摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1269 Determination of Total Chromium and Chromium(vi) in Animal Feeds by Electrothermal Atomic Absorption Spectrometry M. Elisa Soares M. Lourdes Bastos and Margarida A. Ferreira laboratory of Toxicology and Laboratory of Bromatology Faculty of Pharmacy University of Oporto R. Anibal Cunha 164 4000 Oporto Portugal An electrothermal atomic absorption spectrometric method was developed for the determination of total and hexavalent chromium in animal feeds. For measuring total chromium the samples were digested in a mixture of three acids (HN0,-HCI-HF). Chromium in its hexavalent state was dissolved in 0.01 mol I-' NaOH solution. A mixture of Pd+Mg was used as a chemical modifier. Extensive validation of the proposed method was carried out both by the standard additions method and by analysis of National Institute of Standards and Technology Standard Reference Material 1548 Total Diet.The detection limits were 0.53 and 0.42 pg I-' for total and hexavalent chromium respectively. Measurements can be made over a linear range between 0.53 and 50 and 0.42 and 50 pg I-' for total and hexavalent chromium respectively. The relative standard deviations were 4.5 and 7.0% for total and hexavalent chromium respectively hence the proposed method is satisfactory for routine analyses. In 70 samples that were analysed the mean levels found were 1.93 pg g-' (0.20-6.87 pg g-') and 230 ng g-' (10.0-780 ng g-I) for total chromium and chromium(vi) respectively. The large variations in the levels of metal found in the analysed feeds shows diverse contamination of the raw materials used in the preparation of the samples.Keywords Chromium; chromium(vr); electrothermal atomic absorption spectrometry; animal feeds Chromium is a biologically important element that is involved in glucose and lipid metabolism,' but it is also considered to be toxic. It is only toxic with excessive intake and the toxicity is mainly dependent on the oxidation state.2 The estimated safe and required dietary intake of Cr is 0.05-0.20 mg d-' .3 Particularly in its hexavalent oxidation form Cr is a major water pollutant usually as a result of industrial effluents. It accumulates in soils and therefore can contaminate crops and other plants or can be incorporated into these systematically from the Several vegetable materials such as grains seeds or leaves are constituents of animal feeds and pre-mixes that are supplemented with mineral additives and other nutrients.Hence it is important to monitor feeds for metallic elements that are toxic or even if they are essential elements to avoid problems of toxicity when their concentrations exceed the safe levels. The measurement of Cr levels is especially important for the reasons of the toxicity of CrV' and the biodependence of Cr"' . Several methodologies for the quantification of Cr have been described in the literature namely in biological and in Also several attempts have been made to quantify Cr both in its state as an essential element of nutrition (Cr"') and in its toxic state (CrV') particularly in water14-19 and in One of the most reliable techniques to measure metallic species at ng 8-l and sub-ng g-' levels is electrothermal atomic absorption spectrometry (ETAAS).However use of this method specifically to control Cr"' and CrV1 levels in feeds has not been reported in the literature. The proposed method was developed to provide a viable procedure to determine Cr in feedstuffs in order to verify contamination levels particularly the hexavalent content. Initially an evaluation of the method of sample preparation for dissolution of the total Cr using a digestion procedure from which good results have been obtained with another complex matrix was carried out.22 For extraction of CrV' a procedure previously applied by other workers to soils was adopted2'; this procedure was accurately evaluated in the present work.Next charring and atomization temperature studies were performed as well as studies of the range of calibration. The final phase of the work was establishing the quality assurance of the method both by the standard additions technique and by analysing a reference material. Subsequently the method was applied to the quantification of total Cr and CrV' in 70 feedstuff samples. Experimental Reagents To avoid contamination all poly( tetrafluoroethylene) (PTFE) materials pipettes micropipette tips autosampler cups and calibrated flasks were immersed for 24 h in freshly prepared 15% v/v HN03 and then rinsed thoroughly with doubly de-ionized water before use. All the acids and NaOH were of Suprapur grade (Merck).Ammonium nitrate was pro analysi grade (Merck). Chromium(II1) standards were prepared daily from a 1000mg I-' solution (Titrisol Merck). A solution of 1000 mg 1-' chromium(v1) was prepared by dissolving 2.829 g of K2Cr207 (AnalaR Merck) in 11 of de-ionized water; from this other diluted standard solutions were prepared daily. The chemical modifier used was a mixture of Pd(NO,) (Merck) and Mg(N03)2 (Merck) prepared by mixing equal volumes of solutions containing 3 and 2 g 1-' respectively in 15% v/v HNO3. The reference material was National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1548 Total Diet obtained from The Office of Reference Materials Laboratory of the Government Chemist Teddington UK. Apparatus A Perkin-Elmer HGA-700 furnace installed in a Model 1100B atomic absorption spectrometer with deuterium arc back- ground correction was used equipped with an AS-60 auto- sampler and an Epson EX-800 printer. The analyses were carried out using platforms ( Perkin-Elmer part No.AAPB109-324) inserted into pyrolytic graphite coated graphite tubes (Perkin-Elmer part No. AAPB109-322). A single- element hollow cathode lamp (Perkin-Elmer part No. AAPC 303-6039) was operated at 25 mA and all data were taken at the 357.9 nm wavelength. The integration time was 5 s. The slit-width was 0.7nm and argon was used as the purge gas with an internal flow rate of 300 ml min-'. Readings from the spectrometer were taken using the peak area mode.1270 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 Sample Preparation The samples used in the present study were of different compositions but heterogeneous with respect to particle size some of them being granulated feeds depending on the animal species for which they were intended. Thus the first step consisted of grinding the samples to a particle size ofjust 1 mm. Samples of approximately 0.5 g were accurately weighed in duplicate. One sample was digested in a 50 ml PTFE container with a mixture of concentrated HN03 (10 ml) HCl(5 ml) and HF (1 ml) and then heated just to dryness on a hot-plate. Further portions of HNO (10 ml) and HCl(5 ml) were added to the residue and the mixture was again evaporated to dryness. This residue was dissolved in 0.5 ml concentrated HNO and diluted with water to 25 ml in a calibrated flask.The total Cr was measured in this solution. The other 0.5g portion was placed in a 15 rnl poly(propy1ene) tube 9 ml of 0.01 mol 1-' NaOH solution added the cap fitted on and the tube then shaken horizontally in an oscillating agitator for 17 h at room temperature in order to extract the Crv' as has been described elsewhere.20 After the addition of 1 ml of 1 mol 1-' NH4N0 solution the sample was shaken briefly and centrifuged for 15 min (3000 rev min-'). The CrV' was measured in the supernatant liquid. Results and Discussion Optimization Programme for ETAAS The following solutions were used to optimize the ETAAS method an acidic CrV1 standard solution (50 pg 1-'); an acidic Cr"' standard solution (50 pg 1-'); a solution of the acid digested samples; an alkaline CrV' standard solution (50 pg 1-') prepared under the same alkaline conditions as the feed samples; and a solution resulting from the alkaline extraction of CrV' from the feeds. These solutions were used to establish the best charring and atomization temperatures and to obtain the most repeatable and sensitive signals in the shortest analysis time.The autosampler was programmed to pipette sequentially the modifier followed by the standard solution or the sample solution and to dispense them together onto the platform. The charring and atomization curves obtained for total Cr and Crvl in acidic and alkaline medium are shown in Figs. 1 and 2 respectively. For both situations the optimum tempera- tures were 1600 and 2500 "C for the charring and atomization respectively.The optimized ETAAS programme used is sum- marized in Table 1. As can be observed in the graphs the temperatures chosen correspond to the best signals for the five situations studied. As was expected the standard acidic solutions give the same readings for the Cr"' and CrV' stock solutions. Nevertheless the absorbance signals obtained for CrV' in the alkaline medium are lower. Thus the furnace programme adopted was the same al C (c1 e s n Q 0.6 - 0.5 0.4 0.3 - 0.2 - - - t d 1100 1300 1500 1700 1900 2100 2300 2500 2700 2900 Temperat u rePC Fig. 1 Charring and atomization graphs for total Cr in acidic CrV' and Cr"' standard solutions and in a solution of the digested samples O6 t 0.5 t al 5 0.4 2 0.3 0.2 0.1 e 2 Standard I 1100 1300 1500 1700 1900 2100 2300 2500 2700 2900 TemperaturePC Fig.2 Charring and atomization graphs for CrV1 in an alkaline Cr"' standard solution and in an alkaline extract of CrV1 from animal feeds Table 1 Optimized ETAAS programme Step Temperature/" C Ramp/s Hold/s Dry 1 130 15 15 Dry 2 300 15 15 Char 1600 30 30 Atomize 2500 0 5 Clean out 2650 2 3 for measuring total Cr and CrV' but it was necessary to use the appropriate standard solutions ie. in acidic and in alkaline media prepared for the CrV' standard. Analytical Curve and Detection Limit To evaluate the optimum calibration range for total Cr and for CrV' standard solutions of CrV' in acidic and alkaline media of from 0 to 100.0 pg I-' were used. Linearity was observed over the concentration range 0.53-50.0 pg 1-' for total Cr and 0.42-50.0 pg 1-' for Cr".To calculate the detection limits 20 determinations were carried out in 0.2% v/v HNO and in 0.01 mol 1-' NaOH- 1 mol I-' NH4 NO3 solution for total Cr and Crvl respectively. The values were calculated as the concentration corresponding to three times the standard deviation (SD) of the background noise and were respectively 0.53 and 0.42 pg 1-'. Validity of the Method Digesting difficult matrixes with a three-acid mixture (HN0,-HCl-HF) has produced good results.22 The use of high-purity grade acids and carrying out the digestion in PTFE containers ensures that the sample pre-treatment can be carried out without contamination. This was confirmed by including a blank assay in each batch of samples using the same amounts of acids and submitting these to the digestion procedure.To extract the Cr present in its hexavalent state the method of solubilization with NaOH solution that had been applied to soils was adopted,20 after testing this procedure with the present matrix. Along with the wet digestion procedure and the sample extraction procedure the method of standard additions was also performed for both Cr determinations. Standard acidic CrV' solutions were added to the feed samples and subjected to the over-all digestion procedure. Also alkaline CrV' standard solutions were added to the feed samples and again subjected to the alkaline extraction procedure. The results obtained in these studies are presented in Table 2. As can be observed the recovery study shows that the absorbance readings for the samples can be compared directly with the standard analytical curves obtained for total Cr and for Crvl.Validation of the digestion procedure was confirmed by1271 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Table 2 Statistical results for the recoveries obtained by the standard additions method Parameter Total Cr CrV' Concentration/pg 1- ' 5.0 10 25 2.5 5.0 10 6 6 6 10 10 10 n Recovery & SD (YO) 9 6 f 0 95+ 1 96+3 94+3 9 6 f 3 95+ 1 Table3 Comparison of recoveries of Cr"' and Cr'" added to one feed sample Cr found CrV' Cr"' Total Cr in sample/ added/ added/ found/ Recovery pgg-' pgml-' pgml-' 1gg-I (%I 3.8 5.0 0.0 8.3 90 3.8 10.0 0.0 14 102 3.8 0.0 5.0 8.7 98 3.8 0.0 10.0 13.8 100 Table 4 Principal mineral constituents in feedstuffs (Decreto-Lei No.57/1985 Diario de Republica No. 54 I Serie 06/03/85) used in the interference studies Major Minor Mineral Concentration (YO) Mineral Concentration/pg g- ' Phosphates 0.8 Cobalt 10 Calcium 1 .o Copper 175 Sodium 0.15 Manganese 250 Potassium 0.5 Zinc 250 Iron 0.12 analysing a reference material. Although it was impossible to obtain a reference material with the same composition as the animal feed samples NIST SRM N-1548 Total Diet was used. As the certified value was given only for total Cr (1.43 pg g-I) the acid digestion procedure was performed on ten 0.5 g portions and the mean value of the metal concentration obtained was 1.39 k0.13 pg g-' signifying a recovery of 99 _+ 3% (mean & SD). It is assumed that after digestion with strong acids Cr will be in its hexavalent form.To compare the recoveries of Cr"' and CrV1 added to feed samples and determined as total Cr a sample was spiked with standards of both species submitted to acid digestion and the total Cr measured. The results are summarized in Table 3. It can be concluded that after acid digestion the determination of total Cr does not depend on the oxidation state of the metal. As a consequence in the present procedure CrV1 standards prepared in an acidic or alkaline medium whichever was convenient were acceptable. The precision of the analytical method was evaluated by measuring the absorbance signals 20 times on the same digested sample and in the same alkaline extract of the sample. For evaluation of the precision of the over-all procedure readings of 20 different digested aliquots and of 20 different alkaline extracted samples were performed.The relative standard devi- ations (RSDs) ranged between 2.2 and 2.7% for the analytical method and between 4.5 and 7.0% for the over-all procedure for total Cr and Crv' respectively. These results are fully acceptable values bearing in mind the digestion step and all the manipulations involved. To study the interferent effects of the principal mineral constituents present in feedstuffs on the measurement of Cr mixed standard solutions containing the mean recommended levels for the various species (see Table 4) were ~repared.'~ For total Cr three aliquots of the standard mixed solution were added to three different concentration levels of standard Cr acidic solutions and the absorbance signals were measured under the analytical conditions described above.For Cr" three 1 ml aliquots of the mixed solution of minerals were Table5 Deviations (YO) from the expected values for total Cr and CrV' obtained in the interference studies Added concentration/pg 1- ' 5 10 25 Total Cr 6.3 5.7 6.7 Species (n=4) (n=4) (n = 4) CrV' 6.0 10.0 7.5 added to 0.01 moll-' NaOH solution (9 ml) and three different concentration levels of the aqueous CrV' standards and the mixtures shaken horizontally in an oscillating agitator for 17 h at room temperature. After adding 1 ml of 1 mol I-' NH4N0 solution the absorbance signals were obtained. The absorbance readings were interpolated on the analytical curves established independently for total Cr and Cr".The deviations from the expected values were always acceptable (< lo%) and are presented in Table 5. Preconcentration of the metal which is required for the analysis of waterlg was not needed because the levels of both total Cr and CrV' that were present in the feedstuffs were high enough to enable quantification without any enrichment steps. Applications A wide range of feeds a total of 70 samples were analysed by application of the ETAAS methodology developed. As would be expected compared with total Cr CrV' was present at low concentrations the mean value found being 230 ng g-' (range 10.0-780 ng g-'). The RSD was 60.9% which shows the great variability of levels present in the range of samples analysed. Nevertheless considering the potential toxicity of the metal in this oxidation state and its bioavailability it is advisable to establish where this potential danger to animal health is arising.For total Cr the mean value found was 1.93 pg g-' (0.20-6.87 pg g-I) the RSD being 77.2%. Considering that almost all of the metal is present as Cr"' and that in this form the metal is an essential element and poorly absorbed by the gastro-intestinal tract these high levels do not cause a toxicity problem for animals. It can be assumed that these different values result from the different degrees of contamination of the soils where the crops had been grown. Also the different technological procedures used in the preparation of the animal feeds can affect the oxidation state of the element.Conclusions An accurate and precise method to quantify total Cr and CrV1 in animal feeds is presented. The ETAAS conditions are the same for both determinations which facilitates the analyses being carried out using the same furnace programme this aspect being a very important criteria in routine analyses. This work received financial support from Junta Nacional de Investigaqiio Cientifica (Centro de Analise do Alimento) and AssociaqZo Nacional das Farmacias.1272 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1 2 3 4 5 6 7 8 9 10 11 12 13 References Katz S. A. and Salem H. J. Appl. Toxicol. 1993 13 217. Goyer R. A. Casarett and Doullk Toxicology Pergamon Press Oxford 4th edn. 1991. National Research Council (US) Subcommittee Recommended Dietary Allowances National Academy Press Washington DC 10th edn.1989. Cary E. E. Allaway W. H. and Olson 0. E. J. Agric. Food Chem. 1977 25 300. Morris B. W. Griffiths H. Hardisty C. A. and Kemp G. J. At. Spectrosc. 1989 10 1. Shan X.-q. Yan Z. and Ni 2.-m. At. Spectrosc. 1990 11 116. Paschal D. C. and Bailey G. G. At. Spectrosc. 1991 12 151. Cary E. E. and Rutzke M. J . Assoc. Off. Anal. Chem. 1983 66 850. Cary E. E. J. Assoc. 08. Anal. Chem. 1985 68 495. Johnson C. D. and Weaver C. M. J. Agric. Food Chem. 1986 34 436. Stoddard-Gilbert K. and Blincoe C. J. Agric. Food Chem. 1989 37 128. Cary E. E. and Kubota J. J. Agric. Food Chem. 1990 38 108. Miller-Ihli N. J. and Greene F. E. JAOC Int. 1992 75 354. 14 15 16 17 18 19 20 21 22 23 Syty A. Christensen R. G. and Rains T. C. At. Spectrosc. 1986 7 89. Johnson C. A. Anal. Chim. Acta 1990 238 273. Thomas O. Gallot S. and Naffrechoux E. Fresenius’ J. Anal. Chem. 1990 338 241. Gammelgaard B. Jons O. and Nielsen B. Analyst 1992 117 637. Peixoto C. M. Gushikem Y. and Baccan N. Analyst 1992 117 1029. Sperling M. Yin X. and Welz B. Analyst 1992 117 629. Furtmann K. and Seifert D. Fresenius’ J. Anal. Chem. 1990 338 73. Milacic R. Stupar J. Kozuh N. and Korosin J. Analyst 1992 117 125. Soares M. E. Bastos M. L. and Ferreira M. A. J. Anal. At. Spectrom. 1993 8 655. Decreto-Lei No. 5711985 Diario da Reptiblica No. 54 I Serie 6 March 1985. Paper 4/00216D Received January 14 1944 Accepted June 24 1944

ISSN:0267-9477    

DOI:10.1039/JA9940901269

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1273 Packed Glassy Carbon Tube Atomizer for Direct Determinations by Atomic Absorption Spectrometry Free from Background Absorption* Kuniyuki Kitagawa,t Masahide Ohta Tetsuya Kaneko and Shin Tsuge Department of Applied Chemistry School of Engineering Nagoya University Furo-cho Chikusa-ku Nagoya 464-01 Japan An electrothermal atomizer consisting of a long glassy carbon tube packed with activated charcoal was evaluated for the direct determination of trace elements in biological materials by atomic absorption spectrometry in the absence of any background absorption. The maximum temperature of the atomizer of 2280°C was maintained continuously by an electric power supply of about 7 kW. Under such isothermal conditions few chemical interferences and high accuracy (relative error < 8%) and precision (RSO < 3.5%) were obtained in the direct determination of trace elements (Cu Mg and Mn) in National Institute of Standards and Technology biological Standard Reference Materials by solid sampling atomic absorption spectrometry. Keywords Electrothermal atomizer; solid sampling; biological material; atomic absorption spectrometry; background absorption Several electrothermal atomizers with the ability to separate analytes have been reported.'-'' Separation of atomic vapours had been reported in earlier studies on super high temperature gas chromatography by Sokolov and c o - ~ o r k e r s .~ ~ - ' ~ As in gas chromatography in general separative atomizers can be categorized into two types open tubular at~mizersl-~ and packed column atomizer^.^-'^ The former type has been applied to the separation of atomic vapours which would reduce inter-element effects.Conversely the latter type has been mainly applied to the removal or separation of non- atomic species from the atomic vapours which overcomes the problems associated with background absorption in the direct analysis of biological samples using electrothermal atomization. An open tubular molybdenum atomizer operating at a high temperature of up to 1800°C has been applied to the sequential separation of metal vapours in the analysis of National Institute of Standards and Technology (NIST) biological Standard Reference Materials ( SRMS).~ However a charring stage is indispensable when using this type of atomizer in order to remove the organic matrices.Therefore the charring temperature and time must be optimized. This is also the case for direct analyses using conventional electro- thermal atomizers. On the other hand separative column atomizers (the packed do not require the inclusion of a charring stage. However they have a limitation as to the temperature attainable which is below about 1400 "C since an alumina tube is used for the column. More recently a molybdenum tube atomizer packed with tungsten powder has been applied to the separation of atomic vapours at tempera- tures of up to 1820°C.'5 In the present study a glassy carbon tube (Tokai Carbon GC30) was employed as the heating material so as to obtain higher temperatures above 2000 "C.Glassy carbon is less permeable than graphite and does not form carbides. For this reason a glassy carbon tube has been used to determine the diffusion coefficients of metal vapours." From the direct analysis of NIST biological SRMs the accuracy of the proposed procedure was evaluated in conjunction with a study of the capability of the glassy carbon tube atomizer packed with activated charcoal to remove background absorption. * Presented in part at the 54th Symposium of the Japanese Society t Present address Research Center for Advanced Energy for Analytical Chemistry Mito Ibaraki Japan ( 1993). Conversion Nagoya University Nagoya Japan. Experiment a1 Apparatus A schematic diagram of the glassy carbon tube atomizer developed for the present study is shown in Fig.1. A glassy carbon tube (Tokai Carbon GC30) of 6.5 mm i.d. 8.6 mm 0.d. and 20 cm in length (1) is supported vertically by ring (4) and cone ( 5 ) graphite electrodes which are fixed at upper and lower stainless-steel discs respectively. The glassy carbon tube Fig. 1 Schematic diagram of the glassy carbon tube atomizer 1 Tokai Carbon GC30 glassy carbon tube (20 cm x 6.5 mm i.d. x 8.6 mm 0.d.); 2 0.67 g of activated charcoal (30-60 mesh); 3 graphite stopper 1.5 mm in thickness with 10-20 holes of 0.8 mm in diameter; 4 graphite ring cut into two halves; 5 graphite cone; 6 water-cooled stainless- steel chamber 64 mm id.; 7 graphite sample cup (30 p1 inner volume); 8 tungsten wire of 0.8 mm diameter; 9 sample inlet; 10 stainless-steel plunger of 2 mm diameter; 11 and 12 water inlets; 13 and 14 argon inlets; 15 AAS signal observation holes of 2 mm diameter; 16 spring; 17 quartz window; 18 to monochromator; 19 to variable transformer; and 20 mica plate1274 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 is mounted in a cylindrical atomization chamber of stainless steel the wall of which (6) is water cooled (12).The glassy carbon tube is packed with 0.67g of activated charcoal of 30-60 mesh (2) the packing length being 45 mm. The bottom end of the packing is plugged with a graphite disc of 1.5 mm thickness (3) through which 10-20 holes with diameters 0.8 mm are drilled in order for the vapours eluting from the packing to pass through. A quarter of the way up from the bottom end of the glassy carbon tube two holes of diameter 1.5 mm are drilled to allow passage of the light beam in order for atomic absorption to take place.Argon purge gas flows outside the glassy carbon tube at a flow rate of 1 1 rnin-l. Argon carrier gas is forced to pass through the packing layer at a flow rate of 0.1-0.4 1 min-'. The glassy carbon tube is heated continuously by passage of a large alternating current of 220 A at a voltage of 32.5 V. A 30 pl graphite sample cup is suspended by a tungsten wire of 0.8 mm diameter (8) which is fitted to a stainless-steel plunger of 2 mm diameter (10). In order to avoid overheating the top of the glassy carbon tube and the bottom stainless- steel disc are water A conventional atomic absorption spectrometer was used for the atomic absorption spectrometric (AAS) measurements.A deuterium lamp was also used for monitoring background absorption. The radiation from the hollow cathode and deu- terium lamps were merged into a beam using a grid mirror. Phase rectification was employed for discrimination between the AAS and non-specific signals. The resulting intensity measurements were converted into absorbance by a microcom- puter programmed in a C-language (MSX C Version 2.1). The temperature of the glassy carbon surface was measured around the holes used for observation of the AAS signals by using a digital optical pyrometer. The instruments used for this work are listed in Table 1. Procedure Prior to measurement the glassy carbon tube atomizer was thermally equilibrated by passing current continuously for more than 15 min.After this has been done electrical control and switching for the drying and charring stages as is necessary for conventional electrothermal atomizers are not necessary. After removing the sample inlet screw (9) lop1 of a test solution were quickly added by a micropipette into the Table 1 Instrumentation Spectrometer Photomultiplier tube Power supply for lamps Electronics for signal processing Hollow cathode lamps Deuterium lamp Microcomputer Program Micropipette Shaker Optical pyrometer graphite cup suspended at the sample inlet position of the tube the screw was set the cup allowed to stand there for about 30 s in order to evaporate the solvent and then inserted into the hot region which is a few mm above the top of the packing layer by lowering the plunger.The drying process could be dispensed with for volumes of less than 10 pl. After 2500. 2410 2000 0 9 5 1500 2 !!? E r-" 1000 600 kd I t I I I I J 0.5 1 2 3 5 10 Electric power/kW Fig. 2 Dependence of the temperature (T) of the glassy carbon tube on the electric power (P) The temperature was measured around the AAS observation hole by a digital optical pyrometer O" 0 A Techtron AA4 Czerny-Turner mount Hamamatsu Photonics R106. 400-600 V f = 50 cm slit-widths 50-200 pm Laboratory-made current regulators 0 4 8 12 16 Time/s Fig. 3 Change in absorbance profile at 324.8 nm for Cu (100 ng) with carrier gas flow rate at STP (a) 130ml min-'; (b) 174 ml min-'; (c) 420 ml min-'; temperature of glassy carbon tube = 1985 "C Laboratory-made phase-sensitive rectifiers 3.6 I I Hitachi HLA-3 (Cu Mn Mg Cd and Zn) Hamamatsu Photonics L233 (Pb) current 2-8 mA Hamamatsu Photonics current d 12 mA Matsushita FS4600F fitted with a hard disk drive (Logitec SHD-40J) and a laboratory- made analogue-to-digital converter card (12 bits) designed by the electronics workshop by the Middle Branch of the Japan Society for Analytical Chemistry Version 1.2 (Japan ASCII) Laboratory prepared written in MSX-C Gilson digital micropipette Yamato MT-11 Minolta TR-630 2.1 x 2.0 2 a 1.5 2 1.0 in Y 2 I] a 1 0 100 200 300 400 500 Carrier gas flow rate/ml min-' Fig.4 Dependence of the absorbance peak height and area on carrier gas flow rate 0 peak area; 0 peak height. The conditions are the same as in Fig. 3JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1275 atomization the next run could be carried out after cooling the sample cup for about 2 min at the sample inlet port. A slurry method was used for sampling powders. An aliquot of NIST SRM 1571 Orchard Leaves or NIST SRM 1577 Bovine Liver was dispersed using a mechanical vibrator in 1 5 or 10 ml of a 0.1 moll-' HN03. The resulting suspension or slurry was taken into the graphite cup using the micropip- ette. In order to allow larger sample particles to be taken into the tip with good reproducibility the top of the micropipette tip was cut off to form a larger entrance bore of = 1.5 mm in diameter. The atomization procedure for the slurry samples was the same as that for the solution samples. Stock standard solutions of 1000 mg 1-l of Cu Mg and Mn were prepared by dissolving the pure metals in HN03 and those of Cd Pb and Zn were commercially available standard solutions (Wako Pure Chemicals). Test solutions were prepared by adjusting the acidity to 0.1 moll-' HNO just before use.All reagents used were of analytical-reagent grade. Results and Discussion Electric Power The electro-thermal performance of the glassy carbon tube atomizer developed was estimated by the following measure- ments. The dependence of the temperature on electric power is shown in Fig. 2. The slope of the rectilinear curve obtained on a logarithmic scale was found to be This suggests that radiative heat loss from the glassy carbon plays a major role in determining the temperature the slope should be approxi- mately % based on the Stefan-Boltzmann law if radiation alone regulates the temperature.The maximum temperature of 2280°C was obtained with a power of 7 kW the limit of the power supply. From Fig. 2 a temperature above 2400 "C which would be sufficiently high for atomization of most elements could be expected to be attained at 10 kW. In the present work the inner wall of the atomization chamber was not polished to a mirror finish however a temperature above 2400°C could expected to be obtained at a lower power after such a modification. 0.4 0.2 a 0 5 0 f! 2 0.4 0.2 Optimization of Carrier Gas Flow Rate The change in the profile of the atomic absorption peak for Cu uersus the carrier gas flow rate is demonstrated in Fig. 3 and the dependence of the peak height and area on the carrier gas flow rate in Fig.4. The peak absorbance becomes smaller at lower carrier gas flow rates (Fig. 4). This could be due to the interaction of the analyte vapour with the packing layer leading to a decrease in the recovery of the analyte atoms. With an increase in the carrier gas flow rate the peak absorbance increases and shows a roll-over effect. At the higher flow rates the residence time of the atoms becomes shorter leading to a decrease in the absorbance. Accordingly a carrier gas flow rate of around 170 ml min-' was considered to be optimum. However the optimum flow rate will change with the atomizer temperature and the type of analyte atoms. As seen in Fig. 4 the absorbance in terms of peak area also decreases in a fashion similar to that for the peak absorbance.This suggests that the optical thickness of the absorption path of 6.5 mm (15 in Fig. 1) is insufficient. Capability of Removing Background Absorption Typical absorbance peak profiles which were obtained with the glassy carbon tube atomizers in the presence and absence of the packing material heated at a relatively low temperature are shown in Fig. 5. The test sample was a solution of Cd to which a rotary vacuum pump oil and a liquid synthetic detergent (Natera Lion Oils and Fats) had been added as a matrix to cause background absorption and as a surfactant (and which also causes background absorption) respectively. Strong background absorption is observed for the open tubular atomizer [Fig. 5(c)]. On the other hand background absorp- tion is drastically decreased for the packed atomizer [Fig.5 ( d ) ] . Non-atomic species such as carbonaceous particulate matter resulting from the organic matrices are trapped as charcoal on the surface of the packing layer and any other eluting species are non-absorbing at the analytical wavelength. In addition the atomic absorption peak obtained with the packed atomizer [Fig. 5(b)] becomes greater than that obtained with the open tubular atomizer [Fig. 5(a)] the net area for the latter which is estimated by subtracting the 0 2 4 6 8 0 2 4 6 8 Ti me/s Fig. 5 Typical absorbance peak profiles (A=228.8 nm) with [(a) and (c)] open and [(b) and (d)] packed glassy carbon tube atomizers measured with (a) and (b) a hollow cathode lamp (atomic + background absorption); (c) and ( d ) a deuterium lamp (background absorption) Sample is 3 pl of 100 pg ml-' Cd solution + 5 pl of vacuum pump oil + 2 pl of synthetic detergent (Natera Lion Oils and Fats); carrier gas flow rate 240 ml min-' STP; temperature of glassy carbon tube = 1233 "C1276 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 Time/s Fig. 6 Typical absorbance peak profiles obtained through repeated runs (a)-@) hollow cathode lamp; ( d ) - ( f ) deuterium lamp. The sample and conditions are the same as those in Fig. 5. The profiles were run in alphabetical order. Integrated absorbance 0.47 s for (a); 0.51 s for (b); 0.46 s for (c); 0.23 s for (d); 0.16 s for (e) and 0.09 s for (f) 2 4 6 8 0 2 4 6 8 l-ime/s Fig. 7 Typical absorbance peak profiles (1=217.5 nm) for Pb with [(a) and (c)] open and [(b) and (d)] packed glassy carbon tube atomizers measured with (a) and (b) a hollow cathode lamp (atomic + background absorption); (c) and ( d ) a deuterium lamp (background absorption).Sample is 3 ~1 of aqueous 20 pg ml-' Pb solution+ 5 pl of vacuum pump oil +2 pl of synthetic detergent (Natera Lion Oils and Fats); carrier gas flow rate 487 ml min-'; temperature of glossy carbon tube= 1534°C background absorption becomes smaller. This effect arises mainly because the degree of atomization becomes higher after the non-atomic species containing the analyte element comes into contact with the heated graphite packing which has a large surface area. An interesting phenomenon was observed for the open tubular glassy carbon tube atomizer.Typical peak profiles obtained through repeated measurements are shown in Fig. 6. The background absorption gradually decreases in magnitude run by run while the atomic absorption remains almost the same (see the values for integrated absorbances in the figure caption). This suggests that the inner wall of the glassy carbon tube gradually became coated with carbon resulting from the pyrolysis of the organic matrices. The coating layer of carbon acts as a packing layer decomposing non-atomic species such as carbonaceous particulate matter into carbon. This is ana- logous to wall-coated open tubular columns used for gas chromatography. Typical peak profiles of atomic and background absorption signals for a sample of Pb mixed with oil are shown in Fig.7. These were obtained with open tubular and packed glassy carbon tube atomizers heated to a moderately high temperature. The results are similar to those for Cd the background absorp- tion is removed by the packed atomizer. Robinson and co-workers have also pointed out the decrease in the back- ground absorption in the direct atomization of lead sulfate samples using a packed graphite a t ~ m i z e r . l ~ * ~ ~ However loss of atoms through leakage would be a potential problem in their atomizer since graphite is highly permeable. Glassy carbon is preferable for the atomizer material as it is much less porous. Peak profiles of typical atomic and background absorptions for Cu are shown in Fig. 8. These were obtained by direct 0.3 ( a ) 0.1 O.* I I 0 20 30 50 Time/s Fig.8 Typical absorbance peak profiles (3 = 324.7) for Cu on direct atomization of 0.25 mg NIST SRM 1577 Bovine Liver containing 100 ng of Cu with a packed glassy carbon tube atomizer with (a) a hollow cathode lamp; (b) a deuterium lamp; carrier gas flow rate 670 ml min-'; temperature of glassy carbon tube= 1750 "C atomization of NIST SRM 1577 Bovine Liver with the packed glassy carbon tube atomizer heated to 178OoC a higher temperature than that used for Pb. No background absorption exists [Fig. 8 ( b ) ] . In Fig. 9 are shown typical peak profiles for Mg. No background absorption occurs in the direct atomization of the NIST SRM 1571 Orchard Leaves with the packed glassy carbon tube atomizer [Fig. 9(e)]. Similar peak profiles wereJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1277 Fig. 9 Typical absorbance peak profiles (A=285.2 nm) for Mg on direct atomization of (a) and (d) aqueous solution (50 ng of Mg); (b) and (e) 8 pg of NIST SRM 1571 Orchard Leaves containing 49 ng of Mg using the packed glassy carbon tube atomizer; (c) and (f) 8 ng of NIST SRM 1571 Orchard Leaves containing 49 ng of mg using the open glassy carbon tube atomizer; (a)-(c) hollow cathode lamp; (d)-(f) deuterium lamp; carrier-gas flow rate 725 ml min-'; temperature of glassy carbon tube=2160 "C Table 2 Results for the NIST Standard Reference Materials Sample Element SRM 1571 Orchard Leaves* c u Mg Mn SRM 1577 Bovine Livert c u c u With standard Standard solution/ addition/ Certificate value/ CLgg-' Pg g-' Pg g-' 11.1 12.6 12+ 1 87.0 98.5 91 +9 138$ 193 f 10 150 0.64 0.59 0.62 & 0.02 - - * T > 2000 "C and RSD for measurement points < 3.5% n 2 3.t T= 1820°C and RSD for measurement points > 15% n 2 3. $ Peak height measurements (all other values from peak area measurements). observed for the test solution and the powder sample [Fig. 9(a) and (b)] although a slight distortion can be seen for the latter. Fig. 9(f) shows that the open tubular glassy carbon tube atomizer also has some ability to remove the background absorption as described for Cd. Peak broadening occurs in the packed atomizer [compare Fig. 9(b) and (c)]. This is mainly due to interaction of the analyte atoms with the packing layer. Thus from the point of view of peak broadening the open tubular glassy carbon atomizer would be preferable.Because of the smaller surface area for interaction however a longer tube length would be required for the open tubular column to remove the back- ground absorption when large amounts organic sample matrices are loaded. Direct Determinations of Cu Mg and Mn Analytical curves were prepared for standard solutions and for the NIST reference materials using the standard additions technique with the packed glassy carbon tube atomizer. For Cu a slight chemical interference was indicated by the differ- ence in the slope between the two types of calibration curves. High precision in terms of relative standard deviation (RSD) was obtained less than 3.5% RSD at most of the measurement points. The results are listed in Table 2.At a lower atomizer temperature of 1820"C the relative error exceeds 22% and the RSD is greater than 15% n 2 3 for the direct determination of Cu in NIST SRM 1577 Bovine Liver. The recovery could be low owing to retention on the packing layer. At a temperature above 2000"C much better precision and accuracy were obtained for the direct determination of Cu Mg and Mn in the NIST Orchard Leaves the RSD being <3S% and the relative error <8%. The packed glassy carbon tube atomizer when operated at a high temperature of above 2000°C proved to have the capability of removing background absorption leading to high precision and accuracy in the direct determinations of trace elements in powder samples. The sensitivities for 1 % peak height absorption were 3.3 ng for Cu 1.0 ng for Mn and 0.53 ng for Mg.Further improvement is necessary to obtain sufficient sensitivity for determinations at the pg kg-' level. The lifetime of the glassy carbon tube was longer than 200 h when used at temperatures of above 2000°C. The authors thank S. Takahashi and K. Tachibana for manufacturing the complicated atomization chamber and T. Watanabe and T. Imura for skillful glass-blowing. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 References Ohta K. Smith B. W. and Winefordner J. D. Spectrochim. Acta Part B 1982 37 343. Ohta K. Smith B. W. and Winefordner J. D. Anal. Chem. 1982 54 320. Ohta K. Smith B. W. and Winefordner J. D. Microchem. J. 1985 32 50. Ohta K. Yamanaka N. Inui S. Winefordner J. D. and Mizuno T. Talanta 1992 39 1643. Yanagisawa M. Kitagawa K. and Tsuge S. Spectrochim. Acta Part B 1982 37 493. Yanagisawa M. Suzuki H. Kitagawa K. and Tsuge S. Spectrochim. Acta Part B 1983 38 1143. Kitagawa K. Takeuchi T. and Yanagisawa M. Anal. Sci. 1989 5 445. Kitagawa K. Mizutani A. and Yanagisawa M. Anal. Sci. 1989 5 539. Yanagisawa M. Ida K. and Kitagawa K. Anal. Sci. 1989,5,765. Yanagisawa M. Katoh K. and Kitagawa K. Anal. Sci. 1990 6 471. Kitagawa K. and Takeuchi T. Anal. Chirn. Acta 1973 67 457. Sokolov D. N. Vakin N. A. and Kalmykov Yu. D. Zauod Lab. 1969 35 158. Sokolov D. N. and Vakin N. A. Zauod Lab. 1970 36 1314. Sokolov D. N. J. Chromatogr. 1970 47 320. Ohta K. Yamanaka N. and Inui S. Analyst 1993 118 1031. Robinson J. W. Slevin P. J. Hindman G. D. and Wolcott D. K. Anal. Chim. Acta 1972 61 431. Robinson J. W. Rhodes L. and Wolcott D. K. Anal. Chirn. Acta 1975 78 474. Paper 4102092H Received April 8 1994 Accepted June 17 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901273

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1279 Speciation of Inorganic Mercury(i1) and Methylmercury by Vesicle-mediated High-performance Liquid Chromatography Coupled to Cold Vapour Atomic Absorption Spectrometry 6. Aizpun Maria Luisa Fernandez E. Blanco and Alfred0 Sanz-Medel* Department of Physical and Analytical Chemistry University of Oviedo Julian Claveria 8 33006-Oviedo Spain A novel high-performance liquid chromatography (HPLC) separation coupled to cold vapour atomic absorption spectrometry (CVAAS) detection for the speciation of inorganic mercury(i1) and methylmercury is described. The mercury species can be successfully separated within 8 min by using a vesicular mobile phase of didodecyldimethylammonium bromide (DDAB) in acetate buffer containing 5% v/v of modifier methyl cyanide (acetonitrile) and 0.005% v/v of 2-mercaptoethanol.The stationary phase is a &-bonded silica column previously modified by the passing of a solution of DDAB. Detection limits of 0.1-0.2 pg I-' of mercury were achieved after off-line preconcentration of the aqueous samples using C Sep-pack cartridges modified with 2-mercaptoethanol solutions. This vesicle-enhanced HPLC-CVAAS approach has been applied to the speciation of inorganic mercury(i1) and methylmercury in spiked sea-water and human urine. Recoveries obtained ranged between 91 -1 03% both for the inorganic and organic mercury species. Keywords Vesicles; high-performance liquid chromatography; cold vapour atomic absorption spectrometry; mercury speciation and preconcentration ; sea-water and human urine It is now well known that the toxicological and biological effects of trace elements depend on their chemical forms in the sample.' In particular inorganic mercury is converted into the much more toxic methylmercury compound in the environment by a number of biological processes. The latter form is of particular concern because of its enhanced toxicity lipophilic- ity bioaccumulation and volatility compared with inorganic mercury. These facts assumed special importance following earlier pollution incidents such as that at Minamata Bay in Japan and were decisive in realizing that analytical measure- ments of the more toxic forms would be more meaningful than total element determination.2 A number of methods have been applied to the determination of mercury species in different matrices.The strategies most widely used involve gas chromatography (GC) separation with different types of detectors including the electron capture detector ( ECD);3 atomic absorption spectrometry (AAS);4 and microwave-induced plasma atomic emission spectrometry (MIP-AES).' More recently mercury speciation has been also carried out by coupling high-performance liquid chromatogra- phy (HPLC) with specific detectors such as AAS,6-1' induc- tively coupled plasma atomic emission spectrometry (ICP- AES)" and ICP-mass spectrometry ( ICP-MS).13,'4 Although the ICP-AES and ICP-MS detectors have unique analytical capabilities for speciation their high instrumental and running costs make them more difficult to be adopted widely as common chromatographic detectors.Conventional nebuliz- ation AAS is the most popular specific detection system,15 owing to its simplicity inexpensive instrumentation and ready availability. The sensitivity of this detector in its conventional use is however insufficient for the speciation of very low concentrations of mercury species in some real samples (e.g. natural water samples). Therefore high sensitivity sample introduction techniques [e.g. post-column derivatization of mercury species to form a cold vapour (CV)] and perhaps preliminary preconcentration have to be used to improve the analytical sensitivity of the HPLC-AAS Reversed-phase liquid chromatography using 2-mercapto- ethan~l'~~'~-'* (or other molecules with the -SH in an aqua-organic mobile phase have been proposed for separation.However the introduction of these mobile phases containing relatively high percentages of organic * To whom correspondence should be addressed. solvents into plasmas results usually in a decrease in sensi- tivity higher plasma background increased instability and even eventual extinction of the plasma.'' Therefore the use of alternative HPLC mobile phases which do not use organic solvents such as in micellar liquid chromatography (MLC) could be advantageous.20 In fact MLC has shown some important advantages including enhanced selectivity versa- tility rapid gradient elution capability low toxicity low cost and the ability to simultaneously chromatograph both hydro- philic and hydrophobic solutes.21 Unfortunately MLC also suffers from some drawbacks including loss of efficiency22 and solvent strength.23 Although MLC has been extensively studied for separations during the last years no attention has been paid to the use of mobile phase systems containing other surfactant-based organized assemblies such as vesicles.Very recently it has been established that HPLC separation followed by HG-ICP-AES detection can be a synergic combination uia the use of di- dodecyldimethylammonium bromide (DDAB) vesicles as mobile phases for the speciation of toxic arsenic species.24 In this paper the new strategy of vesicle-mediated HPLC separations is coupled to vesicle-enhanced CVAAS detection as it has been shown that CV generation of mercury can be improved in DDAB vesicles.z5 This vesicular HPLC-CVAAS technique is applied to mercury speciation in sea-water and human urine samples after adequate preconcentration in c18 cartridges modified with 2-mercaptoethanol.Experimental Apparatus A Knauer Model 6400 HPLC pump with an attached sample injection valve equipped with a 100mm3 loop were used for eluent delivery and sample introduction. The analytical column was a Spherisorb ODS 2 (250x4.6mm id.) packed with 10 pm C18-bonded silica stationary phase previously modified by passing DDAB solution as described below. A four channel peristaltic pump HP4 Minipuls 2 Gilson and a laboratory-made gas-liquid separatorz6 constituted the con- tinuous CV generator. A schematic diagram of the whole HPLC-CVAAS system is presented in Fig. 1. An ultrasonic device from Sonics and Materials Model VC (250 W) was used for DDAB vesicle preparation.1280 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 mobile phase Pump Hydrochloric acid (1%) Fig. 1 Schematic diagram of the coupled HPLC-CVAAS system for mercury speciation C,* column A Unicam Model PU 9400X atomic absorption spec- trometer equipped with a T shaped absorption quartz cell (8 mm i.d. 12 cm length) was used for absorption measure- ments of mercury vapour at 253.7 nm. Reagents Methylmercury stock solution (100 mg I-' of Hg) was obtained by dissolving the appropriate amount of methylmercury chlor- ide salt (Merck) in 10cm3 of acetone which was made up to 100 cm3 with ultrapure Milli-Q water. This stock solution was stored in a glass bottle at 4 "C.Inorganic mercury was obtained from Merck as a 1000 mg dm-3 Hg solution. Working standard solutions of both mercury compounds were freshly prepared daily by diluting the stock solutions with ultrapure Milli-Q water. The DDAB vesicular solution (lo-' mol dm-3) was pre- pared by dissolving 0.4626 g of DDAB (Fluka) in 100 cm3 of Milli-Q water and sonicating this solution (with a power output of 60 W) for 10 min. Sodium tetrahydroborate solution (1 YO m/v) was prepared by dissolving 1 g of NaBH ( Probus Barcelona) in 100 cm3 of 0.1 % m/v NaOH solution. Filtration of the solution through a Whatman grade 4 filter paper before use was carried out. This solution was stored at 4 "C and prepared weekly. Hydrochloric acid solution (1% v/v) was prepared from concentrated HCl (Merck) and Milli-Q water.The mercury- selective complexing agent 2-mercaptoethanol was obtained from Merck and used without further purification. The HPLC grade methanol and methyl cyanide (acetonitrile; Romile Chemicals) were used. All other chemicals were of analytical-reagent grade and distilled and de-ionized (Milli-Q system Millipore) water was used throughout. Procedures HPLC column modification The C bonded silica reversed-phase column was modified by passing a total of 500 cm3 of a DDAB (1 x mol dm-3) aqueous solution in 50% methanol at a flow rate of 1 cm3 min-'. Milli-Q water was then passed through the column for 30 min at the same flow rate. This modified column was kept in water when not in use. Mercury speciation Vesicular mobile phases were prepared by dissolving the appropriate amount of DDAB in water containing 0.005% v/v of 2-mercapthoetanol 5% v/v acetonitrile and buffered with ammonia acetate (0.01 mol dmP3) at the desired pH.The mobile phase was degassed by ultrasonicating for 30 min prior to use. This mobile phase was then continuously pumped through the analytical column at a flow rate of 1.5 cm3 min-' and 100mm3 of the working solutions of the mercury com- pounds were injected for analysis. The eluent at the exit of the HPLC column was first mixed with a 1% v/v HC1 solution and then mixed with the 1% m/v NaBH solution for the mercury CV generation which was continuously swept through the gas-liquid separator to the T quartz cell of the atomic absorption spectrometer by a stream of argon (250 cm3 min-I) see Fig.1. All separations were performed at room temperature under isocratic conditions. Each separation was attempted under several different combinations of organic modifier DDAB vesicle concentrations pH etc. The best chromatographic resolution of the various set of conditions tested was obtained using the separation conditions summarized in Table 1. Optimum experimental conditions finally selected for CV generation and AAS detection after preliminary investigations are also summarized in Table 1. Peak heights from the chroma- tograms were used in all mercury quantifications. Sample collection and pre-treatment Sea-water samples were collected in a Spanish coastal region (Gijon) of the Cantabric Sea and immediately acidified (by the addition of ultrapure nitric acid to have a final pH of 2) for storage in pre-cleaned polypropylene bottles.The samples were filtered through a Millipore 0.45 pm membrane. The pH was adjusted to 7 with diluted ammonia before the speciation analysis. Human urine samples were filtered through a Millipore 0.45 pm membrane. Preconcentration step Preconcentration of mercury compounds from water and urine samples was carried out using Sep-pack C18 [trichloro(octa- decyl)silane chemically bonded to Porasil A] cartridges modi- fied with 2-mercaptoethanol solution as described below. Cartridges were activated by washing with 7cm3 of meth- anol which was subsequently displaced with 7 cm3 of ultrapure water. Then the cartridge was modified by passing through it 10 cm3 of 0.5% v/v 2-mercaptoethanol aqueous solution.Volumes of 100cm3 of the samples (standard solutions sea- water and human urine) were then pumped through the modified cartridge by using a Gilson Minipuls 2 peristaltic pump at a flow rate of 10 cm3 min-'. The mercury species Table 1 HPLC-CVAAS Experimental conditions for the mercury speciation by Chromatography Column Temperature Sample volume Mobile phase Flow rate NaBH cv HCI Ar carrier Wavelength Lamp current Slit-width AAS C,,-bonded silica 10 pm particle size 250 x 4.6 mm i.d. (modified with 1 x rnol dmT3 DDAB) Room temperature 100 mm3 10 mmol dm-3 ammonium acetate buffer + 2-mercaptoethanol+ 1.5 cm3 min-l 5% acetonitrile +0.005°/a 2 x mol dm-3 vesicle of DDAB pH=5 1% m/v NaBH (in 0.1% NaOH) flow rate 1% v/v flow rate 1 cm3 min-' Flow rate 250 cm3 min-' lcm3 min-' 253.7 nm 6 mA 0.5 nmJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1281 loaded in the cartridge were eluted with 10 cm3 of acetonitrile. These acetonitrile solutions were rotary evaporated and the residue was dissolved in 1 cm3 of the mobile phase ( x 100 preconcentration factor) and 100 mm3 of this solution were injected for HPLC determinations. Results and Discussion Chromatographic Separation The CI8 column was previously modified with DDAB as detailed above. The DDAB coating formed proved to be very stable and resistant to water and to the mobile phase passing Although organic solutions e.g. methanol could remove the coating the modified column was very durable in the recommended vesicular operation.In fact no significant column behaviour indicating any sort of degradation was observed after several months of daily usage. In order to investigate the effect of the pH of the mobile phase the retention times of both solutes under study were measured in the pH range 4-612,16 using a mobile phase consisting of 0.002 mol dm-3 ammonium acetate 5% v/v acetonitrile 0.005% v/v 2-mercaptoethanol and 2 x mol dm-3 DDAB vesicles. As expected when the pH of the mobile phase was varied from 4 to 6 no significant change of retention time of the neutral mercaptoethanol- mercury complexes12*'6 was observed. A mobile phase of pH 5 was then chosen for further studies. After fixing the pH the effect of increasing buffer concen- trations in the mobile phase on retention times of solutes was investigated.The results obtained (Fig. 2) indicate that reten- tion times tend to decrease as the concentration of ammonium acetate increased up to an ammonium acetate concentration above of 0.01 mol dm-3. At higher buffer concentrations the retention time change was very small Therefore a concen- tration of 0.01 mol dm-3 was chosen. This concentration is considerably lower than that generally used (0.06 mol dm-3) in reversed-phase HPLC with conventional hydro-organic mobile phases.12 The use of such high salt concentrations generated in our system produced irreproducibilities in the measured retention times and very long column equilibration times. The effect of increasing the acetonitrile concentration in the vesicular mobile phase on the separation of the two mercury compounds investigated was also studied.The results were compared with those obtained in the absence of DDAB (using a typical reversed-phase system with an unmodified ClS column). The comparative results observed are given in Table 2 which shows that for both HPLC systems investigated (conven- tional and vesicular) the overall peak broadening and retention 1 I I 1 I 0 0.005 0.01 0.015 0.02 0.025 [Ammonium acetatel/mol I ' Fig.2 Effect of buffer concentration on retention times A Inorganic mercury and B methylmercury. Mobile phase 0.005% 2-mercaptoethanol 5% acetonitrile 2 x lop4 DDAB vesicles in ammonium acetate buffer (pH=5); flow rate 1 cm3 min-'. All other experimental conditions as described in Table 1 times of inorganic mercury and methylmercury were sensitive to the percentage of acetonitrile added to the mobile phase.As expected using conventional reversed-phase HPLC the retention times were reduced by increasing the amount of acetoni trile present. However high percentages of ace toni trile greater than 25% were needed in order to obtain acceptable resolution values for inorganic and organic mercury. As shown in Table 2 acetonitrile addition also decreased the retention times in the vesicular HPLC system but in this case good separations were achieved even in the absence of the organic modifier. This finding can be of great importance for further coupling of the exit of the column to conventional nebulization ICP-AES or ICP-MS detect01-s.~~ In addition when the same amount of organic modifier was used in both systems (see the results for 0-5% acetonitrile in Table 2) lower retention times were obtained for the vesicular chromatography.It should be realized that the surface of the C stationary phase used had been previously modified by the surfactant molecules from the DDAB solution. Therefore a partially coated polar stationary phase will result which allows for not only classical hydrophobic interactions but also for electrostatic interactions with ionizable solutes. These will distribute between the modified stationary phase and the charged vesicles into the mobile phase.23 Considering the nature of the solutes for separation (neutral complexes of mercury with 2-mercaptoethanol) only hydrophobic inter- actions should occur. Therefore the observed decrease in the retention times in the presence of acetonitrile and DDAB could be rationalized by the fact that hydrophobic attractions of the neutral solutes by the surfactant-modified stationary phase are weaker than those observed with the conventional unmodified C18 stationary phase.The dependence of the solute retention time on the surfactant concentration in the mobile phase was evaluated. To do so the influence of using or not using sonication of the surfactant mobile phase (presence and absence of mono-dispersed DDAB vesicles in the mobile phase respectively) as a previous step to the chromatographic separation was investigated. Results obtained in these experiments have been plotted in Fig. 3 and show that the retention time decrease observed with surfactant addition only takes place if mono-dispersed DDAB vesicles are present in the mobile phase (ie.with previous sonication of the surfactant solutions). This effect of vesicles in the separation is stronger for the solute with stronger retention time (inorganic mercury which will form mercury 2- mercaptoethanol neutral species and will be the last species to elute (see Fig. 4). Finally shorter retention times were obtained with increas- ing mobile phase flow rates (resulting in greater peak height) and a 1.5 cm3 min-' flow rate was selected as optimum for the speciation analysis. Fig. 4 illustrates that a mobile phase consisting of 2 x mol dm-3 DDAB vesicles buffered at pH 5 with ammonium acetate (1 mmol dm-3) containing 0.005% v/v mercaptoethanol 5% acetonitrile and delivered at 1.5 cm3 min-l allowed a most adequate isocratic separation of methylmercury and inorganic mercury in an aqueous mixture.Analytical Characteristics of the Method In order to evaluate the precision of this novel chromato- graphic speciation method eight injections (100 mm3) of a standard mixture containing known amounts of the two mer- cury compounds concentration (200 pg dm-3 in the element) were made. The relative standard deviation (RSD%) of the peak height results calculated for inorganic mercury were always worse than those for methylmercury at 20ng of the metal level (Table 3). The observed detection limits (as calcu- lated for a signal-to-noise ratio of three) for an injected volume of 100 mm3 are also included in Table 3 and show values of 10-20 yg dmP3 rather insufficient for the determination of1282 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 I J Table 2 Effect of the percentage of organic modifier in the mobile phase on the retention times (t,) and resolution (Rs) 2 min ~~~ Acetoni trile (%I 25 20 15 10 5 2.5 0 Conventional reversed phase HPLC* tr (Hg)/min 5.30 5.35 5.83 7.50 14.17 14.47 > 30 t (MeHg)/min 6.25 6.33 6.80 7.50 12.45 12.97 > 30 Vesicular HPLCT - Rs tr ( Hg )/min tr (MeHg)/min Rs 0.05 0.84 0.30 0.00 0. so 10.50 7.33 1.50 0.60 12.97 8.43 1.98 -_ 17.40 11.32 1.95 * Column (250 x 4.6 mm i.d.) 10 pm bonded silica; mobile phase 0..005% 2-mercaptoethanol in 0.06 mol dmP3 ammonium acetate buffer t Column (250 x 4.6 mm i.d.) 10 pm c18 bonded silica (modified with lop3 mol dmP3 DDAB); mobile phase 0.005% 2-mercaptoethanol mol dm-3 vesicles of DDAB in 0.01 mol dmP3 ammonium acetate buffer (pH 5) flow rate 1 cm3 min-'. Other experimental conditions (pH 5) flow rate 1 cm3 min-'.Other experimental conditions as described in Table 1. 2 x as described in Table 1. usual levels of methylmercury and inorganic mercury in natural water samples (e.g. sea-water where preconcentration should then be used). Calibration graphs for each of the mercury species were worked out and turned out to be linear over several orders of magnitude ranging from the detection limit to 400 pg dm-3 (maximum concentration tested). I ~ 0 0.5 1 .o 1.5 2.0 2.5 [DDA81/10-'mol I-' Fig. 3 Effect of DDAB concentration in the mobile phase on retention times DDAB vesicles (solid line) and DDAB without sonication (broken line).A Inorganic mercury and B methylmercury. Experimental conditions are given in Table 1 2 Time - Fig. 4 Typical chromatogram of a standard mixture containing 100 pg dm-3 of each mercury specie 1 methylmercury and 2 inorganic mercury. Experimental conditions as described in Table 1 Preconcentration of Mercury Species It is well known that the sulfydryl groups (-SH) exhibit a relatively high affinity for mercury. This fact prompted us to investigate the possibility of increasing the detectability of mercury species in our vesicle-mediated HPLC separation by on-line complexation preconcentration. In this vein Sep-pak cartridges packed with CI8 were modified (impregnated) with the chelating agent 2-mercaptoethanol as described under Procedures.Known amounts of inorganic mercury and methylmercury were added to ultrapure water and preconcentrated on a Sep- pak C cartridge modified by passing through an aqueous solution of 5% v/v of mercaptoethanol. This modified cartridge proved to be adequate for preconcentration of up to 400 ng of both types of solutes (recovery 91 & 5%) and was used for the off-line column extraction-preconcentration of the sought species in sea-water and urine samples. For such samples at the pH around 7 the two species of interest were extracted quantitatively in the modified cartridge and so this pH was used here for further work. Samples (100 cm3) were used in the test for preconcentration in the cartridge with a loading rate of 10 cm3 min-'.The loaded mercury species were then recovered quantitatively from the cartridge for injection in the HPLC column as detailed already under Procedures (Preconcentration step). Analysis of Real Samples The complete analytical procedure (preconcentration and determination by vesicular HPLC-CVAAS) described was applied to the speciation of the two mercury compounds in sea-water. To do that aliquots of acidified sea-water (100 cm3) samples were spiked with concentrations between 0-4 pg dm-3 of inorganic and methyl mercury. The spiked samples were Table 3 Figures of merit of the vesicle-mediated HPLC-CVAAS method (without preconcentration) Mercury species DL/pg dm - RSD YO)^ Linear range/pg dm-3 t,/min Methylmercury 10 3.5 10-400 4.83 Inorganic mercury 16 6.9 16-400 7.15 * Detection limit calculated as 3 times the baseline noise.t Relative standard deviation 8 replicates of each mercury species injected.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1283 Table 4 Recoveries of mercury species added to sea-water (100 cm’) DL for methylmercury with 100 1 preconcentration 0.10 pg dm-3; and DL for mercury with 100 1 preconcentration 0.16 pg dm-3 Methylmercury Inorganic mercury Absolute amount Recovery Concentration Absolute amount Absolute amount Recovery added/pg dm - added/ng recovered/ng* (Yo) recovered/ng* (Yo) - 20 - 0 0 0 2 200 177 89 184 92 3 300 272 91 27 1 90 4 400 362 91 359 91 * Hg found in the preconcentrated solution (1 cm3). Table 5 Recoveries of mercury species added to human urine (100 cm3) Methylmercury Concentration Absolute amount Absolute amount added/mg dm-3 added/pg recovered/pg 0 0 0 0.100 10 10.3 0.300 30 28.6 0.003 0.300 0.310* 0.004 0.400 0.405* Inorganic mercury Recovery Absolute amount Recovery (Yo) recovered/pg (%) - 103 95 103* 101* - 11.4 27.0 0.272* 0.371* - 114 90 91* 93 * * After preconcentration procedure. then filtered through a 0.45 pm filter and their pH adjusted to 7 with diluted ammonia to be analysed immediately.Recoveries obtained of the added mercury spikes were always higher than 90% as shown in Table4. Very small amounts of inorganic mercury only (about 0.2 pg dm-3) were detected in the natural sea-water samples used in these recovery experiments. The blank values for each mercury species were estimated by parallel experiments applying the same preconcentration- speciation procedure to 100 cm3 of Milli-Q water Again only inorganic mercury was detected in the blank solution at concentrations of 0.4k0.2 pg dm-3 according to the con- tamination of the reagents (namely 2-mercaptoethanol used without purification).Natural sea-water samples were treated as unknown samples and analysed for inorganic and methyl- mercury following the recommended procedure. The applicability of the proposed vesicle-mediated method for the mercury speciation in human urine was also evaluated. As real samples of the human urine available did not contain detectable mercury and methylmercury they were spiked with concentrations between 0.1-0.3 mg dm-3 of the two mercury compounds. The spiked samples were filtered and analysed for both species without preconcentration.Recovery values for different levels of the two species are shown in Table 5. Parallel experiments were performed with samples of 100 cm3 with low levels of spiked samples which were preconcentrated and analysed following the procedure described above. Recoveries obtained ranged between 91-103% (see Table 5). In brief the recommended strategy offers a reliable analytical method to speciate toxic mercury at very low concentration levels in sea-water and urine which could be extended to other environmental and/or biological samples. Conclusions The coupling of a vesicle-mediated HPLC separation with CVAAS detection of mercury atoms has proved to be a novel rapid and efficient method for the speciation of methylmercury and inorganic mercury(11).These two species can be separated in less than 8 min with DDAB vesicular mobile phases and a DDAB previously modified c18 bonded silica stationary phase. The presence of vesicles in the mobile phase seems to play a role in the neutral mercury species finally formed for HPLC separation and this role could be similar to that described23 for micelles in MLC. This ‘vesicular’ chromatography makes it possible to carry out the separation of the two mercury compounds using very low percentages of organic modifier in the mobile phase as compared with those required in conven- tional reversed-phase chromatography. This means improved detection performance using specific (atomic) detectors particularly using plasmas.27 A new method for the off-line preconcentration of mercury species by solid-liquid extraction with sep-pack cartridges modified with 2-mercaptoethanol solutions is also proposed here.This technique exhibits much higher efficiencies for recovery of mercury compounds than other recent approaches.” The modified c18 material has shown to be an effective solid-phase complexing agent (extractant) which pro- vides virtually quantitative retention of both investigated mer- cury species from sea-water and human urine samples and thus it should be adequate for the on-line preconcentration HPLC-CVAAS speciation of mercury in environmental and biological samples. These vesicle-mediated strategies could be extended to the speciation of other toxic metals of environmental concern,2 particularly those’ using HPLC-plasma detection.28 1 2 3 4 5 6 7 8 9 10 11 12 13 References Robinson J.B. and Tuovinen 0. H. Microbiol. Rev. 1984,48,95. Environmental Analysis Using Chromatography Interfaced With Atomic Spectroscopy Eds. Harrison R. M. and Rapsomanikis S. Ellis Horwood Chicester 1989. Horvart M. Byrne A. R. and May K. Talanta 1990 37 207. Fischer R. Rapsomanikis S. and Andreae M. O. Anal. Chem. 1993 65 763. Bulska E. Emteborg H. Baxter D. C. Frech W. Ellingsen D. and Thomassen Y. Analyst 1992 117 657. Van Loon J. C. Anal. Chem. 1979 51 1139A. Munaf E. Haraguchi H. Ishii D. Takeuchi T. and Goto M. Anal. Chim. Acta 1990 235 399. Fujita M. and Takabatake E. Anal. Chem. 1983 55 454. Lupsina V. Horrat M. Jeran Z. and Stegnar P. Analyst 1992 117 673.Rencede M. C. R. Campos R. C. and Curtis A. J. J. Anal. At. Spectrom. 1993 8 247. Sarzanini C. Saccchero G. Aceto M. and Abollino A. Anal. Chim. Acta 1994 284 661. Krull I. S. Bushee D. S. Schleicher R. G. and Smith S . B. Jr. Analyst 1986 111 345. Huang C. and Jiang S. J. Anal. At. Spectrom. 1993 8 681.1284 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 14 15 16 17 18 19 20 21 22 23 Shum C. K. Pang H.-M. and Houk R. S. Anal. Chem. 1992 64 2444. Ebdon L. Hill S. and Ward R. W. Analyst 1987 112 1. Evans O. and McKee G. D. Analyst 1988 113 243. Holak W. Analyst 1982 107 1457. Hempel M. Hintelmann H. and Wilken R. D. Analyst 1992 117 669. Boor A. W. and Browner R. F. Anal. Chem. 1982 54 1402. Suyani H. Heitkemper D. Creed J. and Caruso J. Appl. Spectrosc. 1989 43 962. Dorsey J. G. De Echegaray M. T. and Landy J. S. Anal. Chem. 1983,55924. Dorsey J. G. Khaledi M. G. Landy S. L. and Lin J-L. J. Chromatogr. 1984 316 183. Ordered Media In Chemical Separations eds. Hinze W. L. and Armstrong D. W. American Chemical Society Washington DC 1987. 24 Liu Y. M. Fernandez Sanchez M. L. Blanco Gonzalez E. and Sanz-Medel A. J. Anal. At. Spectrom. 1993 8 815. 25 Sanz-Medel A. Fernandez M. R. ValdCs-Hevia M. C. Aizpun B. and Liu Y . M. Talanta 1993 40 1759. 26 Menendez Garcia A. Sanchez Uria E. and Sanz-Medel A. J. Anal At. Spectrom. 1989 4 581. 27 Aizpun B. Fernandez M. R. and Sanz-Medel A. J. Anal. At. Spectrom. 1993 8 1097. 28 Element Specific Chromatographic Detection by Atomic Emission Spectroscopy Uden P. C. American Chemical Society Washington DC 992. Paper 4/01 724B Received March 23 1994 Accepted June 10 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901279

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1285 Determination of Trace Impurities in High-purity Graphite by Electrothermal Atomic Absorption Spectrometry and Inductively Coupled Plasma Atomic Emission Spectrometry* 2. Hladky and M. Figera Department of Analytical Chemistry Faculty of Chemical Technology Slovak Technical University Radlinskeho 9 812 37 Bratislava Slovakia The process for the determination of B and Si in high-purity graphite used for spectral analysis elaborated here is based on matrix combustion without any loss of analyte in an oxygen atmosphere with the addition of alkali. Sub-boiling distilled nitric acid and water were used to dissolve ashes. Silicon B Cd and Cu were determined by inductively coupled plasma atomic emission spectrometry and the other elements (Cr Co Mo V Ni and Ti) were determined by electrothermal atomic absorption spectrometry.Carbon is the most commonly used material in the production of electrodes for emission spectral analysis as well as being the material of electrothermal atomizers used in atomic absorption analysis. The level of impurity should not be greater than 10-5-10-6%. The purifying processes used in the production of carbon electrodes for spec- troscopy are adequate for obtaining these values for most elements with the exception of B Si and those elements whose carbides are extremely non-reactive. Precision given as the relative standard deviation for both Si and B at the ppm level was within 1O0/o. The accuracy of the whole procedure was controlled using spiked samples.Keywords Silicon and boron determination ; high-purity graphite; inductively coupled plasma; electrothermal atomic absorption spectrometry The one component common to electrothermal atomizers in atomic absorption spectrometry (ETAAS) and electrodes used in atomic emission spectrometry (AES) are graphite tubes bars or differently formed electrodes. The analytical performance of a measuring system is very dependent on the shape and properties of the material from which it is constructed. Some of the most important properties required mainly by the tube material of an electrothermal atomizer have been described Graphite most closely satisfies these requirements and there- fore is the most widely used material. However even electro- graphite carbon does have some limitations the most important of which are its porosity and that some elements can form highly refractive carbides or nitrides on its ~urface.~ Therefore the use of a suitable method with low limits of determination of impurities is necessary and is described in this paper.Experimental Chemicals and Solvents Standard solutions of B Si Cd Cu V Ni Cr Co Mo and Ti with concentrations of 1 mg ml-' were used for the preparation of the calibration solutions and the standard additions for spiked samples. All solutions were prepared using beakers made of poly- (tetrafluoroethylene) (PTFE) and micropipettes with tips made of polyethylene. Solutions were kept in these beakers that had been conditioned for a few days with purified de-ionized water and with the blank solutions. High-purity water was produced from de-ionized water with the Barnstead Nano-Pur System ( Wilhelm Werner Germany) and secondarily purified by distillation with apparatus made completely of PTFE (Berghof Germany).All measurements were carried out in a clean laboratory. Processing of the Samples The determination of B Si and the other elements was performed on samples prepared by the following methods.6 *Presented at the XXVIII Colloquium Spectroscopicum Internationale (CSI) York UK June 29-July 7 1993. Method A Graphite (5-10 g) was combusted in an oxygen atmosphere in the apparatus shown in Fig. 1 for 2 h. Sub-boiling distilled nitric acid and water were used to dissolve the residual ashes and to bring the volume up to 10m1 respectively. Method B Matrix combustion was carried out as in Method A in an oxygen atmosphere without any loss of analytes with the addition of alkali occurring in the solid phase (graphite sintered by 0.1 g of solid NaHCO,).Method C This is similar to Method B except that the addition of alkali occurred in the liquid phase (graphite saturated by the solution Quaflz tube Oven maximum 600°C Sintered filter Pt boat cylinder (flow 2- Fig. 1 Apparatus used for combustion in oxygen atmosphere Table 1 Measurement conditions for ICP-AES Optics Grating Resolution Wavelength range Power/W Nebulizer Observation height/mm Argon consumption/l min- ' Uptake rate/ml min-' Computer Czerny-Turner arrangement (f= 1 m) Holographic 2400 grooves mm-' 13 pm (first order) 165-800 nm 1200 glass Meinhard 15 14 (on the whole) 1.6 (without peristaltic pump) PDP-l1/03 (built into the spectrometer) language ARLEB1286 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 Table 2 Measurement conditions for ETAAS ~~~ ~ ~ Element Wavelength/nm Bandwidth/nm Lamp current/mA TD/t,*/”C S - l TA/tAt/°C S - ’ Cr 357.9 0.5 8.0 1200/25$ 260015 c o 240.7 0.2 12.0 1250/25$ 250015 Mo 313.3 0.2 12.0 1400/20 2800/3 v 318.5 0.2 12.0 1450120 280013 Ni 232.0 0.2 9.0 11 50/25$ 260015 Ti 365.4 0.2 13.0 1500/20 280013 * TD and t temperature and time of decomposition respectively. t TA and t A temperature and time of atomization respectively. 10.1% solution of Mg(N0,)2 used as chemical modifier. of 0.1 g of NaHCO in 0.5 ml H20 and dried under IR radiation). Samples prepared using each of the methods (A B and C) were spiked by standard additions of B Si and Cd before com bus tion.The efficiency of this procedure was tested on synthetic samples composed from boron nitride and carbide and silicon carbide as the most stable compounds of these elements. Instrumentation The analytical measurements were carried out by inductively coupled plasma atomic emission spectrometry (ICP-AES) using a sequential spectrometer ICP 3510 (ARL USA) under the conditions summarized in the Table 1 or by ETAAS using a Pye Unicam (UK) SP9 atomic absorption spectrometer fitted with a PU 9090 Data Graphic System and PU 9095 Video Furnace Programmer using the standard operating conditions summarized in Table 2. Results and Discussion Because the level of impurities in graphite should not be higher than to (level required for production of high- purity graphite suitable for use in the production of graphite tubes for ETAAS) we could make the choice of the spectral line with satisfactory low detection limits.7 The program SCAN provided by the ICP-manufacturer can be used to scan around a chosen spectral line.The lowest quantity determinable (LQD equal to the limit of determination in solution) and background equivalent con- centration (BEC) values for suitable lines were determined by the standard programme limit of detection (provided by the manufacturer) using a solution with a content of 10 pg m1-l. Results are summarized in Table 3. As can be seen from Table 3 the best lines for B Si and Cd determination seem to be 249.678 251.611 and 228.802 nm Table 3 Analytical characteristics for some suitable lines Element B B B Si Si Si Cd Cd Cd c u c u linelnm 182.640 208.959 249.678 212.452 251.61 1 288.158 214.438 226.502 228.802 324.757 224.700 LQD */pg ml - ’ 0.1000 0.0200 0.0125 0.1000 0.0500 0.1000 0.01 60 0.01 50 0.0200 0.0056 0.0200 BEC/pg ml - ’ 2.000 0.400 0.250 2.000 1 .ooo 2.000 0.348 0.300 0.121 0.150 0.400 ~~ * The lowest quantity determinable (k the concentration which yields a signal differing from that of the blank by 100 where cr is the standard deviation of the blank intensity) values were obtained with synthctic solutions of elements and represents only ‘theoretical’ values and will differ from that obtained using real sample solution with a known content of elements.4 ~ 1 0 ~ I 1 o4 104 1 o4 0 2 49.631 249.669 249.706 Wavelengthtnm Fig. 2 Spectra obtained around B line 249.678 nm for 0 sample; 0 ultrapure water; 0 blank; and A standard 10 ppm of B. Integration time 1 s and scan range k0.039 nm respectively which were also free of possible spectral coinci- dence in the sample analysis. Figs. 2 3 and 4 show the results of the spectral scans around these lines for the samples ultrapure water (sub-boiling distilled water from PTFE appar- atus) blank (composed of all reagents used in the determi- nation) and 10 ppm standards. The scan for Cu at 324.754 nm is not shown because it was similar to that for Cd i.e. free of spectral coincidence and the signal was lower than the limit of detection. Calibration graphs were constructed for the concentration range 0-10 mg 1-l (B Si Cd and Cu) for the ICP-AES method.1 o4 104 1 o4 1 o4 0 251.564 251.602 251.689 Wavelengthh m Fig.3 Spectra obtained around S line 251.611 nm for 0 sample; 0 ultrapure water; 0 blank; and A standard 10 ppm of Si. Integration time 1s and scan range k0.039 nmJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1287 7.5 x lo4 6 x lo4 - v) C 3 Y .- 2 4.5 x lo4 F 4? .e 3 1 0 4 c ._ - > C al c - 1.5 x lo4 0 228.775 228.793 228.830 Wave le ngt hlnm Fig. 4 Spectra obtained around Cd line 228.802 nm for 0 sample; 0 ultrapure water; 0 blank; and A standard 10 ppm of Cd Table 4 Results for the determination of some elements in high-purity graphite calculated on the mass of the solid material; '<' indicates that the measured value was lower than limit of determination calculated on the mass of sample Element B Si Cd c u Cr c o Ni Mo V Ti ICP-AES/ng g-' 332.3 -t 7.1 25444.3 & 127.2 < 20.0 0.7 < 6.0 & 0.4 ETAAS/ng g- - 4.35 & 0.54 12.69k 1.08 45.51 k4.41 63.59k4.51 374.05 k 55.73 823.00 k 27.98 Table5 Recovery of standard additions of B Si and Cd from the spiked samples by ICP-AES (samples were spiked with standard solutions of B Si and Cd whose volumes and concentrations were recalculated on the mass of samples so that concentration added was 10.00 pg ml-' in the final solution) Similarly the calibration graphs for determinations by ETAAS were prepared for the concentration ranges of O-lOpgl-' (Cr Co and Ni) O-lOOpgl-l (Mo and V) and 0-1000 pg I-' (Ti). The signals were measured as the peak heights.Table 4 gives the results of analyses of samples for a Slovakian producer of graphite used for spectral analysis. The samples were prepared by the different methods A B and C and results obtained by ICP-AES from these samples and from the spiked samples were compared with one another (Table 5). In Table 4 are summarized the results obtained using method C. This method of preparation of samples was used because it gave the best recoveries 97% on average. Conclusion The proposed method for the determination of trace impurities in high-purity graphite for spectral use has been applied for the determination of B Si Cd Cu Cr Co Mo V Ni and Ti in spectral electrographite. The accuracy and efficiency of this method for the treatment of the samples were tested on spiked samples with recoveries from 74 to 97% depending on the preparation methods used and with the use of boron nitride and carbide and silicon carbide as the most inert compounds of B and S .Precision as the relative standard deviations (RSD) for Si and B were within & 10% at the ppm level and & 15% at the ppb level for the other elements calculated from three indepen- dent determinations from different portions of the samples. Methods B and C have been shown to be suitable for the determination of impurities in high-purity graphite. References Products for Analysis Technology Ringsdorff Werke GmbH Bonn FRG E 555/89e. Lersmacher B. and Knippenberg W. F. Philips Tech. Rev. 1977 37 189. L'vov B. V. Spectrochim. Acta 1961 17 761. Dymott T. C. Wassall M. P. and Whiteside P. J. Analyst 1985 110 467. Luguera M. Madrid Y. and Camara C. J. Anal. At. Spectrom. 1991 6 669. De Keyser W. L. and Cypres R. Bull. Centre Phys. Nucleare Univ. Bruxelles 1952 35 28. Winge R. K. Fassel V. A. Peterson V. J. and Floyd M. A. Inductively Coupled Plasma Atomic Emission Spectroscopy An Atlas of Spectral Information Elsevier Amsterdam 1985. Method of Recovery of Recovery of Recovery of Average recovery preparation B/pg ml Si/pg ml - ' Cd/pg ml - ' w.) A 7.65 k0.28 6.78 k 0.61 7.81 0.36 74.1 B 8.96k0.33 7.72k0.56 8.15k0.25 82.8 C 9.48 0.25 9.13 k 0.48 10.51 0.56 97.1 Paper 31039358 Received July 7 1993 Accepted June 6 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901285

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1289 Determination of Trace Amounts of Thallium in Nickel-based Alloys by Electrothermal Atomic Absorption Spectrometry Rajananda Saraswati* N. R. Desikan and T. H. Rao Defense Metallurgical Research Laboratory Kanchanbagh P. O. Hyderabad-500 258 lndia A method is described for determining trace amounts of thallium in nickel-based alloys by electrothermal atomic absorption spectrometry (ETAAS). The alloy was dissolved by two methods one a microwave digestion procedure and the other a hot-plate dissolution method. Different acids were used for each of the two different methods. For microwave digestion a combination of H2S04 HNO and HF acid was used whereas for hot-plate dissolution a combination of H,SO HF and H202 was used.Ascorbic acid (C6H&@6) was used as chemical modifier. Comparable results for TI are achieved by the two methods of dissolution. The installation of a stabilized temperature platform in the graphite furnace tube improves the absorbance signal and the precision of the measurement. The detection limit was 0.1 ng ml-' of TI ( 3 ~ ~ ) . The accuracy of the method was elucidated through the analysis of certified nickel-based alloys. Keywords Thallium; nickel-based alloy; atomic absorption spectrometry Nickel-based heat-resisting alloys are used extensively in the aerospace industry for the construction of turbine blades. The presence of certain trace elements in nickel-based alloys can have serious effects on the mechanical and physical properties.Thallium is one of the deleterious trace elements that reduces the workability elongation and creep-rupture life of nickel- based alloys when present even at trace levels.'-3 The determi- nation of trace elements including T1 in these alloys has been a challenge because of the complexity of the matrices. Many methods are available for the determination of trace amounts of T1; electrothermal atomic absorption spectrometry (ETAAS) is among the most widely Sample preparation is one of the important steps in the determination of T1 by ETAAS. Serious signal depression can be caused by the presence of HCI' or HC104.' Matrix interference is another difficulty that has to be overcome in the accurate determination of T1. The use of H2S04' has been suggested for the elimination of HCl interference; use of pyrolytic graphite coated graphite tubes could reduce the interference due to HC104.10 Chemical modi- fication with organic or inorganic compounds can be used for the suppression of matrix The objective of this work was to develop a microwave dissolution method for nickel-based alloys for the determi- nation of T1 and to study the effect of chemical modifiers on the accurate determination of T1 in certified reference materials.This work also proves the merits of using a stabilized tempera- ture platform furnace to determine the trace amounts of T1. Experimental Instrumentation A GBC Model 902 atomic absorption spectrometer equipped with a GBC Model GF 2000 graphite furnace was used. A Visimax I1 hollow cathode lamp (HCL) was used as a T1 light source and a deuterium lamp was used for the background correction.Pyrolytic graphite coated graphite tubes and pyro- lytic graphite platforms supplied by GBC Instruments Australia were used. High purity argon gas was used as the purge gas with a flow rate of 50mlmin-'. Solutions were injected into the graphite furnace by a GBC PAL 2000 auto- sampling system. Integrated absorbance (peak area) values were used in measurements which were recorded with an Epson Lx-800 printer. Microwave-digestion was carried out in a micro-wave oven * Present Address Inorganic Analytical Research Division National Institute of Standards and Technology Gaithersburg Maryland 20899 USA. Toshiba Model E855 BTC that operated between 72 and 720 W with 81 W increments. The Teflon perfluoroalkoxy (PFA) vessels used for the dissolution had a volume of 200 ml and tight-fitting screw-cap lids.Reagents High-purity Suprapure grade chemicals from Merck were used. The lab-ware was used only after soaking in 10% HNO for several hours and then equilibriating in Milli Q water (18 MR resistance; Millipore). A stock standard solution containing 1000 pg ml-' of T1 was reported by dissolving specpure thallium(1) nitrate or sulfate in nitric or sulfuric acid. Working T1 standard solutions were prepared every day by serial dilution of the stock stan- dard solution. Digestion Procedure Wet digest ion This procedure is similar to that described e1~ewhere.l~ Two samples from each of standard reference material (SRM) 897 SRM 898 and SRM 899 [Trace Alloys; National Institute of Standards and Technology (NIST) Gaithersburg MD USA] and certified reference material (CRM) 345 and CRM 346 [IN 100 Alloy; Bureau of Analysed Standards (BAS)] together with two blanks were prepared.An accurately weighed (1 g) sample of nickel-based alloy was placed into a platinum crucible and moistened with distilled water and 7.0 ml of concentrated HF and 7.0 ml of concentrated H2S04 were added. The crucible was covered with a platinum lid and heated gently for 3 h on a hot-plate till 90% of the sample was dissolved. Then 2 ml of 30% H202 were added and heated again gently for 1 h to dissolve the sample completely. The crucible was allowed to cool to room temperature and 5 ml of 5% m/v ascorbic acid were added. The contents of the crucible were carefully transferred into a 100 ml polypropylene Cali- brated flask and diluted up to the mark with de-ionized water.Microwave-oven digestion procedure Two samples from each of SRM 897 898 899 and CRM 345 346 together with two blanks (one blank for each set) were prepared. Samples of nickel alloy (1 g) were accurately weighed into the PFA microwave digestion vessels. To each vessel were added 15 ml of 10% v/v HNO 5 ml of concentrated H2S04 and 1 ml of concentrated HF. The vessels were capped tightly and were placed (six at a time) in the microwave-oven. The 1 h digestion programme was set with 40% power for 10 min,1290 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 60% power for 30 min and 40% power for 20 min. Following the digestion programme the samples were left for 15 min in the microwave oven and then removed from the microwave oven and placed in a tray of ice water for about 30 min.Then the vessels were carefully uncapped their contents transferred into 100 ml polypropylene calibrated flasks 5 ml of 5% m/v ascorbic acid were added and the solution was diluted to volume with distilled water. Instrumental Analysis Procedure The above alloy solutions were analysed by injecting 20-50 pl portions into a pyrolytic graphite coated graphite tube using the optimized conditions for drying ashing and atomizing steps as shown in Table 1. The corresponding blank solutions were also analysed. The calibration graphs were obtained by injecting different amounts of T1 standard solution (Fig.1). The solutions were also analysed by injecting onto the platform located in the graphite tube using the instrumental conditions detailed in Table 1. Blanks were run regularly and their values were subtracted to obtain the net absorbance values. Results and Discussion Atomization of T1 suffers seriously from the severe interference of matrix elements and acids in ETAAS. Chemical modifiers Table 1 in Ni-based alloys Optimum analytical conditions for thallium determination Instrument mode Beam mode Wavelength/nm Slit-width/nm Lamp current/mA Integration time/s Sampling mode Furnace Furnace step temperature/ No. "C 1 80 2 110 3 400 4 500 (550)* 5 1800 (2200)* 6 2500 Ramp time/ 5 10 15 15 2 5 S Absorbance BC on Double beam 276.8 0.5 (0.7)* 10.0 (8.0)* 3.0 Auto sampling Hold time/ Gas flow/ s mlmin-' 5 50 10 50 15 50 15 0 2 0 3 50 Read on No No No No Y No * Platform.0.30 0.25 0.20 a C $ 0.15 0 Q .i? 0.10 0 5 10 15 20 25 30 Concentrationhg mi-' Fig. 1 platform; B without platform Calibration graph obtained for T1 standard solutions; A with enhance the volatility of unwanted elements of the matrix and stabilize the analyte during the pyrolysis stage. Sulfuric acid and palladium are frequently used as effective chemical modi- fiers for the determination of T1. In the present study the solutions were prepared in an acid combingtion that included H2S04 to avoid the interference which results from HC1 or HC104 as decomposition acids. The interference of the alloy matrix could be removed without any separational technique by using ascorbic acid as a modifier.Studies were conducted to optimize the ascorbic acid concentration. It was found that 0.25 g of ascorbic acid is sufficient to suppress the matrix effect of 1 g of nickel alloy sample (in 100ml). In the presence of H2S04 T1 forms the stable oxide instead of the volatile halide. It was observed that 6-8% v/v H2S04 is optimum for the best absorbance signal. The effect of vigorous heating at high temperatures during the digestion step was monitored. It was observed that excess heating at higher temperature during the dissolution process resulted in evaporation of T1; therefore all precautions were taken to avoid this. Ashing temperature atomizing temperature and time all play a very important role in the absorbance of T1 in nickel- based alloys.The effect of varying ashing temperature on the absorbance of T1 was studied keeping the atomizing tempera- ture constant. Fig. 2 shows the influence of ashing temperature on the integrated absorbance of T1 in nickel-based alloys. The maximum absorbance signal of TI was observed at an ashing temperature of 500°C. The absorbance of T1 decreased at lower and higher ashing temperatures. This may be explained on the basis that at low pyrolysis temperatures T1 is co-volatilized with the matrix resulting in a reduction in T1 residence time and therefore atomic absorption of thallium decreases. At 500"C the reduction or decomposition of these compounds will become maximum and formation of T1 atoms will be higher; accordingly the absorbance of T1 is increased. However at higher ashing temperatures T1 compounds will evaporate as molecules and the result is a decrease in atomic ab~orpti0n.l~ The effect of varying the atomizing temperature while keep- ing the ashing temperature constant was also investigated. Fig.3 shows the influence of atomizing temperature on the absorbance of T1 in nickel-based alloys. An atomizing tempera- ture of 1800 "C (without platform) and 2200 "C (with platform) produced maximum absorbance for T1. The effect of installation of a L'vov platform which was designed to fit in the graphite furnace tube was investigated. It was observed that by installing the platform the absorbance of thallium was enhanced by almost a factor of three (Fig. 3 ) . 0.20 0.16 a EJ 0.12 e d a 0.08 0.04 0 A I I I 200 400 600 800 1000 Te m pe ra t u rePC Fig.2 Absorbance of TI as a function of ashing temperature at an atomizing temperature of 1800 "C; A with platform; B without platformJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1291 8 g 0.12 e s a 0.08 a 0.04 0 1400 1600 1800 2000 2200 2400 TemperaturePC Fig.3 Absorbance of T1 as a function of atomizing temperature at an ashing temperature of 500 "C; A with platform; B without platform Table 2 Absorbance and RSD values of the calibration standards; n= 10 Concentration of Mean absorbance standard/ng ml- (platform value) 5 0.014 (0.051) 10 0.030 (0.106) 20 0.06 1 (0.196) 30 0.089 (0.288) Table 3 Results of T1 determinations in Ni-based alloys Concentration Concentration Sample found/pg 8-l certified/pg g-' SRM 897 0.50 & 0.02 0.51 & 0.03 SRM 898 2.70 k 0.05 2.75 & 0.02 SRM 899 0.25 & 0.01 0.252 & 0.003 CRM 345 1.90 & 0.05 < 0.2 CRM 346 1.98 k 0.04 (2)* * Information value.This is attributed to the platform which delays volatilization until the gas phase is at a higher unchanging temperature. This enhancement in the absorbance of T1 would help to enable the determination of T1 in complex matrices such as nickel-based alloys at low levels rather than with atomization from the tube wall. The precision was also improved by the installation of the platform in the furnace tube. The relative standard deviation (RSD) was about 2-10% without platform and 0.2-3% after installation of the platform. The mean absorbance values and RSD values were obtained from 10 repetitive measurements of the absorbance of T1 in the same sample solution.Measurement Uncertainty The calibration curve prepared in the range 0-30ngml-I of T1 was linear (Table 2). The detection limit defined as the T1 concentration corresponding to three times the standard devi- ation of the blank is 1.0 ng m1-l. The accuracy of this method was tested by determining TI in SRM 897 898 and 898 and CRM 345 and 346. The comparison of results obtained by this method (Table 3) shows good agreement with the certified values and no evidence of bias. Microwave digestion of nickel-based alloys with H,SO HN03 and HF in a pressurized PFA vessel provides an alternative to the conventional hot-plate wet digestion pro- cedure for the determination of T1. In the presence of H2S04 and ascorbic acid matrix interference can be suppressed to the required levels.It is evident from the absorbance values obtained that the installation of a stabilized temperature platform in the graphite furnace would help in extending the analysis to further lower levels of T1. The authors thank the Director Defense Metallurgical Research Laboratory for giving permission to communicate this work. References 1 2 3 4 5 6 7 8 9 10 11 12 13 Holt T. D. and Wallace W. Int. Met. Rev. 1976 21 1. Heardridge J. B. and Nicholson R. A. Spectrochim. Acta Part B 1984 39 551. Wood D. R. and Cook R. M. Metallurgica 1963 67 109. Griepink B. Sager M. and Tolg G. Pure Appl. Chem. 1988 60 1425. Marks Y. J. Welcher G. G. and Spellman R. J. Appl. Spectrosc. 1977 31 9. Kujirai O. Kobayashi T. Ide K. and Sudo E. Talanta 1983 30 9. Welcher G. G. Kriege 0. H. and Marks J. Y. Anal. Chem. 1974 46 1227. Koirytohann S. R. Glass E. D. and Lichte F. E. Appl. Spectrosc. 1981 35 22. Fuller C. W. Anal. Chim. Acta 1976 81 199. Slavin W. Carnrick G. R. and Manning D. C. Anal. Chim. Acta 1982 138 103. Shan X. Ni Z. and Zhang L. Talanta 1984 31 150. Manning D. C. and Slavin W. Spectrochim. Acta Part B 1988 43 1157. Kujirai O. Kobayashi T. and Sudo E. Fresenius' 2. Anal. Chem. 1979 297 398. Paper 4/02 794 I Received May I 1 I994 Accepted July 18 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901289

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1293 Determination of Aluminium in Infusion Solutions by Inductively Coupled Plasma Atomic Emission Spectrometry-a Critical Comparison of Different Emission Lines Sebastian Recknagel Ullrich Rosick and Peter Bratter* Hahn-Meitner lnstitut GmbH Department of Trace Elements in Health and Nutrition Glienicker Strasse 100 D- 74 7 09 Berlin Germany Seven emission lines for the determination of Al in infusion solutions by means of inductively coupled plasma atomic emission spectrometry were compared. For this comparison wavelength scans of different samples were made in order to study the background in the region of each line and to inspect for interference structures arising from molecular bands of water and argon. Background equivalent concentrations and detection limits were calculated for the emission lines at 167 237 394 and 396 nm.The detection limits were 1.2 pg I-' for the 167 nm line 8.8 pg I-' for the 237 nm line 10 pg I-' for the 394 nm line 4.7 pg I-' for the 396 nm line and > 10 pg I -' for the other three lines (226 308 and 309 nm). The A1 content of various infusion solutions was determined on different days using the three emission lines with the lowest detection limits (167 237 and 396 nm). Day-to-day-variation was approximately 5% in all cases. As a final step the Al contamination of 25 infusion solutions for parenteral nutrition was evaluated by using the two most suitable emission lines (1 67 and 396 nm). No statistically significant differences between the results were detected.Advantages and disadvantages of these two emission lines for the determination of Al are discussed. Keywords Inductively coupled plasma atomic emission spectrometry; aluminium determination; infusion solutions It is now well known that for patients with an elevated intravenous supply of A1 and/or a disturbed renal excretion of Al the body burden of the element cannot be disregarded as harmless. There are two main groups of patients in danger from the toxic effects of Al dialysis patients and patients undergoing long-term parenteral nutrition. Therefore it is of vital interest to monitor continuously A1 contamination of the infusion solutions and of the haemodialysis preparations as well as the serum samples of the patients concerned in order to check the actual body burden.The serum A1 level of healthy people today is below 3.5 pg 1-1 (median value 1.5 pg 1-l)l and within the range of the detection limit of inductively coupled plasma atomic emission spectrometry ( ICP-AES).2-7 To deter- mine concentrations of A1 of c 3.5 pg 1-l in serum samples electrothermal atomic absorption spectrometry (ETAAS) is the method of choice. In infusion solutions for total parenteral nutrition A1 concentrations of up to 12000 pg I-' have been found.* Higher concentrations of A1 in infusion solutions can be determined better by ICP-AES rather than by ETAAS because of the greater dynamic range of the former method. Most workers who have described the determination of A1 by ICP-AES in different materials used the 396nm emission line.2,4,9-13 On the other hand the determination of A1 has sometimes been carried out with the 394 line,3.9-11 the 2376*7714 or the 309 nm line.4 Matusiewicz and Barnes successfully used the 308 nm emission line to measure A1 in biological materials after electrothermal vaporization of the ~amp1es.l~ However for the direct determination of A1 in aqueous solutions this line is not appropriate because of the strong OH- inter- ference.16-19 The aim of the present work was to compare the different emission lines that have been proposed to find out which lines are best suited for the determination of A1 in infusion solutions. Experimental Apparatus For ICP measurements the monochromator of a Jobin Yvon 70 Plus spectrometer (Instruments S.A.Longjumeau France) * To whom correspondence should be addressed.with a 40.68 MHz r.f. generator was used. The resolution of the spectrometer was 0.009 nm as given by the manufacturer (grating 2400 lines mm- '). The monochromator system and the optical interface between the plasma and the entrance to the spectrometer was purged with nitrogen (spectrometer 4 1 min-'; and interface 0.6 1 min-'). The incident power of the r.f. generator was 1.2 kW the argon flow rates were 12 1 rnin-' (outer gas) 0.3 1 rnin-l (intermediate gas) and 0.7 1 min-' (aerosol carrier gas). The intermediate- and aerosol carrier gas flows were controlled by a mass flow controller (MKS Instruments Andover MA USA). For nebulization a concen- tric pneumatic nebulizer (Meinhard type) was used with a sample flow rate of 1 mlmin-1 (Gilson Minipuls 2 peristaltic pump).Instrument control and data aquisition were performed by a Siemens PCD-2M personal computer running with the ISA software J-YESS 4.01. This software package works through peak searching which was used throughout this investigation. In this mode 7-1 1 points around the theoretical line maximum were measured with an integration time of 0.5 s. The actual peak maximum was then calculated using a Gaussian fitting procedure. Reagents and Samples Calibration for the determination was based on a 1 g1-l A1 standard solution [Al(NO3) in 0.5 moll-' HN03 Merck Darmstadt Germany]. The samples to be measured were infusion solutions for parenteral nutrition from different pharmaceutical manufacturers. The determination of A1 is highly susceptible to contami- nation due to the ubiquity of A1 in the laboratory atmos- phere.20,21 Therefore preparation of the samples was carried out under clean-room conditions. The samples were diluted and acidified only there was no digestion of the fluids before measurement.For dilution vessels made of Teflon-tetrafluor- ethylene-perfluorethylene (FEP Nalge Rochester NY USA) were used. These vessels were pre-cleaned with nitric acid (2%) and distilled water. The samples were taken with pre-cleaned pipette tips and diluted with distilled water. Except for the NaHCO solution all samples were acidified (to about pH 1) with HNO (65% Suprapur Merck) before measurement.1294 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Emission Lines Investigated The six most sensitive emission lines for the determination of Al as given by Boumans,22 are the 396 394 309 308 237 and 226 nm lines.These lines and the emission line 167 nm described by Uehiro et a1.16 were investigated in the present study. As a first step 61 step scans with a step width of 4pm for a wavelength measured in the first optical order (A> 310 nm) and 2pm for wavelengths measured in the second optical order (A < 3 10 nm) were performed for different samples 1 argon; 2 distilled water; 3 500 pg 1-1 of A1 in water; and 4 a 1.25% serum protein solution with an A1 content of 25 p 1-l. A 61-step scan corresponded to a wavelength scan over 0.244nm in the first optical order and over 0.122nm in the second optical order. In a second step slopes of the analytical curves background equivalent concentrations (BEC) and detection limits (DL) were determined for the emission lines 396 394 237 and 167 nm on different days.Background measurement was carried out at the following positions 167 k0.0167 nm (for phosphate solutions; + 0.0557 nm instead of + 0.0167 nm); 237 _+ 0.0250 nm; 394 _+ 0.0670 nm; and 396f0.0535 nm. The third step was the determination of A1 in four infusion solutions calcium gluconate 10% (dilution factor 50) potassium lactate 1 mol 1-1 (20) the artificial colloid solution Gelifundol(10) and NaHCO 8.4% (lo) using the standard additions technique with two additions. The standard additions technique was used in order to avoid sample transport interferences. The samples were measured nine times on 5 d.The determination of A1 in these samples was performed at the emission lines 396 237 and 167 nm. For comparison 25 infusion preparations were analysed at the 396 and the 167nm lines. Comparisons of the determination by ICP-AES with ETAAS have been published else where.*^^^ Results and Discussion 61-Step Scans The results of the 61-step scans for the A1 emission lines investigated are shown in Fig. 1. It can be seen that the emission lines 308 and 309 nm are not sufficently sensitive for the determination of 25 pg 1-1 of A1 in a sample. No difference can be seen between the scans for water and for the 25 pg 1-1 infusion solution. In addition there is strong interference near the peak maxima of the two above mentioned emission lines.In the case of the 308 nm emission line Nygaard et a1.17-19 have described this interference as being due to a molecular band structure of OH-. It was possible to see this interference when measuring argon (without pumping) water and the infusion solution. The interference on the 309 nm emission line is also due to a molecular band structure of OH-. This interference has been mentioned by Winge et a1.24 The 226 nm emission line is also not particularly sensitive and it has a complex background with a weak interference on the peak maximum. This can be seen in Fig. l(b) in the scan of the water sample. Because of these disadvantages of the emission lines discussed above the following investigations were only carried out on the four remaining lines which showed no interference on the peak maxima.The matrix interference of organic sample components on the 167nm line will be dis- cussed later. Sensitivity Background Equivalent Concentrations and Detection Limits Determination of slope of the analytical curve (sensitivity) BEC and DL was achieved by measuring doubly distilled water and three standard solutions (50 200 and 500 pg I-' of Al). Each sample was measured five times apart from water which was evaluated ten times. Measurements were carried out on five successive days with two measurements a day (but on the last day only one measurement). The mean values [+ standard deviation (SD)] calculated from nine independent measurements are summarized in Table 1. As can be seen from the values the emission line at 167 nm had the lowest DL.It was lower than the DL of the 396 nm line by a factor of four. Detection limits of the two other lines were higher than that for 396nm by a factor of two. As expected the SDs of the DLs were approximately 30% for all emission lines. Background equivalent concentrations for the 237 and 396 nm lines were higher than that for 167 nm by a factor of 40 but half as large as that for the 394 nm line. When samples with a high Ca content (> 1000 mg 1-l) were measured BECs of the emission lines at 394.401 and 396.152 nm increased by a factor of five because of the strong Ca emission lines at 393.366 and 396.847 nm which interfere with the two A1 emission lines. The relative standard deviation (RSD) of the slope of the analytical curve is of greater interest than the value of the slope itself because the latter depends on the high voltage of the photomultiplier.The SD is a measure of the run-to-run stability of the measurement on the emission line being exam- ined. It is about 6% for the emission lines at 394 and 396 nm 11% for 237 nm and 26% for 167 nm. The reason for the higher value for the 167nm line is not known. A possible cause could be oxygen in the spectrometer which absorbs light in the far ultraviolet (UV) region (;1<200nm). This was excluded by measuring Ge-containing samples at the 164 nm line [(SD of the slope about 10% (data not published)]. Consequently it was necessary to recalibrate the spectrometer from time to time when measuring at the 167 nm line with the calibration technique. When working with the standard additions technique this problem does not arise as can be seen when the values for the residual error of the method for the above-mentioned emission lines are compared (see Table 1).In this case there were no significant differences between the values for the different emission lines. Within-run and Between-run Variation For four different infusion solutions nine independent determi- nations of A1 were made on five different days on the three different emission lines at 167 237 and 396 nm. The results of these measurements are summarized in Table 2. Statistical analysis with the Bartlett test showed no difference in the variances among the four emission lines (Bartletts x2= 3.3 for two degrees of freedom and therefore p < 0.05). No differences between the relative standard deviations (RSD%) could be Table 1 Comparison of performance of four prominent emission lines for the determination of A1 in pure water samples ~ ~~ Between-run variation* Within-run SD Wavelength/ Detection limit/ BEC/ Sensitivity/ Recovery variation of sensitivity/ of method/ (Yo) arbitrary units per pg 1 - ' I % I-' nm Pg I-' K 3 - l arbitrary units per pg 1-' 167.020 1.2k0.40 9.9 & 1.9 1.4fO.36 99.6 f 4.4 + 0.045 237.312 8.8 k 2.3 424 62 1.0 f 0.12 102.0 _+ 2.9 0.014 394.401 1Of 3.3 864 f 104 0.63 & 0.041 101.2f2.7 7 0.005 396.152 4.7k1.7 446k50 1.15 & 0.0'7 100.7 5 3.2 f 0.010 3.8 2.3 2.6 2.9 * For n = 9 independent measurement cycles.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1295 I / I 60 40 20 0 500 ( C) +- L3V .- 8 600 500 .- +- - 400 300 200 100 960 920 880 840 800 I I 5 15 25 500 ( b ) 1 450 400 350 300 250 600 (dl 100 ’ I I I I 1 660 640 620 35 45 55 Step number Fig. 1 A argon; B water; C sample 1 (1 +4); D 500 pg 1-’ of A1 (for the sake of clarity the individual samples have been shifted) Details of the 61-step scans of four samples at the A1 emission lines (a) 167; (b) 226; (c) 237; ( d ) 308; (e) 309; (f) 394.401; and (g) 396.152 nm.Table 2 Determination of A1 in infusion solutions of the A1 emission lines 167 237 and 396nm; in each case nine independent determi- nations were performed A1 content (+SD)/pg 1-’ Sample Dilution 167.020 nm 237.312 nm 396.152 nm Calcium gluconate 1 + 50 5087 f 249 5008 & 255 5139 f 272 Gelifundol l + 1 0 906f46 865f53 895f27 Potassium lactate 1 + 20 2350 f 101 2370 f 168 2310 f 122 Sodium hydrogen 1 + 10 572 f 22 584 f 28 604 & 32 carbonate shown for the samples.The mean value was 5.2%. Comparison of within-run and between-run variation for the A1 emission lines evaluated did not lead to any conclusion being drawn as to which emission line should be preferred for the determi- nation of A1 in biological samples. Comparison of the 167 and the 396 nm Lines The concentrations of A1 in the samples investigated by means of the two different wavelengths are summarized in Table 3. Agreement was found for most of the preparations. The values measured at the 167 nm line as compared with those measured at the 396 nm line are shown in Fig. 2. The statistical calcu- lation done by means of the reduced major axis method25 shows the equivalence of the results obtained at the two emission lines evaluated.Sample 9 1034 was excluded because the A1 content of this sample was much higher than all of the other values measured. Interferences The interference of Ca on the 396 nm line mentioned above is normally not a problem even when there are high concen- trations of Ca in the solutions being analysed provided that1296 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Table 3 A1 content (pg 1-l) of various infusion solutions for parenteral nuixition by means of ICP-AES using the emission lines 167 and 396 nm ~ ~~ ~ Sample No. 91009 91020 91021 91023 91027 91028 91029 91034 91038 91039 91041 91042 91043 9 1045 9 1046 91047 91048 91050 91051 91052 91055 91056 91059 91073 91075 Sample (manufacturer) Periplasmal 3.5% (Braun) Sodium hydrogen carbonate 8.4% (Braun) Potassium-L-malate 17.21 % (Braun) Calcium-Braun 10% (Braun) Inzolen PSA (Kohler) Inzolen PSI (Kohler) Inzolen Infantibus (Kohler) Sodium glycerophosphate (Pfrimmer) Gelifundol (Biotest Ph.) Serum protein solution 5% (DRK) Serumar 20 ml (Armour-Pharma) Serumar 250 ml (Amour-Pharma) Vitamin C 500 mg (Pharma Hameln) Calcitrans 10% (Fresenius) Potassium lactate (Fresenius) Calcium-Braun 10% (Braun) Calcium gluconate 10% (Ph.Hameln) Sodium chloride (Pfrimmer) Potassium phosphate (Pfrimmer) Potassium chloride (Pfrimmer) HAES-steril 10% (Fresenius) Lipofundin MCT 20% (Braun) Seltrans (Fresenius) Potassium phosphate (Braun) Sodium glycerophosphate (Pfrimmer) Dilution 1 :2 1:6 1 10 1 10 1:lO 1 10 1 10 1 10 1 4 1:4 1 4 1:3 1 4 1 10 1:lO 1 10 1 10 1:2 1:3 1:2 1:4 1:2 1:3 1:2 1 10 167.020 nm 22 999 1385 4955 2088 886 666 11998 865 61 197 66 986 3934 1940 6372 462 1 12 3758 15 86 39 77 77 1355 396.152 nm 37 92 1 1399 4957 2086 877 590 12143 892 75 235 76 958 3549 2071 6204 4426 12 3853 18 71 32 82 56 141 1 background correction is done on both sides of the A1 peak.However in another investigation problems arose with the determination of A1 in Ca-containing solutions at the 396 nm line. In this determination measurements were made five times at the emission lines 167 and 396 nm and approximately 10% more A1 was found with the 396 nm line than with the 167 nm line in the original calcium gluconate solution (lo% dilution 1 + 49).The reason for this over-estimation has not been clarified yet and needs to be evaluated in a future investigation. Uehiro et al. have described an interference of Fe on the 167 nm emission line.16 The interference factor given by these workers was 0.0018 g of A1 per g of Fe. In an earlier published investigation the present workers found a slightly lower inter- ference factor of 0.0013 g of A1 per g of Fe.23 There is another disturbance near the A1 peak at 167 nm which results from phosphates. This is shown in Fig. 3. The distance between the two peaks is 0.03 nm. Problems can arise when measuring phosphate-containing solutions and the back- ground is extrapolated from the position of the P peak (position A in Fig.3). No problems arose when the correct background measurement position was set to the right of the P peak (position B in Fig. 3). More important than the interference of P are molecular bands that appeared when solutions with a high C content were analysed. The intensity of these molecular bands is dependent not only on the C content of the solution but also on the structure of the organic compound. Alcohols e.g. ethanol and methanol show much higher molecular bands than other organic compounds such as acetic acid or amino acids.23 For example the molecular bands that result from ethanol at a concentration level of 1 moll-' of C in a solution containing 500 pg 1-' of A1 are shown in Fig. 4. Furthermore the intensity of these molecular bands also depends on the aerosol carrier gas flow rate.With a low flow rate (0.5 1-' min) the A1 to C signal ratio is much higher than with a higher flow rate (e.g. 0.8 1 min-'). These molecular bands can cause high values which are misleading when samples are analysed using the standard additions technique. A reduced aerosol carrier gas flow of 0.5 1 min-' was therefore used in the present study for the determination of A1 in infusion solutions with a high C content. In addition to the above-mentioned spectral interferences there are two other possible sources of interference (i) sample transport interferences and (ii) chemical interferences e.g. alkaline interferences because of ionization effects in the plasma. These were avoided by using the standard additions technique throughout this study.Conclusion For the determination of A1 in infusion solutions which are representative of a great variety of biological samples in the 5000 r Al content/pg I-' (12=167 nm) Fig.2 Comparison of methods for determining A1 in infusion solu- tions for parenteral nutrition using the emission lines at 167 and 396 nm 1 13 25 37 49 61 Step number Fig. 3 Interference of P near the A1 emission line at 167 nm solution contained 15.8gl-' of Pod3- and 1.25mg1-' of Al. See text for explanation of positions A and BJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 700 600 r 500 4- .- v) C Q) 4- .E 400 > .- t l - 2 300 200 100 - - - - - - - 0 I I 166.98 167.03 167.08 Wavelengt hlnm 1297 Fig. 4 Variation of aerosol carrier gas flow at the 167 nm emission line sample 500 pg 1-' of A1 in water containing ethanol (1 moll-' of C); A 0.5; B 0.6; C 0.7; and D 0.8 1 min-' gas flow composition of their matrices only two of the seven emission lines evaluated have detection limits below 5 pg 1-I.These two lines 167 and 396 nm are well suited for measuring A1 in biological samples within the existing guidelines for maximum permissible A1 concentrations in infusion solutions (10 pg 1-l). For the determination of A1 in serum of untreated patients ICP-AES is not sensitive enough because the concentration of A1 in serum of healthy people is below 3.5 pg I-' hence ETAAS has to be used. For the analysis of infusion solutions which according to the results presented could contain in some cases widely varying concentrations of Al the use of ICP-AES is more suitable than ETAAS.Compared with the latter ICP-AES has a wider dynamic range and therefore sample preparation is less susceptible to error (e.g. contami- nation) when samples to be analysed have to be highly diluted. Owing to the wide variations in the matrices of the samples analysed throughout this study the standard additions tech- nique was applied. The reason for this was to avoid systematic errors resulting from undetected non-spectral interferences. When series of samples have more uniform matrix composi- tions the less time-consuming calibration technique of using standard solutions can of course be used. Both of the recommended wavelengths have their advantages and disadvantages. The detection limit of the 167nm line is lower than that of the 396 nm line.On the other hand measuring at an emission line in the vacuum UV is more difficult because of possible problems such as air in the system or soiling of optical components of the spectrometer. When samples with a high Fe content or a high content of organic C are to be evaluated it is better to work at the 396 nm line because of the interferences on the 167 nm line. For determining A1 in samples with a high Ca content the 167nm line is more appropriate because of the strong Ca emission line near the A1 line at 396 nm. References 1 Gramm H.-J. Bratter P. Rosick U. Bohge P. and Recknagel S. Infusionsther. Transfusionsmed. in the press. 2 Schramel P. Wolf A. and Klose B. J. J. Clin. Chem. Clin. Biochem. 1980 18 591. 3 Lichte F.E. Hopper S. and Osborn T. W. Anal. Chem. 1980 52 120. 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Sanz-Medel A. Roza R. R. Alonso R. G. Vallina A. N. and Cannata J. J. Anal. At. Spectrom. 1987 2 177. Berner Y. N. Shuler T. R. Nielsen F. H. Flombaum C. Farkouth S. A. and Shike M. Am. J. Clin. Nutr. 1989 50 1079. Coni E. Bellomonte G. and Caroli S. J. Trace Elem. Electrolytes Health Dis. 1993 7 83. Violante N. Petrucci F. Delle Femmine P. and Caroli S. Microchem. J. 1992 46 199. Recknagel S. Bratter P. Chrissafidou A. Gramm H.-J. Kotwas J. and Rosick U. Infusionsther. Transfusionsmed. 1994 21 266. Allain P. and Mauras Y. Anal. Chem. 1979 51 2089. Narayaman P. Csanady G. Wegscheider W. and Knapp G. J. Anal. At. Spectrom. 1989 4 347. Leflon P. Plaquet R. Mornibre A. and Fournier A. Clin. Chim. Acta 1990 191 31. Brenner I. B. ICP Inf. Newsl. 1992 18 289. Arniaud D. ICP In5 Newsl. 1990 16 39. Coni E. Stacchini A. Caroli S. and Falconieri P. J. Anal. At. Spectrom. 1990 5 581. Matusiewicz H. and Barnes R. M. Spectrochim. Acta Part B 1984 39 891. Uehiro T. Morita M. and Fuwa K. Anal. Chem. 1984,56,2020. Nygaard D. D. Chase D. S. Leighty D. A. and Smith S. B. Anal. Chem. 1984 56 424. Nygaard D. D. Sotera J. J. Spectroscopy 1988 314 22. Nygaard D. D. Chase D. S. and Leighty D. A. Appl. Spectrosc. 1983 37 432. Rosick U. Recknagel S. Bratter P. in Mineralstofle und Spurenelemente in der Ernahrung des Menschen ed. Bratter P. and Gramm H.-J. Blackwell Wissenschaft Berlin 1991 p. 104. Ericson S. P. At. Spectrosc. 1992 13 208. Boumans P. W. Line Coincidence Tables for Inductively Coupled Plasma Atomic Emission Spectrometry Pergamon Press Oxford 1980 vol. 1. Recknagel S. Rosick U. Tomiak A. and Bratter P. in 6. Colloquium Atomspek-trometrische Spurenanalytik ed. Welz B. Perkin Elmer GmbH ifberlingen 1991 p. 841. Winge R. K. Peterson J. and Fassel V. A. Appl. Spectrosc. 1979 33 206 Feldmann U. Schneider B. Klinkers H. and Haeckel R. A. J. Clin. Chem. Clin. Biochem. 1981 19 121. Paper 41029731 Received May 18 1994 Accepted July 14 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901293

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1299 Quantitative Analysis of Electroniclgrade Anhydrous Hydrogen Chloride by Sealed Inductively Coupled Plasma Atomic Emission Spectroscopy* Tracey Jacksier Air Liquide Chicago Research Center 5230 S. East Avenue Countryside IL 605253787 USA Ramon M. Barnes Department of Chemistry Lederle Graduate Research Center University of Massachusetts Box 345 70 Amherst MA 01 003-45 10 USA Anhydrous hydrogen chloride at concentrations of up to 100% has been introduced into a sealed inductively coupled plasma system for quantitative spectrochemical analysis. Vapour-phase sampling of monobutyltin trichloride was developed to calibrate tin and carbon. The addition of chlorine as a modifier gas was required in a ratio of 1 :1 to maintain stable and reproducible emission signals.Under flowing conditions detection limits for tin and carbon were 49 and 271 (ng g - ' ) respectively in a 16% v/v HCI-CI argon plasma. Impurities identified qualitatively included Al C Ca Cr Fe Ni and Sn. In addition the bulk temperature of the plasma was determined to be 10500+1500 K. Keywords Sealed discharge; gas analysis; inductively coupled plasma; hydrogen chloride; emission spectroscopy A trend in very large scale integration (VLSI) technology requires ultra high purity chemicals in fabrication process steps. Chemical vapour deposition (CVD) and reactive ion etching (RIE) are particularly sensitive to impurities present in the process gases and impurities on the silicon surface from prior process steps. These impurities diffuse into the bulk semiconductor and lead to the formation of various types of defects.Therefore these impurities have an impact on product quality and reliabilit~.'-~ For example contaminants such as Fe Ni and Cu can generate defects that are responsible for yield losses caused by leakage current^.^ Anhydrous HCl is used to etch the silicon wafer surface prior to epitaxial crystal g r ~ w t h ~ and the trace element content is therefore of critical importance. Gases used in VLSI fabrication are specified with metal contamination in the lower ng g-' range. The production of ultra-pure chemicals therefore requires precise measurement of impurities at or below this level. The analysis of trace metal contamination in HCl includes various chemical preparation techniques direct analysis and preconcentration methods to improve sensitivity in the determination of trace impurities.For most metal determination techniques the metal impurit- ies must first be transferred into a liquid phase by bubbling HCl through aqueous solutions (hydrolysis) to preconcentrate the sample regardless of whether the impurities are distributed in the gaseous liquid or solid phase. The maximum concen- tration is determined by the solubility of HCl(g) in water (ix. solubility of HCl at 60 "C is 56.1 g per 100 g H20).6 The concentration of HCl dissolved in the aqueous solution can be determined by volumetric titration. Dilution of the HCl(1) is necessary before inductively coupled plasma (ICP) analysis owing to the change in physical properties of the sample solution and the relative sensitivity for the analyte (i.e.signal suppression). Additionally high concentrations of HCl may decrease the plasma excitation temperature resulting in loss of analyte ~ensitivity.~ Reducing the sensitivity loss by dilution can be overcome with evaporation of the sample. However many sample handling steps increase the risk of Contamination and elemental loss. The direct analysis of concentrated HCl using inductively coupled plasma atomic emission spectroscopy (ICP-AES) or * Presented in part at 1993 Pittsburg Conference and Exposition Atlanta GA USA March 8-12 1993 electrothermal atomic absorption spectroscopy (ETAAS) has been used to eliminate elemental loss during evaporation.8 However the direct analysis of HC1 by ICP-AES suffers from signal suppression of analytes such as As and P resulting in lower sensitivity.Lower analytical sensitivity is also observed in the analysis of concentrated acids with ETAAS. Additionally corrosion of parts of the analyser occurs in ETAAS. An analytical method was developed by Bridenne et aL8 in which metallic impurities were determined by ICP-AES after hydrolysis acid evaporation and reconstitution in aqua regia. Approximately 50 ng cm-3 were reported for Cr Cu Fe and Ni in HCI with an average relative standard deviation (RSD) of ~ 7 % . However the many sample handling steps necessary in this procedure increased the risk of sample contamination. An on-line extraction method was developed by Schramg to introduce gaseous HC1 directly into the ICP.The gaseous sample was passed into a mixing chamber positioned under the aerosol tube of the ICP torch. The sample gas flow was regulated by a peristaltic pump. Feeding the gas directly into the plasma reduced the risk of sample contamination since the gas reached the plasma directly without further chemical or physical proces~.~ Two major difficulties were identified; matrix effects and calibration. The direct calibration of impurities was impossible because many metals did not form sufficiently volatile compounds to matrix match standards. Total reflection X-ray fluorescence (TXRF) spectrometry has been used for the analysis of HF HC1 HNO H,SO and NH3.2 However this method was developed for aqueous acids and sample evaporation was required to remove the solvent.Despite efforts current state-of-the-art techniques are inad- equate to determine ultratrace metal impurities in gaseous HC1. At present HC1 gas purity is established indirectly from the quality of the epitaxy obtained.' Thus additional effort to improve methodologies and sensitivity is warranted. The sealed inductively coupled plasma (SICP) has been developed as an analytical spectrochemical ~ o u r c e . ~ ~ - ~ ~ It exhibits a number of advantages which suggest its suitability for application to the analysis of HCl. Low flow rates (< 70 cm3 min-l) result in limited sample consumption and dilution. No sample preparation is required because anhydrous HC1 is a gas at room temperature. Finally the enclosed source makes the safe introduction of HCl contamination free.The SICP analysis of HCl is presented in this paper.1300 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Qualitative detection of metal impurities in HCl as well as quantitative results for the determination of C and Sn are reported. Additionally the temperature of an HCl-Cl,-Ar discharge is estimated. Experiment a1 Instrumentation The instrumentation used was described previous1y.lo3l2 The gas handling system was modified (Fig. 1) to allow simul- taneous mixing of Cl anhydrous HCl and a standard into the Ar gas stream. Hydrogen chloride (Alphagaz Walnut Creek CA USA) was introduced through an electronic mass flow controller F4 (0-100cm3 min-l Model 1159B MKS Instruments Andover MA USA) C1 (Alphagaz) through F3 (0-100 cm3 min-l Model 1159B MKS Instruments) Ar through F1 (0-1000cm3 min-l Model 1159B MKS Instruments) and the standard through F6 (0-20 cm3 min-' Model 1159B MKS Instruments).To eliminate the possibility that the impurities were the result of particle generation by valves or the gas handling system a 0.01 pm in-line particle filter (Membralox Model 28235G Alcoa Separations Technology Pittsburgh PA USA) was installed. After the filter was in place the gas handling system was purged with Ar for 48 h to remove entrained air. Before experimental use the filter was conditioned for 48 h by passing 10 cm3 min-l of HCl through it. The discharge was generated at 40.68 MHz (RF Plasma Products Model HFS 5000D 0-5 kW Marlton NJ USA) by using a 2 turn induction coil (66.5 mm diameter 3.2 mm 0.d.copper tubing) around the 65 mm SICP ~0ntainer.l~ The end- on view of the centre of the discharge was imaged through a 100mm focal length quartz lens (2cm diameter Oriel) onto the entrance slit (50 pm) of a 0.35 m Czerny-Turner mono- chromator (Heath Model EU-700/E) with 1180 grooves mm-' grating blazed for 250 nm. Emission signals were detected with an RCA 4832 photomultiplier tube operated with an anode voltage of 1100 V. The output from the photomultiplier tube was amplified with a programmable picoammeter (Kiethley Model 18000-20 Kiethley Instruments Cleveland OH USA). A low pass filter and offset circuit were installed to condition the signal. Data were collected with a data acquisition board (DT 2905 Data Translation Marlboro MA USA) and IBM PS/2 Model 80 computer (International Business Machines Armonk NY USA).Operating Procedure An Ar plasma is formed under flowing conditions (typically with 100 cm3 min-l) at approximately 1.0 kW. Immediately after plasma initiation HCl is added to the Ar stream. The HC1 concentration is then increased by simultaneously increas- ing the HC1 flow and decreasing the Ar flow. Once the desired volume ratio of HCl is achieved the applied r.f. power is increased to 1.4 kW. As the flow of Ar is decreased a slight change in tuning is required to minimize the reflected power. The flowing plasma is maintained for approximately 30 min to stabilize the discharge. The sealed ICP operates in two modes gas flowing and non-flowing (static) at atmospheric pressure. For static oper- ation the discharge is isolated with valves V13 and V5 (Fig. 1) from the gas handling system.Temperature Determination The excitation temperature is one of the main physical param- eters that characterizes the excitation mechanism in equilib- rium plasmas. The temperature was obtained by plotting a graph of log (1A/gA) as a function of the excitation energy E,, of the upper level of the transition. If E is expressed in cm-l the resultant Boltzmann plot slope represents -l/(kT). In this expression I is the intensity of the line at wavelength A from the excited level of energy E,, g is the statistical weight of this level and A is the transition probability . l5 Neutral iron (FeI) is generally selected because its emission lines possess desirable characteristics for precise temperature I I V13 f Standard container Vent $.- Sealed ICP Fig. 1 Gas handling system used for the introduction of standards and samples into the sealed ICP Fl-F6 mass flow controllers; V1-V19 shut off valves; and CV check valveJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY determinations and large transition probability data sets are available.16 The criteria used in selecting suitable Fe I lines for the temperature measurement were (i) no significant spectral overlap of the lines should occur (ii) line intensities should fall within the dynamic range of the detector (iii) transitions to lower energy non-resonant ground state levels to avoid self-absorption and (iv) maximal spread in upper energy levels to minimize error in measurements of the slope of the tempera- ture p l ~ t .~ ' ' ~ The National Institute & Standards and Technology (NIST) transition probability and energy level values were used for the temperature measurements." The transition probabilities given by O'Brian et al. also were used.lg Calibration Diffusion tubes had been used previously with the SICP for the calibration of As and P in silane."*'' However diffusion tubes were not available as calibration standards for C or Sn. Therefore vapour-phase sampling was investigated as an alternative source for calibration. Mono but yl- tin trichloride ( Aldrich Chemical Milwaukee WI USA) MBTTC a liquid with a room temperature vapour pressure of 5.3 Pa is neither pyrophoric nor toxic and was used as both the Sn and C standard.The MBTTC was pipetted into a 316 1 stainless-steel sample cylinder (Whitey 316LSS 150 cm3 Model 316L-HDF4-150) fitted with an outage tube adapter with a standard 26.4cm long tube stub (hereafter referred to as a bubbler). A known concentration of the MBTTC standard was introduced into the gas stream just prior to entering the discharge container. The MBTTC was carried by the argon flow in the tube stub as it bubbled through the sample and out through the outage tube adapter. The bubbler connection to the handling system is illustrated in Fig. 1. A mass flow controller (F6) located before the bubbler was used to meter the standard flow through the discharge. A 2.3 kPa crack pressure Teflon coated check valve (Model SS-4C-KZ-TS-1/3 Nupro Willoughby OH USA) was placed immediately before the stream merging point with the plasma gas.The calibration standard was added from the bubbler to the HCl-containing gas stream. The concentration of the standard was varied by changing the total gas flow of the Ar gas stream. Additionally since the discharge container has a volume of 100cm3 information could be obtained for both absolute concentration ng and ppb (m/m). A calibration function was obtained by varying the concentration of Sn from 431 (2.3 pg Sn per g HC1) to 2777 ng (15 pg Sn per g HC1) and C from 174 (0.94 pg Sn per g HCl) to 1123 ng (6.1 pg Sn per g HCl). The lower concentration range was determined by the flow of the available mass flow controllers. Results are reported as the average of three replicate experiments per Sn (or C) concentration.Analyte equilibration time was also evaluated by introducing 414 ng of Sn (as MBTTC) into the discharge with a total gas flow rate of 61 cm3 min-'. The net Sn emission was monitored for 6 min. Equilibrium was reached after 2.5 rnin and remained constant for 2.5 min. A 4 min sample introduction time was adopted. The analyte purge time was evaluated by introducing 414 ng of Sn into the plasma gas stream with a total gas flow of 61 cm3 min-'. The net Sn emission was recorded for a total of 25 min. The Sn emission was decreased after 15 min. Therefore a 20 min purge time between samples was adopted. The calibration standard was allowed to flow through the discharge container for 4 min before emission intensities were recorded at either C I 247.857 nm or Sn I 326.234 nm lines.All experimental determinations were made in triplicate. Results and Discussion Maximum Hydrogen Chloride Addition Hydrogen chloride concentrations of less than 22% v/v in the discharge can be maintained with a minimum of 400 W for NOVEMBER 1994 VOL. 9 1301 100 l o 0 i 400 600 800 1000 1200 1400 1600 1800 2000 PowerAiV Fig. 2 Maximum HCl content as a function of net r.f. applied power. Data obtained by introducing the largest tolerable concentration of HCl into the discharge with a fixed r.f. power both flowing and static operating modes. Increasing the HC1 content to 38% v/v is possible by increasing the forward applied power to greater than 500 W. Formation of a 100% v/v HCl plasma at powers of > 1.6 kW can be accomplished in a 65 mm discharge container.However plasma stability for concentrations > 80% v/v is limited to approximately 20 min. Hydrogen appears to diffuse through the quartz container and causes arcing to the induction coil. The relation between r.f. power and maximum obtainable HCl is illustrated in Fig. 2. Impurity Identification The discharge was generated initially with 5% HCl in Ar (v/v) at 1.0kW. As the concentration of HC1 in the plasma was increased the power was increased to maintain the stability and diameter of the plasma. A concentration of 26% v/v HC1 at 1.5 kW was obtained. Although higher HC1 concentrations can be sustained the discharge requires powers of approxi- mately 2.0 kW to maintain a constant plasma size. Low power however results in a longer discharge container lifetime.The monochromator wavelength was scanned from 200-900nm and impurities identified in the ICP spectra by verifying the presence of at least three prominent lines for each element when possible except C for which only the C I 247.857 nm line was used. Impurities identified included Al As B C Ca Cr Cu Fe Ni and Sn (Figs. 3 and 4). Owing Si I 1 0 0 0 ~ fl .- a 700 750 I 650 ' I I I 250 255 260 265 Wavelengt h/n m Fig.3 Emission spectrum of Si I and Fe I1 (250-265nm) in 26% HCl v/v in Ar at 1.5 kW in a 65 mm discharge container. The peak at 257 nm is attributed to Clz molecular emissionI21302 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 - C cn E Lo 0 u! Iz 650 625 600 305.00 310.00 315.00 320.00 325.00 330.00 Wavelengt h/n m - I I I - Fig.4 Emission spectrum of A1 I Sn 1 and Cu I (305-330 nm) in HCl 26% v/v in Ar at 1.5 kW in a 65 mm discharge container. The broad molecular emission has been attributed to chlorineI2 to the previous transfer of arsine and boron trichloride in the gas handling system As and B emission appear in all spectra. Since the coil is in contact with the container wall Cu from the induction coil is believed to be diffusing through the quartz container and thereby producing significant emission.12 The molecular emission in Figs. 3 (251 nm maximum) and 4 has been identified as C1 molecular emission and has been dis- cussed e1~ewhere.l~ Additionally the analyte impurity emission was observed to decrease from flowing to static conditions. Therefore all further analyses were conducted in the flowing mode.Effect of HCl and Modifier on Net Sn Signal Owing to the physical changes in the size of the plasma as a function of HCl concentration HCI concentrations (16,20 and 24%) needed for standardization were investigated. Introducing more than 15% C1 into the 20% HCl-Ar plasma or more than 4% C1 into the 24% HC1-Ar plasma was not possible. Additions of up to 26% C1 were introduced into a 16% HC1-Ar plasma. When compared with the 20% and 24% HC1-Ar plasmas the l6Y0 HC1-Ar plasma was more robust and therefore chosen for further investigation. For a 16% v/v HCl-Ar plasma the concentration of C1 was varied from 6.5 to 26% for a constant Sn concentration of 414 ng. For C1 concentrations below 16% Sn appeared to deposit inside the container.The Sn sensitivity increased with increas- ing C1 concentration above 16%. To minimize the contri- bution of the impurities present in the C1 to the HC1 analysis the smallest concentration of C1 necessary to prevent depos- ition was adopted. Therefore standardization conditions were set at 16% HCl v/v 16% Cl v/v in Ar at 1.4 kW with a total flow of 61 cm3 min- through the discharge. Temperature Determination All the Fe I spectral lines from the 16% v/v HC1 with 16% C1 v/v in Ar discharge viewed 1 cm from the centre of the discharge fitting all the criteria are listed in Table 1. A line with an r2 value of 0.88 was obtained (Fig. 5 ) that yielded a temperature of 10 500 f 1500 K. With the O’Brian et al. trans- ition pr~babilities,’~ the calculated temperature was 11 800 2400 K (r2 = 0.77).Since no data exist to suggest that the SICP is in local thermal equilibrium (LTE) with 32% molecular gas in Ar the Fe I temperatures may not be a unique measure of the energy characteristics of the SICP. Table 1 for temperature determination Excitation energy,18 gAi8 and wavelengths for Fe I lines used A/nm 294.79 300.10 304.76 305.91 356.54 351.01 358.12 360.89 361.88 363.15 364.78 373.49 373.71 374.56 374.95 375.82 376.38 381.58 382.04 382.59 Eexclcm - 34329 34017 33507 33096 35768 35379 34844 35856 35612 35257 34782 33695 27 167 27395 34040 34329 34547 38175 33096 33507 €54 2.2 2.8 2.9 2.0 7.8 18.0 23.0 10.0 9.5 8.6 6.1 20.0 1.5 1.2 13.0 10.0 6.2 16.0 12.0 8.9 IFlowing 13 19 28 10 11 32 68 12 23 23 22 74 32 36 41 28 16 14 45 31 log(lA/gA) 3.24 3.31 3.47 3.18 2.70 2.80 3.02 2.64 2.94 2.99 3.12 3.14 3.90 4.05 3.07 3.02 2.99 2.52 3.16 3.12 5.0 1 1 4.5 F\ 4.0 5 3.5 3.0 07 2 2.5 Y 1.5 2.0 1 1.01 1 I I I I 1 1 26000 28000 30000 32000 34000 36000 38000 40000 E Jcm - ’ Fig.5 Temperature determination for 16% HC1 with 16% C12 in argon at 1.4 kW using Fe I lines viewed 1 cm from the centre of the discharge slope =0.0001362 -t 15%; intercept = 7.739 However the Fe I temperature reflects the ability of the discharge to populate excited levels in the energy range under ~0nsideration.l~ Cali bra tion A linear fit to the tin data gave a slope of 0.6803_+0.013 and an intercept of 0 with an r2 value of 0.9938 (Table2).The detection limit for Sn in the gas stream was 49 ng 8-l. The detection limit was calculated as 3 times the standard deviation of the background divided by the slope of the calibration func- tion.The background value was obtained from the least concentrated sample. For C a linear fit gave a slope of Table 2 Tin calibration 16% HCl-l6% C12 in Ar; 1.4 kW for Sn I 326.234 nm CSnl/ng CSnl/pg g-’ Average _+ SD/ng RSD (Yo) 43 1 2.3 376+45 12.1 1058 5.7 783 & 49 6.3 1659 9.0 1170+75 6.4 223 1 12.0 1487 +_ 77 5.2 2777 15.0 1853 k 79 4.3JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1303 Table 3 Carbon calibration 16% HCI-16% Cl in argon; 1.4 kW for C I 247.857 nm CCl/ns CCl/PS g - Average -t SD/ng RSD (Yo) 174 0.94 23+4 16.6 428 2.3 54+4 8.3 67 1 3.6 78f4 5.1 903 4.9 112f2 1.9 1124 6.1 139+1 0.3 The advantages of the SICP have been extended to the analysis of reactive gases.Additionally detection limits for C and Sn have been demonstrated to be competitive with conven- tional ICP-AES. 0.1228+0.003 with an intercept of 0.2368k2.55 and an r2 value of 0.9983 (Table 3). The detection limit for C was calculated as 271 ng g-'. These SICP detection limits are comparable with those of the conventional ICP for C (0.01 ppm m/v) and tin (30 PPb m/vh Conclusion With the SICP the direct analysis of 16% HCl was accomplished with minimal sample dilution and no sample preparation. Although HC1 is a corrosive gas the SICP provided an enclosed system for safe analysis. The direct analysis of HCl is an on-going research area and experiments are planned to compare data with hydrolysis experiments.' An equal concentration of Clz to HCl was required to prevent deposition of Sn and C standards in the discharge.Doping the HCl with Sn and C resulted in a detection limit of 49 and 271 ppb respectively in a discharge containing 16% hydrogen chloride v/v. Additionally absolute detection limits of 50.2 ng and 9.1 ng were obtained for C and Sn respectively. Extrapolation of these detection limits to a discharge contain- ing pure HCl results in detection limits of 306 ppb m/m and 1694 ppb m/m for Sn and C respectively. The repeatability of these data suggests that vapour phase sampling is an alternative to diffusion tubes for calibration. One limitation of this approach is the unavailability of vapour pressure data of suitable compounds. Additionally many com- pounds do not have an ideal vapour pressure.This would require heating and or cooling of suitable compounds and a more complex delivery system would be needed. However the use of diffusion tubes may provide better repeatability owing to a decreased dependency on temperature variation. The use of diffusion tubes for elements that do not form hydrides is under investigation. Iron impurities present in the HC1 allowed the temperature of the discharge to be estimated at 10 500+ 1500 K. The large deviation for the temperature measurement is due to exper- imental error. The value is somewhat lower than a similar discharge in C1,.20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 This research was supported by Air Liquide. Experiments were performed at the University of Massachusetts.References Hardwick S . J. Lorenz R. G and Weber D. K. Solid State Technol. 1988 10 93. Prange A. Kramer K. and Reus U. Spectrochim. Acta Part B 1991 46 1385. Faix W. G. Schramm W. Vix F. Weichbrodt G. and Hendelman R. Fresenius 2. Anal. Chem. 1988 329 847. Roth H. J. and Neunteufel P. Proceedings of the Satellite Symposium to ESSDERC 1989 Berlin. Pivonka D. E. Appl. Spectrosc. 1991 45 597. Handbook of Chemistry and Physics ed. R. C. Weast The Chemical Rubber Company Cleveland OH 61st edn. 1980. Thompson M. and Barnes R. M. in Inductively Coupled Plasmas in Analytical Atomic Spectrometry eds. Montasser A. and Golightly D. W. VCH Publishers New York 1992 ch. 5. Bridenne M. Carre M. Coffre E. Marot Y. and Simondet F. presented at the 1992 Winter Conference on Plasma Spectrochemistry San Diego CA January 1992 poster WP35. Schram J. Fresenius. Z . Anal. Chem. 1992 343 727. Jahl M. J. Jacksier T and Barnes R. M. J . Anal. At. Spectrum. 1992 7 653. Jahl M. J. and Barnes R. M. J. Anal. At. Spectrom. 1992,7 833. Jacksier T. and Barnes R. M. J. Anal. At. Spectrum. 1992,7,839. Jacksier T. and Barnes R. M. Appl. Spectrosc. 1994 48 382. Jacksier T. and Barnes R. M. Spectrochim. Acta Part B 1993 48 941. Jarusz J. Mermet J.-M. and Robin J. P. Spectrochim. Acta Part B 1978 33 55. Kubota A. Fijishiro Y. and Ishida R. Spectrochim. Acta Part B 1981 36 697. Blades W. M. and Caughlin B. L. Spectrochim. Acta Part B 1985 40 579. Corliss C. H. Bozman W. R. Experimental Transition Probabilities for Spectral Lines of Seventy Elements NBS Monograph 53 U.S. Government Printing Office 1962. O'Brian T. R. Wickliffe M. E. Lawler J. E. Whaling W. and Brault J. R. J . Opt. SOC. Am. B 1991 8 1185. Jacksier T. and Barnes R. M. Spectrochim. Acta Part B in the press. Paper 4/0005 71 Received January 5 1994 Accepted July 18 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901299

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Direct Solid Sample Analysis in a Moderate-power Argon Microwave-induced Plasma With Spark Generation Yong-Nam Pak Department of Chemistry Education Korea National University of Education Cheong- Won Chungbuk 363-797 Korea S. R. Koirtyohann Department of Chemistry University of Missouri Columbia MO 65207 USA 1305 A spark-Ar microwave-induced plasma (MIP) atomic emission spectrometry system has been developed for the direct analysis of solid samples. Particles of the sample are generated with a spark and swept into a moderate-power Ar MIP connected to a sequential spectrometer. Low-Alloy Steel and Aluminium Alloy Standard Reference Materials were analysed by the proposed system and the detection limits for most elements of interest were around 10 pg g-’ or less which are comparable to a spark-Ar inductively coupled plasma (ICP) system.Precisions were in the range 3-1 1 O/O which are 2-3 times higher than those of the spark-ICP system. Hence spark-MIP AES is a viable alternative technique to spark-ICP-AES. Keywords Argon microwave-induced plasma; spark atomic emission spectrometry; direct solids analysis When a solid sample is analysed it is advantageous to use the solid sample directly rather than dissolving it because of the time and expense involved as well as the risk of contamination dilution of sample components and the difficulties of dissolving some samples. Also typical nebulization of a sample solution delivers only 1-2% of the total solution into the atomizer and the remainder goes to waste.Arc/spark discharges,’-’ laser a b l a t i ~ n ~ ? ~ and electrothermal vap~rization’.~ have all been used to introduce solids samples into inductively coupled plasmas (ICP) and less widely into direct current plasmas (DCP). The microwave-induced plasma (MTP) has become established as an efficient excitation source and applications to various samples have frequently appeared in the literature. In addition to aqueous sample recent develop- ments have involved using an MIP as an ionization source for a mass spectrometer in MIP mass spectrometry ( MS).l3-I5 However the conventional low-power MIP has shown some limitations in the vaporization of samples. Thus the MIP has been used either as an excitation source combined with an external atomization ~ o u r c e ~ ~ ~ ’ or only gaseous samples have been introduced.More recent developments of a higher power (500 W or more) MIP18-21 have provided wider applications for aqueous samples. However only a few applications of MIPS to solid sample analysis have been reported. Gelhausen and Carnahan22 used a 500 W He MIP for solid sample analysis using a solid powder injection device. Other workers have used a spark/arc d i ~ c h a r g e ~ ~ ~ ~ or laser v a p ~ r i z a t i o n ~ ~ - ~ ~ to introduce solid samples into a low-power MIP. Two papers that have been published on an spark-Ar MIP at low power (about 100 W) are by Helmer and W a l t e r ~ . ~ ” ~ ~ They used a low-power (130 W) plasma in a TE013 tapered The effluent from the spark passed into an MIP and the emission from the eroded metals in the microwave discharge was monitored.They found that the linearity of the results obtained for the concentrations of several elements was acceptable (r values around 0.98) but because of the low power the system was subject to numerous interferences and particles that were not completely melted even after they had passed through the plasma were found.24 However a moderate-power plasma is expected to excite solid aerosols more efficiently. Combination of a spark and a moderate-power Ar MIP should provide an economical yet efficient method for the direct atomic emission spectrometric analysis of solid samples. Experiment Instrumentation A block diagram of the experimental system is shown in Fig. 1. A spark is generated inside the sampling chamber which is electrically connected to an a.c.spark power supply. The spark is formed between the two electrodes as a point-to-plane configuration with the plane usually being the sample. A jet electrode is used the characteristics of which are explained under Results and Discussion. The particles generated are sent to the plasma through a mixing chamber to eliminate any large particles. The Ar-MIP vaporizes and re-excites the particles introduced. Since the plasma torch and cavity have been explained in a previous paper,lg they will not be discussed again here. The spectral signal is monitored by a sequential slew scanning spectrometer. A computer controls the operation of the spectrometer and can also statistically manipulate the data and provide reports of the results.It was convenient to generate the spark in a spectrograph arc/spark stand (Model number 16-300 Jarrel Ash Franklin MA USA). The stand provides a metal cage with convenient connections for the electrical leads and contains two clamps for the sample and the counter electrodes. The spark stand ensures safe operation and reduces the stray r.f. electric field generated by the spark. A separate sampling chamber was required to enclose the spark providing a convenient sample mounting and con- trolling the gas stream that sweeps the particles into the plasma. The sampling chamber was made from poly(tetra- fluoroethylene) (PTFE) end-caps and a quartz tube body and Fig. 1 +- U Plasma gas Block diagram of the experimental set-up1306 L Clamp JOURNAL OF ANALYT1CA.L ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 I -Graphite electrode Gas out Gas in (tangential) lass chamber Fig. 2 t Gas in (Gas sweep) Schematic diagram of the sampling chamber is shown schematically in Fig. 2. The dimensions of the glass cylinder are 25 mm i.d. and 15 mm high. Tangential gas inlet and outlet tubes (4mm i.d.) were attached at the side of the chamber. The top piece of PTFE was 35 mm 0.d. with a 7 mm id. hole so that a sample could be sparked through the hole. The bottom piece of PTFE had either 4 mm or a 3/16 in hole depending on the counter electrode used. For a glassy carbon electrode a 4 mm i.d. hole was employed. Alternatively a carbon electrode with a 3/32 in hole drilled through the centre could be used.Rubber O-rings were used to seal any gas leaks The sample with a flat area 1-2cm in diameter was placed on top of the chamber. A carbon electrode in the upper clamp of the spark stand held the sample in place and provided electrical connection. The sharpened counter electrode was held in the lower clamp at on appropriate distance from the sample. A Perkin-Elmer inductively coupled plasma unit (PE Model 6500 Norwalk CT USA) was modified to house the MIP. The ICP load coils and spray chamber of the ICP torch were removed and the MIP cavity mounted on the optical axis. The plasma was usually viewed axially with the axis of the plasma perpendicular to the plane of the entrance slit. This could cause the hot plasma gas to enter the spectrometer and damage its components.Thus a vertical stream of air which cut off the hot plasma gas was placed between the spectrometer and the plasma. The PE Model 6500 spectrometer is equipped with gratings for operation in the visible and ultraviolet (UV) regions. The monochromator is a Czerny-Turner type with a focal length of 608 mm. The gratings are holographic and the reciprocal linear dispersion is 0.65 nm mm-I in the UV and 1.3 nm mm-' in the visible region. A typical bandpass of 0.02 nm was used for these experiments Reagents and Samples The samples used in this study were conducting metals such as National Institute of Standards and Technology (NIST Gaithersburg MD USA) Standards Reference Materials (SRM) Low-Alloy Steel (1261a-1264a) and Aluminium (601-604 and 7075). The counter electrode used in the spark was a graphite rod of spectroscopic grade (Union Carbide Sommerville NJ; USA) although some of the early work was done with a glassy carbon electrode (Sigri Sommerville NJ USA).The Ar gas used was 99.995% pure and was obtained from Matheson (Joliet IL USA). Procedures A conductive sample was sanded with a 120 grit belt to smooth the surface and cleaned with acetone to remove any organic material The sample was then placed on top of the spark chamber and pressed by the clamps against the O-ring. It was sparked for 3 min including a 30 s pre-spark. After three burns or less a new sample area was used and the gap was rechecked and adjusted if necessary. Whenever the spark chamber was opened Ar gas was flushed for 1 min before the run to remove any entrained air.After approximately 50 burns the mixing chamber was cleaned and the connecting tubes were replaced with new ones because particles of previous samples tend to build up in them. After approximately 300 burns the surface of the auxiliary electrode was cleaned and polished again. Results and Discussion Optimization of the Spark-Ar MIP Selection of the optimum conditions was based on the best signal-to-background ratio (S/B) precision and sensitivity obtained. Although several experimental parameters were inter-related attention was given to one condition at a time while the rest were kept the same. The parameters selected were firstly the plasma parameters such as power sample gas and tangential (or plasma) gas flow rate. The spark conditions such as power number of breaks per half-cycle (BHC) capaci- tance reactance and the distance for the spark gap were considered next.The last parameter investigated was the mode of introduction of the sample gas ix. in a tangential or gas- sweep mode. When the sample gas flows through the side arm of the sampling chamber it is in the tangential mode and when through the drilled counter electrode it is in the gas- sweep mode (see Fig. 2). The optimum conditions used are summarized in Table 1. Plasma The data were collected from the centre of the plasma with axial viewing. Visual observation of the Ar MIP revealed a 'doughnut' shape similar to an Ar ICP with the working conditions used (Table 1). Since most of the elements under investigation gave the best S/B at the centre the plasma was observed along the axis at the centre.The signal-to-noise ratio (S/N) as a function of the power of the Ar MIP is shown in Fig. 3. The linear increase in S/N with power was observed for the Low-Alloy Steel sample using the Fe 238.2 nm line. It is obvious that a higher power should be more beneficial in the analysis of solid samples. However occasional arcing inside the tuning stubs was observed at a higher power. Once arcing has developed it burns the tip of the inside tuning rod causing an impedance mismatch and quickly quenches the plasma or damages other parts in a few seconds. Thus a relatively low and safer power 250-300 W was used in the Ar MIP. This means that sensitivity was reduced by a factor of 2-3 under the present conditions. Table 1 Optimum operating conditions for Low-Alloy Steel and Aluminium in spark-Ar MIP with the gas-sweep configuration Plasma conditions- Plasma Ar flow rate/l min-' Sample Ar flow rate/l min-l Power forward/W Power reflected/W Distance from the tip Integration time/s of the sample channel to the plasma/cm Spark conditions- Inductance/pH Capacitance/nF Breaks per half cycle Spark gap/mm Variac power setting Breakdown voltage/kV Pre-spark/s Low-Alloy Steel 40 2.5 2 4 30 13.4 30 1 .o 0.35 300 > 5 5 5 Aluminium 3 10 2.5 1 4 25 10.7 30JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1307 300 400 500 Power/W Fig. 3 S/N ratio as a function of power for Ar MIP. Low-Alloy Steel sample was studied at 238.2 nm Signal changes with the tangential and sample gas flow have been studied and the results obtained are shown in Figs.4 and 5 respectively. An increase in the tangential gas flow deceases the signal virtually linearly A minimum tangential flow of about 1 1 min-' was required to keep the plasma off the wall of the torch. Thus this minimum tangential gas flow of 11 min-' was used in all experiments. The sample gas flow shows a maximum at 0.35 1 min-l. Spark The current flow could change during the run because the counter electrode burns away changing the distance between the two electrodes. Thus careful control of the spark was needed by adjusting the power and the auxiliary controlling gap until the current flow of the spark was determined by the 30 I 1 0 1 2 3 4 5 6 7 Tangential gas flow rate/l min-' Fig.4 Intensity as a function of the plasma tangential gas flow in the spark-Ar MIP 17 I 1 3 16 .- a 5 1 \ 0.3 0.4 0.5 0.6 Sample gas flow rate/l min-' Fig.5 spark-Ar MIP Intensity as a function of the plasma sample gas flow in the auxiliary gap. The spark gap was changed and the behaviour of the spark was monitored with an oscilloscope. If the pattern looked unstable the auxiliary gap was increased until the breaking voltage pattern remained almost the same. Once the auxiliary gap was set the power was changed to obtain stable breaking patterns and to determine the number of BHC. Then under a fixed power setting (fixed transformer and capacitor settings) the inductance was changed to optimize the con- ditions for the best S/B and S/N. The sample gas configuration was changed and the whole process was repeated for the spark.Analytical curves for the Low-Alloy Steels SRMs were then obtained and examined. If the relative standard deviations (RSD) were too large or if the correlation coefficients (r) were too small to be acceptable (below 0.95) for a certain element then a new spectral line was sought. Finally the sample was changed from the Low-Alloy Steel to the Aluminium Alloy and optimization of the spark conditions as well as the gas configurations repeated. Temporal behaviour of the spark-MIP There are two types of signal changes one of which is from the spark or noise within a run (short-term noise) and the other is between the runs (long-term noise). To study noise within a burn a steel sample was sparked and analysed with an Ar MIP for 3 min which is the maximum setting of the timer on the instrument.Two types of temporal behaviour were observed and are shown in Fig. 6. Curve A represents the Cr Ni Si Cu and Fe group of elements which shows a relatively flat response with time. Curve B which represents the Al Mn and Mg group of elements shows a 20-30% decay in the signal intensity over the 3 min. The same results have been reported for a spark-Ar ICP,28 except for Mn. The signal begins to rise when the spark starts and reaches a plateau in 20-30 s. Thus it was necessary to pre-spark a sample for 30 s before the data were collected and processed. The Ar spark shows a 20-30% decrease in intensity for Mn under the tangential configuration and the decay is more striking for previously sparked sample surfaces.This is partly due to fractional distillation as it is known that Mn distils from steel during furnace melts. However this does not explain completely why the signal decays for one group of elements (B group) and not for another (A group). There are conflicting reports regarding this phenomena. Ekimoff and W a l t e r ~ ~ ~ suggested that the change in signal is not due to changes of composition with time but rather to changes in the total amount of material produced. However Strasheim and Blum3' 1000 1 ln 500 w .- t 3 t 2 $ 0 c .- - > ln C a c. .- e - 500 - - 1000 I I I 0 1 2 3 4 Ti me/m i n Fig.6 Curves showing two types of temporal behaviour of Low- Alloy Steel in the spark-Ar MIP A Cr Ni Si Cu and Fe; and B Al Mn and Mg1308 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 noticed a change in the surface composition with time during sparking. The second type of noise (long-term noise) which occurs between the runs was examined by repeatedly sparking the same sample surface several times. While a sample is being sparked the surface characteristics are expected to change which in turn could alter the mechanism or efficiency of sampling. If indeed such phenomena occur a large change in the signal as the sampling frequency increases could be expected. After five burns the signal changed dramatically and was reduced by more than 50% for the Aluminium Alloys. When a new sample surface was used the signal was immedi- ately restored to the original value. In the present experiments a new surface was used after a sample surface had been sparked three times or less.Sample gas configuration The RSDs of two different configurations are compared in Table 2 using a major element as an internal standard. Again the values were obtained after triplicate observations except for the major elements. A comparison between the major elements clearly shows that the gas-sweep configuration is a more precise method. The S/B study for the two sample gas configurations is listed in Table 3. To understand the difference the sample transport rate was examined by collecting the particles from a trap installed just Table 2 Comparison of RSDs between two sample gas configurations in the spark-Ar MIP RSD(%) Element Line/nm Low-Alloy Steel- Fe 238.2 Ni 227.0 c u 224.7 c o 228.6 Cr 267.7 A1 396.15 Mn 257.6 Aluminium Alloy- A1 396.2 Fe 238.2 c u 324.75 Mg 279.55 Si 253.0 Mn 257.6 Cr 267.7 Gas sweep 6.8 7.5 1.2 2.7 3.4 7.7 4.6 7.7 13.7 8.5 11.9 7.2 10.4 5.4 Tangential 9.5 6.4 4.8 - - 8.8 - 12.6 24.4 11.5 10.7 - - - Table 3 Comparison of S/B ratios for two sample gas configurations in the Spark-Ar MIP Element Line/nm Aluminium Alloy- Mn 257.6 Si 251.6 Mg 279.5 c u 324.75 Low-Alloy Steel*- Cr c u Mo Ni * From ref.28. Tangential 7.9 (1.03/0.13) 4.1 (1.85/0.45) 2.1 (0.247/0.12) 3.4 (0.84/0.25) Gas sweep 14 (1.38/0.10) 10 (2.90/0.28) 7 (0.41/0.06) 23 (2.44/0.11) before the plasma. The sample transport rate of the Low-Alloy Steel for the gas-sweep configuration was determined to be 0.50pgs-' while for the tangential flow it was 0.2Opgs-'.This partly explains why the gas-sweep configuration gives higher S/B values. The other reason is possibly due to the smaller particle sizes produced in the gas-sweep configur- ation.28 The background and signal both favour the gas-sweep mode to achieve higher S/B values. The background is measured from the intercept of the total intensity versus concentration curve. Other investigators28 have also reported that the gas-sweep configuration gives lower RSD values and detection limits than does the tangential mode. Analysis of Steel and Aluminium Samples With the Spark-Ar MIP The excitation ability of a moderate-power Ar MIP was investigated by examining several different transitions of Mn with excitation energies ranging from 3.07 to 13.24 eV. The analytical curves of the 257.61 (Mn 11) 293 (Mn 11) and 403.3 nm (Mn I) lines are shown in Fig.7. The intensity ratios of Mn to Fe for the 259.9 nm line are plotted with respect to concentration (%). The y-intercept does not pass through zero which suggests that a portion of the background remains uncorrected. The backgrounds were corrected by subtracting the intensity at a wavelength near to the analytical line used for the analysis. From Fig. 7 it is clear that an Ar MIP is an efficient excitation source for all different energy levels. Other analytical curves show similar linearity. The correlation coefficients (r) of the analytical curves for a spark-Ar MIP for Mn Cu and Cu are 0.9995 fO.0001 0.9980f0.0023 and 0.9996 f 0.0001 respectively. The r values obtained in the spark-Ar MIP system are higher than the spark-He MIP31 but less than the spark-Ar ICP.28 Detection Limits and Precision The detection limits (DL) for the Low-Alloy Steel and Aluminium SRMs in the spark-Ar MIP are listed in Table4 along with the S/B values and noise.Most of the DL values were around 10 pg 8-l in both types of sample. The RSDs of two different samples were obtained from three replicate measurements of the relative intensities and are shown in Table 5. Three replicates were chosen because this is the typical number of observations made in a real experiment. The RSDs of the major components such as Fe in Low-Alloy Steel and A1 in Aluminium Alloy SRMs were obtained by measuring the signal intensities more than 15 times. A comparison between 13 11 al LL i s 4- $ 5 .- 4- 2 7 >.v) .- - 3 1 0 0.5 1.0 1.5 2.0 Concentration (%) Fig. 7 Analytical curves of Low-Alloy Steels for different Mn trans- ition in the spark-Ar MIP showing the intensity ratio of Mn to Fe (at 259.9 nm) versus concentration 0 Mn 257.61; @ Mn 403.3; and A Mn 293 nmJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1309 Table 4 Detection limits for the spark-Ar MIP Element Low-Alloy Steel- Cr Ni Mn c u Si Aluminium Alloy- Fe Si Mg Mn c u Zn Line/nm 267.7 227.2 259.0 224.7 253.0 238.2 251.6 279.55 257.6 324.8 213.86 Concentration (YO) 0.30 0.32 0.69 0.098 0.228 0.205 0.18 0.39 0.079 0.29 0.029 SIB 9.3 5.2 12 32 15 4 3 150 176 11 0.3 SIN 1000 1400 480 1000 480 770 400 29000 2300 91 470 Detection limit (%) 0.00090 0.00070 0.0043 0.00043 0.0015 0.00080 0.0014 0.00040 o.Ooo10 0.0096 0.00019 Table 5 RSD values of Low-Alloy Steel and Aluminium with the gas- sweep configuration in the spark-Ar MIP Element line/nm Low-Alloy Steel- Fe 238.2 Ni 227.0 c u 224.7 c o 228.6 Cr 267.7 A1 396.15 Aluminium Alloy- A1 396.2 Fe 238.2 c u 324.75 Mg 279.55 Si 253.0 Mn 257.6 Cr 267.7 Concentration (YO) 95 0.14 0.098 0.048 0.066 0.24 < 94 0.52 4.38 1.56 0.13 0.81 0.24 RSD (Yo) 6.8 7.5 1.2 2.7 3.4 7.7 7.7 13.7 8.5 11.9 7.2 10.4 5.4 Table6 Comparison of the DLs and RSDs for the spark-Ar MIP with the spark-Ar ICP in the Low-Alloy Steel Element DL(%)- Cr Mn A1 Ni c u Zn* RSD (Yo)- Cr Mn A1 Mg* * In aluminium.Spark-MIP 0.0009 0.0043 0.00027 0.00070 0.00043 0.00019 3.4 4.6 7.7 10.7 Spark-ICP 0.0024 0.0030 0.00020 0.00030 0.00050 0.00020 2.1 1.4 1.1 3.1 Fe in Low-Alloy Steel SRMs and A1 in Aluminium SRMs shows that the RSD for Fe was slightly lower than that of Al.Most of the RSD values in Aluminium were greater than 7%. However the RSDs in the Low-Alloy Steel are around 5% except for A1 (396.3 nm) and Ni (221 nm) where there is a featured background. In conclusion Low-Alloy Steel samples give better precisions than the Aluminium Alloys. This could be due to either the excitation in the plasma or the sampling by the spark. Other s t ~ d i e s ~ ’ ~ ~ have reported the possibility of change in performance with different samples which seems to be the more likely explanation. Earlier work carried out by Prell and Koirtyohann2* with the spark-Ar ICP demonstrated good precision less than 5% RSD for Low-Alloy Steel and Aluminium samples.The detec- tion limits were mainly less than 0.0010% or 10ppm. The RSDs and detection limits for the two systems are compared in Table 6. The RSDs of the Ar MIP are 2-3 times worse than the Ar ICP. In particular the long-term stability (longer than 3min) of the Ar MIP is worse than that of the Ar ICP. However the Ar MIP and the ICP show comparable detection limits. It is surprising that even the 300 W Ar MIP is as efficient as a 1.5 kW ICP in analysing solid samples. Therefore a moderate power Ar MIP combined with a spark can be a very efficient tool for analysing solid samples directly. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 References Codeo A. G. Lopez M. T.D. Seco J. L. J. and Cobo I. G. J. Anal. At. Spectrom. 1992 7 11. Ono A. Saeki A. and Chiba K. Appl. Spectrosc. 1987 41 970. Marks J. Y. Fornwalt D. E. and Yungk D. E. Spectrochim. Acta Part B 1983 38 107. Human H. G. C. Scott R. H. Oakes A. R. and West C. D. Analyst 1976 101 265. Aziz A. Broekaert J. A. C. Laqua K. and Leis F. Spectrochim. Acta Part B 1984 39 1091. Leis F. and Laqua K. Spectrochim. Acta Part B 1978 33 27. Ishizuka I. and Owamino Y. Spectrochim. Acta Part B 1983 38 519. Matousek J. P. and Orr J. B. Spectrochim. Acta Part B 1976 31 875. Sneddon J. Spectroscopy 1986 1 34. Beenakker C. I. M. and Boumans P. W. J. M. Spectrochim. Acta Part B 1978 33 53. Michlewicz K. G. and Carnahan J. W. Anal. Chem. 1986 58 3122. Brown P. G. Haas D. L. Workman J. M.Caruso J. A. and Fricke F. L. Anal. Chem. 1987 59 1433. Wilson D. A. Vickers G. H. and Hieftje G. M. Anal. Chem. 1987 59 1661. Creed J. T. Davidson T. M. Shen W. Brown P. G. and Caruso J. A. Spectrochim. Acta Part B 1989 44 909. Park C. J. Pak Y. N. and Lee K. W. Anal. Sci. 1992 8 443. Matousek J. P. Orr J. B. and Selby M. Talanta 1986 35 11. Stahl R. G. Brett L. and Timmins K. J. J. Anal. At. Spectrom. 1989 4 337. Haas D. L. and Caruso J. A. Anal. Chem. 1984,56 2014. Pak Y. N. and Koirtyohann S. R. Appl. Spectrosc. 1991 45 1132. Wu M.-G. and Carnahan J. W. J. Anal. At. Spectrom. 1992 7 1249. Okamoto Y. Anal. Sci 1991 7 283. Gelhausen J. M. and Carnahan J. W. Anal. Chem. 1991 63 2430. Helmer D. and Walters J. P. Appl. Spectrosc. 1984 38 392. Helmer D. and Walters J.P. Appl. Spectrosc. 1984 38 399.1310 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 25 Uebbing J. Ciocan A. and Niemax K. Spectrochim. Acta Part B 1992 47 601. 26 Uebbing J. Ciocan A. and Niemax K. Spectrochim. Acta Part B 1992 47 611. 27 Leis F. and Laqua K. Spectrochim. Acta Part B 1979 34 307. 28 Prell L. J. and Koirtyohann S. R. Appl. Spectrosc. 1988,42 1221. 29 Ekimoff D. and Walters J. P. Anal. Chem. 1981 53 1644. 30 Strasheim A. and Blum F. Spectrochim. Acta Part B 1973 28 13. 31 Koirtyohann S. R. and Pak Y. N. Bull. Kor. Chem. SOC. 1994 in the press. 32 Nickel H. and Srachova J. Fresenius’ Z. Anal. Chem 1972 260 229. Paper 4/00043 I Received January 1 1994 Accepted March 10 1994JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1311 Aboal-Somoza Manuel 469 Absalan G. 45 Adams F. 151 Aizpun B. 1279 Akman Siileyman 333 Aller A. J. 871 Alvarado Jorge 1223 Alves Luis C. 399 Amarasiriwardena Dula 199 Anderson David R. 67 Anderson S. E. 263 Anderson Stan T. G. 1107 Anghel Sorin D. 635 Anzano Jesus M. 125 Argentine Mark D. 199 1121 Arnold J. T. 263 Arriagada Lorna 93 Arruda Marco A. Z. 657 Avila Akie K. 543 Baaske Bernd 867 Back M. H. 45 Barciela-Alonso Carmen 469 Barinaga Charles J. 1053 Barnes Ramon M. 199 981 Barrios Carlos 535 Barshick Christopher M. 83 Barth Peter 773 Baxter Douglas C. 297 Baxter Malcolm J. 727 Bayne Charles K. 83 Beauchemin Diane 509 Becerra Jose 535 Begley Ian S. 171 Belarra Miguel A. 125 BeneS Petr 303 Bermejo-Barrera Adela 469 Bermejo-Barrera Pilar 469 483 Bernasconi G.151 Berndt H. 861 Berndt Harald 39 193 Betti Maria 385 Bettinelli Maurizio 805 Biffi Claudio 443 Blanco Gonzalez E. 281 Blanco E. 1279 Bloxham Martin J. 935 Boge Edward M. 369 Bolland David T. 1255 Botelho Gloria M. A. 1263 Botto Robert I. 905 Boughriet A. 11 35 Bowins Robert J. 1233 Branch Simon 33 Brandt R. 1063 Bratter Peter 1293 Brenner I. B. 737 Briand Alain 17 Broekaert Jose A. C. 1015 1063 Brown Nicole V. 363 Bruhn Carlos G. 535 Bruno Sergio N. F. 341 BudiE Bojan 53 Burakov V. S. 307 Byrne John P. 913 Cabon J. Y. 477 Cai Xiangjun J. 697 Cairns Robert O. 881 Camara Carmen 291 Campbell Michael 187 Campos Reinaldo C. 341 1263 483 1121 1299 483 CUMULATIVE AUTHOR INDEX JANUARY-NOVEMBER 1994 Carey Jeffrey M. 975 Carrion Nereida 205 217 Caruso Joseph A.145 957 975 Castillo Juan R. 125 311 Cervera Maria Luisa 651 Chakrabarti Chuni L. 913 919 Chakrabarti C. L. 45 Charnley Norman R. 1185 Chartier Fredkric 17 Cheam Venghout 315 Chirinos Jose 237 Cimadevilla Enrique Alvarez- Cabal 117 Cleland Sandra L. 975 Cobo I. G. 223 Coedo Aurora G. 223 1111 Conver T. S. 899 Cooper 111 C. B. 263 Cordos Emil A. 635 Cornejo Silva G. 93 Cornelis Rita 945 Crain Jeffrey S. 1273 Crews Helen M. 615 727 Cserfalvi Tamas 345 Cujes Ksenija 285 Curtius Adilson J. 341 543 Dadfarnia Shayessteh 7 Dahl Kari 1 Dams Richard 23 177 187 815 1075 1243 Dean John R. 615 de Boer Jan L. M. 1093 De Kimpe Jurgen 945 de la Guardia Miguel 651 Dennis John 727 Denoyer Eric R. 927 Deram L. 1135 Deruaz D. 61 Desrosiers Roland 3 15 Donard Olivier F. X.1143 Doner Giileren 333 Dorado M. Teresa 223 11 11 Du Xiaoguang 629 Duan Yixiang 629 Durrant Steven F. 199 Ebdon Les 33 611 615 939 Elgersma Jaap W. 619 Eljuri Elias 205 Elmahadi H. A. M. 547 Emteborg Hiikan 297 Epler Katherine S. 79 Evans E. Hywel 939 Evans R. Douglas 985 Evans Susan 1249 Fadda Sandro 519 Fariiias Juan C. 841 Farrer Humphrey N. 1107 Fecher Peter A. 1021 Feinendegen Ludwig E. 791 Feldmann Ingo 1007 Fell Gordon S. 457 Feng Xinbang 823 Fernandez de la Campa M. R. Fernandez Alberto 205 217 Fernandez Maria Luisa 1279 Ferreira Margarida A. 1269 Fischer Johann L. 623 Fischer W. 257 375 FiSera M. 1285 Fisher Andrew S. 611 Florian K. 257 1263 23 1 Fonesca Rodney W. 167 Foster Robert D. 273 Foulkes Michael E. 615 Frentiu Tiberiu 635 Gallego Mercedes 657 663 691 Garcia Alonso Jose Ignacio Geertsen Christian 17 Ghazy Shaban E.857 Giessmann Ulrich 1007 Gilmutdinov Albert Kh. 643 Giovanonne Bruno 1209 Glatz Jean-Paul 1209 Golloch Alfred 867 971 Goltz Douglas M. 919 GomiSEek Sergej 285 Gonzalez Urcesino 535 Goode Scott R. 73 965 Goossens Jan 177 187 Gower Stephen A 363 369 Gras Nuri T. 535 Gray Alan L. 1179 Greb Ulrich 1075 Greenfield S. 565 Greenway Gillian M. 547 Gregoire D. Conrad 393 605 Griffin Steven T. 697 Grohs James 927 GiiCer Seref 797 Hadgu Negassi 297 Halls David J. 1177 Haraldsson Conny 1229 Harnly James M. 419 Harrison W. W. 991 1039 Hatterer Andre 525 Hauptkorn Susanne 463 Heitmann U. 437 Hernandez Cordoba Manuel 553 1167 Hese A. 437 Hiernaut Tania 385 Hill Steve J. 935 Hinds Michael W. 451 Hladky Z.1285 HlavaEek I. 245 251 HlavaEkova I. 245 251 Hoffmann Erwin 685 1237 Holcombe James A. 167 415 Hollenbach Mark 927 Horlick Gary 593 823 833 Houk R. S. 399 Hoult Gavin 7 Howe Alan M. 273 Hu Yanping 213 701 Huang Benli 779 Huang Zhuoer 11 Hudnik Vida 53 Hughes Dianne M. 913 Hutton J. C. 45 Hutton Robert C. 385 881 Imai Shoji 493 759 765 Inagaki K. 116 1 Ince Ahmet T. 1179 Isaevich A. V. 307 Ito Tetsumasa 1001 Itriago Ana 205 Jacksier Tracey 1299 Jackson Jason G. 167 Jakubowski Norbert 193 1007 Janssens K. 151 Jaramillo Victor H. 535 1209 1217 913 919 1015 Jepkens Brigitte 193 Jin Qinhan 629 851 Jones Delwyn G. 369 Jung Gerhard 1075 Kabil Mohamed A. 857 Kaneko Tetsuya 1273 Kantor Tibor 707 Karagozler A. Ersin 797 Katskov Dmitry A. 321 431 Kimber Graham M. 267 Kirschner Stefan 971 Kitagawa Kuniyuki 1273 Kloner A.737 Kmetov Veselin 443 Koch Lothar 385 1209 1217 Kogan Valentina V. 451 Koirtyohann S. Roy 997 Koirtyohann S. R. 1305 Kojima Isao 1161 Kolihova Dana 303 Kondo S. 1161 Koppenaal David W. 1053 Koropchak J. A. 899 Krahenbuhl Urs 1249 Kratzer Karel 303 Krieger Brian L. 267 Krivan Viliam 463 773 Kroft Marilyn 927 Krug F. J. 861 Krushevska Antoaneta P. 199 981 1121 Kubova Jana 241 1173 Kujirai Osami 751 Kumamaru Takahiro 89 Kurfurst U. 531 Laborda Francisco 727 Lacour Jean-Luc 17 Lamoureux Marc M. 919 Larsen Erik H. 1099 Laser Bernd 1075 Lazik C. 45 Le Bihan A. 477 Lechner Josef 3 15 Lee Julian 393 Leis F. 1063 Li Yongquan 679 Li Zhikun 679 Liang Yan Zhong 669 Liezers Martin 1 179 Lile E. S. 263 Littlejohn David 1255 Liu X.R. 833 Lonardo Robert F. 1195 Long James V. P. 1185 Lopez Garcia Ignacio 553 1167 Lopez Jose C. 651 Lopez-Gonzalvez M. Angeles Lord 111 Charles J. 599 Lourdes Bastos M. 1269 Ludke Christian 685 1237 Luecke Werner 105 Lyven Benny 1229 Ma Yizai 679 Mamich Stephen 927 Manickum Colin K. 227 Manninen Pentti K. G. 209 Manzoori Jamshid L. 337 Marais Pieter J. J. G. 321 431 Marchante Gayon Juan Marcus R. Kenneth 1029 1045 29 1 713 Manuel 1171312 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Marcus R.K. 45 Marin Sergio R. 93 Marshall William D. 1153 Martin Fabienne M. 1143 Martinez-Garbayo Maria Paz Martinsen Ivar 1 Massey Robert C. 615 Mauchien Patrick 17 McAllister Trevor 427 McCrindle Robert I. 321 431 McLeod Cameron W. 67 751 McNutt Robert H.1233 Mendes Paulo C. S. 663 Mermet Jean-Michel 17 61 Mezei Pal 345 Michel Robert G. 501 1195 Milagros Gomez M. 291 Miller-Ihli Nancy J. 605 1129 Milton Dafydd M.P. 385 Minnich Michael G. 399 Misakov P. Ya. 307 Mixon Paul D. 697 Moenke-Blankenburg Lieselotte 1059 Moens Luc 177 187 815 1075 1243 Mohl Carola 791 Momplaisir Georges-Marie Montoro Rosa 651 Moreda-Piiieiro A. 483 Moreda-Piiieiro J. 483 Moreno Rodrigo 841 Mori Toshio 159 Mostafa Mohamed A. 857 Moulton Gary P. 419 Mousty Francis 719 Munthe John 1229 Murillo Miguel 205 217 237 Nagengast Anton 1021 Nagulin K. Yu. 643 Nakahara Taketoshi 159 Nakamura Yoshisuke 751 Naoumidis A. 375 Naumenkov P. A. 307 Nevoral Vladislav 241 Ni Zhe-ming 669 Nickel H. 257 375 NiedergesaiD Rainer 107 1 Nobrega J. A. 861 Nolte Joachim 1059 O’Haver Thomas C.79,419 Ohman Peder 1229 Ohorodnik S. K. 991 Ohta Masahide 1273 Okamoto Yasuaki 89 Okamoto Yukio 745 Okochi Haruno 751 125 1087 217,841 1203 1153 Olson Lisa K. 975 O’Neill Peter 33 Outred Michael 381 Outridge Peter M. 985 Ozdemir Yuksel 797 Pagliosa Giorgio 1209 Pak Yong-Nam 1305 Palacios M. Antonia 291 Papaspyrou Manfred 791 Patriarca Marina 457 Pauwels J. 531 Payling Richard 363 369 Peachey Russell M. 267 Pepelnik Rudolf 1071 PCrez-Arantegui J. 3 11 PCrez Parajh Juan M. 111 Petit de Peiia Yaneira 691 Petrucci G. A. 131 Petty John D. 267 Pilger C. 1063 Platzner Isaac 719 PolakoviEova Jozefa 1173 Polettini Albert0 L. 719 Pollmann D. 1063 Popescu Adrian 635 Potts Philip J. 1185 Poussel E. 61 Prange Andreas 1071 Pretorius Warren G. 939 Prudnikov Evgeniy D.619 Pyrzynska Krystyna 801 Quentmeier Alfred 355 QuerrC G. 311 Rademeyer Cornelius J. 623 Radziuk Bernard 1 Raikov S. N. 307 Raith Angelika 1045 Rasmussen Gert 385 Recknagel Sebastian 1293 Reed Nicola M. 881 Reija Carmen 651 Reyes Olga 535 Richner Peter 985 Rivoldini Alessandro 5 19 Robert Robbie V. D. 1107 Robles L. C. 871 Rodriguez Aldo A. 535 Romon-Guesnier Sabine 199 RonEeviC Sanda 99 Rosenberg Rolf J. 713 Rosick Ullrich 1293 Rowland Stephen J 939 Rubio J. 151 Riimmeli Mark H. 381 Sala Jose V. 719 Salbu Brit 1 SaIeemi Abdollah 337 Salit Marc L. 997 Salud Seremi 535 Santelli Ricardo E. 663 1087 Sanz-Medel Alfredo 11 1 117 Schaldach Gerhard 39 Scheie Andrew J. 415 Schneider Germar 463 Schoknecht G. 437 Scholze Horst 1237 Schumann Thomas 1059 Schwarzer Rudolph 431 Schwuger Milan J.791 Segal I. 737 Sekerka Ivan 315 Selby Mark 267 Sena Fabrizio 1217 Sharp Barry L. 171 Sheppard Brenda S. 145 Shick Charles R. Jr. 1045 Shtepan Aleksander M. 321 Silva M. M. 861 Silva R. B. 861 Siroki Marija 99 Sjostrom Sten 17 Skole Jochen 685 Slowick Jeffrey J. 951 Smit Henri C. 619 Smith B. W. 131 1039 Smith Clare M. M. 419 Smith David H. 83 Smith Fraser O. 267 Smith Monty R. 1053 Smith Trevor A. 67 Soares M. Elisa 1269 SpEvaEkova VEra 303 Steers Edward B. M. 381 Steffan I. 785 1117 Stephens Roger 675 Stevenson C. L. 131 StreSko Vladimir 241 1173 Stuewer Dietmar 193 1007 Sturgeon Ralph E. 493 605 Sturup Stefan 1099 Su Evelyn G. 501 Sugawa Kazumitsu 89 Sutton Robert L. 1079 Sy T. 437 Takahashi Katsuyuki 751 Tan Yanxi 1153 Telgheder Ursula 867 971 Thoby-Schultzendorff Dominique 1209 Thomas Christopher L.73 965 Thomassen Yngvar 1 Thompson K. Clive 7 Tittarelli Paolo 443 805 Tittes Wolfgang 1007 1015 Tolg Gunther 1015 1063 Tomlinson Medha J. 957 Trincherini Pier R. 719 Tsalev Dimiter L. 405 Tschopel P. 1063 Tsuge Shin 1273 231 281 1279 1015 759 765 Turak Elvan E. 267 Turk Gregory C. 79 997 Uchida Hiroshi 1001 Uden Peter C. 951 Ulens Katia 1243 Valcarcel Miguel 657 663 691 ValdCs-Hevia y Temprano van der Velde-Koerts Trijntje Van Winckel Stefaan 1243 Vandevelde Leon 1243 Vanhaecke Frank 187 Vanhoe Hans 23 177 187 815 Varga Imre 707 Veber Marjan 285 Verbeek Alistair A. 227 Verrept Peter 1075 Versieck Jacques 23 Viiias Pilar 553 1167 Vincze L. 151 Vujicic G. 785 11 17 Wade Jeffery W. 83 Walden W.O. 1039 Walter Serge 525 Wang Jiansheng 957 Wang Jiazhen 679 Wang Jin 1153 Wang Mohui 1195 Wang Xiaohui 679 Wang Xiaoru 779 Wang Ying 851 Wartel M. 1135 Webb C. 263 Weiss ZdenEk 351 Wiederin Daniel R. 399 Williams J. C. 697 Williams John G. 1179 Williams Jr. J. C. 697 Willie S. N. 759 Wiltshire Guy A. 1255 Winefordner J. D. 131 1039 Worsfold Paul J. 611 935 Wrobel Katarzyna 117 281 Xiao Grace 509 Yang Chenlong 779 Yang Pengyuan 779 Yang Wenjun 85 1 Yu Lijian 997 Yuan Xianglin 851 Yuzefovsky Alexander I. 1195 Zakharov Yu. A. 643 Zander A. T. 263 Zaray Gyula 707 Zhang Hanqi 851 Zhang Zhanxia 213,701 Zheng Jianguo 213 701 Zhu Jim J. 905 Thuang Zhixia 779 Zilkova Jana 303 Zilliacus Riitta 713 M. C. 231 1093 1203

ISSN:0267-9477    

DOI:10.1039/JA9940901305

出版商:RSC    

年代:1994

数据来源: RSC