Environ Engg and Policy 1 (1998) 103–107 Q Springer-Verlag 1998

Low-pressure reverse osmosis as a process for treatment of anionic

pollutants in water environment
Chavalit Ratanatamskul, Kazuo Yamamoto

103
Abstract The recent development of new generation Ci,m pconcentration at membrane surface
Low-Pressure Reverse Osmosis (LPRO) or nanofiltration bi pconvective coefficient;
membranes have received attraction for application in ci pconcentration of ion inside the membrane;
the field of wastewater and water treatment through an Ci pconcentration of ion in the external solution;
increasingly stringent regulation for drinking purpose zi pvalence of ion;
and water reclamation. In this research, the application fX peffective charge density of membrane;
on treatment of anionic pollutants (NO3 –, NO2 –, Cl –, x pdistance inside the membrane
SO4 2–, HPO4 2– & H2PO4 –) was done under very much
lower pressure range of operation (0.49 to less than 1
0.1 MPa) than any other previous research has done. Introduction
Moreover, the extended Nernst-Planck equation was used Nowadays, it is recognized that environmental problems
for analysis of the experimental data by considering ef- caused by anionic pollutants have been concerned to be a
fective charge density of the membranes. very important topic. Many kinds of anionic pollutants
play important roles in contaminating drinking water and
List of symbols reused water, yielding an unacceptable quality of the wa-
ji pthe flux of the i-th ion through the membrane ter product. The most well-known example of drinking
based on membrane pore area water pollution is a high concentration of nitrate ion in
Dx pmembrane thickness; groundwater which is utilized for drinking purpose in a
CD pDonnan potential at the membrane interface; water-deficient area. Also, the common use of chloramine
C pelectric potential; as a disinfectant to inactivate bacteria activity and as a
Ci,p peffluent concentration of ion in the external solu- way to reduce trihalomethane-formation potential in long
tion; water distribution systems in some countries causes high-
jv pvolume flux based on membrane area; er concentration of nitrite ions.[1] Nitrite ion can affect
ui pmolar mobility; the human health directly, which is different from nitrate
vi pspecific volume; ion that has to change into the form of nitrite before be-
P phydraulic pressure; coming toxic to the human body. A very strict EC guide-
F pFaraday constant; line for nitrite ion is 0.1 mg/l for drinking purpose. In
R pgas constant or rejection; contrast, the WHO guideline set a more flexible level of
T ptemperature; nitrite ion of 3 mg/1, which is easier to be achieved.
ai pactivity; Moreover, the problem of regrowth of bacteria or slime
Ak psurface porosity; formation in the cooling piping system for industrial use
g pactivity coefficients in membrane and in external and in other water distribution systems due to enough
solution; available trace-ions of nitrogen and phosphorus has to be
DP ptransmembrane pressure difference; controlled. The studies done by Herson et al. (1984) 2 and
A ppermeability coefficient; Sathasivan et al. (1995) 3 suggested that lowering the con-
centration of phosphate ion might be a possible way to
prevent or control the regrowth of bacteria in the distri-
bution systems. Furthermore, the biocorrosion of the wa-
Received: 29 January 1998
ter distribution system due to the utilization of sulfate
C. Ratanatamskul ion by the bacterial regrowth also reduces the perform-
Department of Environmental ance and lifetime of industrial materials and then in-
Engineering, Chulalongkorn University, Bangkok 10330 creases the cost of reparation and maintenance of the
Thailand equipment.
e-mail: [email protected] In order to minimize the amount of anionic pollutants
to achieve a high quality of reused water as required
K. Yamamoto
Dept. of Urban Engineering, The University of Tokyo,
through a necessity and through a more stringent regula-
Tokyo, Japan tion, a possibly promising method of the new-generation
LPRO (Low-pressure reverse osmosis) or nanofiltration
Correspondence to: C. Ratanatamskul membranes as thin film composite membranes has been
 Environ Engg and Policy 1 (1998)

investigated in this research under very low pressure

range of operation. Lower pressure operation is very at-
tractive in reducing capital and operation costs and can
provide an easier way for system maintenance. At pres-
ent, little information is known for the operation of the
LPRO membranes in treating diluted aqueous solution
containing nitrite, nitrate and phosphate ions under very
much lower pressure especially less than 0.5 MPa. The
LPRO membranes used in this research are negatively- Fig. 1. Schematic diagram of the experimental set-up: 1pfeed
charged membranes. Therefore, there is a possibility for water, 2ppump, 3pNF membrane module, 4pby-pass, 5preg-
separating ions according to their valences by the electri- ulating valve, 6ppermeate
104 cal repulsion force.
tion. The schematic diagram of the experimental set-up is
2 shown in Fig. 1. The operating pressure and recirculation
Experimental methods rate were adjusted and controlled by means of by-pass
In this research, the experiment has been divided into and regulating valves. The recirculation rate was kept in
two parts. The first part dealt with the study on the re- the range of 1.5–1.7 l/min. The permeate fluxes of mem-
jection performance of anions in a mixed solution by na- branes were measured after one and two hours to ascer-
nofiltration membranes under very low pressure range of tain a pseudo steady-state of operation. The permeate
operation. And the second part related to the experiment samples were collected after two hours of operation. The
using numerical simulation to predict the rejection of concentration of all types of anions in the samples
single monovalent anion. Three types of nanofiltration (HPO4 2 – & H2PO4 –, SO4 2–, Cl –, NO2 – and NO3 –) were
membranes used in this study are listed in Table 1. The analyzed by the indirect UV-detection method of HPLC
membrane module used here is a flatsheet type C-10T (Shimadzu LC-6A). This method utilized IC-Al column.
(Nitto Denko Co.) and the effective membrane surface Phtalic acid of 0.5 mm (pHp4.6) was used as liquid
area of the module is 60 cm 2. For the first part of the re- phase. The detection wavelength of UV is 260 nm.
search, the composition of the feed solution of mixed an-
ions was prepared as shown in Table 2. In this study, a 3
relatively high concentration of anions was initially se- Results and discussion
lected to test the possibility of the rejection performance
for various kinds of anionic pollutants by three types of 3.1
nanofiltration membranes. The pH of the feed solution Flux characteristics of three types of NF membranes
was adjusted to 7.2 by NaOH and the temperature was in From Fig. 2, the fluxes of three types of NF membranes
the range of 24–27.5 7C. For the second part of the re- were investigated under steady state after two hours. It
search, single chloride solution of 35.5 mgCl/l and single was found that the fluxes increase with transmembrane
nitrate solution of 20 mgN/l were prepared with sodium pressure. The membrane type NTR-7250 gave the highest
chloride and potassium nitrate, respectively. The operat- flux of 1.1!10 –5 m/s at the operating pressure of
ing transmembrane pressures used were in the range 0.49 MPa whereas the membrane types NTR-729HF and
from 0.49 to less than 0.1 MPa, which is a considerably NTR-759HR produced the fluxes of 6.35!10 –6 and
low pressure range for nanofiltration membrane opera- 4!10 –6 m/s, respectively under the same operating pres-
sure. General transport equation to predict the flux of
water across RO membrane is JvpA* (Dp–Dp). However,
Table 1. Characterization of three types of nanofiltration mem- in this experiment, osmotic pressure, Dp, was considered
branes
negligible because of dilute solution. The water permea-
Membrane Type Skin Layer-Material Rejection of NaCl* tion coefficient, A, obtained under very low operating

NTR-7250 Polyvinyl alcohol 70%

NTR-729HF Polyvinyl alcohol 93%
NTR-759HR Aromatic polyamide 99.5%

* Catalog value at 1.5 MPa, 0.15% NaCl

Table 2. Composition of the feed solution

Composition Concentration

KH2PO4 10 mgP/l
NaNO2 20 mgN/l
KNO3 20 mgN/l
NaCl 30 mgCl/l
MgSO4 7 H2O 30 mgSO4/l
Fig. 2. Fluxes of three types of NF-membranes
C. Ratanatamskul, K. Yamamoto: Low-pressure reverse osmosis

pressures here are 2.27!10 –5, 1.51!10 –5, 8.24!10 –6 m/s of Cl –, NO2 –, and NO3 – apparently for membrane types
– MPa for membrane types NTR-7250, NTR-729HF and NTR-729HF and NTR-7250. However, the effect of trans-
NTR-759HR, respectively. membrane pressure to the rejection percentage is consid-
erably insignificant in operating with membrane type
3.2 NTR-759HR, which is known as the tightest among these
The performance of anion rejection three types of NF membranes. It has been recognized
Figures 3, 4 and 5 illustrate the performance of the three that multivalent ions can be rejected better than unival-
types of NF membranes in rejecting anions such as ent ions in the case of RO membranes. From this result,
HPO4 2-& H2PO4 –, SO4 2–, Cl –, NO2 –, and NO3 – in the di- it appears that HPO4 2-& H2PO4 – and SO4 2– could be re-
lute solution as function of transmembrane pressure after moved better than Cl –, NO2 –, and NO3 – (univalent ions)
the filtrate time of 120 minutes. The percentage of rejec- with all types of membranes under very low operating
tion increased with transmembrane pressure in the cases pressures. 105
Moreover, the results on monovalent anionic rejection
in the research (in the case of Cl –, NO2 –, and NO3 –) also
suggests that the solute permeate flux seems independent
of the water flux and it can be considered relatively con-
stant as the rejection of these monovalent anions in-
creased with higher operating transmembrane pressure.
From the data obtained in this research, the membrane
type NTR-759HR gave the highest rejection performance
for all types of anionic pollutants with rejection percent-
ages of nearly 100% for HPO4 2-& H2PO4 – and SO4 2–,
89–98% for Cl –, 83–97% for NO3 – and 83–93% for NO2 –,
respectively. The membrane type NTR-729HF yielded the
second best performance of anionic removal as about
100% for HPO4 2-& H2PO4 – and SO4 2–, 43–82% for Cl –,
Fig. 3. Performance of anionic rejection by membrane type 11–60.5% for NO3 – and 11–66.5% for NO2 –, respectively.
NTR-7250 In the case of membrane type NTR-7250 with the highest
flux of permeate resulted in the anionic removal as
89–95% for HPO4 2-& H2PO4 –, 96–100% for SO4 2–,
15.5–58% for Cl –, 0–22% for NO3 – and 0–37% for NO2 –,
respectively.
Due to the obvious sensitivity of rejection perform-
ance on operating transmembrane pressure in the case of
monovalent anions (Cl –, NO2 –, and NO3 –), the prediction
of this phenomena has been done further with the follow-
ing numerical simulation.

4
Numerical simulation for rejection of single anionic
pollutant in dilute solution
Previously, some transport models of charged reverse-os-
Fig. 4. Performance of anionic rejection by membrane type mosis membranes have been proposed and also separa-
NTR-729HF tion dependency upon feed concentration, membrane
charge density and permeate volume flux have been indi-
cated. However, few reports have experimentally com-
pared the experimental data with the theory. Here, the
extended Nernst-Planck model was applied for the analy-
sis of the experimental data of dilute ionic solution under
very low pressure range of operation by considering the
effective charge density of the membranes.

4.1
Transport phenomena equation for charged RO membrane
The extended Nernst-Plank equation proposed by Dresn-
er (1972) 4 is as follows,
jip–ciuiRT7(d(lnai)/dx)
–ziciuiF7(dC/dx)–ciuivi(dP/dx)cbicijv (1)
5
Fig. 5. Performance of anionic rejection by membrane type The third term in Eq. (1) is mostly neglegible.
NTR-759HF The electroneutrality condition is expressed as;
 Environ Engg and Policy 1 (1998)

In external solution:
A zi Cip0 (2)
In membrane:
A zi cicfXp0 (3)
The Donnan equilibrium at an interface between the
membrane and the external solution can be assumed as
(gci/gCi) 1/zipExp (P(F/RT)7DCD) (4)
From (1) to (3);
106
A (zi/ui (ci–Ci,p)7jv)
dC/dxp (5) Fig. 7. The comparison of the model simulation with the ex-
F A (zi 2Ci
perimental data for rejection of chloride ion by membrane type
(ci–Ci,P) jv zici F (dC/dx NTR-729HF
dci/dxp – (6)
uiRT RT
And (jv7Ak)pADP (7)
Rp1–Ci,p/Ci,m (8)
The calculation can be done by numerically integrat-
ing equations from (2) to (8) by Rung-Kutta method.

5
Numerical simulation results and discussion

5.1
Effect of transmembrane pressure on ionic rejectionFi-
gures 6 to 8 show the relationships between rejection and
transmembrane pressure of dilute chloride and nitrate Fig. 8. The comparison of the model simulation with the ex-
solution. The concentration of feed chloride and nitrate perimental data for rejection of nitrate ion by membrane type
ions in this experiment were 35.5 mgCl/l and 20 mgN/l, NTR-729HF
respectively. The comparison of the simulation data with
the experimental data utilizing membrane types NTR-
7250 and NTR-729HF were also shown. Rejection in- were 1.3!10 –4 and 9.94!10 –15 m in the case of mem-
creased with an increase in transmembrane pressure. It is brane types NTR-729HF and NTR-7250, respectively.
apparent that the model could fit the experimental data However, the value of fX/Cm–parameter depended on
very well at transmembrane pressure higher than the type of feed ionic solution. The observed values were
0.1 MPa. However, when transmembrane pressures are 51 and 25 in the cases of membrane types NTR-729HF
less than 0.1 MPa, a small deviation from the model was and NTR-7250, respectively with dilute chloride solution.
observed. Two transport parameters of Dx/Ak (membrane And it was 35 in the case of NTR-729HF with dilute ni-
thickness/surface porosity of the membrane) and fX/Cm trate solution.
(effective charge density of membrane/influent concentra-
tion) were obtained. Since the parameter of x/Ak is com- 5.2
mon for the same membrane, the values obtained here Concentration profiles of ions inside the membrane
In this study, sodium chloride of 1 mmol/l was selected
as a model for anionic pollutants. Concentration profiles
of sodium and chloride ions across the membrane were
numerically simulated. It was found that the concentra-
tion of chloride ion significantly decreased with the dis-
tance inside the membrane. In the case of NTR-7250, it is
apparent that when much lower pressure operation of
0.067 MPa or less was employed as shown in Fig. 9, more
thickness of membrane might be required to reach the
same reduction as in the case of higher pressure opera-
tion. From Figs. 9 & 10, the approximately 1,100 times
higher concentration of sodium ion than that of chloride
ion inside the membrane is caused by the electric poten-
Fig. 6. The comparison of the model simulation with the ex- tial occurred inside the membrane. The electric potential
perimental data for rejection of chloride ion by membrane type here tends to repulse chloride ion, a co-ion to membrane
NTR-7250 surface charge, from entering inside the membrane. In
C. Ratanatamskul, K. Yamamoto: Low-pressure reverse osmosis

6
Conclusions
1. In the application of nanofiltration membranes under
very low pressure operation (0.49 to less than
0.1 MPa), rejection efficiency of more than 90 percent
for all kinds of anionic pollutants could be achieved
with the membrane type of NTR-759HR at transmem-
brane pressure at least 0.196 MPa. Therefore, this tech-
nology can be considered as a promising method in
reducing the cost for operation and system-mainte-
nance, when compared with the traditional reverse os-
mosis that much higher operating pressure always has 107
to be applied to obtain the same rejection efficiency.
Fig. 9. The Concentration gradient of chloride ion across mem- 2. The extended Nernst-Planck equations could fit the
brane type NTR-7250 experimental data well, however, a small deviation
from the prediction by the equations at transmem-
brane pressure less than 0.1 MPa was observed. The
reason for this deviation is still under investigation.
3. This membrane application under very low pressure
can be considered as a clean technology since it helps
to reduce environmental impact and health risks
caused by anionic pollutants, whereas operating cost
and energy consumption can also be minimized.

References
Short report (1995) J Wat SRT-Aqua 43 : 315–322
Herson DS et al. (1984) Proc Wat Qual Technol Conf, Denver,
p 309
Sathasivan et al. (1995) Proc Asian Regional Conf on water
quality and pollution control, Philippines, pp 587–594
Dresner L (1972) Desalin 10 : 17–26
Fig. 10. The Concentration gradient of sodium ion across mem- Tsuru T et al. (1991) J Chem Eng Jap 24 : 511–524
brane type NTR-7250

contrast, the electric potential results in accumulating so-

dium ion, which is a counter-ion to membrane-surface
charge, inside the membrane. Moreover, increasing trans-
membrane pressure causes an increase in electric poten-
tial. Therefore, it resulted in higher accumulation of sodi-
um ion and higher reduction of chloride ion inside the
membrane as shown in Figs. 9 and 10.