Desalination 247 (2009) 33–44

Prediction of boron transport through seawater reverse

osmosis membranes using solution–diffusion model

Phan Vu Xuan Hung, Seung-Hee Cho, Seung-Hyeon Moon*

Department of Environmental Science and Engineering
Gwangju Institute of Science and Technology
Tel. +82-62-970-2435; Fax: +82-62-970-2434; email: [email protected]
Received 13 November 2008; revised 17 December 2008; accepted 24 December 2008

Abstract
One of the key transport parameter—permeability, is being used as an intrinsic characteristic of mem-
brane for reverse osmosis process prediction. In this study, the prediction of reverse osmosis process in
terms of boron rejection was carried out in two steps. First, the permeability of boron existing as boric acid
B(OH)3 and as metaborate ion B(OH)4− through seawater reverse osmosis membranes was estimated in three
types of feed matrix: DI water, NaCl and artificial seawater. The second step was to develop a computer
program using boric and borate permeabilities as inputs to predict the boron rejection at different operating
conditions. The governing equations were derived from the solution–diffusion model. The effect of tempera-
ture was quantitatively determined employing a representative constant and incorporated with the program.
The prediction showed good agreement with experimental data obtained from a lab-scale plate-frame type
reverse osmosis system.
Keywords: Boron removal; Reverse osmosis membranes; Desalination; Solution–diffusion model; Prediction

1. Introduction have high rejection for most of solutes in sea-

In recent years, reverse osmosis (RO) tech- water, the removal of some micro toxicants
nology has been widely applied to desalinate including boron is not effective [2]. The prob-
seawater as well as brackish waters from saline lem of high boron concentration was observed
aquifers and rivers [1]. Though RO membranes after commissioning a seawater reverse osmo-
sis (SWRO) desalination plant in Eliat, Israel,
in 1997. Farmers using the post-treated seawa-
ter for irrigation noticed poisoning of crops
*Corresponding author. and partly discolored leaves due to the toxicity

Presented at the 2nd joint workshop between the Center for Seawater Desalination Plant and the European
Desalination Society, Gwangju Institute of Science and Technology, Korea, October 8–9, 2008.

0011-9164/09/$– See front matter © 2009 Published by Elsevier B.V.

doi:10.1016/j.desal.2008.12.010
34 P.V.X. Hung et al. / Desalination 247 (2009) 33–44

of boron [3]. According to the World Health H

Organization (WHO) regulation, the limit of O

RO Membrane
boron concentration in permeate of SWRO B
desalination plants is 0.5 mg/L [4]. This req-
uirement could not be accomplished by con-
ventional configuration of SWRO desalination
plants. Therefore, several modifications and
post-treatment methods have been developed
to enhance boron rejection in SWRO desalina-
tion plants [5–9].
However, the investigation of boron trans-
port mechanism and boron transport parame-
ters was not effective for the actual RO system H2O
since the fraction of boric acid and borate
was not considered separately. Therefore, in Fig. 1. The different existence of boric acid and
borate causes the difference in rejection ability of RO
this study, the transport of boron in RO mem- membranes.
branes was predicted in terms of individual
permeabilities of boric acid and metaborate
ion using solution–diffusion model, and the rejection of boron by RO membrane because
predicted results were compared to actual boric acid molecule is very small, uncharged
experimental results. and non-polar. The average radius of boric
acid is around 2.1 Å, just double the size of
water [10] making it able to form bridges of
hydrogen with the active groups of the mem-
2. Background
brane and diffuse in a similar way to that of
2.1. Boron chemistry in seawater water [12]. On the other hand, the rejection
Boron chemistry in aqueous solution pro- of borate by RO membrane is much more
vide critical principles to investigate the boron effective because of its negative charge and
transport mechanism and improve the boron hydrated layer (Fig. 1).
rejection. Boron exists in seawater and in
other water bodies almost as boric acid and 2.2. Solution–diffusion model
its conjugate base, metaborate ion. Boric acid
is a very weak monobasic acid and acts exclu- The solution–diffusion model has emerged
sively by hydroxyl ion (OH−) acceptance rather over the past 20 years as the most widely
than proton (H+) donation [10]: accepted explanation of transport in dialy-
sis, reverse osmosis, gas permeation and per-
vaporation [13]. The appropriate interpretation
B(OH)3 + 2H2 O ⇔ B(OH)4− + H3 O + of transport mechanism and simplification in
mathematical expressions made it popular in
In seawater, pKa of the boric acid ranges numerous research papers and especially mem-
from 8.4 to 9.2 depending on salinity and tem- brane and desalination textbooks [7,14–18].
perature [11]. In addition, pH of seawater nor- The mass transport through reverse osmo-
mally varies from 7.5 to 8.5 that make boric sis membrane was well rationalized by the
acid predominant in seawater. This caused low solution–diffusion model, in which permeants
 P.V.X. Hung et al. / Desalination 247 (2009) 33–44 35

dissolve in the membrane material and then at very different extents as mentioned above.
diffuse through the membrane down a chemi- Consequently, boric permeability and borate
cal potential gradient. The permeants are permeability should be simultaneously esti-
separated because of the differences in their mated from experimental data, which was the
diffusion rates through RO membranes [14]. fundamental hypothesis of this study. (For
This approach along with other appropri- convenience, the permeability of boron exist-
ate assumptions, the governing equations of ing as boric acid B(OH)3 and as metaborate
solution–diffusion model applied to RO proc- ion B(OH)4− was referred as boric permeability
ess were derived elsewhere [13]. Two most sig- and borate permeability with mass unit was
nificant derived equations are the equations mg-B equivalent.)
of water transport (Eq. 1) and salt transport
(Eq. 2) using the concept of water permeabil-
ity and solute permeability as intrinsic proper- 3. Materials and methods
ties of membranes. 3.1. Material and apparatus
Jw = Wp ( ΔP − Δπ ) (1) All chemicals used are pure grade chemi-
cals. NaCl, MgCl2.6H2O, CaCl2.2H2O, KCl
were provided by DC Chemicals (Korea);
J s = S p ΔC (2)
MgSO4 was provided by Junsei (Japan); and
where Jw is the volumetric flux [m/d]; Js is the Boric acid was provided by Sigma-Aldrich
gravimetric solute flux [kg/m2/d]; Wp is the (USA). Artificial seawater (ASW) used in
water permeability [m/d/Pa]; Sp is the solute experiments contained around 19174 mg/l chl-
permeability [m/d]; ΔP and Δπ are the hydrau- oride, 10031 mg/l sodium, 2442 mg/l sulfate,
lic pressure difference and osmotic pressure 1468 mg/l magnesium, 491 mg/l calcium and
difference between two sides of membranes 393 mg/l potassium (34000 mg/l total dissolved
[Pa] and ΔC is the solute concentration differ- solids) which represented the average compo-
ence between two sides of membranes. sition of seawater. Boric acid was spiked in
the feed solution at 5 mg/l as boron. Two com-
mercialized polyamide thin-film composite
SWRO membranes—SHN and SR, obtained
2.3. Boric and borate permeability from Woongjin Chemicals (Korea) were used
In the study of boron transport parame- in this study. Detailed specifications of these
ters, Sagiv [19] and Taniguchi [20] used boron two membranes at standard test condition for
permeability as the representative for both spiral-would type (800 psi, 25°C, 32,000 mg/l
boric and borate. That permeability was not NaCl feed solution, 8% recovery) are summa-
an intrinsic property of membranes because rized in Table 1 according to the vendor.
at least it depended on feed pH. Hyung [21] The lab-scale membrane unit (OSMON-
did a thorough study of boron transport ICS) was used for all experiments. The mem-
parameters of SWRO membranes, also esti- brane element was rectangular shape, 14 cm
mated boron permeability first and used it to in length and 9.5 cm in width and effective
estimate boric and borate permeability. How- area around 125 cm2. The feed spacer, mem-
ever, two species of boron— boric and borate brane and permeate spacer were sandwiched
could be considered as two different species in between two stainless steel plates and sealed
because they are rejected by RO membranes with two silicon rubber rings. The membrane
36 P.V.X. Hung et al. / Desalination 247 (2009) 33–44

Table 1
Speciation of SR and SHN membranes

Stabilized salt
Flowrate (m3/d) rejection Boron rejectionb Description

RE8040a-SR 22.7 99.6% 90% High rejection RO membrane

element for seawater and high
salinity well water
RE8040-SHN 22.7 99.75% 92% Ultra-high rejection RO membrane
element for seawater and high
salinity well water
a
Element with 8-inch outer diameter and 40-inch long.
b
At pH 8 and 5 mg/l boron feed with the standard test condition.

was fixed very tightly in test cell by pressuriz- conductivity of reservoir and permeate were
ing at 900 psi by hydraulic hand pump (P19, recorded every 20 minutes during 1 hour exp-
SPX Corporation, MN, USA). Feed solu- eriment. In addition, reservoir and permeate
tion stored in a 5L reservoir was circulated samples were also taken to analyze boron con-
and pressurized by a high-pressure pump. centration. Boron concentration was analyzed
Since both concentrate and permeate were by inductively coupled plasma–mass spectros-
circulated to the reservoir and the volume of copy method (ICP-MS) [22].
permeate samples was less than 1% of total
feed volume, feed concentration change was
neglected reasonably. Feed temperature was 3.2. Boric permeability and borate
controlled and maintained by a cooling cir- permeability estimation
culating bath (J-LTB, JISICO, Korea). Reten- In this study, a new hypothesis was appl-
tate pressure was monitored with a pressure ied in which boron flux was considered as
gauge (WIKA) located on the retentate line. the sum of independent boric flux and borate
Retentate flowrate was measured using a flux:
hydraulic flow meter (Dwyer, Mich., USA).
Permeate flowrate was measured using Opti- J boron = J boric + J borate = Sboric ΔCboric
flow 1000 Digital flow meter (Humonics Inc., (3)
CA, USA). + Sborate ΔCborate
In this study, the effects of pH, tempera-
ture, salinity, and boron concentration were where Jboron, Jboric, Jborate are the fluxes of
considered. In pH change experiments, which boron, boric and borate [kg-B/m2/d], respec-
were used to estimate boric and borate per- tively; Sboric and Sborate are boric and borate
meability, pH was adjusted by 1M of NaOH permeability [m/d], respectively; ΔCboric and
and 1M of HCl solution. Temperature change ΔCborate are boric and borate concentration
experiments were conducted to test the eff- difference between two sides of membranes
ect of temperature and investigate the rela- [kg/m3], respectively. The series of three
tionship of permeability and temperature. values (Jboron, ΔCboric, ΔCborate) were obtained
For one data point, permeate flowrate, pH, from experiments and used as inputs for a
 P.V.X. Hung et al. / Desalination 247 (2009) 33–44 37

nonlinear regression method in which the calculate temperature correction factor (TCF)
plane Jboron = SboricΔCboric + Sborate ΔCborate was used for adjusting salt permeability at differ-
optimized. From the regression plane, boric ent temperatures:
permeability Sboric and borate permeability
Sborate were determined. Boric and borate
concentration in feed and permeate were
S p (T ) ⎛ C C⎞
TCF = = exp ⎜ − ⎟ (7)
computed from the well-known buffer Eq. S p (298.15) ⎝ 298.15 T ⎠
(4) and mass balance Eq. (5).
or in alternative form:
Cborate
pH = pK a + log (4)
ln(TCF ) =
C
−
C (8)
Cboric 298.15 T
Cboric + Cborate = Cboron (5)
where TCF is the temperature correction
factor; Sp(T) and Sp(298.15) are the perme-
where pKa is the dissociation constant of boric abilities at temperature T and 298.15K [m/
acid; Cboron, C boric and Cborate are the boron, d]; C is so called temperature effect constant
boric and borate concentration in either side (TEC) [K], characteristic of membrane bar-
of membrane [kg-B/m3]. The dissociation rier material and a type of permeant. With
constant pKa of boric acid depends on salin- the determination of permeability at different
ity and temperature according to the follow- temperatures by the above method, TCF was
ing empirical Eq. [11]: obtained and fitted to Eq. (8) to estimate the
TEC for boric and borate transport. The TEC
ln( K a ) = ( −8966.90 − 2890.53S
1/ 2
− 77.942S was then incorporated in prediction program
3/2 2
to recalculate the permeability at varying tem-
+ 1.728S − 0.0996S ) / T peratures.
1/ 2
+ (148.0248 + 137.1942S + 1.62142S )
1/ 2
+ ( −24.4344 − 25.085S − 0.2474S ) ln(T )
1/ 2 3.4. FORTRAN program for prediction
+ ( 0.053105S )T (6 )
The equations and algorithm [7] used for
programming is shown in Fig. 2.
where Ka is the acid constant of boric acid; S
Practically, the permeate flowrate also
and T are the salinity [kg-TDS/m3] and tem-
can be obtained from the similar calcula-
perature [K] of solution.
tion by using water permeability. However,
this study strongly focused on the hypoth-
esis of boric and borate permeability esti-
3.3. Temperature effect mation and used it to confirm the prediction
The effect of temperature on performance of boron rejection, so the permeate flowrate
of a reverse osmosis process is explained by obtained from experiments was inputted to
numerous literatures [7,19,21]. Feed tempera- the program. The program was simple and
ture affects the diffusion rate of permeants convenient to incorporate with the whole
through RO membranes. According to Wilf program to predict the performance of a rev-
[7], the following equation was applied to erse osmosis process.
38 P.V.X. Hung et al. / Desalination 247 (2009) 33–44

In put: Sboric, Sborate,

Cf, Qp, A, pH, pKa Adjust Sboric, Compute Cp, R,
Assume: Cp=0, Sborate by
AFC=Cf TCF

Start New AFC

New Cp: Cp’

ΔC = AFC − C p

J boron = Sboric ΔCboric + Sborate ΔCborate

J boron Qp C p' − C p
< 10−3
NO
Cp = R= Cp
Jw Qf
YE
Cf ⎛ 1 ⎞ S
AFC = Ln⎜ ⎟
R ⎝1− R ⎠ Final
results
printed out

Fig. 2. FORTRAN calculation program.

4. Results and discussion regression. The constant variance test results

displays whether or not the data passed or
4.1. Permeability estimation failed the test of the assumption that the vari-
Data obtained from pH change experi- ance of the dependent variables in the source
ments of three types of feed solution were population is constant regardless of the value
fitted to equation (3) to estimate boric perme- of the independent variables. The power, or
ability and borate permeability. The success of sensitivity, of a regression is the probability
nonlinear regression performed by SigmaPlot that the model correctly describes the relation-
is decided by four critically statistical param- ship of the variables, if there is a relationship.
eters. The coefficient of determination, R2, is For a power test, alpha (α), the acceptable
measure of how well the regression model probability of incorrectly concluding that the
describes the data. The normality test results model is correct, was chosen at 0.05 in this
display whether the data passed or failed the study. The regression results were summarized
test of the assumption that the source popu- in Tables 2 and 3 for SR and SHN membrane,
lation is normally distributed around the respectively. Those results suggested that

Table 2
Nonlinear regression results of SR membrane

Experiment Boron in DI water Boron in NaCl 32000 mg/l Boron in artificial seawater
Statistical parameters R2 0.983 0.893 0.863
Normality test Passed Passed Passed
Constant variance test Passed Passed Passed
Power (with α = 0.05) 0.997 0.871 0.815
Boric permeability (m/d) 0.8372 0.3942 0.2221
Borate permeability (m/d) 0.0376 0.0399 0.0131
 P.V.X. Hung et al. / Desalination 247 (2009) 33–44 39

SR membrane, DI water feed SHN membrane, ASW feed

0.003 1e-3
Boron flux (kg/m2/d)

Boron flux (kg/m2/d)

0.002 8e-4

0.001 6e-4

0.000 3e-3
4e-4 4e-3

)
/m 3
2e-3

)
/m 3
3e-3

(kg
2e-4

(kg
5e-3 1e-3
3e-3

ric
4e-3

ric
3e-3

bo
ΔC 2e-3 2e-3
2e-3 ΔC

ΔC

bo
1e-3

ΔC
borate (kg/m 3
)
borate (kg/m 3 1e-3
)

Regression plane
Regression plane
Experimental data
Experimental data

Fig. 3. Representative regression planes for SR and SHN membranes.

boric permeability was approximately seven 4.2. Temperature effect

to twenty times higher than borate permeabil- The determination of TEC for SHN mem-
ity for SR and SHN membranes at different brane according to Eq. (8) was demonstrated
condition. Representative regression planes in Fig. 4. Experiments were conducted with
were demonstrated in Fig. 3. ASW feed solution at four points of tem-
The satisfaction of four statistical param- perature 20oC, 25oC, 30oC and 35°C; TCF
eters indicated that the experimental data well was calculated by Eq. (7). TECs of boric and
fitted Eq. (3). Therefore, boric and borate borate of SHN membranes obtained from
permeabilities were appropriately estimated. regression results were 4,717.5 and 4,361.2,
The boric and borate permeability tended to respectively. TECs of boric and borate of
decrease when ionic strength of feed solution SR membranes obtained from regression
increased from DI water to NaCl and ASW. results were 3,565.4 and 3,549.3, respectively.
The explanations were still unclear but it was The equivalence in magnitude of boric and
assumed that the existence of Na+ and espe- borate TEC in both membranes implied
cially divalent actions Ca2+ and Mg2+ in ASW that the effect of temperature on boric and
might contribute to the higher rejection since borate transport might be similar; and those
they can form the complex with certain portion was higher than TEC of water proposed at
of boric and borate. In terms of comparing 2500–3000 in [7]. It implied that the boron
SR and SHN, it is obvious that SHN is better flux increased faster than water flux as tem-
for boron rejection because of lower permea- perature increased; and boron rejection
bility in three types of feed solution (only one decreased consequently, consistent with both
exception for borate in ASW). That coincided experimental and predicted results.
with the test result of the vendor (Table 1).
Even though membrane characterization was
not carried out in this study, it is thought that
the difference in permeability originated from 4.3. Prediction by FORTRAN program
the difference in surface properties and pore- The success of permeability and TEC
size distribution of two membranes. determination provided intrinsic parameters
40 P.V.X. Hung et al. / Desalination 247 (2009) 33–44

0.6
0.6
Boric-SHN
Borate-SHN
Boric-SR
Borate-SR
0.4
0.4

ln(TCF)
0.2 0.2

ln(TCF)
0.0 0.0

-0.2 -0.2

-0.4 -0.4
0.00320 0.00324 0.00328 0.00332 0.00336 0.00340 0.00344 0.00320 0.00324 0.00328 0.00332 0.00336 0.00340 0.00344

1/T (K -1 ) 1/T (K -1 )

Fig. 4. Temperature effect constant determination for SHN and SR membranes.

100.0
100.0
Boron in DI water Boron in NaCl solution

90.0 90.0
Boron rejection (%)

Boron rejection (%)

80.0 80.0

70.0 70.0

60.0 60.0
Experiment
Experiment
P redict ion P redict ion
50.0 50.0
6.0 7.0 8.0 9.0 10.0 11.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0
pH pH

100.0 100.0
Boron in ASW Boron in ASW, pH = 8.1− 8.3

90.0 90.0
Boron rejection (%)
Boron rejection (%)

80.0 80.0

70.0 70.0

60.0 60.0
Experiment Experiment
P redict ion P redict ion
50.0 50.0
7.0 7.5 8.0 8.5 9.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0
pH Tem perature (o C)

Fig. 5. Prediction of boron rejection of SR membrane.

 P.V.X. Hung et al. / Desalination 247 (2009) 33–44 41

100.0 100.0
Boron in DI wat er Boron in NaCl solut ion

90.0 90.0
Boron rejection (%)

Boron rejection (%)

80.0 80.0

70.0 70.0

60.0 60.0 Experiment

Experiment
P redict ion P redict ion
50.0 50.0
6.0 7.0 8.0 9.0 10.0 11.0 12.0 6.0 7.0 8.0 9.0 10.0 11.0
pH pH

100.0 100.0
Boron in ASW Boron in ASW , pH = 8.1− 8.3

90.0 90.0
Boron rejection (%)

Boron rejection (%)

80.0 80.0

70.0 70.0

60.0 60.0
Experiment Experiment
P redict ion P redict ion
50.0 50.0
7.0 7.5 8.0 8.5 9.0 9.5 10.0 15.0 20.0 25.0 30.0 35.0 40.0
pH Tem perature ( oC )

Fig. 6. Prediction of boron rejection of SHN membrane.

in order to input to the FORTRAN program the results of temperature change experiment
in predicting the boron rejection at different was also shown at the bottom left graph of
operating conditions. First, the results from Fig. 5 and 6. For both experimental and pre-
pH change experiments, which already used to dicted, the boron rejection tended to decline
estimate the permeability were compared with as temperature increased. It confirmed that
predicted data and presented in Figs. 5 and 6 boric permeability and borate permeability
for SR and SHN membrane, respectively. Both increased faster than water permeability did
experimental and predicted data show that via temperature. For all cases, the predicted
boron rejection increases as feed pH increases data matched well with experimental data
due to the increase of borate fraction. The (relative error almost fell in the range of 1–
sharp change was also observed at the adja- 3%); so the hypothesis of boric and borate
cent of pKa of boric acid. The comparison of permeability estimation was supported; the
42 P.V.X. Hung et al. / Desalination 247 (2009) 33–44

100.0 100.0
Experiment wit h boric
90.0 90.0 Experiment wit h borat e
Boron rejection (%)

Boron rejection (%)

80.0 80.0

Experiment
70.0 P redict ion 70.0

60.0 60.0
Experiment
P redict ion
50.0 50.0
4.5 5.0 5.5 6.0 6 .5 4.0 5.0 6.0 7.0
Feed boron conc. (mg/l) Feed boron conc. (mg/l)

Fig. 7. Prediction of boron rejection of SR membrane.

feasibility of Eq. (7) for TCF and the accuracy feed concentration was observed. Figure 8
of FORTRAN program were also confirmed. demonstrated the experimental and predicted
In this study, three more experiments data of the experiment conducted with SHN
were conducted to compare experimental membrane in artificial seawater with the var-
data and predicted data. The first one was iation of salinity. In this case, the predicted
performed with SR membrane and DI water data at a low feed salinity were higher than
feed solution containing only boric acid at experimental data. It implies that boric and
pH 6.5. The second one was similar to the borate permeabilities might increase as salin-
first one but all the boric acid content was ity decreases, the similar observation when
converted to borate by increasing pH to comparing boric and borate permeabilities
11.0. The results shown in Fig. 7 indicate in DI water, NaCl and artificial seawater
that the predicted data still did not go far (Table 3). The variation of boric and borate
from the experimental data. In addition, the via salinity permeability should be studied
insignificant variation of boron rejection via further to unravel these gaps.

Table 3
Nonlinear regression results of SHN membrane

Experiment Boron in DI water Boron in NaCl 32000 mg/l Boron in artificial seawater
2
Statistical parameters R 0.924 0.980 0.995
Normality test Passed Passed Passed
Constant variance test Passed Passed Passed
Power (with α = 0.05) 0.975 0.996 0.997
Boric permeability (m/d) 0.3704 0.3505 0.1986
Borate permeability (m/d) 0.0548 0.0356 0.0244
 P.V.X. Hung et al. / Desalination 247 (2009) 33–44 43

100 Nomenclature
Boron in art ificial seawat er
A Membrane area, m2
90 AFC Average feed concentration, kg/m3
Cborate Borate concentration (as boron),
Boron rejection (%)

80 kg (B)/m3
Cboric Boric concentration (as boron),
70
kg (B)/m3
Cboron Boron concentration (determined
by ICP-MS), kg (B)/m3
60
Experiment
Jborate Boron flux as borate, kg (B)/m2/d
P redict ion Jboric Boron flux as boric acid,
50 kg (B)/m2/d
10.0 20.0 30.0 40.0 50.0 Jboron Boron flux, kg (B)/m2/d
Feed salinity (g/l)
Js Salt flux, kg/m2/d
Fig. 8. Prediction of boron rejection of SHN Jw Water volume flux, m3/m2/d
membrane. or m/d
NDP Net driving pressure, Pa
5. Conclusions or kg/m/s2
Qf Feed flowrate, m3/d
The prediction of boron transport through Qp Permeate flowrate, m3/d
SWRO membranes was conducted with SR S Salinity , g/L or kg/m3
and SHN membrane using a lab-scale plate- Sp Salt permeability, m/d
frame RO system. The effects of pH and tem- T Temperature, K
perature on boron rejection were found to be Wp Water permeability, m/s/Pa or
consistent with previous study. Further, boric m2s/kg
and borate permeabilities were investigated sep- ΔCborate Borate concentration (as boron)
arately and simultaneously using a simplified difference between two sides of
solution–diffusion model and nonlinear regres- membranes, kg (B)/m3
sion. Temperature effect was also quantitatively ΔCboric Boric acid concentration (as
determined through an empirical temperature boron) difference between two
effect constant. The prediction by FORTRAN sides of membranes, kg (B)/m3
program well matched with experimental obser- ΔP Pressure difference between two
vation. This study can be easily incorporated sides of membrane, Pa or
in the complete prediction of RO process per- kg/m/s2
formance and applied to a pilot or large-scale Δπ Osmotic pressure difference
system with some correction factors. between two sides of
membrane, Pa or kg/m/s2
Acknowledgements
This research was partly supported by a References
grant (code#C106A1520001-06A085600222) [1] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G.
from Plant Technology Advancement Georgiadis, B.J. Marinas and A.M. Mayes,. Sci-
Program funded by Ministry of Construction ence and technology for water purification in the
& Transportation of Korean government. coming decades, Nature, 452 (2008) 301–310.
44 P.V.X. Hung et al. / Desalination 247 (2009) 33–44

[2] Y. Magara, T. Aizawa, S. Kunikane, M. Itoh, the elimination of boron by means of reverse
M. Kohki, M. Kawasaki and H. Takeuti, The osmosis, Desalination, 140 (2001) 145–152.
behavior of inorganic constituents and disin- [13] J.G. Wijmans and R.W. Baker, The solution-dif-
fection by products in reverse osmosis water fusion model: a review, J. Membr. Sci., 107 (1995
desalination process, Water Science and Tech- ) 1–21.
nology, 34 (1996) 141–148. [14] R.W. Baker, Membrane technology and appli-
[3] C. Fritzmann, J. Lowenberg, T. Wintgens and cations, John Wiley & Sons Ltd, West Sussex,
T. Melin, State-of-the-art of reverse osmosis des- 2004.
alination, Desalination, 216 (2007) 1. [15] S.K. Gupta, Analytical design equations for
[4] Boron in Drinking water, WHO, 2003. reverse osmosis systems, Industrial and Engi-
[5] J. Redondo, M. Busch and J. P. De Witte, Boron neering Chemistry Process Design and Develop-
removal from seawater using FILMTEC (TM) ment, 24 (1985) 1240–1244.
high rejection SWRO membranes, Desalination, [16] R. Rautenbach and F.P. Helmus, Some consid-
156(200X) 229–238. erations on mass-transfer resistances in solution-
[6] Y. Magara, A. Tabata, M. Kohki, M. Kawasaki diffusion-type membrane processes, J. Membr.
and M. Hirose, Development of boron reduc- Sci., 87 (1994) 171–180.
tion system for seawater desalination, Desalina- [17] N.M. Al-Bastaki and A. Abbas, Predicting the
tion, 118 (1998) 25. performance of RO membranes, Desalination.
[7] M. Wilf, The guidebook to membrane desalina- 132 (2000) 181–187.
tion technology, Balaban Desalination Publica- [18] J. Mallevialle, P. Odendaal and M.R. Wiesner,
tions, L’Aquila, 2007. Water treatment membrane process, McGraw-
[8] M. Taniguchi, Y. Fusaoka, T. Nishikawa and Hill, Lyonnaise des Eaux, 1996.
M. Kurihara, Boron removal in RO seawater [19] A. Sagiv and R. Semiat, Analysis of parameters
desalination, Desalination, 167 (2004) 419–426. affecting boron permeation through reverse
[9] N. Nadav, Boron removal from seawater reverse osmosis membranes, J. Membr. Sci., 243 (2004)
osmosis permeate utilizing selective ion exchange 79–87.
resin, Desalination, 124 (1999) 131–135. [20] M. Taniguchi, M. Kurihara and S. Kimura,
[10] A. Earnshaw and N. Greenwood, Chemistry of Boron reduction performance of reverse osmo-
the elements, Butterworth-Heinemann, 1997. sis seawater desalination process, J. Membr. Sci.,
[11] A.G. Dickson, Thermodynamics of the disso- 183 (2001) 259–267.
ciation of boric acid in synthetic seawater from [21] H. Hyung and J.-H. Kim, A mechanistic study
273.15 to 318.15 K. Deep Sea Research Part on boron rejection by sea water reverse osmosis
A Oceanographic Research Papers. 37 (1990) membranes, J. Membr. Sci., 286 (2006) 269–278.
755–766. [22] R.N. Sah and P.H. Brown, Boron determination
[12] M. Rodriguez Pastor, A. Ferrandiz Ruiz, M.F. - A review of analytical methods, Microchemical
Chillon and D. Prats Rico, Influence of pH in Journal, 56 (1997) 285–304.