PROCESS DESIGN FOR DROWNING-OUT

CRYSTALLIZATION OF LITHIUM HYDROXIDE

MONOHYDRATE

M. E. Taboada1,
, T. A. Graber1, L. A. Cisternas1, Y. S. Cheng2 and K. M. Ng2

1
Departamento de Ingenieria Quimica, Universidad de Antofagasta, Antofagasta, Chile.
2
Department of Chemical Engineering, The Hong Kong University of Science and Technology,
Clear Water Bay, Hong Kong.

Abstract: A drowning-out crystallization process was designed to produce purified lithium

oxide monohydrate. The study started with the development of the ternary phase diagram for
the LiOH – water –ethanol system. This was followed by process conceptualization integrating
crystallization, downstream processing and solvent recovery. LiOH . H2O crystals with a mean
size between 0.3 mm and 0.4 mm depending on the operating conditions and compositions
were obtained.
Keywords: lithium hydroxide; drowning-out crystallization; process design.

INTRODUCTION hydroxide solution has inverted solubility and

Lithium salts can be obtained from the Salar
de Atacama brines in the north of Chile
where the mining of inorganic salts is of
global significance. These include KCl,
K2SO4, LiCO3, NaNO3, Na2SO4, KNO3,
H3BO3 among others. Lithium hydroxide is
usually produced by reacting lithium carbon-
ate and calcium hydroxide in aqueous sol-
ution. It is then recovered as a monohydrate


Correspondence to: salt which is used in the production of lubricat-
Dr M.E. Taboada, ing greases, inks and Li battery (Kirk and
Departmento de Ingenieria
Othmer, 2000). The conventional method to
Quimica, Universidad de
Antofagasta, Casilla 170,
recover lithium hydroxide monohydrate is
Antofagasta, Chile. evaporative crystallization since its solubility
E-mail: mtaboada@uantof.cl does not change much with temperature.
However, the energy required for evaporative
DOI: 10.1205/cherd06251
crystallization is high. Also, it is difficult to con-
0263–8762/07/ trol the particle size distribution of the final
$30.00 þ 0.00 product by manipulating the evaporation rate
of the solvent. Process alternatives are there-
Chemical Engineering fore highly desirable. For example, a recent
Research and Design
study suggests membrane electrolysis as
Trans IChemE, the purification method for lithium hydroxide
Part A, September 2007 from reactor effluent (Ryabtsev et al., 2004).
The objective of this research was to design
# 2007 Institution
a drowning-out crystallization process as a
of Chemical Engineers
process alternative for producing lithium
hydroxide (Cisternas et al., 2006; Berry
et al., 1997; Rajagopal et al., 1991). This
study started with an investigation of the
phase behaviour of the LiOH –water –ethanol
system at different temperatures varying from
288.15 to 308.15 K. An aqueous lithium
a small variation in solubility with temperature
(Taboada et al., 2005a, b). Ethanol was
chosen as the drowning out agent in part
due to its miscibility with water. The drowning
out crystallization process using ethanol as
drowning out agent was then designed. Par-
ticle size distribution of the crystals resulting
from different compositions of drowning-out
agents was measured to determine its effect
on the downstream filter costs.
SLE PHASE DIAGRAM
The design of crystallization processes is
governed by the thermodynamic behaviour
of the components present. The SLE data of
lithium hydroxide –water –ethanol system at
several temperatures were collected exper-
imentally. The materials used in the exper-
iments were lithium hydroxide monohydrate
(99% pure), absolute ethanol (99.5% pure),
and distilled water with conductivity of
0.05 mS cm21. The solubility experiments
were carried out using a rotary water tempera-
ture bath with a holder containing eight 90 mL
glass flasks. Various ternary mixtures, in
excess of lithium hydroxide monohydrate,
were prepared. They were stirred at
298.15 K for 72 h to reach equilibrium and
allowed to settle for another 24 h. The compo-
sition of the saturated liquid was analysed
after filtering out the solids. The solubility
results are presented on a phase diagram
(Figure 1) and can be represented by the

1325 Vol 85 (A9) 1325– 1330

1326 TABOADA et al.
Figure 1. Ternary isothermal phase diagram of LiOH þ ethanol þ H2O
system at T ¼ 298.15 K.
following empirical model. Here, S is the solubility of lithium
hydroxide in wt% and E is ethanol concentration in wt%,
where 60% , E , 100%.
Ln½S ¼ 2:408925 þ 2:0613(E) (1)
At 298.15 K, the solubility of lithium hydroxide in water is
11 wt%, point (1) in Figure 1. With addition of ethanol, the
solubility decreases gradually. It drops to 1.91 wt% when
about 60% of ethanol is present in the mixture, revealing
that ethanol might be a potential drowning out agent. Lithium
hydroxide monohydrate has a mass ratio of lithium hydroxide :
water of 24 : 18, which can be normalized to 57% : 43%. There
may exist a double saturation point for LiOH and LiOH . H2O
at point D at this temperature. The dotted line connecting point
D to approximately 57% LiOH–43% H2O (corresponding to
LiOH . H2O) forms a compartment (L þ S1). Another dotted
line connecting point D and 100% LiOH divides the area into
two compartments (L þ S1 þ S2) and (L þ S2). Here L, S1
and S2 denote the liquid and the two solid phases,
LiOH . H2O and LiOH, respectively. In the (L þ S1) compart-
ment, only LiOH . H2O will crystallize. In the (L þ S1 þ S2) com-
partment, the liquid, the hydrated and the anhydrous salts are at
equilibrium. Both anhydrous and monohydrated lithium hydrox-
ides may be present for the compositions in this region. In the
(L þ S2) compartment, only LiOH will crystallize. However,
this region is expected to be very thin. In this case, point D
almost coincides with the ethanol vertex and the L þ S2
region becomes the C2H5OH–LiOH edge. This study will only
focus on the L þ S1 compartment where the drowning out crys-
tallization process will take place. To confirm the identity of S1 in
the (L þ S1) compartment, precipitation crystallization exper-
iments were carried out. Aqueous ethanol solution, point (2) in
Figure 1, was added to the saturated lithium hydroxide solution
and solids were precipitated out. Three mixtures with compo-
sitions along the line connecting points (1) and (2) were obtained
at ethanol mass fractions of 0.24, 0.31, and 0.38 as shown in
Figure 1. X-ray diffraction experiments confirmed that the
solids at equilibrium from the above three compositions were
indeed LiOH . H2O. The three blue dotted lines in Figure 1 are
the corresponding tie-lines. To the best of our knowledge,
there is no solvate between lithium hydroxide and ethanol.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A9): 1325–1330
Temperature Effect on Solubility of Lithium
Hydroxide
Next, the temperature effect on the solubility of lithium in
water and in water þ ethanol (0.1 ethanol/water mass frac-
tion) systems were studied, respectively. The solubility data
at 288.15, 293.15, 298.15, 303.15 and 308.15 K were
obtained using the procedure described above. The results
are shown in Figure 2. It is clear that the solubility of lithium
only slightly depends on temperature at all in the tested
range. Drowning crystallization was thus proposed as an
alternative process for lithium hydroxide purification.
Based on the above results, an empirical correlation of the
solubility of lithium hydroxide, S (%wt of solute), the concen-
tration of ethanol, E (%wt of ethanol), and temperature, T (K)
was developed:
S ¼ (1:0178  0:0065T )  (1:4402
 0:045T ) eE=40:118 (2)
CONCEPTUAL DESIGN
A drowning-out crystallization process to purify lithium
hydroxide using ethanol as drowning-out agent was designed
according to the heuristics and systematic procedure devel-
oped by Berry et al. (1997). Based on thermodynamics
phase behaviour of lithium hydroxide –water –ethanol
system, the simplest Class I separation process was syn-
thesized. The base case involves the adding of drowning
out agent to selectively crystallize lithium hydroxide mono-
hydrate. It is followed by downstream system as well as the
solvent recovery system. The process flow diagram is
shown in Figure 3(a) while Figure 3(b) shows the process
paths on the LiOH –water –ethanol ternary phase diagram.
The stream numbers on the phase diagram correspond to
those on the flow sheet.
The process would produce 6.26 ton h21 of pure lithium
hydroxide monohydrate with ethanol as the antisolvent.
6.34 ton h21 of impure lithium hydroxide monohydrate
crystals are fed to the dissolver. Together with the recycle
stream 5, stream 3 has a flow rate of 10.71 ton h21 of lithium
hydroxide. The dissolver effluent (stream 3) is then filtered to
remove insoluble particles, which are assumed to be the only
Figure 2. Solubility of lithium hydroxide in binary system (†) and tern-
ary system (ethanol þ water) (B) at 288.15, 293.15, 298.15, 303.15
and 308.15 K.
 PROCESS DESIGN FOR DROWNING-OUT CRYSTALLIZATION 1327

goes to downstream process which consists of a filter and a

dryer. The slurry (stream 8) is separated into a wet solid
stream (stream 9) and a filtrate stream (stream 13) by filtration.
Stream 12 is mixed with stream 13 and fed to the distillation
column for solvent recovery. The stream 5 of 77.13 ton h21,
containing 9.23 wt% dissolved lithium hydroxide, is recycled
to the dissolver.
The distillate, which contains 90 wt% ethanol (stream 15),
is cooled and mixed with the make up stream and fed back
to the crystallizer. Finally, the pure lithium hydroxide mono-
hydrate is recovered after drying. A computer code (HSC
chemistry for windows) was used to calculate the enthalpy
data listed in Table 1. The energy requirement in the distilla-
tion step is 8.1 Gcal h21. Energy integration was not
attempted in this study. Leaving out the auxiliary units, the
drowning-out crystallization process consists of a dissolver,
a crystallizer, two filters, a dryer and a distillation column.

Downstream Unit Operations

Figure 3. (a) A schematic process flow diagram for the purification of
LiOH by drowning out crystallization using ethanol as the drowning
out agent and (b) the process paths depicted on an isothermal
phase diagram at 258C.
impurity in the system. A make up stream of pure ethanol
(stream 17) is mixed with stream 16 and added to the salt sol-
ution (stream 6) to induce the precipitation of LiOH . H2O in
the crystallizer. The crystallizer has two outlets: one solvent
outlet (stream 12) and one slurry outlet (stream 8). Stream 8
The crystal size distribution affects cake porosity and per-
meability which in turn impact the load of the downstream fil-
tration system (Wibowo et al., 2001). The PSD in the MSMPR
depends on a combination of crystallization kinetics and crys-
tallizer residence time (Jones, 2002). In the case of drowning
out crystallization, flowrate of drowning out agents relates to
the crystallizer residence time. It alters the rate of generation
of supersaturation that results in different PSD. Therefore,
crystallization experiments were carried out to investigate
the effect of various flowrate and compositions of aqueous
ethanol on PSD and filter cost.
Two drowning out agents, 70% and 90% ethanol in water,
were added to the saturated lithium hydroxide solution at
different flowrates. First, 993.75 g of saturated lithium hydrox-
ide solution, which contains 11 wt% lithium hydroxide, was
prepared in a 2L jacketed crystallization vessel. This vessel
is equipped with a helical stirrer that maintains the stirring
speed of 500 rpm throughout the experiment. The tempera-
ture of the mixture was controlled by the external water

Table 1. Material and energy balances for the ethanol process.

w/w%
Enthalpy, Q
Stream T, 8C Ton h21 LiOH . H2O(s) LiOH(aq) C2H6O H2O Gcal h21

1 25.00 6.34 99.25 0.00 0.00 0.00 28.30

2 25.00 0.35 0.00 0.00 0.00 100.00 1.33
2́ 78.50 0.35 0.00 0.00 0.00 100.00 1.31
3 97.79 57.05 0.00 14.24 0.00 85.68 222.78
4 40 0.05 0.00 1.42 0.00 8.58 0.14
5 101.20 50.35 0.00 9.00 0.00 91.00 193.10
6 40.00 57.00 0.00 14.25 0.00 85.75 225.74
7 73.64 16.16 0.00 0.00 90.00 10.00 20.17
8 42.86 9.58 65.00 2.54 6.09 26.37 39.59
9 38.00 6.73 92.59 0.54 1.29 5.58 29.76
10 68.00 0.46 0.00 0.00 18.91 81.09 1.29
11 68.00 6.27 99.36 0.58 0.00 0.00 28.11
12 42.86 90.48 0.00 7.26 17.40 75.35 206.50
13 38.00 2.85 0.00 7.26 17.40 75.35 9.93
14 42.86 62.43 0.00 7.26 17.40 75.35 216.04
15 78.50 12.08 0.00 0.00 89.93 10.07 20.59
16 78.50 12.08 0.00 0.00 89.93 10.07 19.97
17 25 0.09 0.00 0.00 100.00 0 0.11

(s): Crystallized monohydrated lithium hydroxide, (aq): aqueous lithium hydroxide.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A9): 1325– 1330
1328 TABOADA et al.

circulation bath. Next, 400 g of a solution of 90 : 10 ethanol : -

water was added to the system at a fixed flowrate to induce
precipitation. Three flowrates of 7, 30 and 60 mL min21
were tested. Stirring was maintained for 15 min and the sol-
ution was allowed to settle for another 15 min. The crystals
were then collected by vacuum filtration and air-dried at
323.15 K in a rotary system to avoid agglomeration. The
temperature was kept low to avoid evaporation of water of
hydration in LiOH . H2O. The same experiment was repeated
using a 70 wt% ethanol as drowning out solvent, and a 949 g
saturated lithium hydroxide solution. Each experiment was
carried out in duplicate.
The dry crystals were then sieved into 13 size ranges and
weighed: , 75, 75 –90, 90 –106, 106 –125, 125–150, 150 –
212, 212 –250, 250–300, 300 –425, 425–600, 600 –850,
850 –1000 and .1000 mm. The cumulative PSD of monohy-
drated lithium hydroxide crystals produced from 70% and
90% ethanol drowning out solvents are shown in
Figure 4(a) and (b), respectively. The crystal mean sizes
are found within 0.31 mm and 0.43 mm for both cases. Cor-
relations of the PSD curves were obtained by applying the
exponential type formula given by Drzymala and Hussin
(2005), and the models of Rosin– Rammler and Gates –
Gaudin –Schuhmann (Macias-Garcı́a et al., 2004) as shown
in Equations (3), (4) and (5) respectively.

(100i  (100L)i )h
Cumulativepercentage ¼ (hi1)
(3)
100"

i #
L
Cumulativepercentage ¼ 1  exp  (4)
h

i
L
Cumulativepercentage ¼ (5)
h

L is the mesh size in millimeters, h and i are parameters

Figure 4. Cumulative particle size distribution of LiOH . H2O crystals (different for each model) obtained by fitting the experimental
from drowning-out crystallization processes using (a) ethanol at 70% data. Table 2 presents the experimental conditions and the
and (b) 90% in water as drowning-out solvent at different flowrates. corresponding parameters and Table 3 presents the cake

Table 2. Parameters for equations (3)–(5).

Drowning out solvent 70% ethanol

Flowrate mL min21 7 30 60

Models DH RR GGS DH RR GGS DH RR GGS

h 9.70 0.45 0.87 21.63 0.38 0.84 28.60 0.38 0.85

i 2.92 3.03 1.14 3.18 3.22 1.02 3.50 3.54 1.04
R2 0.9962 0.9957 0.9026 0.9877 0.9880 0.8565 0.9994 0.9995 0.8617
Standard error 0.0245 0.0258 0.1221 0.0456 0.0450 0.1555 0.0109 0.0103 0.1524
Drowning out solvent 90% ethanol

Flowrate mL min21 7 30 60

Models DH RR GGS DH RR GGS DH RR GGS

H 20.88 0.39 0.85 77.70 0.34 0.83 12.35 0.47 0.89

i 3.22 3.27 1.03 3.99 4.01 0.95 3.41 3.52 1.24
R2 0.9996 0.9996 0.8728 0.9978 0.9979 0.8391 0.9993 0.9991 0.9035
Standard error 0.0084 0.0076 0.1442 0.0202 0.0200 0.1662 0.0106 0.0116 0.1210

DH ¼ Drzymala and Hussin; RR ¼ Rosin–Rammler and GGS ¼ Gates– Gaudin–Schuhmann.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A9): 1325–1330
 PROCESS DESIGN FOR DROWNING-OUT CRYSTALLIZATION 1329

Table 3. Cake properties for crystals obtained from each experiment.

Drowning out solvent 70% ethanol 90% ethanol

Flowrate, mL min21 7 30 60 7 30 60

L, mm 0.34 0.33 0.40 0.43 0.31 0.35

Porosity 0.38 0.38 0.37 0.38 0.37 0.38
Permeability, m22 1.02e210 9.16e211 8.10e211 8.79e211 5.96e211 1.31e210
Filter area, m2 0.34 0.34 0.35 0.35 0.36 0.34

Table 4. Equations for calculation of filter area and cake properties.

Parameters Equations

Filter area: Darcy’s law: Q DP (6)

¼
A m(x=k þ Rm t)

Permeability modified Blake– Kozeny equation (7)

2
1 13 M2 1 13
k¼  2
 ¼   D2p
180 (1  1) M1 180 (1  1)2
P
m
3
Di fi
i¼1
Cake porosity 1 ¼1P
m mP (8)
(Di kDl)3 H(Di kDl)fi þ(1=n) ½(Di þkDl)3 (Di kDl)3 H(Di kDl)
i¼1 i¼1

where
P
m
(Di þkDl)2 (1(3=8(kDl=)Di þkDl))fi
4(781o )kDl Pm
i¼1
n¼1þ 13
i¼1
½D3i (Di kDl)3 H(Di kDl)fi

Cost of filter C ¼ exp[11.27 2 1.3408(ln A) þ 0.0709 (ln A)2] (9)

10 , A , 1500 ft2

properties, such as porosity and permeability. The filter areas was obtained when 90% ethanol was used as drowning out
and the cake properties are calculated using the Darcy’s law agent at 30 mL min21. Under the microscope, agglomeration
[equation (6)], modified Blake–Kozeny equation [equation was observed on 20-mesh sieve for all the tests. The most
(7)] (MacDonald et al., 1991), where the porosity is from significant agglomeration was observed with 90% ethanol
the correlation developed by Ouchiyama and Tanaka and for flow rate of 60 mL min21.
(1984) [equation (8)]. The relevant equations are listed in We assume that a rotary vacuum filter operating under
Table 4. There is not a clear trend for the effect on particle vacuum at DP ¼ 25 000 Pa is used in the process for the
size against the flow rate of drowning agent. For the case filtrate of 2.85 ton h21. If the filter resistance, Rm, equals
of 70% ethanol, it seems that changing the drowning out 1  1010 m21, a filter area of 0.36 m2 costing about USD
flow rate did not give significant effects on the PSD. The 60 000 is required. More detailed cake properties and the
effect was more obvious when 90% ethanol was used. A parameters for the filter are summarized in Table 5. Since a
filter cake with the lowest permeability of 5.96  10211 m22 two-outlet crystallizer is used in this process, the majority of
the mother liquor is separated prior to the filter. The slurry
of a high solid-to-liquid ratio (stream 8) therefore minimizes
Table 5. Cake properties and filter parameters. the filter area. The purchase price of the rotary vacuum
filter could be estimated using the correlations in equation
Cake properties (9) and was corrected to year 2006 with the M&S Index of
Mean particle size, Dp (m) 4.32  1024
1241.1 as quoted in Chemical Engineering.
Porosity, 1 0.38
Permeability, k (m2) 5.96  10211
Solid density (kg m23) 1510
Cake thickness (m) 0.08 Solvent Recovery System
Filtrate properties Due to the presence of drowning out solvent, extra effort
Filtrate flow rate, Q (m3 s21) 7.92  1024 and cost for the solvent recovery system are inevitable.
Viscosity, m (kg ms21) 1.0  1023 The dimensions and cost of the distillation column were esti-
Liquid density (kg m3) 817.97 mated based on the Fenske –Underwood –Gilliand method
Filter parameters and the procedure suggested by Peters et al. (2003). The cal-
Pressure difference, DP (Pa) 25 000 culated column dimensions and the purchase price are listed
Filter resistance (m21) 1  1010 in Table 6. If tray columns with bubble-cap trays made of
Filter area, A (m2) 0.36
Purchase price (US$) 94 000 stainless steel 304 were used, the purchase prices of them
would be about US$ 292 000.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A9): 1325– 1330
1330 TABOADA et al.

Table 6. Dimensions and purchase price of the distillation column. REFERENCES

Distillation column (US$) 292 000a
Column diameter, m 3.6
No. of trays
Column height, m 13.35
a
The price is updated to year 2005 using the Marshall & Swift
equipment index of 2005 ( ¼ 1241.4).
CONCLUSIONS
Drowning-out crystallization is proposed as a process
alternative for the purification of lithium hydroxide mono-
hydrate. It is designed based on the thermodynamic beha-
viour of the LiOH –water –ethanol ternary system and the
properties of the filter cake. In addition, a distillation column
is included to recover and recycle the drowning out agent.
Similar design issues are expected for crystallization pro-
cesses of organics, chiral compounds, polymorphs, among
others. Also, no attempt was made to perform energy inte-
gration, optimization of filtration and crystallization oper-
ations, and so on. Further efforts in these directions are
underway.
NOMENCLATURE
A the filtration area, m2
Di the particle size in interval i, m
kDl the mean particle diameter, m
E mass percentage of ethanol (wt%)
fi the number fraction of particles in interval i
h parameter of PSD models
H the Heaviside function
i parameter of PSD models
K permeability, m2 s21
L mesh size, mm
M1 the first moment of the particle size distribution
M2 the second moment of the particle size distribution
DP the pressure difference across the filter
Q the volumetric flowrate of filtrate in, m3 s21
Rm specific resistance of the filter, m21
S solubility of lithium hydroxide, wt%
T temperature, K
x cake thickness, m
Greek symbols
1 cake porosity
m viscosity of mother liquor, kg m21 s21
1o the average porosity of uniformly sized spheres
Berry, D.A., Dye, S.R. and Ng, K.M., 1997, Synthesis of drowning-out
crystallization-based separations, AIChE J, 43(1): 91–103.
22 Cisternas, L.A., Vasquez, C.M. and Swaney, R.E., 2006, On the
design of crystallization-based separation processes: Review and
extension, AIChE J, 52(5): 1754– 1769.
Drzymala, J. and Hussin, A., 2005, Mathematical equations for
approximation of separation results using the Fuerstenau up-
grading curves, International Journal of Mineral Proccessing, 76:
55–65.
Jones, A.G., 2002, Crystallization Process Systems (Butterworth-
Heinemann, Oxford).
Kirk, R.E. and Othmer, D.F., 2000, Kirk-Othmer Encyclopedia of
Chemical Technology, Vol 15 (Wiley, New York).
MacDonald, M.J., Chu, C.F., Guilloit, P.P. and Ng, K.M., 1991, A gen-
eralized blake-kozeny equation for multisized spherical particles,
AIChE J, 37(10): 1583– 1588.
Macı́as-Garcı́a, A., Cuerda-Correa, E.M. and Dı́az-Dı́ez, M.A., 2004,
Application of the Rosin–Rammler and Gates–Gaudin–Schu-
hmann models to the particle size distribution analysis of agglom-
erated cork, Materials Characterization, 52: 159– 164.
Ouchiyama, N. and Tanaka, T., 1984, Porosity estimation for random
packings of spherical particles, Ind Eng Chem Fundam, 23: 490.
Peters, M.S., Timmerhaus, K.D. and West, R.E., 2003, Plant Design
and Economics for Chemical Engineers, 5th edition (McGraw Hill,
London).
Rajagopal, S., Ng, K.M. and Douglas, J.M., 1991, Design and econ-
omic trade-offs of extractive crystallization processes, AIChE J,
37(3): 437 –447.
Ryabtsev, A.D., Nemkov, N.M., Kotsupalo, N.P. and Serikova, L.A.,
2004, Preparation of high-purity lithium hydroxide monohydrate
from technical-grade lithium carbonate by membrane electrolysis,
Russian Journal of Applied Chemistry, 77(7): 1108–1116.
Taboada, M.E., Galleguillos, H.R., Graber, T.A. and Álvarez-Benedı́, J.,
2005a, Density, viscosity, refractive index and electrical conductivity
of saturated solutions of the lithium hydroxide þ ethanol þ water
system at 298.15 K, and thermodynamic description of the solid–
liquid equilibrium, Fluid Phase Equilib, 235: 104–111.
Taboada, M.E., Véliz, D.M., Galleguillos, H.R. and Graber, T.A.,
2005b, Solubility, density, viscosity, electrical conductivity, and
refractive index of saturated solutions of lithium hydroxide in
water þ ethanol, J Chem Eng Data, 50: 187– 190.
Wibowo, C., Chang, W.C. and Ng, K.M., 2001, Design of integrated
crystallization systems, AIChE J, 47(11): 2474–2492.
ACKNOWLEDGEMENTS
The authors wish to thank CONICYT for financing this study (FON-
DECYT projects # 1020892, # 1040299 and CICITEM). Additional
support from the Research Grants Council (HKUST602703) is also
gratefully acknowledged.
The manuscript was received 27 December 2006 and accepted for
publication after revision 26 March 2007.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A9): 1325–1330