摘要:

408 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Conferences and Meetings 1987 Winter Conference on Plasma and Laser Spectrochemistry January 12-16, 1987, Lyon, France The 1987 Winter Conference on Plasma and Laser Spectrochemistry will feature recent developments in this field. The conference topics will include the various types of plasmas (ICP, MIP, DCP and GDL) and hyphenated methods such as ICP-MS, chromatography, flow injection and Fourier transform spectroscopy. A symposium will be devoted to different aspects of laser spectrochemistry (laser induced atomic fluorescence, intracavity laser absorption, laser enhanced ionis- ation spectrometry). The Conference will cover fundamental aspects, technological developments and applications, along with manufacturer seminars.Activities (plenary lectures, poster presentation, manufacturer information stands and accommodation) will all take place at the same location, so that a fruitful exchange can occur between recognised experts, users and customers. The official language of the Conference will be English. A large number of ple- nary lectures will allow the presentation of the state of the art in plasma and laser spectrochemistry. Although some oral presentations will be accepted, poster presentations are highly recommended in order to facilitate exchange of informa- tion and to overcome language problems. The presentation of high quality posters will be encouraged by several awards (compact disc players and other awards). The companies involved in plasma and laser spectrochemistry will have the opportunity to participate in the exhibi- tion and to deliver seminars in the form that they think the most appropriate.They will also have the opportunity to organise customers’ meetings. For further information contact J. M. Mermet, 1987 Winter Conference, Lab- oratoire des Sciences Analytiques, Bat. 308, Universite Claude Bernard-Lyon I, 43 Boulevard du 11 nov. 1918, 69622 Villeurbanne Cedex, France. 38th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy March 9-13,1987, Atlantic City, NJ, USA The 38th Pittsburgh Conference, together with its associated exposition of modern laboratory equipment, will be held in the Atlantic City Convention Centre and the adjacent Atlantis and Trump Plaza Hotels. The Scientific Programme will include sessions on Computer-aided Microscopy and Analysis, Occupational Health and Safety in the Laboratory, Reflectance Infrared Spectroscopy, Instrumentation and Automation of Environmental Sample Analysis, Nuclear Magnetic Resonance in Solids, Multi- dimensional Separations, The Analytical Chemistry Opportunity in Process Instrumentation, Hybrid Analytical Techniques Involving Surface Analysis, Future Directions in Mass Spectrometry, Chemometrics in the Computer Inte- grated Laboratory, Stationary Phase Structure and Retention in Reversed Phase Liquid Chromatography, New Developments in Fourier Transform Mass Spectrometry, Separation Sciences and Technology: Metals, LC - MS and SFC - MS, Advances in Raman Spectro- scopy, Detectors for LC and SFC, Immo- bilised Reagents in Chemical Analysis, Quality and Productivity in the Analytical Laboratory and Inductively Coupled Plasma Mass Spectrometry.For further information contact The Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy Inc., 437 Donald Road, Pittsburgh, PA 15235, USA.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 409 Users’ Meeting: Plasma Spectrometry March 2627, 1987, Dortmund, FRG A meeting will be held at the Institut fur Spektrochemie und angewandte Spek- troskopie, in collaboration with the “Deutsche Arbeitskreis fur Angewandte Spektroskopie (DASp) ,” of users and manufacturers of instruments for plasma spectral analysis (ICP, CMP, DCP, MIP) at which participants can report on their experiences and problems and exchange information.The intention is that partici- pants will be able to give short talks, which will be discussed according to topic area. The meeting begins on 26th March 1987 at 2 p.m. and ends on 27th March 1987 at 1.30 p.m., and there is no charge. On both days lunch will be available in the canteen of the Institut fur Arbeitsphysilogie, which is near the Institute. Hotel reserva- tions can be made through the Dort- munder Verkehrsverein (Quartiernach- weis), Konigswall 18, 4600 Dortmund 1 (telephone 0231 140341). Applications and proposed titles of brief presentations should be submitted by 1st March 1987. For further informa- tion contact Dr. J. A. C. Broekaert, Institut fur Spektrochemie und ange- wandte Spektroskopie, Postfach 778, D-4600 Dortmund 1, FRG.XXV Colloquium Spectroscopicum Inter- nationale June 21-26, 1987, Toronto, Canada The XXV CSI will be held at the Hilton Harbour Castle, Toronto, Canada. This North American CSI is sponsored by the Spectroscopy Society of Canada, the Society for Applied Spectroscopy (USA) and the National Research Council of Canada. Nobel Laureates Dr. Gerhard Herz- berg and Professor Arthur L. Schawlow will each present a plenary lecture. Invited lectures on current research topics will be given by approximately 35 young spectroscopists who are making major contributions to the field of atomic and molecular spectroscopy, including: N. Armstrong (Univ. of Arizona), G. I. Bekov (Academy of Sciences, USSR), T. Berthoud (Centre d’fitudes Nucleaires, France), M.Blades (Univ. of British Columbia), M. A. Bolshov (Academy of Sciences, USSR), J. A. C. Broekaert (Inst. f. Spektrochemie & Angewandte Spektros- kopie, FRG), D. C. Compton (Standard Oil Company), G. De Loos (Lab. voor Analytische Scheikunde, The Nether- lands), N. J. Dovichi (Univ. of Wyom- ing), R. Garrel (Univ. of Pittsburgh), J . M. Harris (Univ. of Utah), J. A. Holcombe (Univ. of Texas), D. E. Honigs (Univ. of Washington), S. Houk (Iowa State Univ.), B. Huang (Chang- chun Inst. of Applied Chemistry, China), T. Imasaka (Kyushu Univ. of Japan), K. Kitagawa (Nagoya, Japan), L. B. McGown (Okalahoma State Univ.), J. W. McLaren (National Research Coun- cil Canada), R. Miller (Unilever, UK), J. M. Ramsey (Oak Ridge National Lab.), J. P. Reilly .(Indiana Univ.), A.Scheeline (Univ. of Illinois), D. C. Schram (Philips Research Labs. , The Netherlands), R. Sturgeon (National Research Council Canada), T. Vo-Dinh (Oak Ridge National Lab.), I. M. Warner (Emory Univ.) and E. S. Yeung (Iowa State Univ.). For further information on the pro- gramme contact Dr. J. D. Wineforder, Department of Chemistry, University of Florida, Gainesville , FL 32611, USA. Symposia are planned for after the Colloquium, with four confirmed to date: ICP-MS; Line Spectra of the Elements; Graphite Furnace Atomic Spectroscopy; and FT and Raman Spectroscopy. There will be an exhibition of scientific instrumentation, services and publica- tions. For exhibition information contact either Dr. Andrew T. Zander, Perkin- Elmer Corporation (MS905), 761 Main Avenue, Norwalk, CT 06859-0905, USA or Dr.Andrew W. Boorn, Sciex, Incor- porated, 55 Glen Cameron Road, Thorn- hill, Ontario L3T 1P2, Canada. A social programme is being prepared and will include a dinner, receptions and tours; Wednesday June 24th is an excursion day. For any further information, including registration, contact Mr. L. Forget, Conference Services Office, National Research Council Canada, Ottawa, Ontario K1A OR6, Canada. Inductively Coupled Plasma Mass Spec- troscopy: Post-CSI Symposium June 28-30, 1987, Lake Muskoka, Canada A symposium, sponsored by the Spectro- scopy Society of Canada in conjunction with the XXV Colloquim Spectroscopi- cum Internationale will be held on appli- cations and development of ICP-MS as an analytical technique immediately follow- ing the XXV CSI being held in Toronto, June 21-26, 1987.The symposium will take place at the Muskoka Sands Inn, a resort and conven- tion centre two hours drive north of Toronto on scenic Lake Muskoka. Jim McLaren (National Research Council of Canada) and Chris Riddle (Ontario Geo- logical Survey) are the Symposium Chair- persons. Six invited speakers will present their latest development work in the opening session: Don Douglas (SCIEX), Alan Gray (University of Surrey), Gary Hor- lick (University of Alberta), Sam Houk (Iowa State University), Jean-Michel Mermet (Universitk Claude Bernard, Lyon) and George Vickers (Indiana Uni- versity). Short original “applications- oriented” papers are being sought from active ICP-MS users for the afternoon session of the first day.It is intended that all symposium presentations shall be original and of high quality as refereed publication of the proceedings will follow the symposium. Paper titles are requested now, a working abstract by April 1987 and the full text of the paper is required by June 12, 1987. The second day will involve break-out discussion groups with the invited speak- ers and a final, moderated panel discus- sion on the future of ICP-MS. The total cost, including registration, will be approximately $450.00 Canadian ($350.00 US) per person and includes accommodation (2 nights, double occu- pancy) with full board, an opening night dinner-cruise on Lake Muskoka, a lake- side barbecue, all symposium literature and a copy of the published proceedings. The day following the symposium is Canada Day, a national holiday, and for those wishing to stay on a conference room rate will be available.Spouses and family members will find plenty to do at the Muskoka Sands. To ensure a productive climate, and due to the physical limitations of the Muskoka Sands, registrations will be limited. To reserve your place and to be placed on the mailing list for further announcements please contact Dr. Chris Riddle, Chief Analyst, Geoscience Lab- oratories, Ontario Geological Survey, 77 Grenville Street, Room 1117, Toronto, Ontario M7A 1W4, Canada. Graphite Furnace Atomic Absorption: Post-CSI Symposium June 28-July 2, 1987, Huntsville, Ontario, Canada A conference dealing with various aspects of graphite furnace atomic absorption will be held following the XXV CSI meeting.Three full days of talks and panel discus- sions are planned in addition to selected opportunities to enjoy the surrounding countryside. A number of the world’s leading researchers in GFAA have been invited to present their work. Two open panel discussion sessions are included in the programme to permit topical coverage which may be of interest to the attendees but not covered in enough detail during the preceding day. These sessions have been designed ( a ) to encourage participation by all who are in attendance and (b) to move in the direc- tion dictated by the interests of the audience. The meeting will be flexibly structured to allow ample time for discussion amongst the speakers and conferees. The slate of invited speakers and topics will broadly cover the field of graphite furnace atomic absorption.The areas explored by the meeting will include: (1) fundamental processes occurring in the furnace with their applications and implications toward analyses; (2) new analytical approaches410 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 for GFAA (e.g. , direct solids analysis and absolute analysis); and (3) future analy- tical prospects for GFAA. Panelists and speakers include the fol- lowing invited scientists: H. Berndt (FRG), A. A. Brown (UK), C. Chakra- barti (Canada), H. T. Delves (UK), K. Dittrich (GDR), H. Falk (GDR), W. Frech (Sweden), J. M. Harnly (USA), J. A. Holcombe (USA), C. Huie (USA), K. W. Jackson (Canada), T. Kantor (Hungary), S. R. Koirtyohann (USA), U. Kurfurst (FRG), R.Lovett (USA), E. Lundberg (Sweden), B. L’vov (USSR), N. J. Miller-Ihli (USA), G. Muller-Vogt (FRG), H. Ortner (Austria), C. J. Rademeyer (RSA), T. Rains (USA), G. D. Rayson (USA), D. E. Shrader (USA), D. Siemer (USA), W. Slavin (USA), J. Sotera (USA), R. E. Sturgeon (Canada), D. Styris (USA), V. Sychra (CSSR), G. Tessari (Italy), B. Welz (FRG) and W. Wend1 (FRG). There is a $100 (CND) registration fee and accommodation will be approxi- mately $116 (CND) per night (single) or $96 (CND) per night (double occupancy). These prices include lodging, gratuities and meals during the conference. Group social functions as well as dining on-site have been made an integral part of the three days to maximise the opportunity for interactions between conferees.A mixer will be held Sunday evening and a group social event is planned for Wednes- day evening. Hidden Valley Resort Hotel is located on Peninsula Lake and is equipped with jogging trails, boating, tennis courts and swimming pools. With the exception of racquet sports and boat rental, all facilities are available free of charge to registered guests of the hotel. Hotel and meeting room space is limited and registrations will be accepted on a first-come first-served basis. Regis- tration information will be contained in the final CSI bulletin to be distributed in January, 1987 or more information can be obtained by writing to Dr. James Holcombe, Department of Chemistry, University of Texas, Austin, Texas 78712, USA or Dr. Ralph Sturgeon, Division of Chemistry, National Research Council of Canada, Ottawa, Ontario K1A OR9, Canada.Second Surrey Conference on Plasma Source Mass Spectrometry July 6-8, 1987, Guildford, Surrey, UK This is an advance notice of the Second Surrey Conference, Short Course and Workshop. Further details are available from Dr. A. L. Gray, Department of Chemistry, University of Surrey, Guild- ford, Surrey GU2 5XH, UK. Second Beijing Conference and Exhibition on Instrumental Analysis Conference October 20-23, 1987; exhibi- tion October 19-25, 1987, Beijing, China The objective of the Second BCEIA is to promote academic exchanges on instrumental analysis and friendly rela- tionships between scientists of various countries and technical and trade co-oper- ation between Chinese and foreign com- panies.Symposia will be held on electron microscopy, mass spectrometry, spectro- scopy, chromatography, radio- and microwave spectroscopy and electro- analytical chemistry. The Symposia will cover theories of analysis, new methods and techniques in instrumental analysis, research and development on instrumen- tation and their applications to industry, agriculture and all other areas. The official language of the conference will be English. Unpublished papers covering the areas given above are invited. Accepted abstracts (two pages) will be compiled and published in English in book form prior to the Conference. Authors should submit titles to the BCEIA General Service Office before December 31,1986. The Second Circular, pre-printed abstract form, registration form, hotel reservation form and post- conference tour booking form will be available in January 1987. During the Conference a large scale exhibition of scientific instruments will be held, in which companies from all over the world will be exhibiting their latest products on electron microscopy, mass spectrometry, spectroscopy, chromato- graphy, radio- and microwave spectro- scopy and electroanalytical chemistry. Space will be provided for exhibitors to hold technical seminars and business talks. The space for the exhibition and seminars will cover an area of 10 000 m2. The stand rents range from US$2 500 to US$5000 per unit. Interpreters will be available for an extra fee for exhibitors. The Conference Registration Fee is US$150, which will include admission to the conference and exhibition, a welcom- ing reception, entertainment, a copy of the abstract book containing all contri- buted papers and souvenirs. There will be an accompanying persons programme arranged by the Beijing Tour Agency. For further information contact the General Service Office, Second BCEIA, Room 4205, South Bldg. , Beijing Exhibi- tion Center Hotel, Beijing, China; telex: 20056 BCEIA CN.

ISSN:0267-9477    

DOI:10.1039/JA986010408b

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

410 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Future Issues will Include- Determination of Metals in Poly(viny1 chloride) by AAS-Miguel A. Belarra, Jesus M. Anzano, Felix Gallarta and Juan R. Castillo Laser Excited Fluorescence (LAFS) as a Real Analytical Method. Part 2. Investi- gation of a Carbon Tube Atomiser for the Determination of In, Ga, Al, V and Ir Traces by LAFS-Klaus Dittrich and Hans- Joachim Stark Studies of a Low-noise Laminar Flow Torch for ICP-AES. Part 2. Noise Power Studies and Interference Effects-John Davies and Richard D. Snook Self-matrix Effects as a Cause of Calibra- tion Curvature in ICP-AES-Michael H. Ramsey, Michael Thompson and Stephen J. Walton Direct Atomic Spectrometric Analysis by Slurry Atomisation. Part 1. Optimisation of Whole Coal Analysis by ICP-AES- Les Ebdon and John R.Wilkinson Effect of Torch Size on an 18 MHz Inductively Coupled Plasma-Bryan D. Webb and M. Bonner Denton KPIKa Ratios for Rare Earth Compounds Using Radioisotope Induced X-ray Flu- orescence-K. J. Borowski, F. s. Tham and J. L. Preiss Spray Deposition versus Single Drop Deposition for Calibration of an Electro- static Accumulation Furnace for Elec- trothermal Atomisation AAS-G. Torsi and F. Palmisano Atomic Absorption Spectrometric Deter- mination of Lead by Generation of Its Covalent Hydride-J. Aznarez, J. C. Vidal and R. Carnicer Population Distribution of Atomic Uranium in the Afterglow of a Pulsed Hollow Cathode Discharge-Y. Demers, J. M. Gagne and P. Pianarosa Silicate Rock Analysis by Energy Disper- sive XRF Using a Cobalt Anode X-ray Tube. Part 2. Practical Application and Routine Performance in the Determina- tion of Cr, V and Ba-P. J. Potts, P. C. Webb, J. S. Watson and D. W. Wright Langmuir Probe Measurements of Plasma Potential and their Correlation with Mass Spectral Characteristics in ICP-MS-A. L. Gray, R. S. Houk and J. G. Williams Atomic Spectrometry Update The Update in the February issue is- Environmental Analysis-Les C. Ebdon, Malcolm S. Cresser and Cameron W. McLeod

ISSN:0267-9477    

DOI:10.1039/JA9860100410

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 411 Applications of Spark-source Mass Spectrometry in the Analysis of Semiconductor Materials A Review Jozef Verlinden,* Renaat Gijbels and Freddy Adams Department of Chemistry, University of Antwerp (U.I.A.), B-2610 Wilrijk, Belgium Introduction Bulk analysis Sample preparation Qua I ita t ive an a I ys is Quantitative analysis Detection limits Determination of carbon, nitrogen and oxygen Analysis of layers Analysis of microsamples Keywords: Review; semiconductors; trace analysis; spark-source mass spectrometry Introduction The progress in materials science and the rapid growth of modern technology over the past decades has led to the production of a variety of high-purity materials and conse- quently also to a frequent demand for compositional analysis down to the sub-p.p.m.level. At present, for instance, there is a need to characterise a variety of new materials for the microelectronics industry, based on high-purity gallium , arsenic, selenium, tellurium, indium, cadmium, etc. Several analytical techniques such as atomic emission spectrometry (AES) , atomic absorption spectrometry (AAS) , neutron activation analysis (NAA) or activation analysis using other activation modes, secondary ion mass spectrometry (SIMS) and spark-source mass spectrometry (SSMS) are currently employed for this purpose. Ideally, the analytical technique selected should have a panoramic elemental capa- bility with low detection limits and maximum selectivity. None of the above-mentioned techniques, except SSMS, meets all these requirements simultaneously.Emission spectrometry is frequently used for the analysis of 4-5 N purity materials, but for high-purity semiconductors the technique is useful only in combination with pre-concentra- tion procedures. Also, the elemental coverage of OES is limited. Solution techniques such as ICP emission spec- trometry and AAS have lower sensitivity owing to limited sample concentrations and chemical blanks. Atomic absorp- tion spectrometry can be used for analyses down to the sub-p.p.m. level and yields accurate results. The method has no multi-element capability, however, and the number of analyses on a given sample is fixed by the number of elements to be determined. SIMS has low (sub-p.p.m.) detection limits for a large number of elements.The sensitivity, however, depends on the experimental conditions (primary ion type, reactive gas pressure above the sample surface, secondary ion mode) and when these are kept constant the sensitivities vary over 4-5 orders of magnitude from one element to another. The result is that even semi-quantitative analysis is difficult to achieve and the detection limits are poor for some elements. On the other hand, SIMS is the method of choice for sensitive * Present address: Metallurgie Hoboken-Overpelt, B-2710 Hoboken, Belgium. surface, lateral and in-depth analyses of semiconductors and other materials. Values much below the p.p.b. level can be determined by activation analysis. Doping ranges of 1014 atoms ~ m - ~ (i.e., 2 p.p.b.a.) or lower are not unusual in silicon and for such measurements the method is most suitable.These favourable detection limits can be attained, however, only for specific elements whereas for other impurities the method is less effective. This technique also appears to be less attractive if the matrix gives rise to excessively active radionuclides. Other methods that are promising for semiconductor analysis are inductively coupled plasma mass spectrometry (ICP-MS) or emission spectrometry (ICP-AES) , laser source mass spectrometry (LSMS) and glow discharge mass spec- trometry (GDMS). Until now, however, very few applications of these techniques in semiconductor analysis have been reported. It was the demands of semiconductor technology that stimulated the initial developments of SSMS1 and the tech- nique was first employed in 1954 for the determination of impurities in germanium, antimony and silicon.2 In 1959, the first commercial instruments became available,3 but after a period of rapid development and applications in the 1960s, activities have declined.The reason for this resides in some drawbacks of the method: it takes a long time to carry out an analysis down to the p.p.b. level, i.e., from 1 to 2 d are required per analysis; when no standards are available the results are accurate only to within a factor of 3, in which event the method is only semi-quantitative; the precision with photoplate detection is usually 30% and even worse if the impurities are heterogeneously distributed; and the equip- ment is expensive and needs highly trained personnel. On the other hand, the method has some unique features that make it most useful for various applications, namely, all elements (H-C1 or Li-U) can be detected simultaneously; some elements, especially at high mass, will be below the detection limit because of lower sensitivity and others, e.g., tantalum, are difficult to determine at low concentrations because they are used in the construction of the source; the method has a nearly uniform sensitivity for all elements; very low (p.p.b.) detection limits can be attained; the mass spectral signal is linearly proportional to the concentration over at least five orders of magnitude; the sample preparation is simple with minimum risk of contamination; milligram amounts of sample can be analysed; no pre-knowledge is required of the412 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL.1 sample type, the elements to be analysed or their concentra- tion range; and carbon, nitrogen and oxygen, which are difficult to measure at low concentrations by other methods, can be determined down to the sub-p.p.m. level when special precautions are taken. Hence, for panoramic survey analysis SSMS has, as yet, not been surpassed by any other analytical technique and there- fore it is expected that the method will remain one of the main tools for the analysis of high-purity materials in spite of its drawbacks. The only manufacturer of the instrument today is Japan Electronics and Optics Laboratories, JEOL (Japan), and renewed interest in this technique in industrial labora- tories is reflected by increasing sales figures.4 A recurrent problem facing the analyst in the microelec- tronics industry is the detection and determination of impuri- ties present on the surface and in thin active regions of microelectronics devices.Indeed, during various stages of the fabrication process surface impurities may diffuse into junc- tion regions and degrade the performance of the device. Because in SSMS the spark transfers energy to a small volume of the specimen, the method can also be used for localised analysis. The lateral resolution is inferior to that obtained in SIMS and Auger electron spectroscopy. However, it has detection limits 3-5 orders of magnitude better than those of Auger electron spectroscopy and for a number of elements also better than those in SIMS, especially with thick layers or large spots on the surface.Moreover, surface roughness does not impede the analysis. Hence SSMS may serve as a unique link between macro- and micro-analytical techniques. In this review, the applications of SSMS in the bulk, layer, surface and microsample analysis of semiconductor materials are discussed in order to demonstrate the possibilities of the method with regard to their chemical characterisation. Also, applications to the analysis of gallium and indium, which are important for the preparation of some microelectronics materials, are considered. For a comprehensive discussion of the technique itself reference is made to the literature.4-6 Bulk Analysis Sample Preparation For the analysis of semiconductors minimum sample prepara- tion is required.The cutting and cleaning of pieces of material to be used as electrodes, followed by pre-sparking in the ion source under vacuum, is common practice. Procedures have been described for cleaning the surfaces of GaAs,7-10 Si,11J2 GellJ2 and InP9 electrodes. Typical electrode dimensions are length 10 mm, width 1-2 mm and thickness 1-2 mm. After mounting the electrodes in the ion source and vacuum pumping, suitable spark parameters should be selected allowing stable sparking to be produced between the elec- trodes. No difficulties have been reported in achieving this for any semiconducting material. In general, the analysis itself is preceded by pre-sparking of the electrodes to remove surface contaminants and oxides from the electrodes; further, this reduces or even eliminates memory effects from samples previously sparked by coating the spark housing, electrode holders and slits with a layer of the matrix under study.Good laboratory practice would imply the use of a different ion source chamber, electrode holders and accelerating slits for different matrices. Because of its low melting-point (30 “C), special electrode preparation is necessary for gallium, in order to prevent the electrodes from fusing together while sparking. Ahearn13 dipped high-purity silicon electrodes in liquid gallium. A small amount of the latter adheres and the major components of the mass spectrum of gallium and silicon are recorded. Brown et a l l 4 have analysed gallium by supporting it on the end of high-purity graphite electrodes. Wolstenholme~~ used a modi- fied ion source for cooling the gallium samples, such that, via two flexible metal strips, the heat generated during sparking is conducted to a Dewar vessel containing liquid nitrogen.Since then, cooling of the samples has become a general practice in the analysis of gallium.1624 A procedure for the preparation of gallium electrodes was described by Nalbantonglu.18 The liquid gallium is transferred into a cylindrical hole in a PTFE mould and cooled in liquid nitrogen. The method works well but it cannot be entirely excluded that surface contaminants are introduced. Therefore, the same author proposed another method19 in which the gallium is brought into a graphite crucible and sparked against a pure graphite counter elec- trode.Nyari and Opauszky20 loaded the gallium sample in a specially shaped gold spoon and used a gold wire as a counter electrode. Shabanova et a1.21 used the “frozen drop” method; the sample is molten at low pressure and then quenched in order to avoid the redistribution of impurities by segregation. The frozen drop is then sparked against a suitable counter electrode. For indium (melting-point 156 “C), electrodes can be cut and sparked as such. However, cooling of the sample is recommended as this results in a smoother surface and therefore a more stable spark.22 The frozen drop method has also been applied to this matrix.23 Powdered samples that cannot be pressed into strong electrodes are usually mixed with a binder.For silicon powder ultrapure graphite24 and gallium25 have been used as binders. For the bulk analysis of solids two electrodes of the sample itself are sparked against each other and only exceptionally is a counter electrode or substrate material used. Qualitative Analysis After mounting the electrodes in the sample holders and vacuum pumping, the next step is to produce a spark. The ion source provides a repetitive series of high-voltage breakdowns between the electrodes. It uses an oscillating potential difference of 20-100 kV at 0.5-1 MHz, which is generated in short pulse trains of variable length (20-200 ps) and with a repetition frequency of 1-104 s-1. The sparking parameters are chosen on an empirical basis so as to achieve a stable spark with a satisfactory level of ion generation.For materials with a low melting-point the spark parameters should be kept as moderate as possible in order to prevent the samples from overheating. Further, smaller differences in elemental sensitivity are encountered at lower spark voltages. In nearly all bulk analyses the ion-sensitive emulsion (photoplate) was used as the detector. A typical mass spectrum shows lines due to ions of various natures: singly charged ions, multiply charged ions, charge-exchange ions, polyatomic ions (An+), heterogeneous compound ions and residual gas ions. Fig. 1 shows part of the mass spectrum of an arsenic matrix illustrating the occurrence of some of the ion types mentioned.26 The singly charged ion lines, correspond- ing to the isotopes of the matrix or impurity elements present in the sample, are by far the most abundant and normally the only ones used for analytical purposes.The other lines are important only because they may interfere with lines of interest. When the singly charged line suffers interference, the analysis may be based on the doubly or triply charged ions of the impurity element or, when the spectrum is well known, the degree of interference may be calculated. In this respect it may be of interest to refer to a study by Becker and Dietze27 on the occurrence of some molecular ions in the spectrum of silicon and germanium: both elements have a tendency to form a series of polyatomic ions. The mass spectra of the matrices discussed in this paper have been shown and commented on in various papers2~13~18,19,*4,26,2~33 and a list of possible interfer- ences from matrix lines has also been re~orted.22,2~?26,31,32 In semi-quantitative analysis, it is usually intended to detect the impurities in a short time and to calculate the detection limits for impurities not observed.This is mostly carried out byJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 413 0 10000 20000 30000 qO000 50000 60000 70000 80000 90000 100000 I I I 1 I I 1 10 15 20 25 30 35 40 Mass x v) c a c t -c - 100 + nl NO e-2 I00 110000 120000 130000 lclO000 150000 160000 170000 180000 190000 200000 1 1 1 I I 1 I I I I I I I 45 50 55 60 65 70 75 80 85 90 95 100 105 Mass 0 cd I I I 1 I I I I 1 I 110 120 130 140 150 160 170 180 190 200 Mass x v) t a c c c - In" a 0 m 2 I I I I I I I I I I I I I 300000 310000 320000 330000 3 q O O O O 350000 360000 210 220 230 240 250 260 270 Mass Fig.1. Mass spectrum of a typical arsenic sample obtained at 600 nC. Mass range 0-40, 4&110, 110-205 and 205-28026414 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 comparing the photoplate of interest with a calibration plate. Concentrations are calculated by comparison of the impurity line with a matrix line or with the line of an impurity element of known concentrations. Visual determination usually yields uncertainties of up to a factor of 3 for the experienced eye; differences in sensitivity are usually not taken into account. Quantitative Analysis For quantitative multi-element analysis, accuracy is the main concern.For such analyses successive exposures that usually vary from 0.0001 to 1000 nC in a 1 : 3 : 10 series are obtained on a photoplate. After development, line blackening is converted into the number of ions striking the plate following linearisa- tion procedures described in the literature .5J4 A recording microphotometer is essential for quantitative measurements of the position (mass) and blackening of the lines. Rapid data reduction from plates with a large number of lines can only be accomplished when the microphotometer is interfaced to a computer. Once the ion intensities have been determined, it is assumed that the value obtained is proportional to the concentration of the element. In bulk analysis it is found that the proportionality factors are about the same within a factor 0.3-3 for most elements.This is illustrated in Table 1 for elements in a GaAs and In matrix.22JS From Table 1 it follows that quantitative analysis based on a simple comparison of the impurity lines with the line of an element of known concentra- tion such as a matrix line or an impurity determined by another method may lead to considerable inaccuracy. Hence, accurate results can only be obtained if these factors, called relative sensitivity factors (RSF), are known with a reasonable degree of accuracy. Table 1. Relative sensitivity factors for elements in GaAs and In matrices22.35 Element GaAs In Cr . . . . . . . . 1.6 Fe . . . . . . . . 1.5 Ni . . . . . . . . c u . . . . . . . . 1.3 Zn . . . . . . . . 1.2 Cd . . . . .. . . In . . . . . . . . Sn . . . . . . . . 1.8 Sb . . . . . . . . Te . . . . . . . . 1.0 T1 . . . . . . . . Pb . . . . . . . . Bi . . . . . . . . - - - - - - - 1.6 0.47 0.47 0.77 1.5 1 .o 0.69 0.67 4.6 1.3 2.0 - - Relative sensitivity factors are usually determined by analysing a standard sample. As the mass spectrometric response is linearly proportional to the impurity concentration over 5-6 orders of magnitude,lOJ6 one standard sample is sufficient for calibration. Unfortunately, no multi-element standard reference materials are available for semiconductor analysis. The solution is then to use self-prepared standards or samples in which the impurity concentrations were deter- mined by other methods such as AAS, OES or others. A mass spectrometric approach yielding accurate and precise results without reference samples is the isotope dilution technique.However, this method is very time consuming. Gauneau et aZ.37 used 0.3 mm thick ion implanted InP and GaAs samples as standards. In these particular experiments Be and B were implanted but the method can be extended to cover a wide range of elements. With proper standards fairly accurate results, within 30% of the true value, have been reported for various elements in GaAs10J8 and In.22 In many instances, however, the results are reliable only within a factor of 2, even after calibration. This is mostly due to a heterogeneous distribution of elements in the sample. Also, relative sensitiv- ity factors are dependent on various experimental parameters4 and for accurate results the experimental conditions at which the relative sensitivity factors were determined should be reproduced as well as possible for each new analysis. In Table 2 reports dealing with the semi-quantitative and quantitative analysis of semiconductors and the metals gallium and indium are summarised.Detection Limits The main feature of SSMS in the analysis of high-purity materials is its high detection power. In general, the p.p.m. level is reached at exposures of 1-10 nC, i.e., after only ca. 5 min of sparking. The most important factors limiting ultimate sensitivity are an intense halo on the photoplate, which extends at the high mass side of the matrix lines, and a general, more uniform blackening of the entire spectrum. Other limitations are spectral interferences and for the determina- tion of H, C, N and 0 the residual gas pressure in the ion source.The halo is due to a secondary emission process: as the ions strike the plate with kinetic energies of 20-30 keV, secondary positive ions are sputtered away, and are deflected by the magnetic field to other photoplate locations. As these secondary ions have much lower energy than the primary ions, their penetration into the emulsion is restricted to a relatively superficial layer. The matrix lines receive a very high ion flux and build up a considerable charge, with the result that further incoming ions are deflected away to higher mass; this is a second effect contributing to the halo. The background Table 2. Reported analyses of semiconductors and of the metals Ga and In Qualitative and semi-quantitative Quantitative Matrix analysis (RSF) C, N and/or 0 As .. . . . . . . 26,32,65 CdTe . . . . . . 31 Ga . . . . . . . . 15-19,60 17 16,20 GaAs . . . . . . 8,29,31,39-44 7,10,35,38,45,46 7-10,4653 GaP . . . . . . . . Ge . . . . . . . . In . . . . . . . . InAs . . . . . . InP . . . . . . . . Sb . . . . . . . . Si . . . . . . . . Te . . . . . . . . 35,45 2,11,19,28,31,42, 12,53 60,63,67,68 56,60 22,61,62 37 9,12,53 2,19,28,60 2,11,24,28,29,31, 12,19,25,35,53,57 12,31,49,53,57, 33,41,42,56,69 58 66 Comparison with other methods 26,42,64 21,42,70 7-10,29,37-39, 42,45,46,54 45 12,42,53 22,42 9,12,37,53 12,29,33,42,49, 53,57,59 64,66 Detection limits 26,42,64 4,18,21,42 8,28,29,33,37,39, 42,50,51,54,55 13,51 11,12,31,42,53 22,23,42 55 37 11,24,25,28,39,42, 59 64JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL.1 Table 3. Detection limits (p.p.m.a.) in various matrices 415 Element Li . . . . Be . . . . B . . . . F . . . . Na . . . . Mg . . . . A1 . . . . Si . . . . P . . . . s . . . . c1 . . , . K . . . . Ca . . , . sc . . . . Ti . . . . v . . . . Cr . . . . Mn . . . . Fe . . . . co . . . . Ni . . . . c u . . . . Zn . . . . Ga . . . . Ge . . . . As . . . . Se . . . . Br . . . . Rb . . . . Sr . . . . Y . . . . Ga4 0.001 0.001 0.001 0.002 0.02 0.002 0.002 0.002 0.002 0.002 0.003 0.002 0.002 0.002 0.003 0.003 0.003 0.003 0.003 0.002 0.004 0.004 0.006 N.a. 0.009 0.003 0.006 0.006 0.02 0.01 0.003 GaAs55 0.005 0.01 0.001 0.003 1 0.03 0.003 0.01 0.003 0.05 0.003 0.01 0.01 0.003 0.005 0.003 0.01 0.003 0.002 0.001 0.01 0.002 0.002 N.a.0.3 N.a. 0.2 0.06 0.05 0.1 0.003 Gel1 0.01 0.01 0.002 0.003 0.01 0.2 0.01 0.01 0.01 0.05 0.01 0.01 0.01 0.02 0.01 0.002 0.1 0.1 0.1 0.1 0.2 0.2 1 N.a. 1 1 0.01 0.01 0.01 - - Sill 0.003 0.01 0.003 0.003 0.01 0.01 0.01 N.a.* 0.05 0.01 0.01 0.01 0.01 0.01 0.01 0.01 1 0.03 0.1 0.05 0.005 0.005 0.006 0.01 0.003 0.006 0.006 0.3 0.02 - - Element Zr . . . . Nb . . . . Mo . . . . Ru . . . . Rh , . , . Pd . . . . Ag . . . . Cd . . . . In . . . . Sn . . . . Sb . . . . Te . . . . I . . * . c s . . . . Ba . . . . La . . . . Hf . . . . Ta . . . . w . . . . Re . . . . 0 s . . . . Ir . . . . Pt . . . . Au . . . . Hg . . . . T1 . . . . Pb . . . . Bi . . . . Th . . . . u . . . . Ga 0.007 0.004 0.02 0.01 0.003 0.01 0.007 0.01 0.004 0.01 0.007 0.01 0.004 0.004 0.04 0.04 0.02 0.03 0.008 0.01 0.008 0.02 0.02 0.02 0.007 0.01 0.02 0.006 0.006 - GaAs 0.01 0.003 0.02 0.02 0.03 0.02 0.006 0.02 0.005 0.01 0.006 0.01 0.003 0.003 0.02 0.003 0.01 0.3 0.004 0.002 0.003 0.002 0.003 0.001 0.004 0.005 0.002 0.003 0.001 0.001 Ge 0.02 0.01 0.03 0.01 0.005 0.01 0.01 0.003 0.001 0.003 0.002 0.003 0.001 0.001 0.1 0.03 0.003 0.002 0.002 0.003 0.001 0.003 0.003 0.01 0.01 0.001 0.001 - - - Si 0.006 0.003 0.01 0.01 0.003 0.01 0.006 0.03 0.03 0.003 0.002 0.003 0.001 0.001 0.002 0.001 0.003 0.03 0.003 0.002 0.002 0.002 0.003 0.01 0.01 0.001 0.002 0.001 0.001 0.001 * Not applicable.Table 4. Possibilities of multi-element analysis by SSMS, NAA and spectrochemical (SC) analysis in silicon, gallium arsenide and germanium4* Number of impurities determined Analytical Detection method limit, YO Silicon Gallium arsenide Germanium SSMS .. . . . . 10-7 38 34 36 10-6 68 67 57 NAA . . . . . . 10-9-10-10 30 9 11 10-8 44 11 15 10-7 50 15 19 10-6 56 18 28 sc . . . . . . . . 10-8 2 2 2 10-7 7 8 8 10-6 20 10 19 darkening is much less when the exposure is taken with a low ion current. For sensitive multi-element analysis it is necessary to reduce or, if possible, to avoid the formation of the halo. In the literature several procedures have been proposed to achieve this g0a1,4~5,71 such as bleaching the emulsion to remove the top layer followed by internal development, selecting appro- priate developers and development temperatures, cutting the plate to prevent the matrix lines from reaching the emulsion, painting a conducting film at positions where the matrix lines appear, placing a magnetic strip behind the plate and using gelatin-free plates.The cutting procedure is often very useful as it allows sub-p.p.m. to p.p.b. detection limits to be attained even for elements near the matrix lines. The second limitation, i.e., the general blackening of the ion-sensitive emulsion, is due to non-focused particles and also to the faint light emission of the mass-separated ion beams when they hit the photoplate. This background only becomes apparent at large exposures of at least a few hundred nanocoulombs, and may be reduced by improving the quality of the vacuum in the analysers. The problem of overlapping element lines is unimportant as, with the exception of indium, all elements have a unique (no isobaric interference) isotope.In general, when working at a mass resolution of ca. 5000, most other interferences are resolved, although line broadening due to overexposure may lead to unexpected overlapping of mass lines. Solutions to this type of problem have been discussed above. In general, detection limits of 10 p.p.b. should be reached for most elements and instances where it is worse than 0.1 p.p.m. are rather rare. Table 3 shows detection limits reported for impurities in the matrices Ga, Ge, Si and GaAs. Table 4 compares the number of elements that can be determined with a given limit of detection in silicon, germanium and gallium arsenide by SSMS, neutron activation analysis and spectro- chemical analysis,@ methods often used for bulk analysis of semiconductors. From Table 4 it follows that SSMS is the method of choice for the multi-element analysis of GaAs.The method also has better sensitivity for many elements in the other matrices, especially in germanium. Karpov and Alimarin42 made a comparison of SSMS detection limits with those of spectrographic, photometric, activation and polarographic analyses for elements in silicon, germanium, indium, arsenic, gallium arsenide and gallium. Similar comparisons were also made with graphite furnace AAS and OES for elements in tellurium and arsenic,64 withJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 416 SIMS for silicon and gallium arsenide,72 with NAA and OES for silicon59 and with OES for gallium ar~enide.5~ Yudelevich et aZ.70 made a comparison of multi-element methods of analysis for high-purity gallium.Zolotov and Grasserbauer73 discussed the typical techniques used for trace analysis of semiconductor materials and detection limits for the various methods were also given. internal wall of the ion source chamber to a sufficiently low temperature leads to condensation of the gases (cryogenic pumping). Liquid nitrogen (80 K) cooled metal surfaces are effective in reducing the pressure of hydrocarbons, water vapour and carbon dioxide but gases such as N2,02 and CO do not adequately freeze out.50>51 Therefore, liquid helium (4 K) pumping is often applied for removing these gases; pressures down to 4 X 10-10 Torr can then be attained.53 In addition it is necessary, after the etching, rinsing and drying procedures, to clean the sample surfaces further by vigorous pre-sparking under high-vacuum conditions._53 Descriptions of helium cryogenic pumps used for measuring low levels of C, N and 0 can be found in the litera- t~re.~7~50~51~53 An exact detection limit for these elements cannot be given as this requires the analysis of a “zero concentration” sample.However, the lowest calibrated con- centrations that have been analysed in GaAs are 2 x 1015 atoms cm-3 of oxygen and 1 x 1016 atoms cm-3 of carbon.7 Sub-p .p.m. concentrations have also been reported for these elements in InP, Ge and GaA~.8,9J2>50>51~53 Clegg et al. 12 have compared detection limits for oxygen with those of other techniques: electrical measurements, vacuum and inert gas fusion, NAA, infrared spectrophotometry and charged-par- ticle and gamma-photon activation analysis.Determination of Carbon, Nitrogen and Oxygen The determination of trace amounts of carbon, nitrogen and oxygen in the sub-p.p.m. range is an important problem in semiconductor research. In routine mass spectrometric analy- sis with conventional equipment there is a large blank for these elements of not less than 10-100 p.p.m. This background is due to the residual gases in the ion source, such as H20, C02 and hydrocarbons, and also to adsorbed surface layers and oxide layers on the sample. In order to reduce both the contribution from adsorbed layers and residual gases, various procedures have been applied. Baking out the source over- night with the sample in position before each analysis leads to ion source pressures of about 10-8 Torr when a conventional oil diffusion pump is used.8351 This pressure may be further reduced by using pumps with a higher speed.48 Cooling an Table 5.Surface and layer analysis of semiconductor materials by SSMS Counter electrode Layer thickness/ ym Depth of 1 crater/ym 5 1-10 0.1 Met hod * XY RS RS Remarks Detection limit 0.1 p.p.m. Ref. 75 50,96 97 Matrix Si . . . . Au point Pt, C, A1 Au point Sample surface coated Distribution profile, Distribution profile, with Cu comparison with NAA multi-layer 5 x 1011 at. cm-2 2 x 10-12 g 8 X 10-l2 g Distribution profile Contamination study Epitaxial layer 0.02 p.p.m.a. Dry residue of ultrapure 10-9-10-12 at. -% Contamination study 0.01 p.p.m.a.Distribution profile Distribution profile Contamination study 1011 at. cm-2 Comparison with ISS Mg contamination Pre-concentration 10-9-10-12at. -yo water 3 x 1011 at. cm-2 Surface 13.2 XY 78 26 XY 78 Surface <2 Surface 10 Surface Surface 0.3 XY Drop XY RS XY XY 78 98 100 101 102 103 Ta (0.04 x 3 mm2) Si (0.37 x 0.052 cm2) 1.7 2 Ta Si A1 Ta (0.5 mm dia.) Ta (0.5 mm dia.) 2.5 9 9.8 Surface Surface Surface XY P.a. RS RS RS 72 92 80,79 80 79 13 0.2 1.7 1.3 1.4 Ge . . . . Graphite Surface 1 Graphite 1 18 Surface Ge point Surface RS 0.1 p.p.m, XY Distribution profile N.a. Sputtered layers RS Contamination study 1012 at. cm-2 XY In monolayer 0.04 monolayer 96 87 104 79 13 GaAs . . . . Au 3 XY A1 (0.05 x 2 mm2) 3-5 0.5-1 XY 1014 at. cm-3 A1 10 XY 10-5 at.-% GeGa(0.35 x 0.3cm2) Surface XY Contamination study 8 x 10-12g RS Quantitative deter- 5-80 XY Distribution profile Surface RS Contamination study 3 x 1011 cm-2 30 RS Comparison with electrical measurements mination of S A1 (0.05 x 2mm2) 3-50 0.5 XY Distribution profile 10-4 at.-% 26 RS Distribution profile 0.05 p.p.m.a.GaP . . . . Point 5-80 XY Three-layer structure * XY = X - Y scan; RS = rotating scan; N.a. = normal analysis; P.a. = point analysis. 72 99 87 100 101 87 94 79 79 79 86JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 417 Analysis of Layers The physical and chemical properties of many materials are dramatically affected by the presence of a thin film on the surface and by the distribution of impurities in the film and in the underlying substrate.Therefore, there is a need for the analytical characterisation of microvolumes in solids. Modern surface chemical instrumentation includes numerous tech- niques, among which Auger electron spectroscopy, X-ray photoelectron spectroscopy (XPS) , ion scattering spectro- scopy (ISS), Rutherford backscattering spectrometry (RBS) and SIMS are the most common. The specific drawbacks of each of the techniques necessitate the combination of several of them in order to gather comprehensive analytical informa- tion about the specimen and to ascertain the results of quantitative analysis. In addition to the techniques mentioned above, SSMS may be very useful when surface layers need to be analysed sensitively for a survey of all elements of the Periodic Table.The method is applicable when the surface roughness prevents the use of other techniques. Depth profiling is possible, especially when relatively thick layers (tens of micrometres) are to be analysed. Applications of SSMS for the layer, surface and in-depth analysis in various materials have been discussed in several reports.4274-77 For layer analysis sample scanning is essential for accumulating sufficient ion charge for semiconductor impurity analysis and limits the depth of penetration to ca. 1 pm. Single pulse analysis would detect only 0.1% impurity levels and without the instrumental modifications described by Derzhiev et al. 77 one pulse would still penetrate the sample to a depth of ca. 1 pm. A variety of geometries have been utilised for layer analysis.Usually a sharply pointed electrode is positioned opposite to the plane surface of the sample electrode which contains the microvolume to be examined (point-to-plane geometry). A series of analyses can be run at any location on the sample to obtain a depth profile. In addition, the sample can be moved in the X - Y directions This can be done either manually or automatically.78 Another method is the rotating scan technique, where the sample in the form of a disc is given both a rotation and a translation movement.51,79980 The specimen is sparked with a fixed counter electrode in such a way that fresh material is sampled and ionised continuously, and a spiral groove is eroded in the disc. In the plane-to-plane geometry a broad counter electrode is sparked against the sample surface. The electrode may have a smaller cross- section than the sample surface, thus allowing a scan to be made,81 or the same cross-section, thus avoiding edge effects .82 Various reports document the influence of the spark parameters, the material and structure of the sample and the counter electrode on the crater dimensions, detection limits, accuracy and depth resolution.74,77983-96 Most of these studies deal with the analysis of semiconductor materials.83-96 The attainable lateral resolution is limited to 10-100 pm, although in some experiments a better resolution has been reported for silicon.97 The depth resolution (depth of one crater) is usually ca.1 Vm or better when automatic scanning is used. A special method for the in-depth analysis of semiconductor layers was proposed by Yudelevich and co-~orkers,98~99 based on chem- ical layer-by-layer etching, evaporation of the etching solution Table 6.Sensitivity with which a mono-isotopic element is determined in silicon layers of various thickness (Layer area 1 cm2; A = lo-’ ) Visual detection, Photometric detection, Layer thicknesdwm at.-% at.-% 3 2 x 10-7 7 x 10-6 1 6 x 10-7 2 x 10-5 5 x 10-4 1 x 10-3 4 x 10-2 0.1 6 x 10-6 2 x 10-4 Monolayer to a minimum volume, adding an internal standard and freezing. The remaining drop is then sparked against a suitable counter electrode. In this way a depth resolution of ca. 0.01 pm has been obtained. Spark-source mass spectrometry can also be used for surface analysis as it enables one to distinguish homogeneously distributed bulk impurities and impurities present on the sample surface.Consequently, mass lines due to surface contaminants decay in intensity as the spark erodes into the electrode. Clegg and Millett79 have shown that it is possible to determine surface contaminants down to 3 x 1011 atoms cm-2, i.e., ca. one thousandth of a monolayer. According to Chupakhin and Ramendik,93 it is possible to detect a surface contamination equivalent to 3 x 10-5 of a monolayer. In practically all SSMS surface studies mechanical scanning is applied in order to integrate ion signals from a sufficiently large surface area. Table 5 summarises applications of SSMS for the analysis of layers and surfaces of semiconductor materials. Both counter electrodes of the same material as the sample and of different metals, including refractory metals, have been used for layer and surface analysis.The use of a counter electrode that easily atomises, such as aluminium, has been proposed if maximum depth resolution is required.81 For the same purpose, in surface analysis, the sample was coated with a thin layer of a pure meta1.97 The detection limits in layer analysis depend on the volume of the layer to be analysed. The minimum concentration, cmin (p.p.m.a.), that can be detected in a layer of thickness h (pm) and area d (pm2) can be calculated from where A is the matrix atomic mass, Z the isotope abundance (YO), N is Avogadro’s number and p is the density of the matrix (g cm-3). A is the useful ion yield, or more precisely the number of atoms that must be atomised in order to deliver one ion at the detector.The value of A depends on the specimen material and on experimental parameters such as breakdown voltage, slit settings and other experimental fa~tors.83J05-10~ In general, A decreases with increasing breakdown voltage and narrower slit widths, with the exception of silicon, where A increases with increasing breakdown voltage.83 Usually, its value lies in the range 10-6-10-8. In the above equation n is the smallest number of ions necessary to give a just detectable line on the photoplate. Its value was estimated to be 3 X lo3 with naked eye detection and l o 5 for photometric detecti0n.9~ Table 6 shows the attainable detection limits in silicon for layers of 1 cm2 area and of various thickness.Detection limits can, of course, deteriorate when the mass spectral contribution of impurities from the counter electrode becomes serious. In this respect Bedrinov et al.88 studied the influence of the atomisation of the counter electrode on the detection limits in Si and GaAs. For layers of sufficient thickness SSMS may be more powerful than SIMS if specific elements are to be detected or in more general panoramic contaminant identification studies.72 Of course, SSMS cannot give the nanometre depth resolution possible by SIMS for the isolation of surface impurities and impurity depth profiling into the bulk of the sample. It is also significant that SSMS consumes approximately 10 000 times more material than SIMS for a given impurity sensitivity.A particular method for the analysis of bulk samples via thin-layer analysis was described by Yudelevich and co- workers.lo~Jo9 The method involves the elimination of the matrix and evaporation of the solution containing the impuri- ties on a specially prepared substrate of high-purity silicon. The silicon with the dry residue layer is then sparked. In this way detection limits were improved by 1-3 orders of magni- tude compared with direct mass spectrometric analysis for impurities in ultrapure water, cadmium and tin. It was suggested that the method can also be applied after matrix418 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 separation by distillation for the matrices tellurium and chlorides of Si, Ge, As and Sb, Quantification is possible after the introduction of a reference element or by applying isotope dilution.Also, heterogeneity effects are minimised by taking a larger sample mass. On the other hand, by using chemical methods for separating the matrix from impurities, the possibility that contaminants are introduced is increased. Analysis of Microsamples Spark-source mass spectrometry has obvious potential advan- tages for the analysis of microsamples because of its high sensitivity, simple sample preparation and multi-element capability. The detection limits depend on the amount of sample available. In principle, microgram samples can be analysed with detection limits at the p.p.m. level.1Os In practice, such analyses are difficult and the smaller is the sample the greater are the problems encountered in holding the sample and sparking it.4 Very few applications of microsample analysis have been reported for semiconductors.Brown et al. 14 analysed 2-mg diodes by inserting them into a split electrode of high-purity graphite. Ahearn13 used a holder and counter electrode of high-purity silicon for the analysis of small silicon crystals and concentrations down to the sub-p.p.m. level were determined. With very small samples it is usually not possible to obtain a series of graded exposures on the photoplate before the sample is entirely consumed. Chastagnerllo developed a single exposure method based on image broadening that allows the estimation of impurity concentrations over a range of four orders of magnitude, Absolute detection limits are very low, ca.10-12 g.111J12 This work was sponsored by the Ministry of Science Policy, Belgium, within research grant PREST UIA/03. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. References Dempster, A. J., Proc. Am. Phil. SOC., 1935, 75, 755. Hannay, N. B., and Ahearn, A. J., Anal. Chem., 1954, 26, 1056. Craig, R. D., Errock, G. A., and Waldron, J. D., Adv. Mass Spectrom., 1959, 1, 136. Ramendik, G., Verlinden, J., and Gijbels, R., in Adams, F., Gijbels, R., and Van Grieken, R., Editors, “Inorganic Mass Spectrometry,” Wiley, New York, 1987, in the press. Ahearn, A. J., Editor, “Trace Analysis by Mass Spec- trometry,” Academic Press, New York and London, 1972. Bacon, J. R., and Ure, A. M., Analyst, 1984, 109, 1229. Brozel, M. R., Clegg, J.B., and Newman, R. C., J. Phys. D , 1978,11, 1331. Brice, J. C., Roberts, J. A., and Smith, G., J. Muter. Sci., 1967, 2, 131. Blackmore, G. W., Clegg, J. B., Hislop, J. S., and Mullin, J. B., J. Electron. Muter., 1976, 5 , 401. Clegg, J. B., Graigner, F., and Gale, I. G., J. Muter. Sci., 1980, 15, 747. Nazarenko, V. A,, in Alimarin, I. P., Editor, “Analysis of High-Purity Materials,” Israel Program for Scientific Trans- lations, Jerusalem, 1968, p. 112. Clegg, J. B., Gale, I. G., and Millett, E. J., Analyst, 1973, 98, 69. Ahearn, A. J., Proc. 10th Colloq. Spectrosc. Int., Washington, DC, 1963, 769. Brown, R., Craig, R. D., James, J. A., and Wilson, C. M., paper presented at the Conference on Ultrapurification of Semiconductor Materials, Boston, April 1961.Wolstenholme, W. A., Appl. Spectrosc., 1963, 17, 51. Fitzner, E., Chem. Rundsch., 1965, 18, 389. Kurthy, J., Acta Chim. Acad. Sci. Hung., 1978, 96, 209. Nalbantonglu, M., Adv. Mass Spectrom., 1966, 3, 183. Nalbantonglu, M., Chim. Anal., 1966,48, 148. Nyari, I., and Opauszky, I., Fresenius Z . Anal. Chem., 1981, 309, 274. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. Shabanova, L. N., Shelpakova, I. R., Gerasimov, V. A., and Yudelevich, I. G., Izv. Sib. Otd. Akad. NaukSSSR, Ser. Khim. Nauk, 1977, 14, 55. Liu, X. D., Verlinden, J., Adams, F.. and Adriaenssens, E., Bull. SOC. Chim. Belg., 1986, 95, 309. Shabanova, L. N., Shelpakova, I. R., and Yudelevich, I. G., Izv.Sib. Otd. Akad. Nauk SSSR, Ser. Khim. Nauk, 1979, 6 , 126. Lasne, C . , Gazet-Talvande, J., and Barraud, Y., Prog. Crystal Growth Charact., 1984, 8, 151. Hurrle, A., and Dietl, J., Fresenius 2. Anal. Chem., 1981,309, 277, Liu, X. D., Verlinden, J., Adams, F., and Adriaenssens, E., Anal. Chim. Acta, 1986, 180, 341. Becker, S., and Dietze, H. J., Int. J. Mass Spectrom. Ion Phys., 1983, 51, 325. Hannay, N. B., Science, 1961, 134, 1220. Kane, P. F., and Larrabee, G. B., Editors, “Characterization of Semiconductor Materials,” McGraw-Hill, New York, St. Louis, San Francisco, London and Sydney, 1970, p. 110. Vidal, G., Int. J . Mass Spectrom. Ion Phys., 1972-73, 10,204. Andreani, A. M., Brun, J. C., Mermoud, J. P., and Stefani, R., Meth. Phys. Anal. GAMS, 1971, 7, 269.Brown, R., Craig, R. D., Elliott, K. M., Adv. Mass Spectrom., 1961, 2, 141. Honig, R. E., in Cali, J. P., Editor, “Trace Analysis of Semiconductor Materials,” Pergamon Press, Oxford, London, New York and Paris, 1964, p. 169. Schuy, K. D., andFranzen, J., Fresenius Z . Anal. Chem., 1967, 225, 260. Ahearn, A. J., in Meinke, W. W., and Scribner, B. F., Editors, “Trace Characterization-Chemical and Physical,” NBS Monograph No. 100, National Bureau of Standards, Washing- ton, DC, 1967, p. 347. Van Hoye, E., Adams, F., and Gijbels, R., Talanta, 1976,23, 789. Gauneau, M., Rupert, A., Minier, M., Regreny, O., and Coquille, R., Anal. Chim. Acta, 1982, 135, 193. Clegg, J. B . , in Makram-Ebeid, S., and Tuck, E. B., Editors, “Semi-insulating 111-V Materials,” Shiva, Nantwich, Cheshire, 1982, p.80. Gauneau, M., Spectra2000, 1979, 7,52. Huber, A. M., Morrilot, G., Merenda, P., andLihn, N. T., in Evans, C. A., Powell, R. A., Shimizu, R., and Storms, H. A., Editors, “Proceedings of the 2nd International Conference on Secondary Ion Mass Spectrometry (SIMS 11) ,” Springer-Ver- lag, Berlin, Heidelberg and New York, 1979, p. 91. Huber, A. M., and Moulin, P., J. Radioanal. Chem., 1972,12, 75. Karpov, Yu. A., and Alimarin, I. P., J. Anal. Chem. USSR, 1979,34, 1085. Look, D. C., J. Appl. Phys., 1977, 48, 5141. Owens, E. B., and Giardino, N. A., Anal. Chem., 1963, 35, 1173. Ahearn, A. J., Trumbore, F. A., Frosch, C. J., Luke, C. L., and Malm, D. L., Anal. Chem., 1967,39, 351. Martin, G. M., Jacob, G., Hallais, J. P., Grainger, F., Roberts, J.A., Clegg, J. B., Blood, P., and Poiblaud, G., J . Phys. C, 1982, 15, 1841. Walters, D. C., and Look, D. C., Anal. Lett., 1983, 16, 1427. Socha, A. J., and Williardson, R. K., in “Proceedings of the 14th Annual Conference on Mass Spectrometry, Dallas, 1964.” Glinskikh, V. M., Melashvili, V. A., Ordzhonikidze, K. G., and Samadashvili, 0. A., J. Anal. Chem. USSR, 1974,29, 89. Jansen, J. A. J., and Witmer, A. W., Spectra2000, 1979,7,70. Jansen, J. A. J., and Witmer, A. W., Fresenius Z . Anal. Chem., 1981, 309, 262. Konishi, F., and Nakamura, N., Adv. Mass Spectrom., 1971,5, 547. Clegg, J. B., and Millett, E. J., Philips Tech. Rev., 1974, 34, 344. Kane, P. F., Anal. Chem., 1966, 38, 29A. Glavin, G. G., Goyushina, V. G., Kaplan B. Ya., Notkina, M. A., Solodovnik. S. M., and Khotin, B.A., in Alimarin, I. P., Editor, “Analysis of High-Purity Materials,” Israel Program for Scientific Translations, Jerusalem, 1968, p. 137. Mahalingam, T. R., Murugaiyan, P., Sonni, K., and Venkates- warlu, Ch., Report No. BARC-797, Bhabha Atomic Research Centre, Bombay, 1975, p. 7.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 419 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73 * 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. Beske, H. E., Frerichs, G., and Melchers, F. G., Fresenius 2. Anal. Chem., 1973, 267, 99. Beske, H. E., Fresenius 2. Anal. Chem., 1971, 256, 103. Kane, P. F., Chem. Tech., 1971, 532. Murugaiyan, P . , Pure Appl. Chem., 1982, 54, 835. Short, H. G., and Keene, B. J., Talanta, 1966, 13, 297.Vidal, G., Galmard, P., and Lanusse, P., Meth. Phys. Anal, GAMS, 1968, 4, 404. Crocker, I. H., and Wray, L. W., Can. Spectrosc., 1971,16,77. Adriaenssens, E., “Extraction Metallurgy 85,” Institution of Mining and Metallurgy, London, 1985, p. 979. Taylor, J. B., Calvert, L. D., Despault, J . G., Gabe, E. J., and Murray, J. J., J. Less-Common Met., 1974, 37, 217. Potapov, M. A., Chupakhin, M. S., Shtanov, V. I., and Zlomanov, V. P., J. Anal. Chem. USSR, 1978, 33, 820. Vastel, J., and Valeriano, P., Rev. Tech. Thomson-CSF, 1971, 3, 489. Dietze, H. J., and Zahn, H., ZIF Mitteilungen, 1980, 46, 60. Hurrle, A., and Dietl, J., Proc. Electrochem. SOC., 1980, 80-85, 106. Yudelevich, I. G., Gilbert, E. N., and Shelpakova, I. R., Zh. Anal. Khim., 1981, 36, 2393.Cornu, A., Adv. Mass. Spectrom., 1968, 4, 401. Verlinden, J., Vlaeminck, R., Adams, F., and Gijbels, R., in Katz, W., and William, P., Editors, “Applied Materials Characterization,” Materials Research Society, Pittsburgh, PA, 1985, p. 331. Zolotov, Yu. A., and Grasserbauer, M., Pure Appl. Chem., 1985, 57, 1133. Liebich, V., and Mai, H., Adv. Mass Spectrom., 1974, 6,655. Honig, R. E., Thin Solid Films, 1974, 31, 89. Malm, D. L., in Murt, E. M., and Guldner, W. G., Editors, “Physical Measurements and Analysis of Thin Films,” Progress in Analytical Chemistry, Vol. 2, Plenum Press, New York, 1969, p. 148. Derzhiev, V. I., Ramendik, G. I., Liebich, V., and Mai, H., Znt. J . Mass Spectrom. Ion Phys., 1980, 32,. 345. Ramendik, G. I., Tatsii, Yu. G., and Chupakhin, M.S . , J. Anal. Chem. USSR, 1973, 28, 653. Clegg, J. B., and Millett, E. J., Acta Electron., 1975, 18, 27. Clegg, J. B., Millett, E. J., and Roberts, J. A., Anal. Chem., 1970,42, 713. Chupakhin, M. S . , Ramendik, G. I., and Kruchkova, 0. I., Zh. Anal. Khim., 1969, 24, 965. Swenters, K., Verlinden, J., and Gijbels, R., Spectrochim. Acta, Part B , 1985,443,769. Swenters, K., PhD Thesis, Universitaire Instelling Antwerpen, Wilrijk, 1986. Ramendik, G. I., Derzhiev, V. I., Chupakhin, M. S . , Liebich, V., and Mai, H., J. Anal. Chem. USSR, 1975,30, 1609. Ramendik, G. I., Chupakhin, M. S . , Tatsii, Yu. G., and Derzhiev, V. I., J. Anal. Chem. USSR, 1974, 29, 202. Bedrinov, V. P., and Belousov, V. I., J. Anal. Chem. USSR, 1976,31, 1701. Bedrinov, V. P., Ukrainskii, Yu.M., Fursov, V. Z., and Chupakhin, M. S . , J. Anal. Chem. USSR, 1972, 27, 535. Bedrinov, V. P., Ukrainski, Yu. M., and Chupakhin, M. S . , J. Anal. Chem. USSR, 1972, 27, 1729. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. Chupakhin, M. S . , Ramendik, G. I., and Yavriyan, A. N., Zh. Anal. Khim., 1970, 25, 1301. Chupakhin, M. S . , Ramendik, G . I., and Venitsianov, V. E., Zh. Anal. Khim., 1970, 25, 855. Chupakhin, M. S., Ramendik, G. I., and Kryuchkova, 0. O., Zh. Anal. Khim., 1969, 24, 352. Chupakhin, M. S . , and Ramendik, G. I., Dokl. Akad. Nauk SSSR, 1969, 184, 1372. Chupakhin, M. S . , and Ramendik, G. I., J. Anal. Chem. USSR, 1972,26, 1241. Dorokhov, A. N., Shelpakova, I. R., and Yudelevich, I. G., J. Anal. Chem. USSR, 1975,30,672. Gerasimov, V. A,, Saprykin, A. I., Shelpakova, I. R., and Yudelevich, I. G., J . Anal. Chem. USSR, 1978, 33, 998. Jansen, J. A. J., and Witmer, A. W., Fresenius Z . Anal. Chem., 1981, 309, 305. Hickam, W. M., and Sandler, Y. L., in Bregman, J . I., and Dravnieks, A., Editors, “Surface Effects in Detection,” Spar- tan Books, Washington, DC, 1965, p. 189. Yudelevich, I. G., Shelpakova, 1. R., Shabanova, L. N., Gerasimov, V. A., and Korda, T. M., J. Anal. Chem. USSR, 1976, 31, 474. Yudelevich, I. G., and Shelpakova, I. P., Mikrochim. Acta, 1978, I, 547. Bedrinov, V. P., Belousov, V. I., and Kharomova, V. N., J. Anal. Chem. USSR, 1974, 28, 2132. Clegg, J. B., and Millett, E. J., Proc. SOC. Anal. Chem., 1974, 11, 52. Saprykin, A. I., Shelpakova, I. R., Chanysheva, T. A., and Yudelevich, I. G., Zh. Anal. Khim., 1983, 38, 1238. Shelpakova, I. R., Saprykin, A. I . , Chanysheva, T. A., and Yudelevich, I. G., J. Anal. Chem. USSR, 1983, 38,439. Kutil, J., and Urvalkova, D., Collect. Czech. Chem. Commun., 1968, 33, 983. Verlinden, J., PhD Thesis, Universitaire Instelling Antwerpen, Wilrijk, 1984. Adams, F., Verlinden, J., and Gijbels, R., in “Proceedings 11, 6th International Symposium on High-Purity Materials in Science and Technology, Dresden, 1985,” p. 1. Gerasimov, V. A., Shelpakova, I. R., and Rudaya, N. S . , J. Anal. Chem. USSR, 1979,34, 13. Yudelevich, I., and Shelpakova, I., in “Proceedings 11, 6th International Symposium on High-Purity Metals in Science and Technology, Dresden, 1985,” p. 269. Shelpakova, I., Saprykin, A., Yudelevich, I., Martin, A., and Rosin, A., in “Proceedings 11, 6th International Symposium on High-Purity Materials in Science and Technology, Dresden, 1985,” p. 276. Chastagner, P., Anal. Chem., 1969, 41, 796. Morrison, G. H., and Skogerboe, R. K., in Morrison, G. H., Editor, “Trace Analysis, Physical Methods,” Wiley-Inter- science, New York, 1965, p. 16. Ramendik, G. I., in Niinisto, L., Editor, “Euroanalysis IV,” Akademiai Kiad6, Budapest, 1982, p. 57. Paper J6f11 Received February 27th) 1986 Accepted June 12th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100411

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 421 Determination of Phosphorus by Graphite Furnace Atomic Absorption Spectrometry Part 1. Determination in the Absence of a Modifier Adilson J. Curtius,” Gerhard Schlemmer and Bernhard Welzt Department of Applied Research, Bodenseewerk Perkin-Elmer & Co GmbH, D-7770 Uberlingen, FRG In the absence of a modifier a substantial part of the analyte is lost prior to its atomisation. Phosphorus can, however, be stabilised by reaction with graphite in the condensed phase. Uncoated polycrystalline electrographite offers a much larger number of active carbon sites and hence much more favourable conditions for phosphorus determination than pyrolytic graphite coated tubes. The number of active carbon sites can be further increased if oxygen is added to the argon purge gas during certain stages of thermal pre-treatment.The best characteristic mass values obtained in the absence of a modifier [ca. m, = 20 ng (0.0044 A s)-l] are, however, substantially inferior to those in the presence of modifiers. Keywords: Phosphorus determination; graphite furnace atomic absorption spectrometry; graphite surface reactions; oxygen purge gas; volume effect Phosphorus is an element that is not typically determined by atomic absorption spectrometry. One of the reasons for this is certainly that the resonance lines for this element, originating from the 4S03,2 term, are all in the vacuum UV between 167.2 and 178.8 nm, a range that is not readily available in conventional atomic absorption spectrometers. L’vov and Khartsyzovl proposed therefore to use the phosphorus lines at 213.5 and 213.6 nm, which originate from the 2D0312 and *D05/2 terms, respectively. These excited levels are only 1.4 eV above the 4S03/2 term so that a reasonable population can be expected at temperatures of ca.2900 K, which are typically used for atomisation. L’vov and Khartsyzov reported a population of about 0.4% for the 2D05,2 level and found a detection limit of 0.2 ng in their graphite cuvette, heated to a constant temperature of 2700 K. After the introduction of a commercially available elec- trodeless discharge lamp for phosphorus,2 Ediger3 proposed a set of standard conditions for the determination of this element in the HGA graphite furnace using the 213.5 - 213.6 nm doublet.In later work, Ediger et ~ 1 . ~ investigated the effect of the matrix on the absorption signal and found that most phosphorus compounds, except for calcium phosphate, give essentially no response at all unless sufficient lanthanum is added as a modifier. Lanthanum was also used by most other workers to stabilise phosphorus for determination by elec- trothermal atomisation atomic absorption spectrometry. One of the few exceptions is the determination of phosphorus in oils reported by Pr6vGt and Gente-Jauniaux,S where the oil matrix apparently stabilises the phosphorus sufficiently so that the addition of lanthanum is not necessarily required. The first careful investigation of factors influencing the determination of phosphorus in electrothermal atomisation atomic absorption spectrometry was published by Persson and Frech.6 These workers found that there are two main problems associated with the determination of this element. These are pre-atomisation losses of phosphorus in the form of suboxides and/or the dimer molecule, and the fact that phosphorus needs not only to be atomised but also excited for its determination at the non-resonance doublet at 213.5 - 213.6 nm.These workers concluded that, unless a phase of high * Present address: Departamento de Quimica, Pontificia t To whom correspondence should be addressed. Universidade Cat6lica do Rio de Janeiro, Rio de Janeiro, Brasil. thermostability is formed by the addition of a modifier, reproducible results can be obtained only if the heating rate and the final temperature of the furnace as well as the atmosphere inside the graphite tube can be controlled during the course of the determination.The most suitable system for phosphorus determination should permit the introduction of the sample into an atomiser pre-heated to a high temperature. One such system is the graphite cuvette used by L’vov and Khart syzov. 1 In this work we wanted to investigate the influence of the graphite tube material and the graphite surface on the determination of phosphorus. No modifier was added to abstract the analyte - carbon interaction from other effects that could possibly occur in the presence of concomitants. The “platform effect” was also investigated, i.e., the effect that is obtained if the analyte volatilisation is shifted in time until the temperature of the tube wall and gas atmosphere have stabilised.Also investigated was the influence of oxygen added to the purge gas at different stages of the thermal pre-treatment of phosphorus. Experimental Instrumentation A Perkin-Elmer Zeeman 3030 atomic absorption spec- trometer equipped with a Zeeman graphite furnace and HGA-600 programmer was used throughout this work. An electrodeless discharge lamp for phosphorus was operated from an external power supply at 8 W. The monochromator was set to 213.6 nm and the slit width to 0.7 nm. A typical temperature programme for the determination of phosphorus is given in Table 1. This programme is optimised for platform sampling and used for wall sampling without change. The third step of this programme was varied between 200 and 2400 “C, but all the other steps were held constant to establish the thermal pre-treatment curves.Similarly, the fifth step was varied but the other steps were unchanged in order to obtain the atomisation curves. Integrated absorbance values were used exclusively for signal evaluation throughout this work. Materials and Reagents Argon. 99.996% purity and argon containing 1% V/V oxygen (Linde, FRG) . Phosphorus stock solution, 1000 mg 1-1. Prepared from dibasic ammonium phosphate, (NH4)2HP04, and diluted with422 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Table 1. Temperature programme used for the determination of phosphorus Internal Timels gas flow- Step Furnace rate/ 90 1 10 300 1 120 15 10 300 2 3 1350 1 30 300 4 200 10 10 300 5 2650 0 5 0 6 2700 1 4 300 7 20 1 8 300 number temperature/"C Ramp Hold ml min-l Read - - - - * - - * Key activated at this stage in the programme.0.2% V/V nitric acid; 20 p1 of a solution containing 50 pg ml-l were used for most determinations. The following graphite tubes and platforms were used: Uncoated polycrystalline electrographite tube. Perkin-Elmer Pyrolytic graphite coated tube. Perkin-Elmer Part No. L'vov platform, pyrolytic graphite. Perkin-Elmer Part No. L'vov platform. Laboratory-made from polycrystalline Total pyrolytic graphite tubes. Ringsdorff-Werke, Bonn, Part No. BO 070 699. BO 091 504. BO 109 324. electrographite tubes. FRG. Results and Discussion Influence of the Graphite Surface Several workers have reported that the tube material has an influence on the sensitivity of the determination of phospho- rus.PrCvBt and Gente-Jauniaux5 found that a new tube gives a very low signal for the first determinations and that after 30 min of "ageing" by repeated injections of oil the sensitivity increases gradually. They assumed that these phenomena are due to a modification of the graphite structure. Ediger et al.4 found a difference in the sensitivity for phosphorus between uncoated and pyrolytic graphite coated tubes that persisted after the addition of increasing amounts of lanthanum as a modifier. Persson and Frech6 reported that the sensitivity for phosphorus was highly dependent on the type as well as the age of the graphite tubes used. We therefore investigated uncoated and pyrolytic graphite coated tubes made of polycrystalline electrographite, and platforms made of total pyrolytic graphite and polycrystalline electrographite, respectively, as atomisation surfaces for phosphorus.The characteristic mass values found for the two types of tubes and the various tube - platform combinations are summarised in Table 2. These data show that the characteristic mass for phosphorus may vary by more than an order of magnitude depending on the graphite surface, the pyrolysis temperature and the atomisation conditions (wall or platform atomisation). Two different pyrolysis temperatures were selected for the m, values in Table 2, 200 and 1350 "C. The best characteristic mass is always obtained at 200 "C because the analyte losses are least at this low temperature, but this is not a useful pyrolysis temperature.A temperature of 1350 "C was selected because this temperature is close to the second maximum if uncoated polycrystalline graphite is used. This pyrolysis temperature would be suitable for the removal of most concomitants, and it is typically used when phospho- rus is determined in the presence of a modifier.7>8 However, there are increasing analyte losses compared with the 200 "C pyrolysis temperature if no modifier is used. The dependence of the integrated absorbance signal for phosphorus on the thermal pre-treatment temperature is shown in Figs. 1-3 for the different tubes and tube - platform combinations. Table 2. Influence of platform and/or tube material on the characteristic mass [mdng (0.0044 A s)-1] obtained for phosphorus with different pyrolysis temperatures; atomisation temperature 2650 "C Pyrolysis temperature/"C Platform Tube material * material* - EG - PGC PG EG PG PGC EG EG EG PGC 200 1350 30 55 220 No signal 40 210 45 650 20 45 17 110 * EG = polycrystalline electrographite; PG = pyrolytic graphite; and PGC = pyrolytic graphite coating.0.15 v) ,y, B 0 500 1000 1500 2000 2500 Tern peratu re/"C Fig. 1. Thermal pre-treatment and atomisation curves for 1 yg of P, deposited on to the tube wall: A, uncoated polycrystalline electro- graphite tube; B, pyrolytic graphite coated tube; and C, atomisation in an uncoated polycrystalline graphite tube, pyrolysis temperature 1350 "C 0.15 C / I 500 1000 1500 2 Tern peratu rePC '000 2500 Fig. 2. Thermal pre-treatment and atomisation curves for 1 yg of P, deposited on to a pyrolytic graphite platform: A, in an uncoated polycrystalline electrographite tube; B, in a pyrolytic graphite coated tube; and C, atomisation in an uncoated polycrystalline graphite tube, pyrolysis temperature 200 "C The difference between the two tube materials is most pronounced when the phosphorus reference solution is deposited directly on, and volatilised from the tube wall ("wall sampling"), which is depicted in Fig.1. The lowest sensitivity (highest characteristic mass) for all the conditions evaluated is obtained with a pyrolytic graphite coated tube, whereas an almost ten times better sensitivity is obtained in an uncoated polycrystalline electrographite tube. For pyrolysis tempera- tures higher than 600 "C the sensitivity starts to fall off and reaches a minimum at ca. 1000 "C.For even higher pyrolysis temperatures, however, the sensitivity increases again to a maximum at 1500-1600 "C. It should be noted that a signal for phosphorus is recorded even after 30-s pyrolysis at tempera- tures as high as 2400 "C.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 423 0.25 1 \ 0 500 1000 1500 2000 2500 2800 Tern peratu re/”C Fig. 3. Thermal pre-treatment and atomisation curves for 1 pg of P, deposited on to a polycrystalline electrographite platform: A, in an uncoated polycrystalline electrographite tube; B, in a pyrolytic graphite coated tube; and C, atomisation in an uncoated poly- crystalline graphite tube, pyrolysis temperature 1350 “C The difference between the two tube materials, uncoated and pyrolytic graphite coated polycrystalline electrographite, is much less pronounced when the phosphorus reference solution is deposited on, and volatilised from a platform (“platform sampling”).This is depicted in Fig. 2 for a platform made of pyrolytic graphite and in Fig. 3 for a platform made of polycrystalline electrographite. A significant drop in sensitiv- ity is found with increasing pyrolysis temperatures and there is no indication of any increase in sensitivity with pyrolysis temperatures higher than 1000 “C when a pyrolytic graphite platform is used in an uncoated tube, as was observed for wall sampling (see Fig. 1). This increase in sensitivity with a maximum of 1500-1800 “C, however, becomes clearly appar- ent when the platform is made of polycrystalline electrograph- ite (Fig. 3).The sensitivity obtained for phosphorus in this maximum depends on the tube in which the platform is used. This shows that not only is the surface on which the sample is deposited important, but so is the surface of the tube that the analyte meets only after its volatilisation. The highest sensitiv- ity, however, is again obtained with rather low thermal pre-treatment temperatures of only 600 “C or less. It is interesting to note that the tube material becomes insignificant for low pyrolysis temperatures and the sensitivity is deter- mined predominantly by the surface on which the analyte is deposited. It has already been clearly pointed out by Persson and Frech6 that one of the basic problems associated with the determination of phosphorus, apart from pre-atomisation losses, is in the use of non-resonance lines.This means that phosphorus must not only be atomised but also brought into an excited state, and the relative population of these energy levels is very much temperature dependent. A higher inte- grated absorbance signal for phosphous can therefore be due to a larger number of atoms being produced, i.e., a higher atomisation efficiency, and/or to a higher population of the metastable state because of a higher effective temperature of the analyte atoms. The differences between the curves for wall sampling and platform sampling when all surfaces are pyrolytic graphite (curves B, Fig. 1 and Fig. 2) and when they are uncoated polycrystalline electrographite, respectively (curves A , Fig.1 and Fig. 3), can be fully ascribed to the “platform effect.” The sensitivity obtained with platform sampling is for all thermal pre-treatment temperatures higher than, or at least equal to that obtained with wall sampling from the same surface. This platform effect is caused by the fact that the temperature of the platform lags behind that of the graphite tube into which it is inserted, so that the tube and the gas atmosphere have stabilised in temperature when the analyte is volatilised.9JO This results in a higher effective temperature in the graphite tube, which leads to a more effective dissociation of molecular species into analyte atoms and/or a higher population of the metastable phosphorus levels that can absorb radiation at the 213.5 - 213.6 nm doublet.All the other differences between the individual curves must be related to surface effects in one way or another. It has already been mentioned that, in addition to the influence of the surface on which the analyte is deposited and from which it is volatilised, there is an effect from the tube surface that the analyte meets only after its volatilisation. This influence becomes apparent when curves A and B are compared in Figs. 2 and 3, respectively. It is well documented that uncoated polycrystalline electrographite has a much higher porosity and a much greater number of reactive carbon sites than pyrolytic graphite.1l Both of these properties can contribute to preventing a larger amount of phosphorus from being lost during the thermal pre-treatment step and/or to keep it longer at the surface during the atomisation step so that a higher effective temperature is reached when it is released into the vapour phase.L’vov et a1.12 used their macrokinetic theory of sample vaporisation to explain the influence of the tube surface on the phosphorus sensitivity by the difference between sample volatilisation from the bulk of a porous body (uncoated tube) and from an open surface (pyrolytic graphite coated tube), respectively. The shift of the phosphorus volatilisation to a higher temperature on more porous surfaces increases the sensitivity in spite of the lower volatilisation rate. In addition to this physical effect, however, there may be a chemical effect and it should be pointed out that these two mechanisms need not be competitive but may well be complementary.The most striking phenomenon is certainly the apparent stabilisation of phosphorus at thermal pre- treatment temperatures above 1000 “C when the analyte is deposited on a polycrystalline electrographite surface. This stabilisation reaches a maximum at 1500-1600 “C, but a signal for phosphorus is obtained even after 30 s of pyrolysis at 2400 “C. Phosphorus forms a “metallic” carbide P&13 which certainly contributes to the prevention of analyte losses. It is unlikely, however, that this carbide is stable up to 2400 “C. Phosphoric acid is reported to form electrolytic lamellar compounds with graphite in the presence of oxygen and hydrogen.14 At low temperatures, oxygen is chemisorbed on graphite and forms carbon - oxygen compounds that provide active sites for attracting phosphorus.Hydrogen must be present in order to stabilise the graphite - phosphate com- pounds. The formation of active sites is catalysed by water and this means that the surface of uncoated polycrystalline electrographite should provide conditions for intercalation if aqueous samples are analysed.15316 It is known that most electrolytic lamellar compounds cannot be decomposed com- pletely under ordinary conditions, and a significant residue is retained with extreme tenacity.14 It is likely that the formation of such residue compounds is responsible for the retention of phosphorus on polycrystalline graphite at such high tempera- tures.KoreEkova et al. 15 reported a very similar behaviour for arsenic when it was determined in uncoated graphite tubes. Whereas the formation of lamellar and residue compounds can explain the stabilisation of phosphorus up to high temperatures if the analyte is in contact with graphite in the condensed phase, it cannot be applied to heterogeneous reactions. The observation that a higher sensitivity is obtained over essentially the entire temperature range in a p l y - crystalline electrographite tube when the phosphorus solution is volatilised from a pyrolytic graphite platform (Fig. 2) can in424 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 principle be due to several different reactions. One explana- tion would be that the more reactive polycrystalline graphite influences the partial pressure of oxygen and carbon mon- oxide in the gas atmosphere, as proposed by several work- ers,6J7J* which, in turn, favours the dissociation of phospho- rus oxides into atoms.Another explanation would be that volatilised phosphorus oxides are reduced to the element more effectively at the more reactive graphite surface. A third explanation would be that volatilised phosphorus compounds, either suboxides or the dimer, are adsorbed or chemically bound to the reactive graphite surface on collision,19 and "stick" to the surface until a higher temperature is reached at which atomisation and excitation are more effective. This third explanation is strongly supported by investigations using scanning electron microscopy.7 Effect of Oxygen We also carried out some experiments in which argon enriched with oxygen (1% 02) was used as the purge gas in different stages of the temperature programme.In a pyrolytic graphite coated tube and after pyrolysis at 1350 "C, no signal exceeding the background signal was found for 2 pg of phosphorus, and this was independent of the programme stage in which the oxygen was applied. When phosphorus was determined in an uncoated polycrystalline electrographite tube, however, oxy- gen addition to the purge gas in most instances brought about an improvement in the sensitivity. The extent of this improve- ment was dependent on the thermal pre-treatment tempera- ture used and the programme stage in which oxygen was applied. The influence of oxygen added to the purge gas during the drying stages at 90 and 120 "C on the integrated absorbance signal of phosphorus is depicted in Fig.4, which shows that the effect does depend on the pyrolysis tempera- ture applied in the thermal pre-treatment stage after the addition of oxygen. An increase due to oxygen could be observed only if a pyrolysis temperature higher than 700 "C was used and the enhancement was most pronounced for pyrolysis temperatures of 1000 "C or higher. When oxygen was applied during the "cool" step to 200 "C between the pyrolysis and atomisation stages, the sensitivity increase was effective over the entire range of pyrolysis temperatures as can be seen in Fig. 5. A similar curve, but with a substantially less pronounced enhancement over that without oxygen was observed when oxygen was introduced during the pyrolysis stage itself.Most striking, however, was the observation that in both instances an addition of oxygen to the purge gas, during the pyrolysis stage or during the cool step before atomisation, never had any influence on the first determination after oxygen was applied, but only on the 0.1 5 cn ? 6 0.10 2 s: u D -0 0.05 L w 4- - 0 t 200 500 1000 1500 2000 2500 Tern peratu rePC Fig. 4. Thermal pre-treatment and atomisation curves for 1 pg of P de osited on to the wall of a polycrystalline electrographite tube: A, 12 oxygen in argon purge gas during the drying stages; B, thermal pre-treatment with pure ar on purge gas (curve from Fig. 1); and C, atomisation curves with (f) and without (0) oxygen added to the purge gas second determination. It was also found that the effect of oxygen, applied in the pyrolysis stage or later, was suspended as soon as a clean step at 2700 "C was used for an extended period of time of 20 s or longer after the atomisation cycle. If oxygen is applied during the drying steps, however, the increasing effect is observed even in the first run, and the duration of the clean step at 2700 "C had no influence, i.e., the effect was not suspended.The effect of oxygen on the sensitivity and/or the appear- ance and the shape of the absorption pulse of the analyte element is the subject of several publications. Phosphorus, however, has not been investigated until now, but it can be expected that some similarities exist with the behaviour of other elements.Beaty et a1.20 found a loss in sensitivity for a number of elements on the addition of oxygen when the determination was carried out in a pyrolytic graphite coated tube. In uncoated polycrystalline electrographite tubes, however, oxygen caused an increase in sensitivity for three elements (indium, lead and zinc), and no loss for the other elements investigated. Sturgeon and Berman17 found that oxygen had little influence on the sensitivity of 12 elements in uncoated and pyrolytic graphite coated tubes. The shift in the appearance temperature, however , was noticeable, and more pronounced for pyrolytic graphite coated tubes. This shift in the appearance temperature was studied carefully by Hol- combe and CO-workers.21-23 Chemisorbed oxygen and the resulting alteration of the graphite surface, i.e., the active spots at this surface, is proposed by these workers to be responsible for the observed phenomena.Reversible (anneal- able) alteration of the surface morphology of the tube occurring at high temperature in the presence of oxygen is also considered by Sturgeon and Berman17 to give rise to specific analyte - surface interactions. The majority of their data, however, in their opinion are consistent with a mechanism that is a response to the suppression of thermal dissociation of analyte oxides by high oxygen partial pressure (Po,). This is essentially along the same lines as the theory of L'vov and Ryabchuk24 who suggested that absorption pulse shifting occurred as a result of a decrease of condensed-phase thermal dissociation rate of analyte oxides in reponse to an increase in The increase in the integrated absorbance signal of phos- phorus in the presence of oxygen observed in this work was not accompanied by a shift in the appearance time.This does not imply, however, that oxygen has no influence on the poz_* 0.20 v) $ 0.15 u m 2 2 n 0.10 u w 2 0 w - 0.05 0 500 1000 1500 2000 2500 Pyrolysis tern peratu re/"C Fig. 5. Thermal pre-treatment curves for 1 pg of P deposited on to the wall of a polycrystalline electrographite tube: A (0), 1% oxygen in argon purge gas during cool sta e to 200 "C with atomisation signal after second oxygen addition; B (d), same as A but atomisation signal after first oxygen addition; and C (A), argon purge gas425 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY.DECEMBER 1986, VOL. 1 atomisation mechanism. It has been discussed that at the 213.5 - 213.6 nm phosphorus lines the appearance of an absorption signal means the appearance of excited atoms. Ground-state phosphorus atoms cannot absorb radiation at these lines so that a change in their’ appearance remains undetected. A late shift in the appearance of ground-state phosphorus atoms may preferably translate into an increase of the integrated absorbance because more atoms become available and they appear at a higher temperature, which is more favourable for their excitation, The observation that oxygen, when applied during the pyrolysis or cool stages before atomisation, has no influence on the first phosphorus determination precludes a direct involvement of oxygen in the atomisation of phosphorus.This means that the increased sensitivity for phosphorus in the presence of oxygen cannot be due to a decrease of the thermal dissociation of analyte oxides during atomisation. It also cannot be due to oxygen chemisorbed at the graphite surface, which is unlikely anyway if pyrolysis temperatures of more than 1000 “C are applied, because oxygen is desorbed rapidly at these temperatures. The mechanism which is most likely to occur is that an activated carbon surface is created by desorption of previously chemisorbed oxygen, i.e., of surface oxides, as CO and C 0 2 at elevated temperatures. This ‘activated carbon surface then stabilises phosphorus in the thermal pre-treatment and atomisation stages. This is in agreement with the model of Holcombe and Droessler23 and with one of the conclusions of Sturgeon and Berman,l7 that alterations of the surface morphology of the tube at elevated temperature in the presence of oxygen may account for specific analyte - surface interactions.L’vov and Ryabchuk24 similarly concluded that an enhanced initial oxygen content in the argon during the preliminary “low temperature” tube treatment should favour a sharp increase of the reactivity of carbon. They then suggested, however, that a more efficient purification of argon from the O2 and OH impurities in the subsequent heating of the tube to the atomisation temperature accounts for the enhanced sensitivity effect found for some elements (not including phosphorus) under these conditions. This mechan- ism should come into effect even for the first determination, which is not in agreement with the observations made in this study.Surface oxides of graphite are best formed at temperatures of 400-500 “C, and at 900-1000 “C most of the oxygen is split off again as CO and C02.11 Whilst discussing the effects observed in our work it should be kept in mind that the oxygen applied in one stage is typically still present in the furnace at the beginning of the next stage of the temperature pro- gramme. Oxygen present during the drying stages is adsorbed on to the graphite surface during the drying and the following temperature ramp to the pyrolysis stages. If the pyrolysis temperature does not exceed 700 “C, the oxygen is not desorbed until the atomisation stage, i.e., no active carbon centres are created during thermal pre-treatment. If higher pyrolysis temperatures are applied, oxygen is desorbed during this stage leaving active carbon centres that can now stabilise the phosphorus leading to the observed higher integrated absorbance values.If oxygen is added during the pyrolysis or the cool stages, however, it will be preferentially adsorbed during the cooling period to 200 “C and desorbed again only during the atomisation cycle. This means that the analyte element is already lost before the active carbon centres are created, which can therefore stabilise phosphorus in the next cycle only. The annealing effect after extended treatmefit of the tube at 2700 “C is in agreement with this model, as it is observed only if oxygen is applied during the pyrolysis or cool stage, and active carbon centres are created for the following determination only.At high temperatures the increasing mobility of carbon atoms on the surface brings about a gradual rearrangement of the graphite crystal lattice resulting in a decrease in the number of active centres.24 The activated surface can react with phosphorus in two ways. It can react in the condensed phase by favouring the formation of lamellar compounds and thereby reduce low- temperature losses of phosphorus in the form of gaseous molecules. The activated surface can also be considered to be “stickier” for gaseous phosphorus molecules (e.g., suboxides or P2) so that more of them will be chemisorbed after their volatilisation and therefore not lost but finally atomised.Volume Effect If the graphite surface reacts directly with the analyte in the condensed phase, the surface area that comes into contact with a given analyte mass should have an influence on the integrated absorbance. A comparison was therefore made between the signals obtained for 5 p1 of a solution containing 200 mg 1-1 of P and for 20 pl containing 50 mg 1-1 of P, both corresponding to an absolute mass of 1 pg of phosphorus, but distributed over a different wet surface area. The experiments were carried out with and without the addition of oxygen to the purge gas during the drying stage. The ratios of the integrated absorbance signals obtained under the different conditions are summarised in Table 3 for an uncoated polycrystalline electrographite tube using wall and platform sampling.No data are included for a pyrolytic graphite coated tube because no signals were obtained that exceeded the noise level. Ratios greater than one are obtained in all instances, which means that a greater integrated absorbance is obtained for the larger volume (larger wet surface area), and also for the surface treated with oxygen in comparison with the untreated surface. Table 3 also shows that the volume effect is greater (1.8) for the oxygen treated surface than it is for the untreated surface (1.3) when wall sampling is applied. The effect is less pronounced when the analyte is deposited on a pyrolytic graphite platform, but there is also a volume effect for this sampling technique. As the volume effect cannot be due to an interaction of volatilised phosphorus species with the tube wall, this means that a substantial number of reactive carbon sites must also be available on the pyrolytic graphite platform surface.This is not surprising as a large number of edge- carbon planes are exposed in the machining process of these platforms, and these are considerably more reactive than the basal plane, thus allowing for easier intercalation of various compounds into the graphite structure.25 Substantial corrosive attack was found on platforms after repeated determinations of phosphorus7 so that an increasing number of reactive sites may be created with time. Table 3. Ratio of the integrated absorbance signals obtained for phosphorus (1 pg) using two different volumes of solution (5 or 20 pl) and argon or argon containing 1% oxygen during the drying steps Integrated absorbance ratio for volumes of 20 p1 to 5 PI- Purge gas Ar Ar + O2 EG* tube, wall sampling .. . . 1.3 1.8 EG* tube, PG* platform . . . . 1.3 1.5 Integrated absorbance ratio for Ar + O2 to Ar- Injected volume 20 pl 5 PI EG* tube, PG* platform . . . . 1.4 1.2 2.0 1.4 EG* tube, wall sampling . . . . * EG = polycrystalline electrographite; PG = pyrolytic graphite.426 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 0.10 I L - I I I I 0 100 200 300 400 Number of determinations Fig. 6. Integrated absorbance for 1 pg of P over the lifetime of a graphite tube. Each data point is the average of 20 determinations: A, uncoated polycrystalline electrographite tube and wall sampling; and B, pyrolytic graphite coated tube with inserted pyrolytic graphite platform When the second set of data, that for the effect of oxygen treatment, are compared for different volumes of solution (i.e., different wet surface areas), the ratios are clearly greater for the larger volume.The effect is also much more pronounced for the uncoated polycrystalline electrographite surface than it is for the pyrolytic graphite platform surface. This shows, as proposed earlier, that the sensitivity increase is primarily due to a reaction of the activated surface with the analyte in the condensed phase, and not to a purification of the gas atmosphere. Tube Endurance Experiments In several experiments we found evidence that phosphorus, even in the absence of any concomitants, has a corrosive effect on the graphite tubes.’ We therefore carried out endurance experiments in which masses of 1.0 and 0.2 pg of P as the dibasic ammonium phosphate in dilute nitric acid solution were determined, using the temperature programme given in Table 1.The change in the integrated absorbance of 1 pg of P over time for two such tubes is shown in Fig. 6. The uncoated polycrystalline electrographite tube, in which the analyte was deposited on the tube wall, broke after 272 determinations, whereas the pyrolytic graphite coated tube with inserted pyrolytic graphite platform failed after 349 determinations. These numbers are clearly smaller than those typically found for matrix-free solutions, particularly when platform atomisation is considered.Usually the lifetime of a tube is twice as long when platform sampling is used compared with wall sampling,25 and the pyrolytic graphite coating should have an additional effect. The pronounced destruction of the platform surface and of the whole graphite tube can be seen very clearly in the scanning electron micrographs depicted in Fig. 7 (a and b). The dependence of the corrosion on the concentration of phosphorus was demonstrated in an experi- ment with 0.2 pg of P, which is not shown here, where the tube and platform were still in reasonably good condition after 410 determinations. The early failure of the graphite tubes is without doubt due to reactions of phosphorus with carbon in the condensed phase and also to a vapour phase attack.Hennigl4 has pointed to the formation of residue compounds from lamellar compounds that are retained by the graphite with extreme tenacity. During their eventual thermal decom- position at very high temperatures, these residue compounds form volatile carbon compounds, which appears to be the mechanism of graphite tube corrosion. More important than the actual lifetime of a tube until breakage occurs is the analytically useful lifetime.25.26 This is the time over which no significant change in sensitivity and reproducibility of the determination is observed. For the uncoated polycrystalline electrographite tube no such change occurred so that the analytically useful lifetime is identical with the total lifetime. For the pyrolytic graphite coated tube Fig. 7.Scanning electron micrographs of corroded platform and tube surfaces after 349 determinations of phosphorus (1 pg of P): (a) Bottom of pyrolytic graphite platform cavity with severe pitting; and (b) injection hole area of the broken tube exhibiting disintegrated polycrystalline graphite (lower left corner) and secondary deposition of nodular graphite and platform sampling, however, a dramatic increase in the integrated absorbance is found towards the end of its lifetime, but well before it actually breaks. This phenomenon is undoubtedly due to the progressive corrosion of the pyrolytic graphite platform surface and of the pyrolytic graphite coating of the tube. This could be shown very clearly on inspection of tube and platform using scanning electron microscopy.7 It is interesting to note that the integrated absorbance values obtained towards the end of the tube life are very close to those obtained for a polycrystalline electrographite plat- form in an uncoated tube.This shows that a pyrolytic graphite coated tube behaves like an uncoated tube after the coating is corroded, which is obvious. It also proposes, however, that a pyrolytic graphite platform, after substantial corrosion, behaves similarly to a polycrystalline electrographite plat- form. This means that, apparently, a comparably large number of active carbon sites are exposed on pyrolytic graphite on pronounced corrosion. Conclusion Phosphorus forms a number of rather volatile compounds, such as suboxides or the dimer molecule P2, which may lead to substantial analyte losses unless special precautions are taken.Moreover, because its resonance lines are in the vacuum UV, phosphorus has to be determined on non-resonance lines and the element needs not only to be atomised but also excited to a metastable state from which the absorption originates. This requires additional energy so that the effective temperature in the graphite furnace plays a decisive role in the determinationJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 427 of phosphorus. In practice this means that the analyte must be retained in the condensed phase as long as possible so that the highest achievable temperature is obtained at the time of analyte volatilisation. This will ensure the most effective dissociation into atoms and the highest possible population of the excited state.The thermal stability of phosphorus, i.e., the extent to which it can be retained in the condensed phase to a high enough temperature so that a reasonable sensitivity is obtained, depends very much on the availability of reactive carbon sites. Pyrolytic graphite coated or total pyrolytic graphite tubes have a relatively small number of reactive sites at their surface, which results in a very ineffective stabilisation and consequently in a very low sensitivity for phosphorus. Uncoated polycrystalline electrographite tubes offer more favourable conditions for the determination of this element. The situation can be further improved if the analyte is deposited on a platform made of polycrystalline graphite, which further delays volatilisation until the graphite tube and the gas atmosphere have stabilised in temperature. The addition of oxygen to the purge gas during drying or thermal pre-treatment steps increases the number of active centres at the graphite surface substantially so that phosphorus can be stabilised much more effectively.The stabilisation of phosphorus on the graphite surface is primarily a condensed phase reaction that leads to the formation of lamellar and residue compounds. Additional stabilisation was found to proceed via the gas phase in a heterogeneous reaction. Volatilised phosphorus species are chemisorbed at active carbon sites, “stick” to the surface for some time and are in this way retained to a higher temperature where atomisation and excitation are more effective.Even the best sensitivity obtained for phosphorus under the most favourable conditions, however, is still clearly inferior to that obtained with the addition of a modifier. The determina- tion of phosphorus without a modifier will therefore not be of practical analytical importance. We nevertheless believe that these investigations have provided a great deal of information about the reactions between phosphorus and graphite, which may be of general interest and may also help to understand the reactivity of other elements. Some of the observations made in this study, particularly those on heterogeneous reactions between gaseous species and the graphite surface, may also be of importance if a graphite furnace is used as a detecting device for chromatographic systems.The authors are grateful to H. M. Ortner and W. Birzer, Metallwerke Plansee GmbH, Reutte, Austria, for providing the scanning electron micrographs. This research was sup- ported by the Conselho Nacional de Desenvolvimento Cientifico e Tecnologico (CNPq), Brasil. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. References L’vov, B. V., and Khartsyzov, A. D., Zh. Prikl. Spektrosk., 1969, 11, 9. Barnett, W. B., Vollmer, J. W., and de Nuzzo, S . M., At. Absorpt. Newsl., 1976, 15, 33. Ediger, R. D., At. Absorpt. Newsl., 1976, 15, 145. Ediger, R. D., Knott, A. R., Peterson, G. E., and Beaty, R. D., At. Absorpt. Newsl., 1978, 17, 28. PrevGt, A., and Gente-Jauniaux, M., At. Absorpt. Newsl., 1978, 17, 1. Persson, J. A., and Frech, W., Anal. Chim. Acta, 1980,119,75. Welz, B., Curtius, A. J., Schlemmer, G., Ortner, H. M., and Birzer, W., Spectrochim. Acta, Part B, 1986, 41, in the press. Curtius, A. J., Schlemmer, G., and Welz, B., J. Anal. At. Spectrom., accepted for publication (Part I1 of this study). L’vov, B. V., Pelieva, L. A., and Sharmopolski, A. I., Zh. Prikl. Spektrosk., 1977, 27, 395. Slavin, W., Manning, D. C., and Carnrick, G. R., At. Spectrosc., 1981, 2, 137. Huettner, W., and Busche, C., Fresenius Z. Anal. Chem., 1986, 323,674. L’vov, B. V., Bayunov, P. A., and Ryabschuk, G. N., Spectrochim. Acta, Part B, 1981, 36, 397. Benesovsky, F., in “Ullmanns Encyklopadie der technischen Chemie ,” Fourth Edition, Volume 9, Verlag Chemie, Wein- heim, 1975, p. 122. Hennig, G. R., in Cotton, F. A., Editor, “Progress in Inorganic Chemistry,” Volume 1, Interscience, New York, 1959, p. 125. KoreCkovA, J., Frech, W., Lundberg, E., Persson, J. A., and Cedergren, A., Anal. Chim. Acta, 1981, 130, 267. Frech, W., and Cedergren, A., Anal.Chim. Acta, 1976,82,93. Sturgeon, R. E., and Berman, S . S . , Anal.Chem., 1985, 57, 1268. Frech, W., Lindberg, A. O., Lundberg, E., and Cedergren, A., Fresenius Z. Anal. Chem., 1986, 323, 716. Schlemmer, G., and Welz, B., Fresenius Z . Anal. Chem., 1986, 323, 703. Beaty, M., Barnett, W., and Grobenski, Z . , At. Spectrosc., 1980, 1, 72. Salmon, S. G., Davis, R. H., and Holcombe, J. A., Anal. Chem., 1981, 53, 324. Salmon, S. G., and Holcombe, J. A., Anal. Chem., 1982, 54, 630. Holcombe, J. A., and Droessler, M. S . , Fresenius Z . Anal. Chem., 1986,323, 689. L’vov, B. V., and Ryabchuk, G. N. , Spectrochim. Acta, Part B, 1982,37, 673. Welz, B., Schlemmer, G., and Ortner, H. M., Spectrochirn. Acta, Part B, 1986, 41, 567. Ortner, H. M., Schlemmer, G., Welz, B., and Wegscheider, W., Spectrochim. Acta, Part B, 1985, 40, 959. Paper 56/41 Received June 3rd, 1986 Accepted July 28th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100421

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 429 Determination of Trace Amounts of Nickel and Cobalt in Silicate Rocks by Graphite Furnace Atomic Absorption Spectrometry: Elimination of Matrix Effects with an Ammonium Fluoride Modifier Rokuro Kuroda, Toshihiko Nakano, Yasuharu Miura and Koichi Oguma Laboratory for Analytical Chemistry, Faculty of Engineering, University of Chiba, Chiba, Japan Trace amounts of nickel and cobalt have been determined in a variety of standard rocks by graphite furnace atomic absorption spectrometry after fusion with a mixture of lithium carbonate and boric acid. The presence of ammonium fluoride in the reaction medium served to remove severe matrix effects, allowing aqueous hydrochloric acid solutions to be used as calibration standards.Results are given for a variety of standard rock samples. The sensitivities are 14 and 25 pg for nickel and cobalt, respectively, with respect to 1% atomic absorption. Keywords: Atomic absorption spectrometry; cobalt and nickel determination; rock analysis; graphite furnace; ammonium fluoride modifier The decomposition of silicates by fusion with lithium fluxes has found increasing use during the past decade, particularly in atomic absorption and X-ray fluorescence spectroscopy.192 Lithium metaborate is generally used when dissolution of the sample is necessary after decomposition. This solution tech- nique has been widely advocated by Ingamells,3 who used it for the rapid photometric analysis of silicate rocks for major and minor elements, and has been adapted to flame atomic absorption ~pectrometry.~7 Omangs proposed a method for the determination of Si, Al, Ti, Fe, Ca, Mg, Na and K in various silicates by atomic absorption spectrometry, in which a mixture of lithium carbonate and boric acid was used for decomposition, allowing rapid dissolution of the fusion cake with hydrochloric acid to be achieved.Barredo and Diez9 used the same flux and added EDTA to improve the stability of the rock solution for measurements by atomic absorption spec- trometry. The same decomposition method has been applied to silicate rock analysis for major and minor elements by spectrophotometric and atomic absorption spectrometric flow injection analysis. 1 ~ 3 With the advent of modern electrothermal atomic absorp- tion spectrometry, the use of the lithium fluxes has been extended to the trace analysis of rocks and related materials In this work we attempted to determine trace amounts of nickel and cobalt in silicate rocks by graphite furnace atomic absorption spectrometry coupled with the fusion technique.Ammonium fluoride was used as a matrix modifier to remove the matrix effects that seriously impaired this determination. Results are given for the determination of both metals in a variety of standard rocks and these results are compared with the recommended values. Experimental Reagents A standard solution of nickel (1.002 mg ml-1 of Ni in 1 M hydrochloric acid) for atomic absorption spectrometry was obtained from Kanto Chemical Co. (Tokyo, Japan). The standard solution of cobalt (1.004 mg ml-1 of Co in 1 M hydrochloric acid) was prepared by dissolving 0.404 g of cobalt chloride hexahydrate in 1 M hydrochloric acid and diluting to 100 ml with the same acid.This solution was standardised by titration with EDTA disodium salt using xylenol orange as the metal indicator. Ammonium fluoride solution was prepared by dissolving 3.6 g of ammonium fluoride in 1 M hydrochloric acid and diluting to 100 ml with the same acid. This was prepared just before use. Apparatus A Shimadzu Model AA-646 atomic absorption spectrometer was fitted with a deuterium background corrector, a GFA-4 furnace atomiser and a Model U-135 chart recorder. Back- ground correction was carried out for all measurements. The radiation sources used were a Hitachi HLA-4S 2083004 multi-element hollow-cathode lamp (Cr, Cu, Fe, Mn, Ni) for nickel and a Hamamatsu TV L-233-27NU single-element hollow-cathode lamp for cobalt, The settings for the spectrometer were as follows: lamp current, 9 mA for cobalt and nickel; wavelength, 240.7 nm for cobalt and 232.0 nm for nickel; band width 0.19 nm; and readout as peak height. The atomisation programmes are summarised in Table 1.The protection gas used was argon. A regular high-density graphite tube (200-54520) and a pyrolytic Table 1. Analytical conditions Drying Ashing Atomisation Cleaning Ni c o Ni co Ni c o Ni c o Temperature/"C . . . . . . 150 500 700 2400 2600 2800 2900 Tim& 30 20 4 3 Mode . . . . . . . . Ramp Step Step Step gas flow-rate/lmin-1 . . . . 1.5 1.5 0 1.5 . . . . . . . .430 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL.1 graphite tube (200-54525) (Shimadzu) were used for cobalt and nickel , respectively. Procedure Place a 200-mg sample in a 30-ml capacity platinum crucible. Add 500 mg of 1 + 1 lithium carbonate - boric acid and mix. Dry the contents for 5 min with a gentle flame and then fuse the mixture for 15 min at 1000 "C. Dissolve the cooled melt in 1 M hydrochloric acid (with magnetic stirring) and dilute to exactly 50 ml with the same acid. Store in an air-tight polyethylene bottle. Take a 5-ml aliquot of the rock solution and transfer it into a polystyrene vial of 13 ml capacity with a screw-cap. Add exactly 5 ml of the ammonium fluoride solution to the vial with an Eppendorf type of pipette and mix well.Inject 10 p1 of the mixture into the furnace and proceed according to the programme given in Table 1. Results and Discussion It can be expected that the fusion method will not suffer any matrix effects, because a uniform content of lithium borate should act as a spectrochemical buffer to minimise interfer- ences due to differences in the matrices. With respect to nickel and cobalt, however, the calibration graphs obtained with the blank flux solution and 1 M hydrochloric acid solution differed considerably and the values obtained for several standard rocks with both calibration standards were biased significantly from the recommended values. Therefore, we decided to modify the matrices with ammonium fluoride, which reacts with both silicic acid and the borate, leading to their -Ashing - -Atomisation -+ 0.2 B I I I I I 1000 1500 2000 2500 Temperature/"C Fig.1. Effect of ashing temperature (2400 "C atomisation) and atomisation temperature (500 "C ashin ) on the signal from nickel. A( A ) In 1 M HC1- 0.5 M NH4F solution $0.30 ng Ni); B( A) in 1 M HC1 - 0.5 M NH4F solution containing lithium borate flux (0.30 ng Ni); C(0) in 1 M HCl - 0.5 M NH4F solution containing a silicate rock (AGV-1; andesite) and lithium borate flux. Volume of solution injected, 10 p1. Solutions prepared in accordance with the Procedure regarding the concentration of each constituent 0.2 0) u C rn + g 0.1 n a 0 - Ashing - -Atomisation - I I I 500 1000 1500 2000 2500 Tem peratu re/"C Fig. 2. Effect of ashing temperature (2600 "C atomisation) and atomisation temperature (700 "C ashin ) on the signal from cobalt.A( A) in 1 M HCI - 0.5 M NH4F solution 6.80 ng Co); B( A) in 1 M HCl - 0.5 M NH4F solution containing lithium borate flux (0.80 ng Co); C(0) in 1 M HCI - 0.5 M NH4F solution containing a silicate rock (JB-1; basalt) and lithium borate flux. Volume of solution injected, 10 PI. Solutions prepared in accordance with the Procedure regarding the concentration of each constituent volatilisation during the ashing process. Fig. 1 shows the effect of ashing and atomisation temperatures on the response from nickel in (A) 1 M hydrochloric acid, (B) 1 M hydrochloric acid + flux and (C) a rock sample solution that is 1 M in hydrochloric acid and contains flux. In all instances ammo- nium fluoride was added prior to the atomic absorption measurements.From the ashing and atomisation curves it appears that the ashing and atomisation of nickel proceed similarly in the presence of fluoride for (A) nickel alone, (B) nickel and flux and (C) a rock sample solution (standard rock AGV-1; andesite). The peak height for the rock solution at 2400 "C (ashing at 500 "C) as calibrated with the standard solutions (A and B) reasonably accounts for its nickel content calculated from the recommended composition. Nickel can be determined accurately on the basis of a calibration graph constructed using a series of aqueous nickel standard solu- tions. Fig. 2 shows similarly the ashing and atomisation curves for cobalt. In this instance the behaviour of cobalt is also the same for cobalt alone, cobalt with flux and a rock solution (JB-1; basalt) in the presence of ammonium fluoride.Cobalt can be determined accurately again on the basis of a calibration graph constructed from a series of aqueous cobalt standard solutions at an ashing temperature of 700 "C and an atomisation temperature of 2600 "C. The addition of ammonium fluoride makes the excess of borate and silicic acid in the sample volatilise as boron trifluoride and silicon tetrafluoride , respectively. The matrices become simplified, being converted into a lithium chloride or fluoride. Similar decomposition and furnace atomic absorp- tion measurements without fluoride were conducted by Salles and Curtius.15 They found that the atomic absorption signal for lead and barium was higher in the blank than in aqueous media, and the opposite was found for cobalt, chromium and copper.The effect of silica was found to be significant particularly for barium, making it necessary to prepare barium calibration solutions containing silicic acid, and to match the concentrations of the samples. This is tedious and almost impracticable for the analysis of a variety of silicate rock samples. The results obtained for many different types of standard silicate rocks are given in Tables 2 and 3 for triplicate independent determinations. The fused masses of these rocks are always dissolved completely in 1 M hydrochloric acid to give clear solutions and remain unchanged for a long period of time . l o This implies satisfactory destruction of carbonaceous Table 2. Determination of Ni in standard rocks Recommended$ or proposed value, Rock* Average found, p.p.m.t p.p.m.17 JG-1 .. AGV-1 BCR-1 G-2 . . GSP-1 MAG-1 QLO-1 RGM-1 sco-1 SDC-1 SGR-1 . . 8.6, 7.1, 7.8( 7.8) . . 15.9,18.1,13.2 (15.7) . . 11.1,10.0,10.4(10.5) . . 3.5, 4.1, 4.0( 3.9) . . 9.4, 9.5, 8.6( 9.2) . . 52.1,56.0,64.8(57.6) . . 3.3, 3.3, 2.9( 3.2) . . 3.6, 2.5, 3.2( 3.1) . . 25.4,33.6,28.3(29.1) . . 44.0,47.9,48.5 (46.8) . . 24.6,23.2,27.7 (25.2) 6 15,17,18.5 10,13, 15.8 3.5,5, 5.1 9,10,12.5 54 5.5 6 30 36 34 * JG-1 = granodiorite; AGV-1 = andesite; BCR-1 = basalt; G-2 = granite; GSP-1 = granodiorite; MAG-1 = marine mud; QLO-1 = quartz latite; RGM-1 = rhyolite; SCo-1 = shale; SDC-1 = mica schist; SGR-1 = shale. t Values of three independent determinations. $ Recommended values are given in italics.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 431 Table 3.Determination of Co in standard rocks Rock* JB-1 . . JG-1 . . AGV-1 BCR-1 G-2 . . GSP-1 MAG-1 QLO-1 RGM-1 sco-1 SDC-1 SGR-1 Average found, p.p.m.t . . 37.3,38.0,41.0(38.8) . . 3.7, 4.2, 4.1 ( 4.0) . . 14.0,16.6,17.2(15.9) . . 29.4,33.8,38.2(33.8) . . 4.0, 5.1, 4.5( 4.5) . . 6.8, 8.1, 6.7( 7.2) . . 19.1,24.1,20.3(21.2) . . 7.3, 9.5, 6.7( 7.8) . . 2.7, 2.1, 2.3( 2.4) . . 10.4,11.3, 9.7(10.5) . . 16.8,20.0,20.6(19.1) . . 10.7,11.0, 9.1(10.3) Recommended$ or proposed value, p. p.m. 17 38.4 4 14.1, I S . 7 , I 6 36,36.3 4.6,5, 5.5 6.4,7, 7.8 20 7.4 2.3 I1 17 12.5 * JB-1 = basalt; JG-1 = granodiorite; AGV-1 = andesite; BCR-1 = basalt; G-2 = granite; GSP-1 = granodiorite; MAG-1 = marine mud; QLO-1 = quartz latite; RGM-1 = rhyolite; SCo-1 = shale; SDC-1 = mica schist; SGR-1 = shale.7 Values of three independent determinations. $ Recommended values are given in italics. materials found in mud and shales and of the chromite, ilmenite, tourmaline, chlorite minerals, etc., which are particularly resistant to acid dissolution. For comparison purposes, the working values17 for these standard rocks are also given in Tables 2 and 3. As several compilations have been published for presenting working values, two or three values are listed as recommended or proposed values. Because of difficulties in performing trace analyses, even the recommended or proposed values do not coincide with each other, although they are close, and it is not known which value is the most reliable.However, the present results are within the range of recommended and proposed values or are in reasonable agreement with the recommended or proposed values. The nickel content of QLO-1, RGM-1, SDC-1 and SGR-1 appears to differ from the proposed or recommended values. For the former two rocks, which are low in nickel, more data are required in order to establish recommended values. For SDC-1 and SGR-1 Abbey18 has given 47 p.p.m. for the former and 29 p.p.m. for the latter as usable values for nickel; these values are close to those reported here. The relative standard deviation (n = 10) obtained by replicate atomic absorption measurements of a standard rock solution (SCo-1, Ni 30 p.p.m.) was 8.9%, and a value of 7.2% was obtained for cobalt (SDC-1, Co 17 p.p,m.).The proposed method may have wide applicability to the determination of trace amounts of metals in silicate rocks, as can be seen from the variety of rocks listed in Tables 2 and 3. The sensitivities of the method are 14 pg of Ni and 25 pg of Co for 1% absorption. The method is rapid, taking only 20-30 min to complete the fusion, so that 1 h is sufficient for sample preparation prior to the atomic absorption measurements. References 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. Bennett, H., Analyst, 1977, 102, 153. Bock, R., “A Handbook of Decomposition Methods in Analytical Chemistry,” International Textbook Co., Glasgow , 1979. Ingamells, C. O., Anal. Chem., 1966, 38, 1228. Boar, P. L., and Ingram, L. K., Analyst, 1970,95, 124. Jeffery, P. G., and Hutchison, D., “Chemical Methods of Rock Analysis,” Third Edition, Pergamon Press, Oxford, New York, Toronto, Sydney, Paris and Frankfurt, 1981. Van Loon, J. C., and Parissis, C. M., Analyst, 1969, 94, 1057. Verbeek, A. A., Mitchell, M. C., and Ure, A. M.,Anal. Chim. Acta, 1982, 135, 215. Ornang, S. H . , Anal. Chim. Acta, 1969, 46, 225. Barredo, F. B., and Diez, L. P., Talanta, 1976, 23, 859. Kuroda, R., Ida, I . , and Kimura, H., Talanta, 1985, 32, 353. Kuroda, R., Ida, I., and Ogurna, K., Mikrochim. Acta, 1984, I , 377. Mochizuki, T., and Kuroda, R., Analyst, 1982, 107, 1255. Mochizuki, T., Toda, Y., and Kuroda, R., Talanta, 1982, 29, 659. Bettinelli, M., Anal. Chim. Acta, 1983, 148, 193. Salles, L. C., and Curtius, A. J., Mikrochim. Acta, 1983, 11, 125. Zhou, L., Chao, T. T., and Meier, A. L., Anal. Chim. Acta, 1984, 161, 369. Govindaraju, K., Geostand. Newsf., 1984, 8 , Special Issue. Abbey, S., Geostand. Newsl., 1980, 4, 163. Paper J6f40 Received May 27th, 1986 Accepted June 30th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100429

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

433 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Rapid Screening Method for the Determination of Platinum and Palladium in Geological Materials by Batch Ion-exchange Chromatography and Graphite Furnace Atomic Absorption Spectrometry Charles H. Branch* and Dawn Hutchison British Geological Survey, 64 Gray's Inn Road, London WC1X 8NG, UK A method is proposed for the rapid separation of platinum and palladium from the major elements in geological materials. Following digestion with aqua regia, platinum and palladium are adsorbed on to the anion exchanger Dowex 1-X8 in the batch mode. After washing and isolation, the resin is ignited and the residue dissolved in nitric acid followed by aqua regia. Platinum and palladium are determined simultaneously by dual channel atomic absorption spectrometry using graphite furnace atomisation. Keywords: Platinum and palladium determination; geological materials; ion-exchange chromatography; graphite furnace a tom isa tion; a tom ic a bso rp ti0 n spectrometry The abundance of platinum and palladium in geological materials is normally so low that pre-concentration is necessry prior to their determination. Various traditional pre-concen- tration procedures are available, including fire assay,lJ solvent extraction,s5 coprecipitation6 and column ion- exchange7J techniques.Determinative end-points include flame1.3-5 and electrothermal atomisation2.6>9 atomic absorption spectrometry, plasma atomic emission spectrometry7 and amperometric detection.8 In recent years methods have been published in which batch ion exchangelOJ1 and batch sorptionlzJ3 have been used to separate elements of interest.They were then either eluted for spectrophotometricl3 and flame atomic absorption analysis1O or a slurry of the chelating medium was sampled for electrothermal atomisation atomic absorption analysis. 11913 The chloro complexes of platinum and palladium are both strongly adsorbed on anion exchangers from dilute hydrochloric acid. This paper describes the development of a batch ion-exchange scheme for the rapid separation and pre-concentration of platinum and palladium from acid digests of geological materials. Platinum and palladium are determined simultaneously by carbon furnace atomic absorption spectrometry using automatic sample solution injection, Experimental Reagents All reagents were of analytical-reagent grade unless specified otherwise.Distilled water was used throughout. Hydrochloric acid, sp. gr. 1.18. Nitric acid, sp. gr. 1.42. Hydrogen peroxide, 100 volume. Hydrofluoric acid, 40%. Sodium peroxide, 98%. Aldrich Chemical Company. Dowex 1-X8 (chloride form), 100-200 mesh. Platinum stock solution, 1000 pg ml-1. Dissolve 2.2769 g of Johnson Matthey Specpure ammonium chloroplatinate in 10% V/V hydrochloric acid and dilute to 1000 ml with 10% V/V hydrochloric acid. Palladium stock solution, 1000 pg ml-1. Dissolve 1 .OOOO g of Johnson Matthey Specpure palladium metal sponge in the minimum volume of concentrated nitric acid. Add concen- trated hydrochloric acid and evaporate to a small volume. Repeat the evaporation with concentrated hydrochloric acid * Present address: Department of Trade and Industry, Ashdown House, 123 Victoria Street, London SWlE 6RB, UK.twice more. Finally, dilute to 1000 ml with 10% V/V hydrochloric acid. Equipment The equipment used consisted of a Camlab Variomag 15-place electronic stirrer, an Instrumentation Laboratory (IL) 951 dual-channel atomic absorption spectrometer with VDU printer readout, an IL 555 electrothermal atomiser fitted with a single-piece pyrolytically coated graphite tube, and an IL 254 Fastac autosampler for automatic sample injection into the graphite furnace. Development of the Method Batch ion-exchange separation Platinum and palladium metals readily form stable anionic chloride complexes that are retained on strongly basic anion-exchange resins from hydrochloric acid media.Amounts of 20 pg of platinum and palladium in 20-50 ml of hydrochloric acid (1-1070 VIV) were magnetically stirred with between 0.25 and 1 g of Dowex 1-X8 for times of between 5 and 60 min. After mixing, unabsorbed platinum and palladium remaining in solution were quantitatively determined by graphite furnace atomic absorption spectrometry. The results obtained showed that platinum and palladium were quantitatively adsorbed on the resin between all the experimental limits selected above. A 15-min stirring time was standardised upon. Platinum and palladium can be eluted from a column of Dowex 1 as ammine complexes using an ammonia - ammo- nium chloride mixture. 14 Initial results indicated quantitative recoveries of platinum and palladium from pure solutions using the batch mode and a 0.2 M ammonia - ammonium chloride mixture and a stirring time of 30 min.Preliminary results on standard reference materials showed a good correlation with certified values. However, when new bottles of resin were used low, erratic (30-60%) recoveries of platinum and palladium were achieved. In an effort to overcome this problem, the concentration of the ammonia - ammonium chloride eluent and the stirring times were increased. As the concentration of the ammonia - ammonium chloride increased, the sensitivity of the graphite furnace determination decreased markedly, masking any improved recoveries. To test higher concentrations of the eluent and to eliminate undesirable ammonium chloride sublimation15 pro- duced at the atomisation stage, ammonium chloride was decomposed with nitric acid.After evaporation of the eluate to dryness the residue was moistened with concentrated nitric434 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 acid and evaporated to dryness again. This residue was dissolved in aqua regia, evaporated, diluted and analysed by graphite furnace atomic absorption spectrometry. Eluate concentrations of up to 2 M ammonia - ammonium chloride were examined but low, erratic (30-60°/0) recoveries of platinum and palladium were still obtained. No explanation was found for the variable behaviour of different bottles of the same resin. As an alternative method of recovery, the resin containing the adsorbed platinum and palladium was ignited in a porcelain crucible and the resultant mass dissolved in nitric acid - aqua regia.Greater than 90% recoveries of platinum and palladium were achieved. The added advantages of this approach were that this solution contained no added salts, so background correction was unnecessary, and the sensitivity of platinum and palladium measurements in hydrochloric acid is better than that in ammonia - ammonium chloride solution. Procedure Weigh 2.5 g of sample into a silica crucible or porcelain dish and heat at 600 "C for 1 h. Transfer the ignited material into a 400-ml beaker, add 30 ml of concentrated hydrochloric acid, cover with a watch-glass and digest for 30 min on a hot-plate. Remove from the heat, allow to cool slightly, add 10 ml of concentrated nitric acid and mix well.When all visible signs of reaction have ceased, return the beaker to the hot-plate and heat gently until no further reaction can be seen. Rinse the watch-glass and the walls of the beaker with distilled water and evaporate to incipient dryness. Add 5 ml of concentrated hydrochloric acid and again evaporate to incipient dryness. Add 0.3 ml of concentrated hydrochloric acid (or sufficient to wet the whole residue), 15 ml of water and heat gently to dissolve all soluble salts. Transfer into a centrifuge tube, dilute to 30 ml ( ~ 1 % V/V hydrochloric acid), mix thoroughly and spin off the insoluble matter. Transfer the supernatant liquid into a 50-ml poly- propylene beaker containing a magnetic stirring bar. Add 0.3 g of Dowex 1-X8 (0.1 g of resin per 10 ml of solution) and mix on a magnetic stirrer for 15 min.Allow the resin to settle, then decant and discard the sample solution. Add 20 ml of 1% hydrochloric acid wash solution to the retained resin and again magnetically stir for 15 min. Allow the resin to settle, decant and discard the wash solution. Remove the stirring bar from the beaker, rinse with water and transfer the resin into a 10-ml porcelain crucible. Allow the resin to settle, draw off the excess of water, dry the resin on a hot-plate, then carefully ignite the dried resin. Allow the crucible to cool, add 1 ml of concentrated nitric acid and evaporate to a small volume. Add a further 1 ml of concentrated nitric acid and 3 ml of concentrated hydrochloric acid.Heat gently to promote a smooth reaction. When the reaction has subsided, increase the heat and evaporate to a small volume. Add 1 ml of concentrated hydrochloric acid and evaporate to a small volume (approximately 0.1 ml). Add 5 ml of water (a smaller volume may be used if required) and determine platinum and palladium using the furnace pro- gramme outlined in Table 1. Instrumental parameters By experimentation it was found that the optimum atomisa- tion temperatures for platinum and palladium using the graphite furnace were 2750 and 2600 "C, respectively. For simultaneous measurement of the two elements the higher temperature was chosen. Similarly, the maximum ashing temperature that could be used for the two elements was determined by the more volatile palladium and was fixed at 950 "C.The most sensitive platinum line at 265.9 nm was used for all platinum measurements. The most sensitive palladium lines at 244.7 and 247.6 nm are about ten times more sensitive than the platinum line. The palladium line at 340.5 nm is five times less Table 1. Graphite furnace operational parameters for the simul- taneous determination of platinum and palladium Stage Automated operation Drying . . . . . . . . 175 "C: deposition time Ashing . . . . . . . . 175 to 550 "C: 25 s 550 to 950 "C: 15 s Step atomisation . . . . . . 950 to 2750 "C: 5 s hold sensitive than the other palladium lines but is still twice as sensitive as the platinum line. It is more intense than the other two palladium lines and a more stable base line can be obtained.This line was used unless stated otherwise. The lamp currents and band passes used were those recommended by the manufacturer. The ultimate sensitivity of the platinum and palladium determination depends on the age of the graphite tube and the deposition time selected on the autosampler. As a guide, practical limits of determination (signal equal to three times the background noise) using a 10-s deposition time (equivalent to approximately 20 ~l of solution) were 0.01 pg ml-1 for platinum and 0.005 pg ml-1 for palladium. For a 2.5-g sample mass and a final analytical solution of 5 ml, these represent practical limits of determination of 20 ng g-1 of platinum and 10 ng g-1 of palladium. Results and Discussion Analysis of Geological Materials PTO-1 It is difficult to find certified reference materials with reasonable levels (pg g-1) of platinum group metals (PGMs) that are not atypical and do not contain abnormally high levels of copper, nickel, iron or sulphur.The South African National Institute of Metallurgy standard PTO-1 approximates to an ultrabasic rock type and contains 3.74 pg g-l of platinum and 1.53 pg g-1 of palladium. The PGMs occur as discrete PGM minerals and are also present in solid solution in the sulphides of iron, nickel and copper. An aqua regia dissolution should be sufficient to solubilise all the platinum and palladium present. PTO-1 as supplied came in four separate splits. Each split was mixed on an inversion mixer for 1 h prior to taking a 2.5-g sample mass. Hydrogen peroxide can be used as an alternative to nitric acid in conjunction with hydrochloric acid for the in situ generation of chlorinating species.It has the advantage of giving a more complete reaction in the cold than nitric acid and any free hydrogen peroxide readily breaks down on heating to give water and oxygen. Table 2 shows the results obtained for PTO-1 using a hydrochloric acid - hydrogen peroxide dissolu- tion. Inductively coupled plasma source mass spectrometry (ICP-MS) results on the same sample extracts are shown for comparison. Two sequential batch resin extractions were carried out on each sample solution to check the exchange capacity of the resin in the presence of major concentrations of other elements. There is generally good agreement between the GFAAS and ICP-MS measurements.The low recoveries for the second resin extraction (approximately 1% of the first extraction) indicate that only a single resin extraction is necessary for real sample solutions. As the values obtained were lower than the certified values for PTO-1, the exercise was repeated using a true aqua regia mixture, i. e. , 3 parts of hydrochloric acid and 1 part nitric acid. The residue from the primary aqua regia extraction was digested with HF in a PTFE beaker followed by aqua regia. The residue from this digestion was fused with sodium peroxide in a nickel crucible after the method of Pitts et al.16 We had found in earlier experiments that the nickel oxide obtained on leaching the fusion cake in water contained all theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL.1 435 ~~ ~~ Table 2. Platinum and palladium results for PTO-1 using a hydrochloric acid - hydrogen peroxide dissolution Pt/pg g-1 Pdlpg g- GFAAS * ICP-MSt GFAAS* ICP-MSt Split 25 1$ 25 15 2$ 15 25 1 2.71 0.03 3.16 0.02 1.12 N.d. § 1.15 N.d. 2 2.88 0.03 3.04 0.03 1.15 N.d. 1.14 N.d. 3 2.74 0.04 3.12 0.04 1.13 N.d. 1.24 N.d. 4 2.79 0.02 3.04 0.02 1.28 N.d. 1.21 Ned. Mean 2.78 0.03 3.09 0.03 1.17 - 1.19 - * GFAAS = graphite furnace atomic absorption spectrometry. t ICP-MS = inductively coupled plasma source mass spectrometry. $ 1 and 2 = sequential resin extractions. § N.d. = not detected. Table 3. Platinum and palladium results for the phase dissolution of PTO-1 Pt/pg g-1 Pdlyg g-1 ~ Primary, HNO, - HCl Split GFAAS ICP-MS 1 3.38 3.42 2 3.48 3.69 3 3.80 4.00 4 3 .OO 3.14 Mean 3.42 3.56 Secondary, HF, GFAAS 0.15 0.11 0.13 0.11 0.12 Tertiary, N@*, GFAAS N.d. 0.06 0.06 0.03 0.04 Primary, HN03 - HC1 GFAAS ICP-MS 1.31 1.33 1.25 1.24 1.44 1.53 1.38 1.27 1.35 1.34 Secondary, HF, GFAAS 0.13 0.14 0.15 0.15 0.14 Tertiary, N a A , GFAAS 0.03 0.03 0.03 0.03 0.03 available platinum and palladium.Therefore, the nickel precipitate from the sample fusion was separated from the alkaline solution, acidified with hydrochloric acid, evaporated to small volume and then treated with aqua regia. Each of the three solutions obtained above was analysed separately. A single resin extraction was made on each solution and the results of the analyses of the separate dissolutions are shown in Table 3. The ICP-MS results on the same solution are shown for comparison.The recoveries obtained using a primary nitric acid - hydrochloric acid dissolution were better than those achieved with a hydrogen peroxide - hydrochloric acid digest and show good agreement with certified values. As virtually all the platinum and palladium occurred in the first two digestion phases, the exercise was repeated using a single, combined HF - aqua regia dissolution. Table 4 shows the results obtained for platinum and palladium. Palladium results obtained using discrete nebulisationl7 are also shown. Measured aliquots of the residual analytical solution follow- ing graphite furnace analysis were evaporated to dryness on a steam-bath, the residues dissolved in 0.5 ml of 1% V/V hydrochloric acid and 200-p1 duplicates taken for flame atomic absorption analysis.By experiment 200 pl of a standard solution had previously been shown to give a peak-height signal equivalent to the steady-state signal when continuously aspirating the standard solution. The most sensitive palladium wavelength at 244.7 nm was used. The sensitivity for platinum was not good enough to permit its determination by discrete nebulisation. The mean values for platinum and palladium compare favourably with the mean totals of the primary and secondary attacks in Table 3. Discrete nebulisation values obtained using flame atomic absorption, although slightly lower, are also comparable. Other materials The availability of other PGM certified reference materials is limited to single-type materials. Those available are the Canadian certified materials PTA-1 (platiniferous black sand), PTC-1 (flotation sulphide concentrate) and PTM-1 Table 4.Platinum and palladium results using an HF - aqua regia dissolution Pdlpg g- I Pthg g- 7 Discrete Split GFAAS GFAAS nebulisation 1 3.37 1.34 1.28 2 3.37 1.60 1.54 3 3.59 1.45 1.41 4 3.52 1.53 1.42 Mean 3.46 1.48 1.41 (nickel copper matte). The samples were ignited and dissolved in HF - aqua regia. During ignition the samples were stirred at regular intervals to prevent clotting. Any residue was fused with sodium peroxide as described previously and analysed separately. The samples were analysed in duplicate and the results are shown in Table 5 . Inductively coupled plasma mass spectrometry values, after conventional fire assay procedures, are shown for comparison.The precision of the duplicate determination of platinum in PTA-1 is poor. The mean value of 2.85 pg g-1 is close to the certified value and could therefore reflect sample inhomo- geneity and the fact that a 2.5-g sub-sample mass is not large enough in this instance. The precision for the other samples is good, although the palladium values for PTC-1 and the platinum values for PTM-1 are low relative to the certified values. However, the mean value of 4.4. pg g-1 obtained for platinum in PTM-1 does agree with that recently reported by Kritsotakis and Tobschall.18 It was thought possible that the exchange capacity of the resin had been exceeded. A second batch ion-exchange extraction was carried out on the primary sample solutions but no additional platinum recoveries were found for PTM-1, although an additional 10% of platinum and palladium were found for PTC-1.This improved the accuracy of the determination of platinum in PTC-1 but made little difference to the palladium accuracy. Because of the abnormally high percentages of Cu, Ni and Fe in these samples, it was suspected that there could be a suppressive effect during graphite furnace atomisation caused by these metals being present in the final analytical solutions.436 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Table 5. Platinum and palladium values for PTA-1, PTC-1 and PTM-1 Pt/pg g-1 Pd/pg g -I This method Certified ICP-MS, Sample value HF - aqua regia Na202 fire assay N.d. - PTA- 1 3.05 3.7 PTC-1 3.0 2.7 N.d.2.4 PTM-1 5.8 4.3 N.d. 5.1 2.0 N.d. 2.5 N.d. 4.5 N.d. This method Certified ICP-MS, value HF - aqua regia Na202 fire assay N/A* 0.03 N.d. - 0.04 N.d. 12.7 6.8 0.02 11.7 7.1 N.d. 8.1 7.9 N.d. 7.7 8.0 N.d. * N/A = no certified value available. The Cu, Ni and Fe concentrations in the final analytical solutions were measured by flame atomic absorption spec- trometry. Standard platinum and palladium solutions contain- ing these levels of Cu, Ni and Fe were then analysed by graphite furnace atomic absorption spectrometry. Relative to pure standard solutions the recoveries ranged from 97 to 104% for platinum and from 94 to 104% for palladium. There are no reference materials that contain low (ng g-1) levels of PGMs. To prepare low-level standards by dilution of higher level powdered standards is not feasible owing to the inherent problems of sample and diluent density differences and hence diluted sample inhomogeneity.In a recent paper by Gladney et aZ.,19 platinum and palladium values were quoted for the USGS Geochemical Exploration Reference Samples GXR-1 to GXR-6 and were far higher than those expected in this type of material. On checking the original source20 of these values, it was found that the platinum results were based on a spectrophotometric technique and the palladium values on either a colorimetric field or an emission spectrographic method. A recent publica- tion by Govindaraju21 gave a compilation of working values for 170 reference materials including the GXRs, but no platinum and palladium figures were quoted for the GXRs.The highest values quoted were 10 ng g-1 for platinum in PCC-1 and 14 ng g-1 for palladium in W-1. Using the method presented in this paper, the amounts of platinum and palladium found in the GXRs were both less than 10 ng 8-1. By inference these values are in agreement with those in Govindaraju’s compilation. Conclusion A rapid screening method has been presented for the identification of platinum and palladium anomalies in geolo- gical materials. If required, absolute values of these metals in anomalous samples can then be determined by a more traditional method using larger sample masses and a fire assay procedure. data. This paper is published with the permission of the Director, British Geological Survey, NERC. 1. 2. 3. 4. 5 . 6. 7. 8.9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. References Moloughney, P. E., Talanta, 1980, 27, 365. Aruscavage, P. J., Simon, F. O., and Moore, R., Geostand. Newsl., 1984, 8, 3. Diamantatos, A., Anal. Chim. Actu, 1983, 147, 219. Pohlandt, C., Talanta, 1979, 26, 199. Pohlandt, C., and Hegetschweiller, H., S. Afr. Nut. Znst. Metall., Rep., No. 1940, 1978. Sighinolfi, G. P., Gorgoni, C., and Mohamed, A. H., Geostand. Newsl., 1984, 8, 25. Brown, R. J., and Biggs, W. R., Anal. Chem., 1984, 56,646. Rocklin, R. D., Anal. Chem., 1984,56, 1959. Haines, J., and Robert, R. V. D., S. Afr. Counc. Miner. Technol., Rep., No. M34, 1982. De Mora, S. J., and Harrison, R. M., Anal. Chim. Acta, 1983, 153, 307. Isozaki, A., Kumagai, K., and Utsumi, S., Anal. Chim. Acta, 1983, 153, 15. Slovak, Z., and Docekal, B., Anal. Chim. Acta, 1980,117,293. Terada, K., Matsumoto, K., and Inaba, T., Anal. Chim. Actu, 1984, 158, 207. MacNevin, W. M., and Crummett, W. B . , Anal. Chem., 1953, 25, 1628. Barredo, F. B., Polo, C. P., and Diez, L. P., Anal. Chim. Acta, 1977, 94,283. Pitts, A. E., Van Loon, J. C., and Beamish, F. E., Anal. Chim. Acta, 1970, 50, 181. Cresser, M. S., Prog. Anal. At. Spectrosc., 1981, 4, 219. Kritsotakis, K., and Tobschall, H. J., Fresenius 2. Anal. Chem., 1985, 320, 15. Gladney, E. S . , Burns, C. E., and Roelandis, I., Geostand. Newsl., 1984, 8, 119. Allcott, G. H., and Lakin, H. W., U.S. Geol. Surv., Open-File Rep., No. 78-163, 1978. Govindaraju, K., Geostand. Newsl., 1984, 8, 3. The authors thank A. R. Date of the Analytical Chemistry Research Group, British Geological Survey, for the ICP-MS Paper J6l31 Received April 29th, 1986 Accepted June 23rd, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100433

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 437 Determination of Silicon in Gallium Arsenide bv Electrothermal Atomisation Atomic Absorption Spectrometry Using the L'vov PI atfo r m * Marco Taddia University of Bologna, "G. Ciamician " Chemical Institute, Via Selrni 2, 1-40 726 Bologna, ltaly Electrothermal atomisation atomic absorption spectrometry with a L'vov platform was investigated for the determination of silicon in gallium arsenide. The sample was decomposed with nitric acid - hydrochloric acid (3 M + 1 M), calcium nitrate was added as a matrix modifier and the diluted solution was atomised from a tantalum carbide coated tube and platform. Ashing at 1650 "C and deuterium arc background correction allowed the non-specific absorbance to be eliminated. For a 40-mg sample the detection limit (3a) is 4.0 yg g-1, equivalent to 5 x 1017 atoms cm-3.Results obtained for silicon-doped gallium arsenide samples are presented. Keywords: Electrothermal atomisation atomic absorption spectrometry; platform atomisation; gallium arsenide analysis; silicon determination One of the most valuable materials for advanced microelec- tronics is gallium arsenide, its importance being related to a wide range of applicability arising from its energy band structure. Owing to the capability of GaAs integrated circuits to work in a speed range that is not otherwise available at present, they are currently used for high data rate communica- tion, in high-speed computers, etc. Silicon may be an undesirable impurity in GaAs crystals grown in quartz containers.' Conversely, it may be delibe- rately added to obtain either n- or p-type doped material showing specific properties.Silicon in GaAs is usually determined by secondary ion1.2 and spark-source mass spec- trometry.3 Recently, methods based on differential-pulse polarography and spectrophotometry as heteropoly-acids have been proposed.4 Although atomic absorption spectrometry with electrother- mal atomisation (ETA-AAS) is a well established technique in GaAs analysis,sfj currently only a limited amount of research on silicon determination has been conducted.7J A detection limit of 5 X 1015 atoms cm-3 has been reported,7 but the small number of data make the evaluation of this result difficult. Another procedure8 requires drying and ashing operations to be repeated several times, which decreases its convenience.These facts reflect the difficulties that can arise in this kind of determination. High background molecular absorbance, pro- duced by species such as GaC1,S is hard to compensate for in some instruments by using a deuterium lamp.9 A further complication arises from the formation of Sic(,, and SiO,,,.1° The condensed-phase dissociation of the carbide has been suggested" to be the rate-limiting step for the atomisation of silicon in uncoated tubes. On the other hand, volatilisation of silicon as SiO has been considered responsible for the anomalous decrease in the silicon signal after ashing at 1600 K, long before the appearance temperature of 2150 K.l0 Platform atomisation for the determination of silicon was first proposed by Frech and Cedergrenlo in order to reduce the losses of silicon as volatile compounds.The sensitivities obtained by atomising under isothermal and non-isothermal conditions at 2900 K were compared. It was found that the sensitivity improved by a factor of 2.5 under isothermal conditions. In this manner, Frech and Cedergrenlo provided reasonable evidence that platform atomisation helps to reduce the gradual removal of SiO during the ashing step. Further, the reactivity of the graphite substrate in determin- ing the oxygen partial pressure inside the furnace must be * Presented at the 24th CSI, Garmisch-Partenkirchen, FRG, September 1985. taken into account.12 For this reason we carried out a preliminary investigation13 to examine the dependence of the silicon absorbance on the inert gas flow-rate, ashing tempera- ture and the amount injected for uncoated, pyrocoated and tantalum carbide coated graphite tubes.Atomisation from uncoated and, to a greater extent, from pyrocoated tubes yielded calibration graphs showing anomalous bending near the origin, curving upwards. In contrast, with TaC-coated tubes a linear response was observed with the injection of up to 32 pg of silicon. These results and others related to the different dependence of the signal on the gas flow-rate further indicate the role of oxygen in determining the sensitivity. Some hypotheses were made13 on the basis of possible reactions between the species involved, with attention paid to the ternary phases Ta - 0 - C and Ta - Si - C and to the kinetics of silicon atoms release from carbide-coated t~bes.14~15 A comparison of the results obtained for the different substrates examined indicated that sampling from TaC-coated surfaces should provide reliable results in the determination of silicon.This paper reports a method for the determination of silicon in Si-doped GaAs samples, based on isothermal atomisation from a TaC-coated tube and platform using calcium nitrate as a matrix modifier. The conditions for the adequate elimina- tion of non-specific absorbance have been systematically investigated. Experimental Apparatus A Perkin-Elmer 372 atomic absorption spectrometer equipped with an HGA-500 graphite furnace and a deuterium arc background corrector was used. The operating parameters are given in Table 1.A Pye Unicam silicon hollow-cathode lamp (19 mA, 251.6 nm) was employed at the minimum spectral band width of 0.2 nm. With the above operating current, which was 1.3 times the maximum recommended by Table 1. HGA-500 graphite furnace conditions Step Temperature/"C Ramp/s Hold/s D r y . . . . . . 220 1 30 A s h . . . . . . 1650 1 20 Atomise* . . . . 2700 0 8 Cool . . . . 400 1 10 Clean . . . . 2750 0 4 Cool . . . . 20 1 30 * The argon flow-rate during this step was reduced to 70 ml min- l .438 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 the manufacturer, the signal to noise ratio improved by a factor of 1.7. No loss of sensitivity was caused by increasing the current, Running this lamp with a higher current, in order to achieve a reasonable signal to noise ratio, may be peculiar to the PE 372 instrument and similar spectrometers having a relatively poor light throughput.Tantalum carbide coated tubes and platforms were pre- pared from standard uncoated tubes and pyrolytic L'vov platforms (PE 121091) by following the method reported by Zatkal6 with slight modifications.13 These included the reduction to a half of the total concentration of oxalic acid in the tantalum soaking solution and the execution of a single tube treatment instead of two. The latter step permitted the resistivity change of the graphite tube to be controlled, making a re-calibration of the optical pyrometer unnecessary. In order to minimise the high furnace blank resulting from the tantalum treatment, a number of thermal cycles varying from five to ten was normally required.The sensitivity to silicon, which is a measure of the tantalum coating reproducibility, varied by as much as 30% from tube to tube. Molecular absorption spectra were obtained by using a hydrogen hollow-cathode continuum source, because the deuterium arc cannot operate independently for this purpose on the PE 372 spectrometer. The 249.2 nm wavelength from a Perkin-Elmer copper hollow-cathode lamp was used to check the efficiency of the deuterium background correction system under the conditions described. Contamination control is a serious problem in the deter- mination of silicon at trace levels. Polypropylene and PTFE were acid washed. An Abich steel mortar (Prolabo 749322) was used to grind the GaAs samples.We confirmed that the mortar did not introduce detectable silicon impurities by processing Si-free samples. Reagents A stock standard solution (1000 mg 1-1) was prepared by diluting Merck titrisol standard SiC14 solution in NaOH with de-ionised, distilled water. This standard is guaranteed to contain 1.000 +_ 0.002 g of Si and is currently used for the ETA-AAS determination of Si.14 Working standards (40.0 pg ml-1 of Si) were freshly prepared from the stock solution by appropriate dilution. Suprapur nitric and hydrochloric acids (Merck) were used to dissolve the GaAs. Tantalum powder (99.7% mlm) (Riedel de Haen) and hydrofluoric acid (Baker Analysed MOS grade reagent) were employed to prepare the tantalum soaking solution. Procedure Wash the gallium arsenide sample with acetone to remove trace amounts of grease and grind it in a steel mortar for faster acid attack.Accurately weigh 40.00 mg of pulverised sample in a PTFE test-tube. Add 2 ml of nitric acid - hydrochloric acid (3 M + 1 M), close the tube and heat it at 100 "C for 2 h. After dissolution add 50 p1 of 5% mlV calcium nitrate solution and 2 ml of de-ionised distilled water. Atomise at least four replicate 10-pl portions of the sample solution and measure the peak height. Determine the silicon concentration from a calibration graph obtained by using a Si-free sample. Results and Discussion Background Studies The absorption spectrum of the decomposed sample solution (Fig. 1) was obtained point by point by measuring the absorbance signal for a hydrogen source at different wavelengths during atomisation at 2700 "C after ashing at 1000 "C.The ashing temperature was deliberately kept low in order to retain the bulk of the matrix before atomisation. The spectrum shows the maximum absorbance between 249 and 251 nm, where the band system of the diatomic species GaCl occurs. These bands are caused by the electronic transition C13.r; - x12+.17 The spectrum is similar to that reported by Dittrich,l8J9 who vaporised the sample from the uncoated graphite wall and first assigned the bands to the above species. Nevertheless, it is difficult to establish the GaO and GaOH contributions to the above spectrum. The monoxide seems to absorb in the same region (maximum absorbance at 240 nm),lSJ9 although there is a discrepancy with regard to the results of other workers.17 Moreover the possible presence of a hydroxide molecule, GaOH (Do = 4.4 eV),20 already hypothesised in flames20J1, cannot be overlooked.For the reasons stated above, no substantial advantages were 0 240 250 260 Wavelength/nrn Fig. 1. Absorption spectra produced by 100 pg of GaAs (HN03 - HC1) vaporised from a tantalum-carbised platform. Absorbance measured at 2700 "C after ashing at 1000 "C 0.4C 0.30 a) C (D + 2 0.20 a 0.10 0 1200 1600 2000 2400 Fina I ashi ng tern pe ratu re/"C Fig. 2. Effect of ashing temperature on the peak absorbance of 10 ng of Si + 100 pg of GaAs (HNO - HC1) + 6 pg of Ca at (A) 249.2 nm and (B) 251.6 nm, without blckground correction using the L'vov platformJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL.1 439 achieved by dissolving the GaAs sample in nitric acid alone. According to other literature spectra18 the peak molecular absorbance was lower but the volatility of the matrix, in the absence of chloride, was also reduced. The optical correction of high molecular absorbance by means of a conventional deuterium lamp is difficult to achieve as the degree of overlap between the analyte atomic spectral line and the individual molecular rotation line is normally unknown.9 As a conse- quence, the ashing step parameters must be carefully opti- mised in order to remove the bulk of the matrix before atomisation. Preliminary attempts to eliminate the non-specific signal by increasing the ashing temperature gave an indication of relevant silicon losses.Therefore, calcium, which was already known to enhance the silicon signal,*2 was added as a matrix modifier. The role of calcium is discussed under Matrix Modifier. The effect of the ashing temperature on the absorbance at 249.2 and 251.6 nm, without background correction, is shown in Fig. 2. The background signal is eliminated between 2000 and 2100 "C, where the silicon signal is strongly reduced even in the presence of the matrix modifier. It appears evident that satisfactory results cannot be achieved without background correction. The effect of the ashing temperature on the background-corrected absorbance at both wavelengths is shown in Fig. 3. The background signal at 249.2 nm is entirely compensated for when ashing at 1550 "C.In order to illustrate the benefits of platform atomisation, we also show, in Fig. 3, the results obtained by measuring the corrected absorbance at 249.2 nm as a function of the ashing temperature, when samples were atomised from TaC-coated tubes without a platform. In this instance a small non-specific signal ( A = 0.018 Ifi 0.006, n = 3) still persists after ashing at 1550 "C. Perhaps the higher temperature in the vapour phase resulting from the platform atomisation promotes more extensive dissociation of GaCl, GaOH and GaO. The relatively low background signal resulting from the wall atomisation after ashing at 1000 "C may be caused by the time lag between the heating of the tube and the platform. With regard to the results shown in Fig. 3, it should be noted that the signal at 249.2 nm is not an accurate measure of the molecular absorption at the silicon wavelength. However, as the molecular absorption at 249.2 nm is greater than that at 251.6 nm (Fig.l ) , if the background system can cope with the molecular absorption at 249.2 nm it will be able to cope with the lower background absorption at the silicon wavelength. It is an interesting coincidence that the copper line used for background measurements identifies with the GaCl maximum reported in the literature.17 As can be seen from Fig. 3, the background signal is eliminated at 1550 "C. Although a certain loss of the analyte occurs at 1650 "C, we found it preferable to ash at this temperature in order to compensate for the increase in resistance observed during the ageing of the tube, which would result in a lower actual tube temperature during the ash stage.This can probably be ascribed to a partial loss of tantalum carbide or poorer contact with the graphite end cones. With the above precaution the signal at 251.6 nm given by the Si-free GaAs samples corresponded to that given by the nitric acid - hydrochloric acid mixture alone during the whole lifetime of the tube. This was considered to be a reliable check of efficient background removal. Matrix Modifier A rapid investigation to establish the optimum calcium concentration was performed. First we tested, in the absence of GaAs, the value reported by Thompson et al.,22 which corresponded to 1000 pgml-1 of calciumin the sample solution, then 500 and 2000 yg ml-1 of calcium.The results indicated that the enhancement of the silicon signal (+36%) given by 500 pg ml-1 should not be improved by further increasing the calcium concentration. The percentage enhancement depen- ded on the ashing temperature (+60% at 1000 "C), but we were forced to ash at 1650 "C in order to eliminate the background signal. Between the two mechanisms proposed to explain the calcium effect,22 that related to the reduction of p(02) in the gas phase also seems consistent with our recent experiences.13 Thus calcium, which is probably atomised via the thermal dissociation of the oxide, should work as an oxygen scavenger and reduce the silicon losses as SiO(g).23 Further experimental evidence suggesting that calcium probably behaves as an oxygen scavenger was gained from a comparison between the calibration graphs obtained by atomising silicon standards from pyrocoated tubes (Fig.4). Pyrocated tubes were selected as it is known13.23 that with these tubes interference from oxygen is greater than with normal uncoated tubes. The uncoated tubes have been shown to be more reactive towards the oxygen at lower tempera- tures.12 With the pyrocoated tubes, the linear portion of the calibration graph shows a dramatic displacement with respect to the origin (Fig. 4). This reflects a mechanism similar to that postulated for tin,23 which involves the capture of the oxygen by the silicon atoms. On adding calcium the bending was eliminated, indicating a higher atomisation efficiency of silicon. This may reasonably be considered to be due to the action of calcium atomised directly from the surface of the tube.Before considering calcium as a matrix modifier, mag- nesium nitrate was first tested. The enhancement effect was less pronounced, presumably owing to the lower thermal stability of MgO. In fact, the calculated thermal dissociation temperature of MgO within the tube is 1750 "C, whereas for calcium it is 1950 "C.23 r 1 0.20 1000 1400 1800 2200 Final ashing temperature/"(= Fig. 3. Effect of ashing temperature on the background-corrected peak absorbance of 6 ng of Si + 100 p of GaAs (HN03 - HCI) + 6 pg of Ca. Measurements were made at ?A) 249.2 nm and (B) 251.6 nm with the platform and at (C) 249.2 nm without the platform 0 10 20 30 Sihg Fig. 4. form): (A) without calcium; and (B) with 5 pg of Ca Calibration graphs for silicon (pyrolytic graphite, no plat-440 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 A problem that arose during the optimisation of the thermal programme was overheating of the furnace thermal block due to the high temperatures required. A cooling step at 400 "C inserted between the atomisation and cleaning steps alleviated the problem.The above relatively high cooling temperature was chosen in order to keep the platform incandescent and to facilitate the subsequent cleaning at 2750 "C. Calibration and Matrix Effect Calibration graphs were obtained by adding known amounts of standard to gallium arsenide solutions (10 mg ml-1) prepared from silicon-free samples. The linear range extends up to 1.4 pg ml-1 of silicon, which corresponds to 14 ng of silicon injected into the furnace (Fig.5). The least-square correlation coefficient for an eight-point graph was 0.999 and the characteristic amount of silicon was 0.15 ng. It should be emphasised this corresponds to only a 35% decrease in sensitivity with respect to the pure standards atomised from uncoated tubes (internal gas flow-rate 50 ml min-I), as reported in the Operator's Manua1.24 A comparison of the slope for the decomposed sample solution with that of the calibration graph using HN03 - HC1 (1.5 M + 0.5 M) yields a ratio that is a measure of the so-called matrix effect under the measurement conditions described here. The ratio was 0.84 with calcium present and 0.61 without calcium (Fig. 5). Therefore, the beneficial effect of the matrix modifier is further indicated.Using peak-area measurements the characteristic amount of silicon was almost the same (0.14 ng) and the slope ratio (Fig. 6 ) was 0.71. The detection limit (30, n = 10) for silicon in gallium arsenide was 4.0 yg g-1 (peak height), which is equivalent to 5 X 1017 atoms cm-3. By measuring the integrated absorbance the detection limit was 8.0 pg g-1. The above results were related to the lower reproducibility of the peak area measure- ments (relative standard deviation = 30%, n = 10) at the blank level (0.10 vg ml-1 of silicon), as a consequence of the significant difficulty in zeroing the instrument. Further, high furnace blank values given by old TaC-coated tubes, previ- ously used for silicon, may adversely affect the detection limit.This memory effect probably arises from a partial retention of the analyte by the tube on atomisation. Presumably the reason for the retention of silicon is the formation of a ternary phase 0.50 0.40 Q, 0.30 0 a e s n a 0.20 o.ia 0 4 8 12 16 Si/ng Fig. 5. Calibration graphs for silicon obtained by measuring the peak absorbance: (A) in 1.5 M HN03 + 0.5 M HC1; (B) with 100 pg of GaAs + 6 pg of Ca; and (C) with 100 pg of GaAs. Using the L'vov platform. The blank was subtracted from the sample readings Ta - Si - C, which is known to exist in the 0.5-10% carbon range, with stability at 5% carbon.25 In conclusion, the peak-height measurement mode was preferred as it gave a better detection limit and a lower matrix effect. Accuracy The accuracy was tested by adding known amounts of determinant to pulverised Si-free GaAs before dissolving the samples.Three series of replicate determinations were carried out on 40-mg samples spiked with 0.40,0.80 and 1.60 pg of Si. The recoveries were 0.39 2 0.05 pg ( n = 4), 0.80 k 0.11 pg (n = 7) and 1.59 k 0.04 pg (n = 3), respectively. Analysis of Real Samples The method was applied to the determination of silicon in Si-doped GaAs samples. The results obtained are presented in Table 2 referring to different zone lengths solidified (ylL, where y = seed end distance and L = total crystal length); the relative standard deviation for the over-all procedure is 11-16%, depending on the silicon concentration. The results confirm how, during the growth of a GaAs crystal from a melt, silicon impurities are rejected by the solid and are accumulated by the liquid ( K < 1, where K = distribution coefficient) .26 In conclusion, by using TaC-coated tubes and platforms, the ETA-AAS technique may be successfully applied to the analysis of silicon-doped GaAs samples, provided that matrix matched standards are used.A combination of platform atomisation, matrix modification and TaC coatings did not totally remove the interference of the GaAs matrix on silicon, even when peak-area measurements were used, although the O.*O I 0.60 v) 3 0.40 .- v) 0.20 0 8 16 Silng 4 Fig. 6. Calibration graphs for silicon obtained by measuring the integrated absorbance: (A) in 1.5 M HN03 + 0.5 M HCl; and (B) with 100 pg of GaAs + 6 vg of Ca using the L'vov platform.The blank was subtracted from the sample readings Table 2. Determination of silicon in Si-doped GaAs samples Silicon found/ No. of Sample YlL pg g-1* determinations A . . . . . . 0.27 10 f 1.6 14 B . . . . . . 0.40 13 k 1.9 6 C . . . . . . 0.61 16 k 2.3 7 D . . . . . . 0.81 37 k 4.0 5 * Mean k standard deviation.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 441 extent of interference was reduced. The detection limit will probably be improved by analysing a larger sample mass, but this will require the application of Zeeman-effect background correction27 or a spectrometer with a more efficient deuterium system. Even so, the analysis of undoped or semi-insulating material (9-300 ng g-1 of Si)7 will require pre-concentration procedures.This work was supported by the National Research Council (CNR) under a contract related to the Progetto Finalizzato per la Chimica Fine e Secondaria. The author thanks Dr. L. Zanotti (CNR-MASPEC, Parma, Italy) for providing gallium arsenide samples and Mr. J. Galitzin for correcting the English in the manuscript. 1. 2. 3. 4. 5. 6. 7. 8. References Thomas, R. N., Hobgood, H. M., Barrett, D. L., and Eldridge, G. W., in Rees) G. J., Editor, “Semi-insulating 111-V Materials,” Proceedings of Meeting, Nottingham, 1980, Shiva, Nantwich, Cheshire, p. 76. Homma, Y., Kurosawa, S., Yoshiota, Y., Shibata, M., Nomura, K., and Nakamura, Y., Anal. Chem., 1985,57,2928. Yemenidjian, N., and Lombos, B. A., J . Cryst. Growth, 1982, 56, 163. Liu, R.-S., and Yang, M.-H., paper presented at the 24th Colloquium Spectroscopicum Internationale, Garmisch- Partenkirchen, FRG, 1985.Dittrich, K., Mothes, W., and Weber, P., Spectrochim. Acta, Part B, 1978,33, 325. Yudelevich, I. G., Beisel, N. F., Papina, T. S., and Dittrich, K., Spectrochim. Acta, Part B , 1984, 39, 467. Clegg, J. B., in Makram-Ebeid, S . , and Tuck, E. B., Editors, “Semi-insulating 111-V Materials,” Proceedings of Meeting, Evian, 1982, Shiva, Nantwich, Cheshire, 1982, p. 80. Toshiba Corp., Jpn. Kokai Tokkyo Koho, JP 82 26 734, 1982; Chem. Abstr., 1982, 97, 48953t. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. Hendrikx-Jongerius, C., and De Galan, L.) Anal. Chim. Acta, 1976,87,259. Frech, W., and Cedergren, A., Anal. Chim. Acta, 1980, 113, 227. Sturgeon, R. E., and Berman, S. S., Anal. Chem., 1985, 57, 1268. Sturgeon, R. E., Siu, K. W. M., and Berman, S. S . , Spectrochim. Acta, Part B, 1984, 39, 213. Taddia, M., Anal. Chim. Acta, 1986, 182, 231. Muller-Vogt, G., and Wendl, W., Anal. Chem., 1981, 53, 651. Lythgoe, D. J., Analyst, 1981, 106, 743. Zatka, V. J . , Anal. Chem., 1978, 50, 538. Rosen, B., “Spectroscopic Data Relative to Diatomic Mol- ecules,” Pergamon Press, Oxford, 1970, pp. 150 and 153. Dittrich, K., Talanta, 1977, 24, 725. Dittrich, K., Anal. Chim. Acta, 1978, 97, 59. Bulewicz, E. M., and Sugden, T. M., Trans. Faraday SOC., 1958,54830. Daidoji, H., Bunseki Kagaku, 1980, 29, 389; Chem. Abstr., 1980, 93, 212561d. Thompson, K. C., Godden, R. G., and Thomerson, D. R., Anal. Chim. Acta, 1975, 74, 289. L’vov, B. V., and Ryabchuk, G. N., Spectrochim. Acta, Part B, 1982,37, 673. “HGA-500 Graphite Furnace Operator’s Manual, Standard HGA Conditions,” Perkin-Elmer, Norwalk, CT, 1978. Nowotny, H., Lux, B., and Kudielka, H., Monatsh. Chem., 1956, 87, 447. Fairman, R. D., and Oliver, J. R., in Rees, G. J., Editor, “Semi-insulating 111-V Materials,” Proceedings of Meeting, Nottingham, 1980, Shiva, Nantwich, Cheshire, 1980, p. 83. Slavin, W., Carnrick, G. R., Manning, D. C., and Pruszkowska, E., At. Spectrosc., 1983, 4, 69. Paper J5l60 Received December 13th, 1985 Accepted July 3rd, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100437

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Analysis of Solid Samples by Graphite Furnace Atomic Absorption Spectrometry Using Zeeman Background Correction* 443 Glen R. Carnrick, Barbara K. Lumas and William 6. Barnett Perkin-Elmer Corp., 901 Ethan Allen Highwa y, M/S 906, Ridge field, CT 06877, USA Although graphite furnace atomic absorption spectrometry is one of the most sensitive techniques for trace element determinations, the analysis of solid materials can be challengingly difficult. As no dissolution is used, none of the sample matrix is removed before the introduction of the sample into the furnace, and this can result in severe vapour-phase interferences. Typically, quantification has been achieved by the use of standard additions or by comparison with a known reference material.Using stabilised temperature platform furnace technology and simple aqueous standards, we have determined Cr in plastic film, Pb in flexible PVC and Cu in NBS Standard Reference Material bovine liver. Agreement was obtained with the values provided by the sample suppliers. The precision was found to be about 66% (relative standard deviation). The solid materials analysed contained relatively high levels of analyte, requiring the use of alternative wavelengths and in two instances an increased gas flow during atomisation. The effect of gas flow on sensitivity was examined and 19 wavelengths were characterised for use with Zeeman background correct ion. Keywords: Graphite furnace atomic absorption spectrometry; Zeeman background correction; solid samples Much of the early work in atomic spectroscopy was concerned with solid samples.The use of arc or spark excitation together with a direct reader is inherently a technique for the direct analysis of solid samples. However, this technique can be plagued by poor precision and difficulties in calibration. With the growth of flame emission and flame atomic absorption spectrometry, the aspiration of solutions became the primary means for performing analyses. However, flame techniques do not offer the sensitivity required for the analysis of many solids. The graphite furnace is much more sensitive and is capable of the direct analysis of solids. Some solid sample applications using wall atomisation, peak absorbance measurements and simple aqueous standards for quantification are reported in the literature.These include the determination of Au in polyester fibres,l Cu in fingernails2 and Cd in rocks and sediments.3 However, owing to the different analyte volatility in the standard and the solid material, the use of simple standards and direct calibration was often not possible. Many workers have utilised matrix-matched solid materials for calibration in order to overcome volatility problems. Ebdon and Pearce ,4 using peak absorbance measurements and matrix modification, found better reproducibility using slurries in the determination of As in coal. They obtained generally good results using either standards (containing nickel and magnesium matrix modifier) or a coal sample for calibration, although neither procedure provided satisfactory results for samples with a high ash content.Busheina and Headridges analysed nickel-based alloys for Cd, In and Zn using a standardised alloy for calibration. Baker and Head- ridge6 determined Bi, Pb and Te in Cu, also utilising a standardised material. Some workers have been able to compensate for differences in analyte volatility by using the method of additions for the analysis of solid samples. Lord et al.7 determined Al, Cr, Cu, Pb and Zn in freeze-dried freshwater mussels, Graf-Harsanyi and Langmyhrs determined Cr in freeze-dried serum and Eames and Matousekg determined Ag in silicate rocks. In 1981, advances in furnace technology lead Slavin et al. lo * Part of this material was presented at the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlantic City, NJ, February 25-28th, 1985.to develop a new furnace concept. By combining several of these advances together into what was named the “stabilised temperature platform furnace” (STPF) , atomisation condi- tions were created such that the integrated absorbance was essentially independent of the matrix. The STPF concept utilised a L’vov platform, matrix modification and rapid heating to achieve analyte volatilisation into a steady-state atmosphere, pyrolytically coated graphite tubes to improve atomisation efficiency, fast digital processing to monitor accurately rapidly changing furnace peaks and signal integra- tion to compensate for varying rates of volatilisation. Using the STPF concept, thedirect furnace analysis of many complex samples using simple standards is possible, rather than relying on matched standards or the method of additions.In the analysis of solid samples, some workers have used only portions of the STPF technology and as a result have achieved limited success. Chakrabarti and co-workers11J2 used a platform but peak absorbance measurements in the analysis of NBS Standard Reference Materials bovine liver and oyster tissue, and the method of additions was required. Frech et al. 13 used integrated signals but wall atomisation in the determina- tion of Pb and Bi in stainless steel; their results were 20% lower than the expected values. Atsuya and Itoh14 used Zeeman background correction, peak absorbance measure- ments and a miniature cup in a cup furnace to analyse NBS reference materials; for some determinations simple aqueous standards could be used whereas others required the use of the method of additions or other standardised materials for quantification.Headridge and Riddington15 had similar prob- lems in the analysis of glass. Voellkopf et aZ.16 recently used STPF technology to determine successfully As, Cd, Cr, Co, Pb and Mn in soil or hay using aqueous standards. Solid samples were atomised from a pyrocoated graphite cup that was inserted into the graphite tube. In this paper, the direct determination of Cu, Cr and Pb in some solid samples with the graphite furnace using STPF conditions and Zeeman background correction is considered. Some of the problem variables associated with the analysis of solid samples were investigated and some possible solutions are offered.In this work we chose to study samples that had relatively high levels of analyte, in contrast to the usual emphasis on ultimate detection limits when using the graphite furnace.444 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Experimental The solid sampling cup, graphite furnace tube and cup- handling tool used for these experiments, shown in Fig. 1, are identical with those used by Voellkopf et al. 16 The tube is the same as the normal pyrolytically coated tubes except that there is a hole in the centre to accommodate the sampling cup, which can be removed to add and weigh the sample. In addition, special Zeeman contact cylinders with a hole to accommodate the sampling cup are required.The sampling cup is inserted into and removed from the furnace tube using a special plastic tool that fits into a hole in the top of the cup. The cup can be weighed on a conventional microbalance both before and after the sample is added. A matrix modifier can be added either outside the furnace or after the cup is placed in the tube. In the latter instance an autosampler can be used to add the matrix modifier. All experimental work was performed using a Zeeman/5000 instrument equipped with an HGA-500 graphite furnace. The Zeeman/5000 has the fast digital processing required to monitor furnace peaks accurately. A background-corrected absorbance measurement is made every 16 ms. A Series 7000 computer running the HGA Graphics I1 software package17 was used to record the data.A pyrometer was used to verify the pre-treatment and atomisation temperature settings. Problems Associated with Solid Sampling When dealing with solid samples, several problems may be encountered. As the sample is not diluted, there is the possibility of high matrix concentration, which can lead to chemical interferences and high background absorbances. Although Zeeman background correction offers accurate correction for high background absorbances, the loss of light due to background absorption will result in a deterioration of the signal to noise ratio and hence the detection limit. The magnitude of these problems will depend, or course, on the analyte being determined and on the matrix volatility. When a graphite furnace solid sample procedure is de- veloped, the mass of sample that is used is determined by the sensitivity of the available analytical lines, the concentration of the analyte of interest and the amount of background absorption produced during absorption.If small sample masses are requried , sample inhomogeneity may adversely Fig. 1. Solid sampling cup and cup insertion tool affect the precision. For example, NBS standard reference materials seldom claim good homogeneity below 250-mg sample amounts. For some applications homogeneity at submilligram mass levels is required or precise analysis will not be possible. High analyte concentrations present a unique problem for the graphite furnace. The usual emphasis when performing analyses with the graphite furnace is on better sensitivity and better detection limits.However, with no sample dilution possible and with a practical lower limit on sample size, it may be necessary to find a means of reducing sensitivity for some solid sample types. Adding matrix modifiers to solids can be more difficult than adding them to samples in solution because it may be difficult to achieve good contact between the modifier and the sample. For solid samples, the absolute calibration of the auto- sampler or pipette used to dispense the standard solutions is important. Usually, absolute pipette accuracy is not very important as the samples and standards are all pipetted with the same device. However, when analysing solid samples, the accuracy of the determinations is directly dependent on pipette accuracy.For these studies we determined the autosampler accuracy by pipetting water and weighing it on a microbalance. Physically placing solid samples , particularly powdered samples, into the furnace has in the past been a tedious task. The solid sampling cup that will be described solves this problem. As has been evident in the literature, standardisa- tion is also a serious problem. With the use of STPF conditions, it is possible in many instances to use aqueous solutions to standardise the instrument. This ability to use aqueous standards for calibration when analysing solid samples is a significant advance over the use of matched solid standards. Sampling Cup Characteristics The sampling cup is similar to the L'vov platform in that it delays the heating of the sample until the tube has reached a 0.5 0.4 Q, 2 0.3 -9 % 3 0.2 0.1 0 1 2 Time/s 3 Fig.2. Comparison of peak shape and sensitivity for 0.4 ng of Pb atomised from a L'vov platform or from a solid sampling cup. A matrix modifier of 200 pg of (NH4)*HP04 and 10 pg of Mg(N03)2 was used Table 1. Standard conditions for sampling cup characteristic mass study Temperature/"C Element WaveIength/nm SWnm Pre-treatment Atomisation Modifier Pb . . . . 283.3 0.7 900 1800 200 pg of (NH4)2HP04 + Cu . . . . 324.8 0.7 1000 2300 - Cr . . . . 357.9 0.7 1600 2500 50 pg of Mg(N03)2 50 P8 of Mg(N03)2JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 445 more steady-state temperature. Its mass is approximately twice that of the conventional platform, so it would be expected to delay the appearance of the peak slightly longer than the platform.The conditions used are listed in Table 1. Fig. 2 shows that with Pb, a volatile element, there is about a 1-s delay in the appearance of the peak, but the profile shapes are essentially identical. When more refractory elements are determined, it was expected that broadening and increased tailing would occur owing to the slower heating of the more massive sampling cup. This was the case when a Cu standard was atomised from the sampling cup, as shown in Fig. 3. The comparison of the two atomisation sites for Cr produced some surprising results. Fig. 4 shows that while the appearance of the Cr peak is delayed using the sampling cup relative to the platform, the peak is narrower and actually tails less than 0.2 I ?$ ,’ - Platform mo = 11 pg Time/s Figure 3.atomised from a L’vov platform or from a solid sampling cup Comparison of peak shape and sensitivity for 0.5 ng of Cu Table 2. Characteristic mass comparison of L’vov platform with sampling cup molpg per 0.0044A.s Element Platform Sampling cup Pb . . . . . . 9.5 10.3 c u . . . . . . 11.0 14.5 Cr . . . . . . 3.3 3.6 when atomisation is from a platform. The reason for this is not fully understood. It is possible that when analyte volatilisation occurs at high temperatures, the ends of the graphite tubes are hotter with the sampling cup than with a platform. This would result in a more rapid rate of dissociation and less analyte condensation and re-volatilisation. A higher temperature at the end of the tube may be partially the result of the maximum power heating cycle that occurs when the sampling cup is used.Normally, the photodiode, used to set the cutoff of maximum power heating, views the interior of the graphite tube. With the sampling cup, the photodiode may also view the wall of the cup, which is cooler than the graphite tube. This probably causes the tube wall, and particularly the ends, to be heated to a higher temperature than when a platform is used. Table 2 compares the characteristic mass values obtained with the platform and the sampling cup. In all instances the characteristic mass measured with the sampling cup is slightly poorer, i.e. , slightly higher than that measured with the L’vov platform. We believe that this is because of slightly different spatial atmosphere temperatures between the two techniques.Reducing Graphite Furnace Sensitivity When solid samples are analysed in the graphite furnace, primary resofiance lines are often too sensitive. Gong and Suhr3 used both the primary 228.8-nm line and the 326.1-nm 0.4 0.3 W C m n L 0.2 n $ a 0.1 0 Fig. 4. 1 2 3 4 5 6 7 8 9 Time/s 0 Comparison of peak shape and sensitivity for 0.25 ng of Cr atomised from a L’vov platform or from a solid sampling cup. A matrix modifier of 50 pg of Mg(NO& was used ~ ~~~ ~ ~~~ Table 3. Relative wavelength sensitivity Cr . Lowest Wavelength/ energy Maximum mo/Pg Element nm level/K Slithm Energy absorbance per 0.0044 A-s Pb . . . . 283.3 0 0.7 58 1.6 10 202.2 0 0.7 14 1 .o 230 261.4 7819 0.7 56 0.9 1000 368.3 7819 0.7 64 2.6 2400 364.0 7819 0.7 54 1.6 6500 357.9 0 0.7 64 1.4 3.5 360.5 0 0.7 64 1.9 6.0 427.5 0 0.7 65 2.6 10.5 429.0 0 0.7 65 2.6 14.5 301.7 8095 0.7 59 1.6 220 520.8 7593 0.4 56 2.5 400 298.6 8095 0.7 56 1.5 450 Cu .. . . 324.7 0 0.7 62 0.66 11 327.4 0 0.7 61 0.74 15 216.5 0 0.2 30 0.45 60 222.6 0 0.2 36 0.65 120 249.2 0 0.7 54 1.2 420 244.2 0 0.7 49 1.9 2200 * 34 0.3 2250 224.4 - 0.2 * Probably a ground-state line.446 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 line in the determination of Cd in solid rock and sediment samples. Grobenski et al. 18 used the alternative 794.7-nm line for the determination of Rb in NBS Standard Reference Material spinach. The use of alternative lines has been common practice in flame atomic absorption spectrometry for many years, but it has not been common in graphite furnace analyses where the ultimate detection limit is usually the goal.The relative sensitivities of alternative wavelengths listed in flame method manuals may not be directly transferable to Zeeman back- ground correction, depending on the type of Zeeman splitting that occurs. Although not usually a problem, each line should also be checked to determine the maximum absorbance. It may be necessary to consult wavelength reference tables to find additional lines that are suitable. For many elements, there are a number of acceptable alternative wavelengths that provide a wide range of sensitivi- ties. Using the conditions listed in Table 1, characteristic mass values and maximum absorbances were determined for a number of Cu, Cr and Pb lines.These are shown in Table 3. Also shown are the energy values displayed on the spectro- photometer when the monochromator was peaked on the various wavelengths. Energy is calculated as E = (1000 - PMT voltage)/lO. As instrumental noise is inversely propor- tional to PMT voltage, the energy value may be used as an indication of relative noise. A decrease of 6 units of energy corresponds to an approximate 3-fold decrease in light intensity and a 1.7-fold increase in noise. For Cu, the characteristic mass varies from 11 pg using the 324.7-nm resonance line to a characteristic mass of 2250 pg using the 224.4-nm line. This provides a range of 200-fold in sensitivity and illustrates the opportunity offered for analysing samples with a wide range of Cu concentration.Similar studies of lead and chromium provided lines with sensitivity ranges of 650 and 125, respectively. Unfortunately, not all elements have such a variety of suitable lines. Cadmium has its principle resonance line at 228.8 nm, and only the 326.1-nm line, which is about 330 times less sensitive, has a suitable sensitivity for graphite furnace use. 100 R I 1 '\\\- Cr 520.8 nrn Q : I ' \ \ > .- c \ -i) -- - - - _ _ \ Pb - 364.0 nrn b----- 2ot Grobenski et a1.18 and Voellkopf et a1.16 have also reduced the furnace sensitivity by using an internal gas flow through the furnace during atomisation. This reduces the residence time of the atoms and therefore the sensitivity. The use of an internal gas flow is contrary to usually recommended STPF conditions, as the introduction of a gas flow may have a cooling effect, thus upsetting the steady-state temperature conditions in the tube.Therefore, the effects of an internal gas flow should be checked carefully during the method develop- ment phase. The effect of gas flow on the measured peak area for chromium and lead is shown in Fig. 5. A reduction in signal of about 5- to 10-fold is possible by increasing the internal gas flow from 0 to 300 ml min-1. Analysis of Samples The analytical conditions used for the determination of three elements in three different matrices are summarised in Table 4. Matrix modification with solid samples may cause problems 0.1 5 n 0.10 0 m f $ n a 0.05 0 10 yl of 250 mg I-' Pb 1 2 3 4 Tim e/s Fig.6. Peak profiles from a Pb standard [containing 200 pg of (NH&HP04 and 10 Fg of Mg(NO,),] and from a solid PVC sample using the solid sampling cup 0.6 , 0 100 200 Internal gas flow/ml min-' 300 0 1 2 Time/s 3 Fig. 5. Effect of internal gas flow during atomisation on Pb and Cr sensitivity Fig. 7. Peak profiles from a Cr standard and from a solid plastic film sample Table 4. Summary of analytical conditions Temperature/"C Gas flow/ Element Sample Wavelengthhm ml min- Pre-treatment Atornisation Cr . . . . . . . . Plastic film 520.8 300 1200 2500 Cu . . . . . . . . Bovine liver 244.2 0 950 2100 Pb* . . . . . . . . Flexible PVC 364.0 100 750 1800 * A matrix modifier of 200 pg of (NH4)2HP04 and 10 pg of Mg(N03), was used in conjunction with the aqueous standard.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL.1 447 Table 5. Summary of analytical results Resultdpg g-l This Comparison Element SampIe type study values* Cr . . . . Plasticfilm 1 468 524 Plastic film 2 550 575 Plastic film 3 728 727 (NBS 1677a) Cu . . . . Bovineliver 155 158 k 7 Pb . . . . FlexiblePVC 8000 8400 * Provided by the sample suppliers. due to poor or irreproducible contact with the solid material. Even with adequate surface contact the reaction of the modifier may be different or ineffective. Therefore, for most of this work matrix modifiers were avoided and pre-treatment temperatures were determined by careful examination of the thermal stability of the solid materials. For example, pre- treatment studies indicated that Pb was stable up to temperat- ures above 750 "C in PVC.In order to ensure the stability of the standard at this temperature, a modifier consisting of 200 yg of ammonium phosphate and 10 yg of magnesium nitrate was used with the aqueous standards. All of these samples contain matrices that are relatively easy to destroy in the graphite furnace, reducing some of the problems of high matrix concentration. However, the concen- tration of analyte in all three instances was high, requiring the use of insensitive, alternative lines and, in two instances, a higher than usual internal gas flow. Results The results obtained are shown in Table 5. In all instances, good agreement was obtained with the values provided by the sample suppliers and obtained using a variety of other techniques.For the chromium values, the reference value provided by the supplier was in terms of milligrams per square metre. We took the original samples, which were round, measured their diameter using a Scherr Tumico Comparator, calculated their area, weighed them and thus converted the values into micrograms per gram as shown here. The bovine liver results were corrected for a 2.5% residual moisture content. The precision for four replicates was determined to be about 6 6 % (relative standard deviation). Fig. 6 shows typical signals obtained when determining Pb in a flexible PVC sample. The Pb signal from the PVC sample contains several peaks that may be due to the manner in which the plastic decomposed as the temperature of the sampling cup increased.A signal containing multiple peaks does not necessarily cause any difficulty when STPF conditions are used, because the signal is integrated under conditions of temperature stability in the tube. With such a sample, it is unlikely that accurate determinations of this material could be made using the method of additions and peak absorbance measurements. The peak profile from Cr in the film exhibited a slower rate of volatilisation, resulting in a broader peak than the standard but, unlike the Pb in the PVC, did not contain any multiple peaks. This is evident in Fig. 7. assumed that all solid materials will in the future be easily analysed in a furnace using STPF technology. The chief advantage of solid sampling is lower absolute detection limits and the elimination of the time-consuming sample decomposi- tion step.There are, however, limitations with solid sample procedures. Some samples that contain high concentrations of a refractory matrix may not be easily determined. High matrix concentrations may cause interferences that may not be overcome or may require extensive method development. Although we found the sampling cup to be convenient to use, solid sampling is still labour intensive as no automatic weighing and/or sample insertion devices are available. During our study we have found that alternative wavelengths and an internal gas flow allow the determination of higher concentrations than would otherwise be possible. In all instances we used aqueous standards for calibration, in contrast to much of the previous work where matched solid standards or the method of additions was used for calibration.The samples we used were relatively homogeneous. For some applications sample homogeneity may affect the precision when small sample sizes are used. Using Zeeman background correction, we have determined the maximum absorbance and characteristic mass value for a number of Pb, Cr and Cu lines. Some of the Pb and Cr lines examined are non-resonance lines and the sensitivity of such lines is highly dependent on the vapour-phase temperature. No attempt was made in this study to establish the optimum atomisation temperature for each of these non-resonance lines. We thank our colleagues F. J. Fernandez, J. D. Kerber and W. Slavin for their constructive discussions and suggestions. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. References Kerber, J. D., At. Absorpt. Newsl., 1971, 10, 104. Barnett, W. B., and Kahn, H. L., Clin. Chem., 1972, 18,923. Gong, H., and Suhr, N. H., Anal. Chim. Acta, 1976,81, 297. Ebdon, L., and Pearce, W. C., Analyst, 1982, 107, 942. Busheina, L. S., and Headridge, J . B., Anal. Chim. Acta, 1982, 142, 197. Baker, A. A., and Headridge, J. B., Anal. Chim. Acta, 1981, 125, 92. Lord, D. A., McLaren, J. W., and Wheeler, R. C., Anal. Chem., 1977,49, 257. Graf-Harsanyi, E., and Langmyhr, F. J., Anal. Chim. Acta, 1980, 116, 105. Eames, J. C., and Matousek, J. P., Anal. Chem., 1980, 52, 1248. Slavin, W., Carnrick, G. R., and Manning, D. C., At. Spectrosc., 1981, 2, 137. Chakrabarti, C. L., and Li, W. C., Spectrochim. Acta, Part B, 1980, 35, 93. Chakrabarti, C. L., Wan, C. C., and Li, W. C., Spectrochim. Acta, Part B, 1980, 35, 547. Frech, W., Lundberg, E., and Barbooti, M. M., Anal. Chim. Acta, 1981, 131, 45. Atsuya, I . , and Itoh, K., Spectrochim. Acta, Part B, 1983, 38, 1259. Headridge, J. B., and Riddington, I. M., Analyst, 1984, 109, 113. Voellkopf, U., Grobenski, Z . , Tamm, R., and Welz, B., Analyst, 1985, 110, 573. Barnett, W. B., and Carnrick, G. R., At. Spectrosc., 1984,5, 210. Grobenski, Z., Weber, D., Welz, B., and Wolff, J., Analyst, 1983, 108, 925. Discussion Although we have obtained accurate determinations for a few elements in three different sample types, it should not be Paper JA6l7 Received April 15th, 1986 Accepted July 8th) 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100443

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Effects of Ageing of Pyrolytically Coated Tubes on the of Refractory Elements by Electrothermal Atomisation Absorption Spectrometry Michel Hoenig" 449 Determination Atomic Institute for Chemical Research, Ministry of Agriculture, Museumlaan 5, 1980 Tervuren, Belgium Frank Dehairst ANCH, Free University of Brussels, Pleinlaan 2, 1050 Brussels, Belgium Anne-Marie de Kersabiec Laboratoire de Geochimie et Metallogenie (U.A. CNRS 796), University of Paris Vl, Place Jussieu 4, 75252 Paris Cedex 05, France The quality of the pyrolytic coating of graphite tubes is shown to be of prime importance in the electrothermal atomisation atomic absorption spectrometric determination of refractory elements in complex matrices. For molybdenum in a plant matrix the shortcomings due to progressive degradation of the coating can be circumvented by signal integration.The problems encountered with chromium in several matrices are related to the irreproducible quality of the coating between different sets of tubes supplied by the same manufacturer. For the determination of barium in sea water, pyrolysis at 1800 "C efficiently eliminates background absorption. With an increase in tube porosity, due to degradation of the coating, there is a drastic increase in the background signal that can be corrected only with a Zeeman device. Keywords: Electrothermal atomisation atomic absorption spectrometry; graphite tube coating; molybdenum; chromium; barium Since the commercialisation, in the early 1970s, of graphite furnace atomisers, concern has arisen about the variability of the surface properties of graphite tubes and their progressive degradation with increasing numbers of firings.Although the use of spectrographic graphite has specific advantages, such as increasing mechanical resistance with temperatures up to 2500 "C and good reduction properties, several drawbacks exist. Firstly, graphite can interact with some elements to produce stable carbides and/or interlamellar compounds. Both of these phenomena may impede the volatilisation of the analyte and change its original atomisation process. Secondly, graphite is porous and the atomiser wall is therefore subject to infiltration by the solution. Also, at high temperatures vapours can diffuse through the atomiser wall.Coating of the tube wall with pyrolytic graphite, which is non-porous and highly impermeable to gases at the tempera- tures generally applied, has greatly reduced these problems However, the thin pyrolytic graphite layer itself becomes increasingly porous with the tube lifetime as a result of the influence of strongly acidic solutions and high atomisation temperatures. Other coatings and impregnation materials have been studied in order to improve the atomiser surface properties. Metals that have an elevated melting-point and form stable interstitial carbides with graphite (Ta, Mo, Nb and W) have been used for this purpose.3-9 However, improvements obtained by these treatments are too case-specific to be universally applicable. Thirdly, the quality of the pyrolytic coating may be highly variable and is a function of the number of active sites per unit area.This number depends on the initial number of nucleation centres available on the graphite surface during the coating process.10 Oxygen provided by the injected sample solution will bind to carbon preferentially on these active sites. During atomisation this carbon is released in the vapour phase as CO, increasing the active site oxygen-binding capacity. Also, analyte and matrix elements will compete for the active sites. * To whom correspondence should be addressed. t Research Associate at the Belgian National Fund for Scientific Research. Such processes are variable during the tube lifetime and can favour or inhibit certain reactions. They can also modify the initial process of analyte atomisation.In practice, it is extremely difficult to distinguish between the different processes involved and to identify which are controlled mainly by active site availability.11 The determination of more volatile elements is less affected by these problems when using platforms in solid pyrolytic graphite. The surface properties of these platforms are much less variable with time. However, with refractory elements a high heating rate is required for the atomisation step. At present this can be achieved only in the case of analyte volatisation from the tube wall. Thus, for refractory elements the analysis results will depend essentially on the quality of the pyrolytic coatings. Recently, totally pyrolytic graphite tubes have been tested,l2J3 showing for the refractory element vanadium an improved lifetime , sensitivity and signal consistency.l4 However, for Varian atomisers such tubes are not commer- cially available. In this paper we discuss three examples where progressive degradation of the pyrolytic coating introduces difficulties in the determination of refractory elements. Experimental A Varian AA-1275 BD spectrometer, equipped with a GTA-95 atomiser and an automatic sample dispenser, was used. Pyrolytically coated graphite tubes (Varian 63-100002- 00 and Le Carbone-Lorraine 453-30-258) were used. Some tests were performed with tantalum-impregnated tubes.* The elements studied were determined at the following wavelengths: Mo, 313.0 nm (band pass 0.5 nm); Cr, 357.9 nm (band pass 1 nm); and Ba, 553.6 nm (reduced height slit, band pass 0.5 nm).Details on the electrothermal programmes are given in Table 1. For barium, additional measurements were per- formed with a Hitachi Zeeman 2-7000 system, using a similar electrothemal programme to that given in Table 1. The optical pyrometer used was a Leeds and Northrup Model 86-27.450 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Table 1. Furnace parameters TemperaturePC (timels) Element Stage Drying* Pyrolysis Cooling Atomisation t Mo . . . . . . . . Ramp 115 (15) 1500 (20) 100 (6.7) 2700 (1.3)$ Cr . . . . . . . . Ramp 115 (15) 1200 (10) - 2600 (0.7)$ Ba . . . . . . . . Ramp 115 (15) 1800 (20) 100 (8.1) 2900 (1.4)$,§ 2800(1.4)$,1 Hold 115 (10) 1500 (5) - 2700 (4) Hold 115 (10) 1200 (5) - 2600 (3) 2900 (3) 0 2800 (3)l Hold 115 (10) 1800 (7) - * For a 2 0 4 sample.t Gas stop in all instances. $ Maximum heating rate. 0 Varian tubes. Le Carbone-Lorraine tubes. 3000 1 0 m J'i 0 1000 2000 3000 Tern peratu re settingPC Fig. 1. Temperature evolution during tube lifetime. Line, theoret- ical curve. Symbols: B, new tube; A, after 65 firings; 7 , after 100 firings; and 0, after 500 firings (with evident macroscopic degrada- tion) Results and Discussion Effect of Tube Ageing on the Temperature Achieved We investigated whether tube ageing results in a reduction in temperature with a given power setting, as observed by other workers. 12-14 A new tube was weighed and the temperature achieved measured for power settings in the range 800-2800 "C.Temperature was measured with an optical pyrometer focused on the inner tube wall, through the injection port. Measured temperatures did not differ by more than 5% from those selected (Fig. 1). These temperature measurements were repeated on the same tube after about 100 injections of strongly acidified sea water. It is shown below (see Barium in Sea Water) that such a treatment rapidly increases the background signal as a result of inefficient elimination of matrix salts at the initially selected pyrolysis temperature. After this treatment the mass of the tube decreased by only 1.6% (from 0.917 to 0.902 mg) and inspection of inner tube wall revealed hardly any macroscopic degradation of the pyro-coating. The temperature measure- ments did not change significantly compared with the initial measurements.Further, a tube showing evident macroscopic degradation, principally around the injection port, also attained temperatures that did not differ significantly from the selected temperatures (about 500 firings). In these instances, however, uncertainty exists as to changes in the heating rate. Thus, evidence is given here that tube-age effects are not due to temperature discrepancies between the achieved and selected temperatures, but rather to alterations to the tube surface. 0.2 a, C m -? a 2 0.1 0 100 200 Firings Fig. 2. Peak-height absorbance for molybdenum measurements during the atomiser lifetime. A, 200 pg of Mo, 6% HNO,; B, plant matrix solution (grass); and C, plant matrix solution + 200 pg of Mo Molybdenum in a Plant Matrix The changing state of the pyrolytic coating with increasing number of firings affects the molybdenum absorbance signal, as shown in Fig.2. In simple nitric acid solution the peak absorbance decreases strongly after the first 100 firings (Fig. 2, A). This fact can be ascribed to the progressive degradation of the pyrolytic graphite layer and has been observed by others.15J6 However, in the presence of plant matrix constitu- ents (K, Ca, Mg, Na and P) the molybdenum absorption signal increases progressively with increasing number of firings (Fig. 2, B). Spiking of the plant matrix with molybdenum results in a similar evolution of absorbance with tube lifetime, but with the absorbance shifted towards higher values (Fig. 2, C ) . Thus, enhancement of the signal with tube lifetime concerns only the molybdenum initially present in the plant matrix.Such anomalous results are not due to insufficient correction of background absorption. Indeed, the background produced by a 20-p.1 injection of plant matrix solution never exceeded an absorbance of 0.2. At the molybdenum wavelength used, such a value is easily corrected by the deuterium arc device. The processes occurring here prevent the correct determination of molybdenum in the plant matrix, be it by direct calibration or standard additions method. For the plant matrix we inspected the molybdenum absorb- ance - time profile during atomisation, with an increasing number of firings. With continuing degradation of the pyrolytic coating, the increase in peak absorbance is accom- panied by peak narrowing.No significant change in peak area is observed. In this instance integration of the molybdenum signal is required. Indeed, deterioration of the pyrolytic graphite layer affects the absorbance signal of molybdenum in simple nitric acid solution, the plant matrix and the molyb- denum-spiked plant matrix in a similar way (Fig. 3, A, B andJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY. DECEMBER 1986, VOL. 1 45 1 B D - - - -- - - -- - - A C .- - - - - -- - - -- 0.2 0 I I I 100 200 Firings Fig. 3. Peak-area absorbance for molybdenum measurements during the atomiser lifetime. Lines as in Fig. 2 C). The small decrease in the analyte signal can now easily be controlled by periodic re-calibration. 17 Chromium in Simple Acid Solution and an Animal Matrix Although chromium is a more volatile element than molyb- denum, the problems observed during its determination in animal matrices can in some instances be overcome only with prea t diffi ciil tv for two different sets of Varian graphite tubes.A similar evolution was observed in the peak-area absorbance measure- ment mode. For one set of graphite tubes (Fig. 4) the determination of chromium is possible with periodic re- calibration in order to cope with small variations in surface properties during the tube lifetime. For the other set the situation is much more complicated. It is observed in Fig. 5 that signals having different amplitudes at the start tend towards the same level with increasing number of firings. The same evolution is observed for graphite tubes from the same set, impregnated with tantalum according to the procedure described by Zhtka.8 In such a situation a correct quantitation of chromium is not possible.This stresses the paramount importance of the quality of the pyrolytic coating. This quality depends essentially on number of active sites, which may vary from one set of tubes to another, as a result of an inhomogeneous or irreproducible coating with pyrolytic graphite. In a simple nitric acid medium chromium probably forms a Cr-0-C bond, and is released in the vapour phase as CrO. In some instances this oxide may not dissociate completely as a result of the large amount of CO simul- taneously released from the active sites.10 Also, using 51Cr as a radiotracer, Veillon et al.18 have shown that a significant fraction of the chromium can be irreversibly retained by the graphite even at very high atomisation temperatures.This was observed both for uncoated and pyrolytically coated graphite tubes. Further, the importance of retention seemed to depend on the matrix type. Barium in Sea Water There is interest in the development of procedures for the direct determination of barium in sea water.19720 Reliable barium data obtained in the past were all obtained by mass spectrometry and show the concentrations to range between 5 and 25 pg 1-1.21 Such concentrations are, in principle, directly measurable with modern electrothermal devices. Firstly, using a simple nitric acid solution, we established optimum conditions for barium determination (Table 1).As for molybdenum, we included a cooling step in the tempera- ture programme. After the cooling stage, the longer heating- ramp to the chosen atomisation temperature permits better isothermal conditions during the atomisation step and thus enhances the analyte peak-height signal (by ca. 30% for Ba). 0.2 8 + 0.1 5: a a a 0 50 100 150 200 Firings Fig. 4. Peak-height absorbance for chromium measurements during the atomiser lifetime using one set of graphite tubes. A, 50 pg of Cr, 2% HN03; B , 100 pg of Cr, 2% HN03; C, animal matrix solution (fish tissue); and D, animal matrix solution + 50 pg of Cr I a, C m e g 0.1 a 0 50 100 Firings Fig. 5. Peak-height absorbance in chromium measurements during the atomiser lifetime using a different set of graphite tubes to that in Fig.4. Lines as in Fig. 4 As signal enhancement is not observed in the peak-area mode, peak narrowing occurs owing to an increased rate of atomisa- tion of the analyte. Using these conditions the characteristic mass of barium obtained is 6.5 pg for an absorbance of 0.0044. For a 2O-pl sample, we therefore consider that barium concentrations of 2 2 pg 1-1 can be measured with acceptable accuracy. This is reasonable, considering the range of barium concentrations in the oceans. Secondly, we studied the background absorption generated by the sea-water matrix. Although the background level is weak at the wavelengths used,22 light scattering and/or molecular absorption can still induce erratic results with the Varian system because background correction is impossible to perform at the barium analytical line.We followed the evolution of the background signal as a function of pyrolysis temperature (Fig. 6). For a new pyrolytically coated tube it appears that the largest fraction of the background signal is already eliminated at 1000 "C, but a weak non-specific absorption persists up to 1650 "C. The atomic signal for pyrolysis between 1000 and 1650 "C shows enhancement exceeding the contribution of the background signal. This enhancement must be due to non-spectral interferences. At higher pyrolysis temperatures the background signal disap- pears entirely. These observations are similar to those of Conley et al. 19 No loss of barium is observed up to a pyrolysis temperature of 1800 "C. Therefore, the use of pyrolysis temperatures between 1700 and 1800 "C should allow the determination of barium in sea water without correction for spectral interferences. Thirdly, barium calibration graphs were established for both simple nitric acid medium and acidified sea water.The452 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 Q) C m 0.25 s a n 0 I I I 1000 1500 2000 500 Pyrolysis tern peraturePC Fig. 6. only; and AA, barium atomic signal Char study for barium (125 pg) in sea water. BG, background 0.3 0 20 40 60 80 Firings Fig. 7. Peak-height absorbance for barium measurements during the atomiser lifetime. A, 100 pg of Ba, 2% HN03, Varian ap aratus; B, Ca. 60 Pg of Ba, acidified sea water, Varian apparatus; C, 5& pg of Ba, 2% HNq3, Hitachi apparatus; and D, ca.620 pg or' Ba, acidified sea water, Hitachi apparatus slopes in both media are very close, indicating the absence of non-spectral interferences. However, these findings apply only for a very limited lifetime of the tube. As opposed to simple nitric acid solution, which shows a slow and progressive decrease in barium absorbance signal (Fig. 7, A), sea water shows a drastic increase in absorbance (Fig. 7, B) (both Le Carbone-Lorraine and Varian graphite tubes). This enhance- ment must be due to the changing properties of the pyrolytic graphite layer. With usage of the tube and hence an increase in porosity, matrix elements may be retained and they interact with either the analyte or the graphite. With interaction with the analyte, suppression of barium ionisation could occur, resulting in an increase in the analyte signal.Ionisation of barium at temperatures higher than 2300 "C can be significant, as suggested by Ottaway and Shawl3 and Sturgeon and Berman.24 Therefore , we added an easily ionisable element (potassium) after pyrolysis and a cooling stage for barium determinations in simple nitric acid. This increased the barium atomic signal by 15% at most, indicating that ionisation is not the major contribution to this observed enhancement. Thus, it appears that as a result of interactions of the constituents of sea water with the graphite, the efficiency of elimination of these matrix constituents during pyrolysis decreases with tube lifetime. This was confirmed by measuring potassium, a major constituent of sea water, using the barium electrothermal programme.The determination of potassium in a sea-water sample, using a new tube, showed no significant signal. With an aged tube (80 firings) a very high potassium signal was produced, confirming the decrease in matrix evacuation efficiency during pyrolysis. Blank firing of the aged tube produced no signal, indicating no memory effects. Finally, the use of a Hitachi 2-7000 system with Zeeman background correction confirmed these results. As for the Varian system without background correction, simple nitric acid solution showed a very slow decrease in the barium absorbance signal with tube lifetime (Fig. 7, C). For a sea-water matrix this signal remains nearly constant during the entire tube lifetime, indicating efficient background correc- tion (Fig.7, D). Conclusions Progressive degradation of the graphite tube pyro-coating appears to be the major factor affecting the determination of the refractory elements Mo, Cr and Ba in complex matrices. For molybdenum in a plant matrix, use of the peak-area measurement mode permits an accurate determination. For chromium, erratic inhomogeneity of the pyro-coating quality can dramatically affect its determination. For barium in sea water, direct determination seems impossible, owing to the increasing inefficiency of matrix elimination during tube lifetime. As no significant change in the temperatures achieved over the lifetimes of the pyro-coated tubes occurs, we conclude that the phenomena must be due to pyro-coating degradation inducing variability in the interactions of the analyte and/or matrix components with the graphite. 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 Aspila, K. J., Chakrabarti, C. L., and Bratzel, A. P., Anal. Chem., 1972,44, 1718. Kantor, T., Clyburn, S. A., and Veillon, C . , Anal. Chem., 1974,46,2205. Cioni, R., Mazzucotelli, A., and Ottonello, G., Anal. Chim. Acta, 1976,82,415. Kuzovlev, I. A., Kuznetsov, Y. N., and Sverdlina, 0. A., Zavod. Lab., 1973,39,428. Ortner, H. M., and Kantuscher, E., Talanta, 1975,22,581. Runnels, J. H., Merryfield, R., and Fisher, H. B., Anal. Chem., 1975,47, 1258. Stiefel, T., Schulze, K., Tolg, G., and Zorn, H., Anal. Chim. Acta, 1976, 87, 67. Zatka, V. J., Anal. Chem., 1978, 50, 538. Norval, E., Human, H. G. C., and Butler, L. R. P., Anal. Chem., 1979. 51,2045. RubeSka, I., personal COmmUnication. Koretkovii, J., Frech, W., Lundberg, E., Person, J. A., and Cedergren, A., Anal. Chim. Acta, 1981, 130, 267. Dymott, T. C., Wassall, M. P., and Whiteside, P. J., Analyst, 1985, 110, 467. Littlejohn, D., Duncan, I., Marshall, J., and Ottaway, J. M., Anal. Chim. Acta, 1984, 157, 291. Littlejohn, D., Duncan, I., Hendry, J . B. M., Marshall, J . , and Ottaway, J . M., Spectrochim. Acta, Part B , 1985,40, 1687. Sturgeon, R. E., and Chakrabarti, C. L., Anal. Chem., 1977, 49, 90. Sneddon, J., and Fuavao, V. A., Anal. Chim. Acta, 1985,167, 317. Hoenig, M., Van Elsen, Y . , and Van Cauter, R., Anal. Chem., 1986, 58, 777. Veillon, C., Guthrie, B. E., and Wolf, W . R., Anal. Chem., 1980, 52, 457. Conley, M. K., Sotera, J. J., and Kahn, H. L., Report No. 11, Instrumentation Laboratory, Wilmington, DE, 1979. Epstein, M. S., and Zander, A. T., Anal. Chem., 1979,51,915. Chan, L. H., Drummond, D., Edmond, J. M., and Grant, B., Deep-sea Res., 1977, 24, 1613. Hoenig, M., and Wollast, R., Spectrochim. Acta, Part B, 1982, 27, 399. Ottaway, J. M., and Shaw, F., Analyst, 1976, 101, 582. Sturgeon, R. E., and Berman, S . S . , Anal. Chem., 1983, 55, 190. Paper J5l61 Received December 16th, 1985 Accepted July 14th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100449

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 453 Direct Microcomputer Controlled Determination of Zinc in Human Serum by Flow Injection Atomic Absorption Spectrometry Kirsten Wiese Simonsen, Bent Nielsen, Arne Jensen and Jan Rud Andersen" Department of Chemistry AD, Royal Danish School of Pharmacy, 2 Universitetsparken, DK-2 100 Copenhagen, Denmark A procedure is described for the direct determination of zinc in human serum by fully automated, microcomputer controlled flow injection atomic absorption spectrometry (FI-AAS). The FI system is pumpless, using the negative pressure created by the nebuliser. It only consists of a three-way valve, programmable from the microcomputer, t o control the sample volume. No pre-treatment of the samples is necessary. The limit of detection is 0.14 m g 1-1, and only small amounts of serum (<150 pi) are needed forthe analysis.The accuracy, as estimated by analysing sera with known amounts of zinc, is excellent, and the between batch precision is ca. 5% RSD. Peak-area measurements are used in the procedure; as a consequence the quantitation can be carried out by use of an aqueous calibration graph. The sample throughput is 40 h-1, making the procedure well suited for routine clinical analysis. Keywords: Zinc determination; flow injection; flame atomic absorption spectrometry; human serum; microcomputer control Zinc is an essential element that is active in a wide variety of enzymes and hormones. In man, it influences growth rate and bone formation, wound healing, development and function of reproductive organs and the integrity of the skin.In the human body the element is present in mg kg-1 concentrations in tissues and fluids.1 Serious zinc deficiency symptoms may arise even after short periods without adequate supply of the element. Children and infants constitute a particular risk group, as their requirements for zinc are large due to their rapid growth. Numerous examples of human zinc deficiency have been identified worldwide, but it was not recognised until recently that the bioavailability of zinc in the presence of phytate and fibres, and in certain infant formulae, is decreased. 1 Once a zinc deficiency exists, the level of zinc in serum or plasma is often decreased in comparison with the normal level, hence the measurement of serum or plasma zinc is useful in diagnosing the deficiency.The conventional method for quantitating zinc in serum is flame atomic absorption spec- trometry (FAAS).* Sometimes comparatively large volumes are required, 3-5 ml corresponding to ca. 10 ml of whole blood, or alternatively the samples are diluted with subse- quent loss of sensitivity and risk of contamination. To improve this situation, procedures with smaller sample requirements are desired for a number of purposes. In this context, the combination of flow injection3 and atomic absorption spec- trometry (FI-AAS) has many appealing characteristics.4 Together with the limited serum volumes necessary, the high precision and the potential for fast direct analysis with the possibility of complete automation makes it an ideal candidate for clinical purpbses.When the patient is a child, or when frequent blood sampling is needed, the inherent limited sample requirement of FI-AAS becomes a definite and almost mandatory advantage. Flow injection AAS systems have previously been des- cribed for the determination of zinc in serum. Rocks et a1.5 used a manual system comprising a pump and a septum injector, and matrix-matched standards were required for the quantitation. A pumpless system harnessing the negative pressure of the nebuliser for the propagation of the carrier was devised by Attiyat and Christian.6 A sample loop, with the inherent loss of sample in filling the loop, and sample dilution were included in their procedure. Sherwood et aL7 used a controlled-dispersion procedure in their determinations.A * To whom correspondence should be addressed. pump plus two carrier reservoirs were necessary in order to maintain a stable flame. In this paper we describe a completely automated FI-AAS procedure for the determination of zinc in human serum. The instrumentation is a simple construction controlled by a microcomputer. The FI system has no pump or injection port in the usual sense. The propagation of the carrier stream and sample to the detector, the AA spectrometer, is driven by the negative pressure of the nebuliser, and the sampling is performed by means of a single three-way valve operated by the microcomputer. The analytical signals are recorded on a strip-chart recorder and their areas collected by the computer, The quantitation is made by comparing the analytical print- outs with an aqueous calibration graph, and excellent results in terms of accuracy, precision and detection limit are obtained. Experimental Instrumentation A Perkin-Elmer Model 460 atomic absorption spectrometer operated under the conditions given in Table 1 was used as a detector. The atomisation signals were displayed on a Radiometer REC 80 strip-chart recorder, and their areas were printed out on a Commodore 4023 matrix printer.Samples were drawn from a Samplomat autosampler (Struers, Copen- hagen, Denmark) by an N Research Type 161 TO31 three-way valve (Lambda Electronics, Stockholm, Sweden) through a 2.5 cm long polyethylene tube of 0.5 mm inner diameter. The dispersion tube was of the same type.Integration of the analytical signals was carried out by a laboratory-built device. (Peak-height measurements could be obtained simul- taneously, but they were not used in this investigation, see below.) The microcomputer controlling the autosampler, three-way valve, integrations and printouts was a Commodore 64 (Fig. 1). The computer program,8 written in BASIC, was stored on tape and read in from a Commodore Datasette. Reagents Potassium chloride and glycerol used in preparing the artificial serum9 for the detection limit study were of analytical-reagent grade, purchased from E. Merck, Darmstadt, FRG. A certified 1 g 1-1 zinc reference solution (Titrisol, E. Merck) was used, and aliquots of this were appropriately diluted to454 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL.1 Table 1. Instrumental characteristics of the components constituting Table 2. Characteristics for the determinatian of zinc in serum the FI-AAS system Samplerequirement . . . . . . ca. loop1 AAspectrometer . . . . Hollow-cathode lamp current Analyticalline . . . . . . Spectral band pass . . . . Mode . . . . . . . . Flame . . . . . . . . . . Perkin-Elmer 460 . . 15 mA . . 213.4 nm 0.7 nm . . Continuous, absorbance (TC1) . . Air - acetylene . . Air flow-rate . . . . . . . . Acetyleneflow-rate . . . . . . Carrier . . . . . . . . . . Carrierflow-rate . . . . . . Sampletubelength . . . . . . Dispersion tubelength . . . . Tubeinnerdiameter . . . . . . Dispersion . . . . . . . . 5.5 1 min-1 3.2 1 min-1 Milli-Q water ca. 6.5 ml min-l 2.5 cm 32 cm, 0.5 mm 1.9 Peak max.detector -1 I I I Integrator 1 I I I I -J Autosam pler U I I Carrier Fig. 1. controlled by the microcomputer Schematic diagram of the FI-AAS system: broken line yield working standards. Milli-Q water, which is a type I ultrapure water prepared using a Milli-Q de-ionisation unit (Millipore, Bedford, MA, USA), was used throughout. The carrier was also Milli-Q water. Contamination Control All laboratory ware (sample and working standard containers, sample cups, pipette tips etc.) were cleaned before use by soaking for at least one week in 4 M nitric acid, followed by washing with copious amounts of Milli-Q water. When necessary, drying was carried out in a class 100 laminar air flow clean bench (TL 2448, Holten LaminAir, Allerod, Denmark).Samples The human serum samples used in establishing the precision and accuracy of the method were stored at -20 "C prior to analysis. One of these was an aliquot of a sample that had been used for a number of years as a control for serum zinc at the Department of Clinical Chemistry, University of Copen- hagen, Herlev Hospital, Herlev, Denmark, i.e., the zinc concentration of this sample was well established. The NBS Reference Material 8419 bovine serumlo (National Bureau of Standards, Washington, DC, USA) was also distributed and stored in a frozen state. The Seronorm Trace Element, batch number 105, from Nyegaard Diagnostics, Oslo, Norway, was freeze-dried and was reconstituted with Milli-Q water before analysis. Analytical Procedure The analyses were carried out using the conditions given in Tables 1 and 2.The serum samples were measured directly after thawing without dilution or other pre-treatments. Two aliquots of each sample were analysed, and blanks were run between each sample to prevent cross contamination and Sample voiume . . . . . . . . Samplethroughput 40 h-1 Detection limit 7.2 ng Typically 50 pl (may be varied via the computer) . . . . . . . . . . . . . . Quantitation . . . . . . . . Byaqueauscalibrationgraph memory effects. As the first sample aliquot is diluted to a small extent by the carrier left in the sample tube from the preceding blank, only the second sample aliquot was used for quantitation. Experience has shown that this precaution secures better precision, accuracy and detection limits.All three signals, originating from the blank and the two sample aliquots, respectively, are in fact registered by the microcom- puter; if it was desired, a minor change in the software would allow for the discrimination of the two first, unused signals. The sample waste thus corresponds to the volume of the first sample aliquot, typically 50 p1. The quantitation was carried out by comparing the area of the transient sample signal with a calibration graph obtained by running a series of aqueous standards, followed by a linear least-squares treatment of the data. The latter graph must be established daily, as the small variations in gas pressure during the lifetime of the acetylene cylinder influences the negative pressure of the nebuliser and thus the carrier and sample flow-rate.The slope of the calibration graphs obtained with the same acetylene cylinder may gradually alter by as much as 10% during its useful lifetime, whereas only changes of ca. 1% were observed within one day. Such small changes are not important for the accuracy, see below. System Description The FI-AAS system is illustrated in Fig. 1. The autosampler is controlled by the microcomputer via an interface. The crucial part, the three-way valve, is open for the carrier when the voltage is zero, and for the sample when a voltage is applied. As previously mentioned, no pump is necessary, as sample and carrier are sucked to the burner by the negative pressure created by the nebuliser. With our experimental conditions, this corresponds to a flow-rate of ca.6.5 ml min-1. The sample volume is determined by the time that the three-way valve is open for the sample (this time is controlled by the Commo- dore 64), and by the flow-rate. We typically use a sample volume of 50 pl. The three-way valve is so small that it is conveniently placed directly on the sampling arm of the autosampler, thereby reducing sample waste and dispersion. A limited dispersion will reduce peak width and hence carry-over and thus increase the sampling frequency. The dispersion obtained with the described instrumentation is 1.9, allowing 40 samples per hour to be analysed by the procedure given above, i.e., two sample injections plus one blank injection, with only little attention from the operator, see Table 2. The actual Perkin-Elmer 460 AAS was not equipped with a computer interface such as an RS 232, BCD, or similar digital outputs. We therefore used the 1-V analogue recorder output, and transferred the signals to the microcomputer via a laboratory-built analogue and digital interface. A diagram of this interface is shown in Fig.2. The analogue part is an amplifier with zero adjustment. The signal from the amplifier is transported to a precision analogue integrator (HA 2900, chopper stabilised; Harris Semiconductor, Brussels, Belgium) and peak detector (laboratory-built:), each having its own sample and hold circuit serving as memories for the peak area and peak height, respectively. (The integrator and peak detector built into the Perkin-Elmer 460 from the factory are not easily accessible for remote control.) The signals are relayed, one at a time, to an analogue to digital converter withJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL.1 455 Switch a I I’ I ’ - 1 - 2 BCD L 4 output 8 logic 10000 Fig. 2. detector; 4, sample and hold; 5, zero adjustment; and 6, Jgital voltmeter with tri-state control logic Schematic diagram of the interface between the s ectrorneter and the microcomputer: 1, integrator; 2, sample and hold; 3, peak max. tri-state BCD output. A Intersil DM-4100D 4.5 digits volt- meter (Date1 Intersil, Mansfield, MA, USA) is used for this purpose. Finally, the signals are transmitted to the Commo- dore 64 in BCD code, one digit at a time (four bit parallel digit serial). 11 Thus, the microcomputer controls the autosampler, the three-way valve, the data collection, the data print out and also keeps track of the sample history. Results and Discussion Apart from the simplicity and limited sample requirements of the present analytical method, its main advantages are the ’ease of quantification, the precision and the accuracy.The quantitation is performed by comparing the analytical signal to an aqueous calibration graph. This is only possible when peak-area measurements are used. The alternative method of peak-height , or absorbance, measurements calls for a matrix-matched standard graph, i.e., a calibration graph obtained in serum, or for the method of standard additions. The reason for this is that viscosity differences between serum and water will have a profound effect on the peak height, as this detection method measures a concentration rather than an amount.For peak-height measurements, the discrete signals are proportional to the instantaneous concentration of the atomised analyte in the flame, and even small differences in viscosity will manifest themselves as peak-height differences. The peak areas, on the other hand, are unchanged. Further- more, the method of standard additions is time consuming, and sera used for the production of matrix-matched calibra- tion graphs should ideally be low in zinc; such sera are not easily available. We do recognise that artificial sera can be made, but we favour the use of a water based standard graph for reasons of simplicity, convenience and reduction in the risk of contamination, and thus recommend the utilisation of peak-area measurements. An estimate of the precision of the method was obtained by analysing aliquots of the same serum sample several times daily on eight consecutive days. In this fashion the within-batch precision was found to be 2-3% RSD, and the between-batch precision ca.5% RSD. The sample was from a healthy person and contained 1.06 k 0.05 mg 1-1 of zinc ( N = 8). The accuracy was established by analysing three different sera with known contents of zinc. Firstly, a control serum obtained from Herlev Hospital having a zinc concentration of 0.99 k 0.03 mg 1-1: on three different days we found for the same serum 0.99 k 0.05 mg 1-1 ( N = 15), 0.98 f 0.04 mg 1-1 ( N = 16) and 1.00 f 0.04 mg 1-1 ( N = 6). Secondly, the NBS Reference Material 8419 bovine serum6 having a recommen- ded zincvalue of 1.1 k 0.1 mg 1-1: we found 1.10 k 0.01 mg 1-1 (N = 2).Thirdly, the freeze-dried control serum Seronorm batch number 105, with reported preliminary information values of zinc as determined by FAAS ranging from 0.72 to 1.02 mg 1-1: we found 0.78 k 0.03 mg 1-1 ( N = 15). The limit of detection, defined as the blank value plus three times the standard deviation of the blank value, is 7.2 ng. This value was calculated from 20 determinations of the blank,12 which was an artificial serum.9 For a 50-pl sample this corresponds to a zinc concentration of 0.14 mg 1-1. In adults, the normal range of serum zinc is 0.75-1.20 mg 1-1, making the present method ideally suited for routine screening of human sera. In conclusion, we have presented a fully automated FI-AAS method for the determination of zinc in human serum that by most standards is ideal in being fast, cost effective, accurate, precise and sensitive, with only small sample requirements.The method relies on peak-area measurements and the daily construction of an aqueous calibration graph is mandatory. References Casey, C. E., and Hambidge, K. M., in Nriagu, J. O., Editor, “Zinc in the Environment,” Part 11, Wiley, New York, 1980, Smith, J. C., Jr., Butrimovitz, G. P., and Purdy, W. C., Clin. Chem., 1979,25, 1487. Rfiiitka, J., and Hansen, E. H., “Flow Injection Analysis,” Wiley, New York, 1981. Tyson, J. F., Analyst, 1985, 110, 419. Rocks, B. F., Sherwood, R. A., Bayford, L. M., and Riley, C., Ann. Clin. Biochem., 1982, 19, 338. pp. 1-27.456 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, DECEMBER 1986, VOL. 1 6. Attiyat, A. S., and Christian, G. D., Clin. Chim. Acta, 1984, 137, 151. request. 7. Sherwood, R. A., Rocks, B. F., and Riley, C., Analyst, 1985, 110,493. 8 . Computer program may be obtained from the authors on request. 9. Derry, J. E., McLean, W. M., and Freeman, J. B., 1. Parent. 11. Detailed diagrams may be obtained from the authors on 12. American Chemical Society Committee on Environmental Improvement, Anal. Chem., 1980,52, 2242. Ent. Nutr., 1983, 7, 131. 10. Veillon, C., Lewis, S . A., Patterson, K. Y., Wolf, W. R., Harnley, J. M., Versieck, J., Vanballenberghe, L., Cornelis, R., and O’Haver, T. C., Anal. Chern., 1985,57, 2106. Paper J6l43 Received June 16th, 1986 Accepted August 8th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100453

出版商:RSC    

年代:1986

数据来源: RSC