Research Article

Received: 21 July 2008 Revised: 8 October 2008 Accepted: 6 November 2008 Published online in Wiley Interscience: 30 December 2008

(www.interscience.wiley.com) DOI 10.1002/jctb.2106

Purification of glycerol/water solutions from

biodiesel synthesis by ion exchange: sodium
removal Part I
Manuel Carmona,a José L. Valverde,a Angel Pérez,a Jolanta Warcholb and
Juan F. Rodrigueza∗

Abstract
BACKGROUND: In this study, equilibrium and kinetic data of the ion exchange of sodium from glycerol–water mixtures on the
strong acid resin Amberlite-252 were obtained. Basic parameters for the design of ion exchange units for the purification of the
crude glycerol phase from biodiesel production have been determined.

RESULTS: Equilibrium uptake of sodium ions with the strong acid ion exchanger Amberlite-252 was studied at three
temperatures. The Langmuir equation and the mass action law model were used to fit the experimental equilibrium data.
Equilibrium constants and thermodynamic parameters were obtained at each temperature. Kinetic experiments were carried
out to evaluate the effective diffusion coefficients of sodium on the resin Amberlite-252 in glycerine–water media.

CONCLUSIONS: Equilibrium results indicate that this process is favourable and also that its selectivity decreases with temperature
increase from 303 to 333 K. Both models were able to fit the experimental equilibrium data. Kinetic experiments showed that
the rate of mass transfer in this binary system is high. An Arrhenius type equation allowed the correlation of effective diffusion
coefficients and temperature. The results indicate that the macroporous resin Amberlite-252 could be useful for removal of
sodium ions from glycerine/water solutions with a high salt concentration.

c 2008 Society of Chemical Industry

Keywords: glycerol; ion exchange; adsorption; biodiesel

INTRODUCTION where R, R and R are hydrocarbon chains with 12 to 22 carbon

Today, the environmental impact of fossil fuels is receiving much atoms.
attention because their combustion releases millions of tonnes of The alkali NaOH and methanol, because of their low prices,
CO2 into the atmosphere each year, which is widely believed to are frequently used as catalyst and reactant, respectively, in this
be contributing to global warming.1 In addition, the dependence process.2 As can be seen, in the forward reaction to produce
of humans on this kind of energy together with the continuous the biodiesel fuel, glycerol is inevitably obtained as a by-product.
changes in the price of petroleum has promoted research into low Biodiesel manufacture yields 10 wt % of glycerol. The process is
cost alternative energy sources, which are environmental friendly. normally carried out at 60 ◦ C. At the end of biodiesel synthesis
One of these alternative energies is obtained by the combustion the reaction product is separated into two phases, biodiesel and
of methyl esters of fatty acids, which are known as biodiesel. glycerol, which settles at the bottom of the reaction vessel.3
Biodiesel is made from the transesterification of vegetable oils The glycerol stream containing 50 wt % glycerol leaving the
and animal fats with methanol or another alcohol in the presence separator contains methanol, catalyst and soap. This stream is a
of an acidic, alkali or enzymatic catalyst to produce alkyl esters low cost product, although its disposal is quite difficult.1 Refining
(Equation (1)). of glycerol started with an acid treatment to split the soaps into
free fatty acid and salts. Fatty acids are not soluble in glycerol
and are separated from the top and recycled to the process; then,

∗ Correspondence to: Juan F. Rodriguez, Department of Chemical Engineering,

University of Castilla – La Mancha, Avda. de Camilo José Cela s/n, 13004 Ciudad
Real, Spain. E-mail: juan.rromero@uclm.es

a Department of Chemical Engineering, University of Castilla – La Mancha, Avda.

de Camilo José Cela s/n, 13004 Ciudad Real, Spain

b Department of Water Purification and Protection, Rzeszów University of

738

(1) Technology, 6 Powstańców Warszawy Str., 35-959 Rzeszów, Poland

J Chem Technol Biotechnol 2009; 84: 738–744 www.soci.org

c 2008 Society of Chemical Industry
Purification of glycerol/water solutions from biodiesel synthesis www.soci.org

methanol is recovered by vacuum distillation. At this point, the

Table 1. Properties of Amberlite 25216
salts remain with the glycerol, which is in aqueous solution and
its purity is about 85 wt %.1 – 4 Further glycerol refining is required Particle Resin capacity Apparent
before it can be used as a raw material for the manufacture of size, for sodium, Surface Void density,
∞
Active D n (mol area, fraction ρR
other products. This purification process allows a product to be
group (m) kg−1 dry resin) (m2 kg−1 ) (%) (kg m−3 )
obtained with purity up to 99.5–99.7% using vacuum distillation
to eliminate water, followed by ion exchange for salt removal.1 The Sulfonic 4.94 × 10−4 4.83 2.487 7.2 1.323
rapid growth of biodiesel production motivated by government
incentives is resulting in a worldwide surplus of glycerol that
depresses its price. Although glycerol has more than 1000 uses,
hydroxide PA grade quality were supplied by Panreac (Barcelona,
including many applications as an ingredient or processing aid
Spain). Demineralized water with conductivity lower than 5 µS was
in cosmetics, explosives, toiletries, personal care, drugs and food
used to prepare the synthetic glycerol solutions with 10 wt % of
products, it has become a low-value byproduct or a waste with
water simulating the conventional solutions obtained in biodiesel
an attached disposal cost. Additionally, the above applications
production processes.
required a highly purified product and crude glycerol is suitable
only for mixing with feed for animals.5
The valorization of the crude glycerol, while avoiding the ap- Ion exchange resin
plication of expensive purification processes, will allow increased Amberlite 252 (Rohm and Haas Co. (Barcelona, Spain)), a
profitability of the biodiesel plants. One promising and econom- macroreticular sulfonated polystyrene-divinylbenzene resin, was
ical alternative is the transformation of glycerol into high value selected for sodium removal from glycerine/water mixtures
products such as (mono- and di-fatty acids) by glycerolysis with because of its excellent behaviour in the uptake of alkali metal
methyl esters at temperatures between 200 and 210 ◦ C.6 Crude ions, such as K and Cs, from non-aqueous media.15,16 The main
glycerol may also be used as a source of renewable energy once a physical properties are given in Table 1.
decrease in water and salts content is achieved. The resin was pre-treated and converted to the H+ -form by
Methods such as ion exchange,7 adsorption8 and membrane repeated treatment in a column with 1.0 mol L−1 NaOH and
technologies9,10 have been applied for the elimination of sodium 1.0 mol L−1 of HCl solutions. Then it was thoroughly rinsed with
chloride from aqueous media. Regarding organic solutions, there demineralized water.17,18
is a lack of information in the literature about the appropriate
technology that should be applied. Ion exchange seems to be the Analytical methods
most suitable method for small scale applications to treat liquids The sodium content of solutions was analysed by atomic ab-
with a very low salt content because of its simplicity, effectiveness, sorption spectrophotometry with a Varian (Madrid, Spain) 220 AS
selectivity, recovery and relatively low cost.9 – 11 spectrophotometer. The standard uncertainty and reproducibility
As discussed above, impurities contained in the glycerol are of measurements was found to be ±0.1%.
water (about 10 wt %) and NaCl (concentration close to 1 mol L−1 ).
A typical ion exchange process involves two steps: loading and Equilibrium experiments
regeneration. For sodium removal, the stream containing the
The experimental setup consisted of seven 250 cm3 Pyrex con-
salt is passing through a fixed bed of resin filled with a strong
tainers hermetically sealed and mechanically agitated, submerged
acid cationic exchange resin. Sodium ions are exchanged with
in a temperature-controlled thermostatic bath kept constant at
protons initially on the resin until the resin exchange capacity
303, 318 and 333 ± 0.1 K. Different known masses of resin, in the
is exhausted. Then, the spent resin is regenerated by using
H+ -form, were put in contact with 100 cm3 of glycerine/water
a concentrated acid solution.12 – 14 Thus, it is clear that for an
(90/10 by weight) containing a total concentration of 1 mol L−1 .
optimized design of ion exchange fixed bed, one must be able
The synthetic mixture was prepared by mixing different amounts
to predict both the loading and regeneration processes. Thus,
of a 1 mol L−1 solution of NaCl with solutions containing HCl
modelling of the ion exchange equilibrium is an a priori step for
1 mol L−1 to vary the initial sodium content and therefore, the
successful optimization of the design and operating conditions of
amount of ion exchanger required for the experiments. The ac-
a fixed bed ion exchange column or for its scale-up. Nevertheless,
curacy of resin weighting was ±0.0001 g. The suspension formed
there is a lack of fundamental information about basic parameters
by the resin and solution was vigorously agitated at 200 rpm by
for the design of such ion exchange systems, especially when an
means of a multipoint magnetic stirrer. To ensure that equilibrium
organic media is considered.
was reached the experiments were left stirring overnight. At the
The aim of this work is to obtain ion exchange equilibrium data
end of this period, the mixtures were filtered to remove the ion
of sodium on the strong-acid cation resin Amberlite 252 in H+
exchange resin. Then, the filtrate was analysed for sodium content
form at different temperatures. The basic equilibrium parameters
as described above. The resin phase concentration in equilibrium
of the Na+ /H+ system were obtained by applying the Langmuir
with the liquid phase was obtained by means of the following
equation and the mass action law. Also, the diffusion coefficients
mass balance equation:
of sodium in the resin were calculated at the same temperatures.
(C0 − C ∗ ) · V
q∗ = · 103 (2)
W
EXPERIMENTAL SECTION
Chemicals where C0 and C ∗ are the initial and equilibrium concentration of
Glycerine (C3 H8 O3 ) with a water content of 2 wt % was sodium in the liquid phase (mol L−1 ), respectively; q∗ denotes
supplied by Sigma-Aldrich (Madrid, Spain) with a purity the resin phase equilibrium concentration of sodium (mmol g−1
739

>97%. Sodium chloride, hydrochloric acid (37 wt %), and sodium of dry resin), V and W are the initial volume of the synthetic

J Chem Technol Biotechnol 2009; 84: 738–744

c 2008 Society of Chemical Industry www.interscience.wiley.com/jctb
 www.soci.org M Carmona et al.

glycerine/water solutions (L) and the weight of dry ion exchange 5

resin (g), respectively.
4
Batch kinetics studies
Effective diffusion coefficients were studied by measuring the rate

q*(mmol g-1)
of sodium uptake from the liquid phase by the resin in a well-mixed 3
tank. The reactor was hermetically sealed and equipped with four
baffles, a standard turbine stirrer with six flat blades, a conductivity
2 Experimental data
cell, and a heating system formed by a temperature-controlled T = 303 K
T = 318 K
thermostatic bath. The effective diffusion coefficient was evaluated T = 333 K
at 303, 318 and 333 K. A high agitation rate was applied to 1 Theoretical curves
T = 303 K
ensure that the film mass transfer resistance was negligible. In T = 318 K
this way, internal diffusion within the resin particle was the only T = 333 K
0
controlling step. In a typical experiment, pre-wetted resin (6 g) 0.0 0.2 0.4 0.6 0.8
was placed in contact with a synthetic mixture of glycerine/water
C*(mol L-1)
(90/10 w/w) (0.5 L) containing a total concentration 0.1 mol L−1 of
sodium chloride. The mixture was stirred for 2 h and the solution Figure 1. Ion exchange isotherms of sodium ions on Amberlite 252 in
conductivity was monitoring continuously and recorded. glycerine/water mixtures at 303, 318 and 333 K at a total concentration of
1 mol L−1 . Langmuir treatment.

RESULTS AND DISCUSSION

Although it seemed clear that ion exchange could be a suitable Table 2. Equilibrium parameters of the system H+ /Na+ for the
technique to remove sodium ions from a liquid phase, it was sodium removal with Amberlite 252 by the Langmuir model
important to know the available resin capacity of Amberlite 252
T KLang qT (mmol g−1
when a glycerin phase is used, and the influence of temperature System (K) (L mol−1 ) dry resin) R2
on the ion exchange process equilibrium. To fit the experimental
data, Langmuir and mass action law models were used, both 303 8.49
being widely used to describe the equilibrium of adsorption or ion H+ /Na+ 318 6.47 4.87 0.987
exchange, respectively. 333 5.68
m 


exp
Equilibrium qObs − qTheor
Obs
It is well known that the Langmuir model describes satisfactorily R2 = 1 − i=1 exp
, where qObs and m are the average

m
 exp 
adsorption processes.19 Nevertheless, it can be applied empirically exp
qObs − qObs
to predict ion exchange processes.20 The Langmuir isotherm i=1
(Equation (3)), contains the parameters qT and KLang , the total experimental concentration in the solid phase and the total number of
useful capacity of the ion exchanger and the Langmuir equilibrium experimental data, respectively.
constant, respectively:
KLang qT C ∗
q∗ = (3) of the system. Water seems to be the only solvating agent for the
(1 + KLang C ∗ ) ion exchange sites and determines the behaviour of the system. In
Figure 1 shows the equilibrium isotherms at 303, 318, and 333 K this system, as in other water/solvent mixtures, water is preferred
of sodium ions in glycerin/water solutions on Amberlite 252. The by the resin over glycerol. Therefore, the solvent/water molecular
curves have been drawn using the parameters obtained by fitting ratio inside the resin differs from that of the bulk solution, thus
the experimental data to Equation (3). the loaded and entering ions are not surrounded by the same
As expected, sodium uptake is favoured at lower temperature. mixture. It is clear that the hydrophilic interaction between the
In spite of the high concentration and viscosity of this solution, and ionic groups and the polar part of the solvent molecules is stronger
also that this glycerol phase is formed at 333 K during the transes- than the hydrophobic one, and not only determines the capacity of
terification process. This temperature is commonly used in the last the resin but also its ion exchange equilibrium behaviour. Thus, the
step of glycerol purification in the biodiesel production process. capacity of the resin not fully accessible in pure organic solvents,
The Langmuir parameters of the system Na+ /H+ at each in this case (mixtures glycerol–water) is completely available.21,22
temperature are shown in Table 2. As can be seen, the Langmuir Nevertheless, a more accurate interpretation of the physical
equilibrium constants are greater than 1.0, indicating that the reality of the process can be achieved by using a model based
sodium uptake is favourable at the temperatures studied. on the mass action law. An ion exchange reaction is defined as
This capacity is similar to the value reported by Sigma-Aldrich for a reversible exchange of ions between a liquid phase containing
aqueous solutions (Table 1). Nevertheless, this value is higher than counter-ions, which are different from those initially in the solid
that obtained by Pérez-Collado et al.16 for caesium removal from phase. Let us consider the following ion exchange process for a
polyols and that obtained by de Lucas et al.15 for the potassium binary system (Equation (4)) where the ion exchanger (r) is initially
removal from the same liquid phase. in the A-form and the counter-ion in solution is B. The counter-ions
Thus, it is important to point out that this result agrees with exchange occurs, and the ion A in the ion exchanger is partially
previous work indicating that the presence of water in any replaced by B,
concentration in a mixture with an organic solvent exerts a stronger
740

β+ β+
influence than that of the organic solvent on the overall behaviour r + αBs
βAα+ ⇔ βAα+
s + αBr (4)

www.interscience.wiley.com/jctb

c 2008 Society of Chemical Industry J Chem Technol Biotechnol 2009; 84: 738–744
Purification of glycerol/water solutions from biodiesel synthesis www.soci.org

1.0 As can be seen, all the points are above the diagonal and
the curve shapes allow the assumption that the system Na+ /H+
exhibits ideal behaviour for both phases. Thus, the Equation (7) can
0.8
be simplified to:

0.6 yB (1 − xB )
KAB (T) = · (8)
yNa+

(1 − yB ) xB
0.4 Experimental data
T = 303 K
T = 318 K
The standard thermodynamic properties of each binary system
T = 333 K were obtained using the following thermodynamic relations:27
0.2 Theoretical curves
T = 303 K
T = 318 K
◦ RT
T = 333 K GAB = − ln KAB (9)
0.0 αβ
0.0 0.2 0.4 0.6 0.8 1.0 ◦ ◦ ◦
GAB = HAB − T · SAB (10)
xNa+

Figure 2. Ion exchange isotherms of sodium ions on Amberlite 252 in where R is the ideal gas constant, G◦ , H◦ and S◦ are changes
glycerine/water mixtures at 303, 318 and 333 K at a total concentration of in free energy, enthalpy and entropy, respectively.
1 mol L−1 . Ideal Mass Action Law treatment. Combining Equations (9) and (10):

1
where α and β are the valences of the ionic species A and B, KAB (T) =   (11)
αβ  ◦
respectively. HAB ◦
e R · − SAB
At equilibrium, the solid and liquid phases contain both T
competing counter-ion species, A and B. Electroneutrality is nec-
essarily maintained in this reaction whether in the ion exchanger The equilibrium experimental data at the three temperatures
or aqueous solution. This indicates that counter-ion exchange studied were simultaneously fitted to the model given by
occurs in equivalent amounts and the ionic concentration for a Equations (8) and (11). Thus the three unknown parameters of
binary system can be expressed as ionic fractions as follows: the model, the maximum capacity (q0 ) and the thermodynamic
properties H◦ and S◦ were calculated.
CA∗ CB∗ Predicted curves for each temperature are shown in Fig. 2.
xA = ; xB = (5)
N N Thermodynamical properties and equilibrium parameter for the
q∗ q∗B system Na+ /H+ at each temperature applying the ideal mass
yA = A ; yB = (6) action law model are given in Table 3.
q0 q0
As can be seen, the resin capacity is fully available for exchange
where yB and xB represent the ionic fraction of the ion B in the in this mixed media. Besides, this model confirms the preference
solid and liquid phases, respectively. N is the total concentration of this ion exchanger for sodium ions and that the selectivity
of the co-ions in the solution phase (mol L−1 ); and q0 is the useful decreases with temperature. The equilibrium constant reported in
capacity of the resin in the system studied (mmol g−1 of dry resin). the literature for the same ionic system using Amberlite IR-120 at
An important factor that should be considered in a compre- 303 K is about 1.7. This value is almost five times lower than that
hensive theory of ion exchange is the non-ideal behaviour of exhibited in this study for Amberlite 252.28 This result confirms
ions in the solid and liquid phases. This can be a consequence of that the equilibrium constant is strongly affected by the presence
ion–ion and ion–solid interactions, ionic charge, ionic radii and of glycerine, favouring sodium uptake.29
other molecular constants.23 Therefore, the equilibrium constant The ion exchange selectivity is depending on the solvation
for the exchange reaction (4) assuming real behaviour for both of counter-ions and also on the solvation of ion exchange sites,
phases could be expressed by the following equation: which can be affected by the electrostatic field generated by both
counter-ions and the extent of ion-pair formation between an
 β ion-exchange site and a counter-ion.30 Both Amberlite IR-120 and
yBα (1 − xB )β γA
KAB (T) = · · Amberlite 252 have the same type of ion exchange site with a
(1 − yB )β xBα γA
 α  α−β high charge density (—SO− 3 ), which produces strong electrostatic
γB q0 fields without ion-pair formation. Thus, the higher selectivity of
· · (7)
γB N the system Na+ /H+ for Amberlite 252 can be attributed to the
lower solvation of the sodium by glycerol compared with water.
where γ and γ are the activity coefficients of each ion in the ion On the other hand, the negative value of G◦ indicates that
exchanger and in the solution, respectively. this ion exchange process is both feasible and spontaneous; the
Liquid phase activity coefficients in aqueous solutions were negative value of H◦ confirms that the ion exchange processes
suitably calculated from the Pitzer limiting law; the activity are exothermic and whereas, the negative value of S◦ suggests
coefficients in the solid phase can be calculated using the Wilson that the randomness decreased in the liquid/solid interface when
equations that consider the combined effects of the differences in the protons are eluted from the solid by the sodium ions initially
molecular size and intermolecular forces.24 – 26 present in the liquid phase.
Figure 2 shows the equilibrium isotherms in dimensionless form Applying the Langmuir model to obtain the thermodynamical
for the system Na+ /H+ at 303, 318, and 333 K in glycerol/water properties provides the following values of H◦ , −11.233 kJ mol−1
741

solutions on Amberlite 252. and S◦ −19.455 J mol−1 K−1 . These values allow the same

J Chem Technol Biotechnol 2009; 84: 738–744

c 2008 Society of Chemical Industry www.interscience.wiley.com/jctb
 www.soci.org M Carmona et al.

Table 3. Equilibrium and thermodynamical properties of the system H+ /Na+ on Amberlite 252 obtained by fitting the experimental data to the
Ideal Mass Action Law

System T (K) KAB q0 (mmol g−1 dry resin) H◦ (kJ mol−1 ) S◦ (J mol−1 K−1 ) G◦ (kJ mol−1 ) R2

303 9.88 −5.77

H+ /Na+ 318 6.80 4.34 −19.93 −46.71 −5.07 0.988
333 4.85 −4.37

thermodynamical conclusions to be drawn. Finally, if one compares 1.0

the coefficient of determination (R2 ) obtained with the two models,
it can be concluded that the system under study is perfectly
0.8
described by either model.

Kinetics 0.6
Experimental Data

F(t)
A homogeneous model which assumes that a quasi-homogeneous T = 303 K
T = 318 K
phase exists inside the solid particle was employed in previous 0.4 T = 333 K
work with satisfactory results to obtain the effective diffusion Theoretical Curves
coefficients.15,22 This is a useful way to obtain effective diffusion T = 303 K
0.2 T = 318 K
coefficients even in the case of this macroporous resin (Amberlite T = 333 K
252) with similar properties to those of a gell form resin.31,32
Intraparticle transport in ion exchange resins is generally 0.0
controlled by the interdiffusion of counter-ions. The most general 0 1000 2000 3000 4000 5000 6000 7000
solution which supposes intraparticle controlled diffusion from a Time (s)
well-stirred solution of limited volume was given by Crank:33 Figure 3. Kinetics of ion exchange of sodium ions on Amberlite 252 in
glycerine/water mixtures at 303, 318 and 333 K and a total concentration

∞
Mt 6α(α + 1) exp(−p2n Deff t/R2P ) of 0.1 mol L−1 .
F(t) = = 1− (12)
M∞ n=1
9 + 9α + p2n α 2

where Mt and M∞ are the amounts of sodium ions in the particles

at time t and at infinitity, respectively; t is the kinetic time (s); RP means of the solver tool in an in-house spreadsheet in Excel using
is the average radii of the particle size (m); and, pn are non-zero a non-linear fitting method. This iterative method is solved using
roots defined by the following equation. an objective and a constrained cell. The objective cell was the sum
of each squared absolute difference between experimental and
3pn calculated values of F. The constrained cell was obtained as the
tan(pn ) = (13)
3 + α · p2n sum of each relative error between the experimental value and
the calculated value divided by the experimental F value.
Taking into account that the kinetic process ends when
To test the effect of the temperature on sodium uptake, a
equilibrium is achieved, the dimensionless ratio of the volumes of
set of experiments were carried out using a synthetic mixture
solution and sphere (α) is dependant upon the partition factor K:
of glycerol/water (90/10 w/w) containing a total concentration
V 0.1 mol L−1 of sodium chloride. Figure 3 shows the kinetics of the
α= (14) sodium uptake with temperature.
W·K
As can be seen, the higher the temperature, the faster the kinetic
If the initial bulk solution is free of the ions present initially on process. That is, while equilibrium is attained in 1 h at the highest
the resin, a mass balance between the both phases leads to the temperature, the system needs more than 2 h to reach equilibrium
following relationship: at the lowest temperature. The model proposed was able to fit

properly the experimental data (Table 4).
Mt qB CA · V W CA
F(t) = = ∗ = ∗
= ∗ (15) According to previous work,15 effective diffusion coefficients of
M∞ qB CA · V W CA sodium uptake from mixtures of glycerol/water (90/10 w/w) are in
the order of magnitude of those obtained for potassium removal
The partition coefficient was found using Equations (2) and
from crude polyols with this same resin at higher temperatures.
(8) and solving for CB∗ . This procedure was carried out at each
temperature studied. Thus, the equation for α as a function of This result confirms that the macroporous resin performance is
temperature becomes: also good in this mixed solvent.16,34 – 37 The effective intraparticle
diffusivities obtained are in the range found in the literature
V Deff = 1 × 10−12 m2 s−1 for small size ions present in organic
α=  · 103 (16)
q∗A  media, and also for the uptake of high molecular weight molecules
W · ∗ from aqueous media.38 – 40
CA T
The increase of the ion mobility and the decrease in the retarding
Intraparticle effective diffusivity values (Deff (m2 s−1 )) were forces acting on the diffusion ions results in an increase of Deff
742

determined by fitting the experimental data to Equation (12) by with temperature.41

www.interscience.wiley.com/jctb

c 2008 Society of Chemical Industry J Chem Technol Biotechnol 2009; 84: 738–744
Purification of glycerol/water solutions from biodiesel synthesis www.soci.org

-22
Table 4. Effective and pore diffusion coefficients of sodium uptake
by Amberlite 252 in H+ form at different temperatures. The average
deviation between the theoretical values of the fitting and the
experimental data using the homogeneous model are also included -23
Temperature Deff ·1012 Ave. Dev. Dp 1011
(K) (m2 s−1 ) (%) (m2 s−1 )

Ln(DP)
-24
303 1.59 3.91 1.57
318 3.90 2.23 8.23
333 5.11 3.85 8.84
-25

m  
Av.Dev.(%) = ABS FObsi exp
− FObsi theor
FObsi theor × 100 m; m:
i=1
total number of experimental data -26
2.9 3.0 3.1 3.2 3.3 3.4
3
1/T·10
The dependence of the effective diffusion coefficient on Figure 4. Correlation between temperature and pore diffusion coefficients.
temperature is often represented by an Arrhenius equation:41,42

−Ea i.e. 3–11 kJ mol−1 for trivalent chromium diffusion into activated
Deff = D0 exp (17)
RT carbon, and 8–23 kJ mol−1 for water diffusion on mango slices at
different stages of maturity during air drying.41,42
where D0 is the frequency factor (m2 s−1 ), Ea is the activation energy As shown in the equilibrium section, the number of resin active
(J mol−1 ), R the universal gas constant and T the experimental sites exchanged decreased with temperature, indicating the higher
absolute temperature (K). selectivity of this material for the entering ion at lower operating
Although the resin has been considered for modelling purposes temperature, despite the maximum usable capacity seeming to
as in quasi-gel form, it is defined by the manufacturer as be fixed. Thus, it is possible that this selectivity has an important
macroporous. Thus it seems appropriate to give a value of the effect on the effective diffusion coefficients. In a previous paper15
pore diffusion coefficient from the values of Deff , Table 4:43,44 an empirical correlation between the effective diffusion coefficient 
q∗ 
dq  and the final conversion of the solid active sites (yB = qB  ),
0
B T
Deff = DP (18) was employed, where q∗B is the equilibrium concentration of the
dCB T
entering ion in the solid phase at the equilibrium state (mmol g−1
where 2 −1 of dry resin); and q0 is the useful capacity of the resin in the system

DP is the porosity diffusion coefficient (m s ) and studied (mmol g−1 of dry resin).
dqB dCB T is the derivative of the equilibrium equation at each
temperature.  
The derivative of Equation (8) gives: q∗B  s
Deff |T = D∞ (22)
q0 T

dqB  Nq0 KAB
= (19)
dCB T [N + (KAB − 1)CB ]2 where D∞ and s are constants not dependent on the temperature.
Fitting the theoretical effective diffusion coefficients and the
Porosity diffusion coefficients were obtained by combining conversions obtained at each kinetic experiment to the above
Equations (17) and (18). These values are reported in Table 4. equation, the following values of D∞ 2.848 × 10−14 m2 s−1 and
Nevertheless, the above expression has been simplified in the s −23.864 were obtained with the determination coefficient R2
literature to the equilibrium constant when the isotherm is equal to 0.99.
considered as linear.44 Thus: The negative value of s indicates that this ion exchange process
 is favoured at lower temperatures as discussed in the equilibrium
dqB  ∼ section. Thus, if kinetics with the same external concentration
= KAB |T (20)
dCB T were carried out at different temperatures, the highest conversion
would be achieved at the lowest temperature. In this case, the
Using this simplification and combining Equations (11), (17) process would be logically the slowest.
and (18), the following relationship between the pore diffusion
coefficient and the thermodynamical standard properties is found:

CONCLUSIONS
1 Ea + H S Equilibria for the sodium uptake from solution of glycerine/water
LnDP = − + + LnDo (21)
T R R 90/10 w/w using Amberlite 252 in H+ -form was favourable over the
temperature range studied (303–333 K). As expected, the selec-
As seen in Fig. 4, the Arrhenius equation allows adequate tivity decreased with temperature. Both, empirical and theoretical
correlation of the temperature with the pore diffusion coefficients. models based on the Langmuir or mass action laws were able to fit
The following values of D0 , 1.089 m2 s−1 and Ea 68.435 kJ mol−1 , the experimental equilibrium data. The ideal behaviour exhibited
were obtained with a coefficient of determination (R2 ) of 0.962. The by this ionic system allowed the observation that both equilibrium
743

value of the activation energy is in the range found in literature, models led to similar equilibrium parameters. Finally, the stan-

J Chem Technol Biotechnol 2009; 84: 738–744

c 2008 Society of Chemical Industry www.interscience.wiley.com/jctb
 www.soci.org M Carmona et al.

dard thermodynamical properties indicated that this process was 18. de Lucas A, Zarca J and Cañizares P, Ion-exchange equilibrium of Ca2+ .
exothermic, spontaneous and feasible. Mg2+ . K+ . Na+ and H+ ions on Amberlite IR-120: Experimental
Kinetic experiments carried out at different temperatures determination and theoretical prediction of the ternary and
quaternary equilibrium data. Sep Sci Technol 27:823–841 (1992).
showed that this ion exchange process was faster at higher 19. Langmuir. J Am Chem Soc 37:1139 (1915).
temperatures. In addition, a kinetic homogeneous model allowed 20. Coetzee JW and Rejaldien MZ, Technical note the recovery of base
effective diffusion coefficients to be obtained at each studied metals by ion exchange resin. Mineral Eng 14:433–437 (2001).
temperature with satisfactory results. An Arrhenius type equation 21. de Lucas A, Valverde JL, Romero MC, Gómez J and Rodrı́guez JF,
The ion exchange equilibrium of Na+ /K+ in nonaqueous and
allowed the correlation of effective diffusion coefficients with the mixed solvents on a strong acid cation exchanger. Chem Eng Sci
temperature of the kinetic studies. The activation energy was of 57:1943–1954 (2002).
the same order of magnitude as those reported in the literature. An 22. Rodriguez JF, de Lucas A, Carmona M and Cañas F, Application of ion
empirical equation based on the final conversion of the solid active exchange to purify acarbose from fermentation broths. Biochem
sites allowed reproduction of the effective diffusion coefficients. Eng J 40:130–137 (2008).
23. Velayudhan A and Horvath C, Adsorption and ion exchange isotherms
The results confirmed that the macroporous resin Amberlite 252 in preparative chromatography. J Chromatogr A 663:1–10 (1994).
could be a good choice to remove sodium ions from glycerol/water 24. Vo BS and Shallcross DC, Modelling solution phase behaviour in
solutions with a high salt concentration. multicomponent ion exchange equilibrium involving H+ , Na+ ,
K+ , Mg2+ , and Ca2+ ions. J Chem Eng Data 50:1995–2002 (2005).
25. Pitzer KS, Activity Coefficients in Electrolyte Solutions. CRC Press, Boca
Raton, pp 75–153 (1991).
ACKNOWLEDGEMENT 26. Wilson GM, Vapor-liquid equilibria XI. A new expression for the excess
An introduction to this work was presented at the IEX-2008 and free energy of mixing. J Am Chem Soc 86:127–130 (1964).
included in IEX 2008 Proceedings. 27. de Lucas A, Zarca J and Cañizares P, Ion-exchange equilibrium in
a binary mixture. Models for its characterization. Int Chem Eng
34:486–497 (1994).
28. de Lucas A, Cañizares P and Zarca J, Binary ion exchange equilibrium
REFERENCES for Ca2+ , Mg2+ , K+ , Na+ and H+ ions on Amberlite IR-120. Chem
1. Gerpen JV, Biodiesel processing and production. Fuel Process Technol Eng Technol 16:35–39 (1993).
86:1097–1107 (2005). 29. Okada T, Nonaqueous ion-exchange chromatography and
2. Meka P, Tripathi V and Singh RP, Synthesis of biodiesel fuel from electrophoresis approaches to nonaqueous solution chemistry and
safflower oil using various reaction parameters. J Oleo Sci 56:9–12 design of novel separation. J Chromatogr A 804:17–28 (1998).
(2007). 30. Okada T, Nonaqueous ion-exchange chromatography I. Role of
3. Meher LC, Vidya D and Naik SN, Technical aspects of biodiesel solvation in anion-exchange resin. J Chromatogr A 758:19–28
production by transesterification – a review. Renewable Sustainable (1997).
Energy Rev 10:248–268 (2006). 31. Helfferich F, Ion Exchange. McGraw Hill Book Co. Inc, New York (1962).
4. Bournay L, Casanave D, Delfort B, Hillion G and Chodorge JA, New 32. Huang T and Li K, Ion-exchange kinetics for calcium radiotracer in
heterogeneous process for biodiesel production: a way to improve batch system. Ind Eng Chem Fundam 12:50–55 (1973).
the quality and the value of the crude glycerin produced by biodiesel 33. Crank J, The Mathematics of Diffusion, Clarendon Press, Oxford,
plants. Catal Today 106:190–192 (2005). pp 93–95 (1975).
5. Ahn E, Koncar M, Mittelbach M and Marr R, A low-waste process for 34. Melis S, Markos J and Cão G, Ion-exchange equilibria of amino acid on
the production of biodiesel. Sep Sci Technol 30:2021–2033 (1995). a strong acid resin. Ind Eng Chem Res 35:1912–1920 (1996).
6. Noureddini H and Medikonduru V, Glycerolysis of fats and methyl 35. Marcelin G, Cronauer DC, Vogel RF, Prudich ME and Solash J, Shale oil
esters. J Am Oil Chem Soc 74:419–425 (1997). denitrogenation with ion exchange. 2. Evaluation of ion-exchange
7. El-Naggar IM, El-Dessouky ME and Aly HF, Ion-exchange equilibrium of adsorbents and resin treatment procedures. Ind Eng Chem Proc Des
alkali metal/hydrogen ions on amorphous cerium (IV) antimonate. Dev 25:747–756 (1986).
React Funct Polym 28:209–214 (1996). 36. Yan TY and Shu P, Regeneration of ion-exchange resin in nonaqueous
8. Pivovarov S, Adsorption of ions onto amorphous silica: Ion exchange media. Ind Eng Chem Res 26:753–755 (1987).
model. J Colloid Interface Sci 319:374–376 (2008). 37. Rodrı́guez JF, de Lucas A, Leal JR and Valverde JL, Determination of
9. Inoue H, Removal of sodium ion from bile acid solution using intraparticle diffusivities of Na+ /K+ in water and water/alcohol
diffusional dialysis with cation exchange membrane. Sep Purif mixed solvents on a strong acid cation exchanger. Ind Eng Chem Res
Technol 33:189–197 (2003). 41:3019–3027 (2002).
10. Savari S, Sachdeva S and Kumar A, Electrolysis of sodium chloride 38. Yoshida H, Shimizu K and Kataoka T, Adsorption of amine and paints
using composite poly(stryrene-co-divinylbenzene) cation exchange on H-form resin from electrodeposition wastewater. AIChE J
membrane. J Membr Sci 310:246–261 (2008). 36:1815–1820 (1990).
11. Leng B, Method and apparatus for purifying glycerine using 39. Slater MJ, The Principles of Ion Exchange Technology, Butterworth-
filter membranes and deionisations means. UK Patent GB Heinemann, Oxford (1991).
2437516A (2007). 40. Yoshida H, Yoshikawa M and Kataoka T, Parallel transport of BSA
12. Gärtner RS and Witkamp GJ, Regeneration of mixed solvent by ion by surface and pore diffusion in strongly basic chitosan. AIChE
exchange resin: Selective removal of chloride and sulfate. Sep Sci J 40:2034–2044 (1994).
Technol 40:2391–2410 (2005). 41. Mohan D, Singh KP and Vinod VK, Trivalent chromium removal
13. Baciocchi R and Chiavola A, Ion exchange process in the presence of from wastewater using low cost activated carbon derived from
high sulphate concentration: resin regeneration and spent brine agricultural waste material and activated carbon fabric cloth. J
reuse. Water Sci Technol: Water Supply 3:35–41 (2006). Hazard Mater B135:280–295 (2006).
14. Valverde JL, de Lucas A, Carmona M, Pérez JP, Gozález M and 42. Corzo O, Bracho N and Alvarez C, Water effective diffusion coefficient
Rodrı́guez JF, Minimizing the environmental impact of the on mango slices at different maturity stages during air drying. J
regeneration process of an ion exchange bed charged with Food Eng 87:479–484 (2008).
transition metals. Sep Purif Technol 49:167–173 (2006). 43. Flury M and Gimmi T, Solute diffusion, in Methods of Soil Analysis, Part
15. de Lucas A, Cañizares P and Rodrı́guez JF, Ion-exchange kinetics for 4, Physical Methods. Soil Science Society of America, Madison, USA
the removal of potassium from crude polyols on strong acid resins. (2002).
Sep Sci Technol 32:1805–1820 (1997). 44. Li P and Sengupta AK, Intraparticle diffusion during selective ion
16. Pérez-Collado M, Molero C, Larrondo F, Sánchez P, Rodriguez L and exchange with a macroporous exchanger. React Funct Poly
Rodrı́guez JF, Removal caesium catalyst from polyols by ion 44:273–287 (2000).
exchange on Amberlite 252. React Funct Polym 64:139–150 (2005).
17. Costa E, de Lucas A and González ME, Ion Exchange. Determination of
744

interdiffusion coefficients. Ind Eng Chem Fund 23:400–405 (1984).

www.interscience.wiley.com/jctb

c 2008 Society of Chemical Industry J Chem Technol Biotechnol 2009; 84: 738–744