spectral imaging and electrochemical study on the response mechanism of ionophore-based polymeric...

Feature Article

Spectral Imaging and Electrochemical Study on the ResponseMechanism of Ionophore-Based Polymeric MembraneAmperometric pH SensorsRobert Long, Eric Bakker*

Department of Chemistry, Auburn University, Auburn, Alabama 36849;* e-mail: [email protected]

Received: October 29, 2002Final version: February 11, 2003

AbstractThe response features of amperometric ionophore-based chemical sensors were evaluated in regular ion-selectivemembranes, poly(vinyl chloride) plasticized with o-nitrophenyl octyl ether, with a H�-selective chromoionophore asmodel ionophore. Direct imaging experiments upon discrete applied voltage pulses revealed that the diffusioncoefficients in such membranes are very similar to that found in membranes under zero current conditions, at about10�8 cm2 s�1. In the potential range that defines the limiting current, it was found by these imaging experiments thatthe ionophore is saturated at the membrane surface, as expected. The observed concentration profiles could be welldescribed with established diffusion theory. At more extreme potentials, the observed diffusion profiles weredifferent, and could be explained by the extraction of additional, uncomplexed hydrogen ions into the membranephase. The results were correlated to the amperometric response behavior of the same electrodes. At the limitingcurrent, Cottrell fits were performed, and similar diffusion coefficients were found as in the imaging experiments. ThepH-dependent normal pulse voltammetric responses were analyzed in terms of their half wave potential, and found toshow a Nernstian pH response, suggesting that they function in close analogy to their potentiometric counterparts.The breakdown of Nernstian pH response at low pH is explained by extraction of electrolyte into the membraneunder these conditions, which renders the interface non-polarizable and leads to loss of voltammetric features. Theresults adequately support the previously published response mechanism of such ionophore-based amperometricsensors, including the origin of the limiting current, the capability of such membranes to be responsive to ion activities,the requirement of applying baseline potentials between pulses, and the reason for the altered selectivity at extremepotentials.

Keywords: Spectral imaging, Amperometric pH sensing membranes, Cottrell equation, Normal pulse amperometry

1. Introduction

Amperometric sensors based on ion transfer voltammetry,inspired by the important work of Koryta [1], have beenstudied by growing number of researchers, for example byGirault for the miniaturization and detection in fluidicsystems [2 ± 4], by Cammann for miniaturization [5], byHorvath and Horvai to study the voltammeric behavior ofplasticized membranes and the relationship to ion-selectiveelectrodes [6, 7], by Marecek for the determination ofionophores [8], by Senda for introducing pulsed amperom-etry and gelified membranes with more stable interfaces [9,10], by Sawada and Osakai for lithium detection in artificalserum [11], and Samec for fundamental characteristics ofion-selective membrane materials [12]. They are known tofunction inmanyways in analogy tometal electrodes [3, 13].Nonetheless, the underlying chemistry is very different,since the applied potential forces the ion of interest todistribute unevenly across a liquid ± liquid interface, oftenwith the assistance of an ionophore dissolved in the organicphase. Unfortunately, potentiometric and amperometricsensors based on the same membrane materials and

ionophores have rarely been compared to each otherdirectly by the same research groups. Indeed, ion-selectiveelectrodes have long ago successfully transitioned fromliquid membrane electrodes used in academic settings tominiaturized, polymer-based, engineered devices that areused in clinical laboratories all over the world [14]. Incontrast, the corresponding amperometric sensors haveonly recently been evaluated in fluidic platforms [15] orwithmodified materials [5, 16, 17]. This discrepancy in the speedand direction of development is perhaps partly due to thedifferent academic environments pursuing research in thoseareas, with analytical chemists focusing more on potentio-metric sensors, and fundamental physical chemists on theiramperometric counterparts. On the other hand, the sensingmaterial of amperometric sensors was required to be asimple organic liquid phase in order to guarantee a lowresistance and high diffusion coefficients in the organicphase. Such membranes are known to respond to ionicspecies with very similar response mechanisms as theirmetal electrode counterparts. Consequently, recent effortshave focused on following the advances made with metalelectrodes, by introducing microinterfaces for enhanced

1261

Electroanalysis 2003, 15, No. 15-16 ¹ 2003 WILEY-VCH Verlag GmbH&Co. KGaA, Weinheim DOI: 10.1002/elan.200302806

mass transport and consequently achieving lower detectionlimits [15]. This approach requires one to deal with simpleorganic solvents, which are difficult to handle and are proneto leach into the sample, and that must limit the lifetime ofthe sensors. Success was obtained in gelifying the organicliquidswith a small amount of poly(vinyl chloride) for easierhandling and to obtain a sharper liquid ± liquid interface [10,18].Recently, we have started to study amperometric ion

sensors based on the very samemembranematerials as theirpotentiometric counterparts in an effort to critically com-pare both transduction principles to each other. Owing tothe high polymer content in the plasticized PVC mem-branes, the diffusion coefficients of extracted ions and otheractive membrane components are, with about 10�8 cm2 s�1,known to be much smaller than in pure organic liquids [19].As a consequence, it was suggested that the currentobserved in the amperometric experiments is typicallygiven by limited mass transport in the organic, not aqueousphase [20]. Koryta has previously also described systems(containing the ionophore monensin) where the sameprocesses was assumed to be rate limiting, by utilizing alow ionophore concentration in the organic phase and a highelectrolyte concentration in the aqueous phase [21]. Be-cause of this alternate response mechanism, it was postu-lated that these amperometric sensors resemble the corre-sponding ion-selective electrodes much more closely thanmany previously studied systems [22, 23]. The membranescontained no added ion-exchanger, but a high concentrationof inert lipophilic salt. Normal pulse voltammetry, ratherthan cyclic voltammetry, was suggested as a preferredinterrogation method for these sensing membranes sincerepeatable current readings require the re-establishment ofa membrane void of extracted ions between discretepotential pulses [23]. The sensing selectivity was found tobe dependent on the magnitude of the observed current,which was used for multianalyte detection purposes [22]. Atheoretical model was developed that explained the exper-imental results satisfactorily. In particular, the observedcurrent in a normal pulse voltammetric experiment wascompared to the function of the concentration of ion-exchanger in a potentiometric membrane [22].This article introduces a more direct approach to under-

standing amperometric ion sensors based on ionophoresdissolved in hydrophobic polymeric membranes. Real timespectroscopic imaging experiments were performed on pHresponsive ion-selective membranes containing a H�-selec-tive chromoionophore. The results are explained in light ofthe current response theory for these amperometric sensors,and correlated to the corresponding normal pulse voltam-metric responses. The excellent selectivity [24], imagingcapability [19] and well studied behavior in ion-selectivemembranes [25] make these H�-chromoionophores idealcandidates for a critical comparative study of amperometricand potentiometric transduction principles.

2. Experimental

2.1. Reagents

The salts used for aqueous solutions were puriss quality orbetter and were dissolved in Nanopure distilled water. Highmolecularweight poly(vinyl chloride) (PVC),o-nitrophenyloctyl ether (NPOE), tetradodecylammonium tetrakis(4-chlorophenyl) borate (ETH 500), Selectophore gradetetrahydrofuran (THF), 9-(diethylamino)-5-octadecanoyli-mino-5H-benzo[a]phenoxazine (ETH 5294), potassium tet-rakis[3,5-bis(trifluoromethyl)phenyl] borate (KTFPB), cit-ric acid, boric acid and all salts were obtained from FlukaChemical Corp. (Milwaukee, WI). Sodium hydroxide,sodium phosphate monobasic and sodium phosphate diba-sic were obtained from Fisher Scientific (Fair Lawn, NJ).

2.2. Membrane Preparation

All membranes were prepared by dissolving PVC andNPOE (1 :2 by weight) into Selectophore grade THFalongwith various amounts of two other components. Forpotentiometric membranes, 10 mmol/kg of the chromoio-nophore ETH 5294 and 5 mmol/kg KTFPB was added. Theamperometric ISE membranes contained 10 wt% ETH 500and 10 mmol/kg ETH 5294, while the ring membranes wereprepared in complete analogy, but with 0.5 mmol/kg ETH5294 to reduce the optical density. This cocktail was shakenmechanically for a fewminutes and then poured into a glassring with an inner diameter of 22 mm mounted on amicroscope slide. The solvent THF was allowed to evapo-rate overnight leaving membranes that were approximately200 �m thick.

2.3. Instrumentation

A Model AFRDE5 potentiostat (Pine Instruments, GroveCity, PA) was used with a three electrode system tomeasurecurrent versus applied potential. The Normal Pulse Am-perometric data were recorded using LabView 5.0 software(National Instruments, Austin, TX) on a Macintosh com-puter equipped with a 16-bit data acquisition board (Na-tional Instruments, Austin, TX) [23]. The imaging data werecollected using a Nikon Eclipse E400 microscope equippedwith a Pariss imaging spectrometer (LightForm, Inc., BelleMeade, NJ) and a Model 4920 Peltier cooled CCD camera(Cohu, Inc., San Diego, CA) [26].

2.4. Electrodes

A saturated KCl, Ag/AgCl reference electrode containing a1.0 M lithium acetate bridge electrolyte was used for allmeasurements. The potentiometric and amperometricworking electrodes were prepared by mounting a piece ofthe ion-selective membrane onto a Philips electrode body

1262 R. Long, E. Bakker

Electroanalysis 2003, 15, No. 15-16 ¹ 2003 WILEY-VCH Verlag GmbH&Co. KGaA, Weinheim

(IS-561, Glasbl‰serei Mˆller, Z¸rich, Switzerland) with ani.d. of 4.0 mm and conditioning the electrode overnight inthe 0.1 M NaCl inner filling solution. This inner solutioncomposition is not optimal for the potentiometric sensor,but was chosen for maximum consistency in the experimen-tal comparison. The potentiometric measurements weremade with a two electrode setup, while the amperometricsystem required a three-electrode setup with a Pt cagecounter electrode immersed in the sample with the ion-selective working electrode. The sample was 0.10 MNaCl in6.6 mMcitric acid, 11 mMboric acid and 10 mMphosphoricacid, buffered with varying amounts of HCl to encompassthe desired pH range.The imaging experiments were done in a cell with a small

Ag/AgCl working electrode separated from the referenceelectrode and much larger Ag/AgCl counter electrode bythe ring shaped ISE membrane. Otherwise, this setup issimilar to that reported in the literature for spectropoten-tiometric studies [19]. The membrane was cut into a donut-shaped ring with an o.d. of 8 mm and an i.d. of 6 mm andmounted onto a polished Plexiglass cell bottom, in thecenter, forming two concentric rings separating the IFS andidentical sample. The outer circle (34 mm i.d.) was a circularrubber gasket. The cell was covered with a 2�� 1/8�� piece ofquartz, fitted with a plastic cover slip and mounting screws.The rubber gasket and ISE membrane form two chambersbetween the Plexiglass and quartz which are filled throughchannels with an identical 0.1 M NaCl solution buffered topH 6.8 with 10 mMNa3HPO4� 10 mMNaH2PO4.Ag/AgClwire working and counterelectrodes were inserted throughthe fluidic channels and placed in close vicinity of the ringmembrane and connected to the potentiostat. The cell wasthen mounted under a microscope and a section of the ISEmembrane, approximately 430 �m deep, was studied intransmission mode as a function of time at a specifiedapplied potential. The spectral camera recorded trans-mission spectra between 390 and 860 nm. Absorbance datawere calculated by comparing the initial transmittance ofthe cell without mounted membrane ring to the trans-mittance of the final setup at a given time. For the imagingexperiment polarized at � 1000 mV, the absorbance changewas calculated relative to the transmittance with mountedmembrane before application of the voltage pulse (in whichcase the chromoionophore was in its deprotonated form).This approach gave similar results, but with a lower noise inthe relative absorbance profiles.

3. Results and Discussion

The response mechanism of polymeric membrane ampero-metric ion sensors was studied here with the H�-selectivechromoionophore ETH 5294 as a model system. Thischromoionophore has been thoroughly studied earlier foruse in optical and potentiometric polymeric sensors andlikely represents an adequate system to study the mecha-nism of such amperometric sensors. Potentiometric sensorsbased on ETH 5294 are known to yield excellent pH

electrodes owing to the high selectivity of this chromoiono-phore to hydrogen ions [25, 27]. On the other hand, theabsorbance and fluorescence properties of this compoundhave repeatedly been used to fabricate optical sensing filmson the basis of competitive ion-exchange or coextractionequilibria [24, 26, 28]. In these cases, the extent of proto-nation of the chromoionophore is followed spectroscopi-cally and related to the ion concentration of interest. Thissame chromoionophore has recently also been utilized toimage the concentration profiles within ion-selectivemembranes under zero current conditions [19, 29]. Adrastic pH change on one membrane side, for example,promotes either the exchange with interfering ions [19] orthe coextraction of hydrogen ions and anions [29], whichgradually alters the concentration profiles within the entiresensing film. These processes were imaged in real time andyielded diffusion coefficients of the chromoionophore aswell as an improved fundamental understanding of poten-tiometric sensors.In thematrix of interest, PVC plasticized with o-NPOE in

a ratio of 1 :2, the chromoionophore ETH 5294 has recentlybeen shown to exhibit negligible binding affinity to alkalimetals [30]. Its pKa value in this hydrophobic environmenthas been determined as 14.8 by utilizing the so-calledsandwich membrane method [30]. Its lipophilicity is nor-mally adequate in basic and neutral pH values, but has beenreported to be insufficient under acidic sample conditions[24]. The absorbance maxima for the protonated andunprotonated forms are around 660 nm and 560 nm, re-spectively [24].With the exception of its somewhat less thanoptimal lipophilicity, this chromoionophore is expected tobe an adequate system for fundamental studies of ampero-metric sensors based on assisted ion transfer into apolymeric membrane.In accordance with earlier work involving alkali metal

selective ionophores [20, 22], membranes containing ETH5294 were measured with normal pulse voltammetry. Thepurpose of alternating each incrementally applied potentialwith a baseline potential (here at 0 V) is to re-establish themembrane surface region devoid of extracted ions beforeeach potential pulse. Figure 1 shows a typical normal pulsevoltammogram for a PVC±NPOE membrane containingthe chromoionophore ETH 5294 in contact with a pH 6.0solution. As with earlier studies on the basis of otherionophores, a fairly wide limiting current region is observedbetween � 600 and � 800 mV. It has been suggested thatthe current in these systems is predominantly dictated by thelimited mass transport of ions from the sample ±membraneinterface into the polymer bulk (and not from the samplebulk to the membrane surface). Consequently, the limitingcurrent region shown in Figure 1 is believed to reflect thelimited availability of ionophore in the membrane. As itsconcentration is exhausted, the interfacial concentration ofextracted hydrogen ions can no longer increase, even atmore extreme applied potentials. Eventually, the appliedpotential is sufficiently large so that an ionophore-mediatedprocess is no longer necessary to extract sample cations inthe membrane. This is thought to mark the end of the

1263Ionophore-Based Polymeric Membrane Amperometric pH Sensors


limiting current region shown in Figure 1, and the selectivityof the membrane is now no longer reflected by theionophore binding affinity but by the relative lipophilicityof the cations. Indeed, it results in a drastic selectivity change(lipophilic ions are preferred over hydrophilic ones) relativeto the voltammetric response at lower currents (whereionophore-mediated selectivity is observed). The effect wasexploited to use suchmembranes formultianalyte detectionpurposes [22].Figure 2A shows the current response upon three applied

potentials of increasing magnitude, separated by a longerbaseline potential, in the range of the limiting current(� 500 mV). Obviously, the current decays continuouslywithin one pulse and does not appear to reach a limitingvalue. With classical amperometric metal electrodes, thelimiting current is given by mass transport from the samplebulk to the electrode surface, and the processes can bedescribed in an analogous fashion. In that case, however, asteady-state concentration profile across the stagnant dif-fusion layer, and hence a steady-state current, is eventuallyachieved. Here, it is believed that the rate limiting step isdiffusion into the membrane interior. Diffusion coefficientsare known to be rather small (on the order of 10�8 cm2 s�1

[19]), and themembrane thickness is on the order of 200 �m.It would take many hours to reach a steady state current. Inprinciple, the current response for this situation (at thelimiting current) may be described in complete analogy to apotential step experiment with a classical metal electrode:the interfacial concentration is suddenly forced from zero to

a constant valuedue to the appliedpotential, and the currentis given by diffusion of the reaction product away from theinterface into the sample bulk. The Cottrell equation isknown to describe this process (it is normally formulated todescribe a potential step experiment at a metal electrode):

i t� � � nFAD1�2C�

�1�2t1�2�1�

where i(t) is the observed current as a function of time, n isthe number of electrons (in this case, the charge of�1 of thehydrogen ion), F is the Faraday constant, A is the surfacearea of the electrode,D is the diffusion coefficient,C* is theconstant phase boundary concentration of the diffusingspecies (in this case that of the protonated chromoiono-phore), and t is the time after applying the potential step.With the experiment shown in Figure 2A, the interfacialconcentration C* in Equation 1 is approximated by thechromoionophore concentration (assuming that the diffu-sion coefficients of protonated and unprotonated chro-moionophore are similar). The corresponding Cottrell fit(current versus the inverse of the square root of time) for theuptake current at � 700 mV is shown in Figure 2B. A linearrelationship is found at times larger than 200 ms, and a leastsquares analysis for the last 500 ms of the pulse yields adiffusion coefficient of 1.0� 10�8 cm2 s�1. This value is inexcellent agreement with literature values found underzero-current conditions [19]. This suggests that electrical

Fig. 1. Normal pulse voltammogram for a PVC±NPOE membrane containing ETH 5294 and the inert lipophilic salt ETH 500 incontact with a 0.1 M NaCl sample buffered at pH 6.0. Inner solution: 0.1 M NaCl. Uptake times: 1 s; Stripping times: 30 s at 0.0 V.



migration is a minor contribution to the mass transport of theprotonated chromoionophore in the membrane, likely be-cause of the very high concentrations of inert lipophilic saltused in themembrane. In this experiment, the influence of theiR drop across the membrane bulk was not compensated for,and this could introduce an error. However, in the limitingcurrent region, this error is expected to have only a smallinfluence on the Cottrell experiment since a small change inthe applied phase boundary potential would have littleconsequence on the phase boundary concentration of thediffusing species. This would not be the case with potentialvalues outside the limiting current window. The Cottrell

experiment was here only analyzed at times larger than 0.5 s,where the current decay was comparatively small. With amembrane resistance of 10 kOhm, the observed currentchange of 8 �A (see Figs 2 and 3) would give a potentialchangeof ca. 80 mV.This ismuch smaller than thewidthof thelimiting current plateau (see Fig. 1) and perhaps sufficentlysmall to ensure that the phase boundary concentrations didnot change appreciably in the course of the experiment.The hydrogen ion extraction and diffusion processes into

polymeric membranes doped with ETH 5294 upon theapplication of an external voltage were further studied bydirect spectroscopic imaging. In analogy to earlier zero-

Fig. 2. A) Current responses for experiment shown in Figure 1 in the indicated potential range (limiting current region). B) Cottrell fitfor current response shown in Figure 2A upon application of � 700 mV (see Eq. 1). Other conditions as in Figure 1.

A)

B)



current experiments [19], the setup involved a ring-shapedion-selective membrane that could be imaged under themicroscope. The ring membrane separated two bufferedsolutions that were each in contact with aworking and a pairof reference and counter electrodes, in analogy to earlierreports [22]. Time-dependent absorbance spectra wererecorded with an optical microscope fitted with a UV/visspectrometer and CCD detector.Figure 3 (top) shows spatially resolved absorbance spec-

tra as a function of time upon the application of a constantpotential of � 500 mV. The profiles were described with theknown diffusion equation [31]:

A x� t� ��A1xl� 2�

��n�1

A1cos�n��n

sinn�xl

� �exp �Dn2�2t

l2

� �

(2)

where A(x,t) is the absorbance as a function of distance xfrom the interface and time t andAl is the absorbance of thedisturbance at distance l.

All theoretical curves shown in Figure 3 (top) wereobtained with Equation 2, with the parameters Dorg�1.05� 10�8 cm2 s�1 and l� 0.10 cm.Obviously, all theoreticalcurves originate at the same calculatedpoint at the interface,suggesting that this experiment was performed in thelimiting current region where all chromoionophore mole-cules at the interface become protonated. Note that it isimpossible in this experiment to accurately image thesurface region spectroscopically because of optical scatter-ing effects [19]. The diffusion coefficient obtained from thisexperiment is in good agreement with literature valuesobtained under zero-current conditions [19], and is also inexcellent agreement with the Cottrell fit discussed above.Consequently, the earlier notion that such polymericmembrane electrodes function according to mass transportprocesses within the polymer phase, not the sample phase,appears to be consistent with the experimental dataobtained here.Figure 3 (bottom) shows the same experiment as for the

top figure, but at a more extreme potential of � 1000 mV.Here, obviously, the concentration profile appears to be

Fig. 3. Time dependent absorbance of a PVC±NPOE membrane containing ETH 5294 (protonated form, at 650 nm] in contact with apH 6.8 solution upon external application of a single (top) � 500 mVor (bottom) � 1000 mV potential pulse. Solid lines in the top figureaccording to Equation 2.



different from the ideal case shown in Figure 3 (top). Theprotonated chromoionophore starts to assume its maximumvalue for an increasingly larger distance away from theinterface as time progresses. This can be explained by thesubstantial extraction of uncomplexed hydrogen ions intothemembrane owing to themore extreme applied potential.In this case, the proton gradient would no longer be dictatedby the gradient of protonated chromoionophore alone.Diffusion appears to be much more rapid than for the caseshown in Figure 3 (top). Similar effects have been observedearlier in zero-current transport experiments of simple acidsin membranes containing this chromoionophore [29]. Thisbehavior is also fully in linewith the earlier interpretation ofthe extraction of undissociated electrolyte at potentialsbeyond the limiting current region [22]. Note that the phaseboundary concentration of protonated chromoionophorecontinuously decreases with time. This effect is believed toreflect a gradual concentration polarization of total chro-moionophore in the membrane, which originates frommembrane transport effects [32]. While this principle doesnot appear to be problematic in the normal function ofamperometric ion sensors, which are pulsed at short times, ithas been used by Lindner [33] to estimate the concentrationofmembrane components remaining in themembrane.Alsonote that the potentials applied in the ring membraneexperiments are not directly comparable to the ones appliedin the Philips electrode body because of the differences inthe experimental setup (different membrane resistance andnon-uniform membrane thickness owing to the ring shape,and a smaller concentration of chromoionophore). Un-

fortunately, the non-uniform membrane thickness andlower ionophore concentration made it difficult to recorduseful normal pulse voltammograms directly with the ring-shaped membranes, which would have enabled a directcorrelation between spectroscopic imaging information andamperometric response. Nonetheless, the correlation to theseparate amperometric experiments appears to be convinc-ing.The basic mechanism of the amperometric sensors

studied here can be summarized as follows. Upon theapplication of a given interfacial potential, ions are forced topartition from the sample into the organic phase boundaryregion in order to satisfy the phase boundary potentialcondition. Since the sample solution is well buffered, iondepletion on the sample side is unimportant. Instead, ionsextracted into the polymeric membrane phase will sponta-neously diffuse in direction of the membrane bulk, which isdevoid of those ions. This ion flux leads to a replenishment atthe phase boundary, which dictates the observed current.Initially, at mild potentials, the extraction and diffusion ofions in the polymeric membrane is assisted by the neutralionophore. As the applied potential increases, the iono-phorewill eventually saturate, that is, there will be a limitingconcentration of ionophore at the phase boundary beyondwhich no hydrogen ions can be extracted into the polymericmembrane. This defines the limiting current. In thispotential range, the observed current decay for a givenpulse can be adequately described with the Cottrellequation because the phase boundary concentration ofionophore is constant as a function of time. Further increase

Fig. 4. Normal pulse voltammograms for a PVC±NPOE membrane containing ETH 5294 and the inert lipophilic salt ETH 500 incontact with sample solutions of various pH. Other conditions as in Figure 1.



of the applied potential leads eventually to the extraction ofuncomplexed ions that no longer require the assistance ofthe ionophore to be extracted into the polymer. In order toobtain repeatable current readings in a practical experi-ment, normal pulse voltammetry is a recommendedmethod[23]. The baseline potential applied intermittently will allowthe ions extracted during a previous applied potential pulseto be adequately displaced again from the membrane.If these notions are accurate, it should be possible to use

suchmembranes containing aH�-ionophore tomeasure pHvoltammetrically. Indeed, the phase boundary concentra-tion of extracted ions is thought to depend directly on theapplied interfacial potential and the pH. If the membrane iseffectively renewed before a given applied potential so thatthe membrane diffusion layer thickness is always identical,the resultant current should be a direct function of pH. Thisis in agreement with recent theory describing standardhydrogen ion transfer potentials in biphasic systems con-taining a lipophilic base [34]. Figure 4 shows differentnormal pulse voltammetric response curves as a function ofthe sample pH. Obviously, the half-wave potentials appearto be directly dependent on pH. At those equal half wavepotential currents, the phase boundary concentration ofextracted hydrogen ions is thought to be identical fromvoltammogram to voltammogram. Consequently, theyshould ideally represent a direct Nernstian relationshipbetween half-wave potential and the sample pH, in accord-ance to the phase boundary potential equation. The half-wave potentials (at the constant current of � 10 �A) areseparately plotted as a functionof the samplepH inFigure 5,lower trace. The relationship is indeed Nernstian in a widepH range, with an experimental slope of � 58.5 mV pH�1.

Interestingly, the samemembranes were found to show verypoor slopes of ca. � 15 mV pH�1 when measured potentio-metrically. Indeed, these membranes contain no deliber-ately added ion-exchanger, but a very high concentration ofinert lipophilic salt, and are therefore not optimized forpotentiometric experiments. The complete lack of pHresponse can probably be best explained by some leachingof the tetraphenylborate from the membrane, thus creatinga membrane with gradual anion-exchange, rather thancation-exchange properties, or by the presence of someimpurities. In contrast, the voltammetric response is ana-lyzed at relatively high currents where a small residual ion-exchange property of the membrane is without conse-quence. Indeed, a visual inspection of the applied potentialsin Figure 4 required to reach much smaller currents, below1 �A, suggests that the response is much less sensitive in thisrange, which is consistent with the potentiometric results.Since the half-wave potentials represent the point where theionophore is thought to be half protonated at the phaseboundary (giving half the limiting current), the voltammet-ric responsewas compared to potentiometricmeasurementson membranes that contained 50 mol-% cation-exchangerwith otherwise the same composition. The potentiometricresponse is shown in the upper trace Figure 5 and is nownearly identical to the voltammetric pH responses shown inthe same figure.The normal pulse voltammograms as shown in Figure 4

start to exhibit unusual behavior at low pH. The limitingcurrent gradually starts to increase at pH 4 until, at pH 3, acomplete loss of any limiting current is observed. While thisseems to be surprising at first, it can be explained bycomparing it to the known behavior of the potentiometricand optical counterparts at low pH. If no external potentialsare applied, PVC±NPOE membranes containing the ion-ophore ETH 5294 are known to exhibit anion-interferencestarting around pH 3.5 (see Fig. 5, upper trace). Thisoriginates from a spontaneous coextraction of protons andchloride ions into the polymeric membrane [35]. With ion-selective electrodes, this results in a breakdown of theNernstian pH response slope and defines the upperdetection limit (see Fig. 5). With the voltammetric sensorsdiscussed here, spontaneous coextraction of electrolyte intothe membrane effectively renders the interface non-polar-izable: a substantial phase boundary concentration ofhydrogen ions is spontaneously present at all times at bothsides of the interface. This leads to a featureless voltam-metric response curve and to breakdown of the NernstianpH response range as well (see Figs 4 and 5). Voltammetricsensors appear to be more susceptible to this effect thantheir potentiometric counterparts, since the latter show aNernstian response that are extended by about a one pHunit (see Fig. 5).

4. Conclusions

Spectral imaging experiments on amperometric hydrogenion sensing membranes revealed that the limiting current

Fig. 5. Bottom data: galvanostatic potential values (at i�� 10 �A) of data shown in Figure 4 as a function of the samplepH. Solid line with slope of � 58.5 mV/pH. Top data: Potentio-metric zero-current response of corresponding ion-selectivemembranes containing the cation-exchanger NaTFPB in additionto ETH 5294.



region of the voltammogram is given by saturation of theionophore at the phase boundary. The time dependentdiffusion profiles could be conveniently correlated tocommon diffusion theory, and yielded diffusion coefficientsthat agreed with experiments on ion-selective membranesunder zero current conditions. The imaging results alsosuggested that the observed current decay curves under thesame conditions may be described by the Cottrell equation,and the correspondence was indeed good, with very similardiffusion coefficients as for the imaging experiments. Thisgood correlation suggested that the underlying responseprinciples of the amperometric ion sensors are well under-stood.As expected, and in contrast to regular amperometricsensors, the voltammetric responses were analyzed galva-nostatically and found to relate to the sample pH in aNernstian fashion. Their response range was compared tothat of the corresponding ion-selective electrodes, and theorigin of the upper detection limit at low pH was discussed.The response characteristics and diffusion profile measure-ments obtained here are in agreement with the expectedresponse mechanism formulated earlier.

5. Acknowledgement

The authors would like to acknowledge financial supportfrom the Petroleum Research Fund (administered by theAmerican Chemical Society) in support of this research.

6. References

[1] Z. Samec, V. Marecek, J. Koryta, M. W. Khalil, J. Electroanal.Chem. 1977, 83, 393.

[2] H. J. Lee, C. Beriet, H. H. Girault, J. Electroanal. Chem.1998, 453, 211.

[3] H. J. Lee, H. H. Girault, Anal. Chem. 1998, 70, 4280.[4] C. Beriet, H. H. Girault, J. Electroanal. Chem. 1998, 444, 219.[5] D. Henn, K. Cammann, Electroanalysis 2000, 12, 1263.[6] V. Horvath, G. Horvai, E. Pungor, Mikrochim. Acta 1990, I,

217.[7] V. Horvath, G. Horvai, Anal. Chim. Acta 1993, 273, 145.[8] V. Marecek, H. Janchenova, M. Brezina, Anal. Chim. Acta

1991, 244, 15.

[9] T. Osakai, T. Nuno, Y. Yamamoto, A. Saito, M. Senda,Bunseki Kagaku 1986, 38, 479.

[10] T. Osakai, T. Kakutani, M. Senda, Bunseki Kagaku 1984, 33,E371.

[11] S. Sawada, H. Torii, T. Osakai, T. Kimoto, Anal. Chem. 1998,70, 4286.

[12] J. Langmaier, K. Stejskalova, Z. Samec, J. Electroanal. Chem.2001, 496, 143.

[13] J. Koryta, P. Vanysek, M. Brezina, J. Electroanal. Chem. 1977,75, 211.

[14] V. V. Cosofret, M. Ernosy, T. A. Johnson, R. P. Buck, R. B.Ash, M. R. Neuman, Anal. Chem. 1995, 67, 1647.

[15] H. J. Lee, C. M. Pereira, A. F. Silva, H. H. Girault, Anal.Chem. 2000, 72, 5562.

[16] M. A. Roberts, J. S. Rossier, P. Bercier, H. Girault, Anal.Chem. 1997, 69, 2035.

[17] J. S. Rossier, M. A. Roberts, R. Ferrigno, H. H. Girault,Anal.Chem. 1999, 71, 4294.

[18] V. Marecek, M. P. Colombini, J. Electroanal. Chem. 1988,241, 133.

[19] B. Schneider, T. Zwickl, B. Federer, E. Pretsch, E. Lindner,Anal. Chem. 1996, 68, 4342.

[20] S. Jadhav, E. Bakker, Anal. Chem. 1999, 71, 3657.[21] J. Koryta, G. Du, W. Ruth, P. Vanysek, Faraday Discuss.

Chem. Soc. 1984, 77, 209.[22] S. Jadhav, E. Bakker, Anal. Chem. 2001, 73, 80.[23] S. Jadhav, A. J. Meir, E. Bakker, Electroanalysis 2000, 12,

1251.[24] E. Bakker, M. Lerchi, T. Rosatzin, B. Rusterholz, W. Simon,

Anal. Chim. Acta 1993, 278, 211.[25] V. V. Cosofret, T. M. Nahir, E. Lindner, R. P. Buck, J.

Electroanal. Chem. 1992, 327, 137.[26] I. Tsagkatakis, S. Peper, E. Bakker, Anal. Chem. 2001, 73,

315.[27] E. Bakker, E. Pretsch, Anal. Chem. 1998, 70, 295.[28] K. Seiler, W. Simon, Anal. Chim. Acta 1992, 266, 73.[29] E. Lindner, T. Zwickl, E. Bakker, B. T. T. Lan, K. To¡ th, E.

Pretsch, Anal. Chem. 1998, 70, 1176.[30] Y. Qin, E. Bakker, Talanta 2002, 58, 909.[31] J. Crank, The Mathematics of Diffusion, Oxford University

Press, New York 1993.[32] M. L. Iglehart, R. P. Buck, G. Horvai, E. Pungor, Anal.

Chem. 1988, 60, 1018.[33] B. D. Pendley, E. Lindner, Anal. Chem. 1999, 71, 3673.[34] V. Gobry, S. Ulmeanu, F. Reymond, G. Bouchard, P.-A.

Carrupt, B. Testa, H. H. Girault, J. Am. Chem. Soc. 2001, 123,10684.

[35] E. Bakker, M. N‰gele, U. Schaller, E. Pretsch, Electro-analysis 1995, 7, 817.



spectral imaging and electrochemical study on the response mechanism of ionophore-based polymeric...

Documents