PSU na-
membrane samples and the average values together with their standard nofibers was 0.97 ± 0.20 µm (Fig. 2(F)). The water contact angle (θw)
deviations were calculated. of PSU ENMs was 115.9 ± 1.6° (Fig. 3(A)). This value is higher than
The void volume fraction of the ENMs, defined as the volume of the that of the phase inversion PSU membrane, 80° [40]. By incorporating
inter-fiber space related to the total membrane volume, was determined PDMS in the PSU nanofiber, the θw value was increased (see Fig. 3(A))
by measuring the density of each ENM using isopropyl alcohol (IPA), because of the low surface energy of PDMS polymer chain coated and
which penetrates inside the inter-fiber space of the ENMs sample and adhered to the surface of the PSU nanofibers due to the curing cross-
distilled water, which cannot enter into the inter-fiber space [1]. linked reaction between PDMS and the curing agent [45]. As shown in
The size of the mean inter-fiber space, the minimum and maximum Fig. 2(B), the PSU-PDMS-0.5 ENM (0.5 g PDMS in 40 mL hexane) ex-
of the prepared ENMs, as well as the cumulative filter flow curves were hibited a relatively smooth surface as the micro-grooves along the fiber
determined by using a capillary flow porometer (POROLUX™ 100) axis were almost filled with PDMS component resulting in an increase
based on the gas-liquid displacement, within a pressure range of of the average fiber diameter (1.24 ± 0.20 µm), as can be seen in Fig. 2
0–0.7 MPa (compressed air) at a room temperature. POREFIL® was used (F). As it was expected, the θw value of the PSU-PDMS-0.5 ENM was
as the wetting liquid agent. In these measurements, the effective increased significantly to 147.6 ± 3.0° (Fig. 3(A)) due to the low sur-
membrane area was 2.6 cm2. The followed procedure was described face energy of PDMS coating and the intrinsic smoother surface of the
elsewhere [16]. Briefly, the ENM was first wetted by the wetting liquid PDMS modified nanofibrous structure. Regarding the PSU-PDMS-1
and the S-shaped wet curve was obtained by measuring the air flow rate sample shown in Fig. 2(C), slight adhesions among the adjacent nano-
as a function of the applied transmembrane pressure. At least three tests fibers were formed due to the increase of PDMS concentration while the
were performed for each ENM sample and the size of the inter-fiber nanofibers surface became smoother and the average fiber diameter
space was calculated by means of Washburn equation [46]: was increased to 1.39 ± 0.36 µm. By further increasing the PDMS
content (i.e. 2 and 4 g PDMS in 40 mL hexane), PDMS agglomerations
4γ cos θ
P= were gradually generated within the nanofibrous structure as indicated
D (1)
by the red circles in Fig. 2(D) and (E). These agglomerations could
where P is the applied pressure, γ is the surface tension of the wetting decrease the void volume fraction of the resultant PSU-PDMS ENMs
liquid, θ is the contact angle of the wetting liquid (in this case zero) and reducing therefore the MD permeate flux. Furthermore, the average
D is the inter-fiber space size. Subsequently, the air flow rate was fiber diameter of PSU-PDMS-2 and PSU-PDMS-4 decreased to
measured through the dry ENM sample at different pressures to obtain 1.10 ± 0.23 µm and 1.13 ± 0.20 µm, respectively, which might be due
the dry curve. According to Darcy's law, the gas flow rate measured to the PDMS polymer chains that were more disposed to form PDMS
through a dry membrane is proportional to the applied transmembrane agglomerations rather than uniformly adhered on the fiber surface after
pressure and the gas permeability was determined from the slope of the the evaporation of hexane at the higher PDMS concentration, and then
plotted dry curve [46]. resulting in the decrease of the average fiber diameter. In addition, the
increase of the surface roughness reduced the contact area of the water
2.5. Direct contact membrane distillation (DCMD) experiments droplet deposited on the surface structure for the θw measurement
leading an enhancement of the measured θw values [48]. As shown in
The resultant ENMs were applied in desalination by DCMD. The Fig. 3(A), the θw values of the PSU-PDMS-1, PSU-PDMS-2 and PSU-
schematic diagram of the used DCMD experimental setup is presented PDMS-4 ENMs before applying the cold-press treatment were
in Fig. 1. The effective membrane area of the membrane module used 151.7 ± 1.7°, 153.9 ± 2.1° and 155.3 ± 1.6°, respectively. This result
for DCMD experiments was 25 cm2. In this study, DCMD test was indicated the obvious increment of θw toward the increase of the PDMS
conducted with 30.0 g/L NaCl aqueous solution as feed, and the concentration due to the surface roughness effect [49]. Therefore, the
permeate tank was distilled water. The feed temperature was varied promising superhydrophobicity could make PSU-PDMS ENMs good
from 50 to 80 °C by means of a heating system (Techne, model TE-8D) candidates for MD applications.
and the feed solution was circulated at a flow rate of 0.6 L/min by a Fig. 3(B) showed the photographs of the dynamic water-adhesion
double-head peristaltic pump (Watson Marlow 323). The permeate side behavior on the PSU-PDMS-1 ENMs surface. As it can be seen, the water
was maintained at 20 °C by a chiller (Polysciences, model 6206 T) and droplet at the needle tip contacted with the surface could leave it with
the permeate solution was circulated by the double-head peristaltic extremely low adhesion confirming the low surface energy of PDMS
pump with a flow rate of 0.4 L/min. During the DCMD test, the coating on the membrane surface. Furthermore, the self-cleaning
permeate flux was obtained by the weight increase of the permeate tank property of PSU and PSU-PDMS-1 ENMs were conducted by observing
set upon a digital balance (Gibertini Europe 3000), which was con- the cleaning behavior of water droplets (controlled by a dropper to be
nected to a computer that recorded the values of both the mass of the about 10 µL) on the dry membrane surface, which was scattered with
produced water and temperatures every 30 s. The salt concentration of hydrophilic graphene oxide (GO) powder as contaminant and tilted
the feed and permeate solutions was measured during each DCMD ex- with a very low angle of about 5°. For clarity, water droplets were dyed
periment by an electrical conductivity meter (Metrohm model 712) and by methyl blue. As shown in Fig. 3(C), the PSU ENMs surface could not

310
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

Fig. 1. Schematic diagram of the DCMD setup: (1) feed tank, (2)
permeate tank, (3) heating thermostat, (4) chiller, (5) double
head peristaltic pump, (6) membrane module, (7) Pt-100 probes,
(8) flow meter, (9) glass heat exchangers, (10) precision balance,
(11) computer.

get rid of the GO powder due to the fact that water droplets were di- that must be applied onto distilled water before it penetrates into a dry
rectly adhered to the membrane surface. By introducing the PDMS membrane pore “inter-fiber space for ENM” [51]. It is one of the crucial
coating, PSU-PDMS-1 sample exhibited an effective self-cleaning pro- structural characteristics that guarantee the stability of the membrane
cess (i.e. the hydrophilic GO powder could first be dissolved in water performance during long-term MD operation. As shown in Fig. 5(a), the
droplets and then rolled and slid off the surface easily (Fig. 3(D)) fol- PDMS modified ENMs exhibited relatively larger LEPw values than those
lowing the “roll-to-clean” mechanism [45]. of the PSU ENMs subjected to the same applied cold-press post-treat-
FTIR spectroscopy analysis was employed to confirm the surface ment pressure. Moreover, the LEPw of PSU-PDMS-x exhibited an in-
chemical compositions of the prepared ENMs. The results are shown in creasing tendency toward the increment of PDMS concentration, which
Fig. 4. The peaks around 1586, 1502 and 558 cm−1 (C˭C aromatic ring could be attributed partly to the enhancement of the surface hydro-
stretching), 1488 cm−1 (CH3-C-CH3 stretching), 1295 cm−1 (S˭O phobicity. Additionally, with the increase of the cold-press post-treat-
stretching), 1244 cm−1 (C-O-C stretching) and 1151 cm−1 (C-SO2-C ment pressure, the LEPw of all ENMs were gradually increased due to
symmetric stretching) were detected for the typical PSU sample [39]. the formation of a tightly packed and thin nanofibrous structure.
For PDMS, the peaks that appeared at 2962, 2904, and 1257 cm−1 were Fig. 5(b) showed the variation trend of the membrane bulk void
assigned to the symmetric stretching, asymmetric stretching and de- volume fraction of the PSU and PSU-PDMS-x ENMs after cold-press
formation vibrations of the -CH3 group, respectively, while the char- post-treatment. It can be clearly observed that the bulk void volume
acteristic bands at 1058 cm−1 and 1010 cm−1 were assigned to the Si- fraction of all the membranes were decreased with the increase of the
O-Si and the -CH3 rocking and Si-C vibrations appeared at around cold-press post-treatment pressure, which was in accordance with the
788 cm−1 [50]. It is worth noting that all these characteristic peaks of membrane thickness behavior. With the incorporation of PDMS, a sig-
both PSU and PDMS were presented in the PSU-PDMS-x ENMs. Ad- nificant increase of the void volume fraction was observed, but for a
ditionally, with the increase of PDMS concentration, the peak in- PDMS content higher than 0.5 g in 40 mL hexane the void volume
tensities of PDMS in PSU-PDMS-x gradually enhanced, whereas those of fraction was reduced. This could be due to the adhesion of the adjacent
PSU component at 1586, 1502, 1488, 1295, 1151 and 558 cm−1 ex- nanofibers as it was stated previously. Compared to the PSU ENMs, the
hibited a remarkably decreasing tendency. Thus, based on the obtained higher void volume fraction values of the PSU-PDMS-x may be attrib-
FTIR spectra it can be confirmed the successful modification of PSU uted to the physical flexibility of PDMS due to its extremely low Young's
ENMs with PDMS coating. modulus [52,53]. In other words, the PSU-PDMS-x ENMs showed a
relatively lower compactness under the same pressure than the PSU
3.2. Cold-press post-treatment and PDMS concentration effects on MD ENMs. The minimum bulk void volume fraction, 65.5 ± 6.6%, was
characteristics of ENMs obtained for the PSU-PDMS-1-12 ENM prepared under the highest cold-
press post-treatment.
As it was previously mentioned, the PSU ENMs and the modified The effect of the cold-press post-treatment on the membrane
PSU-PDMS ENMs were cold-pressed at room temperature under dif- structural characteristics can be seen in Fig. 6. As it can be seen, the
ferent pressures ranging from 2 to 12 MPa in order to improve their nanofibers of the PSU-PDMS-1-2 could maintain their intrinsic cylind-
dimensional integrity and reduce the wetting phenomenon in MD rical morphology after 2 MPa cold-pressing. As the pressure was in-
process. Their structural parameters, namely, the membrane thickness, creased from 4 to 12 MPa, the PDMS coated nanofibers were gradually
the liquid entry pressure of water (LEPw), the void volume fraction, the fused together. This confirmed the increased tendency of the LEPw va-
mean size of the inter-fiber space and its distribution, and the gas lues with the increase of the cold-press post-treatment pressure. This
permeability of the ENMs were determined. As it was expected, the observed PDMS fusion has a decisive influence on the MD permeability
membrane thickness decreased when the pressure was increased of the ENMs as it will be discussed later on. It must be mentioned that
(Table 1). The effect of the post-treatment pressure on the LEPw and the PSU-PDMS-4 (4 g of PDMS in 40 mL hexane) ENM having a surface
void volume fraction values for PSU, PSU-PDMS-0.5, PSU-PDMS-1 and full of PDMS agglomerations and fusion of PDMS coated nanofibers,
PSU-PDMS-2 ENMs is shown in Fig. 5. LEPw is the minimum pressure even when it was subjected to the lowest cold-press post-treatment

311
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

Fig. 2. Representative FESEM images of (A) PSU, (B) PSU-PDMS-0.5, (C) PSU-PDMS-1, (D) PSU-PDMS-2 ENMs, and (E) PSU-PDMS-4 ENMs before cold-press treatment. (F) The
corresponding average fiber diameter of PSU and PSU-PDMS-x ENMs.

pressure, was not considered for further analysis. For the same reason, size of the inter-fiber space of the PSU-y and PSU-PDMS-x-y ENMs can
only the PSU-PDMS-2-y ENMs subjected to the two smaller pressures (2 be determined. The results are summarized in Table 1. It can be clearly
and 4 MPa) were considered for further characterization and DCMD concluded that the mean size and size distribution of the inter-fiber
desalination tests. It is noteworthy that the water contact angles of PSU- space of all ENMs exhibited a decreasing trend as the cold-press post-
PDMS-x ENMs before and after cold-press post-treatment did not treatment pressure was increased. Moreover, it is worth noting that the
change significantly due to the physical flexibility of PDMS (compare maximum size of the inter-fiber space was reduced more significantly
Table 1 and Fig. 3). Although the PDMS fusion would be formed under than the minimum size when the cold-press post-treatment pressure
higher pressure on the membrane surface, the water contact angles of was increased. This was attributed to the formation of the tightly
the PSU-PDMS-0.5-y ENMs were still larger than 140°, and those of the packed ENMs structure with a reduced thickness as stated previously.
PSU-PDMS-1-y and PSU-PDMS-2-y ENMs were still larger than 150° The mean inter-fiber size of the PSU-PDMS-x membranes prepared
exhibiting self-cleaning property because of the existence of the low under the same cold-press post-treatment pressure was decreased with
surface energy of PDMS fusion and the intrinsic roughness from the the increase of PDMS concentration. This indicates the effect of the
randomly stacked fiber morphology (as shown in Fig. 6B-E). formed PDMS agglomerations and the fusion of the PDMS coated na-
The cumulative filter flow curves of the ENMs prepared at different nofibers to each other on the size of inter-fiber space as it was discussed
PDMS concentrations and cold-press post-treatment pressures are pre- previously. The mean inter-fiber size of the PSU ENMs were smaller
sented in Fig. 7. From these curves, the mean, minimum and maximum than all the PSU-PDMS-x ENMs due to the PDMS adhesion effect, which

312
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

cross-section morphologies of the PSU-PDMS-x-4 ENMs, it is clearly

confirmed that the PDMS modification of PSU nanofibers was not only
achieved on the nanofibrous surface layer, but the inner nanofibers
were also wrapped by the PDMS coating (Fig. 8B4–D4). Furthermore,
the PDMS fusion occurred on the partial surface layers, rather than in
the internal nanofibrous layers, as shown in Fig. 8B2B3–D2D3, which
fully certified the fact that the PSU-PDMS-x-y ENMs have relatively
larger bulk void volume fractions and mean inter-fiber sizes than the
PSU-y ENMs due to the physical flexibility of PDMS under the cold-
press post-treatment.
Gas permeability of dry ENMs was determined using the capillary
flow porometer [46]. The results are shown in Table 1. It was observed
that the gas permeability of all ENMs decreased with the increase of the
cold-press post-treatment pressure. This is due to the decrease of the
void volume fraction and size of the inter-fiber space. It is interesting to
note that the decrease of the gas permeability with the applied cold-
press post-treatment pressure was drastically higher for the PSU ENMs
than for the PSU-PDMS-x ENMS (see Table 1). For example, a decrease
of 7455 L m−2 s−1 bar−1 was observed for the PSU ENMs, whereas a
decrease of 1753 L m−2 s−1 bar−1 was obtained for the PSU-PDMS-0.5
ENMs, when the cold-press post-treatment pressure was increased from
2 to 12 MPa (Table 1). The increase of the PDMS content reduced sig-
nificantly the gas permeability values (i.e. 8646, 4934 and
724 L m−2 s−1 bar−1 for the PSU-PDMS-0.5, PSU-PDMS-1 and PSU-
PDMS-2, respectively, when the cold-press post-treatment pressure was
4 MPa). This is due to the reduction of both the void volume fraction
and the size of the inter-fiber space with the PDMS content. It is worth
Fig. 3. (A) Water contact angles variation for the ENMs prepared with different PDMS
quoting that gas permeability of the prepared ENMs in this study were
concentrations before cold-press treatment; (B) images of the dynamic water-adhesion extremely higher than those of the commercial PVDF MD membranes
behavior on the PDMS-PSU-1 surface; (C and D) the self-cleaning property of PSU (C) and (253 L m−2 s−1 bar−1) [20].
PDMS-PSU-1 ENMs (D). For a better visibility, water was dyed by methyl blue and hy-
drophilic graphene oxide powder was used as a surface contaminant. 3.3. DCMD experiments

In this study, the effects of the PDMS concentration and cold-press

post-treatment pressure on the DCMD desalination performance was
investigated. Prior to cold-press post-treatment pressure, the feasibility
of DCMD operation of the PSU-PDMS-2 ENMs without cold-press post-
treatment was evaluated. The results showed that the electrical con-
ductivity of the permeate changed from 4.224 μS/cm to 50 μS/cm
within 2 h of DCMD process at a temperature difference of 50 °C due to
its insufficient dimensional integrity and low LEPw (lower than 0.15 bar
in all cases). By using a cold-press post-treatment, the LEPw values of
PSU-PDMS ENMs were improved significantly (see Fig. 5(a)). Fig. 9
shows the obtained permeate flux and salt rejection factor of the PSU-
PDMS-x-y ENMs as a function of the applied temperature difference (30
− 60 °C). As it was expected, a higher temperature difference induced a
higher permeate flux (Fig. 9(a.1, b.1 and c.1)), due to the enhancement
of the transmembrane vapor pressure difference [55]. Additinally, for
the same temperature difference the permeate flux of the PSU-PDMS-x
ENMs (x > 0.5) was gradually decreased with the increase of the cold-
Fig. 4. FTIR spectra of PDMS and PSU, PSU-PDMS-0.5, PSU-PDMS-1 and PSU-PDMS-2 press post-treatment pressure for PDMS. This is attributed to the re-
ENMs before cold-press treatment. duction of the bulk void volume fraction, the size of the inter-fiber
space and the potentially increased formation of the PDMS coated na-
was in agreement with the variation trend of the bulk void volume nofibers fusions under higher cold-press post-treatment pressures. For
fraction and the lower compactness of the PSU-PDMS-x ENMs due to the PSU-PDMS-0.5-y no clear trend was detected between the permeate
the physical flexibility of PDMS (see membrane thickness values in flux and the applied cold-press post-treatment pressure. On the other
Table 1) [52,53]. hand, under the same temperature difference and the cold-press post-
The representative cross-section FESEM images of PSU-4 and PSU- treatment pressure, the permeate flux was decreased with the increase
PDMS-x-4 ENMs shown in Fig. 8 confirmed further the observed of the PDMS concentration. For instance, the permeate flux of the PSU-
structural membrane characteristics. As can be seen in Fig. 8A1–A4, the PDMS-0.5-4, PSU-PDMS-1-4 and PSU-PDMS-2-4 ENMs at the tempera-
compact PSU-4 ENM exhibited a randomly oriented 3D nonwoven na- ture difference of 50 °C was 22.3, 21.5 and 14.7 kg/m2 h, respectively.
nofibrous structure, and the inner nanofibers presented a circular cross- This is in agreement with the gas permeability trend and it is attributed
section with micro-wrinkled skin and sponge-like structure core, which to the reduction of both the void volume fraction and the size of the
was attributed to the relatively slow diffusion velocity of DMF solvent inter-fiber space with the PDMS content.
entrapped inside the electrospinning fluid jet [54]. Regarding to the For sake of comparison, the PSU-8 ENM was used to evaluate the
DCMD performance and the results are also shown in Fig. 9(a.1). The

313
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

Table 1
Structural characteristics (thickness, mean size and size distribution of the inter-fiber space, gas permeability and contact angle of water, θw) of the prepared PSU-y and PSU-PDMS-x-y
ENMs.

Membranes Thickness (μm) Mean size of inter-fiber space (μm) Size distribution of inter-fiber space (μm) Gas permeability (L m−2 s−1 bar−1) θw (°)

PSU-2 101.9 ± 5.9 2.045 0.931–4.412 8228 ± 16 113.2 ± 1.1

PSU-4 93.2 ± 7.6 1.623 1.118–3.605 3961 ± 23 111.8 ± 0.8
PSU-8 74.8 ± 10.4 1.023 0.770–1.896 2391 ± 15 110.6 ± 1.2
PSU-12 56.1 ± 5.0 0.730 0.491–1.051 773 ± 6 109.2 ± 2.1
PSU-PDMS-0.5-2 109.9 ± 8.5 3.691 2.253–11.630 8908 ± 38 146.3 ± 1.2
PSU-PDMS-0.5-4 95.9 ± 5.1 3.362 2.094–7.805 8646 ± 15 145.1 ± 0.9
PSU-PDMS-0.5-8 82.5 ± 6.0 2.971 1.867–7.110 8038 ± 14 143.2 ± 1.8
PSU-PDMS-0.5-12 70.7 ± 2.4 2.453 1.778–6.562 6353 ± 6 141.7 ± 2.1
PSU-PDMS-1-2 111.4 ± 6.1 3.470 2.070–7.523 6347 ± 4 151.5 ± 2.7
PSU-PDMS-1-4 97.1 ± 5.5 2.819 1.583–5.686 4934 ± 13 151.1 ± 1.1
PSU-PDMS-1-8 83.9 ± 7.8 2.699 1.577–4.203 4594 ± 14 150.9 ± 1.5
PSU-PDMS-1-12 76.6 ± 5.2 2.299 1.530–3.412 3699 ± 4 150.2 ± 0.7
PSU-PDMS-2-2 124.2 ± 7.3 3.139 1.100–5.817 909 ± 4 153.9 ± 2.1
PSU-PDMS-2-4 107.6 ± 6.4 2.786 0.852–4.828 724 ± 2 151.7 ± 1.3

(a) lower salt rejection factors than the PSU-PDMS-x-y ENMs and it was
PSU ENMs
1.0 PSU-PDMS-0.5 ENMs dramatically decreased with the increase of the temperature difference,
PSU-PDMS-1 ENMs indicating that the PSU-8 ENM suffered wetting during the DCMD
PSU-PDMS-2 ENMs process. This result confirmed the insufficient hydrophobicity of this
0.8
ENM and its lower LEPw value (Fig. 5a). The salt rejection factor of PSU-
LEPw (bar)

PDMS-2-2 and PSU-PDMS-2-4 did not change significantly (i.e. it was

0.6 maintained the same up to 40 °C and increased by only 0.01% at 60 °C),
while the salt rejection factor of PSU-PDMS-1-4 is 0.5% higher than that
0.4 of PSU-PDMS-1-2 and the salt rejection of PSU-PDMS-0.5-4 is about 2%
higher than that of PSU-PDMS-0.5–2 ENMs at 60 °C. It must be pointed
0.2 out that with the increase of PDMS concentration, the salt rejection
factors of PSU-PDMS-x-y ENMs increased due to the increase of surface
hydrophobicity under the same cold-press post-treatment pressure (see
0.0
0 2 4 6 8 10 12 14 θw values in Table 1). The salt rejection factors of both PSU-PDMS-1-4
Pressure (MPa) and PSU-PDMS-2-4 ENMs were higher than 99.99%, and the salt re-
jection factor of PSU-PDMS-2-2 ENMs was higher than that of PSU-
(b) PDMS-1-2 ENMs. This means that at low cold-press post-treatment
100
PSU ENMs pressure, the PDMS concentration should be enough high in order to
Void Volume Fraction (%)

90 PSU-PDMS-0.5 ENMs obtain a homogeneous coating of the PSU fibers, resulting in an ex-
PSU-PDMS-1 ENMs cellent rejection factor (higher than 99.9%). The PDMS content of 0.5 g
PSU-PDMS-2 ENMs
in 40 mL of hexane provides an uneven modification of PSU nanofibers.
80
Consequently, the salt rejection factor increased significantly with the
increase of the cold-press post-treatment pressure from 2 to 4 MPa, but
70 no enhancement was observed when the pressure was increased from 4
to 12 MPa.
60 In general, a higher cold-press post-treatment pressure and/or a
higher PDMS concentration enhanced the salt rejection factor except for
50 the PSU-PDMS-0.5-y ENMs that presented an irregular variation trend
with the increase of the cold-press post-treatment pressure due to the
40 above mentioned not homogeneous PDMS coating. Additionally, the
0 2 4 6 8 10 12 14 PSU-PDMS-1-2 ENMs still showed salt rejection factors lower than
Pressure (MPa) 99.99% because the cold-press post-treatment pressure of 2 MPa was
not sufficiently high to increase its LEPw value more than 0.5 bar. It is
Fig. 5. (a) LEPw and (b) void volume fraction of PSU, PSU-PDMS-0.5 and PSU-PDMS-1
important to point out that the DCMD experiments were carried out for
ENMs as a function of the applied cold-press post-treatment pressure.
the same ENM from the lowest to the highest temperature difference
(i.e. from ΔT = 30 °C to ΔT = 60 °C). Therefore, the observed decrease
permeate flux of the PSU-8 ENM was greater than that of all PSU-PDMS- of the salt rejection factor in Fig. 9(a.2, b.2 and c.2) is also due to the
x-8 ENMs although the PSU-8 ENM exhibited a relatively lower bulk membrane wetting with the DCMD operating time. The cold-press post-
void volume fraction, inter-fiber space size and gas permeability (see treatment applied under 4 MPa or higher results in excellent salt re-
Fig. 5 and Table 1). This is due to the PDMS coating and nanofibrous jection factors (i.e. higher than 99.99%) for the PSU-PDMS-x ENMs due
fusions as it was discussed based on the FESEM analysis and the im- to their superhydrophobicity, self-cleaning property and the good LEPw
provement of the hydrophobic character of the PSU-PDMS-x-y ENMs. In values. However, their corresponding permeate fluxes were lower
addition, as the mass transfer is inversely proportional to the membrane compared to the other ENMs.
thickness, the PSU-8 ENMs, which has a smaller membrane thickness, Based on the above results, continuous DCMD desalination of the
presented lower mass transfer resistance and a higher permeate flux. PSU-4 and PSU-PDMS-x-4 ENMs were performed over a period of 12 h
The salt rejection factors are plotted in Fig. 9(a.2, b.2 and c.2) a operation using a 30 g/L NaCl aqueous solution as feed and a tem-
function of the applied temperature difference. The PSU-8 ENM showed perature difference of 50 °C. As shown in Fig. 10, both the PSU-4 and

314
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

Fig. 6. Representative surface FESEM images of (A) PSU-4, (B) PSU-PDMS-1-2, (C) PSU-PDMS-1-4, (D) PSU-PDMS-1-8 and (E) PSU-PDMS-1-12 ENMs after cold-press post-treatment.

PSU-PDMS-0.5-4 ENMs presented competitive and stable permeate flux potentially increased formation of the PDMS coated nanofiber fusions
of about 23.4 ± 0.4 and 21.6 ± 0.3 kg/m2 h, respectively. However, a compared to the PSU-PDMS-1-4 ENM. The PSU-PDMS-2-4 ENM also
gradual increase of the permeate concentration was observed from the exhibited a stable permeate concentration due to its excellent hydro-
beginning of the DCMD process that could be due to the low LEPw and phobicity.
hydrophobicity of these ENMs. The permeate concentration registered Compared to the desalination performance of some membranes used
for these two ENMs was higher than 0.15 g/L NaCl within 6 h of DCMD in MD [9–11,20,29,56–58], the above obtained stable DCMD perfor-
operation (see Fig. 10), indicating wetting of their inter-fiber space. mance (i.e. DCMD permeate flux and permeate concentration) sug-
With the increase of the PDMS content in 40 mL hexane, the PSU- gested that the PSU-PDMS-x-y (x ≥ 1, y ≥ 4) ENMs are promising for
PDMS-1-4 ENMs maintained a stable permeate flux of about MD desalination. For example, a permeate flux of 7.2 kg/m2h (3.5 wt%
21.5 ± 0.2 kg/m2 h and low permeate concentration (i.e. electrical NaCl feed, ΔT = 40 °C) [20] was obtained for some commercial PVDF
conductivity variation 4.224–4.523 µS/cm) over a period of 12 h DCMD flat-sheet MD membranes, 12.4 kg/m2h (1 M NaCl feed, ΔT = 50 °C)
operation. This was attributed to its superhydrophobic character, self- [56] was reported for PVDF hollow fiber membranes, 11–12 kg/m2h
cleaning property and its rather higher LEPw compared to the other (3.5 wt% NaCl feed, ΔT = 60 °C) [57] was achieved for PVDF ENMs
PSU-PDMS-x-4 ENMs as it was mentioned previously (Fig. 5(a)). The and 5.75 kg/m2h (3.5 wt% NaCl feed, ΔT = 63 °C) [10] for modified
obtained DCMD flux of the PSU-PDMS-2-4 ENM was lower than that of PVDF ENMs. When using other polymers, Lalia et al. [11] considered
the other ENMs (i.e. 15.6 ± 0.3 kg/m2 h) due to its lower bulk void hot-pressed polyvinylidenefluoride-co-hexafluoropropylene (PVDF-
volume fraction, smaller inter-fiber space size, larger thickness and the HFP) ENMs and claimed permeate fluxes of 20–22 l/m2h (1 wt% NaCl

315
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

100 100

Cumulative Filter Flow (%)

Cumulative Filter Flow (%)

(a) PSU-PDMS-0.5-2
PSU-PDMS-0.5-4
80 80 PSU-PDMS-0.5-8
PSU-PDMS-0.5-12
PSU-2
PSU-4
60 PSU-8 60
PSU-12

40 40

20 20
(b)
0 0
500 1000 1500 2000 2500 3000 3500 4000 4500 5000 1000 2000 3000 4000 5000 6000
Inter-fiber space size (nm) Inter-fiber space size (nm)
100 100
(d)
Cumulative Filter Flow (%)

(c)

Cumulative Filter Flow (%)

80 80 PSU-PDMS-2-2
PSU-PDMS-1-2
PSU-PDMS-2-4
PSU-PDMS-1-4
60 PSU-PDMS-1-8 60
PSU-PDMS-1-12

40 40

20 20

0 0
1000 2000 3000 4000 5000 6000 7000 1000 2000 3000 4000 5000 6000
Inter-fiber space size (nm) Inter-fiber space size (nm)
Fig. 7. Cumulative filter flow of the ENMs prepared at different PDMS concentrations and cold-pressures: (a) PSU-y, (b) PSU-PDMS-0.5-y, (c) PSU-PDMS-1-y and (d) PSU-PDMS-2-y
ENMs.

Fig. 8. FESEM cross-section images of (A1-A4) PSU-4, (B1-B4) PSU-PDMS-0.5-4, (C1-C4) PSU-PDMS-1-4 and (D1-D4) PSU-PDMS-2-4 ENMs at different magnifications.

316
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

(a.1) (a.2)
35
PSU-8

Permeate Flux (kg/m h)

2 30 PSU-PDMS-0.5-2
PSU-PDMS-0.5-4
25 PSU-PDMS-0.5-8
PSU-PDMS-0.5-12
20

15

10

0
30 40 50 60
Δ T (ºC)

(b.1) (b.2)
35
PSU-PDMS-1-2
Permeate Flux (kg/m h)

30 PSU-PDMS-1-4
2

PSU-PDMS-1-8
25 PSU-PDMS-1-12

20

15

10

0
30 40 50 60
Δ T (ºC)

(c.1) (c.2)

Fig. 9. Permeate flux (a.1, b.1 and c.1) and salt rejection factor (a.2, b.2 and c.2) as a function of temperature difference for the prepared ENMs: (a) PSU-8 and PSU-PDMS-0.5-y, (b) PSU-
PDMS-1-y and (c) PSU-PDMS-2-y ENMs (30.0 g/L NaCl aqueous solution as feed, permeate temperature = 20 °C, feed flow rate = 0.6 L/min and permeate flow rate = 0.4 L/min).

feed, ΔT = 60 °C), Dong et al. [29] studied fluoroalkylsilane grafted entrapped under the feed-ENM interface will be lost resulting in the
PVA ENMs exhibiting permeate fluxes of 25.2 kg/m2h (3.5 wt% NaCl wetting of the surface nanofibrous layer. Subsequently, the water droplets
feed, ΔT = 40 °C), Leitch et al. [9] used novel high-porosity membranes will continuously penetrate into the inter-fiber space of the poorly hy-
based on bacterial nanocellulose aerogel and reported a permeate flux drophobic PSU nanofibrous structure, until complete wetting the ENM. In
of 22.92 kg/m2h (35 g/L NaCl feed, ΔT = 40 °C) and Hwang et al. [58] contrast, as shown in Fig. 11(B), due to the entire 3D superhydrophobic
for commercial PTFE membrane in DCMD process obtained a permeate nanofibrous structure and self-cleaning surface, the hot salt water droplets
flux of 22.89 kg/m2h (3.5% NaCl feed, ΔT = 40 °C). will be immediately rolled and slid off the ENM surface during the DCMD
Fig. 11 schematically illustrated the wetting tendency of the PSU and operation. Even after the wetting of the top surface nanofibrous layer
PSU-PDMS-1-4 ENMs. In the case of the PSU ENM (Fig. 11A), the hot salt another stable feed-ENM interface with air and/or vapor entrapped un-
water droplets will be immediately adhered to PSU nanofibers once the derneath it will be promptly created making the PSU-PDMS-1-4 ENM
DCMD process is carried out, and the air and/or vapor that normally is excellent long-term DCMD applications.

317
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

Permeate Flux (kg/m h) 24 void volume fraction and the size of the inter-fiber space were reduced,
2

22 but still larger than those of the PSU-y ENMs due to the physical flex-
ibility of the PDMS under pressure. To further enhance the LEPw of the
20
PSU-PDMS-x-y ENMs, in next research work we propose the use of
18 different solvents to enhance the compactness of PSU ENMs and in-

Permeate Concentration (g/L)

16 corporate hydrophobic inorganic nanoparticles into the PDMS layer to
14 increase the surface roughness and hydrophobicity and reduce the in-
0.15
terfiber space of these ENMs.
0.12 During the DCMD tests, the PSU-y ENMs suffered wetting due to its
PSU-4
PSU-PDMS-0.5-4 0.09 insufficient hydrophobicity and low LEPw values. Therefore, the cold-
PSU-PDMS-1-4
0.06 press post-treatment for PSU-PDMS ENMs is a crucial process to prepare
PSU-PDMS-2-4
high and good performance membranes. The PSU-PDMS-x-y ENMs with
0.03
x ≥ 1 and y ≥ 4 were found to be promising in MD because of their
0.00
stable DCMD performance (i.e. very low electrical conductivity of the
0 2 4 6 8 10 12 permeation during 12 h DCMD operation). The higher PDMS con-
Operation Time (h) centration or the larger cold-press post-treatment pressure resulted in
Fig. 10. Permeate flux and permeate concentration of PSU-4, PSU-PDMS-0.5-4, PSU- lower permeate fluxes. The PSU-PDMS-1-4 ENM was able to maintain a
PDMS-1-4 and PSU-PDMS-2-4 ENMs as a function of the MD operation time (30 g/L NaCl stable permeate flux of about 21.5 kg/m2h and a high quality water
aqueous solution as feed, ΔT = 50 °C, feed flow rate = 0.6 L/min and permeate flow rate permeate (i.e. electrical conductivity variation 4.224 – 4.523 µS/cm
= 0.4 L/min). over a period of 12 h DCMD operation) for a feed NaCl solution of 30 g/
L and a temperature difference of 50 °C without detecting wetting of the
A inter-fiber space. This demonstrated the feasibility of designing and
wetting tendency

preparing self-cleaning PSU-PDMS ENMs for MD applications.

Start of DCMD process

Hot salt water

Acknowledgements
Time

Fast wetting of the The authors are thankful to the financial support of the Spanish
nanofibrous layers Ministry of Economy and Competitiveness (MEC) (CTM2015-65348-C2-
2-R). X. Li is thankful to the support of the Program for New Century
Excellent Talents in University (NCET-13–0725), Chinese Universities
Scientific Fund and a fellowship from the China Scholarship Council.
PSU ENMs
References
water vapor permeate

B
Start of DCMD process

[1] M. Khayet, T. Matsuura, Membrane Distillation: Principles and Applications,

Hot salt water

Hot salt water

Elsevier, 2011.
Time [2] E. Drioli, A. Ali, F. Macedonio, Membrane distillation: recent developments and
perspectives, Desalination 356 (2015) 56–84.
Prevent wetting due to the [3] A. Alkhudhiri, N. Darwish, N. Hilal, Membrane distillation: a comprehensive re-
self-cleaning surface view, Desalination 287 (2012) 2–18.
[4] M. El-Bourawi, Z. Ding, R. Ma, M. Khayet, A framework for better understanding
membrane distillation separation process, J. Membr. Sci. 285 (2006) 4–29.
[5] P. Wang, T.-S. Chung, Recent advances in membrane distillation processes: mem-
brane development, configuration design and application exploring, J. Membr. Sci.
474 (2015) 39–56.
PSU-PDMS-1-4 ENMs [6] Y.D. Kim, K. Thu, N. Ghaffour, K.C. Ng, Performance investigation of a solar-as-
sisted direct contact membrane distillation system, J. Membr. Sci. 427 (2013)
Fig. 11. Schematic wetting behavior of (A) PSU ENMs and (B) PSU-PDMS-1-4 ENMs in 345–364.
[7] Y. Zhang, Y. Peng, S. Ji, Z. Li, P. Chen, Review of thermal efficiency and heat re-
the continuous DCMD process.
cycling in membrane distillation processes, Desalination 367 (2015) 223–239.
[8] A. Ashraf, H. Salih, S. Nam, S.A. Dastgheib, Robust carbon nanotube membranes
directly grown on Hastelloy substrates and their potential application for membrane
4. Conclusions
distillation, Carbon 106 (2016) 243–251.
[9] M.E. Leitch, C. Li, O. Ikkala, M.S. Mauter, G.V. Lowry, Bacterial nanocellulose
Novel self-cleaning PSU-PDMS-x-y ENMs with competitive DCMD aerogel membranes: novel high-porosity materials for membrane distillation,
performance have been prepared by electrospinning PSU nanofibers Environ. Sci. Technol. Lett. 3 (2016) 85–91.
[10] J.A. Prince, G. Singh, D. Rana, T. Matsuura, V. Anbharasi, T.S. Shanmugasundaram,
followed by PDMS modification and cold-press post-treatment. The Preparation and characterization of highly hydrophobic poly(vinylidene fluoride) –
results showed that PDMS adhesion and/or agglomeration was formed Clay nanocomposite nanofiber membranes (PVDF–clay NNMs) for desalination
on the ENM surface with the increase of PDMS concentration, inducing using direct contact membrane distillation, J. Membr. Sci. 397–398 (2012) 80–86.
[11] B.S. Lalia, E. Guillen-Burrieza, H.A. Arafat, R. Hashaikeh, Fabrication and char-
self-cleaning property and superhydrophobicity due to the low surface acterization of polyvinylidenefluoride-co-hexafluoropropylene (PVDF-HFP) elec-
energy of PDMS and the surface hierarchical roughness. trospun membranes for direct contact membrane distillation, J. Membr. Sci. 428
The cold-press pre-treatment pressure had a decisive influence on (2013) 104–115.
[12] Y. Liao, R. Wang, M. Tian, C.Q. Qiu, A.G. Fane, Fabrication of polyvinylidene
the membrane surface morphology and structural characteristics. The fluoride (PVDF) nanofiber membranes by electro-spinning for direct contact
PDMS fusions was gradually formed on the surface of the nanofibrous membrane distillation, J. Membr. Sci. 425 (2013) 30–39.
layer depending on the applied cold-press post-treatment pressure, [13] J. Lee, C. Boo, W.H. Ryu, A.D. Taylor, M. Elimelech, Development of omniphobic
desalination membranes using a charged electrospun nanofiber scaffold, ACS Appl.
while the cross-section FE-SEM images showed that the inner nanofi- Mater. Interfaces 8 (2016) 11154–11161.
bers were also wrapped by the PDMS coating, but the PDMS fusions was [14] L.D. Tijing, J.-S. Choi, S. Lee, S.-H. Kim, H.K. Shon, Recent progress of membrane
not formed in the internal nanofibrous layers. distillation using electrospun nanofibrous membrane, J. Membr. Sci. 453 (2014)
435–462.
With the increase of the cold-press post-treatment pressure, the
[15] M. Essalhi, M. Khayet, Self-sustained webs of polyvinylidene fluoride electrospun
LEPw of the PSU-PDMS-x-y ENMs was improved significantly, while the nanofibers at different electrospinning times: 1. desalination by direct contact

318
X. Li et al. Journal of Membrane Science 542 (2017) 308–319

membrane distillation, J. Membr. Sci. 433 (2013) 167–179. [37] A. Venault, Y. Chang, D.-M. Wang, D. Bouyer, A. Higuchi, J.-Y. Lai, PEGylation of
[16] M. Essalhi, M. Khayet, Self-sustained webs of polyvinylidene fluoride electrospun anti-biofouling polysulfone membranes via liquid-and vapor-induced phase se-
nano-fibers: effects of polymer concentration and desalination by direct contact paration processing, J. Membr. Sci. 403 (2012) 47–57.
membrane distillation, J. Membr. Sci. 454 (2014) 133–143. [38] H. Yu, Y. Cao, G. Kang, J. Liu, M. Li, Q. Yuan, Enhancing antifouling property of
[17] Y. Liao, R. Wang, A.G. Fane, Engineering superhydrophobic surface on poly(viny- polysulfone ultrafiltration membrane by grafting zwitterionic copolymer via UV-
lidene fluoride) nanofiber membranes for direct contact membrane distillation, J. initiated polymerization, J. Membr. Sci. 342 (2009) 6–13.
Membr. Sci. 440 (2013) 77–87. [39] P.S. Singh, S. Joshi, J. Trivedi, C. Devmurari, A.P. Rao, P. Ghosh, Probing the
[18] Y. Liao, R. Wang, A.G. Fane, Fabrication of bioinspired composite nanofiber structural variations of thin film composite RO membranes obtained by coating
membranes with robust superhydrophobicity for direct contact membrane dis- polyamide over polysulfone membranes of different pore dimensions, J. Membr.
tillation, Environ. Sci. Technol. 48 (2014) 6335–6341. Sci. 278 (2006) 19–25.
[19] Y. Liao, C.H. Loh, R. Wang, A.G. Fane, Electrospun superhydrophobic membranes [40] P. Arribas, M. Khayet, M. García-Payo, L. Gil, Self-sustained electro-spun poly-
with unique structures for membrane distillation, ACS Appl. Mater. Interfaces 6 sulfone nano-fibrous membranes and their surface modification by interfacial
(2014) 16035–16048. polymerization for micro-and ultra-filtration, Sep. Purif. Technol. 138 (2014)
[20] X. Li, X. Yu, C. Cheng, L. Deng, M. Wang, X. Wang, Electrospun superhydrophobic 118–129.
organic/inorganic composite nanofibrous membranes for membrane distillation, [41] M. Tian, Y. Yin, C. Yang, B. Zhao, J. Song, J. Liu, X.-M. Li, T. He, CF4 plasma
ACS Appl. Mater. Interfaces 7 (2015) 21919–21930. modified highly interconnective porous polysulfone membranes for direct contact
[21] Y.C. Woo, L.D. Tijing, W.G. Shim, J.S. Choi, S.H. Kim, T. He, E. Drioli, H.K. Shon, membrane distillation (DCMD), Desalination 369 (2015) 105–114.
Water desalination using graphene-enhanced electrospun nanofiber membrane via [42] B. Cortese, S. D'Amone, M. Manca, I. Viola, R. Cingolani, G. Gigli,
air gap membrane distillation, J. Membr. Sci. 520 (2016) 99–110. Superhydrophobicity due to the hierarchical scale roughness of PDMS surfaces,
[22] L.D. Tijing, Y.C. Woo, W.G. Shim, T. He, J.S. Choi, S.H. Kim, H.K. Shon, Langmuir 24 (2008) 2712–2718.
Superhydrophobic nanofiber membrane containing carbon nanotubes for high- [43] M. Jin, X. Feng, J. Xi, J. Zhai, K. Cho, L. Feng, L. Jiang, Super‐hydrophobic PDMS
performance direct contact membrane distillation, J. Membr. Sci. 502 (2016) surface with ultra‐low adhesive force, Macromol. Rapid Commun. 26 (2005)
158–170. 1805–1809.
[23] E.J. Lee, A.K. An, T. He, Y.C. Woo, H.K. Shon, Electrospun nanofiber membranes [44] C.H. Xue, Z.D. Zhang, J. Zhang, S.T. Jia, Lasting and self-healing superhydrophobic
incorporating fluorosilane-coated TiO2 nanocomposite for direct contact membrane surfaces by coating of polystyrene/SiO2 nanoparticles and polydimethylsiloxane, J.
distillation, J. Membr. Sci. 520 (2016) 145–154. Mater. Chem. A 2 (2014) 15001–15007.
[24] X. Li, C. Wang, Y. Yang, X. Wang, M. Zhu, B.S. Hsiao, Dual-biomimetic super- [45] M. Cao, D. Guo, C. Yu, K. Li, M. Liu, L. Jiang, Water-repellent properties of su-
hydrophobic electrospun polystyrene nanofibrous membranes for membrane dis- perhydrophobic and lubricant-infused “slippery” surfaces: a brief study on the
tillation, ACS Appl. Mater. Interfaces 6 (2014) 2423–2430. functions and applications, ACS Appl. Mater. Interfaces 8 (2015) 3615–3623.
[25] H. Ke, E. Feldman, P. Guzman, J. Cole, Q. Wei, B. Chu, A. Alkhudhiri, R. Alrasheed, [46] A. Jena, K. Gupta, Characterization of pore structure of filtration media, Fluid/Part.
B.S. Hsiao, Electrospun polystyrene nanofibrous membranes for direct contact Sep. J. 14 (2002) 227–241.
membrane distillation, J. Membr. Sci. 515 (2016) 86–97. [47] L. Wang, C.L. Pai, M.C. Boyce, G.C. Rutledge, Wrinkled surface topographies of
[26] X. Li, L. Deng, X. Yu, M. Wang, X. Wang, C. García-Payo, M. Khayet, A novel electrospun polymer fibers, Appl. Phys. Lett. 94 (2009) (151916-151916-151913).
profiled core–shell nanofibrous membrane for wastewater treatment by direct [48] X. Feng, L. Jiang, Design and creation of superwetting/antiwetting surfaces, Adv.
contact membrane distillation, J. Mater. Chem. A 4 (2016) 14453–14463. Mater. 18 (2006) 3063–3078.
[27] T. Zhou, Y. Yao, R. Xiang, Y. Wu, Formation and characterization of polytetra- [49] M. Liu, Y. Zheng, J. Zhai, L. Jiang, Bioinspired super-antiwetting interfaces with
fluoroethylene nanofiber membranes for vacuum membrane distillation, J. Membr. special liquid-solid adhesion, Acc. Chem. Res. 43 (2010) 368–377.
Sci. 453 (2014) 402–408. [50] I.-J. Chen, E. Lindner, The stability of radio-frequency plasma-treated poly-
[28] F. Guo, A. Servi, A. Liu, K.K. Gleason, G.C. Rutledge, Desalination by membrane dimethylsiloxane surfaces, Langmuir 23 (2007) 3118–3122.
distillation using electrospun polyamide fiber membranes with surface fluorination [51] K. Smolders, A.C.M. Franken, Terminology for membrane distillation, Desalination
by chemical vapor deposition, ACS Appl. Mater. Interfaces 7 (2015) 8225–8232. 72 (1989) 249–262.
[29] Z.Q. Dong, B.J. Wang, X.H. Ma, Y.M. Wei, Z.L. Xu, FAS grafted electrospun poly [52] D. Fuard, T. Tzvetkova-Chevolleau, S. Decossas, P. Tracqui, P. Schiavone,
(vinyl alcohol) nanofiber membranes with robust superhydrophobicity for mem- Optimization of poly-di-methyl-siloxane (PDMS) substrates for studying cellular
brane distillation, ACS Appl. Mater. Interfaces 7 (2015) 22652–22659. adhesion and motility, Microelectron. Eng. 85 (2008) 1289–1293.
[30] H. Maab, L. Francis, A. Al-Saadi, C. Aubry, N. Ghaffour, G. Amy, S.P. Nunes, [53] L. Wang, Q. Gong, S. Zhan, L. Jiang, Y. Zheng, Robust anti‐icing performance of a
Synthesis and fabrication of nanostructured hydrophobic polyazole membranes for flexible superhydrophobic surface, Adv. Mater. 28 (2016) 7729–7735.
low-energy water recovery, J. Membr. Sci. 423–424 (2012) 11–19. [54] B. Ding, J. Lin, X. Wang, J. Yu, J. Yang, Y. Cai, Investigation of silica nanoparticle
[31] M. Nosonovsky, B. Bhushan, Biomimetic superhydrophobic surfaces: multiscale distribution in nanoporous polystyrene fibers, Soft Matter 7 (2011) 8376–8383.
approach, Nano Lett. 7 (2007) 2633–2637. [55] M. Khayet, Membranes and theoretical modeling of membrane distillation: a re-
[32] Y. Lee, S.H. Park, K.B. Kim, J.K. Lee, Fabrication of hierarchical structures on a view, Adv. Colloid Interface Sci. 164 (2011) 56–88.
polymer surface to mimic natural superhydrophobic surfaces, Adv. Mater. 19 [56] N. Mokhtar, W. Lau, A. Ismail, W. Youravong, W. Khongnakorn, K. Lertwittayanon,
(2007) 2330–2335. Performance evaluation of novel PVDF–Cloisite 15A hollow fiber composite mem-
[33] L. Feng, Y. Zhang, J. Xi, Y. Zhu, N. Wang, F. Xia, L. Jiang, Petal effect: a super- branes for treatment of effluents containing dyes and salts using membrane dis-
hydrophobic state with high adhesive force, Langmuir 24 (2008) 4114–4119. tillation, RSC Adv. 5 (2015) 38011–38020.
[34] S. Wang, L. Jiang, Definition of superhydrophobic states, Adv. Mater. 19 (2007) [57] C. Feng, K.C. Khulbe, I. Matsuura, R. Gopal, S. Kaur, S. Rarnakrishna, A. Khayet,
3423–3424. Production of drinking water from saline water by air-gap membrane distillation
[35] B. Chakrabarty, A. Ghoshal, M. Purkait, Ultrafiltration of stable oil-in-water emul- using polyvinylidene fluoride nanofiber membrane, J. Membr. Sci. 311 (2008) 1–6.
sion by polysulfone membrane, J. Membr. Sci. 325 (2008) 427–437. [58] H.J. Hwang, K. He, S. Gray, J. Zhang, I.S. Moon, Direct contact membrane dis-
[36] H.A. Tsai, D.H. Huang, R.C. Ruaan, J.Y. Lai, Mechanical properties of asymmetric tillation (DCMD): experimental study on the commercial PTFE membrane and
polysulfone membranes containing surfactant as additives, Ind. Eng. Chem. Res. 40 modeling, J. Membr. Sci. 371 (2011) 90–98.
(2001) 5917–5922.

319