A

sa1wsibd©

K

1

prMgSaTsm12oa2t

0d

Biosensors and Bioelectronics 22 (2006) 145–152

Potentiometric sensor for atrazine based on amolecular imprinted membrane

G. D’Agostino ∗, G. Alberti, R. Biesuz, M. PesaventoDipartimento di Chimica Generale, Universita degli Studi di Pavia, Via Taramelli 12, 27100 Pavia, Italy

Received 24 January 2006; received in revised form 21 April 2006; accepted 4 May 2006Available online 3 July 2006

bstract

A molecular imprinted polymer (MIP) membrane for atrazine, not containing macropores, was synthesized and implemented in a potentiometricensor. It is expected to work like a solid ISE (where the specific carrier are the imprinted sites) the specific carrier being the imprinted site. Thective ion is the protonated atrazine, positively charged. To form this species the determination is carried out in acidic solution at pH lower than.8, in which atrazine is prevalently monoprotonated. At these conditions the membrane potential increases with atrazine concentration over aide concentration range (3 × 10−5 to 1 × 10−3 M). The slope of the function E versus log c is about 25 mV/decade, showing that the atrazine form

orbed on MIP is the biprotonated one. The detection limit is determined by the relatively high concentration of atrazine released by the membrane

n the sample solution at the considered conditions. It seems to be independent of the atrazine concentration in the internal solution of the sensor,ut it depends on the acidity of the solution. The response time is less than 10 s and the sensor can be used for more than 2 months without anyivergence.

2006 Elsevier B.V. All rights reserved.

selec

bmppbat2fi

ctSi

eywords: Molecular imprinted polymer; Potentiometric sensor; Atrazine; Ion

. Introduction

The molecular imprinted polymers for triazinic herbicides,articularly atrazine, were the subject of many investigations inecent years with interesting application in chemical analysis.

IPs are used for solid phase extraction (SPE) and chromato-raphic separation (Matsui et al., 1995, 1997; Muldoon andtanker, 1997; Olsen et al., 1998; Takeuchi et al., 1999; Lanzand Sellergren, 1999; Bjarnason et al., 1999; Ferrer et al., 2000).he use of these polymers for the recognition in chemical sen-ors has been also proposed. Some authors prepared polymericembranes for conductimetric and capacimetric (Piletsky et al.,

995, 1998; Sergeyeva et al., 1999; Panasyuk-Delaney et al.,001), for piezoelectric detection (Pogorelova et al., 2002), andther signal transduction methods, for example through the bulk

coustic wave and fluorescence (Liang et al., 2000; Jenkins et al.,001) Despite the relatively simple transduction of the poten-iometric signal, only a few sensing devices of this kind have

∗ Corresponding author. Tel.: +39 0382 987 580; fax: +39 0382 528 544.E-mail address: girolamo.dagostino@unipv.it (G. D’Agostino).

pstaara

956-5663/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.bios.2006.05.014

tive electrode

een developed. All of them were based on the use of very thinembranes or films (Kitade et al., 2004; Zhou et al., 2005), with

roblems due to difficulties in preparation, reproducibility andossible interferences. Recently a sensor has been developed,ased on a MIP polymeric film deposited on the gate surface ofn ion-sensitive field-effect transistor, obtaining good results inerms of specificity and low detection limit (Pogorelova et al.,002). Potentiometric sensors have been previously proposedor charged analytes, for example for nitrate anion based onmprinted conducting polymers (Hutchins and Bachas, 1995).

Atrazine is protonated, and as a consequence, positivelyharged in aqueous solution at sufficiently low pH, its protona-ion constant being log Ka = 1.7 (Smolkova and Pacakova, 1978;kopalova and Kotoucek, 1995). The combination of the pos-

tively charged species with the imprinted membrane shouldroduce a variation of the membrane charge. This can be mea-ured potentiometrically by a conventional ISE device, in whichhe membrane is placed between an internal filling solution

t constant composition in contact with a reference electrode,nd the sample solution (Craggs et al., 1974). This configu-ation is very simple, and gives a Nernstian response to thenalyte concentration, which is very convenient for quantifi-

146 G. D’Agostino et al. / Biosensors and Bioelectronics 22 (2006) 145–152

eme 1

cto

bdwnftmaoMwa

2

2

lt9(wau

2

wo2fTlaauaoTstamb(et

1ofH

2

w0ed

(

c13

t0trj

3

3

fftdlit

b

tt3dt

Sch

ation. Moreover it reduces potential interferences, for examplehose from redox active substances (Hutchins and Bachas, 1995),r generally from substances active at the transductor.

In the present work, a molecular imprinted polymeric mem-rane for atrazine was prepared, The membrane was formedirectly at the end of a small Teflon tube which can be filledith a solution at constant composition, in contact with an inter-al reference electrode, as in a classical potentiometric cellor ion selective electrode (ISE). (Craggs et al., 1974), likehat depicted in the Scheme 1 reported below. This preparation

ethod for potentiometric sensors for charged species is herepplied because it is simple and straightforward. The affinityf the protonated form of atrazine for the imprinted sites in theIP membrane can be different from that of the neutral atrazine,hich is the template. For this reason the effect of the solution

cidity is particularly investigated in the present work.

. Experimental

.1. Materials

Methacrylic acid [79-41-4] and ethylene glycol dimethacry-ate [97-90-5] (Sigma–Aldrich)) were distilled in vacuum prioro use in order to remove stabilizers. Atrazine [1912-24-] (Sigma–Aldrich)) and 2,2′-azobisisobutyronitrile [78-67-1]AIBN) (Sigma–Aldrich), were of reagent grade and were usedithout any further purification. All other chemicals were of

nalytical reagent grade and the solutions were prepared withltrapure water (Milli-Q).

.2. Preparation of the MIP membrane

The molecular imprinted polymer membranes for atrazineas prepared from a reagent mixture obtained by mixing 48 �lf methacrylic acid, 95.2 �l of ethylene glycol dimethacrylate,5 mg of atrazine and 28 mg of AIBN. The mixture was uni-ormly dispersed by sonication (sonic bath model Transsonic420—Elma) and then deaerated with nitrogen for 10 min. Ethy-

ene glycol dimethacrylate acts not only as cross-linker, but alsos solvent. A 25 �l aliquot of reagent mixture was placed inhome made Teflon device that controls the shape of the liq-

id during the polymerization. This device was introduced inn oven (model M710) at 70 ◦C for 17 h. A glassy membrane isbtained with 5 mm diameter, tightly fixed at the end of a smalleflon tube 10 mm long. This is directly used for assembling theensor without any further manipulation of the membrane. Thehickness of the membrane can be varied by placing differentmounts of reagent mixture into the Teflon device for the poly-erization. The template was removed by washing the mem-

rane successively in 10 ml of a methanol/acetic acid solution4:1, v/v, of 98% methanol and pure acetic acid) for three times,ach time for 1 h, then in 10 ml of water/acetic acid (5:1) for threeimes for 1 h and finally in 10 ml of pure water for three times for

pIts

.

h. The membrane was conditioned in 3.7 × 10−4 mol l−1 aque-us atrazine solution (pH 1.5, HCl solution, KCl 0.02 mol l−1)or 16 h. In some cases the membrane was treated in 0.1 mol l−1

Cl before conditioning.

.3. Sensor construction and potential measurements

The Teflon tube bearing the membrane at the end was filledith a solution containing 3.7 × 10−4 mol l−1 atrazine, HCl.03 M (pH 1.5) and saturated with KCl. A Ag/AgCl referencelectrode was contacted with the internal filling solution to trans-uce the potentiometric signal.

All the potential measurements were carried out at 25.0 ◦C±0.1 ◦C) in the cell as shown in Scheme 1.

Test solution: HCl pH 1.5, KCl 2 × 10−2 mol l−1, atrazine;Membrane: MIP membrane.

The cell potential was measured at different atrazine con-entration in the test solution in the range 1.7 × 10−7 to.7 × 10−4 mol l−1. All pH adjustments were made with HCl7% or NaOH 50%.

The liquid junction potential Ej at the interface betweenhe filling solution of the double junction reference electrode,.34 mol l−1 KNO3, and the test solution at the usual composi-ion depends on the proton activity according to the experimentalelation: Ej = −24aH = −24γHcH. The correction for the liquidunction potential was done when appropriate.

. Results and discussion

.1. Potentiometric response of the MIP membrane

The molecular imprinted polymer membrane was preparedrom a reagent mixture containing only methacrylic acid as theunctional monomer. Not any porogen solvent was used, in ordero avoid the formation of macropores through which a rapidiffusion of atrazine across the membrane would take place. Aarge fraction of cross-linker, EDMA, was employed, 88%, andn this way a rigid membrane, resistant to stress, and well sealedo the Teflon tube, was obtained.

Some examples of the potentiometric response of MIP mem-ranes are displayed in Fig. 1.

Increasing concentrations of atrazine (cadd) were added tohe test solution. The potential increases immediately afterhe atrazine addition at concentrations higher than around× 10−5 M, in the case of the membrane conditioned asescribed in the experimental part. For comparison the poten-iometric response obtained with a similar membrane, but not

reviously conditioned in atrazine solution, is reported in Fig. 1.t shows that in the first point, at cadd = 1 × 10−7 mol l−1, a longime is required to reach the equilibrium potential. Besides, thelope of the function E versus log cadd increases with the con-

G. D’Agostino et al. / Biosensors and B

Fig. 1. Cell potential at different atrazine concentrations in function oft −7

1(

ctamtvsctla

M

gcaott

a0at1pKctp

utim4ipbAwsdm

1wstd

The membrane potential was constant at 0.1 mV after equili-

TS

MMM

M

A

ime. Conditions reported in Scheme 1. Atrazine concentration 1.7 × 10 to.7 × 10−4 mol l−1. Membrane 0.5 mm thick. (�) MIP membrane conditioned;*) MIP membrane non conditioned; (–) NIP membrane conditioned.

entration, being in any case much lower than that obtained withhe conditioned membrane. Probably when the concentration oftrazine in the membrane is low, as it is in the not conditionedembrane, some atrazine is sorbed from the test solution into

he membrane, producing a potential decrease. It has been pre-iously reported, for example in the case of a potentiometricensor for nitrate (Hutchins and Bachas, 1995), that the sensoronditioning, reduces the potential drift due to the diffusion ofhe template in the bulk of the membrane and shortens the equi-ibration time, even in the case of relatively thick membranes,

s seen in Fig. 1.

A membrane synthesized, washed and conditioned like theIP membrane but in the absence of template (NIP membrane)

b

c

able 1tandardization curve of different atrazine MIP membranes, at pH 1.5, in KCl 0.02, a

Standardization lineE = E0′

add + Nlog cadd

E(p)

(mV)DL(10−5 M)

embrane 0.5 mm E = 211(3) + 24.2(7)log c 99 2.3embrane 1 mm E = 135(2) + 32.2(5)log c −3.5 5.0embrane 0.5 mm,treated with 0.1 M HCl

E = 171(4) + 22(1)log c 62.7 1.2

embrane 0.5 mm,atrazine concentrationinternal solutionc = 5 × 10−5 M

E = 177(5) + 26(1)log c 61.0 3.2

verage of four differentmembranes 0.5 mm

E = 209(12) + 23(3)log c 106(1) 3(1)

ioelectronics 22 (2006) 145–152 147

ave a much lower potential variation in function of the atrazineoncentration, as seen in Fig. 1. This indicates that atrazine inter-cts only weakly with the NIP membrane because of the absencef specific molecular interaction sites. Besides, the potential ofhe not imprinted membrane is considerably lower than that ofhe MIP membrane, for example 24 mV instead of 99 mV.

The experiments reported in Fig. 1 were carried out incidic solution, HCl 0.031 mol l−1 (pH 1.5) containing KCl.02 mol l−1. Atrazine is at least partially protonated at pH 1.5,ccording to the protonation constant reported in the litera-ure (Smolkova and Pacakova, 1978; Skopalova and Kotoucek,995), while it is not protonated at neutral pH. Indeed not anyotentiometric response was obtained in neutral 0.02 mol l−1

Cl. This shows that the membrane potential depends on theoncentration of charged species, similarly to the usual ion selec-ive electrodes (Bard and Faulkner, 1980). The effect of theroton concentration will be further discussed below.

Other membranes were treated with 0.1 mol l−1 HCl, after thesual washing with acetic acid/methanol, and before the condi-ioning procedure. The E versus log cadd straight line is shownn Table 1. While the slope was similar to that obtained for the

embranes not treated with HCl 0.1 M, E0′and E(p) were about0 mV lower. In this case, 2 days were required for condition-ng in 3.6 × 10−4 mol l−1 atrazine solution. The lower measuredotential could be related to the higher protonation of the car-oxylic groups in the membranes treated in more acidic solution.nyhow the potentiometric response of the membranes treatedith 0.1 mol l−1 HCl to increasing atrazine concentration was

imilar to that of the not treated ones both in terms of slope andetection limit. After the HCl treatment, the selectivity of theembrane was improved, as it will be shown below.A lower potential was measured using a thicker membrane,

mm instead of 0.5 mm, as seen in Fig. 2. The equilibration timeas near to that of the thinner membranes, only a few seconds,

uggesting that the potentiometric equilibrium is established athe interface membrane–aqueous solution in the case of the con-itioned membranes.

ration.The sensors can be used for at least 1 month with no modifi-

ation of the potentiometric response. The sensors were stored

nd results from Eq. (3)

Linearized Eq. (1)

Slope Ord. origin E0′(mV)

∑K

potAtr,jcj (10−5)

3.4(1) × 107 1.9(1) × 103 228.2 3.5(2)1.3(1) × 104 0.5(1) 121.3 4.1(9)5.8(2) × 106 85(7) 205.8 1.5(4)

2.8(2) × 106 31(1) 196.4 1.1(0)

8(2) × 107 1.2(4) × 103 239(3) 1.9(8)

148 G. D’Agostino et al. / Biosensors and B

F0t

i0

cTtcsHoTtwtoatiftc

3

ctfatsitwcao1ieads

3a

i

capat

TS

S

S

2

A

ig. 2. Standardization curves of atrazine at different conditions. (©) Membrane.5 mm thick, pH 1.5; (×) membrane 0.5 mm thick, pH 7; (�) membrane 1 mmhick, pH 1.5.

n aqueous solution containing 3.6 × 10−4 mol l−1 atrazine,.02 mol l−1 KCl and 0.031 mol l−1 HCl.

Atrazine diffuses through the MIP membrane because of theoncentration difference at the two interfaces of the membrane.he diffusion was checked in the following way: the Teflon

ube carrying the membrane, previously conditioned in a con-entrated atrazine solution, was filled with 0.1 ml of atrazineolution 3.6 × 10−4 mol l−1, containing KCl 0.02 mol l−1 andCl 0.031 mol l−1, and the membrane was contacted with 1 mlf solution at the same composition, but not containing atrazine.he contact was maintained for 48 h at room temperature. After

his time, the concentration of atrazine in the outer solutionas 5 × 10−6 mol l−1. The determination was made spectropho-

ometrically at λ = 225 nm, at which the molar absorptivityf atrazine is 2.07 × 104 cm−1 l mol−1. This demonstrates thatctually some atrazine diffuses from the inner to the outer solu-ion. The diffusion rate is relatively low, since the concentration

n the inner and outer solution is still different after 48 h. This isavorable for the potentiometric determination, and is due in parto the absence of macropores in the membrane. The diffusionan affect the detection limit.

wzsd

able 2tandardization curve of atrazine MIP membranes 0.5 mm thick, in KCl 0.02 M and

Standardization lineE = E0′

add + Nlog cadd

E(p)

(mV)DL(10−5 M)

imazine membranewashed in methanol/acetic acid

E = 223(4) + 29(1)log c 95.6 4.65

imazine Membranewashed in methanol/acetic acid and treatedin HCl 0.1 M

E = 81(1) + 4.0(2)log c 62.70

-OH atrazine membranewashed in methanol/acetic acid and treatedin HCl 0.1 M

E = 68.5(2) + 0.64(5)log 66.6

metrine membranewashed in methanol/acetic acid and treatedin HCl 0.1 M

E = 68.2(1) + 0log c 68.1

ioelectronics 22 (2006) 145–152

.2. Selectivity of the atrazine MIP membrane

The effect of increasing concentrations of simazine (6-hloro-N2,N4-diethyl-1,3,5-triazine-2,4-diamine), a chlorinatedriazine with a structure very similar to atrazine, was evaluatedor a MIP membrane washed only with methanol/acetic acidnd for a membrane further treated with 0.1 mol l−1 HCl solu-ion before conditioning in concentrated atrazine solution. Thetandardization curve obtained for the membrane only treatedn methanol/acetic acid is reported in Table 2, it is similar tohat obtained for atrazine (see Table 1). On the contrary, thereas not any potential variation for simazine concentration in the

ase of the membrane treated with HCl 0.1 M. The results arelso reported in Table 2 for comparison. The same holds for twother triazines: ametryne (N2-ethyl-N4-isopropyl-6-methylthio-,3,5-triazine-2,4-diamine), which contains a –SCH3 groupnstead of a chlorine atom, and has protonation properties differ-nt from atrazine (log Ka = 4.1), and 2-hydroxy-atrazine bearingn –OH instead of –SCH3 or chlorine, which is the principalegradation product of atrazine. Evidently the membrane is morepecific for atrazine after treatment in HCl 0.1 mol l−1.

.3. Response of the MIP membrane based sensor totrazine concentration

Some typical Nernstian plots, E versus log cadd are reportedn Fig. 2.

At pH 7, the slope is always near to zero, while at pH 1.5 theurve is composed of two straight lines, one of which, at lowertrazine concentration, has slope near to zero and the other has aositive slope near to 25 mV/decade. The slope and the ordinatet the origin determined by fitting the linear part of the calibra-ion plot to the Nerst equation are reported in Table 1, together

ith the potential corresponding to the part with slope equal to

ero (E(p)). The detection limit (DL), i.e. the intercept of the twotraight lines (Buck and Lindner, 1994) is also reported. Fourifferent membranes were synthesized and used at similar con-

pH 1.5, for different triazines

Linearized Eq. (1)

Slope (107) Ord. origin E0′(mV) ΣK

potAtr,jcj (10−5)

3.75 Not significantlydifferent from 0

230 Not significantlydifferent from 0

and B

doed

tmwetbHdr1s

iavnao

tcf

wbcrtc

β

ttTabmbK

d

tic

1

e

rftftc

sibsi

losi

3

bF

Aciof the curve E versus log cadd with slope different from zero arereported in Table 3, together with those obtained from Eq. (3). Atthe pH of the experiments, the liquid junction potential variationis low, so that not any correction was made.

G. D’Agostino et al. / Biosensors

itions in a short time (2 days) to investigate the reproducibilityf the membranes. The average values obtained for the param-ters are reported in Table 1, with the corresponding standardeviation, which is relatively low.

The slope (N) of the Nernstian part of the standardiza-ion curve E versus log cadd is similar for all the considered

embranes, and is near to 25 mV/decade. The same slopeas obtained with an ISFET/MIP based sensor, with a differ-

nt imprinted polymer (Pogorelova et al., 2002). According tohe classical model for ISE this should indicate that a dou-ly charged ion is responsible for the potentiometric signal.owever atrazine is not doubly protonated at the considered con-itions, as calculated from the protonation constant of atrazineeported in the literature, log Ka = 1.7 (Smolkova and Pacakova,978; Skopalova and Kotoucek, 1995). Only 63% of atrazine inolution is monoprotonated at pH around 1.5.

E0′ is noticeably different in some of the experiments reportedn Table 1. For example, when an internal solution at lowertrazine concentration is used, a lower potential is obtained. Thisariation corresponds to the concentration difference in the inter-al solution. The detection limit is only slightly affected by thetrazine concentration in the internal solution, probably becausef the very slow diffusion rate.

Considering that interfering cations may be present, and thathe potentiometrically active form of atrazine is ionic, say z-harged or z-protonated, the following relationship can be usedor modeling the potentiometric response of the electrode:

E = E0 + N ln([HzAz+] + ∑K

potAtr,jc

z/zIj )

E = E0 + N ln

(cadd

αHzA+

∑K

potAtr,jc

z/zIj

)

N = RT

nF

(1)

here z is the charge of the species which actually distributesetween the two phases, HzAz+. Kpot

Atr,j is the atrazine/jion (withharge zj) potentiometric selectivity coefficient, and αHzA is theatio of the total atrazine concentration in the solution phase tohe concentration of the z-protonated form (side reaction coeffi-ient of protonated atrazine):

αHzA =∑

βza[H]z

βza[H]z

cAtr = ∑[HzAz+] = [HzAz+]

∑βza[H]z∑βza[H]

βza = [HzAz+]

[H]z[A](2)

za is the z-protonation constant of atrazine, and the summa-ion is extended from z = 0 to n (n is the maximum protona-ion number), and β0a = 1. Charges are omitted for simplicity.he concentration of the active species depends on the totaltrazine concentration, and on the acidity of the solution. Possi-le interfering species are protons and atrazine released by theembrane, the concentration of which, indicated by the sym-

ol cr, is unknown. Its potentiometric selectivity coefficient ispotAtr,Atr = 1.Eq. (1) may be written in the following linearized way, to

ifferentiate the two terms in the logarithm, one depending on

F0(

ioelectronics 22 (2006) 145–152 149

he atrazine concentration added to the solution and the otherndependent of it, but possibly depending on the atrazine con-entration released by the membrane:

0E/N = 10E0/N

(cadd

αHzA+

∑K

potAtr,jc

z/zIj

)(3)

N is assumed to be 29.5 mV/decade at 25 ◦C, near to thexperimental value.

The results obtained by the linearization procedure are alsoeported in Table 1. The logarithm of the slope of the linearizedunction is E0 − N log αHzA, equal to that obtained directly fromhe function E versus log cadd(E0′). It is also reported in Table 1or comparison. The differences are due to the slope of the func-ion, which is slightly different from 29.5, as seen in the secondolumn of Table 1.

The term accounting for the interferences, expressed as theummation

∑K

potAtr,jcj , obtained from the parameters of Eq. (3),

s reported in Table 1. It is similar in the different experiments,ecause the composition of the solutions was the same. It islightly lower in the case of lower atrazine concentration in thenternal solution.

The detection limit, defined as the intercept of the two straightines in the E versus log cAtr plot, is relatively high. On the basisf the experiments described in this paragraph the interferingubstances responsible for this high detection limit cannot bedentified.

.4. Effect of proton concentration on the sensor response

Some curves E versus log cadd obtained with the same mem-rane, but at different acidity of the test solution are reported inig. 3.

It can be seen that the potential is higher at higher acidity.t pH 4, the potential does not increase at increasing atrazine

oncentration, while at lower pH the curve E versus log cadds composed of two straight lines. The parameters of the part

ig. 3. Standardization curves of atrazine at different acidities. Membrane.5 mm thick, treated in HCl 0.1 M. (�) pH 1.0; (�) pH 1.45; (�) pH 1.8;*) pH 4.0.

150 G. D’Agostino et al. / Biosensors and Bioelectronics 22 (2006) 145–152

Table 3Standardization curve of atrazine MIP membrane 0.5 mm thick, treated in 0.1 M HCl, in KCl 0.02 at different pH, and results from Eq. (3)

Standardization lineE = E0′

add + Nlog cadd

E(p) (mV) DL (10−5 M) Linearized Eq. (1)

Slope (107) Ord. origin (102) E0′(mV)

∑K

potAtr,jcj (10−5)

ppp

dwoaiep

E

p

E

to

irnsotettvi

cshi

atbitt1tiaoo

aarpi

pptewita

3

saultspem1c

The membrane electrical resistance at different atrazine con-centration in the external solution is reported in Fig. 5. Themembrane resistance is relatively low, and decreases at concen-tration of atrazine higher than 3 × 10−5 mol l−1, which is near

H 1.00 E = 218(10) + 23(3)log c 115 3.30H 1.40 E = 214(6) + 25(1)log c 104 4.90H 1.85 E = 198(9) + 24.2(2)log c 96 6.60

The slope of the curve E versus log cadd is almost indepen-ent of the pH, while the ordinate at the origin (E0′) decreasesith increasing pH. This can be ascribed to the protonationf atrazine, if the z-protonated form, is that potentiometricallyctive. Its concentration becomes higher at higher proton activ-ty, so that the corresponding potential should also increase, asxperimentally observed. If the completely deprotonated formrevails, E0′ depends on the solution pH as here reported

0′ = E0 + N log βza − N z pH

The plot of E0′ versus pH is actually a straight line, with thesearameters obtained by least square method:

0′ = 269.2(6) − 27.3(4)pH

The expected slope is 59.16 mV/decade, since z correspondso the charge of the ionic form of atrazine. This is the same effectbserved in the case of the E versus log catr plot.

The difference between the constant potential E(p) and E0′s constant at different pH, around 105 mV, a value welll cor-esponding to Nlog(DL). It can be deduced that protons doot influence significantly the detection limit. The other pos-ible interfering ion is the atrazine released by the membrane,r diffusing from the internal solution to the sample solution,he potentiometric selectivity coefficient of which, K

potAtr,Atr, is

qual to 1, as atrazine is the primary ion. So the interferenceerm represents the concentration released by the membrane athe interface membrane/aqueous solution. This is a reasonablealue compared to the detection limit, and is near to that reportedn Table 1, obtained at fixed pH.

Since the detection limit increases slightly at decreasing H+

oncentration, it is possible that more atrazine is released inolution at higher pH. This could explain why at sufficientlyigh pH, for instance at pH 4, the membrane potential does notncrease at increasing atrazine concentration.

The effect of the acidity on the potentiometric response of thetrazine MIP membrane is seen also in Fig. 4, in which the poten-ial in function of the solution pH is reported for different mem-ranes, conditioned in the usual way but not containing atrazinen the test solution, so that the measured potential correspondso E(p). The cell potential reported in ordinates is corrected forhe liquid junction potential variation at pH lower than around.2, as reported in the experimental part. At pH lower than 2.0he potential increases at decreasing pH, while at higher pH it

s constant. The potential versus pH function for pH lower thanround 2 is a straight line with slope near to −23 mV/decade, andrdinate at the origin approximately 135 mV. One of these lines,btained from one of the experiments in Fig. 4, is here given

Fa

15(1) 23(6) 241 1.5(4)5.9(4) 5(2) 229 0.9(4)2.4(2) 4(1) 218 1.5(4)

s an example: E = 136(1)–24(1)pH (R2 = 0.991). These resultsre in good agreement with those obtained in the experimentseported above, and the explanation is that the concentration ofrotonated atrazine, which is the potentiometrically active form,s higher at higher acidity.

The effect of the pH variation on a NIP membrane is com-letely different, as shown in Fig. 4 for comparison. Here theotential increases at increasing pH, contrary to what observed inhe case of the MIP membrane. This could indicate that protonsasily diffuse through the membrane not containing atrazine,hile their mobility is reduced in presence of atrazine. This

ncreases the conductivity of the membrane. It will be seen belowhat actually the resistance of the membrane is lower at highertrazine concentration in the solution phase.

.5. Electrical characterization of the membrane

The MIP membrane was characterized by the impedancepectroscopy technique (De Marco and Pejcic, 2000) at differenttrazine concentration in the external solution. The instrumenttilized was a Solartron SI 1260 impedance/gain-phase ana-yzer, in the 0.1 Hz–10 MHz frequency range. The usual cell,hat reported in Scheme 1, was investigated. The impedancepectrum shows that only one resistance and capacitance inarallel configuration constitute the equivalent circuit of thelectrochemical system considered, to be attributed to the poly-eric material. The electrical capacitance of the membrane was

3.4 pF, and does not significantly vary in function of the atrazineoncentration.

ig. 4. Dependence of the potential of cell 1 on the pH of the test solution. Nodded atrazine. (�,�,©) MIP membrane 0.5 thick; (*) NIP membrane 0.5 thick.

G. D’Agostino et al. / Biosensors and B

Fc

tvcas

4

oibwmAal

stttnsevapadrbTbttp

A

(

R

B

B

B

C

D

F

H

J

K

L

L

M

M

M

O

P

P

P

P

SL.M., Panasyuk, T.L., El’skaya, A.V., 1999. Conductimetric sensor for

ig. 5. Resistance of the conditioned MIP membrane at different atrazine con-entration obtained by impedence spectroscopy.

o the detection limit of the potentiometric determination. Theariation of the resistance is almost linear at the considered con-entrations. The decrease of the membrane resistance with thetrazine concentration is probably sufficiently high to reduce thelope of the Nernst function with respect to 59.16 mV/decade.

. Conclusions

A potentiometric sensor for atrazine was developed, basedn a molecular imprinted polymer membrane. It works as anon selective electrode highly specific for atrazine. The mem-rane was rigid enough to bear the filling solution in contactith the internal reference electrode. To obtain a good potentio-etric response, the membrane must be previously conditioned.n interesting characteristics of the potentiometric sensor is thatshort time, only a few seconds, is required to reach the equi-

ibrium potential.Since the potential variation is produced by the charged

pecies, the solution pH must be lower than pH 1.7 to ensurehe protonation of atrazine,. Protons do not interfere accordingo an ion exchange mechanism, but they have an influence onhe measured potential, since the concentration of the proto-ated species of atrazine increases with increasing acidity. Thelope of the potentiometric curve E versus log cadd is lower thanxpected for a monocharged ion. This is probably due to theariation of the electrical resistance of the membrane with thetrazine concentration. Atrazine is sorbed more strongly at lowH, at which the carboxylic groups in the polymeric membranere completely protonated, as demonstrated by the fact that theetection limit slightly decreases with pH. This is probably theeason for which at pH higher than approximately 1.7 the mem-rane potential does not change with the atrazine concentration.he detection limit is around 2 × 10−5 mol l−1, and it appears toe determined by the distribution coefficient of atrazine betweenhe acidic aqueous solution and MIP. Thus the best way to lowerhe detection limit is using membranes with stronger sorbingroperties at the considered conditions.

cknowledgment

This work was financially supported by the project FIRB01MURST, Italy).

S

ioelectronics 22 (2006) 145–152 151

eferences

ard, A.J., Faulkner, L.R., 1980. Electrochemical Methods. John Wiley &Sons, New York (Chapter 2.4).

jarnason, B., Chimuka, L., Ramstrom, O., 1999. On-line solid-phase extrac-tion of triazine herbicides using a molecularly imprinted polymer forselective sample enrichment. Anal. Chem. 71, 2152–2156.

uck, R.P., Lindner, E., 1994. Recommendations for nomenclature of ion-selective electrodes (IUPAC Recommendations 1994). Pure Appl. Chem.66, 2527–2536.

raggs, A., Moody, G.J., Thomas, J.D.R., 1974. PVC matrix membrane ion-selective electrodes. J. Chem. Educ. 51, 541–544.

e Marco, R., Pejcic, B., 2000. Electrochemical impedance spectroscopyand X-ray photoelectron spectroscopy study of the response mechanismof the chalcogenide glass membrane iron(III) ion-selective electrode insaline media. Anal. Chem. 72, 669–679.

errer, I., Lanza, F., Tolokan, A., Horvath, V., Sellergren, B., Horvai, G.,Barcelo, D., 2000. Selective trace enrichment of chlorotriazine pesticidesfrom natural waters and sediment samples using terbuthylazine molecu-larly imprinted polymers. Anal. Chem. 72 (16), 3934–3941.

utchins, R.S., Bachas, L.O., 1995. Nitrate-selective electrode developedby electrochemically mediated imprinting/doping of polypyrrole. Anal.Chem. 67, 1654–1660.

enkins, A.L., Yin, R., Jensen, J., 2001. Molecularly imprinted polymersensors for pesticide and insecticide detection in water. Analyst 126,798–802.

itade, T., Kitamura, K., Konishi, T., Takegami, S., Okuno, T., Ishikawa, M.,Wakabayashi, M., Nishikawa, K., Muramatsu, Y., 2004. Potentiometricimmunosensor using artificial antibody based on molecularly imprintedpolymers. Anal. Chem. 76, 6802–6807.

anza, F., Sellergren, B., 1999. Method for synthesis and screening oflarge groups of molecularly imprinted polymers. Anal. Chem. 71, 2092–2096.

iang, C., Peng, H., Nie, L., Yao, S., 2000. Bulk acoustic wave sensor forherbicide assay based on molecularly imprinted polymer. Fresenius J.Anal. Chem. 367, 551–555.

atsui, J., Miyoshi, Y., Doblhoff-Dier, O., Takeuchi, T., 1995. A molecularlyimprinted synthetic polymer receptor selective for atrazine. Anal. Chem.67, 4404–4408.

atsui, J., Okada, M., Tsuruoka, M., Takeuchi, T., 1997. Solid phase extrac-tion of a triazine herbicide using a molecularly imprinted synthetic recep-tor. Anal. Commun. 34 (3), 85–87.

uldoon, M.T., Stanker, L.H., 1997. Molecularly imprinted solid phaseextraction of atrazine from beef liver extracts. Anal. Chem. 69, 803–808.

lsen, J., Martin, P., Wilson, I.D., 1998. Molecular imprints as sorbents forsolid phase extraction: potential and applications. Anal. Commun. 35 (10),H13–H14.

anasyuk-Delaney, T., Mirsky, V.M., Ulbricht, M., Wolfbeis, O.S., 2001.Impedometric herbicide chemosensors based on molecularly imprintedpolymers. Anal. Chem. Acta 435, 157–162.

iletsky, S.A., Piletskaya, E.V., El’skaya, A.V., Yano, K., Karube, I.,Parhometz, Y.P., El’skaya, A.V., 1995. Atrazine sensing by molecularlyimprinted membranes. Biosens. Bioelectron. 10, 959–964.

iletsky, S.A., Piletskaya, E.V., Sergeyeva, T.A., Panasyuk, T.L., El’skaya,A.V., Levi, R., Karube, I., Wulff, G., 1998. Imprinted membranes forsensor technology—opposite behavior of covalently and noncovalentlyimprinted membranes. Macromolecules 31, 2137–2140.

ogorelova, S.P., Bourenko, T., Kharitonov, A.B., Willner, I., 2002. Selectivesensing of triazine herbicides in imprinted membranes using ion-sensitivefield-effect transistors and microgravimetric quartz crystal micro. Analyst127, 1484–1491.

ergeyeva, T.A., Piletsky, S.A., Brovko, A.A., Slinchenko, E.A., Sergeeva,

atrazine detection based on molecularly imprinted polymer membranes.Analyst 124, 331–334.

kopalova, J., Kotoucek, M., 1995. Polarographic behaviour of same s-triazine herbicides and their determination by adsorptive stripping voltam-

1 and B

S

52 G. D’Agostino et al. / Biosensors

metry at hanging mercury drop electrode. Fresenius J. Anal. Chem 351,650–655.

molkova E.Jr., Pacakova, V., 1978. Liquid chromatographic separation andbehaviour of some substituted s-triazines on a CN-bonded stationaryphase. Chromatographia 11, 698–702.

T

Z

ioelectronics 22 (2006) 145–152

akeuchi, T., Fukuma, D., Matsui, J., 1999. Combinatorial molecular imprint-ing: an approach to synthetic polymer receptors. Anal. Chem. 71,285–290.

hou, Y., Yu, B., Levon, K., 2005. Potentiometric sensor for dipicolinic acid.Biosens. Bioelectron. 20, 1851–1855.