1Sample introduction1.1PreconcentrationThe published articles or conference presentations that have been included in this section have been divided into two broad categories. The first of these contains a survey of procedures in which the sample pretreatment involved either: discrete sampling; continuous flow; segmented flow; use of fluid flow characteristics for mixing; controlled residence times; mass transfer between phases (solid phase extraction, precipitation or liquid–liquid extraction); or direct introduction of the liquid phase into the spectrometer. This first section is simply titled 'flow injection'. The second section contains preconcentration procedures that do not fall within this rather broad definition of FI.1.1.1Flow injectionDevelopments in the last two or three years in the use of FI techniques with ETAAS have been reviewed.1Topics covered included air-segmentation and new procedures for precipitation and liquid–liquid extraction (LLE) as well as the use of the knotted tubular reactor as a sorption medium. Several other English language reviews have appeared2–5as well as ones in Russian,6Chinese7and Japanese.8The use of FI procedures with AAS in dissolution testing of oral dosage forms has been reviewed.9Stability constants have been determined10based on FI-ion exchange with FAAS.An FI microsampling system for the determination of Fe in soils by FAAS has been developed.11The LOD was 0.4 µg ml−1and the throughput 300 h−1. Copper was determined in beer12at 30 samples h−1. Honey samples were diluted in a continuous flow manifold13with a solution containing HCl and lanthanum for the determination of Ca, Mg and Zn by FAAS.The majority of the published papers concerning FI with atomic spectrometry in this review period are concerned with some aspect ofpreconcentration and/or matrix separation. Precipitated metal hydroxides were collected on zirconia spheres in a micro-column,14re-dissolved in 70 µl of dilute HNO3and determined directly by FAAS. Detection limits of between 0.1 and 0.5 µg l−1for Cu, Ni, Pb and Zn were obtained. Zinc was preconcentrated15as the thiocyanate complex on polyurethane foam (packed into a micro-column) and determined by FAAS following elution with water–acetone–nitric acid (68 + 30 + 2). For 3 min loading, the LOD was 0.8 µg l−1.Liquid–liquid extractionis less popular than solid phase extraction, probably because of the difficulty of phase separation; however, a few LLE methods have been described. Chinese workers determined (a) Pb in biological specimens16by extraction of the iodo-complexes into IBMK, and (b) Cr and Ni in human hair17by extraction of the DDC complexes into the same solvent. In both cases quantification was by FAAS. In a 'wetting film' system the oxine complex of Cu was extracted18into a film of IBMK on the walls of a PTFE reactor followed by back extraction into a slug of HNO3solution sandwiched between two air segments. Even though the eluent was delivered to the FAA spectrometer at 0.6 ml min−1, the throughput was 30 h−1with an enrichment factor of 43 for an LOD of 0.2 µg l−1. Four parallel hollow fibres were impregnated19with bis(2-ethylhexyl)phosphate in kerosene for the preconcentration of Pb from an aqueous formate buffer solution. Again, elution was with dilute HNO3. Wang and co-workers described20a procedure in which metals were retained by an iminodiacetate–ethylcellulose membrane mounted on either side of the flow channel. The eluent was 300 µl of dilute HNO3for the determination of Cd, Co, Cu, Mn, Ni, Pb and V by FAAS or ICP-OES with LOD between 0.2 and 1 µg l−1and enrichment factors from 7 to 18. Chelation was also used to retain Cd21on a poly(aminophosphonic acid) resin for determination in mussels. The solution LOD was 0.6 µg l−1for a sample volume of 3.4 ml with preconcentration factors ranging from 16 to 47 for sample volumes of 3.4–10 ml. The eluent was dilute HCl. The procedure was also applied in the determination of Co in steel.22The analyte was retained at pH 2 and eluted with dilute HNO3. Several groups of Chinese workers have devised preconcentration methods based on chelation with a solid phase extractant packed into a microcolumn: these include the determination of Pb in nickel sulfate by retention on YZ 50102,23the determination of Ag in geological materials by retention on triphenylphosphine-loaded glass beads,24the determination of Cu, FeIIIand Ni in caustic soda by retention on 'D401 chelating resin',25and the determination of copper by retention of the complex with 8-hydroxyquinoline-5-sulfonic acid on an anion-exchange resin, D-290.26Holcombe and coworkers compared27the performance ofpoly(l-cysteine)with that of 8-hydroxyquinoline. Both chelating agents were immobilized on controlled pore glass. The analytes were Cd, Cu and Pb and the matrices were synthetic sea-water, cobalt and nickel and Seawater CRM. Recoveries of 50 µg l−1of Cd and Pb from synthetic sea-water were quantitative for both resins; however, for solutions containing 10 000-fold excess of cobalt and nickel, the recoveries of Cd and Pb were low with the oxine material, but quantitative for the poly(l-cysteine) material. Neither material gave acceptable recovery for Cu. For the determination of Cd and Pb spikes in the CRM, the recoveries were variable. Recoveries of Cd were high, though that of Pb was acceptable with CASS-1 but low for NASS-2. Breakthrough curves were used to investigate the nature of the various binding sites on the materials and the various binding constants and capacities were estimated. Some preliminary results28for the behaviour of poly(l-aspartic acid) immobilized on gold-plated minigrids have been obtained. Sea-water was the matrix for one of the relatively fewsorbent extractionFAAS procedures published in this review period.29The analytes, Co, Fe, Mn, Ni and Zn, were retained as the complexes with 5,7-dichloro-8-hydroxyquinoline on C18silica and eluted with acidified (pH 2) methanol. Sensitivity enhancements of between 60 and 80 were obtained with LOD ranging from 0.5 to 4 µg l−1for a 1 min loading time at 7 ml min−1. The procedure was suitable for the determination of Mn and Zn in sea-water samples, but the concentrations of Co, Fe and Ni were too low to be determined. There was no discussion of the effect of sample volume on the LOD.In thedetermination of CrIIIand CrVI,30the CrVIwas retained as the APDC complex on the interior surface of a PEEK tube and then eluted with IBMK. The CrIIIwas retained from a solution of potassium hydrogen phthalate on C18; CrVIwas not retained, but could be preconcentrated downstream via the APDC–PEEK tubing procedure. The CrIIIwas eluted with methanol. For 5 ml sample volume the LOD were 0.9 and 0.5 µg l−1for CrIIIand CrVI, respectively. Preconcentration, following ion chromatographic separation, on a solid-phase extraction cartridge31with AAS detection allowed the quantification of both cations (Ca and Mg) and anions (halides, nitrate and sulfate), apparently. Cyclamate was determined32by anindirect continuous flow FAAS procedurein which oxidation to sulfate was followed by precipitation of the lead salt. Following collection on a filter and washing with ethanol, the precipitated lead sulfate was dissolved in ammonium acetate solution. The Ca and Mg content of brines was determined33by a combined low- and high-pressure system. The metals were retained as ion pairs of the xylenol orange chelates with tetrabutylammonium hydroxide on C18in the low pressure system and eluted with methanol and delivered to the spectrometer via a hydraulic high pressure nebulizer.The Ni and Sn content of brass was determined34by a procedure in which themetals were dissolved electrolyticallyin a flow system with collection of the electrolyte in the autosampler cup of an AA spectrometer with ETA. Apparently a sample could be analysed every 2 min. The procedure was used to determine35Al in aluminium alloys, and Cu and Zn in brass and bronze, and has been further developed36for the determination (in metal alloys) of Cu, Pb and Zn by FAAS and Fe, Ni and Sn by ETAAS.Adams and co-workers have devised a number of preconcentrationprocedures for use with ETAAS. Blood was analysed37for Cd and Pb; the complexes with ammonium diethylphosphorodithioate were retained on the inner walls of a knotted PTFE reactor followed by elution with 35 µl of methanol. The determinations were made with a pyrolytic graphite tube pre-treated with iridium. The LOD were 0.2 and 2 ng l−1for Cd and Pb, respectively. Platinum was determined38in the same matrix by an analogous procedure in which the analyte was retained as the APDC complex. For a 90-s preconcentration at 8.8 ml min−1, a preconcentration factor of 112 and an LOD of 10 ng l−1were obtained. The performance characteristics of the procedure were improved39bypre-coating the reactor with the chelating agent. This avoids the need for reagent addition, there are no losses to tubing walls outside the reactor, there is no need for a pre-fill step between samples of different concentrations and it allows for better optimization of the processes of reagent sorption and analyte preconcentration. This new procedure was applied39to the determination of Cu and Mn in some biological RMs. The chelating reagent was 1-phenyl-3-methyl-4-benzoyl-5-one, which was first pumped through the reactor as a 0.05% solution for 30 s at 2.5 ml min−1. The excess was removed by air pumped at 5 ml min−1for 15 s. The sample was delivered to the reactor at 2–2.5 ml min−1for 30 s and the liquid was flushed out with air. The adsorbed chelates were eluted with 30–40 µl of methanol delivered by an air carrier directly to the pyrolytic graphite-coated atomizer. The enhancement factors were 34 and 21 and the LOD were 6 and 5 ng l−1, for a sampling frequency of 22 h−1. Single cell protein, cod muscle and freeze-dried animal blood were accurately analysed.A microcolumn of Muromac-A-1 chelating resin was incorporated40into the autosampler capillary for the preconcentration of Cd, Co and Ni from sea-water matrices. Sample volumes of 800–1800 µl were loaded through the resin and the retained analytes eluted with 20% HNO3; LOD were 0.1, 7 and 33 ng l−1for Cd, Co and Ni, respectively. Three saline water RMs (NASS-4, CASS-3 and SLEW-1) were accurately analysed. A somewhat similar procedure was subjected41to a Simplex optimization procedure. The microcolumn was packed with silica gel functionalized with 1-(di-2-pyridyl)methylene thiocarbonohydrazide, the analyte was Ni and the eluent 2 mHNO3. The optimized procedure had an LOD of 60 ng l−1and a throughput of 36 h−1for 60 s preconcentration.A solid phase extractant, PbO2, highly selective for lead consisting of a crown ether macrocyle immobilized on silica gel was used in the analysis42of some biological RMs. The material retained Pb over a wide range of acidities (0.08–3 mol l−1HNO3) and the only interferences were from barium, potassium and strontium, whose ionic radii are similar to that of Pb. For a loading time of 20 s at 3 ml min−1, with elution into 46 µl of 0.03 mol l−1EDTA (delivered by an air carrier to the furnace) an enhancement factor of 23 and an LOD of 2 ng l−1were obtained. The retention efficiency was 70% and the throughput was 23 h−1. For the determination of Co43a column of fullerene (C60) was used to retain the APDC complex. An autosampler cup was used as the interface between the FI system and spectrometer. The analyte in a 25-ml sample volume was eluted with 500 µl of IBMK, delivered with a nitrogen carrier, and the LOD was 8 ng l−1. A similar procedure has been applied44for the determination of Bi in steels and aluminium alloys. The analyte was retained on activated carbon as the diethyldithiophosphoric acid complex. The aluminium or iron complexes were washed from the column with 500 µl of dilute HCl and then the analyte complex was eluted to an autosampler cup with 500 µl of ethanol. The enhancement factor for 10 ml of sample (loaded at 2.5 ml min−1) was initially given as 37 and the LOD was 50 ng l−1. When the work was published,45the enhancement factor was corrected to 14. Retention of Al and Cu on a microcolumn of Chelex-10046followed by elution with 1 ml of dilute HNO3was used in a procedure for the analysis of dialysis concentrates. For the determination of Cd in sea-water,47a column of C18retained either the complex with 4-(2-pyridylazo)resorcinol or 2-(2-pyridylazo)-5-dimethylaminophenol. A 50 µl portion of the methanol eluent was injected into the graphite furnace. The LOD was between 2 and 4 ng l−1. In a somewhat similar procedure, a C18column retained48the APDC complex of Cd from a sea-water matrix. The analyte was eluted with 80 µl of methanol for a LOD of 42 pmol in 2 ml (0.2 ng l−1). Chinese workers49determined Cd, Cu and Pb retained as their 5-sulfo-8-hydroxyquinoline complexes on a C18column. The analytes were eluted with a 180 µl mixture of HCl and HNO3and a 30 µl sub-sample was transferred by an air carrier to the atomizer. The enrichment factors were 22, 28 and 26 and the LOD were 0.7, 4.2 and 5.4 ng l−1for Cd, Cu and Pb, respectively. Burgueraet al. devised a flow-based procedure50for the determination of Bi in blood in which the sample was first digested in a flow-through reactor in a microwave oven, the analyte being separated by precipitation with tin(ii)from alkaline solution and collected in a knotted coil in an ultrasonic bath, and finally the precipitate was dissolved in HNO3and a 20 µl sub-sample was introduced into the atomizer. The LOD was 0.4 µg l−1. Kruget al.51,52preconcentrated Pb electrochemically on a W‐coil atomizer, which was temporarily functioning as the cathode in a flow-through electrolysis cell. For a 2-min preconcentration at 1 ml min−1, an enrichment factor of 25 and an LOD of 0.2 µg l−1were obtained.Martinez and co-workers have developedFI-SPE procedures for use with ICP-OES.53,54Bismuth was determined53in urine by retention of the 8‐hydroxyquinoline complex on an Amberlite XAD-7 column, with elution by HNO3, with an LOD of 30 ng l−1(100 ml solution) by use of an ultrasonic nebulizer. However, for the determination of Pb in tap water,54retention of the APDC complex on a knotted reactor with dissolution in 4 mol l−1HCl was employed. For a 10-ml sample the LOD was 200 ng l−1(again ultrasonic nebulization was employed). An iminodiacetate–agarose adsorbent, IDA Novarose, was used55to accumulate Cd, Co, Cr, Cu, Fe, Mn, Ni and Zn at pH 4–8, prior to detection by ICP-OES. The retention of the trivalent cations (Fe and Cr ) was slow, whereas the divalent cations could be loaded at flow rates up to 80 ml min−1. The procedure was applied to river, tap and lake waters.Chinese workers56were able, apparently, to glue Au to a PVC nylon 6 resin microcolumn, prior to elution with thiourea and detection by ICP-OES. Following digestion in 30 ml of 50%aqua regia, 3 ml of 10 g l−1 animal glue were added and, after making up to 100 ml, a sub-sample was injected in the FI manifold for direct transport to the column. The LOD was 2 µg l−1.Two solid phase extraction columns in series57allowed cationic (on Chelex 100) and anionic (on AG-1 X-8) species of Cu and Mn to be preconcentrated. Following elution with dilute HCl the elements were determined by ICP-OES. The procedure was applied to the speciation of these elements in milk. On the addition of an acetate buffer, casein and other macromolecules precipitated; analysis of the filtrate and precipitate allowed the casein-bound metal content to be determined.An FI system was used58to introduce a 25-µl slurry sample in an air carrier down a glass tube inserted into an axial hole in a carbon electrode into anarc or spark. Sediment samples were ground and sieved to particle sizes of less than 38 µm and slurried in nitric acid with lanthanum oxide in an ultrasonic bath. The analyte was Al, and 40 determinations per hour were possible against aqueous standard solutions. Chinese workers59used FI as the interface betweencapillary isotachophoresisand ICP-OES. Unlike capillary electrophoresis, which has a flow rate of 0.02–1 µl min−1, the eluent from the isotachophoresis system flows at about 1 ml min−1, which makes interfacing with the spectrometer relatively simple. The system was used to separate and determine Cu and Cu–EDTA. The combination of ion chromatography with plasma source spectrometric detection has been critically compared5with other techniques such as NAA, GD-MS and ETV-ICP-MS.1.1.2Off-line preconcentrationPublications in this section have been divided into two broad categories: the first of these contains descriptions of procedures that would be relatively easy to interface directly with an atomic absorption or emission spectrometer, and the second contains descriptions of procedures that would be difficult to interface with such a spectrometer. Within each of these broad categories, the material is sub-divided into sections dealing with FAAS, ETAAS and ICP-OES.Baker's yeast (Saccharomyces cerevisiae) immobilized on sepiolite (the calcareous internal shell of the cuttlefish) was used60to preconcentrate Fe and Ni. Sample volumes of up to 500 ml were processed at 2.5 ml min−1. Elution was with 1 mol l−1HCl and the LOD were 60 and 90 µg l−1for Fe and Ni, respectively. A brass SRM was analysed. The procedure has also been applied61to the determination of Cd, Cu and Zn in river and sea-waters. The cupferron complexes of Cd and Pb were collected62on activated carbon in a batch procedure for the analysis of urine. After filtering and drying the retained metals were dissolved in dilute HNO3and determined by FAAS with the aid of a slotted tube atom retarder. The LOD were 0.03 and 0.3 µg l−1for Cd and Pb, respectively, for a 210 ml sample and 1.5 ml HNO3solution. A melamine–urea resin was used63in a Cr speciation procedure. The resin only retains CrVI, and thus this species can be preconcentrated and determined after elution with sodium acetate solution. Following oxidation with H2O2, total Cr was determined by the same procedure and thus CrIIIwas found by difference.Chinese workers have described a number of procedures including: (a) solid phase extraction on polyaminophenol resin to accumulate iron;64(b) solid phase extraction on dithizone-loaded silica to concentrate Cu; (c) the co-precipitation of Cd, Co, Mn and Ni with either copper pyrrolidinedithiocarbamate (PDC) or magnesium 8-hydroxyquinoline; and (d) the collection of Pb in soy sauce on activated carbon. Vietnamese workers described65a procedure for the determination of Au by FAAS following either precipitation with tellurium or extraction from an acid solution into IBMK. A liquid–liquid extraction procedure in which the analytes (Fe, Mo, Sn and V) were extracted from acid solution in the presence of methylenedisalicylohydroxamic acid into IBMK containing tributyl phosphate was used to analyse66food, sediment and crude petroleum.Arobotic sample pre-treatmentprocedure has been devised67which not only preconcentrates the analyte (Cu) as the PDC chelate on the interior of a knotted reactor but also performs sample weighing and digestion. The procedure was applied to the analysis of environmental and biological materials by ETAAS. Preconcentration on Dowex 1X8 or Chelex-100 has been used68to preconcentrate trace elements in high purity gallium, arsenic and arsenic oxides. The desorbed elements were determined by ETAAS, FAAS, NAA and ICP-OES. A diethylaminoethylcellulose anion exchange resin was used69to collect CrVIwhile other cations were collected by an iminodiacetate resin. The eluted metals (Cd, Co, Cr, Ni and Pb) were determined in some pharmaceutical substances by ETAAS or TXRF. Mineral waters were analysed70for Cd, Co, Ni, Pb and V by retention on a cellulose material modified with 8-hydroxyquinoline-5-sulfonic acid. The analytes in a 20-ml sample volume were eluted into less than 1 ml and determined by ETAAS. The LODs were 4, 100, 40, 60 and 200 ng l−1for Cd, Co, Ni, Pb and V, respectively. Elution with IBMK was used71to remove Cd, Co, Cr, Cu, Fe, Mn, Ni and Pb from a column of immobilized sodium DDC or APDC with direct introduction of the solvent into the atomizer. The LOD ranged from 40 to 300 ng l−1. The same solvent was used72to elute organo-Sn compounds from an Amberlite XAD-2 column impregnated with tropolone. The LOD was 10 ng l−1. Polyethylene powder conditioned with 1-(2-pyridylazo)-2-naphthol retained73the complexes of Cd, Cu, Pb and Zn prior to elution with perchloric acid solution. The procedure was applied to the analysis of haemodialysis concentrates. Japanese workers (describing their results in a Chinese journal) showed: (a) that In, Rh, Ru, Sr and V could be collected74as precipitates with chitosan and that the LOD for Ru was 60 ng l−1; and (b) that Cu and Pb could be collected75as the PDC complexes on a miniature membrane filter which was then dissolved in methylcellosolve prior to analysis by ETAAS.Polyaniline76extracted Cd, Cu, Pb and Sb from acid digests to which potassium iodide had been added in a procedure whose goal was to separate the analytes from potentially interfering matrix components (the volume of eluent was greater than the sample volume). The analytes were eluted with dilute HNO3and determined by ICP-OES; the recovery for Sb was about 75%. Coal fly ash (CRM) and a sea plant material were analysed. For the determination of Y, Bi, Th, U and other REE in steel,77all the metallic components of the sample were retained from a solution containing oxyethylidenediphosphonic acid (OEDPPA) on a column of tetraphenylmethylenediphosphine oxide on macroporous styrene–divinylbenzene. The iron was removed by rinsing with acid solution (0.5 mol l−1HNO3), followed by the analytes with 0.5 mol l−1OEDPPA at pH 4.5 for determination by ICP-OES or ICP-MS. Recoveries were >80% for all analytes except Th, for which the recovery was 65%. The 'metalfix chelamine resin' was used78to retain Cd, Ni and Pb from pH 4 solutions with elution by 1 mol l−1HCl. The eluent was analysed by ICP-OES and the procedure was applied to the analysis of fir wood. In the determination of organo-P (as phosphino-polycarboxylates) in oil production waters, silica-immobilized C18was used79to preconcentrate the analyte species and separate inorganic P. The analyte species were eluted with a boric acid buffer at pH 9 and determined by ICP-OES and ICP-MS with either cross-flow or ultrasonic nebulization (for which the LOD was about a factor of 2.5 better—20 µg l−1for ICP-OES and 0.5 µg l−1for ICP-MS). Japanese workers have devised procedures for: (a) the preconcentration,80by a factor of 1000 at µg l−1concentrations, of As on a column consisting of the iron(iii)complex of lysine polyacetic acid; and (b) the preconcentration of Cd, Co, Cu, Ni, Pb, V and Zn,81by factors of about 100 from sea and river waters, on a column containing xanthurenic acid immobilized on silica gel. In the former procedure, As was determined by ICP-OES, and in the latter both ICP-OES and ETAAS were used.Theliquid–liquid extraction of Pb82from an acid solution containing zinc hexamethylenedithiocarbamate into 2,6-dimethylheptan-4-one was used for sample pre-treatment in the analysis of tap and ground water samples by FAAS. The Pb was back-extracted into an aqueous acid solution containing copper. The LOD was 70 ng l−1. For the determination of Zn in sugar by FAAS,83the analyte was extracted as the xanthate complex into a surfactant phase and then back-extracted as the EDTA complex when the solution was cooled. A column of Amberlite XAD-484retained the complexes of Cd, Co, Cu, Ni and Pb with 1-nitro-2-naphthol in a procedure for the analysis of chemical grade potassium salts by FAAS. Total Hg in drinking water and methylmercury in air were determined85by an ETAAS procedure in which the complexes with 2,3-dimercaptopropane-1-sulfonate was retained on Sep-Pak C18cartridges. After elution with methanol, a sub-sample of 50 µl was introduced into the furnace. The LOD was 50 ng l−1. The complexes were eluted with HNO3in acetone, which was evaporated to near-dryness and the residue taken up in HNO3. The LOD ranged from 20 to 60 ng l−1. Chinese workers developed a procedure for the indirect determination of Si in a cobalt compound by extraction of the heteropolymolybdate into IBMK. After washing the organic layer to remove excess isopolymolybdate, the molybdenum was determined by FAAS with direct introduction of the organic solvent. The interference from phosphorus was overcome by a prior extraction of the phosphomolybdate into ethyl acetate. The LOD for Si was 0.2 mg l−1. Also in the Chinese literature are descriptions of: (a) a flotation procedure86for the determination of Ag, Cd, Cu and Pb in which the metals were collected from a mixture of potassium iodide, methylene blue and toluene; (b) a procedure for the preconcentration of Cu on microcrystalline naphthalene loaded with dithizone;87and (c) a procedure for the determination of Cd by co-precipitation with copper sulfide. In the first procedure, the analyte was eluted with DMF, which was introduced directly into the flame. Reductive precipitation with selenium as the collector88formed the basis of a method for the determination of Au, Pd, Pt and Rh in geochemical samples by FAAS and ICP-OES.Complexation andsorption on activated carbon89was used in a procedure for the determination of Bi and Mo by ETAAS. The complexing agent was dithiophosphoric acidO,O-diethyl ester and the analytes were retained on a bed of activated carbon on a filter paper. After drying, mixing with HNO3, drying and mixing with dilute HNO3, a 20 µl subsample was transferred to the atomizer. The procedure was applied to the analysis of steels, and the possible interference from the iron was overcome by reduction to iron(ii)with ascorbic acid. The same collection procedure was used90for the preconcentration of As from natural waters as the heteropolymolybdate. After collection by filtration, the 100 mg of carbon was suspended in 5 ml of 0.1 mol l−1acetic acid containing 0.02% of Zr and 10 µl was transferred to the furnace for measurement at 197.2 nm. The LOD was 20 ng l−1for a 1 l sample. A Chinese researcher has described91a procedure for the determination of Tl by retention of TlIIIon a tributylphosphate resin. The analyte was oxidized with a mixture of iron(iii)and H2O2and eluted with a solution of ammonium sulfite and ascorbic acid. The LOD was 3 µg l−1.Coprecipitationwith dithiophosphates was used92to preconcentrate As, Cu, Pb and Se from digests of biological materials for determination by ICP-OES. The LOD ranged from 5 to 400 µg kg−1. Chinese workers93separated Au by precipitation of the complex with malachite green on microcrystalline naphthalene which was collected, packed into a column and the Au eluted with 6 mol l−1HNO3. The eluent was evaporated to near-dryness and the residue taken up in HCl for determination by ICP-OES. A somewhat similar procedure94has been used for the determination of Zn. The 1-(2-triazolyazo)-2-naphthol complex was retained on a Sep-Pak C18cartridge and then eluted with ethanol. The ethanol was evaporated and the residue taken up in a mixture of HNO3and H2SO4and the Zn determined by ICP-OES with an LOD of 20 µg l−1. For the determination of Cd, Co, Cu, Fe, Ni, Pb and Zn in saline matrices (haemodialysis concentrates) the analytes were extracted as the complexes with 1-pyrrolidinedithiocarbamic acid onto 30 µm Amberlite XAD-2 resin. After filtering and washing, the resin was slurried with 1% Triton X‐100 solution and introduced directly into the spectrometer for determination by ICP-OES. Solid phase extraction has been combined with liquid–liquid extraction95for the determination of Sc in red mud down to an LOD of 10 µg kg−1. The analyte was first retained on a column of Dowex 50W-X8 from which several other components were eluted first before Sc and Y and other lanthanides. The eluate was extracted with di(2-ethylhexyl)phosphoric acid and the Sc back-extracted with NaOH solution for determination by ICP-OES.The As content of air was determined96by absorption in acid and then evaporation to dryness in the presence of graphite powder, which was then analysed by AES.Some preliminary results for the determination of organohalogen compounds in waters have been presented.97,98Both methods are based on the use of a solid phase extraction procedure (either on activated charcoal or a polyacrylate fibre) followed by detection in a helium MIP by OES.1.2Chemical vapour generationTsalev has reviewed recent developments with AAS detection.99Applications to speciation by both chromatographic (HPLC and GC) and non-chromatographic procedures are included. Sanz-Medelet al. have reviewed100the use oforganized surfactant assemblies(micelles, vesicles) and emulsions in procedures for AAS. Topics covered include the generation of volatile hydrides and ethylides, CV Hg generation and the 'synergic effects of using vesicles to improve the separation of reversed-phased HPLC and the detectability of AAS by on-line vesicular hydride generation'.1.2.1Fundamental studies in hydride generationExperimental conditions for the determination of As by AFS using FI-HG were optimized101based onmathematical modelling. The use of AFS restricted the range of parameters that could be varied, as the borohydride solution flow rate and concentration were fixed to ensure a constant generation of H2. Furthermore, in those experiments where an argon purge gas was added prior to the gas–liquid separator, the total flow of argon to the atomizer was kept the same. It was concluded that the optimal sensitivity and throughput were obtained with minimal GLS headspace volume and maximum carrier solution flow rate. For the HG-AAS determination of Se, optimum conditions were found with the aid of two 24experimental designs.102Helium (400 ml min−1) was used as the purge gas, and both H2(300 ml min−1) and O2(40 ml min−1) were added. The LOD was between 0.7 and 2 ng, but the sample volume was not specified (nor was the nature of the GLS). In the determination of Sb by FI-HG-AFS103it was found that a cooled GLS (10 °C) with reduced headspace gave 'taller and narrower' signal peaks and also stabilized the baseline. Other innovative features of the procedure included the addition of both a surfactant (Triton X-100) and a silicon antifoam agent, which eliminated matrix effects (samples included urban particulate matter, tap and sea waters).Carrero and Tyson have further developed104their procedure forgenerating hydrogen selenide from the surface of an anion-exchange resin. Sample solution containing SeIVwas first passed through the resin to preconcentrate the analyte, then borohydride was passed through to load a controlled amount of the reagent, and finally acid was passed to generate the hydrogen selenide which was subsequently trapped on the interior surface of a graphite furnace atomizer. The entire procedure was automated with a FIAS 200 FI system. For a 20 ml sample volume, the LOD was 4 ng l−1. There have been several reports of the use ofnebulizers and spray chambers as HG devicesand GLS for ICP-OES determinations. Tao and Sturgeon105modified a Meinhard-type nebulizer by the insertion of a capillary tube into the sample introduction channel through which the acidified sample was introduced. The borohydride solution was introduced via the conventional channel and the solutions mixed just prior to nebulization. A Scott-type spray chamber was used. The very short residence times in the solution phase allowed the determination of Se in the presence of 5% nickel, 2.5% cobalt and 20 mg l−1copper. The LOD was 2 µg l−1and the method was validated by the analysis of a nickel oxide CRM. The procedure was also used with ETAAS with in-atomizer collection of the hydrogen selenide. Chinese workers106described a 'multi-functional cyclone nebulizer–hydride generator' for the determination of As, Hg, Pb and Sb and107a combined HG-ultrasonic nebulizer based on a CETAC U-5000AT device. Both procedures were used in conjunction with ICP-OES. A moving-bed HG device, also for ICP-OES, has been devised.108Solid borohydride and an organic acid were co-immobilized on a moving strip onto which the sample solution was directed. The device was used in the determination of As, Se and Sb in a Chinese tea. It has been shown109that H2Se can be collected (at 200 °C) on a gold wire in a manner analogous to the amalgam trapping of mercury. When heated (600 °C) in a stream of H2, H2Se was released and detected by AFS. Electrochemical HG has been used for the determination of As by ETAAS110and quartz tube atomization AAS.111In the former procedure, the generated hydride was collected for 2 min and then transported to the atomizer to give an LOD of 20 ng l−1for a 200 µl sample volume. The atomizer was pre-treated with iridium, and 400 firings were possible without any change in performance. For the latter procedure, an LOD of 10 ng l−1(continuous flow made) was obtained with a throughput (FI mode) of 36 h−1. To study themechanism of atom formation of As in a flame, Tesfalidetet al.112constructed a miniature H2–O2burner inside the cavity of an electron spin resonance spectrometer. Radical recombination in the flame was identified as the atomization mechanism. The procedure was also used to study the spatial distribution of H radicals in a quartz tube atomizer. Thespatial distribution of Sb atomsin such an atomizer has been measured113in cross-section by using a charge coupled device detector. In an unheated flame-in-tube atomizer, the highest free atom density was at the centre and the distribution was not influenced by the purge gas and O2delivery rates, but was significantly affected by the position of the O2delivery capillary tip. In an externally heated tube the atom distribution was more homogeneous. At high analyte concentrations (into the roll-over region of the calibration), the atom density was higher both near the walls and at the centre. The presence of polyatomic particles on whose surface free atom decay was induced was considered responsible. Chinese workers, who studied the atomization of As114in a low-temperature atomizer (25–200 °C) with AFS detection, concluded, from experiments with a dual generation system (one for AsH3and one for H2), that the atomization mechanism involved collision with hydrogen radicals.Chinese workers have also investigated115the role of organic acids in the HG determination of Pb. In a medium containing potassium dichromate as oxidant, the interference from the resulting CrIIIwas overcome by the addition of lactic and malonic acids. A holed T-tube atomizer (flame heated) has improved116the tolerance for interference by other hydride-forming elements in the determination of As, Bi, Sb and Se compared with those observed in a conventional flame-heated quartz tube atomizer. Five 2 mm holes were evenly spaced along the length of the atomizer. Presumably the holes were in the bottom of the tube facing the flame. With regard to LOD, the only significant decrease was for As, for which the LOD in the holed tube of 0.2 µg l−1compared with 0.06 µg l−1for the conventional tube. The linear ranges were improved for all elements except Sb, which was the element that showed least improvement in interference tolerance, being more tolerant only to Se. Chinese workers investigated117the interference of tellurium on the HG-AFS determination of Se and showed that the interference could be removed by the addition of FeIII. Camero and Sturgeon118trapped the hydrides of As, Se and Sb on the interior surface of graphite furnace atomizer by electrostatic deposition. The generation/collection efficiencies were estimated to be 80, 71 and 62%, respectively, values which seem a little lower than those for collection on an iridium-coated surface.Workers at Xiamen University have a number of interesting developments to report.119First, they confirm theformation of ZnH2(see alsoref. 120). Second, they report thegeneration of CoH2and NiH2, and third they report an enhancement in sensitivity in the determination of Cd, Tl and Zn by HG-ICP-OES by the presence of Co, Ni and Te. The degree of enhancement was increased as the length of tubing between the GLS and the spectrometer was increased. This is in contrast to the behaviour observed in the absence of the enhancers.Several publications have described the results ofefforts to determine more than one hydride-forming analyte per sample. The problems to be overcome include: (a) the difficulty of adjustment of the various oxidation stages, particularly the production of SeIVand AsIIIfollowing oxidative sample pretreatment; and (b) the mutual interferences in both generation and atomization stages. In a procedure entitled 'simultaneous determination of As, Hg, Se and Sb',121the various hydrides were generated batchwise and trapped cryogenically (liquid nitrogen). The trapped species were simultaneously evaporated by electrical heating and transported to a He MIP for quantification by OES. The procedure was also applied to the determination of Bi, Pb and Sb.122A similar concept has been described for the determination of As, Sb, Se and Sn bygas-phase molecular absorption spectrometry. First Sn hydride was generated at low acidity, then the hydrides of As, Sb and Se were generated at higher acidity. All hydrides were trapped in a liquid nitrogen cryotrap and then volatilized simultaneously. A diode array spectrometer was used and multiple linear regression was applied to deconvolute the spectra obtained. The wavelengths of maximum absorption were 190, 198, 22 and 194 nm, respectively, and the LOD were 50, 20, 100 and 1000 µg l−1. Russian workers123claimed to have determined As, Hg and Se simultaneously by continuous flow HG-ICP-OES. Samples (skim milk powder and freeze-dried bovine liver) were decomposed by HNO3at 160 °C under pressure. The LOD were 0.2, 1 and 0.06 µg l−1for As, Hg and Se, respectively. For the determination of As, Bi, Ge, Sb, Se, Sn and Te in 30% zinc sulfate solution, Rigby and Brindle124divided the analytes into two groups. In the first, Se and Te were determined following heating in the presence of HCl for 25 min; in the second, As, Bi, Ge, Sb and Sn were determined after the addition ofl-cysteine andl-histidine (and some HCl). The hydrides were generated in a continuous flow (CF) system with determination by ICP-OES. Detection limits in the original sample solution were 3, 3, 8, 2, 3, 2 and 38 µg l−1for As, Bi, Ge, Sb, Se, Sn and Te, respectively. For the determination of As, Bi, Sb, Se and Te in nickel, Feng and Fu125also divided the analytes into two groups. The samples were dissolved in HNO3and the analytes co-precipitated with lanthanum hydroxide, filtered and redissolved in HCl. In the first group, As and Sb were determined following the addition of HCl, thiourea and ascorbic acid. In the second group, Bi, Se and Te were determined after the addition of HCl. The analysis was performed by HG-AFS, giving LOD in nickel of 0.1, 0.4, 0.2, 0.1 and 0.1 mg kg−1for As, Bi, Sb, Se and Te, respectively. A HG-ETAAS procedure with trapping on an iridium coating has been developed for the determination of As, Sb and Se.126The hydrides were generated in a CF system in a 2.5 m titanium reaction loop. The procedure was applied to the analysis of a number of environmental and clinical materials; in the determination of 1 mg kg−1of Sb in soil, up to 600 mg kg−1of As did not interfere. A procedure for the determination of As and Se in soluble coffee by ICP-OES has been developed.127Following oxidative pretreatment, the SeVIwas reduced to SeIVwith HCl; presumably the As remained as AsV. Steel was analysed for Se and Te by HG-AFS128following acid decomposition and removal of potentially interfering cations by cation exchange. Several instrumental developments for the determination of multiple hydride forming elements have been described.107,129As, Hg and Se were extracted130from coal bysub-critical waterprior to determination by chemical vapor generation AFS. Arsenic and Se were determined131in foodstuffs, after microwave assisted digestion in HNO3, by AAS. Selenium was reduced with HCl (after removal of the nitric acid by heating to dryness); arsenic was reduced with hydroxyammonium chloride, potassium iodide and ascorbic acid. Chinese workers132have developed a method for the determination of As and Hg in biological samples and patent medicines by HG-AFS. The pre-treatment conditions were strongly oxidizing.1.2.2Generation of other volatile compoundsLopez-Molinero and co-workers have devised procedures for the determination of As after volatilization as the trifluoride,133,134the chloride135and the bromide.136Both AAS135and ICP-OES133,134,136have been used for quantification. It is not clear what the advantages over HG might be—possibly greater freedom from matrix interferences, as the LOD would appear to be much poorer than those obtained with HG. The reactions are carried out in a batch mode in concentrated H2SO4. It has been shown that Ge can be determined via the tetrachloride136–138by ICP-OES, this time in a continuous flow procedure. The LOD was 0.2 µg l−1. Chloride in water was determined139by volatilization of the chlorine formed on reaction with permanganate in sulfuric acid medium with quantification by MIP-OES. The LOD was 10 µg l−1. A somewhat similar procedure was used for the determination of both chloride and organochlorine compounds in waters,140except that the final determination was by either dc or rf GD-OES. The LOD were between 0.1 and 0.5 µg l−1. Organochloride and organobromide compounds were determined by MIP-OES141after conversion to hydrogen chloride or bromide and trapping in NaOH solution. Organic species were separated from inorganic species by passage through a column of activated charcoal. The LOD were 8 and 3 µg l−1for Br and Cl, respectively. The ratio of15N∶14N in soils has been determined by OES.142Ethylation has been used143as the basis of the determination of Bi by FI-AAS. The LOD was 0.8 µg l−1and the procedure was successfully applied to the analysis of urine, for which only a simple 1 + 1 dilution was needed as sample pre-treatment.1.2.3Vapour generation of individual elements1.2.3.1ArsenicBoth AsIIIand AsVwere determined in sea-water144after preconcentration on activated alumina as the complexes with quinolin-8-ol-5-sulfonic acid by HG-AAS. The retained species were eluted with HCl; AsVwas not reduced. The LOD were 0.05 and 2 µg l−1, respectively for a 2 ml sample volume. Chinese workers have reduced AsVwithl-cysteine for determination by HG-AAS;145reduced AsVwith KI and ascorbic acid for the analysis of foods and beverages;146and used target factor analysis for the determination of four As species in waters.147Muinozet al. devised148a microwave-assisted distillation procedure for the separation of inorganic As species from seafood products prior to determination by HG-AAS. The concentrations found ranged between 0.05 and 1 mg kg−1. The As contents of a large number of water samples from the US, China and Canada have been determined149by HG-ETAAS with in-atomizer trapping on a palladium-coated cuvette. For a 25-ml sample the LOD was 0.3 ng l−1.Both AFS and AAS were evaluated for the determination of As in wine and beer by HG.150The only pre-treatment was de-gassing and sample solutions were injected directly in the 6 mol l−1HCl carrier in the FI manifold. The LOD was 0.3–0.5 µg l−1. Chinese workers digested the samples under auto reflux for the analysis of oil by HG-AFS,151and passed the hydrides through a solution of permanganate to selectively remove stibine in the determination of As in antimony trioxide.152Inorganic As species were selectively determined by HG-ICP-OES.153By control of the solution acidity, only AsIIIwas determined; AsVwas then reduced withl-cysteine (for 12 h) and total As was determined. Chinese workers determined As in 67 food coal-tar dyes by HG-ICP-OES,154and As in sediments after co-precipitation with aluminium hydroxide.155In the former study HCl and KI were used to reduce AsV, whereas, in the latter study, only HCl was used.The generation of volatile halides has also been used as the basis for the separation of As from unwanted matrix components. The generated species include the fluoride,134the bromide136and the chloride.135In a study of As in urine, HG was used156though the type of spectrometry was not specified. The investigators found elevated concentrations in the urine of some subjects which could not be assigned to a high consumption of seafood and they concluded 'that additional factors relevant in the exposure to As are still unidentified'.Arsenic species have been separated by HPLC with HG-AAS detection.157In this study, UV photolysis was found to be superior to microwave-assisted decomposition in the post-column reaction scheme to convert non-hydride active forms of arsenic to hydride active forms.1.2.3.2BismuthA comprehensive study of reaction media for the determination of Bi by HG-ICP-OES has been undertaken.158Tartaric acid was found to be the most effective medium in terms of efficiency of generation and control of interferences. The LOD was 0.3 µg l−1and the method was validated by the accurate analysis of some RM (water and silicate) and was applied to the analysis of iron ore and coal fly ash. Interferences from a number of transition metals and other hydride forming elements (and mercury) were evaluated. Chinese workers159generated bismuthine from a slurry of the sample (geological materials mixed withaqua regia) on the addition of 0.8% sodium borohydride solution. The LOD was 60 µg l−1for determination by AFS. The interference from copper was eliminated by the addition of solid thiourea. Both copper and nickel interferences were considerably reduced on masking with 1-(2-thiazolylazo)-p-cresol.160A batch generation system was used with detection by AAS. Urine was analysed by AAS143following ethylation of the Bi. The LOD in urine was 2 µg l−1.1.2.3.3CadmiumTo remove interferences in the determination of Cd in waste waters and sewage sludge by HG-AAS with room temperature atomization Munos-Olivas and Camara161added KCN (0.5%) to the borohydride solution. This must have produced considerable amounts of potentially hazardous HCN on merging with the acid carrier stream. The LOD was 50 ng l−1.1.2.3.4GermaniumThe effects of various acids on the determination of Ge by HG-ICP-OES have been evaluated.162It appears that tartaric might have been the best of the various acids investigated (hydrochloric, citric, oxalic, acetic and sulfosalicylic) in terms of interference tolerance. The LOD were all in the range 0.1–0.3 µg l−1regardless of which acid was used. Chinese workers also investigated the optimum conditions for HG,163this time for AFS detection. They found that phosphoric acid, or a mixture of phosphoric acid and a mineral acid, gave the best generation efficiency and freedom from interferences. The gas-phase interference from As was eliminated by passing the vapors through a column of HgCl2. The LOD was 0.4 µg l−1. The technique of HG-AFS was also used in a speciation procedure164in which inorganic Ge and carboxyethylgermanium sesquioxide were determined in health drinks. As was discussed above, Ge has been determined by ICP-OES after volatilization as the chloride.137,138The authors found that the LOD for HG and chloride generation (CG) were similar (around 0.2 µg l−1) but that CG was much more selective.1.2.3.5LeadBrindle and co-workers determined Pb in calcium carbonate materials (coral) by HG-DCP-OES.165Ferricyanide was added as oxidant to aid in the production of Pb in the +4 oxidation state. The LOD was 0.7 µg l−1. A specially designed gas–liquid separator was used. Chinese workers166,167also used ferricyanide in a procedure for the determination of Pb by HG-AFS with an LOD of 0.2 µg l−1. The procedure was applied to the analysis of geochemical RM. A mixture of phenanthroline, potassium thiocyanate and oxalic acid was used to mask interferences. Malekiet al. described168a procedure for the generation of plumbane on injection of a sample solution into a reaction vessel containing solid borohydride and solid tartaric acid in a heating block at 65 °C. The gases evolved were swept by a stream of nitrogen into one end of a flame-heated tube in an AA spectrometer. The LOD was a modest 4 µg l−1.1.2.3.6AntimonyA FI-HG-AAS procedure was developed for the determination of Sb in plant materials169in which the SbVproduced by digestion with oxidizing acids was reduced to SbIIIby KI–ascorbic acid. The LOD was 0.01 µg l−1. The sample digestion procedure was further developed170for the analysis of lipid-rich materials. The method was validated by the analysis of bovine liver and pig kidney CRM and applied to the determination of Sb in pigeon eggs, bream and deer livers. The LOD, based on the dry powder, was 7 ng kg−1. Dietzet al.171cryogenically trapped the stibine produced from a batch reactor, containing acetic acid, into which the aqueous sample and aqueous alkaline borohydride were separately introduced. The hydride was re‐volatilized in a stream of N2and the Sb detected by MIP-OES. The LOD was 0.09 µg l−1. A procedure for the determination of Sb in brass has been devised.172Potentially interfering matrix elements (cobalt, copper, iron and nickel) were precipitated as the hydroxides.1.2.3.7SeleniumThere is still considerable interest in the development of methods for the measurement of Se in a variety of materials by HG with atomic spectrometry. As it is impossible to generate hydrogen selenide from SeVI, any strongly oxidizing sample pre-treatment has to be followed by a reduction step in which the Se is reduced to SeIV. For the determination of Se in urine a microwave-assisted procedure with nitric acid and hydrogen peroxide was developed.173Fumes of NOx, which cause a spectral interference in the subsequent HG-AAS determination, were eliminated by the addition of urea. Meat was digested in a mixture of nitric, sulfuric and percholoric acids,174steel was dissolved in nitric, hydrochloric and perchloric acids,128and hair was digested in nitric and perchloric acids.175For each of these procedures, the SeVIwas reduced to SeIVby heating in the presence of HCl. Li and co-workers noted173that organoselenium compounds were partially decomposed by this process, and concluded that a simple procedure for distinguishing inorganic selenate from inorganic selenite in the presence of organoselenium compounds was not possible. Presumably, distinction between borohydride active 'inorganic' forms of Se and borohydride unreactive 'organoSe' would not be possible either. A procedure for the selective determination of selenite has been developed176in which six different extraction solutions were evaluated and the Se was preconcentrated and separated by retention on an anion-exchange resin. An anion-exchange resin was used177to remove the interference from copper (retained as anionic chloro complexes) and to preconcentrate the Se as selenotrisulfide. Chinese workers178devised a double HG method in which the Se was converted to hydrogen selenide from a relatively large sample volume and trapped in a relatively small volume of alkaline hydrogen peroxide solution, from which hydrogen selenide was generated and detected by AFS. The LOD was 6 ng l−1. By trapping H2Se on a hot gold wire109Se was preconcentrated to give an LOD of 5 ng l−1. The trapped hydride was released at 600 °C into a stream of H2and detected by AFS. Soluble Se-species in agricultural drainage waters and soil sediment extracts were distinguished on the basis of resistance to oxidation.179Selenium(iv)was determined by AAS following HG from the sample solution with no pre-treatment. OrganoSe was determined by HG-AAS after oxidation with alkaline persulfate, and finally total Se was determined after oxidation with 30% H2O2at 90 °C, followed by reduction to SeIVby persulfate (20 min at 90 °C). The reduction of SeVIby a reagent that normally behaves as a strong oxidant is an interesting finding.1.2.3.8TinA FI-HG-AAS procedure was developed for the determination of Sn in sea-water,180with a LOD of 0.1 µg l−1for a 500 µl sample.1.2.3.9TelluriumSteel samples were decomposed with nitric, hydrochloric and perchloric acids.128Following precipitation with hydroxide and re-dissolution in HCl, the sample solution was passed through a strong cation-exchange resin. The eluent was mixed with HCl and boiled for a few minutes and the Ge (and Se) were determined by HG-AFS.1.2.3.10MercuryRussian181and Czech182workers have reviewed the determination of Hg. The first of these concentrates on the AA determination in soils, while the second is concerned with the determination of organoHg compounds in various environmental samples. Russian workers have also reported183on the determination of Hg in biological materials from cadavers. A new mussel tissue SRM has been issued.184The CV-AAS determination of Hg was used in support of studies of the efficiency of polymer-enhanced ultrafiltration,185a new technique for the removal of heavy metals from aqueous solutions. Provided that 10% HCl was added, the presence of the water-soluble polymer polyethylimine did not cause any interference. A procedure for thecharacterization of thiol (–SH) binding groupshas been developed186in which free HgIIand thiol-bound HgIIwere distinguished on the basis of their reactivity towards borohydride solutions of different concentrations. Quantitative reduction of HgIIto Hg0takes place with a specific amount of sodium tetrahydroborate according to the 'stoichiometric reaction of mercury with alkaline NaBH4', but the reduction of HgII–thiol complexes may require up to 6 orders of magnitude molar excess of borohydride. The mercury was determined by AFS. On the other hand,187the 'sulfide interference' was overcome by treating the sample solution with solutions of sodium hydroxide, sodium hypochlorite (NaClO) and copper sulfate, added in that order, followed by the addition of 10% tin chloride in 0.5 mol l−1sulfuric acid. The determination was by AAS. A sequential injection AAS procedure has been developed188in which borohydride was used as the generating agent. Small volumes of sample in low acidity HCl solution (0.05 mol l−1) and reagent (15–30 µl, 0.2–1%) separated by air segments were first drawn into the holding coil. Then vapor generation was initiated on reversing the flow and flushing the solutions through a flow-through gas–liquid separator. The LOD was 0.1 µg l−1and the consumption of reagents, compared with the FI procedure, was decreased by a factor of 25. Preliminary information189about a new miniaturized AFS system in which the thermally decomposed samples were introduced from a metal ETV device has been provided. An MIP-AES system with amalgam trapping preconcentration is being developed.190Russian workers have developed191anamalgam trapping deviceconsisting of a column of silica or alumina granules coated with sponge gold.191Several studies have focused on the extraction/digestion procedure for the determination of total Hg. Three different acid extraction procedures were evaluated192for the CV-AAS analysis of soils. Some of the extracts were exposed to UV radiation for up to 4 h. Mercury vapour was generated by the addition of 3% sodium borohydride solution to the sample solution diluted 1 + 1 with HCl. For the analysis of sediments, mercury was solubilized by microwave-assisted extraction in a sealed vessel with nitric acid.193Borohydride solution was again used, but only at a concentration of 0.2 %, with detection by AAS. Coal was analysed194following microwave digestion inaqua regiaby FI-CV-AAS and FI-CV-ICP-MS. The results were compared with those obtained with a LECO pyrolysis system and by NAA. The LOD were 80 ng l−1and 6 ng l−1for AAS and ICP-MS, respectively. Urine was analyzed195by a procedure in which bromate, bromide and HCl were added prior to digestion in a flow through microwave system. Quantification was by AAS with the PerkinElmer FIMS (flow injection mercury system). The LOD was around 50 ng l−1. For the determination of Hg in blood and plasma, room temperature bromination was used after overnight digestion or heating for 4 h. Elemental Hg was produced by reduction with tin(ii)and quantification was by AFS, with a throughput of 20 h−1. The LOD were 0.5 and 0.9 nmol l−1for plasma and blood, respectively. To avoid some sources of systematic error in the analyses of biological materials, a single vessel procedure has been developed.196The samples were digested with a mixture of nitric and sulfuric acids; the Hg vapour generated on the addition of SnIIwas quantified by AAS. A similar procedure has been used197for the analysis of bovine kidney. Following the acid digestion, 30% H2O2was added. The final determination was by AAS. Chinese workers198analysed foodstuffs by first mixing the solid sample with vanadium pentoxide and nitric acid and leaving overnight. Then sulfuric acid was added and the mixture heated to 140 °C. Potassium permangante was then added until a pink coloration was observed, and finally the excess permanganate was removed with hydroxyammonium chloride. Quantification was by AAS. A somewhat similar procedure was used199by other Chinese workers for the determination of Hg in household waste materials A microwave assisted digestion procedure has been devised200for the analysis of foliage by CV-AFS. Water samples201were analysed (AAS) by direct introduction into an FI system in which the sample carrier of 3% HCl merged with 3% SnCl2in 10% HCl. Total Hg in hydrocarbons and natural gas condensate has been determined by CV-AFS.202Samples were vaporized at 400 °C and all mercury species were collected on a gold trap at 200 °C. On heating to 900 °C metallic Hg was released. The procedure was applicable to the determination of mercury chloride, methylmercury chloride, ethylmercury chloride, phenylmercury chloride, dimethylmercury, diethylmercury, and diphenylmercury. Atmospheric particulates were analyzed203for their mercury content by pyrolysis-gold amalgamation-thermal desorption-AFS. It was found that results for denuder-based methods were higher than those of a conventional filter procedure, as mercury-bearing gold particles were flaking off the gold-coated denuder surfaces.In an effort to improve the LOD of an ICP-OES procedure for the determination of Hg, the sample solution was preconcentrated204by retention of the Hg on an anion-exchange column loaded with 1,5-bis[(2-pyridyl)-3-sulfophenylmethylene]thiocarbonohydrazide. The Hg was eluted with 2 mol l−1nitric acid and merged with a steam of tin(ii). The LOD was 4 µg l−1, which is not as low as can be achieved by AAS or AFS. For the determination of Hg in cosmetics,205an APDC complex was retained on a C18solid phase extractant by reaction with tin(ii)and detection by AFS. The LOD was single digit µg l−1. Trebleet al.have devised206a batch procedure in which the mercury was retained from a solution containing iodide on a cation-exchange resin impregnated with quinine. The mercury was desorbed by thiourea in HCl and Hg vapor was generated on the addition of alkaline potassium borohydride. The LOD was 1 µg l−1.Someconflicting resultshave been reported for the determination of inorganic-Hg (i‐Hg) and methylmercury (m-Hg) based on the reactions with borohydride and tin(ii). Burguera and co-workers described a procedure for the determination of i‐Hg and total-Hg in urine.207The basis of the method is that i‐Hg is determined by CVAAS after the addition of tin(ii)and total-Hg is determined following the microwave-assisted oxidation of organoHg with persulfate in a FI system. The LOD was 0.1 µg l−1. Thus, the authors assumed that organoHg compounds do not react with SnIIto give mercury vapour. These workers also applied the procedure for the determination of the same species in fish egg oil.208Samples were handled as emulsions, though this time the Hg vapor was generated with borohydride, and thus the authors assumed either that this reagent produced a species which did not absorb at 253.7 nm or that borohydride did not react with organoHg compounds. A similar concept was used by Rio-Segade and Bendicho209for the analysis of biological and environmental samples. Persulfate in sulfuric acid was used as the oxidant and the mercury vapour was generated on the addition of SnII. Distinction between i-Hg and total-Hg was based on heating the reactants, in a CF system, to 85 °C. At this temperature organomercury compounds were decomposed to i‐Hg, whereas at room temperature only i‐Hg reacted with SnII. These authors also claimed210to be able to determine m‐Hg, which had been selectively extracted (ultrasound assisted) from fish tissue by 2 mol l−1HCl, by forming Hg vapour on the addition of borohydride. Treatment with 5 mol l−1HCl extracted i‐Hg from which Hg vapor was generated by adding SnII. They also claimed in a later publication211that Hg0was generated by the post-column merging of a stream of borohydride (0.01%) with the column eluent (10 mmol l−1tetrabutylammonium bromide and 25 mmol l−1sodium chloride in 60% methanol) containing separated i‐Hg and m‐Hg. While it appears to be agreed that SnIIgenerates Hg0only from i-Hg, it is not clear whether borohydride generates Hg0from organoHg compounds (specifically methylmercury). Work in progress (C. D. Palmer and J. F. Tyson212) supports the hypothesis that methylmercury hydride is produced when borohydride reacts with m-Hg and that this compound is stable with respect to decomposition to Hg0at room temperature in a quartz tube atomizer and so no signal is observed when borohydride reacts with m-Hg. Chwastowskaet al.213claimed to be able to determine both i-Hg an alkyl-Hg following 'reduction by SnII'. The species were separated from the sample matrix with the aid of resin loaded with 2-mercaptobenzothiazole. The resin retained both species, which were subsequently eluted with a solution of thiourea in HCl. Inorganic-Hg was first determined, and then cadmium was added to displace Hg from organoHg compounds which reacted with SnIIto give Hg0. This is a modification of the procedure originally described by Magos.214The LOD was 10 ng l−1. Taoet al.used215a TMAH sample dissolution procedure for the determination of i-Hg in biological tissues. The i-Hg was released by the addition ofl-cysteine and reduced to Hg0on the addition of SnII. The LOD was 0.1 µg l−1.1.3NebulizationAs usual, there has been a large contribution to the literature concerning sample introduction in this review period. Among the papers that have reviewed or given an overview of sample introduction is one by Montaseret al.216containing 202 references, that gave the basic concepts of sample introduction for ICP spectrometry. A paper that challenged many of the 'accepted' theories has been presented by Olesik.217Instead of regarding the spray chamber as a droplet size filter that, when used in conjunction with a pneumatic nebulizer, contributes to the very low sample transport efficiency (often 1–2%), he claimed that the low transport is caused by many other processes including droplet–droplet collisions and coagulation and evaporation of aerosol and solution from the spray chamber walls. Since larger droplets have higher momentum, they cannot follow the gas flow and hence impact on the side of the spray chamber. Therefore, by decreasing the size of the droplets by either using a lower uptake rate, by using a solvent that is more volatile than water, or by heating the aerosol to cause evaporation, analyte transport rates could approach 100%. Other factors that affect the transport rate include acid or salt matrices that alter the extent of evaporation. A variety of nebulizer–spray chamber assemblies were assessed. A more applications-based review has been produced that concentrated on biota analysis by flow injection (FI) coupled with atomic spectrometry.218The review contained 95 references and information about the type of samples analyzed and summaries of the procedures.The introduction of liquids to microwave induced plasmas (MIP) has continued to attract considerable attention. This has been achieved using a variety of sample introduction devices.An ultrasonic nebulizer(USN) was used to introduce water samples at a rate of 35 µl min−1into an argon MIP operated at 150 W.219Figures of merit including linear range, precision, LOD and accuracy were reported. The LOD for 31 elements were between 0.3 and 1000 µg l−1and the accuracy was verified by the analysis of certified water samples. The same sample introduction system has been used to introduce acid digests of a reference material (pine needles) into a vertically positioned, axially viewed, aerosol-cooled plasma.220The torch cooling system for the aerosol generation and circulation was described in detail. Ahydraulic high pressure nebulizer(HHPN) used in conjunction with a desolvation unit has been used to introduce different Cr species eluting from a high-performance liquid chromatography (HPLC) column.221The sample was introduced to a BST Rutin C18RP column and the mobile phase was 15% methanol, 0.1 mmtetrabutylammonium acetate, 0.1 mmammonium acetate and 1 mmphosphoric acid. The linear range covered two orders of magnitude and the absolute LODs were 13 ng for CrIIIand 18 ng for CrVI. Severe interferences from easily ionizable elements were observed during the analysis of real samples. Athermospraynebulizer constructed in-house, used in conjunction with a desolvation device, also made in-house, has been described.222The design and construction of both devices was reported in detail. Under optimum operating conditions of 100 W forward power, a support gas flow rate of 0.75 l min−1, carrier gas flow rate of 0.8 l min−1, observation height 8 mm above the top of the torch, sample uptake rate of 1.5 ml min−1, capillary temperature 320 °C, aerosol spray chamber temperature 200 °C and condenser cooling water temperature 0 °C, LODs were found to be in the range 0.019–426 µg l−1and precision was 0.4–2% RSD. Inclusion of up to 20% methanol in the analyte solution was found not to disturb the plasma, hence offering the potential for coupling with HPLC.Very little in terms of novelty has been reported for sample introduction to flame atomic absorption spectrometers (FAAS). Two papers written in Chinese have reported the determination of Zn in milk or milk powder. One223used an emulsifying agent and reported that the results were in agreement with those obtained using a dry-ashing procedure. The other224used pulse nebulization with injection volumes of 100 µl. This latter paper reported a LOD of 28 ng ml−1, an average recovery of 99% and precision at 3.3%. No interferences were reported. Another Chinese paper reported the determination of several analytes (Cd, Cr, Cu, Fe, Mn, Ni, Pb and Zn) in air after passing them through a micropore membrane filter at a rate of 80 l min−1.225After collection, the membranes were acid digested and the resulting solutions analyzed by FAAS. Digestion was reported as being complete and recoveries using the standard additions method were 91–110%. Total Cr and CrVIhas been determined in cigarette ash and smoke using FI and HHPN.226The FI approach involved a sorption preconcentration system. The system was described in more detail (in Hungarian) in another paper.227The majority of novel sample introduction systems described in the literature have concerned ICP-AES detection. A wide variety of different nebulizer types have been described. Asonic spray nebulizer(SSN) has been developed that has an orifice diameter of 250 µm.228The orifice was made of a polyamide material to prevent metal contamination and allowed a gas flow rate of 1 l min−1. A comparison with a conventional concentric nebulizer was made and it was found that the LOD were similar, although the sample uptake was very much reduced (1–50 µl min−1compared with 850 µl min−1for the conventional nebulizer). Absolute sensitivity was improved by a factor of 13.Direct sample insertion (DSI)is a technique that is still being used by some authors because it enables both liquid and solid samples to be analysed directly. A review of the technique containing 79 references has been made by Sing.229The review provides an overview of the instrumentation, operating parameters, figures of merit and applications. An application of DSI in which several analytes (Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb and Zn) were determined by ICP-AES in sea-water and ground water samples has also been presented.230The presence of several salts and common solvents such as hydrochloric, nitric, sulfuric and phosphoric acids was found to decrease the emission intensity. The extent of the matrix interferences was element dependent.Direct injection high efficiency nebulizers(DIHEN) have been described in two presentations.231,232The device enables a transport efficiency of 100% to be achieved, whilst introducing the sample at a flow rate of 1–100 µl min−1. The analysis of 'difficult' samplese.g., petroleum products, was described in one paper,231and nebulizer diagnostic techniques based on optical patternation were used to characterize the aerosol in the other.232An assortment of other nebulizers have been used. These include a modified concentric glass nebulizer capable of analysing high salt content samples.233It is a modification of both a standard Meinhard and the LB nebulizer. It is reported as being capable of aspirating a saturated solution of sodium chloride (26.54%) for two hours without clogging and the stability is superior to that of the LB nebulizer. The analytical figures of merit are also favourable when compared with the Meinhard, with higher SBR and lower LOD. A comparison of several different nebulizer types [high efficiency nebulizer (HEN), microconcentric nebulizer (MCN), Micromist and a conventional concentric nebulizer] operating at low sample uptake rates has been made.234Parameters investigated included the gas back pressure, the free liquid aspiration rate, primary and tertiary aerosol droplet size and the solvent and analyte transport rate. It was found that the conventional nebulizer produced a coarser aerosol and subsequently had a lower solution transport rate through the spray chamber and a lower sensitivity than the other nebulizers do. Overall, it was found that the HEN provided the lowest limits of detection. Amicrowave thermospray nebulizer (MWTN)has been developed that operates by breaking the liquid stream by the expansion of the solvent vapour inside a capillary when heated by a microwave field.235It was found that a certain level of dissolved ions must be present in the sample for nebulization to occur and that the analysis of pure water was not possible. An interesting characteristic of the nebulizer is that the higher the dissolved salt content, the finer the aerosol produced. For samples containing dissolved ions, the LODs and sensitivities obtained using this nebulizer were improved when compared with a conventional nebulizer. A modified nebulizer has been described that minimizes transition element interferences with selenium hydride generation (HG) ICP-AES.105A capillary tube is placed in the sample introduction channel of a conventional Meinhard nebulizer. This carries the acidified solution into the sodium tetrahydroborate solution that is introducedviathe normal sample introduction channel. Gas–liquid separation is performed by a conventional Scott double-pass spray chamber. It was found that this small modification enabled Se at the 0.5 mg l−1level to be determined without interference from 50 g l−1nickel, 25 g l−1cobalt and 20 g l−1copper (all introduced as their 2+ salts). The LOD was 2 µg l−1, precision was typically <2% and the method was validated by the analysis of NIST certified material nickel oxide.Ultrasonic nebulizers(USN) have again found wide usage for a number of applications. A USN has been used in conjunction with a long torch to improve the sensitivity in an axially viewed horizontal ICP.236Sample transport efficiency was increased when compared with a cross flow nebulizer and hence net signal intensity increased and the intensity and fluctuation of the background decreased, thereby yielding improved LODs and dynamic ranges. The device was used during the analysis of a water CRM. Since the sample transport rate using a USN is substantially higher than with conventional nebulizers, desolvation is often necessary to prevent plasma perturbation. A paper that has addressed this has been published by Allenet al.237A microporous membrane desolvator (Cetac MDX-100) was used whilst determining the effects of sodium on the analytical signal of several analytes. The authors concluded that the enhancement of the signal caused by the sodium was greater in the presence of the desolvator but that the mechanism by which the sodium interfered was the same. A HG-USN system has been described that is capable of determining both hydride forming analytes and other analytes simultaneously.107Using the spray chamber as the gas–liquid separator, it was found that the LODs and SBR of the HG-USN device were the same as, or superior to, HG and USN separately. A USN has been used for the determination of Bi in urine using a FI preconcentration technique.53After digestion of the urine using nitric acid, the solution was taken to near dryness on a sand bath. After re-extraction with nitric acid, the white ash was taken up inaqua-regia, diluted, mixed with ethanolic 8-hydroxyquinoline adjusted to pH 5 and passed through a column. The Bi–8′-hydroxyquinoline complex was eluted with 2 mnitric acid. The LOD was 0.03 µg l−1, precision at 2 µg l−1was 2.5% and the recoveries for 1–8 µg l−1were in the range 93–100.5%. Matrix effects caused by the use of a USN have been discussed by Budic.238The effects of calcium, potassium, sodium, nitric acid and other matrices during the determination of several analytes were discussed. It was found that, in the presence of 20% nitric acid, the transport of the sample decreased by 8% when compared with water and that the presence of 1 mg ml−1matrix elements decreased the sample transport efficiency by 10%. Consequently, this led to a reduction in emission intensity. It was also found that the presence of potassium enhanced both the electron density and excitation temperature, whereas the presence of the other ions had the opposite effect.Matrix effectshave also been discussed for a number of nebulizer–spray chamber designs. An overview of nebulizer diagnostics, including a discussion of tools such as optical imaging (e.g., high speed photography), particle imaging velocimetry, optical patternation, Fraunhofer laser diffraction, phase Doppler particle analysis, rainbow refractometry, absolute intensity of scattered light and the ratiometric technique, has been made.239Another overview of matrix effects and how they can be minimized and compensated for has been made by Mermet.240There have been several papers that addressed the effects of acids on the determination of analytes. In one paper, the transient effects of different acids on the signals of several analytes were discussed.241It was found that the presence of the acid changed the extent of aerosol evaporation in the spray chamber and hence the sample transport rate. The effect was dependent on the type of acid used. A similar paper by the same authors stated that, as the concentration of nitric acid was increased over the range 0–25%, the sample transport rate decreased, but that the nebulizer gas flow rate also had a large effect.242At low flow rates (0.7–1 l min−1) there was a decrease in transport efficiency up to a nitric acid concentration of 2%, but there was no further decrease up to a concentration of 25%. At gas flows higher than 1.3 l min−1there was a continual decrease in transport efficiency. It was therefore concluded that the acid dependent changes in aerosol properties occurred mainly during the transport of the aerosol through the spray chamber. Acid effects for a number of low-flow sample introduction methods was discussed by Todoli and Mermet.243One method used a MCN with a Scott double-pass spray chamber, another two used desolvation systems comprising two Liebig condensers in series or a porous PTFE membrane, respectively, both using MCNs, and the last system used a DIN. Liquid flow rates ranged from 5 to 120 µl min−1and a number of acids including 0.9 mnitric, hydrochloric and sulfuric and 3.6 mnitric acid was tested. It was found that for the MCN with the spray chamber, the acid effects increased as the flow rate decreased. The opposite was true for the desolvation systems and the DIN. At a flow rate of <30 µl min−1no effects were observed if a conventional desolvation system was used at a temperature of 160 °C. In another similar paper, Todoli compared the performances of several nebulizer designs (pneumatic concentric micronebulizer, HEN, Micromist, MCN and DIN) with respect to acid introduction.244Aerosol characteristics and figures of merit, such as LOD and background equivalent concentration (BEC), were compared. It was concluded that the DIN suffered fewer matrix effects. The analysis of plant material has also led to some matrix effects that were removed if higher plasma power was used,e.g., 1450 W.245Depressions in emission intensity from the analytes As, Cd, Co, Cr, Cu, Ni, Pb and Se were observed in matrices containing calcium, iron, magnesium, manganese, phosphorus, potassium and sodium and this depression increased with increasing concentration of the concomitant species. A correction for volatility differences between organic sample analytes and standards in organic solutions has been made.246The method is based upon the measurement of the analyte signal at two different spray chamber temperatures. The technique required no prior knowledge of the chemical structure of the analyte and tests with organosilicon and organochlorine compounds were found to reduce the error by a factor of between 2 and 30.Spray chambersare the second, but equally important, component of most sample introduction systems. These come in a variety of designs, many of which have been discussed in the literature during this review period. Several designs (a Scott double-pass made of Ryton and three cyclone spray chambers made of glass, polypropylene and PTFE) were compared for their characteristics of the tertiary aerosol, solvent transport rate, analyte transport rate and net emission intensity, LOD and BEC.247Interferences produced by the presence of widely used acids were also evaluated. In all cases, a glass concentric nebulizer was used to introduce the sample. Results indicated that the glass and polypropylene chambers produced the most coarse tertiary aerosols, but also had the greatest solution transport rates, highest emission intensity, and lower LODs and BECs than the other chambers. The position of the nebulizer was found to be critical for the cyclone spray chambers. Acid effects were found to increase as the tertiary aerosols became more fine and so the Scott double-pass spray chamber exhibited the worst performance. The cyclonic spray chambers were also found to exhibit better short-term stability. A doughnut-shaped spray chamber that had an internal volume of 102 cm3has been described by Liuet al.248A Meinhard nebulizer mounted radially at the entrance arm produced a spray that was split into two flows by a central hollow cylinder. These flows were then recombined at the exit of the spray chamber and transported to the plasma. This spray chamber was found to exhibit higher sample transport efficiency, more stable internal pressure, higher signal intensity, lower BEC and improved precision and LODs when compared with a conventional Scott style spray chamber. Another spray chamber design that reputedly had improved washout characteristics (with a four order of magnitude change in 60 s), improved sensitivity by an order of magnitude and lower LODs has also been reported.249Unfortunately, the design of the chamber was not discussed in the abstract. It was stated, however, that the long and short term stabilities were also improved. The performance of three different spray chambers used in conjunction with a USN has been compared.250A conventional Fassel type, a Cetac and a spray chamber designed and constructed in-house (details given) were the types compared for the analysis of transient signals. The performance was judged in terms of the washout period required, with shorter times indicating a better performance. The spray chamber designed in-house was found to be the best by far with a washout period 1/5th that of the Fassel and 3/5th that of the Cetac. In addition, it had a similar sensitivity to the Cetac which, in turn, was superior to the Fassel. Consequently, the Cetac and the chamber designed in-house had improved LODs when compared with the Fassel.Computational and simulationpapers that attempt to model the processes within the spray chambers have also been published. Two papers by Berndtet al. discussed the application of such programmes.251,252By transferring the physical flow domain (the spray chamber) to the computational domain, a grid consisting typically of 50 000–150 000 volumes was constructed, then mathematical algorithms were used to obtain numerical solutions for the individual volumes, which enabled a distribution of velocity components and pressure to be constructed. In this way it is possible to calculate the equations of motion for each droplet through the spray chamber. The authors have constructed data for both a cyclone and a Scott style spray chamber. The results of the computational model were in good agreement with those obtained experimentally, with a calculated transport efficiency of 2.6% compared with an efficiency of 2.7% obtained experimentally. Despite these encouraging results, the authors stressed that the need to build and test prototype spray chambers still exists. A paper that discussed the simulation of atomic and ionic absorption and emission spectra for thermal plasma diagnostics has also been published.253Doppler, van der Waals, Stark and instrumental broadenings were all taken into account and the effect of self absorption was also considered. The input parameters were pressure, temperature, electron density and the emitting species number density. The simulation method was applied to the diagnosis of volatilization phenomena in thermal plasma spraying processes and the selection of diagnostic lines and instrumental and data processing parameters. It was also used to evaluate the accuracy of the temperature and density distributions calculated from spectroscopic measurements.1.4Solid samplingSolids analysis utilizing both AAS and AES continues to be a significant area of research. A number of instrumental configurations are used for the introduction of either slurried or solid samples into flames, furnaces and plasmas. As in previous years, many of the publications in this area are application based and these reports are summarized inTable 1.Solid samplingType of atomizationType of samplingElem.MatrixSample treatment/commentsRef.CV-AASSlurry, prepared using ultrasonic mixingHgBiological, environmentalTotal Hg determined in samples using flow injection. Samples suspended in 15% HNO3(mussel tissue, aquatic plant), or 9 + 1 15% HNO3–15% HCl (river sediment, sewage sludge). Results were comparable with those obtained after microwave digestion276ETAASSlurryAs, Cd, Cu, PbFoodSlurries prepared in 0.15% agar as a stabilizing agent. Phosphate (for Pb and Cd) and Pd–MgNO3(for As) used as modifiers277ETAASSlurryBa, Cu, Fe, Pb, ZnTea leavesBa and Pb determined using standard additions, Cu, Fe and Zn determined using aqueous standards. Results from slurry analysis of SRM compared well with those obtained using ICP-AES after microwave digestion. Sensitivity and simplicity found to be the benefits of slurry method; however, for determination of several analytes ICP-AES may be faster278ETAASSlurryBeRice and flourLOD 1.4 pg obtained, no details of method given in abstract279ETAASSlurryTeInSbPalladium nitrate used as modifier. LOD 0.4 µg g−1. Results agreed well with solution ETAAS and ICPMS280ETAASSlurry, automatic preparation using Ar bubblingCo, Cu, NiSediments and soilsGround samples suspended in HF, samples containing Cu and Ni partially digested using microwave. Fast heating programme used281ETAASSlurry, automatic preparation using ultrasonic probeSbGeologicalChemometric techiques used for the optimization of procedure259ETAASSlurry, automatic preparation using ultrasonic probe, manually prepared using hand-held ultrasonic probe or ultrasonic bathPbMussel tissue and SRMsPreparation procedures compared for extraction of Pb. Quantitative extraction into liquid only possible using ultrasonic probe. This helps avoid sedimentation and volumetric errors. However, analytical results obtained using automatic preparation are superior258ETAASSlurry, prepared by ultrasonic mixingCo, Cu, NiBotanicalCoulter particle analyser used to study the effect of grinding and ultrasonic agitation on slurry analysis. 1 h grinding produced similar size distribution for all materials282ETAASSlurry, prepared by ultrasonic mixingCu, FeOld manuscripts6 mm discs punched from paper samples, ground in a mortar, suspended in 5 ml H2O and sonicated for 2 min. 20 µl samples taken for analysis. Calibration by aqueous standards and by standard addition to clean filter paper used. RSDs 0.8–10% for Fe and 0.3–7.6% for Cu. Laser ablation ICP-MS was used for metal distribution studies283ETAASSlurry, prepared in an ultrasonic bathCdCalcium drugsA Mo-tube atomizer was employed with thiourea as chemical modifier. LOD 17 pg ml−1. Results in good agreement with those obtained from acid digested samples284ETAASSlurry, prepared manually and with ultrasonic mixingAsBaby foodSamples, 0.2–2 g, suspended in 5 ml solution containing 0.1% m/v Triton X‐100, 1% v/v HNO3, 20% v/v H2O2, 0.3% v/v nickel nitrate. 20 µl aliquots taken for analysis during stirring. Fast furnace programme used. Method of standard additions used. LOD 23 pg285ETAASSlurry, prepared using ultrasonic mixingCdBiologicalSample (10–35 mg) mixed with 1 ml 60% PTFE (as modifier), 0.4 ml conc. HNO3and 0.2 ml 0.5% plant glue solution. The mixture was agitated ultrasonically for 60 mins. LOD 0.12 pg, RSD (n= 5) 7.8% for 1 ng ml−1286ETAASSlurry, prepared using ultrasonic mixingCo, Cr, NiWheat flourSample (150 mg) mixed with 5 ml 15% HNO3–10% H2O2and agitated ultrasonically for 15 min. 20 µl aliquot taken for analysis. LODs 23–44 ng g−1, RSDs 4–8% (nvalue not given)287ETAASSlurry, prepared using ultrasonic mixingPbBiologicalAmmonium phosphate modifier and glycerol used. LOD 6.8 pg288ETAASSolidCdEquine muscleDirect analysis used for screening of tissues. Results obtained for direct solid analysis and for analysis of digested samples were closely correlated and were statistically no different289ETAASSolidCu, Fe, MnFlour SRMSample was vaporized in one graphite furnace and the generated aerosol was transported in Ar and electrostatically deposited on a platform in a second graphite furnace. Multi-element analysis carried out via continuum source coherent forward scattering spectrometer. Results obtained in agreement with certified values290ETAASSolidPbMuscle tissueAnalysis of samples contaminated with gun-shot residue. Solid sampling suitable for determination of Pb level in non-contaminated samples and for the discrimination between original Pb content and Pb derived from gun-shot291ETAASSolidSbPVCSamples (2.5-3.5 mg) were inserted into enlarged furnace injection hole through a paper cone. RSD (n= 14) = 12.6%, which was considered satisfactory for screening purposes.256ETAASSolidVarious (11)Tungsten trioxide and tungsten blue oxideHigh background due to vaporizing oxides eliminated by use of hydrogen purge gas during pyrolysis. LODs 0.07 (Mg, Na, Zn)–2 (Ni) and 0.01 (Mg, Na, Ni)–1.7 (Fe) ng g−1for tungsten trioxide and tungsten blue oxide, respectively292ETAASSolidVarious (11)Tungsten powderDirect analysis of high purity tungsten powders with an ETAAS system equipped with a transversely heated atomizer, deuterium background correction and solid sampling introduction by the 'boat' technique. LODs in range 0.01–4 ng g−1. RSDs (n= 5) 1.1–30.8%. (See also 293)294ETAASSolidVarious (15)Titanium barsSample cut from Ti bar and etched with HNO3/HF. Samples (<30 mg) placed onto platform and inserted into furnace. LODs from 0.02 ng g−1(Mg) to 30 ng g−1(Sn) (See also 294 and 295).296ETAASSolidVarious (9)Aluminium oxide powderSample (0.06–6 mg) placed onto a pyrolytic graphite coated platform and inserted into furnace. Platform re-coated every 5–6 firings to reduce background absorbance. LODs 25, 3.5, 2, 10, 0.8, 0.25, 1.5, 12 and 0.5 ng g−1for Co, Cr, Cu, Fe, K, Mg, Mn, Ni and Zn, respectively. (See also 294 and 296)295ETAASSolidZnGeological RMCalibration graphs prepared using different solid RMs had different slopes. To overcome errors, cluster analysis was used to select materials of similar composition and three-dimensional calibrations using intensities, masses and contents were used. The latter was the preferred approach297Flame AASSlurryCaMaize flourSolution of flour with 0.15% gelatin, LOD 0.12 mg l−1298Flame AASSlurryCa, Fe, K, Mg, Na, ZnVegetable tissuesOptimized conditions found for slurry analysis. 50 mg lyophilized vegetable material slurried for 5 min using ultrasound and diluted to 50 ml with 0.720 mHNO3. RSDs <3% for all analytes. LODs 2.2, 1.8, 3.4, 8.7, 2.1 and 1.7 mg l−1for Ca, Fe, Mg, Na, K and Zn, respectively299Flame AASSlurryCuChinese medicinal herbsPowdered sample (1.5 g) suspended in 25 ml 0.15% agar solution. After shaking, 4 ml aliquot mixed with 1 ml 0.15% agar and 1-5 ml Cu (II) standard solution (10 µg ml−1) and H2O to 25 ml. LOD 57 ng ml−1, RSD (n= 6) <2.5%.300Flame AASSlurryMnChinese medicinal herbsSample treatment identical to that described in 300. LOD 13.6 ng ml−1301Flame AASSlurryZnMilk fat5 surfactants compared for preparation of homogeneous emulsion. Triton X‐100 and NaDBS preferred302Flame AASSlurry, prepared from sample shaken with zirconia beads, particle size <0.8 µmCa, Cu, K, Mg, Na, ZnHairAcid predigestion used as a pre-treatment stage. Glycerol and Viscalex HV30 compared as wetting agents. Wetting agents were only beneficial for dilute slurries. LODs 5, 3.5, 3.1 and 20 mg kg−1for Ca, Cu, Fe and K, respectively, and 1.7, 0.6, and 3.5 µg kg−1for Mg, Na and Zn, respectively303ICP-AESCup-in-torchPbFingernailsRe cup inserted 1.5 cm above the carrier gas inlet. Drying and ashing carried out in-situ304ICP-AESDirect insertionCu, Fe, Mg, Mn, ZnWood pulpSolid samples were introduced into ICP using pyrolytically coated graphite probe after in-situ treatment with HCl and NaF. Samples were dried and ashed by heating probe prior to plasma ignition. LODs 50–1000 pg. Method suitable for screening of samples, significant time saving over dissolution methods obtained305ICP-AESDirect insertionVarious-An automated direct insertion device was described. Solution, powder and slurry sampling were compared and slurry sampling was found to be the most promising. Formation of carbides was eliminated by addition of SF6to the plasma gas306ICP-AESDirect insertion, sample preconcentrated on activated carbonCd, Cu, Pb, ZnWatersMetal complexes of oxalic acid, iminodiacetic acid or 8-hydroxyquinoline were collected on activated carbon. Loose particles of carbon were either loaded into direct insertion cup or complexes were adsorbed onto a direct insertion probe with an activated carbon cap. The specially machined cap produced the better results. Recoveries were poor although 8-hydroxyquinoline was the most promising complexing agent307MIP-AESSlurry ETV on W coilAg, CdWaste waterSamples were vaporized on W-coil. LODs 16 µg l−1for Ag, 1 µg l−1for Cd308ICP-AESETVVarious (16)Graphite, silicon carbideSolid sample, 20 µm for graphite and <5 µm for SiC, placed on platform and inserted into ETV unit. Freon added to Ar carrier gas309ICP-AESSlurryRare earthsLanthanum oxides——ICP-AESSlurryVarious (9)Milk and infant formulaSimultaneous analysis of slurry samples310ICP-AESSlurry ETVAl, Ti, YSi3Ni4powdersSamples were analysed with and without PTFE as a fluorinating agent. 90% of a 100 µg sample could be vaporised with no loss of analyte. LODs from 0.11 (Al) to 0.09 µg g−1(Ti) with RSD 1.9–4.2%311ICP-AESSlurry ETVRare earths (14)Lanthanum oxideSamples were analysed with PTFE emulsion as a fluorinating agent. LODs from 2 (Yb) to 130 ng ml−1(Ce) with RSD <5%312ICP-AESLaser ablationAl, Fe, MgEnvironmental samplesEffect of matrix, chemical and physical form of analyte and laser wavelength were studied264,313ICP-AESLaser ablationVariousVariousDescription of a novel bulk solid sampling system314ICP-AESSlurry ETV on W‐coilVarious (11)Al-based ceramic powdersSamples were vapourized on a W-coil. LODs between 0.01 (Mg) and 8.5 µg g−1(Co)315ICP-AES/ICP-MSLaser ablationVariousAmorphous solidsSpatial resolution of 50 µm obtained. RSDs 0.3–0.7% obtained for Al and Si with Ba as internal standard, and 1.5% for B with Si as internal standard316Spark AESSolidVariousSteelSteel sample is embedded in an ingot of Sn allowing analysis of small samples. Results in good agreement with those obtained from bulk samples270Nolteet al.254considered bothlaser ablation and slurry sampling for ICP-AESin their study of the influence of simultaneous versus sequential measurement of the signal, background and internal standard. The experiments were carried out using an array spectrometer allowing collection of simultaneous spectra; this permitted the use of a single data set to estimate the differences between measurements made in simultaneous and sequential modes. Simultaneous background correction and simultaneous internal standardization gave a mean RSD of 3.6%, sequential background correction and simultaneous internal standardization produced an average RSD of 6.1% and sequential background correction and sequential internal standardization produced an average RSD of 6.4%.The use ofsolid sampling for ETAASas a screening technique can provide a significant time saving over conventional digestion methods. Belarraet al. carried out a theoretical evaluation using 18 000 simulated results.255The optimum conditions for analysis were determined based on the type and frequency of outliers commonly encountered in solid sampling ETAAS. The median was found to be a more useful result than the mean due to the reduced effect of outliers. The minimum number of samples required for a screening method was determined as 5–20 for a guaranteed recall of 95% (recall = no. of correct identifications ÷ no. of attempted identifications) when the median was used. This sampling rate corresponds to between 15 min and 1 h of work, a significant reduction in time compared with the preparation of liquid samples. The same workers successfully applied solid sampling to the screening of antimony in PVC.256Slurry sampling for ETAAShas proved to be a popular technique for the analysis of solids with easy automation. Pioneer of the technique, Miller-Ihli, discussed the advantages of slurry analysis with ETAAS and provided evidence that the method can be successfully applied to ETV-ICP-MS.257Ultrasonic mixing, either manual or automated, is commonly the preparation method of choice for slurry sampling. The maintenance of a homogeneous slurry allows the sampling of both liquid and solid from sample cup to furnace. Amoedoet al. considered the extraction of Pb from solid samples using ultrasound with subsequent analysis of the supernatant alone.258Probe sonication allowed quantitative recoveries of Pb from several plant and animal SRMs. Several chemometric techniques were used by Cal-Prietoet al. for the optimization of a method for the determination of Sb in geological samples using automated slurry sampling ETAAS.259Plackett–Burman designs were used to assess the influence of sample mass, volume, HNO3concentration, Triton X‐100 concentration, ultrasonic probe power and sonication time, and to optimize the variables. A two-way ANOVA was used to evaluate the effects of sample cups and replicates. Control charts were also used to monitor graphite tube performance. In addition, two pipetting options and three quantification methods were studied. The usefulness of the optimized method was demonstrated by the accurate analysis of 5 SRMs.A review (115 references) oflaser ablation (LA) samplingby Russoet al. was published during the period of this review.260Calibration and optimization, fractionation, sensitivity enhancements, mass loading and particle transportation were discussed. A single calibration graph was successfully used for the determination of major elements in geological materials in LA-ICP-AES by Kanickyet al.261Sc and Y were used as internal standards and limestone and silicate samples were analysed. The same authors262,263measured the acoustic signals produced during laser ablation ICP-AES using a Nd:YAG laser at 266 nm. It was found that the acoustic signal could be used to focus the laser on the solid surface by monitoring the maximum in the acoustic signal. The intensity of the acoustic signal could also be used to determine the size of the crater produced during ablation.A systematic investigation into the effects of the matrix, chemical and physical form of the analyte and laser wavelength was carried out by Motelica-Heinoet al.264Synthetic samples were prepared from different crystalline compounds of Al, Fe and Mg spiked in SiO2or CaCO3pressed into pellets and analysed using LA-ICP-AES using a Nd∶YAG laser at 1064 and 266 nm. The experiments demonstrated the strong dependence of the LA-ICP-AES response factor on the chemical form of the analyte and on the bulk matrix composition. These effects were also dependent on the laser wavelength, and use of the UV laser did not lead to any improvement in minimization of the effects. In contrast to chemical effects, there was no effect on the response factor from the grain size or binding pressure of the pressed pellets. Matching of mineralogical and chemical composition of the matrix and of the chemical forms of the analytes for calibration standards is recommended to avoid systematic errors in quantitative analysis. The effect of the gas environment on laser ablation characteristics was investigated by two groups of workers. The effect of five noble gases (He, Ne, Ar, Kr and Xe) in the laser ablation sampling chamber was studied by Leunget al.265An enhancement of ICP emission intensity for laser sampling in He and Ne and a decrease for Kr and Xe relative to Ar was observed. Therefore, the use of He as the sample chamber gas can significantly improve sensitivity. Tharejaet al.266used an intensified CCD to image the plumes produced by the laser ablation of aluminium and PTFE substrates in Ar, He, air and O2at various pressures. The effect of ambient atmosphere on the expanding front of the plume was investigated photographically. Stratification of the plasma was observed at moderate laser intensities for both the metal and the polymer materials.The nature of the aerosols produced from the laser ablation of polymer materials using a UV laser was investigated by Todoli and Mermet.267Ca, Sn and Ti were determined in PVC and PE samples and the aerosols compared with those obtained from glass samples. For Ca, but not Sn or Ti, the aerosol particle size was dependent on the chemical form of the analyte. A second PVC sample, containing 11 elements, was also studied, and Al, C and Na exhibited different behaviour with particle size with respect to the remaining elements. Carbon was mainly present in gaseous form and in particles of less that 3 µm in diameter; this observation precludes its use as an internal standard to compensate for variations in the ablation process for polymers.Individual fluid inclusions were analysed using LA-ICP-AES.268The laser radiation is used to drill the solid sample until the fluid inclusion is reached. Emission from the excited atoms and ions in the fluid was analysed using a spectrometer equipped with a pulsed and gated multichannel detector. Na∶K, Na∶Ca and Na∶Li intensity ratios were measured. Calibrations were established using a range of standards, including glasses and fluid inclusions. LODs achieved were suitable for the determination of ions in inclusions.A review (79 references) ofdirect sample insertion for ICP-AESby Sing appeared this year.229A useful overview of the technique including instrumentation, operating parameters, system response, analytical figures of merit and applications is provided.A critical evaluation ofdc arc spectrometry for the analysis of solid samplesby Florianet al.269concluded that the modernized dc arc AES method, combining fibre optics and a multi-channel spectrometer, is a ready alternative to existing solid sampling methods for AES and AAS. Grientschniget al. proposed a procedure for the analysis of small samples of steel using spark discharge AES.270The procedure involved embedding the sample in an ingot of pure Sn. Results from the analysis of three steel samples prepared by the embedding method were in agreement with those obtained on bulk samples of the same steels. The method is suitable for samples having a diameter of greater than 6 mm.A review (20 references) in Japanese by Shimizu271considered the use ofglow discharge (GD)for multielement surface analysis. Shimizuet al.272assessed the resolution of GD-AES for the depth profiling of chemical species in anodic alumina films. The films, acting as cathodes, were sputtered in Ar at 3–5 Torr. Emission associated with the sputtered species (Al, B, Cr, H, P, W) was monitored throughout the experiment with a sampling interval of 0.01 s. The depth resolution was reported as being comparable to SIMS profiling of similar films.The use of plasma etching for the characterization of non-conducting materials by GD-AES was investigated by Barshicket al.273Studies were performed in Ar containing 0.01% by weight of CF4. When a conducting cathode such as Cu was used, the sputtering rate was decreased by a factor of five compared with the use of pure Ar as support gas. When a non-conducting glass was analysed, however, fluoride radicals formed in the discharge reacted with the substrate to form volatile SiF4, which was spontaneously released into the gas phase, carrying Cu and U with it. As a result, enhancements of 50 and 30% were observed for the determination of Cu and U, respectively, in glass.Workers in Marcus' group274conducted studies to determine the practical benefits of mixed discharge gases (Ar and He) in the bulk and depth-resolved analysis of solids using an RF GD-AES source. A number of parameters were assessed for both conductive and non-conductive sample matrices. It was observed that the addition of He to Ar did not improve detection limits in the bulk analysis of conductive materials. However, in the analysis of non-conductive materials, the addition of He was found to enhance analyte emission intensity without significantly influencing the sputtering characteristics. For depth resolution, optimized conditions, especially gas pressure, derived for pure Ar can be improved through the addition of He to the discharge.Marshall275evaluated the impact of lamp control parameters on the quality of both bulk and depth profile analysis by GD-AES. The lack of current and voltage stability was shown to have a larger influence on the quality of calibration curves, bulk check standard results and quantitative depth profile analysis than do similar fluctuations in pressure. Therefore, control modes that achieve constant current and voltage at the expense of constant pressure are recommended.1.5Electrothermal vaporization1.5.1ET-AASModification of the graphite tube surface with metals, or use of metallic atomizers, continues to be of interest in ET-AAS. Rademeyer and de Jager317discussed the use of 'permanent' metallic modifiers in ETAAS. The process of sputtering the metal onto the graphite surface was investigated. Physical characteristics of the graphite surface after sputtering with noble metals were evaluated and reported. Fundamental studies on the stabilizing effect of sputtered modifiers and solution modification were carried out. A permanent iridium modifier deposited on tungsten and zirconium-treated platforms was evaluated by Slaveykovaet al.318using XRF, ESCA and SEM. Two atomization processes were observed and explained by the differences between the Ir–W and Ir–Zr interactions and surface distribution. A similar study on the behaviour of As on tubes coated with a range of metals was carried out using XRD and SEM.319The performance of tube treatments in the determination of As with a palladium modifier were compared. Optimum sensitivity and precision were found with a Ta coating. Portable AAS systems based on tungsten-coil atomizers continue to be the subject of several presentations and publications. Temperature measurements of gas and coil surface temperatures have been made320and descriptions of systems are given.321,322Several papers comparing the performance ofchemical modifiers for ETAAShave provided some useful information for users of this technique. For the determination of Sr in biological fluids, La was found to provide optimal performance in an evaluation of 12 modifiers, including Pd.323Pd was found to be the best modifier for the determination of Cd in sewage sludge and Sn in PVC after an evaluation of background effects.324Mixed modifiers for particular applications also received some attention.325–3271.5.2ETV-ICP-AESFew papers have been published during the period of this review reporting fundamental studies with ETV-ICP-AES. A number of publications on the use of the technique for the direct analysis of solid samples have appeared and they are discussed in the previous section of this review. Work from Jones' group on the use of tungsten coils for vaporization was reviewed as conference presentations last year328and has subsequently been published.329The efficiency of sample introduction into an ICP via electrothermal vaporization was investigated using three methods by Kantor and Gucer.330The methods used were: (i) a direct method based on deposition of aerosol particles and analyte vapour by mixing the vaporization product with supersaturated steam and subsequently condensing the mixture followed by the determination of the amount of analyte collected; (ii) indirectly by dissolution and analysis of material deposited on interface components; and (iii) by calculation from line intensities when using ETV and pneumatic nebulization using Hg as a reference element. Using a 200 ml min−1Ar carrier flow, the transport efficiencies calculated were 67–76% for medium volatility elements such as Cu, Mg and Mn and 32–38% for volatile elements such as Cd and Zn. However, the addition of CCl4vapour to the internal Ar flow resulted in transport efficiencies of 67–73% for all five elements studied.1.5.3In-torch vaporization ICP-AESKaranassioset al.evaluated six vaporization chambers for the introduction of micro-samples to an ICP.331The chambers ranged in shape from conical to oval, had internal volumes ranging from 3.4–18 cm3and were tested using in-situ generated smoke. The best results were obtained with a 6.5 cm3internal volume chamber. The sample was deposited onto a Re filament forming a three-coil loop and inserted into the vaporization chamber attached to the ICP torch. Good analytical performance was obtained with the optimized system for both liquid and slurried samples.