Forward Osmosis: Principles, Applications, and Recent Developments
Forward Osmosis: Principles, Applications, and Recent Developments
Forward Osmosis: Principles, Applications, and Recent Developments
Review
Received 30 December 2005; received in revised form 30 May 2006; accepted 31 May 2006
Available online 6 June 2006
Abstract
Osmosis is a physical phenomenon that has been extensively studied by scientists in various disciplines of science and engineering. Early
researchers studied the mechanism of osmosis through natural materials, and from the 1960s, special attention has been given to osmosis through
synthetic materials. Following the progress in membrane science in the last few decades, especially for reverse osmosis applications, the interests
in engineered applications of osmosis has been spurred. Osmosis, or as it is currently referred to as forward osmosis, has new applications in
separation processes for wastewater treatment, food processing, and seawater/brackish water desalination. Other unique areas of forward osmosis
research include pressure-retarded osmosis for generation of electricity from saline and fresh water and implantable osmotic pumps for controlled
drug release. This paper provides the state-of-the-art of the physical principles and applications of forward osmosis as well as their strengths and
limitations.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Osmosis; Forward osmosis; Direct osmosis; Desalination; Reverse osmosis; Pressure-retarded osmosis
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2. Osmotic processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.1. Classification of osmotic processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.2. Draw solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3. Concentration polarization in osmotic processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.1. External concentration polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.2. Internal concentration polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.2.1. Modeling concentrative internal concentration polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.2.2. Modeling dilutive internal concentration polarization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.3. Influence of internal concentration polarization on water flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4. Membranes and modules for forward osmosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.1. Membranes for forward osmosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.2. Membrane modules and devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.2.1. Plate-and-frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2.2. Spiral-wound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2.3. Tubular . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.2.4. Hydration bags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Abbreviations: CP, concentration polarization; DOC, direct osmotic concentration; DS, draw solution; FO, forward osmosis; MD, membrane distillation; NPDES,
national pollutant discharge elimination system; OD, osmotic distillation; PRO, pressure-retarded osmosis; RO, reverse osmosis; TDS, total dissolved solids; TKN,
total Kjeldahl nitrogen (sum of organic nitrogen and ammonia); TMDL, total maximum daily load
∗ Corresponding author. Tel.: +1 303 273 3402; fax: +1 303 273 3413.
0376-7388/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2006.05.048
T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87 71
Fig. 1. Solvent flows in FO, PRO, and RO. For FO, P is approximately zero and water diffuses to the more saline side of the membrane. For PRO, water diffuses
to the more saline liquid that is under positive pressure (π > P). For RO, water diffuses to the less saline side due to hydraulic pressure (P > π).
T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87 73
Fig. 4. Illustrations of driving force profiles, expressed as water chemical potential, µw , for osmosis through several membrane types and orientations [48]. (a) A
symmetric dense membrane. (b) An asymmetric membrane with the porous support layer facing the feed solution; the profile illustrates concentrative internal CP.
(c) An asymmetric membrane with the dense active layer facing the feed solution; the profile illustrates dilutive internal CP. The actual (effective) driving force is
represented by µw . External CP effects on the driving force are assumed to be negligible in this diagram.
(πbulk ) is higher than the osmotic pressure difference across layer also substantially affects mass transfer of water across the
the membrane (πm ) due to external CP and that the effec- membrane. The concentration profile across an asymmetric or
tive osmotic pressure driving force (πeff ) is even lower due composite FO membrane for concentrative internal CP is illus-
to internal CP. Furthermore, similar to the operation of heat trated in Fig. 5a. C1 and C5 are the concentrations of the bulk
exchangers, operation of FO in a counter-current flow configu- feed and draw solution, respectively; C2 and C4 are the con-
ration (feed and draw solution flowing tangential to the mem- centrations of the feed-membrane and draw solution-membrane
brane but in opposite directions) provides constant π along interfaces (resulting from external CP), respectively; and C3 is
the membrane module and makes the process more efficient the concentration at the active layer–support layer interface.
[56]. Adopting the models that were developed by Lee et al. [30],
Loeb et al. [50] introduced a simplified equation to describe
3.2.1. Modeling concentrative internal concentration the water flux during FO without consideration for membrane
polarization orientation:
Unlike RO, for which water transport through the membrane
is mostly affected by the hydraulic resistance created by the 1 πHi
Jw = ln (2)
membrane structure, in FO, internal CP in the porous support K πLow
Fig. 5. (a) Concentrative internal CP and (b) dilutive internal CP across a composite or asymmetric membrane in FO.
T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87 75
bags for water purification and osmotic pumps for drug delivery
[26,63].
Even considering their limitations, the most readily avail-
able semi-permeable polymeric membranes are flat sheet mem-
branes. For continuous flow operation of an FO process, flat
sheet membranes can be used in either a plate-and-frame con-
figuration or in a unique spiral-wound configuration.
4.2.1. Plate-and-frame
The simplest device for packing flat sheet membranes is a
plate-and-frame module. Plate-and-frame modules can be con-
structed in different sizes and shapes ranging from lab-scale
devices that hold single, small-size membrane coupons to full-
scale systems that hold more than 1700 membranes. While the
design and construction of large plate-and-frame heat exchang- Fig. 9. Flow patterns in a spiral-wound module modified for FO. The feed solu-
ers is well established, the construction of large plate-and-frame tion flows through the central tube into the inner side of the membrane envelope
membrane modules is more complicated. Two of the main limi- and the draw solution flows in the space between the rolled envelopes. Figure
tations of plate-and-frame elements for membrane applications adapted from [36].
are lack of adequate membrane support and low packing den-
One of the main limitations in operating a spiral-wound
sity. Lack of adequate membrane support limits operation to low
membrane element for RO is the inability to induce flow in the
hydraulic pressure and/or operation at similar pressures on both
permeate channels (i.e., the inner side of the envelopes) of the
sides of the membrane (requiring relatively high process con-
membrane element for the purpose of cleaning or backwashing.
trol). Low packing density leads to a larger system footprint,
A unique application of FO for backwashing of spiral-wound
higher capital costs, and higher operating costs (labor for mem-
RO membranes was investigated by Sagiv and Semiat [64].
brane replacement). Other limitations of the plate-and-frame
Upon stopping the RO process after desalination, water diffused
configuration include problems with internal and external seal-
by osmosis from the permeate channels back into the feed
ing, difficulty in monitoring membrane integrity, and a limited
side of the membrane, thus cleaning the membrane through
range of operating conditions (e.g., flow velocities and pres-
osmotic backwashing. In their investigation, a spiral-wound
sures) [18].
RO membrane was purposely scaled with saturated CaCO3
solution. At predetermined intervals, the system was stopped
4.2.2. Spiral-wound
for 20 s to allow osmotic backwashing to occur. Preliminary
Commercially marketed spiral-wound membrane elements
results (Fig. 10) indicate that almost complete recovery of water
(e.g., spiral-wound RO elements) are operated with only one
flux could be achieved.
stream (the feed stream) flowing under direct control of its flow
velocity tangential to the membrane. The permeate stream flows 4.2.3. Tubular
very slowly in the channel formed by the two glued membranes The use of tubular membranes (tubes or hollow fibers) for
and its composition and flow velocity are controlled by the prop- continuously operated FO processes is more practical for three
erties of the membrane and the operating conditions. Therefore,
in its current design, spiral-wound membrane elements cannot
be operated in FO mode because the draw solution cannot be
forced to flow inside the envelope formed by the membranes.
Mehta [36] designed and successfully tested a unique spiral-
wound element for FO. Both outside-in and inside-out operation
can be used. In Fig. 9 (inside-out operation), the draw solution
flows through the spacers and between the rolled membranes, in
the same way that a feed stream flows in a spiral-wound element
for RO. However, unlike RO elements, the central collecting tube
is blocked halfway through so that the feed solution cannot flow
to the other side. Instead, an additional glue line at the center
of the membrane envelope provides a path for the feed to flow
inside the envelope. In this configuration, the feed flows into the
first half of the perforated central pipe, is then forced to flow into
the envelope, and then flows out through the second half of the
perforated central pipe. It is worth noting that this configuration
can also be efficiently used for PRO. The draw solution outside Fig. 10. Effect of osmotic backwashing on the performance of a spiral wound
of the envelope can be pressurized similar to the way it is done RO membrane after desalination. The feed solution consisted of 0.5% NaCl and
in spiral-wound membrane elements for RO. super-saturated CaCO3 . Figure adapted from [64].
78 T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87
main reasons. First, tubular membranes are self-supported. This study on direct potable reuse of wastewater in advanced life sup-
means they can support high hydraulic pressure without defor- port systems for space applications [5,18,19], and an investiga-
mation and they can be easily packed in bundles directly inside tion on concentration of liquids from anaerobic sludge digestion
a holding vessel. Second, it is much simpler to fabricate tubu- at a domestic wastewater treatment facility [4]. It is worth not-
lar modules and packing density is relatively high. Third, these ing that in most wastewater treatment applications FO is not the
modules allow liquids to flow freely on both sides of the ultimate process, but rather a high-level pretreatment step before
membrane—a flow pattern necessary for FO [65]. Furthermore, an ultimate desalination process.
it is the authors’ opinion that hollow fiber membranes may be
most suitable for FO because there is no need for a thick support 5.1.1. Concentration of dilute industrial wastewater
layer as in flat sheet RO membranes. This will result in reduced One of the first feasibility studies of FO in an industrial appli-
internal CP and enhanced performance. cation for wastewater treatment was published in 1974 [2] and
The main difference between hollow fiber and tubular mem- 1977 [3]. The objective was to use a low energy process to
branes is the flow regime that can be achieved on the bore side. In treat industrial wastewater containing very low concentrations of
hollow fibers (internal diameter <1 mm), only laminar flow can heavy metals for possible reuse. A bench-scale system was used
be achieved, and therefore, mixing at the membrane surface is to study the feasibility of using newly commercialized cellulose
limited. Under these conditions, external CP, fouling, and scal- RO membranes to concentrate dilute real or synthetic wastewater
ing can reduce mass transport across the membrane. In tubular streams containing copper or chromium. Not aware of the effects
membranes (internal diameter ≥2 mm), turbulence can be easily of internal CP in RO membranes, the authors observed water
achieved and CP, fouling, and scaling reduced. fluxes ranging from zero to approximately 4.5 l/m2 h—much
That said, to date, the use of tubular semi-permeable mem- lower than the calculated fluxes of 10–17 l/m2 h from the mass
branes (RO membranes) for FO is limited [12,13,58]. Tubular transfer equation and manufacturer data for the membranes
semi-permeable membranes for RO are not as common in the tested in RO mode under equivalent conditions. Attempts to
market as flat sheet semi-permeable membranes. This is mainly investigate the effects of external CP on flux by varying the feed
due to the shift of the membrane industry to develop flat sheet and draw solution flow rates yielded inconclusive results.
RO membranes rather than tubular RO membranes as a result of Simulated seawater was used as the draw solution because
excessive fouling and membrane integrity problems with hollow it is a potentially inexpensive source available in coastal areas.
fiber RO membranes. At the time when tubular membranes are Passage of sodium chloride from the artificial seawater and
specially designed for FO applications, the manufacturer will diffusion of feed contaminants towards the draw solution
have to take into consideration membrane characteristics and occurred at a higher rate than expected. Relative salt passage
hydraulic conditions for ultimate optimization of tubular mem- was 1 g NaCl for every 11.5–688 g water passage in the opposite
branes for specific FO applications. direction. Different approaches to enhance salt rejection were
investigated including chemical treatment of the membrane
4.2.4. Hydration bags with polyvinyl methyl ether and thermal treatment (tempering)
The hydration bag is another configuration of flat sheet FO by immersing the membranes in hot water (60–93 ◦ C) for up
membrane. It is a double lined bag; the internal bag is made of to 4 min. While chemical treatment showed no effect on flux or
an FO membrane and is filled with draw solution (e.g., flavored rejection, thermal treatment resulted in elevated salt rejection
sucrose) and the external bag is a sealed plastic bag containing but decreased water flux.
the FO bag and the feed water to be treated. The application of Due to the poor performance of the RO membranes, further
hydration bag technology is further described in Section 5.1.5. pilot-scale testing of the process was called off. The authors
concluded that membranes must be improved and specifically
5. Modern applications of forward osmosis ‘tailored’ for the FO process to be feasible for water treatment.
Forward osmosis has been studied for a range of applica- 5.1.2. Concentration of landfill leachate
tions. Commercial applications, though still limited, are emerg- Landfill leachate is a highly variable feed solution that
ing in the water purification field (e.g., extraction bags) and in presents a particularly difficult treatment challenge, especially
the pharmaceutical industry (e.g., osmotic pumps). The follow- when high effluent quality is required. It is a complex solution
ing section summarizes past and present applications of FO in consisting of four general types of pollutants: organic com-
wastewater treatment and water purification, seawater desalina- pounds, dissolved heavy metals, organic and inorganic nitro-
tion, food processing, pharmaceutical applications, and power gen, and total dissolved solids (TDS). A typical composition of
generation. leachate is given in Table 1. The simplest treatment for landfill
leachate is to process it in a wastewater treatment facility. How-
5.1. Wastewater treatment and water purification ever, wastewater treatment facilities generally focus on treating
organics, heavy metals, and nitrogen. They often have no treat-
Several modern applications of FO in the field of wastewater ment for TDS, and in some cases, treatment facilities even
treatment have been published in the literature. These include increase TDS concentration [6].
an early study on concentration of dilute industrial wastewater Two commercially available treatment processes that are
[2,3], an investigation on treatment of landfill leachate [6,7], a able to efficiently remove TDS from wastewater are mechanical
T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87 79
Table 1
Raw leachate and final effluent water quality characteristics [6]
Contaminant (metals or salts) Untreated leachate (g/l) Recovered water from leachate (g/l) Rejection (%) NPDES TMDL (g/l)
Contaminant Untreated leachate (mg/l) Recovered water from leachate (mg/l) Rejection (%) NPDES TMDL (mg/l)
ND = not detected.
evaporation (e.g., vapor compression, vertical tube falling film, The success of the pilot-scale system led to the design and
horizontal tube spray film, forced circulation) and membrane construction of a full-scale system (Fig. 11). In the full-scale
processes. A comprehensive evaluation of vapor recompression system, the raw leachate is collected and pretreated before water
mechanical evaporation, RO, and FO revealed that FO can be is extracted in six stages of FO cells. A three-pass (permeate-
very effective in treating landfill leachate [6]. In 1998, Osmotek staged) RO system produces a stream of purified water for land
constructed a pilot-scale FO system to test the concentration application and a reconcentrated stream of draw solution at
of landfill leachate at the Coffin Butte Landfill in Corvallis, approximately 75 g/l NaCl. The concentrated leachate is solidi-
Oregon [6,7]. Because this landfill is located in an area that fied before disposal [6].
receives more than 1400 mm per year of rainfall, approximately Between June 1998 and March 1999, the treatment plant
20,000–40,000 m3 of leachate is generated annually. In order treated over 18,500 m3 of leachate, achieving an average water
to meet the National Pollutant Discharge Elimination System recovery of 91.9% and an average RO permeate conductivity of
(NPDES) total maximum daily load (TMDL), the leachate must 35 S/cm [6]. The concentrations of major contaminants in the
be treated to a TDS level lower than 100 mg/l prior to land appli- feed and effluent of the full-scale FO system are summarized in
cation. Table 1. Most contaminants had greater than 99% rejection and
An FO pilot system was tested for 3 months using Osmotek’s final effluent concentrations were substantially lower than the
CTA membrane (Section 4.1) and NaCl as the draw solution. NPDES TMDL levels.
Water recoveries of 94–96% were achieved with high contam-
inant rejection. Flux decline was not apparent during the pro- 5.1.3. Direct potable reuse for advanced life support
cessing of raw leachate; however, a flux decline of 30–50% was systems
observed during processing of concentrated leachate. Almost Long-term human missions in space require a continuous and
complete flux restoration was achieved after cleaning. self-sufficient supply of fresh water for consumption, hygiene,
80 T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87
Table 2
Water flux through one FO membrane and four RO membranes in FO mode [18]
Membrane used Water flux (l/m2 h)
CTA—Osmotek 17.4
Osmonics—CE 1.90
Osmonics—CD 1.97
Hydranautics—LFC1 0.54
Hydranautics—LFC3 0.66
100 g/l NaCl solution was used as the draw solution and DI water as the feed.
Fig. 12. Flow diagram of the original NASA DOC test unit. Three waste streams are pretreated in DOC#1 and DOC#2. The draw solution is reconcentrated and
drinking water is produced by the RO subsystem [18].
T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87 81
Fig. 14. Water flux as a function of draw solution (DS) concentration during FO concentration of (a) pretreated centrate and (b) non-pretreated centrate (b) [4]. Flux
decline between trials is due to organic and suspended solids fouling. Flux was markedly restored after a short cleaning cycle with sodium hydroxide.
82 T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87
decline between each trial (for trials #1 through #4) is the result bags can be placed directly in the source water or they can
of membrane fouling over time. After the short cleaning cycle be suspended in another sealed plastic bag that holds the
(between trials #4 and #5), the flux was completely restored to its source water—providing better mobility and autonomy to the
initial level. A similar performance was observed when raw (not user. In recent years, the military procured hydration bags for
pretreated) centrate was used as the feed solution (Fig. 14b). emergency and relief efforts around the world [23]. Yet, there
However, the extent of flux decline between trials was much is debate among experts whether hydration bags provide water
more apparent and was attributed to greater fouling caused by treatment per se because the product is not pure water but a
high concentrations of suspended solids (Fig. 14). Some flux sweet drink that can only be used for specific applications.
restoration after cleaning was achieved; however, complete flux
recovery was not obtained. Phosphorous rejection exceeded 99% 5.2. Seawater desalination
and ammonia and total Kjeldahl nitrogen (TKN) rejections were
approximately 87% and 92%, respectively. Color and odor com- Several patents have been awarded for different methods and
pounds were almost completely rejected. It is expected that a systems for water desalination by FO [41–43,45–47,70–73];
combination of FO and RO (for draw solution reconcentration) however, most of them have not matured or proven feasible.
will result in even higher rejections of pollutants from the final Very few peer-reviewed publications on FO desalination could
effluent. be found. Kravath and Davis [21] investigated desalination of
Atlantic Ocean seawater by FO using cellulose acetate flat sheet
5.1.5. Forward osmosis for source water and hollow fiber membranes and glucose solution as a draw
purification—hydration bags solution. Kessler and Moody [44] and Moody and Kessler [74]
The use of FO for water purification was briefly introduced modeled and tested similar applications of FO for desalination.
in Section 4. The concept of hydration bags was developed for Their objective was to develop a batch desalination process for
military, recreational, and emergency relief situations when reli- emergency water supply on lifeboats, not as a continuous pro-
able drinking water is scarce or not available [63,69]. Hydration cess for seawater desalination.
bags are one of the few commercial applications of FO. Although In recent bench-scale studies [22,47,48], it was demonstrated
slower than other water purification devices, FO hydration bags that when using a suitable FO membrane (e.g., the FO CTA mem-
require no power and only foul minimally, even when used with brane) and a strong draw solution (highly soluble ammonia and
muddy water. The high selectivity of an FO membrane ensures carbon dioxide gases), seawater can be efficiently desalinated
that in most situations and for most sources of water, the per- with FO. The draw solution was formed by mixing together
meating water is free of microorganisms, most macromolecules, ammonium carbonate and ammonium hydroxide in specific
and most ions. proportions [48]. The salt species formed include ammonium
In the hydration bags, an edible draw solution (e.g., a sugar bicarbonate, ammonium carbonate, and ammonium carbamate.
or beverage powder) is packed in a sealed bag made of a Analysis of the process has shown that an osmotic pressure driv-
semi-permeable FO membrane [63]. Upon immersion of the ing force (π) as high as 238 bar for a feed water with a salt
bag in an aqueous solution, water diffuses into the bag due to concentration of 0.05 M NaCl, and as high as 127 bar for a feed
the osmotic pressure difference and slowly dilutes the initially water with a salt concentration of 2 M NaCl, can be achieved
solid draw solution. At the end of the process the diluted with the ammonia/carbon dioxide draw solution [48]. This is a
draw solution can be consumed as a sweet drink containing rather high driving force considering that 2 M NaCl is equivalent
nutrients and minerals. In this regard, hydration bags represent to brine from seawater desalination at approximately 70% water
an ultimate treatment process; not a pretreatment process. recovery.
For small, personal devices (Fig. 15), the process can take A schematic drawing of the novel ammonia–carbon dioxide
3–4 h to completely hydrate a 12 oz beverage. The extraction FO process is illustrated in Fig. 16. Water is extracted from sea-
water and dilutes the ammonia–carbon dioxide draw solution.
Upon moderate heating (near 60 ◦ C), the draw solution decom-
poses to ammonia and carbon dioxide. Separation of the fresh
product water from the diluted draw solution can be achieved
by several separation methods (e.g., column distillation or mem-
brane distillation (MD)). The degasified solution left behind is
pure product water and the distillate is a reconcentrated draw
solution available for reuse in the FO desalination process.
Bench-scale FO data demonstrates that the ammonia–carbon
dioxide FO process is a viable desalination process. Salt rejec-
tions greater than 95% and fluxes as high as 25 l/m2 h were
achieved with the FO CTA membrane with a calculated driv-
ing force of more than 200 bar [22]. Although this is a relatively
high flux, much greater flux is actually expected for such a high
driving force. Further analysis of the results [48] has indicated
Fig. 15. Illustration of water purification hydration bag [69]. that the performance ratio (defined as experimental water flux
T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87 83
include decreased dosing frequency, more regular drug concen- der. Exit ports can range from simple, straight channels to more
tration in the bloodstream, enhanced bioavailability, improved complicated design configurations [27].
patient compliance, and reduced side effects [25]. For example, When the osmotic pump is brought into contact with an
delivering therapies to specific cells can be an obstacle in treat- aqueous solution or wet environment, water diffuses through
ing neurodegenerative diseases, such as Parkinson’s disease and the membrane into the draw solution compartment. As pressure
Huntington’s disease. Targeting treatment to the brain is diffi- builds up, it expands the draw solution compartment, pushes
cult because the blood–brain barrier blocks many drugs [78]. In the piston, increases the pressure in the drug compartment, and
other cases, a drug can have low oral bioavailability (e.g., it can consequently induces the release of the drug through the ori-
have low solubility or limited permeability) in the gastrointesti- fice. The rate of water diffusion is the most important design
nal tract [25]. aspect of an osmotic pump because it dictates the rate at which
Controlled- or modified-release of drugs is possible through the drug will be released. Theeuwes and Yum [26] summarized
the use of osmotic pumps. Osmosis offers several advantages the principles of design and operation of osmotic pumps and
as a driving force for constant pumping of drugs, including methodically described the design criteria for constant delivery
accurate mass transfer. For example, when considering targeted rate.
treatment, osmotic pumps can deliver medicine directly to the The volumetric flux of water into the draw solution com-
cerebrospinal fluid, where it is ultimately taken up into neurons. partment of osmotic pumps can be described by the simplified
The development of osmotic pumps was pioneered with the Kedem–Katchalsky equation (Eq. (1) in Section 2). When the
commercialization of the ALZET® osmotic pump for animal pump is placed in an environment with an osmotic pressure of
studies in the mid-1970s [26]. More recently, osmotic principles πe , π can be expressed as
have been applied to human therapy, resulting in the develop-
ment of drug delivery systems such as the DUROS® system [27], π = πDS − πe (7)
OROS® Push-PullTM , L-OROSTM , and EnSoTrol® [25] that can where πDS is the osmotic pressure of the draw solution. P can
release drugs continuously for up to 1 year. Recently, osmotic be expressed as
pumps have been studied as regulated systems for the acquisi-
tion, metering, buffering, delivery, and assay of fluid biological P = Pd − Pe (8)
samples [79].
where Pd is the internal pressure difference associated with the
The principal components of a typical osmotic drug-delivery
drug delivery through the orifice and Pe is the pressure nec-
system are illustrated in Fig. 17. The osmotic pump is contained
essary for the piston to compress the drug reservoir. To achieve
in a titanium alloy cylindrical reservoir that is 4 mm in diameter
constant water flux, and hence constant delivery rate, the right-
and 40 mm in length. This reservoir protects the drug molecules
hand side of Eq. (1) must be constant. This condition can be met
from enzymes, body moisture, and cellular components that
only if both terms, σπ and P, are constant (assuming that
might deactivate the drug prior to release. A polyurethane mem-
the membrane permeability coefficient is constant)
brane covers one end of the reservoir. Like other semi-permeable
membranes, it is permeable to water but almost completely σ(πDS − πe ) = constant (9)
impermeable to ions. The osmotic engine (i.e., the draw solu-
tion) occupies a portion of the cylinder behind the membrane. and
The draw solution is most often NaCl and a small amount of Pd − Pe = constant (10)
pharmaceutical excipients in a tablet form. An elastomeric pis-
ton separates the draw solution from the drug formulation in the Eq. (9) can be satisfied if (1) the amount of draw solution in the
drug reservoir. The drug may either be a solution or a suspen- osmotic pump can ensure saturation in the water solution, (2) the
sion and either aqueous or non-aqueous in nature. The drug must osmotic pressure of the environment is negligible, and (3) the
be stable at body temperature (37 ◦ C) for extended periods of reflection coefficient is constant. Implantable osmotic pumps
time, usually from 3 months to 1 year. The drug exit port is a in the human body can easily meet these three requirements
small orifice located at the opposite end of the titanium cylin- [26]. Starting from solid or very concentrated draw solution, and
assuming a small drug reservoir, the draw solution can be kept
saturated throughout the delivery time period and at much higher
osmotic pressure than the osmotic pressure in the human body.
The values of P in Eq. (10) can be reduced to almost zero by
controlling the orifice size and by selecting a highly deformable
partition between the draw solution and drug compartments [26].
Even so, when a low molecular weight salt is used as a draw
solution, π will generally always be much greater than P.
Fig. 17. Cross-section of the implanted DUROS® System [27]. A polyurethane
membrane covers one end of the reservoir. The osmotic engine (i.e., the draw
5.5. Osmotic power—pressure-retarded osmosis
solution) occupies a portion of the cylinder behind the membrane. An elas-
tomeric piston separates the draw solution from the drug formulation in the drug
reservoir. Upon diffusion of water into the osmotic engine, the piston is pushed Renewable energy can be extracted wherever two streams of
and the drug is released through the drug outlet orifice. different salinity or different chemical potential meet. Consid-
T.Y. Cath et al. / Journal of Membrane Science 281 (2006) 70–87 85
6. Concluding remarks
[52] L. Song, M. Elimelech, Theory of concentration polarization in cross- ater recovery process for advanced life support systems, in: Proceedings
flow filtration, J. Chem. Soc., Faraday Trans. 91 (1995) 3389–3398. of the 35th International Conference on Environmental Systems (ICES),
[53] S.S. Sablani, M.F.A. Goosen, R. Al-Belushi, M. Wilf, Concentration Rome, Italy, 2005.
polarization in ultrafiltration and reverse osmosis: a critical review, [68] BiogasWorks, Nutrient utilization and anaerobic digestion, 2000. http://
Desalination 141 (2001) 269–289. www.biogasworks.com/Index/Nutrient%20Utilization%20&%20AD.htm.
[54] M. Elimelech, S. Bhattacharjee, A novel approach for modeling con- [69] HTI, Military/X Pack, 2005. Electronic Source: http://www.
centration polarization in crossflow membrane filtration based on the hydrationtech.com.
equivalence of osmotic pressure model and filtration theory, J. Membr. [70] W.T. Hough, Process for extracting solvent from a solution, US Patent
Sci. 145 (1998) 223–241. 3,532,621 (1970).
[55] M. Mulder, Basic Principles of Membrane Technology, 2nd ed., Kluwer [71] W.T. Hough, Solvent extraction into a comestible solute, US Patent
Academic Publishers, Dordrecht, The Netherlands, 1997. 3,702,820 (1972).
[56] S. Loeb, M.R. Bloch, Countercurrent flow osmotic processes for the [72] J. Yaeli, Method and apparatus for processing liquid solutions of suspen-
production of solutions having a high osmotic pressure, Desalination 13 sions particularly useful in the desalination of saline water, US Patent
(1973) 207–215. 5,098,575 (1992).
[57] G.T. Gray, J.R. McCutcheon, M. Elimelech, Internal concentration polar- [73] K. Lampi, E.G. Beaudry, J. Herron, Forward osmosis pressurized device
ization in forward osmosis: role of membrane orientation, Desalination and process for generating potable water, US Patent 6,849,184 (2005).
197 (2006) 1–8. [74] C.D. Moody, J.O. Kessler, Forward osmosis extractors, Desalination 18
[58] G.D. Mehta, S. Loeb, Performance of Permasep B-9 and B-10 mem- (1976) 283–295.
branes in various osmotic regions and at high osmotic pressures, J. [75] M.D. Rosa, F. Giroux, Osmotic treatment (OT) and problems related to
Membr. Sci. 4 (1979) 335–349. the solution management, J. Food Eng. 49 (2001) 223–236.
[59] S. Loeb, Pressure-retarded osmosis revisited: the prospects for osmotic [76] R.D. Scott, Dehydration and packaging of foodstuff, US Patent
power at the dead sea, in: Proceedings of the Euromembrane ’95 Con- 3,858,499 (1975).
ference, University of Bath, UK, 1995. [77] J.R. Herron, E.G. Beaudry, C.E. Jochums, L.E. Medina, Osmotic con-
[60] I. Goosens, A. Van-Haute, The use of direct osmosis tests as complemen- centration apparatus and method for direct osmosis concentration of fruit
tary experiments to determine the water and salt permeabilities of rein- juices, US Patent 5,281,430 (1994).
forced cellulose acetate membranes, Desalination 26 (1978) 299–308. [78] E. Singer, New technologies deliver in treating neurological diseases—
[61] R.W. Baker, Membrane Technology and Applications, 2nd ed., John RNA interference, osmotic pumps may solve drug delivery problems,
Wiley & Sons, Ltd., New York, NY, 2004. Nat. Med. 10 (2004).
[62] C. Biberdorf, Filter in a pouch, The Warrior (2004) 3. [79] Y.C. Su, L. Lin, A water-powered micro drug delivery system, J. Micro-
[63] Hydration Technologies Inc., Hydration bags—technology overview, electromech. Syst. 13 (2004) 75–82.
Electronic Source: http://www.hydrationtech.com/merchant.mv?Screen= [80] S. Loeb, T. Honda, M. Reali, Comparative mechanical efficiency of
CTGY&Store Code=MHTI&Category Code=TECH-FO. several plant configurations using a pressure-retarded osmosis energy
[64] A. Sagiv, R. Semiat, Backwash of RO spiral wound membranes, Desali- converter, J. Membr. Sci. 51 (1990) 323–335.
nation 179 (2005) 1–9. [81] Energy from the ocean, A report prepared for the Subcommittee on
[65] J. Mallevialle, P.E. Odendaal, M.R. Wiesner, Water Treatment Membrane Advanced Energy Technologies and Energy Conservation Research,
Processes, McGraw-Hill, USA, 1996. Development, and Demonstration of the Committee on Science and
[66] P.O. Wieland, Design for human presence in space: an introduction Technology, U.S. House of Representatives, Ninety-fifth Congress, Sec-
to environmental control and life support systems, NASA George C. ond Session, Prepared by the Congressional Research Service, Science
Marshall Space Flight Center, Alabama, 1994. http://flightprojects.msfc. Policy Research Division, 2002.
nasa.gov/book/rp1324.pdf. [82] S. Loeb, F.V. Hassen, D. Shahaf, Production of energy from concen-
[67] T.Y. Cath, V.D. Adams, A.E. Childress, S.J. Gormly, M.T. Flynn, trated brines by pressure-retarded osmosis. II. Experimental results and
Progress in the development of direct osmotic concentration wastew- projected energy costs, J. Membr. Sci. 1 (1976) 249–269.