摘要:

Analyst, February, 1983, VoL. 108, $9. 171-177 171 Trace-metal Determinations in Concentrated Electrolyte Solutions-a Comparative Study Peter R. Skidmore and Susan S. Greetham British Ever Ready Company Ltd., Group Technical Centre, St. Ann’s Road, Tottenham, London, N15 3TJ The results presented here are typical of those obtained during an extended study of methods for determining trace amounts of cationic contaminants in the two most commonly used dry-cell electrolytes (aqueous 7 M solutions of zinc chloride and potassium hydroxide). The results show that when dilution effects are taken into account there is little to choose between inductively coupled plasma atomic-emission and flame atomic-absorption spectroscopy as far as sensitivity is concerned but the precisions of results obtained from a direct reading (simultaneous) inductively coupled plasma spectrophotometer are significantly better than those obtained by flame atomic absorption, and the time taken to analyse samples for a number of elements is greatly reduced by the simultaneous inductively coupled plasma method. A t lower levels, significant reductions in sensitivity of the electrothermal atomic-absorption technique, amounting to 3-4 orders of magnitude in some instances, are caused by the presence of large excesses of these electrolytes, but where ultimate sensitivity is of less importance than sample throughput this technique has advantages over electroanalytical techniques.Where sensitivity, high precision and/or low capital costs are the prime consideration then electroanalytical techniques have much to offer.Keywords ; Direct current plasma ; inductively coupled plasma ; atomic- absorption spectrometry ; trace elements ; electrolyte solutions Because no one analytical instrument or technique can be universally applicable it is frequently necessary for the analyst to conduct both technical and financial evaluations of a number of techniques so as to determine the optimum combination of techniques for a particular purpose. This paper presents some typical results taken from the large number obtained during an extended study of methods for the determination of trace cations in common “dry-cell” electrolytes primarily using atomic-spectroscopic techniques but with a brief survey of some electroanalytical techniques for a few of the cations studied.The techniques of particular interest were emission spectroscopy using direct current (d.c.) and radiofrequency plasmas and absorption spectroscopy using flame and electrothermal atom sources. Experimental Each technique was examined as a separate entity using solutions prepared in appropriate matrices. Generally, it was considered that the preparation and examination of calibration graphs obtained from solutions made from the purest readily available materials would provide enough information for our purposes. Materials and Apparatus Zinc chloride solutions. Prepared by dissolving 99.999% (59Zn) zinc metal (Koch-Light Ltd.) in the minimum practicable amount of UltraR hydrochloric acid (Hopkin and Williams Ltd.) and reducing both the volume and free acid content of the solutions under reduced pressure in a rotary evaporator.Prepared by dissolving Selectipur potassium hydroxide pellets (E. Merck) in de-mineralised water in PTFE beakers and stored in polyethylene bottles. Following the recommendations of Atkinson and Hoylel standards made in alkaline solutions were stabilised by the addition of 25 mg of EDTA per 100 ml of solution. I t was The solutions were then stored in Pyrex flasks. Potassium hydroxide solutions.172 SKIDMORE AND GREETHAM : TRACE-METAL DETERMINATIONS Analyst, VoZ. 108 decided at the beginning of this work that where solutions would be diluted prior to aspiration through a nebuliser, common laboratory practice would be to neutralise the potassium hydroxide with a mineral acid at the time of dilution. This is particularly important in the inductively coupled plasma (ICP) technique where nebuliser, spray chamber and torch are all prone to attack by solutions of high pH.In such instances, to minimise the possibility of contamination the matrix solutions were made by dissolving UltraR potassium chloride in de-mineralised water. The ordinary commercial 1 mg ml-1 standard solutions of metal ions supplied by BDH and Fisons were diluted to provide the required concentrations of analyte; the solutions supplied by Hopkin and Williams were not used because they are generally made up in perchloric acid. During our initial evaluation of the ICP technique we found that the presence of perchlorates caused signal suppression for many elements and conversations with ICP users in the metallurgical and petrochemical industries confirmed that this effect is well known, although as far as we are aware, not previously reported.New glassware and plastic containers were used throughout and each item was leached with approximately 5 M nitric acid solution prepared from UltraR nitric acid and de- mineralised water for at least 7 d prior to use. The graphs shown in Figs. 1 and 2 were obtained by feeding data into a Tektronix 4054 microcomputer, fitting a first- or second-order polynomial and plotting the results on a Tektronix 4662 plotter. In some instances higher order polynomials gave better correlation coefficients but these were ignored because it is reasonable to assume that, except for third- order polynomials needed to fit the sigmoid curves produced by readily ionised species (e.g., potassium and barium), most spectroscopy graphs will be first order over a short range and second order over extended ranges.Direct Current Plasma This work was carried out using a Spectrospan I11 multi-channel d.c. plasma spectro- photometer . Preliminary trials with solutions of potassium hydroxide produced enormous distortion of the plasma, which we believe is caused by a vast increase in the number of ions producing what appears to be a very large, misshapen potassium plasma. Gilbert2 reported both signal and background enhancement when determining trace elements in sea water by this technique, effects which we believe to have similar origins as large numbers of both sodium and chloride ions will be produced from salt water matrices.The effect was not eliminated by sensible dilution so this part of the work was discontinued. Trials with 7 M solutions of zinc chloride showed severe memory effects and it was found necessary to dilute the solutions to 0.25 M (3.4% m/V) to eliminate these effects. A series of 16 solutions equivalent to 0-25 p.p.m. in 7 M zinc chloride solution were made up in 0.25 M zinc chloride + nitric acid, the instrument was calibrated using the 0 and 25 p.p.m. solutions and each solution was aspirated in turn in ascending order of concentration. The results of one run, with re-calibration immediately before the run, are shown in Fig. 1. Radiofrequency (Inductively Coupled) Plasma This work was carried out using a Hilger Analytical E964 multi-channel direct-reading spectrometer with a Plasmatherm 2500 5 kW radiofrequency (RF) generator, Meinhard nebuliser and Fassel-type torch.Because the geometry of the system precludes the possibility of sample ions becoming part of the plasma no difficulties were experienced with either matrix. Preliminary trials with solutions containing 5, 2.5 and 1% total dissolved solids showed that whereas the nebuliser will cope easily with 5% solids, severe fouling of the injector tip occurs at this concentration, some fouling occurs with 2.5% solutions but the 1% solution passed through the system without causing any fouling; subsequent work was carried out using solutions containing 1% of total dissolved solids. In each matrix 15 solutions were made up with analyte concentrations chosen to cover the range 0-104, which is lo5 times the sensitivity of the particular element.Because the Meinhard nebuliser is prone to blockage each solution was carefully examined by transmitted light before aspiration to confirm its freedom from particulate matter. The results are shown in Table I and Fig. 2.February, 1983 I N CONCENTRATED ELECTROLYTE SOLUTIONS 173 Atomic Absorption (Flame) For this part of the work we used a Varian Techtron AA5 atomic-absorption spectro- photometer with a 10-cm slot burner and Hilger hollow-cathode lamps. Air - acetylene flames were used for all elements except tin where, following the advice of L’vov et u Z . , ~ we used a fuel-rich air - hydrogen flame and obtained a five-fold increase in sensitivity and a great reduction in noise.Because the nebuliser burner system has an effective limit of about 10% dissolved solids in the aspirated solutions and because zinc chloride in aqueous solutions may hydrolyse on dilution, standards equivalent to 0-150 p.p.m. in the original 7 M solutions were made up in 0.7 M zinc chloride - 1 M hydrochloric acid solution (ten-fold dilution, 9.54% solids) and 1.4 M potassium chloride - 1 M hydrochloric acid solution (five-fold dilution, 10.44y0 solids). Aqueous standards were made up in 1 M hydrochloric acid. Each set of standards was aspirated in ascending then descending order with a “mid- point” (absorbance about 0.4) solution aspirated at regular intervals to check for drift and the signals averaged over 30 s.The results are shown in Table 11. Atomic Absorption (Electrothermal) The Varian Techtron AA5 used for the flame atomic-absorption determinations was fitted with a Varian Techtron CRA63 carbon-tube furnace of the “mini-Massmann” pattern, the output from the AA5 recorder socket was fed into a Time Electronics TS lOON high-speed integrator, which was triggered by wiring the remote control start of the integrator in parallel with the atomise step indicator light of the CRA63 control box. The particular advantages of this configuration are: the integration time does not include the dry and ash stages and therefore background noise is reduced ; integration always starts at precisely the same time in each cycle; zero offset can be applied to the integrator to compensate for base line drift over long periods of time; and because the output from the AA5 is effectively linear up to 35 mV, the 30 mV input setting on the integrator can be used to enable measurements to be made on samples that are “off-scale” on the 10mV meter fitted as standard to the AA5.A series of nine standards covering the range 0.005-10 pgml-l were made up in each electrolyte and 10 measurements were made on each solution. The medians of each series of measurements were taken and are shown in Table 111. Table IV compares the limits of detection found in practice with those claimed by the manufacturers for analytes in aqueous solutions. McArthur* recommends the use of 30% m/V aqueous ammonium nitrate solution to reduce the effects of halide volatilisation.We have found this to be efficacious, but it is more convenient to abort the determination between the dry and ash stage, add 5 p1 of saturated aqueous ammonium nitrate solution and recommence the determination cycle. This pro- cedure does, however, extend the time taken for one determination and in many laboratories this may be unacceptable. Polarography and Voltammetry A rapid evaluation of polarographic (pulse and differential pulse) techniques for the “easy” elements, antimony, cadmium, copper, lead and tin, in 7 M matrix solutions using a Rletrohm E506 polarograph showed that the high electrolyte concentrations used reduced the sensi- tivity of ordinary pulse polarography to such an extent that no signal was observable at the 1 p.p.m. level. Differential-pulse polarography on the other hand gives useful results to better than 0.1 p.p.m.Anodic-stripping voltammetry using a hanging mercury drop electrode gives excellent results to at least 0.01 p.p.m. for all the elements investigated. During the deposition period the solutions were stirred for 2.5 min, rested for 30 s and stripped using a 0.1 pA mm-l sensitivity setting. The peak shapes and definitions were such that significant improve- ments could be obtained by extending the deposition time and/or increasing the sensitivity. Obviously the tin and lead peaks cannot be separated in these matrices but this problem could be overcome by changing the supporting electrolyte after the rest period. Arsenic can be determined at very low levels by using a rotating gold disc electrode.174 SKIDMORE AND GREETHAM : TRACE-METAL DETERMINATIONS Analyst, Vol.108 30 25 6 20 z 2 10 CI 15 $! (1 Fig. 1. D.c. plasma - Cchelle monochromator drift. Results Direct Current Plasma This instrument has been modified by the manufacturers in an attempt to improve stability; nevertheless the calibration graphs shown in Fig. 1 demonstrate that excessive drift still occurs and the only way to obtain acceptable results would be to re-calibrate at frequent intervals. TABLE I RESULTS OBTAINED BY INDUCTIVELY COUPLED PLASMA - OPTICAL Element As .. .. Cd . . .. cu . . .. Fe . . .. Mg . . .. Ni .. .. Pb . . . . Sb .. .. Sn . . .. Zn . . .. EMISSION SPECTROMETRY Matrix Range, p.p.m. Slope 0-500 538 0-200 312 0-200 307 0-500 1995 0-100 1223 0-100 1161 0-500 1654 0-200 1382 0-200 1499 0-500 1450 0-200 1617 0-200 1 840 0-50 2 132 0-100 1525 0-100 1499 0-500 694 0-500 496 0-500 500 0-500 535 0-200 362 0-200 339 0-500 877 0-200 540 0-200 55 1 0-500 89 1 0-100 530 0-100 490 0-200 994 0-100 1022 2: 2: 22 2: 2; 2: 22 2: 2: ZnC1, ZnC1, ZnC1, ZnC1, ZnC1, ZnC1, ZnC1, ZnC1, ZnC1, Intercept 803 523 794 2 833 90 247 3 639 1021 1050 1637 1891 1039 699 418 4 390 1160 506 1297 1591 1075 1425 2 361 1652 7 024 561 279 680 744 165February, 1983 IN CONCENTRATED ELECTROLYTE SOLUTIONS 175 Inductively Coupled Plasma In each instance the correlation coefficient to a least-squares straight-line fit was better than 0.999.The concentration range, slope and intercept are shown in Table I, where it can be seen that generally a reduction in sensitivity occurs in both matrices for all elements except iron.This reduction is probably largely due to viscosity and surface tension effects and underlines the importance of matrix matching when comparing samples with external standards. The enhancement of the iron signal is probably brought about by the presence of chloride as the only common factor in both the zinc chloride and potassium chloride solutions : the aqueous standards generally are made from nitrate solutions and contain very little chloride. The enhancement is illustrated in Fig. 2 where the arsenic graphs are shown as typical for comparison. Further work is needed to clarify this point. 350 1 0 50 100 150 200 p.p.m. Fig. 2. ICP matrix effects. Atomic Absorption (Flame) As was expected, all the elements except cadmium showed considerable non-specific absorption in the potassium chloride matrix, none of the elements except lead showed non-specific absorp- tion in the zinc chloride matrix and none of the elements showed non-specific absorption in the dilute acid matrix.The results in Table I1 are absorbance readings corrected for both background and blank absorbances. There are marginal signal enhancement and suppression effects for some of the analytes but nothing of any real significance. Atomic Absorption (Electrothermal) The first thing that one notices about these results is the reduction in sensitivity for some elements whilst other elements appear to be virtually unaffected. Table IV shows that the sensitivity for cadmium is reduced by four, copper by three and tin by two orders of magni- tude, whereas iron, nickel and lead show negligible effects.The results are much as would be expected from this well tried technique.TABLE I1 Concentration/ mg ml-1 30 20 15 10 7.5 5.0 2.5 2.0 1.0 0.6 0.5 0.3 0.2 0.1 ATOMIC ABSORPTION (FLAME) ABSORBANCE READINGS Element (wavelength/nm) I L 5 Sn (224.6) Cd (288.8) Cu (324.8) Fe (248.3) Ni (232.0) Pb (217.0) Sb (217.5) - - - - - - -7 HCI* KCl* ZnCly* HCl* KCl* ZnC1,* HCl* KCl* ZnCI,* HCl* KCl* ZnCl,* HCl* KCl* ZnCl,* HCl* KCl* ZnC1,* HCl* KCI* ZnCl,* 0.604 0.791 0.747 0.737 0.964 0.867 0.820 0.450 0.593 0.527 0.529 0.507 0.467 0.409 0.400 0.827 0.828 0.803 0.616 0.606 0.595 0.509 0.469 0.478 0.271 0.389 0.342 0.283 0.284 0.276 0.717 0.456 0.369 0.236 0.209 0.610 0.611 0.586 0.324 0.318 0.307 0.267 0.240 0.254 0.158 0.210 0.189 0.143 0.146 0.143 0.376 0.357 0.167 0.157 0.138 0.134 0.089 0.100 0.073 0.074 0.315 0.131 0.096 0.084 0.068 0.169 0.175 0.162 0.069 0.066 0.062 0.058 0.047 0.053 0.041 0.044 0.044 0.033 0.027 0.030 0.108 0.041 0.029 0.027 0.017 0.084 0.030 0.027 0.022 0.016 0.051 0.018 0.016 0.013 0.009 0.040 0.013 0.006 0.011 0.004 0.020 0.017 0.007 0.004 0.005 0.004 0.004 0.003 0.966 0.894 0.861 0.660 0.669 0.424 0.311 0.388 0.375 0.260 0.255 0.287 0.213 0.194 0.194 0.162 0.134 0.129 0.130 0.148 0.109 0.067 0.064 0.065 0.076 0.056 0.035 0.033 0.038 0.023 0.020 0.016 0.004 0,013 0.016 0.011 0.008 0.096 0.008 0.005 * Matrices: HCI = 1 Y hydrochloric acid; KCl = 1.4 Y potassium chloride, 1 M hydrochloric acid; and ZnC1, = 0.7 M zinc chloride, 1 M hydrochloric acid, 0.196 0.122 0.098 0.064 0.033 0.011 0.007 0.004February, 1983 IN CONCENTRATED ELECTROLYTE SOLUTIONS 177 The results listed in Table I11 show acceptable linearity except for iron where there is considerable scatter; we found that this scatter improved with furnace tube use and is almost certainly due to contamination of the tubes during manufacture.Major improvements TABLE I11 ATOMIC ABSORPTION (ELECTROTHERMAL) PEAK AREAS (INTEGRATOR COUNTS) Element (wavelength/nm) Cd (228.8) Cu (324.8) Concen trationl mg ml-I ZH** i(OH+;nz* 10 834 873 2499 2496 5.0 417 423 1248 1248 2.0 171 168 498 498 1 .o 57 90 249 249 0.5 33 45 123 125 0.1 24 24 0.05 0.01 0.005 Fe (372.0) Ni (232.0) r---4- 7 r--- KOH* ZnC1,* KOH* ZnCI,* 1668 1302 3655 1281 960 1128 1833 690 759 855 723 525 588 789 366 447 474 720 183 219 414 639 60 102 207 321 30 30 45 96 24 45 Pb (217.0) r A - 1 KOH* ZnC1,* 6264 6244 2631 2596 1332 1246 651 586 225 109 93 58 30 28 Sn (224.6) ---7 KOH* ZnC1,* 1671 4086 838 2043 334 816 167 408 17 42 21 x3 204 * Matrices: KOH = 7 M potassium hydroxide and ZnC1, = 5 M zinc chloride.could be brought about by repeatedly firing the tubes for several seconds at maximum temperature prior to use. However, this procedure reduces the life of the tube to an un- acceptable extent and is not to be recommended. Conclusions Although the electroanalytical methods were only briefly considered, both differential- pulse polarography and anodic-stripping voltammetry seem worthy of further study, especially because in comparison with the spectroscopic techniques the apparatus is cheap to buy, relatively cheap to operate and does not require fixed fume-extraction systems to protect the operator. Considering labour and capital costs and the relative performances of the various systems, it is obvious that there is no one perfect technique. However, for the large laboratory the best combination of techniques would be a direct-reading inductively coupled plasma system and an electrothermal atomic-absorption spectrophotometer, whereas for the smaller labora- tory, a combination of flame and electrothermal atomic absorption would suffice. In both laboratories, the electroanalytical techniques could prove useful. TABLE IV ATOMIC ABSORPTION (ELECTROTHERMAL) LIMITS OF DETECTION FOUND FOR ELECTROLYTE SOLUTIONS AND CLAIMED BY MANUFACTURERS FOR AQUEOUS SOLUTIONS Element/ng ml-l Matrix Cd c u Fe Ni I’b Sn . . 0.03 0.08 0.6 2.0 0.8 1.0 . . 500 100 10 5 10 100 ZnC1, . . . . 500 100 5 50 10 50 HKtH : : References 1 . 2. 3. 4. Atkinson, A., and Hoyle, W. C., Appl. Spectrosc., 1979, 33, 37. Gilbert, T. K., Env. Meas. Lab. Env. 0. (US Department of Energy), 1979, EML-356, 429-62; L’vov, B. V., Katskov, D. A., Kruglikova, L. V., Orlov, N. A., and Polzik, L. K., Zh. Anal. Khzuuz., McArthur, J . M., Anal. Chim. A d a , 1977, 93, 77. Chem. Abstr., 1979, 91, 128810. 1975, 30, 1861. Received September 9th, 1982 Accepted October 26th, 1982

ISSN:0003-2654    

DOI:10.1039/AN9830800171

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

178 Analyst February 1983 Vol. 108 pp. 178-188 Simultaneous Multi-element Analysis by Carbon Furnace Atomic-emission Spectrometry John Marshall David Littlejohn and John M. Ottaway Department of Pure and Applied Chemistry University of Strathclyde 295 Cathedral Street Glasgow G1 1XL James M. Harnly Nutrient Composition Laboratory Beltsville Human Nutrition Centre US Department of Agriculture Belts-ville M D 20705 USA Nancy J. Miller-lhli and Thomas C. O'Haver Department of Chemistry University of Maryland College Park MD 20742 USA The application of carbon furnace atomic-emission spectrometry (CFAES) to simultaneous multi-element analysis has been investigated using a direct-reading spectrometer system. A computer-controlled wavelength modulation system employing a quartz refractor plate is used to provide automatic back-ground correction in both the single and multi-element modes.Detection limits obtained using a three-step square-wave modulation waveform with this system are comparable to those previously obtained using the rotating sector method of modulation. The linear range of calibration graphs has been extended to 4-5 orders of magnitude by measurement of emission intensities off the centre of the line profile. The potential of CFAES as a technique for simultaneous multi-element analysis is demonstrated by the determination of trace elements in NBS standard reference materials and orange and pine-apple j uice samples. Keywords A tomic-emission spectrometry ; electrothermal atomisation ; simul-taneous multi-element analysis ; wavelength modulation ; f r u i t juice The renewal of interest in optical emission spectrometry brought about primarily by the development of plasma emission sources has led to an increasing demand for systems with a capability for simultaneous multi-element analysis.As a result a fresh impetus has been given to the development of multi-element spectrometer systems not only for atomic-emission but also for atomic-fluorescence and atomic-absorption measurements. Ideally such a spectrometer system should be readily adaptable to flame plasma or furnace atomic spectro-metric measurements without sacrificing the advantages of any particular technique. The complementary nature of these atom cells should be emphasised as it is unlikely that any one technique will be capable of providing a solution to all of the analytical problems encountered.Although plasma emission sources have enjoyed considerable popularity as a result of the appearance of suitable commercially available multi-channel detection systems, the development of similar instrumentation based on flame and furnace atomisation is still restricted to research laboratories. A number of have described flame-based systems for simultaneous multi-element analysis and the subject has been reviewed by Busch and Morri~on.~ Fewer publications are found in the literature concerning the use of electro-thermal atomisers for simultaneous multi-element analysis because such atomisers are still primarly associated with the technique of line source atomic-absorption spectrometry.As a consequence much research has been devoted to the adaptation of instrumentation to accommodate multi-channel line source atomic-absorption measurement The applica-tion and extension of such systems is limited by the dependence on multi-element hollow-cathode lamps (and hence the choice of elements that can be determined at the same time), deteriorating signal to noise ratios and/or design factors in the method of detection particu-larly with regard to background correction. Whilst satisfactory for specific applications it would appear that the line source atomic-absorption technique does not provide the most convenient solution to the general problem of simultaneous multi-element determination based on electrothermal atomisation MARSHALL et al. 179 A number of pub1icationsl4-l6 have outlined the development of a single-channel continuum-source atomic-absorption system which employs an Eimac xenon arc lamp as the primary light source and a high-resolution 6chelle monochromator incorporating a wavelength modula-tion device to provide automatic background correction.More recently,17 a background-corrected simultaneous multi-element atomic-absorption spectrometer (SIMAAC) was described which allowed the determination of up to 16 elements without significant signal to noise loss compared with a single-element operation. Flame or furnace atomisation could be used with this instrument and a computer-controlled wavelength modulation system was used to achieve background correction simultaneously on all channels. A single-channel version of the khelle spectrometer which is the basis of the SIMAAC system has been used in conjunction with an alternative system of wavelength modulation to improve detection limits in carbon furnace atomic-emission spectrometry (CFAES) .la The multi-channel version of the 6chelle is essentially a direct-reading polychromator and it would be expected that the SIMAAC system could be modified to allow simultaneous multi-element atomic-emission measurements from a carbon furnace.The application of the SIMAAC system to CFAES measurements is described in this paper. The utility of the computer-controlled wavelength modulation system employed in the SIMAAC instrument was investigated for CFAES in regard to sensitivity and to the extension of the linear range of calibration graphs.19 The performance of CFAES in the multi-element mode is demonstrated by the determination of trace elements in NBS standard reference materials.Experimental Instrumentation In terms of instrumental parameters very little modification of the system is required for CFAES measure-ment. The optical separation of the atomic-emission signal from the continuum tube wall background signal is regarded as one of the most important factors in the optimisation of sensitivity in CFAES.20 In this work a lens of focal length 20 cm was used to produce a 1 1 image of the carbon furnace on the monochromator entrance slit. Although it has been reported that direct tube wall emission has no appreciable effect on the continuum-source atomic-absorption (CSAA) signal,21 it is clear that efficient optical baffling of the furnace image will not be disadvantageous in the absorption mode.I t is conceivable that the same optical system could be used in both the absorption and emission modes and that changing from one technique to the other would require only that the Eimac lamp be switched on or Off. Optimum sensitivity and calibration graph linearity in continuum-source atomic absorption is achieved using high resolution and hence narrow spectrometer slit widths. Relatively wide slit widths are employed for CFAES measurements for optimum performance with the khelle spectrometer where an automatic method of background correction is available.la Thus in the single-element mode a change in slit dimensions is required prior to the use of either technique.The exit slit is located in a removable cassette and can be adjusted in a few seconds. In the multi-element mode up to 20 slits of a pre-specified height and width are fixed in the cassette face. Once inserted these slits cannot conveniently be altered. The multi-element cassette used in the SIMAAC system is fitted with 25-pni slits which are necessary for the most advantageous use of the CSAA technique. The use of narrow slits for CFAES measurement as predetermined by the custom-built nature of the cassette will result in a decrease in light throughput and signal to noise ratio in the multi-element mode. Additionally the individual spectral lines selected for the multi-element cassette were chosen to give optimum CSAA sensitivity.In a number of instances these wavelengths were different from the optimum CFAES wavelengths. Thus of the 16 elements that can be deter-mined simultaneously using the SIMAAC system,17 only nine of these were feasible in terms of useful analytical sensitivity by CFAES determination in the multi-element mode. Clearly, for optimum CFAES performance in the multi-element mode a second cassette with appropri-ate slit dimensions and wavelength selection would be required. A Princeton Applied Research 128A lock-in amplifier was used for phase-sensitive detec-tion of the atomic-emission signal in the single-channel mode when the computer data The SIMAAC system and its operation have been described p r e v i o u ~ l y . ~ ~ ~ ~ 180 MARSHALL et a,?. SIMULTANEOUS MULTI-ELEMENT Analyst VoZ.108 acquisition system was not employed. In this instance a modulation detection frequency of 40 Hz was employed to provide a direct comparison with previous results.ls Background-corrected atomic-emission signals were measured from the analogue output of the lock-in amplifier on a Perkin-Elmer PE56 potentiometric recorder. A modulation frequency of 56 Hz was employed when the computer was used for data acquisition. Software It is necessary in continuum-source atomic-absorption measurements using the SIMAAC system to compute an absorbance - time array for each channel from the raw intensity data. For emission measurement the background intensity must be subtracted from the total emission intensity to give the net intensity due to analyte and thus the program was modified to allow an intensity subtraction rather than an absorbance calculation.Other software modifications were minor and related to the alteration of the data presentation format to accommodate results in relative intensity units rather than in absorbance units. The output format allowed the display of the net atomic-emission signal in arbitrary intensity units up to a maximum value of 10000 which was adequate for most measurements. Five peak-height and six peak-area measurements per atomisation were provided simultaneously for each channel in the final data presentation in an analogous manner to that described previ-ously for continuum-source atomic absorption.l9 Electrothermal Atomisation The Perkin-Elmer HGA 2000 furnace was used to obtain most of the results but an HGA 500 was also used.The former furnace is similar in design to the HGA 72 furnace employed in previous studies in CITAES,la and differs only in minor respects concerning the flexibility of the programme controller. Standard unmodified graphite tubes were used either with or without platform atomisation as indicated in the text. Sample volumes of 50 p1 were introduced manually to the furnace. Drying temperatures of 100 and 300 "C for 40 s were employed for tube wall and platform atomisation respectively. A compromise ashing temperature of 900 "C for 20 s was used in the multi-element mode but otherwise book values were employed. The maximum temperature (about 2700 "C) was set in the atomisation stage for 12 s in all instances. Argon was employed as the furnace purge gas and measurements were made in the gas stop mode.A 3-s cleaning step at maximum tempera-ture under gas flow conditions was also programmed to remove any involatile residues after atomisation. Peak-height and peak-area data were produced simultaneously using the SIMAAC system. Thus integration limits obtained empirically were defined for each channel prior to measure-ment. The furnace programmer was used to provide a trigger for the initiation of data acquisition over the defined integration time. When the HGA 2000 furnace was used it was necessary to trigger the data cycle 18 s from the end of the ashing stage of the programme. As the total data acquisition time was 30 s this reduced the total integration time available in the atomisation stage to 12s using this atomiser.When the HGA 500 furnace was employed it was found possible to initiate the data acquisition cycle at the start of the atom-isation stage of the furnace programme. This atomiser was used in the multi-element mode under the following conditions drying stage 110 "C for 30 s; ashing stage 900 "C for 20 s; atomisation stage 2700 "C for 8 s ; and cleaning stage 2850 "C for 3 s. An AS-1 auto-sampling system was used to introduce 20-4 samples to the HGA 500 furnace. A much more rapid emission signal is observed using the HGA 500 than with the HGA 2000 furnace and integration limits were separately optimised. Reagents and Sample Preparation Aqueous standard solutions of the appropriate element were prepared in a dilute acid medium as required from atomic-absorption stock standard solutions.For multi-element analysis aqueous calibration standards were prepared in the range 1 ng ml-l to 100 pgml-l for minor elements and 10 ng ml-l to 1000 pg ml-l for major elements from multi-element stock solutions supplied by Spex Industries Ltd. The NBS standard reference materials and orange and pineapple juice samples were digested in triplicate in concentrated nitric acid and diluted with distilled water using the procedure described previ0us1y.l~ All reagents and gases were of the highest available purity February 1983 ANALYSIS BY CARBON FURNACE AES Results and Discussion 181 Modulation Waveform Wavelength modulation has been widely recognised as an accurate technique for the correction of spectral background in atomic spectrometry.In the SIMAAC system wave-length modulation is provided by means of a quartz refractor plate positioned behind the entrance slit of the spectrometer. The oscillation of the refractor plate can be controlled using externally generated waveforms e.g. by computer allowing a greater degree of flexi-bility in the operation of the system. I t has been predicted from theory22 and verified e~perimentally~~ that an improvement in signal to noise ratio of up to a factor of 1.8 can be achieved using a square as opposed to a sinusoidal waveform. Previous work using sine2* and square ~ a v e l * * ~ ~ modulation waveforms for CFAES has been difficult to compare owing to the use of different monochromators and furnaces. Using the SIMAAC system a direct comparison of modulation waveforms can be made under the same experimental conditions.I t is of particular interest to determine whether similar detection limits can be obtained for CFAES by three-step square wave modulation using a refractor plate system to those obtained using square wave modulation provided by the rotating quartz sector method.18~~~ CFAES detection limits obtained using the SIMAAC system in the single-element mode with the HGA 2000 atomiser are presented in Table I. The results were obtained using a lock-in amplifier to provide a direct comparison with detection limits obtained using the rotating quartz sector shown in column 6 of Table I. The detection limits obtained for these elements using square wave modulation with the refractor plate system are comparable to those obtained using the rotating quartz sector and seem to indicate that there is little practical difference between the two methods of modula-tion in terms of absolute sensitivity.TABLE I CFAES DETECTION LIMITS OBTAINED I N THE SINGLE-CHANNEL MODE USING THE SIMAAC SYSTEM 50-p1 sample volumes using standard tube or platform atomisation as indicated. Detection limitlpg 1-1 r-Modulation SIMAAClHGA 2000 fichelle - HGA 7; Element Wavelength/nm Tube waveform using lock-in amplifier system25 Chromium . . 425.43 Platform Square 0.013 0.023 Iron . . 371.99 Standard Square 0.33 0.35 Iron . . 371.99 Standard Sine 0.56 -Manganese . . 403.08 Standard Square 0.08 0.066 Copper . . 324.75 Standard Square 0.18 0.10 The detection limits for iron shown in Table I illustrate the difference in signal to noise ratio obtained using sine and square wave modulation.The difference in detection limit was found to be 1.7-fo1d1 which is very close to the theoretical factor of 1.8.22 A detection limit of 0.27 ng ml-l was obtained for copper using the SIMAAC computer data acquisition system, and this was sufficiently close to the value of 0.18 ng ml-l obtained using the lock-in ampli-fier to suggest that signal to noise ratios will be maintained using this type of detection system for CFAES. However computer control of the modulation waveform offers greater flexibility of operation because when a bigaussian waveform is employed it is possible to extend calibration graph linearity to high concentrations.lg In the multi-element mode, detection limits were impaired by non-optimum slit and wavelength conditions.However, calculations to correct for these factors indicate that approximately the same sensitivity as in the single-channel mode will be attained with optimum conditions in the multi-element mode. It has been demonstrated that signal to noise ratios are not significantly poorer in multi-channel operation for continuum-source atomic absorption,17 and this is almost certainly the situation for CFAES 182 MARSHALL et a,?. SIMULTANEOUS MULTI-ELEMENT Analjst VoZ. 108 Extension of Analytical Calibration Graphs in CFAES It has been demonstrated that the linear range of analytical calibration graphs in atomic-absorption spectrometry can be extended using the SIMAAC system.lg Intensity measure-ments are made at predetermined intervals across the absorption profile using the wavelength modulation system which is controlled by a bigaussian waveform generated by the computer.A series of absorbances can be computed from these intensity measurements for a single atomisation. Thus for a series of analytical standards a number of calibration graphs can be produced that have overlapping linear ranges. The method takes advantage of the fact that greater growth curve linearity can be obtained by measurement of intensity off the centre of the line profile. This increase in linearity is achieved at the expense of sensitivity and as absorbance values are computed further away from the line centre the signal to noise ratio gradually deteriorates.In principle this type of measurement could be used to extend the linear range of CFAES calibration graphs. This possibility was investigated for a number of elements in both single and multi-element modes. tl Emission line profile A A A A A A A A A A A A A A 1 2 3 4 . A . 17 18 1920 6 Sample points ' Wavelength modulation interval b Fig. 1. Distribution of data points for CFAES intensity measurements at different positions on the emission line profile. The principle of operation of the system in the emission mode is illustrated in Fig. 1. The emission intensity is measured at 20 pre-selected points across the wavelength modulation interval. The background intensity represented by lT1 is measured on either side of the line profile at the extremes of the modulation interval.This value is subtracted from I, the line plus background intensity to yield the net atomic-emission intensity. The combinations of data points used to calculate the net emission intensity were the same as those used for atomic absorptionlg and are given in Table 11. The intensity calculated on the basis of all TABLE I1 POSSIBLE COMBINATION OF INTENSITY POINTS FOR COMPUTING EMISSION INTENSITIES (1 - 11) Case 1 2 3 4 5a 5b 5c 5d 5e 5f :; 5i I , 1-5 + 16-20 1-4 + 17-20 1-3 + 18-20 1-2 + 19-20 1 + 20 1 + 20 1 + 20 1 + 20 1 + 20 1 + 20 1 + 20 1 + 20 1 + 20 I, 6-1 5 7-14 8-13 9-12 10 + 11 9 + 12 8 + 13 7 + 14 6 + 15 5 + 16 4 + 17 3 + 18 2 + 19 Combinations used in this work 1 -2 3 4 5 February 1983 ANALYSIS BY CARBON FURNACE AES 183 20 points (case 1) corresponds to the normal calibration graph produced by a standard wave-length modulation system and would be expected to give the best signal to noise ratio.Cases 2 4 are essentially the same but with fewer intensity points and would yield a poorer signal to noise ratio with similar calibration graph characteristics to case 1 . Iii this study, cases 1 and 5e-5i were used as this set of six calibration graphs had been found to provide 4-6 orders of magnitude of linear dynamic range in the absorption mode.lg These combina-tions are indicated in Table I1 and will be referred to herein as graphs 1-6. 4 .- z 3 C * .r 2 cn J 1 0 - 2 -1 0 1 2 3 Log (solution concentration/pg rn1-l) 1 -2 -1 0 1 2 3 Log (solution concentration/pg rn1-l) Fig.2. CFAES calibration graphs obtained for chromium using the SIMAAC system in the single-element mode and the HGA 2000 furnace. Data points used for computing emission intensities as in Table 11. Peak-area measurements a t 425.43 nm. l i g . 3. CFAES calibration graphs obtained for manganese under the sainc conditions as used for the results in Fig. 2 but a t 403.08 nm. Calibration graphs for chromium magnesium iron manganese copper and sodium were obtained using the SIMAAC system in the single-element mode and some of these are shown in Figs. 2-5. I t was found that the linearity of CFAES calibration graphs could be extended by up to two orders of magnitude by the use of edge-of-line intensity measurements.I t is clear from Fig. 2 that graphs of lower sensitivity i.e. graphs 4-6 provide this improvement in calibration graph linearity in CFAES. Relatively little improvement in linearity is obtained between graphs 1 and 3. In continuum-source atomic absorption relatively narrow 25-pm slits are used and the resolution “window” is small in comparison with the line width. Slit widths are much larger for CFAES measurement in order to provide optimum sensitivity, but in this instance the resolution is relatively poor. The window of observation is larger 6 1 - 1 0 1 2 3 I I 4 I 0 - 2 -1 0 1 2 3 Log (solution concentration/yg mi-’) Log (solution concentration/yg rnl-’) Fig. 4. CFAES calibration graphs obtained for iron a t 371.99 nm.Conditions as in Fig. 2. Fig. 5 . CFAES calibration graphs obtained for copper at 324.75 nin using peak-area measurements. Conditions as in Fig. 2 184 MARSHALL et al. SIMULTANEOUS MULTI-ELEMENT Analyst Vol. 108 in comparison with the width of the line than in the absorption mode and thus measurements at the pre-determined points are average intensities over a larger wavelength range. Thus a pseudo-edge-of-line calibration technique is used for CFAES in this mode because in graphs 2 and 3 the over-all signal to noise ratio is still dominated by the effect of the peak maximum line intensity. Further away from the line centre this effect will be less and the window will contain a greater proportion of background intensity which is essentially constant over the wavelength modulation interval concerned.Hence only graphs 4-6 where intensity measurement is made sufficiently far away from the line centre offer significant improvements in linearity. It would appear that a redistribution of data points would increase the amount of useful information that can be derived from emission-line profiles to minimise the similarity of graphs 1-3. The point spacing used was selected for CSAA but there is no reason why this could not be changed to best accommodate emission measurements. This could be achieved by reducing the number of points at the line centre and increasing the number of points in the wings of the line profile. This would serve to improve the continuity of the linear ranges between graphs.One of the interesting features of the calibration graphs produced was that an increase in sensitivity was observed at relatively low concentrations on graph 2 with respect to graph 1. The background intensity on graph 1 is measured at points 1-5 and 16-20 whereas on graph 2 only points 1 and 20 are used. Thus on graph 1 background intensity is measured much closer to the emission-line profile than on graphs 2-6. This means that analyte line intensity derived from the wings of the line profile at points 2-5 and 16-19 is included in the background intensity measurement resulting in a lower net background-corrected atomic-emission intensity. The background intensity for all other graphs is computed at the extreme ends of the modulation interval and in these instances the effect will be minimal.Thus graph 2 was found to be more sensitive for almost all elements than graph 1 even at low concentrations, and is linear to slightly higher concentrations than graph 1. In most instances however, graphs 1-3 are sufficiently close together that only one of the three is required to ensure a continuous calibration range. This appeared to be the case for both peak-height and peak-area measurements. In general peak-area measurements provided superior linearity to peak-height measurement s. -2 -1 0 1 2 3 Log (solution concentration/pg mi-’) Fig. 6. CFAES calibration graphs obtained for copper at 324.75 nm using peak-height measurements. Other conditions as in Fig. 2. Improved calibration graph linearity has been predicted for CFAES with the use of signal integration26 and the results obtained in this study confirm this view.Figs. 5 and 6 show calibration graphs obtained for copper by peak-height and peak-area measurements. The graphs are closer together for the peak-height data and appear to level off at lower con-centrations than the corresponding peak-area measurements. Additionally six graphs are provided by the computer for peak-area measurement whilst only five are provided for peak-height measurement. Consequently signal integration was adopted as the mode of measure-ment most suitable for the extension of calibration graph linearity in CFAES. Using the SIMAAC system it is possible to extend calibration graphs in both the single-and multi-element modes. This is extremely important in terms of the development o February 1983 ANALYSIS BY CARBON FURNACE AES 185 CFAES as multi-element analyses demand a wide linear dynamic range as a result of the variation in concentration between elements and from sample to sample.Hence sensitivity problems resulting from dilution in order to measure high concentrations can be avoided using this technique. Calibration graphs for chromium and potassium in the multi-element mode using the HGA 500 furnace are given in Figs. 7 and 8. Linearity up to 1000 pg ml-l is achieved which is typical of the elements investigated and demonstrates the feasibility of the calibration graph extension technique as applied to CFAES in the multi-element mode. 0 1 2 3 Log (solution concentration/yg rn1-l) Fig. 7. CFAES calibration graphs obtained for chromium using the SIMAAC system in the multi-element mode and the HGA 500 furnace.Data points used for computing emission intensities as in Table 11. Peak-area measurements a t 357.9 nm. I I -2 -1 0 1 2 Log (solution concentration/yg rn1-l) Fig. 8. CFAES calibration graphs obtained for potassium a t 404.41 nm. Other con-ditions as in Fig. 7. Limitations of the Technique It was found that care had to be taken in the selection of the amplification factor for each channel to accommodate the emission signal within the range of the analogue to digital converter (ADC). There is a threshold minimum voltage at the input to the ADC below which no signal will be measured. In CFAES the signal increases by several orders of magnitude from low to high concentration values and it is possible to saturate the ADC if the amplification factor selected is too high.In the absorption mode the difference between a high intensity and a lower intensity is measured and the range of the ADC is not a significant factor in the measurement. It is possible to measure chromium by CFAES from concentra-tions below 0.1 ng ml-1 to almost 1000 pg ml-1 with this system but it is not possible to measure at the extreme ends of this range (about 7 orders of magnitude dynamic range) using a single amplification factor. The range of the ADC in such instances may determine the effective concentration range that can be measured for a given amplification factor. Where careful selection of amplification is employed for individual channels it is likely that the range afforded by the ADC will be suitable for most CFAES applications.The most serious limitation of the technique for the measurement of CFAES signals at high concentrations (500-1 000 pg ml-l) is the broadening of the emission-line profile across the entire modulation interval. This effect has also been found to be the limiting factor in CSAA linearity.19 When the line broadens across the modulation interval it is no longer possible to obtain a true background measurement as analyte intensity will also be measured at the background data point positions. The background-corrected intensity observed will therefore be less than it would have been in the absence of line broadening and calibration graph curvature will occur. The modulation interval used in the present work of about 0.1 nm (depending on wavelength) is sufficiently wide that this effect is normally observed only at concentrations in excess of 500 pg ml-l.Although this problem could be avoided by increasing the modulation interval the practicality of removing samples of higher concentration from the furnace may imply that the range of calibration graph linearity offered by the SIMAAC system is more than adequate for most CFAES applications 186 A fialyst VoZ. 108 Analysis of NBS Standard Reference Materials The purpose of this work was to demonstrate the potential of the carbon furnace as an emission source for simultaneous multi-element analysis. It has already been indicated that as a result of inappropriate slit widths and wavelength selection in the SIMAAC system, sensitivities obtained in the multi-element mode were poorer than those achieved in a single-channel operation.Therefore samples containing relatively high concentration levels in terms of absolute CFAES sensitivity of elements of interest were chosen to demonstrate the multi-element analysis capability of CFAES and also the application of the calibration graph extension technique. The elements for which suitable certificate values were available were sodium potassium chromium calcium copper and iron. The application of CFAES to multi-element analysis was demonstrated by the determination of these elements in NBS standard reference materials and orange and pineapple juice samples. MARSHALL et al. SIMULTANEOUS MULTI-ELEMENT - 1 TABLE I11 E8 .9 .' . SIMULTANEOUS MULTI-ELEMENT ANALYSIS OF STANDARD REFERENCE MATERIALS BY CFAES Chromium/vg g-' Sodium/pg g-l Calciumlpg g-' Sample CFAES value CFAES value CFAES value CFAES value (-,-A--. - 7- --A- 7 r---Certificate Certificate Certificate Certificate Wheatflour 2.6 f 0.2 2.0 j 0.3 0.76 f 0.16 - 9.0 f 0.8 8.0 f 1.5 208 f 34 190 & 10 Rice flour . . 2.3 f 0.2 2.2 f 0.3 0.43 0.07 - 6.0 f 1.6 6.0 f 1.5 135 f 4 140 f 20 Bovine liver . . 185 f 8 193 f 10 1.0 f 0.4 0.088 f 0.012 2400 f 350 2430 f 130 128 f '2 124 f 6 Orange juice 0.44 f 0.1 0.29 f 0.1 0.14 f 0.06 - 4.1 f 0.92 4.3 f 1.2 86 f 1 92 f 5 Pineapple juice 0.59 f 0.12 0.45 f 0.15 0.12 & 0.01 0.16 f 0.06 2.9 f 1.4 2.7 f 1.6 105 f 8 98 f 5 Iron (single-channel result)t/ Copperlpg g-l Iron (multi-element result)*/ Potassium % v g g-1 vg g-' ,-- -.--A__- 7 7-* 7 CFAES Certificate value CFAES Certificate value CFAES Certificate value Whear flour .. . . . . 0.01 f 0.002 0.13 0.004 11.5 6.1 18.3 f 1.0 15.2 f 0.5 18.3 f 1.0 Rice flour . . . . . . 0.09 f 0.01 0.11 f 0.002 9.7 f 2.7 8.7 f 0.6 6.4 0.6 8.7 f 0.6 Bovine liver . . . . . 1.2 f 0.22 0.97 f 0.06 187 f 80 Orange juice . . . . . . 0.23 f 0.037 0.19 f 0.01 Pineapple juice . . . . 0.057 f 0.004 0.072 f 0.004 6.6 f 3.3 268 8 345 f 7 268 rt 8 4.6 f 2.3 1.22 - -- - -* Wavelength 302.06 nm in multi-element mode. t Wavelength 371.99 nm in single-element mode. The results of the analyses are presented in Table 111. The values given are the average results for triplicate digests and analyses.Certificate values are quoted for NBS standard reference materials wheat rice flour28 and bovine liver.29 The values presented for the juice samples were previously obtained by continuum-source atomic-absorption spectro-metry using plane atomisation.lg In general good agreement was obtained between the CFAES results and the certified values. The spread of results is illustrated in Fig. 9. A relatively wide dynamic range of five orders of magnitude of concentration is exhibited February 1983 ANALYSIS BY CARBON FURNACE AES 187 Some problems that contributed to a deterioration in the quality of the analytical results produced were identified. Spectral interference was found to be a problem for the deter-mination of iron at 302.1 nm which was the wavelength used for multi-element analysis.Extremely poor precision was obtained for iron owing to a deterioration in the signal to noise ratio a t this non-optimum wavelength and as a result of over-correction for the additional spectral background. The use of the optimum CFAES wavelength a t 371.99 nm under single-element conditions provided better precision and results closer to the certificate values, as can be seen from a comparison of the results for iron in Table 111. Recent work suggests that the differences between the CFAES results and the certificate values may result from low recoveries from the sample preparation procedure. The samples were analysed by continuum-source atomic-absorption spectrometry using platform atomisa-tion and the recoveries are presented in Table IV.High blanks were produced by the digestion method and the values reported for the CFAES results have blank values subtracted. The recoveries obtained for other elements were less than 100% and appeared to correlate with the slight discrepancies observed between the CFAES results and the certificate values. The samples were analysed directly in the multi-element mode by CFAES without prior method development in the single-element mode. Although acceptable results can be obtained using this approach initial development of the analytical method in the single-element mode may be preferable for the diagnosis of any spectral or chemical interference problems. However this approach may be considered prudent for all such atomic spectro-metric systems and is not considered to be a specific limitation of CFAES.TABLE IV RECOVERIES OBTAINED IN THE ANALYSIS OF NBS STANDARDS USING CONTINUUM-SOURCE ATOMIC-ABSORPTION SPECTROSCOPY (CSAAS) Recovery yo i i G Z i c = Element SRM SRM Calcium . . 90 94 Iron 86 83 Potassium . . 77 78 Sodium* 207 232 Copper . . 95 91 * Not blank corrected. Conclusions It is concluded that the use of a background-corrected direct-reading polychromator such as the SIMAAC instrument offers considerable potential for the development of a multi-element detection system for carbon furnace atomic-emission spectrometry. The extension of analytical calibration graphs in CFAES considerably enhances the versatility of the tech-nique and affords a linear dynamic range of up to five orders of magnitude in concentration for a number of elements.Clearly applications of CFAES already described will be feasible in the multi-element mode and this may significantly widen the potential of the technique. The application of an inductively coupled plasma - wavelength modulation echelle spectro-meter system to marine analysis has recently been described.30 Hence the type of detection system described in this paper is applicable to flame plasma or furnace emission sources as well as flame and furnace continuum-source atomic absorption and as such offers a flexible approach to multi-element analysis by atomic spectrometry. This work was made possible by the US Department of Agriculture which provided the Financial support from the Pye Foundation (for D.L.) and the facilities for this research.SERC (for J.M.) is gratefully acknowledged 188 MARSHALL et al. References Mavrodineanu R. and Hughes R. C. Appl. Opt. 1968 7 1281. Strasheim A. and Human H. G. C. Spectrochim. Ada Part B 1968 23 265. Mitchell D. G. Jackson K. W. and Aldous K. M. Anal. Chem. 1973 45 1215A. Johansson A. and Nilsson L. E. Spectrochim. Acta Part B 1976 31 419. Felkel H. and Pardue H. L. Anal. Chem. 1977 49 1112. Morrison G. H. Anal. Chem. 1977 49 106. Chuang F. S. Natusch D. F. S. and O’Keefe K. R. Anal. Chem. 1978 50 525. Ullman A. H. Pollard B. D. Boutilier G. D. Bateh R. P. Hanley P. and Winefordner J . D., Busch K. W. and Morrison G. H. Anal. Chem. 1973 45 712A. Pickford C. J. and Rossi G. Analyst 1973 98 329. Lundberg E. and Johansson G.Anal. Chem. 1976 48 1922. Alder J . F. Alger D. Samuel A. J. and West T. S . Anal. Chim. Acta 1976 86 301. Salin E. D. and Ingle J . D. Appl. Spectrosc. 1978 32 596. Zander A. T. O’Haver T. C. and Keliher P. N. Anal. Chem. 1976 48 1166. Harnly J. M. and O’Haver T. C. Anal. Chem. 1977 49 2187. O’Haver T. C. Harnly J . M. and Zander A. T. Anal. Chem. 1977 49 665. Harnly J . M. O’Haver T. C. Golden B. and Wolf W. R. Anal. Chem. 1979 51 2007. Ottaway J. M. Bezur L. and Marshall J. Analyst 1980 105 1130. Harnly J. M. and O’Haver T. C. Anal. Chem. 1981 53 1291. Littlejohn D. and Ottaway J. M. Analyst 1977 102 553. O’Haver T. C. Harnly J . M. and Zander A. T. Anal. Chem. 1978 50 1218. O’Haver T. C. Epstein M. S. and Zander A. T. Anal. Chem. 1977 49 458. Koirtyohann S. R. Glass E. D. Yates A. D. Hinderberger E. J. and Lichte F. E. Anal. Chew., Epstein M. S. Rains T. C. and O’Haver T. C . Appl. Spectrosc. 1976 30 324. Bezur L. Marshall J. Ottaway J . M. and Fakhrul-Aldeen R. Analyst submitted for publication. Littlejohn D. and Ottaway J. M. Can. J. Spectrosc. 1979 24 154. “NBS Certificate of Analysis Standard Reference Material 1567 Wheat Flour,” National Bureau “NBS Certificate of Analysis Standard Reference Material 1568 Rice Flour,” National Bureau of “NBS Certificate of Analysis Standard Reference Material 1577 Bovine Liver,’’ National Bureau McLaren J . W. and Bermann S. S. APPZ. Spectrosc. 1981 35 403. Anal. Chem. 1979 51 2832. 1977 49 1121. of Standards Washington DC 1978. Standards Washington DC 1978. of Standards Washington DC 1977. Received August 25th 1982 Accepted September 20th 1982 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30

ISSN:0003-2654    

DOI:10.1039/AN9830800178

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst February 1983 Vol. 108 $9. 189-195 189 Determination of Trace Impurities in Sodium Coolant from a Fast Breeder Reactor by Inductively Coupled Plasma Atomic-emission Spectrometry Thomas Berry Kenneth C. Macleod Alistair C. Christie and James A. Cunningham United Kingdom Atomic Energy Authority (Northern Division) Dounreay Nzicleav Powev Development Establishment Thurso Caithness K W14 7TZ Trace impurities in samples of the liquid sodium coolant from the Dounreay Fast Reactor are determined by iritluctively coupled plasma (ICP) atomic-emission spectrometry of vacuum cl.istillation residues. A 72-channel vacuum spectrometer with a 1-kW argon plasma source fed by a coaxial capillary nebuliser simultaneously measurc.s 59 elements with a precision varying from ~ 1 % a t concentrations 20 times the detection limit to 30% at detection levels.The ICP torch box is enclosed in a fume-cupboard fitted with a filtered extract system to allow safe handling of the radioactive solutions. Comparisons of some results haw been made with spark-source mass spectro-metry d.c. arc emission spectrography neutron-activation analysis and atomic-absorption spectrometry. Keywords Liquid sodium ; sodium distillation ; inductively coupled plasma ; atomic-emission spectrometry ; impurity determination The Prototype Fast Reactor at the Dounreay Site of the UKAEAl is the only UK reactor that uses pure liquid sodium metal as the medium for transferring heat from the reactor core to the steam-raising plant. The measurement of impurities in the sodium coolant is an essential part of the Fast Breeder Reactor development programme leading to the design of the Commercial Fast Reactor.Corrosion and erosion can lead to the presence of dissolved and particulate material in t l ~ sodium and a proportion of the dissolved species adsorbs on to the surfaces contacted. An indication that some elements are present as particulates is given by significant differences in their concentrations whilst other elements do not vary between samples taken simultaneously. Ingress of foreign material is unlikely but there are some oil seals and the circuit is opened for maintenance; the impurities monitored cover all the elements that could be incorporated accidentally. Particularly in the primary heat transfer circuit where the coolant encounters the intense neutron flux in the reactor core absorption of neutrons occurs with the production of activa-tion products.Elements having a high fast neutron capture cross-secti ln are consequently undesirable but even for lower neutron capture rates the activation products may be un-desirable. Some examples of undesirable reactions are as follows (1) 1°B(n,cc)'Li high capture cross-section ; (2) 6Lijn,~)~H gaseous product that can diffuse into structural materials and react with sodium; and (3) 6oNi(n,p)60Co y-active product of long half-life. The sodium is analysed currently for 70 elements and this paper is confined to a description of the simultaneous determination of 59 elements in distillation residues by inductively coupled plasma atomic-emissio I spectrometry (TCP-AES).Prior to the introduction of the ICP-AES method a t Dounreay 2 years ago a combination of neutron-activation analysis, d.c. arc emission spectroscoply m d flame atomic-absorption spectrometry were used. Some samples have been analysed by spark-source mass spectrometry. Other workers in addition 10 the earlier methods used at Dounreay have described the use of visible and ultraviolet light absorption spectrophotometry or colorimetry,2 X-ray fluorescence spectrometry3 and graphite rod atomic-absorption spe~trometry.~ The ICP-AES technique described here is generally more precise has lower detection limits for corn-parative sample sizes and provides more results simultaneously than the alternative reported methods. Crown Copyright 190 BERRY et al.DETERMINATION OF TRACE IMPURITIES IN Analyst VoZ. 108 Experimental Apparatus The spectrometer and ICP source were supplied by Applied Research Laboratories and details of a similar instrument have been published elsewhere.5 The main differences in the Dounreay Quantovac from those given in Table I of reference 5 are as follows (1) only the grating with 1080 lines mm-1 is fitted; (2) there is no de-salter on the nebuliser but the argon is saturated with water vapour by passage through a bubbler; (3) the Scott-type spray chamber is used; (4) there are 71 channels compared with the 18 listed; and (5) the mini-computer is a DEC PDP 11/03 having 32K words RAM and ARL SAP-11 software is used. The wavelengths used for ICP-AES analysis are listed in Table I; the other channels are associated exclusively with the mono-alternance spark source used for metal billet analysis or do not apply here (e.g.carbon sulphur) and are not given. The 72nd channel is avail-able for fitting an additional phototube or monochromator. Sixty elements can be read simultaneously with printout of results in concentration units corrected for spectral-line interferences. A fume-cupboard has been fitted to enclose the ICP torch box which permits operation with slightly radioactive solutions including the solutions from the distilled sodium residues. Owing to its mass the fume-cupboard has to be supported by the floor so shock-absorbing seals interface it to the ICP torch box. The fume-cupboard extract removes the gases and entrained sample aerosol from the torch via a bank of high-efficiency particle removal filters and a 20-m stack to atmosphere.The fume-cupboard and discharge arrangements comply with the relevant safety requirements. Sampling Obtaining a representative sample is difficult-theoretically the whole primary sodium circulates through the reactor core once every 5min at full power so good mixing should result. Local variations in purity can occur at the sodium - blanket gas interface and at any particular corrosion or erosion sites stagnant pockets and leaks. On-line sodium samplers are provided and are depicted schematically in Fig. 1; they allow four heated sample crucibles to be washed by sodium at a rate of 18 1 min-l into the sampling chamber for 4 h. The sodium is then drained from the chamber leaving the crucibles full of sodium.The cooled crucibles containing solid sodium are transferred in a sealed transit vessel to the sodium analysis laboratory for further work. The sodium samplers are attached to an argon-atmosphere positive-pressure glove-box into which the crucible holder can be withdrawn for access. Air cooling jacket To sodium drain tank Fig. 1. Sodium sampler February 1983 SODIUM COOLANT FROM A FAST BREEDER REACTOR BY ICP-AES 191 Samples for oxygen hydrogen or carbon analysis require extra precautions to avoid atmospheric pick-up and a separate facility is used to fill a tube rather than crucibles in this instance. The crucibles used have included nickel and alumina but are currently stainless steel with fused silica liners; some pick-up of the main crucible element occurs so this must be excluded from the analysis schedule.Solution Preparation The sodium samples when first removed from the reactor are y-active and shielding must be provided and handling precautions taken. The sodium itself becomes y-active from the ingrowth of sodium-24 and -22 isotopes the former providing the major contribution but having the shorter half-life of 15 h. The sodium is distilled from the sodium oxide and other remaining impurities prior to analysis with the following advantages the sodium metal need not be dissolved; the impuri-ties are pre-concentrated giving good detection limits at a total salt concentration of only a few milligrams per millilitre; most of the y-activity is found in the distillate; and the distilla-tion residue is suitable for neutron-activation analysis (NAA) coupled with y-spectrometry ; without sodium separation the NAA detection limits would be unsatisfactory owing to the high y background.If present in elemental form certain elements for example cadmium caesium potassium and zinc distil with the sodium and accurate results for these elements must be derived from undistilled samples. Pick-up of oxygen to give a total of about 300 p.p.m. of oxygen is arranged prior to distillation to retain better the impurities with the undistilled sodium oxide. The transit vessel containing four crucibles of sodium in their carrier is placed in an argon-atmosphere positive-pressure glove-box in the sodium analysis laboratory and immediately prior to distillation two crucibles are transferred into the sodium distillation vessel (Fig.2), which is then closed The distillation vessel is removed from the glove-box and connected to the distillation rig by the upper coupling; the rig is evacuated up to this coupling before opening the Saunders valve. Fig. 2. Sodium distillation pot. The base of the sodium distillation vessel is heated electrically and vacuum is provided by a rotary pump and mercury vapour diffusion pump with a liquid nitrogen trap. The sodium from two crucibles (8 g) is distilled at a pressure of 0.1 N m-2 at the diffusion pump using a programmed heating cycle that culminates in a thermocouple temperature of 500 "C. The distillate collects on a cold finger and on the upper walls of the vessel 192 Analyst VoZ.108 The combined residues in the two crucibles amount to about 10 mg and are dissolved by boiling in 1 N nitric acid under reflux for 1 h and the final solution volume is made up to 5.0 ml with 1 N nitric acid; PTFE laboratory ware is used to minimise pick-up of boron, aluminium calcium and other elements. The dissolution efficiency is checked by total y-counting and high-resolution y-spectroscopy allowing recoveries to be calculated for certain elements. Concentrated nitric or hydrochloric acid has been used to dissolve some sample residues and has given the same results as 1 N nitric acid. This indicates that no intractable carbides oxides or other compounds are present in significant amounts. The sodium in the remaining two crucibles is available either for dissolution without prior distillation of sodium to allow determination of impurities by atomic-absorption spectro-metry or by ICP-AES or is distilled to provide duplicate residue impurity results by ICP-AES.Mercury is determined without prior sodium distillation by cold vapour atomic-absorption spectrometry. The combination of the decay of the y-activity occurring since removal from the reactor and distillation of the sodium leaves less than 10 nCi ml-l of analysis solution which is acceptable with the ICP system and fume-cupboard arrangement described. The blank solution is usually prepared by leaching clean crucibles as for the sample solutions. The longer procedure of placing crucibles in the inverted position in the on-line sampler subjecting them to the distillation procedure and counting prior to leaching has little effect on the blanks obtained with the preferred silica crucibles.BERRY et at?. DETERMINATION OF TRACE IMPURITIES IN Measurement In order to achieve the required detection limits precision and accuracy the sample transport system and ICP torch must be thoroughly clean. The ICP must have been ignited long enough to have reached thermal equilibrium and the gas flow for sample injection should be adjusted with a calibrated flow meter prior to running the calibration standards. The argon flow-rates are adjusted to coolant 10.5 1 min-l plasma 1.5 1 min-l and sample transport 1.0 1 min-l and the uptake rate for 1.0 N nitric acid of 0.8 ml min-l through the peristaltic pump is checked.The pump should be adjusted so that no flow is possible with the pump stopped thus ensuring that there is no aspiration effect leading to increased sample flow-rat es. The optical alignment is checked for two elements near the slit frame ends following the method in the operating manual and adjusted by moving the entrance slit if required. De-ionised water is used to prepare solutions and nitric acid is re-distilled in silica equip-ment before use. Glass pipettes are used otherwise the laboratory ware is of PTFE. The master standard solutions are 1 g 1-1 in 1.0 N nitric acid or exceptionally 1.0 N hydrochloric acid where required for stability and a number of intermediate mixed standards are prepared and diluted to working standard strength. The analysis solutions are analysed with the minimum of delay as there is a possibility that some elements could precipitate on standing if present in sufficient concentration.The control solution contains 0.2 pg ml-l of each element. A preliminary check of the sodium content of the analysis solutions is made and an equivalent concentration of Specpure sodium nitrate is included in the calibration standards, blanks and control solutions. The ICP torch has no constriction in the sample injection capillary and can handle concentrated sodium nitrate solutions without blocking. The argon supply to the nebuliser is bubbled through water which ensures that the nebuliser can handle concentrated solutions without crystallisation occurring. Three-point calibration is used usually at zero 2.5 and 5.0pgml-l and the working graphs are linear over this range.Ten replicates of 10-s integrations of the millivolt signal are made and the computer program produces the standard deviations. Twice the standard deviation of the zero standard element signals converted into micrograms per millilitre is used to define the detection limits. If the sensitivity and detection limits are close to normal values the calibrations are incorporated in the computer program. The blank and sample solutions are measured then the calibration is repeated and finally the control solutions are measured. Usually only 2 ml of solution is available for ICP analysis and it is all consumed. Further samples taken at the same time may be presented for comparative purposes February. 1983 SODIUM COOLANT FROM A FAST BREEDER REACTOR BY ICP-AES 193 At the present composition of the sodium.atomic-line interferences are not excessive and the correction factors can be incorporated in the computer program . Elements contributing line interferences are given in Table I . Should the concentration of individual elements increase significantly then possible line interferences are re-assessed and any additional corrections are applied . The interference correction factors are checked some time prior to the analysis using 1 mg ml-1 solutions. allowance being made for the impurities present in these . The control sample results for individual elements are plotted on Shewart mean charts to verify that the analytical method is functioning satisfactorily.6 TABLE I ICP DATA AND COMPARISON OF RESULTS FROM DIFFERENT ANALYTICAL TECHNIQUES FOR ELEMENTS I N SODIUM METAL Concentration.p.p.m. Relative -.A- Wavelength/ standard NAA* MSt D.c. arc AASS Other nm Line interference error "i; . Element ICP-AES if - * As Au . . B Ba Be Bi Ca . . Cd . . Ce . . co . . Cr cs . . c u . . Eu Fe . . Ga . . Gd Ge . . Ho . . In K La Li Lu . . Mo Nb Nd Ni P Pb Pd Pr Pt Rb Rh Ru . . Sb s c Se . . Si . . Sm . . Sn . . Sr . . T a Tb Th Ti TI Tm v w Y Yb Zn . . Zr . . ;; ;; * * . . ~ 0 . 0 3 . . 1.3 . . ~ 0 . 4 . . 0.03 . . <0.1 . . 0.3 . . <0.001 . . (0.05 . . 1.3 . . (0.01 . . <0.15 . . < o m . . 0.10 .. 0.4 . . <0.02 . . <0.005 . . t 0 . 0 5 . . 2.7 . . <0.17 . . <0.002 . . <0.04 . . <0.003 . . <0.4 . . 1.1 . . <0.3 . . <0.001 . . 0.2 . . 0.05 . . <0.1 . . <0.01 . . <0.02 . . 0.3 . . <0.2 . . 0.4 . . <0.05 . . <0.07 . . <0.3 . . <0.4 . . <0.2 . . <0.15 . . <0.6 . . <0.001 . . t 0 . 7 . . 1.2 . . t 0 . 1 . . <0.2 . . 0.004 . . <0.1 . . <0.009 . . t 0 . 0 1 . . <0.05 . . <0.3 . . <0.3 . . <0.004 . . <0.3 . . to.001 . . <0.001 1.1 . . <0.002 0.012 1.3 < 0.05 0.032 <8 <0.1 0.01 0.10 <8 <0.01 2.8 <0.024 1.2 <0.01 < 0.1 <0.17 <1 <0.005 <0.1 <0.1 1.4 1.5 1.0 <0.02 0.2 <0.1 <0.1 < 0.04 <3 <2 <0.07 < 0.03 <I 0.15 < 0.6 <0.3 <2 <5 <0.3 < 0.1 t O .l <1 <O.l <0.1 <0.1 < 0.04 <0.03 <0.5 <2 0.08 <0.04 < 0.01 < 0.08 <0.14 <0.1 0.14 <0.1 0.3 <0.1 <0.1 <0.1 < 0.1 t 0 . 5 <0.3 < 0.1 t0.04 <0.2 <0.1 ~ 0 . 6 < 0.1 <0.1 <0.1 <0.5 0 . 3 < 0.04 <1 <2 <2 <2 <0.05 t0.5 <0.1 < 0.03 <0.1 <0.5 t 0 . 1 <0.4 <0.1 <0.1 t O . l ~ 0 . 0 0 3 t 0 . 0 2 <0.1 <2 < 0.3 <0.5 1.2 <0.1 < 0.3 <0.2 2.6 1.1 <0.1 0.2 <0.1 < 0.4 0.3 <0.5 <0.5 338.289 308.315 197397 242.795 0.02 249.678 233.527 313.042 223.061 315.887 226.502 446.021 228.616 286.257 455.530 324.754 353.171 369.265 393.050 259.940 403.298 303.285 265.118 345.600 325.856 766.491 433.373 650.784 261.542 279.553 257.610 281.615 319.498 401.215 231.604 178.287 220.361 360.955 422.533 292.979 780.023 343.489 379.896 206.838 364.279 196.026 251.612 425.640 189.980 407.771 301.254 350.917 374.119 337.280 377.572 376.133 311.071 400.875 371.029 369.420 213.856 349.621 05 W 0.67 0.067 0.067 0.4 0.13 0.67 0.54 0 s .Mn Cr. Nb. 05 Ta Er. Ho. Mn. Pd 1.8 Hf 0.47 0.067 Ce Ce. Cr. Th 0.27 0.40 Sm Sm Mo. Tm Ti 0.67 0.013 Ti Er E r Ti c u 0.40 * Neutron-activation analysis .t Mass spectrometry . $ Atomic-absorption spectrometry 194 BERRY et a,?. DETERMINATION OF TRACE IMPURITIES IN Analyst VoZ. 108 Results and Discussion The results of current analyses of the primary sodium coolant by ICP-AES are presented in Table I where comparisons are included with other analysis techniques. The control results charts are not included but recoveries are within the expected range based on the known standard deviations for each element. The ICP-AES precision obtained is expressed as the relative standard error of the mean of ten replicates and is the coefficient of variation divided by 3.16. The individual precisions for the alternative method results have not been included but in general the neutron-activation analysis mass spectrometric and d.c.arc precisions are worse by at least an order of magnitude. Flame atomic-absorption spectro-metric results are similar in precision to the ICP-AES results when related to the detection limits . Table I includes 61 elements but the detection limits obtained for caesium and lithium are inadequate for the present purpose and atomic absorption is used to measure these. The ICP-AES precisions vary as a function of the detection limit and the number of interference corrections applied. The detection limits for 34 of the elements equal or improve upon values obtained previously by other methods. Complete validation of the ICP-AES results in these instances is not possible otherwise agreement with the other techniques is very good, especially as each technique involved independent operators and standards.The ICP-AES analysis wavelengths are not always the most sensitive available either to exclude lines having unacceptable interference for our applications or to accommodate all the elements in the slit frame. Manufacture of the sodium by ICI was started in 1967 but the impurity levels reported show that the bulk material is still very pure. Interferences are likely to give an enhanced signal so the ICP-AES results can be regarded as maximum concentration values present. The advantage of having 65 elements lines on the spectrometer is apparent when the concentration of interfering elements is needed to calculate a spectral-line interference correction. Even so the spectrometer was not designed to measure osmium and hafnium which are two of the elements that could interfere.Pre-diction of ICP-line interferences for all wavelengths on the Quantovac is not possible from published literature and in the early days spark-source intensities’ were the only guide. The situation has improved with the publication of data for some prominent ICP emission lines8 but limited with some exceptions to the region 230-680 nm; ten of the Dounreay ICP wavelengths are not included in this publication. Hydride generation equipment has been obtained to reduce the detection limits for arsenic, bismuth tin antimony selenium tellurium germanium and lead but no results are available at present. By saturating the injector argon with water vapour and using a constant-bore sample tube in the ICP it is possible to increase the concentration of sodium in the solution to 10% m/V without salt build-up on the torch or nebuliser tip.This capability provides an opportunity to measure the relevant impurities in sufficiently aged sodium without prior distillation and should be useful for analysing cold trap samples of relatively high oxygen content that cannot be distilled in the normal manner. It will also allow further investigation of the extent to which individual elements distil over with the sodium; however there is a loss of sensitivity by this technique which increases the detection limits by an order of magnitude. Alternative nebulisers are being evaluated including a fritted glass nebuliser9 and a PTFE high-solids nebuliser.1° The authors thank many colleagues in the UKAEA Northern Division and at AERE Harwell for their contributions to this work particularly at Dounreay members of the PFR Operations Group and Sodium Laboratories. In this context useful information has resulted from discussions with Messrs. W. B. Bremner K. R. Melhuish J. McKie and D. Skea. References 1. 2. Moore R. V. Grieves J. O. Henry K. J. Cambell R. H. Evans A. D. Bishop J. F. W. Kemp, Silverman L. “International Series of Monographs in Analytical Chemistry,” Volume 44 Pergamon E. F. and Scott R. W. Nucl. Eng. Int. 1971 16 629. Press Oxford 1971 February 1983 SODIUM COOLANT FROM A FAST BREEDER REACTOR BY ICP-AES 195 Scheinder H. Griinhauser M. Nagel G. Nold E. Schafer A. and Schumann H. KFK 2267 1976 Garbett K. Goodfellow G. I. and Marshall G. B. Anal. Chim. Acta 1981 126 135 and 147. Dahlquist R. L. and Knoll J. W. AppZ. Spectrosc. 1978 32 1. Davies 0. L. and Goldsmith P. L. Editors “Statistical Methods in Research and Production,” Harrison G. R. “Wavelength Tables,” Second Edition MIT Press Cambridge MA 1969. Boumans P. W. J. M. “Line Coincidence Tables for Inductively Coupled Plasma Atomic Emission Layman L. R. and Lichte F. E. Anal. Chem. 1982 54 638. Ebdon L. and Cave M. R. Analyst 1982 107 172. 3. 4. 5. 6. 7. 8. 9. 10. (in German) UKAEA Risley Translation No. 3061 1976. Fourth Revised Edition Longmans London 1977. Spectrometry,” Volumes 1 and 2 Pergamon Press Oxford 1980. Received August 2nd 1982 Accepted September 28tk 198

ISSN:0003-2654    

DOI:10.1039/AN9830800189

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

196 Analyst February 1983 Vol. 108 fip. 196-203 Determination of Trace Elements in Biological Materials Using a Hollow-cathode Discharge: Comparative Study of Matrix Effects Sergio Caroli Oreste Senofunte and Pietro Delle Femmine Istituto Superiore d i Sunit; Laboratovco d i Tossicologia Viale Regina Elena 2'-!+ * ) I ) J !j I-]? W P Italy An emission source that does not appear to be significantly prone to mstr-: effects is the hollow-cathode discharge. In order to elucidate its effectiL potential a number of elements (aluminium arsenic calcium copper galliuni and zinc) were determined in the presence of various compounds (ortho-phosphoric acid sodium nitrate and potassium nitrate) over a wide range of concentrations and internal ratios. The same elements were determined in liquid samples resulting from mineralisation of biological materials (kidney, liver brain and blood of mice).The results sliowed that t h e hollow-cathode emission source is affected by this type of interference to a lesser extent than atomic-absorption spectrometry and arc-emission spectrography. Keywords Hollow-cathode discharge ; enzission spectroscopy ; atomic-absorp-tion spectrometry ; matrix effects Over the last decade an enormous almost revolutionary change has occurred in most industrial-ised countries in the control of environmental pollution and the protection of human health. In this respect governmental policies are nowadays much more directed towards prevention than in the past. We have on the one hand an improved knowledge of deleterious effects caused by chemicals and on the other steadily increasing technological progress.As a consequence it is necessary to determine ever smaller concentrations with higher accuracy and precision to meet the requirements of the modern analytical world. From this point of view spectroscopic methods are of immense value even though their potential in trace and sub-trace analysis can be seriously affected by matrix differences. More or less complicated procedures are therefore often necessary in order to overcome this detri-mental influence. The simultaneous presence of other elements is in fact a potential cause of interferences whenever quantitative determinations are carried out. Chemi-cal interferences classified in the first group are generally said to cause variations in analytical signals as they alter the population density of free ground-state atoms through the formation of more or less dissociable compounds.Ionisation effects belong to the second group and are due to the fact that matrix elements can modify the electron density of the source and thus influence the ratio of neutral to ionic species of the analyte in the excitation zone. This over-simplification may be deceptive to some extent as the mechanism of these interferences and measures for their prevention are not yet fully understood. Even though there are other phenomena of minor importance that appear to be matrix-related it can be assumed that the two mentioned above account for the main difficulties normally encountered when performing quantitative determinations by spectrochemical methods.Spectral interferences are not taken into account here although they are of great importance, as the purpose of this investigation was to elucidate the extent to which analytical signals generated by known amounts of analytes can be altered by the simultaneous presence of other species. In fact spectral overlap can make a line unsuitable for quantitative determinations, but the difficulty is usually overcome by resorting to other more convenient lines or by adopting appropriate correction methods. An emission source that does not appear to be very prone to matrix effects is the hollow-cathode discharge (HCD) .1-6 The analytical potential of this excitation device is inherently powerful for a wide range of applications including the determination of trace elements in complex samples.An investigation of the intrinsic sensitivity of the HCD to matrix varia-tions was therefore undertaken. Moreover in order to have a better over-all evaluation of the performance of the HCD the results were compared with those obtained by two other basically different spectroscopic methods namely atomic-absorption spectrometry (AAS) and arc-emission spectrography (AES) . For the sake of simplicity inter-element effects can be grouped into two categories CAROLI SENOFONTE AND DELLE FEMMINE Experimental 197 A two-step approach was adopted for investigating the real relevance of matrix effects in the HCD. In the first phase six representative elements namely aluminium arsenic calcium, copper gallium and zinc were subjected to analysis using an HCD in the presence of very simple synthetic matrices consisting of various amounts of orthophosphoric acid sodium nitrate and mixtures of the latter with potassium nitrate over the entire composition range.The second part of the research was devoted to the analysis of the same elements directly in samples resulting from mineralisation of biological materials. Measurements were performed by using for each method the best procedures available for circumventing the influence of matrix differences and thus contributing to the elucidation of the over-all ability of the three techniques for the determination of trace elements in biological samples. The procedure adopted for preparing the series of samples necessary for the second phase of the study in-cluded the following main steps (;) five male DBA/2 mice healthy and of similar age and mass, were killed the blood was collected by cardiac puncture and various organs (brain liver and kidney) were excised; (ii) blood and organs of the same type were then pooled and subjected to wet ashing with hot concentrated nitric acid (Merck Suprapur) until clear solutions were obtained on cooling; (iii) half of each mineralisation solution was diluted with doubly distilled water to 150 ml and stored for blank determinations; (iv) the second half of each solution was added with standard solutions containing the six elements under investigation at concentrations that after dilution to 150 ml with doubly distilled water of the final mixtures their respective amounts were much higher than those correspondingly determined for the blanks.All measurements were performed in triplicate and the results were averaged. TABLE I ELEMENT CONCENTRATIONS AND MOLAR RATIOS OF ANALYTE TO MATRIX COMPONENTS FOR SYNTHETIC SAMPLES ANALYSED USING AN HCD Element Concentration p.p.m. Matrix molar ratios Al as Al(NO,) . . . . 4 M* + xH,PO, x = 10 25 50 100 250, 500 1000 As as As,O . . 12 Ca as CaCl . . . . 12 M + xNaNO, x = 33 66 100 1000 Cu as Cu(NO,) . . 4 M + x[yNaNO + (1 - y)KNO,I < Y < 1, Ga as Ga(NO,) . . 12 x = 33 66 100 1000 Zn as Zn(NO,) . . 12 * M = Al As Ca Cu Ga Zn. Synthetic solutions were analysed using the HCD only. In fact the effects caused by phos-phate sodium and potassium on the performance of the other two techniques have been described at length in several papers e.g.references 7-10 whereas there is a lack of data for the first emission source for analysing solution residues. It was decided to investigate in detail the effects of the above three chemical species as they are known to cause many difficul-ties in the determination of trace elements by AAS and AES and are also amongst those TABLE I1 CONCENTRATIONS OF ANALYTES ADDED TO SOLUTIONS RESULTING FROM MINERALIS ATION Concentration p.p.m. , Element HCD AAS AES A1 . . 4 4 4 As 12 12 -* Ca . . 12 4 4 cu 4 4 4 Ga . . 12 4 4 Zn . . 12 0.4 1 * Determination impossible owing to spectral interferences 198 CAROLI et al. DETERMINATION OF TRACE ELEMENTS IN Analyst VoZ. 108 elements generally found in relevant concentrations in biological materials.Table I sets out concentrations of elements and the molar ratios of the analyte to the matrix. With biological materials the six elements were added to the solutions resulting from wet ashing as reported in Table 11. The concentrations differ slightly however from those selected for synthetic solutions in AAS and AES. In fact owing to the inherently different operating principles underlying the three methods it is obvious that for a given element analytical parameters such as spectral wavelength or analyte concentration usually cannot coincide. The experimental conditions have to be optimised separately for each method this being much more a stringent requirement to guarantee the reliability of the comparison than a formal coincidence of the above parameters.The experimental details concerning each of the three techniques are given below. TABLE I11 EXPERIMENTAL CONDITIONS FOR THE INVESTIGATION OF MATRIX EFFECTS WITH THE HCD Conditions Vacuum spectrograph SPV lm/800 KSV Electric source unit HVG 2 RSV Spectrograph entrance slit 30 pm Concave grating 1 m radius 1200 grooves nun-' blazed at 5' 52' Paschen - Itunge mounting Dispersion 0.78 nni mm-' at 300 nm Emulsion Kodak SA-1 plates Developer Kodak D-19 (4 min 20 f 0.1 "C) Exposure time 60 s Microdensitometer MD-100 Jenoptik Jena Carrier gas argon 440 Pa Current intensity 200 mA Spectral lines/nm Al (I) 396.1 As (I) 235.0 Cu (1) 329.0; (I) 327.4; (I) 324.7; (11) 224.7 Ga (I) 417.2 Zn (I) 213.9; (11) 317.2 Ca ( I ) 397.4; (11) 393.4 Hollow-cathode Spectrography A 0.4-cm3 volume of sample solution was introduced into the cathode cavity with a micropipette.Cathodes were then subjected to infrared irradiation under continuous rotation in order to deposit uniformly solution residues on the bottom and on the lower half of the cathode wall (Fig. 1). The dried residues were finally calcined in a muffle furnace for 20 min at 200 "C. Steel hollow cathodes were preferred for their resistance to chemical agents and their favourable thermal and electrical conductivity characteristics. Table 111 summarises the main information for this excitation source. Fig. 1. Procedure for filling hollow cathodes with solutions and drying before discharge February 1983 Atomic-absorption Spectrometry BIOLOGICAL MATERIALS USING A HOLLOW-CATHODE DISCHARGE 199 Instrumentation and operating conditions are listed in Table IV.When operating in the electrothermal mode use was made of tubes made of non-pyrolytic graphite. These allowed higher precision to be achieved even though their lifetime is shorter than that of tubes made of pyrolytic graphite. Background correction was applied for all elements except calcium for which it was completely unnecessary. On the other hand it was found to be indispensable for the determination of gallium and aluminium. Whenever operating with the graphite furnace it was purged with high-purity argon (interrupted flow mode). TABLE IV EXPERIMENTAL CONDITIONS FOR THE INVESTIGATION OF MATRIX EFFECTS WITH AAS Perkin Elmer Model 430 with deuterium background corrector and chart recorder Model 56.Lamp current Slit width/ Wavelength/ Atomisation/ Element intensity/mA nm nm excitation* A1 . . 22 0.7 309.3 A As . . -t 0.7 193.7 A CaS . . 18 1 .o 422.7 B cu 30 1.0 324.7 C Ga . . 20 0.7 287.4 A Zn . . 18 1.0 213.9 C * A HGA-76 graphite furnace; B dinitrogen oxide - acetylene C air - acetylene. 7 Electrodeless discharge lamp 8 W. 1 Background correction not applied. Arc-emission Spectrography Volumes of 10 cm3 of the sample solutions were added with equal amounts of solutions containing 10% of sodium sulphate and 10% of sulphuric acid and subsequently dried under infrared irradia-tion and calcined in a muffle furnace at 200 "C for 20 min.The residues were ground with care, placed in the electrode cup and then discharged. Table V reports the instrumental paramenters adopted for this emission source. TABLE V EXPERIMENTAL CONDITIONS FOR THE INVESTIGATION OF MATRIX EFFECTS WITH AES Conditions Element Current intensity/ A Spectrograph PGS2 Zeiss A1 . . 8 Arc generator Optica Ca . . 8 Entrance slit 20 pm plane grating cu . . 8 Ebert mounting Zn 8 650 grooves nim-l blazed at 5" 35' Ga . . . . 8 Intermediate diaphragm 5.0 Analytical gap 3 mm RingsdorH HW 0134 anode cups Emulsion Kodak SA- 1 plates Developer liodak D-19 (4 min 20 f 0.1 "C) Microdensitonieter MD-100 Jenoptik Jena Exposure time/ Wavelength/ S nm 10 309.3 10 422.7 10 324.7 50 287.4 10 334.5 Results and Discussion Synthetic Samples with the HCD Measurements of analytical signals for the six elements using the HCD revealed a consider-able dependence on matrix composition.At first each element was investigated in-dependently i.e. each solution contained only the element and the matrix components. The general pattern remained the same even when all six elements were contained in the same solu-tion. In this instance it was necessary to use the same concentration for each of the 200 CAROLI et al. DETERMINATION OF TRACE ELEMENTS IN Analyst Vol. 108 (12 p.p.m.) in order to maintain the analyte to matrix ratios reported in Table I. An example of this dependence is given in Fig. 2 which illustrates the behaviour of one line for each element in the same solution containing phosphate.I 1 -1.0 - 2.0 -3.0 Log( [ MeI/[H,P041) Fig. 2. Dependence of analyte emission intensity on phosphate anion concentration (HCD). A Cu(1) 324.7 nm; B Al(1) 396.1 nm; C Ga(1) 417.2 nm; D Zn(1) 213.9 nm; E Ca(1) 397.4 nm; and F As(1) 235.0 nm. This is however a false effect that can be easily explained if the particular mechanism operating in the HCD is taken into account. In fact according to the procedure described above salt residues form a deposit on the inner surface of hollow cathodes and their thickness depends on the absolute amount of matrix components if the analyte content is constant. As a consequence the percentage distribution of analyte atoms on the surface layer of the deposit decreases with increasing thickness of the latter.Fig. 3 gives a schematic representation of this phenomenon. l + l / Deposit \ 1 + 3 / Cathode wall Analyte 0 Matrix Fig. 3. Surface dilution of analyte atoms in the sample deposit on the inner wall of hollow cathodes. As the mechanism of hollow-cathode excitation is based on sputtering of the cathode surface through bombardment of noble gas ions it is understandable that an increase in matrix con-tent will lead to a decrease in the number of analyte atoms available for the sputtering process for unit surface area in unit time. This phenomenon can pass unnoticed if total amounts of residue are insufficient to build up deposits of sufficiently different thicknesses. Proof of this was achieved by taking one element as a reference and calculating intensity ratios.In this instance the ratio of analyte to reference is constant in the sample independent of deposit thickness. This was indeed so as shown by Fig. 4 demonstrating that the behaviour of the intensity ratios of two elements is unaffected by composition variations. Obviously this is true whichever element is chosen as the internal standard thus demonstrating that intensity ratios are virtually independent of phosphate anion variations Febmary 1983 BIOLOGICAL MATERIALS USING A HOLLOW-CATHODE DISCHARGE 20 1 0 A - - @ . 0.6 0 0 n 0 o o c 0'4 1 B 0 0.2 r;.:/ u~ -0.4 -0.6 -0.8 I 8 I O A I D O -1.0 -2.0 -3.0 Log ([Me]/[ HJPO~I 1 Fig. 4. Dependence of analyte emission intensity ratios on matrix concentration [Ga(I) as reference, 417.2 nm HCD].A Cu(1) 324.7 nm; B Al(1) 396.1 nm; C Zn(1) 213.9 nm; D Ca(1) 397.4 nm; and E As(1) 235.0nm. From a general point of view this is not surprising as analyses of solutions using an HCD are in practice always carried out on the dried residues i.e. on solid samples. The ratio of analyte to total matrix is therefore of outstanding importance for analyte determinations whereas the nature of the matrix components is much less important. This was ascertained in many other instances for trace elements contained in metal alloys,2s6 accuracy of determination not being affected by matrix variations. Another possibility for overcoming the phenomenon discussed above is to discharge hollow cathodes until the samples contained therein are totally consumed. In this instance the analyte will contribute to emission to an extent that depends only on its absolute amount, regardless of deposit thickness.A possible disadvantage is that the presence of large amounts of matrix will require long exposure times for discharging all of the sample. Therefore the spectral background will be considerably higher which can hinder determinations. Identical qualitative results were obtained when working with solutions containing the six elements and various amounts of either sodium nitrate or potassium nitrate or both at different ratios. Even in this instance the use of reference elements results in the intensity ratios being independent of matrix amount. Variation in type of cation (sodium or potassium) on the other hand does not seem to cause any alteration in the absolute signal.One example of this is given in Fig. 5 for some representative lines of copper. It is also deducible that a decrease in 3.0 2.0 m - m -I 1 .o 0% KN03 50% 100% NaN03 0 -0.2 -0.4 -0.6 -! -5 0 1 . -0.8 C U n " n 1 I -1.5 -2.0 -2.5 -3.0 Log([Me1/INaNO31) Fig 6. Dependence of analyte emission intensitv ratios on matrix concentration for 100% KN03 0% NaN03 Fig. 5. Copper emission intensity as a singly ionised spccies [Ga( I) as reference, function of matrix composition. Analyte to 417.2 nm HCD]. A Zn(l1) 317.2 nm; 13 Cu(I1) matrix molar ratio == 1 + 33 (HCD). Cu(1) 224.7 nm; and C Ca(1I) 393.4 nm. 324.7 nm A 1st min; and B 2nd min. Cu(1) 327.4 nm C 1st niin; and I) 2nd niin. Cu(1) 329.0 nm E 1st min; and F 2nd min 202 Analyst VoZ.108 emission intensity occurs as discharge proceeds. This is logical as a progressive diminution of deposit within the cathode ensues. An identical pattern is followed by all other spectral lines at all ratios of analyte to matrix over the entire sodium nitrate composition range. Attention is drawn to the fact that spectral lines emitted by singly ionised analyte species also seem to be unaffected by matrix composition to the same extent as neutral atoms. It is well known that HCD spectra obtained using argon as the carrier gas are due essentially to emission from neutral atoms. However a few lines of first ionisation can also be detected, which are thought to originate mainly from Penning ionisation provided that analyte atoms are present at sufflciently low concentration and possess an ionisation potential lower than the energy of argon metastables.11 There is no doubt that the over-all situation is much more complicated than this description implies.Nonetheless the absence of local thermal equili-brium in the plasma and the existence of at least three clearly identifiable groups of electrons with different energies12 provide sufficient evidence to state that Saha’s equation cannot be reliably applied to HCDs. The experimental findings strongly support the relative freedom from matrix effects for singly ionised atoms also. An example of this behaviour is illustrated in Fig. 6. CAROLI et aZ. DETERMINATION OF TRACE ELEMENTS IN Mineralisation of Samples with the Three Techniques Determination of the six elements after addition to solutions resulting from wet ashing of biological materials confirmed the general pattern followed by synthetic solutions when using the HCD.Considerable differences in the amounts of analytes were determined on the other hand by the other two techniques. The four biological materials investigated are in fact characterised by different contents in mineral salts as summarised in Table VI. TABLE VI AVERAGE MINERAL CONTENT OF BIOLOGICAL SAMPLES Concentration/pg g-l organ mass , Organ Ca Fe K Mg Na Blood 70 453 2940 58 5060 Brain . . 130 40 3490 160 2330 Liver . . 19 66 1510 220 630 Kidney . . 57 70 3390 200 2185 The data reported in Table VI are averages of many determinationslO carried out by AAS, one technique being sufficient for a rough evaluation of the concentration levels of relevant elements even though each determination can obviously be affected in turn by the nature of matrix under study.On the other hand blank determinations of the six elements were carried out in each instance with the same spectroscopic technique used for establishing the degree of recovery of the elements added. In all instances the procedure adopted for ascertain-ing the role played by different matrix compositions relied on the use of the standard additions method for one of the four samples and measurement of the elemental concentration for the other three using the calibration graph obtained for the first. The same procedure was then repeated in turn for the other samples.In this way it was possible to evaluate the degree of dependence of each spectroscopic method on matrix variations. Although this procedure only gives qualitative results the information provided is nevertheless sufficient for ranking the inherent influence of matrix composition on each method. Table VII reports for the three methods the range of deviation from nominal values of added elements. It is apparent that one set of calibration graphs is sufficient with the HCD to determine all elements reliably, independent of which organ was chosen the variation being virtually of the same order of magnitude as that of the relative standard deviation. For AAS and AES on the other hand, this is not so and therefore it is necessary to employ calibration graphs whenever the sample matrix varies.Emphasis is laid on the fact that with the HCD the use of reference spectral lines is for the reasons mentioned above essential and compensates for virtually all variations in total amount of matrix. In AES in contrast no beneficial effects were obtained by using interna February 1983 BIOLOGICAL MATERIALS USING A HOLLOW-CATHODE DISCHARGE TABLE VII 203 RANGE OF PERCENTAGE VARIATIONS ON NOMINAL VALUES FOR ELEMENT DETERMINATION IN BIOLOGICAL MATERIALS Results based on 16 individual measurements in each instance. HCD AAS AES 7 Element mo Lriation,*o/ orel % A1 . . 5-15 2 20-30 4 16-70 9 As 10-25 5 20-35 4 Ca 8-15 3 30-40 3 50-80 8 c u 5-12 3 15-20 5 20-80 6 Ga 3-7 2 16-25 5 30-90 8 Zn * . 5-16 4 20-40 4 15-90 7 - -standards.This is in agreement with the fact that this essentially thermal source is affected to different extents by the matrix depending on the analyte and reference considered and also on the experimental conditions. Conclusions The results clearly show that the HCD tends to be substantially independent of matrix effects in contrast to the other two methods considered provided that reference elements are employed. This can be explained by the fact that the transport of sample into the discharge zone and excitation in a water-cooled HCD are essentially the consequence of a mechanical process (ion bombardment) with little contribution from thermal effects. Atomisation is therefore hardly affected by the simultaneous presence of other elements and generally requires no or only partial corrections.This characteristic makes the hollow cathode particularly suitable for the analysis of trace substances in biological samples not only because the analytical results are more reliable as matrix differences are insignificant but also because no complicated procedures are required in order to circumvent these otherwise detrimental effects. The over-all evaluation process is therefore greatly simplified and no serious sources of external contamination are introduced. Although there is a lack of systematic research in this field these preliminary findings are in line with what has generally been experienced with sputtering sources. They can therefore contribute to a better comprehension of problems that have not yet been overcome. 1. 2 . 3. 4. 5. 6. 7. 8. 9. 10. 1 1 . 12. References h q u a I<. Purr APPl. ChF?tl. 1977 49 1595. Anzani lt. and Trucco R. Paper presented at X 11' Smi. Spettrochini. iIIlI Montccatini 3-6 Caroli S. Scnofonte O. .\limonti X. and Zimiiier I<. Spmtvosc. Ltptt. 1!)8 I 14 575. Caroli S . Alimonti .I. mid Petrucci I?. .4naf. Chini. Actn 1882 136 Bci!). Caroli S. Xlimonti R. Delle Fernmine P. and Shukla S IC. Anal. Chinz. Actu lW2 136 %25. Alimonti A. Caroli S. and Violante N. Spectvosc. L r t t . 1980 13 313. Van Schouwenburg J . Ch. and van der Wey A. D. Anal. Chini. Actii 19G6 36 243. Ramakrishna T. V. Robinson J . W. and West P. W. A n n l . Chiru. ,4cta l!)Mi 36 57. Halls D. J . and Towishend A, Axal. CJtim. Rcta 1966 36 278. Fu\va I<. l'ulido P. Mckay R. arid Vallee B. L. Amzl. Chfm. 1964 36. 2407. Coburn J . W. and I<av E, Appl. Phys. I,&. 1971 18 438. Howorka F. and Pahl JI. 2. Natuvfovscti. Tpil - 4 1972 27 1429. October 1968. Heceived July 14tA 1982 Accepted Azigztst 18th 198

ISSN:0003-2654    

DOI:10.1039/AN9830800196

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

204 Analyst February 1983 Vot. 108 j!@. 204-212 A Critical Comparative Study of Atomic-spectrometric Methods (Atomic Absorption Atomic Emission and Inductively Coupled Plasma Emission) for Determining Strontium in Biological Materials Alfredo Sanz-Medel and Rosa Rodriguez Roza Department of Analytical Chemistry Faculty of Chemistry University of Oviedo Oviedo Spain and Conchita Perez-Conde Department of Analytical Chemistry Faculty of Chemistry Universidad Complutense Madrid Spain As part of a general research plan aimed at establishing tolerable doses of strontium and its distribution in selected soft tissues of rats and hamsters treated with stable strontium a comparative study on the analytical per-formance of flame atomic absorption flame atomic emission and inductively coupled plasma (ICP) emission for biological strontium assays especially in blood serum has been carried out.Optimum conditions for the various methods were established and analytical performance characteristics were evaluated for each method in terms of limits of cletection dynamic range, selectivity and precision attainable using the same basic instrument (the Perkin-Elmer ICP/5000). ICY emission spectrometry appeared to be the best method as i t provides a sensitivity of about 100 times better than the next most sensitive method a linear calibration graph over five orders of magnitude good precision in real sample analysis and virtual absence of spectral or chemical interferences (although nebulisation and transport effects have to be allowed for) from those elements and organic matrices common in biological materials.A dinitrogen oxide - acetylene flame in the presence of a sodium content approximating the blood sodium content increases sensitivity and selectivity when compared with an air - acetylene flame flame emission being about four times more sensitive than flame atomic absorption for both flames. Results for the determination of strontium in blood serum brain and liver o f rats and hamsters treated with stable strontium are also reported. Keywords Strontium determination ; biological materials ; flame atomic-absorption spectrometry ; flame atomic-emission spectrometry ; inductively coupled plasma emission spectrometry Determination of strontium in body fluids and tissues has taken on remarkable importance over the past few years following the release of radiostrontium strontium-90 in nuclear tests and the awareness of its dangerous retention in bone (where it becomes an internal radiation source).I t is now well documented that a straightforward approach in attempting the removal of radiostrontium is by the administration of stable strontium a trace element that is present in diet and tissues. The therapeutic use of a stable carrier however involves the administration of this element in at times large amounts and over a period of several days.' Unfortunately toxic effects have been observed in rats fed with a diet enriched with stable strontium. The origin of many aspects of such toxicity is still unknown2s3 and much more work is needed for the assessment of the role of strontium in health and disease.To cope with the necessary analyses and control for these studies the various methods of analytical atomic spectrometry appeared to be the most promising techniques. Application of atomic spectrometry to the determination of strontium in body tissues and fluids is still comparatively scarce.4 Strontium can be determined in biological materials by flame-emis-sion s p e c t r ~ r n e t r y ~ ~ ~ but this determination is carried out mainly by atomic absorption using a flame718 or with a graphite furnace when sensitivity is the limiting factor e.g. normal strontium levels in blood serumg or in hard tissues.1° Inductively coupled plasma (ICP) emission spectrometry is becoming more and more im-portant for the determination of trace elements in the bio-medical field.l SANZ-MEDEL RODRIGUEZ ROZA AND PEREZ-CONDE 205 Some papers illustrate the use of ICP atomic-emission spectrometric (ICP-AES) techniques for simultaneous multi-element analyses of about 20 elements (including strontium) in plant material12.l3 and in hard tissues.l* Pehlivanian et al.15 have studied in detail the application of ICP-AES to the determination of major and trace elements in blood serum.As part of a general research plan aimed at establishing tolerable doses of strontium and the distribution of strontium in selected soft tissues of rats and hamsters treated with stable strontium a comparative study of the analytical performance of flame atomic absorption, flame atomic emission and ICP-AES for biological strontium assays especially in blood serum, has been carried out.The possibility of using the same basic optical system afforded by the commercial instrument employed in this work makes possible a sound comparison from which the best technique can be chosen. Experiment a1 Apparatus The Perkin-Elmer ICP/5000 system equipped with a recorder to optimise operating condi-tions was used. Most of the experiments on flame atomic absorption and flame atomic emission were carried out in the atomic-absorption mode of this instrument using the same 5.5-cm optical path dinitrogen oxide - acetylene slot burner (to facilitate comparison of the analytical characteristics obtained by each method). Only for experiments on interferences occurring in air - acetylene flames was a conventional 10.4-cm optical path air - acetylene slot burner used.ICP emission measurements were carried out in the ICP mode of the ICP/5000 system for which some of the specifications and optimised operating parameters for strontium determina-tions of real samples are given in Table I. TABLE I INSTRUMENTATION AND PLASMA OPERATING CONDITIONS Spectrophotometer ICP source Optimum conditions for strontium determination Sequential scanning 2 kW-27.12 MHz Plasma power . . 1.15 kW < 0.03-nm band pass Glass concentric Gas flows-Holographic ultraviolet nebuliser Plasma . . . . 16 1 m i r ' Auxiliary . . . . 0.8 1 min-I grating Fused silica torch Nebuliser . . . . 0.55 1 min-l at 36 p.s.i.a. Observation height . . 18 mm Reagents Nitric acid 70% m/ V . Atomic-spectroscopy grade.TricMoroacetic acid 10% m/V. Stock solutions of cations (strontium sodium potassium calcium magnesium and iron). Prepared from the carbonates (E. Merck) by dissolution in a small volume of nitric acid and dilution to the desired volume with doubly distilled and de-ionised water (which was used for preparation of all the more dilute solutions). Stock solution of phosphorus 5 000 p,p.m. Prepared by dilution of analytical-reagent grade phosphoric acid with doubly distilled de-ionised water. Stock solution of chloride 50000 p.p.m. Prepared by dilution of analytical-reagent grade hydrochloric acid with doubly distilled de-ionised water. Methods of Sample Preparation Blood serum analysis dilution procedure For flame atomic-absorption and -emission analysis samples of serum (1-2 ml accurately measured) are diluted 1 + 1 with doubly distilled de-ionised water and nebulised directly into the dinitrogen oxide - acetylene flame or into the air - acetylene flame.The absorbance or the emission is measured at 460.73 nm. normal human serum diluted 1 + 1 with water was used as calibration standards (see Table I1 for operating conditions). As the level of strontium in human serum i 206 SANZ-MEDEL et aZ. CRITICAL COMPARISON OF ATOMIC AnaZyst VoZ. 108 For ICP emission analysis samples of serum (0.1-0.5 ml accurately measured) are diluted 1 + 10 1 + 20 1 + 40 etc. with water depending on their strontium content and nebuliser characteristics (clogging of the concentric nebuliser that was used occurred after very few sample analyses when dilution was less than 1 + 4).After the appropriate dilution the sample is fed into the plasma and its emission measured. Aqueous strontium standards in 0.1% nitric acid are used for calibration when the dilution factor is 1 + 20 (usually employed in our samples) or higher. When such a high dilution is not possible normal human serum diluted in the same way as the samples and whose normal level of strontium has been determined by the standard additions method can be used to compensate for negative nebulisation and transport inter-f erences. As a previous deproteinisation of the serum could eliminate the unwanted presence of high concentrations of organic compounds this method of sample preparation was also tested. A 1.5-ml volume of trichloroacetic acid solution was used per 3 ml of serum and the supernatant liquid after centrifuging was analysed for strontium by atomic-absorption spectroscopy.A loss of sensitivity of 50% was observed when compared with the signal obtained for the same strontium content in serum diluted 1 + 1 (as mentioned previously). Therefore and con-sidering the additional manipulation required deproteinisation was discarded and sample dilution adopted. Soft tissues analysis decomposition procedure The dissected rat organs i.e. brain and liver are stored frozen in clean plastic bags. Samples of 2-4g of the cooled wet tissue are weighed and placed in a digestion flask. (Previous drying of the samples was eventually discarded for our purposes not only due to the risk of analyte losses in the drying but because it was observed that the nitric acid attack was less violent and more reliable if the samples are wet.A small portion of the analysed organ is used separately to determine the total water content by drying it at 100 "C overnight if results referring to dry sample mass are required). A constant amount (5 ml) of concentrated nitric acid is added and the samples are digested on a hot-plate at a temperature of about 100 O C , until only the lipid fraction remains undigested. The solutions are boiled gently almost to dryness and removed from the hot-plate before any charring occurs. A few millilitres of cold water are then added resulting in a clear yellow solution above which oily drops are formed. This undigested lipid fraction solidifies on standing and is filtered off or centrifuged to separate.To remove any analyte that may be trapped the precipitate is thoroughly washed with hot water cooled and re-filtered. The combined solutions are diluted to 25.0 ml with water and aspirated into the plasma using the operating conditions given in Table I (flame methods lacked the necessary sensitivity to analyse the strontium content in these samples). Results and Discussion Optimum Working Conditions The influence of the various instrumental parameters (observation height gas flow etc.) on the flame atomic-absorption and -emission signals of strontium was studied at the resonance line of the neutral atom 460.73 nm with an aqueous 2 p.p.m. solution. Air - acetylene and dinitrogen oxide - acetylene flames were investigated and optimum working conditions sought in both instances.Table I1 summarises the conditions found for the best signal to noise ratio. TABLE I1 OPTIMUM OPERATING CONDITIONS FOR FLAME ATOMIC-ABSORPTION AND -EMISSION SPECTROSCOPY WITH DINITROGEN OXIDE - ACETYLENE AND AIR - ACETYLENE FLAMES Oxidant Acetylene Wavelcng th/ Lamp intensity/ flow-rate/ flow-rate/ Slit width/ Observation height, Method nm Flame m .4 I min-1 1 min-1 n in relative units Atomic absorption . . l(i0.7 Air - C,H 40 13.4 2 0.4 2 460.7 N,O - CZH 20 14.8 7.5 0.4 5 Atomic emission . . 460.7 Air - C,H - 13.4 2 0.14 2 460.7 NZO - CZH - 14.8 7.5 0.14 February 1983 SPECTROMETRIC METHODS FOR SR IN BIOLOGICAL MATERIALS 207 The influence of wavelength] excitation power nebulisation and height above the load coil on the ICP signal Isr and background intensities I,% was studied by aspirating into the plasma a 1 p.p.m.aqueous solution of strontium (only when studying less intense strontium lines were 10 p.p.m. solutions used). Among the various known emission lines of strontium the four most sensitivex6 were initially selected for ICP determinations. The effect of changing radiofrequency power nebuliser argon pressure and observation height above the coil was investigated for each of the four lines by studying the net strontium signal intensity ISr and background intensity] IB1 of each particular line under varying conditions. The signal to background ratio one of the most important factors to be considered in selecting the best of the available analyte wavelengths] can be expressed quantitatively in terms of the background equivalent concentration (BEC) .The BEC is defined as the concentration of analyte that gives an intensity equal to the intensity of the plasma background at the analytical wavelength used and may be used to estimate detection limits.17 We evaluated the BEC equal to (Is/Isr) CRr under all conditions studied concluding that instru-mental parameter changes produce similar effects on the net emission intensity and background intensity of the four spectral lines increasing the radiofrequency power from 0.9 to 1.5 kW gives a continuous increase of signal and background ; increasing nebuliser argon pressure from 26 t o 40 p.s.i.a. leads to a steady decrease in the background while the strontium signal in-creases to a maximum value and then decreases again; when the height is increased from 10 to 20 mm above the coil a similar effect is observed the signal reaches a maximum value while the background intensity decreases (except for the 216.59 and 215.28 nm lines where from 16 mm upward the background and noise start to increase).The optimum operating conditions Le., the minimum value of the REC were worked out in this way and results are given in Table 111. As can be seen the resonance line of strontium(1) at 407.77 nm is the most sensitive one and was then selected (if spectral overlap interferences were observed in real sample analysis we could resort to the other three lines). The best operating plasma conditions for this line are given in Table I.TABLE I11 OPTIMUM OPERATING CONDITIONS AND BEC DETERMINED FOR THE FOUR STRONTIUM LINES STUDIED Instrumental parameters A BEC f Plasma Nebuliser (argon) observation pressure/ Iil; power; Csr used (z car)* Wavelength/nm height/mm p.s.i.a. kW p.p.m. p.p.m. 407.77 18 36 1.15 0.14 0.007 42 1.55 18 38 1.10 0.26 0.0 1 216.59 14 36 1.25 8.80 0.30 215.28 12 32 1.20 3.40 0.40 Sensitivity and Precision In order to compare quantitatively the sensitivity attainable in the strontium determination by the various techniques studied the limits of detection for optimum operating conditions were determined in each instance. The detection limit (DL) was defined here as the strontium concentration required to produce a net line signal twice that of the standard deviation of the background signal : C Y DL = 20 - Isr In the flame atomic-absorption and -emission tests (with both types of flames) a strontium solution with CSr = 0.2 p.p.m.was prepared containing sodium ion (as the nitrate) in a con-centration similar to that found in blood serum (3400 p.p.m. of sodium which will prevent strontium ionisation in the flames). The blank used was a sodium nitrate solution of the same concentration whose atomic absorption (or emission) was registered ten times to find out the background standard deviation C T ~ 208 SANZ-MEDEL et al. CRITICAL COMPARISON OF ATOMIC Analyst Vol. 108 In the ICP tests the limit of detection was evaluated in two ways by the equation DL = 2a;:’*BEC corresponding to the above definition as a;;’. = ag/IB and by the manufacturer’s recommended method.ls A 0.1% nitric acid solution was used here as a blank to evaluate 0;;’.(0.005) and a DL of 0.00007 p.p.m. of strontium was calculated. The manufacturer’s method (twice the standard deviation of the values obtained by measuring a strontium con-centration one fifth of the BEC ten times with a 3-s integration time) gave a DL value of 0.00008 p.p.m. of strontium. Table IV summarises the results obtained and clearly demon-strates that the ICP method provides a sensitivity about 100 times better than the next more sensitive technique (dinitrogen oxide - acetylene flame atomic emission). Flame atomic emission is more sensitive than the atomic absorption in either flame the dinitrogen oxide -acetylene flame being more efficient than air - acetylene in producing neutral atoms of strontium (in a 3400 p.p.m.solution of sodium). The reproducibility of the strontium determinations was evaluated for comparative pur-poses by determining 0.2 p.p.m. of strontium in a 3400 p.p.m. sodium solution by the different methods. As illustrated in the last columns of Table IV at this concentration of strontium (which was the usual one in our samples of serum) the more sensitive the technique the better reproducibility is attained. Moreover with the ICP method ten replicate analyses of a solution containing 0.003 p.p.m. of strontium (unattainable for the flame methods) and 0.1% nitric acid gave a relative standard deviation of l.2y0 and & 1.5% when determining 0.05 p.p.m. of strontium in serum diluted 1 + 20 with water.Ten replicates and a 1-s integration time were used for all tests. TABLE IV COMPARATIVE VALUES FOR SENSITIVITY AND PRECISION OF EACH METHOD Dctcction limit/ Mcthod Flame ng ml-’ Atomic absorption . . Air - C,H 40 NZO - CZH 17 Atomic emission . . Air - C,H 11 N,O - C2H2 6 ICP . . . . . . - 0.08 Standard Dynamic range/ deviation, ng ml-1 p.p.m.* 200-4 000 0.0145 85-4 000 0.004 6 55-2 000 0.003 4 25-1 600 0.002 4 0.4-105 0.0006 Cocfficient of variation %* 7.27 2.41 1.63 1.30 0.30 * Deterniination of 0.2 p.p.m. of strontium in a 3400 p.p.m. sodium solution analyscd tcii timcs. Linear Calibration Range Analytical calibration graphs based on net strontium line signals of gradually diluted strontium standards (in a constant matrix of 3400 p.p.m.of sodium for the flame methods and in a matrix of 0.1% nitric acid solution for ICP) were obtained. As illustrated in the last column of Table I V the results demonstrate that if we consider that a reasonable assumption is that the minimum concentration required for quantitative determination is that corresponding to 5 times the detection limit,17 the effective dynamic range of the flame methods covers about one order of magnitude of concentrations (slightly wider for tlic flame atomic emission) while the ICY method provides an effective dynamic range of from 0.0004 to 100 p.p.m. (more than five orders of magnitude). These results also confirm that self-absorption producing curvature in the upper range is less severe in the ICP than in the flames.Interference Studies Rousselet et aL8 have reviewed the numerous interactions encountered in the determination of strontium by atomic-absorption spectrometry in biological media. After their study of the effect that sodium calcium lanthanum and phosphoric and other acids exert on the strontium atomic-absorption signal it can be concluded t h a t ionisation condensed phase or refractory compound formation are mainly responsible for such interactions. We have systematically studied the effect on the signals of 2 p.p.m. of strontium of increas-ing the concentrations of major inorganic elements found in blood serum i.e. sodium potass-ium magnesium calcium chlorides and phosphates (the influence of up to 100 p.p.m. of iron was also checked and was found to be unimportant in all three methods) February 1983 SPECTROMETRIC METHODS FOR SR IN BIOLOGICAL MATERIALS 209 The results obtained by atomic-absorption spectrometry confirm the findings of Rousselet et aLs alkali and alkaline earth metals enhance the strontium atomic-absorption signal because they retard the analyte ionisation equilibrium in flames.In the air - acetylene flame calcium appeared as the most effective enhancer (a positive 25% enhancement when 2000 p.p.m. of the element were present while sodium or potassium at the same concentration increased the initial strontium signal by only 10-15y0). As expected in the dinitrogen oxide - acetylene flame this ionisation phenomenon becomes so evident that the strontium signal increases by five times over its initial value in the presence of 900 p.p.m.of potassium or 2000 p.p.m. of sodium; above these specified concentrations of alkali metals a plateau is reached. Our results on chloride and phosphate interferences are similar to those described by Rousselet only in the air - acetylene flame. When using the dinitrogen oxide - acetylene flame we observed no interference for 300 p.p.m. of phosphorus and 10000 p.p.m. of chloride (con-centrations well above those expected in blood and soft tissues) using the optimum conditions (Table 11). However the higher noise of the dinitrogen oxide - acetylene flame cannot be considered a serious disadvantage compared with the air - acetylene flame as the signal to noise ratio (see DL values) is twice as good with the dinitrogen oxide - acetylene flame.More-over the precision observed when analysing relatively low levels of strontium i.e. 0.2 p.p.m., considerably favours this last flame as shown in Table 111. In conclusion and against Rousselet et aZ.’s suggestions,* we strongly recommend the use of the dinitrogen oxide -acetylene flame for direct atomic-absorption measurements of strontium in the blood serum of strontium-treated subjects. In these samples owing to their high sodium ion content the ionisation interference is virtually eliminated although for accuracy it is advisable to ensure a fixed sodium concentration in samples and standards of about 3400 p.p.m. Analogous experiments to those described for the flame atomic absorption have been carried out for the atomic emission.All the effects observed on adding alkali or alkaline earth cations are virtually the same as those observed for atomic absorption proving that the interferences are of the same nature the population of neutral strontium atoms is increased by the presence of easily ionisable elements. Increasing amounts of chloride steeply decrease the strontium signal with air as the oxidant. Negative errors the magnitude of which depends on the instrumental conditions are also observed in the presence of phosphoric acid. In a dinitrogen oxide - acetylene flame however neither chloride (up to 10000 p.p.m.) nor phosphate (300 p.p.m. maximum tested) ions interfered under the optimum conditions. I t was checked that nitric acid would be the acid of choice to carry out attacks and extractions on biological material because its presence even at high relative concentrations does not disturb the strontium signals.The influence of the above-mentioned cations and anions on the determination of strontium using the ICP was initially studied with the aid of the “graphics” utility of the ICP/5000 system. Spectra in a range of &- 0.25 nm around the ionic line at 407.77 nm were generated and compared for 2 p.p.m. of strontium in the absence and in the presence of each foreign ele-ment. At the concentrations usually found in blood serum neither the cations nor anions tested produced spectral interferences. Background intensity did not increase at either side of the analyte line and no spectral overlap was detected (a small enhancement observed in the strontium peak intensity on adding 200 p.p.m.of calcium was eventually traced to impurities of strontium in the calcium carbonate used). Further studies were undertaken using the “single” utility of the instrument (Le. determin-ing a single element) to analyse 2.0 p.p.m. of strontium in the presence of increasing concentra-tions of each one of the foreign ions tested. The instrument was calibrated using 1.0 and 2.0 p.p.m. strontium standards prepared in 0.1% nitric acid. Results of these analyses (as percentage recoveries) are given in Table V which show that only very high concentrations of sodium or chloride significantly affect the strontium determination. The observed progressive signal decrease with increasing matrix concentration is attributable to nebulisation and trans-port changes with regard to the standards used.This was verified in two ways (Table V last three columns). (i) If calibration is carried out with strontium standards prepared in an approximately similar matrix (a 3 400 p.p.m. sodium solution) such interferences disappeared. (ii) If the automatic internal standard technique available in the software of the “single” utility is utilised (analysis of 0.1 and 0.5 p.p.m. of strontium by adding 100 p.p.m. of nickel to all solutions and using its emission line at 231.7 nm as an internal standard) interferences from high concentrations of chloride or sodium are also eliminated 210 Element Na . . K c1 P . . Ca Mg . , Fe . , SANZ-MEDEL et al. CRITICAL COMPARISON OF ATOMIC TABLE V Analyst Vol.108 INFLUENCE OF CATIONS AND ANIONS PRESENT IN BLOOD SERUM, ON THE EMISSION OF 2 p.p.m. OF STRONTIUM (Ir = 100) Recovery yo Concentration, p.p.m. 3 400 1 600 360 180 90 10 830 5 775 3610 211 84 56 200 90 50 100 50 25 100 50 30 Compound used NaNO, KNO, HCI CaCO, Internal standard O.lq& HNO 3400 p.p.m. of Na technique 88.5 - 96 96 89 97.5 97.9 -100 99 103 -89 97.0 100 91 97.8 100 92 - 100 102.5 102.7 104.5 102.1 101.5 -99 100.2 100.5 99.8 102 -98 100.0 100 103.4 100.5 -100.5 100 102 100 102 100 Application to Samples The proposed flame atomic-absorption and ICP-AES methods are currently being applied to the determination of strontium in serum of rats whose drinking water was treated with varying amounts of strontium chloride (above 1000 p.p.m.of the element). Atomic absorption works well for this type of analysis and analogous results were obtained by using a quicker calibration graph technique or the more accurate standard additions technique.'* The strontium con-centrations observed in the serum samples were in the range 0.2-1.2 p.p.m. (increasing with the length of treatment of the rat). With such strontium levels a dilution factor of 1 + 20 (0.01-0.06 p.p.m. of strontium) has always been used in ICP-AES determinations and the results obtained agree well within TABLE VI INFLUENCE OF DILUTION OF SERUM ON STRONTIUM RECOVERY Calibration was with 2 p.p.m. of strontium in 0.1% nitric acid. Serum dilution p.p.m.p.p.m. Error % Strontium added Strontium found, 1 + 40 1.00 0.998 0.2 0.20 0.197 1.5 0.06 0.059 1.7 1 + 20 1 + 10 1 + 5 1.00 0.20 0.06 1 .oo 0.20 0.06 1 .oo 0.20 0.06 0.980 2.0 0.194 3.0 0.059 1.7 0.899 10.1 0.184 8.0 0.058 3.3 0.806 19.4 0.170 15.0 0.057 5. February I983 SPECTROMETRIC METHODS FOR SR IN BIOLOGICAL MATERIALS 211 & 5%’ with flame atomic-absorption values for the same samples. Typical nebulisation and transport problems were observed for less dilute serum samples. The organic matrix influence was assessed by studying the effect of decreasing the serum content in the final solution on the recovery of known amounts of strontium. Table VI gives some results of such an effect a consistent negative error whose magnitude decreases with decreasing serum or strontium content becomes apparent.These results are easily explained in terms of the balance of two effects a negative transport interference (which grows with serum concentration) and a positive interference due to the small strontium content of the normal human serum used as a “matrix” (as confirmed through the “graphics” obtained for dilute normal serum). This small strontium concentration (0.05-0.08 p.p.m. in undiluted serum) is not negligible compared with the lowest amount of strontium added in the recovery tests and in these instances it virtually offsets the negative matrix effect. The soft tissues we have analysed for strontium so far include a few samples of brain and liver of strontium-treated rats.Total absence of spectral interferences on the strontium signal in such samples was checked through the “graphics” utility (Fig. I ) and analysis was carried out by the standard additions technique to allow for transport effects in our concentric nebu-liser. The strontium content found in such samples lies in the range 0.3-1.5 pg g-l of wet tissue in brain and 0.15-0.6 pg g-1 in liver. We are currently experimenting with other digestion procedures (different from that given in the experimental section) because of the presence of relatively high concentrations of undigested organic matter that burns in the vicinity of the induction coil and clogs the central orifice of the plasma torch after about six sample analyses. c 0 In .-.-E Lu I -0.5 nm-Wavelengthhm Fig.1. ICP/5000 graphic spectral display for 1, strontium aqueous standard of 0.06 p.p.m. ; 2 treated rat brain solution; 3 treated rat liver solution; and 4 treated rat serum (diluted 1 + 20). It is worth noting that it was not necessary to resort to background correction in any analysis. The strontium spectra provided by the “graphics” utility did not show any increase in the background level on the addition of inorganic interferents tested or when real samples were analy sed for strontium. Ma Antonia Palacios’s assistance with the dissolution methods for soft tissues and Dr. Munoz Tejedor’s medical advice on strontium metabolism and provision of samples are gratefully acknowledged. References 1. 2. Catsch il. in Lenihan T. AI. Editov “Strontium Metabolism,” Academic Press Sew York London, Iishirsagar S.G. J. Nzrtr. 1976 106 1475. 196G p. 260 212 3. 4. 5. 6. 7. 8. 9. 10. 1 1 . 12. 13. 14. 15. 16. 17. 18. SANZ-MEDEL RODRIGUEZ ROZA AND PEREZ-CONDE Mufioz Tejedor D. Rev. Fisiol. Univ. Navurra submitted for publication. Annu. Rep. Anal. A t . Spectrosc. 1974-1980 4-10. Harrison G. E. Nature (London) 1958 182 792. Webb M. S. W. and Wordingham M. L. Anal. Chim. A d a 1963 28 450. David D. J. Analyst 1962 87 576. Rousselet F. El Solh N. and Girard M. L. Analusis 1975 3 44. Beck F. Janovskova J. and Moldan B. A t . Absorpt. Newsl. 1974 13 47. Helsby C. A. Talanta 1977 24 46. Mermet J. M. and Hubert J. Prog. Anal. A t . Spectrosc. 1982 5 1. Dahlquist R. L. and Knoll J . W. Appl. Spectrosc. 1978 32 1. McQuaker N. Ti. Kluckner P. D. and Chang G. N. Anal. Chem. 1979 51 888. Kluckner P. D. and Brown D. F. “lnternational Winter Conference San Juan Puerto Kico 1980,” Pehlivanian E. Mermet J. M. and Robin J. J. Bzophys. Med. Nucl. 1980 4 247 (abstract ICP Boumans P. W. J . M. “Line Coincidence Tables for ICP Atomic Emission Spectrometry,” Pergamon Boumans P. W. J . M. Opt. Pura Apl. 1978 11 143. Palacios Corvillo M. A. Sanz-Medel A. and Gallego R. Quim. Anal. in the press. (abstract ICP Inj. Newsl. 1981 7 89). Inf. Newsl. 1981 7 90). Press Oxford 1980. Received July 14th 1982 Accepted August 16th 198

ISSN:0003-2654    

DOI:10.1039/AN9830800204

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst February 1983 Vol. 108 pp. 213-224 Investigations on Atomisation Mechanisms of 213 Volatile Hydride-forming Elements in a Heated Quartz Cell Part 1. Atomisation of Arsine” Gas-phase and Surface Effects; Decomposition and Bernhard Welz and Marianne Melcher Department of Applied Research Bodenseewerk Perkin-Elmer 6. Co. GmbH 0-7170 i’berlingen Federal Republic of Germany The atomisation of gaseous hydrides in a heated quartz cell is not caused by a thermal decomposition but by collision with free hydrogen radicals. These radicals are formed in a reaction with oxygen at temperatures above 600 “C. In a “clean” environment the concentration of radicals is well above the equilibrium concentration because their formation is a much faster process than their recombination.Several materials however can catalyse radical recombination and therefore have a depressing effect on the observed signal. In the absence of hydrogen arsine is not atomised but thermally decomposed, probably with the formation of As and As,. Keywords A tomic-absorption spectrometry ; hydride generation technique ; atomisation mechanisms ; arsine atomisation and decomposition ; hydrogen radicals The question of the atomisation mechanism for gaseous covalent hydrides such as arsine in heated quartz tubes as frequently used in the hydride generation atomic-absorption spectro-metric technique is not discussed in most publications. The general opinion however, appears to be that the hydrides are atomised by thermal decomposition. Thompson and Thoresby,l for example refer to an “electrothermal atomisation” of arsenic in a heated quartz tube and Verlinden et a1.2 used an electrically heated silica tube for the “thermal decomposition of arsine in an argon atmosphere.” There are however some inconsistencies that cast doubt on this mechanism one of which is the temperature itself.Whereas in heated quartz tubes typically temperatures around 800 “C are found to be optimum for the atomisation of arsenic3-5 and other volatile hydride-forming elements temperatures of 1 700-1 800 “C are necessary to atomise arsine6 or selenium hydride’ in a graphite tube furnace. The second fact is that addition of oxygen or air to the purge gas increases the sensitivity,s-ll and the highest sensitivities are found for extremely fuel-rich hydrogen - oxygen (or hydrogen - air) flames burning inside unheated quartz tube a t o m i s e r ~ l ~ - ~ ~ with an optimum hydrogen to oxygen ratio of 5 l.8911114 The third fact is the pronounced influence of the quartz cell surface on the signal obtained for the hydride-forming elements.15-17 Evans et a1.16 suspected the “need for a catalytic film of the element on the silica surface before consistent response is achieved,” and they refer to the “fragile nature of this catalytic film.” Dgdina and RubeSka14 studied the atomisation of selenium hydride in a cool hydrogen -oxygen flame burning in an unheated quartz tube.They showed that hydride atomisation is caused by free radicals generated in the flame of their system. Rased on interference studies we have concluded earlier18 that the same mechanism of atomisation should take place in an electrically heated quartz tube atomiser for arsenic and selenium.In this work we carried out several investigations with different purge gas mixtures in a commercially available hydride system with an electrically heated quartz cell atomiser, as well as with a modified system that allows the introduction of small amounts of gases through a septum. With this system we studied the atomisation of pure arsine in various * Parts of this material were also presented a t the 8th FACSS Meeting Philadelphia P A 1981 and at the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy Atlantic City N J 1982 214 gas atmospheres. decomposition. observed with volatile hydride-forming elements in heated quartz cells.WELZ AND MELCHER INVESTIGATIONS ON ATOMISATION OF Analyst VoZ. 108 The results indicate clearly that the atomisation of arsine is not a thermal Mechanisms are proposed that satisfactorily explain the various phenomena Experimental Apparatus A Perkin-Elmer Model 4000 atomic-absorption spectroyhotometer equipped with an external electrodeless discharge lamp power supply and a Model 56 recorder was used for all investigations. Perkin-Elmer electrodeless discharge lamps were used for all elements and the operating parameters and the instrument settings were according to the manufacturer’s recommendations. A Perkin-Elmer Model MHS-20 mercury - hydride system was used for all experiments except for those with pure arsine. For the latter a series of experimental set-ups were used, as shown schematically in Fig.1. Commercially available parts of the MHS-20 were used where possible to keep the results as comparable as possible. This was specifically true for the heated quartz cell and the reaction flask. The gas flows of the experimental set-ups were controlled by rotameters that were calibrated for argon and hydrogen via soap-bubble flow meters. The MHS-20 and its temperature control system have been described in detail elsewhere. l8 Gases and Reagents 4 v.p.m. of oxygen 10 v.p.m. of nitrogen and 4 v.p.rn. of water. Argon. Argon (kinde) was the standard welding quality 99.996% pure containing 900 “C -hv A NaBH4-Gas -. C Gas-Gas ~. Quartz tube F! Fig. 1. Experimental set-ups to study the atomis-ation of arsine gas.A Arsine gas was injected between the reaction flask and the heated quartz cell. In some expcriments a “blank” reaction was carried out simultaneously in the reaction flask. €3 Two heated quartz cells in a tandem arrangement. Arsine was injected into the gas stream to the first cell and in some experiments hydrogen was added between the first and the second cell. Atomisation signal was only measured in the second cell. C A quartz tube heated partially with a Bunsen burner, was installed in the gas line to the heated quartz cell. In some experiments hydrogen was added between the quartz tube and the heated quartz cell. “Gas” means Ar Ar + 1% 0 or Ar - H mixtures, depending on experimental conditions February 1983 VOLATILE HYDRIDE-FORMING ELEMENTS.PART I 215 This was obtained in cylinders (Messer-Griesheim) and for our experiments the mixture was filled into smaller gas containers under atmospheric pressure and aliquots were taken from these containers with a syringe. From the relative molecular mass of arsine (77.95) it can be calculated that 1 p1 of the arsine - argon mixture contains approximately 17 ng of arsenic. Sodium tetrahydroborate(ll1). A 3% m/V solution was prepared from sodium tetra-hydroborate(II1) powder (Riedel-de Haen) by dissolving it in de-ionised water and stabilising it with lo/ m/V sodium hydroxide. The solution was always filtered before use and could be stored for only a few days. These were prepared by diluting Titrisol concentrates (Merck) according to the manufacturer's recommendations.Working standards were prepared by further dilution with an appropriate acid and dilute standards were prepared fresh daily. Argon containing 1% of oxygen. Arsine 0.5% in argon containing less than 2 v.p.m. of hydrogen. Argon S1 (Linde) was obtained in cylinders. Stock standard solutions of hydride-forming elements. Hydrojuoric acid 40% m/V. Analytical-reagent grade (Merck). Results Influence of Purge Time Using an MHS-20 hydride system we found that the purge time with an inert gas to remove the air prior to the addition of the sodium tetrahydroborate(II1) solution had a significant influence on the sensitivity of all hydride-forming elements at least at lower quartz cell temperatures (Fig. 2). This phenomenon was also reported by Meyer et al.15 for selenium in a different system.100 .-40 .- I 4-0 L C) p I I I I I I I I I :d) v I I I I r5 - 700 1000 700 1000 700 1000 700 1000 700 1000 700 1000 Quartz-cell temperature/"(= Fig. 2. Influence of thc purge time with argon on the relative sensitivity of A 25-s purge time; (a) Sn; ( h ) -1s; ( c ) Sb; hydride-forming elements in a heatcd quartz cell (MHS-20). 0 60-s purge time; 25- or 80-s purge time Ar + l o 02. ( d ) Bi; ( P ) Se; and (f) Te. At 700 O C and with a 60-s purge time no signals were obtained for antimony arsenic, selenium and tin ancl only 50% or less of the maximum sensitivity was found for bismuth and tellurium. Even at a quartz tube temperature of 1000 O C optimum sensitivity could not be obtained for arsenic selenium tellurium ancl tin with a 60-s purge time.Vsing a shorter purge time of 25 s prior to the addition of the sodium tetrahydroborate(II1) solution, maximum sensitivity was obtained for all six hydride-forming elements at least at tempera-tures above 800 "C. In the hydride system used an inert gas (nitrogen or argon) bubbles through the sample solution during the purge stage. IVe therefore suspected that the inert gas stream drives the dissolved air out of the sample solution and that the air has some influence on the atoniisa-tion. To confirm this assumption we carried out additional experiments to obtain sample solutions that are low in dissolved air and measured them in the liydride system in the usua 216 Analyst VoZ.108 way. In one set of experiments nitrogen or argon was bubbled through the solutions for 5 min outside the hydride system prior to the determination. In a second experiment the solutions were heated nearly to boiling for 15 min before they were connected to the hydride system. In all experiments we obtained sensitivities very close to those observed for a 60-s purge time in Fig. 2 but now independent of the purge time used. This confirms that the air dissolved in the solution for measurement seems to play an important role in the atomisa-tion of the hydride-forming elements at least at lower quartz cell temperatures and that the air is removed from the solution during a 60-s purge time. To investigate further the influence of oxygen on the sensitivity of the volatile hydride-forming elements and to obtain better control over the amount of oxygen present in the system during atomisation we replaced the purified nitrogen or argon with a special argon gas containing 1% of oxygen as purge gas.We studied the sensitivity for arsenic and tin, the two elements for which the purge time had the strongest influence. As can be seen in Fig. 2 the full sensitivity was obtained for both elements at 700 "C and there was no influence of temperature on the signal as long as it was above 600 "C. Below that temperature no signal could be obtained even in the presence of oxygen. This effect of oxygen had already been observed in 1974 by Goulden and Brooksbank,* who found that an increase in sensitivity particularly for arsenic and antimony could be obtained if oxygen was added to the gas stream.They also found that the tube must be heated to a minimum temperature of about 700 "C in order to obtain a good response whereas above this temperature variations in temperature had no marked effect. Pierce et al.1° used an argon flow-rate of 200 ml min-l and added 1.2 ml min-l of air through the manifold of their automated system. They also found that for the determination of selenium temperature is not critical between 700 and 900 "C. Vijan and Woodg used a similar ratio of 400 ml min-l of argon and 3.9 ml min-l of air but a cell temperature of only 570 "C and obtained a factor of two increase in sensitivity for selenium owing to the addition of air. When air or oxygen is added to the purge gas a small pale flame can be observed burning inside the quartz tube at both sides of the gas inlet.This was also reported by Vijan and Wood9 who in addition found that an increase in the length of the quartz tube had no or even an inverse effect on the sensitivity of selenium. They speculated that the effective path length is determined by the burning hydrogen slug inside the cell rather than by the length of the cell itself. All of these observations suggest that a heated quartz cell in the presence of a small amount of air or oxygen behaves in the same way as a fuel-rich hydrogen - oxygen flame burning in an unheated quartz tube. WELZ AND MELCHER INVESTIGATIONS ON ATOMISATION OF Experiments with Pure Arsine The previous experiments suggest that oxygen has an effect on the atomisation of hydride-forming elements at lower quartz cell temperatures.At higher temperatures however, thermal dissociation of the hydrides cannot be excluded as an alternative atomisation mechanism. To investigate this further we performed a series of experiments with a special gas mixture of 0.5% of arsine in argon. For these experiments a modified hydride system was used (Fig. 1A) that allowed the injection of small gas volumes through a septum with a syringe. The heated quartz cell however was unchanged to make the results more easily comparable. The system also allowed the generation of arsine from a solution in the con-ventional way without modification for comparative measurements. To check the sensitivity of the system arsine was generated in the conventional manner in the reaction vessel from an acidic solution by addition of sodium tetrahydroborate(II1) solution and signals of about 0.35 absorbance unit were obtained for 50 ng of arsenic (Fig.3A). With an empty reaction vessel and injecting 3 pl of the arsine - argon mixture (corre-sponding to about 50 ng of arsenic) through the septum into the argon flow to the heated quartz cell no measurable signal was obtained (Fig. 3B). In the next set of experiments a blank reaction with only 1.5% m/V hydrochloric acid and sodium tetrahydroborate(II1) solution and no arsenic was carried out in the reaction vessel. Injection of 3 p1 of the arsine - argon mixture through the septum into the tubing between the reaction vessel and the heated quartz cell while the blank reaction was going on resulted .As expected only a minor blank signal was obtained (Fig.3C) February 1983 VOLATILE HYDRIDE-FORMING ELEMENTS. PART I 317 0.7 0.6 0.5 2 0.4 $ 0.3 al m e 2 0.2 0.1 0 A Fig. 3. Atomisation signals for arsenic under different experimental conditions. A 50 ng of As evolved from an acidic solution; B 50 ng of As injected as ASH (3 pl) purge gas Ar; C, blank signal from 1.50; rn/V HC1 with sodium tetrahydroborate(II1) ; D 50 ng of As injected as ASH (3 p1) while a blank reaction was going on; and E 5 ml of arsine (approximately 85000ng of As) injected purge gas AT no blank reaction. in a sharp narrow signal of about 0.7 absorbance unit (Fig. 3D). The peak-height repro-ducibility of these signals was not very good and depended on the speed of injection and other experimental parameters.The peak area however agreed satisfactorily with that obtained for aqueous standards from solution (Fig. 3A). This indicates the same degree of atomisation for both arsenic injected as arsine gas and liberated from an acidic solution by reaction with sodium tetrahydroborate( 111). I t is necessary however that a “blank reaction” is carried out simultaneously with the arsine injection. In a final experiment a much larger volume (5 ml) of the arsine - argon mixture was injected through the septum into the argon flow to the heated quartz cell with no blank reaction going on in the reaction vessel. This time a small signal of 0.06 absorbance unit was obtained (Fig. 3E) which shows that a small percentage of the injected arsine is atomised at 900 “C in a pure argon atmosphere.The sensitivity and therefore the atomisation efficiency however are approximately four orders of magnitude lower than in experiments with a simultaneous blank reaction or evolution from acidic solution. From these experiments the question arises of which component is formed during the reaction between hydrochloric acid and sodium tetrahydroborate(II1) that initiates or supports the atomisation of the arsine in the heated quartz cell. It could be shown that neither water vapour nor hydrochloric acid or sodium tetrahydroborate( 111) fumes had any effect on the atomisation of arsine. Hence there is only the hydrogen left that could support the atomisation. The experiments with the arsine - argon mixture were therefore repeated with a mixture of hydrogen and argon as carrier gas instead of pure argon and “normal” atomisation signals were obtained immediately even without a blank reaction going on in the reaction vessel.The signal height for arsenic was found to depend both on the total gas flow through the heated quartz cell and on the hydrogen to argon ratio. Without optimising these two para-meters too carefully the best sensitivity and reproducibility were found for a flow-rate of 100ml min-l of hydrogen and 200mlmin-l of argon. The signals obtained under these conditions for different volumes of the arsine - argon mixture injected into the carrier gas stream are shown in Fig. 4. They exhibit reasonable reproducibility and linearity and a sensitivity that is at least as good as that found for acidic standard solutions.This experi-ment demonstrates that hydrogen is an essential parameter for the atomisation of arsine in a heated quartz cell 218 Analyst Vol. 108 From the earlier experiments with the purge time we know that oxygen also has an influence on the sensitivity of arsenic at least at lower quartz cell temperatures. We there-fore repeated several of the experiments with the arsine - argon mixture. Firstly argon containing 1% of oxygen was used as a purge gas with no hydrogen added and no measurable signal was obtained as with pure argon. This indicates that oxygen alone cannot support the atomisation of arsine in a heated quartz cell. Hydrogen was therefore added to the carrier gas for all further experiments.The quartz cell temperature was varied between 650 and 1000 "C and a mixture of 100 ml min-l of hydrogen and 200 ml min-l of argon with and without 1% of oxygen was used as the purge gas. Aliquots of the arsine - argon mixture were injected into the carrier gas as described earlier. When a pure argon and hydrogen mixture was used as the carrier gas a dependence of the signal from the quartz cell temperature was found that was identical, within experimental error with the curve in Fig. 2 for a 60-s purge time. When hydrogen and argon containing 1% of oxygen were used as the purge gas mixture however maximum sensitivity was obtained between 650 and 1000 "C. This demonstrates once again that hydrogen is essential for the atomisation of arsine but that oxygen is also required at least at lower quartz cell temperatures.The only remaining question is what happens to the arsine in a quartz tube heated at 900 "C if it is not atomised (or atomised only to a negligible percentage). To investigate this further we used two identical quartz cells in a tandem arrangement as shown in Fig. lB, where the gases coming out of the first cell were transferred into the second. Only the second cell was mounted in an atomic-absorption spectrometer to measure the absorbance signals. There was also a provision made to intro-duce an additional gas via a Y-connection between cells one and two. To check the sensitivity of the arrangement 3 p1 of the arsine - argon mixture were injected into the carrier gas stream of 100 ml min-l of hydrogen and 200 ml min-l of argon while the first cell was not heated.An average signal of about 0.6 absorbance unit was obtained which reflects the "normal" sensitivity of the system (Fig. 5A). In a second run, the experiment was repeated with the first quartz cell heated to 900°C as well and the signal observed in the second cell decreased to an average value of 0.04 absorbance unit, which is about 7% of the previous sensitivity with the unheated cell (Fig. 5B). Increasing WELZ AND MELCHER INVESTIGATIONS ON ATOMISATION OF This cell was always heated to 900 "C. 0.6 0.5 8 0.4 0.3 I= rn en 9 Q 0.2 0.1 0 yr c Time -+ Fig. 4. Signals for A 0.5 B 1.0 and C 1.5 p1 of arsine corresponding to approximately 8 17 and 25ng of As injected into a stream of 100 ml min-l of H + 200 ml min-' of Ar.Cell temperature 900 "C. 0.6 0.5 8 0.4 $ 0.3 0.2 0.1 0 0 9 2 A C D hhh. Time __+ Fig. 5. Atomisation signals for arsenic in the second quartz cell of a tandem arrangement. A 3 p1 of ASH (50 ng of As) first quartz cell not heated; B 3 pl of ASH (50ng of As) first quartz cell heated to 900 "C; C 10 p1 of ASH (170 ng of ,4s) first quartz cell heated to 900 "C. A B and C carrier gas 100 ml min-1 of H j- 200 ml min-l of Ar. D Same as C, carrier gas 200mlmin-l of Ar; 100 ml min-l of H added after the first quartz cell February 1983 VOLATILE HYDRIDE-FORMING ELEMENTS. PART I 219 the injected volume of the arsine - argon mixture from 3 to 10 p1 increased the observed signal proportionally (Fig.5C). Finally we used only an argon flow-rate of 200 ml min-l as the carrier gas through the first cell and added the 100 ml min-l of hydrogen to the carrier gas between the two quartz cells. This resulted in a further reduction of the signal for 10 pl of arsine - argon mixture to an average of 0.05 absorbance unit which is about one third of the sensitivity obtained when the hydrogen also passed through the first quartz cell (Fig. 5D). This indicates that a small percentage of the arsenic injected into the carrier gas as arsine passes through tlie heated quartz cell in a form that can be measured by atomic absorption in a second heated quartz cell in the presence of hydrogen. This percentage however is about three times higher if hydrogen is added to the carrier gas before it enters the first heated quartz cell than if hydrogen is added after it has left the first cell.This shows clearly that the arsine certainly does not pass through the heated quartz cell without being de-composed. If we do not see an atomisation signal for arsenic when arsine is injected in pure argon carrier gas and passes through a heated quartz cell this means that arsine is decomposed, but not atomised to any significant extent. The experiments were repeated with niucli larger volumes of arsine in a modified arrange-ment where tlie arsine was passed through a single heated quartz cell and the outcoming gases were collected in a gas reservoir. The collected gases were then swept back through the heated quartz cell if necessary with addition of hydrogen.In this experiment we obtained the same result vix. a three to four times higher signal is obtained when the arsine was first decomposed (atoniised) in the presence of hydrogen than in pure argon. In a further experiment we looked for condensed arsenic or arsenic compounds in the quartz cell after a large volume of arsine was decomyosed in pure argon at 900 "C. On dismantling the cell and washing it with nitric acid we found that approximately 2./ of the inserted arsine had condensed in some form on the walls of the cell itself and about Sq at the quartz windows. Finally we used an arrangement in which a quartz tube was inserted in the tubing between the septum for arsine injection and the heated quartz cell (Fig. 1C). Heating the quartz tube with a Hunsen burner resulted in approximately the same decrease in sensitivity as observed with the two quartz cells in sequence in the previous esperinient.After a large volume of arsine had been passed through the quartz tube heated at one end using the hydrogen - argon mixture as purge gas the cool end was heated with a Bunsen burner and a fairly broad peak was observed in the heated quartz cell. This means that part of the arsenic condenses in sonie form at the cooler end of the heated quartz tube. By heating it in an atmosphere of argon and hydrogen it can not only be volatilised but also to some extent atomised in the heated quartz cell. Repeating the same esperinient in pure argon as carrier gas and adding the hj-drogen onlj-between the quartz tube and the heated quartz cell however gave an almost identical result.This means that tlie condensed arsenic species can also be re-volatilisecl in a pure argon atmosphere and that the hydrogen is needed only for the atomisation. Influence of the Quartz Cell Surface The influence of the surface of the heated quartz cell on the sensitivity observed for volatile liydride-forming elenients is well documented. Ileyer ct d . l s reported a signal suppression for selenium that they attributed to increasing devitrification of thc quartz due to trace amounts of sodium hjdroxide in the carrier gas and to trace aniounts of burnt-in metals. They also inserted a piece of platinuni foil in the quartz cell which caused t1:e same kind of signal suppression. Evans ct (71.16 reported for the determination of antimony, arsenic and tin that a separate quartz cuvette must be kept specificallj- for each element and must be pre-conditioned for tliat elenient by repeated applications of the most concentrated working standard solution until a constant response is obtained.\Ye have also seen from time to time that a new untreated quartz cell gives a low response for volatile hydride-forming elements (Fig. &I). Heating the cell for 24 11 at 1 000 "C usually could solve this probleni independent of the elenient to be determined. 1j.e also investigated several acid treatments for the quartz cell and found that rinsing in 40:; hydro-fluoric acid for approximately 15 min is tlie most effective procedure that brought the optimum sensitivity inimcdiatelj.for each hydride-forming element (Fig. tiB) . The hydro 220 Analyst VoZ. 108 fluoric acid cleaning procedure was usually also successful for longer used quartz cells that were contaminated by sample components and therefore showed low sensitivities. WELZ AND MELCHER INVESTIGATIONS ON ATOMISATION OF 0.3 B Time -b Fig. 6. Signal recorded for 50ng of As. A New untreated quartz cell; B quartz cell washed with 40% m/V HF. Cell temperature 900 OC. To investigate the surface problem further a quartz cell was sand-blasted (outside only) and then used for an arsenic determination without cleaning and no signal was obtained. Placing a cell in a sand-blasting chamber so that it could collect the dust but without being sand-blasted resulted in the same total signal suppression.Washing the quartz cell in hydrofluoric acid or only in dilute nitric acid restored the original sensitivity immediately, which suggests that the “dust” is the inhibiting species. Inserting a small untreated quartz tube into the heated quartz cuvette conditioned with hydrofluoric acid resulted in a 75% decrease in the arsenic signal. Inserting the same quartz tube after it was treated with hydrofluoric acid had no influence and gave the full sensitivity for arsenic. In both instances the quartz tube was left in the hot environment long enough for it to reach the same temperature as the cuvette itself. A graphite tube placed in the heated quartz cell caused the same extent of signal suppression as the untreated quartz tube. More detailed experiments were carried out with small graphite rods or plates placed in different parts of the heated quartz cell which showed that the influence depends strongly on the location of the graphite pieces.No influence was found for graphite placed in the gas inlet or exactly in the middle of the heated quartz cell. Almost no influence was also found for graphite placed at the end of the heated quartz cell close to the windows. A strong signal suppression however was observed when the graphite parts were placed about half way between the gas inlet and the end windows or if a longer graphite rod that protruded more towards the ends was placed in the middle of the heated quartz cell. In addition there is an influence of the type of graphite on the degree of signal suppression.Solid pyrolytic graphite which has a very dense surface shows only a minor signal suppression for arsenic whereas ordinary isotropic graphite has a much more pronounced influence on the sensitivity. I t should be mentioned that all of these effects are reversible which means that the original sensitivity is restored immediately when the graphite piece is removed from the quartz cell. Different metal wires placed in the heated quartz cell also caused signal suppressions that were dependent on the metal used. The suppression was 70% for platinum 50% for iron and 25% for copper. These effects were not completely reversible because frequently some metal was burnt into the quartz surface. Measurement of Effective Temperature A possible explanation for the role of oxygen that enhances the sensitivity for all hydride-forming elements at lower quartz cell temperatures (Fig.2) would be that it reacts with the hdyrogen and thereby increases the effective temperature in the gas phase of the quartz cuvette. In fact a pale flame can be seen inside the heated quartz cell under certain con-dit ions. L’vov et nZ.19 have proposed the two-line method for measuring the effective temperature in an atomic vapour by comparing the absorbance of a ground-state line with that of a lin February 1983 VOLATILE HYDRIDE-FORMING ELEMENTS. PART I 22 1 that terminates on a slightly elevated energy level. Among the volatile hydride-forming elements tin with its lines at 286.3 and 284.0 nm is almost ideally suited for this purpose. For this pair of lines the effective temperature Teff (K) can be calculated according to 2143 1% (4.9AIl4) Terr = where A is the absorbance of the ground-state line at 286.3 nm and A the absorbance of the metastable line at 284.0 nm.Table I shows the results of these gas-phase temperature measurements for tin at quartz cell temperatures between 700 and 1000 "C with pure argon and argon containing 1% of oxygen as the purge gas and purge times of 20 and 60s prior to the addition of sodium tetrahydroborate(II1). Using pure argon as the purge gas and a 60-s purge time no absorbance could be measured for the 284.0-nm line at 700 and 800 "C and therefore no temperature could be calculated. For a given quartz cell temperature the effective gas-phase temperatures agree with each other within less than &20 "C for the different experi-mental conditions with and without oxygen.The only exception is the surprisingly low value of 639 "C for argon containing 1% of oxygen and a 60-s purge time at a cell temperature of 700 "C. At low quartz cell temperatures the effective temperatures are in good agreement with the set temperatures and they tend to be lower at higher quartz cell temperatures with a maxi-mum deviation of about 100 "C at 1000 "C (see Fig. 7). The effective temperatures in the presence of oxygen however are in no instance significantly higher than those measured in the absence of oxygen. The enhancing effect of oxygen on the sensitivity of tin and the other volatile hydride-forming elements can therefore not be due to a temperature increase in the gas phase of the atomiser cell.In Fig. 7 the relative sensitivities obtained for tin under different experimental conditions are plotted against the effective temperature. The pattern of this graph is almost identical with that in Fig. 2 which suggests that oxygen supports the atomisation according to some other mechanism. Discussion From the above it is apparent that (a) hydrogen is essential and plays an active role in the atomisation of gaseous hydride-forming elements in a quartz cell heated at 1000 "C; (b) oxygen supports the atomisation in the presence of hydrogen at least at lower quartz cell temperatures; (c) in the presence of oxygen the effective temperature of the vapour phase in the heated quartz cell is not higher than in pure argon; ( d ) maximum sensitivity can be obtained only in the "clean" environment of a quartz cell conditioned with hydrofluoric acid, etc.; placing various materials in the quartz cell may cause dramatic signal suppressions; and (e) in the absence of hydrogen arsine is decomposed in a heated quartz cell but is not atomised; the arsenic species formed in that thermal decomposition are in part volatile and/or can be re-volatilised and in the presence of hydrogen atomised. From these results we conclude that the atomisation of volatile hydride-forming elements in a heated quartz cell must be due to collisions with free H radicals according to This low value was obtained repeatedly and could well be the correct one. MH. + H + MH,- + H, MH + H + M + H 2 TABLE I CALCULATED EFFECTIVE TEMPERATURE OF THE GAS PHASE UNDER DIFFERENT EXPERIMENTAL CONDITIONS FOR TIN Calculated effective temperature of the gas phase/"(= Quartz cell l'urge gas Ar Purge gas Ar + 1% O2 A temperature/ - I A 1 "C 20-s purge time 60-s purge time 20-s purge time 60-s purge time 700 721 N.d.* 698 639 800 740 N.d.* 731 747 '300 843 826 852 855 1000 908 873 908 908 * N.d.= not determined 222 WELZ AND MELCHER INVESTIGATIONS ON ATONISATION OF Analyst Vol. 108 700 800 900 1 000 Effective tern peratu re/”C Fig. 7. Relative sensitivity for tin under differ-ent experimental conditions plotted against the effective temperature of the vapour phase. a, Measured effective temperature; the solid line reflects the most probable linear relation between the quartz cell temperature and the effective vapour-phase temperature.0 Sensitivity for Sn 60-s purge time with Ar. A 20-s purge time with Ar. 0 60-s and 20-s purge time with Ar + 1% 0,. The H radicals are probably formed by the same reactions with oxygen as proposed by DCdina and RubeBka14 for the fuel-rich hydrogen - oxygen flame: H + O + O H + O 0 + H + OH + H OH + H + H,O + H This easily explains the role of oxygen and the effect of the purge time on the sensitivity of virtually all hydride-forming elements. The strong influence of temperature on the sensi-tivity could be explained by a different mechanism for radical formation. Quartz has a strong catalytic effect at temperatures around 1000 “C and H radicals can be formed by decomposition of hydrogen molecules at the quartz surface.Unsaturated oxygen atoms at the surface bind hydrogen molecules and an H atom is torn away. The free enthalpy of this reaction is certainly much lower than that of a direct fission of the hydrogen molecule. There must be an equilibrium between hydrogen in the gas phase and the hydroxyl groups at the quartz surface. In the presence of hydrogen only saturation of the surface should occur with the formation of H radicals in the gas phase which should decrease asymptotically to the equilibrium concentration. At low quartz tube tempera-tures e.g. 700 “C this equilibrium concentration of H radicals seems not to be high enough for nearly quantative atomisation whereas it does appear to be high enough at 1000 “C. Such an active participation of the quartz cell surface in the generation of H radicals could also explain the severe depressing effect on the signal observed for the hydride-forming elements in the presence of several surface contaminants.Another possible explanation for the influence of the temperature on the sensitivity how-ever is that even in “pure” argon the H radicals are formed by reaction with oxygen. The amount of oxygen that is always present in argon of the quality used for these purposes should be large enough to produce a sufficient amount of H radicals. Under these “low oxygen” conditions however temperature should have a significant influence on the rate of radical production and hence on the amount produced February 1983 VOLATILE HYDRIDE-FORMING ELEMENTS. PART I 223 Because the probability of free atom formation from the gaseous hydrides is proportional to the number of collisions with free radicals the atomisation efficiency should increase with increasing number of radicals.In the heated quartz cell the number of radicals should increase with increasing temperature as well as with increasing oxygen concentration (up to a certain optimum). This behaviour is clearly reflected in the sensitivity curves in Fig. 2. The decreasing dependence of the sensitivity on temperature and oxygen content for the heavier elements within the same group (arsenic antimony and bismuth and selenium and telluriumj can be explained by a higher collision efficiency for the larger and less stable hydrides. Because radical recombination is a considerably slower process than their formation a concentration well above equilibrium values can be expected in the heated quartz cell.This is one of the reasons why relatively low oxygen concentrations can produce a sufficient radical density for the atomisation of the hydrides. A number of materials however strongly catalyse the recombination of H radicals. Among these are a number of metals and for the metals investigated in our experiments the catalytic effect decreases in the order platinum > iron > copper. This however is exactly the order in which the depressing effect of metal wires placed in the heated quartz cell decreased. I t can therefore be assumed that the interference of these metal wires in the quartz cell is a catalytic effect on the radical re-combination.The reduced lifetime and hence reduced concentration of H radicals then causes the decrease in sensitivity. The same mechanism can also be assumed for the graphite parts inserted in the heated quartz cell. It is known that gaseous liydrides can be trapped quantitatively in charcoal so that a direct reaction of the hydride with the graphite cannot be excluded a priori. The experiment in which graphite rods were placed on different parts of the quartz cell including the gas inlet shows that this mechanism is not very likely. The fact that the depressing effect of graphite is highest when the rods are placed some distance from the gas inlet suggests that this is the area where most H radicals are formed and that the signal suppression is again a catalysis of radical recombination.The effect of unconditioned or sand-blasted quartz cells and of the unconditioned tube inserted into a conditioned cell can all be explained by the same effect. All the “dust” or “dirt” forms a “catalytic film” on the quartz tube surface that speeds up radical recombina-tion and therefore reduces the concentration of H radicals in the atomiser cell. A similar catalytic film is probably produced by most other contaminants that reach the heated quartz cell and are deposited on or burnt into the surface. Only a quartz suriace that has been well conditioned c.g. by treatment with hydrofluoric acid is sufficiently free of active spots to ensure the well above equilibrium concentration of H radicals that is required for the atomisation of the gaseous h ydrides.In the absence of hydrogen arsine is not atomised in a heated quartz cell or to be more precise the atomisation efficiency is reduced by about four orders of magnitude. Part of the arsirie passes through the hot quartz cell in a form that can be atomised later in a second heated quartz cell in the presence of hydrogen and part of the arsenic is condensed and deposited at cooler parts in some form. The condensed arsenic species can be easily re-volatilised by heating and also be atomised in the presence of hydrogen in a heated quartz cell. As arsinc decomposes at temperatures above 300 “C it is very unlikely that it can pass a quartz cell heated at 900 “C without being dissociated. Probably arsine is decomposed but not atoniised and the species that are most likely to be formed are the tetramer and dimer whereby the equilibrium As + 2 As + 4 As is temperature dependent.Arsenic atoms have a high probability of occurrence only at temperatures above 1700 “C and this is the temperature that has been found optimum for atomisation of arsine in a graphite tube.G At lower teinyeratures the diatomic molecule As has the highest prob-ability20 arid higher associates may be found at temperatures below 1000 “C. It can therefore be assumed that at the \very low concentration levels that are considered here part of the arsenic can travel over longer distances in such a molecular form or even in microcrystalline form such as “fl!~ dust.” Part of the arsenic may also be deposited on cooler parts and be re-volatilised by heating as the sublimation point of 633 “C is low.In th 224 WELZ AND MELCHER presence of hydrogen the arsenic may then also be atomised by collision with H radicals according to the equations and ASH + H +As + H, Another species that could possibly be formed in the atmosphere of the heated quartz cell is an oxide such as As,O,. The probability of oxide formation is certainly higher in the absence of hydrogen than in its presence because an excess of hydrogen reduces the partial pressure of oxygen substantially at elevated temperatures. The experiments with two heated quartz cells in sequence (Fig. 1B) showed that a %fold higher signal for arsenic is obtained in the second cell when the arsine is decomposed in the presence of hydrogen than in the absence of hydrogen (Fig.5C and D). Approximately the same ratio was obtained in other experimental arrangements and for higher arsine concentra-tions. This could possibly indicate a partial reaction of arsenic to the oxide in the absence of hydrogen. As + H +ASH + As Conclusion The atomisation of gaseous hydrides in a heated quartz cell is caused by collision with H radicals. The radicals are generated in a reaction with oxygen at elevated temperatures, and temperature plays an important role if the oxygen concentration is very low. The H radical concentration in the quartz cell atomiser is well above the equilibrium level because generation of radicals is a much faster reaction than their recombination. However this is only true if the quartz surface is well conditioned and free of active spots.In the presence of graphite several metals or simply a “catalytic film” on the quartz cell surface radical recombination is speeded up considerably and a much closer to equilibrium concentration of H radicals may only be present leading to significant decreases in sensitivity. The heavier elements within a group of the Periodic Table have a higher collision efficiency and therefore exhibit a less pronounced dependence on the oxygen concentration and temperature. In the absence of hydrogen arsine and probably also the other hydride-forming elements are thermally decomposed in the heated quartz cell but not atomised. The most probable species formed from arsenic are As and As,. In this form arsenic can he transported with the carrier gas over larger distances and be atomised in a heated quartz cell in the presence of hydrogen.The authors thank Prof. B. V. L’vov (Leningrad Polytechnic Institute Leningrad USSR), Prof. A. Cedergren and Dr. W. Frech (University of Umei Sweden) Dr. J. D6dina (Czecho-slovak Academy of Sciences Prague) and Dr. I. RubeSka (Geological Survey of Czecho-slovakia Prague) for useful discussions and valuable suggestions. The collaboration of B. Huber is gratefully acknowledged. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. References Thompson A. J . and Thoresby P. A. Analyst 1977 102 9. Verlinden M. Deelstra H. and Adriaenssens E. Talanta 1981 28 637. Chu R. C. Barron G. P. and Baumgarner A. W. Anal. Chem. 1972 44 1476. Fishman M. and Spencer R. Anal. Chem. 1977 49 1599. Welz B. and Melcher M. Spectrochim. Acta Part B 1981 36 439. Maher W. A. Anal. Chim. Acta 1981 126 157. McDaniel M. Shendrikar A. D. Reiszner K. D. and Wcst P. W. Anal. Chew. 1976 48 2240. Goulden P. D. and Rrooksbank P. AnaE. Chem. 1974 46 1431. Vijan P. N. and Wood G. R. Talanta 1976 23 89. Pierce F. D. Lamoreaux T. C. Brown H. Ti. and Fraser R. S. Afipl. Spectrosc. 1976 30 38. Andreae M. O. Anal. Chem. 1977 49 820. Siemer D. D. and Hageman L. Anal. Lett. 1975 8 323. Siemer D. D. Koteel P. and Jariwala V. Anal. Chem. 1976 48 836. Dcdina J . and Rubeqka I . Spectrochim. A d a Part B 1980 35 119. Meyer A. Hofer C. Tolg G. Raptis S. and Knapp G. Fresenius 2. Anal. Chem. 1979 296 337. Evans W. H. Jackson F. J. and Dellar D. Analyst 1979 104 16. Reamer D. C. Veillon C. and Tokousbalides P. T. Anal. Chem. 1981 53 245. Welz B. and Melcher M. Anal. Chim. Acta 1981 131 17. L‘vov R. V. Katskov D. A. and Kruglikova L. P. Zh. Prikl. Spektrosk. 1971 14 784. KoreEkovA J . Frech W. Lundberg E. Person J . A. ancl Ccclergren A. A n d . Chim. Acta 1981, Received August 24th 1982 Accepted September 29th 1982 130 267

ISSN:0003-2654    

DOI:10.1039/AN9830800213

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst February 1983 Vol. 108 $9. 225-234 225 A New Design of Graphite Furnace for Rapid Cycle Electrothermal Atomisation - Atomic-absorption Spectrometry Mohammad-Hossein Bahreyni-Toosi and John B. Dawson Department of Medical Physics General Infirmary Leeds LS1 3EX The design and performance of a new furnace developed for electrothermal atomisation - atomic-absorption spectrometry is described. The furnace consists of a segmented graphite rod. Sample solutions may be atomised directly from the rod without prior drying and pyrolysis with the result that the time interval between sample injections is reduced to <30 s. The performance of the segmented-rod atomiser with direct sample atomisation was investigated for the determination of silver gold aluminium bismuth, cadmium chromium copper mercury manganese lead vanadium and zinc in complex matrices containing salts and organic material.In all instances sensitivities comparable to those associated with conventional furnace systems were achieved. The coefficient of variation ranged from 4.0% for cadmium in 0.1 M nitric acid to 9% for gold in 1% sodium chloride solution. Keywords ; A tonzic-absorption spectrometry ; electrothermal atomisation ; seg-mented graphite rod ; direct atomisation We have previously reported preliminary studies of a technique developed for the determina-tion of elements in blood plasma protein fractions by electrothermal atomisation - atomic-absorption spectrometry (ETA - AAS). In this technique' the customary drying and pyroli-sation stages prior to atomisation are omitted and the time interval between the injection of samples is reduced to 30 s.When we attempted to apply the technique to long tube furnace systems the results were not satisfactory. \$7e attribute this failure to the increased size of the furnace with its slower non-uniform heating rate and longer vapour-residence time compared with our laboratory built system. In this paper we present details of the con-struction of the new atomiser a segmented graphite rod and also the results of further experience in its use. Experimental Apparatus The atomic-absorption spectrophotometer was a laboratory assembled system with deuterium-arc background correction (Fig. 1 ) . The light source was a hollow-cathode lamp driven at 400 Hz by 0.9 ms current pulses of 5-20 mA each.Electronic signal processing provided output signals corresponding to the intensities of the hollow-cathode lamp the deuterium arc and the corrected atomic-absorbance signal. The furnace was manu-factured from Ringsdorff spectroscopic carbon (Type RW 003). I t consists of a graphite cylinder with a transverse hole as the atomising chamber and a hole orthogonal to the chamber as the sample injection port. The surface area of the atomisation chamber is increased by cutting a screw thread (5 BA) into its surface. The ends of the cylinder are recessed to accommodate the flat ends of the cylindrical furnace support electrodes. This arrangement ensures good electrical contact and protects the contact surface: from environ-mental gases. The small mass of the furnace (30 mg) and effective electrical connections lead to rapid and uniform heating of the furnace.Nitrogen or argon (2-4 1 min-l) is used as the shielding gas for the furnace and electrode assembly. Normally a new grahpite furnace is prepared for use by vaporising concentrated nitric acid in the furnace to create an absorbing surface. However when determining some elements e.g. aluminium and vanadium it was found necessary to coat the furnace with pyrolytic graphite by adding methane to the shield gas. The furnace was driven by a laboratory built programmable power supply capable of providing up to 300 A at 11.5 V d.c. Details of the segmented-rod atomiser (SRA) are shown in Fig. 2 226 Monochromator BAHREYNI-TOOSI AND DAWSON NEW DESIGN OF AnaZysf VoZ.108 -+-+- - -Lens Fig. 1. Apparatus used in the investigation of atomiser performance. Comparative analyses were carried out using the same system with a laboratory built long tube (4 cm) Massmann type furnace fitted with extra water cooling to reduce the operating-cycle time and with a Unicam SP2900 flame atomic-absorption spectrophotometer operated with instrument settings as recommended in the manufacturer’s handbook. Transient signals were studied with the aid of a digital storage oscilloscope (Type 0s 4040 Gould Advance Ltd.) ; furnace temperature measurements were made with a photodiode system (Type RS308-067 R.S. Components Ltd.) calibrated against an optical pyrometer (Type 13422, Cambridge Industrial Instruments Ltd.) . Reagents Two general sample types were used in the trials of the atomiser blood plasma and plasma fractions separated by a previously described procedure2 ; and inorganic solutions with high salt content.Detail of atomiser I-5 mm-i Fig. 2. Segmented rod atomiser F e b wary 1983 227 AnalaR or Aristar grade reagents were used to minimise the trace element background signal. The eluting solution (0.05 M Tris - hydrochloric acid 0.2 M sodium chloride pH 7.4) used in the separation of blood plasma protein fractions by ion-exchange column chromato-graphy (DEAE-Sepharose CL6B Pharmacia Fine Chemicals) was further purified by passing twice through a column packed with Chelex 100 (Bio-Rad Laboratories). A nickel-based alloy BCS 387 was chosen as a representative metallurgical sample. The composition of the alloy was Ni 42.9% Cr 12.46y0 Mo 5.83y0 Ti 2.95% and Fe 36%.Standard solutions were prepared by appropriate dilutions of "atomic-absorption spectroscopy" stock solutions (BDH Chemicals Ltd.) and made up to a final acidity of 0.1 M with nitric acid. GRAPHITE FURNACE FOR RAPID CYCLE ETA - AAS Procedure As the purpose of this paper is to demonstrate the wide range of potential applicability of the segmented-rod atomiser rather than its optimised performance for a particular analysis, details of sample preparation procedures will not be presented here. Where relevant, additional information will be given in the Results and Discussion section. Results and Discussion Evaluation of Furnace Performance The characteristics of the SRA will be illustrated by comparing its performance with that of a Massniann type system when both are applied firstly to the determination of copper in either 0.1 M nitric acid or in the ion-exchange column eluting buffer and secondly to the determination of bismuth in either 0.1 M nitric acid or in solutions of nickel-based alloys.Typical operating conditions of the furnaces are summarised in Table I. In the rapid heating mode the full atornising current is applied to the SRA without prior drying and pyrolysis of the sample. The cycle time for the SRA is much shorter than that of the conventionally operated tube furnace (1-2 min) particularly when used in the rapid heating mode (-4 min). The simplicity and speed of operation of the SRA are significant advantages of the system.The energy consumption of the SRA in either conventional or direct atomisa-tion modes (-3000-4500 J) is also much less than that of a conventionally operated tube furnace (-14000 J). TABLE I TYPICAL FURSACE OPERATIKG COXDITIONS FOR THE DETERMINATION OF COPPER AND BISMUTH Paraiiie tcr Segiiientetl rod-Current/:\ . . . . . . Titiie/s . . . . . . ?'cmpcniturc/ I< . . . . injectioiils . . . . . . Currcnt/A\ . . . . . . Tinic/s . . . . . . Tempcrnturc/I< . . . . injectionls . . . . . . r 1. I line iutcrval bctwecn sample Tube furnace- 1 iiiie interval bctivccii sample Stepwise heating f A Dry Ash Atomise 7 3 7u-13i\ --z cu . . 37 35 ,?0 45 300 300 . . 10 10 10 10 1 . 1 1.0 . . 360 350 920 670 7970 3800 . . 15 15 40 35 2iO 350 .. 30 30 40 30 5 5 . . 370 350 790 670 3 870 2400 - - . . 100 - - -Rapid heating : atomise rA.-c u Bi 7 300 300 2970 2800 1 . 1 1 .0 _- 25 - 30 Performance data based on peak-height measurements using a chart recorder with 0.3 s full-scale response tinie of the SKA and tube furnace are shown in Table 11. No satis-factory results were obtained when the tube furiiace ivas used in the rapid heating mode. The rciisons for this failure will be discussed when the oscilloscope recordings of the atomiser absorhancc signals are considcrcd. Tlie copper and bisniuth sensitivities of the two furnac 228 BAHREYNI-TOOSI AND DAWSON NEW DESIGN OF Analyst VoZ. 108 TABLE I1 COMPARATIVE PERFORMANCE OF ATOMISING SYSTEMS FOR THE DETERMINATION OF COPPER AND BISMUTH Tube furnace Segmented rod A 1 Stepwise heating Stepwise heating Rapid heating Parameter r---h-7 c u Bi Bil c u ---7? c u Sample volume/pl .. . . . . 20 10 5 5 5 5 Sensitivitylpg 1 % absorption-A* 21 62 16.5 39.5 20 33.5 Bt . . 26 22.5 24 C$ . . 76 44 40 Detection limit/pg-A 8 11 4.9 10 6.0 8.5 B 11 5.0 5.0 c 13 11 10 A . . 4.2 5.5 4.0 5.0 4.0 5.4 B . . 6.0 5.5 5.5 * . 8.8 8.6 7.1 c * Matrix A 0.1 M nitric acid. t Matrix B Tris - hydrochloric acid - sodium chloride eluting buffer. $ Matrix C 1% solution of BCS 387 in 1 M nitric acid - 1 M hydrofluoric acid. Coefficient of variation at 20 x detection limit %-systems and of the modes of operation are similar. The larger sample volumes that can be accommodated by the tube furnace facilitate measurements on more dilute solutions.In all instances the sensitivities for both elements in complex matrices are about 20% less than those for simple solutions. The detection limits for copper and for bismuth by the three systems are again comparable. However overall the performance of the SRA is on average 25:/ better than that of the tube furnace. The reproducibility of the determination of copper is better than that for bismuth for all systems and measurements of simple solutions are more precise than those where there is a complex matrix. Comparison of SRA with a Tube Furnace The absorption traces generated during the atomisation of bismuth from simple and complex matrices using the tube furnace and the SRA are shown in Fig.3. When the tube furnace was used with conventional stepwise heating a satisfactory signal from bismuth was obtained. There are however considerable differences between the waveforms generated by simple and complex matrices. The atomic-absorption peak heights are practically identical but the peak area from the complex matrix is about 30% less than that from the simple matrix. This suppression of signal was also found when the absorbance signals were observed using the slower responding chart recorder in place of the oscilloscope (Table 11). No clear atomic-absorption signal from bismuth in either simple or complex matrix was obtained when the tube furnace was used in the direct-atomisation mode. Inspection of the deuterium and hollow cathode lamp traces reveals very irregular patterns with super-imposed oscillations.These patterns were attributed to condensation and vaporisation of sample water and salts on the walls of the tube as the furnace gradually heated. The oscillo-scope traces of absorption in the SRA are more regular than those of the tube furnace. There is no difference in the appearances of the atomic-absorption signals from simple or complex matrices and from stepwise heating or direct atomisation. The amplitude of the atomic-absorption signal from the SRA is comparable to that from the tube furnace but the peak area of the former is only about 200/ of the latter. The background absorption signals, deuterium and hollow cathode lamp from the SRA are of considerable interest. In the step-wise mode the complex matrix generates a substantial background at the same time as the bismuth signal; this background signal is not present in the direct-atomisation mode.In the latter mode there is a large initial signal followed by a small peak before the release of bismuth occurs. In both modes some background absorption occurs after the bismut February 1983 GRAPHITE FURNACE FOR RAPID CYCLE ETA - AAS 229 signal has been generated. From these experiments we concluded that tube furnaces are not likely to be suitable for use in the direct-atomisation mode and that interference effects in the SRA used in the direct-atomisation mode are likely to be very different from those in conventional systems. Tube furnace atomiser Segmented rod atomiser A r I \ Stepwise heating Direct atomisation Stepwise Direct -5s' -5s- Power Power Power Power heating atomisation Fig.3. Oscilloscope traces of absorption signals produced during the atomisa-tion of bismuth (0.5 ng X = 223.1 nm) from (A) 0.1 M nitric acid and (B) nickel-based alloy BCS 387. Arrows indicate the instant of maximum bismuth absorption. Furnace operating conditions as in Table I. Comparison of Modes of Operating the SRA A series of experiments was carried out to determine the effect of the matrix on the deter-mination of copper in plasma protein fractions using the SRA. Fig. 4 presents oscilloscope recordings of the absorption signals produced when the SRA is operated in three different modes as follows stepwise heating i.e. drying pyrolysis and atomisation ; drying followed directly by atomisation ; and direct atomisation without drying and pyrolysis.The inflections in the deuterium traces for the nitric acid solutions are electrical artefacts. In the example of the deuterium traces generated by the buffer solution there is a progression from a single peak (arising from the inorganic matrix when stepwise heating is used) through to one showing an additional peak (owing to smoke from organic material) to the trace produced by direct-atomisation showing an organic smoke peak that was enhanced by the vaporisation of water in addition to the inorganic peak. The hollow cathode lamp traces show that the maximum of the copper absorption peak occurs after the vaporisation of the matrix material. The traces for the hollow cathode lamp signals generated by the buffer solution when stepwise heating drying only and direct atomisation are used are practically identical in terms of vaporisation of the inorganic matrix and copper.The background corrected atomic-absorption signals are very similar in form irrespective of heating programme though the magnitude of the signal from the buffer solution is about 25% less than that from 0.1 M nitric acid. The origin of this reduction in signal was not investigated but is likely to arise from the loss of copper as a volatile compound before the temperature is sufficient to produce atomisation. The recordings shown in Fig. 4 indicate that when the SRA is used in the direct-atomisation mode the analytical signal produced is the same as when stepwise heatin 230 BAHREYNI-TOOSI AND DAWSON NEW DESIGN OF Stepwise heating Drying only Direct atomisation Power Power Power 1.1 s 1.1 s I H 1.1 s H 1 D2 signals HCL signals C 0 = 50% 1.0 s Time AA signals V Oa2 L 1.0 s Time z L J.I A-P-- I J J Analyst Vol. 108 Fig. 4. Oscilloscope traces of absorption signals produced during the atoniisation of copper (0.5 ng X = 324.7 nm) from ('4) 0.1 M nitric acid and (B) Tris - hydrochloric acid - sodium chloride buffer. Furnacc operating conditions as in Table I. Arrows indicate the instant of maximum copper absorbance. is used. sample are unnecessary. Hence in this instance the time and care required for drying and pyrolysing the Temperature Dependence of Absorbance Signals Fig. 5 shows the rise in furnace temperature (as measured by a photodiode detector system calibrated against an optical pyrometer) when a current of 300 A is passed through it and the corresponding absorbance profiles for the release of bismuth and copper using direct 3 000 2 500 ?2 2000 a Q E $ 1500 1000 500 I$ Power off J 0.1 a, 9, 2 n a V m 3 Tim e/s Fig.5. Time dependence of (A) SRA temperature and absorbance signals of (€3) bismuth (0.5n.g) and ( C ) copper (0.5 ng) from 0.1 M nitric acid in the direct-atomisation mode. Furnace current M 300 A Febrzwy 1983 GRAPHITE FURNACE FOR RAPID CYCLE ETA - AAS 231 atomisation of simple solutions. The maximum heating rate of the furnace is 3200 K s-l and temperatures up to approximately 3800 K may be achieved by extending the heating period.In the course of measuring the furnace temperature it was found that heating rates and atom-vapour residence time in the furnace were influenced by the shield gas flow-rate and the electrode cooling water flow; these were maintained as constant as was practicable. The shapes of the absorbance graphs are of interest. Atomic absorption by both bismuth and copper is first detected at -1800 K but the maximum signals occur at different time intervals later (bismuth 0.25 copper 0.5 s). The duration of the bismuth absorbance signal is <0.5 s and that for copper approximately 2 s. This difference probably reflects the difference in volatility of the elements. Though the peak height for bismuth is greater than that of copper the area of the copper peak is greater and leads to the greater sensitivity of the latter element (Table 11) when measurements are made using the slower response chart recorder.Effect of Graphite Surface on Absorbance Signals Early in the development of the SRA direct-atomisation system it was observed that the interior surface of the furnace played an important role in the success of this approach. This effect is demonstrated in Fig. 6 where as the furnace ages the change in sensitivity and reproducibility in the determination of copper in a Tris - hydrochloric acid - sodium chloride eluting buffer is presented. In operating the SRA in the conventional mode with stepwise !5 Fig. 6. Effect of SR.\ usage and nitric acid pre-treatment on sensitivity and precision in the dctcrmination of copper (0.375 ng) in buffcr solution.A Stepwise heating un-treated SR.4 ; 13 direct-atomisation untrcatcd SH.1; and C direct-atomisation with nitric acid pre-treatment of SRX. The bars indicate the standard errors of the means of seven measurements. heating the sensitivity of the analysis gradually falls with increasing use (-80,; reduction after 120 firings). Initiallji the standard error of the mean of seven observations (SEM) is approximately 30/ after which the precision remains constant at -294 for -90 firings and then deteriorates rapidly to -I-*( after 120 firings. \!'hen used in the direct-atomisation mode the SKA performance is initially worse than that of conventional operation with a sensitivity that is 30% lower and has an SEN of 6.50,;).After about 50 firings the SRA sensitivity and precision are the same for both direct and conventional operation. There-after the performance of the direct-atomisation system remains constant and after 120 firings its sensitivitl- is 100,; greater than that of the conventional niode with a SEN of 27;. Clearly the direct-atomisation approach benefits from the ageing of the furnace. I t was found that this process could be quickened in new furnaces by vaporising concentrate 232 AnaZyst VoZ. 108 nitric acid. In this procedure the furnace was first cleaned by heating it through several atomisation cycles then 10 pl of concentrated nitric acid were injected and atomised; this operation was repeated three times before the SRA was used for analysis. Fig. 6 shows the performance of an SRA after such treatment.I t can be seen that from the first firings of the SRA the sensitivity and reproducibility are the same as the best obtained with con-ventional operation or a “naturally” aged furnace. The performance of the acid-treated SRA remains practically unchanged throughout its lifetime (-200-300 firings). I t should be noted however that these in-depth studies were carried out on a single sample type i.e., copper in a buffer solution. Nevertheless we believe that acid treatment is generally beneficial. There are however some elements e.g. aluminium and vanadium for which a different treatment such as coating with pyrolytic graphite and the addition of a wetting agent to improve sample dispersion is necessary to achieve satisfactory analyses.It is clear that the nature of the graphite surface is important to the successful operation of the SRA and further investigation of the mechanism of its effect is required. BAHREYNI-TOOSI AND DAWSON NEW DESIGN OF Analytical Applications The potential range of application of the SRA with direct atomisation was explored by a series of pilot experiments. Most detailed studies were directed at the determination of copper in plasma protein fractions and of bismuth in nickel-based alloys and have been presented earlier in this paper. Further information on other analyses carried out with the SRA will now be presented. TABLE I11 PERFORMANCE DATA OF SEGMENTED-ROD ATOMISER USING DIRECT-SAMPLE ATOMISATION WITH SAMPLE VOLUME 2 OR 5 pl AND CYCLE TIME 25 S Element/ nm Matrix Ag Nickel-based alloy BCS 387 .. 328.1 0.1 M nitric acid . . Au Sodium chloride solution ii yo) : 242.8 Distilled water . . . . . . Al Rat plasma and plasma fractions 309.3 0.1 M nitric acid Ri Nickel-based alloy; BCS id.7 223 1 0.1 M nitric acid Cd,. Rat whole blood and plasma : 228.8 0.1 M nitric acid Cr Tris - HCI buffer - 0:2 M N k l :: 357.9 0.1 M nitric acid . . . . . . Cu, 324.8 0.1 M nitric acid . . . . . . Hg Tissue culture medium . . . . 253.7 0.1 M nitric acid . . . . . . Mn Sea water . . 249.5 Doubly distilled wa&r ’ ’ . . Pb Nickel-based alloy BCS iH’7 283.3 0.1 xi nitric acid . . . . . . 1.I Human plasma fractions and buffer Sensi tivitylpg, 1 ‘x absorption . . . . 3.0 . . . . 2.5 . . . .20 . . . . 11.6 . . . . 25 . . . 17.8 . . . . 40 . . . . :i:i5 . . . . 0.48 . . . . 0.35 . . . . 33 . . . . 4.5 solution . . 24 . . . . 20 . . . . 80 - . . . . . . . . I .5 . . . . 1.2 . . . . I i . 8 ._ 26 > :ilSj 1.0 11 nitric acid . . . . . . . . . . 2 3 1 Zn Hiim;in plasma fractions and buffer solution . . ll.2!l 21:;.9 0.1 JI nitric acid . . . . . . . . . . 0.22 Dctrctiori limit (D.L.)/PK 1.0 0 . X 8.0 4.5 7.0 X.5 0.7 0.8 1 0 1 0 24 SX 5.0 6.0 8 0 -1.3 1 3.0 4.2 100 0 . 1 6 lJ.12 Cocfficien t of variation a t 2 0 x D.L. 7, 6.0 4.5 9.0 7.5 7.0 5.6 ij.4 5.5 ?.I 4.0 4.5 5.8 5; 4.0 6.0 5.8 4.8 4.!l 5.3 -4. ! 1 .;> 4.8 Literature valuesJ Sensi tivi ty/pg, 17 ihsorptioii r 20 50 40 1 20 30 200 L 20 200 1 D.L./’ Pg 0.1 10 r 20 0.1 In 2 7 00 0.2 2 1 (KJ 0.05 The results obtained by peak-height measurements using a chart recorder for the deter-mination of a variety of elements in several matrices are summarised in Table 111.It should he noted that these are not fully validated analytical methods but serve to indicate the range of possible applications of the SKA with direct sample atomisation. In all instances the sensitivity of the SRA direct-atomisation system was close to the literature values quoted for simple solutions of the analyte elements. Apart from chromium mercury and lead the presence of sample matrix leads to suppression of the analytical signal. The average coefficient of variation of single measurements on simple solutions was about 5% and ranged from 4 to 7 0 / ; the corresponding values for complex matrices were about 6% with a range from 4.5 to 9%.As manual sample dispensing was used throughout a signifi-cant proportion of the observed error may come from that source though clearly complex matrixes also contribute to reduced precision February 1983 Matrices Containing Organic Material The segmented-rod atomiser was developed in response to the need for rapid analyses created by the large number of samples arising from the separation of plasma protein fractions by ion-exchange column chromatography. To verify the SRA - rapid heating technique for the determination of copper and zinc in plasma protein fractions the major copper- and zinc-containing fractions were analysed by both continuous-nebulisation flame atomic-absorption spectrometry and rapid heating electrothermal atomisation.In view of the matrix effects (Table 11) standard solutions of copper and zinc were prepared using the eluting fluid as diluent. The correlation coefficients between the methods for both copper and zinc are excellent. The slope of the linear-regression line for copper however is 0.92 and suggests a real difference between the results obtained by the two methods This difference may arise from a non-linear relation-ship between the results obtained by the two methods at the higher concentrations. How-ever as the primary purpose in developing the SRA direct-atomisation system was the determination of low concentrations of copper and zinc in fractionated plasma the difference at higher concentrations was not investigated further.Even so the errors are not large and the recoveries for the whole separation and element determination procedure (using flame or SRA as appropriate for 12 samples) were copper 99.8 * 1.9% (mean * s.d.) and zinc 105.7 & 6,9y0. The elevated recovery of zinc reflects the problems of contamination that arise with this element. GRAPHITE FURNACE FOR RAPID CYCLE ETA - AAS 233 The results obtained by both methods are presented in Fig. 7 . b. ~ A= 1.48 x 10-3 5 = 0.9204 /. r = 0.999 7 I 1 - 0.1 E 4 f .- *.' c 2 0.05 8 -s v) U 2 a 0.0 I/' A= 1.12 x 10-4 B= 0.9788 '' r = 0.9986 / I I le'' A- 7n I / Line of equality 7-Line of equality / * Fig.7. Comparative performance of SRA direct-atomisation system with continuous nebulisation flame atomic absorption for the determination of copper and zinc in plasma protein fractions. Correla-tion coefficient r. Regression line y = A + B x . Aluminium cadmium chromium and mercury were also determined in organic matrices. The successful determination of aluminium in rat tissue blood plasma (diluted ten-fold with eluting buffer) and plasma fractions necessitated the use of argon as the shield gas pyrolytic coating of the SRA furnace the addition of an optimum amount of O.Olyo Triton X-100 and the raising of the atomisation temperature to about 3200 K. The pyrolytic coating was produced by introducing methane into the shield gas for the first five firings of a new atomiser.The coating was maintained by firing the atomiser twice in the argon - methane atmosphere after every 50 analytical firings. In comparison with data from the literature the atomic-absorption sensitivity is good but there is approximately 30% suppression by the complex matrix. Cadmium was readily determined in rat blood and plasma following a 1 + 19 dilution with warm (-35 "C) 0.1% Triton X-100 using the SRA with slightly less power (280A 1 s) than was uscd for copper. The sensitivities €or chromium and mercury were improved by the presence of complex matrices. The reasons for these effects have not been investigated but with mercury it is likely that the matrix formed complexes with the element and thereby reduced its pre-atomisation volatilisation 234 BAHREYNI-TOOSI AND DAWSON Inorganic Matrices There was 20% suppression of the silver signal and an enhancement of 40% of the lead signal when determined in a 1% solution of the alloy.Sodium chloride solutions are difficult matrices for many analyses and test not only the characteristics of the atomiser but also the effective-ness of the background correction. Gold and manganese were successfully determined in such a matrix though there was considerable suppression of the signal (gold -40y0 manganese -20%). As a further test of the atomiser performance attempts were made to determine vanadium. Results comparable to literature values were obtained by replacing the usual nitrogen shielding gas with argon and coating the rod with pyrolytic graphite.For the atomisation of vanadium the furnace temperature was raised to >3300 K by applying full power (300 A) for 1.7 s. In addition to bismuth silver and lead were also determined in nickel-based alloys. Conclusion This work has shown that the performance of the SRA used in the direct-atomisation mode is comparable to that of a conventional tube furnace. There are two significant advantages of the former over the latter the time interval between sample injections is considerably reduced to (30 s; and the SRA is simple easy to operate and inexpensive. The SRA direct-atomisation system appears to be of general applicability and interference effects are no worse than in conventional systems and in some instances could be less. The general indications are that the relative importance of the basic physical and chemical parameters of ETA - AAS using SRA direct-atomisation differs from that found in stepwise atomisation using a tube furnace. These require further investigation in order that the relative merits of the two techniques may be more clearly defined. The authors thank W. R. Nall and A. Stubbs of the Bragg Laboratory Sheffield for collaboration in the analysis of nickel-based alloys and their colleagues D. C. Chilvers G. P. Davitt D. J. Ellis G. W. Fisher A. F. Johnston and A. D. Kersey for assistance in the con-struction and operation of the atomic-absorption atomiser systems. References 1. 2. 3. Bahreyni-Toosi M. H. Dawson J. B. and Ellis D. J. Analyst 1982 107 124. Dawson J. B. Bahreyni-Toosi M. H. Ellis D. J. and Hodgkinson A. Analyst 1981 106 153. Fuller C. W. “Electrothermal Atomization for Atomic Absorption Spectrometry,” Chemical Society, Received August 26th 1982 Accepted September 28th 1982 London 1977

ISSN:0003-2654    

DOI:10.1039/AN9830800225

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

Analyst February 1983 Vol. 108 $9. 235-243 235 Determination of Lead in Blood by Flame Atom ic-f I uorescence Spectrometry Prem R. Sthapit and John M. Ottaway and Gordon S. Fell Department of Pure and Applied Chemistry University of Strathclyde Cathedral Street Glasgow G1 1XL Department of Clinical Biochemistry Royal Infirmary Glasgow G4 OSF A simple rapid method is described for the determination of lead in blood. Dilution 1 + 4 with Triton X is the only sample preparation required and measurements are carried out using a purpose-built atomic-fluorescence spectrometer with a nitrogen-separated air - acetylene flame. The prepara-tion and operation of the lead electrodeless discharge lamps used as the excitation source have been optimised by a ten-factor Simplex procedure.A detection limit of 6 pg 1-1 has been achieved for lead in aqueous solution. No significant chemical interferences were observed from the major constituents of the blood matrix and a second continuum source is used to achieve auto-matic background correction for scattered radiation. Aqueous lead stantlards are used for calibration. ;\ccuracy was established by satisfactory comparison with values reportctl for quality control blood samples. Keywords Flame atomic-Puorescence spectrometry ; lead determination ; blood analysis ; electrodeless dischnvge lamps In 1979 Micliel ct al.1 described a flame atomic-fluorescence instrument system that allowed the development of simple rapid procedures for the determination of cadmium in blood and urine. The essential features of the instrument were a double monochromator to reduce stray a second light source based on a xenon arc to permit accurate automatic back-ground correction for scattered radiation a photon counting detection system and an electrodeless discharge lamp (EDL) light source prepared and operated under optiniised conditions.3 With this instrument cadmium can be determined in urine by direct aspiration of the acidified sample into a separated air - acetylene flame and in blood following only 1 + 4 dilution with Triton X.Although also used for research this instrument has been used extensively for routine cadmium determinations. The sensitivity of atomic fluorescence has meant that pre-concentration steps are not required and the use of flame atomisation allows a more rapid t liroughput of samples than any other technique of comparable sensi-tivity.Although an atomic-fluorescerce instrument with tlie required design features is not commercially available the system was assembled almost exclusively from standard components and could be used in many laboratories that are required to carry out population surveys for occupational or environmental cadmium exposure. The use of the xenon arc in this instrument to generate atoriiic-fluorescence signals has also been demonstrated4 and, with wavelengtli modulation for automatic background correction metals such as copper, iron and magnesium present at higher concentrations in biological materials can be deter-mined. Following tlic successful tlevelopnicnt of methods for the deterinination of cadniiuni in biological materials it was considered desirable to investigate tlie development of a metliod of similar simplicity and speed for the determination of lead in whole blood.Yethods most commonly used for this determination at present depend on electrotliermal atomic-absorption spectronwtry,5t6 wliicli is relatively slow compared with methods based on flame atomisation. A rapid flame atoniic-fluorescelice spectrometric procedure with the required sensitivitj-would clearly be of great benefit in reducing tlie cost and time of population surveys designed to investigate occupational or environmental exposure to lead. Tlie introduction of an atomic-fluorescence method for lead requires the preparation of a lead electrodeless discharge lamp of suitable sensitivity and stability.In our laboratory tlie preparation of EDLs of optimised and reproducible performance is carried out under conditions identified by a simples optiinisation procedure.397 Several attempts have been made to produce lead EDLs with satisfactory performance but onlj- a recent study involving lead bromide as the fill niaterial has provided an optimised EDL capable of the sensitivit!. and stabilitj- required for the deteriiiination of lead in most blood samples. Tlie preparation of tlie lead EU 236 STHAPIT et al. DETERMINATION OF Analyst Vol. 108 and the subsequent development of a rapid procedure for the determination of lead in whole blood is described here. The developments described confirm that a flame atomic-fluorescence instrument can offer rapid accurate routine analyses of both lead and cadmium in clinical materials.Experimental Instrumentation A detailed description of the flame atomic-fluorescence instrument and the components used in its assembly has been given previously.1 The lead EDLs were operated in a Broida 3/4-wave 210L cavity with power from a Mark I11 microwave generator (both supplied by Electromedical Supplies). The continuum light source for background correction was an EM1 Varian VIX-300 UV xenon arc. Light from the two sources was introduced alternately into the flame via a mechanical chopper assembly (laboratory constructed) with a modulation frequency of 300 Hz. Air at a controlled temperature was passed through the microwave cavity to stabilise the output of the excitation light source.The nebuliser (mixing chamber unit) was a standard Perkin-Elmer system and was operated with a laboratory-constructed capillary burner and flame separator. A nitrogen-separated air - acetylene flame operated under stoicheiometric conditions was used throughout this work with a 15 1 min-l flow of nitrogen separating gas. The nebuliser uptake rate was 9.5 ml min-l for aqueous standard solutions and a correction factor of 1.19 was used to allow for the reduced uptake rate of blood samples diluted 1 + 4 with Triton X. The monochromator was a Spex 1672 double monochromator and was set at 283.3nm with a slit width of 0.5 mm. The 9789 QB EM1 photomultiplier was set at an operating voltage of 1110 V and the Ortec Brookdeal 5 CI photon-counting system was used at a 1-s integration time and a 55% duty factor.The A and B channels of the photon counter were synchronised to the chopper modulation unit alternating the two light sources by a photo-diode-derived reference signal. The instrument was generally allowed a 30-min warm-up period for stabilisation. Accurate setting of the background correction system was achieved by varying the intensity of the xenon arc light source whilst monitoring the A-B signal on the photon counter.1 During this procedure a solution giving a scattered light intensity significantly greater than that obtained from real samples was nebulised into the flame. The solution used was 2% m/V calcium orthophosphate. Results for sample solutions and standards were taken from an average of ten readings.Blank values were always checked whilst nebulising de-ionised water. Preparation of Lead Electrodeless Discharge Lamps The method used for the preparation of EDLs was similar to that described by Michel et aL7 EDL blanks were cut from Vitrosil quartz tubing of 10 mm 0.d. and 8 mm i.d. One end of the tube was sealed and 4 cm from this end a narrow capillary constriction was drawn to allow easy sealing. Above the constriction 10 cm of quartz tubing was left for handling purposes. The optimisation of the method used to prepare the EDL is described under Results and Discussion but the basis of the procedure is as follows. The blanks were chemically cleaned and then heated under a partial vacuum in an argon atmosphere to remove volatile impurities. A known amount of lead (25-2000 pg) was introduced as lead bromide and a small amount (20 pl) of water was added.The water was removed under vacuum and the lead salt in the bulb of the quartz tube was partially sublimed by the microwave discharge. The bulb was then sealed under vacuum in an argon atmosphere. The vacuum system is based on that described by Gleason and Perte18 as modified by Michel et al.' A pressure below Torr was achieved routinely within the 30-min interval between the making of each EDL. Reagents purity de-ionised water was used for the preparation of all solutions. chloric acid. in 0.04 M hydrochloric acid. All signals were recorded as counts per second using a 1-s count period. All reagent solutions were prepared from reagents of the highest available purity.High-A stock solution of 1000 pg ml-l of lead was prepared from lead nitrate in 0.04 M hydro-Standard solutions of lead of the required concentrations were prepared dail February 1983 LEAD IN BLOOD BY FLAME AFS 237 Blood Sample Preparation Standard disposable syringes and needles were used to take blood samples by venepuncture. The blood was kept in plastic bottles containing potassium EDTA as an anticoagulant. Direct aspiration of blood samples into the air - acetylene flame was not possible. Two dilution - treatment procedures were investigated and found to be equally acceptable. Procedure 1 Blood (1 ml) was diluted with 4 ml of 0.1% Triton X (to ensure complete haemolysis). After mixing the diluted blood was centrifuged (5 min at 3000 rev min-l) to remove cellular debris.To allow for the slower rate of uptake of the diluted blood compared to aqueous lead standards a correction factor of 1.19 was applied. A supernatant volume of approximately 4.5 ml was sufficient for measurement of ten l-s readings under stabilised conditions. Procedure 2 Aristar grade) (1 ml) with simultaneous mixing on a vortex rotamixer. was centrifuged (5 min at 3 000 rev min-1). the flame and was sufficient for ten l-s readings. necessary in this instance. The supernatant fluid was aspirated directly into the flame. Blood (1 ml) was diluted with 3 ml of water and treated dropwise with nitric acid (25% V / V , The resulting solution The supernatant was again aspirated directly into No uptake rate correction factor was Results and Discussion Optimisation of the Construction and Operation of Lead Electrodeless Discharge Lamps As the application of the Simplex algorithm to the optimisation of the construction and operation of cadmium EDLs has been described in detail in previous publication^,^,^ and the same procedures were used here only those considerations relating specifically to the con-struction of lead EDLs will be discussed.In the procedure used in these investigations ten variables were considered to have an important influence on the properties of the final EDL produced. The simplex method is thus particularly convenient for optimisation purposes as it allows a large number of variables to be included with a minimum of experimental work. The variables studied and the optimum conditions identified are summarised in Table I.These can be understood from the following brief description of the method. I t is not practicable to vary the material diameter or length of the EDL blank which were as described above or the nature of the fill gas. Argon was used as it is now generally accepted to offer the best compromise between a high radiant output and a long lifetime. The lead compound used and its method of introduction are very important variables but these can only be fixed by preliminary experimentation. In order to maintain the best possible reproducibility of EDL preparation it is preferable to introduce the lead as an aqueous solution. Preliminary experiments with water-soluble lead compounds such as lead acetate lead nitrate and lead perchlorate failed completely as the discharge of the EDLs TABLE I FACTORS INCLUDED IN THE SIMPLEX OPTIMISATION THEIR PRACTICAL LIMITS AND THE OPTIMUM CONDITIONS DETERMINED BY THE SIMPLEX Factor Symbol Practical limits Optimum conditions Mass of lead .. . . . . . . . Time under vacuum after water removed . . Argon pressure under microwave excitation during preparation . . . Microwave power for discharge during prepara-tion . . . . Time interval for microwave discharge during preparation . Time for EDL to cool before evacuation . . Time under vacuum after cooling period . . Final argon fill pressure . . EDL operating microwave power . . EDL operating temperature . . 25-2000 pg 0-60 rnin w, 11 A1 0-40 Torr PI 0-200 w 12 1-60 s t 3 0-60 min 0-60 rnin t4 A2 0-40 Torr p2 0-200 w T 0-200 "C 961 pg 25 s 7.7 Torr 119 w 13 s 229 s 115 s 7.3 Torr 65.0 W 200 " 238 STHAPIT et al.DETERMINATION OF Analyst Vol. 108 could not be initiated. Amongst the insoluble lead compounds considered lead bromide was found to give the best results. Consequently an accurate mass of lead bromide (factor Wl) weighed by electromicrobalance was introduced directly into the bulb of the EDL and was followed by 20 pl of water. Two variables were defined concerning the pressure of the argon fill gas the argon pressure during the microwave excitation used at the preparation stage (factor A,) when the fill material is partially sublimed and the final pressure of argon (factor A,) before the bulb of the EDL is sealed.When the microwave excitation is applied during the preparation stage the lead bromide added to the EDL blank is partially decomposed to lead and bromine and some part of the breakdown products are removed from the EDL bulb. It has been noted previously3 and it was confirmed for lead that this is a critical stage in the preparation of EDLs. Hence the following five variables were investigated the time the EDL spends under vacuum after removal of the water (factor tl) the time of the microwave discharge (factor t,) and the microwave power used (factor Pl) the time allowed for the EDL to cool after the discharge is discontinued (factor t3) and the time the EDL is held under vacuum after the end of the cooling period (factor t4). Finally as the EDLs are operated in a thermostatically controlled 3/4-wave Broida cavity, the microwave power (factor Pz) and the temperature (factor T ) are the two variables that affect the performance of the EDLs produced by the above procedure.Therefore having limited the search to a single compound (lead bromide) ten factors were investigated and the practical limits of these factors are shown in Table I. The performance of each EDL was evaluated in terms of its atomic-fluorescence detection limit for lead using the instrumentation described above. Detection limits were defined as the concentration that gives a signal equal to twice the background noise. In this system the noise is defined as the square root of the total signal in the background channel of the photon c0unter.l As the EDL preparation and operation involve ten variables eleven initial EDLs were constructed using a matrix and accompanying equations as defined by Yarbro and Demir~g.~ The Nelder and Meadlo variable-size simplex algorithm was then used to move the simplex and a PET computer program was used to identify the next set of vertices.The simplex was assumed to be terminated when the step size of each factor was small and not much greater than the reproducibility of the measurement of the factors and where improvement in the lead detection limit was negligible. The ten-factor simplex search generated vertices that involved the construction of 80 EDLs. The optimum conditions are given in Table I, but very little change was achieved after vertex No. 67.EDL Performance When the simplex was terminated the detection limit achieved for the atomic-fluorescence determination of lead was 25 pg 1-1. The optimum conditions identified by the simplex were then used to prepare a further nine EDLs over a period of 3 months to test the reproducibility of the construction procedure. The relative standard deviation achieved was 35% (see Table II) with a mean value of the detection limit of 19pgl-l and a range from 12 to 31 pg 1-l. This is considered satisfactory although slightly inferior to the reproducibility TABLE I1 REPRODUCIBILITY AND DETECTION LIMITS OF LEAD EDLs ‘Ten identical J3DLs were prepared using the optiinuni factors levels (from Table 1) mtl the reproducibility was evaluated in terms of their detection limit measured as men-tioned in the text with a l-s counting time.EDL Detection limitlpg 1-1 1 12 2 18 3 30 4 14 5 23 Mean detection limit . . Standard deviation . . Relative standard deviation EDL Detection limitlpg 1-’ 6 20 7 13 8 31 9 16 10 15 19 pg 1-1 35% 6.81 pgl-February 1983 LEAD IN BLOOD BY FLAME AFS 239 achieved with cadmium3 and seleniumll EDLs. When the simplex was started the heater used to thermostat the EDL during operation had a maximum temperature of 200 “C. At a later stage a re-designed heater allowed temperatures up to 280 “C to be used and when the EDL operating temperature was re-evaluated using the optimum EDLs a routine detection limit of 12 pg 1-l could be achieved at 230 “C. The development of the method for the determination of lead in blood was carried out with EDLs of this performance.In subse-quent work with a heater of greater stability at higher temperatures and a slightly improved EDL a detection limit of 6 pg 1-1 has been achieved. This could be of some importance to the determination of low levels of lead in blood as will be mentioned later. The relationship between relative standard deviation and lead concentrations for aqueous solutions is shown in Fig 1. At 25 pg 1-1 the precision becomes better than 10% and this is taken as the determination limit. The calibration graph is linear over more than three orders of lead concentration (from 10 to lo4 pg 1-1) and easily covers the range of lead con-centrations found in blood samples after 1 + 4 dilution. Analytical Investigations As has been previously noted for the determination of cadmium by atomic-fluorescence spectrometry,l an enhancement of lead fluorescence by mineral acids such as nitric sulphuric and hydrochloric acid was observed.The effect of hydrochloric acid is illustrated in Fig. 2, from which it can be seen that the lead signal becomes constant at hydrochloric acid con-centrations above about 0.04 M. To counteract this effect and also to improve the stability of dilute lead solutions all standards were prepared in 0.04 M hydrochloric acid. A similar acid enhancement phenomenon observed in measurements of the laser-enhanced ionisation spectrometry of indiuml2 has been ascribed to a burner “memory” effect and the charac-teristics observed with both cadmium and lead suggest that this may be the cause of the above enhancements.When a solution of hydrochloric acid (without lead) is aspirated into the burner immediately after the aspiration of an aqueous lead solution (without hydro-chloric acid) a transient lead atomic-fluorescence signal is observed. It appears that part of the lead in the aqueous solution is absorbed on the inside of the burner or mixing chamber, and that on subsequent aspiration of hydrochloric acid this is desorbed and passes into the flame. This problem is avoided when 0.04 M hydrochloric acid is added to the aqueous standards. The effects of various inorganic ions on the fluorescence signal for 200 pg 1-1 of lead were investigated. Concentrations from 1 to 1000Opgml-l of Na+ K+ Ca2+ Mg2+ Fe2+ Fe3+, F- C1- NO3- and Mo,O,,~- were examined.No effects were observed at interferent concentrations below 100 pg ml-1 except for Ca2+ and Mg2+. Above this value a marked enhancement was observed in nearly every instance if the background correction system was not used. As long as the A typical example that for Ca2+ is illustrated in Fig. 3. .- 4-0 W $ 15 ’0 nr c 10 > .- c -0 5 a 0 Concentration of lead/pg I-’ Fig. 1. Precision of the atomic-fluorescence measurement of lead as a function of concentra-tion 240 STHAPIT et al. DETERMINATION OF Analyst Vol. 108 background correction system was used no interferences were observed from the above ions, provided that the solutions also contained 0.04 M hydrochloric acid as illustrated for Ca2+ in Fig.3. As indicated in a previous paper,l maintenance of the accuracy of the scatter back-ground correction system is achieved in a very simple manner. The scatterings test solution (2% calcium orthophosphate) is aspirated into the flame and the intensity of the xenon arc adjusted until the difference between the signals from the EDL and the xenon arc A and B channels on the photon counter respectively is reduced to zero. The photon counter used has an A-B display mode and this adjustment is therefore made most conveniently on this position. Once balanced the light intensities are stable for long periods but if drift is sus-pected a balance check can be performed in a few seconds without any alteration to the opera-tion of the instrument and this is usually carried out every 30 min.2 300 r I v) v) 4- 5 2200 . 8 0 5 2000; ‘ I 1 1 1 I 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Concentration of HCVM -I v) v) 4- 5 2400 0 0 K 0) . g 2000 1 600 1 10 100 1000 10000 Concentration of Ca/pg mi-’ FiE. 3. Effect of calcium on the atomic-Fig. 2. Effect of hydrochloric acid concentration fluorescence signal for 0.2 pg ml-’ of lead: on the atomic-fluorescence signal for 0.2 mg 1-’ of A without continuum source background lead. correction; and €3 with background correction. Analysis of Blood Samples To test the accuracy of the atomic-fluorescence method for the determination of lead in blood three experimental evaluations were carried out (a) standard additions comparisons, (b) recovery experiments and (c) analysis of quality control samples circulated in the UK by the Heavy Metals Laboratory University of Surrey Two sample treatment processes were investigated as described under Experimental a 1 + 4 dilution with Triton X and a sample treatment with nitric acid.In the former instance a correction factor was required for the reduction in uptake rate on nebulisation of the supernatant. The correction factor was found to be 1.19 and acceptable results were achieved for lead in the same way as reported previously for cadmium.l The treatment with nitric acid has also been investigated for the determination of cadmium in and has provided an improvement in detection limit for that procedure. A major advantage in the present instance is that the uptake rate of the supernatant formed after nitric acid treatment is identical with that for aqueous standards and a correction factor is consequently unnecessary.Standard additions calibration graphs for three blood samples treated by procedure 1 are compared with an aqueous (0.04~ hydrochloric acid) calibration graph in Fig. 4. The measurements for the samples were corrected for the reduction in uptake rate before plotting Fig. 4. The close similarity of the slopes suggests that there are no residual chemical inter-ferences following this sample treatment. Four samples were analysed by direct comparison with an aqueous calibration graph and also by the standard additions procedure. The results are given in Table I11 and show satisfactory agreement. In Table IV the results of some recovery experiments are given which also indicate acceptable accuracy.The 1 + 4 dilution procedures adopted in this work mean that the detection limit achieved for blood samples is five times the aqueous solution detection limit reported above. The detection and determination limits for lead in blood are therefore 60 and 125 pg l-l rcspec February 1983 LEAD IN BLOOD BY FLAME AFS 241 3 000 2 500 c I 2000 500 0 0 50 100 150 200 250 3 Concentration of Pb/pg I-' Fig. 4. Standard additions calibration graphs for three blood samples compared with a calibration graph prepared from lead standards in 0.04 M hydrochloric acid. tively. Therefore although acceptable precision cannot be achieved for very low blood lead levels of (100 pg l-l acceptable results can be obtained for the identification of occupational or excessive environmental exposure to lead.The precision measured as a function of blood lead concentration is shown in Fig. 5 which indicates a slight deterioration in performance com-pared with the measurement of aqueous solutions. Nineteen samples circulated by the Special Advisory Service Quality Control Scheme operated by the Heavy Metals Laboratory, University of Surrey were analysed using both the sample preparation procedures and using aqueous lead solutions as standards. The results for eighteen of the samples are shown in Table V and are compared with the means and standard deviations of results returned from the participating laboratories. One sample (No. 24) could not be analysed with confidence as its value was too near the determination limit of the present atomic-fluorescence method.Samples 21 /27 22/35 23/39 25/36 28/37 29/40 30/38 were apparently duplicates circu-lated without the knowledge of the participating laboratories. Almost all of the atomic-fluorescence results fall within one standard deviation of the mean result. Regression analysis between the group means and the results of the two AFS procedures produced the following data: Procedure 1 y = 1.058 x - 23.99 (Y = 0.9864) Procedure 2 y = 1.057 x - 10.18 (Y = 0.9742) Although the results obtained by using the two procedures appear to be equally acceptable, procedure 1 was generally preferred. Using the treatment with Triton X a clear super-natant liquor was always obtained whereas some samples occasionally produced cloudy solutions after treatment by procedure 2.TABLE I11 ANALYSIS OF FOUR BLOOD SAMPLES BY STANDARD ADDITIONS AND BY DIRECT COMPARISON WITH AQUEOUS LEAD STANDARDS Lead concentration/pg 1-' r Sample No. Standard additions Aqueous standardisation 1 64 62 2 98 92 3 600 610 4 657 67 242 STHAPIT et al. DETERMINATION OF TABLE IV RESULTS OF RECOVERY EXPERIMENTS FOR THE ADDITION OF LEAD Analyst Vol. 108 STANDARD SOLUTIONS TO WHOLE BLOOD SAMPLES Lead added/pg 1-l Lead found/pg 1-l 0 67 50 88 100 156 200 258 400 479 800 858 1200 1171 1600 1765 Recovery yo 75.2 93.4 96.6 102.5 99.0 92.4 105.0 -Conclusions The atomic-fluorescence method described is based on the direct aspiration of diluted blood into an air - acetylene flame and calibration with acidified aqueous standards.Provided that optical background correction for scattered radiation is applied the method appears to be entirely free from matrix interferences. Although the determination limit of 125 pg 1-1 in blood used for much of this work means that the method is unsuitable for the precise measurement of blood levels in much of the general unexposed population the method is potentially useful for large-scale and rapid screening of the blood lead levels of those suspected of being environmentally or industrially exposed to lead. The effect of the recent improve-ment in detection limit to 6 pg 1-1 and consequent improvement in determination limit to about 60 pg 1-1 in blood is currently under investigation.The fact that similar atomic-fluorescence methods have now been described for both lead and cadmium makes this tech-nique highly suitable for surveys of industrial workers. I t has been pointed out previously14 that the routine measurement of cadmium could be extremely important for subjects likely to be treated for lead toxicity. 100 300 500 700 900 1100 1300 Concentration of Pb/vg I-' Fig. 5. Precision of the determination of lead in blood using the 1 + 4 dilution procedure with Triton X as a function of the lead concentration of the original sample (A). For comparison the precision obtained with standards prepared in 0.04 M hydrochloric acid is given (B). Minor changes in the blood preparation procedure are currently being investigated in an attempt to lower the lead determination limit.Use of volumes of sample greater than 1 ml would certainly allow this using the present procedures as a 10-s counting time or a lower dilution factor could be employed. However it is considered that the removal of large February 1983 LEAD I N BLOOD BY FLAME AFS TABLE V ANALYSIS OF QUALITY CONTROL WHOLE BLOOD SAMPLES BY FLAME ATOMIC-FLUORESCENCE SPECTROMETRY 243 QC sample No. (1982 series) 21 22 23 25 27 28 29 30 31 32 33 34 35 36 37 38 39 40 Laboratory mean f SDIpgl-I 820 f 41 441 f 46 600 f 45 306 f 33 816 f 48 248 f 29 385 f 31 495 f 46 691 f 64 646 f 43 472 f 46 772 f 39 440 f 41 302 f 29 240 f 33 455 f 33 640 f 130 382 f 48 Atomic-fluorescence result/pg 1-' - Procedure 1 Procedure 2 853 852 449 41 1 589 605 275 292 858 854 241 243 368 398 458 514 745 718 666 633 489 407 787 862 417 472 259 289 263 256 511 63 1 52 1 564 430 41 8 volumes of blood (up to 10 ml) is probably inconvenient or unacceptable in routine screening programmes and anyway in this instance the current determination limit would probably be considered adequate.One other advantage not shared by many widely used atomic-absorption procedures is the relatively wide linear calibration range. Thus no difficulty is experienced with blood lead concentrations above 600 or 1000 pg 1-I. Attempts to improve the lower limit of determination particularly by the construction of a more intense lead EDL light source are under active study.The authors acknowledge the support of the Scottish Home and Health Department for the purchase of the major items of equipment. They are also grateful for the assistance of Dr. B. G. Cooksey who developed the computer program used in the simplex optimisation studies, and to Mr. G. Henderson for constructing an improved temperature control box for the EDL thermostat. Financial support from the British Council (for P.R.S.) is also gratefully acknowledged. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. References Michel 13. G. Hall $1. I,. Ottaway J. hT. and Fell G. S . Analyst 1979 104 491. Michel I<. G. Hall M. L. Rowland S. A. I<. Sneddon J. Ottaway J . M. and Fell G. S. .4nalyst, hlichel I<. G. Ottaway J. JI. Sneddon J. and Fell G. S. Analyst 1978 103 1204. Michel R. G. Sneddon J. Hunter J. I<. Ottaway J . &I. and Fell G. S. Awalyst 1981 106 288. Stoeppler Jl. Ikandt I<. and Rains T. C. Aiza/-yst 1978 103 714. Hendry J . J{. AT. Fell G. S. and Ottawa? J . JI. .4waiysf to be submittcd. Rlichcl R. G. Coleman J . and Winefordtier J . D. Spcctvochini. Acta Pard B 1978 33 195. Gleason W. S. and Pertel K. Rev. Sci. InstYzmz. 1971 42 1638. Yarbro L. A. and Deming S. N. Anal. Chin?. Acta 1974 73 391. Keldcr J. A. and Mead I<. Coniput. J. 1965 7 308. Rlichel R. G. Ottaway J . M. Sneddon J . and Fell G. S. Analyst 1979 104 687. Trask T. O. and Green I<. R. Anal. Chrnr. 1981 53 320. Ekaneni I:. Rarnard C. Ottamav J . M. and Fell G. S. to be published. Fell G. S . Ottaway J . &I. and Husscin F. 13. R. BY. J. I d . Jlcd. 1977 34 106. 1979 104 505. Received Septeitibev 3vd 1982 Accepted September 21st 198

ISSN:0003-2654    

DOI:10.1039/AN9830800235

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

244 Analyst February 1983 Vol. 108 pp. 244-253 Determination of Lead in Whole Blood by Electrothermal Atomic-absorption Spectrometry Using Graphite Probe Atomisation Shree K. Giri Charles K. Shields David Littlejohn and John M. Ottaway Department of Pure and Applied Chemistry University of Strathclyde Cathedral Street Glasgow G1 1XL Atomisation from a pyrolytic graphite probe placed in a hot and constant-temperature HGA 70 furnace was used for the direct determination of lead in diluted whole blood. Substantial reductions in the classical vapour-phase interference effects by up to 2 yo m/ V magnesium chloride and calcium chloride and 1.5% m/V sodium chloride allowed the use of aqueous standard solutions for analytical calibration. Good agreement with national (UK) mean values was obtained for the analysis of quality control blood samples.The analytical precision is equivalent to that with conventional atomisation but with improved sensitivity. Keywords Lead determination ; whole blood analysis ; electrothermal atomisa-tion ; atomic-absorption spectrometry ; Probe atomisation The development of reliable methods for the determination of lead in whole blood has received considerable attention in clinical laboratories over the past decade. In the UK an inter-laboratory quality control scheme has substantially improved the accuracy of blood lead determinations. As a result there is now widespread confidence in the methods used to monitor occupational exposure to lead. A variety of procedures have been applied to measure blood lead concentrations by atomic spectrometry as indicated by the two other papers in this series.lV2 Atomic-absorption spectrometry with electrothermal atomisation (ETA - AAS) continues to be the most widely used technique.Although significant chemical and spectral interferences from constituents of the blood matrix are encountered with con-ventional atomisation sample preparation and standardisation and measurement procedures have been devised that compensate for these effects. A procedure involving the deproteinisa-tion of blood with nitric acid and the use of pooled blood standards3 has been successfully used at Glasgow Royal Infirmary for some years. Standard additions with or without deproteinisation have also been applied. In both methods interference due to the suppression of the lead atomic-absorption signal by the chloride salts of alkali and alkaline earth metals in blood is only overcome by the standardisation procedure.Accurate results are therefore obtained but at the expense of comparatively long analysis times and/or the risk of contami-nation. Lead atoms are produced at a comparatively low tube temperature similar to that required for the volatilisation of the matrix molecules. As the lead and matrix species co-exist in a low-temperature vapour conversion of analyte atoms to lead halide molecules is thermo-dynamically favourable. There is therefore a decrease in the lead atomic-absorption signal in comparison with an interferent-free solution of the same lead concentration. Many reagents have been used in an attempt to control the interference mechanism.The most popular involves the addition of a matrix modifier to remove the halide interferent prior to the atomisation stage Ammonium and phosphate salts are considered the most satisfactory reagents for matrix modification. Manning and Slavin4 used ammonium nitrate to reduce interferences when measuring lead with molybdenum-coated and pyrolytic graphite-coated graphite tubes. Hinderberger et al .5 combined the suppression-reducing effects of ammonium and phosphate by the addition of 1% of ammonium dihydrogen orthophosphate to blood and urine samples for the determination of lead and cadmium. An alternative approach to the control of halide salt interferences on the determination of lead involves the use of platform atomisation which delays volatilisation of the sample until a later time in the atomisation stage.Lead atoms and matrix molecules then exist in a vapour heated to a higher temperature than that obtained under conventional tube wall The chemical interferences caused by halide salts occur in the vapour phase GIRI SHIELDS LITTLE JOHN AND OTTAWAY 245 atomisation. The platform concept introduced by L’VOV,~ is becoming increasingly popular for atomic-absorption mea~urements.~-~ Another paper1 illustrates the satisfactory appli-cation of platform atomisation to the determination of lead in whole blood. The procedure involves peak-area measurement following deproteinisation of the blood matrix with nitric acid. However platform atomisation alone is not always successful in reducing matrix interferences to an acceptable level.Kaiser et aZ.1° found that platform atomisation was unsatisfactory for the determination of lead in synthetic solutions containing 500 pg ml-l of sodium and potassium. Also May and Brumbaughll reported the failure of a platform and maximum power heating with a Perkin-Elmer HGA 500 atomiser to remove interference effects in the analysis of fish tissue and fish blood digests for lead. Significant improvement in the control of halide interference was achieved only when both groups used platform atomisation and ammonium dihydrogen orthophosphate as a matrix modifier for the analysis of blood5 and fish1’ samples. To date the successful determination of lead in blood and other chloride salt-containing matrices by ETA - AAS requires at least the addition of a matrix modifier to the sample.l2 Other forms of sample preparation are often required even if platform atomisation and signal integration are used.I t would clearly be desirable if blood could be analysed either directly, or following only dilution with distilled water. The analysis time and risk of contamination would then be considerably reduced. Recent papers13-17 have described a novel form of electrothermal atomisation that may prove more successful than platform atomisation in the control of halide interference effects, and so allow the dilution-only analysis of a variety of samples. The concept probe atom-isation involves the introduction of a sample on a metal wire or graphite probe into an already heated furnace tube.The volatilisation and atomisation of the analyte are therefore separated from the heating of the vapour which can be raised to a high and constant temperature prior to introduction of the sample. L’vov and Pelieva13 and Manning et aLf4 reported the use of a tungsten wire probe in this manner. Major limitations of this type of probe were the small amount of sample that could be deposited on the wire and the lifetime of the probe at high furnace temperatures. A graphite probe has proved to be more satisfactory in both respects. Manning and Slavinf5 observed marked reductions in the interference of sodium sulphate and magnesium chloride on aluminium by volatilisation of up to lop1 of sample solution from a small graphite strip introduced into a pre-heated HGA 500 atomiser.The aluminium sensitivity was also improved in comparison with tube wall atomisation. In our laboratory probe atomisation in Perkin-Elmer HGA 70 and HGA 72 furnace atomisers has resulted in improved ETA - AES sensitivity for a number of elements owing to the rapid atomisation of the analyte species into the hot furnace v a p ~ u r . l ~ ~ ~ ~ Preliminary studies indicated that no interference on lead was observed for up to 2% m/V magnesium chloride and calcium chloride.16 Volatilisation in the high-temperature furnace environment allowed greater dissociation of the lead chloride molecules than could be achieved following tube wall or platform atomisation. In this paper the application of probe atomisation to the determina-tion of lead in whole blood is described.Dilution of the sample with distilled water is the only sample treatment required. The removal of interference effects permits calibration with aqueous lead standard solutions and the use of peak-height measurements. Experimental A Perkin-Elmer 306 atomic-absorption spectrometer fitted with a deuterium arc lamp for background correction was used in conjunction with a modified Perkin-Elmer HGA 70 atomiser. Atomic-absorption signals at 283.3 nm were recorded on a Servoscribe 541.20 flat-bed recorder. Probe Atomisation A 9 mm diameter hole was cut in the front of the HGA 70 furnace housing to allow the introduction of a pyrolytic graphite probe through the hole in the furnace housing and into the graphite tube. A 5 mm long by 2 mm wide slot was cut in the graphite tube at th 246 GIRI et al.DETERMINATION OF PB IN WHOLE BLOOD BY Analyst VoZ. 108 centre (see Fig. 1). Tubes made of pure pyrolytic graphite were used,fs with dimensions similar to those of conventional HGA 70/72 tubes (i.e. 9mm iid. 53 mm long). The entrance slot was created with a small file by enlarging the conventional injection hole. Probes were also prepared from pure pyrolytic graphite and were typically 50 mm long by 2-3 mm wide and 300-400 pm thick with a 4 x 3 mm sample head at one end capable of holding up to 50 pl of sample solution. The design of the probe is also illustrated in Fig. 1. If graphite with a thickness greater than 400 pm was used to make a probe the sample head was filed to a thickness of approximately 400 pm. The experimental arrangement is shown in Fig.2. The probe was attached to a Perkin-Elmer HGA 72 solid-sampling tool held by a lens holder on an optical bench mount. The mount was positioned on a 25-cm optical bench placed at right-angles to the furnace workhead. The probe was manually moved into and out of the furnace by movement of the probe mount along the optical bench. A second optical mount was situated closer to the furnace to act as a positioning "stop" for the probe. This mount ensured reproducible placement of the probe in the atomiser tube and prevented the probe head from touching or breaking the tube wall opposite the entrance slot. With this arrange-ment only the probe sample head entered the tube and was heated. The height of the probe was adjusted so that the probe head was close to but not touching the bottom surface of the atomiser tube (Fig.3). In this position it does not interrupt the hollow-cathode and deuterium arc light beams and radiation emitted by the heated probe does not affect signal measurement during the atomisation sequence. I t should be noted that other types of spectrometers may need additional optical masking at the monochromator entrance slit to reduce the effective slit height in order to prevent emission from the hot probe entering the monochromator. A short graphite tube 9 mm 0.d. and about 30 mm long was used as an entrance tube and was placed in the hole in the furnace housing at right-angles to the atomiser tube. The entrance tube was used as a drying zone for evaporation of the sample droplet and was positioned close to but not touching the atomiser tube.It also reduced the rate of loss of inert gas from inside the workhead during atomisation and was found to have an enhancing effect on the magnitude of atomic-absorption signals. The modification of the atomiser tube and the preparation of probes were greatly assisted by the use of pure pyrolytic graphite.l* Standard graphite is too brittle and was not as easily shaped and filed as pyrolytic material particularly in the formation of the thin probe head. In addition it was observed that the pyrolytic tubes gave a higher temperature at any voltage setting compared to standard tubes. A substantial increase in lifetime was also obtained by use of pyrolytic graphite probes and tubes. The same probe could be used for a number of weeks.The lifetime of the tube was mainly restricted by damage at the ends, where contact with the conduction cones was made. In our experience this is a routine problem with the HGA 70/72 atomiser design. The procedure used to carry out the drying ashing and atomisation steps of the furnace pro-gramme with the graphite probe was as follows. A sample aliquot of 10 20 or 5Opl was deposited on the probe head with a micropipette and the probe was moved into the drying zone while the atomiser tube was heated to an appropriate temperature. For aqueous solutions a temperature of 450 "C was suitable but when an ashing step is required as with blood samples the atomiser was set to the same temperature as that required for ashing for convenience. For blood an ashing temperature of 600-800 "C was used and during the drying stage this temperature was used with the probe held slightly further away from the atomiser tube than for the normal aqueous solution drying procedure.In both instances, drying was generally completed in times comparable to a conventional furnace operation. When drying was complete the probe was moved into the atomiser tube itself for ashing at 600-800 "C. The probe was then removed from the furnace assembly and the atomiser tube heated to the required atomisation temperature. A period of 8-10 s was required to ensure an almost constant tube temperature with the HGA 70 power supply. Accordingly a total atomisation time of 13 s was selected and the probe was introduced directly into the hot atomiser tube after exactly 10 s.The inert gas flow was not stopped until a few seconds before introduction of the probe in order to enhance tube lifetimes. Gas stop during measurement maximises the signal sensitivity. Lack of contact between the atomiser tube and the probe ensures that the probe is heated principally by black-body radiation from the tube wall. The furnace programmes used for aqueous and blood samples are given in Table I Fig. 1. Pyrolytic graphite tube with entrance slot and probe used in these studies. Fig. 2. Experimental arrangement for the introduction of the probe into the HGA 70 furnace atomiser head. Fig. 3. Position of probe inside the atomiser tube during the ashing and atomisation stages. [to face page 24 February 1983 ELECTROTHERMAL AAS USING GRAPHITE PROBE ATOMISATION TABLE I 247 FURNACE PROGRAMMES FOR PROBE ATOMISATION WITH AN HGA 70 FURNACE Sample volume .. Dry: Temperature . . Time Ash : Temperature . . Temperature . . Time . Atomise : Time . . Gas mode . . Background correction . . Aqueous programme Diluted blood programme 10-50 p1 Probe positioned outside tube heated to 450 "C 30 s for 20 pl 10 or 20 pl Probe positioned outside tube heated to 600-800 "C 40-60 s for 10 p1; 60-90 s for 20 pl Not required Not required Probe into tube a t 600-800 "C 45 s 1900-2400°C 1900 "C Total duration of 13 s. and probe introduced after 10 s Gas stop for final 5 s of atomisation stage Furnace heated to equilibrium temperature - Deuterium arc Preparation of Whole Blood Samples The blood samples analysed were mainly quality control material sent to Glasgow Royal Infirmary as part of the National Quality Control scheme supervised by the Heavy Metals Laboratory University of Surrey.A volume of 200 pl of whole blood was transferred with a micropipette into a small plastic ampoule containing 800 p1 of distilled water. Blood left in the pipette tip was washed out using the water - blood mixture to ensure quantitative transfer. The capped ampoule was then shaken by hand for a few seconds and the diluted blood was analysed by injecting 10 or 20 p1 on to the sample head of the probe. Aqueous lead standard solutions in the range 10-200 pg 1-1 were prepared by dilution of a stock lead nitrate solution with distilled water.Diluted blood and aqueous standard solutions were analysed using the diluted blood furnace programme given in Table I. Interference Study Lead solutions containing 0.1-2y0 m/V calcium chloride and magnesium chloride and 0.2-1 .ti yo m/ V sodium chloride were prepared with the appropriate analytical-reagent grade salts. Distilled water was used for all solutions. Lead atomic-absorption signals were measured with the aqueous and diluted blood furnace programmes given in Table I. Temperature Measurements 1100). assuming an emissivity of 1 for pyrolytic graphite. Probe and tube wall temperatures were measured with an Ircon optical pyrometer (Series Temperatures in the range 1300-2500 "C were recorded on a chart recorder, Results and Discussion Probe Atomisation for Lead A small viewing hole was drilled a few millimetres above the entry slot of one of the pyrolytic graphite tubes to measure the heating rate of the sample head when the probe was placed inside the hot atomiser tube.The optical pyrometer was focused through the viewing hole on to either the tube or probe surface. Temperature measurements obtained when the atomiser was heated to 2300 "C are given in Fig. 4. The probe was apparently heated to 2 100-2 200 "C within 0.5 s and for the probe dimensions given previously the final equilibrium temperature of the probe was only 50-100 "C below that of the tube wall. The probe heating rate measured by this method is subject to error as the pyrometer does not distinguish between radiation emitted by the probe and radiation from the tube wall scattered or reflected by the probe.However it is likely that the small dimensions of the atomisation surface (i.e. 4 mm x 3 mm x 400 pm) ensure that the probe head is rapidly heated to temperatures close to the vapour-phase temperature in the tube. This view is supported b 248 2300 y 2200 2 2100 ; 2000 E z !-1900 GIRI et aZ. DETERMINATION OF PB IN WHOLE BLOOD BY Analyst VoZ. 108 A -a I ' B e - I I I --Probe in -4 I 3 2 1 Probe Time/s in Fig. 5. Probe atomisation of 2 0 - 4 aliquots of (A) 100pgl-l of lead (B) 100 pg 1-1 of lead + 1.5% NaCl and (C) 1.5% NaCl blank. Measured at 283.3 nm with background correc-tion and an atomisation temperature of 2 300 "C.oL1500 I709 1900 2100 2300 2500 ' 0 TemperaturePC Fig. 6. Effect of tube temperature on the sensitivity and precision of probe atomisa-tion of 20 p1 of 100 pg 1-l lead in aqueous solution. The effect of sample volume on signal magnitude and precision is shown in Table 11. Volumes of 10 20 and 50 p1 were injected on to the probe and lead solution concentrations were chosen to give a total sample mass of 2 ng of lead in each instance. There was not February 1983 ELECTROTHERMAL AAS USING GRAPHITE PROBE ATOMISATION TABLE I1 1.6 8 1.2 e a 0.8 249 ---EFFECT OF SAMPLE VOLUME ON LEAD ABSORBANCE WITH PROBE ATOMISATION Lead concentration/ Relative standard Volume/ pi Pg 1-1 Mean absorbance deviation yo 10 200 0.33 2.1 20 100 0.37 1 .Q 60 40 0.32 2.6 great difference in the magnitude of the absorbance signals for each combination or in the precision of measurement.A volume of 20 pl gave slightly better sensitivity and precision and was used for most of the experiments described in this paper. 2.0 I 0’4 0 ; 100 200 300 400 500 Pb concentration/pg I-’ Fig. 7. Calibration graph for the determination of lead in aqueous solution using probe atomisation (283.3 nm; 20 pl). If the probe heating rate is much faster than conventional tube wall or platform atomisa-tion as is suspected it is possible that the response time of many chart recorders and some spectrometers may be too slow. This would undoubtedly cause the curvature of calibration graphs to be worse than normal and may also result in a decrease in sensitivity.We have already noted that the lead sensitivity obtained by probe atomisation in the HGA 70 furnace is equivalent to that achieved under optimum conditions with a Perkin-Elmer HGA 500 atomiser.16 The lead calibration graph for probe atomisation of aqueous lead solutions is shown in Fig. 7 and is found to be linear to at least 0.8 absorbance unit. The precision of signals measured in the construction of the lead calibration graph is illustrated in Fig. 8 in the form of a Ringbom plot of relative standard deviation against concentration and d o 100 200 300 400 500 Concentration/pg I-’ 0 0.5 1 .o 1.5 2.0 Absorbance Fig. 8. Ringbom plotsxs with respect to both absor-bance (A) and concentration (B) for the determination of aqueous lead solution using probe atomisation 250 GIRI et al.DETERMINATION OF PB IN WHOLE BLOOD BY Analyst VoZ. 108 ab~0rbance.l~ The deviation of the concentration plot from that of absorbance at 200 pg 1-1 of lead is due to the onset of curvature in the graph of absorbance against concentration. A precision of better than 5% in concentration terms is achieved for 25-35Opg1-l of lead with a 20-4 injection volume. A substantial non-linear portion of the calibration graph could therefore be satisfactorily used for lead analysis if required. Interference Studies Relevant to Blood Lead Analysis Three inorganic components of blood known to cause interference on lead atomic-absorption measurements were studied viz. magnesium chloride calcium chloride and sodium chloride. The diluted blood furnace programme given in Table I was used to obtain lead atomic-absorption signals with the exception that a tube temperature of 2100 "C was chosen for atomisation in the magnesium chloride and calcium chloride experiments.There was slight interference by magnesium chloride on 100 pg 1-1 of lead at 1670 "C but the effect was removed at 2100 "C for up to 2% m/V magnesium chloride (Fig. 9). No interference was observed for up to 2% m/V calcium chloride at the same temperature as shown in Fig. 10. Typical lead signals are given in Fig. 11 which illustrates that no blank signals were measured for the magnesium chloride and calcium chloride solutions. 0.30 0 0 B 0 0.5 1 .o 1.5 2.0 MgCI2 concentration % m/V Fig. 9. Effect of magnesium chloride on the determination of lead (100 pgl-l; 10 pl) using probe atomisation a t atomisa-tion temperatures of (A) 1670 "C and (R) 2 100 "C.Sodium chloride exhibits a different interference effect on lead in that it is spectral and not chemical. Blank signals were measured for sodium chloride concentrations greater than 0.5% ?n/V even with deuterium arc background correction which could compensate for 0.6-0.7 A of background. A study of the time-resolved absorbance signals of lead sodium chloride and lead plus sodium chloride showed that temporal resolution of the lead and sodium chloride signals could be achieved with probe atomisation (Fig. 5). rhere was oniy a slight contribution by sodium chloride to the lead peak-height signal eveii at concentra-tions of 1.5% m/V sodium chloride.The presence of sodium chloride in a lead solution caused a change in the precision of measurement. In general both the magnitude of the sodium chloride background absorption and the value of the lead relative standard deviation (RSD) increased as the furnace temperature was increased. Fig. 12 shows the change in the RSD of signals for 100 pg 1-1 of lead in ly0 mjV sodium chloride. At 2000 "C the R.SD was loyo whereas at 1700 "C it was reduced to almost lye as good as for a sodium chloride-free solution. - _ nu a 0.35 I I 0 0.5 1 .o 1.5 2.0 CaCI2 concentration YO mlV Fig. 10. Effect of calcium chloride on the determination of lead (100 pg 1-l; 20 pl) using probe atomisation a t an atoniisa-tion temperature of 2 100 "C February 1983 ELECTROTHERMAL AAS USING GRAPHITE PROBE ATOMISATION )+ , 251 CaCI2 blank _.1 C) e) Fig.11. Typical chart recordings (283.3 nm) for the probe atomisation of 100 pg 1-1 of lead at (u) and (b) 20 p1 of 100 p g 1-l of lead in (a) the absence and (b) the presence of 1 % CaCl,, (c)-(f) 10 p1 of 100 pg I-' of lead in (c) the absence and (e) the 2 100 "C. and blanks for water and 1 % CaCI,. presence of 1% MgCl, and blanks for ( d ) water and (f) 1% MgCl,. Whole Blood Analysis The concentration of halide salts in whole blood is much lower than the levels investigated during the interference study. I t should therefore be possible to analyse whole blood directly without any sample treatment provided that the background correction system can correct for background signals generated during atomisation by the organic and inorganic components of the whole blood matrix.Satisfactory signals were obtained when 10-p1 volumes of whole blood were injected on to the probe and analysed directly. However a procedure involving a five-fold dilution with water was adopted for the analysis of most quality control blood samples for two reasons. A small volume of viscous whole blood was always retained by the pipette tip with direct sampling. I t was easier and more accurate to pipette the diluted blood solution. The concentration of lead in the blood samples provided was generally greater than 100 pg 1-1 and was often in the 250-1 000 pg 1-1 range. Five-fold dilution of blood samples reduces the measurement concentration to the optimum range of most atomic-absorption spectrometers and allows the use of a 20-4 sample volume.1 700 1800 1 900 2000 Tern peratu re/"C Fig. 12. Effect of atomisa-tion temperature on the relative standard deviation of the measurement of 20 p1 of 100 pg 1-' of lead in the presence of 1% m/V NaC1 252 GIRI et aZ. DETERMINATION OF PB IN WHOLE BLOOD BY Analyst VoZ. 108 The results of the preliminary investigations indicated that the optimum furnace tempera-ture for the determination of lead in a matrix containing magnesium chloride calcium chloride and sodium chloride was 1900 "C. The sensitivity and precision of lead signals were satisfactory at this compromise temperature. The diluted blood furnace programme given in Table I was therefore used for the analysis of diluted blood samples and aqueous lead standard solutions.The recoveries of lead added to pooled whole blood with an intrinsic lead concentration of 145 pg 1-1 are shown in Table 111. Recoveries were satisfactory in the range 91.2-101% with the exception of the initial sample where a 116% recovery was obtained on addition of 50 pg 1-1 of lead. TABLE 111 RECOVERY OF LEAD ADDED TO POOLED BLOOD (145 p,g I-') Five-fold dilution of blood with water; aqueous standards. Known increase in concentrationlpg 1-1 50 200 300 400 600 800 Measured increase in concentrationlpg 1-1 Recovery yo 58.0 116 189.4 94.7 285.4 95.1 373.7 93.4 606.0 101.0 729.8 91.2 The probe atomisation method described in this paper was used by an experienced and a novice analyst to determine the lead concentration of a number of quality control blood samples.The results obtained by both analysts are given in Table IV together with the national mean value and standard deviation for each specimen. Good agreement was achieved between the probe and national results for all samples as is shown in Fig. 13. The error bars illustrate the concentration range equivalent to one standard deviation for the national mean results. Both analysts had one erroneous result on different samples which was thought to be due to incorrect sampling of the appropriate whole blood specimens. The precision of the method for whole blood analysis is illustrated in Fig. 14 by a graph of relative standard deviation of concentration against actual blood lead concentration. Precisions of better than 5% were obtained for samples in the range 250-900+ pg l-l which is the range of interest in monitoring occupational exposure to lead.The actual lead con-centrations measured in the diluted blood samples were actually five times lower than the values shown in Fig. 14. If better precision and/or sensitivity is required a t lead concentra-tions of less than 250 pg 1-1 this could probably be achieved by the use of a lower dilution factor or by direct analysis of the whole blood sample. TABLE IV ANALYSIS OF QUALITY CONTROL WHOLE BLOOD SAMPLES BY PROBE ATOMISATION Quality control National 21 820 f 40 22 441 f 46 23 600 f 44 24 130 f 27 25 306 f 33 27 816 f 47 28 248 f 29 29 385 f 31 31 691 f 64 32 646 f 43 number meanlpg 1-1 30 495 f 45 Lead concentration by probe AAS*/pg 1-1 I Experienced analyst Novice analyst 832 475 ';g) 33 1 224 408 510 725 689 -830 420 560 105 300 770 230 370 455 * Five-fold dilution of blood with water; aqueous standards February 1983 ELECTROTHERMAL AAS USING GRAPHITE PROBE ATOMISATION 253 850 800 700 600 5 500 400 .o 300 -ol 3- .al -m C 4-0 = 200 100 0 100 200 300 400 500 600 700 800 Strathclyde probe resultdyg I-’ 0 0 0 100 200 300 400 500 600 700 800 900 Actual Pb concentration/pg I-’ Fig. 14. Relative standard deviation for the measurement of different concentrations of lead in whole blood samples. Fig. 13. Comparison of lead concentration for quality control samples obtained by probe atomic-absorption spectrometry and the national mean values for the samples analysed.The successful development of a simple method for the direct determination of lead in whole blood supports the view that probe atomisation could be of considerable importance in the continuing evolution of electrothermal atomisation. Rapid vaporisation from the probe surface into a hot furnace environment has improved the atomic-absorption sensitivity for lead with the HGA 70/72 furnace and the substantial reduction in vapour-phase inter-ference effects has allowed the use of aqueous standards for blood lead analysis. The probe concept is an extremely simple and convenient form of atomisation and is undergoing further development at the University of Strathclyde in both atomic-absorption and atomic-emission studies.This work was made possible by the provision of pure yyrolytic graphite tubes by Pye Unicam Ltd. and PhiliDs Research Laboratories. Financial sumort from the Pve Founda-t ion 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. (for D.L.) and the ‘British Council (for S.K.G.) is gratefully’ikknowledged. References Hendry J. €3. M. Fcll G. S. and Ottaway J . M. Analyst to be submitted. Sthapit P. It. Vcll G. S. and Ottaway J . M. Analyst 1983 108 235. Stoeppler M. Brandt K. and Rains T. C. Analyst 1978 103 714. Manning D. C. and Slavin W. Anal. Chem. 1978 50 1234. Hinderberger E. J. Kaiser J . L. and Koirtyohann S. R. At. Spectrosc. 1981 2 1 . L’vov B. V. Spectrochim. Acta Part B 1977 33 153. Slavin W. Manning D. C. and Carnrick G. R. At. Spectrosc. 1981 2 137. Carnrick G. R. Slavin W. and Manning D. C. Anal. Chem. 1981 53 1866. Slavin W. and Manning D. C . Spectrochim. Acta Part B 1980 35 701. Kaiser M. L. Koirtyohann S . R. Hinderberger E. J. and Taylor H. E. Spectrochim. Acta Part May T. W. and Brumbaugh W. G. Anal. Chem. 1982 54 1032. Halliday M. C. Houghton C. and Ottaway J. M. Anal. Chim. Acta 1980 119 67. L’vov B. V. and Pelieva L. A. Zh. Anal. Khim. 1978 33 1225. Manning D. C. Slavin W. and Myers S. Anal. Chem. 1978 51 2375. Manning D. C. and Slavin W. Paper presented a t Pittsburgh Conference on Analytical Chemistry, Giri S. K. Littlejohn D. and Ottaway J. M. Analyst 1982 107 1095. Marshall J. Giri S. K. Littlejohn D. and Ottaway J. M. Anal. Chim. Acta in the press. Lersmacher B. Ger. Offen. 2 949 275 (Cl. GOINZI/03) June 25th 1981. Van Dalen H. P. J. and De Galan L. Analyst 1981 106 695. B 1981 36 773. Atlantic City NJ 1982. Received September lsl 1982 Accepted September 21st. 198

ISSN:0003-2654    

DOI:10.1039/AN9830800244

出版商:RSC    

年代:1983

数据来源: RSC

摘要:

254 Analyst Febmary 1983 Vol. 108 pp. 254-260 Analysis of Used Lubricating Oils for Wear Metals by Wavelength Dispersive X-ray Fluorescence Spectroscopy Edward Searle and Christopher M. Thompson Research Laboratory London Transport Executive 566 Chiswick High Road London W4 5RR Analysis of used lubricating oils for wear metals by X-ray fluorescence spectro-scopy has always been difficult when the determination of many elements has been required. Problems have been mainly caused by the nature of the sample, selection of standards and presentation of samples and standards to the instrument in a suitable form. The method described largely overconies these problems is simple and convenient to use and is suitable for a wide range of lubricating oils. Repeatability is good and readily available standards can be used for calibration purposes.A correlation progratnnie has been carried out using flame atomic-absorption and direct-reading emission spectroscopy on chromium tin lead copper and iron. Keywords Wear metals ; lubricating oils ; wax medium ; X-ray jluorescence spectroscopy Identification and determination of wear-metal particles in used lubricating oil is of value in assessing wear in various parts of the engine and/or transmission system of a vehicle. The analysis can provide data that are of major assistance in the efficient operation and mainten-ance of these mechanical units. Analysis of wear-metal particles in used lubricating oils has been carried out by flame at omic-absorption spec t roscopy,1-6 electrothermal at omic-absorp t ion spectroscopy optical emission spectro~copy,l~~$~ inductively coupled plasma emission spectro~copy,~ scanning electron microscopy - X-ray analysis,* and X-ray fluorescence spectroscopy.8 The results obtained by the use of these techniques have not always been reliable mainly owing to the non-homogeneous nature of the sample and to the fact that instrumental response has been shown to be dependent upon the size of the particles present in the ~ a m p l e .l - ~ ? ~ $ ~ * ~ Various procedures have been devised to overcome these difficulties but these have not always been convenient to carry out when the determination of several elements in a large number of samples was req~ired.2,~-6,9 It was shown that the results of analyses carried out directly on oil samples by X-ray fluorescence spectroscopy gave inconsistent results and that the variations depended on the time that elapsed between preparation of the sample and carrying out the analysis.This was caused by sedimentation of the particulate matter in the sample. Sedimentation has been prevented by converting the oil samples into a solid form by mixing with a suitable wax.l0 This paper describes a method for converting oil samples into a solid form by mixing with a suitable wax and a procedure for the determination of wear metals in used lubricating oils that utilises this method to overcome the problems described above. Experimental Apparatus The following apparatus and conditions were used Siemens SRS 200 X-ray fluorescence spectrometer fitted with a 3 kU' chromium anode tube helium flush sample-rotation facilities controlled by a PDP 11-34 computer; sample cups fitted with graphite bases containing apertures of 34 mm diameter; argon - methane (90 + 10) carrier gas for the gas flow propor-tional counter; helium flush gas 99% ; PTFE or PTFE-lined polyproprylene mould assembly; box shaker for oil samples; water-bath heated to 65-70 "C; electric hot-plate; and top-pan balance SEARLE AND THOMPSON 255 Reagents Powdered polyethylene wax melting-point 117 "C density 0.94 g ~ m - ~ at 20 "C (Wilkins Campbell & Co.Ltd. Type 6450) and Conostan metallo-organic standards to cover the range 10-500 p.p.m. (Continental Oil Co. Ponca City OK USA) were used. Proced ure Heat the sample at 65-70 "C in a water-bath for 1 h and then shake the heated sample for 2 min in a box shaker or other suitable shaker.Weigh to the nearest 0.01 g approximately 10 g of the sample and record the mass. Add an equal mass of the powdered polyethylene wax; stir and mix well. Heat the mixture on an electric hot-plate in a fume hood until the mixture melts. Stir well and pour quickly into a PTFE- or PTFE-lined polypropylene mould, 10 mm deep x 50 mm diameter pre-heated to 100 O C until the mixture is level with the top of the mould. Allow to cool to room temperature and remove the sample from the mould after standing for 1.5 h. Analyse the sample in a helium atmosphere using a chromium anode X-ray tube operated at 55 kV and 50 mA. Rotate the sample during analysis. The following ele-ments have been determined tin lead molybdenum nickel titanium manganese iron copper and chromium.The instrumental conditions employed for the determination of these ele-ments are shown in Table I. The oil sample - wax disc can be continuously exposed to X-rays for periods of up to 16 h without excessive decomposition. Element Tin . . . . . . Lead . . Molybdenum . . Nickel . . Titanium . . . . Manganase * . Iron . . . . Copper . . Chromium . . . . TABLE I INSTRUMENTAL CONDITIONS Titanium Crystal X-ray line primary-beam filter Germanium Germanium Lithium fluoride 100 Germanium Lithium fluoride 100 Lithium fluoride 100 Lithium fluoride 100 Germanium Lithium fluoride 100 La La Ka Ka K a Ka Ka Ka Ku -* -+ t + + + + + -Collimator 0.15" 0.15" 0.15" 0.15" 0.15" 0.15" 0.15' 0.15" 0.15" * Indicates no primary-beam filter in use.t Indicates primary-beam filter in use. Calibration Calibration was carried out using Conostan oil standards covering the range 10-500 p.p.m. Calibration samples were prepared by mixing the appropriate oil standard with the wax and proceeding as described above. Results and Discussion Preliminary Studies Wear-metal particles are usually not distributed evenly throughout the used-oil samples and the distribution varies on standing. The larger particles settle towards the bottom of the container faster than those which are smaller. Some used-oil samples are also viscous and contain significant proportions of unburnt carbon diesel oil sludge and water from the engine coolant.These factors cause problems when a representative sample is required for analysis. The procedure employed when taking a sample from the engine or transmission system is, therefore very important and it is desirable to take the sample when the unit is in operation or immediately afterwards. The choice of standards for calibration is limited to readily available homogeneous oil based materials. Heterogeneous standards are not readily available and are not easy to prepare when determination of several elements is required. Preparation of Well Mixed Samples It was essential to reduce the viscosity of the oil samples in order to achieve adequate mixing of the wear metal particles in the oil. This was done by heating the samples in a water-bath a 256 SEARLE AND THOMPSON ANALYSIS OF USED LUBRICATING Analyst VoZ.108 65-70 "C for 1 h followed by vigorous shaking of the heated samples for 2 min either by hand or in an electrically driven box shaker. Use of the box shaker was more convenient and reliable. Sedimentation of Wear-metal Particles in Oil Samples Sedimentation of wear-metal particles resulted in an apparent increase in the concentration of the metals over a period of time. It was therefore important to overcome this effect in order to obtain reliable results. Sedimentation was prevented from taking place by mixing the oil sample with a suitable wax to form a solid disc.10 Several types of wax were investigated but the most suitable was a polyethylene wax melting-point 117 O C density 0.94 g cm-3 at 20 "C type 6450 supplied by Wilkins Campbell & Co.Ltd. This wax was free from metal contamination and mixed well with oil. When the wax was mixed with oil melted and allowed to cool to room temperature, a solid mass was rapidly formed. Various compositions of oil and wax were examined to determine the optimum oil to wax ratio with respect to sample dilution reduction in sensitivity and resistance to degradation by X-rays during long exposure periods. The optimum ratio was found to be one part of oil to one part of wax. Particle Size Effects Variations in the particle size of the wear metals can exert a considerable influence upon the magnitude of the spectra of the elements p r e ~ e n t . ~ ~ 3 ~ ~ ~ 6 * ~ It has been shown that the instru-mental response produced by some materials whose particle size was greater than about 63 pm, fell to zero when attempts were made to determine these materials by atomic-absorption spectroscopy or by inductively coupled plasma emission spectroscopy .5 It has also been shown that when powders were analysed by X-ray fluorescence spectroscopy the response from the component being determined increased as the particle size decreased and eventually became almost constant when the particle size had been sufficiently r e d u ~ e d .~ This problem was investigated by milling well mixed portions of used oil samples for 30 min in a micronising mill prior to treatment with the wax and comparing the results obtained from portions of the same samples that had not been milled.It has not been possible to determine whether or not the milling process produced a reduction in the size of the particles present in the oil and detailed effects of particle-size variations have not yet been established but will be investigated at a later date The results are shown in Table 11. TABLE I1 EFFECT OF MILLING ON THE ANALYSIS OF USED OIL SAMPLES Lead p.p.m. Copper p.p.m. Tin p.p.m. Chromium p.p.m. Iron p.p.m. +-7r---h--7r-h- -7-r- 7 Sample Urimillcd Millet1 Unmilled Millctl Unrnillcd Millcd Uninilled Millcd liiimilled Milled 1 31 33 101 100 6 14 7 7 254 253 2 14 10 39 38 6 9 4 2 197 123 3 20 23 60 49 1 3 2 6 99 95 Determination of Elements with Low Atomic Numbers X-rays emitted from elements with atomic numbers of less than 22 have low energies and these X-rays can be readily absorbed by the sample matrix elements with high atomic numbers and the end window of the gas flow proportional counter.This problem was particularly severe with silicon and aluminium. During this work it was found that the standard deviations obtained for these elements were much larger than those obtained for other elements investigated. Analytical results obtained for these elements at low concentra-tions were found to be too unreliable for practical use. Repeatability paring five discs. A sample of used diesel engine oil was analysed on separate days by four analysts each pre-Analysis of used transmission oil The results are shown in Table 111 February 1983 OILS FOR WEAR METALS BY WAVELENGTH DISPERSIVE XRF TABLE I11 REPEATABILITY 257 Mean Element determinations p.p.m.Number of concentration, Lead . . . . . . 20 24 Copper . . 20 93 Tin . . . . 20 4 Iron . . . . 20 125 Chromium . . 20 19 Relative Standard standard deviation deviation, p.p.m. % 0.4 1.6 1.4 1.5 0.2 5.0 2.3 1.8 2.7 14.7 samples afforded similar results. Analysis of the reverse sides of the discs incorporating these samples produced identical results indicating that the wear-metal particles were evenly distributed throughout the discs. With tin the concentration in the sample was below that of the lowest concentration in the standards used for calibration ( i e . 10 p.p.m.). With chromium the net count rate at the concentration present in the sample was low. This accounted for the higher relative standard deviations obtained for these elements when compared with the figures obtained for the other elements.Matrix and Inter-element Effects A number of used lubricating oil samples were analysed utilising calibrations derived from dissolved standards of different types of lubricating oil. A number of samples prepared by diluting standards in unused additive-free lubricating oil were also analysed under the same conditions. The base oil used in the additive-free oil samples was the same as that used in the used additive-containing oil samples. The results and details of the standards and additives contained in the oils used for calibration and analysis are given in Tables IV and V. The results obtained using standard matrices 2 and 3 indicated that some inter-element effects derived from the wear-metal particles themselves were exhibited by tin.TABLE IV COMPOSITION OF THE STANDARDS AND MATRICES USED FOR CALIBRATION PURPOSES Standard matrix number Base oil 1 Conostan 2 Monograde 3 Monograde 4 Multigrade Additives present p.p.m. None Calcium 3800 Zinc 1000 Phosphorus 1000 Calcium 3800 Zinc 1000 Phosphorus 1000 Magnesium 2 400 Calcium 640 Zinc 1000 Phosphorus 1000 Other standard Ag Al B Ba Cd, Mg Mo Na Ni, P Si Ti V Zn, plus Ca a t 5 times the level of the other elements elements present Composition of standards All standard elements of equal concentration in each standard to cover the range 500-10 p.p.m. Silicon Silicon 2 p.p.m. Silicon 5 p.p.m.Silicon 10 p.p.m. Silicon 20 p.p.m. Silicon 50 p.p.m. Silicon 100 p.p.m. As in standard matrix 3 above All standard elements of equal concentration in each standard to cover the range 500-10 p.p.m. Fe = 10 Cu = 4, Sn = Pb = Cr = 2 p.p.m. Fe = 25 Cu = 10, Sn = Pb = Cr = 5 p.p.m. Fe = 50 Cu = 20, Sn = Pb = Cr = 10 p.p.m. Fe = 100 Cu = 40, Sn = Pb = Cr = 20 p.p.m. Fe = 250 Cu = 100, Sn = Pb = Cr = 50 p.p.m. Fe = 500 Cu = 200, Sn = Pb = Cr = 100 p.p.m. As in standard matrix 3 abov TABLE V ANALYSIS OF USED LUBRICATING OILS CONTAINING ADDITIVES AND UNUSED ADDITIVE-FREE LUBRICATING BASE OIL UTILISING CALIBRATIONS DERIVED FROM STANDARDS DISSOLVED I N VARIOUS Sample 1* 2 7 4 5 6 it s 9 10 11 12 Lead p.p.m. : standard matrix number r I b 1 2 3 4 13 7 4 4 7 4 0 0 1 2 7 5 5 38 34 41 36 45 38 46 40 20 16 15 14 10 4 0 O(2) 12 7 1 2(3) 15 10 12 l l ( 1 0 ) 27 23 21 19(20) 54 52 57 49(50) 100 99 113 96(100) Copper p.p.m.: standard matrix number c -7 1 2 3 21 17 18 15 9 5 5 2 27 23 24 20 21 30 33 28 20 26 28 23 60 57 67 59 13 12 12 9(4) 10 9 16 12(10) 20 21 30 25(20) 38 42 52 45(40) 93 103 117 104(100) 195 206 226 202(200) Iron p.p.m. : standard matrix number 1 2 3 18 18 19 31 31 31 63 65 71 372 342 354 276 268 287 99 100 104 10 20 16 52 63 62 61 72 104 103 113 114 274 264 254 638 536 543 4 24 43 102 536 433 154 19(10) 90(25) 104(50) 168(100) 413( 200) 823(500) Chromium, standard number -h-- 1 2 4 7 0 0 1 4 49 54 38 43 2 8 0 3 2 7 8 13 18 23 42 49 95 102 104 Samples 1-6 inclusive were used lubricating oil samples and contained the following additives calcium 3000 p.p.m zinc 1000 p.p.m.phosphorus t The figures in parentheses for samples 7-12 lnclusive are the actual concentrations of the elements present in the samGles. Samples 7-12 were prepared for samples 1-6 inclusive February 1983 OILS FOR WEAR METALS BY WAVELENGTH DISPERSIVE XRF 259 This has largely been overcome by careful use of the available computer software. The above results also indicated that the additives present in the various base oils used for calibration purposes gave rise to inter-element effects especially in the case of iron. The high results obtained when standard matrix 4 was used for calibration purposes appear to be caused by the presence of magnesium in the additive used in this base oil.In this work inter-element effects have been minimised by using standards dissolved in the same base oil containing similar concentrations of additives to those present in the used oil samples for calibration purposes and by careful use of available computer software. The high results obtained for copper in samples 7 and 8 and for iron in samples 7 8 and 9 were possibly caused by the presence of a tube line which could not be completely removed by filtering. Correlation with Other Analytical Techniques A correlation programme was carried out on a series of monograde diesel engine oils contain-ing the additives calcium zinc and phosphorus with approximate concentrations of 3 800, 1 000 and 1 000 p.p.m.respectively using direct-reading optical emission spectroscopy (OES) and flame atomic-absorption spectroscopy (AAS). The OES work utilised the rotating disc technique. Calibration was carried out using standards dissolved in the same medium as the samples which also contained the additives with the concentrations as listed above. The AAS work utilised a dry-ashing technique followed by dissolution of the residue in hydrochloric acid. The solution was filtered to remove any silicon present and analysed for the required elements. The X-ray fluorescence (XRF) results were obtained using standard matrix 3 above for calibration purposes. The results are shown in Table VI. TABLE VI CORRELATION PROGRAMME OF THE ANALYSIS OF WEAR MATERIALS IN USED LUBRICATING OILS BY XRF AAS AND OES SPECTROSCOPY The optical emission spectroscopic analysis was carried out by the British Rail Technical Centre, Derby.Chromium 7+ Sample AAS XRF 1 9 6 2 1 3 3 1 3 4 54 57 5 39 43 6 4 6 7 AAS 27 31 77 355 307 121 ~~ Iron -h--7 OES XIiF 20 19 28 31 74 71 355 354 322 287 106 104 Concentration p.p.m. A - Copper Tin Lead -7 7+ r-* ---l AAS OES XIiF OES XR1; AAS 0 1 3 S XRF 23 22 18 - 0 13 16 4 9 7 5 - - 0 7 1 0 0 29 28 24 3 0 11 11 5 36 34 33 - 2 28 46 41 30 32 28 4 0 39 53 46 52 60 67 ( 1 8 21 21 15 The matrix matching calibration procedure used in this correlation programme enabled reasonable agreement to be obtained between the AAS OES and XRF techniques for the determination of chromium iron and copper.The concentration of tin in the samples was too low to enable any conclusions to be drawn about the reliability of the XRF method for the analysis of this element. The agreement obtained for lead was not as good as that obtained for the other elements analysed. The lead content of samples 1-3 inclusive was rather low and was near to the limit of detection for the XRF method. The ratio of the peak to background count rate in the XRF method was low and so the analytical error was greater than at higher lead concentrations. Conclusions This method overcomes the problem of sedimentation using a novel sample preparation technique. The procedure is rapid and simple to carry out and is suitable for a range of engine and transmission oils.Prolonged exposure to X-rays appears to cause no serious damage to the discs and does not give rise to any contamination of the equipment. The discs can also be reused for further analysis. The method is suitable for analysis of elements which have atomic numbers greater than 14 but is not considered to be suitable for the determination of silicon o 260 SEARLE AND THOMPSON aluminium. The analytical results obtained have been shown to be comparable to those obtained using other techniques especially when matrix matching between samples and standards was employed. The authors thank P. T. Corbyn and H. Bauer of the British Rail Technical Centre for carrying out the optical emission spectroscopic analysis detailed above. References 1. 2. 3. 4. 5. 6. 7. “ASTM Methods for Emission Spectrochemical Analysis,” Third Edition American Society for 8. 9. 10. Jackson D. I<. Salama C. and Dunn R. Can. Spectrosc. 1970 15 17. Saba C. S. and Eistentraut K. J. Anal. Chern. 1979 51 1YS7. Taylor J . H. Bartels T. T. and Crump N. L. A n d . Chem. 1971 43 1780. Brown J . R. Saba C. S. and Rhine W. E. Anal. Chem. 1980 52 2365. Fuller C. W. Hutton R. C. and Preston B Analyst 1981 106 913. Langmyhr I;. J. Analyst 1979 104 993, Testing of Materials Philadelphia 1960 614. Johari O. Corvin I . Samudra A. V. and Staschke J. lnd. Lubr. Tribol. 1979 31 172. “Accuracy in X-ray Spectrochemical Analysis as Related to Sample Preparation,” E. G. & G. Ortec, Corbyn P. F. and Bauer H. personal communication. Oak Ridge TN USA. Received May 24th 1982. Accepted October 22nd 1982

ISSN:0003-2654    

DOI:10.1039/AN9830800254

出版商:RSC    

年代:1983

数据来源: RSC