摘要:

Use of the AT,’ Signal as a Diagnostic Tool in Solid Sampling Electrothermal Vaporization Inductively Coupled Plasma Mass Spectrometry Journal of Analytical Atomic Spectrometry FRANK VANHAECKE,* GABOR GALBACS,~ SYLVIE BOONEN LUC MOENS AND RICHARD DAMS Laboratory of Analytical Chemistry Ghent University Institute for Nuclear Sciences Proeftuinstraat 86 B-9000 Ghent Belgium The utility of the Ar2+ signal (at mass-to-charge ratio rn/z=80) as a diagnostic tool in solid sampling electrothermal vaporization inductively coupled plasma mass spectrometry (ETV-ICP-MS) is reported. Simultaneous monitoring of the argon dimer signal with the signal(s) of the analyte element(s) indicated that non-spectral interferences caused by matrix components co-volatilizing with the analyte element(s) can strongly affect the analyte signal profiles in solid sampling ETV-ICP-MS of samples of biological or environmental origin.This observation led to a more profound understanding of why for a given matrix the signal profiles strongly differ from one element to another and why for a given element the signal profile is seen to be strongly dependent on the matrix. These matrix effects were also observed to cause a curvature in the sample mass response curves (analyte signal intensity as a function of sample mass). It is shown that at least in some instances the use of the Ar2+ signal as an internal standard allows (i) this non-linearity to be corrected for and (ii) accurate analysis results to be obtained. Finally it is demonstrated that simultaneous registration of the argon dimer and the analyte signal(s) is useful during optimization of ashing and vaporization temperatures.Keywords Argon dimer (Ar2+ ); solid sampling; electrothermal vaporization; inductively coupled plasma mass spectrometry; matrix efects Since inductively coupled plasma mass spectrometry (ICP-MS) combines low detection limits with multi-element capabilities and a high sample throughput it is a highly desirable technique for trace element determination in a variety of matrices. Originally ICP-MS was mainly intended for the analysis of aqueous samples but at present there is an increasing interest in the direct analysis of solid samples (solid sampling). Direct analysis of solid materials is of course important for materials that cannot or only with great difficulty be brought into solution.Moreover in general solid sampling limits the neces- sary amount of often laborious and time-consuming sample pre-treatment leading to a reduced risk of contamination or analyte losses and finally as samples are analysed without dilution lower limits of detection are also to be expected. Next to laser ablation ICP-MS which has evolved into a well established solid sampling technique over the past 10 electrothermal vaporization (ETV) ICP-MS also offers pos- sibilities for the direct determination of trace and ultratrace elements in solid samples. * Senior research assistant of the Belgian National Fund for Scientific Research. t Present address Attila J6zsef University Department of Inorganic and Analytical Chemistry D6m Ter 7-P.O.Box 440 H-6720 Szeged Hungary. Both Voellkopf et al.’ and GrCgoire et aL6 reported on the analysis of solid samples using ETV-ICP-MS but both research groups preferred to analyse slurries rather than dry solid samples. Wang et aL7 explored the feasibility of ‘real’ solid sampling ETV-ICP-MS and highlighted some difficulties hampering routine use of this technique at present. They concluded that in order to obtain optimum accuracy external calibration using a solid certified reference material (CRM) with a composition as similar as possible to that of the sample should be used. Argentine and Barnes8 used ETV-ICP-MS for the direct determination of volatile and non-volatile impurities in semiconductor-grade organometallic materials and process chemicals.The technique showed great potential particularly for the non-volatile impurities since complete separation of the (volatile) matrix from the analytes could be obtained by application of a suitable temperature programme. Quantification of the latter impurities was performed using aqueous standards and the results obtained compared within a factor of two with the results obtained using electrothermal vaporization inductively coupled plasma atomic emission spec- trometry (ETV-ICP-AES) and ICP-AES after decomposition. Since for the volatile impurities the ETV-ICP-MS results were degraded owing to the occurrence of severe non-spectral interferences caused by co-volatilization of matrix and analyte ETV-ICP-MS was assessed to be inferior to ETV-ICP-AES for the determination of volatile components. In our laboratory however trace amounts of As and Se (both giving rise to volatile components evaporating from the sample at tempera- tures below SOOOC) were successfully determined in CRMs of plant and environmental Although also for these matrices important non-spectral interferences could be estab- lished several calibration methods were demonstrated to offer possibilities for accurate quantifi~ation.~ The use of an internal standard was observed to be advantageous (standard additions methods) to imperative (external calibration methods) depending on the calibration method used.The present paper reports on the results of a more funda- mental study in which the Ar,’ signal was systematically monitored and used as a diagnostic tool in solid sampling ETV-ICP-MS.Application of the AT,+ signal intensity as an indication of plasma loading effects and non-spectral inter- ferences was introduced for pneumatic nebulization ICP-MS by Beauchemin et a1.12 Recently the potential of this approach was confirmed by Chen and Houk,” who extended its appli- cation range by also using a variety of other polyatomic ions. GrCgoire et aL6 used the argon dimer signal to monitor matrix effects in slurry sampling ETV-ICP-MS allowing them to select an appropriate calibration strategy for the determination of trace metals in samples containing substantial matrix components. In the present study it is shown that systematic monitoring Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1047of the Ar,+ signal in solid sampling ETV-ICP-MS allows a more profound understanding of the signal profiles (signal intensity as a function of time) observed.Registration of the Ar2+ signal was also shown to be useful during optimization of ashing and vaporization temperatures. Finally it is demon- strated that at least in some instances the Ar2+ signal can be used as an internal standard correcting for matrix-induced signal suppression effects such that the necessity of adding an appropriate elemental standardg can be avoided. EXPERIMENTAL Instrumentation The ETV system used is a commercially available graphite furnace of the boat-in-tube type (SM-30 Grun Analytische Mess-Systeme Ehringhausen Germany). Although this device was originally designed for solid sampling Zeeman atomic absorption spectrometry (AAS) some simple modifications (described el~ewhere’~) sufficed to make it compatible for use with both ICP-AES and ICP-MS.Sample holders (‘boats’) can be easily and reproducibly loaded into the cylindrical furnace with the aid of a pair of tweezers sliding on a rail which is rigidly mounted in front of the furnace. After loading the sample one end of the furnace is closed using a shutter kept in position by a catchspring. During operation an Ar flow the flow rate of which is controlled by a mass flow con- troller (Model 5876 Brooks Instruments Veenendaal The Netherlands) is swept through the furnace transporting vapor- ized ‘sample particles’ into the central channel of the ICP. The multi-step temperature programme of the furnace (Table 1) is controlled by a computer program developed in-house while Table 1 spectrometer Operating conditions for the ETV system and the ICP mass (a) ETVsystem- Type SM-30 Grun Analytische Mess- Temperature programme the multi-step temperature Systeme programme consisted of (1) (2) boosting step (1 s) ‘drying’ step (30 s at about 120 “C) during which the power applied is twice as high as for the ashing step allowing the heating rate of the furnace to be increased at moderate temperatures and hence i i stable ashing temperature to be obtained rapidly intermediate step (11 s at the ashing temperature) to switch the valve to the ‘measuring’ position and allow the plasma to stabilize (3) ashing step (30 s) (4) (5) vaporization step (15 s) ( 6 ) intermediate step (7) to switch the valve to the ‘venting’ position cleaning step (2 times 3 s at about 2700 “C) (b) ICP mass spectrometer- Type Rf power Sampling depth Carrier gas flow rate Auxiliary gas flow rate Plasma gas flow rate Lens voltages Sampling cone Skimmer cone Perkin-Elmer SCIEX-Elan 5000 10 mm 0.3-1.2 1 min-’ 1.0 1 min-’ 12 I min-’ Tuned using pneumatic nebulization-no further tuning required when switching to solid sampling ETV-ICP-MS 1000-1300 W P -58.1 V B +9.7 V s2 -9.0 v El +4.0 V Nickel 1.0 mm orifice diameter Nickel 0.75 mm orifice diameter the temperature can be monitored using an optical pyrometer especially designed for use with this type of ETV system (PY20 Grun Optik Ehringhausen Germany).The ETV system was coupled to a Perkin-Elmer SCIEX-Elan 5000 ICP mass spec- trometer via 10mm id silicone rubber tubing.In order to (i) reduce the amount of deposition of vaporized sample material on the torch the interface and the lens stack and (ii) avoid degradation of the interface pump oil a three-way valve was used to vent vapours generated during drying ashing and cleaning steps. Operating conditions are summarized in Table 1. Measurements Measurement parameters are summarized in Table 2. In order to obtain a compromise between fast hopping between the nuclides monitored (to obtain a representative image of the corresponding signal profiles) and an efficient use of the total measuring time (high ratio of actual measuring time to mass spectrometer settling time) 30ms was used as the dwell time per measuring point.” Although a quadrupole filter can only transmit ions of one mass-to-charge ratio at each moment during the acquisition and the voltages applied to the quadru- pole rods have to be changed in order to hop between the nuclides of interest during one measurement the term ‘simul- taneous monitoring’ is used throughout the text in order to indicate that both the analyte signal(s) and the Ar,+ signal were recorded during the same firing of the furnace.In order to protect the electron multiplier against excessively high count rates the OmniRange device was used for the monitoring of both the Ar2+ signal intensity and analyte signal intensities giving rise to excessively high count rates. The OmniRange device can selectively and reproducibly reduce the sensitivity of the mass spectrometer by varying the ion transmission efficiency at the exact time a given mass-to-charge ratio is being measured.The OmniRange setting for the argon dimer [and if necessary for the analyte(s)] was always selected such that the Ar2-‘ signal intensity was more or less comparable to the intensity of the analyte signal in the matrix under consider- ation. The three-way valve was switched manually to the ‘measuring position’ 11 s before the start of the vaporization stage while the measurement itself was started 1 s before the beginning of the vaporization stage (Table 1). The total measurement time exceeded the duration of the transient signals monitored. Integration over a more limited time interval could be carried out using Microsoft Excel 5.0.Samples All the solid materials investigated during this study are CRMs of biological or environmental origin Mussel Tissue (BCR CRM 278) Olive Leaves (BCR CRM 062) Rye Grass (BCR CRM 281) Sea Lettuce (BCR CRM 279) and Tomato Leaves (NIST SRM 1573) of biological origin and Light Sandy Soil (BCR CRM 142) and River Sediment (BCR CRM 320) of environmental origin. RESULTS AND DISCUSSION Fig. 1 shows the AT,’ signal profiles during the vaporization step for (a) an empty sample holder for (b) a sample holder Table 2 Measurement parameters Dwell time 30 ms Scanning mode Peak hop transient Readings per replicate Points per spectral peak 1 Total measurement time ~ 3 0 s Sweeps per reading 1 800/number of nuclides monitored 1048 Journal of Analytical Atomic Spectrometry December 1995 Vol.10Fig. 1 8oAr2+ signal intensity as a function of time during the vaporization stage of the multi-step temperature programme for 0 (empty sample holder) 0.527 and 2.030mg of solid sample of Sea Lettuce (BCR CRM 279) containing about 0.5mg of Sea Lettuce and (c) containing about 2 mg of Sea Lettuce. The marked difference between the profiles under consideration indicates that also in solid sam- pling ETV-ICP-MS the argon dimer can be used as an indication of matrix effects. Both for the empty sample holder and for the solid samples the Ar2+ signal profile shows a strong suppression at the beginning of the vaporization. The magnitude of this suppression was seen (i) not to be affected by the sample mass (Fig. 1) and (ii) to increase with an increasing difference between the ashing and the vaporization temperature.Hence it is believed that since the Ar2+ signal intensity strongly depends on the carrier gas flow rate (Fig. 2) this suppression is to be ascribed to sudden changes in the carrier gas flow rate caused by rapid heating of the furnace. Fig. 1 shows that for an empty sample holder the Ar2+ signal intensity continuously increases and finally reaches a more or less stable value. When analysing a liquid standard solution e.g. 5-30 pl of a 200 pg I-' Cd standard solution the Ar2+ profile was observed to be identical with that of an empty sample holder. For a solid sample however the Ar2+ signal profile shows important alterations. Since these alterations become more important with increasing sample mass and since the over-all shape of the Ar2+ signal profile during vaporization strongly depends on the matrix under consideration there is little doubt that also in solid sampling ETV-ICP-MS the Ar2+ signal can be used as an indication of non-spectral interferences caused by volatilizing matrix components and hence as an indication of the actual sensitivity of the instrument at each moment during vaporization.This suppression of the * 0.25 0.5 0.75 1 1.25 u - Carrier gas flow ratd min-' Fig. 2 'OArz+ signal intensity monitored as a function of the carrier gas flow rate at three rf powers during the vaporization stage of the multi-step temperature programme for samples of about 0.7 mg of River Sediment (BCR CRM 320) (vaporization temperatures 1400 "C) Ar2+ signal intensity could for example be caused by a cooling effect on the plasma caused by the introduction of (possibly macromolecular) matrix components leading to a decreased probability of Ar2+ formation and/or ionization.Evaporation of gaseous products from the matrix during vaporization could possibly also give rise to Ar2+ signal suppression owing to (i) a dilution of the carrier gas (ii) a change in the position of the Ar,+ maximum density zone in the plasma leading to an alteration of the extraction efficiency and/or (iii) a widening of the central channel of the ICP. Simultaneous monitoring of the argon dimer and the signal from analyte elements allowed a better understanding of the signal profiles observed to be obtained. During previous appli- c a t i o n ~ ~ .~ ~ it was observed that for a given material the signal profiles observed strongly depend on the analyte element and that for a given analyte element the signal profile strongly depends on the matrix. The results of simultaneous monitoring of the argon dimer with As Cd and Co respectively are shown for Olive Leaves (BCR CRM 062) in Fig. 3(a)-(c). A closer look at the analyte profiles concerned indicates that in each instance the actual shape of the profile is determined by both the analyte element (appearance time and duration of signal) and the matrix (suppression of the analyte signal by co-volatilizing matrix components). For As for example the first dip in the Ar2+ profile coincides with a dip in the 75As+ profile and it is clear that also the remainder of the 75As+ profile is strongly affected by the variation in the Ar2+ intensity (or more correctly the actual sensitivity of the mass spec- trometer).It is important to stress that the 75As+ signal profile is not perturbated by the presence of 40Ar35C1+ since no significant alteration in the shape of the signal profile is observed on addition of about 100 pg of C1 per mg of sample (about a 140-fold excess in comparison with the original C1 0 2.5 5 7.5 10 12.5 15 Timds 5 Fig. 3 "Ar,+ and M+ signal intensity as a function of time during the vaporization stage of the multi-step temperature programme for solid samples of Olive Leaves (BCR CRM 062) where M+ is (a) 75As+ (b) l1'Cd+ and (c) 59C0+ Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1049content 0.7 mg g- ').Similar signal profiles were also observed for e.g. Hg Sb and Se. For Cd on the other hand [Fig. 3(b)] condition that sample masses were chosen as close as possible to one an~ther.'?'~ - - the appearance time and the signal duration are such that the ll1Cd+ profile exactly fits under a peak in the AT2+ profile and hence apparently does not suffer from signal suppression. Co finally is less volatile than As and Cd and hence its signal is only detected later after the start of the vaporization than for more volatile elements. Since during the presence of the Co+ signal the Arz+ signal only shows a continuous increase in the signal intensity also for Co no dips in the profile are observed. It is also interesting that after about 5 s the sensitivity of the instrument gradually increases (as indicated by the continu- ously increasing argon dimer intensity) such that the tailing of the As and Co peaks (apparently) becomes more prominent.A more realistic representation of the signal profile can however be obtained by plotting the ratio of the signal intensity of the analyte element to the signal intensity of the argon dimer as is illustrated in Fig. 4 for Co. Fig. 4 clearly indicates that about 13 s after the start of the vaporization the signal ratio (59C0+ 80Ar2+) has already decreased to about 5% of its maximum value such that in this instance the acquisition duration and integration range can be limited to for example 15 s. Since it was demonstrated in Fig. 1 that the matrix effects encountered become more important with increasing sample mass it is clear that these matrix effects also cause curvature of the sample mass response curves (analyte signal intensity as a function of sample mass).It was demonstrated in a previous paper" that for As and Se this curvature could be corrected for by using Sb as an internal standard on condition that an aliquot of (liquid) Sb standard was brought into the sample holder and dried under an infrared lamp before solid sample loading. The use of Sb as an internal standard therefore corrected for matrix effects and allowed As and Se to be determined accurately in CRMs of both biological and environ- mental origin either by external calibration methods using liquid as well as solid standards (CRMs with similar matrix composition and analyte content) or by standard additions methods.If no internal standard was used accurate quantifi- cation was only possible using standard additions and on ' 140 120 0.8 + 2 2 0.4 a?. 0 In 0.2 n h c .- c 100 2 E 80 - c p f Q) 60 .E 20 0 - 0 2.5 5 7.5 10 12.5 15' Tirnds Fig. 4 Signal ratio (59C0+ "Ar,+) and 59C0+ signal intensity as a function of time during the vaporization stage of the multi-step temperature programme for about 0.7mg of solid sample of Olive Leaves (BCR CRM 062). In order to facilitate the evaluation of the importance of the memory effect the horizontal line indicates 5% of the maximum value for the (59C0+ *'Ar,+) signal ratio In general selection of a suitable internal standard is not self-evident. Most importantly an appropriate internal stan- dard should show an analogous (furnace) chemistry and should of course only be present at negligible levels in the samples under consideration.Although only of secondary importance with solid sampling ETV-ICP-MS in general internal stan- dardization for correction for matrix-induced signal suppres- sion or enhancement and for improving the precision has been observed to be most efficient if the mass number of the internal standard was chosen to be relatively close to that of the analyte element(^).'^-'' Since a suitable internal standard should fulfil all three conditions simultaneously it is clear that selection of such an internal standard sometimes poses an insurmountable problem. It was found however that also Ar2+ shows potential as an internal standard as is illustrated in Fig.5 in which the 7 5 A ~ + signal intensity and the ratio (75A~+ Ar2+) calculated by summation of the individual signal ratios for every reading are plotted as a function of the sample mass for Sea Lettuce. Since the signal ratio (75A~+ Ar2+) increases linearly with increasing sample mass it is clear that application of the Ar2+ signal as an internal standard allows the curvature of the 75As+ sample mass response curve to be corrected for accurately . Table 3 presents the results obtained for the quantification of As (by single standard addition) in Sea Lettuce (BCR CRM 279) using (i) Sb9 and (ii) Ar2+ as an internal standard. The results obtained were not deteriorated by the presence of C1 as even on addition of amounts of C1 exceeding the original content of the sample no 40Ar35C1+ interference on the 7 5 A ~ + signal could be established.' All the results were obtained under the same experimental conditions (e.g.calibration by single standard addition five sub-samples per replicate meas- uring sequence) discussed in detail el~ewhere.~ Comparison of these results with one another and with the certified value clearly demonstrates that in this instance Ar,+ corrects as successfully as Sb for the matrix-induced suppression observed such that its application as an internal standard allows accurate 14 I h .r C 12 .- i 6 a c .G 4 0 0.5 1 1.5 2 2.5 3 Sample mass/mg 300 .- C Fig. 5 75As+ signal intensity and (75A~+ "Ar,+) signal ratio as a function of the sample mass (solid sample) of Sea Lettuce (BCR CRM 279).The dashed line is only intended as an indication of the trend observed Table3 Results (pgg-') for the determination of As in Sea Lettuce (BCR CRM 279); calibration by single standard addition (two indepen- dent determinations for each internal standard used) Mean (SEM)* Mean (SEM)* Certified value using Sb as using Ar,+ as 3.09 0.20 3.18 (0.22) 2.96 (0.34) 3.24 (0.17) 3.31 (0.28) k 95 YO confidence limit internal standard internal standard * Standard error of the mean. 1050 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10results to be obtained. Owing to the transient nature of the signals involved application of Sb as an internal standard for the determination of As and/or Se implies (i) integration of the analyte signal intensity and the Sb+ signal intensity separately and (ii) division of the integrated signal intensities obtained (75A~+ 12'Sb+ and/or 82Se+ 12'Sb+).For this approach it is mandatory that the internal standard (Sb) shows a completely analogous behaviour to the analytes (As and Se). When using the argon dimer as an internal standard the ratio (As' ATz+) is calculated for every reading individu- ally and the final 'signal ratio' is obtained by summation of all the individual ratios. This approach offers the advantage of real-time correction for non-spectral interferences and should be valid more universally since similarity of signal profiles is no longer requested. Finally it should be stressed that for example in order to evaluate the importance of the differences in mass number between an analyte element and the argon dimer it is advisable to check the utility of the Ar,+ signal as an internal standard for each analyte and each matrix under consideration e.g.by plotting the ratio (M' Ar2+) as a function of the sample mass. During the analysis of materials of environmental origin registration of the argon dimer has also proved to be useful during the optimization of the vaporization temperature. For some materials of environmental origin (e.g. soils or sludges) ashing at a moderate temperature (200-400°C) does not provide an effective removal of the matrix. For the determi- nation of relatively volatile elements (As Cd Se) it is therefore advisable to work at a moderate vaporization temperature in order to volatilize the latter analytes before the major matrix components.Fig. 6(a)-(c) shows the signal profiles for 'llCd+ and Ar2+ for River Sediment (BCR CRM 320) at about 1370 1550 and 1700 "C respectively. The Ar2+ signal profiles indi- .; 1.4 . . . . n...X a . . . I . . . . 1 L . L 5 0 2.5 5 7.5 10 12.5 15 17.5 20 0 2.5 5 7.5 10 12.5 15 17.5 20 v > v) 0 c .- c .- 0 2.5 5 7.5 10 12.5 15 17.5 20 Timds Fig. 6 '*Ar,+ and 'llCd+ signal intensity as a function of time during the vaporization stage of the multi-step temperature programme for solid samples (about 0.7 mg) of River Sediment (BCR CRM 320) at a vaporization temperature of (a) 1370 (b) 1550 and (c) 1700 "C cate an increasing amount of co-volatilization of the matrix on increasing the vaporization temperature from 1370 to 1700°C.Since weighing of the empty sample holders before and after each analysis of a sample of one of the environmental materials investigated (Light Sandy Soil or River Sediment) systematically indicated a significant loss of mass a (chemical) reaction between the solid sample material and the sample holder material (pyrolytically coated graphite) could possibly contribute to the matrix effect observed. No significant loss of mass was established for the sample holders when analysing materials of biological origin. It is important in order to limit (i) the amount of matrix-induced signal suppression (ii) the deposition of sample material on the torch the interface and the lens stack and (iii) the degradation of the interface pump oil that no excessive amount of matrix material co-volatilizes with the analyte(s) as is for example to a very large extent the case in Fig.6(c). Hence simultaneous registration of the analyte signal (as an indication of analyte vaporization) and the argon dimer (as an indication of the amount of co-volatilizing matrix material and/or the extent to which the chemical reaction between the sample and the graphite sample holder proceeds) is useful during the optimization of the vaporization temperature for matrices of environmental origin. Simultaneous registration of the argon dimer also provides an explanation of why in these instances the signal intensity does not reach a constant value starting from a given vaporization temperature (as is the case for the samples of biological origin in~estigated):~,~~ the signal intensity starts to decrease at excessively high vaporization temperatures as a result of matrix effects.Since for environmental samples an appropriate selection of the vaporization temperature can significantly reduce the amount of matrix components co-volatilizing with the ana- lyte(s) (e.g. As Cd or Se) and hence the extent to which these analyte signal intensities are suppressed careful optimization of the vaporization temperature also has a direct bearing on the curvature of the corresponding sample mass response curves (signal intensity as a function of the sample mass). As a result of this optimization the curvature of the sample mass response curves was observed to be much less pronounced for environmental than for biological matrices,16 since for the latter at least for the elements investigated no conditions could be found leading to a complete separation of volatiliz- ation of analyte(s) and matrix respectively.For biological matrices on the other hand optimization of the ashing temperature is important. An optimum ashing temperature provides an efficient removal of the matrix with- out analyte losses. Since the Ar,' signal profile gives an indication of matrix effects it is clear that the ashing tempera- ture also exerts an influence on the Ar2+ profile such that registration of the argon dimer gives an indication of the effectiveness of the ashing step. Simultaneous registration of the analyte signal (to ensure that no analyte is lost during ashing) and the argon dimer (to control the effectiveness of the ashing step) during vaporization following ashing at suc- cessively increasing ashing temperatures is therefore useful during optimization of the ashing temperature.CONCLUSIONS It is shown as has already been demonstrated for pneumatic nebulization ICP-MS by Beauchemin et aZ.12 and Chen and Houk13 and for slurry sampling ETV-ICP-MS by GrCgoire et u Z . ~ that also for solid sampling ETV-ICP-MS monitoring of the Ar2+ signal provides a direct indication of plasma loading effects and non-spectral interferences. Monitoring of the Ar2+ signal intensity not only allows differences in the shape of signal profiles between different analytes and/or matrices to be explained but is also of use for optimization of Journal of Analytical Atomic Spectrometry December 1995 VoZ.10 1051operational conditions. Finally since it was demonstrated that the Ar,’ signal may at least in some instances be used as an internal standard to correct for matrix-induced signal suppres- sion effects the necessity of adding an appropriate elemental internal standard (similar in furnace chemistry and only present at a negligible level in the sample) for accurate quantification may be circumvented. Further research is presently being carried out in order to evaluate the universality of this approach. REFERENCES Gray A. L. Analyst 1985 110 551. Arrowsmith P. Anal. Chem. 1987 59 1437. Denoyer E. R. Fredeen K. J. and Hager J. W. Anal. Chem. 1991 63 445. Ulens K. Moens L. Dams R. Van Winckel S. and Vandevelde L. J. Anal. At. Spectrom. 1994 9 1243. Voellkopf U. Paul M. and Denoyer E. R. Fresenius’ J. Anal. Chem. 1992 342 917. Grkgoire D. C. Miller-Ihli N. J. and Sturgeon R. E. J. Anal. At. Spectrom. 1994 9 605. Wang J. Carey J. M. and Caruso J. A. Spectrochim. Acta Part B 1994 49 193. 8 9 10 11 12 13 14 15 16 17 18 19 Argentine M. D. and Barnes R. M. J. Anal. At. Spectrom. 1994 9 1371. Vanhaecke F. Boonen S. Moens L. and Dams R. J. Anal. At. Spectrom. 1995 10 81. Boonen S. Vanhaecke F. Moens L. and Dams R. Spectrochirn. Acta Part B 1995 in the press. Moens L. Verrept P. Boonen S. Vanhaecke F. and Dams R. Spectrochim. Acta Part B 1995 50,463. Beauchemin D. McLaren J. W. and Berman S. S. Spectrochim. Acta Part B 1987 42 467. Chen X. and Houk R. S. J. Anal. At. Spectrom. 1995 10 in the press. Verrept P. Dams R. and Kurfurst U. Fresenius’ J. Anal. Chem. 1993 346 1035. Denoyer E. R. At. Spectrosc. 1994 15 7. Galbacs G. Vanhaecke F. Moens L. and Dams R. unpub- lished work. Thompson J. J. and Houk R. S. Appl. Spectrosc. 1987 41 801. Doherty W. Spectrochim. Acta Part B 1989 44 263. Vanhaecke F. Vanhoe H. Dams R. and Vandecasteele C. Talanta 1992 39 737. Paper 51040771 Received June 23 1995 Accepted August 21 1995 1052 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10

ISSN:0267-9477    

DOI:10.1039/JA9951001047

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Chelation Preconcentration with Resin Analysis by Direct Sample Insertion Inductively Coupled Plasma Spectrometry ROBIN RATTRAY AND ERIC D. SALIN* Department of Chemistry McGill University 801 Sherbrooke St. West Montreal Quebec H3A 2K6 Canada Batch preconcentration with Chelex-100 followed by direct analysis of the analyte-laden resin by direct sample insertion inductively coupled plasma atomic emission spectrometry (DSI-ICP-AES) is described. The performance of the technique is element specific. Quantitative retention of Cu Zn Cd and Pb on the resin is achieved but only for Cu and Zn does the ratio of the signal before and after preconcentration approach the theoretical preconcentration factor. This observation is mainly caused by the adverse effect of the remnants of the resin after ashing on the excitation properties of the plasma.This is clearly shown by monitoring the ratio of the intensity of a Pb ionic line to a Pb atomic emission line. If the ashing temperature is increased Cd and Pb are prematurely vaporized in the ashing stage which is performed with inductive heating in the graphite DSI probe. Increasing the radiofrequency power sustaining the ICP improves the performance of the technique. Keywords Chelation; preconcentration; resin analysis; direct sample insertion; inductively coupled plasma spectrometry The use of chelating resins for preconcentration and matrix elimination is now popular for lowering detection limits and avoiding interferences in analytical atomic spectrometry. Most applications involve selective retention of the analyte on a resin column followed by elution with an appropriate solution.However the elution step contributes to dilution of the analyte the degree of dilution being dependent on the volume in which the analyte is removed. In addition some resins such as those based on poly(dithiocarbamate) are reluctant to release the bound analyte and the resin is usually dissolved prior to Dissolution is undesirable as it is labour intensive time consuming and a potential source of contamination. Milley and Chatt' directly analysed Chelex-100 resin using instrumental neutron activation analysis to determine the concentration of 15 trace elements in acid rain samples. They mentioned several of the disadvantages associated with analysis involving column elution and suggested that the approach should be generally applicable to other techniques amenable to the analysis of solid samples.Van Berkel and Maessen6 later reported the direct analysis of analyte-laden poly(dithi0- carbamate) resin using electrothermal vaporization (ETV) sample introduction for inductively coupled plasma atomic emission spectrometry (ICP-AES). They found that compro- mise conditions for ashing had to be chosen to avoid vaporiz- ation losses or matrix effects from the pyrolysis products of the resin. Good results were nevertheless obtained for Cu Zn Cd and Pb in a urine certified reference material. However the signal improvement factor achieved was not reported. A commercially available system based on analysis of the resin-bound analyte has been manufactured by CETAC * To whom correspondence should be addressed.I Journal of 1 Analytical 1 Atomic Spectrometry Technologie~.~ In their implementation of the concept sub- micrometre sized resin beads are mixed with the sample to retain the analyte. The mixture is then pumped through a hollow fibre bundle which traps the resin. Finally the resin beads are washed from the fibre bundle to a direct injection nebulizer (DIN) for introduction into the ICP. The system can be operated in a batch mode wherein about 1 ml of mixture can be collected off-line and later analysed. This mode permits the generation of a steady-state signal which may be more convenient to process with most spectrometer systems than a transient signal. Again however the preconcentration factors and therefore detection limits obtained are not optimum because of the dilution that occurs while removing the resin from the fibre bundle.There are other advantages to performing the analysis of the resin-bound analyte. Unlike column-based systems each sample is presented with a fresh portion of resin which makes resin regeneration unnecessary and avoids memory effects from the resin. Also by eliminating the elution step it may be easier to design more selective resins since only the chemistry of retention as compared with retention and elution is of concern. Direct sample insertion (DS18v9) sample introduction for ICP spectrometry seemed to be a suitable technology for direct analysis of the resin from a chelation preconcentration." The graphite probes commonly used can easily accommodate the milligram amounts of resin that would be used and the resin can be conveniently dried and ashed inductively in situ using methodology and instrumentation developed previously." The sample introduction efficiency for DSI is 100% so any precon- centration advantage procured with the resin should be pre- served.Based on these premises the investigations described in this paper were carried out. EXPERIMENTAL The Plasma Therm ICP and the modified Thermo Jarrell Ash spectrometer that were used have been described previously.'2 Briefly the modifications to the spectrometer included the installation of a galvanically driven quartz refractor plate behind the entrance slit for rapid background correction and a diode laser at the zero order for experimental alignment.I3 The stepper motor-driven DSI device that was used was built and operated with essentially the same characteristics as that described previ0us1y.l~ The temperature of the graphite DSI probe during the ashing step was measured with a Pyro Micro- Optical Pyrometer (Pyrometer Instrument Northvale NJ USA).Table 1 lists some of the important DSI ICP and data acquisition parameters that were used in this study. The data acquisition software used (SF20) was written by G. Leghe; an earlier version of this program with many of the salient capabilities is commercially available (Trulogic Systems Oakville Ontario Canada). The elements monitored were Cu Zn Cd and Pb at 324.7,213.8,228.8 and 220.3 nm respectively. Data processing was carried out with user-written programs Journal of Analytical Atomic Spectrometry December 1995 Vol.10 1053Table 1 DSI ICP and data acquisition parameters Insertion distance Insertion time Radiofrequency (r.f.) power Reflected power Viewing height Plasma gas flow rate Auxiliary gas flow rate Integration time per point Background measurement position 3 mm above TOLC* 20 s 1.0-1.75 kW <low 17 mm above TOI,C* 12-16 1 min-l 1.8-2.0 1 min-' 20 ms -0.1 nm below peak * TOLC = top of load coil. Insertion distance measured relative to top of DSI probe. and Lab Calc (Galactic Industries Salem NH USA). The peak area of the background corrected signal was used exclusively for quantification. Multi-element solutions approximately 1 YO in nitric acid were prepared by dilution of 1000ppm atomic absorption standards (Fisher Scientific) with 18 M a deionized distilled water obtained from a Milli-Q water purification system ( Millipore Bedford MA USA).Working solutions were freshly prepared each day. The preconcentration procedure used was as follows 1.0 ml of the multi-element solution contained in a 1.5 ml polypropylene flip-top microcentrifuge tube was adjusted to pH 9 by the addition of 75 pl of 2.0 mol 1-1 aqueous ammonia. The Chelex-100 resin (100-200 mesh) (Bio- Rad Laboratories Richmond CA USA) was converted into the ammonium form as described by Marino and 1ngle.l" A 100 pl aliquot of the resin slurry (about 7.5 mg of dry resin) was taken with an Eppendorf micropipette the tip of which had been cut off to enlarge the orifice and allow the resin to be able to pass through.This aliquot of resin slurry was added to the standard solution and mechanically agitated for 30 min. This time was found to be insufficient for complete extraction of Pb and was later increased to 3 h. The supernatant (100 pl) was analysed by DSI-ICP-AES to determine the extent of completion of the extraction process. A nitric acid blank M YO) was treated in an identical fashion. Resin samples were dried and ashed in the graphite DSI probe using the induction field in the load coil of the IlCP according to the protocol in Table 2. Two consecutive drying steps were necessary to prevent the slurry from boiling violently and being ejected from the probe. Two arshing schemes as described in Table2 termed 'low temperature' and 'high temperature' were initially studied.During the 'low- temperature' ashing step the graphite DSI probe barely glowed to a dull red colour suggesting a temperature of approximately 650 "C. This temperature is only approximate since the lowest temperature that could be measured with the optical pyrometer used was 700°C. However a good correlation was obtained between the net rf power applied (forward minus reflected) and the measured probe temperature; interpolation of that curve also indicated that the lowest ashing temperature used was about 650°C. If milder ashing conditions were used the plasma would be extinguished on insertion of the probe. Under the 'high-temperature' conditions the temperature was meas- ured as 815 "C. Two higher ashing temperatures (950 and Table 2 Drying and ashing protocol Forward Reflected Stage power/W power/W Duration/s Drying 1 25 8 30 Drying 2 50 16 60 Ashing (815 "C) 140 50 60 Ashing ( % 650 "C) 70 25 90 1080 "C) were also investigated later. The aqueous samples did not require ashing and were only dried as described in Table 2.RESULTS AND DISCUSSION Calibration at 1 kW The system was calibrated at 1 kW with 50 pl aliquots of aqueous Standards. Linear calibration graphs with log-log slopes of unity were obtained from 10 ppb (1 ng) to 1 ppm (100 ng). The signal produced by Cu at 1 ppm (100 ng) just saturated the detector; this upper limit was later extended by modifying the timing parameters in the SF20 data acquisition software. Detection limits (3s) were calculated from the slopes of these graphs and an estimate of the blank noise.These limits obtained at 1 kW were 0.2 0.2 0.6 and 0.6 ppb for Cu Zn Cd and Pb respectively. The ICP was initially operated at 1.0kW to generate the following data. However it was later realized that it was more prudent to use higher powers to avoid any problems with memory effects in the probe especially for Cu the least volatile of the four elements. The probe was also 'burned' in the plasma between runs to detect and remove any residual analyte. Analysis of the supernatant from an extraction of 1 ml of a 100 ppb multielement solution revealed that quantitative extraction was being achieved for all elements except for Pb. As mentioned before these initial experiments used a relatively short extraction time (30 min).By increasing the extraction time to 3 h all the elements were quantitatively removed from solution suggesting that the kinetics of the extraction for Pb are slower than for the other elements. However when the resin was transferred to the DSI probe and analysed lower values than expected (1 m1/100 pl= 10) for the signal improve- ment factor were observed. In fact for Cd and Pb the signal was suppressed to or even below the level seen with a 100 p1 aliquot of the same solution (Table 3). Further experiments were done to determine the cause of these apparent discrepancies. Possible sources of the losses observed included premature volatilization and/or physical ejection of the resin from the probe on insertion. However when aqueous standards were subjected to the 'low- temperature' ashing protocol applied to the resin analysis no significant losses were observed even for Cd the most volatile element studied.When ashed at 815°C about 70% of the Cd and 20% of the Pb were lost but Cu and Zn were not appreciably affected. Fig. 1 summarizes the effect of ashing temperature on the signal obtained from a 100 pl aliquot of a solution 50 ppb in Cu and Zn and 100 ppb in Cd and Pb. When the resin ashed at 650 "C was inserted into the plasma qualitative changes in the ICP were seen. An intense green flash from the top of the probe of about 1 s duration was Table 3 Preconcentration with Chelex-100-DSI-ICP-AES ( 1 kW) average peak area (n = 3) (arbitrary units) CU Zn Cd Pb Solution analysis- 100 p1 100 ppb 20 214 3185 3638 568 100 pl supernatant nd* nd 36 130 from 100 ppb extraction Extraction (%) 100 100 99 75 Analysis of resin- 1 ml 100 ppb 136 100 12 758 3893 306 Blank 450 206 nd nd Signal improvement 6.5 3.7 1 0.6 factor * nd =No peak detected.1054 Journal of Analytical Atomic Spectrometry December 1995 Vol. 101.2 1 1 600 700 800 900 I000 1100 Ashing temperaturd'c Fig. 1 Variation of ashing temperature (aqueous standards). A Cu; B Pb; C Cd; D Zn followed by a pale orange glow which lasted several seconds further up in the 'tail' of the plasma. These emissions were due to products of resin pyrolysis; they were also observed when the blank resin but not the aqueous standards were analysed. For example the green flash was almost certainly due to emission from the C2 Swan bands arising from components of the hydrocarbon skeleton of the resin.At higher ashing tem- peratures this green emission was not observed suggesting that the concentration of these species was dramatically reduced or eliminated. The following experiments were done to test the hypothesis that these species were adversely affecting the excitation properties of the plasma. The pure resin was first ashed (65OOC) in the probe as described before. A 100 pl spike of the 100 ppb standard multi- element solution was then dried on top of the resin ash and the probe was inserted into the ICP. The response obtained was compared with the response to the standard solution alone and the response from the pure resin. These experiments were performed in triplicate; the averaged results are presented in Table 4. The results in Table4 and the previous experiments imply that the presence of the resin ash was suppressing the analyte signal and that the losses were not due to premature volatiliz- ation during the ashing stage.In addition at the Cd I Pb I1 and Zn I channels the background (off-line) signal intensity from the ashed resin was essentially the same as from the aqueous standards so the decrease in sensitivity was not due to a change in spectral characteristics. However at longer wavelengths (e.g. in the Cu I channel at 324.8 nm) a 'dip' in the background was seen which suggests that the plasma temperature has probably changed. This is similar to the well known 'pressure pulse' phenomenon observed after vaporization in many ETV-ICP-AES system^.'^.^^ These observations reinforce the hypothesis that the decrease in signal is at least partially due to the effect of the products of pyrolysis of the resin on the excitation properties of the ICP.This hypothesis is supported further by the fact that the severity of the interference observed increases in the same order as the excitation or ionization potential for the analytical lines that were used (Table 4). Similar conclusions were drawn by the previous study using electrothermal vaporization and the poly(dithi0carbamate) resin6 During these experiments (at 1 kW) it was noted that the background-corrected signal for Cu did not return to the baseline and there was a slight memory effect. Also the peak shapes for Zn were ragged although they were much sharper with the resin ash present.Because of these observations no further experiments were carried out at 1 kW. Variation of RF Power In an attempt to confirm and overcome the matrix interference a series of experiments were performed at various rf power levels. At each level the response to 1OOpl of a 100 ppb solution (10 ng of analyte) was determined. As expected an increase in sensitivity was observed at elevated power (Fig. 2). The ratio of the response of these aqueous solutions (100 pl) to the response of the resin extract of 1 ml of the same solution (treated as described under Experimental) was calculated. While improvements in this ratio were generally observed as the rf power level was raised (Fig. 3) the signal improvement was still well below the theoretical preconcentration factor of 10 particularly for Cd and Pb.These graphs represent only one experiment per data point which may explain the erratic trend of the Pb curve. The trend of higher ratios at high power is however applicable to all elements. Variation of Plasma Observation Height A short study of the effect of varying the position in the plasma that was viewed by the spectrometer was carried out to find out if the interferences could be spatially resolved. The ICP was operated at 1.25 kW and the results are summarized in Fig. 4. The emission intensities decreased as the viewing height increased (Fig. 4) and there was no appreciable change in the ratio of the signals from aqueous standard and analyte-loaded resin. Qualitatively however the signals seem to be smoother and more reproducible when measured at 25mm above the top of the load coil.As a result of the previous experiments the calibration with aqueous standards was later repeated at 1.5 kW in order to enhance the removal of the analyte from the probe and to reduce the matrix effects which appeared to be less severe at higher power. Again the graphs were linear with log-log slopes near unity. Detection limits were not calculated but the slopes of these graphs are four times greater for Cd and Zn and two times greater for Cu and Pb than at 1.0 kW. Variation of Ashing Temperature A high ashing temperature is desirable to remove as much of the interfering resin pyrolysis products from the probe as possible. The effect of ashing at a higher temperature was therefore examined.A multi-element solution (1 ml) 5 ppb in Cu and Zn and 10 ppb in Cd and Pb was extracted with Chelex-100 as described under Experimental and the resin was ashed at both 650 and 815°C. A 1% nitric acid blank was also taken but the resin from extraction of this blank was only ashed at the lower temperature to minimize loss of Cd or Pb. The experiments were carried out in triplicate at 1.5 kW and the results (averaged) are presented in Table 5. Table 4 Effect of resin ash on analyte signal (1 kW). Signal (peak area) is the average of three measurements (arbitrary units) Pure 100 p1 Resin ash+ Suppression Excitation Species resin 100 ppb 100 p1 100 ppb ("/) potentiallev CU' 200 20 214 16 159 20 3.82 Cd' nd* 3638 1594 56 5.42 Zn' nd 3185 459 86 5.79 Pb" nd 568 nd 100 7.42 * nd =No peak detected.Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1055v) 0 0 0 3 25- u * - . - - 1 1.2 1.4 1.6 1.8 v) 0 0 51 1100 1000 s 900 I 800 700 a a 600 Power/W . - . . . . . . . 1 1.2 1.4 1.6 1.8 Fig. 2 Effect of rf power on signal from liquid standards (a) Cu; (b) Zn; (c) Cd; and ( d ) Pb 0.6 0.5 0.4 0.3 0.2 3.5 3 2.5 2 1.5 1.4 1.3 1,2 1.1 1 1.2 1.4 1.6 1.8 1 1.2 1.4 1.6 1.8 PowerkW Fig. 3 Effect of variation of rf power on intensity ratio (a) Cu; (b) Zn; (c) Cd; and (d) Pb 300 i I *-*--I- v) 0 2 50 1 1 " I t\ I % 500 &-- 16 18 20 22 24 26 16 18 20 22 24 26 Viewing heighvmm above TOLC Fig. 4 Viewing height variation. A Standard and B resin 1056 Journal of Analytical Atomic Spectrometry December 1995 Vol.10Table 5 Effect of ashing temperature (1.5 kW). Values are averages of three measurements peak area in arbitrary units Table 7 Characteristics of species used and experimental conditions in intensity ratio study 650°C ash 815 "C ash Blank (650 "C ash) 100 pl 50 ppb (Cu Zn) 100 pl 100 ppb (Cd Pb) Signal ratio (650 "C) Signal ratio (8 15 "C)t RSD (Yo) RSD (Yo) c u 21 011 15 22 073 2 2674 2457 7.5 7.9 Zn 7450 17 8751 6 1505 808 7.4 9.0 Cd 936 7 nd* nd nd 1499 0.62 0 Pb 211 23 243 3 nd 137 1.5 1.8 * nd = No peak detected. t Blank-extracted resin ashed at 650 "C used for blank correction. Qualitative differences were observed in the response to the two ashing methods. For Cu when the resin was ashed at the lower temperature a 'spike' in the signal was always observed 0.8 s after probe insertion and 2.0 s before the main peak.After the higher temperature ash this feature was either greatly diminished or absent from the signal profile (Fig. 5). The spike is probably due to the physical ejection of some analyte with the sudden vaporization of resin pyrolysis prod- ucts that still remain in the probe after a low-temperature ash. It is likely that all elements are being affected but the effect is obvious for Cu despite the short residence time in the plasma owing to its low excitation potential (Table 4). The low vola- tility of this element also allows good temporal resolution of this effect from the main analyte peak. A comparison of the peak areas suggests that the loss due to physical ejection is about 5%.It can be seen from Table 5 that quantitatively Cu Zn and Pb showed only slight improvements after ashing at a higher temperature suggesting that the interfering species were still not completely removed in the ashing process. For Cu under low-temperature ashing conditions the total peak area (i.e. spike plus main analyte signal) was used. Cd was completely lost when ashed at 815°C but was seen under the milder conditions. Also noteworthy is the fact that the precision of the measurements was much better under higher temperature ashing conditions possibly because of the absence of the disruptive effect of the rapid volatilization of resin pyrolysis products as the probe entered the plasma. Several strategies were adopted in order to monitor the entry of the resin pyrolysis products into the plasma.The emission signal from the C I line at 193.1 nm was monitored during both ashing methods to determine the extent of the hydrocarbon load (in the form of C) in the plasma. However neither this line nor the H I line at 486.1 nm proved to be particularly useful as the signals tended to plateau after a low resin load (2 mg) in the DSI probe. The ratio of the intensity of emission lines arising from ionic and atomic species of the same element has been used as a diagnostic tool in AES.'*-'' Previous workers with resin analy- sis by ETV-ICP-AES used the Cu I1 224.7 nm:Cu I 324.7 nm ratio to verify that the excitation conditions were in fact changing. The Cu I1 line at 224.7 nm was not available with our spectrometer but another Cu I1 line at 213.6nm was C u I CuII P b I P b I I Wavelength/nm 324.8 213.6 405.8 220.4 Ionization po tential/eV na* 7.7 na 7.4 Excitation potential/eV 3.8 8.5 4.4 7.3 Energy sumt/eV 3.8 16.2 4.4 14.7 Ashing conditions- Forward power/W 70 140 200 260 Temperature/" C - 650 815 9 50 1080 Reflected power/W 25 50 75 100 * na =Not applicable. t Ionization potential plus excitation potential?' Table 6 Effect of ashing temperature on Cu atomic line ionic line intensity ratio accessible in the existing Zn I channel by adjusting the position of the refractor plate at the exit slit assembly.The 'energy sums' (ionization energy plus excitation energy") of these two lines are comparable (224.7 nm= 15.9 eV; 213.6 nm= 16.2 eV22) and similar results to the previous work with ETV- ICP-AES were expected.However when the measurements were made with DSI-ICP-AES there was no difference between the Cu 1:Cu I1 intensity ratio calculated from aqueous stan- dards (100 ply 50 ppb) alone in the presence of the resin ashed at 650 or 815 "C (Table 6). This apparent discrepancy can be explained by considering the vaporization characteristics of the species of interest. The appearance time of the peak maximum is a measure of the volatilization rate and is 0.1 1.5 1.9 and 2.8 s after probe insertion for Cd Pb Zn and Cu respectively. However the pyrolysis products are rapidly volatilized approximately 1 s after insertion based on visual inspection and the 'spike' observed in the Cu signal profile (Fig. 5). This suggests that the elements of higher volatility that co-vaporize with the resin breakdown products would be more prone to detrimental effects of varying plasma excitation conditions.Cu on the other hand vaporizes much later and is therefore at least partially temporally resolved from this interference. A channel for a PbI emission line (405.8 nm Table 7) was therefore installed in the spectrometer and the experiment repeated to compare the behaviours of these two elements of differing volatilization properties. A series of ashing temperatures were used to note more carefully any trends in performance. Fig. 6 clearly shows that as the ashing temperature was increased the Pb line intensity ratio increased dramatically (approximately six-fold) whereas the Cu line intensity ratio remained essentially constant.These data confirm and explain the experimental data presented earlier. CONCLUSIONS Direct analysis of the resin from a Chelex-100 type preconcen- tration by DSI-ICP-AES does not seem universally applicable to trace analysis because of the matrix effects of the resin ash on the plasma excitation conditions. Elements of lower vola- tility are affected less because they are temporally separated from the interfering matrix which volatilizes shortly after insertion. Some compensation can be achieved by working at high rf power but variation of the viewing height does not appear to be beneficial. Using a high ashing temperature Cu I peak area* Cu I1 peak area* Ratio Cu I Cu I1 Aqueous standard (dried only) 25 450 2120 12.0 Aqueous standard + resin (650 "C ash) 22 630 1880 12.0 Aqueous standard + resin (8 15 "C ash) 26 010 2170 12.0 * Peak areas in arbitrary units.Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 10576000 4000 2000 0 0 .- c E 20 - ----On-peak sign& 0 2 4 6 8 10 12 14 16 Time/s / Fig. 5 Effect of ashing temperature on Cu signal from loaded resin (1.5 kW) (a) approximately 650 "C and (b) 815 "C results in losses of all but the least volatile elements. The pyrolysis products of other resins or solid phases used in preconcentration may not have as detrimental an effect and may be more suitable to this application. Although not investi- gated here the beneficial aspects of the elimination of the sample matrix may still make the technique feasible for the analysis of certain sample types.Also worthy of investigation may be the use of larger sample volumes to provide higher preconcentration factors since the resin capacity is high enough for the sample volume to be increased by at least an order of magnitude with an expected signal improvement of approxi- mately two orders of magnitude to be expected for less vola.tile elements such as Fe Mn Co and Ni. This work was supported by the Ontario Ministry of the Environment and Energy (Project 5746) and was also facili- tated by the assistance of Fisons Instruments. R.R. gratefully acknowledges financial support from the University of the West Indies and the Canadian International Development Agency. 25 30 k c .- ---a *!15 j 10 fi 3 ,/ f l r ; 600 700 800 900 1000 1100 Ashing temperat urd0C Fig.6 Pb I1:Pb I; and B Cu 1:Cu I1 Variation of intensity ratios with ashing temperature A REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Murthy R. S. S. Horvath Z. and Barnes R. M. J. Anal. At. Spectrom. 1986 1 269. Barnes R. M. and Genna J. S. Anal. Chem. 1979,51 1065. Barnes R. M. Fodor P. Inagaki K. and Fodor M. Spectrochim. Acta Part B 1983 38 245. Fodor P. and Barnes R. M. Spectrochim. Acta Part B 1983 38 229. Milley J. E. and Chatt A. J. Radioanal. Nucl. Chem. Articles 1987 110 345. Van Berkel W. W. and Maessen F. J. M. J. Spectrochim. Acta Part B 1988 43 1337. Smith F. Wiedierin D. and Gjerde D. Winter Conference on Plasma Spectrochemistry Sun Diego January 1994 Paper 131 0. Salin E. D. and Horlick G. Anal. Chem. 1979 51 2284. Karanassios V. and Horlick G. Spectrochim. Acta Rev. 1990 13 89. RbiiEka .I. personal communication. Rattrdy R. Minoso J. and Salin E. D. J. Anal. At. Spectrom. 1993 8 1031. Legere G. and Burgener P. ZCP ZnJ Newsl. 1988 13 521. Ren J. M. Legere G. and Salin E. D. Appl. Spectrosc. 1993 47 1953. Sing R. L. A. and Salin E. D. Anal. Chem. 1989 61 163. Marino D. F. and Ingle J. Anal. Chem. 1981 53 292. Gunn A. M. Millard D. L. and Kirkbright G. F. Analyst 1978 103 1066. Ng K. C. and Caruso J. A. Anal. Chim. Acta 1982 143 209. Boumans P. W. J. M. Theory of Spectrochemical Excitation Plenum Press New York 1966. Mermet J. M. Spectrochim. Acta Part B 1989 44 1109. Carre M. Poussel E. and Mermet J. M. J. Anal. At. Spectrom. 1992 7 791. Murillo M. and Mermet J. M. Spectrochim. Acta Part B 1987 42 1151. Robinson J. W. Handbook of Spectroscopy CRC Press Cleveland OH 1974. Paper 51008285 Received February 2 1995 Accepted July 24 1995 1 058 Journal of Analytical Atomic Spectrometry December 1995 Vol. I0

ISSN:0267-9477    

DOI:10.1039/JA9951001053

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Speciation of Inorganic Selenium Using Flow Injection Hydride Generation Atomic Fluorescence Spectrometry Journal of Analytical Atomic Spectrometry D. W. BRYCE A. IZQUIERDO AND M. D . LUQUE D E CASTRO Department of Analytical Chemistry Fuculty of Sciences University of Cbrdoba 14004 Spain Two methods based on flow injection atomic fluorescence have been developed in order to speciate Se as Se'"-SeV'. Both methods use hydride generation of Sew with atomic fluorescence detection as the derivatization-detection step. In the first method two sample plugs are injected simultaneously in series so that the first plug passes straight to the detector to determine Se'". The second plug passes through a focused microwave device where Se"' is reduced to Sew prior to its conversion into the hydride.The Se"' content is then given as the difference between the two results. In the second method a mini-column is used to retain both the Se species; Sew and Sen are then eluted sequentially with formic and hydrochloric acids respectively. The columns can be used for preconcentration of the analytes as well as for sampling. Both methods show exceptional sensitivity [limits of detection (3s) of 0.04 pg 1-'1 and wide linear range up to 50 pg I-' with excellent linearity (9 = 0.999) and reproducibility (relative standard deviation<5%). The methods have been used in an intercomparison exercise for the Measurements and Testing Programme and have been applied to the determination of the analytes in tap water and haemodialysis samples. Keywords Flow injection; hydride generation; atomic fluorescence spectrometry; selenium speciation In the last couple of decades information on the speciation of certain elements (particularly Hg Se As and Cr) has become increasingly important as the effects of the different forms are better understood.Selenium is one such case with special importance owing to its ambivalent behaviour it is essential at low concentrations but toxic at high concentrations with a relatively small difference between these levels.'" This coupled with the fact that the concentration levels found are extremely low means that sensitive and accurate methods are needed for speciation of this element. Methods based on sophisticated expensive instruments such as isotope dilution mass spe~trometry,~ stimulated Raman scattering5 and hydride generation inductively coupled plasma mass spectrometry6 have been reported for toxic elements in the last decade.Speciation in all instances is achieved by the use of a powerful separation technique prior to the introduction of the sample into the instruments. Several flow injection (FI) methods for the speciation of Se have been developed recently including microwave-assisted hydride generation atomic absorption spectrometryY7.* and FI cathodic- and anodic-stripping v~ltamrnetry.~~'~ Atomic fluorescence has been used recently for the determi- nation of Hg and the hydride forming elernents."*l2 These methods have the advantages of excellent sensitivity with wide linear ranges are relatively free from interferences and have little memory effects. Small dedicated instruments based on atomic fluorescence which provide excellent features of the methods thus developed with accessible prices are on the market at present.PS Analyti~al'~.'~ have developed two detectors (Merlin and Excalibur) for the determination of Hg and Se Cd and As and the excellent features of these detectors have been reported in several paper^.'^-'^ The work described in the present paper had two main objectives. These were firstly to design a sensitive FI method for the speciation of Se for use in an intercomparison exercise on speciation of Se in water samples organized by the Measurements and Testing Programme (Community Bureau of Reference formerly BCR) and secondly to develop a system incorporating mini-columns to fulfil a triple aim i.e. to achieve the speciation of inorganic Se and carry out the preconcen- tration and the sampling steps in an automated continuous way.EXPERIMENTAL Apparatus An Excalibur atomic fluorescence detector (PS Analytical Orpington Kent UK) fitted with (a boosted discharge) hollow cathode lamp for Se (Photron Narre-Warren Victoria Australia) and a specific ultraviolet (UV) filter to allow transmission of the Se atomic fluorescence spectrum in con- junction with a solar blind photomultiplier was used. A Microdigest 301 focused microwave system (Prolabo Paris France) was used together with two Gilson (Worthington OH USA) Minipuls-3 programmable pumps fitted with rate selectors and three Rheodyne 5041 injection valves (two of which were connected in order to function as a double injection valve).Teflon tubing of id of 0.5mm was used in order to construct the FI manifold. Reagents and Solutions All reagents were of analytical-reagent grade. Ultrapure water obtained from a Millipore (Milford MA USA) Milli-Q system was used throughout. Hydrochloric acid of 6 and 1 mol I-' (Panreac Barcelona Spain) 1.5% m/v NaBH solution (Aldrich Milwaukee WI USA) in 0.1 mol 1-' NaOH solution and formic acid (Merck Darmstadt Germany) were used as the carriers in the system. The Na2Se03 (Merck) and Na2Se0 (Aldrich) stock solutions of 1 g1-I were prepared in Milli-Q water. Working solutions were prepared daily by appropriate dilution in Milli-Q water. Argon 5N (SEO Barcelona Spain) and hydrogen (Carburos Metalicos Barcelona Spain) were used in order to flush the hydrides formed to the detector.Mini-column Teflon tubing of 2mm id was filled with Dowex-1 (chloride form) strongly basic anion exchanger (200-400 mesh) (Sigma St. Louis MO USA) (with cotton wool placed at each end). The exchanger was conditioned with 0.1 moll-' HCl before being thoroughly washed with water (Milli-Q) and then used for the retention-preconcentration of the analytes. Procedure The manifold used for the speciation of Se based on dual simultaneous injection is shown in Fig. l(a). The loops of the Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1059I HCI I I / GLS NaBH. - CR IV (b 1 pp Sample 2 CHOOH HCI + w Fig. 1 FI manifolds for the speciation of selenium based on (a) simultaneous dual injection of the sample and (b) sequential elution of Se'" and Se"' where HC1=6moll-'; NaBH4=1.5% (in NaOH 0.1 moll-'); D=detector; MC=minicolumn; CHOzH=2 moll-'; G = argon; IVl-.3 =injection valves 1-3; M =microwave system; CR = cooling reactor; GLS = gas-liquid separator; PP = peristaltic pump; sample 1 = Se in HCl 6 moll-'; sample 2 = Se in H20; and W = waste (for method (b) HCl=2 mol 1-' NaBH4= 1.0'3'0 G=argon+ hydrogen) double injection valve are filled with the same sample in 6 moll-' HCl and injected simultaneously into the carrier stream also of 6 mol 1-' HCl.The plug injected via valve 1 passes to the confluence point where it merges with the stream of NaBH solution where SeIV is converted into the hydride form before continuing to the gas-liquid separator where the hydride is flushed to the detector by a stream of argon giving rise to a signal corresponding to the Se" concentration in the sample.Meanwhile the plug injected via IV passes through a reactor placed in the chamber of a focused microwave device where SeV' is reduced to Se". This plug then continues through another reactor placed in an ice-bath (in order to reduce the temperature of the sample) before continuing to the confluence point where the Se all in the form of Se" forms the hydride as explained above thus giving rise to a signal corresponding to the concentration of SeIV+SeV' initially present in the sample. The concentration of Sev' is then calculated as the difference between the two results. In the second method based on the selective elution of SeIV and Se" the speciation is performed after a step where the two oxidation states are retained on a mini-column prior to selective elution.The manifold used in this method is shown in Fig. l(b). In this case three simultaneous injections are used (with valves 2 and 3 being a double injection valve thus allowing valve 1 to be operated independently for the precon- centration of the analytes). The loops of the valves IV1 IV and IV3 are filled with the sample (this time without HCl) CH02H and HCl respectively. The three valves are changed to the inject position simultaneously so that the three plugs are inserted into the carrier stream (in this case H,O). The sample plug is carried to a mini-column containing the Dowex-1 strongly anionic exchanger situated in the carrier channel where SeIV and Sew are retained.As the plug of CH0,H passes through the column it selectively elutes Se'". This plug of CH02H containing SeIV then continues to the first confluence point where it mixes with a stream of 6 mol 1-' HCl before passing through a reactor placed inside the micro- wave device and then through a cooling coil (this step is not actually necessary as no reaction is taking place but it keeps the manifold simple i.e. omits the need for a splitting point). From here the stream is mixed with one of NaBH solution at the second confluence point where the Se" forms the hydride before being passed from the gas-liquid separator to the detector by the stream of argon giving the signal for Se". Following this the plug of HC1 injected via valve 3 passes through the column eluting the SeV1.This stream mixes with the stream of HC1 before passing through the microwave device where the Sevl is reduced to Se". This once cooled mixes with the stream of NaBH solution forming the hydride and then giving the signal for the SeV1 present in the initial sample. Signal measurement was carried out on the basis of peak height. RESULTS AND DISCUSSION Fundamental Steps of the Atomic Fluorescence Spectrometric Determination The determination step (derivatization and monitoring) was common to both methods and was based on the formation of the hydride derivative of Se"' and atomic fluorescence of the analyte in a hydrogen flame. Organoselenium compounds such as selenomethionine and selenocystine would not give any signal as they are not reducible under the experimental conditions used. The determination can be split into three sub- steps.Firstly the hydride has to be formed which is carried out in the presence of NaBH in an acidic medium. In the second step the hydride and the hydrogen formed are swept out of the gas-liquid separator by a stream of argon into a chemically generated hydrogen diffusion flame (the flame is maintained by the excess of hydrogen produced in the reaction between NaBH and HC1). In the third step the hydride is atomized in the flame and the atoms are detected by atomic fluorescence spectrometry. These detectors and gas-liquid sep- arators have been well documented in the literat~re.".'~.~~ Optimization Determination step The univariate method was used throughout the optimization process.For the optimization of this common step the FI variable that was studied was the flow rate the chemical variables were the NaBH and HC1 concentrations. The final variable that was vital was the flow rate of the argon. As the overall flow rate of the FI system was increased so the sensitivity increased although at values above 3.5 ml min-' the hydrogen flame became large which was characterized by a high signal-to-noise ratio zs has been reported previously in the literature.' Clearly the sampling frequency was also found to increase with the increased flow rates. As a compro- mise between sensitivity signal-to-noise ratio and sampling frequency a flow rate of 2.9 ml min-' was chosen permitting good sensitivity signal-to-noise ratio and sampling frequency.The NaBH concentration was studied in the range 0.5-2.0% m/v (in 0.1 mol 1-' NaOH solution). At concentrations lower than 1.2% m/v it was difficult to maintain the hydrogen flame as insufficient H2 was formed. Above 1.2% m/v the analytical signal improved very slightly and so 1.5% m/v was chosen as the optimum concentration. The concentration of HCl had a minimal effect (less than k 2%) on the analytical signal in the range 1.5-8 moll-'. However as the reduction of SeV' is carried out optimally in 6 moll-' HC1," this concentration was chosen. Although it was necessary for the samples to be in 6 mol I-' HCl the HCl concentration in the carrier stream could be as low as 2 mol 1-' which was sufficient to maintain 1060 Journal of Analytical Atomic Spectrometry December 1995 VoE.10the flame; however this resulted in signals for the blank (as a plug of 6mol1-' HC1 was being injected into a stream of 2 moll-' HCl) thus increasing the limit of detection and so 6 moll-' was used. The flow rate of argon was also of prime importance. As the flow rate was increased so the peak height and sampling frequency increased although above 1000 ml min-' the argon flow rate was so great that it extinguished the hydrogen flame. As a result a flow rate of 900 ml min-' was chosen as optimum. One other stage that was common to both methods was the reduction of SeV'. A method for the on-line reduction of SeV' aided by a focused microwave device had been developed previously by workers in this laboratory,20 and was adapted for this method.The HCl concentration had to be 6 moll-' (below this concentration incomplete reduction of SeV' was observed and at concentrations above 6 moll-' no increase was observed). Microwave power of 50 W and a flow rate through the microwave device of 1.0ml min-' were also necessary for the complete reduction of Se". One modification of the determination step was developed with the object of making method 1 less expensive. In this case the NaBH concentration was decreased to 1.0% m/v (i.e. a 33.3% reduction in NaBH,) the HC1 concentration was reduced to 2mol1-' (i.e. a 66.7% reduction in HCl) and a gas mixture of argon-hydrogen (540+ 3 ml min-') was used to flush the hydride to the detector. In this way the additional flow of H gas was used to maintain the hydrogen flame.The flow rate was increased to 3.9 ml min-'. These adaptations resulted in a more cost efficient rapid method which showed slightly poorer sensitivity and reproducibility but was suitable for routine analysis. This method is subsequently called 'method l B whereas the original method 1 is referred to as 'method 1A'. Speciation based on dual simultaneous injection method 1 The speciation of Se'v-Sev' in method 1 was achieved by controlling the dynamic system. The concentration of the different oxidation states is obtained by the simultaneous injection of two plugs into the system. The first plug gives the Se" concentration and the second plug after reduction of SeV' to SeIV gives the concentration of Se" + SeV' from which the Sev' concentration can be obtained by the difference in the two results.The optimum injected volume for both plugs was 600 p1. Below this value the peak height was smaller whereas above this value the peak height remained more or less constant although the peaks tended to be a lot broader with the possibility of forming double peaks which affected the repro- ducibility. The samples were prepared in 6 moll-' HCl in order to facilitate the reduction of Sev' to Se". The conditions of the reduction process have been described previously2' and were maintained for the present work. The length of the cooling reactor [CR in Fig. l(a)] was studied over the range 0-400 cm. The purpose of this reactor was two-fold firstly to cool the stream coming out of the microwave system and secondly to allow the first peak to return to the baseline prior to the start of the second peak thus improving reproducibility. A length of 200cm was found to be optimum as it allowed sufficient separation of the peaks without being too detrimental to the sampling frequency. Speciation based on selective elution of Se" and Se" method 2 In this case speciation was achieved by on-line retention of the two species on a mini-column prior to their selective elution.After a literature search Dowex-1 strongly anionic exchanger (chloride form) 200-400 mesh was used as the packing material. The column was prepared as described under Experimental. For this system an additional channel ( H20) had to be added to the manifold in which the mini-column was placed.Again an injection volume of 600 p1 for the sample was used and an overall flow rate and flow rate through the microwave system were maintained at the values given for method 1. With initial concentrations of 3 and 6mol1-' for CHOzH and HCl respectively the elution volumes of the two acids were studied between 100 and 1000p1. In order to check that elution was complete the eluent was injected three times after every injection of SeIV in order to make sure that all of the Se" had been eluted after the first injection of CH02H. This cleaning step was repeated for the study of the elution of SeV1 with HC1. For CH02H the peak height increased as the elution volume was increased up to 600 p1 above which a slight decrease in peak height was observed. Similarly for the elution volume for Se" i.e.the injected volume of HC1 the peak height increased as the injected volume was increased up to a volume of 350 pl above which a slight decrease was observed. For all further experiments 600 and 350 pl were used for CH02H and HCl respectively. The lengths of the reactors between the three injection valves were as short as possible (i.e. 10 cm). This was sufficiently long to enable the Se to be retained on the column prior to the elution by CH02H and HCl and also allowed the first peak to return to the baseline before the second peak started. As far as the concentrations of CH0,H and HCl were concerned the CH20H concentration was studied from 1 to 3 moll-'. Below 2 mol 1-' the elution was incomplete (ca. 75%) whereas above 2mol 1-' 100% elution was observed.As a result 2 moll-' was chosen as the optimum concentration. For the elution of Se" 6mol1-' HCl was used providing 100% elution. Concentrations below this value were not studied as the Sev' had to be in 6moll-' HCl in order for it to be reduced to Se" and concentrations above this were not studied as they would have no advantages over 6mol1-'. Hence 6mol1-' was used for all further experiments. The flow rates of the channels containing H 2 0 and HCl were each set at 0.5 ml min-' so that when the two channels combined at the confluence point the overall flow rate i.e. the flow rate through the microwave system was maintained at its optimum value of 1 .O ml min - ' . The details of the optimization process i.e. the variables studied the ranges studied and the optimum conditions are all outlined in Table 1.Table 1 Optimization of variables Type of variable Common variables Method 1A Method 1B Method 2 Variable Overall flow rate/ml min-' Argon flow rate/ml min-' [ HCl]/mol 1 - [NaBH,] (YO m/v) [ NaOH]/mol 1 - ' Injected volume 1/p1 Injected volume 2/pl Reactor length/cm Argon flow rate/ml min-' Hydrogen flow rate/ml min-' [HCl]/mol 1-' [NaBH,] (YO m/v) Overall flow rate/ml min-' Injected volume l/pl Injected volume 2/pl Injected volume 3/pl [CHO,H]/mol 1-' [HCl]/mol 1-' H,O flow rate/ml min-' HCl flow rate/ml min-' Range studied 1-4 1 00- 1000 1-8 1-2 100-lo00 100-1000 0-350 loo- 1000 0- 10 1-6 0.5-2 1-5 100-1000 100-1000 100-1000 1-3 - 6 0.5 0.5 Optimum value 2.9 900 6 1.5 0.1 600 600 200 540 3 2 1 3.9 600 600 3 50 2 6 0.5 0.5 Journal of Analytical Atomic Spectrometry December 1995 Vol.10 1061Interferences Once the systems had been optimized a literature search was carried out in order to know which species usually interfere in the atomic fluorescence spectrometric determination of SeIV. This revealed that Co Bi Pd Ni As Pb and Cu are the most common interferents. Various concentrations of these foreign species were then added to a solution containing 2 pg 1-' Se" with the object of determining the highest tolerable ratio of foreign species to analyte with +5% of the original signal being considered as having no interference. Method 1A was used to determine the interference effects. The results obtained are as follows (foreign species to analyte ratio tolerated without interference) Cu > 500 1; Co > 1000 1; Pb > 1000 1; Bi 500 1; Ni > 1000 1; As"' 1000 > 1; Pd 250 1.These relatively high tolerance levels particularly for Cu can be explained by the high acid concentration that is used. This ensures Cu remains in solution thus preventing it from destroying the SeH which could occur if Cu was allowed to precipitate out. The method shows excellent tolerance to foreign species. This is partly due to the fact that the hydrides are separated from the carrier stream as a gas prior to detection. Only Pd interferes strongly (at a foreign species to analyte ratio of 500 1 a reduction in peak height of 65% was observed) but is not considered important as Pd is very unlikely to be present in any great amounts in the samples being investigated.Features of the Method When the optimum value of each of the variables had been found the characteristics of the methods were studied. For the calibration curves of the two methods based on dual simultaneous injection (methods 1A and lB) nine standard solutions each of Se" SeV' and SerV+SeV1 were prepared in 6mol1-' HC1 with concentrations ranging from 0.15 to 50 pg 1-' of each species and were injected in triplicate into the FI systems. The methods both showed exceptionally good linearity with coefficients of correlation r2 better than 0.999 in all cases although method 1A had a slightly better sensi- tivity. In order to study the reproducibility of the results for Se" from the first and second peaks and SeV' from the second Table 2 Features of the proposed methods peak 11 solutions of both low (0.5 pg 1-') and high (25 pg 1-') Se" and SeV' concentrations were injected in triplicate.The reproducibility expressed as YO RSD was excellent in all cases although slightly better for method 1A. The limits of detection were expressed as 3s for the low level reproducibility study for method 1A and 3s for the blank signal (injected 22 times) for method 1B. Again method 1A showed slightly better character- istics with limits of detection of 0.04 and 0.06 pg 1-' and 0.11 and 0.13 pgl-' for Se" and SeV' in methods 1A and lB respectively. Sampling frequencies of 40 and 50 h-' were observed for methods 1A and lB respectively. These tests were all repeated for the method based on the selective elution of SeIV and SeV' (method 2) with the exception of those which were unnecessary (i.e.calibration and reproduc- ibility of Se" in the second peak). A sampling frequency of 15 h-l was observed for this method. The method showed a limit of detection of 0.07 pg 1-'. In addition to these tests method 2 was also studied to check its suitability for the preconcentration of Se" and Sevl. A solution of 0.025 pg 1-' of SeIV and SeV' was preconcentrated ten-fold prior to the elution step. This was carried out by filling the loop of valve 1 with sample and injecting it ten times before the loops of the double injection valve were filled with CHOzH and HCl and injected to elute the preconcentrated analytes. The reproduc- ibility of the ten-fold preconcentration process expressed as YO RSD of 11 repetitions was excellent showing the potential of the mini-columns.All the equations coefficients of correlation and reproducibilities are shown in Table 2. Applications Method 1A was used in an intercomparison exercise organized by the Measurements and Testing Programme (formerly the BCR) on the speciation of Se in water samples. Two solutions (five bottles each) of 2-9 and 20-70 pg 1-' of SeIV and SeV' were analysed. A separate batch of the same samples was also determined by FI cathodic-stripping voltammetry2' showing good agreement between the two methods. The results obtained are shown in Table 3. Following this method 2 was applied to the determination Method Analyte 1A Se" 1st peak Se" 2nd peak Sevl 2nd peak 1B SeIV 1st peak SeIV 2nd peak SeV' 2nd peak 2 SeIV Sev' Equation y = 5.70~ + 1.29 y = 5 .0 1 ~ + 0.93 ~ = 4 . 8 9 ~ + 1.34 ~ ~ 4 . 7 7 ~ - 1.97 y= 4 . 2 9 ~ - 2.39 JI = 4.27~ - 2.1 1 y = 3.00~ + 0.75 y= 1.60~-0.37 Linear range/ Pug 1-' 0.1-50 0.1-50 0.1-50 0.25-50 0.25-50 0.25-50 0.15-50 0.15-50 Regression coefficient ( r 2 ) 0.9995 0.9988 0.9995 0.9990 0.9984 0.9988 0.9993 0.9999 Limits of detection (3s)/ Pg I-' 0.04t 0.06 - 0 . l l t 0.13$ 0.07 0.06 Reproducibility RSD (%); n = l l * 5.44 (0.5) 2.98 (25) 4.76 (0.5) 4.03 (25) 5.26 (0.5) 4.93 (25) - - - 5.70 (0.025)g 6.75 (0.025)$ * At the concentration levels given in parentheses (pg 1-l). For SeIV. $ For Sevl. 0 For a sample preconcentrated by a factor of ten repeated 11 times. Table 3 Intercomparison study; all results are in p g 1-1 Method 1A FI Solution Se" Se'" + SeV' SeV' SeIV SeIV + SeV1 SeV' Low concentration 5.2 k 0.2 12.4 & 0.9 7.3 0.9 5.2 f 0.7 11.9 k 2.6 6.7 _+ 2.4 High concentration 35.3 f 1.3 78.8 k 3.7 43.5 f 3.7 37.8 f 4.2 80.8 14.8 43.1 +_ 7.4 1062 Journal of Analytical Atomic Spectrometry December 1995 Vol.10Table 4 Results (pg 1-') for analyses of spiked tap water samples Se" added Sevl added SeIV found SeV1 found 0.5 0.5 0.51 0.52 2 2 2.06 1.93 10 10 9.50 9.78 0.1 0.1 0.09" 0.104" 0.05 0.05 0.045+ 0.055-f 0.015 0.0 15 0.014+ 0.014t * After preconcentration by a factor of five. After preconcentration by a factor of ten. Table 5 Results (pug I-') for analyses of haemodialysis samples after preconcentration by a factor of ten Sample A B Se" found 0.32 0.08 SeV1 found 0.09 0.06 of Se" and SeV1 in tap water.Even after preconcentration by a factor of 20 no Se was found in the water. Thus Se" and Sev' at concentrations of 0.015 0.05 0.1 0.5 2 and 10 pg 1-1 were added to the water and the recoveries checked (preconcen- trating the first three solutions on the column prior to elution). Finally two samples used in the haemodialysis procedure in a local hospital were analysed. The maximum legal Se (total) concentration in such samples is 90 pg 1-'. These samples along with the a tap water sample spiked with 0.05 pg1-' were analysed after a preconcentration factor of ten. All the results obtained for both methods are shown in Tables 4 and 5. CONCLUSION Two extremely sensitive methods and one slightly less so (suitable for routine analysis) with exceptionally good linearity and linear ranges reproducibility and sampling frequencies have been developed and applied to real samples.The develop- ment of mini-columns for the retention and specific elution of Se" and SeV' allows the speciation of Se to be carried out giving an SeV1 concentration that is independent of the Se" concentration which is therefore more accurate than when the concentration of SeV1 is calculated as a difference. The methods developed show increased automation in com- parison with other more expensive complex methods reported recently for speciation of Se. The methods reported here follow the trend of increased automation with steps such as precon- centration elution reduction speciation and detection all being carried out on-line.In addition the methods compare favourably in terms of automation limits of detection linear range sampling frequency simplicity and cost (particularly method lB which shows 33.3 and 66.7% saving in the amounts of NaBH and HC1 used respectively as well as a 25% saving in time) with other methods reported recently. The mini-columns developed should be easily applicable to use in field sampling studies after investigations of resin capacity and the effect of humic/fluvic acids and pH of the sample have been carried out. PS Analytical is thanked for the loan of an Excalibur atomic fluorescence selenium detector. One of the authors (D.W.B.) would like to thank the Measurements and Testing Programme (formerly BCR) for a doctorate grant covering the expenses incurred during his stay in Spain.REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Schroder H. A. Frost D. V. and Balassa J. J. Chronic Dis. 1970 23 227. Robinson W. O. J . Assoc. Off. Agric. Chem. 1933 16 423. Underwood E. J. Trace Elements in Human and Animal Nutrition Academic Press New York 4th edn. 1977. Tanzer D. and Heuman K. Anal. Chem. 1991 63 1984. Leong M. B. D'Silva A. P. and Fassel V. A. Anal. Chem. 1986 58 2594. Buckley W. T. Budac J. J. Godfrey D. V. and Koening K. M. Anal. Chem. 1992 64 724. Cobo-Fernandez M. G. Palacios M. A. Chakraborti D. Quevauviller Ph. and Camara C. Fresenius' J. Anal. Chem. 1995 351 438. Pitts L. Worsfold P. J. and Hill S . J. Analyst 1994 119 2785. Bryce D. W. Izquierdo A. and Luque de Castro M . D. Fresenius' J. Anal. Chem. 1995 351 433. Bryce D. W. Izquierdo A. and Luque de Castro M. D. Anal. Chim. Acta. 1995 308 96. Stockwell P. B. and Corns W. T. Int. Lab. 1994 19 33. Ebdon L. Goodall P. Hill S. J. Stockwell P. B. and Thompson K. C. J. Anal. At. Spectrom. 1993 8 723. Stockwell P. B. Thompson K. C. Henson A. Temmerman E. and Vandecasteele C. Int. Lab. 1989 14 45. Corns W. T. Stockwell P. B. Ebdon L. and Hill S . J. J. Anal. At. Spectrom. 1993 8 71. Stockwell P. B. and Corns W. T. J. Autom. Chem. 1993 15,79. Corns W. T. Ebdon L. Hill S. J. and Stockwell P. B. J. Autam. Chem. 1991 13 267. Stockwell P. B. and Corns W. T. Hydrocarbon Asia 1993 October 36. Corns W. T. Ebdon L. Hill S. J. and Stockwell P. B. Analyst 1992 117 717. Ebdon L. Corns W. T. Stockwell P. B. and Stockwell P . M. J. Autom. Chem. 1989 11 247. Stockwell P. B. and Corns W. T. Analyst 1994 119 1641. Bryce D. W. Izquierdo A. and Luque de Castro M . D. Analyst 1995 120 2171. Paper 5/02687C Received April 27 1995 Accepted August 11 1995 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1063

ISSN:0267-9477    

DOI:10.1039/JA9951001059

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Determination of Mercury by Electrothermal Atomic Absorption Spectrometry Using Different Chemical Modifiers or a Slurry Technique Journal of Analytical Atomic Spectrometry I. KARADJOVA P. MANDJUKOV S. TSAKOVSKY AND V . SIMEONOV Faculty of Chemistry University of Sofa 1 J. Bourchier Sofa 11 26 Bulgaria J. A. STRATIS AND G. A. ZACHARIADIS School of Chemistry Aristotle University Thessaloniki 540.06 Greece The applicability of different chemical modifiers for thermal stabilization and ETAAS determination of mercury is studied. The modifier effect is strongly influenced by the type of acid and acid content in the sample solution. A method is described for mercury determination in tuna fish and sediment after high pressure digestion with nitric acid using thioacetamide (TAC) as chemical modifier.The method permits determination of 0.5 pg g- ' mercury. A simpler and faster procedure using a slurry technique for the determination of mercury in reference materials (tuna fish spinach cabbage and sediments) was also evaluated. Optimal instrumental parameters for mercury determination in slurries are presented. On the basis of the results obtained a new procedure which allows determination of 0.1 pg g- ' mercury was developed. Keywords Mercury; electrothermal atomic absorption spectrometry; modiJer; slurry technique Cold vapour atomic absorption spectrometry (CVAAS) with or without amalgamation is one of the most widely used methods for mercury determination because of its high sensi- tivity and reliability. However this technique requires additional equipment to the conventional instrumentation and a large number of reagents are used which can give rise to a relatively high blank value.Solid-phase microsampling is not possible and also some matrix interferences on vapour gener- ation have been observed. Electrothermal atomic absorption spectrometry (ETAAS) is another common and relatively sensitive method for mercury determination although the high volatility of mercury and some of its compounds even at room temperature offers a lot of difficulties in its analytical application. A number of investigations have been presented for mercury stabilization in the graphite furnace. Thermal stability of mercury has been achieved by using different modifiers intro- duced into the furnace as a solution (NH4)2S,1 H202 and HCl,2-5 Te di~hromate,~ etc.Grobensky et aL8 proposed another scheme for mercury stabilization; injection of the modifier (PdCl,) into the furnace followed by appropriate heating and after cooling injection of the sample. However some of the conclusions of the papers cited above are ambigu- ous and depend strongly on the particular analysis. For optimal ETAAS determination of mercury efficient thermal stabilization is necessary in both the drying and pretreatment steps. In our opinion only chemical reactions can stabilize mercury in the drying step; the role of the modifier is in the formation of a thermally stable compound of mercury preventing reduction of Hg2+ to Hg'. In this study the use of different organic and inorganic modifiers (Pd as PdCI2 Ce as Ce(NH4),(N0,) TAC (thio- acetamide CH,CSNH,) and mixture of H202 and HCl) were investigated and compared for thermal stabilization and ETAAS determination of mercury in different reference mate- rials.Optimal instrumental parameters for determination of Hg after high pressure digestion with nitric acid using TAC as modifier are presented. A simple procedure for analysis of Hg after slurry preparation is also proposed. The results obtained by the above techniques are compared with those obtained by the conventional cold vapour technique. EXPERIMENTAL Apparatus The AAS measurements were carried out on a Perkin-Elmer Zeeman 3030 atomic absorption spectrometer coupled with an HGA 600 atomizer. The light source was an electrodeless discharge lamp for mercury.The spectral bandpass and the wavelength used were as recommended by Perkin-Elmer. Uncoated and pyrolytic graphite coated graphite tubes were used as atomizers. Solutions (10-20 pl) were introduced into the graphite furnace using an AS-60 autosampler (Perkin- Elmer). Atomic absorption signals were recorded on an Anadex printer. The apparatus for the cold vapour determinations included a 100 ml Dreschler bottle a humidity trap with Mg(C104)2 as desiccant and a 15 cm cylindrical cell mounted on the AA spectrometer. Reagents Nitric acid hydrochloric acid and hydrogen peroxide used for sample digestion and slurry preparation were of Suprapur grade (Merck). All other reagents were of analytical-reagent grade. Mercury stock solution 1000 mg 1-' was Spectrosol grade (Merck).Modifier solutions were prepared by dissolving appropriate amounts of PdCl Ce(NH,),(NO,) or TAC in doubly distilled water. Sample Preparation Procedure 1 wet digestion An accurately weighed amount (about 500mg dry mass) of spinach cabbage sediment or tuna fish was placed in a Teflon crucible and 7.0ml of concentrated HNO were added. The crucible was covered placed in a pressure bomb and left for 3 h at a temperature of 90°C. The resulting solution was transferred into a 10 ml (25 ml for CVAAS) calibrated flask and diluted to the mark with doubly distilled water. The procedure for digestion of the sediments was the same except that a mixture of 4 ml of concentrated HNO + 2.5 ml Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1065of concentrated HC104+ 1.5 ml of concentrated HF was used for the acid digestion and the temperature was raised to 145 "C.Table 1 Maximum pretreatment temperatures for the loss-free deter- mination of mercury (0.1 pg ml-') obtained with different Pd or Ce concentrations Procedure 2 slurry preparation The dried samples were ground in an agate ball-mill until the particle size was < 50 pm. An accurately weighed amount of sample (about 50-100 mg) was placed into a polyethylene vessel and mixed with 1 ml of an aqueous solution containing 6% v/v HC1+2% v/v HN03+4% v/v HzOz. Finally 0.1 ml of an aqueous solution of 0.5% v/v Triton X-100 was added and the slurry was prepared after manual shaking. Atomic Absorption Measurements E TAAS An aliquot of 10-20 p1 of the digested sample solution or the slurry solution was injected into the graphite furnace.The chemical modifiers were injected either separately or pre-mixed with the tested solution. CVAAS An aliquot of 10 ml of the digested sample solution (procedure 1) was transferred into the Dreschel bottle with 15 ml of doubly distilled water and a suitable portion of a 5% m/v solution of SnCl was added to reduce mercury. The physico- chemical and instrumental conditions were as described in ref. 9. RESULTS AND DISCUSSION ETAAS Determination of Mercury Pd as PdCl and Ce as Ce(NH4)2(N03)6 as modijiers Palladium and cerium were chosen as representatives of two types of modifiers. It is known that Pd exists as Pdo and Ce as CeO during the pretreatment step. Previous investigations showed that with these modifiers the losses in the drying step were strongly dependent on the type of acid and acid content in the sample solution.For this reason for model investi- gations aqueous standard solutions of mercury were used. Various Pd and Ce concentrations from 0.1-500 pg ml-' (stoichiometric ratio Pd:Hg) were studied for thermal stabiliz- ation of 0.1 pg ml-' Hg. As can be seen from the pretreatment and atomization curves (Fig. l ) even very low concentrations 0.15 -$ 0.13 0 $ 0.11 a 3 0.09 2 0 c) = 0.07 0.05 A o 200 400 600 aoo 1000 1200 Ternperat ure/"C Fig. 1 A Pretreatment and B atomization curves for mercury stabil- ization (20 pl of 0.1 pg ml-' Hg) in presence of different concentrations of Pd (10 pl) 0 0.2 pg ml-'; 0 1 pg ml-'; 0 50 pg ml-'; and .500 pg ml-' Pd/pg ml - T/"C Ce/pg ml - ' T/"C 0.2 200 0.2 200 0.5 300 0.5 250 1 350 1 300 5 3 50 5 3 50 50 400 250 3 50 500 400 500 400 of Pd or Ce stabilize mercury up to 200°C. Enhancement in the concentrations of Pd or Ce leads to an increase in the maximum pretreatment temperatures (Table 1). Also for modi- fier concentrations in the range 0.1-50 pg ml-' no difference in the sensitivity for mercury determination (integrated absorbance was used in measurements) is observed when maximum power or 1 s ramp time in the atomization step was used. An atomization temperature of 850°C gives the best sensitivity. Higher atomization temperatures decrease sensi- tivity owing to convection losses between the pretreatment and atomization step in connection with the high volatility of mercury species.The presence of Pd or Ce at 500 pg ml-' level allows higher pretreatment temperatures to be used about 400 "C but the sensitivity is lower. It might be assumed that a high modifier content promotes the trapping of mercury in a matrix crystal lattice thereby decreasing its equilibrium vapour pressure. This requires higher atomization temperatures and even with maximum power convection losses prior to the atomization step are remarkable. The absorbance signal profiles and appearance time for mer- cury as can be expected depend on the concentration of Pd or Ce. Using aqueous standard solutions and Pd as modifier losses during the drying step are around 25% (Table 2) and do not depend on Pd concentration. Probably the reduction of Pd in the graphite furnace prevents some losses of mercury.The same is almost true for standard solutions of mercury in the presence of hydrochloric acid with Pd or Ce as modifiers. Unfortunately in the presence of nitric acid (nitric acid content around 10% v/v in the sample solution) losses of mercury during drying step are around 80% even with 500 pg ml-' Pd or Ce as modifiers (Table2). A probable explanation is the formation of Hg(N03) in the presence of nitric acid which decomposes at a much lower temperature (79 "C) in compari- son with HgC1 (mp 276°C and bp 302"C).10 Therefore it is impossible to employ the modifiers Pd or Ce for mercury determination in nitric acid digested samples. Mixture of HC1+ H202 + Pd and TAC as modijiers A mixture of HCI and H,Oz as modifier especially for the drying step has been pr~posed.~ This modifier permits a maximum pretreatment temperature of 240 "C.Taking into account our previous investigations it is evident that the mixed modifier HCl + H202 + Pd will ensure a maximum pretreat- Table 2 Mercury losses during the drying step (130 "C) using Pd as modifier mercury concentration 0.05 pg ml-' 20 pl sample aliquot Peak area (arbitrary units) t/s Pd 5 pg ml-'* Pd 50 pg ml-'* Pd 500 pg ml-'t 15 0.071 0.073 30 0.066 0.068 60 0.053 0.055 0.065 0.03 5 0.014 * Dissolved in 10% v/v HC1. t Dissolved in 10% v/v HNO,. 1066 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10ment temperature of around 350 "C without losses during the drying step. This modifier was used for the determination of mercury in various samples (tuna fish and sediments) dissolved after high-pressure digestion with nitric acid.Unfortunately in the presence of a high nitric acid content (>2% in the sample solution) this modifier does not work. Marked losses of mercury were observed when these samples were introduced into the graphite furnace. Further investigations showed that in the presence of nitric acid the modifier which gives suitable pretreatment stabiliz- ation as well as temperature stabilization during the drying step is TAC with an optimal concentration of 10 g 1-'. The TAC hydrolyses in acidic solution according to the reaction CH3CSNH2 + 2H20 + H + +CH3COOH + NH4+ + H2S and thus permits the formation of HgS during the drying step. The role of this modifier is also to reduce the excess of nitric acid and thus to prevent the formation of thermally unstable Hg(N03)2.According to the pretreatment and atomization curves (Fig. 2) obtained for mercury standards with high nitric acid content and TAC as modifier the maximum pretreatment temperature is 230°C. This modifier was used in the determi- nation of mercury in reference samples tuna fish and sediments prepared according to procedure 1. Modifier solution 10 pl of l o g 1-' TAC was injected separately and mixed with the sample solution in the furnace using the AS-60 autosampler. For both tuna fish and sediment using instrumental conditions summarized in Table 3 and TAC as modifier the results for the determination of mercury were in good agreement with v) 0.05 \ 0 m 0.04 8 Ll 'p 2 0.03 2 CI) L = 0.02 0.06 r- - - - - A f I \" b I L 0.01 1 I 1 I 0 200 400 600 800 1000 1200 Temperature/"C Fig.2 A Pretreatment and B atomization curves for mercury deter- mination in presence of different matrices 0 tuna fish acid digestion sample with TAC as modifier; 0 sediment IAEA 158 slurry technique; and 0 cabbage IAEA 359 slurry technique Table 3 Instrumental parameters for mercury determination the certified values. The detection limit obtained was 0.5 pg g- ' (3s criteria six parallel determinations). The standard additions method is recommended for calibration. Results obtained for tuna fish and sediments are compared with those obtained with the cold vapour technique and good agreement was achieved (Table 4). ETAAS Determination of Mercury by Slurry Atomization During any digestion procedure used for sample preparation in mercury determination mercury losses may occur.An alternative approach for determining such a volatile element is direct solid microsampling or slurry atomization. Slurries were prepared according to procedure 2. On the basis of some preliminary results a mixture of 6% vjv HC1+2% v/v HN03 + 4% H202 ensures thermal stabilization of leached mercury in the drying step and also is very suitable for slurry homogenization. The addition of Triton X- 100 prevents slurry precipitation for a few minutes. Slurries are manually shaken before every manual injection into the graphite furnace. The optimal instrumental parameters were obtained using pretreat- ment and atomization curves for every sample treated accord- ing to procedure 2 (typical examples for sediment and cabbage are depicted in Fig.2) and are summarized in Table 3. Better sensitivity was achieved when uncoated graphite tubes were used as atomizers. In order to reach higher pretreatment temperatures without mercury losses Pd (50 pg ml- ') was mixed with the slurries before injection. The results obtained are almost the same as without Pd for tuna fish cabbage and spinach slurries. The pretreatment temperature of 240 "C achieved without Pd is good enough to minimize non-specific absorption. Also it is better to use a 1 s ramp time in the atomization step than maximum power. For slurries obtained from sediments the atomization temperature is critical. It is important to atomize mercury before the matrix element atomization so an atomization temperature of 900°C is used.In the presence of Pd as modifier a higher atomization temperature has to be used and thus very high values of non- specific absorption were observed (Fig. 3). As a conclusion for mercury determination by slurry atomization under the experimental conditions described there is no need to use Pd as modifier. Slurries have different physicochemical properties compared with aqueous solutions and it is clear that it is best to calibrate against certified reference material prepared in the same way as the sample. For slurries obtained from sediments this is the only possibility. Results achieved using slurry atomization with such calibration agreed very well with results achieved after acid digestion and ETAAS or CVAAS determination of Step Drying Pretreatment Atomization Cleaning Conditions T/"C Ramp/s Hold/s Gas flow/ml min-' T/"C Ramp/s Hold/s Gas flow/ml min-' T/"C Ramp/s Hold/s Gas flow/ml min-' T/"C Ramp/s Hold/s Gas flow/ml min-' Wet digestion (tuna sediment) 130 10 10 300 240 10 20 300 850 1 8 0 2200 1 3 300 Slurry (tuna spinach cabbage) 130 10 10 300 220 15 25 300 900 1 8 0 2300 2 4 300 Slurry (sediment) 130 10 10 300 220 20 35 300 900 1 8 0 2500 3 5 3 00 Journal of Analytical Atomic Spectrometry December 1995 Vol.10 1067Table 4 Comparison of results for mercury content (pg g-') in reference materials; n=6 BG 0.1 ( a ) ETAAS Reference material wet digested sample slurry technique Tuna fish (IAEA 350) 3.72 k0.06 3.97 k 0.08 Cabbage (IAEA 359) < DL 0.1 5 0.07 Spinach (IAEA 331) < DL 0.1910.07 Pine needles NIST SRM 1575 < DL 0.17+0.06 River sediment NIST SRM 1645 1.24 + 0.09 Sediment (Quasimeme 1993-94) 1.05f0.07 1.22f0.09 0.97 & 0.07 Certified CVAAS value 3.82 +0.08 3.9 0.1 1 f 0.08 NA* 0.17 k 0.08 NA - 0.15 1.1 1.03 k 0.08 NA - * NA = not available.0 8 Timds Fig. 3 ETAAS determination of Hg in sediment IAEA 1588 by slurry technique. Effect of atomization temperature ( xt) on absorbance signal (AA) profile for Hg (solid line) and background absorbance (BG) signals profile (broken line) at a pretreatment temperature of 300 "C (a) T = 900 "C; and (b) T = 1600 "C mercury (Table4). For slurries obtained from tuna fish cab- bage or spinach the standard additions method also might be used for calibration.The slurry analysis results for these samples compare well with concentrations obtained using a method based on acid digestion and analysis by ETAASl or CVAAS (Table 4). The detection limit for mercury by slurry technique is 0.1 pg 8-l ( 3 s criteria n= 10). CONCLUSION Some of the difficulties associated with mercury determinaf ion in some sample types e.g. soils vegetation and biota can be overcome by using chemical modifiers or a slurry technique. The application of well known modifiers such as Pd or Ce does not always provide a ready solution for any type of matrix. For nitric acid digested samples correct results for mercury content can be achieved by using TAC as modifier. The slurry sampling technique is proposed as a useful and effective method for rapid quantitative mercury determination even in a complex sediment matrix. REFERENCES 1 2 3 4 5 6 7 8 9 10 Ediger R. At. Absorpt. Newsl. 1975 14 127. Issaq H. and Zielinski W. Anal. Chem. 1974 46 1436. Alder J. and Hickman D. Anal. Chem. 1977 49 336. Owens J. and Gladney E. Anal. Chem. 1976,48 787. Lendero L. and Krivan V. Anal. Chem. 1982,54 579. Analytical Methods for Furnace Atomic Absorption Spectroscopy Perkin-Elmer Uberlingen Germany 1979. Rattonetti A. Instrumentation Laboratory Inc.; Report No. 12 Wilmington MA 1980. Grobenski Z. Erler W. and Vollkopf U. At. Spectrosc. 1985 6 91. Zahariadis G. and Stratis J. J. Anal. At. Spectrom. 1991 6 239. Handbook of Chemistry and Physics 56th edn. CRC Press Cleveland OH 1975-1976. Paper 51002788 Received January 17 1995 Accepted June 7 1995 1068 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10

ISSN:0267-9477    

DOI:10.1039/JA9951001065

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

Investigation of Automated Determination of Germanium by Hydride Generation Using in situ Trapping on Stable Coatings in Electrothermal Atomic Absorption Spectrometry* HERMANN 0. HAUG AND LIAO YIPINGt Forschungszentrum Karlsruhe Institut f u r Technische Chemie Postfach 3640 0-76021 Karlsruhe Germany Sequestering and in situ concentration of Ge hydride in the graphite furnace can be automated by using a highly stable trapping reagent to replace the Pd modifier. In a systematic study two groups of trapping reagents which require only a single application i.e. carbide-forming elements (Zr Nb Ta or W) and noble metals (Ir Pd-Ir) were investigated and trapping temperature curves were measured. It was shown that effective trapping of germane is possible on Zr-coated tubes and platforms at trapping temperatures of 550-750 and 600-800 OC respectively. Trapping temperatures should not exceed 650 "C (the 'critical temperature') because at temperatures higher than 650 "C errors in absorbance values could occur owing to an adsorptive 'carry-over effect'.Good signal stability was observed over more than 400 complete trapping and atomization cycles and a precision of better 3% was obtained. Comparatively small signals were observed for the Nb- Ta- and W-coatings. Ir-coated graphite tubes allowed trapping of germane at lower temperatures (400-500 "C) but the signals were small and of low stability compared with those for the Zr coating. Characteristic masses of about 54 pg of Ge on Zr-coated graphite tubes (peak height) and 108 pg of Ge on Zr-coated platforms (integrated absorbance) were observed and the calibration graphs were linear up to 4 ng of Ge on both tubes and platforms.The detection limit was 18 pg of Ge for a 1 ml sample volume using flow injection hydride generation. The method was tested by applying it to the determination of Ge in sediment geological and low-alloy steel certified reference materials. Keywords Germanium hydride; in situ hydride trapping; graphite furnace atomizer; stable trapping reagent; atomic absorption spectrometry The determination of volatile hydride-forming elements by atomic absorption spectrometry (AAS) using hydride gener- ation combined with the graphite furnace as both the hydride trapping and preconcentration medium and the atomization cell has been reviewed by Yan and Ni.' The technique has been described in a number of and can lead to significantly enhanced sensitivities and effectively eliminates interferences in the atomizer.In Fig. 1 an attempt is made to summarize and illustrate the working ranges for the determi- nation of Ge by the application of different AAS detectors including hydride generation; it can be seen that an inherent preconcentration of the analyte can be achieved by the in situ trapping of the hydride. * Presented in part at the Colloquium Analytische t On leave from Peking University Beijing People's Republic Atomspectroskopie Konstanz Germany April 1995. of China. Journal of Analytical Atomic Spectrometry Most of the hydride techniques published have used batch or continuous hydride generation followed by transfer of the gaseous hydrides into the pre-heated graphite tube via a quartz capillary which was manually inserted into the injection h ~ l e .~ - ~ Continuous hydride generator? as well as flow- injection system^^.^ have been used to automate the Ge hydride generation step. The transfer of the hydride into the graphite tube via a quartz capillary mounted on the autosampler arm was reported by Li et al.,' while Tao and Fang' used a separate mechanism for moving and inserting the quartz capillary into the furnace. It is known that Pd on the surface of the graphite tube or platform acts as an efficient adsorber for the volatile hydrides of the elements of Groups IVA-VIA of the Periodic Table leading to an effective deposition of the analyte element on the Pd-treated s ~ r f a c e .~ . ~ - l ~ A catalytic dissociation on the Pd was suggested by Sturgeon et aL8 Pd significantly improved the sensitivity and precision of the in situ trapping of the hydrides in the graphite furnace. Pd-treated graphite tubes and platforms were successfully applied to the technique of in situ trapping for germane (GeH,) by Doidge et ~ l . ~ Zhang et al. Tao and Fang,' and Ma et a1.,6 and in our previous work.2 The application of the Pd solution could be carried out automatically as part of the furnace cycle by the PTFE capillary of the sample dispenser and the quartz capillary for the introduction of the hydride was manually inserted and removed.' Alternatively an automatic insertion and removal of the quartz capillary by the autosampler could be used after applying the Pd solution manually.A fully automated process combining hydride generation plus graphite furnace has been limited by the requirement of a high and constant trapping efficiency over many firings. i l l l l l 7 I / 1 / 1 ' / / I I l i l l l l l I 1 1 1 1 1 1 1 1 I I I Solution flame Solution ETAAS HG-flame t---. HG-ETAAS 0.0 1 0.1 1 10 100 1000 Concentration/ng ml-' Fig. 1 Concentration ranges for the determination of Ge by AAS techniques and detection limits (DL) (calculated for 1 ml of solution) Solution in flame AAS x=DL (ref. 20); solution ETAAS (vol. 20 pl) (ref. 2) x=DL (ref. 21) Pd modifier; HG-flame AAS (vol. 1 ml) x = DL (ref. 15); and HG-ETAAS (vol. 1 ml) x=DL (this work) Zr-coated tube Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1069It appears that the Pd trapping reagent must be applied to the graphite surface before each hydride trapping cycle.Pipetting of the Pd solution by the quartz capillary was disregarded by Shuttler et al." because of the danger of catalytic decomposition of the hydrides by any residual Pd. A second mechanical device for the movement of the quartz capillary as used by Tao and Fang' could be avoided by trapping the hydrides on a more durable coating of the graphite surface. A simple and elegant approach has been the appli- cation of the highly stable mixed Pd-Ir trapping reagent for As Se Bi which had to be applied only once when a new graphite tube or platform was installed." Stable signals for at least 300 complete hydride generation and furnace cycles were reported for Se." For trapping of germane on Ir-Mg- coated platforms ( Perkin-Elmer transversly heated furnace) Schlemmer and Feuerstein12 reported about 80-100 trapping cycles before the Ir-Mg coating had to be re-applied.Ni et ~ 1 . ' ~ and Yan and Nil4 used Zr-coated graphite tubes for trapping Sn and Pb hydrides and found tube lifetimes of 150-200 firings. Insertion of the quartz capillary into the graphite tube and its removal were performed automatically. The Zr-coated graphite tube was preferred by Ni et ~ l . ' ~ 'because the Zr remained on the graphite tube and the enhanced trapping effect for Sn lasted for more than 100 firings'. Ge hydride generation using acetate buffer solutions was reported by Castillo et a1.15 and by Halicz.16 Sodium acetate- acetic acid at pH 4-5 allowed the difficulties with the narrow acidity range for the generation of germane and interferences from foreign ions to be circumvented.The acetate buffer was also used in our previous investigation of the in situ collection of Ge hydride on Pd modifiers. As the reaction medium 0.2moll-' phosphoric acid was used by Jin et al.17 in batch hydride generation and by Nakahara and Wasa18 in continuous hydride generation. Interferences from transition metals were reduced by the addition of 0.1 moll-' thiourea" for the determination of Ge in iron meteorites. Brindle et ~ 1 . ' ~ suggested 0.02 moll HNO + 0.4% L-cysteine for reduction of interferences from transition elements e.g. for Ge in copper and iron samples.In the present work a comparative search was made for trapping reagents for Ge hydride that only needed to be applied once to a new graphite tube or platform. In addition hydride trapping on coatings with carbide-forming elements (Zr Nb Ta W) as well as noble metals (Ir Ir-Mg and Pd-Ir) was studied. EXPERIMENTAL Instrumentation Measurements were performed using a Varian SpectrAA-800 Zeeman atomic absorption spectrometer with a GTA-100 graphite furnace and programmable sample dispenser (Varian Techtron Mulgrave Australia). The Ge wavelength of 265.2 nm was selected from a Ge hollow cathode lamp (Varian) operated at 5 mA. Pyrolytic graphite coated partitioned graphite tubes (Varian Part No. 63-100012-00) and pyrolytic graphite coated graphite tubes with an integrated pyro- lytic graphite platform (Varian Part No.63-100123-90 Schunk Kohlenstofftechnik Germany) were used; henceforth these will be referred to as 'graphite tubes' and 'platforms' respectively. Both peak heights and peak areas were usually recorded. Hydride generation was accomplished using a VGA-76 continuous vapour generation accessory (Varian Techtron) as well as a FIAS-400 flow injection unit and AS-90 autosampler ( Perkin-Elmer Uberlingen Germany) operated by a separate computer. Reagents All chemicals were of analytical-reagent grade or better. De-ionized water was obtained from a Milli-Q System (Millipore Eschborn Germany). A Ge standard solution (Alfa-Johnson Matthey Karlsruhe Germany) with a Ge concentration of 1000 pg ml-' in 5% HNO was diluted with 1 mol 1-' HNO to provide a working stock solution of 10 pg ml-'. Working standard solutions were prepared daily by further dilution e g . in 0.1 mol I-' acetate buffer for the parameter studies.The 0.1 mol 1-' acetate buffer pH 4 was a 5 + 1 mixture of 0.1 mol 1-' acetic acid (Suprapur Merck Darmstadt Germany) and 0.1 mol I-' sodium acetate solution. For analysis of samples with complex matrices 0.02 moll-' HNO +0.4% L-cysteine (Merck) or 0.02 mol I-' phosphoric acid (Suprapur Merck) + 0.4% L-cysteine was used to reduce interferences. Mixed palladium nitrate-magnesium nitrate solution was used as a chemical modifier and was obtained by dilution of appropriate aliquots of a palladium nitrate solution (Merck) and a solution of Suprapur Mg(N0,),.6H20 (Merck) in de-ionized water.The mass applied on both graphite tubes and platforms was 15 pg of Pd+8 pg of Mg(NO,),. A 0.6-1.0% m/v NaBH solution was prepared by dissolving the appropriate amount of NaBH (Merck) in 0.1% NaOH (Suprapur Merck) and was used without filtration. The solutions used for coating with Zr and W were 0.02 moll-' ZrOC1 and 0.02 mol 1-' Na2W04 respectively and were prepared from their salts (Merck). For Nb- and Ta-coating commercial standard solutions of 10 mg ml-' (in 2% HF; Johnson Matthey) were used; for Ir-coating a 1 mg ml-' standard solution (Johnson Matthey) was used. Sample Preparation Sample solutions were prepared from the geological certified reference materials G-2 (granite) and AGV-1 (andesite) and the NIST Standard Reference Material (SRM) 1646 Estuarine Sediment.Microwave-heated high-pressure dissolution (MLS System Buchi Goppingen Germany) was applied 0.500 g of the sample was weighed into the PTFE liner of the pressure vessel and 2.8 ml of concentrated HNO,+0.5 ml of H202+ 1.7 ml of HF were added. The closed vessels were heated in three steps i.e. 5 min at 500 W; cooling and pressure release addition of 1 ml of concentrated HNO,; 4min at 600 W and 1 min without heating; 3 min at 600 W. The solution was diluted to 25 ml with 0.1 moll-' HNO and filtered after allowing it to stand overnight so that any non-silicate residues could settle. Low-alloy steel reference materials (NIST SRMs 361 and 363) were digested with aqua regia by the following procedure 0.200g of the steel sample was weighed into a 100ml PFA calibrated flask and 5 ml of aqua regia (1.2 ml of concentrated HNO + 3.6 ml of concentrated HC1) were added.The flasks were heated in a water-bath to 90°C for about 3 h. The solutions were diluted to 100 ml with 1 moll-' HNO and filtered after the residues had settled. The final solutions for all determinations were obtained by diluting the above sample solutions in 0.02moll-' HNO3-O.4O/0 L-cysteine or 0.02 moll- ' H,PO,-O.4% L-cyst- eine solution and were then used for the hydride generation step. Coating of the Graphite Tubes and Platforms Coating with Zr Nb Ta or W A simple coating method was proposed by Iwamoto et which did not involve soaking the whole graphite tube but rather injecting a 0.01 moll-' sodium tungstate solution into the graphite tube followed by a complete drying/ashing/ 1070 Journal of Analytical Atomic Spectrometry December 1995 Vol.10atomization cycle for liquid samples. We modified the coating procedure by injecting a 50 pl aliquot of the coating reagent solution (for Zr Nb Ta or W) and lop1 aliquots on to a platform respectively. After drying the residual reagent was run through a heating cycle (pyrolysis at 1200 "C for 20 s; atomization at 2500 "C for 3 s) and the procedure was repeated to build-up a certain mass of coating i.e. 2-3 pmol on graphite tubes and 1-2 pmol on platforms. The metal carbides are assumed to be formed during the heating procedure. Coating with Pd-Ir Ir or Ir-Mg According to the procedure of Shuttler et al.," 50pl of a mixture of 0.05% m/v Pd and 0.05% m/v Ir in 0.1 mol I-' nitric acid were injected into the graphite tube or 10 pl on to the platform slowly dried and heated in a reduction step at 1200°C for 20 s and 2000°C for 3 s.The complete injection and heating cycle was repeated so as to obtain a total mass of 50 pg of Pd and 50 pg of Ir on the graphite tube and 20 pg of Pd and 20 pg of Ir on the platform. The analogous procedure using a 0.1 YO m/v Ir solution or 0.1 YO Ir + 0.02% Mg afforded 100 pg of Ir or 100 pg of Ir+20 pg of Mg on the graphite tube and 40 pg of Ir or 40 pg of Ir + 8 pg of Mg on the platform. Hydride Generation Hydride generation was accomplished using two different generators the VGA-76. continuous unit and the FIAS-400 flow injection system. Using the Varian VGA-76 generator a PTFE sample injection valve (six-port with sample loop; Latek Heidelberg Germany) was introduced into the sample channel of the peristaltic pump into which a carrier solution of the required acid concentration (e.g.the acetate buffer solution) was pumped. The calibrated sample loop of 1 ml was filled by a second pump. As is shown in the flow diagram in Fig. 2 the acidic carrier/sample stream merges with the NaBH flow then with an Ar carrier gas flow to strip the liberated hydrides from the reaction mixture as it is flowing through the reaction coil to the gas-liquid separator. A second Ar gas flow was introduced into the separator to avoid condensation in the Sample -1 n n Carrier NaBH4 W -1- - ETAAS II Ar Fig. 2 Basic flow diagram for the sample injection-hydride generator system used for GeH generation and in situ trapping in the graphite furnace Table 1 Ge hydride generation conditions transfer tubing.23 All tubing downstream of the peristaltic pumps was of PTFE.For studying the signal stability on the coated tubes or platforms (over 400 cycles) the VGA-76 generator was run in a continuous mode with the Ge standard added to the carrier solution. In this instance the hydride was introduced into the furnace and trapped for 30 s. In a later stage of this study (during analysis of samples) hydride generation was accomplished using the Perkin-Elmer FIAS-400 flow injection system and AS-90 autosampler which has been described by Welz and Sch~bert-Jacobs,~~ and was controlled by a separate computer. The standard and sample solutions were injected from the automatically filled sample loop (1 ml) into the carrier solution stream at selected intervals. With this unit condensation occurred in the transfer tubing connected to the gas-liquid separator after about 30-50 runs.The hydrides and reaction gases were transferred into the graphite furnace and the hydrides were thermally decomposed on the pre-heated coated graphite surface where the analyte element was collected in situ and concentrated before the atomization step. The tip of a piece of quartz capillary of about 1 mm id (1.8 mm od) which was connected via a suitable length of PTFE tubing (2.0 mm id) to the outlet of the gas- liquid separator was inserted through the sample hole of the graphite tube and was held in close contact with the opposite wall or platform surface respectively. The end of the quartz tube had two bevels to both sides of the tip to provide an unobstructed outlet for the gases when placed in the graphite tube.At trapping temperatures >700"C a very small H2 flame was observed at the hot tip of the quartz capillary when it was lifted out of the graphite tube which extinguished after a few seconds. The advantage of a closed low-volume capillary system and gas-liquid separator ( 1 ml) compared with some batch hydride designs" is that entrained air is flushed out within a few seconds even during start-up. Progress with respect to practical use was the fixation of the platform within the graphite tube in the form of the 'forked platform' or the similar 'integrated platform' now available.A manually introduced quartz capillary used for some of the initial measurements could be simply rested on the platform in its fixed position in the same way as on the tube wall. An important step towards automation was the hydride transfer into the graphite tube uia the quartz capillary mounted in place of the PTFE capillary tubing of the sample dispenser arm. Changes in the AAS software allowed controlled hot-injection of the gaseous hydrides. The Ge hydride generation parameters were adjusted to a flow rate of the carrier solution of about 3-4mlmin-' and the NaBH concentration was varied between 0.5 and 1% at a flow rate of 1 mlmin-' for the VGA-76 unit and up to 2.5 ml min-' for the FIAS-400 unit. The aim was to achieve a low hydrogen production rate so as to increase the residence time of the hydrides in the furnace and a small back-pressure in the gas-liquid separator.The routinely selected conditions are summarized in Table 1. Solution For parameter studies (VGA-76) Carrier Reductant Total Ar gas flow Carrier For analysis of sediment and geological samples (FIAS-400) Reductant Ar gas flow Concentration Flow rate/ml min - ' 0.1 moll-' acetate buffer (pH 4) 0.6% NaBH in 0.1% NaOH 3 1 90 0.1 mol 1-' HN0,-0.4% L-cysteine or 0.2 mol 1-' 4 2.5 H,PO,-0.4% L-cysteine 1 Yo NaBH in 0.1 YO NaOH 50 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1071Procedure for in situ Trapping Before the first hydride generation cycle the graphite tube or platform was coated with the selected adsorber for hydride trapping as described above.The sequence of operations describing the hydride generation collection and atomization is as follows. The sample loop was filled with the sample (or standard) solution after which the hydride generator was started according to the conditions given in Table 1 as was the automatic furnace cycle with the temperature programme given in Table 2. After the graphite tube reached the selected trapping temperature (e.g. 600 "C) the quartz capillary was inserted automatically and the sample injection valve was then switched for the sample volume being carried into the generator by the acidic carrier solution. After throughput of the sample volume the reaction of the acidic carrier and reductant flow was continued for an additional 30 s to ensure the complete flushing of the reaction gases into the graphite furnace.The quartz capillary was removed automatically from the furnace at the end of the collection time and the furnace programme was continued with an Ar flush of the graphite tube followed by the atomization step (with gas stop) and a final cleaning step. a 0.2 c 5 2 0.1 a RESULTS AND DISCUSSION Use of Pd-treated Graphite Surfaces for in situ Trapping of the H ydrides In the first series of experiments the hydride generation and trapping cycle was started with injection of Pd-magnesium nitrate solution into the graphite tube or on to the platform which was then dried and thermally pre-treated at the hydride collection temperature. This application of Pd was performed before each hydride trapping/atomization cycle because the Pd was evaporated at the high atomization and cleaning temperatures of each cycle.In this work the trapping of Ge as its hydride on the Pd-Mg pre-treated graphite tubes or platforms is considered as the reference case when compared with trapping on other reagent-coated graphite tubes or platforms. An efficient collec- tion of Ge on the Pd-Mg pre-treated surface of the pyrolytic graphite tube or platform was obtained over a wide range of trapping temperatures from below 200°C to higher than 800°C. More detailed results on the Pd-Mg applications will be discussed in the context of the other investigations. - - Test of Long-term Stability of Zr- Nb- Ta- W- Ir- Ir-Mg- and Pd-Ir-coated Graphite Tubes and Platforms for in situ Trapping of Ge Hydride Pre-treatment of the graphite surface with metals that form thermally stable carbides has been proposed to overcome difficulties in the determination of Ge because of losses caused by the formation of volatile compounds (with 0 S Cl etc.) during thermal pre-treatment steps.Coating with elements such as Zr ( platforms") W ( t ~ b e s ; ~ ~ ~ ~ ~ plat- f o r m ~ ~ ~ ) Mo ( t ~ b e s ; ~ ' ~ ~ platforms33) and Ta34-36 improved the sensitivity and precision for Ge in aqueous and organic extractant solutions. Ma et aL6 reported the trapping of germane using La- Ta- W- and Zr-coated normal graphite tubes compared with pyrolytic carbon coated (pyrC) graphite tubes. These workers used Pd modifiers on the coated surfaces and obtained a good response on Ta- W- and Zr-coated tubes but even higher sensitivity on the pyrC tubes.Trapping temperature curves In a second series of experiments the trapping of the Ge hydride was studied on new pyrolytic graphite tubes as well as on platforms that had only been coated once with Zr Nb Ta W Ir Ir-Mg or Pd-Ir according to the coating procedures described above. The influence of the deposition temperatures for in situ collection of germane on the coated tube wall is shown in Fig. 3 and for sequestering on the coated platforms in Fig. 4. Each set of trapping temperature curves is compared with the trapping achieved with a pre-injected Pd-Mg modifier which was applied before each trapping cycle. The Pd modifier treated surfaces obviously resulted in the highest response for trapping of germane. Trapping and atomization from the coated tube walls (Fig.3) showed relatively good efficiency for Zr between 550 and 750 "C (atomization temperature 2500-2600 "C); the signals decreased below 550°C. With Nb- Ta- and W-coated tubes only a very low response was observed. Typical signals are shown in Fig. 5. At trapping temperatures higher than 800"C the signal stability deteriorated. Trapping temperature curves on coated platforms are sum- marized in Fig. 4; the absorbance signals from the Zr-coated platform reached a maximum at trapping temperatures between 700 and 800°C and the efficiency decreased rapidly 0.3 I 10.9 0.6 a V z 3 0.3 2 0 200 300 400 500 600 700 800 900 Trapping temperaturePC Fig. 3 Trapping temperature curves for Ge hydride (2 ng Ge) on graphite tubes coated with 3 Ir-Mg; 4 Ir; 5 Nb; 6 Zr; 7 W; compared with graphite tubes with 2 pre-injected Pd-Mg modifier or 1 without any coating or modifier.Peak height mode Table 2 Graphite furnace parameters for Ge hydride trapping and atomization on the coated surfaces used in subsequent parts of this work Temperature/"C Ir- (or Pd-Ir-) coating Zr- (or Nb- Ta- W-) coating Step tube/platform tube/platform Ramp/s Hold time/s Ar sheath gas/l min-' 1 Hydride introduction 400/500 600/700 5 50* 31- 2 Ar flush 400/500 600/700 10 3 3 Gas stop 400/500 600/700 2 0 4 Atomization 2200/2400 2 5OO/2500 1 2-4 0 5 Cleaning 2200/2400 2550/2600 - - - 2 3 * Depending on the volume of the sample loop (e.g. 1 ml). 7 Proprietory adjusted for hot-inject mode; total gas flow through the furnace.1072 Journal of Analytical Atomic Spectrometry December 1995 Vol. 100.20 8 0.16 0 m e 0.12 5 n 2 0.08 2 CJ c 0.04 - 0 /x 3 1 I I I 4 * 100 200 300 400 500 600 700 800 900 Trapping temperature/"C Fig. 4 Trapping temperature curves for Ge hydride (2 ng Ge) on pyrolytic graphite platforms coated with 3 Ir-Mg; 4 Ir; 5 Zr; 6 W compared with 2 pyrolytic graphite platforms with pre-injected Pd-Mg modifier; or 1 without any coating or modifier. Peak area mode 0 1 Time/s 2 Fig. 5 Signal graphs obtained after in situ trapping of Ge hydride (2 ng Ge) and atomization on graphite tubes coated with 1 Zr (6OO/26OO "C trapping/atomization temperatures) and 2 Nb (7OO/25OO "C) 0 1 2 Timels 3 Fig. 6 Signals obtained after in situ trapping of Ge hydride (2 ng Ge) and atomization on platforms coated with 1 Zr (700/26oO "C trapping/ atomization temperatures) and 2 W (600/2500 "C) below 600°C.The Ge signals (Fig. 6) from the Zr platform showed a small shoulder. Less effective GeH trapping on W- Nb- and Ta-coated platforms was observed in the range between 600 and 750"C with a double peak for the Ta-coated platform. The temperature curves for graphite tubes with Ir Ir-Mg or Pd-Ir coatings show that Ge trapping occurs at lower temperatures compared with the Zr coating. However the signals (Fig. 7) were small and showed considerable tailing and the absorbance decreased after a relatively small number 0.2 Q C e 5 s 0 0.9 Q C e 5 2 0 0 1 2 3 Time/s Fig. 7 Signals obtained after in situ trapping of Ge hydride (2 ng Ge) and atomization on graphite tubes coated with 1 Pd-Mg (500/2600 "C trapping/atomization temperatures) and 2 Ir (5OO/22OO "C) of firings eg.during recording of the trapping temperature curve (even at atomization temperatures below the limiting value of 2400 "C). Very small Ge signals with considerable tailing were observed for the trapping of germane on Ir- Ir-Mg- and Pd-Ir-coated platforms. The signals were again not stable for more than a small number of firings. The carbide-forming elements were found to shift the Ge absorption peaks towards shorter atomization times where- as the noble metal coatings shifted the signals to longer atomization times with the longest delay by the Pd modifier. Hydride generation and in situ trapping of Ge hydride was most effective with Zr-coated tubes or platforms with respect to fully automatic sample processing.This coating was investi- gated further for long-term stability and suitability for analysis of samples. Signal stability on Zr-coated surfaces After only a single coating with the trapping reagent the reproducibility of the absorbance and integrated absorbance signals of Ge was tested in 400 complete in situ trapping/ atomization cycles on the coated graphite tubes and platforms the hydride being generated in a continuous mode (from a 2 ng ml-' solution) and trapped for 30 s (equivalent to a mass of 2.8 ng of Ge). In Fig. 8 the stability of the Ge absorbance signal and the precision is shown for hydride trapping on a Zr-coated tube (3 pmol Zr; 750/2600 "C trapping/atomization temperature).The curves indicate that the stability of this coating as a trapping reagent was sufficient for at least 400 complete hydride trapping and atomization cycles. The precision of the absorbance signals (n = lo) showing some variations is better than 3%. 0.20 0.15 Q 0 2 0.10 5 n 0.05 0 -12 ~ 19- h I E 100 200 300 400 No. of firings Fig. 8 Signal repeatability and precision of 400 complete cycles of Ge hydride (2.8 ng Ge) in situ trapping and atomization on graphite tubes coated with 3 pmol Zr as trapping reagent (750/26oO "C trapping/ atomization temperature) Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1073The stability of the Ge signal was also followed for the in situ trapping of germane on a Zr-coated platform over 400 firings and a precision of better than 4% was obtained for the integrated absorbance signals.According to our experience during the subsequent analysis of samples some changes in sensitivity (of 10-20%) occurred with some Zr-coated plat- forms between about 200 and 300 firings and a more frequent reslope of the calibration graph beyond 180 firings is therefore recommended. Calibration graphs and carry-over efSect Calibration graphs were established for in situ trapping of GeH on Zr-coated graphite tubes and platforms. There was only a very small adsorptive ‘carry-over effect’ for Ge hydride collection compared with the ‘carry-over effect’ that was observed at higher trapping temperatures in our with Sn hydride trapping. By ‘carry-over effect’ it is understood that a very small fraction of the Ge is deposited at the tip of the quartz capillary and that this small fraction is carried over to the next measurement of Ge sample or blank.The ‘carry-over effect’ for Ge trapping is very small or negligible at trapping temperatures of 600-650 “C and increases at higher tempera- tures. Thus Ge trapping should not exceed the ‘critical tem- p e r a t ~ r e ’ ~ ~ of 650-680 “C. The calibration graphs for Ge were linear up to about 4 ng of Ge on both the Zr-coated tubes (peak height) and Zr-coated platforms (peak area). Analytical Figures of Merit Trapping efJiciency The efficiency of the in situ trapping on the coated surfaces was compared with that obtained by direct injection of an aqueous Ge standard solution on the same coated surface. Trapping efficiencies for germane (based on peak area) on the Pd-Mg (modifier) treated tube and platform of about 70 and 87% respectively were observed. On the Zr-coated surfaces the trapping efficiencies were 61 and 60% on the tube and platform respectively.Characteristic mass The characteristic mass m as determined from the slopes of the lower linear part of the calibration graphs is summarized in Table 3. A comparison of the sensitivity suggests that the atomization of Ge from the Zr-coated tubes or platforms is considerably lower than from the Pd modifier treated surfaces. A characteristic mass (based on peak height at A = 0.0044) of 12 pg of Ge using a Pd modifier on graphite tubes was reported by Doidge et aL3 and of 16 pg of Ge using a Pd-Mg modifier in our previous work.’ Using platforms Tao and Fang’ found a characteristic mass (at Aint = 0.0044 s) of 49 pg of Ge with a Pd modifier while a value of 38 pg of Ge was reported in our previous paper’ with a Pd-Mg modifier.Table3 Characteristic mass m0 for Ge hydride trapping on Zr-coated graphite tubes and platforms Graphite tube Zr-coated Pd-Mg modifier Pd modifier Zr-coated Pd-Mg modifier Pd modifier Platform m0 m o (peak height)/pg (peak area)/pg Ref. 54 170 This work 16 55 2 3 12 - 32 110 This work 10 38 2 49 5 - However in terms of the automated in situ collection of germane in electrothermal AAS the Zr-coated tubes or plat- forms can both be recommended because of the good long- term trapping stability of the single coating which is effective for > 400 firings. Detection limits and precision Calculated detection limits for in situ trapping of Ge hydride generated from acetate buffer solution based on the relative standard deviation of the reagent blank (3s) and a sample volume of 1 ml were 9 pg of Ge on a Zr-coated tube (peak height) and 18 pg of Ge on a Zr-coated platform (peak area).The precision of the determination using a Zr-coated graphite tube is better than 2% for determinations at the 1 pg I-’ level. The reagent blank was very small and is estimated to contain 0.1 ng of Ge using a 1 ml sample volume. Determination of Ge in Geological and Sediment Samples There is a lack of reference materials with a certified Ge content. The geological reference standards G-2 (granite) and AGV-1 (andesite) as well as the NIST SRM 1646 Estuarine Sediment were analysed for their Ge content using both the 0.02 mol 1-1 HNO3-0.4% L-cysteine and the 0.02 mol 1-1 H3P04-0.4% L-cysteine solutions as reaction medium for the same sample solution.The results of the sample analyses using standard additions are summarized in Table 4. It is interesting that the Ge concentration values obtained for the geological and sediment samples using the 0.02 moll-’ HNO,-0.4% L- cysteine solution were higher than with the 0.02 mol I-’ H3P04-0.4% L-cysteine solution as reaction medium. Com- paring standard additions with the slopes of the calibration graphs there was no significant interference from the sample matrix and a Ge spike added before the dissolution procedure was recovered quantitatively.The difference between the two media was an additive error which could not be corrected by standard additions. The Ge concentration values obtained for the AGV-1 and G-2 standards are lower than the ‘recommended concentration’ values in the compilation of geostandards by Gladney et However our values show better agreement with the ‘method mean’ values measured by atomic absorption listed in the same compilati~n.~~ Determination of Ge in Steel Samples Steel samples in the form of the NIST SRMs 361 and 363 were digested with aqua regia. After dilution the concentration of Ge in the final solutions was in the range 0.8-1.2 ng ml-’ which resulted in a peak absorbance A of about 0.2. The measured Ge concentration values are also listed in Table4.The Ge values from the hydride generation from 0.02 moll- HN03-0.4% L-cysteine and 0.02 moll-’ H,PO,-0.4% L-cyst- eine media show satisfactory precision but are low compared with the non-certified values of Ge in the steel SRMs. The Ge content found for SRM 361 was confirmed by an indepen- dent determination using inductively coupled plasma mass spectrometry. Further investigation of the factors influencing the Ge hydride generation process is required and additional optimiz- ation for complex matrices such as minerals and alloys is necessary. CONCLUSION A single addition of Zr solution as trapping reagent and thermal pre-treatment of the carbide-forming element on the pyrolytic graphite surface allows the efficient in situ trapping 1074 Journal of Analytical Atomic Spectrometry December 1995 Vol.10Table 4 Determination of Ge in geological sediment and low alloy steel certified reference materials by standard additions using flow injection hydride generation in situ collection on Zr-coated graphite tubes. (All values in pg g-’). Values in parentheses indicate number of determinations Material Geological samples AGV- 1 via peak height via peak area via peak height via peak area NIST SRM 1646 via peak height via peak area NIST SRM 3619 via peak height via peak area NIST SRM 363 via peak height via peak area G-2 Sediment sample Low-alloy steels Found Hydride generation using 0.02 moll-’ HNO + 0.4% L-cysteine 1.23 & 0.02 (4) 1.17f0.03 (4) 1.01 kO.01 (4) 1.01 kO.01 (4) 1.9f0.2 (2) 1.9f0.1 (2) 35k2 (6) 34k1 (6) 26k1 (2) 25$1 (2) 0.02 moll - ’ H3P04 t 0.4% L-cysteine 1.07rt0.01 (2) 1.03 & 0.01 (2) 0.97 f 0.01 (4) 0.98 f 0.01 (4) 1.6k0.1 (2) 1.6k0.2 (2) 35f1 (2) 3 5 f l (2) 3 1 f l (2) 2 8 f l (2) Recommended concentration3’ 1.25f0.13* 1.14 k 0.15* AA ‘mean valuey3’ 1.06t 0.95-t * ‘Recommended concentration’ values in ref.38. t Atomic absorption ‘method mean’ values in ref. 38. $ ‘Non-certified‘ concentration value by NIST. 9 A value of 36 kg g-’ of Ge was obtained for this SRM by ICP-MS. of Ge hydride (with subsequent atomization) for more than 400 complete trapping/atomization cycles. Instrument and software requirements for automating the hydride generation and trapping of the gaseous hydrides in the graphite furnace are simplified by the application of such a long-lasting trapping reagent for coating of a new graphite surface and repetitions of such coatings during the lifetime of the graphite tube or platform do not seem to be necessary.Zr-coated graphite tubes as well as Zr-coated platforms can be recommended for the effective in situ trapping and concen- tration of Ge hydride because of the reasonably high sensitivity and good long-term stability of the absorbance signals. The trapping of germane on surfaces coated with other carbide- forming elements or with noble metals (Ir and Pd-Ir) was very poor and the Ge signals decreased after a relatively small number of firings. The coating from a carbide-forming element allows higher atomization and clean-out temperatures to be used. The technique of Ge hydride generation and trapping in the graphite furnace provides very low limits of detection of Ge because reagent blanks are usually very low for this element.A wide range of concentrations can be analysed by employ- ing different volumes of the sample loop (and different Ge wavelengths). Further investigation of the factors influencing the hydride generation process which was beyond the scope of this work would be required to enable Ge hydride generation to be used for samples with complex matrices such as minerals and alloys. The authors thank Varian Techtron Australia for providing the software modifications for sample introduction at relatively high temperatures. REFERENCES 1 2 3 Yan X.-p. and Ni Z.-m. Anal. Chim. Acta 1994 291 89. Haug H. O. and Ju C. J. Anal. At. Spectrom. 1990 5 215. Doidge P.S. Sturman B. T. and Rettberg T. M. J. Anal. At. Spectrom. 1989 4 251. 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Zhang L. Ni Z.-m. and Shan X.-q. Spectrochim. Acta Part B 1989 44 751. Tao G. and Fang Z. J. Anal. At. Spectrom. 1993 8 577. Ma Y. Han Y. and Ariguli R. Fenxi Huaxue 1994 22 586. Li S. Wang R. Ni Z.-m. Zhang L. Yan X. He B. Shan X. Jing S. and Han H. Guangpuxue Yu Guangpu Fenxi 1992 12 (4) 117. Sturgeon R. E. Willie S. N. Sproule G. I. and Berman P. T. Spectrochim. Acta Part B 1989 44 667. Zhang L. Ni Z.-m. and Shan X.-q. Spectrochim. Acta Part B 1989 44 339. Sturgeon R. E. Willie S. N. and Berman S. S. Anal. Chem. 1987 59 2441. Shuttler I. L. Feuerstein M. and Schlemmer G. J. Anal. At. Spectrom. 1992 7 1299.Schlemmer G. and Feuerstein M. paper presented at the Colloquium Analytische Atomspectroskopie CANAS’95 Konstanz Germany April 1995. Ni Z.-m. Hang H.-b. Li A. He B. and Xu F.-z. J. Anal. At. Spectrom. 1991 6 385. Yan X.-p. and Ni Z.-m. J . Anal. At. Spectrom. 1991 6 483. Castillo J. R. Lanaja J. and Aznarez J. Analyst 1982 107 89. Halicz L. Analyst 1985 110 943. Jin K. Terada H. and Taga M. Bull. Chem. SOC. Jpn. 1981 54 2934. Nakahara T. and Wasa T. Microchem. J. 1994 49 202. Brindle I. D. Le X.-c. and Li X.-f. J . Anal. At. Spectrom. 1989 4 227. Welz B. Atomic Absorption Spectrometry VCH Verlag Weinheim 2nd edn. 1985 pp. 284 and 285. Varian Manufacturer’s Literature Varian Australia 1994. Wamoto E. Shimazu H. Yokota K. and Kumamaru T. J. Anal. At. Spectrom. 1992 7 421. Sturman B. T. Appl. Spectrosc. 1985 39 48. Welz B. and Schubert-Jacobs M. At. Spectrosc. 1991 12 91. Zheng Y. and Zhang D. Anal. Chem. 1992 64 1656. Kolb A. Mueller-Vogt G. and Stoebel W. Spectrochim. Acta Part B 1987 42 951. Gao Y.-q. and Ni Z.-m. Huaxue Xuebao 1982,40 1021. Gao Y.-q. and Ni Z.-m. Xiyou Jinshu 1982 1 57. Zheng Y. and Zhang D. Fenxi Ceshi Tongbao 1988 7(6) 41. Benzo Z. Cecarelli C. Carrion N. Alvarez M. A. Rojas C. and Rosso M. J. Anal. At. Spectrom. 1992 7 1273. Bao C. Cheng X. Li Y. Wei Y. and An Z. Guangpuyue Yu Guangpu Fenxi 1991 11(5) 56. Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 107532 Bao C. Cheng X. Liu C. and Wei Y. Fenxi Huaxue 1992 20 429. in the press. 33 Criaud A. and Fouillac C. Anal. Chim. Acta 1985 167 257. 34 Mino Y. Shimomura S. and Ota N. Anal. Chim. Acta 1979 37 Haug H. O. and Liao Y. Spectrochim. Acta Part B 1995 50 38 Gladney I!. S. Jones E. A. Nickell E. J. Roelandts I. Geostand. Newsl. 1992 16 111. 107 253. 35 Hoquellet P. and Lebeyrie N. Analusis 1975 3 505. 36 Sohrin Y. Isshiki K. Kuwamoto T. and Nakayama E. Talanta Paper 5104463 D Received July 10 1995 1987 34 341. Accepted August 22 1995 1076 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10

ISSN:0267-9477    

DOI:10.1039/JA9951001069

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

T r Automated Sampling System for the Direct Determination of Trace Amounts of Heavy Metals in Gaseous Hydrogen Chloride by Atomic Absorption Spectrometry BERND BAASKE AND URSULA TELGHEDER* Department of Instrumental Analytical Chemistry University of Duisburg Lotharstr. 1 4705 7 Duisburg Germany An analytical procedure is described for the determination of iron nickel chromium and manganese in gaseous hydrogen chloride by means of a modified atomic absorption spectrometer. An automated sampling system allows a 'quasi' on-line monitoring of the investigated gas. The introduction of the gas was regulated by a magnetic valve which is connected to the autoprobe of the spectrometer and to a control unit. Typical parameters e.g. the temperature programme and the gas injection volume were optimized.Different types of calibration were carried out using element standard solutions standard additions of gaseous standards and standard solutions as well as gaseous standards. The detection limit for the described procedure was 39 pg for iron 8.7 pg for nickel 1.1 pg for chromium and 0.4 pg for manganese (temperature = 25 "C) using the standard calibration with element standard solutions. The reproducibility of the absorbance signals for the elements in gaseous hydrogen chloride varies between 5.3 and 13% (n= 10). Keywords Electrothermal atomic absorption spectrometry; analysis of reactive gases; automated gas sampling system; determination of heavy metals The analysis of reactive gases is of increasing importance particularly because of their use in the production of semiconductors.The types of metallic impurities in gases have also not been clarified. Reactive gases are generally contained under pressure in steel cylinders. The cylinder and the cylinder valve must be considered together with the sample as it is not possible to dispose of the packaging. The pressure of the gas must be reduced to atmospheric pressure for the purpose of analysis. The sources of contamination could be the valve system the corrosive gas itself or both. The analysis of the metallic contaminants of gases must not only take into account particle- bound impurities but also components in liquid and gaseous states. Flaherty et al.' have published results from the examin- ation of the dynamics of particles and metal contamination from gas cylinders regulators and valves.They found that for corrosive gases metallic impurities exist in the form of univer- sally distributed particles and that the cylinder valve is a significant source of metal contamination. Hence different methods exist for the investigation of the contaminants; for example indirect methods such as filtration and hydrolysis. It is possible to differentiate between particle-bound impurities and components in liquid and gaseous states by using these two methods. Faix et aL2 described two analytical procedures for the determination of particle-bound trace metals in high- purity hydrogen chloride. They used polycarbonate filters with * To whom correspondence should be addressed. Journal of Analytical Atomic Spectrometry a pore diameter of 0.05 pm for sampling.The subsequent determination of Br Mg Mn Na Sb Sn Te Ti and Zn was carried out by instrumental neutron activation analysis (NAA) and the determination of Cr Cu Fe Mn and Ni by electrother- mal atomic absorption spectrometry (ETAAS). Denyszyn et al. accomplished the sampling of trace metal impurities in gaseous hydrogen chloride in two steps ( 1 ) particulate material was collected on a membrane filter and (2) aerosol or gaseous material was collected in a Greenburg-Smith impinger. The determination of Fe Cr Ni and Cu was performed by ETAAS. The concentrations of the metal impurities measured in the gas-phase filters were in most instances 25-50% higher than the background measurements of the filters. The impinger data were higher than the filter results. Cui et aL4 have investigated trace amounts of Al Ca Cr Cu Fe U Mn Na Pb and Zn in high-purity gases such as N H2 HCl NH3 B2Hs SiH ASH and PH,.They discussed different sampling methods including the conditions for the absorption of the gas in solution and for matrix separation. Subsequently ETAAS methods were established. The determination of ASH in PH has been investigated by Scharf et aL5 The gas was absorbed in impinger gas-bottles filled with HNO and the solutions were analysed by hydride generation AAS with NaBH as the reductive reagent. Another possibility of sampling is the absorption of the trace metals on a solid material.6 Relatively few direct methods have been published for the determination of metallic contaminants particularly in reactive gases.Hutton et ~ 1 . ~ investigated the direct determination of As in SiH4 by inductively coupled plasma mass spectrometry (ICP-MS). They used an alloy sample cone which operated at a higher temperature than the commercially available nickel cones. The addition of hydrogen to the carrier gas further increased the analyte signal level. Schram' described a bypass-backflush balancing system for the direct introduction of gaseous hydro- gen chloride connected to an inductively coupled plasma atomic emission spectrometry (ICP-AES) instrument. A widely applied technique is the use of a sealed ICP (SICP) for the analysis of SiH4,11 HC112 and chlorine13 as described by Barnes and co-workers. The SICP provides complete sample containment. Only a small volume of the toxic gas needs to be used.The plasma is sustained with the discharge gas and sample enclosed together inside a quartz container. Initial investigations into the development of a microwave- induced plasma atomic emission spectrometry (MIP-AES) system to determine trace metals in gases have been de~cribed.'~ The atomic emission spectra were generated by using a medium power microwave-induced plasma with a rectangular-type cavity. A continuously running sampling system was developed for the direct determination of the iron concentration in gaseous hydrogen chloride. An analytical procedure was described for the determination of iron in gaseous hydrogen Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1077chloride by means of a modified atomic absorption spec- trometer.15 The gas inlet was realized by a special graphite capillary.Typical parameters were optimized and the cali- bration was carried out using iron standard solutions. The gas injection was carried out manually with a gas-tight syringe. Based on this fundamental knowledge the automation of this modified atomic absorption spectrometer and further improvements to the procedure are described here. EXPERIMENTAL Sampling The sampling procedure is important and difficult for the continuous analysis of reactive gases. The material that will be in contact with the sample gas must be inert and the transport without any variations. In principle a gas introduc- tion system using a peristaltic pump15 is suitable for this purpose because it is possible to regulate the gas flow after opening the hydrogen chloride cylinder during a measurement without moving any valves. On the other hand the pulsation of the pump does not allow a constant gas flow.Hence the calculation of the amount of gas during a monitoring process is very difficult if not impossible. For further investigations a system with a reducing valve made of stainless steel suitable for hydrogen chloride (Fig. l) was used. After passing the reducing valve (4) the gas flow is divided into two parts one leading to a mixing chamber ( 6 ) where the sample gas could be mixed with argon (2) or any calibration gas (3) the other to a scrubber (10). The sample gas flow is regulated by a flow meter ( 5 ) to ensure a constant analyte flow. The gas introduction is regulated by a magnetic valve (8).The flow meter and all tubes which come into contact with the reactive gases are made of non-metallic materials such as poly (propylene) poly (tetrafluoroethylene) (PTFE) or glass. Trace amounts of water vaporized from the scrubber that back-diffuse into the system are adsorbed by tubes filled with silica gel (7). 9 " I 7 Fig. 1 Gas introduction system 1 HCl; 2 Ar; 3 Fe(CO) in Ar; 4 reducing valve; 5 flow meter; 6 mixing chamber; 7 tube filled with silica gel; 8 magnetic valve; 9 washing bottle; 10 scrubber; and 11 Solaar 939 AA spectrometer with graphite furnace (Unicam) For the analysis of corrosive gases on atomic absorption spectrometer (Unicam) with a graphite furnace and with an autoprobe unit was used.The graphite tubes used for the autoprobe technique had two holes one for sample introduc- tion and a second wider hole drilled into the tube at 90" to the former hole. A suitable gas inlet was realized by a special quartz capillary connected to a tube made of poly(propy1ene) (Fig. 2). The outer diameter of the quartz capillary is 2 mm and the inner diameter is 1 mm. The length of the capillary is 80mm. The magnetic valve and the driving motor for the autoprobe are connected to a control unit. This control unit is a potentiometer which adjusts the rate of feeding for the capillary. The electric pulse to start the motor is given by the spectrometer programme. During the transport of the capillary to the graphite tube the motor has to operate against a spring which sustains the sledge in position during the injection step.The motor has been switched off by a light barrier before it starts. Subsequently the motor will be activated and the capillary will be led back by the spring. The dwell time of the capillary in the graphite tube and hence the sample introduc- tion can be fixed by the spectrometer programme. In order to ensure that the magnetic valve opens at the moment when the capillary is just at the hole of the graphite tube a period of retardation is set by a second potentiometer. This retardation has to be considered in the spectrometer programme. Reagents Germany.) Argon 4.6 99.996%. (Messer Griesheim Duisburg Hydrogen chloride 2.8 99.8%. (Messer Griesheim.) Hydrogen chloride 5.0 99.999%. (Messer Griesheim.) Iron calibration solution (1.000 g I-') iron(111) nitrate in 0.5 moll-' nitric acid. (Bernd Kraft Duisburg Germany.) horizontal position screw B sledge / / I vertical adjustment depthtadjustment screw A injection hole screw c slot for the probe q- Autoprobe graphite tube c I Fig.2 Autoprobe unit capillary 1078 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10Nickel calibration solution (1.000 g 1-I) nickel(I1) nitrate in 0.5 moll-' nitric acid. (Bernd Kraft.) Copper calibration solution (1.000 g 1- l) copper(r1) nitrate in 0.5 mol 1-' nitric acid. (Bernd Kraft.) Manganese calibration solution ( 1.000 g 1-I) manganese(I1) nitrate in 0.5 moll-' nitric acid. (Bernd Kraft.) Dilutions were carried out in ultrapure water (Milli-Q-Plus Millipore Eschborn Germany) and the resulting solutions were stored in poly( propylene) bottles.Sub-boiled hydrochloric acid 37% pro analysi. (Merck Darmstadt Germany.) Calibration gas iron pentacarbonyl 6.3 ppm (v/v). (Messer Griesheim.) -$ 0.30 2 0.25 -e 0.20 -0 Q 0.15 0.10 0.05 - E - - Instrumentation - - - -. .- .- Flow meterlgas mixer. Suitable for hydrogen chloride (PTFE) range 0-5 ml min-' and 2-26 ml min-' (Novodirect Kehl Germany). Scrubber. Buchi 412 (Biichi Flawil Switzerland). Spectrometer. Solaar 939 AA spectrometer with Solaar GF90 Sampler. Solaar FS 90 (Unicam). electrothermal atomization unit (Unicam Kassel Germany). RESULTS AND DISCUSSION Experimental Parameters The elements iron nickel chromium and manganese in gaseous hydrogen chloride were determined with the automated ETAAS system described above.The parameters for the deter- mination of iron in gaseous hydrogen chloride will be discussed as an example. The measurements were carried out at the resonance wavelength of 248.3 nm with a slit-width of 0.2 nm. Deuterium background correction was used. Atomization could be achieved at a temperature of 2300°C. In contrast to earlier rneasurement~,'~ absorbance on the spectral line for iron was observed if pyrolytic graphite coated graphite tubes were used. It is assumed that traces of iron in the form of iron carbide are deposited on the walls of the graphite tube. Further investigations to explain this phenomenon are required. The injection volume can be controlled by opening the magnetic valve and depends on the opening time of the valve (see Fig.1); for all investigations a volume of 0.43 ml was used. Finally the temperature programme of the furnace which was used for the determination of trace amounts of iron in gaseous hydrogen chloride is shown in Table 1. Calibration of the Analytical System When performing a calibration it is important that the matrix of the sample and standards be as similar as possible. In this instance gaseous hydrogen chloride is not available as a calibration gas. However iron pentacarbonyl diluted in Ar can be used as a gaseous calibration standard. The decomposition of this gaseous standard starts at 50°C. Initially calibration was carried out with dilute hydrochloric acid standard solution^.'^ It was assumed that the iron atoms produced from the solutions and those produced from the gas have the same spatial distribution in the graphite tube and that the temperature is constant during both peaks.Furthermore the matrix of the gaseous hydrogen chloride has to be similar to the matrix of the dilute hydrochloric acid solutions after drying and ashing of the solutions. Additionally it would be sensible to investigate a gaseous standard in order to exclude the influence of the matrix. Gaseous iron penta- carbonyl diluted in Ar was used for calibration because no certified gaseous hydrogen chloride standard is available. In order to compare different calibration methods and to evaluate the best fitted type of calibration (closer to the sample matrix) the following calibration procedures were investigated. (i) Calibration with dilute hydrochloric acid standard solu- tions.For this calibration six standard solutions [sub-boiled hydrochloric acid (1 + 99) with concentrations of iron in the range 0.16-0.8 ng per 20 p13 were analysed. (ii) Standard additions of gaseous iron pentacarbonyl diluted in Ar [ciron = 6.3 ppm (v/v)] and dilute hydrochloric acid standard solutions. For this type of calibration a constant volume of standard solutions with concentrations in the range 0.02-0.8 ng of iron per 20 pl was injected into the graphite tube. After drying (T= 120°C) ashing (T= 1200°C) and cool- ing (T=30O0C) a constant volume of gaseous iron penta- carbonyl (mFe = 0.084 ng) was added. Subsequently the whole sample was atomized at 2300°C and the graphite tube was cleaned at 2500°C.(iii) Calibration with gaseous iron pentacarbonyl diluted in Ar [ciron = 6.3 ppm (v/v)] . The different calibration graphs are shown in Fig. 3. The intensities of the absorbance signals correspond to the peak areas. The average was calculated from six results. The iron concentrations were calculated in terms of absolute mass for the possibility of comparison. Supposing that iron pentacarbonyl reacts as an ideal gas the volume of 1 mol of iron is 22.414 1. An iron concentration of 6.3 ppm (v/v) is equivalent to 6.3 pl of iron per litre of Ar or 2.81 x lo-' mol of iron. Taking into consideration the gram- 0.35 A 1 x / / ,"' A:y = 0.4031~ + 0.0173 B:y = 0.3227~ +0.0258 0 ' C:y =0.3207~ +0.0034 0 d 0 - 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Absolute mass of irordng Fig.3 Calibration graphs by A gaseous Fe(CO) in Ar; B standard additions of gaseous Fe(CO)5 in Ar and dilute HCl standard solutions; and C dilute HC1 standard solutions Table 1 Temperature programme for the determination of trace amounts of iron in gaseous hydrogen chloride Step Temperature Argon gas flow rate/ TemperaturerC Time/s increasePC s - ml min-' Sample introduction 1100 4 1100 Adsorption of trace amounts of iron on the graphite tube 1100 4 1100 Removal of the gaseous hydrogen chloride 1100 7 1100 Atomization 2300 8 TC* Cleaning 2500 5 TC* 200 0 300 0 300 * TC Temperature controlled by an optical sensor. Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1079molecular mass of iron (M = 55.85 g mol-l) the concentration of mass is 0.0157mg of iron per litre of Ar.The absolute masses in Fig. 3 depend on the injection time and the flow of the gas. Fig. 3 shows that the slopes of the calibration graphs obtained with dilute hydrochloric acid standard solutions and by the standard additions of gaseous iron pentacarbonyl-Ar and dilute hydrochloric acid standard solutions are almost identical. The small difference shows that the influence of the matrix is minimal. Thus the method of standard additions as described above is the most suitable type of calibration for the analysis of hydrogen chloride particularly since the result obtained by the extrapolation of the regression line to the abscissa (0.086 ng of iron) corresponds to the calculated mass (0.084ng of iron). Hence the calibration for other elements such as nickel chromium and manganese could be carried out with dilute hydrochloric acid standard solutions or by standard additions. The detection limit was found to be 39pg of iron [standard deviation s = 11.4% (n = 6 ) ] by calibration with standard solutions and 26 pg of iron [ s = 12.9% (n=4)] by calibration with gaseous iron pentacarbonyl-Ar.In both instances the detection limit (DL) was calculated by DL = 3sS-I (1) (s = standard deviation of the blank S = sensitivity of the calibration and 3 =statistical value). Calculation of Element Concentrations in Gaseous Hydrogen Chloride The determination of iron nickel chromium and manganese was carried out by the standard calibration procedure using the parameters described above. All the analytical signals were calculated by means of peak areas.The element concentrations in gaseous hydrogen chloride are shown in Table 2. Investigation of Gaseous Hydrogen Chloride of Different Quality Reactive gases are generally contained under pressure in steel cylinders. The cylinder and the cylinder valve must be con- sidered together with the sample as it is not possible to dispose of the packaging. The pressure of the gas must be reduced to atmospheric pressure for the purpose of analysis. The sources of contamination could be the valve system the corrosive gas itself or both. Hence the analysis of gaseous hydrogen chloride with purities of 99.999 and 99.8% respectively could show whether there is a difference between the content of contami- nants. Fig. 4 shows the dependence of the absorbance signal of iron on the timing of sampling.The sampling of hydrogen chloride was carried out every 70 s after opening the valve. In both instances the absorbance was highest just after opening the hydrogen chloride cylinder and then decreased slowly to a constant value. Thus a steady state was reached about 7 min after opening the gas cylinder. It was found that at this time no significant difference between the iron concentration in gaseous hydrogen chloride with purities of 99.999 and 99.8% respectively could be observed. Hence it seems that the valve system is one of the main sources for the contamination of gaseous hydrogen chloride with iron irrespective of the purity of the gas. CONCLUSION The described ETAAS system with an automated sampling system allows the determination of iron nickel chromium and manganese in gaseous hydrogen chloride.The problem of calibration has been discussed. A comparison of calibration Table 2 Element concentrations in gaseous hydrogen chloride (T= 25 "C) [values in parentheses are relative standard deviations (W) n= 101 ( E integrated absorbance) Element concentration Element Sensitivity1 DL*/pg in gas/pg 1-' E Pg-l Iron 0.0003 39 19.5 (5.3) Nickel 0.0004 8.7 3.4 (13) Chromium 0.0016 1.1 < DL-f Manganese 0.0043 0.4 < DLt * DL = Detection limit. t Below the detection limit. 0.6 1 4 8 12 16 20 24 Tirne/min Fig.4 Dependence of the integrated absorbance on the timing of sampling with standard aqueous solutions as well as standard additions and calibration with gaseous standards has shown that the influence of the matrix is minimal.This fact is interesting for the calibration for other elements for which no calibration gas exists. Nevertheless in those instances calibration using stan- dard solutions is possible. Further investigations with different gases used in the microelectronics industry are planned. This work is part of a JESSI-project and is supported by Messer Griesheim GmbH. The authors are grateful to Dr. Eschwey Messer Griesheim GmbH. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Flaherty 1;. T. John L. and Amato A. F. Solid State Technol. 1992 35 S1. Faix W. G. Schramm W. Vix F. Weichbrodt G. and Henkelmann R. Fresenius' 2. Anal. Chem. 1988 329 847. Denyszyn R. B. Yin I. H. and Bandy T. paper presented at Microcontamination West Conference Anaheim 1989. Cui X. Xu X. Yan X. and Lang W. Bandaoti Xuebao 1989 10( 12) 94s. Scharf H. Hahn E. and Emrich G. 2. Chem. 1990,30(3) 107. Miyazaki K. and Nakagawa K. Koatsu Gasu 1992,29 281. Hutton R. C. Bridenne M. Coffre E. Marot Y. and Simondet F. J. Anal. At. Spectrom. 1990 5 463. Schram J. Fresenius' J. Anal. Chem. 1992 343 727. Jacksier T. and Barnes R. J. Anal. At. Spectrom. 1992 7 839. Jacksier T. and Barnes R. Spectrochim. Acta Part B 1993 48 (6/7) 941. Jahl M. J. and Barnes R. J. Anal. At. Spectrom. 1992 7 833. Jacksier T. and Barnes R. J. Anal. At. Spectrom. 1994 9 1299. Jacksier T. and Barnes R. Spectrochim. Acta Part B 1994 49 (8) 797. Kirschner S. Golloch A. and Telgheder U. J. Anal. At. Spectrom. 1994 9 971. Baaske B.. Golloch A. and Telgheder U. J. Anal. At. Spectrom. 1994 9 867. Paper 51039021 Received June 16 1995 Accepted August 18 1995 1080 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10

ISSN:0267-9477    

DOI:10.1039/JA9951001077

出版商:RSC    

年代:1995

数据来源: RSC

摘要:

CUMULATIVE AUTHOR INDEX JANUARY-DECEMBER 1995 Abell I. D. 591 Aboal-somoza Manuel 227 Adams Freddy C. 11 1 Alexandrova Anka 799 Allen Lloyd A. 267 Alvarado JosC 483 Alvarez Nestor 487 Alvarez Walter Oliver L. 487 Amarasiriwardena Dulasiri 505 Arbore Philippe 381 Arpadjan Sonja 799 Arruda Marco A. Z. 55 501 Asami Naoto 999 Asensio Jesus Sanz 975 Axner Ove 539 Baaske Bernd 1077 Barciela-alonso M. C. 247 Barnes Barbara S. 177 Barnes Ramon M. 505 935 Baroni. U 555 Baxter Douglas C. 711 769 Beatriz de La Calle Guntiiias Becker Johanna Sabine 637 Becker-Ross Helmut 61 127 Beissler Hermann 885 Belazi Abd Ulhafid 233 Bellini Alessandra 433 Bermejo-Barrera Adela 227 Bermejo-Barrera Pilar 227 247 Bernal J. Galban 975 Betti Maria 381 Bettinelli Maurizio 555 Bizzarri. G 555 Boonen Sylvie 81 1047 Bordel-Garcia Nerea 3 11 649 Borodin Alexander V.703 Boulos Maher I. 935 941 Bratter Peter 487 Briggs R. J. 1033 Brockhoff Carol A. 443 Broekaert Jose A. C. 583 849 Brown Garrett N. 527 Brueggemeyer Thomas W. 177 Brunetto M. R. 343 479 Bryant M. F. 295 Bryce D. W. 1059 Bulska Ewa 49 Burden Trevor J. 259 Burguera J. L. 343,473,479 Burguera M. 343,473 479 Bye Ragnar 803 809 Cabon J. Y. 993 Cabrera Horacio P. 511 Caldwell Kathleen L. 367 Camara Carmen 321 815 871 Carrero P. 343 479 Caruso Joseph A. 7 601 853 Castle Laurence 303 Cavalli Paolo 885 Cernohorsky TomaB 155 Cervera M. L. 353 Chakraborty Ruma 353 Chan W. T. 295 Chekalin Nikolai 539 Chen Chih-jung 955 Chen Hengwu 533 941 M. 815 897 145 247 1011 101 1 671 1019 Chen Xiaoshan 837 Cho Jung H. 335 Cho Kyu H.335 Cielo P. 643 Cimadevilla Enrique Alvarez- Cabal 149 Cloud Jacques 287 Coedo Aurora G. 449 859 Corns Warren T 287 Cossa Daniel 287 Costa-Fernandez Jose M. 649 Creed John T. 443 Crews Helen M. 303 625 Crighton J. 591 Cristiano Ana Rita 483 Crowe John B. 177 Curtius Adilson JosC 329 483 Dams Richard 81 569 575 Das Arabinda K. 353 Dauchot J. P. 1039 Davidson Christine M 233 241 Deaker M. 423 de Bikre Paul 395 de Gendt Stefan 681 689 de la Calle Guntinas M. de la Guardia Miguel 353 Di Marco Marco 1003 Diemiaszonek Robert C. 661 Dietze Hans-Joachim 637 897 Doleialovh Katefina 763 Donard 0. F. X. 865 Dorado Teresa 449 859 DUlivo Alessandro 969 1003 Dunemann Lothar 655 Dyakov Alexey O. 703 Dybdahl Bjorn 769 Dyvik Geir 769 Ebdon Les 3 17 Ebihara Mitsuru 25 Efstathiou Constantinos E.221 Ek Paul 121 Enger Jonas 539 Entwistle Andrew 395 Epifanie Arnaud 923 Evans R. Douglas 595 619 Fairman Ben 281 Fang Zhaolun 533 Fang Zheng 359 FariAas Juan C. 511 Fell Gordon S. 215 Feng Liang 875 Fernandez B. 859 Fernandez-Garcia Matilde 671 Ferr6n-novais M. 247 Fisher Andy 519 Florek Stefan 61 127 145 Fodor Peter 609 Fordham Peter J. 303 Fornari Roberto 433 Frech Wolfgang 7 11 769 Fukushima Masami 999 Furuta Naoki 25 Gaillat Ana 935 941 Galbacs GBbor 1047 Gallego Mercedes 55 501 Gallignani M. 343 479 Garcia Alonso J. Ignacio 381 Gijbels Renaat 849 Golloch Alfred 161 Gomes Anne-Marie 923 1019 1047 Beatriz 111 321 Gomez Gomez M. M. 89 Goodall Phillip 3 17 Gramshaw John W. 303 Grazhulene Svetlana S. 161 Greenfield Stanley 183 Greenway Gillian M.929 Gregoire D. Conrad 823 1027 Grotti Marco 325 Guern Y. 993 Guo Gang-ping 753 Guo Xiao-wei 987 Guo Xu-ming 987 GutiCrrez Ana Maria 871 Hahn Lothar 777 Halls David J. 169 Hang W. 689 Haraguchi Kensaku 999 Harnly James M. 187 197 Harrison Iain 215 Harrison W. W. 689 Harville Tina R. 671 Hattingh Cornelius J. 727 Haug Hermann O. 1069 Hayashi Yasuhisa 37,439 He Bin 747 Hecq M. 1039 Heitkemper Douglas T. 177 Held Andrea 849 Hill Steve J. 317 409 519 Hinds Michael W. 527 Hintelmann Holger 619 Hochstrasser Chantal 947 Hohann Volker 677 Houk R. S. 267 837 Hu Bin 455 Huang Meng-fen 31 HuldCn Stig-G&an 121 Hutton R. C. 591 Hutton Robert C. 929 Hwang Chorng-jev 3 1 Imai Shoji 37 439 Imbert Jean-Louis 93 Ingelbrecht Chris 849 Inoue Yoshinori 363 Ito Saburo 999 Ito Tetsumasa 843 Ivaska Ari 121 Izquierdo A.1059 Jackson Kenneth W. 43 Jacquiers-Roux Dimitri 777 Jakubowski Norbert 583 Jantzen Eckard 105 Jedral Wojciech 49 Jiang Shiuh-jen 31 963 Jiang Zucheng 455 Jin Qinhan 875 Jin Qun 875 Johansson Magnus 71 1 Jones Alice V. 785 Jones Phil 281 Jurcek Petr 947 Kabil Mohamed A. 733 Karadjova I. 1065 Kawabata Katsuhiko 363 Keating Gillian E 233 Kerrich Rob 99 Khvostikov Vladimir A. 161 Kim Ha S. 335 Kim Hyo J. 335 Kinard W. F. 295 Kirschner Stefan 161 Knutsen Einar 757 Koch Lothar 381 Kopajtic Zlatan 947 Kotrebai Mihaly 505 Kotrly Stanislav 155 763 Kozma Laszlo 631 Krushevska Antoaneta 505 Lajunen Lauri H. J. 117 Lam Joseph W. H. 551,981 Lampugnani Leonardo 969 Larrea Mada T. 511 Lasztity Alexandra 505 Le Bihan A. 993 Le Cor Yann 721 Ledergerber Guido 947 Lee Gae H.335 Lee Kee B. 335 Le Garrec H. 993 Lerat Yannick 137 Liang Yan-zhong 699 Littlejohn David 215 233 241 Lobinski Ryszard 11 1 Lund Walter 405 803 809 Lunzer Florian 311 1019 Luque de Castro M. D. 1059 L'Vov Boris V. 703 Lyon Ian C. 273 Madon Lydie 923 Madrid Yolanda 321 815 Maher W. 423 Mahmood Tariq M. 43 Mandjukov P. 1065 Manzoori Jamshid L. 881 Mao X. L. 295 Marawi Isam 7 Marcus R. Kenneth 671 Martin Theodore D. 443 Martinez-Soria M. Teresa 975 Martinsen Ivar 757 Marunkov Alexander 539 Masera Eric 137 Massart D. Luc 207 Matusiewicz Henryk 981 Matveev Oleg 885 Mauchien Patrick 137 Mazzucotelli Ambrogio 325 McCartney Martin 233 McCrindle Robert I. 399 McCurdy Ed 303 McLaren James W. 371 551 McLeod C. W. 89 Mester Zoltan 609 Methven Bradley A.J. 551 981 Mile Brynmor 785 Miyazaki Akira 1 Mizuno Seiichiro 415 Moenke-Blankenburg Moens Luc 81 569 575 1047 Molle C. 1039 Monteiro Maria In& C. 329 Montoro Rosa 459 Moreda-Piiieiro Antonio 227 Moreda-Piileiro Jorge 101 1 Muller Hans 777 Muller Victor 681 Muller-Vogt German 777 MuAoz Olivas Riansares 865 Naka Hirohito 823 Nakagawa Kohichi 999 Nakamura Susumu 467 1003 253 Lieselotte 655 101 1 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10 1081Negretti De Bratter Virginia E. Nelms Simon M. 929 Nemet Bela 631 Ni Zhe-ming 493 699 747 Niemax Kay 563 Nishiyama Yasuko 439 Nolte Joachim 655 Novichikhin Alexander V. 703 Ogata Toshio 999 Okuhara Kyoichi 37 Olson Lisa K. 7 Omenetto Nicolo 885 Outridge Peter M. 595 Owen Linda M.W. 625 Paama Lilli 117 Pang Ho-ming 267 Parent Magali 575 Park Yang S. 335 Parry Susan J. 303 Parsons Patrick J. 521 Paschal Daniel C. 367 Pasullean Benyamin 241 Pellegrini Giovanna 969 Penninckx Wim 207 Peramaki Paavo 1 17 Pereiro-Garcia Rosario 31 1 Perera Indral K. 273 Perez Conde M. Concepcion Perez-corona M. Theresa 321 Perkins Charles V. 253 Petrucci Giuseppe A. 885 Petzold G. 371 Piiri Lindy 117 Pin Christian 93 Piperaki Efrosini A. 221 Pitts Les 409 519 Polzik Leonid K. 703 Powell J. J. 259 Praler Frank 677 Prange Andreas 105 Proulx Pierre 935 941 Qiao Huan-cheng 43 Qin Yong-chao 455 Quetel C. R. 865 487 649 671 1019 87 1 Quijano M. Angeles 871 Rademeyer Cornelius J. 399 Radziuk Bernard 127 197 415 Raith A. 591 Rattray Robin 829 1027 1053 Reinicke Albrecht 487 Ren J.M. 1027 Riondato Jorgen 569 Rivas C. 343 479 Robb Paul 625 Roberts David J. 721 Rodel Giinther 127,415 Rollin Stefan 947 Romanova Natalia 739 Rondon C. 343 Rowlands Christopher C. 785 Russo R. E. 295 Sabsabi M. 643 Sadler Daran A. 253 Saito Kengo 37 Salin Eric D. 829 1027 1053 Sanjuan Jane 287 Sanz-Medel Alfredo 149 281 311 649 671 1019 Saprykin Anatolij I. 897 Sarrette Jean-Philippe 923 Saverwijns Steven 575 Schelles Wim 68 1 Schmecher Gisela R. 61 Schnurer-Patschan Christoph Schoppenthau Jorg 655 Schulze Gerhard 487 Schumann Joachim 677 Schumann Thomas 655 Sena Fabrizio 381 Seubert A. 371 Shan Xiao-Quan 791 Shemirani Farzaneh 881 Simeonov V. 1065 Skaugset Nils Petter 739 Skogstad Asbjorn 739 Slavin Walter 521 Smeyers-Verbeke Johanna 207 Smichowski Patricia 815 739 739 563 Smith Clare M.M. 187 Smith Monica M. 349 Sniatecki K. 615 !pezia. S 555 Sramkova Jitka 763 Stenz Herbert 127 Stockwell Peter B 287 317 Stratis J. A. 1065 Sturgeon Ralph E. 981 Stuwer Dietmar 583 Styris David L. 527 Sumi Masao 999 Sun Han-wen 753 Suzuki Yoshihito 363 Taddia Marco 433 Tanaka Toshiyuki 37 439 Tang Shida 521 Tanner Scott D. 905 Tao Hiroaki 1 Taylor Andrew 1033 Taylor P. D. 259 Taylor Philip D. P. 395 849 Tejedor Wedleys 459 Telgheder Ursula 161 1077 Telouk Philippe 93 Thomaidis Nikolaos S. 221 Thomas Christoph 583 Thomas P. 615 Thomassen Yngvar 739 Thompson Diana 303 Thompson K. Cllive 317 Thompson R. P. H. 259 Ting Bill G. 367 Tischendorf Reinhard 61 Tomlinson Medha J. 601 853 Trassy Christian C. 661 Tsakovsky S.1065 Tsalev Dimiter L. 1003 Turner Andrew D. 721 Turner Grenville 273 Uchida Hiroshi 843 Uchino Tomonori 25 Uggerud Hilde 405 Valcarcel Miguel 55 501 Van Grieken RenC 681 689 Van Staden Jacobus F. 727 Van Straaten Mark 849 Vanhaecke Frank 81 569 1047 Vanhoe Hans 575 Vankeerberghen Peter 207 Velez Dinoraz 459 Villeneuve Janice Y. 619 Walder Andrew J. 395 Wang Jian-shang 7 601 853 Wang Yun-zhou 359 Warren Arnold R. 267 Wautelet M. 1039 Wei Wen-ching 955 Weiss Zdengk 891 Wen Bei 791 Wendl Wolfgang 777 Wernli Beat 947 Wetzig Klaus 677 White Mark A. 349 Wibetoe Grethe 757 Wickstr~m Torild 803 809 Willie Scott N. 981 Wilson H. Kerr 349 Woller Agnes 609 Wolnik Karen A. 177 Worsfold Paul 409 519 Wrobel Katarzyna 149 Xie Qian-li 99 Xu Dong-qun 753 Xu Shukun 533 Yamamoto Kouei 415 Yang Hueih-jen 963 Yang Mo-hsiung 955 Yang Peng-yuan 699 Yang Wei-min 493 Ybaiiez Nieves 459 Yiping Liao 1069 Zachariadis G.A. 1065 Zamboni Roberto 969 1003 Zeiher Michael 41 5 Zeng Yun’e 455 Zhang Hanqi 875 Zhang Ke 359 Zhang Yuan-fu 359 Zoorob Grace 853 Zybin Aleksandr 563 1082 Journal of Analytical Atomic Spectrometry December 1995 Vol. 10

ISSN:0267-9477    

DOI:10.1039/JA9951001081

出版商:RSC    

年代:1995

数据来源: RSC