Department of Chemical and Food Engineering, Federal University of Santa Catarina, 88040-970 Florianpolis, SC, Brazil
Technological Federal University of Paran, 85884-000 Medianeira, PR, Brazil
a r t i c l e
i n f o
Article history:
Received 23 July 2014
Received in revised form 6 March 2015
Accepted 8 March 2015
Available online 13 March 2015
Keywords:
Polymeric membranes
Non-aqueous systems
Organic solvents
Membrane conditioning
n-Hexane
Alcohols
a b s t r a c t
Several pre-treatments and solvents were tested through permeation in commercial polymeric membranes (UH004, UP005, UP010 and UH050 Microdyn-Nadir) with different
molar mass cut-offs (4 kDa, 5 kDa, 10 kDa and 50 kDa, respectively), in order to evaluate
their efciency and stability through non-aqueous solvent permeation. After pre-treatments and permeations, membranes were characterised with contact angle, Fourier
Transform Infrared Spectroscopy (FTIR), and Field Emission Scanning Electronic
Microscopy (FESEM) to evaluate structural integrity. The results of solvent permeation suggest that selectivity of the polymeric matrix depends not only on pore size, but also, in
great extent, on the interaction between solvent and polymer. A strong relation can be
noted between the conditioning length and permeability for ethanol pre-treatments.
Permeability to n-hexane increased from 4 to 18 times after pre-treatment, depending
on the time of exposure to ethanol and n-hexane. Characterisation analyses show no signicant changes on the membranes surface. In some cases, discrepancies observed among
permeate uxes and contact angles might be an indicative of the occurrence of swelling
and plasticisation. However, results suggest the feasibility in the use of these membranes
for the recovery of solvents in the oil industry, if suitable process parameters are chosen.
2015 Elsevier Ltd. All rights reserved.
1. Introduction
The use of polymeric membranes is already a feasible
technology for water and aqueous solutions treatment.
Although there are many studies on the preparation of solvent resistant membranes, only few are commercially
available as Borsig, PolyAn, Duramem, Puramem and
SolSep membranes. Most of the commercial available
polymeric membranes are produced for aqueous systems
Corresponding author. Tel.: +55 48 3721 2529.
E-mail address: di.luccio@ufsc.br (M. Di Luccio).
http://dx.doi.org/10.1016/j.eurpolymj.2015.03.010
0014-3057/ 2015 Elsevier Ltd. All rights reserved.
and they present some drawbacks to industrial implementation in non-aqueous solvent-based processes. The
problems are generally related to membrane lack of stability, due to the changes caused by the solvents in its structure, and low permeate uxes [13]. Therefore, studies in
this eld could benet industry segments from edible oil
processing to pharmaceuticals and petrochemical for its
reduced energy consumption and process simplication.
Many authors suggest that the pre-treatment of polymeric membranes by immersion in organic solvents can
prevent the pore collapse during permeation and enable
the solvent entrance into the pores. This enhances the
493
Table 1
Manufacturers data for the membranes tested.
Membrane
UH004
UP005
UP010
UH050
Material
MMCO (kDa)
T max (C)
pH range
Retention (%)
PESH
4
95
014
9299a
PES
5
95
014
9198a
PES
10
95
014
6385b
PESH
50
95
014
7588c
494
Fig. 1. Schematic diagram of the permeation assays with the solvents used in pre-treatment.
Table 2
Solvent viscosities.
Solvent
Water
Ethanol
iso-Propanol
n-Hexane
1.002
1.078
1.960
0.326
proportional to the applied pressure differential, and inversely proportional to uid viscosity and bed length. This is
the basis for Darcys Law, given by the Eq. (1).
Q K DP
A l DL
where Q is the permeate ow rate (m3 s1); A, the membrane area (m2); K is the permeability of ltration media;
DP is the pressure differential (Pa); l is the uid viscosity
(Pa s); and L is the bed length (m).
Eq. (1) can be simplied to Eq. (2). The plot of experimental solvent ux (J) vs the pressure difference applied
to the process (DP) yield the membrane permeability,
obtained by the plot slope, usually expressed in
L m2 h1 bar1. Thus, we can note that the permeability
depends on the solvent viscosity. If the solvent permeability is multiplied by solvent viscosity (Table 2), expressing
both in the appropriate units, the normalised permeability
is obtained, which corresponds to K/DL.
J Lp DP
Table 3
Membrane permeability to different solvents studied.
Normalised permeability (1014 m)
Water
Ethanol
iso-Propanol
n-Hexane
UH004
UP005
UP010
UH050
UH004
UP005
UP010
UH050
20.7 2.1
11.0 0.9
4.4 0.9
3.0 0.8
27.8 3.8
29.6 6.7
3.6 1.2
6.1 1.3
97.8 10.0
41.2 3.8
8.4 0.9
52.1 3.2
250.3 17.0
118.4 8.2
29.4 3.1
81.8 6.4
5.7 0.6
3.3 0.3
2.4 0.5
0.3 0.1
7.7 1.0
8.9 2.0
1.9 0.7
0.5 0.1
27.2 2.8
12.3 1.1
4.6 0.5
4.7 0.3
69.7 4.8
35.5 2.5
16.0 1.7
7.4 0.6
495
Fig. 3. Permeate uxes for pre-treatment solvent permeation assays in (a) UH004, (b) UP005, (c) UP010 and (d) UH050, respectively.
496
results in grouping and enlargement of hydrophobic clusters, thus it might affect the pore sizes as well. Therefore,
the clustering effect results in a slight decrease of total
pore number but with a small enlargement of pores, thus,
increasing non-polar solvent uxes.
On the other hand, swelling and plasticisation are to
some extent opposing phenomena, yet coexistent and
interdependent. Swelling phenomenon is the penetration
of the solvent in the polymeric matrix. According to Ebert
and Cuperus [19], when dense membranes swell, the free
volume increases. This would make dense membranes
more open, resulting in lower rejections and higher uxes.
On the contrary, in porous membranes, since the solvent
penetrates in the polymeric matrix, it swells and tends to
press the pore walls, resulting in narrower pores. In turn,
the phenomenon of plasticisation can be attributed to
swelling stresses on the polymeric matrix. The addition
or sorption of some species, in this case, the organic solvent, causes an enhancement of local segmentation motion
of the polymer chains, which leads to a weakening or softening effect characterised by a reduction in glass transition
temperature and modulus. This increase in local segment
motion results in enhanced transport rate of all the penetrants to be separated, resulting in losses in membrane
selectivity. Furthermore, plasticisation also leads to
acceleration in polymer physical aging rate, with thickness
decrease and thus, signicant ux reduction [20].
However, if swelling and plasticisation takes place, the
effect on the ux is hardly predictable and the membrane
can even become unstable [21].
To summarise, clustering effect, swelling and plasticisation are phenomena that can coexist and, although it
seems that swelling and plasticisation are opposing phenomena, they can both lead to lower uxes, yet only plasticisation results in less selectivity. As to clustering, it is
desirable that it may occur mainly on pre-treatment stage.
However, while exposed to the solvents, depending on
their interaction with the polymer, membranes are susceptible to all the aforementioned phenomena.
Only UH004 (Fig. 3a) showed the expected behaviour of
uxes (water > ethanol > iso-propanol > n-hexane). For the
other membranes, iso-propanol presented the lowest
uxes. In UP010, at 6 bar, it was almost seven times lower
than n-hexane ux and almost ve times lower in UH050,
at 1.5 bar. For UP010, n-hexane ux is higher than ethanol
and iso-propanol uxes. This fact does not occur for UP005
neither for UH050. This fact might be a considerable indication that not only is pore size playing an important role
in non-aqueous solvent ux, but also the interaction
between the solvent and membrane material. Whereas
UH050 has the largest pore size from all the membranes
tested, its material is slightly different (UP010 PES;
UH050 PESH, more hydrophilic).
According to Coutinho et al. [8], the two determining
factors for membrane efciency are size exclusion and
chemical interactions between the solvent and/or solute
and the membrane active layer. Studies by Yang et al. [9]
with non polar solvents using polymeric membranes also
showed that the transport mechanism through these was
not only function of solvent viscosity or diffusion through
membrane pores, afrming that there is some interaction
between the solvent and the membrane, which is determinant for permeate ux. These ndings corroborate with
data in the present study. Although the iso-propanol molecule is smaller than n-hexanes, the latter had greater
uxes.
The results for permeation of treatment solvent are also
corroborated by the contact angle measurements (Fig. 4).
The highest increases in contact angles were noticed for
membranes pre-treated and permeated with n-hexane.
Thus, the membrane afnity with non-polar molecules
increased and facilitated their mass transport through the
pores. However, although the hydrophobic character with
n-hexane treatment is greater than that obtained with
other solvents, the permeate uxes for this solvent were
not the highest, probably due to pore size inuence on permeation. The highest contact angle was observed for
UP010 after treatment and permeation with n-hexane.
This membrane also presented n-hexane uxes higher than
ethanol and iso-propanol. Therefore, an increase in UP010
afnity with n-hexane is again highlighted.
The interaction mechanisms between solvent and
membrane are thought to undergo a reorganisation of
the polymeric chains, resulting in larger pores and
correspondingly lower rejections of dissolved components
[16].
The polymer/solvent interactions can be expressed in
terms of their solubility parameters. The overall solubility
parameter (Hansen solubility parameter), d, can be calculated as a function of the individual contributions related
to dispersive forces, polar interactions (dipoledipole
forces) and hydrogen bonding, respectively dd, dp and dh:
0:5
497
Water
Ethanol
iso-Propanol
n-Hexane
PES
a
dd
dp
dh
15.6
15.8
15.9
14.9
19.6
16.0
8.8
6.8
0.0
10.8
42.3
19.4
16.4
0.0
9.2
47.8
26.5
23.8
14.9
24.2
Vm (cm3 mol1)
v (dimensionless)
DdSP (dimensionless)
18.0
58.4
76.5
130.8
4.4
0.5
0.3
4.9
33.7
11.1
9.0
14.9
[22].
DdSP
q
dd;P dd;S 2 dp;P dp;S 2 dh;P dh;S 2
On the same way, Jansen et al. [22] reports an interaction parameter (v), to verify the matching between solvent
and polymer. This parameter can be calculated through the
equation:
v 0:34
Vm
dP d2
RT
Table 5
Permeability of each tested membrane to n-hexane after pre-treatments 58, with ethanol and iso-propanol.
Permeability (L h1 m2 bar1)
Pre-treatment
UH004
UP005
UP010
T5
T6
T7
T8
2 h ethanol + 2 h n-hexane
2 h ethanol + 12 h n-hexane
12 h ethanol + 2 h n-hexane
12 h ethanol + 12 h n-hexane
4.5 0.4
7.7 0.3
14.3 0.6
16.4 0.5
4.5 0.3
5.8 0.5
13.3 1.2
18.3 1.1
21.8 1.5
52.0 2.5
68.7 3.7
117.2 5.8
UH050
15.2 1.3
33.4 1.4
189.6 5.2
272.3 6.5
T5
T6
T7
T8
2 h iso + 2 h n-hexane
2 h iso + 12 h n-hexane
12 h iso + 2 h n-hexane
12 h iso + 12 h n-hexane
14.0 0.9
11.2 1.0
14.4 1.1
29.4 1.7
13.1 1.2
11.3 0.8
8.7 0.7
11.0 0.9
49.3 4.2
69.0 2.8
104.7 7.8
29.4 2.2
5696.4 8.5
3466.5 12.7
5128.9 25.2
4817.6 22.2
498
Fig. 5. Contact angles for the membranes pre-treated with ethanol (a)
and iso-propanol and (b) after n-hexane permeation assays.
499
Fig. 6. FESEM micrographs of UH050, new membrane and after Treatment 8 with ethanol.
Fig. 7. FTIR spectra for UP010 membrane before and after the pre-treatments.
500
These groups are not present in the base polymer, and their
intensity decrease after pre-treatment. Thus, they suggest
the presence of preservative agents. Belfer et al. [33] also
found these peaks in Millipore PES membranes and
concluded that they were related to preservatives existing
on the new membranes, as the peaks disappeared after
several washing procedures, with water and ethanol.
Vibrational bands around 3300 cm1 were also expressive in new membranes, with drastic reductions after pretreatments. According to Hajatdoost et al. [32], this region
is assigned to the stretching of OH groups and hydrogen
bonds, and is related to membranes water absorption.
Socrates [30] afrms that polyethersulfone may exist in
equilibrium with up to 2% of water, thus, bands due to
water will normally also be observed in the spectra of this
polymer. Therefore, after conditioning, the extinction of
these peaks can be noticed, which indicates the removal
of membranes humidity, making it more hydrophobic.
These results corroborate with contact angle increases
after conditioning, as seen in Figs. 4 and 5.
4. Conclusions
The membrane pre-treatment and permeation with
various solvents suggests that the permeability of the polymer depends not only on the pore size, but also on the
interaction of solvent and polymer. The lowest uxes and
permeabilities were obtained with iso-propanol permeations. On the other hand, ethanol uxes were high for
UP005 (ethanol uxes were higher than water uxes),
UP010 and UH050. Hence, the tested membranes present
potential for use with non-aqueous solvents with relatively
brief pre-treatments.
The study of the inuence of exposure time and solvent
on pre-treatment of polymeric commercial ultraltration
membranes shows the possibility of increasing permeate
uxes through the increase of the immersion times in a
combination of different solvents. Each membrane exhibits
better permeability results with a different pre-treatment,
which are, only in some cases, corroborated by the increase
in contact angles. The results evidenced that the ux
increase is strongly inuenced by the time of exposure to
ethanol than by the time of exposure to n-hexane. It can
be stated that the pre-treatment increased the contact
angle for all the membranes tested, consequently increasing hydrophobicity. Moreover, characterisation showed
no substantial changes in membrane structure
morphology.
Furthermore, given the correlation of permeability and
contact angles only in a few cases, it is believed that other
variables affect the conditioning efciency, such as polymeric constitution, molar mass cut-off, and the interactions between solvent and membrane. Therefore, the
efciency of the process would be a balance of all these
variables.
Acknowledgements
Authors are grateful to CAPES and CNPq for nancial
support, Central Laboratory for Electronic Microscopy
References
[1] Ochoa NA, Pagliero C, Marchese J, Mattea M. Ultraltration of
vegetable oils. Degumming by polymeric membranes. Sep Purif
Technol 2001:41722.
[2] Rafe A, Razavi SMA. Water and hexane permeate ux through UF
polysulfone amide membrane. Desalination 2009;236:3945.
[3] Darvishmanesh S, Degrve J, Van Der Bruggen B. Comparison of
pressure driven transport of ethanol/n-hexane mixtures through
dense and microporous membranes. Chem Eng Sci 2009;64:
391427.
[4] Koseoglu SS, Engelgau DE. Membrane applications and research in
edible oil industry: an assessment. J Am Oil Chem Soc
1990;67(1990):23949.
[5] Van Der Bruggen B, Geens J, Vandecasteele C. Fluxes and rejections
for nanoltration with solvent stable polymeric membranes in
water, ethanol and n-hexane. Chem Eng Sci 2002;57:25118.
[6] Tsui EM, Cheryan M. Characteristics of nanoltration membranes in
aqueous ethanol. J Membr Sci 2004;237:619.
[7] Araki MS, Coutinho CM, Gonalves LAG, Viotto LA. Solvent
permeability in commercial ultraltration polymeric membranes
and evaluation of the structural and chemical stability towards
hexane. Sep Purif Technol 2010;71:1321.
[8] Coutinho CM, Chiu MC, Basso RC, Ribeiro APB, Gonalves LAG, Viotto
LA. State of art of the application of membrane technology to
vegetable oils: a review. Food Res Int 2009;42:53650.
[9] Yang XJ, Livingston AG, Santos LF. Experimental observations of
nanoltration with organic solvents. J Membr Sci 2001;190:4555.
[10] Firman LR, Ochoa NA, Marchese J, Pagliero CL. Deacidication and
solvent recovery of soybean oil by nanoltration membranes. J
Membr Sci 2013;431:18796.
[11] Darvishmanesh S, Degrve J, Van der Bruggen B. Mechanisms of
solute rejection in solvent resistant nanoltration: the effect of
solvent on solute rejection. Phys Chem Chem Phys 2010;12:
1333342.
[12] Sarmento LAV, Spricigo CB, Petrus JCC, Carlson LHC, Machado RAF.
Performance of reverse osmosis membranes in the separation of
supercritical CO2 and essential oils. J Membr Sci 2004;237:716.
[13] Dijkstra MFJ, Bach S, Ebert KA. Transport model for organophilic
nanoltration. J Membr Sci 2006;286:608.
[14] Darvishmanesh S, Degrve J, Van Der Bruggen B. Performance of
solvent-pretreated polyimide nanoltration membranes for
separation of dissolved dyes from toluene. Ind Eng Chem Res
2010;49:93308.
[15] Othman R, Mohammad AW, Ismail M, Salimon J. Application of
polymeric solvent resistant nanoltration membranes for biodiesel
production. J Membr Sci 2010;348:28797.
[16] Van der Bruggen B, Geens J, Vandecasteele C. Inuence of organic
solvents on the performance of polymeric nanoltration
membranes. Sep Sci Technol 2002;37(4):78397.
[17] Roudman AR, DiGiano FA. Surface energy of experimental and
commercial nanoltration membranes: effects of wetting and
natural organic matter fouling. J Membr Sci 2000;175:6173.
[18] Flory P. Principles of polymer chemistry. Ithaca, New York: Cornell
Univ. Press; 1953. p. 688.
[19] Ebert K, Cuperus FP. Solvent resistant nanoltration membranes in
edible oil processing. Membr Technol 1999;107:58.
[20] Visser, T. Mixed gas plasticization phenomena in asymmetric
membranes. PhD thesis, (2006) University of Twente, The
Netherlands. <http://purl.utwente.nl/publications/57627>.
[21] Buonomennaa MG, Golemmea G, Jansenb JC, Choib SH. Asymmetric
PEEKWC membranes for treatment of organic solvent solutions. J
Membr Sci 2011;368:1449.
[22] Jansen JC, Darvishmanesh S, Tasselli F, Bazzarelli F, Bernardo P, Tocci
E, et al. Inuence of the blend composition on the properties
and separation performance of novel solvent resistant
polyphenylsulfone/polyimide nanoltration membranes. J Membr
Sci 2013;447:10718.
[23] Geens J, Van der Bruggen B, Vandecasteele C. Characterisation of the
solvent stability of polymeric nanoltration membranes by
measurement of contact angles and swelling. Chem Eng Sci
2004;59:11614.
501