JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRYy AUGUST 1 99 1 VOL. 6 187R ATOMIC SPECTROMETRY UPDATE-ADVANCES IN ATOMIC ABSORPTION ANDFLUORESCENCESPECTROMETRYAND RELATEDTECHNIQUES Steve J. Hill" Department of Environmental Sciences Polytechnic South West Plymouth Devon PL4 8AA UK John 6. Dawson Department of Instrumentation and Analytical Science UMIST P. 0. Box 88 Manchester M60 1 QO UK W. John Price 75 Amberley Close Holne Cross Ashburton Devon UK Ian L. Shuttler Bodenseewerk Perkin-Elmer GmbH Postfach 101 764 0-7770 Uberlingen Germany Julian F. Tyson Department of Chemistry University of Massachusetts Amherst MA 01 003-0035 USA Summary of Contents 1 Atomic Absorption Spectrometry 1 .l. Flame Atomizers 1.1 .l. Fundamental studies 1.1.2. Interference studies 1.1.3. Sample introduction 1.1.3.1.Discrete procedures 1.1.3.2. Atom-trapping techniques 1.1.3.4. Sample introduction by flow injection 1.1.3.5. Solid sample introduction 1.1.4. Chromatographic detection 1.2. Electrothermal Atomizers 1.2.1. Atomizer design and surface modification 1.2.2. Sample introduction 1.2.3. Fundamental processes 1.2.4. Interferences 1.2.5. Developments in technique 1.3. Chemical Vapour Generation 1.3.1. Hydride generation 1.3.2. Preparation-separation of volatile organometallic compounds for chemical vapour AAS 1.3.3. Mercury by cold vapour generation 1.4.1. Light sources 1.4.2. Optics 1.4.3. Detectors 1.4.4. Background correction 1.4.5. Continuum source and simultaneous multi-element AAS 1.5.1. Instrument control 1.5.2. Data processing 1.5.3. Chemometrics 1.4. Spectrometers 1.5.Instrument Control and Data Processing 2 Atomic Fluorescence Spectrometry 2.1. Discharge Lamp Excited Atomic Fluorescence 2.2. Laser-excited Atomic Fluorescence Spectrometry 2.2.1. Electrothermal atomization 2.2.2. Low pressure atomization systems 2.3. Inductively Coupled Plasma Excited Atomic Fluorescence 2.4. Studies of Flames and Plasmas Using Laser-induced Fluorescence 3 Laser-enhanced Ionization 3.1. Systems Operating at Atmospheric Pressure 3.2. Systems Operating at Reduced Pressure 4 Atomic Magneto-optical Rotation Spectrometry * Review Co-ordinator to whom correspondence should be addressed.188R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 In previous years the third and fourth ASU reviews in JAAS have been devoted to ‘Instrumentation’ and ‘Atomization and Excitation’ covering developments in equipment-procedures and fundamental processes respectively.Many of the advances in analytical atomic spectrometry reported in the ASU references concern research in both of these areas and duplication of coverage was becoming difficult to avoid. As a separate ASU deals with all aspects of XRF and atomic mass spectrometry it has been decided to create two new reviews on ‘Atomic Emission Spectrometry’ and ‘Atomic Absorption Spectrometry Atomic Fluorescence Spectrometry and Related Techniques’. The review on ‘Atomic Emission Spectrome- try’ may be found in JAAS Volume 6 Issue 4. The present review describes developments in all aspects of atomic absorption spectrometry atomic fluorescence spectrometry and related techniques including fundamental processes and instrumentation reported in the Atomic Spectrometry Updates References in JAAS Volume 5 (90/1159-90/4179) and Volume 6 (91/1-91/2702).The full references names and addresses of authors can be readily found from the Atomic Spectrometry Update References in the relevant issues of JAAS. However as an additional service to readers an abbreviated form of each literature reference quoted (except for those to Conference Abstracts) is given at the end of the review. Comments on the new format of the reviews are welcome. 1. ATOMIC ABSORPTION SPECTROMETRY 1.1. Flame Atomizers 1.1.1. Fundamental studies The relatively few publications that have appeared during this review period concerned with fundamental studies of processes in flames for analytical applications indicates the maturity of this topic.Studies of atomization mech- anisms particularly those carried out in the presence of matrix species attract more interest and they are considered in section 1.1.2. A method for measuring the collisional widths of absorp- tion lines developed by O’Haver and Chang (see J. Anal. At. Spectrom. 1989 4 165R and section 1.5.2) has now been published (90/3399). The procedure is based on the shapes of calibration curves generated from absorbance data produced by fitting with the help of a computer a model for the transmission profiles obtained when a continuum source is used. A precision of f 0.1 pm was obtained for the elements from Cd to Sr with a spectrometer having a bandpass of between 5 and 12 pm.A few papers have appeared concerned with the analyti- cal utility offlames other than those employing air-C2H2 or N20-C2H2. The determination of Cd in various non-ferrous alloys was improved by the use of an air-C3HS/C4H10 flame (90/1527) and the rapid determination of Cu Mn and Ni in steels with the help of an air-C,H flame was made (91/1264). The measurement of A1 using an O2 enriched air-C2H2 was reported (9 111448). A characteristic concen- tration of 2.5 mg 1-I of A1 was obtained at a C2H2 flow rate of 4.17 1 min-l and an 02:C2H2 ratio of 0.58. Air-CH3(CH2)4CH3 or air-C,H CH3 flames were used for the determination of Fe in copper-nickel alloys after liquid-liquid extractive separation (9 1 /I 367). A character- istic concentration of 0.13 mg 1-’ of Fe was obtained.The use of the N20-C3H8 flame has been described for the determination of Ca and Mg in natural and waste waters (91/910 9111 1 12) and the determination of Mn in steels (9 I/ 1 264). Also of interest in this section is the detection of combustion products by a variety of techniques. Molecular fluorescence stimulated by radiation from an ArF laser was used to monitor NO (90/1851). Laser-induced fluores- cence has also been used to monitor OH in low pressure ( 1- 10 bar) flat flames (90/4 1 79) and both OH and CN have been detected by this technique (9 1/C63 1). This presenta- tion described the use of resonance enhanced multiphoton ionization to detect HCO and CH3 radicals while major species (such as CO C02 and NO) were detected by FT’IR spectrometry.The use of molecular beam sampling with MS detection was also described. 1.1.2. Interference studies Despite the maturity of flame-based spectroscopic proce- dures there is a steady output of publications concerned with the study of interference effects. As in previous reviews (J. Anal. At. Spectrum. 1990,5 19 1 R) much of this literature is concerned with methods for removing or compensating for the various effects thereby ensuring accurate analyses. A comprehensive demonstration of a variety of interfer- ences has been provided by Pszonicki and Skwara (90/2235) as part of a demonstration of the utility of a procedure for overcoming them. The method is a combina- tion of the standard additions and successive dilution methods with the additional feature that if the concentra- tion indicated by the extrapolation procedure is imprecise an excess of a known interferent is deliberately added to swamp the effect of the unknown interference due to some matrix species.The method does not correct for additive interferences such as those due to background or spectral overlap. A number of enhancement effects have been reported. In the determination of Al in an O2 enriched air-C2H2 flame (see section 1.1. l) the characteristic concentration was improved to 1.2 mg l-’ by the addition of a variety of organic reagents containing a phenyl ring with hydroxyl and/or carboxyl groups in the ortho position (9111448). These included aluminon chrome azurol S sulphosalicylic acid and hydrogen phthalate.For the determination of Rh in organic matrices (90/3390) the addition of ethanoic acid produced a detection limit of 0.2 mg 1-’ for atomization in an air-C2H2 flame. In the direct determination of P (90/2230) via measurement of Co following the extraction of the ion pair between molybdophosphate and bis[2-(5- chloro-2-pyridylazo)-5-diethylaminophenolato]cobalt(1~) 2- propanone was found to produce an enhancement owing to an increased aspiration rate. Two other solvents ethanol and methanol were found to increase the aspiration rate but without a corresponding improvement in sensitivity. This was attributed to a decrease in flame temperature. Chinese workers (9 112542) reported remarkable enhancements in the sensitivity for Ru on the addition of oxidizing agents.Factors of 10 17 19 19 and 23 were found for bismu- thate(v) peroxodisulphate(vx) iodate(v) bromate(v) and cerium(xv) respectively. The combination of cerium(xv) and bromate(v) was found to enhance the signal by a factor of 37 with a corresponding detection limit of 20 ng ml-l. Further confirmation of the enhancement effect of amines on the A1 signal in an air-C2H2 flame (J. Anal. At. Spectrom. 1989 4 165R) was provided (911929). TheJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 189R addition of tetrabutylammonium chloride at a concentra- tion of 6% m/m in the final solution gave a detection limit of 4.8 mg 1-1 for the determination of A1 in silicon. The standard additions method was used to compensate for the interference due to iron and calcium.Other Chinese workers reported on the sensitizing action of sodium dodecyl sulphate (9 1/406 9 1/242 1 91/2423). The first report concerned the determination of Cr in plating bath sludges. It was found that dissolution in l0/6 sulphuric acid with the addition of sodium sulphite removed the interference from copper iron magnesium and nickel. The second report (9112421) showed that the sensitivity for Ni could be increased by the addition of sodium dodecyl sulphate (SDS) cetyltrimethylammonium chloride or aniline. The effects of aniline and either surf-actant were found to be additive. The third report (9112423) indicated that the linear range for the determination of V in steels was increased by the addition of SDS. Yan and Zhang have proposed a mechanism to explain this surfactant effect (9012076).In contrast to earlier explanations based on aerosol ionic redistribution (see for example Kornahrens H. Cook K. D. and Armstrong D. W. Anal. Chem. 1982 54 1325) the model was based on reverse micelle forma- tion. It was suggested that the analyte ions in solution attract a group of surfactant molecules whose hydrophilic ends of opposite charge to that on the metal ion species are orientated towards the ion in the centre of the micelle. During the aerosol generation processes each individual analyte entity is eventually completely surrounded by surfactant molecules and on desolvation in the flame they provide a highly efficient micro-environment for the pro- duction of atoms. In addition potentially interfering species of opposite charge should be excluded from the micelle giving rise to an effective protecting mechanism.Supporting evidence for this was presented for the measure- ment of Ca in the presence of a number of anionic interferences including phosphate and aluminium. The addition of 0.02 mol dm-3 SDS completely restored the signal magnitude to that obtained for the same concentra- tion of Ca in a solution containing no potential interferent. Refinements of an earlier method (J. Anal. At. Spectrom. 1989,4 165R) for the correction of physical interferences in a nebulizer by a method based on Fraunhofer diffraction have been made by Kitagawa and Okuyama (9 1/1122). The size of the laser beam dimension was restricted by a 0.2 mm aperture and an improved computer program was used to estimate the background signal more accurately.The procedure was tested on solutions of Cu in sulphuric acid of concentration up to 4 mol dm-3. A comprehensive study of the likely interferences encoun- tered in the determination of Cr in argillites (a type of oil- shale containing 10-20% organic matter 2O-4O0/o clay minerals and 50-55Oh aleurite) was made (90/3591). It was found that accurate results could be obtained with atomiza- tion in an air-C2H2 flame if potassium thiocyanate was added at a concentration of 0.4Oh. The addition of per- chloric acid (0.3%) was found not only to help with interference reduction but also increased the sensitivity. The possible benefits of the addition of dodecylamine hydrochloride were also investigated but the performance was considerably inferior to that of the thiocyanate-per- chloric acid combination.It was suggested that the addition of these reagents had a beneficial effect on the flame chemistry and promoted the formation of volatile chromyl chloride. A characteristic concentration of between 0.07 and 0.2 mg 1-l of Cr was obtained. Chinese workers (90/3450) found that sulphosalicylic acid prevented inter- ferences from aluminium silicon titanium and zirconium in the determination of Mn in alloys. Russian workers reported (91/914) similar benefits from the addition of thiourea to natural- and waste-water samples for the determination of Bi. A detection limit of 0.8 mg 1-l was obtained. For the determination of Mo in the same sample types (91/1293) it was found that the addition of sodium sulphate and ammonium chloride reduced the interferences from calcium iron nickel and zinc.German workers (91/987) found that both hydrochloric acid and strontium were required to remove the depressive effect of 1- hydroxyethylidene- 1,l -biphosphonate on the determina- tion of Ca. The effect was also alleviated by the addition of EDTA. The effectiveness of several procedures for over- coming the depression caused by inorganic phosphate on the determination of Ca were studied by Welch et al. (90/4033). The target analysis was the determination of Ca in urine for which a method based on the addition of lanthanum (20 g 1-l) followed by atomization in an air-C2H2 flame was found to be superior in accuracy to procedures involving (a) the addition of potassium with atomization in an N20-C2H2 flame and (b) the addition of smaller concentrations of lanthanum and atomization in an air-C2H2 flame.The effectiveness of some releasing agents for the removal of the phosphate depression on the determination of Mg has been studied (91/1049). The continuous titration method was used (J. Anal. At. Spec- trom. 1986 1 137R) to study the effects of calcium lanthanum praseodymium and strontium on atomization in an air-H2 flame. Pyrogallol has again been shown (90fC 1753) to be effective in the removal of the depressive effects of phosphate and silicate in the determination of Ba Ca Mg and Sr (J. Anal. At. Spectrom. 1988,3 142R). The continuous titration procedure was also used in this study. The interference of aluminium in the determination of noble metals (Ag Au Ir Pd Pt and Rh) has been investigated (91/1570) and its beneficial effect on the determination of Mn in soil samples in which the presence of iron caused inaccuracies noted (9 111 399).In this study a correlation between releasing action and ionization poten- tial was observed. Posta and Skucs have now published details (911284) of their study of the effect of alkali and alkaline earth metal salts on the atomization of some transition metals in the air-C2H2 flame. As was previously noted (J. Anal. At. Spectrom. 1989 4 165R) the interference arises from the volatilization characteristics of the solid aerosols formed from the pairs of salts. These workers concluded that the presence of magnesium sulphate should be avoided because of its specific interference action. A number of years after drawing attention to the problem (see Thompson K. C.and Wagstaff K. Analyst 1980 105 641) Thompson has presented further results (9UC8181) on the role of trace concentrations of silicate in the determination of Fe and Mn in samples relating to the water supply industry. Again specific instrument dependence was found and for matrices other than potable waters the N20-C2H2 flame was recommended. A steady state kinetic model for the so- called ionization suppression effect has been developed (9 1/C 1797). The model which takes into consideration the radiative recombination collisional excitation and de- excitation of K ions in the presence of excess of sodium accounted for the observed enhancements which were larger than predicted on the basis of ionization suppression due to increased electron number density.Koirtyohann and co-workers have studied (9 1 /C 1 900 90/2200) the unexpected interferences produced by certain oxygen-containing compounds on the atomic absorption signal of a number of REEs and first row transition elements in the N20-C2H2 flame. The effects mainly suppressions were observed for various organic acids amino acids carbohydrates sulphuric and phosphoric acids and are characterized by showing an abrupt slope change at mole ratios of concomitant to analyte of between 1 1 and 2:1. The fact that many of the compounds in question190R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRYy AUGUST 199 1 VOL. 6 contain oxygen and decompose instead of vaporizing when heated was thought to be significant. The effects could be overcome by re-optimization of the fue1:oxidant ratio and were considerably less severe when the sample contained other salts.A partial explanation in terms of solid-phase reactions after desolvation but prior to vaporization was offered. The effects of different mineral acids on the determination of Cd and Pb have been reported (9111497). There appeared to be little difference between hydrochloric nitric perchloric and sulphuric acids and between fuel-rich and fuel-lean flames. Klunder and Boss have again reported (91/C1863) on their studies of ligand interferences in fuel- rich flames. To study the rate of atomization thought to be affected by many interferents the sample was introduced by a single-droplet generator and the atomic population interrogated by AFS with pulsed HCL excitation.Chinese workers (9 1/1155) have examined the quantum chemical parameters of 14 organic complexing agents and studied the possible correlation between an enhancing effect on the atomization of Yb and (a) the charge density on the frontier atom and (b) the bond order between the frontier atom and its neighbouring atom. Huzino has developed a method for the determination of Ca in calcium silicate (91/1294). The sample was ground to a particle size of less than 4 pm and suspended in water. To the suspension was added hydro- fluoric acid and carbon black (J. Anal. At. Spectrom. 1989 4 165R) at a concentration of 0.02% m/v. A linear calibration function was obtained for atomization in an air-C2H2 flame for Ca concentrations up to 500 mg I-'.1.1.3. Sample introduction The methods used for the production and transport of aerosols in analytical instrumentation have been reviewed by Browner (90/3246). The properties that give aerosols their particular usefulness in interfacing bulk liquids sample solutions with various detection systems were discussed and some specific examples of analytical applica- tions given. Reference to flame atomizers was brief and mainly from a historical point of view however the article contains much useful information about aerosol generation and transport and should be read by all practising analytical spectroscopists who rely on aerosols as an integral part of their instrumental techniques. Willis et al.have discussed some practical aspects of handling aerosols for samples containing high concentrations of dissolved solids (9 1 /847). In line with suggestions made by the Instrumental Criteria Sub-committee of the Analytical Methods Committee Analytical Division Royal Society of Chemistry (Anal. Proc. 1984 21 4 9 the decrease in absorbance with time while nebulizing a solution containing an easily atomized metal in a refractory halide matrix was used as a perform- ance parameter. Two different types of spray chamber and burner design were evaluated. Superior performance was obtained from a burner design with a flared slot. In addition to spraying solutions of sodium chloride containing a few mg 1-1 of Cu as the test element a cola drink spiked with A1 and a used gear-box oil diluted (1+4) with aviation kerosene were also used.Drop-size measurements were not made but the relative depression of the Mg signal in the presence of aluminium was used as an indicator of aerosol characteristics. The roles of the position of impact bead and of paddles were also studied and some suggestions made for the basis of comparison of the safety aspects of burners for use with N20-C2H2 flames. This paper provides a useful set of results with which those from future studies in this area may be compared. It is likely that with the rise in interest in flow injection introduction procedures (see later) studies of the tolerance of any new system to blockage and to solute volatilization interferences will undoubtedly be made.Several reports of the use of double capillary nebulization have appeared. Chinese workers (90/352 1) described a system for the determination of some hydride-forming elements and of Hg. The method was described as simple and rapid and was applied to the determination of As Bi Hg and Sb in geological samples. The lower limit of the working range was found to correspond to 10 mg kg-' for As and to 1 mg kg-l for the other elements. A study of the flame profiles of Ca Cr Cu and Fe in the presence of several organic solvents (introduced in one branch) and a non-ionic surfactant (introduced with the analyte in the other branch) has been made by Bagdi et al. (91/942). The solvents included IBMK acetone benzene toluene xylene isobutyl acetate and tetrachloromethane. The presence of the organic solvents (apart from tetrachloromethane) had a negligible effect on the influence of the surfactant poly- oxyethylene glycol nonylphenyl ether.Tetrachloromethane was found not unexpectedly to decrease markedly the sensitivity for Ca and Fe. The results were compared with the nebulization of emulsions of organometallic com- pounds stabilized with the surfactant (see J. Anal. At. Spectrom. 1987 2 182R and J. Anal. At. Spectrom. 1990 5 193R) and the differences observed were attributed to the interaction of the surfactant with the inorganic analyte species. The use of a similar device was proposed by Watling et al. (9111463) for the addition of an ionization suppressant to flames into which organic solvent samples were being introduced. It was found that the procedure resulted in considerable savings in costs and time over the addition of organic-based suppressants and these workers suggested that the procedure should be useful for the determination of wear metals and additives in oils.Stupar has also described such a procedure for use in conjunction with an ultrasonic nebulizer device (91/1103) in which the two solutions are mixed on the face of the piezoelectric transducer. Possible applications to flow injection (FI) methods and for internal standardization in ICP-OES were suggested. Further studies by Berndt on the hydraulic high pressure nebulizer have been described (90/1929 91lC498). The latter report indicates that organic solvents can be success- fully introduced into flame atomizers and thus the device can be used as an interface between HPLC and atomic spectrometry detectors (see under 1.1.4).The use of the system in the FI mode with sample volumes of just 1 or 2 ply was mentioned. Gustavsson has reviewed the design and application of various devices for interfacing HPLC and FI procedures with atomic spectrometry (91/1056 91/1098). Details of these devices have appeared in previous Updates (J. Anal. At. Spectrom. 1990 5 195R and J. Anal. At. Spectrom. 1989 4 167R). Robinson and Choi described a system for the production of Hg atoms (9013353) which used a thermospray nebulizer to introduce the sample solution into a circular air-C2H2 flame. The flame gases were entrained in the vertically mounted side arm of a quartz T-cell. The sensitivity was 0.2 mg 1-l and the detection limit was 10 ng ml-1. Credit must go to workers at Dalhousie University for publishing in this field of considerable maturity what appears to be a completely new way of generating an aerosol for introduction into a flame atomizer namely by the use of an atmospheric pressure helium plasma (90/3495).A 90 kHz discharge was produced between an upper electrode (copper or chromel) mounted 10- 1 5 mm above the surface of the sample solution contained in an 1 1 mm diameter vial into which a counter electrode was inserted. Some details of the capability of the power supply were provided in an earlier publication concerned with sputtering (Stephens R. J. Anal. At. Spectrom. 1988,3 1 137). The signal was found to increase with both sodium chloride concentration andJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL.6 191R power. However the rate of transport to the flame was only 0.09 ml min-l and thus the sensitivity and detection limits were correspondingly poorer than those obtained with conventional nebulization. In a study of droplet size no diffraction pattern was observed with the laser Fraunhofer method but solute volatilization interferences were ob- served when Ca was determined in the presence of phosphate or aluminium. The mechanism of aerosol forma- tion was therefore considered to be mechanical disruption of the surface followed by rapid evaporation of the solvent. 1.1.3.1. Discrete procedures. The supposed advantage in terms of improved detection limit which accrues by the use of a procedure in which a relatively large volume of sample (adequate for conventional continuous nebulization) is taken through a preconcentration procedure which pro- duces a limited volume (so that discrete nebulization is employed) was evaluated critically by Beinrohr (90/3958).The theoretical argument was supported by results for the determination of Fe and it was concluded that the improve- ment in detection power of the preconcentration procedure may be more than offset by the reduced sensitivity and poor precision of the discrete nebulization step. Workers con- templating the development of such a procedure are advised to read this paper closely. In the meantime a procedure for the determination of Mo in geological and biological materials in which pulse nebulization into a flame atomizer was used has been developed by Monte and Curtius (9013479).A preconcentration procedure involv- ing the sorption of the Mo complex with the 0,O-diethyl ester of dithiophosphoric acid onto activated carbon with release into nitric acid gave rise to only 1 or 2 ml of solution from which 100 pl sub-samples were dispensed by a micropipette. Rocks described such an analysis (90/C4041) the determination of Cu and Zn in 100 pl samples from a hospital baby care unit. It was pointed out that the discrete nebulization procedure avoided the risk of contamination from reagents. Chinese workers (9 1/374) described the use of this same volume of sample solution in the determination of Cu Mg Mn and Zn in aluminium alloys. The final solution contained strontium as a releasing agent and was 1-2% m/v in dissolved solids; an air-C2H2 flame was used.In contrast Gruber and Herbauts (9 1 / 1 490) used an N20-C2H2 flame for the determination of A1 in solutions containing high concentrations of dissolved solids. However by the addition of tetrabutylammonium chloride Chinese workers (91/1469) were able to use an air-C2H2 flame in the determination of A1 in silicon. What appears to be the standard volume for discrete nebuliza- tion 100 p1 was taken. With Zeeman-effect background correction a detection limit of 8 mg 1-l was obtained and it was implied that this was an improvement on the value of 1 1.5 mg 1-l obtained with conventional nebulization. The Chinese workers mentioned in the previous section de- scribed the adaptation of the branched capillary (90/352 1) to a discrete nebulization mode (90K2046).As well as suggesting the possible use for chemical vapour generation procedures the device was used for the addition of releasing agents and ionization suppressors. Russian workers (91/C488) described the use of two discrete procedures. Either sample volumes of 1 pl were introduced via the nebulizer or samples were introduced on a probe inserted directly into the jlame (or plasma). It appears as though evaporation from the probe was assisted by either ultrasound or external heating (see also J. Anal. At. Spectrom. 1990 5 193R and J. Anal. At. Spectrom. 1989 4 167R). Also of interest is a report of the combination of discrete nebulization with molecular emission spectrometry for the determination of B in steels (90/2221).Further applications of the injection device constructed by Futekov et al. have been developed (90/4010). In addition to the introduction of a discrete sample volume the procedure uses ensemble summation of successive signals to improve the SIN ratio (J. Anal. At. Spectrom. 1987 2 183R). The determination of Bi Cu and Zn at concentrations of 5 0.5 and 0.25 mg kg-l respectively in sodium chloride and of Cu Fe and Pb at concentrations of 0.47 3.7 and 0.1 mg kg-' respectively in ammonium fluoride was achieved. Further details of the device are given later in the flow injection section. A procedure for the determination of Au with direct sample introduction into the flame by means of a silica boat was described (91/418).After extraction with IBMK from an acid mixture a 50 pl sample was placed in the boat which was transported by means of a rotating turntable. However a precision of only 15% RSD was claimed. Discrete amounts of Co were introduced by evaporation of some volatile derivatives (91/7) by a procedure appar- ently developed independently by two Spanish research groups but published in one paper. A number of com- pounds were studied but the final procedure adopted for the determination of Co in steel bronze and brass was based on derivatization with trifluoroacetylacetone. A 5 pl sample of the final solution was injected into a heated vessel mounted externally to the spray chamber. A similar procedure has been used for the introduction of the trifluoroacetylacetone derivatives of Cr and Fe as well as of Co (91/2576).Further details of the procedure are given in section 1.3 (Chemical Vapour Generation). 1.1.3.2. Atom-trapping techniques. Chinese workers (90/1263) have reviewed the use of both water-cooled atom traps (WCATs) and slotted tube atom retarders (STARS). In contrast to previous review periods there would apear to be no other publications or conference presentations con- cerned with the WCAT. Several researchers have described the use of a STAR. Bums et al. (9013418) used an FI method for the determination of Pb in copper-based alloys. Although the detection limit (0.13 mg 1-l) was degraded in comparison with that for continuous nebulization (0.09 mg l-l) a throughput of 50 samples h-l was possible and the life of the quartz tube was extended.Zhang and Yuan (9 1/C2 150) described a procedure for the determination of Pb in water in which sulphydryl cotton fibres were used for preconcentration and matrix isolation. Sensitivities 2-3 orders of magnitude higher than those obtained with conventional flame techniques were claimed. The perform- ance of a commercially available device has been described on a number of occasions (90/C1750,90/C3037,91/2401). Improvements in detection limits by up to a factor of 4 were obtained for a number of elements normally deter- mined in fuel-lean to stoichiometric air-C2H2 flames. 1.1.3.4. Sample introduction by flow injection. The inter- est in FI procedures noted in last year's Update (J. Anal. At. Spectrom. 1990,5 194R) has been sustained.Several of the prominent workers in this area contributed to a review book (Flow Injection Atomic Spectroscopy ed. Burguera J. L. Marcel Dekker New York 1989) from which some chapters have been individually abstracted (9 1/155 9 1/ 1 3 1 1 9 1/ 1 3 12). The book while presenting useful material to illustrate the scope of flow injection atomic spectrometry (FIAS) contains few references after 1986 and thus is somewhat out of date in terms of recent developments in what has proved to be a rapidly growing area of research activity; not just for flame-based techniques but also for most other types of analytical atomic spectro- metry. Tyson has reviewed the relevant 1988 and 1989 literature (9 1 / 1498). Attention was drawn to the variety of FI procedures described for preconcentration and separation of analyte and matrix species.This aspect of FIAS has been further elaborated in a series of conference presentations describing ongoing work in Tyson's research group (90K1606 90/C3038 91/C1854 91K2100) and was also highlighted by RhiiCka (911C1853). Features of a new192R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1991 VOL. 6 commercially available HAS system introduced by Perkin- Elmer have been described by company personnel on several occasions (90/1537 90/C2050 90/C2908 90/C2982 9 1/C788 9 1/C1746 9 1/C1883 9 1/C2098). Carbonell et al. have investigated some findamental parameters of the basic FI manifold interfaced to an FAA spectrometer (9 1/3 10). These parameters included flow rate nature of the carrier stream (use of surfactants or of organic solvents) and the use of compensating fluid flow introduced at a connector just prior to the nebulizer.No major new effects were observed and it was found that the use of a water carrier stream with no compensation gave satisfactory results for the determination of Cu Mn and Pb in sewage. Kuban et al. (9 1 /2523) also used a simple single- line manifold for the determination of Mg in aluminate luminophors and Zhou et al. (91/1273) determined Ca K Mg and Na in surface waters. A manifold containing a UV-visible absorption spectrophotometer in series with an FAA spectrometer has been used for the determination of pairs of components in silicate rocks (90K2038). A scheme for the eventual determination of ten constituents was described.A third group of Chinese workers ( 19/ 1 52 1) used this manifold design for the determination of Li in brine and of Pd in geological samples following extraction of the mercaptobenzothiazole complex into 4-methylpentan-2- one. The single-line manifold has also been used for the introduction of slurries. For the determination of Cu Mn and Pb in sewage sludges (91/310) the samples were partially digested in sealed Pyrex glass containers in a microwave oven. After dilution to volume with water a 500 pl sub-sample was injected. Accurate results were obtained for the analysis of two reference materials and detection limits of 60 50 and 300 ng ml-l for Cu Mn and Pb respectively were claimed. For the determination of Fe and Zn in foods (9 1/ 1 584) slurries were prepared by grinding the dry sample to pass through a 1-2 mm sieve followed by suspension in water.A specially designed Babington V- groove nebulizer was used with an additional oxidant inlet. Again 500 p1 sub-samples were injected. Calibration was effected by injection of slurries of various concentrations of a white bean homogenate whose composition with respect to the analytes had been established by analysis following dry ashing and dissolution in dilute (1 + 1) hydrochloric acid solution. Detection limits of 0.6 and 0.3 mg 1-l for Fe and Zn respectively were obtained. Bums et al. (9013418) determined Pb in copper by injection of the sample into a single-line manifold with a flow rate slightly above that required by the suction developed by the nebulizer.The excess of fluid was diverted to waste at a T-connector just prior to the nebulizer. With the aid of a slotted tube atom retarder a detection limit of 0.13 mg 1-l for a 250 pl injection volume was obtained. Workers at Xiamen University (9 1 /C2 142) demonstrated the possibility of determining Cu in aluminium with an FI system incorporating on-line electrolytic dissolution. Gradient chambers have been used for the generation of concentration gradients and dilution of samples. Chinese workers (9112483) used the rise curve produced on switch- ing from blank to concentrated standard (see Tyson J. F. and Appleton J. M. H. Talanta 1984 31 9) to calibrate for the determination of Mg in waters and geological materials. Good accuracy and precision were claimed. De la Guardia et al.(90/4011) used a merging zones manifold (to add lanthanum) for the determination of Ca K Mg and Na in water. The well-stirred gradient chamber was located between the merging point and the spectrometer. The use of two such chambers of volumes 827 and 2245 pl was evaluated. Calibrations up to several hundred times less sensitive than normal could be obtained thereby extending the working range by more than two orders of magnitude. These workers were unable to use the exponential concen- tration-time relationship which is obtained for a step change in concentration for single-standard calibration but obtained satisfactory results from empirical relationships of the form logc=A + tlogB where c and t are concentration and time respectively and A and B are experimentally determined coefficients.Workers from the same research group (91/25 14) used the same manifold for the determina- tion of Pb in gasoline. In this procedure the mixing chamber served to produce an emulsion of the injected sample. On- line de-metallation of the alkyl compounds by reaction with iodine in petroleum spirit was used to permit the determi- nation of total lead. As the sensitivities for tetraethyllead and tetramethyllead were considerably different in the absence of the iodine reagent it was suggested that this could be exploited as the basis for a method of speciation. After establishing the calibration function for each of the species two measurements on each sample were made (with and without the iodine). Results for the analysis of synthetic mixtures appeared to be encouraging.Initial results on the use of wide-bore tubing (1.5-5.0 mm id.) for on-line dilution have been presented (90/C 1606 9 1 /C790). Prob- lems due to differences in physical properties between the injected solutions and the carrier stream were encountered. For the determination of some clinically relevant elements Araujo and Lima (9011549) used three different FI mani- folds to cover the range of concentrations encountered. On- line dilutions by stream splitting and merging of diluent were made. A procedure incorporating on-line dialysis has been developed for the determination of total and free Ca in milk (van Staden J. F. and van Rensburg A. Analyst 1990 115 605) which has been extended to include the determination of Cl(9 1 / 1 5 1 4).Total Ca was determined by solution spectrophotometry and free Ca by FAAS. The dialysis unit separated the free Ca and for an injection volume of 30 pl provided an on-line dilution allowing determinations over the range 100-1 500 mg 1-l. Interfer- ence from phosphate was overcome by the use of an N20-C2H2 flame and an ionization suppressant (2.5 mg ml-l potassium solution) was merged just after the dial y ser. The possibility of exploiting on-line chemical reaction for the development of indirect procedures was discussed by Hansen (90lC2867) under the title of ‘FIA Conversion Techniques’. The benefits of the controlled kinetic features of FI were stressed. Esmadi and Attiyat (9 1/1528) described a procedure for the determination of either ammonia cyanide or thiosulphate. A distilled water carrier stream was aspirated by nebulizer suction at 3 ml min-l through a manifold which contained a column (2.8 mm x 7 cm) of solid silver chloride and glass beads (1.9 mm).The dissolved silver was detected at a downstream FAA spectro- meter. Detection limits of approximately 10-6 mol dm-3 were obtained for an injection volume of 200 pl. Li and Zhang (911424) determined Si in steels by an indirect procedure based on the formation of Molybdenum Blue off- line. A detection limit for Si of 0.25 mg 1-’ was reported. Procedures for the determination of two drugs (methadone and chlordiazepoxide) have been developed in the continu- ing programme of research in this area by Valcarcel’s group at Cordoba (91112 91/2469).The manifold contains a reductor column of zinc or cadmium metal which releases an equivalent amount of metal ions into the carrier stream on reaction with a reducible species. For the determination of chlordiazepoxide (9 1/ 12) the sample solution had a pH of 3 and so a measurable zinc blank signal was obtained. This was subtracted from the signal obtained on injection of the sample solution (200 pl). The working range was 2.0-25 mg 1-l with a sampling frequency of 150 h-l. As the reaction is specific to the N-oxide group of the target molecule the method is selective for chlordiazepoxide inJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 193R the presence of other 1,4-benzodiazepines. For the determi- nation of methadone (91/2469) a cadmium column was used and a 90 pl sample volume was injected giving a working range of 5-50 ng ml-I.The procedure was used to assay the drug in pharmaceu- tical preparations and in urine. For the latter sample the drug was extracted from alkaline solution into dichloro- methane which was evaporated and the residue taken up in distilled water. The majority of HAS publications in this review period concern the performance of on-line preconcentration and matrix separation procedures. Although a small number of papers have appeared in which liquid-liquid extraction (LLE) or precipitation have been used by far the greatest activity is in the development of solid-phase extraction (SPE) procedures. Kuban and co-workers have described (9 11323) an LLE manifold coupled indirectly to a flame spectrometer and its application to the determination of Cu in water from power station cooling circuits (9 1 / 1 25 5).The analyte was extracted at pH 5.05 as the APDC complex into IBMK and after phase separation 100 or 1 50 pl sub-samples were injected into an air carrier stream for direct introduction to the spectrometer. The phase separator was a flat membrane type and it was found that microporous Teflon had a superior performance to that of hydrophobized cellulose. The use of the FI valve in this position in the manifold provides an interface between the extraction manifold and the transport manifold. In general for FI-LLE-FAAS there is a mismatch between the optimum flaw rates for these two functions. Efficient extractions are only obtained at low flow rates and in order to obtain a real preconcentration it is necessary that the flow rate of the organic phase be less than that of the aqueous sample phase.Depending on the volume injected the working range was 20-900 or 2-500 ng ml-l with a sample throughput of 40 h-l. These workers concluded that although the procedure had performed satisfactorily it had no advantages in comparison with the use of solid-phase extractants. Tyson and co-workers have reported on some further attempts (90/1470 91K1864) to exploit on-line precipitation for sample clean-up and pre- concentration. The possibilities for the removal of interfer- ence effects were demonstrated (90/1470) for the determi- nation of Ca in some synthetic rock solutions containing aluminium phosphorus and silicon.The Cordoba group has published another method for determining trace metals in mineral samples (J. Anal. At. Spectrom. 1990 5 195R) with a procedure for the measurement of Co by precipita- tion with 1-nitroso-2-naphthol (90/1176). A merging stream manifold was used in which the sample stream (suitably buffered and dosed with citrate masking agent) flowing at 7 ml min-l merged with the reagent stream flowing at 1 ml min-l. After passing through a 70 cm reaction coil the precipitate was collected on a stainless-steel filter. Dissolu- tion was effected by forward flushing with ethanol. A stream-switching valve located between the filter and the spectrometer diverted the matrix components to waste rather than passing them through the nebulizer. By taking a 100 ml sample volume it was possible to achieve a detection limit of 1 ng ml-1 at a sampling frequency of 1 h-l.However by reducing the sample volume to 10 ml a more reasonable throughput of 40 h-l was possible with a detection limit of 6 ng ml-I. Accurate analyses were obtained for the determination of Co at concentrations of between 2 and 20 mg kg-l in silicate rocks. An indirect precipitation procedure for the determination of chlorhexidine has been developed (9 1/2 189) by another Spanish group. The precipitate produced on injection of sample solution into an ammoniacal copper carrier stream was retained on a paper or plastic filter and subsequently dissolved in nitric acid. A working range of 5-20 mg 1-1 was obtained for Cu. Aspects of the design and application of in-line columns for solid-phase extraction (SPE) procedures have been considered by Fang (90/C2047).By combining SPE with chemical vapour generation it was possible to achieve detection limits of 1 2 and 2 ng 1-1 for Bi Hg and Se respectively. Use of columns of suitably small dimensions allowed on-line enrichment factors of 25-30 to be achieved for sample volumes and throughputs similar to those obtained for conventional introduction. Coetzee et al. (9014016) devised a method for the determination of Ag in bore-hole water at concentrations down to 4 ng d - l . The column (1 mm x 3 cm) contained basic alumina and was inserted in a single-line manifold between the valve and the nebulizer. The carrier stream was propelled by a dual piston HPLC pump which could deliver either sample solution a water carrier or an ammonia regenerant solution.After loading the column at 5 ml min-l the accumulated Ag was removed by injection of the eluent solution (500 pl of 2 mol dm-3 nitric acid) into the water carrier at an optimized flow rate of 5 ml min-l. It was found that the pH of the sample solution was critical with a maximum retention at pH 4.0 whereas the sampling flow rate was not. In order to avoid low recoveries of Ag due to the formation of ammine complexes the column was washed thoroughly with water after regeneration. The total analytical cycle time appeared to be about 10 min. High concentrations of various calcium magnesium and sodium salts (0.1-0.5 mol dm-3) did not interfere. Bennet and Dean (91/C801) presented preliminary results for the separation of Co as the chloro complex from iron and nickel which are potential spectral interferents.An injection valve was used to control the volume of sample solution loaded onto the column rather than rely on a procedure based on constant delivery from a pump for a known time. An evaluation of a number of possible solid-phase reagent materials for the preconcentration of A1 in dialysis concentrates has been made by Periero Garcia et al. (90/3478). It was concluded that Chelex 100 and Amberlite IRA-400 were the most suitable materials although it was pointed out that the latter material cannot be used when the ethanoate concentration exceeds 40 g l-l and requires a chloride concentration of at least 3 g 1-1 in both standards and samples. The sample solution (up to 25 ml) was introduced by injection through a septum into a buffer carrier stream.A switching valve located between the column and the spectrometer diverted matrix components to waste during sample loading. The eluent was 100 pl of sodium hydroxide solution. The performance of the method was evaluated for the determination of A1 in a dialysis concentrate which contained less Al than could be detected by direct analysis by ETAAS. Accurate recoveries were obtained for concentrations of spikes of between 50 and 190 ng ml-l. The performance of another chelating resin containing iminodiacetate functional groups Muro- mac A-1 has been evaluated by Hirata and co-workers (90/1968). A fully automated manifold incorporating five 3- way and two 2-way solenoid valves was constructed in which the column (4 x 7 mm) was mounted directly on the outside of the spray chamber.The operation of the manifold appeared to consist of loading the sample through the column by timed switching of the appropriate valves followed by blowing out the sample solution remaining in the manifold between the switching valve (where the eluent was introduced) and the column by switching in a stream of argon. Unless this was done a reduced peak height sensitivity was obtained depending on the length of tubing between the eluent switching valve and the column (lengths of up to 100 cm were investigated). This effect was discussed in terms of dispersion during the loading but no comments were made on the possible role of dispersion of the leading edge of the eluent stream during elution which194R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL.6 seems a more likely explanation. The column was used to preconcentrate Cd Cr Cu Fe Mn and Zn with enrichment factors of between 90 and 180 obtained for 20 ml sample volumes. The sampling rate was 13 h-l and the detection limits were between 0.14 and 2.1 ng ml-l. Accurate analyses of some biological reference materials for Cd and Cu were made. Kuban et al. have reviewed the use of chelating material for FI preconcentration (91/2239) and made a detailed comparison of some materials containing bound oxine or diethylenetriaminetetraethanoic acid (DTTA) functional groups (91/898) for the retention of Cu. A six-way five- column injection device was used which contained five columns (3 mm x 2.6 cm) in parallel.Samples were buffered to pH 4.2 and loaded for up to 5 min at a flow rate of up to 11 ml min-l although 5 ml min-l were typically used. The accumulated Cu was eluted in the reverse direction by a stream of 2 mol dm-3 hydrochloric acid. Enrichment factors of up to 100 were obtained with sampling frequen- cies of up to 70 h-l. Kuban et al. demonstrated another useful feature of FI-SPE-FAAS namely the greatly in- creased linear range. By control of the volume of sample loaded linear calibrations were obtained from the detection limit (0.3 ng ml-l) to 200 pg ml-l (60 s load time). Good agreement between the FI procedure and an ETAAS procedure was obtained for the analysis of power station cooling circuit water.A simple version of this procedure has been described by another Czechoslovakian group (91/245) in which a column of the oxine material was incorporated into the nebulizer capillary. A known volume of buffered sample solution was loaded onto the column by the action of nebulizer suction. After aspirating a water wash the copper was eluted with 2 mol dm-3 hydrochloric acid solution. Peak area was used. as the quantitative parameter. The dynamic range was from the detection limit of 0.3 ng ml-l to 5 pg ml-l (breakthrough) with precisions better than 5% RSD for concentrations above 25 ng ml-l. A similar procedure was described by Karakaya and Taylor (9011 923) for the determination of Cd in urine. A column of alumina was used and the flow rate was externally con- trolled by means of a peristaltic pump.A detection limit of 1 ng ml-l was obtained. One problem for on-line micro-column procedures is the dimensional instability of some resin- based solid phase extractants. The cycle of swelling and shrinking which occurs when the pH of the solution is changed can lead to blockage of flow lines or frits on the one hand and the appearance of voids and channels on the other. Several publications have reported on the use of silica-based materials. Fang and Welz (90/ 1 175) described a procedure which used 8-hydroxyquinoline immobilized on controlled pore glass (CPG) in ultra microcolumns fabricated from Eppendorf pipette tips. The column volume was only 120 pl and only 1.6 ml of sample solution were required which gave a throughput of 120 h-l.Knotted tube conduits were used to minimize the dispersion arising from flow in the tubing connecting the column and the spectrometer and it was found that a conical reactor shape gave a lower dispersion than a cylindrical packed-bed reactor. Enrich- ment factors of between 25 and 31 were obtained for Cu Cd and Pb and the potential of the procedure for the determination of these elements in sea-water was demon- strated by recovery from synthetic samples. The repetitive determination of Cu at a concentration 100 ng ml-l had an RSD of 1 Soh (n= 8 1). Some preliminary results for the use of this material for the determination of A1 have been presented (9 1/C785). The precipitation of A1 from solutions of high pH required to get efficient retention was pre- vented by the addition of malonate as an auxiliary complexing agent.Enrichment factors of up to 70 were obtained. Other workers have immobilized 8-hydroxy- quinoline-5-sulphonic acid on CPG (9 1 / 1 077) and used this material in a simple single-line unidirectional flow manifold for the determination of Cu Cd Hg Mg Pd and Zn. For a 5 ml sample volume detection limits of 0.5,0.6,25 1.0,lO and 1 .O ng ml-l respectively were obtained for a sampling rate of 20 h-l. Metals were eluted from the column by the injection of 100 pl of 2 mol dm-3 hydrochloric acid solution at a flow rate of 2 ml min-’. The column was also loaded at this flow rate. The capacity of the extractant was found to be 5 x 10-3-8 x mol g-l. The use of a commercial prototype preconcentration unit was described for the determination of Pb in tap water (90/3512).A 3 x 50 mm column containing 8-hydroxyquinoline immobilized on silica was mounted in the injection loop of a six-port rotary valve. The sample solution was merged with a buffer solution containing a cocktail of masking agents (0.2 mol dm-3 boric acid 2% triethanolamine 2% thiourea and 2% acetylacetone) to prevent the interference of aluminium copper iron and zinc. After loading the column was inserted into a water carrier stream by rotation of the valve and the residual buffer components were eluted. The retained Pb was eluted by the injection from a second valve of 1 mol dm-3 hydrochloric acid. A number of tap water samples containing between 5 and 80 ng ml-l of Pb were analysed giving results which were statistically indis- tinguishable from results obtained by a collaborating laboratory using ETAAS.Lancaster and RdiiEka described an SPE procedure in which a C,*-based support material was used (911C1859). The first step in the procedure was to load the ligand onto the column this was followed by the analyte solution and finally by a solvent of reduced polarity to elute the metal ligand complexes formed in the column. The use of a commercially available automated system which incorporates a column of cellulose with iminodiacetic acid functional groups has been described (90/C2984). Also of interest in this section are some closely related procedures. Bulgarian workers are developing two lines of research that bear more than a superficial resemblance to FI.Havezov and Detcha described preliminary work with a controlled dispersion device (91lC1763). In this technique an injection valve is not used but discrete volumes of liquid are intercalated by controlled stream switching and pump speed. The ‘dosage system’ developed by Futekov has been further modified (90140 10). This merging stream manifold system is a cross between a pulse nebulization system and a controlled dispersion system. The driving force for liquid transport was provided by nebulizer suction which drew fluid either from a reservoir of wash solution or from a sample vial in the autosampler tray or both. The wash reservoir was isolated by an electronically controlled valve and the autosampler probe was also electronically con- trolled. The sequence of operations was that with the valve open and the probe up a wash solution was delivered to the nebulizer.The probe was then inserted into a sample and the valve was closed. The air in the branch of the manifold from the merging point to the autosampler provided a bubble which prevented dispersion of the leading edge of the sample slug. The probe was withdrawn after a con- trolled time and shortly after the rear of the sample slug passed the merging point the valve was opened again allowing wash solution to flow. The net result was that a discrete sample volume sandwiched between two air bubbles sandwiched between the wash solution was intro- duced into the nebulizer. There was therefore no on-line dilution. The system was used to determine Bi and Zn in sodium chloride and Cu Fe and Pb in ammonium fluoride.Sample solutions containing up to 30% m/m dissolved solids were successfully handled. Schulze et al. (9 1 /C 1698) compared results obtained for the determination of Cu in various water samples with three different methods includ- ing FI-SPE-FAAS. By using a system of three columns inJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1991 VOL. 6 195R parallel a throughput of 160 h-I with an enrichment factor of 15 was obtained. A number of FI-AAS papers are concerned with the on- line derivatization of the analyte to a volatile derivative. These results are considered in more detail in section 1.1.3. For example methods for the determination of Bi in turbine blades (a matrix with a high concentration of nickel) (91/C1782) Bi in various non-ferrous metals (90/35 11) and Se in blood (9012 194) have been described.A procedure which might be considered gas-phase FI(90/1989) has been evaluated for the determination of Hg and some organom- ercury compounds by Baeyens and Leermakers. The vapour phase analyte species were trapped on gold coated glass beads and after preconcentration thermally desorbed. Under these conditions the organomercury compounds decomposed with the release of Hg for detection by conventional cold vapour absorption. It was found that not all of the Hg was released and a second desorption at a somewhat higher temperature was required. As there was a correlation between the Hg compound and the extent of release during the two steps of the desorption some information about speciation was obtained.1.1.3.5. Solid sample introduction. Few papers concerned with the introduction of solids directly into flames have appeared in this review period. As has been noted already (see above) the Ca content of a calcium silicate material was determined by the nebulization of a suspension of the ground material in water (9 1 ~ 2 9 4 ) and FI procedures have been used to analyse partially digested sewage sludge slurries (9 1/3 10) and some food slurries (91/1584). Morales et al. (9 1/247 1) described a procedure for the determination of Co Fe Mo and Ni in petroleum industry catalysts by slurry nebulization of the partially digested samples after heating with a mixture of nitric and phosphoric acids in closed vessels in a microwave oven into an air-C,H flame.1.1.4. Chromatographic detection The hydraulic high pressure nebulizer has been adapted as a method of coupling HPLC and FAAS (9 1 / 1254). Compared with conventional nebulization an increase in peak area of an order of magnitude was obtained for Cu. The impor- tance of the position of the impact bead was stressed. The use of the system was illustrated by the separation of FeI1 and Fell1 on a 10 pm 150 mm Spherisorb column. The mobile phase was methanol-0.05 mol dm- ammonium ethanoate (30 + 70) adjusted to pH 4.0 with ethanoic acid and the flow rate was 1.0 ml min-'. Iron speciation by HPLC with both electrochemical and FAAS detection has been described (9 1 /C 1 80 1 ). Robinson and Choi (9 11292) described a thermospray interface for HPLC-FAAS.The system was used for the determination of chlorophylls (via Mg absorbance) and in particular the monitoring of duckweed chlorophyll whose destruction appeared to correlate with the presence of heavy metals. The possibility of using this phenomenon as an indicator of heavy metal presence was suggested. An improved performance over this type of coupling was obtained for the determination of ionic alkyllead corn- pounds from a system in which post-column ethylation was used (90/1990). Sample components were separated on 5 pm Nucleosil CI8 with aqueous 80% methanol containing 0.06% of ammonium tetramethylenedithiocarbamate at 1 ml min-' and ethylated by reaction with sodium tetraethyl- borate. The ethylated species were detected in a quartz tube furnace after gas-liquid separation. Detection limits for alkyllead compounds were about 0.1 ng with good discrimi- nation over inorganic lead (the sensitivity for PbI1 was several orders of magnitude less).A half-replicate 25 factorial design was used for optimization of the post- column reactor-separator-atomizer system. A post column derivatization system for the determination of arseno- betaine arsenocholine and tetramethylarsonium cations has been devised (9 1 / 1 5 7). The analytes were separated by HPLC and the methanolic eluent nebulized by the thermos- pray effect and pyrolysed in a methanol-O2 diffusion flame to give the hydride on reaction with excess of H2. The arsine was then atomized in a cool 02-H2 diffusion flame. As AsV was also derivatized and no signal was obtained without the post-thermospray H2 it was concluded that the generation of arsine was thermochemically mediated.A reversed-phase HPLC system directly coupled to a spectrometer has been used to characterize the metal composition of metallothionein isoforms (90129 14). The analytes were Cu Cd and Zn and the matrices included a crude extract of turkey hen liver. The speciation of organic Hg compounds in fish by capillary GC with AAS detection was described (9011 209). The column eluate was pyrolysed at 700 "C to give a limit of detection for methylmercury of 0.1 ng which corresponds to 0.04 mg kg'l in the sample. Some preliminary information about a GC procedure for the determination of organolead and organotin compounds with a quartz tube atomizer has been provided (90/C2 103).Kitagawa and co-workers de- scribed a 'separative column atomizer' (901333 1) consisting of an alumina tube packed with graphite granules coated with gold or platinum or mixed with tin powder. The ability of the column to separate the vapours of Cd Hg and Pb was studied for the temperature range 800- 1400 "C. The device was used for the determination of Hg (90/3337) in various rat tissues. Also of interest in this section is a brief review of atomic spectrometry detection for chromatography (9 1/2 176). 1.2. Electrothermal Atomizers Electrothermal atomization is a mature technique and one would expect the number of papers in the literature to be reaching a plateau if not decreasing. The opposite seems to be the case.During the period of this review the main area of interest is that of sample introduction in particular methods of preconcentration. Clearly analytical scientists are being asked to determine ever decreasing concentra- tions in smaller and smaller samples. While electrothermal atomization has the clear advantage of being able to handle small samples directly we now appear to have reached the point where the detection limits of ETAAS are no longer adequate. While many of the reports in this year's review show considerable imagination and innovation there is still a large group following in the footsteps of early workers and arriving at the same destination though somewhat later! The opening comments from last year's review (J. Anal. At. Spectrom.1990 5 179R) still apply. Littlejohn (9 1/C735) indulged in some crystal ball gazing in reviewing the state and status of ETA in AS. The increase in our knowledge of ETA chemistry combined with the instrumental sophistication at our disposal has made it possible to perform analyses previously considered to be difficult with a minimum of sample preparation and without the need for complex calibration procedures. Apart from the use of ETA in AS the graphite atomizer has become an important vaporization and atomization system for other techniques such as AES furnace non-thermal excitation spectroscopy (FANES) ICP ICP-MS and laser- excited AFS. 1.2.1. Atomizer design and surface modijication Interferences in the double-tube design discussed last year (J. Anal. At. Spectrom.1990,5 179R) were studied by Xuan (9 1/1467). The interference of calcium chloride cobalt HC1 HNO HC104 iron nickel magnesium chloride and196R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRYy AUGUST 199 1 VOL. 6 sodium chloride on the determination of Pb and Cd was studied but no conclusions were presented in the abstract. Chan (91/384) applied this design to the determination of Pb in steel and waste-waters. Compared with the conven- tional graphite tube interferences from calcium chloride HN03 iron and sodium chloride were reduced and sensi- tivity increased by 1.42-fold. Only one paper discussing probe atomization from Chinese workers (9 1/1441) was received during the prepa- ration of this review. These workers prepared the probe by carbonization of a graphite rod followed by coating with zirconium and used it to determine Au in geological samples. The detection limit and sensitivity were improved and interferences reduced using the probe.Some work continues on different graphite tube designs. The use of metal- or metal carbide-coated graphite surfaces has been investigated by Russian Chinese and Japanese workers. This review year coatings of tantalum seem to be the most popular. Wu et al. (91/1438) deter- mined Sc in geological soil and sediment samples with a tantalum foil-lined graphite tube with lanthanum and ammonium hydrogen carbonate as chemical modifiers. The sensitivity was enhanced approximately 1 %fold tube life- times were greater than 600 cycles and the method showed no memory effect with a lower atomization temperature good accuracy and effective elimination of many interfer- ences.Long (91/450) appeared to use a tantalum-coated device similar to a L'vov platform inserted into a graphite tube to determine As in food-grade phosphoric acid. A prior extraction was performed into 4-methylpentan-2-one be- fore back-extraction into 0.4 mol dm-3 HN03. The internal gas was 8-10% Hz in Ar which prolonged the useful life of the tantalum-coated device. Hamid (90/3297) applied tantalum carbide-coated graphite tubes to the determina- tion of Ga and T1. For T1 the pyrolysis temperature could be increased from 400 to 1000 "C and most of the interferences from inorganic acids were eliminated. A tungsten-coated graphite tube has been used by Zheng and Zhang (911407) to eliminate the interferences from 5000-fold amounts of chlorides for the determination of Li with ammonium nitrate or diammonium hydrogen phos- phate as chemical modifiers.A method for the determina- tion of Li in blood was developed. Nater and Burau (9011 97 1) examined the use of lanthanum- tungsten- and zirconium-treated graphite tubes for the determination of silicon. The zirconium-treated graphite tubes improved the atomization characteristics of Si in the presence of alkali metal fluorides as chemical modifiers and gave the best results. Sensitivity was 50 pg of Si at a confidence interval of 95%. No evidence of silicon carbide formation was observed and over 200 samples could be analysed without any degradation of the zirconium-treated surface as shown by no change in the sensitivity.Pyrzynska (9 1/62) examined the use of electrographite pyrolytic graphite coated electro- graphite tungsten- and zirconium-coated graphite tubes for the determination of Cr. The best results were obtained with pyrolytic graphite coated electrographite. In a novel approach Baxter and Frech (90/3328) applied the amalgam forming properties of Hg with platinum directly in a graphite atomizer. The Hg vapour generated by reduction with tin(@ chloride was swept in a flow of Ar through a platinum-lined graphite atomizer to permit in situ preconcentration. Atomization of the Hg was carried out under gas-stop conditions in order to maximize sensi- tivity. For 50 ml sample volumes of synthetic sea-water detection limits (20) below 2 ng 1-l were obtained.Atnashev (90/3266) reviewed molybdenum and tungsten metal electrothermal atomizers compared with graphite atomizers with and without ballast or platform accessories. Prudnikov (9 1/968) examined the performance of micro- furnaces (0.5-3 mm diameter and 5 mm long) fabricated from graphite tantalum and niobium. Higher sensitivity and reduced interferences were claimed. Nakamura and Kubota (91/11) proposed a method to extend the dynamic range of a calibration graph obtained from a tungsten-strip atomizer by calculating the initial mass of an analyte on the atomizer using an atom formation model. The model is based on the assumption that the rate of atom formation given by a simple Arrhenius-type expression is a function of the number of atoms in a sample on the heater and the heater temperature.To extend the dynamic range further the light beam from the hollow cathode lamp was adjusted so as to pass through a region of low atom density above the heater. Ivanova et al. (91/230) and Khavezov et al. (9 1/234) used the tungsten coil electrothermal atomizer originally proposed by Berndt (J. Anal. At. Spectrom. 1990 5 179R) to determine As Sb and Sn in gold (91/230) and alkali and alkaline earth elements in am- monium paratungstate (9 1/234). The tungsten coil elec- trothermal atomizer was found to tolerate higher concen- trations of chloride ions than conventional graphite elec- trothermal atomizers (91/230) and to be more suitable for the low level determination of Ba Ca and Mg in a carbide forming matrix such as ammonium paratungstate (9 1/234).Welz et al. (9012255 901397 1) continued their scanning electron microscopy studies on surfaces from ETAAS for total pyrolytic graphite (TPG) (9012255) and glassy carbon tubes (90/3971). The inner surface of TPG tubes was found to be fairly rough owing to adhering residues of the polycrystalline electrographite substrate on which the TPG was deposited. These residues were vaporized on subsequent dry heating and recondensed on the tube surface in the form of nodules and flakes but disappeared entirely when the tube was heated with an analyte solu- tion. All the experiments were continued for about 550 atomization cycles and all tubes appeared to be in good condition at that point although the sensitivity for V in dilute HN03 and Mo in iron had decreased by approxi- mately 50%.The signal for P in lanthanum disappeared almost entirely and that for Cu in HC104 was very erratic throughout the experiment. It was not always possible to correlate a change in analytical performance with a vis- ible corrosion pattern. De-lamination exfoliation and pitting were typical types of corrosion and primary faults caused by irregular crystal growth the most likely sites for corrosive attack. However the extent of the corrosion was dependent upon the matrix. Glassy carbon tube surfaces (901397 1) were found to vary depending upon the manufacturing date. Corrosive attack on the surface was found to depend on the matrix and its oxidizing power. High iron and lanthanum concentrations were found to be more corrosive than perchloric acid.A frequent form of attack was the removal of the protective skin which is responsible for the low reactivity of glassy carbon with oxygen followed by penetration into and enlargement of the micropores. Clear evidence was found for catalytic graphitization which gives rise to the forma- tion of intercalation compounds which eventually resulted in preferential oxidation and vaporization of this carbon type. Kitagawa and Tanahashi (901395 1) investigated the atomization behaviour of Co Fe and Ni from metal strip atomizers. The various atomizer metals molybdenum tantalum and tungsten showed different atomization peak profiles. The tantalum strip surface covered with tantalum oxide caused tailing of signal peaks which could be reduced by atomization in an Hz atmosphere owing to reductive reactions however for Co and Ni with a tungsten surface the peaks were exceptionally broad.It was postulated that a solid solution was formed between the analyte atoms and the metal surface.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 197R 1.2.2. Sample introduction The sheer volume of references received for review and limitations of space preclude any detailed description of many of the procedures described. The reviewer (X.L.S.) of this subject has attempted to select those that show particular novelty or wide applicability and promise for the future. One paper worth consulting is that by Sturgeon (90/2087) where sample introduction was considered from the point of view of analysing marine samples; some pertinent questions regarding the different forms of sample introduction into an electrothermal atomizer are addressed.The perennial topic of calibration in solid sampling was discussed by Baxter and Frech (91/1507) who compared solid sampling with the cup-in-tube technique and a constant temperature two-step electrothermal atomizer. Consistently lower vapour-phase temperatures and greater interferences were found in the former. The use of a convective gas flow during atomization and non-resonance lines as a means of reducing sensitivity were shown to lead to a greater potential for interferences. Non-resonance lines were more sensitive to matrix-induced changes in the temperature interval in which atoms were formed compared with resonance lines.Atsuya et al. (9 1/ 1509) compared three standard materials NXST Tomato Leaves coprecipitations prepared with magnesium oxinate and nickel-dimethylgly- oxime (DMG)- 1 -(2-pyridylazo)-2-naphthol (PAN) as cali- bration standards for solid sampling. Both the NIST and synthetic material prepared from coprecipitation with magnesium oxinate were suitable standard materials for the direct determination of Cu in several biological materials. Low results were obtained using coprecipitations prepared with nickel-DMG-PAN. This same group Akatsuka and Atsuya (90134 13) used a synthetic reference material prepared by coprecipitating cobalt with magnesium(@ 8- quinolinate to determine vitamin BI2 as Co in solid pharmaceutical samples with the cup-in-tube technique. For a sample mass of 1 mg the detection limit for Co was 0.15 pg kg-* (4 pg kg'l of vitamin BIZ).Low and Hsu (91/1510) collected Pb from airborne aerosols on membrane filters and analysed samples from the filters directly with graphite platform boats and Zee- man-effect solid sampling electrothermal atomization. The distribution of Pb between the sub-samples from each filter was generally homogeneous with RSDs of 11-37% but typically t15Oh for samples with 8 h sampling times. Luecker et al. (90/3994) described a combined solid sampling autoprobe system and applied it to the determina- tion of Cd Cu Mn Pb and Zn in biological matrices. A large number of papers and abstracts are now begin- ning to be generated on slurry sampling. Those workers who have been extolling the virtues of this approach to sample introduction in electrothermal atomization for the last few years should feel rewarded that the topic is starting to become more widely accepted.Miller-Ihli (90K 1687 91/C692) discussed the benefits and limitations of an automated ultrasonic agitation system for biological samples including fluids which may contain sediments. The approach provides significant time savings compared with conventional ashing procedures and minimizes the risk of sample contamination and analyte loss prior to analysis. Carnrick et al. (90/3295) described the design and use of a commercially available automated ultrasonic probe sampler accessory which requires no physical modifications to the electrothermal atomizer or solution autosampler. This system was used by Jordan et al.(90/3294) to determine Mn in RMs. Approximately 10 mg of sample were weighed directly into the autosampler cups and 1 ml of a suspending solution containing dilute nitric acid and Triton X-100 added. Owing to the relative high concentra- tions of Mn in the materials the less sensitive 403.1 nm line was used. No chemical modifier or pyrolysis step was used in the electrothermal atomizer programme and results were within the 95% confidence limits of the certificate values. Bradshaw and Slavin (90/C1688 90/2090) described the application of this equipment to the determination of As Pb Se and T1 in coal and fly ash RMs. Detection limits were of the order of 0.5-1.0 pg g-l in the solid sample with precisions ranging from 2- 10% depending upon the amount of solid sample delivered to the atomizer.Schmiedel et al. (90/3414) considered the problems of solid or slurry sampling for the determination of Mo Ru Rh and Pd in radioactive simulated waste samples carried out in glove boxes and hot cells. The slurry technique was preferred; samples were diluted either with a graphite powder or as a direct suspension in a chemical modifier. Recoveries were between 91 and 102% with RSDs of 4-8% for solid and 91-103% with RSDs of 14-20% for slurry sampling respectively. Haraldsen and Pougnet (90/ 1 342) determined Be in coal slurries. The ETA autosampler was modified to allow continuous magnetic stirring of the coal slurry. Aqueous standards were used for calibration with magnesium nitrate as the chemical modifier.Good preci- sion approximately 5% RSD and accuracy were obtained at the 3 pg g-l level. Pyrolysis and chemical modification were found not to be required by Bendicho and de Loos- Vollebregt (9 1 / 195) for the determination of Cu Co Cr Fe Mn and Ni in slumes of glass samples. The same workers used a slurry extraction procedure (9 111 96) for these glass materials. The samples weighed into autosampler cups were suspended in 3% HF and the slurries stirred by passing a flow of Ar through them. The Ar flow achieves effective mixing of the slurry and removes part of the silicon as silicon fluoride. Homogeneous slurries were obtained with- out the use of stabilizing or dispersing agents and calibra- tion was achieved using aqueous standards. Russian workers Kharlamov and Karyakin (901326 l) applied a novel approach of electric spark colloidal disper- sion to produce slurries from monolithic samples of steels and alloys to determine Bi Pb Sb and Sn.A similar method of spark induced ablation and dispersion in water was used by L'vov and Novichikhin (91145). The results for Cu Fe Mg Mn Pb and Zn in high-purity aluminium were in good agreement with certified values. Detection limits were in the range 2-50 ppb. Shengjun and Holcombe (9111550) utilized unicellular green algae to preconcentrate Cu ions from sea-water and river samples; 6 mg of algae were mixed with 10-100 ml of sample. The resultant algae pellet separated by means of a centrifuge was resuspended in 1 ml of OS0h nitric acid and analysed as a slurry by ETAAS. The values determined were within the limits of certification for the RMs SLRS-1 and CASS- 1.The advantages of an air- ashing stage to avoid the build-up of carbonaceous residues following slurry atomization was investigated by Ebdon et al. (9 1/6). Not surprisingly similar results to those using air or O2 ashing in the clinical field were found. Air ashing significantly reduced interferences from non-specific ab- sorption and allowed a wider range of ashing temperatures without sample loss. Chromium Co Mn and Pb were determined in tomato leaves human hair mussel Chlorella and sewage sludge RMs. Good agreement was obtained with the certified values. Nakamara et al. (90/3361) examined over 30 different surfactants as dispersing agents for slurry analysis. Molyb- denum sulphide-aluminium oxide catalyst samples were suspended in an aqueous solution containing triethanolam- ine and nonylphenylpolyoxyether.Marecek and Synek (9 1/845) reported on the determination of Ga in a-alumina powder. The slurry 0.25% m/v was suspended in etha- nol-water (9 + 1). For calibration standard oxides were prepared either with Ga occluded within the alumina matrix or adsorbed on alumina present in the a- or y-forms.198R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRYy AUGUST 199 1 VOL. 6 It was found that alumina decreased markedly the volatility of Ga and increased the sensitivity approximately 8-fold compared with aqueous solutions containing no alumi- nium. This was considered to be owing to the mini-platform effect of the alumina on the atomizer wall. Kane and Epstein (90/C4078) considered the implica- tions of sampling statistics with respect to slurry sampling for the ETAAS analysis of geochemical reference materials.While slurry sampling has a number of advantages for these materials the measurement precision is highly dependent upon a homogeneous analyte distribution in the samples. However they concluded that even in those instances where slurry sampling results in unacceptable inaccuracies mineral speciation information can frequently be postu- lated from the data. In a conference report Welz and Sperling (90/Cl683) asked the relevant question 'Solid and slurry sampling in GFAAS-is it worth the effort?' They reviewed the advantages and disadvantages of both ap- proaches and concluded that for samples that are relatively easily brought into solution slurry sampling is not rec- ommended though for refractory samples such as coal and ash which usually have relatively high trace element concentrations slurry sampling is of real advantage and matrix effects are minimal.Russian workers (90/3047) proposed a method for the determination of As Bi Sb and Se in high purity water by in situ concentration of the hydrides on the inner surface of an electrothermal atomizer. The detection lim- its were 2 0.03 0.05 and 3 pg 1-l for As Bi Sb and Se respectively. Deng and Liang (9 1 / 126 1 ) examined the trapping of hydrides of As Bi Ge Pb Sb and Sn directly in a graphite electrothermal atomizer. Absolute detection limits were in the range 60 pg-2.1 ng with RSDs of the order of 0.35-2.4% but with no indication of the concen- tration range over which these were measured. Sturgeon et al.(90/3393) discussed the efficient trapping of volatile hydrides of As Bi Sb Se and Sn on microgram masses of palladium and other platinum group metals in a graphite electrothermal atomizer. Absorption occurs at relatively low temperatures (200 "C) and analytical figures of merit were improved over hydride trapping on the bare tube wall. Absolute limits of detection (30) ranged from 8 pg (Se) to 43 pg (As). The catalytic reactivity of platinum group metals promotes low-temperature deposition of the hydrides by dissociative chemisorption. Haug and Chon- ghua (90/415 1) compared the direct introduction of aque- ous Ge solutions into a graphite electrothermal atomizer equipped with a pyrolytic graphite L'vov platform using chemical modifiers of palladium nitrate or mixed palla- dium and magnesium nitrates with hydride generation techniques onto a pre-reduced coating of palladium or palladium-magnesium nitrate. The collection temperature was 700-800 "C.Recoveries were of the order of 80-90% with detection limits of 30 pg absolute and 3 ng 1-* relative assuming a 10 ml sample volume for the hydride generator. No conclusions were presented from an analytical point of view concerning the relative merits of the two approaches. Zhang et al. (90/3396) applied their trapping procedure for volatile hydrides on palla- dium discussed last year (J. Anal. At. Spectrom. 1990 5 179R) to the determination of Bi Ge and Te in geologi- cal materials and natural water samples.Astruc et al. (911240) discussed a simplified theory of on-line discontinuous detection in liquid chromatography linked with electrothermal atomization. This theory makes provision for the effects of interface measurement on peak shape. Measurement of peak area was found to be essential as the peak height was sensitive to many parameters. Experimental evidence for the theoretical conclusions were presented for the determination of bu- tyltin species. Baeckstroem and Danielsson (9 1 /252 9 1/1303) described a continuous-flow two-step extraction system linked with an electrothermal atomizer for the concentration and separation of Cd Co Cu Fe Ni and Pb in sea water. In the first step the metals were extracted as dithiocarbamates into Freon 113 and in the second step they were back-extracted into a dilute aque- ous mercury(rr) solution. The sample throughput was 30 h-1 with concentration factors of the order of 50-100-fold and extraction yields in the range 80-107%.Detection limits were less than 10 ng 1-l. The development and potential of coupling FI on-line sorbent preconcentration with electrothermal atomization was discussed in several presentations by Welz and co-workers (9 1/C787 91/1433). It was pointed out that while FI as the name implies handles flowing streams FI preconcentration and matrix separation on solid sorbents is a discontinuous process which fits well with the discrete non-flow through nature of ETAAS. While the loading time is wasted with flow through detectors waiting for elution this time can be fitted perfectly into the cycle time of an electrothermal atomizer programme. Lead was used as a model element (9 1 / 1433) with diethyldithiocarbamate (DDTC) as the chelating agent.The chelate was formed on-line and loaded for 60 s onto a 15 pl conical micro-column of bonded silica with octadecyl functional groups (C18) The column was washed with water and eluted with ethanol into a 75 pl collector prior to introduction into the electrothermal atomizer. A 26-fold enhancement of peak area compared with direct introduction was achieved with a precision of 1.9% RSD for 0.1 pg 1-l of Pb and a detection limit (30) of 0.0003 pg 1-*. The procedure was applied to the determination of Pb in sea-water RMs. The almost complete removal of the matrix allowed measure- ments to be made without chemical modifiers.In view of the many and varied manual preconcentration pro- cedures that are starting to be reported in the literature the potential to automate these methods using on-line FI techniques presents an exciting opportunity for the future. Sneddon (9 1 /64 9 1 / 1 529) examined the precision and accuracy of introducing aqueous solutions of Cd Cu and Mn as dry aerosols by pneumatic nebulization and an electrothermal vaporizer linked to an impaction-electro- thermal atomizer AAS system. The efficiency of the impaction device was assessed (9 1/64). On an uncoated electrographite surface particles < 50 pm were collected and particles >85 pm were not. Increasing the jet-to- impaction distance from 2 to 8 mm improved the collec- tion efficiency of particles >50 pm.Russian workers (9 1/469) examined the problem of increasing the sensitiv- ity of ETAAS to allow the direct sampling of gas-aerosol components in clean rooms and laboratories in the semi- conductor industry. They attempted to decrease losses and increase the residence time of the atomized particles in the gas phase. The most significant factors influencing the localization and confinement of the atoms in the atomizer volume were the direct losses caused by the jet- like wear and diffusion processes as well as chemical losses due to oxidative processes. Chinese workers (91/421) determined Ag in waters by electrolytic precon- centration on a tungsten wire cage-shape multilayer disc electrode prior to introduction of this electrode into an electrothermal atomizer.The detection limit was 0.005 pg 1-l and no interferences were observed in the presence of more than 30 foreign ions. The use of a pyrolytic graphite L'vov platform as a cathodic macro-electrode for the preconcentration of Cd prior to introduction into a graphite tube for conventional atomization discussed in last year's review (J. Anal. At. Spectrom. 1990 5 179R) has now been published (90/3505).JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1991 VOL. 6 199R 1.2.3. Fundamental processes A number of the conference reports discussed last year (J. Anal. At. Spectrom. 1990 5 179R) with respect to the use of MS to investigate atomization mechanisms have been published. McAllister (9014 142) showed that gas-phase oxides of As Ga and In may be formed and lost during thermal pre-treatment prior to atomization.Prell et al. (90141 53) studied the real-time mass spectra of gas-phase species in an effort to elucidate the mechanisms controlling vaporization atomization and interferences on Sr by calcium. The results suggest that Sr atoms are produced by desorption of the Sr created by dissociative chemisorption of the oxide. Strontium precursors included the carbide oxide and hydroxide. During atomization Sr was lost by carbide and cyanide formation. The low- and high-tempera- ture carbide losses imply two different formation mecha- nisms. Calcium interferes by increasing low-temperature strontium carbide and hydroxide losses and in shifting the appearance of free Sr to a lower temperature.Hassell and Holcombe (91K757) discussed some of the problems associated with interpreting the mass spectra obtained using secondary ion mass spectrometry (SIMS) to elucidate ETA reaction mechanisms. The signal intensity in SIMS spectra is affected by (z) diffusion (ii) loss to the gas phase (iii) change in analyte morphology and (iv) reactions. These problems can be clarified by use of a cycling temperature programme SIMS technique at varying heating rates. This technique enables one to determine if the reaction is thermally reversible and/or if morphological changes are occurring. Wang et al. (90/3339) studied the atomization mechanism of Cu from a graphite surface with the aid of thermal desorption MS and computer simulations. Results indicated that the desorption of Cu atoms from the graphite surface at individual active sites is the rate-limiting step.Comparing the atomization of Cu metal and Cu oxide powders with aqueous solutions revealed that Cu oxide molecules are dispersed prior to reduction by carbon. The following reactions are proposed for aqueous solutions C~(NO~)~(s)+CuO(ads)~Cu(ads)-~Cu(g) Computer simulations and experimental results indicated that the Cu dimer may form from gas-phase reactions at higher concentrations but that the gaseous dimer is not generated from the graphite surface. L'vov and co-workers have been the most prolific of groups during the period of this review concerning the proposed reduction of oxide by carbides (ROC) mechanism to account for the interaction of metal oxides and carbon within a graphite atomizer.A number of the conference reports mentioned last year (J. Anal At Spectrom. 1990,5 179R) have now been published and are recommended to those readers with an interest in this area of electrothermal atomization research. L'vov (90/209 1) discussed the back- ground to the formation of this mechanism. The mecha- nism is essentially a reduction process based on two concurrent reactions MXO,(S/l) +yC(s)+xM(g) + Yco M(g) + .as)-+MCz(g) MCAg) + (Z/Y)M,O,( S/l)+( 1 + xz/y)M(g) + S O The first of these reactions occurs on the surface of the graphite and the second on that of the oxide. Two assumptions had to be introduced in order to validate this mechanism. Firstly the presence of defects on the graphite surface favouring the formation of excess of gaseous carbides and secondly the assumption that irrespective of the saturated vapour pressure the oxide is reduced via formation of gaseous metal atoms.This mechanism is discussed extensively with respect to the phenomenon of spike formation during the atomization of Al and Mn (90/4141). It was shown that the reduction of Al and Mn oxides by carbon causes a delay in the vaporization of gold and palladium injected separately into the same atomizer. This effect was attributed to the formation of a carbon film on the metal particles of gold and palladium owing to the decomposition of the excess of Al and Mn gaseous carbides formed at the moment of the appearance of the spikes. This effect was observed when mixtures of gold and Mn or palladium and Al were pipetted onto one site as well as onto different sites in the graphite tube (91/216) which in the opinion of these workers tends to confirm the above theory.Bendicho and de Loos-Vollebregt (9 1 / 1 92) examined and characterized the appearance of spikes on the absorption signal of Ai using different sample introduction techniques such as thermospray deposition and suspension and solu- tion introduction. The results were interpreted with respect to the carbothermal reduction of oxides. The distribution of the sample on the surface of the atomizer after drying governs the formation of spikes. L'vov et al. (91/1595) studied the effect of the structure and nature of the distribution of the dry residue of a sample on the formation of absorption signal spikes.Visual observations were made of the change in structure of a dry residue spot of krb203 during carbothermal oxide reduction confirming the simul- taneous occurrence of the process over the entire surface of the sample spot. An increase in the particle size of the sample results in a higher amplitude for the absorption signal spikes and the ROC process becomes continuous for separate Yb203 particles weighing > 10 pg. The gas-phase transfer of metals from a graphite probe to a furnace wall during low-temperature pyrolysis of oxides of Bi Co Cr Cu In Mn Ni and Pb was observed by L'vov et al. (90/3378). The temperatures of the transfers were 400-840 K and agreed well with values calculated for the chemical reduction of oxides by carbon to the metal.The correlation between theoretical and experimental temperatures con- firmed the carbide mechanism of the process. Gilchrist et al. (9 l/C67 1) also studied the vaporization and atomization of Ag As Au Cu Mn Pb Se and Sn from a graphite probe inserted into a constant temperature graphite atomizer. Arsenic Pb Se and Sn were lost in molecular form at temperatures for which no atomic absorption signals were observed. The results for Pb and Sn were consistent with a mechanism where the condensed oxides are vaporized prior to their equilibrium dissociation in the gas phase. Arsenic and Se were rapidly lost from the graphite probe but were found to undergo desorptiodadsorption with the graphite wall prior to their equilibrium dissociation in the gas phase.Silver Au Cu and Mn were found to be atomized directly from the graphite probe surface without the formation of molecules. Katskov and Kopeikin (901 1529) presented experimental results to confirm the validity of an earlier proposed quasi-equilibrium model of metal oxide atomiza- tion in ETAAS. Atomization was found to depend upon the concentration of O2 directly above the sample surface in addition to the temperature and thermodynamic properties of the metal and the oxide. The partial pressure of O2 coming from outside into the boundary layer over the sample the equilibrium pressure of the metal in the boundary layer and the theoretical atomic absorption were determined for a series of oxide decomposition reactions. Russian workers (9 1 /276) reviewed the explosive atomiza- tion of material i.e. spike formation in ETAAS.Fu and Li (9 1/382) investigated the atomization mechanism of Pb in a graphite atomizer using a time-resolved technique. The proposed process involving reduction of the oxide to metallic Pb by carbon on the surface of the graphite tube rather than the thermal decomposition of the oxide in the vapour phase seems to agree with the findings of L'vov. The thermodynamic modelling of processes in electro-200R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 thermal atomizers has also been reviewed by Russian workers (9013265). Gilchrist et al. (90/4143) discussed the gas-phase thermodynamic equilibrium model proposed to account for the shifts in absorbance pulses of As Cr Pb Se Sn and Zn when H2 or CO are added to the Ar purge gas or ascorbic acid to aqueous solutions.The results are consi- dered to be consistent with the hypothesis that the absor- bance pulses are shifted owing to the production of H2 and CO from pyrolysis of the ascorbic acid. The gas-phase dissociation equilibrium of the analyte oxides are disturbed by the addition of H2 CO or C02 to the gas phase. Hydrogen or CO shifts the absorbance pulses to lower appearance temperatures and C02 has the opposite effect. The energetics of desorption of solution and vapour-phase deposited analytes from various graphite surfaces discussed last year (J. Anal. At. Spectrom. 1990 5 179R) has now been published (9115). Lynch et al. (9115) studied the energetics of atomization of Ag Au Cu and Pb in pyrolytic graphite coated uncoated electrographite and glassy carbon tubes following deposition of the analyte as a solution (primary site) and as an aerosol vapour (secondary site).For Cu and Pb an apparent first-order rate of release occurs from all tube types and from both primary- and secondary- deposition sites suggesting that both elements desorb from the surface as individual atoms. Activation energies for Ag and Au were independent of furnace tube type but dependent on the deposition mode. The data from these elements suggest the formation of micro-droplets or caps (primary site) and islands (secondary site) with desorption occurring from the droplet surface or at the metal-graphite interface. Sturgeon et al. (9013393) calculated Arrhenius energies and used them to characterize analyte-substrate interactions for the adsorption of volatile hydrides of As Bi Sb Se and Sn on pg masses of palladium and other platinum group metals (PGMs).The thermal desorption of all analytes from the PGMs was reported to be first order. With respect to the direct analysis of solids or slurries there is little understanding of the chemistry of analyte release and chemical modijication. Jackson and Qiao (901C30 10) considered that when an analyte metal is adsorbed on a refractory particle in a graphite atomizer free atom production can be considered to occur in distinct stages (i) the analyte desorbs (either as the element or some compound) from the particle; (ii) the analyte undergoes gas- heterogeneous- or condensed-phase reactions within the atomizer; and (iii) the analyte is released as atoms. These workers applied the model of McNally and Hol- combe (Anal.Chem. 1987 59 1105) which enables the prediction of the order of analyte release from an examina- tion of the shape of the absorbance-time profile. While no results were presented in the abstract it is to be hoped that this work results in a publication and provides further insight into the chemistry of atomization from slurries. In a series of papers Chinese workers Yan and Lin (9013463 9013588 911381 9111028 9111450) proposed an improve- ment to the Smets method of calculating the rate constant of atom formation (9013588) that avoids the assumption of steady-state first-order kinetics for atom formation and atom loss.They used this improved method to study the atomization process of Al Co Cu Fe and Mo (9111028) and Ag and Na (9111450). Sadly so little information is given in the abstracts from the Chinese journals that it is impossible to say what conclusions were reached. Schron (9011957) discussed the thermodynamic data available concerning gas equilibria chemical reactions in thermal plasmas with regard to their importance in signal interfer- ences in AES AAS or AFS. Thermodynamic calculations covering 42 elements and more than 150 of their halides are included in this report. In general atomization of their halides directly depends on the partial pressure ratio pH21pHX2 (where H = hydrogen and X = halide). The conference report on loss mechanisms discussed last year (J.Anal. At. Spectrom. 1990 5 179R) has now been published (90/3397). In the published paper a theoretical model for the diffusion of analyte species inside a tube atomizer is developed with diffusion considered as the dominant process. The predictions of this model deviate from L'vov's during the initial phase of atomization but the difference disappears in the tail of the pulse. Schwab and Lovett (90/3966) used computer simulations to model the atomic emission signal from graphite furnace atomic emission systems. The excited states were found to be created by absorption of continuum radiation presumably from the tube walls. The time between the absorbance and emission peak maxima was a function solely of the diffusion coefficient of the analyte for a known temperature ramp however questions remain about the temperature dependence of diffusion.Evidence for re-deposition on the graphite surf-ace could be obtained by simulating published absorbance-emission peaks and the position and nature of the emission peak was proposed as a useful tool for graphite furnace diagnostic studies. In a conference report Lovett (9 1/C663) discussed atomization in graphite furnaces and considered the influence of non-uniformity in the atomiza- tion process with the aid of computer simulation. Rademeyer and Vermaak (901C2907) attempted to eluci- date the atomization mechanisms of Si in a graphite atomizer by recording the molecular spectra of gaseous molecular species formed during atomization. No reference spectra could be found in the literature and thermodynamic and kinetic parameters were determined for reactions leading possibly to the formation of oxides and carbides.No actual atomization mechanism is postulated in the abstract. X-ray diffraction and thermodynamic calculations were applied by Jiang et al. (911991) in order to study the atomization mechanism of B and the enhancement of sensitivity produced by a chemical modifier of strontium nitrate. The formation of B atoms is due to the sublimation of solid B derived from the reduction of B203 by carbon. The enhancement of the B signal by SI-(NO~)~ is due to the formation of SrB6 before atomization which decreases the volatilization losses of B206 and retards the formation of B4C. Havezov and Russeva (901291 6) examined the atomi- zation of P from the platform and in the presence of nickel- ( 11) and lanthanum(II1) chemical modifiers by measuring the slopes of the ln(absorbance) versus T-' curves.In both instances the atomization mechanism is considered to be A deteriorated graphite surface could be repaired by treatment with Zr0Cl2. Thermogravimetric and X-ray diffraction techniques were used by Haron et al. (9 1 1 1072) to study the differences in thermal breakdown of hexa- chloroplatinic acid and cis-Pt(NH,),C12. These showed that a solution of H2PtC16 decomposed to yield volatile PtC1 before being converted into metallic Pt. A solution of cis- Pt(NH3)2C12 was converted into metallic Pt in a single step. The differences in the thermal decomposition mechanisms could be one of the reasons for the lower signal obtained from H2PtC1 solutions compared with cis-Pt(NH3),C12 solutions.The appearance temperature of Pt was examined by King et al. (911C794) who found that the form of platinum in the sample affected the appearance tempera- ture. Grigic et al. (9013981) studied the influence of N,N- tetramethylenedithiocarbamate (TMDTC i. e. pyrrolidin- 1-yldithioformate) and tetraethylthiuram disulphide (TETD) on the thermal stability of Hg with X-ray diffrac- tion analysis of the compounds formed. All studies con- firmed in both inert and oxidizing atmospheres the formation of HgS prior to atomization. Radioactive 76A~ was used by Krivan and Arpadjan (9014002) to study the behaviour of As"' and AsV in a graphite ET atomizer. Matrix systems including hydrochloric and nitric acids P2(g)=Wg); P(g)=P*(g)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL.6 201R sodium chloride urine serum and a tetramethylammon- ium hydroxide solution of hair were examined. Various modifier combinations allowed significant stabilization during thermal pre-treatment but in samples containing chloride the presence of HN03 was necessary to stabilize As"'. The conference report from L'vov concerning the theore- tical calculation of characteristic mass values discussed last year (J. Anal. At. Spectrom. 1990 5 179R) has now been published in English (9111453) and Russian (90/3379). L'vov et al. (9013356) also calculated the coeficients of sensitivity losses (2) for the analytical lines of 19 elements when using Zeeman-effect background correction.Calcula- tions were performed without allowing for the emission line width and taking the Doppler profile into consideration. The algorithm was used to calculate the Voigt integral and the hyperfine structure of the lines was taken into account. The coefficients of sensitivity losses for Ca Be Li and Mg lines were dependent on the emission linewidth of the light source. The determined experimental values for the major- ity of lines exceeded the known available values however good agreement between the calculated values and those obtained experimentally by these workers was found. Guo (9OlC2045) used the characteristic masses (m,) to determine Cd and Pb in biological matrices by absolute analysis (ie. without standards). The methods were carefully optimized and STPF conditions applied but the abstract is confusing as to which conditions were critical to enable standardless analyses to be achieved.Frech and Baxter (91/1458) in the published paper derived from the conference report dis- cussed last year (J. Anal. At. Spectrom. 1990 5 179R) considered the temperature dependence of atomization efficiencies in graphite atomizers. The atomization effici- ency (eat) defined as 1 00(m,[calc]/rn,[exp]) was calculated for a spatially and temporally isothermal two-step atomizer for the following elements Ag Al Au Bi Cd Ga Ge In Mn Pb Se Sn Te and T1. For some elements comparative measurements were made with a Massmann-type atomizer operated under STPF conditions. The elements Al Au Ga Ge In Mn Se Sn and Te were found to have temperature dependent atomization eficiencies and are thus susceptible to forming stable volatile compounds at lower atomization temperatures. Improved atomization efficiencies were found for some of these elements when a pre-atomization cool-down step was incorporated into the temperature programme using the Massmann-type atomizer.Generally the results with the two-step atomizer verified the validity of the method and the fundamental constants used to calculate the theoretical m values. The workers demon- strated that comparing characteristic mass values is not a valid method of comparing different atomizer designs and argue that a better criterion for evaluating atomizer performance is atomization efficiency although they clearly stated that from a practical analytical viewpoint detection limit performance is far more important than the characteristic mass.This paper is recommended to those readers interested in this topic for its clarity and style of presentation. These workers are to be congratulated for including an extensive appendix which discusses in some detail many of the underlying factors involved in deriving and calculating theoretical characteristic mass values. Us- ing an alternative approach Zhou and Yan (91/1146) calculated theoretically the coefficient /3 between numbers of ground-state atoms and the peak height absorbance in a GTA-95 atomizer. They compared these theoretical charac- teristic mass values with the values given by the manufac- turer to assess atomization efficiency. For volatile elements such as Cd and Zn the efficiency was found to be greater than 80% 20-50% for medium volatile elements such as Ag Cr Cu Mg and Ni and less than 1Ooh for less volatile elements such as Al Ti and V.Harnly (90lC2994) performed Fourier transforms of transient electrothermal atomizer absorbance-time profiles of species in order to yield the amplitude and phase angle arrays. The amplitude array was found to be concentration dependent while use of the phase angles from the lowest frequencies was an extremely sensitive indicator of subtle changes in the position and shape of the furnace signals. Applying principle component analysis to the phase angle arrays enabled changes to be observed between samples as a function of the type and strength of the acid matrix presence of an interfering agent and age of the graphite atomizer.The summation of ETAAS signals was applied by Berndt and Schaldach (91/236) to increase the detection limits so that there was no need for prior separation or preconcentration. The accuracy of the results obtained were confirmed by separation and preconcentration. Wegsche- ider et al. (9111252) reduced the raw data from ETAAS absorbance-time profiles to eight descriptors and applied a fractional factorial design method to derive empirical models useful in the prediction of sensitivity from single atomization curves. The models were applied to assess the interferences on Rh Se and Sn. The prediction of sensitivi- ties from the typical changes in peak shapes due to frequently encountered interferents such as aluminium chromium iron magnesium sodium and zinc were gener- ally within 5-1 5% relative error.1.2.4. Interferences Electrothermal atomization has been commercially avail- able for 20 years and is an area of research starting to suffer from the 're-inventing the wheel' syndrome. Nowhere is this more pronounced than in the examination of interferences and chemical modifiers. This reviewer has to voice concern at some of the work accepted for publication. The responsi- bility for this situation lies not solely with the authors but also with the referees of papers for scientific journals. A criterion for a scientific publication is that of originality and sadly this criterion is being overlooked rather too often. While palladium in all its various forms continues to dominate the field there are many areas and elements that require different chemical modifiers. The potential of vanadium(v) proposed by Tsalev et al.and discussed last year (J. Anal. At. Spectrom. 1990 5 179R) has now been published (90/4 1 45). Kelemen et al. (9 1 /844) examined molybdate vanadate and molybdovanadate as chemical modifiers for Ba. These ions form complexes with interfer- ing phosphate and silicate ions and carbides. This last effect may improve the sensitivity for Ba by suppressing the interactions between Ba and graphite. A combined V205-Si modifier was employed by Bishop (90/3347) to determine Ba in sea-water. This combined chemical modifier aided injection sample drying graphite tube lifetime and elimi- nated most of the sea-water components in a slow pyrolysis at 1 150-1 200 "C.With atomization at 2600 "C and a gas flow rate of 60 ml min-' the characteristic mass was 5.6 pg per 0.0044 A s and the detection limit (la) was 2.5 pg in a 25 pl sample. For sea-waters with a Ba concentration of 5.6-28 pg l-l precision was 1-2% and accuracy 2-3%. Work on phosphate modifiers continues to generate large numbers of papers especially in the clinical field where these modifiers do not suffer from the precipitation prob- lems that palladium based chemical modifiers can have. Subramanian (90/2233) examined mixtures of 5 g 1-l (NH4)2HP04 + 0.4 g 1-l (NHJ2Cr2O7 + 2% v/v HN03 and 50 g 1-l ascorbic acid+0.2 g 1-l iron as iron(n1) nitrate as possible chemical modifiers for the determination of Sn in aqueous leachates from copper pipes soldered with lead-tin (NH&HP04 + 0.5 g 1-' Mg(NOj)2 + 0.2% V/V HN03,2 g 1-'202R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL.6 solder. The first chemical modifier was preferred based on its higher sensitivity and thermal stability. The detection limits were 1.7 and 12.8 pg 1-l for absorbance and integrated absorbance measurements respectively. It is not clear from the paper why a palladium based chemical modifier which is generally recommended for Sn was not included in the study. Yet the paper is recommended as one where the sheer volume of analytical data presented leaves no doubt that the method is reliable and robust. A whole collection of reports from various workers (9 1/250 90/C2 100 9 1/422 90/337 1) discussed the use of phosphate modifiers with and without 0 ashing for the determination of Cd and Pb in whole blood and other biological matrices.Nothing that has not already been published for over ten years is presented in these papers. Huang et al. (90/3474) found that the use of ascorbic acid overcame an interference from iron on the determination of Cr in soil samples after acid digestion with HN03 HF and HC104. For the determination of V Bauza et al. (90/C294 1) claimed that ascorbic acid increased the thermal stability of the V signal with a pyrolysis temperature of 600 "C but depressed the peak height sensitivity. This depression could be overcome by adding ammonium sulphate. A group of Chinese workes proposed the use of calcium nitrate as a chemical modifier for A1 (91/169) B (90/C2048) Be (91/165) Dy (90/C2048) Ge and Sn (91/181).In all instances calcium nitrate was claimed to increase the thermal stability of the analyte enhance sensitivity allow lower atomization temperatures and reduce interferences. The use of alkali metal fluorides as chemical modifiers was examined by Nater et al. (90/3330 90/1971) for the determination of A1 (9013330) and Si (90/1971). An HF-CsF mixture was used for A1 to form AlF3 which sublimes at 129 1 "C. The subsequent gas-phase atomization of AlF3 occurred rapidly and produced sharp clean peaks. The use of CuF was also examined and appeared to eliminate interferences from organic acids but at the expense of shortened tube lifetimes. For Si an HF-LiF mixture was employed. Conversely Wu et al.(91/918) discussed the interference of fluoride on the determination of Al Ca and heavy metal ions in environmental samples. The choice of chemical modifier and electrothermal atomization conditions for the determination of A1 is still a popular topic. Quinonero et al. (90/325 1) found increased sensitivity for A1 when using Mg(N03) instead of NH4H2P04. The paper by Redfield and Frech (90/2056) accounting for the pre-atomization losses of A1 was dis- cussed last year (J. Anal. At. Spectrom. 1990 5 179R). However when a laboratory with over 11 years experience of determining A1 in serum (91/C753) is still refining the method clearly the chemistry of A1 in a graphite electro- thermal atomizer is still open to discussion. Cunningham et al. (91lC753) found a narrow plateau in the ashing profile around 1400 "C with a considerable reduction in sensitivity for temperatures of less than 1300 "C. The variation between different tubes could move the signal off this plateau. Magnesium nitrate did not change the shape of the profile but 1% v/v HN03 did produce a flatter ashing profile.Triton X-100 was found to cause a reduction in peak and integrated absorbance for aqueous standards but not for serum samples. This effect was to some extent overcome by the addition of KN03 to the standards but the investigators were still not satisfied with the minor daily differences shown between KN03 and serum based standards. The interference of indium in the atomization of Ag Bi Cd Sn and TI in the presence of hydrobromic acid was investigated by Yudelevich et al.(90/2228) with the aid of a two-step atomizer. The results indicated that the indium interference is due to the formation of InBr which leads to a higher bromine concentration in the gas phase thus shifting the reaction equilibrium towards the formation of metal bromides. These bromides are removed from the observation volume by diffusion before their dissociation is complete. Several different chemical modifiers have been proposed for a variety of elements during the last year. Aluminium nitrate was proposed for Be (91/49) in the presence of HF and HN03. For the determination of Bi (9 1/289) cadmium and lead improved the sensitivity palladium impaired sensitivity and cobalt molybdenum and zinc produced double peaks. Several different modi- fiers were tested for B (91/C805) but none overcame the memory effects associated with this element and all showed a similar level of performance. The best sensitivity was obtained with calcium chloride.The matrix interference of iron(Ir1) on Cr was overcome by pneumatic nebulization into the graphite atomizer (90/3513). The mechanism of interference was claimed to be a physical process resulting from the 'occlusion' effect of iron(1Ir). The different degrees of interference of carbon black on the atomization of Al(OH)3 and A1203 (9 1 /7 1 7) and iron oxides (9 1/ 1 127) were exploited by Yoshimura and Huzino (9 1/717 9 1/1127) in order to allow the differential determination of the various species. The matrix effect of boron on the determination of Fe (90/3416) in primary circuit cooling of a nuclear power plant was investigated.Styris and Prell (91/C672) used MS to investigate the possible mechanisms of action of magnesium nickel and palladium chemical modifiers. Modification of the analyte vaporization behaviour was shown to be due to interfer- ences in the formation of low temperature gas-phase oxides hydroxides carbides and polymers of the analyte. The principal mechanisms involve homogeneous condensed phase reactions. These workers stated that a clearer under- standing of chemical modifier mechanisms is required in order to advance their use further. Sakurada et al. (90/4020) examined the mechanism of stabilization of Pb with a palladium chloride chemical modifier using X-ray photo- electron spectroscopy.Lead as lead chloride in the presence of palladium chloride was reduced to a metallic form at 200 "C and remained at 600 "C after all the chlorine had been removed. The action of the modifier was discussed in terms of Pb reduction and alloy formation with palladium. The application of reduced palladium was discussed by Schra- der and Beach (91/C674). While they claimed that there is substantial evidence that reduced palladium is the neces- sary intermediate species to achieve optimum analyte stabilization there is also a large body of evidence that shows that unreduced palladium also works extremely well. There is a need in this area for a thorough investigation of the mechanisms of palladium chemical modification. Erard (9 1 / 1066) examined a variety of chemical mod~ers for the determination of Se in cereals with Zeeman-effect background correction.The palladium and magnesium combination was the most suitable for wheat samples containing approximately 1.0 pg 1-1 of Se but for those containing 0.1 pg 1-' the modifier was unsuitable because of double atomization peaks that were thought to be due to poor atomization cloud uniformity. Chinese workers (9111 70) claimed that a mixed chemical modifier of ammo- nia solution and palladium increased the sensitivity of determinations for As Cd Cu and Pb. From the abstract it appears that ammonia solution was added to the strong acid (HzSO4) sample solutions to form ammonium salts prior to the addition of the palladium. Bozsai and Karpati (9 1/286) discussed the high sulphate interferences on As and Se in mineral waters.This paper is obviously the precursor of the conference abstract discussed last year (J. Anal. At. Spec- trom. 1990 5 179R) and a fuller investigation of the interference has now been published (9 1 /7 1). Sampson (9 1/C8 14) discussed the use of a mixedpalladium-platinum chemical modifier for the determination of As and Se inJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 203R biological materials containing phosphate and iron using continuum source background correction. The platinum catalyses the reduction of phosphate and retards the atomization of iron thereby reducing the interferences but has a depressive effect on the sensitivity which can be retained by the addition of palladium. However this method is not completely foolproof as shown by the statement that integrated absorbance was subject to a residual interference which did not affect peak absorbance.The action of platinum in combination with CO to reduce the spectral interference from iron on the determina- tion of Se was discussed last year by Peile et al. (see J. Anal. At. Spectrom. 1990 5 179R) and resulted in correspon- dence between Radziuk et al. and Peile et al. (91/849). Radziuk et af. (9 11849) considered that the effects claimed for CO are due solely to the presence of platinum although Peile et al. (911849) believe that a more complex mecha- nism is involved than a simple temporal separation of iron from Se. Martin and Williams (90/2057) found that a combined chemical modifier of palladium and copper nitrates with an H2-Ar purge gas allowed the determination of Se in microwave digests of bovine liver against aqueous standards.This method was developed for bovine liver as this was considered to be a more complex tissue sample than the heart tissue which was to be analysed in a further study. Interested readers are referred to the paper by Fairhurst et al. (Fairhurst J. Lloyd B. and Delves H. T. Anal. Chim. Acta 1987 97 197) which discussed the acid digestion of heart tissue for Se determinations. In a tungsten atomizer copper was found by Fudagawa and Kawase (911936) to be the most suitable chemical modifier to prevent losses of Se. The thermal stability of As was assessed by Krivan and Arpadjan (9014002) using radioactive isotopes and for a variety of chemical modifiers including nickel molyb- denum palladium and tungsten.Palladium was also exam- ined in mixtures with hydrogen peroxide. No conclusions were given in the abstract as to which was the preferred chemical modifier. Subramanian (9013 5 53) directly deter- mined As in urine with platform atomization and a chemical modifier of either palladium or palladium plus ascorbic acid. Smeyers-Verbeke et al. (91/846) found that a chemical modifier of 6 pg of palladium nitrate in combina- tion with 500 pg of ammonium nitrate allowed the direct determination of Cd in undiluted urine against aqueous standards. This is a lower amount of palladium than the 50 pg recommended earlier by Yin et al. for the determination of Cd in biological materials (Yin X.Schlemmer G. and Welz B. Anal Chem. 1987,59 1462). However Smeyers- Verbeke et al. (911846) claimed that the use of a large amount of palladium should be avoided in order to reduce the background absorbance due to sodium chloride. Although palladium and/or magnesium nitrate are excel- lent chemical modifiers contamination of the modifier with the analyte element of interest can cause severe problems for determinations at very low levels. Bulska et af. (9014 148) found that for the determination of Cd in serum the endogeneous Cd contamination in the palladium-mag- nesium nitrate modifier could be removed by pre-injecting the modifier and heating to 1100 "C. After cooling the sample was injected and a maximum pyrolysis temperature of 800 "C applied. The very features that are an advantage for the use of palladium or other metals as chemical modifiers can cause problems when one has to determine elements in a matrix of palladium or platinum.Popova and Bratinova (9013482) described the determination of Pb in diaminodichloropalladium and diaminodinitropalladium. Thermal analysis X-ray diffraction and infrared spectro- metry indicated that the decomposition process takes place at 356 "C for diaminodichloropalladium and at 222 "C for diaminodinitropalladium with the principal product of this process being metallic palladium. No reducing agent was required in order to produce metallic palladium. Arpadjan et al. (90/4 146) discussed the determination of Ag Cd Co Fe Ir Ni Mn Pb Rh and Ru in pure platinum and palladium. Three groups (9013332 90/4150 911474) have investi- gated the interference of chloride matrices on the determina- tion of Mn.Stobbaerts and Deelstra (911474) found that interferences from KCl and NaCl in foods and biological materials could be overcome by using a suppressor such as L-ascorbic acid. In contrast to many other published observations they claimed that contamination of samples during preparation is negligible. Hulanicki et al. (9014 1 50) found that a chemical modifier consisting of an acidic component (phosphoric acid) and a basic component (calcium ions) combined with platform atomization im- proved the accuracy of Mn determinations in the presence of inorganic salts. While the use of the platform alone improves the determination chemical modification com- bined with platform atomization had the greatest effect.Pressurized atomization in Ar at 304 kPa was employed by Fazakas and Zugravescu (9 1/59) for the determination of Pb in the presence of 0.1-5% NaCl. The depression of the Pb signal was not observed at this pressure and the effect was ascribed to a longer residence time in the atomizer due to pressurized atomization producing a higher vapour temperature and therefore better dissociation of molecular compounds. In a conference report Fisher et al. (911C779) discussed the use of reducing gases as gaseous chemical modifiers for the determination of V and Al. Problems were encountered by Temminghoff (90/4 1 58) in the determina- tion of Cd in digested plant material with a modifier of 1% NH4H2P04 0.2% Mg(N03)2 and 0.04% AlC13.Automati- cally pipetting the sample solutions followed by the modi- fier produced double peaks due to problems with the air gap between the two solutions on the platform. The failure of the solutions to mix and the presence of an air bubble caused sputtering during drying and double peaks from wall and platform atomization. This problem was solved by adding 5% v/v butanol as a surfactant to the modifier solution. This allowed the solutions to flow and mix on the platform and produced single peaks. Dittrich and Fuchs (9012060) used a chemical modifier of lanthanum ions to stabilize the phosphate ion for the determination of P by FANES and mofecular non-thermal excitation spectrometry (MONES) (9013483) of the PO and HPO species. With FANES a detection limit of 90 pg was achieved which is an improvement of approximately 60- fold compared with conventional commercially available ETAAS equipment.Ajayi et al. (9012229) compared the temperature gradients and chemical interferences on Cd Mn and Pb for one normal and two modified tubes. A higher vapour temperature and lower chemical interfer- ences from magnesium and sodium chlorides were ob- served when the tube temperature gradient was reversed through a reduction in the wall thickness towards the ends of the tube. However these signals show more noise as a result of the greater emission continuum from the ends of the tubes which may well have a deleterious effect on detection limits. 1.2.5. Developments in technique Multi-element analysis using ETAAS is an area that was previously the preserve of a few specialized research groups throughout the world but interest is increasing and several novel approaches besides the established but not widely used continuum source atomic absorption spectrometry (CSAAS) are starting to find their way into the literature.In a conference report Harnly (911C743) touched upon the Achilles heel of multi-element techniques with a title that204R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1 99 1 VOL. 6 included the phrase 'the search for photons in the UV'. Both CSAAS and ETAES suffer from poor detection limits at wavelengths lower than 280 nm compared with conven- tional ETAAS. Several improvements have been suggested such as pulsing the continuum source detection with a linear photodiode array (PCS-LPDA) and hollow anodes or cathodes with the FANES technique.All three approaches have detection limits comparable to ETAAS and are compatible with multi-element determinations. The appli- cation of CSAAS to the analysis of biological materials was discussed by Miller-Ihli (9012 199 9013584). Various sam- ple preparation methods were used for ETA including direct analysis wet ashing dry ashing and slurries. The use of compromise atomization conditions did not appear to limit significantly the accuracy and precision and good agreement was found with RMs. Moulton et al. (901C3033 9013499) discussed in more detail the modifications to a PCS-LPDA-CSAAS system and wins the prize for the longest acronym in this year's review. With flame atomiza- tion LPDA detection resulted in an improvement in the signal-to-noise ratios by a factor of 3-9-6.2 for the Cd Co Fe Ni and Zn resonance lines between 2 13.9 and 248.3 nm.A pulsed source in conjunction with LPDA improved the signal-to-noise ratios by a factor of 5.8-8.5. Two Russian workers (9011520 9013412) appear to have developed a multi-element ETAAS system using a continuum source and separation of the elements as volatile compounds. In order to solve the problems of sample introduction and atomization the elements were isolated as volatile com- pounds followed by gas-chromatographic separation. The separated compounds were fed individually into the atom- izer in a stream of inert gas. This technique appears to be applicable to non-dispersive modes of AE AF and AA using non-selective detection. A fast sequential ETAAS instrument has been described by Batie et al.(901C3014). With this new design it is claimed that up to 24 elements can be determined in a single sample in rapid sequence. The application of this instrument coupled with a glow-discharge atomizer to the determination of powders was described by Carter et al. (90/C3012) and by Bernhard et al. (90/C3013) to the determination of titanium and nickel based alloys. Taga et al. (9013362) proposed a simultaneous ETAAS system dedicated to the determination of Ag and Cu. The reso- nance radiation from Ag (328.07 nm) and Cu (327.40 nm) hollow cathode lamps were introduced through the same spectral bandpass of the monochromator. When the mixed radiation was passed through the atomizer two peaks corresponding to Ag and Cu appeared in the absorbance- time profile.The addition of palladium or platinum improved the separation of the peaks without mutual interference and by measuring the peak heights the simulta- neous determination of Ag and Cu could be achieved. A single-channel time-divided simultaneous multi-element AAS system was described by Nakamura and Kubota (9012 196). A multi-element hollow multi-cathode lamp for Al Cu and Fe was used with a tungsten strip atomizer and no monochromator. The power to the three cathodes was supplied in the order Cu A1 and Fe. Consequently the emission beam for each element has the same light path but occurs in a distinct time sequence. By forming suitable volatile molecular species in an ETA non-metallic elements can be measured.Gomez et al. (90/3508) evaluated ET molecular absorption spectrometry for the determination of fluorine in biological samples after mineralization and from the same group Corvillo et al. (91/1585) applied this method to the determination of fluoride in sea-water. Aluminium monofluoride is formed in a graphite atomizer and the molecular absorption at 227.45 nm was measured. The interference by chloride was removed by adding a chemical modifier of NH4N03 with 0.01 mol dm-3 Al+O.Ol mol dm-3 Sr. In sea-water the detection limit was 8-10 ng ml-1 with a peak height precision of 5-7% for 0.2 ,ug I-' of F-. Yu and Huang (901C2037) determined Br by measuring the molecular absorbance due to InBr at 248.3 nm with radiation from a chromium lamp in an electrothermal atomizer.A graphite platform greatly reduced the background due to I n 0 and increased the signal-to-noise ratio 1 0-fold. The difficulties of determining chlorine and bromine by FANES were discussed by Dittrich et al. (90/C315 1). The presence of easily ionizable elements such as potassium and sodium changes the electronic pressure of the helium discharge which affects the intensity of the atom and ion lines. Naumann et al. (9013398) applied comprehensive fluctuation analysis to the investigation of the compromise conditions required by the FANES source. These showed that the optical conductance of the spectrometer has to be improved especially for wavelengths t 2 10 nm in order to exploit the potential detection power of FANES fully.In a series of reports Harnly et al. (90/C3036,90/C3 149 9013498) outlined the operation of a FANES system with the furnace QS a hollow anode (HA-FANES). The cathode is a carbon rod which runs down the centre of the atomizer. This arrangement results in a discharge with excellent stability and no need for electrical shielding. The emission signals were dependent on the current density and viewing position with more intense signals obtained when viewing from the top and left of the discharge. The detection limits for Cd and Cu were 0.8 and 1.5 pg respectively which are comparable to those obtained with conventional FANES. Yizai and Harrison (901C2049) proposed a tungsten loop furnace (WLF) coupled with a hollow cathode discharge (HCD) and AES which is much simpler to build and operate than the FANES system.The technique was applied to the determination of B in glass at a concentration of approxi- mately 0.04 mg 1-' with a precision of (5%. Chakrabarti et al. (9013348) proposed a new analytical technique of pulsed and transient modes of atomization using a commercially available cathodic spluttering atom- izer. The transient mode uses a high-power discharge for the rapid production of analyte atoms whereas the pulsed mode uses a discharge of much shorter duration and higher power to achieve even faster atomization. The faster kinetics of atom formation result in better sensitivity than for steady-state atomization. The sensitivity of the tech- nique is sometimes similar to or greater than that of ETAAS but no analytical results are presented in the abstract.The preliminary analytical characteristics of furnace atomization plasma excitation spectrometry (FAPES) for Ag were presented by Blades and co-workers (901C3083 9013975 911C662). A detection limit (30) of 0.3 pg was achieved. This source combines the high transport effici- ency characteristic of a graphite electrothermal atomizer with the powerful excitation characteristics associated with a high temperature plasma gas. The source has been improved from that originally described by using an isothermal integrated contact cuvette (9 1/C662). Hadeishi and Le Vay (9111508) described the construc- tion of stable long lasting high intensity discharge lamps operated at high frequency between a few hundred Hz and 150 kHz that were used to excite the spectra of refractory elements by utilizing their halides. The lamps are small enough to fit between the poles of a magnet and conse- quently can be used in the direct analysis of solids by Zeeman-effect AAS. A multiple mode laser diode was employed by Ng et al.(9111548) as the line source for ETAAS. Several emission lines are supplied from a single diode with a line width narrower than the absorption profile. The separation between the lines was approximately 0.3 nm. These characteristics are beneficial for simulta- neous background correction and peak ratio measurements.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 205R The detection limit for Li in 10 pl sample volumes was 0.6 pg l-l which is comparable to that obtained from hollow cathode ETAAS systems.The linear dynamic range was approximately two orders of magnitude. Butcher et al. (90lC303 1) presented the first use of a commercial graphite furnace and probe atomizer with laser-enhanced ionization (LEI). Laser-enhanced ionization uses a tunable laser source to excite atoms to an energy level near the ionization continuum while a d.c. electric field is applied across the atom cell (see also section 3). After excitation a fraction of the atoms ionize generating electrons and positively charged ions. The electrons move in the electric field to the cathode and generate a current that is the signal used to determine elements by LEI. The sample on the probe was inserted into the graphite atomizer after a constant temper- ature had been reached with the aim that atomization under near isothermal conditions should minimize gas-phase reactions between the analyte and its matrix.The potential of the method was evaluated by the analysis of biological RMs. A multichannel background correction technique for pulsed laser-excited atomic fluorescence spectrometry with ETA (ET-LEAFS) was described by Sjostrom (90/4156). A detection limit of 0.05 pg of Ga in aqueous solution was obtained for the corrected signal. It was shown that black- body radiation scattered laser light and molecular fluores- cence can be corrected for by the simultaneous background correction technique. McCaffrey and Michel (9 1 / 1 5 33) applied capacitive-discharge heating to a small anisotropic graphite tube which provided heating rates of up to 65 "C ms-l with a final temperature of 2500 "C.Preliminary investigations with Pb indicated that sample deposition on the unbroken lamellar graphitic planes provided about 1.5 times the peak area and peak height sensitivity obtained from sample deposition across the lamellar planes. The vapour from a graphite ETA was introduced through the anode tube into a magnetron rotating d.c. arc for AES measurements by Slinkman and Sacks (9111 547). This ensured adequate sample-plasma interaction and produced AES detection limits for Cd Cr Cu Fe Mg Mn and Zn in the ppb range. Kitagawa et al. (90/3331) described a novel approach to separate the atomic vapours of Cd Hg and Pb prior to AA measurements. The method consists of a separative column atomizer (SCA) which is an alumina tube packed with graphite granules coated with gold or platinum or mixed with tin and heated to 800-1400 "C.Different retention characteristics and peak broadening were observed for different packings. These workers claimed that retention of the vapour through the SCA might be useful for the separation of co-existing elements and removal of their interferences. No analytical results were presented in the abstract. Forsyth (91/258) described the optimization of a combined gas chromatographic-AAS system equipped with a quartz tube atomizer for organotin determinations. The atomizer was heated at 800 "C with a 60% H2-40% air gas mixture the calculated detection limits (30) were 7.3 pg of Sn as Me3BuSn 8.2 pg of Sn as Me2Bu2Sn 16.6 pg of Sn as MeBu3Sn and 17.2 pg of Sn as Bu,Sn.Haug (91/54) overcame the problems of poor reproduci- bility in a commercial automatic sampler system for ETAAS by continuous de-gassing of the dispenser rinse solution. This resulted in improved performance by avoid- ing the development of gas bubbles in the dispenser syringe. Brueggemeyer and Fricke (90/C303 5) discussed the possi- bility of increasing the dynamic range in ETAAS by using more of the information contained in the atomic peak profiles. Because the deviations from linearity become more pronounced as the absorbance increases the upper points on the peak may be reduced in size while the points at lower levels (the leading and trailing edges) remain largely unaffected. Hence it is possible to calibrate using selected parts of the peaks with a variety of options such as partial peak area point by point regression analysis dAldt compar- isons and multivariate pattern recognition techniques.The only drawback to this procedure is the need for both the sample and standard peaks to have the same temporal position and a similar shape however the workers claim that this can be overcome by using for the standard a known amount of analyte in the sample matrix of interest. A number of workers continue to investigate the possibil- ities of reducing the length of ETA programmes. Zugra- vescu and Fazakas (9 111 392) found the maximum pyrolysis temperature for Ag in carbonated mineral waters to be inadequate and nebulized the samples directly into the graphite tube at 150 "C followed by atomization without any pyrolysis step. The detection limit was 1.5-3 pg with a precision of 2-3%.The pyrolysis step was removed and the drying step optimized by Hinds et al. (90lC1684) for the determination of Pb in soil slurries. This reduced the ETA cycle time from 160 to 40 s. Analysis of a high carbon content soil showed good Pb recoveries even when no modifier (normally a palladium-magnesium nitrate mix- ture) was used. By making alterations in the software and firmware controlling the autosampler Slavin et af. (90/2089) developed fast STPF methods for several ele- ments in over ten different RMs. Total cycle times including autosampler washing and sample uptake were less than 60 s. As a chemical modifier and the pyrolysis step were omitted the background signals were larger and use of Zeeman-effect background correction was usually required although continuum source background correction was not tried.Bradshaw and Slavin (90/2090) used fast sampling STPF conditions with Zeeman-effect background correc- tion to determine As Pb Se and T1 in coal and fly ash slurries. The slurries were prepared automatically by ultrasonic mixing and the ETA programmes omitted the pyrolysis step and chemical modification. With a cycle time of approximately 60 s and calibration against aqueous standards reasonable agreement was obtained with the certificate values for the SRM materials examined. Hinds (91/C675) found that for the determination of trace elements in precious metals fast atomizer programmes were possible.The inclusion of a chemical modifier and pyrolysis step was unnecessary owing to the chemical modifier action of the matrix metals for volatile analytes. Shuttler and co-workers (90/C30 1 1 9 l/C688 9 ~ C 7 8 9 ) discussed a range of tests that can be used as quality assurance (QA) procedures to test instrument performance. The determination of Pb in sodium sulphate (90/C3011) illustrated the differences in performance of different platform systems and how well STPF conditions were achieved. The performance of a previously proposed QA procedure (Slavin W. Manning D. C. and Carnrick G. R. Talanta 1989,36 17 1) was examined in an interlabora- tory exercise (91/C789). This allowed an assessment of the stability of ETAAS instruments and performance of differ- ent platform designs in a wide variety of laboratories.1.3. Chemical Vapour Generation This section will present developments in the understand- ing instrumentation and technique of chemical vapour generation as used in conjunction with AAS detection. Applications to specific analyses of which there are many particularly environmental and biological samples and to systems employing AFS and AES detection will be reported in other ASU reviews. In a new development as an alternative to the detection of free analyte atoms hydrides of the analyte have been absorbed in chromogenic solutions and determined by UV-visible molecular absorption spec- trometry (Ge 91/88; and As and Ge 911440).206R ,JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 1.3.1. Hydride generation Fundamental studies of chemical interferences in the liquid phase occurring during the generation of metal hydrides have been undertaken since the early 1970s notably by Smith (Analyst 1975 100 300).It was concluded that 'before this type of accessory [hydride generation (HG)] for AA for the determination of As Bi Ge Sb Se Sn or Te is used it is necessary to ensure that the matrix being analysed does not interfere'. Any analyst new to HG would be well advised to consult Smith's paper along with more recent work dealing with specific elements and matrices. Sanz and co-workers (91/1435 911C1788) have carried out an in- depth study of the interference of iron and mercury in the determination of Sb both experimentally and by means of a computerized speciation model. They observed that in a variety of acidic media as the iron concentration (<0.1 mol 1-l) increased initially there was no reduction in the AAS signal for Sb; at higher concentrations the signal decreased with increasing iron to a level (>0.3 mol 1-l) above which the signal was again independent of iron concentration.With mercury however there was no interference at concentrations up to approximately 0.5 mol 1-l. The difference in the effect of iron and mercury was attributed to the formation of complexes in the reacting solution whereby the actual redox potential of the Fell1 -Fell system was higher than that of the Sbv-SblI1 system while that of Hgl1 -Hg was lower. Gas-phase interferences by arsenic and selenium in the determination of Te have been overcome by preferentially vaporizing arsenic in the presence of Br2-H2S0 and reducing Se to a non-interfering state with KI (90K4083); other interferents were removed by precipitation with lanthanum hydroxide.The mechanisms whereby hydrides decompose into free atoms are not fully understood. It has been suggested (9 1/168) that in an electrically heated quartz cell atomization of some hydrides is not solely a gas-phase process but that the cell surface also plays a part. The atomization of Ge from GeH in an electrically heated quartz tube was studied by Bing et al. (9 1/C2 139) and found to be different from that of ASH or SeH,. At temperatures below 960 K most of the GeH decomposed on the surface of the tube above that temperature O2 in the gas entering the tube caused serious interference by the formation of GeO.Based on a theoretical model Quingle et al. (91K2140) confirmed the conclusion of Welz and Melcher (Analyst 1983 108 213) that the essence of thermal decomposition of arsine is a free radical process with turning points in the atomization efficiency when the concentration ratios of H2 and O2 are 1 :2 and 2 1. General studies of interferences included one on the metal ion enhancement of the rate of hydrolysis of sodium tetrahydroborate in an acetate buffer at pH 5.0. This effect has consequences for the timing of operations in the determination of As"' (90/1164) by both manual and continuous flow methods. Interferences in the determina- tion of Bi by nickel cobalt and copper were found (9111 74) to arise in the liquid phase while those by antimony lead and selenium occurred in the gas phase.The latter effect was overcome by heating the intake tube of the quartz atomizer and the former by addition of 8-hydroxyquinol- ine and thiourea. The effects of 26 interfering species on the determination of As and Se were examined by Castillo et al. (90/ 1227). They found widespread interference the extent of which was greater when a flame rather than a quartz tube was used for atomization. The interference of copper and nickel in the determination of As and Se has been studied in depth by Janssen (90/ 19 10). Correction by standard addi- tions was found to be problematical. When the matrix composition is not known determination by ETAAS was recommended. Preconcentration and reduction of interfer- ence effects was achieved by extraction of As and Se with dioctyltin dinitrate solution in CHCl and butanol (91/1034).The elements were stripped from the extract with 2 mol dm- hydrochloric acid prior to determination by HG-AAS. The detection limits for As and Se were 0.4 and 0.8 ng ml-l respectively. The determination of Pb in samples frequently requires pre-oxidation; potassium hexa- cyanoferrate(II1) (91/180) or nitroso-R salt (9111970) have been recommended as oxidants to be followed by oxalic acid (9 1 / 1 80) and ammonium cerium nitrate (901C2044 9 1 / 1496) as interference suppressing agents. In order to determine total Sn in organotin compounds it was neces- sary to treat the compounds with bromine in carbon tetrachloride (90/3484); the hydride generating solution contained approximately 1.2% water in DMF.Atomization was in an electrically heated quartz tube and the detection limit was 1.5 ng ml-l. A volatile species of Cd presumed to be the hydride has been generated by carrying out the reaction with sodium tetrahydroborate in DMF in the presence of diethyldithiocarbamate (90/ 1 994). The detec- tion limit was 9.1 ng of Cd. The interference of selenium and tellurium on the cold vapour determination of Hg was used as the basis of an indirect method for the determina- tion of Se and Te (90/3049) and gave detection limits of the order of 1 x with a relative error of 10% in the determination of 0.2 pg of Se or Te. Minor modijications to discrete sample processing systems for hydride generation have been found to give worthwhile improvements in performance.A commercially available hydride generation cell was modified to facilitate sample and reagent introduction and washout without removal of the cell and interruption of the Ar supply to the HS- entrained air flame (90lC2877). The modification speeded routine analysis. Xuan and Chen (91/131) reduced the sample volume required for an analysis from 5-25 ml to 20 pl by mounting a micro-cup inside the HG cell and injecting sample and reagents into it. The absorption signal was measured in a heated quartz micro-cell. The need for a carrier gas in an HG system was overcome (911348) by adding sodium hydrogen carbonate to the sample solution prior to generation of the hydrides in order to prevent explosion.The detection limits for As Bi and Sb were 7 14 and 5 ng respectively. When a 'long-seam' (slotted quartz tube) quartz atomizer was used (91/448) for the determina- tion of Sn with an N2-H2-entrained air flame gas phase interference by e.g. antimony arsenic and selenium was much smaller than when an electrically heated quartz tube was employed; the usable acid concentration range was also greater. The sensitivity was 0.64 ng ml-l per 1% ab- sorption and the RSD 4.9%. The sensitivity of discrete systems can be improved by collecting the generated hydride in a cold trap (liquid air cooled) followed by its rapid release into the atomizing cell. This approach was used in the determination of Ge in iron meteorites (911241) and gave a detection limit of 0.03 pg g-l in the original meteorite.Flow injection is one of the most convenient ways of mechanizing sample handling in HG systems. In addition it reduces sample and reagent consumption can lower detection limits and speeds analysis (90/C2982). Flow systems can facilitate elaborate sample pre-treatment prior to the generation of the volatile compound. Selenium hydride generation usually takes place in acidified samples. The range of acceptable sample pH has been widened to neutral and alkaline solutions by adding sodium tetrahydro- borate before acidifying the solution with hydrochloric acid (91/C746 91/2549). Marshall (911457) has described three variations of u basic manifold whereby interferences in HG may be eliminated. In one system pre-reduction with potassium iodide was used in the determination of elements present in207R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL.6 several oxidation states (see also 91/48). This may also be used as a means for speciation of for example As (9 1/1380) by adjustment of the pH of the sample. In another interferences by noble metals during the determination of As Bi Sb and Se were reduced by the addition of tellurium. Finally interferents in the determination of As and Se were removed by means of an in-line ion-exchange column (see also 9 1/C776). General matrix related interferences were overcome by means of automation of the standard additions process. In-line ion-exchange columns were used for analyte preconcentration in the determination of Bi and Se at ultratrace levels in natural waters (9 1/1076 91/2653).Bismuth was extracted on chelating ion exchangers CPG- 8Q and Se on anion exchanger D-201; both were eluted with 1 mol dm-3 hydrochloric acid directly into the HG system. Detection limits were 1 ng 1-l of Bi and 2 ng 1-l of Se. Separation of the gaseous hydride from the liquid stream is a key step in achieving the best results from FI-HG-AAS. Three designs of gas-liquid phase separators have been examined (91/C774) viz. a glass U-tube a silicon rubber tubular membrane and a PTFE sheet membrane. All were found to be satisfactory for the determination of As in a water CRM (IAEAN-4). The membrane separators gave dry uncontaminated gas streams; the sheet membrane had the shortest rise and memory times. A PTFE sheet membrane was also found to be satisfactory (901351 1) for the determination of Bi in NIST SRM 361 Special Low Alloy Steel AISI 4340 a leaded gun metal and BCS CRM 347 Electronic Flowsolder the detection limit was 0.17 ng of Bi.Stibine was generated electrochemically in a thin layer electrolytic cell using FI and a microporous PTFE tube was used to separate the gas from the liquid carrier stream prior to transfer into an N2 stream for AA measurement in a heated quartz tube (90/C2035). It was claimed that the system suffered from less interference and contamination and was fast and convenient. The application of the system to the determination of As has also been investigated (9K2137). Interferences from Cull SeIV and Hg2+ were observed but could be reduced by the use of appropriate electrode materials.The elimination of Hg2 + interference on the determination of As required the removal of the former by chelation with thiourea and vitamin C. From time to time graphite furnaces have been used as the atomizing cells in HG-AAS. When operated in a constant flow mode at 1800-2300 "C the sensitivities for As Sb and Se were similar to those obtained by conven- tional quartz tube (91/233). Tin however was strongly adsorbed on the graphite inlet tube. A 1000-fold excess of arsenic antimony bismuth selenium tellurium and tin as interfering elements produced no more than a 12% depres- sion of the signal. Reactions between water and H2 and heated graphite gave rise to an increased background. Greater sensitivity can be achieved by collecting the analyte element in the furnace prior to atomization in order to generate a transient signal. The collection of As Bi Sb Se or Sn may be either as the hydride at a relatively low temperature 200 "C (90/3393) or as the element after thermal decomposition of the hydride at a higher tempera- ture 800 "C (9013047 90/4 15 1).Detection limits were in the range 10-40 pg. Hydride collection was effected by means of adsorption directly on the graphite surface or chemisorption on a palladium or other platinum group element coating deposited on the graphite furnace wall [Bi Ge and Te (90/3396 91/Cl858)]. Detection limits ranged from 8 pg for Se to 43 pg for As (90/3393). The higher temperature collection of the elements on the untreated graphite inner surface of the furnace gave detection limits for As Bi Sb and Se in high-purity water of 2.0,0.03 0.05 and 3.0 ng ml-l respectively. Germanium was collected at 800 "C on a platform or graphite foil treated with a palladium-magnesium modifier (90/4 15 1).The reproduci- bility of the determination was 2% RSD with a detection limit of 30 pg ('3 ng 1-l) in the hydride generating solution. 1.3.2. Preparation-separation of volatile organometallic compounds for chemical vapour AAS The use of volatile phases can in some instances greatly enhance the eficiency of sample introduction. In a wide ranging study Castillo et al. (90/C2945 9 1/C 1779) exam- ined the utility of chelates as the volatile phase in analytical atomic spectrometry. The compounds synthesized were acetylacetonates of Al Cr Cu Fe Mn Ni and Zn; trifluoroacetylacetonates of Al Co Cr Cr and Fe; hexa- fluoroacetylacetonates of Co and Fe; and dithiocarbamates of Co and Cu.The compounds were synthesized in an FI system collected on a sintered glass filter which could be heated to the sublimation temperature of the chelate (=250 "C) and carried in a nitrogen gas flow into a heated silica atomization tube for determination by AAS. Chromatographic separation prior to determination by AAS may be used either for speciation of organometallic compounds present in the sample or as a means of separating organometallic compounds synthesized during the sample pre-treatment from an interfering matrix. This approach has been proposed by Rigin (9013412) as a basis for dispersionless and multi-element analysis using non- selective detection. Speciation of organotin compounds in water was effected by extracting the compound into pentane as diethyldithiocarbamate complexes (901342 1).After drying under reduced pressure the complexes were derivatized to form pentylated alkyltin compounds that were separated by GC and determined by quartz furnace AAS. Detection limits were 0.1-0.4 ng of Sn for the various organotin species. High-performance liquid chromato- graphy has been used to separate alkyllead compounds (90/ 1 990) and anion-exchange chromatography to separate biogenic organoarsenic compounds (90/C30 1 6). In both instances post-column treatment of the emuent was neces- sary before determination by AAS could be carried out. Ionic alkyllead compounds were separated in a flow system as their tetramethylenedithiocarbamate complexes ethy- lated purged from the reaction mixture and volatilized into a quartz tube furnace; detection limits approached 0.1 ng. Organoarsenical compounds were photo-oxidized on-line with low intensity mercury lamp irradiation (maxi- mum=254 nm) to arsenate and fed to a continuous flow arsine generator.Sample streams were mixed with potas- sium persulphate in sodium hydroxide solution prior to entering the photoreactor for 20 s. Arsine was extracted from the reaction mixture through a Teflon membrane. The detection limit was 5 ng ('50 ppb). Tetramethyl-and tetraethyllead in petrol were separated and determined by selective volatilization of the sample ( 10-50 pl) in a special reactor at a temperature of 1 50- 180 "C directly into an FAAS system (901 1 174).The detection limit was 3.0 ng of Pb. Tin and organotin species were selectively reduced at pH 2 to their corresponding volatile hydrides (90/1504). The hydrides were collected in a liquid nitrogen cooled trap from which they were fractionally volatilized into an Ar-H,-entrained air flame for the determination of Sn by AAS. The detection limit was 0.3 ng and the RSD was =2.5%. Methylmercury in aqueous samples has been determined by reaction with sodium tetraethylborate in order to convert the non-volatile monoe- thylmercury into gaseous methylethylmercury (9 1/892). The methylethylmercury was then thermally desorbed from the column and analysed by cryogenic gas chromatography with cold vapour AFS.The method also allowed the simultaneous determination of labile Hg" species through208R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 the formation of diethylmercury and of dimethylmercury which was not ethylated. The detection limit for methyl- mercury was 0.6 pg (= 0.003 ng 1-l) for a 200 ml sample. 1.3.3. Mercury by cold vapour generation Three methods of cold vapour AAS viz. batch FI continuous flow all with or without amalgamation were evaluated by Welz and Schubert-Jacobs (90/C205 5 ) with respect to sensitivity freedom from interferences cost speed and ease of operation. The batch system was the most sensitive but suffered from the greatest interference was the slowest and used the most reagent. Flow injection was the fastest used least reagent and suffered least interference but owing to the small sample volume was the least sensitive.Continuous flow with collection of Hg by amalga- mation combined the freedom from interference and ease of operation of the FI system with larger sample volumes to give a method well suited for determining Hg at the 0.1 pg 1-l level. In a series of papers Daniels and Wigfield reported a detailed examination of the determination of Hg by a cold vapour method. They considered reagent volume optimization (9 1 /259) acidic versus alkaline reduction (9 1/260) signal reporting options (9 1/26 1) and instrumen- tal adjustments leading to increased sensitivity (9 1 /262). Particular attention was given to studying the effects of pH and of cysteine on the reduction procedures; the feasibility of speciating Hg in these sulphydryl-complexed and free forms was demonstrated.A number of modest changes in technique that have led to worthwhile improvements in performance in particular circumstances have been reported. For example following oxidation of the sample with 0.4 mol dm-3 H2S04 and reduction with SnCl the Hg vapour generated was flushed out of the reaction vessel by injection of water (91/2496). The gas was dried by passage through a condenser; a detection limit of 0.001 ppb was achieved. The detection limit in the determination of Hg by cold vapour generation was significantly lowered when the Hg vapour generated by tin(@ chloride reduction was collected on platinum (90/3328) or gold (90/2231) in a graphite furnace and subsequently atomized by raising the furnace temperature under gas stop conditions. Using 50 ml samples detection limits were of the order of 0.1 ng 1-l.In some circumstances Hg may be directly extracted from aqueous or gaseous samples by amalgamation on a column filled with HAuC1,- modified packing material (9 11903). The Hg was released by heating and determined by cold vapour AFS. The detection limit was 4 x g. The thermal desorption behaviour of metallic Hg and some organomercury compounds from a gold column was studied by Baeyans and Leermakers (90/1989). They found that the proportion of the Hg released from its different compounds was determined by the temperature dependence of the decomposition process and diffusion of Hg in gold metal. Under optimum conditions of temperature and carrier gas recoveries ranged from 100% for metallic Hg to 80% for methylmercury chloride.A calibration procedure for the determination of Hg by CVAAS in conjunction with collector preconcentration has been described (9 1/25 13). A special valve linked a closed Hg-saturated gas flow with an open flow of the carrier gas to the AA spectrometer. By this means defined volumes of Hg-saturated gas were injected into the system and consequently the problems of contami- nation and loss associated with manual methods were avoided. In the region of 1 ng of Hg within-run precision was better than 1 Yo day to day precision better than 2% and results for various biological RMs agreed with the given values. Wigfield (9 1/C687) quantified adsorption losses in the connecting tube of cold vapour generating apparatus for 1 1 types of tubing.Losses as great as 40% were observed for tubing 30 cm long with 5 mm id. at a flow rate of 500 ml min-l. Munaf et al. (9013046) found that matrix interference by anions cations and organic compounds in the determina- tion of Hg was much reduced when a continuous microflow system with tin(@ chloride in sodium hydroxide solution as the reducing reagent was used. In more recent work (91/C1697) the same system was used with sodium borohydride in sodium hydroxide solution as the reducing agent sodium peroxodisulphate as the oxidizing agent and cadmium chloride as the catalyst; the detection limit was 0.15 pg 1-l. The system was applied to the continuous monitoring of Hg in waste water.From a study of reaction time and matrix effects Ngim et al. (9011949) found that interference effects in the microdetermination of Hg in undigested biological samples could be considerably re- duced and peak height sensitivity increased by using a reaction mixing time of 1 min before bubbling with carrier gas in order to liberate the Hg. Reagent consumption was also significantly reduced. The interference of silver in the determination of Hg can be severe because silver ions may be preferentially reduced to the metal by the reagents and the metal itself can block the release of Hg vapour by amalgamation. Although FI methods reduce the problem the use of a masking agent thiourea or increased acid concentration was recommended (9 1/C 1857) when silver is present in considerable excess. The interference of I - in the determination of Hg was used as the basis of an indirect method for the determination of I- (90/1286).A known and excess amount of Hg2+ was added to the sample to form a stable complex with I-. The free Hg2+ was reduced to Hg with a nitrogen containing sulphate which did not react with the complex. The liberated uncomplexed Hg was then determined by AAS. The recovery of I- from water samples was 92- 1 OOOh. 1.4. Spectrometers 1.4.1. Light sources Hieftje (9 1 / 1 5 1 7) has made a critical comparison of sources and detection methods and there has been one comprehen- sive review with 156 references (90/3322) on theory production and operation of EDLs as light sources in atomic spectroscopy. There is also a review (91/1086) on the use of a continuous wave tunable dye laser in high resolution atomic spectroscopy.The advantage of such devices is that they eliminate Doppler widths of spectral lines. The replacement of hollow cathode sources by compact and inexpensive tunable semi-conductor diode lasers was discussed at length by Hergenroeder and Niemax (9 1 / 12 10). It was claimed that 60% of the elements now measured by conventional AAS could be determined with such lasers if the second harmonic wavelengths generated in non-linear crystals are included. High intensity arc lamps have been described by O’Haver et al. (91X767) specifically for continuum source AAS (see section 1.4.5) and by Zander et al. (90/1974) for general purposes. The latter operating with Ar achieves its improved luminance by special electrode design.Angus et al. (90/C1814) showed that enhanced emission from an HCL can be achieved by the addition of a small amount of N2 to the normal fill gas. Work is still proceeding on the establishment of electron densities and temperatures. The growth of radiation inten- sity with time (‘temporal evolution’) of a pulsed HCL has been investigated by Griffin (911C1952) and the processes of evaporation and excitation using different metals in the hollow cathode have been examined by Mierzwa (91/C2112). From discussions on the characteristics of two Russian types of HCL for AA analysis (911340) it appears that theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRYy AUGUST 199 1 VOL. 6 209R strong self-absorption observed in one type over a particu- lar range of currents is not observed in the other.Construc- tional differences are not detailed and the reasons for such variations in performance are therefore not clear. Complet- ing a series of studies on the influence of rotating magnetic fie& on an HCL discharge the effects of rotation perpendi- cular to the hollow cathode axis have now been described (90/2258). The radiative properties of the discharge varied with field strength fill gas discharge current and the cathode element although it is said that effects on the line intensities of analytes were generally small. The behaviour particularly ionic:atomic line intensity ratios and current-voltage dependence of low and high voltage internal anode HCL discharges have been investi- gated by Phillips et al.(9111452). Of all the noble gases Ne provides the highest intensity of Cu I 324.7 nm in both high and low voltage versions. As a secondary radiometric standard in space applications a Pt-Ne HCL has been investigated (9 1 17 1 6). This produces a uniform distribution of lines over the range 11 5-350 nm. 1.4.2. Optics An equation which more rigorously treats the phenomenon of stray light has been derived from Beer’s Law (9 113 1) and the parameters derived from fits of the equation to AA data allow quantitative measurement of the proportion of stray light in a particular AA experiment. It has been re- discovered (901C2885) that the measuring range in AA can be extended by ‘manipulating the angles of the burner head with respect to the HC beam’.Rotation of the burner head was often used 20 years ago and the burner stem of at least one early commercial instrument was calibrated in degrees for this purpose. Two papers referred to improvements in gratings. Coma aberration of conventional holographic grating (where fringes are generated using plain and spherical wavefronts) are largely eliminated if the fringes are generated using aspherical wavefronts as reflected from a spherical mirror. It was claimed (9012267) that resolution and energy throughput are both improved. Varied line space (VLS) gratings which substantially improve imaging in grazing incidence spectrometers cannot be made using holographic techniques. It has been shown however (9 117 12) that VLS and holographic gratings can be made to approximate to each other in an imaging sense to the lowest order and that they are therefore interchangeable although the groove profiles may be different.The use of optical waveguides in the development of both AAS and AES techniques was reviewed and discussed by Bauner and de Laffolie (9 111 064). 1.4.3. Detectors A new generation of charge transfer devices (901C2957) provides a range of very high performance detectors from single element replacements for the PMT to arrays with some millions of detector elements. The use of unique read- out modes destructive and non-destructive read-out rapid scan etc. make these particularly attractive in the develop- ment of new methodology in the atomic spectroscopy field. 1.4.4. Background correction de Loos-Vollebregt pointed out (911C2117) that when in Zeeman-effect background correction the a.c.magnetic field is parallel to the optical axis (i.e. ‘longitudinal’ field) no polarizer is required and therefore loss of half the primary source intensity is avoided. In recent work a C-type longitudinal magnet (strength 8 kG at 100 Hz) was used with a transverse heated integrated contact tube. The dynamic range of Zeeman analytical curves can be extended by a factor of about five by changing the shape of the magnetic field so that intensity measurements can be made at three different field strengths (zero intermediate and high) giving additional curves of reduced slope. The limitations of presently available background correction systems using continuum source or inverse a.c.Zeeman effect were enumerated by Schlemmer (9 11C2 1 18). A new system based on the platform work of Frech and the transverse heated tube with longitudinal a.c. Zeeman egect as described by de Loos-Vollebregt (see above) is said to give improved light throughput and detection limits and to be ‘Perkin-Elmer’s contributions to the next decade of graphite furnace analysis’. Arriaga and Stephens (9013954) detected structured inter- ferences from extraneous absorption lines which lie close to the analyte line of interest in a.c. Zeeman-effect background corrected AAS by harmonic analysis of the signals obtained. Calculations were performed based on a model using normal Zeeman splitting and measurements made for the iron-platinum interference at 217.9 nm and the interfer- ence between the isotopes of lithium at 670.8 nm.The analysis showed that the amplitudes of the harmonic frequencies vary with the position and half-width of the interfering line. It appeared that even two lines of identical profile and centre frequency which overlap completely could still be distinguished provided that they did not show the same Zeeman multiplet. Amplitude ratios could serve as a parameter to detect the occurrence of such interfer- ences over a limited concentration range and provide an instrumental method of warning that errors may have occurred in the measurements. The paper discussing the influence of furnace radiation on Zeeman AAS signals discussed last year (J. Anal. At. Spectrom. 1990 5 179R) has now been published (9014 1 44).Li (901C2950) examined molecular spectra produced by Co-NaC1 Fe-NaC1 Ni-NaC1 and Se-Fe systems in a continuum source background corrected ETAAS instrument. Examples of insufficient good and overcorrection were found. For the Ni-NaC1 system the 232.0 nm Ni line is adjacent to the 234.2 nm maximum absorption band of NaCl and the non-symmetry of the absorption band causes overcompensation. These workers considered that the Se-Fe system is similar to Ni-NaC1 and proposed that the over-compensation seen with this system is due to the molecular absorption of FeCl. Yet as Martinsen et al. have shown this over-compensation is most probably due to iron atomic line absorption (Martinsen I. Radziuk B. and Thomassen Y. J. Anal. At. Spectrom. 1988 3 1013). 1 A.5. Continuum source and simultaneous multi-element AAS Work has continued on CSAAS since the earliest papers of O’Haver (ARAAS 1979 9 41) principally in the Univer- sities of Maryland (901C1773 901C1775 9011997 901C3033 9013499 9 1K767 9 1K1988) and Florida (901 199 1 9013982).The wavelength modulating systems of the original instruments have latterly been replaced by photodiode arrays and this is the part of the system which differs the most between these two laboratories. The array used in Florida provides a range of 2.5 nm per line with an acquisition time for a single spectrum of 22 ms. Since last years review when the use of a graphite furnace was reported the air-C2H2 flame (9011991) and ICP (9013982) have all been employed as the atom source. Maryland’s 128- or even 256-pixel array gives a range of 0.2-0.4 nm and an acquisition time of less than 5 ms is possible with a fast data system. Also associated with the latter laboratory is a new pulsed xenon arc lamp (911C767) giving a peak intensity of 260 times normal when operated at 200 A and210R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL.6 resulting in improvements in a signal-to-noise ratio (and therefore by implication also in detection limit) by factors of 6-8.5 for Co Fe and Zn. Considerations for minimum noise levels with this type of system have been set out separately (901C1775) and patents have been applied for (91/1421). O’Haver (911C1861) has also shown that im- proved signal-to-noise ratios result when the spectral width of the light source is comparable to that of the absorption line (instead of much less as required by Beer’s Law) and has discussed the implications for CSAA.Harnly (911C1882) pointed out that with diode array detectors read-out noise rather than diode shot-noise adversely affects detection limits. Shot noise can be made dominant again if source intensity is increased or if a spectrometer with improved light throughput can be employed e.g. a low- resolution monochromator may offer a 25-fold improve- ment over an Cchelle. The use of both line and continuum sources in simulta- neous AA spectrometers has been discussed by Golloch (9 11 1 193). In the former case HCLs were coupled withfibre optics and the monochromator was replaced by a polychro- mator with a wide entrance slit. The performance of this assembly was said to be comparable to that of a conven- tional AA instrument.Multi-element AAS using a continuum source and a wavelength-modulated kchelle polychromator was described by Miller-Ihli (9013584) who gave details of a prototype instrument being used for graphite furnace analyses of biological materials. An kchelle spectrometer in conjunction with a charge injection device has also been used to achieve multiplex detection with continuum source and electrother- mal atomization (9 1 11 549). Absorbances were computed by scanning through all orders of the Cchelle spectrum. Two simultaneous AA spectrometers using conventional light sources have been described by manufacturers. A Hitachi instrument (911245 1) with graphite furnace and Zeeman-effect background correction uses four hollow cathodes and four ‘independent gratings’ in what is pre- sumed to be a static system.Good performance is claimed for As Pb Sb and Se. A dynamic system from Thermo Jarrell Ash (911C2096) is capable of measuring two ele- ments at a time by ETA and four by FAAS. Rapid drives for both wavelength and source selection are synchronized. The entire wavelength range from 185 to 900 nm can be covered in 20 ms so that more than one wavelength and light source can be monitored during even a fast transient ETA signal. 1.5. Instrument Control and Data Processing 1.5.1. Instrument control The requirement for more efficient and rapid methods of analysis has led to the development of a number of sequential systems for FAAS. In one such design (90x30 1 5 ) the analysis of up to 24 elements in a single sample has been reported in a 4 s per element cycle.All elemental and flame conditions were controlled by a computer which also captured stored and reported calibration and sample data. In addition it was possible to change flame conditions automatically from air-C2H2 to N20-C2H2 if required. In one of the new commercial designs rapid galvanometer drives have been employed to perform simultaneous multi- element analysis (9 11C1865). In this instance one galvano- meter drive oscillated an HCL selector mirror capable of alternating the direction of light from each of four lamps many times per second. The lamps up to a total of eight were configured in two tiers on a vertically adjusted dual focal array.Synchronized with the lamp selection mecha- nism a second galvanometer drive oscillated a grating between the corresponding wavelengths. All parameters of the spectrometer optical furnace atomizer autosampler and preparation station were programmed using an external personal computer equipped with trademark software. This software has also been used for the simultaneous determi- nation of element pairs (As-Se and Pb-TI) in soils and waters (90/C3041). The speed of the system is optimized using three characteristics firstly two elements are mea- sured in a single atomization secondly the sample drying stage is eliminated through the use of aerosol deposition and finally sample preparation including chemical modifi- cation is not only automated but is carried out during the furnace cycle and therefore without increasing the analysis time. Following standardization a set of quality control samples are automatically run to verify both the calibration and the method.At each point decision making software determines the next action. Once verification is complete samples are prepared and run along with a single recovery sample for each. Recoveries are calculated and reported for each sample. Sample data are saved to disk only after a successful verification has been performed otherwise the calibration and verification are repeated along with the sample batch. In another commercial design 12 elements per minute have been determined (90/C3013) although the system also uses non-thermal atomization for the analysis of undissolved solid samples (901C30 14).This system uses the integral computer to monitor the internal standard to improve accuracy and a correction known as Pamiers method to calculate concentration ratios. Automated systems for the routine application of ETAAS have also been described. In one such system based on an IBM-AT computer features included automated dilution when the calibration range was exceeded; automated correc- tion throughout the run when standard sensitivity or baseline drift occurred; a novel sample blocking mode; automated recognition of spiked blanks unspiked and spiked samples reference samples standards and sample blanks; programmability to include the order in which solutions (samples modifiers blanks etc.) were picked up by the autosampler; the ability to perform automated digestions in the furnace using a separate digestion solution such as nitric acid; the ability to perform a tube treatment prior to each atomization using a separate treatment solution; and the ability of the autosampler to prepare standards in autosampler cups from a stock solution (911C1949).The last of these features was reported to provide better accuracy than existing commercial systems due to the larger volumes used. A computer controlled apparatus for on-line element preconcentration for FAAS has been described (9 1 /C 1 8 1 6). The system permits the pre-selection of any enrichment factors for each sample the print-out of the preconcentra- tion protocol the automatic element enrichment of up to 20 samples and off-line or on-line operation with AAS. The elements Al Cd Co Cu Fe Ga Mn Ni Pb V and Zn could be enriched from various sample materials and volatile hydrides of As Sb Se and Sn could also be produced. An automated on-line column preconcentration technique for FI-AAS has also been developed which enables enrichment factors of between 90- and 180-fold for seven elements at a sampling rate of 13 h-l (9011948).A second report (9112563) described the automation of an FI system for the hydride generation of Se and subsequent determination by AAS. Pre-treatment of the sample and the details of the automated equipment were reviewed. An automated solid sampling analysis system consisting of a powder sampler an integrated microbalance a transport and handling system for the sample boat a microcontroll- ing unit for all necessary steps and an appropriate AAS instrument has been described (9111 506) and its potential for routine metal determinations discussed.A novel design of Cchelle polychromator has been used to develop a flexible bench top multi-element analyser. Radia- tion at 120 wavelengths was transferred by optical fibres toJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 I VOL. 6 21 1R a bank of photomultiplier tubes. Up to 15 wavelengths could be monitored simultaneously and the instrument programmed to make rapid measurements at up to 60 wavelengths without alteration of the optical fibre selection (9 1/C766). The instrument was equipped with wavelength- modulation background correction facilities which permit operation in simultaneous multi-element analysis mode by continuum source AAS or ETAAS.All procedures in this system are controlled by a microcomputer with programs written in Pascal and pre-compiled at execution. Steady- state or transient signals at different wavelengths could be displayed simultaneously on the computer screen. Peak height and peak area measurements for analyte and background signals are produced for each channel and the data processing facilities are compatible with commercial spreadsheet packages. 1.5.2. Data processing A number of publications have dealt with the use of microprocessors for calibration. In one such paper the principle of piecewise fitting of a curve to a calibration graph was discussed with the aid of both a BASIC computer program and a flow diagram for a programmable calculator (9 111 525).A method for evaluating procedures for approxi- mation of calibration graphs has also been reported (9013388). Six approximation methods as used by major companies producing AAS instruments were evaluated using a model representation of the real absorbance minus the concentration dependence. It was clear from the results of this study however that the function which proved best for approximation over a wide range of real calibration graph characteristics using three reference solutions was not necessarily the best function to use when there was a much more limited change in the characteristics. A compar- ison of least squares and least mean squares techniques in AAS has also been made to investigate the problem of outliers in calibration (9 1/26 10).The factors contributing to the optical limitations in AAS have been reviewed (91/1012). An equation to correct the effects of these limitations was proposed where Ad and A are the measured and the corrected absorbances (xI12)E is the full half width of the geometrical slit (xIl2)* is the full half width of a profile of full foot with xl c is the concentration of an absorbing layer of effective thickness 1 and a is a factor given by a=( 1 +a2/ln2)4 where a is the classical damping factor. The equation is reported to extend not only the linear part of the calibration graph but also modulate its slope. A method for estimating the collisional widths of atomic absorption lines in atmospheric pressure flames (see section 1.1) has been reported by O'Haver et al.(90/3399). The method is based on the shapes of calibration graphs formed from absorbances determined by fitting a computer model to transmission profiles measured with a continuum source. A measurement precision of f 0.1 pm was reported for elements Cd (228.80 nm) to Sr (460.73 nm) using a spectrometer with a spectral bandpass of approximately 5- 12 pm. The application of a multi-curve calibration technique in AAS has been used for the determination of 6Li and 'Li isotopes (9111043). Because of the non-linearity of the calibration graph the results obtained proved unsatisfactory with regular calibra- tion methods and so a series of standards were divided into several segments and a series of calibration graphs fitted to quadratic equations.The optimum calibration graph for a specific sample is selected based on computer computation. Finally although log-log plots are commonly found in the literature their mis-use and mis-interpretation can lead to incorrect conclusions for example about the linear dynamic range of a calibration. The mathematical and statistical implications of such transformations have been discussed (9 1 /C682) using examples from atomic spectroscopy. Methods for the estimation of random errors and detec- tion limits have been the subject of a review with 20 references (9 111 5 15). A comparison was made between theoretical and experimental detection limits for FAAS based on instrumental standard deviation and non-instru- mental errors. A number of publications have also concen- trated on quality assurance programmes.Although in the past many such studies have been based on controlling large scale routine analysis an approach for controlling the quality of a modest number of analytical results has been discussed (91K1529). In addition a number of more specific applications of quality control such as commercial software packages for ETAAS data (9 l/C1832) quality control schemes for geochemical data (90/4043,90/C4057) reference sample data bases (90/C4100) and soil and vegetation analysis (90/C4060) have been reported. In another such paper dealing with quality control procedures for monitoring whole blood and plasma trace metal levels by AAS (91/1088) a number of methods were discussed in detail. These included the use of SRMs and in-house standards for calibration Levey-Jennings charts West- gard's multi-rule quality control schemes and additional information obtainable from cumulative sum charts in order to detect analytical drift from the mean and standard deviation index or z-score plots to show analytical bias.The general application of computers in data processing has been the subject of several publications particularly where automatic processing of results by computer facili- tates a shortening of analysis time increases the infonna- tion content or improves the accuracy of results (9 1/ 1074 90/ 1 5 16). The transfer and manipulation of atomic absorp- tion data into third-party software has also been widely reported (90/C1826 9011 932 90/C2869). The implementation of fast Fourier transforms via the use of residue factored look-up tables has also been investigated (90/1976).The principles of factored look-up tables were reviewed and a gate-level pipelined adder architecture presented. The basics of the quadratic residue system was also discussed. Advanced ways of handling laboratory data interchange and storage problems such as the Analytical Data Interchange and Storage Standards (ADISS) project have also been reported (90K2938). Monte Carlo procedures have been used to study the effects of activation energy and frequency factors in ETAAS (91K1745). Signal summation has been used for the improvement of performance and an increase in the amount of sample in the direct AAS analysis of solids (9 1/2506). Time-resolved AAS signals of several individually weighed solid samples were stored and later summed by personal computer to yield a signal covering all the individual samples with an S / N which was improved according to the Ni law.This made it possible to detect low concentrations in a number of samples without the need for preliminary preconcentration steps. The distributions of trace elements in powdered samples has also been the subject of a statistical treatment (9K1824). Here it was found that the distributions are frequently asymmetric occasionally with several maxima resulting from a small mass fraction with a high analyte content. Statistical treatment using the Poisson distribution has important consequences in for example the prepara- tion of reference materials the determination of minimum sample amounts required to obtain normally distributed results and for the methodology of solid sample analysis.1.5.3. Chemometrics A major aim of chemometrics is to aid the optimization and212R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 interpretation of results in a more time effective way. The use of simplex and its various modifications factorial designs with multivariate response functions signal pro- cessing and pattern recognition methods all fall under this general heading. Although the area has been reviewed (91/C737) each of these areas has been the subject of a number of publications. Simplex methods are widely used to maximize the S/B ratio of selected peaks although the choice of response function step size and starting condi- tions and the difficulties encountered with factors that have different significance may sometimes limit the utility of the technique.A set of programs have been presented which enable the simplex optimization of a process that involves ten interdependent variable factors (9 1/945). The programs written in Applesoft BASIC are fully listed. They are based on the modified simplex method and are suitable for use in optimizing any instrument process or situation that involves 2- 10 variables. Evolving factor analysis has been applied to FI analysis data (9 1 / 1 162). The concentra- tion profiles and spectra of Bi3+ and the bismuth chloride complexes BiC12+ through BiC13-were estimated from injection of bismuth perchlorate into a flowing stream of 1.0 mol 1-l HC1.The estimated spectra compared favour- ably with previously published spectra of the complexes. An exploratory study on medium-based robust regression methods for linear calibration in AAS analysis has been conducted (90/1984). A strategy was developed either to avoid the influence of outliers or to reject them in the case of homoscedasticity in routine analysis. Several robust methods based on the median i.e. the single median the repeated median and the least median of squares were introduced and evaluated. The methods were used for the analysis of real data obtained from AAS studies. The application of different multivariate and discriminant analyses and factor analyses have been used to characterize polluted soils by AAS (91/23 1) while principal components analysis correspondence factor analysis and spectra map analysis have been applied to spectroscopic data (90/29 15).In both instances the information obtained by displaying the objects and variables in each reduced factor space was compared and the redundant information shown. An objective evaluation of precision requirements for geoche- mical analysis using analysis of variance has also been discussed (90lC4065). The partial least squares method(PLS) and its application in analytical chemistry has been the subject of a review of 58 references (9013459). Baxter and Oehman (90/3974) have presented the possibility of mathematically resolving spectral interferences using multicomponent standard addi- tions and PLS modelling. The method was applied to the determination of Au in the presence of cobalt at the 242.8 nm wavelength with continuum source background correc- tion (90/3974) and As in the presence of aluminium and phosphates (9 VC749).More accurate results were ob- tained by the application of this procedure than by conventional single-component standard additions methods. However it should be noted that the successful application of this procedure requires knowledge of the sample constituent causing the spectral interference and that its concentration can be increased by standard addi- tions. The possibility of resolving spectral interferences in ETAAS by applying multivariate calibration has been investigated (9013974). Resolution is achieved using multi- component standard additions (the so-called generalized standard additions method or GSAM) combined with PLS modelling.The multivariate calibration method PLS-GSAM was described and its use illustrated by application to the determination of Au by ETAAS in the presence of cobalt at the 242.8 nm wavelength where severe spectral interference problems are observed using continuum source background correction. The inherently multiparameter character of atomic spectra has also been used to illustrate the utility of multilinear or regression- based calibration techniques (90lC2922). A critical assess- ment was made of the use of the 1024 pixels available on a conventional linear photodiode array when used with a spectrally segmented photodiode array spectrometer. Prin- cipal component analysis has been reported for a number of applications including a classification of polluted and unpolluted mussels using data obtained by ETAAS (90/1180) and in data analysis of heavy metal pollution in the sea (1911354).A mathematical treatment based on the analysis of the atomization pulse and its derivatives has been reported for the description of the atomization in a graphite furnace (90/1985). By observing that the absorbance rate presents a maximum and a minimum at times t and t, respectively and that for all the experimental situations studied the temperature had reached a plateau before tpt. the rate equations were solved using boundary conditions at t rather than at the beginning of the pulse. Simple and physically valid equations were obtained for the determina- tion of the rate constants (k and kJ for the formation and loss of atoms.Monte Carlo algorithms and the computa- tional power of a supercomputer have been combined with spatially resolved atomization profiles obtained by using a spatial isolation wheel (9 l/C1983). This combination of experiment and theory has led to an insight into the energetics of processes occurring in the graphite tube which are not available directly from conventional techniques such as Arrhenius plots. Finally the nature of the signal from a graphite furnace has been investigated using computer simulations of atomization and diffusive removal (90/3966). 2. ATOMIC FLUORESCENCE SPECTROMETRY In previous reviews atomic fluorescence spectrometry (AFS) has been presented under several headings princi- pally as inductively coupled plasma and laser-excited AFS (LAFS) (ASU-Atomization and Excitation J.Anal. At. Spectrom. 1990 5 189R and 205R). These continue to be the most active research areas though in terms of practical applications cold vapour methods for Hg gener- ation of hydrides (or other volatile compounds of the analyte element) and gas chromatographic systems com- bined with conventional light sources appear to be the most widely used. Sophisticated light sources such as the ICP or laser do not readily transfer into the routine analytical laboratory. Atomic fluorescence spectrometry is at its most useful when conventional light sources are used and when there is either very little sample matrix or the analyte element can be separated from it. As atomic fluorescence is used along with molecular fluorescence as a diagnostic aid in studies of combustion and of the plasma this use of fluorescence will also be presented in this review.As an aid to the practical application of AFS a table of data on 76 elements collected from 322 references has been compiled by Winefordner’s Group at the University of Florida (9011 186). The table includes the excitation and fluorescence wavelengths atom reservoir excitation source limit of detection and comments on the types of samples analysed. A review of Chinese publications on AFS has been prepared by Yingqi (9013470).JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 213R 2.1. Discharge Lamp Excited Atomic Fluorescence The light sources used in practical AFS are generally EDLs or pulsed HCLs.An exception was the use by D'Silva and Zamgow (901C3230) of a helium afterglow discharge detec- tor (HeAD) into which H2S and PH3 were introduced in order to produce intense emission of S and P. This radiation was used to excite the fluorescence of S and P as they were atomized by a second HeAD through which the effluent of a GC system was passed. By this means the effect of the intense emission of CO in the HeAD was reduced. When a xenon$ashtube was used to irradiate Cd atoms in an ICP (90/3300) approximately 87% of the fluorescence signal observed at 643.8 nm was related to double resonance fluorescence. This phenomenon is a disadvantage in analytical applications owing to its large potential for interferences. Ionic absorbance orfluorescence are rarely used analytically however a technique for generating intense ion emission has been described (9 11C2 146).When a commercial Ca HCL was driven with a 1.5 ps 28 A current pulse the intensity of the Ca I1 393.3 nm line was enhanced by a factor of more than 1 x lo6. Many of the papers to be reviewed here rely upon cold vapour procedures or gas chromatography for vaporizing the analyte prior to atomization in a flame or plasma. By these means the analyte element is separated from its matrix and the background signal that the latter may generate is thereby eliminated. Acceptable S / N ratios can thus be achieved with less intense excitation sources than for example a laser. This approach was followed by Rigin (911247 9 11848). A comprehensive scheme for the genera- tion of volatile compounds of analyte elements was devel- oped.The method was based on the autoclave decomposi- tion of a bulk sample using xenon tetrafluoride with subsequent conversion of the analyte elements into volatile compounds that were then isolated as a gaseous analytical concentrate. Atomization of the gas phase was by an air- cooled ICP. It was found that interferences were eliminated and the system could be calibrated with pure compounds of the element without the need for matrix matching. As the analyte was determined free of its matrix this method was proposed as a possible approach to absolute analysis by AFS. Non-dispersive systems with EDL excitation HG and atomization in an Ar-H,-entrained air flame are a feature of the AFS determination of As Bi Hg Sb and Se.Such systems have been applied after appropriate sample pre- treatment to the analysis of geological material (90/3469) food products (911446) and copper (91158). A dual channel system has been described by Guo et al. (901 1832). Spectral interference in the determination of Bi was overcome by using a bromide EDL in place of the usual iodide lamp. The detection limits for As Bi Hg and Sb in geological samples were 0.1 0.001 0.002 and 3.0 ppm respectively. The performance of a flame-based system was improved by passing the liberated hydrides through an electrically heated quartz tube furnace operating at 900 "C prior to entering the flame. These systems have been used in the analysis of geochemical samples (9 1 /4 1 1 9 1/432) and coal fly ash fish liver and muscle and sea-water (9014 152).Electrodeless discharge lamps may be constructed to operate at either microwave- or radio-frequencies. The latter mode gave improved performance in the determination of Sn in tungsten steel and molybdenum steel by HG non-dispersive AFS (9013524). The detection limit was 3.4 x g of Sn. Instrumentation specifically constructed for the determi- nation of Hg in air and water by cold vapour AFS has been described by Stockwell et al. (90lC2985). The instrument has a wide linear dynamic range and high sensitivity. In an FI system (90140 19) organo-mercurials were photo-decom- posed by passing the aqueous sample through a quartz tube and irradiating it with a 400 W UV lamp. Inorganic Hg was reduced to elemental Hg by tin@) chloride and the Hg released was extracted into Ar through a porous PTFE membrane.Fluorescence at 253.7 nm was excited by a low pressure Hg lamp and measured with a solar blind PMT. The detection limit was 0.18 ppb of Hg from 600 pl of sample. Organo-mercurials have also been decomposed using bromine monochloride following the elimination of interfer- ing organic substances with acidic potassium permanganate (911C2145). Detection limits by cold vapour AFS of eight different organic and inorganic forms of Hg were 2.3 x pg ml-l and recoveries were between 94 and 105%. A pulsed Hg HCL was used in conjunction with cold vapour generation and atomization in an MIP (911C2147). A detection limit of 3 ppb of Hg was achieved under optimized operating conditions.Two conference papers (9 1 1C20 1 6 9 1 /C2092) have recommended that AF should be considered as an alternative to AA for the measurement of Hg using standard and EPA methods of sample prepara- tion and Hg liberation. As the feasibility of determining Hg by AF has been long established the fact that it has not evolved into a standard method poses an intriguing question. The ICP as an atomizer in AFS has the advantage over the flame of being able to atomize refractory materials. Nebulization eficiency is an important factor in determin- ing the sensitivity of both techniques when analysing solutions. The use of an ultrasonic nebulizer in conjunction with an ICP resulted in superior detection limits in the sub- ppb range (911C486). The addition of alcohol to aqueous solutions is another well-known technique for producing a greater proportion of smaller droplets when the solution is nebulized. This approach has been applied to the determi- nation of Eu in yttrium oxide (9 11C2122).The tendency to form refractory oxides in the tail plume of the ICP was also reduced and a detection limit of 24 ng ml-l of Eu was obtained. A commercial ICP-AFS system (Baird AFS 2000) has been evaluated for the multi-element analysis of impurities in uranium (901C1890) and the determination of precious metals in geological samples (9 11725). Detection limits for Au Pd and Pt were 1 0.5 and 12 ng ml-l respectively. Calibration graphs were linear over 5 orders of magnitude. Analyses of CRMs were satisfactory and long- term reproducibility was f 5Oh.A pulsed HCL ( 5 kHz 300 mA) AFS system was developed by Klunder and Boss (901C1607) for use with a uniform-size droplet generator in order to study ligand interferences on Co signals in fuel-rich flames. The vapour cloud produced by individual droplets was observed. Changes in the rates of the desolvation and vaporization processes for different complexes were determined with a view to relating them to changes in free atom formation. 2.2. Laserexcited Atomic Fluorescence Spectrometry General reviews of LAFS have been published by Fujiwara (9 11140) and Omenetto (9112678). Winefordner et al. have carried out a theoretical study of single atom detection by LAFS or LEI (901C1583 9012207). They concluded that from practical considerations none of the conventional analytical approaches can achieve single-atom detection with the possible exception of an ionization method coupled with atomization in vacuum.Arising from the absence of reference materials with certified metal contents in the 0.1 pg g-l or 1 fg ml-l region the practicality of absolute analysis by LAFS has been examined by Omenetto (91x1743). The feasibility of using photodissociation lasers in LAFS has been investigated by Simeonsson et al. (901C1587 9 1/2445). Such lasers produce coherent line radiation214R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1991 VOL. 6 which is inherently tuned to atomic transitions and thus overcomes the problems of linewidth and de-tuning that arise with tunable dye lasers. Experiences in the measure- ment of T1 and In were reported and compared with other atomic spectrometric methods. In an analytical system as complex as LAFS it is inevitable that processes other than the one generating the analytical signal should occur and thereby modify the response.The presence of unvaporized droplets in the analytical volume irradiated by the laser could modify the analytical response by absorbing laser energy their explo- sive evaporation (91/2223) or focusing the laser beam to cause dielectric breakdown and the generation of a plasma (9 1 /2 2 2 5). 2.2.1. Electrothermal atomization In a conference paper (90/C2099) Michel et al. reported that for elements of medium to high volatility detection limits with ET-LAFS were in the 0.1-100 fg range. These values were 1-4 orders of magnitude more sensitive than those for ETAAS in the same furnace.Petrucci et al. (90/C1621) using flame and furnace atomization have also reported detection limits in the femtogram range. In addition they undertook an initial evaluation of ionization detection of the LAFS signal generated in several atomiza- tion sources; the outcome should be of considerable interest. The technique which has been termed resonance detection is predicted to have a spectral selectivity of the order of 10 pm. Furnace atomization systems usually operate on discrete samples however Womach et al. (91/293) have developed a system for continuous sample introduction into a ceramic atomization furnace. The system was optimized for the determination of Pb and achieved a detection limit of 2 ng ml-* with a linear calibration graph over three orders of magnitude.A much lower detection limit 0.05 pg ml-l was obtained with discrete sample atomization and a frequency-doubled dye laser pumped by a high repetition rate Cu vapour laser (90/1954). By means of a capacitor discharge heated anisotropic graphite furnace a heating rate of 60 000 "C s-l was achieved in the determination of T1 in NIST biological samples (91/C1991). An integration time of 80 ms gave a detection limit of 5 fg of T1. The performance of three lasers nitrogen copper vapour and Nd:YAG as pumps for tunable dye lasers used in the determination of Fe Ga In Ir Pb and T1 by ETAFS has been studied by Vera et al. (90/2236). Detection limits for Ir ranged from 0.5 fg with the copper vapour laser to 10 pg with the nitrogen laser for Ga they were 2 pg with the copper vapour laser and 25 fg with the Nd:YAG laser and for Pb the detection limit was the same 3 fg for both nitrogen and Nd:YAG lasers.The same group of workers (90/1986) carried out a study of the ET-LAFS determina- tion of Ga In and yb by double resonance excitation and achieved detection limits of 1 2 and 220 fg respectively. However single-resonance excitation direct line fluores- cence remained the preferred scheme for less volatile elements with strong direct line fluorescence such as Fe Pb and T1. When a laser was used in place of a discharge lamp for excitation of In atomized from a tungsten spiral atomizer (90/1936) the detection limit was lowered by two orders of magnitude to 1 x g.Even under optimal conditions of synchronized laser and atomizer pulsing there was considerable attenuation of the fluorescence signal. The background and the S/N ratio of an AFS system are closely linked hence when narrow bandpass filters (1 nm) were used to exclude background radiation in a non- dispersive instrument the noise was reduced and detection limits improved to 1 fg (90/4155). The S / N ratio was also improved when the signal level was increased by raising the pulse rate of the exciting radiation from 100 to 500 Hz (90/C1586 90/C2996); a reduction in the detection limit of T1 from 3 to 1 fg was achieved. Methods have been developed for the correction of background e.g. Zeeman effect (90/C2996 9 1/C1993) or the simultaneous measure- ment of scattered and black-body radiation at wavelengths adjacent to that of the fluorescent radiation (90/4156 9 ~ 2 6 2 2 ) .In an examination of the analysis of real samples by ET-LAFS Michel et al. (90/C2099) found that the only significant background signals they detected were black- body emission from the furnace and stray light caused by the scatter of laser radiation off surfaces of the instrument. Scatter by matrix species could be removed by proper charring of the sample prior to atomization. While calibration graphs in ET-LAFS are linear over 4-6 orders of magnitude non-linearity may appear at high concentrations of analyte. The on-set of non-linearity was believed by Dougherty et al. (91/1534) to be caused primarily by post-filter effects. For non-resonance LAFS long lifetimes for the terminating energy levels in the fluorescence processes were postulated to increase the probability of post-filter effects.Conditions that increased the residence time of atoms in the furnace (gas stop and platform atomization) produced the same effect. Successful practical applications of ET-LAFS have in- cluded the determination of Hg in air at the pg m-3 level (9 1/ 163) and Co Fe Ni and Pb in high-purity Si02 at the 1 x 10-4-1 x level (91/339). Michel et al. (91/C734) analysed a variety of SRMs in order to ascertain the magnitude of background signals and the accuracy and precision of the determination of Mn Pb and T1. Results were comparable with those obtained using ETAAS. The same paper reported great sensitivity in the determination of F in water and urine by laser-excited molecular fluores- cence of the magnesium fluoride molecule vaporized in the graphite furnace.Lead and T1 were successfully determined in nickel-based alloys by direct solid sampling graphite furnace LAFS (90/C1676 91/1426 911C1938). Ultimate detection limits in the range 1 x 10-4-1 x lo-' pg g-l using a 1 mg sample were predicted. The RSDs were 7-14% with an accuracy in the measurement of nickel-based alloy SRMs of = 10%. 2.2.2. Low pressure atomization systems For the determination of isotopes by LAFS it is necessary that the exciting laser has a narrow linewidth and the sample is atomized at low pressure. Niemax (90/3403) using a sample mounted in a vacuum chamber compared four techniques for this purpose excitation in an atomic beam orthogonal to the laser beam saturation spectros- copy off-resonance multi-photon spectroscopy and reso- nant multi-photon spectroscopy.The first technique gave especially efficient excitation. In other studies employing closed systems the same group used laser (Nd:YAG) ablation as a means for atomizing metallic samples for LAFS analysis (9 1 /95 9 1/C6 18). With reliable lasers and time-gated detection systems fluorescence intensities could be measured with small standard deviations but were strongly dependent on the pressure of the noble gas buffer. The dynamic range covered four orders of magnitude and detection limits were of the order of 0.3 pg for B Cr Mg Mn and Si in low-alloy steels copper and aluminium. Using this system these workers demonstrated (9 1/22 12) that internal standardization can be applied successfully to the determination of elements by LAFS of microplasmas produced by laser ablation of solid samples.The ratios of the signals from ground-state atoms of Cr Mg Mn and Si when probed by laser-induced fluorescence (LIF) were found to a first approximation to be independent of theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY7 AUGUST 199 1 VOL. 6 215R matrix the amount of ablated material and plasma temper- ature. These ratios therefore could be used as a basis for internal standardization. Russian workers have been active in investigating the merits of atomization in low-pressure environments. Bo1'- shov et al. claimed (91/271) that for some sample types vacuum ETA aflords almost 100% eficiency without matrix eflects.The technique incorporating a graphite vessel with quartz windows for atomization at 10 mTorr and 3000 "C and a Nd:YAG laser for excitation was used to determine Co in high-purity tin vegetables and quartz and also Ru in geological samples. A commercially available cathodic sputtering cell designed for AAS was adapted for LAF measurements (90lC1624). An improvement in detection limits of 1-2 orders of magnitude over AAS was achieved. By normalizing the LAF signal from Fe to the AES signal from Cu reasonable linearity across a series of SRM brasses of widely varying composition was achieved. Sputtering was also used in a study of the direct analysis of high-purity materials by LAFS (91/1598). In the determination of Pb in pure copper and Si in pure germanium detection limits were 2-3 ng g-l.The study also investigated the possibili- ties of a planar magnetron as an atomizer and LAFS detection as a means for the determination of both surface and volume contamination of solids. There was reasonable agreement between theoretical and experimental calibra- tion graphs. The sensitivity of the determination of Co by hot hollow cathode AES (91/338) was improved by 3-4 orders of magnitude when the discharge process was used to atomize dry solution residues and an LAF signal was generated by a Nd pumped dye laser. Smith et al. (90/1189) employed a commercially available HCL as the atom reservoir and excitation by a copper laser pumped dye laser for the determination of Pb by LAFS. By virtue of the very low background emission and excellent atomization charac- teristics of the glow discharge a detection limit of 1.8 ag was achieved.A pulsed glow discharge was used as an atomizer for LAFS by Click et al. (90/3343 911C1999) and Khvostikov and Grazhulene (90/C3 142). By firing the laser 0.1 ms after the end of the discharge pulse the fluorescence signal was generated when the atom population was still large but the background emission had become negligible. Pulse repetition rates of the order of 10 Hz were employed. Detection limits for Ir and Pb were 20 and 0.5 pg respectively. Gas analysis by LAFS is relatively undeveloped owing in part to the fact that the resonance transitions of most gases lie in the VUV region for which intense lasers are not available.Bol'shakov et al. (9 1/873) have circumvented this problem by populating metastable states of the analyte species in a plasma at a gas pressure of 4 kPa and then exciting fluorescence from that state with a laser. The detection limit for Ne in helium was 0.3 ppb Ne in argon < 100 ppm and N2 in helium and argon 1 ppb after oxidation to NO2. 2.3. Inductively Coupled Plasma Excited Atomic Fluores- cence The ICP as an emission source in AFS is a means of achieving an intense multi-element source for a wide range of elements. Greenfield and co-workers have continued their pioneering work (see Anal. Proc. 1984 21 61) exploring the possibilities of dual plasma (atomizer-source) systems. They have confirmed (91/C823) using the dual plasma system that a reduction of spectral interferences can be achieved by replacing AES detection with AFS.A small effect due to increased background noise was noted for a few elements when high concentrations of interferents were nebulized into the atomizer plasma. Improvement in the optical and sample transfer efficiency of the original instrument led to a 3-fold decrease in the detection limit (90/C3221). The system was applied to the determination of refractory elements (Mo V W and Zr) following preconcentration on Amberlite IRA-93 (90/1467 90/1748 9013503). The purpose of the work was to develop a preconcentration procedure for the simultaneous determina- tion of a number of refractory elements. The method was suitable only for Mo and W and achieved detection limits of 15 and 65 ng ml-1 with a precision of 5.2 and 4.3% respectively.From a comparison of a boosted discharge HCL with an ICP employed as excitation sources for AF in the tail flame of an ICP the same workers (90/3969) concluded that when the lamps were over-run there was no difference in the radiance of the sources. The lamps gave rise to lower blank standard deviation values but poorer sensitivity owing to inefficient irradiation of the analytical volume in the plasma. An air ICP has been investigated by Seltzer and Green (90K1585) as an atom source for ICP- LAFS. It was suggested that the higher ground state atom population and superior thermal properties of the air ICP could lead to lower detection limits. The background continuum was lower than that of the Ar ICP but there was considerably more band structure in the air ICP.2.4. Studies of Flames and Plasmas Using Laser-induced Fluorescence The fluorescent lifetime of the Na 589.18 nm line in atmospheric pressure CH4-0,-diluent (N2 or Ar) flames was measured by Alfono (90/2272) using a high repetition rate (3.8 MHz) picosecond dye laser with gated photon counting detection. The lifetime was found to change from 0.61 to 0.80 ns over a 3 mm region above the flame (CH4-02-N2) reaction zone. At the same location in the flame the lifetimes were 1.10 ns with Ar as the diluent and 0.56 ns when N2 was used. A hollow cathode atom/ion beam source was used to make accurate measurements of absolute transition probabilities by means of LAFS and FTS (90/C159 1). Time resolved LAF measurements of radiative lifetimes in the range 2 ns-2 ps were accurate to within 5%.The intrinsic advantage of FTS over scanning spectropho- tometry of simultaneous measurement on all spectral elements facilitated the measurement of branching ratios in complex atomic spectra throughout the spectral range from the UV to IR. Absolute transition probabilities were derived from lifetime and branching ratio data. Two-photon LIF has been used to detect oxygen atoms in the gas phase above a heated catalytically stabilized combuster (9 1/7 13); excitation was at 226 nm and detection at 777 nm. Strong interfering background signals were reduced by spatial and spectral filtering. The detection limit for oxygen was estimated to be 5 x 1 OI3 atoms cm-'. Atomic oxygen has also been detected using the four-wave mixing technique of third harmonic generation (9 112224).Third harmonic generation provides state selective low aperture detection with sensitivity comparable to other optical methods. Ground-state oxygen was generated using a microwave discharge and detected by focusing a 391 nm dye laser beam into the post-discharge region with measure- ment at 130 nm. The detection limit was 5 x l O I 3 atoms (approximately 2 ppm at atmospheric pressure). Measurement of OH in flat stoichiometric CH,-air flames burning at 1 2 5 7 and 9.6 bar have been carried out by LIF and resonance coherent anti-Stokes Raman scattering (CARS) (9014 179). Temperature and OH concentrations were derived from the data. Neither method was superior to the other; e.g. resonance CARS might be chosen because of its coherent nature or LIF because of its two dimensional imaging capabilities.Dunn-Rankin et al. (9 1 /2225) have examined sources of possible error in gas temperature measurements made by CARS in the presence of liquid216R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 droplets. The droplets can cause dielectric breakdown of a plasma by focusing the CARS laser beam. These effects reduce the peak signal intensity and raise the background intensity. Correction procedures were devised whereby the systematic error was reduced to within one standard deviation of the drop-free measurements of temperature. Rensburger et al. (90/1973) examined in depth the systematic errors in the determination of temperatures in low-pressure flames by LIF arising from spectral bias time delay temporal sampling rate optical depth and rotational level dependence of the fluorescence quantum yield for each radical (OH NH CH and CN).The severity of these problems is different from each radical and must be evaluated experimentally. The discrepancies between esti- mates of temperature made with and without correction for the systematic errors could be as much as several hundred degrees. Laser atomic fluorescence in conjunction with atomic emission was used to probe the spatial dependence of atom and ion density in an Ar ICP as a function of aerosol transport rate (90/4 1 54). Fluorescence responsivities (sig- naYmass of analyte) either remained constant or in- creased as sample aerosol transport rate increased while ion emission responsivity decreased.The changes in AE and AF responsivities with aerosol loading were spatially dependent. Laser-induced fluorescence of Ba and Sr was used to study the three-dimensional spatial distribution of atoms in a three electrode d.c. plasma (90/C1817 90/3492). It was found that the distribution of barium ion ground states was determined primarily by flow dynamics round the Ar plasma jet. The portion of the plasma where emission was generated was small com- pared with the over-all spatial distribution of the analyte indicating that only a small fraction of the total analyte present contributes to the emission signal. The presence of Na and other easily ionized elements did not alter the spatial distribution of Ba but did cause ionization suppression.3. LASER-ENHANCED IONIZATION Laser-enhanced ionization (LEI) uses laser radiation to excite atoms to levels from which ionization by thermal processes is increased significantly i. e. the current flowing between electrodes placed in the atom vapour increases when the vapour is laser irradiated. Direct laser ionization (laser photoionization) i.e. situations where all the ioniza- tion energy is supplied optically will also be considered. This latter technique is most commonly used in association with low-pressure systems such as resonance ionization mass spectrometry (RIMS) (see Atomic Spectrometry Update-Inorganic Mass Spectrometry and X-ray Fluores- cence J. Anal. At. Spectrum. 1990 5 250R). There is a low but sustained level of interest in LEI and associated techniques.The processes occurring during LEI are com- plex and will doubtless be a fruitful field for study for some time to come. The widespread exploitation of LEI as a universal analytical method however does not appear likely in the foreseeable future despite its potential for high sensitivity. This review is based on approximately 20 published papers and ten conference presentations. Previous reviews can be found in Atomic Spectrometry Update-Instrumen- tation J. Anal. At. Spectrum. 1990 5 152R and Atomic Spectrometry Update-Atomization and Excitation J. Anal. At. Spectrum. 1990 5 180R. A thorough general review of atomic ionization spectrometry has been published by Kuzyakov and Zorov (91/1134). The principles fea- tures advantages and applications of the technique are presented.Laser-enhanced ionization in flames has been reviewed by Axner and Rubinsztein-Dunlop (90/ 1849 90/3953 9 1/99) and high-resolution laser-induced ioniza- tion spectroscopy by Bushaw (90/332 1). Conference review papers have been presented by Winefordner (9 1/C568) Rubinsztein-Dunlop (9 1/C62 1) and Green and Seltzer (91/C2021). 3.1. Systems Operating at Atmospheric Pressure Practical applications of LEI in flames have been described by Turk and Kingston (90/C1765). The NIST (Washington DC USA) programme for the certification of SRMs has employed LEI based methods for several years. It was found (9 ~ 1 4 2 5 ) that interference from alkali and alkaline earth elements present in biological samples necessitated the removal of those elements by chelation chromatography.In two papers (9 ~ 2 4 2 6 9 1/2427) Axner and co-workers reported investigations of the interference effect of sodium. To reduce the effect a two-laser system was used one to ionize sodium while the other probed a volume 4 mm higher in the flame for the analyte where the population of sodium atoms was depleted. This approach reduced the sodium signal by up to 5-fold. An alternative approach to improve discrimination against sodium was careful selec- tion of laser power. The optimum power is determined by factors that are specific to each experimental set-up. This latter approach was derived from a fundamental theoreti- cal and experimental study of one-step LEI transitions for atoms in flames (9112213).A simplified expression for the one-step LEI signal strength as a function of excitation wavelength was derived. From this expression general element-independent predictions of the most sensitive unsaturated one-step LEI transitions were made for 13 elements. Other workers (9 1 /C2 144) have examined the interference of radicals e.g. OH CH and C2 in an air-C2H2 flame on the LEI detection of trace elements in the UV spectral region. They found for example that a strong OH band causes serious deterioration in the detec- tion of Pb at 283.3 nm and advised that fundamental studies of laser ionization of radicals are an important aspect of LEI determination of elements with high ioniza- tion potentials. For the determination of Cs in cadmium mercury telluride by single and double-step excitation schemes (90/3048) nebulization of the sample solution into an air-C4H,,,-C3Hs flame was compared with evaporation from a heated pyrolytic graphite rod into the flame.For both techniques the detection limits and linear dynamic ranges were 0.1 pg and 1 pg ml-’-10 pg ml-l respectively; accuracy and precision were good for the determination of 1 x 10-9-1 x lO-’Oh Cs. The lower limit of the linear range of a flame-based LEI method for the determination of Mg was reduced by two orders of magnitude to 1 ng 1-l when the conventional two-electrode system was replaced by a three-electrode arrangement (90/ 1 523). This configuration facilitated differential signal measurement whereby the level of electrical interference from the pulse voltage generator of the nitrogen pumping laser was reduced.Using a pre-mixed air-C,H2 flame Turk and Travis (91/2204) investigated random signal fluctuations of simultaneously detected LEI and LIF. They found a greater degree of noise correlation between the two techniques than between either technique and laser power fluctuations. They proposed a scheme for background correction in LEI by utilizing the LIF signal of the interferents and the noise correlation between LEI and LIF. Axner and Sjostrom have examined spectral broadeningJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 217R of the LEI signal from Sr (460.733 nm) in a flame (91/C623 9 1/723) when optical saturation conditions prevail (laser power 11 kW cm-2). They observed a narrow peak on top of a broader peak.The latter was attributed to the theoretically predicted effect whereby the width of the measured peak increases as the square root of the light intensity. It was proposed that the narrow peak arose from ionization of atoms outside the primary (laser beam defined) interaction region by radiation scattered from the laser beam. This effect is particularly significant when beams of small diameter and high intensities are used and can lead to errors when number densities ionization yields signal enhancement and saturation effects are studied by LEI. It has been proposed (9 1/C 1994) that the spatial distribution of ionization may be studied by observing the LEI signal as a function of the voltage between the detection electrodes. As the voltage applied to the cathode increases the effective electric field extends further from the cathode into the flame; the burner being at ground potential.Progressively more of the flame and its ion content is sampled as the voltage increases. Graphite probe furnace LEI has been proposed (90/C1620 90/C303 1) as a means of minimizing gas-phase reactions between the analyte and its matrix. Biological reference materials were used to evaluate the potential of the system for the analysis of real samples. Michel et al. (901C2099) followed a similar approach but additionally used the probe as an electrode and thus avoided sample re- condensation on the electrode. Plasmas provide alternative but difficult to exploit atomizers for LEI. In the past the tail flame of an ICP has been used for this purpose (e.g.91/2460) rather than the AE analytical zone. Some recent work (90/C1623) however has been directed at utilizing the latter. By momentarily switching off (1.5 ms) the r.f. power (1 kW) in synchronism with firing of the laser an ionization signal could be detected with water-cooled copper electrodes. The electrical conductivity properties of an MIP have been examined with a view to its use as a source for LEI (90/1485). Argon and Ar-active N2 plasmas were investigated. In the latter plasma suppression of the signal by thermally ionized Group 1A elements was encountered. 3.2. Systems Operating at Reduced Pressure The potential and limitations of isotope ratio measure- ments using ETA into a low pressure noble gas atmosphere and resonant Doppler-free two-photon LEI spectroscopy have been examined (90141 12).It was concluded that the technique could be an alternative analytical technique to mass spectrometry provided physical effects such as laser instablity and gas-temperature fluctuations which can give rise to systematic errors are eliminated. The precision of the absolute measurement of isotopes was 10-1 5% but for isotope ratios it was 2-5%. A special vacuum chamber with a rapid sample turn around time and incorporating a graphite tube furnace was constructed by Stevenson et al. (901C1622) in order to facilitate studies designed to characterize and optimize resonance ionization spectroscopy (RIS) as an analytical method. The factors affecting the atomization eflciency of a vacuum furnace system for atomic ionization spectrometry have been studied by Russian workers (9011223).They found that the molecular collision number in a tube atomizer was a quadratic function of the ratio of the tube length to its internal diameter and concluded that vacuum atomization requires a stricter approach to establishing the accuracy of the analysis than conventional methods involv- ing atomization in gaseous media. Laser-enhanced ionization was used to generate ions in a laser microprobe system in which the vaporization wave- length was tuned in resonance with the analyte (Cu or Na) absorption band (90/C 1638). A 70-fold enhancement of the ion intensity from the resonant transition was observed. The acoustic wave generated by the laser plume in the He atmosphere was used to monitor the total amount of material vaporized during each laser pulse.Laser ablation of sodium tungstate in vacuo was used in the study of laser power effects on LEI measurements of the Na D-line (9 1 /722 9 K2033). Laser-enhanced ionization spectral broadening occurred as the power of either the ablating or exciting laser was increased. Ion intensity showed a linear response to changes in laser power. At 1 atm of He severe collisional broadening was observed. High resolution and low sensitivity occurred with low excitation intensity; the reverse was true at high intensity. 4. ATOMIC MAGNETO-OPTICAL ROTATION SPECTROMETRY Atomic magneto-optical rotation spectrometry (AMORS) also known as coherent forward scattering (CFS) uses the rotation of the plane of linearly polarized resonance radiation by a magnetized atomic vapour to generate an analytical signal.The birefringent and dichroic properties of the vapour are a consequence of the Zeeman effect in the atom. The analytical signal may be recorded either as a change in the intensity of radiation transmitted by a system of crossed polarizers or as the angular rotation of the plane of polarization of the incident radiation. The latter may be observed directly (91lC462; see also Monnig C. A. Madison R. T. and Hieftje G. M. Appl. Spectrosc. 1990 44 216) or computed from measurements of transmitted intensity (91/796). A review of developments in AMORS was presented in Atomic Spectrometry Update-Atomiza- tion and Excitation J. Anal. At. Spectrom. 1990 5 210R. Research activity in AMORS is currently at a low level and is undertaken in probably no more than ten laboratories worldwide.As far as can be judged at the present time AMORS will remain a laboratory novelty until some as yet unrecognized outstanding advantage of the technique over established techniques is revealed. Some advantages of AMORS over other spectroscopic techniques particularly with respect to background correction have been reviewed by Stephens in a conference presentation (9 VC689). To date the lowest detection limits obtained by AMORS have been comparable to those obtained by AAS using the same instrumental system (e.g. 9 1/462). A detection limit for Ag of 1 ppb was reported for a Voigt (transverse magnetic field) configuration system employing ETA (90/C2043).Errors resulting from background absorption and fluctuations in the intensity of the HCL were reduced by the use of a rotating polarizer with a stationary analyser. The ratio of the energies transmitted when the planes of the two polarizers were parallel and orthogonal was calculated to give a corrected signal (see also Dawson J. B. Duffield R. J. Kersey A. D. Hajizadeh-Saffar M. and Fisher G. W. J. Anal. At. Spectrom. 1987 2 233). Zhao et al. (91/2447) using a pulsed HCL examined the influence of light source S/N ratio continuum background and struc- tured interference on the determination of Au in rocks and minerals. A detection limit of 14 pg and satisfactory precision and accuracy were achieved. Simultaneous multi- element analysis by AMORS using a crossed polarizer configuration with a continuum light source (xenon arc) has been reported by Hermann et al.(9 1 / 199). In this arrange- ment the AMORS process becomes in effect a high- resolution spectrometer which transmits radiation from the source only at resonance wavelengths. Its effectiveness is limited by the extinction coefficient of the crossed polariz-218R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1 99 1 VOL. 6 ers emission from the graphite furnace and the quadratic nature of the response curve. The instrument which was a development of one previously reported (9.011 172) was applied to the determination of Ca Cd and Mg. used light sources only weakly emit such lines. Extended working ranges were achieved by the use of very weak absorption lines for the measurement of high concen- trations.This approach to the determination of major constituents is not feasible in AAS because the commonly LOCATION OF REFERENCES The full list of references cited in this Update have been published as follows 90/1159-90/C1460 J. Anal. At. Spectrom. 1990 5(4) 167R-178R. 90/1461-90/2277 J. Anal. At. Spectrom. 1990 5(6) 2 15R-242R. 90/2278-901359 1 J. Anal. At. Spectrom. 1990 5( 7) 278R-32 1 R. 9013591-90/4179 J. Anal. At. Spectrom. 1990 5(8) 361R-378R. 91/1-911825 J. Anal. At. Spectrom. 1991 6(1) 41R-68R. 91/826-91/C1687 J. Anal. At. Spectrom. 1991 6(3) 109R-136R. 91/C1688-91/2702 J. Anal. At. Spectrom. 1991 6(4) 153R-186R. Abbreviated forms of the literature references quoted (excluding those to Conference Proceedings) are given on the following pages for the convenience of the readers.The full references names and addresses of the authors and details of the Conference presentations can be found in the appropriate issues of JAAS cited above. Abbreviated List of References Cited in Update 9011164. Analyst 1989 114 1159. 9011171. J. Anal. At. Spectrom. 1989 4 525. 9011172. J. Anal. At. Spectrom. 1989 4 529 9011174. J. Anal. At. Spectrom. 1989,4 539. 9011175. J. Anal. At. Spectrom. 1989 4 543. 9011176. J. Anal. At. Spectrom. 1989 4 547. 9011180. Anal. Chim. Acta 1989 220 135. 9011186. Appl. Spectrosc. 1989 43 376. 9011189. Appl. Spectrosc. 1989 43 873. 9011209. Fresenius Z. Anal. Chem. 1989,334,27.9011223. Zh. Anal. Khim. 1988 43 2132. 9011227. Microchem. J. 1989 39 2 13.9011263. Fenxi Shiyanshi 1988 7( lo) 36. 9011286. Jilin Daxue Ziran Kexue Xuebao 1989 (l) 78. 9011342. Analyst 1989 114 1331. 9011467. Anal. Proc. 1989 26 382 9011470. Anal. Proc. 1989 26 375. 9011485. Spec- trosc. Lett. 1989 22 461.9011504. Bunseki Kagaku 1989 38 T119.9011520. Zh. Anal. Khim. 1989,44,257.9011523. Zh. Anal. Khim. 1989 44 374. 9011527. Zavod. Lab. 1989,55(4) 44.9011529. Zh. Prikl. Spektrosk. 1989,50,20. 9011537. Labor Praxis 1989 13 406. 9011549. J. Trace Elem. Electrolytes Health Dis. 1989 3(2) 97. 9011832. J. Geochem. Explor. 1989 33 237. 9011849. Trace Elem. Anal. Chem. Med. Biol. Proc. Int. Workshop Sth 1988,43. 9011857. Nucl. Instrum. Methods Phys. Res. Sect. B 1988 (Publ. 1989) B40-B41( l) 635. 9011910. Landwirtsch. Forsch. 1988 41 217. 9011923.Arch. Toxicol. Suppl. 1989 13 (Biol. Monit. Exposure Response Subcell. Level Toxic Subst.) 124.9011929. CLB Chem. Labor Betr. 1989 40 299. 9011932. ASTM Spec. Tech. Publ. 1988 973 8. 9011936. Vysokochist. Veshchestva 1988 (6) 153.9011949. Proc. ICMR Semin. 1988 8(Proc.-Asia-Pac. Symp. Envi- ron. Occup. Toxicol. 1987) 259. 9011954. Spectrochim. Acta Part B 1989 44 939. 9011957. Spectrochim. Acta Part B 1989 44 965. 9011968. Anal. Chim. Acta 1989 221 65.9011971. Anal. Chim. Acta 1989,220,83.90/1973. Appl. Opt. 1989 28 3556. 9011974. Appl. Opt. 1989 28 3968. 9011976. Appl. Opt 1989,28 4 167.9011984. J. Anal. At. Spectrom. 1989 4 605. 9011985. J. Anal. At. Spectrom. 1989 4 613. 9011986. J. Anal. At. Spectrom. 1989 4 619. 9011989. J. Anal. At. Spectrom. 1989 4 635.9011990. J. Anal. At. Spectrom. 1989 4 641. 9011991. J. Anal. At. Spectrom. 1989 4 647. 9011994. J. Anal. At. Spectrom. 1989 4 661. 9011997. J. Anal. At. Spectrom. 1989 4 673. 9012056. J. Anal. At. Spectrom. 1989 4 685. 9012057. J. Anal. At. Spectrom. 1989 4 691. 9012060. J. Anal. At. Spectrom. 1989 4 705. 9012076. J. Anal. At. Spectrom. 1989 4 797. 9012087. Spectrochim. Acta Part B 1989,44 1209. 9012089. Spectrochim. Acta Part B 1989 44 1237. 9012090. Spectrochim. Acta Part B 1989 44 1245. 9012091. Spectrochim. Acta Part B 1989 44 1257. 9012194. Analyst 1990 115 275. 9012196. Analyst 1990 115 283. 9012199. Spectrochim. Acta Part B 1989 44 1221. 9012200. Spectrochim. Acta Part B 1989 44 1213. 9012207. Spectrochim. Acta Part B 1989 44 1397. 9012221. Zh. Anal. Khim.1989 44 1382. 9012228. Ta- lanta 1989 36 657. 9012229. Talanta 1989 36 805. 9012230. Talanta 1989 36 955. 9012231. Talanta 1989 36 999. 9012233. Talanta 1989 36 1075. 9012235. Talanta 1989 36 1265.9012236. Talanta 1989,36 129 1. 9012255. Spectrochim. Acta Part B 1989 44 1125. 9012258. Spectrochim. Acta Part B 1989 44 1191. 9012267. Appl. Opt. 1989 28 4375. 9012272. Appl. Opt. 1989 28 5010. 9012914. J. Chromatogr. 1989 482 87. 9012915. Anal. Chim. Acta 1989 225 69. 9012916. Anal. Chim. Acta 1989 219 127. 9013047. Vysokochist. Vesh- chestva 1989 (4) 177. 9013048. Vysokochist. Veshchestva 1989 (4) 186. 9013049. Vysokochist. Veshchestva 1989 (4) 191. 9013246. Microchem. J. 1989 40(1) 4. 9013251. Microchem. J. 1989 39 344. 9013261. Zavod. Lab. 1989 55(8) 36. 9013265. Zavod.Lab. 1989 55(9) 27. 9013266. Zavod. Lab. 1989,55(9) 37.9013294. At. Spectrosc. 1989 10(6) 165. 9013295. At. Spectrosc. 1989 10(6) 170. 9013297. At. Spectrosc. 1989 10(6) 192. 9013300. Appl. Spectrosc. 1989 43 7 19. 9013321. Prog. Anal. Spectrosc. 1989 12 247. 9013322. Prog. Anal. Spectrosc. 1989 12 369. 9013328. Anal. Chim. Acta 1989 225 175. 90/3330. Anal. Chim. Acta 1989,225,233.9013331. Anal. Sci. 1989 5,445.9013332. Anal. Sci. 1989,5,439.9013337. Anal. Sci. 1989 5 765. 9013339. Anal. Chem. 1989 61 2652. 9013343. Anal. Chem. 1990,62 157.9013347. Anal. Chem. 1990 62 553. 9013348. Anal. Chem. 1990 62 574. 9013353. Spectrosc. Lett. 1989,22 763.9013356. Zh. Prikl. Spektrosk. 1989 51 559. 9013361. Bunseki Kagaku 1989 38 399. 9013362. Bunseki Kagaku 1989 38 403.9013371. Bunseki Kagaku 1989 38 734. 9013378. Zh. Anal. Khim. 1989 44 794. 9013379. Zh. Anal. Khim. 1989 44 802.219R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 9013388. Zh. Anal. Khim. 1989 44 810. 9013390. Z. Chem. 1989 29 21 7. 9013393. Spectrochim. Acta Part B 1989,44 667.9013396. Spectrochim. Acta Part B 1989,44 751. 9013397. Spectrochim. Acta Part B 1989 44 759. 9013398. Spectrochim. Acta Part B 1989,44,77 1.9013399. Spectrochim. Acta Part B 1989 44 795. 9013403. Fresen- ius 2. Anal. Chem. 1989 333 702. 9013406. Fresenius Z. Anal. Chem. 1989 334 11 5. 9013412. Fresenius Z. Anal. Chem. 1989 335 15. 9013413. Fresenius 2. Anal. Chem. 1989 335 200. 9013414. Fresenius 2. Anal. Chem. 1989 335 195. 9013416. Fresenius Z. Anal. Chem. 1989 335 392. 9013418 Fresenius 2.Anal. Chem. 1989 335 566. 9013421. Fresenius Z. Anal. Chem. 1989 335 769. 9013450. Fenxi Huaxue 1989 17 441. 9013459. Fenxi Huaxue 1989 17 662. 9013463. Fenxi Shiyanshi 1989 8( I) 30.9013469. Fenxi Shiyanshi 1989,8(2) 6 1.9013470. Fenxi Shiyanshi 1989,8(4) 105.9013474. Fenxi Shiyanshi 1989,8(5) 60.9013478. J. Anal. At. Spectrom. 1990 5 15. 9013479. J. Anal. At. Spectrom. 1990 5 21. 9013482. J. Anal. At. Spectrom. 1990 5 35. 9013483. J. Anal. At. Spectrom. 1990 5 39. 9013484. J. Anal. At. Spectrom. 1990 5 45. 9013492. J. Anal. At. Spectrom. 1990 5 99. 9013495. J. Anal. At. Spectrom. 1990 5 121. 9013498. J. Anal. At. Spectrom. 1990 5 139. 9013499. J. Anal. At. Spectrom. 1990 5 145. 9013503. Analyst 1990 115 53 1. 9013505. Analyst 1990 115 539.9013508. Analyst 1990 115 553. 9013511. Analyst 1990 115 567. 9013512. Analyst 1990 115 571. 9013513. Yejin Fenxi 1988 8(1) 35. 9013521. Yejin Fenxi 1988 8(4) 10. 9013524. Yejin Fenxi 1988 8(5) 21. 9013584. Talanta 1990 37 119. 9013588. Talanta 1990 37 167. 9013591. Talanta 1990 37 519. 9013951. Anal. Sci. 1990 6 87.9013953. Spectro- chim. Acta Part B 1989 44 837. 9013954. Spectrochim. Acta Part B 1989 44 867. 9013955. Spectrochim. Acta Part B 1990 45 85. 9013958. Spectrochim. Acta Part B 1990,45 13 1.9013966. Spectrochim. Acta Part B 1990,45 281. 9013969. Spectrochim. Acta Part B 1990 45 341. 9013971. Spectrochim. Acta Part B 1990,45,367.90/3914. Spectrochim. Acta Part B 1990,45 48 1. 9013975. Spectro- chim. Acta Part B 1990 45 493. 9013981. Anal. Chim.Acta 1989 226 203.9013982. Anal. Chim. Acta 1989,227 331. 9013994. Fresenius Z Anal. Chem. 1989 335 176. 9014002. Fresenius 2. Anal. Chem. 1989 335 743. 9014010. Fresenius 2. Anal. Chem. 1989 335 971. 9014011. Fresenius Z. Anal. Chem. 1989 335 975. 9014016. Fresenius J. Anal. Chem. 1990,336,20 1~9014019. Anal. Sci. 1990,6 91.9014020. Bunseki Kagaku 1989,38 407.9014022. Bunseki Kagaku 1990,39 T5.9014033. Clin. Chem. 1990 36 351. 9014112. Spectrochim. Acta Part B 1990 45 15. 9014141. J. Anal. At. Spectrom. 1990 5 163. 9014142. J. Anal. At. Spectrom. 1990 5 171. 9014143. J. Anal. At. Spectrom. 1990 5 175. 9014144. J. Anal. At. Spectrom. 1990 5 183. 9014145. J. Anal. At. Spectrom. 1990 5 189. 9014146. J. Anal. At. Spectrom. 1990 5 195. 9014148. J. Anal. At. Spectrom.1990 5 203. 9014150. J. Anal. At. Spectrom. 1990 5 209. 9014151. J. Anal. At. Spectrom. 1990 5 215. 9014152. J. Anal. At. Spectrom. 1990 5 225. 9014153. J. Anal. At. Spectrom. 1990 5 231. 9014154. J. Anal. At. Spectrom. 1990 5 239. 9014155. J. Anal. At. Spectrom. 1990 5 249. 9014156. J. Anal. At. Spectrom. 1990 5 261. 9014158. J. Anal. At. Spectrom. 1990 5 273. 9014179. Appl. Opt. 1990 29 1560. 9115. J. Anal. At. Spectrom. 1990 5 31 1.9116. J. Anal. At. Spectrom. 1990 5 321. 9117. J. Anal. At. Spectrom. 1990 5 325.91111. Analyst 1990,115,939.91112. Analyst 1990 115 943. 91131. Appl. Spectrosc. 1990 44 280. 91145. At. Spectrosc. 1990 11(1) 1. 91148. At. Spectrosc. 1990,11(1) 24. 91149. At. Spectrosc. 1990 11(2) 78. 91154. Spectros- copy (Eugene Oregon) 1990 5(3) 45.91158. Anal. Lett. 1989 22 1609. 91/59. Anal. Lett. 1989 22 1369. 91162. Anal. Lett. 1989 22 2847. 91164. Anal. Lett. 1989 22 2887. 91171. Talanta 1990 37 545.91188. Microchem. J. 1990 41 29. 91195. Mikrochim. Acta 1989 II(4-6) 201. 91199. Mikrochim. Acta 1989,3 197.911131. Yejin Fenxi 1989,9 36.911140. Bunseki 1989 3 199. 911155. Analyst 1990 115 605. 911157. Anal. Chem. 1990 62 1161. 911163. Anal. Chem. 1990 62 1452. 911165. Fenxi Huaxue 1989 11 399. 911168. Fenxi Huaxue 1989 17 673. 911169. Fenxi Huaxue 1989 17 692. 911170. Fenxi Huaxue 1989 11 727. 911174. Fenxi Huaxue 1989 17 835. 911180. Fenxi Huaxue 1989 17 940. 911181. Fenxi Huaxue 1989 17 96 1. 911192. Spectrochim. Acta Part B 1990 45 547. 911195. Spectrochim. Acta Part B 1990,45 679. 911196. Spectrochim.Acta Part B 1990 45 695. 911199. Spectrochim. Acta Part B 1990 45 763. 911216. Zh. Anal. Khim. 1990 45 506. 911230. J. Anal. Chem. 1990 336 320. 911231. Fresenius J. Anal. Chem. 1990 336 394. 911233. Fresenius J. Anal. Chem. 1990,336,480. 911234. Fresenius J. Anal. Chem. 1990 336 484. 911236. Fresenius J. Anal. Chem. 1990 336 567. 911240. Anal. Chim. Acta 1990 228 129. 911241. Anal. Chim. Acta 1990 228 139. 911245. Anal. Chim. Acta 1990 230 163. 911247. Anal. Chim. Acta 1990 231 85. 911252. Anal. Chim. Acta 1990 232 301. 911258. Sci. Total Environ. 1989 89 299. 911259. Sci. Total Environ. 1989 89 319. 911260. Sci. Total Environ. 1989 89 325. 911261. Sci. Total Environ. 1989 89 33 1. 911262. Sci. Total Environ. 1989,89 337. 911271. Zavod. Lab. 1989,55(9) 43.911276.Zavod. Lab. 1989 55( 1 l) 31. 911288. Acta Chim. Hung. 1989 126 391. 911289. Acta Chim. Hung. 1989 126,347. 911292. Spectrosc. Lett. 1989 22 779. 911293. Spectrosc. Lett. 1989 22 1333. 911310. J. Assoc. Ofl Anal. Chem. 1990,13 389. 911323. Chem. Listy 1990,84 376.911338. Zh. Prikl. Spektrosk. 1990 52 359. 911339. Zh. Prikl. Spektrosk. 1990 52 363. 911340. Zh. Prikl. Spektrosk. 1990 52 374. 911348. Yankuang Ceshi 1989 8 217. 911374. Guangpuxue Yu Guangpu Fenxi 1989 9( 11 55. 911381. Guangpuxue Yu Guangpu Fenxi 1989 9(2) 45. 911382. Guangpuxue Yu Guangpu Fenxi 1989 9(2) 48. 911384. Guangpuxue Yu Guangpu Fenxi 1989 9(2) 56. 911386. Guangpuxue Yu Guangpu Fenxi 1989 9 60. 911406. Guangpuxue Yu Guangpu Fenxi 1989 9(4) 68. 911407. Guangpuxue Yu Guangpu Fenxi 1989 9(6) 54.911411. Lihua Jianyan Huaxue Fence 1988 24 103. 911418. Lihua Jianyan Huaxue Fence 1989 25 6.911421. Lihua Jianyan Huaxue Fence 1989,25 18.911422. Lihua Jianyan Huaxue Fence 1989 25 21. 911424. Lihua Jianyan Huaxue Fence 1989 25 43. 911432. Lihua Jianyan Huaxue Fence 1989 25 97. 911440. Lihua Jianyan Huaxue Fence 1989 25 205. 911446. LihuQ Jianyan Huaxue Fence 1989 25 271. 911448. Lihua Jianyan Huaxue Fence 1989 25 291. 911457. Rep. - MINTEK 1990 M388 10 pp. 911462. Report 1988 INDU/DC/GMH/TR-88-33; Order No. AD-A1 98799 19 pp. 911469. Vysokochist. Veshchestva 1990,1 165.911474. Bull. SOC. Chim. Belg. 1989 98 513. 911712. Appl. Opt. 1990,29,47.911713. Appl. Opt. 1990,29 11 1.911716. Appl. Opt. 1990 29,295 1.911117. Nippon Kagaku Kaishi 1989 (7) 1094. 911722. Appl. Spectrosc. 1990 44 871. 911723. Appl. Spectrosc. 1990 44 864. 911725. J. Geochem. Explor. 1989,34 245.911845. J. Anal. At. Spectrom. 1990 5 385. 911846. J. Anal. At. Spectrom. 1990 5 393.911847. J. Anal. At. Spectrom. 1990 5 399. 911848. J. Anal. At. Spectrom. 1990 5 407. 911849. J. Anal. At. Spectrom. 1990 5 413. 911873. J. Anal. At. Spectrom. 1990 5 549. 911892. Can. J. Fish. Aquat. Sci. 1989 46 11 31. 911898. Collect. Czech. Chem. Commun. 1989 54 1785. 911903. Fenxi Ceshi Tongbao 1989 8(2) 29. 911910. Gidrokhim. Mater. 1989 105 106. 911914. Gidrokhim. Mater. 1989 105 100 911918. Hua Husueh 1987 45(1) 23. 911929. Jilin Daxue Ziran Kexue Xuebao 1989 (2) 127. 911938.220R JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 199 1 VOL. 6 Kidorui 1989 14 38. 911942. Magy. Kem. Foly. 1989,95 275. 911945. Rep.-MINTEK 1989 M392 20 pp. 911968. Vestn. Leningr. Univ. Ser. 7 Geol. Geogr. 1989 (l) 80. 911987. Z. Med. Laboratoriumsdiagn 1989 30 388. 911991. Acta Chim. Sin. (Engl. Ed.) 1989 (9 437. 9111012. Delta J. Sci. 1987 11 88. 9111028. Huaxue Xuebao 1989,47 1139.9111034. Izv. Khim. 1989,22,282. 9111043. Jisuanji Yu Yingyong Huaxue 1989 6 205. 9111049. J. Serb. Chem. SOC. 1988,53,699.9111056. Kem. Tidskr. 1988 100(9) 51. 9111064. Labor Praxis 1990 14 32.9111066. Mitt. Geb. Lebensmittelunters. Hyg. 1989,80 452. 9111072. Pertaniku 1989 12,79.9111074. Probl. Anal. Khim. 1989 9 237. 9111076. Quim. Anal. (Barcelona) 1989 8 191. 9111077. Quim. Anal. (Barcelona) 1989 8 159. 9111086. Sci. Prog. (Oxford) 1989,291 35 1.9111088. Sci. Total Environ. 1989 89 353. 9111098. TrAC Trends Anal. Chem. (Pers. Ed.) 1989,8 336.9111103. Vestn. Slov. Kem. Drus. 1989 36 75. 9111112. Acta Hydrochim. Hydrobiol. 1989 17(5) 501. 9111122. Bunko Kenkyu 1989 38 187. 9111127. Chem. Express 1990 5 129. 9111134. Crit. Rev. Anal. Chem. 1988 20 221. 9111146. Gaodeng Xuexiao Huaxue Xuebao 1989 10 1086. 9111 155. Huazhong Shqan Daxue Xuebao Ziran Kexue- ban 1988 22 443. 9111162. J. Chemom. 1989 3 455. 9111193. Proc. Chem. Con$ 1988,41 123.9111210. TrAC Trends Anal. Chem. (Pers. Ed.) 1989 8 333. 9111252. Chemom. Intell. Lab. Syst. 1990,7,28 1.9111254. Chroma- tographia 1990 29 254. 9111255. Collect. Czech. Chem. Commun. 1989 54 2683. 9111261. Fenxi Ceshi Tongbao 1989 8(3) 37. 9111264. Freiberg Forschungsh. B 1988 266 91. 9111273. Huanjing Wuran Yu Fangzhi 1989 11(4) 43. 9111293. Khim. Tekhnol. Vody 1989 11 531. 9111 294. Kinki Daigaku Rikogakubu Kenkyu Hokoku 1989 (25) 125. 9111303. Mar. Chem. 1990 29 33. 9111311. Pract. Spectrosc. 1989 7(Flow Injection At. Spectrosc.) 225. 9111312. Pract. Spectrosc. 1989 7(Flow Injection At. Spectrosc.) 293. 9111354. Chemom. Intell. Lab. Syst. 1989 7 163. 9111367. Izv. Vyssh. Uchebn. Zaved. Khim. Khim. Tekhnol. 1989 32 46. 9111380. Metodi Anal. Acque 1988 8 3. 9111392. Rev. Chim. (Bucharest) 1989 40 619. 9111399. Tap Chi Hoa Hoe 1989 27 7. 9111421. US Pat. Appl. US 428,529 01 Mar. 1990. 9111425. J. Anal. At. Spectrom. 1990 5 595. 9111426. J. Anal. At. Spectrom. 1990 5 603. 9111433. J. Anal. At. Spectrom. 1990 5 639. 9111435. J. Anal. At. Spectrom. 1990 5 65 1.9111438. Fenxi Huaxue 1989 17 1068. 9111441. Fenxi Huaxue 1989 17 1116. 9111448. Fenxi Huaxue 1990,18,72.91l1450. Fenxi Huaxue 1990 18 174. 9111452. Spectrochim. Acta Part B 1990,45,621. 9111453. Spectrochim. Acta Part B 1990,45,633.91/1458. Spectrochim. Acta Part B 1990,45 867.9111463 Spectro- chim. Acta Part B 1990 45 955. 9111467. Yejin Fenxi 1989 9(6) 24. 9111469. Yejin Fenxi 1989 9(6) 58. 9111490. Analusis 1990 18(3) 12. 91114%. Anal. Chim. Acta 1990,231 1 5 1.9111497. Anal. Chim. Acta 1990,233 165. 9111498 Anal. Chim. Acta 1990 234 3. 9111506. Fresenius J. Anal. Chem. 1990,337,248.9111507. Fresen- ius J. Anal. Chem. 1990 337 253. 9111508. Fresenius J. Anal. Chem. 1990 337 264. 9111509. Fresenius J. Anal. Chem. 1990 337 294. 9111510. Fresenius J. Anal. Chem. 1990 337 299. 9111514. Fresenius J. Anal. Chem. 1990 337 393. 9111515. Fresenius J. Anal. Chem. 1990 337 412. 9111516. Fresenius J. Anal. Chem. 1990 337 551. 9111517. Fresenius J. Anal. Chem. 1990,337,528.9111521. Yankuang Ceshi 1989 8(4) 300. 9111525. Lihua Jianyan Hauxue Fence 1989,25(6) 367.9111528. Anal. Lett. 1990 23 1069. 9111529. Anal. Lett. 1990 23 1107. 9111533. Appl. Spectrosc. 1990 44 9 19. 9111534. Appl. Spectrosc. 1990 44 934. 9111547. Anal. Chem. 1990 62 1656. 9111549. Anal. Chem. 1990 62 1983. 9111548. Anal. Chem. 1990 62 1893. 9111550. Anal. Chem. 1990 62 1994.9111570. Guangpuxue Yu Guangpu Fenxi 1989,9(6) 50. 9111584. Talanta 1990 37 71 1. 9111585. Talanta 1990 37 719. 9111595. Zh. Anal Khim. 1990 45 920. 9111598. Zh. Anal. Khim. 1990 45 942.9112176. J. Chin. Chem. SOC. (Taipei) 1990 37 313. 9112189. J. Pharm. Biomed. Anal. 1 989 7 144 1. 9112204. Spectrochim. Acta Part B 1990 45 409. 9112212. Spectrochim Acta Part B 1990,45 537.9112213 Spectrochim Acta Part B 1990,45 56 1.9112223. Appl. Opt. 1990,29,334.91/2224. Appl. Opt. 1990 29 3135. 9112225. Appl. Opt. 1990 29 3150. 9112239. Chem. Listy 1990,84,934.9112401. Spectroscopy (Eugene Oreg.) 1990,5(6) 4 1.9112421. Yejin Fenxi 1989 9 4. 9112423. Yejin Fenxi 1989 9 51. 9112426. Appl. Spectrosc. 1990 44 1 1 17. 9112427. Appl. Spectrosc. 1990 44 1124. 9112445. Spectrochim. Acta Part B 1990 45 1025. 9112447. Spectrochim. Acta Part B 1990 45 1075. 9112451. Bunko Kenkyu 1990 39 88. 9112460. Anal. Chem. 1990 62 2506. 9112469. Anal. Chim. Acta 1990 234 433. 9112471. Anal. Chim. Acta 1990 235 405. 9112483. Guangpuxue Yu Guangpu Fenxi 1990,10(1) 38. 9112496. Lihua Jianyan Huaxue Fence 1990 26 33. 9112506. Fresenius J. Anal. Chem. 1990,337,275.9112513. Fresenius J. Anal. Chem. 1990,337 860. 9112514. Fresen- ius J. Anal. Chem. 1990 338 9. 9112523. Chem. Pap. 1990 44 339. 9112542. Huanjing Huaxue 1987 8(6) 46. 9112549. J. Autom. Chem. 1989,11 156.9112563. J. Trace Elem. Electrolytes Health Dis. 1990 4 41. 9112576. Microchem. J. 1990,42 103.9112610. Z. Chem. 1990,30 222. 9112653. Kexue Tongbao 1990 35 526. 9112678. Quim. Anal. (Barcelona) 1989 8 247.