摘要:

Journal of Analytical Atomic Spectrometry (Including Atomic Spectrometry Updates - Formerly ARAAS) JAAS Editorial Board* Chairman: L. C. Ebdon (Plymouth, UK) J. Brew (London, UK) M. S. Cresser (Aberdeen, UK) D. L. Miles (Wallingford, UK) B. L. Sharp (Aberdeen, UK) M. Thompson (London, UK) A. M. Ure (Aberdeen, UK) *The JAAS Editorial Board reports t o the Analytical Editorial Board, Chairman J. D. R. Thomas (Cardiff, UK) JAAS Advisory Board F. C. Adams (Antwerp, Belgium) R. M. Barnes (Amherst, MA, USA) L. Bezur (Budapest, Hungary) R. F. Browner (Atlanta, GA, USA) S. Caroli (Rome, Italy) L. de Galan (Delft, The Netherlands) J. B. Dawson (Leeds, UK) K. Dittrich (Leipzig, GDR) W. Frech (Umeii, Sweden) K. Fuwa (Tokyo, Japan) A. L. Gray (Guildford, UK) S. Greenfield (Loughborough, UK) G.M. Hieftje (Bloomington, IN, USA) G. Horlick (Edmonton, Canada) J. J. LaBrecque (Vienna, Austria) 6. V. L'vov (Leningrad, USSR) J. M. Mermet (Villeurbanne, France) Ni Zhe-ming (Beijing, China) N. Omenetto (lspra, Italy) E. PlSko (Bratislava, Czechoslovakia) R. E. Sturgeon (Ottawa, Canada) R. Van Grieken (Antwerp, Belgium) A. Walsh, K. 6. (Victoria, Australia) B. Welz (Uberlingen, FRG) T. S. West (Aberdeen, UK) Atomic Spectrometry Updates Editorial Board Chairman: *M. S. Cresser (Aberdeen, UK) R. M. Barnes (Amherst, MA, USA) N. W. Barnett (Plymouth, UK) *J. Brew (London, U? *A. A. Brown (Cambridge, UK) J. C. Burridge (Aberdeen, UK) J. B. Dawson (Leeds, UK) *L. C. Ebdon (Plymouth, UK) H. J. Ellis (Ross-on-Wye, UK) J. Fijalkowski (Warsaw, Poland) D. J.Halls (Glasgow, UK) S. J. Haswell (London, UK) *D. A. Hickman (London, UK) G. M. Hieftje (Bloornington, IN, USA) S. J. Hill (Plymouth, UK) H. Hughes (Anglesey, UK) P. N. Keliher (Villanova, PA, USA) K. Kitagawa (Nagoya, Japan) *D. Littlejohn (Glasgow, UK) C. W. McLeod (SheHield, UK) K. W. Jackson (Saskatoon, Canada) F. J. M. J. Maessen (Amsterdam, The Nether- lands) *J. Marshall (Middlesbrough, UK) *D. L. Miles (Wallingford, UK) J. M. Mermet (Villeurbanne, France) E. Norval (Pretoria, South Africa) I. Novotny (Brno, Czechoslovakia) P. E. Paus (Oslo, Norway) P. R. Poole (Hamilton, New Zealand) T. C. Rains (Washington, DC, USA) J. M. Rooke (Leeds, UK) G. Rossi (lspra, Italy) I. RubeSka (Prague, Czechoslovakia) W. Slavin (Norwalk, CT, USA) R. Stephens (Halifax, Canada) J.Stupar (Ljubljana, Yugoslavia) A. Taylor (Guildford, UK) M. Thompson (London, UK) J. F. Tyson (Loughborough, UK) *A. M. Ure,jAberdeen, UK) B. Welz (Uberlingen, FRG) J. B. Willis (Victoria, Australia) *B. L. Sharp (Aberdeen, UK) *Members of the ASU Executive Committee Editor, JAAS: Judith Brew The Royal Society of Chemistry, Burlington House, Piccadilly, London WIV OBN, UK. Telephone 01 -734 9864. Telex No. 268001 US Associate Editor, JAAS: Dr. J. M. Harnly US Department of Agriculture, Beltsville Human Nutrition Research Center, BLDG 161, BARC-EAST, Beltsville, MD 20705, USA. Telephone 301-344-2569 Advertisements: Advertisement Department, The Royal Society of Chemistry, Burlington House, Piccadilly, London W1V OBN. Telephone 01-437 8656. Telex No. 268001 Journal ofAnalytical Atomic Spectrometry IJAAS) (ISSN 0267-9477) is published eight times a year by The Royal Society of Chemistry, Burlington House, London WlVOBN, UK.All orders accompanied with payment should be sent directly to The Royal Society of Chemistry, The Distribution Centre, Blackhorse Road, Letchworth, Herts.SG6 lHN, UK. 1987 Annual subscription rate UK f180.00, Rest of World f202.00, USA $356.00. Air freight and mailing in the USA by Publications Expediting Inc., 200 Meacham Avenue, Elmont, NY 11003. USA Postmaster: send address changes to Journal of Analytical Atomic Spectrometry (JAAS), Publications Expediting Inc., 200 Meacharn Avenue, Elmont, NY 11003. Second class postage pending at Jamaica, NY 11431. All other despatches outside the UK by Bulk Airmail within Europe, Accelerated Surface Post outside Europe.PRINTED IN THE UK. @The Royal Society of Chemistry, 1987. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form, or by any means, electronic, mechanical, photographic, recording, or otherwise, without the prior permission of the publishers. Information for Authors Full details of how to submit material for publication in JAASare given in the Instructions to Authors in Issue 1. Separate copies are available on request. The Journal of Analytical Atomic Spectrometry (JAAS) is an international journal for the publi- cation of original research papers, short papers, communications and letters concerned with the development and analytical application of atomic spectrometric techniques.The journal will be published bimonthly, will include com- prehensive reviews of specific topics of interest to practising atomic spectroscopists and will incorporate the literature reviews which were previously published in Annual Reports on Analytical Atomic Spectroscopy (ARAAS). Manuscripts intended for publication must describe original work related to atomic spec- trometric analysis. Papers on all aspects of the subject will be accepted, including fundamental studies, novel instrument developments and practical analytical applications. As well as AAS, AES and AFS, papers will be welcomed on atomic mass spectrometry and X-ray fluoresc- ence/emission spectrometry. Papers describing the measurement of molecular species where these relate to the characterisation of sources normally used for the production of atoms, or are concerned, for example, with indirect methods of analysis, will also be acceptable for publication.Papers describing the development and applications of hybrid techniques (e.g., GC-coupled AAS and HPLC - ICP) will be parti- cularly welcome. Manuscripts on other subjects of direct interest to atomic spectroscopists, including sample preparation and dissolution and analyte preconcentration procedures, as well as the statistical interpretation and use of atomic spectrometric data will also be accept- able for publication. There is no page charge. The following types of papers will be con- sidered. Full papers, describing original work. Short papers. the criteria for originality are the same as for full papers, but short papers generally report less extensive investigations or are of limited breadth of subject matter.Communications, which must be on an urgent matter and be of obvious scientific importance. Communications receive priority and are usually published within 2-3 months of receipt. They are intended for brief descriptions of work that has progressed to a stage at which it is likely to be valuable to workers faced with similar problems. Reviews, which must be a critical evaluation of the existing state of knowledge on a parti- cular facet of analytical atomic spectrometry. Every paper (except Communications) will be submitted to at least two referees, by whose advice the Editorial Board of JAAS will be g'uided as to its acceptance or rejection.Papers that are accepted must not be published else- where except by permission. Submission of a manuscript will be regarded as an undertaking that the same material is not being considered for publication by another journal. Manuscripts (three copies typed in double spac- ing) should be addressed to: Judith Brew, Editor, JAAS The Royal Society of Chemistry, Burlington House, Piccadilly, London WIV OBN, UK US Associate Editor, JAAS Dr. J. M. Harnly US Department of Agriculture, Beltsville Human Nutrition Research Center, BLDG 161, BARC-EAST, Beltsville, MD 20705, USA or AH queries relating to the presentation and submission of papers, and any correspondence regarding accepted papers and proofs, should be directed to the Editor or US Editor (addresses as above).Members of the JAASEditorial Board (who may be contacted directly or via the Editorial Office) would welcome comments, suggestions and advice on general policy mat- ters concerning JAAS. Fifty reprints are supplied free of charge.J ou r n a I of Ana I yt ica I Atomic Spectrometry (Including Atomic Spectrometry Updates - Formerly ARAAS) JAAS Editorial Board* Chairman: L. C. Ebdon (Plymouth, UK) J. Brew (London, UK) M. S. Cresser (Aberdeen, UK) D. L. Miles (Wallingford, UK) B. L. Sharp (Aberdeen, UK) M. Thompson (London, UK) A. M. Ure (Aberdeen, UK) *The JAAS Editorial Board reports t o the Analytical Editorial Board, Chairman J. D. R. Thomas (Cardiff, UK) JAAS Advisory Board F. C. Adams (Antwerp, Belgium) R. M. Barnes (Amherst, MA, USA) L.Bezur (Budapest, Hungary) R. F. Browner (Atlanta, GA, USA) S. Caroli (Rome, Italy) L. de Galan (Delft, The Netherlands) J. B. Dawson (Leeds, UK) K. Dittrich (Leipzig, GDR) W . Frech (UmeA, Sweden) K. Fuwa (Tokyo, Japan) A. L. Gray (Guildford, UK) F. Greenfield (Loughborough, UK) G. M. Hieftje (Bloomington, IN, USA) G. Horlick (Edmonton, Canada) B. V. L'vov (Leningrad, USSR) J. M. Mermet (Villeurbanne, France) Ni Zhe-ming (Beijing, China) N. Omenetto (Ispra, Italy) E. PlSko (Bratislava, Czechoslovakia) R. E. Sturgeon (Ottawa, Canada) R. Van Grieken (Antwerp, Belgium) A. Walsh,.<K. B. (Victoria, Australia) B. Welz (Uberlingen, FRG) T. S. West (Aberdeen, UK) Atomic Spectrometry Updates Editorial Board Chairman: *M. S. Cresser (Aberdeen, UK) R.M. Barnes (Amherst, MA, USA) N. W. Barnett (Plymouth, UK) *J. Brew (London, UK) *A. A. Brown (Cambridge, UK) J. C. Burridge (Aberdeen, UK) J. B. Dawson (Leeds, UK) J. R. Dean (Norwich, UK) *L. C. Ebdon (Plymouth, UK) H. J. Ellis (Ross-on-Wye, UK) J. Fijalkowski (Warsaw, Poland) D. J. Halls (Glasgow, UK) S. J. Haswell (London, UK) *D. A. Hickman (London, UK) G. M. Hieftje (Bloomington, lN, USA) S. J. Hill (Plymouth, UK) H. Hughes (Anglesey, UK) P. N. Keliher (Villanova, PA, USA) K. Kitagawa (Nagoya, Japan) *D. Littlejohn (Glasgow, UK) K. W. Jackson (Saskatoon, Canada) F. J. M. J. Maessen (Amsterdam, The Nether- lands) *J. Marshall (Middlesbrough, UK) *D. L. Miles (Wallingford, UK) J. M. Mermet (Villeurbanne, France) E. Norval (Pretoria, South Africa) I. Novotny (Brno, Czechoslovakia) P.E. Paus (Oslo, Norway) P. R. Poole (Hamilton, New Zealandl T. C. Rains (Washington, DC, USA) J. M. Rooke (Leeds, UK) G. Rossi (lspra, Italy) I. RubeSka (Prague, Czechoslovakia) A. Sanz-Medel (Uviedo, Spain) *B. L. Sharp (Aberdeen, UK) W. Slavin (Norwalk, CT, USA) R. Stephens (Halifax, Canada) J. Stupar (Ljubljana, Yugoslavia) A. Taylor (Guildford, UK) M. Thompson (London, UK) J. F. Tyson (Loughborough, UK) *A. M. Ure,(Aberdeen, UK) B. Welz (Uberlingen, FRG) J. B. Willis (Victoria, Australia) *Members of the ASU Executive Committee Editor, JAAS: Judith Brew The Royal Society of Chemistry, Burlington House, Piccadilly, London W I V OBN, UK. Telephone 01 -734 9864. Telex No. 268001 US Associate Editor, JAAS: Dr. J. M. Harnly US Department of Agriculture, Beltsville Human Nutrition Research Center, BLDG 161, BARC-EAST, Beltsville, MD 20705, USA.Telephone 301-344-2569 Advertisements: Advertisement Department, The Royal Society of Chemistry, Burlington House, Piccadilly, London W I V OBN. Telephone 01-437 8656. Telex No. 268001 Journal ofAnalytical Atomic Spectrometry (JAAS) (ISSN 0267-9477) is published eight times a year by The Royal Society of Chemistry, Burlington House, London WIVOBN, UK. All orders accompanied with payment should be sent directly to The Royal Society of Chemistry, The Distribution Centre, Blackhorse Road, Letchworth, Herts. SG6 IHN, UK. 1987 Annual subscription rate UK €180.00, Rest of World f202.00, USA $356.00. Air freight and mailing in the USA by Publications Expediting Inc., 200 Meacham Avenue, Elmont, NY 11003.USA Postmaster: send address changes t o Journal of Analytical Atomic Spectrometry (JAAS), Publications Expediting Inc., 200 Meacham Avenue, Elmont, NY 11003. Second class postage pending at Jamaica, NY 11431. All other despatches outside the UK by Bulk Airmail within Europe, Accelerated Surface Post outside Europe. PRINTED IN THE UK. @ The Royal Society of Chemistry, 1987. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form, or by any means, electronic, mechanical, photographic, recording, or otherwise, without the prior permission of the publishers. Information for Authors Full details of how to submit material for publication in JAASare given in the Instructions to Authors in Issue 1.Separate copies are available on request. The Journal of Analytical Atomic Spectrometry (JAAS) is an international journal for the publi- cation of original research papers, short papers, communications and letters concerned with the development and analytical application of atomic spectrometric techniques. The journal is published eight times a year, includes com- prehensive reviews of specific topics of interest to practising atomic spectroscopists and incor- porates the literature reviews which were pre- viously published in Annual Reports on Analy- tical Atomic Spectroscopy (ARMS). Manuscripts intended for publication must describe original work related to atomic spec- trometric analysis. Papers on all aspects of the subject will be accepted, including fundamental studies, novel instrument developments and practical analytical applications.As well as M S , AES and AFS, papers will be welcomed on atomic mass spectrometry and X-ray fluoresc- encelemission spectrometry. Papers describing the measurement of molecular species where these relate to the characterisation of sources normally used for the production of atoms, or are concerned, for example, with indirect methods of analysis, will also be acceptable for publication. Papers describing the development and applications of hybrid techniques (e.g., GC-coupled AAS and HPLC - ICP) will be parti- cularly welcome. Manuscripts on other subjects of direct interest to atomic spectroscopists, including sample preparation and dissolution and analyte preconcentration procedures, as well as the statistical interpretation and use of atomic spectrometric data will also be accept- able for publication.There is no page charge. The following types of papers will be con- sidered. Full papers, describing original work. Short papers: the criteria for originality are the same as for full papers, but short papers generally report less extensive investigations or are of limited breadth of subject matter. Communications, which must be on an urgent matter and be of obvious scientific i m portance. Corn m u n ications receive priority and are usually published within 2-3 months of receipt, They are intended for brief descriptions of work that has progressed to a stage a t which it is likely to be valuable to workers faced with similar problems. Reviews, which must be a critical evaluation of the existing state of knowledge on a parti- cular facet of analytical atomic spectrometry. Every paper (except Communications) will be submitted to a t least two referees, by whose advice the Editorial Board of JAAS will be guided as to its acceptance or rejection. Papers that are accepted must not be published else. where except by permission. Submission of a manuscript will be regarded as an undertaking that the same material is not being considered for publication by another journal. Manuscripts (three copies typed in double spac- ing) should be addressed to: Judith Brew, Editor, JAAS The Royal Society of Chemistry, Burlington House, Piccadilly, London W1V OBN, UK Dr. J. M. Harnly US Associate Editor, JAAS US Department of Agriculture, Beltsville Human Nutrition Research Center, BLDG 161, BARC-EAST, Beltsville, MD 20705, USA or All queries relating to the presentation and submission of papers, and any correspondence regarding accepted papers and proofs, should be directed to the Editor or US Editor (addresses as above). Members of the JAASEditorial Board (who may be contacted directly or via the Editorial Office) would welcome comments, suggestions and advice on general policy mat- ters concerning JAAS. Fifty reprints are supplied free of charge.

ISSN:0267-9477    

DOI:10.1039/JA98702FX005

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

March 1987 JASPE2 2(2) 87-262 Journal of Analytical Atomic Spectrometry THIRD BIENNIAL NATIONAL ATOMIC SPECTROSCOPY SYMPOSIUM, BRISTOL, UK, 23-25 JULY, 1986 CONTENTS 87 89 95 105 115 125 131 135 143 151 157 163 167 171 177 185 189 197 201 205 21 1 217 221 227 233 239 245 Foreword-E. J. Newman A Physicist's View on Current Questions in Atomic Spectrometry-Leo de Galan Recent Advances in the Theory of Atomisation in Graphite Furnace Atomic Absorption Spectrometry: the Oxygen - Carbon Alternative. Plenary Lecture-Boris V. L'vov Mathematical Correction of Systematic Temporal Background-correction Errors for Graphite Furnace Atomic Absorption Spectrometry-James M. Harnly, James A. Holcombe Determination of Phosphorus by Graphite Furnace Atomic Absorption Spectrometry. Part 2.Comparison of Different Modifiers-Adilson J. Curtius, Gerhard Schlemmer, Bernhard Welz Atomisation Characteristics of Lead Determined in Alumina Matrices by Slurry - Electrothermal Atomisation Atomic Absorption Spectrometry-Regina Karwowska, Kenneth W. Jackson Direct Atomic Spectrometric Analysis by Slurry Atomisation. Part 2. Elimination of Interferences in the Determination of Arsenic in Whole Coal by Electrothermal Atomisation Atomic Absorption Spectrometry-Les Ebdon, Huw G. M. Parry Electrothermal Atomisation Atomic Absorption Spectrometric Determination of Inorganic and Methylated Arsenic after Pre-concentration by Hydride Generation and Trapping the Hydrides in a Cerium(lV) - Iodide Absorbing Solution-Dimiter L. Tsalev, Petko B. Mandjukov, J. A. Stratis Graphite Furnace Atomic Absorption Spectrometric Determination of Selenium in Plant Materials Following Combustion in a Stream of Oxygen-Charles A.Shand, Allan M. Ure Direct Electrothermal Atomistion Atomic Absorption Spectrometric Determination of Selenium in Whole Blood and Serum with Continuum-source Background Correction-Khalid Saeed Determination of Lead in Biological Materials by Atomic Absorption Spectrometry Sensitised with Hydride Generation-Milagros Bonilla, (the late) Lourdes Rodriguez, Carmen Camara Determination of Vanadium in Urine by Electrothermal Atomisation Atomic Absorption Spectrometry-Pilar Bermejo-Barrera, F. Bermejo-Martinez, J. A. Cocho de Juan Determination of Gold in the Presence of Platinum and Palladium by Electrothermal Atomisation Atomic Absorption Spectrometry-Krystyna Brajter, Krystyna Stonawska Between-batch Variability of Thermal Characteristics of Commercially Available L'vov Platform Graphite Tube Atomisers and Analytical Accuracy in Electrothermal Atomisation Atomic Absorption Spectrometry-Ian L.Shuttler, H. Trevor Delves Atomic Spectrometric Methods (Atomic Absorption and Inductively Coupled Plasma Atomic Emission) for the Determination of Aluminium at the Parts per Billion Level in Biological Fluids-Alfred0 Sanz-Medel, Rosa Rodriguez Roza, Ricardo Gonzalez Alonso, Alejandro Nova1 Vallina, Jorge Cannata Correction of Matrix Effects in Inductively Coupled Plasma Atomic Emission Spectrometry by Interactive Power Adjustment-Michael Thompson, Michael H. Ramsey, Barry J. Coles, Chong Ming Du Determination of Tungsten and Molybdenum a t Low Levels in Geological Materials by Inductively Coupled Plasma Mass Spectrometry-Gwendy E.M. Hall, Chang J. Park, J. C. Pelchat Determination of Arsenic by Hydride Generation in the D.C. Plasma Atomic Emission Spectrometer-Determination of Arsenic(ll1) and ArsenicW) as Total Arsenic-Charles Boampong, Ian D. Brindle, Claudio M. Ceccarelli Ponzoni Methods for Improving the Sensitivity in Flame Atomic Absorption Spectrometry-Alistair A. Brown, D. J. Roberts, K. V. Kahokola Interface System for Directly Coupled High-performance Liquid Chromatography - Flame Atomic Absorption Spectrometry-Les Ebdon, Steve Hill, Philip Jones Indirect Atomic Absorption Determination of Chloride by Continuous Precipitation of Silver Chloride in a Flow Injection System-Pi I ar Martinez J i menez, Mercedes Ga I leg 0, M ig uel Va Ica rcel Flow Manifold for Automated On-line Dilution of Standards for Flame Atomic Absorption Spectrometry and its Use in a Null Measurement Method-Stephen R.Bysouth, Julian F. Tyson Pneumatic Nebulisers-Poor Pumps and Inferior Sub-samplers-lgnacio Lopez Garcia, Clare O'Grady, Malcolm Cresser High-resolution Fourier Transform Atomic Spectrometry-Anne Thorne Atomic Magneto-optical Rotation Spectroscopy (AMORS) Using a Rotating Polariser and a Segmented Graphite Rod Atomiser-John B. Dawson, Roger J. Duffield, Alan D. Kersey, Mohsen Hajizadeh-Saffar, Graham W. Fisher Charge-transfer Excitation Processes in the Grimm Lamp-Edward B. M. Steers, Richard J. Fielding Role of Energy Dispersive X-ray Fluorescence Spectrometry in Process Analysis of Plastic Materials-Peter L.Warren, Olivier Farges, Murray Horton, Joe Humber Continued inside back coverSHORT PAPERS 249 Determination of Dissolved Sulphite in Groundwaters by Inductively Coupled Plasma Atomic Emission Spectrometry -Kathryn Lewin, J. Nicholas Walsh, Douglas L. Miles 251 Excitation of Gallium, Indium, Selenium, Tellurium, Arsenic and Antimony in a Helium Microwave-induced Plasma-Katherine J. Timmins 253 Use of 1,5-Bis(di-2-pyridylmethylene)thiocarbonohydrazide as an Extracting Reagent for the Determination of some Transition Metal Ions by Atomic Absorption Spectrometry-A. Bustos, F. Sanchez-Rojas, C. Bosch Ojeda, A. Garcia de Torres, J. M. Can0 Pavon 257 Graphite Furnace Atomic Absorption Spectrometric Screening Method for the Determination of Aluminium in Haemodialysis Concentrates-Jan Rud Andersen 261 Direct Determination of Cobalt in Acetic Acid Extracts of Soils by Graphite Furnace Atomic Absorption Spectrometry- Margaret C.Mitchell, Michael L. Berrow, Charles A. Shand The Fourth Biennial National Atomic Spectroscopy Symposium will be held at the University of York on 1988 28 June - 1 July 1988 1988 The symposium will provide a forum where interesting and useful applications of atomic spectroscopy can be reported and discussed. In addition to plenary, invited and submitted lectures, a particular feature of the meeting will be the presentation of posters. There will also be an exhibition and a full social program for delegates and their guests. This meeting is organised by the Atomic Spectroscopy Group, Analytical Division of the Royal Society of Chemistry and the Spectroscopy Group of the Institute of Physics. Further information can be obtained from the Secretary of the organising committee: Dr R Miller, Unilever Research, Port Sunlight Laboratory, Quarry Road East, Bebington, Wirral, Merseyside L63 3JW Circle 003 for fu'rther information Typeset and printed by Black Bear Press Limited, Cambridge, England

ISSN:0267-9477    

DOI:10.1039/JA98702BX007

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JASPE2 l(8) 737-834,211 R-242R (1 987) December 1987 Journal of Analytical Atomic Spectrometry Including Atomic Spectrometry Updates CONTENTS NEWS AND VIEWS 737 737 738 739 741 741 741 743 743 744 Editorial-Les Ebdon Fourth Biennial National Atomic Spectroscopy Symposium-David Hickman Fourth BNASS Student Bursaries Conference Report Book Review AS U High I ig h t s- D av i d H i c k m a n 1987 Hilger Spectroscopy Prize Conference and Meetings Gordon F. Kirkbright Bursary Fund Papers in Future Issues PAPERS 745 765 773 785 793 80 1 805 809 813 819 823 829 833 Matrix-effect Observations in Inductively Coupled Plasma Mass Spectrometry- Samantha H. Tan, Gary Horlick Effect of Operating Parameters on Analyte Signals in Inductively Coupled Plasma Mass Spectrometry-Margaret-Anne Vaughan, Gary Horlick, Samantha H.Tan Influence of Matrix Effects on Methods for the Quantification of Major and Impurity Elements in Brass Using Secondary Ion Mass Spectrometry (SIMS)-Frank Michiels, Freddy Adams Spatially Resolved Absorbance Profiles Detailing the Selenium Vaporisation Process in Electrothermal Atomisers-Maureen S. Droessler, James A. Holcombe Investigations of Interferences in Graphite Furnace Atomic Absorption Spectrometry Using a Dual-cavity Platform. Part 2. Influence of Sodium Chloride and Nickel Chloride on the Atomisation of Lead-Bernhard Welz, Suleyman Akman, Gerhard Schlemmer High-resolution Gas Chromatography With Graphite Furnace Atomic Absorption Spectrometry as the Detection System-Olle Nygren Determination of Trimethyllead Salts in Blood Using High-resolution Gas Chromato- graphy - Graphite Furnace Atomic Absorption Spectrometry-Olle Nygren, Carl-Axel N i lsson Indirect Determination of Cationic Surfactants in Frozen Squid by Flame and Electrothermal Atomisation Atomic Absorption Spectrometry-Pedro Martinez Gonzalez, Carmen Camara Rica, Luis Polo Diez Determination of Manganese, Calcium, Magnesium and Potassium in Pine (Pinus Caribaea) Needle Samples by Flame Atomic Absorption Spectrometry With Slurry Sample Introduction-Nereida Carrion, Zully A.de Benzo, Elias J. Eljuri, Franco Ippoliti, Daniel Flores Simplex Optimisation of a 5-kW Nitrogen-cooled Argon Inductively Coupled Plasma for Maximum Signal to Background Ratios and Minimum Matrix InterferenceGlyn L. Moore, Reinhard G. Bohmer Investigation of Small Volume Cloud Chambers for Use in Inductively Coupled Plasma Nebulisation-Phillip L.Kempster, Jacobus F. Van Staden, Henk R. Van Vliet A Capacitively Heated Tungsten Spiral Atomiser for Atomic Fluorescence Spectro- metric Analysis-Boris V. Arkhangelskii, Alex S. Gonchakov, Svetlana S. Grazhulene COMMUNICATION On the Determination of Analyte Transport Efficiency in Inductively Coupled Plasma Atomic Emission Spectrometry-Gertjan Kreuning, Frans J. M. J. Maessen ATOMIC SPECTROMETRY UPDATE 21 1R Minerals and Refractories-David A. Hickman, Joan M. Rooke, Michael Thompson 231R References Typeset and printed by Black Bear Press Limited, Cambridge, Englandii The Chemical Analysis of Water: General Principles and Techniques 2nd Edition by A.L. Wilson and D. T. E. Hunt, Water Research Centre, Medmenham Hardcover 704pp ISBN 0 85186 797 9 Price f55.00 ($99.00) RSC Members f36.00 This new edition covers the considerable developments which have taken place in the eleven years since the first edition was published, in the measurement of water quality with particular reference to methods for estimating and controlling possible errors in analytical results. Ordering: Non-RSC Members should send their orders to: The Royal Society of Chemistry, Distribution Centre, Blackhorse Road, Letchworth, Herts SG6 1 HN, UK. RSC Members should send their orders to: Brief Contents: Information Requirements of Measurement Programmes; Sampling; The Nature and Importance of Errors in Analytical Results; Estimation and Control of the Bias of Analytical Results; Estimation and Control of the Precision of Analytical Results; Achievement of Specific Accuracy by a Group of Laboratories; Reporting Analytical Results; The Selection of Analytical Methods; General Precautions in Water-Analysis Laboratories; Analytical Techniques; Automatic and On-Line Analysis; Computers in Water Analysis.The Royal Society of Chemistry, Membership Manager, 30 Russell Square, London WC1 B 5DT, UK. ROYAL SOCIETY OF C HE M I STRY lnformat ion Services Circle 004 for further information ROYAL SOCIETY OF CHEMISTRY Analytical Division East Anglia Region Atomic Spectroscopy and Automatic Methods Groups SPECTROSCOPY ACROSSTHESPECTRUM: Analytical Applications of Spectroscopy 12-15 July 1987 University of East Anglia Norwich, UK Incorporating The First International Near Infrared Spectroscopy Conference (Organised by the International Committee for Near Infrared Spectroscopy) University of East Anglia, 12-1 7 July 1987 Affiliated Groups: Association of British Spectroscopists British Mass Spectrometry Society International Committee for Near Infrared Analysis United Kingdom Chemometrics Discussion Group The aim of the meeting is to bring together spectroscopists from many different disciplines with the prospect of an interchange of ideas and methods.The meeting will be organised in three parts. General sessions, poster ses- sions and parallel specialist sessions. There will be an equipment exhibition and a social programme. An internationally recognised group of specialists have been invited to present the plenary and keynote lectures which cover the areas of: Combined techniques, Data Analysis and Fourier transform spectroscopy.Parallel sessions are planned in the following areas: near infrared spectroscopy, atomic absorption spectroscopy, mass spectrometry, NM R spectroscopy, microwave spec- troscopy, infrared spectroscopy, process control, and chemometrics. FURTHER INFORMATION Dr. C. S. Creaser, School of Chemical Sciences, University of East Anglia, NORWICH NR4 7TJ, UK. 2 for further informationRamon M. Barnes, Editor Department of Chemistry GRC Towers University of Ma8sachusetts Amherst, MA 01003-0035 Tel. (41 3) 545-2294 Objectlve The ICP Information Newsletter is a monthly journal published by the Plasma Research Group at the University of Massachu- setts and is devoted exclusively to the rapid and impartial dissem- ination of news and literature information related to the devel- opment and applications of plasma sources for spectrochemical analysis.Background ICP stands for inductively coupled plasma discharge, which dur- ing the past decade has become the leading spectrochemical excitation source for atomic emission spectroscopy. ICP sources are also applied commercially as an atom and ion cell in atomic fluorescence spectrometry and as an ion source for mass spec- trometry. The popularity of this source and the need to collect in a single literature reference all of the pertinent data on ICP stimulated the publication onhe ICP Information Newsletter in 1975. Other plasma sources, such as microwave induced plas- mas and direct current plasma jets, have also grown in popularity and are included in the scope of the lCP Information Newsletter.S c o p As the only authoritative monthly journal of its type, the ICP Information Newsletter is read in more than 40 countries by scientists actively applying or planning to use the ICP or other types of plasma spectroscopy. For the novice in the field, the ICP Information Newsletter provides a concise and systema- tic source of information and background material needed for the selection of instrumentation or the development of new meth- odology. Editorial The ICP Information Newsletter is edited by Dr. Ramon M. Barnes, Professor of Chemistry, University of Massachusetts at Amherst, with the assistance of a 20-member Board of National Correspondents composed of leading plasma spectroscopists.The Board members from around the world report news, view- points, and developments. Dr. Barnes has been conducting plasma research on ICP and other discharges since 1968. Healso serves as chairman of the Winter Conferences on Plasma Spectrochemistry. Regular Fartuna Original submitted and invited research articles by ICP and plasma experts. Complete bibliography of all major ICP publications from 1961 to the present. Abstracts of all ICP papers presented at major US and interna- tional meetings. *First-hand accounts of ICP developments from na- tions around the world. Special reports on microwave and other plasma progress. Calendar and advanced programs of plasma meetings.Publication of plasma-related patents. Technical translations and reprints of critical foreign- Critical reviews of plasma-related books. language ICP papers. Conference Activities The /CP /#formation Newsletter has sponsored five international meetings on developments in atomic plasma spectrochemical analysis since 1980 in San Juan, Orlando, San Diego, Leysin, Switzerland, and Kailua-Kona, HI. Meeting proceedings have ap- peared as Developments in Atomic Plasma Spectrochemical Analy- sis (Wiley), Plasma Spectrochemistry and Plasma Spectrochemistry II (Pergamon Press) as well as in special issues of Spectrochimica Acta, Part B and Journal of Analytical Atomic Spectrometry. Subscription Information Subscriptions are available for 12 issues on either an annual or volume basis.The first issue of each volume begins in June and the last issue is published in May. For example, Volume 12 runs from June 1986 through May 1987. Back issues beginning with Volume 1, May 1975 are also available. To begin a subscription, complete the attached order form, and submit it with prepayment or purchase in- formation. For additional information please call (41 3) 545-2294 or contact the Editor. Detach and send to: ICP Information Newsletter, Dr. Ramon M. Barnes Department of Chemistry, GRC Towers, University of Ma8sachusett8, Amherst, MA 01003-0035 Telephone (413) 545-2294 Start a subscription for the following issues (complete): [ ] Volume(s) - (June 198-- May 198-) or [ ] 198- (January-December) I enclose: [ ] prepayment or [ ] purchase order (No. ) or [ ] Send invoice. Current subscription rates are $49 (North America), $69 (Europe, South America), or $75 (Africa, Asia, Indian/Pacific Ocean Areas, Middle East, and USSR). Back issues rates available on request.I I I I I I I I I I I I I I I I I I I I I I I I I I I I FOLD HERE I JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAR‘87 READER ENQUIRY SERVICE FOLD HERE I I t I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1 Postage will be paid by Licensee Do not affix Postage Stamps if posted in Gt. Britain, Channel Islands, N. Ireland or the Isle of Man I I I BUSINESS REPLY SERVICE Licence No. WD 106 Reader Enquiry Service Journal of Analytical Atomic Spectrometry The Royal Society of Chemistry Burlington House, Piccadilly LONDON WIE 6WF England I I I I I I I I 1 I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1 I I I I I I I I I I I I I I I I I ! I I I I ! I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I

ISSN:0267-9477    

DOI:10.1039/JA98702FP009

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 87 Foreword Third Biennial National Atomic Spectroscopy Symposium : Bristol, UK, July 23rdD25th, 1986 A Special issue of TheAnaZyst (February 1983) was devoted to a series of papers dealing with new developments in atomic spectroscopy, that were presented at the First Biennial National Atomic Spectro- scopy Symposium (BNASS) held in Shef- field in July 1982. A similar special issue (May 1985) dealt with material from the 2nd BNASS held in Leeds in July 1984. The BNASS series of meetings has completed a memorable double with this, the first Special Issue of the Journal of Analytical Atomic Spectrometry, which provides a record of much of the original work presented at the 3rd BNASS held at the University of Bristol last summer.The Biennial National Atomic Spectro- scopy Symposia are organised jointly by the Atomic Spectroscopy Group of the Analytical Division, Royal Society of Chemistry, and the Spectroscopy Group of the Institute of Physics. The 3rd BNASS differed from its predecessors by being integrated with the 7th SAC Con- ference, organised by the Analytical Divi- sion of the RSC, which ran from the 20th to the 26th July. The abbreviation SAC is derived from the Society for Analytical Chemistry, the forerunners of the RSC’s Analytical Division, who founded the highly acclaimed series of SAC Confer- ences. These are international confer- ences, dealing with all aspects of ana- lytical chemistry, and they occur every three years. Inevitably, every six years an SAC Conference and a BNASS will both be due to occur and just as inevitably the former would contain a large contribution from analytical atomic spectroscopy, and so it was decided to combine them in 1986.The outcome of this experiment was generally judged by the organisers and the delegates to have been successful. The symposium opened on the Wed- nesday afternoon, when SAC delegates traditionally attend social events, indus- trial visits and up-date courses. About 200 SACIBNASS delegates attended the Association of British Spectroscopists Lecture given by Professor Leo de Galan entitled “A Physicist’s View on Current Questions in Atomic Spectrometry.” The first poster session for 3rd BNASS was also held during the afternoon and was well attended, and discussion was catalysed by the suitable refreshments provided by courtesy of ARL Ltd.and Chelsea Instruments Ltd. The second poster session was held during Thursday. In all, 57 poster themes were displayed and the standard of presentation was exceptional. Although it was hard to make a choice, the award for the best poster from a student was won by Simon Sparkes from Plymouth Polytechnic on the subject of slurry atomisation. The Thursday and Friday plenary lec- tures of the SACIBNASS meeting had been specially selected to appeal to atomic spectroscopists. Regrettably, Professor L’vov from Leningrad, who was expected to speak on Thursday, had to cancel. However, his place was filled most ably at short notice, and some sacrifice of his family holiday, by Dr. Jean-Michel Mermet who spoke on “Mixed Gas or Air ICPs: Toys or Tools?” Friday’s plenary lecture was given by Professor G.Tolg whose subject was “Extreme Trace Analysis of the Elements-The State of the Art Today and Tomorrow.” The lecture programme consisted of 33 presentations, timetabled in two parallel streams, on Thursday and Friday. Of these, eight had been invited to give overviews and themes, and together with the contributed papers and posters they gave the delegates a comnprehensive account of recent advances in atomic spectroscopy. The invited lecturers and their topics were: V. Sychra et al., “Advances in metal- based electrothermal atomisers”; K. Dittrich, “The use of lasers and other non-thermal excitation in atomic spectroscopy for trace analysis”; Anne Thorne, “Fourier transform atomic spectroscopy”; R.D. Snook, “Torch con- figuration and designs for inductively coupled plasma atomic emission spec- trometry”; M. s. Cresser, “Pneumatic nebulisers-poor pumps and inferior sub- samplers?”; A. R. Date, “ICP-MS: the best thing in analytical chemistry since chopped light?”; G. J. Oliver, “X-ray fluorescence analysis in the ceramic and allied industries”; and D. A. Hickman, “Analysis in forensic science using atomic spectroscopy. ” The lecture sessions covered elec- trothermal atomisation, non-thermal excitation, background correction, Fourier transform atomic spectroscopy, atomic emission spectroscopy, nebulisers and sample introduction, ICP-MS, X-ray techniques and other aspects of applied atomic spectroscopy. It has always been a particular aim of BNASS to provide younger workers in the field of atomic spectroscopy with an opportunity to participate fully in a major conference.The organisers were very pleased that many students were able to attend. We are most grateful to BDH, Chelsea Instruments, Hilger Analytical, ICI, Pye Unicam and Shell for providing funds to be used as student bursaries. It was particularly welcome and appropriate that the first student to be awarded a bursary by the Kirkbright Memorial Fund, Curtiss Monnig of the University of Indiana, attended this symposium. The members of both the SAC86 Ex- ecutive Committee and the 3rd BNASS Organising Committee are to be congratu- lated for the quality and content, scientific and social, of the meeting, Particular thanks are due to Dr. Neil Barnett, Chairman of the 3rd BNASS Scientific Programme Committee, and his col- leagues, for arranging an interesting, lively and well balanced programme. We look forward with enthusiasm to the Fourth National Biennial Atomic Spec- troscopy Symposium, for which plans are now being made, to be held at the University of York in 1988. We believe readers will be interested and enlightened by the contents of this Special Issue of JAAS, individual copies of which may be purchased from the Royal Society of Chemistry, Distribution Centre, Blackhorse Road, Letchworth, Herts., SG6 lHN, UK. E. J. Newman Chairman of 3rd BNASS Organ ising Corn M ittee

ISSN:0267-9477    

DOI:10.1039/JA9870200087

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 89 A Physicist‘s View on Current Questions in Atomic Spectrometry Leo de Galan Laboratorium voor Analytische Chemie, Technische Universiteit Deli?, de Vries van He ystplantsoen 2, 2628 RZ Delft, The Netherlands Simple considerations and elementary physics are used to address four questions raised in connection with current developments in analytical atomic spectrometry. How can we increase the speed of ETA-AAS? What are the limitations to sample introduction in ICP-AES, also with respect to solid samples? What are the merits of hig h-resolution ICP-AES? How can non-metals be determined directly? Keywords: Multi-element atomic absorption spectrometry; high-resolution atomic emission spectrometry; solid samples; non-metal determination ~~ If anything, the recent launching of this journal bears witness to the vitality of analytical atomic spectrometry.Another indication is the continuous flow of new instrumental tech- niques proposed in the literature. In a recent report the author has presented an overview and a personal assessment of current developments.1 In the present paper four key topics are selected for a more detailed discussion based on simple calculus and elementary physics. High-speed Electrothermal Atomisation Atomic Absorption Spectrometry For some time to come electrothermal atomisation will the probe directly at atomisation temperatures, but attempts have been somewhat disappointing. Probably, the mass of even the smallest probe (5 mg) is so much larger than that of the dried sample (0.1 mg) that rapid heating to atomisation temperatures is not possible and the signals become broad and weak.Another possible technique for high-temperature sample introduction is droplet spraying against a pre-heated furnace wall. There seems no reason why the furnace temperature could not be raised from the 400 K used in the FASTAC system to a much higher temperature with a corresponding gain in cool-down time. Possibly, this could reduce the cycle time of ETA analysis to 0.5 min, or one sixth of the present remain the method of choice for the ultratrace analysis of metals, most conveniently in combination with atomic absorp- tion spectrometric detection (ETA-AAS). With the advent of the platform technique and accurate background correction, the only drawback is the slow speed of ETA-AAS, which is due to the combination of two shortcomings: the long cycle time of the furnace analysis and the sequential, single-element nature of the atomic absorption measurement.Efforts to increase the speed of ETA-AAS must, therefore, attack either or both of these limitations. as only 10 s are used to collect the signal during the atomisation stage [Fig. l(a)]. Incidentally, it is this low duty cycle that allows Zeeman instruments to employ magnets value. 2500 The typical cycle time of ETA of 3 min is the more annoying 700 without water cooling. Major fractions of the time are spent in drying (1 min), ashing (1 rnin) and in cooling and sample introduction (a further 1 rnin), Unfortunately, these stages cannot be ignored without punishment.Slow drying is necessary for the simple reason that 10 pl of liquid solution evaporate to 100 ml of vapour at atomisation temperatures and crash volatilisation would sweep the sample out of the 1 ml large furnace. In fact, by the same token, a total salt content of 1% by itself almost fills the furnace volume if it were instantaneously atomised without removing a maj or portion in the preliminary ashing step. Nevertheless, we can cut the cycle time of ETA in half, if we introduce the sample rapidly at an elevated temperature of, for example, 700 K, because we gain the drying time at the front end and the cool-down time at the tail end of the cycle [Fig. l(b)]. One possibility, extensively described in the literature, is the probe technique, provided we heat the probe with one sample outside the furnace to 700 K while the previous sample is being ashed and atomised inside the furnace.It might be thought that we can also obviate the ashing step by introducing * Association of British Spectroscopists’ Lecture, presented at the Third Biennial National Atomic Spectroscopy Symposium (BNASS) , Bristol, UK, 23rd-25th July, 1986. (a) 10 s At o rn i se from platform Ash 60 s f- 10 pl liquid background 10 ml vapour desolvate 60 s 3 y 100 2 +- 1 2 (z1 1 2 Timeim in 3 Fig. 1. Heating cycle in electrothermal atomisation: (a) typical times needed for the various steps in the cycle; and ( b ) possible reduction realised with high-temperature sample introduction90 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 2.0 AAS ZAA &8 kG Dip ZAAH 6-8 kG I 1 1 I 2.5 5.0 7.5 10.0 Concentration/mg I-’ Fig. 2. Extending the dynamic range of ETA-AAS by the simul- taneous measurement of two calibration graphs (ZAA and ZAAH) of different sensitivity4 for the Pb 283.3-nm line from an EDL Table 1. Composition of the ICP vapour Rate/ Density/ * Species Introduction pmol s-1 cm-3 Ar . . . , . . llmin-1 loo0 1.5 x 1018 Ar 5 x 1015 Ar+ H20 . . . . 0.01 x3mlmin-1 25 4 x 1016 H,O Matrix . . . . 1% 0.1 1014 Na+ 1014 O+ Analyte . . . . lpgml-l 10-5 1010 Detection limit 1 ng ml-1 10-8 107 The second option to increase the speed of ETA is to use the brief atomisation time more efficiently by progressing from a single-element measurement to a simultaneous multi-element measurement.If we remain with the well-proven absorption technique, we are confronted with three obstacles. Firstly, we must provide for a primary radiation source and a poly- chromator detection system that emits and records spectral lines of more than one element simultaneously. It would seem that a combination of multi-element hollow-cathode lamps and beam-combination techniques2 allows an extension from one element to ten elements. Admittedly, higher demands are placed on the resolving power of the polychromator to isolate the spectral lines of interest from nearby unwanted lines. Still, this is technically feasible. The second problem is the limited dynamic range of the absorption measurement at the single wavelength prescribed by the hollow-cathode lamp.Adequate precision can only be obtained over about 1.5 decades.3 For simultaneous multi- element determinations this means that the concentrations of the various analyte elements in the sample must conform to the sensitivity of their transitions used, which is clearly impractical. Extension of the dynamic range can be realised by various means, as follows. (i) For each element the polychromator can be programmed to select lines of varying sensitivity; this adds to the complexity of the system and for some elements the choice of spectral lines is rather limited. (ii) If an a.c.-driven magnet is used in a Zeeman instrument, the power supply can be modified to allow measurement at three rather than two field strengths, so that the background corrected signal is measured at the peak (ZAA) and towards the wing (ZAAH) of the absorption line; as illustrated in Fig.2 the dynamic range is thus extended by another decade to a total span of 2.5 decades.4 (iii) Both the primary radiation source problem and the dynamic range limitation have been solved by an elegant extension of the same principle in the wavelength-modulated continuum-source high-resolution Cchelle spectrometer.5 By measuring increasingly away from the absorption peak the dynamic range could be extended to five decades. The disadvantage of even the brightest continuum sources used is their low intensity and therefore poorer detection limits towards the far-UV (h < 250 nm). Apparently, the foregoing two obstacles might be over- come, if we replace the absorption measurement by an emission measurement.The furnace itself has too low a temperature for elements with resonance lines in the far-UV (such as Zn and Cd), even if they are released at the highest temperature of 3300 K. Much better results are obtained when furnace atomisation is combined with post-excitation in another discharge, such as an ICP, a microwave discharge or a hollow-cathode discharge. The latter combination, described in the literature by the acronym FANES,6 seems the most profitable one, although it remains to be seen how interfer- ence effects are accommodated. However, there remains a third obstacle. Whatever the measurement technique used, simultaneous multi-element analysis implies a single set of “compromise” conditions for the entire sample, a situation well known in ICP analysis.The realisation of such compromise conditions is by far the most serious challenge to multi-element ETA analysis.7 New furnace designs and materials, such as pyrolytic graphite, might solve this problem in the future. Sample Introduction in Inductively Coupled Plasma Atomic Emission Spectrometry The three common, pneumatic nebulisers (cross-flow, concen- tric and V-groove) used for introducing solution into an inductively coupled plasma (ICP) exhibit two disadvantages: a modest stability and a low efficiency. The first effect is at least partly responsible for a lower than desired precision of ICP analysis, whereas the second feature is wasteful of sample and perhaps responsible for the fact that detection limits do not surpass the ng ml-1 level.What can we do to improve the situation? The instability is not due to statistical fluctuations in the number of droplets entering the ICP. If we assume that the nebulisation chamber transmits 1% of the liquid uptake rate of 3 ml min-1 in droplets with a diameter of 5 pm, then we calculate the droplet introduction rate as 107 s-1. Even with a time constant as small as 0.1 s, the statistical variation of the Poisson distribution is only 0.1%. The actual, much larger instability is, therefore, an inherent property of the nebuliser - chamber combination. Since variations in sample introduction affect all elements equally, an obvious remedy is the use of an internal standard and, indeed, this technique is gaining popularity in ICP analysis.However, the advantage gained when the internal standard is selected in accordance with the analyte under consideration,s demonstrates that the imprecision of ICP data resides at least partly in the discharge itself. On the other hand, plasma instability may well be due to the nebuliser, because the intensity variation of argon lines increases significantly with the introduction of the sample. Therefore, a reconsideration of the nebuliser - chamber combinations used for ICP is certainly worthwhile. Such a reconsideration should also include the low effi- ciency (1%) of aerosol transport through the nebulisation chamber. It would seem that the sensitivity of ICP analyses could easily be increased by an order of magnitude by simply raising this value to 10% or more.In practice, this is not so simple. There is an upper limit to the amount of solvent that can be introduced into the discharge. Even when introduced in the vapour state, too much solvent extinguishes the plasma. The explanation may be appreciated from the data in Table 1, which present the composition of the plasma under current operating conditions. Apparently, the atoms, ions and elec- trons in the discharge result predominantly from the plasma carrier gas (Ar) and this feature explains the stability of the ICP against sample influences. Indeed, at the introduction rate mentioned above the solvent contributes only a few per cent. of the total number of atoms and electrons in the plasma. A further increase by a factor of ten would mean thatJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 91 the discharge is carried to a large extent by electrons from the solvent, which requires additional energy and makes the discharge very sensitive to variations in solvent introduction and sample interchange. It might be argued that solvent overloading can be readily avoided by adding a desolvation stage to the sample introduc- tion system. Indeed, this is the general practice with devices that employ increased solution introduction rates, e.g., by heating the nebulisation chamber or by using ultrasonic nebulisers. The resulting increase in analytical sensitivity and the corresponding decrease in the detection limits are well documented,g but such data invariably refer to academic solutions, containing only analyte.With real samples it has been observed that interferences can only be maintained at an acceptable level as long as the total salt content is kept below 0. YO. This is a factor of ten less than permitted at the ten-fold lower, conventional solution introduction rate. Apparently , there is an upper limit not only to the acceptable solvent loading of the plasma, but also to the total salt entering the discharge. The reason is two-fold. Referring again to Table 1, we see that if an easily ionised matrix element such as Na is introduced at a much higher rate than 0.1 pmol s-1, it contributes significantly to the total number of electrons and thus influences the plasma discharge. This is an undesirable situation that has contributed much to the bad reputation of d.c.plasmas. For less easily ionised matrices there may exist other limitations. If the desolvation system allows more and larger droplets to pass into the ICP, the number and the size of the resulting sample crystals also increase over their values under standard operating conditions. The latter values are readily calculated. As was remarked earlier, for a solvent uptake rate of 3 ml min-1 a 1% efficient nebulisation system gives rise to 107 5 ym droplets per second or a total solvent introduction rate of 0.5 mg s-1. For a salt content of 1% the corresponding figure for the desolvated sample is 5 pg s-1 distributed over l o 7 crystals per second with a diameter of 1 ym. There are indications that these figures approach the upper limits for the current medium power ICP! For example, in the determina- tion of wear metals in oil, the particulate material must remain under 5 pm to warrant calibration with dissolved standards.Also, the total salt introduction rate of 5 pg s-1 compares well with similar data in other techniques, as follows. ( i ) In the by now almost forgotten carbon arc, 5 mg of sample were volatilised freely, and erratically, in 5 min, constituting a rate of 15 pg s-1. (ii) In ETA analyses 1% of sample in 10 p1 amounts to 0.1 mg; if only one tenth survives ashing and is consecutively atomised in 1 s the introduction rate is 10 pg s-1. (iii) In cathodic sputtering with a hollow-cathode or a glow discharge the maximum sputtering rate is 30 pg s-1.10 The quoted upper limits of ym particle size introduced at a total rate of a few pg s-1 also provide a sobering perspective for the direct analysis of solid samples by either ICP or ETA. For the general, not extremely volatile, sample this is only possible if the above limits are not exceeded.Clearly, weighing 10 pg of sample into an ETA furnace seems just as impractical as grinding solids to a diameter of 1 pm particle size. This pessimistic view is confirmed by reports in the literature. With standard commercial sources accurate results have only been obtained for finely dispersed slurries (in the case of the ICP) of for volatile (organic) samples (in the case of ETA).11 For the more typical samples the minimum require- ment is calibration against matched standards. 12 High-resolution Inductively Coupled Plasma Spectrometry Spectral interferences are, perhaps, the last remaining prob- lem of ICP analysis using emission spectrometric detection.They make it necessary to select a “proper” analysis line depending on the sample composition and, in turn, frequently deteriorate the detection limits in comparison with the low values achieved with the most sensitive transition. Attempts to relieve or overcome the problem are, therefore, easily understandable. A drastic way out is to replace the emission technique by another detection mode, notably fluorescence mass spec- trometry. Indeed, the fluorescence spectrum is much “cleaner” than the emission spectrum. The difficulty is the appropriate primary source. Unless a cheap, stable, tunable, far-UV laser becomes available, the propsects for ICP-AFS remain at best uncertain.Mass spectrometry seems to be in a better position, since a free-lying isotope is available for every element, and if this is not the most abundant one, the high sensitivity of mass spectrometry can accommodate that. Unfortunately, the ICP vapour does not only contain atoms and ions, but also molecular fragments. Because they are weak emitters, their low abundance creates few problems in ICP emission, but in mass spectrometry molecular ions are detected just as sensitively as atomic ions. Considering again Table 1, we see that the matrix exceeds the analyte by 4 to 7 orders of magnitude. Consequently, interference from rnol- ecular masses can only be avoided if the matrix is atomised to six 9s. This is difficult to realise.If we remain with emission as the detection mode, the obvious direction is towards even higher optical resolution than the value of 10 pm realised in current top-quality instruments. Two possible solutions have been indicated. A modest reduction of the spectral band pass down to 3 pm is realised with a monochromator equipped with an Cchelle grating used in high order.13 The latter condition requires the inclusion of an order sorter to remove order overlap, which adds to the complexity and presumably to the cost of the instrument. A more exciting proposal is the use of a Fourier transform instrument, where the monochromator is replaced by a Michelson interferometer (Fig. 3). Indeed, the theory of the Michelson interferometer relates the distance traversed by the moving mirror, xmaX, to the band pass, Ah, as x,,, = 1/2Aa = h2l2Ah where u is the wavenumber, so that a distance of 2 cm already corresponds to a spectral band pass of only 1 pm (at a wavelength of 200 nm).Simultaneously, the spacing between two data readings is determined by the minimum wavelength as Ax = 1/20,, = h,i,/2 Consequently, data must be read every 0.1 pm to extend the wavelength range down to 200 nm, so that high demands are placed on the (laser controlled) stability of the device. Incidentally, the number of data read remains equal to the number of spectral channels: or about 20000 in the present example. However, there is a fundamental objection to the use of an FT instrument in the ultraviolet region of the spectrum for \I/ Detector Fig. 3. Schematic diagram of a Michelson interferometer92 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 Band width 30 pm 3 Pm Table 2. Fellgett (dis)advantage in Fourier transform spectrometry Configuration Signal Noise S/N “Gain” Scanning . . . . . . I , (VARx)0,5 Ix/(VARx)0.5 - . . . . . . . . nIx (nVART)O.s f i I ~ / ( V A R ~ ~ s ~ ~ ( V A R X / V A R T ) O . ~ IR: VART= VAR, . . - (nVARx)0.5 fiZJ(VAR) . fi uv: VART= nI . . - (n n1)O.S Zx/IO.5 VQi Table 3. Determination of non-metals Fraction ionised, Element h,,,/nm eV 8000K Ediss (MO)/ T = Zn . . . . 213 As . . . . 194 S . . . . 182 I . . . . 178 P . . . . 178 C . . . . 165 Br . . . . 154 C1 . . . . 135 8 5 4 2 6 11 2 2 0.18 0.04 0.02 0.02 0.03 0.01 0.01 0.003 Boltzmann factor T = 6000 K 1 0.33 0.15 0.11 0.11 0.04 0.01 0.001 T = 12 000 K ~160 110 90 90 50 30 10 analytical purposes.For each item of data collected, the integrated intensity over the complete spectrum is measured. This simultaneous recording is known as Fellgett’s advantage in FT instruments developed for the infrared region, but turns out to be a disadvantage for the UV spectrum. This is demonstrated in Table 2, where the signal to noise ratios of various configurations have been collected. The first row presents reference data for the sequential registration of the spectral channels in a conventional scanning instrument. In the FT instrument the integrated signal is magnified by the number of channels ( n ca. 20000) and so is the variance. Consequently, the gain in signal to noise (the square root of the variance) is equal to the square root of the channel number n multiplied by the ratio of the variance of one channel, VARx, over that in the integrated spectrum, VART.It is this ratio which is the critical parameter. For the detectors used in the IR region the variance is independent of the recorded intensity, so that VARx = VART and, hence, there is a Fellgett advantage of The photomultiplier used in the UV, however, exhibits a variance proportional to the recorded intensity and, as a result, the Fellgett “advantage” becomes equal to the square root of the ratio of Ix, the intensity in the channel X under consideration, to the average intensity, I , in the spectrum. Clearly, in ICP analysis the abundance of strong lines from the argon and the matrix makes the average intensity larger than the intensity of a weak line from a trace analyte.The consequence is a deterioriation of the signal to noise ratio depending on the strength of the matrix spectrum. This result does not diminish the value of the FT-UV instrument for fundamental studies requiring high resolution, but makes it less attractive for analytical purposes.14 The final question to be answered is the real merit of high resolution for the problem of spectral interferences con- sidered here. To this end Fig. 4 presents the experimentally observed profile, when a spectral transition is measured with increasingly higher resolution. The representative example refers to the combination of a Lorentzian atomic line profile with a width of 3 pm and a triangular monochromator profile resulting from entrance and exit slits of equal width.15 Initially, for a large band pass, the experimental profile remains triangular and becomes narrower in proportion to the spectral band pass.Once the value of 10 pm achieved in current instruments is surpassed, however, the triangle becomes distorted and the experimental profile slowly approaches the true line profile. Consequently, we need a ten-fold increase in optical resolution to achieve a further Wavele ng th/pm Fig. 4. Intensity distribution recorded when an atomic spectral line (half width 3 pm) is measured with a spectral band width decreasing from 30 to 1 pm three-fold reduction in profile width. It is true that the conclusion depends on the width (much less on the actual shape) of the atomic spectral lines, but recent data show this value to range from 2 to 7 pm,l3 so the example in Fig.4 is not unfavourable. It would appear that the modest reduction of spectral interferences offered by high-resolution instruments could also be realised to a large extent by a software program aiming at deconvoluting the spectrum for instrumental broadening or curve fitting of overlapping bands. A first attempt in this direction is the software discrimination between spectral lines and spectral background.16 Determination of Non-metals Except for the vacuum spark, atomic spectrometry is almost exclusively concerned with the determination of metals and metalloids. If non-metals are determined it is either indirectly, or, for the elements I, P and S , with lower sensitivity.Now the sensitivity of an atomic spectrometric determination can be described as the product of four factors: S = A p (1 - a) exp (-E/kT) where A is the transition probability that determines the inherent strength of a transition, p is the degree of dissocia- tion, a is the degree of ionisation and the exponential function is the Boltzmann factor describing the population of the energy level from which the transition starts and that has an energy E above the ground state. There is no reason why the first factor, A , should be much lower for non-metals than for metals. For the other three factors relevant data are collected in Table 3. The low dissociation energies for oxides in the third column of the table illustrate that atomisation need not be a problem for the halogens, sulphur and phosphorus.Carbon (and nitrogen) are much more strongly associated with oxygen, but their sensitive determination is generally precluded by high blank values anyway. By contrast, the ionisation energy of the non-metals is high, because it must exceed the highest excitation energy for which the resonance wavelength in the second column is an inverse measure. Consequently, the elements are little ionised even at a high temperature of 8000 K. This explains why mass spectrometric detection (as in ICP-MS) offers no solution for the determination of non-metals. The final two columns in the table list the Boltzmann excitation factors for emission relative to the value of zinc atJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 93 6000 K. The latter value is chosen, because it represents the situation in the ICP where zinc can be determined with adequate sensitivity. The data for As, S, I and P explain the lower sensitivity observed for these elements, although the emission intensities might still be adequate. Clearly, the excitation factor alone accounts for a reduction in emission sensitivity by two to three orders of magnitude in the case of Br and Cl. To restore the situation the excitation temperature must be raised to about 12 000 K as shown by the final column in the table. Such a high temperature is realised in spark discharges, but not in the ICP. Possibly, sufficiently high excitation temperatures can be obtained in noble gas discharges at reduced pressure and power.For example, in a 100-W microwave discharge at a few torr an electron temperature of 40 000 K generates sufficient emission of halogen atoms and ions to permit gas-chromato- graphic detection. Unfortunately, the gas kinetic temperature in the extremely non-thermal low-pressure discharge is so low (less than 1000 K) that the sample can only be introduced after preliminary volatilisation and even then interferences are notorious.17 Increasing the pressure and the power of the discharge promotes the kinetic energy, but simultaneously reduces the electron temperature to the slightly non-thermal situation present in the ICP. It appears, therefore, that the determination of halogens in condensed samples by emission spectrometry is impossible. A potentially viable alternative is absorption spectrometry, provided we use a resonance transition starting from the ground state, for which E = 0.The combination of low dissociation energy, high ionisation energy and high excitation energies of other levels favours the formation of a high concentration of ground-state atoms. As is clear from the second column in Table 3 the resonance transitions that must be used in the absorption measurement lie in the vacuum ultraviolet region, so that suitable radiation sources must be developed and the atomiser must be operated in a noble gas atmosphere. Neither demand seems unsurmountable. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. References de Galan, L., Anal. Chem., 1986, 58, 697A. Salin, E. D., and Ingle, J. D., Appl. Spectrosc., 1978,32,579. van Dalen, J. P. J., and de Galan, L., Analyst, 1981, 106,695. de Loos-Vollebregt, M. T. C., de Galan, L., and van Uffelen, J. W. M., Spectrochim. Acta, Part B, 1986,41, 825. Harnly, J. M., and O’Haver, T. C . , Anal. Chem., 1981, 53, 1291. Falk, H., Hoffmann, E., and Ludke, Ch., Spectrochim. Acta, Part B , 1984,39,283. Lewis, S. A., O’Haver, T. C., and Harnly, J. M., Anal. Chem., 1985, 57,2. Lorber, A., and Goldbart, Z . , Anal. Chem., 1984, 56,37. Boumans, P. W. J. M., and de Boer, F. J., Spectrochim. Acta, Part B , 1975,30, 309. Caroli, S . , Prog. Anal. At. Spectrosc., 1983, 6, 253. Vollkopf, U., Grobenski, Z . , Tamm, R., and Welz,, B., Analyst, 1985, 110, 573. Shao, Y., and Horlick, G., Appl. Spectrosc., 1986, 40, 385. Boumans, P. W. J. M., and Vrakking, J. J. A. M., Spectro- chim. Acta, Part B , 1984, 39, 1239. Faires, L. M., Spectrochim. Acta, Part B , 1985, 40, 1473. de Galan, L., and Winefordner, J. D., Spectrochim. Acta, Part B , 1968, 23, 277. Taylor, P., and Schutyser, P., Spectrochim. Acta, Part B , 1986, 41, 81. Matousek, J. P., Orr, B. J., and Selby, M., Spectrochim. Acta, Part B , 1986,41, 415. Paper J6l63 Received July 23rd, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200089

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 95 Recent Advances in the Theory of Atomisation in Graphite Furnace Atomic Absorption Spectrometry: the Oxygen - Carbon Alternative* Plenary Lecture Boris V. L'vov Department of Analytical Chemistry, Polytechnical Institute, Leningrad 195251, USSR Two approaches to the problem of incomplete analyte atomisation in graphite furnace atomic absorption spectrometry based on the formation of either monoxides or carbides in the gas phase are compared. The discussion of the first approach provides an explanation of an apparent discrepancy between the comparatively high (lo-8-10-7 atm) free oxygen partial pressure in the sheath gas and the absence of its effect on atomisation under typical analytical conditions. Within the second approach we have revealed, experimentally studied and theoretically interpreted the major features of the appearance in graphite furnaces of gaseous carbon at enhanced levels compared with equilibrium concentrations.The theory of gaseous carbide formation developed on this basis has provided an explanation for many unusual effects caused by the furnace material, sample vaporisation techniques and sheath gas (Ar vs. N2) on the shape and magnitude of analytical signals, and theoretically substantiated the advantages of the stabilised temperature platform furnace. Keywords: Graphite furnace atomic absorption spectrometry; graphite activation; metal - O2 interaction kinetics; non-equilibrium gaseous carbon; gaseous carbides State of the Art Today, almost 30 years after the birth of graphite furnace atomic absorption spectrometry (GFAAS) ,1 and about 15 years after the advent of the first commercial electrothermal atomisers, we are in a position to maintain that this method has won the recognition of analytical chemists all over the world as one of the most sensitive, selective and simple analytical techniques.The automation and computerisation of instrumentation, the refinement of the atomisation technique and the use of the Zeeman effect to take into account spectral interferences have made AA spectrometers a reliable tool for completely automated analysis. The concept of the stabilised temperature platform furnace (STPF) introduced by Slavin2J in the early 1980s, and presently enjoying widespread use, has brought us close to the development of an absolute method of analysis.4 A possibility has emerged for the first time in the history of instrumental methods of excluding from analytical procedures the use of standards and even regular calibration using pure analyte solutions.Against the background of these truly revolutionary5 technical and methodological achievements one would con- sider all the more anachronistic the slow progress in the theory of atomisation, the absence of consensus and, sometimes, even of any opinion whatsoever regarding the variety of unusual effects in GFAAS, some of which are outlined below. The reason for the anomalously low sensitivity for a number of elements in graphite furnaces, in particular for boron and the lanthanides, remains unclear.The reasons for many of the effects of furnace material on atomic absorption are also unknown.6.7 For example, when evaporated from the wall of an uncoated graphite tube, the characteristic masses of Al, Ge, Si and Sn are smaller than those obtained with pyrolytic graphite coated tubes, whereas for V, Mo and Ti the reverse is true.6 Nothing can be said about the reasons for the enhancement of the signal when analytes are vaporised from the platform and lanthanides in a Ta-foil lined furnace.8 An explanation cannot yet be suggested for the enhanced *Plenary lecture at SAC 86, the 7th SAC International Conference on Analytical Chemistry, and BNASS, the Third Biennial National Atomic Spectroscopy Symposium, Bristol, UK, 20-26 July, 1986. (Unable to be presented.) Pre-printed from the April 1987 issue of The Analyst.sensitivity for some elements (Pd, Ge) in nitrogen compared with argon, and for differences in the vaporisation rates of B, Ba, Sr, Er, Eu, Sc, Ti, V and some other elements in nitrogen and argon.9 There is no consensus as to the interpretation of the molecular spectra obtained with graphite furnaces. Some authorslOJ1 believe monoxides to be responsible for the spectra observed in the vaporisation of oxygen-containing compounds of Al, Ba, Sr and Ca, whereas othersl2J3 attribute them to monocyanides and carbides. The purpose of this paper is to discuss the status of the problem and to explain these and some other unusual phenomena based on the results of our recent studies. In agreement with most researchers, we believe that the key to understanding the mechanism of these phenomena lies in the interaction of the analyte with oxygen and carbon.Let us consider this problem from the viewpoints held by the proponents of the oxygen and carbon hypotheses. Oxygen Differences in Evaluating the O2 Partial Pressure The partial pressure of free oxygen present in graphite furnaces has recently become a subject of heated debate between several research groups, Three different approaches to this problem have surfaced, namely, an equilibrium approach,14-16 based on assuming thermodynamic equilibrium to exist between the carbon and the sheath gas; a hypothetical approachl7J8 involving an (a priori) assumption of a linear variation of the quantity -l0glOp(O2, atm) from 10 to 20 in the range 1500-2500 K, and an empirical approachl9--21 based on direct and indirect measurements of the oxygen partial pressure.An analysis of the results and arguments presented in some of these papers has revealed a number of aspects that are either misunderstood or left out of the considerations. Total Oxygen Pressure in Graphite Furnaces Some comparisonsl8.21 of the hypothetical with the empirical pOz scales neglect the fact that in both direct and indirect measurements the total pressure of oxygen is determined, i.e., P(02)z = P(02) + 0.5p(O) * . (1)96 JOURNAL OF rather than p ( 0 2 ) . At high temperatures these quantities may differ substantially. Indeed, at 2500 K, p ( 0 2 ) = 10-20 atm corresponds top(0) = 1.4 x 10-12 atm, so that p(O2)X = 0.7 x 10-12 atm.Thus the difference between the empirical and hypothetical scales at 2500 K constitutes only five orders of magnitude rather than twelve18321 (Table 1). Note that this difference remains approximately constant over the range 1500-2500 K. Arguments Against the Empirical p ( 0 2 ) Scale and their Criticism The main argument advanced against the empirical scale in which p(02) varies from 10-6 to 10-7 atm over the range 1500-2500 K is the thesis18 that at such oxygen pressures the atomisation of a number of elements (B, Si) forming thermally stable metal - oxygen molecules is practically impossible. In order to estimate the degree of dissociation of the monoxides, the existence of the following equilibrium is tacitly assumed: MO(g) = M(g) + 0 . . . . * - (2) It has, however, been pointed out19 that as we are speaking here about the excess oxygen pressure above the equilibrium level, “this conclusion is valid only in the case where the partial pressure of element in the gas phase at each moment of time is much less than that of oxygen, i e . , p(M) << p z ( 0 2 ) .In the opposite case, practically all of the oxygen present will be bound in a gaseous oxide MO, the purge gas becoming ‘pure’ for the excess amount of the element.” It is this which accounts for the noticeable absorption of silicon and boron18 when they are introduced into the furnace in amounts (2 X 10-10 and 5 x 10-9 mol, respectively) exceeding by far the total amount of free oxygen present in the furnace under these conditions (7 X 10-12 mol, 2500 K, argon flow 0.8 cm3 s-1).13 Another argument put forward against the empirical scale is the so-called buffering effect of the excess CO that is assumed18721 to maintain a low and constant concentration of oxygen in the furnace, irrespective of its original content in the sheath gas.This buffering effect, however, is based on the same heterogeneous equilibrium C(S) + 0.502=CO . . . . - - (3) ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 that underlies the purely thermodynamic approach. 1 4 ~ 6 As shown convincingly by Sturgeon et aZ. ,21 this equilibrium is not reached. Moreoever, contrary to the conclusions drawn by the above workers,21 their data on the partial pressures of CO and C02 prove persuasively that the partial pressure o f free oxygen corresponding to the equilibrium co,=co+o .. . . (4) setting in at temperatures above 2200 K is in good agreement with earlier measurements.1992° This is obvious from Table 2, which presents the values of ~ ( 0 ~ ) ~ calculated by us. Thus all methods for the direct or indirect determination of the O2 partial pressure in furnaces with even a low argon flow (ca. 0.4 cm3 s-l), without exception, yield results that agree well with one another. This supports the validity of the empirical scale. Kinetic Limitations of the M - 0 2 Interaction In evaluating the equilibrium M(g) + 0 2 = MO(g) + 0 . . . . ( 5 ) the finite time (T~) of atom residence in the furnace, which restricts substantially the number of collisions between interacting particles, is not taken into account. The impor- tance of this factor in gas-phase equilibrium studies has been pointed out by Holcombe et aZ.22 as early as 1979, although this investigation was carried out using rod rather than tube atomisers.Table 3 presents the lowest concentrations and partial pressure of oxygen required for a single collision of the metal atom with the O2 molecule to occur, calculated by us for different experimental conditions. The number of collisons was calculated by well known relationships,23 assuming for the mean radius and mean molar mass of the colliding particles the values 1.5 x 10-10 m and 0.05 kg mol-1 for T =: 2000 K, p = 1 atm. As seen from Table 3, the minimump(02) that still can be of significance for process (5) is ca. 10-9 atm ,under gas-stop conditions, ca. 10-8 atm with a flow of 0.8 cm3 s-1 and ca.10-7 atm with a flow of 5 cm3 s-1. At still lower values of p(Oz) the Table 1. Total oxygen pressure in graphite furnace Argon p(Oz),/atm flow-rate/ Scale Reference cm3 s-1 1500 K 2000 K 2500 K Equilibrium . . . . 14 - 2 x 10-1s 5 x 10-15 5 x 10-13 Hypothetical. . . . 17 - 2 x 10-11 1 x 10-11 5 x 10-13 Empirical . . . . 19 0.8 5 x 10-7 2 x 10-7 4 x 10-8 20 0.4-2.1 1 x 10-6 2 x 10-7 4 x 10-8 Table 2. Calculation of p(O& from measurementsz1 of p(CO,)lp(CO) ratio. In a T-shaped pyrocoated tube, at stationary tem- perature, and under an argon flow-rate of 0.4 cm3 s-1 P(CO2) T/K p(C0) K*latm p ( O)/atm p ( Oz)/atm p ( Oz),/atm 2118 5.1 x 10-2 5.07 X 10-6 2.6 X 10-7 2.9 X 10-8 1.6 x 10-7 2378 5.9 x 10-3 1.42 x 10-4 8.4 x 10-7 1.2 x 10-8 4.3 x 10-7 2620 3.6 X 10-4 1.68 X 6.0 X lo-’ 5.7 X 3.0 x 10-7 * Equilibrium constant for reaction (4).Table 3. Minimum values of concentration (nmin) and partial pressure (pmin) of gaseous reagent for a single collision of analyte atom with a reagent molecule in an HGA-type furnace Sheath gas Gas flow-rate/ nminl flow mode cm3 s-1 zR/s molecule cm-3 pmin/atm Gasstop . . . . - 0.25 1.2 x 1010 3.3 x 10-9 Mini-flow . . 0.8 0.05 6.2 x 1010 1.7 X 10-8 Fullflow.. . . 5.0 0.01 3.1 x 1011 8.3 x 10-8JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 97 metal atoms leave the furnace without undergoing a single collision with the O2 molecules. Taking into account the kinetic limitations of reaction ( 5 ) determined by the rate constants of the metals involved, the minimum values ofp(O2) at which chemical equilibrium can be obtained may be higher still.Hence, any debate on the reasons for the variation ofp(02) in the sheath gas below the above levels has no practical significance. Attempts at a thermodynamic evaluation of reaction (5) under these conditions appear to be just as useless. Conclusions Thus, the residual partial pressure of oxygen in a pure sheath gas (<1 X lO-3% 0,) with even a low flow-rate (0.4 cm3 s-1) through the graphite furnace and for temperatures up to 2500 K is not less than 10-7 atm. When measuring metal vapour pressures less than 10-7 atm this may result in a reduced senstivity for some elements because of the formation of MO molecules, The effect of O2 becomes more pronounced for higher sheath gas flow-rates or for higher 0 2 partial pressures in the sheath gas.At low furnace temperatures typical of the vaporisation of volatile elements (Zn, Bi, Pb) this may affect not only the peak area but also its position. This is connected with the effect of O2 on the process of oxide vaporisation.19 At the higher temperatures required for the vaporisation of In , Ga, Sn and Si the concentration of O2 near the furnace wall decreases, with the result that the effect of 0 2 is confined to the furnace axis and manifests itself in a reduced peak area. This conclusion is in accord with the experiments of Byrne et al. 16 At metal vapour partial pressures considerably in excess of 10-6 atm the influence of O2 on the peak area is negligible, even with intense sheath gas flow-rates, owing to the purification effect.19 The only consequence of this effect is the appearance of a curvature in the calibration graph when the amounts of O2 and M in the gas phase become comparable.For the typical analytical conditions of temperatures above 2200 K and the gas-stop mode, the residual oxygen pressure in pure sheath gas in a pyrocoated tube is below 10-8 atm.19 Its effect on the monoxide formation may be neglected even at the furnace axis as firstly, the analyte partial pressure in the furnace, as a rule, is greater than 10-8 atm and, secondly, the number of the metal atom collisons with 0 2 molecules becomes insufficient for equilibrium ( 5 ) to occur. Thus free oxygen p,ressure variations in the sheath gas cannot account for the numerous effects of the furnace material and sheath gas on the sensitivity and vaporisation rates of analytes under typical analytical conditions.Other possibilities should be explored. Carbon Historical Background Our interest in the formation of stable gaseous compounds of metals with carbon as a possible reason for the incomplete atomisation of elements in carbon-containing media began more than 10 years ago. In particular, in a paper24 published in 1976, we suggested that the decrease of the integrated absorbance signal for a large number of elements vaporised from a graphite rod in a hydrogen flame with an addition of 3-8% C2H2 is due to the formation of MC,-type compounds. Later, in 1979, in a report12 dealing with the problem of monocyanide formation in AAS, we stated that: “For a longer period of time we have been living in an ‘oxidising’ world and have got used to the idea that monoxides and hydroxides represent the only obstacles in our way to solving the problem of complete and over-all atomisation.With the advent of high-temperature reducing flames and graphite furnaces a hope began to dawn that free carbon present in atomisers of these kinds would help to solve this problem. However, this did not happen. Having eleminated the previous obstacles, we stumbled upon others. The reducing medium did not remain inert with respect to free atoms and ‘issued,’ in place of monoxides and hydroxides, their carbon analogues, i. e . , dicarbides and monocyanides. In this respect, the reducing world turned out to be a symmetric image of the ‘oxidising’ one.7 y There was sufficiently firm ground for drawing such a conclusion. It may be recalled that systematic studies of gaseous carbides by high-temperature mass spectrometry started more than 25 years ago? During this period the structural and thermodynamic characteristics of gaseous carbides of many elements were thoroughly investigated. As seen from Table 4,26 carbide molecules were found to be formed with elements of all sub-groups of the Periodic Table, with the exception of IIB and VIIIA. Apart from the compounds listed in Table 4, some elements are known to form other carbides, e . g . , MC3. Of particular interest are MC2 molecules. According to the view first put forward by Chupka et al.25 the C, radical in MC2 may be regarded as pseudo-oxygen.Therefore, the MC2 molecules should be considered as typical compounds of the elements as are their gaseous monoxides. In most instances these molecules are the most thermally stable. An exception is the VIIIB sub-group elements, for which the most thermally stable are the MC molecules, and also some element of the IVB sub-group, for which the most stable molecules of the MC4 type. At the same time it has to be admitted that for most elements the fraction of carbide molecules under thermo- dynamic equilibrium is very small, with the exception of elements with energies of dissociation Do (M-C2) 2 660 kJ mol-1 (Ce, Hf, La, Pr, Si, Th and U). In the example of aluminium, for instance, at 2000 K the equilibrium partial Table 4. Known gaseous carbides.26 The symbols indicate that the carbide exists Periodic Carbide sub-group Element* MC MC2 MC4 M2C MzCz Table IA .. . . H IB . . . . Cu IIA . . . . Be Ba IIIA . . . . B A1 Ga IIIB . . . . Sc Y Ln An IVA . . . . Si Ge IVB . . . . Ti Zr Hf VA . . . . P VB . . . . V Nb VIA . . . . S Se VIB . . . . Cr Mo VIIA . . . . Hal VIIB . . . . Tc VIIIB . . . . Ru Rh 0 s Ir Pt + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + * Ln = lanthanides; An = actinides; Hal = halogens.98 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 pressure ratio p(MC2)/p(M) is 5 x 10-6 (Do = 477 kJ mol-1) and for phosphorus it is 5 X 10-4 (Do = 510 kJ mol-1).It is apparently for this reason that workers in the field of pyrometallurgy and carbon chemistry studying the mechan- isms of carbon interaction with metals disregard the possibility of gaseous carbides being involved in these processes. It is for the same reason also that in the late 1970s we stopped halfway in the development of this version. Fortunately, in our studies in the early 1980s of the mechanism by which pg amounts of A1203 are vaporised in graphite furnaces we have succeeded in revealing27 a remark- able effect consisting in the appearance of fast spikes of signals against the main smooth pulse, which originate from the thermal dissociation of A1203 (Fig. 1). In a subsequent investigation of this effect it was shown2s30 to be connected with the reduction of the oxide by carbon, and established that the transfer of carbon from the furnace walls to the oxide involves A12C2 molecules the concentration of which is comparable to that of the A1 atoms.Thus, contrary to the above-mentioned thermodynamic calculations, we came to the same conclusions as when studying the reasons for incomplete vapour atomisation in electrothermal atomisers. Thus, all the data suggest that during atomisation carbon in no way behaves by the rules, and that it is more active toward metals than is so under thermodynamic equilibrium. This implies that it should also vaporise more easily than graphite, which is at thermodynamic equilibrium. These arguments have been put forward by us26 in a preliminary form in 1984. In the section dealing with the reasons for the remarkably high activity of carbon it was stated that: “For the ideal single crystal of graphite, a hexagonal lamellar structure is charac- teristic.Each of the carbon atoms in the plane of a layer is bonded to three immediately neighbouring atoms. The bonds to the atoms of the adjacent layers are considerably weaker, such that they can be neglected. In practice, as a consequence of the numerous defects in the crystal structure in connection with the small dimensions of the grains of polycrystalline graphite, the presence of impurities, and particularly the 1.2 1 1.0 n 0.4 t O.* h 0 J I t 20 40 60 Timeis Fig. 1. Vaporisation of 1 pg of A1 as Al(NO& from the wall of the pyrolytic graphite coated tube heated at a rate of 17 K s-1 in the temperature range 180&2800 K under a stopped-flow of Ar activation of the surface by foreign gases and metal vapours, a certain proportion of the atoms on the surface of the tube will prove to be bonded not to the three neighbouring atoms, but to a smaller number, such that their detachment from the defective layer becomes far more probable.Of course, if activated graphite is heated for a sufficiently long period, the number of defects will decrease as a result of recrystallisation, such that the vapour pressure of the carbon will gradually approach the equilibrium value. Therefore, the greatest excess of the pressure over the equilibrium pressure should be expected either with continuous activation of the surface by foreign gases (for example, O2 present as an impurity in the argon) or at the initial moment of heating activated graphite, even under vacuum conditions.’’ In order to account for the formation MC2-type gaseous carbides in amounts comparable to the free atom concentra- tion it was assumed26 that the C2 partial pressure during sample atomisation in the graphite furnace at. 2200 K exceeds by four to five orders of magnitude the tabulated (thermo- dynamic equilibrium) value. Despite the fact that these estimates had support from some of the observations of other workers, including mass spectrometric measurements, many workers considered this hypothesis with extreme scepticism. Thus, the behaviour of carbon in the course of graphite furnace operation became the key problem in further studies into the mechanism of the atomisation of analytes in GFAAS.Its solution has been found to be an easier task than we had anticipated. Determination of Gaseous Carbon in Graphite Furnaces Experimental We used three different methods to measure the content of the carbon vapour present (Table 5): (i) direct determination31 of & molecule concentration, n(C2), from the absorption and emission of the (0,O) Swan band head; (ii) indirect determi- nation of the concentration of n(C2) in a nitrogen-sheathed furnace from the absorption and emission of the (0,O) CN band-head based on the equilibrium C&) + N2 = 2CN(g) . . . and (iii) indirect determination32 of the concentration of n(C2) by successively measuring the absorption of the A12C2 band maximum and of the A1 auto-ionisation line at 193.6 nm.The C2 concentration was calculated from the equilibrium (7) The aluminium was introduced in the graphite furnace either as Al(N03)3 solution or in the form of metal granules of about 1 mg. In contrast to the preceding rneasurernents,32 we have taken account of the background absorption near the 206 nm band and 193.6 nm line. Although the method of direct determination of the C2 content is preferable over the indirect techniques, its use is restricted to temperatures above 2700 K. Using all three methods permitted the whole temperature range of interest to be covered (1800-3000 K) for both sheath gases, namely, argon and nitrogen. The application of independent methods improved the reliability of the data obtained. Table 5. Methods of gaseous carbon (G) determination Method of Background measurement Band or line/nm correctionhm Absorption .. . . C2516.5 - Absorption . . . . CN388.3 - Absorption . . . . A1193.6 200 Al2G 206 202 Emission . . . . G516.5 516.8 Emission . , . . CN388.3 388.6 Sheath gas Ar - N2 3000 0.04 Ar - N2 2700 0.04 N2 2500 0.07 N2 2400 0.07 Ar 1800 0.07 Ar 1800 0.07 Tmi,/K Slit widthhmJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 99 Equipment. All measurements were carried out on a Perkin-Elmer Model HGA-500 atomiser fitted with a Model 5000 AA spectrometer. The absorption and emission signals were recorded with a 3500 data station and a graphics plotter. Standard graphite tubes both with and without pyrolytic graphite coating and platforms made of anisotropic pyrolytic graphite were used in the experiments. High-purity argon and nitrogen, containing not more than lO-3% 02, served as the sheath gases.Procedure. The signals were measured in the fast tempera- ture ramp mode (ca. 2000 K s-1) under stopped-flow conditions. The absorbance and emission signals were measured at a constant furnace temperature for 60 s. At temperatures above 2600 K the recording time was limited to 40 s because of overheating of the atomiser. In the emission studies the signal measurement at the band maximum was followed by determining the background emission from the heated walls of the furnace, primarily from its ends. Fig. 2 illustrates absorbance and emission traces of the C2 band in Ar at 3000 K, Fig. 3 displays similar records for the CN band in N2 at 2600 K and Fig.4 shows absorbance traces of the A12C2 206-nm band and A1 193.6-rim line, together with the background in Ar at 1800 K. The aluminium was introduced into the furnace before heating in the form of metal granules. Results and discussion The traces in Figs. 2 and 3 reveal clearly that in all instances the signals connected directly or indirectly with the gaseous carbon concentration in the furnace are at a maximum at the onset of constant furnace temperature. Subsequently, these signals decay gradually, tending to constant values. Let us assume that the signals obtained in pyrolytic graphite coated furnaces reach their equilibrium values characteristic of the given temperature within 60 s (or 40 s for T 3 2700 K). The excess concentration of the constituent in question can then be quantitatively evaluated for each moment of time by relating the running values of the signals to their minimum values for the tails.In the emission measurements we took the background into account by preliminarily subtracting the background signal from each running value. The resulting graphs plotted from the data in Figs. 2 and 3 and characterising the variation of the excess concentration of the C2 molecules with heating time are displayed in Figs. 5 and 6. It can be 0.05 1 ( a ) I a, 0.04 0.03 e s 0.02 n a 0 0.01 0 0.1 c- : I .- ‘E 1 UI 516.5 nm I I 1 1 1 I 1 0 1 2 3 4 5 6 7 8 Time/s Fig. 2. function of time in a pyrolytic graphite coated tube at 3000 K in Ar (a) Absorption and ( b ) emission of the C2 516.5-nm band as a readily seen that the absorption and emission measurements yield close results when carried out under the same conditions.The results displayed in Fig. 7 would at first glance appear somewhat unexpected. In contrast to the behaviour of the C2 lines in Figs. 5 and 6, the calculated excess concentration of the C2 molecules at 1800 K in the presence of A1 vapour continues to increase slightly after the stationary furnace temperature has been reached. The ratio n ( C 2 ) h ~ ( C z ) ~ ~ in this instance was calculated by the expression The numerical coefficient in equation (8) was determined at 2500 K for the ratio n(C2)ln(C2),, derived from Fig. 6. The 0.08 0.07 0.06 2 0.05 n 5 0.04 67 J3 (u m a 0.03 0.02 0.01 0 0.07 .; 0.06 2 0.05 I .= 0.04 e 0.03 0.02 E LU 0.01 0 3 0 .- .- 0 2 4 6 8 10 12 14 16 18 Ti me/s (a) Absorption and ( b ) emission of CN 388.3-nm band as a Fig.3. function of time in a pyrolytic graphite coated tube at 2600 K in N2 0.6 0.5 0.4 0.3 - - - - s 0.2 - + 0.1 - z ’ m I I I (6) 0.5 0.4 0.3 0.2 0.1 - - - - 200 nrn - IY I 0 20 40 60 Ti meis Fig. 4. Absorption of (a) A12C2 206-nm band and (b) Al 193.6-nm auto-ionisation line as a function of time in a pyrolytic graphite coated tube at 1800 K in Ar100 Absorbance ( a ) - Ern iss ion (b) I I c .J JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 2 1 PO 2- 2 f l 1 I 0 1 2 3 4 Tirne/s Fig. 5. Time dependence of the ratio n(CJ/n(C&, calculated from (a) absorption and ( b ) emission data of Fig. 2 equilibrium constant K(A12C2) for reaction (7) and the equilibrium partial pressure p(C2)eq were taken from tables.33 Apart from obtaining the dependence of n/neq as a function of time and temperature of the pyrolytic graphite coated furnace, we measured the radial distribution of the ratio n/neq and also the dependence of n/neq on furnace type (i.e., with and without a pyrolytic graphite coating) and sheath gas used.Bearing in mind the higher sensitivity of the emission compared with the absorption method, the major part of these experiments consisted in emission studies of the CN band. An analysis of the totality of the data obtained brings us to the following conclusions: (i) In the initial stage of furnace heating the concentration of the gaseous carbon (C2 molecules) is much higher than the equilibrium value.As the temperature increases, the maxi- mum value of this excess (n/neq) drops from about 105 at 1800 K to a factor of 2-3 at 3000 K (Fig. 8). (ii) At a constant A1 vapour pressure (ca. 10-3 atm at 1800 K) in the furnace the relative excess of C2 concentration, n/neq, increases slightly in time from about 1 x 104 to 4 x 104 (Fig. 7). (iii) In the absence of metal vapour (a “clean” furnace) the maximum absolute concentration of the C2 molecules remains approximately constant above 2400 K (Fig. 9). (iv) The excess of C, molecules increases rapidly from the axis to the wall of a clean furnace (Fig. 10). As the height of the zone viewed by the monochromator is about 2.5 mm, the true variation of n/neq along the furnace radius is probably much stronger than would follow from Fig.10. (v) The ratio n/neq behaves differently in time for clean furnaces with and without pyrolytic graphite coating at equal temperatures (Fig. 11). Initially the ratio dneq in the pyrolytic graphite coated furnace is greater than that for the uncoated furnace. A few seconds later the pattern changes, namely, the ratio n/neq in the pyrolytic graphite coated furnace is greater than that for the uncoated furnace. The total excess carbon, J(n/neq)dt, measured during the signal recording time (60 s) 700 600 500 400 300 200 100 0 . G o 700 600 500 400 300 200 100 0 0 1 2 3 4 5 Ti rne/s Fig. 6. Time dependence of the ratio n(C$n(C,),, calculated from (a) absorption and (b) emission data of Fig. 3 60 000 50 000 40 000 U c“ 2 30000 20 000 10 000 0 20 40 60 Timeis Fig.7. Time dependence of the ratio n(C,)ln(G& in the presence of A1 metal calculated using equation (8) from data of Fig. 4 for the uncoated furnace is a few times greater than that for the coated furnace. (vi) For the same clean furnace, the value of n/neq obtained in nitrogen is less than that for argon (Fig. 12). Insignificant initially, this difference reaches a factor of 1.5-2 about 2-5 s after the heating is switched on. Causes of the Appearance of Non-equilibrium Gaseous Carbon Let us turn back to the reasons for such a remarkable behaviour of carbon in graphite furnaces and attempt to relate it with other much more natural and universally adopted ideas concerning the properties of graphite. Among these is, in particular, the active chemisorption of oxygen that is revealed at low temperatures.Subsequent heating results in a desorp- tion of oxygen, not in the form of 02, but rather as CO molecules produced in the dissociative chemisorption of O2 and, to a lesser extent, of COz molecules. The desorption ofJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 101 0’ I I I I I I 1800 2000 2200 2400 2600 2800 3000 TemperatureiK Fig. 8. pyrolytic graphite coated tube as a function of temperature Maximum relative concentration of C2 molecules in a I I I I I 1800 2000 2200 2400 2600 2800 3000 TemperaturelK Fig. 9. Maximum absolute concentration of C2 molecules in a pyrolytic graphite coated tube as a function of temperature. The dashed line corresponds to a probable change in concentration these molecules starts at 600-700 K and is complete above 1300-1500 K.The chemisorption of O2 on graphite occurs over a relatively small fraction of surface area, which does not exceed 10%.34 Therefore, it may be assumed that the desorption of CO and C 0 2 involves the detachment from the surface of the crystal lattice of only single atoms (in the form of CO2 molecules) or of pairs of neighbouring atoms (in the form of CO molecules). The detachment of single carbon atoms (or of pairs of neighbour carbon atoms) results in a rupture of three (or five) bonds. If we recall that the mean number of bonds per carbon atom is initially (for a thermodynamically perfect lattice) 1.5, it becomes clear that the desorption of chemi- sorbed oxygen brings about a decrease of this number.The detachment of C02 molecules creates in the lattice three times, and that of CO molecules twice, the number of “defective” carbon atoms with one rather than 1.5 bonds per atom. As the enthalpy of vaporisation of “defect-free” graphite is 716 kJ mol-1 at 2000 K,33 the difference in the detachment energy (AE) between a normal and defective carbon atom should be 239 kJ mol-1. This entails a difference in the carbon partial pressure of a factor exp(AE/RT), which is 1 x 105 for 2500 K, 1.7 x 106 for 2000 K and 2.1 x 108 for 1500 K. Hence, the presence of defects in the graphite crystal structure should eventually increase the volatility of carbon compared with the perfect structure. These arguments are valid for any type of carbon-containing materials, both for a perfect single crystal, which may be 2.7 ‘ I I I I I I I 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Distance from axis/mm Fig.10. n(CN),, in a pyrolytic graphite coated tube at 2600 K in Nz Radial dependence of the maximum value of n(CN)I 20 U c“ ‘c 10 0 I I Coated I 0 1 2 3 4 5 Timels Fig. 11. used at 2600 K in N2 Time dependence of the ratio n(CN)/n(CN),, on the tube 0.3 I Nitrogen Ba c kg rou nd . Argon l? - 0 1 2 3 4 5 6 7 8 9 1 0 Timels Fig. 12. Emission of C, 516.5-nm band in Ar and N2 at 3000 K considered to approximate the surface of the pyrolytic graphite coating, and for ordinary polycrystalline graphite. The only difference between them, other conditions being equal, is that the degree of surface activation of polycrystalline graphite is higher than that of pyrolytic graphite coated material.The reason for this is, mainly, the greater real surface area of polycrystalline material. When pre-activated material is heated, the first atoms to vaporise are the carbon atoms in the defective structure. As the defect concentration decreases gradually, the gaseous carbon content approaches the equilibrium level.102 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 The effect of the activation of carbon by oxygen on its subsequent behaviour and, in particular, on its vaporisation deserves further theoretical and experimental study. However, even at the present time, some of the results already obtained by us together with available literature data may be considered as lending support to the above reasoning.In particular, in complete agreement with the above, the total absolute amount of carbon released from furnace walls remains practically constant irrespective of the furnace temperature (item iii). The total amount of surface carbon for a standard furnace with an inner and outer surface area of 12.3 cm2 is 4 X atoms, corresponding to a mass of 1 pg. According to the measurements of Sturgeon,35 a single heating of the furnace for 6 s to 2000 K reduces its mass by about 0.2 pg. The major loss of carbon under these conditions is due to the desorption of C02 and CO. When heated to 2650 K, the mass loss is 0.9 pg for a pyrolytic graphite coated furnace (in agreement with the calculation) and 1.5 yg for an uncoated furnace. In the latter instance the area of the activated surface exceeds clearly its geometric size.As the desorption of C02 and CO starts at 600 K,36it should be expected that the activation of graphite above this temperature is not as efficient as it is below 600 K. In order to elucidate the effect of the graphite activation temperature on the subsequent vaporisation of defective carbon, we carried out the following experiments. The same furnace was period- ically heated and cooled (with the power supply set in the automatic mode) while continuously recording the CN band absorbance. The heating conditions were constant (2600 K, 10 s), whereas the cooling temperature was varied from 2300 to 600 K (10 s). When in the cooling stage, nitrogen (5 cm3 s-1) was passed through the furnace, whereas during the heating this nitrogen flow was stopped.As follows from Fig. 13, at 2300 K there is practically no surface activation, at 1300 K the activation efficiency increases and only at 600 K does it reach a maximum. Thus an analysis of the available theoretical and experi- mental data, combined with our own measurements, suggests that the appearance of excess gaseous carbon over the equilibrium content during the heating of graphite tubes originates directly from the preliminary activation of the graphite surface by oxygen. An additional contribution to graphite surface activation may come from metal vapours, particularly at sufficiently high partial pressures. This is supported by the experiments with aluminium granules. Interpretation of Unusual Effects in GFAAS The enhanced content of gaseous carbon in graphite tubes provides favourable conditions for the formation of gaseous carbides in concentrations above the equilibrium level.This is the most substantial conclusion that can be drawn from the experiments. Only now can the hypothesis of the formation of gaseous carbides in amounts comparable to the free metal atom concentration be transferred from the realm of fantasy and fiction to that of solid fact. Only now can it be used as a basis to explain the above-mentioned observations of the effect of furnace material, sample vaporisation techniques and sheath gas on the vaporisation rate and the degree of analyte atomisation and thereby transfer them from the category of “puzzling” phenomena to that of “obvious” facts.Consider some of these effects. (i) As shown in Fig. 11, the gaseous carbon content in the initial stage of heating of the pyrolytic graphite coated furnace is somewhat greater, and a few seconds after the maximum temperature has been reached, smaller than that of the uncoated furnace. This accounts for the difference in the effects of the pyrolytic graphite coating on the sensitivity for elements of different volatility when vaporised from the 0.2 0, C (0 2 2 0.1 a D 0 0 20 40 60 Timeis Fig. 13. Effect of cooling temperature on absorption of CN band at 2600 K furnace wall. For more volatile elements (Al, Sn, Si and Ge) the degree of gaseous carbide dissociation in a pyrolytic graphite coated furnace is lower, and for low-volatility elements (Mo, V and Ti), which take time to vaporise, higher than that for uncoated furnaces.6.7 (ii) The use of the platform delays the vaporisation of the analyte compared with its vaporisation from tlhe wall. At the moment of analyte vaporisation from the platform the content of gaseous carbon that is predominantly released from the wall is substantially lower than that released during the ramp heating.As a result, the degree of dissocation of gaseous carbides increases considerably, approaching unity for most elements.4 This accounts for the noticeable enhancement of sensitivity provided by the platform (Table 6). For Go, Cu, K, Li, Mg, Mn, Na, Ni, Pb, Pd and Rb the same increase in the sensitivity by a factor of 1.4 k 0.1 corresponds to the increase in the vapour density at the optical axis expected to occur if the platform is used.For the other 20 elements this effect is complemented by an enhanced degree of gaseous carbide dissociation. Note the absence of any correlation between the magnitude of the effect and the energy of monoxide dissocia- tion, D,(MO), which is still more evidence for the non-oxygen nature of the phenomenon. (iii) The three- to nine-fold discrepancy between the calculated and experimental values of characteristic mass, mo, for Er, Eu, Si and Ti4 when vaporised from the wall is likewise accounted for by the incomplete dissocation of the gaseous carbides. The dissociation energies37 of these molecules range from 540 ( E u q ) to 661 kJ mol-1 (Sic2). The discrepancy between the calculated and experimental values of mo reaching a factor twelve for Ba and three for Sr4 when evaporated from the wall, is primarily due to the formation of gaseous monocyanide molecules, which are fairly stable for these metals.Although the amount of free nitrogen present in an argon-sheathed furnace does not exceed 0.1% during the vaporisation of these elements, the concentration of CN is noticeable owing to the enhanced concentration of gaseous carbon. Our experiments9 showed the degree of dissociation of BaCN and SrCN under these conditions to be about 0.2 and 0.5, respectively. (iv) Elements forming unstable monocyanides and stable gaseous carbides may have higher sensitivities with nitrogen as the sheath gas, as the concentration of excess carbon in nitrogen due to the formation of CN molecules is somewhat lower than that in argon.Our measurements suggest that Pd and Ge should be among such elements. Even when vaporised from the platform, the difference in their sensitivity is 1.3- to 1.5-fold (Fig. 14). (v) The formation of gaseous carbides during sample atomisation may be accompanied by their partial decomposi- tion on the surface of the sample particles involving the appearance of a carbon film.38 This process becomes particu- larly noticeable if the sample particles become colder than the furnace walls because of intense vaporisation of the material.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 103 Table 6. Effect of platform on sensitivity. Atomisation at maximum power heating in Ar under stopped-flow conditions Do( M0)37/ mo(W * Do(MO)37/ mow> * Element kJ mol-1 T/K mo(P) Element kJ mol-1 T/K mo(P) Ag .. . . 220 2000 3.3 Li . . . . 340 2700 A1 . . . . 500 2700 2.0 Mg . . . . 360 2400 As . . . . 480 2500 1.8 Mn . . . . 410 2500 Au . . . . 220 2100 2.2 Na . . . . 250 2100 Be . . . . 440 2700 5.2 Ni . . . . 360 2700 Bi . . . . 340 2100 3.0 Pb . . . . 370 2200 Cd . . . . 280 1800 4.8 Pd . . . . 280 2700 c o I . . . . 360 2600 1.3 Rb . . . , 250 2100 Cr . . . . 390 2700 1.7 Sb . . . . 380 2600 c s . . . . 290 2200 3.1 Set . . . . 420 2200 c u . . . . 260 2600 1.4 Si . . . . 790 2900 Fe . . . . 405 2600 1.6 Sn . . . . 530 2500 Ga . . . . 380 2600 5.5 Tet . . . . 390 2200 Ge . . . . 650 2700 7.5 T1 . . . . 310 2100 In . . . . 320 2300 2.5 Zn . . . . 270 2000 K . . . . 280 2100 1.4 * mo(W) and mo(P): characteristic masses for analyte vaporised from the wall and from the platform.t 1 pg of Ni was added. 1.3 1.4 1.5 1.4 1.5 1.5 1.4 1.3 2.1 2.4 1.6 2.8 2.1 2.8 2.0 Nitrogen A 0.3 a, C 0.2 e 0 n 6 0.1 0 0 1 2 3 4 Timeis Fig. 14. of the uncoated tube at 2750 K in Ar and N2 Absorption signals for 10 ng of Ge vaporised from the wall The appearance of the carbon film results, in its turn, in a reduced rate of sample vaporisation. For this reason the higher concentration of the gaseous carbides observed in glassy carbon tubes compared with pyrolytic graphite coated furnaces (which is supported by a difference in the characteris- tic masses) is accompanied also by a reduced vaporisation rate of these elements.7 We believe that this could explain also the reduced vaporisation rate of many elements in graphite tubes compared with Ta-lined furnaces ,8 and also the differences between the vaporisation rates of the same elements ob- served9 in graphite furnaces in Ar and N2.In nitrogen the formation of the carbon film occurs less intensely than in argon, other conditions being equal, because of a slightly lower concentration of gaseous carbon and, hence, of gaseous carbides. This effect is illustrated in Fig. 15 for Ba vaporisa- tion, for pyrolysis and atomisation temperatures of 1500 and 2900 K and a 1-s ramp time. (vi) The decisive role of gaseous carbides and, for a number of elements, of monocyanides is in good accord with 0~r12J3 interpretation of the molecular spectra observed in graphite furnaces during the vaporisation of Al, Ba, Sr and Ca as originating from A12C2, AlCN, BaCN, SrCN and CaCN.A similar spectrum to that of A12C2 with a long-wavelength maximum at 255 nm is observed39 for GaC2, InC2 and TlC2, with long-wavelength maxima at 252, 272 and 269 nm, respectively.39 Strange as it might seem, some ~orkers40-~3 0.8 0.7 0.6 0.5 5 0.4 2 0.3 0.2 0.1 0 al m s 0 1 2 3 4 5 6 7 8 9 10 Time/s Fig. 15. Absorption signals for 30 ng of Ba vaporised from the wall of the pyrolytic graphite coated tube in Ar and N2. The signal in N2 was enlarged 3.3-fold still continue to interpret these and some other spectra recorded in graphite furnaces as due to monoxides. Our repeated attempts to obtain any of the well known44 monoxide spectra in the graphite furnace did not meet with success. (vii) BaC2 and SeCz molecules have been observed in noticeable concentrations by Styris45y46 in a mass spectro- metric study of the products of vaporisation of the oxygen- containing Ba and Se salts from the graphite surface under vacuum.The temperatures at which the signals appear (1200 K for BaC245 and 400-500 K for SeC246) agree with the onset of the carbothermal reduction of BaO and Se02 calculated by the scheme in reference 47. General Conclusions Our analysis of the two approaches to the problem of atomisation of elements in graphite furnaces, which assume the formation of either oxygen- or carbon-containing com- pounds in the gas phase, shows that under typical analytical conditions only the second approach is capable of providing a satisfactory explanation of the incomplete atomisation of elements.The consideration of the effect of oxygen has substantiated the validity of the empirical scale for the free oxygen partial pressure, which is supported by all direct and indirect measurements. It is explained why free oxygen does not affect the atomisation of elements under typical con- ditions, and also why its effect on atomisation becomes104 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 noticeable under intense sheath gas flow-rate or when there is a high content of O2 in sheath gas, particularly a t moderate temperatures (below 1500 K). As for the effect of carbon, we have for the first time revealed , experimentally studied and theoretically interpreted the major features observed in the appearance and distribu- tion over the furnace of gaseous carbon in enhanced compared with equilibrium concentrations.This provided a solid basis to substantiate the decisive role of gaseous carbon-containing compounds such as carbides and monocyanides in the atomisation of elements in graphite furnaces, and to explain numerious instances of the effect of furnace material, sample vaporisation techniques and sheath gas on the analytical signal. An interpretation has been found of the positive effect obtained from the use of platforms and pyrolytic graphite coated furnaces as means for suppressing many of these drawbacks and favouring the highest sensitivity. Some other aspects of the atomisation problem have not been touched upon in this paper, in particular, the major mechanisms of sample vaporisation (thermal dissociation48-50 and carbothermal reduction27-30 of oxides) where the above concepts prove to be of paramount importance.Also outside the scope of this paper are the applications of these concepts to the theory of some industrial high-temperature processes involving the carbothermal production of metals from ores,479s1 specifically of iron, and the catalytic graphitisation, oxidation and gasification of carbon-containing materials.38 However, the very list of the processes reflected on a microscale in the phenomena observed in graphite tubes leaves no doubt that GFAAS has presented to the researcher not only a perfect method of analysis but an elegant tool for exploring the world. The world we live in. Note Added on the Proofs The recent scanning electron microscopic studies by Welz et aZ.52 showed the formation of carbon hollow shells (blisters) after atomisation of 20-25 pg of La from the pyrolytic graphite platforms.This fact is in accordance with our explanation of the reduced vaporisation rate of many elements in graphite tubes. The author is grateful to Martin Mojica who carried out most of the experiments on gaseous carbon and to Leonid Polzik for fruitful comments on the manuscript. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. References L’VOV, B. V., Inzh. Fiz. Zh., 1959, 2(2), 44; 1959, 2(11), 56. Slavin, W., and Manning, D. C., Anal. Chem., 1979, 51,261. Slavin, W., and Carnrick, G. R., Spectrochim. Acta, Part B , 1984, 39, 271. L’vov, B. V., Nikolaev, V. G., Norman, E.A., Polzik, L. K., and Mojica, M., Spectrochim. Acta, Part B , 1986,41, 1043. Ottaway, J. M., At. Spectrosc., 1982,3,89. De Loos-Vollebregt, M. T. C., and de Galan, L., Spectrochim. Acta, Part B , 1984, 39, 449. Schlemmer, G., and Welz, B., Fresenius Z. Anal. Chem., 1986, 323,703. L’vov, B. V., and Pelieva, L. A., Can. J . Spectrosc., 1978,23, 1. L’vov, B. V., and Mojica, M., paper presented at the 2nd Regina1 Conference of Analytical Chemistry, Krasnoyarsk, June 16-20, 1986. Hutton, R. C., Ottaway, J. M., Epstein, M. S., and Rains, T. C., Analyst, 1977, 102, 658. Tsunoda, K., Fujiwara, K., and Fuwa, K., Anal. Chem., 1978, 50, 861. L’vov, B. V., in Kirkbright, G. F., Editor, “XXI Colloquium Spectroscopicum Internationale and 8th International Confer- ence on Atomic Spectroscopy, Keynote Lectures,” Heyden, London, 1979, p.152. L’vov, B. V., and Ryabchuk, G. N., Zh. Prikl. Spektrosk., 1980, 33, 1013. 14. 15. 16. 17. 18. 19. 20. 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. Chakrabarti, C. L., Chang, S. B., and Roy, S . E., Spectrochim. Acta, Part B , 1983, 38,447. Chang, S. B., Chakrabarti, C. L., Huston, T. J., and Byrne, J. P., Fresenius 2. Anal. Chem., 1985,322, 567. Byrne, J. P., Chakrabarti, C. L., Chang, S. B., Tan, C. K., and Delgano, A. H., Fresenius Z. Anal. Chem., 2986, 324, 448. Frech, W., Persson, J. A., and Cedergren, A,, Prog. Anal. At. Spectrosc., 1980, 5 , 279. Cedergren, A., Frech, W., and Lundberg, E., Anal. Chem., 1984, 56, 1382.L’vov, B. V., and Ryabchuk, G. N., Spectrochim. Acta, Part B , 1982,37,673. Sturgeon, R. E., Siu, K. W. M., and Berman, S . S . , Spectrochim. Acta, Part B , 1984,39,213. Sturgeon, R. E., Siu, K. W. M., Gardner, G. J., andBerman, S . S., Anal. Chem., 1986,58, 42. Holcombe, J. A., Eklund, R. H., and Smith, J. E., Anal. Chem., 1979,51, 1205. Kikoin, A. K., and Kikoin, I. K., “Molecular Physics” (in Russian), Nauka, Moscow, 1976, p. 137. Katskov, D. A., Kruglikova, L. P., L’vov, B. V., and Polzik, L. K., Zh. Prikl. Spektrosk., 1976,25,918. Chupka, W. A., Berkowitz, J., Giese, C. F., and Inghram, M. G., J. Phys. Chem., 1958, 62, 611. L’vov, B. V., Zh. Anal. Khim., 1984,39, 1953. L’vov, B. V., and Savin, A. S., Zh. Anal. Khim., 1982, 37, 2116. L’vov, B. V., Dokl. Akad. NaukSSSR, 1983,271, 119. L’vov, B. V., and Savin, A. S . , Zh. Anal. Khim., 1983, 38, 1925. L’vov, B. V., and Savin, A. S., Zh. Anal. Khim., 1983, 38, 1933. L’vov, B. V., Novotny, I., and Pelieva, L. A., Zh. Prikl. Spektrosk., 1980, 32,965. L’vov, B. V., andyatsenko, L. F., Zh. Anal. Khim., 1985,40, 626. Gurvich, L. V., Khachkurusov, G. A., Medvedev, V. A., et al., “Thermodynamic Properties of Individual Substances” (in Russian), Nauka, Moscow, 1978-1982. Walker, P. L., Rusinko, F., and Austin, L. G., Adv. Catal., 1959, 11, 133. Sturgeon, R. E., personal communication. Dollimore, J., Freedman, C. M., and Harrison, B. H., Carbon, 1970,8, 587. Krasnov, K. S., Editor, “Molecular Constants for Inorganic Compounds” (in Russian), Khimiya, Leningrad, 1977. L’vov, B. V., Dokl. Akad. Nauk SSSR, 1985,283,1415. L’vov, B. V., Norman, E. A., and Polzik, L. K., Zh. Prikl. Spektrosk., 1987, in the press. Fuwa, K., Haraguchi, H., and l’sunoda, IC., in Fuwa, K., Editor, “Recent Advances in Analytical Spectroscopy, Proceedings of the 9th International Conference on Atomic Spectroscopy,” Pergamon Press, Oxford, 1982, p. 119. Dittrich, K., Vorberg, B., Funk, J., and Beyer, V., Spectro- chim. Acta, Part B , 1984, 39, 349. Dittrich, K., Spivakov, B. Ya., Shkinev, V. M., and Vorob’eva, G. A., Talanta, 1984, 31, 341. Sedykh, E. M., and Belyaev, Yu. I., Prog. Anal. At.. Spectrosc., 1984, 7 , 373. Pearse, R. W. B., and Gaydon, A. G., “The Identification of Molecular Spectra,” Chapman and Hall, London, 1976. Styris, D. L., Anal. Chem., 1984, 56, 1070. Styris, D. L., Fresenius Z . Anal. Chem., 1986,323,710. L’vov, B. V., Zzv. Vuzov. Chern. Metallurgiya, 1986, NO. 1,4. L’vov, B. V., andRybchuk, G. N., Zh. Anal. Khim., 1981,36, 2085. L’vov, B. V., andFernandez, G. J. A., Zh. Anal. Khim., 1984, 39,221. L’vov, B. V., Ryabchuk, G. N., and Fernandez, G. J. A., Zh. Anal. Khim., 1984, 39, 1206. L’vov, B. V., and Yatsenko, L. F., Zzv. Vuzov. Chern. Metallurgiya, 1986, No. 5, 1. Welz, B . , Curtius, A. J., Schlemmer, G., Ortner, H. M., and Birzer, W., Spectrochirn. Acta, Part B, 1986, 41, 1175. Paper J6llOO Received July 14th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200095

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 105 Mathematical Correction of Systematic Temporal Background-correction Errorsfor Graphite Furnace Atomic Absorption Spectrometry* James M. Harnlyt US Department of Agriculture, ARS Beltsville Human Nutrition Research Center, Nutrient Composition Laboratory, Beltsville, MD 20705, USA James A. Holcombe Department of Chemistry, University of Texas at Austin, Austin, TX 78712, USA The time interval separating the sample and reference measurement gives rise to background-correction errors when the background is changing rapidly. Use of one, two, three or four reference points produces temporal error functions proportional to the first, second, third and fourth derivative, respectively. Combining the appropriate error function and the reference absorbance profile, the error associated with each absorbance value can be computed.The resulting error profile can be used as a diagnostic tool or can be subtracted from background-corrected absorbances to provide a temporally corrected absorbance profile. Keywords: Background correction; graphite furnace atomic absorption spectrometry; electrothermal atomisers; atomic absorption spectrometry Correction of graphite furnace analytical signals for non- specific broad-band absorption, more commonly referred to as background absorption, is essential for accurate determina- tions. The graphite furnace achieves excellent characteristic concentrations and detection limits by total vapourisation of the analyte into a small, relatively confined volume during the atomisation step.Unfortunately, the remaining inorganic constituents of the sample also are vapourised. Thus, each atomisation potentially gives rise to two, rapid, transient signals: the analytical signal from the element of interest and the non-specific signal from molecular absorption and/or scattering by the rest of the sample components. The greatest difficulty in measuring the analyte signal occurs when the analyte and background absorption signal are coincident. As the graphite furnace is the method of choice for trace metal determination in almost every conceivable sample matrix, the background signal is frequently much larger than the analytical signal. As illustrated in Fig. 1 for the determina- tion of Pb in a 2% NaCl solution, the analyst is often faced with the classical analytical problem of measuring the differ- ence between two large signals, i .e . , the difference between the total absorbance signal (analyte and background) and the background absorbance signal. This measurement is com- pounded by the fact that both of the signals are changing rapidly. The need for accurate background-correction measure- ments has spurred the development of a number of commer- cial and prototype instruments. The oldest method involves the use of two sources.1 A continuum source is used in addition to the hollow-cathode lamp and the two beams are superimposed. Two more recent methods are the Zeeman splitting2-4 and the Smith - Hieftje techniques.5 A successful prototype system has incorporated a single high intensity continuum source and wavelength modulation (SIMAAC) .6 7 7 All four background-correction methods are conceptually identical. As illustrated in Fig. 2 for a short interval during the atomisation step, each method makes repeated, rapid, sequential measurements of the total and background absor- bance. These measurements are more commonly referred to as the sample (total) absorbance and the reference (back- * Presented at the Third Biennial National Atomic Spectroscopy 7 To whom correspondence should be addressed. Symposium (BNASS), Bristol, UK, 23rd-25th July, 1986. f - \ 2.0 Q) c tu 1.0 .f! 8 2 0 -0.1 0.18 0.36 0.54 0.72 0.90 Time/s Fig. 1. Absorbance profiles for: A, 150 ng mi-1 Pb in H20; B, 150 ng ml-1 Pb in 2% NaCl (total signal); and C, 150 ng ml-1 Pb in 2% NaCl (background, or reference, signal) ground) absorbance signals.For one commercial Zeeman instrument, the repetition frequency in Fig. 2 is 60 Hz with 16.7 ms between the reference measurements and 8.3 ms between the sample and reference measurements (i.e., the sample measurement is spaced evenly between the reference measurements). These frequencies are similar for SIMAAC and some two-source instruments. Other two-source instru- ments maintain a repetition frequency of 60 Hz but shift the sample measurement towards one of the reference measure- ments ( i e . , an uneven spacing of the sample between the reference measurements). Commercial instruments employ- ing the Smith - Hieftje method have a repetition frequency of 12 Hz with 4 ms between the sample and the reference measurements.Temporal errors arise from the non-simultaneous measure- ment of the sample and reference signals. As shown in Fig. 2 reference measurements are made before and after the sample measurement. For accurate background corrections, however, the reference absorbance must be known at the106 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 I s 1 2 3 4 5 5 6 Time 4 Fig. 2. Illustration of repetitive, sequential data acquisition of the sample absorbances (&-,!&) and reference absorbances (RI-&,) as a function of time for a short interval during the atomisation of a sample with analyte and background absorbance present same time as the sample measurement. Ideally, a mathemat- ical function with the same shape as the background-absor- bance function would be fit to the reference measurements and the background absorbance could be predicted at any time.Unfortunately this ideal background absorbance func- tion does not exist. Estimates of the backgromd absorbance at the time of the sample measurement can be made in a number of ways.8.9 Each method of computing the background absorbance, however, will have inherent, systematic errors. The error function for each background correction computation method can be determined and used to evaluate the accuracy of the correction. The simplest approach to background correction is to subtract the nearest (temporally) reference measurement (either reference measurement when there is even spacing). This method, dubbed the asymmetric mode, will provide an estimate which is either too high or too low if the background absorbance is changing. Larger errors occur when the background changes faster and when the time interval between the sample and reference measurements is larger.It has been shown mathematically that the temporal error for the asymmetric mode is proportional to the first derivative of the background absorbance function and the sample - reference time interval.8 Fortunately, it was not necessary to evaluate the background absorbance function to validate the equation. The next logical step is to employ two reference measure- ments which bracket the sample measurement. The bracket- ing mode provides accurate estimates when the background changes linearly but is in error if there is a non-linear change.The temporal error for the bracketing mode is proportional to the second derivative of the background absorbance function and the square of the sample - reference time interval.8 Fitting a quadratic equation to three reference measure- ments and a cubic equation to four reference measurements surrounding the sample measurement provides better esti- mates of the background absorbance when it is changing in a non-linear fashion. The temporal errors for a quadratic - cubic polynomial fit to three and four points are proportional, respectively, to the third and fourth derivative of the background absorbance function and the sample - reference time interval taken to the third and fourth power.9 Again, the error expressions were derived without evaluating the back- ground absorbance function. The temporal errors for the four background correction computation modes were compared using computer models and a simulated background.9 These models have confirmed that the temporal error equations are valid and that the maximum temporal errors are decreased significantly by using more reference points and higher degree polynomials.The degree of improvement (reduced temporal error) using more reference points is dependent on the model and on the rapidity and the maximum absorbance of the background signal. Using a skewed Gaussian peak as a computer model for the background absorbance signal (with a width of 1.8 s and a maximum absorbance of l.O), the maximum temporal errors have a ratio of 3300 : 75 : 7 : 1 for the asymmetric, bracketing, quadratic(3) and quadratic(4) modes, respectively.9 Simulat- ing background with a quartz refractor plate gave a repetitiou: background signal with a sinusoidal shape.The most rapid simulated background absorbance signals (with a width of 0.05 s and a maximum absorbance of 0.3) gave maximum temporal errors with a ratio of 3 1 : 8 : 3 : 1 for the four computation modes.9 In general, for both models, the ratios grew smaller with decreased width of the background absorbance signal. As predicted by the modelling, systematic and significant reduc- tions in the temporal errors were observed experimentally with the use of real samples. It was, however, not possible to measure the error ratios as the temporal errors for the quadratic(4) mode were hidden in the base-line shot noise.This paper considers the use of the previously derived temporal error equations as a means of computing the temporal error associated with each background-corrected absorbance. The appropriate derivative of the background absorbance function is calculated from the reference measure- ments because the true background absorbance function is not known. In this manner, for each computation mode, the linear array of sample and reference measurements (Fig. 2) can be used to construct three absorbance profiles as a function of time: the total absorbance and background absorbance profiles shown in Fig. 1 and a temporal error prcfile. The point-by-point subtraction of the background profile from the total absorbance profile yields the standard background- corrected absorbance profile.The temporal profile can be used either as an overlay on the background-corrected profile, to predict the times of greatest inaccuracy, or it can be subtracted point-by-point from the background-corrected profile to yield a temporally corrected and background- corrected absorbance profile. The use of temporal error profiles as a means of correcting the analyte signal in the presence of a large background absorbance signal is also investigated. The effect of the number of reference measurements used, the different means of computing derivatives, the background frequency and the sample - reference time interval will be considered. The accuracy and precisions of the temporally corrected asymmet- ric, bracketing, quadratic(3) and quadratic(4) modes are compared using computer modelling, simulated background and the determination of Pb in 2% NaC1.Experimental Instrumentation The data presented in this paper were obtained using a prototype wavelength modulated continuum-source spec- trometer (SIMAAC). For these studies SIMAAC was used in the single-element mode. The system has been described in detail previously. 10 An HGA-500 graphite furnace and power supply (Perkin-Elmer Corporation, Ridgefield, CT, USA) were used for all determinations. Data Acquisition The data acquisition program permitted the operator to specify the modulation frequency. Unless otherwise stated the modulation frequency was 56 Hz.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 107 During a single cycle of the quartz refractor plate, 40 intensity measurements were made: ten measurements at the centre of the absorption profile, ten measurements to the right of the absorption profile (no analyte absorption), ten measurements at the centre again and ten measurements to the left of the profile. Individual intensity data were converted into absorbance data using a stored reference intensity. The ten measurements at each location were summed to give a single value. Thus, the resulting absorbance data were equivalent to acquiring a sample and reference absorbance at 56 Hz (twice the refractor plate frequency of 28 Hz). Temporally, the sample measurement was evenly spaced between the preceding and following reference measure- ments.Data Reduction Background-corrected absorbances were computed in five different ways to imitate five modes of background correction: the asymmetric, bracketing and quadratic - cubic fit to three, four and six points. The data were handled as a block of 12 absorbances: six consecutive pairs of sample (Sl-S6) and reference @I-&,) absorbance measurements where S3 was the sample measurement of interest (Fig. 2). For the calculations, the reference (R1-&) and sample (S3) measure- ments were assigned absorbances of A1-& and A,, respec- tively. Upon completion of the calculations, the program selects the next sample measurement and re-defines the preceeding and trailing references, i.e., the references are re-defined as S, = S, + for n = 2-6, and the next consecutive pair of measurements are selected as R1 and S1.The values of Al-A6 and A, are similarly re-defined. The two most common modes of computing background- corrected absorbances are the asymmetric and bracketing mode.8 In the asymmetric mode the preceding or trailing reference absorbance is subtracted from the sample ab- sorbance : In the bracketing mode,* both of the bracketing reference absorbances are used. AASm=As-A3 . . . . . . (1) A B R A K = A s - ( T ) . A3 + A4 . . . . Three other modes were used which fit a quadratic or a quadratic - cubic equation to three (A2-A4), four (A2-A5) or six (A1-A6) reference absorbances.9 In all three instances, a reference absorbance, AREF(n) (n = 3 , 4 or 6), is computed for the time of the sample measurement. The background-correc- ted absorbances for the quadratic fits is: Each of the reference absorbances was computed using the simplified least-squares approach described by Savitsky and Golay11 where 6 .. . (4) n= 1 where C, are the coefficients and N is the normalisation constant listed in Table 1. This approach fits a quadratic equation to three points and a quadratic - cubic equation to four or more points. The coefficients and normalisation constants are the same for a quadratic or cubic p01ynomial.l~ Throughout this paper the quadratic(4) and quadratic(6) modes will be referred to for simplicity and to be consistent with previous papers. The first four background-correction modes have the following error functionsW 15 d 3 A ~ ~ EQUAD(3) = -- At3 . . . . 32 dt3 (7) where ABG is the background absorbance - time function, and At is the time between the sample and reference measurement and 2At is the time between reference measurements.The general situation for uneven spacing of the sample between references will be considered later in this paper. No error function was computed for the quadratic fit to six points. The first two derivatives were computed in two ways. The true first and second derivatives were obtained by differentia- tion of the quadratic equation fit to four points using the simplified least-squares approach. The coefficients and nor- malisation constant are listed in Table 1. Approximations of the first two derivatives were also computed by successive differences of the absorbances or the differences, i.e. , as then ~ A B G = A3-A4 . . * . * * (9) -- dt 2At where the distance between the reference points has been left at 2At to be consistent with equations (5)-(8). Following this pattern, approximations of the second, third and fourth derivatives were computed. The solutions, again, lent them- selves to notation in the simplified least-squares form as shown in Table 2. The derivative centred around the sample point is desired. The second and fourth derivatives are centred to either side. In both situations, the average of two derivatives, centred to each side, were used; e.g. , second derivatives using points A1-A3 and A2-A4 were averaged. Correction of the background-corrected absorbances for temporal errors was accomplished by subtracting the appro- priate transient absorbance error from the computed ab- Table 1.Coefficients for simplified least-squares fit* dldt QUAD QUAD QUAD QUAD Coefficient (3) (4) ( 6 ) (4) c1 . . . . 0 0 -3 0 c2 . . . . -1 -1 7 -3 9 12 -1 C, . . . . 6 c, . . . . 3 9 12 1 -1 7 3 c, . . . . 0 c, . . . . 0 0 -3 0 N . . . . . . 8 16 32 10 * See equation (4). d2ldt2 QUAD 0 1 -1 -1 1 0 2 (4) Table 2. Coefficients for derivatives with finite time interval (2At)* First Coefficient derivative c* . . . . . . 0 c2 . . . . . . 0 c, . . . . . . 1 c, . . . . . . -1 c5 . . . . . . 0 c, . . . . . . 0 N . . . . . . 2At * See equation (4). Second derivative 0 0 1 0 -2 1 1 -2 0 1 0 0 4Atz Third derivative 0 1 -3 3 -1 0 8At3 Fourth derivative 1 0 -4 1 6 -4 -4 6 1 -4 0 1 16At4108 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 sorbance. Thus, the corrected absorbance for the asymmetric mode was equation (1) minus equation (5) or AAsm - EASYM. The temporally corrected absorbances for the other modes were computed in an analogous manner. Background Simulation A repetitive background absorbance signal was simulated by removing the graphite furnace and inserting a quartz refractor plate, mounted on a galvanometer, into the optical path. This method had been described previously.9 The refractor plate was modulated sinusoidally and caused increased attenuation of the light beam through reflection and refraction as the plate moved further from a position perpendicular to the light path. Sample Atomisation The furnace atomisation parameters are listed in Table 3.Solutions were atomised from the furnace wall to obtain the sharpest possible peaks with the most rapid transient signals. In this manner, the worst possible conditions were sought in terms of rapid background absorbance transients. A 2% NaCl solution was prepared using Specpure NaCl (Jarrell-Ash, Waltham, MA, USA). Solutions of ca. 150 ng ml-1 of Pb in 2% NaCl were prepared by diluting 50 pl of a 5 yg ml-1 Pb solution with 1600 pl of the 2% NaCl solution in the autosampler cups. Results and Discussion Computing the Temporal Error Profile The temporal error functions were computed as described under Experimental. Initially, the first two derivatives were computed in two different ways (see Experimental for the specific equation). The first approach computed all the derivatives by difference, i.e., the first derivatives as the differences between successive reference measurements and the second derivatives as the differences between successive first derivatives.In this manner, the actual data were used but the derivatives were expected to be biased low as a finite time interval (2At) was used. The second approach was to fit a quadratic - cubic equation to 3,4 or 6 reference measurements bracketing the sample and to compute the first and second derivative as the appropriate derivative of the quadratic or quadratic - cubic equation evaluated at the time of the sample measurement. While a true derivative was obtained (At+ 0), it was a concern that the lack of fit of the quadratic - cubic equation might lead to erroneous values.This, however, did not prove to be so. Table 4 compares the first derivatives computed by approxi- mation and as the true derivative of quadratic - cubic equations fit to three, four or six points. As expected, the true derivatives are larger (9.3, 11.4 and 42.8% for three, four and six points, respectively) with greater uncertainty between runs (5.5% by approximation as compared with 8.5,8.6 and 16.4% for the true derivatives). Thus, derivatives of the quadratic(3), quadratic(4) and quadratic(6) modes are more accurate with twice the uncertainty. The much larger values of the quad- ratic(6) mode suggests that the use of more reference values, at a time when the background absorbance was changing rapidly, significantly changed the fit at the centre of the function.The second derivatives, by approximation and by differ- ence, were identical when the same number of points were used: i.e., the four reference points surrounding the sample yield two second derivative approximations (one to either side of the sample) whose average is equal to the second derivative of the quadratic fit to the same four points. In retrospect, it can be shown mathematically that the two are identical as the true second derivative of a quadratic equation is a constant. The second derivatives of the quadratic(3) and quadratic( 6) modes gave values and precisions comparable to the quadratic(4) mode. The third and fourth derivatives were computed only by difference. True third and fourth derivatives would require a quartic - quintic polynomial to be fit to the data (a minimum of five points).The higher order polynomial would allow too many degrees of freedom and erroneous behaviour of the equation between the data points. Use of a quartic - quintic polynomial with a greater number of points would produce an undesired filtering effect. Tracings for the first four background-correction modes and their error profiles for the determination of 2% NaCl (with no analyte present) at the Pb wavelength are shown in Fig. 3. As there was no analyte present, the shape of the tracing is due to shot noise and transient errors. The error profiles were based on derivatives computed from successive differences. The 2% NaCl gave a non-specific background absorbance of 1.04 with a peak half-width of ca. 0.8 s. The positive going temporal error, [Fig. 3(c), (f) and (i)] was 0.074,0.031 and 0.010 for the asymmetric bracketing and quadratic(3) modes, respectively, as printed by the computer.Differences between the com- puter values and the traces in Fig. 3 are a result of the filtering effect of the strip-chart recorder. The error for the quad- ratic(4) mode was not distinguishable. An overlay of the temporal error profile and the analyte absorbance profile can be a very useful diagnostic tool as illustrated by Fig. 3. If the analyte peak and either the positive or negative temporal error peaks are coincident, then the inaccuracy of a peak-height determination is established. It can be seen in Fig. 3 that the main peaks for the asymmetric and bracketing modes were due to transient errors.Because the sum of the error profile is zero, integration of the analyte peak over the entire interval (or over the major excursions of the error function) should, in theory, provide an accurate result. Subtraction of the Transient-error Profile Fig. 3 illustrates the subtraction of the temporal-error profiles [frames ( c ) , 0, ( i ) and (Z)] from the background-corrected absorbance profiles [frames ( b ) , ( e ) , (h) and ( k ) ] to give the Table 3. Atomisation conditions Sample position . . . . . . Drying:Time . . . . . . Temperature . . . . Ashing:Time . . . . . . Temperature . . . . Atomisation:Time . . . . Temperature . . Internalgas flow: Gas . . . . Rate. . . . Furnace wall 30 s ramp, 30 s hold 110°C 5 s ramp, 5 s hold 500 "C 0 s ramp, 2 s hold 2200 "C Argon 20 ml min-1 (during atomisation) Table 4.Comparison of first derivatives computed for atomisation of 2% NaCl (283.3 nm)* Method of computing first derivative Positive maximum f(t) - f(t + At) . . . . 0.1574+0.0086 d -$UAD(3) . . . . . . 0.1721 ? 0.0146 d -QUAD(4) . . . . . . 0.1754 k 0.0151 dt d -QUAD(6) . . . . . . 0.2248 k 0.0370 dt At * Average of ten atomisations.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 109 0.08 0.06 0.04 > v) C Q) 4- .- w .; 0.02 2 0 v) 0 -0.02 -0.04 0.04 0.02 0, C m + o a a n -0.02 -0.04 - - 1 0 1 2 3 4 5 (C-ii c - 0.08 0.06 0.04 0, C 4 0.02 P a n 0 -0.02 -0.04 ( C - i i i ) r - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 (b-i - I 0 1 2 3 4 ( b-ii ~ F 11111 - 1 0 1 2 3 4 5 0.02 0 -0.02 -0.04 L (b-iii) - 0 1 2 3 4 ( d-iii L - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 -1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 Tim e/s Fig.3. Atomisation of 2% NaCl at 283.3 nrn showing the transmitted intensity signal (a) and signals for the ( b ) asymmetric, (c) bracketing, (d) quadratic(3) and ( e ) quadratic(4) mode for (i) the background-corrected signal, (ii) the temporal error and (iii) the background- and temporally corrected signal (i. e . , i - ii) temporally and background-corrected profiles [frames ( d ) , (g), 0') and (m)] for a 2% NaCl solution (at 283.3 nm). Data for the same atomisation are presented in Table 5 . It can be seen that, although the computer read-out to the strip-chart recorder was slowed (ca. by a factor of four), the most extreme excursions were still too rapid for the recorder to track quantitatively.Each of the derivatives shown were computed by difference. Subtraction of the temporal errors had little effect on the integrated signal (as the 2-s integration period included the entire background absorbance peak) but had a significant effect on the peak-height measurement for the asymmetric and bracketing modes. The peak height was relatively unaffected for the quadratic(3) mode. Frames (h), (i) and 0') of Fig. 3 show that a significant temporal error was still occurring but it was less than the peak resulting from the increased noise. As the source intensity was attenuated by ap roximately a factor of ten, the base-line noise increased by ?-. 10 Thus, the positive and negative going peaks observed in frames (g), 0') and (m) were nearly identical and were the110 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 Table 5.Results for atomisation of 2% NaCl Computation Maximum Minimum ff mode Area/A s absorbance absorbance absorbance ASYM . . . . 0.0018 0.1888 -0.0502 0.0200 BRAK . . . . 0.0038 0.0566 -0.0295 0.0079 QUAD(3) . . 0.0025 0.0256 -0.0314 0.0071 QUAD(4) . . 0.0019 0.0235 -0.0306 0.0067 QUAD(6) . . 0.0002 0.0260 -0.0393 0.0070 1st Derivative . . -0.0020 0.1576 -0.0236 0.0156 2ndDerivative . . 0.0019 0.0569 -0.0095 0.0043 3rdDerivative . . 0.0006 0.0242 -0.0101 0.0022 4thDerivative . . 0.0001 0.0048 -0.0015 0.0005 derivative . . 0.0038 0.0566 -0.0295 0.0079 derivative . . 0.0019 0.0235 -0.0306 0.0067 derivative . . 0.0020 0.0235 -0.0306 0.0067 derivative .. 0.0019 0.0232 -0.0311 0.0067 ASYM - 1st BRAK - 2nd QUAD(3) - 3rd QUAD(4) - 4th Table 6. Results for atomisation of de-ionised water* Computation Maximum Minimum U mode AredA s absorbance absorbance absorbance ASYM . . . . -0.0003 0.0139 -0.0141 0.0049 BRAK . . . . -0.0002 0.0129 -0.0141 0.0047 QUAD(3) . . -0.0002 0.0131 -0.0151 0.0049 QUAD(4) . . -0.0002 0.0130 -0.0149 0.0048 QUAD(6) . . -0.0002 0.0123 -0.0136 0.0045 IstDerivative . . O.OOO0 0.0061 -0.0044 0.0018 2ndDerivative . . 0.0000 0.0011 -0.0008 0.0003 3rdDerivative . . O.oo00 0.0010 -0.0010 0.0004 4thDerivative . . 0.0000 0.0002 -0.0001 0.0001 derivative . . -0.0002 0.0129 -0.0141 0.0047 derivative . . -0.0002 0.0130 -0.0149 0.0048 derivative . . -0.0002 0.0130 -0.0150 0.0048 derivative .. -0.0002 0.0129 0.0148 0.0048 * Average of ten atomisations. ASYM - 1st BRAK - 2nd QUAD(3) - 3rd QUAD(4) - 4th I I l l t l I - 1 0 1 2 3 4 5 result of increased base-line noise. The temporal error for the quadratic(4) mode was negligible and the profile and measured parameters were essentially unchanged. Close inspection of the data in Table 5 shows that the temporally corrected asymmetric mode was identical with the bracketing mode and the temporally corrected bracketing mode was identical with the quadratic(4) mode. In retrospect, these equivalences are not illogical and can be shown mathematically. The results for the temporally corrected quadratic(3) mode were quite similar to the quadratic(4) mode but there is no mathematical equivalence.Instead, the agreement of the temporally corrected results suggests that the temporal errors were eliminated or reduced below a detectable level. The remaining noise characteristics were the result of the inherent shot noise. The atomisation of a de-ionised water blank (Fig. 4 and Table 6) shows that the temporal-error corrections had a negligible effect on the base-line noise characteristics. The area, maximum, minimum and standard deviation were unchanged. This is not too surprising as the temporal correction is simply a different means of computing the reference absorbance. The sample absorbance is unchanged and the reference absorbance, although computed from 2-6 values, is essentially unfiltered. Thus, the temporally correc- Table 7. Noise levels for simulated background Computation mode 1 Hz 2 Hz 5 Hz ASYM .. . . 0.0072 0.0134 0,0326 derivative . . 0.0044 0.0045 0.0064 BRAK . . . . 0.0044 0.0045 0.0064 BRAK - 2nd derivative . . 0.0045 0.0045 0.0045 QUAD(3) . . 0.0045 0.0045 0.0047 QUAD(3) - 3rd derivative . . 0.0045 0.0045 0.0045 QUAD(4) . . 0.0045 0.0045 0.0045 derivative . . 0.0045 0.0045 0.0045 ASYM - 1st QUAD(4) - 4th 10 Hz 0.0502 0.0130 0.0130 0.0050 0.0070 0.0051 0.0050 0.0050 0.01 al c 5 0 fl 2 2 -0.01 -0.02 0.01 8 $ 0 e 2 -0.01 v) -0.02 - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 - 1 0 1 2 3 4 5 -10 1 2 3 4 5 Ti m e/s Fig 4. Atomisation of water blank showing the total transmitted intensity signal (a) and signals for the ( b ) asymmetric, ( c ) bracketing, ( d ) quadratic(3) and ( e ) quadratic(4) mode for (i) the background-corrected signal, (ii) the temporal error and (iii) the background- and temporally corrected signal (ie., i - ii)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARC€ ted noise levels are not statistically different from the background-corrected noise levels.Simulated Background Absorbance Simulating a background absorbance signal with a quartz refractor plate allows the amplitude and frequency of the background to be controlled. Unlike background occurring in the graphite furnace, the simulated background is repetitive, allowing a statistical evaluation. The results in Table 7 show the computed base-line standard deviation for a background absorbance varying sinusoidally from zero to 0.6 at frequen- cies of 1, 2, 5 and 10 Hz.These frequencies correspond to background peak widths of 0.5, 0.25 , 0.1 and 0.05 s-1 and absorbance change rates of 1.7, 3.4, 8.4 and 17.0 s-I, respectively. Simulated background between 1 and 5 Hz covers the background absorbance conditions normally encountered with graphite furnace atomic absorption spec- trometry. Simulated background at 10 Hz is significantly worse than any reported measurements.12 The background-corrected noise levels are similar to those previously reported.9 The temporally corrected values, except for the asymmetric modes, reduce the error to the level expected from the base-line noise (k0.0045) of the system. Only at 10 Hz does an apparent increase in the base-line noise appear in the temporally corrected absorbances as a result of temporal errors.At lower frequencies, the base-line noise is dominated by the shot noise. A more extensive examination of the frequency response of the first four background-correction modes is shown in Fig. 5. The background absorbance, varying sinusoidally from 0 to 0.46, was modulated at frequencies from 0.1 to 90 Hz. It can be seen that all four background-correction modes can eliminate temporal errors below 0.5 Hz, i.e., background absorbance signals greater than 2 s wide using the sine wave model. The limiting noise below 0.5 Hz is shot noise. Background at higher frequencies (1-10 Hz) is more accu- rately eliminated using more reference points in the absor- bance computation with a decrease in noise for an increasing number of points. Above 10 Hz, the noise levels increase with increasing frequency. The decrease in noise above 50 Hz is a result of biasing.The experimental results in Fig. 5 are consistent with theoretical predictions using computer modelling. The fre- quency response shown in Fig. 6 is predicted with a computer simulated sampling frequency of 60 Hz (one reference and one sample measurement acquired in 0.0167 s) and a sinussidal background absorbance signal. The slopes of the curves are 1 , 2, 3 and 4, respectively, for the asymmetric, bracketing, quadratic(3) and quadratic(4) modes of background correc- tion computation. Figs. 5 and 6 show that at low frequencies (4 Hz) background signals can be approximated by a quadratic equation and temporal errors can be corrected. Below 5 Hz, the time interval between four reference measurements (0.05 s) is less than the half-width of the background signal.The quadratic or quadratic - cubic equations fit exactly the experimental data and little smoothing occurs. At higher frequencies (> 10 Hz) , the background absorbance signal half-widths are less than 0.05 s and the background signals become indistinguishable from noise. Consequently, at higher frequencies, the quadratic and quadratic - cubic equations are primarily a smoothing function. As background absorption is a mass fluctuation of V'noise, smoothing will not alter the noise level. By definition, however, the amplitude of llf noise is inversely proportional to the square root of the frequency. Thus, the background-correction modes in Figs. 5 and 6 are most effective at low frequencies where the background fluctuation is most severe.Use of more reference values extends the effectiveness of the background correction to higher frequencies, up to a point. 1987, VOL. 2 111 Q, 0.1 v) 0 W C m .- e g 0.01 v) D m - - W m 0.001 I I 0.1 1 .o 10 100 Background signal frequencyiHz Fig. 5. Effect of the apparent background absorbance frequency of simulated background signals on the base-line absorbance noise for the (A) asymmetric, (B) bracketing, ( C ) quadratic(3) and (D) quadratic(4) modes of background correction. Absorbances com- puted at 60 Hz with 0.00833 s between the sample and reference measurements 1 .o W .- 0 0.1 W C m + a m al C .- - 0.01 m 100 0.001 0.1 1 .o 10 Rzwknrniind cinnal froniionrvlH7 Fig. 6. Computer model predicted effect of the frequency of back round absorbance si nals on the base-line absorbance noise for the ?A) asymmetric, (BY bracketing, (C) quadratic(3) and (D) quadratic(4) modes of background correction. Absorbances com- puted at 60 Hz with 0.00833 s between the sample and reference measurements Lead in 2% Sodium Chloride Solution The determination of 150 ng ml-1 of Pb (283.3 nm) in 2% NaCl solution was examined in detail.Fig. 7 shows the background-corrected and temporally corrected absorbance profiles computed using the asymmetric, bracketing, quad- ratic(3) and quadratic(4) modes. In addition the predicted envelope of the base-line absorbance noise at the 3a level has been traced in each frame and the signal for the same concentration of Pb in de-ionised water is shown in frame (d).The horizontal axis is the number of the absorbance computa- tions starting at the beginning of the atomisation cycle with 56 absorbances computed each second (0.018 s per absorbance). The predicted noise envelope is based on the absorbance noise level measured prior to the start of atomisation and the112 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 ~ ~ Table 8. Absorbance signals for 150 mg ml-1 of Pb in 2% NaCl solution* Background-correction mode Measurement Signal ASYM BRAK QUAD(4) Peak height, A . . . . . . Background corrected 0.320 50.025 0.214 f0.019 0.211 kO.020 Temporal error 0.216 40.025 0.108 f0.055 0.096 kO.063 Peak area/A s (2-sinterval) . . . . . . Backgroundcorrected 0.0236 k 0.0031 0.0249 f 0.0027 0.0206 f 0.0031 Temporal error 0.0025 ? 0.0065 0.0043 5 0.0065 0.0008 k 0.0066 0.0211 k 0.0012 0.0001 k 0.0021 Peak area/A s (optimised interval) .. . . Background corrected - - Temporal error - - * Results of ten atomisations. 8 'p Y -0.2 - -0.2 - 0 0.18 0.36 0.54 0.72 u 0 0.18 0.36 0.54 0.72 Ti me/s Fig. 7. Atomisation of 150 ng ml-l Pb in 2% NaCl solution showing the absorbance - time profiles for the (a) asymmetric, ( b ) bracketing, (c) quadratic(3) and (d) yadratic(4) modes; (1) background-correc- ted, (2) temporal error, ( ) temporally corrected absorbance profiles and (4) 30 absorbance noise envelope and ( 5 ) trace of 150 ng ml-1 Pb in de-ionised water [(d) only] assumption that the absorbance noise will change inversely with the square root of the source intensity (the shot noise limited situation).The solution of Pb in NaCl was atomised from older pyrolytically coated graphite tubes (not the higher density Ringsdorff tubes) using the maximum power mode. Thus, conditions were used which maximised the rate of atomisa- tion. These conditions also maximise the chemical inter- ference of NaCl on Pb. The main concern of this study is to assess the accuracy of the measurement of the free atom absorbance signal and does not attempt to account for errors from chemical interferences which may arise due to the particular sample - matrix combination. The temporal error was significant for the asymmetric mode, considerably less for the bracketing mode and was negligible for the quadratic(3) and quadratic(4) modes.The noise envelope shows that the background peak trailed the analyte peak by ca. 0.090 s. Without the noise envelope, the increased base-line absorbance noise could have been mis- taken for the analytical signal or transient errors. The consistency of the noise signal for all four modes and its position within the noise envelope eliminated any doubt as to its origin. 0.3 0.2 a m : 0.1 e $ P Q 0 -0.1 0 0.18 0.36 0.54 0.72 Time/s Fig. 8. Atomisation of 150 ng ml-1 Pb in 2% NaCl. A, Range of values for ten repeat atomisations temporally corrected using the quadratic(4) mode; and B, 3a noise envelope for the first atomisation of the series A profile composite for ten atomisations of Pb in a 2% NaCl solution is shown in Fig. 8. The shaded area encloses the absorbance limits for ten quadratic(4) profiles while the noise envelope is taken from Fig.7, the first atomisation in the series, Some variation in the envelope was expected between atomisations, consequently, the only significant excursion outside the envelope was the Pb peak with a maximum at number 16 or 17. From an instrumental point of view, the noise peaks were not large enough to interfere with the peak-height determination at this Pb concentration, using a simple peak-picker routine. At lower Pb concentrations, however, the peak-picker routine would begin to select the noise maxima. The results of peak-height and -area measurements for this series of determinations of Pb in NaCl is shown in Table 8. The temporally corrected modes are not shown to eliminate redundancy.It was shown earlier that the bracketing mode is equivalent to the temporally corrected asymmetric mode and the quadratic(4) mode is the temporally corrected bracketing mode. Peak. area was integrated over the whole atomisation cycle (2 s) and over the optimum interval for quadratic(4). For the optimum interval, integration was started when the signal exceeded a threshold of four times the base line absorbance noise level (40). Peak-height determinations had precisions of k8.1, k8.9 and -t9.4% while the peak-area (2 s) determinations had precisions of k13.1, k9.9 and f13.4%. The optimised peak-area (ca. 0.2 s) determination had a precision of 25.7%.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 113 Table 9. Predicted maximum temporal absorbance errors for different sample - reference time intervals (At,) Time interval (At,)* Computation mode 0.00833t 0.00417 0.00208 0.00083 ASYM .. . . 0.136 0.068 0.034 0.014 BRAK . . . . 0.035 0.026 0.015 0.007 QUAD(3) . . 0.020 0.011 0.006 0.002 QUAD(4) . . 0.004 0.003 0.002 0.001 * Atl + At2 = 0.01667 s (60 Hz). t Experimental data from Fig. 6. In this instance, optimisation of the integration limit elimi- nated a major portion of the background signal which had increased base-line noise. If the background peak had overlapped the analyte peak then the 40 threshold for detecting the analyte signal would have been much higher and the precision of the integrated signal would have been poorer. Reducing the Time Interval The temporal-error functions [equations (5)-(8)] were evalu- ated in a general form where the sample measurement was separated from the preceding reference measurement by Atl and from the trailing reference measurement by At2, The time interval between reference measurements was held constant at 2At (Atl + At2 = 2At) to be consistent with previous derivations.The general evaluation gave d2ABG At1 At2 EBRAK = ~- . . . . . . (11) dP 2 d 3 A ~ ~ AtlAt2(2Atl + At2)(3Atl + 2At2) . . (12) 16(At1 + At2) %UAD(3) = - dt3 X d4ABG ‘%UAD(4) = - dt4 192 (At1 + At2)2 (13) In all four equations, the temporal errors approach zero as either Atl or At2 approach zero. A temporal error approach- ing zero, however, is a mathematical artifact. Shortening the time interval can provide a useful reduction in temporal errors, but there is a limit imposed by the instrumentation. The At1 or At2 values can only be zero with multiple detectors and the inherent problems of a second noise source and the balancing of the detector signals.Using a single detector, Atl or At2 can only become very small using fast electronics and a small time constant. The requirement for a reduced time constant produces an inherently less precise signal. Thus, there is a trade-off involved in shortening the time interval. Equations (lo)-( 13) are easily evaluated for all the factors except for the appropriate derivatives of the background absorbance function. As this function will vary between interferents and between atomisations, it is difficult to evaluate. For the specific situation of 2% NaCl (Fig. 7), the first four derivatives have values of 1.63 x 101 s-1, 1.01 x 103 s-2,7.37 x 104 s-3 and 2.83 x 106 s-4.Using these values, equations (lo)-( 13) have been evaluated in Table 9 for time intervals of 0.00833, 0.00417, 0.00208 and 0.00083 s: i.e., the reference measurements were kept 0.01667 s apart (60 Hz) while the time interval between the sample and reference measurements was 112, 114, 118 and 1/20 of the cycle. In every instance, the temporal errors decrease as the sample measurement is moved closer to the reference measurement. The decrease in the error is most significant for the asymmetric mode. With a time interval of 0.00083 s the asymmetric error is less than that for the bracketing and quadratic(3) modes at 0.00833 s. However, if the bracketing, quadratic(3) or quadratic(4) modes of data handling are applied to the data collected with a 0.00083 s time interval, temporal errors are reduced by factors of 2, 7 and 14, respectively, as compared with the asymmetric mode. As the half-width of the background absorbance signal decreases (the apparent background frequency increases) the higher derivatives will increase at a faster rate than the lower derivatives. Consequently, for a background absorbance signal that occurs faste; than the one shown in Fig. 7 (an intensity half-width of 0.8 s, an apparent frequency of 1.25 Hz), all four background-correction modes will become more comparable although the same ranking will still occur. For the simplistic sine wave model of a background absorbance signal and a 0.000833-s interval between the sample and reference measurements, the temporal errors for the asymmetric mode will be comparable to those for the bracketing, quadratic(3) and quadratic(4) modes at 20, 26 and 30 Hz, respectively. For the quadratic(3) and quadratic(4) modes, the limiting factor is the time interval covered by three or four reference measure- ments as compared with the half-width of the background signal. One of us (J. A. H.) would like to acknowledge the partial financial support provided by National Science Foundation Grant ChE 84-09819. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. References Koirtyohann, S. R., and Pickett, E. E., Anal. Chem., 1966,28, 585. Koizumi, H., and Yasuda, K., Anal. Chem., 1975, 47, 1679. Kozumi, H., Yasuda, K., and Katayama, M . , Anal. Chem., 1977, 49, 1106. Fernandez, F. J . , Myers, S. A., and Slavin, W., Anal. Chem., 1980, 52, 741. Smith, S. B., and Hieftje, G. M . , Appl. Spectrosc., 1983, 37, 419. Harnly, J. M., and O’Haver, T. C . Anal. Chem., 1977, 49, 2187. Harnly, J. M., O’Haver, T. C . , Colden, B., and Wolf, W . R., Anal. Chem., 1979, 51, 2007. Harnly, J. M. and Holcombe, J . A., Anal. Chem., 1985, 57, 1983. Holcombe, J. A., and Harnly, J . M . , Anal. Chem., 1986, 58, 2606. Harnly, J. M., O’Haver, T. C., Golden, B. M., and Wolf, W . R., Anal. Chem., 1979, 51, 2007. Savitsky, A., and Golay, M. J. E., Anal. Chem., 1964, 36, 1627. Grobenski, Z., Lehmann, R., Radziuk, B., and Voellkopf, U., At. Spectrosc., 1984, 5 , 87. Paper J61.53 Received July 15th, 1986 Accepted October 30th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200105

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 115 Determination of Phosphorus by Graphite Furnace Atomic Absorption Spectrometry Part 2.* Comparison of Different Modifierst Adilson J. Curtius,* Gerhard Schlemmer and Bernhard WelzO Department of Applied Research, Bodenseewerk Perkin-Elmer & Co. GmbH, 0-7770 Uberlingen, FRG Various modifiers were investigated with respect to the highest applicable pyrolysis temperature and best characteristic mass for phosphorus and also their influence on the tube lifetime and the reproducibility of integrated absorbance values. Palladium alone or with the addition of calcium was found to be the best modifier for the determination of phosphorus by graphite furnace atomic absorption spectrometry. The maximum applicable pyrolysis temperature is 1400 “C.The best sensitivity is obtained for atomisation from a pyrolytic graphite platform in a pyrolytic graphite-coated tube. The characteristic mass is around 5.5 ng (0.0044 A s)-1 with Zeeman-effect background correction. Lanthanum, yttrium and nickel modifiers permit essentially the same pyrolysis temperature and give comparable sensitivity but not short- and long-term stability of integrated absorbance signals. Lanthanum exhibits in addition a pronounced memory effect and attacks graphite when used at high concentration. Mechanisms involved in the various reactions are discussed and experiments with a dual-cavity platform were carried out to distinguish between possible mechanisms. Keywords : Phosphorus determination; graphite furnace atomic absorption spectrometry; platform atomisation; palladium modifier; lanthanum modifier L’vov and Khartsyzovl proposed the use of the non-resonance doublet at 213.5 - 213.6 nm for the determination of phosphorus in graphite furnace atomic absorption spec- trometry (GFAAS).As the atom population at the excited levels from which these lines originate is highly temperature dependent, the sensitivity of phosphorus becomes equally temperature dependent. Another problem in the determina- tion of this element originates from the tendency of phos- phorus to form gaseous compounds of high thermal stability. Persson and Frech2 therefore concluded that reproducible results for phosphorus can be obtained only if the heating rate and the final temperature of the furnace and also the atmosphere inside the graphite tube can be controlled during the course of the determination.Introduction of the sample into an atomiser pre-heated to a high temperature such as that used by L’vov and Khartsyzovl should therefore be among the most suitable systems. Commercially available graphite furnaces, however, are not based on the system of L’vov. Ediger et aZ.3 found that most phosphorus compounds except calcium phosphate gave essen- tially no signal in their Massmann-type graphite furnace,4 unless sufficient lanthanum was added as a modifier. In earlier works we investigated the determination of phosphorus without a modifier and found that graphite surface, pyrolysis temperature and atomisation conditions have a substantial influence, and the sensitivity can vary by more than an order of magnitude.However, even under the most favourable conditions large amounts of phosphorus are lost in some form prior to atomisation so that a determination without a modifier is not of practical analytical importance. Most analysts have therefore used a modifier for the determination of phosphorus and lanthanum, which was first recommended by Ediger,6 is the most widely applied.7-lo By analogy with the determination of arsenic,llJ2 nickel was also * For Part 1 of this series, see reference 5. t Presented at the Third Biennial National Atomic Spectroscopy $ Present address: Departamento de Quimica, Pontificia § To whom correspondence should be addressed. Symposium (BNASS), Bristol, UK, 23rd-25th July, 1986. Universidade Cat6lica do Rio de Janeiro, RIO de Janeiro, Brazil.used as a modifier for phosphorus.13J4 Saeed and Thomas- sen15 listed 21 elements that enhanced phosphorus sensitivity and they found that aluminium, barium, cerium, cadmium,co- balt, caesium, iron, lithium, thorium and zirconium are all more effective than lanthanum or nickel. Their data are, however, based on peak height and therefore do not neces- sarily reflect atomisation efficiency. Others investigated the treatment of graphite tubes with zirconium1618 as an alterna- tive to the addition of a modifier, but Kubota et al. 18 obtained satisfactory results only if they injected zirconium solution before each atomisation cycle in the treated tube. The aim of this work was to find optimum conditions for the determination of phosphorus using GFAAS.A number of modifiers including palladium, which was found to be more universal than the others,19 was investigated with respect to sensitivity obtained for phosphorus with various modifier masses using wall and platform atomisation. Reproducibility and long-term stability of the integrated absorbance values were investigated for some of the modifiers, in addition to the influence of modifiers on the lifetime of graphite tubes and on spectral interferences when a deuterium arc, instead of a Zeeman-eff ect background corrector, is used. Additional experiments were carried out to clarify the reaction mechan- isms involved in the stabilisation of phosphorus by some of the modifiers. Table 1. Temperature programme for the determination of phosphorus.Step No. 4 was omitted in some experiments. The temperature in steps 3 and 5 was vaned to establish the pre-treatment and atomisation curves Step Furnace No. temperature/’C 1 90 2 120 3 1350 4 200 5 2650 6 2700 7 20 Time/s ~ ~ Ramp Hold 1 10 15 10 1 30 10 10 0 5 1 4 1 8 Internal gas flow/ml min- 300 300 300 300 0 300 300116 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 Experimental Instrumentation A Perkin-Elmer Model Zeemad3030 atomic absorption spectrometer with an HGA-600 Zeeman graphite furnace and an electrodeless discharge lamp for phosphorus operated at 8 W was used for most of the experiments. Some of the measurements were carried out on a Perkin-Elmer Model Zeemad5000 atomic absorption spectrometer with an HGA- 500 Zeeman graphite furnace.The wavelength was set at 213.6 nm, and all instrumental parameters were according to the manufacturer's recommendations.20 The temperature pro- gramme typically used for graphite furnace determination, unless stated otherwise, is as given in Table 1. The cooling step to 200 "C before atomisation was used for the thermal pre-treatment and atomisation curves only. Samples were injected automatically using Perkin-Elmer AS-40 and AS-60 autosamplers, except for the dual-cavity platform experiments for which manual sample injection had to be used. All data including the time-resolved signals were plotted on a Perkin-Elmer PR-100 printer or a Hewlett-Packard 7225A graphics plotter. Signal evaluation throughout this work was based exclusively on integrated absorbance values.Graphite Tubes and Platforms An uncoated polycrystalline electrographite tube (EG) (Per- kin-Elmer, Part No. B 0070 699), a pyrolytic graphite coated tube (PC) (Perkin-Elmer, Part No. B 0091 504), a L'vov platform of pyrolytic graphite (PG) (Perkin-Elmer, Part No. B 0109 324) and a dual-cavity platform, custom-made (Rings- dorff-Werke, Bonn, FRG) and described in detail else- where,21 were used. Reagents and Solutions Phosphorus stock solution, 5000 mg 1-1. Prepared from dibasic ammonium phosphate, (NH&HP04, and diluted with 0.2% V/V nitric acid. Lanthanum stock solution, 50000 mg 1-I. Prepared by dissolving lanthanum nitrate hexahydrate in 0.2% V/V nitric acid. Nickel stock solution, 10 000 mg 1-*. Prepared from nickel nitrate by dissolution in 0.2% V/V nitric acid.Palladium stock solution, 10000 mg 1-1. Prepared by dissolving palladium powder (Merck, Darmstadt, FRG, No. 12486) in concentrated nitric acid and diluting with de-ionised water. Calcium stock solution, 50000 mg 1-1. Prepared from calcium nitrate tetrahydrate by dissolution in 0.2% V/V nitric acid. Iron stock solution, 50 000 mg 1-1. Prepared from iron(II1) nitrate by dissolution in 0.2% V/V nitric acid. Thorium stock solution, 10000 mg 1-1. Prepared from thorium nitrate pentahydrate by dissolution in 0.2% V/V nitric acid. Magnesium stock solution, 50000 mg 1-1. Prepared from magnesium nitrate hexahydrate by dissolution in 0.2% V/V nitric acid. Tungsten stock solution, 20 000 mg 1-1. Undiluted Titrisol concentrate (Merck).Yttrium stock solution, 10000 mg 1-1. Prepared from yttrium(II1) nitrate pentahydrate by dissolution in 0.2% V/V nitric acid. All solutions were adjusted to contain a final concentration of 0.2% V/V nitric acid. A lo-@ volume of phosphorus reference solution and 10 p1 of modifier solution were used for all experiments, except for the dual-cavity platform where 5 1.11 of each were used because of the smaller volume capacity and to prevent mixing of the solutions. The phosphorus solution was always injected first and the modifier second. Results and Discussion Pre-treatment - Atomisation Curves and Characteristic Mass When phosphorus is determined without the addition of a modifier, the type of graphite that is used (EG or PC tubes) and the technique (wall or platform atomisation) have a very pronounced influence on phosphorus sensitivity and on the shape of thermal pre-treatment curves.5 EG can apparently stabilise part of the phosphorus to very high temperatures in the form of lamellar or residue compounds.Most of the analyte is, however, lost at low temperatures in the form of gaseous molecules without being atomised, and this is largely independent of the graphite material used. When 10 pg of lanthanum are used as a modifier, the influence of the tube material on the pre-treatment and atomisation curves, as can be seen in Fig. 1, is not very pronounced. It is interesting that there is essentially no difference in the thermal pre-treatment curves for an EG and a PC tube when wall atomisation is used. A very pronounced difference was found in the absence of a modifier under these conditions.5 Higher sensitivities are obtained when the sample is deposited on and volatilised from a platform.This meets expectations, because for its determination at 213.5-213.6 nm phosphorus must not only be atomised but also excited. Platform atomisation with its higher effective gas-phase temperatures offers more favourable conditions for a higher atom population at the excited levels. Considering the thermal pre-treatment curves in Fig. 1, there is a minor but interesting dependence of the phosphorus signal on the tube material. The decrease for pyrolysis temperatures above 1400 "C is steeper in the PC tube, with or without a platform, and no signal for phosphorus is obtained in this tube on atomisation following pyrolysis at temperatures above 1800 "C.In an EG tube, with or without a platform, the signal decrease with increasing pyrolysis temperature is less pronounced, and even after a pyrolysis at 2000 "C there is still enough phosphorus left to produce a significant signal during atomisation. This suggests an additional stabilisation of phosphorus at reactive sites on the graphite tube surface, for example via carbon - phosphorus - lanthanum compounds, as suggested in earlier work.22 0.15 1 v) ? g 0.10 + $ n m m 'Is) c 2 0, c 0.05 0 1000 2000 Tern peratu rePC 3000 Fig. 1. Thermal pre-treatment and atomisation curves for 0.2 pg of phosphorus using 10 pg of lanthanum as the modifier. 1, EG tube, wall atomisation; 2, PC tube, wall atomisation; 3, EG tube, platform atomisation; and 4, PC tube, platform atomisationJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 .i...; 117 IC L 0 1000 2000 Tern pe rat u re/"C 3000 Fig. 2. Thermal pre-treatment and atomisation curves for phospho- rus using platform atomisation in a PC tube. 1 , l .O pg of phosphorus, no modifier; 2,0.2 pg of hosphorus, 10 pg of lanthanum modifier; 3, 0.2 pg of phos horus, 29pg of palladium modifier; and 4, 0.2 pg of phosphorus, 2g pg of nickel modifier Nickel and palladium were among the modifiers investi- gated in more detail. The thermal pre-treatment and atomisa- tion curves are depicted in Fig. 2 and compared with those obtained with lanthanum, and without a modifier. Only the results for platform atomisation in a PC tube are shown because this appears to be the most favourable environment for the determination of phosphorus.Both nickel and palladium stabilise phosphorus up to pyrolysis temperatures of about 1400 "C, the same as those for the lanthanum modifier, and the sensitivities obtained for all three modifiers are comparable. Characteristic mass values for phosphorus were determined with a number of modifiers, including a palladium and calcium nitrates mixed modifier, which was found to be particularly useful for the determination of phosphorus in biological materials.23 The results obtained using wall atomisation in an EG tube and platform atomisation in a PC tube are summarised in Table 2. The best values are obtained for lanthanum, yttrium and the palladium and calcium mixed modifier, all in a PC tube with a PG platform.Some of the modifiers were also investigated using wall atomisation in a PC tube and platform atomisation in an EG tube, but the results were inferior in all instances. The characteristic mass values obtained with a given modifier differ by not more than a factor of two when the various atomisation techniques and tube materials are used. This similarity in characteristic mass values is, however, found for integrated absorbance values only and not to the same extent for peak-height evaluation. For all modifiers investi- gated here the peaks become significantly sharper on going from an EG to a PC tube and from wall to platform atomisation, as is shown in Fig. 3 for the palladium modifier. This suggests that phosphorus compounds are more strongly retained by the EG material.In another set of experiments tubes and platforms coated with a layer of tantalum carbide (TaC) by physical vapour deposition24 (PVD) were investigated. Without the addition of a modifier, the phosphorus signal is higher in these tubes compared with EG or PC tubes if low thermal pre-treatment 0.3 0.4 r--- a 0 c ClJ 0.2 e 2 a 0.1 a 0 0 1 3 0 1 3 0 1 3 Time/s Fig. 3. Atomisation signals for 0.2 pg of phosphorus using 20 yg of palladium modifier. A, EG tube, wall atomisation; B, PC tube, wall atomisation; and C, PC tube, platform atomisation Table 2. Characteristic mass (m,) for phosphorus with a variety of modifiers (modifier mass 20 yg unless stated otherwise) using wall atomisation in an EG tube and platform atomisation in a PC tube mJng (0.0044A s)-1 Modifier Ca .. . . Fe . . . . La . . . . Mg . . . . Ni . . . . Pd . . . . Pd+5pgCa . . Th . . . . w . . . . Y . . . . Wall atomisation, Platform atomisation, EG tube PC tube . . 13 22 10 . . 7 5.5 19 . . 8.5 6.0 . . 10.5 7.0 . . 9.5 5.5 . . 12 25 . . 15 23 . . 8.5 5.5 - . . - . . temperatures are applied. This suggests that low-temperature losses of phosphorus are reduced, as graphite is less available for the reduction of phosphorus compounds to volatile species. With increasing pyrolysis temperatures, however, the phosphorus signal decreases rapidly, and no increase in sensitivity is found above 1000 "C, as with EG.5 This increase was attributed to the formation of lamellar compounds of phosphorus with graphite.We believe that the TaC coating only prevents reactions of phosphorus compounds with carbon and cannot really stabilise the phosphorus. The thermal pre-treatment curve for phosphorus in a TaC-coated tube with a TaC-coated platform is depicted in Fig. 4. Also shown are the decomposition and atomisation curves for the same system with the addition of lanthanum and palladium, respectively, as the modifiers. With lanthanum a slightly higher integrated absorbance is obtained in the TaC-coated tube compared with that measured in a PC tube, but the values start to decrease at pyrolysis temperatures above 1200 "C, compared with 1400 "C in the PC tube. This also suggests a role for carbon in the stabilisation of phosphorus by the lanthanum modifier at high temperatures, which is hindered by the TaC-coated surface. Essentially the same curve as in a PC tube, but with higher integrated absorbance values, is obtained in the TaC-coated tube with the palladium modifier. This shows that TaC coating of graphite tubes does not bring about any real advantage as the coating cannot replace the addition of a modifier.Moreover, the sensitivity started to decrease soon after the first use of the tubes and the small sensitivity advantage of new TaC-coated tubes over PC tubes disappeared within 30-50 determinations. Modifier Mass and Appearance Time All experiments described so far were carried out with the same mass of modifier (20 pg), except those in which118 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 0.20 0.15 v) 9 z +? s : 0.10 m U c !! m C ” - 0.05 3 c. 2 l ‘A fi 3 0 Temperatu rePC Fig. 4. Thermal pre-treatment and atomisation curves for 0.2 pg of phosphorus using a TaC tube and platform. 1, No modifier; 2,20 pg of palladium modifier; and 3, 10 pg of lanthanum modifier 0.25 0.20 v) 9 +? s 2 0.15 a m W 0.10 E m C c - 0.05 a 2 Fig. 5. Influence of lanthanum modifier mass on the integrated absorbance signal for 0.2 pg of phosphorus. 1, EG tube, wall atomisation; 2, PC tube, wall atomisation; 3, EG tube, platform atomisation; and 4, PC tube, platform atomisation lanthanum or a mixed modifier were used. Ediger et aZ.3 reported in their early work that the phosphorus signal depends strongly on the lanthanum concentration. However, their data are based on peak-height measurement, as are those of most others7-9 who typically favoured the use of higher lanthanum concentrations, The influence of lanthanum modifier mass on the phospho- rus signal in PC and EG tubes with and without a platform is depicted in Fig.5. When no modifier was used the highest sensitivity for phosphorus was found for wall atomisation in an EG tube and the lowest sensitivity in a PC tube.5 This tendency persists when small masses of lanthanum modifier (1 pg) are added. The effect of a 100-fold increase in the mass of modifier is clearly dependent on the graphite tube surface. In an EG tube the integrated absorbance signal is little affected by the modifier mass. In a PC tube, however, the sensitivity increases almost linearly with the logarithm of the modifier mass over the range investigated.Similar behaviour is found for the palladium modifier, as depicted in Fig. 6. This shows a pronounced dependence of phosphorus stabilisation and atomisation on the tube material for wall atomisation. Based on previous experience with this element,5 it is not surprising that a higher signal is obtained in an EG tube. It is surprising, however, that with both modifiers the curves for the EG and the PC tube intersect at a modifier mass of about 30 pg. The significantly different slope of the plot of integrated absorbance versus modifier mass for the two tube materials suggests that the stabilisation in a PC tube is due predominantly to direct reaction of analyte and modifier, whereas in an EG tube the much larger number of active carbon sites available on an EG surface are also involved in analyte stabilisation. As phosphorus must not only be atomised but also thermally excited, a higher integrated absorbance does not necessarily imply a higher efficiency of free atom formation.Nevertheless, the possibility that the higher reactivity of the EG surface may also favour the formation of stable gaseous molecules between carbon and phosphorus should be taken into account. This may in effect lead to reduced free atom formation in tubes of this material compared with that in PC tubes in the presence of high modifier masses. The appearance time for phosphorus, i. e., the time after the start of the atomisation stage at which the phosphorus signal exceeds twice the base-line noise, was measured for modifier masses of 1, 10 and 100 pg (Table 3).Although these data are not easy to measure with high accuracy and should therefore not be over-interpreted, they can nevertheless provide useful information. One of the evident trends is that with small modifier masses (1 and 10 pg), while the sensitivity increases the phosphorus signal appears essentially at the same time in both types of tube. With high modifier masses the signal measured in the PC tube is clearly delayed compared with that measured in the EG tube. As a late appearance also means a higher temperature, this shift can lead to a more effective excitation of phosphorus atoms and may well account for the higher integrated absorbance found. Table 3. Appearance time for phosphorus in the presence of increasing masses of lanthanum or palladium measured using wall and platform atomisation in PC and EG tubes Appearance time/s Lanthanum modifier Palladium modifier Wall atomisation Platform atomisation Wall atomisation Platform atomisation Modifier mass/pg EG PC EG PC EG PC EG PC 1 0.34 0.34 0.52 0.37 0.49 0.49 0.63 0.57 10 0.40 0.34 0.49 0.48 0.46 0.55 0.69 0.63 100 0.43 0.52 0.60 0.57 0.49 0.72 0.98 0.86JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 119 O ; Table 4. Appearance time for phosphorus with different masses of several modifiers using platform atomisation in a PC tube. Analyte mass, 0.2 pg of phosphorus L I 1 I l l l l l I 1 1 1 I I I I I 5 10 50 1I Appearance timels Modifier mass/pg Ca Fe La Ni Pd Th W Y 1 0.43 0.55 0.37 0.63 0.57 0.57 10 0.43 0.52 0.48 0.75 0.63 0.86 0.46 0.52 100 1.01 0.75 0.57 0.98 0.86 0.92 0.66 0.55 0.15 v) 9 e $ n 0 C 0.10 m -0 4- I cn 0.05 - Fig.6. Influence of palladium modifier mass on the integrated absorbance signal for 0.2 pg of phosphorus. 1, EG tube, wall atomisation; 2, PC tube, wall atomisation; 3, EG tube, platform atomisation; and 4, PC tube, platform atomisation 0.25 0.20 In 5 u C 0.15 2 n m -0 4- I p 0.10 t. C - 0.05 0 1 5 10 50 100 500 Modifier rnass/yg Fig. 7. Influence of modifier mass on the integrated absorbance signal for 0.2 pg of phosphorus using a PC tube and platform atomisation The results for platform atomisation in a PC tube meet expectations very well as in all instances the signal was higher than that obtained for wall atomisation from the same tube material (see Figs.5 and 6 ) , and it appeared typically about 0.1 s later from the platform. This means that the gain in sensitivity is essentially due to the delay in atomisation, which leads to a higher population of the excited levels from which absorption takes place. The only instance in which a distinct difference between the two modifiers was found was platform atomisation in an EG tube. With the palladium modifier (Fig. 6) the data points for platform atomisation in the two types of tubes essentially coincide. This means that the platform eliminates the effects of these two tube materials on the analyte signal. With the lanthanum modifier (Fig. 5), however, the curve for platform atomisation in an EG tube closely follows the curve for wall atomisation in the same tube.This means that the signal is much more dependent on the surface with which the analyte comes into contact after its volatilisation than on the surface with which it is in contact in the condensed phase. Such phenomena have previously been observed for other ele- ment+; however, this is contrary to the observations made for phosphorus without the addition of a modifier.5 We have no explanation for this behaviour, but it suggests that lanthanum alone, without the participation of reactive carbon sites, is not a very effective modifier for phosphorus. This is also supported by the substantially earlier appearance of the phosphorus signal in the presence of lanthanum compared with that for the palladium modifier under any of the conditions used.The experiments with the TaC-coated tube and platform described in the previous section lead to the same conclusion, that palladium is a more effective modifier for phosphorus than lanthanum. The influence of the mass of several other modifiers on the signal for phosphorus was investigated, but only for platform atomisation in a PC tube because this appeared to be the most promising combination. The results are depicted in Fig. 7 and the corresponding appearance times are given in Table 4. It is obvious from these data that most of the modifiers exhibit a behaviour similar to that of lanthanum or palladium under these conditions. The integrated absorbance of phosphorus increases substantially with increasing modifier mass in all instances except for that of small masses of calcium.Similarly, except for yttrium, increases in modifier mass shift the appearance of the phosphorus signal to later times. This means that, at least under conditions where the interaction of graphite can be expected to be relatively small, compound formation between phosphorus and the modifier is the dominant mechanism of stabilisation, preventing analyte losses prior to the atomisation stage. Further, large masses of modifier retain the analyte longer in the condensed phase so that it is atomised into a hotter environment, which results in increased excitation efficiency. It is also apparent from Fig. 7 that some of the modifiers recommended in the literature do not provide sufficient sensitivity for the determination of phosphorus to be of practical interest.Also investigated, but not shown in Fig. 7 or Table 4, was the previously mentioned mixed modifier composed of 20 pg of palladium and 5 pg of calcium as the nitrates. The integrated absorbance obtained for phosphorus with this120 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 I I I mixed modifier was around 0.17 A s, which is essentially the sum of the values obtained for the individual modifiers, calcium and palladium. This suggests that the mechanisms for the action of these two modifiers are different and additive in their stabilising power for phosphorus. The appearance time for the phosphorus signal was approximately 1.0 s, which is the latest appearance time found in this mass range for any of the modifiers investigated.This mixed modifier apparently pro- vides excellent stabilisation of phosphorus, which leads to a high efficiency of atomisation and excitation and makes it particularly attractive for practical applications.23 I 1 I I Memory Effect of Modifier With high masses of lanthanum, peak broadening and increased tailing are observed, which suggests that phospho- rus is retained in the tube by some process. Closely related to this effect are most probably two other phenomena observed with the lanthanum modifier. The first is that several determinations in the presence of lanthanum are required before the phosphorus signal reaches its final high value. The second is that the enhancing effect of lanthanum on the phosphorus signal persists if after repeated application of lanthanum phosphorus is determined without further addition of a modifier.Both phenomena are shown in Fig. 8; 20 determinations of phosphorus with the addition of 20 pg of lanthanum were followed by an additional 40 determinations of phosphorus without a modifier. The most pronounced “memory” effect was found in EG tubes with the sample pipetted on the wall where the phosphorus signal was altered permanently after a single addition of lanthanum. No memory effect was observed in PC tubes using masses of 100 pg of lanthanum (10 p1 of a 1% solution, three replicate injections). For higher lanthanum masses, however, the memory effect becomes apparent in this tube also, probably because the pyrolytic graphite coating is rapidly destroyed under these conditions, as has also been shown in earlier work.22 The memory effect is also detectable when the sample is deposited on a PG platform, but it depends on the surface of the surrounding tube.Fig. 9 shows the effect for platform sampling in an EG tube. In the first determina- tion after three applications of the lanthanum modifier, the phosphorus signal drops by about 0.045 to 0.075 A s, regard- less of the previously applied modifier mass. In all following determinations without a modifier, the signal remains remark- ably constant, which indicates that the graphite surface has been altered irreversibly. If the sample is deposited on a PG platform in a PC tube, the memory effect is measurable (see Fig. 8), but the phosphorus signal does not reach a stable value as it does in an EG tube, In another experiment (not shown here), the platform was taken out of the pyrolytic graphite coated tube after 20 determina- tions with added lanthanum and inserted in a new PC tube for subsequent determinations of phosphorus without a modifier.The signals for phosphorus were only slightly lower in this experiment, which shows that in this instance the lanthanum is mainly retained on the platform and that there is little influence from the PC tube. From these results it appears likely that lanthanum reacts with phosphorus in at least two different ways. Firstly, there must be a direct reaction that leads to stabilisation without the participation of carbon. This stabilisation is likely to be dominant in PC or TaC-coated tubes. The second, however, must also involve carbon and may well include lamellar compounds of phosphorus and lanthanum. KoreCkova et al.12 proposed the formation of such compounds for arsenic when lanthanum was used as a modifier. This kind of reaction should be dominant in EG tubes. It is interesting that only lanthanum shows such a pronoun- ced memory effect. With nickel or palladium, the full sensitivity for phosphorus is obtained in the first determina- tion with the modifier added, and not only after a set of ten or more determinations as with lanthanum. Even in an EG tube one “blank” determination is sufficient to remove completely the effect of the mqdifier and bring the phosphorus signal back to its original low value. This precludes the formation of stable compounds between nickel or palladium and carbon or any permanent alteration of the graphite surface due to these modifiers.Tube Lifetime and Reproducibility Most workers use fairly high lanthanum concentrations such as 1% mlV to obtain the best sensitivity for phosphorus. When 10-pl aliquots are used, this corresponds to a modifier mass of 100 pg per determination. In previous work22 we observed severe corrosion of tubes and platforms at such concentra- tions, and the analytically useful lifetime was reduced to less than 50 determinations. In addition, after a number of determinations, very rapid spikes appeared occasionally in the background channel, which disturb the measurement and make a reasonable signal evaluation difficult or even impos- sible. Rapid tube corrosion and signal distortion are among the reasons why we decided in this work not to aim for the highest sensitivity but to obtain reasonable tube lifetimes instead.Fig. lO(a), ( 6 ) and (c) shows the results of typical lifetime experiments carried out with three of the modifiers, lan- 0.3 v) 1 L 0, C c - 4 0 4 0.2 s n m Fig. 8. Integrated absorbance signal for 0.2 pg of phosphorus using a PC tube and platform atomisation; 20 determinations with the addition of 20 pg of lanthanum modifier each, no modifier added over the next 40 determinations 0.2 I- With La+- Without La-4 n” 9 1 0.1 0 A 0 X 0 X 0 A 3 A 0 A 0 A l, A 0 3 A 0 0 A Number of determinations Fig. 9. Integrated absorbance signal for 0.2 pg of phosphorus using an EG tube and platform atomisation. Three determinations each with the addition of lanthanum modifier, no modifier added over the next ten determinations.A, 100; B, 25; C , 10; D, 2.5; and E, 1.0 pg of lanthanumJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 121 0.15 0.10 0.05 0 0.15 cn 4 g 0.10 +? $ 0 n 0 0.05 2 Q, c. - 0 0.15 0.10 0.05 B 0 solution 1 + + + +++ + + I + 4- I .!! 10 g + $ i 1 + + + - a I E ~ + ++++ + -/-+- + ;-+:+: 4’ .- + 4- m n - k+ ++ + +-+/-+-+-+ + - -+ - 7+- ++ ++ + + + a 1 1 1 I n omoo /--New solution ! v 5 + + *+-+-- 9 + ++ 4 + ++ + - --+- ‘72 -+:+++-++ - + - ++-+ + + ++ ++ + + 1 I I I 0 100 200 300 400 Number of determinations Fig. 10. Endurance test over 430 determinations of 0.2 pg of phosphorus using a PC tube and platform atomisation. Each data point is the average of ten determinations and the relative standard deviation over ten determinations, respectively.( a ) 20 pg of lan- thanum modifier; ( b ) 20 pg of nickel modifier; and ( c ) 20 pg of palladium modifier thanum, nickel and palladium, respectively. The experiments were typically carried out completely unattended overnight without correcting for any drift in sensitivity or for sample evaporation. In two experiments, those with lanthanum and palladium, a fresh solution was prepared in the morning towards the end of the series to detect evaporation losses, which would lead to an increased analyte concentration in the solution. The same mass of modifier, 20 pg per determination, was used in each of the experiments. From Fig. lO(a) and (6) it is apparent that the results for lanthanum and nickel are fairly similar. Firstly, there is a substantial decrease in sensitivity for phosphorus over the 440 determinations, which is almost compensated by an increase in analyte concentration resulting from evaporation of the solvent.This is evident from the last 100 measurements with lanthanum as modifier, made using a freshly prepared solution. Even if the rapid change in sensitivity over the first 40 determinations is disregarded, the signal drops by one third over about 400 determinations. Secondly, the relative stan- dard deviation measured with either modifier is typically between 4 and 5% for each set of ten determinations, and the precision deteriorates with decreasing sensitivity and increas- ing number of determinations. Neither modifier is very satisfactory with respect to long- and short-term signal stability.The pattern obtained for the palladium modifier, which is depicted in Fig. lO(c), is clearly different. If the data are corrected for solvent evaporation, there is essentially no drift in the integrated absorbance for phosphorus over 440 determi- nations. In addition, the average values for ten determinations each show, after an initial phase of slight drift, a very high consistency. Finally, the relative standard deviation is typic- ally around 2.5%, which is roughly a factor of two better than with the other two modifiers, and there is no sign of a deterioration of the precision towards the end of the experi- ment. Thus, for routine applications, palladium appears to be clearly superior to the other modifiers investigated here.The question regarding the relatively poor performance of the lanthanum and nickel modifiers has yet to be addressed. Corrosion is not very likely to be the reason for drift and imprecision because none of the tubes were actually at the end of their useful lives after the 440 determinations of phos- phorus. Also, at least for the lanthanum modifier, the worst drift occurred during the first 3040 determinations, and the long-term stability improved clearly after some 250 determi- nations. This may indicate that a certain mass of lanthanum must be incorporated in the graphite surface and/or the surface must be modified in some way by the lanthanum before a good performance can be expected. This “etching” effect cannot be achieved sooner, however, by applying higher lanthanum concentrations, as this only leads to rapid corro- sion.22 Soaking of graphite tubes with lanthanum solution was found to improve the performance for several elements, including aluminium, beryllium26 and silicon .27 We therefore also investigated this technique for the determination of phosphorus.An EG platform was immersed in a 5% m/V lanthanum solution for 4 d and then inserted into a PC tube. When phosphorus was determined in this tube - platform system without the addition of a modifier, the initially high signal decreased steadily with the number of determinations made (Fig. 11). If, however, 10 pg of lanthanum were added to the analyte solution in each determination, integrated absor- bance values with remarkably little short- and long-term variations were obtained, Comparably good results (not shown here) were obtained with a PG platform soaked in lanthanum solution.Whereas the initial values for phosphorus without added modifier were much lower, indicating that a smaller amount of lanthanum was retained on this platform, the signal reached the same high values after lanthanum had been added to the analyte solution. This experiment indicates that active centres on the graphite surface have to be covered with lanthanum before optimum performance can be expec- ted. Further, the addition of a modifier in each determination appears to be essential to long-term stability of the integrated absorbance. The other modifiers were not investigated to the same extent, mainly because they were found at an early stage to bring about little improvement in sensitivity for the determi- nation of phosphorus.A phenomenon was observed, however, in the course of the experiments with increasing mass of some of the modifiers. If masses of 100 pg or more of lanthanum, tungsten or yttrium were applied repeatedly, the reproducibility decreased substantially. On inspection of the tube it became apparent that the platform was clearly deformed in all these instances. It has been shown in earlier work22 that this phenomenon is due to excessive intercalation, which causes layer swelling and exfoliation of the lamellar pyrolytic graphite. With tungsten, the platform was, in addition, completely covered with a white layer. It can be assumed that large masses of all three modifiers cause severe122 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 b-0.2 pg P only- I- 0.2 pg P + 10 pg La-4 lC-----3.05% I I I I I I 20 30 40 50 60 0 10 Number of determinations Fig. 11. Integrated absorbance signal for 0.2 pg of phosphorus using a PC tube and atomisation from a platform soaked in 5% lanthanum solution for 4 d; 20 determinations of phosphorus without the addition of a modifier, the following 40 determinations with the addition of 10 pg of lanthanum modifier each 0 0 0 I I I 5 10 15 Number of determinations Fig. 12. Integrated absorbance signal for 0.2 pg of phosphorus in the presence of 25 pg of lanthanum modifier using a dual-cavity platform. First determination, phosphorus without a modifier; second determi- nation, nitric acid blank; determinations 3-15 phosphorus with lanthanum modifier.A, Phosphorus and lanthanum modifier mixed in the same cavity; and B, phosphorus and lanthanum modifier separated in different cavities 0.3 0) C m + :: $ 0 I I I I I 0 1 2 3 4 5 Timeis Fig. 13. Atomic absorption signal for 10 vg of lanthanum using PC tube, platform atomisation and temperature programme for phospho- rus atomisation; wavelength 365.0 nm. Dotted line is the background absorbance corrosion of graphite tubes and platforms, and that even smaller masses of tungsten or yttrium lead to imprecision and drift of signals similar to that observed for lanthanum. Iron cannot be recommended either because large masses of this modifier have been shown previously28 to cause severe corrosion of graphite tubes.Dual-cavity Platform Experiments A platform with two cavities for separated injection and volatilisation of analyte element and interferent or modifier has been used in previous work21 to investigate reactions in the condensed phase and in the gas phase of a graphite tube. If care is taken that the analyte and modifier do not mix in the condensed phase, the dual-cavity platform should help to distinguish between the different mechanisms of stabilisation proposed for lanthanum and other modifiers. The influence of lanthanum on phosphorus when the solutions are pipetted into separate cavities, and when mixed, is depicted in Fig. 12. In the latter instance high sensitivity is obtained in the first determination and the sensitivity increases further over the next 5-10 atomisation cycles, as usual, until a fairly constant value is obtained.If phosphorus and lanthanum are pipetted into separate cavities the inte- grated absorbance increases gradually before it reaches a level which is not much lower than that when the solutions are mixed. Two aspects of this behaviour are particularly remarkable. The first is that the integrated absorbance increases at a fairly constant rate. This precludes a pure gas-phase mechanism such as that proposed by Eklund and Holcombe29 for lead and chromium, and later by L’vov and Ryabchuk30 for phospho- rus. They suggested that the enhancement of the analyte sensitivity in the presence of excess of lanthanum can be attributed to purification of the purge gas in that lanthanum binds free oxygen into stable gaseous compounds, and thus enhances the dissociation of analyte oxides.A much more likely explanation for the gradual increase in phosphorus sensitivity is that lanthanum is volatilised in part in each atomisation cycle and reacts with active carbon centres in the other cavity. In this way the lanthanum modifier alters the surface gradually until a kind of a saturation or coverage is finally obtained. Whereas no lanthanum could be detected in the phosphorus cavity using ED-XRF, an alteration in surface morphology became clearly apparent on inspection with scanning electron microscopy.22 The second aspect is that the very first phosphorus determination is clearly enhanced when lanthanum is in the other cavity.This effect does not contradict the proposed mechanism; it only shows that this is not the only one because an alteration of the surface can only take effect after the atomisation of lanthanum, i.e., in the second determination of phosphorus: A possible explanation for the enhancement of the first phosphorus signal would be a direct gas-phase interaction of a phosphorus species with lanthanum. Another explanation would be a reaction of lanthanum with oxygen species as proposed by L’vov and Ryabchuk30 and an enhancement of phosphorus sensitivity because of the reduced oxygen partialJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 123 pressure. Both mechanisms would require that lanthanum is atomised before the phosphorus species are volatilised.This, however, is not very likely because we know from earlier works that essentially all the analyte is lost at temperatures around 1200 “C under the conditions of this experiment. The first atomisation signal for lanthanum, however, can only be recorded after about 0.8 s, as can be seen in Fig. 13, which corresponds to a temperature near 2000 “C. Any interaction of gaseous lanthanum with phosphorus or oxygen that in turn influences phosphorus atomisation is therefore very unlikely because the events take place at different times. The explanation that is, in our opinion, most likely is that phosphorus which is volatilised in the form of an oxide or as the dimeric molecule, P2, at relatively low temperatures does not leave the graphite tube instantaneously but reacts in part with the lanthanum in the other cavity. It has been shown in previous work21725 that multiple collisions of the volatilised analyte with the graphite tube wall can be expected before it is carried out of the tube.These collisions can be “inactive,” i.e., the analyte bounces back into the absorption volume, or they can lead to all kind of reactions, depending on the analyte and the surface. In the experiments with the dual-cavity platform, the cavity that contains the modifier is part of the surface, and therefore participates in the reaction. With volatilised phos- phorus species that collide with the lanthanum modifier in the condensed phase, the formation of lanthanum phosphate or, more generally, of mixed oxides of lanthanum and phosphorus is a very likely reaction.Such compound formation would keep part of the phosphorus from being removed and stabilise it to the high temperatures that are necessary for its atomisation and excitation. It should be mentioned briefly that the integrated absor- bance for phosphorus in the dual-cavity platform experiments never reached the same value that is obtained when analyte and modifier are mixed. This means that even the combined mechanisms, the “etching” of the phosphorus cavity by lanthanum from the gas phase and the reaction of gaseous phosphorus species with lanthanum in the condensed phase, do not match the stabilising power obtained when the solutions are mixed. Essentially the same result was obtained in the soaking experiment described in the previous section, where only soaking and the addition of a modifier resulted in the highest sensitivity and good reproducibility of the signal.Similarly, only between 70 and 80% of the integrated absorbance that is found for the mixed solutions was obtained in a number of additional experiments. Among these was pipetting a modifier-free phosphorus solution into the cavity that had previously contained the lanthanum modifier. When the dual-cavity platform was removed from the graphite tube and inserted into a new tube, about 70-80% of the orginal signal was obtained for a modifier-free phosphorus solution from each of the two cavities. This indicates that the distribution of lanthanum over the entire platform is fairly uniform. It also suggests that it is mainly the “etching” of the platform that determines the signal for phosphorus.If the phosphorus solution is pipetted on the wall of the tube from which the dual-cavity platform was removed, the signal was clearly enhanced over that for phosphorus without an added modifier. This shows that lanthanum actually alters the surface of the entire tube and not only that of the platform. Of the other modifiers considered in this work, only palladium, calcium and mixtures of the two were investigated in dual-cavity platform experiments. The most significant difference between these modifiers and lanthanum is that there is no influence of the number of determinations on the signal for phosphorus. In other words, the final sensitivity is in any event, with the solutions mixed or with the analyte and the modifier in different cavities, obtained in the first determina- tion.The integrated absorbance for phosphorus mixed with the modifier and for pipetting into different cavities, respec- tively, are given in Table 5. It is apparent that the modifier in the other cavity has an influence on the phosphorus signal, and this influence increases with increasing modifier mass. The appearance temperature for palladium is around 1250 “C, and essentially all of the phosphorus is lost at this temperature in some molecular form if platform atomisation is used and no modifier added.5 The stabilisation therefore cannot be due to a gas-phase interaction but must rather be due to a reaction of volatilised phosphorus species with the palladium modifier in the condensed phase, as already discussed for lanthanum.The type of compound formed between phosphorus and palladium in the gas-condensed phase reaction and when the solutions are mixed is not known. In an additional set of experiments, phosphorus was pipetted into the cavity that had previously contained the palladium modifier, but no detectable signal was obtained. This was true even after ten cycles in each of which 50 pg of palladium were added. Palladium is apparently not retained in the tube to any significant extent, and it does not alter the surface of the platform in a way that measurably affects phosphorus atomisation. Calcium, on the other hand, is transported from one cavity to the other via the gas phase. After ten atomisation cycles with calcium in one cavity, phosphorus, when pipetted into the other cavity, gave a signal that was clearly greater than that for phosphorus without a modifier.This transport and accumulation of calcium, however, appears to have no significant influence on the phosphorus signal in the dual-cavity platform experiments because no increase in signal occurred with the number of determinations. There may be two reasons for this: firstly, Table 5. Dual-cavity platform experiments with the palladium and/or calcium modifiers. All integrated absorbance values are averages of ten determinations Cavity I* Cavity II* Integrated absorbance Relative sensitivityt , % - P + lOPd P 10 Pd - P + 20Pd P 20 Pd - P + 50Pd P 50 Pd - P+5Ca P 5 Ca - P+25Ca P 25 Ca - P - P P + (20 Pd + 5 Ca) (20 Pd + 5 Ca) P + (20Pd + 25Ca) (20 Pd + 25 Ca) 0.119 f 0.003 0.027 k 0.003 0.140 k 0.010 0.037 f 0.006 0.171 k 0.008 0.054 f 0.009 0.050 f 0.004 0.017 ? 0.003 0.048 k 0.003 0.024 -t 0.003 0.145 k 0.011 0.052 ? 0.004 0.148 k 0.007 0.053 k 0.012 23 26 31 34 50 36 36 * P = 0.2 pg of phosphorus.All numbers are pg of modifier. Modifiers in parentheses are mixed before injection. t Integrated absorbance of mixed analyte and modifier is set to 100%.124 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 there is no influence from the calcium mass on the phosphorus sensitivity (compare Fig. 8), and secondly, the stabilisation via reaction of gaseous phosphorus species with calcium in the condensed phase appears to be much more effective. In essence, all the modifiers investigated here react with phosphorus by direct compound formation, which may also include the participation of active carbon sites.There was no indication of any gas-phase interaction of analyte and modifier or of a modifier with oxygen. There is, however, evidence for reactions of gaseous phosphorus species with the modifier in the condensed phase for all modifiers investigated here. Conclusion The exact mechanism of stabilisation, i.e., the compound formed between the analyte and the modifier, is not known for any of the modifiers investigated. Some mechanisms, however, could be excluded as a result of this work. Among these is the “scavenging” mechanism,29JO according to which lanthanum purifies the purge gas binding free oxygen in stable gaseous compounds, thus favouring analyte dissociation.This scavenging may occur but it has no detectable influence on phosphorus atomisation. It has been shown in earlier work22 that, contrary to other reports, even the repeated addition of large masses of lanthanum does not lead to the coating of the surface with a dense layer of lanthanum carbide.26127731 Large masses of lanthanum, however, do cause severe disintegration of graphite and exfoliation of pyrolytic graphite layers. At the same time part of the lanthanum is retained on the graphite surface, probably in the form of residue compounds.22~32 To bring this into context with the findings of this work, it may be stated that one of the reactions of lanthanum is to “cover” the active carbon sites on the graphite surface, and to prevent direct carbon - phosphorus interaction. The predominant stabilising mechanism, however, is compound formation between phosphorus and lanthanum, which is supported by the observation that the appearance of phosphorus absor- bance coincides with that of lanthanum when the two components are mixed.The other modifiers investigated here act in a similar way through compound formation with phosphorus in the con- densed phase. The type of compound formed, however, may be different and was not investigated here. This compound formation prevents pre-atomisation losses of phosphorus in the form of gaseous molecules and, hence, increases the atomisation efficiency. In addition, if platform atomisation is used, higher modifier masses in general cause the appearance of phosphorus atoms to be delayed, which leads to a higher relative population of the excited levels.When the practical aspects of phosphorus determination are considered, lanthanum appears not to be the modifier of choice. Although the sensitivity obtained for phosphorus and the maximum applicable pyrolysis temperature are compara- tively good, there are too many limitations associated with the use of this modifier. Nickel was found to be not much better than lanthanum except that no memory effect was observed. However, a spectral interference was observed during the determination of phosphorus in the presence of larger masses of nickel if a continuum-source background corrector was used. Most of the other modifiers were excluded at an early stage because they did not provide a comparable sensitivity for phosphorus. Coating graphite tubes with a layer of tantalum carbide resulted in increased sensitivity at low temperatures only.High pyrolysis temperatures could be applied only if a modifier was used in addition to the carbide coating, which is considered to be of no advantage. The only modifier that was found to be distinctly different is palladium, alone or mixed with calcium. Palladium gives essentially the same sensitivity as lanthanum or nickel and permits the use of the same high pyrolysis temperature. It does not show any excessive sensitivity drift or memory effect, however, and gives much better short- and long-term stability of the phosphorus signal. An additional advantage is that palladium appears to be a fairly universal modifier applicable to most of the main group elements.19 The authors are grateful to H.M. Ortner, Metallwerke Plansee GmbH, Reutte, Austria, for providing graphite tubes and platforms coated with a layer of tantalum carbide by PVD.24 This research was supported by the Conselho Nac- ional de Desenvolvimento Cientifico e Tecnoldgico (CNPq) , Brazil. References 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. 27. 28. 29. 30. 31. 32. L’vov, B. V., and Khartsyzov, A. D., Zh. Prikl. Spectrosk., 1969, 11, 9. Persson, J. A., and Frech, W., Anal. Chim. Acta, 1980,119,75. Ediger, R. D., Knott, A. R., Peterson, G. E., and Beaty, R. D . , A t . Absorpt. Newsl., 1978, 17, 28. Massmann, H., Spectrochim. Acta, Part B, 1968, 23,215. Curtius, A. J., Schlemmer, G., and Welz, B., 1. Anal. At. Spectrom., 1986, 1, 421. Ediger, R. D . , At. Absorpt. Newsl., 1976, 15, 145. PrCvGt, A,, and Gente-Janniaux, M., At. Absorpt. Newsl., 1978, 17, 1. Slikkerveer, F. J., Braad, A. A., and Hendrikse, P. W., At. Spectrosc., 1980, 1, 30. Langmyhr, F. J., and Dahl, I. M., Anal. Chim. Acta, 1981,131, 303. Welz, B., Vollkopf, U., and Grobenski, Z . , Anal. Chim. Acta, 1982, 136,201. Ediger, R. D., At. Absorpt. Newsl., 1975, 14, 127. KoreCkovB, J., Frech, W., Lundberg, E., Persson, J. A., and Cedergren, A., Anal. Chim. Acta, 1981, 131, 267. Hogen, M. L., Cereal Chem., 1983,60, 403. Russeva, E., Havezov, I., Spivakov, B. Y., and Shkinev, V. M., Frensenius Z . Anal. Chem., 1983,315,499. Saeed, K., and Thomassen, Y., Anal. Chim. Acta, 1981,130, 281. Havezov, I., Russeva, E., and Jordanov, N., Fresenius 2. Anal. Chem., 1979, 296, 125. Lin, S. W., and Julshamn, K., Anal. Chim. Acta, 1984, 158, 199. Kubota, T., Ueda, T., and Okutani, T., Bunseki Kagaku, 1984, 33, 633. Schlemmer, G., and Welz, B., Spectrochim. Acta, Part B, 1986, 41, 1157. “Techniques in Graphite Furnace Atomic Absorption Spectro- photometry,” Part No. 6993-8150, Perkin-Elmer, Ridgefield, CT, 1985. Welz, B., Akman, S . , and Schlemmer, G., Analyst, 1985,110, 459. Welz, B., Curtius, A. J., Schlemmer, G., Ortner, H. M., and Birzer, W., Spectrochim. Acta, Part B, 1986, 41, 1175. Curtius, A. J . , Schlemmer, G., and Welz, B., J. Anal. At. Spectrorn., 1987, in the press, 56/11 1. Ortner, H. M., Krabichler, H., and Wegscheider, W., in Welz, B., Editor, “Fortschritte in der atomspektrometrischen Spuren- analytik,” Band 1, Verlag Chemie, Weinheim, 1984, p. 73. Schlemmer, G., and Welz, B., Fresensius 2. Anal. Chem., 1986, 323, 703. Runnels, J. H., Merryfield, R.. and Fisher, H. B., Anal. Chem., 1975, 47, 1258. Lo, D . B., and Christian, G. D., Can. J . Spectrosc., 1977, 22, 45. Welz, B., Schlemmer, G., and Ortner, H. M., Spectrochim. Acta, Part B, 1986, 41, 567. Eklund, R. H., and Holcombe, J. A., Anal. Chim. Acta, 1979, 108,53. L’vov, B. V., and Ryabchuk, G. N., Spectrochim. Acta, Part B, 1982,37, 673. Anderson, A., At. Absorpt. Newsl., 1976, 15, 71. Hennig, G. R., in Cotton, F. A., Editor, “Progress in Inorganic Chemistry,” Volume 1, Interscience, New York, 1959, p. 125. NOTE-Reference 5 is to Part 1 of this series. Paper J6l71 Received August 8th, 1986 Accepted October 24th, I986

ISSN:0267-9477    

DOI:10.1039/JA9870200115

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 125 Atomisation Characteristics of Lead Determined in Alumina Matrices by Slurry = Electrothermal Atomisation Atomic Absorption Spectrometry* Regina Kawvowska and Kenneth W. Jacksont Department of Chemistry, University of Saskatchewan, Saskatoon, Saskatchewan S7N OWO, Canada The electrothermal atomisation characteristics of lead adsorbed on and occluded in two types of AI2O3 slurry are examined. The work was undertaken to understand better the atomisation of lead from soil slurries, where the metal is mostly adsorbed on the active surface of clay particles. ”High-temperature alumina“ (HTA) is nearly anhydrous AI2O3, and “low-temperature alumina” (LTA) is a mixture of hydrated oxides. The HTA matrix does not seriously affect the atomisation of lead and its quantitative determination is possible using platform atomisation and aqueous standards if an Mg + P043- matrix modifier is added to samples and standards.However, lead is strongly retained by LTA and kinetic studies indicate that atomisation may occur via the breaking of covalent Pb-0 bonds between lead and the particles. Consequently the atomisation appearance time is delayed, and if lead is occluded in LTA systematic errors will occur. However, if lead is adsorbed on LTA, it could still be determined with aqueous calibration standards if a matrix modifier is used. This is similar to the way in which lead is incorporated in soil. Keywords: Atomic absorption spectrometry; lead determination; slurry - electrothermal atomisation; solid samples; environmental analysis The feasibility of analysing solids in the form of slurries by electrothermal atomisation atomic absorption spectrometry (slurry - ETA-AAS) has been demonstrated in several previous papers.1-3 The direct analysis of solids can be very difficult if matched calibration standards are needed to minimise interferences, because reference materials that provide a suitable range of analyte concentrations are rarely available. However, the exact composition of the matrix may be less important than the need to atomise samples and standards into a nearly constant temperature gaseous environ- ment.Slavin and Carnrick4 showed that generally such conditions can be achieved when analysing solutions. Requirements are a pyrolytic graphite L’vov platform, careful temperature control, rapid signal processing to follow accu- rately the absorbance versus time characteristics, the measure- ment of peak areas rather than peak heights, usually a matrix modifier to delay analyte vapourisation and efficient back- ground correction.In an earlier paper3 we demonstrated that these nearly isothermal conditions can also be obtained during slurry - ETA-AAS , using deuterium-arc background correc- tion and measuring absorbance peak areas. The atomisation of 1 ng of lead from 50 pg of A1203 yielded quantitative results when compared with aqueous calibration standards, provided that a non-pyrolytic graphite platform was used. A matrix modifier was not needed. The analysis of soil is a useful application of slurry - ETA-AAS.1.2 Trace metals are either adsorbed on or occluded within the soil particles and it was seen during a study of alumina slurries3 that the atomisation characteristics of lead could be different if the analyte was occluded rather than adsorbed.The alumina used previously was prepared by the high-temperature (900 “C) calcination of an aluminium salt, and at this temperature nearly anhydrous A1203 is obtained5 (the hydration process is too slow to occur during the short time that slurries are stored). This material is called “high- temperature alumina” (HTA). However, when aluminum salts are heated to just below 600 “C, “low-temperature alumina” (LTA) is formed. This is a mixture of incompletely deactivated hydroxides of general formula A1203.nH20 (0 < n < 0.6).5 In this work we have extended the study to * Presented at the Third Biennial National Atomic Spectroscopy t To whom correspondence should be addressed.Symposium (BNASS), Bristol, UK, 23rd-25th July, 1986. LTA, in order to assess whether the more active surface has a greater retention effect on the Pb2+ cation, and hence whether matrix interferences can still be overcome. This is particularly applicable to soil, where the cations of trace metals such as lead are likely to be chemisorbed on the active surfaces of clay particles.6.7 Lead atomisation was also studied in the presence of larger amounts of A1203 than were used previously. Experimental Apparatus For accurate ETA-AAS work it is important for the time constant of the signal processing system to be sufficiently rapid to avoid distortion of the transient signal .3,43 A Perkin-Elmer Model 2280 atomic absorption spectrometer was modified to provide fast signal processing (response time 1/60 s).The output from the pre-amplifier immediately after the photo- multiplier tube was fed to one channel of a 12-bit A/D board (Model DT2801, Data Translation, Marlborough, MA, USA), which was mounted in a 320K IBM-PC microcom- puter. This also enabled rapid deuterium-arc background correction to be performed. An essential feature for this work was the simultaneous monitoring of graphite furnace wall temperatures using an optical pyrometer (Series 1100, Ircon, Niles, IL, USA). The output from the pyrometer was fed to a second channel of the A/D board.Data acquisition was triggered by the ETA power supply read pulse, which signals the start of the “atomisation” heating stage. Hence, back- ground-corrected absorbance, background absorbance, ETA wall temperature and time were correlated accurately. During wall atomisation, the inside tube wall temperature was measured by the usual method of focusing the optical pyrometer through the dosing hole. However, for L’vov platform atomisation this method would lead to errors owing to the lower temperature of the platform compared with the wall. Therefore, it was necessary to view the outside wall of the tube instead, at the risk of some measurement error if the outside wall is cooler than the tube interior. For L’vov platform atomisation, pyrolytic graphite-coated furnace tubes (Perkin-Elmer, Part No.2901822) were used with pyrolytic graphite platforms (Part No. B0121091). For part of this work, non-pyrolytic platforms prepared from cut-down standard graphite tubes (Part No. 2901820) were also used. Pyrolytic graphite coated tubes (Part No. 2901821) were used for wall atomisation.126 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 ~~~ ~ Table 1. Instrumental operating parameters ETA stage Parameter Dry Ramptimeh . . . . 20 Holdtimek . . . . 50 TemperaturePC . . 130 Wavelength . . Spectral band width Lamp current . . Purgegas. . . . * Platform atomisation. t Wall atomisation. Char Cool Atomise Clean 10 10 O , * l t 5 30 20 5 5 600 200 2200,*2300t 2500 . . . . . . 283.3 nm . . . . . . 0.7nm . . . . . . 8mA .. . . . . Ar 0.4 0.2 8 51 m n L 0.4 2 0.2 (a) 0 1 2 3 4 5 Time/s Fig. 1. Wall atomisation of 1 ng of Pb (a) adsorbed on A1203 particles and ( b ) occluded in A1203 particles. Programmed atomisa- tion temperature = 2300°C. A, Aqueous Pb solution; B, HTA; C, LTA; and D, non-specific absorption Reagents Analytical-reagent grade chemicals were used throughout. The distilled, de-ionised water used had no measurable lead concentration. Stock standard lead solution (1000 mg 1-1). This was prepared by dissolving Pb(N03)2 in 1% HN03. Low-temperature alumina (LTA) with lead adsorbed on its particles. Heating A1(N03)3.9H20 in a muffle furnace at 500°C for 5 h produced amorphous LTA.9 Appropriate masses of this material were mixed with the diluted standard lead solution, the solution was evaporated to dryness and the residue was heated at 500°C for 5 h.Lo w-temperature alumina with lead occluded in its particles. Appropriate amounts of solid A1(N03)3.9H20 and lead stock standard solution were mixed, the solation was evaporated to dryness and the residue was heated at 500°C for 5 h. High-temperature alumina with lead adsorbed on its par- ticles. High-purity A1203 (Certified Reagent, Fisher Scien- tific) was ground to reduce the particle size to <10 pm. Appropriate masses were then mixed with the diluted stock standard lead solution, the solution was evaporated to dryness and the residue was heated at 1000°C for 5 h. High-temperature alumina with lead occluded in its particles. Appropriate amounts of solid A1(N03)3.9H20 and lead stock standard solution were mixed, the solution was evaporated to dryness and the residue was heated at 1000°C for 5 h.Matrix modifier solution. This contained 1 g of Po43- + 0.1 g 1-1 of Mg(N03)2 and was prepared from (NH4)2HP04 and Mg(N03)2. Procedure Slurries were prepared by weighing an appropriate amount of the alumina into a 50-ml beaker and adding 20 ml of distilled, de-ionised water. When necessary, this contained a suitable volume of matrix modifier solution. The slurry was homo- genised by stirring magnetically for 5 min, and while stirring aliquots were micropipetted (Eppendorf) into the ETA. Except where stated otherwise, aliquots contained 1.0 ng of lead + 50 pg of alumina. It was shown in the past2 that slurries could be introduced into an ETA with high precision in this way, provided that the particle diameter was <20 pm.The particle diameters of all the prepared alumina samples were ground to <10 pm. For the analysis of solutions, aliquots of the diluted stock standard lead solution contained 1 ng of Pb. When matrix modifier was added to slurries or to the aqueous lead standard solution, each aliquot contained 100 pg of A new graphite tube was used for each series of exper- iments. In each run slurries were analysed, in triplicate, in order of increasing amounts of alumina. Absorbance peak areas were measured, and in all instances they were within the linear region of the calibration graph. Either wall or L’vov platform atomisation was used and the operating parameters are given in Table 1. In the atomise step, the gas stop mode was used, except when studying wall atomisation reaction rates, where an argon inner gas flow-rate of 50-100 ml min-1 was used. Deuterium-arc background correction was used for all measurements.P043- + 10 I.18 of Mg(N03)2. Results and Discussion Wall Atomisation Platform atomisation is required for quantitative determina- tions, but wall atomisation is more useful for studying the atomisation processes in an ETA.3 This is because the more pronounced differences in absorbance peak shape can be related to wall temperatures, and these can be measured much more easily than platform temperatures. 10 The instrumenta- tion used in this work allows the transient absorbance signal to be followed accurately and to be correlated with the wall temperature. (measured with the optical pyrometer).Fig. 1 shows absorbance signals obtained from the pyrolytic graphite wall atomisation of various samples containing 1 ng of lead and 50 pg of A1203. The deuterium lamp non-specific absorbance signal is also shown. For comparative purposes, the absorbance of 1 ng of lead in the absence of A1203 (i. e . , the aqueous stock standard lead solution) is included (peak A). In Fig. l ( a ) , adsorption of lead on HTA (peak B) caused no significant increase in the appearance temperature ( Tapp). For the aqueous lead standard, Tapp is 1050 K, which is in good agreement with the value reported by Sturgeon et al.11 For lead adsorbed on HTA it is 1100 K. The peak profile shows that atomisation of lead is slower when it is adsorbed on HTA, but the effect is not large.Conversely, when lead is adsorbed on LTA [signal C in Fig. l(a)], atomisation is delayed significantly compared with the aqueous standard. Also, two absorbance peaks are seen (Tapp = 1380 and 2100 K, respectively). The processes involved in the release of lead were studied experimentally by measuring activation energies (E,) , using the Arrhenius relationship between In k and 1/T. Two methods that use the initial part of the absorbance peak were applied. In the method of Sturgeon et a1.12 In AT is plotted against 1/T, where AT is the absorbance at temperature T. Smets’ method13 involves the calculation of the rate constant k from the ratio of the absorbance at temperature T to the integral of the remainder of the absorbance peak (this integral was calculated accurately through the signal processing software), and plotting of In k against 1/T.Results from both methods are given in Table 2. It was necessary to ensure that the rate of atom removal was greater than the rate of atom formation.14 This was achieved through the use of an inner gasJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 127 D 0.1 Table 2. Experimental activation energies (E,) - Experimental E, values*/kJ mol-l Sturgeon Temperature et al. 's Smets' Sample range/K method12 method13 AqueousPbstandard . . . . 115&1300 193 k 15 188 f 26 Pb adsorbedonHTA . . . . 1200-1500 137 k 40 197 f 30 Pb adsorbed on LTA , . . . 1470-1600 446 k 32 472 k 30 * Precision limits are k the standard deviation.I A B 0 1700 1900 2100 2300 2500 Atomisation tern peratu re/"C Fig. 2. Effect of programmed atomisation temperature on the integrated absorbance of 1 ng of Pb adsorbed on A1203 (platform atomisation). A, Aqueous Pb solution; B , HTA; and C, LTA. Matrix modifier was added to all samples flow (50-100 ml min-1) during atomisation and a relatively slow wall heating rate (250 "C s-1). Frech et al. 14 showed that both methods should give similar results and, except in one instance, our findings support this (Table 2). For the aqueous lead sample, the E, value is in close agreement with the enthalpy of vapourisation of lead (192 kJ mol-1),15 but the much larger value for lead adsorbed on LTA indicates a different reaction mechanism. It is likely that lead is chemisor- bed on LTA through covalent Pb-0-A1 bonding,6.7 and the bond energies are of the same order of magnitude as those for the formation of covalent Pb-0 bonds.16 It is probable that vapourisation of lead occurs via the breaking of the Pb-0 bond.The mechanism of Pb release from HTA is inconclu- sive. Smets' method gives a value in close agreement with the aqueous lead standard, which would not be surprising if the adsorptive forces are only physical in nature. However, Sturgeon et al. 's method gives an apparently anomalous result. The large relative standard deviation of this value may indicate that several mechanisms are involved in the release of lead. In Fig. l(a) the second absorption peak appearing at 2100 K from the atomisation of lead from LTA may be caused by sintering of the LTA during the heating cycle in the graphite furnace.This could lead to occlusion of some of the lead in particles of LTA. This peak appears at the same temperature as that for non-specific absorption from the evaporation of the alumina matrix (peak D), suggesting that occluded lead is retained until the A1203 lattice is broken. It is not certain that the method of preparing the alumina samples with lead occluded in their particles led to a homogenous distribution of lead throughout the particles. Possibly, during the heating process used to prepare the alumina, some diffusion of lead atoms towards the surface occurred. Nevertheless, signal C in Fig. l(6) shows little of the analyte being released before the alumina matrix vapourises, indicating very strong retention of lead.This is good evidence for the complete occlusion of lead in the LTA, because the signal is very different from the corresponding signal for lead 0.5 1 ~ 1700 1900 21 00 2300 2500 Atomisation temperatu rd"C Fig. 3. Effect of programmed atomisation temperature on the integrated absorbance of 1 ng of Pb occluded in A1203 (platform atomisation). A, Aqueous Pb solution; B, HTA; C, LTA; and D, LTA without matrix modifier (matrix modifer was added to A, B and C ) 0.6 (a' /E 2150 r - - 1920 1530 y ?! - 2150 e 1920 2 0 1 2 3 4 5 Time/s Fig. 4. Platform atomisation of 1 ng of Pb (a) adsorbed on HTA particles and ( b ) adsorbed on LTA particles. Programmed atomisa- tion temperature = 2200°C. A, Aqueous Pb solution with matrix modifier; B, A1203 without matrix modifier; C, A1203 with matrix modifier; D, non-specific absorption in the absence of matrix modifier; and E the temperature curve adsorbed on LTA [signal C in Fig.l(a)]. For lead occluded in HTA [signal B in Fig. l(b)], Tapp is 1300 K, which is only slightly higher than Tapp for lead adsorbed on HTA, i. e., lead is easily removed from the alumina particles at temperatures far below .the temperature of evaporation of alumina. Pos- sibly, much of the lead diffused towards the surface of the HTA particle during the high-temperature (1000 "C) heating during synthesis. The presence of a small second peak in signal B [Fig. l ( b ) ] coinciding with that for lead occluded in LTA indicates that HTA may not be completely anhydrous. In these atomisation studies it is unlikely that LTA is converted into HTA during the charring or atomisation process.The conversion takes much longer than the short treatment times in the ETA. Pyrolytic Graphite Platform Atomisation Throughout this discussion, the "atomisation temperature" is the temperature programmed into the atomisation step of the ETA power supply. This is, of course, generally higher than It 1s clear from the above results that the LTA matrix has a much greater effect on the atomic absorption of lead than the HTA studied previously.3 If wall atomisation was used for TaPP:128 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 0.6 - al C 0.4 * e $ n a 0.2 ' / D 1 2150 0 1 2 3 4 5 Ti rnek Fig. 5. Platform atomisation of 1 ng of Pb occluded in LTA particles.Programmed atomisation temperature = 2200 "C. A, Aqueous Pb solution with matrix modifier; B, LTA without matrix modifier; C, LTA with matrix modifier; and D, the temperature curve determining lead in LTA slurries, calibration standards with a carefully matched matrix would be needed. In order to assess the feasibility of quantitative analytical determinations, plat- form atomisation was investigated. Absorbance signals were obtained for 1 ng of lead in the presence of 1000,500,250,100 and 50 pg of A1203. Both types of alumina (LTA and HTA) were studied. At all atomisation temperatures (1700-2400 "C) the amount of LTA or HTA matrix had no significant effect on absorbance peak areas, Aliquots containing 1000 pg of alumina were obtained by micropipetting 50 pl of a slurry containing 20 mg ml-1 of alumina.The effect of larger amounts of alumina was not studied, because the high viscosity then made micropipetting difficult. For all further experimental work aliquots contained 1 ng of lead in the presence of 50 pg of HTA or LTA. Previous work3 indicated that a matrix modifier such as Mg + Po4+ would be needed for quantitative recovery if a pyrolytic graphite platform was used. The effect of atomisa- tion temperature on integrated absorbance using pyrolytic platform atomisation was studied for the standard aqueous lead solution, lead adsorbed on HTA and lead adsorbed on LTA. Matrix modifier [lo0 pg of P043- + 10 pg of Mg(N03)2] was added to all samples. At atomisation temperatures greater than 1900 "C, all three samples gave similar absorbances (Fig.2), indicating a good recovery of lead compared with the matrix-modified aqueous standard. As expected, all of the absorbances decreased at higher atomisation temperatures owing to greater diffusional losses. Previously,3 at an atom- isation temperature of 2000"C, lead adsorbed on HTA without a matrix modifier gave a low recovery compared with the matrix-modified aqueous standard. Hence, for the quanti- tative determination of lead adsorbed on either LTA or HTA, aqueous standards should be satisfactory provided that matrix modifier is added to all samples and standards, and provided an atomisation temperature above 1900°C is used. The corresponding curves for lead occluded in A1203 (Fig. 3) show poor recoveries at low atomisation temperatures when com- pared with the matrix-modified lead standard.The recoveries decrease in the other lead occluded in HTA with matrix modifier, lead occluded in LTA with matrix modifier and lead occluded in LTA without matrix modifier. Then at tempera- tures 22200 "C, all recoveries are higher than for the aqueous standard. The recovery for lead occluded in HTA is only slightly higher, and hence aqueous standards should be acceptable. As already discussed, however, much of the occluded lead in the HTA sample may have diffused close to the surface of the particle. Earlier,3 when no matrix modifier was used, occluded lead was incompletely recovered from HTA at an atomisation temperature of 2000°C. At 2200"C, lead occluded in LTA without matrix modifier gives an integrated absorbance about 17% higher than the aqueous standard.The importance of the matrix modifier is shown in Fig. 4. At the optimum programmed atomisation temperature of 2200 "C the absorbance peak for lead adsorbed on HTA [Fig. 4(a)] appears before the conditions have become nearly isothermal and the integrated absorbance is only 0.150 A s. With matrix modifier the atomisation has been delayed until conditions are nearly isothermal, and the absorbance is 0.250 A s. This peak now appears at about the same time as for the matrix-modified aqueous Pb standard [Fig. 4(b)], which has a very similar absorbance (0.245 A s). The stronger retention of lead by LTA causes the absorbance to be delayed into the nearly isothermal region even without matrix modifier [Fig.4(b)]. However, the matrix modifier seems to facilitate the release of more lead at a lower temperature (i.e., with matrix modifier, the first lead peak is larger and the second peak is smaller). If the second lead peak is due to its occlusion when the particle sinters in the ETA, then the matrix modifier may be preferentially binding the lead and inhibiting occlusion. The signal for lead adsorbed on LTA with matrix modifier [C in Fig. 4(b)] has an absorbance of 0.260 A s. When lead is occluded in LTA (Fig. 5 ) , the effect of matrix modifier in enhancing the size of the first absorbance peak compared with the second is even more pronounced, but there is the apparently anomalous behaviour whereby a greater absor- bance is obtained without the matrix modifier (Fig.3). When lead is occluded in LTA (Fig. 5), the second peak is the larger, i.e., most of the analyte is retained until the matrix vapourises. Compared with the standard lead solution, there is an approximately 1-s delay before this second peak. Apparently, this delay leads to more efficient atomisation. It has been shown17.18 that the standard graphite tube has a relatively uniform axial temperature distribution during the period of ramp heating, and then a temperature gradient is established whereby the tube becomes cooler towards its ends. Therefore, it is not likely that a higher vapour temperature is leading to enhanced atomisation at longer times. Rather, it may be that more lead is released from the LTA particles as the platform temperature continues to increase.Further work will be necessary to understand this effect better, which could lead to a systematic error in quantitative determinations. The occlusion of lead in HTA was not studied further, owing to the uncertainty about the distribution of lead in the particles. Non-pyrolytic Graphite Platform Atomisation Previously,3 using a non-pyrolytic graphite platform, similar integrated absorbances were obtained for aqueous lead, lead adsorbed on HTA and lead occluded in HTA without any matrix modifier. This was attributed to the more reactive surface of non-pyrolytic compared with pyrolytic graphite, and to heavier than usual platforms, which delayed the atomisation of all samples to the nearly isothermal region. The previous results were obtained from 1 ng of lead in the presence of 50 pg of A1203, and larger amounts of HTA have now been investigated (100,250,500 and 1000 pg). A second lead peak was now seen and high recoveries were obtained compared with the aqueous standards.All amounts of LTA (even 50 pg) showed this effect whether lead was absorbed or occluded. Hence, it was concluded that a non-pyrolytic graphite platform was unsuitable for the quantitative analysis of most alumina slurries, Conclusion When lead is adsorbed on LTA particles, conditions have been established whereby quantitative results should be obtained with aqueous standards provided that a pyrolytic graphite platform is used and an Mg + PO4+ matrix modifier is added to all samples and standards. This provides an insight into the likely behaviour of lead in soil, where the metal is mostly adsorbed on the surface of clay particles through covalent bonds via 0 and OH.6>7 Possibly the soil particlesJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 129 behave in the same way as LTA where the mechanism of lead release is through the breaking of Pb-0 bonds. However, the soil matrix is complicated further by the presence of organic material, and this causes the atomisation temperatures to be lower. 19 In materials where lead is occluded, systematically high results are likely, owing to the very strong retention of lead by an active matrix such as LTA. For such applications other matrix modifiers may be satisfactory, and this will be examined. The investigation of the temperature dependence of absor- bance peak area (Fig.3) shows that occluded lead is not released from LTA at 1700°C in the absence of a matrix modifier. This presents the possibility of speciation, i.e., the mode of incorporation of lead in a refractory matrix might be established from absorbance studies. We are grateful to the Natural Sciences and Engineering Research Council of Canada for financial support of this work. 1. 2. 3. 4. 5. References Jackson, K. W., and Newman, A. P., Analyst, 1983,108,261. Hinds, M. W., Jackson, K. W., and Newman, A. P., Analyst, 1985, 110,947. Karwowska, R., and Jackson, K. W., Spectrochim. Acta, Part B , 1986,41, 947. Slavin, W., and Carnrick, G. R., Spectrochim. Acta, Part B , 1984, 39,271. Linsen, B. G., Editor., “Physical and Chemical Aspects of Adsorbents and Catalysts,” Academic Press, London and New York, 1970, Chapter 4. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. Paul, E. A., and Huang, P. M., in Hutzinger, O., Editor, “The Handbook of Environmental Chemistry,” Volume 1, Part A, Springer, Berlin, 1980, p. 67. Huang, P. M., in Hutzinger, O., Editor, “The Handbook of Environmental Chemistry,” Volume 2, Part A, Springer, Berlin, 1980, p. 47. Lundberg, E., and Frech, W., Anal. Chem., 1981,53, 1437. Funaki, K . , and Shimiza, Y., Kogyo Kagaku Zasshi, 1959,62, 788; Chem. Abstr., 1962, 57, 8158d. Chakrabarti, C. L., Wu, S., Karwowska, R., Rogers, J. T., Haley, L., Bertels, P. C., and Dick, R., Spectrochim. Acta, Part B, 1984,39,415. Sturgeon, R. E., Siu, K. W. M., and Berman, S . S., Spectrochim. Acta, Part B , 1984,39,213. Sturgeon, R. E., Chakrabarti, C. L., and Langford, C. H., Anal. Chem., 1976,48, 1792. Smets, B., Spectrochim. Acta, Part B, 1980, 35,33. Frech, W., Zhou, N. G., and Lundberg, E., Spectrochim. Acta, Part B , 1982, 37, 691. Weast, R. C., Editor, “CRC Handbook of Chemistry and Physics,” Sixty-second Edition, CRC Press, Boca Raton, FL, Szekely, J., Evans, J. W., and Sohn, H. Y., “Gas - Solid Reactions,” Academic Press, New York, 1976, p. 34. Falk, H., Glismann, A., Bergann, L., Minkwitz, G., Schubert, M., and Skole, J., Spectrochim. Acta, Part B , 1985,40, 533. Rademeyer, C. J., Human, H. G. C., and Faure, P. K., Spectrochim. Acta, Part B , 1986,41, 439. Hinds, M. W., Karwowska, R., and Jackson, K. W., poster presented at the XXIV Colloquium Spectroscopicum Inter- nationale, September 15th-20th, 1985, Garmisch-Partenkir- chen, Abstract No. TU E 085. 198 1- 1982, Paper JA6l12 Received July 29th, 1986 Accepted October 30th, I986

ISSN:0267-9477    

DOI:10.1039/JA9870200125

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 131 Direct Atomic Spectrometric Analysis by Slurry Atomisation Part 2.* Elimination of Interferences in the Determination of Arsenic in Whole Coal by Electrothermal Atomisation Atomic Absorption Spectrometryt Les Ebdon and Huw G. M. Parry Department of Environmental Sciences, Plymouth Polytechnic, Drake Circus, Plymouth PL4 8AA, UK The analysis of whole powdered coal without sample dissolution has been investigated. Techniques have been established for the determination of arsenic by slurry injection into an electrothermal atomiser for atomic absorption spectrometry. The effects of the furnace programme, type of background correction, form of atomisation and slurry matrix have been investigated. Calibration was achieved using aqueous standards with sample destruction performed in situ in the furnace.The coal was slurried in nickel nitrate (matrix modifier), magnesium nitrate (ashing aid), nitric acid and ethanol (wetting agent). The use of conventional continuum-source and Smith - Hieftje background correction systems was compared for correction of a broadened aluminium line interference at the most sensitive As line (193.7 nm). Only the latter was effective, because the interfering line is within the monochromator band pass. A carefully designed heating programme to separate temporarily the arsenic and aluminium absorption peaks was also advantageous. The less sensitive 197.2-nm line yielded comparable results with either form of background correction. Complete recoveries of arsenic were achieved when the particle size was reduced to 30 pm.Samples could be injected from either constantly agitated slurries or thixotropically stabilised suspensions. A number of certified reference material coals were analysed (NBS SRM 1632a and 1635 and BCR No. 40). Excellent agreement was obtained between the results from the preferred slurry method and certificate value. Keywords: Arsenic determination; electrothermal atomisation; atomic absorption spectrometry; slurry atomisation; background correction The toxicity of arsenic and its compounds is well documented and therefore there is a need for a rapid method that allows the levels of arsenic in coal to be monitored. Arsenic is present in coal in trace amounts and monitoring is essential, e.g., for the food industry and where there is concern over quality control, catalyst poisoning and environmental pollution.The very high tonnages of coal used and possibilities of concentration during conversion processes mean that even ng g-1 levels may be significant. Coal is a notoriously difficult matrix to bring into solution. Some methods are slow and produce only incomplete dissolu- tion, and other more rapid digestion methods, e.g., with perchloric acid and hydrofluoric acid, are potentially hazar- dous. Additionally, there is a likelihood that volatile elements such as arsenic may be lost on dissolution. Therefore, any technique that avoids this stage would be favourable. X-ray fluorescence spectrometry1 is a rapid technique, but does not possess the required sensitivity for direct trace element analysis of coal.Spark-source mass spectrometry2 and neu- tron activation analysis (NAA), on the other hand, are both capable of determinations of elements at concentrations down to 0.1-1 pg g-1, but the techniques are slow and, in NAA, the equipment is not readily available to most laboratories. Atomic absorption spectrometry is conventionally regarded as a solution-based technique, but recent developments in solid sampling have made atomic absorption using elec- trothermal atomisation (ETA-AAS) a promising approach, especially as the necessary instrumentation is available in many laboratories. Owing to its volatility, arsenic is frequently determined via hydride generation,3 but transition metals cause some inter- * For Part 1 of this series, see J .Anal. At. Specfrom., 1987, 2, 39. t Presented at the Third Biennial National Atomic Spectroscopy Symposium (BNASS), Bristol, UK, 23rd-25th July 1986. ferences in this approach, which is in any event limited to solution samples.4 Therefore, methods whereby coal can be introduced directly into a furnace atomiser, in a slurry form, without the need for lengthy ashing or digestion stages, appear preferable. Such a slurry method has been described by Ebdon and Pearce,s who used nickel nitrate (as matrix modifier), magnesium nitrate (as ashing aid), nitric acid (to provide an acid environment) and ethanol (wetting agent) to prepare the slurries. Positive errors were noticed at the more sensitive analytical resonance arsenic line, 193.7 nm, however, and this was later attributed to a broadened aluminium-line spectral interference,6 first reported by Riley.’ This type of interfer- ence is referred to as structured background. Such interfer- ences cannot be corrected by conventional continuum-source background correction systems, which by definition assume that the background is unstructured across the band pass of the spectrometer.When the variation in background is caused by the fine structure of molecular electronic spectra or, as in this instance, the broadened edges of other elemental atomic, or ionic, absorption lines, especially if the element is present in high concentrations, problems may arise. Depending on the shape of the background signal, when the averaged back- ground absorption is subtracted from the atomic absorption signal it can result in under- or over-correction.* It has been reported that such problems can be alleviated by using one of two alternative background correction systems, namely those of Zeemang or Smith and Hieftje.10 This paper reports an investigation of the graphite furnace using Smith - Hieftje background correction.Experimental An instrument fitted with both deuterium-arc and Smith - Hietfje background correction systems (IL 655 furnace fitted to a Video 12 spectrometer; Instrumentation Laboratory, Warrington, UK) or an instrument fitted with deuterium132 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 1.000 a C Table 1. Spectrometer and furnace operating parameters Video 12 instrument: furnace programme using pyrolytically coated cuvettes- Dry Ash Atomise Timels .. . . . . 30 40 30 45 5 5 TemperaturePC . . 80 180 900 1200 2200 2200 InternalN,gas . . Spectrometer parameters: flow-rate(SCFH)* 20 20 20 5 5 5 Wavelength . . 193.7 and 197.2 nm Bandpass . . . . 1.0nm SP-9 instrument: furnace programme using pyrolytically coated cuvettes- TemperaturePC Hold timels Ramp 1 Dry . . . . 200 20 8 2 A s h . . . . 900 30 7 3 Ash . . . . 1200 5 5 4 Atomise . . 2300 10 0 5 Clean . . . . 2800 3 0 Gas flow-rate 3 1 min-l during dry and ash stages 0.5 1 min-1 during atomisation stage Spectrometer parameters: Wavelength 193.7 and 197.2nm Bandpass . . 1.0nm * Standard cubic feet per hour. , . - . . 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 Tirne/s Fig. 1. Plot of signal versus time for arsenic atomisation using deuterium-arc background correction at 193.7 nm.A, Corrected signal; B, total signal; C, arsenic absorption peak; and D, interfering peak ~~ Table 2. Determination of As in reference coals by direct slurry atomisation using deuterium-arc background correction Aslue e-1 Wavelengthhm Sample NBS SRM 1635 NBS SRM 1632a 193.7 Slurry 1.1 k 0.5 20.0 k 0.8 197.2 Slurry 0.38 k 0.10 5.0 k 0.60 - Certificate value 0.32 k 0.13 9.3 k 1.0 hollow-cathode lamp, continuum source, background correc- tion, SP-9 furnace fitted to an SP-9 spectrometer with PU 9095 video furnace programmer and SP-9 computer (Pye Unicam, Cambridge, UK) were used. Full operating details are given in Table 1. Slurry samples were injected into the graphite furnace using precision micropipettes (Gilson Pipetman P; Anachem, Luton, Bedfordshire, UK).Chemicals and Reagents Three certified reference material coals were analysed: sub-bituminous coal (NBS SRM 1632a; National Bureau of Standards, Washington, DC, USA), bituminous coal (NBS SRM 1635) and European coal (BCR No. 40; Community of Bureau of Reference, Brussels, Belgium). All reagents used were of analytical-reagent grade (BDH Chemicals, Poole, Dorset, UK) and all solutions were prepared with doubly distilled, de-ionised water. Stock standard arsenic solution, 1000 mg 1-1. Prepared by the addition of As203 (1.3200 g) to water (25 ml) and nitric acid (50 ml) and heating until dissolved. After dilution with water to 1000 ml, the solution was stored in previously acid-washed polyethylene bottles.Working standard arsenic solutions. Prepared by taking various aliquots of the stock standard solution and matching with the slurry reagent. Sample Preparation Coal (0.5-1 g) was weighed accurately into a 50-ml poly- propylene screw-capped bottle, to which 10 ml of slurry reagent [containing 10 g 1-1 of Ni(N0&.6H20, 10 g 1-1 of Mg(N03)2.6H20, 50 ml l-1 of concentrated nitric acid and 100 ml 1-1 of ethanol] were added. A magnetic stirrer bar was inserted, the bottle was sealed and the contents were stirred until the coal was a homogeneous mixture (typically 5 min). Aliquots of the suspension slurry (10-30 pl) were taken using a micropipette, whilst stirring continuously. When extra grinding was required, this was achieved in a vibrating agate ball mill (Mark 2 Vibrating Mill; Beckman- RIIC, Glenrothes, UK).Particle size measurements were made by the electrical zone-sensing method (Coulter Counter TA 11; Coulter Electronics, Luton, Bedfordshire, UK). In order to prepare a thixotropic slurry, 1 g of sample was added to 50 ml of distilled water. The mixture was stirred and 3.0 ml of 0.1% mlVsodium hexametaphosphate solution were added. This was followed by the addition of 0.1 ml of H2 antifoam agent (Allied Colloids, Bradford, UK) and 1 ml of Viscalex HV30 thixotropic thickening agent (Allied Colloids). The pH was adjusted to 7.5 by addition of ammonia solution, and, once settled, the gel was transferred to a 100-ml calibrated flask and diluted to the mark. Standards were prepared using the same method. Results and Discussion The initial ETA-AAS results for arsenic in NBS SRM 1632a and 1635 using a deuterium arc as a background corrector did not show good agreement with the certified value as reported by Ebdon and Pearce.5 This, however, was in agreement with Riley,7 who attributed the positive errors at the 193.7-nm arsenic line to the broadened aluminium-line, so-called structured background. With this in mind, a furnace programme was developed (Table 1) that included an atomisation step ramped over 5 s, thereby enabling the volatile arsenic to be atomised before the more refractory aluminium.The resulting printout from the instrument (Fig. 1) shows temporal separation of the arsenic and aluminium peaks. However, despite this separation, the continuum background corrector failed to correct for the aluminium interference (Table 2) and large positive errors were observed.When the analysis was repeated using the less sensitive 197.2-nm line, the results for NBS SRM 1635 were within the limits of the certificate value (Table 2). This is not unexpected as the 197.2-nm line is free from any spectral interference from aluminium, -thereby illustrating that both background correc- tors are capable of correcting for the background at this wavelength. The values obtained for NBS SRM 1632a were low, so a new slurry was prepared in which the coal had been ground in a vibrating ball mill for 35 min to reduce the particle size from 100 to 30 pm. When the above experiments were repeated with the ground coal, the recovery of arsenic from the coalJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL.2 133 Table 3. Effect of particle size on arsenic determination at 197.2 nm As in NBS SRM Sample 1632alyg g-1 Slurry of coal as received . . 5.0 f 0.60 Slurry of groundcoal . . . . 9.2 f 1.0 Certificatevalue . . . . 9.3 f 1.0 - ~~ ~ Table 4. atomisation using Smith - Hieftje background correction Determination of arsenic in reference coals by direct slurry As& g - 1 NBS SRM nm Sample 1635 1632a (ground) 193.7 Slurry 0.4 f 0.1 8 . 2 f 1.1 9.5 f 0.7 197.2 Slurry 0.39 k 0.1 6.0 k 0.8 9.2 k 1.0 Certificatevalue 0.32k 0.13 9.3 f 1.0 9.3 k 1.0 Wavelength/ NBSSRM NBSSRM 1632a 1.000 1 . . *. ’ _... I 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 Tim e/s Fig. 2: Plot of signal versus time for arsenic atomisation using Smith - Hieftje background correction at 193.7 nm.A, Corrected signal; B, total signal; C, arsenic absorption peak; and D, corrected interfering peak matrix was enhanced and at 197.2 nm acceptable accuracy was obtained (see Table 3). This is in agreement with the findings of Fuller,ll who studied slurry atomisation by ETA-AAS, and suggests that arsenic was not fully released from large particles in the brief atomisation stage. The possibility of correcting for structured background by the use of the correction system developed by Smith and Hieftjelo was then investigated because, whereas the 197.2- nm line appeared to be interference free, the 193.7-nm line is about twice as sensitive. The above work was repeated using the Smith - Hieftje background correction system in place of the deuterium arc.The results, shown in Table 4, show that with this background corrector, even at the 193.7-nm line good agreement with the certified values is obtained. This demonstrates the effectiveness of this correction system in compensating for the structured background at 193.7 nm. The effective correction of the aluminium peak can be clearly observed in the atomisation printout (Fig. 2) where the background (broken line) still shows the interfering line but the solid line shows that the corrected signal is not affected. Similar results were observed at the 197.2-nm line, which was free of spectral interference. Again, the arsenic recoveries were enhanced by grinding the coal. The slow atomisation ramp proved advantageous at both lines because the relatively slow modulation frequency of the Smith - Hieftje corrector caused problems with rapidly changing high levels of back- ground.This work was repeated on the SP-9 instrumentation fitted with continuum-source background correction. Errors were observed at the primary resonance line (193.7 nm) whereas the correct value was achieved at the “interference free” line (197.2 nm). Table 1 lists the operating parameters and Table 5 gives the results obtained. Analysis of Reference Coal BCR No. 40 The levels of arsenic in BCR No. 40 were determined. The particle size was reduced to below 25 pm and the coal was Table 5. Determination of arsenic in NBS SRM 1632a using deuterium hollow-cathode lamp background correction (SP9 spectrometer) W aveleng th/nm Sample Aslpg g-1 197.2 Slurry 9.8 k 0.8 193.7 Slurry 21.5 k 0.9 Certificate value 9.3 * 1.0 Table 6.Furnace programme for delayed atomisation cuvettes Atomise Dry Ash Time/s . . . . , . 30 40 35 40 5 5 TemperatureI’C . . . , 70 110 550 1000 2300 2300 N2flow-rate(SCFH)* . . 20 20 20 5 5 5 * Standard cubic feet per hour Table 7. Determination of arsenic in BCR No. 40 coal with different forms of background correction at 193.7 nm - Deuterium arc Sample Smith 1 Hieftje Slurry . . . . 12.3 k 1.0 12.8 -t 0.8 Certificate value 13.2 + 1.1 suspended in the slurry reagent as previously described. The conventional cuvettes previously used in the IL furnace were now replaced by the delayed atomisation cuvettes (DAC), in which the centre of the cuvette is thicker than the edges, which leads to slower heating of the centre compared with the ends.In effect, atomisation occurs into a hotter environment, which should reduce the level of non-specific absorption, which is important when considering slurry loading and the determina- tion of volatile elements. The furnace programme was re-optimised, which resulted in a higher atomisation temper- ature (see Table 6). The results with these new cuvettes and using both background correctors are shown in Table 7. It is obvious that both background correctors produce results that are in excellent agreement with the certified values. This implies that the structured background due to aluminium, observed with the previous coals, does not present a signifi- cant problem in BCR No.40. The aluminium interference still persisted for the other coals in the new cuvettes and therefore the reduced interference seems to be due to the higher As : A1 ratio in BCR No. 40 compared with the other coals. Thixotropic Slurries Another method by which slurries have been analysed is by preparing the slurry in a thixotropic medium as described by Fuller and Thompson. 12 Although this does not improve the dispersion of the coal, it may be more suitable for certain auto-samplers as a thixotropic slurry remains homogeneous without constant agitation. The resulting slurries were stable for up to 24 h, but the results were not as reproducible owing to the hydrophobic - hydrophilic interaction between the slurry and the pipette tip, whereby not all the slurry was expelled.It is expected that a pneumatically operated auto-sampler would improve the reproducibility. Conclusion The results illustrate how the problem of structured back- ground, encountered by other workers, can be overcome in the determination of arsenic in whole coal by use of a Smith - Hieftje background correction system. The slurry injection technique described offers a simple and rapid form of analysis, eliminating the need for lengthy contamination-prone sample134 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, MARCH 1987, VOL. 2 preparation stages, thereby increasing the speed of analysis, and is readily adaptable to existing instrumentation. Although there was some decrease in sensitivity using Smith - Hieftje background correction for arsenic (a factor of 1.3 compared with continuum-source correction), the limit of detection at 193.7 nm was still superior by a factor of 2 to that obtained at 197.2 nm using deuterium-arc background correction.The limit of detection using a 10% m/Vcoal slurry was 0.03 Fg g-1 of arsenic in coal, which is adequate for all practical purposes. Slurry atomisation has been shown to be more rapid than dissolution approaches, accurate provided that care is taken over background correction procedures and the furnace programme is optimised, and to offer excellent sensitivity. The authors acknowledge the support of H.G.M.P. by the SERC and British Coal under the CASE Studentship Scheme. References 1. Denton, C. L., Himsworth, G., and Whitehead, J., Analyst, 1972, 97, 461. 2. 3. 4. 5. 6 . 7. 8. 9. 10. 11. 12. Jackson, P. F. S., and Whitehead, J., Analyst, 1966,91, 418. Wilkinson, J. R., Ebdon, L., and Jackson, K. W., Anal. Proc., 1982, 19, 305. Ebdon, L., and Wilkinson, J. R., Anal. Chim. Acta, in the press. Ebdon, L., and Pearce, W. C., Analyst, 1982, 107, 942. Ebdon, L., Pearce, W. C., and Riley, K. W., Analyst, to be submitted. Riley, K. W., At. Spectrosc., 1982, 3, 120. Bauslaugh, K., Radziuk, B., Saeed, K., and Thomassen, Y., Anal. Chim. Acta, 1984, 165, 149. De Loos-Vollebregt, M. T. C., and De Galan, L., Prog. Anal. At. Spectrosc., 1985, 8, 47. Smith, S. B . , and Hieftje, G. M., Appl. Spectrosc., 1983, 37, 419. Fuller, C. W., Analyst, 1976, 101, 961. Fuller, C. W., and Thompson, I., Analyst, 1977, 102, 141. Paper J6/93 Received October 9th, 1986 Accepted November 25th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200131

出版商:RSC    

年代:1987

数据来源: RSC