ISSN:0003-2654    

DOI:10.1039/AN98207FX005

出版商:RSC    

年代:1982

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98207BX007

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

Febrzcary, 1982 SUMMARIES OF PAPERS IN THIS ISSUE ... 111Summaries of Papers in thisDevelopment and Optimisation of Atom Cells for SensitiveCoupled Gas Chromatography - Flame Atomic-absorptionSpectrometryThe development of four novel atom cells for the determination of volatileorganometallic compounds by coupled gas chromatography - flame atomic-absorption spectrometry is described. Tetraalkyllead compounds provided amodel system in the optimisation of the four atom cells by the variable step-size simplex method. The effects of the various parameters on analyticalperformance are discussed. In the most sensitive system presented theeffluent from the chromatograph was fed to a small hydrogen diffusion flameand the atoms from this flame were swept into a flame-heated ceramic tube.This simple, readily demountable arrangement enjoys the advantages ofcontinuous operation associated with flames but, because of the relatively longatomic residence times, gave detection limits of 17 pg for both tetraethyl- andtetramethyllead.These limits are superior to any previously reported for agas chromatographic - atomic-absorption spectrometric technique, includingthose employing electrothermal atomisation.Keywords : Gas chromatography ; flame atomic-absorption spectrometry ;simplex optimisation ; trace metal speciation ; alkyllead compound determi-nationL. EBDON and R. W. WARDDepartment of Environmental Sciences, Plymouth Polytechnic, Drake Circus,Plymouth, Devon, PL4 8AA.and D. A. LEATHARDDepartment of Chemistry, Sheffield City Polytechnic, Pond Street, Sheffield, S1 1WB.Analyst, 1982, 107, 129-143.Study of the Use of Soil Suspensions for the Determinationof Iron, Manganese, Magnesium and Copper in Soils by FlameAtomic-absorption SpectrometryA method is described for the direct, routine atomic-absorption spectrometricdetermination of copper, iron, manganese and magnesium in soil samples ofthe terra rossa and peat types.An investigation was made of the factors in-fluencing the atomisation efficiency of these elements when suspensions of soilsamples were aspirated into the flame. Particle size, flame temperature andposition in the flame were found to be critical in determining the fractions ofparticular elements atomised. Special emphasis was given to the preparationof the soil suspensions, which is the most critical step in the whole analyticalprocedure.Magnetic and ultrasonic devices were used for stirring purposes.The latter proved to be more efficient, particularly when suspensions of highclay content soils are being prepared. An average standard sample made foreach soil type was used for calibration. Test analyses of two sets of soilsamples showed that the majority (80-90%) of samples can be analysed with anaccuracy of f20%. This should be acceptable in most applications wherea large number of samples are to be analysed. Considerable amounts of timeand chemicals can be saved. The method was also found to be suitable for thedetermination of lithium, calcium, strontium, barium, aluminium, chromiumand titanium.Keywords : Iron, manganese, magnesium and copper determination ; soilsamples ; soil suspensions ; game atomic-absorption spectrometryJ. STUPAR and R.AJLECJoief Stefan Institute, E. Kardelj University, 61000 Ljubljana, Yugoslavia.Analyst, 1982, 107, 144-156iv SUMMARIES OF PAPERS IN THIS ISSUERapid Hydride Evolution - Electrothermal AtomisationAtomic-absorption Spectrophotometric Method for DeterminingArsenic and Selenium in Human Kidney and LiverFebruary, 1982A rapid semi-automated electrothermal atomisation atomic-absorption spectro-photometric procedure has been developed for the determination of arsenic andselenium in human liver and kidney specimens. The sample is digested with amixture of nitric and perchloric acids. The nearly dry residue is taken up inhydrochloric acid.The arsenic or selenium in the hydrochloric acid solution isconverted into its hydride with sodium tetrahydroborate(II1). The hydride isdecomposed and atomised in an electrically heated silica furnace, and theatomic-absorption signal is measured at the appropriate resonance wavelengthsof arsenic and selenium. Both of the elements can be determined in the bio-logical samples in the range 50-500 ng per gram of wet sample. The method wasapplied to the determination of arsenic and selenium in about 40 autopsysamples of human kidney (cortex and medulla) and liver taken from Canadianadults living in the Great Lakes Region of Ontario. The arsenic levels in allof the samples analysed were found to be < 10 ng per gram of wet sample; themedian selenium levels in the cortex, medulla and liver were found to be0.84, 0.31 and 0.39 mg per kilogram of wet sample, respectively.Keywords : Arsenic and selenium determination ; hydride evolution ; electro-thermal atomisation ; atomic-absorfition sfiectrophotometry ; human kidneyand liverK.S. SUBRAMANIAN and J. C. MERANGEREnvironmental Health Centre , Health and Welfare Canada, Tunney 's Pasture,Ontario, Canada, K1A OL2.Analyst, 1982, 107, 157-162.Reduction of Matrix Interferences in the Determination of Lead inAqueous Samples by Atomic-absorption Spectrophotometry withElectrothermal Atomisation with Lanthanum Pre-treatmentA previous paper described a lanthanum pre-treatment procedure for over-coming matrix interferences in the determination of lead in drinking water.In this paper, the amounts of lanthanum and nitric acid employed have beenoptimised such that the technique is now applicable to a wide range of aqueoussamples, for example river waters, borehole waters, sewage effluents and tradeeffluents. The technique has been tested and found to be satisfactory forsamples containing up to 1 150 mg 1-1 of chloride, 1420 mg 1-1 of sulphate,760 mg 1-1 of sodium and 1530 mg 1-1 total hardness (as calcium carbonate),The optimum pre-treatment conditions for samples was 1% V / V nitric acidand 0.05% m/ V of lanthanum (as lanthanum chloride), which completelyovercame suppressive interferences in the determination of lead, and gave afurnace tube lifetime of approximately 600 firings.This investigation enableda close study to be made of the processes by which interferences are overcome bynitric acid - lanthanum matrix modification, and some possible mechanisms arepresented and discussed.Keywwds : Lead determination ; aqueous samples ; lanthanum pre-treatment ;electrothermal atomisation ; interference mechanismsM. P. BERTENSHAW and D. GELSTHORPEDirectorate of Scientific Services, Severn-Trent Water Authority, NottinghamRegional Laboratory, Meadow Lane, Nottingham, NG2 3HN.and K. C. WHEATSTONEDirectorate of Scientific Services, Severn-Trent Water Authority, Abelson House,2297 Coventry Road, Sheldon, Birmingham, B26 3PU.Analyst, 1982, 107, 163-171February, 1982 SUMMARIES OF PAPERS IN THIS ISSUEA Study of Pneumatic Nebulisation Systems for InductivelyCoupled Plasma Emission SpectrometryThe analytical performance of different pneumatic nebulisers and cloudchambers for inductively coupled plasma emission spectrometry is reported.A vortex cloud chamber and a double-pass cloud chamber were compared foruse with concentric glass nebulisers.An all-plastic double-pass cloudchamber was preferred. Two new nebulisers for solutions containing highlevels of dissolved solids or slurries are described. One of these, machinedentirely from inert plastic, gave an improved performance compared with aglass concentric nebuliser and no problems were encountered with thenebulisation of solutions containing 20% m/V of dissolved solids or slurries.Keywords : Pneumatic nebulisers ; inductively coupled plasma ; atomic-emission spectrometry ; cloud chambers ; slurriesL.EBDON and M. R. CAVEDepartment of Chemistry, Sheffield City Polytechnic, Pond Street, Sheffield, S1 1WB.Analyst, 1982, 107, 172-178.VApparent and Real Reducing Ability of Polypropylene inCold-vapour Atomic-absorption Spectrophotometric Determinationsof MercuryExperiments undertaken with polypropylene reaction flasks in connectionwith the Perkin-Elmer MHS-1 MercuryjHydride System show that poly-propylene very quickly adsorbs tin(I1) chloride, which is not then removed byusual routine rinsing. This can cause mercury to be lost rapidly from solu-tions stored in uncleaned flasks as well as giving signals for mercury in cold-vapour atomic-absorption methods even if reductant is not added.Thisapparent ability of polypropylene to reduce mercury(I1) is rapid and ceaseswith careful cleaning of the apparatus. This apart, the polypropylene reac-tion flasks have a real, relatively slow, but considerable ability to reducemercury( II), which was especially conspicuous if solutions without preserva-tives were agitated in them.Keywords : Mercury (11) determination ; reduction by polypropylene ; atomic-absorption spectrophotometry ; cold-vapour methodA. KULDVEREGeological Survey of Norway, P.O. Box 3006, N-7001 Trondheim, Norway.Analyst, 1982, 107, 179-184.1,2,4,6-Tetraphenylpyridinium Perchlorate as a Reagent forIon-association Complex Formation and Its Use for theSpectrophotometric Determination of ThalliumThe synthesis, characteristics and applications of 1,2,4,6-tetraphenylpyri-dinium perchlorate (TPPP) as a reagent for the formation of ion-associationcomplexes is described.This reagent forms a 1 : 1 complex with TlC1,- (Amax.310 nm, molar absorptivity 3.14 x 104 1 mol-1 cm-1) that is slightly solublein water and can be extracted with isopentyl acetate with an extractionefficiency of 97.9%. The TlC14- - TPP+ complex obtained is useful for thespectrophotometric determination of thallium in the concentration range0.05-0.8 p.p.m., and the method is applicable to the determination of thalliumin sphalerites and zinc concentrates.Keywords : 1,2,4,6-Tetraphenylpyridinium perchlorate ; ion-associationcomplex ; thallium determination ; spectrophotometryT.PEREZ RUIZ, C. SANCHEZ-PEDRERO and J. A. ORTURODepartment of Analytical Chemistry, Faculty of Science, University of Murcia,Murcia, Spain.Analyst, 1982, 107, 185-189vi SUMMARIES OF PAPERS IN THIS ISSUESpectrophotometric Determination of Nitrate in Vegetable ProductsUsing 2-sec-ButylphenolFebruary, 1982A procedure is described for the determination of nitrate in vegetable products,based on the quantitative reaction of nitrate and 2-sec-butylphenol in sulphuricacid (5 + 7), and the subsequent extraction and measurement of the yellowcomplex formed in alkaline medium. The colour reaction is sensitive andstable and absorbances measured at 418 nm obey Beer’s law for concentrationsof nitrate-nitrogen between 0.13 and 2.50 pg ml-l. Various possible inter-ferents in vegetable products do not interfere with the nitration of 2-sec-butylphenol.Recoveries of nitrate from vegetable products were satisfactoryand the standard deviation for the whole procedure was 1.41% for the 42determinations; the detection limit of the method is 1.3 p.p.m. for nitrate-nitrogen.Keywords : Nitrate determination ; vegetable products ; 2-sec-butyl~henol ;spectrophotometryAKIO TANAKA, NORIHIDE NOSE and HISAO IWASAKISaitama Institute of Public Health, Kamiokubo-Higashi, 639-1, Urawa, Saitama,Japan.Analyst, 1982, 107, 190-194.Determination of Thiol Concentrations in Haemolysate byResonance Raman SpectrometryA simple and sensitive method for determining the thiol concentration inhaemolysate, using the resonance Raman spectrum of the product of thereaction of the lysate with 5,5’-dithiobis( 2-nitrobenzoic acid) (Ellman’sreagent), is described. The method uses a signal due to the haem ring ofhaemoglobin as an internal calibrant.In six separate determinations of thethiol concentrations in the lysate from a normal male volunteer, the averagethiol concentration was 1820 p~ and the relative standard deviation was4.5%. Standard addition of glutathione to four lysate samples confirmed thisresult and indicated an acceptable precision. To obtain a comparison withanother technique, clear lysates were prepared by precipitation of haemo-globin and addition of sufficient glutathione to maintain thiol levels comparableto those of coloured lysates. Agreement between the resonance Ramanmethod and a spectrophotometric method was acceptable in this instance.The method is relatively free from the standard interferences found forspectrophotometric assays and is selective. The mean value for the lysate fromsix normal male volunteers (1920 f 88 p ~ ) and from seven patients withrheumatoid arthritis (3 240 f 774 p ~ ) were significantly different, suggestingthat this measurement may have relevance in the study of disease processes.Keywords : Thiol determination ; haemolysate ; resonance Raman spectrometryJ. C. BANFORD, D. H. BROWN, A. A. MCCONNELL, C. J. MCNEIL andW. E. SMITHDepartment of Pure and Applied Chemistry, University of Strathclyde, ThomasGraham Building, Glasgow, G1 1XL.R. A. HAZELTON and R. D. STURROCKCentre for Rheumatic Diseases, Baird Street, Glasgow, G4 OEH.Analyst, 1982, 107, 195-199

ISSN:0003-2654    

DOI:10.1039/AN98207FP013

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

February, 1982 SUMMARIES OF PAPERS IN THIS ISSUEStudies in Chemical Phase Analysis. Part 111. Determination of theSolubilities of Certain Elements and Compounds Pertinent to Steels inOrganic Solvent - Halogen Mixtures with Particular Emphasis onManganese Silicon NitrideviiTo assist in the development of methods for the determination of manganesesilicon nitride in steels containing aluminium or niobium nitrides, the solubili-ties of iron and manganese silicon nitride have been determined in methanoland methyl acetate each containing a halogen or an interhalogen compound.Such studies indicate that the best solvent for the determination of aluminiumnitride or niobium nitride in the presence of manganese silicon nitride is iodinetrichloride - methyl acetate under reflux.However for the isolation ofmanganese silicon nitride together with the more stable nitrides, iodine -methyl acetate is the best solvent.The solubilities of iron and certain other elements and of some compounds ofiron and manganese have been determined in iodine - methanol solution( 1 + 10 m/V). The solvent is recommended for the isolation of iron(I1) andmanganese(I1) oxides from steels but iron(I1) and manganese(I1) sulphides andcementite are appreciably attacked by the solvent.Keywords : Iron and manganese silicon nitride solubilities ; ovganic solvent -halogen mixtures ; steelsG. T. ABOU ZEID and J. B. HEADRIDGEDepartment of Chemistry, University of Sheffield, Sheffield, S3 7HF.Analyst, 1982, 107, 200-205.Determination of Residues of Synthetic Pyrethroids in Fruit andVegetables by Gas - Liquid and High-performance LiquidChromatographyA multi-residue method for the determination of synthetic pyrethroids in fruitand vegetables is presented.After extraction with hexane - acetone, thepyrethroids are separated from co-extractives by a partition process andchromatography on a silica gel column and quantitatively determined byelectron-capture gas - liquid chromatography and/or high-performance liquidchromatography using an ultraviolet spectrophotometric detector.Keywords : Synthetic pyrethroids ; gas - liquid chromatography ; high-perform-ance liquid chromatography ; fruit and vegetables ; residuesP. G. BAKER and P. BOTTOMLEYDepartment of Industry, Laboratory of the Government Chemist, Cornwall House,Stamford Street, London, SEl 9NQ.Analyst, 1982, 107, 206-212.Selective Spectrophotometric Determination of Cephalosporinsby Alkaline Degradation to Hydrogen Sulphide and Formation ofMethylene BlueShort PaperKeywords : Cephalosporins ; alkaline degradation ; hydrogen sulphide ; methyl-ene blueM.A. ABDALLA and A. G. FOGGChemistry Department, Loughborough University of Technology, Loughborough,Leicestershire, LE11 3TU.and C. BURGESSGlaxo Operations (UK) Ltd., Barnard Castle, County Durham, DL12 8DT.Analyst, 1982, 107, 213-217viii SUMMARIES OF PAPERS IN THIS ISSUEDetection and Determination of Coumarin in Hydrocarbon OilsFebruary, 1982Short PafierKeywords : Hydrocarbon oils ; markers ; coumarin determination ; fluorescenceD.B. LISLE, J. MACNAB and K. L. H. MURRAYDepartment of Industry, Laboratory of the Government Chemist, LGC Unit,National Physical Laboratory, Queens Road, Teddington, Middlesex, TW11 OLW.Analyst, 1982, 107, 217-220.Fluorescence Cell Design and Use to Determine Cryde Oil in WaterShort PaperKeywords : Spectrojluorimetry ; fluorescence cell design ; crude oil ; waterfiollutionP. JOHN, E. R. McQUAT and I. SOUTARDepartment of Chemistry, Heriot-Watt University, Riccarton, Currie, Edinburgh,EH14 4AS.Analyst, 1982, 107, 221-223.Simple, Sensitive and Simultaneous Determination of Some SelectedInorganic Anions by High-performance Liquid ChromatographyShort PaperKeywords : Anion determination ; high-performance liquid chromatography ;ultraviolet detectionJ. P. DE KLEIJNFood Inspection Service, Westerbrink 3, 9405 B J Assen, The Netherlands.Analyst, 1982, 107, 223-225.Liquid - Liquid Extraction of Indium(II1) and Thallium(II1) fromSuccinate Solution with Trioctylamine pndTricaprylmethylammonium SaltShort PaperKeywords : Indium(III) and thallium(III) separation ; solvent extraction ;liquid ion exchangers ; succinate solutionS. D. SHETE and V. M. SHINDEAnalytical Laboratory, Department of Chemistry, Shivaji University, Kolhapur416 004, India.Analyst, 1982, 107, 225-230

ISSN:0003-2654    

DOI:10.1039/AN98207BP017

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

FEBRUARY 1982 The Analyst Vol. 107 No. 1271 Development and Optimisation of Atom Cells for Sensitive Coupled Gas Chromatography = Flame Atomic-a bsorption Spectrometry L. Ebdon and R. W. Ward Department of Environmental Sciences, PZymouth Polytechnic, Drake Circus, Plymouth, Devon, PL4 8A A and D. A. Leathard Department of Chemistry, Shefield City Polytechnic, Pond Street, Shefield, S1 1 WB The development of four novel atom cells for the determination of volatile organometallic compounds by coupled gas chromatography - flame atomic- absorption spectrometry is described. Tetraalkyllead compounds provided a model system in the optimisation of the four atom cells by the variable step- size simplex method. The effects of the various parameters on analytical performance are discussed.In the most sensitive system presented the effluent from the chromatograph was fed to a small hydrogen diffusion flame and the atoms from this flame were swept into a flame-heated ceramic tube. This simple, readily demountable arrangement enjoys the advantages of continuous operation associated with flames but, because of the relatively long atomic residence times, gave detection limits of 17 pg for both tetraethyl- and tetramethyllead. These limits are superior to any previously reported for a gas chromatographic - atomic-absorption spectrometric technique, including those employing electrothermal atomisation. Keywords : Gas chromatography ; pame atomic-absorption spectrometry ; simplex optimisation ; trace metal speciation ; alkyllead compound determi- nation I t is now generally recognised that the form of occurrence of trace metals is a primary factor in controlling their behaviour and fate in the environment and that different chemical and physical species of trace metals have different toxicological properties.This recognition has given an impetus to trace metal speciation studies. An obvious way forward in such speciation studies lies in utilising the power of separation of the various chromatographic techniques. Having achieved separation, detection of the species becomes of major importance. Detection methods for analytical gas and liquid chromatography typically provide either general or selective information. For metal-cont aining species atomic spectroscopy has the attributes that it can moaitor any of a large number of elements, with a high degree of sensitiv- ity and specificity, and it can be made compatible with the range of currently available chromatographic techniques.The use of atomic-spectroscopic detection for chromatography has been reviewed by several authors.lS2 We have recently reviewed a number of environ- mental applications and the advantages of different approaches to hybrid chromatographic - atomic spectroscopic trace metal ~peciation.~ The use of such specific detection allows less than optimum chromatographic separations to be tolerated with a consequent saving in time for sample clean-up and analysis. If two species co-elute and only one contains the metal of interest, the use of metal-specific detection means that only the metal-containing species is detected.Thus complete chromatographic separation is not required, only separation of the species containing the metal of interest being necessary. We have recently reviewed the particular advantages and disadvantages of various chroma- tographic - atomic spectroscopic coupling^,^ and therefore in this paper consideration is con- fined to coupling gas chromatography with atomic-absorption spectrometry. Flame atomic- absorption spectrometry offers simplicity of atom cell and operation. The instrumentation 129130 EBDON et aZ. : DEVELOPMENT AND OPTIMISATION OF Analyst, VoZ. I07 is relatively inexpensive and is readily available in most laboratories concerned with environ- mental monitoring. Flame atomic-absorption spectrometry suffers, in most published reports, from relatively high detection limits.Atomic-absorption spectrometry with electrothermal atomisation utilises more expensive instrumentation in which the atom cell is primarily de- signed for small (10-100 pl), discrete, condensed-phase samples, and the heating of such devices, normally a graphite furnace, is typically not continuous. Hence modification of the atom cell is required before it can accept a continuous gas stream from a gas chromatograph. The main advantage of this detection system lies in the excellent limits of detection obtained, although the technique yields short linear working ranges. Probably the most obvious way to link a gas chromatograph with an atomic-absorption spectrometer is to pass the effluent directly into the nebulisation chamber of the flame. This was first demonstrated by Kolb et using a short piece of heated tubing. With a standard burner system it was possible to separate and detect specifically the various alkyllead com- pounds in petrol.Various workersk7 have utilised similar couplings to determine tetraalkyl- lead compounds in petrol and air. Morrow et aZ.* also used a heated interface from a gas chromatograph to both flame atomic-absorption and emission detectors to measure silicon using an oxygen - acetylene or dinitrogen oxide - acetylene flame. This permitted the identification of silylated species in the presence of non-silylated species. The gas-chromatographic effluent was passed directly to the burner head by a heated metal transfer line connected to a gas union threaded into the side of the burner.This approach minimised peak broadening due to mixing with the large gas volume in the nebulisation chamber, encountered in the previous coupling systems. A short manifold positioned inside the burner head was used to distribute the chromatographic effluent evenly along the air - acetylene flame. Cokers used this system to separate and identify the individual alkyllead compounds in petrol. Wolf10 used the same type of interface for the speciation of various volatile chromium chelates. have utilised graphite furnaces as the atom cell for the determination of tetraalkyllead compounds. These all give superior detection limits to the flame cells reviewed above, presumably as a result of the longer residence times of the atoms in the furnace atomiser.Quartz T furnaces, both flame and electrothermally heated, have also been used,l4S15 again with improved detection limits compared with the above flame cells. Bye et aZ.7 demonstrated that for the furnace technique the detectability for tetraethyl- and tetramethyl- lead was improved 75- and 100-fold, respectively. Radziuk et aZ.14 showed that as a gas- chromatographic detector for alkyllead compounds a graphite furnace is 50 times more sensitive than a flame. Flame and electrothermally heated quartz furnaces have been found to be 15 times more sensitive than a conventional flame.14 Hence it appears that graphite furnace atomic-absorption spectrometry has been considered to be the most sensitive gas- chromatographic detector for alkyllead compounds.The best reported detection limits appear to be those of De Jonghe et aZ.,16917 of 90 and 40 pg for tetraethyl- and tetramethyllead, respectively, using a graphite furnace atomiser. Cokers used a slightly different method of interfacing. Various Preliminary Considerations In conventional flame atomic-absorption spectrometry of solutions the nebulised sample is continuously entering the atom cell. Hence the residence time of the atoms in the radiation path is not critical in achieving an analytical response. In coupled gas chromatography- atomic-absorption spectrometry, even though there is a continuous gaseous stream entering the atom cell, the sample itself is transient, so that the residence time of atoms in the atom cell is a vital consideration.If the problem of the short residence time in the flame were to be overcome, then it appeared to us that advantage could be taken of the simplicity of the flame coupling without undue deterioration in detectability. The determination of alkyllead compounds, in addition to being important from an environ- mental and toxicological viewpoint, offers a suitable model system for trace metal speciation studies. Both standards and samples are available that require only dilution with a suitable solvent prior to analysis. Preliminary work in which the effluent from the gas chromatograph was delivered to the mixing chamber of a conventional air - acetylene flame for atomic-absorption spectrometry confirmed that although such an approach was feasible it involved unnecessary dilution of theFebruary, 1982 ATOM CELLS FOR COUPLED GC - FLAME AAS 131 effluent.Detectability was immediately improved when the effluent was delivered directly to the flame, and at this stage the best results were obtained by attempting to direct the effluent along the slot of a 10-cm air - acetylene flame, as shown in Fig. 1. Visual monitoring of the luminous flame produced as the solvent flare impinged on the flame in this arrangement, termed atom cell I in this paper, indicated that only approximately 3 mm of the available path length was being utilised. Clearly, a slower burning flame would be expected to allow more of the available path length (10 cm) to be used. The air - propane flame has a burning velocity of 0.45 m s-1, compared with 1.6 m s-l for the air - acetylene flame.18 The use of an air - propane flame gave an immediate %fold increase in the response to lead.The increased lead response gained corresponded to the increased residence time of the atoms in the light path. It was considered that the response might be further improved by using a tube to hold the atoms in the atom cell longer. The use of such tubes has long been recognised,lg and more recently Watling20 used a slotted tube in an atomic-absorption flame to increase not only sensitivity but also precision. Delvesz1 used a tube suspended above the flame in conjunction with a micro-sampling system to give increased sensitivity by increasing atom residence times in the light path. Therefore, in an effort to increase atom residence times, atom cell I1 (Fig.2) employed a ceramic tube suspended over the atomic-spectrometer burner. Preliminary experiments with this atom cell showed that a 5-fold increase in response to lead over atom cell I had been achieved without optimisation of the system. Fig. 1. Atom cell I. A, Air - propane burner head; B, steel clip; and C, glass- lined interface tube. Fig. 2. Atom cell 11. A, Ceramic tube; B, air - acetylene burner head; C, aluminium supports; and D, glass-lined interface tube. A further atom cell (atom cell 111, Fig. 3) retained the ceramic tube to help increase residence times, and incorporated the refinement of the interface tube passing through the burner and Fig. 3. Atom cell 111. A, Ceramic tube; B, air - acetylene burner head; C , aluminium supports; and D, glass-lined interface tube.132 EBDON et al.: DEVELOPMENT AND OPTIMISATION OF Analyst, VoZ. 107 terminating just above the burner slit directly below the hole in the ceramic tube. The advant- age over the atom cell I1 was that the nitrogen carrier gas possessed some directional impetus towards the hole in the ceramic tube. The distance above the burner slit at which to terminate the interface tube was set at 1 mm so as to ensure direct delivery of the effluent to the primary reaction zone of the flame rather than the unburnt gases. In atom cells I1 and I11 the air - acetylene flame performed two functions: firstly atomisa- tion of the alkyllead species and secondly heating the ceramic tube to prevent condensation of the atomic species. However, in these cells the exhaust gases of the flame passed down the ceramic tube.In an attempt to prevent this, and hence increase the residence time of atoms in the tube, in the final atom cell (atom cell IV, Fig. 4) the atomisation and tube-heating func- tions were separated. A small hydrogen diffusion flame, burning on the end of a glass-lined stainless-steel T-piece atomised the alkyllead species. The air - acetylene flame was used mainly to heat the ceramic tube but it also performed an auxiliary function of ensuring the hydrogen diffusion flame remained alight. Fig. 4. Atom cell IV. A, Ceramic tube: B, air - acetylene burner head; C, aluminium supports; D, glass-lined T-piece; E, hydrogen diffusion flame; F, effluent stream from column; and G, auxiliary hydrogen flow.Optimisation Studies True comparison of the analytical performance of different atom cells can only be made when those atom cells have been rigorously optimised. Unfortunately, this is not always done, probably because a factorial design of optimisation experiment may be both tedious and time consuming. If attempts are made to give certain factors priority the factorial optimisa- tion may be made more rapid but there is the attendant risk of not obtaining the true optimum. The modified simplex method of Nelder and Mead22 offers a most elegant and rapid solution to the problem. This method has recently been demonstrated as a viable optimisation technique in atomic spectroscopy by Ebdon et aZ.23 and reviewed by Deming and Parker.24 Nelder and Mead’s modification of the original sequential simplex optimisation procedure of Spendley et aZ.25 has been widely applied. This variable step-size simplex22 allows more rapid attainment of the optimum, prevents achievement of a false optimum and permits a closer definition of the optimum. If the number of continuously variable parameters is n, then an n-dimensional simplex is constructed, defined by n + 1 points in factor space, and each of these parameters varied according to the simplex algorithm.22 The choice of the initial step size is critical in this optimisation procedure. Yarbro and Deming26 have shown it to be preferable to initiateFebruary , 1982 133 the procedure with a large step size. This ensured the exploration of maximum factor space prior to the simplex collapsing on to the optimum.These workers26 described a matrix and the accompanying equations that can be used to construct the initial simplex. It was therefore decided to optimise, for the alkyllead model system, the four atom cells out- lined above using the variable step-size simplex method, in order to compare their utility for coupled gas chromatography - atomic-absorption spectrometry. Peak height was chosen as the criterion of merit and, subject to base-line separation being achieved, no consideration of resolution or column efficiency was attempted. It was considered that by using peak height rather than peak area a better indication of the system likely to yield the lowest limit of detection would be given. The simplex was terminated when no significant difference was seen in peak height between successive new vertices.A univariate search,27 in which n-1 of the parameters were held constant and the remaining one varied across its allowed range of values as the peak height was measured, was used to confirm the success of the simplex optimisation. This search also indicated the influence of each parameter on the performance of the atom cell. The results and insights gained from this optimisation are discussed below. ATOM CELLS FOR COUPLED GC - FLAME AAS Experimental Apparatus An SP192 atomic-absorption spectrometer (Pye Unicam, Cambridge) provided with a deuter- ium-arc background corrector and a rapid response interface (SP198) was used as the detector. The separation of the alkyllead compounds was achieved using a Series 104 gas chromatograph (Pye Unicam) fitted with a glass column (1.5 m x 4 mm i.d.) packed with 5% Carbowax 20M on Chromosorb 750 (80-100 mesh).Samples were injected directly on to the column using a microlitre syringe (Scientific Glass Engineering, Melbourne, Australia). The output from the rapid response interface was displayed on a chart recorder (Philips PM9221; Pye Unicam) and processed by a computing integrator (DP88; Pye Unicam) capable of yielding peak-height or peak-area information in addition to retention times. The interface between the gas chroma- tograph and atom cell was achieved using glass-lined tubing (0.76 mm i d . ; Phase Separations, Queensferry, Clwyd) connected via a stainless-steel low hold-up union. The interface was heated by a method similar to that of Quimby et aZ.28 with the temperature monitored by a thermocouple.The ceramic tubes used in atom cells 11, I11 and IV were constructed of recrystallised alumina ('l'hermal Syndicate, Wallsend, Tyne and Wear) and their dimensions are given in Table I. The hollow-cathode lamp current used was 4.0 mA and the 283.2-nm lead line was monitored with a band pass of 0.8 nm. The retention times for the two alkyllead species studied, obtained using optimum conditions for each atom cell (see Table 11), are given in Table 111. TABLE I CERAMIC TUBE SPECIFICATIONS Length/mm Outer diameter/mm Inner diameterlmm Hole diameterlmm 110 16.0 12.5 10.0 110 14.5 10.0 8.0 110 12.5 8.0 6.0 110 10.0 6.25 4.0 Reagents Standards were prepared from stock solutions of both tetraethyl- and tetramethyllead (Associated Oc tel, Ble tchle y) by dilution with 2,2,4-t rime th ylpent ane (analytical-reagent grade). Construction of Atom Cells The four atom cells outlined above were designed to be as simple as possible in order to aid rapid coupling and decoupling of the chromatograph and the spectrometer.In the first (atom cell I, Fig. l ) , the effluent from the gas chromatograph passed via the heated glass-lined tubing directly to the burner head, and impinged laterally on the analytical flame. Atom cell I1 (Fig. 2) had a ceramic tube supported over the analytical flame on aluminium rods. There was134 EBDON et a,?. : DEVELOPMENT AND OPTIMISATION OF Analyst, VoZ. I07 TABLE I1 CENTROID VALUES OBTAINED FOR THE SIMPLEX VARIABLES IN THE FOUR ATOM CELLS Atom cell Variable Nitrogen flow-rate/ml min-' .. .. .. .. Propane flow-ratell min-l . . .. .. .. Acetylene flow-rate/l min-' . . .. .. .. Air flow-rate/l min-' . . .. .. .. . . Hydrogen flow-rate/ml min-' .. . . .. Chromatographic column temperature/"C . . .. centre of optical path/mm . . .. .. .. Separation between air - propane burner and Ceramic tube - air - acetylene burner separation/ Ceramic tube - hydrogen diffusion burner separa- mm . . .. .. * . .. .. .. tionlmm . . .. .. .. .. .. I I1 0.11 - - 0.62 3.2 4.7 25 41 - - 175 163 4.1 - - 4.3 I11 IV 80 64 - - 0.52 0.54 3.9 3.7 40 165 159 - 4.3 10.9 - 0.25 TABLE I11 RETENTION TIMES OBTAINED FOR EACH ATOM CELL UNDER OPTIMUM CONDITIONS Retention time/s r A \ Atom cell Tetramethyllead Tetraethyllead I .... 44 64 I1 . . .. 31 50 I11 . . .. 16 26 IV .. .. 26 34 a hole in the bottom of the tube. The effluent from the gas chromatograph passed via the heated glass lined interface tube so that it impinged on the analytical flame at right-angles directly below the hole. In atom cell I11 (Fig. 3), the ceramic tube was arranged over the analytical flame as for atom cell I1 but the interface tube passed through the burner and emerged from the burner slit directly below the hole in the ceramic tube. In atom cell IV (Fig. 4), the ceramic tube arrangement over the analytical flame was the same for atom cell I1 except that the hole was in the side of the tube, ie., at right-angles to the burner head. The gas-chromatographic effluent passed along the interface tube to a T-piece into which an auxiliary flow of hydrogen was introduced.A hydrogen diffusion flame was burnt on the end of the T-piece. This flame was aligned with the centre of the hole in the ceramic tube. Initial Studies with Atom Cell I The effect of the normally constant parameters on analytical performance was considered for atom cell I. The length of the interface tube was varied between 0.18 and 0.8 m and no change in the lead response was found. If insufficient thermal insulation of the interface tube was used then, as its length increased, an increase in base-line noise was found. By ensuring that the interface tube temperature was isothermal with the oven the latter effect, apparently due to mains-borne noise arising from the thermostating of the oven, was eliminated. Inter- face tubes of different internal diameter (0.25,0.51,0.76 and 1.27 mm) were investigated.The best peak-height response to lead (see Fig. 5 ) was obtained for the 0.76 mm i d . tube and this diameter tubing was used for all subsequent work. Simplex Optimisation of Atom Cell I The continuously variable parameters investigated for this system were nitrogen carrier gas flow-rate, air flow-rate, fuel gas flow-rate, chromatographic column temperature and the distance of separation between the burner head and the centre of the optical path. The centroid of the optimum range predicted by the simplex optimisation is given for each variable in Table 11. That the optimum had been achieved was then confirmed using the univariate search procedure.Results and DiscussionFebruary, 1982 ATOM CELLS FOR COUPLED GC - FLAME AAS 136 6 2 14 0, a) r .- Y m 2 10 1 0 0.25 0.50 0.75 1.0 1.25 Interface tube internal diarneter/rnrn Fig. 5. Variation of peak height response for lead with internal diameter of interface tube. A, Tetramethyllead; and B, tetraethyl- lead. E y 10 r 0, a) t Y .- z 6 n 2 20 60 100 140 Nitrogen f low-ratelm I r n i n - ' 2 3 4 Air flow-rate/l rnin-' 0.1 0.2 0.3 Propane flow-rate/l min-' 80 120 160 2 6 10 14 Temperature/"C Separation/mrn Fig. 6. Confirmation of the simplex optimisation for the five variables studied for atom cell I. The shaded areas indicate the regions identified as optimum by the simplex method. (a) Nitrogen flow-rate; (b) air flow-rate; ( c ) propane flow-rate; ( d ) chromatographic column temperature; and (e) separation between the air - propane burner and the centre of the optical path.A, Tetramethyllead; and B, tetra- ethyllead.136 EBDON et al. : DEVELOPMENT AND OPTIMISATION OF Analyst, VoZ. 107 Fig. 6 [(a)-(e)] demonstrate the success of the simplex optimisation procedure. The shaded region on each graph identifies the optimum range predicted by the simplex procedure for each variable. Fig. 6(a) shows the carrier gas flow-rate to be a critical variable and confirmed the simplex-predicted optimum. At first sight it appeared strange that the optimum nitrogen flow-rate was so low, considering that the aim was to use as much of the available path length in the flame as possible. The low burning velocity of the air - propane flame conferred the advantage of longer atom residence times, but also a certain lack of laminarity.Thus high carrier gas flow-rates resulted in severe distortion of the flame profile, on occasion out of the light path, and hence yielded lower responses to lead. Fig. 6(b) and (c) illustrate an interesting effect at high oxidant to fuel gas ratios. The univariate search for oxidant flow-rate [Fig. 6 ( b ) ] , shows that at high air flow-rates the peak-height response to lead for tetraethyllead decreased, probably owing to a reduced flame temperature and hence reduced atomisation efficiency. This effect was not noticed for tetramethyllead. Although not observed using this atomic-absorption detector, the solvent peak overlaps the tetramethyllead peak and thus may act as a secondary fuel supply aiding atomisation.A similar effect was noted for the univariate search for propane flow-rate [Fig. 6(c)], where at less than 100 ml min-l of propane the peak height for tetraethyllead decreased whereas for tetramethyllead it increased. This again would appear to be related to the effect of the solvent on the atomisation process. The univariate search for the chromatographic column temperature [Fig. 6 ( d ) ] confirmed the pre- dicted optimum range. This temperature range was the highest possible compatible with full base-line resolution of the peaks whilst also yielding the most rapid analysis time. This be- haviour was also observed in the succeeding simplex optimisations. Fig. 6(e), showing the distance of separation of the burner head and the centre of the optical path, is consistent with the ready atomisation of lead even in the relatively cool air - propane flame and again confirms the success of the simplex procedure.Initial Studies with Atom Cell I1 The effect of ceramic tube diameter, being a non-continuous variable, was investigated prior to the simplex optimisation of the atom cell. From the results obtained using four different diameter tubes (see Table IV), it can be seen that the 6.25 mm i.d. ceramic tube gave a 2-fold increase in response to lead over the 12.5 mm i.d. tube. The fact that the smallest internal diameter ceramic tube gave the best response may be attributed to increased viscous drag from the tube walls and that the tube has less unilluminated space. TABLE IV EFFECT OF INTERNAL DIAMETER OF CERAMIC TUBE ON LEAD RESPONSE Relative lead response Ceramictube A \ i d ./mm Tetramethyllead Tetraethyllead 12.6 1.0 1.0 10.0 1.1 1.1 8.0 1.6 1.7 6.26 1.9 2.1 Simplex Optimisation of Atom Cell I1 Having chosen the 6.25 mm i.d. ceramic tube for use with this system, the continuously variable parameters were optimised. These variables were nitrogen carrier gas flow-rate, oxidant flow-rate, acetylene flow-rate, chromatographic column temperature and the separa- tion of the ceramic tube and the air - acetylene burner head. The centroid values for the optimum range obtained for each variable from the simplex optimisation are given in Table 11. The univariate searches [Fig. 7(a)-(e)] illustrate the success of the optimisation procedure; the shaded regions on the graphs again depict the optimum range of values in the final simplex set.The carrier gas flow-rate [Fig. 7(a)] proved to be an extremely critical variable with a narrow optimum range. The lateral velocity of the carrier gas at the centroid was 1.46 m s-1, which was just below the maximum burning velocity of the air - acetylene flame. As the lateral velocity increased above this value the possibility of the nitrogen gas stream passing through the air - acetylene flame increased, thus reducing the number of atoms entering the ceramic tube and hence reducing the response to lead. The univariate searches for the oxidant andFebruary , I982 ATOM CELLS FOR COUPLED GC - FLAME AAS 137 fuel gas flow-rates [Fig. 7 ( b ) and (c)] demonstrated both to be important variables with the optimum ranges being at those values which gave the most stoicheiometric flame.The univariate search for the chromatographic column temperature [Fig. 7 ( d ) ] confirmed the con- clusions made during the preceding optimisation. Fig. 7(e) demonstrates that the optimum ceramic tube - air - acetylene burner separation was the smallest possible that would allow a stable flame to be burnt. Minimisation of this separation obviously maximises the possibility of lead atoms entering through the hole in the ceramic tube. 20 15 . E 5 10 E 0 a, .- (0 a 5 20 60 100 Nitrogen flow-rate/ml min-' 20 15 5 20 15 . E E 0 a, Y a, a .- 1c 5 4 5 Air flow-rate/l rnin-' 0.3 0.5 0.7 Acetylene flow-rate/ I rnin- ' 60 100 140 180 Temperatu re/"C 20 15 .Eo E Y l o m al r .- a, a 5 2 6 10 Sepa rat ion/rn m Fig. 7. Univariate searches for the five variables studied in the simplex optimisation for atom cell 11. The shaded areas indicate the optimum ranges predicted by the simplex method. (a) Nitrogen flow-rate ; (b) air flow-rate ; (c) acetylene flow-rate ; (d) chromatographic column temperature ; and (e) separation between the air - acetylene burner and the ceramic tube. A, Tetramethyllead; and B, tetraethyllead.138 Analyst, VoZ. I07 Simplex Optimisation of Atom Cell I11 The continuously variable parameters of this system were the same as for the atom cell 11. The centroid values of the simplex predicted optimum ranges are given in Table I1 with the univariate searches in Fig. 8(a)-(e). The optimum range for the nitrogen carrier gas flow-rate EBDON et al.: DEVELOPMENT AND OPTIMISATION OF 20 15 2- 5 5 lo w r 0, .- m a 5 - 20 50 100 Nitrogen flow-rate/ml min-' 20 15 . 5 E Y 10 0) a AZ .- a a 5 20 15 . 5 2 lo 2 0) .- Y m P 5 3 4 5 Air flow-rate/l min-' 100 150 Temperature/"C Acetylene flow-rate/l min-' 5 10 Separatiodmm Fig. 8. Univariate search confirmation of the simplex optimisation for the five variables in atom cell 111. (a) Nitrogen flow-rate; (b) air flow-rate; (c) acetylene flow-rate ; (d) chromatographic column temperature; and (e) separation between the air - acetylene burner and the ceramic tube. A, Tetramethyllead; and B, tetra- ethyllead. The shaded areas indicate the optimum ranges predicted by the simplex method.25 20 E 3 15 t 0 a¶ .s Y m .- 10 5 20 60 100 Nitrogen flow-ratelm1 min-’ A I I I L 3 4 5 Air flow-rate/l min-’ 20 - 15 - E Y Y 10- s 01 a¶ r .- a¶ 0.5 - T T Acetylene flow-ratell min-’ :d) 15 E 3 E cn a¶ r .- Y 10 n a¶ 5 10 30 50 70 Hydrogen flow-rate/mI min-’ Fig. 9. Confirmation of the simplex optimisation for the seven variables studied for atom cell IV. The shaded areas indicate the regions identified as optimal by the simplex method. (a) Nitrogen flow- rate; (b) air flow-rate; (G) acetylene flow-rate; (d) hydrogen flow-rate ; (e) chromatographic column temperature; (f) separation between the air - acetylene burner and ceramic tube; and (g) separa- tion between the hydrogen diffusion burner and the ceramic tube. A, Tetramethyllead; and B, tetra- ethyllead. 50 100 150 200 Tem perat ure/”C 5 10 15 20 0 2 4 Separation/mm Separation/mm140 EBDON et al.: DEVELOPMENT AND OFTIMISATION OF Analyst, VoZ. 107 [Fig. S(a) J was much higher than in the previous system. High nitrogen flow-rates will give a more laminar gas column much less susceptible to directional fluctuations and hence giving the atoms a better chance of entering the ceramic tube. The slight advantage of atom cell I11 is presumably related to the increased control over the entry of atoms into the tube. Fig. 8(b) and (c) illustrate the critical nature of the oxidant and fuel gas flow-rates. The other optimum values were very similar to those observed for atom cell 11, except a smaller flame was preferred because the flame no longer swept the atoms into the tube, and rather the role of the flame gases in sweeping out the tube was emphasised.Simplex Optimisation of Atom Cell IV The continuously variable parameters for this atom cell were the same as for atom cell I1 with the addition of the hydrogen flow-rate and the hydrogen diffusion flame burner - ceramic tube separation. Fig. 9(u.)-(g) illustrate the success of the simplex optimisation procedure. Fig. 9(a) shows that the nitrogen flow-rate is a critical parameter; it can be seen that the simplex has found the flow-rate range that gives the best compromise peak-height values for both lead species. The apparent anomaly at low flow-rates for the tetramethyllead may be related to the co-elution of the solvent mentioned earlier. Fig. 9(b) and (c) illustrate that again the simplex has identified the optimum air and acety- lene flow-rates as those consistent with a stoicheiometric flame.Fig. 9(d) illustrates that the hydrogen flow-rate is the least critical of the gas flow-rates in this atom cell. Indeed, the major role of the hydrogen flame appeared to be to prevent the appearance of large solvent peaks, caused by uncorrected molecular absorption, and to aid the formation of reproducible narrow peaks. Fig. 9(f) shows that the pre- dicted optimum range for the ceramic tube - air - acetylene burner separation occupied part of the optimum range as shown by the univariate search. The occurrence of a wide optimum range makes it difficult, although not vital, to identify closely the optimum range. Fig. 9(g) indicates that the separation between the hydrogen burner and the ceramic tube is a critical parameter, but again this has been predicted successfully.It is understandable that as the separation increased the response to lead, for both tetraethyl- and tetramethyllead, decreased because of the increased chance of the atoms not entering the ceramic tube through the hole. The fact that lead response was decreased if the hydrogen diffusion burner was placed inside This effect is illustrated in Fig. lO(u) and (b). 0 1 2 I Min I I M- *hK T I t T t I w Time S \ r I T I A - T I 0 1 Min Fig. 10. Chromatograms of 2,2,4-trimethylpentane, tetramethyllead and tetraethyllead obtained with atom cell IV: (a) under optimum conditions; and (b) under the same conditions but with zero hydrogen flow-rate. A, Injections of 1.0 pl of 2,2,4-trimethylpentane; B, injection of 1.0 p1 of tetraethyllead (1.0 ng of lead) and tetramethyllead (1.0 ng of lead) in 2,2,4-trimethylpentane.I, Point of injection; S, 2,2,4-trimethylpentane; M, tetramethyiIead ; and E, tetraethyllead.February, 1982 ATOM CELLS FOR COUPLED GC - FLAME AAS 141 the ceramic tube (defined as a negative separation on the figure) suggests that the air - acetylene flame was not merely heating the ceramic tube. It appears that the air - acetylene flame also played a part in ensuring that the hydrogen flame remained burning so as to give efficient atomisation. Calculation of Detection Limits equation where XL is the limit of detection and SB the standard deviation of the blank. The use of a chromatographic technique gives the advantage that from the retention times of the species a signal may be expected at a certain time after injection of sample.Thus, for this coupled gas chromatographic - atomic-absorption technique it would seem reasonable to measure the background variation at the retention time of the chromatographic peak. As an example of this method, Fig. 11 shows the traces obtained using atom cell I for 10,5,4 and 3 ng of lead injected as tetraethyl- and tetramethyllead. This atom cell gave detection limits of 2.0 and 1 .O ng for tetraethyl- and tetramethyllead, respectively. In atomic spectroscopy, detection limits generally quoted in the literature are based on the x, = 2s, D 0 1 2 I Min 1 I B Time Fig. 11. atom cell I. species, and D, 3 ng as Pb of each species. Chromatograms of tetramethyllead and tetraethyllead near the detection limit with A, 10 ng as Pb of each species; B, 5 ng as Pb of each species; C, 4 ng as Pb of each M, Tetramethyllead; and E, tetraethyllead.The detection limits for the four atom cells are given in Table V. The introduction of the ceramic tube above the air - acetylene flame had a profound influence on the response of the atom cell to lead. Also, the separation of the atomisation process and heating of the ceramic tube functions, in atom cell TV, gave a further increase in the reponse to lead. Both increases in response were caused by increasing the residence time of the atoms in the atom cell. Conclusion The use of a flame atom cell as an atomic-absorption detector for gas chromatography has the advantages of simple and well understood operation and its continuous mode of operation142 EBDON et al.: DEVELOPMENT AND OPTIMISATION OF TABLE V Analyst, VoZ. 107 LINEAR WORKING RANGES AND DETECTION LIMITS (AS LEAD) FOR THE FOUR ATOM CELLS Detection limitlpg A I \ Atom cell Linear rangelng Tetramethyllead Tetraethyllead I .. .. 10-300 1000 2 000 I1 .. .. 1.0-50 58 75 I11 .. .. 0.8-20 48 71 IV .. . . 0.1-15 17* 17* * Equivalent to 82 fmol of compound. is ideal for dealing with continuous sample streams. The use of the ceramic tube device above the flame increased atom residence times so that the detection limits obtained were better even than those reported for graphite furnace systems.17 As the optimum height of the ceramic tube above the air - acetylene burner is now known, the atom cell has been further simplified by using a stainless-steel knife-edge support clip fixed on to the burner (Fig.12), which is readily demountable to permit rapid decoupling of the instruments. No additional sophisticated heating is required as for a furnace, useful tube lifetimes are in excess of 6 months and observ- able memory effects are absent. Hence further advantages of the tube-in-flame detector developed are low cost and great robustness. This atom cell is now being applied to a number of environmental applications. Fig. 12. Atom cell IV with fixed-height tube support. A, Ceramic tube; B, air- acetylene burner head ; C, stainless-steel knife-edge ceramic tube support; and D, glass-lined T-piece. We thank Pye Unicam Ltd. for the grant of equipment, Pye Unicam and the Science Research Council for an SRC CASE Studentship (to R.W.W.) and Associated Octel Ltd.for the provision of samples. 1. 2. 3. 4. 5. 6. 7. 8. 9. References Fernandez, F. J., At. Absorfit. Newsl., 1977, 16, 33. Van Loon, J . C., A~zal. Chem., 1979, 51, 1139A. Ebdon, L., Ward, R. W., and Leathard, D. A., Anal. Proc., in the press. Kolb, B., Kemmner, G., Schleser, F. H., and Wiedeking, E., Fresenius 2. Anal. Ckem., 1966,221, 166. Chau, Y. K., Wong, P. T. S., and Saitoh, H., J. Chromatogr. Sci., 1976, 14, 162. Katou, T., and Nagawa, R., Bull. Inst. Environ. Sci. Technol., 1974, 1, 19. Bye, R., Paus, P. E., Solberg, R., and Thomassen, Y., At. Absorpt. Newsl., 1978, 17, 131. Morrow, R. W., Dean, J . A., Shults, W. D., and Guerin, M. R., J. Chromatogr. Sci., 1969, 7, 572. Coker, D. T., Anal. Chem., 1975, 47, 386.Febraary , 1982 ATOM CELLS FOR COUPLED GC - FLAME AAS 143 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. Wolf, W. R., Anal. Chem., 1976, 48, 1717. Segar, D. A., Anal. Lett., 1974, 7 , 89. Robinson, J . W., Kiesel, E. L., Goodbread, J . P., Bliss, R., and Marshall, R., Anal. Chim. Ada, Radziuk, B., Thomassen, Y., Van Loon, J. C., and Chau, Y. K., Anal. Chim. Ada, 1979, 105, 255. Radziuk, B., Thomassen, Y., Butler, L. R. P., Van Loon, J. C., and Chau, Y . K., Anal. Chim. Ada, Chau, Y . K., Wong, P. T. S., and Goulden, P. D., Anal. Chim. Acta, 1976, 85, 421. De Jonghe, W., Chakrabarti, D., and Adams, F., Anal. Chim. Acta, 1980, 115, 89. De Jonghe, W. R. A., Chakrabarti, D., and Adams, F. C., Anal. Chem., 1980, 52, 1974. Price, W. J ., “Spectrochemical Analysis by Atomic Absorption,” Heyden, London, 1979. Fuwa, K., and Vallee, B. L., Anal. Chem., 1963, 35, 94. Watling, R. J., Anal. Chim. Acta, 1978, 97, 395. Delves, H. T., Analyst, 1970, 95, 431. Nelder, J. A., and Mead, R., Comput. J., 1965, 7, 308. Ebdon, L., Cave, M. R., and Mowthorpe, D. J., Anal. Chim. Acta, 1980, 115, 171. Deming, S. M., and Parker, L. R., CRC Crit. Rev. Anal. Chem., 1978, 7 , 187. Spendley, W., Hext, G. R., and Himsworth, F. R., Technometrics, 1962, 4, 441. Yarbro, L. A., and Deming, S. N., Anal. Chim. Acta, 1974, 73, 391. Michel, R. G., Coleman, J., and Winefordner, J . D., Spectrochim. Acta, Part B, 1978, 33, 195. Quimby, B. D., Uden, P. C., and Barnes, R. M., Anal. Chem., 1978, 50, 2112. 1977, 92, 321. 1979, 108, 31. Received August 14th, 1981 Accepted September llth, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700129

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

144 Analyst, February, 1982, Vol. 107, pp. 144-156 Study of the Use of Soil Suspensions in the Determination of Iron, Manganese, Magnesium and' Copper in Soils by Flame Atomic-absorption Spectrometry J. Stupar and R. Ajlec JoZef Stefan Institute, E. Kardelj University, 61000 Ljubljana, Yugoslavia A method is described for the direct, routine atomic-absorption spectrometric determination of copper, iron, manganese and magnesium in soil samples of the terra rossa and peat types. An investigation was made of the factors in- fluencing the atomisation efficiency of these elements when suspensions of soil samples were aspirated into the flame. Particle size, flame temperature and position in the flame were found to be critical in determining the fractions of particular elements atomised.Special emphasis was given to the preparation of the soil suspensions, which is the most critical step in the whole analytical procedure. Magnetic and ultrasonic devices were used for stirring purposes. The latter proved to be more efficient, particularly when suspensions of high clay content soils are being prepared. An average standard sample made for each soil type was used for calibration. Test analyses of two sets of soil samples showed that the majority (80-90%) of samples can be analysed with an accuracy of k2Oy'. This should be acceptable in most applications where a large number of samples are to be analysed. Considerable amounts of time and chemicals can be saved. The method was also found to be suitable for the determination of lithium, calcium, strontium, barium, aluminium, chromium and titanium.Keywords : Iron, manganese, magnesium and copper determination ; soil samples ; soil suspensions ; flame atomic-absorption spectrometry Direct introduction of solid materials into the flame for atomic-absorption spectrometric measurements has been attempted by many investigators, usually in order to avoid the tedious and time-consuming dissolution of the samp1e.l-22 The procedure has proved to be particularly advantageous when finely powdered, chemically resistant materials (ores, sediments, soil, etc.) and/or a large number of samples are to be analysed. The first experiments on the direct atomisation of solids in the flame were reported by Gilbert1 in 1962, who measured emission spectra of a soil suspension in oxygen - hydrogen and oxygen - acetylene flames produced by a large bore Beckman atomiser - burner.He concluded that several major and also trace elements present in the soil can be quantitatively determined by this method. However, he pointed out that, in order to obtain accurate results, the effects of the physical and chemical nature of the sample, solvent and of the flame should be carefully investigated. Masson2 applied the method of Gilbert to the determination of potassium in plant material, and Kashiki and OshimalO introduced alumina catalyst powder samples into the flame for cobalt and molybdenum determination. A few papers appeared on the determination of suspended metal particles in lubricating 0ils.*9~9~9~~ Fullerls determined trace metals in titanium oxide pigments employing a combination of flame and electrothermal atomisation of the suspended samples.Harrison and Juliano13 reported the determination of tin in different tin compounds (oxides and sulphides) and tin ore concentrate using a turbulent air - hydrogen flame. Willis16 studied in detail the atomisation of suspensions of geological samples. He found that the results for one particular element could vary by a factor of two between rocks of widely differing types. He recommended the technique for such purposes as geochemical prospecting, where this accuracy could be accepted. Langmyhr19s20 and L'vov17 produced excellent reviews of the state of the art in the direct analysis of solids by atomic-absorption spectroscopy. O'Reilly and Hicks21 presented an elaborate study of the direct analysis of coal samples using a slurry atomic-absorption spectro-STUPAR AND AJLEC 145 metric technique.Employing an air - acetylene flame, they were able to determine seventeen major, minor and trace elements with an accuracy of 5-25y0. Van Loon22 produced an atomic vapour of magnesium, manganese and copper by the direct heating of iron, brass and silica samples in an air - acetylene flame and simultaneous aspiration of concentrated hydrochloric acid. The aim of our study was to develop a method for the direct, routine atomic-absorption spectrometric determination of total iron, manganese, copper and magnesium in peat soil and terra rossa soil samples by aspiration of suspensions into the flame. Factors that influence the precision and accuracy of measurement of soil suspensions, as well as problems associated with the preparation of appropriate standards, are discussed.The possibility of determining other major and minor elements in soil was also investigated. Experimental Apparatus A Varian-Techtron, Model AA-5, atomic-absorption spectrometer was used throughout this work. A standard nebuliser - spray chamber system and the burner heads for air - acetylene (10-cm slot) and dinitrogen oxide - acetylene flames (5-cm slot) were employed. A supple- mentary system was designed for mixing air and dinitrogen oxide in order to prepare various oxidant gas mixtures. Suspensions of soil samples were prepared in 100-ml Erlenmeyer flasks and were stirred with a magnetic device during aspiration into the flame.Instrumental parameters employed in the atomic-absorption spectrometric measurements of iron, manganese, copper and magnesium are presented in Table I. In general, the same absorption line, spectral band pass and lamp current were used for both suspension and solution measurements. A hydrogen lamp was employed for background absorption measurements. An oxidant mixture of dinitrogen oxide (30%) - air (70%) was found to be superior for suspen- sion measurements, as far as sensitivity and stability of measurements were concerned. A dinitrogen oxide - acetylene flame appeared to be less suitable for these elements owing to a much higher flame gas velocity and noise of the absorption signal. Air - acetylene or di- nitrogen oxide - acetylene flames were used for atomisation of solutions. The 372.0-nm iron line was selected for absorption measurement because of its suitable sensitivity and linear response.This was connected directly to the nebuliser. TABLE I INSTRUMENTAL PARAMETERS EMPLOYED IN ATOMIC-ABSORPTION SPECTROMETRIC MEASUREMENTS OF SOIL SUSPENSIONS Wavelength/ Element nm 372.0 279.5 324.7 285.2 Iron . . .. . . 372.0* Manganese . . . . 279.5* Copper . . .. . . 324.7* Magnesium . . . . 285.2* Spectral band pass/ nm 0.33 0.33 0.17 0.17 0.33 0.33 0.17 0.17 Lamp current1 mA 9 9 5 5 3 3 3 3 Flame Air - acetylene Air - N,O - acetylene Air - acetylene Air - N,O - acetylene Air - acetylene Air - acetylene N,O - acetylene Air - N,O - acetylene Working range/ pg ml-l 0-100 0- 60 0 - 5 0 - 3 0 - 2 0 - 1 0- 20 0- 15 * Solutions of soil samples obtained by conventional decomposition.Preparation of Samples Preparation of samples is an essential stage in the direct atomic-absorption spectrometric analysis of solids in flames, particularly in dealing with soils. These differ significantly in their chemical (mineral) and physical (particle size) composition. Soils are generally character- ised by three different particle size fractions: less than 2 pm, organic matter, clay; 2-20 pm, silt; and 20-2000 pm, sand. Clay, silt and sand contain a great variety of minerals whose proportions vary tremendously between soils of different types. Under dry conditions clay146 STUPAR AND AJLEC: SOIL SUSPENSIONS IN DETERMINATION OF Analyst, Vol. 107 minerals are liable to form aggregates (silica chains of various lengths) that disintegrate in the presence of water and certain electrolytes to small segments.In our study we concentrated on two soil types common in the northern part of Yugoslavia, peat soil (from the area around Ljubljana) and terra rossa (from the Istrian peninsula). There is a substantial difference in the physico-chemical nature of these soil types. An illustration of the typical grain size composition of these soils is given in Table 11. It was found that in the terra rossa clay fraction the minerals bohmite, haematite, gibbsite and anatase are pre- dominant. In contrast, peat soil contains more organic matter, illite being the most character- istic mineral. TABLE I1 PARTICULATE COMPOSITION OF SOIL SAMPLES Organic matter, clay, % Sample* (<2 Pn-4 P4B .... 53.9 P13B .. .. 42.3 P2B .. .. 47.9 P63G . . .. 20.5 P73G .. .. 22.7 P71G .. .. 34.9 T 3 . . .. .. 60.4 T 5 . . .. .. 60.0 T 7 . . .. .. 60.6 T 9 . . .. .. 75.8 T11 .. .. 62.0 T13 .. .. 59.0 * P = Peat soil and T = terra rossa soil. Silt, yo (2-20 pm) 39.6 47.4 46.0 43.8 42.0 51.9 16.6 17.8 17.0 10.2 15.4 18.4 Sand, % (20-2000 pm) 4.5 10.3 6.1 35.7 35.8 13.2 23.0 22.2 22.4 14.0 22.6 22.6 Grinding Grinding is the first step involved in the preparation of soil samples after their collection. The purpose of this is to reduce the particle size of the sand fraction, which contains particles up to 2 mm in diameter. A wet method of grinding (using ethanol) was performed in an agate ball-mill (Fritsch; 25 balls, 14 mm in diameter).When the procedure was completed, samples were dried in an oven at 105 "C and crushed with an agate pestle and mortar. Par- ticle-size analysis of soil samples after grinding showed that the largest particles present in the samples did not exceed 40 pm in diameter. From the economic point of view it is important to select an optimum time of grinding. For this investigation a series of five samples (terra rossa) were ground, one portion for 0.5 h and the other for 1 h. Suspensions [0.25y0 m/V, six of each sample] in 20% V/V propan-2-01- water were prepared and aspirated into the flame, and the absorbances of iron, magnesium and manganese were recorded. It may be concluded from the results shown in Table I11 for iron and those obtained for magnesium and manganese that grinding the samples for 0.5 h under the conditions specified TABLE I11 VARIATION OF THE ABSORBANCE OF IRON IN AN [AIR (70%) - DINITROGEN OXIDE (30%)] - ACETYLENE FLAME WITH TIME OF SAMPLE GRINDING Sample: terra rossa, 0.25% wz/V suspension.Grinding time/h r ~~ 1 1 0.5 A A f I f -l Relative standard Relative standard 6 0.3430 2.44 0.333 2 2.07 2.86 8 0.4075 1.56 0.361 6 1.98 11.25 12 0.3567 1.92 0.334 2 2.56 6.31 1 0.1776 1.71 0.1572 1.03 11.44 16 0.3987 2.02 0.353 6 2.25 11.31 Sample A , deviation, % AO.5 deviation, % 100 (A, - A O J / A l , %February, 1982 FE, MN, MG AND Cu IN SOILS BY FLAME AAS 147 is adequate for all practical purposes. The absorbances measured are on average 6-15y0 lower in comparison with those after grinding for 1 h, but the precision of the measurement (relative standard deviation) of major elements (iron, magnesium) is not seriously affected.On the other hand, the precision of the measurement of some minor (manganese) and trace elements may vary considerably with the sample grinding time. ic Preparation of soil suspension It primarily influences sample transport by changing the solution flow and/or the drop-size distribution. In atomisation of suspensions, the solvent plays an even more important role, as it also acts as a stabiliser and may affect the chemical or physical state of the sample. In the course of this study the following organic solvents and water - organic solvent mixtures were investigated: methanol - water (O-lOOyo V/V methanol) ; ethanol - water (O-lOOyo V/V ethanol) ; propan 2-01 - water (O-lOOyo V/V propan-2-01) ; acetone - water (O-lOOyo V/V acetone) ; and 4-methylpentan-2-one.A detailed study of the mechanism of nebulisation of suspensions and the role of the solvent in these processes will be reported in a separate paper. In the routine analytical application of the suspension technique, the following criteria were considered in choosing an appropriate solvent : low cost; low metal content; and moderate viscosity and low surface tension. On this basis methanol and propan-2-01 were selected for trials in which 0.25y0 m/V suspen- sions of a terra rossa soil sample were prepared in both solvents and solvent - water (20 and 50% V/V of solvent) mixtures. Suspensions were mixed for 10 min on a magnetic stirrer and left until the next day. They were stirred for 1 min before measurement and for half of the period of aspiration into the flame as shown in the recorder traces (yo absorption) for iron measurements presented in Fig.1. Arrows denote where stirring of the suspensions was stopped. When stirring of the suspensions was stopped, settling of the larger particles occurred. This process is less pronounced in viscous solvents, which is well illustrated in Fig. 1 (50% mixtures of water and solvent and pure propan-2-01). The stability of the absorption signal is generally better in propan-2-01 and improves with increasing percentages of solvent in the mixture for both propan-2-01 and methanol. Sensitivity of measurement is generally higher in less viscous solvents of low surface tension (methanol, acetone). The sensitivities (relative to water) of the common organic solvents used in this work are listed in Table IV.It is interesting to note that the sensitivity of iron Choice of solvent. The role of the solvent in flame spectrometry is well established. 100% 7 60 s H Time __+ Fig. 1. Recorder traces showing sensitivity and stability of soil suspension measurements in various organic solvents : (a) methanol - water; (b) propan-2-01 - water. Sample, terra rossa, 0.25% m/ V ; flame, [air (70%) - dinitrogen oxide (30%)] - acetylene; and element, iron.148 STUPAR AND AJLEC: SOIL SUSPENSIONS IN DETERMINATION OF Analyst, VoZ. 107 and magnesium differed considerably in the same solvents. A variation in sensitivity may also be observed with soil samples of a different type. The choice of solvent in suspension analysis would therefore depend on whether sensitivity or precision of measurement is the limiting factor in a particular application.In most of the experiments reported a 20% V/V propan-2-01- water mixture was employed as a compromise between sensitivity and stability of the absorption measurements. TABLE IV NORMALISED SENSITIVITIES OF IRON AND MAGNESIUM IN VARIOUS SOLVENTS , RELATIVE TO WATER Sample: peat soil, 0.25% m/V suspension. Flame: [air (70%) - dinitrogen oxide (30%)] - acetylene. Solvent Magnesium Iron Water .. .. .. .. .. .. .. 1 1 Methanol . . .. .. .. .. . . .. 2.3 3.2 Ethanol . . .. * . .. .. .. .. 1.8 2.4 Propan-2-01 . . .. .. .. .. .. .. 1.1 1.5 Propan-2-01 (20y0) -water .. .. .. . I . . 1 1.4 Acetone . . .. .. .. .. .. .. 0.7 2.7 Heptanone or 4-methylpentan-2-one . . .. .. 1.7 2.4 Mixing of suspensions. The general purpose of mixing is to provide a uniform distribution of a powdered sample in a suitable solvent for reproducible aspiration into the flame. Homo- genisation of samples was accomplished by utilising a magnetic stirrer or an ultrasonic device. Particular attention was devoted to the mixing of soil suspensions as some adverse effects upon their atomisation were observed. In the first experiments dealing with peat soil samples, the following procedure of suspension preparation was employed: soil was mixed with the solvent for 15 min using a magnetic stirrer, left overnight (24 h) and stirred again for 1 min just before aspiration into the flame.Absorption signals obtained in this way were reasonably stable for all four elements investigated. In Fig. 2. the stability and reproducibility of iron absorption signals on spraying a solution and a soil suspension are compared. It is evident that the reproducibility of measuring sus- pensions is worse by factor of two, which is acceptable and does not represent a limiting factor in suspension analysis. 60 s H A .3ucl A B . I I Time Fig. 2. Recorder traces comparing the stability and reproducibility of absorption measurements of soil suspensions and true solutions for iron at 272.1 nm. A, Solution, 30 pg ml-l of iron, 3 = 0.223 3 f 0.86%; and B, soil suspension, 0.25% m/V peat soil, Z = 0.171 7 f 1.7%.February , 1982 149 Measurement of soils of a different type (terra rossa) employing the same mixing procedure produced a continuous rise in the absorption signal with duration of mixing.This phenomenon was observed in more or less pronounced form with the majority of samples of this type of soil. Even stirring for 1 h was not adequate to stabilise the absorption signal. The magnetic stirrer was then replaced by an ultrasonic immersion device (125 W of ultrasonic power output) and the suspensions were treated for different periods of time (1-60 min). The suspensions were left for approximately 24 h and stirred again for 1 min prior to the absorption measure- ment using a magnetic stirrer. The results of this experiment are illustrated in Fig. 3 where iron absorbance is plotted against the time of stirring.Ultrasonic agitation of the suspension appeared to be more efficient in breaking larger clay aggregates into their original fragments. It may be assumed (see Fig. 3) that equilibrium in the suspension is obtained after approxi- mately 5 min of ultrasonic agitation even for most of the unfavourable samples of this terra rossa type of soil (high clay content). Recorder traces due to iron absorption are shown from measurements of two different types of soil, peat and terra rossa. Parallel suspensions were prepared from each sample and mixed for 5 min, one on a magnetic stirrer and the other employing an ultrasonic device. The following day all four suspensions were mixed again with the magnetic stirrer for 1 min before the measurement. After approxi- mately 2 min of continuous monitoring of the absorption signal, stirring was interrupted while the suspension was still aspirated and the signal recorded. On the basis of the absorption signals recorded the following conclusions can be drawn.FE, MN, MG AND Cu IN SOILS BY FLAME AAS This was confirmed by the experiment illustrated in Fig. 4. 0.30 (0 + z Q 0.20 D 0.lOt I , , , , 0 10 20 30 40 50 60 Time of stirring/min Fig. 3. Effect of stimng on the absorbance Suspension, terra rossa of iron in the flame. B Time __+ 0.25% m/V in 20% propan-2-01; flame, [air (70%) - dinitrogen oxide (30%)] - acetylene. Fig. 4. Recorder traces for (a) iron absorption A, ultrasonic (125 W) ; and B, magnetic. (1, 150; 2, 90; 3, 70; and 4, 30 pg g-l of iron) on aspirating suspensions of various soil types [ (b) terra rossa No.9 and ( c ) peat soil No. 671 and application of different methods of mixing: A, magnetic stimng and B, ultrasonic agitation. Flame, [air (70%) - dinitrogen oxide (30%)] - acetylene; 0.25% m/ V suspension in 20% propan-2-01- water. (i) For highly organic soils (peat for example) low in clay content, magnetic stirring is almost equivalent to ultrasonic agitation in producing a stable suspension. There is not much difference in the atomisation efficiency of iron, which implies an identical particle size distribu- tion in the suspension. Clay-rich soil types, such as terra rossa, seem to form very stable aggregates that disintegrate only partially under the influence of magnetic stirring. The equilibration of particle sizes in the suspension is rather a slow process (continuous slow in- crease in the absorption signal).Intense ultrasonic vibrations are more efficient in this respect. An almost complete disintegration of clay aggregates is achieved in a few minutes, which results in an extremely high atomisation efficiency of the elements from the suspension.160 STUPAR AND AJLEC : SOIL SUSPENSIONS IN DETERMINATION OF Analyst, vd. I07 (ii) A small addition of an electrolyte such as sodium carbonate to the suspension was found to be beneficial, substantially increasing the atomisation efficiency of the elements studied. Experiments were made on peat soil samples; the results are presented in Table V. The reason for the enhancing effect of sodium carbonate has not yet been closely investigated. However, two possible explanations of the phenomenon seem feasible: an influence on the particle size distribution in the suspension by electrostatic action ; and/or influence on the vaporisation process in the flame by altering the chemical composition.TABLE V RELATIVE ATOMISATION EFFICIENCY OF IRON, MANGANESE, MAGNESIUM AND COPPER IN AN [AIR (70%) - DINITROGEN OXIDE (30%)] - ACETYLENE FLAME; EFFECT OF THE PRESENCE OF SODIUM CARBONATE IN THE SUSPENSION Sample: peat soil in 26% V/V propan-2-01 - water. Relative atomisation efficiency Fe i i z z T G 7 Sample Na,CO, Na,CO,* B2 . . . . 0.294 0.383 B4 .. . . 0.352 0.381 B13 .. . . 0.208 0.279 G67 .. .. 0.289 0.345 G73 . . . . 0.180 0.262 G71 .. . . 0.262 0.337 Mn Na&O, Na,CO,* 0.241 0.383 0.295 0.367 0.188 0.245 0.323 0.341 0.172 0.256 0.185 0.216 % i G z - i z cu Without Na,CO, 0.234 0.255 0.193 0.225 0.142 0.216 With Na&O,* 0.297 0.298 0.260 0.267 0.253 0.329 Without 0.302 0.294 0.250 0.304 0.271 0.205 NaaCO, With Na&O,* 0.414 0.421 0.329 0.462 0.387 0.297 * Amount of Na,CO, added = 33 mg per gram of soil.Atomisation of Suspension Atomisation of suspensions in flames is in principle similar to atomisation of solutions and involves sample transport to the flame, vaporisation of sample in the flame and dissociation of the vaporised sample. A new quantity ea,rel. introduced by WillislS is of principal interest in suspension analysis. It is defined as the relative atomisation efficiency, i.e., the ratio of the number of free atoms of the analyte in the ffame when suspensions are sprayed to the number of free atoms in the same flame volume when the same amount of analyte solution is sprayed: ..* ' (1) ca,rel. = e a / ~ - .. .. where €8 is the atomisation efficiency of an element in the suspension and is the atomisation efficiency of an element in the solution. The distribution of atoms inside the flame should be considered for a proper understanding of ea,rel.. Suspensions are obviously more complicated than solutions as far as atomisation is concerned. Some additional parameters such as grain size, chemical composition of the sample and concentration of the suspension may be vital in influencing the atomisation efficiency of an element in the suspension. In this paper only practical aspects of the atomisation of soil suspensions will be discussed; a comprehensive treatment of this process will be published separately. Measurement of e R d Dried and grmnd soil samples (peat soil) were analysed by a conventional atomic-absorption spectrometric procedure following an acid (nitric - perchloric - hydrofluoric) decomposition.Residues, if found, were fused with sodium carbonate, dissolved in hydrochloric acid (1 + 1) and analysed separately. The true concentrations (micrograms per gram) of the elements concerned were calculated as an average of six parallel determinations. Apparent concentrations of elements were obtained by measurement of the suspensions. Three parallel suspensions of each sample were made in a 25% propan-2-01 - water mixture, stirred for 16 Inin, left overnight and stirred again for 1 min prior to aspiration into the flame.February, 1982 F E , MN, MG AND Cu IN SOILS BY FLAME AAS 151 Suspensions were measured against standard solutions prepared in the same solvent.Back- ground absorption and the solvent-soluble fraction were measured for each particular element and subtracted from the reading when necessary. A measurable background absorption was only observed for the magnesium absorption line. The Ea,rel. value was thus calculated by dividing the apparent concentration of an element by its true concentration. Solvent-soluble Fraction When a suspension is prepared, a certain proportion of the elements present in the soil dis- solves to form a true solution. Thus the apparent concentration of particular atoms in the flame will be a result of atomisation of the solution and of the solid sample.As the relative contribution of the soluble fraction to the total absorption signal is generally different from that of the solid fraction, large variations of the ratio of soluble to solid fractions between samples and the standard may considerably affect the accuracy of the analytical results. Solvent-soluble fractions of the elements manganese, iron, magnesium and copper were deter- mined, varying the following experimental parameters : (a) Type of soil: higher concentrations of the elements were found in the solution from terra rossa samples (high clay content) in comparison to peat soil samples (low clay content ) (b) Element : the elements investigated showed considerably different solubilities (in the order magnesium > iron > manganese > copper).(c) Suspension preparation : ultrasonic agitation of the suspension in comparison with mag- netic stirring produced a 2-3-fold increase in the solvent-soluble fraction (magnesium, iron). (d) Solvent: the solubility of the elements in organic solvent (propan-2-01, methanol) - water mixtures decreases with the percentage of organic solvent and becomes negligible at and above 50% of organic solvent in the mixture. The results of this investigation would thus imply the use of 50% propan-2-01 (or methanol) in routine soil analyses as a logical solution of the above-mentioned problem. In addition to this, 50% propan-2-01- water produced the most stable absorption signals (see Fig.l), and the reduced sensitivity would not be a drawback, at least for major and minor elements. Ea,ml. as a Function of Various Parameters As the efficiency of atomisation is directly correlated with the accuracy of determination by suspension analysis, it is of prime importance to investigate the effects of sample and flame parameters upon Eg,rel,. Variation of Ea,rel. with suspension concentration is shown in Table VI. A 4% m/V suspension concentration was considered to be a practical upper limit as frequent clogging of the capillary occurs at higher concentrations. It is evident that suspension concentration is not a critical factor; only magnesium shows a slight decrease in Ea,rel. at 1 and 2% m/V suspension concentrations. TABLE VI RELATIVE ATOMISATION EFFICIENCY OF COPPER, MAGNESIUM, MANGANESE AND IRON IN FLAMES ; VARIATION WITH SUSPENSION CONCENTRATION Sample: peat soil in 25% V/V propan-2-01- water.Flame: [air (70%) - dinitrogen oxide (30%)] - acetylene and air - acetylene (for copper). Suspension concentration, 0.25 0.50 1 .oo 2.00 3.00 4.00 1% m/V Relative atomisation efficiency* A I 7 cu Mg Mn Fe 100 100 96 93 99 95 100 91 95 100 100 83 99 95 100 - - - - - - - - 99 * Maximum value for each element normalised to 100.152 STUPAR AND AJLEC: SOIL SUSPENSIONS IN DETERMINATION OF Analyst, VoZ. 107 In order to establish the effect of the size of the grains in a soil sample upon Ea,rel., one of the peat soil samples was fractionated by the wet precipitation technique into four grain size groups, less than 2 pm, 2-10 pm, 11-20 pm and 2140 pm, and Ea,rel.was measured for the elements concerned in each of the fractions. Fig. 5 illustrates the variation of Ea,rel. with grain size, assuming a symmetrical distribution of particles in each group. As this is obviously not true, a certain variation in the shape of the curves may be expected. However, it can be concluded from the general trend of these curves that deviations in grain size distribution between samples, particularly within the fraction of less than 10 pm, would produce major changes in Ea,rel. and subsequently result in a large analytical error. A variation in mineral composition within each fraction would additionally alter values of Ea,rel., and was in fact observed with samples of various soil types (peat soil, terra rossa).0.6 0.4 - r 5 w 0.2 0 2 10 20 30 Particle diametedpm Fig. 6. Relative atomisation efficiency as a function of particle size of: A, copper; B, iron; C, magnesium; and D, manganese. Sample, peat soil, 0.25% m/V manganese, iron and magnesium and 1 % m/ V copper suspension ; solvent, 25% V / V propan-2-01 - water; flame, air - acetylene (copper), [air (70%) - dinitrogen oxide (30%)] - acetylene (iron, manganese and mag- nesium). In Fig. 6 the influence of height in the flame (time the particle spends in the flame) and flame temperature upon Ea,rel. of copper, magnesium and manganese are represented. Copper seems to be fairly easily liberated from soil minerals as it shows the least effect of either of these variables. Magnesium and manganese experienced an increase, with height in the flame, of Ea,rel.particularly in the cooler air - acetylene flame. Segments higher in the flame where Ea,rel. reaches its limiting value are normally preferable for suspension measure- ments. The results of this investigation clearly indicated the need for careful control and optimisa- tion of these parameters in direct routine analysis of soil samples. Analysis of Soil Samples On the basis of the above experimental evidence, an attempt was made to use the suspension technique in routine soil analysis. Two sets of soil samples of different geochemical origin (peat soil, terra rossa) were selected to test the precision and accuracy of the suspension technique. Samples were ground (see under Grinding) and dried at 105 "C prior to analysis.The true concentrations of copper, iron, manganese and magnesium were determined by wet dissolution and a conventional atomic-absorption spectrometric procedure. Direct analyses were performed on suspensions prepared in 20y0 and 50% propan-2-01 - water mixtures. Generally 0.25% m/ V suspensions were employed for iron and magnesium determinations,February, 1982 FE, MN, MG AND Cu IN SOILS BY FLAME AAS 153 0.20 0.15 - e a-0.10 0.05 0 B 8 10 12 14 16 18 20 22 24 / 8 10 12 14 16 18 20 22 24 Height in flame/mm Fig. 6. Variation of relative atomisation efficiency with height of measurement in the flame. Sample, peat soil, 1% m/V suspension in 25% V/v propan-2-01 - water. A, Magnesium; B, manganese; and C, copper. Solid lines, air - acetylene flame; and broken lines, [air (70%) - dinitrogen oxide (30%)] - acetylene flame.whereas 1-2% m/V suspension concentrations were found to be the most convenient for measurement of manganese and copper. The latter was measured in an air - acetylene flame, whereas for the other three elements a [dinitrogen oxide (30%) - air (70y0)] - acetylene flame was preferred. Selection of appropriate standards for calibration presented the most delicate problem in direct soil analysis. Significant variations in mineral, chemical and particle size composition between soils of different origin would entail the use of a separate set of standards for each particular soil type if reasonable accuracy is to be expected. Our approach was to prepare for each soil type one average sample from which a standard suspension for calibration could be made.The average sample was prepared from equal portions of all of the samples in a particu- lar set (six samples of peat soil and fifteen samples of terra rossa). The elemental content of the standard was determined by conventional atomic-absorption spectrometric and other analytical methods. The same preparation procedure (grinding, mixing of the suspension, etc.) was applied for the standard as for the other samples in the series. The standard suspen- sion was measured with each set of identical samples. The concentration of a particular ele- ment in the sample was calculated taking a simple linear relationship between the absorbances of the standard and of the sample. Background absorption was substracted from the reading if neccessary but no allowance was made for the solvent-soluble fraction in 20% propan-2-01 - water. The standard suspension was assumed to have an average chemical, mineral and grain-size composition of the whole series of similar samples.Thus an approximately equal number of the results should be biased positively and negatively with respect to the true value. The con- cept of an average sample as a standard was first tested on a set of six peat soil samples. Suspensions in 20% propan-2-01 - water were mixed for 15 min on a magnetic stirrer, left over- night and stirred for approximately 1 min before the measurement. No difficulties such as capillary clogging, drifting of the absorption signals, etc., were observed during the measure- ments.The resulks of this trial are collected in Table VII where deviations from the true values [A (yo)] and Ea,rel. are also given. It is evident that the concept of the standard is fairly well demonstrated and close correlation between Eg,rel. and A can be observed. Using the same procedure some difficulties arose in obtaining reproducible results. Anomalous behaviour of A similar test was made with a series of fifteen terra rossa soil samples.TABLE VII CONCENTRATIONS OF IRON, MANGANESE, MAGNESIUM AND COPPER IN PEAT SOIL DETERMINED BY ATOMIC-ABSORPTION SPECTROMETRY WITHOUT DECOMPOSITION OF THE SAMPLES AND WITH MAGNETIC STIRRING Fe v Sample* p.p.m. A, % B2 55 800 + 12.5 B4 34 100 +6.l B13 20 100 - 13.6 G67 30 100 -3.9 G7 1 34 700 - 1.1 G73 21 800 - 16.7 Standard G 28600 0 Ea,rel.0.38 0.38 0.28 0.34 0.34 0.26 0.34 Mn Concentration, p.p.m. 2 760 240 200 210 460 690 714 A* % +24.6 +35.5 -9.3 + 17.0 - 20.0 -6.3 0 capel 0.38 0.37 0.25 0.34 0.22 0.26 0.27 cu r - Concentration, p.p.m. 14 650 15 650 11 600 10 100 10 250 6 600 6 980 A* % + 13.2 + 13.9 - 4.6 +1.1 + 22.0 - 6.6 0 Ea,rel. 0.30 0.30 0.26 0.27 0.33 0.26 0.26 Concentration, p.p.m. 40 61 33 42 32 28 34.7 A, % - 6.7 - 1.9 - 26.4 + 7.5 -33.0 -9.6 0 1 ta,rel. 0.41 0.42 0.33 0.46 0.30 0.39 0.40 * Suspensions in 20% propan-2-01 - water. 2 TABLE VIII cd E 8 m M Z CONCENTRATIONS OF IRON, MANGANESE, MAGNESIUM AND COPPER IN TERRA ROSSA SOIL DETERMINED BY ATOMIC-ABSORPTION SPECTROMETRY WITHOUT DECOMPOSITION OF THE SAMPLES AND WITH ULTRASONIC AGITATION Fe Mn Sample* 1 2 3 4 6 6 7 8 9 10 11 12 13 14 15 Standard Concentration, p.p.m.25400 56 500 53 000 38 500 55 000 50 300 46400 77 600 67 900 49 600 39 900 60 200 44 200 41 400 72 100 56 260 A, % - 37.4 -4.7 - 6.0 - 33.1 - 4.4 - 13.4 -25.3 + 5.6 - 10.5 - 11.9 - 13.8 + 3.6 -25.3 - 12.1 +5.6 0 Ea,rel. 0.36 0.60 - - 0.47 0.43 0.61 - - 0.62 - - 0.50 0.53 Concentration, p.p.m. 710 1060 860 660 900 900 680 840 660 1100 1600 1600 1330 1700 860 1092 A, % -26.8 -4.6 - 1.6 - 13.9 - 13.6 - 13.6 - 15.4 + 14.3 + 9.6 -9.1 - 0.2 - 13.8 + 0.6 +9.1 + 3.6 0 a Ca,rel. 0.31 0.39 - 0.34 0.47 0.48 - - - 0.35 - - 0.61 0.40 Mg I 1 Concentration, p.p.m. A, % Ea,rel. 2 950 -46.8 0.21 -6.6 0.36 6 950 6 800 -7.6 - 4 200 -34.3 - 6 100 0.0 - 6 360 -13.4 0.34 4 600 -17.6 - 6 150 +13.6 0.40 8 660 +23.7 0.41 6 400 -8.9 - 5 160 -17.6 - 6 400 f11.7 0.41 4 700 -29.2 - 3 850 -9.1 - 7 300 +26.7 0.43 5 9SO 0 0.34 cu Concentration, p.p.m.29.0 41.0 36.5 33.0 37.0 33.0 66.0 61.0 38.6 33.0 39.6 40.0 37.5 49.0 41.2 - A* % - 30.4 - 2.6 - 6.4 - 10.4 - 16.3 +6.8 - + 16.6 +2.3 + 10.3 +0.7 + 14.1 0 - 8.6 - 36.2 - 1.4 0 0.68 -I 0.61 - - 2 Q 0.66 - - 0.64 0.66 * Suspensions in 20% propan-2-01 - water.Febrzlary, 1982 FE, MN, MG AND Cu IN SOILS BY FLAME AAS 165 10 20 30 40 50 60 Variation of the analytical error (deviation from the true value) with time of stirring suspension. Sample, terra rossa No. 9, 0.25% m/V suspension in 20% V / I' propan-2-01 - water ; determination of iron ; flame, [air (70%) - dinitrogen oxide (30%)] - acetylene. Ti me of sti rri ng/mi n Fig. 7. some particular samples in the series was observed.If such samples were mixed for different times large variations in the results were noted. In Fig. 7 the variation of the analytical error is plotted as a function of the time of stirring the suspension. With these samples rather noisy absorption signals were observed, showing a continuous rise during prolonged aspiration into the flame. When stirring was interrupted the absorption signal dropped rapidly, the remain- ing signal being substantially improved in stability. Ultrasonic agitation was found to be effective in solving this problem, as described under Mixing of suspensions. Therefore, the analysis of terra rossa samples was performed on suspensions treated (for 2 min) with an ultrasonic vibrator. After approximately 24 h the suspensions were measured under continu- ous stirring employing a magnetic device. Although the variability of the measured signals was acceptable, the results of the analysis given in Table VIII shows a definite negative bias.The reason for the systematic error was found to be in the relatively large proportions of the elements in the soluble form (up to 40% of the total amount). The solvent-soluble fraction of the elements would have to be measured for each sample and the standard if an efficient correction were to be performed. To avoid this time-consuming operation, 50% V / V propan-2-01 - water suspensions were prepared with the same set of samples. In this instance the solvent-soluble fraction could be neglyted and the stability of the absorption signals was slightly improved.The results of this trial are summarised in Table IX, where the negative and positive devia- TABLE IX CONCENTRATION OF IRON, MANGANESE, MAGNESIUM AND COPPER IN TERRA ROSSA SOIL DETERMINED BY ATOMIC-ABSORPTION SPECTROMETRY WITHOUT DECOMPOSITION OF THE SAMPLES AND WITH ULTRASONIC AGITATION Sample* 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Standard Fe Concentration, p.p.m. 38 600 58 200 51 700 42 700 55 200 51 700 47 800 79 200 64400 56 900 47 500 54 100 53 300 50 100 61 700 56 260 ------+- A, % -4.9 -1.8 -7.3 - 25.6 - 3.9 -11.0 - 22.9 + 7.9 -15.1 + 0.9 + 2.6 - 7.0 - 9.9 + 6.3 - 9.7 0 Mn Concentration, p.p.m. 840 1240 890 580 970 900 670 900 470 1210 1610 1480 1270 1610 890 1092 v- A, Yo + 12.3 + 2.8 - 24.7 - 5.3 -13.2 -17.1 + 23.1 - 7.3 + 0.2 + 6.5 -15.3 - 17.5 + 4.1 + 7.0 0 -13.3 Mg Concentration, - p.p.m.3 900 6 900 5 700 4 100 6 250 6 350 4 850 6 700 6 400 7 400 6 150 5 200 6 000 4 750 6 600 5 980 As % - 30.1 + 7.9 -8.8 - 35.9 + 1.9 -13.5 - 12.8 +23.5 - 2.6 + 5.2 -1.6 + 7.1 - 10.3 +12.4 + 14.4 0 cu Concentration, p.p.m. - 34.0 48.5 33.0 24.5 - 33.0 28.0 52.0 42.0 41 .O 34.5 39.5 41.0 36.0 47.0 41.2 A, % -18.3 + 14.8 -4.3 -29.6 - 19.7 -28.5 + 9.3 -15.9 - 2.7 - 33.5 + 10.1 0.0 - 3.1 + 9.1 0 - * Suspension in 50% propan-2-01 - water.156 STUPAR AND AJLEC tions are represented in an almost equal proportion for manganese and magnesium. This con- firmed the validity of the average standard suspension for calibration purposes for manganese and magnesium; however, for copper and iron a considerable negative bias was observed on average.Conclusion Determination of copper, iron, manganese and magnesium in samples of two different soil types directly by aspiration of suspensions into the flame seems feasible for routine analyses where an accuracy of determination of &20% can be tolerated. In fact, the results of this investigation showed that 50-70% of the values have errors of less than &lo%. From a 2% m/V suspension in 20% propan-2-01- water adequate absorption signals were obtained for lithium in the air - acetylene flame and calcium, chromium, aluminium, titanium, strontium and barium in the dinitrogen oxide - acetylene flame. No measurable background absorption was recorded with these elements. Addition of potassium chloride (to give 2 mg ml-l of potassium) was necessary to increase the barium absorption signal.The main advantage of direct soil analyses as compared with the more elaborate total dissolution method can be summarised as follows: the time required for the analysis is reduced considerably; lower costs for chemicals and equipment ; and increased laboratory safety. Other simple methods such as nitric acid digestion can compete with the proposed technique only in the speed of operation, but the running costs and the accuracy of the former were found to be on average less satisfactory. The possibility of determining some other elements was also investigated. We are indebted to DT. F. Lobnik from the Department of Agriculture, University of His advice during the The authors gratefully acknowledge financial support for this work from the Boris KidriE Ljubljana, for providing soil samples of defined mineral composition. experiments and preparation of the manuscript are gratefully appreciated. Foundation. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. References Gilbert, P. T., Anal. Chem., 1962, 34, 1025. Mason, J. L., Anal. Chem., 1963, 35, 874. Sipicin, S. A., KirjuSkin, V. V., and Ermolaev, A. A., Zavod. Lab., 1965, 31, 253. Burrows, J . A., Heerdt, J. C., and Willis, J. B., Anal. Chem., 1965, 37, 579. Sipicin, S. A., KirjuSkin, V. V., and Kuklina, N. Ja., Zh. Anal. Khim., 1966, 21, 779. Venghiattis, A. A., At. Absorpt. Newsl., 1967, 6 , 19. Venghiattis, A. A., and Whitlock, L., At. Absorpt. NewsZ., 1967, 6 , 135. Bartels, T. T., and Slater, M. P., At. Absorpt. Newsl., 1970, 9, 75. Kriss, R. H., and Bartels, T. T., At. Absorpt. Newsl., 1970, 9, 78. Kashiki, M., and Oshima, S., Anal. Chim. Acta, 1970, 51, 387. Coudert, M. A., and Vergnaud, J. M., Anal. Chem., 1970, 42, 1303. Taylor, J . H., Bartels, T. T., and Crump, N. L., Anal. Chem., 1971, 43, 1780. Harrison, W. W., Juliano, P. O., Anal. Chem., 1971, 43, 248. Govindaraju, K., Hermann, R., Mevele, G., and Chouard, C., At. Absorpt. Newsl., 1973, Govindaraju, K., Mevele, G., and Chouard, C . , Anal. Chem., 1974, 46, 1672. Willis, J. B., Anal. Chem., 1975, 47, 1752. L’vov, B. V., Talartta, 1976, 33, 109. Fuller, C. W., Analyst, 1976, 101, 961. Langmyhr, F. J., Talanta, 1977, 24, 277. Langmyhr, F. J., Analyst, 1979, 104, 993. O’Reilly, J . E., and Hicks, D. G., Anal. Chem., 1979, 51, 1905. Van Loon, C. J., Spectrosc. Lett., 1979, 12, 543. 12, 73. Received July 27th, 1981 Accepted September 18th, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700144

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

Analyst, February, 1982, Vol. 107, $9. 157-162 157 Rapid Hydride Evolution - Electrothermal Atomisation Atomic-absorption Spectrophotometric Method for Determining Arsenic and Selenium in Human Kidney and Liver K. S. Subramanian and J. C. Meranger Environmental Health Centre, Health and Welfare Canada, Tunney's Pasture, Ontario, Canada, KIA OL 2 A rapid semi-automated electrothermal atomisation atomic-absorption spectro- photometric procedure has been developed for the determination of arsenic and selenium in human liver and kidney specimens. The sample is digested with a mixture of nitric and perchloric acids. The nearly dry residue is taken up in hydrochloric acid. The arsenic or selenium in the hydrochloric acid solution is converted into its hydride with sodium tetrahydroborate(II1). The hydride is decomposed and atomised in an electrically heated silica furnace, and the atomic-absorption signal is measured at the appropriate resonance wavelengths of arsenic and selenium.Both of the elements can be determined in the bio- logical samples in the range 50-500 ng per gram of wet sample. The method was applied to the determination of arsenic and selenium in about 40 autopsy samples of human kidney (cortex and medulla) and liver taken from Canadian adults living in the Great Lakes Region of Ontario. The arsenic levels in all of the samples analysed were found to be < 10 ng per gram of wet sample ; the median selenium levels in the cortex, medulla and liver were found to be 0.84, 0.31 and 0.39 mg per kilogram of wet sample, respectively. Keywords : Arsenic and selenium determination ; hydride evolution ; electro- thermal atomisation ; atomic-absorption spectrophotometry ; human kidney and liver The determination of arsenic and selenium in biological materials at or below the milligrams per kilogram level has received considerable attention in recent years owing to the potential public health problems they present. Arsenic, for example, is a cumulative poison and carcinogenf; also, it adversely affects the nervous system, cardiovascular system and the genetic system.2 Selenium, in trace amounts, appears to be essential for both man and beast and its deficiency has been linked to some types of muscular degenerati~n.~ Although definitive evidence is lacking, selenium is also reported to cause both acute and chronic poisoning.The major chronic effects appear to be liver damage, anaemia, restricted growth and shortening of life span.4 The human kidney and liver have been found to be the main storage areas for both arsenic2 and ~elenium.~ Therefore, the analyses of these two organs may provide an index of exposure to both arsenic and selenium in humans. The hydride evolution - electrothermal atomisation technique is one of the most sensitive techniques available for measuring the ultratrace levels of arsenic and selenium found in the kidney and liver of unexposed adults. Recently Vijan and co-workers proposed an electrothermal atomisation atomic-absorption spectrophotometric method via hydride generation for determining arsenic5 and selenium6 in vegetable matter.As this approach seemed simple, sensitive, selective and rapid, we explored the feasibility of adapting it to the determination of arsenic and selenium in some autopsy samples of human kidney (cortex and medulla) and liver. The results are reported in this paper. Experimental Apparatus A Varian Techtron, Model AA-5, atomic-absorption spectrophotometer equipped with a Cathodeon arsenic (or selenium) hollow-cathode lamp was used for measuring the levels of arsenic or selenium. An electrically heated open-ended silica tube, 15 cm long and 1.2 cm i d .158 SUBRAMANIAN AND MfiRANGER: RAPID HYDRIDE EVOLUTION - Analyst, 'c/'Ot?. I07 with a 4 mm diameter inlet tube fused in the middle, was used for atomising arsenic and selen- ium in the gaseous stream.The argon flow-rate was regulated by means of a calibrated flow meter. The temperature of the silica tube was regulated by means of a 0-110 V Variac. The furnace was mounted on the burner head and aligned in the usual manner to let maximum light from the hollow-cathode lamp reach the detector. A Technicon Autosampler with a 40- sample capacity, a proportionating pump and manifold were used in conjunction with a 10-mV strip-chart recorder for achieving automatic operation as described in a previous paper.? The manifold is represented schematically in Fig. 1. 4 A - HCI, 3.9 mi min-' (As, 2 M; Se, 6 Air, 3.9 ml min-' 1% NaBH4,2 ml min-' Waste- Gas - liquid M) r ; 15-turn coil I Proportionating I Pump A Air, 1 ml min-' Sample, 3.9 mi min-' HCJ, 1.2 ml min-' (As, 2 M; - Se, 6 M) A t AsH3/SeH4 Atomic-a bsorption Sampler spectrometer Fig.1. Schematic representation of Autosampler, proportionating pump and atomic-absorption spectrophotometer. Reagents Arsenic(IT1) stock soZw!ion, 1 OOO mg 1-l. Dissolve exactly 0.1320 g of high-purity arsenic(II1) oxide (NBS Standard Reference Material 83c, dried at 100 "C for 2 h) in 2 ml of 1 M sodium hydroxide solution. Add 25 ml of water followed by 4 ml of 1 M hydrochloric acid, dilute to 100 ml with high-purity water and store the solution in a pre-cleaned polyethylene bottle. Dissolve exactly 0.1406 g of analytical-reagent grade selenium(1V) oxide (BDH Chemicals) in the minimum amount of 10% ultrapure hydro- chloric acid. Dilute to 1 1 with 10% hydrochloric acid and store in a 1-1 screw-capped linear pol yet h ylene bottle. For both of the above reagents, solutions of lower concentrations were prepared fresh daily by serial dilution of the stock solutions.Sodium tetra~ydroborate(III) solution, 1 yo m/V. Dissolve 10 g of sodium tetrahydro- borate(II1) (Alfa Inorganics) and 2 g of sodium hydroxide in 11 of high-purity water. Keep refrigerated in a tightly capped linear polyethylene container until ready for use. The solution is stable at least for 1 week under these conditions. SeZenium(1V) stock solution, 1000 mg 1-l. All other reagents and solutions used were of the highest purity available. Samples The autopsied kidney (cortex and medulla) and liver samples in duplicate came from 42 adults who had died at the Kingston General Hospital, Kingston, Ontario.All of the samples appeared to have come from normal individuals based on the presence of recognised histo- pathological changes. The autopsies were carried out within 48 h after death from the same parts of the liver or kidney tissue, using stainless-steel scalpels, as different concentrations of the same element have been found within the same tissue.* The samples were shipped frozen to the laboratory in individual polyethylene bags and stored in a freezer at -20 "C until analysed. Prior to analysis the samples were thawed and brought to room temperature (22 "C). The tissues were then washed three times with high-purityFebrMary, 1982 ELECTROTHERMAL ATOMISATION AAS FOR AS AND SE 159 water. The excess of water was shaken off and any remaining moisture was blotted off with Kimwipe disposable wipers (Type 9OO-S, Kimberly-Clark of Canada Ltd., Toronto, Ontario).As the entire sample was used for analysis no homogenisation was carried out. Analytical Procedure Sample decomposition The samples were weighed out (0.90-1.3 g) accurately into 50-ml Pyrex-glass beakers, which had been cleaned with ultrapure nitric acid as described in a previous publicati~n.~ The samples were then digested with 8 ml of concentrated nitric acid (Baker Ultrex grade) and 2 ml of concentrated perchloric acid (BDH Chemicals, Aristar grade) on a sand-bath until a clear, colourless solution was obtained. The digestate was then allowed to evaporate to fumes of perchloric acid and diluted to 10 ml with 0.5 M hydrochloric acid. About 3 4 ml of the solution were transferred into the 5-ml disposable polypropylene cups placed in the Auto- sampler.The blanks and the reference standards (bovine liver, NBS Standard Reference Material 1577) were also taken through the same procedure. Measurement The atomic-absorption spectrophotometer was allowed to warm up and the silica furnace was allowed to attain thermal equilibrium (usually 30 min) under the optimised operating conditions given in Table I. The vent was closed to the lowest position in order to prevent air turbulence in front of the cell opening and thereby minimising base-line drift. The argon was turned on at the pre-determined flow-rate. The manifold tubes were inserted into the reagent solutions, hydrochloric acid and sodium tetrahydroborate(II1) solution, as shown in Fig.1. The proportionating pump was started. When the system had attained equilibrium (in about 20 min) as indicated by the minimum base-line noise on the strip-chart recorder the spectro- photometer was adjusted to zero absorbance. The Autosampler containing the test solution was switched on and the arsenic(II1) hydride [or selenium(1V) hydride] vapour was carried by the argon stream through a U-shaped spray trap into the resistance-heated silica tube where it was atomised. The atomic-absorption peaks of the blanks, calibration standards, reference standard and samples were recorded. The concentration of arsenic or selenium in the test solution was obtained by reference to linear working graphs prepared from the calibration standards. TABLE I OPTIMUM OPERATING CONDITIONS FOR DETERMINING ARSENIC AND SELENIUM Parameter Arsenic Selenium Line/nm .. .. .. . . . . 193.7 Band width/nm . . .. .. .. 0.3 Hollow-cathode lamp current/mA . . 7.0 Damping.. .. .. .. . . Maximum (D) Argon flow-rate/ml min-l . . . . 450 Sample time/min . . . . . . 1.0 Wash timelmin . . .. . . .. 2.0 Recorder span, full-scale/mV . . .. 5.0 Chart speed/cm min-1 . . . . .. 0.5 196.0 0.3 4.5 Maximum (D) 400 1.0 2.0 5.0 0.5 Results and Discussion Sample Decomposition In agreement with other workers,1°-12 we found that wet digestion with the nitric acid- perchloric acid mixture yielded the optimum recoveries of arsenic and selenium. The relative proportions of the acids stated in the procedure were chosen so that the digestion would be effected predominantly by nitric acid.The use of the perchloric acid was merely to facilitate oxidation of the resistant fatty material and the organoarsenic or organoselenium compounds. As nitric acid has been shown to interfere with the hydride generation of a r ~ e n i c ~ ~ , ~ ~ it should be removed from the solution by boiling it off. Also, the digestion with the nitric acid - perchloric acid should not be taken to complete dryness in order to minimise any losses of160 SUBRAMANIAN AND MBRANGER: RAPID HYDRIDE EVOLUTION - Analyst, VoZ. 107' selenium.11 Therefore, in this work the decomposition with the nitric acid - perchloric acid mixture was continued only until the solution was nearly (but not completely) dry. Inter- ference from the small amount of residual perchloric acid is unlikely because as much as 15% perchloric acid did not cause any change in sensitivity for either arsenic or selenium.16 Optimisation of Instrumental Variables The manifold tube sizes were selected by trial and error until the optimum sensitivity was attained.A 1-min sampling and 8-min washing cycle governed by a two-lobed cam was found to give absorption peaks with the least distortion and that also returned smoothly to a reason- ably steady base line. Introduction of air into the system through the two manifold tubes made the peaks sharper and also allowed faster return to the base line. The optimum argon flow lay in the range 0.35-0.50 1 min-l for both arsenic and selenium. Flow-rates lower than 0.35 I min-1 produced distorted peaks while flow-rates higher than 0.5 1 min-1 decreased the sensitivity because of the combined effects of dilution and shortened residence time.The Variac voltage was adjusted manually until the optimum atomic-absorption signal was obtained for the arsenic or selenium, whichever was being determined. Optimisation of Solution Conditions We were interested in a rapid, routine method capable of handling large numbers of samples and therefore one with a wide tolerance in operating conditions and concentrations of reagents. In the final procedure sample volumes could vary from 3 to 5 ml, the concentration of hydro- chloric acid pumped through the manifold could be 2 M or greater for arsenic(II1) and 6 M or greater for selenium(IV), and the tetrahydroborate(II1) concentration could vary from 0.5 to 5.0% for arsenic(II1) and from 0.5 to 1.0% for selenium(1V).Interferences At ten times the detection limit of either arsenic(II1) or selenium(IV), no interferences occurred from at least 5000 mg 1-1 each of Na+, K+, Mg2+, Ca2+, Mn2+, AP+, NO3-, C104-, C1-, B r , SO4% and Po43- and 100 mg 1-1 each of Ba2+, Cd2+, C$+, Cr6+, Fe2+, Mo6+, Ti4+, V5+ and Zn2+. For example, with arsenic(III), Cu2+, Co2+ and Ni2+ could be tolerated only up to 5 mg 1-l; Hg2+ and Pb2+ only up to 2 mg 1-1; Bi3+, Sn2+ and Te2+ up to 0.4 mg 1-l; and Sb3+ up to 0.08 mg 1-l. For selenium(IV), the per- missible interference levels were Cu2+, Co2+, Ni2+, Pb2+ and Te2+ 1 mg l-l, Hg2+ 0.2 mg 1-1 and Bi3+ and Sn2+ 0.1 mg 1-l. These interferences could be prevented or at least minimised by separating the arsenic(II1) and selenium(1V) using coprecipitation with lanthanum hydrox- ide.16 Interferences did occur from a number of other elements.Analytical Parameters The sensitivity (concentration corresponding to 0.0044 A), detection limit (concentration corresponding to three times the standard deviation of the blank value) and the linear range for arsenic in 2 M hydrochloric acid were 1.0, 1.0 and 0-60 ng ml-l, respectively. The corre- sponding values for selenium(1V) were 1.0, 1.0 and 0-50 ng ml-l. Thus, in a 1.0-g sample made up to a final volume of 10 ml one can determine as low as 10 ng of arsenic or selenium per gram of sample and as high as 600 ng g1 of arsenic or 500 ng g-l of selenium. The lower limit of detection may be improved by decreasing the sample volume or by increasing the sample size.Similarly, the upper limit may be extended by increasing the final sample volume or by decreasing the sample size. For samples analysed on the same day, the coefficient of variation at the 95% confidence interval was 21, 12 and 7% at 5, 10 and 20 times the detection limits, respectively, of both arsenic and selenium. Considering the levels involved the precision was deemed satisfactory. Analyses of bovine liver (NBS Standard Reference Material 1577) gave values of 47 & 5 ng g-l of arsenic and 920 & 40 ng g-l of selenium as against the certified/provisional values of 60 ng g1 of arsenic and 1100 ng g-l of selenium, respectively. This corresponded to recoveries of 75 and 84% for arsenic and selenium, respectively.The rather lowFebruary, 1982 ELECTROTHERMAL ATOMISATION AAS FOR As AND SE 161 recovery obtained for arsenic and selenium in bovine liver could not be attributed to any inter- ference effect as the major and trace elements in bovine liver were present below the inter- ference levels noted earlier. Also, such low recoveries may not be due to volatilisation losses because some random samples of liver and kidney spiked with arsenic and selenium showed good recoveries. As can be seen from Table 11, the recoveries of arsenic and selenium were 90-105 and 95-1 lo%, respectively. TABLE I1 RECOVERY OF ARSENIC AND SELENIUM FROM SOME HUMAN KIDNEY AND LIVER SAMPLES Arsenic Sample type Sample No. Liver . . .. 9159 9174 9182 9196 9252 Medulla .. .. 9207 9221 9249 Cortex . . . . 9242 9250 9253 Spikelng kg-l* 25 10 25 10 25 25 25 25 10 10 25 Recovery, % 100 90 100 90 90 105 90 90 90 100 90 Selenium Spikelng g l * Recovery, % 100 97 50 105 100 110 50 98 100 95 100 97 100 102 100 98 50 100 50 110 100 105 L I 1 * Corresponds to mass of wet sample. Application to Samples The proposed method was applied to the determination of total arsenic and selenium in 44 samples of liver, 39 samples of cortex and 42 samples of medulla. All of the samples analysed contained arsenic below the detection limit of the method, i.e., 10 ng g-l of wet sample. The levels of arsenic reported in human kidney and liver by various workers17-19 exhibit wide variations. This makes comparison of our results with those of other workers rather difficult.However, our findings are in agreement with those of Brune et aZ.,19 who from careful work with control samples found values of 3.0 and 4.0 ng g-l of arsenic in human liver and kidney cortex, respectively. Thus, the concentration of arsenic found in the liver and kidney of unexposed Canadians appears to be extremely low. Table I11 gives the median and extreme values of selenium in the liver, cortex and medulla samples. In general, these values are in good agreement with those in the literature for unexposed humans.17-19 The histogram in Fig. 2 shows the frequency distribution of selen- ium in the liver, cortex and medulla samples analysed in this work. Note that nearly 60% of the liver samples and 50% each of the medulla and cortex samples contain selenium in the range 0.35-0.51, 0.10-0.20 and 0.51-0.90 mg k g l of wet sample, respectively.The liver and kidney tissue samples and the NBS bovine liver were analysed by Barringer Magenta Limited, Rexdale, Ontario, Canada, under contract No. 714. TABLE I11 SELENIUM CONCENTRATION IN KIDNEY AND LIVER FROM SOME CANADIAN ADULTS Selenium concentrationlmg k g l * Number of samples A \ Sample type analysed Median Extremes Cortex . . .. .. 39 0.84 0.22-1.30 Medulla.. .. .. 42 0.31 0.10-0.82 Liver . . .. .. 44 0.39 0.20-0.65 * Corresponds to mass of wet sample.162 SUBRAMANIAN AND MBRANGER A L Medulla 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 8 4 0 0.5 1 :o 1.5 Selenium concentration/mg kg-’ Fig. 2. Histograms showing levels of selenium in autopsied kidney and liver samples of some Canadian adults. References Hernberg, S., in Hiatt, H.H., Editor, “Origins of Human Cancer,’’ Volume 4, Cold Spring Harbor Laboratories, New York, 1977, pp. 147-157. Fowler, B. A., Ishinishi, N., Tsuchiya, K., and Vahter, M., in Friberg, L., Nordberg, G. F., and Vouk, V. B., Editors, “Handbook on the Toxicology of Metals,” Elsevier, Amsterdam, 1979, Lansford, M. G., and Calabrese, E. J., Med. Hypotheses, 1979, 5, 877. Glover, J., Levander, O., Parizek, J., and Vouk, V. B., iy, Friberg, L., Nordberg, G. F., and Vouk, Elsevier, Amsterdam, 1979, pp. 556-577. Vijan, P. N., Raynen, A. C., Sturgiss, D., and Wood, G . R., Anal. Chim. Acta, 1976, 82, 329. Vijan, P. N., and Wood, G. R., Talanta, 1976, 23, 89. Subramanian, K. S., and Sastri, V. S., Talanta, 1980, 27, 469. Molokhia, M. M., and Smith, H., Arch. Envivon. Health, 1967, 15, 745. Subramanian, K. S., and Mkranger, J. C., . I d . J. Environ. Anal. Chem., 1979, 7 , 25. Gorsuch, T. T., Analyst, 1959, 84, 135. Bajo, S., Anal. Chem., 1978, 50, 649. Pahlavanpour, B., Pullen, J. H., and Thompson, M., Analyst, 1980, 105, 274. Pierce, F. D., and Brown, H. R., Anal. Chem., 1977, 49, 1417. Kang, H. K., and Valentine, J . L., Anal. Chem., 1977, 49, 1829. Agemian, H., and Thomson, R., Analyst, 1980, 105, 902. Bedard, M., and Kerbyson, J. D., Can. J . Spectrosc., 1976, 21, 64. Iyengar, G. V., Kollmer, W. E., and Bowen, H. J. M., “The Elemental Composition of Human Johnson, C. A., Lewin, J. F., and Fleming, P. A., Anal. Chim. Acta, 1976, 82, 79. Brune, D., Nordberg, G., and Wester, P. O., Sci. Total Environ., 1980, 16, 13. pp. 293-319. V. B., Editors, “Handbook on the Toxicology of Metals, Tissues and Body Fluids,” Verlag Chemie, Weinheim/Bergstrasse, 1978. Received August 19th, 1981 Accepted September 3rd, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700157

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

Analyst, Febwary, 1982, Vol. 107, $9. 163-171 Reduction of Matrix Interferences in the 163 Determination of Lead in Aqueous Samples by Atomic-absorption Spectrophotometry with Electrothermal Atomisation with Lanthanum Pre-treatment M. P. Bertenshaw" and D. Gelsthorpe Directorate of ScientiBc Services, Severn-Trent Water A uthority, Nottingham Regional Laboratory, Meadow Lane, Nottingham, NG2 3HN and K. C. Wheatstone Directorate of Scientijic Services, Severn-Trent Water Authority, A belson House, 2297 Coventry Road, Sheldon, Birmingham, B26 3P U A previous paper described a lanthanum pre-treatment procedure for over- coming matrix interferences in the determination of lead in drinking water. In this paper, the amounts of lanthanum and nitric acid employed have been optimised such that the technique is now applicable to a wide range of aqueous samples, for example river waters, borehole waters, sewage effluents and trade effluents. The technique has been tested and found to be satisfactory for samples containing up to 1 150 mg 1-1 of chloride, 1420 mg 1-1 of sulphate, 760 mg 2-1 of sodium and 1530 mg 1-1 total hardness (as calcium carbonate).The optimum pre-treatment conditions for samples was 1% V/V nitric acid and 0.05% m/ V of lanthanum (as lanthanum chloride), which completely overcame suppressive interferences in the determination of lead, and gave a furnace tube lifetime of approximately 600 firings. This investigation enabled a close study to be made of the processes by which interferences are overcome by nitric acid - lanthanum matrix modification, and some possible mechanisms are presented and discussed. Keywords : Lead determination ; aqueous samples ; lanthanum $re-treatment ; electrothermal atomisation ; interference mechanisms Analysis by atomic-absorption spectrophotometry using eiectrothennal atomisation is at present the fastest and most convenient means of determining lead at low concentrations (it?., microgram per litre levels) in aqueous samples.Unfortunately, the method of direct analysis is prone to severe negative bias owing to interferences from the sample matrix. These effects may be measured and minimised by standard addition,l but this procedure increases the time of analysis to such an extent that the technique may become unfavourable when compared with alternative methods of trace metal analysis, such as chelation - solvent extraction or evaporation followed by flame atomic-absorption spectrophotometry.Various methods for eliminating interferences in lead atomisation have been proposed ; these not only include the use of chemical releasing agents and matrix modifiers, such as EDTA,2 ascorbic a ~ i d , ~ , ~ phosphoric acid,5s6 thiourea,' ammonium nitrates-lo and mineral acids,11-15 but also certain graphite-coating materials such as carbides of molybdenum,5~s~1s tantalum,1°J7 and lanthanum. l6 P-~O It is evident from the literature that controversy exists concerning the nature and mechan- isms of interference phenomena ; for instance, some workers consider the well documented interference from chlorides to be due to vapour phase reactions during at~misationll~l~s~~ whilst others favour occlusion of analyte by a solid p h a ~ e .l ~ , ~ l With such uncertainty in determining causal effects and the likelihood that interference mechanisms may be dependent on instrumentation and conditions used,9s22 it seems likely that most of the reagents listed above are applicable only to a limited range of atomiser types. There are convincing reportsl61l8-20 that lanthanum may be used to overcome suppression of the lead signal owing to the sample matrix during electrothermal atomisation. There is also * Present address : Directorate of Scientific Services, Malvern Regional Laboratory, 141 Church Street, Malvern, Worcestershire, WR14 2AA.164 BERTENSHAW et al. : REDUCTION OF MATRIX INTERFERENCES Analyst, VoZ. I07 considerable e v i d e n ~ e l l - ~ ~ ~ ~ ~ that nitric acid is effective in removing unwanted matrix prior to atomisation of the analyte.In our previous paper,l8 concentrations of 0.01 yo m/V of lanthanum and 1% nitric acid were used successfully for the automated determination of lead in drinking waters. Under these conditions, over 500 firings were obtained from furnace tubes before they required replacement. However, it was suspected that although these concentrations of lanthanum and nitric acid were sufficient for drinking waters, they may be insufficient to cope with more difficult sample matrices. At higher lanthanum concentrations, however, furnace-tube life was found to be markedly reduced; for example, only 50-100 firings were obtained when a concentration of 0.1% m/V of lanthanum was used.This was not so for manual injection by syringe where sample solution was dissipated over a much wider surface area in the furnace tube. Other workersl6J7 have reported similar findings concerning degrada- tion of carbon in the presence of excessive amounts of lanthanum, although the presence of lanthanum in moderate amounts has been claimed to prolong furnace lifetime.24 Therefore, it is desirable to establish the minimum concentration of lanthanum to be used that is consistent with overcoming the suppression of the lead signal in a particular sample matrix. Experimental Reagents Nitric acid, 70% ml V , atomic-s~ectroscopy grade. Lanthanum chloride (LaCl,. 7 H20) , analytical-reagent grade.Lead nitrate standard solzltion (1 OOO mg 1-1 of lead), atomic-spectroscopy grade. Apparatus The electrothermal atomiser used was an Instrumentation Laboratory IL555 controlled- temperature furnace fitted with a single-piece, non-pyrolytically coated carbon furnace tube (Cat. No. 42411). Nitrogen was used as purge gas. Samples were introduced by means of an IL254 Fastac automatic sampler and injection device using disposable polystyrene sample cups of 4-ml capacity. An IL251 spectrophotometer fitted with deuterium hollow-cathode lamp background correction and a Juniper lead hollow-cathode lamp was used for absorption measurements. Output from the spectrophotometer was monitored using a Linseis LS4 flat- bed chart recorder. Fast-response traces of background absorption and background-corrected lead absorption were recorded simultaneously on a Medelec FOR-4.2 oscilloscope via a specially constructed "sample-and-hold" circuit.Procedure The manufacturer's recommended settings for the spectrophotometer were used for back- ground-corrected lead absorption measurement at a wavelength of 217 nm in the single-beam mode. The furnace programme used for automated, spray-injection of sample was as shown in Table I. TABLE I FURNACE PROGRAMME Stage Setting Drying cycle .. .. . . 175 "C; 5 s 350 "C to 650 "C; 25 s Atomisation . . .. . . 650 "C to 2300 O C ; 2.5 s (ramp) Cleaning cycle . . .. . . Hold at 2300 "C; 7.5 s Ashing cycle . . .. . . 175 "C to 350 O C ; 5 s Note that the furnace control unit used was modified to incorporate a 2.5-s temperature ramp on atomisation in order to improve the sensitivity over that obtained with the 5-s ramp used in previous work.18 A 5-s deposit time was used with the automatic sampler for spray injection of approximately 20 pl of sample.Samples were stored in polyethylene containers and made 0.1% V/V with respect to nitric acid. Standard solutions were prepared daily in polypropylene calibrated flasks and preserved in 0.1% V/V nitric acid. Prior to analysis, samples and standards were adjusted to containFebruary, 1982 IN DETERMINATION OF PB IN AQUEOUS SAMPLES BY AAS 165 certain test concentrations of nitric acid and lanthanum (as lanthanum chloride) ; combinations of the following concentrations were tested: 0.1,0.5 and 1.0% V/V nitric acid with 0.00, 0.01, 0.03, 0.05 and O.lOyo m/V of lanthanum.Lanthanum impregnated furnace tubes1*S2O were not used, but where analyses were to be carried out in the presence of lanthanum, tubes were pre-conditioned with five replicate injections of 0.5% m/V lanthanum solution. No tube pre-treatment was employed for analyses carried out in the absence of lanthanum. Several injections of lead standard solution (50 pg 1-1 of lead) containing appropriate test concentrations of nitric acid and lanthanum were made until adequate precision was achieved (relative standard deviation less than 5%) prior to analysis of the samples. Samples were analysed in random order with a standard solution (50 pg 1-1 of lead) placed after every three samples. The pooled mean absorbance of the standard for each sample run (at least ten kings) was used for calibration. All results quoted are means of duplicate injections.Results and Discussion A range of samples, which included river waters, borehole waters, a trade effluent and a sewage effluent, were selected for the investigation. The samples represent a much wider range of compositions with respect to their major constituents than those reported previously18 ; the concentrations of six major species of potential interest in connection with suppression of the lead signal are given in Table 11, together with total hardness data. Lead concentrations in these samples were determined by the accepted method of standard additions1 and the con- centrations in all samples were then adjusted to contain 50 pg 1-' of lead to enable direct com- parison with standard solutions of the same concentration.Table I11 gives recovery data for 50 pg 1-1 of lead in samples using combinations of nitric acid and lanthanum, with concentra- tions as specified above. In order to elucidate the effects of individual matrix constituents on lead recoveries, correla- tions between data given in Tables I1 and I11 were examined. For drinking water samples investigated in our earlier work,18 it was suspected that total hardness was the main factor responsible for the suppression of the lead signals, although it was considered that this was not the only factor. However, recoveries for samples reported here show a poor correlation with total hardness data given in Table 11. A much better correlation is seen when recoveries are compared with sulphate concentrations.Therefore, in Table I11 samples are arranged in order of increasing sulphate concentrations. This is exemplified by comparing samples D and E; recovery in sample D, which contained high sulphate and moderately high total hardness (460 mg 1-1 and 498 mg I-l, respectively), was generally much worse than in sample E, which contained relatively low sulphate and very high total hardness (126 mg 1-1 and 1499 mg l-l, respectively). The dominant effect of sulphate in suppressing the lead signal is further sub- stantiated in data given in our previous paper; (see Table VII in reference 18), which indicated the effect of total hardness on suppression; in fact, suppression also correlates with sulphate concentration. TABLE I1 CONCENTRATION OF MAJOR CONSTITUENTS IN THE SAMPLES TESTED Concentrationlmg 1-1 Sample A .... B .. .. c .. .. D .. .. E . . .. F .. .. G .. .. H .. .. Sample type River water River water River water River water Trade effluent Sewage effluent Borehole water Borehole water Borehole water Borehole water Borehole water Ca*+ Mg*+ 33 5 68 14 100 24 173 16 560 24 54 6 128 19 148 36 210 48 300 63 465 90 Total hardness 103 228 349 498 1499 160 398 518 722 1009 1532 (CaCO,) Na+ 13 32 47 93 109 96 19 20 210 360 7 60 K+ 5 3 7 36 8 8 14 6 15 9 8 c1- 18 44 63 98 1120 120 36 31 267 590 1150 so,*- 48 88 187 460 126 154 86 257 600 860 1420166 BERTENSHAW et al. : REDUCTION OF MATRIX INTERFERENCES Analyst, VoZ. I07 TABLE I11 PERCENTAGE RECOVERIES OF LEAD IN SPIKED SAMPLES IN THE PRESENCE OF VARIOUS CONCENTRATIONS OF NITRIC ACID AND LANTHANUM Samples are spiked to a .concentration of 50yg 1-1 of lead, and are from left to right. arranged in order of increasing sulphate content Nitric acid Lanthanum Recovery of lead, yo* concentration, concentration, A % V / V %m/V ‘A G B E F C H D I J K 0.1 0 64 72 68 70 61 50 37 24 24 18 17 0.01 103 98 99 79 79 83 45 36 25 32 30 0.03 111 96 96 80 95 88 49 38 30 23 18 0.05 97 89 89 67 90 91 56 63 23 20 14 0.10 90 88 86 88 58 73 46 21 20 20 23 > 0.5 0 66 80 76 84 74 65 55 41 40 41 43 0.01 97 106 101 97 96 87 71 50 51 51 62 0.03 101 108 105 100 89 110 93 70 71 65 57 0.05 101 114 114 105 106 109 99 108 101 111 75 0.10 97 110 98 105 102 106 115 108 75 91 58 1.0 0 77 79 77 89 82 69 49 41 40 55 54 0.01 95 100 96 100 98 79 61 66 51 61 70 0.03 101 115 108 104 96 110 88 75 88 82 85 0.05 101 107 106 112 101 116 100 99 96 107 101 0.10 96 102 98 93 97 106 109 108 97 108 98 Original lead concentration Sulphateconcentration/mgl-l 48 86 88 126 154 187 257 460 600 850 1420 before spikinglygl-l... . 2.8 1.0 (1.0 2.2 13.3 2.1 d . 0 <1.0 <1.0 < L O <1.0 Note that percentage recoveries are subject to uncertainty arising from the determination of lead in both the spiked and un- spiked samples. The percentage recoveries quoted here are the mean of duplicate analyses of both spiked and unspiked samples and it is estimated that the 95% confidence interval of each of the recovery results is approximately 10-12%. Samples F, C and D showed a decrease in recovery on increasing lanthanum concentration from 0.05 to 0.1% at low nitric acid strength; this was probably a result of background over- correction as there was insufficient nitric acid present to “ash” the added chloride matrix.Overall, use of 0.5% nitric acid substantially improved recoveries but 1% nitric acid acted most f avourably. However, using only O.Olyo of lanthanum, together with 0.5 or 1% nitric acid, recoveries of 95-106y0 were obtained in samples containing as much as 154 mg 1-1 of sulphate, 1120 mg 1-1 of chloride and 1500 mg 1-1 total hardness (as calcium carbonate). As these represent the highest levels of chloride and hardness investigated, it may be concluded that the presence of lanthanum, even at very low levels, is sufficient to counteract interference from chloride or hardness pro- vided that an adequate concentration of nitric acid (Le., greater than 0.5% V / V ) is used.Samples containing higher levels of sulphate required both more lanthanum and more nitric acid to overcome suppression such that in the presence of 1% nitric acid, 0.03% m/V of lanthanum was sufficient to compensate for approximately 200 mg 1-1 of sulphate whereas 0.05y0 m/V of lanthanum was required for 1400 mg 1-1 of sulphate. Full recovery was observed with 0.1% m/V of lanthanum in samples tested and therefore this amount is probably excessive for the concentrations of sulphate tested. Zatkal’ used tantalum coating in preference to lanthanum because this forms an inert carbide ; although 350-400 fulngs were obtained from furnace tubes, potential interferences remained the same as those in untreated tubes.Poldoski16 has used an additional molyb- denum coating to combat reaction between lanthanum carbide and water. Although this treatment appears to be satisfactory, tube lifetimes were not determined but were considered to be in excess of 300 firings, with lanthanum replenishment in samples at a concentration of 0.05% m/V of lanthanum. claim only 150 determinations from lanthanum or zirconium coated tubes. L a g a ~ ~ ~ obtained between 200 and 400 firings from furnace tubes treated with samples containing 0.1% m/V of lanthanum compared with about half that number from untreated tubes. Thompson et aLZ0 obtained 200 to 300 firings from furnace tubes fired with samples containing 0.125% m/V of lanthanum.Our earlier work1* showed that about 500 firings could be achieved using either 0.01% m/V of lanthanum with automated injection or 0.1% mJV of lanthanum with manual injection. The work reported in this paper indicates that a concentration of 0.05% m/V of lanthanum and 1% V/V nitric acid should be used for automated injection of samples with more difficult matrices ; under these conditions about 600 firings were obtained from the furnace tubes used. All samples tested gave unacceptable recoveries in the absence of lanthanum. Runnels etFebruary, 1982 overcoming suppressive interferences on the lead signal. mean recoveries for all samples under various conditions of matrix modification (Fig. 1). IN DETERMINATION OF PB IN AQUEOUS SAMPLES BY AAS 167 Data given in Table I11 clearly demonstrate the effects of nitric acid and lanthanum in This can be summarised as pooled ?! 50 Q1 30 10 0 0.1 0.5 1 .o Concentration of nitric acid, % V N Fig.1. Average recovery versus concentration of nitric acid for different Lanthanum concentration: A, 0 ; B, 0.01; C, 0.03; See Table I11 for individual results. levels of lanthanum. D, 0.05; and E, 0.10% m/V. Fig. 2 shows, in more detail, the effect of nitric acid and lanthanum on atomisation traces for sample K made up to contain 50 pg 1-1 of lead; this sample contained the highest concentra- tion of sulphate and of dissolved salts of all the samples tested. In the absence of lanthanum [Fig. 2(a) and ( b ) ] , increasing nitric acid concentration not only reduced the background absorption signal but also increased the appearance temperature of lead from approximately 800 to 10oO "C.This temperature region was virtually free from background interference in the presence of 1% V/V nitric acid [Fig. 2(b)], but the lead peak was still considerably sup- pressed [compare with Fig. 2(e)]. In the presence of 0.1% m/V of lanthanum and only 0.1% V/V nitric acid [Fig. 2(c)] the background corrector clearly had difficulty in accom- modating the spiky background absorption signal owing to additional chloride matrix (from lanthanum chloride) and produced negative peaks on the corrected signal as a result of electronic over-correction; addition of 1% V/V nitric acid and 0.1 yo m/V of lanthanum [Fig 2(d)] gave a smooth background signal of smaller amplitude that was easily corrected electronically so that the lead peak could be measured.In this instance, the lead peak showed a similar profile to that of a synthetic standard containing only lead (50 pg 1-1 of lead), nitric acid and lanthanum chloride [Fig. 2 ( e ) ] . The data presented above show several consistencies with observations made by other workers. Firstly, Salmon et have recently discussed the effects of chemisorbed oxygen on lead atomisation. They proposed that blocking of active sites on graphite by chemisorbed oxygen may cause a shift to a higher temperature release mechanism. As the optimum oxygen absorption temperature of 500°C is close to the usual ashing temperature for lead determination, the partial pressure of oxygen during ashing may largely determine the mechanism of lead atomisation and hence the appearance temperature.This perhaps explains the observed in- crease in appearance temperature where, in the absence of lanthanum, the nitric acid concentra- tion was increased from 0.1 to 1% V/V, thus increasing the oxygen pressure [Fig. 2(a) and ( b ) ] . Using high temperature equilibrium calculations, Frech and Cedergren12 have shown that a high partial pressure of oxygen may allow formation of gaseous lead oxide unless ashing takes place at a sufficiently high temperature to decompose any sodium nitrate formed as a result of the reaction between sodium chloride and nitric acid. In this work, an ashing temperature of 650 "C was used, which is well above the stability temperature of 380 "C given for sodium Oxygen adsorption on graphite is only temporary as chemisorbed oxygen is completely removed at about 950 oC.26 Carbide coatings, however, are more permanent and apparently exhibit similar behaviour in controlling atomisation mechanisms.26 This is borne out by168 BERTENSHAW et al.: REDUCTION OF MATRIX INTERFERENCES Analyst, VoZ. I07 temperature profiles in Fig. 2; in the presence of lanthanum, only the higher appearance temperature is evident [Fig. 2(d) and (e)]. Fig. 2. Absorption traces for sample K containing 50 pg 1-l of lead under vari- ious conditions of matrix modification. Broken line, background absorption signal; and solid line, correc- ted signal. (a) 0.1% nitric acid (no lanthanum); (b) 1% nitric acid (no lanthanum); (c) 0.1% nitric acid + 0.1% lanthanum; (d) 1% nitric acid + 0.1% lanthanum; and (e) trace of 50 pgl-1 lead standard solution containing 1% nitric acid + 0.1% lanthanum and also tempera- ture profile as measured by IL555 control unit.Secondly, excess of nitric acid is instrumental in removing chloride from the matrix by This helps to prevent the following (2) Volatilisation of undissociated lead chloride,ll which is otherwise seen as a lowering of appearance temperature of the dissociated leadl3; this offers an alternative explanation to the lower temperature lead peak in Fig. 2(a). volatilisation as hydrogen chloride during ashing. chemical, physico-chemical and spectral interferences.February, 1982 IN DETERMINATION OF PB IN AQUEOUS SAMPLES BY AAS 169 (ii) Formation of chloride crystal, which may occlude lead atoms at normal appearance temperatures.(iii) Volatilisation of sodium chloride, which may cause large molecular absorption at the normal appearance temperature of lead ; this may cause problems with electronic background correction if A further observation concerns interference owing to calcium ~ h l o r i d e , ~ , ~ , ~ ~ which is reported to be unaffected by the presence of nitric acid.l5 As the boiling-point of calcium chloride is in excess of 1600 OC2' and in this work no background signal was perceptible on atomising this salt, it is likely that the interference on the lead signal is caused by persistence of crystallinity of the salt beyond the atomisation temperature of lead resulting in occlusion of analyte.By its nature this type of interference would be expected to produce variable levels of suppression in different tyes of atomisers, as demonstrated in Table IV. as in Fig. 2(c). REPORTED SUPPRESSIONS OWING TO CALCIUM CHLORIDE IN RELATION TO ATOMISER TYPES CaCl, concentration/ Source Atomiser type Matrix mg 1-1 Suppression, % reference CRA 63 .. .. 20% HNO, 100 21 15 1000 60 HGA72 .. . . 0.1% HCl 277 36 3 HGA 2200 . . . . 0.1% HNO, 277 10 10 5 550 17 IL455 . . .. - 338 0 6 IL 555 . . . . 1% HNO, 1387 45 This paper Interference owing to sulphate, principally sodium sulphate, has also been widely reported.5,6s10~12s15~1g~2g As with calcium chloride, sodium sulphate interference appears to be unaffected by the addition of nitric acid,15 although reduced to some extent by addition of ammonium nitrate,lO and shows a highly variable degree of suppression with different atomisers (Table V) .Frech and Cedergren12 reported on a large background signal from sodium sulphate even after ashing at 1000 "C; this demonstrates the high temperature stablity of sodium sulphate, which may therefore act as an interferent by physical inhibition of lead atomisation owing to remaining crystals. Backman and Karlssona proposed a mechanism of non-volatile lead sulphide formation for interference observed in the direct analysis of sulphide-bearing steels and alloys. However, this mechanism is unlikely to be active in samples containing a large excess of oxyanion as provided by nitric acid. TABLE V REPORTED SUPPRESSIONS OWING TO SODIUM SULPHATE IN RELATION TO ATOMISER TYPES Na,SO, concentration/ Atomiser type Matrix mg 1-l Suppression, yo CRA 63 .... 20% HNO, 100 22 1000 50 HGA 7 4 .. .. - 22 18 8 872 68 HGA 2100 . . .. HNO, 148 86 HGA 2200 . . . . 0.1% HNO, 1480 89 5916 94 IL 455 . . .. - 3 300 12 IL 555 . . . . 1% HNO, 740 20 Source reference 15 19 6 10 5 This paper Although specific interference tests using sodium sulphate alone in the IL furnace did not indicate the potential severity of sulphate interference, data presented in this and other work (Table V) show that sulphate can cause large suppressions when acting in combination with other matrix constituents. This supports the conclusions of Regan and Warren4 who suggest that individual interferents are probably modified by other species present in the matrix.If occlusion of analyte within inorganic matrix crystals is the principal mechanism of interference,170 BERTENSHAW et aZ. : REDUCTION OF MATRIX INTERFERENCES AnaZyst, VoZ. 107 as proposed by Krasowski and Copeland21 and supported in this work, then processes of crystal nucleation and disruption are of primary importance in determining the extent of suppression. It is easily envisaged that these processes may be largely controlled by rates of heating, atomiser type, graphite-tube wall condition and matrix constituents present that may provide sites for crystal nucleation. have shown that lanthanum in the presence of nitric acid effectively overcomes matrix interferences. In particular, Anderssonlg has demon- strated the control of sulphate interference with lanthanum, which is supported in our work by data given in Table 111.Also, in this study a specific interference test using 500 mg 1-1 of calcium (as the chloride), showed that suppression was substantially reduced in the presence of lanthanum. If, as appears to be so, lanthanum is effective in overcoming sodium sulphate and calcium chloride interferences, both of which are unaffected by addition of nitric acid15 and appear to result from occlusion of analyte, then a common mechanism of operation is suggested. The fact that lanthanum reacts to form a carbide coating on the walls of carbon furnace tubes has been used in the analysis of carbide-forming elements, such as beryllium, and in this context acts by preventing formation of analyte ~ a r b i d e .~ ~ ~ ~ ~ However, lead does not form a stable carbide, so this is clearly not the reason for lanthanum’s effectiveness in the determination of lead. Coating materials such as molybdenum and tantalum form stable and inert inter- stitial carbides but lanthanum carbide is known to react with ~ a t e r . ~ ~ , ~ ~ s ~ ~ , ~ 7 Runnels et aZ.25 have proposed the formation of lanthanum hydroxide and hydrocarbons when an aqueous sample is introduced into a lanthanum carbide coated furnace. It is this property, that lanthanum may enter the solution from the furnace tube walls, which allows direct interaction with the sample matrix. This may be chemical interaction, but in the light of proposed inter- ference mechanisms, is more likely to be due to inhibition of microcrystalline salt formation.The reported deterioration of graphite in the presence of an excess of lanthanum16,25,29 is a direct result of the reaction between lanthanum carbide and water.25 Although it would appear that lanthanum as well as carbon is lost by this process, Runnels et point out that lanthanum hydroxide is reduced to carbide at temperatures of around 1950 “C. Therefore, it is only necessary to replenish a certain portion of the lanthanum coating with samples intro- duced into the furnace. Our earlier work18 and previous Conclusions Suppression of the lead signal during determinations by carbon furnace atomic-absorption spectrophotometry may be overcome in many difficult sample matrices by the addition of nitric acid and lanthanum. Nitric acid acts by removing chloride from the matrix, whilst lanthanum appears to alter the crystallinity of matrix salts during ashing, thus preventing occlusion of the analyte.Additionally, both nitric acid and lanthanum may cause deactiva- tion of sites on the graphite that would otherwise favour a lower temperature atomisation mechanism. Using a matrix of 1% V/V nitric acid and 0.05% m/V of lanthanum, full recovery of lead was obtained for samples containing up to 1 150 mg 1-1 of chloride, 1420 mg 1-1 of sulphate, 760 mg 1-1 of sodium and total hardness up to 1530 mg 1-1 (as calcium carbonate). Under these conditions of sample treatment the furnace tube lifetime was approximately 600 firings. The authors thank W. F. Lester, Director of Scientific Services, Severn-Trent Water Author- ity, for permission to publish this work.References 1. Ranson, L., and Orpwood, B., “An Evaluation of an Electrothermal Device for the Determination of Lead and Cadmium in Potable Water,” Water Research Centre Technical Report No. 49, 1977. 2. DolinSek, F., and Stupar, J., Analyst, 1973, 98, 841. 3. Regan, J. G. T., and Warren, J., Analyst, 1976, 101, 220. 4. Regan, J. G. T., and Warren, J., Analyst, 1978, 103, 447. 5. Hodges, D. J., Analyst, 1977, 102, 66. 6. Callio, S., At. Sfiectrosc., 1980, 1, 80. 7. Ohta, K., and Suzuki, M., Fresenius 2. Anal. Chem., 1979, 298, 140. 8. Ediger, R. D., At. Absorpt. Newsl., 1975, 98, 127. 9. Manning, D. C., and Slavin, W., Anal. Chem., 1978, 50, 1234.February, 1982 IN DETERMINATION OF PB IN AQUEOUS SAMPLES BY AAS 171 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. Halliday, M. C., Houghton, C., and Ottaway, J. M., Anal. Chim. Acta, 1980, 119, 67. Shaw, F., and Ottaway, J . M., Analyst, 1974, 99, 184. Frech, W., and Cedergren, A., Anal. Chim., 1977, 88, 57. Czobik, G. J., and Matousek, J. P.. Anal. Chem., 1978, 50, 3. Fordham, A. W., J . Geochem. Explor., 1978, 10, 41. Hageman, L. R., Nichols, J . A., Viswanadham, P., and Woodriff, R., Anal. Chem., 1979, 51, 1406. Poldoski, J. E., Anal. Chem., 1980, 52, 1147. Zatka, V. J., Anal. Chem., 1978, 50, 538. Bertenshaw, M. P., Gelsthorpe, D., and Wheatstone, K. C., Analyst, 1981, 106, 23. Andersson, A., At. Absorpt. Newsl., 1976, 15, 71. Thompson, K. C., Wagstaff, K., and Wheatsone, K. C., Analyst, 1977, 102, 310. Krasowski, J. A., and Copeland, T. R., Anal. Chem., 1979, 51, 1843. Fuller, C. W ., “Electrothermal Atomisation for Atomic Absorption Spectrometry,” The Chemical Churella, D. J., and Copeland, T. R., Anal. Chem., 1978, 50, 309. Lagas, P., Anal. Chim. Acta, 1978, 98, 261. Runnels, J. H., Merryfield, R., and Fisher, H. B., Anal. Chem., 1975, 47, 1258. Salmon, S. G., Davis, R. H., Jr., and Holcombe, J . A., Anal. Chem., 1981, 53, 324. Weast, R. C . , Editor, “Handbook of Chemistry and Physics,” Fifty-fourth Edition, Chemical Rubber Co. Press., Cleveland, Ohio, 1974. Segar, D. A., and Cantillo, A. Y., in Gibb, T. R. P., Editor, “Analytic Methods in Oceanography,” American Chemical Society, Washington, D.C., 1975, p. 75. Backman, S., and Karlsson, R. W., Analyst, 1979, 104, 1017. Society, London, 1977, p. 61. Received June 29th, 1981 Accepted September 4th, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700163

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

172 Analyst, February, 1982, Vol. 107, f$. 172-178 A Study of Pneumatic Nebulisation Systems for Inductively Coupled Plasma Emission Spectrometry L. Ebdon” and M. R. CaveT Department of Chemistry, Shefield City Polytechnic, Pond Street, Shefield, S1 1 WB The analytical performance of different pneumatic nebulisers and cloud chambers for inductively coupled plasma emission spectrometry is reported. A vortex cloud chamber and a double-pass cloud chamber were compared for use with concentric glass nebulisers. An all-plastic double-pass cloud chamber was preferred. Two new nebulisers for solutions containing high levels of dissolved solids or slurries are described. One of these, machined entirely from inert plastic, gave an improved performance compared with a glass concentric nebuliser and no problems were encountered with the nebulisation of solutions containing 20% m/ V of dissolved solids or slurries.Keywords : Pneumatic nebulisers ; inductively coupled plasma ; atomic- emission spectrometry ; cloud chambers ; slurries The problem of sample introduction is increasingly being recognised as a major limiting factor in the application of the inductively coupled plasma (ICP) for analytical optical emission spectrometry. The difficulties of pneumatic nebulisation with the low argon flow-rates avail- able, the so-called “injector gas flow-rate,” have prompted the investigation of alternative approaches to nebulisation, such as ultrasonic nebulisation,l and of alternative means of sample introduction, such as vaporisation from a carbon rod2 or hydride generation.3 Pneu- matic nebulisation remains, however, the most popular form of sample introduction as it is simple, inexpensive and robust, without being particularly prone to memory effects.Provided that precautions are taken, matrix effects are minimal. Pneumatic nebulisation is, in general, compatible with the ICP without the need for desolvation. The miniaturised concentric pneumatic nebuliser* remains probably the most popular nebuliser. Conventional concentric nebulisers constructed from metal, as used in flame spectrometry, are of limited use with the ICP. The flow of gas required to operate them is too high for the optimum operating conditions of the p l a ~ m a . ~ For use with an ICP, such nebu- lisers are usually constructed from glass and consist of a narrow annulus surrounding a narrow sample capillary.The high-velocity flow of the nebuliser gas in the annulus produces reduced pressure at the capillary tip, drawing sample through the capillary to the tip where it is shattered into a fine aerosol. Scotts has described with clarity the construction of such a nebuliser. More recently improvements to this type of nebuliser, designed to reduce salt build-up and droplet size, have been evaluated.’ The former problem of salt blockages can also be ameliorated by wetting the injector gas stream.8 In this design two capillaries are mounted at right-angles, one acting as the sample uptake capillary and the second carrying the gas flow. The accurate positioning of the capillaries is vital in order to achieve efficient nebulisation, and initially this was achieved by careful adjustment of the capillaries. More than one uptake capillary may be used mounted close to the gas jet.1° Recently the original cross-flow design has been modified to make it more robust and of a fixed geometry.llSl2 The fine capillaries were replaced with more substantial glass tubes drawn down to fine jets at the ends.The jets were then positioned so that a given solution uptake rate was obtained at the required gas flow and pressure. The tubes were then fused together by a glass bar so that, once constructed, the nebuliser required no further adjustment. Meddings et aZ. ,I2 using such a modified design, with careful control of gas flows, have been able to obtain a relative standard deviation of 0.5% in routine analysis in their laboratory.Ape1 et aZ.13 have developed a novel nebuliser in which the sample was pumped on to a fritted glass disc through which the nebuliser gas passed, producing an aerosol of sample solution. The nebuliser was highly efficient but could handle only small sample solution flow-rates and some memory effects and drift were observed. Plymouth, Devon, PL4 8AA. Cross-flow nebulisers9 are also widely used with the ICP. * Present address : Department of Environmental Sciences, Plymouth Polytechnic, Drake Circus, t Present address : Department of Chemistry, University of Massachusetts, Amherst, Mass. 01003, USA.EBDON AND CAVE 173 A major disadvantage of cross-flow and concentric nebulisers is that the sample solution has to pass through a narrow capillary (about 0.5 mm i.d.) , so that care must be taken that sample solutions contain no small particles that can block the nebuliser.It has also been found that solutions containing a high salt content (5-10%) can cause a build-up of material at the capillary tip , impairing nebuliser performance and eventually causing blockage. Babingtonl* originally described a nebuliser in which the gas issued from a small orifice into a flowing solution. Such a device is clearly less prone to clogging by suspended solids or solutions of high salt content. Nebulisers based on this principle have recently been proposed for use with the ICP1"17 and the growing number of conference reports of their use testifies to their apparent popularity.18 The use of a cloud chamber appears to be imperative with these nebulisers before the aerosol is injected into the ICP. The cloud chamber essentially acts as a sorting device, allowing only the droplets below a certain size to reach the neb~1iser.l~ Hence the role of the cloud chamber must be considered in any study of nebulisers, as others have previously d e m o n ~ t r a t e d .~ ~ ~ ~ The two most popular types of cloud chamber are the double-pass cloud chamber described for ICP work by Scott et aZ.,21 in which the aerosol is forced down an inner tube and back up an outer concentric tube before exit, and the vortex-type cloud chamber, in which the aerosol is shot tangentially into a cyclone chamber, the larger droplets falling to the drain at the base and the aerosol leaving from a tube a t the top of the chamber.In the study reported here, two concentric glass nebulisers were combined with a double- pass and a cyclone cloud chamber and their performances were assessed. Neither of these nebulisers is suitable for high-solids or slurry nebulisation, or for use with solutions that contain hydrofluoric acid. Two Babington-type nebulisers were therefore designed to be suited to these conditions, the second being made entirely from PTFE and assessed using an all-plastic double-pass cloud chamber. The basis of assessment of the nebulisers and chambers was both nebulisation efficiency (the ratio of aerosol produced to solution uptake per unit time) and the emission signal to noise ratio in a plasma. The plasma was operated under optimum conditions as identified by simplex optimisation. Experimental The free-running r.f.generator, demountable torch, spectrometer and read-out system used have been described previously.22 All quoted power levels refer to power in the plasma as measured calorimetrically.22 Nebuliser Construction One commercial all-glass concentric nebuliser was used (Meinhard, Type T-230-A2 ; J. E. Meinhard Associates, Tustin, Calif., USA) and a similar nebuliser constructed in the manner described by Scott.6 A glass and plastic high-solids nebuliser (Fig. 1) was constructed from glass tubing (10 mm o.d.), which was sealed at one end so as to produce a short (1.5 mm long) capillary. The sealed end was then ground away, forming a 90" groove, until the end of the capillary became open and an argon flow of 1 1 min-l a t about 30 lb in-2 back-pressure was obtained.,4 sample introduction tube was then constructed (as shown in Fig. 1). The two parts were mounted in a former and a thermosetting plastic resin was poured around them, allowing the glass components to be fixed in the correct position. The second high-solids nebuliser was constructed in one piece from a length of PTFE rod (13 mm o.d.), which was machined as shown in Fig. 2. The sample was fed to the groove at the end of the rod using a length of PTFE tubing (1.0 mm id.). The solution supplied to the high-solids nebulisers was pumped by a small peristaltic pump (Mark IV, 60 rev min-l; Schuco Scientific, London) at 1.65 ml min-l. The concentric nebulisers were either operated at their natural uptake rates (commercial, 2.5 ml min-l; laboratory constructed, 0.9 ml min-l) or pumped in the same way at 1.65 ml min-l. Cloud Chambers The above nebulisers were combined with two different cloud chambers. The first, a vortex- type chamber from a flame spectrometer (SP900; Pye Unicam, Cambridge), is shown in Fig.3. The diameter of the cyclone chamber at the top was about 110 mm and the height of the chamber 175 mm. The second cloud chamber was constructed from two concentric plastic174 EBDON AND CAVE : STUDY OF PNEUMATIC NEBULISATION Analyst, VoZ. 107' Glass sample inlet tube A Glass gas inlettube Plastic body D Fig. 1. Glass and plastic high-solids nebuliser. SamDle inlet I . I I D I I I Gas 'inlet PTFE body __*I + I "== :@ 90" n M A = l mm B -0.4 mm C = 3 m m 0 = 2 m m Fig.2. PTFE high-solids nebuliser. cylinders (see Fig. 4), the outer (55 mm i.d.) being sealed at one end and fitted with an outlet and a drain, and the inner (23 rn i.d.) accepting the nebulisers via an appropriate bung. The gap between the open end of the inner cylinder and the closed end of the outer could be adjusted within wide limits. To plasma To plasma Sampl 1 To drain To drain Fig. 3. Vortex cloud chamber. Fig. 4. Double-pass cloud chamber.February, 1982 SYSTEMS FOR ICP EMISSION SPECTROMETRY 176 Measurement of Nebulisation Efficiency The uptake of aqueous solutions from a measuring cylinder was recorded over a given period of time during which the aerosol produced at the outlet end of the cloud chamber was collected in two pre-weighed U-tubes filled with silica desiccant and connected in series.The U-tubes were then weighed again (no significant increase in the mass of the second U-tube was ever observed). Plasma Operating Conditions Table I. The conditions for the measurement of aluminium and manganese emission are given in TABLE I PLASMA OPERATING CONDITIONS I Wavelength/nm . . .. .. .. Injector gas (Ar) flow-rate/l min-1. . Plasma gas (Ar) flow-rate/l min-l . . Coolant gas . . .. .. .. .. Coolant gas flow-rate/l min-l . . . . Viewing height*/mm .. .. . . Power in plasma/kW . . .. .. .. . . * Height above the load coil. Element Al(1) Mn(I1) Mn(1) 396.1 257.6 403.1 A 1 0.60 0.50 0.58 1.7 1.7 11.3 6.2 Ar N2 12.4 32 18 27 0.40 0.42 0.42 Reagents All chemicals were of analytical-reagent grade.Results and Discussion Comparison of Concentric Nebulisers and Cloud Chambers Table I1 summarises the results obtained in comparing the two concentric nebulisers at their natural uptake rates, and the two cloud chambers. The vortex, or cyclone, cloud chamber produced a greater proportion of aerosol than the double-pass cloud chamber for both nebulis- ers, but the gain in signal to noise ratio was not as large. This indicated that some larger droplets were reaching the plasma when the vortex chamber was used. This resulted in increased noise levels, offsetting some of the signal gain, and some condensation of water drop- lets in the capillary tip of the injector. The latter gave rise to some spitting problems. By replacing the capillary tip injector with a jet-type injector (see Fig.5) this condensation was prevented. The jet injector produced a less well defined aerosol stream compared with the narrow filament of aerosol produced by the more laminar flow conditions obtained with the capillary injector. Hence reduced amounts of analyte passed into the central channel of the plasma, as illustrated by results for aluminium (see Fig. 6). The jet injector produced both a reduced signal and increased noise on both background and analytical signals. TABLE I1 COMPARISON OF CONCENTRIC NEBULISERS AND DIFFERENT CLOUD CHAMBERS Nebulisation Signal to noise ratio Concentric nebuliser Cloud chamber efficiency, yo for 10 pg ml-l of Al Laboratory constructed . . . . Double-pass 4.0 6.1 Laboratory constructed .. . . Vortex 5.5 7.6 Commercial . . .. . . Double-pass 0.97 6.4 Commercial . . .. . . Vortex 2.9 9.7176 EBDON AND CAVE: STUDY OF PNEUMATIC NEBULISATION Analyst, VoZ. 107 Fig. 5. Injector tubes. These problems negated the small advantage to be gained from the cyclone chamber and the double-pass cloud chamber was preferred for later experiments. A comparison of our own laboratory-constructed nebuliser and the commercial nebuliser showed that it is possible to construct simply concentric nebulisers that give analytical results comparable to those obtained with the best commercial nebulisers. Further, this was achieved at a lower uptake rate of 0.9 ml min-l compared with 2.5 ml rnin-l. The enhanced nebulis- ation efficiency ensured that equivalent amounts of aerosol were delivered to the plasma with a consequent saving in sample consumption when using our nebuliser. Although the results using the cyclone chamber do not appear to be fully consistent, presumably because of the effects noted above, there does seem to be a further gain in signal to noise ratio per unit uptake rate for the laboratory-constructed nebuliser, suggesting that the types of droplet sizes pro- duced are in the desired size range.Performance of High-solids Nebulisers A comparison of the performances of the laboratory-constructed concentric nebuliser and the glass and plastic high-solids nebuliser is perhaps best obtained by inspection of Fig. 7. The latter nebuliser produced a 60% larger signal for the 6 pg d-l manganese solution when sample was pumped to it at 1.65 ml min-l and the concentric nebuliser was operated at its natural uptake rate of 0.9 ml min-l. Both nebulisers were operated, in this instance, with the double-pass cloud chamber and capillary injector and using the 257.6-nm manganese ion line.Unfortunately, pulses from the simple, and inexpensive, peristaltic pump used with the high- solids nebuliser caused much greater noise on both the background and analytical signals. P A n (3); Min Distilled water ’ Ld Time Fig. 6. Comparison of signal and noise levels from jet and capillary injector tubes for 10 pgml-I of aluminium. A, Capillary injector; and B, jet injector. Time Fig. 7. Comparison of concentric and glass - plastic high-solids nebylisers using 6 pg ml-l of manganese. A, Concentric nebuliser; and B, high-solids nebuliser.Febrzcary, 1982 SYSTEMS FOR ICP EMISSION SPECTROMETRY 177 The results were, however, sufficiently promising to lead us to construct the PTFE high- solids nebuliser in which the sample entry was placed nearer the gas exit orifice to reduce the pulsation effect.Fig. 8 shows the success of this approach using the 403.1-nm manganese atom line in an all-argon plasma and a 1 pg ml-l manganese solution. The signal for the all- PTFE nebuliser was about 20% higher than that obtained with the concentric nebuliser, similar noise levels being achieved with both nebulisers. Time Fig. 8. Comparison of (A) PTFE high-solids nebuliser and (B) glass concentric nebuliser using 1 pg ml-l of manganese. The PTFE nebuliser coped well with high-solids solutions and even with slurries without blocking, as illustrated in Fig.9. A 1 pg ml-l manganese solution containing 20% m/V of sodium chloride was pumped at 1.65 ml min-l to both nebulisers, and again the atom line was monitored. The enhancement of the observed manganese emission signals with high levels of sodium can be seen, as can some relative improvement of the signal obtained with the con- centric nebuliser consequent upon pumping. The concentric nebuliser quickly clogged, how- ever, and this is clearly seen in the erratic and noisy signal obtained. A P% Time Fig. 9. Comparison of high-solids and glass con- centric nebulisers using a 1 pgml-l manganese solu- tion containing 20% m/V sodium chloride. A, PTFE high-solids nebuliser ; and B, glass concentric nebuliser.Conclusion These experiments have confirmed the low levels of efficiency of nebulisers used for ICP emission spectrometry and the importance of considering other aspects of the sample introduc-178 EBDON AND CAVE tion system such as the cloud chamber and injector tube. A double-pass cloud chamber and a capillary injector tube, with a capillary at least 20 mm long, were found to be preferable. Although well suited to flame atomic-absorption spectrometry, concentric pneumatic nebu- lisers may not retain their current popularity in ICP emission spectrometry, because the low flows of nebulisation gas available have led to the development of miniaturised nebulisers that are not optimal for solutions of high salt content. The ability of the ICP to atomise efficiently samples with high dissolved solids contents, and even slurries, suggests that nebulisers designed to generate aerosols from such samples, as well as from conventional dilute aqueous solutions, will ultimately be recognised as optimal for this source.The simply constructed all-PTFE nebuliser described here has demonstrated competitive performance with established nebu- lisers, plus the ability to handle solutions of high solids content. The combination of an inert plastic nebuliser and an all-plastic cloud chamber offers promise for work with hydrofluoric acid solutions, which cannot be aspirated with glass concentric or cross-flow nebulisers. This PTFE nebuliser is currently being used to aspirate slurries into the plasma and no problems with clogging have been encountered.We thank the Science Research Council and the London Scandinavian Metallurgical Company, Rotherham, for a grant which made this work possible. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. References Wendt, R. H., and Fassel, V. A., Anal. Chem., 1965, 37, 920. Gunn, A. M., Millard, D. L., and Kirkbright, G. F., Analyst, 1978, 103, 1066. Thompson, M., Pahlavanpour, B., Walton, S. J., and Kirkbright, G. F., Analyst, 1978, 103, 568. Meinhard, J . E., ICP Inf. Newsl., 1976, 2, 163. Ebdon, L., Cave, M. R., and Mowthorpe, D. J., Anal. Chim. Acta, 1980, 115, 179. Scott, R. H., ICP Inf. Newsl., 1978, 3, 425. Bogdain, B., ICP Inf. Newsl., 1978, 3, 491. McQuaker, N. R., Kluckner, P. D., and Chang, G. N., Anal. Chem., 1979, 51, 888. Kniseley, R. N.. Amenson, H., Butler, C. C., and Fassel, V. A., Appl. Spectrosc., 1974, 28, 285. Donohue, D. L., and Carter, J. A., Anal. Chem., 1978, 50, 686. Novak, J . W., Lillie, D. E., Boorn, A. W., and Browner, R. F., Anal. Chem., 1980, 52, 576. Meddings, B., Kaiser, H., and Anderson, H., Paper presented at International Winter Conference, Developments in Atomic Plasma Spectrochemical Analyses, San Juan, January, 1980. Apel, C. T., Bieniewski, T. M., Cox, L. E., and Steinhaus, D. W., ICP Inf. Newsl., 1977, 3, 1. Babington, R. S., Pop. Sci., May 1973, 102. Suddendorf, R. F., and Boyer, K. W., Anal. Chem., 1978, 50, 1769. Wolcott, J . F., and Sobel, C. B., Appl. Spectrosc., 1978, 32, 591. Garbarino, J. R., and Taylor, H. E., ApPl. Spectrosc., 1980, 34, 584. “Annual Reports on Analytical Atomic Spectroscopy,” Volumes 8, 9 and 10, Royal Society of Novak, J. W., and Browner, R. F., Anal. Chem., 1980, 52, 792. Greenfield, S., McGeachin, H. McD., and Chambers, F. A., ICP I n , . Newsl., 1977, 3, 117. Scott, R. H., Fassel, V. A., Kniseley, R. N., and Nixon, D. E., Anal. Chem., 1974, 46, 75. Ebdon, L., Mowthorpe, D. J., and Cave, M. R., Anal. Chirn. Acta, 1980, 115, 171. Chemistry, London, 1979, 1980, 1981. Received August 17th, 1981 Accepted September llth, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700172

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

Analyst, February, 1982, Vol. 107, pp. 179-184 179 Apparent and Real Reducing Ability of Polypropylene in Cold-vapour Atomic-absorption Spectrophotometric Determinations of Mercury A. Kuldvere GeoZogicaZ Survey of Norway, P.O. Box 3006, N-7001 Trondheim, Norway Experiments undertaken with polypropylene reaction flasks in connection with the Perkin-Elmer MHS- 1 Mercury/Hydride System show that poly- propylene very quickly adsorbs tin(I1) chloride, which is not then removed by usual routine rinsing. This can cause mercury to be lost rapidly from solu- tions stored in uncleaned flasks as well as giving signals for mercury in cold- vapour atomic-absorption methods even if reductant is not added. This apparent ability of polypropylene to reduce mercury(I1) is rapid and ceases with careful cleaning of the apparatus.This apart, the polypropylene re- action flasks have a real, relatively slow, but considerable ability to reduce mercury( 11) , which was especially conspicuous if solutions without preserva- tives were agitated in them. Keywords Mercury( I I ) determination ; reduction by polypropylene ; atomic- absorption spectrophotornetry ; cold-vapour method In the course of analytical work at the Geological Survey of Norway (NGU) mercury signals have been observed in the cold-vapour atomic-absorption spectrophotometric determination of mercury with our mercury vaporisation system, the Perkin-Elmer MHS-1, without using any reducing agent. Such signals have received little attention in the literature. Koirtyohann and Khalill reported mercury signals due to reduction by polypropylene when using a plastic syringe procedure for mercury determination.They claimed that the reduction was caused by 2,6-di-tert-butyl-4-methylphenol, commonly called butylated hydrox-y toluene (BHT), which is added as an antioxidant to the polypropylene formulation by the manufacturer. Heiden and Aikens,2 using the Hatch and Ott m e t h ~ d , ~ later found that BHT does not reduce mercury(I1) rapidly, ie., not within the time of determination (1 min), and they attributed the mercury( 11) reducing ability to the polypropylene itself. Gutenmann et aL4 have also observed similar signals from mercury without any added reducing agent. They explained their observation by the volatilisation of mercury(I1) chloride with subsequent photolytic decomposition in the absorption cell to yield atomic mercury, an explanation that has later been refuted by Koirtyohann and Khalil.1 At NGU, we use 100-ml polypropylene flasks as reaction vessels for the reduction of mercury samples using tin(I1) chloride solution.In addition, the immersion tube as well as the manifold of the MHS-15 is of polypropylene. Therefore, the anticipated reducing ability of polypropylene in the apparatus under daily use for mercury determination was in great need of careful investi- gation. This paper presents the experimental results concerning the reduction of mercury(I1) by the rinsed and cleaned polypropylene flasks used as reaction vessels in connection with the Perkin- Elmer MHS-1 Mercury/Hydride System. The term “rinsed” will be used here to mean a thorough, short routine rinsing of reaction flasks as well as immersion tube and magnetic stirrer with 3% V/V hydrochloric acid and water as recommended by the manufacturer,5 whereas “cleaned” refers to the cleaning process given under Procedure in this paper.Experimental Apparatus A Perkin-Elmer, Model 403, atomic-absorption spectrophotometer, equipped with an MHS-1 Mercury/Hyd ride System, Model 056, recorder and a Perkin-Elmer mercury electrodeless dis- charge lamp, was used. Argon was used as purge gas and 100-ml polypropylene flasks as the reaction vessels where the reduction of mercury(I1) with tin(I1) chloride was carried out. The operating parameters are given in Table I.180 KULDVERE : APPARENT AND REAL REDUCING ABILITY OF Analyst, Vol.I07 TABLE I INSTRUMENTAL PARAMETERS Spectrophotometer parameters- Light source .. .. .. .. .. EDL Wavelength/nm .. .. .. . . 253.6 Recorder full-scale/A . . .. .. . . 0.25 Slit (spectral slit width) /nm .. .. . . 3 (0.2) Recorder response (time constant)/s . . . . 2 (1) Chart speedlmm min-l . . .. .. . . 10 RangelmV . . .. .. .. .. . . 5 Recorder fiarameters- MHS-I fiarameters- Programme .. .. .. .. .. . . HgII Sample volume/ml . . .. .. .. .. 31 Temperature of silica cell/"C . . .. . . 250 Reduction solution/ml . . .. .. . , 2.5 Reagents and Solutions Glass-distilled water and chemicals of analytical-reagent grade were used throughout this work, unless stated otherwise. Potassium permanganate (mercury free), 5% m/V solution. One drop of this solution, added per determination in the following experiments, is equal to approximately 50 pl or 0.016 mmol of potassium permanganate.A 10% m/V solution of tin(I1) chloride (SnC1,.2H20) in 5% V/V hydro- chloric acid. To free the solution of any contaminating mercury, argon was bubbled through it. The solution contained 5% V/V nitric acid and O.Olyo m/V potassium dichromate as preservatives6s7 and was prepared by suitable dilution from a stock solution* (lo00 mg 1-l; Titrisol, Merck). Solutions containing 0.1 and 3.0 M of nitric acid were prepared from 65% acid (Merck) by appropriate dilutions with water and were used without standardisation. Reducing agent. Merczlry(I1) working standard solution, 0.1 mg 1-l. Nitric acid solutions. Procedure To ensure that all parts of the apparatus that had been in contact with tin(I1) chloride were free from trace amounts of the reductant, the immersion tube with magnetic stirrer and pellet disk as well as the reagent capillary and all reaction flasks were washed thoroughly with detergent solution using a suitable brush, and then with water.Afterwards, they were immersed in 6 M hydrochloric acid for 1 d, then washed with water and immersed for several hours in 3 M nitric acid and washed with water again. Thereafter the cleaned parts were immersed in doubly distilled water for 2 d to remove the last trace amounts of the adsorbed reductant. To study the mercury losses from dilute solutions stored in cleaned reaction flasks, 30 ml of water, 0.1 or 3.0 M nitric acid solution were dispensed into separate flasks followed by 1 ml of the mercury(I1) working standard solution.In some experiments one drop of the potassium permanganate solution (about 50 pl) was added. After appropriate standing times mercury signals were measured in the usual manner (Table I), i.e., 2.5 ml of the reducing agent were added according to the procedure. To measure the mercury signals, which appeared without added tin(I1) chloride reagent, only the reagent pump was disconnected if routinely rinsed apparatus was used. When using the cleaned apparatus the reagent capillary was also disconnected and cleaned. In both instances, 2.5 ml of water or 0.1 or 3.0 M nitric acid were added manually instead of the re- ductant and in addition to the sample volume (31 ml), to uncleaned and to cleaned reaction flasks to give final volumes equal to the series of experiments performed with added reductant.To study the effect of agitation on mercury loss from aqueous solutions stored in cleaned reaction flasks, 30 ml of water and 1 ml of mercury working solution were transferred into the reaction flasks. To half of them one drop of permanganate solution was added. The flasksFebruary, 1982 POLYPROPYLENE IN COLD-VAPOUR AAS DETERMINATION OF HG 181 were allowed to stand for different periods of time, some of them without and some with vigor- ous mechanical stirring. The results of these experiments are shown in Figs. 1 and 2 and in Tables II-IV and are discussed below. Results and Discussion On using polypropylene flasks continuously in routine work as reaction vessels, the reducing power of the MHS-1 system was found to be sufficient to reduce mercury standard solutions [3.2 p.p.b.(parts per lo9)] without adding tin(I1) chloride (Table 11). This seemed very curious at first ; however, after about 7 subsequent determinations with the same flask without added reductant (reagent pump disconnected) the response from mercury began to decrease and, after a further 15 determinations, nearly disappeared. It was also observed that reac- tion flasks that were allowed to stand unused for weeks gave inferior first signals in these determinations (Table 11). TABLE I1 ABSORBANCES FOR 100 ng OF MERCURY(II) IN AQUEOUS SOLUTION WITHOUT ADDED REDUCTANT USING THE ROUTINELY RINSED AND CLEANED POLYPROPYLENE APPARATUS The average absorbance of two runs for 100 ng of mercury(I1) with added tin(I1) chloride reagent is 0.101 (single determinations of 0.100 and 0.101) both with and without 0.016 mmol of potassium permanganate solution per determination Absorbance A I 1 Routinely rinsed apparatus Cleaned apparatus, Flask no.KMnO, not added 1 0.001 0.001 2 0.000 0.000 3 0.000 0.001 4 5 6 0.046t Pyrex Erlenmeyer flask o.oo2t KMnO, not added 0.102 0.099 0.100 0.103 0.100 0.101 0.062* 0.075* 0.016 mmol of KMnO, added- per determination 0.000 0.001 0.001 0.000 0.001 0.001 * The very first signals after the uncleaned flasks had been allowed to stand unused for t The cleaned vessels had been exposed five times to SnC1, in routine work and then 2 weeks. allowed to stand for 3 d in air; these are the first signals thereafter.On the basis of these observations and from the literature reports reviewed,ls2s4 it was felt that the reason for these signals might be trace amounts of tin(I1) chloride adsorbed on the appara- tus surfaces or the reducing ability of polypropylene itself. To verify one of these hypo- theses the apparatus was cleaned as described under Procedure. No response for mercury from solutions without added reductant was obtained using the cleaned apparatus (Table 11). The results of the above experiment may be taken as proof of the inability of the polypropy- lene material in the apparatus to reduce mercury(I1) rapidly, i e . , within the time of the determination (2 min), as well as the inability of eventual additives, such as BHT in the polypropylene material, to reduce mercury(I1) rapidly.The latter is in agreement with the report from Heiden and Aikens,2 which found that the BHT does not reduce mercury(I1) rapidly. It seems that polypropylene has an ability to adsorb tin(I1) chloride instantaneously, to resist its rinsing as well as to retard its oxidation by air. A cleaned polypropylene flask and a Pyrex 150-ml Erlenmeyer flask that were used for five determinations each in routine work and that had had the same rinsing treatment as usual in routine work were allowed to stand for 3 d in air without use. The determination then carried out without the use of the reducing agent, with the polypropylene flask showed about 50% of the normal mercury response for 100 ng of mercury(II), whereas the same experiment with the Erlenmeyer flask resulted in no response from mercury (Table 11).182 KULDVERE : APPARENT AND REAL REDUCING ABILITY OF Analyst, VoZ. I07 At the NGU, the polypropylene flasks are used as reaction flasks into which a mercury solution is transferred immediately before measurement ; sometimes they stand overnight, but as we always add one drop of permanganate solution in such instances, no erratic values have been observed under these conditions. Mercury standard solutions, which were also analysed without any added permanganate, immediately after transferring into an uncleaned reaction flask (i.e., routinely rinsed) showed good agreement with corresponding determinations per- formed with added permanganate (see for example previous work8 and/or the comment in Table 11).Low and inconsistent readings were occasionally observed for standards on standing without permanganate. The reason for this is clear from the above, as regards readings obtained with uncleaned flasks. As the experiments with cleaned reaction flasks have shown (Table 11), polypropylene does not reduce mercury(I1) rapidly, i.e., within 2 min. It seemed reasonable to ask if it does reduce, but slowly. Some experiments, therefore, have been performed on this (Figs. 1 and 2 and Tables I11 and IV). To ensure that the cleaning process described under Procedure in this paper was effective enough to remove all adsorbed tin(I1) chloride, some experiments were carried out with doubly cleaned reaction flasks (i.e., the cleaning procedure was applied twice). As shown in Table 111, agreement with results obtained with singly cleaned flasks is good.Admittedly, if tracer studies had been carried out with tin-113, for example, a simpler cleaning procedure could probably have been proposed. It is hoped that this will be carried out at a later stage. Fig. 1 shows the loss of mercury from solutions stored in open, cleaned polypropylene reac- tion flasks without permanganate. The greatest loss of mercury occurs from the aqueous solution (line C). This is in agreement with frequent reports (see for example references 6,7 and 9) that nitric acid has a preservative effect on mercury(I1) ions and on metal ions in general. Nevertheless, as regards The loss from 3.0 M nitric acid begins after about 24 h (line A).TABLE I11 LOSSES OF MERCURY FROM VIGOROUSLY STIRRED AND NON-STIRRED SOLUTIONS IN POLYPROPY- LENE REACTION FLASKS IN THE ABSENCE OF PERMANGANATE Absorbance* Agitated solutions Singly cleaned Doubly cleaned A I I Standing time/h flasks flasks 0 0.58 0.084 (0.099) 0.082 (0.101) 1 0.080 (0.100) 0.078 (0.101) 4 0.056 (0.098) 0.060 (0.100) 18 0.049 (0.100) 0.051 (0.099) Non-agitated solutions I Singly cleaned Doubly cleaned flasks flasks 0.101 0.100 0.099 (0.100) 0.097 (0.101) 0.094 (0.101) 0.096 (0.099) 0.091 (0.100) 0.089 (0.099) 0.076 (0.100) 0.080 (0.098) * Mercury, 100 ng, in aqueous solution. The figures in parentheses represent the absorbances of the same samples but in the presence of one drop of permanganate solution (0.016 mmol). TABLE IV MERCURY LOSSES ON STANDING FROM DIFFERENT SOLUTIONS IN THE PRESENCE OF PERMANGANATE Standing time*/h A I 1 Solutions 0 24 48 72 3.0 M nitric acid .. . . 0.101 0.101 0.099 0.100 0.1 M nitric acid . . . . 0.100 0.099 0.101 0.100 Aqueous .. .. . . 0.102 0.100 0.098 0.094 0.100 0.093 * All values are absorbances from solutions containing initially 3.2 p.p.b. of mercury(I1) with 0.016 mmol of potassium permanganate added per determination.February, 1982 POLYPROPYLENE IN COLD-VAPOUR AAS DETERMINATION OF HG 183 Timelh Fig. 1. Losses of mercury from (A) 3.0 M and (B) 0.1 M nitric acid solutions and (C) aqueous solution on standing in cleaned polypropylene reaction flasks in the absence of permanganate. Mercury- (11) concentration, 3.2 p.p.b. the storage of a mercury(I1) solution (3.2 p.p.b.) in polypropylene containers, preservation can- not be achieved for any length of time, especially not in low molarity nitric acid, as illustrated in Fig.1 (line B); an oxidising agent, such as permanganate, is also necessary. When these experiments were repeated (Table IV) in the presence of one drop of potassium permanganate solution (0.016 mmol of potassium permanganate) no loss of mercury occurred from the 0.1 and 3.0 M nitric acid solutions on standing for 72 h; a 5% loss from the aqueous solution was ob- served on standing for 72 h; none was observed after 48 h. In all experiments carried out in the presence of permanganate only 0.016 mmol of potassium pennanganate per determination was added. Fading of the pink colour on the third day in the aqueous solution, which is likely, suggests that there would have been no loss from this if more permanganate had been added. The amount of permanganate used, however, was found to be appropriate under the measuring conditions chosen.The commencement of the loss of mercury from aqueous solutions can be seen in detail in Fig. 2. A 15-min standing time without permanganate did not result in any loss of mercury. A longer standing time may cause mercury to be lost if the measurement itself is made without addition of the one drop of pennanganate solution (line A). However, if permanganate was added just before the determination (line B), then solutions stored without pennanganate 0.10 0.08 0.06 0.04 0 e a 0.02 t 0 5 10 15 20 25 30 35 Ti me/mi n Fig. 2. Loss of mercury from aqueous solutions stored in cleaned polypropylene reaction flasks in absence of permanganate.A, No potassium permanganate added a t any point; and B, potassium permanganate added just before measurement. Mercury(I1) concentration, 3.2 p.p.b.184 KULDVERE stayed at full strength for about 35min. This indicates that even if some mercury(I1) is already reduced after 15 min, this is, for a given time, accessible to permanganate oxidation and available for determination. Finally, it can be concluded that the behaviour of polypropylene is different from that of polyethylene as regards the agitation of mercury(I1) solutions in containers of the respective materials. As shown in Table 111, mercury is lost much more rapidly from agitated solutions stored in cleaned polypropylene reaction flasks without pennanganate than from non-agitated solutions.When using polyethylene flasks for sample storage, agitation, on the contrary, is one of the means of preventing mercury losses from solutions where addition of preservatives is not desirable as reported by Mahan and Mahan,lo because oxygen retards reduction of mercury(I1) in this instance. Conclusions Even if polypropylene has a considerable real and, in addition, an apparent ability to reduce mercury(II), the polypropylene reaction flasks can be used in the determination of mercury with the Perkin-Elmer MHS-1 without any risk of losing mercury. The analyst, however, must be aware of the fact that aqueous solutions of mercury(II), especially in uncleaned reaction flasks in the absence of an oxidant such as pennanganate, must be analysed immediately; in the presence of a slight excess of permanganate no loss of mercury occurs from aqueous solu- tions on standing for 2 d. As regards the signals from mercury during the determinations without added reducing agent when using the uncleaned polypropylene apparatus, it is conceivable that this phenome- non can be attributed to the ability of the polypropylene to adsorb tin(I1) chloride rapidly, to retard its oxidation by air as well as to resist the rinsing of the last trace amounts of reductant, which causes the premature reduction of mercury(I1). The author thanks Dr. R. Boyd of the Geology Division of the NGU for revision of the English. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. References Koirtyohann, S. R., and Khalil, M., Anal. Chem., 1976, 48, 136. Heiden, R. W., and Aikens, D. A., Anal. Chem., 1979, 51, 151. Hatch, W. R., and Ott, W. L., Anal. Chem., 1968, 40, 2085. Gutenmann, W. H., Lisk, D. J., and Grier, N., Bull. Environ. Contam. Toxicol., 1972, 8, 138. Perkin-Elmer MHS-1 Mercury/Hydride System, Operator's Manual, Bodenseewerk Perkin-Elmer, Christmann, D. R., and Ingle, J. D., Jr., Anal. Chim. Ada, 1976, 86, 53. Feldman, C., Anal. Chem., 1974, 46, 99. Kuldvere, A., and Andreassen, B. Th., At. Absorpt. Newsl., 1979, 18, 106. Lo, J. M., and Wahl, C. M., A n d . Chem., 1975, 47, 1869. Mahan, K. I., and Mahan, S. E., Anal. Chem., 1977, 49, 622. 1977. Received May llth, 1981 Accepted August 25th, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700179

出版商:RSC    

年代:1982

数据来源: RSC