摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 373 Direct Determination of Lead in Soils and Sediments by Atomic Absorption Spectrometry Employing a Graphite Capsule in Flame Atomiser Janez Stupar Univerza E. Kardelja, lnstitut "Joief Stefan " Ljubljana, 61 I I I Ljubljana, Jamova 39, Yugoslavia The construction of a graphite capsule in flame atomiser is described for the direct atomic absorption spectrometric determination of lead in soils, sediments and bone ash. A single calibration graph can be used for different materials, prepared by atomisation of different masses of various standard reference materials. The absolute limit of detection of lead at the 283.3-nm line was 2 x 10-3 pg and the reproducibility was 8-20% (RSD). The effects of the physical properties of the graphite and the effect of the heating power upon the lead absorbance - time profiles were investigated.These parameters were found to have a direct influence on the sensitivity, accuracy and precision of the measurements. Keywords: Atomic absorption spectrometry; atomisation of solids; capsule in flame; lead determination; standard reference materials, soils, sediments The determination of lead in soils and sediments is often required for the monitoring of atmospheric or water pollution. Preparation of solutions from these samples for atomic absorption measurements is troublesome and time consum- ing. In addition, there is the likelihood of sample contamina- tion during chemical treatment. To avoid these problems, attempts were made to develop a direct method of analysis.The suspension technique was found to be unsuitable due to the low lead levels in the dilute suspension and the poor accuracy of the technique, 1 Combination of an electrothermal and a flame atomiser was investigated for the purpose. The idea for this combined atomisation technique came from Borzov et af.2 who published the first paper, in 1969, on the application of a graphite furnace and an air - acetylene flame for efficient atomisation of solid samples (carbonate and oxide ores) and solutions for the determination of copper. A few years later, a new approach was made in this direction by Kackov et a1.3 The atomiser described in their paper was referred to as a capsule in flame atomiser. The essential difference from the previous system was in the placing of the sample inside a graphite cylinder filled with graphite powder. The sample vaporised inside the graphite cylinder during its rapid heating to a temperature of 2000-2500 K.Vapours (molecules andlor atoms) were released into the flame by diffusion through the porous graphite walls. The integrated absorbance [Ai (absorbance seconds)] was measured imme- diately above the capsule. The essential feature of this closed system is in the prevention of part of the sample being thrown out of the furnace during rapid heating, which is usually experienced in open systems. Since then, several studies have been carried out by different groups in the Soviet Union to investigate the analytical applicability of this novel technique to the direct analysis of various solid samples by AAS and AFS.The essential observations are summarised in the following para- graphs. Kackov et al.3 measured 27 elements of high, medium and low volatility in samples of different origin (for example, geological, ceramics, oxides and biological) using a graphite capsule and airhitrous oxide - acetylene flames. Even elements with boiling-points exceeding 3500 K, such as titanium, silicon and molybdenum, were vaporised satisfac- torily in the capsule. The details of the instrument, capsule design and sample preparation were presented in the paper. Typically, a 50-mg sample was used for a single measurement. The samdles were diluted with the graphite powder to aid the chemical reduction and to prevent formation of a compact bead upon melting.Complete vaporisation from the matrix of the element being determined can be accomplished in fractions of a second to some tens of seconds depending on the volatility of the element and the heating power. The high sensitivity of the technique permitted the determination of elements in a concentration range of 10-1-10-3 pg g-l. A precision of 7-15740 was reported, depending predominantly on the homogeneity of the sample. L'vov et af.4 and Pelieva et af.5 studied the optimisation of the parameters involved in atomic absorption measurements employing a graphite capsule in flame atomisation device. Flame-gas velocity, flame geometry, distance between the capsule and the burner head and the heating power (current) were the most critical parameters in determining the inte- grated absorbance (Ai).Differences in the electrical proper- ties between capsules were found to influence significantly the precision of the measurement. The characteristics of the electronic registration system required were also considered in the paper. In a review paper L'vovs summarised the main features of the graphite capsule in flame atomiser for the direct analysis of solids by AAS. One of these was the calibration procedure. If integrated absorbance, Ai, is measured, the effects of the kinetics of liberation and transport of the analyte to the flame are almost absent. The necessary requirements are as follows: complete vaporisation of the element determined from the sample, constant residence time of the atoms in the flame and a linear relationship between the absorbance and atom concentration in the flame (no) over the range of no experien- ced in the atomisation process.Thus, in principle, any substance with a known amount of element to be determined can be used for calibration. Razumov and Korovina7-9 studied the distribution of the absorbing atoms in the flame above the heated capsule. Convection currents flowing around the hot graphite capsule cause an enrichment of the atom concentra- tions at certain distances above the capsule, but close to it (up to 2-3 mm) a depletion was observed. It was further pointed out that in addition to diffusion (flame-gas velocity and temperature), the convection flow should be considered in calculating the effective concentration of absorbing atoms (no!) in the observed region of the flame.Reducing the flame-gas velocity and temperature by using a town gas - air mixture instead of air - acetylene improved the sensitivity of Cu, Mn, Te, Cd and Bi by factors of 2-3. Determination of a number of major and trace elements in plant and soil samples was reported. Several paperslC12 have dealt with the use of a graphite374 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 capsule in flame atomiser in the analysis of geological (rock and ore) samples. The most interesting and common observa- tion in these papers referred to the shape of the absorption signals, Synthetic standards prepared from graphite powder and a standard solution exhibited narrow single peak absorp- tion signals. However, rock and ore samples produced absorption signals which consisted of several peaks that appeared at higher temperatures.The samples were mixed with graphite powder in ratios from 1 : 1 to 1 : 1000 to enable free vaporisation of the element from the sample matrix. Silicate and carbonate rocks and sulphide ore samples were successfully analysed (Cu, Pb, Cd, In, Te, Sb, Bi and T1) by this technique. Lithium and rubidium were determined in silicate rocks12 by measurements of integrated emission (Ei). MenSikov et al. 13 used a graphite furnace in flame atomiser for the direct determination of gold in ores and other samples. Employing a 500-mg sample mass, a limit of detection of 3 X 10-2 kg g-1 was obtained. Kantor ef al. 14 used a graphite cup (Varian CRA-63) with an air - acetylene flame atomiser for the determination of sodium in alumina. The calibration graphs for aqueous standards and the alumina samples were different.Kantor et al. 15 employed an electrically heated graphite furnace in conjunction with a flame AAS instrument to investigate the thermochemical properties of various com- pounds. The aim of the present work was to construct and develop our own combined capsule in flame atomiser, and to evaluate its performance for the direct analysis of various solid materials (soils, sediments and bone ash) by AAS. Experimental Spectrometer and Atomiser A simple atomic absorption spectrometer was used in this work, assembled in our laboratory from the following parts: a Hilger prism monochromator, a Varian Techtron Pb hollow- cathode lamp operated at 5 mA, a Varian AA-5 lamp power supply, a Hamamatsu R-213 photomultiplier tube, a Varian indicator module Type IM-6D and a Varian Model 9176 10-mV chart recorder with full-scale response less than 0.5 s. The atomiser (see Fig.1, 1-7) employed for atomisation of powdered solid samples was a slot (two parallel slots) 5 burner (1),16 producing a 25 mm long air - acetylene flame (7), on top of which a special device (2) was fixed. The device, machined from brass, enabled the capsule to be clamped firmly and moved in or out of the flame. The graphite capsule (6) was clamped between two water-cooled brass blocks (3 and 4), which also served as electrical contacts (5). Heating of the capsule was provided by a locally made power unit,17 which supplied a maximum of 600 A at 10 V.The heating rate, as well as the final temperature of the capsule, was controlled by the voltage on the primary of a step-down transformer. Six different positions can be selected as follows: position 1, 5.2 V; 2, 5.6 V; 3, 6.1 V; 4, 6.6 V; 5 , 7.0 V and 6, 7.4 V. Capsules, 30 mm in length, were machined from graphite rods (Ringsdorff) 6.14 mm in diameter (see Fig. 2). Four graphite types of different physical properties were investi- gated: RWO (d = 1.75 g cm-3), RWI 103 ( d = 1.55 g cm-3), RWII 203 (d = 1.53 g cm-3) and RWI (d = 1.57 g cm-3). Most of the analytical measurements were performed using RWI 103 and RWII 203 material. Before use, new capsules were heated for 3 s (2000 K) to remove trace amounts of lead. Temperature Measurements The temperature of the graphite capsule (outside wall temperature) was measured at the centre of the sample compartment area using an optical pyrometer (Pyrowerk, C - 10 mm Fig.2. sample compartment Graphite capsule: A, capsule; B, graphite stopper; and C, 2600 2400 2200 2000 z 2 f 1600 1800 4- $ 1400 1200 1000 - x - - x - - C X' 0 1 2 3 4 5 6 7 8 9 10 Timeis Fig. 1. Details of the graphite capsule in flame atomiser: 1, burner; 2, device to hold the capsule; 3, water inlets; 4, water outlets; 5 , electrical contacts; 6, capsule; and 7, flame Fig. 3. Variation of capsule temperature with time for different heating powers and graphite materials; RWI 103, solid lines and RWII 203, broken lines. Voltages on step-down transformer: A, 5.2; B, 5.6; and C, 6.1 VJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 375 RWI 103 Hannover GmbH). The temperature of the capsule was measured each second from 2 to 10 s after application of the electrical voltage. The temperature - time functions of two graphite materials used in the experiments were measured, varying the voltage on the primary of the step-down trans- former. The results of these measurements are presented in Fig. 3. A substantial difference in heating rate as well as in the final equilibrium temperature between these materials can be observed at the same applied voltage. The final temperature is primarily the consequence of the different electrical resistivity of the materials. The temperature gradient in the horizontal position was small and a fairly uniform temperature was obtained in the sample compartment region.The temperature of the capsule was found to change gradually with use (number of firings). RWI RWll 203 RWO Standards and Samples Four different standard materials were used for preparation of the calibration graphs: IAEA Soil-5 standard reference material (Pb 129 p.p.m.); NBS River Sediment SRM 1645 (Pb 714 p.p.m.); IAEA A-3/1 Calcinated Animal Bone standard reference material (Pb 6.75 p.p.m.); and lead nitrate aqueous standard solutions. Six different soil samples (terra rossa and peat soil) and six river and lake sediments of unknown lead content were prepared and analysed by the described technique. Soil and sediment samples were dried (105 "C) , and crushed and ground to a particle size of <45 pm prior to analysis.Solid material (5-25 mg) was transferred into the capsule for a single measurement by a weighing spoon. Lead nitrate standard solutions (10 pl) were placed in the capsule using an Oxford micropipette and the capsule was heated at 90 "C for 5 min. Optimisation of Atomisation Parameters Several studies , particularly on optimisation of flame parameters, have been carried out by L'vov et al. ,4 Pelieva et aZ.5 and Razumov and Korovina.7.8 It is evident from these studies that the sensitivity of the measurements is inversely proportional to the flame-gas velocity. As a slot burner was employed in this study, the minimum flow of air and acetylene was selected to maintain a stable flame. These were 3.5 1 min-l of air and 0.7 1 min-1 of acetylene.The position of the capsule in the flame was at the flame centre, 14 mm above the burner head. The absorbance at the 283.3-nm lead line was measured immediately above the capsule. Integrated absorbance, Ai, was derived from the absor- bance - time profiles obtained on the chart recorder by graphical integration of the area under the A versus time graphs, employing a planimeter technique. It was observed that the lowest position on the step-down transformer (position 1) was adequate to vaporise all of the lead (as a solution of lead nitrate) from the capsule within 8-10 s. To achieve the same effect with sediment and particularly soil Time - Fig. 4. Absorbance - time profiles for 0.6 pg of Pb (aqueous nitrate solution), showing the effect of different graphite characteristics samples, a higher heating power (position 2 or 3) was necessary.The major objective of the investigation in regard to optimising the atomisation parameters was to select the most suitable graphite material. Capsules of equal dimensions (see Fig. 2) were prepared from the four different graphite materials. These were used for atomisation of lead nitrate aqueous solution under identical conditions (heating power etc.). The results of this experiment are presented in Fig. 4. As a result of the different electrical (specific electrical resistivity) and physical properties (porosity etc.) of the graphite materials investigated , they heated up at different rates and provided different resistances to diffusion vapour through the thin capsule walls.The various graphite proper- ties are particularly well established for RWI 103 ( d = 1.55 gcm-3) and RWII 203 (d = 1.53 gcm-3). The latter has avery high specific electric resistivity and the heating rate is thus slow (see Fig. 3) and as a consequence, lead atoms are liberated at a lower rate over a longer period of time. Wider absorbance - time profiles were characteristic of this material. In contrast, RWI 103 and RWO ( d = 1.75 g cm-3) have very small and fairly similar electrical resistivities, but substantially different porosities. RWI 103 is a highly porous material while RWO is a rather compact type of graphite, and thus provides a high resistance to lead molecules (atoms) diffusing through the walls. Lead absorbance - time profiles characteristic of these materials reflected these properties very well (note the tailing of the traces in Fig.4). Graphite capsules prepared from RWI 103 and RWII 203 material were selected for further experiments. Various soil and sediment samples were atomised using these capsules. Typical lead absorbance - time profiles are illustrated in Figs. 5 and 6. Similar trends in the shape of the absorbance peaks as those observed for lead nitrate solution were displayed by the soils and sediments. A short term high lead atom concentration was built up in the flame when employing RWI 103 graphite, while a moderate lead atom concentration spread over a longer period of time resulted from the use of RWII 203 graphite. In the preparation of a calibration graph it is desirable that any material of known lead content could be used as a standard. According to L'vov6 this is possible if three requirements are met: complete vaporisation of lead from the 0.3 0.2 0.1 t $ 0 f! 2 2 m 0.2 0.1 0 IAEA Soil-5 standard RWI 103 -L 10 s (bl - RWll203 Clav soil Near town (el RWll203 Time - Country RWI 103 ( f ) RWll203 Fig.5. Typical absorbance - time profiles for the determination of Pb; atomisation of soil Sam les em loying capsules of different graphite materials. (a) 0.61; (g) 0.62; 6) 0.32; ( d ) 0.41; (e) 0.12; and cf) 0.14 pg of PbJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 376 0.6 0.5 t0.4 8 5 0.3 2 m n v) 0.2 0: ( (a) RWll203 t 1 0 s ' I Time + Fig. 6. Typical absorbance - time profiles for the determination of Pb; atomisation of sediment samples employin capsules of different graphite type: (a) Sample No.1, 1.86 pg of Pb; t b ) sample No. 8,0.56 pg of Pb; (c) sample No. 1, 1.7 pg of Pb; and ( d ) sample No. 8,0.77 pg of Pb 150 v) ? * 100 s 0 c m n m rn 4- 2 g 50 4- - I I I 0 0.5 1 .o 1.5 Total amount of Pb/pg Fig. 7. Calibration graphs for determination of Pb, atomisation of solid samples. and 0 , NBS .River Sediment; and 0, IAEA Soil-5 standard; and A, IAEA Calcinated Animal Bone. Capsule material: solid line, graphite RWII 203; and broken lines, graphite RWI 103 capsule, constant time of residence of lead atoms in the flame and a linear relationship between nopb and absorbance over the range of nopb observed in the flame during the atomisation cycle. The experiments confirmed that no residual lead was present in the capsules after 10 s of heating.A constant residence time of atoms in the flame was assured by good flame stability. The linearity of the relationship between nOpb and absorbance was examined at the 283.3-nm Pb line for the particular spectrometer employed by spraying aqueous lead standard solutions into the flame. It was found that the q ) p b 0~ A relationship started to deviate from linearity at nOpb values corresponding to absorbances in excess of 0.25. Therefore the use of RWI 103 graphite capsules would have resulted in a different calibration graph for each of the materials employed (soils and sediments), at least at higher total lead contents in the capsule. This was actually confirmed by the fact that the calibration graphs for soils and sediments experienced a certain amount of curvature (see Fig.7, broken lines). An additional reason for this is likely to be found in the failure of the recorder to follow the fast changes of absorbance in a very short time. Therefore, for analytical purposes, the calibration graphs were again prepared using RWII 203 graphite capsules. In this instance, as can be seen in Fig. 7 (solid line), the three different standard reference materials resulted in the same calibration graph. In order to evaluate the analytical applicability of the graphite capsule in flame atomisation technique, several soil and sediment samples were analysed. Five portions (5-25 mg) of each sample were atomised, and the mean values and standard deviations were calculated.The same samples were also analysed by conventional flame AAS. Two portions (300 mg) of each sample were dissolved in a mixture of HCl - HN03 - HF, diluted with water to volume and sprayed into the air - acetylene flame. Background absorption was measured and subtracted from the total absorbance signal. The results are summarised in Tables 1 and 2. It may be concluded from Table 1 that good agreement between techniques (with one exception) was obtained for soil lead contents. It is interesting to emphasise that the precision of measurement of conventional AAS is not better than that of the direct atomisation technique, particularly at low lead levels in soil. In the sediment analysis (see Table 2), the discrepancy between the results is substantial, the precision of measure- ment of conventional AAS is far better than that obtained by direct atomisation. The latter can be at least partly explained by the low masses of the sample (5-10 mg) used for single determinations, associated with the poor homogeneity of the samples.These should be diluted with graphite powder before atomisation. Table 1. atomisation and conventional flame atomisation AAS Determination of lead in soils by graphite capsule in flame Mean valuelpg g-1 and RSD, Yo Graphite capsule in flame direct Sample atomisat ion Clay . . . . . . . . 50 f 13 Sand . . , . . . 16+8 Terra rossa Terra rossa Peat soil (near highway) . . 6 2 2 17 Peatsoil . . . . . . 26 k 14 (Zemunik) . . . . 22+ 11 (Novigrad) . . . . 21 f 20 Dissolution, flame atomisat ion 53 + 14 20+ 11 21 f 19 2 3 2 8 97 & 5 30 2 10 Table 2.flame atomisation and conventional flame atomisation AAS Determination of lead in sediments by graphite capsule in Mean valuelpg g-1 and RSD, YO Graphite capsule in flame direct Sample atornisat ion No. 1 . . . . . . 224 k 9 No. 8 . . . . . . 165 2 24 No. 9 . . . . . . 34 k 20 27 + 15 No. 10 . . . . . . No. 11 . . . . . . 61 k 12 No. 12 . . . . . . 27 f 18 Dissolution, flame atornisat ion 197 & 3 134 2 4 39 L 6 21 k 11 54 k 8 24 k 6JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 377 Conclusions The graphite capsule in flame atomisation technique is rapid and free from contamination problems. It can be used successfully for monitoring lead in soils, sediments and some biological samples. Absolute sensitivity of lead in these materials based on 1% absorption for 1 s is 2 X 10-3 pg.A single calibration graph can be used for all three materials investigated (soil, sediment and bone ash). A small amount of curvature is observed at high absorbance values, which is due to the non-linear relationship between the lead atom concen- tration in the flame (nOPb) and absorbance ( A ) at higher absorbance values. Reproducibility and the accuracy of the method are strongly influenced by the sample homogeneity due to the low sample masses used (5-15 mg) for a single determination. This was found to be particularly critical with most of the sediment samples, whilst a satisfactory agreement was obtained with conventional AAS for soil samples. The selection of an appropriate graphite material (purity, porosity and electrical resistivity) is important in determining the shape of the lead absorbance - time profile.Most of the graphite capsules can be used for up to 50 lead determinations. The author thanks the “Boris Kidrit” Foundation for provid- ing financial support, and Mr. Bruno Hiitsch of Ringsdorff- werke GmbH, Bad Godesberg, FRG, for supplying the graphite material. The author is also indebted to Professor S. R. Koirtyohann, University of Missouri, Columbia, MO, USA for making some valuable suggestions. I would also like to acknowledge the help and encourage- ment of the late Professor Lado Kosta, with whom I collaborated for many years. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. References Willis, J. B., Anal. Chern., 1975, 11, 1752. Borzov, V. P., L’vov, B. V., and PljuS, G. V., Zh. Prikl. Spektrosk., 1969, 11, 217. Kackov, D. A . , Kruglikova, L. P., and L’vov, B. V., Zh. Anal. Khim., 1975, 30, 238. L’vov, B. V., Kruglikova, L. P., and PljuS, G. V., Zh. Prikl. Spektrosk., 1971, 15, 975. Pelieva, L. A., Muzikov, G. G., and Sarnopoljski, A. J . , Zh. Prikl. Spektrosk., 1976, 25, 414 L’vov, B. V., Talanta, 1976, 23, 109 Razumov, V. A., and Korovina, L. M., Zh. Prikl. Spektrosk., 1978 29, 730. Razumov, V. A., Zh. Prikl. Spektrosk., 1977 26, 948. Razumov, V. A., Zh. Anal. Khim., 1977, 32, 383. Zaguzin, V. P., Karamanova, N. G., and Pogrebnjak, J . F., Zh. Prikl. Spektrosk., 1978, 28, 963. Karamanova, N. G., and Pogrebnjak, J. F., Zh. Anal. Khirn., 1979, 34, 213. Karamanova, N. G., and Pogrebnjak, J. F., Zh. Prikl. Spektrosk., 1980, 32, 9. MenSikov, V..J., Malih, V. D., and Sestakova, T. D., Zh. Anal. Khim., 1974, 29, 2132. Kantor, T., Bezur, L., and Pungor, E., Mikrochim. Acta, 1981, I , 289. Kantor, T., Pungor, E., Sztatisz, J . , and Bezur, L., Talanta, 1979, 26, 357. Stupar, J., Spectrochim. Acta, Part B, 1976, 31, 263. DolinSek, F., and Stupar, J., Analyst, 1973, 98, 841. Paper 5611 9 Received March 19th, 1986 Accepted May 30th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100373

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 379 Determination of Dissolved Inorganic Selenium(1V) and Selenium(V1) Species in Natural Waters by Hydride Generation Atomic Absorption Spectrometry S. C. Apte* and A. G. Howardt Department of Chemistry, The University, Southampton, Hampshire SO9 5NH, UK A method is described for the determination of dissolved total inorganic selenium and dissolved selenium(lV) species in natural waters. Reduction of selenium(lV) in the sample with sodium tetrahydroborate(ll1) results in the evolution of hydrogen selenide. This gas is pre-concentrated in a cryogenic trap, re-volatilised and atomised in an electrically heated quartz furnace aligned in the light beam of an atomic absorption spectrometer. Selenium(V1) species are determined after a pre-reduction with hydrochloric acid.The technique can be employed for the analysis of samples containing between 0.2 and 5 ng of selenium. It is directly applicable over the concentration range 10-200 ng I-' of selenium but higher concentrations can be determined following suitable dilution. The over-all detection limit of the method is ca. 3 ng 1-1 of selenium. Keywords: Selenium determination; cryogenic pre-concentration; atomic absorption spectrometry; h ydride genera tion; na tura I waters The metalloid selenium is a trace constituent of natural waters, occurring in deep ocean waters at concentrations of typically 50-100 ng 1-l.1 Its environmental chemistry is complicated by the existence of a variety of different and interconvertible selenium species making up its biogeochemical cycle.Although methylated selenium species have been found in the terrestrial environment, they do not appear at present to be important components of the marine environment2 and most systems are dominated by inorganic selenium(1V) and sel- enium(V1) species. l v S 5 As these species have differing toxi- cities and reactivities and are assimilated by organisms at different rates, it is desirable that analytical methods should be available to determine selectively the two major inorganic species. For the study of marine systems, such methods should be capable of determining selenium species in the concen- tration range 1-300 ng 1-1 of Se. Procedures for the determination of selenium at low levels have largely centred on fluorimetric5-7 and gas chromato- graphic methods8-11 involving the formation of piazselenol derivatives of selenium(1V).The selectivity of such methods to the quadrivalent oxidation state has been exploited to give information on chemical speciation.5.9 In order to gain adequate sensitivity, however, such methods require lengthy pre-concentration procedures (such as coprecipitation or solvent extraction) and pre-treatment steps that greatly reduce sample throughput. Additionally, the piazselenol reagents are suspected carcinogens. Hydride generation coupled with atomic spectrometry is a sensitive, rapid method that can be readily applied to the determination of most metalloid elements in the low pg 1-1 range (see, e.g., references 12-14). In order to achieve the sub-yg 1-1 detection limits required for the analysis of natural waters, it is necessary to employ a pre-concentration step.This can be conveniently achieved by concentration of the evolved hydride by cryogenic trapping.15.16 In spite of the merits of the hydride generation approach, very little work has been published on low-level selenium methods. This paper describes a relatively simple hydride generation AAS method for the determination of inorganic selenium species in natural waters. Information on oxidation state is obtained by exploiting the selectivity of the tetraborohy- drate(II1) reduction step to the selenium(1V) oxidation state. * Present address: Water Research Centre, P.O. Box 16, Henley t To whom correspondence should be addressed. Road, Medmenham, Buckinghamshire, UK.A sample throughput of better than five samples per hour has been achieved as a result of hydride generator design and the incorporation of a PTFE cryogenic trap. These modifications improve on past methods, giving a highly sensitive and reliable analytical procedure for the speciation of selenium in natural waters. Experimental Reagents and Glassware All chemicals were of analytical-reagent grade unless stated otherwise. Glass- and plasticware were cleaned by soaking in dilute (1 + 9) nitric acid and were rinsed with distilled water before use. Stock standard solutions (1000 mg 1- of Se) were prepared from sodium selenate and selenious acid dissolved in dilute (1 + 9) hydrochloric acid and were cross-checked by flame AAS. Standard solutions of lower concentration were prepared by dilution of the stock solution on the day of use.Analytical-reagent grade sodium tetrahydroborate( 111) was used to prepare unstabilised solutions in distilled water (optimised concentration, 4% rnlV). These solutions were stable for the course of a working day. The selenium content of the sodium tetrahydroborate(II1) is variable and makes a significant contribution to the blank measurements. It is therefore recommended that when setting up the method, batches of the reagent are screened for selenium to permit the selection of a low concentration material for future work. Concentrated hydrochloric acid was purified before use by sub-boiling distillation. Sulphanilamide solution (2% mlv) was prepared by dissolution in the minimum volume of 6 M hydrochloric acid and then diluted with distilled water.Apparatus The apparatus consisted of three distinct stages: a hydride generator, a cryogenic trap and an atomisation - detection system. The last two stages have been described in detail in previous publications13.17 and only a brief description will therefore be given here. Hydrides are generated from solution in a purpose-built generator [Fig. l(a)] consisting of a cylindrical chamber having a medium porosity frit at its base. A side port, which is capped with a silicone-rubber septum (replaced every 20-30 determinations), allows the introduction of tetrahydrobo- rate(II1) solution into the acidified sample. The generation chamber is connected to the rest of the apparatus via a ground-glass joint at the top of the chamber. When in use, the380 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 Silicone - rubber septum 3 I 5 mm i.d. I 2 cm 14 Borosilicate glass I AAS light path ( C) ! t Nitrogen carrier gas ( a ) Fig. 1. Details of the (a) hydride generator, ( b ) cryogenic trap and ( c ) quartz-tube atomiser 2.5 ng Fig. 2. Typical recorder output showing peak profiles generated from selenium(1V) standards sample is held in the chamber by a combination of upward gas pressure (supplied by the nitrogen carrier gas) and surface tension. Once generated, the hydride is trapped at liquid nitrogen temperature in a glass U-tube packed with ca. 3 X 20 X 0.1 mm strands of PTFE [Fig. l(b)]. On completion of the trapping stage, the liquid nitrogen Dewar vessel is removed to allow the trap to warm to room temperature.The eluting hydride is then swept into an electrically heated quartz tube [Fig. l(c)] having a thermally shielded inlet. The furnace tube is aligned in the optical path of a Varian AA 175 AB atomic absorption spectrometer fitted with an EM1 9783 B photomul- tiplier tube giving an extended response in the UV region. Throughout the work the instrument was operated with deuterium lamp background correction. The resulting spect- rometer output was recorded on a Tekman TE 200 y - t chart recorder. A typical recorder output from the system is shown in Fig. 2. The following spectrometer operating conditions were used throughout the study: source, hollow-cathode lamp; operating current, 5 mA; wavelength, 196.0 nm; and spectral band pass, 1.0 nm.Recommended Procedures The sample is split into two aliquots. Inorganic selenium(1V) is determined directly on one aliquot and total inorganic selenium is determined on the other portion after reduction of the inorganic selenium(V1) to the quadrivalent state. The selenium(V1) concentration is given by the difference between the two values. Selenium( I V ) With the carrier gas flow-rate set to approximately 50 ml min-1, the sample aliquot (usually 50 ml) is pipetted into the reactor and the pH is adjusted to a hydrogen ion concentration of approximately 4 M by the addition of 25 ml of hydrochloric acid. Sulphanilamide solution (1 ml) is also added in order to mask any interference arising from the presence of nitrite.20 The reactor is connected to the cryogenic trap assembly and the carrier gas flow-rate adjusted to 200 ml min-1.After the solution has been purged for 60 s, the trap is allowed to cool in liquid nitrogen for 1 min prior to the injection of 5 ml of sodium tetrahydroborate(II1) solution. A glass syringe with a stainless-steel needle is used for this step, with 5 ml of solution taking ca. 15 s to inject. After a period of 90 s, to allow the hydride to be collected in the trap, the liquid nitrogen Dewar vessel is removed and the trap allowed to warm to room temperature, releasing the hydrogen selenide into the atomiser. The resulting spectrometer output is recorded. Between determinations, the reduction chamber is rinsed with distilled water and the trap is warmed in hot water for ca.3 min in order to remove any condensed water droplets. Quantitation is by means of peak-height response and the method is calibrated against standards prepared in distilled water. Total inorganic selenium A 50-ml sample is pipetted into a conical flask, 50 ml of re-distilled hydrochloric acid are added and the solution isJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 381 0.20 a, t CD a ln & 0.14 2 0.08 0 5 10 HClirnol I-' NaBH4, "/" miV Volume addediml I l r 0.08 0 1 2 3 Flow-rate/ml min 1 Trapping timeimin Fig. 3. Optimisation of experimental conditions. Effect of changing: (a) hydrochloric acid concentration (NaBH, concentration 2% mlV, gas flow-rate 200 ml min- 1); ( b ) sodium tetrahydroborate(II1) concentration (HCI concentration 4 moll-', NaBH4 volume 5 ml, gas flow-rate 200 ml min-I); ( c ) sodium tetrahydroborate(II1) volume (HCl concentration 4 mol 1-1, NaBH, concentration 2% mlV, gas flow-rate 200 ml min-1); ( d ) carrier gas flow-rate (acid strength 4 mol 1-l, NaBH, 5 ml of 2% mlvsolution); and (e) trapping time (HCl concentration 4 mol 1k1, gas flow-rate 200 ml min-l, NaBH, 5 ml of 2% mlV solution) Table 1.Experimentally optimised conditions Hydrochloric acid concentration . . . . . . . . 4 rnol I- 1 Sodium tetrahydroborate( 111) concentration: 50-ml sample . . . . . . . . . . . . . . 4% m/V 25-ml sample . . . . . . . . . . . . . . 2% mlV Sodium tetrahydroborate(II1) volume . . . . . . 5 ml Carrier gas flow-rate . . . . . . . . . . . . 200 ml min-I Trapping time .. . . . . . . . . . . . . 90 s ~ Table 2. Performance characteristics Linear detection range . . . . 0-5 ng Precision [n = 10, 25 ml containing 5 ng of Se(IV)] . . . . . . . . 5.1% Sample volumelml ng ngl-I . . . . Detection limit (30) 25 0.15 6 50 0.16 3 Sample throughput . . . . 5 per hour 0 5 10 15 20 Timeih Fig. 4. Conversion of selenate (SO pg I-' of Se) to selenium(1V); 6 rnol I-' hydrochloric acid at 60°C selenium(V1) had first to be reduced to selenium(1V). This was achieved by warming the sample solution at 60°C in 6 M hydrochloric acid. Reduction was essentially complete in 4 h (Fig. 4), and recovery experiments employing both distilled water and sea-water samples spiked with selenate (40 pg 1-1 of Se) gave recoveries of 98.7 and 99.670, respectively. warmed at 60 "C for 4 h.After cooling, 1 ml of sulphanilamide solution is added and the solution is transferred to the reactor for analysis as described above. Results and Discussion Instrumental Design and Optimisation of Conditions Hydride generation The hydride generator employed in this method was designed to achieve efficient mixing of sample and reagents and to allow rapid stripping of the evolved hydride from sample solutions. In order to reduce adsorption problems, the generation chamber and all other glass surfaces were acid washed and treated with trimethylchlorosilane dissolved in light pet- roleum. The optimisation of experimental conditions was based on the peak-height response from selenium(1V) standards (200 ng 1-l). The effects of varying the hydrochloric acid concentra- tion, sodium tetrahydroborate(II1) concentration, carrier gas flow-rate and trapping time are depicted in Fig.3. Optimised conditions for both 25- and 50-ml sample volumes and system performance data are summarised in Tables 1 and 2, respec- tively. In agreement with the findings of other workers, no response was obtained from selenium(V1) standards, thereby permitting the selectivity of the tetrahydroborate(II1) hydride generation step to the quadrivalent oxidation state to be exploited to give information on inorganic selenium specia- tion. In order to determine total inorganic selenium, all Pre-concentration Problems were experienced in selecting a trap packing that did not adsorb significant amounts of the hydride.Both glass beads and glass-wool were unsatisfactory in this respect. A trap packing consisting of PTFE strands did, however, give reproducible results. Additional problems were experienced in selecting a drying agent for gas streams. Calcium chloride and sodium hydroxide adsorbed the hydride, owing to the formation of selenides, and a water trap consisting of a glass U-tube immersed in a salt - ice slurry was effective but caused severe signal drift as ice built up in the tube. The problems were overcome by omitting a discrete gas stream drying stage and allowing the water to freeze in an unpacked section of the cold trap. The ice was removed by immersing the trap in hot water between determinations. A tomisation The hydride was atomised in a quartz tube [Fig.l(c)] heated by means of a purpose-built electric furnace consisting of two "coffin-shaped" halves made from alumina cement in which hand-wound wire elements were embedded. Power was supplied to the furnace by means of a variable transformer. The furnace was capable of a maximum temperature of ca. 1100 "C, although the optimum temperature range for atomi- sation was ca. 800-1000°C. After extended use (approxi- mately 3 months), the performance of the quartz tube atomiser was found to deteriorate, giving rise to noise and poor precision. By this stage the quartz surface generally showed signs of significant erosion and the tube was therefore replaced.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 382 Table 3. Results of spiking recovery experiments on sea water Amount of Se(1V) addedlng n Recoverylng 5 4 5.0 k 0.1 5 4 5.1 f 0.1 Table 4.Analysis of some real water samples Sample location Total Selng 1 - 1 Se(IV)/ng 1-’ Loch Ewe, N.W. Scotland: 26/3/83 58 4 6/4/83 46 6 12/4/83 50 <3 18/7/83; Salinity, ‘/o : River Tamar, England, <2 161 69 13 154 31 21 12 1 21 30 89 21 Interferences With one notable exception (nitrite’s) previous work has shown the method to be essentially free from interferences from chemical species at the concentration levels expected to occur in natural waters.19 In this study, a nitrite concentration of 8 pg 1-1 completely suppressed the signal response from a 200 ng 1-1 selenium(1V) solution. Signal suppression was also evident during total selenium determination in sea-water samples, even when the initial nitrite concentration was low.Nitrite interference was completely masked by the addition of sulphanilamide solution prior to hydride generation. 18 Bias evaluation by spiking sea-water samples (Table 3) showed excellent freedom from interference. It is recommended that such standard addition experiments should be performed periodically to check for possible interference effects. This is particularly so with samples expected to contain any poten- tially interfering species. Analytical Performance Data The calibration graph was linear from 0 to 5 ng, corresponding to a concentration range of 0-100 ng 1-1, if a 50-ml sample size is employed. Within-batch precision was evaluated by repli- cate analysis of ten 5-ng selenium(1V) standards (100 ng 1-1 of Se) and was 5.1%.This was not significantly affected by the pre-reduction step employed for the total selenium determina- tion. The detection limit (based on three times the standard deviation of ten blanks) was 3 ng 1-1. These detection limits can be improved simply by increasing the sample volume to 100 ml or greater and re-optimising the volume of tetrahydro- borate added. Sample throughput was greater than five samples per hour. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. Sample Storage and Analysis of Real Samples The described methods have been used to determine selenium speciation in a number of natural waters and some typical results are given in Table 4. For speciation analysis, samples should ideally be analysed as soon as possible after collection.If delays are unavoidable, it is recommended that samples are filtered as soon as possible after collection and then rapidly frozen.’:’ Conclusions The method described provides a simple, sensitive and relatively rapid means of determining inorganic selenium in natural waters. It has been proved effective for the analysis of natural water samples. This work was supported by a studentship (to S.C.A.) from the Science and Engineering Research Council. References Measures, C. I., Grant, B. C., Magnum, B. J . , and Edmond, J . M., in Wong, C. S . , Boyle, E., Bruland, K. W., Burton, J. D., and Goldberg, E. D., Editors, “Trace Metals in Seawater,” NATO Conference Series IV, Volume 9, Plenum Press, New York, 1983. Cutter, G. A., Science, 1982, 217, 829. Measures, C. I., and Burton, J. D., Nature (London), 1978, 273, 293. Robberecht, H., and Van Grieken, R., Talanta, 1982,29,823. Takyanagi, K., and Wong, G. T. F., Anal. Chim. Acta, 1983, 148, 263. Rankin, J. M., Environ. Sci. Technol., 1973, 7, 823. Chau, Y. K . , and Rikley, J. P., Anal. Chim. Acta, 1965,33,36. Measures, C. I., and Burton, J. D., Anal. Chim. Acta, 1980, 120, 177. Uchida, H., Shimoishi, Y., and Toei, K . , Environ. Sci. Technol., 1980, 14, 541. Shimoishi, Y., Anal. Chim. Acta, 1973, 64, 465. Shimoishi, Y., and Toei, K., Anal. Chim. Acta, 1978, 100, 65. Goulden, P. D., and Brooksbank, P., Anal. Chem., 1974, 40, 1431. Howard, A. G., and Arbab-Zavar, M. H., Analyst, 1981,106, 213. Pierce, F. D., Lamoreaux, T. C., Brown, H. R., and Fraser, R. S . , Appl. Spectrosc., 1976, 30, 38. Cutter, G. A,, Anal. Chim. Acta, 1978, 98, 59. Piwonka, J . , Kaiser, G., and Tolg, G., Fresenius Z. Anal. Chem., 1985, 321, 225. Apte, S. C., and Howard, A. G., J . Anal. At. Spectrom., 1986, 1,221. Cutter, G. A., Anal. Chim. Acta, 1983, 149, 391. McDaniel, M., Shendrikar, A. D., Reiszner, K. D., and West, P. W . , Anal. Chem., 1976, 48, 2241. Paper J611.5 Received March 18th1 1986 Accepted April 15th1 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100379

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 383 Automation of Molecular Emission Cavity Analysis. Determination of Phosphorus lbrahim H. El-Hag and Alan Townshend Department of Chemistry, University of Hull, Hull HU6 7RX, UK The molecular emission cavity analysis (MECA) technique has been fully automated and the new instrument was used for the determination of phosphorus anions in inorganic samples after a rapid batch ion-exchange process. The system consists of an automatic sample dispenser and a microprocessor-controlled timer, which, via a moveable cavity device attachment, controls the movement and the residence time of the cavity in a hydrogen - nitrogen - air flame (where the green HPO emission is monitored at 528 nm) and the cooling time of the cavity outside the flame between successive injections.The relative standard deviation for phosphates is improved from 4.5% for the manual MECA method to 0.9% for the automatic method (using peak-height measurements) and the detection limit is 2.5 ng of phosphorus. Keywords: Molecular emission cavity analysis; phosphorus determination; polyphosphates The reliable determination of phosphorus in a wide variety of materials is of considerable significance in modern analytical practice. Flame photometric methods based on the green HPO emission in hydrogen-based flames have been used for the determination of phosphorus in inorganic samples. 1.2 In these methods, a preliminary column ion-exchange procedure is needed to overcome the depressive effects of metal ions on the green emission.Other flame and non-flame methods based on phosphorus line emission and absorption have also been reported but they suffer generally from the effects of radiation absorption by oxygen and flame gases in the ultraviolet region where the emission and absorption line spectra of phosphorus are strongest.3-6 Molecular emission cavity analysis (MECA)’ has been applied for the sensitive determination of several non-metals and metalloids. The first MECA study of phosphorus was that of Bogdanski,R who reported a detection limit of 10 p.p.m. of P when phosphoric acid was injected into a stainless-steel cavity. No green emission was observed from dilute solutions of metal phosphates because of cationic depressive effects. This problem was partially solved by Osibanjo9 by adding perchloric or sulphuric acid solutions to the analyte.Per- chloric acid overcame the depressive effects of the alkali metals but failed with the alkaline earth metals. Sulphuric acid removed all the depressive effects, and a method for the determination of phosphorus in detergents has been based on this procedure.10 However, the S2 emission from the sulphuric acid extends weakly down to the region of the HPO emission, and this can cause problems at low phosphorus concentrations when the phosphate and sulphur peaks are not time resolved. The reproducibility of MECA is usually dependent on operator skill. When the cavity is introduced into the flame, its temperature increases rapidly. The rate of increase of temperature and the ultimate temperature reached by the cavity depend on several factors, including the material and size of the cavity and its holder, the effectiveness of the cavity holder as a heat sink and the flame composition, factors that are essentially constant.They also include the position and residence time of the cavity in the flame, the initial tempera- ture of the cavity, the time allowed for solvent evaporation before cavity introduction and the rate at which the cavity is introduced into the flame. This latter group of factors must be optimised and controlled carefully, especially when emissions are produced very quickly (within 1-2 s) after introduction of the cavity into the flame. The search for improved precision in the MECA determi- nation of inorganic phosphorus compounds led to the design and construction of the automated analyser that is described in this paper.The technique is based on the automatic injection of microlitre volumes of the analyte into a carbon cavity after a rapid batch ion-exchange process and the microprocessor- controlled movement of the cavity into and out of the flame. Experimental Reagents Stock solutions (1000 p.p.m.) of phosphorus as inorganic phosphorus compounds were prepared using doubly distilled water. Further dilutions were made as required. Dowex 50W-X8 (H+) cation exchanger (BDH Chemicals) was used for the pre-treatment of the samples before their injection into the cavity. Automated MECA Instrument The automated instrument (Fig. 1) consists of a moveable cavity device (A), controlled by a microprocessor (B), a Varian ASD-53 automatic sample injector with rotating sample carousel (D) and control unit ( C ) , a Pye Unicam SP900 flame emission spectrometer (E) and an Oxford 3000 recorder The moveable cavity device (MCD) (Fig.2) is used to introduce the cavity reproducibly and automatically into and out of the flame. It consists of a synchronous motor that is connected by a flexible joint to a gear box. The gear box is attached to an egg-shaped cam, which has four microswitches riding on its periphery. These microswitches are connected to a microprocessor-controlled timer. Two of these micro- switches are used to stop the current to the motor when the cavity is inside or outside the flame, so that the cavity remains in or out of the flame for a pre-determined period, programmed in via the microprocessor. A third microswitch is used as a delay switch to increase the cooling time of the cavity, when it is necessary, for any desired length of time. The fourth microswitch synchronises the performance of the automatic sample injector with the MCD.As it rotates, the cam tilts the cavity from the vertical position (position of injection of the samples) to the horizontal position into the flame (Fig. 2). The movement of the cavity out of the flame occurs in a reverse fashion. The microprocessor-controlled timer was used to control these periods of time successively. It consists of an INS 8060N monolithic, 8-bit , N-channel microprocessor with an address capability of 64K. It has direct memory access, two sense inputs, multiprocessor, bus logic and a serial input - output (F) *384 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 Fig. 1. Automated MECA analyser: (A) MCD; (B) microprocessor - timer; (C) sample injector control unit; (D) sample injector and carousel; (E) spectrometer; and (F) chart recorder Fig. 2. Moveable cavity device moving from injection position into flame (the cavity illustrated is vitreous carbon, not that used in the present investigation) port. The instructions are capable of operating the multi- addressing modes of the immediate programme counter. A Varian ASD-53 automatic sample dispenser (Varian Techtron, Springvale, Australia) was used for the injection of solutions into the MECA cavity. As the ASD-53 was designed basically to function with carbon furnace atomic absorption instruments, a simple modification of its electrical circuitry was needed to adapt it to MECA, so that a pulse sent from the MCD via the fourth microswitch initiated the injection process.The ASD-53 included a rotating carousel module, which held up to 25 samples, and an injector control unit. In addition, for its carbon furnace function, water and inert gas flows need to pass through the controller en route to the furnace. When these flows decrease below set limits, the dispensing unit is closed down. Although water and inert gas flows were not needed for the MECA measurements, it was Table 1. Optimum conditions used for the determination of inorganic phosphorus compounds by automated MECA Condition H2 flow-rate .. . . . . . . . . N2 flow-rate . . . . . . . . . . Air flow-rate . . . . . . . . . . Distance of cavity front inside the flame Cavity centre to burner top distance . . Residence time of cavity in flame . . Coolingtimeofcavity . . . . . . Wavelength . . . . . . . . . . Slit width . . . . . . . . . . * 1-2 nm in the presence of sulphate. Optimum value . . . . 2.91min--l . . . . 5.0lmin-1 . . , , 5.51min-1 . . , . 6mm* . . . . 4mm . . . . 10s . . . . 90s . . . . 528nm . . . . 0.3nm decided not to interfere with the electrical circuitry to override the control signals. Instead, a flow-rate of 2.5 1 min-l of tap water and 800 ml min-1 of air were passed through the controller, and then led to waste. A Pye Unicam SP900 atomic absorption - flame emission spectrometer with an EM1 S-11 photomultiplier tube was used for the emission measurements. The Cavity The cavity used in this study was a 4 x 4 mm diameter carbon cavity at the end of a 37 mm long carbon rod (Poco Graphite, Decatur, TX, USA). The rod is housed in an aluminium cavity holder, which was attached to the MCD (Fig.1). Recommended Procedure To a 25-ml aliquot of the phosphorus-containing solution (5-100 pg ml-1 of P) in a 50-ml beaker are added 5 g of ion-exchange resin, and the mixture is stirred on a magnetic stirrer for 5 min. The solution is transferred into a vial in the carousel unit for the automatic MECA measurements. At the beginning of the analysis of the first sample, the micropro- cessor unit sends a pulse signal via the MCD to the control unit of the ASD-53 to begin the injection of the samples into the cavity.The cavity then moves into a pre-selected region in the flame by the action of the MCD. After the emission signal has been recorded, the cavity is returned to its original position for cooling before the next sample is injected. After the 25th sample has been analysed, the injector unit sets itself at the standby position ready for a new carousel to be loaded. The analysis time is a factor of the heating and cooling times of the cavity between the successive injections. The optimum analysis time in this study is 2 min for each reading. The optimised conditions for the automated determination of inorganic phosphorus compounds are given in Table 1. Procedure for Fertiliser Analysis A l-g sample of fertiliser is accurately weighed, transferred into a 250-ml conical flask and 200 ml of doubly distilled water are added.The mixture is digested on a hot-plate for 30 min, shaking every 3 min. The resulting solution is filtered through a Whatman No. 42 filter-paper and diluted to volume with doubly distilled water in a 250-ml calibrated flask. A 12.5-ml portion is diluted to 25 ml with doubly distilled water in a 50-ml calibrated flask, 2 g of Dowex 50W-X8 cation-exchange resin are added and the mixture is stirred with a magnetic stirrer for 5 min. A 5-p.1 aliquot of the solution is analysed by automated MECA with the cavity at the edge of the flame.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 385 60 50 > E 40 > v) C a, c-. .- 30 .- c 0 v) v) .- .- E 20 w 10 Ib) 0 6 4 Timeis Fig.3. (a) before and ( b ) after batch ion exchange Emission - time profiles from 40 pg m1-I of P as NaH2P04: 100 90 80 70 > E .g 60 . v) C e .5 50 C 0 .- 40 .- E w 30 20 10 0 50 100 150 200 250 Amount of phosphorusing Fig. 4. Calibration graphs for (A) sodium phosphite; (B) sodium hypophosphite; (C) hypophosphorus acid; (D) sodium dihydrogen phosphate; (E) sodium tripolyphosphate; and (F) sodium pyrophos- phate 60 50 > ‘r 40 E .w .- v) c 30 .- c 0 .- $ 20 E .- w 10 0 Amount of phosphorusing Fig. 5. sodium dihydrogen phosphate; and (C) sodium trimetaphosphate Calibration graphs for (A) sodium tetrametaphosphate; (B) Results and Discussion When a dilute solution of sodium phosphate is injected into a carbon cavity, only a faint green HPO emission is obtained [Fig.3(a)]. However, if the solution is first treated with a few grams of cation-exchange resin for 30 s, an intense green emission is obtained, which gives a single emission - time peak [Fig. 3(b)]. Therefore, treating all sample solutions with the ion-exchange resin, as described under Recommended Pro- cedure, before automated MECA, gives sensitive, reprodu- cible, single-peak emissions from a wide range of phosphates. Linear calibration graphs are obtained for various phosphorus anions in the range 25-250 ng of P in 5 pl (Figs. 4 and 5). Some analytical characteristics are summarised in Table 2. The calibration graphs for ortho-, pyro- and tripolyphos- phate are almost identical, and those for tri- and tetrameta- phosphate are also similar.The t, values for all the com- pounds are also similar. This indicates the likelihood that the major species generating the HPO emission is common to all these compounds, and is most likely to be orthophosphate produced by hydrolysis of the polyphosphates in the acidic solution on heating in the flame. It also indicates that it might be possible to determine total phosphorus(V) anions from a single MECA measurement. Phosphite and hypophosphite show increased sensitivities compared with the phosphates, and their t , values are smaller, indicating volatilisation at a lower cavity temperature, because of their more ready decomposition with release of volatile products such as phosphine. Such an increase in sensitivity for emissions produced at lower cavity temperatures has previously been observed for sulphur compounds,7 and is a manifestation of the Salet phenomenon.11.12 A major advantage of the automated instrument is the improvement in reproducibility, resulting from the improved control over the introduction of the cavity into the flame.The relative standard deviation for ten measurements of 250 ng of phosphorus as phosphate has been improved from 4.5% for the system operated by manual sample injection and manual cavity introduction to 0.9% for the automated procedure. Likewise, the values for 250 ng of phosphorus as phosphite have improved from 10-20% to 2.6%. The faster emission from the phosphite emphasises the need for reproducibly controlling cavity introduction into the flame. The limits of detection (20) were 2.5 ng of P for phosphite and 7.5 ng of P for phosphate in 5-pI samples.The technique was applied to the determination of phos- phorus in a fertiliser sample (Standard Sample No. 1 for Fertilizer Analysis, Fertilizer Manufacturer’s Association). The certificate value was 4.8% phosphorus. The automated MECA procedure also gave 4.8%. The relative standard deviation was 3.6% for ten measurements on a final fertiliser solution. The fertiliser solution also contains sulphate, which gives a blue S2 emission. However, if the cavity position is changed from the centre of the flame (the coolest region) to close to the edge (where the flame is hotter and more oxidising) the blue emission is almost completely eliminated. The sensitivity for phosphorus in that cavity position is 33% less than in the flame centre.Conclusions The automated device improves the precision of MECA determination of phosphorus, and enables series of determi- nations to be carried out without any operator intervention. The slowest part of the operation is the cooling of the cavity before injecting the next sample. This could be speeded up by installing a cold-air blower, or by use of a water-cooled ~ a v i t y . ~ The device can also be used to determine sulphur compounds, which will be described later.386 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 Table 2. Analytical characteristics of the inorganic phosphorus compounds investigated Compound NaH2P02 . . . . . . Na2HP03.5H20 . . . . (NaPO& . . . . . . (NaP04)3.10H20 .. Na4P2O7.10H20 . . . . Na5P3OI0 . . . . . . NaH2P04.2H20 . . . . H3P02 . . . . . . Sensitivity/ mV ng-I P 0.30 0.31 0.40 0.16 0.24 0.19 0.21 0.20 Relative standard deviation, O h * 3.2 3.2 2.6 2.9 2.5 2.4 2.7 2.2 * Based on ten measurements of 150 ng of phosphorus. t Time from cavity insertion into flame to maximum emission signal. t,/st 1.1 1.1 0.8 1.6 1.8 1.8 2.0 2.0 Detection limithg 3.0 3.0 2.5 7.5 10 10 10 10 The authors thank Mr. S. Travers for constructing the MCD arld Mr. G. E. Palin for the design and construction of the microprocessor-controlled timer. I. H. El-Hag thanks the Government of the Sudan for the award of a research scholarship. References 1. 2. 3. 4. 5. Syty, A., Anal. Lett., 1971, 4, 531. Elliott, W. N., and Mostyn, R. A., Analyst, 1971, 96, 452. Skogerboe, R. K . , Gravatt, A. S . , and Morrison, G. H., Anal. Chem., 1967, 39, 1602. Greenfield, S., Jones, 1. L., and Berry, C. T., Analyst, 1964, 89, 713 Greenfield, S., McGeachin, H. McD., and Smith, P. B., Talanta, 1975, 22, 553. 6. 7. 8. 9. 10. 11. 12. Havezov, I., Russeva, E., and Jordanov, N., Fresenius Z . Anal. Chem., 1979, 296, 125. Bogdanski, S. L., Burguera, M., and Townshend, A., CRC Crit. Rev. Anal. Chem., 1981, 10, 185. Bogdanski, S . L., PhD Thesis, University of Birmingham, 1973. Osibanjo, O., PhD Thesis, University of Birmingham, 1976. Osibanjo, O., Al-Tamrah, S. A., and Townshend, A., Anal. Chim. Acta, 1984, 162, 409. Salet, G., C. R. Acad. Sci., 1869, 68, 404. Salet, G., Bull SOC. Chim. Fr., 1870, 14, 182. Paper J6f I 4 Received March 6th, 1986 Accepted April 22nd, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100383

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 387 Quality Assurance of Analytical Data, with Special Reference to the Determination of Lead and Cadmium in Biological Samples Chandrakant B. Pandya, Tushar S. Patel, Gaurang M. Shah, Natubhai G. Sathawara, Veljeebhai G. Patel, Dinesh J. Parikh and Barid Baran Chatterjee National Institute of Occupational Health, Meghani Nagar, Ahmedabad 380 0 16, India An atomic absorption spectrometric system encorporating a Delves cup was used for the determination of lead and cadmium in blood, and simple conventional flame atomic absorption spectrometry (AAS) following the dry ashing of kidney cortex samples in a muffle furnace, using a temperature control programme, was used for the determination of cadmium. Quality control (QC) samples were run and analytical data were subjected to strict statistical assessment.The results were found to be satisfactory for lead in blood, Pb-B ( Y = 0.9809~ + 1.4201 pg I-’ and r = 0.98621, for cadmium in blood, Cd-B ( Y = 0.9180~ + 0.2871 pg I-’ and r = 0.9813) and for cadmium in kidney cortex ( Y = 0.9726~ + 1.7874 mg kg-1 and r = 0.9872). Keywords: Quality control; lead determination; cadmium determination; atomic absorption spectrometry; biological samples There is increasing demand for the development of improved analytical methodologies, and as a result various advances in the physical sciences are soon incorporated into analytical equipment, providing the analyst with a selection of methods for given purposes. The choice of analytical methods depends on many factors, e .g . , the nature and size of the samples, the nature of the substances being determined and the concentra- tions at which they are expected to be present, the interfering substances likely to be encountered, the level of specificity and sensitivity of method, the nature and complexity of the preparatory steps and the cost involved per sample analysed. The role of trace metals in human health and diseases, the need for assessment of anthropogenic environmental expo- sure to a variety of metals widely used or being used increasingly in manufacturing processes and for their determi- nation in tissues and biological fluids, is of interest to toxicologists and analytical chemists. There is thus a need to devise increasingly sensitive methods for the determination of these metals at ultra-trace levels in samples of limited size.The reliable determination of trace metals in biological samples still remains an analytical problem of some concern, and of growing importance. This paper describes the determi- nation of lead and cadmium in certain human tissues. Various procedures had been tried for the determination of these metals in biological materials, and it was evident that atomic absorption spectrometry had distinct advantages over other methods. In 1979, the National Institute of Occupational Health (NIOH), Ahmedabad, India, had agreed to participate in a UNEP/WHO pilot project on “Assessment of human expo- sure to lead and cadmium through biological monitoring.” During the initial stages of this project, a programme was initiated to investigate the use of the Delves cup technique for the determination of lead and cadmium in blood and a simple conventional flame AAS for the determination of cadmium in kidney cortex.Experimental A pooled sample of human blood was obtained from the city hospital, which was haemolysed and stored. This blood sample was used to prepare calibration graphs for lead and cadmium using the standard additions method. We received quality control (QC) samples (blood and kidney cortex) periodically from 1979 to 1985 under the UNEPIWHO programme, from the Co-ordinating Institute (CI), National Institute of Environmental Medicine and Karolinska Institute (Department of Environmental Hygiene), Stockholm, Sweden. These samples were stored in a deep freeze at -20°C until required for analysis.Instrumentation An atomic absorption spectrophotometer (PE-373 AAS) equipped with background correction facilities, Delves cup assembly, hollow-cathode lamps for lead and cadmium and a PE-56 chart recorder, was used. Preparation of Spiked Blood Samples Stock solutions of lead and cadmium (1000 p.p.m.) were prepared by weighing exactly 1.000 g of the powdered metals (99.99% purity, Ventron, Alfa Products, Beverly, MA, USA) and dissolving in a minimum volume of 1 + 1 V/V acid (HN03 for lead and HC1 for cadmium). The final dilution to 100 ml was made with 1 M HN03 for both metals. Lead ( 5 and 20 p.p.m.) and cadmium (0.05, 0.20 and 0.25 p.p.m.) standard solutions were prepared daily by serial dilution of the stock solutions with de-ionised water.The blood samples were accurately spiked with the standard solutions, as shown in Tables 1 and 2 for lead and cadmium, respectively. Table 1. Preparation of spiked blood samples for use as standard additions of lead Volume of blood takedm1 4.80 4.80 4.80 4.80 4.80 4.80 Distilled waterhl 0.20 0.10 0.00 0.10 0.05 0.00 Pb standard solution added 0.1 ml of 5 p.p.m. 0.2 ml of 5 p.p.m. 0.1 ml of 20 p.p.m. 0.15 ml of 20 p.p.m. 0.2 ml of 20 p.p.m. - Pb content of spi ked blood sarnplel I%-’ 0 100 200 400 600 800 Table 2. Preparation of spiked blood samples for use as standard additions of cadmium Cd content of Volume of Distilled Cd standard spiked blood blood/ml water/ml solution added sarnple/pg I - 0 4.80 0.20 - 4.80 0.10 0.1 rnl of 0.05 p.p.m.1 .o 4.80 0.00 0.2 ml of 0.05 p.p.rn. 2.0 4.80 0.10 0.1 mlof0.2p.p.m. 4.0 4.80 0.00 0.2 ml of 0.2 p.p.rn. 8.0 4.80 0.00 0.2rnlof0.25p.p.m. 10.0388 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 Sample Treatment Blood samples were brought to room temperature prior to analysis. They were then thoroughly homogenised in a vortex mixer. Lyophilised, powdered, kidney cortex samples were kept overnight at 90-100 "C and then cooled. A portion (50 mg) of the dried kidney cortex powder was taken in a porcelain crucible and dry ashed in a programmable muffle furnace. The muffle furnace was programmed for the temperature to rise at a rate of ca. 50 "C h-1 until a maximum temperature of 450 "C was gradually reached. The samples were kept at this temperature for ca.20 h to ensure complete ashing, after which the muffle furnace was turned off. The samples were slowly brought to room temperature whilst still inside the furnace. The residues were weighed and dissolved in 15 ml of 1 M HN03. Instrumental Analysis A Delves cup assembly was installed in the flame compart- ment of the AA spectrometer. A 10-pl sample of blood was placed in a pre-conditioned, metal free, clean nickel cup, using an Eppendorf pipette. Five cups were used for each blood sample. The sample cups were held in a sample tray and dried in an oven at 140 "C for about 2 min. The tray was taken out of the oven, the cups removed and then placed individually in a loop of the Delves cup assembly. The combustion of the organic matter in the sample was accomplished close to the flame area of the burner.When combustion had ceased, the cups were inserted into the flame, just below the opening of absorption tube and the absorption signals were then measured. The 283- and 228-nm lines were used for lead and cadmium determinations, respectively. The mean value of five readings for each sample was used for further treatment of the (RSD = 4.9%) 0.032 0.035 0.033 2 0.0015 (RSD = 4.6%) 0.013 k 0.001 (RSD = 7.8%) Time - Fig. 1. indicating the good reproducibility of analysis by Delves cup AAS Typical peaks recorded for cadmium in three blood samples, data. Any samples showing values 10% above the relative standard deviation (RSD) of all the five determinations were subjected to re-analysis. Fig. 1 shows typical peaks for the determination of cadmium in blood by the Delves cup technique.The first peak of each trace is for the cadmium in the blood. The true cadmium peak was separated from the smoke peak, generated by the high organic content of the blood matrix, by complete combustion of the organic matter near to the flame area of the burner prior to inserting the cup into the flame. The second peak is due to molecular absorption of blood electrolytes, e.g., sodium and potassium. For low cadmium blood levels, the second peak overlaps and dominates the true cadmium peak but, with experience, the true cadmium peak can be separated and identified by skilful operation of the recorder. Conventional flame AAS was used for the determination of cadmium in the kidney cortex samples prepared in the manner described above.Statistical Treatment of the Analytical Data The reference values for various sets of QC samples received periodically from CI were not known to us, but we eventually received them as feedback, alongwith comments on our reported values. These data were used further for the statistical evaluation. UNEP/WHO provided certain guide- lines for the statistical treatment of the analytical data, to determine whether results were accepted or rejected. A detailed account of the statistical treatment of data has been produced by Cederlofl and information can also be found in the UNEPNHO report.2 However, the details given in Table 3 may be useful, as the main statistical calculations given here are based on them. The equation used for calculating the acceptability, inter- vals is given as: c= q m / 2 ) -6; .. . . * * (1) where 2 is the standard normal cumulative probability, an independent variable in the normal integral function. The value of 6; can be obtained from the equation s:=a-;,[-+--] 1 (xi-.u)2 . . . . n (n-l)S,2 where 69 = standard error of the estimate Y ; hYx = error of the method; x i = low level (LL) or high level (HL) value; Sx = estimated standard deviation of the values ; X = mean of the x-values; and n = number of observations. The second unknown variable in equation (1) is Z(mi2). This can be obtained from the equation d = 6 9 [ q - m / 2 ) + Z([3/2)+1 - * * * (3) where d = difference between the mean x and the upper/lower criteria line; 2(m,2) = the 2 value for half the a error; and Z(p/2) = the 2 value for half the p error (a and p represent the probabilities of errors).For 90% power (p = loo/,) p/2 will be 0.05, corresponding to a 2 value of 1.645.1 Inserting values into equation (3), Z (m/2) may be obtained and equation (1) can be solved. Table 3. Accepted parameters for statistical evaluation of analytical data MAD line from the Assumed error Applied working range regression line Y = x Determination of method (LL to HL) set as indicated Pb-B 10 pg 1-1 100-400 vgl-1 Y = x * (0.1x + 20) pg 1-1 Cd-B 0.5pg1-l 1-15 pg 1-1 Y = x k (0.1x + 1) pg1--' Cd-kidney 3.0mg kg-1 50-400 mg kg-1 Y = x k (0.15~) mg kg-1 cortex (dry mass) (dry mass)JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 389 0.28 0.24 a m g 0.20 5 0.16 n a 0.12 0.08 0.04 4 A / / 0 20 40 60 80 100 Cadmiumipg 1-1 Lead/pg 1-1 I 1 , I I l l 0 200 400 600 800 1000 Fig.2. Typical calibration graphs for lead (A) and cadmium (B) in blood using the standard additions method. Results obtained were 2.9 pg I-' of cadmium and 120 pg I-' of lead 400 350 7 300 - 0) 5. $ 250 - m p 200 c 0 8 150 [r 100 50 I I I 1 I I 1 50 100 150 200 250 300 350 400 Reference value/pg I - 1 Fig. 3. Regression line based on 82 reported values for lead in blood. The outer solid lines indicate the MAD lines and the dotted lines the acceptance lines. The statistical evaluation of data shown is based on the parameters given in Table 3 and the regression equation given in the text 12 - 11 10 - 9 - 3 8 - - r [J, 5 - 9 7 - e 5 - u 0 U $ 6 - 4 - 3 - 2 - 1 - 1 2 3 4 5 6 7 8 9 1 0 1 1 12 Reference value/pg I-' Fig.4. Regression line based on 64 reported values for cadmium in blood. The outer solid lines indicate the MAD lines and the dotted lines the acceptance lines. The statistical evaluation of data shown is based on the parameters given in Table 3 and the regression equation given in the text 400 350 r b, 300 Y E" 2 250 3 m - 8 200 C a L al 5 150 U 100 50 Results and Discussion Many workers have found the use of the Delves cup with the flame AAS technique to be satisfactory for determining lead and cadmium at very low concentrations.>s It is claimed to be a rapid method with sufficient sensitivity, accuracy and reproducibility. As no prior treatment of the samples is required, there are no opportunities for loss or contamination.Typical calibration graphs for lead and cadmium in blood obtained by the standard additions method are shown in Fig. 2. Inspection of the analytical data from the standard additions of lead and cadmium to blood show that the RSD values are 8% at a lead concentration of 120 pg 1-1, (Le., zero addition of Pb) and 10% at a concentration of 920 pg 1-1, ( L e . , with the addition of 800 pg 1-1 of lead to blood), and 4% at a cadmium concentration of 2.9 pg 1-1 (i.e., zero addition of cadmium) and 2% at a concentration of 12.9 pg 1-1 ( i e . , with the addition of 10 pg 1-1 of cadmium to blood) to the samples and the average recoveries of the lead and cadmium were 107O/0, range 105-110%, and 104%, range 98-1 lo%, respectively.50 100 150 200 250 300 350 400 Reference valuelmg kg-1 Fig. 5. Regression line based on 48 reported values for cadmium in kidney cortex. The outer solid lines indicate the MAD lines and the dotted lines the acceptance lines. The statistical evaluation of data shown is based on the parameters given in Table 3 and the regression equation given in the text Fig. 3 shows the regression line for blood lead, with its data points well within the acceptance region. The regression equation: Y = 0.9809~ + 1.4201 with r = 0.9862 was satisfactory. The outer solid lines indicate the maximum allowable deviation (MAD) lines and the dotted lines are the acceptance lines. Both of these lines were fixed using the parameters given in Table 3 and equation (1). Evaluating at the LL (x = 100 pg l-l), the regression equation gave a function value of 99.5 pg 1-1, which fell within the 72.8 and 127.2 pg 1-1 range, the accepted interval.Evaluating at the HL (x = 400 pg 1- l), the regression equation gave a function value of 393.8 pg 1-1, which was within the390 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 accepted interval of 343.6-456.4 pg 1-1. Thus, both the function values were well within the appropriate accepted intervals and hence the results were accepted. Similarly, Fig. 4 shows the regression line for blood cadmium, with all its data points well placed within the acceptance region. The regression equation, Y = 0.9180~ + 0.2871 and r = 0.9813 was satisfactory. Both the function values were accepted, i.e., 1.7 pg 1-1 at the lower evaluation point of Cd-B (x = 1.5 pg l-l), which is within the 0.5-2.5 pg 1-1 range, and 11.3 pg 1-1 at the higher evaluation point of Cd-B ( x = 12 pg 1-I), which is within the 10.1-13.9 pg 1-l range.Fig. 5 shows that the data obtained for cadmium in kidney cortex were also acceptable as they fell well within the acceptance lines. The regression equation obtained, Y = 0.9726~ + 1.7874 with r = 0.9872, was satisfactory. If evaluation is carried out at the LL (x = 50 mg kg-I), the regression equation gave a function value of 50.4 mg kg-l, v:hich fell within the accepted interval of 43.8-56.3 mg kg-l and evaluation at the HL ( x = 400 mg kg-1) gave a function value of 390.8 mg kg-1, within the accepted interval of 341.6-458.4 mg kg,-l and hence the results were accepted.Conclusions The analytical results obtained for the determination of lead and cadmium in blood using the Delves cup technique and flame AAS and for determining cadmium in the kidney cortex by dry ashing followed by conventional flame AAS, showed good correlation with the values of the reference materials used for quality control of the analytical procedures. The Delves cup technique is a simple, rapid and sensitive method, capable of providing reproducible results for the determination of trace levels of lead and cadmium. The complete dry ashing of kidney cortex samples at a relatively low temperature preven- ted the loss of cadmium by volatilisation that is encountered when higher temperatures in the muffle furnace were used. The residue was dissolved in 1 M HN03, and when analysed by a standard flame AAS method provided accurate and repro- ducible results.The reference samples of blood and kidney cortex were provided by the CI, Stockholm, under the UNEP/WHO programme. The training on the use of Delves cup technique with flame AAS was also provided through UNEPIWHO funds. The authors thank Mr. Birger Lind, Research Chemist of the Industrial Hygiene Department, Karolinska Institute, Stockholm, Sweden for the specialist training of the chemist for this specific project. 1. 2. 3. 4. 5. 6. 7. 8. References Cederlof, R., “Some Aspects on the Power of Quality Control Procedures,” working paper to the meeting of the UNEP/ WHO pilot project on assessment of human exposure to pollutants through biological monitoring, metal component, Stockholm, 27-30 May, 1980. Marie, V., Editor, “Assessment of Human Exposure to Lead and Cadmium Through Biological Monitoring,” UNEP/WHO report, National Swedish Institute of Environmental Medicine, Karolinska Institute, Stockholm, Sweden, 1982, pp. 27-31. Delves, H. T., Analyst, 1970, 95, 431. Fernandez, F. J., and Kahn, H. L., At. Absorp. Newsl., 1972, 11,33. Olsen, R. D., and Jatlow, P. J., Clin. (:hem., 1972, 18, 1312. Pandya, C. B., Patel, T. S., Shah, G. M., Parikh , D. J., and Chatterjee, S . K., in Shanker Das, M., Editor, “Trace Analysis and Technological Development,” Wiley Eastern, New Delhi, India, 1983, pp. 195-199. Friberg, L., and Marie, V., Environ. Res., 1983, 30, 95. Elinder, C. C., Friberg, L., Lind, B., and Jaworid, M., Environ. Res., 1983, 30, 233. Paper J6137 Received March llth, 1986 Accepted May 9th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100387

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 391 An External Quality Assessment Scheme for Trace Elements in Biological Fluids Andrew Taylor and R. J. Briggs Supra-regional Assay Service Trace Elements Laboratory, Department of Clinical Biochemistry, St. Luke‘s Hospital, Guildford, Surrey GUI 3NT and Robens Institute of Industrial and Environmental Health and Safety, University of Surrey, Guildford, Surrey GU2 5XH, UK The role for external quality assessment (EQA) schemes is illustrated by published results, which are probably inaccurate. The organisation and management of an EQA scheme for trace elements in biological fluids is described to show how the analytical performance of participating laboratories can be determined. The wider roles fulfilled by EQA schemes are discussed with examples of how the trace elements EQA scheme has been used to demonstrate the reliability of analytical data.Keywords: Quality assessment; trace elements; biological fluids Recent developments in analytical techniques make it possible to determine the concentrations of elements at low (ng ml-1) levels using samples of only 5-50 pl. 1 However, measurements at these levels may be complicated by problems such as contamination during sample collection, storage or analysis, adsorption of analyte from sample on to the container and interferences associated with the complex nature of a biologi- cal matrix. The influence of improved control over contamina- tion and other factors has been well illustrated by the work of Versieck and Cornelis .2-4 Their various reports have demon- strated that reference values published from different labora- tories are now showing some consistency at concentrations that are generally considerably lower than were stated less than ten years ago.Despite recognition of these problems, results are still reported which, if examined critically, appear to be extremely unlikely. Among the anomolous results presented in one study5 were the “normal” serum lithium concentrations of 0.48 and 0.78 mmol 1-1. These results are typical of those found in subjects receiving prophylactic lithium to treat depressive illness6 and would not be found among many of the general population. A second series of results that is difficult to interpret is from work with neonates.’ Although there are few other studies with which to compare, some of these data also appear to be quite unrealistic.The concentrations of gold in serum (mean k SD) were 1.2 k 0.45 yg g-1, which are similar to those found in infants of mothers who received intramuscular injections of gold throughout pregnancy.8 The concentrations of copper were equivalent to levels found in cases of copper toxicity and quite dissimilar from those reported by other workers.9 Other unlikely results can be found in the published literature. In recent years a different situation has developed with ultra-trace elements, where probable concentrations in blood are less than 10 pg 1-1. It is difficult to achieve reliable results for measurements made close to analytical detection limits, particularly where contamination is not easy to avoid.The establishment of reference ranges for healthy and other populations requires care and may be facilitated by inter- laboratory comparisons of results.3-4 External quality assessment (EQA) schemes should, if properly used by participants, prevent the publication of results of doubtful validity. Furthermore, participants with performance judged to be satisfactory by their EQA data can generally be confident about their results from samples having very low concentrations. ‘0.11 An EQA scheme for trace elements in biological fluids was established to provide objective evaluation of the analytical performance of labora- tories that carry out these measurements. This paper describes the operation of the scheme and its application to biomedical analytical atomic spectrometry.Experimental The combination of analytes and sample types that are included in the scheme are given in Table 1. The approximate number of participants is 110 (September 1985) and these include laboratories from more than 15 countries throughout the world. Not all participants are included for every analyte - matrix combination. The scheme is organised to give a succession of six-month cycles that commence in April and October each year. Every month, three specimens of each matrix are sent, as appro- priate, to the participants. Each cycle includes, therefore, a total of 18 specimens. (The programme for aluminium in water and dialysis fluids is a little different with fewer samples, sent bimonthly, throughout an annual cycle.) Samples of serum, blood and urine are prepared for distribution as described below.All laboratory-ware is cleansed by soaking overnight in 10% V/V HCl and rinsing thoroughly with de-ionised water. Batches of tubes used for storage and distribution of samples after preparation are screened and have been shown not to cause adventitious contamination to the contents. Serum. Chelex 100 ion-exchange resin (Bio Rad Labora- tories Ltd.) is added to a single batch of sterile horse serum (Sera-Lab Ltd., Crawley Down, Sussex, UK) to reduce the endogenous concentrations of Cu and Zn to low to normal levels. Resin, 50 g 1-1, is added and the serum continuously stirred for 24 h. The suspension is centrifuged and the serum decanted into nine calibrated flasks.The concentration of Au, Cu and Zn are increased in eight of these pools by the addition of calculated amounts of standard solutions (BDH Spectrosol solutions) to the chelex-treated serum. A series of nine pools Table 1. Analytes and samples included in the Trace Elements Quality Assessment Scheme. Figures in parentheses give the approximate number of participants for that analyte - sample combination (September 1985) Waterldialysis Serum Blood Urine fluid Aluminium (50) Cadmium (17) Cadmium (12) Aluminium (35) Copper (65) Lead (30) Mercury (20) Gold (6) Selenium (12) Zinc (75)392 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 Table 2. Blood Pb results in samples prepared after erythrocyte lysis using saponin or ultrasonication. Results are mean k SD of all values reported from two distributions Sonicated samples Saponified samples Pb added/ Mean * SD/ Recovery, Pb added1 Mean k SD/ Recovery, pmol 1- 1 pmol 1- O/O pmol 1- pmol I- O/O - 0 0.27 k 0.059 0 0.28 k 0.047 - 0.8 1.12 k 0.059 106 1 .o 1.27 k 0.055 100 1 .o 1.41 k 0.090 114 1.6 1.96 + 0.122 106 2.0 2.35 k 0.104 104 - - - 2.8 3.04 k 0.106 99 3.2 3.47 k 0.160 100 0 0 0 .0.. 0.. 0 ...... ......... I 1 <X> Blood lead/pmol I-' I 1 0 1 .o 2.0 3.0 4.0 Sample number . . . . . . . . . . 72 Mean . . . . . . . . . . . . . . 3.07 pmol l-1 Standard deviation . . . . . . . . . . 0.21 pmol I-' Range of results . . . . . . . . . . 2.67-3.62 p o l 1-1 Numberofvaluesreported . . . . . . 26 Coefficient of variation . . . . . . . . 6.73% Trace Element Quality Control: Blood Lead Results 2 3.62 4 2.90 5 3.15 8 3.03 10 2.95 15 2.93 17 2.92 25 2.84 26 3.04 31 3.15 35 2.98 36 2.93 38 2.67 23 3.35 47 3.28 50 2.94 64 3.19 67 3.12 70 3.00 72 2.80 78 3.06 79 3.23 84 3.34 89 3.11 93 3.33 103 2.93 Fig.1. A page from a monthly report for lead in blood showing the histogram, calculations and tabulation of results alongside the appropriate laboratory code number supplemented with Al and Se are similarly prepared but without the preliminary chelex treatment. The samples are mixed and dispensed into labelled tubes (2.0-ml trace-element tubes, Tek Lab, Sacriston, Durham, UK). Blood. Human blood is collected from a volunteer and contains dipotassium ethylenediaminetetraacetic acid , 2 mg ml-1, to prevent coagulation. (Samples are screened for Australia Antigen and HTLV I11 antibodies before any further preparation is undertaken.) A comparison of ultra- sonic energy and the use of saponin to produce lysis of erythrocytes was made where two portions of blood from the same donor were subjected to both techniques and thereafter treated identically.No difference between the results reported by participants were found with respect to precision and recovery (Table 2), and erythrocyte lysis is routinely accompli- shed by the addition of saponin to the blood. The haemolysed blood is transferred into calibrated flasks and the concen- trations of Cd and Pb augmented as for elements in serum. The blood is mixed and then dispensed into labelled tubes. Urine. Human urine collected from volunteers who have been screened and found not to carry Australia Antigen or HTLV I11 antibodies is placed into calibrated flasks and supplemented with Cd and Hg.Samples are mixed and dispensed into labelled tubes. All samples of serum, blood and urine are subjected to gamma-irradiation (minimum dose 24 kilogray) to destroy any bacterial contamination that may have occurred during preparation. These specimens are stored at -20 "C until despatched. To determine that samples have been properly prepared and that there has been no contamination or other cause for between-tube variation, the elements included in the EQA scheme are measured in this laboratory in at least ten specimens from each batch prior to despatch. Preparation of the serum samples is carried out at the beginning of a six-month cycle and each of the nine pools is distributed for analysis on two separate occasions to allow determination of between-batch precision. Blood and urine samples are prepared monthly.Accurate supplementation of the pools with the analytes permits subsequent calculations of recoveries of the amounts added. Analysis of Results Participants are asked to return their results within one month of the date of despatch. Results received before that time are used to prepare reports for each sample - matrix - analyte. An initial calculation of the mean and standard deviation (SD) is made and any result outside the range "mean k 3SD" is omitted from the report. This monthly report gives the consensus mean, SD and coefficient of variation, a histogram of distribution and a tabulation of results (Fig.1). At the end of a 6-month cycle an end of term report is prepared, which summarises the 18 results for each sample - matrix - analyte combination and presents an assessment of analytical performance based upon parameters of: proximity to consensus mean (x - X), difference between the results for samples analysed on two occasions (Xl - X2) and recovery of added analyte (%R). For each of these parameters we have adopted the procedure developed by Yeoman (described by Vahterlo) for the establishment of targets or markers of satisfactory performance. These targets are based on what is necessary for clinical purposes and what can be achieved with available analytical techniques. A graphical example of this approach is shown for serum zinc in Fig.2. Limits are selected at a high and a low level (k1.5 pmol 1-1 at 20.0 pmol l-1 and kl.0 pmol l-1 at 4.0 pmol l-1, for Zn in serum) and these are joined to give targets that cover all concentrations. A second series of targets, closer to the mean, is drawn from limits that are half of those of the first set. In this way inner (A) and outer (A + B + C) target zones are established (Fig. 2). The limits selected for each analyte for the preparation of these zones are given in Table 3. Those shown for Pb in blood were first used by Yeoman,'O others were suggested by us. Fig. 3 presents results for proximity to the consensus mean ( X - X) for three laboratories. These examples demonstrate: accurate perform- ance, laboratory ( a ) , where all results are close to the mean and are within the targets; good performance but with an obvious negative bias, laboratory ( b ) ; poor accuracy where few results are within the target ranges, laboratory (c).Fig. 4 shows how a similar display can be used for the assessment of precision (Xl-X2). This example is also for Zn in serum. Whitehead12 recommends that EQA schemes should include a calculation of a performance score. We have used the scoring system elaborated by Yeortian for the Supra- regional Assay Service and adopted for the CEC Blood LeadJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 393 2.0 1.0 1.0 2.0 zone zone A I l l - (a) --. 0 c - 0 0 o o 'Mear 0 ' 0 B 0 0 - \ 0 . 0 - 10 20 Mean/pmol 1-l Fig. 2. Preparation of target zones for the assessment of perform- ance.The example here is for Zn in serum: 0, points used to delineate the outer zone and H, points used to delineate the inner zone 2.0 1.0 1.0 2.0 Table 3. Limits used for each matrix - analyte combination for the preparation of graphs to show accuracy and precision, e.g., Figs. 2-4 - (c) 0 ,p 0 . - - - 0 . Mean 0- - \ - 0 ' 0 0 - ==r Assay Serum AVprnol I - ' . . , . SerumAu/pmolI-l . . , . Serum Cu, Zn/pmolI-l . . Serum Se/pmolI-l . . . . Blood Cdhmol I-' . . . . Blood Pb/pmoll- . . . . UrineCdhmolI-' . . . . Urine Hghmol 1- . . . . Inner limits 0.2 at 1 .0 0.4 at 4.0 0.25 at 4.0 0.60 at 16.0 0.50 at 4.0 0.75 at 20.0 0.06 at 0.75 0.10 at 2.00 4 at 50 7.5 at 200 0.07 at 0.48 0.12 at 2.90 5 at 50 12 at 300 5 at 50 15 at 500 Outer limits 0.4 at 1 .0 0.8 at 4.0 0.5 at 4.0 1.2 at 16.0 1.0 at 4.0 1.5 at 20.0 0.12 at 0.75 0.20 at 2.00 8 at 50 15 at 200 0.14 at 0.48 0.24 at 2.90 10 at 50 24 at 300 10 at 50 30 at 500 I L (b) 2.0 10 20 30 Fig.3. Graphical display of targets used to assess proximity to consensus mean for zinc in serum. Laboratory (a) = satisfactory performance; laboratory ( b ) = good accuracy but a low bias; and laboratory ( c ) = poor accuracy 2.0 t 0 1 0 A i i 0 10 20 30 Mean/pmol I - ' Fig. 4. Graphical display of assessment of between-batch precision, samples analysed on two separate occasions. This example is for Zn in serum I 1 1 1 1 1 1 1 200 400 2.0 4.0 Pb added/pmol 1-1 Cd addedinmol 1-1 I 1 1 1 A1 added/pmol 1-1 2.0 4.0 4.0 8.0 Au addedipmol l - 1 10 20 30 Cu addedipmol 1-1 Cd in urine 100 200 Cd addedlnmol 1-1 10 20 30 Zn added/pmol 1-1 Hg in urine 1 1 200 400 Hg addedinmol 1-1 Fig.5. Comparisons between consensus mean values (endogenous levels) and s ike levels for Cd and Pb in blood, Al, Au, Cu and Zn in serum and Ed and Hg in urine. Data are from samples distributed between April and September 1985. Regression equations and correlation coefficients are as follows. Blood, Cd Y = -9.847 + 0.9941 X X , r = 0.9851; Pb Y = 0.037 + 0.9514 X X , r = 0.9974. Serum, Al Y = 0.049 + 1.0197 x X, r = 0.991 : Au Y = 0.576 + 1 .0054 X X , r = 0.9767; Cu Y = 0.461 + 0.9637 x X , r = 0.998; Zn Y = -1.149 + 1.0235 X X , r = 0.9996. Urine, Cd Y = 1.725 + 0.9158 X X , r = 0.9846; Hg Y = 5.487 + 0.9957 x X , r = 0.9897394 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL.1 Table 4. Results for Al, Cu and Zn obtained on two samples of serum: (a) within 1 month of preparation and (b) 6 months later after storage at -20 "C Aluminium Copper Zinc No. of Mean c v , No. of Mean c v , No. of Mean c v , results k SD/pmoll- YO results k SD/pmol I-' YO results k SD/pmolI-l YO (4 31 1.27* k 0.38 29.9 62 17.76 f 2.29 12.9 66 13.04 k 1.48 11.3 (b)30 1.18* k 0.20 16.9 53 17.81 +- 1.71 9.6 59 12.84 f 1.26 9.8 (a)30 3.46 kO.61 7.6 59 23.54 k 2.45 10.4 64 2.70 k 0.99 36.7 (b)29 3.44 k0.51 14.8 55 23.98 t- 2.63 11.0 58 2.87 f 1.16 40.4 * Significant difference p = c0.05. Table 5. Mean inter-laboratory coefficients of variation ("/o) at the concentrations shown, from 1975 to 1985 Serum Blood Urine Al Au c u Zn Cd Pb Cd Hg Date 2-3 pmol I - ' 5-10 pmoll-1 23-27 pmol I-' 14-17 pmol I-' 80-120 nmol I-' 2-3 pmol I-' 50-100 nmol I- * 250-300 nmoll- * 1975 - 1978 1981 41.6 14.0 9.5 12.5 19.8 7.4 31.9 27.8 1982 42.4 16.9 11.2 12.4 24.1 6.6 18.7 17.9 1983 32.0 11.1 8.3 14.1 22.2 6.6 24.8 22.0 1984 16.4 19.1 8.7 9.4 18.9 6.8 30.3 20.0 1985 16.1 23.0 9.0 10.6 14.9 5.5 17.5 23.0 - - - - - 19.2 8.9 - - - - - - 8.9 17.5 Survey.13 From the 18 results returned by a laboratory for, for example, Zn in serum, during a cycle, a performance score is calculated from the sum of: the percentage of results within the inner zone of the X - X graph, the percentage of results within the outer zone of the X - X graph, the percentage of results within the inner zone of the X1-X2 graph, the percentage of results within the outer zone of the Xl-X2 graph and the percentage of recoveries within the range 90-1 1O0/o.A perfect score would be 500, we suggest that good performance is indicated by a score of 360. As analytical techniques improve this target score can be increased or made more difficult to achieve by refinement of the zones shown in Fig. 2 . Results and Discussion An external quality assessment scheme will be of use only if certain fundamental conditions are fulfilled. These include (i) the distribution of specimens that are homogenous both internally and from tube to tube, (ii) samples that are stable throughout their anticipated period of use and (iii) the performance of valid computations with results. Careful preparation and preliminary analysis of a number of samples from every batch has ensured that the specimens distributed for the trace element EQA scheme are homogeneous. Specimens of sera from the same pools are analysed twice with intervals of up to six months between the two distributions.As shown in Table 4 there is no evidence of instability or deterioration of the samples and with the exception of one set for A1 (where the second set of data are actually superior), results for all elements show no significant changes over this length of time. Other specimens included in the scheme are not stored for long periods. Some aspects of assessment of performance used in this trace element EQA scheme are calculated by reference to the consensus mean and this value may not necessarily be a good representation of the true concentration, particularly if there are only a small number of results.Georges, l4 however, has demonstrated that the consensus mean is generally adequate even when there are few participants. Furthermore, good accuracy and the validity of the consensus mean will be demonstrated if this value and the spike level are identical. Fig. 5 shows data from the period April to September 1985 and indicates good agreement for almost all of the elements. The inference from these results is that the consensus mean is valid except for Au in serum. This last analysis is carried out by no more than five participants and results require very cautious interpretation. Mean values from results obtained using similar methods have not been routinely calculated but the code numbers of laboratories using particular methods are made available to participants so that they can determine group-method means for their own use.The International Federation of Clinical Chemistry (IFCC) expert panel on Nomenclature and Principles of Quality Control, described a series of objectives that external quality control procedures should seek to achieve.15 The work and results of the Guildford trace elements EQA scheme has been applied to all of these objectives. Investigation of Analytical Methods in Use The programmes for Cu and Zn in serum have sufficient numbers of participants for results to be used to prepare a comparison of methods used for sample preparation. It was shown that protein precipitation with trichloroacetic acid gave higher results (mean difference = +10.9%) than when sera were diluted with water or butanol and that the use of ETA-AAS produced low results for copper (mean difference = -3.9%).Assessment of New Methods and Verification of Experimental Results New methods for the measurement of trace elements are continually under development. To show that a novel pro- cedure gives satisfactory results, experiments to demonstrate acceptable accuracy and precision should be carried out. The analysis of EQA samples with results compared against the consensus or group-method means, provides an independentJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 395 check of the performance of the new procedure. Results from investigations that have involved the trace element EQA scheme have been presented in many publications, for example in references 17-19.Furthermore, the analytical aspects of clinical investigations have been demonstrated to be reliable by the acceptable performance of the laboratory in the EQA scheme.20 Presentation of “State of the Art” and Stimulation for Improved Performance The survey of performance prepared at six-month intervals identifies groups of laboratories whose scores are consistently high. It is evident that analytical performance can be maintained at standards appropriate or superior to those demanded for effective use of the results by clinicians and other workers. These standards of high performance are indicative of the best that can be achieved for the determi- nation of trace elements in biological samples with the type of equipment generally available to hospital and other labora- tories and represent the “state of the art” for such analytical procedures. Laboratories that fail to attain suitable (not necessarily the best) levels of performance should examine their procedures, methods, instrumentation etc.to determine which factors should be changed or improved in order that their analytical data match that which has been shown to be possible. A good example of poor performance revealed by the trace elements EQA scheme is afforded by the determination of A1 in serum. This measurement is important in the management of patients with chronic renal disease. These subjects are likely to absorb A1 from contaminated dialysis fluids or from orally administered hypophosphataemic agents, with consequent toxicity.When the unsatisfactory analytical situation was realised, considerable efforts to bring about an improvement were made. The six-month reports have shown that the number of laboratories that achieve the target score for the determination of A1 in serum has increased in the last three years and that the inter-laboratory variation between results has gradually declined.21 Many factors are involved in this trend but regular monitoring of performance with demon- stration of the variation between results from laboratories has been one such factor.19 Changes in inter-laboratory variation during 1975-1985, which reflect the improved performance for the measurement of trace elements in biological fluids, are given in Table 5.The results show that for certain elements there has been little change in the spread of results between laboratories but that for analyses that have received special attention for statutory or other purposes, such as Pb in blood and A1 in serum, the situation has gradually improved. Additional data for these latter determinations have been given elsewhere.21 Education Management of any EQA scheme provides opportunities for the collection of large amounts of information and data pertinent to the analyses included in the scheme. This can be made available to participants by direct communication, in special reports or publications or by the provision of scientific meetings. Those responsible for the organisation of the trace elements EQA scheme are frequently asked by participants for information and advice to help them to solve particular problems.Meetings have also been organised, including an international conference on the role of aluminium and other trace elements in renal disease. While these aspects of the EQA scheme are important in the work of laboratories concerned with measurements of trace elements the most important function is to supplement any internal quality control procedures that may be in operation. The principles and practices described here were elaborated for the assessment of analytical performance with respect to trace element determinations in biological samples. They are also relevant to different applications and other techniques, in addition to atomic absorption spectrometry.We are grateful to Mr. N. Smith, Occupational Medicine Laboratory, Health and Safety Executive, London, for his help with the lysis of blood cells by ultrasonication. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. References Delves, H. T., Clin. Endocrin. Metab., 1985, 14, 725. Versieck, J., and Cornelis, R., N . Engl. 1. Med., 1980, 302, 468. Versieck, J., and Cornelis, R., Anal. Chim. Acta, 1980, 116, 217. Versieck, J., Crit. Rev. Clin. Lab. Sci., 1985, 22, 97. Bierenbaum, M. L., Fleischman, A. I., Dunn, J., and Arnold, J., Lancet, 1975, i, 1008. Taylor, A., Clin. Endocrin. Metab., 1985, 14, 703. Platin, L. O., and Meurling, S . , “Trace Element Analytical Chemistry in Medicine and Biology,” Walter de Gruyter, Berlin, 1980, pp. 243-254. Cohen, D., Orzel, J., and Taylor, A,, Arthritis Rheum., 1981, 24, 104. Pleban, P. A., Numerof, B. S., and Wirth, F. H., Clin. Endocrin. Metab., 1985, 14, 545. Vahter, M., “Assessment of Human Exposure to Lead and Cadmium through Biological Monitoring,” National Swedish Institute for Environmental Medicine and Department of Hygiene, Karolinska Institute, Stockholm, Sweden, 1982. Friberg, L., and Vahter, M., Environ. Res., 1983, 30, 95. Whitehead, T., “Quality Control in Clinical Chemistry,” John Wiley, Chichester, 1976. Department of The Environment, “The UK Steering Group EC Blood Lead Survey. European Community Screening Programme for Lead, United Kingdom Results for 1981,” Department of The Environment Pollution Report No. 18, HMSO, London, 1983. Georges, R. J., Ann. Clin. Biochem., 1985, 22, 283. Buttner, J., Borth, J., Boutwell, J. H., Broughton, P. M. G., and Bowyer, R. C., Clin. Chim. Acta, 1978, 83, 189F. Taylor, A., and Bryant, T. N., Clin. Chim. Acta, 1981,110,83. Shaw, J . C. L., Bury, A. J., Barber, A., Mann, L . , andTaylor, A., Clin. Chim. Acta, 1982, 118, 229. Sthapit, P. R., Ottaway, J. M., and Fell, G. S . , Analyst, 1983, 108, 235. D’Haese, P. C., Van de Vyver, F. L., de Wolff, F. A., De Broe, M. E., Clin. Chem., 1985, 31, 24. Sutton, A. M., Harvie, A., Cockburn, F., Farquarson, J., and Logan, R. W., Arch. Dis. Child., 1985, 60, 644. Taylor, A., Starkey, B. J., and Walker, A. W., in Taylor, A., Editor, “Aluminium and Other Trace Elements in Renal Disease,” Balliere Tindall, Eastbourne, 1986, pp. 26G273. Paper J6l6 Received February 13th1 1986 Accepted June 2nd1 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100391

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 397 Blood Cadmium Determination-Results of an External Quality Assessment Scheme Bryan J. Starkey, Andrew Taylor and Arthur W. Walker Supra-regional Assay Service, Trace Elements Laboratory, Department of Clinical Biochemistry, St. Luke's Hospital, Guildford, Surrey GU 1 3NT and Robens institute, University of Surrey, Guildford, Surrey GU25Xff, UK Laboratory performance with respect to blood cadmium determination in the Guildford Trace Elements External Quality Assessment Scheme has been examined. Between March 1982 and September 1985 over 20 laboratories participated in the scheme and more than half submitted results regularly. A methodology survey showed that, whilst more laboratories used electrothermal atomic absorption spectrometry (ETA-AAS), sample pre-treatment, standardisation procedures and furnace temperature programmes varied enormously.Mean values for recovery of cadmium added to blood during each 6-monthly cycle were between 91 and 102% but varied more widely for individual samples (81-1 17%). Inter-laboratory coefficients of variation usually exceeded 20% at cadmium concentrations below 100 nmol I-' ( 1 1.2 ug 1-1) but were lower (12-20%) at higher concentrations. No demonstrable changes in these parameters during the period of study were evident and it is concluded that laboratory performance with respect to cadmium analysis requires considerable improvement. Keywords: Blood cadmium determination; quality assessment scheme The health hazards of increased cadmium absorption from industrial processes and environmental pollution are well recognised.1.2 Absorbed cadmium is excreted very slowly3 and continued exposure is associated with pulmonary, renal and hepatic damage. Measurement of Cd in blood has been shown to be of value in the assessment of occupational cadmium exposure (e.g., battery manufacture, welding and metal refining)4.' and environmental cadmium exposure (c.g., tobacco smoke, vegetables from cadmium-rich soils and shellfish). 6.7 Many methods for blood cadmium determination have been described. They include those dependent upon chelation of cadmium by pyrrolidiness with subsequent organic extraction and measurement by either flame atomic absorption spec- trometry (FAAS)9 or ETA-AAS.8 Others9 have used the microsampling technique of Delves10 to avoid laborious sample preparation.Alternative techniques, including flu- orescencell and anodic-stripping voltammetry,12 have been used, but most of the methods reported more recently have been based on ETA-AAS.1.3-'7 The multiplicity of methods in use for blood cadmium determination has led to a great diversity of results, as evidenced by the wide range of reference intervals quoted.2 The lack of suitable reference material with assigned values for blood cadmium concentration renders the assessment of analytical performance difficult. This led to inclusion of blood cadmium in the Guildford Trace Elements External Quality Assessment Scheme. The results from this scheme are presented here, and the performance of laboratories is discussed. Experimental Apparatus All glass- and plastic-ware used for preparation of control material was immersed in hydrochloric acid (10% VlV) for 24 h, rinsed in de-ionised, reverse osmosis treated water and dried in a dust-free cabinet prior to use.Trace metal free polycarbonate tubes for the distribution of samples were obtained from Teklab, Durham and each batch was checked for Cd contamination prior to use. Cadmium measurements were performed using a PU9000 furnace atomic absorption spectrometer with autosampler (Pye Unicam, Cambridge). Reagents Nitric acid (Aristar), saponin and cadmium nitrate (1 mg ml-1) standard solutions were purchased from BDH Chemicals, Poole, Dorset. Blood was drawn from healthy volunteers and EDTA (dipotassium salt) was used as an anticoagulant (final concentration 2 mg ml-1).Water was purified before use by de-ionisation and reverse osmosis (Elgastat Spectrum System, Elga, High Wycombe, Bucks). Operation of the Quality Assessment Scheme The preparation and distribution of samples, the statistical treatment of the results and the assessment of laboratory performance have been described in detail elsewhere. 18 The general outline of operation is summarised in Fig. 1. Methodology Survey Laboratories were asked, by means of a questionnaire, for details of their method for blood cadmium determination. Information was requested concerning sample pre-treatment, assay standardisation, instrumentation and, where applicable, furnace conditions in use. 0 t her Investigations To determine the contribution of inter-sample variability to the range of results returned by scheme participants, 15 samples from each of three distributed batches of material were assayed in the authors' laboratory.Results and Discussion Data on the mean recoveries of added cadmium (Table 1) demonstrated that, for each 6-monthly period, the mean recovery of cadmium from blood approximated to 100% (91.3-101.8%), implying that the additions of cadmium to blood were accurate. Mean recoveries found for individual samples distributed during each 6-monthly period were, however, much more variable (81-117%). This was particu- larly evident where the additions of cadmium were small and398 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 probably reflects the greater difficulty experienced by labora- tories in measuring accurately low [less than 50 nmol 1-1 (5.6 pg I-l)] cadmium concentrations.The inaccuracies of the laboratory analysis of low cadmium concentrations are further exemplified in Fig. 2, which shows the inter-laboratory variation of results for cadmium determi- nation in different concentration ranges during consecutive time periods. The inter-laboratory coefficients of variation at Cd concentrations of less than 50 nmol l-1 (5.6 pg 1-1) varied considerably (1746%) but usually approximated to 40%. At higher cadmium concentrations (100-300 nmol 1-1) these coefficients of variation were in the range 12-20%. The contributions of sample-to-sample variation, sample quality and sample stability to the range of results found by participants were investigated by assaying samples from three distributions covering a wide concentration range (Table 2).To simulate distribution within the scheme, these samples were stored initially at -20 "C, and then for 15 d at 4 "C prior to analysis. Samples were analysed within the same analytical batch. The coefficients of variation found were similar to those for within-batch precision data generated by repeated analysis of a single sample, implying that samples can be considered as identical. Inter-laboratory coefficients of variation were higher by a factor of 2.6-7.7. Whilst it could be argued that samples assayed in this manner would be expected to give much better precision than that found in an inter-laboratory comparison, the discrepancy found is large.Between-batch precision data from the authors' laboratory demonstrate that the between-batch coefficient of variation is normally less than twice that found within-batch at the same concentration. The results from the methodology survey are summarised in Tables 3 and 4. To maintain anonymity the laboratory codes shown are not those used in the Quality Assessment Scheme. Pre-treatment of samples prior to analysis (Table 3) varied enormously; from no treatment as in the Delves cup pro- cedure, to protein precipitation or dilution in complicated solutions. Despite the use of matrix modifiers (e.g., Triton), ashing agents (e.g., ammonium nitrate) or matrix removal (e.g., by precipitation with nitric acid), no laboratory used aqueous standards.All laboratories used either matrix- matched standards or standard additions techniques, presu- mably because the effects of the blood matrix on assay procedures could not be completely eliminated. All but one laboratory used ETA-AAS for the analysis, reflecting the greater sensitivity achievable by this technique compared with flame AAS. No laboratories in the group used the older chelation-based methodologies. Furnace instruments from five different manufacturers were in use and the wide diversity of furnace cuvettes, temperature programmes and modes of background correction used are summarised in Table 4. Six laboratories used furnace cuvettes of pyrolytically coated graphite rather than uncoated tubes, presumably to confer lower detection limits to their assays.No laboratory, however, Healthy volunteer 350 ml blood + 1 g EDTA Lyse (Saponin, 0.4 g) 100 ml 100 ml +[Cdl, + [Cdl, + 100 ml + IcDli i 4 2-ml aliquots y-Irradiation Laboratories I i I Results G ui ldford Mean, SD Discard (mean k 3 SD) outliers IJ Monthly report to laboratories (mean, SD, CV) Collate 1 6-Monthly report w.r.t. 9 sample pairs ( a ) Accuracy w.r.t. consensus means (6) Precision w.r.t. duplicate samples (c) Recovery (4 Laboratory ranking spiked value - endogenous value ( Cd addition Fig. 1. Operation of the scheme Table 1. Recovery of added cadmium from blood Mean No. of recovery, Range,t April '82-Sept. '82 . . 21 24 100.6 88-115 April '83-Sept. '83 . . 17 20 101.8 94-117 April '84-Sept. '84 . . 22 21 97.3 86-107 Period participants N* YO YO Oct.'82-March '83 . . 20 18 98.2 85-115 Oct. '83-March '84 . . 21 20 93.4 82-108 Oct. '84-March '85 . . 18 18 91.3 82-100 April '85-Sept. '85 . . 17 19 94.0 81-112 * Nis the mean number of results per sample pair. Cadmium added 1- Range of mean cadmium recovery from each sample distributed. to samples varied from zero to 300 nmol 1-1. 50 40 30 8 >- u 20 10 I I I I I I April Sept. April Sept. April Sept. April Sept. 'a2 '82 'a3 '83 184 'a4 '85 '85 Fig. 2. Inter-laboratory coefficients of variation at different cadmium concentrations. The results represent the mean inter-laboratory coefficients of variation (Yo) during consecutive 6-monthly periods for different ranges of blood cadmium concentration: A, 0-50; B, 50-100; C, 100-150; D, 150-200; and E, 200-300 nmol l-lJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 399 Table 2.Comparison of intra- and inter-laboratory precision Intra-laboratory * Inter -laboratory Mean1 CV, Mean/ CV, N nmoll-1 % Ni- nmoll-I Yo 15 8.2 10.5 6 6.6 17.5 15 135.1 2.2 9 134.4 17.1 15 302.0 2.2 9 299.8 17.0 * Analyses were performed within the same analytical batch in the t In the inter-laboratory study, N is the number of laboratories each intra-laboratory study. returning a result. used a stabilised temperature platform. The use of such a technique19 may help circumvent some of the matrix interfer- ences incumbent in blood cadmium measurement. The great diversity of temperature programmes reflect, in part, the different instruments in use. In blood cadmium determi- nation the ashing phase is critical.Cadmium may be lost at temperatures as low as 400"C, which is too low to ensure complete removal of organic matter. Several laboratories in the group used addition of phosphate salts to stabilise cadmium and prevent its co-volatilisation with organic material, enabling ashing temperatures of up 650°C to be used. Great care is essential, however, to ensure that minor variations in the ashing temperatures chosen do not lead to Table 3. Methodology survey-sample pre-treatment Laboratory code A B C D E F G H I J K L M N 0 P Q Sample diluent 30% 2,2' diaminodiethylamine - 30% HN03 - 30% water Triton X-100 - (NH4)3P04 5% HN03 (pptn.) Triton X-100 - (NH4)2HP04 Triton X-100 - NH4N03 - EDTA 0.02% Triton X-100 - 3% HN03 Triton X-100 - (NH4),S04 - NH4N03 5% HN03 (pptn.) Conc.HN03 digestion - (NHJ3P04 dilution Triton X-100 0.5% Triton X-100 - 5% HN03 0.2% Triton X-100 - 0.5% (NH4)3P04 None (Delves cup) 20% HN03 (pptn.) 1% NH4HzP04 NH4H2PO4 - HN03 10Yo (NH4)2HP04 Pre-dilution (diluent : sample) 4: 1 2 : 1 6 : 1 4 : 1 11 : 1 10: 1 24 : 1 5 : 1 5 : l NSt 40 : 1 5 : 1 NS 12: 1 1 : l 5 : 1 - Source of standard* Merck BDH BDH BDH BDH BDH BDH BDH BDH BDH BDH BDH BDH Cd metal BDH Merck Merck * All laboratories use standards in a blood matrix. t NS = Not specified. Table 4. Methodology survey-furnace conditions Dry Ash Atomise Laboratory Code Instrument A Hitachi B Perkin- Elmer C Perkin- D Perkin- E Perkin- F Varian G IL Elmer Elmer Elmer I PYe Unicam K IL L IL N Perkin- Q IL Elmer Tube t Y Pe TaO, coated Pyro- coated Uncoated Uncoated Pryo- coated Uncoated Uncoated Uncoated Pyro- Pyro- Pyro- Pyro- coated coated coated coated Temperature/ "C 120 150 110 85 100 130 110 150 100 160 80 120 150 (FASTAC) Manual 80 110 70 Time/s 60 20 20 10 30 5 5 15 20 5 2 10 - Drying 5 30 5 Temperature1 "C 300 600 400 500 350 600 ( 0 2 ) $ 400 (02) 500 550 280 400 300 450 650 650 225 275 287 Time/s 20 30 30 10 30 10 15 10 15 10 30 (cool) - 20 5 15 10 30 30 Temperature/ Background "C Time/s correction* 1300 7 (GS)? Z 2 100 3 Z 2200 3 D2 2000 1 D2 1850 5 S/H 2100 4 D2 D2 1500 - 1750 1 D2 1400 5 D2 * Z = Zeeman; D2 = deuterium; S/H = Smith - Hieftje.t GS = Gas stop. $; O2 = Oxygen ashing.400 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, OCTOBER 1986, VOL. 1 cadmium loss. Deuterium background correction was used by most of the group.Zeeman correction, however, used by only two laboratories, has been advocated as a more accurate means of background correction for urinary cadmium deter- mination20 and may likewise improve the accuracy of the determination of cadmium in blood. Examination of the methods used in the light of perfor- mance within the scheme allows no firm recommendations to be made with regard to the best analytical procedure for cadmium determination. Although each 6-monthly report ranks laboratories for analytical performance with respect to accuracy relative to concensus mean values, precision with respect to duplicate samples and recovery of added cadmium, no laboratory performed consistently well in all these respects, and no single laboratory performed better than the other laboratories in every 6-monthly cycle.The great diversity of sample pre-treatment procedures, instrumentation and temperature programmes in use probably contributes to the wide variation in results which, from studies outlined here, is unlikely to be due to sample-to-sample variability, poor sample quality or sample instability. The poor laboratory performance may be attributed to contami- nation of samples within the laboratories, to procedural losses during sample pre-treatment or the use of inappropriate temperature programmes that may lead to cadmium loss during ashing or to inadequate background correction. There is a real need for reliable quality assurance material with agreed, assigned values for cadmium concentration.The use by laboratories of the same internal quality control material with well defined acceptance limits would do much to improve laboratory performance and reduce inter-laboratory varia- tion. In conclusion, current laboratory performance with respect to cadmium determination is poor and requires considerable improvement. This paper highlights the need for laboratories to pay strict attention to the quality of the results they produce. The authors thank Ms. Heather Scott for her excellent secretarial assistance. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. References Perry, H. M., Jr., and Erlanger, M. W., J. Lab. Clin. Med., 1974, 83, 541. Friberg, L., Piscator, M., Nordberg, G. F., and Kjelstrom, T., “Cadmium in the Environment,” CRC Press, Cleveland, OH, 1974. Lauwerys, R., “Criteria (DoselEffect Relationships) for Cad- mium,” Pergamon Press, Oxford, for Commission of the European Communities, Luxembourg, 1978. Scott, R., Mills, E.A., Fell, G. S . , Husain, F. E. R., Yates, A. J., Paterson, P. J., McKirdy, A . , Ottaway, J. M., Fitzgerald-Finch, 0. P., Lamont, A., and Roxburgh, S . , Lancet, 1976, ii, 396. Webb, M. A. H., Chettle, D. R., Al-Haddad, I. K., Downey, S. P. M. J., and Harvey, T. C.,Ann. Occup. Hyg., 1982,25,33. Carruthers, M., and Smith, B., Lancet, 1979, i , 845. Kjellstrom, T., Environ. Health Perspect., 1979, 28, 169. Beevers, D. G., Campbell, B. C., Goldberg, A . , Moore, M. R., and Hawthorne, V. M . , Lancet, 1976, ii, 1222. Ulander, A., and Axelson, O., Lancet, 1974, i, 682. Delves, H. T., Analyst, 1970, 95, 431. Michel, R. G . , Hall, M. L., Ottaway, .I. M., and Fell, G. S., Analyst, 1979, 104, 491. Matson, W. R., Roe, D. K . , and Garritt, D. E., Anal. Chem. , 1965, 37, 1598. Perry, E. F., Koirtyohann, S. R., and Perry, H. M., Jr., Clin. Chem., 1975,21, 626. Lagesson, V., and Andrasko, L., Clin. Chem., 1979,25, 1948. Pleban, P., and Pearson, K. H., Clin. Chim. Acta, 1979, 99, 267. Stoeppler, M., and Brandt, K., Fresenius 2. Anal. Chem., 1980,300,372. Subramanian, K. S . , and Meranger, J . C., Clin. Chem., 1981, 27, 1866. Taylor, A., and Briggs, R. J., J. Anal. At. Spectrom., 1986, 1, 391. Slavin, W., Manning, D. C., and Carnrick, G. R., At. Spectrosc., 1981, 2, 137. Subramanian, K. S . , Meranger, J. C., and MacKeen, J. E., Anal. Chem., 1983, 55, 1064. Paper J6l16 Received March 13th, 1986 Accepted April 28th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9860100397

出版商:RSC    

年代:1986

数据来源: RSC