Supercritical Gel Drying: A Powerful Tool for

ARTICLE
Tailoring Symmetric Porous PVDF-HFP
Membranes
S. Cardea,*,† A. Gugliuzza,‡ M. Sessa,† M. C. Aceto,‡,§ E. Drioli,‡,§ and E. Reverchon†
Department of Chemical and Food Engineering, University of Salerno, Via Ponte Don Melillo 1,
84084 Fisciano, Italy, Research Institute on Membrane Technology, ITM-CNR, c/o University of Calabria, Via Pietro
Bucci 17C, 87036 Rende (CS), Italy, and Department of Chemical Engineering and Materials, University of Calabria,
Via Pietro Bucci 17C, 87030 Rende, Italy
Downloaded via NATL TAIWAN UNIV SCIENCE & TECHLGY on November 14, 2019 at 03:42:03 (UTC).

ABSTRACT In this work, poly(vinylidene fluoride) copolymer with hexafluoropropylene (PVDF-HFP) membrane-like aerogels have
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

been generated for the first time. PVDF-HFP gels have been prepared from polymer-acetone solutions by adding various amounts
of ethanol. A series of supercritical drying experiments have been performed at different pressures (from 100 to 200 bar) and
temperatures (from 35 to 45 °C) and at various polymer concentrations (from 5 to 12 wt %). The effects of the process conditions on
the membrane morphology have been evaluated, and structure-property relationships have been found. In all cases, the membranes
exhibit interconnected structures with nanosized pores and high porosity, leading to reduced resistance to the gas mass transfer and
high hydrophobic character of the surfaces. These membrane-like aerogels promise to form a new class of highly hydrophobic porous
interfaces, potentially suitable to be used in membrane operations based, for example, on the contactor technology.

KEYWORDS: membranes • aerogels • supercritical CO2, interfaces • contactors

INTRODUCTION ness of the films, reducing the period of operational time.

T he copolymer of poly(vinylidene fluoride) with hexaflu- Recently, PVDF-HFP membranes have also been generated
oropropylene (PVDF-HFP) is an acid-resistant, inert, by a new technique in which supercritical carbon dioxide
and semicristalline material. Traditionally, PVDF mem- (SC-CO2) replaces the liquid nonsolvent (i.e., SC-IPS process)
branes find large application in advanced fields of contactor (22-24). Compared to the dry-wet process, the SC-IPS
technology (1), catalysis (2-4), biomedicine (5-8), as well process can give several advantages: SC-CO2 substitutes the
as transductors (9) for the polymorphism that characterize nonsolvent, reducing the potential pollution; the membrane
this material. Because of these important technological is obtained without additional post-treatments because SC-
applications, many works have been developed with the aim CO2 completely extracts the solvent; it is possible to modu-
of producing porous PVDF-HFP structures (1, 10-20), late the membrane morphology, cells, and pore sizes simply
emphasizing often the events controlling the phase-separa- by changing the operative conditions (22-34). The results
tion phenomena (1, 13, 19). However, porous PVDF mem- obtained in the case of PVDF-HFP membranes confirmed
branes prepared according to traditional dry-wet processes the versatility of the process (through a change in the SC-
rarely exhibit well-controlled morphology and chemistry, CO2 solvent power, either a leafy morphology or a cellular
which are characteristics necessary to make them the ideal structure was obtained (22), but skinned surfaces were
interfaces for membrane operations such as, for example, always obtained 22-24).
contactors (1). Non-well-defined pore size and pore distribu-
Another interesting process that could be a valid alterna-
tion, coalescence phenomena, low surface, and overall
tive for the preparation of porous membranes is the gel
porosity influence, in turn, the final performance of the
membranes, producing interfaces not suitable for promoting drying process (35-37). Unlike the phase-inversion process,
the uniformity and high productivity of the process (21). In the starting sample is a gel (not a solution) and the porous
addition, the lack of hydrophobicity due to the use of structure (i.e., aerogel) is formed during the gelation process;
hydrophilic pore formers affects significantly the waterproof- this process can assure a uniform and symmetric skinless
nanostructured aerogel morphology, but some drawbacks
can seriously affect gel formation. The surface tension of the
* Corresponding author. Tel.: +39(0)89964232. Fax: +39(0)89964057. E-mail:
scardea@unisa.it.
solvent to be eliminated can cause the collapse of the gel
Received for review October 7, 2008 and accepted December 10, 2008 polymeric structure (due to the cohesive forces between the
†
University of Salerno. liquid solvent and the polymeric nanosized network), leading
‡
Research Institute on Membrane Technology, ITM-CNR, c/o University of to a partially nonporous structure, and long processing times
Calabria.
§
Department of Chemical Engineering and Materials, University of Calabria. are usually necessary. For these reasons, the removal of the
DOI: 10.1021/am800101a solvent from the gel, without damaging the overall network
© 2009 American Chemical Society of the polymer, represents an attractive strategy for the

www.acsami.org VOL. 1 • NO. 1 • 171–180 • 2009 171

Published on Web 01/08/2009
 realization of well-patterned aerogels that could be useful as
porous membranes.
ARTICLE
In literature, some attempts to prepare membranes by
means of the gel drying process have been reported by Cho
and Lee (35). PVDF gels were prepared from PVDF/γ-
butyrolactone solutions by using a hot-air-assisted drying
process. Nonporous structures characterized by polymeric
particles were obtained. Dasgupta et al. (36, 37) studied two
different drying processes to obtain membranes. In the first
work (36), the gels were obtained from PVDF/organic diester
solutions that were then immersed in a cyclohexane bath
for 12 h to replace the solvent into the gels. The process was
repeated for 6-7 days. Then, the gels were dipped into
methanol for 1 day and, finally, dried in a vacuum at 60 °C
for 3 days, yielding partially porous structures. In a subse-
quent work, Dasgupta et al. (37) tried to avoid the collapse FIGURE 1. Experimental apparatus.
of the structure starting from a PVDF/camphor solution. In
Table 1. Composition of the Gels Prepared
this way, they removed the solvent exploiting its tendency
acronym PVDF-HFP [wt %] acetone [wt %] ethanol [wt %]
to easily sublimate under near-“freeze-drying” conditions.
However, the drying time was longer. A new method to A 5 60 35
avoid the collapse of the structure and to obtain porous B 7 60 33
structures in a shorter time was presented by Daniel et al. C 10 60 30
(38); a SC-CO2-assisted drying process was used for the D 12 60 28
formation of syndiotactic polystyrene aerogels finalized to
99.8%, Sigma-Aldrich, St. Louis, MO); CO2 (purity 99%) was
the production of polymeric manufacts. The aerogel forma- purchased from Società Ossigeno Napoli (Napoli, Italy). Gases
tion was obtained in a single step, preventing the structure (CO2, CH4, and O2) for permeation experiments were used as
of the gel from collapsing. received (purity 98%, Pirossigeno Company, Cosenza, Italy).
In the present work, a SC-CO2-assisted drying process is Ultrapure water was used for contact-angle measurements
(filtered by the USF ELGA plant). A total of 12 measurements
proposed for the first time as a powerful nonconventional
were done and averaged, and the plateau values were com-
approach to fabricating porous PVDF-HFP membranes. The pared. All materials were processed as received.
supercritical fluid has been used to dry the gel, resulting in Preparation of Aerogels Like Membranes. Gels of PVDF-HFP
stable aerogels having structures like that of skinless mem- were prepared by dissolving the polymer in acetone under
branes with high overall porosity. This technique provides vigorous stirring at concentrations ranging from 5 to 12 wt %.
Then, different amounts of ethanol were added to the homo-
further multifaceted advantages with respect to the tradi-
geneous solution warmed at 50 °C (Table 1). Each polymer
tional phase-inversion techniques: uniform nanosized po- solution (i.e., dope) was placed in a formation cell (steel caps
rous PVDF-HFP membranes can be generated in only one with a diameter of 2 cm). The formation cell was reposed in a
step without using hydrophilic pore formers; stable mem- freezer at -20 °C until a gel was formed (∼1 h). The drying
brane networks can be achieved by removing simulta- process was performed in a laboratory apparatus equipped with
a 316 stainless steel high-pressure vessel with an internal
neously all solvents in an ecofriendly way; the solvent
volume of 80 mL, in which SC-CO2 contacts the polymer gel in
recovery can be performed by simply depressurizing the a single pass (Figure 1). The gel was placed in the vessel, which
system; the methodology results in reduced costs and was closed and filled from the bottom with SC-CO2 up to the
shorter time. Changes in the final membrane morphology desired pressure using a high-pressure pump (Milton Roy Mil-
and pore size, transport, and surface properties have been royal B). The time elapsed before SC-CO2 contacted the gel was
around 60 s. The apparatus operated in a continuous mode for
analyzed as a function of process parameters such as the
1 h, i.e., with a constant CO2 flow rate of 1.5 kg h-1. At the end,
polymer concentration, pressure, and temperature. The the vessel was slowly depressurized for 10 min and the aerogels
aerogels prepared according to this new approach exhibit a were depleted from the support. The operative conditions of
good performance when working as membranes. The su- pressure and temperature were modified, ranging from 100 to
percritical-assisted drying approach promises, therefore, to 200 bar and from 35 to 45 °C, respectively.
Gelation Boundaries. A phase diagram of the PVDF-HFP/
become a strategic alternative to the traditional methods for
acetone/ethanol system was obtained by dropwise addition of
the fabrication of a new class of porous PVDF-HFP mem- the nonsolvent (ethanol) under continuous stirring at 25 °C
branes suitable for advanced chemical separations. (Figure 2). The gelation point was visually observed by means
of the increased viscosity of the solution (13). The maximum
concentration of polymer in the starting solution was 20 wt %
MATERIALS AND METHODS in acetone. Further additions of polymer were limited either by
Materials. PVDF-HFP (number average 199 × 103, weight the excessive viscosity of the solution, which produced inef-
average 353 × 103, polydispersity 1.8, density 1.78 g cm-3) was ficient stirring, or by a too strong inherent turbidity of the
kindly supplied by Solvay SA (Ixelles, Belgium); acetone (purity polymer dopes. The diagram in Figure 2 is thus an isotherm (at
99.5%, Sigma-Aldrich, St. Louis, MO) was used as the solvent 25 °C) for the ternary system PVDF-HFP/acetone/ethanol. The
and ethanol as the nonsolvent for formation of the gel (purity gelation boundary “moves” inside the diagram through a

172 VOL. 1 • NO. 1 • 171–180 • 2009 Cardea et al. www.acsami.org

 ARTICLE
FIGURE 2. Phase diagram of the PVDF-HFP/acetone/ethanol system
at 25 °C with a gelation line and with processed gel compositions
(A, 5%/60%/35%; B, 7%/60%/33%; C, 10%/60%/30%; D, 12%/60%/
28%).

FIGURE 4. DSC analysis performed on polymer powder and mem-

branes: (a) first heating run; (b) cooling run.

atomic force microscopy (AFM; Nanoscope III Digital Instru-

ments, VEECO Metrology Group). AFM was operated in tapping
mode by a tip attached to the end of an oscillating cantilever
across 10 × 10 µm2 of sample surface at a rate of 1.0 Hz. The
pore size, pore distribution, and surface porosity were estimated
from AFM images. The pore size was a log-normal distribution
FIGURE 3. IR spectra collected on polymer powder and membrane- for all membranes; the cumulative distribution function versus
like aerogels at different polymer concentrations. the ascending pore size was a straight line on log-normal
probability paper. The mean pore size was calculated from the
change in the temperature of the solution. In particular, upon log-normal plot, and the pore distribution was expressed by the
warming of the solution (for example, up to 50 °C), the gelation probability density function [df(dp)/ddp, µm-1] according to
boundary shifts toward the PVDF-HFP/ethanol axis and the ref19.
homogeneous region largely increases. Upon cooling of the The overall porosity (
), expressed as the “void volume
solution (for example, at -20 °C), the gelation boundary shifts fraction” inside the membrane, was calculated from the appar-
toward the acetone apex, leading to a smaller homogeneous ent density of the membrane (Fm) and the density of the
region. PVDF-HFP powder (Fp ) 1.78 g cm-3). The apparent mem-
Aerogel Crystallinity. Changes in the aerogel crystallinity brane density was determined by measuring the volume and
were detected for all samples by thermal analysis using Dia- the weight of the film. This operation was performed on 10
mond Pyris differential scanning calorimetry (DSC; Perkin samples for each experimental condition with a standard devia-
Elmer, Waltham, MA). To obtain an identical thermal history, tion of 0.02.
each sample was heated from 50 to 200 °C at 15 °C min-1, To confirm the estimated porosity values and evaluate the
cooled down to 50 °C, and heated again up to 200 °C. The interconnectivity degree into the bulk of the films, specimens
weighed amount of sample for each experiment was ap- of various samples were filled by a liquid characterized by a low
proximately 8.0 mg. The heat flow was normalized by the surface tension (Fluorinert FC40, 3M Novec) and the void
sample weights. All transitions were compared with those of volume fraction was calculated from the weight and density of
the polymer powder. IR spectra were also collected in transmis- the solvent adsorbed inside the open pores.
sion mode at a resolution of 4 cm-1 from each sample using a Evaluation of Gas Transport through the Membranes.
Spectrum One System (Perkin Elmer, Waltham, MA) and com- Permeation experiments to some gases (CO2, CH4, and O2) were
pared with that of the untreated polymer powder. also performed at 1 bar and 25 °C according to the fixed
Morphology and Topography Analyses. PVDF-HFP aero- volume-pressure increase mode. Four specimens of mem-
gels were cryofractured by a microtome (Bio-optica SpA, Milano, brane samples with an effective area of 2.14 cm2 were tested.
Italy; model Microm HM 550 OMVP). The samples were sputter The experimental error was e10%.
coated with gold and observed by a scanning electron micro- Contact-Angle Measurements. The wetting degree was
scope (SEM; model LEO 420, Assing, Italy) to study the film estimated by dynamic contact-angle experiments. The tech-
structure. The membrane topography was estimated using nique used consisted of the sessile drop method by a CAM 200

www.acsami.org VOL. 1 • NO. 1 • 171–180 • 2009 173

ARTICLE

FIGURE 5. (a) Gel obtained with 10 wt % of PVDF-HFP. (b and c) Membrane-like aerogel with 10 wt % of PVDF-HFP obtained by supercritical
drying. (d) PVDF-HFP nonporous polymeric film obtained by air drying.
contact-angle meter (KSV Instruments Ltd., Helsinki, Finland). Membrane-like Aerogels. Gelation Process. The
The water droplets were deposited on the membrane surface gel drying process allows one to obtain polymeric aerogels,
using a microsyringe with an automatic dispenser; the images
were captured by a digital camera. An average of 15-20 i.e., to eliminate the liquid solvent, avoiding the collapse of
readings were carried out for each specimen, and the mean the gel solid structure; one of the possible aerogel structures
value was then calculated. is formed by a nanostructured network of polymeric fibers
(38).
RESULTS AND DISCUSSION Using SC-CO2 drying, various PVDF-HFP gels were pre-
Gelation Region for the PVDF-HFP/Acetone/ pared by changing the polymer concentration from 5 to 12
Ethanol System. The gelation boundary of PVDF-HFP
wt % and keeping the concentration of acetone constant at
in acetone/ethanol solutions was determined by identifying
60 wt % (Table 1). The concentration of the ethanol was
the equilibrium gelation points, which are the compositions
varied proportionally to the polymer concentration. At 50
at which the homogeneous solutions form solid structures
°C, the polymer dope compositions lie in the homogeneous
embedded in the liquid system (gels) at 25 °C. Experimental
region; the subsequent freezing of the solution at tempera-
determination of the gelation region was limited to a maxi-
tures down to -20 °C leads the mixtures into the gelification
mum polymer concentration of 20 wt % to avoid inefficient
stirring because of a very high solution viscosity (Figure 2). region that is more extended at lower temperatures, as
As is possible to observe in Figure 2, the gelation line of the discussed in the paragraph on “gelation boundaries”. The
solution is very close to the solvent apex, confirming mixtures remain in the gelation region when raised again
the tendency of the copolymer PVDF-HFP to produce these to the ambient temperature (25 °C) (Figure 2).
kinds of structures as already observed in the literature The PVDF tendency to form gel is mainly due to the
(11, 22, 39-40); indeed, the ternary phase diagrams show semicrystalline character of the polymer. The solutions move
large gelation regions for these kinds of polymers, where quickly inside the crystallization region when aided by a soft
precipitation by solid-liquid demixing predominates on that nonsolvent like ethanol. This enters the polymer dopes into
liquid-liquid, resulting in membranes with different mor- the crystallization region where a solid-liquid demixing
phologies (22, 41). occurs. Gels with high crystalline character are then formed.

174 VOL. 1 • NO. 1 • 171–180 • 2009 Cardea et al. www.acsami.org

 ARTICLE

FIGURE 6. SEM (magnification: 10 000×) and AFM (area scan size: 10 × 10 µm2) images of the membrane surfaces obtained at 200 bar and
35 °C: (a and b) 5 wt %, (c and d) 7 wt %, (e and f) 10 wt %, and (g and h) 12 wt % of PVDF-HFP.

With this regard, IR spectroscopy and thermal analysis nonpolar R phase. This is confirmed by both the disappear-
provided useful information about the changes occurring in ance of the melting peak at 112 °C and the reduced enthalpy
the membrane crystallinity as the polymer and ethanol of fusion of the aerogels. Despite the highest thermodynamic
concentrations were varied in the mixture (Figures 3 and 4). stability and predominance of the R phase, a significant
The dissolution of the polymer in acetone reduces drasti- presence of β and γ forms is observed for the aerogels
cally the crystalline domains in favor of the most stable and formed from gels A and D, as indicated by the increased

www.acsami.org VOL. 1 • NO. 1 • 171–180 • 2009 175

ARTICLE

FIGURE 7. SEM images of the cross sections of membrane-like aerogels prepared at 200 bar and 35 °C: (a) 5 wt %, (b) 7 wt % (c) 10 wt %, and
(d) 12 wt % of PVDF-HFP.

vibrational modes typically ascribed to the stretching bands acetone/ethanol/SC-CO2 mixture is formed. Also, this mix-
of the β and γ forms (Figure 3). Further confirmation of the ture presents a surface tension near zero, and it can be
higher crystalline content is given by a somewhat larger removed from the nanosized polymeric network without
enthalpy of fusion (48 J g-1) than that estimated for the other mechanical stress for the structure because no cohesive
aerogels (44 J g-1) during the first heating run (Figure 4) as forces are present.
well as by the changes in the supercooling of the crystalliza- Parts a-c of Figure 5 show an example of a PVDF-HFP-
tion occurring during the cooling run. Indeed, different based gel obtained starting from a mixture at 10% in
positions of the crystallization peaks are present. Both polymer, 60% in acetone, and 30% in ethanol. The aerogel
aerogels A and D exhibit a crystallization temperature close formed by the SC-CO2-assisted process (Figure 5b,c) pre-
to 100 °C, slightly higher than that of the polymer powder serves both the dimension and shape exhibited by the gel
(96 °C). In contrast, higher crystallization temperatures were before the drying treatment, respectively (Figure 5a). On the
estimated for the other two aerogels (B and C). This suggests contrary, the same gel, dried in hot air, exhibits a completely
that the crystalline nucleus compositions of aerogels A and collapsed dense structure (Figure 5d). The SC-CO2-assisted
D are comparable. At higher polymer content, the high drying process is, therefore, meant to overcome the draw-
supersaturation degree of the solution promotes the nucle- back related to the collapse of the polymer network. Indeed,
ation and growth of various crystalline forms, resulting in a the fast removal of the solvent avoids the weakening of the
high polymorphic degree of the aerogel. When the polymer matrix and then the destruction of the aerogel. In addition,
concentration is reduced, a proportional amount of nonsol- the high SC-CO2 solvent power together with the absence
vent is added to the solution, making favorable environ- of surface tension at supercritical conditions avoids the
ments for the nucleation of polymorphic species. presence of traces of solvent inside the aerogels at the end
Effect of Supercritical Fluid Processing. The of the process. The result is a well-dried aerogel having the
obtainment of supercritical conditions plays a crucial role structural stability necessary to work as a membrane.
in the fabrication process of stable membrane-like aerogels. Effect of the Polymer Concentration. All polymer
As described above, SC-CO2 rapidly eliminates all of the dopes that we used are inside the gelation boundary at 25
solvents, transforming the gel in aerogel and preserving the °C, with different solvent/nonsolvent ratios. Shock freezing
nanosized polymeric structure. Indeed, CO2 at supercritical quickly enters the dopes in the gelation area, leading to
conditions can be considered as a zero-surface-tension denser structures at higher degrees of supersaturation (i.e.,
compound; when it diffuses inside the PVDF-HFP gels, an at higher polymer concentrations) as shown in Figure 6. All

176 VOL. 1 • NO. 1 • 171–180 • 2009 Cardea et al. www.acsami.org

 ARTICLE
FIGURE 9. Evaluation of the changes in the geometric mean pore
size measured from AFM images: (a) cumulative distribution of the
pore sizes; (b) probability density function curves built up from the
pore sizes measured from AFM.

lation, resulting in a bigger critical nucleus size and then in

larger porosity. As a result, aerogels like skinless membranes
are formed.
All membranes obtained from PVDF gels by the SC-CO2-
FIGURE 8. Relationships between the overall void volume fraction assisted drying process exhibit uniform and symmetric
and (a) the polymer amount per membrane surface, (b) the surface structure across the overall section of the layers, as well
porosity estimated by AFM analysis, and (c) the geometric mean pore
size as a function of the polymer concentration. illustrated by SEM images (Figure 7); indeed, SEM analysis
was performed on different zones of each sample to testify
membranes exhibit fibrous structures, but nodulelike mor- to the uniformity and symmetry of the structures obtained.
phology occurs significantly on membranes prepared from This result was also confirmed by other analyses performed
higher polymer concentration mixtures. SEM and AFM on the membranes (DSC, AFM, and IR): each analysis was
images show more compacted and skinned structures be- performed several times on different zones of each mem-
cause the initial polymer contents in the mixture are 10 and brane, resulting in comparable results.
12 wt % (Figure 6c,d); however, the particulate-like mor- When the cross sections of all films are compared, an
phology predominates on the top layer of the membrane increase in the overall porosity can be appreciated with
surfaces, suggesting a dependence from the local polymer decreasing polymer concentration. An inverse proportional-
concentration and the tendency of the polymer to minimize ity between the void volume fraction (
, %) and the amount
the surface energy, forming solid spheres. Small amounts of polymer per unit of surface (G, g m-2) was, therefore,
of acetone tend to evaporate before contacting the super- estimated (Figure 8a). On the contrary, a linear increase
critical fluid. This could produce a further increase in the between the overall and surface porosities was observed
polymer concentration at the gel/air interface, resulting in (Figure 8b). This suggests a uniform increase in the forma-
an increased degree of supersaturation and leaving nodule tion of liquid micelles during the nucleation stage of gelation,
aggregates. leaving a slightly larger and uniformly distributed pore size
In contrast, open and reduced spherulitic structures as the polymer concentration decreased (Figure 8c). Changes
characterize the membranes derived from solutions at lower in the porosity are, therefore, the result of a gradual tendency
polymer concentration (5 and 7 wt %; Figure 6a,b). This fact of the pore size to increase with decreasing polymer con-
suggests a slower occurrence of crystallization-induced ge- centration, as was well quantified by AFM measurements

www.acsami.org VOL. 1 • NO. 1 • 171–180 • 2009 177

ARTICLE

FIGURE 11. SEM images of PVDF-HFP membrane-like aerogels

FIGURE 10. Effect of the pressure on the final morphology and the obtained at 200 bar and 45 °C: (a) whole section; (b) internal
cell size of membrane-like aerogels prepared from gels with 10 wt structure.
% of PVDF-HFP at 35 °C and two different operating pressures: (a)
200 bar; (b) 100 bar.
which remains constant (i.e., approximately 88%; Figure
(Figures 8c and 9). All plots of the median ranks versus the 10). This result can be explained considering that a pressure
increasing pore sizes on the log-normal probability paper variation only changes the solvent power of SC-CO2, which
were linear with correlation coefficients g0.90r2 (Figure 9a). tends to increase with the pressure. Operating in this way,
Unimodal pore distributions were estimated for all mem- it is possible to dry the gel in a shorter time without
branes, even if a broadening of the distribution was obtained influencing the membrane structure. In particular, the drying
for membranes prepared at 5 wt % of polymer (Figure 9b). time decreased down to 30 min as the pressure increased
This fact could be ascribed to coalescence phenomena from 100 to 200 bar. This result confirms that the aerogel
occurring in the polymer-lean region during gel formation. structure is formed during the gelation process and not
The obtainment of values of overall porosity up to 95% during the drying process. On the contrary, an increase in
with the complete suppression of macrovoids is an impor- the temperature from 35 to 45 °C reduces dramatically the
tant result with respect to traditional PVDF-HFP mem- membrane porosity from 90% to 40%, resulting in a strong
branes, where fingerlike structures are often combined with densification of the polymeric matrixes across the overall
dense skin layers. On the other hand, a uniform distribution section (Figure 11). Specifically, the gel thickness decreases
of the void volume fraction through the overall film is a sensibly from 2 to 0.4 mm for membranes prepared from
crucial factor for the productivity of most of the membrane polymer dopes at 10 wt %, 45 °C, and 200 bar (Figures 10a
operations, but it gains particular importance if the mem- and 11b), resulting in compact structures.
brane is used to equip, for example, contactor devices. This result could be due to the volatility of the liquid
Effect of the Pressure and Temperature. Experi- solvents, which is responsible for an increase in the polymer-
ments were performed at different pressures, keeping the rich phases during the membrane formation; because the
other process parameters constant (35 °C and 10 wt % of boiling point of acetone is 56.5 °C, a nonnegligible vaporiza-
PVDF-HFP). A decrease in the pressure from 200 to 100 tion at a temperature of 45 °C is expected. This fact
bar does not affect the overall porosity of the membranes, produces a decrease in the solvent/nonsolvent ratio, result-

178 VOL. 1 • NO. 1 • 171–180 • 2009 Cardea et al. www.acsami.org

 for example, the inhibition of spreading phenomena. Liquid
water must be prevented from penetrating inside the mem-

ARTICLE
brane pores to keep the interface of the two media constant
at the mouth of each single channel of the films. Thus, a high
hydrophobic character of the surface has to be assured.
Dynamic contact-angle experiments performed on the sur-
face of the various membrane-like aerogels produced in this
work confirmed the high hydrophobic character of the
membranes (Figure 12b). As the polymer concentration
decreased, an increase of the contact-angle values up to 143
( 3° was observed, yielding useful information about the
waterproofness of the membrane surfaces. A hysteresis of
66 ( 5° was calculated between the advancing and receding
angles, because of surface roughness effects, responsible for
the capture of hydrophobic air inside the membrane pores
and, consequently, for the enhanced water repellence with
increasing pore size (1). These aerogels are, therefore,
suggested to form an interesting class of new membranes
having the necessary structure-property characteristics for
equipping contactor devices.

CONCLUSIONS
The SC-CO2-assisted gel drying process is a promising
approach for the fabrication of aerogels having characteris-
tics useful for membranes equipping contactor devices. This
FIGURE 12. Transport property relationships assessed for PVDF-HFP technique offers the powerful advantage of producing stable
membranes prepared by the SC-CO2-assisted drying process: (a) symmetric membranes with nanosized pores, high porosity,
correlation between the overall void volume fraction and rising gas
mass transfer; (b) estimation of the wetting degree on the surface
and porous surface, avoiding the collapse of the structure
of all PVDF-HFP membrane-like aerogels. due to the absence of surface tension of SC-CO2. The
polymer concentration is a crucial controlling parameter for
ing in a reduction of the liquid micelle size and then in a changing the structural aspects of the films. The membrane-
major densification of the final structure. like aerogels exhibited, in turn, modulated pore size and
Structure-Property Relationships. PVDF mem- porosity. These morphological achievements are responsible
branes achieved by the SC-CO2-assisted drying process were for the enhanced transport and surface properties of the
tested to establish their suitability to work as nonselective membranes. In addition, these membrane-like aerogels
interfaces for advanced applications such as contactor op- exhibit polymorphic character, the crystalline composition
erations. One of the most important requirements of the of which could be usefully changed as a function of the
contactor technology is the high mass transfer through the solvent/nonsolvent ratio used. Finally, the manufacturing
membrane. The target of high productivity can be achieved approach appears less expensive and ecosustainable with
if a high overall porosity characterizes the membranes, respect to traditional manufacturing approaches.
resulting in a reduced resistance to mass transport.
Indeed, an increase in the overall porosity has been REFERENCES AND NOTES
estimated with decreasing polymer concentration (Figure 8a) (1) Gugliuzza, A.; Drioli, E. J. Membr. Sci. 2007, 300, 51>62.
and rising surface porosity (Figure 8b) for all membrane-like (2) Buonomenna, M. G.; Drioli, E.; Nugent, W.; Prins, L.; Scrimin, P.;
Licini, G. Tetrahedron Lett. 2004, 45, 7515–7518.
aerogels. This characteristic produces a reduced resistance (3) Bottino, A.; Capannelli, G.; Comite, A. Desalination 2002, 146, 35–
to the mass transfer into the bulk of the films and, therefore, 40.
high gas permeations (Figure 12a). When the permeation is (4) Bottino, A.; Capannelli, G.; Comite, A.; Di Felice, R. Desalination
2002, 144, 411–416.
compared to the various gases, the membranes exhibited a (5) Guenard, V.; Valentini, R. F.; Aebischer, P. Biomaterials 1991, 12,
somewhat higher permeation to methane to be ascribed to 259 –263.
the lowest molecular weight of this gas. Despite the fact that (6) Chen, H.; Soldani, G.; Galletti, P. M.; Goddard, M. ASAIO J. 1992,
38, 201–206.
a convective flux can characterize the gas transfer through (7) Grandien, J. D. U.S. Patent 4,203,847, 1980.
these membranes, a small Knudsen component in the mass- (8) Karakulski, K.; Morawski, W. A.; Grzechulska, J. Sep. Purif.
transfer mechanism cannot be excluded. Changes in the Technol. 1998, 14, 163–173.
(9) Gregorio, R.; Cestari, M. J. Polym. Sci., Part B: Polym. Phys. 1994,
intrinsic structural parameters of the membrane-like aero- 32, 859 –870.
gels also influence positively the final surface properties of (10) Lin, D. J.; Chang, H. H.; Chen, T. C.; Lee, Y. C.; Cheng, L. P. Eur.
the films in terms of controlled mutual liquid-membrane Polym. J. 2006, 42, 1581–1594.
(11) Young, T. H.; Cheng, L. P.; Lin, D. J.; Fane, L.; Chuang, W. Y.
interactions. Indeed, another important achievement for Polymer 1999, 40, 5315–5323.
processing hydrophobic membranes in contactor devices is, (12) Deshmukh, S. P.; Li, K. J. Membr. Sci. 1998, 150, 75–85.

www.acsami.org VOL. 1 • NO. 1 • 171–180 • 2009 179

 (13) Cheng, L. P.; Lin, D. J.; Shih, C. H.; Dwan, A. H.; Gryte, C. C. J. 2006, 46, 188 –197.
Polym. Sci., Part B: Polym. Phys. 1999, 37, 2079 –2092. (30) Matsuyama, H.; Yamamoto, A.; Yano, H.; Maki, T.; Teramoto, M.;
ARTICLE
(14) Wang, D.; Li, K.; Teo, W. K. J. Membr. Sci. 2000, 178, 13–23. Mishima, K.; Matsuyama, K. J. Membr. Sci. 2001, 194, 157–163.
(15) Wang, M. D.; Li, K.; Teo, W. K. J. Membr. Sci. 1999, 163, 211– (31) Reverchon, E.; Cardea, S. J. Supercrit. Fluids 2005, 35, 140 –146.
220. (32) Temtem, M.; Casimiro, T.; Aguiar-Ricardo, A. J. Membr. Sci. 2006,
(16) Yan, L.; Wang, K.; Ye, L. J. Mater. Sci. Lett. 2003, 22, 1713–1717. 283, 244 –252.
(17) Tomaszewska, M. Desalination 1996, 104, 1–11. (33) Temtem, M.; Casimiro, T.; Aguiar-Ricardo, A.; Mano, F. J. J.
(18) Feng, C.; Shi, B.; Li, G.; Wu, Y. J. Membr. Sci. 2004, 237, 15–24. Supercrit. Fluids 2008, 43, 542–548.
(19) Khayet, M.; Feng, C. Y.; Khulbe, K. C.; Matsuura, T. Polymer 2002, (34) Reverchon, E.; Cardea, S.; Rapuano, C. J. Appl. Polym. Sci. 2007,
43, 3879 –3890. 104, 3151–3160.
(20) Chen, N.; Hong, L. Polymer 2002, 43, 1429 –1436. (35) Cho, J. W.; Lee, G. W. J. Polym. Sci., Part B: Polym. Phys. 1996,
(21) Gugliuzza, A.; Aceto, M. C.; Macedonio, F.; Drioli, E. J. Phys. Chem. 34, 1605–1611.
B 2008, 34, 10483–10496. (36) Dasgupta, D.; Manna, S.; Malik, S.; Rochas, C.; Guenet, J. M.;
(22) Reverchon, E.; Cardea, S. Ind. Eng. Chem. Res. 2006, 45, 8939 – Nandi, A. K. Macromol. Symp. 2005, 222, 175–180.
8945. (37) Dasgupta, D.; Nandi, A. K. Macromolecules 2005, 38, 6504 –6512.
(23) Cao, J. H.; Zhu, B. K.; Ji, G. L.; Xu, Y. Y. J. Membr. Sci. 2005, 266, (38) Daniel, C.; Alfano, D.; Venditto, V.; Cardea, S.; Reverchon, E.;
102–109. Larobina, D.; Mensitieri, G.; Guerra, G. Adv. Mater. 2005, 17,
(24) Huang, S.; Wu, G.; Chen, S. J. Membr. Sci. 2007, 293, 100 –110. 1515–1518.
(25) Reverchon, E.; Cardea, S. J. Membr. Sci. 2004, 240, 187–195. (39) Lin, D. J.; Chang, C. L.; Huang, F. M.; Cheng, L. P. Polymer 2003,
(26) Cardea, S.; Gugliuzza, A.; Schiavo Rappo, E.; Aceto, M.; Drioli, 44, 413–422.
E.; Reverchon, E. Desalination 2006, 200, 58 –60. (40) Stephan, A. M.; Renganathan, N. G.; Gopukumar, S.; Teeters, D.
(27) Xu, Q.; Pang, M.; Peng, Q.; Li, J.; Jiang, Y. J. Appl. Polym. Sci. 2004, Mater. Chem. Phys. 2004, 85, 6 –11.
94, 2158 –2163. (41) Van de Witte, P.; Dijkstra, P. J.; Van den Berg, J. W. A.; Feijen, J.
(28) Xu, Q.; Pang, M.; Peng, Q.; Li, J.; Jiang, Y.; Wang, H.; Zhu, M. J. Membr. Sci. 1996, 117, 1–31.
J. Appl. Polym. Sci. 2005, 98, 831–837.
(29) Reverchon, E.; Schiavo Rappo, E.; Cardea, S. Polym. Eng. Sci. AM800101A

180 VOL. 1 • NO. 1 • 171–180 • 2009 Cardea et al. www.acsami.org