91

1040-8347/95/$.50© 1995 by CRC Press, Inc.

Critical Reviews in Analytical Chemistry, 25(2):91–141 (1995)

I. INTRODUCTION

In 1976, Jagner and Graneli1 reported anovel analytical technique for the determi-nation of metal traces that they called poten-tiometric stripping analysis (PSA) becauseanalyses based on the oxidation of speciespreviously deposited on an electrode by oxi-dants carried convectively to the electrodesurface had not yet been included amongelectroanalytical techniques by the Interna-tional Union of Pure and Applied Chemists(IUPAC). However, as admitted by its pro-ponents themselves, the technique should bereferred to more accurately as chrono-potentiometric stripping analysis. This tech-nical alternative arose from polarographicmethods (more specifically, from anodicstripping voltammetry, ASV). In both tech-niques, metals in a sample are electrolyti-cally concentrated by deposition on an elec-trode (usually a rotating mercury filmelectrode) prior to analysis proper. The two,however, differ in the way deposited metalsare stripped and the analytical signal is ob-

tained. In ASV, stripping is done electro-chemically (Figure 1) by applying a usuallylinear potential scan to the working elec-trode over a given period during which thecurrent circulating by the electrode is re-corded as a function of the applied potential.When such a potential equals the oxidationpotential of one of the deposited metals, themetal in question is stripped from the elec-trode, which is accompanied by an increasein the measured current. Each metal is thusidentified by the presence of a maximum inthe current/potential recording obtained, asthe position of the maximum (Ep) is charac-teristic of each metal and its height (ip) isproportional to the metal concentration insolution. The signal is overlapped with anon-Faradic background current originatingfrom the electric charge at the electrode-solution interface, which is the greatest hurdleto be overcome in order to lower determina-tion limits. The effect of such a current canbe lessened by using several variants of ASVbased on the application of nonlinear poten-tial ramps; such variants include alternate

Potentiometric Stripping Analysis: A Review

J. M. Estela, C. Tomás, A. Cladera, and V. CerdàDepartment of Chemistry, University of Balearic Islands, 07071 Palma de Mallorca,Spain

ABSTRACT: A bibliographic review (150 references) on potentiometric stripping analysis(PSA) is performed. Theoretical, instrumental, analytical applications and advantages, andinferences of other modern PSA techniques are considered, like derivative PSA, constant-currentPSA, multichannel potentiometric monitoring stripping analysis, differential PSA, constant-current enhanced PSA, derivative adsorptive PSA, kinetic PSA and reductive PSA.

Implementation of PSA in flow systems is also considered, namely continuous-flow andflow-injection systems.

KEY WORDS: potentiometric stripping analysis (PSA), background, instrumentation,applications, flow systems, continuous flow, flow injection analysis.

92

current ASV (acASV) and differential pulseASV (dpASV), which provide substantiallyimproved detection limits.

In PSA, however, no control is made ofthe potential of the working electrode duringmetal stripping (Figure 2), which is accom-plished by using a chemical oxidant in solu-

tion — usually Hg(II) or dissolved oxygen.The working conditions are set in such away that the rate of oxidation of depositedmetals remains constant throughout the strip-ping process; such a rate is determined bythat of oxidant diffusion from the solution tothe electrode surface. Under these condi-

FIGURE 1. Timing of anodic stripping voltametry analysis.

93

FIGURE 2. Timing of potentiometric stripping analysis.

tions, the analytical signal is recorded bymonitoring the potential of the working elec-trode as a function of time. The curves thusobtained can be interpreted as being pro-vided by a redox titration of deposited met-als in which the titrant is added over them ata constant rate. The distance between thetwo consecutive equivalence points in a curvewill be proportional to the metal concernedin solution, whereas the potential of the cen-tral zone (E0) will be characteristic of it.

The most salient feature of stripping tech-niques is that dissolved metals concentrateat the working electrode during the elec-

trodeposition step (zone 1), thereby substan-tially lowering their detection limits. In ad-dition, the sensitivity can be adjusted to theparticular requirements by choosing an ap-propriate electrodeposition time. The PSAtechnique is comparable to ASV in terms ofsensitivity but lags slightly behind acASVand dpASV in this respect.2 On the otherhand, it features several major advantagesover voltammetric techniques:

1. Potentiometric stripping can be imple-mented by using straightforward equip-ment such as a three-electrode cell, a

94

high-impedance operational amplifier,an x/t recorder, and a potentiostat. Useof the potentiostat can be simplified tooperating at a single potential (e.g.,–1.25 V vs. SCE, where all metals suit-able for analysis will be reduced andhydrogen formation avoided). In addi-tion, times can be measured morereadily and precisely than microcurrentsand no potential ramp need be used (incontrast with ASV), which results indiminished instrumental costs.

2. Both ASV and PSA are multielementtechniques. The width of ASV bandsand hence discrimination betweendifferent elements is a function of theanalyte concentration and the potentialscan rate. This somehow complicatesthe analysis of samples containing ratherdifferent concentrations of the speciesto be determined because adequate reso-lution can only be achieved by applyinga slow potential ramp (which lengthensanalyses) or altering the scan rate dur-ing stripping. Because the electrodepotential in PSA is controlled by anoxidation process, the “scan rate” isself-optimized, so signal discriminationis more than adequate whatever theanalyte concentration ratios. This, how-ever, has one major limitation. Becausethe electrode potential remains virtu-ally constant during stripping until theanalyte concerned is depleted, thoseelements being deposited at the poten-tial in question will continue to be de-posited until the analyte is fully stripped.The end result is that the signal for anelement depends, however slightly, onthe concentration of the elements thatare stripped before it.

3. Potentiometric stripping analysis hasproved to be feasible in samples withionic strengths down to 10–4 M, as wellas polar organic solvents such as pro-panol and acetic acid, and in the pres-

ence of electroactive organic speciesprovided they are not deposited on (andhence do not alter) the electrode orchange the rate of oxidation of depos-ited elements. In contrast to ASV, nocurrent is drawn through the sampleduring the stripping phase.

4. The structure of the thin mercury filmchanges during preelectrolysis becauseof the sustained increase in film thick-ness. Frequently, the film is also af-fected by adsorbents or nitrogenbubbles. The net effect is that the trans-port rate of analytes into the mercuryfilm differs slightly between the analy-sis of a sample as such and from astandard aliquot. In PSA, the rate oftransport of oxidants is similarly af-fected, thus partly compensating for thiseffect. This also holds with changes inthe electrode rotation rate. Neither ef-fect is offset, for example, in ASV.

5. As in ASV, PSA signals overlap with abackground signal due to charge cur-rents at the electrode/solution interface.However, PSA background signals areless significant.

6. PSA has also proven suitable for theanalysis of heavy metals at concentra-tions in the range 0.1 to 1.0 ppm, whereno deaeration is required. The constantoxygen concentration in the sample canbe advantageously used for oxidationduring stripping. Due care should beexercised, however, that the analytesolubility in the mercury phase is notexceeded. In addition, samples must bebuffered in the acid region duringpreelectrolysis in order to avoid theformation of insoluble or irreversiblehydroxo species.

On the other hand, PSA also has severalpitfalls, some of which are common to alltechniques involving mercury film elec-trodes. Thus:

95

1. Like all other thin mercury film tech-niques, PSA is affected by the forma-tion of intermetallic compounds. Thus,the 1:1 copper-zinc intermetallic com-pound poses severe interferences,which, however, can be overcome bythe addition of gallium.

2. One unique disadvantage of PSA is thedecrease in the oxidant concentrationduring preelectrolysis. This shortcom-ing can be circumvented by making theelectrode surface small relative to theoverall sample volume.

3. The analytical signals provided bymercury film electrodes are markedlyinfluenced by the electrode’s history.In PSA, the use of Hg(II) as the oxidanteliminates the risk of destroying themercury film between consecutiveanalyses; during stripping, the poten-tial of the working electrode will auto-matically stop before it reaches the re-gion of mercury oxidation (zone 3 inFigure 2). Formation of, for example,calomel on the film surface, is thushindered. Also, there is no risk of oxi-dation of the glassy carbon surface.Using an oxidant other than Hg(II) con-siderably increases the risk of themercury film being destroyed, so that itmust be regenerated more frequently.Fortunately, the electrode can beregenerated in situ if desired andanalyses performed by using thestandard-addition method.

4. Stripping analysis, both potentiometricand voltammetric, is particularly wellsuited to the determination of heavymetals in liquid samples, no pretreat-ment of which is often needed. Thetime-consuming step of analyses in suchconditions is plating. This has madeautomating the technique mandatory.On the other hand, plating can be fur-ther expedited by using microproces-sor-controlled units enabling rapid ac-quisition and processing of stripping

data; use of such units has led to newPSA variants of improved sensitivity,selectivity, and expeditiously. Theadded use of continuous-flow and flow-injection systems for this purpose con-tributes to further increased throughputand selectivity.

5. One other major limitation of ASV andPSA is that direct stripping analyseswith adequate sensitivity are only fea-sible for a small number of analytes.This is particularly true of PSA whendissolved oxygen is used as the oxi-dant. One way of extending applicationto a wider range of analytes entailsimproving deposition (whether anodicor cathodic) and/or the stripping stepby using an electrode other than that ofmercury film or an oxidant differentfrom Hg(II) and dissolved oxygen, byaltering the stripping solution or byusing an alternative technique to recordor process the analytical signal.

II. VARIANTS OF PSA TECHNIQUE

The PSA techniques can be classifiedinto the following variants.

A. Derivative PotentiometricStripping Analysis (dPSA)

This variant of PSA was developed byJagner and Aren2 in order to facilitate evalu-ation of the analytical signal by using itsderivative. The signal is obtained in the sameway as in conventional PSA. The dPSA tech-nique involves preconcentrating metalanalytes in a thin mercury film covering aglassy carbon electrode and subsequentlymeasuring the electrode potential subject tocontrolled transport of oxidant to the elec-trode surface. After plating, the potential ofthe working electrode is recorded with theaid of an operational amplifier coupled as avoltage monitor. The time derivative of the

96

signal is registered on a second recorderchannel by means of derivative circuitry.The dE/dt vs. t graph thus obtained (Fig-ure 3) exhibits maxima at those points wherea conventional PSA curve would show asharp variation of the potential with time.The distance between two consecutivemaxima corresponds to an analytical signalequivalent to the plateau length in conven-tional PSA but is easier to determine with ahigher precision.

B. Constant-Current StrippingAnalysis (CCSA)

Whereas some authors regard this tech-nique as a variant of PSA,3 others claim thatit should be called “chronopotentiometricstripping analysis”.2 In this technique, the

metal analyte is stripped by a constant oxi-dizing current passed through the workingelectrode rather than by a chemical oxidant.In both PSA and CCSA, the time needed forthe analyte to be oxidized is directly propor-tional to the metal ion concentration (Fig-ure 4). This technique has been used inten-sively by Renman et al.4 in flow systems, aswell as in some special applications, includ-ing the determination of lead in gasoline5

and the use of polymer-modified electrodes.3

As in voltammetry,6 passing an electric cur-rent during stripping gives rise to interfer-ences from electroactive species present insamples; such interferences, however, canreadily be overcome, particularly in flowsystems, by subjecting a matrix other thanthat of the sample to stripping (i.e., using thematrix-exchange technique) or employing aphysically or chemically modified electrode.

FIGURE 3. Timing of derivative potentiometric stripping analysis.

97

FIGURE 4. Timing of constant current stripping analysis.

C. Multichannel PotentiometricMonitoring Stripping Analysis(MSPSA)

This technique was originally developedand subsequently used intensively byMortensen et al.7 The stripping time iselectrochemically enhanced by using a com-puterized data acquisition technique, viz.,multichannel potentiometric monitoring (Fig-ure 5) in conjunction with potentiometricstripping analysis (MSPSA). After a single,short deposition period, a substantial frac-tion of the accumulated metal may beforced to undergo several oxidations and

rereductions in a precisely timed sequence.The computer acquires and adds up the ana-lytical signals, viz., the number of time units(clock pulses) resulting from the oxidationsteps within the preselected potential win-dow. Thus, even after a short plating period,a relatively small amount of preconcentratedmetal may produce a significant analyticalsignal. The feasibility of enhancing signalsby using computerized multiscanning in con-junction with voltammetric stripping analy-sis has been demonstrated beyond doubt.The extent to which the analytical signal canbe enhanced depends heavily on how effi-ciently freshly oxidized metals can be recov-

98

FIGURE 5. (I) Potential vs. time behavior of working electrode during redissolutionof three amalgamated metals. Ea – Ec is the potential window studied. (II) Computermemory section: the data storage area starting at address A0 holds a record ofaccumulated clock pulse counts. (III) The resultant multichannel potentiogram. (FromMortensen, J.; Ouziel, E.; Skov, H. J.; Kryger, L. Anal. Chim. Acta. 1979, 112, 297–312. With permission.)

ered (rereduced) after each cyclic scan. Ifthe time available for oxidized metal to es-cape from the working electrode by diffu-sion in a quiet solution is short, its recoverywill be quite high. Optimal signal enhance-ment can be achieved by using fast anodicscans; the oxidation potential is scanned onlyas far as required to obtain the signal, andthis is followed by a prompt return to thereduction potential. Similarly, in multi-scanning PSA, chemical oxidation shouldproceed rapidly, followed by resumption ofpotentiostatic control at the reduction poten-tial. As in potentiometric stripping, the rateof the oxidation process may be controlledby the amount of oxidant added to the solu-tion; a high recovery of metals can be ex-pected if a proportionally large excess ofoxidant is used. This technique is suitable

for stripping analysis with preconcentrationtimes of 60 to 90 s at a mercury film elec-trode and provides linear responses from 1 to100 µg/l Cd(II) and Pb(II). The detectionlimit falls to ~5 ng/l for a preconcentrationtime of 30 min.

D. Differential PotentiometricStripping Analysis (DPSA)

This is a computer-assisted variant ofPSA originally developed by Kryger.8 InDPSA, as in PSA, stripping of precon-centrated analytes is caused by some oxidantin the sample solution being transported tothe working electrode, and the process isrecorded potentiometrically. If the rate ofstripping is high relative to that at which the

99

newly stripped material can escape (by dif-fusion or convection) from the vicinity ofthe working electrode, a high concentrationregion of analyte is created around the work-ing electrode during the stripping step. TheDPSA technique exploits the formation ofsuch a region: after plating is finished,potentiostatic control is stopped and the po-tential of the working electrode is recordedas a function of time with the aid of a micro-computer. The electrode potential is (Fig-ure 6), however, allowed to undergo only asmall change (∆E′ . 10 to 50 mV) and, assoon as a preset potential threshold is reached,potentiostatic conditions are resumed over ashort period at a plating potential slightlyanodic of the previous one, ∆E. In this way,a substantial amount of newly oxidized ma-terial can be replated and reoxidized in asubsequent stripping step going from the newplating potential across the selected poten-tial window. The procedure is repeated untilthe entire potential range of interest has been

covered. With a suitable choice of potentialwindows, the stripping signal at any poten-tial interval is recorded several times and theresults are accumulated in the computermemory. Hence, for a given plating period,a signal enhancement is likely to result. Theprocess is analogous to the multiscanningeffect that provides the increased sensitivityof differential pulse stripping voltammetryrelative to the linear sweep technique. Thedifferential potentiogram obtained is essen-tially the derivative of time with respect topotential, and where the stripping potentio-gram exhibits a plateau signalling the strip-ping of a component, the differentialpotentiogram shows a maximum (Figure 6).The signals for trace elements such as cad-mium and lead, which exhibit transport-con-trolled potentiometric stripping, can be en-hanced by using a scheme involving multiplestripping and rereduction of preconcentratedanalytes, the detection limits for which arebelow 5 × 10 –10 M if a 60-s plating time is

FIGURE 6. Principle of differential potentiometric stripping analysis. Curve a, normal potentialvs. time behavior during stripping of a plated component; curve b, potential vs. time behaviorduring differential potentiometric stripping; curve c, differential stripping potentiogram. (FromKryger, L.; Anal. Chim. Acta. 1980, 120, 10–30. With permission.)

100

reduction” cycles, so the stripping time isextended. Zie and Huber9 used rotating mer-cury film electrodes, Cd(II) during strippingand dissolved oxygen as oxidant to developand thoroughly test this technique, the foun-dation of which is inspired by catalytic strip-ping as applied to ASV and CCSA in orderto improve the sensitivity. The cathodic ca-talysis process is very strongly influencedby the prevailing hydrodynamic conditions.In order to achieve the maximum possiblecatalytic effect, stripping should be carriedout in a quiet solution so as to ensure theformation of a high concentration zone offreshly stripped analyte in the vicinity of theelectrode surface. The CCEPSA techniqueis more sensitive than conventional PSA byat least one order of magnitude. This en-hancing factor is equally applicable toCCEPSA detection limits. Figure 7 showssome typical stripping curves for Cd(II) ob-tained by using this technique.

F. Derivative AdsorptivePotentiometric Stripping Analysis(dAPSA)

This technique, another variant of PSA,was originally developed by Jin and Wang,10

who called it “derivative adsorption strip-ping potentiometry”. The dAPSA techniquewas conceived to extend the application ofPSA to organic compounds and some inor-ganic elements (e.g., iron, cobalt, and nickel)that cannot be electrolytically precon-centrated on mercury. It exploits the adsorp-tive capacity of some organic compoundsand inorganic complexes to preconcentratethem at an electrode. The adsorbed com-pounds are subsequently stripped by the ef-fect of an oxidant or reductant. The processinvolves the following reactions:

Osol Oads (1)

Oads + ne– Rads (2)

m R ads + n′ Ox → m Oads + n′ Red (3)

used. The accuracy of this technique wastested on a biological reference material. LikePSA, the DPSA technique is insensitive toreversible redox couples present in solution.

The technique is somehow related tomultiscanning PSA; however, the latter usesa single plating potential and the potentialinterval for each scan is on the order ofseveral hundred millivolts, so cadmium andlead, for example, may be stripped in thesame scan. This results in an unwanted cor-relation of the cadmium recovery with thelead concentration: a high concentration oflead forces the working electrode to remainat the stripping potential of lead for a longtime. At such a potential, newly strippedcadmium can escape from the working elec-trode by diffusion-convection, so there willbe a poor recovery of this metal betweenscans. In DPSA, the magnitude of ∆E′ iskept sufficiently small, so cadmium and leadare not stripped in the same scan and theprevious correlation vanishes. The correla-tion problem in the multiscanning techniqueis overcome by allowing the magnitude ofthe stripping interval to increase gradually.Thus, the component with the most cathodicstripping potential is determined by multiplescanning; then, another component is in-cluded in the scan, and so on. This “inter-rupted stripping” can be considered a crudetype of DPSA, but requires prior knowledgeof the stripping potentials involved. Also,achieving substantial analyte recoveries inmultiscanning potentiometric analysis entailsstripping in a quiet solution, which is unnec-essary with DPSA.

E. Constant-Current-EnhancedPotentiometric Stripping Analysis(CCEPSA)

In this technique, a constant cathodiccurrent is applied to the electrode systemduring the chemical stripping step in order toforce freshly stripped analyte to be redepos-ited into the mercury film. Some of thestripped species undergo several “oxidation-

101

where O and R denote oxidized and reducedspecies, respectively; the subscripts sol andads the solution phase and adsorption phase,respectively; and Ox the oxidant and Red itsreaction product.

Potential (E) and time (t) data are digi-tized, converted to dt/dE values, and plottedagainst the potential, which results in in-creased sensitivity and resolution. The tech-

FIGURE 7. Effect of the imposed cathodiccurrent on the shapes of the stripping curves.(a) 0; (b) 3.5; (c) 4.5; (d) 5; (e) 5.75 µA. 1.0 ×10–6 M Cd2+ in pH 5 acetate buffer; depositiontime (td) 2.0 min; deposition potential (Ed) –0.9V vs. SCE. S = 4.8 s for curves a, b, c, and dand 20 s for curve e. (From Zie, Y.; Huber,C. O.; Anal. Chim. Acta. 1992, 263, 63–70.With permission.)

nique is applicable to quiet and stirred solu-tions alike. Equations for the reversible pro-cess involved were derived and tested byusing 1-hydroxyanthraquinone as analyte andHg(II) as oxidant in deaerated solutions.

G. Kinetic Potentiometric StrippingAnalysis (KPSA)

This automated variant of PSA origi-nated from the findings of Cladera et al.11 instudying various oxidants and working elec-trodes for the determination of mercury byPSA. They found the shape of the strippingcurves for mercury deposited on a gold elec-trode by using the periodate/iodide systemas oxidant to deviate from conventional PSAcurves and resemble kinetic profiles moreclosely (Figure 8). Pertinent calibrationcurves can therefore be constructed by usingordinary kinetic treatment methods such asthose of the initial rate and fixed potential.

The overall process can be interpreted asfollows: in the electrolysis step, mercury isreduced on the gold surface, with which itamalgamates:

Hg2+ + 2 e– → Hg(Au) (4)

When the amount of deposited mercury islarge enough, a layer of unamalgamatedmercury may also be formed in addition tothe previous one. During stripping, the curveobtained on addition of periodate afterpreelectrolysis is consistent with a processin which mercury is either oxidized with avery slow kinetics or not oxidized at all. Inthe presence of iodide, addition of periodateto an acid medium results in the release ofiodine, which gives rise to the kinetic curvefor mercury oxidation through the followingreaction:

Hg(Au) + I2 + 2 I– → HgI42– (5)

When the amount of mercury present is quitelarge, the metal ion cannot penetrate the elec-

102

FIGURE 8. Kinetic potentiometric stripping analysis. Potentiograms obtained andslopes calculated at different Hg(II) concentrations. (From Cladera, A.; Estela, J. M.;Cerdà, V. J. Electroanal. Chem. 1990, 288, 99–109. With permission.)

trode completely, so, initially, the situationis equivalent to working with a pure mercuryelectrode; this gives rise to the typical pla-teau of PSA curves, the potential of whichcan be calculated by applying the Nernstequation to the HgI4

2–/Hg system. Once theentire mercury surface is depleted or if theamount deposited is small enough to pen-etrate the gold electrode, the redox processmay be limited by the kinetics of transferfrom the bulk electrode to the interface. Thisphenomenon, together with slower transferkinetics than those of the chemical oxidationreaction, accounts for the appearance of ki-netic curves after the typical plateau (or evenfrom the beginning of stripping if the amountof deposited mercury is rather small). Thisinterpretation is supported by the fact thatonly the copperized graphite electrode givesrise to the expected plateau: that for copperfollowed by that for mercury. As the copper

has already been stripped by the time themercury is, the latter does not need to diffusethrough the electrode, so it is only reoxidizedin a step controlled by the chemical processand diffusion of the oxidant. The shape ofthe kinetic curves obtained, with two dis-tinct slopes, can be explained by assumingthe electrode potential to be determined firstby the mercury concentration at the elec-trode surface, which in turn depends on itsrate of diffusion through the gold. Once allthe mercury has been oxidized, the slopechanges again and becomes steeper as itreaches the plateau yielded by the blank.

H. Reductive PotentiometricStripping Analysis (RPSA)

This modification of PSA was devel-oped in order to extend application of con-

103

ventional PSA to those analytes that cannotbe deposited cathodically owing to their lowsolubility in mercury or markedly cathodichalf-wave reduction potentials. Such ele-ments may occasionally be preconcentratedanodically and determined by cathodic strip-ping voltammetry. In addition, interferencesmay be overcome in some favorable casesby switching from anodic to cathodic strip-ping analysis. The RPSA technique was firstimplemented by Christensen and Kryger12

using the determination of manganese aschemical model; the metal was precon-centrated by anodic oxidation at a platinumelectrode and stripped by using hydroquinoneas reductant (Figure 9). In this way, manga-nese was determined at concentrations in themicrogram per milliliter range. The accu-racy of the technique, tested on a standardbiological material, is quite satisfactory.

In later work, Christensen et al.13 dem-onstrated the suitability of amalgamatedmetals as reductants in RPSA. The amal-gams were electrolytically generated fromdissolved metals in a mercury pool. Duringstripping, the reductant, which was storedinside the working electrode, reacted withsparingly soluble mercury compounds of theanalytes preconcentrated at the electrodesurface (Figure 9). With amalgamated so-dium, the technique proved to be suitable forthe determination of selenium and sulfur atthe 10–7 M level with a 1- to 2-min precon-centration. Halides can be determined at the10 –6 M level with a few seconds ofpreconcentration and use of a less powerfulreductant such as amalgamated zinc.

III. THEORETICAL FOUNDATION OFPOTENTIOMETRIC STRIPPINGANALYSIS

After the potentiostat is switched off oncethe electrodeposition of metals in a PSAexperiment is finished, the potential of theworking electrode rises rapidly until it reachesthe nearest oxidation potential for the depos-

ited metals. At this point, the metal in ques-tion is reoxidized and the potential remainsvirtually constant until the metal has beenstripped completely. Then, the potential risesrapidly again until that of the next metal isreached. The process is repeated until alldeposited metals have been stripped, afterwhich the potential continues to rise untilequilibrium with the bulk solution is reached.

FIGURE 9. Time sequences of potentials gen-erated during electrolysis by three-statepotentiostat module. (A) Oxidative potentiomet-ric stripping analysis. T1 + T2 + T3 is the elec-trodeposition time. (B) Reductive potentiomet-ric stripping analysis with water-soluble reducingagent. T2 + T3 is the electrodeposition timeand T1 is the electrode regeneration time. (C)Reductive potentiometric stripping analysis withelectrolytical generated amalgamated metal asthe reducing agent. T1 is the electrode regen-eration time, T2 is the amalgam creation time,and T3 is the electrodeposition time. (FromChristensen, J. K.; Keiding, K.; Kryger, L.;Rasmussen, J.; Skov, H. J. Anal. Chem. 1981,53, 1847–1851. With permission.)

104

in the bulk solution, DMi the diffusion coef-ficient of the metal ion in solution, δ1 thediffusion layer thickness during the elec-trodeposition time (tdep), and S the effectivesurface area of the working electrode.

During stripping, the following genericreaction takes place:

Mi(Hg)+ (n/m) Aj(ox) → Min+ + (n/m) Aj(red)

(8)

where Aj denotes the potential oxidantspresent in solution. Again, a diffusion layeris formed, only by Aj this time. On the as-sumption of rapid electrode reactions andthe diffusion of Aj to the electrode as beingthe rate-determining step, the overall amountof A j that reacts at the electrode during thestripping step can be expressed as

Aj(ox)tot ∝ [A j(ox)]DAj(ox)τδ2–1S (9)

where [Aj(ox)] is the oxidant concentrationin the bulk solution, DAj(ox) its diffusion co-efficient, τ the stripping time, and δ2 thediffusion layer thickness during stripping.As in the previous case, concentrations inthe bulk solution can be assumed to remainconstant during stripping. A combination ofEquations 7 and 9 yields an expression forthe overall time required for the completestripping of Mi(Hg):

τ i ∝M

in+[ ]tdep

δ2δ

1–1

n

m j

DA j OX( ) A j OX( )[ ]∑ (10)

according to which the signal obtained at agiven metal concentration will increase withincreasing electrodeposition time and de-creasing concentration of oxidants in solu-tion. On the other hand, the thicknesses ofthe diffusion layers formed during elec-trodeposition and stripping (δ1 and δ2, re-spectively) are directly related to the stirringof the solution in such steps. If stirring is

As a result, each deposited metal gives riseto a horizontal segment (a “plateau”) in thepotential-time curve. The potential at whicheach plateau appears is characteristic of eachmetal and its length gives the time the metaltakes to be oxidized. The rate at which eachmetal is oxidized is determined by the diffu-sion of the dissolved oxidant to the elec-trode, its oxidizing power, and the reactionkinetics. Consequently, it will depend on thenature and concentration of the oxidant, aswell as the experimental conditions (stirring,nature and surface of the electrode, etc.).

The phenomena involved in the elec-trodeposition-stripping cycle at a mercuryfilm electrode have been studied theoreti-cally14,15 in order to derive equations for thepotential-time curves provided by the PSAtechnique.

During electrodeposition, the followinggeneric reaction takes place at the electrodesurface:

Min+ + n e– → M i(Hg) (6)

where Min+ denotes the metal ions that can be

reduced at the potential of the working elec-trode, the elemental forms of which aremercury soluble. If Equation 6 is rapidenough, the concentrations of metals in thevicinity of the electrode start to fall rapidlyand a diffusion layer is established betweenthe electrode surface and the bulk solution.On the other hand, because the electrodeshave a fairly small surface, concentrations inthe bulk solution can be assumed to remainconstant throughout the experiment. Underthese conditions, according to Fick’s law ofdiffusion, the overall amount of metal that isdeposited on the electrode over an intervaltdep is given by the following proportionalityrelationship:

Mi(Hg) ∝ [M in+]DMitdepδ1

–1S (7)

where Mi(Hg) is the overall amount of de-posited metal, [M i

n+] the metal concentration

ox ox

105

maintained constant in both steps, δ1 and δ2

will be virtually identical, so they can beeliminated from Equation 10. On the otherhand, if stirring is stopped during stripping,δ1 will be smaller than δ2, so the signal willbe increased as a result. However, this sen-sitivity-enhancing procedure is only recom-mended when a highly reproducible stirringsystem is available.

Finally, according to Equation 10, onconstancy of the hydrodynamic conditions,oxidant concentration, and electrodepositiontime, the signals obtained in a series of ex-periments will be proportional to the metalconcentration in solution. This relationshipis the basis for application of the PSA tech-nique to quantitative analysis.

The equations above rely on the assump-tion that the rate-determining step of theprocess is the diffusion of species from thebulk solution to the electrode surface, whichinvolves considering any processes poten-tially occurring in the thin mercury filmformed on the electrode to have a negligibleinfluence. However, such processes must beconsidered if an accurate equation for thepotential-time curves is to be derived.

The potential of the working electrode atany time is given by the Nernst equation:

E = E 0 +RT

nFln

M in+[ ]s

Mi

Hg( )[ ]s

(11)

where [M in+] is the concentration of dis-

solved metal in the vicinity of the elec-trode and [M i(Hg)]s that of amalgamatedmetal in the mercury layer at the elec-trode-solution interface. As a rule, as oneof the metals, Mi, starts to be stripped, twodiffusion layers are formed: one in thesolution and the other in the mercury filmcovering the electrode. After a transitioninterval, a steady state is reached whereconcentrations at the electrode-solutioninterface, and hence the electrode poten-

tial, remain essentially constant until alldeposited metal is stripped. On the assump-tion that diffusion in the mercury film onthe electrode was the rate-determining step,Chau et al.15 derived the following equa-tion for the potential-time curve obtainedfrom the metal stripping:

E = E 0 +RT

nFln

2l

π D0

+

RT

nFln

t

τ – t

(12)

where l is the thickness of the mercury filmcovering the electrode, D0 the diffusion co-efficient for dissolved ions, and τ the timeneeded for deposited metal to be fullystripped.

So far, no mention has been made ofelectric bilayer phenomena at the electrode-solution interface. Such phenomena can beinterpreted as the appearance of excess elec-tronic charge at the electrode surface duringapplication of the electrodeposition poten-tial in order to counter the bilayer capaci-tance. When the potentiostat is switched offin order to start stripping, the excess elec-tronic charge can give rise to the followinggeneric reaction:

e–exc

+ (l/m j) A j(ox) → (l/m j) A j(red) (13)

In contrast to the above reactions, wherethe potential is determined by electrons atwell-defined energy levels, Equation 13 pro-vides no constant-potential zones, but rathera semiexponential background signal thatadds to the signal resulting from the strip-ping of the metals.

The literature abounds with theoreticalstudies aimed at elucidating, predicting, andchecking for the phenomena involved in PSA.Mortensen and Britz16 derived a model forpredicting the background signal (E vs. t) inPSA and concluded that, in the absence ofoxidizable materials at the working electrode,the E-t function is determined by the bilayer

106

capacity, which is strongly affected by dis-solved surfactants.

Labar and Lamberts17 demonstrated thefeasibility of controlling the transport ofoxidants to the surface of the working elec-trode (PSA at a stationary electrode) andfound that, when linear diffusion of oxidiz-ing species was the rate-determining step atthe working electrode-solution interface andthe physical properties of the medium wereassumed to lower the diffusion coefficient ofsuch species, the relative sensitivity of thetechnique was increased by a factor of 50 forall common ions assayed by PSA (Cu, Cd,Zn, Pb, Bi, Tl, and Ga).

Hussam and Coetzee18 reported a theo-retical treatment based on the assumption ofan initial parabolic concentration gradient ofthe metal in mercury and stripping in a stirredsolution, which is the usual condition forPSA. They derived and experimentally vali-dated equations for the transient potential asa function of time, as well as for the transi-tion time, and established the theoreticalbases for generating stripping pseudo-polarograms by PSA:

τ =n

2

CMn+0

CHg1+0

DMn +

DHg1+

td (14)

and

E t = E 0 –RT

nFln

Dτδl

–RT

nFln 1 – φ( ) (15)

where td is the deposition time; τ the transi-tion time; C 0

Mn+ and C 0Hg2+0 the bulk concen-

trations of the metal and mercury ions, re-spectively; DMn+ (or D) and DHg2+ thediffusion coefficients of the metal and mer-cury ions, respectively, in the mercury phase;l the mercury film thickness; δ the diffusionlayer thickness; Et the transient potential;φ = (nF/RT)(E d – E0), Ed being the reactionpotential; and all other symbols have theirusual meanings.

The results obtained by using conven-tional electrodes and fiber microelectrodes

showed the latter to minimize the interfer-ences arising from bilayer perturbations orprecipitation of metal derivatives on the sur-face of the working electrode.

Hoyer and Kryger19 carried out a theo-retical and experimental study of the poten-tial-time transient in PSA by using a rotatingmercury-film electrode. The data obtainedby digital simulation using the Cranck-Nicholson finite-difference method werequite consistent with the experimental re-sults and were used to derive relationshipsbetween peak width and signal duration inreversible analyte systems. Severe signaloverlap was found to result in distorted com-posite signals.

Xia et al.20 also performed theoreticaland experimental studies on film potentio-metric analysis using stirred and quiet so-lutions of copper(II) and lead(II), and de-rived and validated equations descriptiveof the transition time in both types of so-lutions:

τ = Kωptp2C0* (16)

for quiet solutions, and

τ = k–1[Ox]–1D0(ωp/ωs)1/2tpC0* (17)

for stirred solutions, where τ is the transitiontime, C0* the concentration of metal analytein solution before preelectrolysis, [Ox] theoxidant concentration at the electrode sur-face, k the rate constant, ωp and ωs the rota-tion speed during preelectrolysis and afterthe potentiostat is disengaged, tp thepreelectrolysis time, D0 the diffusion coeffi-cient of the metal analyte, and

K = 0.30D07/3V–1/3A2k–2[Ox]–2V–2 (18)

A being the surface area of the electrode andV the volume of the diffusion layer.

Labar et al.21 extended their studies onfilm PSA by investigating the variables thatcontrol the preconcentration and strippingsteps using a rotating glassy carbon disk

107

electrode (rde), Pb(II) as the analyte, andFe(III) as the oxidant. The results obtainedwere compared with previously reporteddata and, except for the influence of the rderotation speed on each step and on the ana-lytical parameters, experiments and theorywere in close agreement. Discrepancies inthe effects of the rotation speed were inves-tigated by potentiostatic coulometry andvoltammetry in regard to the precon-centration step: the effects of the rotationspeed were found to arise from the physicalbehavior of the rde. Use of the standard-addition or internal-standard method wasrecommended in preference over calibra-tion curves for analytical purposes owingto the occurrence of an activation period inthe electrodeposition step.

Garai et al.22 reported a theoretical treat-ment for PSA as regards anodic stripping ofmetals in quiet solutions involved in revers-ible and quasireversible electrode reactionsby using mercury film electrodes. The ex-perimental results were quite consistent withcalculated data, and derivations based onvarious assumptions (e.g., stripping in stirredsolutions, a diffusion layer of constant thick-ness) were compared. These authors sup-ported the recommendation by Labar et al.as regards recording the transition time for aquiet solution, but pointed out that the re-ported δHg/δMe = 20 (where δ is the diffusionlayer thickness) was too high a ratio for aquiet solution, and assigned it a value of 5 to10, depending on the rotation speed duringplating. The equation derived for the transi-tion time was

τ =z

Meδ

HgD

Rc

00

zHg

δMe

DHg

cHg

t el (19)

where τ is the transition time, z the numberof electrons exchanged in the electrode reac-tion, D the diffusion coefficient, δ the diffu-sion layer thickness, tel the preelectrolysistime, and c the concentration of metal ormercury ions in solution, which is depen-

dent on the mercury film thickness (this is atvariance with the values reported by Chau etal.15 and proportional to the first power ofthe charge of the metal ion, in contrast to thepeak current observed in ASV).

On the other hand, in calculating thefunction relating the potential and time in aPSA experiment, Garai et al. assumed themetal to distribute uniformly within themercury film. This assumption differs fromthose of Hussam and Coetzee18 and De Vriesand Van Dalen,23,24 who postulated a para-bolic metal distribution in the mercury film.A uniform distribution of the metal appearsto be more realistic according to Garai et al.because the literature almost unanimouslydemonstrates that a homogeneous metal dis-tribution in the amalgam can be reached in avery short time. The equation

E = E 0 +RT

zMe

Fln

2d

DR

π( )1/ 2 + lnt1 / 2

τ – t

(20)

and that derived by Hussam and Coetzeeshow that the E-t functions have identicalforms, irrespective of the assumption regard-ing the metal distribution in the amalgamand the stripping conditions (whether strip-ping is done in a stirred or quiet solution).The two equations differ by a shift in thepotential of 9 or 14 mV, depending onwhether the electrode process involves theexchange of one or two electrons, respec-tively.

Jin and Wang10 derived and experimen-tally validated equations for the derivativeof time with respect to potential, as well asthe peak half-width for a reversible processin adsorption stripping potentiometric analy-sis in both stirred and unstirred solutions onthe assumption of strongly adsorbed oxi-dized and reduced forms and obeyance ofthe Langmuir isotherm. The equation for anoxidant in a stirred (or unstirred) solution isof the form

108

dt

dE

p

=nFτ

4RT (21)

where

τ =ΓRt a

′ n

mkD

OX2 / 3v – 1 / 6ω

01/ 2c

OX(22)

for stirred solutions, and

τ =δΓR t a

′ n

mD

OXc

OX(23)

for unstirred solutions, τ being the transienttime; m and n′ the stoichiometric coeffi-cients for the reaction between the reducedspecies adsorbed at the electrode and theoxidized species, respectively, Dox (cm2/s)the diffusion coefficient of the oxidant, ν(cm2/s) the kinematic viscosity, ϖ0 (rad/s)the angular velocity in the bulk solution, cox

(mol/cm3) the oxidant concentration, ΓR (mol/cm 2) the surface concentration of theadsorbed reduced species, and ta thepreconcentration time.

At the peak half-height,

dt

dE=

1

2

dt

dE

p

=nFτ8RT (24)

so the peak half-width will be given by

W1/ 2 =3.52RT

nF(25)

(at 25°C, W1/2 = 90.6/n mV).A comparison of the equations for the

peak potential, peak half-width in adsorp-tion chronopotentiometry, and adsorption inPSA with oxidation in a stirred or unstirredsolution reveals that their forms are identi-cal. So is that for the dt/dE function, exceptfor the sign and the definition of the transi-tion time.

Garai et al.25 developed and experimen-tally checked the theory of dPSA, based onthe equation

dE

dt=

RT τ + t( )zF

2 t τ – t( )(26)

as well as differential potentiometric analy-sis, which designates the technique based onthe derivative of time with respect to poten-tial against the potential function as themeasured signal. Notwithstanding the vastapplication and flexibility of both techniques,no related mathematical expressions have sofar been reported. The technique called“differential potentiometric analysis” byKryger8 does not correspond exactly to thefunction discussed by Garai et al. as differ-ential potentiometric stripping analysis; how-ever, the latter closely approximates the read-out obtained by the former method, as wellas multiscanning PSA as conceived byMortensen et al.7 and Renman et al.4

Zie and Huber9 derived the equation forthe transient potential-time curves inCCEPSA:

E t =E 0 +RT

nFln

2l

πD0( )1/ 2

+

RT

nFln

t1 / 2

τc

– t

(27)

where l is the mercury film thickness, D0 thediffusion coefficient of the metal ion in so-lution, t time, and

τ c =0.62D 0

2 /3v–1ω1/ 2tdC *Ox

kC *Ox –I c

nFA(28)

the stripping time for catalytic current PSAwith stirred deposition and quiescent strip-ping conditions, Ic being the constant ca-thodic current applied during the strippingstep, td the deposition time, COx the oxidant

t

109

concentration in the bulk solution, k theoxidation rate constant, D0 the diffusion co-efficient of dissolved metal ions, ν the kine-matic viscosity, and ω the angular velocityof the solution phase, which is the same asthat for conventional PSA. From Equation27 it follows that, whether oxidation of de-posited amalgam is effected by a constantoxidation flux or a “reduced” oxidant flux,the shape of the transient potential-timecurves will be the same.

IV. INSTRUMENTATION IN PSA

The basic experimental set-up neededfor a potentiometric stripping experimentcomprises a potentiostatic circuitry, a three-electrode electrochemical cell, and an im-pedance recorder (Figure 10). However, theinception of novel PSA variants and attemptsat improving their performance calls for someextent of automation. This has been imple-mented in various ways, including commer-cially available instruments such as the Ra-

diometer ISS 820 Ion Scanning System orthe Radiometer PSU 20 TraceLab potentio-metric stripping unit, as well as customizedmicrocomputer-controlled configurationsbuilt from electrochemical equipment typi-cally available at laboratories (Figure 11).

Potentiometric stripping, particularly atvery high stripping rates, very short platingtimes, or very low concentration levels ofthe determinants, entails recording relativelyrapid potential changes. In addition to chemi-cal and hydrodynamic conditions affectingdetectability in a given determination, thetime resolution of the analog strip-chart re-corder is also greatly influential. Severalauthors have found it advantageous to usedigital signal processing in this context. Thiscan be done by using a customized dedicatedmicroprocessor system26 or a minicomputerwith satellite process controlling measure-ments.27 Most often, commercially availablemicrocomputers are employed,4,11,28–33 andthe potential of the working electrode issampled at frequencies from 729 to 26 kHz.4

By equipping the electrochemical module

FIGURE 10. Basic instrumentation for potentiometric stripping analysis.

110

FIGURE 11. Block diagram of a computerized instrumental for PSA experiments with galvanic orchemistry stripping. (From Cladera, A.; Estela, J. M.; Cerdà, V. J. Electroanal. Chem. 1990, 288,99–109. With permission.)

with separate memory, which is updatedin hardware during the recording of thestripping step, an extremely high acquisi-tion rate (660 kHz) was achieved with asystem interfaced to an Apple IIe micro-computer.30

A. Electrochemical Cells andElectrodes

The earliest working electrodes used inPSA were of the dropping mercury type;1

however, they soon proved impractical andwere superseded by the rotating glassy car-bon rod electrode, in which the carbon rodwas coated with a mercury film, and a plati-num wire was used as the counterelectrodeand calomel as the reference electrode.2 Thethree-electrode suite was connected to a glassor polyethylene vessel of 20 to 100 ml that

allowed samples to be deaerated if desiredand an oxygen-free atmosphere to be main-tained by bubbling a nitrogen or argon stream(Figure 12).

The working electrode has received thegreatest attention in PSA studies and appli-cations. Although the glassy carbon rod elec-trode has to date been the most widely usedin both analytical applications and theoreti-cal studies, several alternative electrodes havebeen tested in order to improve the applica-bility, reproducibility, sensitivity, and selec-tivity of this technique (Figure 13).

For Hg(II) determination,34 the glassycarbon rod electrode is coated with a copperfilm and potassium permanganate is used asthe oxidant.

A copper coating over a gold film elec-trode has also been used for the indirectdetermination of chlorine-containing spe-cies.35

111

FIGURE 12. Electrochemical cell for potentio-metric stripping analysis. 1, rotating film mercuryglassy-carbon electrode; 2, reference electrode;3, Pt auxiliary electrode; 4, N2 trap.

FIGURE 13. Working electrodes used for potentiometric stripping analy-sis. Electrodes a, b, and d have a total electrode area of 8 mm2 andelectrode c, 110 mm2. (From Jagner, D.; Arén, K. Anal. Chim. Acta. 1978,100, 375–388. With permission.)

112

In RPSA,13 the working electrode is usu-ally a small mercury pool that is placed in aconical cavity in the bottom of the cell whenan amalgamated metal is employed as a re-ductant in the stripping step. If a dissolvedchemical reductant is used instead, a glassycarbon or platinum electrode is fit for thepurpose (e.g., in the determination of Mn(II)by use of dissolved hydroquinone as reduc-tant and a Pt electrode because of its higherreproducibility relative to glassy carbon.

Arsenic(III) can be determined by usinga gold electrode or a glassy carbon electrodecoated in situ with gold plus arsenic duringthe preelectrolysis step.36 The coating film isobtained by adding Au(III) to the sample,which also acts as the oxidant during thestripping step.

Dexiong et al.37 used a rotating glassycarbon disk electrode with a mercury filmcoating for the adsorption PSA for germa-nium, where Alizarin Red complex is formed

and adsorbed at the electrode, thereby in-creasing the sensitivity.

In 1984, Schulze and Frenzel38 introducedhigh-modulus carbon fibers as working elec-trodes for PSA. They used four types of suchelectrodes (Figure 14): single-fiber, cut-fi-ber, fiber-bundle, and cut-bundle electrodes,the small surface area of which resulted indecreased background signals and henceexcellent signal-to-noise ratios. Also, becauseof their small size, fiber electrodes (particu-larly the cut-fiber electrode) offer majoradvantages for processing small sample vol-umes and as detectors for flow systems.

Baranski and Quon39 used a mercury-coated carbon fiber microelectrode as theworking electrode in the microdeterminationof heavy metals; they used microelectro-chemical cells (Figure 15) and an amalgam-ated gold wire as reference electrode.

Frenzel40 designed a microcell (Figure16) consisting of a glassy carbon tube in-

FIGURE 14. Carbon fiber electrodes. (A) single-fiber electrode; (B) cut-fiber electrode; (C) fiberbundle electrode; (D) cut-bundle electrode. (a) Silver wire; (b) epoxy resin; (c) mercury; (d) carbonfiber; (e) cyanacryl glue. (From Schulze, G.; Frenzel, W. Anal. Chim. Acta. 1984, 159, 95–103. Withpermission.)

113

FIGURE 15. Electrochemical cell used inmicroanalysis. (a) Working electrode (carbonfiber sealed in polyethylene); (b), analyzedsolution; (c), reference electrode (amalgam-ated gold wire). (From Baranski, A. S.; Quon,H. Anal. Chem. 1986, 58, 407–412. With per-mission.)

tended to act as an auxiliary electrode. Thetube was stuck on a nipple in the center of aperspex cylinder. In this way, a small bea-ker-like vessel was formed with an approxi-mate volume of 20 µl — sample volumes aslow as 5 µl could be used, however. A labo-ratory-made Ag/saturated AgCl referenceelectrode was inserted upward into theperspex block and connected to the cell viaa 0.1-mm bore in the center of the nipple. Adiaphragm was made by plugging a smallportion of quartz wool into the bore. Thefiber working electrode was held by an ordi-nary laboratory stand and was carefully low-

ered into the beaker until the tip was dippedinto the sample solution.

Based on the results reported by Alberyand Bruckenstein,42 who demonstrated thecomplete hydrodynamic equivalence of thewall-tube electrode and the rotating disk elec-trode, Kapauan41 constructed a PSA cell byusing a wall-tube electrode configuration witha built-in centrifugal pump (Figure 17). Thereproducibility of the system was tested byusing 0.4 µg/ml solutions of zinc, cadmium,and lead at a plating voltage of –1.37 V vs.Ag/AgCl that was applied for 10 s. The rela-tive standard deviations of the measured pla-teau lengths in six runs were 1.1, 0.6, and1.8% for zinc, cadmium and lead, respec-tively.

Alternative types of working electrodeused for enhanced selectivity include chemi-cally modified electrodes (CMEs) and physi-cally (membrane-coated) modified elec-trodes. A dimethylglyoxime chemicallymodified graphite paste electrode was usedfor the determination of Ni(II).43 The elec-trode was made by mixing spectroscopic-grade graphite powder, dimethylglyoxime,and DC200 silicone oil in a 1-ml polyethyl-ene syringe. The surface of the working elec-trode (3 mm2) was renewed daily by press-ing out of the syringe a 1-mm layer of pasteand removing it with filter paper. Electricalcontact was effected by means of a silverwire inserted in the paste. The potentiomet-ric stripping determination of Ni(II) withthis graphite-paste CME involves three steps;the first and second are similar to those inconventional voltammetric stripping deter-minations, viz., chemisorption of Ni(II) ionson the CME surface and reduction of thepreconcentrated nickel at a sufficiently nega-tive potential. In the third step, however,nickel reduced on the electrode surface isoxidized chemically by atmospheric oxygen.

Two general types of polymer-modifiedelectrodes (PMEs) have also been used instripping analysis in order to improve theselectivity (by protecting the surface of the

114

FIGURE 16. Schematic representation of a microliter-capacity cell. (From Frenzel, W. Anal.Chim. Acta. 1987, 196, 141–152. With permission.)

FIGURE 17. Cross section of a wall-tube PSAcell. (From Kapauan, A. F.; Anal. Chem. 1988,60, 2161–2162. With permission.)

115

working electrode from adsorptive interfer-ences): specific and nonspecific. SpecificPMEs are ion-exchanging polymers (iono-mers) that selectively preconcentrate theanalyte within the polymer, whereas non-specific PMEs control access to the elec-trode surface by acting as diffusion barri-ers. The perfluorosulfonate cation-exchangeresin Nafion has been used as a specificPME material in both anodic strippingvoltammetry (ASV) and PSA44 for the de-termination of heavy metals in various en-vironmental and clinical samples. For non-specific PMEs, cellulose acetate dialysismembrane-modified mercury film elec-trodes (CM-MFEs) have been used in ASVand PSA.3 The nonspecific cellulose ac-etate PME material is more advantageousin routine applications than is the specificNafion PME material, primarily as a resultof significant preconcentration by the lat-ter. Six or more replicates per sample arerequired to obtain a steady signal using aNafion-modified MFE in ASV, and con-secutive samples exhibit carryover.45 Thenonspecific cellulose acetate dialysis mem-brane-modified MFE does not precon-centrate analyte so severely, so it requiresfewer replicates per sample and minimizescarryover. The main disadvantage of usinga CM-MFE arises from the presence of arelatively thick membrane at the redox sur-face, which results in diminished sensitiv-ity. However, the sensitivity of a CM-MFE(1000-amu cutoff) is lower than that of anunmodified MFE by a factor of ~18 in ASVbut only ~6 in PSA.

For a nonspecific polymer such as cellu-lose acetate, dialysis occurs across the solu-tion/membrane-electrode interfaces. Thedriving force of dialysis is the concentrationgradient across the membrane, where themembrane’s permeability governs partition-ing between it and adjacent phases. The fluxJ across the membrane is given by

J =dx

dtCs = P C s – Cd( ) (29)

where x is the direction normal to the mem-brane surface, t time, P the membrane per-meability, and Cs and Cd the analyte concen-trations on the sample and detector side ofthe membrane surface, respectively (bothdiffer from the bulk concentrations except atequilibrium.46 The electrochemical drivingforce across the membrane gives rise to asteeper concentration gradient from thechange in oxidation state on amalgamation(in a 1:1 stoichiometry). The use of CCSAwith a dialysis membrane-modified electrodecancels an opposing gradient of divalentcations within the membrane (i.e., the analytevs. the Hg(II) oxidant), thereby increasingthe dialysis efficiency.

Wang and Tian47 assessed the perfor-mance of screen-printed electrodes forvoltammetric and potentiometric strippingmeasurements of trace metals with a view totheir exploitation for single-use decentral-ized testing. Mercury-coated carbon elec-trodes screen-printed on a plastic strip werefound to perform comparably to conventionalhanging mercury drop and mercury-coatedglassy carbon surfaces. Reproducible mea-surements of lead in 100-µl drops were thusobtained, and a detection limit of 30 ng/mlwas estimated following a 10-min precon-centration. Convenient quantitation of leadin urine and drinking water was achieved inthis way. A TraceLab potentiometric strip-ping unit (the PSU20 model from Radiom-eter) furnished with an ordinary Ag/AgClelectrode and a platinum wire auxiliary elec-trode, a SAM20 sample station, and an IBMPS/2-55X computer were used to obtainpotentiograms. In addition to having a greatpotential for single-use decentralized clini-cal or environmental testing, the highly stableresponse of screen-printed electrodes makethem especially attractive for routine, low-cost, centralized operations.

Subsequently, Wang and Tian48 used amercury-free disposable lead sensor basedon PSA at a gold-coated, screen-printed elec-trode. The combination of gold-coated car-bon strips and PSA was found to yield an

116

analytically attractive behavior in contrast tomany earlier unsuccessful attempts at moni-toring lead without the involvement of mer-cury. Changes in peak intensity and position(relative to mercury-coated strips) providenew selectivity dimensions.

V. POTENTIOMETRIC STRIPPINGANALYSIS IN FLOW SYSTEMS

The implementation of PSA in flow sys-tems was first proposed by Anderson et al.49

as a logical response to the inception ofmicrocomputers, the high data-acquisitioncapabilities of which allowed the pre-electrolysis time to be substantially short-ened; as a result, throughput was after thatdetermined by how expeditiously samplechangeover and electrode cleaning weredone. This limitation could obviously beovercome by using a flow system, with theadded advantage that the stripping step neednot involve the sample — in contrast to batchoperation — so use of a suitable strippingsolution (and a matrix-exchange technique)was bound to result in enhanced sensitivityand selectivity in PSA, irrespective of thecomposition of the sample concerned.

The theoretical foundation of PSA in flowcells was established by Anderson et al.49

from equations previously derived for MiCl+,MiCl2, . . . , and DMi by Hanekamp andNiewerk50 for diffusion-controlled PSA. Theamount of Mi(n) reduced and simultaneouslyamalgamated, Mi(Hg), by potentiostatic depo-sition on a mercury-coated glassy carbon elec-trode in a thin-layer cell is given by:

i =1

m

∑ Mi

Hg( ) =t= 0

t dep

∫i =1

m

∑ M i n( )[ ]DMi

2 / 3( )break ?? γν1 / 3ω u deplν–1( )β

dt(30)

where tdep (s) is the time of potentiostaticdeposition, Mi(n) the molar concentrationsof reducible Mi(n) species (e.g., MiCl+,

MiCl2, . . . ), DMi (cm2/s) the diffusion con-stants for such species, ω (cm) the cell spacerwidth, ν (cm2/s) the kinematic viscosity,l (cm) the electrode length, udep (cm/s) themean linear flow during potentiostatic depo-sition, and γ and β two numerical constants.

The amount of amalgamated metal thatis reoxidized (stripped), in millimoles, overa given interval t2 – t1 (s) is given by:

mMi

Hg( )stripped

=t

1

t2∫

k = 1

j

∑ Author l o g i c a l b r e a k??

nmk

–1A

kox( )[ ]D

Aid ox( )

2 / 3( )γν 1 / 3ω ustrip

lν –1( )β

dt

(31)

where [Ak(ox)] is the molar concentration ofone of the j different oxidants present in thestripping solution and capable of diffusion-controlled oxidation of M i(Hg) according tothe reaction

M i Hg( ) + nm k

–1Ak ox( ) → M i n( ) + nm k

–1Ak red( )

(32)

and ustrip is the mean linear velocity duringstripping.

If t1 and t2 are used to designate thebeginning and end, respectively, of Mi(Hg)stripping, the potentiometric stripping signalfor Mi at tMi,strip = t2 – t1 can be represented by

tM i,strip= tdep M i n( )[ ]DM i

2 /3 u depustrip–1( )β

break ??

k =1

j

∑ nmk–1A

kox( )D Ak ox( )

2 / 3[ ] – 1

(33)

provided the kinematic viscosity is the samein the sample as in the stripping solution.From Equation 33 it follows that the potentio-metric stripping signal for Mi(n) will be inde-pendent of the overall electrode surface area.

–1

117

Because in most potentiometric experi-ments the concentrations of the major samplematrix components (e.g., H+ and Cl–) are keptconstant and the composition of the strippingsolution is the same from one analysis to thenext, Equation 33 can be rewritten as

tM i,strip∝ tdep M n( )[ ] udepustrip

–1( )β(34)

which is the fundamental equation for quan-titative PSA in flow cells. Based on previousfindings of Levich,51 Hanekamp and VanNieuwerk50 calculated β = 0.5 for a thin-layer cell as opposed to the experimentalvalue of 0.3 obtained by Swartzfager.52

The potential of the working electrode,E (V), during the stripping of Mi(Hg) isgiven by

E = E0 + RT ln 10( )n

–1F

–1log M i

n +( ) M i Hg( )( )[ ]–1

(35)

where E0 is equivalent to the half-wave po-tential in polarography and (M i

n+) is thesteady-state activity of M i

n+ accumulated atthe electrode surface during stripping. FromEquation 35 it follows that the strippingpotential is independent of the mercury filmthickness. In PSA, however, the glassy car-bon working electrode is usually precoatedwith a film of mercury, so the relative in-crease in film thickness during consecutiveanalyses can be neglected. A side-reactioncoefficient, α = [M(N)]/[Mn+], due to rapidcomplexation reactions at the electrode sur-face with different ligands thus needs to beintroduced. If L is used to designate the pre-dominant complexing agent at the electrodesurface and the free concentration of L ischanged from [L]I to [L]II in two consecutiveexperiments, the stripping potential shift forM i

n+, ∆E (V), will be

∆E = RTn –1F–1 ln 10( )

BREAK ?? log αL[ ]= L[ ]t

– log αL[ ]= L[ ] m

( )(36)

provided complexation is not the rate-deter-mining step in the combined oxidation-com-plexation reaction.

PSA as a detection technique has beenused primarily in two types of flow systems:continuous-flow (CF) and flow-injection (FI)systems. Both CFSPSA and FIPSA systemsare highly automated and differ in the factthat the sample is continuously driven to thedetection cell during the preelectrolysis stepin the former, whereas a sample volume isinjected into a carrier stream that leads it tothe detector (where preelectrolysis is con-ducted) in the latter.

CFSPSA configurations typically con-sist of a flow-cell accommodating a three-electrode system, in addition to a microcom-puter-controlled, six-step PTFE inlet valvefor sample or reagent inlets and a propulsionsystem that can be a peristaltic pump or awater-column suction unit via which the flowrate can be regulated. The remaining ele-ments are those typical of conventional PSAand include a potentiostat (or, occasionally,a galvanostat) and an x-t recorder with ahigh-impedance input or, more commonly, amicrocomputer for data acquisition and pro-cessing. The computer can also be used toactuate the six-way valve, engaged and dis-engage the potentiostat (galvanostat), con-trol a sampler, and enact a series of predeter-mined steps. In some instrumental setups,several of these operations are performed bycommercially available dedicated instru-ments (e.g., the Ion Scanning System ISS820from Radiometer).

The first step in a CFSPSA applicationinvolves conditioning the working electrode.For example, a mercury film electrode canbe conditioned prior to insertion into theflow cell or in situ (by having the valvecirculate the mercury solution through thecell and applying a potential program toensure deposition of the film). Then, thesample is circulated through the cell byswitching the valve and the potentiostat issimultaneously engaged at the desired elec-

118

trodeposition potential and flow rate over apreset interval. Subsequently, the strippingstep can be carried out by disengaging thepotentiostat and either using a chemical oxi-dant previously added to the sample or pass-ing an electric current. Alternatively, strip-ping can be accomplished by changing thematrix; for this purpose, the valve is switchedbefore the potentiostat is disconnected in

order to circulate a suitable stripping solu-tion containing an oxidant or pass an electriccurrent. After the sample channel is flushed,the cycle can be repeated by inserting a freshsample.

The literature abounds with reportedapplications of CFSPSA systems. Andersonet al.49 used a thin-layer cell (Figure 18)accommodating a mercury-coated glassy

FIGURE 18. (A) Flow cell. A, PTFE block with glassy carbonelectrode and sample inlet/outlet; B, PTFE cell spacer; C, Ptcounterelectrode; D, PTFE block; E, reference electrode compart-ment. (B) Details of cell spacer, glassy carbon electrode, and sampleinlet/outlet. (From Anderson, L.; Jagner, D.; Josefson, M. Anal.Chem. 1982, 54, 1371–1376. With permission.)

119

carbon working electrode, and matrix-ex-change stripping, for the selective determi-nation of Tl(I) in the presence of a 104-foldexcess of Cd(II), provided the stripping so-lution contained at least 1 M ammonia. Theyemployed the flow system (Figure 19) todetermine Pb(II) in wines by using a matrix-modifying solution containing 1 M CaCl2,0.1 M HCl, and 0.25 nM Hg(II) in 40%ethanol.

Later, Jagner et al.53 employed CFSPSAsystems for the determination of mercury(II)in various samples of clinical (urine) andenvironmental interest (sediments), as wellas digests from biological materials (orchardleaves and fish muscle), by use of a goldworking electrode. The preelectrolysis stepwas carried out in a sample previously con-ditioned with ammonia and iodide, whilestripping was done in an acidified bromidesolution containing Cr(VI). By using mer-cury standard solutions, a detection limit of2 nM was achieved after a 90-s preelectrol-

ysis, the dynamic range encompassing nearly3 decades. Copper(II) and silver(I) werefound to interfere in a 1000- and fivefoldexcess, respectively, over mercury(II).

Mannino54 used a CFSPSA system simi-lar to one previously employed by Andersonet al.49 in order to determine the optimalconditions for the separation and determina-tion of lead and tin in fruit juices and softdrinks. Preelectrolysis was carried out in astrongly acidic medium using a coated glassycarbon working electrode, while strippingwas done in a deaerated ammonium nitratesolution at pH 4.6. By using the stopped-flow operational mode during stripping, leadand tin could be determined at concentra-tions as low as 0.1 ppb.

Eskilsson and Turner55 used CFSA (Fig-ure 20) with PSA detection (Figure 21) forthe determination of manganese(II) over theconcentration range 2 nM to 30 µM in natu-ral waters, as well as total manganese, fol-lowing pretreatment with hydroxylamine.

FIGURE 19. Flow system designs. (A) Continuous flow; (B) flow injec-tion. (From Anderson, L.; Jagner, D.; Josefson, M. Anal. Chem. 1982,54, 1371–1376. With permission.)

120

FIGURE 20. Block diagram of a PSA flow system. (From Eskilsson, H.; Turner, D. R. Anal. Chim.Acta. 1984, 161, 293–302. With permission.)

FIGURE 21. PSA flow cell. (a) Perspex block with reference electrode compartment and 1-mmflow channel; (b) teflon block containing glassy carbon working (we) and counter (ce) electrodes;(c) clamp to hold cell together. (From Eskilsson, H.; Turner, D. R. Anal. Chim. Acta. 1984, 161, 293–302. With permission.)

121

The experimental procedure involved re-moval, chemical clean-up, and subsequentregeneration of the mercury film formed overthe glassy carbon working electrode aftereach measurement, as well as using a CaBr2-saturated solution for stripping. In this way,the main problem posed by the CFSPSAdetermination of manganese — too rapidstripping rates arising from an incompletelyirreversible reoxidation of Mn(II)-amalgam-ated manganese and resulting in broad, ill-defined peaks — was circumvented. The rateof chemical oxidation during stripping is acrucial parameter that is controlled by theconcentration of oxidant in the strippingsolution and its rate of transport to the elec-trode (the flow rate or rotation speed). Thetwo parameters are adjusted in such a way asto ensure a low stripping rate for both batchand flow measurements. The oxidant con-centration can be lowered for flow measure-ments by using saturated CaBr2 as the strip-ping solution. This approach is necessitatedby the difficulty involved in transporting adeoxygenated solution through flexible tub-ing, which is highly permeable to oxygen —the problem does not emerge if oxygen isinsoluble in the solution used, as is the casewith the calcium bromide solution. The oxi-dant in the stripping solution is assumed tobe residual oxygen. On the other hand, inter-ferences arising from manganese-copper in-teractions in the mercury electrode can besuppressed by adding zinc or gallium, de-pending on the concentration of interferingcopper.

Dyrssen et al.56 studied the effect of us-ing concentrated solutions of salts on thestripping time in computerized flow PSA.They demonstrated that, for 1:1 salts, theeffect was mainly due to salting out of theoxygen, while for 1:2 salts, the effect wasmainly due to salting out of the oxygen,while for 1:2 salts, the effect was due to bothsalting out and an increased viscosity ofoxygen (η) in the salt solution. They alsoshowed the stripping time to be stronglycorrelated with the η/S ratio, where S de-

notes the solubility of oxygen in the saltsolution.

Almestrand et al.57 determined cadmium,lead, and copper in milk and milk powderwith minimal pretreatment by using a highlyautomated CFSPSA system that enabledcontrol of the pump rate, electrolysis timeand potential, and opening and closing ofvalve inlets, in addition to digital evaluationof stripping times. Samples were diluted five-fold with Suprapur hydrochloric acid andelectrolyzed for 0.5 to 4 min prior to strip-ping (also in Suprapur HCl). The analyticalresults were consistent with the certifiedvalues for three milk powder referencesamples. The detection limits for cadmium,lead, and copper in milk samples achievedafter 4-, 1-, and 0.5-min preelectrolysis were0.8, 4, and 8 µg/l, respectively.

Renman et al.4 developed a computer-assisted system for PSA measurements in aCFS configuration using a chemical oxidantor a constant electric current for stripping,and a new cell design (Figure 22). The thin-layer cell was designed for accommodatingTeflon-embedded glassy carbon electrodesin a spring-regulated holder, allowed readyfastening, and exhibited leak-free behavior.Subsequently, the design was used for thedetermination of molybdenum(VI) in seawater58 by potentiostatic adsorption of the8-quinolinol complex of the metal onto amercury-film electrode and subsequent re-duction of the complex by constant-currentstripping in 5 M CaCl2.

Huiliang et al.59 used carbon fiber elec-trodes in flow potentiometric and constant-current stripping analysis applications. Thesignal-to-noise ratio was approximately 1.6times higher for an 8- to 10-µm carbon fiberrelative to the glassy carbon disk electrode.The use of fiber electrodes for the deter-mination of metal ion concentrations by acurrent-measuring technique such as amper-ometry or voltammetry often poses instru-ment problems arising from the very lowcurrents to be monitored. This problem doesnot apply to PSA because the technique’s

122

FIGURE 22. Flow cell assembly. (a) Calomel reference; (b) reference com-partment lid; (c) reference compartment; (d, e) platinum tube flow inlet andoutlet; (f) ceramic plug; (g) polyethylene spacer; (h) spacer holder; (i) glassycarbon rod; (j) teflon body; (k, m) titanium holder; (o) spring; (n) holder top.(From Renman, L.; Jagner, D.; Berglund, R. Anal. Chim. Acta. 1986, 188, 137–150. With permission.)

123

performance is independent of the surfacearea of the working electrode. For this rea-son, fiber electrodes were recognized as suit-able for use in PSA.38,39,60 In contrast to thepotentiometric stripping signal, the magni-tude of the constant-current stripping signalis dependent on the surface area of the work-ing electrode. Consequently, fiber electrodescan be assumed to be inferior to glassy car-bon disk electrodes and other types ofmacroelectrodes used in CCSA. However,based on the experimental results obtainedby using fiber electrodes with chemical orelectrical stripping in addition (usually) to amatrix-exchange technique, in the determi-nation of lead in urine,61 cadmium and leadin whole blood62 with carbon fiber electrodes,arsenic(V) in sea water with a gold-coated,platinum-fiber electrode,63 silver(I) with aplatinum-fiber electrode,64 gold(III) withcarbon and platinum fiber electrodes,66 andZn(II), Cd(II), and Pb(II)44 or organic com-pounds such as erythromycin67 in urine byuse of modified Nafion carbon fibers andmercury-coated fiber electrodes, they are asgood as or even better than other types ofelectrodes typically used in PSA.

In flow injection PSA (FIPSA) (Figure23), fixed-size aliquots of the unknown areinserted at preset times into a carrier flow-ing stream by using an autosampler operat-ing in conjunction with a sample-loop/chro-matographic switching valve combination.With proper consideration of variables suchas the tubing diameter and flow rate, thesample plug is carried along the systemwith very limited dispersion into the carrieron either side. The injected plug can bepassed by an electrode where plating of themetal ions to metal occurs. After the plughas passed by this point, potentiometricstripping can occur in the carrier environ-ment. The working glassy carbon electrodeis usually plated with mercury in situ byadding mercury ion to the sample, thus lead-ing to the formation of analyte amalgamduring the plating process.

The theoretical foundation of the FIPSAtechnique, as established by Hu et al.,68 is asfollows: the limiting current il at a flow-through planar electrode was formulated byLevich51 as

il = nkFCbvfx (37)

FIGURE 23. Block diagram, FIA/PSA instrument. (From Hu, A.; Dessy, R. E.; Granéli, A. Anal.Chem. 1983, 55, 320–328. With permission.)

124

where n is the number of electrons involvedin the process; F the Faraday constant; Cb thebulk concentration of the electroactive spe-cies; vf the flow velocity; k a complex func-tion of the diffusion coefficient, the kine-matic viscosity of the fluid, and the electrodegeometry; and x a constant calculated theo-retically to be 1/3 for the tubular electrodeand 1/2 for the planar electrode.

For a given set of experimental condi-tions, k is constant, so

i = Kv fx (38)

where K = nkFCb.The amount of analyte Q deposited in a

glassy carbon electrode over the electrolysisperiod can be calculated from

Q = iltelec (39)

where telec is the electrolysis time.In flow injection analysis, the sample is

injected into the flow stream as a plug. There-fore, the time during which it is exposed tothe electrode surface determines the effectiveplating time, which can thus be defined as

t elec =Vs

vf

(40)

where Vs is the sample volume and νf theflow velocity.

Appropriate combinations of Equations38 to 40 yield

Q = i l t elec = i l

Vs

vf

(41)

Q = KVsv fx–1 (42)

According to Equation 42, the amountof analyte plated out is inversely proportionalto the flow rate.

If the limiting current can be measuredat various flow rates, a plot of log (νf) vs. log

(il) will give a straight line of slope x. Like-wise, if the amount of analyte deposited canbe experimentally determined, a plot of Qvs. log (νf) will also give a straight line ofslope x – 1.

Buffle et al.69 related the rate of oxida-tion to the oxidant concentration by

dN

dt=

AD Cb – C0( )H

(43)

where dN/dt is the rate of oxidation, A theelectrode surface area, D the diffusion coef-ficient, H the diffusion layer thickness, Cb

the bulk oxidant concentration, and C0 theoxidant concentration at the electrode sur-face.

If the oxidation process is diffusion-con-trolled and the bulk oxidant concentration ismuch larger than that at the electrode sur-face, Equation 43 simplifies to

dN

dt=

ADC b

H(44)

Consequently, the rate dN/dt remainsconstant provided the oxidant concentra-tion Cb remains unchanged. Bruckensteinand colleagues70,71 and Jagner and col-leagues.1,2,26,72,73 investigated the matter andfound the simplification to be warranted.The following relationship was found toexist between Q (the amount of analytedeposited) and tstrip (the time need for theamalgamated metal to be oxidized):

Q =ADC

bt

strip

H(45)

Therefore, the time required for stripping ametal from the electrode surface can be usedto quantify the concentration of the metal inquestion. Because

KVsv fx–1

nF=

ADCbt

strip

H(46)

a plot of log (tstrip) vs. log (νf) will also givea straight line of slope x – 1.

125

Research into FIPSA systems has beenconcerned primarily with the design of highlyautomated systems, the performance of elec-trochemical flow cells, and procedures forincreasing the selectivity (by use of the ma-trix-exchange technique during stripping) andsample throughput.

Hue et al.74 proposed an FIPSA systemfor the automatic determination of copper,cadmium, and lead in ground water andevaluated the performance of the matrix-exchange technique for the separation ofclose oxidation peaks. Heavy metals com-monly found in ground water, such as lead,thallium, cadmium, bismuth, and tin, wereassayed in order to explore the scope andlimitations of the method. Data generatedduring the stripping step were digitized andacquired at a high speed by a minicomputerequipped with a satellite processor, and thedigital numbers obtained for each time inter-val were mapped into a data file, therebyincreasing the memory cell’s content. Eachmemory cell represents a channel of a mul-tichannel analyzer. This provides the firstderivative of the stepped potential scan, yield-ing “peak”-type data in a direct fashion. Thenumber of counts accumulated in thememory-mapped multichannel analyzer de-pends on the oxidation rate of the analyte aswell as the data acquisition rate of the ana-log-to-digital converter used. The recordednumber of counts on a channel (or cell) isproportional to the data acquisition rate andresolution of the converter (in volts per chan-nel). It is inversely proportional to the oxida-tion rate, that is, the slope of the potential-time curve (in volts per second). This lineardependence of the observed signal on thedata acquisition rate will have a unique slopethat is characteristic of the oxidation rate ata given point on the potential time curve.Metals with different oxidation rates shouldexhibit a different dependence of the signalmagnitude on the data acquisition rate. Therelative magnitude of the signal-to-baselineratio remains constant regardless of the dataacquisition rate used. However, the absolute

magnitude of the signal-to-baseline differ-ence does increase as the data acquisitionrate is raised. In other words, the minimumquantity the technique can detect can be re-duced by using a high data-acquisition rate.The best analogy is the improvement incounting statistics in radiochemistry that re-sults from an increased observation period.

Schulze et al.75 investigated the use ofthe matrix-exchange technique in FIPSA. Forthis purpose, they measured the strippingpotentials of cadmium, lead, thallium, andtin in various supporting electrolytes. Thedata obtained were used to optimize the com-position of the carrier solution for the simul-taneous determination of mixtures of theseelements. Complexing agents such as cit-rate, tartrate, and ammonia led to a dramaticpotential shift for bivalent ions, thus remov-ing hindrances on the simultaneous determi-nation. Among the nearly 40 different sup-porting electrolytes tested, NH3/KOH wasfound to be so good that the four elements(Cd, Tl, Pb, and Sn) could be determinedsimultaneously even at concentrations dif-fering by as much as one order of magni-tude.

Frenzel and Bratter76 proposed a FIPSAsystem featuring extremely short residencetimes (less than 1 s) and a proportionallyhigh throughput (up to 200 samples per hour)(Figure 24). Up to four elements could bedetermined simultaneously at concentrationsfrom a few micrograms per liter to severalhundredths. On-line sample manipulation(dilution and matrix modification) was pos-sible by using one- and two-channel flowsystems. The utility of FIPSA was evaluatedby comparing the response, sensitivity, andseveral practical aspects of four differentflow-through cells: a Metrohm EA 1096 wall-jet detector, a laboratory-made wall-jet cell,a thin-layer cell, and an open thin-layer cell.The ensuing method was used successfullyfor the fast sequential quantitation of zinc,lead, and copper in tap water, and the directdetermination of lead and cadmium in aciddigests from biological samples with no pre-

126

FIGURE 24. Two-channel system for on-line sample manipula-tion. (From Frenzel, W.; Bratter, P. Anal. Chim. Acta. 1986, 179,389–398. With permission.)

treatment. Frenzel et al. checked for the theo-retical prediction of the FIPSA technique ofa linear signal dependence on the residencetime of the sample; they used times in the0.5- to 600-s range and different injectionvolumes, flow rates, and metal ion concen-trations. The prediction was found to holdfor all four types of cell. In contrast to thetheoretical predictions for flow-injectionanodic stripping voltammetry, the strippingtime in FIPSA is inversely proportional tothe flow rate as long as this is not alteredbetween deposition and stripping. This is aresult of the increased plating efficiency athigh flow rates being offset by the likewisefaster transport of oxidant during stripping.In all cases, experimental verification wasobtained for flow rates in the 0.2- to 6-ml/min range except for the open thin-layercell, where the maximum useful flow ratewas 2 ml/min (the filter strip was flushedaway at higher flow rates). The question ofwhich cell was best suited for FIPSA couldnot be answered unambiguously, however.The shape and length of the stripping curves

were nearly identical under comparable ex-perimental conditions. With respect to re-sponsiveness and carryover, the laboratory-made wall-jet cell proved superior. Forpractical reasons, the open thin-layer cell(Figure 25) was found to be preferable be-cause disassembling for cleaning and remov-ing air or hydrogen bubbles was obviouslyunnecessary.

Locascio and Janata77 designed and testedan automatic system for the determination oflead(II) and cadmium(II) in the 10–3 to 10–6

M range. It consisted of solid-state integratedcircuitry containing a chemically sensitivefield-effect transistor (CHEMFET) and atransistor control switch (Figure 26). Flowof solutions, application of potentials duringthe electrode preparation, plating and strip-ping periods, and data acquisition were con-trolled by an inexpensive microprocessor.Sample volumes of 200 µl containing themetals at concentrations in the above-mentioned range could be processed in lessthan 2.5 min with a relative standard devia-tion below 10%. The detection limit was at

127

FIGURE 25. Schematic diagram of open thin-layer cell.GCE, 1 mm glassy carbon working electrode; Pt, 0.2-mmauxiliary electrode; SCE, saturated calomel referenceelectrode; F, filter strip. (From Frenzel, W.; Schulze, G.Analyst. 1987, 112, 133–136. With permission.)

the time restricted by the low sampling rateof the microprocessor used (a CommodoreVIC20). Also, the upper value of the usableconcentration range was limited by the maxi-mum practical concentration of oxidant.

An electrochemical flow-cell suitable(Figure 27) for use with dual detection byPSA and atomic absorption spectrometry ina flow injection system was characterized bySchulze from its flow pattern and dispersionmeasurements.78 The deposition efficiencycould be altered from 1 to 2% to 24% byusing a glassy carbon and a carbon felt elec-trode, respectively. Application to the deter-mination of lead in water showed that, afterenrichment on carbon felt, the sensitivity forflame atomic absorption spectrometry wasincreased by one order of magnitude (en-richment factors of ~5 and ~30 at sampling

frequencies of 60 and 12 h–1, respectively).Dual detection allowed some errors arisingfrom apparatus malfunctioning or a wrongsample treatment in one of the methods to beidentified, thus providing accuracy checks.

Matuszewski et al.31 reported an FIPSAsystem using an IBM PC/XT computer forrapid digital recording of potential-timecurves. A straightforward high-volume wall-jet cell79 was used as detector cell (Figure28). Investigations of the influence of elec-trode pretreatment and solution deliverymode (i.e., by gravity flow, pumping, etc.)revealed these parameters to have a signifi-cant effect on the magnitude and precisionof stripping times. The optimized systemwas applied to the simultaneous determina-tion of cadmium and lead in geologicalsamples.

128

FIGURE 26. Circuit diagram of the integrated potentiometric strip-ping system. The auxiliary electrode (Aux) is connected to the outputof operational amplifier 1 and supplies the necessary current tomaintain a constant potential on the reference electrode (Ref). AD/A converter that is connected to the positive input of the sameamplifier sets the potential of the reference electrode. S1a and S1bare switched by a DPTD relay, and S2 is switched independently byan SPDT relay. When S1a, S1b, and S2 are in the positions shownin the diagram, the CHEMFET is in the plating mode. When S1a,S1b, and S2 are in their alternative positions, the CHEMFET is in thestripping mode. The output voltage of operational amplifier 2 isrecorded during the stripping cycle. The value of feedback resistor Ris 25.9 kΩ. The part of the circuit inside the dashed line area is onone integrated silicon chip. (From Locascio, L.E.; Janata, J. Anal.Chim. Acta. 1987, 194, 99–107. With permission.)

VI. INTERFERENCES INPOTENTIOMETRIC STRIPPINGANALYSIS

PSA interferences can be classified intotwo broad categories: those arising fromthe presence of metals and those intro-duced by organic substances in thesamples.

The interferences posed by metals can inturn be classified into four groups:80

1. Class 1 interferences arise frominterferents, I, that decreased the ana-lytical signal of the analyte, A, througha parallel oxidative action of ionic Itoward the deposited A. Obviously, Imust be a stronger oxidant than A in

129

FIGURE 27. Flow-through cell for potentiometric stripping analy-sis. (a) working electrode; (b) reference electrode; (c) auxiliaryelectrode. (From Schulze, G.; Koschany, M.; Elsholz, O. Anal.Chim. Acta. 1987, 196, 153–161. With permission.)

the given ionic medium. The followingreactions take place in parallel duringthe stripping of A:

A(Hg) + (n/2) Hg2+ → An+ + (n/2) Hg(I)(47)

m A(Hg) + n Im+ → m An+ + n I(Hg)(48)

where M(Hg) denotes amalgamated M.The magnitude of the signal decrease

will depend on the relative reaction-rate constants of the heterogeneous Re-actions 47 and 48, and the [Im+]/[Hg2+]ratio.

The proportionality between theanalyte concentration is generally pre-served. Typical cases of class 1 inter-ferences are observed with the follow-ing A-(I 1,I2, . . .) pairs: Zn-(Cd, Tl, In,Pb, Bi), Ga-(Cd, Tl, Pb), Cd-(Pb, Bi),Tl-(Pb, Cu, Bi), In-(Pb, Cu, Bi), Pb-(Cu, Bi), and Cu-Bi. On the other hand,no class 1 interferences have been de-tected for Zn-(In, Pb) in 0.1 M HCl.

2. Class 2 is only observed if I is lessoxidant than A and is capable of beingplated, giving a stripping plateau pre-ceding that of A. An increase in theanalytical signal is observed that can beattributed to the direct reduction of An

130

FIGURE 28. (A) Manifold used in flow injection measurements. C, carrier solution;R, reagent solution; D, detector cell. (B) Configuration of the measuring cell in flowinjection measurements. W, glassy carbon working electrode with perpex cap in wall-jet arrangement; R, reference electrode; C, counter electrode. (From Matuszewski,W.; Trojanowicz, M.; Frenzel, W. Fresenius Z. Anal. Chem. 1988, 332, 148–152. Withpermission.)

at the electrode surface by amalgamatedI, which takes place in two simultaneousreactions during the stripping of I:

I(Hg) + (m/2) Hg2+ → Im+ + (m/2) Hg(49)

n I(Hg) + m An+ → m A(Hg) + n Im+

(50)

In most instances, for the same timeinterval, Reaction 50 yields largeramounts of A(Hg) during the strippingof I than does the following reactionduring the deposition step:

An+ + n e– → A(Hg) (51)

In contrast to class 1 interferences,class 2 interferences may be readilysuppressed or completely circumventedby a judicious choice of the deposition

potential. Thus, using a deposition po-tential between the stripping potentialsof I and A will prevent deposition ofthe former.

The proportionality between the ana-lytical signal and the analyte concen-tration is generally preserved unlessother classes of interferences are in-volved.

Typical class 2 interferences are ob-served with the following pairs: Cd-Zn,Tl-Zn, In-Zn, Pb-(Zn, Cd, In), Cu-(Zn,Cd, In, Pb, Tl), and Bi-(Zn, Cd, Tl, In,Pb).

3. Class 3 interferences are observed whenA and I form one or more intermediatecompounds. These interferences arereadily distinguished from class 1 andclass 2 interferences because, in mostcases, they appear at relatively lowinterferent concentrations. They usu-ally result in a dramatic decrease in the

131

analytical signal. An increased signal isto be expected, however, if the inter-metallic compound is stripped at thesame potential as the analyte. Interme-tallic compounds are known to behaveas separate metals and are reportedlyoxidized at potentials between the oxi-dation potentials of the individual met-als or at the same potential as one of theconstituent metals.81

In the presence of low interferentconcentrations, the analytical signal islinearly related, but not proportional, tothe analyte concentration because ei-ther part of the analyte is consumed bythe interferent to form the intermetalliccompound or this is stripped togetherwith the analyte. The analytical signaldecreases or increases by a constantamount that is proportional to theinterferent concentration.

Typical instances of class 3 interfer-ences with a decrease in the analyticalsignal are observed with the followingpairs: Zn-(Cu, Ni, Co), Ga-(Cu, Ni, Co,Zn), and Cd-(Cu, Ni, Co), while a sig-nal increase is observed with the Cu-Znpair. This can only be partly ascribed toa class 3 interference, as the predomi-nant interference is of class 2.

4. Class 4 interferences encompass allinteractions not amenable to classes 1through 3. They may originate fromvarious phenomena:

A. Chemical interaction between Aand I prior to the deposition step(e.g., inhibition of the depositionof A by coprecipitation with basicsalts of I). Typical instances areobserved with the following pairs:Zn-[Cr(III), Al] in acetate buffer atpH 4.0, Ga-[Cr(III),Al], Cd-[Cr(III), Al], Tl-[Cr(III), Al, Ga)],and In-(Al, Ga). In all cases, addi-tion of the interferent after thedeposition step has virtually noeffect on the analytical signal.

B. Formation of a separate crystallinephase of I on the surface of themercury layer may also consider-ably influence the analytical signalin either direction. In most cases,E-t recordings are accompanied bynoise that obscures the signal. Typi-cal instances are observed with thefollowing pairs: Tl-(Ni, Co), In-(Ni, Co), Pb-(Ni, Co), and Bi-Ni.

C. Interferences observed mainly withCa, Mg, and Mn are not amenableto either of the above describedphenomena. Such interferences areobserved with the following pairs:Zn-(Mn, Ca, Mg), Ga-(Mn, Ca,Mg), Tl-(Ca, Mg), Bi-Al, and Bi-Cu — this is in contradiction withclass 1 Cu-Bi, as α-class 2 Bi-Cuwould be expected).

Not only the proportionality, but alsothe linearity between the analytical sig-nal and the analyte concentration seemto be affected by class 4 interferences.The relationship between the analyticalsignal and the analyte concentrationshould be checked in each separate in-stance.

It should be stressed that mixed-classedinterferences are also likely to occur. Onetypical example is the Cd-Ga pair, where class2 and 3 interferences may compete with eachother, resulting in an unusual, consistentlynonmonotonic interference pattern.

The specialized literature includes vari-ous procedures for the elimination of inter-ferences due to the presence of metals. Thus,Danielson et al.82 overcame the strong inter-ference of copper in the determination ofzinc by adding gallium, which forms a morestable intermetallic compound with copper,thus allowing zinc to be oxidized at its nor-mal stripping potential. Copper(II) can thusbe determined in the presence of zinc fol-lowing deposition at a potential where Zn(II)is not reduced.

132

Hoyer and Kryger83 used the generalizedstandard-addition method (GSAM) to resolveoverlapping signals and correct those affectedby the formation of intermetallic intermedi-ates because the normal SAD provided in-accurate results if the analytical strippingsignal was overlapped by one or more sig-nals resulting from the stripping of nonanalytecomponents. In addition, if an amalgamatedmetal takes part in the formation of an inter-metallic compound with plated nonanalytecomponents, care should be exercised toeliminate this effect prior to application ofthe normal SAD.

On the other hand, these interferencescan readily be overcome in many cases.Nearly coincident stripping signals can of-ten be well resolved by using another me-dium; also, their relative magnitudes canmany times be controlled via the platingconditions. Flow techniques with mediumexchange provide an additional means forcontrolling the selectivity.

As regards the interferences posed bythe presence of organic matter (OM) in PSA,the technique is much less markedly affectedby adsorbed OM than are other strippingtechniques such as ASV because, when OMcovers the electrode, the diffusion of boththe metal ion and the oxidant are similarlyhindered, so the effects tend to cancel eachother out. Moreover, organic redox com-pounds do not interfere with the analysis, someasurements of the electroactive fractionof the metal can be performed in complexmedia, which results in less complicatedsample pretreatments.

The adsorption of proteins and surfac-tants or the accumulation of reaction prod-ucts at the electrode surface can result in agradual loss of electrode activity when amercury electrode is used for the determina-tion of heavy metals in complex mixtures.Obvious ways of improving the selectivityin PSA include using modified electrodes toprotect their surfaces from adsorptive inter-ferences, a matrix-modifying solution, or theSAM in the analyses.

With voltammetric techniques, solublereversible redox couples may pose interfer-ence problems. The equilibrium concentra-tions of such couples at the working elec-trode are determined by the controlledelectrode potential according to the Nernstequation. When the electrode potential ex-ceeds the characteristic potential of thecouple, a rapid, transport-controlled redoxconversion takes place and a Faradaic signalis obtained as a result. Reversible couplesmay often exist as intermediate productswhen natural samples containing traces oforganic matter are subjected to voltammetricanalysis. With pulse voltammetric tech-niques, where rapid potential steps are em-ployed, the effect is more pronounced thanwith slow potential sweeps.

In PSA, where the sample contains ex-cess oxidant, dissolved reversible couplesonly exist while potentiostatic control ismaintained (i.e., during the plating period).During stripping, when potentiostatic con-trol is abandoned, only the oxidized formcan exist, the reduced form being convertedso rapidly that no characteristic plateau orpeak is observed. The transport rate of oxi-dant toward the working electrode is of littleconsequence here because the reduced formis not preplated on the electrode and there-fore is at a low concentration relative to theconcentration of oxidant immediately avail-able at the electrode. Nevertheless, the ana-lytical signal may be decreased as a result ofthe organic compound concerned having areduction potential similar to the analyteoxidation potential. The decrease is usuallyinsignificant, provided the SAM is used forquantitative work.

VII. APPLICATIONS OFPOTENTIOMETRIC STRIPPINGANALYSIS

PSA is not as general an analytical tech-nique for the determination of metal tracesas is graphite-furnace atomic absorption spec-troscopy, for example. However, in samples

133

containing high concentrations of inorganicsalts, the PSA technique seemingly surpassesspectroscopic techniques for the determina-tion of some elements. Typical samples arebody fluids, mineral acid digests from bio-logical materials, saline waters, and chemi-cals for purity testing. Table 1 summarizesselected applications reported after the lit-erature review by Jagner14 was published.Of special note in this context is the recentwork of Ostapczuk et al.6,84 and Labar,85 whoassessed the present potential and limita-tions of the determination of trace elementsby PSA in various types of samples of bothclinical and environmental interest, and re-ported the results obtained in interlaboratoryquality-control experiments.

A second area of application for PSA isthe determination of the different chemicalforms of trace elements in samples,14 in com-mon with nearly every electroanalytical tech-nique. By proper adjustment of the potentio-

static deposition potential, the working elec-trode can be made to respond only to hydroor chloro complexes of the trace metal andnot to the EDTA complex, for example.Applications of this type include those re-ported by Jagner and Aren86 and Fayyad,87

who determined the conditional stabilityconstants for the EDTA complexes of Pb(II)and Bi(III), and the dissociation potentialsof the Bi-NTA and Bi-EDTA complexes(NTA denotes nitrilotriacetic acid), respec-tively, using computerized flow-injection.

The PSA technique can also be used asa complementary analytical technique inconjunction with atomic absorption spectros-copy for analysis of a wide variety of samplesin order to comply with the results dictatedby national legislation boards, which areincreasingly demanding that samples be ana-lyzed by two different analytical techniques.For many toxic trace metals, PSA seems tobe the best complement.

134

135

136

137

138

REFERENCES

1. Jagner, D.; Graneli, A. Anal. Chim. Acta. 1976,83, 19–26.

2. Jagner, D.; Aren, K. Anal. Chim. Acta. 1978,100, 375–388.

3. Aldsstadt, J. H.; Dewald, H. D. Anal. Chem.1993, 65, 922–926.

4. Renman, L.; Jagner, D.; Berglund, R. Anal.Chim. Acta. 1986, 188, 137–150.

5. Jagner, D.; Renman, L.; Wang, Y. Anal. Chim.Acta. 1992, 267, 165–169.

6. Ostapczuk, P. Anal. Chim. Acta. 1993, 273,35–40.

7. Mortensen, J.; Ouziel, E.; Skoy, H.; Kryger,L. Anal. Chim. Acta. 1979, 112, 297–312.

8. Kryger, L. Anal. Chim. Acta. 1980, 120, 19–30.

9. Zie, Y.; Huber, C. O. Anal. Chim. Acta. 1992,263, 63–70.

10. Jin, W.; Wang J. Anal. Chim. Acta. 1991,252, 59–64.

11. Cladera, A.; Estela, J. M.; Cerdà, V. J.Electroanal. Chem. 1990, 288, 99–109.

12. Christensen, J. K.; Kryger, L. Anal. Chim.Acta. 1980, 118, 53–64.

13. Christensen, J. K.; Kryger, L.; Mortensen, J.;Rasmussen, J. Anal. Chim. Acta. 1980, 121,71–83.

14. Jagner, D. Analyst. 1982, 107, 593–599.

15. Chau, T. C.; Li, D.; Wu, Y. L. Talanta. 1982,29, 1083–1087.

16. Mortensen, J.; Britz, D. Anal. Chim. Acta,1981, 131, 159–165.

17. Labar, C.; Lamberts, L. Anal. Chim. Acta.1981, 132, 23–33.

18. Hussam, A.; Coetzee, J. F. Anal. Chem. 1985,57, 581–585.

19. Hoyer, B.; Kryger, L. Talanta. 1985, 32(9),839–844.

20. Xia, Y. H.; Chou, T. C.; Mo, J. Y. Anal.Chim. Acta. 1989, 222, 263–268.

21. Labar, C.; Muller, R.; Lamberts, L. Electro-chim. Acta. 1991, 36(14), 2103–2108.

22. Garai, T.; Mészáros, L.; Bartalits, L.; Locatelli,C.; Fagioli, F. Electroanalysis. 1991, 3, 955–961.

23. Vries, W. T.; Van Dalen, E. J. Electroanal.Chem. 1964, 8, 366.

24. Vries, W. T.; Van Dalen, E. J. Electroanal.Chem. 1967, 14, 315.

25. Garai, T.; Nagy, Z.; Mészáros, L.; Bartalits,L.; Locatelli, C.; Fagioli, F. Electroanalysis.1992, 4, 899–904.

26. Graneli, A.; Jagner, D.; Josefson, M. Anal.Chem. 1980, 52, 2220.

27. Hu, A.; Dessy, R. E.; Graneli, A. Anal. Chem.1983, 55, 320.

28. Almestrand, L.; Jagner, D.; Renman, L.Talanta. 1986, 33, 991.

29. Hoyer, B.; Kryger, L. Talanta. 1986, 33, 883.

30. Hoyer, B.; Skov, H. J.; Kryger, L. Anal. Chim.Acta. 1986, 188, 205.

31. Matuszewski, W.; Trojanowics, M.; Frenzel,W. Fresenius Z. Anal. Chem. 1988, 332, 148–152.

32. Cladera, A.; Estela, J. M.; Cerdà, V. J. Autom.Chem. 1990, 12, 108–112.

33. Cladera, A.; Estela, J. M., and Cerdà, V.Talanta. 1991, 38, 1475–1479.

34. Jagner, D. Anal. Chim. Acta. 1979, 105, 33–41.

35. Xie, Y.; Huber, C. O. Anal. Chem. 1991, 63,208–212.

36. Jagner, D.; Josefson, M.; Westerlung, S. Anal.Chem. 1981, 53, 2144–2146.

37. Dexiong, F.; Peihui, Y.; Zhaoliang, Y.Talanta. 1991, 38(12), 1493–1498.

38. Schulze, G.; Frenzel, W. Anal. Chim. Acta.1984, 159, 95–103.

39. Baranski, A. S.; Quon, H. Anal. Chem. 1986,58, 407–412.

40. Frenzel, W. Anal. Chim. Acta. 1987, 196,141–152.

41. Kapauan, A. F. Anal. Chem. 1988, 60, 2161–2162.

42. Albery, W. J.; Bruckenstein, S. J. Electroanal.Chem. 1983, 144, 105–112.

139

43. Trojanowicz, M.; Matuszewski, W. Talanta.1989, 36(6), 680–682.

44. Huiliang, H.; Jagner, D; Renman, L. Anal.Chim. Acta. 1988, 207, 17–26.

45. Aldstadt, J. H.; Dewald, H. D. Anal. Chem.1992, 64, 3176–3179.

46. Martins, E.; Bengtsson, M.; Johansson, G.Anal. Chim. Acta. 1985, 169, 31–42.

47. Wang, J.; Tian, B. Anal. Chem. 1992, 64,1706–1709.

48. Wang, J.; Tian, B. Anal. Chem. 1993, 65,1529–1532.

49. Anderson, L.; Jagner, D.; Josefson, M. Anal.Chem. 1982, 54, 1371–1376.

50. Hanekamp, H. B.; Van Niewkerk, H. J. Anal.Chim. Acta. 1980, 121, 13–22.

51. Levich, V. G. Physicochemical Hydrodynam-ics. Prentice-Hall: Engelwood Cliffs, N. J.,1962.

52. Swartzfager, D. G. Anal. Chem. 1976, 48,2189–2192.

53. Jagner, D.; Josefson, M.; Aren, K. Anal. Chim.Acta. 1982, 141, 147–156.

54. Mannino, S. Analyst. 1984, 109, 905–907.

55. Eskilsson, H.; Turner, D. R. Anal. Chim. Acta.1984, 161, 293–302.

56. Dyrssen, D.; Eskilsson, H.; Haraldsson, C. J.Electroanal. Chem. 1989, 262, 161–165.

57. Almestrand, L.; Jagner, D.; Renman, L.Talanta. 1986, 33(12), 991–995.

58. Hua, C.; Jagner, D.; Renman, L. Anal. Chim.Acta. 1987, 192, 103–107.

59. Huiliang, H.; Hua, C.; Jagner, D.; Renman, L.Anal. Chim. Acta. 1987, 193, 61–69.

60. Jennings, V. J.; Morgan, J. E. Anal. Proc.1983, 20, 276.

61. Huiliang, H.; Jagner, D.; Renman, L. Talanta.1987, 34(6), 539–542.

62. Almestrand, L.; Jagner, D.; Renman, L. Anal.Chim. Acta. 1987, 193, 71–79.

63. Huiliang, H.; Jagner, D.; Renman, L. Anal.Chim. Acta. 1988, 207, 37–46.

64. Huiliang, H.; Jagner, D.; Renman, L. Anal.Chim. Acta. 1988, 207, 27–35.

65. Huiliang, H.; Hua, C.; Jagner, D.; Renman, L.Anal. Chim. Acta. 1988, 208, 237–246.

66. Huiliang, H.; Jagner, D.; Renman, L. Anal.Chim. Acta. 1988, 208, 301–306.

67. Hua, C.; Jagner, D.; Renman, L. Talanta.1988, 37(7), 525–529.

68. Hu, A.; Dessy, R. E.; Granéli, A. Anal. Chem.1983, 55, 320–328.

69. Buffle, J.; Pelletier, M.; Monnier, D. J.Electroanal. Chem. 1973, 43, 185.

70. Brukenstein, S.; Bixier, J. W. Anal. Chem.1965, 37, 786.

71. Bixier, J. W.; Bruckenstein, S. Anal. Chem.1965, 37, 391.

72. Jagner, D. Anal. Chem. 1978, 50, 1924.

73. Jagner, D. Anal. Chem. 1979, 51, 342.

74. Hu, A.; Dessy, R. E.; Granéli, A. Anal. Chem.1983, 55, 320–328.

75. Shulze, G.; Bonigk, W.; Frenzel, W. FreseniusZ. Anal. Chem. 1985, 322, 225–269.

76. Frenzel, W.; Bratter, P. Anal. Chim. Acta.1986, 179, 389–398.

77. Locascio, L. E.; Janata, J. Anal. Chim. Acta.1987, 194, 99–107.

78. Schulze, G.; Koschany, M.; Elsholz, O. Anal.Chim. Acta. 1987, 196, 153–161.

79. Frenzel, W.; Schulze, G. Analyst. 1987, 112,133–136.

80. Psaroudakis, S. V.; Efstathiou, C. E. Analyst.1987, 112, 1587–1591.

81. Heineman, W. R.; Mark, H. B.; Wise, J. A.;Roston, D. A., in Laboratory Techniques inElectroanalytical Chemistry; P. T. Kissingerand W. R. Heineman, Eds.; Marcel Dekker:New York, 1984; p. 156.

82. Danielson, L. G.; Jagner, D.; Josefson, M.;Westerlung, A. Anal. Chim. Acta. 1981, 127,147.

83. Hoyer, B.; Kryger, L. Anal. Chim. Acta. 1985,167, 11.

84. Froning, M.; Mohl, C.; Ostapczuk, P.Fresenius J. Anal. Chem. 1993, 345, 233–235.

85. Labar, C. H. Electrochim. Acta. 1993, 38(6),807–813.

140

86. Jagner, D.; Aren, K. Anal. Chim. Acta. 1982,137, 201.

87. Fayyad, M. Electroanalysis. 1990, 2, 631–635.

88. Pheiffer Madsen, P.; Drabaek, I.; Sorensen, J.Anal. Chim. Acta. 1983, 151, 479–482.

89. Mannino, S. Analyst. 1983, 108, 1257–1260.

90. Guo, X. Fenxi Huaxue. 1983, 11, 362–364.

91. Bogoslovskii, V. V.; Gabor Klatsmanyi, P.;Pungor, E. Magy. Kem. Foly. 1984, 90, 282–284.

92. Pereira-Ramos, J. P.; Chivot, J.; Sarantopoulos,F.; Godard, R. Analusis. 1984, 12, 249–255.

93. Obiols, J.; Ribera, L. Afinidad. 1984, Novem-ber–December, 513–517.

94. Wang, E.; Sun, W.; Yang, Y. Anal. Chem.1984, 56, 1903–1906.

95. Wang, G.; Ma, M.; Yan, J. Huaxue Xuebao.1984, 42, 283–287.

96. Wang, Z.; Wang, S. Fenxi Huaxue. 1984, 12,796.

97. Mannino, S.; Bianco, M. J. Micronutr. Anal.1985, 1, 47–53.

98. Jaya, S.; Prasada Rao, T.; Prabhakara Rao, G.Talanta. 1985, 32, 1061–1063.

99. Wang, E.; Sun, W. Anal. Chim. Acta. 1985,172, 365–370.

100. Zhu, C.; Yan, H. Huanjing Kexue. 1985, 6,28–32.

101. Jaya, S.; Rao, T. P.; Rao, G. P. Bull.Electrochem. 1986, 2, 131–133.

102. Manino, S.; Bianco, M. Riv. Soc. Ital. Sci.Aliment. 1986, 15, 393–396.

103. McKinnon, A.; Scollary, G. Analyst. 1986,111, 589–591.

104. Fayyad, M.; Tutunji, M.; Ramakrishna, R. S.;Taha, Z.H.A. Analyst. 1986, 111, 471–473.

105. Psaroudakis, S. V.; Efstathiou, E. Analyst.1989, 114, 25–28.

106. Wu, G. Xiamen Daxue Xuebao ZiranKexueban. 1986, 25, 734–736.

107. Noren, J. G.; Hulthe, P.; Gillberg, C. Swed.Dent. J. 1987, 11, 45–52.

108. Fayyad, M.; Tutunji, M.; Ramakrishna, R. S.;Taha, Z. Anal. Lett. 1987, 20, 529–535.

109. Rozali bin Othman, M.; Hill, J. O.; Magee,R. J. Fresenius Z. Anal. Chem. 1987, 326,350–353.

110. Fayyad, M. Anal. Chem. 1987, 59, 209–211.

111. Wang, Y.; Zhang, S. Fenxi Huaxue. 1987, 15,31–35.

112. Huang, P. Fenxi Huaxue. 1987, 15, 574.

113. Wang, Z.; Wang, T. Zhonghua YufangyixueZazhi. 1987, 21, 297–298.

114. Teng, J.; Feng, D.; Zhou, D. Jinan LiyiXuebao. 1988, 3, 41–45.

115. Fayyad, M.; Tutunji, M.; Taha, Z. Anal. Lett.1988, 21, 1425–1432.

116. Huiliang, H.; Jagner, D.; Renman, L. Anal.Chim. Acta. 1988, 207, 17–26.

117. Zhou, J.; Yu, X.; Zhang, S. Gongye WeishengYu Zhiyebing. 1988, 14, 15, 29–32.

118. Xiang, Y. Zhongguo Tiaowiepin. 1988, 4, 28–30.

119. Jin, L.; He, P.; Wu, B. Fenxi Shiyanshi. 1988,7, 14–16.

120. Zhou, J.; Zhou, H.; Hu, Y.; Li, D. Jinan LiyiXuebao. 1988, 3, 92–94.

121. Yang, P.; Li, D. Jinan Daxue Xuebao. 1989,1, 31–36, 59.

122. Huang, P.; Sun, P.; Liu, Y. Lihua JianyanHuaxe Fence. 1989, 25, 265–267.

123. Psaroudakis, S. V.; Efstathiou, E.; Hadjiioannou,T. P. Anal. Lett. 1989, 22, 1753–1763.

124. Sun, X.; Zhang, M.; Guo, S. Huaxi Yike DaxueXuebao. 1989, 20, 324–326.

125. Wang, S.; Lin, Q.; Zhou, J.; Feng, D.; Li, D.Jinan Daxue Xuebao. 1989, 1, 80–81.

126. Xian, Y.; Li, X.; Lian, T.; Zhu, X.; Wang, X.Gongye Weisheng Yu Zhiyebing. 1989, 15,226–229.

127. Wang, G.; Li, X.; Zhu, S. Fenxi Shiyanshi.1989, 8, 58–59.

128. Zhou, J.; Feng, D.; Li, Y.; Xie, Y. JinanDaxue Xuebao. 1989, 3, 52–56.

129. Jin, L.; Su, R.; Fang, Y. Fenxi Shiyanshi.1990, 9, 12–15.

141

130. Zhang, Y.; Han, J. Zhonghua LaodongWeisheng Zhiyebing Zazhi. 1990, 8, 350–351.

131. Wang, Y. Shanghai Huanjing Kexue. 1990,9, 31.

132. Yin, T. Turang Xuebao. 1991, 28, 108–114.

133. Nanos, C. G.; Karayannis, M. I. Fresenius J.Anal. Chem. 1991, 340, 253–257.

134. Coetzee, J. F.; Ecoff, J. M. Anal. Chem. 1991,63, 957–963.

135. Wang, D.; Liu, H. Fenxi Huaxe. 1991, 19,1104.

136. Feng, D.; Zhou, J.; Feng, J.; Chen, J. JinanDaxue Xuebao Ziran Kexue Yu Yixueban.1991, 12, 56–60.

137. Sun, Q.; Hou, S.; Chen, Z.; Zhu, J.; Liu, A.Fenxi Huaxue. 1991, 19, 1408–1410.

138. Zhang, G.; Wang, S.; Wang, H. Yejin Fenxi.1991, 11, 1–3.

139. Zhang, P. Fenji Ceshi Tongbao. 1990, 10,50–53.

140. Mannino, S.; Brambilla, M. Ital. J. Food Sci.1992, 4, 47–51.

141. Ostapczuk, P. Clin. Chem. 1992, 38, 1995–2001.

142. Qin, W.; Guo, G.; Zhang, M. Fenxi Huaxue.1992, 20, 1361.

143. Suturovic, Z.; Marjanovic, N. Zito Hleb. 1992,19, 215–219.

144. Marin, C.; Ostapczuk, P. Fresenius J. Anal.Chem. 1992, 343, 881–886.

145. Cai, Z.; Qi, D. Fenxi Huaxe. 1992, 20, 862.

146. Sun, B.; Yang, S. Huanjing Wuran YuFangzhi. 1992, 14, 30–32.

147. Zhou, J.; Mei, Q.; Zhang, S.; Yuan, X.Gongyye Weisheng Yu Zhiyebing. 1992, 18,173–176.

148. Jagner, D.; Sahlin, E.; Axelsson, B.; Ratana-Ohpas, R. Anal. Chim. Acta. 1993, 278, 237–242.

149. Yuan, Y.; Wang, Y.; Gao, C. Haiyang YuHuzhao. 1993, 24, 45–50.

150. Cladera, A.; Estela, J. M.; Cerdà, V. Talanta.1990, 37, 689–693.