Professional Documents
Culture Documents
PII:
DOI:
Reference:
www.elsevier.com/locate/memsci
S0376-7388(15)30046-6
http://dx.doi.org/10.1016/j.memsci.2015.07.018
MEMSCI13837
Unravelling
flux
behavior
of
superhydrophilic
loose
Heverlee, Belgium
b
Graphical abstract:
1. Introduction
Sustainable wastewater treatment calls for a new paradigm shift to emphasize on
resource recovery from wastewater in addition to contaminants removal [1, 2]. The
development of advanced technology and practices potentially enables cost-effective
resource recovery, including water, nutrients, bioplastics and energy from the
industrial wastewater [1]. For example, in textile wastewater, a high amount of
organic matters (i.e., dyestuffs) can be applied for energy production or recovered for
further dyeing [3-6]. Furthermore, a high salinity (~6.0% NaCl or ~5.6% Na2SO4) in
the textile wastewater endows its potential reuse as a draw solution for forward
osmosis/pressure retard osmosis or an agent for base/acid generation by bipolar
membrane electrodialysis, expediting the close of water and material loop [7, 8].
However, direct treatment and reuse of textile wastewater can cause the severe fouling
of forward osmosis membrane or ion exchange membrane due to dye adsorption, thus
requiring the effective fractionation of dye and salt in textile wastewater [9, 10].
Nanofiltration (NF)
200-1000 Da and pore size of ~0.5 to 2.0 nm, exhibit one of the most competitive
separation and purification processes in textile wastewater treatment [11]. The unique
separation mechanism of NF membranes, which involves size exclusion and
electrostatic repulsion, allows for its wide applications in textile wastewater treatment,
endowing high selectivity of dye molecules as well as partial permeation for inorganic
salt solutes [7].
The performance of NF membranes can be severely limited by membrane fouling and
concentration polarization [12-18]. Generally, in textile wastewater treatment,
membrane fouling caused by cake layer formation and pore blocking plays a
dominant role for the permeation flux decline [19-21]. To optimize the operational
conditions of NF membranes for dye/salt separation, Koyuncu et al. intensively
investigated the operation factors involving membrane flux decline in textile
wastewater, indicating that the increasing cross-flow velocity can be a pronounced
factor to improve the membrane flux by alleviating the membrane fouling and
concentration polarization [22]. Furthermore, the adsorption and accumulation of dyes
in the membrane pore structure, mainly resulting from hydrophobic interaction or
electrostatic attraction, gives rise to the high risk of irreversible fouling, which in turn
induces a dramatic flux decline [23]. Inevitably, this requires an increase in frequency
of chemical cleaning, which may shorten the membrane lifespan. Apart from dye
adsorption and accumulation, Van der Bruggen et al. demonstrated that the high
salinity in the textile wastewater has a negative impact on the membrane flux, related
to the high osmotic pressure caused by high salt rejection of commercially available
4
In this work, the flux of the superhydrophilic loose NF membranes (Sepro NF 6 and
NF 2A, Ultura) in a single-component (i.e., NaCl or dyes) solution was studied at
variable pressures and concentrations. Furthermore, the flux decline of these NF
membranes was systematically investigated for simulated textile wastewater (i.e.,
dye/salt mixture). In addition, interactions between salt, dyes and the membrane
surface were related to flux decline by characterizing the membrane surface after
fouling. The combination of diafiltration and post-concentration for textile wastewater
was performed to understand the membrane flux behavior in the overall treatment
process, in view of assessing an industrial application.
Two thin film composite NF membranes with different salt rejections were kindly
supplied by Ultura (USA). The properties of the tested NF membranes are shown in
Table 1.
Sepro NF 2A
Sepro NF 6
Semi-aromatic
polyamide
Semi-aromatic
polyamide
Contact angle ()
21.7 1.4
14.3 0.9
Permeability (Lm2h1bar1) at
25 C
10.1
16.7
529
847
21.2
7.3
3.78
5.06
Process pH limitation
3.0-10.0
3.0-10.0
measured at 6 bar; the pure water flux was recorded to determinate the flux recovery
ratio ():
J w,c
J w,0
100
(1)
where Jw,0 is the pure water flux of original NF membranes, and Jw,c is the pure water
flux of NF membranes after cleanning.
(2)
where P denotes the transmembrane pressure, represents the osmotic pressure
difference between bulk and permeate solution, is the solvent viscosity, Rm is the
intrinsic membrane resistance, Rf is related to the hydraulic resistance caused by
fouling, such as cake layer formation and pore blocking, and Rcp is the resistance
originating from the concentration polarization of salts.
For a multi-component solution, the osmotic pressure difference () is calculated by
superimposing the contribution from each individual component [33]:
(3)
For the dilute solution, the osmotic pressure can be estimated by the Vant Hoff
equation:
(4)
where Ci is the molarity of solute, Rg is the universal gas constant, and T is the
thermodynamic (absolute) temperature.
In Equation 2, the intrisnic membrane resistance (Rm) was calculated from the flux of
pure water, the value of Rf was estimated from independent filtration tests using
individual dye solution, and the resistance caused by the concentration polarization of
10
NaCl (Rcp) was determined by the independent filtration of pure NaCl solutions.
salt permeation [7, 27, 34]. However, a further investigation of the flux behavior is
essential because of the complexity of the dye/salt system, with enhanced interaction
among salt, dye, and membrane surface. As a reference, the flux behavior was studied
in single-component solutions. Fig. 2 shows the flux of Sepro NF membranes in the
NaCl solutions at variable salt concentrations and operation pressures.
(A) 200
(B) 200
Mea.:
Cal.:
180
2 g/L
2 g/L
5 g/L
5 g/L
10 g/L
10 g/L
20 g/L
20 g/L
40 g/L
40 g/L
Pure water
160
140
140
Mea.:
Cal.:
2 g/L
2 g/L
5 g/L
5 g/L
10 g/L
10 g/L
20 g/L
20 g/L
40 g/L
40 g/L
Pure water
Flux, Lm h
-1
160
-1
120
-2
-2
Flux, Lm h
180
100
80
120
100
80
60
60
40
40
20
20
0
0
10
10
Fig. 2 Flux of Sepro NF membranes in NaCl solution filtration. (A): Sepro NF 6; (B):
Sepro NF 2A (Dot: experimental data; Solid line: theoretical flux calculated without
taking concentration polarization into account)
As illustrated in Fig. 2 and Supplementary Fig. S1, the flux of Sepro NF membranes
in the NaCl solutions was positively related with the operation pressures. However,
the flux of NF membranes decreased with the salt concentration at any fixed pressure.
For instance, at 10 bar, the flux of Sepro NF 6 and NF 2A dropped by 22.8% and 53.3%
in the whole range of salt concentration, respectively. This is due to the following
factors:
(1) Firstly, the viscosity of salt solution. As shown in Supplementary Fig. S2, the
12
viscosity of NaCl solutions follows the Jone-Dole equation, and increases with
the salt concentration [35]. Indeed, the solution viscosity increased by 6.38%
when NaCl concentration increased from 0 to 40 gL-1, which will negatively
impact on membrane permeate flux.
(2) Secondly, concentration polarization. In Fig. 2, theoretical flux values have
been calculated taking into consideration of the change in solution viscosity as
well as osmotic pressure difference across the membrane (see Supplementary
Fig. S3). However, the calculation did not take account of concentration
polarization effect. The difference between the theoretical and experimental
results thus revealed the role of concentration polarization. Generally, the
concentration polarization resistance can be negligible at low salt
concentrations. As shown in Fig. 3, Sepro NF 6 had an obvious effect of
concentration polarization at salt concentrations ranging from 20 to 40 gL-1,
whereas Sepro NF 2A experienced a larger impact of concentration
polarization in NaCl solutions from 5 to 40 gL-1. Furthermore, the additional
hydraulic resistance caused by concentration polarization was significantly
greater for Sepro NF 2A. The greater concentration polarization effect for
Sepro NF 2A can be explained by its relatively higher salt rejection compared
to that of Sepro NF 6 (Table 1). Once again, this reveals the advantage of
highly salt-permeable membranes. The low rejection nature of such
membranes allows solutes to easily pass through them, which reduces the
osmotic pressure difference across the membrane and the concentration
13
polarization of salts.
Rm=2.3908E13 m
-1
4 bar
6 bar
8 bar
10 bar
0.3
(B)
Rm=4.1762E13 m
-1
2.5
13
13
0.4
0.2
0.1
3.0
-1
(A)
-1
m )
0.5
0.0
2.0
4 bar
6 bar
8 bar
10 bar
1.5
1.0
0.5
0.0
0.1
0.1
10
10
14
0.05 g/L
1.0 g/L
(B)
0.2 g/L
2.0 g/L
-2
100
80
60
40
20
Congo red
80
60
40
20
Direct red 23
140
0
-1
120
-2
100
Flux, Lm h
-2 -1
0.2 g/L
2.0 g/L
60
40
20
Congo red
-1
-2
100
Flux, Lm h
Flux, Lm h
-2 -1
120
Flux, Lm h
0.05 g/L
1.0 g/L
0
80
140
0
80
60
40
20
0
Water
0.5 g/L
80
-1
Water
0.5 g/L
120
Flux, Lm h
Flux, Lm h
-2 -1
(A) 140
Direct red 80
0
60
40
20
Direct red 23
0
80
60
40
20
Direct red 80
0
Fig. 4 Membrane flux of pure dye solutions for NF membrane at viable operation
pressures and dye concentrations. (A): Sepro NF 6; (B): Sepro NF 2A
The permeability of Sepro NF membranes in the dye solutions showed no obvious
decline at a low pressure (linear relationship between the flux and pressure). However,
as the operation pressure increased, the membrane flux deviated from the linear
relationship, possibly due to the severe compaction of the dye cake layer which
increased the fouling resistance (Rf). In some severe fouling cases (2 gL-1 direct red
23 for both NF membranes), increasing pressure above the threshold value did not
result in enhanced permeability. In the current study, the threshold pressure of Sepro
NF membranes in dye filtration lied at 6 bar. Similar type of limiting flux behavior
has been previously reported for surfactants [37] and macromolecules [38]. Indeed, a
limiting flux value of ~ 40 Lm-2h-1 was observed for both Sepro NF 6 and NF 2A at
6 bar. This type of membrane independence suggests that the threshold flux value was
likely governed by foulant-foulant interaction instead of foulant-membrane interaction
15
[38].
In order to obtain an explanation for the mechanisms of membrane fouling, the
fouling resistance of Sepro NF membranes for these three dyes was calculated, as
shown in Fig. 5. In general, Sepro NF 2A appeared to have a stronger antifouling
property than Sepro NF 6, as indicated by its relatively lower hydraulic resistance
caused by membrane fouling (Rf). At low dye concentrations, Rf for Sepro NF 2A was
relatively small compared to its intrinsic membrane resistance. In contrast, Sepro NF
6 had higher values of Rf despite of its more hydrophilic membrane surface (Table 1).
A plausible explanation may be its relatively large pore size allows dye molecules to
block its membrane pores more easily, leading to greater permeability drop. The
presence of larger nodules on the surface of Sepro NF 6 can also possibly facilitate
the deposit of dye molecules on the valley regions, resulting in a higher hydraulic
resistance of Sepro NF 6 [31]. Furthermore, Sepro NF 6 is less negatively charged
compared to Sepro NF 2A, which may allow the negatively charged dye molecules
(due to their -SO3- groups) to approach Sepro NF 6 more easily. At very high dye
concentrations (e.g., 2.0 gL-1 Congo red or direct red 23), similar Rf values were
observed. In the latter case, fouling was more likely dominated by cake layer
formation (e.g., see SEM images of Sepro NF membranes fouled by 2000 ppm direct
red 23 solution in Fig. 6), where foulant-foulant interaction prevails over
foulant-membrane interaction [39]. Specifically, Sepro NF 2A had a higher calculated
hydraulic resistance in the direct red 23 solution, compared to Sepro NF 6. This is
mainly due to the complexity of the direct red 23 solution (~70% impurity, including
16
5.8% Cl-, 12.6% SO42-, 5.2% HCO3- and other intermediate during dye synthesis),
diminishing the flux of Sepro NF 2A through a higher salt retention.
0.05 g/L
0.2 g/L
0.5 g/L
1.0 g/L
2.0 g/L
(A) 1.4
1.2
1.0
0.8
0.8
0.6
0.6
0.4
0.4
Direct red 23
0.0
5
-1
Rm=2.444E13 m
-1
13
0.2
Congo red
Rf, 10 m
Rf, 10 m
-1
0.0
5
3
2
1
Congo red
Rm=4.0078E13 m
-1
Direct red 23
3
2
1
0.9
0
0.9
0
0.6
0.6
0.3
0.05 g/L
0.2 g/L
0.5 g/L
1.0 g/L
2.0 g/L
1.2
1.0
0.2
13
(B) 1.4
Direct red 80
0.3
Direct red 80
0.0
0.0
2
Fig. 6 SEM images of Sepro NF membranes fouled by 2000 ppm direct red 23. (A):
Sepro NF 6; (B): Sepro NF 2A
In order to clarify the mechanism of the flux decline for Sepro NF membranes in dye
solutions, digital pictures of NF membranes cleaned by rinsing with water and the
17
FTIR measurements are shown in Figs. 7 and 8, respectively. As shown in Fig. 7, the
fouled Sepro NF 6 was less effectively cleaned after water rinsing, with a small
amount of dye deposited on the membrane, indicating that the reversible membrane
fouling may control the flux behavior. By contrast, the surface of Sepro NF 2A
appeared to be better cleaned, once again confirming its better antifouling ability.
Furthermore, the FTIR spectra in Fig. 8 indicate no obvious alteration in the observed
functional groups, from which it can be concluded the dyes had no chemical reaction
with the surface of Sepro NF membranes. This is in direct contrast to some earlier
studies claiming that chemical interactions (e.g., covalent bonds) between dye and the
surface of polyamide membrane occurs [40].
Fig. 7 Digital pictures of fouled NF membranes after water rinsing. (A): Sepro NF 6;
(B): Sepro NF 2A
Fouled by direct red 23
Fouled by direct red 80
Fouled by congo red
Original
Absorpiton
1800
(B)
Absorpiton
(A)
1600
1400
1200
Wavenumber (cm-1)
1000
800
1800
1600
1400
1200
1000
800
Wavenumber (cm-1)
Fig. 8 FTIR spectra for fouled NF membranes. (A): Sepro NF 6; (B): Sepro NF 2A
18
(A)
100
80
40
-2
70
(B)
50
Flux, Lm h
-2
Flux, Lm h
-1
90
60
-1
120
60
50
40
30
20
30
20
10
10
0
0
0
10
20
30
Concentration of NaCl, gL
40
-1
10
20
30
Concentration of NaCl, gL
-1
40
Fig. 9 Flux of Sepro NF membranes in the dye/salt binary mixtures at 6 bar. (A):
Sepro NF 6; (B): Sepro NF 2A
As shown in Fig. 9, the flux of membrane in the dye/salt mixture decreases with the
salt concentration. This is due to the increase in osmotic pressure difference between
the feed and permeate side of NF membrane (see Supplementary Fig. S4), lowering
the driving force. This is consistent with previous studies [7, 8, 22, 41].
It is worth noting that in the filtration of direct red 23 or Congo red/NaCl mixture,
Sepro NF membranes had an extremely low flux (< 5.0 Lm-2h-1) at high salt
concentrations. As shown in Supplementary Fig. S4, the osmotic pressure difference
levels off at higher salt concentrations, indicating that the driving force almost keeps
constant. However, the measured membrane flux in dye/NaCl solutions with different
salt concentrations is much lower than the calculated theoretical flux. This implies
19
that Sepro NF membranes may have different fouling mechanisms in dye/salt binary
mixture, which are responsible for the greatly enhanced flux decline compared to the
theoretical predictions. One plausible mechanism is 'cake-enhanced concentration
polarization', originally proposed by Hoek and Elimelech [42]. The formation of an
unstirred cake layer of foulant may promote concentration polarization in the porous
cake layer and lead to an increase of the osmotic pressure difference and a significant
flux decline [42-47]. Furthermore, the increased salt concentration (thus higher ionic
strength) weakens the electrostatic repulsive force between foulant-foulant and
foulant-membrane [12]. As shown in Fig. 10, precipitation of dye occurs with the
increasing salt concentration, since the presence of background electrolytes can
effectively shield the electrostatic double layer interaction [12] as well as increase the
hydrophobicity of dyes [48]. This implies a reduced stability of dye molecules which
promotes more severe membrane fouling.
Fig. 10 Digital pictures for the dye/NaCl mixtures with different salt concentrations.
(A): direct red 80; (B): Congo red; (C): direct red 23
105
100
95
90
85
80
Direct red 80
Direct red 23
Congo red
Dye
Fig. 11 Flux recovery of fouled NF membranes after water rinsing or HCl cleaning
As shown in Fig. 11, Sepro NF membranes have a high flux recovery (>95.6%) after
rinsing with pure water. This is due to their superhydrophilicity which reduces the
affinity of dye molecules on the membrane surface as well as no chemical interactions
between dye molecules and membrane surface [31]. Compared to Sepro NF 2A,
Sepro NF 6 had a lower flux recovery ratio, even though it had a more hydrophilic
surface. This is ascribed to the fact that a trace amount of dye foulant deposited on the
surface of Sepro NF 6, which can be confirmed by SEM images in Fig. 12. After
rinsing with water, Sepro NF 2A exhibits a smooth surface with a ~99% flux recovery.
AFM measurement also shows that the membrane surface can be cleaned with water,
indicating the physical cleaning can be an acceptable approach for flux recovery of
Sepro NF membranes, especially for Sepro NF 2A.
21
(A1)
(B1)
Dye foulant
Impurities
(A2)
(B2)
(A3)
(B3)
Fig. 12 SEM and 3D AFM images of cleaned NF membrane surfaces by water rinsing.
(A): Sepro NF 6; (B): Sepro NF 2A
Additionally, cleaning membranes with HCl gave rise to a nearly complete recovery
(>99.4%) of the water flux for both Sepro NF membranes. Fig. 13 shows that both
Sepro NF membranes had a clean surface, only with a small amount of crystalline
impurity deposited on the membrane surface. Therefore, Sepro NF membranes had an
22
excellent resistance to fouling. HCl cleaning offered a more effective route for the
complete flux recovery of Sepro NF membranes, compared to rinsing with water.
(A1)
(B1)
Impurities
(A2)
(B2)
(A3)
(B3)
Fig. 13 SEM and 3D AFM images of cleaned NF membrane surface by HCl cleaning.
(A): Sepro NF 6; (B): Sepro NF 2A
Fig. 14 shows the flux of Sepro NF membranes for direct red 80/NaCl binary mixture
in diafiltration and post-concentration.
Mea. (Diafiltration)
Mea. (Concentration)
Flux, Lm-2h-1
100
3 times
1 time 2 times
90
(B) 100
Cal. (Diafiltration)
Cal. (Concentration)
Cake
layer
CECP
70
1 time 2 times
60
Postconcentration
Diafiltration
3 times
4 times
40
Postconcentration
Diafiltration
20
50
Cake
layer
50 CECP
30
60
Cal. (Diafiltration)
Cal. (Concentration)
80
4 times
80
70
Mea. (Diafiltration)
Mea. (Concentration)
90
Flux, Lm-2h-1
(A) 110
10
0
100
200
300
400
500
600
200
400
600
800
1000
Fig. 14 indicates that the flux of both Sepro NF membranes had a similar tendency. In
the initial stage of diafiltration, the large gap between the measured and calculated
flux of Sepro NF membranes resulted from the cake-enhanced concentration
polarization. As the diafiltration proceeded, the decrease of salt content mitigated the
cake-enhanced concentration polarization phenomenon, enhancing the driving force
for water transfer through the NF membranes. Furthermore, the reduction of the salt
concentration also alleviated the formation of a cake layer, since the salt effect on the
dye aggregation was limited, as demonstrated in Fig. 10. As the concentration of salt
in the feed was reduced to a fixed small content, the cake-enhanced concentration
polarization phenomenon can be ignored. Fig. 15 schematically shows the evolution
24
Following the diafiltration, the post-concentration procedure was performed. The flux
of Sepro NF 6 and NF 2A in this stage slightly decreased from 86.6 to 81.9 Lm-2h-1
and from 56.2 to 52.4 Lm-2h-1, after a concentration factor of 4.0, respectively. Due
to the high salt permeation of Sepro NF membranes, the concentration of NaCl in the
feed had a slight increase (see Supplementary Fig. S5), and no obvious adverse effect
on promoting cake-enhanced concentration polarization occurred. Therefore, as
schematically demonstrated in Fig. 16, the mild difference between the calculated and
measured flux of Sepro NF membranes was due to the evolution of dye cake layer in
post-concentration with different concentration factors.
25
4. Conclusions
In this study, two superhydrophilic loose NF membranes (Sepro NF 6 and NF 2A
membrane, Ultura) were applied for dye/NaCl fractionation with a full investigation
on flux behavior. It was demonstrated that in the treatment of textile wastewater with
a high salinity, the loose NF membranes can generally suffer the concentration
polarization and the formation of dye cake layer. Particularly, the pore blocking and
dye deposition can be partially involved in the flux decline of loose NF membranes
with a lager pore size and nodules. The enhancing hydrophilicity of surface for Sepro
NF membranes can reduce the physical or chemical adsorption of dye. Furthermore,
in the highly loaded textile wastewater, a specific fouling mechanism, denoted as
cake-enhanced concentration polarization, can significantly deteriorate the
membrane flux through increasing osmotic pressure as well as exacerbating the
formation of dye cake layer.
Application of diafiltration for a dye/salt binary mixture indicates that cake-enhanced
concentration polarization played a dominant role for the low flux at the initial step.
26
Acknowledgments
J. Lin and W. Ye would like to acknowledge the support provided by China
Scholarship Council of the Ministry of Education, China. Michle Vanroelen from
CIT, KU Leuven, is acknowledged for performing Ion Chromatography
measurements. Ultura (USA) is greatly thanked for supplying the NF membrane
samples.
References
[1] J.S. Guest, S.J. Skerlos, J.L. Barnard, M.B. Beck, G.T. Daigger, H. Hilger, S.J.
Jackson, K. Karvazy, L. Kelly, L. Macpherson, A new planning and design paradigm
to achieve sustainable resource recovery from wastewater 1, Environ. Sci. Technol.,
43 (2009) 6126-6130.
27
[2] P.L. McCarty, J. Bae, J. Kim, Domestic wastewater treatment as a net energy
producercan this be achieved?, Environ. Sci. Technol., 45 (2011) 7100-7106.
[3] E. Fernando, T. Keshavarz, G. Kyazze, Simultaneous co-metabolic decolourisation
of azo dye mixtures and bio-electricity generation under thermophillic (50 C) and
saline conditions by an adapted anaerobic mixed culture in microbial fuel cells,
Bioresour. Technol., 127 (2013) 1-8.
[4] S. Kalathil, J. Lee, M.H. Cho, Efficient decolorization of real dye wastewater and
bioelectricity generation using a novel single chamber biocathode-microbial fuel cell,
Bioresour. Technol., 119 (2012) 22-27.
[5] S. Burkinshaw, C. Graham, Recycling of exhausted reactive dyebaths, Dyes
Pigments, 28 (1995) 193-206.
[6] Y. Yang, L. Xu, Reusing hydrolyzed reactive dyebath for nylon and wool dyeing,
Am. Dyestuff Rep., 85 (1996) 27-34.
[7] J. Lin, W. Ye, H. Zeng, H. Yang, J. Shen, S. Darvishmanesh, P. Luis, A. Sotto, B.
Van der Bruggen, Fractionation of direct dyes and salts in aqueous solution using
loose nanofiltration membranes, J. Membr. Sci., 477 (2015) 183-193.
[8] B. Van der Bruggen, B. Daems, D. Wilms, C. Vandecasteele, Mechanisms of
retention and flux decline for the nanofiltration of dye baths from the textile industry,
Sep. Purif. Technol., 22 (2001) 519-528.
[9] R. Valladares Linares, Z. Li, V. Yangali-Quintanilla, Q. Li, G. Amy, Cleaning
protocol for a FO membrane fouled in wastewater reuse, Desalin. Water Treat., 51
(2013) 4821-4824.
28
membranes
review:
Recent
advances
and
future
prospects,
Highlights
- A loose nanofiltration (NF) membrane was employed for dye/salt separation
- CECP mechanism substantially deteriorated the flux of loose NF membrane in
dye/salt mixture
- The loose NF membrane had an excellent antifouling against dye due to its
superhydrophilicity
- Water rinsing can be an effective method to recover the flux of loose NF membrane
34