摘要:

December 1 983 Vol. 108 No. 1293 The Analyst Atomic-absorption Spectroscopy as a Tool for the Determination of Inorganic Anions and Organic Compounds A Review Manuel Garcia-Vargas Miguel Milla and Juan Antonio Perez-Bustamante" Department of Analytical Chemistry Faculty of Sciences University of Cddiz Cddiz Sfiain Summary of Contents Introduction 1. Indirect methods 1.1. Methods based on the formation of compounds in solution 1.1.1. Reaction of precipitation 1.1.1.1. Determination of the metal excess 1.1.1.1 .a. Precipitation as sulphate determination of sulphur 1,l. 1.1. b. Precipitation as sulphide determination of sulphur 1.1.1.1 .c. Precipitation as chloride determination of chlorine 1.1.1.1 .d. Other types of precipitation determina-tion of organic compounds 1.1.1.2.a.Dissolution with EDTA solution deter-mination of sulphur 1.1.1.2.b. Dissolution with ammonia solution : determination of chlorine 1.1.1.2.c. Dissolution with iodide solution deter-mination of iodine 1.1.1.2.d. Dissolution with mineral acids deter-minations of sulphur and ammonia 1.1.1.2.e. Direct atomisation of the precipitate : determination of sulphate 1.1.1.2. Determination of the metal precipitated 1.1.2. Metal complexes 1.1.2.1. Analyte as ligand 1.1.2. I .a. Charged metal complexes determina-tions of cyanide sulphite pyrophos-phate fluoride and organic compounds 1.1.2.1 .b. Neutral metal complexes determina-tions of cyanide thiocyanate halides and organic compounds 1.1.2.2. Ion-association complexes 1.1.2.2.a.Inorganic anions determinations of perchlorate nitrate and iodide 1.1.2.2.b. Organic compounds. * To whom correspondence should be addressed. 141 1418 GARCIA-VARGAS et aZ. AAS FOR DETERMINATION OF ArtaZyst VoZ. 108 1.1.3. Heteropolyacid compounds determination of phosphate 1.1.3.1. Binary heteropolyacids 1.1.3.1.a. Single extraction 1.1.3.1 .b. Extraction - decomposition 1.1.3.1 .c. Precipitation of molybdophosphoric acid 1.1.3.2. Ternary heteropolyacids 1.1.4. Redox reactions 1.1.4.1. Inorganic anions determinations of iodide iodate, nitrate and chlorate 1.1.4.2. Organic compounds 1.2. Methods based on the formation of compounds in the atomisation cell 1.2.1. Suppression or enhancement of the atomic-absorption signal of a metal 1.2.1.1.Constant concentration of the metal added 1.2.1.1 .a. Alkali metals determination of fluorine 1.2.1.1 .b. Alkaline earth metals determinations of phosphate sulphate fluoride and organic compounds 1.2.1.1 .c. Transition metals determinations of fluoride sulphate and ammonia Variable concentration of the metal added 1.2.1 .%a. Atomic-absorption inhibition titration : determinations of sulphate and phos-phate 1.2.1.2.b. Atomic-absorption inhibition - release titration determination of fluoride Molecular-absorption spectrometry determination of halo-gens 1.2.1.2. 1.2.2. 1.3. Methods based on the formation of volatile compounds determi-nation of halogens 2. Direct methods 2.1. Non-metallic elements 2.1.1 Flame methods 2.1.1.1. Atomic-absorption spectroscopy determinations 2.1.1.2.Molecular-absorption spectroscopy determinations 2.1.2. Electrothermal methods determinations of phosphorus, of phosphorus sulphur and iodine of phosphorus and sulphur sulphur and iodine 2.2. Organic compounds 3. Conclusions Keywords Review ; inorganic anions ; organic compounds ; indirect methods ; atomic-absovption spcctvoscofi December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1419 INTRODUCTION Atomic-absorption spectroscopy (AAS) which has developed rapidly in recent years is mainly characterised by high speed sensitivity and selectivity in the determination of most metallic elements. In order to extend the range of application of AAS to other elements such as F, C1 Br I 0 S N P and C much attention has been paid to the development of direct and indirect methods of analysis.1-8 Considerable attention has also been devoted to the applica-tion of indirect methods permitting the determination of organic cornpounds.3~4~9p10 The reader interested in this field is strongly advised to examine the Annual Reports on Analytical Atomic Spectroscopy (published annually by the Royal Society of Chemistry since 1971) and Fundamental Reviews from Analytical Chemistry (published biennially on AAS since 1970).An up-to-date bibliography can be found in the 6-monthly bibliography published in Atomic Absorption NewsZetter (at present Atomic Spectroscopy). The direct methods available for the determination of the elements listed above by AAS are few and of low sensitivity (>5 pg ml-l per 1% absorption3p4) so that they cannot be readily determined at trace levels in solution.This is due to the fact that they exhibit their main resonance lines in the vacuum ultraviolet region (below 190 nm) and therefore they cannot be directly determined with a conventional instrument covering the spectral range from 190 to 850 nm. Further there are great difficulties in making absorption measure-ments in the vacuum ultraviolet region owing mainly to absorption by air absorption by the flame itself lack of an appropriate source and lack of transparency in the optical components. On the other hand when higher wavelengths (secondary or non-resonance lines) are used the sensitivity is poor. Therefore indirect methods are employed more by necessity than by choice.No systematic study of the determination of organic compounds by AAS appears to have been undertaken. Of the methods reported some are very selective and rely on the forma-tion of a particular complex and its subsequent extraction whereas others utilise a more general reaction that can be applied to the determination of a range of compounds. On the other hand a few direct methods have been proposed for the determination of organic compounds. A critical survey of AAS methods described for the determination of anions and organic compounds in various materials from 1965 up to early 1982 is presented here. The atomic-absorption methods used for the determination of anions and crganic com-pounds may be broadly classified into two main groups indirect and direct.The methods are readily seen as direct if the atomic absorption of the analyte is related to its concentration. In contrast indirect methods are based on the chemical reaction between the analyte and one or several species one of which is a metal easily measurable by AAS and a relationship is established between the metal atomic absorption and the concentration of the analyte. Obviously the applicability of these methods depends on the extent to which the selectivity for the determination of the analyte can be retained via the chemical reaction used before the final absorbance measurement. Difficulties can also arise from non-stoicheiometric reactions. On the other hand indirect methods are rarely used for routine analysis because most of them are difficult to automate.Taking into account the chemical reaction involved in each procedure indirect methods are further divided into methods based on the formation of compounds in solution and in the atomisation cell and of volatile compounds. The direct methods are classified as flame and flameless. In this review the methods considered are discussed critically on these bases. 1. INDIRECT METHODS The application of these methods to the determination of non-metallic elements and organic compounds involves carrying out a suitable chemical reaction. This reaction is essential for the performance of the procedure as problems arising from the atomic-absorption deter-mination are easily circumvented. Therefore the choice of one or several suitable reagents (one of which is a metallic element) to react with the analyte is very important.In these methods the atomic signal of a metal (consumed or unconsumed in the reaction) is directly or inversely related to the concentration of the analyte. These methods are divided into methods of formation of compounds in solution formation of compounds in the atomisation cell and formation of volatile compounds 1420 GARCIA-VARGAS et aZ. AAS FOR DETERMINATION OF Analyst VoZ. 108 These are classified into methods based on a reaction of precipitation with a metal ion, formation of metal complexes formation of heteropolyacid compounds and redox reactions. 1.1. Methods Based on the Formation of Compounds in Solution E DTA Ammonia Iodide Dilute SOf sol. sol. sol. acid detn. 1.1.1. Reaction of Precipitation These methods are usually based on the reaction of an anion (inorganic or organic species) with a solution of a cation of appropriate concentration to yield an insoluble compound.This compound should have a very low solubility. In these methods the maximum molar concentration of the anion in the samples must be exceeded by the molar concentration of the cation in order to obtain complete precipitation. The mixture is then filtered centri-fuged or decanted and the unconsumed metal or the metal incorporated in the precipitate is determined by AAS. In general these methods are specific only if the medium in which the anion is precipitated does not contain other anionic species that might also be precipitated. Interferences from cations in the anion-containing solution can be easily avoided by using a cation exchanger.These methods are divided into two groups determination of either the metal excess or the metal incorporated into the precipitate. In both instances the atomic-absorption signal of the metal is adequately related to the concentration of the analyte precipitated. A general scheme of the procedures employed for the determination of inorganic anions is illustrated in Fig. 1. Bay2- (aq) Anion (aq) Metal ion solution Pb Ag(NH&+ (aq) A g l l (aq) Metal ion (aq) Anion - metal ion ppted. SOa2- detn. CI- detn. Separation I- detn. 1 Excess of metal ion (aq) 1 ;2i bl- detn. Ba,Pb Pb,Hg,Zn Ag Anion - metal iodsolid) Fig. 1. General outline of procedures for the indirect determination of inorganic anions by means of their precipitation.1.1.1 .l. The determination by AAS of the unreacted metal ion used as the precipitation reagent in the supernatant solution is the basis of this method which is more often employed than the alternative methods based on the determination of the precipitated metal after dissolution of the precipitate as less manipulation is required. The standard solutions needed for the construction of calibration graphs are generally prepared from the anion standard solutions treated in the same way as the unknown samples. This procedure is more time consuming than the direct use of the metal standard solutions but it avoids errors due to the nature of the precipitation reaction and also some operational errors. Determination of the metal exces December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS.A REVIEW 1421 In this indirect method the procedures described have been systematised according to the anion involved in the reaction of precipitation. 1.1.1.1 .a. Sulphate ions are usually precipitated by a definite amount of barium chloride solution in a strongly acidic medium. The excess of barium remaining in solution is then measured by flame AAS at the 553.5-nm resonance line after several hours of digestion of the precipitate to ensure quantitative precipitation. It has been pointed out1 that this indirect method allows the determination of 5-60 pg ml-l of SO,” using a barium ion solution of concentration equal to or lower than 100 pg ml-l of Ba(I1). When barium standard solutions are used for the construction of calibration graphs the sulphate recovery ranged between 98.0 and 100.5%.11 A number of determinations of sulphur in different samples are summarised in Table I.Precipitation as sulphate determination of sdphur. TABLE I DETERMINATION OF SULPHUR BY PRECIPITATION AS BARIUM OR LEAD SULPHATE AND MEASURE-MENT BY FLAME AAS OF THE EXCESS OF METAL IN SOLUTION AT THE 553.6-nm (Ba) Sample Soil extracts* . . Potable water* . . Plants . . . . Natural water . . Polluted water . . Textile extracts . . NH,HCO Biological materials Fuel oils . . . , Surface waters, soils plants . . Water . I OR 283.3-nm (Pb) RESONANCE LINES Precipitation Separation of Metal measured ; medium excess of metal flame Comments Reference t Filter Ba; air - CaH2 Recovery is 96.5-103.8% 11 .. pH 2.2 buffer$ Filter Ba; air - C,H 12 . . HNO - HCIO - HCI Filter Ba; air - C,H Triple slot burner is used. Recovery is 0.02 N HC1 Decant Ba; N,O - C,H BaSO allowed to stand for 18 h. Relative 94.9-102.2% 13 standard deviations ranged from 22-1y0 for 13-508 wg ml-1 of SO,*-Method compares favourably with both turbi-dimetric and gravimetric methods 14 Centrifuge Ba; air - C,H t 0.002 N HCl Filter or decant Ba; air - C,H BaSO allowed to stand for a t least 12 h Filter Ba; air - C,H 40 g of sample decomposed with boiling water. Recovery 95.&98.8% t 0.1 M H N O ~ Centrifuge Ba; N,O - C,H S is oxidised by concentrated nitric acid a t 250 “C - Centrifuge Ba;N,O-C,H S is oxidised to sulphate by 30% H,O, after application of Schoniger’s oxygen flask method .. 40% ethanol Centrifuge Pb; air - coal PbSO allowed to stand for a t least 12 h. Either the 217.0- or the 283.3-nm Pb lines are used. Recovery 95-100% gas . . 25% ethanol Centrifuge Pb; air - CaHa SO oxidised to sulphate by 3% H,O1 Sulphides sulphospinels . . 0.05 M HCl Filter Ba; air - C,H Method applied to small size samples of ca. 10 mg. Brines . . . . . t Filter Ba;air-C2H Burner must be cleaned often and Ba originally present in brine must be taken into account Standard deviation is 1.47% * The approximate concentration of sulphate is calculated by measuring the conductivity of a sample solution. t At the pH of the solution. 3 Chloroacetic acid - Na,COa buffer solution. 15 16 17 18 19 20 21 22 23 The influence of foreign ions on the determination of sulphates via the barium- AAS procedure has been examined by several workers.In general to overcome ionisation inter-ferences an alkali metal is added to both standards and sample~.1~,1~*~7*~8 Varley and Chinll have shown that Na K and Mg in soil extracts cause no measurable interference a t any level on 50 pg ml-l of neither was there any interference from either Mn or Fe(I1) at concentrations of 100 pg ml-l. Ca and P do not interfere a t 30 and 10 pg ml-l respectively, but if they are present at higher concentrations it has been recommended by several workers that both samples and standard solutions should be matched with a lanthanum solution. MontieP2 has reported that in the air - acetylene flame errors due to the presence of alkali or alkaline earth metals on the atomic absorption of barium are corrected by the addition of Ca2+ ions to the standards and by the presence of sodium in the buffered precipitation medium.Moreover high concentrations of calcium enhance the atomic signal of barium owing to spectral interference of the CaOH band.13~1~ It has been pointed out that either backgroun 1422 GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Analyst Vol. 108 correction or the dinitrogen oxide - acetylene flame may be used to minimise the CaOH inter-ference.14 Bataglial has indicated that to overcome interference from aluminium and perchloric acid the addition of strontium in the first instance and matching both samples and standards for perchloric acid in the second are necessary.I t has been pointed out by Ametani22 that Fe(II) Cu(II) Cr(II1) and Cd(I1) do not significantly affect the barium atomic signal when sulphur is determined in minerals such as CuCr,S3C1 and CdCr2S,. However for samples containing titanium both unknown and standard solutions should be matched for titanium concentration.22 Column-packed alumina has been employed for the prior separation of sulphates from other ions in brines in order to avoid interferences in the atomic absorption of barium. 23 The precipitation of sulphate as PbSO in water - ethanol media has also been proposed (Table I). Sulphur may be determined at concentrations as low as 1 pg ml-l of SO,,- or 2 pg ml-l of SO when PbSO is precipitated from 4oYo2O or 25YO21 V/V ethanolic solution, respectively.The effect of different ions on the lead atomic signal in the air - acetylene flame was studied by Rose and Boltz21 with 5 pgml-l of SO present. C1- NO,- NO3-, Clog- Na K NH,+ Ca and Mg do not interfere at levels up to 500 pg rnl-l or acetate, EDTA Ba or A1 at levels of 250 20 10 or 10 pg ml-l respectively. Iron(II1) and phosphate should be absent. To overcome interference from phosphate Little et al.,O added aluminium solution so that up to 10 pg ml-l of P in the final solution can be tolerated by this procedure, The determination of sulphate in fresh water involving precipitation as BaSO in acidic medium precipitation of the excess of Ba2+ ions with chromate in ammonia medium followed by determination of the unconsumed chromium by AAS has been de~cribed.~, 1 .l.1.1. b. Precipitation as sulphide determination of sulphw. Yoshida and Takaha~hi,~ determined micro and submicro amounts of sulphides in lake and river waters by addition of Hg(I1) (3-60 pg of Hg for 0-9.5 pg of S2-) to a 100-ml sample in sulphuric acid medium, addition of tin(I1) chloride solution warming in a water-bath a t 40 "C and determination of the reduced free mercury in an absorption cell by electrothermal AAS. The method allows sulphide ion to be determined at levels down to 0.2 pg 1-1 with a relative standard deviation of less than 294. C1- B r and F- can be tolerated at levels up t o 10 mg in the determination of 0.23-0.31 pg of S2- but 1-2 pg of I- SCN- or S,O,,- can not. Interferences from cations that form stable metal sulphides have also been reported26 when sulphur is distilled as H,S and determined by measuring the signal due to excess of zinc after precipitation with Zn(I1) in ammonia medium.The recovery of zinc was about 83% from sodium sulphide standards. The method is simple and relatively free from interferences. The precipitation as PbS has also been used for the microdetermination of sulphur in organic compounds (5-10 mg of thiols sulphides disulphides and thiocarbonyls) with an average recovery of 99Y0.,7 Chlorides are precipitated with a solution of silver nitrate and the unconsumed silver is determined by AAS at the 328.1-nm resonance line. Pintal has shown that for the analysis of a solution containing between 5 and 100 pg ml-l of C1- the addition of 300 pg ml-1 of Ag(1) solution is necessary.Bromide and iodide interfere seriously. Another source of error in this indirect method is the decomposition of AgCl by ultraviolet radiation. However the results obtained compare favourably with those obtained by both titrimetric2*-32 and gravimetric33 methods. Analyti-cal applications of the precipitation as silver(1) chloride are summarised in Table 11. An interesting method for the determination of micro amounts of chlorides has been described by Tofuku and Hirand.37 The method is based on the addition of increasing known amounts of silver nitrate solution to aqueous ethanolic solutions of chlorides and the deter-mination of the supernatant silver in an air - acetylene flame at 328.1 nm. Extrapolation of the graph of absorbance vemm silver content gives the end-point at which chlorides in the samples have been precipitated by the added silver.The recovery for 2-10 pg of C1- ranged from 80 to 100~o. An analogous method of end-point determination using a short-path burner head and a mercury hollow-cathode lamp to produce radiation at the 338.3-nm secondary resonance line of Ag has also been reported.38 The method has been applied to samples with C1 contents from 1.2 to 236 pg ml-1 in distilled surface ground treated and waste waters (recovery 92-108%). 1.1.1.1 .c. Precipitation as chloride determination of chlorine December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1423 TABLE I1 DETERMINATION OF CHLORINE BY PRECIPITATION AS SILVER CHLORIDE AND MEASUREMENT BY FLAME AAS OF THE REMAINING AMOUNT OF METAL IN SOLUTION AT THE 328.1-nm RESONANCE LINE Sample NH,HC03 .. Plant material . . . . Cr chalcogenide spinels . . Plant liquors Serum Redwines Vinegar . . Poly(viny1 chloride) . . Aminesalts . . . . Cr plating solution . . NaOH Semiconductor crystals . . Separation of excess of metal Filter Filter Filter Filter Centrifuge Filter Filter Centrifuge Centrifuge Centrifuge Filter Filter Comments Reference Recovery 95.2-100.5~0 40 g of sample decomposed in boiling water. Results compare favourably with those obtained by Mohr’s titrimetric method. Detection limit 0.01% of C1- in bulk plant 28 33 17 Illdirect method more satisfactory than gravimetric method Precision is better than that achieved by titrimetric mcthods.Recovery 99.9-100.1% 29 Indirect method checked by a standard mercurimetric method indicates maximum difference of 1.08% in chloride concentration 30 Indirect method simpler more rapid and precise than the Volhard’s titri-metric method adopted by OIV (Office Internationale de la Vigne et du As above 31 Chlorine released in Sch8nigcr’s oxygen flask combustion reduced to chloride by 3076 H,O,. Recovery 98-100.4y0. Indirect method com-Indirect method compares favourable in accuracy and precision with mercurimetric and Fajan’s titrimetric procedures. Recovery 98.6-103.7% 32 Total chlorine determined after reduction of chlorate with sulphite. Recovery 99-104% 36 Acetone (28:6 V / V ) added. Relative standard deviation 0.12% for AgCl coprecipitated with l3aS0,.Relative error <10% for 0.002-0.2~0 CI content 40 Vin). Recovery 99-101.8y0 34 pares favourably with X-ray fluorescence and Schiiniger’s analyses 35 0.02% NaCl 39 1.1.1.1 .d. Other types of precipitation determination of organic compounds. Several methods for the determination of organic compounds by means of precipitation reactions have been proposed. Table I11 summarises the species determined and the characteristics of the gravimetric and spectrophotometric procedures employed. 1.1.1.2. These indirect methods involve the formation of an insoluble compound and its further separation. In most instances the precipitate is dissolved and the metal determined by AAS. An advantage of these methods with respect to the above methods is that the amount of cation added is not so critical.The pro-cedures described have been systematised into five groups according to the method of dissolu-tion employed to dissolve the precipitate. 1.1.1.2.a. Dissolution with EDTA solution determination of sulphztr. This method has been applied to the determination of sulphates by precipitation with barium ion solution. The BaSO is dissolved with EDTA solution in alkaline media and determined at the 553.8-nm resonance line of barium in an air - acetylene flame. Calibration graphs may be obtained from barium chloride standard solutions containing 1500 pg ml-l of Na to over-come ionisation effects. Under these conditions the recoveries of sulphur in soil extracts ranged between 98.7 and 100.4y0 (Arrayan Temuco and Corte Alto soils) and 79.2 and 101 .9yo (Puerto Octay soil) Wollin49 uses sulpliate standard solutions treated as the samples for the calibration graphs.Under these conditions the recovery of sulphur in sulphamide thiourea ammonium sulphate and Azure A ranged from 99.5 to 101.1 yo. The determination of total sulphur in 0rganic~~9~O and inorganic49 materials by precipita-tion as BaSO in the presence of La(II1) solution (to avoid interference from phosphate) has been reported. However addition of lanthanum has been claimed to be unnecessary as precipitation in hydrochloric acid medium will rule out any barium phosphate precipitation.** On the other hand the determination of readily soluble sulphate and total sulphur in soil5l and inorganic sulphate in urine52 has been carried out in the absence of lanthanum solution with recoveries of sulphur of 97-103 and 98% respectively.Determination of the metal precipitated However greater manipulation is required 1424 GARCIA-VARGAS et al. AAS FOR DETERMINATION OF TABLE I11 Analyst Vol. 108 DETERMINATION OF ORGANIC COMPOUNDS VIA PRECIPITATION AND MEASUREMENT BY FLAME u s OF THE EXCESS OF METAL Compound Matrix Oxalic acid . . . . Urine Phenylacetylene . . . . Water CHI, theobromine Na salicylate mercaptobenzothiazole, K ethyl xanthate organic acids . . -NN’-Diphenylguanidine (DFG) Sulphonamides . . . Tablets! suspensions, injections Non-ionic surfactants . . Water Quantitative Measured relationship metal CaC,04 Ca AgCECPh Ag Insoluble Ag compounds (DFG),Cd14 Cd Insoluble Ag compounds Adsorbed surfactant Mo on Ba molybdophosphate Comments Reference Method can be employed as a routine micro-method 41,42 43 Method could also be adapted to the deter-mination of alkylated barbiturates and mercaptans 44 Zn(SCN),B- or CdBr,,- can be also used.Method may be compared to back-titration of excess of metal with EDTA Indirect method is selective simple and precise and can favourably compare with USP XIX procedure 46 An empirical factor must be determined for each batch and multi-component system owing to non-stoicheiometric nature of 45 precipitate 47 1.1.1.2.b. Dissolution with ammonia solution determination of chlorine. The precipitation of chlorides with silver solution dissolution of the precipitate in ammonia solution and determination of the silver in the resulting solution is the basis of the method.Both air -a ~ e t y l e n e ~ ~ s ~ ~ and air - hydrogen55 flames have been employed. Calibration graphs are constructed from standard silver solutions in ammonia. Bromide and iodide interfere in the chloride determinati~n.~~ This method has been applied satisfactorily to the determination of chlorine (0.2-100 pg ml-l) in selenium (via distillation as hydrogen chloride)53 and to the determination of chloride in ZnO and PbO (C1 content 0.02-3%)54 and in titanium with prior dissolution of the sample in acidic medium. Smith and N e ~ s e n ~ ~ have indicated that this method when applied to the determination of chlorides in eleven quaternary amine chlorides gives low recoveries for C1- (93.5-98.2y0).It is therefore less convenient than those methods based on the determination of the un-reacted silver. However this effect may possibly arise from inadequate protection of the AgC1 precipitate from direct ultraviolet light. An interesting determination of trace amounts of C1- in semiconductor crystals has been developed by Gladysheva et aLgO The procedure requires the coprecipitation of AgCl and BaSO followed by dissolution of the AgCl in ammonia. Relative errors are reported to be 4 0 % for 0.002-0.2~0 of chlorine. 1.1.1.2.c. Dissolution with iodide solution determination of iodine. Iodine in organic compounds (implying previous conversion into iodide by redox reactions) has been deter-mined via dissolution of AgI with iodide solution by flame AAS.The method has good sensitivity for Ag (0.1 pg ml-1) and consequently for iodide.56 Chloride and bromide interfere seriously. 1.1.1.2.d. Dissolution with mineral acids determinations of sulphur and ammonia. Ray et aLZ6 proposed another indirect method for the determination of sulphides in flooded acid sulphate soils (0.29 mg of S2- per gram of sample) by means of ZnS precipitation in ammoniacal medium and dissolution of the precipitate in hydrochloric acid. The method is practically free from interferences as sulphide is distilled as hydrogen sulphide. The precipitation with silver ions has been successfully applied to the analysis of rnercapfan~.~~,~* The silver precipitate is filtered off and digested with nitric acid.A single calibration graph prepared from silver nitrate is utilised. Pure thiol standards are not required. The method is relatively free from interferences but substances that reduce silver ions to metal or form insoluble silver salts may interfere December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1425 Danchick et aZ.59 have shown that the coprecipitation of ammonium molybdophosphate with a known amount of thallium molybdophosphate followed by dissolution of the precipi-tate provides the basis for an indirect atomic-absorption method for the determination of ammonia. The dissolved molybdate is determined at 313.2 nm in an air - acetylene flame. The method shows poor selectivity heavy metals being precipitated as molybdophosphates.SCN- sol. S032-detn. CN- detn. v v V 1.1.1.2.e. Direct atomisation of the precipitate determination of sulphate. Siemer et a,?.GQ described the determination of sulphates in natural waters (32-2640 pg ml-l of SO,%) by precipitation with lead(I1) nitrate in 50-75% V/V ethanol - water using a modified Woodriff furnace for direct lead determination at the 405.7-nm non-resonance line. They showed that barium cannot be satisfactorily employed even at temperatures as high as 2400 "C. Large amounts of calcium and chloride do not interfere but phosphate chromate iodide and arsenate cause serious interferences. 1.1.2. Metal Complexes In these methods the analyte reacts with either one or two reagents giving rise to the formation of a charged or neutral metal complex or an ion-association complex.The metal is then determined by AAS after filtration or liquid - liquid solvent extraction. The ion-association or metal complex formed should be very stable. Generally calibration graphs are constructed from standard anion solutions prepared in the same way as the unknown samples. Common interferences arise from other anionic species that might also be complexed. An obvious interference is due to the presence of the metal used for the indirect determination in the anion-containing solution. In these instances the atomic absorption of the metal in the unreacted sample has to be determined. - SCN-complex (org) Anion (aq) I I Cd Neutral metal complex (org) Fe(lll) Ion-pair association (aq) cu Fig.2. General outline of procedures for the indirect determination of inorganic anions via the formation of metal complexes 1426 GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Analyst Vol. 108 These methods are classified into two groups those based on the formation of a metal complex with an analyte as ligand and those based on the formation of an ion-association complex with an analyte as counter ion. An outline of procedures for the analysis of in-organic anions is given in Fig. 2. 1.1.2.1. Analyte as ligand These methods make use of stoicheiometrically formed complexes of metals that can be isolated determined by AAS and quantitatively related to the complexed anion. Inter-ferences arise from the presence of metals that can form more stable complexes with the anion to be analysed or from species forming more stable complexes with the cation employed for the atomic-absorption measurements.In both instances separations are required. The reported procedures have been organised according to the charge of the compounds formed. 1.1.2.1 .a Charged metal complexes determinatiolzs of cyanide sulphite pyrophosphate, fluoride and organic compowzds. They are based on the reaction of an anion-containing solution with a metal (in solution or in the solid state), a metal oxide or a metal salt in their solid states. The complexed metal is then normally determined in aqueous solution by either electrotherma161 or flame62-78 AAS. Cyanides have been determined indirectly by measuring the atomic signal of Age1@ Cu63#6* or Nis5 in the complexes Ag(CN), Cu(CN) 3- or Ni(CN) 42- respectively (Table IV).The most sensitive procedure has been described by Jungreis and Ain.61 The method uses silver wool placed on a membrane filter in a closed system. The sample solution (0-100 pg 1-1 of CN-) is injected from a 35-ml syringe. Silver is determined in the effluent by injecting a 100-p1 volume of the sample into the graphite furnace. C1- Br- I- F- SCN- SO," and NO3- do not interfere but S,032- does. These procedures are rapid and simple. TABLE IV SENSITIVITIES OF INDIRECT AAS DETERMINATION OF ANIONS BY FORMATION OF CHARGED METAL COMPLEXES Anion Sample CN- . . Water CN- . . Water CN- . . Water CN- . . Sewage effluents CN- . . . . Biological materials S 0 P - . . . . Water P,o,~- . . Water F- .. . . Organic compounds Substance used for Sensitivitylpg ml-1 complex formation anion per 1 yo absorption Ag wool* 0-0.lt Cu (I) cyanide 1-1ot NiC1 solution$ 1 HgO suspension 139 Ag pieces 0.5 Basic Cu carbonate 0.5 Copper sulphate solution7 0.87-8.7 t Fe(II1) - SCN-11 0.5-67 Reference 61 62 63 64 65 66 79 80 * HGA 2100 graphite furnace was used. 3 Excess of Ni(I1) is eliminated by precipitation with Na sulphide. 5 Detection limit defined as the concentration (in micrograms per millilitre) that produces an absorption 'I[ Copper(I1) ions are extracted by a 10% ( V / V ) liquid ion exchanger into IBMK or ethyl acetate. 11 Excess of Fe(II1) is extracted as thiocyanate into IBMK. Range of determination in micrograms per millilitre.signal equivalent to twice the standard deviation in the noise fluctuation of the background signal. Jungreis and AnavP proposed a method for the determination of sulphite that appears to be useful in air-pollution analysis. The procedure is based on the reaction of sulphite (119-833 pg) with red HgO (about 120 mg) at pH 11 to yield Hg(SO,),,-. The atomic signal of the complexed Hg is measured at 253.7 nm in an air - acetylene flame. The method is affected by I- S203,- and SCN- but SO," NO,- and CN- do not interfere. A very sensitive method for the determination of (or SO,) based on the disproportionation reaction Hg,,+ + 2SOS2- -+ Hg + Hg(SO,)," has been developed by Marshall and Midgley.8l The procedure monitors the Hg released at 40 "C in the above reaction using the cold vapour atomic-absorption technique.Thi December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1427 method can determine sulphite at levels up to 5 ng. The selectivity is good as some anions may be tolerated at high foreign ion to sulphite ratio COS2- 10000 1; SO,2- and NO,-, 1000 1 ; F- and HP0,- 100 1 ; C1- 10 1 ; and S2- (5 1) ; NH,+ did not interfere at a ratio of 3000:l. West and Lorica82 showed that iodide could be extracted as a cadmium complex with tributyl phosphate in isobutyl methyl ketone (IBMK) and utilised in AAS. The only atomic-absorption spectrophotometric method reported for the determination of pyrophosphate ions in the presence of other condensed phosphates is based on its inhibitory effect on the extraction of copper(I1) by Amberlite LA-1 [N-dodecyl(trialkylmethyl)amine] into isobutyl methyl ketone or ethyl acetate.79 The extracted copper is monitored at the 324.7-nm line.The method is simple rapid and selective. Tetrameta- and trimeta-phosphates phosphates sulphates chlorides and acetates do not interfere. Some species may be determined by their masking action in the solvent extraction of a metal complex. This effect has been used by Kidani and ItoSo for the determination of fluoride by means of extracting the Fe(II1) - SCN- complex into isobutyl methyl ketone. The decrease in the Fe atomic absorption in the organic layer is quantitatively related to the concentration of fluoride (0.5-6 pg ml-l of F-). Interferences occur owing to metals forming fluoride complexes more stable than that of Fe(III) but anions other than cyanide do not interfere.Also EDTA may be determined by its masking effect on the extraction of a Cu(I1) - quinolin-8-01 complex into isobutyl methyl ketone at pH 6.5.‘j In these methods it is evident that the test substance must form a more stable compound with the metal than the extracting agent. These methods might be also applied to the determination of organic compounds provided that they act as ligands. A summary of indirect procedures for the determination of organic TABLE V DETERMINATION OF ORGANIC COMPOUNDS BY FORMATION OF METAL CHARGED COMPLEXES AND MEASUREMENT BY FLAME AAS OF THE COMPLEXED METAL Measured Compound Sample metal Quantitative relationship Comments Reference Histidine . . . . . . Effluent of liquid - liquid EDTA .. Streptomycin chromatography EDTA NTA . . . . Effluent of liquid - liquid &-Amino nitrogen . . . . Plasma chromatography EDTA NTA . . . . Effluent of high-speed liquid chromatography EDTA . . . . Water and waste water Phenols (21 compounds) . . Water Cu Cu complex 1 pg of histidine can be Ni detected 67 Ni - EDTA and release of Detection limit near 4 pg of complexed Ni by pH EDTA per g of compound. adjustment Phosphates do not interfere 68 cu Cu complexes Detection limit 8 x lo-’ mmol of EDTA or ETA 69 Cu Solubilisation of copper Plasma cleprotenised with phosphate and filtration Cl,C-COOH filtered and neutralised with alkali TO Cu Cu complexes Detection limits are 10.7 and 23.6 rig of Cu for KTA and EDI’A respectively 71 Cu Solubilisation of copper Lower limit of analysis for at pH 10 and filtration EDTA is 1 x 1 0 F M 72 Co Co complex extraction Recovery 91.5-104.50/,.into IBMK Cu(II) I’e(II1) and PhKH, interfere 73 Biuret . . . . . . Fertilisers and urea Cu In strongly alkaline Although a stoicheiomptric relationship does not exist, a linear relation between Cu and biuret may be obtained 74 75 solution Cu forms a biuret complex; Cu(OH), is filtered off Amino acids - - Cu Schiff bases with Range of determination: salicylaldehyde are 1.8-15.0 kg ml-l of glycine 76 formed complexed with Cu and Schiff base - Cu -bathophenanthroline complex is extracted itlto IBMK Noscapine Drugs Alcohols -Cr Add Reinecke’s salt and -extract into chloroform 77 Cr Add CrI and pyridine to Methanol added to benzene alcohol in benzene solution before nebulisation 78 solutio 1428 GARCIA-VARGAS et al.AAS FOR DETERMINATION OF Analyst Vol. 108 compounds by formation of charged complexes is given in Table V. Of the procedures described some use a selective reaction,67-69,71,72,74,75,77 while others use a general reaction that may be applied to the determination of a range of compounds.70,73,76,78 1.1.2.1 .b. Neutral metal complexes determinations of cyanide thiocyanate sulphide halides and organic compounds. These methods involve the formation of an uncharged metal com-plex which is frequently separated by solvent extraction and the complexed metal in the organic solution is then determined by AAS. These procedures have a greater sensitivity than the procedures described previously.However the sensitivity can also be enhanced by the choice of a favourable Vaq Vorg ratio. Danchick and Boltzs3 proposed an accurate and highly selective method for the indirect determination of cyanides (sensitivity 0.06 pg ml-l of CN-). The method is based on the extraction into chloroform of the dicyanobis( 1,lO-phenanthroline) iron (11) complex, evaporation of the organic solvent re-dissolution in ethanol and atomic absorption of the associated iron at the 248.3-nm resonance line. The formation of a highly insoluble silver cyanide precipitate is the basis of a second method for the determination of cyanides, described by the same workers.83 The silver cyanide is quantitatively precipitated and the silver excess in solution is determined at 328.1 nm after centrifugation.This method has a higher simplicity and sensitivity (0.03 pg ml-l of CN-) than the method based on the ferroin reagent. A selective method for determining thiocyanate has also been described by Danchick and B01tz.~~ They proposed two procedures based on the quantitative formation of the dithio-cyanatodipyridine - copper( 11) complex and its selective extraction into chloroform. In one procedure the organic layer is sprayed into an air - acetylene flame and the Cu atomic signal at 324.7 nm is measured (sensitivity 0.2 pg ml-l of SCN-). Toxic combustion products may be formed. In the other method the organic phase is evaporated almost to dryness diluted with ethyl acetate and the Cu atomic absorption measured (324.7 nm; air - acetylene).In this procedure the sensitivity is greater (0.05 pg ml-l of SCN-) and the method seems less hazardous. There are few interferences in the determination of SCN-. Christian and Feldman6 used the masking effect of sulphide (up to 3 x mmol) on the extraction of 3.1 x mmol of copper as a copper - quinolin-8-01 complex to propose an indirect procedure for determining S2- based on the direct atomisation of the organic phase. Belcher et aLa5 determined chlorides at very low levels by extracting phenylmercury (11) chloride into chloroform evaporation and re-dissolution in ethyl acetate. The Hg concentra-tion is measured at 253.65 nm using an air - acetylene flame. As little as 0.015 pg ml-l of C1- can be determined in 250 ml of sample.Alternatively but less convenient the phenyl-mercury(I1) chloride may be extracted into isopropyl acetate and this solution is sprayed into the flame. Both methods have some advantages with respect to other AAS procedures for determining chloride they are more sensitivive and less affected by the presence of other ions. However bromide and iodide are listed as interferents as they react quantitatively with the reagent to form the corresponding phenylmercury(I1) halides which are then extracted together with the chloride complex. This of course offers the possibility of a total halide determination followed by stepwise elimination of iodide and bromide by known chemical reactions to give the individual halide content of a sample. Chuchalina et a1.86 developed a method for the indirect determination of trace amounts of C1- Br- or I- based on the addition of an excess of Hg(I1) ions to the halide solution percolation through a column packed with a KU-2x8 cation exchanger and atomic absorption of the Hg in the effluent.The detection limits are 2.1 x 10-10 4.8 x 10-10 and 7.6 x 10-log of C1- B r and I- respectively. The method has also been applied with excellent results to the determination of many organic compounds (Table VI). The metal associated with the organic compound is frequently determined by AAS after filtration88-90,92,94,96 or solvent e x t r a c t i ~ n ~ ~ ~ ~ ,93,97,98,10(t-102 and a relationship is obtained between the metal absorption and the amount of organic substance. 1.1.2.2. Ion-association complexes The basis of this method is the enhancement of the solvent extraction of a charged metal complex by formation of an ion pair with the analyte.The quantitative reaction is normall December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1429 TABLE VI DETERMINATION OF ORGANIC COMPOUNDS BY FORMATION OF NEUTRAL METAL COMPLEXES AND MEASUREMENT BY FLAME u s OF THE COMPLEXED METAL Measured Compound Matrix metal Quantitative relationship Characteristics of method Reference Thiol groups . . . . Proteins Protein - S - Hg p - 1-5 pg of Hg can be Hg hydroxybenzoate determined in a 100-cm complex is formed. absorption cell. Recovery Excess of unreacted 96.0-100.9~0. Detection reagent is separated by limit 0.1 pg ml-1 of Hg. gel filtration Mg Ca Fe(II) Mn(II) Ni, (Sephadex G-25) Cu Zn and Cd do not interfere a7 Esters .. . . . . -Acidanhydrides -Biacetyl . . -Ketones and aldehydes of low relative molecular mass -Primary amines . . Water Secondary amines . . Water Aliphatic secondary amines -Histidine methionine CS Air CS Air CSs Air CS CC14 C,H, Anthranilic acid . . . Water Fe Fe Ni Hydroxamic acid formed is complexed with Fe(II1) was 92.4-105% 88 and filtered off. The Fe(II1) complex determined in filtrate Hydroxamic acid formed Recovery 97.5-102.8%. is complexed with Fe(II1) Acetone HOAc PhNH,, and filtered off. PhNO, PhOH K+ Mgl+, Iron(II1) complex Can+ EtOAc do not determined in filtrate interfere. Cu(I1) and AI(II1) interfere 89 Recovery for nine esters Ni - dimethylglyoxime filtered and dissolved in 5-200 pg ml-1 of biacetyl complex formed is can be determined 90 HNOI Cu Cu - thiosemicarbazone Recovery 96.7-101.1%.complex is formed, extracted into benzene 10-fold excess interfere 91 and diluted with ethanol Co(II) Ni Ag and Zn in a cu Schiff's bases are formed 1 mg of amine can be complexed with Cu determined. Secondary filtered off and excess and tertiary amines in a of Cu determined in 30-fold mass excess do not filtrate or precipitate interfere but Zn Mg and dissohed in nitric acid Ca do 92 and analysed for Cu Cu Cu - dithiocarbamate is 1.1-6.6 pg ml-1 of formed and extracted Et,NH.HCl can be into IBMK determined. Recovery 93.8-103.2 yo. Lead( 11). Fe(III) Mg(II) MeNH,.HCI, Me,N.HCI Ph,NH.HCl and aniline sulphate do not interfere 93 0.6-8.5 pg of Ni per mole Ni Ni - dithiocarbamate is formed filtered and of amine can be determined.dissolvkd in HNO - HCI Recovery 98.7-100.7% 94 Cu Schiff's bases formed are 3-30 pg ml-1 of compound Cu complexed with copper can be determined 95 CS trapped in KOH and Sensitivity 0.009 g m-* of H$ removed with Cd. CS 96 Copper - xanthate is formed filtered and dissolved formed and extracted CS per 10 ml of solution 97 into isoamyl acetate formed and extracted determined. Sensitivity Cu Cu - dithiocarbamate Detection limit 7 pg of Cu Cu - dithiocarbamate 5-50 pg of CS can be into toluene 0.29 p,g ml-I of CS? 98 Cu Cu - dithiocarbamate Determination limit solution evaporated to 2.52 x mg of CS, dryness and dissolved in HNO 99 line formed and extracted recovery Co Co(I1) - anthranilic 3-22 pg ml-l of compound acid - bathophenanthro-into IBMK determined with 99.5% 100 Na diethyldithiocarbamate (DDC) .. . . . . Human serum and urine Cu Cu chelate formed Down to 5 pg 1-l of DDC and extracted into CCI can be determined 10 1430 GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Analyst VOZ 108 Compound Matrix Ammonium tetramethylene-dithiocarbamate (APDC) . . W'ater TABLE VI-continued Measured metal Quantitative relationship Cu or Co Cu (or Co) chelate formed and extracted into IBMK Quinolin-8-01 . . . . -Non-ionic surfactants . . Sea water Cu Cu - quinolin-8-01 formed andextractedinto IBMK or ethyl acetate surfactant adduct is formed and extracted into benzene Co Co - thiocyanate -Free fatty acids .. . . Blood serum Co Cobalt complex formed Soap Vegetable oil Na Na oleate formed Characteristics of the method Reference Concentration of M APDC can be determined 6 Concentration of 2 x M quinolin-8-01 can be determined 6 Procedure better than that given in Ref. 47 as no empirical factor is required. Precision better than 10% 102 lii-585 pmol I-* can be determined. Method less tedious than gas chromato-WPhY 103 104 -related to the formation of a metal - organic reagent positive complex with an anion. The atomic absorption of the complexed metal is usually measured in an air - acetylene flame and related quantitatively to the associated analyte.The selectivity of these methods depends essentially on both the extracted complex compound and the conditions under which extraction is carried out. They are divided into inorganic anions and organic compounds, according to the chemical nature of the analyte. I?%organic anions determinations of perchlorate nitrate and iodide. Two organic reagents have been employed for the formation of ion-association pairs 1,lO-phenanthroline (phen) and 2,9-dimethyl-l 10-phenanthroline (neocuproine) . The only method for the atomic-absorption determination of C104- has been described by Collinson and Boltz.lo5 The method is based on the extraction of [(neocuproine),Cu(I)]C104 into ethyl acetate and measurement of Cu atomic signal at 324.7 nm in an air - acetylene flame.Between 0.5 and 5 pg ml-l of Clog- can be determined; 15 pg ml-l of Ac- C1-, SOg2-, A somewhat analogous method for determining nitrates (0.4-4 pg ml-l) has been pro-posed,lo6 but it includes extracting the ion pair into isobutyl methyl ketone.lO6 The method is relatively free from interferences as indicated by several workers.lO6-108 This indirect method has been applied to the determination of nitrates nitrites (by oxidation with cerium sulphate) and nitro-groups (by oxidation with permanganate) in inorganic and organic compoundslo8 and nitrates in plants.1°9 Iodide (0.5-5 pg ml-1) can be extracted as (phen),Cd'II into nitrobenzene and the atomic absorption of the chelated cadmium is measured at 228.8 nm in an air - acetylene flame.ll0~ll1 S042- CN- C1- Br03- F- and B02- do not interfere but 104- Clop, C103- NO3- NO2- and Br- do.Yamamoto et al.l12 have described a less sensitive analogous method for iodide based on the (phen),Fe(II) chelate. The calibration graph is linear from 10.1 to 40.4 pg ml-l of I-. The same foreign ions interfere in both procedures. 1.1.2.2.b. Organic compounds. Indirect atomic absorption methods have been developed for the determination of organic compounds via the formation and extraction of ion pairs. A special characteristic of these procedures is that they may be applied to organic compounds having a negative113-l18 or p0sitivell9-l~~ charge. A summary of the analytical applications and characteristics of the procedures is shown in Table VII. 1.1.3. These methods are based on widely known reactions of some anions with molybdate and other ions in solution yielding heteropolyacids of definite chemical composition which are usually extracted into an organic solvent.The metal is normally determined by AAS either in the organic layer or in aqueous solution after suitable back-extraction. This method shows good sensitivity because according to the heteropoly-compound formed 12,127-149 11150 or 10151 Mo atoms are associated to every P atom. However Parsons et aZ.15 claim that 15 atoms of Mo are extracted for each atom of P although no evidence for the anomaly is given. 1.1.2.2.a. K and Fe(I1) may be tolerated in the determination of 3 pg ml-l of C104-. Heteropolyacid Compounds Determination of Phosphate A general scheme of methods for determining phosphates is given in Fig.3 December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW TABLE VII 1431 CHARACTERISTICS OF PROCEDURES FOR DETERMINATION OF ORGANIC COMPOUNDS BY FORMATION OF ION-ASSOCIATION COMPLEXES AND MEASUREMENT BY FLAME AAS OF THE COMPLEXED METAL Measured Pentachlorophenol (PPh). . Water Fe(248.3) Fe(I1) - (phen) - (PPh) Up to 3 x M PPh Compound Sample metal (nm) Quantitative relationship Comments Reference formed and extracted has been studied 113 into nitrobenzene Salicylic acid (SA) . . Water Fe(248.3) Fe(I1) - (phen) - (SA), formed and extracted into nitrobenzene 114 2-Hydroxynaph thoic acid (HNA) . . Water Ni(232) Ni(I1) - (phen) - (HNA) From 8 x 1 0 F to 4 x M formed and extracted into nitrobenzene Naphth-2-01 does not HNA can be determined.interfere 115 Phthalic acid (PhA) . . Water Benzylpenicillium (BP) . . Catechol (C) . . . . Water Cu(324.7) Cu(1) - (neocuproine) -(PhA) formed and extracted into IBMK From4 x 10-e to4 x lo-' M PhA can be determined. Both isomers of PhA do not interfere a t the same concentration level of PhA 116 117 Cd(22R.R) Cd(I1) - (phen) - 18.6-111.6 Vg ml-' of BP (BP) formed and extracted Starch lactose dextrin can be determined. and Na saccharin do not interfere 118 Cu(324.7) CuC is formed and 11.0-176.2 yg ml-' of extracted with catechol can be chloride into CHCI a t pH 10 trioctylmethylammonium determined 119 Anionic detergents (AD) . . Fresh water Cu(324.7) Cu(1) - (phen) - (AD) is Na dioctylsulphosuccinate formed and extracted into IBMK can be determined in the range 3 yg I-' to 2.5 yg ml-'.Na stearate Mn0,- Co, Ni Fe(III) EtOH C10,-, 104- SCN- Cr,O,a- and dithiocarbamates a t about 5 yg I-' do not interfere. Precision is 5% for 25 yg of Na dioctylsulphosuccina tc per litre 120,121 Linear alkylhenzcne sulphonate (LAS) . . . . Watcr Quaternary amines . . -Aliphatic amines (AA) . . -cu(:bi2.7) Cu(T) - (thiourea) - Calibration graph is linear in the range 0.001-20 pg ml-I of LAS. Coefficient of variation is 2.4% for 0.1 yg ml-' of LAS 122 (LAS) is formed Cu(342.7) Amine is complcxed with Method requires a 7-4 VM solution of the ammonium excess of (DDS) ; unreacted DDS is cornplexed with Cu(1) -(phen) and extracted into IBMK dioctylsulphosuccinate complex 123 Co(240.7) Co(I1) - (SCN) - (AA), is forrried extracted into can be determined 124 lienzenc and diluted with methanol 50-650 ~ r g of AA pvr 5 ml n-Bu tylscopolammonium bromide and tannate .IjkJOd uriric.;ind fac.cc.s Co(240.7) Co(I1) - (SCN) - (aminr) 0.8-1.5 nig o f conipoi~iid is formed and extracted can be deterrriiucd 125 into CHCI,. Extract is evaporated and dissolved in IBMK H yoscine-N-hu tyl hromide (H) . . . . T'liarmaccritic;il Co(240.7) Co(1I) - (SCN) - (H) is I.inc.sr rd;itioiiship olitairicd prcpara tions formed and extracted into 1,2-dichloroethane to -7 x lo-' M. No i l l the range of 5 j c iritcrfererices from cxcipien t s 12 1432 H5PSb2M010040 H4PVMo11040 (aq) GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Analyst Vol.108 6CgH70N.P205 H3PMo12040 (aq) Z4MOO3.11 H2O ( ppted. 1 IAbout pH 1 P detn. + P detn. Back-extraction with ammonia sol. 1 1 Ammonia solution MoWI) (aq) Mo Mo(VI) (aq) 1.1.3.1. Binary heteropolyacids Phosphate has been widely determined using the formation of molybdophosphoric acid (H3Phh)12040) when a molybdate solution is added to a phosphate-containing solution in a strong acid medium. After several minutes the molybdophosphoric acid is usually extracted into an organic solvent which is then washed repeatedly with citrate ~ o 1 ~ t i o n ~ ~ ~ - ~ ~ ~ with dilute a ~ i d ~ ~ ~ y ~ ~ ~ ? ~ ~ ~ - ~ ~ ~ or with water131 to remove the excess of molybdate transferred to the organic phase. The phosphorus content is related quantitatively to the atomic-absorption measurements of Mo at 313.2 nm by either flame or electrothermal techniques.Usually the specificity of these methods is greatly dependent on both the extraction conditions and the extractant employed. Extensive use of molybdophosphoric acid analytically has been made (Table VIII). These procedures are classified into methods based on single extractions extraction -decomposition and precipitation of the heteropolyacid according to the separation procedure used for the isolation of the molybdenum associated with the phosphorus. These methods involve the determination of Mo by AAS in the organic phase. Three different types of organic solvents have been used a l ~ o h o l s ~ ~ ~ - ~ ketones130,132,134i1j2 and ester~.~33~13~-141 In general oxygenated solvents are the best extractants for heteropolyacids.Calibration graphs are usually constructed from phosphate standard solutions prepared in the same way as samples. Zaugg and Knox127-129 have examined the extraction of molybdophosphoric acid in several alcohols butan-1-01 isobutanol hexan-2-01 heptan-1-ol and octan-2-01. They found the last to be the best because it extracts less free molybdate. The method is sensitive using an air -acetylene flame (detection limit 0.015 pg ml-l of Y) but it is subject to interferences mainly from As and Si. 1.1.3.l.a. Single extraction December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW TABLE VIII 1433 DETERMINATION OF PHOSPHORUS BY EXTRACTION OF MOLYBDOPHOSPHORIC ACID FOLLOWED BY ATOMIC ABSORPTION OF MOLYBDENUM AT 313.2 nm Range of determination/ vg ml-1 of analyte 5-200 (P) Atomisation technique Air - C2Ha Air - C,H, Air - C2Ha Air - C,H, Air - C2Hz Air - C2H, Air - C2HB NZO - CZHz N,O - C,HZ N2O - CaH, N2O - CZH, NSO - CZH, HGA 70 graphite furnace Air - C,H, Air - C,H, N,O - C,H, Electro-thermal Sample Optimum pH Solvent used* Comments Reference IBMK Octan-2-01 IBMK IBMK 390.3-nm line is also used.Standard addition method (SAM) is employed for Mo measurement Plasma . . 1.7-2.1 152 127,128 130 131 132 133 134 135 136 137 138 140 141 142 143 144 145 148 Biological materials; fresh and sea water . . . . 1.0-1.1 Up to 6 (PO,'-) 100-200 (P)? Other lines used were 379.8,386.4 and 390.3 nm.Sensitivity is 0.04 Wg ml-1 of PO4*- at 379.8 nm Milk and blood serum . . . . 1.25 ml HClO, Hystamine diphosphate, pyridoxal phosphate, triphenyl-phosphorus . . 5 ml HClO, Recovery 97.9-106.0%. SAM is used Up to 2.5 (P) Sample digested with H,S04 - HNO, prior to determination Bottom sediments . . Acid medium Extract is cooled and dissolved in IBMK Benzophenone Urine 1 M HCl IBAc 0.5-2.5 (P) Minimum determinable concentration is 0.1 vg ml-1 of P Excess of Mo is determined HF is added to avoid interferences from Nb Ta Ti V W and Zr Wines 1-2 Fe steel . . 6 N HNO, IBMK 10-70 (Mo) IBAc 0.05-1.0 (P) Synthetic mixtures 0.96 M HCI IBAc 0.083-1.0 (P) Recovery 89.1-102%. Method is adequate for sequential determination of P and Si Synthetic mixtures .. 0.7 IBAc u p to 1.0 (P) Sensitivity 0.01 pg ml-l. Recovery 85.7-110y0. Method is adequate for sequential determination of P and Si Precision is 3.56 or 1.55% for 0.011 or 0.052% P respectively Fe steel . . Acid medium BAc Up to 0.35 (PO,a-) BAc 0.01-1.0 (P) A1 alloy . . 1 Sensitivity 0.02 pg ml-I. 83.3-133.6 Yo Recovery Silicon structures . . 0.7 N HNO EtAc 0.006 (P)$ Si is previously removed with HF. Results obtained agree with those obtained by neutron-activation analysis Pure water . . 0.02 ml 7 N H2SO' BAc 1-4 p.p.b. (Mo) 10 pg of P in 0.4-ml samples of ultra-pure water determined with errors below f 30%. Back-extracted with 4 N NH, Steel, cast iron 5ml HNOs (1 + 2) IBAc 0.01-0.1 (P) Organic layer is evaporated and molybdophosphate dissolved in dilute nitric acid.Recovery 88-111y0 Synthetic mixtures . . 2 M HCI DE 0.1-1.2 (P) Recovery 92-110%. Back-extracted with ammonia buffer 10 separate determinations on same sample gave results ranging from 0.11 to 0.49% P20,. Back-extracted with ammonia buffer Rock 1 DE - pentan-1-01 1-10 (Mo) (5:l) Blood and bovine serum . . 0.67 M HCI DE - pentan-2-01 (5 1) Back-extracted with ammonia buffer. Procedure has a relative standard deviation greater than spectrophoto-metric method * IBMK isobutyl methyl ketone; IBAc isobutyl acetate; BAc butyl acetate; EtAc ethyl acetate; DE diethyl ether. t Given in micrograms of P. % Sensitivity in micrograms per millilitre of phosphorus 1434 A.nalyst Vol 108 Isobutyl methyl ketone has been used as an extracting agent,130-132J52 mainly for high phosphate concentrations.Few data on interference effects have been reported. The standard additions method has been employed to overcome matrix effects.130J52 The most suitable methods for the indirect determination of phosphate make use of the extraction of molybdophosphoric acid into esters such as isobutyl a~etatel~~*l~~-137 and but yl acetate. l38-l40 Kirkbright et ~ 1 . l ~ ~ have worked out a rapid and simple method for the indirect sequential determination of P and Si by the AAS of Mo heteropolyacids. The determination of P (sensitivity 0.007 pg ml-l of P) is almost specific because of the selectivity of isobutyl acetate in the extraction of molybdophosphoric acid.The presence of a 100-fold mass excess of the following ions produces no interferences on 10 pg of P Al Au Bi Ca Cd Co(II) Cr(III), Fe(III) Ni Pb Mg Mn Se(IV) Te(IV) Ti(IV) Zn F- EDTA NO3- As(V) Ge(IV), Sb(V) and V(V). A 10-fold mass excess of W(V1) causes no interference in the P determina-tion. Another important scheme has been devised for the sequential determination of P As and Si by using the heteropoly chemistry of Mo.l3' Molybdophosphoric acid is also extracted into isobutyl acetate (sensitivity 0.01 pg ml-1 of P). The method is rapid simple accurate and practically free from interferences from several foreign ions. With the exception of Ti and Zr, which lead to a low recovery of P none of the other 40 ions tested interfered at a foreign ion to P ratio of 80.The non-interfering effect of W(V1) and the interfering effect of Ti(1V) on the P determination is in disagreement with the observations made by Kirkbright et ~ 1 . l ~ ~ This difference could probably be explained as being related to the different experimental con-ditions under which the molybdophosphoric acid is extracted. Butyl acetate has proved also to be an excellent extractant for molybdophosphoric acid in the determination of phosphate.138-141 Bernal et al.140 compared butyl acetate with other organic solvents such as isobutyl methyl ketone acetophenone diethyl ether butan-1-01, amyl alcohol and isoamyl alcohol. They found butyl acetate to be the best because of its specific selectivity and as it gives virtually a 1 0 0 ~ o extraction in a single step.Twenty-four ionic species have been shown not to interfere up to a foreign ion to P ratio of 10000 1. Si(IV) W(V1) and Ge(1V) do not interfere up to a ratio of 8000 1 and Ge(1V) up to 6000 1. GARCIA-VARGAS et al. AAS FOR DETERMINATION OF No appreciable amounts of free molybdate are extracted into isobutyl acetate. 1.1.3.1 .b. Extraction - decomposition. The principle of these methods involves the extraction of the molybdophosphoric acid into esters,142,143 ethers144 or into mixtures of ethers and followed by its re-extraction with ammonia ~ o 1 ~ t i o n ~ ~ ~ ~ ~ ~ ~ or alternatively, by evaporation of the organic layer and dissolution of the solid residue in nitric acid.143 The Mo liberated into the aqueous solution is then determined by flame143-145 or e l e ~ t r o t h e r r n a l ~ ~ ~ ~ AAS.An advantage of these decomposition procedures is that the standard solutions needed for the construction of calibration graphs may be prepared directly from aqueous molybdate solutions. Furthermore the atomic-absorption determinations of organic solutions are subject to a number of drawbacks In spite of the fact that the signal intensity may be en-hanced its stability is usually poor and the results become irreprodu~ib1e.l~~ ,lg5 A very sensitive back-extraction method used for the sequential indirect determination of P and As in pure water has been described by Ro~enb1um.l~~ The method relies on the use of butyl acetate as the extractant for the molybdophosphoric acid and the determination of Mo in ammonia solution is then carried out in a graphite furnace.As low as 25 pg ml-l of P may be determined. Other schemes have also been devised for the sequential determination of P and Si1449145 and have been applied to rock analysis.145 This last procedure created a controversy in relation to its unsuitability for the analysis of Si in r0cks.1~69~~~ 1.1.3.1 .c. Precipitation of molybdoPhosphoric acid. Melton et ~ 1 . l ~ ~ have used the gravi-metric quinolinium molybdophosphate method of phosphate precipitation and dissolution in ammonia solution for determining P in fertilisers (5.5148.07% P,05 content). The deter-mination of Mo is carried out in an air - acetylene flame at its 320.9-mm secondary resonance line. The method for the determination of macro-amounts of Pis rapid and simple and its accuracy compares favourably with the AOAC method.149 December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS.A REVIEW 1435 1.1.3.2. Ternary heteropolyacids These methods are based on the formation of molybdovanadylphosphoric150 or molyb-doantimonylph~sphoric~~~ acids followed by their extraction into isobutyl methyl ketone or butyl acetate respectively. In the former the determination of the V absorbance at the 318.5-nm line in a dinitrogen oxide - acetylene flame is related quantitatively to the P content. The detection limit is 0.0005% of P. In the latter the atomic absorption of Mo at 313.3 nm or of Sb at 217.6 nm is measured in a graphite furnace. The working graph for P is found to be linear over the range 0.01-0.15 pg (Mo measurements) or 0.1-2 pg (Sb measurements).The molybdovanadylphosphoric acid method150 has been applied satisfactorily to the determina-tion of P in steel and iron. Elements usually present in these samples do not interfere. The molybdoantimonylphosphoric acid method151 has been applied to the determination of trace amounts of P in sulphur and in phosphosilicate glass with satisfactory results. Interferences from Bi Ca Co Fe(III) Mn(II) Pb Se(IV) Si Te(1V) and Zn at a 100-fold mass excess are almost negligible but As(III) Ge and Cr(V1) interfere seriously. 1.1.4. Redox Reactions Oxidising or reducing substances may sometimes be determined by reaction with either one or two reagents (one of which is a metal ion) that can be oxidised or reduced by the substances involved.A separation procedure is then used to isolate either the reduced or the oxidised form of the added metal ion. In general these methods are not very selective because com-plete absence of other oxidising or reducing agents is required. They are divided into inor-ganic anions and organic compounds according to the chemical nature of the analyte. 1.1.4.1. Inorganic anions determinations of iodide iodate nitrate and chlorate mequiv.) have been determined by their reaction with Cr(V1) in an acidic medium and then extraction of the unreacted Cr(V1) into isobutyl methyl ketone from a 3 M HC1 medium. The increase in the atomic signal of the Cr(II1) in aqueous solution or the decrease in the atomic absorption of the Cr(V1) in the organic phase may be quanti-tatively related to the iodide concentration.6 The reduction of Se(1V) to elemental Se by iodide (up to 10-3mmol) in an acidic medium followed by AAS of Se has been also pro-p 0 ~ e d .~ ~ ~ s ~ 5 ~ Iodate (up to 3 x mequiv.) has been determined by oxidation of Fe(I1) to Fe(II1) and extraction of the latter into diethyl ether from 9 M HCl solution followed by flame AAS of Fe in the organic layer.6 The presence of air in the aqueous solution is critical for an accurate method and it is necessary to deaerate all solutions with nitrogen to prevent air oxidation of the Fe(I1). Hassanl55 has proposed an indirect method for determining nitrates (0.2-2 mg of nitrate-nitrogen) by reduction with Cd metal in a 0.05-0.1 M HCl medium. The dissolved cadmium is aspirated into an air - acetylene flame and measured at its 228-nm line.The reaction of nitrate under the prescribed conditions proceeds through a four-electron reduction process. No interferences occur in the presence of up to a 100-fold mass excess of C1- C2042-, C0,2- and NH4+. The method has been applied with excellent results to inorganic and organic nitrates. Mitsui and F ~ j i m u r a l ~ ~ proposed a simple and rapid method for the indirect determination of chlorates. The method is based on the reduction of chlorate (0.012-2.24 pg ml-l of C103-) to chloride with Fe(II) precipitation of silver chloride and its dissolution with ammonia. The method is not very selective. 1.1.4.2. Organic compounds Some organic compounds have also been determined indirectly by redox reactions.Two types of reactions have been employed. One involves the reduction of the analyte with a metal in solution and the metal precipitate is dissolved in nitric a ~ i d ~ ~ ~ ~ ~ ~ - and determined by AAS or the unreduced metaP6O is determined by AAS. The other is based on initial oxidation with periodic acidggs161 or potassium pennanganate,le2 precipitati~n~~,l~l or complexationls2 with a metal ion solution and determination by AAS of the metal precipitate after dissolu-tion999161 or liquid - liquid solvent extraction.162 In general the reported procedures have been applied to functional groups (Table IX). Iodides (up to 5 x 2N03- + 4Cd + 10H+ = N,O + 4Cd2+ + 5H2 1436 Compound Aldehydes . Acetaldehyde . . Pol yh ydric alcohols . . . . Aliphatic and aromatic 1,2-diols .. Nitro compounds Sugar Folic acid . . . . GARCIA-VARGAS et aZ. AAS FOR DETERMINATION OF Analyst VOZ. 108 TABLE IX DETERMINATION OF ORGANIC COMPOUNDS BY REDOX REACTIONS Matrix Measured metal Quantitative relationship Characteristics of the method Reference Water Add Tollen's reagent and 0.25-4 ymol ml-' of compound Ag dissolve Ag precipitate in can be determined. Recovery Limit of detection is 0.0019 mg of acetaldehyde HNO is 95-103% 157 Sample is treated with Tollen's reagent; the precipitated Ag is Ethanol filtered and dissolved in HNO 99 Ethanol Alcohols react with HIOl to Limits of determination are (mg): yield iodate which is precipitated glycerol 0.16- lactose 0 82. with silver nitrate. Precipitated sacchar&e 1 193.glucbse 0:28. Ag is dissolved in ammonia tartaric add,'l.i2; benzdin 1.67 99 Ag Water Pb Diols are oxidised with HIO,. 0.4-4 ymol ml-l can be Iodate is then precipitated with determined. Method is rapid and Pb(I1) and this Pb dissolved in nitric acid Recovery 93.3-101.3% 161 Water Nitro compounds are reduced Metal ions interfere strongly, with Zn. Tollen's reagent is added and the precipitated Ag dissolved in HNO into benzene. Resorcinol selective but precision is poor. but their effect can be avoided by extraction of nitro compounds benzoin p-aminophenol a i d ninhydrin also interfere 159 Plant materials c u Cu(I1) is added in alkaline 0.1-0.5 mg ml-l of dextrose can medium Cu 0 removed and be determined. Recovery unreactid ($11) determined 99.6-102.4%.Method compares favourably with AOAC method 160 - Ni Acid is oxidised with Mn0,- Linear range is 1.2-20 yg of Ni the product complexed with per mole of folic acid. No Ni and extracted into IBMK interference is observed from some organic substances 162 1.2. Methods Based on the Formation of Compounds in the Atomisation Cell These methods are based on the fact that some anions form refractory compounds with metal elements at the atomisation cell temperature. They are divided in two groups depend-ing on whether the absorption measurements are monitored on either atoms or molecules. An important characteristic of these methods as compared with those described previously is that the chemical reaction used to establish a quantitative relationship between the metal and anion is carried out in the atomisation cell itself and not in the solution.Therefore they are, in general more rapid as no prior separation of the chemical species of the metal to be deter-mined is required. 1.2.1. This type of method involves measurement of the suppression or enhancement of the atomic signal of a given metal by its reaction in the atomisation cell with the anion to be determined. These methods are rapid but their selectivity is affected by the presence in the analyte-containing solution of other species that might suppress or enhance the absorbance of the metal added. However in order to overcome interferences from metal cations the sample should be treated previously with a cation-exchange re~in.l63-~6~ Similarly in some instances, potentially interfering anions should be removed either by an anion-exchange resinla' or by precipit ation.ls5 The methods are classified into two groups according to the addition of a constant or variable concentration of metal ion to the anion-containing solution.Suppression or Enhancement of the Atomic-absorption Signal of a Metal 1.2.1.1. As a rule the calibration graph is constructed using a suitable concentration of metal to obtain a linear change of absorbance (with negative slope) in the presence of increasing amounts of the analyte at the concentration level expected to be present in the samples. The anion concentration in the sample solution is determined by adding the same concentration of metal as in the calibration graph measuring the absorbance by AAS and relating this measurement to the calibration graph.In this indirect method the described procedures have been classified according to the nature of the metal ion added. Constant concentration of the metaL adde December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1437 Gutsche et al.170 described a rapid micro-method (0.1 pl of sample is employed) for the indirect determination of fluorine by reacting sodium vapour with fluorine-containing compounds at temperatures of about 800 "C. The decrease in the atomic Na concentration gives a relative and specific (Cl- Br- and I- do not interfere practically) value for the amount of fluorine present. The method is also very sensitive (detection limit 0.8 ng) and may be employed as the basis of a fluorine detector device in the gas-chromatographic determination of trifluoroethanol.1.2.1.1 .b. Alkaline earth metals determinations of phosphate szclphate fluoride and organic compounds. Phosphates (4-20 pg ml-l of P) and sulphates (10-30 pg ml-l of S) can be determined by the depressing effect they have on the atomic absorption of 20 pg ml-1 of Ca at the 422.7-nm line in an air - acetylene flameel El Shaarawy171 pointed out that the sensitivity is highest at P to Ca ratios lower than 0.5 and although the sensitivity increases on using an air - hydrogen flame an air - acetylene flame is preferred because interferences are fewer. The method has been satisfactorily applied to the analysis of P in foods171 and rock phosphate.lG3 A practical method for the determination of phosphate (up to 10 pg ml-l of P) in aqueous and biological media has been proposed by E ~ h o v a l ~ ~ employing a constant concentration of added Sr and keeping a Sr to P ratio of greater than 1.7.Absorbance measurements are made at the 460.7-nm Sr line. The method has been used for determining P in rock phosphate.ls4 Bond and O'DonnelP5 have utilised the depressing effect of F- ions on the Mg atomic signal to determine fluoride (0.2-20 pg ml-l of F-) in an air - coal gas flame. The method is sub-ject to interferences mainly from phosphate sulphate and a high concentration of mineral acids. Aluminium lanthanum acetate and oxalate also interfere at concentrations greater than 1000-fold the fluoride concentration. Few other cations interfere at a significant level, their interference being compensated for by the addition of an excess of cation to both the sample and the standard solutions.M) and ribonuclease (<3 pg ml-l) may be determined by their depressive action on the absorbance of 25 and 14 pg ml-l of Ca, respectively. 1.2.1.1 .c. Transition metals ; determinations o fjzcoride sulphate and ammonia. A procedure based on the enhancement of Zr atomic absorption at 360.1 nm (500 or 2000 pg ml-l of Zr are employed) by fluoride in the dinitrogen oxide - acetylene flame has been reported by Bond and O'D0nne1l.l~~ The method is less sensitive (5-200 pg ml-l of F-) than the Mg-based method described previously but allows a higher degree of freedom from interferences. In the presence of phosphate which interferes in the Zr method a similar enhancement of Ti atomic signal allows the determination of fluoride in the range 40400 pg ml-l by addition of 400 pg ml-l of Ti and measuring at the 364.3-nm Ti line.165 Kunishi and Ohno173 have developed an atomic-absorption inhibition - release method for the determination of the sulphate in rhodium-plating baths.The method involves a decrease of the Fe(II1) atomisation in the plating solutions and a corresponding increase in Fe absorb-ance when a standardised La(II1) solution is added. The authors propose the following mechanism : 1.2.1.1 .a. Alkali metals determination of fluorine. Christian and Feldman6 have reported that glucose ( Rh(S04),'2r8-3)- + Fe3+ + Rh(S04).'a"-3)- - Fe3+ 3La3+ + 3S042-tfree) + La2(S04), Rh(S04),(2n-3)- - Fe3+ + (2n/3)La3+ + (n/3)La,(S04) + Fe3+ + Rh3+ At the point where La(II1) /SO4,- is approximately 2/3 the Fe absorbance increases steeply and sharp breaks occur.The amount of total sulphate is quantitatively determined by extrapolation at the position of these breaks. The enhancement of Zr absorbance in a dinitrogen oxide - acetylene flame at its 360.1-nm line by ammonia has been employed by Bond and Willis174 for determining ammonia. The calibration graph covers 1.7-85 pg ml-1 of NH,. There are no interferences from NO3- and S042- but Po4,- does interfere. Mitsui and Fujimural75 reported an interesting method for the indirect determination of trace amounts of ammonia (up to 20 pg ml-l of NH,+) by the enhancement effect on the atomic absorption of silver in an air-acetylene flame.The recovery ranged from 98 to 100%. Calcium and zinc interfere 1438 Analyst Vol. 108 1.2.1.2. In recent years new indirect determinations based on sharp shifts in the magnesium atomic absorption have been developed. These methods are divided into two groups atomic absorption inhibition titration (AAIT) and atomic-absorption inhibition - release titration (AAIRT) . 1.2.1.1 .a. Atomic-absorption inhibition titration determinations of sulphate and phosphate. The inhibition titration experiment involves the addition of an Mg(I1) titrant solution to a stirred solution of the anion from which metal cations have been removed. The titrated solution is simultaneously aspirated into an air - hydrogen flame and the atomic signal of the magnesium is monitored.In the AAIT method the experimental conditions such as a relatively cool flame the hydrogen - air flow ratio and the beam height are optimised. However the most critical parameters are the flame temperature and the anion to metal molar ratio.166J69 Magnesium is selected as the titrant metal owing to both its atomic-absorption sensitivity and the large inhibition that can be brought about by anions forming Mg refractory compounds. Important advantages of this method are speed simplicity and sensitivity. However the devices used proved to be suitable for studying analytical interference effects and flame reaction~.~~~*l~8 A rapid and sensitive method for the determination of sulphate [Fig. 4 ( a ) ] a t trace levels (1-20 pg ml-l of SO,2-) has been described by Looyenga and Huber.166 In the absence of trace amounts of interfering ions (e.g.phosphate and silicate) the method is applicable to GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Variable concentration of the metal added Typical calibration graphs are given in Fig. 4. 0.6 0.4 m e s n a 0.2 0.6 a, 0.4 f 2 n 4 0.2 80 60 8 40 4? s n a 20 1 .o 2.0 3.0 Time/mi n I 0.3 1.0 2.0 3.0 4.0 Time/mi n I I I 2.0 4.0 6.0 8.0 10.0 Volume of Mg solution/ml ,I Volume of t---*le Volume of Mg solution/ml HCI solution/ml Fig. 4. Typical graphs recorded in AAIT (~9 b1S7 and c1"9 and AAIRT (P9). (a) Titrations of 2.0 4.0 and 8.0 pg ml-l SO,2- with 50 pg ml-l Mg solution; titrant flow-rate, 1.49 ml min-l. ( b ) Titration of 8 pg ml-1 P043- with 200 pg ml-l Mg solution; for a b and c see text.(c) Titration of 1.0 pg ml-l SiO, 4.0 pg ml-l Po43- and 20 pg ml-l SO4,-with 50 pg ml-l Mg solution; titrant flow-rate 2.03 ml min-l; for A B and C see text. ( d ) Titration of 2.0 pg ml-l F- with 50 pg ml-l Mg solution followed by 0.001 M HC1. Aspiration rates MgCl, 2.0 ml min-l; HC1 1.05 ml min-1. AA fluoride analytical signal; solid line smoothed recorder trace December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1439 concentrations as low as 0.1 pg ml-1 of SO,2-. Crawford et aZ.167 studied an Mg-based AAIT method for ortho- pyro- tri- tetra- and hexaphosphates. The shape of the calibration graph for orthophosphate [Fig. 4(b)] suggests the formation of refractory compounds with 2 3 and 4 Mg atoms for each atom of P.At points a b and c the Mg to PO,,- ratio remains essentially constant and any of these points may serve as an end-point for the titration even in the presence of sulphate. The procedure has been used to determine phosphate satisfactorily in river water and detergents. A similar method has been used by Nyangababo and Hamya168 for determining phosphate in fresh waters. A very rapid sensitive and selective method for the simultaneous titration of silicate, phosphate and sulphate has been r e p ~ r t e d . l ~ ~ J ~ ~ The titration graph shows three distinct rectilinear segments (A B and C) with different slopes [Fig. 4 ( c ) ] corresponding to silicate, silicate and phosphate and silicate phosphate and sulphate respectively.The method is sensitive enough to determine these three anions in most waste waters as well as silicate and sulphate in surface and drinking waters. Fong and Huber1s9 have developed a magnesium-based AAIRT method for the determination of fluoride (0.1-10 pg ml-l of F-) in solution. It is based on the distinctive pH dependence of the MgF inhibition reaction [Fig. 4 ( d ) ] . The sample is adjusted to pH 5.2 & 0.2 and an excess of MgCl is added until a pre-determined signal magnitude is obtained [ A in Fig. 4 ( d ) ] . The addition of Mg is then stopped and the addition of 0.001 M HC1 is begun. With decreasing pH, a distinct upward shift in the Mg atomic-absorption signal is obtained corresponding to the amount of Mg released from the inhibition of fluoride [AA in Fig.4(d)]. This effect permits the determination of fluoride in the presence of equal or larger amounts of sulphate and other anions. I t has even been suggested that the procedure can also be employed for the standard-isation of an acid solution. The numerical value of Do (the number of milliequivalents of the acid solution added that are necessary to reach a constant value in the Mg atomic signal) is inversely proportional to the acid concentration and is obtained from the Mg absorbance versus volume of reactants as observed in Fig. 4 ( d ) . 1.2.2. Molecular-absorption Spectrometry Determination of Halogens Molecular-absorption spectrometry (MAS) has seldom been applied to the determination of anions.lsO~l*l In this section those MAS methods are described which are based on the forma-tion of stable metal - halogen molecules in the atomisation cell by nebulisation of the anion-containing solution to which a metal ion in excess has been added.The molecular absorption of the compound formed is then measured at the absorption peak by means of a deuterium lamp. An important drawback of these indirect methods is the serious background absorption.lsO Tsunoda et aZ.lSo have determined nanograms of C1- and Br- by means of molecular absorp-tion of AlCl at 261.4 nm (sensitivity 0.12 ng of Cl-) and AlBr at 279.0 nm (sensitivity 1.1 ng of Br-) in a carbon-rod furnace. Aluminium solutions containing Fe(II1) + Sr(I1) or Co(I1) + Sr(I1) may be employed to decrease background absorption as well as a two-channel spectro-photometer.The method is rapid simple sensitive and relatively free from interferences of foreign ions. However interferences from Br- or I- on the AlCl absorption and I- on the AlBr absorption are not reported. These workers have applied this method with good results, to the determination of chlorine in NBS orchard leaves and chlorine or bromine in solutions containing trace amounts of organo-oxychlorine and bromine compounds. A similar procedure has also been describedlsl for the determination of chloride (up to 28.6 pg ml-l of C1-) by molecular absorption of InCl at 267.2 nm. Silicate must be removed or previously determined. 1.2.1.1 .b. Atomic-absorption inhibition - release titration determination of Jluoride. 1.3. Methods Based on the Formation of Volatile Compounds: Determination of Halogens In this section the methods associated with the formation of gaseous compounds fed into the atomisation cell are described.The methods involving the formation of PH are not con-sidered here as they have been reviewed widely in the literature on volatile hydrides. Gutsche et al.la2 developed a method for the specific determination of trace amounts of fluoride via gaseous SiF,. The atomic absorption of Si is measured at its 251.6-nm resonance line using either a dinitrogen oxide - acetylene flame (detection limit 30 pg of F-) or a graphite furnace (detection limit 0.17 pg of F-) 1440 GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Analyst Vol. 108 A rapid but not very sensitive and selective method has been employed for determining chloride (50-200pgml-l of C1) in minute volumes of inorganic and organic compounds in solution by means of gaseous CuCl.lS3 Iodides (1 x 10-6-5 x M) have been determined by heating solutions containing I-and Hg(NO,) in a graphite tube.TWO Hg absorption peaks appear because the decompo-sition temperature of HgI is higher than that of Hg(NO),),.18* Cyanide S2- and S,0,2-interfere but in the presence of H,O do not. Interfering cations may be removed by extraction as metal quinolin-8-olates. 2. DIRECT METHODS In these methods the direct atomic signal measurement of an element is related to its con-centration. They are divided into methods based on non-metallic elements (inorganic anions) and on metallic elements (organometallic compounds and metal chelates) according to the nature of the measured element.2.1. Non-metallic Elements The direct determination by AAS of non-metallic elements (halogens 0 S N P and C) involves special difficulties. Some of these problems arise from the fact that these elements exhibit their principal resonance in the vacuum ultraviolet region where an intense absorption by the atmosphere and the flame itself is observed (Table X). Therefore measurements in this spectral zone requires an atomic-absorption spectrophotometer whose optical system and instrumental assembly can be evacuated or purged with an inert gas. Although a vacuum monochromator has been employed a nitrogen or argon purged monochromator of greater luminosity can be used as ~ e l l . l 8 ~ Such instruments are not readily available in many analytical laboratories.However because of the very high absorption in the ultraviolet region particular attention should be paid to desorption processes that can distort the analyti-cal signal. Linked with this is the strong non-specific absorption to be expected in this range, which must certainly be taken into account in measuring. Therefore background correction is essential which may be carried out by alternative measurement of the radiation from a line source and that of either a continuous source or a non-resonance line. TABLE X MAIN RESONANCE LINES AT WAVELENGTHS OF LESS THAN 200 nm OF NON-METALLIC ELEMENTS * Element Linelnm Element Linelnm F 95.2 95.5 0 130.22 c1 134.7 138.0 S 180.7 182.4 182.6 Br 145.0 148.9 157.6 N 119.96 I 178.3 183.0 P 177.5 178.3 178.8 C 156.1 165.7 * From references 3 4 and 190.Other difficulties arise from the tendency of 0 P N S and C to form molecular species even at high temperature~.l~~-~~~ Nevertheless this fact may be made use of for the determination of non-metallic elements.186p187 With a conventional instrument non-metallic elements can be determined in the region of the spectrum above 200 nm by either atomic or molecular absorption using a non-resonance line of the element or an adequate band of the molecular species of the analyte. However the sensitivity under these conditions is generally lower than that obtained measuring at wave-lengths below 190 nm. In addition the sensitivity is highly temperature dependent in some instances .I89 Apart from the problems connected with the analytical line difficulties are also encountered in finding intense and stable emission sources for non-metallic volatile elements such as halo-gens S and P.Hollow-cathode lamp (HCL) sources do not have sufficient intensity to permit their use fo December I983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1441 non-metallic elements with the existing optical arrangement.lgl Also it is difficult with ele-ments of high volatility to ensure the manufacture of reproducible stable sources.192 Although electrodeless discharge lamp (EDL) sources have circumvented some drawbacks of the sealed-off HCLs for these elements in relation to the emission intensity signal to back-ground intensity ratio stability and limited operating lifetime other problems still exist.For example there is an appreciable self-absorption when EDLs are employed. As reported by Kirkbright and some difficulty has been experienced in achieving intense emission at the resonance lines below 200 nm for P S and I particularly if source line broaden-ing owing to self-absorption is to be minimised. This self-absorption can be partly eliminated by operating the source at lower powerlg2 or a t low pressure.lg4 TABLE XI SENSITIVITIES DETECTION LIMITS AND TOLERANCE LIMITS OF PHOSPHORUS SULPHUR AND IODINE FOR SOME ANALYTICAL LINES EMPLOYED I N DIRECT METHODS BY FLAME u s Ele-ment Source * P . . DDL P HCL P HCL P -P EDL P EDL P EDL S DDL S DHCL S DHCL S DHCL S EDL I EDL I DHCL I _. EDL I CDL Wavelength/ nm 246.0 213.5/213.6 214.9 213.5/213.6 177.5 178.3 178.8 207 182.6 182.0 180.7 180.7 206.2 183.0 183.0 183.0 Detection Flame type Sensitivity t limit t Tolerance limits for foreign ions 1 Reference NZO - CZH, -290 540 -4.8 5.4 8.8 10 5 3 1 9 1300 14 1 2 6 2 0 --180 29 21 37 -10 5 5 30 600 22 25 9 NH,+ Na K Mg Ca Mn Cu Zn (1:1.2) Many cations interfere As above Al Ca Cu Fe K Na (1 1) I-; not reported for other ions A1 Ba Ca Cd Cu K Li Mg Mo Na Ni, Z< Cl,’Br,’F kO,’ S)O,IL Bb,- )cO,;-, EDTA (1 40) I-; not reported for other ions § § § § Al Cu K Mg Mn Mo Na Ni Zn F C1, Br Pod8- (1 5) § § Cu Co Cr K Na Ni Mg Mo Zn A1 ~ 4 1 ~‘o);’F %I & 6oa-’ P~,s-,’so>-(1:lOO) § 186 197 197 198 191 191 191 187 192 192 192 185 199 192 195 200 * DDL deuterium discharge lamp.CDL capillary discharge lamp of He - I vapour. t Definitions and units are given i i Tabl; IV. $ In parentheses P to foreign ion mass ratio except for reference 186 which is molar ratio. f Not reported. Kirkbright and Wilsonlg2 described a type of demountable hollow-cathode lamp (DHCL) as an alternative sharp line source for P S and I and compared it with the corresponding micro-wave excited EDLs for the same elements. The most significant improvement in performance characteristics is observed for the sulphur DHCL source. However using a DHCL the resonance lines for P in the 177-179 nm spectral range are rather weak but the 213.5 - 213.6 nm and 252.5 nm lines show appreciable intensity.From the elements given in Table X it is to be noted that physical conditions are not favour-able for the determination by AAS of certain gases (N and 0) and C.lgo Because of the low degree of dissociation an unwanted high theoretical detection limit is expected for these elements. A particularly high experimental cost is to be expected for elements with resonance lines below 105 nm (e.g. F) as for this range there are no materials for windows.Lg0 These methods are classified into two groups flame and electrothermal methods according to the procedure used to obtain the atom population 1442 GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Analyst V d . 108 2.1.1. Flame Methods The strong radiation absorption in vacuum ultraviolet by the atmosphere and most flames made the use of wavelengths above 200 nm in the early stages of the direct determination of non-metallic elements by flame AAS compulsory and until recently it has been assumed that only poor sensitivity could be attained.However since the description by Kirkbright and ~ ~ - w o r k e r s ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ of the N,-separated dinitrogen oxide - acetylene flame the determination of these elements at their more sensitive resonance lines shorter than 200nm became possible. This flame shows a remarkable transparency at these wavelengths owing to the absence of absorbing oxygen species in the interconal zone of the flame and the separation of the oxidising outer mantle by N shielding.Although the dilution effects are less severe with other flames such as air - acetylene and nitrogen - hydrogen entrained air these latter flames show a transparency too low to be used below 190 nm. In addition chemical and physical interferences are more serious in these cooler f l a m e ~ . ~ ~ ~ s ~ ~ ~ J ~ ~ A correlation of sensitivities detection limits and tolerance limits of foreign ions for P S and I using different flame types is given in Table XI. 2.1.1 . l . Atomic-absoyption spectroscopy determinations of phosphoms s.ulPhw and iodine It can be seen that the most suitable lines for the direct determinaion of P are in the spectral region between 177 and 179 nm. No background emission is detected at this spectral zone.191 In order to circumvent the difficulties arising from work in the vacuum ultraviolet the less sensitive non-resonance lines at the 213.5 - 213.6 nm doublet have also been employed (Table XII).The determination of P by AAS does not exhibit the severe interferences that are given in flame emission but it is considerably less sensitive.lg8 TABLE XI1 ANALYTICAL APPLICATIONS OF DIRECT ATOMIC-ABSORPTION TECHNIQUES FOR DETERMINATION OF NON-METALS Element Wavelength/ Sample determined nm Nucleotides . . P 246.0 Foodstuffs P 178.3 Detergents . . P 214.3 NBS orchard leaves P 213.6 Biological materials (NBS) P 214.9; 213.6 Gasoline . . P 214.2 BCS steels . . P 213.6 BAM steels . . P Oils fats . . P 213.5/213.6 Gasoline . . P 213.6 Amino acids and derivatives . . S 207 Oils .. . . S 180.7 HGA 2100 Graphite profile tubes HGA 2100 Graphite profile tubes Graphite furnace HGA 76B HGA 2100 Standard used Hap04 (NH4)ZHP04 H,PO4 * CaHPO t KHZP04 t Triphenyl phosphate 1 NH,HzP04 (NH,),HPO4 t,§ Ammonium phosphateT Tricresyl phosphate $ HZSO4 Dibenzyl disulphidell Rat thyroid . . I 183.0 Graphite furnace KI * Matrix interference is compensated for by the addition method. t A solution of La(NO,) is previously injected into the graphite furnace. Standard dissolved in P-free gasoline. Concentration of element in analysed sample Reference 3.39-574 mg ml-1 of P 186 0.73-1.340/; (m/m) 191 13.8 and 18.7% (m/m) 201 0.21% (m/m) 202 0.19-1.10% (m/m) 203 0.005-0.01 g gal-' of P 0.010-0.062% (m/m) 205 Down to 0.002% (m/m) 206 1-100 pg ml-I of P 3-65.2 pg ml-I of P 204 207,208 209 1-12 mol of S per mole of compound 187 0.106-2.68% (m/m) 193 210 94-117 mg of I per 100 g Zeeman correction is used.~ '1[ Standards and samples are ashed with MgSO and P is converted into MgzP20s and dissolved in HNOS. 11 Standards dissolved in IBMK. Sulphur and iodine are determined at their 180.7- and 183.0-nm resonance lines respectively (Table XII). Although a higher sensitivity is obtained for iodine at 178.2 nm the recorded emission intensity from the EDL source is lower a t this line than a t 183.0 nm and the pre December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1443 cision of the measurements thus becomes lower.211 On the other hand the Leipert amplifica-tion procedure based on the reactions - Br, I + 103-103- + 51- + 6H+ + 31 + 3H,O provides a 38-fold increase in sensitivity as reported by Kirkbright et al.lg6 There is however, a loss of selectivity when substances capable of oxidising iodide to elemental iodine are present.Nevertheless interfering cations may be removed by cation exchange. The sensitivity for both sulphur and iodine can be enhanced by solvent e x t r a c t i ~ n l ~ ~ ~ ~ ~ or by addition of a miscible organic solvent to the aqueous s01ution.l~~ This fact is fully accounted for by the improved nebuliser efficiency obtained with the organic solvent. 2.1.1.2. Molecdar-absorption spectroscopy determinations of phosphorw and sulphzcr Both P186,189 and S187v188 show a great tendency to form molecular species with oxygen in most atomisation - excitation media the determination of these elements at the 246.0 nm (PO) and 207 nm (SO,) molecular bands also being possible (Table XII).TABLE XI11 SENSITIVITIES DETECTION LIMITS AND TOLERANCE LIMITS OF PHOSPHORUS SULPHUR, IODINE AND BROMINE FOR SOME ANALYTICAL LINES EMPLOYED I N DIRECT METHODS BY ELECTROTHERMAL Element* Wavelength/nm Atomisation device Sensitivity? P P P P P P s s s s s I I I I I I I I Br 177.5 213.6 213.5/213.6 214.2 213.5/213.6 213.5/213.6 180.7 180.7 182.0 182.6 216.9 178.2 183.0 183.0 183.0 183.0 183.0 206.1 206.16 148.86 Graphite furnace HGA 2100 0.075 f HGA 2100 0.075 f HGA 2100 -HGA 76B -HGA 76 -Graphite furnace Graphite furnace 0.42 yg Graphite furnace 0.68 yg Graphite furnace 1.5 yg Microwave discharge -in quartz cell 1.3 x loe f 5.4 x lo7 3 Graphite tube 0.02 Graphite tube 0.04 Graphite furnace HGA 2000 0.1 Cathode sputtering 1 2 cell Platinum loop 9 1.2 x lo8 3 HGA 2000 3.5 Graphite cuvette -Special graphite -furnace An EDL source is used in all instances.t Definitions and units are given in Table IV. $ Sensitivity given in absorbance per gram. 6 Not reDorted. Detection limitt Tolerance limits for foreign ions11 Reference - § 214 202 0.1 1 § 217 0.2 f § 204 0.3 7 Mg Na K As (1 100); Si (1 50) 0.27 f 2 ng § 207 218 - 5 214 - § 219 219 219 0.3 yg as H,S or 3 188 0.7 yg as SO2 - § 211 NaCl Na,HPO (1 100) 211 - NaNO (1 1000) ; KCl Na,SO,, - § 214 0.4 6 C1 Br Na (1 100) C1 Br F NO,- (1 50) 210 221 18 Interfering cations are removed by cation exchange and some anions via oxidation of iodide to iodine and solvent extraction followed by reduction of iodine to iodide 220 10 § 2 ng 0 1.5 ng § 210 215 190 La(N6& solution is employed for correcting matrix interferences.11 Values in parentheses are the P to foreign ion mass ratio 1444 Analyst VoZ. 108 2.1.2. Electrothermal Methods Determination of Phosphorus Sulphur and Iodine The first attempts to determine P by electrothermal AAS were made by Walsh212 in the early 1960s and later by L’vov and Khartsyzov213 in their experiments with a graphite cuvette.The direct determinations of S and I by electrothermal techniques were reported by L’vov and K h a r t s y z ~ v ~ ~ ~ s ~ ~ ~ and by Berezin.,16 In a fuel-rich N,-separated dinitrogen oxide - acetylene flame three problems arise211 (i) a relatively large volume of sample is required for nebulisation; (ii) the sensitivity is limited by the difficulty of achieving high sample concentrations in the cell owing to dilution by the support gas; and (iii) at short wavelengths a degradation of the signal to noise ratio occurs owing to fluctuating absorption at the flame - atmosphere interface. Therefore the use of non-flame atomisation devices overcomes the limitations imposed by the nature of the flames the absorp-tion by flame gases is eliminated and compared with flames reduced noise levels as well as higher sensitivities can be a c h i e ~ e d l ~ ~ s ~ ~ ~ (Table XIII) .As the inert purge gas either nitrogen or argon can be utilised. However argon has been reported to produce a two-fold increase in sensitivity for P with respect to On the other hand the sensitivity is greatly affected by temperature and tube The use of graphite profile tubes has been proved to increase the sensitivity of many elements by a factor of 3-5 over that of standard cylindrical GARCIA-VARGAS et al. AAS FOR DETERMINATION OF The graphite furnace is the most widespread technique of electrothermal atomisation. TABLE XIV COMBINATION OF CHROMATOGRAPHIC AND ATOMIC-ABSORPTION SPECTROMETRIC TECHNIQUES FOR ANALYSIS OF ORGANIC COMPOUNDS Chromatographic* Compound Matrix method Lead alkyls .. Gasoline HPLC Metallothionein Protein HPLC Tin tetraalkyls, alkyltin chlorides HPLC or GC Triphenyltin . . Biological LC Samples Tetraalkyl lead . . Gasoline LC Tributyltin oxide (I), triphen yl tin chloride (11) . . Synthetic TLC resin emul-sion paints Alkyl selenium . . GC Tetraalkyl lead . . Air GC Alkyl mercury . . Fish tissue GC Ga and In organometallic compounds . . GC AAS Pb; flame Cd Zn or Cu; flame Sn; flame Sn; flame Pb; flame Sn; electrothermal Se; flame Pb; electrothermal Hg ; electrothermal Ga or In; flame Comments Determination at pg level Gel-permeation column is used Detection limit 11-19 pg. Sensitivity 0.9- 1.5 wg Column packed with A1,0,. Compound to be determined is previously extracted into CHCI -methanol Detection limit ca.10 pg of Pb Detection limit 0.2 pg g-l. Average recovery for 1 g of paint containing 10 mg of compound (I) or (11) was 74.7-82.1 and 87.1-90.1%, respectively Determination at pg level Detection limit 0.3 pgml-l for 0.5-g samples Determination at wg level. Standard deviation 0.06% Reference 226 227,228 229 230 231 232 233 234,235 236 237 * HPLC high-performance liquid chromatography; GC gas chromatography; LC liquid chromatography; TLC thin-layer chromatography. Other electrothermal atomisation devices are the platinum loop22o and the cathode sputter-ing ce11.212s221 This latter offers the advantage of a longer retention time for the atom cloud so that the absorption peak can be measured with standard equipment.For iodine221 good reproducibility may be obtained and interferences may be compensated for by matching the composition of both samples and standards. The analytical applications of the determination of P and S in di‘fferent samples are sum-marked in Table XII December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1445 2.2. Organic Compounds Organic substances can be determined by direct AAS methods if they contain a metal ele-ment in their molecule. Vitamin B, contains one atom of cobalt per molecule. Therefore a method has been pro-posed for the determination of this vitamin in pharmaceutical dosage forms based on the atomic absorption of cobalt at the 242.5-nm 1ine.222p223 Dimethylpolysiloxanes have been determined in fats and and juices and beer225 via atomic signal measurement of Si at its 251.6-nm resonance line.However direct AAS methods for the determination of organic compounds are widely employed for the rapid analysis of organometallic compounds by a hybrid analytical technique based on the GC - AAS coupling. In Table XIV some procedures based on this principle are summarised. 3. CONCLUSIONS A comparative study of the relative analytical utilisation of the described methods is pre-sented in Table XV. From this table the following conclusions can be drawn. TABLE XV COMPARATIVE STUDY ON THE APPLICABILITY OF THE DESCRIBED METHODS Method Inorganic anions Organic compounds Indivect-Precipitation 17.3 4.4 Heteropoly compounds 10.4 -Volatile compounds 1.2 -Complex formation 12.9 16.9 Redox reactions 2.4 2.8 Direct atomisation 0.3 0.4 Total 53.5 Total 24.5 Flame 5.6 6.8 Electrothermal 9.3 Total 14.9 Total 6.8 Overall total 68.4 Overall total 31.3 Dived--1.Most investigations have been devoted to the determination of inorganic anions as can be concluded readily from the publication of twice as many papers related to their determina-tion as those devoted to organic compounds. 2. The indirect methods have found much greater application than direct methods in the determination of both inorganic and organic anions. They are more tedious and time consum-ing than the direct methods but their methodology is much better established. For the indirect methods precipitation reactions have been used most for determining inorganic anions while organic compounds have been determined mainly by means of complex formation reactions and a large number of indirect methods are suitable for the determination of organic functional groups.They were developed in the early 1970s and are still faced with many research and technical problems. Their use for organic analysis is a consequence of the employment of the atomic-absorption spectro-photometer as a chromatographic detector. We have attempted to prove how powerful a tool AAS is for the analysis of inorganic anions and organic compounds. However much effort has to be made is the future to increase the applicability of AAS methods to the fast-growing field of organic analysis and to improve further the development of the direct methods in connection with recent advances in instru-ment ation.3. 4. The direct methods are used a third less than the indirect ones. References 1. 2. 3. Pinta M. Methodes Phys. Anal. GAMS 1970 6 (3) 268. Gold’shtein M. M. and Yudelevich I. G. Zh. Anal. Khim. 1976 31 801. Kirkbright G. F. and Johnson H. N. Talanta 1973 20 433 1446 GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Analyst Vol. 108 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65.66. 67. 68. 69. 38. Kirkbright G. K. and Sargent M. “Atomic Absorption and Fluorescence Spectroscopy,” Academic Press London 1974. Williams W. J . “Handbook of Anion Determination,” Butterworths London 1979. Christian G. D. and Feldman F. J. Anal. Chim. Acta 1968 40 173. Gannibal L. F. Elina N. A. Taravkova E. K. and Ingura V. A. Metody Izuch. Sostava Svoistv. Begak 0. Yu. Eustrop’ev K. K. Pronkin A. A. and Sosmovskaya 2. N. &I. Vses. Simp. Kushiro H. Tanako Y. and Fukui I. Bunseki Raiburari 1981 3 197. Suzuki K. T. Kagaku no Ryoiki Zokan 1981 133 79. Varley J. A. and Chin P. Y. Analyst 1970 95 592. Montiel A. Trib. CEBEDEAU 1972 25 292. Bataglia 0. C. Cienc. Cult. (Sao Paulo) 1976 28 672. Galle 0. K. and Hathaway L. R. Afipl. Spectrosc. 1975 29 518.Borden F. Y. and McCormick L. H. Soil Sci. SOC. Am. Proc. 1970 34 705. Dunk R. Mostyn R. A. and Hoare H. C. A t . Absorpt. Newsl. 1969 8 79. Sinha R. C. P. Singhal K. C. and Banerjee B. K. Technology (Sindri India) 1969 6 198. Magyar B. M. and Santos F. Helv. Chirn. Acta 1969 52 820. Lautenbacher H. W. and Baker H. W. A m . Lab. 1972 4 45. Little I. P. Reeve R. Proud G. H. and Lulham A. J . Sci. Food dgric. 1969 20 673. Rose S. A. and Boltz D. F. Anal. Chim. Acta 1969 44 239. Ametani K. Bunseki Kagaku 1974 23 745. Conto M. I. and Adilson A. J. Appl. Spectrosc. 1980 34 228. Iosof V. and Calinescu E. Rev. Chim. (Bucharest) 1979 30 958. Yoshida Z. and Takahashi M. Mikrochinz. Acta 1977 1 459. Ray R. C. Nayar P. K. Misra A. K. and Sethunathan N. Analyst 1980 105 984.Hassan S. S. M. and Eldesouki M. H. Mikrochim. Acta 1979 2 27. McHugh J. B. and Turner J. H. U.S. Geol. Surv. Prof. Pap. 1980 1129-A-1 Fl-F3. Ezell J. B. Jr. At. Absorpt. Newsl. 1967 6 84. Bartels H. A t . Absorpt. Newsl. 1967 6 132. Gil de la Pena M. L. and Garrido M. D. A n . Bromatol. 1975 27 269. Smith R. V. and Nessen M. A. Microchem. J . 1972 17 638. Ametani K. Bull. Chem. SOC. Jpn. 1974 47 242. Garrido M. D. Llaguno C. and Garrido J. Am. J . Enol. Vitic. 1971 22 44. Truscott E. D. Anal. Chem. 1970 42 1657. Ozawa T. Tokyo-toritsu Kogyo Gijutsu Senta Kenkyu Hokoku 1980 9 115. Tofuku Y. and Hirand S. Bunseki Kagaku 1971 20 142. Gambrell J. W. A t . Absorpt. Newsl. 1972 11 125. Listskaya G. S. Zaets V. V. and Zatsarevyi V. M. Metody Kontrolya Kach.Prod. Khim. Prom-sti., Gladysheva V. P. Nurtaeva A. K. Mal’tseva T. A. and Vinogradova G. I. Zavod. Lab. 1982 45, Menache R. Clin. Chem. 1974 20 1444. Welsch E. P. and Chao T. T. Anal. Chim. Acta 1975 76 65. Smith R. V. and Bailey D. L. Anal. Chim. Acta 1974 73 177. Gupta H. K. L. and Boltz D. F. Microchem. J . 1971 16 571. Zawadski H. Szymendera J. Krzysztof K. and Karczynski F. Chem. Anal. (Warsaw) 1981,26,237. Hassan S. S. M. J . Assoc. Ofi. Anal. Chem. 1981 64 1158. Sheridan J. C. Lau E. P. K. and Senkowski B. Z. Anal. Chem. 1969 41 247. Galindo G. G. Appelt H. and Schalcha E. B. Soil. Sci. SOG. A m . Proc. 1969 33 974. Wollin A. At. Absorpt. Newsl. 1970 9 43. Roe D. A. Miller P. S. and Lutwak L. Anal. Biochem. 1966 15 313. Oeien A. Acta Agric. Scand.1979 29 71. Meyer J. L. and Rundquist R. T. Biochem. Med. 1975 12 398. Reichel W. and Acs L. Anal. Chem. 1969 41 1886. Westerlund-Heherson U. A t . Absovpt. Newsl. 1966 5 97. Fujinuma H. Kasama K. Takeuchi T. and Hirano S. Bunseki Kagaku 1970 19 1487. de Almedia M. A. T. M. de Moraes S. and Barberio J. C. Relat. Inst. Energia Atom. IEA-285, Gupta H. K. L. and Boltz D. F. Microchem. J . 1971 16 571. Marhevka J. S. and Siggia S. Anal. Chem. 1979 51 1259. Danchik R. S. Boltz D. F. and Hargis L. G. Anal. Lett. 1968 1 891. Siemer D. D. Woodriff R. and Robinson J. Appl. Spectvosc. 1977 31 168. Jungreis E. and Ain F. Anal. Chim. Acta 1977 88 191. Jungreis E. Isr. J . Chem. 1969 7 583. Jungreis E. and Kraus S. MiRrochim. Acta 1976 1 413. Manahan S. E. and Kunkel R.Anal. Lett. 1973 6 547. Tsongas M. and Kevatsis A. Vet. Hum. Toxicol. 1979 21 190; Chern. Abstr. 1979 91 18745~. Jungreis E. and Anavi Z. Anal. Chim. Acta 1969 45 190. Slavin W. and Schmidt G. I. J . Chromatogr. Sci. 1979 17 610. Hurtubise R. J. J . Pharm. Sci. 1974 63 1131. Manahan S. E. and Jones D. R. Anal. Lett. 1973 6 745. Mineral. Gorn. Povod. Kol’sk. P-va Apatity 1980 74; Chem. Abstr. 1981 95 72602g. Pokhimii Neorgan. Ftoridov Dnepropetrovsk 1978; Chem. Abstv. 1979 90 33514g. 1982 35; Chem. Abstr. 1982 96 1925232. 1. 1973; Chem. Abstr. 1974 80 127908j December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1447 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91.92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. Hall F. F. Schneider B. Culp T. R. and Ratliff C. R. Clin. Chem. 1972 18 34. Jones D. R. and Manahan S. E. Anal. Chem. 1976 48 1897. Kunkel R. and Manahan S. E. Anal. Chem. 1973 45 1465. Mitsui T. and Fujimura Y. Bunseki Kagaku 1974 23 1303. Woodis T. C. Jr. and Hunter G. B. J . Assoc. Off. Anal. Chem. 1976 59 22. Corominas L. F. J . Assoc. Ofl. Anal. Chem. 1977 60 1214. Kidani Y. Uno S. and Inagoki K. Bunseki Kagaku 1976 25 514. Minamikawa T. and Matsumura K.Yakugaku Zasshi 1976 96 440. Mitsui T. and Kojima T. Bunseki Kagaku 1977 26 228. Ranjitkar K. P. and Townshend A. Anal. Lett. 1974 7 743. Kidani Y. and Ito E. Bunseki Kagaku 1976 25 57. Marshall G. and Midgley D. Anal. Chem. 1981 53 1760. West P. W. and Lorica A. S. Anal. Chim. Acta 1961 25 28. Danchik R. S. and Boltz D. F. Anal. Chim. Acta 1970 49 567. Danchik R. S. and Boltz D. F. Anal. Chem. 1968 40 2215. Belcher R. Nadjafi A. Rodriguez-Vazquez J . A. and Stephen W. I. Analyst 1972 97 993. Chuchalina L. S. Yudelevich I. G. and Chinenkova A. A. Zh. Anal. Klzim. 1981 36 920. Suzuki M. Coombs T. and Vallee B. Anal. Biochem. 1969 32 106. Minami K. Mitsui T. and Fujimura Y. Bunseki Kagaku 1979 28 513. Minami Y. Mitsui T. and Fujimura Y. Bunseki Kagaku 1980 29 491.Minami Y. Mitsui T. and Fujimura Y. Bunseki Kagaku 1979 28 1717. Minami Y. Mitsui T. and Fujimura Y. Bunseki Kagaku 1981 30 566. Mitsui T. and Fujimura Y. Bunseki Kagaku 1974 23 1303. Mitsui T. and Fujimura Y. Nippon Kagaku Kaishi 1974 10 1908. Oles P. J. and Siggia S. Anal. Chem. 1973 45 2150. Kidani Y. Uno S. and Inagaki K. Bunseki Kagaku 1977 26 158. Saito P. Y. Metal. A B M 1976 32 683. Kneebone B. M. and Freiser H. Anal. Chem. 1975 47 942. Kovatsis A. V. Pharm. Deli?. Epistem. Ekdosis 1978 4 73. Mitsui T. and Fujimura Y. Bunseki Kagaku 1976 25 429. Kidani Y. Osugi N. Inagaki K. and Koike H. Bunseki Kagaku 1975 24 218. Martens F. K. and Heyndrickx A. J . Anal. Toxicol. 1978 2 269. Courtot-Coupez J. and Le Bihan A. Anal. Lett. 1969 2 567. Maynard A.T. and Dodd M. J. Aust. J . Med. Technol. 1974 5 161. Garnys V. P. and Matousek J. P. Clin. Chem. 1975 21 891. Collinson W. J. and Boltz D. F. Anal. Chem. 1968 40 1896. Kumamaru T. Tao E. Okamoto N. and Yamamoto Y. Bull. Chem. Soc. Jpn. 1966 38 2204. Yamamoto Y. Kumamaru T. Hayashi Y. and Otani Y. Bunseki Kagaku 1969 18 359. Houser M. E. and Fauth M. I. Mzcrochem. J . 1970 15 399. Hoshikawa G. and Fudano Y. Kagawa Daigaku Nogakuba Gakujutsu Hokoku 1976 27 111. Yamamoto Y. Kumamaru T. Hayashi Y. and Otani Y. Bunseki Kagaku 1968 17 92. Kumamaru T. Bull. Chem. Soc. Jpn. 1969 42 956. Yamamoto Y. Kumamaru T. Hayashi Y. and Matsuhita S. J . Sci. Hiroshima Univ. Ser. A , Yamamoto Y. Kumamaru T. and Hayashi Y. Talanta 1967 14 611. Yamamoto Y. Kumamaru T. Hayashi Y.and Otsuchi M. Bunseki Kagaku 1969 18 354. Yamamoto Y. Kumamaru T. Hayashi Y. and Otani Y. Bull. Chem. Soc. Jpn. 1969 42 1714. Kumamaru T. Anal. Chim. Acta 1968 43 19. Kumamaru T. Hayashi Y. Okamoto N. Tao E. and Yamamoto Y. Anal. Chim. A d a 1966, Kidani Y. Nakamura K. Inagaki K. and Koike H. Bunseki Kagaku 1975 24 742. Kidani Y. Uno S. Kato Y. and Koike H. Bunseki Kagaku 1974 23 740. Le Bihan A. and Courtot-Coupez J. Bull. SOC. Chim. Fr. 1970 406. Le Bihan A. and Courtot-Coupez J. Analusis 1974 2 695. Matsueda T. and Morimoto M. Bunseki Kagaku 1980 29 769. Alary J. Rochat J. Villet A. and Coeur A. Ann. Pharm. Fr. 1976 34 345. Minami Y. Mitsui T. and Fujimura Y. Bunseki Kagaku 1981 30 475. Minamikawa T. Matsumura K. Kamei A. and Yamamoto M. Bunseki Kagaku 1971 20 1011.Park M. K. Shon Ch. Y. and Shin M. H. Soul Taehakkyo Yakhak Nomunjip 1979 4 140. Zaugg W. S. and Knox R. J. Anal. Biochem. 1967 20 282. Zaugg W. S. A t . Absorpt. Newsl. 1967 6 63. Zaugg W. S. and Knox R. J. Anal. Chern. 1966 38 1759. Linden G. Turk S. and de la Fuente B. T. Chim. Anal. (Paris) 1971 53 244. Kidani Y. Takemura H. and Koike H. Bunseki Kagaku 1974 23 212. Ichinose N. Yamada S. Matsushima H. and Sakurai N. Igaku T o Seibutsugaku 1978. 97 83; de Voto G. Boll. SOC. Ital. Biol. Sper. 1968 44 424. Hernandez Saint-Aubin F. Garrido C. D. and Cabezudo I. A n . Quim. Ser. B 1981 77 370. Inokuma Y. and Endo J. Tetsu T o Hagane 1977 63 1026; Chem. Abstr. 1977 87 94933~. Kirkbright G. F. Smith A. M. and West T. S. Analyst 1967 92 411. Ramakrishna T.V. Robinson J. W. and West P. W. Anal. Chim. Acta 1969 45 43. Suzuki T. Morinaga H. and Sasaki A. Tetsu T o Hagane 1975 63 1063. 1971 35 141. 35 542. Chem. Abstr. 1979 91 116701n 1448 GARCIA-VARGAS et al. AAS FOR DETERMINATION OF Analyst Vol. 108 139. Kumamaru T. Otani Y . and Yamamoto Y. Bull. Chem. Soc. Jpn. 1967 40 429. 140. Bernal J. L. del Nozal M. J. Deban L. and Aller A. J. Talanta 1981 28 469. 141. Yudelevich I. G. Buyanova L. M. Bakhturova N. F. and Korda T. M. Izv. Sib. Old. Akad. Nauk S S S R Ser. Khim. Nauk 1976 5 52; Chem. Abstr. 1977 86 50252e. 142. Rozenblum V. Anal. Lett. 1975 8 549. 143. Bajalordo A. M. and Gomez Coedo A. Rev. Metal. (Madrid) 1973 9 35. 144. Hurford T. R. and Boltz D. F. Anal. Chem. 1968 40 379. 145. Riddle C. and Turek A.Anal. Chim. Acta 1977 92 49. 146. Abbey S. and Maxwell J. A. Anal. Chim. Acta 1978 99 397. 147. Riddle C. and Turek A. Anal. Chim. Acta 1978 99 398. 148. Mimura T. and Wakisaka S. Teikyo Igaku Zasshi 1981 4 143; Chem. Abstr. 1982 96 48537q. 149. Melton J. R. Hoover W. L. Howard P. A. and Green V. S. J . Assoc. 08. Anal. Chem. 1971 54, 149a. “Official Methods of Analysis of the Association of Official Analytical Chemists.” Eleventh Edition. 373. 150. 151. 152. 153. 154. 155. 156. 157. 168. 159. 160. 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195.196. 197. 198. 199. 200. 201. 202. 203. 204. Association of Official Analytical Chemists Washington D.C.,- 1970 Section 2.023. Janousek I. Chem. Anal. (Warsaw) 1979 24 59. Kuga K. and Kitazume E. Bunseki Kagaku 1981 30 164. Parsons J. A. Dawson B. Callahan E. and Potts J . T. Biochem. J . 1970 119 791. Christian G. D. Knoblock E. C. and Purdy W. C. J . ASSOG. 08. Agric. Chem. 1965 48 877. Newberry C. L. and Christian G. D. J . Assoc. Off. Agric. Chem. 1965 48 322. Hassan S. S. M. Talanta 1981 28 89. Mitsui T. and Fujimura Y . Bunseki Kagaku 1974 23 3. Oles P. J. and Siggia S. Anal. Chem. 1974 46 911. Mitsui T. and Kojima T. Bunseki Kagaku 1977 26 182. Mitsui T. and Kojima T. Bunseki Kagaku 1977 26 317. Potter A. L. Ducay E. D. and Macready R.M. J . Assoc. Off. Anal. Chem. 1968 51 748. Tau B. Melius P. and Kilgore M. V. Anal. Chem. 1980 52 602. Kidani Y. Nakamura K. and Inagaki K. Bunseki Kagaku 1976 25 509. Singhal K. C. and Banerjee B. K. Technology (Sindri India) 1968 5 239. Singhal K. C. Banerji A. C. and Banerjee B. K. Technology (Sindri India) 1968 5 117. Bond A. M. and O’Donnell T. A. Anal. Chem. 1968 40 560. Looyenga R. W. and Huber C. O. Anal. Chim. Acta 1971 55 179. Crawford W. E. Lin C. I. and Huber C. O. Anal. Chim. Acta 1973 64 387. Nyangababo J. T. and Hamya J. W. Indian J . Chem. Educ. 1980 7 19. Fong C. C. and Huber C. O. Spectrochim. Acta 1976 31B 113. Gutsche B. Kleinseder H. and Herrmann R. Analyst 1975 100 192. El-Shaarawy M. I. Ind. Aliment. (Pinerolo Italy) 1980 19 477. Ezhova V.M. Lab. Delo 1971 647. Kunishi M. and Ohno S. At. Absorpt. Newsl. 1974 13 29. Bond A. M. and Willis J. B. Anal. Chem. 1968 40 2087. Mitsui T. and Fujimura Y. Bunseki Kagaku 1974 23 449. Lin C. I. and Huber C. O. Anal. Chem. 1972 44 2200. Posta J. and Lakatos I. Magy. Kern. Foly. 1980 86 284. Stojanovic D. D. Durdevic M. D. Vajgand V. J. and Nikolic S. D. Glas. Hem. Drus. Beograd, Huber C. O. U.S. N . T . I . S . PB Rep. No. 22333815 1973 46 pp. Tsunoda K. Fujiwara K. and Fuwa K. Anal. Chem. 1978 50 861. Yoshimura E. Tanaka N. Y. Tsunoda K. Toda S. and Fuwa K. Bunseki Kagaku 1977,26 647. Gutsche B. Rudiger K. and Herrmann R. Spectrochim. Acta 1975 30B 441. Tomkins D. F. and Frank C. W. Anal. Chem. 1972 44 1451. Nomura T. and Karasawa I. Anal. Chim. Acta 1981 126 241.Kirkbright G. F. and Marshall M. Anal. Chem. 1972 44 1288. Haraguchi H. and Fuwa K. Anal. Chem. 1976 48 784. Fuwa K. and Vallee B. L. Anal. Chem. 1969 41 188. Taylor H. E. Gibson J. H. and Skogerboe R. K. Anal. Chem. 1970 42 1569. Persson J.-A. and Frech W. Anal. Chim. Acta 1980 119 75. Becker-Ross H. and Falk H. Spectvochim. Acta 1975 30B 253. Kirkbright G. F. and Marshall M. Anal. Chem. 1973 45 1610. Kirkbright G. F. and Wilson P. J. Anal. Chern. 1974 46 1414. Kirkbright G. F. Marshall M. and West T. S. Anal. Chcm. 1972 44 2379. Childs A. H. and Schrenk W. G. Appl. Spectrosc. 1976 30 507. Kirkbright G. F. West T. S. and Wilson P. J. A t . Absovpt. Newsl. 1972 11 53. Kirkbright G. F. West T. S. and Wilson P. J. A t . Absovpt. Newsl. 1972 11 113.Manning D. C. and Slavin S. A t . Absorpt. Newsl. 1969 8 132. Kerber J. D. Barnett W. B. and Kahn H. L. A t . Absovpt. Newsl. 1970 9 39. Thompson K. C. Spectrosc. Lett. 1970 3 59. Lowe M. D. Sutton M. M. and Clinton 0. E. Appl. Spectrosc. 1982 36 22. Toralballa G. C. Spielholtz G. I. and Steinberg R. J. Afikrochim. Acta 1972 4 484. Ediger R. D. Knott A. R. Peterson G. E. and Beaty R. D. A t . Absorpt. Newsl. 1978 17 28. Langinyhr F. J . and Dahl I. M. Anal. Chim. Acta 1981 131 303. Driscoll D. J. Clay D. A. Rogers C. H. Jungers K. H. and Butler F. E. Anal. Chem. 1978 50, 1979 44 703. 767 December 1983 INORGANIC ANIONS AND ORGANIC COMPOUNDS. A REVIEW 1449 205. 206. 207. 208. 209. 210. 211. 212. 213. 214. 215. 216. 217. 218. 219.220. 221. 222. 223. 224. 225. 226. 227. 228. 229. 230. 231. 232. 233. 234. 235. 236. 237. Whiteside P. J . and Price W. J. Analyst 1977 102 618. Welz B. Voellkopf U. and Grobenski Z. Anal. Chim. Acta 1982 136 201. PrevBt A. and Gente-Jauniaux M. At. Absorpt. Newsl. 1978 17 1. PrevBt A. and Gente-Jauniaux M. Rev. Fr. Corps Gras 1977 24 493. Vigler M. S. Strecker A. and Varnes A. Appl. Spectrosc. 1978 32 60. Kirkbright G. F. and Wilson P. J. At. Absorpt. Newsl. 1974 13 140. Adams M. J. Kirkbright G. F. and West T. S. Talanta 1974 21 573. Walsh A. Colloq. Spectrosc. Int. loth Univ. Maryland 1962. L’vov B. V. and Khartsyzov A. D. Zh. Prikl. Spektrosk. 1969 11 9. L‘vov B. V. and Khartsyzov A. D. Zh. Prikl. Spektrosk. 1969 11 413.L’vov B. V. and Khartsyzov A. D. Zh. Anal. Khim. 1969 24 799. Berezin I. A. Opt. Spektrosk. 1969 26 855. Ediger R. D. At. Absorpt. Newsl. 1976 15 145. L’vov B. V. and Pelieva L. A. Zh. Anal. Khim. 1978 33 1572. Adams M. J. and Kirkbright G. F. Can. J. Spectrosc. 1976 21 127. Manfield J . M. West T. S. and Dagnall R. M. Talanta 1975 21 787. Kirkbright G. F. West T. S. and Wilson P. J. Anal. Chim. Acta 1974 68 462. Berge D. G. Pflaum R. T. Lehman D. A. and Franck C. W. Anal. Lett. 1968 1 613. Diaz F. J. Anal. Chim. Acta 1972 58 455. Doeden W. G. Kushibab E. M. and Ingala A. C. J . Am. Oil Chem. Soc. 1980 57 73. Kacprzak J . L. J . Assoc. Ofi. Anal. Chem. 1982 65 148. Botre C. Cacace F. and Cozzani R. Anal. Lett. 1976 9 825. Suzuki K. T. Pharmacobio-Dyn. 1980 3 S-18. Suzuki K. T. Koen Yoshishu Seitai Seibun no Bunseki Kagaku Shinpojumu 4th 1979 96. Thorburn Burns D. Glockling F. and Harriot M. Analyst 1981 106 921. Manabe M. Wada O. Iwai H. Matsui H. Manabe S. and Ono T. Sangyo Igaku 1981 23 312. Messman J. D. and Rains T. C. Anal. Chem. 1981 53 1632. Kojima S. Nakamura A. and Kaniwa M. Eisei Kagaku 1979 25 141. Radziuk B. and Van Loon J. Sci. Total Environ. 1976 6 251. Chau Y. K. Wong P. T. S. and Goulden P. D. Anal. Chim. Acta 1976 85 421. Chau Y. K. Wong P. T. S. and Saitoh H. J. Chromatogr. Scz. 1976 14 162. Bye R. and Paus P. E. Anal. Chim. Acta 1979 107 169. Shushunova A. F. Demarin V. Y. Makin G. I. Sklemina L. V. Rudneuskii N. K. and Aleksandrov Yu. A. Zh. Anal. Khim. 1980 35 349. Received March 18tk 1983 Accepted June 7th 198

ISSN:0003-2654    

DOI:10.1039/AN9830801417

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

1450 Analyst December 1983 VoL. 108 pp. 1450-1458 Determination of Iron in Semiconductor-grade Silicon by Furnace Atomic-absorption Spectrometry Dona1 A. Stewart and David C. Newton Plessey Research (Caswell) Ltd. Towcester Northamptonshive "12 8EQ The determination of iron in semiconductor-grade silicon for device produc-tion by furnace atomic-absorption spectrometry is described for iron contents down to a level of 5 ng g-1 in slice and polycrystalline material. The silicon matrix is first removed by treatment with hydrofluoric acid - nitric acid. Reagent purification and the selection of the most suitable furnace materials have been studied. Surface iron contamination is assessed separately by a chemical etching technique. Keywords I r o n determination ; furnace atomic-absorption spectrometry ; semiconductor-grade silicon Semiconductor-grade silicon is used in the manufacture of a wide range of integrated circuits in the electronics industry.Fabrication of devices by an ion implanted bipolar process is particularly exacting in the demands made on the purity of the starting material. For ex-ample the presence of metallic impurities such as iron in the silicon has been linked with high device failure rate by Ward.l This paper describes the determination of iron in silicon at a level of a few parts per billion (p.p.b) (one part in lo9) by furnace atomic-absorption spectrometry (FAAS). The original objective was the measurement of total iron in the 3 in diameter slices used for device produc-tion. Subsequently surface contamination of the slices which probably occurs during the slicing lapping and polishing of the single crystal was found to be important and a chemical etching technique was devised to study this aspect.Finally some measurements were also made on the iron content of polycrystalline silicon before growth to a single crystal had taken place. Experimental Technique Principles In this work sensitivity was improved chemically by the dissolution of the silicon in hydrofluoric acid - nitric acid. Evaporation removed the bulk of the matrix and enabled the volume of the solution to be kept to a minimum. This solution could then be analysed directly by FAAS without further treatment. Optimisation of the FAAS measurements in particular selection of the most suitable furnace tubes is discussed in the next section.The iron level in high-grade material can approach the limit of detection by FAAS. Furnace Atomic-absorption Spectrometry Relatively few papers have appeared in the literature concerning low levels of iron in semi-conducting materials. Hoffmeister2 had determined iron contamination by FAAS at a level of 30 ng 1-1 in ultra-pure water used in the electronics industry and the technique was in fact preferred to neutron activation. Volland et aL3 reported a greater than two-fold increase in sensitivity when they compared pyrolytically coated tubes with uncoated ones. They found a dependence upon nitric acid concentration although the equipment (Varian CRA-63) was different from that in this study. Iron has been measured in silica by Fuller.* The matrix was dissolved in hydrofluoric acid and standards were prepared in the same medium.There was little difference between aqueous and hydrofluoric acid standards and no interference was found. Uncoated furnace tubes were probably used although this was not stated. In this work initial tests with pyrolytically coated tubes of American origin gave spurious and erratic results with both iron standards and residues from an actual silicon analysis in which fluoride was also present. This behaviour was found to be characteristic of many different batches of furnace tubes. No improvement resulted when the tubes were subjected to a high temperature cleaning cycle or when the furnace temperature programme was varied and the anomalous behaviour was ascribed to a sporadic release of iron from the tube or coating by the nitric or hydrofluoric acid media.Spurious iron gave a sharply peaked response on the vide STEWART AND NEWTON 1451 display of the spectrometer which could be distinguished from the more rounded peak from an iron standard or silicon sample (Fig. 1). At other times a double peak was seen in which the true and spurious iron signals occurred together. Tests were also made on uncoated tubes from the same manufacturer; spurious signals were not encountered but the sensitivity was reduced to a level that was inadequate for semiconductor requirements (the characteristic concentration was 30 pg per 0.0044 units of absorbance). At this stage uncoated tubes from Ringsdorf were submitted for trial. The sensitivity was intermediate between those of the pyrolytically coated tubes (manufacturers’ characteristic concentration 3 pg) and the uncoated ones evaluated earlier.Although some batches required up to 50 firings before the tube blank was reduced to an acceptably low absorbance value rectilinear graphs of absorbance versus concentration were obtained without difficulty for both the Ringsdorf and American un-coated tubes. Sample introduction into the furnace required care and reproducible absorb-ance values obtained only if the pipette tip was not allowed to touch the sides of the loading hole or the back of the furnace tube. Accidental contact between the tip and any part of the tube often gave a spurious peaked response on the video display of the spectrometer and unacceptably high absorbance values.Provided these precautions were followed manual injection gave a precision of 2 4 % . The absolute limit of detection was 20 pg (signal to noise ratio = 1 1) and the characteristic concentration was 10 pg. An average furnace tube could be used for 400 firings before the performance deteriorated appreciably. 0 3.0 Time/s Fig. 1. Variation in peak shape with (1) a signal from an iron standard and (2) spurious iron peak from a graphite tube. Apparatus An IL Video I Model 551 atomic-absorption spectrometer fitted with an IL 555 controlled-temperature furnace was utilised for most of the work. The atomic-absorption signal is dis-played on a video screen and the background due to non-atomic absorption could also be dis-played simultaneously.The video display was essential for differentiating between good and bad sample injections and for detecting anomalous furnace behaviour as described earlier. It is unlikely that these small differences would have been apparent with an instrument fitted with recorder output or digital display only. Some recent measurements in particular the analysis of acids were made with a Perkin-Elmer 3030 spectrometer HGA 500 furnace and AS-40 autosampler which also features a clear video display of the atomic-absorption signal. Further details of experimental con-ditions are summarised in Table I. Reagent Purification High-pwity water Filtered mains water was fed through a reverse osmosis cartridge on to an activated carbon scavenger and then High-purity water was supplied by an Elga Spectrum RO-1 System 1452 STEWART AND NEWTON DETERMINATION OF IRON IN TABLE I Analyst Vo2.108 INSTRUMENTAL AND EXPERIMENTAL OPERATING CONDITIONS Furnace tubes .. Furnace programme (Fig. 2) . . Analytical line . . Mode Purge gas . . . I Sample introduction . . Calibrated flasks . . Beakers . . . . . . Working environment. . . . Single-piece Ringsdorf uncoated (M.C.P. Electronics Ltd. Inter-metallics Division Alperton Middlesex) Temperaturel'C 80 140 600 780 2400 Step 1 2 3 4 5 Ramp time/s . . . . 20 25 15 20 0 Hold time/s . . 0 0 0 0 5 A maximum ash temperature of 780 "C was selected to prevent any possible loss of iron as either iron(I1) or iron(II1) fluorides (melting points >lo00 "C) 248.3 nm.Atomic-absorption minus background Argon. Flow-rate was reduced to 400 1 h-l to increase the residence time of the free atoms within the furnace tube (no gas-stop facility is provided on the IL 555) An Eppendorf 10-pl pipette with disposable tips was used to introduce 10-pl volumes of solution into the graphite furnace. All pipette tips were cleaned in redistilled nitric acid and high-purity water before use Quartz stoppered flasks capacity 10 cm3 were used to contain the final sample solution before analysis. Dilution to an appropriate value was carried out by mass (Sartorius top pan balance). Before use the flasks were filled with water and the water tested (25-p1 injec-tions) to ensure that they were free from iron (absorbance <0.015 A) Dissolution of the silicon took place in Nalgene PTFE - FEP beakers, 25-cm3 capacity (Techmate Ltd.Luton) . Before an actual analysis several dummy blank determinations were run to ensure that the beakers were free from iron Spectral band width 0.3 nm All sample preparations were carried out in laminar flow clean air cabinets. Retort stands and other supports were constructed of PVC and PTFE to minimise iron contamination through a nuclear-grade ion-exchange cartridge. The water was finally collected in a poly-propylene tank fitted with a 0.2-pm bacteria filter and a small pump was used to allow con-tinuous recirculation at 0.5 1 min-1 through the carbon and ion-exchange cartridges. The system has provided water with an iron level of <0.02 ng ml-l for several months.The function of this system is to deposit a liquid sample in aerosol form on to the furnace tube which is pre-heated to 150 "C and consists of a modified nebuliser and pre-mix system a transfer tube injector tip, vacuum line and drain trap. The sample aerosol which is formed by the nebuliser passes through the vacuum line when a timed pinch clamp opens and closes the line and thus deposits The iron content was measured with an IL 254 Fastac autosampler. 2 000 9 2 1500 z 3 c g 1000 E + 500 20 40 60 80 100 120 140 160 Ti m e/s Fig. 2. Furnace temperature programme December 1983 SEMICONDUCTOR-GRADE SILICON BY FURNACE AAS 1453 the aerosol on to the heated tube. The deposition time may be varied to increase or decrease sensitivity and is so arranged that a time of 1 s would deposit an amount of aerosol approxi-mately equivalent to 1 p1 of solution.Because the aerosol dries almost instantaneously, volumes of solution up to 1000 p1 may be deposited giving a much higher sensitivity than is possible with a conventional autosampler. Once the aerosol has been deposited the normal temperature cycle of the furnace begins. A deposition time of 100 s gave a limit of detec-tion of <0.02 ng ml-l of iron. High-purity acids Semiconductor-grade acids contain significant concentrations of iron and hence need to be further purified by distillation. The iron content of commercially available acids is variable and it may be necessary to select batches from several manufacturers before a satisfactory source is located (Table 11).Initially hydrofluoric and nitric acids were singly distilled in a small platinum still of 60 cm3 capacity. Evaporation of 5 cm3 of hydrofluoric acid and 4 cm3 of nitric acid in a PTFE basin gave a reagent blank of 30 ng of iron (Table 111). If both acids TABLE I1 ANALYSIS OF ACIDS BEFORE PURIFICATION Acid Source Batch Iron contenting ml-l Hydrofluoric . . . . BDH Chemicals A 108 Aristar (40%) B 40 Hydrofluoric . . . Carlo Erba A 228 Erbatron (50 yo) * Selectipur (50%)* Nitric . . . . Carlo Erba A 16 Erbatron (70%) B 16 Hydrofluoric . . . . Merck MOS A 18 * Diluted to 40% hydrofluoric acid. This may then be distilled in a sub-boiling still without the forma-tion of bubbles which could affect the purity of the distillatess TABLE I11 Acid 50% Merck MOS HF; 70% Carlo Erba HNO, 60% Merck MOS HF; 70% Carlo Erba HNO, 40% BDH Chemicals RSE RSE * .Aristar H F . . 70% Carlo Erba RSE HNO . . 40% BDH Chemicals Aristar H F . . . . 70% Carlo Erba HNO, RSE . . PREPARATION OF REAGENT BLANKS Reagent blank Purification (5 ml H F + 4 ml HNO,) Evaporation Single distillation (platinum still) Two successive distillations (platinum still) Two successive distillations 1 (platinum still) One distillation (quartz still) then second distillation (platinum still) i Sub-boiling distillation (PTFE -FEP) then second distillation (platinum still) One distillation (quartz still) then second distillation (platinum still) 30 ng of iron PTFE basin 12 ng of iron Platinum basin 2-5 ng of iron Platinum basin 1 ng of iron PTFE - FEP beake 1454 STEWART AND NEWTON DETERMINATION OF IRON IN Analyst V d .108 were distilled twice and evaporation took place in a platinum basin the blank was reduced to 12 ng of iron. At this stage requirements were for a more plentiful supply of purified acids, together with an appreciably lower reagent blank. Larger volumes of nitric acid were assured by distillation in a simple quartz still of 250 cm3 capacity. Before use the distillate was re-distilled in the platinum still. This in conjunction with a change to hydrofluoric acid of a higher purity specification gave an over-all reagent blank of 2-5 ng. The production of high-purity hydrofluoric acid in large amounts was more difficult. Eventually a sub-boiling distillation unit of PTFE - FEP was constructed as described by Mattin~on.~ With this apparatus 600 cm3 could be distilled in a single batch over a period of 6 d with only a minimum of attention.In general the magnitude of the reagent blank depends upon four factors the purity of the acids initially; the distillation process; the nature of the vessel in which evaporation takes place; and also the time that the purified reagents have been standing before use. If all four factors are optimised including a distillation of reagents immediately before use then a reagent blank of 1 ng may be attained (Table 111). A complete analysis of one particular batch of reagents is listed in Table IV (1-6). The best Sample 1 2 3 4 5 6 7 8 TABLE IV ANALYSIS OF ACIDS AFTER PURIFICATION Starting material BDH Chemicals Aristar 40% hydrofluoric BDH Chemicais Aris’tar 40% hydrofluoric Hydrofluoric acid purified by sub-boiling Carlo Erba Erbatron 70% nitric acid .. Carlo Erba Erbatron 70% nitric acid . . Nitric acid purified by distillation in quartz acid . . acid . . distillation as in (2) as in (5) Merck MOS Selectipur 50%* . . Merck MOS Seiectip;; 50%* . . Iron content/ Purification ng ml-l Used as received 108 Sub-boiling distillation in a PTFE -FEP apparatus5 2 Distillation in a platinum still 0.2 1.5 0.7 0.36 0.34 0.33 Used as received 16 Distillation in a quartz still Distillation in a platinum still Used as received 18 Sub-boiling distillation as in (2) * Diluted to 40% hydrofluoric acid.This may then be distilled in a sub-boiling still without the forma-tion of bubbles which could affect the purity of the di~tillate.~ blank recorded with these reagents was 3 ng. Analysis of another batch of hydrofluoric acid in which the starting material was of higher purity is also included (Table IV 7 and 8). The polypropylene bottles tested originally were suitable only for short-term storage because they slowly released iron into the acids. Purified acids were stored in PTFE - FEP bottles. Surface Analysis of Slices Many workers in the field of silicon technology have employed anodic oxidation as a means of studying the depth - concentration profile of dopant elements.6~7 Thus Tsujii and Kitamme* measured the profile of arsenic in silicon by growing a silicon oxide layer anodically and then dissolving the oxide in dilute hydrofluoric acid.The arsenic in solution was then determined by atomic-fluorescence spectroscopy. In this study it was necessary to measure surface contamination of the slices in an “as received state,” and consequently contact with mounting wax silver paint or metallic elec-trodes as needed for some anodic oxidation procedures was best avoided. HillQ had also measured diffusion profiles of phosphorus in silicon using a radionuclide technique and in this instance sectioning was achieved by chemical etching in nitric acid to which a very small amount of hydrofluoric acid had been added. A similar technique was reported by Vasilevs-kaya and Mal’kov.lo This was studied in more detail and it was found that thicknesses of 60-190 nm could be dissolved by varying the ratio of the acids in the etch (Table V).Layer December 1983 THICKNESS SEMICONDUCTOR-GRADE SILICON BY FURNACE AAS TABLE V OF SILICON REMOVED AS A FUNCTION OF HYDROFLUORIC ACID TO NITRIC ACID RATIO Both acids were doubly distilled. 1455 Hydrofluoric acid Nitric acid Etching time/ Thickness removed/ volumes volumes min nm 1 350 1 20 1 350 3 64 1 230 3 109 1 175 3 125 1 140 3 190 of only 20 nm could be analysed by reducing the etching time from 3 to 1 min. The actual thickness removed which was dependent to some extent on the source of the silicon was checked by weighing or Talysurf measurements. The form of the apparatus is shown in Fig. 3. The slice (area 5-8 cm2) is held by suction to a quartz tube holder and the assembly lowered into the etchant until the slice just touches the surface of the liquid.When the appropriate time has elapsed the slice is raised clear of the liquid and the etchant analysed for iron. This may be repeated after reversing the slice enabling individual measurements to be made of the iron contamination of both the polished (front) and unpolished (rear) surfaces of the slice. To vacuum pump 4 Quartz tube Silicone slice m- PTFE FEPdish 1 Etchant J I Labjack I Fig. 3. Apparatus for surface analysis. Sample Preparation Before Analysis Two separate techniques have been studied. (i) Etching with a hydrofluoric acid - nitric acid mixture to remove up to 60% of the original sample mass.This gave a result for iron content in the bulk material after removal of all possible surface contamination. (ii) Rinsing with propan-2-01 or high-purity water. This gave a measure of the silicon slices in an “as received” state. Method Bulk artalysis A sample of silicon (0.3 g) which has either been etched in hydrofluoric acid - nitric acid or rinsed in propanol-2-01 or water is weighed into a PTFE - FEP beaker then 5 ml of purified hydrofluoric acid are added followed by 4 ml of purified nitric acid in small portions. When the initial vigorous reaction has subsided the solution is evaporated just to dryness on a low-temperature quartz hot-plate with auxiliary heating from an infrared lamp. The residue is taken up in 160 pl of water 25 p1 of purified nitric acid and diluted (by mass) to 500 p1 in a quartz flask.The final volume may need to be adjusted depending upon the iron concentra-tion and the analysis is completed by standard additions as follows. Three 10-p1 aliquots are injected into the graphite furnace. The solution remaining is then “spiked” with a standar 1456 STEWART AND NEWTON DETERMINATION OF IRON IN Analyst VoZ. 108 TABLE VI BULK ANALYSIS OF SILICON Manufacturer Form Pre-treatment Iron contenting g-l A B batch 1 . . 2 3 4 5 6 C batch 1 . . 2 3 4 D batch 1 2 3 4 5 batch 1 . . 2 improved process E batch 1 . . 2 . . * . Slice Slice Slice Polycrystalline Slice Slice Slice 1 2 Slice Slice E* RRt R R R R E (33%) R R R E (63%) E (31%) E (43%) E (62%) E (62%) R R R R R R 640 670 17 50 77 5 35 5 5 30 31 17 15 116 127 200 23 6 193 112 6 20 350 110 * E Sample is etched with hydrofluoric acid - nitric acid.t R Sample is rinsed in propan-2-01 or water only. Results in parentheses represent the percentage of original sample mass removed during etching before analysis. solution to give an addition of 15 ng of iron (30 pl of 0.5 p.p.m. iron). A further three 10-pl aliquots are injected into the furnace and this procedure is repeated for a second standard addit ion. With this method a volume correction must be applied but this is preferable to dividing the solution into three portions with the possibility of contamination.A reagent blank is run in the same PTFE - FEP beaker prior to the analysis of the silicon. Results of the analysis are shown in Table VI and standard additions graphs for a typical sample and blank are given in Fig. 4. This may be attributed to the presence of residual siliceous material in the sample that has not been entirely volatilised by the hydrofluoric acid - nitric acid mixture. On occasion the graphs for sample and blank may have different slopes. 30 15 0 15 30 Iron/ng Fig. 4. Typical standard additions graphs for (1) sample and (2) reagent blank December 1983 SEMICONDUCTOR-GRADE SILICON BY FURNACE AAS 1457 Surface analysis 1-3 min in 2 ml of special etchant (Table V) in a PTFE - FEP beaker.are needed on front and rear surfaces the apparatus (Fig. 3) is used. analysed for iron as described under B d k analysis. cal slopes as the amount of siliceous residue remaining on evaporation is negligible. A silicon slice (area 5-8 cm2) that has been rinsed in high-purity water is etched for If separate results The etchant is then In this instance standard additions graphs for both sample and blank have virtually identi-Results The results from the bulk analysis of silicon are shown in Table VI. The results shown in Table VII represent an initial experiment in which a silicon slice was etched successively in five 2-ml portions of etchant [hydrofluoric acid - nitric acid (1 + 350, V / V ) ] and the individual fractions were analysed for iron. Both sides of the slice were exposed to the etch but it can be seen that the bulk of the iron is located in the first 100 nm of the surface.In later work both sides of the slice were analysed successively and the etch times and acid concentrations adjusted to dissolve 50-100 nm'bf the surface ik-a single step (Table VIII). TABLE VII PROFILING SURFACE ANALYSIS OF SILICON Total thickness removed/nm Iron found/ng Surface area/cm2 Iron/ng cm-2 22 34 15.6 2.2 44 31 15.6 2.0 66 8 15.6 0.5 88 6 15.6 0.4 110 3 15.6 0.2 Reproducibility Reagent blanks The following results for iron content were obtained (1) 2.9 2.7 and 2.9 ng i.e. 2.8 & 0.1 ng; and (2) 1.9 2.3 ng, Five blanks were run on the same day in two different beakers. i.e. 2.1 & 0.2 ng. Manufacturer B D (improved production) TABLE VIII SI NGLE-STEP s u RFACE AN ALY s I s Area/ Thickness Iron content/ Iron content/ Pre-treatment cm2 Surface removed/nm ng cm-2 g-l w* 7.8 Polished 52 <0.12 < 10 Rough 52 0.12 10 w 5.8 Polished 61 <0.17 < 12 Rough 61 0.17 12 w 7.5 Polished 60 9.9t 717 Rough 60 0.4 29 w 7.3 Polished 59 0.48 35 Rough 59 0.14 10 W 8.9 Polished 60 0.12 9 Rough 60 0.12 9 c .. w 8.7 Polished 52 0.40 33 Rough 52 0.12 10 w 8.1 Polished 62 0.62 43 Rough 62 0.25 18 * W Slice rinsed in high-purity water. t Slice shown to be contaminated during processing 1458 STEWART AND NEWTON Silicon. slices Duplicate values obtained from adjacent areas gave results of 6 and 5 ng g-l. Further samples were taken from points 2-5 cm from the original sampling area and gave values of 3 21 and 12 ng g-1.It is likely that these differences are real and investigational work is in progress to study the variation in iron concentration across the slice. Samples were taken from the centre of a slice from manufacturer C. Discussion The outlined method has provided a suitable means for the measurement of iron in semi-conductor-grade silicon slices down to an iron content of 5 ng g-l and to similar levels in poly-crystalline material. The level of iron contamination has been found to vary from 5 to 650 ng g-l dependent upon manufacturer. It has been shown by Wardl that silicon slices with an iron content of less than 20 ng g-l gave high yields when used for device fabrication. Surface analysis with a chemical etch has indicated iron levels of micrograms per gram penetrating the first 100 nm of surface i.e.1000-fold higher than in the bulk material. In almost all instances higher concentrations were detected on the polished side compared with the rough side of the slice. This could indicate contamination by polishing debris or slurry as suggested by Schmidt and Pearce.ll Further work will be directed towards lowering the limit of detection for both surface and bulk analysis. We thank P. J. Ward [Plessey Research (Caswell) Ltd.] for initiating this work and for many helpful discussions and the Directors of Plessey Research (Caswell) Ltd. for permission to publish this paper. This project was carried out with the support of the Procurement Executive Ministry of Defence sponsored by DCVD. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. References Ward P. J. J . Electrochem. Soc. 1982 129 2573. Hoffmeister W. 2. Anal. Chem. 1978 290 289. Volland G. Kolblin G. Tschopel P. and Tolg G. 2. Anal. Chem. 1977 284 1. Fuller C. W. Anal. Chim. Acta 1972 62 261. Mattinson J. M. Anal. Chem. 1972 44 1715. Lanza P. and Buldini P. L. Anal. Chim. Acta 1979 104 139. Buldini P. L. Ferri D. and Lanza P. Anal. Chim. Acta 1979 106 137. Tsujii K. and Kitazume E. Anal. Chim. Acta 1981 125 101. Hill C. Plessey Research (Caswell) Ltd. personal communication. Vasilevskaya L. S. and Mal’kov N. V. Zavod. Lab. 1971 37 1044. Schmidt P. F. and Pearce C. W. J . Electrochem. Soc. 1981 128 630. Received May 3rd 1983 Accepted July 22nd 198

ISSN:0003-2654    

DOI:10.1039/AN9830801450

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst December 1983 Vol. 108 PP. 1459-1465 1459 Furnace Atomisation with Non-thermal Excitation- Experimental Evaluation of Detection Based on a High-resolution Echelle Monochromator Incorporating Automatic Background Correction H. Falk E. Hoffmann and Ch. Ludke Central Institute for Optics and Spectroscopy Academy of Sciences Rudower Chaussee 5 1199 Berlin GDR John M. Ottaway and S. K. Giri Department of Pure and Applied Chemistry University of Strathclyde Cathedral Street Glasgow G1 1XL The analytical potential that could be derived from the combination of a FANES source with a high-resolution wavelength-modulated kchelle mono-chromator has been investigated. Detection limits for elements of high excitation potential are improved using the non-thermal excitation process compared with the thermal excitation processes available in carbon furnace atomic-emission spectrometry.For some other elements of lower excitation potential detection limits are impaired owing to the increased complexity of the background spectrum in the FANES source. Keywords ; A tomic-emission spectrometry ; bchelle monochromator ; electro-thermal atomisation ; low-pressure discharge ; wavelength modulation Recently a new excitation source has been described for use in atomic-emission spectrometry, which couples high excitation energy with an efficient atomisation process. The method has been termed FANES,1-3 or furnace atomic non-thermal excitation spectrometry and involves conventional electrothermal atomisation of samples in a tube atomiser in which a low-pressure gas discharge is simultaneously generated using the graphite tube itself as the cathode.The combination of the highly efficient electrothermal atomisation system with the low-noise excitation dischargeJ4 offers high sensitivity analysis for a wide range of elements using small sample volumes. The usual attractive features of emission sources e.g. high dynamic range (5-6 orders) and the ease of operation in a simultaneous multi-element mode, have also been demonstrated. Compared with the use of electrothermal atomisation as an atomic-absorption atom cell the possibility of determining elements such as halogens with resonance lines in the vacuum ultraviolet using FANES is also exciting. The measurement of atomic-emission signals during electrothermal atomisation without supplementary excitation has also been extensively studied and recently re~iewed.~ Under these conditions the existence of local thermal equilibrium (LTE) has been demonstrated6 and parts per billion (parts per log) (p.p.b.) detection limits have been achieved for a wide range of elements.Using this technique carbon furnace atomic-emission spectrometry or CFAES the best detection limits reported to date have been achieved with a spectrometric system based on a 0.75 m 6chelle monochromator incorporating automatic background correction.' The dchelle spectrometer offers the high resolution favourable for atomic-emission measurements with a high optical conductance (f/13 aperture). Background correction is achieved using square-wave wavelength modulation generated by means of a rotating quartz chopper with four separate quartz quadrants of different thicknes~es.~~~ The chopper is mounted either near the entrance or exit slits of the monochromator and the modulation frequency of 40 Hz is adequate for most atomic-emission signals and allows efficient background correction to be achieved for both the continuum background from the graphite furnace and matrix scatter signals from for example clinical material^.^ Although very low detection limits have been obtained for many elements,7~10 the determination of volatile elements with high excitation potentials (e.g, cadmium zinc and selenium) a t sub 1460 FALK et at?.FURNACE ATOMISATION WITH NON-THERMAL Analyst VOZ. 108 parts per billion levels has still not been found possible despite the introduction of platform7J0 and probell atomisation techniques.For the high sensitivity detection of these elements by atomic emission during electrothermal atomisation the FANES approach appears to be attractive if not essential. Measurements of FANES reported to date1-3 have been made with a PG S2 grating spectro-graph (VEB Carl Zeiss Jena). This has a two-channel photomultiplier detection system in place of the photographic plate and was operated in the d.c. registration mode without background correction during signal measurement. Under these conditions detection limits are determined by the fluctuation of the background signal at the position of the atomic line being measured. The 6chelle spectrometer system developed for CFAES obviously offers characteristics highly suited to measurements with the FANES excitation source.The high resolution and background-correction system should both improve detection limits of FANES and in addition the range of elements available by CFAES should be extended by the high excitation energy in the FANES source. To examine these possibilities a short-term collaborative project was set up between our laboratories which allowed the transfer of a FANES source to the University of Strathclyde for coupling with the dchelle spectrometer system. The results of this study which allowed some interesting conclusions to be reached, are reported in this paper. Experimental The dchelle spectrometer system is based on a Spectrametrics dchelle monochromator, modified for wavelength modulation background correction.For experiments using wave-length modulation (WM) the instrumental configuration was exactly as described previ-ously.7 To provide comparison with earlier FANES results obtained using a d.c. system, measurements were also made using intensity modulation (IM). In this instance a rotating chopper disc was mounted between the FANES source and the spectrometer entrance slit. A square-wave modulation waveform with a frequency of 130 Hz was generated and the chopper incorporated a reference signal that was used to synchronise the lock-in amplifier. Hollow-cathode lamps were used for wavelength adjustment by focusing the lamp through the FANES source on to the entrance slit of the spectrometer. Hollow-cathode lamps were also used to measure the linearity range of the detection system and to confirm the linear relationship between the slit aperture (entrance slit width = exit slit width) and the signal amplitude.A schematic diagram of the cross-section of the FANES source used in this study is shown in Fig. 1 which shows only minor modifications to that described earlier.2 This FANES furnace tube is identical in size with that used in the Perkin-Elmer HGA 500 heated graphite atomiser ( i e . 28 mm long and 5.9 mm i.d.) and is connected to electrical power supplies for heating the furnace tube (Uh in Fig. 1) and for exciting the low-pressure gas discharge. In the latter the graphite tube acts as the cathode and a separate anode is introduced as shown in Fig. 1. A mechanical pump is used to reduce the pressure of inert gas (helium or argon) to 1-5 Torr during the discharge process.The discharge current was 30 mA at a voltage of 600 V. The operation of the FANES source is analogous to that used in conventional electro-thermal atomisation. A 20-pl sample aliquot is injected into the furnace and dried and/or ashed as required at atmospheric pressure. During this time the cover over the sample injection hole is removed to allow the released vapours to be cleared from the furnace. The furnace chamber is then sealed and pumped down to 3 Torr of helium (unless otherwise mentioned) and the low-pressure discharge is initiated. This latter process takes approxi-mately 30 s. The atomisation stage is then initiated using the optimised atomisation temperature.The transient atomic-emission signals were recorded on a Servoscribe RE 541 20 strip-chart recorder. After allowing the furnace to cool to ambient temperature, the pressure is raised to atmospheric pressure and the cover removed for introduction of the subsequent sample. The FANES furnace chamber and power supplies are all home built in the laboratories of the Central Institute for Optics and Spectroscopy but are based on con-ventional electrothermal atomisation and low-pressure discharge generating circuits. The heating rate of the electrothermal atomiser can reach a maximum of 2000 "C s-l December 1983 EXCITATION-EXPERIMENTAL EVALUATION OF DETECTION 1461 D Fig. 1. Cross-section of the FANES source. Ua Anode; Uh connections for electrothermal heating of the graphite tube; A graphite tube; B graphite con-tact cylinders ; C removable lid for sample injection; D quartz windows; E rotation arm for changing the graphite tube.The furnace is also water cooled (not shown) to allow rapid introduction of successive samples. Results and Discussion In the investigations reported a range of elements were selected to provide a comparison with detection limits achieved previously with FANES (chromium silver and lead) and CFAES (chromium silver lead zinc and cadmium). Some elements were also chosen with high excitation potentials i.e. with resonance lines in the low ultraviolet wavelengths (zinc, cadmium and selenium) to provide information on potential improvements compared with CFAES and the over-all limitations of the present system.All analyte solutions were pre-pared from AnalaR reagents with 1 0 - 2 ~ nitric acid added and dilution with high-purity distilled water. The effect of sample volume is illustrated in Fig. 2 for cadmium in a helium atmosphere and is typical of electrothermal atomisation. Signal response increases in a more or less linear fashion up to 4 0 ~ 1 after which the signal and reproducibility both deteriorate owing to the greater spreading of the sample in the graphite tube. 20 40 60 80 100 Sample volume/pI Fig. 2. Effect of sample volume on the FANES signals Carrier for 100 pg 1-1 cadmium solution a t 228.8 nm. gas helium; discharge current 20 mA 1462 FALK et a,!. FURNACE ATOMISATION WITH NON-THERMAL Analyst VOZ. 108 Results for the instrument system examined in this study and those for CFAES and previous FANES studies are given in Table I.Detection limits in this work were calculated as the concentration equivalent to three times the standard deviation of the background noise. These results allow the following conclusions to be drawn. 1. The use of the 6chelle spectrometer system with wavelength modulation does not lead to the anticipated dramatic improvement in FANES detection limits. A factor of 2-3 at the most was achieved for silver and chromium. 2. Wavelength modulation does not give an improvement in detection limits compared with intensity modulation under identical conditions of resolution (see cadmium chromium, zinc and lead where WM results are worse than IM). 3. A spectral resolution of 10-20 x lo3 appears to be sufficient for FANES in order to achieve the best detection limits.4. Whilst significant improvements in detection limits compared with CFAES are achieved for cadmium selenium and zinc detection limits are actually poorer than those reported for CFAES for chromium. The results were found to be entirely related to the nature of the background signal generated in the FANES source. Despite the use of the automatic background correction device a small residual background signal was generated in all instances as indicated in TABLE I DETECTION LIMITS ACHIEVED WITH THE FANES ~CHELLE SPECTROMETER SYSTEM USING INTENSITY MODULATION' (IM) OR WAVELENGTH MODULATION (WM) Element Silver. . Cadmium Chromium Zinc . . Lead Lead Selenium a .Wavelength/ nm 328.7 228.8 425.4 213.9 405.7 283.8 196.0 Measure-ment mode IM IM WM WM IM IM IM IM WM WM WM WM IM IM IM IM IM WM IM W M IM WM IM IM IM WM WM WM IM R* x 103 15 33 15 33 6.5 16 25 33 16 25 33 6.5 6.5 16 31 45 27 16 16 16 7 17 16 31 40 16 31 40 16 Dt 0.38 0.11 0.38 0.18 1 .a 0.38 0.18 0.07 1.0 0.38 0.18 0.07 1 .o 0.38 0.18 0.07 0.02 0.38 0.38 0.38 0.4 0.38 0.38 0.11 0.04 0.38 0.11 0.04 0.38 Detection limit /pg IBSl mV 1.5 0.8 co.1 0.1 1 .o 5.9 0.3 0.06 1 .o 0.24 0.1 0.02 23 8 3.5 1.1 0.7 <0.02 1.1 0.05 44 0.18 1.64 0.19 0.10 0.5 0.2 0.1 0.05 This work 0.4 2.3 1.1 2.6 1.1 1.1 2.4 10.2 4.1 3.4 3.5 9.1 46 7.5 6.2 9.6 9.2 4.0 5.3 14.2 17.4 12 56 60 20 30 55 800 67 FANES previous work 1.03 CFAES 2.610 3OO1l 10 (357.9 nm)2 --14 (368.0 nm)* 1.21' 2 400" 46;" * R Practical resolving power of the spectrometer.t D Relative optical conductance of the spectrometer. $ IB Relative intensity of background December 1983 EXCITATION-EXPERIMENTAL EVALUATION OF DETECTION 1463 Table I and Figs. 3 and 4. It is clear that wavelength modulation substantially reduces the background signal. The rise in the signal when the low-pressure discharge is switched on is equivalent to the background signal from the FANES source under IM conditions and this is substantially smaller under WM conditions.The detector noise was only comparable to the source background intensity for the narrowest slit widths used. The FANES source back-ground signal was measured independently and was found to consist mainly of a multi-line spectrum. The background intensity as a function of slit width could be described by a power law the exponent of which varied between 1 and 2 depending on the wavelength used. D D C J -t U t Time + Fig. 3. Background and analyte signals measured a t the cadmium (228.8 nm) line for 10 pl of (a) A Background a t (b) A Background a t 20 p g 1-' cadmium solution using (a) wavelength and (b) intensity modulation. 0.2 mV; B FANES switched off; C background a t 0.4 mV; D analyte peaks.0.1 mV chart recorder voltage; B FANES switched off; C background a t 1 mV; D analyte peaks. From these measurements and the observation of a considerable enhancement of the back-ground signal when small amounts of air leaked into the carrier gas it was concluded that the background consists mainly of molecular bands of the plasma gas. These bands con-tained spectral ranges with continuous as well as line characteristics. It is clear from this information why WM does not give improved detection limits compared with IM. The back-ground intensities at the analyte line and background measurement wavelengths selected by the wavelength modulation device are not equal owing to the complex structured background of the FANES source.Consequently detection limits remain dependent on the signal to background ratio rather than the signal to noise ratio. Although wavelength modulation did not allow exact automatic background correction to be achieved a manual correction could easily be made using the measurements illustrated in Figs. 3 and 4. The background (IM) or residual background (WM) from the source can be measured when the gas discharge is in operation and before the sample atomisation sequence is started. Such a procedure would however be inadequate for structured background signals from the analyte matrix. When argon was used as the carrier gas instead of helium the background intensity was remarkably higher at all analyte wavelengths used whereas comparable analyte intensities were obtained 1464 FALK et al.FURNACE ATOMISATION WITH NON-THERMAL - T m rn E O iij t- 7-T F1 m m I C D \ C -f Analyst Vol. 108 Time -b Fig. 4. Background and analyte signals measured at the lead 405.8-nm line for 20 pl of 50 pg 1-1 lead solution using (a) wave-length and (b) intensity modulation. (a) A Background at 5 mV (note it is negative because the FANES background is larger at the analyte wavelength than at the one used for background correction) ; B FANES source off; C background a t 25 mV; D analyte peaks. (b) A Background at 2.5 mV; B FANES source off; C background at 10 mV; D analyte peaks. The best detection limits for cadmium and zinc of 0.03 and 0.1 pg l-l respectively (assuming a volume of 40p1) indicate that the non-thermal excitation process of FANES does allow as p r e d i ~ t e d ~ ~ ~ a higher population of energy levels with larger excitation potentials than the thermal excitation mechanism available in CFAES.These values are commensurate with electrothermal atomisation atomic-absorption spectrometric (ETA-AAS) detection limits and indicate that FANES combined with CFAES would probably be competitive with current ETA-AAS systems. The relatively high detection limit for selenium of 40 pg 1-1 is explained by the very low transmittance of the optical system at the 196.0-nm wavelength. For chromium the FANES detection limit is significantly poorer than the CFAES detection limits and this would also be expected for other less volatile elements with relatively low excitation potentials owing to the greatly increased structured background from the FANES source.Whilst very high resolution does not appear to produce lower FANES detection limits it may be useful with particular sample types to overcome spectral interferences. Generally, wavelength modulation would be more useful if the positions used for background measure-ment can be chosen to give a signal more exactly correlated with the background intensity at the analyte line. With this system this is difficult and could only be achieved by alteration of the angle of incidence of the light beam at the rotating chopper. A more easily adjustable wavelength modulation device would be useful but might be limiting if matrix spectral interferences varied from sample to sample. In order to improve the performance of FANES and obtain lower detection limits and improved background correction it would seem preferable to attempt to reduce the complexity and magnitude of the background itself by using a more perfect vacuum system and a purer gas supply.This work was made possible by the Cultural Exchange Agreement between the Royal Society of the UK and the Academy of Sciences of the GDR and the authors are very gratefu December 1983 EXCITATION-EXPERIMENTAL EVALUATION OF DETECTION 1465 for the opportunity for collaborative study and the financial support provided through this scheme. Financial support from the SRC for the purchase of the kchelle spectrometer and from the British Council (for S.K.G.) is also gratefully acknowledged. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. References Falk H. Hoffmann E. Jaeckel I. and Ludke Ch. Spectrochim. Acta Part B 1979 34 333. Falk H. Hoffmann E. and Ludke Ch. Spectrochim. Acta Part B 1981 36 767. Falk H. Hoffmann E. and Ludke Ch. Fresenius 2. Anal. Chem. 1981 307 362. Falk H. Spectrochim. Acta Part B 1977 32 437. Ottaway J. M. Hutton R. C. Littlejohn D. and Shaw F. Wiss. 2. Karl-Mum Univ. Leipzig, Littlejohn D. and Ottaway J. M. Analyst 1979 104 208. Ottaway J . M. Bezur L. and Marshall J. Analyst 1980 105 1130. Michel R. G. Sneddon J. Hunter J. K. Ottaway J. M. and Fell G. S. Analyst 1981 106 288. Ottaway J. M. Bezur L. Fakhrul-Aldeen R. Frech W. and Marshall J. in Bratter P. and Schramel P. Editors “Trace Element Analytical Chemistry in Medicine and Biology,” Walter de Gruyter Berlin 1980 p. 575. Bezur L. Marshall J. Ottaway J. M. and Fakhrul-Aldeen R. Analyst 1983 108 553. Giri S. K. Littlejohn D. and Ottaway J. M. Analyst 1982 107 1095. 1979 28 357. Received February 25th 1983 Accepted July 29th 198

ISSN:0003-2654    

DOI:10.1039/AN9830801459

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

1466 Artalyst, December, 1983, Vol. 108, pp. 1466-1170 Determination of Manganese in Precipitated Calcium Carbonate Samples Using Candol uminescence Spectrophotometry: Some Problems Associated with Preparation of Standards (The late) Ronald Belcher, E. Roy Clark, Malcolm H. Lloyd" and H el ena P uza nows h a -Tarasi ew i cz Department of Chemistry, University of Aston in Birmingham, Gosta Green, Birmingham, B4 VET The intense yellow candoluminescence given by manganese in a calcium oxide - calcium sulphate matrix when placed in a hydrogen - nitrogen - air flame has been used to investigate the determination of manganese (in the range 0.2-0.6 p.p.m.) in commercial samples of precipitated calcium carbonate. The results obtained by the candoluminescence emission technique have been compared with those obtained by atomic-absorption spectrophotometry and a kinetic spectrophotometric method.Keywords Manganese determination ; calcium carbonate ; candoluminescence spectrophotometry ; standard firefiaration Since the publication of the review on candoluminescence spectrophotometry by Belcher et al.1 in 1978 in which the history and practical applications of the technique were discussed very few other papers have been published. One by Sokolov2 deals with semiconductor applications and another by Dhaher and Kassir3 describes a new technique in which the matrix material of calcium oxide and calcium sulphate is fabricated into the form of a rod, which is then intro- duced into the hydrogen - nitrogen flame. This modification was used for the quantitative determination of manganese, bismuth and antimony in solutions.The use of a calcium oxide - calcium sulphate matrix in candoluminescence emissions has received much a t t e n t i ~ n ~ - ~ and the main parameters that determine reproducibility have been thoroughly investigated. The control of variables such as flame conditions, gas flow-rates and positioning of the matrix is relatively easy but the reproducibility of the matrix preparation is more difficult. Nichols et a1.' investigated manganese candoluminescence in calcium com- pounds but Belcher et aL5 were the first to develop a quantitative method for the determination of manganese over the range 0.1-3.0 ng in a procedure involving the application of 1 pl of the solution containing the activator to the centre of the matrix surface when a reproducibility of between 5 and 10% was obtained. In the surface application technique first described by Belcher et aL4 for the determination of bismuth, the moisture content and porosity of the matrix are critical because candoluminescence depends upon the concentration of the activator at the surface of the matrix exposed to the flame.If the absorption of a given volume of activator solution is not constant for a set of matrices then candoluminescence emission intensities will vary and this will result in poor reproducibility. Many of the applications of candoluminescence published so far have involved this surface technique and are thus subject to these problems, but one application that could avoid this step in the analysis is the determination of manganese in calcium carbonate, a compound that can be directly converted to calcium oxide.The oxide could then be converted into the calcium oxide - calcium sulphate matrix offering a simple procedure using no solutions and only one reagent, i.e., calcium sulphate, thus minimising contamination. The only problem is the production of suitable standards in which the manganese is uniformly distributed. Two approaches are possible. Firstly, standards could be produced by precipitation methods, or secondly, calibration could be made by another method. Both of these possibilities were investigated and are reported in this paper. The preparation of standards was investigated using two procedures and the results obtained are compared with those obtained by atomic- absorption spectrophotometry and a kinetic spectrophotometric method.8 * Present address : Ferro (Great Britain) Ltd., Wornbourne, Wolverhampton.BELCHER, CLARK, LLOYD AND PUZANOWSHA-TARASIEWICZ 1467 Experimental and Results Apparatus The candoluminescence emission intensity of manganese was measured at 580 nm using a Pye Unicam SP 900 flame spectrophotometer (slit width 0.15 mm) fitted with a Meker-type burner and a Pye-Unicam AR 55 chart recorder. The matrix was inlaid into the hexagonal aperture (3 mm deep, hexagonal sides 2 mm) in the head of an Allen screw and the matrix holder was positioned in the flame in a similar way to that reported previou~ly.~,5 A metal cover placed over the matrix holder ensured that no extraneous light reached the spectro- photometer. Gas flows of 2.0,2.0 and 7.25 1 min-l for hydrogen, air and nitrogen, respectively, were controlled by needle valves and monitored using RS Series “Meterate” flow tubes RS2/C, RSS/R and RSS/C (Glass Precision Engineering Ltd.). The burner assembly was as described previously and the horizontal position of the matrix near the edge of the flame was adjttsted until maximum candoluminescence intensity was obtained.This position was kept constant throughout, as was the height of the matrix surface above the burner (1.6 cm). Reagents Calcium oxide used in procedure 1 was prepared from analytical-reagent grade calcium carbonate heated at 800 “C in a muffle furnace with weighing checks to ensure that conversion was complete.Commercial samples of calcium carbonate containing various amounts of manganese were supplied by John and E. Sturge Ltd., Birmingham. Preparation of Stock Solutions De-ionised water was used for the preparation of all solutions. A solution of 57.383 g of calcium nitrate was prepared by dissolving 82.581 g of analytical-reagent grade Ca(N0,),.4H20 in water and diluting to 100 ml. A solution of 5.556 g of sodium carbonate was prepared by dissolving 15.000 g of analytical-reagent grade Na,CO,. 10H,O in water and diluting to 100 ml. The manganese calibration solutions were prepared by appropriate dilution with de-ionised water of various volumes of solution taken from 100 ml of an atomic-absorption standard solution containing 996 pg ml-1 of manganese in 2% nitric acid (Aldrich Chemical Co.Inc.). A solution of 0.1 p.p.m. manganese, prepared from a 10 p.p.m. manganese solution, was used to prepare the manganese-doped matrices in the coprecipitation procedures. Coprecipitation Procedures Procedwe 1 The first procedure used was an adaptation of the “matrix preparation method” described by Belcher et in which the matrix was prepared from a slurry of calcium oxide - plaster of Paris. The modification involved the addition of manganese solution to the de-ionised water used to prepare the slurry so that in the alkaline conditions manganese was precipitated as hydroxide into the calcium hydroxide - plaster of Paris suspension. (a) A 1.5-g mass of the calcium oxide mixed with 0.187 5 g of plaster of Paris was suspended in 11.25 ml of de-ionised water and the mixture was shaken for about 10 min.The resulting suspension after filtration by vacuum until sufficiently dry was inlaid into a set of weighed Allen screws in the manner previously de~cribed.~ After drying in an oven at 110 “C for 13 min the matrices were stored in a desiccator until ready for use. Before each matrix was introduced into the hydrogen - nitrogen - air flame, an excess of matrix material was carefully scraped off with a clean razor blade so as to leave a clean surface, which was level with the top of the screw. The Allen screw plus matrix was weighed and positioned in the flame using the apparatus described above. (b) The procedure given under (a) was then repeated using 11.25-ml amounts of de-ionised water containing amounts of 0.1 p.p.m.manganese solution such that the resulting sets of matrices based on calcium oxide were equivalent to calcium carbonate samples with manganese contents of 0.2,0.3 and 0.4 p.p.m. An average candoluminescence emission intensity for each set of matrices was calculated (Table I). Procedlure (2) In the second method, manganese carbonate was coprecipitated with calcium carbonate from a manganese-doped calcium nitrate solution using sodium carbonate solution. The filtered The plaster of Paris was of technical grade. The candoluminescence emission was then measured.1468 Manganese concentration, p.p.m. 0 0.2 0.3 0.4 0.6 BELCHER et a,?. : DETERMINATION OF Mn IN CaCO, AyzaZyst, VoZ. 108 TABLE I CANDOLUMINESCENCE EMISSION INTENSITIES n = 8 in all instances.Procedure 1 Procedure 2 A ~r -7 r--------A-------- Intensity* Intensity* Mean of corrected for Mean of corrected for 0.0408 11 - 0.0442 24 - 0.0434 33 22 ( 4 4 ) - 0.044 1 39 28 ( 4 6 ) 0.0468 50 39 ( 5 7 ) matrix mass/g Intensity" blank matrix mass/g Intensity* blank - - 0.042 5 42 18 ( A l l ) 0.042 2 60 36 (It211 - - - - - - * One chart division = 0.25 cm. 95% Confidence limits (chart divisions) are given in parentheses. precipitate was then heated to produce manganese-doped calcium oxide, which was subse- quently used to prepare the matrix with plaster of Paris. (c) A 3.5-g mass of calcium carbonate was obtained by adding 66.80 nil of the sodium car- bonate solution to 10.62 ml of the calcium nitrate solution, filtering off 21.0 ml of water, drying in an air oven at 110 "C for 1 h and then conversion into calcium oxide by heating in a muffle furnace at 800 "C for 4 h.A set of matrices was prepared, and the corresponding candoluminescence emission intensities were measured in the same manner as described under Procedure 1 (a) above. ( d ) Samples of calcium carbonate doped with 0.3 and 0.6 p.p.m. of manganese were similarly prepared by using the required proportions of de-ionised water and 0.1 p.p.m. manganese solution in total volumes of 21 ml. These carbonate samples were used to prepare samples of calcium oxide, which in turn were used to prepare two sets of matrices whose candoluminescence emission intensities were measured in the same manner as described under Procedure 1 (a). The results, together with those for samples containing no manganese [Z(c)], are given in Table I.Candoluminescence Emission Intensities of Commercial Samples of Calcium Carbonate Samples were converted into calcium oxide by heating in a muffle furnace under the same conditions as described under Procedure l(a). The matrices were then prepared using the same amounts of calcium oxide and plaster of Paris and the same procedure as previously described. The candoluminescence emission intensities of four samples containing 0.2, 0.3, 0.4 and 0.6 p.p.m. manganese are given in Table 11. Determination of Manganese by Atomic-absorption Spectrophotometry and a Kinetic Spectrophotometric Method A Perkin-Elmer PRS-10,460 atomic-absorption spectrophotometer was used for the deter- mination of manganese using the method of additions.A Pye Unicam SP8-100 spectrophoto- meter was used for the measurement of absorbance in the kinetic spectropnotometric method. Discussion The results for the determination of manganese in the commercial samples of calcium carbonate using atomic-absorption spectrophotometry and the kinetic spectrophotometric method are given in Table 11. There is good agreement between the two methods and both may be used as the basis for the standardisation of the candoluminescence method. Graphs of average candoluminescence emission intensities, using coprecipitation procedures 1 and 2 (Table I) and those for the commercial samples of calcium carbonate (Table II), are shown in Fig. 1, in which intensities are plotted against the manganese contents of calcium carbonate and the manganese contents of the doping co-precipitation procedures.Corrections are made for the candoluminescence emission intensity of "blank" determinations obtained by extra- polation back to zero manganese content. This emission arises from the manganese content of the plaster of Paris. Purer grades of plaster of Paris or freshly precipitated calcium sulphateDecember, 1983 SAMPLES USING CANDOLUMINESCENCE SPECTROPHOTOMETRY 1469 TABLE I1 CANDOLUMINESCENCE EMISSION INTENSITIES AND MANGANESE CONTENTS OF COMMERCIAL SAMPLES OF CALCIUM CARBONATE BY ATOMIC-ABSORPTION SPECTROPHOTOMETRY AND KINETIC SPECTROPHOTOMETRY For caiidoluinincscence intensity measurements n = 13 ; for both atomic-absorption and kinetic spectrophotometry n = 5. Sample B t Sample C$ Sample Df --7 ~ ---- h T,---------A--------7 r------h----- Sample A* h--- __-__ Matrix mass Intensity? Matrix mass Intensity7 Matrix mass Intensity? Matrix mass Intensity7 Meanlg 0.0415 13 0.0422 15 0.0427 21 0.0443 26 Relative standard deviation, ?& .. . . 3.0 23.1 5.2 13.3 4.0 19.0 4.5 15.4 Mean, after correction for blank - 9 - 11 - 17 - 22 95% confidince li&, . . *2 chart divisions . . . . - * Manganese content: 0.205 p.p.m. f 1.1% (r.s.d.) by AAS; 0.224 p.p.m. f 1.6% by kinetic spectrophotometry. t Manganese content: 0.301 p.p.m. f 1.1% (r.s.d.) by AAS; 0.305 p.p.m. & 1.8% ( r d . ) by kinetic spectrophotometry. 3.3% (r.s.d.) by kinetic spectrophotometry. 5 Manganese content: 0.608 p.p.m. f 2.4% (r.s.d.) by AAS; 0.609 p.p.m. 6 3.2% (r.s.d.) by kinetic spectrophotometry. 7 Candoluminescence intensity, 1 chart division = 0.25 cm.Standard deviationig : : 0.0016 3 0.0022 2 0.0017 4 0.0020 4 - *2 - *I - f 2 Manganese content: 0.405 p.p.m. f 0.8% (r.s,d.) by AAS; 0.416 p.p.m. dihydrate (CaS0,.2H20) subsequently heated to 120-130 “C to give CaSO4.4H,O should produce lower “blank” emissions. The peak heights for emissions obtained for coprecipitation procedures 1 and 2 are both higher than expected. This can possibly be attributed to a non-uniform distribution of manganese throughout the particles of the matrices, the manganese being more concentrated near or on the surface. 0.2 0.4 0.6 Manganese concentration, p.p.rn. Fig. 1. Candoluminescence emission intensities obtained using A, procedure 1; B, procedure 2; and C, commercial calcium carbonate.Procedure 1 (Fig. 1A) gives the highest candoluminescence emission and this could possibly arise from an “after precipitation” of manganese hydroxide, which coats the surface of the calcium hydroxide particles. In procedure 2 (Fig. 1B) higher emissions than expected are found. Here, it seems likely that manganese carbonate is more concentrated on the surface of the calcium carbonate than in the centre of the particles. Differences in particle sizes of the precipitates obtained in procedures 1 and 2 could account for the differences in emission intensities. Our results indicate that the preparation of standards by either of these coprecipitation1470 BELCHER, CLARK, LLOYD AND PUZANOWSHA-TARASIEWICZ procedures cannot be used and that the standardisation by alternative methods is thus justi- fied.The kinetic spectrophotometric method is time consuming but satisfactory for these low levels of manganese. The atomic-absorption spectrophotornetric method also involves a dis- solution step but the candoluminescence technique uses only one other reagent, calcium sulphate. The method is relatively simple once a calibration graph has been obtained and it is rapid in the hands of a skilled operator. The larger standard deviation by the candolumines- cence method compared with the other methods is due to the slight variations in gas flows. This problem could possibly be overcome by modification of the sample holder head so as to permit a change-over of samples without cutting off gas flows. One of us (H.P.-T.) thanks Dr. D. J. Harrison, Science Officer, and the British Council in Warsaw for the award of a visiting Research Fellowship. We also thank John and E. Sturge Ltd. of Birmingham for supplying samples of precipitated calcium carbonate and for their interest in this work. Refer en ce s 1. 2. 3. 4. 5. 6. 7. 8. Belcher, R., Nasser, T. -4. K., Shahidullah, M., and Townshend, A., Int. Lab., 1978, Jan./Feb., 45. Sokolov, V. A., Zh. Fiz. Klzinz., 1978, 52, 3090. Dhaher, S. M., and Kassir, 2. M., Ana2. Chew, 1980, 52, 459. Belcher, R., Ranjitkar, K. P., and Townshend, -4., Analyst, 1975, 100, 415. Belcher, R., Karpel, S., and Townshend, A., Talanta, 1976, 23, 631. Belcher, R., Ranjitkar, K. P., and Townshend, A., Analyst, 1976, 101, 666. Nichols, E. L., Howes, H. L., and Wilber, D. T., Carnegie Inst., Washington, Publ., 1928, No. 348. Dolmanova, J. F., Zolotova, G. A., and Ratina, M. A., Zh. Anal. Khim., 1978, 33, 1356. Received May 26th, 1983 Accepted July 4th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801466

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, December, 1983, Vol. 108, Pe. Id7l-ll76 147 1 Low-tem perature Luminescence Spectroscopy Using Conduction Cooling and a Pulsed Source Luminescence Spectrometer Alun T. Rhys Williams and Stephen A. Winfield and James N. Miller Pevkin-Elmer Limited, Beaconsfield, Buckinghamshire, H P 9 1QA Department of Chemistvy, Lozcghborough University of Technology, Loughborough, Leicestershire, LE11 3TU A new conduction cooling device for low-temperature luminescence measure- ments is presented and its use evaluated in a pulsed source luminescence spectrometer, Fluorescence and phosphoresence quantum yields are reported for some polyaromatic hydrocarbons a t 77 K. Keywords : Low-temperature luminescence spectroscopy ; conduction cooling ; pulsed souwe Low-temperature fluorescence and phosphorescence spectroscopy are not popular assay methods amongst analytical chemists.Potentially these techniques offer both increased sensitivity and selectivity of analysis. The latter is obtained by spectral line narrowing at low temperatures and by the use of time-resolved phosphorimetry. Increased sensitivity is predominantly observed for those compounds which phosphoresce rather than those which have relatively high fluorescence quantum yields, Phosphorescence detection limits range from 0.0004 to 4 pg ml-l for polycyclic aromatic hydrocarbons, whilst drugs such as pheno- barbital and procaine have been detected at 7 and 2.5 ng ml-l, respectively. The reluctance to use low-temperature luminescence spectroscopy probably arises from the problems associ- ated with the need to cool samples to liquid-nitrogen temperature, 77 K.Conventional low- temperature accessories use immersion cooling, i.e., the sample in a synthetic fused silica tube is cooled by immersion in liquid nitrogen held in a Dewar flask. Two major problems arise from this method ; poor precision resulting from irreproducible sample positioning and erratic freezing rates and very slow sampling rates, Additional problems include a gradual decrease in signal intensity due to an accretion of ice crystals in the base of the Dewar and a build-up of ice on the outside of the base of the Dewar if the sample compartment has not been adequately purged with dry nitrogen. One other factor that can deter potential users is the relatively high cost of phosphorescence accessories.In addition to a costly synthetic fused silica Dewar some means of mechanically chopping the excitation and emission beams, i.e., a rotating can with its associated motor and control module, is required. Attempts have been made to improve the precision of low-temperature measurements by rotating the sample tubel ; this averages any inhomogeneities in the frozen sample but does not overcome positioning errors. Conduction cooling provides an alternative to immersion cooling as a means of improving the precision and speed of analysis. Ward and co-worker~~~~ discussed two conduction cooling systems. In the first, a copper rod was immersed in liquid nitrogen and the sample was cooled by contact with the rod. Although the speed of analysis was improved by a factor of three over conventional phosphorimetry, the detection limits were similar. Measurements indicated that the operating temperature of the system was 100 K.In the second system the copper rod was cooled by flowing through it liquid nitrogen, the operating temperature being 85 K. In this instance detection limits were lower, though the main advantage was the improved pre- cision and increased speed of analysis. In this paper we describe a conduction cooling device using a high-purity copper rod immersed in liquid nitrogen used in conjunction with a pulsed source luminescence spectro- meter.4 The advantages of using a pulsed source are as follows: increased sensitivity, as a pulsed source produces higher peak intensities in the ultraviolet region than a continuously operated xenon-arc lamp with a mechanically chopped excitation beam; better time resolution , which permits analysis of phosphors with short lifetimes (0.1-50 ms)596 in the presence of1472 RHYS WILLIAMS et aZ.: LOW-TEMPERATURE LUMINESCENCE Analyst, VOZ. 108 relatively long-lived phosphors ; and the ability to discriminate fluorescence and scattered light from long-lived phosphorescence without the use of a mechanical phosphoroscope. The events that occur during excitation of a sample with a pulsed source are shown in Fig. l.'p8 As the observed phosphorescence signal (P) represents only a fraction of the total phosphorescence intensity (PT), the area under the corrected phosphorescence-emission spectrum has to be multiplied by a factor that is related to the characteristics of the pulsed- source electronics and the lifetime of the phosphorescent speciesg The excitation flash width (if) is assumed to be much smaller than the phosphorescence lifetime, the latter being ca.0.5 x 10-3-10 s for organic phosphors. -t*-fg-* Time+ Fig. 1. Events occurring during the excitation of a sample with a pulsed source. A, 'lhc excitation pulse; 73, build-up of luminescence signal I, and then exponential decay; tp, width a t half peak height; t d . delay from beginning of pulse to beginning of observation; t,, gate width of detector. The total observed decay for a single exponential may be expressed by the following equa- tion : where PT is the total phosphorescence, yo is the intensity at zero time and r is the single- exponential decay time.This may be evaluated as The integrated phosphorescence intensity, P, during the time interval tg, at a time td after to is given by the following expression : The fraction of light observed is then given by the ratio of equations (2) and (3) : Equation (4) holds for a single pulse of excitation light; for a pair of pulses with a cycle time of 20 ms, equation (4) becomesDecember, 1983 USING CONDUCTION COOLING AND A PULSED SOURCE 1473 where all times are in microseconds. Experimental Low-temperature fluorescence and phosphorescence measurements were obtained on a Perkin-Elmer, Model LS-5, luminescence spectrometer fitted with a Hamamatsu R928 red- sensitive photomultiplier. Data were recorded on a Perkin-Elmer, Model 3600, Data Station.The spectrometer uses a 9-W xenon lamp pulsed at line frequency, tlie pulse width at half peak height being less than 10 ps. The signals from tlie sample and reference photomultipliers are gated by the microprocessor-controlled electronics and measured by an A-D converter with an 18-bit dynamic range. The reference photomultiplier forms part of the quantum-corrected reference system, which enables corrected excitation spectra to be obtained. In the fluor- escence mode the signals from the sample photomultiplier are gated for the duration of the flash whilst in the phosphorescence mode the gating is delayed so that it no longer coincides with the flash. The discrimination between fluorescence emission and long-lived phosphorescence is difficult to achieve on instruments using a d.c.source, e.g., a 150-W xenon lamp. Fig, 2(a) shows the low-temperature “fluorescence” spectrum of coronene in hexane obtained using a d.c. operated 150-W xenon source. Fig. 2(b) sliows the fluorescence spectrum obtained using a pulsed 9-W xenon source. The difference results from the fact that the peaks above 500 nm are phosphorescence transitions. When using a pulsed source the sample photomultiplier can be gated to look at tlie signal at the instant of the flash and again just before the next flash. By subtracting the signals a fluorescence signal free from any long-lived phosphorescence and dark current is obtained. 80 60 8 e g 4c 2 m 2c 100 80 60 40 20 400 500 600 0 500 Wavelengt hlnm Fig. 2. (a) Low-temperature fluorescence-emission spectrum of coronene in hexane measured with a d.c. operated 150-W xenon lamp.(b) Low-temperature fluorescence-emission spec- trum of the same sample measured with a pulsed 9-W xenon lamp. A !I I I# I I I I I I c l I 2 ‘D Fig. 3. Diagram of the copper conduction cooling rod. A, Sample; B, optical window ; C, hollow cavity; D, hole allowing nitrogen to boil over the sample tubes.1474 RHYS WILLIAMS et d. LOW-TEMPERATURE LUMINESCENCE Aftdyst, VOZ. 108 The sample compartment of the LS-5 was changed to incorporate the conduction cooling low-temperature accessory. The conduction rod was machined from high-purity 99.99% copper with a thermal conductivity of 4.05 J s-l cm-l K-l (Fig. 3). Sample tubes of 4 mm 0.d. and 2 mm i.d. were inserted into a hole drilled in the top of the copper rod with a slot milled to 20 mm deep from the end to serve as the optical windows.The rod is hollowed leav- ing a cavity of 12.7 mm i.d. extending from the open end. A hole is drilled into the cavity allowing nitrogen gas to flow around the sample tube in the slot. Once sealed in position in the sample compartment the flow of nitrogen serves to keep the slot free of moisture from the surrounding air. The bottom part of the copper rod is immersed in 2 1 of liquid nitrogen held in a box constructed of stainless steel, insulated by expanded polystyrene. The accessory is enclosed by the sample compartment cover through which the sample tubes are inserted by means of a removable lid. The inside walls of the windows of the optical module were kept free of ice by a stream of dry nitrogen at a flow-rate of 1 1 min-l. Topping up with liquid nitrogen is necessary every 15-20 min, depending upon the external air temperature.The calculation of the quantum yields was greatly simplified by the use of a desk-top computer interfaced with the spectrometer, with the observed and factored phos- phorescence-emission areas being automatically calculated. Reagents Methylcyclohexane (spectroscopic grade) was fractionated and stored over concentrated sulphuric acid for 1 week. The solvent was then washed with de-ionised water and again fractionally distilled. This process, which was also applied to cyclohexane, greatly reduced the background fluorescence. Methylcyclohexane was preferred as the solvent as it formed a clear glass at 77 K, whilst cyclohexane formed a snow that increased light scattering and hence the background signal.Aqueous solutions require a thicker walled sample tube to withstand the expansion when cooled to 77 K. Stock solutions of the chemicals were prepared and stored at 4 "C. Spectra were recorded with a 2.5-mm spectral band pass and were corrected for instrument response from 250 to 630 mm. Results and Discussion Confirmation of equation (5) for calculating the phosphorimeter factor was obtained in the following way. A transferrin - terbium complex (loA6 and 4 x 10W M, respectively) at pH 8.0 was excited at 280nm and the area under the corrected emission spectrum of terbium measured at various delay ( t d ) and gate (tg) times (Table I). The observed areas were multi- plied by the appropriate factor to produce a mean result of 2513 with a relative standard deviation (RSD) of & 2.03%.The graph shows that almost all the signal from short-lived phosphors is measured whilst only a fraction of a long- lifetime emission is measured. For the same quantum efficiency, short-lived phosphors A T value of 1.22 ms was used. The variation of P,,P for various values of T is shown in Fig. 4. TABLE I CALCULATION OF TOTAL PHOSPHORESCENCE EMISSION FROM OBSERVED AREAS AT VARYING id AND t g USING EQUATION (4) T = 1.22 ms. td/ms 0.03 0.1 0.5 1.0 2.0 3.0 0.1 0.1 0.1 tglms 0.5 0.5 0.5 0.5 0.5 0.5 1 .o 2.0 3.0 0 bserved emission area 823 79 1 596 359 169 78 1286 1851 2087 Total phosphorescence emission* 2508 2553 2570 2423 2589 2512 2495 2493 2477 *Mean = 2513; S.D.50.95 (n = 9).December, 1983 USING CONDUCTION COOLING AND A PULSED SOURCE 1475 (7 ca. s) will show an order of magnitude increase in intensity compared with a long-lived species (T ca. 10-1 s). Fluorescence quantum yields were calculated as described by Rhys Williams et aZ.1° assuming a fluorescence quantum yield of 1.00 for 9,lO-diphenylanthracene at 16 12 % cl‘ 8 - - - ~ 8 8 8 8 8 1 10 100 1000 10000 Time/ms Fig. 4. Variation of PT/P (ratio of the total phosphorescence, PT, to the integrated phosphores- cence intensity, P, for different T values (ms) measured with a t d of 0.1 ms and t , of 1.0 ms. 77 K.ll Because a 2 mm i.d. cuvette is used in the phosphorescence accessory, solute con- centrations of up to five-fold higher were used compared with fluorescence measurements made using 10 mm path length cuvettes. This enabled good quality spectra with high signal to noise ratios to be used in the calculation of quantum yields.The observed peak intensity of the 9,lO-diphenylanthracene increased approximately five- fold on going from 293 to 77 K. This increase was predominantly due to band sharpening, 250 350 450 5150 Wavelengthlnm Fig. 5. (a) Room-temperature excita- tion and emission spectra of 9,lO-diphenyl- anthracene in methylcyclohexane. (b) Low-temperature (77 K) excitation and emission spectrum of the same sample.1476 RHYS WILLIAMS, WINFIELD AND MILLER although an increase in the absorption coefficient cannot be dismissed. Fig. 5 compares the room temperature and 77 K excitation and emission spectra of 9,lO-diphenylanthracene.The results of the quantum yield determination are shown in Table 11. The reproducibility of the quantum yield values at 77 K is & S-lO% (RSD). Fluorescence quantum yields agreed, within experimental error, with those in the literature with the exception of fluorene which Parker and Hatchard12 assumed to have a quantum yield of 0.54 at room temperature and at 77 K. Phosphorescence quantum yields, with the exception of benzene, were also in good agreement with previously determined values. The discrepancy with the benzene is probably due to the level of impurities, as both the quantum yield and lifetime differ signifi- cantly. TABLE I1 LOW-TEMPERATURE FLUORESCENCE AND PHOSPHORESCENCE QUANTUM YIELDS (Qf AND QP) RELATIVE TO ~,~O-DIPHENYLANTHRACENE AT Qf = 1.00 Values in parentheses are literature values.Aex. rangel Compound Solvent Aex./nm nm Lifetimels Observed Qr Observed Qp References Anthracene .. .. MCH* 252 360-490 N.0.t 0.29 (0.27) N.o. 12 Benzene .. . . .. MCH 254 265-340 5.8 ( 8 ) 0.20 (0.21) 0.31 (0.19) 1 2 BenzoDhenone .. .. MCH 250 400-500 6 x N.o. 0.89 (0.84) 13 . . (8 x 10-5) Coronene .. .. .. MCH 303 350-510 9 0.12 0.07 Fluorene . . .. . . MCH 265 275-380 N.o. 0.77 (0.54) N.o. (0.07) 12 Naphthalene .. Hexane 276 300-400 N.o. 0.35 (0.39) N.o. (0.008) 12 Tetraphenylbutadfene . . MCH 346 350-520 N.o. 0.90 N.o. Triphenylene . . . . Hexane 259 330-440 10.5 (15) 0.04 (0.06) 0.25 (0.28) 1 2 Eu(TTA),$ .. .. Ethanol 275 3.6 x 10-4 N.O. 0.13 (0.18) 14 * MCH, methylcyclohexane.t N.o., not observed. $ TTA = thenoyltrifluoroacetonate. Results obtained a t room temperature, 25 “C. No special technique or practice was required in order to maintain reasonable precision using After initial cool-down it was only necessary to add liquid Samples were cooled and reached maximum phos- the conduction cooling device. nitrogen on average at the rate of 1 1 h-l. phorescence intensity 3045 s after insertion. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. References Lukasiewicz, R. J., Mousa, J. J., and Winefordner, J . D., Anal. Chem., 1972, 44, 1339. Ward, J . L., Bateh, €3. P., and Winefordner, J . D., Appl. Spectvosc., 1980, 34, 15. Ward, J . L., Walden, G. L., Bateh, R. P., and Winefordner, J. D., Appl. Spectrosc., 1980, 34, 348. Rhys Williams, A. T., I n t . Lab., 1981, 11, 90. Harbaugh, I<. F., O’Donnell, C. M., and Winefordner, J. D., Anal. Chem., 1971, 45, 381. Harbaugh, K. F., O’Donnell, C. M., and Winefordner, J. D., Anal. Chem., 1974, 46, 1206. Fisher, R. P., and Winefordner, J. D., Anal. Chem., 1972, 44, 948. O’Haver, T. C., and Winefordner, J. D., Anal. Chem., 1966, 38, 1258. Boutilier, G. D., Bradshaw, J. D., Weeks, S. J., and Winefordner, J. D., Appl. Spectrosc., 1977, 31, Rhys Williams, A. T., Winfield, S. A., and Miller, J. N., Analyst, 1983, 108, 1067. Mantulin, W. W., and Huber, J. R., Photochem. Photobiol., 1973, 17, 139. Parker, C. A., and Hatchard, C. G., Analyst, 1962, 87, 664. Gilmore, E. H., Gibson, G. E., and McClure, D. S., J . Chem. Phys., 1955, 23, 399. Baumik, M. L., and Telk, C. L., J . o p t . SOG. A m . , 1964, 54, 1211. 307. Received July 4th, 1983 Accepted July 27th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801471

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, December, 1983, Vol. 108, $9. 1477-1480 147 7 Analytical Potential of Valence State and Ligand Atom Effects in Titanium K X-ray Spectra John B. Jones and David S. Urch Admiralty Marine Technology Establishment, Holton Heath, Poole, Dorset, BH16 6 J U Chemistry Department, Queen Mary College, Mile End Road, London, E l 4NS Changes in peak profiles and in peak energies of titanium Kcc, KPl,3 + K/3’ and K& + KP” X-rays with both valency and ligand environment changes have bekn observed. Both factors affect Kcc but a distinct KP’ feature identifies Ti(I1) whilst ligand atoms can be identified from the KP2,+ KjY energy difference. When the ligands are known, the small shifts of the sharp and intense Ti Ka, peak provide the best indicator of valency. It is suggested that such shifts, together with changes in KP peak profiles, could be used analytically to determine the valence state of titanium and the chemical nature of the atoms that surround it.Keywords : Titanium X-ray spectra ; valence state effeects ; ligand atom effects The wavelengths of characteristic X-ray emissions can be slightly but perceptibly altered by changes in the formal valency or ligand environment of an atom, even if the X-ray results from electronic transitions between core orbitals. More dramatic effects such as larger wavelength shifts and the formation of new peaks are observed as a result of valence shell to core transi- tions. The analytical potential of these effects has, however, been exploited only in a few iso- lated instances, e.g., to determine the co-ordination number of aluminium using Ka,,, shifts‘ or to determine the valency of manganese using the KP’: KP1,3 intensity ratio2 or to determine the chemical nature of a ligand atom attached to a second-row atom using the KP’-KP1,3 energy differen~e.~ Even so there is sufficient variety in the parameters that can be measured and the correlations that can be established to suggest that a detailed study of peak shifts, etc., should yield data that could be useful in analysis by providing structural and valence state information.In this work an examination had been made of the X-ray emission spectra from a wide range of titanium compounds in which both the valence state and the ligand environ- ment of the titanium atom were varied. Experimental The titanium X-ray emission spectra of the following substances were measured : titanium metal, titanium carbide, titanium oxides (TiO, Ti203 and TiO,), titanium sulphide (Ti,S,), titanium trifluoride, potassium hexafluorotitanate and titanates of barium and neodymium.Either commercially available samples of AnalaR purity were used or compounds were pre- pared by standard techniques. The stoicheiometries of the titanates, which were supplied by S.T.L., Harlow, were checked by X-ray diffraction. Discs of all the samples (except titanium metal) were made by pressing the powdered compound with terephthalic acid in a ring press to 5 tons. These discs were irradiated with X-rays from a sealed chromium anode X-ray tube (50 mA, 50 kV) in a Philips PW1410 X-ray fluorescence spectrometer.To obtain the best dispersion of the characteristic radiation, a rubidium acid phthalate crystal (2d = 2612 pm) was used for Ti Karl,, (ninth order) and ammonium dihydrogen orthophosphate (2d = 1064 pm) was used for Ti KP1?3 and Ti KP2,5 (fourth order). The exact positions of the Ka,,, and KP,,, peaks were determined by counting for pre-set periods of time every O.Ol”26’ in the region of the peak maximum. The more complex KP,,, region was measured by scanning the wave- length automatically and recording the fluctuations in count rate (from the ratemeter) graphically. Results and Discussion The energies of Ka,, KP1,3 and KP,,, (with K/3” if seen) are listed in Table I and some typical results for the Kp”-Kp2,5 and KP,,,--KP’ regions are shown in Figs.1 and 2, respectively.1478 JONES AND URCH: ANALYTICAL POTENTIAL OF VALENCE Analyst, VoZ. 108 TABLE I ENERGIES OF TITANIUM X-RAYS X-ray energy/eV b r Titanium metal.. . . 4510.0 4931.0 4960.3 Ti0 . . .. . . 4509.9 4930.4 4955.9 Ti,O, . . .. . . 4509.6 4930.7 4956.6 Ti,S, . . .. . . 4509.8 4930.9 4958.0 TiF, . . .. . . 4509.4 4930.9 4959.6 Tic . . .. . . 4510.0 4931.0 4959.9 TiO, . . .. . . 4509.5 4930.9 4960.8 BaTiO, . . .. . . 4509.2 4930.7 4961.5 BaTi,O, . . . . . . 4509.3 4930.7 4960.8 Nd,Ti,O, . . .. . . 4509.3 4930.7 4960.8 K,TiF, . . .. , . 4509.3 4930.9 4962.4 Compound Kal* Kfi1,3 KfiZ,5 * Kcc, = Kor, -5.6 eV. 7 K r 4953.5 4 945.1 4945.8 4945.3 4945.3 4941.0 Kw,2 The most intense emission feature in the X-ray spectrum of titanium is Kocl,,, which is generated by an electronic transition from the 2p orbitals to a 1s vacancy.The final 2p-1 “hole” configuration is, however, split by spin - orbit exchange coupling into two possible states, which differ in energy by 5.6 eV, 2P, and 2P+ This causes the Ka peak to be split into Kcc, [ls-1(2S+) -+ 2p-1(2P;)] and Kcc, [1s-1(2S,) -+ 2~-l(~P,)], the former having a slightly higher energy and being twice as intense as the latter. Because of its intensity and because it can be resolved from Koc,, chemical effects in Ti Kcx spectra will be discussed in terms of the Koc, peak alone. The shifts observed for Kcxl are very small, the whole range of 0.8 eV corresponding to only 0.06’28, but the sharpness of the Kcx peaks enables their angular position to be determined precisely (&0.01”28).The shifts that are observed, relative to the value of Ti Karl for titanium metal, are a function of both ligand and valence state. These two effects can be disentangled as shown in Fig. 3. From this diagram it can be seen that any effect that would increase the 4 :i”\. 2 n- 4940 4960 4980 E ne rg yleV Fig. 1. Titanium K/3”- Kfi,,& spectra. A 4920 4930 4940 E ne rgyleV Fig. 2. Titanium Kfi’ -ICfll,s spectra from (A) TiO, and (B) TiO. For Ti0 note KP’ at 4916 eV.December, 1983 STATE AND LIGAND ATOM EFFECTS IN Ti K X-RAY SPECTRA 1479 effective charge on the titanium atom, be it either an increase in the electronegativity of the ligands or an increase in the formal valence state, causes a reduction in the Ka, energy. This is because valence shell orbitals, which would, from either cause, experience a reduction in electron density, overlap more with 2p orbitals than with 1s.Thus, as charge is withdrawn from the valence shell both 2p and 1s orbitals became more tightly bound owing to a reduction in electron - electron repulsion, but because of orbital overlap this effect will be greater for 2p than for Is. The energy difference between 2p and Is ionisation.energies, the Ka, and Kaz X-ray energy, will therefore diminish as the titanium atom becomes more positive. Unfortunately, the shifts brought about by changes in ligands and in valence state are com- parable, so that it is only possible to determine the valence state from a knowledge of the Ka shift if the ligand atoms are known or vice versa; the chemical nature of the ligands can only be indicated if the valency is known.It should be noted, however, that the required information about ligand atoms can sometimes be found from KP” and KP,,, spectra. Carbon Sulphur X I Fluorine I x o 0 -0.2 -0.4 -0.6 -0.8 ShiWeV Fig. 3. Shifts of the titanium Ka, peak from titanium metal for different ligands and different titanium valences: e, (11); X , (111); and C, (IV). KPl3 The Ti K/3,,, peak is caused by tfie transition 1s-l -+ 3p-l, but as 3p orbitals are more diffuse than 2p orbitals their overlap with a 1s core hole is less so that KP1,3 peaks are less intense than Ka. For titanium the 3p spin - orbit exchange splitting is smaller than for 2p so that distinct KP, and KP3 peaks are not seen; however, overall KP1,3 is broader than either Ka, or Ka2.This greater width makes it more difficult to locate the Kp1,3 peak maximum than Ka,. Also it is found that the KP1,3 shifts are in fact smaller than Kor, shifts and behave in a different way, the shift for Ti(I1) being greater than those of Ti(II1) or Ti(1V). This shows that other factors apart from the effective charge at titanium help to determine the KP1,3 energy. With an ionisation energy of only about 38 eV it seems most likely that Ti 3p orbitals will be directly influenced by ligand atoms, i.e., they will be subject to chemical bond effects4 The breadth of the Kp1,3 peak may be due to many possible final states being accessed by the “3p” -+ 1s relaxation ; states in which other electronic excitation (“shake-up”) has taken place involving vacant 3d orbitals.The anticipated simple decrease in KP1,3 energy with increasing positive charge at the titanium would then be masked by these other configurational excitation effects. The simplest of these is exchange coupling between unpaired 3d electrons and the final state vacancy in the 3p shell.5 The hole state spin can be parallel or anti-parallel to the 3d electron spin so that there would be two possible energy states associated with 3p5, for Ti(I1) and Ti(II1). These two states give rise to two peaks in the Kp spectrum, the main KP1.3 peak and a low-energy satellite KP’ (the relative intensity of which is, for light transition elements, pro- portional to the number of unpaired d electrons5). The presence of such a peak in the Ti KP spectrum of Ti0 can be clearly seen in Fig.2, with a relative intensity of about 10% of KP1,3 and about 12 eV less energy. By analogy with manganese2 the presence of such a peak can be used diagnostically to indicate the presence of divalent titanium. The single d electron in Ti(II1) does not, however, generate an intense enough feature to be of use analytically.1480 JONES AND URCH KP2.5- KP” These very weak peaks result from electronic transitions from valence band orbitals to the titanium 1s core orbital; the transition will therefore be due to Ti 4p character in these valence band orbitals. However, these orbitals are predominantly ligand in character and so their energies will be determined by the nature of the ligand atom and the effective charge at the ligand atom.If for all titanium oxides and for titanates the oxygen 2p ionisation energy is taken, as a first approximation, to be constant, then it follows that the KP,,, transition energy should decrease in the valency sequence Ti(1V) > Ti(II1) > Ti(I1) because the Ti 1s ionisa- tion energy will be largest for the titanium state with the greatest positive charge. This anticipated trend is in fact observed and can be seen from the results in Table I : TiO, (4960.8 eV), Ti203 (4956.6 eV) and Ti0 (4955.9 eV). A similar trend is found for titanium in different valence states with fluorine ligands, TiF,” (4962.4 eV) and TiF, (4959.6 eV). For a particular valence state increasing the electronegativity of the ligands should also increase the Ti 1s ionisation energy but this trend runs parallel to the increase in ligand valence orbital ionisation energy so that only slight changes may be observed.Actually a small increase is found: for TIC, TiO, and K,TiF, the energies are 4959.9, 4960.8 and 4962.4 eV, respectively. Although the shifts found for KP2,5 are much larger than for the other K peaks of titanium their lack of intensity limits their use for analysis. If, however, both KP,,, and KP” can be observed then the location of Kp” is of use in determining the chemical nature of the ligand atoms. This is because the origin of KP2,5-KP” for first-row transition elements is strictly analogous to that for KP, 3-K/3’ for second-row main group elements,6 i.e., KP2,5 reflects the 4p character in orbitals principally ligand 2p in character and KP” the 4p character in orbitals principally ligand 2s in character. The KP2,5-KP” region of the titanium X-ray spectrum is shown in Fig.1, where it can be seen that KP” is only clearly observed in Ti(1V) compounds. This is because the intensity ratio Kp” : KP2,5. decreases with metal valency’ and also because the KP” -+ KP,,, feature shifts to lower energies with lower valencies, KP” then tending to be lost* on the side of the steeply rising KP,,, peak (TiO, Fig. 1). Fig. 1 shows that for tetra- valent compounds KP” can be found for K,TiF, at about 20eV less than the KP2,5 peak energy, for TiO, and the titanates at about 16 eV less than KP,,,, and for TIC at about 6 eV less than KP2,5. These energy separations can, therefore, be used to determine the nature of the ligand atoms that surround the titanium.Conclusions Shifts in Ti Kcc,, whilst small, can be used to determine the valency of titanium provided that the ligand environment is known. This can be found, at least for Ti(IV), by an examination of the energy separation between KP” and Kp2,5. The presence of low-valency titanium [i.e., Ti(II)] can be confirmed by the presence of a distinct KP’ peak on the low-energy side of K/31,3. Thus if all the K X-ray emission spectra are examined it should be possible to deter- mine both the valency of the titanium and the chemical nature of atoms to which it is bound. The authors thank The Royal Society arid the Central Research Fund of London University One of them (J.B. J.) is grateful to for financial assistance in the purchase of equipment. both the SERC and to S.T.L. (Harlow) for a CASE award and research grant. References 1. 2. 3. 4. 5. 6. 7. 8. Day, D. E., Nature, London, 1963, 200, 649. Wood, P. R., and Urch, D. S., X-ray Spectrom., 1978, 7, 9. Esmail, E. I., Nicholls, C. J., and Urch, I>. S., Analyst, 1973, 98, 725. Ichikawa, K., Nakamori, H., Tsutsumi, K., and Watanabe, T., J. Phys. SOC. Jpn., 1977, 43, 1255. Slater, R. A., and Urch, D. S., J. Chem. SOC. Chem. Commun., 1972, 564. Urch, D. S., J. Phys. C , 1970, 3, 1275. Asada, E., Takiguchi, T., and Suzuki, Y., X-ray Spectrom., 1975, 4, 186. Nemnonov, S. A., and Kurmayev, E. S., Fiz. Met. Metalloved., 1969, 27, 816. Received May 3rd, 1983 Accepted July 4th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801477

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, December, 1983, Vol. 108, $9. 1481-1484 148 1 Quantitative Examination of Polyester = Cotton by Nea r-inf rared P hotoacoustic Spectroscopy Caroline M. Ashworth, Gordon F. Kirkbright and Dominic E. M. Spillane Department of Instrumentation and Analytical Science, U M I S T , Manchester, M60 1QD Quantitative analysis of polyester - cotton has been conducted using near- infrared photoacoustic spectroscopy. Linear calibration graphs were obtained for mixtures of cotton and polyester; the linear regression coefficient for the cotton calibration graph was calculated to be 0.990 and that for poly- ester was 0.998. Keywords : Polyester - cotton analysis ; textile analysis ; near-infrared photo- acoustic spectroscopy Photoacoustic spectroscopy has, in recent years, been applied to the examination of a variety of solid samples.Quantitative analysis by this technique has been undertaken for materials separated on thin-layer chromatography plate~,l-~ the moisture content of a. variety of ~amples,~J the content of pharmaceutical preparations6 and the composition of copolymeric systems. Photoacoustic spectroscopy is well suited to the analysis of condensed-phase systems as it is applicable to a wide variety of sample morphologies. Results have been reported in the litera- ture for highly reflecting materials,* strongly scattering sample^^^^^^^^ and opaque samples.gJ0 It should be noted that the photoacoustic signal is not independent of the form of the sample; however, if care is taken in matching particle and sample size, quantitative spectroscopic analysis may be undertaken.The near-infrared region of the electromagnetic spectrum proves to be extremely useful for quantitative analysis by photoacoustic spectro~copy.~-~ Absorption in this region can, most frequently, be attributed to overtone and combination bands arising from hydrogenic systems, principally C-H, N-H and 0-H. This facilitates the analysis of simple mixtures of organic materials and of organic compounds in inorganic matrices. The majority of modern textile fabrics employ a blend of more than one type of fibre for reasons of cost, texture and wear rate. Textile fabric analysis is of particular current import- ance for tariff classification purposes and official methods are under review by BSI and the appropriate EEC authorities.This has resulted in the development of methods to determine the percentage composition of such fabrics using techniques such as gravimetry,ll near-infrared reflectance spectrophotometry12 and electron spectroscopy for chemical ana1y~is.l~ Recent work14J5 has demonstrated that photoacoustic spectroscopy has some potential for the analysis of dyes and additives to fabrics in the ultraviolet - visible region of the spectrum and the possibility of analysing fibre mixtures using Fourier transform - photoacoustic spectroscopy for use in the mid-infrared region has also been described.15 This paper presents the results of a quantitative study undertaken with polyester - cotton utilising near-infrared photoacoustic spectroscopy. Experimental Instrumentation The near-infrared photoacoustic spectrometer employed for this study has been described e1~ewhere.l~~~~ The system employs a 400-W tungsten filament lamp (Type A1/239, Philips Electrical Ltd., Croydon) as light source, the radiation being focused through a mechanical chopper on to the entrance slit of an f/4 monochromator containing a grating blazed at 2.0 pm.A silicon cut-on interference filter (type D07393F, Davin Optical Ltd., Barnet, Herts.) was placed in the exit slit of the monochromator to prevent transmission of overlapping spectral orders. At the exit slit of the monochromator the radiation was reflected into a laboratory-built photoacoustic ce11.l6 The signal from the cell was taken to a lock-in amplifier (PARC, Model 124A, Princeton Applied Research Corp., Princeton, NJ, USA) and stored in a digital scan recorder (PARC, Model 4101).Hard-copy output was obtained by using an X - Y recorder (Model 25000, Bryans Southern Instruments Ltd., Surrey).1482 ASHWORTH et al. QUANTITATIVE EXAMINATION OF Analyst, vd. 108 The spectrometer was operated in the single-beam mode; correction of the spectral data was All spectra achieved by ratiometry against a previously stored spectrum of carbon black. were scanned at 200 nm min-l at a spectral half-band pass of 0.03 pm. Sample Handling The following six commercially available fabric samples were employed : pure cotton ; 70% cotton, 30% polyester; 50% cotton, 50% polyester; 40% cotton, 60% polyester; 17% cotton, 83% polyester ; and pure polyester. All the samples employed for this study were ground under liquid nitrogen in a mill (SPEX Ind.Inc., “Freezer Mill”). Quantitative measurements were performed with samples that had been sieved to give a particle size of <110 pm. No further sample pre-treatment was per- formed. Calibration standards were obtained by mixing appropriate amounts of the pure ground polyester and cotton samples. Sampling of the commercially available mixtures was performed by removing ca. 30 cm2 of the material and then grinding and sieving as for the pure samples. Samples transferred into the cell were approximately 50-60 mg. Results and Discussion Fig. 1 shows the near-infrared photoacoustic spectra of powdered cotton, powdered polyester and a 1: 1 m/m mixture of these. Comparison of the spectra shows that cotton exhibits an absorption at ca.1.55 pm, which may be assigned to the first overtone of an O-H stretching vibration,l* but the polyester spectrum shows a maximum at ca. 1.72 pm arising from a C-H combination band. In principle, the absorption at these wavelengths referenced to that at ca. 1.35 pm, where all the spectra reach a minimum, should yield quantitative data concerning the composition of the mixture. There is, however, a large contribution to the signal at 1.72 pm from the cotton in the sample and a smaller but still significant contribution from the polyester in the sample at 1.55 pm. It is therefore necessary to apply a correction to the photoacoustic measurements made in the mixtures. Accordingly, a mathematical treatment was applied to calculate the contribution to the photoacoustic signal of each component independently.1.35 1.55 1.72 Wavelength/pm Fig. 1. Normalised photoacoustic spectra of (A) pure cotton; (B) a 1 : 1 m/m mixture of cotton and polyester; and (C) pure polyester.December, 1983 POLYESTER - COTTON BY NEAR-IR PHOTOACOUSTIC SPECTROSCOPY 1483 This treatment assumes that the contributions of the two components to the photoacoustic signal will be additive and therefore we can say that for any mixture of polyester and cotton and .. .. - - (1) .. .. * (2) S (1.55) = C + XP . . . I S (1.72) =yC + P . . .. where S(h) is the measured photoacoustic amplitude at h pm; C is the contribution to the photoacoustic signal at 1.55 pm from cotton alone; P is the contribution to the photoacoustic signal at 1.72 pm from polyester alone; x is the ratio of the photoacoustic signal for 100% polyester at 1.55 pm to that a t 1.72 pm; and y is the ratio of the signal amplitude for 100% cotton at 1.72 pm to that at 1.55 pm and the observed photoacoustic signal at 1.35 pm for a sample is subtracted from all measurements made on that sample in order to compensate for background variations.It can be seen that S(h), x andy can be derived from experimental results and therefore equations (1) and (2) may be solved simultaneously to give values for C and P. Fig. 2 shows the calibration graphs obtained for C and P versus mixture composition for a set of prepared, powdered standards containing mass fractions in the range 0-1 for both com- ponents. The regression coefficients for these graphs are 0.990 for cotton and 0.998 for polyester .These calibrations will provide a useful source of rapid quantitative data for the amount of polyester or cotton in a mixture. If, however, the mixture is a simple binary system of cotton and polyester, it may be more useful to measure the ratio of the components. Fig. 3 shows a calibration graph for percentage cotton/( 100- percentage cotton) versus the ratio C/P. This yields a linear graph with a regression coefficient of 0.9999 from which it is possible to deter- mine directly the ratio of the components in the system. We have employed the calibration in Fig. 3 to determine the cotton composition in four commercial polyester - cotton materials. The dotted lines show the values of C/P obtained for ground samples of these materials and the ratio of cotton to polyester present in the sample 2.0 I d O1 \ o b ' I I A 0 20 40 60 80 100 Cotton, TO 100 80 60 40 20 o Polyester, YO Fig.2. Calibration graphs for polyester - cotton mix- tures. (A) Graph of the contribution to the photo- acoustic signal amplitude a t 1.55 pm from cotton alone (C) versus percentage of cotton present in the sample; (B) graph of the contribution to the photoacoustic signal amplitude a t 1.72 pm from polyester alone (P) versus percentage of polyester present in the sample. 0.6 0.5 0.4 Q- ' 0.3 0.2 0.1 0 0 1 2 3 4 Percentage cotton1 (1 00 - percentage cotton) Fig. 3. Graph of the ratio C / P versus percentage cotton/( 100 - percentage cotton). Observed results for polyester - cotton fabrics containing (A) 70% cotton; (B) 50% cotton; (C) 40% cotton; and (D) 17% cotton.1484 ASHWORTH, KIRKBRIGHT AND SPILLANE which these indicate.The numerical results are summarised in Table I; these show that the values obtained are in general agreement with the composition reported by the manufacturer. The relative standard deviation for these results (ten replicates) was calculated to be ca. 0.1. This relatively high value for the relative standard deviation arises partly from the number of measurements required to perform the analysis but also from uncertainty concerning the homogeneity of the samples themselves ; this naturally renders representative sampling difficult. It was noted that most of the textiles used in the study displayed inhomogeneity in the weave of the fabric.Further, several of the fabrics incorporated raised patterns and measurements made on raised and plain parts of the same fabric indicated that a larger amount of polyester was present in the patterned parts compared with the plain areas. It should also be noted that the sample grinding itself may introduce errors as a result of mechano- chemical degradation. RESULTS FOR TABLE I POLYESTER - COTTON DETERMINATIONS Cotton, % Sample A . . .. . . B .. .. .. c .. .. .. D . . .. .. Manufacturers’ stated I composition 70 50 40 17 Conclusions The results presented indicate that binary mixtures Photoacoustic method 73.5 51.8 31.0 14.9 of textiles may be analysed successfullv using near-infrared photoacoustic spectroscopy. The method @;en abov; uses powdereh samples; measurements can be made, however, on whole (unground) samples of woven materials. This method was employed in an attempt to ensure representative sampling of the whole fabric and to avoid problems when local inhomogeneities were observed.(Further refinements in methodology will be required in order to improve the accuracy and precision of the technique for tariff classification purposes.) References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1 1 . 12. 13. 14. 15. 16. 1 7 . 18. Castleden, S. L., Elliott, C. M., Kirkbright, G. F., and Spillane, D. E. M., Anal. Chem., 1979, 51, Ashworth, C. M., Castleden, S. L., Kirkbright, G. F., and Spillane, D. E. M., J . Photoacoust., 1982, 2152. 1, 151. Fishman, V. A., and Bard, A. J., Anal. Chem., 1981, 53, 102. Castleden, S.L., Kirkbright, G. F., and Menon, K. R., Analyst, 1980, 105, 1076. Jin, O., Kirkbright, G. F., and Spillane, D. E. M., APPZ. Spectrosc., 1982, 36, 120. Castleden, S. L., Kirkbright, G. F., and Long, S. E., Can. J . Spectrosc., 1982, 77, 245. Kirkbright, G. F., and Menon, K. R., Anal. Chirn. Acta, 1982, 136, 373. Adams, M. J., Beadle, B. C., Kirkbright, G. F., and Menon, K. R., Appl. Spectrosc., 1978, 32, 430. Spillane, D. E. )I., PhD Thesis, University of London, 1982. Chalmers, J. M., Stay, B. J., Kirkbright, G. F., Spillane, D. E. M., and Beadle, B. C., Analyst, 1981, “Methods of Test: Quantitative Analysis of Fibre Mixtures,” British Standard 4407 :1975. Sobolewski, M. L., and Hickie, T. S., J . Text. Inst., 1977, 68, 302. Sargent, D. &I., Berni, R. J., and Janssen, H. J., Text. Res. J . , 1976, 46, 763. Davidson, R. S., King, D., Duffield, P. A., and Lewis, D. M., J . Soc. Dyers Colour., 1983, 99, 123. Davidson, R. S., King, D., and Fraser, G., Paper presented at the Third International Conference Adams, M. J., Beadle, B. C., and Kirkbright, G. F., Anal. Chem., 1978, 50, 1371. Castleden, S. L., Kirkbright, G. F., and Long, S. E., Can. J . Spectrosc., 1982, 27, 245. Miller, R. G. J., and Willis, H. A., J . Appl. Chem., 1956, 6, 385. 106, 1179. on Photoacoustic Spectroscopy, Paris, April 1983. Received June 26th, 1983 Accepted J u l y 7th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801481

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, December, 1983, Vol. 108, pp. 1485-1489 1485 Static and Flow Injection Voltammetric Determination of Total Phosphate and Soluble Silicate in Commercial Washing Powders at a Glassy Carbon Electrode Arnold G. Fogg and Geoffrey C. Cripps and Brian J. Birch Chemistry Department, Loughborough University of Technology, Loughborough, Leicestershire, LE 1 1 3T U Unilever Research, Port Sunlight, Merseyside, L62 4XN Anionic detergents have been shown to lower markedly the differential-pulse voltammetric peak currents obtained for 12-molybdophosphate and 12- molybdosilicate in aqueous solutions, but when ethanol - water or acetone - water (1 + 1 V / V ) solutions are used there is no interference. Static differential-pulse voltammetric and flow injection voltammetric procedures have been developed for the determination of total phosphate (involving hydrolysis of polyphosphates) and soluble silicate in commercial washing powders using these partly aqueous conditions.The effect of increasing ethanol and acetone concentrations on the diff erential-pulse voltammograms has been studied. Keywords 1 Diflerential-pulse voltammetry ; flow injection analysis ; phosphate determination ; silicate determination ; washing powder Linear-sweep and diff erential-pulse voltammetric procedures have been reported previously for the determination, at a glassy carbon electrode, of phosphate as 12-molybdophosphate in aqueous solution,l and phosphate, silicate, arsenate and germanate in aqueous acetone solu- tion in which the /3-heteropolymolybdates are stabilised.2 Phosphate has been determined also as molybdovanadophosphate in aqueous solution.3 All these static methods have been adapted for use as flow injection voltammetric procedures, in which the pre-formed hetero- polyacids were injected into reagent blank as eluent3 In this paper, a study of the applications of these voltammetric procedures to the determina- tion of total phosphate and soluble silicate in commercial washing powders is reported.Detergents were found to lower considerably the voltammetric signal obtained for 12-molyb- dophosphate using wholly aqueous solutions and that for p-12-molybdosilicate using aqueous acetone solutions containing 10% V/V of acetone. On the other hand, interference did not occur in the determination of phosphate as /3-12-molybdophosphate in aqueous acetone solution containing 50% V/V of acetone.These levels of acetone are those which were recommended previous1y.l-3 In this study increasing the level of acetone in the silicate determination to 50% V/V was shown to remove the interference from detergents, as had been so with phosphate. In developing the procedures recommended here for the determina- tion of total phosphate and soluble silicate in commercial washing powders, the use of ethanol, which also stabilises the P-heteropolymolybdates and is effective in eliminating interference from the detergents, was preferred to acetone on the grounds of its lower volatility and flammability. Experimental Voltammetric Measurements Voltammograms were obtained using a PAR 174 polarographic analyser (Princeton Applied Research) with three-electrode operation (glassy carbon electrode, platinum counter electrode and calomel reference electrode), For differential-pulse voltammetry a sweep rate of 5 mV s-l, a pulse height of 50 mV and a pulse frequency of 0.5 s were used.Between scans the glassy carbon electrode was cleaned as described previously with 1 M sodium hydroxide solution, and occasionally, as required, by polishing with the addition of water or ethanol on a polishing cloth. Voltammograms were recorded with a Gould HR 2000 x - y - t recorder.1486 FOGG et al. : STATIC AND FLOW INJECTION VOLTAMMETRY Analyst, Vol. 108 Flow injection analysis was applied as described previ~usly.l-~ A Metrohm detector cell (EA 1096), fitted with a Metrohm glassy carbon electrode (EA 286), was used.The detector cell was partially immersed in electrolyte (0.01 M sulphuric acid solution) contained in a beaker and contact with the counter and reference electrodes used in the static work was made by means of salt bridges. Flow of eluent (approximately 2.5 ml min-l) was produced by gravity feed from a reservoir approximately 1 m above bench level and connected to the injection valve by means of tubing of 1.58 mm bore. The injection valve was connected to the detector cell by means of approximately 15 cm of 0.58 mm bore tubing. Eluents were de-gassed by means of a vacuum pump. Injections were made by means of a Rheodyne 5020 sample injection valve with 0.1-ml loop capacity. The glassy carbon electrode was cleaned daily with 1 M sodium hydroxide solution after approximately 50 injections or when changing eluents.The potential of the detector cell was controlled by means of the PAR 174 polarographic analyser; current peaks were recorded with the Gould HR 2000 x - y - t recorder. The use of a glassy carbon electrode in excellent condition is more critical with the static method and it is best to keep an electrode specifically for this purpose. During a static determination, i.e., the running of standard and sample solutions, clean the electrode with 1 M sodium hydroxide solution but do not polish it. For both static and FIA work fresh molyb- date reagent solutions give the best results but they can be used for up to one week. Reagents Dissolve 0.408 g of analytical-reagent grade potassium dihydrogen orthophosphate in water and dilute to 1 1 in a calibrated flask.This solution is 3 x M in orthophosphate. Prepare less concentrated standard solutions by dilution. Weigh accurately about 0.15 g of quartz into a platinum (or nickel) crucible and fuse it with about 2 g of sodium carbonate. Cool the melt, dissolve in water, dilute to 11 in a calibrated flask and store in polythene. (A 3 x M solution of sodium metasilicate was also used as a standard solu- tion after comparison with this solution.) Dissolve 15 g of sodium molyb- date dihydrate and 12 g of tartaric acid in water, add 45 ml of concentrated hydrochloric acid and dilute to 500 ml. A brown colour develops in the reagent solution after a few days; dis- card it 3 d after preparation.[Since completing this work it has been pointed out that ammonium molybdate (AnalaR grade) has a lower blank value for phosphate, silicate and arsenate than has sodium molybdate. The replacement of sodium molybdate by ammonium molybdate should not affect the method but would need to be tested.] Ammonium molybdate - sulphuric acid solution. Dissolve 40 g of ammonium paramolybdate heptahydrate in 500 ml of water in a polythene beaker and add 500 ml of 1 M sulphuric acid solution. Mix and store in polythene. Mannitol solution, 10% m/V. Dissolve 10 g of mannitol in water and dilute to 100 ml. Dissolution and Hydrolysis of Sample Dissolve an amount of commercial washing powder containing approximately 50-150 mg of total phosphate (as orthophosphate, 40-115 mg as P,O,) and 15-30 mg of soluble silicate (as SiO,) in about 90 ml of water and heat at 80-90 "C in a water-bath with stirring until the sample has dissolved.Trace amounts of insoluble silica may remain undissolved. Transfer the solution into a calibrated flask and dilute to 100 ml. Take an aliquot of this solution for soluble silicate analysis. To 20 ml of the solution contained in a flask add 5 ml of 2 M sulphuric acid and heat for 1 h on a boiling water-bath to hydrolyse the polyphosphates. Cool and dilute to 100 ml in a calibrated flask. Take an aliquot of this hydrolysed solution for total phosphate determina- tion. In this study, 0.5 g of detergent powder was taken, a 5-ml aliquot of the unhydrolysed solu- tion was taken for silicate determination and a 5-ml aliquot of the hydrolysed solution was taken for phosphate determination. For better sampling of possibly inhomogeneous deter- gent powders the dissolution of a minimum of 10 g of sample is recommended.Standard orthophosphate solution, 3 x M (285 pg ml-l of PO,3-). Standard silica solution, approximately 0.15 mg ml-l of SiO,. Sodium molybdate - tartaric acid - hydrochloric acid solution.December, 1983 Preparation of Calibration Graphs and Analysis of Samples Determination of silicate OF PHOSPHATE AND SILICATE IN WASHING POWDERS 1487 Add aliquots of standard silicate solution (0-10 ml) or an aliquot of unhydrolysed sample solution containing less than 1.8 mg of 50, to a mixture of 10 ml of ammonium molybdate - sulphuric acid reagent and 23 ml of ethanol.Mix and, after 15 min, add 10 ml of 10% mannitol solution, mix again and, after a further 15 min, dilute to 50 ml in a calibrated flask. Obtain a differential-pulse voltammogram of each solution scanning from 0 to + 0.6 V or inject 100 pl of the solution into a stream of blank solution prepared as above in larger volume but omitting the silicate. Prepare a calibration graph based on the middle and larger differen- tial-pulse peak at +0.28 V or the flow injection voltammetric signal obtained at +0.16 V. Determination of phosphate Add aliquots of standard phosphate solution (0-10 ml) or an aliquot of hydrolysed sample solution containing less than 3 mg of orthophosphate to a mixture of 10 ml of sodium molyb- date - tartaric acid - hydrochloric acid reagent and 15 ml of ethanol.Mix and, after 15 min, dilute to 50 ml in a calibrated flask. Obtain a differential-pulse voltammogram of each solution scanning from -0.05 to +0.5 V or inject 100 pl of the solution into a stream of blank solution prepared as above in larger volume but omitting the phosphate. Prepare a calibration graph based on the second and larger differential-pulse peak at +0.23 V or the flow injection signal at +0.15 V. Results Data illustrating the decrease in size of the differential-pulse voltammetric peaks of 12- molybdophosphate obtained using the purely aqueous procedure and of /3-12-molybdosilicate obtained using the aqueous acetone procedure in which the final concentration of acetone is 10% V/V, in the presence of sodium lauryl sulphate and sodium alkylbenzenesulphonate, are shown in Table I.On a mass basis suppression is greater with the alkylbenzenesulphonate. At higher concentrations of detergent the degree of suppression comes to a limiting value. TABLE I EFFECT OF ANIONIC DETERGENTS ON THE DIFFERENTIAL-PULSE VOLTAMMETRIC SIGNALS OBTAINED USING PREVIOUSLY REPORTED PROCEDURES Phosphate method (purely aqueous)* Silicate method (10% V/V acetone)t Concentration of sodium lauryl sulphate in measuredsolution/gF .. . . .. 0 0.08 0.16 0.24 0.40 0.60 0.80 0 0.12 0.30 0.60 1.15 2.30 DPVpeakcurrentlpA .. .. .. 335 220 155 155 155 150 150 128 120 116 77 65 64 Concentration of sodium alkylbenzene- sulphonate in measured solution/g 1-l . , 0 0.016 0.032 0.048$0.080 0.120 0.260 0 0.004 0.012 0.020 0.040 0.080 DPVpeakcurrentlpA .. .... 335 255 220 195 120 102 80 130 110 67.5 40 30 32.5 A C r -7 r > * Phosphate concentration of measured solution = 3 x t Silicate concentration of measured solution = 2 x lo-' M. M. Typical concentration of detergent in measured solution for commercial samples containing 15% m/m of detergent. For 12-molybdophosphate formed in the purely aqueous procedure1 the first and major differential-pulse voltammetric peak on scanning from zero in the positive potential direction is at +0.14 V, and this is used to determine phosphate. In the aqueous acetone procedure using 50% acetone small peaks are obtained at +0.14 and +0.33 V, but the major peak is at +0.23 V. The peaks at +0.14 and +0.23 V are quite distinct and the increase in importance of the latter peak at the expense of the former can be seen clearly when the amount of acetone added before the determination is increased.A similar result is seen with the addition of ethanol and the peak heights of both peaks are plotted against ethanol concentration in Fig. 1. Studies of the effect of ethanol concentration in the presence of surfactant have also been made and the effect of ethanol in overcoming the depression of peak height is clearly seen in Fig. 1. In the aqueous acetone method2 for silicate (10% V/V acetone) with a potential scan from1488 FOGG et al. : STATIC AND FLOW INJECTION VOLTAMMETRY Analyst, VoZ. 108 I 601 I I I I I 10 20 30 40 Ethanol concentration, % VrV Fig. 1. Effect of ethanol concentration on the differential-pulse voltammetric peaks of 12-molybdophosphate and on the suppression of the action of sodium alkylbenzenesulphonate.Orthophosphate concentration = 6 x M ; sodium alkylbenzenesulphonate con- centration = 0.02 g 1-1 (B and D only). A and B, peak at $0.14 V, dominant in aqueous solution, A, without detergent and B, with deter- gent. C and D, peak at +0.23 V, C, without detergent and D, with detergent. zero in the positive direction a small differential-pulse voltammetric peak appears at +0.14 V but the major signal is a doublet with peaks occurring at +0.28 and +0.34 V. The change in height of the first peak of this doublet for ethanol is plotted in Fig. 2, and the effect of ethanol in overcoming the interference of surfactants is also shown. Sodium perborate is present in many washing powders.In purely aqueous solutions sodium perborate above 0.05 g I-f in the measured solutions was found to interfere with the voltammetric peaks of the heteropolyacids and at 2.5 g 1-1 to suppress them fully. In the presence of acetone or ethanol, however, no interference was experienced from perborate, or from EDTA, which may also be present in washing powders. Results are shown in Table I1 for the determination of total phosphate and soluble silicate in samples of commercial washing powders. Comparison is made with results obtained routinely using the molybdenum blue colorimetric procedures. The aqueous acetone visible spectro- photometric methods, on which the voltammetric methods reported here are based, were developed by Chalmers and Sinclair as selective methods for determining phosphate and silicate in admixture^.^,' No mutual interference between phosphate and silicate was observed in the voltammetric procedures either previously2J or in this work at the levels normally found in washing powders.200 I I 1 I 1 I 10 20 30 40 Ethanol concentration, Yo V/V Fig. 2. Effect of ethanol concentration on the differential-pulse voltammetric peak at +0.28 V of 12-molybdosilicate and on the suppression of the action of sodium alkylbenzenesulphonate. Silicate concentration = 3 x M ; sodium alkylbenzenesulphonate con- centration = 0.020 g 1-1 (B only). A, Without surfactant and B, with surfactant.December, 1983 OF PHOSPHATE AND SILICATE IN WASHING POWDERS 1489 Discussion Anionic detergents have been shown to lower significantly the diff erential-pulse voltammetric peaks of heteropolymolybdates in purely aqueous solutions.As the lowering of signal occurs whether the surfactant is added before or after the formation of the heteropolyacid, the surfact- ant clearly does not inhibit the formation of the heteropolyacid but only its reduction. It is likely that this is caused by adsorption of the surfactant at the electrode surface. The addition of acetone or ethanol, which would be expected to decrease the adsorption of the surfactant on the electrode surface, does in fact remove the interference. TABLE I1 ANALYSIS OF COMMERCIAL WASHING POWDERS Soluble silicate Reproducibility, sample A Phosphate determination, determination % m/m of L I -7 % m/m of P,O,* Si0,’t Total phosphate, Soluble silicate, I L \ r % mlm of P,O, yo mlm of SiOa B C D E F G Routine value (colorimetric method) .. 19.9 6.91 18.5 9.0 18.6 3.91 4.09 6.41 Staticmethod .. .. .. 19.7* 7.05* 18.13 9.6$ lS.6$ 3.62 3.64 6.30 FIAmethod .. .. .. 19.8* 6.90* 18.1$ 9.63 18.23 3.56 4.10 6.07 phosphate determinations: static method, 2.5% ; FIA method, 0.9%. 2.2%; FIA method, 1.2%. * Six determinations with hydrolyses for each determination of total phosphate and silicate were carried out. Coefficient of variation for Coefficient of variation for silicate determinations: static method, t Coefficients of variation (4 determinations) were typically <2% for the static method and <1% for the FIA method. $ Determinations were made on a single hydrolysate of each sample. The method is put forward at this stage as a possible alternative to existing procedures.To obtain adequate sensitivity colorimetric methods usually incorporate the additional step of reduction of the heteropolyacid to molybdenum blue in a boiling water-bath, so that in this respect this method is simpler. The colorimetric method has been automated using an air- segmented continuous flow system and, more recently, using flow injection analysis4 The flow injection technique has been applied voltammetrically in this study but the heteropoly- acids were pre-formed before injection. A procedure for the determination of phosphate by direct injection of phosphate solution into a molybdate reagent stream has recently been re- ported from this laboratory,5 and procedures for the determination of phosphate and silicate by the direct injection of hydrolysed detergent powder solutions are currently being studied. The authors are grateful for financial support for G.C.C. by a Science and Engineering Research Council CASE award with Unilever Research. References 1. 2. 3. 4. 6. 6. 7. Fogg, A. G., and Bsebsu, N. K., Analyst, 1981, 106, 369. Fogg, A. G., Bsebsu, N. K., and Birch, B. J . , Talanta, 1981, 28, 473. Fogg, A. G., and Bsebsu, N. K., Analyst, 1981, 106, 1288. RkZiCka, J . , and Hansen, E. H., “Flow Injection Analysis,’’ John Wiley, New York, 1981, p. 165. Fogg, A. G., and Bsebsu, N. K., Analyst, 1982, 107, 566. Chalmers, R. A,, and Sinclair, A. G., Anal. Chim. Acta, 1965, 33, 384. Chalmers, R. A., and Sinclair, A. G., Anal. Chim. Acta, 1966, 34, 412. Received May 5th, 1983 Accepted July 7th, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801485

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

1490 Analyst, December, 1983, Vol. 108, pp. 1490-1494 Cholate Liquid Membrane lon-selective Electrode for Drug Analysis* Luigi Campanella, Lorenzo Sorrentino and Mauro Tomassetti Institute of Analytical Chemistry, University of Rome, 001 85-Rome, Italy A cholate liquid membrane electrode employing benzyldimethylcetyl- ammonium cholate as sensor was prepared, characterised and applied to the analysis of commercially available drugs containing cholanic acids. The results are comparable to those obtained using a benzoate electrode. Keywords : Cholate ; liquid membrane ; ion-selective electrode ; benzyldimethyl- cetylammonium cholate ; drug analysis The determination of cholic acids is becoming increasingly important for drug control and their quantitative determination is currently performed by high-performance liquid chroma- tography (HPLC) ,l thin-layer chromatography (TLC)2 and gas chromatography (GC)3 and by en~ymatic,~ spectrophotometric,5 radiochemical6 and calorimetric' methods.Each of these methods presents some problems, such as toxicity of the reagents, high cost and complexity of the apparatus or the procedure. In a previous study8 we proposed a potentiometric method based on the use of a liquid membrane electrode indicator containing tributylcetyl- phosphonium benzoate dissolved in nitrobenzene as sensor. The results obtained were of a To the electrometer t PTFE rinas AgCl PT F E housing Internal solution Liquid exchanger 6 PTFE discs Fig. 1. Liquid membrane electrode assembly. lower precision (but almost the same accuracy) compared with those obtained by traditional methods, but the method is both simpler and cheaper.In this paper we present a new liquid membrane electrode indicator, containing a quaternary ammonium cholate salt, benzyl- dimethylcetylammonium cholate, as sensor. This new sensor has been characterised and employed for the determination of the cholic acids in some commercial drugs. Results are compared with those obtained by a benzoate liquid membrane electrode and by the enzymatic - spectrophotometric method of T a l a l a ~ . ~ * Pap": presented at the Symposium on "Electroanalysis in Biomedical, Environmental and Industrial Sciences, Cardiff, 6 8 t h April, 1983.CAMPANELLA, SORRENTINO AND TOMASSETTI 1491 Experimental Reagents Cholic acid, sodium cholate, deoxycholic acid and chenodeoxycholic acid were supplied by Merck, ursodeoxycholic acid by Giuliani and lithocholic acid and benzyldimethylcetylammonium chloride (BDMCACl) by Fluka.All reagents for the enzymatic - spectrophotometric (ultraviolet) tests for bile salts, using a previously reported pro~edure,~ were provided by Nyegaard, Oslo. All reagents were of analytical-reagent grade. Apparatus a recorder (Varian G-14 A2) and an automatic burette (Radiometer ABU-11). calomel electrode was employed as reference electrode. were performed with a Perkin-Elmer 320 spectrophotometer. Potentiometric measurements were carried out using an electrometer (Radiometer PHM64), A saturated Spectrophotometric measurements Fig. 2. Comparison between the response of (a) the cholate electrode with that of (b) the benzoate electrode in standard solutions of sodium cholate with changing cholate concentration (C).A and A': C (initial) = 2.0 x M ; C (after dilu- tion) = 1.9 x 10-SM. B and B': C (initial) = 3.8 x 1 0 - 4 ~ ; C (after addition) = 9.0 x 1 0 - 4 ~ ; C (after dilution) = 3.8 x M. C and C': C (initial) = 4.0 x M ; C (after addition) = 6.8 x M ; C (after dilution) = 4.4 x 1 0 - 5 ~ . D and D': responses of the electrodes for successive increases of cholate concentra- tion, D, C = 6.7 x - 2.2 x 1 0 - 3 ~ ; D', C = 8.0 x 10-4 - 2.5 x 10-3 M. M ; C (after addition) = 4.5 x Procedure Benzyldimethylcetylammonium cholate (BDMCACh) is prepared by the reaction between commercially available BDMCACl in chloroform and an aqueous solution of cholic acid at pH 9. To the chloroform phase diethyl ether is added with stirring and the BDMCACh is1492 CAMPANELLA et al.: CHOLATE LIQUID MEMBRANE Analyst, VoZ. 108 precipitated. The product obtained is purified by recrystallisation and characterised by melting-point determination (104-106 "C) , elemental analysis, thermal analysis, TLC on silica gel, infrared spectroscopy, NMR spectroscopy and X-ray powder diffraction. The electrode assembly characteristics are as follows : electrode body, PTFE ; sensors, BDMCACh (C49H,505N.H20) ; membrane solvent, decan-1-01 (dielectric constant 8.1 relative to vacuum; viscosity 12.5 cP) ; membrane solution concentration, 0.01 M in BDMCACh; internal solution, sodium cholate 0.01 M, potassium chloride 0.01 M ; internal reference electrode, Ag - AgCl- C1-; PTFE discs; and Millipore supports, 1.3 x lop2 m diameter, 1 x m thickness and 2 x lO-'rn pore size.The electrochemical cell is operated under the following conditions: Ag - AgCl- 0.01 M KCl, 0.01 M sodium cholate I] 0.01 M BDMCACh in decan-1-01 11 solution under test 11 saturated calomel electrode. The arrangement details of the electrode are shown in Fig. 1. TABLE I SELECTIVITY CONSTANTS ACCORDING TO THE MOODY - THOMAS METHOD Background level of jn- Kchol- interference/M Benzoate . . . . 0.11 1 x 10-3 Acetate .. . . 8.30 x 1 x 10-2 Nicotinate . . . . 1.92 x 1 x 10-1 Citrate.. .. . . 3.80 x 10-4 1 x 10-2 Oxalate . . .. 4.00 x 10-4 1 x 10-1 Nitrate .. . . 1.37 x 10-2 1 x 10-2 Sulphate . . .. 1.77 x 10-4 1 x 10-1 Chloride .. .. 8.50 x 1.2 x 10-2 Phosphate . . . . 8.70 x 1 x 10-5 Hydroxyl . . . . 0.10 1.6 x 10-3 Results The electrode was checked for electrochemical and analytical characteristics using standard sodium cholate solutions. The response time was 10 s maximum. Fig. 2 shows the response of the cholate electrode compared with that of the benzoate electrode with varying cholate concentration. The linearity range was 4.00 x 10-5-0.01 M; the slope of the calibration graph was -0.0577 (& 0.0006) volts per decade of mean activity, at 16 "C; and the relative standard deviationlo was 0.5% over the same range. The accuracy, for sodium cholate solutions of concentrations ranging between 1 x and 0.01 M, differed according to the technique employed (i.e. titration method, direct potentio- metry, standard additions and Gran's plotllSl2) and the best results were obtained by a Gran's plot.The error was not higher than 3% compared with about 5.5% for the standard additions method and 6% for the other two methods. The electrode was also checked for selectivity constants relative to several common anions; the values, obtained according to the method of Moody and Thomas,l33l4 are shown in Table I. Values of the selectivity constants, K i j , were obtained by measuring the e.m.f. in a solution containing a fixed amount (see Table I) of the interferent j and a varied activity of the primary ion i for which the electrode is selec- tive. The value of Kij is calculated from K i j = ar/afv where z and y are the charges of i and j ; ai and a j are the values that correspond to the intersection of the part of the calibration TABLE I1 LINEARITY RANGE AND SLOPES OF THE CALIBRATION GRAPHS FOR CHOLIC ACIDS DETERMINATION Slope, volts per Acid decade of concentration Linearity range/M Cholic acid .. . . .. -0.0567 4.00 x 10-5-1.00 x 10-2 Deoxycholic acid . . . . . . -0.0598 3.98 x 10-4-3.98 x 10-3 Chenodeoxycholic acid . . .. -0.0593 2.00 x 10-4-3.16 x Ursodeoxycholic acid . . .. -0.0521 1.00 x 10-4-5.01 x 10-3 Lithocholic acid , . .. .. - 0.2000 2.61 x 10-5-7.94 xDecember, 1983 ION-SELECTIVE ELECTRODE FOR DRUG ANALYSIS 1493 TABLE I11 COMPARISON BETWEEN POTENTIOMETRIC DETERMINATIONS WITH A BENZOATE ELECTRODE, A CHOLATE ELECTRODE AND BY ENZYMATIC DETERMINATION OF CHOLANIC ACIDS I N COMMERCIAL DRUGS Each value is the mean of at least five determinations; values in parentheses are standard deviations (yo).Cholanic acid found, '$!, Cholanic Drug acid 1 Chenodeoxy- 2 3 cholic acid 4 Ursodeoxy- cholic acid 5 , Benzoate Choiate electrode electrode Cholate Enzymatic - Nominal (standard (standard electrode spectro- Differences between found and nominal values, yo value, addition addition (Gran's plot photometric & \ % method) (1) method) (2) method) (3) method (4) Method 1 Method 2 Method 3 Method4 71.4 62.1 (9.1) 75.0 (0.0) 76.4 (2.1) 76.4 (2.1) -13.0 +5.0 +7.0 +7.0 50.0 55.0 (8.1) 51.0 (7.1) 51.0 (8.2) 53.2 (3.0) +10.0 +2.O +2.0 +6.4 94.3 108.4 (9.8) 96.8 (9.7) 96.4 (9.9) 97.1 (2.9) +15.0 +2.7 +2.2 +3.0 83.3 82.1 (6.2) 85.8 (0.0) 85.0 (1.5) 82.0 (2.1) -1.4 +3.0 4-2.0 -1.8 56.6 53.8 (6.5) 57.4 (7.0) 58.3 (0.7) 56.6 (1.0) -4.9 +1.4 +3.0 0.0 graph that has an approximately zero slope.This part corresponds to the complete inter- ference by ion j with the Nernstian (or approximately Nernstian) part and corresponds to the electrode function for the primary ion i. For K i j < 1 the electrode responds preferentially to ion i and for Kij > 1 the electrode responds preferentially to ion j . In order to evaluate the possible uses of the electrode in the analysis of aqueous solutions containing the other cholanic acids (deoxy-, chenodeoxy-, ursodeoxy- and lithocholic acid) , all of biological and pharmaceutical interest , linearity concentration ranges and slopes were determined for these unconjugated acids (Table 11). Finally, the electrode was applied to the determination of two cholanic acids (chenodeoxy- cholic and ursodeoxycholic acids) , both of which are contained in a commercial drug used for dissolving biliar gallstones, using the following procedure. A weighed amount of each of five examined drugs was dissolved in water at pH 11 and after sedimentation of a little insoluble matter the solution was filtered and appropriately diluted.Each was then analysed TABLE IV COMPOSITIONS OF THE EXAMINED DRUGS Drug Component Content, yo mlm 1 Chenodeoxycholic acid 71.4 Corn starch 26.6 Aerosil 1.4 Magnesium stearate 0.6 2 Chenodeoxycholic acid 50.0 Lactose 38.4 Starch 2.0 Talc 2.0 Starch - sodium glycolate 6.0 Magnesium stearate 0.8 Precipitated silica 0.8 3 Chenodeoxycholic acid 94.3 Polyvin ylp yrrolidone 3.8 Colloidal silica 1.1 Magnesium stearate 0.8 4 Ursodeoxycnolic acid 83.3 Starch 10.0 Precipitated silica 3.3 Magnesium stearate 3.3 5 Ursodeoxycholic acid 56.6 Lactose 37.7 Pol y vin y lp yrrolidone 3.8 Magnesium stearate 1.1 Colloidal silica 0.81494 CAMPANELLA, SORRENTINO AND TOMASSETTI by an enzymatic - spectrophotometric and by a potentiometric method with a cholate electrode and a benzoate electrode.The better results for precision and accuracy were obtained by standard additions and Gran’s plot methods, using the cholate electrode. In Table I11 experi- mental data, obtained by this sensor, are reported and compared with those obtained by both a benzoate electrodea and enzymatic method.9 In Table IV the nominal percentage compo- sitions of all the examined drugs are reported.Conclusions From the results several conclusions were drawn, evidencing the superiority of cholate electrodes over benzoate electrodes. In general, faster response times are obtained using a cholate electrode-10 s was the maxi- mum time obtained. The accuracy in cholate standard solutions is almost the same as for a benzoate electrode and the best results are obtained by a Gran’s plot. The precision is higher for a cholate electrode than for a benzoate electrode and the linearity range is wider and the minimum detection limits are lower; also the values of the selectivity constants for anions are low enough to prevent any common interference when using the cholate electrode. Considera- tion of the slopes of the calibration graph for the five cholanic acids examined (Table 11) leads to the conclusion that the value of the slope increases as the number of hydroxyl groups in the steroid ring decreases or if one of these hydroxyl groups is in the p-position towards the plane of the ring.The results of drugs analysis are, however, less precise using a cholate electrode compared with an enzymatic - spectrophotometric method but comparably accurate, and the analysis with the cholate electrode is faster and cheaper than that by the enzymatic method; moreover it has no problems of reagent availability and storage. Finally, in comparison with other possible methods such as calorimetry’ and chromato- graphy,3 potentiometry is perhaps less precise but does not require expensive and complicated apparatus or need any pre-treatment of the sample.This work received financial support from the Italian C.N.R. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. References Sian, M. S., and Harding Rains, A. J., Clin. Chim. Acta, 1979, 98, 243. Brusgaard, A., Clin. Chim. Acta, 1970, 28, 495. Setchell, K. D. R., and Matsui, A., Clin. Chim. Acta, 1983, 127, 1. Talalay, P., Methods Biochern. Anal., 1960, 8, 119. Biader Ceipidor, U., Curini, R., D’Ascenzo, G., and Tomassetti, M., Thermochim. Acta, 1981, 46, 269. Minder, E., Karlaganis, G., Schmied, U., Vitins, P., and Gustav, P., Clin. Chim. Acta, 1979, 92, 177. Biader Ceipidor, U., Curini, R., D’Ascenzo, G., and Tomassetti, M., Thermochim. Ada, 1981, 46, 279. Campanella, L., Sorrentino, L., and Tomassetti, M., Anal. Lett., 1982, 15, 1515. Biader Ceipidor, U., Curini, R., D’Ascenzo, G., Tomassetti, M., Alessandrini, A., and Montesani, C., Irving, H. M. N. H., Freiser, H., and West, T. S., Editors, IUPAC, “Compendium of Analytical Mascini, M., Ion-Scl. Electrode Rev., 1980, 2, 17. Moody, G. J., and Thomas, J. D. R., Sel. Rev. Anal. Sci., 1973, 3, 59. Moody, G. J., and Thomas, J. D. R. , Editors, “Ion-selective Electrodes,’’ Merrow Technical Library, Moody, G. J., and Thomas. J . D. R., in Pungor, E., Editor, “Ion-selective Electrodes,” Symposium Received May 6th, 1983 Accepted July 27th, 1983 G . Ital. Chim. Clin., 1980, 5, 127. Nomenclature,” Pergamon Press, Oxford, 1978. Watford, 197 1. held a t MBtrafured, Hungary, 23-25 October 1972, Akademiai Kiado, Budapest, 1973, p. 97.

ISSN:0003-2654    

DOI:10.1039/AN9830801490

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst, December, 1983, Vol. 108, $p. 1495-1499 1495 Anion-exchange Method for Speciation of Arsenic and Its Application to Some Environmental Analyses John Aggett and Rabiya Kadwani Chemistry Department, University of Auckland, Auckland, New Zealand A relatively straightforward two-stage anion exchange method has been developed for the speciation of inorganic arsenic(V) and arsenic(III), mono- methylarsonic acid and dimethylarsinic acid. Arsenic(II1) and dimethyl- arsinic acid are eluted in succession with a carbon dioxide - hydrogen carbonate buffer a t pH 5.5. Separation of monomethylarsonic acid and inorganic arsenic(V) is then obtained by elution with carbon dioxide - sodium chloride solution a t pH 4.0-4.2. The method has been applied to the analysis of sediment interstitial waters, and three species of aquatic plants.Keywords : Anion-exchange method ; arsenic speciation ; environmental analysis In recent years several methods have been published for the separation and analysis of environ- mentally important arsenic species. In these methods separation has been based on selective generation of arsines,l92 sequential volatilisation of generated ar~ines,~-~ conventional ion- exchange chrornatography,6-11 high-performance liquid chromatography12 and ion chromato- graphy.12913 Analyses have been completed through the use of hydride gene~ation~y~s~s~~ and graphite furnace6,8,10-12 atomic-absorption spectroscopy, diff erential-pulse polar~graphy,~ d.c. discharge3 and microwave emi~sion.~ Although speciation by methods based on hydride generation is attractive for those systems to which it is applicable it is of course limited to the determination of those species which can be converted into volatile hydrides.Fortunately, the four species most commonly considered to be of environmental importance at the present time, i e . , arsenate, arsenite, monomethylarsonic acid, and dimethylarsinic acid are amenable to this form of analysis. However, conventional ion-exchange and ion-chromatographic methods appear to possess the potential advantage that it should be possible to extend or modify them to include the analysis of additional environmentally important arsenic species should that become necessary. The purpose of this paper is to report the development and application of a relatively simple anion-exchange method for the speciation of arsenate, arsenite, monomethylarsonic acid and dimethylarsinic acid.As these four arsenic species are weak acids the dissociation constants of which are quite different (Table I) it seemed that separation by anion-exchange chromato- graphy was both logical and possible, However, the first two published ion-exchange methods for speciation of arsenic697 both used cation-exchange chromatography. The mechanism for these separations has not been established and attempts in this laboratory to use them to separate inorganic arsenic( 111) and inorganic arsenic(V) were not successful. Subsequently, Henry and Thorpeg and other workers10911 have published methods that require the use of both cation-and anion-exchange columns.TABLE I DISSOCIATION CONSTANTS OF ARSENIC SPECIES Acid Pkai Pka2 pka, Arsenic acid .. .. . . 2.25 7.25 12.30 Arsenious acid . . .. . . 9.23 Monomethylarsonic acid . . . . 4.26 8.25 Dimethylarsinic acid . . . . 6.25 The method presented here is a two-stage single column anion-exchange method using hydrogen carbonate and chloride as eluate anions. These species appear to have no adverse effects in subsequent analytical procedures. Its successful application is dependent on careful control of pH. It has been applied to the speciation of arsenic in samples obtained during studies on the fate of arsenic released into the Waikato River from geothermal sources in its catchment. For developmental purposes analyses were performed by hydride generation atomic-absorption spectroscopy, although any extension to more general speciation would1496 AGGETT AND KADWANI : ANION EXCHANGE FOR SPECIATION Analyst, VoZ.108 require the use of a more general technique for analysis, such as graphite furnace atomic- absorption spectroscopy or inductively coupled plasma atomic-emission spectroscopy. Experimental Anion-exchange Resins and Procedures Four anion-exchange materials were chosen for the initial investigation, viz. Zerolit FF(ip) SRA 62 (BDH Chemicals Ltd.), Zerolit FF(ip) SRA 70 (BDH Chemicals Ltd.), Amberlyst A-26 (Aldrich) and Whatman DE52 (Whatman). Columns (25 cm long) were prepared in 25-ml burettes using small glass-wool pads to support the resin. In some of these experiments the length of the Whatman DE52 column was reduced to 12 cm.The columns were converted into the hydrogen carbonate form and then prepared for use by elution with the appropriate buffer until the effluent reached the required pH. They were then loaded with 10-ml aliquots of the solutions under study. These solutions normally contained 1 pg ml-l of arsenic. During the elution procedure 5-ml aliquots were collected with an ISCO, Model 273, fraction collector. These aliquots were subsequently analysed by a hydride generation atomic- absorption method1 in which generation was carried out in 1 moll-1 hydrochloric acid. Separ- ate standards were used for the different compounds. Ekates As theoretical considerations suggested that separations were most likely to be achieved where the pH of the eluate was between 4 and 7 three buffer media were investigated as possible eluates, viz., citrate, acetate and carbon dioxide - hydrogen carbonate. Carbon dioxide - hydrogen carbonate buffers were prepared by saturating de-ionised distilled water with carbon dioxide.The solution was left overnight to ensure that equilibrium solubility of carbon dioxide was achieved and then solid sodium hydrogen carbonate was added until the desired pH was reached. The citrate and acetate buffers were prepared by dissolving the appropriate amounts of analytical-reagent grade materials in de-ionised distilled water. The carbon dioxide - ammonium chloride eluate was prepared by dissolving ammonium chloride (10 g 1-l) in saturated carbon dioxide solution. The pH of this was in the range 4.04.2. Speciation in interstitial water from sediments Sediment samples were packed into 50-ml centrifuge tubes immediately after collection and the tubes sealed to prevent intrusion of air.Interstitial water was obtained by centrifuging the samples in a refrigerated centrifuge for 15 min at 10000 rev min-l. This was normally done within 4 h of sample collection. The interstitial water was preserved at pH 2 with hydrochloric acid until speciation was carried out. Immediately prior to speciation the pH of the samples was adjusted to 5.0-5.5 and trace amounts of iron(II1) were removed by extrac- tion into chloroform with acetylacetone; 10-ml aliquots were speciated. Iron(II1) was removed prior to speciation to avoid the possibility that at the pH of speciation it might precipitate as hydrous iron(II1) oxide and in doing so coprecipitate arsenic.Speciation in lakeweeds For speciation of arsenic in lakeweeds, fresh samples of lakeweed (100 g) were cut into small pieces and homogenised with 500 ml of de-ionised water. They were then centrifuged in a refrigerated centrifuge for 1 h at 15000 rev min-l. The supernatant liquid and residue were separated and stored at 4 "C until required for analysis. The supernatant liquid was passed through a 0.45-pm membrane filter immediately before speciation. The solid material obtained by centrifuging the lakeweeds was refluxed with methanol and ethanol in separate procedures. The extracts were evaporated to dryness and the residues re-dissolved in de-ionised water and speciated. Results and Discussion Preliminary experiments were conducted with various combinations of resins and eluates in the pH range 5-7.In these the only promising separations were obtained with carbon dioxide-hydrogen carbonate as the eluate and this was used as the basis for subsequent development .December, 1983 OF As AND APPLICATION TO ENVIRONMENTAL ANALYSES 1497 The behaviour of arsenic(II1) and dimethylarsinic acid on SRA 70 as a function of the pH of the carbon dioxide - hydrogen carbonate buffer is shown in Fig. 1. Neither monomethyl- arsonic acid nor arsenic(V) was eluted by 250 ml of eluate in the pH range 4.8-6.4. These observations are understandable in terms of the dissociation constants of the respective acids and the nature of the buffer solution used for elution. Arsenic(II1) exists as an undissociated molecule over the pH range studied and as a conse- quence is eluted rapidly in a manner independent of pH.At lower pH the elution of dimethyl- arsinic acid overlaps that of arsenic(II1) but as the pH is raised there is greater tendency for dimethylarsinic acid to be retained by the resin, presumably a consequence of dissociation. This is reversed somewhat in the region of pH 6.0-6.4. The cause of this is believed to be the increase in hydrogen carbonate concentration in the eluent in this pH region. In practice, the hydrogen carbonate concentration increased from about 6 x mol 1-1 at pH 5.4 to about 6 x mol 1-1 at pH 6.4. T 2 pH 6.4 pH 4.8 10 20 30 Volume of aliquotlml Fig. 1. Separation of arsenic(II1) and dimethylarsinic acid on SRA70 by elution with carbon dioxide - hydrogen carbonate as a function of pH.1, Arsenic(II1) ; 2, dimethylarsinic acid. The results showed that separation of arsenic(II1) and dimethylars.inic acid by elution with carbon dioxide - hydrogen carbonate was satisfactory in the pH range 5.2-6.0. Elution of monomethylarsonic acid was accelerated and satisfactory separation from arsenic(V) achieved using saturated aqueous carbon dioxide (pH 4.04.2) containing 10 g 1-1 of ammonium chloride. Thus complete separation of the four species was obtained by eluting first with 150 ml (30 x 5-ml aliquots) of carbon dioxide - hydrogen carbonate buffer a t pH 5.5 & 0.3 followed by elution with a further 150 ml(30 x 5-ml aliquots) of carbon dioxide - ammonium chloride buffer solution at pH 4.04.2.Although the pH range for successful separation of arsenic( 111) from dimethylarsinic acid appears small, no problems have been encountered in meeting that specification during subsequent speciation procedures. Similar results were obtained with the other resins although the optimum pH for separation of arsenic( 111) from dimethylarsinic acid with carbon dioxide - hydrogen carbonate eluate varied: SRA 62, 6.0; SRA 70,5.5; A-26,5.2; and DE52, 6.0. The SRA 70 was preferred as it appeared to provide slightly better separation between arsenic(II1) and dimethylarsinic acid1498 AGGETT. AND KADWANI: ANION EXCHANGE FOR SPECIATION Analyst, V d . 108 TABLE I1 RECOVERIES OF ARSENIC SPECIES ON TREATED SRA 70 Species Mean recovery, yo* Arsenic(II1) .. .. .. 97.0 Arsenic(V) . . .. .. 98.1 Monomethylarsonate . . .. 98.8 Dimethylarsinate . . .. 94.8 * Four samples analysed. than did the others. In addition, peaks obtained with SRA 62, which has a lower percentage of cross-linking than SRA 70, were more inclined to tail. Elution was normally carried out at a flow-rate of 1 ml min-l. Increasing this to 4 ml min-1 did not appear to affect the separations. Initial investigation into quantitative aspects revealed a significant problem, i.e., that arsenic(II1) was oxidised to arsenic(V) during elution. This was indicated by the 70-80% recoveries for arsenic(II1) associated with the elution of low concentrations of arsenic in aliquots 35-45 when arsenic(II1) was eluted on its own. Although the cause of this problem was not positively identified the problem was removed when resins were treated with nitric acid (1 moll-l) and ethylenediaminetetraacetic acid (0.1 mol l-l, pH 5) before use. Recover- ies obtained with resin treated in this way are shown in Table I1 and the chromatogram of a synthetic mixture of the four species in Fig.2. The method has been applied to the analysis of the interstitial waters of sediments in Lake Ohakuri, and also to three species of lakeweed, Egeria densa, Largarosiphon major and Cerato- phyllum demersum, from the same lake as part of an investigation into the fate of arsenic released into the aquatic system from geothermal sources in the Waikato River catchment. No significant concentrations of methylated arsenic species were found in the interstitial waters examined and a typical sample (collected 7th July 1982) contained 0.52 pg ml-l of arsenic(II1) and 0.50 pg ml-l of arsenic(V).Chromatograms of the supernatant liquids obtained from centrifuging the lakeweed homo- genates showed that the two major arsenic species were inorganic arsenic(II1) and arsenic(V). No dimethylarsinic acid was detected in any of the supernatant samples analysed. It is possible that small amounts of monomethylarsonate (less than 3% of total arsenic) were present as very small concentrations of arsenic were detected in aliquots 3545 in most samples. However, no maxima were observed in the vicinity of aliquot 36, which suggests that these As (Ill 1 u 10 DMA 20 30 II I MMA I 40 50 L 60 Volume of a I iq u ot/m I Separation of a synthetic mixture of arsenic(II1) (0.5 pg), dime'hylarsinic acid (0.5 pg), monomethylarsonic acid (0.5 pg) and arsenic(V) (1 pg) on SRA7O.1499 December, 1983 OF As AND APPLICATION TO ENVIRONMENTAL ANALYSES 10(1 L .10 20 30 40 50 60 Volume of aliquotlml Fig. 3. Speciation of arsenic extracted by methanol from the solid phase of Largarosiphon major produced by centrifugation of the homo- genate. small peaks could be the result of oxidation of arsenic(II1) during elution. Attempts to clarify the situation by chromatographing samples at 4 "C were not successful. Only one of a number of solid phase lakeweed samples speciated appeared to contain a significant fraction of monomethylarsonic acid and none gave any indication of the presence of dimethylarsinic acid.The chromatogram of the Largarosiphon major sample containing monomethylarsonic acid is shown in Fig. 3. These applications indicate the usefulness of the method. By comparison with other ion- exchange methods for speciation of arsenic it is relatively simple and requires the use of eluates that are unlikely to interfere with subsequent quantitative analyses. Recoveries of the individual compounds are adequate and the only significant limitation appears to be associated with the problem of distinguishing between monomethylarsonic acid present in very small amounts and arsenic(V) formed by oxidation of arsenic(II1) on the ion-exchange column. At a flow-rate of 4 ml min-1 the ion-exchange procedure takes 75 min and it should be possible to make the procedure semi-automatic by coupling the ion-exchange system to the hydride generation system or other alternative analytical systems.Although dimethylarsonic acid was found in lakeweeds, dimethylarsenic acid might be found by other methods. Different digestion methods are at present being investigated for their effectiveness. The authors are grateful for the assistance of NZ Electricity Department personnel in collection of samples and Dr. E. White and Dr. M. H. Timperley, DSIR, for access to the centrifuge for separating interstitial waters from sediment solid phases. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. References Aggett, J . , and Aspell, A. C., Analyst, 1976, 101, 341. Howard, A. G., and Arbab-Zavar, M. H., Analyst, 1980, 105, 338. Braman, R. S., Johnson, D. L., Foreback, C. C., Ammons, J . M., and Bricker, J. L., Anal. Chem., Andreae, M. O., Anal. Chem., 1977, 49, 820. Talmi, Y., and Bostick, D. T., Anal. Chem., 1975, 47, 2145. Yamamoto, M., Soil Sci. SOC. Am. Proc., 1975, 39, 859. Dietz, E. A., and Perez, M. E., Anal. Chem., 1976, 48, 1088. Iverson, D. G., Anderson, M. A., Holm, T. R., and Stanford, R. R., Environ. Sci. Technol., 1979,13, Henry, F. T., and Thorpe, T. M., Anal. Chem., 1980, 52, 80. Pacey, G. E., and Ford, J. A., Talanta, 1981, 28, 935. Grabinski, A. A., Anal. Chem., 1981, 53, 966. Brinckman, F. E., Jewett, K. L., Iverson, W. P., Irgolic, K. J., Ehrhardt, K. C., and Stockton, Ricci, G. R., Shepard, L. S., Colovos, G., and Hester, N. E., Anal. Chem., 1981, 53, 610. 1977, 49, 621. 1491. R. A., J . Chromatogr., 1980, 191, 31. Received July 4th, 1983 Accepted August lst, 1983

ISSN:0003-2654    

DOI:10.1039/AN9830801495

出版商:RSC    

年代:1983

数据来源: RSC