摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 Inductively Coupled Plasma Mass Spectrometry Has It Already?* Gary Horlick Department of Chemistry University of Alberta Edmonton Alberta Canada T6G 2G2 593 The development and current status of inductively coupled plasma mass spectrometry (ICP-MS) is briefly overviewed in the context of Shakespeare’s seven ages of man. It is argued that the technique has reached the age of maturity and that it is now time for reflection on the future directions for ICP-MS and atomic spectrometry and elemental analysis in general. It is suggested that greater effort is required in developing atomic spectrometric methods for the direct analysis of materials and that alternative mass spectrometric methods with glow discharge and electrospray ion sources will be increasingly utilized for elemental analysis. Keywords Inductively coupled plasma mass spectrometry In a now quite famous editorial’ Laitinen presented a brief discourse entitled ‘The Seven Ages of an Analytical Method’.Laitinen drew his inspiration from a passage in the play 24s You Like It’ written by William Shakespeare in 1598. In Act 11 Scene VI12 the passage begins All the world’s a stage And all the men and women merely players They have their exits and entrances; And one man in his time plays many parts His acts being seven ages The passage continues defining the seven ages of man which are summarized in Table 1. In Laitinen’s editorial he para- phrased these ages into those that he felt represented the phases through which an analytical method or technique progresses.These seven ages as defined by Laitinen are summa- rized in Table 2. He suggested that these ages provide an interesting framework within which to evaluate the current developmental status of an analytical technique and as an example he briefly assessed infrared spectroscopy. Since then many researchers and students have used this framework to assess methods and several workers have addressed this topic with specific reference to analytical atomic s p e ~ t r o m e t r y . ~ ~ ~ In this presentation inductively coupled plasma mass spec- trometry (ICP-MS) will be discussed in terms of the seven ages outlined by Laitinen. Table 1 The seven ages of man-Shakespeare The First Second and Third Ages of ICP-MS Tbe Beginning Shakespeare’s first age (the Infant) as depicted by Robert Smirke is shown in Fig.1. (all figures are from ref. 2). This age is the initiation or conception of the idea. In ICP-MS this age occurred approximately in the time frame of 1974-1980 with the application of MS to flames by Hayhurst and Telford6 and to d.c. plasmas by Gray.7 The second age (the School Boy) is shown as depicted by J. A. Atkinson in Fig. 2. This age is that first foray out into the world and represents in Laitinen’s terms the first research instruments. For ICP-MS this occurred in the time frame of 1980-1983. The first paper in ICP-MS is recognized as that by Houk et aL8 entitled ‘Inductively Coupled Argon Plasma as an Ion Source for Mass Spectrometric Determination of Trace Elements’ and published in late 1980. In the same time frame Douglas and French’ were also building their first research instrument at The Institute for Aerospace Studies of the University of Toronto in Canada.The third age (the Lover) is shown in Fig. 3 as depicted by Robert Smirke. At this age one moves more confidently out into the world hoping to be liked and accepted; for ICP-MS this age represents the first commercial instruments. This occurred in 1983-1984 when both SCIEX and VG Instruments brought out and began delivering the first commercial instru- ments. In hindsight there was actually quite a rapid movement of the technique through these first three ages. One talks of demand-pull and technology-push as forces developing a new ~~~~ ~ 1st the Infant 2nd the School Boy 3rd the Lover 4th the Soldier 5th the Justice 6th the Retired Gentleman 7th the Old Man Table 2 The seven ages of analytical method-Laitinen 1st 2nd 3 rd 4th 5th 6th 7th Conception of the idea First reseach instruments First commercial instruments Characterization of the method Maturity Reflection Old age and senescence *Presented as a Plenary Lecture at the 1993 European Winter Conference on Plasma Spectrochemistry Granada Spain January 10-15 1993.Fig. 1 Engraving from a painting by Robert Smirke (1752-1845) ‘At first the infant Mewling and puking in the nurse’s arms.’594 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 Fig. 2 ‘And then the whining school-boy with his satchel And shining morning face creeping like snail Unwillingly to school.’ Drawing by J.A. Atkinson (1752-ca. 1833) Fig.3 ‘And then the lover Sighing like a furnace with a woeful ballad Made to his mistress’ eyebrow.’ Engraving from a painting by Robert Smirke ( 1752-1845) idea or technique but certainly with ICP-MS there was also an aspect of entrepreneurial spirit that accelerated development through these first three ages. Many aspects of these develop- ments were reviewed at a symposium held at the 1993 Pittsburgh Conference noting the tenth anniversary of the first commercial instrumention ( 1983-1993) for ICP-MS.1° The Fourth Age Characterization and Conquest The fourth age is the age of the Soldier (Fig. 4). In this age the new technique knows no bounds and is out to conquer all problems. This fourth age is the crest of the method and is a period of intense research activity. The main goal of research during this age is to achieve a complete characterization of Fig.4 ‘Then a soldier Full of strange oaths and bearded like the pard Jealous in honor sudden and quick in quarrel Seeking the bubble reputation Even in the cannon’s mouth.’ Drawing by J.A. Atkinson (1752-ca. 1833) the method and this occurred for ICP-MS from 1984 to about 1990 although this certainly is an on-going age. In this time frame many detailed studies were carried out and many developments occurred. Operating conditions were established spectral interferences were delineated and matrix effects were investigated. In addition quantitative analysis protocols were established for ICP-MS and all forms of sample introduction systems developed for ICP atomic emission spectrometry (AES) for liquids solids powders and small volume liquid samples were adapted for ICP-MS.An overview and summary of the current status of ICP-MS in all of these areas is provided in three recent book chapters.”-13 The Fifth Age The Age of Maturity A key age in the development of a technique is the age of maturity. This is Shakespeare’s age of ‘the Justice’ (Fig. 5). As depicted in Fig. 5 we have the wise ‘Justice’ being consulted for his wisdom and knowledge one of the more capable men of the realm. But as can be seen from the depiction in Fig. 5 all is not completely well with the ‘Justice’. Somewhere in the time frame of 1991 to 1992 this age was reached by ICP-MS just a little over 10 years from the appearance of the first research instruments.The current state of development of ICP-MS manifests many of the characteristics of a mature technique. The ICP-MS method is now applied to a wide range of analytical determi- nations and problems,13 and the technique is settling in as a routine tool. There is broadened commercialization as compan- ies realize that it is here to stay. In addition to SCIEX (now Perkin-Elmer-SCIEX) and VG Instruments several other companies are now or soon will be marketing ICP-MS instrumentation. These include Finnigan (marketing the Turner instrument) Varian Thermo Jarrel-Ash Yokogawa and Seiko Instruments. At the same time and somewhat ironically theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 595 Fig. 5 'And then the justice In fair round belly with good capon lined With eyes severe and beard of formal cut Full of wise saws and modern instances; and so he plays his part.' Drawing by J.A. Atkinson ( 1752-CU. 1833) development pace has slowed and there has been a loss of fundamental research momentum. This is reflected in a reduction in the number of fundamental papers one sees at conferences and in the literature. In fact we now see in conference proceedings and in the literature a consideration of and an emergence of competitive technologies. That is in addition to the argon-based ICP source a number of other sources are now being used as ion sources for elemental MS. These include mixed-gas I C P S ' ~ ' ~ alternate gas ICPS,'~.'~ microwave induced plasmas (MIPS),'~~~' glow discharges ( G D S ) ~ ' ? ~ ~ and electrosprays ( E S S ) .~ ~ Also it is by no means clear that the quadrupole is the best type of mass spectrometer for elemental MS. Already sector-based instruments are being ~ s e d ~ ~ ~ ~ and research instruments utilizing t i m e - ~ f - f l i g h t ~ ~ ~ ~ and Fourier transform MSZ8 are being proposed and tested for plasma MS. Finally as is hinted at in the picture of the Justice (Fig. 5) with maturity comes the realization that all may not be well and there are certain problems that just will not go away. These problems do exist for ICP-MS and concern spectral interferences matrix effects sample introduction sys- tems for the direct analysis of materials and for the direct speciation of elemental composition and the adequacy of the quadrupole mass spectrometer with respect to resolution and the measurement of transient signals.The Sixth Age The Age of Reflection While the age of maturity is often considered a pivotal age for a technique it is really the sixth age the age of reflection that may be the most important in terms of the evolution or devolution of a technique. At this time ICP-MS has actually entered this sixth age. The Shakespearean sixth age was depicted by Atkinson as a retired gentleman out walking his dog (Fig. 6). This is an age when the gentleman would have Fig.6 'The sixth age shifts Into the lean and slipper'd pantaloon With spectacles on nose and pouch on side His youthful hose well saved a world too wide For his shrunk shank; and his big manly voice Turning again toward childish treble pipes And whistles in his sound.' Drawing by J.A. Atkinson (1752-ca. 1833) ample time to look back at his life assess where he has been and what he has accomplished and in a relaxed way consider what should or could lie ahead. With the assumption that the retired gentleman was involved in ICP-MS he might be contemplating the three questions summarized in Table 3. While ICP-MS has been emphasized in this presentation a question that still needs to be asked is 'What is the best technique for elemental analysis?. Atomic emission remains a versatile capable reliable and relatively low-cost technique for elemental analysis and new develop- ments in sources and in particular CID29*30 and segmented CCD3'*32 image sensor-based spectrometers stand to revol- utionize atomic emission systems.And surely the retired gentle- Table 3 The sixth age the age of reflection Three Questions 1 What is the best technique for elemental analysis? Emission versus Mass Spectrometry 2 What is the best source? ICP MIP GD ES or ... combination sources 3 What is the best spectrometer? Emission Mass Spectrometer Slew-scan Quadrupole Direct reader Sector PDA Cchelle TOF CID Cchelle Ion trap Segmented CCD Cchelle or ... combination spectrometers CCD Cchelle FTS-MS596 JOURNAL OF ANA.LYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 Fig. 7 ‘Last scene of all That ends this strange eventful history Is second childishness and mere oblivion Sans teeth sans eyes sans taste sans everything.’ Drawing by J. A. Atkinson (1752-ca. 1833) man is also contemplating the two questions alluded to earlier ‘What is the best source? and ‘What is the best spectrometer?’ (See Table 3).Much is happening in these areas that could stem the march of time leading towards the seventh age the period of old age and senescence (Fig. 7). None of us go particularly willingly into this seventh age. For a technique old age and senescence can be staved off and time can even be reversed by the appearance of new ideas and concepts. It is particularly at this sixth age of a method that the flow of time can be reversed. In effect new ideas concepts and developments allow one to re-enter the time-line at earlier levels a seven-ages concept suggested by Hieftje’. However while some new developments will rejuvenate the ICP-MS time-line by feedback others will lead to competitive evolutions that will result in the establishment of new time-lines and the rejuvenation of old time-lines.Feedback and New Time-lines One of the ‘problems that won’t go away’ that was mentioned earlier is the direct analysis of solid materials by ICP-MS. This of course is not just a problem for ICP-MS but Table 4 Seven ages feedback and formation of new time lines represents a challenge for both ICP-MS and ICP-AES. In our own laboratory this problem is being addressed by the use of direct sample insertion systems in conjunction with mixed-gas ICP meth~dology.~~ With this approach a diverse range of materials including metal turnings and fillings foods botan- icals refractories and oil can all be directly analysed.The characterization of new sample introduction techniques rep- resents re-entry into the ICP-MS time-line at the fourth age (see Table 4). The recent developments in GDMS can be thought of as being initiated by a feedback evolution of ICP-MS with subsequent rejuvenation of the GDMS time-line. While GDMS actually predates ICP-MS by several years its seven ages time- line was stagnated around the second and third ages in the first half of the 80s. The success and rapid development of ICP-MS lent credibility to the general utilization of MS for elemental analysis. This has lead in turn to the rejuvenation of GDMS. Interest in this technique has developed rapidly in the last few years in part because it is directly applicable to the analysis of solid samples a sample form that is difficult to handle directly with ICPs.In a way this has re-started the GDMS time-line and .there is now considerable activity throughout ages two three and four for this method. Recently the analytical capabilities of ICP-MS and GDMS have been compared.34 Finally another of the ‘problems that won’t go away’ is the need for systems that are applicable to the direct speciation of elemental composition. Such analyses have tended to be implemented by coupling atomic spectrometry detection sys- tems (ie. ICP-AES and ICP-MS) to chromatographic front ends. Earlier the sixth age question was formulated ‘What is the ideal source for elemental analysis?. Perhaps it might be useful to re-formulate the question as ‘What should be the goals of an ideal elemental analysis system for the determi- nation of the composition of solution samples?.By composi- tion I mean that it should be capable of directly determining both cations and anions in a solution and that it should provide the atomic and molecular form of the ions along with their valence state. Thus if vanadium were present in a solution this ideal method should enable the individual con- centrations of V2+ V3+ V02+ and V022+ to be determined and not just provide total vanadium concentration. This method should also allow the anion composition of a solution to be determined thus directly providing for the determination of ions such as Cl- Br- I- Cr,0,2- and SCN-. In addition it should also provide ligand information and information about the degree of complexation such as directly determining Zn(NH,),2+ and Zn(NH3)32+.Well that’s a big ideal. However the potential for such capability in elemental analysis 1st E i 7th 1st 2nd 3rd 4th 5th 6th 7th i t d 3rd 4th 5th 6th ICP-MS Conception of the idea First research instruments First commercial instruments Characterization of the method Maturity Reflection Old age and senescence 7th GDMS Conception of the idea First research instruments First commercial instruments Characterization of the method Maturity Reflection Old age and senescence ESMS Conception of the idea First research instruments First commercial instruments Characterization of the method Maturity Reflection Old age and senescenceJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 597 has already been demonstrated by electrospray MS.23935936.This development represents the establishment of a new seven ages time-line as illustrated in Table 4. Thus while ICP-MS has entered the latter ages of the seven ages time frame it is being rejuvenated by new developments in methodology. In addition a consideration of the limitations of ICP-MS and of the needs of elemental analysis has resulted in the rejuvenation of the time-line of an established method such as GDMS and in the establishment of a new time-line for ESMS. Clearly there remain many challenges and oppor- tunities in developing methods for the determination of the elemental composition of materials as defined in the broadest sense. 1 2 3 4 5 6 7 8 9 10 11 12 13 References Laitinen H. A. Anal. Chem.1973 45 2305. Rouse A. L. The Annotated Shakespeare Greenwhich House New York 1988 p. 360. Barnes R. M. TrAC Trends Anal. Chem. (Pers. Ed.) 1981 1 51. Fassel V. A. Fresenius 2. Anal. Chem. 1986 324 511. Hieftje G. M. Spectrochim. Acta Part B Special Supplement 1989 44 113. Hayhurst A. N. and Telford N. R. Combust. Flame 1977 28 67. Gray A. L. Analyst 1975 100 289. Houk R. S. Fassel V. A Flesch G. D. Svec H. J. Gray A. L. and Taylor C. E. Anal. Chem. 1980 52 2283. Douglas D. J. and French J. B. Anal. Chem. 1981 53 37. Pittsburgh Conference Atlanta GA USA 1993 paper numbers Horlick G. and Shao Y. in Inductively Coupled Plasmas in Analytical Atomic Spectrometry eds. Montaser A. and Golightly D. W. VCH Publishing New York 2nd edn. 1992 551. Douglas D. J. in Inductively Coupled Plasmas in Analytical Atomic Spectrometry eds.Montaser A. and Golightly D. W. VCH Publishing New York 2nd edn. 1992 p. 613. Taylor H. E. and Garbarino J. R. in Inductively Coupled Plasmas in Analytical Atomic Spectrometry 2nd ed. eds. Montaser A. and Golightly D. W. VCH Publishing New York 2nd edn. 1992 p. 651. 978-982. 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Lam J. W. and Horlick G. Spectrochim. Acta Part B 1990 45 1313. Lam J. W. and McLaren J. W. J. Anal. At. Spectrom. 1990 5,419. Evans E. H. and Ebdon L. J. Anal. At. Spectrom. 1990 5 425. Montaser A. Chan S. K. and Koppenaal D. W. Anal. Chem. 1987 59 1240. Koppenaal D. W. and Quinton L. F. J. Anal. At. Spectrom. 1988 3 667. Shen W-I. Davidson T. M. Creed J. T. and Caruso J.A. Appl. Spectrosc. 1990 44 1003. Shen W-l. Davidson T. M. Creed J. T. and Caruso J. A. Appl. Spectrosc. 1990 44 101 1. Harrison W. W. J. Anal. At. Spectrom. 1988 3 867. Shao Y. and Horlick G. Spectrochim. Acta Part B 1991 46 165. Agnes G. R. and Horlick G. Appl. Spectrosc 1992 46 401. Bradshaw N. Hall E. F. H. and Sanderson N. E. J. Anal. At. Spectrom. 1989 4 801. Morita M. Ito H. Uehiro T. and Otsuka K. Anal. Sci. 1989 5 609. Farnsworth P. B. Wu M. Lee M. L. Lee E. D. and Prince J. paper presented at the 1993 Pittsburgh Conference Atlanta GA 1993 paper No. 1342. Hieftje G. M. and Myers D. P. paper presented at the European Winter Conference on Plasma Spectrochemistry Granada Spain January 10-15 1993 paper No. OR3-2. Marcus R. K. Cable P. R. Duckworth D. C. Buchanan M. V. Pochkowski J. M. and Weller R. R. Appl. Spectrosc. 1992 46 1327. Bilhorn R. B. and Denton M. B. Appl. Spectrosc. 1990,44,1538. Pilon M. J. Denton M. B. Schleicher R. G. Moran P. M. and Smith S. B. Appl. Spectrosc. 1990 44 1613. Barnard T. W. Crockett M. I. Ivaldi J. C. and Lundberg P. L. Anal. Chem. 1993 65 1225. Barnard T. W. Crockett M. I. Ivaldi J. C. Lundgerg P. L. Yates D. A. Levine P. A. and Sauer D. J. Anal. Chem. 1993 65 1231. Liu X. and Horlick G. J. Anal. At. Spectrom. in the press. Feng X. and Horlick G. J. Anal. At. Spectrom. in the press. Agnes G. R. and Horlick G. Appl. Spectrosc. in the press. Agnes G. R. and Horlick G. Appl. Spectrosc. in the press. Paper 3/06649E Received December 6 1993 Accepted December 17 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900593

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 599 Determination of Lead and Lead Isotope Ratios in Gasoline by Inductively Coupled Plasma Mass Spectrometry Charles J. Lord 111 Phillips Petroleum Company Research and Development Bartlesville OK 74004 USA An accurate method for determining the concentration of lead in gasoline over the range 0.004-1500 1-14 g-' is described. Tetraalkyllead compounds in the gasoline are converted into water-soluble species that are subsequently extracted into dilute nitric acid. Lead concentrations in the nitric acid extracts are determined using inductively coupled plasma mass spectrometry with a bismuth internal standard. The relative standard deviation of this method is +2.4% for lead concentrations greater than 0.1 pg g-'. Studies with spiked samples and standard reference materials demonstrate excellent quantitative recoveries.The detection limit for lead in a 1 g gasoline sample is 0.004 kg g-'. Gasoline samples as large as 25 g can be extracted to enhance the accuracy and precision of low level (<0.1 pg g-') determinations. Isotope ratio measurements on National Institute of Standards and Technology Standard Reference Material 1636a (Lead in Reference Fuel) yield a 206Pb/207Pb ratio of 1.2230 & 0.0035 (95% confidence interval). The precision of these ratio measurements is suitable for performing both isotope dilution analyses as well as isotopic tracer studies. Keywords Lead in gasoline; inductively coupled plasma mass spectrometry; lead isotope ratios; gasoline analysis; aqueous extraction Environmental and health concerns related to lead exposure have resulted in the phasing out of lead additives in many consumer products such as solders paint pigments and gaso- lines.The use of lead in gasoline has been decreasing steadily over the last decade in many industrialized nations around the world. This decrease has occurred because of governmental regulations limiting the amount of lead that can be added to leaded fuels and because of an increased market demand for lead-free gasoline. Regulations in the United Kingdom' allow a maximum lead content of about 2OOpgg-1 while in the United States the maximum level is 36 pg g-'. Exceptions to the US regulations are aviation and racing fuels which may contain up to 15OOpgg-' of lead. Lead-free gasoline in the US cannot receive any intentional addition of lead but is currently allowed up to 18 pg g-' of lead from incidental contamination during transportation and storage.A variety of methods have been reported in the literature for measuring the concentration of lead in gasoline. These include flame atomic absorption;-7 direct current plasma emission,8 ~olorimetry,'-~' polarography,12 high-performance liquid ~hromatography,'~ X-ray fluore~cence,'~ gas chromatog- r a p h ~ ~ ' electrothermal atomic absorption,16 hydride gener- ation atomic ab~orption,'~"~ and titrimetr~.'~ Each of these procedures suffers from inherent limitations in the analytical technology that they employ. For example the titrimetric procedure" is intended for high concentration lead measure- ments and cannot provide a sufficiently low detection limit to verify that gasolines are legally 'lead-free' (ie.contain < 18 pg g-l). A major shortcoming of the atomic absorp- tion2-7,16-18 and c~lorimetric'-'~ techniques is limited dynamic range (typically two orders of magnitude) which often necessi- tates pre-screening and pre-dilution of samples. The chromato- methods supply interesting information concerning the speciation of alkyllead compounds but are not practical for routine lead analyses. The interpretation of polarographic12 data for gasoline samples appears to be somewhat complex and highly matrix dependent. Optical emission methods using direct current plasma8 or inductively coupled plasma sources offer good dynamic range but require sample dilutions of 25- to 50-fold and are therefore not useful for gasolines that contain less than about 5 pg g-' of lead.X-ray fluorescence spectrometry14 requires little sample preparation however like plasma emission it is not suitable for ultratrace lead analysis. The ability to measure ultratrace concentrations of lead is critical for applications such as the investigation of contami- nated fuel or petrochemical feedstock shipments. For these applications the methods discussed above do not provide adequate performance capabilities. A new method based on inductively coupled plasma mass spectrometry (ICP-MS) has been developed to fill this gap in analytical performance. The ICP-MS procedure described in this paper is more versatile and powerful than any previously published method for the determination of lead in gasoline.With this procedure lead concentrations from ultratrace levels to 1500 pg g-' can be accurately measured. The ICP-MS technique overcomes the limitations of previous methodologies and offers a unique combination of benefits including a dynamic range of more than five orders of magnitude a very low detection limit (0.004 pg g - I) rapid and simple sample preparation excellent accuracy and precision calibration with aqueous standards minimal use of organic solvents and the ability to measure lead isotope ratios. In addition this procedure requires no hardware or software modifications to the ICP mass spectrometer. Experimental Instrumentation A VG PlasmaQuad 2 Turbo-Plus (VG Elemental Winsford Cheshire UK) ICP mass spectrometer was used to perform the lead analyses.Argon for the ICP was obtained from the gas outlet of a 160 1 liquid argon Dewar and all gas flow rates were regulated by mass flow controllers. Sample aerosols were produced with a DeGalan V-groove nebulizer (VG Elemental) to which solutions were metered by means of a peristaltic pump (Gilson Medical Electronics Villiers le Bel France). Voltages on the ion lenses were adjusted to maximize the count rate at m/z 209. Internal standardization was accomplished by adding an aqueous 'O'Bi spike to all blank standard and sample solutions. Ions were detected using an electron multiplier operated in both analogue and pulse count- ing modes. The choice between analogue and pulse counting detection was made automatically by the instrument software. If the count rate for a particular ion exceeded 2 x lo6 counts s-' then only analogue data were collected at that mass.A listing of the instrument operating conditions is given in Table 1. Reagents Analytical-reagent grade iodine and toluene and Optima grade nitric acid (Fisher Scientific Pittsburgh PA USA) were used600 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 Table 1 ICP-MS operating conditions R.f. power Forward Reflected Frequency Ar flow rates Outer Intermediate Carrier Nebulizer DeGalan Solution pump rate Spray chamber Temperature Internal standard Bismuth Isotopic mass Sampler-load coil separation Sampler Skimmer (micro) Scan parameters Dwell time No. of MCS channels No. of sweeps per scan Mass range Electron multiplier Interface cones Detector 1500 W < 5 w 27.12 MHz 14.0 1 min-' 0.50 1 min- 1.05 1 min-' V-groove plastic 1.0 ml min-' Double-pass water jacketed 10 "C Aqueous 0.100 yg g-' 209 12 mm Ni 1.0 mm orifice Ni 0.7 mm orifice 160 ys 512 800 202-211 u Analogue and pulse counting in the sample preparation procedure.Although ultrapure nitric acid is recommended for trace lead analyses analytical-reagent grade acid is an acceptable alternative for samples containing more than 5 pg g-' of lead. The water used in this work was triply de-ionized with the final stage of de-ionization provided by a Milli-Q water purification system (Millipore Bedford MA USA). A 3% m/v iodine solution was prepared by adding 3 g of iodine to a calibrated flask and diluting to 100 ml with toluene.A 10% v/m nitric acid solution was prepared by diluting 100ml of concentrated nitric acid (sp. gr. 1.42) to 1000 g with de-ionized water. A commercial premium unleaded gasoline (Phillips Petroleum Bartlesville OK USA) and Optima grade isooctane (Fisher Scientific) were used as diluents in the lead recovery studies. Instrument calibration standards were prepared from 100 pg ml-' aqueous stock solutions of lead and bismuth (Inorganic Ventures Toms River NJ USA). The 0.100 pg g-' calibration standard was made by diluting 100 pl each of the 100 pg ml-' lead and bismuth stock solutions to 100 g with 10% v/m nitric acid. The calibration blank was prepared by diluting 100 pl of the 100 pg ml-' bismuth stock solution to 1OOg with 10% v/m nitric acid.An Eppendorf (Brinkmann Instruments Westbury NY USA) digital micropipette (10-100 pl capacity) was used to dispense the stock solutions. The accuracy of the method was evaluated using National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1636a Lead in Reference Fuel (NIST Gaithersburg MD USA) which contains tetraethyllead at four concentrations ranging from 11.2 to 764 pg g-'. (Note SRM 1636a is no longer available from NIST however an equivalent set of reference materials is offered in SRMs 2712-2715). The lead isotope ratio measurements were cali- brated against NIST SRM 981 (Common Lead Isotopic Standard). Procedure Gasolines were prepared for analysis as follows. Approximately 1.0 g of sample was placed in a 60 ml polyethylene bottle and the mass was recorded to the nearest 0.001 g (in order to minimize evaporation and obtain a more accurate and stable reading the bottle was capped prior to recording of the mass).Using a micropipette 75 p1 of a 3% m/v iodine in toluene solution were added to the sample. The contents of the bottle were mixed and then allowed to react for 5 min. An appropriate volume of a 100 pg ml-' aqueous bismuth solution was added to each sample as an internal standard. The volume (pl) of the internal standard addition was calcu- lated by multiplying the sample mass (g) by 50. A digital micropipette with a volume resolution of 0.1 pl was used for this operation. Next about 50 g (2547 ml) of 10% v/m nitric acid were added to each bottle the mixture was manually shaken and then placed on a mechanical shaker for 5 min.It was not necessary to record the mass of the nitric acid added because the internal standard automatically compensated for any variation in the mass of the final solution. The concen- tration of bismuth in the final solutions was assigned a value The gasoline (top) layer was allowed to separate from the 10% nitric acid (bottom) layer then about 20 ml of the nitric acid layer were removed for analysis by ICP-MS. Alternatively the sample could have been analysed in the original polyethyl- ene bottle provided that the peristaltic pump was turned off during the insertion and removal of the nebulizer uptake tube (turning off the peristaltic pump prevented gasoline from being aspirated into the plasma as the uptake tube passed through the organic layer).The nitric acid extracts were stable indefi- nitely and could be stored for many months with no loss of sample integrity. The lead concentration in the nitric acid solution was determined from the ratio of the 208Pb signal to the 0.100 pg g-' 'O'Bi signal. The response factor for lead relative to bismuth was normally about 1.2. The exact value for the relative response factor was obtained from measurements on a 0.100 pg g-' aqueous lead calibration standard. Because of the wide linear dynamic range of ICP-MS a single calibration standard is usually sufficient for quantitative analysis. A dilution factor of 50 was used to calculate the lead content of the original gasoline sample. of 0.100 pg g-I. Results and Discussion Despite the inherent advantages of the ICP-MS technique its use for determining lead in gasoline has not been previously reported.A major reason for this is that volatile organic solutions such as gasoline create serious operational difficulties when introduced directly into an ICP-MS instrument. The two most severe problems are plasma instability and carbon build-up on the mass spectrometer interface cones. In order to overcome these difficulties it is necessary to eliminate the organic matrix before the sample is aspirated into the plasma. In the procedure described here tetraalkyllead compounds in gasoline are converted into water-soluble species and then quantitatively extracted into dilute nitric acid. Lead is thereby separated from the gasoline matrix in a simple one-step extraction process.These lead extracts can then be analysed with the standard ICP-MS hardware and operating conditions that are used in routine analyses of aqueous solutions. Optimization of this procedure in terms of extraction time and sample size is discussed in the following two sub-sections. A discussion of the analytical performance capabilities of the method is also presented below. Aqueous Extraction Rate A key feature of the iodine-tetraalkyllead reaction is that the reaction products are water-soluble compounds. In order to optimize the procedure it was necessary to determine the transfer rate of iodized lead from the gasoline phase into the aqueous phase. Because kinetic data for the aqueous extraction process were not previously published a study of lead recovery as a function of extraction time was performed.Extraction times ranging from 3 to 240min were included in the study.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 601 The results of these experiments are shown in Fig. 1. The data indicate that the transfer of lead into the nitric acid solution is a very rapid process requiring less than 3 min for complete quantitative recovery. A conservative extraction time of 5 min was therefore adopted for routine use in the ICP-MS method. Sample Size For samples containing greater than 0.1 pg g-' of lead the optimum sample size is 1 g. This mass of sample allows even the most highly leaded gasolines to be processed using the standard procedure described above. From a stoichiometric viewpoint 75 p1 of a 3% m/v iodine solution is sufficient to convert 1840 pg of lead from the tetraalkyl compounds into the PbI form.For a 1 g gasoline sample that contains the maximum allowable lead content (x 1500 pg g-') there will be a modest excess of iodine reagent. Furthermore the aqueous extraction solutions will always be at least an order of magni- tude under-saturated with respect to PbI solubility. These amounts of sample and reagent were chosen to prevent any inaccuracies due to insufficient reagent availability or the formation of insoluble precipitates during the analysis of leaded gasolines. Although 1 g of sample is recommended for routine gasoline analyses extraction of a larger sample will decrease the dilution factor and should improve the accuracy and precision of ultratrace lead measurements.In order to determine whether lead was quantitatively recovered from larger samples a series of experiments were conducted in which the sample mass was varied while the iodine and nitric acid solutions were main- tained at their standard levels. The test sample in these experiments was a gasoline containing 0.1 12 pg g-l of lead. The results of this work plotted in Fig. 2 demonstrate that excellent quantitative recovery of lead is obtained from samples as large as 25g. The lead recovery curve is linear which indicates that the extraction efficiency is independent of sample size. The slope of the recovery curve is 0.111 pg g-' which is in excellent agreement with the known gasoline concentration of 0.112 pg g-'.The ability to select sample masses of between 1 and 25 g introduces tremendous flexibility into this ICP-MS procedure. Specific improvements in the accuracy and precision of low level lead analyses are discussed further in the follow- ing sections. Limit of Detection Measurements and statistical calculations on samples contain- ing lead at three to five times the estimated limit of detection (LOD) were used to determine both the instrument and method LODs. In this paper the LOD is defined as the minimum concentration of lead that can be measured and Certified value 0 50 100 150 200 250 Extraction time/min Fig. 1 Kinetic study of the extraction of lead from gasoline. A 1 g sample of gasoline containing 11.2 pg g-' of lead was reacted with 75 pl of 3% m/v iodine solution for 5 min followed by the addition of 50g of 10% v/m nitric acid and shaken for lengths of time ranging from 3 to 240min.Quantitative extraction of lead was achieved in less than 3 min 3.0 2.5 cn g 2.0 2 2 a 1.0 $ 1.5 0 (0 1 0.5 20 25 0 5 10 15 Sample mass/g Fig.2 Recovery of lead from a gasoline sample containing 0.112 pg g-' of lead. The recovery curve is linear (rZ=0.9982) the intercept is 0.003 pg and the slope is 0.111 pg g-' which is equivalent to a recovery efficiency of >99% reported with 99% confidence as being greater than zero. The following equation was used in the calculations LOD = t x s where t =single sided t-statistic for a 99% confidence level and s = standard deviation of replicate analytical measurements. Analyses of dilute aqueous lead solutions using our ICP-MS equipment gave an instrument LOD of 0.07 ng 8-l.Multiplying this value by the dilution factor of the method 50 yields an estimated LOD of 0.004 pgg-' for lead in gasoline. The 0.011 pg g-' sample listed in Table 2 is roughly three times the estimated LOD and therefore provides appropriate data for calculating the LOD of the method. Based on these data the calculated LOD for lead in gasoline is 0.004 pg 8-l. This method LOD is in good agreement with the LOD estimated above from measurements of aqueous solutions. A lower LOD can be obtained by extracting lead from a larger gasoline sample. For example if a 25 g gasoline sample is extracted into 50g of 10% nitric acid then an LOD of 0.0002 pg g- ' should theoretically be attainable The ability to achieve this theoretical limit will largely depend on how effectively the low level lead contamination is controlled. At these ultratrace concentrations the purity of reagents and sample containers becomes an important analytical issue.Analytical Precision The precision data obtained from replicate determinations of lead in gasoline over a range of concentrations are presented in Table 2. The 11.2-764 pg g- ' samples were the four concen- Table 2 gasoline using ICP-MS Precision obtained from replicate determinations of lead in Gasoline lead concentration/ 18 g-' 0.01 1 0.01 1 0.1 12 1.12 11.2 18.8 25.1 764 764 1320 1520 Pooled value* Sample mass/ g 1 .o 25.0 1 .o 1 .o 1 .o 1 .o 1 .o 1 .o 2.0 1 .o 1 .o No. of replicates 5 5 5 10 13 5 5 4 5 5 5 RSD* ("/) 10.6 4.4 3.4 3.3 1.4 1.4 0.9 3.9 1.3 0.7 0.5 2.4 *The pooled RSD value is calculated using a degree of freedom weighting procedure. All data except the 0.01 1 pg g-' ( 1 g) sample are included in the calculation. The pooled value represents an average with 52 degrees of freedom.602 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 trations supplied by NIST as SRM 1636a the 0.011- 1.12 pg g-' samples were serial dilutions of the 11.2 pg g-' reference material and both the 1320 and 1520 pg g-' samples were commercial racing fuels. The average relative standard deviation (RSD) over the 0.112-1520 pg g-' range was f2.4%. This pooled average with 52 degrees of freedom was calculated using a statistical weighting procedure and includes the data from the 0.01 1 pg g-' (25 g) sample.The precision of these analyses is similar to that observed in simple aqueous solutions which indicates that the sample extraction procedure does not intro- duce significant variance into the analytical data. Furthermore identical analytical performance is obtained regardless of whether isooctane or commercial gasolines are used in the procedure. This is important because it demonstrates that the ICP-MS method is free from any biases or interferences caused by compositional changes in the gasoline matrix. As expected for measurements near the LOD the analytical uncertainty increases. This increased variation is seen in the 0.011 pg g-' sample of Table 2 which has a h 10.6% RSD. However by processing 25 g of the same gasoline sample the RSD is substantially improved to +4.4%.Analytical Accuracy The accuracy of this procedure was evaluated by analysing NIST SRM 1636a and commercial gasolines of known lead content. The results of this evaluation shown in Table 3 demonstrate excellent quantitative recovery of lead over the entire concentration range of 0.011-1520 pg g-'. The average percent recovery is 100 + 2 for lead concentrations greater than 0.1 pg g-'. For lead concentrations less than 0.1 pg g-' there appears to be a slight negative bias in the data. By processing a larger sample the accuracy of low level measurements is improved. This can be seen by comparing the recovery results (Table 3) for the 1 and 25 g extractions of the 0.011 pg g-' gasoline sample. The dynamic range of this method is greater than five orders of magnitude which allows all gasoline samples to be analysed using exactly the same procedure.In practice this represents a significant time saving because accurate quantitative results can be obtained without having to pre-screen or re-dilute samples. Lead Isotope Ratios Because a mass spectrometer system is used to determine lead in this procedure isotope ratio data are readily available. A typical mass spectrum obtained during the data acquisition step is presented in Fig. 3. The ability to measure lead isotope ratios allows isotope dilution procedures as well as isotope tracer studies to be performed. Isotope dilution analysis can be achieved simply by replacing the bismuth internal standard with a lead isotope spike solution.Very accurate lead determi- nations are attainable with the isotope dilution technique which is based on isotope ratio data from a spiked and unspiked sample pair. In all mass spectrometers the instrument response varies as a function of ion mass. Over a particular mass range this response variation can have either a positive or a negative slope. A positive slope for example results in larger signals from the higher mass ions relative to the lower mass ions. Although the magnitude of this mass bias effect is small over the range of a few mass units it must be accounted for in order to derive accurate isotope ratios. The 206Pb/207Pb isotope ratios measured over a wide range of sample concentrations are listed in Table 4. These data have been adjusted to compensate for the mass bias of our instru- ment.The mass bias correction factor was determined from isotope measurements on aqueous solutions of NIST SRM 98 1 (Common Lead Isotopic Standard). The correction factor applied to the gasoline data is 1.0060. As can be seen in Table4 for lead concentrations greater than 1 pgg-l very precise ratios are obtained. An average relative precision of 0.3% (95% confidence interval) is attained in these cases. For the 0.112 pg 8-l sample the mean 206Pb/207Pb ratio agrees well with the higher concentration samples however the measurement is less precise The 0.011 pg 8-l sample which is close to the LOD of the method 150 1 '08Pb V 202 204 206 208 210 212 m/z Fig. 3 Mass spectrum of lead extracted from gasoline (SRM 1636a).The four lead isotopes and the bismuth internal standard are shown. The concentration of bismuth is 0.100 pg 8-l in the nitric acid extract and the 206Pb/207Pb ratio in the sample is 1.2230f0.0035 Table3 commercial fuels Accuracy of lead in gasoline determinations based on ICP-MS analyses of NIST SRM 1636a (Lead in Reference Fuel) and Gasoline lead concentration/ Pg 8-l 0.000 0.01 1 0.01 1 0.1 12 1.12 11.2 18.8 25.1 764 764 1320 1520 Average* Sample mass/ g 1 .o 1 .o 25.0 1 .o 1 .o 1 .o 1 .o 1 .o 1 .o 2.0 1 .o 1 .o Lead added/ PI; 0.000 0.01 1 0.275 0.1 12 1.12 11.2 18.8 25.1 764 1530 1320 15201 Lead recovered/ Pg < 0.004 0.010 0.270 0.108 1.11 11.1 18.7 25.5 774 1540 1340 1480 Recovery efficiency ("/.I - 91 98 96 99 99 100 102 101 101 102 97 100 +_ 2 * The average recovery efficiency is the mean of all the data except the 0.011 pg g-' (1 g) sample.The error interval represents one standard deviation.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 603 Table 4 Precision of lead isot ,e ratio measurements on gasoline samples over a wide range of concentrations Lead concentration/ g-' 0.01 1 0.01 1 0.112 1.12 11.2 18.8 25.1 764 764 Mean* Sample mass/ g 1 .o 25.0 1.0 1 .o 1 .o 1.0 1 .o 1 .o 2.0 Isotope ratio (206Pb/207Pb) 1.1531 1.2331 1.2191 1.2228 1.2223 1.2223 1.2247 1.221 1 1.2259 1.2230 95% Confidence interval f 0.0283 f0.0103 + 0.0205 0.0043 & 0.0025 + 0.0045 0.0029 f 0.0050 * 0.0022 f 0.0035 * Mean values are calculated from the 1.12-764 pg g-' samples and are weighted using the degrees of freedom associated with each measurement.does not yield very precise or accurate isotope ratios when a 1 g sample is processed. However as discussed above the data for very low concentration samples can be improved by extracting lead from a larger gasoline sample. The results compiled in Table 4 show that when 25 g of the 0.011 pg g-' sample is extracted both the accuracy and precision of the isotope ratio are significantly improved. Isotopic Tracers The lead emitted in automobile engine exhaust retains the original isotopic ratios of the tetraalkyllead fuel additives. This is because no isotopic fractionation of lead occurs during the combustion process. The average 206Pb/207Pb ratio given in Table 4 for NIST SRM 1636a is 1.2230+0.0035 (95% confi- dence interval).This ratio is characteristic of the Mississippi Valley lead ores (J-type lead) and is distinctly higher than the average ratio of 1.182 in the earth's continental crust.20 The SRM 1636a was issued by NIST in 1980 and contains the tetraalkyllead additives that were in use during that time period in the US. Data compiled by Sturges and Barrie21 yield an average 206Pb/207Pb ratio of 1.213 for atmospheric particu- late samples collected in the US during 1982-1985. The similarity between the gasoline and atmospheric lead isotope ratios suggests that automobile exhaust was a major source of airborne lead in the early 1980s. If it is assumed that these atmospheric particulates contained a mixture of lead from both gasoline and crustal sources then using simple algebra it can be calculated that about 75% of the airborne lead was derived from the combustion of leaded fuel in automobile engines.It should be noted that the atmospheric contribution from leaded gasoline has declined sharply in the US since 1986 when major reductions in lead additive concentrations were made mandatory nationwide. The same pipelines and tankers that are used to ship leaded fuels such as aviation gasoline are also used to transport unleaded gasoline diesel fuel and petrochemical feedstocks. Sometimes cross-contamination occurs between the leaded and unleaded products. The isotopic tracer approach described above can be used to identify the source of the contamination. This approach is most effective when each of the potential sources is characterized by a distinctive 206Pb/207Pb ratio.The 206Pb/207Pb ratios in gasoline change periodically as different lead ores are used to manufacture the tetraalkyllead additives. In the early 1980s a ratio of 1.223 was typical while the modern day racing fuels analysed in this study are characterized by a ratio of 1.153. Either of these 206Pb/207Pb ratios could be easily distinguished from the average crustal value of 1.182. Conclusions The ICP-MS procedure is the most versatile method available for determining the concentration of lead in gasoline. The wide dynamic range of ICP-MS allows trace quantities as well as high levels of lead to be determined using the same sample preparation and analysis procedure. Excellent accuracy and precision very low LODs and isotope ratio measurements are achievable with this method.Because organic solvents are not used to dilute the gasoline samples the volume of solvent waste generated during an analysis is minimal. Calibration with aqueous rather than organometallic standards further reduces the production of organic waste. This waste minimization feature is an important environmental benefit of the ICP-MS procedure. The lead extraction procedure ensures that uniform and consistent results are obtained from all samples regardless of compositional variations in the gasoline matrix. Separating the lead from the gasoline matrix also removes the problems associated with analysing organic solutions by ICP-MS such as plasma instability and carbon build-up on the mass spec- trometer interface cones.I thank the management of Phillips Petroleum Company for their support and permission to publish this research. I also thank Greg Gingerich for his careful work in preparing and analysing the many gasoline samples that comprise the basis of this paper. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 References Hunt A. Johnson D. L. Watt J. M. and Thornton I. Environ. Sci. Technol. 1992 26 1513. Lukasiewicz R. J. Berens P. H. and Buell B. E. Anal. Chem. 1975 47 1045. McCorriston L. L. and Ritchie R. K. Anal. Chem. 1975,47 1137. Madec M. and LaVilla F. Rev. Inst. Fr. Pet. 1976 31 687. Berenguer V. Guinon J. L. and De La Guardia M. Fresenius' Z . Anal. Chem. 1979 294,416. Polo-Diez L. Hernandez-Mendez J. and Pedraz-Penalva F. Analyst 1980 105 37. Annual Book of ASTM Standards American Society for Testing and Materials Philadelphia PA 1992 volume 5.02 Method D 3237 p. 624. De La Guardia M. Salvador A. and Berenguer V. Analusis 1981 9 74. Griffing M. E. Rozek A. Snyder L. J. and Henderson S. R. Anal. Chem. 1957 29 190. Annual Book of ASTM Standards American Society for Testing and Materials Philadelphia PA 1992 volume 5.02 Method D 3116 p. 561. Annual Book of ASTM Standards American Society for Testing and Materials Philadelphia PA 1992 volume 5.02 Method D 3348 p. 726. Guifion J. L. and Grima R. Analyst 1988 113 613. Bond A. M. and McLachlan N. M. Anal. Chem. 1986,58 756. Annual Book of ASTM Standards American Society for Testing and Materials Philadelphia PA 1992 volume 5.02 Method D 3229 p. 584. Robinson J. W. Kiesel E. L. Goodbread J. P. Bliss R. and Marshall R. Anal. Chim. Acta 1977 92 321. Scott D. R. Holboke L. E. and Hadeinshi T. Anal. Chem. 1983 55 2006. Aznarez J. Vidal J. C. and Carnicer R. J. Anal. At. Spectrorn. 1987 2 55. Nerin C. Olavide S. and Cacho J. Anal. Chem. 1987 59 1918. Annual Book of ASTM Standards American Society for Testing and Materials Philadelphia PA 1992 volume 5.02 Method D 3341 p. 713. Chow T. J. and Earl J. L. Science 1972 176 510. Sturges W. T. and Barrie L. A. Atmos. Environ. 1989 23 1645. Paper 3/07065 D Received November 29 1993 Accepted January 31 1 YY4

ISSN:0267-9477    

DOI:10.1039/JA9940900599

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 605 Direct Analysis of Solids by Ultrasonic Slurry Electrothermal Vaporization Inductively Coupled Plasma Mass Spectrometry D. Conrad Gregoire Geological Survey of Canada 601 Booth St. Ottawa Ontario Canada KIA OE8 Nancy J. Miller-lhli United States Department of Agriculture Nutrient Composition Laboratory Beltsville MD 20705 USA Ralph E. Sturgeon Institute for Environmental Chemistry National Research Council of Canada Ottawa Ontario Canada K1A OR9 The direct analysis of solids using ultrasonic slurry electrothermal vaporization inductively coupled plasma mass spectrometry is reported. National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1632a Trace Elements in Coal (Bituminous) Total Diet SRM 1548 and National Research Council of Canada LUTS-1 Lobster Hepatopancreas RM were analysed for a number of elements including Ni Cu Cr Pb Mn and Co.As a consequence of matrix effects most analytes were determined using the method of standard additions although Ni Cu Pb and Mn in NIST coal and Pb in LUTS-1 were successfully determined by external calibration using aqueous standards. With the exception of Cr in the coal sample excellent agreement was obtained between the concentration determined and the certified range. Monitoring of the argon dimer during the high temperature vaporization cycle was shown to be an effective means of assessing matrix effects and selecting calibration strategies for individual analytes. Calculated limits of detection range from 0.07 ng g-' for Co to 3.2 ng g-' for Cr in 2 mg samples.Keywords Slurry sampling; electrothermal vaporization; inductively coupled plasma mass spectrometry; ultrasonic mixing Recent studies'-3 have shown that electrothermal vaporization inductively coupled plasma mass spectrometry (ETV-ICP-MS) is effective for the ultratrace and micro-analysis of a variety of materials ranging from arctic snow' to single zircon^.^ The ETV-ICP-MS technique allows for multi-element analysis of microlitre-sized samples while at the same time taking advan- tage of the very high sensitivity of ICP-MS. The direct analysis of solids and/or slurries by any analytical technique offers advantages over more conventional sample preparation. Among these advantages are the reduced sample preparation time the reduced possibility of sample contami- nation increased sensitivity (no dilution) decreased likelihood of analyte loss through volatilization prior to analysis and the selective analysis of micro-amounts of solids. Slurry sampling offers additional advantages for samples that occur naturally as slurries such as milk and blood and for the completion of surveys where the analysis of large numbers of samples such as contaminated soils is required. Furthermore slurry sam- pling combines the benefits of solid and liquid sampling and permits the use of conventional liquid sample handling appar- atus such as autosamplers.Electrothermal atomic absorption spectrometry (ETAAS) has been successfully applied to the analysis of slurries.&" Bendicho and de Loos-Vollebregt l2 and Miller-Ihli6 have reviewed the available literature and quantified the importance of certain factors such as particle size sample density and analyte Accurate results were obtained for the determination of As Fe Mn and Pb in sediment4 and for the determination of many elements in a wide variety of matrices ranging from g l a d 3 to ~pinach.~ Ebdon et analysed coal using slurry nebulization ICP-MS.Agreement between determined analyte concen- tration and the certified values for the semi-quantitative deter- mination of 67 elements in seven reference coal samples was GSC publication No. 10794. within a factor of two. When a more quantitative approach was used excellent agreement was obtained between deter- mined analyte concentration and the certified range for 16 elements.In other studies Totland et ~ 1 . ' ~ determined the platinum group elements and gold in a number of reference materials using slurries nebulized directly into the plasma of an ICP mass spectrometer. Limits of quantitative analysis for the platinum group elements in samples ranged from 0.04 to 0.2 pg g-'. Huang et a1.I6 used slurry sampling for the determi- nation of Zr V Cr W Mo B and Ti by ETV-ICP-AES. By using a suspension of poly( tetrafluoroethylene) these workers16 were able to improve limits of detection for these elements by factors ranging from 7 to 119 over those obtained by simple volatilization of sample into the ICP atomic emission spec- trometer without the use of a fluorinating agent. The analysis of slurries by ETV-ICP-MS has not been reported in the literature.The characteristics of ETAAS and ETV-ICP-MS techniques have been compared by Gregoire et al.17 These workers have shown that the requirements for successful ETV-ICP-MS analysis are perhaps less demanding than for ETAAS. Complete vaporization of sample and efficient transport of volatilized material to the argon plasma (provided complete atomization occurs within the plasma) are all that is required for ETV-ICP-MS. Electrothermal atomic absorption spectrometry on the other hand requires not only complete vaporization but also complete in situ atomization of sample material. For ETV-ICP-MS breakdown of analyte-containing vapour (molecules) or aerosols followed by atomization and ionization ideally takes place within the argon plasma.The ETV serves only as a means of thermally pre-treating and vaporizing (in whatever form) the sample into a stream of argon flowing into the argon plasma. Electrothermal vaporiz- ation ICP-MS promises an effective technique for the analysis of diverse complex materials for a large number of analytes. The objective of this study was to demonstrate the feasibility of direct analysis of solids by ultrasonic slurry ETV-ICP-MS and to highlight the strengths and weaknesses of this approach.606 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 Experimental Instrumentation A Perkin-Elmer Sciex Elan 5000 ICP mass spectrometer equipped with an HGA-BOOMS electrothermal vaporizer was used. The electrothermal vaporizer system was fitted with a Model AS-60 autosampler equipped with a USS- 100 ultrasonic mixing probe.The ultrasonic mixing probe was constructed of high-purity titanium. Pyrolytic graphite coated graphite tubes were used throughout. The experimental conditions for the Elan 5000 and the HGA 600MS are given in Table 1. Optimization of the USS-100 was accomplished by lowering the probe into an autosampler cup containing sample solution and activating the device. While in operation the power level of the ultrasonic probe was adjusted to provide maximum smooth (rolling) agitation of the sample without spillage. A power level of 30 (12 W) and a mixing time of 30 s were selected for optimum performance of the ultrasonic mixing probe. To facilitate accurate sampling of larger particle sizes the Teflon tip provided on the AS-60 autosampler was replaced with a larger thin-walled Teflon capillary tube (id.0.81 mm). Optimization of plasma and mass spectrometer conditions was accomplished using solution nebulization sample intro- duction and aqueous standards (High Purity Standards Charleston NC USA). The HGA-600MS was interfaced to the argon plasma via an 80cm length of 6mm i.d. Teflon tubing. The operation of the HGA-600MS was completely computer controlled. During the dry and char stages of the temperature programme opposing flows of argon gas (300 ml min-') originating from both ends of the graphite tube removed water and other vapours through the dosing hole of the graphite tube. Prior to and during the high temperature or vaporization step a graphite probe was pneu- matically activated to seal the dosing hole.Once the graphite tube was sealed a valve located at one end of the HGA-600MS workhead directed the carrier argon flow originating from the far end of the graphite tube directly to the argon plasma at a flow rate of 900 ml min-'. Some of the analytes studied are considered refractory (Cr and Ni) and require relatively high temperatures for complete vaporization. Accordingly the ETV heating cycle (Table 1) comprises the highest heating rates available and a high Table 1 Instrumental operating parameters and data acquisition parameters ICP mass spectrometer R.f. power Coolant argon flow rate Intermediate argon flow rate Carrier argon flow rate Sampler/skimmer HGA-600MS Electrothermal vaporizer Sample volume Drying stage (2 second ramp) Internal argon flow rate Charring stage (10 second ramp) Internal argon flow rate Vaporization stage Heating rate Time at maximum temperature Clean-up stage 1000 w 15.0 1 min-' 850 ml min - ' 900 ml min-' Nickel 10 pl 110 "C for 50 s 300 ml min-' 400 "C for 50 s 300 ml min-' 2600 "C 2000 "C s - ' 6 s 2700 "C for 8 s Data acquisition Dwell time 10 ms Scan mode Peak-hopping Isotopes monitored per measurement cycle Signal measurement mode Integrated Points per spectral peak 1 5 vaporization temperature.A high temperature clean-up step (2700 "C) was included as part of the temperature programme to minimize the effects of carry-over particularly from un-vaporized matrix material. Preparation of Slurries A weighed portion of reference material (sample) was placed in a 20ml plastic centrifuge tube. A diluent consisting of 5% Ultrex (Baker) HN03 containing 0.005% Triton X-100 was added to the sample and mixed on a vortex mixer.An aliquot of sample slurry was then immediately withdrawn from the centrifuge tube using an Eppendorf pipette and placed into a clean Teflon auto-sampler cup. In preparing the slurries care was taken to ensure that an appropriate ratio of sample diluent was used by taking into account the particle size and the density of the reference material in order to minimize effects arising from poor sampling statistics."." The coal [National Institute for Standards and Technology (NIST) Standard Reference Material (SRM) 1632a Trace Elements in Coal (Bituminous)] slurry was prepared by mixing 4mg of sample per 4ml of diluent.The density of the coal was approximately 0.7 g cmP3 and the particle size was apparently less than 50 pm. A 20 pl aliquot of sample slurry would require a minimum of 0.08 mg ml-' of coal to give 50 particles ensuring a representative sampling of the bulk slurry.'O*ll The NIST SRM 1548 Total Diet was prepared in a similar fashion except that 400 mg of sample were suspended in 4 ml of diluent. With a density of about 1 g cm-3 and a particle size of less than 250 pm only 20 mg ml-' was required to ensure that each 20 p1 aliquot contained 50 particles. This criterion was met with a concentration of 100 mg ml-'. The National Research Council of Canada (NRCC) LUTS-1 Lobster Hepatopancreas reference material (RM) was provided as 10.3 g of homogeneous slurry.The material was diluted by adding (in a calibrated flask) diluent to a total volume of 100 ml. The suspension was sonicated for 30 min in an ultra- sonic bath and following mixing on a vortex mixer a 1 ml aliquot of the diluted LUTS-1 slurry was removed using an Eppendorf pipette and placed in a Teflon autosampler cup. Although no data are available characterizing the actual particle size of the LUTS-1 material we believe it to be less than 50 pm and certainly less than 100 pm. Looking at the guidelines for minimum only 1.3 mg ml-' was needed for 100 pm particles of a density of 1 g cm-2 to provide the minimum 50 particles per 20 pl analytical sample aliquot. Reagents All acids were produced by sub-boiling distillation in Teflon distillation vessels.Distilled water (18 Mi2 cm) was obtained from a Millipore RO system. For ultrasonic slurry ETV- ICP-MS determinations the addition of 5 pl of NASS-3 Open Ocean Sea Water Reference Material (NRCC) (diluted 500-fold) chemical modifier was added to both standard and sample solutions. The addition of this solution provides 0.7 ng of salt containing Na Cl Mg and Sr which acts as a physical carrier'>" ensuring efficient transport of vaporized analyte from the graphite tube to the argon plasma. Prior to use NASS-3 was purified of trace metals using chelating resinslg and diluted to strength with ultra-pure water. The use of a carrier such as NASS-3 results in both an enhancement in signal and a linearization of calibration curves.20,21 Linear calibration curves were obtained for all analytes from the limit of detection to masses giving a signal intensity of approximately lo6 counts s-' in ICP-MS.Absolute limits of detection in pg (parentheses) for the analytes studied' were Ni (0.47); Cr (3.2); Cu (0.42); Pb (0.086); Mn (0.12) and Co (0.14). The limit of detection is defined as the mass of element which produces a response equivalent to three timesJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 607 the standard deviation of the blank. The blank is the integrated response obtained from the vaporization of 5 pl of NASS-3 carrier. A 1 pg mass of analyte results in an integrated signal of from 1000 to 2000 counts depending on the analyte and the experimental conditions. Signal Measurement and Data Acquisition Fig.1 illustrates typical ETV-ICP-MS signal pulses obtained for the volatilization of 1OOpg of analyte in the presence of NASS-3 carrier. Because of the transient nature of ETV- ICP-MS signal pulses the number of analyte isotopes moni- tored was limited to five to ensure recording of accurate analyte signal pulse shapes. The peak-hopping mode of data acquisition was used with a 10 ms dwell time for each analyte isotope monitored. Single (one m/z per mass) sequential measurements were made for each analyte isotope on a con- tinuous basis from the start of the vaporization cycle until the signal returned to baseline values. When the method of standard additions was used for calibration purposes three additions were used in addition to the unspiked sample.The concentration of added aqueous standard was adjusted to bracket the concentration of analyte present in the sample. Analytical results presented in the tables are the means of five separate determinations for each sample or standard addition. Integrated data (counts) for analyte signal pulses were used for all measurements. Fig. 1 also shows the signal pulse for the argon dimer 40Ar40Ar+. The argon dimer is produced in the argon plasma and behaves in a similar manner to'' any other (analyte) ion in the plasma both in terms of non-spectroscopic interferences and plasma effects. Beauchemin et used the argon dimer as an internal standard to correct for changes in signal intensity due to the presence of concomitant elements and/or signal drift.These workers found that the argon dimer could not be effectively applied to correct interferences and drift over the entire mass range but only for an m/z range from 63-114. We decided to investigate the use of the argon dimer as a semi- quantitative indicator of possible matrix suppression (or enhancement) effects. The argon dimer signal pulse (Fig. 1) shows a large increase in signal about 1.8 s into the high- temperature vaporization step. This 'bulge' in signal is the result of the expansion of gases within the graphite tube during rapid heating. This 'pressure-pulse' reflects a change in the sampling depth at the ICP mass spectrometer interface. The utility of monitoring the intensity of this molecular ion is discussed below. Selection of Analyte Isotopes The selection of analyte isotopes was determined by consider- ing a number of factors including the background spectral 25 I 20 tn *.2 15 X tn c 0 c '0 10 5 0 1 2 3 4 5 6 Time/s Fig. 1 ETV-ICP-MS signal pulses for 100 pg of Pb Cu Ni and Cr features of ETV-ICP-MS the concentration of analyte in the sample the abundance of the analyte isotope and possible molecular ion species produced within the ETV or in the argon plasma which are isobaric with the analyte isotope. Of the six elements studied only 55Mn and 59C0 are monoisotopic. Chromium has three isotopes at m/z 52 (83.76% abundance) 53 (9.55) and 54 (2.38). The major isotope of Cr is interfered with by a significant background molecular ion arising from the formation of 40Ar12C+ precluding its use as an analyte isotope.Chromium analyses were completed using 53Cr. Nickel has five isotopes at m/z 58 (67.77) 60 (26.16) 61 (1.25) 62 (3.66) and 64 (1.16). Of the two major (abundance) Ni isotopes 60Ni was selected as the analyte isotope because the more abundant '*Ni is isobaric with "Fe and some of the materials studied may be high in iron. Copper has two isotopes at m/z 63 (69.09) and 65 (30.91). The major isotope of Cu is free of isobaric interferences from other elements and molecular ions produced from matrix components. The 65Cu however is interfered with by a molecu- lar ion produced from the combination of "Mg+ with 40Ar+ to form an argide in the plasma." The four isotopes of Pb are free of isobaric interferences and each can be used for detection. The abundance (YO) of each of these isotopes is '04Pb (1.37); '06Pb (25.15); '07Pb (21.11) and 'O'Pb (52.37).Because of the relatively large variation in concentration of Pb present in the reference materials studied '07Pb was used for the determination of Pb in the coal '08Pb was used for the determination of Pb in LUTS-1 and '04Pb was used for the determination of Pb in the Total Diet RM. Results and Discussion NIST SRM Coal 1632a Reference Material Coal has a fairly simple matrix comprised primarily of carbon with small quantities of inorganic impurities. The analysis of this material by ultrasonic slurry ETV-ICP-MS was not expected to be difficult since the signal pulse for the argon dimer (Fig. 2) was identical with that observed for a vaporiz- ation cycle carried out in the absence of any matrix (Fig.1). The results presented in Table 2 show that external calibration could be used for the analysis of coal for Cu Ni Pb and Mn. The results obtained for these elements were in agreement with the certified values. The result for the determination of Cr however was low indicating a matrix effect or interference causing inaccuracy. Application of the method of standard additions did not yield the correct concentration value for this element. These results together with the knowledge that ana- lyte signals were probably not perturbed (argon dimer) by the presence of matrix indicate that possibly on reaching the argon plasma coal particles physically transported to the argon 20 15 tn *. 9 10 c 3 0 0 5 0 1 2 3 4 5 6 Time/s Fig. 2 ETV-ICP-MS signal pulses for 20 pg of NIST SRM Coal 1632a608 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 Table 2 Analysis of NIST SRM Coal 1632a Trace Elements in Coal (Bituminous) by slurry sampling ETV-ICP-MS Concentration/pg g - c u Cr Ni Pb Mn External calibration 16.8k0.1 22.7 f 2.8 20.0f 3.5 1 1.0 f 0.5 30.8 & 2.2 Certified 16.5+ 1.0 34.4 f 1.5 19.4f 1.0 12.4f0.6 28k2 plasma are not completely vaporized during the short residence time (1-3 ms) of analyte in the argon plasma. A second explanation which is complementary to the above involves the degree of mobilization of analyte into the diluent solution. Of all the elements determined in the coal sample Cr is the least likely to to be extracted into the mobile phase. In Fig. 2 the Cr analyte signal pulse clearly has the latest appearance time and is also the last element to 'peak' during the high- temperature vaporization cycle.The appearance time is defined as the time elapsed from the on-set of the high-temperature vaporization cycle to the point when the analyte signal is first measurable above background noise. Whereas the other elements were partially or perhaps completely extracted from the solid phase (coal) to the liquid phase (diluent) Cr remained partially trapped in the coal matrix. Analysis using either external calibration or the method of standard additions would give incorrect results. NRCC LUTS-1 Lobster Hepatopancreas Reference Material Fig. 3 shows the slurry sampling ETV-ICP-MS analyte signal pulses for Cr Pb Co and Ni for a 2 mg sample size of LUTS-1 RM.As was shown for coal Pb is the most volatile analyte with the earliest appearance time and Cr is the least volatile with the latest appearance time. The ultrasonic slurry ETV- ICP-MS signal pulse for the argon dimer for LUTS-1 is qualitatively different from that in both the coal slurry and the aqueous standard. At about 0.8 s from the start of the high temperature vaporization cycle there is a small peak corre- sponding to the initial expansion of the argon gas flowing through the graphite tube. At about 2.5 s a second much larger peak maximum occurs indicating some signal enhance- 30 25 v) 20 z *' 15 c 3 s 10 5 0 . . . . . . . . . . . . . . ... Ar . I . . . . . . . . ,'... ... . . . . . . . . .. ' -. Pb . . * . . < . 1 2 3 4 5 6 Time/s Fig.3 ETV-ICP-MS signal pulses for 2 mg of NRCC LUTS-1 Lobster Hepatopancreas reference material ment due to volatilized matrix components.The LUTS-1 RM comprises a highly organic matrix composed of complex proteins carbohydrates and lipids. All of these materials are prone to decomposition during the char step of the heating cycle resulting in the formation of a carbon char composed of carbon chains covering a wide range of molecular weights. This material is relatively refractory and vaporizes only at high temperatures. The use of oxygen ashing to reduce the quantity of carbonaceous material present was tried but the alternate gas valve configuration on the ETV unit did not allow for the use of air during the char step without extinguishing the argon plasma during the vaporization step.The results obtained for the analysis of LUTS-1 by ultrasonic slurry ETV-ICP-MS using both external calibration and the method of standard additions are summarized in Table 3. The results obtained by external calibration do not agree with certified values for any element except Pb. Of the four elements studied only Pb is virtually completely volatilized before the second pulse maximum occurs in the argon dimer signal. Because Pb is vaporized before the major matrix components are volatilized from the graphite tube Pb ions in the argon plasma and the mass spectrometer are not subject to signal alteration due to space-charge or other effects. When the method of standard additions was applied excellent agreement with certified values was obtained for all of the elements studied.NIST Total Diet 1548 Reference Material Fig. 4 shows typical analyte signal pulses for the slurry sam- pling ETV-ICP-MS volatilization of 2 mg of Total Diet SRM. Perhaps the most notable feature of this figure is the signal 50 40 cn *' 30 5 20 X v) + 0 10 0 1 2 3 4 5 6 Time/s Fig.4 ETV-ICP-MS signal pulses for 2mg of NIST SRM 1548 Total Diet Table 3 Analysis of NRCC LUTS-1 Lobster Hepatopancreas by slurry sampling ETV-ICP-MS Concentration/pg g - External calibration Standard additions Certified c o 0.17 f 0.02 0.048 & 0.006 0.051 f 0.006 Cr 0.45 f 0.07 0.081 f0.013 0.079 & 0.012 Ni 0.59 k0.03 0.159 f 0.009 0.200 & 0.034 Pb 0.013 kO.001 0.007 k 0.001 0.0 10 f 0.002JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY pulse for the argon dimer.The Total Diet 1548 SRM is a composite material comprising foodstuffs representative of the total dietary intake for the average person. This sample con- tains large amounts of organic material (as does LUTS-1) in addition to substantial quantities of salts. During the high- temperature vaporization step (following the dry and char step) the first component to volatilize is the salt matrix which corresponds to the virtual total suppression of the argon dimer signal at 1.8 s into the vaporization step. At higher tempera- tures and at later times carbonaceous material is volatilized resulting in some enhancement of the argon dimer signal (3.2 s). All of the analyte signal pulses occur during the volatilization of the salt matrix and therefore all should be subject to interference effects in the plasma and/or mass spectrometer.The analytical results obtained for the diet sample using both external calibration and the method of standard additions are summarized in Table 4. Only one elemental concentration for the analytes studied is certified (Cr) for this material but information values are given in the certificate for both Ni and Pb. Additional values were obtained separately for comparison purposes for Cr and Ni using independent analytical methods. The results in Table 4 show that values obtained for Cr Ni and Pb using external calibration do not agree with certified or analysed values. Only Cu was in agreement with the certified range. Although the Cu was volatilized at the same time as the salt matrix in the sample the concentration of Cu in the sample was very high.It has been shown that the severity of interference effects is related to the molar ratio of analyte to matrix component.22 The higher the ratio the smaller the matrix effect. The much higher ratio of Cu to matrix compared with other analytes studied probably resulted in a Cu+ ion signal which was relatively insensitive to interference effects compared with Cr Ni and Pb. When the method of standard additions was used excellent agreement was obtained between determined concentrations and the certified or reference values for all of the elements studied. Comparison of Sample and Calibration Standard Analyte ETV- ICP-MS Signal Pulses Fig. 5 shows a comparison of analyte signal pulses for Ni obtained from the volatilization of sample and aqueous stan- dard.This figure shows that for the Coal and Total Diet RMs the Ni in the sample is volatilized at a later time than is Ni in the aqueous standard whereas for the Lobster Hepatopancreas RM Ni is volatilized at the same time for both sample and aqueous standards. These results may indicate that for the Coal and Total Diet RMs analyte vaporization is delayed (compared with aqueous standard) because of adsorption of analyte onto coal particles or the delayed vaporization of analyte trapped in a salt melt. All of the elements determined in the diet sample experienced delayed vaporization and all elements determined in the Lobster Hepatopancreas sample were volatilized at the same time as were aqueous standards.Only Ni Cr and Cu experienced delayed vaporization for the coal sample whereas Pb and Mn were not delayed relative to MAY 1994 VOL. 9 609 50 40 r 0 1 2 3 4 5 6 Time/s Fig. 5 Comparison of nickel ETV-TCP-MS signal pulses derived from A reference materials and B aqueous standards (a) NIST SRM 1632a; (b) NIST SRM 1548; and (c) NRCC LUTS-1 Lobster Hepatopancreas aqueous standard. Lead and magnesium were perhaps both quantitatively extracted from the coal matrix by the diluent thus preventing delayed volatilization. These data along with Tables 2-4 show that accurate and precise analytical results can be obtained even if the volatiliz- ation properties of the analyte are altered by the presence of matrix components. Conclusion This study has shown that ETV-ICP-MS can be used success- fully for the direct analysis of slurries and that the technology developed for the analysis of slurries by ETAAS is directly transferrable to ETV-ICP-MS.The use of the argon dimer is demonstrated to be useful in monitoring non-spectroscopic or matrix effects and in selecting the appropriate calibration Table 4 Analysis of NIST SRM 1548 Total Diet by slurry sampling ETV-ICP-MS Concentration/pg g - ' External calibration Standard additions Certified c u 2.1 k0.2 2.9 f 0.2 2.6 f 0.3 Cr 0.28 & 0.02 0.11 k0.02 0.094 f 0.014* Ni 0.75 & 0.02 0.29 f 0.02 (0.41 11- 0.30+0.01$ Pb 0.025 & 0.005 0.045 k 0.006 (0.05)t ~~ * Acid digestion determination by ETAAS. t Certificate information values. 1 Acid digestion solution nebulization ICP-MS determination.610 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 strategy for individual analytes for the analysis of samples containing substantial matrix components. Inductively coupled plasma mass spectrometry offers several advantages for the ETV determination of slurries (over ETAAS) including a multi-element capability and the ability to select different analyte isotopes having different natural abundances to extend the range of concentrations determinable on a single slurry sample preparation. Although the presence of large amounts of salts in the samples lead to serious matrix effects use of the method of standard additions gives results in excellent agreement with certified values except for Cr in coal. For refractory materials such as coal the partitioning of analyte between solid and liquid phases is important. Using reported ETV-ICP-MS limits of detection’ for the analytes studied the following limits of detection in ng g-l can be estimated for a slurry sample size of 2mg Co 0.070; Cu 0.21; Cr 3.2; Mn 0.060; Ni 0.24; and Pb 0.012.Clearly the future of ETV-ICP-MS applied to the analysis of slurries is promising. Further work is required on the study of chemical modifiers which could either be used to remove matrix components and/or stabilize analytes such that vaporiz- ation of analyte occurs after matrix components have left the graphite tube. The analysis of organic materials could be improved with the use of air or oxygen ashing to assist in the removal of carbon derived from sample material prior to the high temperature volatilization step.The authors are grateful to the Perkin-Elmer Corporation for the loan of the HGA-600MS and the USS-100. References 1 2 Sturgeon R. E. Gregoire D. C. Willie S. N. Zheng J. and Kudo A. J. Anal. At. Spectrom. 1993 8 1053. Gregoire D. C. and Lee J. J. Anal. At. Spectrom. 1993 9 393. 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Grkgoire D. C. Ansdell IS. Goltz D. M. and Chakrabarti C. L. Chem. Geol. in the press. Epstein M. S. Carnrick G. R. Slavin W. and Miller-Ihli N. Anal. Chem. 1989 61 1415. Bin H. Zucheng J. and Yun’e Z. J. Anal. At. Spectrom. 1991 6 623. Miller-Ihli N. J. Anal. Chem. 1992 64 964A. Miller-Ihli N. J. J. Anal. At. Spectrom. 1988 3 73. Miller-Ihli N. J. J. Anal. At. Spectrom. 1989 4 295. Miller-Ihli N. J. Fresenius’ J. Anal. Chem. 1990 337 271. Miller-Ihli N. J. At. Spectrosc. 1992 13 1. Miller-Ihli N. J. Fresenius’ J. Anal. Chem. 1993 345 482. Bendicho C. and de Loos-Vollebregt M. T. C. J. Anal. At. Spectrom. 1991 6 353. Bendicho C. and de Loos-Vollebregt M. T. C. Spectrochim. Acta Part B 1990 45 695. Ebdon L. Foulkes M. E. Parry H. G. M. and Tye C. T. J. Anal. At. Spectrom. 1988 3 753. Totland M. Jarvis I. and Jarvis K. E. Chem. Geol. 1993 104 175. Huang M. Jiang Z. and Zeng Y. J. Anal. At. Spectrom. 1991 6 221. Gregoire D. C. Lamoureux M. Chakrabarti C. L. Al-Maawali S. and Byrne J. P. J. Anal. At. Spectrom. 1992 7 579. Gregoire D. C. and Sturgeon R. E. Spectrochim. Acta Part B 1993 48 1347. Sturgeon R. E. Berman S. S. Willie S. N. and Desaulniers J. A. H. Anal. Chem. 1981 53 2337. Gregoire D. C. Al-Maawali S. and Chakrabarti C. L. Spectrochim. Acta Part B 1992 47 1123. Ediger R. D. and Beres S. A. Spectrochim. Acta Part B 1992 47 907. Gregoire D. C. Spectrochim. Acta Part B 1987 42 895. Beauchemin D. McLaren f. W. and Berman S. S. Spectrochim. Acta Part B 1987 42 467. Paper 3/068223 Received November 15 1993 Accepted January 10 1994

ISSN:0267-9477    

DOI:10.1039/JA9940900605

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 61 1 Determination of Arsenic Chromium Selenium and Vanadium in Biological Samples by lnductively Coupled Plasma Mass Spectrometry Using On-line Elimination of Interference and Preconcentration by Flow Injection Les Ebdon Andrew S. Fisher and Paul J. Worsfold Plymouth Analytical Chemistry Research Unit Department of Environmental Sciences University of Plymouth Drake Circus Plymouth Devon PL4 8AA UK A method of matrix elimination has been developed that facilitates the determination of analytes such as arsenic chromium selenium and vanadium in biological matrices by inductively coupled plasma mass spectrometry (ICP-MS) without interference from polyatomic ions. The method involves the retention of the analytes as anions on activated alumina (acidic form) in a microcolumn using an on-line flow injection system with simultaneous matrix removal.Analysis of certified reference materials [NlES (National Institute of Environmental Studies) 9 Sargasso Tort-1 Lobster Hepatopancreas Dorm-1 Dogfish Reference Muscle and Dolt-1 Dogfish Liver Tissue] yielded results in good agreement with the certified values although a photolysis step was used for determinations of arsenic in animal based samples in order to destroy organoarsenic compounds. Tests showed close to 100% recovery for all analytes. The limits of detection (3an-,) using a 200 1.11 sample loop were 1.2 6.0 9.0 and 65 ng ml-' for vanadium chromium arsenic and selenium respectively. A preconcentration system was developed for selenium because of the relative insensitivity of ICP-MS to this element.This yielded a detection limit of 1.0 ng ml-' at m/z 78. Keywords lnductively coupled plasma mass spectrometry; flow injection matrix removal; preconcentration; activated alumina; biological matrices Inductively coupled plasma mass spectrometry (ICP-MS) is a technique that although being fairly interference free still suffers from several troublesome interferences the majority of which are derived from polyatomic ions.' One of the most common interfering species is chloride. Various chlorine- containing ions interfere with a number of analytes of particular biological and environmental importance e.g. 35C1160 + on vanadium at m/z 51 35Cl'60H+ on chromium at m/z 52 37C1160+ on chromium at m/z 53 40Ar35Cl+ on arsenic at m/z 75 and 40Ar37C1+ on selenium at m/z 77.For some of these analytes the interference is less severe because it occurs on a relatively minor isotope e.g. '3Cr but for others e.g. 75A~ which are monoisotopic the interference can be extremely troublesome. Other interferences arising from either biological matrices e.g. 40Ar12C+ on chromium at m/z 52 or from the plasma gases e.g. 38Ar38Ar + 40Ar38Ar + and 40Ar40Ar + on selenium at m/z values of 76 78 and 80 respectively also exist. These interferences make some of the first row transition metals and the metalloids arsenic and selenium particularly difficult to determine. The interferences arising because of the presence of chloride ions have been alleviated in a number of ways. The addition of a molecular gas to the nebulizer gas the introduction of a small amount of an organic solvent2 and size-exclusion chromatography' have all been used successfully. A review has been published of chromatography coupled with atomic spectrometry,6 which highlights methods of analyte preconcentration and matrix elimination. In another paper,7 cationic analytes were separated from interfering ions in the matrix by retaining the ions of interest on a microcolumn of chelation-exchange resin.This approach worked well for a number of analytes but others for example arsenic chromium and selenium were not retained completely on this column. Analytes such as these once oxidized by potassium persulfate to higher oxidation states could be retained on an anion- exchange system.Activated alumina has been used previously to retain both anions (acidic and cations (basic form)." In the present paper a method is described which utilizes an activated alumina packed microcolumn to separate and if necessary preconcentrate the analytes of interest from the interfering species present in the matrix which are flushed to waste by the use of an appropriate buffer and switching valves in a typical flow injection manifold. The method facilitates the analysis of biological materials. Experimental Chemicals Tris buffer (1 moll- ') at pH 9 was prepared by dissolving 120 g of Tris( hydroxymethy1)methylamine ( AnalaR Merck Poole Dorset UK) in water and diluting to almost 11. Nitric acid (Fisons PrimaR Loughborough Leicestershire UK) was used to adjust the pH to 9 and the solution was then diluted to volume.The exchange medium used was activated alumina (acidic form Brockmann Grade 1 Merck). Sodium hydroxide (Aristar) and the 1000 pg ml-' stock standards (SpectrosoL) were also obtained from Merck. Potassium persulfate (BioChemika Microselect) and hydrogen peroxide ( puriss) were obtained from Fluka (Fluka Chemie Buchs Switzerland). Instrumentation A flame atomic absorption spectrometer (IL 151 Thermo- Electron Warrington Cheshire UK) was used for some of the preliminary experiments. Most analyses were performed on an ICP-MS instrument (PlasmaQuad PQ2 VG Elemental Winsford Cheshire UK). The flow injection manifold was made in-house and consisted of two valves (Rheodyne Type 50) a peristaltic pump (Gilson Minpuls 3 ) and poly(tetra- fluoroethylene) (PTFE) tubing (0.8 mm i.d.).Preparation of Columns The activated alumina (1 g) was cleaned by boiling with nitric acid (6 mol I-') for 5 min and then rinsing with water. Portions of the cleaned alumina (approximately 0.15 g) were introduced into a glass tube (2.5 cm x 3 mm id.) by a syringe. Plugs of glass wool at either end ensured that the alumina remained in the column.612 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 Sample Preparation Biological samples (0.5000 g) were weighed accurately into PTFE digestion bombs (Savillex NJ USA) and nitric acid (3 ml) and hydrogen peroxide (2 ml) were added. After leaving for 16 h (overnight) the mixtures were placed in a microwave oven and irradiated at medium power (approximately 300 W) for 3 min.After cooling the pressure was released and potass- ium persulfate (0.1 g) was added. The bombs were re-sealed and the digests were microwaved at medium power for a further 2 min. After cooling the digests were diluted using tris with sodium hydroxide (3 ml 60% m/v) added to assist the buffer. The digests were then diluted to volume (50 ml) by the addition of more buffer. Blanks were prepared in a similar fashion but omitting the sample. Procedure Initial experiments to ensure that chromium would be retained on the column were performed using a flame atomic absorption spectrometer. A standard solution of chromium [ S pg ml-I of chromium(II1) nitrate] containing 6% v/v nitric acid 6% v/v 60% sodium hydroxide and 0.2% m/v potassium persulfate was prepared.The standard was diluted to volume using tris buffer. For all work (unless otherwise stated) a sample volume of 200 pl was injected onto the column and the analytes were eluted using a similar volume of nitric acid. A schematic diagram of the equipment used is shown in Fig. 1. The mobile phase used throughout was water. The operating conditions used are detailed in Table 1. Chromium was eluted from the column using nitric acid (1 moll-'). The results indicated that chromium was retained on the column and that there was no discernible breakthrough at this relatively high concentration. Elution proved to be quantitative. Results and Discussion Having demonstrated that the chromium was retained by the column it was decided to transfer the methodology to the ICP-MS instrument so that the four analytes could be deter- mined virtually simultaneously using the time-resolved software on the instrument.The operating conditions used on the ICP-MS instrument are detailed in Table 2. The isotopes at which the elements were determined are as follows 51V; 52Cr and 53Cr; 75As; 77Se 78Se and 82Se; and '151n (internal standard). The time-resolved software is a package which is available on VG instruments that allows a number of analyte isotopes to be monitored with respect to time. This was performed using the peak-jumping mode. Peak areas were used to construct calibration curves. Initially an interference study was performed in which the effects of sodium chloride (1 YO m/v) on a series of vanadium standards were determined.The experiment was performed in four modes i.e. with and without sodium chloride and with Samplehitric acid Waste Fig. 1 Schematic diagram of the separation system Table 2 Operating conditions for the determinations using ICP-MS Detection mode Channels Sweeps Dwell time/ps Gas flow/l min-' Outer Intermediate Aerosol Forward powerw Sample uptake rate/ml min- Nebulizer type Spray chamber Time-resolved analysis 2048 100 320 14.0 1 .o 0.9 1500 1.5 Ebdon Double pass water cooled 1 and without a column present. In all cases the solutions contained nitric acid sodium hydroxide potassium persulfate and buffer. For the experiments in which no column was present valve 2 in Fig. 1 was orientated such that the solutions were directed to the detector. When a column was present valve 2 directed the matrix to waste and was then switched so that the analyte could be eluted for detection.For the ICP-MS determinations the nitric acid eluent contained 100 ng ml-1 of indium which acted as an internal standard. The results obtained are shown in Fig. 2. It can be seen that in the absence of a column the presence of chloride interfered severely with the determination of vanadium but that this interference was reduced substantially when a column was used. Potassium persulfate acts as an oxidizing agent which converts the analytes into oxyanions e.g. Cr"' will be oxidized to CrV1. In turn the persulfate is converted into sulfate. The sulfate then competes with the chloride ions for the active sites on the alumina. Since the sulfate and the analyte oxyanions are both doubly charged and the chloride is only singly charged the chlorides pass through the column to waste.Evaluation of the Method Having demonstrated that the presence of the persulfate enables the separation of chloride ions from the analytes it was necessary to validate the process using certified reference materials (CRMs). The CRMs used were NIES 9 Sargasso (National Institute for Environmental Studies Japan) and Tort-1 Lobster Hepatopancreas Dorm-1 Dogfish reference muscle and Dolt-1 Dogfish Liver Tissue (National Research Council Canada). These were digested using the procedure described earlier and then analysed. The results for the analyses are shown in Table 3. Selenium proved impossible to determine directly because of the inherent insensitivity of ICP-MS to this element arising from the fact that its abundance is split over six isotopes and because it has a very high ionization potential.Although Table 1 Operating conditions used for the determination of chromium by flame atomic absorption spectrometry using a fuel rich flame 0 200 400 600 800 1000 1200 [Vl/ng mi- Wavelength/nm Spectral bandpass/nm Lamp current/mA Sample uptake rate/ml min-' 357.9 0.5 5 3 Fig.2 Effect of sodium chloride on the vanadium signal at m/z 51. A With 1 % m/v C1- ' no column; B with 1 % m/v C1- ' with column; C without added chloride no column; and D without added chloride with column. (As the three approaches give very similar results lines B C and D are practically indistinguishable)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 613 Table 3 Results obtained for the analysis of certified reference materials; n = 4 Material NIES 9 Sargasso Tort-1 Lobster Hepato pancreas Dorm-1 Dogfish Reference Muscle Dolt-1 Dogfish Liver Tissue Analyte V Cr As Se v Cr As Se Cr As Se Cr As Se Result obtained/pg g-' 1.1 *0.1* 0.4f0.1 l l l f 5 NDS 1.2k0.3 2.0 + 0.4 9.3 * 2.0 ND 3.90 & 0.60 6.2 f 2.0 ND ND 3.9+ 1.0 ND Certificate value/pg g-I 1.0 & 0.1 (0.2)t 115+9 (0.05) 1.4 k0.3 2.4 t- 0.6 24.6 & 2.2 6.88 k0.47 3.60 k 0.40 17.7f2.1 1.62 k0.12 0.40 k 0.07 10.1 k 1.4 7.34 + 0.42 * *95% Confidence limits. t Values in parentheses indicate reference values only. $ ND Not determined. calibration curves could be prepared they tended to be 2-3 orders of magnitude higher than the selenium content of the real samples and no biological reference material could be found that contained selenium at the 1OOpgg-I level.A preconcentration procedure was therefore required. This is discussed in a later section. The results for vanadium and chromium showed reasonable agreement with the certified values. However the results for arsenic were erratic. The arsenic value obtained in the Sargasso material showed close agreement with the certified value but the results for the determinations in the fish and shellfish products were low. This was presumably because the arsenic in these samples was present as the extremely stable compound arsenobetaine which is known to withstand many acid digestion procedures.''*12 Several methods have been used previously to destroy arsenobetaine.These include the use of perchloric acid sulfuric acid12 and potassium persulfate. The use of perchloric acid was precluded because of the desire to avoid elevated concentrations of chlorine and because of the safety hazard of using it in a microwave oven. Similarly the use of sulfuric acid could lead to excessive pressure build-up within the microwave bomb. The use of slightly less vigorous reagents such as a potassium persulfate-nitric acid-hydrogen peroxide mixture seemed partially successful in destroying the arsenobetaine but still did not enable 100% recovery to be obtained. An alternative method reported in the literature has been the use of photolysis by ultraviolet (UV) light.'' These workers irradiated sample digests for 1 h using a 1200 W lamp.Other workers have used weaker lamps but longer irradiation period^.'^ It was therefore decided that UV irradiation should be used in an attempt to decompose the arsenobetaine. Use of UV Irradiation to Decompose Arsenobetaine For this study a standard of 100 pg ml-' of arsenic as arsenob- etaine was used. The standard (0.5ml) was digested in the normal way and then transferred quantitatively into a silica tube. Hydrogen peroxide (1 ml) was added and the tube was inserted into an aluminium foil lined box containing a 150 W xenon lamp and irradiated for 8 h. As a comparison a second digest was prepared but this was not irradiated. A control standard of arsenobetaine (1 pg ml-') with the pH adjusted to 9 was also prepared.The results of the analysis of these solutions indicated that arsenobetaine was not retained on the column so passed to waste with the matrix. The digestion procedure alone was found to destroy approximately 28% of the arsenobetaine but the digestion procedure combined with the photolysis destroyed almost 100%. Therefore photolysis was used routinely after the digestion procedure for all subsequent analyses The certified reference materials were reanalysed using this modified preparation procedure. The results of the analyses are given in Table4 and are in good agreement with the certified values. Preconcentration Procedure for Selenium By passing the standards and samples through the column at a flow rate of 2 ml min-' for 5 min an effective volume of 10 ml was preconcentrated.The column was then washed with water for 1 min in the normal manner. The sample loop (100 pl) on valve 1 was then filled with acid and the analytes eluted for detection. In this way a preconcentration factor of 100 should have been achieved. The results of the analysis of a certified reference material are shown in Table 5. Again good agreement with the certified value was obtained although the uncertainty was comparatively high. This was because the determination was still fairly close to the limit of detection. Limits of Detection Limits of detection of the analytes were estimated by preparing multiple (n = 7) blanks using the entire digestion procedure and then determining the concentration equivalent to three times the standard deviation of the analyte signal near to the limit of detection. The results obtained are presented in Table 6.The results for selenium with and without the preconcentration are given. Recovery Tests Although it had already been demonstrated that fairly accurate results could be obtained it was still necessary to determine the recoveries from spiked samples. This was performed in the following manner. A sample (0.5000g) of the Tort-1 Lobster Hepatopancreas was spiked with an aliquot (300 pl) of an aqueous standard containing chromium selenium and vanadium at 10 pg ml-' and 300 pl of arsenic at 100 pg m1-I. The spikes were allowed to soak into the sample for 24 h and then the samples were digested and diluted to 50ml in the normal manner. The levels of the spikes in the digests were therefore 60 ng ml - ' of chromium selenium and vanadium and 600ngml-' of arsenic.The digests were then analysed and the results obtained were vanadium 98 & 4; chromium 106 f 6; arsenic 103 4; and selenium 109 & 12%. Table 4 Results obtained for the determination of arsenic in certified reference materials after photolysis Result obtained/ Certified value/ Material Pg g- Pg g-' Dorm-1 Dogfish Tort-1 Lobster Reference Muscle 17.3 + 2.0 17.7f2.1 He pato pancreas 22.3 t- 1.8 24.6 + 2.2 Table 5 Results obtained for the determination of selenium in certified reference materials after preconcentration Result obtained/ Certified value/ Material Pg g- ' Pg g-' Tort-1 Lobster Hepatopancreas 6.55 & 1.25 6.88 f0.47 Liver Tissue 7.47 k0.61 7.34 k0.42 Dolt-1 Dogfish6 14 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 Table 6 Limits of detection for the analytes Flow injection solution detection limit ( ~ c T ~ - ~ ) / Analyte Isotope ng ml-' v 51 1.2 Cr 53 6 As 75 9 Se 78 65 Set 78 1 (Corresponding detection limit in the solid sample ( ~ C T " - ~ ) / Detection limit for normal solution introduction (3gm-')/ Pg g-l* ng ml-' 0.12 0.10 0.6 0.5 0.9 1 .o 6.5 4.0 1 .o - * For a 1% m/v digest. After preconcentration. Conclusions Arsenic chromium and vanadium can be determined in bio- logical and food matrices after acid digestion in the presence of hydrogen peroxide and potassium persulfate. The pH of the sample was adjusted to 9 and the solution passed through an activated alumina column. Interfering chloride ions were elim- inated and passed to waste because of the presence of sulfate.The analytes retained on the column could then be eluted for detection using nitric acid. If the sample is animal based a further digestion step is required to destroy arsenobetaine. This step involves UV irradiation of the digested sample. For selenium a preconcentration step was required to obtain the detection limits necessary. Despite having a 100-fold pre- concentration step the uncertainty of the analysis was still fairly high because the determinations were still close to the detection limits. For all analytes the procedure was validated using certified reference materials. Reasonable agreement was obtained with the certified values. Tests showed close to 100% recovery for all the analytes in a spiked sample. The authors acknowledge the financial support of the Ministry of Agriculture Fisheries and Food Colney Lane Norwich UK. 1 2 3 4 5 6 7 8 9 10 11 12 13 References Munro S. Ebdon L. and McWeeny D. J. J. Anal. At. Spectrom. 1986 1 211. Evans E. H. and Ebdon L. J. Anal. At. Spectrom. 1989 4 299. Hill S. J. Ford M. J. and Ebdon L. J. Anal. At. Spectrom. 1992 7 719. Branch S. Ebdon L. Ford M. Foulkes M. and O'Neill P. J. Anal. At. Spectrom. 1991 6 151. Lyon T. D. B. Fell G. S. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988 3 601. Ebdon L. Fisher A. S. Hill S. J. and Worsfold P. J. J. Autom. Chem. 1991 13 281. Ebdon L. Fisher A. S. Worsfold P. J. Crews H. M. and Baxter M. J. Anal. At. Spectrom. 1993 8 691. Cook I. G. McLeod C. W. and Worsfold P. J. Anal. Proc. 1986 23 5. Cox A. G. McLeod C. W. Miles D. L. and Cook J. M. J. Anal. At. Spectrom. 1987 2 553. Cox A. G. and McLeod C. W. Anal. Chim. Acta 1986,179,487. Cullen W. R. and Dodd M. Appl. Organornet. Chem. 1988 2 1. Cullen W. R. and Dodd M. Appl. Organomet. Chem. 1988,3,79. Stringer C . E. and Attrep M. Anal. Chem. 1979 51 731. Paper 31061 14K Received October 13 1993 Accepted January 4 1994

ISSN:0267-9477    

DOI:10.1039/JA9940900611

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 615 Determination of the Growth Promoter 4-Hydroxy-3-Nitrophenyl- Arsonic Acid in Chicken Tissue by Coupled High-performance Liquid Chromatography-Inductively Coupled Plasma Mass Spectrometry John R. Dean* Les Ebdont and Michael E. Foulkes Plymouth Analytical Chemistry Research Unit University of Plymouth Drake Circus Plymouth Devon UK PL4 8AA Helen M. Crews and Robert C. Massey Ministry of Agriculture Fisheries and Food Food Safety Directorate Food Science Division Norwich Research Park Norwich UK NR4 7UQ A method has been developed to measure the levels of the growth promoter 4-hydroxy-3-nitrophenylarsonic acid (roxarsone) in samples of tissue from chickens fed on a diet supplemented with this compound. Extracts of the tissue were prepared for analysis by a trypsin enzymolysis digestion technique and bulk matrix separation was performed by anion-exchange column chromatography. Further separation of the roxarsone from other species was achieved using reversed-phase high-performance liquid chromatography.Detection of the arsenical compound was by inductively coupled plasma mass spectrometry. No roxarsone was detected in chickens fed on the supplemented diet with or without a withdrawal period (7 d on base feed after the administration period) at a quantification limit of 25 ng g-'. Recoveries for roxarsone from spiked (standard additions) samples of chicken tissue varied between 85 and 103%. Keywords High-performance liquid chromatography; inductively coupled plasma mass spectrometry; arsenic speciation; 4-hydroxy-3-nitrophenylarsonic acid; chicken tissue; food analysis The metabolism of roxarsone and its toxicity effects have been the basis of a number of studies.In general phenylarsonic compounds are rapidly metabolised and mostly excreted by the urinary system in poultry and other domestic animals when administered parenterally. When given orally a consider- able percentage is excreted in the faeces. After absorption by the gastrointestinal tract 50-75% of the material is excreted within 24 h. Of the 25% remaining excretion is much slower and can take 8-10 d.',' Limited biotransformation appears to occur for the nitrated phenylarsonic acids fed to p o ~ l t r y ' ~ ~ and one minor metabolite detected in the excretion together with roxarsone was 3-amino-4-hydroxyphenylarsonic acid.2 The retention of this particular arsenic species may in part have been responsible for the elevated total arsenic levels in poultry fed roxarsone-supplemented Experimental Reagents Roxarsone (Phase Separations Queensferry Clwyd UK) was prepared at a concentration of 1000 pgml-' as arsenic in doubly de-ionized water and serially diluted prior to use.The roxarsone standard required warming for complete dissolution. A fresh roxarsone standard was prepared each month. All other reagents were of Aristar grade (Merck Poole Dorset UK) where possible. Chicken Samples Chickens reared from 7 d of age on a base ration diet or a diet fortified with roxarsone (36 pg g-' as roxarsone) were analysed. The base diet consisted of 46% wheat 26% soya products 21% barley 5% fish meal and 2% other minor components.The fortified diet was prepared by successive mixing of a 10% roxarsone pre-mix with the base diet. The chickens 30 in total were fed according to the following scheme. Ten chickens (six * Present address Department of Chemical and Life Sciences University of Northumbria at Newcastle upon Tyne UK NE1 8ST t To whom correspondence should be addressed. males four females) were fed a base ration and 20 chickens were fed with the ration supplemented with roxarsone at 10.6 pg g-' (as arsenic). After 49 d on their respective diets the ten chickens fed on the base ration and ten chickens (six males four females) fed on the supplemented ration were slaughtered. The remaining ten chickens (five males five females) were then fed on a base ration for a further 7 d (withdrawal period) prior to slaughter.The mean mass (g) and standard deviation of the chickens fed on base supplemented and supplemented plus 7 d base ration were 2147f315; 2265 & 327; and 2644 f 422 respectively. The breast and leg meat of the chickens was removed homogenized and frozen until required. Roxarsone Extraction Procedure The procedure employed for the extraction of roxarsone from chicken samples was based on a method similar to the simu- lated digestive enzymolysis procedure developed by Crews et a1.- The extraction of roxarsone is based on the enzyme activity of trypsin at pH 8.0 with homogenized chicken tissue. Approximately 1.0 g of homogenized chicken tissue was accu- rately weighed into a glass homogenizer.To this were added 100 mg of trypsin (Type 111 Sigma Chemical Company Poole Dorset UK) and 10 cm3 of 0.1 mol dm-3 ammonium bicarbon- ate solution (Aristar grade Merck) resulting in a solution of pH 8.0. Where necessary a standard roxarsone spike of 50-200ng (as arsenic) was added at this point to the glass homogenizer. Using a poly( tetrafluoroethylene) (PTFE) pestle the sample was dispersed in the bicarbonate-trypsin solution and then quantitatively transferred into a dry plastic boiling tube (previously cleaned in 10% Aristar nitric acid and doubly de-ionized water). The boiling-tube was then covered and held for 4 h in a shaking water-bath maintained at 37 "C. The sample tube was centrifuged at 11 x lo3 rev min-l (iso- thermally at 20 "C) for 20 min in order to separate solids from the supernatant.The solution was then decanted into acid- cleaned dry 25 cm3 beakers carefully avoiding transfer of the fatty mass which was also present. The pH of the supernatant liquid was slowly adjusted to 4.0 using sulfuric acid (10% Aristar grade) from a glass micropipette. At this pH the616 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 roxarsone is in the anionic form. The supernatants were covered and stored overnight in a fridge at 5°C. From the solution 2.0 g aliquots were passed through three strong anion- exchange columns (Bond Elut Jones Chromatography Glamorgan UK) connected in series. At pH 4 non-ionic and cationic species pass through but the anionic roxarsone form is strongly retained on the strong anion-exchange column.The columns were washed sequentially with 5 x 2 cm3 portions of 0.01 mol dmW3 acetate buffer solution (pH 4 prepared from 0.005 mol dmP3 sodium acetate and acetic acid (Aristar grade)] and 2 x 2 cm3 portions of doubly-de-ionized water. Acid cleaned dry 25 cm3 beakers were placed under the Bond Elut columns and the roxarsone was eluted using 4 x 1 g portions of 0.1 mol dm-3 orthophosphoric acid (Aristar grade) whose pH had previously been adjusted to 0.7 using sulfuric acid. At this low pH the roxarsone was eluted quantitatively. The eluent (4.0 g in mass) was syringe filtered through a 0.22 pm filter (Millex-GV Millipore SA Harrow Middlesex UK) into acid-cleaned dry 25 cm3 beakers and covered. Conditioning and cleaning of the columns as described above was found to be critical.Determination of Roxarsone The roxarsone content of the low pH extract was determined by directly coupled reversed-phase high-performance liquid chromatography-inductively coupled plasma mass spec- trometry (HPLC-ICP-MS). Reversed-phase chromatography was used since at the pH of the mobile phase (pH2.5) the roxarsone is in the non-ionic form and the polar character of this species can be used to separate it from other non-ionic arsenic species. The mobile phase was 5% methanol-95% v/v 0.01 mol dm-3 orthophosphoric acid (both kristar grade Merck) which was pumped at a flow rate of 1.0cm3min-' using an HPLC Pump (Model 6000A Solvent Delivery System Waters Milford MA USA). The mobile phase passed through a 175mm3 loop fitted to an injection switching valve (Rheodyne 7125 fitted with an R175 loop Rheodyne Cotati CA USA) on to the column.The column used was a reversed- phase type PEP RPC HR 5/10 plastic coated column (Pharmacia Milton Keynes UK). This was directly coupled to a concentric glass nebulizer within the water-cooled spray chamber of the ICP mass spectrometer (VG Plasmaquad 2 VG Elemental Winsford Cheshite UK). The spray chamber was cooled to 10°C. This instrument was tuned to the mass of monoisotopic arsenic (75As) using a continuously delivered roxarsone solution with a concentration of 50 ng ~ m - ~ to effect the best response prior to any series analysis. Typical operating conditions of the ICP mass spectrometer are shown in Table 1.Samples of the filtered roxarsone-containing extracts were loaded into the 175 mm3 loop and were then injected on to the column at the same time as the single-ion monitoring programme (VG Elemental Winsford Cheshire UK) was initiated. Injections of standard roxarsone solutions (1-20 ng cm-3 As) were used to calibrate the instrument. Peak area measurements were used throughout. Quantification was based upon spiked chicken samples taken through the whole extraction procedure. These were prepared in the low pH orthophosphoric stripping solution employed to extract roxar- sone from the Bond Elut anion-exchange columns. Checks on roxarsone recovery were made using the extracts from spiked samples of base fed only chicken tissue. Chromatograms of two injections of a spiked and unspiked tissue are presented in Fig.1. The absolute retention time for roxarsone was dependent upon the condition of the column and the lengths of the PTFE tube connections from the rheodyne head to the column and the column exit to the nebulizer. For ten separate injections the mean time from injection to the roxarsone peak was 4.10+0.04 min. The relative retention time i.e. the time between solvent front and roxarsone response was usually in the range of 2.9 to 3.2min. A comparison of the standard Table 1 Reversed-phase HPLC-ICP-MS conditions used for the determination of roxarsone in chicken tissue HPLC Column Mobile phase Liquid flow rate/cm3 min-' Pump Injection valve Loop volume/mm3 ICP-MS Forward power/W Reflected power/W Gas flow rates/dm3 min-l Coolant Intermediate Carrier Nebulizer Spray Chamber Data Collection Single ion monitoring Mass/u Dwell time/ps Channels Scan time/s PEP RPC HR 5/10 (Pharmacia) 5% methanol-95% 0.01 mol dm-3 1 .o Waters Model 6000A Solvent Rheodyne Type 7125 fitted with a 175 reversed phase orthophosphoric acid Delivery System R175 loop VG Elemental PlasmaQuad PQ2 1300 < 10 13 0.8 0.82 Concentric glass Meinhard Water cooled 'Scott' type as nebulizer T-30 supplied by VG Elemental 75 163840 2501 409.8 roxarsone solutions with the frequent injection of extracts from spiked and unspiked chicken tissue digests ensured that only roxarsone was being monitored in the sample matrix.Periodic injections of blank and stripping solution was also performed to check that roxarsone was not being retained on or stripped from the column.Results and Discussion The total level of arsenic in the chicken tissue was measured after acid digestion by hydride generation atomic absorp- tion spectrometry. For chickens fed the base ration with no supplementation the arsenic content of the leg tissue was 0.172 k0.047 pg g-' fresh mass and of the breast tissue 0.075 + 0.022 pg g-'. For the chickens fed the supplemented ration for the whole period the arsenic content of the leg tissue was 0.257 + 0.049 pg g-' and the breast 0.134 & 0.036 pg g-'. For the chickens fed the supplemented ration but then the base ration for the final 7 d before slaughter the arsenic content of the leg tissue was 0.187+0.052 pg g-' and of the breast 0.090+0.035 pg g-'. In each case the figures are given as pg g-I of arsenic and on a fresh mass basis the mean is for ten different chickens and uncertainties quoted are +1 standard deviation.The results of the roxarsone recovery experiments are shown in Table 2. These show that a peak (RT 1.2 min) was seen for all samples at the solvent front including the blank taken through the procedure. The extraction procedure was found to account for the majority of the first peak response. No peak was seen at the retention time for roxarsone in the blanks trypsin only and non-spiked roxarsone free fed chicken extracts. Recovery values for roxarsone spikes added to chicken tissue and taken through the procedure were in the range 85-103%. The presence of a larger peak at the solvent front in most chromatograms was indicative of chloride interference.This interference is well do~umented'~'~ and is due to the formation of 40Ar35C1+ the m/z of which is coincident with that of 75As. This was verified by monitoring the response of m/z 77 from the polyatomic 40Ar37C1+. Based on the isotopic ratios of 35C1:37C1 (75:25) it was found that chloride from the procedureJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 I I I I I I I I I I 1 I I 617 Fig. 1 Successive chromatograms for two injections (the time of injection is indicated by an arrow) for (a) a tissue extract containing a roxarsone spike (200 ng of As as roxarsone added to 1 g of chicken tissue and then taken through the extraction procedure). The tissue was a sample of chicken leg tissue (supplemented feed plus final 7 d on base feed); (b) a similar tissue extract with no added roxarsone spikes (1 g of chicken leg); (c) expansion of (b) to show baseline response and absence of quantitative roxarsone peak from chicken tissue where the supplement was withdrawn 7 d before slaughter Table 2 roxarsone recovery from samples taken through the extraction procedure Presence of peaks roxarsone Chicken Trypsin spike (200 ng) used?* added? added? No No No No No Yes No Yes No N O Yes Yes Yes Yes No Yes Yes Yes 5 ng ~ r n - ~ standard roxarsone injection.$ First peak As+/ArCl+ detected? Yes Yes Yes Yes Yes Yes No Second peak roxarsone detected? No Yes No Yes No Yes Yes Recovery (YO) from spike (200 ng) No spike added No spike added No spike added No spike added 98-99 9 1-99 85-103f * Chicken sample used has been base fed only (no roxarsone supplement in diet).t Range covered in all spiked chicken samples. $ Standard injection (not taken through procedure). may account for between 70 and 90% of the response of the first peak observed at the solvent front. No response was seen at m/z 77 at the retention time for roxarsone. When added other arsenic species were found to elute before roxarsone. Roxarsone eluted approximately 3 min after the solvent front whereas arsenite arsenate arsenocholine and arsenobetaine eluted 1-1.5 min after the solvent front and monomethylarsinic acid dimethylarsonic acid and arsanilic acid eluted up to 2min after the solvent front. Results of the roxarsone determination experiments for spiked and unspiked samples of tissue from chickens fed on roxarsone taken through the extraction procedure are shown in Table 3.618 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 Table 3 Results of determination of roxarsone in chickens fed on supplemented diets Sample type* ‘F’ leg (supplement only)? ‘F’ leg (supplement only) + 200 ng spike ‘H’ leg (supplement only) ‘C’ Breast (supplement only + 7 d)§ ‘C‘ Breast (supplement only + 7 d) + 200 ng spike ‘C‘ leg (supplement only + 7 d) Spike only (200 ng as As) Blank (taken through method) roxarsone value/ ng As per gram of tissue N.D.2 173 N.D.$ N.D.2 18693 N.D.2 205 N.D.$ Recovery from 200 ng spikes in tissue % 87 - - - - - 102 - * Examples of chicken tissue types leg and breast samples; individual chickens identified as A-H.t Supplemented diet with roxarsone (no withdrawal period). 2 Not detected based on detection limit of 1.8 ng of As per gram of chicken (3RSD) from 1 ng cmP3 standards. 5 Supplemented diet with roxarsone + 7 d withdrawal period. Table 4 Summary of roxarsone determinations in tissue from chickens fed on base only and supplemented diets Diet Leg samples Detection Detection limit/ limit/ Pg kg-l Breast samples Yg kg-l Base feed only (7-49 d) Not detected (ten samples) 17-25 Not detected (ten samples) 20-25 roxarsone supplemented feed only Detected in three out of ten samples in 7-25 Not detected (ten samples) 25 roxarsone supplemented + 7 d withdrawal Not detected (ten samples) 7-10 Not detected (ten samples) 7- 16 (7-49 d) on base feed the range 55-60 pg kg-’ No roxarsone was identified in the unspiked chicken samples and blank.Based on the 1 ng cm-3 of As roxarsone standard injections a detection limit of 1.8 ng of arsenic per gram of chicken (6.5 ng of roxarsone per gram of chicken) was obtained. Linearity of the roxarsone standards 1 5 and 20 ng cm-3 of As was determined to be 0.985. The results of roxarsone determinations in tissue from chickens fed on base only and suplemented diets are given in Table 4. Conclusions A method for the determination of roxarsone in chicken muscle has been developed. The extraction procedure is based on trypsin enzymolysis digestion at pH 8. The method gave recoveries of 85-103% from spiked samples of chicken muscle and showed that the digestion technique did not affect roxar- sone itself.Separation of roxarsone from the digest matrix was required prior to its determination and the clean-up step employed involved anion-exchange chromatography at pH 4-5. Quantification was accomplished by reversed-phase HPLC using ICP-MS detection giving a quantification limit of 25 ng of roxarsone per gram of muscle tissue. No roxarsone was detected in muscle from chicken fed on a diet supplemented with 36 pg g-’ of the growth promoter following a 7 d with- drawal period. Moody J. P. and Williams R. T. Food Cosmet. Toxicol. 1964 2,695 Peoples S . A. Medical and Biological Effects of Environmental Pollutants National Academy of Sciences Washington D.C. Baron R. R. Proceedings of a Seminar on the Use of Arsenicals in Feeding Stufls Salisbury Laboratories London 1969 p. 107 Peoples S. A. Medical and Biological Effects of Environmental Pollutants National Academy of Sciences Washington D.C. 1977 p. 150 Moody J. P. and Williams R. T. Food Cosmet. Toxicol. 1964 2 707 Crews H. M. Burrell J. A. and McWeeny D. J. J . Sci. Food Agric. 1983 34 997 Crews H. M. Burrell J. A. and McWeeny D. J. 2. Lebensm. Unters. Forsch. 1985 180 221 Crews H. M. Burrell J. A. and McWeeny D. J. 2. Lebensm. Unters. Forsch. 1985 180 405 Tan S. H. and Horlick G. Appl. Spectrosc. 1986 4 445 1977 pp. 155-156 L. and McWeeney D. J. J. Anal. At. 1 10 Munro S Spectrom. Ebdon 986 1 2 References Paper 3/06334K Received October 25 1993 Accepted February 2 1994

ISSN:0267-9477    

DOI:10.1039/JA9940900615

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 619 Theoretical Calculation of the Standard Deviation and Detection Limit in lnductively Coupled Plasma Atomic Emission Spectrometry* Evgeniy D. Prudnikov* Earth's Crust Institute State University St. Petersburg 199034 Russia Jaap W. Elgersma and Henri C. Smit Laboratory for Analytical Chemistry University of Amsterdam Nieuwe Achtergracht 166 1018 WV Amsterdam The Netherlands Standard deviations and detection limits in inductively coupled plasma atomic emission spectrometry (ICP- AES) are theoretically calculated. Theoretical and experimental data are compared. Practical application of the theory is described. The theory helps to provide more insight into the (inter)connections between standard deviation instrumental sensitivity and detection limit.The theory is useful in decreasing method errors. A simple method of estimating the standard deviation of net line signals and detection limits is proposed. Keywords lnductively coupled plasma atomic emission spectrometry standard deviation; detection limit Equations to calculate theoretically the detection limit (c,) in atomic emission spectroscopy in general are described in refs. 1 and 2 whereas the formulae for flame atomic emission spectrometry in particular are presented in refs. 3 and 4. These relations are based on the Einstein-Boltzmann equation for spectral line intensity radiated from plasmas which are in local thermal equilibrium (LTE). In the formulae for atom and ion lines in inductively coupled plasma atomic emission spec- trometry (ICP-AES) the temperature parameter should be replaced by an effective temperature (Tee) because the ICP is considered to be in partial LTE.' Kaiser's statistical criterion is the underlying idea of the formula for cm6 C = k,S,- 'Sbl (1) See Table 1 for definitions and units of symbols.This statistical approach has already been the basis of the formulae for c in AES.2,7,8 Different derivations based on eqn. (1) are used to estimate the c from experimental data. Recently Boumans again emphasized the determination of c in ICP-AES by the 'SBR-RSDB appr~ach'.~-'' The relative variables in this procedure signal-to-background ratio (SBR) and relative standard deviation of the background (RSDB) are objective criteria for instrumental performances.However the empirical methods demand a series of experimental data and in addition insufficiently discriminate between the different kind of errors contributing to sbl. Therefore a further development of the description of c is worthwhile. To gain an understanding of the analytical method it is desirable to describe theoretically the standard deviation of signal measurements one of the most important parameters in atomic spectrometry because it is related to instrumental and method factors element concentration and c,. Measurements of random errors are based on mathematical ~tatistics.'~''~ For a long time the empirical linear approxi- mation between element concentration and standard deviation has been S,=a+bc (2) where S is the standard deviation of an analytical determi- nation in units of concentration; and a and b are the coefficients a in concentration units and b dimensionless.Theoretical calculation of S as a function of the element ~ ~~ ~ * Presented at the XXVIII Colloquium Spectroscopicurn Inter- nationale (CSI) York UK June 29-July 4 1993. concentration and c has been achieved in atomic absorption spectrometry and in flame atomic emission ~pectrometry.'~'~ The aim of this article is to describe theoretically these figures of merit in ICP-AES. A comparison between theoretical and experimental data is rriade to show how instrumental and non-instrumental (method) errors in signal measurements can be distinguished. Theory The five kinds of random errors in different analytical and other natural systems should be distinguished." These fluctu- ations depend on the system and the element concentrations.Their values are proportional to the power 0 3 1 3 and 2.16 The standard deviation (s) of analytical measurements may now be written a d 6 s = [ $ I 2 4- (kd + 2RA2S,cbl)s,C 4- A2S:C2 + kkS$c3 + knS;c4]+ where kk and k are constants R the correlation coefficient for the correlation between net and interfering signals R = 0-1. It was stated in Ref. 17 that when using the calibration function and the value of the analytical signal (v in amperes) u=S,c can be transformed to the equivalent blank value (cbl) cbl= q,l/Sv and sbl= ASvcb if the flicker noises are prevail- ing which is what is usually observed in practice. The first three terms at the right-hand side of eqn. (3) characterize the linear range of measurements whereas the two following terms are determined by the appearance of non- linearity in analysis.Since calibration functions in ICP-AES are linear over wide concentration ranges of five orders of magnitude the two last terms can be disregarded. Thus the following equations are obtained (3) S = [sbt -k (kd + 2RASbl)SvC f A2sv2 C2]' (4) and for s, Substituting eqn. (1) into eqns. (4) and ( 5 ) gives S = [Cm2S,2k,-'+(kd + ~RAC,S,~,-')S,C+A~S~C~]~ ( 6 ) and S = [ c ~ ~ - ~ c - ~ + (kd + 2RAc,S,k,-')SV- 'c- + A']' (7) Equations (4) and (6) and eqns. ( 5 ) and (7) can be applied for theoretical calculations of s and s in ICP-AES respectively. For the present purposes it is assumed that the excitation source is in LTE. Theoretical calculations for the LTE model620 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 Table 1 Definitions and units of symbols Atomic weight of element atomic mass unit Transition probability s Additive error in linear approximation pg ml-' Photomultiplier factor related to gain per stage dimensionless Coefficient of multiplicative error in linear approximation d.1. Intensity of excitation source continuum w cmP2 sr-' nm-' Standard deviation of BA W cm-2 sr-' nm-' Concentration of element pg ml-' Limit of detection pg ml-' Blank concentration of element pg ml-' Effective aperture of the spectral instrument cm2 Average diameter of the plasma region focused on the entrance slit of the spectral instrument cm Exponential function d.1. Electronic charge C Excitation energy of state u J Focal length of the collimator cm Solution flow rate ml min-' Frequency response band width Hz Statistical weight of ground state d.1.Statistical weight of state u d.1. Slit-height cm Planck constant J s-' Dark current of the photomultiplier A Boltzmann's constant J K-' Numerical factor chosen according to the confidence level Coefficient for calculation of shot noise A Gain factor of the photomultiplier d.1. Moles present d.1. Instrumental sensitivity of apparatus A pg- ' ml Standard deviation of the total noise of net line signal A Relative standard deviation of the total noise of net line Standard deviation of background fluctuations A Effective source temperature according to local thermodynamic equilibrium (LTE) model K Transmission factor of optics d.1.Nebulizer gas flow rate ml min-' Slit-width cm Nebulization efficiency d.1. Atomization efficiency d.1. Ionization efficiency d.1. Photomultiplier sensitivity A W-' Instability factor of the excitation and measurement d.1. Spectral slit-width nm Frequency at the centre of the line s-' (d.1.) desired d.1. signal d.1. are well described and show good results.14 For the ICP T is introduced. It is possible to estimate this effective temperature theoretically and thus to obtain supplementary information about excitation processes in the ICP. Comparison Between Theoretical and Experimental Data The standard deviations of the intensities of the Cu I 324.754 nm atom line and the Cd I1 226.502 nm ion line as a function of the element concentration were calculated theoreti- cally.These values were compared with the statistically adapted standard deviations of net line intensities obtained from measurements described in ref. 19; these experimental data were still on hand at our laboratory. So that comparisons can be made relevant experimental conditions of the measurements are cited briefly here. The experiments were performed with the Jarrell-Ash Atom Comp Model 970 ICAP multichannel spectrometer system. The measurements were conducted under optimum conditions for simultaneous multi-element analysis. Back- ground and gross line intensities were measured for acidified aqueous solutions including a blank and a set of reference solutions in the concentration range 1 ng ml-l-100 pg ml-l. A 10 s integration time was employed.The experiments were performed over a period of 4 d. Each day four complete runs of the reference solutions were carried out. The intensity measurements for each reference solution were preceded by quadruplicate measurements of the blank solution. From each of a group of four consecutively measured gross line intensities the mean value of the corresponding four measured back- ground intensities was subtracted resulting in four net line intensities per concentration. The method of calculation included the selection of T of the plasma when the theoretical results were nearest to the experimental data. For calculations the same procedure as described in refs. 14 and 15 was followed. According to refs. 1-4 14 15 and 20 the following equation can be given for instrumental sensitivity S, S = (4.n)- 'hv,g,,g0- 'Au( - exp EJkTeff )dPly T f WH(D/F)2 x 6.1023300Fsa/l/3i(AT,fl (8) and for standard deviation of the background noise sbl SbI2 = 2e,BMdfid + 2ecBMAfy&13 WH(D/F)2Ay + [ y TfAB,WH(D/F )2A.A,]2Af (9) where the first term on the right-hand side of eqn.(9) is the shot noise due to the thermal electron emission of the photo- cathode the second term is the shot noise contribution of the excitation source to the background noise and the third term is the flicker noise of the excitation source. The coefficient of the shot noise factor k d is k,(A) = 2ecBMAf Values of the coefficients and parameters were taken from the and the value of a from ref. 21. The intensity of the plasma source continuum (B,) was calculated theoretically with the aid of data from ref.1. The list of values used in the calculations are as follows n=3.14; h=6.63 x J s-l; vo= 1.32 x 1015 s-' for Cd; yo= 0.93 x 1015 s-l for Cu; gu=4; g0=2; A,=99 x lo8 s-l for Cd; A,=4.1 x lo8 s-l for Cu; E,=8.75 x J for Cd; E,= 6.12 x 10-19J for Cu; k=1.38 x J K-'; T see Table 2; dpl= 1.5 cm; y = 1 x lo4 A W-'; T,=0.3; W=o.0025 cm; H= 0.3 cm; (0/1;)2=2.56 x F,= 1.1 ml min-'; a=0.016; fi= 1; /li = 0.5; A= 112.4 for Cd; A = 63.5 for Cu; T/B = 800 ml min-'; 1 x 10-"A; B,=3 x W cm-2 sr-l nm-'; AII,=0.025 nm; AB,=3 x W cm-'sr-l nm-'; R = 1; A=O.O05; and k,=3. The calculations were carried out for different plasma source temperatures. Thus T could be chosen according to the LTE model of the excitation source (see Table2). It was apparent that the calculated results approached the experimental data most closely for 7000-8000 K.For further considerations Kff = 7500 K was taken. Using the LTE model,T can be estimated from Table 2 as literaturel-3 .14.15.19,20 nT/298= 1; ec= 1.6 X wi9 C; B= 1.3; M = lo6; AfzO.1 HZ; i d = Table 2 Theoretical estimation of the effective temperature (KR) Theoretical results Cu I 324.754 nm Cd I1 226.502 nm T/K S V cJpg ml-' S" cJpg ml-' 6000 4.8 x 1.0 x 8.9 x 4.8 x 7000 1.2 x 10-4 3.5 x 10-4 3.5 x 1 0 - 5 1.2 x 10-3 7500 1.6 x 10-4 2.6 x 10-4 5.9 10-5 7.0 10-4 8000 2.2 x 10-4 1.9 x 10-4 9.5 10-5 4.4 10-4 9000 3.7 x 10-4 1.1 x 10-4 2.0 x 10-4 2.2 x 10-4 12000 9.6 x 10-4 4.4 x 10-5 8.9 10-4 5.8 10-5 Experimental data (graphical method) cdpg ml - ' Cu I 324.754 nm Cd I1 226.502 nm 3 x 10-4 6.5 xJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 621 the mean gas temperature (7J of the ICP. The proposed approach also permits discussion of the physical parameters of the ICP discharge. According to the data in Table2 the temperature of excitation (T,) for the Cu I line is approximately equal to T cu I = 7300 K and for the Cd I1 line T Cd = 7600 K at the same point in the ICP discharge. The results of the calculations confirm the increase in T with the increase in the excitation energy of the level. It should also be noted that the electronic and ionization temperatures ( zlec and respect- ively) have a higher value than T p and T,. The results show the deviations of the ICP from the LTE model. Therefore this approach might be useful for the study of the physical param- eters and the development of the theory of ICP spectrometry.With the values listed above S sbl s and s were computed according to eqns. (8) (9) (4) and ( 5 ) respectively whereas c was given by eqn. (1) S,(CU)= 1.6 x Sb1=1.4X lo-'; kd+2RASb1=lq4X lo-" for R = l S,(Cd)=5.9 x lo-' s(Cu)=(1.9 x 10-16+2.2 x 10-14c+6.5 x lO-l3c2)d s(Cd)=(1.9 x 10-16+8.3 x 10-15c+8.5 x 10-14~2)* s,(Cu)=(7.4 x 10-'cP2+8.8 x 1OP7c-'+2.5 x lo-')* (11) s,(Cd) = (5.4 x ~ O - ' C - ~ + 2.4 x 10-6c-' + 2.5 x lo-')* (12) c,(Cu) = 2.6 x c,(Cd)=7 x pgml-l for k,=3 pg ml-' for k = 3 It is obvious that agreement between theoretical results and the experimental data of s is rather close (Fig. 1 and Table 3). Practical Application It should be remembered that for the calculations of s (Cu) and s (Cd) in eqn.( 5 ) the value for the co-correlation factor R R= 1 was substituted. In ref. 17 the significance of R has been described and illustrated by theoretical and experimental data. For instrumental errors R can have values from 0 to 1. When R =0 there are no contributions of non-instrumental errors to the total noise of signal measurements. When R = 1 it can be seen that eqn. (4) is very similar to eqn. (2) the linear empiric approximation. This value of R confirms the dominant role of the correlation between the standard deviation of the net line signals and the interfering signals. If in practical calculations an experimental value of R> 1 is obtained then R is an indicator of the presence of shot type non-instrumental (method) fluctuations.The linear approximation R = 1 allows for reasons of practical application more simple methods of estimating s and c,.17 In that case s,=a+bc may be converted into s,= (a2 + 2abc + b2c2)* for R = 1. Then by combining eqns. (2) and (7) one obtains (13) s = c,k,- 'c-l + A For the factors a and b from eqn. (2) it follows that To estimate these instrumental parameters standard deviations of only two concentrations have to be available; a low and a high concentration of the linear working range should be taken. Here Boumans' formulations of a similar expression is noteworthy:22 S = ( c ~ ~ - ~ c - ~ + A2)* Table 3 presents three different types of formulae for s as a function of c derived for Cu as well as for Cd.Type I expression for the linear approximation eqns. ( 11) and ( 12) derived from eqn. ( 5 ) . Type 11 expression for the practical application of the linear approximation derived from eqn. (13). Type 111 expression of the fit for the means of the s values of the experimental data from ref. 19. Table 3 also lists the c values calculated according to the above types of functions taking into account that in theory s = 0.33 for k = 3. Fig. 1 (a) and (b) show plots of the mathematical expressions mentioned above for Cu and Cd respectively; i.e. the s values a=c,k,-l; b=A. 0.40 0.35 0.30 C 0 > 0 -a -0 C (0 0 .- 0.25 .- % 0.20 w .? 0.15 4- - 0 CT 0.10 0.05 0.01 I O - ~ 1 0 - ~ 10-2 l o - ' 1 10 Crn \\ J I I I I I I O - ~ :10-3 10-2 lo-' 1 10 CITl Log (concentration/pg ml-') Fig.1 Relative standard deviation s of net line intensities as a function of the log of the concentration for (a) Cu I 324.754 nm and (b) Cd I1 226.502 nm. 0 Linear approximation (Type I); broken line practical application of linear approximation (Type 11); solid line experimental data (Type 111); error bars show the variability of the s values of the sets of measurements (n = 4); c indicates the detection limit622 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 Table 3 factor (k,) Expression of the relative standard deviation of net line intensities (s,) versus concentration (c); and detection limits (c,) for confidence Analysis line/ nm Cu I 324.754 Cu I 324.754 Cu I 324.754 Cu I 324.754 Cd I1 226.502 Cd I1 226.502 Cd I1 226.502 Cd I1 226.502 Type (see text) I I1 I11 - I I1 I11 - Expressions for s ~ = ( 7 .4 ~ lo-’ ~-’-1-8.8 x lo-’ c-’+2.5 x lop5)+ s,= 1 x ~-‘+0.005 Graphical method s,=(i x 10-8 c-2+4 x 10-6 c-1+2.5 x 10-5); ~ = ( 5 . 4 ~ lop8 C2-2.4x ~ = 2 . 1 x lop4 ~-‘+0.005 Graphical method c-’+2.5 x lo-’)* s,=(4 x 10-8 c-2+:1 x 10-5 c-1+ 1.6 x lo+)+ cm(ka = 3 I/ pg ml-’ 2.6 x 10-4 3 x 10-4 3.3 x 10-4 3 x 10-4 7 x 10-4 6.3 x 10-4 6.5 x 10-4 6.6 x lop4 are given as a function of the log of the element concentrations. The figure illustrates the following. (i) The s values calculated according to Type I coincide with the curve of Type 11. Thus eqn (13) can be applied for theoretical calculations of s and c which is a time-saving method for the linear approximation R = 1. (ii) The s values of net line signals for concentrations in the middle of the working range (Type 111) are significantly enhanced with respect to those of corresponding concentrations of Type I and 11.According to the concept of linear approxi- mation R should be considered R> 1. This means that non- instrumental errors due to method factors are responsible for the relatively large enhancements of the s values. (iii) A systematic decrease of s data for the lowest measured concen- tration of Cu [Fig. l(a)]. It is more likely that the applied method of background correction causes this bias. It should be noted that sbl is defined as the standard deviation of the background measurements without the use of the nebulizer. However the largest contribution to the total noise is the nebulizer flicker n o i ~ e .~ ~ ~ ~ Therefore it is under- standable that the c values in Table 3 are much too low i.e. for Cu about 5 times and for Cd 2 times. From the above discussion it is apparent that the applied theory can be used to control the contributions of non- instrumental errors of the analytical procedure. Conclusions Theoretical calculations of the standard deviation and detec- tion limit in ICP-AES have been carried out. The theoretical results for Cu I 324.754nm and for Cd I1 226.502nm are compared with experimental data. The possibilities of the practical use of this theory are shown. The theory helps to gain more insight into the (inter)connections between standard deviation instrumental sensitivity and detection limit. The theory is useful to decrease method errors.A simple method of estimating standard deviation of net line signals and detec- tion limits is proposed. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 References Mandel’shtam S. L. Zh. Prikl. Spektrosk. 1964 1 5. Silberstein Kh. I. Spectral Analysis of Pure Substances Khimiya Leningrad 1971; Adam Hilger Bristol 1977. Winefordner J. D. and Vickers T. J. Anal. Chem. 1964,34 1939. Prudnikov E. D. Zh. Anal. Khim. 1972 27 2327. Blades M. W. in Inductively Coupled Plasma Emission Spectroscopy ed. Boumans P. W. J. M. Wiley New York 1987 pt. 2 ch. 11. Kaiser H. Z . Anal. Chem. 1965 209 1. Boumans P. W. J. M. and Maessen F. J. M. J. Z . Anal. Chem. 1966 220 241. Boumans P. W. J. M. and Maessen F. J. M. J. Z . Anal. Chem. 1967 225 98. Boumans P. W. J. M. Spectrochim. Acta Part B 1990 45 799. Boumans P. W. J. M. Spectrochim. Acta Part B 1991 46 431. Boumans P. W. J. M. Spectrochim. Acta Part B 1991 46 641. Doerffel K. Statistik der Analytischen Chemie Leuna-Merseburg Leipzig 1966. Nalimov V. V. The Application of Mathematical Statistics to Chemical Analysis Eizmatgiz Moscow 1960; Pergamon Press Oxford 1963. Prudnikov E. D. Fresenius’ 2. Anal. Chem. 1981 308 339. Prudnikov E. D. Spectrochim. Acta Part B 1981 36 385. Prudnikov E. D. Analyst 1984 109 305. Prudnikov E. D. Fresenius’ J . Anal. Chem. 1990 337 412. Prudnikov E. D. and Shapkina Y. S. Vestn. State Univ. St. Petersburg 1992 25 38. Maessen F. J. M. J. and Balke J. Spectrochim. Acta Part B 1982 37 37. Prudnikov E. D. Fresenius’ J. Anal. Chem. 1981 308 342. Maessen F. J. M. J. Coevert P. and Balke J. Anal. Chem. 1984 56 899. Boumans P. W. J. M. Spectrochim. Acta Part B 1976 31 147. Second ICP Conference Noordwijk aan Zee The Netherlands ICP lnf. Newsl. 1978 4 211. Paper 3/05056D Received August 20 1993 Accepted January 4 1994

ISSN:0267-9477    

DOI:10.1039/JA9940900619

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 623 Direct Determination of Metals in Oils by Inductively Coupled Plasma Atomic Emission Spectrometry Using High Temperature Nebulization Johann L. Fischer and Cor J. Rademeyer* Department of Chemistry University of Pretoria Pretoria 0002 South Africa A method is described wherein a Babington V-groove nebulizer and heated spray chamber is utilized to nebulize edible oil and lubricating oil directly into an inductively coupled plasma (ICP) to determine the metal concentrations in the oils by ICP atomic emission spectrometry. Factors influencing the attainment of a maximum analytical signal are discussed. Both the methods of standard additions and normal calibration were used and calibration curves were linear in the range 0-10 pg g-' for Fe Cu Cr Ti Al Ni Mg Ag and Na.The following detection limifis are reported Fe 0.032; Cu 0.069; Cr 0.051 ; Ti 0.045; Al 0.328; Ni 0.1 31 ; Mg 0.003; and Ag 0.077 pg g- . Keywords Inductively coupled plasma; high temperature nebulization; nebulizer; oil analysis; organic samples Metals occur naturally in a wide variety of foods. Some of these metals are essential for a healthy diet while others can be toxic e.g. lead mercury and cadmium.l Besides the obvious need to analyse foodstuffs for the toxic metals some bio- essential metals might adversely effect food quality if present in too high concentrations. Trace amounts of copper (0.1-1 pgg-I) and iron (1 pgg-l) for example hasten the rancidity of fats.2 They serve as catalysts in the oxidation of the unsaturated bonds in lipids and cause the rapid deterior- ation of cooking oil and fat-containing foods.' It is therefore essential to monitor the metal content of edible oils and indeed most food stuffs.A further application of the determination of metals in an oil medium is the analysis of lubricating oils to monitor the wear of oil-lubricated machines. This kind of analysis is carried out extensively by the US Air Force and others like the South African Air Force (SAAF) in their Spectrochemical Oil Analysis Programs (SOAP). Various methods for the analysis of metals in oils have been described including neutron activation analysis colorimetry4 and spectrophotometric analysis of the residue after wet or dry ashing.' The favoured techniques have been however those in which the oil sample is directly analysed with or without dilution with an organic solvent.These methods include various spark atomic emission techniques,6 flame7 or non-flame' atomic absorption or fluorescence as well as X-ray fluores~ence.~ The method of choice for the analysis of lubricat- ing oil has long been rotating electrode spark source spec- troscopy which offers high throughput and simultaneous determination of several elements.6 Several workers9-l0 have since described the application of inductively coupled plasma atomic emission spectroscopy (ICP-AES) to the analysis of oils after dilution with a suitable solvent. This method has now largely replaced the rotating electrode spark source method in many laboratories. However a disadvantage of the ICP method is the need to dilute the samples with a low viscosity solvent to facilitate nebulization of the viscous samples and to equalize the nebulization efficiency for samples of varying viscosity.This decreases the overall sensitivity of the analysis and leads to increased limits of detection (LODs) while providing an opportunity for contamination and analyt- ical error. Algeo et a/." described a method whereby lubricating oil samples were nebulized directly into the plasma without prior dilution using a heated Babington nebulizer. This elimin- ated the difficulties caused by the high viscosity of oils at room temperature without introducing the disadvantages of the dilution method mentioned above. In a recent study carried * To whom correspondence should be addressed. out in this laboratory" the hot nebulization of molten wax into an ICP was investigated. This paper describes the appli- cation of that method to the analysis of lubricating and edible oils.Experimental Apparatus and Instrumentation Spectrometer A Spectroflame ICP atomic emission spectrometer manufac- tured by Spectro Analytical Instruments was used. The instru- ment is fitted with vacuum and air polychromators and a sequential monochromator covering the region 200-480 nm. Viewing height does not play a vital role in this optical system since the fibre optics have a viewing cone of approximately 60". The excitation source was an argon plasma operating at a radiofrequency of 27.12 MHz. A fully demountable torch supplied by Spectro was used.Nebulization system The previously described12 nebulization system was used. As mentioned in that article problems were experienced with the reproducibility of the nebulizer. In line with the recommen- dations made in the article the diameter of the gas outlet in the nebulizer nozzle was reduced to 0.25 mm thus increasing the operating pressure of the nebulizer without increasing the gas flow rate. The nebulizer was further modified so that the V-groove actually extended past the nut that clamps the nozzle to the nebulizer body (Fig. 1). This led to easy drainage of Top view Aluminium nozzle Swagelok fittings for 3 mm tubing Front view Diameter =0.40 mm Diameter= 0.25 mm Fig. 1 provide free drainage of sample from the V-groove Modified heated Babington nebulizer with extended nozzle to624 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 excess sample ensuring that nebulization occurred smoothly and reproducibly. The sample introduction system was modified as depicted in Fig. 2. The pressurized sample container used in the work on waxesL2 was replaced with a variable speed peristaltic pump. A Rabbit pump from Rainin Instrument Company was used with Masterflex silicone tubing size 13. The tubing between the pump and nebulizer was fed through a length of 6 mm diameter stainless-steel tubing which was heated on the outside with resistance wire and insulated with asbestos rope. The sample was heated and stirred on a hot-plate while being fed to the peristaltic pump. This system was preferred to the pressurized system because it simplified and sped up sample changes.However the system could not be used for waxes or other samples that are solids at room temperature because the sample cools rapidly in the length of tubing passing through the peristaltic pump. In the work reported here the waste bottle used previouslyL2 with the additional liquid phase was unnecessary since oils are liquids at room temperature. The conventional liquid trap supplied with the Spectro ICP instrument was used. Sample introduction tube The previously described12 sample introduction tube for waxes was unsuitable for the oil analysis since its aperture became clogged by carbon deposits. The normal sample introduction tube supplied with the Spectro demountable torch was used instead. Some carbon build-up did occur on the sample (a) Nebulizer n Heated sample line Thermocouple ainless-steel pipe Variable a.c.source powering resistance wire Oil sample ted and stirred Asbestos rope 6 mm Stainless-steel Pipe Peristaltic tubing Resistance wire wrapped around stainless-steel Fig.2 (a) Modified sample introduction system used for oils and (b) cross-section of heated sample line. (The sample is fed to the nebulizer with a peristaltic pump using a heated sample line) introduction tube but it appeared on the outer edge of the tip and therefore did not cause the aperture to be reduced. This carbon build-up had to be removed periodically since it extended into the plasma thus altering the analytical con- ditions of the plasma. This necessitated the shut-down and re-starting of the plasma after about 20min of operation. Fortunately the analysis was not adversely affected since the plasma conditions were reproducible between successive runs.The introduction of oxygen into the plasma is widely used to prevent the formation of carbon deposits in the torch when organic solvents are introduced into the ICP. This has not yet been tried but it might alleviate the problem and improve the long-term stability of the method. Samples The edible oil used for the investigation was commercially available sunflower oil used for cooking purposes. Additionally three aircraft lubricating oil samples were obtained from the SAAF. These samples (No. 35 No. 36 and No. 38) were ana- lysed by the SAAF as part of their Spectrochemical Oil Analysis Program and were of the same type of oil namely Turbonycoil 600 (JSD OX-27).A hot filtration was carried out under vacuum-water aspirator using a sintered glass funnel to remove a few large metal particles clearly visible to the naked eye. The filtration process was very rapid (about 2min for 50ml) and was carried out on the heated sample just prior to analysis. Standards For the work on sunflower oil the method of standard additions was used for calibration. This was necessary since no blank sample completely free of metals was available. Such a sample in any case would be difficult to obtain. The standards were prepared by weighing the appropriate amount of Conostan D12 900ppm standard oil and adding sunflower oil to a total mass of 1OOg. The following additions were prepared 0.0995 0.5041 1.001 1.9994 and 5.0035 pg g-' of the metals. These mixtures were heated and thoroughly stirred beforehand to yield homogeneous mixtures.When analysing aircraft lubricating oil the objective of the analysis is different from that of the sunflower oil. Here the change in metal concentrations in the sample due to wear of the engine are to be determined. The clean oil used in the engines can therefore be used as a blank thus making a conventional calibration procedure possible. Absolute values for the metal concentrations are therefore not reported rather a value above that of the blank. The standards were prepared in the same way as with the sunflower oil weighing the appropriate amount of standard oil (Conostan D12) and adding clean Turbonycoil 600 (JSD OX-27) to a total mass of 100.00 g.The metal concentrations in the standards were 0.101 0.510 0.998 2.017 and 5.014 pg g-'. Optimization of Working Conditions Influence of temperature on oil viscosity A temperature interval was sought that would yield a sufficiently low viscosity (less than 0.01 N s m-2) to ensure a sufficiently high flow rate through the nebulizer. Additionally the temperature interval had to be chosen so that the viscosity would not be strongly influenced by temperature changes. The influence of temperature on the viscosities of both sunflower oil and Turbonycoil 600 oil was studied by measuring the viscosity of the oil at different temperatures with a Brookfield Synchro-lectric viscometer Model LVF. The viscosity was measured at temperatures in the 20-130°C range.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 625 Influence of oilflow rate on signal intensity For the optimization of the sample flow rate and plasma power setting the system reported on previously,12 employing a pressurized sample chamber was used. This was beneficial since that set-up allowed a larger interval of flow rates to be used which assisted in understanding the influence of sample flow rate on intensity. The sample that was used during the optimization was prepared by weighing 2.2292 g of Conostan D12,900 ppm standard oil into a beaker and adding commer- cial sunflower oil to a total mass of 205.830 g thus yielding an addition of 9.7473 pg 8-l of all the elements. The elements that were determined and the wavelengths used are given in Table 1.All the elements except Na were determined using the monochromator. The background signal was measured at equal distances to the left and right of the peak. The average background under the peak was then calculated as the average of the left and right background. The measurements under a given set of operating conditions consisted of 11 measurements each at the peak and the two background positions. An integration time of 3 s was used. Sodium was determined with the polychromator. No background correction was performed since the polycromator has no scanning facility. In determining the influence of sample flow rate on signal intensity the sample flow rate was varied by adjusting the pressure applied to the sample chamber from 104 mm H20 to 188mm H20.The parameters that were kept constant are given in Table 2. Influence of plasma power on signal intensity The influence of plasma power on signal intensity was deter- mined by measuring the peak and background intensities as described above at power settings between 1.2 and 1.5 kW while keeping all other parameters constant at the values given in Table 2 except that the pressure exerted on the sample chamber was kept constant at 150mm H20. Table 1 Analytical wavelengths used for the analysis of oils ~ Element Fe c u Cr Ti A1 Ni Mg Ag Na Peak wavelength/nm 259.940 324.754 267.716 336.121 396.152 231.604 279.553 328.068 589.592 Background wavelengths/nm relative to peak & 0.028 & 0.028 + 0.027 z0.033 & 0.027 f 0.025 f 0.025 f 0.026 - Table 2 optimization of sample flow rate Instrumental parameters that were kept constant during the Parameter Nebulizer argon pressure/kPa Nebulizer argon flow rate/l min-l Oil temperaturePC Plasma power/kW Plasma gas flow rate/l min-' Intermediate gas flow rate/l min-' Observation height/mm Spray chamber temperaturerc Value 7.0 0.5 1.4 12 0.9 5 108-1 10 213-218 Results and Discussion Sunflower Oil Influence of temperature on oil viscosity The results of the study are depicted in Fig.3. The viscosity q ( N s m-2) and temperature T ("C) is apparently related through a power function q=2.856 T-1.27. From this it can be deduced that the viscosity is less than l o x N s m P 2 above 85°C and by differentiation it is known that between 100 and 115"C the slope of the curve is approximately -0.1 x loM3 N s mV2 "C-'.This would then imply that a working temperature of 110f2"C would yield a viscosity of 7.2 x kO.2 x lop3 N s m-2. Judging from the work done on waxes,12 a viscosity of less than 10 x lop3 N s m-2 constant to within 0.5 x lop3 N s m-2 is sufficient to ensure constant nebulization. Influence of oilflow rate on signal intensity The influence of sample flow rate on intensity and signal-to- background ratio (S/B) was similar for all the elements of interest. In Fig. 4,(u)-(c) these effects are shown for iron. The effect of flow rate on the background signal however was interesting [Fig. 4(a)J. The background increased nearly lin- early up to a pressure of 150 mm H 2 0 and then levelled off.Bearing in mind that this is a general effect for all the elements studied it seems that the background intensity over the whole spectral range increased in this fashion with an increase in sample flow rate. Hence it follows that up to a pressure of 150 mm H 2 0 the increase in flow rate resulted in a significant increase in the mass of sample reaching the plasma. If the effect of flow rate on signal intensity and S/B [Fig. 4 (b) and (c)] is evaluated it is clear that both increase up to a pressure of 164mm H 2 0 and then level off. The increase in the S/B between 150 and 164 mm H,O pressure is especially profound due to the fact that the background does not increase signifi- cantly in this region while the analytical signal does. This further increase in the signal while the background remains almost the same might possibly be ascribed to a more suitable size-distribution of sample particles reaching the plasma.It is therefore concluded that the optimum sample flow rate was reached at a pressure between 150 and 170 mm H20. Influence of plasma power on signal intensity As can be expected an increase in input power resulted in an increase in both signal and background intensity. A near linear relationship was obtained between input power and S/B with S/B decreasing as power increases as indicated in Fig. 5. It was concluded that the lowest possible power setting 1.2 kW would be the most satisfactory. 70 60 50 a $ 40 8 30 4- .- v) v) .- ' 20 10 0.0 -0.5 I -1.0 9 -1.5 y -2.0 'g L1 al > -0 -2.5 '5 -3.0 E -3.5 LL .- I I 1 1 I 0' 1 -4.0 20 40 60 80 100 120 TemperaturePC Fig.3 Influence of temperature on the viscosity of sunflower oil A viscosity; and B first derivative.The first derivative of the trend indicates that temperature does not influence viscosity to a great extent at higher temperatures626 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 L h .c 750 $ 650 2 550 mz 450 .% 350 c v > C $ 250 - Pressure asserted on oil/mm H20 Fig.4 Influence of sample flow rate on (a) background intensity (b) 0 total and 0 net signal intensities and (c) signal-to-background ratio of Fe at 259.940 nm 30 I 20 - 5 - F 0 ' I I I I I 1.20 1.25 1.30 1.35 1.40 1.45 1.50 PowerIkW Fig. 5 A Fe; B Ti; C Cr; D Cu; E Ag; and F A1 Influence of plasma power on signal-to-background ratio for Analysis of sunjlower oil For the analysis of sunflower oil the modified system (Fig.2) was used. This was done in order to simplify the sample changing procedure thus reducing analysis time The modifi- cation also solved some difficulties that were experienced with blockages that occurred in the sample feed line. These block- ages were experienced whenever the system was shut off for a few days and then re-started. A possible cause for the blockages might be that the metal catalysed decomposition of the oil in the stainless-steel tubing formed sticky deposits in the sample feed Iine after a period of time. By using the peristaltic pump set-up the oil did not come into contact with metal except in the nebulizer. It was also found that the nebulizer had to be thoroughly cleaned (by rinsing all parts with hexane) if it was left in contact with oil for any length of time.The pump rate of the peristaltic pump was adjustable so that the optimum pump rate could be determined with ease. It was found that the peristaltic pump was unable to reach as high a flow rate as was possible with the pressurized set-up. Although this meant that the system was not being operated under optimum conditions this set-up proved to be more reliable and easier to operate than the pressurized one. It was therefore decided to use the highest possible pumping rate achievable namely 1.27 g min-'. It is of course possible to increase the pumping rate if a larger diameter peristaltic tubing is used. The analysis was carried out under the conditions listed in Table 3 the analytical wavelengths used were those given in Table 1. As with the optimization 11 intensity measurements were taken at the peak and two background positions respect- ively.The under-peak background was calculated from the average of the two sets of background measurements on either side of the peak. The calibration curves for all the elements were linear with correlation coefficients in excess of 99.9% in almost all cases. Detection limits for the various elements were determined from the background intensity plus three times the standard deviation (SD) of the background. Calculating the LOD with standard additions calibration is somewhat more involved than simply substituting the calculated intensity into the total intensity calibration graph and obtaining the LOD in terms of concentration.The LOD was calculated as the concentration difference between the concentration equivalent of the background intensity and the concentration equivalent of the background intensity plus three times the SD of the background intensity using the total intensity calibration curve. Since off-peak background measurements were made for each sample it was possible to calculate the LOD for every set of background measurements The average LOD as well as the quantification limit at the level k = 10 is given in Table 4 for each element. When the method of standard additions was used the intercept of the net intensity calibration curve with the concen- tration axis yielded the concentration of the element in the sample. In order to determine the precision of the analysis it was necessary to calculate the SD in determining the intercept.This was done by fitting two additional curves I,+a and In-a where I is the net signal intensity of each standard Table 3 Operating conditions employed during the analysis of sun- flower oil Parameter Plasma gas flow rate/l min-' Nebulizer gas flow rate/l min-' Nebulizer gas flow rate) min-' Nebulizer pressure/kPa Plasma power/kW Spray chamber temperature/"C Temperature of sample feed line/"C Oil temperaturePC Value 12 0.9 0.4 7.0 1.2 190 109 90- 105 Table 4 Limits of detection (LODs) (k = 3) and limits of quantification (LOQs) (k = 10) for the various elements in sunflower oil. The average is reported as calculated from six sets of background measurements Element Fe c u Ti A1 Cr Ni Ag Mg Average LOD/pg g- 0.032 0.069 0.045 0.328 0.051 0.131 0.077 0.003 Average LOQ/pg g-l 0.107 0.230 0.150 1.095 0.169 0.438 0.257 0.010JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 627 and cr is the standard deviation of the net intensity. The intercepts of these lines with the concentration axis were at approximately equal distances from the intercept of the net intensity curve. The average of these two distances was then used as the SD of the intercept. A composite SD (oTot) like on was determined from eqn. (1) c is composed of the SDs of the on-peak left background and right background measurements. Using this method the concentrations of the elements in the sunflower oil sample were determined as well as the SD and relative standard deviation (RSD) of each determination.The results are tabu- lated in Table 5 indicating that the metal concentrations in this oil fell in the low to medium concentration range as defined by the standardized electrothermal atomic absorption spectrometry method for determining Cu Fe and Ni in oils.' The LODs possible with this method are the same as the typical metal concentrations of a low metal content oil which suggests that this method is not the most suitable for the analysis of oil with such low metal content (30-100 ng g-'). High temperature nebulization as a means of sample introduction to the ICP mass spectrometer might be capable of determining accurately such low concentrations and deserves investigation. Jet Lubricating Oil For the reasons stated above a study of the temperature dependence of the viscosity of jet lubricating oil was first carried out using the same method as described for cooking oil.From the values obtained it was concluded that the lubricating oil is very similar to cooking oil as far as viscosity is concerned. The function describing the relationship between temperature and viscosity is y = 2.456 T-1.24. From this it is clear that above 85 "C the viscosity is below 10 x N s m-2. From the derivative of this function it is clear that the rate of viscosity change per degree Celsius in the temperature interval 90-105 "C is approximately -0.1 x low3 N s m-20C-1. The Table 5 Concentrations of the various elements in the sunflower oil sample. The standard deviation (SD) was determined as described in the text Element Fe c u Ti A1 Cr Ni Ag Mg RSD Concentration/pg g- ' SDhg g- W) 0.161 0.01 1 6.73 < 0.230 <0.150 < 1.095 <0.169 - - < 0.438 < 0.257 - - - - - - - - - - 0.301 0.002 0.77 same working temperature used for sunflower oil also applies here.The modified sample introduction system used for sunflower oil was also used for this analysis. The operating conditions were not optimized again since the oil is so similar to sunflower oil and since the most important parameter sample flow rate was already at its maximum. In contrast to the analysis of sunflower oil this analysis did not require the use of standard additions because the clean unused oil served as a sample blank. Nevertheless background measurements to either side of the analytical wavelength were made in an effort to show whether or not the results obtainable from the total intensity calibration curve and the net intensity calibration curve differed. The calibration curves for all the elements (both net and total intensity) showed excellent linear dependence with correlation coefficients in excess of 99.9% in almost all cases.Detection limits were calculated using the sample blank as the background signal and using three times its SD. The LODs are tabulated in Table 6 and compare favourably with LODs reported by Brocas13 for the analysis of metals in oils by diluting the oil with different solvents. The concentrations of the different elements in the three samples are shown in Table 7 along with the precision and RSD of each.Also in Table 7 are the results obtained by the SAAF. There is good agreement between the results obtained from hot nebulization and the SAAF values for iron mag- nesium nickel and chromium. For nickel magnesium and chromium the results that did not correspond well were all in the low concentration range. Since the LODs of the SAAF are all higher than those that can be obtained with the proposed method it is to be expected that their determinations will be subject to larger errors at these low concentrations. The copper and aluminium concentrations as given in Table 7 are all higher than the values reported by the SAAF about 40% higher for copper. It is believed that the SAAF method may suffer from a spectral interference originating from molyb- denum emission.Ideally a certified reference standard ought to be used to establish accuracy. Such a sample was unfortunately unobtain- able. As explained previously the object of the analysis of lubricating oil is to determine changes in concentration rather than absolute values. The reasonably good agreement between the methods for most elements and the fact that the discrepanc- ies can be ascribed to faults in the reference results do however suggest that this method might be more accurate. The precision of the method is also acceptable especially with regard to iron and magnesium where RSDs of the order of 2% were obtained at concentrations well under 1 pg g-'. Taking into account that sodium is a difficult element to determine using ICP these high RSDs at concentrations below 1 pgg-' are to be expected.The RSDs of Ti Cr Ni and Ag were high for some samples but considering that they were obtained at concentration levels of the order of 0.1 pg 8-l this might not be so serious. In order to determine the long-term stability of the method Table 6 LODs in oil diluted with different solvents (from ref. 13) Limits of detection (LODs) for the analysis of used aircraft lubricating oil by high temperature nebulization (HTN) compared with Detection limit/pg g-' ~ Method Fe c u Ti A1 Cr Ni Ag Na HTN* 0.006 0.06 1 0.008 0.128 0.0 17 0.018 0.104 0.157 Toluene? 0.22 0.28 3.7 3.3 0.9 - 0.25 - 0.12 0.06 0.3 Kerosene? 0.025 0.015 MIBK+ 0.04 0.03 0.7 Xylenet 0.07 0.035 0.75 0.35 0.2 1 1.3 - - - - - - 0.3 ._ - - * LOD=X,+3Gb; n = l l .LOD = x b + 30b referring to the original undiluted oil.628 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 Table 7 Concentrations of the elements found in the samples of used aircraft lubricating oil; the results obtained by the SAAF are also given Sample 35 36 38 Figure of merit Concentration/pg g- ' SD/M g - ' SAAF results/pg g-' Concentratioq'pg g- ' SD/M g-' RSD (%) SAAF results/pg g-' Concentration/pg g- ' SD/PS g- RSD (%) SAAF results/pg 8-l RSD (Yo) Fe 2.61 1 0.046 1.76 2.519 0.753 0.016 2.16 0.926 3.277 0.108 3.29 3.729 c u 0.356 0.034 9.54 0.245 0.105 0.018 16.9 < 0.080 0.425 0.050 0.298 11.8 Ti 0.08 1 0.022 27.0 - 0.340 0.017 4.86 0.097 0.016 - 16.5 - A1 2.056 0.077 3.74 1.482 < 0.128 - - < 0.493 0.644 0.084 13.1 < 0.493 Cr 0.189 0.01 8 9.76 0.095 0.082 0.009 0.094 0.383 0.033 8.59 0.334 11.4 Ni 0.138 0.019 0.139 0.054 0.016 0.152 0.067 0.025 37.2 < 0.088 13.9 29.9 Ag 0.175 0.018 10.5 <0.135 0.109 0.015 13.3 <0.135 < 0.104 - - t0.135 Mg 1.913 0.040 2.11 1.848 0.260 0.006 2.19 0.406 1.253 0.033 2.63 1.273 Na 0.720 0.189 26.2 - 0.277 0.127 45.6 - 0.255 0.225 88.3 - Table 8 Recovery values calculated for the 2.017 pg g-' standard solution; the analysis was carried out 90 min after calibration Figure of merit Fe c u Ti A1 Cr Ni Ag Mg Na Concentration/pg g - ' 2.114 2.076 2.231 2.127 2.186 2.259 1.969 2.299 2.501 RSD (Yo) 3.14 3.95 3.74 6.11 2.99 3.37 2.55 2.14 17.6 Recovery (%) 104.8 102.9 110.6 105.5 108.4 112.0 97.6 114.0 124.0 and to maintain its accuracy the 2.017 pg g-' standard was analysed after all the samples.This was nearly 90min after the last standard sample had been analysed. Recovery was calculated from this and the values are presented in Table 8 these indicate that the system yields fairly accurate results over a reasonable period of time. The oil analyses described were carried out on oil samples with relatively low viscosity. A viscosity study of a more viscous motor car lubricating oil Castrol GTX (SAE 20W 50) revealed that the viscosity of this oil at 123 "C is quite similar to the viscosity of the oils used in this study at 85°C. The viscosity of Castrol GTX is also related to temperature through a power function q = 170.813 T-2.024 so that it is possible to identify a temperature range where the viscosity is relatively unaffected by small fluctuations in temperature.Conclusion It was shown that the nebulization system reported on pre- viously," can be used to introduce edible oil and lubricating oils directly into the ICP. Factors that determine optimum signals were optimized i.e. temperature of the oil oil flow rate and plasma power. Detection limits attainable with this method for the analysis of edible oils were in the range of low metal content. Detection limits obtained with lubricating oils com- pared favourably with others reported in the literature. to the laboratories of the SAAF for supplying the samples of aircraft lubricating oil and the associated information. 1 2 3 4 5 6 7 8 9 10 11 12 13 References Reilly C. Metal Contamination of Food Applied Science Publishers London 1980. Autoxidation and Antioxidants Volume 11 ed. Lundberg W. O. Interscience New York 1962. Hsgtahl 0. T. and Melsom S. Anal. Chem. 1965 38 1414. Labuza T. P. and Karel M. J. Food Sci. 1967 32 572. Abbott D. C. and Polhill R. D. A. Analyst 1954 79 547. Gambrill G. M. Gassmann A. G. and O'Neil W. R. Anal. Chem.,'1971 23 1365. List G. R. Evans C. D. and Kwolek W. F. J. Am. Oil. Chem. SOC. 1971 48 438. Hendrikse P. W. Slikkerveer F. J. Zaalberg J. and Hautfenne A. Pure Appl. Chem. 1988 60 893. Fassel V. A. Peterson C. A. Abercomie F. N. and Kniseley R. N. Anal. Chem. 1976 48 516. Merryfield R. N. and Loyd R. C. Anal. Chem. 1979 51 1965. Algeo J. D. Heine D. R. Phillips H. A. Hoek F. B. G. Schneider M. R. Freelin J. M. and Denton M. B. Spectrochim. Acta Part B 1985 40 1447. Rademeyer C. J. and Fischer J. L. J. Anal. Atom. Spectrom. 1993 8 487. Brocas J. J. Analusis 1982 10 389. Paper 31045 73K Received July 30 1993 Accepted January 1 1 1994 The authors thank Mr H. Otto of Atlas Aviation associated

ISSN:0267-9477    

DOI:10.1039/JA9940900623

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 629 Comparative Studies of Surfatron and Microwave Plasma Torch Sources for Determination of Mercury by Atomic Emission Spectrometry Yixiang Duan Xiaoguang Du and Qinhan Jin* Department of Chemistry Jilin University Changchun 130023 China Two cavities are compared for the determination of mercury with aqueous sample introduction by an ultrasonic nebulizer. The performances of the cavities surfatron and microwave plasma torch (MPT) are compared with particular emphasis on the characteristics of plasma features the effects of operational conditions tolerance to foreign material or the presence of other elements and analytical figures of merit. The effect of potassium chloride and ammonium chloride on the analytical performance of MIP (surfatron)- AES and MPT-AES has been examined and a possible mechanism is discussed.An improvement of about 3-fold in detection limits was achieved by adding an appropriate amount of 0.05% m/v of ammonium chloride. The detection limits for mercury by surfatron AES and MPT-AES are 0.9 and 1.3 ng ml-’ respect- ively. The precision obtained in this work is below 2.0. Stable discharges were obtained in both cavities with use of a low-power microwave generator. Keywords Microwave-induced plasma surfatron atomic emission spectrometry mercury microwave plasma torch The microwave-induced plasma (MIP) is an attractive exci- tation source for atomic spectrometry since it possesses high electron temperature and relatively low gas temperature allowing atomic and ionic species including non-metal to be excited.’T2 In addition the analyte emission is usually measured against a simpler and less intense background spectrum in the MIP system than that observed with other emission sources such as flame and inductively coupled plasma (ICPj.Moreover the overall cost of the MIP system is low compared with for instance that of an ICP source; the microwave generator for example is readily available less expensive and usually more compact than the radiofrequency generator which is necessary for ICP. Because of these advantages interest in the MIP as an excitation source for atomic emission has been revived in the past decade334. Improvements in plasma torch d e ~ i g n ~ . ~ cavity tuning and coupling7.’ and the development of a moderate- power system%” as well as the microwave-induced nitrogen discharge at atmospheric pressure,12-13 have greatly added to the attractiveness of the MIP source for atomic spectrometry.The analytical performance of MIP sources mainly depends on the properties of the plasma working gas the operational conditions of the plasma sample introduction techniques and the coupling devices for generating plasma. Many aspects of this technique are at present under scrutiny one of these being the design and/or modification of the plasma coupling device. Both cylindrical and rectangular cavities have been developed and used for atomic spectrometry. Among these devices the surface-wave induced plasma device known as ‘surfatron’ developed by Moisan et a2.14 and the microwave plasma torch (MPT) developed by Jin et aL5 serve as promising sources for atomic emission spectrometry (AES j.The performance of such devices for the atomic spectrometric determination of mercury has been studied in this work. Argon was used as the working gas and an ultrasonic nebulizer was used for sample introduction. To illustrate their analytical performance aspects of the sources such as characteristics of the plasma operational condition effects tolerance to foreign material or the presence of other elements and analytical figures of merit are considered. Preliminary tests indicate that stable discharges could be obtained in both of the cavities with a low-power microwave generator. * To whom correspondence should be addressed. Experiment a1 Instrumentation The instrumentation used is shown schematically (Fig.1). The microwave power is delivered at 2450MHz by a microwave power supply (DW-2 Beijing Geological Instrument Factory China). The power transmitted to the cavities is taken as the difference between ‘the incident and reflected power which are monitored by the built-in power meters. The reflected power can be maintained at 0 W by proper tuning of the microwave structures for both surfatron and MPT The microwave struc- tures are mounted on a translation stage which allows adjust- ment for lens focusing. Sample solutions contained in a Pyrex glass cell are introduced into the plasma by using an ultrasonic nebulizer based on a commercial humidifier (CSW-1 Shanto Optic-Electronic Instrument Company China).” This nebul- izer can be operated at a lower carrier gas flow rate than is required by the concentric nebulizer typically used with the MIP-AES system.As the rate of aerosol formation is controlled by the power applied to the piezoelectric crystal of the ultra- sonic nebulizer the nebulizer efficiency is independent of aerosol carrier gas flow rate so that the latter can be indepen- dently adjusted to optimize the residence time of the analyte in the plasma discharge. The sample uptake rate is usually around 0.4 ml min-’ depending on the carrier gas flow rate. A desolvation system was employed in all studies. The desolvation apparatus is similar to that described elsewhere,16 and consists of a quartz heating tube (6 mm i.d. 8 mm Recorder Ultasonic nebulizer Fig.1 plasma torch; PMT photomultiplier tube Schematic diagram of the instrumental array MPT microwave630 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 0.d. x 180 mm long) a tap-water-cooling condenser and a concentrated sulfuric acid desiccator. An ohmic heating element is wrapped outside the surface of the quartz tubing. A transformer is used to control the voltage necessary to maintain the tube at approximately 160°C for drying the aerosol. The water vapour is then removed by the water-cooling condenser at about 8 "C and a concentrated sulfuric acid desiccator held in a glass vessel (20 mm i.d. x 100 mm long). The dried aerosol is introduced directly into the plasma by the carrier gas. Emission from the plasma is focused on to the entrance slit of a monochromator (WDG 30 Beijing Optic Instrument Factory) with a lens assembly; the monochromator entrance slit is usually fixed at 10 pm.The final signals amplified by a photomultiplier tube (PMT) are delivered to a chart-feed recorder (XWT- 164 Dahua Factory Shanghai China) and recorded. The line used for the determination of mercury is at 253.7 nm which gives an optimum signal-to-background ratio. All the data presented here are the background-corrected values. Two different viewing geometries (axial and radial) were used for surfatron and MPT systems respectively. The axial (end-on) method of viewing which is the most common geometry for MIP systems was used for the surfatron device. In this configuration the emitted radiation from the plasma inside the cavity was collected with a lens and focused on to the monochromator with unity magnification.The radial (side- on) viewing mode was used for the MPT as described by Jin et uL5 In this instance the radiation from the plasma was collected with a quartz lens and imaged on to the entrance slit of the monochromator with a magnification of 0.5. Although this optical arrangement was selected for operational con- venience in this work it can be improved by using an appro- priate aperture lens.5 Cavities Surfutron The surfatron used was made of brass based on the original design of Moison et ~ 1 . ' ~ The input and coupling section consists of a Teflon-filled coaxial tube (inner connector) which is terminated by a standard N-type connector at the input end and a circular coupling plate at the other end.The entire input section can be translated up or down by an adjustment screw. This adjustment alters the capacitance associated with the coupling plate and allows impedance matching of the surfatron to that of the transmission line and the generator. The wave- forming structure consists of a coaxial chamber terminated by adjustable cylinders at one end and by a thin brass end-plate at the other. A gap exists between the end-plate and the inner cylinder and surfatron waves are launched from this gap along a quartz tube (2 mm i d . and 4 mm 0.d.). The launching efficiency can be maximized by adjustment of the length of the gap. A 1.0m length of coaxial cable was employed as a transmission line. Once the tuning adjustments are set to near- optimum positions a plasma could be initiated by striking the outlet of the quartz tube with a short burst from a Tesla coil.Initiation of the plasma with a length of tungsten wire although often employed in the Beenakker cavity or MPT is not readily achieved with the surfatron. MPT The microwave plasma torch used in this study consists of three concentric metal tubes and possesses an ICP-torch-like configuration. The outer tube is made of brass (26 mm 0.d. and 22 mm id.) while the intermediate tube (5.5 mm 0.d. and 4.5 i.d.) and central tube (3 mm 0.d. and 2 mm id.) are made of copper. The pIasma support gas is introduced through the intermediate tube and the sample aerosol is introduced through the central tube by a carrier gas. A 1.0m length of coaxial cable was used as a transmission line.A detailed description of MPT structure tuning and ignition is given in ref. 5. Reagents All stock solutions for matrix studies were prepared either from analytical-reagent grade soluble salts or metals dissolved in a minimal amount of acid and diluted to the desired concentration with de-ionized distilled water. The stock solu- tion for mercury (1000 1-18 ml-') was obtained by dissolving HgCl in 1 mol I-' HC1. Volumetric dilution of stock solution was performed daily to obtain working solutions. De-ionized distilled water was used throughout. The purity of argon used as working gas is 99.99%. Results and Discussion Plasma Features Surfutron After initiation the plasma forms in the vicinity of the gap and extends outwards several centimetres to an extent mainly depending on the input power and gas flow rate.The wet aerosol generally caused the plasma to shrink and to cling to the wall of the discharge tube. However a stable plasma could be maintained when a desolvation-dessicator system was used to remove most of the water vapour from the aerosol. Being operated at atmospheric pressure a dry argon plasma has a violet filament appearance while a wet argon plasma is pre- dominately pink. Since the microwave power selected for surfatron operation is only about 50 W one filament plasma was usually observed under this experimental condition. However the number of filaments in the plasma discharge changes with increasing microwave power and carrier gas flow rate as indicated in ref.17. MPT After ignition the plasma forms between the intermediate and central tubes near the top of the torch and extends outwards several centimetres into the air to an extent mainly depending on input power and gas flow rates. The flame-like argon plasma takes a cone shape in which the central channel is less luminous than the principal plasma cone. The argon plasma contains four distinct regions which are distinguishable by colour and by shape. The first region between the top of the torch and the plasma core is robust and brilliant. The second region the plasma core is often used as the observation zone for atomic emission measurements and is brilliant white. The background emission in this portion is stronger than in any other area. The third region between the plasma core and the tail plume gradually merges with the fourth region the tail plume. Both third and fourth regions are better for atomic fluorescence spectrometry as shown in previous work.'*~lg Generally the plasma tail plume is pale yellow. Impedance matching between the torch and the microwave generator is largely independent of changes in plasma param- eters.Therefore an argon plasma formed in an MPT is more uniform and stable than a traditional plasma. Though the MPT plasma has been sustained for long continuous periods (> 12 h) the cavity temperature has never exceeded 50 "C. This increase in temperature is gradual over the entire period of operation. With extended use there is no noticeable change in either the appearance of the plasma or in the emission intensity of the analyte.Optimization of Plasma Parameters As there are several parameters that affect the analytical performance an effort was made to optimize the other exper- imental parameters before optimization of the plasma param- eters. The operational conditions used are those listed inJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 63 1 Table 1 Operational conditions for surfatron and MPT in the determination of mercury Parameter Microwave frequency (MHz) Microwave forward power (W) Reflected power (W) Carrier gas flow rate (ml min-') Support gas flow rate (ml min-') Plasma viewing mode Ultrasonic nebulizer frequency (MHz) Desolvation temperature ("C) Sample solution concentration (ng 1-' of Hg) Sample solution medium Sample solution volume (ml) (batch addition) Sample uptake rate (ml min-') Measurement mode Surfatron MPT 2450 50 0 500 end-on 160 100 0.05% NH4Cl -0.03 moll-' HCI 25 1.4 2450 50 0 600 600 side-on 160 100 0.05% NH4Cl -0.05 mol 1-' HC1 25 1.4 0.38 0.38 Constant signal magnitude Table 1 unless otherwise stated.The analyte concentration for the optimization studies was 0.1 pg ml-'. Microwave power The effect of microwave power on the mercury signals for both surfatron and MPT sources is shown in Fig. 2 when other experimental conditions were as indicated in Table 1. The emission signals of mercury increase with increasing microwave power of 40 W or below. Above this level the power curve for the MPT tends to level off while for the surfatron it increases slowly. Higher power generally led to poorer precision. This was true for both surfatron and MPT systems.In the interest of both signal intensity and precision 50 W was selected as an acceptable power for both surfatron and MPT. This lower power may be attributed to the special characteristics of mercury which is volatile and easily atomized even at lower temperature. Under this power condition both surfatron and MPT can form very stable argon plasmas. Carrier gasjlow rate The MPT has two gas channels namely carrier gas which introduces sample into the plasma and support gas which is used to maintain the plasma discharge. The effect of the support gas flow rate on the determination of mercury was examined in the MPT system in the range from 200 to 700mlmin-'. The mercury emission intensity was doubled when the support gas flow rate increased from 200-400mlmin-1.However in the range from 400-700 ml min-' there was almost no change in intensity. A flow rate of 600 ml min-' was chosen as the optimum value. The effect of carrier gas flow rate on signal intensity for both surfatron and MPT is shown in Fig. 3. Generally with an increase of carrier gas flow rate the amount of sample aerosol introduced into the plasma should be increased and 120.0 I 1 I I > I ,'- I 10 20 30 40 50 60 70 80 90 Microwave forward powerW Fig.2 Effect of microwave power on mercury signal intensity A MPT; and B surfatron. Experimental conditions as listed in Table 1 lead to an increase in signal intensity. This effect is shown (Fig. 3) in carrier gas flow rates ranging from 200 to 400 ml min-' (for surfatron) or 200-500 ml min-' (for MPT).However as expected the higher carrier gas flow rate will results in a shorter residence time of analyte in the plasma and also decrease the analyte number density in the plasma dis- charge. These effects are obvious even at high carrier gas flow rates (Fig. 3). The optimum carrier gas flow rates obtained in this work for surfatron and MPT are 500-600 ml min-' respectively. Generally the MPT system requires a slightly larger carrier gas flow rate than the surfatron does. Sample Solution Medium and Acidity To demonstrate the effects of medium and concentration on the mercury emission intensity and plasma performance both HC1 and HNO which are commonly used in real sample analysis have been examined in surfatron and MPT systems (Fig.4).Both HCl and HNO slightly enhance the mercury emission signals in a concentration range from 0 to 25 mmol 1- '. At higher concentrations the emission signal intensity does not change significantly with increasing concen- tration of acids up to 0.5 moll-'. This is true for both HC1 and HNO also for both surfatron and MPT. However for the surfatron system the concentration of HNO should not be > 0.2 mol l-' otherwise the plasma shrinks significantly and becomes unstable. For the MPT system the emission signal of mercury in HCl medium is slightly higher than that in HNO,. In contrast to the surfatron however for up to 0.5 moll-' which is the maximum concentration of HNO examined no obvious effect was found in the MPT system.Thus the MPT has stronger tolerance to sample solution than the surfatron does. The main reason for this advantage of the MPT source may be its central channel plasma which is of benefit for sample introduction and the interaction between 90.0 1 1 f? 80.0 70.0 .- 8 L . 60.0 50.0 2 40.0 5 30.0 e > Y .- al Y 100 300 500 20.0 700 Carrier gas flow rate/ml min-' 900 Fig.3 Effect of carrier gas flow rate on mercury signal intensity A MPT; and B surfatron. Experimental conditions as listcd in Table 1632 c .- rn C 2 20.0 - JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 ' 40.0 20.0 4 I 1 I I I 1 0 0.10 0.20 0.30 0.40 0.50 0.60 [Acidl/mol I - ' Fig.4 Effects of sample solution medium and acidity on mercury signal intensity A HCl (MPT); B HNO (MPT) and C HCl (surfatron).Experimental conditions as listed in Table 1 I 1 I I 1 I I 0 5 10 15 20 25 30 [KCll/mmol I - ' Effect of concentration of KCl on the analyte signal intensity Fig. 5 using the surfatron system at a microwave power of 80 W 0 0.05 0.10 0.15 0.20 0.25 [NHdCI] (%) Fig.6 Effect of concentration of NH,C1 on the signal intensity A MPT; and B surfatron. Experimental conditions are listed in Table 1 sample and plasma. Furthermore the separate introduction of carrier and support gas into the plasma should also increase the plasma tolerance and stability. Concentration of Potassium Chloride Since the microwave plasma is a substantial departure from the local thermal equilibrium (LTE),' ionization interference can be expected to be significant in the microwave discharge.Although some other plasmas such as ICP are also a departure from the LTE the extent of such a departure is different in different plasma discharges. For example the gas temperature (T,) and electron temperature (T,) in the ICP are 6430 and 7520 K respectively,20 while in the MPT the electron tempera- ture is as high as 21 500 K whereas gas temperatures are generally lower by a factor of 2-10 depending on the spatial position in the discharge and experimental conditions.21 Such a considerable departure from LTE in the MIP or MPT discharge inevitably cause greater ionization interference than in an ICP. The presence of an easily ionized element (EIE) in microwave plasma is likely to result in an increase in the electron number density which will not only affect plasma properties for example increasing the electric and thermal conductivity of the plasma and improving the coupling of microwave energy but also moving the ionization equilibrium of the analyte towards the formation of neutral atoms and hence enhancing the atomic emission of the a n a l ~ t e .~ ~ - ~ ~ This phenomenon was also observed in this experiment when potassium was used as EIE for the determination of mercury. The effect of KC1 concentrations on the mercury signal obtained by the surfatron system is shown in Fig. 5. As the concentration of KC1 increases from 0 to 12mmoll-' the intensity of the mercury line increases by about 70% and remains unchanged in the concentration range from 12-24 mmol 1-I. Analogous results are obtained with the MPT system.However the addition of a relatively large amount of EIE can lead to poor plasma stability at 50 W power. This result suggests that 50 W is too low to maintain the ionization of EIE and plasma stability simultaneously since a steady- state microwave discharge can be sustained only when the power absorbed by the plasma equals the power loss from the In this instance a maximum power of 80 W was used to keep the plasma stability for both the surfatron and MPT systems. Interestingly the addition of KCl not only enhances the analyte signal but also increases the tolerance of microwave plasmas to the effects of interfering elements. This effect is valid for both surfatron and the MPT systems. Matrix effects on the determination of mercury with or without KCl (14 moll-') added are listed in Table 2.The ratio of matrix to analyte in Table 2 for example is 2000 which means that the concentration of a single interferent in the solution is Table 2 Maximum amount of matrix allowed to exist in the determinaiion of mercury* Sample solution 0.1 pg ml-' Hg 0.1 pg ml-' Hg-14 mol 1-' KCl Matrix/analyte ratio 100 200 300 500 1000 1000 1500 2000 3000 Surfatron Al Au Be Cd Co Cr Fe K La Sb Sr Te W Zn Zr Ag Bi C1 Cu Mn Ca Pb Ba As Bi Br Ca Ni I Cr Ge Al Ba Cd Cu Na Pb Se Te W Zn K Mn MPT As Au Cd C1 Co Fe K Na Al Ag Be Br Cr Ge P Si Sr Te W Cu Mn Sb Ca La Pb Se Ba Bi Zn Zr Bi Br Cr I Ba Ca Fe As Cu Ge Na Ni Se Te W Zn Al Cd Mn Pb K * Matrix concentration causing recovery to drop below 95%.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 633 Table 3 Analytical figures of merit for determination of mercury ~~ ~~~ ~ Figure of merit Surfatron* MPT* MIP-AES Without NH4CI With NH,CI With NH,CI Evensen added added added Cavity? TMOIO§ Detection limit (ng ml-') 3.3 0.9 1.3 3 36 Precision (relative standard deviation) (%) 2.0 1.7 1.4 2 4.2 * This work. t Reference 25. 5 Reference 7. 200 pg ml-I since the concentration of analyte examined is 0.1 pg m1-l. To maintain the plasma working stability 80 W power was used throughout the investigation of the matrix. Though signal suppression was frequently observed when the concentration of matrix was higher than the maximum values shown in Table 2 in some cases for example when EIEs were used as interferents signal enhancement was also observed.A few more non-metal elements were tested in the MPT than in the surfatron in the absence of KCl. The addition of KCl significantly increases tolerance to interfering elements. For the determination of mercury the maximum amount of matrix that can be added to a sample solution is raised by factors of hundreds to thousands. This amount is defined as the maxi- mum concentration of matrix that will not cause a significant effect (<5%) on the signal intensity. Addition of Ammonium Chloride Generally the addition of ammonium chloride to a sample solution not only improves the precision but lowers the detection limits. Addition of 0.05% m/v NH4C1 (Fig. 6) increases the mercury signal magnitude by about loo% both in the surfatron and in the MPT system.However in the surfatron system a plateau in the NH,C1 effect curve was found in the concentration range 0.02-0.10%. In the MPT system from 0.02 to 0.2% NH4C1 there is no obvious decrease. This may again demonstrate that the MPT has a stronger tolerance to foreign materials. A quantitative comparison of detection limits and precision with and without the addition of NH4C1 is given in Table 3 for the surfatron system. Compared with the system in the absence of NH,Cl the addition of NH4Cl lowers the detection limits by as much as nearly 3-fold. This improvement of detection limits and pre- cision can be attributed at least in part to the fact that NH4Cl is heat-labile (sublimation temperature about 335 "C). When NH4C1 was introduced into the plasma together with the sample solution sample decomposition was speeded up.This may be the reason for the signal enhancement and the detection limit improvement. Analytical Figures of Merit Many factors contribute to the detection limit attainable with microwave plasma systems such as the nature of the matrix used operating pressure support or carrier gas type and flow rate applied microwave power and plasma viewing mode. The sample introduction system and the plasma sustaining assembly also make important contributions. Under the exper- imental conditions listed in Table 1 the detection limits for the determination of mercury were calculated as prescribed by IUPAC guidelines. For calculation of the standard deviation eleven continuous readings were taken. Similar detection limits for surfatron and MPT systems were obtained (Table 3) although that for MPT is a little poorer.However as discussed above the optical system for the MPT is easily improved by use of a proper lens arrangement. A way of improving the detection limit for the MPT by using the 'end-on' viewing mode is currently being investigated. The linear dynamic range obtained here is about 3 orders of magnitude for the surfatron. A slightly larger dynamic range was obtained for the MPT from about one order of magnitude over the detection limit to an upper level of about 50 pg rn1-I. Though this dynamic range is not ideal it is acceptable for real sample analysis. Conclusion The primary results show that both the surfatron and the MPT are promising excitation sources for atomic spectrometry.The higher tolerance to foreign materials and the excellent plasma working stability make the MPT especially attractive for real sample analysis. This work was sponsored by the National Natural Science Foundation of China. We thank Hanqi Zhang Department of Chemistry Jilin University for comments. 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 Matousek J. P. Orr B. J. and Selby M. Prog. Anal. At. Spectrosc. 1984 7 275. Matusiewicz H. Spectrochim. Acta. Rev. 1990 13 47. Broekaert J. A. C. Anal. Chim. Acta. 1987 196 1. Dahmen J. ICP In$ Newsl. 1990 16 321. Jin Q. Zhu C. Borer M. W. and Heiftje G. M. Spectrochim. Acta Part B. 1991 46 417. Matusiewicz H. Spectrochim. Acta Part B 1992. 47 1221. Haas D. L. and Caruso J. A.Anal. Chem. 1984 56 2014. Brown P. G. Haas D. L. Workman J. M. Caruso J. A. and Fricke F. L. Anal. Chem. 1987 59 1433. Michlewicz K. G. and Carnahan J. W. Anal. Chem. 1985 57 1092. Michlewicz K. G. and Carnahan J. W. Anal. Chem. 1986 58 3122. Cull K. B. and Carnahan J. W. Appl. Spectrosc. 1988 42 1061. Deutsch R. D. and Hieftje G. M. AppI. Spectrosc. 1985 39 214. Urh J. J. and Carnahan J. W. Appl. Spectrosc. 1986 40 877. Moisan M. Patal R. Hubert J. Bloyet E. Leprince P. Marec J. and Ricard A. J. Microwave Power 1979 14 57. Duan Y. Huo M. Du Z. and Jin Q. Appl. Spectrosc. 1993 47 1871. Duan Y. Zhang H. Huo M and Jin Q. Spectrochim. Acta Part B in the press. Bulska E. Broekaert J. A. C. Tschopel P. and Tolg G. Anal. Chim. Acta 1993 276 377. Duan Y. Kong X. Zhang H. Liu J. and Jin Q. J. Anal. At. Spectrom. 1992 7 7. Duan Y. Du X. Li Y. and Jin Q. in preparation. Hanselman D. S. Sesi N. N. Huang M. and Hieftje G. M. Spectrochim. Acta Part B in the press. Huang M. Hanselman D. S. Jin Q. and Hieftje G. M. Spectrochim. Acta Part B 1990 45B 1339. Kawaguchi H. Hasegawa M. and Mizuike A. Spectrochim. Acta Part B 1972 27 205. Skogerboe R. K. and Coleman G. N. Anal. Chem. 1976 48 611A. Zander A. T. and Hieftje G. M. Appl. Spectrosc. 1981 35 357. Lichte F. E. and Skogerboe R. K. Anal. Chem. 1973 45 399. Paper 3/054 74 H Received September 13 1993 Accepted January 1 I 1994

ISSN:0267-9477    

DOI:10.1039/JA9940900629

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 635 Capacitively Coupled Plasma with Tip-ring Electrode Geometry for Atomic Emission Spectrometry. Analytical Performance and Matrix Effect of Sodium Chloride and Potassium Chloride Emil A. Cordos Sorin D. Anghel and Tiberiu Frentiu Department of Chemistry University of Cluj 3400 Cluj-Napoca Romania Adrian Popescu Research Centre for Analytical Instrumentation P.O. Box 71 7 Of. P.5 3400 Cluj-Napoca Romania A radiofrequency (r.f.) capacitively coupled plasma of low to medium power is described and characterized for atomic emission analysis of pneumatically nebulized sample solutions. The plasma is generated at 27.12 MHz in a torch with tip-ring electrode geometry operated at 85-275 W with argon as support gas. The r.f.discharge is obtained at the tip of a sharp platinum electrode placed inside a quartz tube while the second electrode is a ring yutside the tube. The limits of detection for Li Cr Ca Pb Fe Cu Mg Hg Cd and Zn were in the ng ml- to pg ml-’ range. The introduction of NaCl and KCI in concentrations up to 500 pg ml-’ in Na or K produces an enhanced matrix effect for most of the elements the highest values being obtained for easily ionizable elements. The matrix effect is less intense at higher power. Keywords Capacitively coupled plasma atomic emission spectrometry; tip-ring electrode geometry; matrix effect Radiofrequency (r.f.) generated plasmas at atmospheric press- ure are commonly used as spectral sources for a wide range of analytical applications. Most of these sources are inductively coupled plasmas (ICPs).The r.f. capacitively coupled plasma as a physical phenomenon was studied before ICPs. They were not used as analytical spectral sources although some attempts to do so were made about three decades ago.’.’ However in recent years some attention has been given to capacitively coupled plasmas (CCP) generated by microwaves (CMP),* or radiofrequency (r.f.) CCP which could also be maintained at atmospheric pressure and lower power i.e. less than 100 W. Blades and co-workers9-14 have reported the use of an r.f. CCP at powers between 30 and 600 W in combination with an electrothermal vaporization system for detection by atomic absorption spectrometry (AAS) atomic emission spec- trometry (AES) and gas chromatography (GC).The r.f. CCP in combination with a furnace atomizing system known as FAPES (furnace atomization plasma emission spectrometry) proved to be very effective and has been studied both by Smith et ~ 1 . ’ ~ and by Sturgeon and co-w~rkers.’~-~~ A spectroscopic device based on FAPES can be fitted into existing AA and AE spectrometers and such a system is commercially available. An r.f. CCP has been developed by Gross et d2’ for element specific detection in GC. The electrode geometry for the r.f. CCP in the papers mentioned above was coaxial. The r.f. CCP could also be maintained in a tip-ring electrode geometry c~nfiguration~’.~~ and has been ~ s e d ’ ~ ~ to develop a low power spectral source for pneumatically nebulized samples. The r.f. discharge was unipolar having direct contact with only one electrode placed inside a quartz tube while the other electrode was external.The discharge could be easily operated in air or argon. In the present paper some of the analytical data obtained with this source for AES are reported. The support gas was argon. The sensitivities and the limits of detection (LODs) for ten test elements were determined and are compared with the results obtained with a low power capacitively coupled micro- wave atmospheric pressure plasma by Winefordner and co-worker~.~,~ The source was also tested for the determination of Pb in airborne particulate matter in samples collected from an industrial area. The advantages of lower or medium power plasmas are their accessibility and ease of handling. However the analytical signal is subject to higher matrix effects than in higher power plasmas as has been demonstrated by Boumans et ~ 1 .’ ~ and Hwang et ~ 1 . ’ ~ for CMP. Whereas a number of papers on interferences in CMP have been p~blished’~~’ there are rela- tively few data on interferences in the r.f. CCP. Smith et a l l 2 and Sturgeon et al.17 have reported the influence of an NaCl and an NaNO matrix on Ag and on Pb in a FAPES source respectively. Matrix interferences in an r.f. CCP have been reported by West and Hume3’ for Ca Sr and Zn. In the second part of this paper the influence of the most common saline matrices with low ionization potential namely NaCl and KCl on the r.f. CCP mentioned above are investigated. Some of the data obtained are comparable with data obtained using a CMP especially by Murayama et aL3’ and Kitagawa and T a k e ~ k i .~ ~ Experimental Instrumentation A diagram of the experimental set-up is provided in Fig. 1. The plasma was operated with a 27.12 MHz r.f. generator and Plasma Upper ring Focalization torch electrode lens Radiofrequency Fig. 1 Block diagram of the experimental set-up636 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 the power could be varied in steps. The oscillator was 'free running' and was designed and built by the present initially for ICP and later adapted for r.f. CCP. It can generate powers of up to 2 kW but for the present r.f. CCP experiments the power was limited to the 85-275 W range. The r.f. discharge was unipolar and was maintained between a water-cooled electrode with a sharp platinum tip which was connected to the r.f.high voltage. The assembly was placed inside an 18 mm i.d. quartz tube and a counter electrode was placed outside the tube and connected to ground. The outer electrode was a metallic ring 25 mm in diameter placed at a height of about 65 mm from the inner electrode tip. The discharge was ignited by means of a Tesla coil and could be easily maintained at atmospheric pressure in air argon or a mixture of both. The gas flow both for nebulization and plasma support was 0.6 1 min-'. The plasma emission was focused onto the entrance slit of the monochromator by a 110 mm focal length fused- silica lens. Two types of optical systems were associated with the plasma. For recording spectra a Heath EU-700 mono- chromator equipped with an RCA 1P28A photomultiplier and Heath EU-701 high voltage power supply was used.The photocurrent was recorded by a K 201 recorder (Zeiss-Jena). For quantitative measurement the above system was replaced by an optical and data processing system designed for the ICP.3s The system consists of a 1 m focal length scanning monochromator with 25 pm resolution and the associated electronics. Both the monochromator drive and data pro- cessing are accomplished by a Telerom computer Model 3P 86(IEIA). Details of the equipment are given in Table 1. The samples were nebulized by using a Meinhard nebulizer in a 120 ml nebulization chamber. The nebulized samples were introduced into the plasma by means of the device shown in more detail in Fig.2. This device is comprised of two poly( tetrafluoroethylene) (PTFE) pieces which form a mixing chamber where the nebulized sample can be mixed if necessary with a supplemen- tary flow of gas. From here the sample is swept into the base of the plasma uia a trough containing 12 holes each 1.3 mm in diameter concentrically placed around the lower electrode Table 1 Instrumentation and operating conditions Plasma torch Optics for spectra recording Plasma power supply Plasma r.f. generator Model EOP 27.12 MHz 2 kW free running oscillator modified for operating at low power 85-275 W. (Research Centre for Analytical Instrumentation Cluj-Napoca Romania) Laboratory constructed Monochromator Heath EU 700 0.35 m focal length 1200 grooves mm-' grating Photomultiplier RCA 1P28A operated at 600 V.Photomultiplier power supply Heath EU 701 (Heath Co. Benton Harbor MI USA) K 201 Zeiss-Jena Jena Germany Scanning monochromator and Recorder Optics for quantitative determinations photocurrent measurement system Model SMS Czerny-Turner mount 1 m focal length 2400 grooves mm-' grating blazed at 340 nm spectral bandpass 25 pm entrance and exit slits 20 pm. Internal wavelength calibration with a Si hollow cathode lamp (Research Centre for Analytical Instrumentation) 2020 DAF display (IEIA Cluj- Napoca Romania) and Control Data fast printer. Interface laboratory constructed 64 ps data acquisition time in-house software Monochromator driving data acquisition and processing Telerom Computer Model 3P 86 with Intel Lower electrode ' 1 10 mm H Cooling water Fig.2 Schematic diagram of the torch assembly 5mm from the tip of the electrode. The dimensions of the mixing chamber and the diameter of the holes provide for a laminar flow of gas up to 1.6 1 min-' for each port. Reagents Stock solutions (1000 yg ml-') were prepared by dissolution of the high-purity metals or their salts in HNO (1 + 1) for Cd Cu Fe Hg and Pb or in HC1 ( l - t l ) for Cr Mg Zn CaCO and LiCO,. Single element working standards of 50 yg ml-I were obtained by diluting the stock solution with high-purity 2% v/v HNO,. For the determination of Pb in environmental samples the concentration range of the working standards were 1-200 pg ml-'. The dust samples were collected on cellulose membrane filters at an air flow rate of 20 1 min-' and a total air volume of 6701.The filters were dissolved in HNO and diluted to 25 ml with 2% v/v HNO,. For matrix interference measurements stock solutions of 1000 pg ml-' Na or K were prepared from the corresponding amount of high-purity NaCl and KCl. The final concentration of standard solutions was 50 pg ml-' in the element under consideration and 0-500 pg ml-' in the matrix element. The intermediate concentrations for matrices were 10 20 50 100 200,300,400 and 500 pg ml- ' Na (as NaCl) and 500 pg ml-' K (as KC1). All solutions were acidified to 2% v/v with HNO,. Procedure for Polarographic Determination of Lead Since there are no certified materials for air particles with a high Pb content the results obtained with the r.f. CCP were compared with those obtained by polarography.The support electrolyte stock solution for polarographic standards was 5 mol 1-' ammonium acetate containing 1 % hydroxylamine clorhydrate and 0.05% gelatine. Working standards of Pb were prepared in the 10-100 pg ml-' range from the corresponding aliquot of lOOOpgml-' Pb stock solution with 6.5ml of support electrolyte boiled for 10 min and after cooling diluted to 25 ml with distilled water. The Pb samples were prepared from 12.5 ml of original dissolved sample following the above procedure.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 637 Results and Discussion Plasma Shape and Spectra The argon plasma has a filament shape with an intense purple core and a barely visible bluish outer mantle. The discharge extends up to the upper electrode above which the heated gas forms a larger bluish plume.It is very stable mechanically and completely noiseless. The plasma shape and especially the core length is power dependent. At 85 and 125 W the core does not exceed 20 mm in height while at 185 W it extends up to the ring electrode (65 mm) and at 275 W is well above it. The width of the plasma mantle gradually increases with power and at 275 W completely fills the quartz tube. Furthermore if higher power or lower argon flow rates were used the plasma heat would affect the quartz tube. The discharge could also be obtained in the single-electrode configuration without the outer electrode. The r.f. field lines close to other ground points but the plasma takes a broom- like shape. It becomes homogeneous with no core wider and mechanically less stable especially in its upper region.It could not be ignited at a power lower than 185 W. Introduction of concentrated NaCl solutions into the plasma influences its shape but the modifications are dependent both on concentration of Na and power. At a low concentration of Na less than 50 pg ml-' the plasma shape is not affected and only Na emission from the mantle is observed. As the concen- tration of Na increases the core becomes shorter the mantle apparently gets wider and finally at higher concentrations the core completely disappears. The yellow coloured plasma then fills the tube and the analytical signal is considerably dimin- ished. At 85 W the core disappears at a matrix concentration of 400 pgml-' Na while at 135 W this limit is 500 pgml-'.At higher powers matrix concentrations of 500 pg ml-' Na are much better tolerated and the analytical signal is less affected. The influence of KC1 is obviously less than that of NaC1 and for up to 500 pg ml-' K there is no change of the core size and shape even at lower powers. When the plasma is operated without the upper electrode the introduction of NaCl produces a visible emission on the outer side so that the discharge develops the shape of a separated flame. The plasma fills the tube and limits the working range to lower concentrations of Na than were observed for the tip-ring configuration. The plasma background spectra were recorded in the 200-800nm range with distilled water as blank and argon or argon and air as support gas.Similar conditions as for the analytical determinations were used argon flow rate 0.6 1 min-l; plasma power 185 W; nebulizer water intake rate 1.4 ml min-'; nebulization efficiency 5%; and observation height 18 mm from the electrode tip. The prominent feature of the emission spectra are the OH bands at 280-285 nm and 302-317 nm the NO bands from 206 to 272 nm the N2 bands from 330 to 380nm and the Ar lines at 390-450nm. The background from 450 to 800nm is smooth. At the same r.f. power the band intensities are dependent on the proportion of air and water in the support gas. An increase of air content results in an increase of the NO and OH bands and a decrease of the N bands. A higher amount of water results in a decrease in the NO and N2 band intensities and an increase in the OH band and Ar line intensities.Sensitivity and Detection Limits The spectral line of each element was recorded by scanning the spectra over 15 points at 2 pm increments each point being the average of 1000 readings with a data acquisition time of 64 ps. The corresponding wavelength-signal pairs were introduced in a Gauss-type equation from which the peak wavelength and intensity were calculated. In order to obtain background data an extra 50 points adjacent to the spectral line were scanned. The linear range for the elements considered above intro- duced as aqueous aerosols in the 185 W plasma was 2-4 orders of magnitude (2 for Cd Hg and Zn 2.5 for Cr Pb and Fe 3.5 for Ca and Mg and 4 for Li). The sensitivities S and limits of detection (LODs) were defined and calculated according to Boumans and c o - ~ o r k e r s ~ ~ - ~ ~ where X A + B is the gross line signal XB is the background signal and co is the analyte concentration.The normalized sensitivity The (LOD) was calculated using the 3a criteria 3RSDB LOD =r= 3RSDB x BEC (3) where RSDB is the relative standard deviation of the back- ground and BEC the background equivalent concentration. For interference studies a correction was made for the change in nebulization efficiency with the matrix concentration. The gross line signals and background signals were obtained as averages of 16 replicate measurements. Both signals were measured as counts from the analogue-to-digital converter of the data aquisition interface. The background signals were measured at 0.1 nm from the spectral lines.The sensitivities RSDB and LODs are listed in Table 2. All data are for a 185 W plasma. The data from Table 2 show that the elements considered above can be divided into three groups Li Cr Ca Cu and Mg with low LODs Pb and Fe with medium LODs and Hg Cd and Zn with higher excitation energy and poorer (i.e. higher) LODs. There are two orders of magnitude between the last and the first group of LODs. The stability of the plasma Table 2 Sensitivities and LODs in r.f. capacitively coupled plasma at 185 W at an observation height of 18 mm Element Li Cr Ca Pb Fe c u Mg Hg Cd Zn Wavelength/ nm 670.78 425.43 422.67 405.78 371.99 324.75 285.21 253.65 228.8 1 213.81 Excitation eV 1.90 2.90 2.94 3.06 3.30 3.82 4.35 4.90 5.41 5.80 energy/ Sensitivity S/coun ts per pg ml-' 5207 269 2185 84 94 446 2384 27 17 10 ~YIom/ ml pg-' 3.62 0.15 0.52 0.03 0.05 0.57 1.64 0.02 0.02 0.02 RSDB 0.80 0.85 0.80 0.62 1 .oo 1 .oo 1.20 0.80 0.72 0.72 W) LOD/ ng ml-' 15 161 45 540 548 51 22 1070 1115 1325638 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 Table 3 Limits of detection for CMP and r.f. CCP at medium powers Limit of detectionlng ml- ' CMP R.f. CCP Element Li Cr Ca Pb Cu Cd Zn Wavelength/ nm 670.68 425.43 422.67 405.78 324.75 228.81 213.86 Ta tubular torch' (325 W) 5 260 650 2900 90 620 5000 W-Pt solid electrode torch' Present work (520 W) (185 W) 8 15 161 400 45 7600 540 51 3200 1115 6800 1325 - - background radiation is remarkable with an RSDB of about 1% over the entire spectrum.Since there are no analytical data from similar r.f. CCPs the results from this work were compared with those from refs. 5 and 7 where a CMP was used. The reasons for selecting these references were some similarities in torch design (the torches used were of single electrode type) the sample introduction system' and medium plasma power was used. The comparative data are given in Table 3 for seven elements. The LODs show the same trend from element to element in all three columns. However with the exception of Li and Cd (for Ta tubular torch) the LODs for the remaining elements are lower in r.f. CCP than in CMP. This difference could originate in the RSDB values which vary from 1.1 to 6.0% in CMP as compared with about 1 % over the entire spectrum for r.f. CCP.The normalized sensitivities were compared for four elements with those published by Boumans et ~ 1 . ~ ~ for ICP and CMP as average values for a variety of matrices. The data are listed in Table 4 and again display the same trend for all of the plasmas considered high normalized sensitivity for Li and Mg and medium values for Pb and Fe. Determination of Lead in Airborne Particulate Matter The parallel determination of lead in the atmosphere by atomic emission in r.f. CCP and by polarography are in good agreement. The results are presented in Table 5. The samples were collected Table 4 Normalized sensitivities for CMP ICP and r.f. CCP. The data for CMP and ICP are from ref. 24 Wavelength/ CMP* CMPt ICPS R.f. CCP Element nm (600 W) (600 W) (700 W) This work Li 670.68 3.0 0.04 0.30 3.62 Pb 405.78 0.2 0.03 0.03 0.03 Fe 371.99 1 .o 0.20 0.30 0.05 Mg 285.21 2.0 0.40 0.40 1.64 * Cs2S04 matrix (2 mg ml-' Cs).Average of the normalized sensitivity obtained in three matrices water CdS04 and (NH,),HPO (2 mg ml-' Cd and 2 mg ml-' PO,). $ Average data from all four matrices. Table 5 Analytical results for lead in airborne particulate matter R.f. CCP Polarograph y Pb/ Pb/ Sample Fgm1-l (Yo) of air pg ml-' (%) of air P1 85.5 1.86 3.19 87.1 1.90 3.25 P2 119.5 2.02 4.46 121.3 0.85 4.53 P3 128.3 0.98 4.79 131.1 1.50 4.89 c,*/ RSD mgm-3 c,*/ RSD mgmP3 * Concentration of analyte in the solution obtained after filter dissolution; n = 5 from a heavily polluted area and have a high content of Pb well above the maximum permitted value of 0.1 mg of Pb per cubic meter of air.The good agreement between the results obtained by polarography and spectrometry is explained by the absence of interfering components especially easily ionized elements in the solutions of the dissolved sample. Plasma Power and Matrix Interferences The matrix effect was determined according to the method of bourn an^,^^ as the ratio of the sensitivities in the presence and in the absence of the interfering matrix Smatrix/Swater. The matrix effects for two Ca lines one atomic and one ionic as functions of Na concentration at four plasma powers are presented in Fig. 3. A strong enhancement effect was observed for all powers but with a different evolution as matrix concentration increases. For 85 W plasmas the enhancement effect is higher at concen- trations of Na lower than 200 pg ml-' for the Ca atomic line and lower than 50 pg ml-' for the Ca ionic line.At concen- trations higher than these the matrix influence on plasma shape results in a signal decrease. At this plasma power concentrations of Na greater than 400 pg ml-' produce a 20 15 10 5 J 1 0 12 X 5i 8 4 -X 300 400 500 100 200 0 INal/pg mi-' Fig. 3 Effect of concentration of Na in matrix on (a) Ca I 422.67 nm line and (b) Ca I1 393.36nm line at A 85; B 135; C 185; and D 275 W. Observation height 18 mmJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 639 v; 0.03 0.02 0.01 0 depressing matrix effect. At 135 185 and 275 W plasma power the matrix effect decreases with power. The matrix enhance- ment effect is particularly high for the 135 W plasma where it reaches a maximum of 20 for the Ca atomic line and 13 for the Ca ionic line.The matrix effect curves for 185 and 275 W are almost superimposed or very close up to 200 pg ml-1 Na followed at higher Na concentration (300-500 pg ml-') by a greater enhancement effect for 185 W and a buffering effect for 275 W both for atomic and ionic lines. The observation of a matrix effect at low Na concentration should also be noted. The slope of matrix effect plots is appreciable in the 10-20 pg ml-I range and then decreases. The phenomenon is similar to the seeding effect of low matrix concentrations of Na reported by M ~ r a y a m a ~ ~ for a CMP. The net analytical signal follows almost the same trend as the matrix effect as can be seen from Fig.4 where the normalized sensitivities for the two Ca lines are plotted as functions of power and matrix concentration. The resulting curves can be divided into two groups one with matrix concentrations of up to 100 pg ml-' Na and the second with matrix concentrations of greater than 100 pg ml-' Na. The first group shows a constant decrease of normalized sensitivity while the second group has a maximum normalized sensitivity at 135 W. The difference between the two groups can be explained by the relation between matrix effect at 85 W and 135 W the two curves crossing at between 100 and 200 pg ml-' Na. For up to 100 pg ml-I Na the matrix enhance- ment effect decreases with power thereby decreasing the nor- malized sensitivities. Above this value the strong enhancement effect at 135 W produces a maximum normalized sensitivity followed by a constant decrease.The increase of background with power also contributes to the decrease of sensitivity with power. From 85 to 275 W the background increases 4 times for the 422.67 nm line and 1.7 times for the 393.36 nm line. This last value explains the slight increase of normalized sensitivity with power for the Ca ionic line at low concen- trations of Na. The KC1 matrix at 500 pg ml-1 K has a similar effect to - 1 1 A I I I I I I 12 8 4 I 2 - 0 0.6 0.4 0.2 n 50 100 150 200 250 300 PowerW Fig. 4 Normalized sensitivities for (a) Ca I 422.67 nm line; and (b) Ca I1 393.36 nm line in A water and for different matrix concentrations of Na B 20; C 50; D 100; E 200; F 400; and G 500 pg ml-I and H in 500 pg ml-' of K Observation height 18 mm Na at the same concentration on normalized sensitivities except at low power where K has practically no influence.It should also be noted that in the absence of NaCl or KCl matrix the normalized sensitivity for both Ca lines depends very little on power. Observation Height The flame-like shape of the plasma and its changes with matrix concentration imply the optimization of observation height for the species being sought. The normalized sensitivity for two atomic lines Cr I 425.43 and Zn I 213.81 are shown in Fig. 5. They are plotted as functions of observation height measured from the tip of the lower electrode in the absence and in the presence of 200 pg ml-I Na at 185 W plasma power. The two lines were selected for their locations in two relatively distant spectral zones and for their sensitivities about one order of magnitude apart.The analytical signal depends both on obser- vation height and matrix. However there is a relatively large zone starting from 16mm above the electrode where the normalized sensitivity does not change significantly both for pure standards and for matrices. An observation height of 18 mm was selected for the determinations since it provides a zone where changes in plasma shape caused by matrix have a minimum effect on analytical signal. Matrix Effect on Normalized Sensitivity and Limit of Detection The matrix effect on normalized sensitivity and LOD for ten elements was studied. The results for NaCl and KC1 matrix effects RSD of the background and LOD for Li Cr Ca Pb Fe Cu Mg Hg Cd and Zn are summarized in Tables 6 and 7.All data were obtained at a plasma power of 185 W and the emission was measured at a height of 18 mm from the lower electrode. Some of the matrix effects on Ca Cr and Zn have already been described above. For many of the elements considered the presence of NaCl and KC1 matrix has an enhancement effect on the signal which is sometimes appreciable (Li and Ca). For easily ionizable elements e.g. Cr Ca and Cd the enhancement effect is present for all Na and K matrices and is greater in the KC1 matrix. 0.4 0.3 0.2 0.1 r I w 3. 5 10 15 20 25640 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VGL. 9 Table 6 Normalized sensitivities and matrix effect as a function of NaCl and KC1 matrix concentration in the range 0-500 pg ml-l expressed as Na or K content.Plasma power 185 W observation height 18 mm Element Li Cr Ca Pb Fe c u Mg Hg Cd Zn Wavelength/ nm 670.78 425.43 422.67 405.78 371.99 324.75 285.21 253.65 228.81 213.81 Normalized sensitivity/ml pg- Matrix effect SmatriJSwater Na 0 3.62 0.15 0.52 0.03 0.05 0.57 1.64 0.02 0.01 0.01 100 3.46 0.24 0.92 0.03 0.05 0.52 3.41 0.02 0.02 0.01 200 3.47 0.32 1.04 0.03 0.05 0.49 3.30 0.02 0.02 0.02 500 15.66 0.54 2.89 0.01 0.08 0.48 2.43 0.03 0.04 0.02 K 500 18.30 0.61 3.98 0.01 0.09 0.51 2.34 0.03 0.04 0.03 Na 100 0.9 1.5 1.7 0.9 0.9 0.9 2.0 1.1 1.1 0.9 200 0.9 2.0 2.0 0.9 1 .o 0.8 2.0 0.9 1.2 1.3 500 4.3 3.5 5.5 0.5 1.6 0.8 1.5 1.4 2.3 1.6 K 500 5.0 3.8 7.5 0.5 1.7 0.9 1.4 1.4 2.5 1.9 Table7 Relative standard deviation of the background and limit of detection as functions of NaCl and KC1 matrix concentration in the range 0-500 pgml-' expressed as Na or K content.Plasma power 185 W; observation height 18 mm Element Li Cr Ca Pb Fe c u Mg Hg Cd Zn Wavelength/ nm 670.78 425.43 422.67 405.78 371.99 324.75 285.21 253.65 228.81 213.81 RSDB (Yo) LOD/ng ml-' K 0 200 500 500 0 200 500 500 0.8 0.7 0.6 0.8 15 6 2 1 0.8 0.8 0.8 0.8 161 82 61 39 0.8 0.7 0.6 0.8 45 19 6 5 0.6 0.7 0.8 1.0 540 685 1435 1943 1.0 0.8 1.0 1.0 548 447 354 329 1.0 1.0 1.0 0.9 51 58 59 51 1.2 1.3 1.3 1.5 22 12 18 15 0.8 0.8 0.9 0.7 1070 1233 810 732 0.7 0.8 0.9 0.8 1115 1053 507 508 0.7 0.9 0.7 0.8 1325 1250 893 750 Na Na K Similar behaviour for these elements in an NaCl matrix was observed by M ~ r a y a m a ~ ~ (Ca and Cd) and Larson and F a ~ s e 1 ~ ~ (Cr) for a CMP.For Li Fe and Zn the enhancement effect is present only at higher concentrations of Na while lower matrix concentrations have a depressing effect on the signal. For Cu and Pb the matrix has a depressing effect which is rather small. However for these two elements an oscillating matrix effect in the low matrix concentration range (10-5Opg ml-' Na) should be mentioned which is similar to that observed by Sturgeon et a1.I8 in a FAPES source. The matrix effect on Hg is also oscillating but unlike for Cu and Pb it extends over a large range of matrix concentrations probably because of the high ionization potential of Hg. In a single case Mg a strong enhancing effect of up to 100 pg ml-' Na is followed by a decrease in the emitted signal.It should be pointed out that all the data given above were obtained at one observation height and in order to obtain a more complete picture of the matrix effect it would be necessary to take measurements at all plasma heights. The low RSDB of the r.f. CCP studied in this work does not change significantly in the presence of the matrix. The RSDB is about 1 % for all the elements considered at all matrix concentrations. Also the background signal XB is constant at all matrix concentrations for all ten elements. Consequently the LOD is based almost entirely on the matrix effect only. Matrix Effect in the Absence of the Ring Electrode Removing the ring electrode produces a plasma which is more sensitive to the presence of easily ionized elements.The matrix effects in this case are greater and more dependent on the Table 8 Matrix effect of Na on the C3r I 425.43 nm line at observation heights of 10 and 18 mm; plasma power 185 W S m a t d S w a t e r Na concentration/pg ml Observation Ring height/mm 0 100 200 300 400 500 Present 10 1.00 1.35 1.88 1.95 2.16 1.96 Present 18 1.00 1.42 1.85 2.00 2.13 2.72 Absent 10 1.00 1.75 2.11 2.91 3.58 0.92 Absent 18 1.00 1.28 1.30 0.74 0.42 0.15 observation height. An example is given in Table 8 for the influence of Na on the Cr 1425.43 nm line at two observation heights 10 and 18 mm. In the presence of the ring electrode there is an enhancing matrix effect which constantly increases with matrix concen- tration and has values that are quite close for both observation heights.The picture is totally different in the absence of the ring. At an observation height of 10mm the matrix effect increases up to 3.58 for 400pgml-' Na and then abruptly decreases at 500pgml-' Na. At an observation height of 18 mm the matrix effect is first enhancing and then depressing down to 0.15 for 500 pg ml-' Na. It is the change in plasma geometry which generates such effects. Therefore a comparison between the two types of CCP requires a complete set of data at different powers and observation points. Such a study is underway. Conclusions It has been shown that an r.f. CCP operated at medium power and atmospheric pressure with a tip-ring electrode geometry torch could be a valuable spectral source for atomic emission.The LODs are in the ngml-' to pgml-' range and compare favourably with other data obtained for r,f. or microwave plasma torches of medium power. The results obtained for real samples are in agreement with those obtained by a non- spectral method. The source is attractive in view of its lower power requirements reduced gas consumption and good stability. The matrix effect of Na and K on elements introduced into the CCP is rather complex and depends on plasma power matrix concentration and the nature of the element being determined. At the low and medium powers used for the present work Na and K could influence the shape of the r.f. discharge thus modifying the signal emitted from a particular region of the plasma and limiting the working range of low power plasmas.For many of the elements studied NaCl andJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 64 1 KCl have an enhancing matrix effect with very high values for easily ionizable elements in low power plasmas and gener- ally higher for K than for Na. The depressing matrix effect when present is usually low. At medium power 185 and 275 W the matrix effect is lower and is less dependent on matrix concentration thereby making low power plasmas more appropriate for analytical determi- nations in spite of a somewhat lower sensitivity. With a few exceptions the LODs are better in the presence of an Na or K matrix. An explanation of the matrix effect phenomena cannot be made based on the present results without data on plasma temperature and electron number density.However the changes in plasma shape the seeding effect and the great increase of emission signal for easily ionizable elements at low plasma power indicate that the primary cause of the matrix effect is the modification of the plasma gas temperature and its distribution with varying matrix concentration. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 References Cristescu G. D. and Giurgea M. Opt. Spectrosk. 1961 11 424. Mavrodineanu R. and Hughes R. C. Spectrochim. Acta 1963 19 1309. Patel B. M. Heithmar E. and Winefordner J. D. Anal. Chem. 1987 59 2374. Hanamura S. Smith B. W. and Winefordner J. D. Can. J. Spectrosc. 1984 29 13. Zhang Y. K. Hanamura S. and Winefordner J. D. Appl. Spectrosc. 1985 39 226. Vermaak H. Kujirai O. Hanamura S. and Winefordner J. D. Can.J. Spectrosc. 1986 31 95. Patel B. M. Deavor J. P. and Winefordner J. D. Talanta 1988 35 641. Jin Q. Zhu C. Borer M. V. and Hieftje G. M. Spectrochim. Acta Part B 1991 46 417. Liang D. C. and Blades M. W. Anal. Chem. 1988 60 27. Liang D. C. and Blades M. W. Spectrochim. Acta Part B 1989 44 1059. Huang D. Liang D. C. and Blades M. W. J. Anal. At. Spectrom. 1989 4 789. Smith D. L. Liang D. C. Steel D. and Blades M. W. Spectrochim. Acta Part B 1990 45 493. Huang D. and Blades M. W. J. Anal. At. Spectrom. 1991,6,215. Huang D. and Blades M. W. Appl. Spectrosc. 1991 45 1468. 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 Sturgeon R. E. Willie S. N. Luong V. and Berman S . S. J. Anal. At. Spectrom. 1989 4 669. Sturgeon R. E. Willie S. N. Luong V.and Berman S . S. Anal. Chem. 1990 62 2370. Sturgeon R. E. Willie S. N. Luong V. and Berman S. S. J. Anal. At. Spectrom. 1990 5 635. Sturgeon R. E. Willie S. N. Luong V. and Berman S . S. J. Anal. At. Spectrom. 1991 6 19. Sturgeon R. E. Willie S. N. and Luong V. Spectrochim. Acta Part B 1991 46 1021. Sturgeon R. E. Willie S. N. Luong V. and Dunn J. G. Appl. Spectrosc. 1991 45 1413. Gross R. Platzer B. Leitner E. Schalk A. Sinabell H. Zach H. and Knapp G. Spectrochim. Acta Part B 1992 47 95. Tataru E. Anghel S. D. and Popescu A. Rev. Roum. Phys. 1991 36 29. Cordos E. A. Anghel S. D. and Fodor A. paper presented at the XXVII Colloquium Spectroscopicum Internationale Norway June 9-14 1991. Boumans P. W. J. M. de Boer F. J. Dahmen F. J. Hoelzel H. and Meier A. Spectrochim. Acta Part B 1975 30 449. Hwang J. D. Masamba W. Smith B. W. and Winefordner J. D. Can. J. Spectrosc. 1988 33 156. Atsuya I. and Akatsuka K. Anal. Chim. Acta 1980 119 341. Atsuya I. and Akatsuka K. Spectrochim. Acta Part B 1981 36 747. Wunsch G. Czech N. and Hegenberg G. 2. Anal. Chem. 1982 310 62. Wunsch G. Hegenberg G. and Czech N. Spectrochim. Acta Part B 1983 38 1135. Disam A. Tschopel P. and Tolg G. Z . Anal. Chem. 1982 310 131. West C. D. and Hume D. N. Anal. Chem. 1964 36 412. Murayama S. Matsuno H. and Yamamoto M. Spectrochim. Acta Part B 1968 23 513. Kitagawa K. and Takeuki T. Anal. Chim. Acta 1972 60 309. Anghel S. D. Popescu A. Racz F. Tataru E. and Cordos E. A Rev. Chim. (Bucharest) 1989 40 344. Cordos E. A. Darvasi E. Muresan T. Anghel S. D. and Popescu A. Rev. Chim. (Bucharest) in the press. Boumans P. W. J. M. Spectrochim. Acta Part B 1991 46 431. Larson G. F. and Fassel V. A. Anal. Chem. 1976 48 1161. Paper 3/03889K Received July 6 1993 Accepted December 14 1993

ISSN:0267-9477    

DOI:10.1039/JA9940900635

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 643 Analytical Measurement in Electrothermal Atomic Absorption Spectrometry-How Correct is it?* Albert Kh. Gilmutdinov K. Yu. Nagulin and Yu. A. Zakharov Department of Physics The University of Kazan 18 Lenin Str. Kazan 420008 Russia The detection system for atomic absorption spectrometers based on the use of photomultiplier tubes (PMTs) is analysed from the point of view of its ability to provide accurate analytical information. It is shown that absorbance recorded by the system depends not only on the number of absorbing atoms but also on their distribution in the furnace volume. The typical non-uniformity of atomic distribution that occurs in graphite furnaces and its impact on the recorded signal are discussed. The cross-sectional distribution of the intensity of the radiation beam from a primary source was measured at different locations of the spectrometer for different source operating conditions.The distribution is rather non-uniform and can be described by the Gaussian function. An analysis of the joint effect of the radiation and analyte non-uniformity on the absorbance measured is given. The shape of the radiation beam cross-section changes from a circle to an ellipse with increasing lamp current. A new detection system based on the use of a solid-state detector (photodiode array charge coupled device charge injection device) instead of PMTs is proposed. The solid-state detector is located vertically along the monochromator exit slit and allows the detection of spatially resolved absorbances.It is shown that the analytical signal recorded by this new system is proportional to the number of absorbing atoms irrespective of the non-uniformities described above. Keywords Nectrothermal atomic absorption spectrometry; cross-sectional distribution of radiant intensity and analyte; spatially resolved detection Atomic absorbance A used as the analytical signal in atomic absorption spectrometry (AAS) is defined as the logarithm of the ratio of the reference radiant flux (Do to the radiant flux @ that is transmitted through the absorbing layer of the analyte atoms.‘ Detection systems of modern atomic absorp- tion spectrometers are based on the use of photomultiplier tubes (PMTs). The most general expression for the absorbance recorded by the detection system can be expressed as cross-sectional distribution of intensity of the incident beam J ( x y ) .In the ideal situation of uniform distribution of the radiation beam analyte and atomizer temperature and when assuming that the analytical line is a single and infinitely narrow one the absorbance recorded at any instant of time is proportional to the total number of absorbing atoms in the atomizer. In this extreme case the general eqn. (1) is substan- tially simplified and takes the form:2 follows:2 [ J O J - b / 2 J-A* where b is the width of the part of the illuminating beam that is recorded by the PMT R and L are the radius and the length of the furnace respectively J(x y) is the cross-sectional distri- bution of the incident radiant intensity just before the atomizer x and y are the coordinate axes directed along the horizontal and vertical furnace diameter respectively J(A) is a function describing the spectral profile of the analytical line and the limits -A* to A* represent the spectral bandwidth isolated by the monochromator (Fig.1). The absorption coefficient K(A; x y z) depends on the spatial coordinates and wavelength and can be presented as follows:2 K(A; x y z) = c(A T)n(x y z) where n(x y z ) is the local number density of the analyte atoms and c(A T ) is a coefficient depending on the spectral features of the analytical line and temperature 7’. From this general relationship it follows that the absorbance recorded by the detection system based on the use of a PMT is dependent on three groups of factors (1) the spectral characteristics of the analysis line (dependence of J and K on A); (2) spatial distribution of the analyte and temperature within the atomizer (dependence of K on x y z); and ( 3 ) the * Presented at the XXVIII Colloquium Spectroscopicum Internationale (CSI) Post-Symposium on Graphite Atomizer Techniques in Analytical Spectroscopy Durham UK July 4-7 1993.\ J-LIZ J A = (lg e)c(A T)N (2) where LIZ n(z) dz L2 N(cm-’) = is the number of absorbing atoms per unit cross-sectional area along the radiation beam; in the case of a uniform distribution this value is proportional to the total number of analyte atoms in the atomizer. This equation is the actual basis of AAS as an analytical technique. The recorded analytical signal A is directly pro- portional to the unknown number of analyte atoms in this case.An extensive theory allowing calculation of the coefficient c(L T ) for a variety of elements and experimental conditions has been developed by L‘vov.~ The real situation however can differ substantially from that presented by the simple eqn. (2). This is because of the influence of the three groups of factors mentioned above. Of these three factors the effect of spectral features of the analytical line has been investigated most extensively. The results of these numerous investigations have been summarized recently by Gilmutdinov et d2 It has been shown that broadening of the emission line hyperfine structure and collisional shift result in a decrease in the slope and curvature of the analytical curves.644 (a) t y JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 ( b ) ( C) t' - u2 0 42 = -1* A* - b12 b12 Fig. 1 (a) Lateral and (b) longitudinal sections of a graphite tube atomizer and the schematic path of the incident beam of radiation (hatched region); (c) the analysis line isolated by the monochromator within its spectral bandwidth (-A* to A*) In many cases however the effect of these factors is not very pronounced. Out of 20 elements examined in ref. 2 the analyt- ical curves of only five of them revealed a curvature coefficient higher than 2. This observation allowed those workers2 to conclude that the spectral features of the analytical lines are not responsible for substantial curvature of the analytical curves in electrothermal AAS (ETAAS) and that one of the most important factors causing the curvature is the non- uniformity of analyte distributions in the furnace cross-section.There are many publications devoted to measurements of analyte distribution within a graphite Salmon and Holcombe4 devised the spatial isolation wheel (SIW) technique which allowed them to obtain temporally- and spatially- resolved absorbances for nine distinct zones along the vertical diameter of CRA-type mini-furnaces. With the SIW technique these researchers have shown"-9 that the value of the absorbance gradient along the vertical diameter of the graphite mini-tube can vary substantially. For example silver atoms showed practically uniform distribution at any instant in time during at~mization.~ On the other hand extremely high non- uniformity was detected in the case of chromium at~mization:~ at the beginning of atomization the absorbance value near the furnace bottom was about 0.3 while at a distance of about 3mm from the bottom absorbance was equal to zero.Chromium produced an absorbance gradient along the vertical diameter of the mini-furnace of about 1 cm-'. For other elements investigated5 (Cu Fe Ni) the maximum absorbance gradient developed at the initial stages of atomization was approximately equal to 0.5 cm-l. In all of the instances investigated the workers"-' reported normal distribution of the analyte with the region of higher number density being located near the bottom of the furnace where the sample was deposited initially. Spatial non-uniformity of analyte distribution in HGA-type furnaces for both wall and platform atomization has been investigated by Gilmutdinov and co-workers'0-12 using the shadow spectral filming (SSF) technique The characteristic feature of the SSF technique is the capability to record monochrome images of the whole atomizer cross-section which allowed temporally resolved absorbance contour maps within the whole furnace cross-section to be obtained.'@12 Further development of the SSF technique has been performed by Chakrabarti et a1.,13 who replaced the film camera that was employed earlier by a charge-coupled device camera system that allowed them to obtain digital images of atomization events.The concentration gradients of analyte distributions reported by these worker^'^'^ were about the same as those reported for the CRA-type mini-furnaces."-8 It is more import- ant to note however that principally different kinds of non- uniformities of analyte distributions were reported in these Temporally- and spatially-resolved investigations of platform atomization of Ga In and T1'@'2 revealed that the cross-sectional distribution of these atoms are inverse with the region of maximal concentration lying not near the plat- form where the sample was initially deposited but at the opposite top part of the furnace wall.The region of maximum absorbance in the case of A1 atomization lies near the sides of the walls and at the end of atomization moves to the region under the platf~rm.'~*'~ Thus it can be concluded from these spatially resolved results"-13 that during atomization in a graphite furnace any kind of non-uniformity of analyte distri- bution can exist.The absorbance gradients of these non- uniformities expressed in cm-' can be as high as 1 cm-l. For these highly non-uniform distributions the values of the absorbance gradients are as important as the absolute absorbance value. Despite these direct indications of pronounced cross- sectional non-uniformity of analyte distribution within the graphite furnace (second group of factors) the effect of this factor on the measured absorbance has been considered in only two publications.6y2 Holcombe and Rayson6 have carried out a simplified analysis of the influence of the radial distri- bution of atoms on the detected absorbance and later Gilmutdinov et aL2 developed an extensive model describing the dependence of atomic absorption on all kinds of non- uniformities of analyte and temperature distributions. It was concluded that spatial non-uniformity of the absorbing layer affects the analytical curves much more strongly than the spectral features described above. However no publications could be found that were devoted to measurements of the cross-sectional distribution of intensity in the incident radiant beam and the analysis of the influence of these non-uniformities on the measured absorbance (the third group of factors).The main objective of this study was to perform measure- ments of typical non-uniformity of the analyte and the incident beam distributions and to analyse the effect of this non- uniformity on the measured absorbance.Experimental The experiments dealing with measurements of analyte non- uniformity were performed using the Saturn-3 atomic absorp- tion spectrometer with the Graphite-1 electrothermal atomizer and the SSF set-up described elsewhere." The technique allows the recording of the image of the furnace interior in the light of the analytical line. The atomic vapour which is vaporized into the furnace volume creates a shadow picture that is recorded by a video camera. Perkin-Elmer pyrolytic graphite coated graphite tubes and platforms were used in the study of the atomization of Cu. Standard GSORM-type nitrate solutions of copper (OXFI Odessa The Ukraine) were used. Test solutions were prepared by dilution of the stock solution. A 5 p1 volume of the solution was injected into the furnace with a microlitre pipette.In the graphite furnace the ashing temperature was 300°C and the atomization temperature was 2300 "C. Argon gas of the highest available quality was used as the internal purge gas and as the sheath gas. The wall and platform temperatures were measuredJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY with a photodiode sensor using the procedure described elsewhere." The optical arrangement that was used for measurement of the cross-sectional distribution of the intensity of the radiation beam is similar to that used in commercially available atomic absorption spectrometers (Fig. 2). The luminous area of the primary source (1) is projected by the spherical mirror (3) on to the middle of the graphite furnace ( 5 ) with a magnification of 0.63.This intermediate image is brought to a sharp focus on the monochromator entrance slit (9) by the concave mirror (7) with a magnification of 0.7. A high-quality Littrow monochromator with an off-axis parabolic mirror (10) and a flat diffraction grid (1 1) was used in the set-up. The parabolic mirror of 270mm focal length and a diffraction grating with 1800 lines mm-' provide sufficient spectral resolution with minimal aberrations. The flat mirrors (2) (6) and (8) are used for turning radiation i.e. changing the direction of the beam within the set-up. Radiation passed through the monochroma- tor is then projected on to the detector by lens (13) with a magnification of 1.0. The LT-2 type hollow cathode lamps (HCLs) for Ca and Cr operating in continuous mode were used.All the measurements were taken at two lamp currents at the maximum current recommended by the manufacturer and at half the maximum current. For the two lamps the currents were 20 and 10mA respectively. In the emission spectra of the lamps Ca 422.7 nm and Cr 429.0 nm lines were used. The cross-sectional distri- bution of radiant intensity was measured at the location of the two intermediate images of the primary source i.e. at the location of the furnace ( 5 ) and at the radiant detector (14). Because interaction of the incident beam with analyte atoms occurs in the furnace the cross-sectional distribution of inten- sity of the beam was measured at different places within the atomizer. To isolate the analytical lines a glass filter (4) was mounted just before the atomizer.The absence of any spectral lines from the HCLs within the bandpass of the filter other than the lines mentioned above was verified. Therefore spec- tral interferences are not expected. When performing the spatially resolved measurements of intensity at the mono- chromator exit slit the width of the slit was equal to 2 mm. A charge-coupled device linear array [(CCDLA) FPPZ-8L NPO 'Electron' St. Petersburg Russia] mounted parallel to the vertical monochromator slits was used as the radiation detector for the spatially resolved measurements. The array consists of 1024 elements with dimensions 13 x 500pm. The gap between the elements is 6 pm. The CCDLA was mounted on an XYZ translation stage equipped with a micrometer drive for horizontal positioning. To measure the spatial distri- bution of radiant intensity in the beam cross-section scanning of the light spot was carried out in the horizontal direction perpendicular to the optical axis with a step of 0.5 mm.Thus the intensity distribution was investigated with a resolution of MAY 1994 VOL. 9 645 500 pm in the horizontal direction and 20 pm in the vertical direction. To verify the spatial resolution of the spectrometer images of negative test targets were taken. The targets were adjusted to the position of the furnace centre where the intermediate image of the primary source is formed. The optical system of the spectrometer was able to resolve a 6 lines mm-l target which corresponds to a spatial resolution of about 0.1 mm.The resolution is sufficient for investigating spatial non-uniformity in 6 mm i.d. HGA-type furnaces. The array was operated in the electron storage mode with an exposure time of 100 ms. Employing the electronic storage allows optimum use of the CCDLA with minimum noise and wide dynamic range in the course of long exposure. The data were digitized with a 12-bit analogue-to-digital converter. A 2-80 microcomputer (Zilog UK) was used as a dedicated system to control data acquisition. The information was then transferred to the serial port of an AT 386 computer for further processing. Results and Discussion Cross-sectional Non-uniformity of Analyte Distribution The solid line in Fig. 3(u) shows the absorbance profile for 50 ng of Cu recorded on a conventional 'Saturn-3' spectrometer with a spectral bandwidth of 3 nm including resonance lines of Cu at both 324.8 and 327.4 nm.Such an increased value of the bandwidth was taken deliberately because the spectral resolution of the shadow filming set-uplo allows recording of the same atomization process only in the light of these two lines simultaneously. This makes these two different measure- ments conventional and shadow filming compatible with each other. The SSF imaging of the atomization process is presented in Fig. 3(b). It can be seen that vaporization of Cu starts from the bottom of the furnace where the sample was deposited (frames 70-76) and that the cross-sectional distribution of Cu atoms during the initial part of the atomization process are strongly non-uniform the gas-phase concentration of Cu atoms decreases sharply when going from the bottom to the top of the furnace.During the dissipation of the atomic cloud (frames 88-150) the cross-sectional structure of the atomic layer is practically uniform. These results are illustrated quanti- tatively in Fig. 3(c) and ( d ) presenting the absorbance distri- butions along the vertical diameter of the furnace (c) prior to and (d) after the peak absorbance. This is the region that is isolated by the monochromator slits in a conventional spec- trometer. It can be seen from Fig. 3(c) that the absorbance gradients dA/dy along the vertical diameter of the furnace run to values of about 1 cm-' at the initial stage of the atomization. These extremely high concentration gradients of Cu atoms correlate well with earlier results of McNally and Holcombe8 obtained with the CRA-type furnace and simulated Fig.2 Schematic diagram of the atomic absorption spectrometer. Focal lengths for mirrors ( 3 ) and (7) are equal to 150 mm focal length for lens (13) is 60 mm. For details of (1)-(14) see text. Other values are distances (in mm) between the elements of the spectrometer646 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 (b) 1 70 0.9 0.7 0.5 0.3 0.1 2000 Fl 0 3 6 1 5 9 0 3 6 He i g ht/m rn Height/mrn Ti mi e/s 72 74 76 78 81 85 88 150 Fig. 3 (a) Absorbance uersus time profile for 50 ng of Cu using wall atomization (the broken line represents the change in the wall temperature). Absorbance versus height contours at 0.12 s time intervals (c) prior to aind ( d ) after the peak absorbance.(b) Images of the processes recorded using the SSF technique. The figures over the frames are their numbers beginning at the on-set of atomization. The images were recorded at 24 frames sK1 later using the Monte-Carlo appr~ach.'~ Such behaviour of Cu atoms in a graphite furnace was attributed to adsorption/ desorption proces~es.~.'~ Cross-sectional Non-uniformity of the Incident Beam of Radiation Fig. 4 shows the distribution of the radiant intensity from the Cr HCL operating at the optimum current of 10 mA in (a) the cross-section just before the atomizer and (b) at the mono- chromator exit slit where the radiant detector is normally located. It can be seen that the radiation spot just before the furnace has a diameter of about 4 mm and the distribution of intensity within the spot is extremely non-uniform.The inten- sity drops sharply from a maximum value at the centre of the spot to zero at a distance of about 2 mm. A similar distribution is obtained at the detector [Fig. 4(b)]. Since the decreased -4 -4 image of the primary source is projected on to the detector the size of the radiation spot here is smaller and the value of the maximum intensity is somewhat higher than before the furnace. A similar distribution is also obtained for emission with the Ca HCL. Since the monochromator slits isolate a relatively narrow region along the vertical diameter of the furnace [Fig. l(a)] distribution of radiant intensity in the y-direction is of special interest.In order to give a quantitative description of the interaction of the radiation with the analyte vapour in the furnace it is necessary to know what this distribution is. Fig. 5(a) shows the intensity distribution along the vertical diameter of the furnace taken from Fig. 4 (solid line). The broken line in the figure is an approximation of the distribution by the Gaussian function J ( y) = J exp(- ay2) where J is the maximum intensity at y = 0 and a is a parameter related to the width of the function. It can be seen that the Gaussian Fig. 4 Distribution of radiant intensity from the Cr HCL operated at 10 mA (a) just before the atomizer and (b) at the location of the detector. Note that the centre of the tube is 0 on the Yaxis in part (a)JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 647 -3 0 3 - 3 0 3 Y/m m Fig.5 Distribution of radiant intensity from the Cr HCL along the vertical furnace diameter (solid lines) and its approximation (broken lines) by (a) Gaussian and (b) exponential functions. Note that the centre of the tube is 0 on the Yaxis in this figure function provides an excellent approximation of the actual distribution. It should be noted that the data for the X-section of the full distribution presented in Fig. 4 can also be rep- resented by a Gaussian function with a slightly different o! parameter. The same is true for all of the other measurements undertaken in this paper the cross-sectional distribution of radiant intensity at any place within the atomizer and the location of the detector for both Cr and Ca HCLs can be well described with the Gaussian function.Thus it can be concluded that the radiation beam within the atomic absorption spec- trometer is Gaussian. Fig. 5(b) represents the same measured distribution (solid line) and its approximation by a simple exponential function (broken line). The fitting in this case is not as good as in the previous case of the Gaussian function but is still reasonably good. This more simple approximation is used further in the theoretical section. Fig.6 shows the contours of the radiation beam from the Cr HCL at different cross-sections along the atomizer length just before the atomizer (a) in the middle of the tube (b) and immediately after the atomizer (c). The corresponding positions of the cross-sections are indicated in the lower part of Fig.6. All the contours are taken from the corresponding full distri- butions shown in Fig. 4 at the level of 20% of the maximum value of the intensity in the tube axis. It can be seen that the beam cross-sections are not round at any location within the furnace the vertical dimensions are somewhat greater than the horizontal ones. The reason for this is that the optical con- figuration of the spectrometer employs a number of off-axis optic elements that results in the aberrations inherent to off- axis imaging. The shape of the radiation beam within the furnace correlates well with the previous theoretical calcu- lations of the optical system performed by Nagulin et a1.” Theory It was shown under Experimental that both the cross-sectional distribution of the analyte and the radiant intensity are non- uniform.In this theoretical section the effect of these factors on the measured absorbance is analysed. In order to make the analysis more clear the following simplifying assumptions are made. (a) The emission line is considered to be a singlet of infinitely narrow width. With this approximation the profile of an emission line J ( 2 ) is presented by Dirac’s delta-function J ( 2 ) = 6(2). This allows the effect of spatial non-uniformity to be analysed without having to consider the additional effects of spectral features (broadening hyperfine structure etc.). A detailed consideration of the influence of these spectral features on the recorded absorbance has been made previously.2 (b) The path of the incident beam of radiation is parallel to the axis of the furnace.The beam has a round cross-section and strikes the atomizer in its centre. This means that the centre of the beam coincides with the tube axis and has the coordinates x = 0 and y = R [Fig. 7(a)]. In general the position of the centre of the incident radiation may not coincide with the tube axis and the beam may not be cylindrical. (c) The changes of both radiant intensity and analyte distri- bution in the x direction within the width (-b/2 to b/2) isolated by the monochromator slits [Fig. 7(a)] are negligible. Thus both the distribution of the analyte and the intensity are dependent only on the vertical coordinate y. There is no major difficulty in solving the problem without making the above assumptions but the final results would be rather awkward.On the other hand the assumptions allow very clear results to be obtained that are reasonably close to the real situation. With these assumptions the general expression (1) for absorbance is significantly simplified and takes the following form J ( Y ) dY A=lg (3) jo2R J(Y) exp { - c(L T”)) dY As is to be expected in the case of uniform distribution of the analyte and the radiant intensity that is if J(y) = constant and N ( y) = N = constant the equation is simplified further and reduces to eqn. (2). In order to investigate the effects of the above non-uniformity on the measured absorbance it is necessary to specify the particular forms of the functions J ( y ) and N ( y) in eqn. (3). While the actual distribution of the incident radiant intensity along the vertical diameter of the furnace is Gaussian [Fig.5(a)] the more simple approximation of the distribution by an exponential function is also valid [Fig. 5(b)]. For further (a) ( b ) 3 E % E 0 -3 0 3 - 3 0 3 -3 0 3 X/mm Fig. 6 Contours of the radiation beam from the Cr HCL operated at 10 mA (a) just before the atomizer (b) in the middle of the furnace and (c) just after the atomizer. The lower part of the figure shows schematically the places of measurement648 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL. 9 (a) Y X I I 0 R 2R Bottom To P Y - Fig.7 (a) Cross-sections of the furnace and the incident radiation beam; (b) analyte distribution along the furnace diameter (solid line) and its series expansion in the vicinity of radiation beam (broken line). The hatched circle is a qualitative representation of the cross-section of the incident beam consideration the distribution is approximated by the following equation (4) where J is the maximum intensity at the incident beam axis r is the characteristic radius of the beam.Parameter r represents the distance from the beam axis y = R where the intensity falls e = 2.718 times. The lower the value of r the more non- uniform is the distribution of the radiant intensity. On the other hand when r -+ coy J ( y) -+ J = constant which corre- sponds to a uniform distribution of the intensity. With intensity distribution as given in eqn. (4) the value for the total reference flux of radiation Q0 = b jiR J ( t ) dy recorded by the PMT is presented as follows (5) If r<<R which corresponds to an infinitely narrow probing beam this relation takes the more clear form (Do = Jmb2r.The particular form of the analyte distribution N(y) along the vertical diameter of the furnace may vary substantially depending on the element under investigation and the atomiz- ation condition^.^ l3 However it is possible to perform a Taylor series expansion of the function N(y) in the vicinity of the incident beam axis. Truncating after the first term of the expansion we obtain N ( y ) = N(R) + N[y=R(y - R (6) This equation means that in the vicinity of the incident beam axis any kind of analyte distribution can be approxi- mated by the linear function (6) having a slope equal to the local gradient of the analyte distribution N [ y ’ R = ~ ‘;iy) l y = R [Fig.7(b)]. Substitution of expressions (5) and (6) into eqn. (3) gives the equation A = (lg e)cN(R) which relates directly the value of measured absorbance to the non-uniformity of the incident radiation r and the analyte distribution N’. The relationship has two terms and can be presented as follows A = A(true)(N) + AA(N’r) (8) The first term can be called ‘true’ absorbance A(true) because it depends only on the number of absorbing atoms N(R) along the incident beam and is not influenced by any kind of non- uniformity. This is a part of the measured absorbance that provides useful information. The second term AA(N’ r) rep- resents the effect of the non-uniformity of analyte distribution and the radiant intensity.It is interesting to analyse this term in the following particular cases. (1) Uniform distribution of the analyte ”0. It follows from eqn. (7) in this case that the additional term AA is equal to zero for any value of r. That is for uniform analyte distribution measured absorbance depends only on the number of absorbing atoms irrespective of the gradients of the incident radiation. (ii) The probing radiation beam is infinitely narrow r+O. Again in this particular case the additional term AA equals zero. This result has an obvious physical meaning when decreasing the lateral size of the probing beam the recorded absorbance provides information about the number of local absorbing atoms within the beam. This could create an impression that the use of a narrow probing beam of radiation is a practical way of avoiding the undesirable influence of the above non-uniformity.However the major objective of AAS is to provide information about the total number of analyte atoms within an atomizer. This means that in analytical measurements the lateral size of the probing beam must be as close to the furnace diameter as possible. Also the use of a very narrow probing beam will result in an increase of the noise in the system and a reduction of the sensitivity of the measurement. Fig. 8 shows the general dependence of the term AA on the value of cross-sectional gradients of analyte distributions for different values of parameter r. While the term AA in eqn. (8) depends on the analyte concentration gradient in the gas phase CN’ it is more convenient to express this value in terms of absorbance gradient along the vertical furnace diameter A’ = dA/dy using the relation A‘ = (lge)c(A 7’)” (9) This relation follows from eqn.(2) by taking the derivative of both sides of the equation. Therefore Fig. 8 presents the dependence of AA on A’. It can be seen from the figure that when increasing the diameter of the beam the input of the undesirable part AA also increases. Typically the characteristic radius of the radiation beam is about 1.5 mm (see Fig. 5 ) . As has been shown under Experimental the absorbance gradient A’ = dA/dy along the vertical furnace diameter can reach values up to 1 cm-l. As follows from Fig. 8 the correspondingJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MAY 1994 VOL.9 649 1 -0.16 Uniform \'I -0.24 1 I I \ I 0 0.5 1 .o A Fig.8 Dependence of the term AA on absorbance gradients along the diameter of the furnace for different radii ( r ) of the incident beam A r = 0.1; and B r = 0.2 cm value of the additional term AA can be as high as 0.1. This means that if the analyte atoms distributed uniformly in the furnace cross-section produce an absorbance of for example 0.5 the same number of atoms will produce an absorbance of between 0.4 and 0.5 depending on the extent of their cross- sectional non-uniformity. In ETAAS the absorbance signal is transient. Although in the above consideration this fact was not taken directly into account the theory developed here is also valid for non- stationary detection. To obtain the transient absorbance A( t ) it is only necessary to take the current value of the number of absorbing atoms N ( t) and their gradients N'( t).It follows from the above analysis that the actual distri- butions of analyte atoms and radiant intensity within graphite furnaces can be highly non-uniform. When using a detection system that is based on PMTs the measured absorbance is strongly influenced by this non-uniformity. To understand why it happens and how this drawback of the detection system can be overcome the following analysis is performed. Once again it is assumed that the incident beam of radiation is parallel to the atomizer wall. Suppose the beam is divided into n narrow parallel parts of unit area (Fig. 9) the total flux of the incident radiation <Do can then be presented as follows Similarly the radiation flux 0 that passes through the absorb- ing layer is presented as follows n @ = X I i i = l The absorbance recorded by the PMT A(PMT) is then presented as Thus the detection system based on the use of PMTs detects the logarithm of the ratio of the sum of intensities.It has been shown above that the signal detected in such a way depends on all kinds of non-uniformities and is not proportional to the total number of atoms N within the radiation beam. On the other hand this value N can be presented as n Nt= 1 Ni i = l where Ni is the number of analyte atoms within the ith beam of radiation (Fig. 9). It follows from the above consideration that for a very narrow probing beam the measured absorbance is proportional to the number of local atoms.Thus Ni K A i . In turn the local absorbance A i is presented as follows Iio Ai=lg- Ii Therefore the 'true' absorbance depending only on the total number of absorbing atoms can be represented as n I " i = l i=l Ii i = l A(true)= 1 A i = 1 lg-cc N i = N t (14) Thus the true absorbance A(true) which is independent of all kinds of cross-sectional non-uniformities is defined as the sum of the logarithm of the ratio of intensities. Thus the detection system of an atomic absorption spectrometer should provide the spatially resolved detection of the analytical signal. Obviously the detection system based on the use of PMTs cannot provide such spatially resolved data because the PMT can record only total spatially integrated radiant fluxes.For spatially resolved detection of the absorbances a solid-state detector (SSD) can be used. Nowadays SSDs (charge coupled devices photodiode arrays charge injected devices) are widely used in emission spe~trometryl~~'~ and in continuum source AAS.'8*19 In all of the above applications the SSD is placed along the dispersion of the spectral device (polychromator) that is in a horizontal position. In the case of the conventional atomic absorption spectrometer a one-dimensional SSD should be located along the vertical slit of the monochromator (Fig. 9) that is in a turned over position.20 Providing that the incident beam of radiation is parallel to the furnace walls each pixel of the SSD will detect the local intensities I and I i . Local absorbance Ai can be calculated using eqn.(13). Then summing up all the local absorbances using eqn.(l4) the spatially integrated absorbance A( SSD) is obtained. The absorbance calculated in such a way is independent of all the cross-sectional non-uniformities A(SSD) = A(true) Thus the analytical signal A( SSD) obtained on the basis of spatially resolved absorbances provides more accurate analytical information than the signal A( PMT) obtained using the PMT based detection system. Conclusions Over the last two decades one of the major concerns in ETAAS has been to provide sufficient temporal resolution in detection of transient absorbances. This well-recognized demand is now one of the components of the stabilized temperature plat- form furnace (STPF) system developed by Slavin and co- SSD ith pixel a / Fig.9 Schematic presentation of a spatially resolved recording of absorbance (see text for details)650 JOURNAL OF ANALYTICAL ATOMIC SPECTROME.TRY MAY 1994 VOL. 9 workers.21.22 The well-resolved transient absorbance integrated over the duration of atomization (peak area) is considered to be a measure of the unknown quantity of analyte in a sample.21,22 The results presented in this work show that the cross- sectional distribution of both radiant intensity and the analyte can be highly non-uniform within the furnace. This results in the fact that the absorbance recorded by the PMT based detection system A(PMT) is dependent not only on the number of free analyte atoms N but also on their gradients N' and the gradient of the radiation beam r i.e.A( PMT) = f ( N ; N' r) To obtain an analytical signal that would be independent of the above non-uniformities spatially resolved detection is necessary. To achieve this purpose a detection system based on a solid-state detector (photodiode array charge coupled device charge injected device) should be used. The absorbance A(SSD) recorded by such a system depends only on the number of absorbing atoms i.e. A ( SSD) = f ( N ) . Therefore to obtain more accurate analytical information the STPF concept21*22 could be supplemented with the follow- ing two features (i) the absorbance detection should be spat- ially resolved and ( i i ) the spatially resolved absorbances should then be integrated. Thus absorbances that are well-resolved not only temporally but also spatially should be used in AAS.Absorbance that is integrated not only over time but also over the monochromator slit should be used as a measure of analyte concentration in the sample. 1 2 References Welz B. Atomic Absorption Spectrometry VCH Publisher Deerfield Beach 2nd edn. 1985. Gilmutdinov A. Kh. Abdullina T. M. Gorbachev S. F. and Makarov V. L. Spectrochim. Acta Part B 1992 47 1075. 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 L'vov B. V. Spectrochim. Acta Part B 1990 45 633. Salmon S. G. and Holcombe J. A. Anal. Chem. 1979 51 648. Holcombe J. A. Rayson G. D. and Akerlind N. Spectrochim. Acta Part B 1982 37 319. Holcombe J. A. and Rayson G. D. Prog. Anal. At. Spectrosc. 1983 6 225. Droessler M. S. and Holcombe J. A. J. Anal. At. Spectrom. 1987 2 785. McNally J. and Holcombe J. A. Anal. Chem. 1987 59 1105. McNally J. and Holcombe J. A. Anal. Chem. 1991 63 1918. Gilmutdinov A. Kh. Zakharov Yu. A. Ivanov V. P. and Voloshin A. V.. J. Anal. At. Spectrom. 1991 6 505. Gilmutdinov A. Kh. Zakharov Yu. A. Ivanov V. P. Voloshin A. V. and Dittrich K. J. Anal. At. Spectrom. 1992 7 675. Gilmutdinov A. Kh. Zakharov Yu. A. and Voloshin A. V. J. Anal. At. Spectrom. 1993 8 387. Chakrabarti C. L. Gilmutdinov A. Kh. and Hutton J. C. Anal. Chem. 1993 65 716. Black S. S. Riddle M. R. and Holcombe J. A. Appl. Spectrosc. 1986 40 925. Nagulin Yu. S. Pavlicheva N. K. and Uskova A. I. Opt. Mekh. Promst. 1988 6 7. Pomeroy R. S. Sweedler J. V. and Denton M. B. Talanta 1990 37 15. Brushwyler K. R. Furuta N. and Hiftje G. M. Talanta 1990 37 23. Moulton G. P. O'Haver T. C. and Harnly J. M. J. Anal. At. Spectrom. 1989 4 673. Jones B. T. Smith B. W. and Winefordner J. D. Anal. Chem. 1989 61 1670. Gilmutdinov A. Kh. and Nagulin K. Yu. patent pending. Slavin W. Manning D. C. and Carnrick G. R. At. Spectrosc. 1981 2 137. Slavin W. Carnrick G. R. Manning D. C. and Pruszkowska E. At. Spectrosc. 1983 4 69. Paper 3106623 A Received November 3 1993 Accepted January 13 1994

ISSN:0267-9477    

DOI:10.1039/JA9940900643

出版商:RSC    

年代:1994

数据来源: RSC