ANALYST MAY 1986 VOL. 111 489 Electrochemical Approaches to Trace Element Speciation in Waters A Review T. Mark Florence CSIRO Division of Energy Chemistry Private Mail Bag 7 Sutherland N.S. W. 2232 Australia Summary of Contents 1. Introduction 2. Range of applicability of electrochemical speciation methods 2.1 La bi le/i nert d iscri m i nation 2.2 Redox state 2.3 Half-wave potential shifts 2.4 Limitations of electrochemical speciation techniques 2.5 Speciation schemes combining electrochemical and other techniques 2.6 Electrodeposition prior to carbon furnace atomic absorption spectrometry 3.1 The electrodeposition step 3.2 The ASV stripping step 3.3 Comparison of kinetics of dissociation of metal complexes at an electrode and a biomembrane 4. Electrodes for speciation measurements 4.1 Hanging mercury drop electrode 4.2 Thin mercury film electrode 4.3 Jet stream mercury film electrode 4.4 Flow-through cells 4.5 Streaming mercury electrode 4.6 Carbon fibre electrodes 4.7 Chemically modified electrodes 5.1 Polarography 5.2 Anodic stripping voltammetry 5.3 Cathodic stripping voltammetry 5.4 Potentiometric stripping analysis 5.5 Pseudo-polarography 5.6 Modulation waveforms 6.1 Collection and preservation of water samples for speciation measurements 6.2 Selected speciation results for some trace metals in waters 6.2.1 Copper 6.2.2 Lead 6.2.3 Cadmium 6.2.4 Zinc 6.2.5 Manganese 7.Correlation between ASV-labile measurements and toxicity 8. Complexing capacity 9. Conclusions and recommendations for future research 3.Theory of labilehnert discrimination 5. Electrochemical techniques for speciation 6. Some speciation results using electrochemical techniques 10. References Keywords Speciation; trace element analysis; water analysis; electrodes for speciation; electrochemical techniques for speciation 1. Introduction Speciation analysis of an element in a water sample may be defined as the determination of the concentrations of the different physico-chemical forms of the element which together make up its total concentration in the sample. The individual physico-chemical forms may include particulate matter and dissolved forms such as simple inorganic species, organic complexes and the element adsorbed on a variety of colloidal particles (Table 1).All these species can co-exist and may or may not be in thermodynamic equilibrium with one another.'-6 An ionic metal spike added to a filtered natural water sample may take times ranging from hours to months to equilibrate with the natural pool of metal species.1.7-9 For many heavy metals in sea water or river waters the predominant physico-chemical forms are unknown. Cu in sea water for example is believed to exist mainly as undefined, highly stable organic complexes the principal ligands perhaps being porphyrins siderophores or metallothioneins.5JOJ1 In this situation where neither the nature nor the concentrations of the dominant ligands are known it is obviously futile to attempt to apply computer modelling techniques to determine speciation. However for well defined experimental waters or for elements at higher concentrations (e.g.Ca and Mg in sea water) chemical modelling can be a powerful There are two main reasons for studying the speciation of elements in waters-to understand either the biological or the geochemical cycling of the elements.13 Biological cyclin 490 ANALYST MAY 1986 VOL. 111 Table 1. Possible physico-chemical forms of metals in natural waters Physico-chemical form Diameter] Possible example nm Particulate . . . . . . . . . . . . . . . . Retained by 0.45-pm filter >450 Inorganic complex . . . . . . . . . . . . . . CdCl+ PbC03 1 Simple hydrated metal ion . . . . . . . . . . . . Cd(H20)62+ 0.8 Organic complex . . . . . . . . . . . . . . Cu - fulvic acid 2-4 Adsorbed on inorganic colloids .. . . . . . . . . Pb2+/Fe203 10-500 Adsorbed on organic colloids . . . . . . . . . . Cu*+/humic acid 10400 Adsorbed on mixed organichorganic colloids . . Cu2+ - humic acid/Fe,03 10-500 includes bioaccumulation bioconcentration bioavailability and toxicity and geochemical cycling involves the transport, adsorption and precipitation of the element in the water system. It is now well established that no meaningful interpretation of either biological or geochemical cycling can be made without speciation information. 14-16 Each different physico-chemical form of an element (Table 1) has a different toxicity so analysis of a water sample for total metal concentration alone does not provide sufficient information to predict toxicity. For example two rivers may both contain 40 pg 1-1 of total dissolved Cu; if the first has most of the copper adsorbed on colloidal particles there will be little or no effect on aquatic life but if the second river has free Cu(I1) ion as the main species few organisms would survive.Lipid-soluble metal complexes are particularly toxic forms of heavy metals because they can diffuse rapidly through a biomembrane and carry both metal and ligand into the cell.15J7J8 Examples of lipid-soluble complexes are copper xanthates (from mineral flotation plants) copper 8-hydroxyquinolinate (agricultural fungicide) and alkylmercury compounds. 17718 Variation in the speciation of an element will also affect its degree of adsorption on suspended matter its rate of transfer to the sediment and its over-all transport in a water system.Speciation analysis will therefore assist in the prediction of the distance over which a river will be affected by effluent discharged from a point source.19 Speciation measurements have been made by a variety of techniques including electroanalysis ion exchange dialysis, ultrafiltration solvent extraction and computer modelling.’-3 If the measurements are made to study aquatic toxicity then the aim is to determine the toxic fraction of the element i.e., the fraction of its total concentration in the water sample that is toxic to aquatic organisms. For a metal complex this aim will be realised if the fraction of total metal that is reactive at a mercury electrode adsorbed by an ion-exchange resin or measured by some other technique is similar to the fraction that is dissociated at and transported across a biomem-brane.10~4-16 Electroanalysis is a powerful technique for the study of trace element speciation and has been applied to (or is potentially applicable to) about 30 elements Ag As Au Bi Br Cd C1, Co Cr Cu Eu Fe Ga Hg I In Mn Mo Ni Pb S Sb Se, Sn T1 U V W Yb and Zn.1>20>21 Four metals of prime environmental concern Cu Pb Cd and Zn can be deter-mined simultaneously and with great sensitivity. Moreover, the redox potential of an electrode can be varied accurately, precisely and continuously over a wide potential range and the study of the kinetics of metal complex dissociation at an electrode is supported by well established theory.22-28 Of all trace element speciation methods available at present, electroanalysis appears to provide the best opportunity for experimentally modelling the bioavailability of elements and their complexes with organic and inorganic ligands.Trace element speciation in natural waters requires special-ised techniques for the collection filtration storage and analysis of the samples because there is a constant risk of contamination or trace element losses when working with such low concentrations of analyte.193.29 A clean room or at least a laminar flow clean-air cupboard is essential for this type of work. Electrochemical techniques have an important advan-tage in that the sample requires much less handling and is in contact with fewer potential sources of contamination than when other speciation methods such as solvent extraction, dialysis or ultrafiltration are used.2. Range of Applicability of Electrochemical Speciation Methods Electrochemical techniques can be used to provide speciation information based on labilehnert discrimination redox state and half-wave potential measurements. The techniques are applicable to metals non-metals colloidal particles and organic compounds. 2.1. Labilehnert Discrimination The determination of labile (i.e. reactive) metal involves the measurement of the concentration of metal in the water sample that can be reduced at and deposited into a mercury electrode from a stirred solution. Labile metal is usually expressed as a percentage of total dissolved metal and the difference between total and labile metal is termed “inert” or “unreactive” metal.Some electrochemical parameters that affect the percentage of labile metal are deposition potential, electrode rotation (or stirring) rate mercury drop diameter, pulse frequency pH temperature and buffer composition. Under certain conditions labile metal has been found to correlate well with the toxic fraction of metal.10330 Labile metal consists of free metal ion and metal that can dissociate in the double layer from complexes or colloidal particles and hence be deposited in the mercury elec-trode.4,SJl For natural waters anodic stripping voltammetry (ASV) is the technique usually used and it has been applied to labilehnert measurements of Cu Pb Cd Zn Mn Cr T1 Sb and Bi.1 Heavy metal “pseudo-colloids,~7 i.e.colloidal par-ticles of Fe203 Mn02 humic acid etc. with adsorbed heavy metal ions,32 can be treated as a special type of metal complex, and may contribute significantly to some labile metal rneasure-ments. Labilehnert discrimination for some elements may also be made by chemical rather than electrochemical exchange. This approach is particularly advantageous for metals such as Fe which are difficult to determine by direct ASV and where concentrations are too low for polarography. In one pro-cedure,33 labile Fe was determined by treating the sample with bismuth - EDTA and the bismuth liberated by chemical exchange [reaction (l)] was measured with high sensitivity by ASV. Fe3+ + BiY- -+ Fey- + Bi3+ . . , Total Fe was then determined in the same way but after heating the acidified water sample to convert all iron into a reactive state.Another unusual measurement is the determi-nation of labile (or solvent-accessible) sulphur in proteins b ANALYST MAY 1986 VOL. 111 491 ~~ ~ ~ Table 2. Toxicity and electrochemical lability of some species in natural waters’.3.*0 Electrochemical Species Toxicity lability As(lI1) . . . . . . . . . . . . As(V) . . . . . . . . . . . . Cr(I.11) . . . . . . . . . . . . Cr(V1) . . . . . . . . . . . . n(1) . . . . . . . . . . . . Tl(II1) . . . . . . . . . . . . cu2+ . . . . . . . . . . . . CUCI . . . . . . . . . . . . c u c o 3 . . . . . . . . . . . . Cu2+ - fulvic acid . . . . . . . . Cu2+/humicacid-Fe20 . . . . C U ~ + - DMP* . . . . . . . . High Low Low High High Low High High Low Low Medium High * DMP = 2,9-dimethyl-1 ,lo-phenanthroline.High Low Low High High Low High High High Low Medium Low Fig. 1. Effect of HCl concentration on the ASV peak heights of Sb(II1) and Sb(V) the cathodic stripping voltammetric determination of sulphide ion liberated from protein disulphide bonds [reaction (2)].34 1 I I I HC-CH2-S-S- + OH- HC-CH2-S-0- + HS- (2) 2.2. Redox State Determination of the redox state of an element in solution is an important speciation measurement because it can drastic-ally affect toxicity adsorptive behaviour and metal transport (Table 2). Polarography and/or ASV have been used to distinguish between Fe(III)/(II),35 Cr(VI)/(III),36 TI(III)/ (I),25,37 Sn(IV)/(II),25>38 Mn(IV)/(II),39 Sb(V)/(III),25 As(V)/ (III),40 Se(VI)/(IV),41 V(V)/(IV),42 Eu(III)/(II),42 U(VI)/ (IV)43 and I(V)/( -I)? Whereas Cr(V1) is anionic (chromate) and highly toxic Cr(II1) is non-toxic and may exist as anionic or cationic hydrolysed or organic ~pecies.~5 For some other elements however including T1 As and Sb the lower valency state is the more toxic.46 The Mn(IV)/(II) di~crimination~~ measurement is important because fine Mn02 particles cause problems in water supply treatment plants by clogging filters.39 For several elements redox state speciation is actually a special case of labilehnert discrimination as one valency state is electrochemically active and the other inactive within the potential range of the electrode.Unreactive valency states of some elements under certain conditions are As(V) Cr(III), Mn(IV) Sb(V) Sn(1V) and Tl(II1).Determination of the labile valency state of the element in the presence of these unreactive forms can be made by a simple ASV or polaro-graphic measurement.25 Total metal can then be determined after chemical treatment of the sample (e.g. chemical reduction) and the concentration of the inert valency state determined by difference. Some metal ions that are electro-chemically inert because they are extensively hydrolysed in most media e.g. Sb(V) and Sn(IV) become labile when the sample is made strongly acidic.25 Fig. 1 shows the ASV behaviour of Sb(II1) and Sb(V) as a function of acidity. Sb(II1) can be determined in 0.2 M HCl and total antimony in 6-8 M HCl then Sb(V) by difference.Alternatively and preferably, total Sb can be measured after reduction of Sb(V) to Sb(II1) using hydrazine hydrochloride.25 Electrochemical methods for measuring valency state pro-duce redox numbers directly which is preferable to using ion-exchange methods to determine ionic charge. This latter technique may often lead to erroneous interpretations. A commonly used method for distinguishing Cr(V1) from Cr(II1) is to pass the sample through a column of anion-exchange resin. Cr(V1) as Cr042- is adsorbed whereas Cr(II1) is assumed to pass through the column as cationic species.36748 It has been suggested however that anionic Cr(OH)4- is the most common form of Cr in natural waters.49 In the polarographic method for Cr speciation,36 Cr(V1) and total Cr are determined sequentially in acetate buffer with half-wave potentials of -0.3 and -1.8 V vs.SCE respec-tively. 2.3. Half-wave Potential Shifts Shifts in the polarographic half-wave potential or ASV peak potential of metal ions in the presence of complexing agents can provide information about the thermodynamic stability of complexes in solution.42 However quantitative deductions from these shifts which have a sound theoretical basis for well defined experimental solutions containing one or at the most two ligands are inapplicable to natural or polluted waters, which may have many unknown ligands and several metals. Under these conditions quantitative interpretation of the shift is impossible although some qualitative deductions can sometimes be made.For example the ASV peak potential of Cu in sea water is about 0.2 V more negative than the peak in nitrate or acetate media. This shift reflects the relatively high stability of Cu(1) chloro complexes compared with those of Cu(II).50 In high-chloride media Cu is stripped from the electrode in a one-electron reaction to form Cu(1) chloro complexes whereas in nitrate solutioii Cu(I1) is produced in a two-electron stripping step. 2.4. Limitations of Electrochemical Speciation Techniques One of the main limitations of electrochemical speciation methods the inability to measure the concentrations of individual ionic species is common to most speciation techniques. Ion-selective electrode potentiometry (ISE) is the only method that can measure the activity of an individual ion, but the applicability of ISE to water analysis is severely limited by its poor sensitivity.Other electrochemical techniques such as polarography and ASV are dynamic systems that draw current through the solution and disturb ionic equilibria. It is not possible for example to use ASV to distinguish between labile cadmium species such as Cd2+ CdS04 CdC1- and CdC03 which may coexist in a river water sample. A single ASV peak is obtained for a mixture of these Cd species. However other speciation methods including ion-exchange chromatography solvent extraction dialysis and ultrafiltra-tion also disturb the natural ionic equilibria in a water sample during the separation process and all suffer from the same lack of specificity.’ Direct electrochemical speciation procedures are limited to measuring gross be havioural differences of groups of species (Table 3).This applies to the usual labilehnert discrimination 492 ANALYST MAY 1986 VOL. 111 Table 3. Electrochemical lability classification of metal species Degree of lability Examples10,22-24,71.80 Labile . . . . . . . . CdC03 CuC03 PbCI2, Cd - NTA Cu - glycine, Zn - cysteine Cu - citrate Zn - fulvic acid Cd - tannic acid Cu2+/humic - Fe203, Inert . . . . . . . . Pb - EDTA Pb2(0H)2C03, Quasi-labile . . . . . . PbC03 ZnCO, Cu - cysteine, CU - NTA Cu - tannic acid,* Cu - APDC,? Cu - fulvic acid,* Zn - tannic acid * Sea water pH 8.2. t APDC = ammonium pyrrolidinedithiocarbamate (ammonium tetramethylenedithiocarbamate) .and to the effect of deposition potential on ASV peak height.26?51 Other groups of species can be determined by ASV after chemical treatment of the sample (e.g. UV irradiation acidification) ,1-3,52 after physical separations (ion exchange ultrafiltration etc.)53?54 or after chemical exchange reactions.33 Although some deductions can be made about the nature of the species that are likely to occupy these behav-ioural “boxes,” exact conclusions cannot be drawn.55 As results from all these speciation procedures are operationally defined it is most important when publishing speciation methods to report all details of the analysis so that results from different laboratories can be compared. There has been considerable confusion in the literature over the ability of ASV to measure the “existing” or “natural” trace element speciation in a water sample.It is often specified5659 that buffer should not be added to a water sample before measurement of speciation by ASV so as to avoid disturbing the natural ionic equilibria. However because ASV is a dynamic technique it cannot possibly measure the “natural” speciation as the very act of measurement disturbs the equilibrium.3.60 If the aim of the determination is to estimate the bioavailable fraction of the metal some pH other than the natural pH of the water may well give the best correlation between ASV-labile metal and bioavailability (Section 7). The purpose of the measurement should always be kept in mind when designing a speciation procedure.For some purposes, however it may be desirable to avoid changing the pH of the sample.59 Sea water can be analysed without the addition of buffer but some fresh waters have too low an ionic strength, and are too poorly buffered to be analysed directly by ASV.24,61 It is possible to buffer the water and maintain its original pH by bubbling N2 - C02 mixtures of controlled composition through the sample.62363 This is inconvenient, and hydrogen carbonate buffers are poorly poised and contribute little to the sample’s conductivity. An important potential interference in ASV polarography and other electrochemical techniques is the adsorption of organic matter on the mercury electrode.3.58764 An adsorbed layer of organic matter may hinder the diffusion of metal ions, and thus diminish or eliminate the diffusion current and cause a non-linear relationship between stripping current and deposition time.65 Alternatively adsorption - desorption processes by organic dipoles on the mercury surface can yield “tensammetric” peaks when high frequency (a.c.or pulse) voltammetric techniques are used.66,67 These tensammetric adsorption waves have no faradaic component but in ASV are often mistaken for metal stripping peaks as in natural waters they may appear at potentials similar to those found for Cd Pb or 0 1 . 6 ~ They can be readily distinguished from metal peaks because they are absent if a simple d.c. scan is used their peak height is seldom proportional to deposition time their peak potential is very sensitive to pH and they disappear when the sample is UV irradiated.67 Fortunately tensammetric peaks are uncommon in water analysis but analytical chemists need to be aware of their existence.Interference by adsorbed organic matter may be a more frequent problem in ASV speciation analysis although it is often difficult to determine if a metal wave is diminished because of physical interference to diffusion by formation of an inert organo complex or by a combination of the two processes. For a particular sample the measured concentra-tion of ASV-labile metal can only be operationally defined by the instrumental and solution conditions used and in most instances little information can be deduced about the electrode processes involved. For standardising ASV-labile metal measurements it is important to use an ionic metal peak-height calibration graph, rather than to attempt to quantify the results by standard additions (“spiking”) of ionic metal to the sample.A metal spike may equilibrate only very slowly with the natural pool of physico-chemical species of the metal in the sample7 and even when equilibrated the spiking experiment would give total, rather than labile metal. The water blank used to construct the calibration graph should have an ionic composition similar to that of the sample. A special type of interference occurs in ASV as a result of intermetallic formation in the mercury electrode.1 These intermetallic compounds cause depression of the stripping peaks and shifts in peak potential. The most common interference is the depression of the zinc wave by an excess of Cu.In practical water analysis however this is rarely a problem because metal concentrations are low and Zn is usually present in excess of Cu.68 2.5. Speciation Schemes Combining Electrochemical and Other Techniques Speciation information obtained from direct electrochemical analysis (e.g. labilehnert discrimination) can be supplemen-ted by ASV or other measurements after various preliminary treatments of the sample. In this case electroanalysis is simply used as a highly sensitive method of analysis. Some important preliminary speciation steps are as follows. (a) UV irradiation to destroy organic matter.7@+70 If the sample is irradiated at natural pH only metal associated with organic matter will be liberated and the increase in labile metal compared with the unirradiated sample represents metal bound in inert organic complexes or to organic colloids.60~62~71~72 When the sample is acidified (0.02 M HN03) before irradiation all forms of metal, including inorganic colloids are converted into labile species and total metal is obtained.69.71 ( b ) Determination of lipid-soluble complexes.Lipid-soluble metal species are likely to be highly toxic.10J5J7J3 Extraction of a water sample with octan-1-01 or 20% butan-1-01 in hexane or passage of the water through a column of Bio-Rad SM2 resin will remove the lipid-soluble fraction of the metal.71 Analysis of the aqueous phase or column effluent by ASV and subtraction from total metal gives lipid-soluble metal. ( c ) Chelating resin separation.Metal that cannot be removed from a water sample by a column of Bio-Rad Chelex-100 chelating resin represents metal bound in highly stable or inert complexes or associated with colloidal particles.70,71,74~75 However the resin may remove some metal from colloidal particles.71 ( d ) Ultrafiltration and dialysis. These techniques separate species on the basis of molecular size and both can provide useful information about the size distribution of metal complexes and colloids,lJ although contamination can be a problem. In general the smaller a metal complex the higher is its biological activity. Several comprehensive speciation schemes combining ASV and these preliminary treatments have been pro-posed.1-3?7.20>76 The scheme used at present in this laboratory is shown in Table 4 ANALYST MAY 1986 VOL.111 493 Table 4. Speciation scheme for copper lead cadmium and zinc in waters Sample (unacidijied) : Filtrate analysis: Filter through a 0.45-pm membrane filter. Reject particulates and store filtrate unacidified at 4 "C. Aliquot No. Volume/ml Operation 1 20 Acidify to 0.05 M HN03 add 0.1% H202 and UV irradiate for 8 h then ASV* Add 0.025 M acetate buffer (pH 4.7) for fresh waters 3 20 UV irradiate with 0.1 YO H202 at natural pH then ASVT 4 20 Pass through small column of Chelex 100 resin. ASV on effluent$ 5 20 Extract with 5 ml of hexane - 20% butan-1-01. ASV on acidified, UV-irradiated aqueous phase§ 2 10 ASV at natural pH for sea water. Interpretation Total metal ASV-labile metal (3) - (2) = organically bound labile metal Very strongly bound metal (1) - ( 5 ) = lipid-soluble metal * Adjust to pH 4.7 with acetate buffer.t Not valid if [Fe] >lo0 pg 1-1. j Optional step. 0 Dissolved solvent in aqueous phase must be removed first. Reference ele Polythene screw !Ctl *ode Perspix cell Fig. 2. Flow-through cell for electrodeposition on an AAS graphite furnace tube. Reproduced with permission from Anal. Chem. 1980, 52 1570 Electrode Mo (Hg) ~ Diffusion I P layer M2+ + L2-t Ks ML Fig. 3. Diagrammatic representation of the reduction of a metal complex at a mercury electrode. The degree of dissociation of the metal complex ML at the electrode (and hence the lability of the complex) increases with increasing KB and increasing 6 2.6.Electrodeposition Prior to Carbon Furnace Atomic Absorption Spectrometry A novel application of electrochemical techniques to speci-ation studies is the controlled-potential electrodeposition of trace metals on to graphite furnace tubes which are then transferred to an atomic absorption spectrometer for electro-thermal measurement .36,77,78 The advantage of this technique is that elements such as Cr Ni and Co which are difficult to determine by direct ASV can be concentrated from solution by electrolysis using labilehnert discrimination and deter-mined with good precision. The furnace atomisation simply replaces the ASV stripping step. The flow-through electrolysis cell used by Batley and Matousek78 is shown in Fig.2; it was successfully applied to the discrimination of Cr(V1) and Cr(II1) and the labilehnert forms of Ni and Co in natural waters; Another type of flow-through cell using a graphite furnace tube was described for the in situ determination of lead and cadmium in sea water.79 3. Theory of Labilehnert Discrimination 3.1. The Electrodeposition Step The dissociation of a 1 1 complex formed between a divalent metal ion M and a ligand L and the subsequent reduction of M2+ at a mercury electrode may be represented by the following equilibria: M L 2 M2++L2- . . . . . (3) 01 = [ML]/[M][L] = kf/kd . . . . (4) M2+ + 2e-+MO(Hg) . . . . . (5) These reactions are shown diagramatically in Fig. 3. When the complex ML is not itself directly reducible the electrolysis (faradaic) current is due solely to the reduction of M2+ ions dissociated from ML [reactions (3) and ( 5 ) ] .This process leads to a kinetically controlled current and i& the ratio of the kinetic current i k 'to the diffusion current i d is an index of the lability of the complex. The diffusion current is the current observed for the same concentration of metal ion but in the absence of ligand. In the absence of kinetic control ik/id = 1. Turner and Whitfield22 calculated that for ASV at a thin mercury film electrode (TMFE) , ik/id = (1 + q - 1 tanhq)-l . . . . (6) and at a hanging mercury drop electrode (HMDE 494 ANALYST MAY 1986 VOL. 111 where u = kf[L]/k& 7 = 6 D-l(kd + kf[L])t; 6 = diffusion layer thickness (cm); D = diffusion coefficient of the metal ion (cm2 s-1); and ro = radius of HMDE (cm).When iklid >0.99 (i.e. a highly labile complex) it can be shown thatgo Turner and Whitfieldgo suggested the criteria shown in Table 5 for the definition of labile quasi(or partially)-labile and inert (non-labile) complexes in ASV analysis at a rotating disc TMFE. The calculations assume that the ligand L is in large excess. Davison23 calculated the following criterion for a labile complex using ASV at a rotating disc TMFE assuming a diffusion coefficient of 1 X 10-5 cm2 s-1 for the metal ion: labile (iklid > 0.90): For both ASV and polarography the lability of a complex depends not only on its dissociation kinetics but also on the effective measurement time which with the constant electroly-sis time of ASV depends on the time the complex molecule is resident in the diffusion (or reaction) layer and this resident time depends in turn on 6 the thickness of the diffusion layer.The larger the value of 6 the longer is the residence time of the complex in the diffusion layer the greater is the opportunity for dissociation and deposition of metal in the electrode and hence the higher is the fraction of labile The thickness of the diffusion layer is governed principally by rotation rate for a rotating disc electrode (RDE) and by the rate of solution stirring for a HMDE. The Levich equation can be used to calculate 6 (cm) at an RDE23: 6 = 1.62 D ' 3 03-4 ~ ~ ' 6 . . . . . . (10) where D is the diffusion coefficient (cm* s-I) o is the electrode rotation rate (rad s-1) and Y is the kinematic viscosity of the electrolyte (Stokes).For rotation speeds in the range l O x l O 4 rev min-1 values for 6 of 5 x 10-3-5 X cm are obtained. Kinetic control in ASV can be studied by measuring the effect of 03 on ik. A constant value of i k o is obtained in the absence of kinetic control but decreases with increasing o if kinetic effects are significant. Calculation of 6 at an HMDE using a magnetic stirring bar for solution stirring is difficult because of the ill-defined hydrodynamic conditions. However 8 can be determined experimentally for an HMDE by measuring the d.c. diffusion current id in the stirred solution23: id=nFADC/6 . . . . . . (11) where n is the number of electrons involved in the electrode reaction F is the faraday A is the electrode area and C is the concentration of electroreducible species.A typical value for 6 at an HMDE is 2 x 10-3 cm. The diffusion layer thickness at Table 5. Electrochemical lability criteria for lead complexes80 Lead complexes Description Lability criterion* concerned Labile . . . . ik/id>0.99 PbCI+,PbS04 Quasi-labile . . ik/id < 0.99 PbC03 PbOH+ Inert . . . . . . i k / i d = ( l + u ) - l Pb-EDTA log(S1[Ll*) < 2 ik/id > (1 + O)-' Pb - humic acid * = stability constant for 1 1 complex; [L] = concentration of ligand; ik = kinetic current; id = diffusion current; u = P,[L]. a dropping (or static) mercury electrode can be estimated from81 where te is the effective measurement time i.e. the drop time for maximum current d.c.polarography or for pulse tech-niques the duration of the applied pulse before the current is sampled plus the mean of the sampling interval.23 For example if the current is sampled for 20 ms 40 ms into the life of the pulse the average measurement time is 50 ms. With a.c. modulation te is the inverse of the frequency (Hz). The reaction layer thickness p which may be less than the diffusion layer thickness is given from reaction layer theory as26 p.= (Dkf-y . . . . . . (13) Van Leeuwen26 calculated from reaction layer theory assum-ing a large excess of ligand that the following conditions apply for polarographic lability measurements: labile complex kdk,-JteJ>> 1 . . . . . . (14) inert complex kdkf-jtet<< 1 . . . . . . (16) quasi-labile complex kdkf-'tet = 1 .. . . . . (15) For polarographic conditions it can also be shown that23 It is apparent that ASV lability determined during the deposition step depends solely on the kinetic parameters of the metal complex dissociation the concentration of excess ligand and the diffusion layer thickness which in turn is a function of the rate of stirring of the solution or the rotation speed of the electrode. In polarography the diffusion (or reaction) layer thickness is governed by the drop time or in pulse techniques by the pulse width and current sampling times. The value of &lid is not affected by factors such as deposition time sample volume or cell volume. Deposition time may only affect iklid in the special case where an adsorbed substance interferes in a non-linear manner with the rate of electrodeposition.An implicit assumption in the preceding discussions is that the metal complex ML is not directly reducible. However, where electrons are added directly to the complex without its initial dissociation in the diffusion layer direct electrochem-ical reduction of some complexes is known to occur especially at very negative potentials.5,51,71,82 The presence of such complexes in a sample can be detected from the effect of ASV deposition potential Ed on peak current; the peak current will increase continuously with increasing Ed (Fig. 4) instead of increasing from zero to a limiting value over a small range of Ed (Section 5.5). To minimise the chance of directly reducible complexes contributing to the ASV-labile measurement the deposition potential should be just sufficiently negative to yield the maximum peak current for the free metal ion in that medium i.e.just on the plateau of the relevant pseudo-polarogram (Section 5.5). For this reason it is preferable in speciation analysis to determine each element separately, using the minimum deposition potential rather than e.g., measuring Cu Pb Cd and Zn simultaneously with a deposi-tion potential of -1.3 V vs. SCE. Reducible metal ions adsorbed on colloidal particles of humic acid hydrated iron oxide etc. can be treated as a special type of metal complex in the preceding theoretical discussions and metal ions dissociated from pseudo-colloids at the solutionldiffusion layer boundary may contribute to kinetic currents.1770371 Although the involvement of pseudo-colloids in metal deposition has been questioned,59 there seems no reason why ions could not dissociate from thes ANALYST MAY 1986 VOL. 111 495 1 Cell interior I Membrane Cell exterior Deposition potentialN vs. Ag - AgCl Fig. 4. Pseudo-polarogram of copper in sewage plant effluent water particles at the diffusion layer boundary under the influence of the potential gradient. Metal ions are known to dissociate from colloidal particles as a result of the concentration gradient across a dialysis or ultrafiltration membrane,34,83 and metal bonding to the particle is unlikely to be stronger than that involved in some ASV-labile molecular complexes such as Cu - NTA.10>71 Pseudo-colloids however cannot act as directly reducible metal complexes because the diffusion coefficients of colloidal particles are so small (10-7 cm2 s-1) that they would not contribute significantly to the peak current .5 3 3.2. The ASV Stripping Step The preceding discussion has considered only the effect of deposition parameters on the amount of deposited metal and on kinetic currents in ASV speciation analysis. Ideally the relative heights of the stripping peaks for labile and total metal in the sample and hence the calculated percentage of labile metal should be controlled solely by the preliminary elec-trodeposition step i.e. by the amount of metal deposited in the electrode. However under certain circumstances the kinetics of the stripping process (electrooxidation) especially when pulse techniques are used may have a significant even dominant effect on the stripping peak height.This can occur if a complexing agent present in the sample solution but not in the standard affects the stripping chemistry or kinetics.22.27 This situation could arise from a number of causes. If a ligand in the sample solution stabilises an intermediate valency state of the metal leading to a smaller number of electrons being involved in the electrochemical oxidation lower stripping peaks will result: Standard CuO-Cu2+ + 2 e - . . . . . . (18) Sample CuO + 2 C1- -+ CuC12- + e- . . . . . . (19) This has been observed for the ASV determination of Cu in the presence of chloride and some other ligands.50>64J4 The presence of complexing agents or surface-active sub-stances may also affect the kinetics of stripping and lead to a change in peak height especially when pulse techniques are used.85 Buffleu showed how the large surface excess of the oxidised metal ion (compared with the bulk solution) present during the initial stages of stripping can cause precipitation and other chemical reactions at the electrode surface that might affect the stripping peak current.Perhaps the most useful method for ensuring that the ASV-labile measurement is controlled only by the deposition step is to use medium exchange where the sample solution, 1. Facilitated diffusion ML- L2- + M*+. . . l@T@l i- Mz+ 2. Lipid solubility ML Fig. 5. complexes through a biomembrane Diagrammatic representation of the transport of metal after electrodeposition is replaced with a new supporting electrolyte in which stripping is carried 0~t.86,87 The new electrolyte would be chosen to yield reversible reproducible stripping peaks for the element under study.3.3. Comparison of Kinetics of Dissociation of Metal Complexes at an Electrode and a Biomembrane For the study of aquatic toxicity by metals the electrochemical and solution parameters should be chosen so that the ASV-labile fraction of total dissolved metal is similar to the toxic fraction. 10$8 Hydrophilic heavy metal ions are believed to be transported across the hydrophobic space of a biomem-brane by the “shuttle” process of facilitated diffusion (or “host-mediated transport”) where a receptor molecule (e.g., a protein) on the outer membrane surface binds a metal ion.89.90 The hydrophobic metal - receptor complex then diffuses to the interior of the membrane and releases the metal ion into the cytosol where it is trapped perhaps by reaction with a thiol compound.91 The receptor then diffuses back to the outer surface of the membrane ready to collect another metal ion (Fig.5).1OJ617 Alternatively if the metal complex is lipid soluble the much more rapid process of direct diffusion can take place (Fig. 5). Direct diffusion is basically different from facilitated diffusion not only because it is faster but because the ligand is also transported into the cytosol.10 The fraction of total metal in solution that can be transported across a membrane surface is equivalent to the bioavailable or toxic fraction.This in turn depends on the relative affinity of the metal for solution ligands and the receptor molecule (Fig. 5) or on the solution - membrane partition coefficient for a lipid-soluble complex. The process of metal accumulation in an organism by dissociation of a metal complex at a membrane surface, facilitated diffusion of the metal through the membrane and deposition in the cytosol (Fig. 5) has obvious similarities to the process of ASV electrodeposition (Fig. 3) where the metal complex dissociates at the diffusion layer boundary and the metal ion travels through the diffusion layer to the electrode where the metal is deposited. 4. Electrodes for Speciation Measurements The electrodes used most often for routine speciation measurements in natural waters are the hanging mercury drop electrode (HMDE) the thin mercury film (on glassy carbon) electrode (TMFE) and the dropping (or static) mercury electrode (DME).Other electrode systems have been used mainly in research or for special applications. Ion-selective electrodes will not be reviewed here because they are usually insufficiently sensitive for speciation analysis in natural waters although they may have application to polluted waters 496 ANALYST MAY 1986 VOL. 111 4.1. Hanging Mercury Drop Electrode (HMDE) The HMDE is widely used in ASV and speciation analysis. Use of simple d.c. techniques leads to stripping peaks with a drawn-out shape owing to the slow diffusion of metal from the interior to the surface of the mercury drop.For this reason it is necessary to use high-frequency (pulse or a.c.) waveforms for ASV at the HMDE.21 The high-frequency techniques respond only to dissolved metal at the surface of the mercury drop and so the stripping peaks have a sharp theoretical shape. With a 15-min deposition time the limit of detection for Pb using differential pulse ASV (DPASV) at the HMDE was found to be 5 x 10-11 M based on noise levels in the Princeton Applied Research (PAR) Model 174 voltam-meter.25 However reagent blanks usually increase this limit substantially. The use of a static mercury drop electrode (SMDE Section 5.1) instead of the older micrometer screw-type hanging mercury electrode system greatly improves the reproducibil-ity reliability and simplicity of ASV at an HMDE because the mercury drop is formed automatically and its size is very reproducible.81 4.2.Thin Mercury Film Electrode (TMFE) While the use of a HMDE or a SMDE may offer better reproducibility than a TMFE especially for Zn,25 the rotated TMFE is much more sensitive (Table 6). This higher sensitivity is essential for high-purity samples such as open-ocean sea water which cannot be analysed at the HMDE.92793 Whereas pulse techniques are essential for an HMDE ASV at a TMFE can conveniently be carried out using a simple d.c. scan because the mercury film is so thin that dissolved metal is stripped from the film very quickly. Use of differential pulse modulation at a TMFE decreases the limit of detection by a factor of about five over a d.c.scan.94 The glassy carbon TMFE can be rotated (1000-3000 rev min-1) or the solution stirred although rotation gives more precise results.5 The substrate used for a TMFE is nearly always glassy carbon polished to a mirror finish with diamond or alumina dust.33395 Glassy carbon is a commercially available synthetic substance almost as hard as diamond but with good electrical conductivity and a wide potential range.94396 Like all forms of carbon however glassy carbon is rapidly attacked by free halogens,97 so the electrode should never be polarised in the positive region when the solution contains halide ions. In chloride media a film of mercury(1) chloride (calomel) forms on a TMFE if the electrode potential E is more positive than E = +0.026-0.0296 log[C1-]2 V vs.SCE . . (20) Table 6. Relative sensitivity of some electrochemical techniques Limit of detection for Electrochemical technique* . . lead/~25,81,135 D.c. polarography (DME) . . . . . . . . 2 X 10-6 D.c. polarography (SMDE) . . . . . . . . 1 X D.p. polarography (DME) . . . . . . . . 8 X 10-8 D.p. polarography (SMDE) . . . . . . . . 1 X D.p. anodic stripping voltammetry (HMDE) . . 2 X 10-lo S.W. anodicstripping voltammetry (HMDE) . . 1 X 10-lo D.c. anodic stripping voltammetry (TMFE) . . 5 X lo-" D.p. anodicstripping voltammetry (TMFE] . . 1 X lo-" S.W. anodic stripping voltammetry (TMFE) . . 5 X lo-'* * D.c. = direct current; D.p. = differential pulse; S.W. = square wave; DME = dropping mercury electrode; SMDE = static mercury drop electrode; HMDE = hanging mercury drop electrode; TMFE = thin mercury film electrode.This film of calomel seriously degrades the performance of the electrode and is difficult to remove (ethanol is the best solvent) .94 Because even low chloride concentrations lead to calomel formation in general a TMFE should not be polarised at potentials more positive than 0 V vs. SCE.94 When first prepared a glassy carbon electrode should be polished metallographically (diamond dust) and thereafter should only require polishing with wet and dry filter-paper (e.g. Whatman No. 541) after each analysis. If the electrode becomes contaminated with organic matter or metal hydroxides wiping with filter-paper soaked in ethanol or 2 M HN03 respectively, will usually restore the surface.The mercury film should be removed by wiping with filter-paper and not by anodic polarisation,9* as this will degrade the surface if chloride is present .94 The mercury film may be electrodeposited on the glassy carbon substrate by two methods-pre-formed or in-situ deposition. The pre-formed method consists of electrodepos-iting a film of mercury from a stirred mercury(I1) nitrate solution (1 x 10-4 M pH 3-5 -0.6 V vs. SCE for 5-10 min). The plated electrode is then washed briefly and immediately used for analysis of the deaerated sample. A fresh film must be deposited for each sample. The in situ technique simply involves adding an aliquot of 1 x 10-2 M Hg(N03)2 [kept in a dark bottle at pH 3 to prevent autoreduction to Hg(I)] to each sample to give a final concentration of 2 x 10-5-4 x 10-5 M Hg2+.During the deposition step of ASV trace metals and Hg(I1) are reduced simultaneously and codeposited forming a very thin film of a dilute amalgam on the electrode. Measurements are usually made on the second or third deposition - stripping cycle as the first deposition is needed to condition the electrode.95 The mercury film thickness 1 (cm), may be calculated from25 1 = 2.43 x 10-l1 itlr2 . . . . (21) where i is the limiting mercury(I1) ion deposition current (PA) t is the deposition time ( s ) and r is the radius of the electrode surface (cm). Typical mercury film thicknesses used in the in situ technique are 5 x 10-6-10 x 10-6 cm. The in situ 3 cm 4 1 I I 3 I c--Fig. 6. Croat. Chem.Acta 1977 49 L1 Jet stream electrode. Reproduced with permission fro ANALYST MAY 1986 VOL. 111 497 deposition method is much simpler than pre-forming a new film for each sample and avoids the danger of oxidation of the pre-formed film before it can be transferred to the deaerated sample. Oxidised films give erratic results especially for Cu. It has often been claimed5658 that the in situ mercury film cannot be used for speciation studies because the addition of mercury(I1) ions to the sample will change the "natural" speciation and cause an increase in labile metal as a result of Hg2+ exchanging with a metal complex ML and liberating free metal ion M2+ (Section 2.4): ML + Hg2+ + HgL + M2+ . . . . (22) Certainly mercury(I1) forms very stable complexes with many ligands and the exchange reaction [reaction (22)] may readily occur with labile metal complexes in natural waters.However it is unlikely that this exchange reaction would significantly affect many ASV speciation results.60 If the complex ML is sufficiently labile to undergo significant chemical exchange with Hg2+ during the period of the analysis (10-20 min) then it may also dissociate at the electrode surface and yield labile metal. If this occurred the addition of Hg2+ would have no effect on the measured concentration of ASV-labile metal. Recent research using natural waters and synthetic waters containing various ligands showed that Hg2+ rarely has any effect on the ASV determination of labile Cu, Pb Cd and Zn if only natural ligands are present.99 Stewart and Smartloo showed that a glassy carbon TMFE covered with a dialysis membrane gave excellent results for the ASV determination of Cd.Wang and HutchinslOl used a cellulose acetate film to cover a glassy carbon electrode and found that electrode fouling by protein adsorption was greatly minimised. It would be most interesting to apply these covered electrodes to the determination of ionic metals in the presence of large organometallic complexes. A special application of a glassy carbon TMFE is its use in a micro-cell using a 13 mm diameter membrane disc to adsorb the sample (15 p1).102 The membrane disc with absorbed sample (containing Hg2+) is dropped into the cavity of a Perspex block. The base of the cavity has flush-fitting platinum and silver discs acting as auxiliary and .reference electrodes respectively.The glassy carbon working electrode is mounted in a Teflon rod which is made a sliding fit in the Perspex block. When the cell is screwed together the membrane disc is compressed between the glassy carbon electrode and the other two electrodes. An O-ring seals the cell. Oxygen is removed by applying a potential of - 1.4 V vs. Ag - AgCl for 20 s. Conventional ASV-labile measurements can then be made. Discs of an ultrafiltration membrane or Chelex-100 paper can be placed between the glassy carbon electrode and the sample disc to provide additional speciation measurements.102 A similar filter-paper ASV cell using a mercury pool electrode has also been described. 103 4.3. Jet Stream Mercury Film Electrode A new method for transporting sample to the surface of a glassy carbon TMFE was described by Magjer and Branica.104 In this technique instead of rotating the electrode or stirring the solution a flat disc having a conically shaped hole is positioned below the glassy carbon electrode (Fig.6 ) . The disc is then vibrated at high frequency in a vertical plane forcing solution on to the electrode surface in a jet stream. The sensitivity of the electrode is critically dependent on the geometries of the vibrating disc and the conical hole but if these parameters are optimised higher sensitivity than rotation or stirring can be achieved.105 4.4. Flow-through Cells A variety of flow-through cells designed for on-line stripping analysis have been described.'O6-110 These cells often have dual mercury-plated glassy carbon or reticulated vitreous carbon108 electrodes with independent potential control for removal of dissolved oxygen or interfering elements at the upstream electrode.110 Labilehnert speciation measurements are possible with these electrodes although they have seldom been used for this purpose.Flow-through cells used for high-performance liquid chromatography could also be used for speciation measurements in a closed-loop system.78 The wall-jet electrode in which a jet of the sample impinges on the working electrode,lll should provide excellent sensitivity and, because of its rapid hydrodynamic characteristics could yield data on the dissociation kinetics of metal complexes. 4.5. Streaming Mercury Electrode The streaming mercury electrode (SME) first used by Heyrovsky and Kuta42 for oscillographic polarography , involves forcing a thin stream of mercury through a short path of the test solution using a mercury reservoir to provide the necessary pressure.The electrode thus consists of a short, rapidly changing cylinder of mercury. A modified SME that uses less mercury and gives more reproducible results has been described by Florence and Farrar. 112 A unique characteristic of the SME is that the electrode is being constantly and very rapidly renewed so that only fast electrochemical reactions are registered and most important , substances that adsorb on mercury have little or no effect.ll2 Although the use of the DME with high-frequency techniques can also discriminate against slow electrode reactions organic matter (e.g.humic acid) adsorbed on the electrode can seriously affect the results. However with the SME the electrode is being renewed so rapidly that adsorption has little chance to occur and so has a negligible effect on electrode kinetics.112 The SME has not yet been applied to speciation measurements but it may prove especially useful in conjunc-tion with differential pulse or square-wave modulation for measuring free metal ion in the presence of metal complexes and surface-active substances. 4.6. Carbon Fibre Electrodes Electrodes consisting of minute carbon fibres (5-10 pm diameter 0.1-0.3 cm length) either bare or mercury coated, are finding increasing use in electrochemistry.The electrodes exhibit low background current and because of the extremely low cell current the iR drop in the solution is negligible.113-115 These low cell currents provide an analysis that is essentially non-destructive so in vivo analysis e.g. in the brain can be made without damage to the animal.116 Analysis can be carried out in the absence of supporting electrolyte and in aprotic organic solvents.~~4 A two-electrode system may be used thus avoiding the need for a potentiostat which is expensive and is a major source of electronic noise.114 Carbon fibre electrodes have considerable potential for speciation analysis in vivo.117 4.7. Chemically Modified Electrodes Electrodes that have been coated with a chemical that alters their characteristics are now widely used in electrochemistry, and some systems have been applied to electroanalysis.118 The substrate may be platinum,119 carbon pastel20 or glassy carbon.121 Polymers such as poly(viny1bipyridine) and vinyl-ferrocene can be electrodeposited on a platinum electrode or groups such as -Si(CH2)3NHCOCOOH directly bonded to oxide groups on glassy carbon.121 The use of surface-active metal complexes in cathodic stripping voltammetry has produced extremely sensitive methods for some metals. One of the early applications of this technique was to the determination of total and reactive A1 in waters using linear scan voltammetry and the di-o-hydroxyaz 498 ANALYST MAY 1986 VOL. 111 dye Solochrome Violet R5 .I22 Very sharp peak-shaped voltammograms were obtained with a limit of detection of 0.2 yg 1-1 of A1 as a result of adsorption of the aluminium - dye complex on the mercury drop.123 More recent applications have involved the use of adsorbed films of dimethylglyoxime for Ni and Cop4 ammonium tetramethylene dithiocarbamate for Zn,125 catechol for Cu Fe U and V126130 and 8-hydroxy-quinoline for M o p all using cathodic stripping voltammetry at an HMDE.Metal concentrations as low as 10-10 M can be determined with a short deposition time,5 and labile and inert metal species in a water sample can be determined on the basis of their reactivity with the organic ligand.132 5. Electrochemical Techniques for Speciation 5.1. Polarography In natural waters even using differential pulse or square-wave modulation polarography is not sufficiently sensitive for speciation measurements of most elements.Many of the elements of interest (e.g. the toxic elements) are present in the range 10-10-10-8 M whereas polarography is limited to concentrations above 10-7 M. For iodine speciation in sea water however polarography is an ideal technique for determining iodate to iodine ratios.44 Iodate is usually present in sea water at concentrations of about 3 x 10-7 M and, because its reduction involves six electrons a large and sharp polarographic peak is produced. Polarography may also find application for speciation studies of polluted waters that have much higher metal concentrations. The technique is especially useful for valency state discrimination (Section 2.2). The development of the static mercury drop electrode (SMDE) has greatly simplified and improved polarographic analysis (Section 4.1).133 Whereas the conventional (Hey-rovsky) dropping mercury electrode produces a gravity-fed mercury drop of continuously changing area the SMDE has a constant area when the current - voltage curve is recorded, thus essentially eliminating charging current due to drop growth.81 This advantage of the SMDE is achieved by using a wide-bore capillary through which the mercury flow is controlled by a valve that can be opened for variable times. This allows drops of different size to form very quickly. The voltage scan is applied after the valve has been closed and the drop is stationary. After completion of the scan the drop is mechanically detached.Because the charging current is so small with an SMDE the advantages of pulse techniques over a simple d.c. scan are only marginal (Table (9.81 The Metrohm SMDE has outstanding performance and uses inexpensive, wide-bore capillary tubing for the electrode. Bond et aZ.134 designed an efficient high-capacity flow-through cell for use with the EG and G PAR Model 303 SMDE. 5.2. Anodic Stripping Voltammetry Anodic stripping voltammetry is the most widely applicable electrochemical technique for trace element speciation in waters.21 Because of the “built-in” concentration step in ASV, extremely high sensitivity can be obtained. At present the most sensitive commercially available ASV technique is square-wave stripping at a glassy carbon TMFE (Table 6).135 In an unmodified (i.e.no pre-treatment) sample such as open-ocean sea water metal concentrations below lo-” M can be determined although for many analyses the limit of detection is set by the blank and not by the intrinsic sensitivity of the technique.25 Differential pulse voltammetry is a factor of two or three less sensitive than the square-wave method,135 and an HMDE is 5-10 times less sensitive than a TMFE. In most instances ASV calibration for labile metal is best carried out by the use of measurements on separate standard solutions rather than by making standard additions (“spik-ing”) to the test solution. In many natural waters excess of organic matter (e.g. fulvic acid) in the sample will complex the metal spike and the increase in peak height will be related to total rather than labile metal.If the concentration of the spike is high i.e. at least 20 times that of the complexing agents in the sample then the peak-height difference between the first and second spikes can be used to calculate labile metal. These high spikes however may lead to metal contamination of the cell. It is more accurate to use a matched matrix with standard metal concentrations similar to that of the sample. Dissolved oxygen is a serious interferent in ASV and care must be taken to remove it completely. Ideally the ASV cell should be under a positive pressure of oxygen-free nitrogen but if this is not possible the cell should be sealed as well as possible and a rapid flow of inert gas maintained at all times. If the gas flow is too vigorous however solution spray in the cell may cause memory effects.Mechanically detached DMEs pose a special problem because the cell must have a slot to allow for movement of the electrode. It is better to use a high-quality grade of oxygen-free nitrogen than to complicate the system (with the possibility of air leaks through the tubing connectors) by installing an oxygen scrubbing system. Oxygen contamination is much more likely to originate from air ingress into the cell or through tubing than from impurity in the sparging gas. Dissolved oxygen can cause an apparent increase in the Cu and Pb stripping peaks79J03 and in unbuffered solutions a decrease in the Cd peak as a result of the consumption of hydrogen ion at the electrode surface: . . (23) In many supporting electrolytes oxygen contamination is manifested by a broadening of the copper stripping peaks.O2 + 4 H+ + 4 e- -+ 2 H20 . . 5.3. Cathodic Stripping Voltammetry Cathodic stripping voltammetry (CSV) involves the cathodic stripping of an insoluble film of the mercury salt of the analyte (H2L) deposited on the working electrode: deposition Hg + L2- -+ HgL + 2 e- (24) . . stripping HgL + 2 e- + Hg + L2- . CSV has not yet found a great deal of application in trace element speciation. As(II1) and Se(1V) can be determined in the presence of their higher valency states,1,3 sulphide can be measured in a large excess of other inorganic or organic sulphur compounds34J36 and the recently developed adsorp-tion - CSV technique can be used to determine free metal ion plus labile complexes for Ni Co Cu Zn Fe V U and Mo (Section 4.7).5.4. Potentiometric Stripping Analysis Potentiometric stripping analysis (PSA) largely developed by Jagnerl37 in Sweden uses the same initial step as ASV i.e., metal is deposited into a TMFE at a controlled potential. However instead of applying a voltage ramp to oxidise and strip the metal a chemical oxidant (0) in solution is allowed to diffuse to the electrode to oxidise the deposited metal and the potential of the working electrode is followed as a function of timeW deposition M2+ + 2 e- -+ MO(Hg) . . (26) oxidation MO(Hg) + 0 -+ M*+ + R2- . . (27) Well separated potential - time steps are obtained as the metals are successively oxidised by oxidants such as dissolved oxygen or mercury(I1) ion.PSA is much less affected by adsorbed organics than is ASV,137,139 and redox compounds do not interfere with the analysis ANALYST MAY 1986 VOL. 111 499 As(II1) has been determined in the presence of As(V) by PSA,140 but there has been little other interest in applying the technique to speciation analysis. 5.5. Pseudo-polarography A pseudo-polarogram is a plot of ASV stripping peak current versus deposition potential. The half-wave potential (E,) of a pseudo-polarogram of a metal is related to (but not identical with) the polarographic half-wave potential (Eh). The value of E4 - E becomes increasingly positive as the rate constant of the electrochemical reaction increases. 141 Pseudo-polarograms may have a classical polarographic shape or the peak height may increase continuously with deposition potential (Fig.4). This latter behaviour implies that metal complexes are present that are directly reduced, i.e. they diffuse intact to the electrode surface without first dissociating in the diffusion layer to metal ion and ligand.71 Brown and Kowalskil41 demonstrated the application of pseudo-polarography to a study of the speciation of As Cd and Pb in various natural waters. Valental42 used pseudo-polarography to identify Pb carbonato complexes in sea water, while Bubic and Branica143 used the same technique to study the ionic state of Cd in seawater. 5.6. Modulation Waveforms Modulating the d.c. voltage ramp with various waveforms provides increased sensitivity in ASV especially when a mercury drop electrode is used.135 At present the most commonly used modulation waveforms in stripping analysis are differential pulse and square wave.However a.c. and staircase waveforms have also been used.1441145 The use of microcomputers21J46J47 in electrochemical instrumentation allows a wide range of waveforms to be applied to the cell to optimise the analysis in terms of sensitivity and selectivity for a particular sample type. Square-wave and staircase voltam-metry have the advantage over the differential pulse technique that much faster scan rates can be used,145 up to 2 V s-1 with a square-wave frequency of 200 Hz so that a complete voltammogram can be obtained in less than 1 s and on a single drop in polarography .135,148 Differential pulse voltammetry cannot utilise scan rates in excess of 5 mV s-1 so that scanning from the Zn to the Cu ASV peaks takes at least 4 min.A disadvantage of pulse and square-wave techniques is that they are more affected than linear scan voltammetry by substances that adsorb on the mercury electrode.25 Adsorbed layers interfere seriously in differential pulse ASV because of the multiple redox reactions that occur at the electrode during deposition.149 It must be appreciated that different modu-lation waveforms will give different results in labilelinert ASV determinations. 6. Some Speciation Results Using Electrochemical Techniques 6.1. Collection and Preservation of Water Samples for Speci-ation Measurements Detailed instructions have been given for the contamination-free collection of water samples for trace element speciation analysis.1~299150 In general samples should be collected in linear polyethylene bottles which are initially acid cleaned, then reserved for collecting the same type of water. Special procedures are required for some elements such as mercury and iodine.1 The collected water sample cannot of course be preserved by adding acid as this alters the element speciation. Freezing of water samples is also prohibited for trace heavy metal speciation because concentration of the solutes during the freezing process may cause hydrolysis of metal ions and other reactions that are irreversible or only slowly reversible, on thawing.3 The safest preservation procedure is to filter the sample immediately after collection and store the filtered sample at 4 "C.The concentrations of Cu Pb Cd and Zn in both fresh water and sea water samples remained unchanged for several months under these conditions.3~70~151 Reports of serious adsorption of these metals on to polyethylene con-tainers can be traced to the use of ionic spikes (either stable or radioactive) in the water samples to measure such losses.3 Whereas ionic metal rapidly partitions to the walls of the plastic container the naturally present metal in pristine water samples very little of which is in the ionic form has a low affinity for both polyethylene and glass.3.70 In polluted waters, however ionic metal may persist close to the source of pollution or when the complexing capacity of the water is exceeded.In such instances losses may occur on storage. On the other hand storage of ultra-pure waters in polyethylene containers may lead to zinc contamination from the plastic . 3 3 2 Filtration and any other manipulation of a water sample should be carried out in a clean room or a clean air cupboard. Electroanalysis should also be carried out in a clean room, glass electrolysis cells should be siliconised and the cell and electrodes should be rinsed copiously with high-purity water. 153 6.2. Selected Electrochemical Speciation Results for Some Trace Metals in Waters 6.2.1. Copper Computer chemical models for the speciation of inorganic Cu in sea water predict that the carbonato and hydroxy complexes are the dominant species.lJ3154 The computed distribution of these complexes varies widely with the models used,3,155 but the latest calculations153 indicate that in sea water at pH 8.2, 25 "C and a total alkalinity of 2.3 mequiv.kg-1 inorganic Cu exists as CuC030 (82%) CuOH+ + C U ( O H ) ~ ~ (6.5%), Cu(OH)(C03)- (6.3%) CuHC03+ (1.0%) and Cu2+ (2.9%). These species are all believed to be ASV labile.24.155 In a typical fresh water more than 90% of inorganic Cu should be present as CuC03 although some is likely to be associated with colloidal particles of hydrated iron oxide .1,330,156 Coastal surface sea water usually has 40-60% of total Cu (the total copper concentration in surface Pacific water off Sydney is 0.3-0.8 yg 1-115') present as inert organic com-plexes.3.11 These complexes are so stable that they pass essentially unchanged through columns of iminodiacetate (Chelex 100) or thiol resins.71 It has been suggested that the Cu-binding ligands are siderophores metallothioneins or porphyrins.15,71 In unpolluted sea water ASV-labile copper usually comprises less than 50% of total dissolved Cu even at a pH as low as 4.7 (Table 7).3:71>151 Most of the inert Cu is organically bound but a significant fraction is inorganic, probably adsorbed on colloidal particles of hydrated iron oxide which are perhaps coated with humic acid .77849157J58 Most fresh water streams also have little ASV-labile Cu, and the percentage of organically bound Cu is usually Industrially polluted waters sometimes exhibit Cu pseudo-polarograms (Section 5.5) that do not have a plateau but which give continuously increasing peak currents with increases in deposition potential (Fig.4). Such behaviour indicates the presence of directly reducible copper complexes. The determination of the activity of free copper(I1) ion using the Cu ion-selective electrode is unreliable in chloride media. 162 high. 70,159-161 6.2.2. Lead Computer modelling of fresh waters suggests that carbonato species e.g. PbC03 and Pb2(0H)2C03 are the main (ca. 90%) inorganic Pb species,173 whereas in sea water speciation is divided between carbonato complexes (83%) and chloro species (11%).163 Calculations by Turner and Whitfield80 an 500 ANALYST MAY 1986 VOL. 111 ~~ Table 7. Concentrations and ASV-labile fractions of dissolved metals in surface sea water Concentration (ng 1-1) and labile fraction (%) in parentheses13 Metal c u .. Pb . . Cd . . Zn . . Ni . . Fe . . Mn . . Open ocean Near shore . . 120 350 (45) . . 14 250 (25) . . 15 75 (85) . . 10 1500 (50) . . 150 500 (70) . . 60 1500 (20) . . 750 3500 (<20) Valenta142 suggest that in sea water the carbonato and hydroxy complexes are only partially ASV labile. Unlike Cu Pb has a stronger affinity for some inorganic adsorbents especially iron oxide than for organic ligands, and it is likely that in most natural waters with pH above 7 a significant fraction of the Pb is associated with hydrated Fe203.1>3 Batley and Gardnerl51 found that in sea water 40-80% of dissolved Pb was present in the inorganic colloid fraction whereas in some low pH (pH 6.0) fresh waters most of the Pb appeared as an electroinactive inorganic molecular species possibly Pb2(0H)2C03.70 Most natural waters have little ASV-labile Pb (Table 7).31J64 Alkyllead species in natural waters may be determined by ASV165 after selective organic phase extraction.166 6.2.3. Cadmium In sea water Cd is computed to exist as the CdCl+ and CdCl2O complexes (92%) whereas in river water the dominant inorganic forms are Cd2+ and CdC03 depending on pH.1JJ63 A high proportion (>70%) of Cd is ASV labile in both sea water151 and fresh waters (Table 7).70 Because Cd ions are adsorbed on colloidal particles at only relatively high pH,1J very little Cd is present as pseudo-colloids. In anoxic waters, Cd may exist as non-labile CdHS+.3J51J63 Cd contamination during analysis can occur via rubber O-rings or seals and colour-code markings on pipettes.133 6.2.4. Zinc The main Zn species computed to be present in sea water are Zn2+ (27%) chloro complexes (47%) and ZnC03 (17%), whereas in fresh waters the dominant inorganic forms are Zn2+ (50%) and ZnC03 (38%).1JJ63 The carbonato com-plexes of Zn and especially the basic carbonates may have low ASV lability.70J67J68 Only about 50% of the total Zn in sea water and river water is ASV labile (Table 7) or extractable by ammonium tetramethylenedithiocarbamate, even though added ionic Zn spikes are completely extract-able.7.70J68 Open ocean water contains as little as 10 ng 1-1 of Zn at the surface,169 although coastal sea water usually contains 0.5-2 pg 1-1 of Zn as a result of river inputs and sewage outfalls.3J68J70 Zn determinations at the sub-pg 1-1 level are extremely difficult because of contamination problems which may originate from a variety of sources including paint skin, clothing plastics rubber filter membranes reagent chemi-cals and vapour from copying machines.lJ68 The HMDE generally produces more precise results for Zn than does a TMFE because small changes in hydrogen overpotential on a Hg-coated glassy carbon electrode affect the efficiency of Zn electrodeposition.High Cu concentrations depress the Zn ASV stripping peak as a result of the formation of intermetallic compounds in the Hg.171 This interference however is rare in natural water analysis. 6.2.5. Manganese The natural water chemistry of Mn is dominated by non-equilibrium behaviour.3 Oxidation of Mn(I1) to Mn(IV) i.e., to Mn02 is thermodynamically favoured in sea water and high pH fresh waters but the oxidation is extremely slow unless catalytic bacteria are present.172 Colloidal Mn02 is troublesome in water treatment plants because it blocks filters and causes discolouration.Both polarography173 and ASV47 have been used to determine labile Mn [Mn2+ and Mn(I1) complexes] in the presence of electroinactive Mn02. Mn(II1) , formed from the oxidation of Mn(I1) by the algae-produced superoxide radical (02 r) may also be present.172 Knox and Turner39 found that in samples from the Tamar Estuary (S. W. England) the polarographically detectable Mn level varied over a 6-month period from <lo% up to 100% of total manganese (31-252 pg 1-1) (Table 7).7. Correlation Between ASV-labile Measurements and Toxicity Variation in the speciation of trace elements can dramatically change their toxicity. Most studies of the toxicity of heavy metals to fish and other aquatic organisms have shown that the free (hydrated) metal ion is the most toxic form and that toxicity is related to the activity of free metal ion rather than to total metal ~ o n c e n t r a t i o n . 3 ~ 1 0 ~ 1 3 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 7 9 Toxicity usually decreases with increasing water hardness or salinity pre-sumably because of increased metal complexing by inorganic ligands. 14,180 Nature has provided aquatic animals with effective defences against ingested heavy metals which are eliminated via the gut,89J81 or detoxified in the liver kidneys and spleen by a group of high-sulphur proteins the metallo-thioneins which are synthesised in these organs in response to a heavy metal challenge.15 These defences allow the animal to cope with fairly high levels of heavy metals in the food chain and sediment; toxicity occurs only with “spillover,” i.e. when the metal intake exceeds the body’s ability to synthesise metallothionein. Evolution has not however equipped ani-mals to tolerate free metal ion in the water that contacts their gills or other exposed biomembranes. 15 Unpolluted sea water or river water contains very little free metal ion most of the dissolved metal being present as non-toxic complexes (e.g., with fulvic acid) or adsorbed on colloidal particles (e.g., humate-coated Fez03 or fibrils182).Natural waters use these detoxification mechanisms to convert free metal ions into non-toxic forms but considerable damage can be caused close to the source of pollution if the complexing capacity (Section 8) of the water is exceeded. Cu(I1) ions bind initially to marine phytoplankton with a stability constant log pl in the range 10-12 complexing apparently occurring via protein amino and carboxylic acid groups.183 Cu is then transported across the membrane by a carrier protein (facilitated diffusion),1843185 where it reacts with a thiol (possibly glutathione) in the cytosol or on the interior surface of the membrane and is reduced to Cu(I).1*3 Reaction with thiols and thiol-containing enzymes may be a common toxic effect of heavy metals although deactivation of enzymes such as catalase by metal substitution may also be involved.183,184 Although there is considerable evidence that free metal ion is the most toxic metal form the situation is not completely clear.14.16 Some studies suggest that other species such as the Cu hydroxy complex175 and the Cu complexes of citrate and ethylenediamine ,186 are also toxic. In addition lipid-soluble Cu10?15773 and Hg15 complexes are extremely toxic and a step to measure lipid-soluble metal complexes should be included in all trace element speciation schemes for polluted waters (Section 2.5). Attempts to use ASV-labile measurements to determine the toxic fraction of a metal have met with varied suc-cess.3~10,30,176,187~188 Young et al.30 using larval shrimp as a tes ANALYST MAY 1986 VOL. 111 501 Table 8. Correlation between ASV-labile and toxic fractions of copper in sea water using the marine diatom Nitzschia closterium LigandlO* Concentration Fulvic acid . . . . . . . . . . 1 x 10-5 M Tannic acid . . . . . . . . . . 6 x 10-7 M Iron . humic acid colloid§ LAS . . . . . . . . . . . . 0.5mg1-1 8-Hydroxyquinoline . . . . . . 5 x 1 0 - 8 ~ DMP . . . . . . . . . . . . 5~ 20-SM Ethylxanthogenate . . . . . . 2 x 1 0 - 6 ~ . . . . 1.0 + 5.3 mg 1-*§ NTA . . . . . . . . . . . . 2 x 1 0 - 5 ~ Copper/ M x 107 3.2 3.2 3.2 3.2 3.2 0.32 0.32 3.2 ASV-labile fraction,? % -0.6V -1.3V 1.5 2.9 5.5 10.5 70 74 100 100 65 100 64 100 10.5 48 - 2.5 Toxic fraction,$ % 7.5 12.5 60 20 25 >loo >I00 >loo * NTA = nitrilotriacetic acid; LAS = linear alkylbenzene sulphonate; DMP = 2,9-dimethyl-l,lO-phenanthroIine.t pH 8.2 with deposition potential of -0.6 or -1.3 V vs. Ag - AgCl. $ Fraction of added Cu appearing toxic compared with ligand-free solution. § 1.0 mg 1-1 of Fe + 5.3 mg 1-1 of humic acid. Table 9. Complexing capacity of some natural waters Source of water LakeOntario . . . . . . ChapelHillLake . . . . SwissLakes . . . . . . NewportRiver . . . . NeuseRiver . . . . . . Magela Creek Australia . . Pacific Ocean coastal . . t ASV titration with Cu*+. $ pH of titration. § Conditional stability constant of Cu complex. Complexing ~apacity1~191-f M X 106cU2+ .. . . 0.34 . . . . 31 . . . . 2.7 . . . . 0.87 . . . . 0.21 . . * . 0.10-0.46 . . . . 0.02-0.2 pHS Log * K § 7.4 8.6 6.0 5.0 8.8 10.9 7.0 9.7 6.8 9.5 6.0 7.6 4.8 -f 3.0 4.0 t --.3 . c 2 2 2.0 - Gradient = x V Y crr a 1.0 -Y 0.1 0.2 0.3 0.4 0.5 Concentration of copper addedip Fig. 7. Complexing capacity titration of a natural fresh water species found a good corielation between ASV-labile Cu and toxicity whereas Srna et al. 187 reported that ASV gave values that were only half those measured by bioassay. Florence et al. lo found that ASV-labile Cu determined in sea water using a low deposition potential correlated well with Cu toxicity towards the marine diatom Nitzschia closterium when natural complexing agents including fulvic humic tannic and alginic acids and hydrated iron oxide were present in the growth medium.However when synthetic ligands such as nitrolo-triacetic acid (NTA) 8-hydroxyquinoline or ethyl xanthogen-ate were present there was no sensible correlation (Table 8). The fraction of total dissolved Cu removed by a column of Chelex 100 resin grossly overestimated the toxic fraction.10 ASV-labile metal might therefore be a simple and reasonable method for measuring the toxic fractions of metals in natural waters but could be inapplicable if synthetic ligands are present. 8. Complexing Capacity Natural waters contain a variety of metal complexing agents, including fulvic humic and tannic acids lignin and colloidal particles of Fe2O3 A1203 and Mn02.1J89 Polluted waters may contain additional natural and synthetic compounds.The concentrations of these ligands are usually well in excess of those of the metals present and the determination of this excess “metal complexing capacity” is an important water quality parameter because it is a measure of the concentration of heavy metal that can be discharged to a waterway before free metal ion appears.193.190-202 Complexing capacity is determined by titrating the water sample with a heavy metal ion; Cu(I1) is usually chosen as the titrant because it is a common heavy metal ion highly toxic to aquatic organisms .65 Complexing capacity is then defined as the concentration of Cu(I1) ion (moll-1) that must be added to a water sample before free Cu2+ appears.It reflects the concentration of organic and inorganic substances in the water sample both molecular and colloidal that bind (and detoxify) Cu ions. Near-shore surface sea water has a Cu-complexing capacity of about 2 x 10-8 M whereas that of river waters ranges from 1 x 10-8 to 50 x 10-8 M (Table 9). Methods used to measure complexing capacity include bioassays ion exchange on resins or Mn02 ion-selective electrode potentiometry Cu salt solubilisation chemical exchange amperome try and vol tammetry. 7191~92203-~05 Of these methods voltammetry using an ASV titration has been most widely applied. ASV titration consists of adding aliquots of a standard Cu solution to the sample and measuring the Cu ASV peak until the slope of the peak current - Cu concen-tration graph increases to that found for ionic Cu (Fig.7) 502 ANALYST MAY 1986 VOL. 111 Assuming a 1 1 Cu - ligand complex the complexing capacity (C) and the apparent stability constant (* K ) can be found from a plot of the relationship191 [CU]/(CU - [CU]) = [CU]/C + 1/*KC . . (28) where [Cu] is the concentration of free Cu(I1) ion and CUT is the total Cu concentration. Some typical values for C and * K are shown in Table 9. There are several problems associated with the ASV titration method for determining complexing capacity. (i) Some Cu complexes such as Cu - NTA although thermo-dynamically stable are kinetically labile and dissociate extensively in the diffusion layer the complex appearing as free metal ion. These kinetic currents can be corrected for to some extent,156 but the procedure required is fairly com-plex.16 (ii) Organic matter adsorbed on the electrode may cause a depression of the metal ASV peak by hindering electrodeposition even though no actual complex formation takes place.(iii) Formation of the Cu complex may be slow, and several hours may need to be allowed between the additions of Cu titrant. It is often better to add increasing aliquots of standard Cu solution to a series of flasks containing a fixed volume of sample and to allow it to stand overnight before ASV measurement. The problem of electrode fouling by organics could be minimised by the use of linear scan voltammetry at a rapidly dropping mercury electrode or a streaming mercury electrode, provided that the samples have a sufficiently high complexing capacity.Interference from adsorbed organic matter increases in the order differential pulse polarography (DME) < linear scan ASV (TMFE) < differential pulse ASV (HMDE).149 In ASV the electrode is exposed to the organic matter for the period of the deposition time whereas in polarography exposure lasts only for the drop time. In linear scan voltammetry the metal ion has to cross the adsorbed organic layer only once during deposition whereas in differential pulse techniques where multiple redox reactions are invol-ved many crossings of the adsorbed layer must occur (Section 5.6). Van den Berg201 has described a ligand-exchange CSV method for measuring complexing capacity based on com-petition between the natural ligands and catechol for Cu ions, followed by cathodic stripping of the adsorbed Cu - catechol complex (Section 5.3).An estimate of the Cu - ligand conditional stability constant can also be obtained. This procedure has the advantages that the problem of dissociation of the Cu complex in the double layer is eliminated and interference by adsorption or organics may be less severe if the Cu - catechol complex is preferentially adsorbed. However, because of the high stability constant of the Cu - catechol cornplex,149 the method would measure only those ligands which form relatively stable complexes with Cu. Waite and More1200 described a novel amperometric titration method for complexing capacity using Cu(I1) as titrant and a high chloride media to stabilise Cu(1).Ion-selective electrodes measure the activity of free, hydrated metal ion and no other species. If a Cu ion-selective electrode is calibrated with a standard CuSO4 solution in non-complexing media (e.g. nitrate or perchlorate) then even simple complexes such as CuClf CU(OH)~ and CuCO3 will be included in the complexing capacity measurement as they are not sensed by the electrode. This explains why literature results for complexing capacity determined by Cu(I1) titrations using an ion-selective electrode for end-point detection are often much higher than ASV values for similar waters. 1.191 Before any method for measuring complexing capacity is chosen over others it should be shown that it gives a reasonable correlation with bioassay techniques otherwise it will have little value for ecotoxicological studies.The ideal method for measuring complexing capacity would be one where the affinity of the analytical probe for the metal ion titrant would be the same as that of a biomembrane e.g. the gill of a fish for the metal ion. If this ideal situation could be achieved the equilibrium constant for the reaction M L + P S M P + L . . . . ( 2 9 ) where P is the analytical probe and ML is the complex formed between the metal titrant and the natural ligand would be the same as the constant for the reaction M L + B e M B + L . . . . ( 3 0 ) where B is a biomembrane. 9. Conclusions and Recommendations for Future Research Speciation analysis is essential for an understanding of the biological and geochemical cycling of trace elements; simple total element analysis provides little information about these processes.Dividing trace elements in waters into different behavioural classes (speciation “boxes”) is a difficult task when the total concentration is at or below the pg 1-1 level. Electroanalysis especially anodic stripping voltammetry (ASV) is perhaps the most powerful technique available for this exacting branch of analytical chemistry. It must be appreciated however that ASV and polarography are dynamic techniques and cannot possibly measure the “natu-ral” speciation of a trace element in a water sample because the measurement itself disturbs the equilibrium. All electro-chemical speciation results are therefore operationally defined. This characteristic of electroanalysis may actually be an advantage as the interaction of a trace metal with a biomembrane is also a dynamic process and it should be possible to choose the solution and electrochemical paramet-ers so that the kinetics of electrodeposition are similar to the rate of uptake of a trace metal by a biological system.An important point here is that the effective measurement time of different electrochemical techniques (ASV polarograph y , linear scan pulse) varies considerably and hence the kinetic contribution of metal complexes to the analytical signal will also vary with the method used. Full analytical details (including calculation of the diffusion layer thickness) must therefore be reported in all published research on trace element speciation.Two broad areas of the methodology of electrochemical speciation analysis of waters would benefit from further research as follows. 1. The relationship between trace element speciation and aquatic toxicity. Much of the interest in speciation stems from the knowledge that the toxicities of different physico-chemical forms of an element vary enormously and that speciation analysis could possibly be used to determine the potential toxicity of a water system. There is little point in developing, from a purely chemical viewpoint “new” speciation schemes without consideration of their application. If the ultimate aim is directed towards ecotoxicology then development of the speciation method should be carried out in parallel with bioassays in an attempt to achieve the best correlation.On the other hand if the research aim is to study geochemical cycling, then the speciation procedure should be tailored to mimic as closely as possible the relevant adsorption and precipitation processes. These remarks also apply to the development of new methods for measuring complexing capacity; bioassays must be carried out hand-in-hand with the chemistry to ensure the relevance of the data. 2. Electrochemical speciation measurements may often be affected by extraneous substances especially surface-active compounds in solution. In ASV trace metal speciation the interpretation of the results is greatly simplified if it can be assumed that the deposition step alone controls the results ANALYST MAY 1986 VOL. 111 503 i.e.that the kinetics of metal deposition controls the magnitude of the stripping peak and that stripping kinetics are unaffected by ligands that are present in the sample but not the standard solution. Perhaps the best way to ensure that this situation exists is to use a medium exchange technique. This involves depositing metals from the sample solution but replacing it with a simple electrolyte (e.g. acetate buffer) before the stripping step. Research is required to design better cells for medium exchange and to determine the usefulness and application of the procedure. Adsorption of surface-active substances (e.g. humic mat-ter) from the sample solution on to the mercury electrode is one of the most serious complications in electrochemical speciation measurements.Both deposition and stripping currents may be decreased in an unpredictable manner and there is often a non-linear relationship between peak current and deposition time. Because the build-up of an adsorption layer on an electrode is a relatively slow process adsorption has less influence when short ASV deposition times (and short drop times in polarography) are used. To overcome the problem of adsorption the streaming mercury electrode (SME) should be investigated for speciation analysis when total metal concentrations are sufficiently high to allow its use. Because of the extremely rapid renewal of the electrode in a SME adsorptive processes and metal complexes with slow dissociation kinetics usually have little effect on the diffusion current.The SME may be especially useful for complexing capacity titrations. Another promising technique for the elimination of interference by adsorption is to cover the thin mercury film electrode with an ultrafiltration or cellulose acetate membrane. This type of covered electrode may be particularly useful in flow-through cells for continuous moni-toring of waters where electrode fouling is a vexing problem. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. References Florence T. M. and Batley G. E. CRC Crit. Rev. Anal. Chem. 1980 9 219. de Mora S. J. and Harrison R. M. Hazard Assess. Chem. Curr. Dev. 1984 3 1. Florence T. M. Talanta 1982 29 345. Florence T. M. Anal. Proc. 1983 20 552.Florence T. M. J. Electroanal. Chem. 1984 168 207. Burton J. D. Phil. Trans. R. SOC. London Ser. B 1979,286, 443. Batley G. E. and Florence T. M. Mar. Chem. 1976,4,347. Catanzaro E. J. Environ. Sci. Technol. 1976 10 386. Allen H. E. Boonlayangoor C. and Noll K. E. Environ. Int. 1982 7 337. Florence T. M. Lumsden B. G. and Fardy J. J. Anal. Chim. Acta 1983 151 281. Mackey D. J. Mar. Chem. 1983 13 169. Morel F. and Morgan J. J. Environ. Sci. Technol. 1972 6, 58. Whitfield M. and Turner D. R. in Jenne E. A. Editor, “Chemical Modeling in Aqueous Systems,” American Chem-ical Society Washington DC 1979 p. 657. Borgmann U. in Nriago J. O. Editor “Aquatic Toxicology,” Wiley New York 1983 p. 47. Florence T. M. Trends Anal. Chem. 1983 2 162. Turner D.R. Metal Ions Biol. Systems 1984 18 137. Florence T. M. Lumsden B. G. and Fardy J . J. in Kramer, C. J. and Duinker J. C. Editors “Complexation of Trace Metals in Natural Waters,” Martinus NijhofflW. Junk Publish-ers The Hague 1984 p. 411. Florence T. M. Stauber J. L. and Mann K. J. J. Inorg. Biochem. 1985 24 243. Chapman B. M. Jones D. R. and Jung R. S . Geochim. Cosmochim. Acta 1983 47 1957. Batley G. E. Mar. Chem. 1983 12 107. Bond A. M. “Modern Polarographic Techniques in Analy-tical Chemistry,” Marcel Dekker New York 1980. Turner D. R. and Whitfield M. J. Electroanal. Chem. 1979, 103 43. 23. 24. 25. 26, 27. 28, 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44 I 45, 46. 47. 48, 49. 50.51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 1 J Davison W. J. Electroanal. Chem. 1978 87 395. Davison W. and Whitfield M. J. Electroanal. Chem. 1977, 75 763. Batley G. E. and Florence T. M. J. Electroanal. Chem., 1974 55 23. Van Leeuwen H. P. J. Electroanal. Chem. 1979,99,93. Buffle J . J. Electroanal. Chem. 1981 125 273. Zirino A. and Kounaves S. P. Anal. Chem. 1977,49 56. Batley G. E. and Gardner D. Water Res. 1977 11 745. Young J . S . Gurtisen J. M. Apts C. W. and Crecelius, E. A. Mar. Environ. Res. 1979 2 265. Chau Y. K. and Lum-Shue-Chan K. Water Res. 1974 8, 383. Benes P. and Majer V. “Trace Chemistry of Aqueous Solutions,” Elsevier Amsterdam 1980. Florence T. M. J . Electroanal. Chem.1970 26 293. Florence T. M. Biochem. J. 1980 189 507. Leon L. E. and Sawyer D. T. Anal. Chem. 1981 53 706. Batley G. E. and Matousek J. P. Anal. Chem. 1980 52, 1570. Batley G. E. and Florence T. M. J. Electroanal. Chem., 1975 61 205. Phillips S. L. and Shain I. Anal. Chem. 1962 34 262. Knox S . and Turner D. R. Estuarine Coastal Mar. Sci., 1980 10 317. Henry F. T. Kirch T. O. and Thorpe T. M. Anal. Chem., 1979 51 215. Hamilton T. W. Ellis J . and Florence T. M. Anal. Chim. Acta 1979 110 87. Heyrovsky J. and Kuta J. “Principles of Polarography,” Academic Press New York 1966. Kubota H. Anal. Chem. 1960 32 610. Butler E. C. and Smith J. D. Deep-sea Res. 1980,27A 489. National Academy of Sciences “Chromium,” Committee on Biological Effects of Atmospheric Pollutants Division of Medical Sciences National Research Council US National Academy of Sciences Washington DC 1974.Mertz W. and Cornatzer W. E . “Newer Trace Elements in Nutrition,” Marcel Dekker New York 1971. O’Halleran R. J. Anal. Chim. Acta 1982 140 51. Orvini E. Zerlia T. Gallorini M. and Speziali M., Radiochem. Radioanal. Lett. 1980 43 173. Sibley T. H. and Morgan J. J . in Hutchison T. Editor, “Proceedings of the International Conference on Heavy Metals in the Environment,” Volume 1 University of Toronto, Ontario 1975 p. 319. Nelson A. and Mantoura R. F. J . Electroanal. Chem. 1984, 164 237. Figura P. and McDuffie B. Anal. Chem. 1979 51 120. Batley G. E. in Leppard G. G. Editor “Trace Element Speciation in Surface Waters,” Plenum Press New York 1983, p.17. de Mora S. J. and Harrison R. M. Water Res. 1983,17,723. Steinnes E. in Leppard G. G. Editor “Trace Element Speciation in Surface Waters,” Plenum Press New York 1983, Brugmann L. Sci. Total Environ. 1984 37 41. Skogerboe R. K . Wilson S.A. and Osteryoung J. G. Anal. Chem. 1980,52 1960. Laxen D. P. and Harrison R. M. Anal. Chem. 1981 53, 345 * Kramer C. J. Yu Guo H. and Duinker J. C. Fresenius 2. Anal. Chem. 1984 317 383. Goncalves L. M. Sigg L. and Stumm W. Environ. Sci. Technol. 1985 19 141. Florence T. M. and Batley G. E. Anal. Chem. 1980 52, 1962. Copeland T. R. Christie J. H. Skogerboe R. K. and Osteryoung R. A. Anal. Chem. 1973 45 995. Laxen D. P. and Harrison R. M. Sci. Total Environ. 1981, 19 59. Vurnberg H. W. and Raspor B.Environ. Technol. Lett., 1981 2 457. Velson A. Anal. Chim. Acta 1985 169. 273. Bhat G. A. Saar R. A. Smart R. B. and Weber J. H., 4nal. Chem. 1981 53 2275. lacobsen E. and Lindseth H. Anal. Chim. Acta 1976 86, 123. p. 37 ANALYST MAY 1986 VOL. 111 67. Batley G. E. and Florence T. M. J. Electroanal. Chem., 1976 72 121. 68. Florence T. M. J. Electroanal. Chem. 1974 49 255. 69. Batley G. E. and Farrar Y. J. Anal. Chim. Acta 1978 99, 283. 70. Florence T. M. Water Res. 1977 11 681. 71. Florence T. M. Anal. Chim. Acta 1982 141 73. 72. Blutstein H. and Smith J. D. Water Res. 1978 12 119. 73. Ahsanullah M. and Florence T. M. Mar. Biol. 1984,84,41. 74. Florence T. M. and Batley G. E. Talanta 1975 22 201. 75. Florence T. M. and Batley G. E. Talanta 1976 23 179.76. Figura P. and McDuffie B. Anal. Chem. 1980 52 1433. 77. Thomassen Y. Larsen B. V. Langmyhr F. J. and Lund, W. Anal. Chim. Acta 1976 83 103. 78. Batley G. E. and Matousek J. P. Anal. Chem. 1977 49, 2031. 79. Batley G. E. Anal. Chim. Acta 1981 124 121. 80. Turner D. R. and Whitfield M. J. Electroanal. Chem. 1979, 103 61. 81. Bond A. M. and Jones R. D. Anal. Chim. Acta 1980,121, 1. 82. Tuschall J. R. and Brezonik P. L. Anal. Chem. 1981 53, 1986. 83. Guy R. D. and Chakrabarti C. L. in Hutchinson T. Editor, “Proceedings of the International Conference on Heavy Metals in the Environment,” Volume 1 University of Toronto, Ontario 1975 p. 275. 84. Florence T. M. and Batley G. E. J . Electroanal. Chem., 1977 75 791. 85. Reignier M.and Buess-Herman C. Freseriius Z . Anal. Chem. 1984 317 259. 86. Ariel M. Eisner U. and Gottesfeld S . J. Electroanal. Chem. 1964 7 307. 87. Desimoni E. Palmisano F. and Sabbatini L. Anal. Chem., 1980 52 1889. 88. Whitfield M. and Turner D. R. in Jenne E. A. Editor, “Chemical Modeling in Aqueous Systems,” American Chem-ical Society Washington DC 1979 p. 657. Langston W. J. and Bryan G. W. in Kramer C. J. and Duinker J. C. Editors “Complexation of Trace Metals in Natural Waters,” Martinus Nijhoff/W. Junk Publishers The Hague 1984 p. 375. Boudou A. Georgescauld D. and Desmazes J. P. in Nriagu J. O. Editor “Aquatic Toxicology,” Wiley New York 1983 p. 117. 91. Williams R. J. Proc. R. SOC. London Ser. B 1981,213,361. 92. Nurnberg H. W. Valenta P. Mart L.Raspor B. and Sipos L. Fresenius Z. Anal. Chem. 1976 296 350. 93. Mart L. Nurnberg H. W. and Valenta P. Fresenius Z. Anal. Chem. 1980 300 350. 94. Florence T. M. Anal. Chim. Acta 1980 119 217. 95. Florence T. M. J. Electroanal. Chem. 1970 27 273. 96. Van der Linden W. E. and Dieker J. W. Anal. Chim. Acta, 1980 119 1. 97. Cowlard F. C. and Lewis J. C. J. Muter. Sci. 1967 2 507. 98. Clem R. G. Anal. Chem. 1975,47 1778. 99. Mann K. M. and Florence T. M. Sci. Total Environ. in the press. 100. Stewart E. E. and Smart R. B.,Anal. Chem. 1984,56,1131. 101. Wang J. and Hutchins L. D. Anal. Chem. 1985 57 1536. 102. Florence T. M. and Stauber J. L. Scz. Total Environ. in the press. 103. Yoshimura T. and Okazaki S. Fresenius Z. Anal. Chem., 1983 316 777. 104.Magjer T. and Branica M. Croat. Chem. Acta 1977,49 L1. 105. Kramer C. J. Guo-Hui Y. and Duinker J. C. Anal. Chim. Acta 1984 164 163. 106. Heineman W. R. and Kissinger P. T. Anal. Chem. 1980,52, 139 R. 107. Seitz W. R. Jones R. Klatt L. N. and Mason W. D.,Anal. Chem. 1973,45 840. 108. Blaedel W. J. and Wang J. Anal. Chem. 1979 51 1724. 109. Lieberman S . H. and Zirino A. Anal. Chem. 1974 46,20. 110. Wang J. and Dewald H. D. Anal. Chem. 1983 55 933. 111. Gunasingham H. and Fleet B. Anal. Chem. 1983,55,1409. 112. Florence T. M. and Farrar Y. J. Aust. J. Chem. 1964 17, 1085. 113. Cushman M. R. Bennett B. G. and Anderson C. W. Anal. Chim. Actu 1981 130 323. 89. 90. 114. Bond A. M. Fleishmann M . and Robinson J. J. Electro-anal. Chem. 1984 168 299. 115.MacFarlane D. R. and Wong D. K. J. Electroanal. Chem., 1985 185 197. 116. Wightman R. M. Anal. Chem. 1981 53 1125. 117. Ponchon J. L. Cespuglio R. Gonon F. Jouvet M. and Pujol J. F. Anal. Chem. 1979 51 1483. 118. Guadalupe A. R. and Abruna H. D. Anal. Chem. 1985,57, 142. 119. Tackeuchi E. S . and Osteryoung J. Anal. Chem. 1985 57, 1768. 120. Cheek G. T. and Nelson R. F. Anal. Lett. 1978 11 393. 121. Miwa T. Jin L. T. and Mizuike A. Anal. Chim. Acta 1984, 160 135. 122. Florence T. M. Anal. Chem. 1962 34 496. 123. Florence T. M. and Belew W. L. J. Electroanal. Chem., 1969 21 157. 124. Pihlar B. Valenta P. and Nurnberg H. W. Fresenius Z. Anal. Chem. 1981 307 337. 125. van den Berg C. M. Talanta 1984 31 1069. 126. van den Berg C. M. Anal. Chim. Acta 1984 164 195.127. van den Berg C. M. Anal. Lett. 1984 17 2141. 128. van den Berg C. M. and Huang Z . Q. J . Electroanal. Chem., 1984 177,269. 129. van den Berg C. M. and Huang Z . Q. Anal. Chim. Acta, 1984 164,209. 130. van den Berg C. M. and Huang Z . Q. Anal. Chem. 1984, 56 2383. 131. van den Berg C. M. Anal. Chem. 1985 57 1532. 132. van den Berg C. M. Anal. Proc. 1984 21 359. 133. Peterson W. M. Am. Lab. 1979 11 69. 134. Bond A. M. Hudson H. A. and Van den Bosch P. A. Anal. Chim. Actu 1981 127 121. 135. Borman S . A. Anal. Chem. 1982 54,698A. 136. Shimizu K. and Osteryoung R. A. Anal. Chem. 1981 53, 584. 137. Jagner D. Analyst 1982 107 593. 138. Labar C. and Lamberts L. Anal. Chim. Acta 1981 132,23. 139. Jagner D. Josefson M. and Westerlund S .Anal. Chim. Acta 1981 128 155. 140. Jagner D. Josefson M. and Westerlund S . Anal. Chem., 1981 53,2144. 141. Brown S. D. and Kowalski B. R. Anal. Chem. 1979 51, 2133. 142. Valenta P. in Leppard G. G. Editor “Trace Element Speciation in Surface Waters,” Plenum Press New York 1983, 143. Bubic S. and Branica M. Thalassia Jugosl. 1973 9 47. 144. Barrett P. Davidowski L. J. and Copeland T. R. Anal. Chim. Acta 1980 122 67. 145. Turner D. R. Robinson S. G. and Whitfield M. Anal. Chem. 1984 56,2387. 146. Kryger L. Anal. Chim. Acta 1981 133 591. 147. Brown S. D. and Kowalski B. R. Anal. Chim. Acta 1979, 107 13. 148. Osteryoung J. G. and Osteryoung R. A. Anal. Chem. 1985, 57 101A. 149. Varney M. S . Turner D. R. Whitfield M. and Mantoura, R. F. in Kramer C.J. and Duinker J. C. Editors, “Complexation of Trace Metals in Natural Waters,” Martinus Nijhoff/W. Junk Publishers The Hague 1984 p. 33. Cescon P. Scarponi G. and Moret I. Sci. Total Environ., 1984 37 95. Batley G. E. and Gardner D. Estuarine Coastal Mar. Sci., 1978 7 59. Landy M. P. Anal. Chim. Acta 1980 121 39. Bubic S . Sipos L. and Branica M. Thalassia Jugosl. 1973, 9 55. Symes J. L. and Kester D. A . Mar. Chem. 1985 16 189. Allen H. E. and Brisbin T. D. Thalassia Jugosl. 1980 16, 331. Leckie J. O. and Davis J. A. in Nriagu J. O. Editor, “Copper in the Environment Part 1 Ecological Cycling,” Wiley New York 1979 p. 90. 157. Balistrieri L. Brewer P. G. and Murray J. W. Deep-sea Res. 1981 28A 101. 158. Piotrowicz S. R. Springer-Young M.Puig J. A. and Spencer M. J. Anal. Chem. 1982,54 1367. 159. Baudo R. Mem. Ist. Ital. Idrobiol. 1981 38 463. p. 49. 150. 151. 152. 153. 154. 155. 156 ANALYST MAY 1986 VOL. 111 505 160. 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. Buffle J. Anal. Chim. Actu 1980 118 29. Negishi M. and Matsunaga K. Water Res. 1983 17 91. Westall J. C. Morel F. M. and Hume D. N. Anal. Chem., 1979 51 1792. Nordstrom D. K. in Jenne E. A. Editor “Chemical Modeling in Aqueous Systems,” ACS Symposium Series No. 93 American Chemical Society Washington DC 1979, p. 857. Benes P. Koc J. and Stulik K. Water Res. 1979 13 967. Bond A. M. Bradbury J. R. Hanna P. J.Howell G. N., Hudson H. A. and Strother S. Anal. Chem. 1984,56,2392. Colombini M. P. Fuoco R. and Papoff P. Sci. Total Environ. 1984 37 61. Bernhard M. Goldberg E. D. and Piro A. “The Nature of Seawater,” Proceedings Dahlem Konferenzen Berlin 1975, Florence T. M. in Nriago J. O. Editor “Zinc in the Environment. Part I Ecological Cycling,” Wiley New York, 1980 p. 199. Bruland K. W. Knauer G. A and Martin J. H. Nature (London) 1978 271,741. Martin J. H. Knauer G. A. and Flegal A. R. in Nriagu, J. O. Editor “Zinc in the Environment. Part I Ecological Cycling,” Wiley New York 1980 p. 193. Shuman M. S . and Woodward G. P. Anal. Chem. 1976,48, 1979. Stauber J. L. and Florence T. M. Aquat. Toxicol. 1985 in the press. Colombini M. P. and Fuoco R. Talanta 1983 30 901.Andrew R. W. Biesinger K. E. and Glass G. E. Water Res. 1977 11 309. Magnuson V. R. Harris D. K. Sun M. S . and Taylor, D. K. in Jenne E. A. Editor “Chemical Modeling in Aqueous Systems,” ACS Symposium Series No. 93 American Chemical Society Washington DC 1979 p. 635. Sunda W. G. Klaveness D. and Palumbo A. V. in Kramer, C. J. and Duinker J. C. Editors “Complexation of Trace Metals in Natural Waters,” Martinus NijhofflW. Junk Publish-ers The Hague 1984 p. 393. Babich H. and Stotzky G. in Nriagu J. O. Editor “Aquatic Toxicology,” Wiley New York 1983 p. 1. Petersen R. Environ. Sci. Technol. 1982 16 443. Hodson P. V. Borgmann U. and Shear H. in Nriagu J. O., Editor “Copper in the Environment. Part 11 Health Effects,’’ Wiley New York 1979 p.307. Pagenkopf G. K . Environ. Sci. Technol. 1983 17 342. Cross F. A. and Sunda W. G. in Wiley M. L. Editor, “Estuarine Interactions,” Academic Press New York 1978, p. 429. Leppard G. G. Massalski A. and Lean D. R. Protoplasma, 1977 92 289. p. 43. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. Florence T. M. and Stauber J. L. Aquat. Toxicol. in the press. Jennette K. W. Environ. Health Perspect. 1981 40 233. Albergoni V. and Piccinni E. in Leppard G. G. Editor, “Trace Element Speciation in Surface Waters,” Plenum Press, New York 1983 p. 159. Guy R. D. and Kean A. R. Water Res. 1980 14 891. Srna R. F. Garrett K. S. Miller S. M. and Thum A. B., Environ. Sci. Technol. 1980 14 1482. Gatcher R. Lum-Shue-Chan K. and Chau Y. K. Schweiz. Z . Hydrol. 1973 35 252. Langford C. H. Gamble D. S. Underdown A. W. andLee, S . in Christman R. F. and Gjessing E. T. Editors “Aquatic and Terrestrial Humic Materials,” Ann Arbor Science Ann Arbor MI 1983 p. 219. Plavsic M. Krznaric D. and Branica M. Mar. Chem. 1982, 11 17. Hart B. T. Environ. Technol. Lett. 1981 2 95. Neubecker T. A. and Allen H. E. Water Res. 1983 17 1. Tusuhall J. R. and Brezonik P. L. Anal. Chem. 1981 53, 1986. Shuman M. S. and Woodward G. P. Anal. Chem. 1973,45, 2032. Shuman M. S . and Cromer J. L. Environ. Sci. Technol., 1979 13 543. Shuman M. S. and Michael L. C. Environ. Sci. Technol., 1978 12 1069. Shuman M. S. and Woodward G. P. Environ. Sci. Technol., 1977 11 809. Bhat G. A. and Weber J. H. Anal. Chem. 1982,54,2116. Hirose K. and Sugimura Y. Mar. Chem. 1985 16 239. Waite T. D. and Morel F. M. Anal. Chem. 1983,55 1268. Van den Berg C. M. Mar. Chem. 1984 15 1. Wood A. M. Evans D. W. and Alberts J. J . Mar. Chem., 1983 13 305. Davey E. W. Morgan M. J. and Erickson S. J. Limnol. Oceanogr. 1973 23 993. Jardim W. F. and Allen H. W. in Kramer C. J. and Duinker J. C. Editors “Complexation of Trace Metals in Natural Waters,” Martinus Nijhoff/W. Junk Publishers The Hague 1984 p. 1. Van den Berg C. M. in Kramer C. J. and Duinker J. C., Editors “Complexation of Trace Metals in Natural Waters,” Martinus Nijhoff/W. Junk Publishers The Hague 1984 p. 17. Paper A51408 Received November 8th 1985 Accepted December 2nd 198