Author's personal copy Long D. Nghiem , Tzahi Cath

Author's personal copy
Separation and Purification Technology 80 (2011) 315–322
Contents lists available at ScienceDirect
Separation and Purification Technology
journal homepage: www.elsevier.com/locate/seppur
A scaling mitigation approach during direct contact membrane distillation
Long D. Nghiem a,⇑, Tzahi Cath b,⇑
a
b
Strategic Water Infrastructure Laboratory, School of Civil, Mining and Environmental Engineering, University of Wollongong, NSW 2522, Australia
Division of Environmental Science and Engineering, Colorado School of Mines, Golden, CO 80401, USA
a r t i c l e
i n f o
Article history:
Received 22 December 2010
Received in revised form 5 May 2011
Accepted 10 May 2011
Available online 14 May 2011
Keywords:
Direct contact membrane distillation
Membrane scaling
Gypsum
Induction time
Brine treatment
a b s t r a c t
Membrane scaling during the treatment of aqueous solutions containing sparingly soluble salts by direct
contact membrane distillation (DCMD) was investigated. The results reveal that membrane scaling
caused by CaSO4 was more severe than that by CaCO3 or silicate. However, under the experimental condition used in this study and at feed and distillate temperature of 20 °C and 40 °C, respectively, CaSO4
scaling occurred only after a sufficiently long induction time of up to 25 h (corresponding to a saturation
index of up to 1.5). The induction period decreased and the size of the CaSO4 crystals increased as the feed
temperature increased. SEM analysis reveals that prior to the onset of CaSO4 scaling, the membrane surface was relatively clean and was completely free of any large crystals. Subsequently, a simple operational regime involving regular membrane flushing to reset the induction period was developed and
was proven to be effective in controlling CaSO4 scaling. At a low system recovery, the permeate flux
was constant despite the fact that the feed solution was always at a super saturation condition. Results
reported here also confirm the interplay between induction time and the saturation index.
Crown Copyright Ó 2011 Published by Elsevier B.V. All rights reserved.
1. Introduction
Membrane distillation (MD) is a thermally driven desalination
process that involves phase conversion from liquid to vapor on
one side of the membrane and condensation of vapor to liquid
on the other side [1]. The hydrophobic microporous membrane
facilitates the transport of water vapor through its pores while
maintaining vapor–liquid interfaces at the pore entrance, but it
does not participate in the actual separation process. Although
the process of MD is not new, it has only recently been recognized
as a low cost, energy saving alternative to conventional separation
processes such as thermal distillation, nanofiltration (NF) and reverse osmosis (RO) [2–3]. MD has several advantages compared
to other desalination processes for the treatment of saline water
and wastewater [3–4]. Because water is transported through the
membrane only in a vapor phase, MD can offer complete rejection
of all non-volatile constituents in the feed solution; thus, almost
100 percent rejection of ions, dissolved non-volatile organics, colloids, and pathogenic microorganisms can be achieved via the
MD process. But more importantly, due to the discontinuity of
the liquid phase across the membrane, water flux in MD is not
influenced by the osmotic pressure gradient across the membrane.
Consequently, the greatest potential of MD can be realized through
the treatment of highly saline solutions [5]. In fact, it has been
experimentally demonstrated that water flux in the MD process
⇑ Corresponding authors. Tel.: +61 (2) 4221 4590 (L.D. Nghiem), tel.: +1 (303) 273
3402 (T. Cath).
E-mail addresses: longn@uow.edu.au (L.D. Nghiem), tcath@mines.edu (T. Cath).
is almost completely independent of the feed water salinity up to
76,000 mg/L total dissolved solids (TDS), which is twice the salinity
of a typical seawater [6]. MD also requires lower operating pressures than all pressure-driven membrane processes, and particularly RO. In addition, MD requires lower operating temperatures
than conventional distillation, which can facilitate the utilization
of low grade heat [1,7]. The unique ability of MD to utilize low
grade heat from industrial sources (which may otherwise be
wasted) or solar thermal energy provides an excellent platform
for a carbon–neutral desalination process [1].
The driving force for mass transport in the MD process is the vapor pressure difference induced by a temperature difference between the liquid–vapor interfaces on the feed and distillate sides
of the membrane. MD can be employed in four different configurations including vacuum, air gap, sweep gas, and direct contact
membrane distillation [1]. Among these configurations, the direct
contact membrane distillation (DCMD) configuration is well suited
for applications such as desalination or the concentration of aqueous solutions, in which water is the major permeating component
[8–9]. Indeed, DCMD requires the least equipment and is the simplest to operate [8]. MD has potential applications in many areas of
scientific and industrial interest, yielding highly purified permeate
and separating contaminants from liquid solutions. It has been
tested for the treatment of thermally sensitive industrial products
such as concentrating aqueous solution in fruit juices, the biotechnology industry, as well as for wastewater treatment and seawater
desalination [5,10–18]. Because water flux is negligibly dependent
on the feed solution osmotic pressure (or salinity), DCMD is particularly ideal for the treatment of saline solution such as RO concen-
1383-5866/$ - see front matter Crown Copyright Ó 2011 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2011.05.013
Author's personal copy
316
L.D. Nghiem, T. Cath / Separation and Purification Technology 80 (2011) 315–322
trate from inland water recycling or brackish water desalination
processes [4–5,19].
MD has been extensively studied over the last few years given
the growing interest in the development of low energy desalination technologies. Examples of recent and innovative development
in MD research include surface modification of ceramic membranes [20], the synthesis of carbon nanotube membranes [21],
and the development of the vacuum enhanced DCMD [6]. Membrane fouling, and more importantly membrane scaling, have been
identified as a major challenge currently hindering the realization
of full scale MD installation for desalination purposes [15,22].
Membrane fouling is caused by the deposition of dissolved or colloidal organic matter on the membrane surface [23]. The deposited
materials can cause severe wetting of the membrane pores and can
eventually lead to a phenomenon known as pore flooding (the
intrusion of liquid water into the membrane pores) [23]. This phenomenon is further exacerbated if salt crystals can be formed inside the membrane pores. Pore flooding leads to a dramatic
increase in the permeation of both water and salts through the
membrane [24]. Membrane scaling during MD is mostly due to
the precipitation of sparingly soluble salts such as CaSO4, CaCO3,
and silicate directly on the membrane surface. The scaling layer
hinders the mass transport of water vapor across the membrane
and reduces the membrane surface hydrophobicity. Both phenomena can be detrimental to the membrane permeate flux. The scaling of sparingly soluble salts has been the subject of numerous
investigations over the past decade. The rate of scale formation
can be governed by several factors including the degree of supersaturation, temperature, water composition, flow conditions, the
material of the substrate, and the availability of any nucleation
sites [24–26]. Koyuncu and Wiesner successfully correlated the
variability in the morphology of CaSO4 and CaCO3 precipitates occurred during NF and RO filtration processes to feed water composition and operating conditions. Hoang et al. [27] investigated the
effects of temperature on the formation of CaSO4 on stainless steel
pipes. They reported an increase in the rate of CaSO4 crystallization
and a significant decrease in the induction period as the solution
temperature increased. Similar results have also been reported
by Chong and Sheikholeslami [25]. More recently, Gryta [24] conducted a systematic study of the behavior of CaSO4 and CaCO3 scaling in MD. Gryta reported a severe case of pore flooding and
deterioration of the distillate quality due to membrane scaling
and fouling when saline wastewater containing CaSO4 was used
as the feed solution to the MD process [24]. When the author used
tap water as the feed solution, membrane scaling could also be observed; however, he did not observe any deterioration of the distillate quality as evident by the electrical conductivity of the distillate
of below 2.3 lS/cm [24].
Despite the growing number of studies on MD, it is noteworthy
that investigations of membrane scaling behavior and cleaning
strategies remain very scarce. In a typical MD process, since the applied pressure is negligible compared to RO and the feed solution
does not enter the membrane pores, the nature of chemical interactions between membrane and process solutions are not expected
to be the same as that in pressure driven membrane filtration processes [1]. Nevertheless, to date, no studies have been undertaken
to elucidate this important premise of the MD process. When comparing the fouling behaviors during forward osmosis (FO) (another
novel membrane process that involves negligible hydraulic pressure) and RO, it has been previously demonstrated that membrane
fouling in the FO process is almost fully reversible and could be
controlled effectively by optimizing the hydrodynamics of the feed
stream without employing any chemical cleaning reagents [5,28–
29]. Lee et al., [28] attributed this reversible fouling behavior in
FO operation to the absence of a hydraulic pressure as opposed
to a typical RO operation. Given the similarity between FO and
DCMD, one may also expect a similar reversible fouling and scaling
behavior in DCMD operation.
The main objective of this study was to investigate the scaling
behaviors of three sparingly soluble salts, including CaSO4, CaCO3,
and silicate, during DCMD. The decline in permeate flux due to
membrane scaling was related to the kinetic of the crystallization
process and surface morphology of the scaling layer. In particular,
this study aimed to evaluate a strategy to proactively control membrane scaling at an early stage, thus could allow for a sustainable
operation of DCMD without any severe membrane scaling.
2. Materials and methods
2.1. Laboratory-scale DCMD system
DCMD experiments were conducted using a closed-loop benchscale membrane test unit (Fig. 1). The membrane cell was made of
acrylic plastic to minimize heat loss to the surroundings. It was designed to hold a flat-sheet membrane under moderate pressure differential without any physical support. The flow channels were
engraved in each of two acrylic blocks that make up the feed and
permeate semi-cells. Each channel is 3 mm deep, 95 mm wide,
and 145 mm long; and the total active membrane area for mass
transfer is 138 cm2.
Feed solution was continuously pumped from a feed reservoir
to the membrane cell, then through a PVC tube encasing a heating
element before returning back to the reservoir. A temperature sensor was placed immediately before the inlet of the feed to the
membrane cell. The heating element and the temperature sensor
were connected to a temperature control unit that was used to regulate the temperature of the feed solution. MilliQ water was used
as the initial distillate stream. The distillate was circulated from a
2 L reservoir through the distillate membrane semi-cell and back
to the reservoir. The reservoir allowed overflow of excess permeating water into a collecting container, which was continuously
weighed on an analytical balance. Another temperature sensor
was installed immediately at the outlet of the distillate semi-cell.
Temperature of the distillate was regulated using a stainless steel
heat exchanging coil which was directly immerged into the distillate reservoir. Cool water was circulated in the heat exchanging
coil from a centralized chiller. The flowrates of the feed and distillate were monitored using two rotameters and were kept constant
and similar at all times. Electrical conductivity of the distillate was
continuously monitored using an Orion 4 Star Plus portable pH/
conductivity meter (Thermo Scientific, Waltham, MA). All pipes
used in the DCMD test unit were covered with insulation foam to
minimize heat loss. Unless otherwise stated, all experiments were
conducted continuously until they were terminated.
2.2. Microporous membrane
A hydrophobic, microporous membrane (Magna PTFE) was acquired from GE/Osmonics (Minnetonka, MN) for this investigation.
This is a composite membrane having a thin polytetrafluoroethylene (PTFE) active layer on top of a polypropylene (PP) support sublayer. According to the manufacturer, the pore size and porosity of
the membrane are 0.22 lm and 70%, respectively [30]. The membrane thickness is 175 lm, of which the active layer thickness is
approximately 5 lm [6,30].
2.3. Experimental protocol
Certified ACS grade NaCl, CaSO4, Na2SiO3, CaCO3, KHCO3, and
CaCl2 (Fisher Scientific, Pittsburgh, PA) were used in this study.
The feed solution was prepared by dissolving an appropriate
Author's personal copy
317
L.D. Nghiem, T. Cath / Separation and Purification Technology 80 (2011) 315–322
Fig. 1. The DCMD set-up: (1) cooling units; (2) distillate overflow collector; (3) distillate reservoir; (4) recirculating pumps; (5) membrane cell; (6) feed reservoir (with
encased heating element).
2.4. Surface characterization techniques
The morphology of the fouling/scaling layer deposited onto the
membrane surface was examined by a Hitachi TM-1000 Scanning
Electron Microscope (SEM). The membrane samples were dried
in a desiccator and were subsequently analyzed without further
treatment. The scaled membrane samples were handled gently
and without any excessive forces to ensure that the fouling and
scaling layer remained intact.
14
12
2
Permeate flux (L/m h)
amount of chemical into MilliQ water. A feed volume of 10 L was
used and the temperature of the distillate was kept constant at
20 °C in all experiments in this study. A new membrane sample
was used for each experiment in this study. At the completion of
each experiment, the membrane was removed from the membrane
cell. Excess liquid on the membrane surface was allowed to drain
off by gently tilting the membrane coupon. The sample was immediately placed in a desiccator for subsequent surface analysis.
10
8
6
CaCl2 & KHCO3
CaCO3
4
Na2SiO3
2
CaSO4
0
0
10
20
30
40
50
60
Time (hours)
Fig. 2. Permeate flux as a function of time. Feed solution containing (a) 10 mM
(1000 mg/L) CaCO3; (b) 10 mM (1100 mg/L) CaCl2 and 10 mM (1000 mg/L) KHCO3;
(c) 1000 mg/L Na2SiO3; or (d) 2000 mg/L CaSO4. Experimental conditions: Tf = 40 °C;
Td = 20 °C; Qf = Qd = 1 L/min.
3. Results and discussion
3.1. Membrane scaling by different sparingly soluble salts
Water flux as a function of operation time with four different
feed solutions is illustrated in Fig. 2. All four solutions, which were
prepared at, or near, the saturation limit, resulted in the same initial permeate flux. Because inorganic salts cannot be transported
through the membrane, the solutions became supersaturated as
the experiment progressed, causing the precipitation of the sparingly soluble salts. Results presented in Fig. 2, however, reveal that
the three potential scalants resulted in very different membrane
scaling behavior. SEM micrographs of the virgin membrane and
membrane samples at the conclusion of the DCMD experiments
are shown in Fig. 3.
It was somewhat surprising that CaCO3 did not cause any discernible flux decline (Fig. 2). As the experiment involving a pure
CaCO3 solution progressed, the feed solution became cloudy almost
immediately, reflecting the precipitation of CaCO3 in the bulk solution. Fine CaCO3 particles were observed to be suspended in the
feed solution. The experiment was terminated when the feed solution volume decreased to below one litre (equivalent to a concentration factor of more than 10). SEM analysis at the conclusion of
the DCMD experiment revealed the presence of small salt crystals
scattered on the membrane surface. However, these salts crystals
did not completely cover the entire membrane surface area
(Fig. 3b). A similar flux behavior was also observed with a CaCl2/
KHCO3 solution (Fig. 2). In both cases, the initial solution pH was
adjusted to 7.5 to increase the solubility of CaCO3. It is noteworthy
that the content of other impurities in these two solutions was
negligible, which could explain the difference between this investigation and several previous studies in which DCMD was investigated and membrane scaling caused by CaCO3 was reported
[15,23]. In fact, a recent study reported by He et al. [31] has also
reported negligible membrane scaling caused by a pure CaCO3
solution. They explained the negligible impact of CaCO3 scaling
on permeate flux by the fast precipitation rate of CaCO3 and the
transport of CO2 across the membrane [31]. The former would
cause CaCO3 to precipitate in the bulk solution rather than on
the membrane surface while the latter would limit the precipitation of CaCO3 directly on the membrane surface. In good agreement with another study previously reported by He et al. [32],
the feed solution pH sharply decreased to 6 within the first hour
of the experiment, then the pH gradually increased to approximately 7 at the end of the experiment. The observed initial drop
of the feed solution pH could be explained by the precipitation of
calcite and the release of CO2. The adsorption of the acidic gas
CO2 led to a more acidic feed solution [32]. As the rate of calcite
precipitation slowed down and the CO2 desorbed from the feed,
the solution pH increased with time toward the initial pH value.
Karakulski and Gryta [33] reported a gradual permeate flux decline caused by the precipitation of silicate on the membrane surface during DCMD operation. They attributed the observed flux
decline to the clogging of the capillary pores, which would subsequently lead to an increase in mass and heat transfer resistance
across the membrane [33]. A gradual permeate flux decline caused
by Na2SiO3 scaling was also observed in the current study (Fig. 2).
SEM analysis at the end of the experiment confirmed an amorphous scaling layer on the membrane surface (Fig. 3c).
Author's personal copy
318
L.D. Nghiem, T. Cath / Separation and Purification Technology 80 (2011) 315–322
Fig. 3. SEM micrographs of (a) virgin MD membrane; and of the membrane surface after experiments with feed solution containing (b) CaCO3; (c) Na2SiO3; and (d) CaSO4.
Experimental conditions: Tf = 40 °C; Td = 20 °C; Qf = Qd = 1 L/min.
other crystals. Because CaSO4 crystals can grow directly on and
completely cover the membrane surface, a complete loss of permeate flux would be inevitable. The presence of needle shaped CaSO4
crystals on a completely covered membrane surface at the end of
the DCMD experiment involving a saturated CaSO4 solution can
be seen in Fig. 3d.
Continuous monitoring of the distillate quality confirmed that
the flooding of membrane pore did not occur in all experiments
conducted in this study. In all cases, a gradual decrease in the electrical conductivity of the distillate was observed, indicating that
only pure water vapour was transported through the membrane.
In addition, no abnormal increase in the permeate flux was observed. Based on these results and several previous studies
The most severe form of membrane scaling was observed with
CaSO4 (Fig. 2) despite the fact that the CaSO4 scaling layer appears
to be much more porous than that of CaCO3 and Na2SiO3. It is interesting to note that the permeate flux was stable for the first 20 h of
operation, followed by a dramatic drop to almost zero in just a few
hours (Figs. 2 and 3d). The delay in CaSO4 scaling and the severity
of this form of scaling reported here are consistent with the literature [5,31]. In the absence of any scale inhibitors, the precipitation
of CaCO3 occurs instantaneously once a saturation condition has
been reached, whereas there can be a significant induction time
for the precipitation of CaSO4 of up to almost 2 days [34]. However,
once nucleuses are formed, the CaSO4 crystals are able to grow rapidly to form needle shaped structures without the interference of
35
35
o
25
20
15
10
5
o
(a)
0
Feed Temp = 60 C
o
Feed Temp = 50 C
o
Feed Temp = 40 C
30
2
2
Permeate flux (L/m h)
30
Permeate flux (L/m h)
Feed Temp = 60 C
o
Feed Temp = 50 C
o
Feed Temp = 40 C
25
20
15
10
5
(b)
0
1600 2000 2400 2800 3200 3600 4000
Feed concentration of CaSO4 (mg/L)
0
5
10
15
20
25
30
35
40
Time (hours)
Fig. 4. Permeate flux as a function of (a) the calculated CaSO4 concentration in the feed and (b) time at different feed temperatures. Experimental conditions: Td = 20 °C; initial
CaSO4 concentration 2000 mg/L; Qf = Qd = 1 L/min.
Author's personal copy
L.D. Nghiem, T. Cath / Separation and Purification Technology 80 (2011) 315–322
319
Fig. 5. Effect of feed temperature on the morphology of the CaSO4 scaling layer. SEM micrographs (a), (b), and (c) correspond to feed temperature of 40, 50, and 60 °C,
respectively. The magnifications were 250 times for the left and 2000 times for the right micrograph. Experimental conditions: Td = 20 °C; initial CaSO4 concentration
2000 mg/L; Qf = Qd = 1 L/min.
[6,24], it was concluded that the penetration of the feed solution in
liquid form and precipitates into the membrane pores did not
occur.
14
CaSO4 was selected for further investigated in this study, given
its significant membrane scaling propensity in comparison to
CaCO3 and silicate. DCMD experiments were conducted at three
different feed solution temperatures to elucidate the impact of
temperature on the crystallization kinetic of CaSO4. Because CaSO4
cannot be transported across the membrane, concentration of
CaSO4 in the feed solution can be readily calculated based on a simple mass balance equation, assuming that CaSO4 remains in the
aqueous phase. The permeate flux can then be presented as a function of CaSO4 concentration (or the degree of supersaturation) and
as a function of time (Fig. 4). The reported results show a decrease
in the induction period for the precipitation of CaSO4 as the feed
temperature increased. In addition, the sizes of the CaSO4 crystals
also correlated very well to the feed solution temperature. Feed
temperature of 40 °C resulted in the formation of thin needle
shaped crystals (Fig. 5a). Larger CaSO4 crystals could be observed
2
3.2. Scaling behavior of CaSO4
Permeate flux (L/m h)
12
10
8
6
4
500 mg/L
1,000 mg/L
2,000 mg/L
2
0
0
500
1000 1500 2000 2500 3000 3500 4000
Feed concentration of CaSO4 (mg/L)
Fig. 6. Permeate flux as a function of the calculated CaSO4 concentration in the feed
solution with different initial concentration of CaSO4. Experimental conditions:
Tf = 40 °C; Td = 20 °C; Qf = Qd = 1 L/min.
on the membrane surface when the feed temperature was 50 °C
(Fig. 5b); and when the feed temperature further increased to
Author's personal copy
320
L.D. Nghiem, T. Cath / Separation and Purification Technology 80 (2011) 315–322
14
2
Permeate flux (L/m h)
12
10
8
6
4
2
0
0
5
10
15
20
25
30
35
40
Time (hours)
Fig. 7. Permeate flux as a function of time. The experiments were terminated at
10%, 50%, and 90% permeate flux reduction. Experimental conditions: initial feed
CaSO4 concentration 2000 mg/L; Tf = 40 °C; Td = 20 °C; Qf = Qd = 1 L/min.
60 °C, very large CaSO4 crystals were clearly visible on the membrane surface (Fig. 5c). These results are consistent with the general theory of CaSO4 precipitation kinetics and thermodynamics
[32,35]. He et al. simulated the saturation index of CaSO4 as a func-
tion of temperature and found that the saturation index increased
as temperature increased [32]. Their subsequent experimental data
also confirmed a more severe CaSO4 scaling at the inlet of the
membrane module where the temperature was highest [35].
The induction time appears to be a key parameter governing the
crystallization of CaSO4 on the membrane surface. When a
2000 mg/L CaSO4 feed solution was tested, the feed reached a saturation index of 1.5 (corresponding to feed concentration of CaSO4
of 3500 mg/L) prior to the onset of membrane scaling. On the other
hand, when a 500 mg/L CaSO4 solution was tested, the onset of
membrane scaling occurred almost immediately at the point of
saturation (Fig. 6). The onset of membrane scaling occurred after
approximately 53, 43, and 30 h of continuous operation when
the feed concentration of CaSO4 was 500, 1000, and 2000 mg/L,
respectively. In this study, the onset of membrane scale was determined by the point at which the permeate flux began to decline.
Results illustrated in Fig. 6 suggest that a supersaturation condition of CaSO4 is not the only prerequisite for membrane scaling.
In all three cases, membrane scaling can only occur after a sufficient induction time.
The long induction time observed with CaSO4 provides an
opportunity for the control of membrane scaling during DCMD
operation. Operating the DCMD process at a low feed solution tem-
Fig. 8. Effect of induction time on the morphology of the CaSO4 scaling layer. The DCMD experiments were terminated at (a) 10%, (b) 50%, and (c) 90% permeate flux
reduction, respectively. The magnifications were 250 times for the left and 2000 times for the right micrograph. Experimental conditions: Tf = 40 °C; Td = 20 °C; Qf = Qd = 1 L/
min.
Author's personal copy
321
5000
12
4000
2
Permeate flux (L/m h)
14
10
3000
8
6
2000
4
1000
Permeate flux
Feed concentration of CaSO4
2
0
0
0
20
40
60
80
100
Feed concentration of CaSO4 (mg/L)
L.D. Nghiem, T. Cath / Separation and Purification Technology 80 (2011) 315–322
120
Time (hour)
Fig. 9. Permeate flux of five repetitive DCMD tests and the corresponding CaSO4 concentration in the feed solution. The membrane was flushed with MilliQ water at the end
of each test and a fresh 2000 mg/L CaSO4 solution was used for the subsequent test. Experimental conditions: Tf = 40 °C; Td = 20 °C; Qf = Qd = 1 L/min.
6000
2
Permeate flux (L/m h)
14
5000
12
4000
10
8
3000
6
2000
4
Permeate flux
Feed concentration of CaSO4
2
1000
0
0
0
5
10
15
20
25
30
35
Feed concentration of CaSO4 (mg/L)
Membrane flushing with MilliQ water
16
40
Time (hour)
Fig. 10. Permeate flux and the corresponding CaSO4 concentration in the feed solution. The initial feed solution contained 2000 mg/L CaSO4. The membrane was flushed with
MilliQ water after approximately 20 h of operation. Experimental conditions: Tf = 40 °C; Td = 20 °C; Qf = Qd = 1 L/min.
perature can prolong the induction time of CaSO4 crystallization,
hence, delaying the onset of membrane scaling. Another interesting approach is to ‘reset’ the induction period, and therefore prevent the formation of CaSO4 crystals on the membrane surface.
To evaluate this approach, a DCMD experiment was terminated
when the decrease in permeate flux has reached by approximately
10% (Fig. 7); the membrane sample was removed for examination
at the onset of membrane scaling. SEM analysis of the membrane
sample revealed the presence of a few small CaSO4 crystals on
the membrane surface. Although much of the membrane surface
remained clean and available for the transport of water vapor,
these small CaSO4 crystals could potentially act as nucleation sites
for the crystallization of further and larger CaSO4 crystals. When
the experiment was terminated at approximately 50% permeate
flux reduction, patches of cubical CaSO4 crystals were covering
more than half of the membrane available surface. And when the
DCMD process was allowed to proceed further (until approximately 90% permeate flux decline), CaSO4 crystals completely covered the membrane surface, resulting in a severe loss of permeate
flux as can be seen in Figs. 7 and 8. All three experiments showed a
consistent induction period of approximately 25 h (Fig. 7).
3.3. Membrane scaling mitigation strategies
Results reported in Section 3.2 suggest that membrane scaling
caused by CaSO4 can be effectively controlled by removing the
nucleation sites from the membrane surface prior to the onset of
a rapid crystallization and membrane scaling. A series of tests conducted in this study has confirmed the feasibility of this approach.
The DCMD system was challenged with 10 L of 2000 mg/L CaSO4
feed solution. After 20 h of operation, the DCMD process was temporarily interrupted and the membrane was flushed with MilliQ
water. A fresh 10 L batch of 2000 mg/L CaSO4 solution was introduced to the feed reservoir and the DCMD process was resumed
again. The flushing cycle was repeated five times. Water recovery
in each cycle was approximately 30%. Throughout this experiment,
the permeate flux remained constant despite the fact that the feed
solution was always at a supersaturation condition (Fig. 9). It appeared that the initial CaSO4 seedlings (or nucleuses) could be easily removed by MilliQ water. As discussed in Section 3.2, because
the DCMD process was halted after 20 h of operation, which is several hours prior to the onset of a rapid crystallization process, the
total mass of CaSO4 crystal deposited on the membrane surface
was negligible. As a result, it was possible to use MilliQ water to
flush these nucleuses away from the membrane to ‘reset’ the
induction period of the crystallization of CaSO4. Previous research
has also confirmed that the nature of membrane scaling in MD differ substantially from that in pressure driven membrane processes.
Although significant crystallization can occur on the membrane
surface during MD, the scaling layer can be effectively removed
by an acidic solution. Results reported here suggest that membrane
scaling caused by the crystallization of CaSO4 can be effectively
controlled by an operational regime involving regular membrane
flushing. Such regime would be comparable to that used in micro-
Author's personal copy
322
L.D. Nghiem, T. Cath / Separation and Purification Technology 80 (2011) 315–322
filtration where backwashing is initiated on a regular basis. Several
other noteworthy techniques that may be used for resetting the
induction process include the applications of magnetic [36] and
ultrasonic [37] fields. In particular, these techniques can also be
useful to limit the formation of nucleuses in the bulk solution
[36–37]. The approach for controlling CaSO4 scaling described here
may also be applicable to other membrane processes such as FO
and electro-dialysis that are not driven by a high pressure gradient.
To evaluate the recovery limit of the system, another series of
experiments was conducted with an initial feed solution of
2000 mg/L of CaSO4. Unlike the previous series, after flushing the
membrane with MilliQ water, the DCMD operation was resumed
with the existing feed solution. Not surprisingly, scaling eventually
occurred when CaSO4 concentration in the feed solution reached
5500 mg/L, corresponding to a system recovery of 64% (Fig. 10).
It is noteworthy that the induction time of the second cycle was
only approximately 15 h. Results reported here confirm that an increase in the concentration of CaSO4 in the feed solution would
inevitably lead to a shortened induction time (Fig. 10). An oversaturated condition may also lead to the formation of nucleuses in
feed solution, thus, resulting in a more rapid crystallization process. Further investigation of the interplay between induction time
and the saturation index of sparingly soluble salts would be necessary to optimize the scaling control procedure proposed in this
study.
4. Conclusion
Scaling caused by CaSO4 on MD membrane was much more severe than scaling caused by CaCO3 or silicate. However, CaSO4 scaling could only occur after a sufficiently long induction time. The
results showed a decrease in the induction period, and the size of
the CaSO4 crystals increased as the feed temperature increased.
Prior to the onset of CaSO4 scaling, the membrane surface was relatively clean and was completely free of any large crystals. A simple operational regime involving regular membrane flushing to
reset the induction period was proposed and tested. The proposed
regime was proven to be very effective for controlling CaSO4 scaling. At a system recovery of 30%, the permeate flux was constant
despite the fact that the feed solution was always at a supersaturation condition. Results reported here also confirm the interplay between induction time and the saturation index.
References
[1] K.W. Lawson, D.R. Lloyd, Membrane distillation, J. Membr. Sci. 124 (1) (1997)
1–25.
[2] F. Macedonio, E. Drioli, Membrane engineering progresses in desalination and
water reuse, Membr. Water Treat. 1 (1) (2010) 75–81.
[3] A.M. Alklaibi, N. Lior, Membrane-distillation desalination: status and potential,
Desalination 171 (2) (2005) 111.
[4] L. Mariah, C.A. Buckley, C.J. Brouckaert, E. Curcio, E. Drioli, D. Jaganyi, D.
Ramjugernath, Membrane distillation of concentrated brines –role of water
activities in the evaluation of driving force, J. Membr. Sci. 280 (1-2) (2006) 937.
[5] C.R. Martinetti, A.E. Childress, T.Y. Cath, High recovery of concentrated RO
brines using forward osmosis and membrane distillation, J. Membr. Sci. 331 (12) (2009) 31–39.
[6] T.Y. Cath, V.D. Adams, A.E. Childress, Experimental study of desalination using
direct contact membrane distillation: a new approach to flux enhancement, J.
Membr. Sci. 228 (1) (2004) 5–16.
[7] V.A. Bui, L.T.T. Vu, M.H. Nguyen, Simulation and optimisation of direct contact
membrane distillation for energy efficiency, Desalination 259 (1-3) (2010) 29–
37.
[8] K.W. Lawson, D.R. Lloyd, Membrane distillation. II. Direct contact MD, J.
Membr. Sci. 120 (1) (1996) 123–133.
[9] J. Zhang, N. Dow, M. Duke, E. Ostarcevic, J.D. Li, S. Gray, Identification of
material and physical features of membrane distillation membranes for high
performance desalination, J. Membr. Sci. 349 (1-2) (2010) 295–303.
[10] E. Garcia-Castello, A. Cassano, A. Criscuoli, C. Conidi, E. Drioli, Recovery and
concentration of polyphenols from olive mill wastewaters by integrated
membrane system, Water Res. 44 (13) (2010) 3883–3892.
[11] Á. Kozák, E. Békássy-Molnár, G. Vatai, Production of black-currant juice
concentrate by using membrane distillation, Desalination 241 (1-3) (2009)
309–314.
[12] G.W. Meindersma, C.M. Guijt, A.B. de Haan, Desalination and water recycling
by air gap membrane distillation, Desalination 187 (1-3) (2006) 291–301.
[13] J. Phattaranawik, A.G. Fane, A.C.S. Pasquier, W. Bing, A novel membrane
bioreactor based on membrane distillation, Desalination 223 (1-3) (2008)
386–395.
[14] M. Gryta, The fermentation process integrated with membrane distillation,
Sep. Purif. Technol. 24 (1-2) (2001) 283–296.
[15] M. Gryta, M. Tomaszewska, K. Karakulski, Wastewater treatment by
membrane distillation, Desalination 198 (1-3) (2006) 67–73.
[16] T.Y. Cath, D. Adams, A.E. Childress, Membrane contactor processes for
wastewater reclamation in space II. Combined direct osmosis, osmotic
distillation, and membrane distillation for treatment of metabolic
wastewater, J. Membr. Sci. 257 (1-2) (2005) 111–119.
[17] T.Y. Cath, Osmotically and thermally driven membrane processes for
enhancement of water recovery in desalination processes, Desalination
Water Treat. 15 (1-3) (2010) 279–286.
[18] J.-P. Mericq, S. Laborie, C. Cabassud, Vacuum membrane distillation of
seawater reverse osmosis brines, Water Res. 44 (18) (2010) 5260–5273.
[19] X. Ji, E. Curcio, S. Al Obaidani, G. Di Profio, E. Fontananova, E. Drioli, Membrane
distillation–crystallization of seawater reverse osmosis brines, Sep. Purif.
Technol. 71 (1) (2009) 76–82.
[20] Z.D. Hendren, J. Brant, M.R. Wiesner, Surface modification of nanostructured
ceramic membranes for direct contact membrane distillation, J. Membr. Sci.
331 (1-2) (2009) 1–10.
[21] L.F. Dumee, K. Sears, J. Schutz, N. Finn, C. Huynh, S. Hawkins, M. Duke, S. Gray,
Characterization and evaluation of carbon nanotube Bucky-Paper membranes
for direct contact membrane distillation, J. Membr. Sci. 351 (1-2) (2010) 36–
43.
[22] E. Curcio, X. Ji, G. Di Profio, A.O. Sulaiman, E. Fontananova, E. Drioli, Membrane
distillation operated at high seawater concentration factors: role of the
membrane on CaCO3 scaling in presence of humic acid, J. Membr. Sci. 346 (2)
(2010) 263–269.
[23] M. Gryta, Fouling in direct contact membrane distillation process, J. Membr.
Sci. 325 (1) (2008) 383–394.
[24] M. Gryta, Calcium sulphate scaling in membrane distillation process, Chem.
Pap. 63 (2) (2009) 146–151.
[25] T.H. Chong, R. Sheikholeslami, Thermodynamics and kinetics for mixed
calcium carbonate and calcium sulfate precipitation, Chem. Eng. Sci. 56 (18)
(2001) 5391–5400.
[26] I. Koyuncu, M.R. Wiesner, Morphological variations of precipitated salts on NF
and RO membranes, Environ. Eng. Sci. 24 (5) (2007) 602–614.
[27] T.A. Hoang, H.M. Ang, A.L. Rohl, Effects of temperature on the scaling of
calcium sulphate in pipes, Powder Technol. 179 (1-2) (2007) 31–37.
[28] S. Lee, C. Boo, M. Elimelech, S. Hong, Comparison of fouling behavior in forward
osmosis (FO) and reverse osmosis (RO), J. Membr. Sci. 365 (1-2) (2010) 34–39.
[29] R.W. Holloway, A.E. Childress, K.E. Dennett, T.Y. Cath, Forward osmosis for
concentration of anaerobic digester concentrate, Water Res. 41 (17) (2007)
4005–4014.
[30] GE Osmonics Labstore, GE PTFE (TeflonÒ) laminated membranes, 2011;
Available from: <www.geosmolabstore.com>.
[31] F. He, K.K. Sirkar, J. Gilron, Studies on scaling of membranes in desalination by
direct contact membrane distillation: CaCO3 and mixed CaCO3/CaSO4 systems,
Chem. Eng. Sci. 64 (8) (2009) 1844–1859.
[32] F. He, K.K. Sirkar, J. Gilron, Effects of antiscalants to mitigate membrane scaling
by direct contact membrane distillation, J. Membr. Sci. 345 (1-2) (2009) 53–58.
[33] K. Karakulski, M. Gryta, Water demineralisation by NF/MD integrated
processes, Desalination 177 (1-3) (2005) 109–119.
[34] R. Sheikholeslami, H.W.K. Ong, Kinetics and thermodynamics of calcium
carbonate and calcium sulfate at salinities up to 1.5 M, Desalination 157 (1-3)
(2003) 217–234.
[35] F. He, J. Gilron, H. Lee, L. Song, K.K. Sirkar, Potential for scaling by sparingly
soluble salts in crossflow DCMD, J. Membr. Sci. 311 (1-2) (2008) 68–80.
[36] C.Y. Tai, M.C. Chang, R.J. Shieh, T.G. Chen, Magnetic effects on crystal growth
rate of calcite in a constant-composition environment, J. Cryst. Growth 310
(15) (2008) 3690–3697.
[37] E. Dalas, The effect of ultrasonic field on calcium carbonate scale formation, J.
Cryst. Growth 222 (1-2) (2001) 287–292.