Separation and Purification Technology 66 (2009) 347–352

Contents lists available at ScienceDirect

Separation and Purification Technology

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

Application of the Al2 O3 –PVDF nanocomposite tubular ultrafiltration (UF)

membrane for oily wastewater treatment and its antifouling research
Lu Yan a,b,∗ , Sun Hong a , Meng Li Li a , Yu Shui Li b
a
Department of Chemistry, Daqing Petroleum Institute, Daqing 163318, China
b
Department of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China

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

Article history: Nanocomposite tubular ultrafiltration (UF) membrane Al2 O3 –PVDF was prepared by the phase-inversion
Received 2 September 2008 method. Oily wastewater of Daqing oilfield was treated by PVDF and Al2 O3 –PVDF tubular ultrafiltra-
Received in revised form 2 December 2008 tion membranes, respectively. The membrane water permeations of the UF process were analyzed.
Accepted 17 December 2008
Retentions of chemical oxygen demand (COD) and total organic carbon (TOC) were more than 90%
and 98%, respectively. The results indicate that after UF treatment, oil content was below 1 mg/L, sus-
Keywords:
pended solids content was below 1 mg/L, and solid particle median diameters were less than 2 ␮m. The
Nanocomposite membrane
quality of the permeation water met the requirement by oilfield injection or drainage. Fouled mem-
Oily wastewater
Ultrafiltration
branes and washed membranes were analyzed by scanning electron microscopy (SEM) and atomic force
Tubular membrane microscopy (AFM), and fouled membranes were backwashed with different solutions. Results show that
Membrane fouling the addition of nano-sized alumina particles improved membrane antifouling performance, and the flux
recovery ratio of modified membranes reached 100% washing with 1 wt% of OP-10 surfactant solution
(pH 10).
Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved.

1. Introduction subjected to, limits the industrialization of the membrane technol-

ogy.
Oily wastewater has been generated from the development of Membrane fouling is affected by the surface hydrophilicity,
oil fields, especially for those using water injection technologies. i.e., improving hydrophilicity of the membrane can reduce mem-
It is necessary to purify this water so that it can be reused to brane fouling to some extent. Therefore, different methods to
save water resources and to protect the environment. Conventional hydrophilize the membrane surfaces have been investigated, such
oily wastewater treatment methods include gravity separation and as blending hydrophilic polymer with hydrophobic polymer, graft-
skimming, dissolved air flotation, de-emulsification, coagulation ing hydrophilic branches on hydrophobic polymer backbone and
and flocculation, which have several disadvantages such as low the deposition of hydrophilic films on hydrophobic materials.
efficiency, high operation costs, corrosion and recontamination Hydrophilic polymers that can improve hydrophility of hydropho-
problems [1,2]. Also, these methods are not effective in removing bic materials include polymethyl methacrylate (PMMA), poly
smaller oil droplets and emulsions. All of these disadvantages have acrylate (PMA), polyvinyl acetate (PVAC), and cellulose acetate (CA),
promoted the development of new processes for oily wastewater etc.
treatment. In addition to the blending of organic polymers, recent studies
Membrane filtration is playing a more prominent role in the have focused on blending inorganic and organic materials. It has
treatment of oily wastewater due to its advantages: no chemi- been demonstrated that the addition of inorganic filler has led to
cal additives are needed to break the emulsion, high chemical an increase in membrane permeability and a better control of mem-
oxygen demand (COD) removal efficiencies are achieved, and treat- brane surface properties [12–16]. Few studies were reported on
ment facilities are quite compact and fully automated [3]. Many treating oily wastewater with organic–inorganic composite mem-
studies have been done on the oily wastewater treatment with dif- branes. In the present study, a tubular ultrafiltration (UF) module
ferent membranes [4–11]. However, membrane fouling, which all equipped with PVDF membranes modified by inorganic nano-sized
membrane processes—especially for oily wastewater treatment, are alumina particles was used to purify oily wastewater from am oil
field. Cross-flow UF was used as the process. The performance of the
membranes was characterized by their flux and antifouling proper-
∗ Corresponding author. Tel.: +86 459 6504236. ties. The quality of permeation water was evaluated by oil content
E-mail address: [email protected] (L. Yan). and values of COD, SS, and total organic carbon (TOC).

1383-5866/$ – see front matter. Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2008.12.015
348 L. Yan et al. / Separation and Purification Technology 66 (2009) 347–352

Fig. 1. The expermental set-up of cross-flow UF process.

Table 1
Water quality of experiment.

Oil content (mg/L) SS (mg/L) pH Turbidity NTU COD (mg/L) TOC (mg/L) Median diameter (␮m) Temperature (◦ C)

15.5 15.8 8.4 98 634 220 2.035 39.5

Fig. 2. Membrane fluxes changing with the filtration time.

Fig. 4. The result of turbidity removed by UF.

2. Experiments
Al2 O3 –PVDF tubular membranes were prepared by the phase-
2.1. Membrane production
inversion method. A casting solution was prepared by dissolving the
PVDF (19%, by weight of the solution) in the solvent at room tem-
The PVDF used was a commercial product (FR904). Dimethylac-
perature and adding nano-sized Al2 O3 particles (2%, by weight of
etamide (DMAC, >99%, reagent) was used as the solvent. Alumina
PVDF) and other additives (1% sodium hexaphosphate, by weight
nano-sized particles (10 nm) were added to the PVDF solutions.
of PVDF; 3% PVP, by weight of the solution) to the casting dopes
Other additives were sodium hexaphosphate and polyvinylpyrroli-
while stirring. In order to obtain optimal dispersions of the parti-
done (PVP). A mixture of distilled water and ethanol was used as
cles in the polymer solutions, agitation was required for at least
the non-solvent for the polymer precipitation.
24 h. The casting solutions were then kept in the dark for at least

Fig. 3. The result of oil removed by UF. Fig. 5. The result of COD removed by UF.
 L. Yan et al. / Separation and Purification Technology 66 (2009) 347–352 349

24 h to remove air bubbles. Inner pressure tubular membranes were

prepared by using an epoxy resin porous tube as the supporter.
An unmodified PVDF membrane was prepared in order to compare
with Al2 O3 –PVDF membranes.

2.2. Cross-flow ultrafiltration

The experimental set-up used in the cross-flow UF investigation

is presented in Fig. 1. The tubular cross-flow membrane had a fil-
ter area of 0.01256 m2 , 35 kDa molecule weight cut-off (measured
by the protein curve method), and a 5 mm tube in diameter and
80 cm in length. In the process of experiment, membrane fluxes
were calculated as follows:
Q
Fig. 6. The result of TOC removed by UF. J=
At

where J is the permeation flux (L/m2 h); Q is total permeation vol-

ume (L); A is effective membrane areas (m2 ); and t is filtration time
(h).
Rejections for various solutes are defined as R = 1 − (CP /C0 ),
where Cp is the concentration of solute in the permeation and C0 is
the concentration of solute in the feed solution. Both values are
obtained using UV spectrophotometer at different wavelengths:
280 and 225 nm for protein solutions and oily water, respectively.
Treatment was carried out with oil field wastewater as feed
stream. The experiment took oily wastewater from the Daqing oil
field that had been treated by the sedimentation and sand filtra-
tion (twice). Permeating fluxes changing with time were studied
under the operating conditions: transmembrane pressure (TMP)
(0.1 MPa); temperature (T = 30 ◦ C); feed cross-flow velocity (v =
Fig. 7. The result of SS removed by UF.
7.8 m/s). Feed stream properties are listed in Table 1.

Fig. 8. SEM micrographs of fouled membrane (left) and new membrane (right) for PVDF-0.

Fig. 9. SEM micrographs of fouled membrane (left) and new membrane (right) for PVDF-2.
350 L. Yan et al. / Separation and Purification Technology 66 (2009) 347–352

2.3. Water quality analyses

Water quality was assessed by testing the feed and permeation

water of modified and unmodified membranes. COD values were
obtained by the conventional method of heavy potassium chro-
mate. The concentration of TOC was measured using a TOC analyzer
(TOC-VCPN, Japan). Turbidity was obtained by AQ2010 nephelome-
ter. SS values were measured by weight method: 100 mL of water
were put through a 0.45 ␮m microfilter; and the retentate was dried
under temperatures of 103–105 ◦ C until the retention weight did
not change. Suspended particulars diameters were measured by
the mode 19703, liquid particle counting system (American, HIAC).
Oil content was analyzed with a UV spectrophotometer (T6-New
Century, China).

2.4. Fouled membranes analyses

Fouled membranes and washed membranes were observed

through SEM (S-4700, Japan) and AFM (BioScopeTM , American).
Contaminants were analyzed with the uniting of gas chromatogra-
phy and mass spectrum (GC-6890N/MS-5973N, American). Fouled
membranes by oily wastewater were cleaned with clean water, lye,
acidity and surfactant solutions, respectively, so as to investigate
flux recoveries of membranes after backwash.

3. Results and discussion

3.1. Membranes permeability

Wastewater experiments with the effluent from an oil field con-

taining 15.5 mg/L of oil were conducted at 30 ◦ C, 0.1 MPa TMP and
7.8 m/s flow velocity in the set-up shown in Fig. 1. The UF process
ran for 10 h continually. Fig. 2 shows the change in membrane fluxes
with filtration time. At the beginning, permeation fluxes declined
gradually until an invariable flux value was obtained. In the pro-
cess of UF, the fluxes of modified membranes PVDF-2 were always
higher than those of unmodified membranes PVDF-0, which was
also attributed to the membrane hydrophilicity improvement due
to the addition of nano-sized alumina particles into the PVDF. Stable
flux of the PVDF-2 membrane is 138.53 L/m2 h, which is 2.07 times
than that of the PVDF-0 membrane, 66.88 L/m2 h. Flux descend-
ing ratio of the PVDF-2 membrane is only 31.16% and it is 48.97%
for the PVDF-0 membrane. All these indicate that the permeability
and antifouling performances of modified membrane are improved
significantly.

3.2. Water quality analysis

In order to investigate the water quality after the UF process,

feed and permeation water were analyzed, respectively. Results are
shown in Table 1 and Figs. 3–7, respectively.
The data of figures illustrate that the initial permeation water
qualities of the effluents were not good. It was because that the
membranes were new and there was no gelatin on the membranes
surface at the beginning of filtration. Thus, some contaminants Fig. 10. AFM micrographs of fouled membranes.
in the wastewater can get across the bigger membranes pores to
the permeation water at the pressures driving. With the devel-
opment of filtration process, gelatin was formed on the surface and turbidity retention is more than 90% when the filtration get
of membranes due to contaminant’s concentrating and polarizing, to stable state. It also illustrates that the flux improvement has no
which became a dominant factor of the fluxes. The gelatin layer can influence on the retention performance, but enhanced some prop-
prevent contaminants particles from entering membranes pores. erties of the modified membrane. The TOC retention improvement
Although, membranes fluxes decline, the permeation water quali- of the permeation water was outstanding. It can be explained that
ties were meliorating. the addition of hydrophilic Al2 O3 particles in the modified mem-
Figs. 3–7 show that after UF treatment process, the oil content brane made it head off more organic contaminants effectively. As a
is below 0.7 mg/L, suspended solids content is below 0.5 mg/L .The result, the amount of organic contaminants which passed through
TOC retention is more than 95%, and retention of COD is 80–90% the membrane with the permeation water was reduced. The TOC
 L. Yan et al. / Separation and Purification Technology 66 (2009) 347–352 351

Fig. 11. Gas chromatogram and mass spectra for the modified membrane contaminants.

Fig. 12. Gas chromatogram and mass spectra for the unmodified membrane contaminants.

value of the permeate water was decreased. The experimental roughness due to nanometer particles addition. Contaminants par-
results indicate that the quality of the permeation water met the ticles adsorbed on the PVDF-2 fouled membrane surface are smaller
requirement for oil field injection or drainage. Thus, it is clear that than that of PVDF-0 fouled membrane and contaminants also accu-
both membranes are very available in treating oily wastewater. If mulate locally. Comparing two fouled membranes, contaminated
the pollution problem of membrane technology is conquered, this layer is more roughness on the unmodified membrane surface. It
technology will be used more extensively. can be explained as follows: on one hand, membrane’s structure
characteristics is different; on the other hand, the improvement of
3.3. Fouled membranes and contaminants analyses the modified membrane’s hydrophilicity leads to different kinds
and amounts contaminants adsorption on the membrane surface,
3.3.1. SEM analyses which make different structure for contaminants layer.
SEM images of fouled membranes and new membranes are
shown in Figs. 8 and 9. The PVDF-0 new membrane surface is slick
and there is no any impurity on it. However, many contaminants 3.3.2. AFM analyses
particles are adsorbed on the surface of fouled membrane. Agglom- Fouled membranes AFM images and their corresponding cross
erate contaminants (indicated by arrow signs) can be found in the section spectrograms are shown in Fig. 10. From Fig. 10 we can see
local of membrane surface. The PVDF-2 new membrane surface is that a part of membrane pores were jammed by contaminants.

Fig. 13. SEM images of the backwashing membranes.

352 L. Yan et al. / Separation and Purification Technology 66 (2009) 347–352

Table 2 indicates that the hydrophilicity improvement of modified mem-

Permeate flux recovery for washed membranes.
branes reduced its adsorption for contaminants and improved its
Cleaning solution Permeate flux recovery ratio (%) antifouling performance [17].
Unmodified membrane Modified membrane Oily wastewater UF ran for 90 min, and membranes were back-
washed for 10 min with a cleaning solution. Flux recovery ratios
Clean water 88.5 91.5
1% OP-10 95 96.29
after washing with different kinds of cleaning solution are listed in
1% OP-10 solution (pH 10) 97.67 100 Table 2. Flux recovery ratios, defined as permeate flux after wash-
2% Sodium dodecylbenzenesulfate 93.67 96.9 ing divided by initial flux, washed with clean water were about
2% Sodium hydroxide 89.67 95.68 90% for modified and unmodified membranes, which means that
most membrane fouling formed in the process of oily wastewater
treatment is reversible fouling caused by the colloidal layer. Flux
Membranes cross section spectrograms incarnate membrane
recoveries were better for the membrane washing with high pH
porous structure of the cross section. Distance between two apices
values alkali solution. The modified membrane flux recovery ratio
denotes a size of the membrane pore and swing in vertical shows
reached 100% after washing with 1% content of OP-10 surfactant
a deepness of the pore. The apertures and deepness of fouled
solution (pH 10).
membranes were decreased due to contaminants existing. Coor-
dinate data of membranes cross section spectrograms show that
the contaminants thickness adsorbed on the modified membranes 4. Conclusion
surface was less than that of unmodified one. It illuminates that
the antifouling performance of the modified membrane has been Modified membrane permeation performance was improved
improved. significantly without sacrificing its retention properties, and its flux
was about twice that of unmodified one. The flux recovery ratio
3.3.3. Contaminants analyses of modified membrane reached 100% washing with 1 wt% OP-10
Figs. 11 and 12 are gas chromatography and mass spectrum anal- surfactant solution (pH 10). The quality of the permeation water
yses of contaminants for modified and unmodified membranes. from oily wastewater treated by the modified membrane met the
About 44 kinds of contaminations were adsorbed on the mem- requirement for oil filed injection or drainage. It indicated that the
brane PVDF-2, and about 68 kinds on the membrane PVDF-0. Al2 O3 –PVDF nanocomposite membranes have a favorable antifoul-
From adsorption peak areas, modified membrane PVDF-2 was ing performance, with higher flux recovery and less contaminants
less 21.65% than unmodified membrane PVDF-0, which indicates adsorption, and have a wide application in oily wastewater treat-
that amounts of contaminants adsorbed on the modified mem- ment.
brane were reduced. Fouled membranes SEM and AFM analyses
can also confirm this conclusion in qualitative. All these owed to Acknowledgements
hydrophilicity improvement of the modified membrane.
The paper was supported by the National postdoctoral fund and
3.4. Washing of fouled membranes the 863 National Basic Research Program 2008AA06Z304, China.

There are two types of membrane fouling for oily wastewater References
treatment. One is reversible fouling, which is typically caused by
[1] Wang Shengchun, Wen Jianzhi, Wang Hai, et al., J. Membr. Sci. 18 (1998)
deposited solute or colloidal particles on the surface and in the 28–32.
pores. Flux decline caused by reversible fouling can be easily recov- [2] K. Karakulski, A. Kozlowski, A.W. Morawski, J. Sep. Technol. 5 (1995) 197–205.
ered with pure water backwash. The other is irreversible fouling, [3] J. Kong, K. Li, J. Sep. Purif. Technol. 16 (1999) 83–93.
[4] N.M. Kocherginsky, C.L. Tan, W.F. Lu, J. Membr. Sci. 220 (2003) 117–128.
which leads to flux decline due to strong physical or chemical sorp- [5] M. Gryta, K. Karakulski, A.W. Morawski, J. Water Res. 15 (2005) 3665–3669.
tion of solutes and particles on the surface or in the membrane [6] W. Scholz, W. Fuchs, J. Water Res. 14 (2000) 3621–3629.
pores. Flux decline caused by irreversible fouling can be recovered [7] In-Soung Chang, Chang-Mo Chung, Seung-Ho Han, J. Desalination 133 (2001)
225–232.
only by washing with acid or alkali solutions. However, often the [8] T. Mohammadi, M. Kazemimoghadam, M. Saadabadi, J. Desalination 157 (2003)
initial permeability of irreversibly fouled membranes cannot be 369–375.
recovered even with aggressive cleaning methods. [9] K. Scott, R.J. Jachuck, D. Hall, J. Sep. Purif. Technol. 22–23 (2001) 431–441.
[10] W. Scholz, W. Fuchsm, J. Water Res. 34 (2000) 3621–3629.
In this study, flux recoveries of two membranes were good
[11] Lu Xiaolong, Liu Xinmin, Hu Chengsong, et al., J. Membr. Sci. 2 (1998) 33–36.
after backwashing with clean water, and modified membrane’s flux [12] I. Genne, S. Kuypers, R. Leysen, J. Membr. Sci. 113 (1996) 343–350.
recovery was higher slightly than that of unmodified one (Table 2). [13] P. Aerts, A.R. Greenberg, R. Leysen, et al., J. Sep. Purif. Technol. 22–23 (2001)
663–669.
This indicates that most membrane fouling is reversible fouling,
[14] A. Bottino, G. Capannelli, V. D’Asti, et al., J. Sep. Purif. Technol. 22–23 (2001)
which can be overcome by mild cleaning. Washed membrane SEM 269–275.
images (Fig. 13) show that the contaminant residues (indicated [15] A. Bottino, G. Capannelli, A. Comite, J. Desalination 146 (2002) 35–40.
by circles) on the unmodified membrane are more visible than [16] Dar-Jong Lin, Cheng-Liang Chang, Fane-Ming Huang, et al., J. Polym. 44 (2003)
413–422.
the modified membrane after backwashing with clean water. This [17] Lu Yan, Yu Shui Li, Chai Bao Xiang, J. Membr. Sci. 276 (2006) 162–167.
coincides with the flux recoveries for washed membranes. This