ISSN:0003-2654    

DOI:10.1039/AN98611FX017

出版商:RSC    

年代:1986

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98611BX019

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

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

ISSN:0003-2654    

DOI:10.1039/AN9861100489

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST MAY 1986 VOL. 111 507 Determination of Ethanol in Alcoholic Beverages Using a Headspace Procedure and Fuel Cell Sensor* W. J. Criddlet and K. W. Parry Department of Applied Chemistry UWIST P.O. Box 13 Cardiff CF? 3XF UK and 1. P. Jones Lion Laboratories Ltd. Barry South Glamorgan UK A procedure is described that allows the rapid determination of ethanol in a wide variety of alcoholic beverages. Dynamic headspace analysis is employed and a fuel cell sensor is used for the quantitative determination of ethanol. The method is rapid and compares favourably in accuracy with distillation and gas-ch romatograph ic procedures. Keywords Ethanol determination; alcoholic beverages; headspace analysis; fuel cell sensor The approved method of determining the ethanol content of a beverage in the UK and most other countries dates back to the early 1900s and is based essentially on studies made by Thorpe and Brown.1 Tables of data were established that related the specific gravity of distillates to original gravity and thus the ethanol content of the final product.Although improvements have been made over the years in distillation apparatus and in procedures for measuring specific gravity, the method is essentially unchanged. However certain disadvantages are inherent in the procedure including a substantial skill element on the part of the operator carrying out the analysis and the excessive time required for each determination. Cost can also be a factor when expensive products are being examined. In more recent times gas chromatography has become increasingly used for ethanoI assay2 and in some countries is accepted as a standard procedure.However gas chromato-graphs are relatively expensive and require the usual back-up facilities normally associated with a well equipped laboratory, as does the distillation procedure. Other techniques are available that depend either on density measurements3,4 or headspace analysis (Alcoltrol Ac3 L. H. Engineering Co., Stoke Poges Buckinghamshire) but both suffer from high cost and the need for a laboratory environment. In this paper we describe a prototype instrument and associated procedure that can be used to determine ethanol in alcoholic beverages with an accuracy comparable to those of the above-mentioned methods and that hopefully will provide the basis of a commercial portable and simple to use instrument for ethanol determination.procedure for charging the vessels was as follows. The standard vessel was half filled with water and a standard volume of an aqueous ethanol solution (2.0 cm3 5.0% V/V; 1.0 cm3 10.0% V / c 0.5 cm3 20.0% V/V; 0.25 cm3 40.0% VIV) was added to the vessel using a digital pipette e.g., Boehringer BCL 1000 DG and the solution was made up to the mark with water. All of the standard solution was removed from the pipette tip by injecting the sample below the water level and depressing the pipette plunger three or four times. After the addition of two drops of Dow Corning 1520 silicone antifoaming agent the sample vessel was filled identically using the appropriate volume of beverage.An air supply (ca. 300 cm3 min-1) was generated using a small air pump and passed alternately through the two vessels (isolated using the taps shown in Fig. 2) thus generating a dynamic headspace containing ethanol. The ethanol content was measured as described below. Headspace ethanol analysis The air emerging from the vessel was passed through a short polypropylene T-piece pierced by a microlance hypodermic needle fitted to a standard Lion Laboratories (Barry South Glamorgan) fuel cell sensor head. Sampling of the flowing air - ethanol stream was achieved by first pressing the RESET button and then the READ button. This resulted in a standard volume of ethanol - water vapour being drawn into the fuel cell where an electrical potential was developed.This potential was amplified and used as a measure of the ethanol concentration in the vapour. Experimental Instrument Construction The instrument consists essentially of two main components, viz. the headspace generation element and the electronic ethanol detection system. These two components will be considered separately. Headspace ethanol generation Two specially designed and calibrated (100 cm3 at 20 "C) vessels were constructed incorporating inlets and outlets for liquids air and a thermocouple inlet as shown in Fig. 1. The * Presented at Analyticon 85 London September 17-19th 1985. t To whom correspondence should be addressed. - Air - EtOH outlet - thermocouple inlet Calibration mark Fig. 1. vessel Diagrammatic representation of a headspace generatin 508 7 4 t 4 5 3 ANALYST MAY 1986 VOL.111 1 1 LY Fig. 2. Schematic flow diagram of the ethanol analyser. 1 Regulated air supply; 2 two-way ta 3 headspace generating vessel; 4 isolating taps; 5 polypropylene {Liece; 6 sampling valve fuel detector unit and microlance hypodermic needle; and 7 air - ethanol outlet Data Analysis The signal developed in the fuel cell was analysed using a Lion AE-D3 unit which was specially built for this study and derived from the company's microprocessor-controlled Auto Alcolmeter. In the AE-D3 four range controls are utilised for 5.0 10.0, 20.0 and 40.0 VfV. Depending on the nature of the unknown sample i.e. whether beer wine etc. the appropriate selector button is activated and the signal from the fuel cell obtained using the standard vessel is displayed (LCD).Note that the concentration of ethanol in the standard is always constant, but will be displayed as one of the four values indicated above. After calibration the air stream is passed through the sample vessel the new air - ethanol stream is sampled (after ca. 2 min to allow flushing of the previous sample) and the new signal is displayed on the LCD. As the fuel cell output is directly proportional to the ethanol concentration over a wide range of concentrations it is a simple matter to display the unknown ethanol concentration. Temperature Correction The temperatures of both sample and standard were moni-tored continuously using an RS Components Type 610-067 digital thermometer fitted with a Type K thermocouple and any difference in temperature was compensated for mathe-matically (see Discussion).In order to establish the relationship between the tempera-tures of the vessels and the instrument readings both vessels were filled with standard ethanol solutions (100 cm3 0.05-0.20% VfV) and the reference vessel was maintained at 25.0 "C in a thermostatically controlled bath while the temperature of the sample vessel was varied in the approxi-mate range 10-35 "C. The AE-D3 was calibrated in the usual way using the reference vessel and the apparent ethanol content of the sample vessel was recorded using the 5.00% Vf V range. Determination of Ethanol in Various Beers Wines and Spirits The ethanol contents of a wide range of alcoholic beverages were determined using the procedure described above with the appropriate volume of the various beverage types viz., beers (2.00 cm3) table wines (1.00 cm3) fortified wines (0.50 cm3) and spirits (0.25 cm3).Table 1. Determination of the slope (m) of the linear logarithmic plot of temperature versus displayed ethanol concentration Ethanol concentration Correlation Yo VIV Slope coefficient 0.05 0.08 0.08 0.10 0.10 0.10 0.10 0.13 0.15 0.20 0.0291 0.0285 0.0286 0.0272 0.0290 0.0279 0.0265 0.0275 0.0279 0.0290 0.998 0.997 0.999 0.998 0.997 0.999 0.999 0.999 0.999 0.998 Statistical analysis: n = 1O;X = 0.0281; S.D. = 8.27 x C.V. = 2.94% Standard / Unknown Tu Ts Temperaturei'C -Fig.3. Graphical representation of the mathematical involved in temperature compensation Irocess Gas-chromatographic Analysis of Reference Samples The ethanol contents of a wide variety of samples were determined using a Perkin-Elmer 8310 gas chromatograph fitted with a 3 m X 3 mm i.d. stainless-steel column contain-ing 10% Carbowax 20M on 60-80 mesh Celite and a flame-ionisation detector. The column was operated isother-mally at 80 "C using helium as the carrier gas (15 cm3 min-l) and propan-1-01 as an internal standard. Analysis of Reference Samples by Distillation The ethanol contents of all reference samples were also determined by the distillation procedure approved by HM Customs and Excise i . e . using gravimetric density rneasure-ment.5 Results and Discussion The technique of headspace analysis particularly in the context of alcoholic beverages has one major advantage over most other analytical procedures viz.that the number of possible interfering substances in the headspace is greatly reduced i.e. many substances present in beverages are non-volatile and would not contribute to the headspace composition. However the procedure does have two inheren ANALYST MAY 1986 VOL. 111 509 Table 2. Determination of ethanol in various alcoholic beverages Ethanol content,% VIV Classification Type Beers. . . . . . . . . . . Light ale Bitter ale Lager Chinese beer Claret Cream sherry Amontillado sherry White port Ruby port Spirits . . . . . . . . . . Whisky Rum Vodka Vodka Table wines .. . . . . Moselle Fortifiedwines . . . . . . Finosherry Liqueurs . . . . . . . . Irish whiskey cream liqueur Coconut liqueur Headspace analysis Gas chromatography 3.10 3.11 2.90 2.91 4.18 4.16 4.85 4.84 8.5 8.5* 12.2 12.2* 17.6 17.7* 17.7 17.7* 17.8 17.8* 20.2 20.3* 20.2 20.3* 40.6 40.0 40.5 40.0 38.0 37.5 46.4 45.8 17.0 16.9 27.7 27.8 Distillation 3.12 2.91 4.17 4.94 8.4 12.1 17.6 17.8 17.7 20.5 20.2 40.lt 40.1 37.4 45.6 16.91. 27.91. * Independently determined by J. Harvey & Sons Ltd. -t Independently determined by International Distillers and Vintners Ltd. Table 3. On-site results of ethanol determination in 1984 port wine vintage compared with laboratory distillation values. Samples 1-30 are red and 31-36 are white port wine Ethanol concentration,% V/V Sample No.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Distillation 20.0 19.8 20.3 20.2 20.0 19.6 18.8 18.9 17.5 17.3 18.3 19.3 19.9 18.5 19.6 20.0 17.8 19.6 16.0 16.8 19.7 19.7 19.1 19.8 18.7 19.7 20.1 20.1 19.8 18.2 17.7 16.1 16.3 17.6 19.3 19.3 * Sample diluted using river water. Analyser 20.0 19.8 20.2 20.4 20.1 19.6 18.9 19.0 17.9* 17.2* 18.9* 19.6 20.1 18.3 19.9 20.0 18.1 19.6 16.0 16.5 19.7 19.6 19.0 20.0 18.6 19.6 20.0 20.0 20.0 17.9 17.3 16.8 16.5 17.3* 19.7* 18.9 disadvantages viz.extreme temperature sensitivity the concentration of ethanol in the vapour phase increasing exponentially with increase in temperature and the effect of dissolved solids on the composition of the vapour phase i.e., the well known “salting-out” effect. Considering the latter problem no great difficulty was experienced in overcoming this. The simple act of dilution, necessary in any event to produce a headspace ethanol concentration suitably low for use with the fuel cell sensor, overcame the problem the “salting-out” effect being un-detectable at the drink dilutions used. Compensation for temperature variation between the stan-dard and sample solutions is however more complex. As indicated above the concentration of ethanol in a headspace, dynamic or static is exponentially dependent on temperature, the slope of the linear logarithmic plots being independent of concentration in the range 0.05-0.28% V/V.To avoid the necessity for thermostating the two vessels it is necessary to compensate mathematically for any variation. Interestingly, however it is not necessary to know the actual temperatures of the liquids but only the temperature difference between them. The following mathematical treatment illustrates this point. Table 1 shows the slopes of such plots the mean slope of these lines being 0.0281. A graphical representation of the process involved with temperature correction is given in Fig. 3 which shows logarithmic plots for the variation of head-space ethanol (as measured on the AE-D3) for standard and unknown samples.If we let S be the AE-D3 response to a standard at a temperature T (note that S is therefore the value of the range selected) R be the AE-D3 response to an unknown sample at a temperature T and U be the AE-D3 response to the unknown solution at the temperature of the standard solution T, i.e. the required corrected response value then from Fig. 3 and L o g S = m T + C1 . . . . . . (1) LogR = mT + C2 . . . . . . (2) where rn is the slope of the logarithmic plots and C1 and C2 are the intercepts for the standard and unknown lines respec-tively. From equation (2), C2 = logR - mT . . . . . .(3) and Log U = mT + C2 . . . . . . (4) i . e . , Log U = mT + log R - mT, = m(T - T,) + log R .. ( 5 510 ANALYST MAY 1986 VOL. 111 It follows that only the difference in temperature is involved in correction log R and m being determined experimentally. Results obtained using this correction were excellent and data obtained for actual determinations of ethanol (under laboratory conditions) in a wide variety of drinks are shown in Table 2. The agreement between standard ( i . e . gas-chromatographic and distillation) procedures and the head-space procedure described here shows clearly that the method is applicable to most beverages and has a precision of the same order as those of gas chromatography and distillation. The analysis times are considerably shorter if times are measured from the initial preparation of samples and standards. It is worth noting that data obtained for the analysis of high-ethanol beverages (spirits) are less satisfactory in absol-ute terms than for the other beverages studied.We attribute this to difficulties in reproducibly dispensing the small (0.25 cm3) sample volume associated with the analysis of spirits. However for general screening purposes the accuracy achieved coupled with the rapidity of the determination may make the procedure of some use to analysts in the spirits industry. If a precision similar to that obtainable by gas chromatography or distillation is required we do not recom-mend the headspace procedure as described in this paper. Extensive field trials of the determination of ethanol under non-laboratory conditions were undertaken during the port wine vintage of 1984 where a large number of wines were analysed on site in Portugal under the adverse conditions that operate in most port wine-producing farms.Water taken directly from the Douro River was sometimes used for dilution with no adverse affects as the results in Table 3 show. Under these conditions power was provided by the integral sealed, re-chargeable lead - acid battery in the AE-D3 unit which provided power for up to 8 h of continuous use. We would particularly recommend the use of digital pipettes for field use as their plastic construction makes them extremely robust. Turning to the fuel cell and its role as the sensor in this analytical study its mode of operation is such that a two-stage oxidation of ethanol occurs giving ultimately acetic acid as the final product.6 It is apparent therefore that compounds other than ethanol may be involved in similar oxidation reactions, and thus interfere quantitatively with data obtained.This is clearly true for most alcoholic beverages which contain a wide variety of volatile oxidisable organic species notably acetal-dehyde methanol and primary and secondary alcohols up to Cg. However the occurrence of these substances is such that in total they are not likely to exceed 500 mg 1-1 in beer wine or spirits and can therefore be discounted as interfering at a measurable level. The same argument is used to justify the distillation procedure where volatile components would affect the final distillate density and we would concur with this line of thought.It is worth noting the modification to the normal use of digital and similar pipettes. Standard usage was found to give erratic delivery of standard volumes depending on the nature of the beverage used. Certain of the more viscous liquids ( port wines liqueurs etc.) often left small globules in the disposable pipette tips thus affecting the precisely delivered volume. To overcome this problem the total contents of the tip were removed as described under Experimental. Although the precisely displayed volume may not therefore be delivered, reproducible volumes certainly were. Bearing in mind that a comparison procedure is involved in the analytical method described here approximate but reproducible volumes are the basic requirement not absolutely precise volumes as is usually the case.The use of an antifoaming agent is recommended for eliminating the carry-over of liquid from the calibrated vessel to the sampling chamber. Some alcoholic beverages particu-larly beer give solutions that show severe frothing and if no antifoaming agent is added bubbles form in the neck of the calibrated vessels and are carried to the sampling chamber. This clearly must be avoided and the recommended anti-foaming agent is extremely efficient in eliminating this effect. Whilst considering the possible carry-over of liquid from the vessels to the sampling chamber it should be noted that the bubbles of air passed through the solution are coarse and sinters are not used to produce fine bubbles as might be expected in an analytical procedure of this type.Sinters were tested in the first instance but it was found that a fine almost invisible spray was produced that was carried to the sampling chamber by the air flow. This resulted in the solution finding its way into the fuel cell with the obvious disastrous results. Removal of the sinter eliminated this effect and it should be noted that bubbles of large dimensions equilibrate very rapidly with a water - ethanol system.7 It would appear that the headspace procedure compares in precision with both gas-chromatographic and distillation methods for ethanol determinations. It should be pointed out that the determination of ethanol by either of the above established procedures requires a high degree of practical analytical expertise. In addition neither procedure can be conveniently operated outside a laboratory environment.We feel therefore that the procedure described in this paper has much to commend it in terms of simplicity of operation i.e. a relatively low skill requirement and its ability to be operated virtually anywhere and with very short warm-up periods. On these grounds alone the procedure scores well against the established procedures. In conclusion the prototype instrument described in this paper offers a rapid inexpensive (below &3000) and portable means for the determination of ethanol in most alcoholic beverages. It is hoped that it will in due course and with appropriate packaging and automation appear as a commer-cially available instrument for the alcoholic beverage indus-tries. The authors thank Mr. R. W. Goswell of John Harvey and Sons Ltd. and Dr. David Clutton of International Distillers and Vintners Ltd. for samples and analytical data and Dr. H. P. Reader of Cockburn Smithes and Sons Oporto for test facilities in Portugal. 1. 2. 3. 4. 5 . 6. 7. References Thorpe E. and Brown H. T. J . Znst. Brew. 1914 20 569. James A. T. and Martin A. J. P. Biochem. J . 1952,50,679. Stabinger H. Leopold H. and Kratky O. MLF Chem., 1967 98 436. Leopold H. Electronik 1970 19 297. “Laboratory Alcohol Table RDC 80/267/04,” HM Customs and Excise London 1979. Williams P. M. PhD Thesis University of Wales 1978. Jones A. W. PhD Thesis University of Wales 1974. Paper A51305 Received August 27th 1985 Accepted September 24th I98

ISSN:0003-2654    

DOI:10.1039/AN9861100507

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST MAY 1986 VOL. 111 511 Piezoelectric Quartz Crystal Detection of Ammonia Using Pyridoxine Hydrochloride Supported on a Polyethoxylate Matrix* Colin S. 1. Lai G. J. Moody and J. D. R. Thomas Department of Applied Chemistry Redwood Building UWIST P.O. Box 13 Cardiff CFI 3XF UK The use of a nonylphenoxypolyethoxylate (Antarox CO-880) as a support-polymer is confirmed as a means of prolonging the life (to >50 d) of pyridoxine hydrochloride as a sensitive sorbent coating during the piezoelectric crystal detection of ammonia. However the matrix system incurs possible interference from hydrogen chloride gas although except for triethylamine the other gases studied at high levels (sulphur dioxide nitrogen dioxide carbon dioxide and hydrogen sulphide) give only a small piezoelectric signal.The extreme sensitivity of the piezoelectric crystal detection of ammonia to below the pg dm-3 range is inconsistent with the Sauerbrey equation which normally applies to straightforward deposition on piezoelectric transducers. This is a consequence of the low slopes of the log(frequency decrease) versus log(concentration) graphs. Such slopes can be increased by modifying the syringe dilution procedure but there are other more intransigent factors involved. Keywords Ammonia detection; flow injection analysis; pol yethoxylate support; piezoelectric quartz crystal detection; gas sensor Suitably coated piezoelectric quartz crystal detectors form a highly sensitive technique for the detection of trace amounts of atmospheric pollutants. 172 The analyte is selectively sorbed by the coating thereby increasing the mass on the crystal and decreasing the frequency of vibration.The frequency change, AF (Hz) is linearly related to the mass sorbed according to the Sauerbrey equation,3.4 which for AT-cut crystals takes the form Am AF= -2.3 X l O 6 P -A where Fis the initial frequency of the quartz plate (MHz) Am is the mass sorbed (8) and A is the area of the coating (cm2). Thus for a particular experimental set-up the change in frequency can be expressed as AF= KC (2) where C is the analyte concentration (mg dm-3 or pg dm-3 or ng dm-3) and K is a constant that includes the basic frequency of the quartz crystal the area coated and a factor to convert the mass of analyte sorbed into its gas-phase concentration.Ucon 75-H-90000 and Ucon LB-3000X were first used for detecting ammonia in air and were found to have good sensitivity.5 These were followed by coatings of extracts of Capsicum annuurn pods and ascorbic acid with and without silver nitrate,6 and later by L-glutamic acid hydrochloride and pyridoxine hydrochloride (vitamin B6 hydrochloride) ,7 which showed exceptional sensitivity. Supporting the pyridoxine hydrochloride on a matrix of a high relative molecular mass polyethoxylate Antarox CO-880 (nonylphenoxypolyethoxy-late with 30 ethoxylate units) helps to extend considerably the useful lifetime of the detector.8 This paper describes further studies on the detection of ammonia using an AT-cut quartz crystal of 9-MHz resonant frequency coated with pyridoxine hydrochloride in a matrix of Antarox CO-880.The parameters studied include coating methods for applying the sorbent to the electrode surfaces of the crystal the effect of interferences procedural steps in the syringe dilution method for obtaining low concentrations of ammonia test gas and tests of the Sauerbrey equation. ~ ~~~~~~~ ~~~ * Presented at Analyticon 85 London UK September 17-19th, 1985. Experimental Apparatus and Detector Design A laboratory-constructed piezoelectric apparatus9 was assem-bled as shown in Fig. 1 and the operating conditions were optimised. The measuring unit consisted of a frequency oscillator with a buffered output powered by a Weir 400 power supply set at 9 V d.c. The frequency output from the oscillator was measured by a Marconi Type 2431A 200-MHz digital frequency meter.A digital to analogue converter selected the last two digits of the frequency meter output for conversion into an analogue signal to a Bryans Model 28000 chart recorder reading to +1 Hz. The AT-cut quartz crystal with gold electrode (Fig. 1 inset) (Quartz Crystal Co. Wellington Crescent New Malden, Surrey) had a resonant frequency of 9 MHz. The detector cell incorporating this crystal was based on the double impinger cell design of Karmarker and Guilbaultlo wherein the gas sample was split into two streams impinging directly on opposite faces of the coated crystal. The glass-encased cell detector was immersed in a water-bath at 25 k 0.1 "C. Pyridoxine Hydrochloride Coatings Coating solutions consisted of mixtures (1 + 1 V/V) of a 0.2% mlV solution of pyridoxine hydrochloride in ethanol and water (1 + 1 VlV) and a 0.2% mlV solution of Antarox CO-880 in acetone.For the capillary coating approach the mixture was applied to the electrodes cf the quartz crystal with a fine drawn-out capillary tube (melting-point tube). For the brush-coating approach a tiny brush was used to1 apply the mixture to the electrode surfaces in the manner described by Hlavay and Guilbault.11 In each instance the coated crystal was then placed in the oven at 80 "C to evaporate the solvent and crystals were stored overnight or between measurements at this temperature. The coating applied in each instance corresponded to a decrease of about 9.5 KHz in the basic frequency of the crystal and was readily removed with ethanol and water (1 + 1 V/V).The crystal was dried before re-coating. Ammonia Test Samples The ammonia gas test samples were obtained with a 10-cm3 gas-tight Perspex syringe from ammonia vapour over the headspace of dilute (2 M) ammonium solution equilibrated a 5 12 Digital to Recorder - analogue converter ANALYST MAY 1986 VOL. 111 Power -Frequency- Oscillator - supply meter Air - 3 I -f l Perspex block Fig. 1. Schematic diagram of piezoelectric quartz crystal detection apparatus with (inset on left) detail of quartz crystal V - 11111111111111111111llllllllllllllllllll Water-bath (25 _+ 0.1 "C) -= Charcoal = Silica gel= 25 "C. Serial dilution of the headspace gas was effected by syringe dilution12 with ambient air (dry air gave AF = 0).Successive dilutions were delayed by 30-60 s in order to allow ammonia to diffuse throughout the air in the syringe. The concentration of ammonia in the headspace was checked by titration. Thus 10-cm3 samples were slowly injected from the syringe into 20 cm3 of 0.025 M sulphuric acid. The excess of sulphuric acid was titrated with 0.1 M sodium hydroxide solution using methyl orange as an indicator. Thirty replicate samples of the headspace gas contained 32.0 mg dm-3 (s.d. 0.25 mg dm-3) of ammonia. Interfering gases were analysed in a similar way with appropriate absorbents and titrants. For the syringe dilution procedure 9 cm3 of the test gas were expelled from the syringe and air (9 cm3) was sucked into a total volume of 10 cm3 plus the volume of needle and syringe connector (0.23 cm3 s.d.= 0.01). The tip of the syringe needle was closed by piercing into a rubber bung. The mixture in the syringe was allowed to stand for 0.5-1 min in order to allow it to become homogeneous by diffusion. The original test gas was thus 10.23/1.23-times diluted. In the second dilution stage 1.23 cm3 of the first mixture was diluted to 10.23 cm3 giving a mixture of (10.23/1.23)2 times dilution over the original concentration. By repeating the procedure, mixtures of low concentrations could be obtained. Samples for Testing the Sauerbrey Equation For testing the Sauerbrey equation serially diluted samples were taken as above. Additionally sample dilutions were prepared by a procedure involving replacing the syringe needle with a clean one in between each dilution.Thus, commencing with a lo-cm3 syringe-full of headspace ammonia standard (32.0 mg dm-3) [actually 10.23 cm3 after allowing for the volumes of the needle (0.033 cm3) and the connector (0.197 cm3)] the following procedure was adopted for serial dilution: (a) expel 9 cm3 of the ammonia standard and replace the needle with a new one; (b) draw in 9 cm3 of air and allow to mix; (c) expel 9 cm3 of the diluted sample from (b) again replacing the needle with a new one; (d) repeat stage (b) and the concentration of ammonia standard should be 100 times diluted [actual concentration is 32 x (1.197/10.197)2 = 0.441 mg dm-3 = 441 pg dm-3 for a syringe needle connector volume of for example 0.197 cm3]; (e) expel 1 cm3 in order to fill the needle with sample; (f) inject 5 cm3 of the sample into the piezoelectric crystal detector; (8) expel 3 cm3 of the remaining sample from (f) then replace the needle with a new one; (h) pull in 9 cm3 of air and allow to mix; and (i) repeat stages ( 4 (f) (8) and (h).Operation of Piezoelectric Quartz Crystal Detection Apparatus The responses of coated piezoelectric quartz crystals were tested on 5-cm3 samples of appropriate dilutions of the headspace ammonia test samples and the mean decrease in frequency for replicate samples was measured. The diluted samples were injected into a carrier stream of dry (silica gel) air and passed through the quartz crystal compartment (at 25 "C) at a rate of 20 cm3 min-1 by the pump of a Pitman Instruments Model 7069 air sampler (Fig.1). As stated the power supply was set at 9 V d.c. Results and Discussion The manner of operation of the piezoelectric quartz crystal detection system is essentially a gas-phase mode of flow injection analysis. Sensitivity is helped by the highly commen-ded12.13 double impinger detector cell design. This is facili-tated by the sorption and subsequent desorption of ammonia by the pyridoxine hydrochloride: CH2OH CHqOH H+CI- H+CI-Fig. 2 illustrates recorder responses for serially diluted ammonia samples while Fig. 3 confirms the previously reported8 role of Antarox CO-880 as a matrix for prolonging the life of the piezoelectric detector for ammonia. The typical responses shown in Fig. 2 illustrate that although the response of the detecting system is fast the return to the base-line frequency takes several minutes because of the relatively slow desorption of the ammonia.However as mentioned previously,8 fresh samples may be injected before returning to the base-line frequency for it is the immediate decrease in frequency caused by the injected sample that is analytically significant ANALYST MAY 1986 VOL. 111 5 13 It has previously been shown8 that Antarox CO-880 coated on the quartz crystal without pyridoxine hydrochloride was not significantly involved in ammonia sorption but that it did sorb water. The extent of water sorption (only the carrier air stream was dried) for the larger samples used here ( 5 cm3 compared with 1 cm3 in earlier studies8) is shown in Table 1 for a pyridoxine hydrochloride - Antarox CO-880 coated crystal.As expected the larger volume samples used in this study (5 cm3) produced larger frequency changes [ca. 350 Hz for 463 pg dm-3 (Fig. 2)] than were observed in the previous study8 for pyridoxine hydrochloride in an Antarox CO-880 matrix for 1 cm3 samples (215 Hz for 30 mg dm-3). These data compare with 1190 Hz for 1 mg dm-3 and a surprising 386 Hz for 10 ng dm-3 reported by Hlavay and Guilbault4 for pyridoxine hydrochloride alone on the quartz crystal. Good linearity of calibration was obtained (Fig. 3) with correlation coefficients of 0.99 when log AF was plotted against log [NH,]. t TTT Moist air I Time -f .7 -0.012 0.096 20 min 8.8 -Fig. 2. Typical recorder trace of a calibration of ammonia gas using a cpartz crystal coated with pyridoxine hydrochloride and Antarox 0-880.Sam le size 5 cm3. Numbers on peaks are NH3 concentra-tion (pg dm-$ 2.5 N 2.0 I Q 1.5 IY Fig. 3. Calibrations over several days of a piezoelectric quartz crystal for ammonia coated with pyridoxine hydrochloride and Antarox CO-880 and illustrating long functional lifetime. Day A 1; 0 2; H, 4; '7 10; 0 15; 0 2 5 ; x 30; 0 3 6 ; V 46; A 57; + 61; and * 67 Sorbent Coating Methods Capillary-tube and brush-coating approaches were compared for the application to the quartz crystal of pyridoxine hydrochloride alone and of hydrochloride in Antarox CO-880. For the pyridoxine hydrochloride alone the capillary-tube approach gave an uneven coating as the hydrochloride was concentrated in certain areas thereby reducing the reacting surface area.The brush-coating approach gave a more even coating. Both approaches gave visually even coatings for pyridoxine hydrochloride in Antarox CO-880 although the brush-coating approach produced larger frequency changes (Table 2). The brush-coating approach was used for all other results discussed here. Tests of reproducibility of the brush-coating technique were carried out for later coatings (Table 3). The five coatings shown were made on the same quartz crystal the previous coating being removed after each set of calibrations by brushing the crystal surface gently with ethanol - water (1 1 rn/V). In each instance the crystal was oven dried at 80 "C for 30 min after coating as stated in the experimental procedure.The relative standard deviations for the frequency de-creases (Table 3) were generally low being less than 4% for the various concentrations of ammonia with 4.0% for 0.012 pg dm-3 of ammonia and 0.9% for the 463 pg dm-3 ammonia sample. The area of coating (A) was measured with a Quantitiet 800 Image Analyser (Cambridge Instruments Ltd.). The amount of coating material was calculated from Sauerbrey's equation (1). Table 1. Effect of moisture on the quartz crystal brush coated with pyridoxine hydrochloride and Antarox CO-880 Laboratory . . . . . . Dilutionstage 0 1 2 3 4 5 6 7 air AFIHz 60 57 47 35 27 23 22 19 21 . . . . . . . . . . Table 2. Comparison of the capillary tube and brush-coating methods of coating pyridoxine hydrochloride with Antarox CO-880 on quartz crystal electrodes.All results are given as AF (Hz); sample volume of ammonia standard = 5 cm3 Ammonia standard (pg dm-3) Moist Coating method 463 56 6.7 0.80 0.096 0.012 laboratoryair Day I : Capillarytube . . . . . . 278 176 110 63 37 24 28 Capillarytube . . . . . . 185 134 96 59 30 23 20 Capillarytube . . . . . . 164 117 94 56 34 22 24 Brush . . . . . . . . . . 392 200 142 101 66 41 25 Day 5: Brush . . . . . . . . . . 248 139 104 66 50 33 25 Day 10: Brush . . . . . . . . . . 278 192 130 84 52 31 2 514 ANALYST MAY 1986 VOL. 111 Table 3. Tests on the reproducibility of the brush-coating technique with relative standard deviation data for another five coatings by capillary tube coating Response (AFIHz) to NH3 with piezoelectric quartz crystal coated with pyridoxine hydrochloride and Antarox CO-880 Relative standard deviation of similar Relative data for standard capillary tube [NH,]/pg dm-3 1st coating 2nd coating 3rd coating 4th coating 5th coating Mean deviation '/o coating '/o 462.6 339 333 332 337 332 334.8 0.90 6.4 55.62 189 195 198 182 194 191.6 2.9 3.4 6.70 125 140 136 133 138 134.4 3.9 6.1 0.80 87 88 94 90 90 89.8 2.7 5.5 0.096 60 62 65 58 61 61.2 4.0 9.9 0.012 35 33 36 37 34 35.0 4.0 9.9 Area of coating/cm2 0.56 0.51 0.58 0.53 0.51 0.533 5.6 11.7 Amount deposited/pg 28.0 24.3 35.9 25.7 22.8 27.3 19.0 21.4 A F due to coating/Hz 9334 8905 11576 9042 8334 9438.2 12.0 10.7 ~ ~~ ~~~ ~ ~ Table 4.Interferences in the piezoelectric crystal detection of ammonia AFIHz NH3 .. NH3 . . so . . NO2 . . HC1 . . HCl . . coz . . HZS . . TEA* . . Concentration/ Gas mg dm-3 . . . . . . . . 3.85 . . . . . . . . 0.46 . . . . . . . . 101 . . . . . . . . 75 . . . . . . . . 109 . . . . . . . . 13.2 . . . . . . . . 1477 . . . . . . . . 116 . . . . . . . . 5.3 Dry lab. air . . . . . . . . Moist lab. air . . . . . . * TEA = triethylamine. Vitamin B6 coating 408 26 1 21 47 45 19 27 21 1 0 10 -Antarox CO-880 coating 37 23 34 32 2674 255 30 41 43 0 11 Vitamin B6 + Antarox CO-880 coating 320 44 35 1496 143 32 43 383 0 16 -The mass of coating material deposited on the crystal surface varied from 22.8 to 35.9 pg.As the coating material was a 1 + 1 mixture of pyridoxine hydrochloride and Antarox CO-880 the amount of pyridoxine hydrochloride was deemed to be half of the mean value of 27.3 pg that is 13.7 pg with a relative standard deviation of 19%. The frequency decrease due to the coating was 9.44 kHz that is slightly more than the 9.24 kHz for a similar analysis of the capillary tube coating method for which relative standard deviation data are presented in Table 3. The two sets of relative standard deviation data show that the brush-coating approach is more reproducible. Chemical Interferences Interferences in the piezoelectric crystal assay of ammonia from other gases are listed in Table 4 for the pyridoxine hydrochloride - Antarox CO-880 coating and for the separate materials.With a few exceptions these confirm previous observations.7 The generally low interferences from acidic gases on pyridoxine hydrochloride alone are expected as the concentra-tions of the interferents are much higher than the concentra-tion of ammonia gas injected and the observed frequency decreases are much less than those for 3.85 mg dm-3 of ammonia (Table 4). However 5.3 mg dm-3 of triethylamine caused a frequency change of 211 Hz. This is not unexpected, because amines have similar structures and properties to ammonia though Hlavay and Guilbault7 found that trimethyl-amine had no effect. Interference profiles for the mixture of pyridoxine hydro-chloride and Antarox CO-880 as coating material are generally similar to responses recorded for pyridoxine hydrochloride alone except for hydrogen chloride gas (109 mg dm-3) with a frequency decrease of 1496 Hz.This response is due to the reaction between hydrogen chloride and Antarox CO-880 and is even more marked when the crystal is coated with Antarox CO-880 alone. Such sensitivity can be attributed to hydrogen bonding between the hydrogen chloride and the ethoxylate oxygen as confirmed by broadening of the ethoxylate infrared absorption band at 3500 cm-1 (vSH stretching) for Antarox CO-880 in the presence of hydrogen chloride. The piezoeIec-tric interference reaction is reversible as shown by the fast return (ca. 5 s) of the frequency decrease to the base line. This is now being evaluated for hydrogen chloride sensing14 as the frequency decrease for 109 mg dm-3 of ammonia is a much greater response than 400 Hz for 100 mg dm-3 using trimethylamine hydrochloride as substrate coating.l1 Application of the Sauerbrey Equation The analytical utilisation of coated piezoelectric quartz crystal detectors has been based on the assumption that Sauerbrey's equation1 is valid i.e. that the mass increase caused by sorption is directly related to the concentration of the sample in the flowing gas stream and is proportional to the decrease of the resonance frequency. 1.2 However all analysis of previous published data by Beitnes and Schrcbder15 shows that the sensitivity of piezoelectric crystal detectors for flowing gas streams does not obey the Sauerbrey equation. Thus as previously indicated,16 despite the poorer sensitivities that would be expected from incomplete sorption on the crystal coating mixing effects with carrier gas etc.the observed decreases in frequency are often greater than the valu ANALYST MAY 1986 VOL. 111 515 Table 5. Comparison of ammonia present in 5 cm3 of sample according to dilution calculations (for 3rd coating data of Table 3) and ammonia calculated [from the frequency change by the Sauerbrey equation (l)] to be sorbed on the piezoelectric crystal coating Concentration of Ammonia present Ammonia calculated to NH3 in in 5 cm3 of be sorbed on coating sample/pg dm-3 AFIHz samplehg (for A = 0.58 cmZ)/ng 463 332 2320 56 198 280 6.7 136 34 0.80 94 4 0.096 65 0.48 0.012 36 0.06 1030 620 420 290 202 112 2.5 2.0 2 Q Q ol -J 1.5 1 .o -2 -1 0 1 2 Log([ NHsIIw d r r 3 ) Fig.4. Ammonia content of standards deduced by syringe dilution procedures related to frequency decrease (A for Table 5 data and C) compared with ammonia content (assuming 100% absorption) calcu-lated from Sauerbrey equation (1) using observed frequency decrease for the corresponding syringe diluted standards (B for Table 5 data and D). Lines A and B are for the ordinary syringe dilution method (with allowance for needle and connector volume) and lines C and D are for the alternative dilution method with needle replacement (with allowance for connector volume) deduced from Sauerbrey’s equation. Conversely observed decreases in frequency relate according to the Sauerbrey equation to mass changes that are greater than the amount of the sought component actually present in the sample (see Table 5).A graph of the data of columns 1 and 2 of Table 5 (A of Fig. 4) has a much lower slope than that of unity expected for log AF versus log C according to the logarithmic form of equation (2). By changing the dilution procedure from the ordinary serial dilution with allowance for needle and needle connector volumes to one where the needle is changed for each dilution stage (detailed in the Experimental section relating to testing the Sauerbrey equation) the slope of log (sample ammonia concentration) versus log AF is steeper (Fig. 4 C) and nearer to the expectation of the Sauerbrey equation (Fig. 4 D). Even greater anomalies than exist in the above data occur in other reported work e.g.for the 0.01 pg dm-3 ammonia sample of Hlavay and Guilbault7 with a AF of 386 Hz the ammonia calculated to be sorbed by the pyridoxine hydro-chloride coating is about 1 pg compared with the 0.00005 pg deemed to be present in the 5-cm3 sample used. The anomalies are related to the gentle slopes observed for the log AF versus log Cgraphs and some previously reported7 slopes are very gentle e.g. 0.0615 for the coating of L-glutamic hydrochloride and 0.0978 for a coating of pyridox-ine hydrochloride. In this study the slopes of the graphs of Fig. 3 are between ca. 0.15 and ca. 0.2 while that of C in Fig. 4 improves to >0.4. Table 6. Frequency decrease (AF) data for the removal of 463 pg dm-3 of ammonia from a 10-cm3 syringe by successive evacuation and re-filling with dry air after replacing the needle at the end of each evacuation.(The 5-cm3 samples examined for piezoelectric frequency changes corresponded to the volume fraction between 1 and 6 cm3) AF/Hz for different tests ~~~ ~ Re-fill No. 1 2 3 4 0 323 316 320 320 1 28 25 16 28 2 7 7 6 6 3 0 0 0 0 Data such as the above have led Beitnes and Schroderls to investigate systematic errors in the syringe dilution method, but the likelihood of sorptions on syringe walls between dilutions is demonstrated (Table 6) to be an incomplete explanation of the anomalies and suggest that other factors are involved. The problem revolves around the better t’han predicted sensitivities and Beitnes and Schroder15 found that alterna-tive dilution methods such as bottle dilution also give sensitivities that are better than predicted.Although not a complete solution a change in the syringe dilution procedure, as discussed above brings the experimental response nearer to the predictions of the Sauerbrey equation (Fig. 4 C) but there could be other factors not recognised here. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. References Guilbault G. G. Ion-Sel. Electrode Rev. 1980 2 3. Alder J. F. and McCallum J. J. Analyst 1983 108 1169. Sauerbrey G. Z . 2. Phys. 1959 155 206. Sauerbrey G . Z. 2. Phys. 1964 178 457. Karmarkar K. H. and Guilbault G . G. Anal. Chim. Acfa, 1975 75 111. Webber. L. M. and Guilbault G. G . Anal. Chem. 1976,48, 2244. Hlavay J. and Guilbault G. G. Anal. Chem. 1978,50,1044. Moody G. J. Thomas J. D. R . and Yarmo M. A . Anal. Chim. Acta 1983 155 225. Cannard A. J. Moody G. J. Thomas J. D. R. and Yarmo, M. A. unpublished work. Karmarkar K. H. and Guilbault G. G. Anal. Chim. Acta, 1974 71 419. Hlavay J. and Guilbault G. G. Anal. Chem. 1978,50 965. Karasek F. W. and Tienay J. W. J. Chromatogr. 1974 89, 31. Cooke S . West T. S. and Watts P.,Anal. Proc. 1980,17,2. Lai C. S. I. Moody G. J. and Thomas J. D. R. to be published. Beitnes H . and Schrcbder K. Anal. Chim. Acta 1984 158, 57. Lai C. S. I . Moody G. J. and Thomas J. D. R . Anal. Proc., 1985 22 10. Paper A5f386 Received October 28th I985 Accepted November 25th 198

ISSN:0003-2654    

DOI:10.1039/AN9861100511

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST MAY 1986 VOL. 111 517 Correlation Between Fluorescent Polarisation lmmunoassay and Enzyme lmmunoassay of Anticonvulsant Drugs and Stability of Calibration Graphs* Neville Ratnaraj Valerie D. Goldberg and Peter T. Lascelles Department of Chemical Pathology The National Hospital Queen Square London WCI N 3BG UK Quality control materials and serum samples from patients on long-term drug therapy were analysed for anticonvulsant drugs by enzyme immunoassay (EMIT) and fluorescence polarisation immunoassay (TDX). The accuracy and precision of the two procedures were studied and the stability of calibration graphs was evaluated over a 30-d period. The accuracy and precision of both assays were satisfactory over the therapeutic ranges of phenobarbitone primidone phenytoin and carbamazepine and there was a good correlation between the results obtained by EMIT and TDX; for sodium valproate the accuracy and precision of the EMIT assay were poor.Calibration graphs generated by the TDX procedure were found to be stable but with the EMIT procedure calibration graphs for phenobarbitone carbamazepine and sodium valproate showed considerable drift. Keywords Enzyme immunoassa y; fluorescence polarisation immunoassa y; calibration graph stability; anticonvulsant drugs Over the last few years the concept of therapeutic drug monitoring (TDM) has become widely accepted as a valuable aid in the care and management of patients with epilepsy.' Well established techniques used for measuring concentra-tions of anticonvulsant drugs in serum include UV spectropho-tometry,2-6 gas - liquid chromatography (GLC)7-14 and high-performance liquid chromatography (HPLC) ,13-17 but as a result of the increase in the numbers of determinations carried out new methods of analysis have been developed that emphasise speed in addition to precision and accuracy.13 The introduction of radioimmunoassay~~J9 and its adaptation for the analysis of anticonvulsant drugs20 was followed by the development of the related techniques of enzyme immuno-assay fluoroimmunoassay and substrate-labelled fluores-cence immunoassay.21-29 These methods have proved to be particularly well suited for the production of commercial kits and hence for automation under microprocessor control and have enabled the drug monitoring units to keep abreast of a rapidly increasing work load.13 The technique of enzyme immunoassay (EIA) has had an enormous impact in the field of TDM,21 mainly because of the development of EMIT a homogeneous immunoassay system that was introduced by the SYVA Company in 1974. Using a discrete analyser samples can be processed at a rate of between 60 and 100 per hour and the results obtained correlate well with results from standard methods including GLC and HPLC.24JO-34 In 1981 as a further development in the field of TDM the Abbott Company introduced the TDX, which employs the principle of fluorescence polarisation.35J6 Although the principle was first described as early as 1926,37 and adapted for drug determinations in 1973,38 the TDX was the first commercial application of fluorescence polarisation immunoassay (FPIA).Results obtained using this method correlate well with those obtained using other pro~edures.39~40 Both the EMIT and TDX procedures are now widely used in the monitoring of anticonvulsant drugs in the routine situation and the accuracy and precision of both methods are acceptable in the context of routine clinical chemistry which commonly works to a 95% confidence limit. Both methods have been designed to take full advantage of modern instrument technology which is increasingly based on micro-processor control of the function and sequence of the analysis; sampling and dilution procedures can thus be automated by * Presented at Analyticon 85 London UK September 17-19th, 1985. the use of robotic arms and probes.Further introduction of EPROM (erasable program read-only memory) has the advantage that calibration graphs can be stored in memory and their stability monitored over a long period so that a fresh calibration graph does not have to be generated for each run. In order to establish limits for the stability of the EMIT and TDX reagents in the routine analysis of anticonvulsant drugs, we now present a comparison of the two techniques in respect of the stability of calibration graphs over a period of 30 d. We also present data on accuracy and precision obtained by analysing quality control materials and pooled patients' sera for anticonvulsant drugs by both methods. Experimental Patients' Samples Blood samples collected without anticoagulant from patients at the National Hospitals and Chalfont Centre for Epilepsy for the routine measurement of anticonvulsant drugs were used; all the patients were on two or more drugs.After centrifuga-tion the sera were stored at -20 "C. Quality Control Sera Quality control sera prepared by pooling patients' sera known by previous analyses to contain the drugs required were used to establish the within- and between-batch precision of the assays and the stability of the calibration graphs. The patients' sera were divided into two groups. One group was pooled, filtered twice with thorough mixing aliquoted into bottles and deep frozen at -20 "C (QCRa). The other group was subdivided into three batches of sera with subtherapeutic (SL) therapeutic (TL) and toxic (TO) levels of each drug, aliquoted and stored deep frozen (Table 1).In addition a commercial quality control material Ortho Bi-Level Assayed Anticonvulsant/Antiasthmatic Control Set I (Corning Medical) was used to investigate the accuracy of the assays and the stability of the calibration graphs. Methods Kits for the determination of phenobarbitone primidone, phenytoin carbamazepine and sodium valproate by the EIA (EMIT) procedure (SYVA) and FPIA (TDX) procedure (Abbott) were used according to the manufacturers' instruc 518 ANALYST MAY 1986 VOL. 111 Table 1. Therapeutic ranges of anticonvulsant drugs Therapeutic range*/ Drug pmoll-1 Phenobarbitone . . . . 20-130 Primidone . . . . . . 15-60 Phenytoin . . . . . . 28-67 Carbamazepine . . . . 12-50 Sodiumvalproate .. . . 360-600 * Values adapted from the literature and used routinely at The National Hospital. tions and the analyses carried out using an EMIT Auto-Carousel with CP5000 EMIT Clinical Processor (SYVA) and a TDX Analyser (Abbott). Calibration graphs were generated on the EMIT Auto-Carousel and Abbott TDX according to the manufacturers’ instructions. Correlation between the EIA and FPIA methods was examined by analysing randomly chosen patients’ samples for the five anticonvulsant drugs in duplicate on the same day and carrying out regression analysis and the accuracy and precision of the two procedures and the stability of the calibration graphs were assessed using the quality control sera Ortho I and QCRa and the pooled patients’ sera SL TL and TO as follows.The accuracy of the EIA and FPIA methods was investi-gated by analysing samples of the commercial quality control serum Ortho I in duplicate by both procedures. The within-batch precision of the EIA and FPIA methods was determined by analysing 40 samples of the quality control serum QCRa in duplicate by both procedures on the same day and the between-batch precision was assessed by analysing samples of QCRa in duplicate by both procedures in separate routine runs using newly generated calibration graphs for each run. The stability of the calibration graphs for the five anticon-vulsant drugs over a 30-d period and the precision of the assays under these conditions were determined by the following procedure. Using freshly prepared calibration graphs samples of QCRa and Ortho I and one sample from each of the groups of pooled patients’ sera SL TL and TO were determined by both methods.The calibration graphs for the EIA and FPIA procedures were stored in the memories of the CP5000 and TDX respectively and the analyses of the quality control samples and pooled patients’ sera were repeated at 3-d intervals for 30 d using the stored calibration graphs which were retrieved from memory as required. Results and Discussion Figs. 1-5 show the results obtained from the analysis of random patients’ samples by EIA and FPIA. It can be seen that the correlation between the two methods is acceptable over a wide range of concentrations; the correlation coeffi-cient lies between 0.98 and 0.99. Results of the within-batch analysis of samples of Ortho I by the two immunoassay procedures are shown in Table 2.It can be seen that the accuracy of the two immunoassay procedures is acceptable for phenobarbitone primidone phenytoin and carbamazepine although with sodium valproate the mean result of 40 analyses of Ortho I using the EMIT procedure barely falls within the range for the analysis quoted by the manufacturer. On the other hand it can be seen from Table 3 that the accuracy of the EMIT assay is unacceptable for phenobarbitone and sodium valproate with the mean result of 40 analyses of Ortho I falling well outside the manufacturer’s quoted range when between-batch analyses were carried out using the same calibration graphs for a period of 30 d. The -0 5 200 2 n I- > 160 II J I I I I I 0 40 80 120 160 200 240 Phenobarbitone level by EMlTiymol I-1 Fig.1. Correlation of TDX with EMIT assay of phenobarbitone. Analysis of randomly collected patients’ samples n = 40; r = 0.99; slope = 1.07; and intercept = -5.27 70 c I1 60 > 401 Q) -.E 20 a‘ 301 l o t //. I / I I I 1 I 0 10 20 30 40 50 60 70 80 Primidone level by EMIT/ymol I - 1 Fig. 2. Correlation of TDX with EMIT assay of primidone. Analysis of randomly collected patients’ samples n = 40; r = 0.98; slope = 1.01; and intercept = -0.99 100 r / 90 -80 -r I -Z 7 0 -8 60-I-9” 50-U 40-5 30 5. -0) C 0 C .--E 20-10 -0 10 20 30 40 50 60 70 80 90 100 Phenytoin level by EMITiymol 1-1 Fig. 3. Correlation of TDX with EMIT assay of phenytoin.Analysis of randomly collected patients’ samples n = 50; r = 0.99; slope = 0.99; and intercept = 0.7 ANALYST MAY 1986 VOL. 111 accuracy of the TDX procedure is acceptable under these conditions (Table 3). Table 4 shows the results of the within-batch analysis of QCRa by both immunoassay procedures. It can be seen that the within-batch precision of both immunoassay procedures measured by the analysis of QCRa is acceptable for phenobar-bitone primidone phenytoin and carbamazepine with a standard deviation of less than 2.5 pmol l-1 and a coefficient of variation of less than 5%. For sodium valproate however, whereas the coefficient of variation for the analysis of 40 samples by both procedures is less than 570 the standard deviation for the EMIT assay is as high as 15.37 pmol l-1 and for the TDX assay the standard deviation is 9.13 pmol 1-’ 900 7 800 0 5 700 600 - -2 > a 500 a -2 400 2 Tz > 7ij 300 .- 5 200 -u v 100 519 ---------I I 1 I 1 I I a I / 0 10 20 30 40 50 60 70 Carbamazepine level b y EMlTiprnol I-’ Fig.4. Correlation of TDX with EMIT assay of carbamazepine. Analysis of randomly collected patients’ samples y1 = 40; r = 0.98; slope = 0.96; and intercept = -1.41 Only for phenobarbitone and primidone in this experiment are the standard deviation and coefficient of variation lower for the EMIT assay than for TDX. Similar results were obtained from the between-batch analysis of the five drugs by the TDX and EMIT procedures as shown in Table 5 .The performance of both assays is satisfactory for phenobarbitone, primidone phenytoin and carbamazepine but less so for sodium valproate with a standard deviation of 6.35 pmol 1-1 for the TDX assay and 12.64 pmol 1-1 for EMIT under the conditions of the experiment. In the EMIT assay these results are in marked contrast to the results shown in Table 6, obtained from the between-batch analysis of QCRa for all five drugs using the same calibration graphs throughout the 30-d Table 2. Results of within-batch analysis of Ortho Control Set I by TDX and EMIT Drug Method Phenobarbitone . . . . . . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate . . . . TDX EMIT EMIT EMIT EMIT EMIT No.of samples 40 40 40 40 40 40 40 40 40 40 Mean/ pmoll-1 65.24 76.88 33.82 32.24 28.24 29.45 17.28 19.16 330.42 420.87 Standard deviation/ pmol 1-1.10 2.86 0.98 1.89 0.99 1.24 0.78 1.38 10.24 20.38 Coefficient of variation % 1.38 4.28 2.64 4.68 2.86 4.28 2.69 4.88 4.08 9.85 Mean concentration quoted by manufactured pmoll-1 59 t- 12 65 k 13 33 f 6 33 k 7 25 f 5 28 k 6 17 k 3 18 f 3 326 f 62 354 f 69 Table 3. Results of between-batch analysis of Ortho Control Set I by TDX and EMIT using the same calibration graph over a 30-d period Mean concentration Standard quoted by No. of Mean/ deviation/ Coefficient of manufactured Drug Method samples pmol 1- 1 pmoll-1 variation % pmol 1-Phenobarbitone .. . . . . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate . . . . TDX EMIT EMIT EMIT EMIT EMIT 11 11 11 11 11 11 11 11 11 11 62.09 80.73 33.64 30.64 26.64 27.45 16.73 20.64 321.55 547.64 1.58 5.00 1.03 2.11 1.03 1.44 0.65 1.43 11.41 113.08 2.45 6.20 3.05 6.89 3.86 5.24 3.87 6.95 3.55 20.65 59 k 12 65 f 13 33 f 6 33 f 7 25 k 5 28 f 6 1 7 f 3 1 8 k 3 326 f 62 354 k 6 520 ANALYST MAY 1986 VOL. 131 ~~~~ ~ ~ Table 4. Results of within-batch analysis of quality control serum QCRa by TDX and EMIT Drug Method Phenobarbitone . . . .. . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate . . . . TDX EMIT EMIT EMIT EMIT EMIT No. of samples 40 40 40 40 40 40 40 40 40 40 Mean/ pmol I-' 77.63 76.32 22.03 22.20 44.00 40.82 34.40 36.97 375.88 360.35 ~ Standard deviation/ pmoll-1 2.00 1.83 1.07 0.41 0.99 1.80 0.50 1.58 9.13 15.37 ~ Coefficient of variation o/o 2.67 2.40 4.88 1.82 2.24 4.40 1.44 4.26 2.43 4.27 Table 5. Results of between-batch analysis of quality control serum QCRa by TDX and EMIT Drug Method Phenobarbitone . . . . . . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate .. . . TDX EMIT EMIT EMIT EMIT EMIT No. of samples 40 40 40 40 40 40 40 40 40 40 Mean/ pmol I-' 75.15 78.70 22.03 21.38 43.38 42.38 33.65 36.70 368.80 378.40 Standard deviation/ pmol l-1 2.55 3.60 0.62 0.74 1.76 1.27 1.12 1.11 6.35 12.64 Coefficient of variation Yo 3.44 4.57 4.81 3.46 4.07 4.01 3.30 3.04 1.75 4.62 Table 6. Results of between-batch analysis of quality control serum QCRa by TDX and EMIT using the same calibration graph over a 30-d period Drug Method Phenobarbitone . . . . . . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate . . . . TDX EMIT EMIT EMIT EMIT EMIT No.of samples 11 11 11 11 11 11 11 11 11 11 Mean/ pmol - * 71.27 86.18 21 .oo 19.18 42.09 40.36 33.55 40.91 350.73 557.91 Standard deviation/ pmol 1- 1 2.15 7.85 1 .oo 2.32 1.22 3.98 0.82 4.68 10.46 111.36 Coefficient of variation O/O 3.02 9.10 4.76 12.07 2.90 9.86 2.44 11.44 2.98 19.96 period. Whereas the standard deviation for the EMIT procedure is satisfactory for primidone phenytoin and carba-mazepine under these conditions (less than 5 pmol 1-1 in all instances) it is 7.85 pmol 1-1 for phenobarbitone and 111.36 pmol 1-1 for sodium valproate; the range of the coefficient of variation lies between 9.10% for phenobarbitone and 19.96% for sodium valproate which is unsatisfactory.On the other hand it can be seen that the results obtained from the between-batch analysis of QCRa for all five anticonvulsant drugs by the TDX assay under these conditions (shown in Table 6) are comparable to those obtained from the within-batch analyses (Table 4) and the between-batch analyses where a fresh calibration graph was generated for each run (Table 5). Tables 7-11 show the results of the analyses of the pooled patients' sera SL TL and TO for the five drugs by TDX and EMIT using the same calibration graph over a 30-d period. It can be seen that the variation in the results for the pooled sera containing drugs below the therapeutic range (SL) within the therapeutic range (TL) and above the therapeutic range (TO) is random for primidone and phenytoin (Tables 8 and 9).For phenobarbitone and carbamazepine the results of the analysis of the sera TL and TO by EMIT unlike the results obtained by TDX show a pronounced high bias particularly between day 18 and day 30 (Tables 7 and 10). For sodium valproate (Table 11) the variation in the results obtained for the sera SL TL and TO by TDX is apparently random but it can be seen that the results obtained for SL using the EMIT assay increase by a factor of 480 pmol l-1 between day 1 and day 30 and that the test ceases to give a result for serum TL after day 24 and for serum TO after day 3 because the apparent increase goes above the upper limit for the test quoted by the manufacturer (1145 pmol 1-1). The results in Tables 7-10 show that the between-batch precision for the TDX method is acceptable for phenobarbi-tone primidone phenytoin and carbamazepine for the sera TL and TO with a standard deviation of less than 5.5 pmol l-1 and a coefficient of variation of less than 7%.For the EMIT assay the corresponding values in particular for the coeffi-cient of variation are poor except for phenytoin (Table 9). In the sub-therapeutic range it can be seen from Tables 7-10 that both assays yield comparable results for the determination of phenobarbitone primidone phenytoin and carbam-azepine but that the mean of the 11 results from the determination of phenobarbitone by EMIT is considerably higher than the mean of the 11 results from the TDX assay (Table 7). For sodium valproate the between-batch precision of the TDX assay is less satisfactory in terms of standard deviation than for the other drugs analysed under these conditions.Although the coefficient of variation is less tha ANALYST MAY 1986 VOL. 111 521 ~ ~~ Table 7. Analysis of pooled patients’ sera for phenobarbitone by TDX and EMIT using the same calibration graph over a 30-d period Phenobarbitone/pmol 1 - 1 * Sub-therapeutic level (SL) Therapeutic level (TL) Toxic level (TO) Day 1 3 6 9 12 15 18 21 24 27 30 Mean . . . . . . . . . . Standarddeviation . . . . . . Coefficient of variation Yo . . * Mean of three determinations. TDX 16 15 13 13 13 14 14 13 13 14 15 13.91 1.04 7.51 EMIT 18 24 21 19 19 21 21 21 20 23 20 20.64 1.75 8.47 TDX 107 105 104 100 100 99 103 102 102 103 103 102.55 2.34 6.89 EMIT 105 116 121 119 112 115 128 130 127 119 132 120.36 8.30 6.89 TDX 224 227 221 222 215 233 225 229 219 228 226 224.45 5.03 2.24 EMIT 216 232 242 254 244 245 269 265 263 258 291 252.64 20.11 7.96 Table 8.Analysis of pooled patients’ sera for primidone UY TDX and EMIT using the same calibration graph over a 30-d period Primidonelpmoll-l* Sub-therapeutic level (SL) Therapeutic level (TL) Toxic level (TO) Day 1 6 9 12 15 18 21 24 27 30 Mean . . . . . . . . . . Standarddeviation . . . . . . Coefficient of variation Yo . . * Mean of three determinations. TDX 10 10 11 11 9 11 11 11 12 10 12 10.73 0.90 8.43 EMIT 13 12 11 9 10 10 10 10 9 11 10 10.45 1.21 11.61 TDX 47 47 48 48 46 47 46 46 48 44 48 46.82 1.25 2.67 EMIT 50 48 46 40 43 41 41 44 38 41 40 42.91 3.73 8.69 TDX 71 70 70 68 65 65 68 67 70 69 67 68.09 1.92 2.82 EMIT 69 68 66 55 56 56 57 59 54 58 55 59.36 5.55 9.36 ~ Table 9.Analysis of pooled patients’ sera for phenytoin by TDX and EMIT using the same calibration graph over a 30-d period Phenytoin/pmol 1 - 1 * Sub-therapeutic level (SL) Day TDX EMIT 1 19 20 3 16 20 6 19 20 9 18 22 12 18 19 15 18 17 18 16 18 21 16 18 24 16 19 27 17 20 30 16 17 Mean .. . . . . . . . . 17.18 19.09 Standarddeviation . . . . . . 1.25 1.51 Coefficient of variation Yo . . 7.28 7.93 * Mean of three determinations. Therapeutic level (TD) Toxic level (TO) TDX 49 44 48 48 46 48 47 47 47 46 43 46.64 1.80 3.87 EMIT 48 48 48 51 43 44 44 45 48 51 45 46.82 2.79 5.95 TDX 96 96 85 93 95 95 91 93 91 92 90 92.45 3.24 3.50 EMIT 96 96 89 97 82 89 85 85 90 93 85 89.73 5.20 5.79 5% for the analysis of all three sera the standard deviation ranges from 17.25 to 26.66 pmoll-1 (Table 11). However the performance of the EMIT assay in the analysis of serum SL is extremely poor with a standard deviation of 132.72 pmol 1-l and a coefficient of variation of 22.15% and no calculations of any kind could be made on the results of the analysis of the sera TL and TO by this method (Table 11).The results obtained from the within and between-batch analysis of quality control samples with freshly generated calibration graphs show that the performance of both immuno 522 ANALYST MAY 1986 VOL. 111 ~ Table 10. Analysis of pooled patients’ sera for carbamazepine by TDX and EMIT using the same calibration graph over a 30-d period Carbamazepinelymol 1-~~ ~ ~~ Sub-therapeutic level (SL) Therapeutic level (TL) Toxic level (TO) Day 1 3 6 9 12 15 18 21 24 27 30 Mean . . . . . . . . . . Standarddeviation . . . . . . Coefficient of variation % . . * Mean of three determinations.TDX 7 7 7 7 5 5 5 6 6 6 6 6.09 0.83 13.65 EMIT 8 8 7 7 8 8 7 7 7 8 8 7.55 0.52 6.92 TDX 22 21 19 19 19 19 18 19 19 19 19 19.27 0.90 4.69 EMIT 22 22 22 22 24 23 23 22 24 26 27 23.36 1.75 7.48 TDX 50 50 50 49 52 49 48 48 53 49 50 49.82 1.54 3.09 EMIT 55 55 54 50 69 75 60 63 75 89 93 67.09 14.50 21.61 Table 11. Analysis of pooled patients’ sera for sodium valproate by TDX and EMIT using the same calibration graph over a 30-d period Sodium valproate/ymol ] - I * Sub-therapeutic level (SL) Day 1 3 6 9 12 15 18 21 24 27 30 Mean . . . . . . . Standarddeviation .. . . . . Coefficient of variation % . . * Mean of three determinations. TDX 354 354 389 368 368 361 382 382 410 354 382 382.18 17.80 4.78 EMIT 397 474 516 54 1 541 62 1 643 643 569 733 877 599.18 132.72 22.15 Therapeutic level (TL) EMIT TDX -590 643 604 732 590 748 569 805 569 805 618 922 590 679 597 878 625 937 604 >1145 625 >1145 600.18 17.25 2.87 Toxic level (TO) TDX 923 923 923 854 854 902 95 1 902 902 923 909 914.82 26.66 2.91 EMIT 937 1139 >1145 >1145 >1145 >1145 > 1145 >1145 >1145 > 1145 >1145 assay procedures is satisfactory in the context of a clinical chemistry laboratory where the standard of a 95% confidence limit is widely accepted.The results of the analysis of the pooled patients’ sera however show that the calibration graphs generated by the TDX procedure largely remain stable over that period whereas the calibration graphs for EMIT exhibit a considerable drift in the determination of sodium valproate and to a lesser extent phenobarbitone and carba-mazepine marked by a high bias in the results for the determination of these drugs. The fact that the standard deviation and coefficient of variation for the between-batch analysis of quality control sera by the EMIT assay are acceptable when a fresh calibration graph is generated for each run is a further indication that the results shown in Tables 7 10 and 11 are primarily a function of the drift of the EMIT calibration graph under the conditions of the experiment, rather than cross-reactivity with other substances present in patients’ sera.On the other hand the relatively unsatisfactory performance of the TDX assay of sodium valproate com-pared with the performance of the other TDX assays examined in this study may be a function of antibody cross-reactivity with derivatives of fatty acids present in patients’ sera as the fluctuations in the results of the analysis of pooled patients’ sera by TDX shown in Table 11 are purely random in character. The two immunoassay procedures under investigation here gave comparable results for all five anticonvulsant drugs when used for the analysis of random patients’ samples over a wide range of concentration.Nevertheless recent reports of interferences in immunoassay procedures from substances present in patients’ sera have given rise to concern over the specificity of commercial kits for the assay of anticonvulsant drugs.41342 A major problem has been the report of interfer-ence with the EMIT assay of phenytoin by a metabolite of phenytoin or its derivatives.43.44 Problems have also been described although to a lesser extent with the TDX assay for phenytoin .41,45 The increasing development and use of mono-clonal antibodies will undoubtedly improve the specificity of assays for anticonvulsant drugs based on immune reactions in the future although there may be problems still to overcome. The qualities of simplicity and speed already make immuno-assays indispensable in the situation of ever-expanding need and increase in workload for the analysis of anticonvulsant drugs.References 1. 2. 3. 4. Pippenger C. E. Ther. Drug. Monit. 1979 1 3. Broughton P. M. G. Biochem. I . 1956 63 207. Wallace J. E. Biggs J. D. and Dahl E. V. Anal. Chem., 1965 37 10. Fuhr J. Arzneim-Forsch. 1964 14 74 ANALYST MAY 1986 VOL. 111 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Hamilton H. E. and Wallace J. E. in Pippenger C . E., Penry J. K. and Kutt H. Editors “Antiepileptic Drugs: Quantitative Analysis and Interpretation,” Raven Press New York 1978 pp. 175-183. Kupferberg H. J. in Pippenger C. E. Penry J. K. and Kutt, H. Editors “Antiepileptic Drugs Quantitative Analysis and Interpretation,” Raven Press New York 1978 pp.9-17. Grimmer G. Jacob J. and Schafer H. Arzneim.-Forsch., 1969 19 1287. MacGee J. Anal. Chem. 1970 42 421. Toseland P. A. Grove J. and Berry D. J. Clin. Chim. Acta, 1972 38 321. Chambers R. E. and Cooke M. J. Chromatogr. 1977 144, 257. Rambeck B. and Meijer J. W. A. Ther. Drug Monit. 1980, 2 385. Toseland P. A. and Wicks J. F. C. in Richens A. and Marks V. Editors “Therapeutic Drug Monitoring,” Churchill Livingstone London 1981 pp. 85-109. Goldberg V. D. Ratnaraj N. Elyas A. and Lascelles P. T., in Rose F. C. Editor “Research Progress in Epilepsy,” Pitman Press Bath 1983 pp. 462-483. Burke J . T. andThenot J. P. J. Chromatogr. 1985,340,199. Huisman J. W. Clin. Chim.Acta 1966 13 323. Atwell S. H. Green V. A. and Haney W. G. J. Pharm. Sci. 1975 64 806. Soldin S. J. and Hill J. G. Clin. Chem. 1976 22 856: Yalow R. S. and Berson S. A. Nature (London) 1959,184, 1648. Yalow R. S. and Berson S. A. J. Clin. Invest. 1960 39, 1157. Cook C. E. Christensen H. D. Amerson E. W. Kepler, J. A. Tallent C. R. and Taylor G. F. in Kellaway P. and Petersen I. S. Editors “Quantitative Analytic Studies in Epilepsy,” Raven Press New York 1976 pp. 39-58. Marks V. in Richens A. and Marks V. Editors “Thera-peutic Drug Monitoring,” Churchill Livingstone London, Rubenstein K. E. Schneider R. S. and Ullman E. F., Biochem. Biophys. Res. Commun. 1972 47 846. Berchou R. C. Lodi R. A. and Sherman J. A. Ann. Clin. Biochem. 1979 16 205. 1981 pp.155-182. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 523 Goldberg V. Ratnaraj N. Elyas A. and Lascelles P. T., Anal. Proc. 1981 18 313. McGregor A. R. Crookall-Greening J. O. Landon J. and Smith D. S. Clin. Chim. Acta 1978 83 161. Kamel R. S. McGregor A. R. Landon J. and Smith D. S., Clin. Chim. Acta 1978 89 93. Kamel R. S. Landon J. andsmith D. S. Clin. Chem. 1980, 26 1281. Wong R. C. Burd J. F. Carrico R. J. Buckler R. T., Thoma J. and Boguslaski R. C. Clin. Chem. 1979,25,686. Li T. M. Robertson S. P. Crouch T. H. Pahuski E. E., Bush G. A. and Hydo S. J. Clin. Chem. 1983 29 1628. Blake C. and Gould B. J. Analyst 1984 109 533. Hemmila I . Clin. Chem. 1985 31 359. Shaw W. and McHan J. Ther. Drug Monit. 1981 3 185. Kulpmann W. R . and Oellerich M. J. Clin. Chem. Clin. Biochem. 1981 19 249. Jolley M. E. J. Anal. Toxicol. 1981 5 236. Popelka S. R. Miller D. M. Holen J. T. and Kelso D. M., Clin. Chem. 1981 27 1198. Jolley M. E. Stroupe S. D. Schwenzer K. S. Wang C. J., Lu-Steffes M. Hill H. D. Popelka S. R. Holen J. T. and Kelso D. M. Clin. Chem. 1981 27 1575. Perrin F. J . Phys. Radium 1926 7 390. Dandliker W. B. Kelly R. J. Dandliker J. Farquhar J., and Levin J. Zmmunochemistry 1973 10 219. Lu-Steffes M. Pittluck G. W. Jolley M. E. Panas H. N., Olive D. L. Wang C.-H. J. Nystrom D. D. Keegan C. L., David T.-P. and Stroupe S. D. Clin. Chem. 1982,28,2278. Wang S. T. and Peter F. Clin. Chem. 1985 31 493. Haughey D. B. Matzke G. R. Halstenson C. E. and Keane W. F. J. Anal. Toxicol. 1984 8 106. Walson P. D. Brain Dev. 1985 7 116. Flachs H. and Rasmussen J. M. Clin. Chem. 1980,26,361. Aldwin L. and Kabakoff D. S. Clin. Chem. 1981 27 770. Oeltgen P. R. Clark T. Blouin R. A. and Shank W. A., Jr. Clin. Chem. 1984 30 1032. Paper A51430 Received November 20th 1985 AcceDt-ed December 12th. 198

ISSN:0003-2654    

DOI:10.1039/AN9861100517

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST MAY 1986 VOL. 111 525 An Iodine-I25 Radioimmunoassay for the Direct Detection of Benzodiazepines in Blood and Urine Colin P. Goddard A. Howard Stead Peter A. Mason Brian Law and Anthony C. Moffat" Central Research Establishment Home Office Forensic Science Service Aldermaston Reading Berkshire RG74PN UK and Margaret McBrien and Simon Cosby Northern Ireland Forensic Science Laboratory 180 Newtownbreda Road Belfast BT8 4QR UK A radioimmunoassay (RIA) for the direct detection of benzodiazepines in blood and urine is described. It is based on a commercially available antiserum and an easily synthesised radio-iodinated derivative of clonazepam that allows the use of relatively simple gamma-counting procedures. The assay can detect low therapeutic levels of all of the benzodiazepines currently available in the UK in 50-$ samples of blood and urine (1-50 ng ml-I depending on the drug); no prior sample preparation is required.It is inexpensive rapid, simple to perform and is broadly specific for the benzodiazepine class of drugs. The assay offers a most suitable means of screening large numbers of samples of forensic interest for the presence of the benzodi azepi nes. Keywords Radioimmunoassay; benzodiazepine detection; blood; urine The benzodiazepines are a group of chemically related drugs used mainly as hypnotics and sedatives. Forensic interest in these compounds is due primarily to their frequent occurrence in drugsldriving cases,1-3 i.e. those cases where a drug is suspected of having made a significant contribution to the impairment of driving ability.A large number of drugddriving cases are seen by forensic science laboratories each year each case requiring an analysis for unknown drugs and metabolites in very small volumes of blood. With the benzodiazepines in particular the analyst is faced with further problems arising from the fact that there is a steadily increasing number of different compounds available (20 in the UK alone4) and because commercial companies are producing more potent drugs all the time the newer com-pounds are often present at very low concentrations. A rapid and simple screening assay for benzodiazepines that is sensitive and broadly specific to this group of drugs is therefore desirable. Methods currently used to detect benzodiazepines include high-performance liquid chromatography (HPLC) ,5,6 gas -liquid chromatography with electron-capture detection (GLC - ECD)7-9 and radioimmunoassays (RIAp14 and radiorecep-tor assays (RRA)15,16 using tritiated labels.Use of HPLC and GLC for routine screening is relatively time consuming. In comparison immunoassay techniques offer many advantages as methods for the efficient screening of large numbers of biological samples for the presence of chemically related drugs. Unfortunately of those RIAs available for the detection of benzodiazepines most are specific to particular drugs and assays of this type are in general of limited use for drug screening. Radioreceptor assays and screening RIAs capable of detecting a broad range of benzodiazepines are convenient to use but published methods are based on tritiated labels and suffer the disadvantages associated with liquid scintillation counting viz.the need for long counting times the need for prior sample extraction the use of expensive scintillant and problems associated with quenching when coloured biological samples are assayed. Two enzyme immunoassay kits one for benzodiazepines in serum and one for benzodiazepine metab-* Present address Home Office Forensic Science Laboratory, Hinchingbrooke Park Huntingdon PE18 8NP. Crown Copyright 1986. olites in urine are commercially available [Syva (UK), Maidenhead Berkshire] but neither has the sensitivity required to detect low therapeutic levels of all available benzodiazepines. In addition neither can cope with the haemolysed blood samples often encountered in forensic cases.The screening assay described in this paper is based on an iodine-125 labelled derivative of clonazeparn. It is sensitive, broadly specific and allows the direct analysis of forensic samples. Experimental Materials and Equipment All chemicals unless specified otherwise were obtained from BDH Chemicals Poole Dorset . Phosphate buffer (0.1 M pH 7.4) containing 0.2% mlV of bovine y-globulin (Cohn fraction I1 from Sigma Chemical, Poole Dorset) and 0.01% mlV of sodium azide was used throughout the assay. The antisera were obtained from Emit TOX serum benzodiazepine and Emit DAU benzodiazepine metabolite kits [Syva (UK)]. The Emit TOX antiserum was diluted 1 + 299 and the Emit DAU antiserum 1 + 999 with assay buffer immediately before each run of the assay.These dilutions give approximately 50% binding with the amount of radiolabel added to each assay tube. The maximum binding of label to each antiserum was approximately 75-80%. The radiolabelled benzodiazepine derivative [7-1251]-iodoclonazepam (specific activity 5.6 TBq mmol-1; 12.7 MBq pg-1) was prepared from the 7-amino derivative of clonazepam via a diazonium intermediate as previously described17 and stored in methanol at 4 "C. It was diluted with assay buffer to give approximately 10000 counts min-l (550 Bq; 0.099 ng) per 100 p1. Standard solutions of diazepam were prepared in synthetic urine at concentrations of 0 1 2 3 5 10 and 20 ng ml-1. The synthetic urine was prepared by adding 22 g of urea 1.8 g of NaH2P04.2H20 1.1 g of Na2HP04 8.25 g of NaCl 5.2 g of KCl 1.5 g of creatinine and 0.1 g of NaN3 to distilled water and making the total volume up to 1 1 (modified from reference 18).Swine serum was obtained from Flow Laboratories Ltd., Irvine Ayrshire 526 ANALYST MAY 1986 VOL. 111 synthetic urine samples containing 1.5 and 5 ng ml-1 of diazepam. The effects of various urine and blood preservatives on the assay were studied by adding the following preservative tablets (1) to known blank urine samples (2.5 ml) phenyl-mercury(I1) nitrate and sodium fluoride (50 + 100 mg), sodium fluoride and sodium sulphate (300 + 300 mg) and sodium azide (50 mg); (2) to known blank blood samples (1 ml) sodium fluoride and potassium oxalate (37.5 + 18.7 mg) and sodium nitrite (25 mg).The final concentrations of the added preservatives were approximately 5 (urine) and 2.5 (blood) times greater than those recommended for forensic use. The assay was used to measure serum levels of diazepam following oral ingestion of single therapeutic doses of the drug (10 mg) by two volunteers. Blood samples were taken before drug ingestion and then at intervals over a period of 36 h afterwards. The Northern Ireland Forensic Science Laboratory (NIFSL) conducted a trial to compare the results obtained using the described RIA with those obtained by conventional methods. The NIFSL routinely uses GLC with ECD GLC with NPD and HPLC methods for the detection and quanti-fication of benzodiazepines in case samples. The trial involved using chromatographic techniques in parallel with the RIA method on each of 80 samples (including clinical driving and post-mortem cases) where benzodiazepines were suspected.The levels of both parent drug and any major active metabolites were obtained by the chromatographic methods, and these were compared with the total benzodiazepine level as measured by RIA. Polyethylene glycol (PEG M 8000) was obtained from Sigma Chemical and was used to prepare a 27.5% mlV solution in distilled water (550 g of PEG plus 1525 ml of water). Disposable polypropylene microcentrifuge tubes (plastic point) were obtained from Alpha Laboratories Eastleigh, Hampshire. Gamma-counting was performed using an NE 1600 counter (Nuclear Enterprises Beenham Berkshire) which had an efficiency of approximately 50% for iodine-125.Method The dilution for the two commercially available antibodies was determined by plotting dilution curves for each against the purified [7-12~I]iodoclonazepam and measuring the dilu-tion needed to bind approximately 50% of the total radio-activity. Cross-reactivities of a small number of benzodiazepines (chlordiazepoxide diazepam flurazepam lorazepam nitra-zepam oxazepam temazepam and triazolam) were obtained using both the TOX and the DAU antisera at the optimum dilution by comparing the calibration graph of each with that of diazepam. The results indicated that the TOX antiserum was more suitable for a general screening assay owing to a narrower range of cross-reactivities; all further work was performed using only this antiserum.The cross-reactivities of all the benzodiazepines currently available in the UK, together with some of their major active and inactive metabolites were then obtained. The concentration of each drug required to give a 50% depression of binding was measured and expressed relative to the value for diazepam. The extent of cross-reaction of several commonly encoun-tered but structurally unrelated drugs was also measured and compared with the cross-reactivity of diazepam. The depres-sion of binding was measured using solutions containing 100 pg ml-1 of the drug unless stated otherwise. The drugs included acetylsalicylic acid (400 pg ml-I) amitriptyline, amphetamine bromocryptine butriptyline caffeine, codeine ephedrine glutethimide imipramine isoprenaline, lysergic acid diethylamide methadone methylphenobar-bitone mianserin morphine nicotine paracetamol (250 pg ml-I) phenytoin and protriptyline.As diazepam is one of the strongest cross-reacting benzo-diazepines and is also one of the most commonly prescribed, it was used to prepare solutions for the standard curve. Assays were performed in duplicate using the following conditions. Into each microcentrifuge tube were pipetted 100 p1 each of sample radiolabel and antiserum. The 100 p1 of sample consisted of (for urine samples and standards) urine or standard (50 pl) and swine serum (50 pl) or (for blood samples) blood (50 p1) together with synthetic urine (50 pl). This procedure ensured that each tube contained approxi-mately the same amounts of salt protein and water.The tubes were then capped vortexed and incubated at room temperature for at least 1 h. Equilibration was attained after 60 min and was stable for up to 24 h after that time. PEG solution (500 pl) was then added and the tubes were re-capped vortexed thoroughly and then centrifuged at 9000 g for 3.5 min. The supernatant was removed by aspiration and the tubes containing the residual bound fraction were each counted for 60 s in the gamma counter. Whole blood samples were assayed neat and after dilution by factors of 10 and 100 with swine serum. Urine samples were assayed neat and after dilution by factors of 10 100 and 1000 with synthetic urine. Cut-off levels for the assay were determined by the analysis of 100 unpreserved blank urine and blood samples obtained from volunteers not taking any drugs.The conditions of the blood samples varied from fresh unhaemolysed to haemo-lysed/putrefied. Intra- and inter-assay coefficients of variation for the assay were determined by repeated analyses of Results and Discussion The optimum antibody dilutions (ie. those giving 50% binding of the added radiolabel) were measured as 1 + 299 for the Emit TOX antiserum and 1 + 999 for the Emit DAU and these dilutions were used in all further work. Table 1. Cross-reactivities of eight common benzodiazepines relative to diazepam. Determined at 50% depression of binding (equivalent to 2.5 ng ml-1 of diazepam) Relative reactivity Benzodiazepine Emit TOX Emit DAU Diazepam .. . . . . . 1 .O 1 .o Triazolam . . . . . . . 1.4 3.3 Flurazepam . . . . . . 2.8 2.5 Temazepam . . . . . . 4.8 12 Nitrazepam . . . . . . 6.4 8.8 Oxazepam . . . . . . . . 11 38 Lorazepam . . . . . . 36 107 Chlordiazepoxide . . . . 52 115 0 10 20 Diazepam concentrationhg ml-1 Fig. 1. Typical calibration graph for diazepam obtained using radioirnmunoassa ANALYST MAY 1986 VOL. 111 527 Table 2. Cross-reactivities of benzodiazepines and their metabolites with the Emit TOX antiserum Concentration reauired Therapeutic Compound Prazepam . . . . . . . . Diazepam . . . . . . Pinazepam . . . . . . Alprazolam . . . . . . Triazolam . . . . . . N-Desmethyldiazepam . Flurazepam . . . . . . Midazolam . . . . . . Temazepam . . . . . . Tetrazepam .. . . . . Clorazepate . . . . . . Nitrazepam . . . . . . Clobazam . . . . . . Flunitrazepam . . . . . . Medazepam . . . . . . Oxazepam . . . . . . Lormetazepam . . . . . . Ketazolam . . . . . . Loprazolam . . . . . . Clonazepam . . . . . . Lorazepam . . . . . . Desalkylflurazepam . . . . Bromazepam . . . . . . Chlordiazepoxide . . . . Desmethylclobazam . . . . Demoxepam . . . . . . Desmethylchlordiazepoxide Prazepam benzophenone . . Diazepam benzophenone . . 7-Acetylaminonitrazepam . . Clozapine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . * .. . . . . . . . . . . . . . . . . . . . . . . . . . . . Relative activity* 0.60 1 .oo 1.20 1.20 1.40 2.40 2.80 3.20 4.80 5.6 6.0 6.4 10.8 10.8 10.8 11.2 11.2 14.8 21.6 32 36 44 52 52 52 >loo > 100 >loo > 100 >>loo >>loo for 50% depresiion of bindinghg ml-1.5 2.5 3.0 3 .O 3.5 6.0 7.0 8.0 12 14 15 16 27 27 27 28 28 37 54 80 90 110 130 130 130 >250 >250 >250 >>250 >250 ~ 2 5 0 rangel ng ml-lf 8-40 50-2000 --5-25 100-1 500 1-5 -360-850 300-1500 20-150 300-900 10-160 500-2000 -5-20 ---10-100 50-240 30-90 80-150 1000-3000 2500-3500 ------* Determined at 50% depression of binding (equivalent to 2.5 ng ml-I of diazepam).t Approximate range of blood concentrations at the steady state wherever possible; data from references 19 and 20. 100 8 & 2 50 0 I I I 1 102 104 +/ Drug concentrationhg ml-1 Fig. 2. Full range of cross-reactivities for benzodiazepines. B = bound activity; B = bound activity when no unlabelled drug is present. A Prazepam B bromazepam The relative cross-reactivities of eight common benzodiaz-epines were measured for both the Emit TOX and the Emit DAU antisera. The results (Table 1) indicate that the two antisera are very similar but that the spread of cross-reactivity values is less with the Emit TOX antiserum. The most important feature of any screening assay is that the sensitivity towards the least cross-reacting drugs is sufficient to detect them at the required levels.The Emit TOX antiserum is more sensitive towards drugs of lower cross-reactivity and conse-quently it was selected for the development of the general screening assay; this antiserum was used in all further work. A typical assay calibration graph for diazepam is shown in Fig. 1. The relative cross-reactivities of all available benzodiaz-epines together with a number of their active and inactive metabolites were measured using the Emit TOX antiserum. The results are given in Table 2. The full range of cross-reactivities of the 20 benzodiazepines available in the UK,4 from prazepam (the strongest) to bromazepam (the weakest), is shown in Fig. 2.The range of cross-reactivities demon-strated by these results shows that all of the parent benzo-diazepines bind relatively strongly to the antiserum their desalkyl metabolites less so and compounds in which the diazepine ring structure is broken (the substituted benzo-phenones) exhibit greatly decreased binding. Further the range suggests that the assay is sufficiently sensitive to detect all available benzodiazepines. Even low-dose drugs such as alprazolam prazepam and triazolam should be easily detected at therapeutic levels because their cross-reactivities are approximately the same as that of diazepam. Ironically, because of its poor cross-reactivity one of the drugs that may be most difficult to detect at low therapeutic levels is clonazepam the drug from which the radiolabel is derived.Other drugs that might prove difficult to detect in similar circumstances include flunitrazepam and flurazepam. However even with these compounds therapeutic levels will usually be above the assay cut-off values and therefore will be detectable. Of the 20 common non-benzodiazepine drugs that were tested for their cross-reactivity in the assay only morphine at a concentration of 100 pg ml-1 caused a depression in binding greater than the positivehegative cut-off value of the assay (see below). This concentration is much higher than normal fatal levels of the drug. The analysis of 100 unpreserved blank blood samples gave a mean level of background cross-reactivity of 0.28 k 0.18 (S.D.) ng ml-l. Similar analyses of 100 blank urine samples gave a mean value of -0.02 k 0.16 ng ml-1.The mean plu 528 ANALYST MAY 1986 VOL. 111 Table 3. Comparison of RIA and GLC/HPLC analyses of blood and urine samples from clinical driving and post-mortem cases involving benzodiazepines GLC or HPLC resulthg ml-1 Drug Diazepam . . . . . . . . . . . . . Lorazeparn . . . . . . . . . . . . . . Flurazepam . . . . . . . . . . . . . . . . Temazepa-m . . . . . . . . . . . . . . . . Lormetazepam . . . . . . . . . . . . . . . . Chlordiazepoxide . . . . . . . . . . . . . . Bromazepam . . . . . . . . . . . . . . . . Desmethyldiazepam/flurazepam . . . . . . . . Diazepam/desmethyldiazepam/ temazepam . . . . . . . . . . . . . . . . Sample [blood (b) , urine (u)] b b b b b b b b b b b b b b b U U b b b b b b b b b b Parent 30 70 200 420 260 460 460 1250 1000 70 1820 130 160 320 100 1400 450 310 750 340 2180 40 240 2600 400/1000 -300/440/3 10 Met a bolit e * --340 130 90 670 ------+ 390 Diazepam equivalent (total) 1-30 70 142 254 420 298 460 739 1250 1000 70 1820 3.6 4.4 8.9 2.8 38.9 >161 120 156 71 454 3.6 4.6 50 524 548 RIA result/ ng ml-* 98 147 210 250 270 350 570 630 820 850 1000 1230 2.3 8.0 10.4 23 41 1630 2300 120 240 520 3.8 27 32 430 540 * Metabolite levels were not determined unless stated otherwise; - no or only trace amounts of parent or metabolite detected; + metabolite t Corrected for cross-reaction of parent drug and/or its metabolite.detected but not quantified. three standard deviations is 0.82 ng ml-1 for blood and 0.46 ng ml-1 for urine. A single positivehegative cut-off level of 1.0 ng ml-1 was set for both blood and urine samples for simplicity ensuring a >99.7%0 probability of obtaining a true positive result. Coefficients of variation for 1.5 and 5 ng ml-1 of diazepam in synthetic urine were 6.7 and 7.9% intra-assay (n = 20) and 6.4 and 11.4% inter-assay (n = 20) respectively. Of the blood and urine preservatives tested the only one to cause a measurable depression in binding of the radiolabel to the antiserum was the sodium fluoride - sodium sulphate (300 + 300 mg) preservative when added to urine at concentrations five times the normal.However as urinary benzodiazepine concentrations are generally much higher than those seen in blood and as the measured background levels in urine caused by this preservative are only equivalent to approximately 24 ng ml-1 of diazepam it is unlikely that the interference will create any serious analytical problem. Despite this urinary levels of less than 30 ng ml-1 should be treated with caution unless a complementary blood level is available. The results of RIA analysis of blood samples obtained from two volunteers given a single oral dose of diazepam are shown in Fig. 3; peak levels of diazepam in both subjects are over 300 ng ml-1. Most points shown in Fig. 3 were obtained following dilution of samples by a factor of 10 or 100 (shortly after ingestion) demonstrating that the assay is capable of measur-ing therapeutic levels of diazepam in blood with at least an order of magnitude to spare.Because of this assay results must be interpreted with great care as a positive result might easily be obtained in a sample taken long after any pharmaco-logical effects have ceased. In practice it is recommended that the assay is performed both on neat samples of blood and after dilution of samples by factors of 10 and 100 and the higher drug levels found in urine may require sample dilution by factors of 10 100 and 1000 in order to obtain results that can be measured accurately from the calibration graph. Of the 80 samples tested by the NIFSL 55 were positive for benzodiazepines by RIA; 19 gave results (mean 28 ng ml-1 diazepam equivalents; range 1.3-180 ng ml-1) that could not be confirmed independently by GLC or HPLC.The remain-ing 36 samples were positive both by GLC or HPLC and by RIA. Nine of these 36 results were rejected for the compari-son study either owing to lack of data on metabolite concentrations or because the parent drug was only detected in trace amounts by GC or HPLC. Table 3 summarises the data obtained from the remaining 27 cases. The range of drugs identified in the case samples is fairly typical of that occurring in routine toxicology. Parent drug and/or metabolite levels as measured by GLC or HPLC are presented. These results ar ANALYST MAY 1986 VOL. 111 529 1.5 3 4.5 Time after ingestiodh Fig.3. single 10-mg oral dose of the drug Serum concentrations of diazepam in two volunteers given a also expressed as total diazepam equivalents calculated from the cross-reactivities of the respective compounds to allow direct comparison with the RIA results. A statistical comparison of the RIA and GLC or HPLC methods was performed using diazepam and lorazepam as examples. Diazepam provides an example typical of those benzodiazepines that undergo complex metabolism giving rise to one or more active metabolites; its major metabolite in blood is desmethyldiazepam. Lorazepam provides an example of a benzodiazepine that is not extensively metabolised and for which active metabolites have not been identified. Good correlations exist between results from the two methods if the parent drug and its major active metabolite in blood are measured as with diazepam ( r = 0 .7 6 ; ~ <0.005 y = 0.506~ + 259 n = 12). Similarly where the drug has no active metabolites and only the parent compound is measured as with lorazepam there is again a relatively high degree of correlation ( r = 0.85 p C0.02 y = 0.0849~ + 6.99 n = 5). In both instances drug levels measured by RIA are generally higher than those obtained by the chromatographic methods. This is not unexpected as it is likely that the RIA will detect those chemically related metabolites of the benzodiazepines that the more specific GLC and HPLC techniques will not. The only results to show large differences between the two methods were observed with flurazepam.RIA results were much higher than chromatographic results even after allow-ing for the major metabolite desalkylflurazepam. It is possible that other metabolites that cross-react well are present in sufficient amounts to raise the level obtained by RIA. Metabolites such as desethylflurazepam didesethylfluraze-pam and N-1-hydroxyethylflurazepam were not measured by chromatography but should cross-react and may be present in significant concentrations. The combination of the Emit TOX antiserum and the [7-12~I]iodoclonazepam label provides an assay that is specific for the benzodiazepines as a group making it ideal as a general screen for this class of drugs. The examination of case samples has shown that by comparison with the less sensitive chromatographic techniques used the RIA gives no false negative results.There is overall good agreement between the results obtained by RIA and those obtained using these chromatographic methods. However the broad range of cross-reactivities and the interaction of active metabolites with the assay means that quantification based on RIA is of little value unless the identity of the drug is known; even low levels detected in screening may represent significant amounts of drugs such as lorazepam and bromazepam. A positive result can only be a semi-quantitative indication of the presence of a drug and even if a single compound is known to be present an accurate quantification of concentrations should not be attempted by this method unless a calibration graph for the specific drug is prepared.All positive results should be confirmed and the drug identified by more specific procedures such as GLC HPLC or mass spectrometry. Conclusions In conclusion the assay is easy to perform quick to set up and run inexpensive and very reliable. Intra- and inter-assay results confirm its reproducibility. The sensitivity and group specificity of the assay allow low therapeutic levels of all of the benzodiazepines currently available in the UK to be detected in very small volumes (50 p1) of sample without the need for any prior preparation procedures. The antiserum used is commercially available. The assay has been used successfully for testing for the presence of benzodiazepines in a large number of cases submitted for forensic analysis and should prove extremely useful to both the clinical and the forensic toxicologist.1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. References Garriott J. C. DiMaio V. J. M. Zumwelt R. E. andPetty, C. S. J. Forensic Sci. 1977 22 383. Robinson T. A. J. Forensic Sci. SOC. 1979 19 237. Taylor J. F. in Goldberg L. Editor “Alcohol Drugs and Traffic Safety,” Almqvist and Wiksell Stockholm 1981, p. 478. “MIMS Monthly Index of Medical Specialities,” Haymarket Press London August 1984. Osselton M. D. Hammond M. D. and Twitchett P. J., J. Pharm. Pharmacol. 1977 29 460. Mehta A. C. Talanta 1984 31 1. Greenblatt D. J. Divoll M. Moschitto L. J. and Shader, R. I. J. Chromatogr. 1981 225 202. Zlatkis A. and Poole C. F. Editors “Electron Capture, Theory and Practice in Chromatography,” Elsevier Amster-dam 1981 p. 306. Douse J. M. F. J. Chromatogr. 1984 301 137. Gelbke H. P. Schlicht H. J. and Schmidt G. Arch. Toxicol. 1977 38 295. Rutterford M. G. and Smith R. N. J. Pharm. Pharmacol., 1980 32 449. Dixon R. in Langone J. J. and Vunakis H. V. Editors, “Methods in Enzymology Volume 84 Immunochemical Tech-niques Part D Selected Immunoassays,” Academic Press, New York and London 1982 p. 490. Aderjan R. and Schmidt G. 2. Rechtsmed. 1979 83 191. Robinson K. Rutterford M. G. and Smith R. N. J. Pharm. Pharmacol. 1980 32,773. Jochemsen R. Horbach G. J. M. J. and Breimer D. D., Res. Commun. Chem. Pathol. Pharmacol. 1982 35 259. Lund J. Scand. J. Clin. Lab. Invest. 1981 41 275. Goddard C. P. Law B. Mason P. A. and Stead A. H., J. Labelled Cpds. Radiopharm. in the press. Rodgers R. Crowl C. P. Eimstad W. M. Hu M. W. Kam, J. K. Ronald R. C. Rowley G. L. and Ullman E. F. Clin. Chem. 1978,24 95. Schutz H. “Benzodiazepines A Handbook,” Springer-Verlag Berlin 1982. Stead A. H. and Moffat A. C. Human Toxicol. 1983 3, 437. Paper A51393 Received October 31st 1985 Accepted December 9th I98

ISSN:0003-2654    

DOI:10.1039/AN9861100525

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST MAY 1986 VOL. 111 53 1 Indirect Micro-scale Method for the Determination of Desferrioxamine and its Aluminium and Iron Chelated Forms in Biological Samples by Atomic Absorption Spectrometry with Electrothermal Atom isation Pierre Allain Yves Mauras Guilene Beaudeau and Philippe Hingouet La bo ra to ire de Pha rm acolog ie e t Toxic0 log ie Cen tre H ospita lie r Reg iona I e t U n ive rsita ire d 'A ng ers, 49040 Angers Cedex France An indirect micro-scale method is described for the determination of desferrioxamine (DFA) itself and in its iron and aluminium complexed forms (FeA and AIA) in blood plasma and urine. AIA FeA and DFA after its conversion into FeA with an excess of iron are selectively extracted with benzyl alcohol and iron and aluminium are determined in the benzyl alcohol extract by electrothermal atomic absorption spectrometry using partition pyrolytically coated graphite tubes with a cuvette to prevent benzyl alcohol from spreading to the tube extremities.The specificity of the method was assessed and the sensitivity is sufficient for the determination of DFA FeA (44 pg 1-1) and AIA in blood plasma and urine after the administration of desferrioxamine as Desferal. However because of the wide range of concentrations observed in biological samples (from 1 to 300 rng I-' of DFA) the use of two calibration graphs and sometimes a preliminary dilution of the high-concentration samples are required. The use of this indirect micro-scale method for more than 1000 assays without particular problems has allowed the study of DFA pharmacokinetics.Keywords Desferrioxamine determination; aluminium and iron complexes; atomic absorption spectrometry; electrothermal atomisation; biological samples Desferrioxamine (DFA) a well known iron chelator has been used as an aluminium chelator in dialysed patients following the work of Ackrill et al.1 Many papers have reported the efficiency of DFA for Fe and A1 removal in patients but the pharmacokinetics of DFA itself and of its Fe and A1 complexed forms which are referred to as ferrioxamine (FeA) and aluminoxamine (AlA) respectively are not yet well known and there is as yet no suitable method for their determination. The spectrophotometric method of Meyer-Brunot and Keberle2 which has a detection limit of 5 mg 1-1, lacks sensitivity and cannot be adapted to the determination of A1A.High-performance liquid chromatography is a promising method but the results obtained by Cramer et aZ.3 do not allow the determination of DFA A1A and FeA in biological samples although an application of this method to biological samples has recently been described.4 In this paper a micro-scale method consisting in the specific extraction of DFA in complexed forms i.e. FeA and A1A with benzyl alcohol (as used by Meyer-Brunot and Keberle2) is described but instead of using spectrophotometric detec-tion for FeA the concentration of Fe or A1 is measured by electrothermal atomic absorption spectrometry. This indirect method has been adapted to the determination of DFA AIA and FeA at therapeutic concentrations in blood plasma or urine.Experimental The instrument used was a Varian Model 975 atomic absorption spectrometer with a GTA 95 graphite furnace and an autosampler. Partition pyrolytically coated graphite tubes with a cuvette (Varian 63.100008.00) were always used to prevent the spreading of benzyl alcohol to the tube extremi-ties. To improve the benzyl alcohol delivery in the graphite tube 5 1-11 of ethanol taken from the modifier beaker of the programmable sample dispenser were automatically dis-pensed with the blank standards and samples. The main instrument settings are indicated in Table 1. Background correction by a deuterium lamp was always used. Reagents For the conversion of DFA into FeA and AIA iron(II1) chloride solution (0.01 M) and aluminium chloride solution (0.02 M) respectively were used.Instead of using solid sodium chloride as used by Meyer-Brunot and Keberle,* a saturated solution of ammonium nitrate (Prolabo) dissolved in Merck buffer (pH S) giving a solution with a final pH of about 7 was used. DFA as the methanesulphate C26H52N6011S (relative molecular mass 656) was obtained from Ciba Laboratories. Table 1. Main instrument settings for the determination of FeA and AIA Fe (FeA) determination at A1 (AIA) determination at Parameter Low levels High levels Low levels High levels Wavelength/nm . . . . 248.3 248.3 309.3 309.3 Furnace temperature/"C: Charring . . . . . . 700 700 1000 1000 benzylalcohol/yl . . . . 10 5 30 10 1min-l . . . . . . 0 0.5 0 0 Atomisation . . . . 2400 2400 2500 2500 Injected volume of Argon purge gas flow-rate 532 ANALYST MAY 1986 VOL.111 Eppendorf polypropylene microtubes of 1.5 ml were used for the extraction and polypropylene solvent-resistant cups for the autosampler. Procedure For each sample of plasma or urine to be analysed the following two methods were used. Determination of A1A and FeA already present in biological samples To 100 pl of blood plasma or urine in an Eppendorf microtube, 200 p1 of saturated NH4N03 (pH 7) and 750 pl of benzyl alcohol were added. The tube was shaken for 60 s on a Thermolyne apparatus and centrifuged at 10 000 g for 5 min. Then 500 pl of benzyl alcohol were transferred into a cup of the autosampler for the determination of A1 and Fe by graphite furnace atomic absorption spectrometry.Determination of FeA and DFA The same procedure was used except that 25 pl of 0.01 M iron(II1) chloride were added to convert DFA into FeA [lo0 p1 of plasma 25 pl of 0.01 M iron(II1) chloride 200 p1 of NH4N03 and 750 pl of benzyl alcohol]. Standards for the preparation of calibration graphs were prepared by adding to normal blood plasma or urine known amounts of DFA (1.25, 2.5 5 and 10 mg 1-1 low concentrations graph and 12.5 25 and 50 mg 1-1 high concentrations graph). DFA was then converted into FeA by the addition of 25 pl of 0.01 M iron(II1) chloride solution or into AlA by the addition of 25 pl of 0.02 M aluminium chloride solution. The extraction procedure was used as described. The concentrations of FeA and A1A already present in biological samples were measured directly but the concentra-tion of DFA was calculated by taking the difference between the FeA + DFA concentration (after FeCI3 addition) and the FeA concentration.Results Fig. 1 shows the absorbance signals obtained for DFA after its conversion into FeA or AlA by increasing the amounts of FeCI3 or AIC13 added in a constant volume of 25 pl to a 100-p1 blood plasma sample containing 100 mg 1-1 (152.4 prnol l-1) of DFA. Concentrations of 0.01 M of Fe and 0.02 M of Al, sufficient to obtain maximum signals were used for the I I I I Fig. 2 shows typical calibration graphs obtained after the addition of DFA to normal blood plasma at low concentra-tions (2.5,5 and 10 mg 1-1) and high concentrations (12.5,25, 50 and 100 mg 1-1) and its conversion into FeA or AIA subsequently extracted in benzyl alcohol.The absorbances of low and high concentrations of FeA are approximately the same (Fig. 2) because a lower volume of benzyl alcohol was injected and a flow of argon gas was used to reduce the sensitivity for high concentrations (Table 1). For FeA measurements a curvature of the two calibration graphs is observed above 25 and 2.5 mg 1-1 respectively whereas for AIA measurements the response is almost linear up to 10 mg 1-1. We tried to cover the largest scale of DFA and FeA concentrations found in biological samples using the same extraction procedure but under two different sets of condi-tions (Table 1). However when the DFA or FeA concentra-tions were higher than 25 mg 1-1 a preliminary dilution (1 + 2 to 1 + 10) of the biological sample was necessary.The detection limit calculated as equivalent in concentra-tion to twice the standard deviation of the absorbance signal of a sample of blood plasma without DFA was equivalent to 95 pg 1-1 of DFA. The sensitivity defined as the concentration that produces a 0.0044 absorbance for iron was equivalent to 44 pg 1-1 or 0.07 pmol l-1 of DFA. The reproducibility of the method tested by ten successive assays of the same sample, was 2.5 and 8% for DFA concentrations of 12.5 and 1.25 mg 1-1 respectively. The recovery studied by adding 1.25, 12.5 and 25 mg 1-1 of DFA to different biological samples and measuring their concentrations with respect to the calibration graphs gave an average very close to 100%.The DFA concentrations measured in the blood plasma of patients extended from 1 to 300 mg 1-1 (1.5 to 457 pmol 1-I), depending on the doses given (10-80 mg kg-1 body mass) the route of administration (intravenous or intramuscular) and the time of blood sampling. Discussion DFA is more soluble in water than in benzyl alcohol but FeA and AlA are more soluble in benzyl alcohol than in water. This indirect micro-scale method based on the selective extraction of FeA and A1A with benzyl alcohol and the determination of the metal extracted by electrothermal atomic absorption spectrometry is applicable only if Fe3+ and Al3+ added to the sample are not found in the benzyl alcohol phase in the absence of DFA. After addition of 25 p1 of 0.01 M FeC13 or 0.02 M AlC13 0.6 al 2 0.4 e z Q a 0.2 0 25 50 100 Graph C 0.0025 0.005 0.01 0.02 Concentration of AICIJ or FeCIdM Fig.1. Effect of 25 p1 of different concentrations of A1Cl3 and FeC13 on the conversion of 100 mg 1-1 (152 pmol l-1) of DFA into A FeA and B AlA under the conditions described in the text 1 I 1 J Graphs A and B 2.5 5 10 DFA concentration in blood plasma/mg I-' Fig. 2. Calibration graphs for A AIA; and B and C Fe ANALYST MAY 1986 VOL. 11 1 533 solutions to 100 pl of blood plasma or urine without DFA, FeA and AlA we found at an acidic pH of about 2 very small amounts of Fe and A1 in the benzyl alcohol phase but not at pH 5 or higher. Therefore to guarantee optimum conditions of pH 200 p1 of a buffered solution of NH4N03 at pH 7 were added.Under these conditions all the blanks (plasma urine without DFA AlA and FeA) gave similar absorbance values to those obtained with benzyl alcohol. Moreover we checked that substances such as citrate ascorbate and EDTANa2, added to the sample in high concentrations (250 mg 1-I) did not modify the blank value. Hence to our knowledge DFA is the only molecule present in blood plasma and urine able to transfer Fe and A1 into the benzyl alcohol phase and the method described can be considered as specific. In addition to Fe and Al DFA can form a soluble complex in benzyl alcohol with vanadium. However when present at only very low levels in biological samples vanadium does not interfere in the determination of DFA. However if necessary it is also possible to measure indirectly the concentration of the vanadium DFA complex.It is interesting that human serum transferrin also binds vanadium.5 This indirect micro-scale method with a sensitivity of 44 yg 1-1 for DFA is about 100 times more sensitive than the spectrophotometric macro-scale method of Meyer-Brunot and Keberle2 which is based on an FeA molecular determina-tion at 430 nm in which the smallest amount that can be determined with certainty is 5000 pg 1-1. The sensitivity of this indirect micro-scale method could be improved by increasing the volume of benzyl alcohol injected into the graphite furnace or by automatic multiple injections. In contrast the sensitivity could be decreased by reducing the amount of biological samples by a preliminary dilution by increasing the volume of benzyl alcohol used for extraction and by increasing the argon gas flow-rate during atomisation.Therefore the proposed analytical conditions are a compromise trying to cover the large range of DFA and FeA concentrations found in biological samples after DFA (Desferal) administration. It is necessary to emphasise that the use of graphite tubes with a cuvette is essential to prevent benzyl alcohol from spreading to the tube extremities giving very high background signals and non-reproducible specific signals a problem encountered with other organic solvents.6 We have been using this indirect method for 1 year and more than 1000 biological samples have been analysed without particular problems. The results of the study of desferrioxamine pharmacokin-etics in healthy subjects and patients with renal failure on haemodialysis and with haemochromatosis will be published elsewhere. We are indebted to the Fondation Langlois for its support and Mrs. Laisne for typing the manuscript. References 1. 2. 3. 4. 5. 6. Ackrill P. Ralston A. J . and Day J. P. Lancet 1980,2,692. Meyer-Brunot H. G. and Keberle H. Biochem. Pharmacol., 1967 16 527. Cramer S. M. Nathanael B. and Horvath C. J . Chroma-togr. 1984 295 405. Kruck T. P. A. Kalow W. and Crapper McLachlan D. R., J. Chromatogr. 1985,341 123. Harris W. R. and Carrano C. J. J. Inorg. Biochem. 1984, 22 201. Allain P. and Mauras Y. Anal. Chim. Acta 1984 165 141. Paper A5141 7 Received November 12th 1985 Accepted December 1 Oth 198

ISSN:0003-2654    

DOI:10.1039/AN9861100531

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

536 ANALYST, MAY 1986, VOL. 111 Table 1. Dependence of relative fluorescence intensity on the lanthanide concentration Element N* C1/2,t P.P.m. Pr . . . . . . 12 Sm . . . . . . 14 Eu . . . . . . 14 Tb . . . . . . 09 Dy . . . . . . 14 Ho . . . . . . 13 Er . . . . . . 14 Tm . . . . . . 13 Lu . . . . . . 22 * N , number of samples. t CIl2, half-concentration and its standard deviation. $. (Z/Zo)c,O, intercept and its standard deviation 194 k 18 1163 k 132 1201 2 160 141 f 21 1095 f 107 1060 f 89 1026 f 80 1045 f 86 1189 f 117 (I/Zo)c = o,$ % , _ 104.3 k 1.03 93.1 k 1.03 91.6 f 1.04 91.5 k 1.08 102.4 f 1.03 104.8 k 1.03 99.4 k 1.02 103.3 k 1.03 107.0 f 1.04 50 F t 0 500 1000 1500 C, p.p.m. Fig. 1. sity on quencher concentration for Sm, Eu, Dy and Ho Semi-logarithmic dependence of uranyl fluorescence inten- similar (weak) quenching power whereas Pr and Tb behaved as medium quenchers.Table 1 shows the calculated concen- trations for which the relative standard deviation was around 10%. The theoretical intercept value (Illo = 100% for C = 0) was not introduced in the calculation treatment as an observed value but it was calculated by means of the experimental values. In all instances the average of the intercept was around 100% and although their standard deviations were small (1.0-1.1% for each element) a relative standard deviation of 10% was also assumed here. According to Figs. 1-3 the semi-logarithmic dependence, up to a certain limit (30% for Tb in Fig. 3) of the relative fluorescence intensity on the quencher, was assumed to be valid for the nine lanthanides investigated.Comparison of the Quenching Power of the Lanthanides The half-concentration plotted in Fig. 4 for lanthanum and the lanthanides are taken from measurements on fluxes obtained 100 50 L F t 50 t I- I I I 1 0 500 1000 1500 C, p.p.m. Fig. 2. Semi-logarithmic dependence of uranyl fluorescence inten- sity on quencher concentration for Er, Tm and Lu by addition of U and quencher (trivalent lanthanide) before fusion. From Table 2 it can be seen that the quenching power is constant for most of these elements (i.e., 1100 k 300 p.p.m. for Sm, Eu, Gd, Dy, Ho, Er, Tm, Yb and Lu). Pellets containing any of these elements at a concentration equal to the half-concentration were white or near-white, and because of their high CIl2 values, were categorised as weak quenchers.Ce, Pr, Nd and Tb act as medium quenchers (with C112 < 300 p.p.m.) but although pellets containing Pr or Tb were strongly coloured, those containing Ce or Nd were very pale (almost white). As some lanthanides (Ce, Pr, Nd, Tb, Dy and Tm) are reported to yield tetravalent compounds,8 our attention was drawn to those rare earths that may be oxidised during the fusion. Some lanthanide sesquioxides (Ln203) or lanthanide com- pounds prepared from volatile acids give dioxides (Ln02) or at least oxides (LnO, with 1.5 < x < 2) on heating in air (e.g., Ce02, white or pale yellow9; Pr6OI1 or Pro2, blacklOJ1; Tb02, chocolate12; whereas Nd407, proposed by Brauner, is doubtful). Hence, after fusion, a certain amount of Ln(IV) can subsist or be formed in the molten fluorides where the presence of oxygen acts as an oxidising agent, while oxide anions present in the molten flux stabilise the higher valency states.During the pellet solidification, the Ln(II1) - Ln(IV) equilibrium is frozen. In the final flux, covalent Ln(1V) compounds show different quenching powers to the more ionic Ln(II1 j forms.ANALYST, MAY 1986, VOL. 111 537 1000 t Table 2. Comparison of the quenching power of lanthanides in the NaF - LiF flux and their tendency to yield the tetravalent state ' Ce Pr Nd Pm Sm Eu Gd Tb Ho Er Tm Yb Lu DY Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . c1/2, P.P*m.3006 1947 2506 11637 12017 14876 1417 10956 10607 10267 10457 8006 11897 - Colour of pellet* Slightly grey Black Slightly grey Hardly brown Hardly ochre Hardly ochre Strong chocolate Slightly yellow-pink Slightly ochre-pink Pink Slightly yellow Slightly ochre Slightly yellow - Compounds of tetravalent element$ E" t/V vs. Ce 0.00 CeO,, CeF,, Cs3CeF 1.45 Pro,, PrF,, Cs3PrF7 2.65 Nd407 ? Cs3NdF7 2.95 3.35 4.55 6.15 1.35 3.15 4.25 4.35 4.35 5.55 7.35 Tb02, TbF4, Cs3TbF CS3DYF7 Cs3TmF7 * Pellet colour for C = 4000 p.p.m. t E" from Nugent et al.,5 extrapolated to other media,7 reported vs. Ece(m) - c~(Iv). t: Ln(1V) compounds from references 5 and 8-12. 6 ClI2 values from reference 2. 7 CIl2 values from this work. 1 I I I + 0 100 200 300 400 C, p.p.rn. Fig.3. sity on quencher concentration for Pr and Tb Semi-logarithmic dependence of uranyl fluorescence inten- 2000 r 200 t I ' I Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 58 59 60 61 62 63 64 65 66 67 68 69 70 71 Fig. 4. De endence of C1,2 on the atomic number of lanthanides: ., this work; 4, from reference 2 Tests to prove the presence of tetravalent lanthanides were carried out with cerium. Fluxes containing no uranium were spiked with a Ce(II1) standard solution (up to 2000 p.p.m.). After fusion, the pellets containing cerium were not com- pletely soluble in water, but they were completely dissolved in concentrated sulphuric acid (70 "C). The flux leaching solution exhibited the yellow colour of Ce(IV). This test shows the presence of tetravalent cerium in the pellets investigated.Another significant observation, which proved the valency change during the fusion - solidification process, was that the Ce(II1) half-concentration (300 p.p.m.) reported earlier2 was lower than the average CIl2 for the weaker quenching lanthanides (CIl2, 1100 p.p.m.) whereas the half-concen- tration from pellets obtained by spiking Ce(IV) standard solutions was 138 p.p.m.4 This result can be explained by a partial oxidation of the initial trivalent cerium during the pellet preparation. For praseodymium, terbium and neodymium, the low half-concentrations could also be explained on the basis of a modification of their initial valency state. As shown in Table 2, the colours of the pellets spiked with Pr and Tb corresponded to the colour of their dioxides.Pellets containing Nd were pale grey, but information on the colour of neodymium dioxide was not available from the literature. Hence, the presence of Nd(IV) and Ce(IV) (pale, covalent compounds), which would absorb in the wavelength range of the UV absorption of the uranyl ion, were assumed to explain their medium quenching power. However, this behaviour can be explained in the same way for Tb and Pr and by considering also the absorption of the emitted fluorescent light on addition of coloured terbium or praseodymium tetravalent compounds. Comparison of the quenching half-concentration of the rare earths and their ability to form tetravalent compounds was made on the basis of the Ln(II1) - Ln(IV) potential calculated by Nugent et aZ.5 and reported (Table 2) versus the potential of cerium.This classification of potential can be extrapolated to other media7 and thus to the molten flux where Ce(IV), Pr(IV), Tb(IV) and Nd(IV) (to a lesser extent) can be formed or maintained at their valency state because their potentials were near to the Ce(II1) - Ce(IV) potential. The other lanthanides, however, are much more difficult to oxidise in the flux where they behave as weak quenchers in the trivalent state. We thank the Head of the Low Level Radioactivity Section of the IAEA Laboratory Seibersdorf, Dr. 0. Suschny, for his interest in this work. References Veselsky, J. C., and Ratsimandresy, Y., Anal. Chim. Acta, 1979, 104, 345. Veselsky, J . C., Radiochim. Acta, 1982, 29, 53. Price, G. R., Ferretti, R. J., and Schwartz, S ., Atomic Energy Commission Document AECD-2282, IAEA, Vienna, 1945. 1945. Veselsky, J. C., Mikrochim. Acta, 1982, 2, 335.538 5. Nugent, L. J., Baybartz, R. D., Burnett, J. L., and Ryan, L., J. Inorg. Nucl. Chem., 1971, 33, 2053. 6. Degueldre, C. A., IISN 1, Institut Interuniversitaire des Sciences Nuclkaires, Brussels, 1977. 7. Duyckaerts, G., and Gilbert, B., Inorg. Nucl. Chem. Lett., 1977, 13, 537. 8. Spitzin, V. I., Martynenko, L. I., and Kieselev, Ju. M., 2. Anorg. A&. Chem., 1982, 495, 39. 9. Spencer, J. F., J. Chem. SOC., 1915, 107, 1265. ANALYST, MAY 1986, VOL. 111 10. 11. 12. Prandtl, W., and Huttner, K., 2. Anorg. Chem., 1925, 149, 235. Pagel, H. A., and Brinton, P. H. M.-P., J. Am. Chem. SOC., 1929, 51, 42. Urbain, G., and Jantsch, G., C. R. Acad. Sci., 1908, 146, 127. Paper A5110 Received January 4th, 1985 Accepted May 12th, 1985538 5. Nugent, L. J., Baybartz, R. D., Burnett, J. L., and Ryan, L., J. Inorg. Nucl. Chem., 1971, 33, 2053. 6. Degueldre, C. A., IISN 1, Institut Interuniversitaire des Sciences Nuclkaires, Brussels, 1977. 7. Duyckaerts, G., and Gilbert, B., Inorg. Nucl. Chem. Lett., 1977, 13, 537. 8. Spitzin, V. I., Martynenko, L. I., and Kieselev, Ju. M., 2. Anorg. A&. Chem., 1982, 495, 39. 9. Spencer, J. F., J. Chem. SOC., 1915, 107, 1265. ANALYST, MAY 1986, VOL. 111 10. 11. 12. Prandtl, W., and Huttner, K., 2. Anorg. Chem., 1925, 149, 235. Pagel, H. A., and Brinton, P. H. M.-P., J. Am. Chem. SOC., 1929, 51, 42. Urbain, G., and Jantsch, G., C. R. Acad. Sci., 1908, 146, 127. Paper A5110 Received January 4th, 1985 Accepted May 12th, 1985

ISSN:0003-2654    

DOI:10.1039/AN9861100535

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

ANALYST, MAY 1986, VOL. 111 539 Quantitative Determination of Thiourea in Aqueous Solution in the Presence of Sulphur Dioxide by Raman Spectroscopy Heather J. Bowley, Elizabeth A. Crathorne and Donald L. Gerrard BP Research Centre, Chertsey Road, Sunbury-on-Thames, Middlesex Tw16 7LN, UK A method has been developed for the quantitative determination of thiourea in aqueous solutions acidified with sulphuric acid in the presence of sulphur dioxide. The method uses laser Raman spectroscopy and is valid in the presence of high concentrations of sulphur dioxide and inorganic salts. This is particularly important with respect to the use of such solutions for the leaching of gold from ores and concentrates. The method is simple, rapid and accurate to within + I .2% and has been applied to the study of solutions used in leaching experiments.Keywords: Thiourea determination; sulphur dioxide; Raman spectroscopy; gold ores The use of an aqueous solution of thiourea as a leaching agent for gold has been widely reported in the literature.1-3 It offers several advantages over sodium cyanide in that it operates in acidic media, forming a cationic gold complex, and has faster dissolution kinetics. However, a major disadvantage is its tendency to oxidise , eventually forming elemental sulphur and several other decomposition products. This leads to an excessive consumption of the thiourea, which generally makes its use uneconomical in the treatment of gold-bearing ores and concentrates. The addition of sulphur dioxide during the leaching process has been found to decrease the thiourea consumption consid- erably by, it is thought, preventing the oxidation process.&6 In order to evaluate the economic feasibility of a gold leaching operation based on thiourea, it is necessary to have a reliable method for determining its concentration in acidified aqueous solutions containing sulphur dioxide and a range of metal salts.The standard methods for quantitatively determining thiourea in water are: (1) titration with N-bromosuccinimide7; (2) titration with mercury(II1) nitrate8; and (3) potentiometric titration with potassium iodate.9 It was found that iron(I1) ions, present in the ores, interfere with method 1 and sulphur dioxide interferes with methods 2 and 3. An alternative method is therefore required for this type of study, and the technique of Raman spectroscopy offers some distinct advantages for this purpose.Experimental Apparatus Potentiometric titration was carried out using a Metrohm titroprocessor. The titration curve and titre volume were plotted automatically using a Knauer dual-pen recorder. All Raman spectra were recorded using an Anaspec Model 36 laser Raman spectrometer fitted with a Tracor Northern Reticon Type S intensified diode array detector. Procedures Potentiometric titration of thiourea Thiourea is oxidised by potassium iodate in acidic solution to give formamidine disulphide: 6SC(NH2)2 + KI03 + 6H+ .+ 3[SC(NH2)2]22+ + 3H20 + KI The end-point of the titration is shown by a sharp increase in potential at a platinum redox electrode. A known volume of the solution to be analysed (5-20 ml) was pipetted into 50 ml of 1 M phosphoric acid and titrated with 0.017 M potassium iodate solution using the titroprocessor.The thiourea concentration was calculated from a knowledge of the volumes of thiourea solution taken and potassium iodate solution used and the molarity of the potassium iodate solution. Using this technique a thiourea solution, containing a weighed concentration of 7.602 g 1-1 in 0.1 M sulphuric acid, was analysed nine times and gave a mean of 7.57 g 1-1 with a standard deviation of 0.05 g 1-1. Laser Raman spectroscopy The Raman spectrum of an aqueous solution of thiourea, showing the C=S stretching mode of thiourea at 735 em-1, is shown in Fig. 1; the y (OCO) mode of acetic acid at 880 cm-1 is shown in Fig.2. As Raman spectroscopy is not intrinsically quantitative, an internal standard must be used for analyses of this type. Because of the particular application to gold leaching, in this study the aqueous solution of thiourea would contain sulphuric acid, sulphur dioxide, various inorganic salts and oxidation products of thiourea. Hence, the best internal standard for this application is one with a strong Raman band that is well separated from any bands from other components likely to be present and stable in sulphuric acid. The compound chosen was acetic acid, which has a strong Raman band at 880 cm-1 (Fig. 2) that is well separated from thiourea, Reagents All of the reagents used were of analytical-reagent grade or equivalent. Solutions of thiourea were made up in sulphuric acid (0.1 M) in various concentrations between 3 and 20 g 1-1, the mass of the thiourea being determined to three decimal places.Solutions for analysis by Raman spectroscopy also contained an accurately weighed amount of acetic acid as an internal standard (see below). An approximate 1 : 1 molar ratio of acetic acid to thiourea was used. Sodium metabisul- phite was used to dope the thiourea solutions with sulphur dioxide, assuming that 1 mol of sodium metabisulphite forms 2 mol of sulphur dioxide in acidic solution. I 1 t Q, C Q Lo LT 300 400 500 600 700 800 900 1000 1100 1200 Wavenum ber shift/cm - 1 Fig. 1. Raman spectrum of aqueous thiourea540 ANALYST, MAY 1986, VOL. 111 sulphuric acid, sulphur dioxide or common anion bands..The intensity of this band was found to be similar to that of the thiourea band at 735 cm-1 (Fig. 1) and so acetic acid was added to the solutions to be analysed at approximately the same molar concentration as the thiourea. The concentration of thiourea is then obtained from the ratio of the integrated areas of the 735 and 880 cm-l bands, and the known mass of acetic acid added to the solution. The Raman technique was calibrated using a range of standard solutions (Table 1) in order to establish the relationship between the 735 and 880 cm-1 bands. It was found necessary, because of differences in response of the various component diodes of the diode array with changes in laser power, to re-calibrate with a known standard before each analysis, as the ratio of the 735 to 880 cm-1 bands could vary by as much as 5%.It was also necessary to use the same area of the detector for each measurement to increase the accuracy of the method. Leach tests Samples (50 g) of pyrite concentrate (head assay = 17 p.p.m. of gold) were leached with 117 ml of 0.1 M sulphuric acid, containing 10 g 1-1 of thiourea, at 25 "C. The concentrate was stirred at 700 rev. min-1 in a 500-ml flat-bottomed glass vessel with indented baffles to improve the mixing. The potential of the solution was monitored by means of a platinum redox electrode, the oxidation potential being maintained between 380 and 420 mV for the first 3 h by the addition of iron(II1) I I I I I I I 1 500 600 700 800 900 1000 1100 1200 1300 Wavenu m ber s hifi/cm - Fig. 2. Raman spectrum of aqueous acetic acid Table 1.Determination of the thiourea concentration with no sulphur dioxide Thiourea concentratiodg I-' Thiourea concentration Potentiometric (by mass)/g 1-1 titration Raman spectroscopy 3.087 3.1 Standard 5.022 5.1 Standard 9.868 9.9 Standard 9.960 9.9 Standard 5.02 5.0 4.96 10.00 9.9 10.02 19.98 19.9 19.83 sulphate solution. The appropriate mass of sodium metabisul- phite was added in stages over these first 3 h to give the required thiourea to SO2 molar ratio. During the last hour of the test no adjustment was made to the potential of the solution, which fell to between 233 mV (SO2 to thiourea molar ratio of 2: 1 mlm) and 368 mV (no SO2). At the end of the leach the pulp was filtered and the residue washed with 2 x 100 ml of distilled water, the filtrate and washing being combined.The total volume of the solution was then measured. The solution was analysed for thiourea by both potentio- metric titration and laser Raman spectroscopy, the latter after the addition of the appropriate amount of acetic acid as the internal standard. The solution was also analysed for gold by atomic absorption spectrometry and the gold content of the dried residue was determined by fire assay. The percentage of the total gold extracted into solution could therefore be calculated. The mass of thiourea in the feed and product solutions was calculated and the amount of thiourea consumed in the leach test expressed as the mass of thiourea consumed (in kg) per mass of pyrite concentrate used in the test (in tonnes) i.e., kg tonne-1.Two tests were carried out using the conditions described above, but in a 1-1 vessel using 100 g of concentrate and 233 ml of a 50 g 1-1 solution of thiourea in 1 M sulphuric acid. Results and Discussion Determination of the Thiourea Concentration in the Absence of Sulphur Dioxide The concentration of thiourea in five solutions containing 5.02, 10.00 and 19.98 g 1-1 of thiourea (by mass) was determined both by potentiometric titration and laser Raman spectroscopy. The results are given in Table 1. Generally good agreement was found between the amount of thiourea added and the amount found using the two analytical techniques, the maximum error being 1% for the potentiometric and 1.2% for the laser Raman technique. Determination of the Thiourea Concentration in the Presence of Sulphur Dioxide The results are listed in Table 2.Typical traces from the potentiometric titration of a 0.1 M thiourea solution containing different amounts of sulphur dioxide are shown in Fig. 3. At an SO2 to thiourea molar ratio of 1 : 2 the change in potential owing to the reaction of the sulphur dioxide with potassium iodate can be clearly seen (marked as the SO2 end-point) although in practice the end-point of this reaction is difficult to determine. Comparing the trace of thiourea alone with that of a solution containing an SO2 to thiourea molar ratio of 1 : 10, no obvious difference in curve shape is apparent. However, the calculated thiourea concentrations of the two solutions are 7.6 g 1-1 (no SO2) and 8.3 g 1-1 (with SO2). Thus, at low concentrations of SO2 its reaction with potassium iodate cannot be detected from the shape of the potentiometric curve.The fact that a falsely high figure for the thiourea concentration is obtained is not apparent using this method at low sulphur dioxide concentrations. Table 2. Determination of thiourea concentration in the presence of sulphur dioxide Thiourea Thiourea concentratiodg 1-* (by mass)/ concentration/ SOz: thiourea Potentiometric Raman concentration Na2S205 g I-' g I-' molar ratio titration spectroscopy 10.10 12.6 1 : l 23.2,24.1 9.95 10.10 4.6 1 : 2.7 14.6,15.1 9.98 10.10 2.5 1 : 5 13.0,12.5 10.11ANALYST, MAY 1986, VOL. 111 541 No so2 Thiourea: S02=10: 1 mlm Thiourea: S02=2: 1 mim Fig. 3. Potentiometric titration of a 7.61 g 1-1 of thiourea solution Table 3.Leaching of a pyritic gold ore using thiourea Thiourea in feed Test No. (by mass)/g 1 - 1 1 10.12 2 10.12 3 10.00 4 10.00 5 50.03 6 50.03 Na2S205 addedg 0 0.75 1.4 2.8 0 8.0 SO2: thiourea Gold extraction, molar ratio % 73 1 : 2 61 1 : l 62 2 : 1 59 50 1 : 2 67 - - Table 4. Analysis of leach liquors Thiourea consumption*/ kg tonne - 1 Thiourea in feed Thiourea in leach Thiourea in leach SO,: thiourea by potentiometric liquor by potentiometric liquor by Potentiometric Raman Test No. molar ratio titration/g 1 - 1 titratiodg 1-1 Raman spectroscopy/g 1 - 1 titration spectroscopy 1 - 10.0 2.3 2.27 10.2 9.8 2 1 : 2 10.0 3.1 3.35 5.4 4.0 4 2 : 1 9.9 7.1 3.34 (-20.1) 2.8 5 - 49.8 22.5 22.55 25.1 24.9 3 1 : l 9.9 4.1 3.84 (-0.4) 1.0 6 1 : 2 49.8 23.0 22.05 (-0.5) 4.3 * Parentheses indicate that the thiourea concentration determined is apparently greater than the initial concentration of thiourea.A 0.1 M sulphuric acid solution containing 10.10 g 1-1 of thiourea was doped with different amounts of sodium metabi- sulphite to give calculated SO2 to thiourea molar ratios of 1 : 1, 1 : 2.7 and 1 : 5. The thiourea concentration of the solutions was then determined by laser Raman spectroscopy and potentiometric titration. Based on the total volume of potassium iodate used to titrate the samples, thiourea concentrations df 24, 15 and 13 g 1-1, respectively, were obtained by potentiometric titration. In each instance the presence of SOz in the solution was obvious from the shape of the potentiometric curve.The thiourea concentrations obtained for the same solutions by Raman spectroscopy were 9.95, 9.98 and 10.11 g 1-1, respectively, showing that the presence of SO2 did not affect this technique. Results of the Leaching Studies The results of the leaching studies are given in Tables 3 and 4. As the amount of sodium metabisulphite added during the leaching of the pyrite concentrate was increased, the differ- ence between the thiourea concentration of the leach liquors, as determined by the two analytical methods, also increased. With no sulphur dioxide present both methods gave similar thiourea concentrations in the leach liquor, i.e. , 2.27 g 1-1 by Raman spectroscopy and 2.3 g 1-1 by potassium iodate titration. The amount of thiourea consumed was, therefore, similar, 9.8 and 10.2 kg tonne-1, respectively. At an SO2 to thiourea molar ratio of 1 : 2 the thiourea concentration in the leach liquor was 3.35 g 1-1 by Raman spectroscopy and 3.1 g 1-1 by potassium iodate titration, giving thiourea consumptions of 4.0 and 5.4 kg tonne-1, respectively.Increasing the amount of SOz in the leach to a 1 : 1 SOz to thiourea molar ratio and above obviously left some free SO2 in the leach liquor as the analysis by potentiometric titration gave a higher thiourea concentration in this solution than the feed concentration. Analysis using Raman spectroscopy gave542 ANALYST, MAY 1986, VOL. 111 thiourea consumptions of 1.0 kg tonne-’ (SO2 to thiourea, 1 : 1) and 2.8 kg tonne-1 (SO2 to thiourea, 2 : 1). Leach tests using 50 g 1-1 of thiourea in 1 M sulphuric acid gave a greater decrease in the thiourea consumption when sulphur dioxide was present than the tests that used 0.1 M sulphuric acid.With no sulphur dioxide present, the Raman and potentiometric methods again gave similar thiourea consumption figures, 24.9 and 25.1 kg tonne-1, respectively. With an SO2 to thiourea molar ratio of 1 : 2 in the leach, the calculated value for thiourea consumed fell to 4.3 kg tonne-’ by Raman spectroscopy, whereas the potentiometric result indicated the production of 0.5 kg tonne-’, obviously an unacceptable result. The presence of sulphur dioxide in the leach tests using 0.1 M sulphuric acid appeared to decrease the gold extraction. At present, the reason for this is not apparent, but it may relate to the redox potential being allowed to fall over the final hour of the leach in order to minimise the amount of sulphur dioxide left in solution. The assistance of Mr. K. R. Nunn and advice from Mr. R. D. Hancock are gratefully acknowledged. Permission to publish this paper has been given by the British Petroleum Co. plc. References 1. 2. 3. 4. 5. 6. 7. 8. 9. Groenwald, T., Hydrometallurgy, 1976, 1, 277. Hiskey, J. B., Miner. Metall. Process., 1984, 11, 173. Pyper, R. A., and Hendrix, J. L., “Gold and Silver-Leaching, Recovery and Economics,” SME-AIME, 1981, p. 93. Lesoille, M., Br. Pat. 2077247, 1983. Schultze, R. G., J. Met., 1984, 6 , 62. Schultze, R. G., Ger. Pat., 3401 961, 1984. Thibert, R. J . , and Sarwar, M., Microchem. J., 1969, 14,271. Yatsimirsky, K. B., and Artasheva, A. A., Zh. Anal. Khim., 1956, 11, 442. Schultze, R. G., personal communication. Paper A51363 Received October 11 th, 1985 Accepted November 25th, 1985

ISSN:0003-2654    

DOI:10.1039/AN9861100539

出版商:RSC    

年代:1986

数据来源: RSC