Reactive & Functional Polymers 68 (2008) 1218–1224

Contents lists available at ScienceDirect

Reactive & Functional Polymers

journal homepage: www.elsevier .com/locate / react

Removal of chloride ions from an industrial polyethylenimineflocculant shifting it into an adhesive promoter using the anionexchange resin Amberlite IRA-420

Manuel Carmona, Angel Pérez, Antonio de Lucas, Luis Rodríguez, Juan F. Rodriguez *

Department of Chemical Engineering, University of Castilla – La Mancha, Avda. de Camilo José Cela s/n, 13004 Ciudad Real, Spain

a r t i c l e i n f o a b s t r a c t

Article history:Received 4 December 2007Received in revised form 7 May 2008Accepted 12 May 2008Available online 15 May 2008

Keywords:PolyethylenimineIon exchangeAdhesiveFlocculantAmberlite IRA-420

1381-5148/$ - see front matter � 2008 Elsevier Ltddoi:10.1016/j.reactfunctpolym.2008.05.002

* Corresponding author. Tel.: +34 902204100; faxE-mail address: juan.rromero@uclm.es (J.F. Rodr

Aqueous solutions of polyethylenimines (PEI) are usually used in the manufacture processof paper from pulps to improve the physical strength, and the ink and coatings fixation andto facilitate the printing. When PEI is used as a liquid cationic flocculant in water treatmentit is supplied as a cheaper hydrochloride salt. The hydrochloride salt form is easier to han-dle and may be converted into the free amine form for adhesive purposes by some separa-tion processes. Ion exchange is one of the easiest and cheapest ways to remove chloridesfrom different aqueous solutions.The strongly basic ion exchanger Amberlite IRA-420 has demonstrated that it can be usedfor chloride removal. The equilibrium isotherms of chloride ions i PEI–HCl at 20 and 40 �C,and NaCl and HCl at 30 �C in aqueous solution on Amberlite IRA-420 have been obtained.The presence of HPEI+ as coion exerts an important influence on the ion exchange equilib-rium. This behavior can be explained by the competition between the protonated aminegroups of PEI and the quaternary groups of the resin for the chloride ions that tends to pro-mote a slightly non-favorable equilibrium. In the same way, the kinetic studies indicatedthat the chloride ions are slowly removed in presence of polyethylenimine. Finally, theNernst–Planck homogeneous model allowed obtaining the self-diffusion coefficients forboth the chloride and hydroxide ions in the Amberlite IRA-420 and the values are in thesame order of magnitude as those reported in the literature.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Polyethylenimine (PEI) is a highly branched aliphaticpolyamine characterized by the following repeating unit:

ð1Þ

. All rights reserved.

: +34 926 295318.iguez).

The amine groups exist in primary, secondary and tertiaryforms with a branching site at every 3–3.5 nitrogen atomin any given chain segment. The N/C relation is usually1:2 with a broad range of molecular weights between800 and 2 � 106 g/mol. The PEI is a weakly basic water-sol-uble polymer, with a wide variety of technical applicationsbased on its physical and chemical properties, surfaceactivity, and its ability to form complexes with anionic spe-cies, metal ions and metal complexes [1].

The known commercial use is due to the following abil-ities: promotes the binding between similar and dissimilarmaterials; presents a high cationic charge density, over20 meq/g, and can form chelation complexes with heavymetal compounds [2–4]; presents scavenging capabilitiesfor oxides of carbon, nitrogen, sulfur, and volatile

M. Carmona et al. / Reactive & Functional Polymers 68 (2008) 1218–1224 1219

aldehydes; it can also be used in the fields such as bio-chemistry, in gene transfer processes [5] and as a precipi-tating agent for protein purification. In the papermakingprocess, PEI can be used to improve the physical strength,and the ink and coatings fixation and to facilitate the print-ing [6]. It can also be used in medicine due to the PEI anti-cancer activity which inhibits the growth of the cancercells and seems to be suitable as a polymeric matrix foruse as a carrier in the drug delivery systems [1]. In fact,the wide versatility of this compound has made the BASFChemical Company name it as chemical chameleon in itstechnical bulletin.

PEI is manufactured by an acid-catalyzed ring openinghomopolymerization of ethylene imine monomer:

CH2 ⎯⎯⎯⎯ CH2NH

ð2Þ

The ethylene imine polymerization can be carried out inaqueous and anhydrous media. When the goal is to employit as an adhesive promoter the anhydrous process is rec-ommended. After the polymerization process, the residualcatalyst has to be removed from the bulk polymer by a dif-ficult and expensive process. This makes its commercialpresentations have a high cost even in the case of its pre-sentation in diluted aqueous solutions (04% w/w). Whenit is produced in aqueous media, the presence of the acidcatalyst favors its further applications as liquid cationicflocculant that reduces the production cost. For this reasonit would be very interesting to find an easy and cheap wayto remove the catalyst from the PEI obtained in this way.

The commercial use of PEI started out as a cationic floc-culating agent in paper manufacture [1]. Besides, it hasbeen usually used in wastewater treatments as clarifica-tion or oil emulsification agent at concentrations lowerthan 0.05% in water depending on the application. It actswith negatively charged colloids neutralizing the chargesand thus, the electrostatic repulsion between particles isreduced and the polymer bridging in which independentmovement of particles is restricted by the adsorption ofpolymer molecules simultaneously on the neighboringparticles [7]. The PEI solution used as a flocculant exhibitsacidic characteristics indicating that the amine groups ofthe polymer are highly protonated [1]. Thus, these positivecharges should be neutralized in solution by the negativeones that usually are chloride ions.

This material when used as anhydrous or unprotonatedpolymer is a very effective adhesive agent for printing inks,using solutions in inks between 0.5 and 1% by weight. Inaqueous solution, the product is weakly basic and can beonly employed with pigments and ligands which are alkaliresistant. Besides, at these conditions it is widely used inpaper production, being easily adsorbed by the paper dueto its cationic character. The basic nature of amines makesthe polymer easily ionizable. Its high cationic charge re-duces the high amount of anions that usually load the pa-per and interferes with the drainage/retention of inks andother process chemical additives.

As indicated above, the cheapest presentation of poly-ethylenimine is usually supplied as a hydrochloride saltand it can be converted into the free amine form by neu-

tralization with a base. Nevertheless, for specific applica-tions when the objective is the expensive unprotonatedpolyethylenimine, the metal or chloride ions must be com-pletely removed for the final product usage.

Several methods for chloride removal from polyethy-lenimine solutions could be applied. These methods avail-able for the removal of chlorides are ion exchange,adsorption, liquid extraction, membrane technologies,etc. The ion exchange process seems to be most suitablefor small-scale applications because of its simplicity, effec-tiveness, selectivity, recovery and relatively low cost [8–9].The ion exchange process involves passage of the watersolution of protonated polyethylenimine through a resinbed containing strong-base anion exchange resins onwhich chloride ions are exchanged for hydroxyl ions andthe water formation takes place by neutralization of theprotons with the hydroxyl ions eluted until the resin ex-change capacity is exhausted. The spent resin is usuallyregenerated using a concentrated solution of sodium chlo-ride or hydroxide [10,11].

In this case, the problem is the possible competitionthat would be established between the amine groups ofpolyethylenimine and the fixed quaternary ammoniumgroup of the resin for the chloride ions. This competitionleads to two difficulties. One is it can limit the usefulcapacity of the resin depending on the alkalinity of thegroups of both the polymers (soluble and resin). The sec-ond difficulty could be related with the relative dissocia-tion of the chloride ions and its diffusion rate. If thedissociation of chloride anions is low and they remain rel-atively fixed to the polyethylenimine amine groups, therate of exchange would be restricted by the diffusion ofthe large polymer in the solution or even inside the resin,resulting in a diminution of the observed exchange rate.It has been observed in the previous works of this groupdealing with the exchange of large ions or conventionalcations associated to polymer or in nonaqueous media[12].

The aim of this work is to study ion exchange equilib-rium and kinetics between chloride-polyethyleniminesolution and OH� form of the strong anionic resin Amber-lite IRA-420 at different temperatures and thus to evaluatethe possibility to apply the ion exchange technology toconvert the polyethylenimine flocculant solution (Cl�

form) into an adhesive product for printing applications.

2. Experimental procedure

2.1. Chemicals

An aqueous solution of cationic polyethylenimine (50%w/w) was supplied by LAMIRSA S.A., and hydrogen chlo-ride (37% w/w) and 96% pure sodium hydroxide (PRSgrade) were supplied by Panreac. Demineralised waterwas conventionally treated in our laboratory (final conduc-tivity less than 1 ls/cm). A commercial ion exchanger gel-type strongly basic Amberlite IRA-420 supply by Rohm andHaas was used. It is an amine quaternary cross-linked sty-rene/divinylbenzene copolymer. Table 1 shows a summaryof the resin properties.

Table 1Properties of the resin Amberlite IRA-420

Producer Rohm and HaasFunctionality –N+–(CH3)3

Matrix type Polystyrene-DVBStandard ionic form Cl�

Total exchange capacity (meq/g) 3.80Bed Porosity 0.32Wet resin density (g/cm3) 1.15Bed density (g/cm3) 0.68pH operating range 0 – 14Maximum operating temperature 77 �CMean wet particle radius (mm) 0.30 – 0.70

0 200 400 600 800 1000 12000.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Solid

pha

se c

once

ntra

tion

of c

hlor

ide

ion

(eq·

kg-1)

Liquid phase concentration of chloride ion (eq·m-3)

HCl,T = 303 K NaCl, T = 303 K PEI, T = 313 K PEI, T = 293 K

Fig. 1. Isotherms of ion exchange for chloride ion removal using differentco-ions H+, Na+ and polyethylenimine in aqueous media on AmberliteIRA-420. Symbols: empirical data; lines: theoretical model.

1220 M. Carmona et al. / Reactive & Functional Polymers 68 (2008) 1218–1224

Before its utilization, the resin was saturated and regen-erated in successive cycles by using HCl and NaOH solution0.1 M, in order to prepare the resin for further ion ex-change tests and obtain normalized conditions of capacityand ionic form. After that, the resin was dried under con-trolled atmosphere until reaching the constant humiditycontent, about 20%. Furthermore, before each equilibriumexperiment the water content of the resin was measuredin order to ensure that the resin weight in dry basis usedfor calculations was right.

2.2. Equilibrium procedure

The experiment apparatus consisted of nine hermeti-cally sealed 0.25 l Pyrex containers submerged in a con-trolled thermostatic bath. The temperature was keptconstant within ±0.2 �C. Equal masses of resin, in theOH� form, were put in contact with aqueous solutions ofdifferent contents of polyethylenimine comprised between0.0387% and 1.11% w/w and correspondingly different con-centrations of chloride ions ranging between 252 and7575 ppm at their natural pHs that varied between 3.79and 2.85, respectively. Previous experiments allowedestablishing that the equilibrium was ensured in 24 h.After equilibration, resin and solution were separated byfiltration and the final pH and chloride content of the solu-tion were measured.

The chloride concentration in the polyethyleniminesolutions was measured with two alternative and comple-mentary methods. The first using a chloride specific elec-trode model 15 213 3000, of INGOLD with a referenceelectrode model 373-90-WTE-ISE-S7 of INGOLD connectedto a pH-meter model MICRO-pH 2002 of CRISON equippedwith a temperature meter. The other by titration with asolution of silver nitrate (AgNO3) using the previously de-scribed system of electrodes for the detection of voltamet-ric changes and the titration end point. In the case of theequilibrium studies, all the equilibrium data were mea-sured by the more accurate titration method.

The equilibrium resin phase concentration of the chlo-ride was calculated by mass balance according to

n�a ¼VWðC0 � C�aÞ ð3Þ

where C0 and C�a are the initial concentration and the equi-librium ‘‘analytical” concentration of chloride in the liquidphase (mmol/l solution), respectively. n�a denotes the resinphase equilibrium concentration of chloride (mmol/g dry

resin). V is the solution volume (l) and W is the dry resinweight (g).

2.3. Batch kinetics procedure

The rate at which chloride was taken up by the resinwas studied in a discontinuous well-mixed tank. In a typi-cal experiment, a quantity of Amberlite IRA 420 was mixedwith 1 l of chloride-polyethylenimine solution in a baffledglass mixing vessel. The mixture was then stirred for 1 h.The electrode system was introduced inside the reactorin order to follow on-line the evolution of chloride concen-tration in the bulk solution. Small aliquots (2 ml) of thesolution were periodically taken out and the chloride con-tent was analysed by titration to verify the specific elec-trode measurements. The sample size was consideredsmall enough not to alter the solution/resin ratio.

3. Results and discussion

Although it seemed clear that ion exchange could be asuitable technique to remove the chloride ions from poly-ethylenimine solutions, it is important to know the influ-ence of such kind of co-ion on the available resincapacity and also on both the equilibrium and the kineticbehavior.

3.1. Equilibrium treatment

3.1.1. Empirical treatmentFig. 1 shows the equilibrium isotherms at 293 and

313 K of chloride ions in polyethylenimine solutions onthe strong basic ion exchange resin Amberlite IRA-420and the isotherms of sodium chloride and hydrochloricacid. The curves have been drawn using the Langmuirparameters obtained by fitting the experimental data tothe theoretical equation

q� ¼ KqT C�

ð1þ qT C�Þ ð4Þ

Table 2Parameters of the Langmuir equation for chloride removal from the liquidphase by Amberlite IRA-420

Systems Equilibrium parameters Av. Dev. (%)

qT (eq kg�1) K (l eq�1)

Resin–HCl (T = 303 K) 3.8 183.08 17.66Resin–NaCl (T = 303 K) 3.8 6.82 8.18Resin–PEI–HCl (T = 293 K) 2.52 60.20 4.78Resin–PEI–HCl (T = 313 K) 2.73 67.65 3.48

Av: Dev:% ¼Pm

i¼1ABS q�exp � q�theo=q�exp

� �� 100=m; m: total number of

experimental data.

0 8 100.0

0.2

0.4

0.6

0.8

1.0

Poto

natio

n de

gree

of

the

poly

ethy

leni

min

e

pH

PEI/[PEI]0 HPEI+/[PEI]0

642

Fig. 2. Effect of the pH on the protonation degree of thepolyethylenimine.

M. Carmona et al. / Reactive & Functional Polymers 68 (2008) 1218–1224 1221

Langmuir parameters of each system are shown in Table 2.It can be clearly seen that the presence of polyethylen-

imine in the media as co-ion exerts an important influenceon the equilibrium uptake of the chloride ions. The impor-tance of the influence is comprised between two limits. Itis not so weak as the presence of the strong Na+ that re-main completely dissociated almost independently of theamount of hydroxyl anions in the solution, but not asstrong as the almost irreversible reaction between the H+

ion of hydrochloric acid and the OH� released from the re-sin beads. According to the Langmuir parameters, the resinexhibits the same useful capacity when the chloride ionsare completely dissociated in liquid phase, this is whenNaCl and HCl are used. This capacity agrees with the valuereported in Sigma–Aldrich catalog and the capacity foruptaking phenolate [13]. On the other hand, for polyethy-lenimine solutions the results indicate that higher concen-tration of chloride ions are required to spend the total resincapacity or another phenomena could be taking place.

3.1.2. Theoretical treatmentIn the case of NaCl, the reaction between the OH-form

resin (R–OH) and the chloride may be understood as a con-ventional ion exchange reaction:

RAOHþ Cl� $ RAClþ OH� ð5Þ

where bars refer to the resin phase and R denotes the resinpolymeric matrix. In the case of HCl the OH� group leavingthe resin reacts almost irreversibly with the H+ of the bulkgiving an isotherm with the typical pattern of the irrevers-ible isotherms:

RAOHþ Cl� þHþ $ RAClþH2O ð6Þ

If the protonated polyethylenimine (HPEI+) would not ex-ert any influence on the equilibrium of the system, theequilibrium isotherms must agree with that of the sodiumchloride in aqueous solution and that it is not the case.

A competitive equilibrium between the ammoniumgroups of the resin and the polyethylenimine is estab-lished, and taking this complex equilibrium into accountthe following equilibrium equation can be written:

RAOHþHPEIþCl� $ RAClþ PEIþH2O ð7Þ

where HPEI+Cl� represent the ionic pair that the Cl� ionsform with the protonated PEI. The existence of the ionicpairing to some extent was confirmed by conductivitymeasurements with abnormally low values.

Applying the ideal mass action law to Eq. (7), the fol-lowing equation can be written:

K II ¼½RACl� � ½PEI�

½RAOH� � ½HPEIþCl��ð8Þ

and the following equation relating the solid phase chlo-ride concentration with the chloride concentration in solu-tion can be derived:

q� ¼ K IIqT ½HPEIþCl��½PEI� þ K II½HPEIþCl��

ð9Þ

where KII denotes the equilibrium constant and qT denotesthe maximum ion exchange capacity of the resin (meq/g).The amounts of HPEI+ and PEI in the solution are relatedwith the pH by a Hassebalch type equation [1]:

pH ¼ pKa—n log1� a

að10Þ

where pKa is the dissociation constant of HPEI+, n is a con-stant depending on each polyelectrolyte which accountsfor the nearest neighboring interaction and 1-a is the dis-sociation degree of PEI expressed as

1� a ¼ ½HPEIþ�½PEI�0

ð11Þ

[PEI]0 being the initial concentration of PEI in solution.From Eqs. (10) and (11) the concentration of [HPEI+] and

[PEI] as function of pH can be obtained:

½PEI� ¼ ½PEI�0 1þ 10pKa—pH

n

h i�1ð12Þ

½HPEIþ� ¼ ½PEI�0 1� 10pKa—pH

n

h ið13Þ

von Zelewsky et al. [1] found that this kind of polymerexhibits a pKa of 7.69 and n equal to 7 for dissociation de-gree values higher than 0.3. The change with pH of the rel-ative concentration of both PEI forms in solution, PEI andHPEI+, with respect to the initial concentration in solutioncan be observed in Fig. 2. As can be seen, the PEI becomesin less charged as the pH value increases.

Substituting the formulae of PEI and HPEI+ in the massaction law Eq. (9) the following expression is obtained

1222 M. Carmona et al. / Reactive & Functional Polymers 68 (2008) 1218–1224

q� ¼K IIqT 1þ 10 pKa—pH

n

� �1þ K II 1þ 10 pKa—pH

n

� � ð14Þ

Taking into account that the maximum ion exchangecapacity of the resin is known, the unknown parameter isonly the KII for the system PEI-resin-chloride at each stud-ied temperature. These values obtained by fitting theexperimental data to the theoretical Eq. (14) are shownin table 3. In Fig. 3 the good fitting between the experimen-tal and the theoretical data as a function of the equilibriumpH can be observed. These results allow to conclude thatthe maximum ion exchange capacity is not a function ofthe co-ions but it is dependent on the concentration of freecounter-ion in solution which can be limited by the liquidphase equilibrium association of the counter-ion with thesoluble polymer.

For comparison purposes, all the experimental datahave also been represented in dimensionless form (Fig. 4)where X = n*/np and Y* = C*/C0 . It seems clear after thistreatment that there is a competition between the proton-ated amine groups of PEI and the quaternary groups of theresin for chloride ions that make that the equilibrium rep-resented in dimensionless appear as slightly non-favour-able, this conclusion could be also reached looking at thevalues of the ion exchange equilibrium constants whichare lower than one. As can be seen in Fig. 4, the uptakein equilibrium of an ion by a solid is dependent on theco-ion when it is not from a strong acid, and besides theexchange process is promoted by increasing thetemperature.

Table 3Parameters of the ideal mass action law model for chloride removal fromthe liquid phase by Amberlite IRA-420

Systems Equilibrium parameters Av. Dev. (%)

qT (eq � kg�1) KII

Resin–HCl (T = 303 K) 183.08 (l � eq�1) 17.66Resin–NaCl (T = 303 K) 3.8 2.00 3.39Resin–PEI–HCl (T = 293 K) 0.47 10.15Resin–PEI–HCl (T = 313 K) 0.61 9.64

0 10 120.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Solid

pha

se c

once

ntra

tion

(eq·

kg-1)

pH

PEI at 313 KPEI at 293 K

8642

Fig. 3. Reproducibility of the experimental data by the theoretical model.Symbols: empirical data; lines: theoretical model.

3.2. Kinetic treatment

In order to simulate the operation and estimate the sizeof the equipment required for polyethylenimine purifica-tion at industrial scale, one needs to have not only theequilibrium data but also the effective diffusion coeffi-cients. For this reason some experiments were done to ob-tain the self-diffusion coefficient of chloride and hydroxideions in the resin.

Amberlite IRA-420 is described by the manufacturers asa geliform resin. For this reason the kinetic behaviour ofthe chloride-polyethylenimine Amberlite IRA-420 systemshould be properly represented by a homogeneous model.The homogeneous model assumes that there exists a qua-si-homogeneous phase inside the solid particle [14,15].Furthermore, the same solution is also valid, as a limitingcase, for biporous models assuming that the entire masstransfer resistance lies in the microespheres [16].Consider-ing an ion exchange isothermal process between an ion ipresaturating the resin and an ion j entering the resinparticle,

R�þiþ jþ () R�þjþ iþ ð15Þ

The Nernst–Planck theoretical treatment for the ionicfluxes of counter-ions has been employed. The electroneu-trality condition has to be verified. The general equationcorresponding to the flux of the ion i, expressed in termsof the concentration gradient, is then

Ni ¼ �Di 1þ qið1� dÞqT þ qiðd� 1Þ

� �oqi

orð16Þ

where d = Di /Dj is the ratio between the diffusivities of spe-cies i and j, and Dij is the interdiffusion coefficient.The massbalance in the resin particle is:

oqi

ot¼ � 1

r2

oðr2NiÞor

ð17Þ

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

Dim

ensi

onle

ss c

once

ntra

tion

in th

e so

lid p

hase

(Y

)

Dimensionless concentration in the liquid phase (X)

HCl,303 K NaCl, 303 KPEI, 313 KPEI, 293 K

Fig. 4. Dimensionless isotherms of chloride ion removal from solutionswith different co-ions on Amberlite IRA 120. C0 = 0.1 N for HCl.

0 10 20 30 40 50 600.0

0.2

0.4

0.6

0.8

1.0

(C0-C

t)/(C

0-C* )

Time (min)

HCl NaCl PEI-HCl

Fig. 5. Kinetics of chloride uptake in aqueous media. V = 1 l. W = 10 g.T = 303 K. Speed = 500 r.p.m. C0 (Cl) = 15 eq m�3. Solid lines: predictedcurves.

M. Carmona et al. / Reactive & Functional Polymers 68 (2008) 1218–1224 1223

and introducing the flux Eq. (16), we get

oqi

ot¼ 2DiqT

rðqt þ ðd� 1ÞqiÞoqi

or

� þ DiqTðd� 1ÞðqT þ ðd� 1ÞqiÞ

2

oqi

or

� 2

þ DiqT

ðqT þ ðd� 1ÞqiÞo2qi

or2

!ð18Þ

with initial and boundary conditions

t ¼ 0; qi ¼ qr Ci ¼ Ci0 ¼ CT ð19Þ

r ¼ 0; oqior ¼ 0 ð20Þ

and taking into account the resistance in the film to masstransfer

r ¼ Rp; �Dijoqior

r¼Rp

¼ KlðCsi —CBulk

i Þ ð21Þ

where the concentration at the solid resin surface qsi is re-

lated with the concentration of the liquid Csi by an equilib-

rium relation based in the Langmuir equation Eqs. (4) withthe parameter values shown in Table 2.

The mass balance in the bulk of tank is

V � oCBulki

ot¼ 3

Rp� KL � ðCs

i —CBulki Þ �W resin

qresinð22Þ

To describe the mass transfer through the external film, thefollowing relation obtained by Calderbank and Moo-Young[17] for the process of ion exchange using a stirrer tankreactor was employed

KLSc�2=3 ¼ 0:13 � ðP=VÞlq2

� �1=4

ð23Þ

where Sc is the Schmidt number and P/V is the dissipatedpower per unit volume, both terms defined as

Sc ¼ lq � Df ð24Þ

According to Bates and Jonson [18], for turbulent regimenthe power number is constant and can be obtained by

P

qx3d5a

¼ 3:5 ð25Þ

where l and q are the viscosity and the density of the sol-vent, Df is the free diffusivity of the entering ion in solution,and x and da are the speed and the diameter of the stirrer,respectively.

The above system of equations has been solved by finitedifferences [19]. The partial differential Eq. (18) is trans-formed into a set of l ordinary differential equations. In thisway, the ion exchange model is reduced to a stiff set of l + 1ordinary differential equations taking into account the ini-tial and boundary conditions, l being the number of nodesin which the particle is divided. Marquardt’s algorithm wasused as a nonlinear fitting procedure to determinate boththe unknown diffusion coefficients (Di and Dj) inside theparticle [20]. The Rosenbrock method was used to inte-grate numerically the stiff set of ordinary differential equa-tions due to its numerical stability and the few steps ofintegration required to achieve the solution [21]. A Fortran6.0 application was developed for solving this model.

Three experiments were carried out using AmberliteIRA-420 in OH�form at the same initial concentration ofchloride in solution (C0 = 15 eq/m3 and T = 303 K) but withthree different co-ions, Na+, H+ and polyethylenimine. Atthe stirring rate employed for the experiments (500 rpm)diffusion inside the resin bead must be the limiting step.In order to obtain reliable and meaningful values of theintraparticle diffusion coefficients for OH� and Cl� ions,all the experimental data were fitted together by nonlinearregression to the mathematical model described aboveusing the corresponding equilibrium isotherm of each sys-tem. The equilibrium for PEI at 303 K was obtained by athermodynamical approach using the values of the Lang-muir equilibrium constants obtained for both the iso-therms studied. In this way, the following values wereobtained DCl = 1.442 � 10�10 and DOH = 7.695 � 10�10 m2/sfor the chloride and the hydroxide ions, respectively. Theself-diffusion coefficient for chloride ion is smaller thanthat corresponding to hydroxide, indicating its lowermobility inside the bead which is due to its bigger hy-drated ion size.

As can be seen in Fig. 5, the kinetic of chloride removalin the presence of polyethylenimine is slower when astrong co-ion is accompanying the exchanging ion. Thiscan be explained by the equilibrium effect because whenboth the conventional systems reach the equilibrium inless than 10 min the polyethylenimine system needs al-most half hour to achieve it.

As it has been explained in the equilibrium section, inthis system the chloride ions are partially non dissociatedand the existence of the ion pair chloride-polyethyleni-mine limits their diffusion. In this way, we can concludethat the real exchange of the anion is limited by the equi-librium with the soluble polymer and even it is possiblethat its diffusivity at least in the film is also difficulted bya certain linkage to the polymer than makes more difficultits diffusion.

Soldano and Boyd [22] reported the self-diffusion coef-ficient of chloride ion with a value of DCl = 0.35 � 10�10 m2/susing the Dowex 2X6 resin, Watson [23] found that the DCl

1224 M. Carmona et al. / Reactive & Functional Polymers 68 (2008) 1218–1224

can be between 3.0 � 10�10 and 4.2 � 10�10 m2/s for Dowex21 K. Kishore and Verma [24] have used values for effectivepore diffusivity of OH� ions in ion exchange resins in therange between 0.2 � 10�10 and 0.8 � 10�10 m2/s. They se-lected these values modifying the free ionic diffusion coef-ficient by the tortuosity factor. In a recent article, Galinadaand Yoshida [25] reported similar values for phosphateions using a strongly basic ion exchanger (DIAION SA10A)resin. The above values are close to the value obtainedfor both the chloride and hydroxide ions using this treat-ment and the differences can be attributed to the resincharacteristics.

4. Conclusions

Chloride ions from polyethylenimine aqueous solutioncan be removed by using a strong-base anion exchanger.It was observed that the presence of H+PEI as coion exertsan important influence on the ion exchange equilibrium. Inthis separation process, a competition takes place betweenthe protonated amine groups of PEI and the quaternarygroups of the resin due to the chloride ions that evolvesto a slightly non-favourable equilibrium. On the otherhand, the developed theoretical equilibrium model al-lowed to obtain the ion exchange equilibrium constant ateach temperature and confirms that for strong-base anionexchanger the maximum capacity is not a function of thepH or the co-ion characteristics. The kinetic studies indi-cated that the chloride ions are slowly removed whenthe polyethylenimine is in solution due to the ionic pairformation. Self-diffusion coefficients for both the chlorideand the hydroxide ions in the Amberlite IRA-420 were ob-tained by using the Nernst–Planck homogeneous model.These values are in the same order of magnitude as thosereported in the literature. According to the results, thepolyethylenimine properties can be changed from a cheap-er flocculant to an expensive adhesive by an ion exchangeprocess.

References

[1] A. von Zelewsky, L. Barbosa, C.W. Schläpfer, Coordination ChemistryReviews 123 (1993) 229.

[2] P. Cañizares, A. Pérez, R. Camarillo, Desalination 144 (2002) 279.[3] P. Cañizares, A. de Lucas, A. Pérez, R. Camarillo, Journal of Membrane

Science 253 (2005) 149.[4] R. Molinari, S. Gallo, P. Argurio, Water Research 38 (2004) 593.[5] A. Kichler, The Journal of Gene Medicine 6 (2004) S3.[6] S. Zakaria, B.H. Ong, T.G.M. van de ven, Colloids and Surfaces A:

Physicochemical and Engineering Aspects 251 (2004) 31.[7] B. Alince, A.A. Roberston, Colloid & Polymer Science 252 (1974) 920.[8] M. Chabani, A. Bensmaili, Desalination 185 (2005) 509.[9] R.S. Gärtner, G-J.G.-J. Witkamp, Separation and Purification

Technology 40 (12) (2005) 2391.[10] R. Baciocchi, A. Chiavola, Water Science Technology: Water

Supplement 3 (6) (2006) 35.[11] J.L. Valverde, A. de Lucas, M. Carmona, J.P. Pérez, M. Gozález, J.F.

Rodríguez, Separation and Purification Technology 49 (2006) 167.[12] A. de Lucas, P. Cañizares, J.F. Rodríguez, I. Gracia, Chemical

Engineering Journal 66 (1997) 137.[13] M. Carmona, A. de Lucas, J.L. Valverde, B. Velasco, J.F. Rodríguez,

Chemical Engineering Journal 117 (2006) 155.[14] A. De Lucas, P. Cañizares, J.F. Rodríguez, Separation and Purification

Technology 32 (1997) 1805.[15] T. Huang, K. Li, Industrial and Engineering Chemistry Fundamentals

1 (1973) 50.[16] B.G. Neeru, B.G.R. Manoja, K. Gupta, Chemical Engineering Science

19 (1994) 3277.[17] P.H. Calderbank, M.B. Moo-Young, Chemical Engineering Science 16

(1961) 39.[18] J.C. Bates, T.D. Jonson, Ion exchange technology, Proceeding of

IEX’84. Ellis Horwood, Chichester UK, 1984.[19] B.A. Finlayson, Nonlinear Analysis in Chemical Engineering,

McGraw-Hill, New York, 1980.[20] D.W. Marquardt, Journal of the Society of Industrial and Applied

Mathematics 11 (1963) 431.[21] W. Press, S. Teukolsky, W. Vetterling, B. Flannery, Numerical Fortran

Recipes in Fortran, second ed., Cambridge University Press, USA,1992.

[22] B.A. Soldano, G.E. Boyd, Journal of the American Chemical Society 75(1953) 6099.

[23] J. Watson, ORNL Report, 1962, p. 3296..[24] K. Kishore, N. Verma, Chemical Engineering and Processing 45

(2006) 31.[25] W.A. Galinada, H. Yoshida, AIChE Journal 50 (2004) 2806.