摘要:

JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1209 Characterization of Spent Nuclear Fuel Dissolver Solutions and Dissolution Residues by Inductively Coupled Plasma Mass Spectrometry Jose lgnacio Garcia Alonso Dominique Thoby-Schultzendorff Bruno Giovanonne Jean-Paul Glatz Giorgio Pagliosa and Lothar Koch European Commission JRC lnstitute for Transuranium Nements Postfach 2340 761 25 Karlsruhe Germany The direct analysis of dissolved spent nuclear fuel by inductively coupled plasma mass spectrometry (ICP-MS) allows the elemental and isotopic composition of the irradiated fuel to be determined without any chemical separation. A qualitative evaluation was carried out on the effects of fission P-decay and neutron- capture reactions during fuel irradiation. Semiquantitative analysis of fission products and actinides in the spent-fuel dissolver solution and in the dissolved residues was performed by referring to a response curve.Comparison of the semiquantitative data with computer predictions of fission products inventory showed satisfactory agreement. The analysis of small spent fuel samples by ICP-MS was used to assess the type and irradiation of the fuel in pattern recognition studies. Quantitative analysis of the fuel solutions and residues was performed only for selected elements because of the presence of isobaric interferences. Isotope dilution analysis was applied for polyisotopic elements; standard additions with an internal standard was used for monoisotopic elements. Elements determined in the residues included Zr Mo Tc Ru Rh Pd U and Pu.Neodymium was also determined in dissolver solutions of fast neutron-irradiated fuels and the results were compared with those given by thermal ionization mass spectrometry. Keywords Inductively coupled plasma mass spectrometry; spent nuclear fuels; dissolution residues; fission products; isotope dilution For reprocessing of irradiated nuclear fuels aimed at the recovery of fissile material the Purex (Pu and U recovery by extraction) process is commonly applied.' This process is preceded by dissolution of the fuel in nitric acid where U Pu and most of the fission products are dissolved. After dissolution a residue is formed (metallic and oxidic phases of fission products); its composition depends on the fuel dissolution and reactor parameters. Owing to the recent trend to higher fuel burn-ups studies on the composition and dissolution charac- teristics of high burn-up fuels and their possible influence on fuel reprocessing2p3 regarding Pu recovery are needed.The analysis of fuel residues is routinely carried out by various methods including X-ray microanalysis,2 X-ray diffracti~n,~?~ inductively coupled plasma atomic emission spectrometry ( ICP-AES)3 and thermal ionization mass spectrometry (TIMS).3 Elements found in the residue include the low-mass fission products Zr Mo Tc Ru Rh and Pd trace amounts of U and Pu and natural impurities such as Fe Cr and Ni.3 However the use of optical spectrometric techniques for the analysis of fission products will suffer from systematic errors when natural elements are used for calibration for the measure- ment of the nuclear-produced elements. This is due to the different average relative atomic masses of the natural and reactor-produced elements.As a rule the relative atomic masses for polyisotopic fission elements are between 1 and 2% higher than those for the corresponding natural element owing to the neutron-rich isotopes produced by fission. When several elements are measured in the same sample the systematic errors accumulate and an accurate sample composition cannot be obtained. This is especially important for the characteriz- ation of residues of spent fuel dissolution as the relative atomic masses of fission Zr Mo Ru and Pd are higher than those for the natural elements. Previous report^^-^ do not seem to have taken this effect into account.Inductively coupled plasma mass spectrometry (ICP-MS) offers unique possibilities of application in the characterization of spent nuclear fuels as a result of its high sensitivity and multi-isotopic capabilities.'-'' Samples of relatively low activity as very dilute solutions can be handled in a glove- box and both elemental and isotopic data can be gathered quickly for most of the fission products and actinides. Moreover the reagent blanks are extremely low for fission products and actinides except perhaps for Ba and most of these elements have low ionization potentials which allows the use of semiquantitative approaches for their determination. The presence of the uranium matrix does not cause inter- ferences at the levels used in these studies.' However the determination of all fission products by ICP-MS is hindered by the presence of some isobaric inter- ferences that cannot be corrected for because the isotopic abundances in the sample are in principle unknown.Also the isotopic abundances of the elements in the sample have to be taken into account for quantification. Techniques that can be applied for this purpose include isotope dilution analysis. In this paper the advantages and disadvantages of ICP-MS for the characterization of spent fuels are addressed and some examples of its application to different fuels are presented. Experimental Instrumentation The instrument an Elan 250 from SCIEX (Canada) was modified in order to handle radioactive samples in a glove- box.Details of this modification have been presented pre- viou~ly.~ The operating conditions are summarized in Table 1. The ion-lens settings were optimized using Rh. All measure- ments were made in the scanning mode. Quantitative data were obtained at low resolution 1 point u-' and 1 s integration time. Data were transferred to a PC for computation. Qualitative data were obtained in the high-resolution mode 20 points up'. Dissolution of Spent Fuel Spent fuel pellets (approximately 30 g ) were dissolved in 200 ml of nitric acid of various concentrations (3-7 moll-') and the insoluble residue was filtered dried weighed and then dissolved in concentrated HC1-HNO (9 + 1). Sample dilutions and spikes were made by mass to 0.1 mg either in the glove-box or in the hot cell facility.Samples of 10 ml containing between 0.1 and 0.5 mg of spent fuel per gram of solution were transferred to the ICP-MS system for measurement.1210 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Table 1 Operating conditions R.f. power/W Reflected power/W Argon outer/l min-' Argon intermediate plasma/l min-' Argon aerosol carrier/l min - (psi) Sample uptake/ml min-' Nebulizer type Spray chamber Load coil-sampler cone distance/mm Interface pressure/Pa Quadrupole working pressure/Pa Sampler and skimmer cones Ion lens settings Bessel box B lens/V Bessel box P/V Einzel lens El/V Photon stop S2/V 1400 < 5 12 1.4 0.86 1 Meinhard Scott type/double pass 25 (fixed) 266.64 666.6 x Platinum Optimized on lo3Rh 7.0 - 9.0 - 4.0 - 2.0 Reagents and Materials Natural element stock standard solutions were obtained from Spex as 1000 ppm standards and diluted as necessary with 1% nitric acid.99Tc metal was obtained from Los Alamos National Laboratory (USA) as the pure metal and dissolved in nitric acid to prepare a stock standard solution. Pu reference material (NIST SRM 949f) was used for Pu measurements. Nitric acid was of Suprapur grade from Merck and water purified in a Milli-Q system (Millipore) was used throughout. Inactive standards were prepared by simple dilution in acid- washed calibrated flasks. Radioactive samples and standards were diluted by mass in the glove-box or in the hot cell facility. Polyethylene bottles of 10 or 20ml were used for all radio- active material. Procedures for the Analysis of Fission Products and Actinides Depending on the precision and accuracy required different quantification approaches were followed either simple quali- tative and semiquantitative analysis using the instrument response curve isotope dilution analysis using natural elements as spikes or standard additions with an internal standard for monoiso topic elements.Qualitative and semiquantitative analysis Fission-product isotopes and possible isobaric interferences were identified based on cumulative fission chain yields poss- ible neutron-capture reactions and long-lived P-decay products. Semiquantitative analysis was performed using the response curve of the instrument prepared using natural elements. No interferences from the uranium matrix were observed at the 500 ppm U level,5 which is higher than the concentrations of the spent fuel solutions prepared.Also no ionization correc- tions needed to be taken into account as most of the fission products and actinides have low ionization potentials. The semiquantitative results were compared with computer predic- tions using the KORIGEN code'' and were also used to optimize the isotope dilution and standard additions procedures. Isotope dilution analysis Isotope dilution analysis (IDA) was applied to the determi- nation of Zr Mo Ru Pd and Pu in residues of high burn-up uranium oxide fuel dissolution and Nd in solutions of an experimental fuel containing uranium and neptunium oxides irradiated in a fast neutron reactor. For the analysis of the fission products the natural elements were used as spikes as no enriched isotopes were available at the time of the analysis.This is possible because of the different isotopic abundances of the fission-product elements compared with the natural elements. A general error theory for the determination of non- natural elements by ICP-MS was developed." The equation used for IDA was where Cs and Csp are the concentrations of the element in the sample (S) and in the spike (Sp) respectively M and M s are the masses taken from spike and sample respectively Ars and ArSp are the relative atomic masses of the sample and spike elements respectively ASPa is the isotopic abundance (at.%) of the reference isotope in the spike (isotope a) and A; is the isotopic abundance (at.%) of the reference isotope in the sample (isotope b) RM is the isotopic ratio (isotope b isotope a) in the mixture RSp that in the spike and Rs (isotope a isotope b) that in the sample.All isotopic abundances measured were corrected for mass discrimination err0rs.l' At the counting- rate levels measured (between 10 000 and 100 000 ions SKI) no detector dead-time correction needed to be performed. Standard additions with an internal standard The method of standard additions with an internal standard was applied to the determination of monoisotopic Tc and Rh in spent nuclear fuel solutions and residues and U in residues of spent fuel dissolution. The sample is spiked with the pure element and the isotopic ratios to other isotopes already present in the sample are measured before and after the spike.This procedure should compensate for matrix interferences and drift in the instrument when the mass and ionization potential of the internal standard element are close to those of the analyte. For the determination of Tc and Rh Zr Mo Ru and Pd were used as internal standards all of which were already present in the sample. For the determination of U Pu isotopes were used as the internal standard. The equation used for the determinations by standard addition is where CsA and Cst are the concentrations of element A in the sample and in the standard solution respectively expressed in the same units Ms and Mst are the masses taken from the sample and standard solution respectively Rs is the count- rate ratio referred to the internal standard in the sample and RM is the same ratio in the spiked sample or mixture (spiked with a known concentration of the analyte).Results and Discussion Qualitative and Semiquantitative Studies on Spent Nuclear Fuel The concentration and isotopic abundances of fission products in spent nuclear fuel will depend on the reactor operating conditions. High burn-up and a thermal neutron spectrum will favour the occurrence of neutron-capture reactions which increases the difficulty of quantifying the fission products because of increased isobaric interferences. Fuels irradiated under different conditions were studied in order to check the suitability of direct measurement by ICP-MS. In Fig. 1 the ICP mass spectra of the fission product distribution of three different spent fuels analysed are shown.The first [Fig. l(a)] is a U02 fuel enriched originally to 1.42% 235U which was burned in a thermal reactor to a high burn- up (7.25 at.%) the second [Fig. l(b)] is an experimental fuel containing 45% Np and was irradiated in the Phenix reactor (fast neutron spectrum) for transmutation purposes13 and the third fuel [Fig. l(c)] is a commercial UOz fuel with an original enrichment of 3.2% and burn-up of 3.25 at.%. In all instances the z38U2f peak (m/z 119) is visible at the centre of the mass spectrum and shows the symmetry of the fission process. Differences between the fission product distribution are seenJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 121 1 8000 ( a ) 7000 1 I 6000 I 5000 4000 3000 2000 1000 0 80 90 100 110 120 130 140 150 160 170 20000 18000 16000 14000 'i 12000 $ 10000 - 8000 6000 4000 2000 0 I 80 90 loo i i o i i o 130 140 150 160 170 20000 I 18000 16000 14000 12000 10000 8000 6000 4000 2000 0 80 90 100 110 4 .I 125 135 145 155 Fig. 1 Mass spectra of the fission product distribution of three different spent fuels (a) high burn-up experimental fuel irradiated in a thermal neutron reactor; (b) Np-containing experimental fuel irradiated in the Phenix reactor (fast neutron spectrum); and (c) typical commer- cial UO spent fuel clearly for the high-mass fission products where the presence of neutron-capture reactions is mainly observed. As can be seen in Fig. 1 (a) odd-mass peaks are greatly reduced compared with even-mass peaks. For the fuel irradiated in a fast neutron spectrum reactor [Fig.1 (b)] neutron-capture reactions can hardly be detected (compare peaks at m/z 149 and 151 in both figures which correspond to isotopes of Sm with high neutron- capture cross-sections). For the commercial fuel [Fig. l(c)] the situation is intermediate between these two extremes. Most of the fission products can be detected in the ICP mass spectrum except Kr I and Xe which are lost during the process of fuel dissolution. Other elements such as Zr Mo Tc Ru Rh Pd and Te remain partially in the dissolver residue. In Fig. 2 the mass spectrum of a residue from the reprocessing of commercial spent fuel is shown. The high-mass 100000 80000 60000 r - 40000 20000 85 90 95 100 105 110 115 120 125 130 135 140 d z Fig. 2 Mass spectrum of fission products found in the residues from the reprocessing of commercial spent fuel.Isotopes of Zr (90 91 92 93 94 and 96) Mo (95 97 98 and loo) Tc (99) Ru (101 102 and 104) Rh (103) Pd (105 106 107 108 and 110) and Te (126 128 and 130) can be detected fission products are virtually absent from the residue and only those elements cited can be clearly detected. Based on the cumulative fission yields from 235U or 239Pu and taking into account possible neutron-capture and /?-decay reactions all nuclides expected to be present in the solution of spent fuel can be identified.14 Normally after a few years of cooling each fission chain decays to only one stable or long- lived isotope. However mixtures of nuclides can be observed for some masses owing to medium-lived fi-decay or neutron c a p t ~ r e .~ ' ~ For a cooling time of 3-5 years half-lives shorter than that of '44Ce (285 d) need not to be considered for p- decay isobars. Examples of significant fi-decay interferences after e.g. 5 years of cooling are those for 90Sr 134Cs 137Cs 147Pm 1 5 4 E ~ and ls5Eu which will be in equilibrium with the corresponding daughter products. lo6Ru and 144Ce would have virtually disappeared after 5 years of cooling. Also the inter- ference of 85Kr on "Rb disappears as Kr is lost during dissolution. Neutron-capture interferences will affect only those nuclides with large neutron-capture cross-sections (e.g. the lanthanides). One example of significant neutron-capture inter- ference is the formation of 14'Srn from 147Pm by neutron capture (to l4*Prn) and fast fi-decay to 14'Srn.This nuclide can interfere with the measurement of fuel burn-up based on the concentration of 14*Nd produced by direct fission. As a first approximation the low-mass fission products Rb to Cd can be considered free from isobaric interferences for their determi- nation by ICP-MS. The interference of 90Zr on 90Sr or vice versa can be ignored when analysing solutions of residue materials as Sr is not present in the residue (see m/z 88 Fig. 2). Where isobaric interferences occur the accurate determination of those nuclides by ICP-MS can be achieved only after a chemical separation. Semiquantitative analysis of fission products was accomplished by the use of the response curve of the instrument as no interference of U on the response for the fission products was observed at the 500ppm U levels.',' The response curve was prepared using natural element standard solutions.' If the response curve is applied to the data shown in Fig.l(c) the concentration of almost all fission product isotopes can be determined directly from the mass spectrum. Table 2 summar- izes the results obtained for this particular fuel and compares them with computer predictions using the KORIGEN code." Except for those elements which remain partially in the residue1212 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY7 NOVEMBER 1994 VOL. 9 Table 2 Fission product content of UO fuel (pg g-') by ICP-MS compared with the calculated composition obtained with KORIGEN KORIGEN Ratio Element ICP-MS (3960 d cooling time) KORIGEN ICP-MS Rb 33 1 270 0.82 Sr 787 650 0.83 Y 469 390 0.83 Zr 3006 2979 0.99 Mo 2425 2797 1.15 Tc 628 704 1.12 Ru 1294 1740 1.34 Rh 194 387 1.99 Pd 804 914 1.14 c s 1793 1877 1.05 Ba 1161 1535 1.32 La 980 1015 1.04 Ce 1625 1995 1.23 Pr 847 914 1.08 Nd 3204 3366 1.05 Pm 22 7 0.32 Sm 482 684 1.42 Eu 82 98 1.20 or produce b-decay isobars the agreement can be considered satisfactory.Semiquantitative data can be used also as a fast screening method for unknown spent fuel samples (pattern recognition) or as a quick method for the determination of fission product and actinides isotopic abundances and fuel burn-up. To illus- trate this a small fuel particle of 0.5 mm diameter of unknown origin at the time of analysis was dissolved completely in 8 mol I-' nitric acid and the solution was analysed by ICP-MS after simple dilution with 1% nitric acid.The response curve was prepared with natural element fission products Th and U as described before. The results obtained for the actinides in the dissolved sample are presented in Table 3. As can be observed the 235U isotopic abundance is lower than the natural abundance. The Pu isotopic composition does not follow the pattern typical for U02 fuels ( i e . 239 > 240 > 241 > 242) the concentration of 240Pu is higher than that of 239Pu which is never the case with U02 fuels. The ratio of transuranium elements to 238U is 0.0485; this means that almost 5% of the heavy actinide elements is Pu or neutron capture-produced Am and Cm. All these observations indicated that the sample corresponded to a MOX fuel.In order to prove this assump- tion the concentrations of fission products in the range 85-115 u were determined in the spent-fuel solution by com- parison with the response curve. The data were transformed into moles of each isotope. The cumulative fission chain yields Table 3 semiquantitative determination of the actinide elements Pattern recognition studies on unknown spent fuel samples Uranium Plutonium isotopic isotopic Ratio abundances Ratio abundances 2351238 0.0049 2401239 1.1924 (235 + 236)/238 0.0064 2411239 0.5280 2421239 0.2522 Ratio Transplutonium elements uranium (239 + 240 1- 241 + 242 + 243 + 244)/238 0.0485 Concentration of actinide isotopes in the dissolved samplelpg ml- ' 235u 0.84 236u 0.25 238u 170.20 239Pu 2.70 240Pu 3.21 P u - - ' ~ ~ A ~ 1.42 242Pu 0.68 243Am 0.18 244Cm 0.06 Total actinides 179.54 0.011 ' ' ' ' ' 1 80 85 90 95 100 105 110 115 120 m/Z Fig.3 Comparison of theoretical15 (A 239Pu and B 235U) and experimental (0) cumulative fission yield for an unknown spent fuel sample. Most of the fission products are produced by 239Pu fission were determined for each fission chain by assuming that the sum of all concentrations will correspond to 100% of fission (the contribution of fission Kr not determined was considered negligible).The results were compared with the theoretical cumulative fission yields produced by 235U and 239Pu.15 The results obtained are presented in Fig. 3. As can be observed the experimental data overlap clearly with the 239Pu fission yield curve confirming that this nuclide is the main source of fission in the sample.Based on the fission yield curve shown in Fig. 3 the fuel burn-up can be calculated. The integral under the curve on a molar basis corresponded to 10.47 pg ml-I of 239Pu fissioned. Given the total final heavy element concen- tration of 179.54 pg ml-' the burn-up was calculated to be 5.5 1 at.%. If typical characteristics for irradiated mixed oxide (MOX) fuels are considered,16 approximately 7% Pu and 93% natural U in origin and 5% Pu and 5% fission products after irradiation (45 000 MW year t -'),I6 it can be concluded that ICP-MS can be applied in pattern recognition studies on spent fuel samples. Determination of Fission Products and Actinides in Residues of Spent Fuel after Dissolution The first step in the reprocessing of spent nuclear fuel is its dissolution in nitric acid.Part of the transition metal fission products form an insoluble residue that contains also some U and Pu. These Pu losses to the residue affect especially the safeguards aspect of reprocessing whence it follows that the concentration of Pu in the residue has to be determined. The amount and elemental composition of the residue depend mainly on the fuel burn-up. The fission product isotopes can be identified in the mass spectrum (Fig. 2) assuming that there are no significant isobaric interferences. The main elements identified in the residue are Zr Mo Tc Ru Rh Pd U and Pu. Except for 99Tc and lo3Rh7 all elements are polyisotopic and show isotopic abundances different from the natural abundances.This however permits natural elements to be used as spikes and isotope dilution methodologies to be applied for the determination of these fission products. For Pu a highly enriched 239Pu reference material [National Institute of Standards and Technology (NIST) SRM 949f1 was used and the results were compared with those given by ID-TIMS. Determination of Zr Mo Ru and Pd The ICP-MS instrument was calibrated using natural elements in order to determine the mass discrimination factor." Once the instrument had been calibrated the correct isotopic ratiosJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1213 Table 4 Isotopic abundances (at.-%) and relative atomic masses (A,) of Zr Mo Ru and Pd produced by fission Mo Ru Pd Element Mass Zr 90 91 92 93 94 96 A 95 97 98 100 A 101 102 1 04 A 105 106 107 108 110 A Sample 1 (&s)* 4.43 f 0.08 12.42 2 0.1 1 14.59 f 0.13 16.44 f 0.10 19.01 f0.21 33.10 & 0.1 1 93.56 17.44f0.20 22.29 & 0.29 24.32 k0.13 35.95 & 0.16 97.88 26.83 f 0.16 33.13f0.18 40.04 k 0.13 102.44 23.28 f 0.20 40.55 & 0.1 7 17.54 0.24 13.40t 0.15 5.21 & 0.07 106.32 Sample 2 4.36 13.38 15.79 17.92 20.33 28.22 93.40 17.75 22.12 24.15 35.97 97.87 27.25 32.81 39.94 102.43 23.74 39.59 17.60 13.44 5.63 106.34 Sample 3 4.09 9.46 11.76 12.93 14.91 46.85 94.03 16.39 21.40 23.41 38.80 97.97 26.99 33.33 39.68 102.43 22.82 40.94 17.54 13.52 5.19 106.33 Sample 4 7.17 12.07 14.33 15.18 18.23 33.03 93.48 17.39 22.09 24.26 36.26 97.89 27.16 33.23 39.61 102.42 23.34 40.49 17.61 13.39 5.16 106.32 Sample 5 9.18 11.89 14.49 14.57 18.11 31.77 93.38 17.89 22.15 24.23 35.73 97.86 27.18 32.80 40.02 102.43 23.94 39.24 17.83 13.54 5.44 106.33 Sample 6 (Ss) 2.46 k 0.04 16.72 2 0.29 17.64 _+ 0.20 19.1820.19 20.35 & 0.20 23.66 2 0.12 93.23 23.40 k 0.20 23.87 k 0.23 24.47 2 0.16 28.27 t- 0.12 97.53 36.3 1 2 0.20 36.50 & 0.22 27.19 2 0.1 1 102.09 32.05 f 0.82 33.91 2 0.49 17.21 k0.63 11.59 & 0.34 5.23 k 0.41 106.20 * Standard deviation (s) calculated based on the relative error in the measured isotopic ratios applying error propagation theory.12 Values in the same range were obtained for samples 2-5.in the sample and in the spiked sample could be calculated in order to apply the isotope dilution equation. The results obtained for the corrected isotopic abundances and the corre- sponding relative atomic masses of Zr Mo Ru and Pd in six different residues are presented in Table 4.As can be observed the isotopic abundances of the elements and their relative atomic masses are different to those of the natural elements.17 Samples 1-5 were taken from different parts of the same fuel pin (with an average burn-up of 7.25 at.%) and sample six came from the reprocessing of commercial spent fuel (Fig. 2). As can be observed there is a remarkable difference between samples 1-5 and 6. The difference arises from two factors; first in samples 1-5 most of the fission comes from 239Pu which was formed in the reactor by neutron capture of 238U. The fission yields for 239Pu are different to those for 235U which is the main source of fission products in sample 6.Second neutron-capture reactions in samples 1-5 are much more important than in sample 6 as can be shown by the isotopic ratios of odd and even masses of the same element (e.g. Ru 101 and 102 and Pd 105 and 106). On the other hand the isotopic abundances in samples 1-5 are fairly constant for Mo Ru and Pd but not for Zr. This could be due to contamination with natural Zr from the fuel cladding (which would explain the changes in 90Zr) or to isobaric interferences due to neutron capture on "Mo which is the nuclide with the highest neutron- capture cross-section found in the residue. Depending on the relative concentrations of Zr and Mo in the different samples the isobaric interference on 96Zr will change accordingly. Fortunately for commercial spent low burn-up nuclear fuels (sample 6) neutron-capture reactions are much lower for the residue-forming elements and minimum isobaric interferences are to be expected.For the isotope dilution procedure isotopes 90 93 and 96 of Zr 98 and 100 of Mo 102 and 104 of Ru and 106 107 and 108 of Pd were selected for quantitative determination. The radioactive isotopes 93Zr and Io7Pd are not present in the natural elements and can be used as reference isotopes in the sample even at low abundances. In Table 5 the final results obtained for Zr Mo Ru and Pd are summarized. For Zr and Pd the final concentration data obtained using two reference isotopes in the sample have been included for comparison. As can be observed there is good agreement for the Zr and the Pd data using the two isotopes shown in all cases.Unfortunately no alternative technique could be used for comparison for the analysis of fission products. As can be observed given the relative atomic masses of the fission elements presented in Table4 the use of non- mass spectrometric techniques (e.g. ICP-AES) will give a systematic error when the natural elements are used for calibration. However here the presence of systematic errors due to isobaric interferences cannot be ruled out. For example sample 3 contained small amounts of Zr and large amounts of Mo which could explain the large changes in the experimen- tal isotopic abundances for 96Zr compared with samples 1 2 4 and 5. The presence of large amounts of 9 6 M ~ formed from neutron-capture on 95Mo will bias the results for Zr for this sample.Fortunately neutron-capture reactions are not so important for commercial spent nuclear fuel (e.g. sample 6). Determination of Tc and Rh For the determination of Tc and Rh in residues of spent fuel dissolution the method of standard additions with an internal standard was applied. The intensities at rn/z99 and 103 were referred to those at m/z 93 (Zr) 97 (+Mo) 101 (Ru) and 106 (Pd) before and after spiking with the pure elements. Table 5 also shows the concentrations of Tc and Rh found using the different internal standard elements. On the basis of the results for standard additions with internal standard for Tc and Rh analysis there seems to be no difference in the use of Zr Mo Ru or Pd as internal standards. Only for those samples where the concentration of the internal standard element is low (e.g.Pd in sample 6) is there a disagreement owing to the poor precision of the isotopic ratios measured. Determination of U and Pu Uranium was determined by the method of standard additions using plutonium isotopes present in the sample as internal standards. Natural uranium was used for the standard additions procedure. The determination of Pu was carried out using a 239P~-enriched spike for the isotopic dilution procedure. 242Pu was used as the reference isotope in the sample. The results obtained by ICP-MS were compared with those obtained by TIMS after chemical separation of U and Pu and spiking with 233U and 244Pu for isotope dilution analysis. The comparison is presented in Table 6.The isotope dilution procedure developed here was evaluated previously for systematic and random errors." It has been1214 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Table 5 Concentrations of Zr Mo Ru Pd Te and Rh in solutions of spent fuel dissolver residues (pg g-' of solution) Ratio used for quantification 93Zr 90Zr 96Zr 90Zr 1ooMa98Mo 106pd. l08pd 107pd. l08pd 1 0 4 ~ ~ . . lozRu 99Tc 93Zr 99Tc 97Mo 99Tc "'Ru 99Tc Io6Pd lo3Rh 93Zr lo3Rh 97Mo 103Rh. lolRu 103~h 106pd Sample 1 1.18 1.20 5.36 2.37 2.46 5.34 0.26 0.26 0.27 0.27 0.37 0.37 0.37 0.39 Sample 2 1.62 1.65 4.14 2.09 2.16 4.38 0.23 0.23 0.23 0.23 0.35 0.35 0.35 0.35 Sample 3 0.32 0.33 3.48 2.24 2.23 4.64 0.73 0.87 0.89 0.94 0.35 0.37 0.38 0.39 Sample 4 0.71 0.74 3.58 1.60 1.63 3.46 0.70 0.74 0.68 0.72 0.25 0.25 0.24 0.25 Sample 5 0.56 0.57 2.82 1.39 1.42 2.83 0.59 0.63 0.63 0.65 0.22 0.23 0.23 0.23 Sample 6 0.79 0.81 2.68 0.12 0.16 1.14 0.19 0.19 0.19 0.23 0.18 0.18 0.18 0.21 Table 6 in Table 5) Comparison of ICP-MS and TIMS for the determination of U and Pu in residues of spent fuel dissolution (pg 8-l in the solution as Element Method Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 U ICP-MS 0.57 0.61 0.47 0.15 1.28 Pu ICP-MS 0.12 0.46 0.02 0.16 0.06 TIMS 0.544 0.634 0.537 0.152 1.392 TIMS 0.125 0.444 0.018 0.158 0.058 observed that systematic errors due to mass discrimination could be corrected for by the use of natural elements standard solutions.Random error propagation studies showed that the determination of the isotopic abundances and relative atomic masses of the fission products contributed significantly to the final error obtained.12 The use of natural elements as spikes can be justified for the determination of Zr and Pd owing to the favourable isotopic abundances in the samples. However for Mo and Ru the use of an enriched spike would have been recommended. Regarding the analysis determination of trace amounts of U and Pu the only simplification used in the ICP-MS measure- ment was to ignore the contribution of 238Pu to the total Pu concentration.The isotopic composition of Pu in samples 1-5 was determined by TIMS. In all instances the isotopic abun- dance of 238Pu was lower than 2at.%. For the ICP-MS measurements the 238Pu was determined as 238U. Regardless of that simplification the analytical results for U and Pu by ICP-MS showed good agreement with those obtained by TIMS.Determination of Nd in Dissolution of Fast Neutron-irradiated Fuels Several Nd isotopes are commonly used as indicators of fuel burn-up as the fission yields from 239Pu and 235U for these isotopes are well known for both thermal and fast neutron spectra. Typically fuel burn-up is determined using 148Nd. However the direct determination of 148Nd is spent fuel by ICP-MS is hindered by the presence of 14*Sm produced by neutron capture of 147Pm or its P-decay product 147Sm. Fortunately nuclear fuels irradiated in fast neutron reactors show very low neutron-capture reactions which means that most of the isobaric interferences for the lanthanides will be eliminated and the concentration and isotopic abundances of Nd can be studied directly by ICP-MS.Later the data can be applied for fuel burn-up determination using available fission yield databases. Nd is currently determined in our laboratories by TIMS using lsoNd as a spike for the isotopic dilution procedure. The sample of irradiated fuel is diluted and Nd is separated from other elements by a time-consuming ion- exchange procedure. For ICP-MS analysis no separation was performed and natural Nd was used as the spike for simplicity. In the following calculations it was considered that all 144Ce (half-life 285 d) has decayed to 144Nd after 5 years of fuel cooling. For the mass range 142-1 50 the mass-discrimination factor K was determined using a natural Nd standard and the where Rexp and Rtheo are the experimental and theoretical isotopic ratios respectively and AM is the mass difference between the measured isotopes.The value of K was derived from the slope of the corresponding regression line. The results obtained for Nd are given in Table 7. For this mass range the mass discrimination factor is -0.617% per mass unit (s= 0.024%). The corrected isotopic abundances of Nd were deter- mined according to the equation As can be observed in Table 7 the systematic relative errors in the corrected isotopic ratios are now lower than the exper- imental standard deviation of the measured ratios. This value of the mass discrimination factor was used to obtain the correct isotopic abundances of Nd in the sample which are illustrated in Table 8 and compared with those obtained by TIMS after the separation of Nd.As can be observed there is good agreement for the main isotopes but the agreement is poor for 148Nd and IsoNd. This could be due to low but noticeable isobaric interferences from Sm isotopes which will not appear in the TIMS procedure because of the chemical separation of Nd from the other lanthanides. For the ICP-MS isotope dilution procedure 143Nd was used as reference isotope in the sample and 144Nd was used as reference isotope in the natural Nd spike. The use of 142Nd was rejected because of the presence of 142Ce in the sample. The concentration of Nd in the diluted sample was found to be 1.82 ppm which compares favourably with 1.87 ppm found by TIMS. Conclusions ICP-MS can be applied to the characterization of spent nuclear fuels with several advantages over other techniques used inJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1215 Table 7 Determination of mass discrimination factor for Nd Isotopic ratio 143 142 144 142 145 142 146 142 148 142 150 142 Theoretical value' 0.4485 0.8774 0.3056 0.6322 0.2113 0.2069 Experimental value 0.4503 0.8851 0.3111 0.6484 0.2185 0.2178 s r (n=5) 1.22 0.64 1.25 1.06 1.65 2.02 (Exp. - Theo.)/ Theo. (YO) 0.41 0.88 1.80 2.57 3.39 5.26 Mass difference -1 -2 -3 -4 -6 -8 Corrected ratio* 0.4475 0.8743 0.3054 0.6328 0.2107 0.2075 (Corr. - Theo.)/ Theo. (%) -0.21 -0.35 - 0.05 0.09 0.3 1 - 0.29 * After correction for -0.617% bias in the experimental isotopic ratios per unit mass. Table 8 ICP-MS and TIMS Isotopic abundances of fission Nd (at.-%) determined by Mass 142 143 144 145 146 148 150 ICP-MS (&s)* - 25.13 LO.08 22.88 0.09 18.61 k0.14 16.11-tO.20 10.67 t- 0.06 6.59 f 0.07 TIMS 0.00 25.96 23.30 18.81 16.34 9.90 5.69 * Calculated as in Table 4.the nuclear analytical lab~ratory.~ The main advantages are as follows 1. Almost all fission products and actinides can be detected and determined simultaneously by ICP-MS with very similar sensitivity. Only Kr Xe and I cannot be determined in dissolver solutions or residues. However Xe and I can be detected also in the solid fuel using laser ablation sampling.'' 2. The use of ICP-MS allows work with highly diluted solutions reducing the radiation hazard for the operator. 3. The application of a semiquantitative approach based on the instrument response curve agrees well with the theoretical calculation of the fission product inventory and also allows pattern discrimination studies to be performed in very small samples.This could be applied for example in the detection and investigation of clandestine reprocessing activities. 4. Quantitative data for fission products and actinides can be obtained by ICP-MS applying the isotope dilution tech- nique. Natural elements can be used as spikes for the fission product elements. However in some instances error theory predicts that the use of enriched spikes will give better results.12 The comparison of U and Pu determination with a reference method (ID-TIMS) was satisfactory. However several disadvantages of the proposed methods need to be mentioned 1.Isobaric interferences cannot be resolved. This applies preferentially to medium-and short-lived P-emitting nuclides but also to isobaric nuclides produced by neutron capture. The accurate determination of those isotopes can only be performed after chemical separation of the elements concerned. Mathematical correction of the isobaric interferences cannot be made. Even for samples irradiated in fast neutron reactors neutron-capture isobars can be present even at low levels. 2. The accuracy of the proposed methods cannot be tested as no reference materials are available. Comparison with other techniques is not possible for all fission products and actinides as no routine methods are currently developed for the complete analysis of spent nuclear fuels.The proposed methods could be fully accepted if the known sources of error (i.e. isobaric interferences) could be eliminated. In this respect the on-line coupling of an ion chromatograph to the ICP-MS system is being investigated.'* 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 References Campbell D. O. and Burch W. D. J. Radioanal. Nucl. Chem. Art. 1990 142 303. Kleykamp H. J. Nucl. Muter. 1990 171 181. Adachi T. Ohnuki M. Yoshida N. Sonobe T. Kawamura W. Takeishi H. Gunji K. Kimura T. Suzuki T. Nakahara Y. Muromura T. Kobayashi Y. Okashita H. and Yamamoto T. J. Nucl. Muter. 1990 174 60. Ache H. J. Fresenius' J. Anal. Chem. 1992 343 852. Garcia Alonso J. I. Thoby-Schultzendorff D. Giovanonne B. Koch L. and Wiesmann H. J . Anal. At. Spectrum. 1993 8 673. Garcia Alonso J. I. Thoby-Schultzendorff D. and Koch L. in Proceedings of the 15th Annual Symposium on Safeguards and Nuclear Material Management Report 26 EUR 15214 EN European Safeguards Research and Development Association (ESARDA) Ispra 1993 pp. 485-489. Garcia Alonso J. I. Babelot J.-F. Glatz J.-P. Cromboom O. and Kock L. Radiochim. Acta 1993 62 71. Garcia Alonso J. I. Thoby-Schultzendorff D. Giovanonne B. and Koch L. J . Radioanal. Nucl. Chem. Art. in the press. Crain J. S. and Galiimore D. L. Appl. Spectrosc. 1992 46 547. Garcia Alonso J. I. Garcia Serrano J. Babelot J.-F. Closet J.-C. Nicolaou G. and Koch L. in Applications of Plasma Source Mass Spectrometry 11 ed. Holland G. and Eaton A. N. Royal Society of Chemistry Cambridge 1993 p. 193. Fischer U. and Wiese H. W. Verbesserte Konsistente Berechnung de Nuklearen lnventars Abgebrannter D WR-Brennstofle auf der Basis von Zell-Abbrand-Verfahren mit KORIGEN Report 3014 Kernforschungszentrum Karlsruhe 1983. Garcia Alonso J. I. in preparation. Koch L. in Handbook of the Physics and Chemistry of the Actinides ed. Freeman A. J. and Keller C. North-Holland Amsterdam 1986 vol. 4 p. 457. Sellmann-Eggebert W. Pfening G. Munzel H. and Klewe- Nebenius H. Karlsruher Nuklidkarte Kernforschungzentrum Karlsruhe 1981. Wahl A. C. At. Data Nucl. Data Tables 1988 39 1. Bull. In5 I g e . Cadres COGEMA 1990 11 1. De Bievre P. and Taylor P. D. P. Int. J. Mass Spectrom. Ion Processes 1993 123 149. Betti M. Garcia Alonso J. I. Arbore Ph. Koch L. and Sato T. in Applications of Plasma Source Mass Spectrometry I I ed. Holland G. and Eaton A. N. Royal Society of Cambridge 1993 p. 205. Paper 4/00778F Received February 8 1994 Accepted April 6 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901209

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1217 Determination of "Tc in Nuclear Samples by Inductively Coupled Plasma Mass Spectrometry Jose lgnacio Garcia Alonso Fabrizio Sena and Lothar Koch European Commission JRC lnstitute for Transuranium Elements Postfach 2340 76125 Karlsruhe Germany Different approaches for the determination of "Tc in various nuclear samples are described. Technetium has been determined in leachates of spent fuel and vitrified highly active waste using the elemental response curve of the instrument. It has been observed that Tc is preferentially leached from spent fuel under oxidizing conditions (30% H202) indicating that the pertecnectate ion should be the leached species. It has also been determined in spent fuel dissolver solutions by the method of standard additions using other elements present in the sample as internal standards.Zirconium Mo and Ru showed to be adequate internal standards for the determination of Tc in those samples. The formation of several polyatomic ions containing Tc was observed at high Tc concentrations. Peaks at mlz 115 131 and 147 were assigned to TcO' TcO,' and TcO,' the TcO,+ being the predominant polyatomic ion. The ratios of the polyatomic ion to the main "Tc peak were estimated with the help of Rh as internal standard. The influence of plasma operating conditions on the relative abundance of those polyatomic ions has been studied. It was also observed that Tc suffers from a long and persistent memory effect. Keywords lnductively coupled plasma mass spectrometry; "Tc; nuclear waste The isotope 9 9 T ~ is produced by fission in relatively high yields.Owing to the long half-life (213000 years) of this p- emitter and its high solubility in water in the form of TcO,- the possible release of Tc to the environment is a cause of concern. In order to assess possible pathways of Tc in the environment studies have to be performed on the concen- tration and leaching behaviour of Tc in different nuclear waste forms and under various leaching conditions. Nuclear wastes containing Tc include mainly spent nuclear fuel residues of spent fuel dissolution (during reprocessing) medium- and high- activity waste (MAW and HAW) solutions (after reprocessing) and vitrified HAW. From those samples the spent nuclear fuel and the vitrified HAW are normally used for leaching and radionuclide migration studies as those are postulated to be the two main forms for long-term nuclear waste repository.It is clear that for meaningful long-term predictions the concen- tration and distribution of Tc and other long-lived radio- nuclides in the different waste forms have to be known accurately before the leaching and migration studies are performed. It has been shown that 99Tc is difficult to detect in environ- mental samples by radiochemical methods.' For nuclear waste samples cumbersome separation processes have to be applied to isolate Tc from other 0-emitters mainly '06Ru ref. 2 as fi-emission is not mono-energetic. In those cases inductively coupled plasma atomic emission spectrometry (ICP-AES) has been applied for Tc determination.2 The detection limit for Tc by ICP-AES was determined to be about 100 ng ml-' ref.3. Using ICP-mass spectrometry (MS) detection limits for Tc were in the low ngl-I more than three orders of magnitude lower than ICP-AES. In comparison with radio- chemical measurements ICP-MS offers three orders of magni- tude lower detection limits in optimum conditions.' Unlike the determination of 99Tc in environmental samples by ICP-MS,' no separation from 99Ru is necessary for nuclear waste samples. This is because 99Ru is not produced by direct fission in measurable quantities and its formation by decay from 99Tc can be safely neglected because of the long half-life of the parent nuclide. Potential matrix interferences in nuclear samples could arise from the presence of high concentrations of actinide elements (mainly U).In our glove-box modified ICP-MS instrument no matrix effects have been observed from U at 500ppm levels,* which is much higher than in the analysed samples. It has been observed that Tc forms several polyatomic ions and shows a pronounced memory effect. This could be of importance in the determination of low levels of impurities in Tc metal and will have to be taken into account. In this paper the formation of polyatomic ions containing 99Tc is studied under various experimental conditions as well as the influence of the polyatomic ions on the determination of impurities in Tc and the importance of the memory effect for low level Tc determination in nuclear waste samples. Applications for the determination of Tc in spent nuclear fuel dissolver solutions and dissolution residues are described. Semiquantitative approaches were applied to the study of Tc leaching from spent nuclear fuel and vitrified HAW.Experiment a1 Instrumentation The glove-box modified Elan 250 from SCIEX Canada has been described p r e v i o u ~ l y . ~ ~ ~ The plasma and quadrupole operating conditions are summarized in Table 1. Table 1 Operating conditions R.f. power/w 1400 Argon flow rates/l min-' Reflected power/w <5 Outer 12 Intermediate 1.4 Aerosol carrier 0.92 (32 psi) Sample uptake rate/ml min-' 1 Nebulizer type Meinhard Spray chamber Load coil-sampler cone distance/mm 25 (fixed) Interface pressure/Pa 266.64 Quadrupole working pressure/Pa 666.6 x Sampler and skimmer cones Platinum Ion lens settings Scott type double pass Optimized for lo3Rh and 99Tc Lens Range/V Optimum setting (% range) Bessel box B lens 0 +10 Bessel box P 0 -60 Einzel lens El 0 -20 Photon stop S2 0 -20 85 20 40 51218 JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 Reagents and Materials The 99Tc standard was obtained from Los Alamos National Laboratory (NM USA) as pure metal and dissolved in 8 mol 1-1 nitric acid to prepare a stock solution. The lo3Rh was obtained from Spex (Grasbrunn Germany) as 1OOOppm standard solution. All sample and standard dilutions were prepared by Mass into polyethylene bottles. Nitric acid was Merck Suprapur (Darmstadt Germany) and Milli-Q water (Millipore Eschborn Germany) was used throughout. Procedures for Sample Preparation The dissolution of spent nuclear fuel samples is performed normally in 7 mol 1-1 nitric acid in the hot cell facility under reflux." The residue obtained after dissolution is filtered washed dried weighed and then dissolved in a mixture of concentrated nitric and hydrochloric acids (1 + 9).Nitric acid (1 moll-') is used to dilute the samples. The ,,Tc is determined in those samples by the method of standard additions using isotopes of Zr Mo Ru and Pd also present in the sample as internal standards." Leaching of spent nuclear fuel and vitrified HAW was carried out in a hot cell under various experimental conditions. Details of the experimental conditions and extensive discussion of the data will be presented elsewhere.I2 Quantification of Tc and other fission products in the samples was performed by refer- ring to a multi-elemental response curve.Standard Additions with Internal Standard for Tc Determination For a mono-isotopic element A with NAS the number of total atoms in the sample S and NBS the number of total atoms of a selected isotope of another element B; the isotopic ratio R in the sample will be Rs=NAS:NBS taking the ratio to the internal standard isotope of B. When we spike NAS atoms from the sample with NASp atoms from a standard solution of the mono-isotopic element A the isotopic ratio in the mixture R M referred to the internal standard will be RM = "AS + NASp)/NBS Substituting NBs and rearranging for NAs we obtain NAS = NASpRS/(RM - (1) As can be observed eqn. (1) is independent of the nature and concentration of the isotope of element B selected as internal standard as far as it is not present in the spike solution used.When we express eqn. (1) in concentration units then c A S = CASp (MSp/MS) LRSMRM - ( 2 ) where Ms and CAS are the mass and concentration of element A in the sample and CAsp and M those of the standard solution. For the analysis of Tc in spent fuel dissolver solutions or residues eqn. (2) was applied." Results and Discussion Measurement of wTc by ICP-MS Optimization of conditions In order to evaluate the influence of plasma and quadrupole conditions on the sensitivity of the "Tc determination a standard solution was prepared containing about 1 pg g-' of 99Tc and Io3Rh (in these studies Rh was used both as a candidate internal standard and analogue to Tc for ICP-MS).The intensities at m/z 99 and 103 were evaluated for a series of plasma r.f. powers and nebulizer pressures. The results obtained for Tc at 1200 1300 and 1400 W forward power are illustrated in Fig. 1. As can be observed optimum nebulizer pressure for Tc increases with increasing plasma r.f. power. Similar results were obtained previously with this instrument 22 24 26 28 30 32 34 36 Nebulizer pressure/lb in-2 Fig. 1 Influence of nebulizer pressure for various r.f. powers on 99Tc signal; Tc = 1.070 pg g- for other elements.* In the case of Rh optimum conditions were similar to those obtained for Tc. The molar sensitivity ratio of Tc Rh for the different conditions studied is presented in Fig. 2. The molar ratio R was determined according to the equation = (199/ATcCRh )/(1103ARhCTc) where I and I, are the count rates measured at m/z 99 and 103 respectively ATc and ARh are the corresponding atomic masses of the elements and CRh and CTc the concentrations of the elements in the test solution.As can be observed in Fig. 2 the molar sensitivity ratio does not change significantly with the plasma forward power and could be considered constant for a range of optimum nebulizer flow rates. The four ion lenses in our Elan 250 were optimized for both Tc and Rh in the optimum plasma conditions. The data for Tc are shown in Fig. 3 for all ion lenses studied. Optimum conditions which appear in Table 1 were the same for both elements. The molar sensitivity ratio Tc Rh is illustrated in Fig.4. As can be observed the ratio can be considered constant and independent of the optimized ion lens or setting. The net result from this study is that Rh behaves in the same way as Tc regarding plasma and mass spectrometer conditions. The mean molar sensitivity ratio from Figs. 2 and 4 is 1.12. This means that Tc is measured with a somewhat higher sensitivity than Rh. A similar sensitivity for both elements in our modified Elan 250 would normally be expected.* The difference could be explained by the higher ionization potential of Rh (7.7 eV) compared with Tc (7.3 eV). The detection limit for Tc was calculated to be 0.1 ng 8-l based on three times the standard deviation of the background at m/z 99 in the absence of Tc memory effects. The detection limit for Tc degrades seriously when a concentrated Tc standard is measured owing to memory effects (see below).1.3 1 0 0 0.5 ' I I I I I I 22 24 26 28 30 32 34 36 Nebulizer pressure/lb in-2 Fig. 2 Molar sensitivity ratio Tc Rh for different plasma operation conditions (Tc=1.070pg g-' and Rh=0.997 pg g - ' ) 0 1400; @ 1300 and 0 1200 WJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1.3 12 - 0 C h E l lens 0 10 20 30 40 50 60 70 80 90 100 Ion lens setting (% range) 0.7 1 0 . 5 ' ' ' I I ' I ' I 0 10 20 30 40 50 60 70 80 90 100 Ion lens setting (YO range) Fig. 4 Molar sensitivity ratio Tc Rh for different ion lens settings a B; 0 El; 0 P; and a S2. Tc= 1.070 pg g-' and Rh=0.997 pg g-' Formation of polyatomic ions containing Tc In order to study the formation of polyatomic ions from Tc a standard solution containing 925 ppm of Tc and 0.856 ppm of Rh was prepared and measured under different experimental conditions. The use of Rh provides an opportunity to obtain absolute values for the Tc' polyatomic ion ratios given the previously measured molar sensitivity ratio Tc Rh.For the following calculations the matrix interferences from the 925 ppm 99Tc standard solution we assumed to be similar for all peaks measured including 99Tc itself. Fig. 5 shows the mass spectrum obtained for this solution in the m/z ranges 48-50 98-104 114-117 130-133 146-149 and 162-165. As can be observed several polyatomic ions are formed TcOzf being the most abundant. The formation of hydroxide ions is also observed with increasing abundances going from TcO + to TcO,'.The formation of TcO,' and TcO,H+ could not be detected. The ratio polyatomic ions 99Tc+ RXzg9 was determined with the help of the lo3Rh standard. The following expression was used Rx:99 =IxA,,CRh/1.121,0,ARhc== where I is the intensity measured at the mass of the polyatomic ion x. The formation of Tc polyatomic ions was studied for 1200 1300 and 1400 W forward power at different nebulizer flow rates. The results obtained at 1300 W forward power are illustrated in Fig. 6. As can be observed there are no dramatic changes in the formation of polyatomic ions with the nebulizer flow. By increasing the r.f. power from 1200 to 1400 W all 0.0008 0.0007 & 0.0006 .- 0.0005 0.0004 2 0.0003 0.0002 0.0001 0 + 0 0 .- +-' a" TcO,' 1219 40 60 80 100 120 140 160 rn/z Fig.5 ICP mass spectrum of a solution containing 925 ppm and 0.856 ppm of Rh under standard operating conditions of Tc 1 22 24 26 28 30 32 34 36 Nebulizer pressurdlb in-2 Fig. 6 Ratios of polyatomic i~ns:'~Tc+ peak for different nebulizer pressures at 1300 W forward power ratios decreased gradually. The ion TcO,+ is the most abun- dant under all experimental conditions measured. For the standard conditions showed in Table 1 the ratios for the polyatomic ions are summarized in Table 2. Previous results obtained by Crain et a!! are shown for comparison. In our data the main polyatomic ion is Tc02+ while TcO+ was the main ion observed by Crain et al. A lower abundance was observed for the monoxide but higher abundances for the di- and trioxides.However the results are in the same order of magnitude. The reason for the higher oxide levels could be the long distance between the load coil and the sampler cone which in Table 2 Ratios of polyatomic ion:99Tc+ using Io3Rh as internal standard Polyatomic ion (x) m/z Ratio x 99 ( x lo4) Ref. 4 data ( x lo4) Tc2 + 49.5 0.24 TcH + 100 ND* TcO + 115 1.8 TcOH' 116 0.07 Tc02+ 131 5.0 TcO,H+ 132 0.48 Tc03H+ 148 1.6 TcO + 147 2.0 * ND = not detected. 0.3 0.2 3.0 ND 1 .o 0.2 1 .o 0.41220 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 our instrument is 25 mm. Owing to safety reasons in our glove box this distance cannot be optimized.' A practical consequence of these measurements is that in the determination of trace impurities in Tc metal "'In should not be used as internal standard.Other isobaric interferences due to the polyatomic Tc ions can be resolved by the use of alternative isotopes. Memory eflect During the measurement of the 925ppm Tc solution the presence of a persistent memory effect was observed even days after measuring the sample. This could be eliminated only after careful cleaning of all nebulizer glassware (spray chamber and transfer tube) with 10% nitric acid and then absolute ethanol. No other way of removing the memory effect was tested. After measuring the 925ppm Tc solution and nebulizing large amounts of 1% nitric acid a constant background of about 3000 ions s-l was measured at m/z 99. In absence of Tc memory the background at m/z 99 is less than 10 counts s-' in our instrument.Determination of wTc in Nuclear Samples Leachates of spent fuel and vitrified HAW In order to study the leaching behaviour of Tc other fission products and actinides from different nuclear waste forms several leaching conditions have to be tested which mimic possible natural scenarios in long-term geological repositories. For these studies,12 ICP-MS has been applied to characterize the leachates obtained. Owing to the inherent poor reproduc- ibility of the leaching experiments highly accurate analytical methods are not required for this type of study; semiquantit- ative approaches have been followed for the ICP-MS measure- ments. Fig. 7 shows the mass spectrum for the fission product region of one leachate of spent nuclear fuel obtained under oxidizing conditions (leaching with 30% H202).As can be observed 99Tc is leached under those conditions. Other elements clearly detected include Rb (85 and 87) Sr (88 and 90) Mo (95 97 98 and loo) Cs (133 134 135 and 137) and Ba (138). The presence of the 238U2+ peak at m/z= 119 (only about 0.5% of the main 238U+ peak) shows that uranium is also leached under these conditions. Fig. 8 shows the mass spectrum of a leachate of spent fuel in distilled water (reducing conditions). As can be observed very low levels of Mo Cs Ba and Gd isotopes could also be detected. The peak at m/z 99 corresponds to Tc but the level is close to the detection limit. 14 I CS :S cs Ba / I .. 1 80 90 100 110 120 130 140 150 160 m/z Fig. 7 ICP-mass spectrum of a leachate of spent fuel after 1 week of leaching in autoclave with 30% hydrogen peroxide 500 c s cs I 300 00 -.200 m m .- 100 0 80 90 100 110 120 130 140 150 160 m/z Fig. 8 TCP mass spectrum of a leachate of spent fuel in distilled water Finally Fig. 9 shows the mass spectrum in the fission products region of a leachate of the French R7T7 HAW Glass (CEA- CEN Marcoule France) in distilled water. In this sample Rb Sr Mo Tc Cs Ba La Ce Pr Nd and Gd can be detected. The isotopic composition of the elements in the glass shows almost natural abundances in all cases. Only the peaks at m/z 90 (Sr) 99 (Tc) and 135 and 137 (Cs-Ba) show that the sample has a non-natural origin. If the isotopic composition of the fission elements shown in Fig. 7-9 are compared the difference between natural and fission elements becomes clear.The R7T7 HAW glass was prepared by mixing genuine with simulated nuclear waste in order to get a high loading of fission products in the glass while keeping the activity low for easy handling in leaching experiments. The fact that Tc is leached only under oxidizing conditions in spent fuel and is easily leached from HAW glasses suggests that the form in which Tc is leached must be as the Tc0,-ion which is readily present in the glass matrix but not in the fuel where it is present mainly as metallic Tc." Quantification of Tc and other fission products in leachates of spent fuel and HAW glasses was performed by the use of the response curve of the instrument for the m/z range 85-153. No matrix effects up to 500 pg 8-l of U were observed.* The response curve was prepared by measurement of a standard solution containing Rb Mo Ba Nd and Eu in 1% v/v nitric acid.The leaching samples were also acidified with nitric acid (1% v/v) for the ICP-MS measurements. The response curve was in all cases similar to that described previously.8 Under 50 I 45 1- Mo 40t I c s 35 30 13 8 25 20 15 10 5 0 r c Cn 0 Ba 80 90 100 110 120 130 140 150 160 m/z Fig. 9 ICP mass spectrum of a leachate of the R7T7 HAW glass in distilled waterJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1221 the standard conditions of Table 1 the sensitivity increased almost linearly from Rb to Eu for the various nuclides meas- ured. Table 3 shows the results obtained for the concentrations of Tc and other fission products present in the sample shown in Fig.7. Other radioactive isotopes detected included 90Sr 134Cs 135Cs and 137Cs. Based on the data shown here it can be concluded that ICP-MS can provide relative data on the leaching behaviour of different fission products and particularly on that of Tc. In combination with the ICP-MS analysis of the spent fuel or the vitrified HAW leaching kinetics could be studied for most fission products and actinides.12 The fact that no separation is necessary for Tc determination in leachates of spent fuel and HAW glasses compares favourably with classical methods of Tc determination involving cumbersome and time-consuming radiochemical separation methods. Table 3 Semiquantitative determination of fission products in leachates of spent fuel in oxidizing conditions Mass Element Concentration of isotope/ng ml-' 85 87 88 90 91 92 93 94 95 96 97 98 99 100 127 129 133 134 135 136 137 138 Rb Rb Sr Sr Zr Zr Zr Zr Mo Zr-Mo Mo Mo Tc Mo I I cs c s c s Ba c s Ba 22 55 19 28 ND* ND ND ND 42 6 49 55 282 60 ND ND 194 5 60 ND 201 12 * ND = not detected.Spent fuel dissolver solutions Technetium in spent nuclear fuel is mainly present in metallic alloys.1° During the reprocessing of spent fuel using the PUREX process the sample is dissolved in 7 mol 1 - l nitric acid. The dissolution of spent fuel in the nitric acid is not quantitative; an insoluble residue is formed" which contains mainly the noble metal fission products (Ru Rh Pd) plus Zr Mo and Tc and traces of U and Pu." The amount of insoluble residue formed depends on the fuel burn-uplo and on the reprocessing plant operating conditions.In order to study the distribution of Tc and other fission products between the dissolver solution and the insoluble residues under different dissolution conditions five pellets (about 30 g each) of the same spent fuel pin were dissolved in nitric acid of various mo1arities.l' The residue'' and the corresponding spent fuel solution were analysed for Tc and other fission products. The method of standard additions was applied for Tc and Rh determination in those samples. In order to compensate for plasma instabilities and long-term drift other isotopes present in the sample were used as internal standards. Nuclides used as internal standards included 93Zr 97Mo "'Ru and lo6Pd which were already present in the sample (note that 99Ru is not produced by fission so no isobaric interferences are to be expected).Table 4 summarizes the results obtained for the five solutions of spent fuel using the different internal standards. For comparison the data obtained for Io3Rh has been also included. As can be observed similar results could be obtained for all the internal standards except for Io6Pd. This is due to its very low concentration in the spent fuel solution which made the measured isotopic ratios unreliable. Based on our previous results for the analysis of residues," the distribution of Tc between solution and residues for various nitric acid molarities can be evaluated. In Table 5 the final results for the distribution of Rh and Tc between spent fuel residues and solutions are summarized. Based on the measured concentrations of Tc and Rh both in the fuel solution and residue the absolute amounts (mg) of Tc and Rh present in both phases can be calculated (Table 5).As can be observed the amounts of Tc and Rh dissolved seemed to be independent of the nitric acid molarity used. Also given the mass of the pellets dissolved (about 30 g in all cases) the total concentration of Tc and Rh in the solid spent fuel were determined (inventory). Taking into account that all Table 4 Determination of Tc and Rh in solutions of spent nuclear fuel. Comparison of different internal standards Concentration found/pg g-' in solution Element Ratio Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Tc 99 93 99 97 99 101 99 106 Rh 103 93 103 97 103 101 103 106 0.149 0.138 0.140 0.181 0.045 0.043 0.043 0.05 1 0.148 0.159 0.156 0.166 0.045 0.048 0.047 0.049 0.131 0.138 0.137 0.171 0.035 0.036 0.036 0.041 0.113 0.102 0.114 0.122 0.037 0.034 0.037 0.039 0.136 0.130 0.141 0.110 0.039 0.038 0.040 0.034 Table 5 Distribution of Rh and Tc between fuel solution and residue Tc Rh Tc Rh c Tcl CRhI HNOJ residue/ residue/ solution/ solution/ in fuel/ in fuel/ %Tc in %Rh in Sample mol 1-1 mg mg mg mg mg g - ' mg g-' residue residue 1 3 9.71 13.80 9.00 2.72 0.58 0.51 51.9 83.5 2 5 8.17 12.35 9.87 2.99 0.56 0.48 45.3 80.5 3 7 29.27 12.92 8.90 2.37 1.20 0.48 76.7 84.5 4 4 36.45 12.72 7.69 2.52 1.40 0.48 82.6 83.5 5 6 29.30 10.70 9.18 2.63 1.20 0.42 76.2 80.31222 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 pellets come from the same fuel pin with similar burn-up,'O the total amounts of Rh and Tc should be also similar. In fact this was not the case for samples 1 and 2 where the amounts of Tc found in the residues are lower than in the other cases. It is probable that Tc losses due to volatilization after the dissolution of the residue in a nitric acid-hydrochloric acid mixture are responsible for the lower Tc inventory values found for samples 1 and 2. However the total amounts of Rh are very similar in all cases as expected. Finally except for samples 1 and 2 the percentage of dissolved Tc and Rh was probably independent of the nitric acid molarity used. However this was not the case for other fission products like Zr and Mo or actinides like Pu where the amounts found in the residues decreased with increasing nitric acid concentration." 1 2 References Kim C.K. Morita S. Seki R. Takaku Y. Ikeda N. and Assinder D. J. J. Radioanal. Nucl. Chem. Art. 1992 156 201. Adachi T. Ohnuki M. Yoshida N. Sonobe T. Kawamura W. Takeishi H. Gunji K. Kimura T. Suzuki T. Nakahara Y. Muromura T. Kobayashi Y. Okashita H. and Yamamoto T. J. Nucl. Muter. 1990 174 60. 3 4 5 6 7 8 9 10 11 12 Ache H. J. Fresenius' J. Anal. Chem. 1992 343 852. Crain J. S. and Gallimore D. L. Appl. Spectrosc. 1992 46 547. Brown R. M. Long S. E. and Pickford C . J. Sci. Total Environ. 1988 70 265. Kim C. K. Otsuji M. Takaku Y. Kawamura H. Shiraishi K. Igarashi Y. Igarashi S. and Ikeda N. Radioisotopes 1989 38 151. Smith M. R. Wyse E. J. and Koppenaal D. W. J. Radioanal Nucl. Chem. Art. 1992 160 341. Garcia Alonso J. I. Thoby-Schultzendorff D. Giovanonne B. Koch L. and Wiesmann H. J. Anal. At. Spectrom. 1993 8 673. Garcia Alonso J. I. Babelot J. -F. Glatz J. -P. Cromboom O. and Koch L. Radiochim. Acta 1993 62 71. Glatz J. -P. Garcia Alonso J. I. Kameyama T. Koch L. Pagliosa G. Tsukada T. and Yokoyama H. 3rd International Conference on Nuclear and Radiochemistry Vienna Austria September 7-11 1992 paper No W02. Garcia Alonso J. I. Thoby-Schultzendorff D. Giovanonne B. Glatz J. -P. Pagliosa G. and Koch L. J. Anal. At. Spectrom. 1994 9 Glatz J. -P. J. Nucl. Mat. in the press. Paper 4/03264K Received June 2 1994 Accepted July 22 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901217

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF AT\( 4LYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1223 Hydride Interference on the Determination of Minor Actinide Isotopes by Inductively Coupled Plasma Mass Spectrometry Jeffrey S. Crain Analytical Chemistry Laboratory Chemical Technology Division Argonne National Laboratory 9700 South Cass Avenue Argonne IL 60439 USA Jorge Alvarado Environmental Research Division Argonne National Laboratory 9700 South Cass Avenue Argonne IL 60439 USA Hydrogen adducts of the major naturally occurring actinide isotopes 232Th and 238U were studied using an inductively coupled plasma mass spectrometer. The hydride:atomic ion ratios for both elements varied as a function of the parameters that were studied i.e. nebulizer flow rate solution uptake rate and desolvation conditions. When the instrument sensitivity for U and Th was optimized 232ThH+:232Th+ was found to be (3.9 0.2) x 1 OU5 with pneumatic nebulization and (2.1 0 0.07) x 1 0-5 with ultrasonic nebulization.Under the same conditions 238UH + . -238 U + was found to be (3.2k0.2) x using pneumatic and ultrasonic nebulization respectively. Conditions that reduced hydrogen number density and/or increased plasma temperature decreased the h dride:atomic ion ratio. Such conditions are best if 233U and 239Pu are to be determined in the presence of Y3'Th and 238U. Keywords lnductively coupled plasma mass spectrometry; spectral interference; actinide determination and (1.8$-0.1) x Recent work by Smith et al.' has demonstrated that in the absence of interferences inductively coupled plasma mass spectrometry (ICP-MS) with electrothermal vaporization (ETV) sample introduction has greater sensitivity than does radiometry for detection of radionuclides with half-lives in excess of = 300 years.Unfortunately electrothermal vaporiz- ation is unavailable in many laboratories. However if one assumes that the efficiency of sample introduction by solution nebulization runs in the range of 1-lo% then the data presented by Smith et al. suggest that the 'half-life cut off' between ICP-MS and radiometry would be =lo4 years in the event that solution nebulization were employed in place of ETV. Recognizing that this 'half-life cut off' is approximate to within an order of magnitude the following actinide isotopes could be considered as candidates for determination by solution nebulization ICP-MS in place of radiometry u It is commonly assumed that actinide determinations by ICP-MS are free of spectral interference by polyatomic ions.This assumption is reasonable in the case of the major naturally occurring actinide isotopes (232Th and 235,238U). It is also reasonable in the event that the actinides of interest (whether they be natural or anthropogenic) are present in similar concentrations. However if there is a large excess of one or several actinides (as would be the case in most environmental samples where Th and U predominate) then molecular ion interference could occur for minor actinide isotopes (e.g. 232Th160 + on 248Cm + >. Garcia Alonso et d2 have determined the degree of oxide ion formation for various actinides analysed by ICP-MS; MO+:M+ ranged from 0.026 (for Th) to 0.002 (for Cm).In the same publication the authors also described the inter- ference of 238UHf on 239Puf; however they were unable to quantitatively measure the UH+ abundance. Russ and Bazan3 reported that 238UH+:238Uf was x 1 x but they did not examine the parametric behaviour of the hydride. In this work using nebulizers of different efficiency we examined the abun- dance and behaviour of thorium and uranium hydride ions under different ICP-MS operating conditions. Thorium and uranium were selected because they are likely to be the major actinide elements in the majority of natural and man-made materials and as such their hydrides are most likely to interfere with the determination of adjacent actinide isotopes.229,230,232~h 231pa 233-236,238~ 236,237~~ 239,240,242,244~ 243Am 2 4 5 - 2 4 8 2 5 0 ~ ~ and 2 247Bk. Experimental Instrumentation The ICP mass spectrometer used in these investigations was a Fisons PlasmaQuad I1 (Fisons Elemental Winsford Cheshire UK). Two types of nebulizers were employed a V-groove pneumatic nebulizer (supplied by Fisons Elemental) and a U5000-AT ultrasonic nebulizer (CETAC Technologies Omaha NE USA). A water-cooled Scott double-pass spray chamber (also supplied by Fisons Elemental) was used in combination with the pneumatic nebulizer. Samples were delivered to the nebulizers using a Minipuls 3 peristaltic pump (Gilson Medical Electronics Middleton WI USA). Instrument operating con- ditions for pneumatic nebulization (PN) and ultrasonic nebul- ization (USN) are listed in Table 1.These conditions are typical for this particular ICP mass spectrometer. Samples A commercially available standard solution containing 1 g I-' of uranium (Spex Industries Edison NJ USA) was used in Table 1 ICP-MS operating conditions Parameter Forward power/kW Ouler argon/l min-' Intermediate argon/l min-' Aerosol carrier argon/l min- Sample uptake/ml min- ' Condenser temperature/K* Heater temperature/K Ion optical voltages Extractor Collector Lens 1 Lens 2 Lens 3 Lens 4 Pole bias Detector PN* 1.35 13.6 1.5 0.84 1 .o 268.2 1 - 193 2.6 - 3.8 - 32 - 47 - 2250 2.4 - 1.6 USN 1.35 13.6 1.9 0.75 1.7 270.6 453.2 - 260 1.4 - 5.1 1.2 - 1.6 -31 -44 - 2250 * Spray chamber temperature (PN) or condenser temperature (USN).1224 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 this investigation. The isotopic composition of the uranium in the standard solution determined in our laboratory by thermal ionization mass spectrometry (TIMS) is listed in Table 2. Radiochemical analysis of this standard indicated that 239Pu:238U was less than 2.9 x lo-''. A standard solution of thorium was prepared by dissolving 99.9% m/m 232Th02 (obtained from sources at Argonne National Laboratory) and diluting the dissolved solid in 25% v/v HNO,. The diluent was prepared from concentrated high- purity nitric acid (Ultrex Grade J. T. Baker Phillipsburg NJ USA) and 17 MRcm de-ionized water (Super-Q Water Polishing System Millipore Bedford MA USA). The final concentration of the Th standard was determined by isotope dilution TIMS. Radiochemical analysis of the standard indi- cated that 233U:232Th was less than 2.4 x For the measurements described herein aliquots of the uranium and thorium standard solutions were mixed and diluted volumetrically with 5% v/v HN03. Working solutions were prepared such that equal mass concentrations of uranium and thorium were present.The acid concentration was selected to maximize the solubility of t h ~ r i u m . ~ The nitric acid diluent was prepared from 18 MR cm de-ionized water (Nanopure Water Polishing System Barnstead/Thermolyne Dubuque IA USA) and concentrated high purity nitric acid (Optima Grade Fisher Scientific Pittsburgh PA USA). Measurements Tuning and data acquisition On each day of this study the instrument was started and allowed to equilibrate for 1 h.Mass calibration accuracy and temporal stability were tested using a solution of Li Mg Co Y In La Lu Pb and U (10 pg 1-1 of each for PN and 0.5 pg 1-' of each for USN). In general it was unnecessary to re-tune the instrument on a daily basis. Except as noted measurements were acquired by multiple ion monitoring (i.e. 'peak hopping') while an analyte solution containing 50 yg I-' each of Th and U (for PN) or 1 pg 1-' each of Th and U (for USN) was pumped into the nebulizer. Each measurement consisted of ten consecutive integrations and each integration was 30 s in duration. A 20 ns dead-time correction was automatically applied to all integrations. The isotopes of interest and their respective dwell times are listed in Table 3.Parameter behaviour studies Variations in hydride-ion abundance were studied by varying a single operating parameter while all the other parameters shown in Table 1 were held constant. In these studies the analyte solution was continuously introduced. Once the operating parameter of interest had stabilized at a particular setting an additional 60 s was allowed to ensure that the instrument had equilibrated prior to measurement. After each study was complete a measurement at or near the optimum parameter setting was repeated to assess the influence of instrument drift. These data are shown as unconnected points in the parameter behaviour plots. Once the measurements were completed the ratio of back- Table 2 was not detected Isotopic composition of the uranium standard solution.233U Isotope 2 3 4 u 2 3 5 u 2 3 6 ~ 2 3 8 ~ at.-%* 0.0022 0.3265 0.0 170 99.6543 Table 3 Isotopes and dwell times used during multiple ion monitoring mlz 229t 2305 232 233 235 238 239 242 Dwell time/ms* 20 20 2 20 10 2 20 20 Channels per peak 3 3 3 3 3 3 3 3 Analyte Background Background 232ThH+ 232~h + 2 3 5 ~ + 238u + 2 3 8 ~ ~ + Background *Integration time at the specified m/z for a single sweep of the quadrupole over a single acquisition channel. Multiple sweeps were added to obtain each 30 s integration. t Used with USN only. Used with PN only. ground signal to major isotope signal was subtracted from the ratio of hydride signal to major isotope signal e.g. (239:238),,,,,-(242:238) so as to provide a net hydride ion atomic ion ratio.The standard deviation of the net ratio was calculated by error propagation using the variances in the background and gross ratio^.^ The background masses used in this study are given in Table 3; in the case of thorium m/z 229 was used as a background mass during ultrasonic nebuliz- ation because 230Th was detected in the analyte solution. Other studies In instances where all instrument parameters were held con- stant (e.g. when studying the reproducibility of the hydride:atomic ion ratio) a diluent blank was measured in advance of the standard(s) such that background correction was performed by blank subtraction. Ratios were then calcu- lated from net signals. A blank was also measured at the end of the procedure to assess the stability of the background. No significant increases in background were observed by this means.Results and Discussion Spectral Interpretation A mass spectrum of the 50 ygl-' thorium and uranium solution (nebulized pneumatically) is shown in Fig. 1. The spectrum was integrated for 5 min. The peaks at m/z 232 m/z 234-236 and m/z 238 are due to the analytes in solution. The peaks at m/z 233 and m/z 239 are too intense to be due to 233U or 239Pu in the standards thus we attribute them to lo' 7 228 230 232 234 236 238 240 242 m/z * Atom percentages were determined by TIMS. Fig. 1 by PN Mass spectrum of 50 pg I-' Th and 50 pg 1-' U acquiredJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1225 - - I I I 232ThH+ and 238UH+. Note that the peaks at m/z 233 and m/z 239 are fully resolved from the adjacent major peaks thus the abundance sensitivity of the mass spectrometer was adequate for accurate measurement of the hydride:atomic ion ratios.Parameter Behaviour In these studies we found that the parameter behaviours of 232Th+ and 238U+ were very similar as were the behaviours of their hydrides. Therefore in the interest of brevity only the parameter behaviour of 232Th+ and 232ThH + are discussed here. Note that the error bars in the parameter behaviour plots represent one standard deviation in the measurement. Analyte Concentration Fig. 2 shows the variation in net 232Th+ signal and the hydride atomic ion ratio as a function of thorium concentration. These data were obtained through PN. Using the origin as the y- intercept linear regression of the net signal versus concen- tration data yielded a correlation coefficient of 0.9996.In all cases the ratios of the y residuals (i.e. predicted signal less observed signal) to the standard deviation in the observed signals were less than 1.6. A paired two-tailed t-test of the observed and predicted signals5 indicated no significant differ- ences between the two sets of data at the 95% confidence interval. These observations suggest that the calibration plot is linear within experimental error even though the signal reaches z 1.8 x lo6 ions s-'. As such we conclude that our dead-time correction was adequate to compensate for count- ing losses. As indicated by Fig. 2 no statistically significant changes in the hydride atomic ion ratios were observed as a function of concentration.Given that analyte concentration was the only variable herein these data suggest that the hydride ions are formed by an equilibrium process i.e. Th' +HeThH+ where the hydride:atomic ion ratio would be equal to the product of the equilibrium constant and the hydrogen number density. If this were so conditions that increase hydrogen number density would increase the hydride atomic ion ratio as would conditions that increase the equilibrium constant (e.g. by reducing plasma temperature). Desolvation conditions Figs. 3 and 4 show the effect of changing desolvation conditions on atomic ion signal and hydride atomic ion ratio. The data shown in Fig. 3 which are for PN were obtained by varying the temperature of the cooling water entering the jacket around 2.0 I 1.5 + 3 2 1.0 ul 1.m 0 v) ,_ 0.5 N N 0 I 1 I I I I0 20 30 40 50 Th concentration/pg I-' 6.0 5.0 v) I 0 r 4.0 x Y + r 3.0 +' I t- 1 2.0 1 .o Fig. 2 Variation in A 232Th+ signal and B ThH+ :Thf as a function of Th concentration for PN 1.6 c I v) v) + 5 1.4 0 ul 8.0 7.0 v) I 6.0 I 0.8 I 1 I I I J 260 265 270 275 280 285 290 Temperature/K Fig. 3 of spray chamber temperature for PN Variation in A 232Th+ signal and B ThH+:Th+ as a function 1.6 5.0 I v) v) 0 .- 1.4 - 1.3 - 1.2 - 1.1 - ____....' . .. ._ ._ . . . . .. I w U 7 I I I I I I 2 1.01 1.0 x 1. 420 430 440 450 460 470 480 490 - Fig. 4 Variation in A 232Th+ signal and B ThH+:Th+ as a function of (a) heating coil temperature and (b) condenser temperature for USN the double-pass spray chamber.The data shown in Fig.4 which are for USN were obtained by independently varying the temperature of the heating coil [Fig. 4(a)] and the tempera- ture of the cooling water entering the condenser [Figure 4(b)]. Fig. 3 shows that with pneumatic nebulization the atomic ion signal increased as spray chamber temperature was reduced. Fig. 3 also shows that the hydride atomic ion ratio decreased as spray chamber temperature was reduced. The observed increase in atomic ion signal has been attributed elsewhere to condensation of water vapour in the spray chamber6 and a subsequent change in the optimal axial sampling position.' Removal of water vapour has also been shown to reduce the abundance of various oxygen-derived polyatomic ions (e.g. ArO') in ICP mass It stands to reason that removal of water vapour (and thus oxygen) is1226 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 (a) the cause of the reported reduction in oxide ions therefore removal of this water vapour (which will also remove hydrogen) should decrease the hydride atomic ion ratio as observed. Marshall and Hieftjeg have also shown that gas kinetic tem- peratures increase as water is removed from the plasma; this effect may further reduce the hydride:atomic ion ratio by shifting the equilibrium for hydride formation (see the preced- ing discussion) in favour of atomic ions. The data in Fig. 4 for ultrasonic nebulization are similar to those in Fig. 3 in that conditions which reduced plasma water loading increased atomic ion signal and reduced the hydride:atomic ion ratio.Note that USN conditions which gave maximum atomic ion signal gave lower hydride:atomic ratios than those measured under optimal PN conditions. Fig. 4 also indicates that the influence of condenser temperature is less significant than that of heater temperature. This obser- vation is consistent with the work of Tsukahara and Kubota," in which "Co+ and 138Baf signals BaO+ :Ba+ and ArO+ :Co+ were found to be insensitive to condenser tempera- ture up to ~ 2 8 0 K. - - - - Sample uptake rate Fig. 5 shows the influence of sample uptake rate on atomic ion signal and hydride atomic ion ratio for PN [Fig. 5(a)] and USN [Fig. 5(b)]. In the case of PN atomic-ion signal passed through a maximum and the hydride atomic ion ratio increased as the sample uptake rate was increased.The decrease in atomic-ion signal at uptake rates above 1 ml min-l may be exaggerated by negative drift in the analyte signal; however the trend observed here is consistent with that reported else- where.l' The decrease in atomic-ion signal at higher uptake rates is presumably due to increased water l ~ a d i n g ~ . ~ which would offset any increase in the analyte mass transport rate (ix. mass of analyte per unit time). Increased water loading 6.0 5.0 4.0 3.0 T *- -I 2.0 1 I 0 8 1.0 I I I I I I ? 1.0 x e 0.4 0.6 0.8 1.0 1.2 1.4 1.6 - - (. 2.2 L cn v) .- 2.1 N z 2.0 1.9 1.8 1.7 1.6 ( b ) 3.5 > t + I r F 3.0 2.5 2.0 I I I I 1 I 5 . .- 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Solution uptake/mI min-' Fig. 5 Variation in A 232Th+ signal and B ThH+:Th+ as a function of sample uptake rate for (a) pneumatic nebulization and (b) USN would also increase the hydrogen number density and decrease gas kinetic temperature thereby increasing the hydride:atomic ion ratio as observed.In the case of USN [Fig. 5(b)] the atomic ion signal quickly rose to a maximum which was insensitive to further changes in uptake; however the hydride:atomic ion ratio continued to increase (albeit modestly) as uptake rate was increased. The trends in atomic-ion signal and hydride:atomic ion ratio shown here are roughly consistent with the trends in analyte and solvent mass transport reported by Tarr et a1.12 using a home- made USN. A slight decline in atomic ion signal was expected at flow rates above the maximum; however in this case the decline may have been obscured by positive drift in the analyte signal.Nebulizer argonJlow rate Fig. 6 shows the influence of nebulizer argon flow rate on atomic ion signal and hydride:atomic ion ratio for PN [Fig. 6(a)] and USN [Fig. 6(b)]. In both cases the changes in atomic-ion signal are consistent with parameter behaviour reported el~ewhere.'~.'~ Furthermore the increase in the hydride:atomic ion ratio shown here is consistent with the behaviour of oxide This is understandable given the likely relationship between water oxide ions and hydride ions. Reproducibility of Hydride Measurements Table 4 shows the within-day and day-to-day reproducibility of the hydride atomic ion ratios for 232Th and 238U measured using PN and USN. These data were gathered on two non- consecutive days for the conditions shown in Tables 1 and 3.The two days during which the PN data were obtained differed from those during which the USN data were obtained. The ratios given in Table 4 are not corrected for instrumental 0.78 0.80 0.82 0.84 0.86 0.88 0.90 + h 0 v) 1 2 1.6 .- 1.4 r4 1.2 1 .o 0.8 0.6 I 1 I I 1 0 0.68 0.70 0.72 0.74 0.76 0.78 0.80 0.82 Nebulizer argon flow/l min-' Fig. 6 Variation in A 232Th+ signal and B ThH+:Th+ as a function of nebulizer argon flow rate for (a) PN and (b) USNJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1227 Table 4 Within-day and day-to-day reproducibility of hydride atomic ion ratios 232ThH + .232Th + 2 3 8 ~ ~ + .238u + PN USN PN USN Day 1* (3.9k0.2) x lop5 (2.1 k 0 .l ) x (3.2k0.2) x (1.8k0.1) x lo-’ Day 1* (3.76k0.09) x lop5 (2.09 f 0.04) x lo-’ (3.0k0.1) x (1.7k0.1) x lo-’ Day 2* (4.0k0.2) x lo-’ (2.07k0.07) x (3.3k0.1) x (1.81 k0.05) x lo-’ Mean (2 day)? (3.9f0.2) x lo-’ (2.10 k0.07) x lo-’ (3.2f0.2) x (1.8 f O . l ) x lo-’ * Mean and standard deviation of five consecutive measurements each consisting of ten consecutive 30 s integrations. t Mean and standard deviation of 15 measurements taken over two non-consecutive days. mass bias which is typically 1-2% per m/z unit for isotope ratios determined by ICP-MS3 Concurrent determination of 235U:238U during our hydride:atomic ion ratio measurements indicated that the instrumental mass-bias was in the worst case c 11% per m/z unit. Therefore we assume that the hydride:atomic ion ratios presented herein are accurate to within & 11% relative to the ‘true’ ratio.The biases in the hydride:atomic ion ratios may be different from those in the isotope ratios since the hydride and atomic ions undoubtedly originate from different processes in the plasma however in the absence of a hydride:atomic ion ratio standard it is not possible to determine if such differences exist. For the 2 day means shown in Table 4 within experimental error (MH+:M+)Th/(MH+:M+)U was unchanged by the choice of nebulization technique. This observation was expected since the relative degree of hydride-ion formation between the two elements should be solely characteristic of the elements. In addition within experimental error (MH+:M+)PN/(MH+:Mf)USN (i.e. the quotient of the hydride:atomic ion ratios for PN and USN respectively) was equal for thorium and uranium. This observation suggests that high-efficiency desolvation (such as that used with the ultra- sonic nebulizer) was a critical factor in the reduction of the hydrides.Further desolvation e.g. by cryogenic means,I5 might reduce hydride-ion formation even further. However the required detection limits (RDLs) for minor actinides are extremely low e.g. 35 fg 1-l for 239Pu in water,16 while 238U concentrations are typically = 1 pg I-’ in water. Given the ratio of the 239Pu RDL to uranium concentration (i.e. 3.5 x it is unlikely that 239Pu in natural water samples could be determined without hydride interference even using cryogenic desolvation unless the uranium is removed from the sample by chemical separation.This is currently being investi- gated in our laboratory. The authors thank A. Essling F. Smith D. Graczyk and L. Smith for their work in preparing and characterizing the standards used in this study. The authors also acknowledge M. Erickson and J. Poppiti for their support of this work. This work was funded by the Laboratory Management Division of the Office for Environmental Restoration and Waste Management United States Department of Energy. Argonne National Laboratory is operated by the University of Chicago for the United States Department of Energy under contract number W-3 1-109-ENG-38. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 References Smith M. Wyse E. and Koppenaal D. J. Radioanal. Nucl. Chem. 1992 160 341.Garcia Alonso J. Thoby-Schultzendorff D. Giovanonne B. Koch L. and Wiesmann H. J. Anal. At. Spectrom. 1993 8 673. Russ G. and Bazan J. Spectrochim. Acta Part B 1987 42 49. Toole J. Hursthouse A. McDonald P. Sampson K. Baxter M. Scott R. and McKay K . in Plasma Source Mass Spectrometry eds. Jarvis K . Gray A. Jarvis I. and Williams J. Royal Society of Chemistry Cambridge 1990 pp. 155-162. Miller J. C. and Miller J. N. Statistics for Analytical Chemistry 3rd edn. 1993 Ellis Horwood PTR Prentice Hall Chicester pp. 46-50 and 58-60. Hutton R. and Eaton A. J. Anal. At. Spectrom. 1987 2 595. Jakubowski N. Feldmann I. and Steuwer D. J. Anal. At. Spectrom. 1993 8 969. Williams J. in Handbook of Inductively Coupled Plasma Mass Spectrometry eds. Jarvis K. Gray A. and Houk R. Chapman and Hall New York 1992 pp. 71-75. Marshall K. and Hieftje G. J . Anal. At. Spectrom. 1987 2 561. Tsukahara R. and Kubota M. Spectrochim. Acta Part B 1990 45 581. Jakubowski N. Feldmann I. and Steuwer D. Spectrochim. Acta Part B 1992 47 107. Tarr M. Zhu G. and Browner R. Appl. Spectrosc. 1991 45 1424. Zhu G. and Browner R. Appl. Spectrosc. 1987 41 349. Gray A. and Williams J. J. Anal. At. Spectrom. 1987 2 599. Alves L. Wiederin D. and Houk R. Anal. Chem. 1992,64 1164. General Radiochemistry and Radioanalytical Services Protocol Version 2.1 Part B EG&G Rocky Flats Golden CO 1991 p. 10. Paper 4102488 E Received April 27 1994 Accepted July 18 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901223

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1229 Determination of Mercury Isotope Ratios in Samples Containing Sub-nanogram Amounts of Mercury Using lnductively Coupled Plasma Mass Spectrometry Conny Haraldsson Benny Lyven and Peder Ohman Department of Analytical and Marine Chemistry University of Goteborg and Chalmers University of Technology S-412 96 Goteborg Sweden John Munthe Swedish Environmental Research Institute P. 0. Box 4 7086 S-402 58 Goteborg Sweden The determination of sub-nanogram amounts of mercury demands systems of high sensitivity. To determine isotopic ratios a large signal is required to obtain a high number of counted ions. It was found that the precision in the isotope ratio determination in samples containing less than 300 pg of mercury is limited by the counting statistics.With 300 pg of mercury the precision was 0.2% (relative standard deviation) in the 202Hg:199Hg ratio. The detection limit was lowered by concentration of mercury vapour on gold traps and the sensitivity of the instrument was increased 4-6-fold by addition of nitrogen or hydrogen to the central gas flow. Keywords lnductively coupled plasma mass spectrometry; mercury isotope ratios; stable isotopes; precon- centration on gold traps Because of its high toxicity mercury is an environmental problem. In 1989 there were 10300 Swedish lakes where pike contained more than 1 pgg-' of mercury (not recommended for human consumption).' A number of methods exist for determining the concentration of mercury in biological matter e.g.in fish the most common being cold vapour atomic absorption spectrometry after acid digestion.' The concentration of mercury in natural waters is normally below 5 ng I-'. At these concentrations a precon- centration step is required with most analytical methods. This includes a closed reduction purging system in which the mercury is converted into a volatile form by reduction to elemental mercury purged from the solution and transported by a gas stream into either a preconcentration device (usually a gold trap) and then into a detector or directly into a detector. Even with inductively coupled mass spectrometry (ICP-MS) with its low detection limits it is not possible to use the normal nebulizer spray chamber system because the detection limit is too high. This is caused by the low degree of ionization of mercury in a plasma (32%) and the long washout times.3 Further the mercury signal is split over several isotopes with the most abundant being 30%.The detection limit for mercury using the normal nebulizer spray chamber system was found to be approximately 50ng1-I. This agrees with the detection limit of 40ng1-1 reported by Powell et d3 Hence some system has to be applied in order to improve the detection limit when determining mercury in natural waters by ICP-MS. The systems presented to achieve this are based on the cold vapour method. When analysing 100ml samples of natural waters using a cold vapour-ICP-MS system without preconcentration on a gold trap the detection limit was 8 ~ g . ~ When using a gold trap Smith5 obtained a detection limit of 36 pg in a 200 ml sample.Quantification in the two studies mentioned was performed using isotope dilution. This is preferable whenever possible because of the ability to compensate for losses in sample pre- treatment a common problem when determining mercury. In recent years there has been growing interest in studying the environmental turnover of methylmercury species (MeHg). Methylmercury the most toxic form of mercury in the environ- ment is the predominant form of mercury in fish.6 Only in recent years have analytical methods been developed for determination of organic mercury species in natural waters soils and ~ediments.~-'' Methylmercury is present in all com- partments of the environment including air and precipitation. Methylmercury compounds can be determined using aqueous- phase ethylation followed by GC separation pyrolysis and detection as Hgo using cold vapour atomic fluorescence spec- trometry.'"' A further method using sulfydryl-cotton to con- centrate methylmercury has been de~cribed.~ This method has the advantage of being able to analyse a large volume of water.In order to understand the mechanisms when mercury enters the biological system it is not enough to know the concen- tration and speciation in biota and water. For studies of the mechanisms involved in transport uptake and methylation of mercury a tracer is of significant help. Until recently the only tracers available were radioactive mercury isotopes. In a mercury-accumulation study fish were exposed to radioactive '03Hg in various chemical forms i.e.methyl phenyl and inorganic." It is preferable however to use stable isotopes whenever possible and ICP-MS provides the opportunity to use stable isotope tracers. ICP-MS has been used to determine isotope ratios of elements e.g. Pb in order to study atmospheric transport processes.12 Osmium isotope ratios have been used to study different geological properties such as the geochemical signa- ture from a large meteorite in New Zealand boundary ~ha1es.I~ The addition of stable isotopes to the diet and determination of isotope ratios in tissue samples by ICP-MS have been used to study the turnover of magnesium in mice and zinc in h ~ r n a n s . ' ~ ' ~ The work presented here is a part of a project aimed at studying mercury transport in a forest ecosystem using I9'Hg as a tracer substance.The low concentration of mercury species found in natural waters makes it difficult to collect the amount of mercury needed to make an accurate and precise determi- nation of the 202Hg 199Hg isotopic ratio. Further the high toxicity of mercury limits the amount of 199Hg that can be added to an ecosystem. It is therefore important to develop a method capable of precisely determining the '02Hg '"Hg ratio in samples containing small amounts (<0.5 ng) of inorganic and organic mercury compounds. Experimental Reagents The mercury-199 isotope (91.95%) was acquired from Oak Ridge National Laboratory. Two different qualities of argon1230 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 were used standard grade and research grade (> 99.9996%). No difference was observed in using the two types of argon. Nitrogen and hydrogen were both of research grade. The mercury used for calibration was of Suprapur grade (Merck). Apparatus The ICP-MS instrument used was a VG Plasma-Quad 1. The instrument was updated with a fixed torch-box a new sampling interface (PQ 1 +) and new faster electronics. When a standard solution was aspirated using a Meinhart nebulizer and Scott- type spray chamber the sensitivity was about 50 MHz (ppm)-' for indium. The general operating parameters are given in Table 1. When used for mercury measurements the spray chamber- nebulizer system was disconnected and the gas flow from the gold trap was added to the aerosol carrier gas flow and then fed directly to the plasma torch. The purpose of this was to expose the gas to the smallest possible surface in order to reduce carryover problems.That mercury easily adsorbs on to glass walls has been described by Powell et aL3 The experimental system is shown in Fig. 1. To minimize the mercury background from the argon a cleaning trap was connected before the sample trap. The two traps were of identical construction and their assembly is described below. The sample trap was placed inside a furnace constructed by winding a resistance wire (width 2 mm 0.1 mm thick resistance 3.7R) around a Pyrex glass tube of length 50mm and i.d. 8 mm. Inside the furnace next to the sample trap was placed a chromel-alumel thermocouple. The signal from the thermo- couple was used to control the heating.The outlet from the sample gold trap was connected to a small T-piece connector thereby making it possible to add additional gases e.g. hydro- gen or nitrogen and to connect the mercury calibration source. The tubing was connected to a Y-piece connector thereby making it possible to vent enclosed air to the atmosphere which otherwise could extinguish the plasma. The gas was mixed with the aerosol carrier gas and fed directly into the plasma torch. Procedure for the Determination of Mercury Isotope Ratios Before determining the isotopic composition of the mercury in a sample it was transferred to a gold trap. The gold trap was then placed inside the furnace and fitted to the inlet and outlet tubing. Valve G was in the open position so that air in the Table 1 ICP-MS operating parameters Outer gas flow/l min-' Auxiliary gas flow/l min - Nebulizer gas flow/l min-' R.f. power/W Sampling cone orifice/mm Skimmer cone orifice/mm 13 1.3 0.8 1 .o 0.7 1300 Vent I Ar 100 ml min-' I Ar700 ml min-' / I / trap was vented to atmosphere to prevent it from disturbing or extinguishing the plasma (Fig.1). The gold trap was vented for 15 s the valve closed and the heating switched on. After a delay time of 20 s the acquisition began using the parameters given in Table 2. After the acquisition the sample trap was removed and the furnace was allowed to cool to 30°C before the next trap was inserted. Mercury Calibration System A system providing a constant flow of mercury vapour was employed (Fig. 2) which was similar to a system described previously.16 The calibration system was used for two main reasons.First it was used to study the properties of mixed gas plasmas in relation to dry and wet argon plasma. Second the system was used to optimize the instrument settings which is difficult to achieve by means of the transient mercury signal from a gold trap. The mercury vapour system consisted of a 125 ml flat-bottomed bottle containing a thin layer of elemental mercury on the bottom. The outlet of the bottle was con- nected to a peristaltic pump (Gilson Minipuls 2) with tubing (0.127 mm id.) and then connected directly to the argon flow. A thermistor was placed inside the bottle to measure the temperature because the vapour pressure is temperature depen- dent. This system supplied a known stable and easily varied amount of mercury vapour to the system.Gold Trap The final step of sample pre-treatment before determining the isotopic composition of the mercury was to collect the mercury from the calibration device or the mercury extracted from a sample. The gold traps used are capable of quantitatively collecting atomic mercury in a gas ~ t r e a m . ' ~ * ' ~ They consisted of 6mm 0.d. quartz tubes of 120mm length. The adsorbent used consisted of gold-coated crushed quartz glass pieces having a diameter of approximately 0.5 mm. Quartz-wool was used to keep the adsorbent in place. When mercury had been collected on the trap it was stable for several days. After heating the trap was ready to use again without further pre- treatment." Table 2 Acquisition parameters in the determination of mercury isotope ratios Dwell time per point/ms 1.24 Distance between points/u 0.0138 Acquisition time/s 70 Resolution/u Points per peak 21 0.8 at 5% peak height II 'CP-MS Ar 100 ml min-' Per i st a It ic Pump Fig.1 Experimental set-up A flow controller; B cleaning trap; C sample trap; D furnace; E chromel-alumel thermocouple; F temperature control unit; and G de-aeration valve Thermocouple A Fig. 2 Mercury calibration sourceJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 123 1 Methylmercury Determination of isotope ratios of methylmercury was carried out as described above except that a two step preconcentration procedure was used. The sulphydryl-cotton method described by Lee7 was followed by a GC separation methodg and the methylmercury was collected on a gold trap. The results from this experiment will be described elsewhere.Results The signal at m/z 202 from 18Opg of mercury and the temperature profile are shown in Fig. 3. To avoid collecting data from the background signal a delay time of 20s was used between turning the heat on and starting the acquisition. As can be seen in Fig. 3 the signal from mercury desorbed from the gold trap began after approximately 20 s. Addition of Other Gases to the Plasma It was found that the sensitivity of the instrument was surpris- ingly low when detecting mercury using a dry argon plasma. Table 3 shows the signal from liquid nebulization of 4 ng ml-1 mercury in 0.15mol 1-I nitric acid containing 10mg 1-' KMnO compared with the signal from a dry argon plasma with 300 pg min-' of gaseous mercury added.The counts per picogram of mercury reaching the plasma were calculated based on the assumption of a nebulizer efficiency of 2% compared with 100% efficiency when mercury was pumped from the calibration vessel. It is clear that the signal from the mercury reaching a dry plasma was significantly lower than the signal from the same amount reaching a wet plasma. It was decided to try adding hydrogen and nitrogen to the central gas flow because a number of results have been presented that showed an increased sensitivity when other gases were added to the argon plasma.20 The central gas flow was chosen because it was more convenient to add additional gases to this flow than to the outer or auxiliary gas flow.An experiment was carried out in which the gas to the central channel of the plasma was humidified by passing it through a vessel with 5 1 25000 350 - ,-.-...---..*.- I I 4 300 Time/s Fig. 3 Signal and temperature profiles obtained when heating a gold trap amount of mercury 180 pg Table3 Signal for mercury in 0.15 mol I-' HNO with 10mg 1-' KMnO Amount of Signal/counts Signal/counts Pg-' Nebulized liquid 4 ng ml-' 79 000 55 000 - 1 Conditions mercury Dry Ar plasma 300 pg min-' 26 000 2 800 of water containing 0.1 g 1-1 KMnO,. The sensitivity was similar to that observed when nitrogen was added to the plasma. However problems were observed owing to the forma- tion of water droplets in the tubing. The droplets were trans- ported to the torch and caused instability.The signal obtained from 300pgmin-' of mercury when adding hydrogen and nitrogen to the plasma is shown in Fig. 4. When changing the gas or gas flow rate the instrument was retuned to achieve maximum sensitivity with the additional gas. This mainly meant lowering the extract lens voltage with increasing flow rate of the additional gas. The observed signal enhancement was a factor of six for hydrogen added at 1%. With 0.5% nitrogen added the signal enhancement was a factor of four. The increase in sensitivity when hydrogen was added is similar to that observed in electrothermal vaporization ICP-MS by Shibata et They proposed that the increased sensitivity is due to an increased electron number density leading to electron impact ionization and increased ion energy which in turn leads to an increase in collision ionization.The higher ion energy in a mixed gas plasma or a wet plasma compared with a dry plasma is indicated by the lower optimum extraction lens voltage required when using a mixed gas plasma or a wet plasma. The optimum extraction voltages were - 170 V with a wet plasma -270 V with nitrogen added and - 390 V with a dry plasma. No significant difference in isotope ratio precision was observed between using nitrogen or hydrogen as the additional gas. When trying to optimize the forward r.f. power to the plasma the signal decreased by approximately 20% when a power of 1100 W instead of 1300 W was applied and no significant increase occurred when the power was increased to 1500 W provided that the instrument was retuned for maximum sensitivity.Acquisition Method It has been suggested that it is preferable to make many very fast scans instead of a few slow scans over the m/z range of interest.22 The idea behind this is that the scanning speed should exceed the speed of sensitivity fluctuations. In order to investigate this the dwell time and the number of scans were varied the results of which are presented in Fig. 5. No advan- tage with very fast scans can be seen. The results presented in Fig. 5 were calculated as the relative standard deviation of ten isotope ratio measurements with an acquisition time of 70s and m/z range 197.6-203.4 with 20 points per peak. The amount of mercury transported to the plasma was varied and 160 I 1 0 10 20 30 Gas flow/mI min-' Fig.4 Signal obtained for 300 pg min-' of Hg with various amounts of A hydrogen and B nitrogen added to the central argon flow (800 ml min-l)1232 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1.2 1.2 1 1 0 1.0 0.8 0.6 0.4 10 20 30 40 Signal/lo5 counts - A - - - A I 1 1 I Fig.5 Effect of scanning speed on the s (lo) of the isotope ratio (202 199). The s of ten isotope ratio measurements with an acquisition time of 70s m/z range 197.6-203.4 with 20 points u-l. Dwell time per point 0 l O ps; 0 2 0 ps; 0 4 0 ps; 0 8 0 ps; 4,160 ps; A 320 ps; H peak jump with the parameters described in Table 2; the solid line represents the s calculated from counting statistics no gas was used in addition to argon. The dwell time at each point was 10 20 40 80 160 and 320 ps.For peak jumping the dwell time was 1024 ps. No difference in precision was found between the fast and slow scans. The peak jumping parameters are given in Table 2. When counting ions having a Poisson distribution the stan- dard deviation is the square root of the number of ions counted. Assuming that the errors at both masses are independent the relative error in the mass ratio 202 199 can be calculated as L where s is the relative standard deviation and N, and Nzo2 are the number of ions counted at the two masses. This function is included in Fig. 5 to show the theoretically possible precision. As shown in Fig. 5 the influence from quadrupole scanning speed is small. The major source of error when determining isotope ratios of small amounts of mercury is the counting statistics. This emphasizes the importance of high sensitivity.When using peak jumping the number of counts at the m/z values of interest are about 17 times larger compared with scanning where a majority of the time is spent acquiring data at m/z values other than 199 and 202. It is concluded that it is highly preferable to use peak jumping when determining isotope ratios of small amounts of analyte (< 5 ng of Hg). Application When running ten standards consecutively the precision obtained using this method was approximately s,=O.2% (la) for standards containing 300 pg and 0.9% for 50 pg of mercury. The results in Fig. 6 represent a 1 d run. Gold traps filled with approximately 300 pg of mercury extracted from different soil samples treated with 199Hg were analysed.The instrumen- tal parameters are listed in Tables 1 and 2. An addition of 10mlmin-1 of hydrogen was made to the central gas flow. Approximately every third sample was a standard i.e. 300 pg of mercury transferred to a gold trap from the calibration apparatus. The results for the standards are mean ratio (202Hg lg9Hg)= 1.8421 s=0.00295. This shows that the measurement had a 4.1% mass bias. The mass bias was calculated as the ratio between the measured 202 199 mass 1.6 -1 t 0.8 A Fig. 6 202Hg lg9Hg measurements over a 1 d run of (H) standards and ( A ) samples for 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 ratio compared with the actual 202 199 mass ratio in a standard. This mass bias is a consequence of the construction of the instrument which is optimized to give a flat response when plotting the concentration (expressed as a mass per unit volume) versus mass.The result is that the sensitivity increases with mass when expressed in molar units. The results in Fig. 6 have been compensated for the 4.1 YO mass bias in the 202 199 mass ratio. The s ( l a ) value for the standards was 0.16%. The results show that the system described where less than 500 pg of mercury was preconcentrated on gold traps and hydrogen or nitrogen was added to the aerosol carrier gas is suitable tracer experiments using stable mercury isotopes. References SNV Report No. 3593 Swedish Environmental Protection Agency Stockholm 1989. Navarro M. Lopez M. C. Lopez H.and Sanchez M. Anal. Chim. Acta 1992 257 155. Powell M. J. Quan E. S. K. Boomer D. W. and Wiederin D. R. Anal. Chem. 1992 64 2253. Haraldsson C. Westerlund S. and Ohman P. Anal. Chim. Acta 1989 221 77. Smith R. G. Anal. Chem. 1993 65 2485. Westoo G. Acta Chem. Scand. 1966 20 2131. Lee Y. H. Int. J. Environ.oAnal. Chem. 1987 29 263. Lee Y. H. and Iverfeldt A. Water Air Soil Pollut. 1990 56 309. Bloom N. S. and Fizgerald W. F. Anal. Chim. Acta 1988 209 151. Horvat M. Bloom N. S. and Liang L. Anal. Chim. Acta 1993 281 135. Stary J. Kratzer K. Havlik B. PraSilova J. and HanuSova J. Int. J . Environ. Anal. Chem. 1980 8 189. Sturges W. T. and Barrie L. A. Atmos. Environ. 1989 23 2513. Lichte F. E. Wilson S. M. Brooks R. R. Reeves R. D. Holzbecher J. and Ryan D. E. Nature (London) 1986,322,816. Schuette S . A. Hartmann S. C. Ting B. T. G. and Janghorbani M. J. Nutr. Biochem. 1992,3 38. Friel J. K. Longerich H. P. and Jackson S. E. Biol. Trace Elem. Res. 1993 37 123. Hanna G. P. Haigh P. E. Tyson J. F. and McIntosh S. J . Anal. At. Spectrom. 1993 8 585. Braman R. S. and Johnsson D. L. Environ. Sci. Technol. 1974 8 996. Brosset C. and Iverfeldt A. Water Air Soil Pollut. 1989 43 147. Munthe J. Hqaldsson C. Lee Y. H. Parkman H. Ostlund P. and Iverfeldt A. in preparation. Lam J. W. H. and Horlick G. Spectrochim. Acta Part B 1990 45 1313. Shibata N. Fudagawa N. and Kubota M. Spectrochim. Acta Part B 1992 47 505. Furuta N. J. Anal. At. Spectrom. 1991 6 199. Paper 4/01 1441 Received February 24 1994 Accepted May 12 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901229

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1233 Electrothermal Isotope Dilution Inductively Coupled Plasma Mass Spectrometry Method for the Determination of Sub-ng ml-' Levels of Lead in Human Plasma Robert J. Bowins and Robert H. McNutt Department of Geology Mc Master University Hamilton Ontario Canada L8S 4M I An isotope dilution technique coupled with electrothermal volatilization and inductively coupled plasma mass spectrometric detection is reported for the quantitative determination of Pb in human blood plasma with a calculated limit of detection of 16 x g. Sam le preparation is simple requiring only addition of concentrated nitric acid and a measured amount of 2EPb spike in solution. Analysis time is typically 2.5 min the relative standard deviation is <2% (n = 5) calculations require only simple numeric manipulations and the isotopic ratio of the Pb isotopes can be monitored.The Pb concentration of a commercially available dehydrated human plasma was found to be 1.27&0.02 ng ml-' Keywords Lead; isotope dilution; electrothermal inductively coupled plasma mass specfrometry; serum The determination of Pb in biological material has been of increasing interest because it is an environmental pollutant with long lasting toxic effects. Attempts to model the kinetics of Pb distribution within the human body'-3 have been ham- pered by the lack of reliable data for Pb concentrations in plasma of healthy unexposed humans which have ranged between 0.02 and 14.5 ng ml-' (ref. 4). The determination of Pb in plasma at ultra-trace levels is made more difficult by the potential for contamination during sample collection preser- vation storage and preparation for analysis. Inductively coupled plasma mass spectrometry (ICP-MS) provides sufficient sensitivity for ultra-trace determinations and has been successfully applied to the analysis of biological However at ultra-trace levels ICP-MS analysis is often hampered by the need for dilution of the highly saline and protein-rich Electrothermal techniques can be applied to the analysis of ultra-trace elements in biological materials although the necess- ity of using matrix modifiers can lead to contamination prob- lems.The sensitivity of the electrothermal technique can be considerably enhanced by using an ICP-MS instrument as the analyte detector," especially when the analyte can be at least partially separated from a complex matrix by controlled heat- ing.The combination of ICP-MS with electrothermal volatiliz- ation (ET-ICP-MS) would appear to hold great promise in the analysis of trace elernent~'l-'~ in difficult matrices such as human plasma.16*17 Experimental Instrumentation The equipment consisted of a SCIEX Elan Model 250 instru- ment and a Perkin-Elmer Model HGA-2100 graphite furnace. The ICP-MS instrument was operated in the peak hopping mode monitoring the intensities at m/z = 204 206 207 and 208 with 1 measurement per peak 2 repeats per scan a measure- ment time of 30 ms and a dwell time of 110 ms for a total dwell time of 220 ms at each m/z during each second of analysis.In order to determine the optimal settings for the incident r.f. energy argon flow rate and lens settings for a dry plasma and to check operating parameters at the start of the day a small mass (x 10 mg) of metallic Pb was placed in the graphite furnace on a carbon boat. Heating to approximately 575°C resulted in a measured intensity at m/z 208 of %lo5 counts s-' which could be maintained indefinitely. The peak intensity data were stored off-line for later analysis. A summary of typical ICP-MS operating conditions appear in Table 1. Two modifications to the graphite furnace were made the sample hole in the graphite tube was sealed with a tapered Table 1 ICP-MS and graphite furnace settings ICP-MS Aerosol carrier gas flow rate Outer gas flow rate Intermediate plasma flow rate R.f.power Graphite furnace heating cycle Dry Anal y se Clean 1850 ml min-' 13 1 min-' 2 I min-' 1.2 Kw Ramp 20 s ambient to 350°C Hold 30 s at 350°C Ramp 30 s 350-900°C at 18°C per second Hold 10 s at 900°C 10 s at 2300°C plug of spectrographic grade carbon; and one of the quartz windows was replaced with a 6 cm long 12mm 0.d. Pyrex tube whose other end was connected to the ICP-MS via a 45 cm length of 6 mm i.d. Tygon tubing. The large diameter and volume of this exit tube provided a cooling chamber for the hot gases generated by the graphite f~rnace.'~ Samples are introduced by removing the remaining quartz window and carefully inserting a 20 pl volume directly onto the graphite tube wall with a pipette. The quartz window is then replaced and the graphite furnace heating sequence started.The argon carrier gas flow path has not been modified in any way from the original graphite furnace mode of operation. The carrier gas total flow rate is controlled at 1.85 1 min-' by a mass flow controller (Model FM4575; Linde division of Union Carbide). This flow rate is some 50% greater than is typical for the same instrument when analysing nebulized solutions but was found to be optimal for the graphite furnace-dry plasma combination. Reagents Concentrated (14 mol 1-') nitric acid was distilled at sub- boiling temperatures in quartz stills in a Class 100 clean room and stored in Teflon containers and the doubly distilled de-ionized water was from a Milli-Q (Millipore) purification unit. The Pb spike (99.708 f 0.004Y0~~~Pb; Oak Ridge National Laboratories) was made up to 100.00 ng 8-l in 1.4 mol 1-' nitric acid.The 1.4 mol 1-' acid was made by addition of 14 mol 1-' nitric acid to Milli-Q water its relative density was measured at 1.047 and it was found to contain no more than 10 pg g-' Pb solution by ICP-MS analysis using conventional nebulization techniques. Thus the reagent Pb blank is insig- nificant relative to the levels of Pb found in the plasma analysed. Standard solutions of various Pb concentrations1234 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 were made by appropriate dilution by mass of a certified 1000 yg ml-' Pb stock solution (ICP standard solution; SPEX Industries). Preparation of Blood Plasma The method was tested on a reconstituted human plasma (SeraChem Instrumentation Laboratory) a normal clinical chemistry control human serum which was purchased in dehydrated form for use as a reference material.This material which represents the pooled contribution of between 200 and 500 healthy individuals was considered to be an excellent representative of an average human plasma. Plasma samples of 1.00 ml volume were prepared for analysis in 6 ml polypropylene test tubes (FALCON brand Beckton Dickinson) by adding 0.300 ml of concentrated nitric acid and 0.100ml of 104.7 ngml-I z04Pb spike solution in 1.4mol 1-l nitric acid. The resulting solution is 3.1 mol I-' in nitric acid and a precipitate of nitrated proteins forms almost immediately. The sample is shaken for 30 s and then centrifuged for 10 min at 2000 rev min-'.The liquid portion is sampled in 20p1 aliquots and deposited directly onto the walls of the graphite furnace tube for analysis. Analytical Procedure The sample is dried at 350°C and during the 30 s heating ramp from 350 to 900 "C Pb is first detectable (> 3s of average background variability) above baseline at z 575 "C. The slow ramp speed through the analyte appearance temperature results in peaks of about 20 s duration (Fig. 1). The prolonged duration of the analyte peak allows as many as 100 measure- ments at each isotope m/z value to be made and enhances the accuracy and reproducibility of the analytical method. Additionally since analyte intensities are measured before the major interfering elements in plasma are volatilized there is very little change in transient signals in loading of the ICP-MS plasma which allows optimal conditions to be maintained during the period of analyte analysis.By the end of the 40s analytical heating interval intensity of the Pb at all m/z values is z 1% of maximum values. The temperature is then raised to 2300 "C for 10 s to clean the furnace for the next sample. 1 x 1 0 ~ I x I x 204Pb (spike) A I \ 40 60 80 100 Time/s Fig. 1 Pb intensity profiles (counts s-') at 1 s intervals during an analysis of reference serum (1.23 ng ml-' Pb) plus 10.34 ng ml-' '04Pb spike. The smaller unlabelled traces are those of '''Pb and "'Pb; note the long duration of the analyte peaks and the log scale on the vertical axis At no time was the argon gas flow diverted nor was there venting of the sample during the drying or cleaning stages.Venting was not found necessary for a 10 h continued operation of the ICP-MS ( z 200 analyses). Analytical temperatures were taken from the front panel meter and verified by a digital thermocouple held inside the graphite tube during a typical analytical cycle. Isotope Dilution Method Common lead has four isotopes whose masses and abundances are 204 (1.4%) 206 (24.1%) 207 (22.1%) and 208 (52.4%).18 In isotope dilution analysislg a precisely known amount of a minor isotope of the element to be determined is added to the sample. The isotope dilution method is attractive since it is the ratio of two m/z intensities measured concurrently for the same element in the same sample that is used to calculate concentrations rather than the ratio of peak heights measured on consecutive samples.The isotope dilution method is there- fore quite tolerant of variations in sample size matrix composi- tion ion beam instabilities and fluctuations in instrument sensitivities between samples. Since intensities for all Pb isotopes are being measured the isotope dilution equation can be reduced to a two isotope form given by; where c is the concentration of analyte ck is the concentration of the spike A is the relative atomic mass of sample Pb Ak is the relative atomic mass of spike Pb ak and bk are the abundances of isotopes 'a' and 'b' in the spike a and b are the natural abundances of isotopes 'a' and 'b' and R is the experimentally determined ratio of a b. It is clear from this equation that once the spike has been added to the solution the concentrations calculated are independent of instrument sensitivity the amount of sample analysed or subsequent dilution.The determining factors in accuracy are the precision to which the concentration of the spike solution can be deter- mined and the accuracy of the measured ratio a b. Stock Pb solutions for the spike and standard additions experiment were made with 1.4 moll-' HNO with a measured relative density of 1.047. The errors in the method are minimized by weighing solutions containing Pb on a digital balance to three decimal places as a double check on the calibrated additions by pipette. The spike solution concentration uncertainty is < 1 YO and arises from the error associated with weighing mg amounts of '04PbC03 to five decimal places.The spike concentration was checked in a standard additions experiment. The uncertainty in measuring the a b ratio is fixed by the ICP-MS instrumental characteristics which usually has an RSD limit of about & 1%. There are statistical counting errors ( N N1/') inherent in the measured intensities but in the range 104-106 these errors are d 1%. A comprehensive discussion of the limitations of isotope dilution as applied to ICP-MS appears in ref. 20. Results Pb Contents of Average Human Plasma A typical intensity profile of an analytical run of a 201.11 sample of plasma reference sample appears in Fig. 1. The peaks are of much longer duration than those commonly encountered in electrothermal atomic absorption spectrometry 15-20 s com- pared with 2-3 s.This extended peak interval allows many intensity measurements to be made at each of the 4 isotope m/z values being monitored. To calculate the Pb concentration in the sample the sum of the intensity data for each m/z after subtraction of a background (the average of the 20 intensity measurements at each m/z prior to the first appearance of theJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1235 Pb peak) were used. This summation of m/z 206 207 and 208 assures that 298.6% of total plasma Pb is being directly analy sed. The concentration of plasma Pb was calculated using the sum of the net intensities for m/z 206 207 and 208 as representing one isotope i.e. 'a' = net intensity at 204 and 'b' = sum of net intensities at 206 207 208.Inserting known values for the relative atomic masses of common Pb (207.20) and spike Pb (204.01) spike mass in the sample (10.47 ng) the abundance of common (0.014) and spike (0.99708) '04Pb isotope and the abundance of common (0.986) and spike (0.00292) 206Pb + 207Pb + "'Pb isotopes into eqn. (1) results in ( 2 ) (0.9971 - 0.00292R) (0.98613 -0.014) [Pblplasma(ng) = 10.6235K where R is the ratio I 2 0 4 (Izo6+ IZo7 + 1208) where I is the net intensities for the indicated m/z values and K is an instrument calibration factor that incorporates the increase in sensitivity with increase in mass number and was determined from the standard additions experiment (see below). The plasma Pb was assumed to have the isotopic composi- tion of common lead but there is a possibility that actual plasma Pb could have significantly different isotope ratios.Since variation of Pb isotopes arises from the addition of Pb from the radioactive decay of Th or U to a common lead isotope the use of the sum of intensities of m/z 206 207 and 208 and using m/z 204 as the spiked isotope means that the plasma Pb actually measured is at least 98.6% of the actual amount and >98.6% for other plasma Pb. In the case of plasma lead with isotope ratios different from common lead the assumption of common Pb isotope ratios for the isotope dilution equation will yield data that are slightly low but deviations will be < 1 %. To test the effects of acidification on the Pb plasma concen- trations measured 1.00 ml samples were prepared by adding 0.100ml of the 1.4mol 1-1 204Pb spike solution and 0.100 0.300 and 0.600 ml of 14 mol 1-' HN03 resulting in samples which were 1.3 3.1 and 5.0 mol 1-I.The concentrations of Pb determined for these differently acidified plasma samples were 1.29 & 0.02 1.25 & 0.02 and 1.27 f 0.02 ng ml - ' respectively (2s n=5). These data all lie within the 2s error measurement plus the & 1% error for replicate sample preparation. This result confirms that the Pb acid blank is insignificant and that different amounts of acid have no discernable effect on the measured plasma Pb concentration. Addition of 0.100 ml of the '04Pb spike has an associated f 1% error whether deter- mined by the accuracy and reproducibility of the pipette volume (&1%) or determined by weighing the added spike solution to three decimal places ( f l least significant digit in this case & 1 %).This source of error could be lessened by using a larger sample size a more dilute 204Pb spike solution in 14mol 1-1 nitric acid or by more accurate weighing i.e. a four decimal place analytical balance would reduce this uncer- tainty by an order of magnitude. To evaluate procedural consistency and evaluate the instru- mental bias factor K a standard additions test was performed. To 1.00ml of plasma samples which had been acidified to 3.1 mol 1-1 and spiked with 10.47 ng of 204Pb were added 0 1.01 3.02 9.97 and 30.23 ng of Pb. The corresponding plasma Pb concentrations measured were 1.25 k0.02 2.34 k0.03 4.51 k0.07 12.17f0.10 and 34.09k0.19 ng ml-' respectively (2s n=5).Correlation analysis of Pb added to Pb measured for standard additions shows r2=0.999 with a slope of 1.0864 f 0.0016 and an intercept of 1.27 f 0.04. The non-zero intercept value for an addition of 0 ng of Pb is taken to represent the sample blank. This 'blank' Pb value of the reconstituted plasma represents the average serum Pb level of 200-500 adults plus any Pb due to processing dehydration packaging etc. The slope of the standard additions correlation (> 1.00) represents the sum of mass discrimination effects inherent in the ICP-MS instrument and this instrumen- tal bias is accounted for by the correction factor K = 1/1.0864 in the isotope dilution eqn. (2). The highly significant corre- lation coefficient suggests that to check for day-to-day changes in instrument sensitivity it is only necessary to analyse the most concentrated Pb plasma to verify the value for K.Discussion Kinetics of Pb in Plasma The success of this method is dependent on either the quantitat- ive transfer of plasma Pb into the acidified liquid portion of the sample or on the establishment of isotopic equilibration of both spike and analyte species between liquid and solid phases present before any is lost. Studies of the kinetics of Pb movement in plasma using labelled Pb radio tracer^"-^^ show that Pb passes quickly through the plasma into the red blood cells or into various organs with a residence half life of ,< 1 d. The Pb is thought to be held at protein binding sites probably at the termination of highly polar functional groups such as those containing an S atom or a -C=O which leaves it vulnerable to displacement by more active ionic species specifically the H + added during acidification.In fact some workers24 have extracted the plasma Pb using organic ligands followed by immiscible liquid separation techniques. These observations lead to the suggestion that the plasma Pb is not structurally fixed since the short residence time suggests that even in the buffered medium of the blood stream at pHz7.5 it is highly labile. Additionally when plasma samples were made increasingly acidic up to 5.0 moll-' in nitric acid there was no correspond- ing increase in measured Pb concentrations. This suggests that if there was a pH-controlled solid-liquid equilibrium between Pb in the liquid portion and the precipitate formed on acidifi- cation the position of such an equilibrium is well towards the liquid side when the plasma is 1.3 mol I-' and even more so at 5.0 mol 1-I.The 1.27 ng ml-1 value for human plasma Pb reported here is bracketed by the somewhat higher result of 1.7 ngml-I reported by Mauras et a1. who analysed diluted whole plasma by conventional pneumatic nebulization and the slightly lower value of 1.1 ngml-I of Manton and who destroyed the organic fraction by digestion in a high temperature sealed bomb before measuring the Pb by isotope dilution analysis. These differences in plasma Pb levels are most likely a reflection of different plasma sources rather than methodological errors since the short residence time of Pb in plasma makes it susceptible to recent exposures.The value of 1.27 ngml-I is significantly lower than the 5.4 ng ml-1 of Ca~alleri,~' the 4.0 ng ml-' of Gercken and Barnes26 and the 2.7 ng ml-' of DeSilvaZ7 and this scatter of values serves to highlight the difficulty of analysing Pb at the low concentrations found in blood plasma. Behaviour of Pb During Heating in the Presence of Organic Matter The organic fraction of the biological fluids may act as a matrix modifier by providing nucleation sites for gaseous Pb generated within the graphite furnace. Ediger and BeresZ8 have demonstrated that at very low analyte levels matrix modifiers are almost essential to successful analysis and that NaCl and hydrocarbons both major constituents of plasma can serve as useful modifiers.The temperature at which plasma Pb was detected during ramp heating was similar to the temperature of z 575 "C used to heat metallic Pb during the instrumental set-up procedure. This suggests that Pb vapour is being generated from the sample by reduction of plasma Pb to metallic form either on the surface of the graphite tube or in the matrix of pyrolytically degraded proteins. Since the amount1236 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 of NaCl volatilized at T<900"C is small while proteins are severely degraded the long duration of the Pb generated in the graphite tube during heating probably reflects a process of continuous breakdown of various species of hydrocarbons over a range of temperatures.Sensitivity of the Method The net intensity of the '04Pb peak is the sum of the interpolated count rate for each 1 s interval during the time of analysis. For 5 consecutive analyses of 2 0 4 of a typical spiked and acidified (3.1 moll-l) plasma sample the average net intensity was 5.68 k0.28 x lo5 counts which represents an instrumental response of 3.83 x lo3 counts per g. The standard devi- ation (2s) of the average background at m/z 204 for 20 measurements previous to the first appearance of the Pb peak was 20 counts for a typical analytical run. Therefore the limit of detection at 3sb is 60 counts or 16 x 10-l' g and the limit of quantitation at 1osb is 53 x 10- l5 g at m/z = 204 with similar limits existing for the remaining m/z. These limits correspond to plasma concentrations of about 0.001 and 0.003 ngml-l respectively which are much lower than the concentrations of Pb found in human plasma.Conclusions The described method has been found to give consistent and reproducible results for total Pb levels determined to be 1.27 ng ml-I for a commercially available human plasma. The viability of the method apparently depends on the presence of hydrocarbons and possibly NaCl that are an integral part of the blood plasma itself. These may act as matrix modifiers that greatly enhance the transport efficiency of a volatilized Pb sample from the graphite furnace chamber to the ICP-MS. The use of isotope dilution means that plasma samples can be prepared for analysis by a single addition of a suitable acid + spike mixture. The described method requires minimal sample preparation short analytical times simple concen- tration calculations and was found to be very sensitive repro- ducible and free from interferences. We gratefully acknowledge Jim McAndrew for technical assist- ance Dave Chettle and Colin Webber for their critical reading of the manuscript and financial support from the International Lead Zinc Research Organization and McMaster University.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 References Marcus A. H. Environ. Rex 1985 36 473. O'Flaherty E. J. Toxicology and Appl. Pharmacol. 1993 118 16. Leggett R. W. Environ. Health Perspect. 1993 101 598. Versieck J. and Cornelis R. Trace Elements in Human Plasma or Serum CRC Press Boca Raton Florida 1988.Beauchemin D. McLaren J. W. and Berman S. J. Anal. At. Spectrom. 1988 3 775. Delves H. T. and Campbell M. J. J. Anal. At. Spectrom. 1988 3 343. Mauras Y. Premel-Cabic A. Berre S. and Allain P. Clin. Chim. Acta 1993 218 201. Vanhoe H. Dams R. and Versieck J. J. Anal. At. Spectrom. 1994 9 23. Vanhoe H. Vandecasteele C. Versieck J. and Dams R. Anal. Chem. 1989,61 1851. Trentini P. L. Ascanelli M. Zanforlini B. Venturini F. and Bucci G. J. Anal. At. Spectrom. 1993 8 905. Park J. Van Loon J. C. Arrowsmith P. and French J. B. Anal. Chem. 1987,59 2191. Park C. J. and Hall G. E. M. J. Anal. At. Spectrom. 1987 2,473. Osborne S. P. Appl. Spectrosc. 1990 44 1044. Shen Wei-Lung Caruso J. A. Frickle F. L. and Satzger R. D. J. Anal. At. Spectrom. 1990 5 451. Sturgeon R. E. Willie S. N. Zheng J. Kudo A and Gregoire D. C. J. Anal. At. Spectrom. 1993 8 1053. Newman R. A. Osborne S. and Siddik Z. H. Clin. Chim. Acta 1989 179 191. Whittaker P. G. Lind T. Williams J. G. and Gray A. L. Analyst 1989 114 675. Handbook of Chemistry and Physics ed. Weast R. C. CRC Press Cleveland OH 1972. Webster R. K. in Methods in Geochemistry eds. Smales A. A. and Wager L. R. Interscience New York 1960 ch. 7 pp. 202-246. van Heuzen A. A. Hoekstra T. and Wingerden B. J. Anal. At. Spectrom. 1989 4 483. Castellino N. and Aloj S. Br. J. Ind. Med. 1964 21 308. Hursch J. Schraub A. Sattler E. and Hoffmann H. Health Phys. 1969 16 257. Chamberlain A. C. Heard M. J. Little P. Newton D. Wells A. C. and Wiffen R. D. Investigations into Lead from Motor Exhausts United Kingdom Atomic Energy Authority report R9198 HMSO 1978. Manton W. I. and Cook J. D. Br. J. Ind. Med. 1984 41 313. Cavalleri A. Minoia C. Pozzoli L. and Baruffini A. Br. J. Ind. Med. 1978 35 21. Gercken B. and Barnes R. M. Anal. Chem. 1991 63 283. DeSilva P. E. Br. J. Ind. Med. 1981 38 209. Ediger R. D. and Beres S . A. Spectrochim. Acta Part B 1992 47 907. Paper 3/07064F Received November 29 1993 Accepted June 14 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901233

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1237 Electrothermal Vaporization for Simultaneous Multi-element Determination* Erwin Hoffmann Christian Ludke and Horst Scholze lnstitut fur Spektrochemie und angewandte Spektroskopie Laboratorium fur spektroskopische Methoden der Umweltanaiytik Rudo wer Chaussee 5 12489 Berlin Germany New techniques have been developed for analysis of microsamples. A modification of a graphite furnace for furnace atomization non-thermal excitation spectrometry (FANES) was used in electrothermal vaporization inductively coupled plasma mass spectrometry. The figures of merit such as the maximum mass which can be introduced into the plasma (25 pg) the detection limits of a number of elements (Ag 0.02 pg; Co 0.3 pg; Mn 0.2 pg) in concentrated nitrogen acid the dynamical range of calibration curves and the influence of sodium on analyte signals are investigated and compared to the ones obtained for FANES.Keywords Electrothermal vaporization; inductively coupled plasma mass Spectrometry; furnace atomic non-thermal excitation spectrometry; matrix interference; detection limits In analytical research the demand exists for simultaneous multi-element determination in the picogram and sub- picogram range. With decreasing absolute amounts to be determined the risk of systematic error L +0v5 was the first to apply electrothermal vaporization (ETV) to atomic absorption spectrometry (AAS). The graphite furnace AAS (ETAAS) is the most sensitive established atomic spectro- metric method today with detection limits for a majority of elements in the pg dm-3 range and the sample volume at the 10-50 mm3 level.Because of its simple instrumentation it should be the best method for element determination in microsamples. The ETAAS method can be used for a broad variety of problems however many interferences and matrix effects are found arising from the thermochemical processes in the furnace which overlap in time with the sample evaporation molecule dissociation and radiation absorption. Therefore for many analytical problems the analytical precision and accu- racy are insufficient. Improvements can be made by introducing the stabilized platform technique in combination with Zeemann background correction. However one disadvantage of ETAAS has remained. The fact that AAS is fundamentally a single- element method limits its practical applications.Atomic emis- sion spectrometry (AES) inherently has a high multi-element capability which has been the main reason in developing the inductively coupled plasma (ICP) in combination with AES for simultaneous multi-element determination since the mid 1970s. Pneumatic nebulization is commonly used for aerosol production in routine analysis. Efforts have been made to improve the performance of ICP-AES by using alternative methods of aerosol production techniques to reduce the detec- tion limits. Electrothermal vaporization (ETV) is a procedure with a high efficiency of aerosol generation that can be easily obtained as opposed to pneumatic nebulization. In combination with ICP-AES ETV provides an increase in detection power for simultaneous multi-elemen t determination in small volume samples.Spectral interferences are reduced because the solvent and volatile matrices are vaporized before the analyte. Many applications have been reported with modified versions of commercially available atomic absorption equipment used in combination with ICP-AES. Tungsten loop graphite yarn and graphite rod are also used for vaporization of microsamples. Broekaert and Tolg6 and Broekart and Boumans7 have dis- cussed the various approaches. Matusiewicz' reviewed crucial aspects of ETV-ICP-AES in 1986. Sources with a plasma at reduced pressure such as hollow * Presented at the XXVIII Colloquium Spectroscopicurn Internationale (CSI) York UK June 29-July 4 1993.cathode discharge have been enjoying considerable interest for spectrochemical analysis. These plasmas are characterized by an absence of local thermal equilibrium. The electron tempera- tures greatly exceed the gas temperatures. Electrons with high energies are available and excite lines with high excitation energies while the spectral background is low. Therefore the detection limits are lower compared with sources in local thermal equilibrium. A graphite furnace with an additional low pressure discharge applied for analyte excitation was realized in FANES.9 The volatilization takes place in the furnace as in ETAAS and the sample analytes are excited in a low pressure hollow cathode discharge burning inside the graphite tube. Detection limits for simultaneous multi-element determination are obtained in the picogram range.However easily ionizable elements will disturb the low pressure discharge and cause strong matrix effects. Increasing the matrix content decreases the analyte intensities and influences the detection power. Mass spectrometry (MS) is the only multi-element method which permits the sequential or simultaneous determination of all elements with their isotopes and in combination with the ICP (ICP-MS) has detection limits for most elements in the 0.1-1 pg dm-3 range which is two orders of magnitude lower than for ICP-AES. High precision analyses can be made with aid of isotope dilution calibration. Gray and Date" were the first to interface an ETV device to an ICP-MS in 1983. Some years later this technique was applied to practical samples.Geological samples,",12 biological sample^,'^^'^ environmental sample^'^.'^ and sea-water17 have been studied. Several laboratories have investigated the transport loss. Park and Hall" reported matrix dependent sensitivity changes and were the first to attempt to optimize ETV for ICP-MS. Ediger and Steven" were the first to investigate the carrier character- istics of Na matrices and to advance the physical carrier concept in ETV-ICP-MS while Hall et all5 and Gregoire2O used various chemical modifiers for enhancing the analyte signals. Kantor2' gave the explanation of the physical mechan- ism underlying the reported analyte loss in ETV-ICP-MS. According to this explanation clusters arise at low analyte pressure and grow until they condense into particles.This process termed self-nucleation results at high vapour concen- trations in particles sufficiently large to form conglomerates and to deposit on surfaces inside the transport tubes. The construction of the electrothermal vaporizer has an important influence on the extent of transport loss. The vaporization of the analyte together with matrix can also cause suppression of the analyte signals. Gregoire et a1.22 showed that the absolute amount of material reaching the plasma is an important parameter limiting the analytical application of ETV-ICP-MS.1238 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 In this paper the effect of the absolute amount of Na on ETV-ICP-MS signals of a large number of elements is exam- ined.Simultaneously determined detection limits in concen- trated HNO are presented for a number of elements and the maximum amount of matrix to be introduced into the plasma is estimated. The results obtained by ETV-ICP-MS are com- pared with those obtained by FANES. The sample undergoes the same physical and chemical processes on the graphite surface during the drying and ashing stages if the same heating conditions are used for ETV-ICP-MS and FANES but the sample may undergo different interactions with the graphite surface during vaporization. The plasma of the discharge in the hollow cathode may influence the graphite surface so that the active sites may differ considerably within the two systems. Experimental Equipment The ICP-MS system used for this work was a Perkin Elmer SCIEX Elan 5000 (Norwalk CT USA).The ETV device was constructed and built in our laboratory with a similar design to that used for FANES. The ETV unit is connected directly to the plasma torch without an additional sample conduit and valve as depicted in Fig. 1. Gas flows and sample pre-treatment are specially designed for ICP-MS. Sample volumes of 5-50mm3 are deposited into the graphite tube through a dosing hole and argon flows from both tube ends during the drying and ashing stages. While the tube is slowly heated the solution vapours are purged out of the dosing hole. During the pre-treatment stages the ICP is only exposed to dry argon. One gas inlet is radially situated between the sample torch channel and furnace while the other one is axially directed to the graphite tube (see Fig.1). After the sample has been pre- treated the gas inlet between the torch and graphite furnace is closed while the central gas flow is maintained. The dosing opening is automatically closed by a lid and the central gas inlet provides a stable argon flow through the graphite tube. The dried and pyrolysed sample is subsequently vaporized into the flowing gas stream to be carried to the ICP-MS. The FANES studies were carried out using a FANES device in conjunction with an echelle polychromator constructed and built in our l a b ~ r a t o r y . ~ ~ Details of the equipment and its operation are reported by Liidke et al.24 The sample is deposited into the graphite tube dryed and ashed in the same way as in ETV-ICP-MS. Then the dosing opening is closed the FANES device is pumped out argon is filled in and the hollow cathode discharge is ignited.After the dryed and ashed sample has been vaporized according to a special temperature programme the analyte atoms are excited by electrons of the discharge. The intensities of six element lines are measured simultaneously by the detection system of the echelle poly- chromator. Background correction is made by wavelength modulation using a swinging quartz refractor plate mounted behind the entrance slit of the spectrometer. 1- 9 2 3 1 4 / lr' I Fig.1 Coupling of the FANES graphite furnace to the Elan 5000 ICP-MS 1 graphite tube; 2 graphite electrode; 3 lid; 4 gas inlet (0.9 dm3 min-'); 5 intermediate gas; 6 outer gas; 7 plasma torch; 8. interface; 9 sampler cone Reagents Nitric acid (5% v/v) samples were prepared in polyethylene bottles.Multi-element standard solutions were used to spike the HNO with the analytes In Ag Zn Pd Cr Cu Al Pb Cd Co Ni Mn Mg Na S K and Ca. Two sets of samples were spiked with element concentrations of 0.3 1 3 10 30 100 and 300 pg drn-,. One of the two sample sets were used as standards for the calibration in ETV-ICP-MS and FANES and the other was used to investigate the influence of increasing Na content on the analyte signals. Concentrated nitric acid of sub-boiling distilled quality was used for measuring the detec- tion limit. Comparison of ETV-ICP-MS and FANES The FANES and ETV-ICP-MS are methods which use the same technique for converting a sample into vapour. Samples for either technique undergo the same chemical processes at the graphite surface.However performance of FANES is related to complete atomization of the sample while for ETV- ICP-MS the complete transport of the analyte whether atom- ized or not is essential. The sensitivity of FANES depends on the residence time of atoms within the hollow cathode dis- charge. The transport of the analytes has to be rapid and efficient for maximum sensitivity of ETV-ICP-MS. The vapor- ization process is temporally and spatially separated from the ionization source for ETV-ICP-MS but all vaporization and excitation processes occur in the same volume at the same time for FANES. Therefore the flexibility of the parameters regulating analyte vaporization and ionization is considerably greater for ETV-ICP-MS.A further advantage of ICP-MS is the fact that not only elements are detected but also isotope ratios can be calculated from the relative intensity determinations. Results and Discussion Signal Profile The Elan 5000 was used for measuring 13 elements using ETV- ICP-MS. The vaporization of picograms of analytes produces rapid transient signals requiring high temporal resolution for quantification; 10 ms was chosen for the dwell time (Table 1). Typical peak-profiles are shown in Fig. 2. The peak widths are for most of the isotopes about 0.5 s at half height. However some isotopes display double peaks caused by chemical or physical reactions while the vaporization process is running (e.g. Co in Fig. 2). The parameters were optimized in such a way that the ion intensities are measured quickly enough for capturing the full transient signal profile and long enough to minimize the statistical signal noise.The ETV-ICP-MS analyte signals are measured one after the other with a high repetition rate. This is different from FANES where the measurements are truly simultaneous. Typical FANES intensity signals are also shown in Fig. 2. Calibration Curve The most important parameters of the vaporization cell are given in Table 1. For ETV-ICP-MS quantitative amounts of the vaporized sample must be transferred to the plasma by the carrier gas. The FANES atomizer was directly connected with the central channel of the ICP torch without gas tubing and valve. Studies were made of the transport loss in this ETV- ICP-MS equipment.A problem in considering transport loss is how to recognize that loss is occuring. Kantor21 noticed that the degree of loss may depend on the vaporized analyte mass and that this behaviour may then result in non-linear calibration curves. Calibration curves were obtained simul- taneously for a number of elements with multi-element analyte solutions without a concomittant matrix. The non-linearity isJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 Table 1 Parameters related to instrumentation 1239 Parameter Vaporization cell Graphite tube Drying and ashing gas flow Drying temperature/”C Ashing temperature/”C Vaporization temperature rate/”C s-’ Maximum vaporization temperature/”C Multi-element capability Signal shape Background correction Signal measurement Measurement parameters during high temperature gas flow Elements measured simultaneously Generator frequency/MHz Plasma power/kW Auxiliary gas flow/dm3 min-’ Coolant gas flow/dm3 min- ’ Nebulizer type Nebulizer gas flow/dm3 min-’ Dwell time Replicate measurements Sweeps/reading ETV-ICP-MS FANES-furnace Diameter 6 mm length 28 mm pyrolytically-coated specially designed 0.9 dm3 min-’ 100 300 2000 2700 Fast sequential 15 elements Peak 0.5 s at half height None Peak area integration Dwell time 10 ms sweeps/reading 1 In Ag Zn Pb Cd Co Ni Mn Mg S number of replicates 200 Pd Cr Na K Ca ICP-MS 40 1.0 1 15 Meinhard 1 50 ms 1 220 FANES FANES-furnace Diameter 6 mm length 28 mm pyrolytically-coated same as in AAS Low pressure 40 hPa 100 300 1500 2200 Simultaneous 7 elements Peak 0.2-1 s at half height Wavelength modulation Peak area integration 10 ms on spectral line 10 ms beside 1st series Ni Cr Cu Al Mg Mn 2nd series Zn Pb Ag Cd Ca K 3rd series Na Co In spectral line ICP-AES 27.12 1.1 1 14 Cross flow 1 40000 30000 r I u) v) c = 20000 s 5 1.m c 0 10000 0 (a) 2500 I 2000 > c .- 1500 + .- a (D a m .g 1000 - 500 1 2 3 4 5 0 1 2 3 4 5 6 Tim e/s Fig.2 (a) Temporal signal profiles of ETV-ICP-MS produced by vaporization of 1.5 pg of 59C0 and 55Mn and (b) temporal signal profiles of FANES by vaporization of 100 pg of Ca (393 nm) Mg (279 nm) and Ag (328 nm) evident in the calibration graphs for most elements. The degree of curvature varies among the elements. The Pd calibration curve showed the largest curvature of the elements listed in Table 1.On the other hand the plot of S signals versus vaporized mass was linear. The transport of S as gaseous molecule such as SO would explain this behaviour. It is also possible however that S may react with gaseous carbon to form particulates or is adsorbed on the carbon particulates. All elements investigated by FANES had linear calibration graphs for small sample amounts. Detection Limits The detection limits obtainable by ETV-ICP-MS under multi- element conditions are considerably better than typical values reported for ETAAS. However for many elements even better detection limits can be obtained if the operating conditions for the plasma ion optics and mass analyser are optimized exclus- ively for one element. In practice detection power is poorer in the presence of a matrix particularly one producing mass interferences.In our investigations detection limits based on 3sb criterion were simultaneously determined for a number of elements in concentrated nitric acid. The blank signals were measured of the sub-boiling concentrated nitric acid using both ETV-ICP-MS and FANES. The number of measurements used for the calculation of the standard deviation of blank signals (sb) was 15. Three elements namely lead magnesium and cobalt were used to determine optimal conditions in the ETV-ICP-MS. Platinum skimmer and sampler cones were used for these investigations. The detection limits obtained are listed in Table2 and are compared with the values measured using FANES and automatic background correction.A maxi- mum of six elements could be determined simultaneously by FANES. Therefore two series of measurements were made. Although ETV-ICP-MS has a detection power superior to Table 2 obtained by ETV-ICP-MS and FANES Detection limits (DL) of elements in concentrated nitric acid FANES ETV-ICP-MS Element In Zn Pb Cd c o Ni Mn Mg Na S K Ca Ag Wavelengthlnm DL/pg 352 328 213 405 228 240 352 403 279 588 766 393 - 30 3 170 20 40 100 20 6 20 3 3 40 - Isotope 115 107 66 208 114 59 60 55 25 23 34 39 44 DL/pg 0.1 0.02 3 0.4 1 0.3 0.7 0.2 2 5 10 > 50 > 501240 JOURNAL OF ANALYTIC.4L ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 that of FANES there are some elements with lower detection limits in FANES as shown in Table 2. Matrix Effect Electrothermal vaporization typically involves three steps drying at low temperature intermediate temperature ashing during which matrix components can be partially volatilized and a high temperature volatilization.During each of these temperature steps many reactions occur which affect the analyt- ical result. It seems that FANES is more susceptible to certain analytical problems than ETV-ICP-MS. In FANES some matrices can cause losses during the ashing step and can influence the analytical results because of vapour-phase reac- tions between the analyte and matrix during the high tempera- ture step or give false results because the electron temperature decreases if low-ionizable elements are introduced into the discharge plasma while the sample is vaporized. As the ICP can decompose and vaporize most chemical compounds the influence of chemical reactions in the graphite furnace on the analytical results should be lower for ETV-ICP-MS than for FANES.On the other hand temperature suppression in the ICP due to matrix elements with low ionization energy can also occur. The effect of sodium on a large number of elements was studied using ETV-ICP-MS. It was found that all analyses are similarly influenced by increasing concentration of sodium chloride. This behaviour is shown in Fig. 3 for two elements the concentration of which was 1Opg dm-3 each (10mm3 sample volume). The effect of adding sodium enhanced the analyte intensities. These observations indicate that the sodium chloride matrix acted as a physical carrier of the analyte elements and caused a more efficient analyte transport between the vaporizer and ICP.Probably the vaporized sodium chloride condenses into particulates more rapidly than on cool surfaces of the vaporizer.21 The influence of sodium chloride on analytes using FANES also shows the same tendency for all elements but the signals are not enhanced with increasing matrix when measured using ETV-ICP-MS. Fig.4 shows for the four selected elements zinc lead silver and cadmium that the line intensities decrease with increasing sodium chloride concen- tration. This behaviour can be explained by suppression of the electron temperature caused by the introduction of sodium with its low ionization energy into the hollow cathode plasma. In Table 3 the influence of sodium load on analytical signals is listed for ETV-ICP-MS FANES ICP-MS and ICP-AES.The ICP-MS equipment used for these measurements was a Perkin Elmer SCIEX Elan 5000 which was also used for ETV- ICP-MS. Optical emission measurements were made with the ICP-AES system Spectroflame (Spectro Kleve). The param- eters of FANES and ICP-AES are given in Table 1. Electrothermal vaporization produces a short transient pulse as shown in Fig. 2. The sample mass vaporized in the electrothermal vaporizer of both ETV-ICP-MS and FANES may change considerably during the evaporation pulse if the heating rate is high. There is a largest mass analyte and matrix together for stable burning of a plasma. The curves in Fig. 3 show that the ion intensities of the analytes decrease rapidly before the ICP becomes unstable by increasing sodium amounts and the discharge is extinguished.The point of maximum Na concentration in the two curves in Fig. 3 indi- cates the maximum amount mg of sodium chloride that can be tolerated in ETV-ICP-MS which is found to be mg ~ 2 5 pg. Investigations to determine the maximum mass that can be introduced into the plasma of the FANES source have been reported earlier.25 There the maximum mass mg of sodium chloride was derived from intensity curves as shown in Fig. 4 according to the analyte signal being depressed by 10%. The 3.5 ~ 3.0 - .- I 0.4 1 U 0.0001 0.001 0.01 0.1 1 10 100 1000 10000 [Nal/mg dm-3 Fig.3 The influence of sodium chloride on A 55Cr and B 66Ni obtained by ETV-ICP-MS. Eight measurements were carried out at each Na concentration.The RSDs of the averaged relative ion intensit- ies were 3 to 5.5% for Cr and 3 to 6% for Ni 0.2 ' I I I I I 0.01 0.1 1 10 100 1000 [Nal/rng dm ~ Fig.4 The influence of sodium chloride on the determination by FANES of A Zn (213 nm); B Pb (405 nm); C Ag (328 nm); and D Cd (228nm). Eight measurements were carried out at each Na concentration. The RSDs of the averaged relative ion intensities were 4 to 7% for Zn 5 to 7% for Pb 2 to 4% for Ag and 3 to 5% for Cd Table 3 Influence of Na on analyte signals including the data on the N.a concentrations for which the analyte signal is depressed or raised by 10%. Concentrations of analytical elements in 5% v/v HNO, 100 pg dm-j (10 mm3 sample volume) for FANES; 100 pg dm- for ICP-MS; 1 mg dm- for ICP-AES; 10 pg dm-3 (10 mm3 sample volume) for ETV-I[CP-MS FANES ICP-MS ICP-AES ETV-ICP-MS Element Zn Cd Ni c o c u Cr Pb Mg Ag M n A/nm 213 328 326 341 346 324 359 405 403 280 Nal mg dm- 30 40 40 900 1000 1400 1800 40 900 1800 Na/ Isotope mg drrC3 66 100 107 1000 114 500 - - 65 5000 208 5000 - - Na/ Ajnm mg dm- 213 328 226 23 1 228 324 267 220 257 280 1000 10000 5000 7000 5000 lo000 6000 5000 lo000 10000 Isotope 66 107 114 60 59 52 208 - Na/ mg dmP3 1 1 1000 10 2000 10 1 - -JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1241 finding rn,zl.3 pg shows the remarkable difference between ETV-ICP-MS and FANES 13 Park C. J. Van Loon J. C. Arrowsmith P. and French J. B. Anal. Chem. 1987 59 2191. 14 Whittaker P. G. Lind T. Williams J. G. and Gray A. L. Analyst 1989 114 675.J. Anal. At. Spectrom. 1988 3 791. Darke S. A. Pickford C. J. and Tyson J. F. Anal. Proc. 1989 The by the Senatsverwaltung fur Wissenschaft 15 Hall G. E. M. pelchat J. C. Boomer D. W. and Powell M. und Forschung des Landes Berlin and the Bundesministerium fur Forschung und Technologie is gratefully acknowledged. 16 1 2 3 4 5 6 7 8 9 10 11 12 References Tolg G. and Tschopel P. Anal. Sci. 1987 3 199. Tschopel P. Kotz L. Schulz W. Veber M. and Tolg G. Fresenius ' Z . Anal. Chem. 1980 302 1. Gretzinger K. Kotz L. Tschopel P. and Tolg G. Talanta 1982 29 1011. Tschopel P. and Tolg G. J. Trace and Microprobe Techn. 1982 1 1. L irov B. V. Spectrochim. Acta Part B 1984 39 159. Broekaert J. A. C. and Tolg G. in Inductively Coupled Plasma Emission Spectroscopy ed.Boumans P. W. J. M. John Wiley & Sons New York 1987 pt 2 vol. 90 pp. 421-458. Broekaert J. A. C. and Boumans P. W. J. M. in Inductively Coupled Plasma Emission Spectroscopy ed. Boumans P. W. J. M. John Wiley & Sons New York 1987 pt 1 vol. 90 p. 297. Matusiewicz H. J. Anal. At. Spectrom. 1986 1 171. Falk H. Hoffmann E. and Liidke C. Prog. Anal. Spectrom. 1988 11 417. Gray A. L. and Date A. R. Analyst 1983 108 1033. Date A. R. and Cheung Y. Y. Analyst 1987 112 1531. Parkand C. J. and Hall G. E. M. J. Anal. At. Spectrom. 1988 3 355. 17 18 19 20 21 22 23 24 25 26 379. Falkner K. K. and Edmond J. M. Anal. Chem. 1990 62 1477. Park C. J. and Hall G. E. M. Curr. Res. Geol. Survey Canada 1986 86 767. Ediger R. D. and Steven A. B. Spectrochim. Acta Purr B 1992 47 907. Gregoire D. C. Anal. Chem. 1990 62 141. Kantor T. Spectrochim. Acta Part B 1988 43 1299. Gregoire D. C. Lamourenx M. Chakrabarti C. L. and Maawali S . A. J . Anal. At. Spectrom. 1992 7 579. Schmidt K. P. Becker-ROB H. and Florek S. Spectrochim. Acta Part B 1990 45 1203. Ludke C. Hoffmann E. and Skole J. J. Anal. At. Spectrorn. 1994 9 685. Hoffmann E. in Proceedings of the CANAS '93 Colloquium Analytische Atomspektroskopie Oberhof March 15-19 1993 ed. K. Dittrich and B. Wetz Univ. Leipzig and UFZ- Umweltforschungszentrum Leipzig-Halle GmbH Leipzig 1993 pp. 229-232. Paper 31071 38C Received December 2 1993 Accepted May 25 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901237

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1243 Study of Element Distributions in Weathered Marble Crusts Using Laser Ablation Inductively Coupled Plasma Mass Spectrometry Katia Ulens Luc Moens and Richard Dams Laboratory of Analytical Chemistry Ghent University Institute for Nuclear Sciences Proeftuinstraat 86 B-9000 Ghent Belgium Stefaan Van Winckel and Leon Vandevelde SCK= CEN Nuclear Research Centre Boeretang 200 B-2400 Mol Belgium The authentication of marble objects can often be very difficult. As the presence of a naturally produced weathering crust on an artefact can be used as a means of evaluating its antiquity techniques to investigate those weathering layers can be very useful in authentication studies. In this study the depth profiles of Mg Al Mn Fe Zn Sr Ba La and Pb in weathered marble crusts were obtained using laser ablation inductively coupled plasma mass spectrometry.Ablations were performed in the free-running mode with a focused laser. In order to obtain spatial information the samples were analysed directly without any sample prep- aration. Signal drift was corrected by using the %a signal as an internal standard. By plotting the ratios obtained as a function of the depth below the surface significant trends were found. In order to check if quantitative results can be obtained standards produced by adding elemental standard solutions to CaCO powder were used. Accurate concentrations in the sampled micro-spots were obtained for both the minor and trace elements in the homogenized synthetic standards and also for the homogeneously distributed minor elements Mg and Sr in the marble samples.Keywords Marble; calcium carbonate; weathering; laser ablation inductively coupled plasma mass spec- trometry; depth profiling Inductively coupled plasma mass spectrometry (ICP-MS) is a powerful technique for trace and ultra-trace element determi- nation mainly in aqueous solution. Traditionally samples have been introduced into the plasma by using pneumatic nebulization. However the use of alternative sampling devices for direct introduction of liquids and solids into the ICP has also been investigated for many years. A variety of solid sampling systems are available today. Mainly five different systems have been described in the literature slurry nebuliz- ation,lp6 electrothermal vaporizati~n,~?~ direct sample inser- tion,’-’’ laser ablation ( LA)231’-14 and arc nebulization.15~16 The use of a laser beam for evaporating atomizing and ionizing a sample has been known for many years. Laser ablation as a means of sample introduction into an ICP mass spectrometer was first reported by Gray.17 He demonstrated the application of a ruby laser in combination with an ICP mass spectrometer to analyse geological materials. ArrowsmithI8 used an Nd:YAG laser for ablating ceramics metals polymers and glasses into the ICP. Today the benefits of LA-ICP-MS for direct solids analysis are well doc~mented.~’-~~ Many studies have mainly concen- trated on bulk analysis of the sample. However LA as a method of sample introduction also has the potential to reveal spatial information on element distribution in solids.ImalZ6 reported the use of LA-ICP-MS for the analysis of rock samples and also for the analysis of mineral inclusions in the rocks. In the present study LA-ICP-MS is used to investigate the possible gradients of trace elements in weathered marble crusts by profiling from the interior marble to the weathering layer. Different phenomena can occur when marbles (ie. metamor- phosed CaCO,) are subjected to weathering. The outside weathered surface or ‘patina’ generally consists of recrystallized calcite clay minerals zeolites Fe and Mn oxides calcium oxalate and gypsum. Evidence for natural dissolution can also be found on the crystals. In nature the deterioration of rock is a result of physical chemical mechanical and biological proce~ses.~’ Among the mechanisms responsible for the decay of marble chemical action from the exposure to water and airborne atmospheric pollutants is of major importance.The natural and anthropogenic pollutants relevant to stone decay are sulfur compounds nitrogen oxides ozone hydrogen chlor- ide hydrogen fluoride and carbon dioxide. Sulfur dioxide is the primary pollutant involved in sulfate formation on stone. Physical weathering mechanical stresses organic activity and fracturing enhance the deterioration process.” The rate at which the process takes place and the weathering features are a function of several factors the most important being the intrinsic properties of the stone (grain size and mineralogy) and its weathering history (duration of weathering burial conditions heating and cooling cycles action of water influ- ence of living organisms and wind etc.).Since various phenomena can take place on marbles exposed to weathering over a long period of time careful examination of the weathered surface is expected to provide the analyst with some extra information concerning authenticity. The assessement of the authenticity of an artefact should of course never be based on one sample or method. At present several techniques (stable-isotope mass spectrometry 14C-dating cathodoluminescence and LA-ICP-MS) are under investigation as to their usefulness in authentication studies. The investi- gation consists of two parts. One part is concerned with the analysis of weathering crusts on natural quarry samples in order to establish some characteristics of naturally weathered marble crusts.Secondly the surfaces of several Greek and Roman artefacts are examined. In this paper the results obtained when analysing the surface crusts using LA-ICP-MS are reported. Experimental Instrumentation During this work a Perkin-Elmer SCIEX Elan 500 ICP mass spectrometer combined with a Perkin-Elmer Model 320 laser sampler was used. The operating conditions of the ICP-MS system are listed in Table 1. All gas flows were regulated with mass-flow controllers. The ion optics were optimized in the usual way with a multi-element standard solution. These ion optics settings were not altered between solution nebulization and LA. Sample introduction could be easily changed from1244 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 Table 1 laser sampler Operating conditions for the ICP mass spectrometer and the ~~ ~~ ICP mass spectrometer- Forward power/W Reflected power/W Gas flow rates/l min-' Outer Intermediate Carrier Resolution Measurements per peak Measuring mode Measurement time/s Dwell time/ms Laser sampler- Mode Laser energy per pulse/mJ Laser pulse sequence Focus 1250 < 5 14.0 1.4 1.3 Low (+_ 1 u at 10% of the peak maximum) 1 (At peak maximum window+O.l u) Peak-hopping 0.350 10 Free-running 320 Single pulse On sample surface pneumatic nebulization for solutions to LA by disconnecting the nebulizer and spray chamber from the sampling inlet of the ICP torch and replacing them with the transfer tube of the laser sampler.The mass spectrometer can be operated in either a scan mode or a peak-hopping mode. Mass scanning and data acquisition are computer controlled. The laser sampler consists of an Nd:YAG laser operating at its fundamental wavelength of 1064nm. It is a stand alone unit with its own control software and computer. The laser can be operated in either the free-running or the Q-switched mode.21 The laser energy can be varied up to 505 mJ in the free-running mode and up to 385 mJ in the Q-switched mode. The laser beam is deflected by a mirror and is incident on the sample at 90". The sample cell is constructed out of borosilicate glass with a high transmission detachable window fixed 10" from the normal. The cell is continuously flushed with Ar as carrier gas to transport the ablated material directly to the ICP torch.A closed video circuit is used to control the laser focus conditions and to observe the ablation event. Analysed Material In order to investigate the impact of weathering expected to occur in patina layers a variety of naturally weathered marble samples from different ancient quarries was examined Naxos (Na.) Paros (Pa.) Pentelikon (Pe.) and Thasos (Th.) in Greece Afyon (Af.) and Proconnesos (Pr.) in Turkey and Carrara (Ca.) in Italy. The features found for the naturally weathered samples from the different quarries were compared with those found for marble cores taken from authentic ancient marble artefacts. As a result of cooperation with several museums a variety of ancient artefacts could be examined.Sample Preparation and Laser Analysis In principle one can study any sample that fits into the sample chamber and can be evaporated by the laser beam. In this study marble squares with sides of 10-20mm and thickness of 2-5mm were taken perpendicular to the surface of the original quarry samples and of the cores taken from the authentic artefacts. For ease of handling all samples were embedded in an epoxy cast (Durcapan ACM Fluka). After embedding the samples were polished. A free-running laser was used for the analysis since in this free-running mode the pulse produces a narrower and deeper crater than in the Q-switched mode. As a result the free- running mode can provide a higher degree of microsampling for characterizing discrete spots within a specimen whereas the Q-switched pulse mode can provide a wider more represen- tative sampling of the bulk specimen.21 Hence higher spatial resolution is obtained using a free-running laser and this is to be preferred for this work. In order to obtain accurate results a short optimization study concerning the carrier gas flow rate and the conditions under which the laser was operated (laser energy and focusing depth) was carried out.This optimization study resulted in the settings given in Table 1. To obtain small sample spots single-pulse sampling was selected rather than continously pulsing the laser. The laser was focused on the sample surface leading to crater diameters of about 0.1 mm. Concentration profiles were recorded by scanning the laser from the weathered crust to the fresh marble thereby taking a sample with the laser every 0.25 mm in the patina.Laser ablation as a means of sample introduction for ICP-MS leads to transient signals. For each laser shot the total measuring time was 2min covering the entire transient signal. The recording of the signals was started about 10s before the actual laser shot was fired. During each measurement four elements were monitored using the peak-hopping data acqui- sition mode. During the 2 min total measuring time a summed result over 35 succesive repeats (dwell time 10ms per peak) was given every 2 s for the elements monitored. For each of the elements measured the results for the '2 s intervals' were afterwards further summed in order to obtain a final result for the entire transient signal.To correct for shot-to-shot differ- ences in the efficiency of the ablation the signal of 43Ca (isotopic abundance 0.14%) was used as an internal standard. This assumes that the content of CaCO is 100% in each sampling position. This may not always be so especially at the surface owing to the presence of for example gypsum. Using 43Ca as an internal standard can thus bring about some errors however it is believed that using an internal standard is to be preferred above using no internal standard at all. After a preliminary test the following elements and isotopes were chosen for analysis 25Mg 27Al 55Mn 57Fe 64Zn 88Sr 13*Ba 13'La and 208Pb. For these elements a concentration profile in some of the weathered marble samples could be expected.Magnesium has three isotopes 24Mg (79%) "Mg (10%) and 26Mg (11Y0). As 24Mg suffered interference from "C2+ due to the presence of high amounts of C while 26Mg could experience an isobaric interference from 12C14N+ 25Mg was monitored and not the most abundant isotope 24Mg. For the determination of Fe 57Fe (2.14%) was monitored. Iron-56 (91.7%) is the most abundant isotope but was interfered to a large extent by 40Ca160+ and 40Ar160+. Iron-54 interfered with 40Ar14N+ . Although in general when using aqueous solutions 57Fe suffers interference from ArOH' in this case it is the best choice as a result of the limited presence of H owing to the use of a dry plasma. For the determination of Zn Sr Ba La and Pb the most abundant isotope was monitored; A1 and Mn are monoisotopic.Standardization The selection of standards is very important for analysis by LA-ICP-MS. If possible it is always desirable to match the matrices of standards and samples. Pearce et described the preparation of CaC03 standards. They studied the quanti- tative and semi-quantitative analysis of carbonate materials using multi-element synthetic standards both as pressed powder pellets and fused glass discs. In this study Suprapur CaCO (Merck) was used as the matrix material of the pressed powder pellets. This has the advantage over powdered marble of being more pure and more homogeneous. Calibration standards were prepared by adding small volumes of commer- cial single-element standard solutions (Johnson Matthey GmbH Alfa Products) to 5 g of the carbonate. A 5 ml volume of a 1% poly(viny1 alcohol) and 5% poly(ethy1ene glycol) solution were added as binder.Pure water (Barnstead Waterl 18 MR) was added up to a total liquid volume of 15 ml. AfterJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1245 mixing the carbonate thoroughly with the standard solution the binder and the water the slurry was freeze-dried. This modification (of the preparation by Pearce et ~ 1 . l ~ ) makes regrinding in a mortar prior to pressing a pellet unnecessary. About 2.5 g were pressed into a pellet 20 mm in diameter and 2-3 mm thick. Standards with concentrations ranging from 25 to lOOopgg-' for Al Fe and Mg and from 10 to 500 pgg-l for Sr Mn Zn Ba La and Pb were prepared. Results and Discussion Standardization The synthetic standards were analysed first in order to investi- gate whether linear calibration graphs could be constructed. Each standard was analysed five times by firing the laser at five different points. Analysis of the series of standards not only yielded linear calibration curves as shown for Mg Mn and Pb in Fig.1 but the signal ratios obtained also showed excellent repeatability (n = 5) indicative of homogeneity (Table 2). Correlation coefficients of typically 0.999 or better were obtained. These data compare favourably with the calibration graphs con- structed by Perkins et for fused glasses which are in fact homogeneous solid solutions. They however preferred to use the focused Q-switched mode in order for the laser to couple with the glass. As linear calibration graphs and good correle- tion coefficients were obtained the application of the synthetic standards to the quantitative analysis of the marble samples was investigated.Quantification In a second experiment the accuracy of the quantitative results obtained using the calibration graphs was checked by examin- ation of four samples; (i) a CaCO standard from the series of 0.16 I I 0 200 400 600 800 1000 1200 Mg additionlpg g-' 1.0 I I 0 100 200 300 400 500 600 Mn additionlpg g-' 0.8 I (c) ?. Os6 0.4 1 8 0.2 c 1 0 100 200 300 400 500 600 Pb addition/pg g-' Fig. 1. Calibration graphs of the ratio of the signal intensity of the isotope monitored relative to the signal intensity for the internal standard (43Ca) for the pressed powder standards produced by solution additions (a) 25Mg (b) 55Mn (c) zoEPb Table2 RSDs (%) for the signal ratios (n=5) obtained for the synthetic standards Analyte "Mg 57Fe 2 7 ~ 1 Analyte 55Mn 64Zn "Sr 13'Ba 13'La "'Pb Concentration/p g- ' 25 100 250 lo00 4.6 12 3.0 2.8 5.3 4.5 2.7 1.6 6.0 2.9 3.5 1.6 Concentration/pg g - 10 50 100 500 3.6 3.6 1.9 2.6 16 8.0 2.3 3.8 2.2 4.7 3.6 1.1 1.3 2.6 2.7 1.2 1.5 1.4 2.1 0.71 2.1 2.4 1.7 1.3 standards this standard was not used when constructing the calibration graphs; (ii) a powdered marble sample [Marble Carrara Ghent (MCG)30] a 7 kg stone of a Carrara marble was reduced to grains smaller than 0.1 mm and after mixing sub-samples were pressed into 5 g pellets in the same manner as described above; and (iii) two 5mm thick slices from the quarry samples Ca.45 and Af.163 respectively. Each sample was analysed 15 times by firing the laser at 15 different points. For the determination of minor and trace elements the marble samples were also analysed by instrumental neutron activation analysis (INAA) and atomic absorption spectrometry (AAS). A summary of the analytical results obtained by INAA and AAS on samples ( ~ 5 0 0 m g ) of the MCG material has been reported by Freitas et aL3' For the activation analysis of the quarry samples marble cylinders (h = 20 mm d = 15 mm m = 9 g) were analysed. The results obtained for the synthetic standard analysed as an unknown were in excellent agreement with the known values (Tables 3 and 4) for all elements examined. For the MCG sample agreement with the results obtained using INAA and AAS was excellent for Sr and acceptable for Mg (Table 3).For the other elements analysed agreement was only within a factor of three (Table4). Comparison of the results obtained for the quarry samples Ca. 45 and Af. 163 with the INAA and AAS values also showed acceptable agreement for Sr and Mg (Table 3). Major deviations however were found between the LA-ICP-MS and INAA and AAS results for the other elements examined. The larger deviation between the LA-ICP-MS and INAA and AAS results for the MCG sample (powdered) and the quarry samples (not powdered) is mainly related to the natural inhomogeneity of the marble samples which is evident from the high relative standard deviations (RSDs) (see Table 4 for MCG). This can be attributed to the presence of accessory minerals.Moens et a131 and R O O S ~ ~ showed that the natural inhomogeneity of marble even for larger samples (9 g ) taken Table 3 Results for the quantitative analysis (n= 15) of Mg and Sr in a CaCO standard a powdered marble sample (MCG) and two quarry samples (Ca. 45 and Af. 163) Sample Analyte Standard "Mg '*Sr MCG 2sMg "Sr Ca.45 "Mg 88Sr Af. 163 "Mg "Sr Measured/ 255 99 4610 192 4140 169 1760 131 Pg g-' RSD (Yo) 3.3 2.3 2.8 1.5 5.9 2.6 7.3 2.8 Reference value/ 250 100 5100f21 196 f 10 4960 & 200 167 & 7 2050 f 80 122f5 Pg I3-l1246 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1.4 Table 4 Results for the quantitative analysis (n = 15) of Al Mn Fe Zn Ba La and Pb in a CaCO standard and a powdered marble sample (MCG) (a) -! i Mn (Th. 149) Sample Analyte Standard 27Al "Mn 57Fe 138Ba 139La '08Pb 1.2 1.0 0.8 MCG 2 7 ~ 1 55Mn "Fe 13'Ba IJgLa '08Pb -'i - i - \ Measured/ 0-1 249 98 232 99 102 101 104 117 29 365 7.5 3.6 0.97 5.4 RSD (%) 6.3 3.2 9.4 4.2 4.6 4.6 5.5 28 21 67 28 26 50 8.7 Reference/ 250 100 250 100 100 100 100 342 & 34 20.3 f 0.9 223 f 67 w.3 g-' - - 0.50 & 0.02 - within a distance of up to 10cm can be considerable.As the amount of material ablated during a laser shot by LA-ICP-MS is very small results cannot be expected to be in good agreement with the bulk analysis by INAA and AAS on 500 mg (MCG) or 9 g (Ca. 45 and Af. 163) samples even when the average of 15 different laser shots at different places is considered. Agreement between both data sets is however obtained for both Mg and Sr both elements being homo- geneously distributed over the calcite matrix. Calcium Sr and Mg are in fact mainly present in the same matrix material whereas the other elements are particularly concentrated in the accesories.Owing to the inhomogeneous distribution of the trace elements in marbles the potential for the comparison of results between the INAA/AAS and LA-ICP-MS is rather limited. Since however accurate results were obtained for the (homogeneously distributed) minor elements (Sr and Mg) in the marble samples and for both the minor and trace elements in the synthetic standards (Tables 3 and 4) it can be assumed that the LA-ICP-MS results obtained for the trace elements in the marble samples are also accurate but only representative for the micro-spot ablated and not for the bulk marble.It is evident that sample homogeneity could be improved by finely grinding the marble samples and pressing them into pellets. However by so doing it would no longer be possible to scan the laser from the fresh marble to the weathered crust and so obtain depth information about the distribution of trace elements in the weathering layer. As the aim of this study was in essence to investigate whether LA-ICP-MS is a useful technique for establishing gradients in trace elements in weathered marble crusts and not to determine minor and trace element concentrations in marbles the results of the analyses are reported as the ratio of the signal intensity of the isotope monitored relative to the signal intensity obtained for the internal standard (43Ca) and no absolute concentrations are given.Analysis of Quarry Samples Since various phenomena can be expected to take place on marbles exposed to weathering over a long period of time careful examination of the surface from naturally weathered marble rocks sampled in different ancient marble quarries can be useful in establishing characteristics of natural decay. In order to allow differences between the fresh and weathered marble to be used as an indication of antiquity the natural inhomogeneity should be known. Hence when profiling from the interior marble to the weathered crust at least five different points in the fresh marble were sampled when possible. Significant trends for several elements were found. The repeatability of LA-ICP-MS is thus sufficient to obtain depth profiles of elements in weathered marble crusts.Seventeen samples from seven different quarries were analysed. The results show that not every sample exhibits the same features. This is probably due to the differences in texture as well as in weathering history. Although different profiles were registered for the different stones analysed some trends could be estab- lished. If a gradient exists mostly the elements examined are enriched in the patina with a penetration depth varying from less then 0.25 mm [Fig. 2(u)] to 3-4 mm beneath the surface [Fig. 2(b)]. Strontium and Mg on the other hand are some- times leached out [Fig. 2(c)]. More detailed data concerning the sample Af. 163 are given in Fig. 3. Two 'peak' concentrations are found for each of Al Fe and Mn at the surface and 0.5 mm below the surface.This is probably due to different stages that occurred during the weathering process. Although not every quarry sample exam- ined exhibits the same features because of the previously mentioned differences in the texture as well as in the weathering history it can be concluded that LA-ICP-MS can be useful in establishing gradients in trace elements in weathering crusts. 14 1 12 10 8 9 ? . x 6 4 2 0 0.15 0.10 0.05 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Penetration dept h/m m Fig. 2. Established gradients in trace elements [ratio of the signal intensity of the isotope monitored ("Mn 138Ba 57Fe 27Al "Sr 25Mg) relative to the signal intensity for the internal standard (43Ca)] in the damaged layer of naturally weathered quarry samples.Most of the elements examined are enriched in the patina with a penetration depth varying from (a) near the surface to (b) 3-4mm beneath the surface. (c) Sr and Mg are sometimes leached out.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 14 12 10 8 6 4 2 0 0.5 0.4 (D 0.3 u ?. 0.2 0.1 0 1.2 r 1 leO t;l 0.8 0 0.50 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Penetration de pt h/m m Fig.3. Influence of weathering on the quarry sample Af. 163 ratio of the signal intensity of the isotope monitored (27Al "Fey "Mg "Sr I3'Ba 139Lay 55Mn 208Pb) relative to the signal intensity for the internal standard (43Ca) The profiles established are probably a result of the penetra- tion into the surface layer of water containing dissolved ions and suspended clays from soils in contact with the marble.According to Margolis and Showers2* encrustations and stains can either come from precipitation in water and soils in contact with the marble or from leaching of impurities from within the marble. Sabbioni and Z a ~ p i a ~ ~ studied the origin of various components (whether of atmospheric origin or from the car- bonate rock) imbedded in black weathering crusts. They found that the presence of Al Mg Fe Mn and Sr can largely be attributed to the carbonate rock and to some extent to atmospheric aerosol of natural (Al Mg) and anthropogenic (Mn) origin. The origin of Pb however is clearly linked to atmospheric deposition (automobile exhaust gases). Analysis of Ancient Artefacts As a result of cooperation with several museums a variety of ancient artefacts could be examined.Analogous trends were found for the patinas on the artefacts although the observed shifts were often smaller. The quarry samples have probably been exposed during a much longer period and in more severe conditions. In this paper the results for two artefacts are reported in more detail i.e. AB-2 and B-XXl. 24 22 20 18 16 14 12 10 8 6 4 2 2.5 s 9. 2.0 X 1.5 1 .o 0.5 - 0 ' 1 I I I L I 0.05 1 Pb 0.03 Ba ........... 0 0.5 1.0 1.5 2.0 2.5 3.0 Penetration depth/mm 1247 Fig. 4. Influence of weathering on the artefact AB-2 ratio of the singal intensity of the isotope monitored (25Mg 55Mn 27Al I3*Ba 13'La 2oaPb) relative to the signal intensity for the internal standard (43Ca) In Fig.4 the results of a strongly weathered artefact are shown. The AB-2 comes from a private Belgian collection and is a votive relief of Cybele. The goddess is enthroned in a niaskos feet resting on a footstool a lion in her lap. The relief was made after a prototype of about 3 4 0 ~ ~ . The object was carved of Pentelic marble. A high Mg concentration was found on AB-2. This is probably due to a process known as dolomitiz- ation. In this process Ca atoms are slowly replaced by Mg atoms present in the water in contact with the marble.34 The statue B-XXl from the Base1 Museum of Ancient Art and Ludwig Collection represents a torso of a priestess. The statue is presumably a Roman copy and was made after a prototype of about 450 BC. Again several elements are enriched in the patina with a penetration depth varying from 0.5 to 1 mm beneath the surface (Fig.5). For some artefacts studied in this work little influence of weathering is found. This could be a result of acid cleaning which in the past was a common practice of curators to remove the weathering crust and along with it unfortunately also any evidence of antiquity. Very little however is known about the weathering history of the artefacts examined. Conclusions In order to allow differences between the fresh and weathered marble to be used as an indication of antiquity profiles were1248 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 0.9 I I I 1 I 1 I 0.030 i Fe 0.025 - *.*. ................................... Ba 0' I 1 I I 1 I 0 0.5 1.0 1.5 2.0 2.5 3.0 Penetration depth/mm Fig.5.Influence of weathering on the artefact B-XXl ratio of the signal intensity of the isotope monitored (27AI 25Mg 55Mn 57Fe I3'Ba '''Pb) relative to the signal intensity for the internal standard (43Ca) recorded by scanning the laser from the fresh interior marble to the weathered crust. The repeatability of LA-ICP-MS was shown to be sufficient to provide depth profiles of elements in weathered marble crusts. Large and highly significant trends for several elements were found. Not every sample examined showed the same weathering features. This is probably due to the differences in texture as well as in weathering history. The LA-ICP-MS method is thus a promising technique in order to allow differences between the fresh and weathered marble to be used as an indication of antiquity.The authors are highly indebted to the Institute for Scientific Research in Industry and Agriculture (IWONL) and the Interuniversity Institute for Nuclear Sciences (IIKW) for fin- ancial support. We are also very grateful to a Belgian collector and to the Base1 Museum of Ancient Art and History for allowing us to sample their works of art. The valuable advice of Dr. P. De Paepe and Dr. N. J. G. Pearce is also gratefully acknowledged. The authors also acknowledge B. Gilissen and J. Cooymans of VITO (Mol) for their indispensable help with the preparation of the CaCO standards. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 References Williams J. G. Gray A. L. Norman P.and Ebdon L. J. Anal. At. Spectrom. 1987 2 469. Darke S. A. Long S. E. Pickford C. J. and Tyson J. F. Fresenius' J. Anal. Chem. 1990 337 284. Mochizuki T. Sakashita A. Iwata H. Ishibashi Y. and Gunji N. Fresenius' J. Anal. Chem. 1991 339 889. Van Borm W. A. H. Broekaert J. A. C. Klockenkamper R. Tschopel P. and Adams F. C. Spectrochim. Acta Part B 1991 46 1033. Jarvis K. E. Chem. Geol. 1992 95 73. Hartley J. H. D. Hill S. J. and Ebdon J. Spectrochim. Acta Part B 1993 48 1421. Darke S. A. Pickford C. J. and Tyson J. F. Anal. Proc. 1989 26 379. Vollkopf U. Paul M. and Denoyer E. R. Fresenius' J. Anal. Chem. 1992 342 917. Boomer D. W. Powell M. Sing R. L. A. and Salin E. D. Anal. Chem. 1986 58 975. Karanassios V. and Horlick G. Spectrochim. Acta Rev. 1990 13 89.Marshall J. Franks J. Abell I. and Tye C. J. Anal. At. Spectrom. 1991 6 145. Moenke-Blankenburg L. Schumann T. Gunther D. Kuss H.-M. and Paul M. J. Anal. At. Spectrom. 1992 7 251. Pearce N. J. G. Perkins T. W. and Fuge R. J. Anal. At. Spectrom. 1992 7 595. Chenery S. and Cook J. M. J. Anal. At. Spectrom. 1993 8 299. Jiang S.-J. and Houk R. S. Anal. Chem. 1986 58 1739. Jiang S.-J. and Houk R. S. Spectrochim. Acta Part 8 1987 42 9312. Gray A. L. Analyst 1985 110 552. Arrowsmith P. Anal. Chem. 1987 59 1437. Hager J. W. Anal. Chem. 1989 61 1243. Tye C. T. Henry R. Abell I. D. and Gregson D. Res. Dev. 1989 31 76. Denoyer E. R. Fredeen K. J. and Hager J. W. Anal. Chem. 1991 63 454. Furuta N. Appl. Spectrosc. 1991 45 1372. Van Heuzen A. A. Spectrochim. Acta Part B 1991 46 1803. Van Heuzen A. A. Spectrochim. Acta Part B 1991 46 1819. Darke S. A. and Tyson J. F. J. Anal. At. Spectrom. 1993 8 145. Imai N. Anal. Chim. Acta 1992 269 263. Amoroso G. G. and Fassina V. Stone Decay and Conservation Elsevier Amsterdam 1983. Margolis S. V. and Showers W. in Classical Marble Geochemistry Technology Trade eds. Herz N. and Waelkens M. Kluwer Academic Publishers Dordrecht 1988 vol. 153 pp. 233. Perkins W. T. Fuge R. and Pearce N. J. G. J. Anal. At. Spectrom. 1991 6 445. Freitas M. C. Moens L. De Paepe P. and E. Barros J. S. J. Radioanal. Nucl. Chem. 1988 123 273. Moens L. ROOS P. De Rudder J. and Hoste J. in Archaeometry ed. Maniatis Y. Elsevier 1989 pp. 613. Roos P. PhD Thesis University of Ghent 1992. Sabbioni C. and Zappia G. Sci. Total Environ. 1992 126 35. Tucker M. E. Sedimentary Petrology Blackwell Scientific Publications Oxford 1991 p. 147. Paper 4/01 5641 Received March 16 1994 Accepted May 23 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901243

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1249 Improved Boron Determination in Biological Material by Inductively Coupled Plasma Mass Spectrometry Susan Evans and Urs Krahenbuhl lnstitut fur Anorganische Analytische und Physikalische Chemie Universitat Bern 3000 Bern 9 Switzerland Important aspects of boron determination by inductively coupled plasma mass spectrometry are reviewed and extended by additional studies. Determinations on a range of biological materials were performed by external calibration standard additions and isotope dilution. Memory effects were reduced by rinsing with a wash-solution of sodium fluoride. Accuracy was tested with four biological standard reference materials. Accurate and precise determinations can be achieved with all three modes.The fastest and simplest method is external calibration where matrix effects can be controlled with beryllium as internal standard. No improvement resulted with the more laborious standard additions method. The isotope dilution technique requires time-consuming supplementary measurements. Biological material with a boron content 0.06 pg g-' can be determined. Keywords Boron; inductively coupled plasma mass spectrometry; biological material; external calibration; isotope dilution Boron is long known to be an essential element for plants,'g2 whereas its role in human and animal nutrition is still a subject of discu~sion.~ The boron supply in human diet is of increasing interest and the demand for reliable determination of boron in biological material is growing.Inductively coupled plasma mass spectrometry (ICP-MS) is a sensitive analytical technique allowing precise and fast measurements. A number of studies have been published on the determination of boron by ICP-MS.4-9 A major problem of determining boron at the trace level with this technique is a high memory effect. Some workersl0+l1 used wash-out times of several minutes which extended the measuring time while Smith et aL6 suggested the use of a direct injection nebulizer to minimize memory effects. Another problem with ICP-MS determinations are the spectroscopic and non-spectroscopic interferences.12 For- tunately for boron no correction of is0 baric interferences is necessary since no other isotopes with an atomic mass of 10 or 11 exist. No species originating from water acids or plasma gas were found at these masses.' However spectral interference can occur from the 12C overlap if high levels of organic carbon remain in solution after the digestion particularly of biological material.Non-spectroscopic interferences caused by high amounts of matrix elements are a severe problem for light elements such as boron.g The separation of boron from the matrix is indispensable when the sample material is fused with sodium carbonate yielding high salt concentration^.^^^ After acid digestion lower salt concentrations occur and a laborious separation may not be necessary. Three different calibration modes can be applied to deter- mine boron concentrations by ICP-MS. Most commonly external calibration (EC) is used. An internal standard with a mass close to the analyte is recommended to control matrix interference and instrumental shifts.14 When measuring with the standard additions method (SA) matrix effects are cor- rected.However rather large amounts of sample solution are needed since at least two additions are necessary. As the isotopes of an element can be determined separately in ICP-MS a third quantification mode is possible using the isotope dilution (ID) technique. A problem here is mass discrimination which can occur with ICP-MS instruments. For boron with a high mass difference of 10% between loB and IIB instrumental mass bias can be ~ignificant.~,~ Moreover high salt concentrations can affect the "B loB ratio measurement s . ~ This work was performed to achieve a simple and fast method for precise and accurate determinations of boron in biological materials.An effort was made to reduce the memory effect by applying different wash solutions. The degree of spectral overlap from the "C isotope onto the "B signal was monitored. Studies were performed on the non-spectroscopic interferences caused by matrix elements. The suitability of Be and Sc as internal standards for boron determination was assessed. The ID technique was examined in detail for boron analysis. The three different calibration modes were compared and are discussed. Limits of detection and determination were evaluated. Accuracy was tested by analysing standard reference materials (SRMs) with certified values and by comparing with the results obtained by inductively coupled plasma atomic emission spectrometry (ICP-AES).Experimental Materials and Reagents Analytical-reagent grade chemicals and de-ionized bi-distilled water were used throughout. A standard solution of 1000 pg g-' of boron (Merck) was diluted and used for cali- bration purposes. An internal standard of 5 pg g-' was pre- pared from standard solutions of beryllium (Sigma 990 pg ml-') and scandium (Perkin-Elmer AAS Standard 1000 pg ml-I). For ID measurements the SRM 951 from the National Institute of Standards and Technology (NIST) (5137 pg g-' 'lB:"B ratio of 4.0436) and an isotopically enriched bi-isotopic spike from Hempel ("B loB ratio of 0.09890) were used. A 2mg g-' NaF in 0.02 mol I-' HNO was prepared as wash solution. Blank and boron standards for calibration were prepared in 1.5 mol 1-' HNO solutions.Plastic containers were used throughout to avoid boron con- tamination from glass vessels. Preparation of Biological Samples The biological samples (hay wheat flour and freeze-dried beef kidney) were digested with 2 ml of HNO (65%) and 1 ml of H,02 (30%) by using the microwave digestion technique described previ~usly.'~ Per digestion 0.5 g (hay 0.2 g) dry sample material was inserted. After the digestion the solutions were diluted with water to a total mass of 10 g. The hay sample solution was separated from the silica residue by centrifugation. Six to twelve digestion solutions were collected and the com- bined solution was used for all types of determinations. If necessary the solutions were diluted before determination with 1.5mol I-' HNO to a boron concentration of less than 100 ng g-'.1250 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 The NIST SRM 1573 (Tomato Leaves) SRM 1575 (Pine Needles) SRM 1577 (Bovine Liver) and SRM 1568 (Rice Flour) were treated in the same manner as the test material. Two to four aliquots each with 0.2 g of the plant or liver or 0.5 g of the flour material were submitted to microwave digestion and measurement. Preparation of Sample Solutions for Quantitative Determinations In EC and SA 50 pl of the Be/Sc internal standard solution was added to 5.00 g of the sample solution. In SA measurements boron was added as a 10 pg g-' solution. Two additions were made to achieve a 2- and 3-fold signal intensity compared with the zero addition solution.In ID measurements the blank solution was measured after each determination of the sample ratio and the ''B and 'OB blank counts were subtracted from the previous measurement. The instrumental discrimination factor was determined from the ratios of the certified solutions (NIST SRM 951 and the spike) by dividing the measured by the theoretical (certified) ratio. All ratios determined were corrected by this factor. The ratio of the pure unspiked sample was determined to check natural abundance and interferences. To achieve maximum precision the spike solution was added in amounts corresponding to a theoretical "B:'*B ratio of between 0.4 and 1.1. After measurement of the ratio of the spiked sample solution and correction by the discrimination factor the boron concentration was calculated from this cor- rected ratio.The precise concentration of the diluted spike solution was determined by the reverse-ID technique with the certified concentration of NIST SRM 951. Instrumental Settings A Perkin-Elmer SCIEX ICP-MS Elan 5000 with a pneumatic nebulizer (cross flow) was operated at 1.00 kW (r.f. power) with a sample uptake rate of 1.0ml min-'. The gas flows were set to a plasma flow-rate of 15 1 min-' intermediate flow-rate of 0.8 1 min-' and nebulizer flow-rate of 0.8-0.9 1 min-'. To improve counting statistics the dwell time on each boron isotope was varied according to the abundance. Thus a measuring time of approximately 60- 100 s per sample resulted. Measurements were performed by peak hopping at the highest resolution mode of 0.6 u (peak width at 10% of the maximum intensity).Between two measurements the sampling system was rinsed with the NaF wash solution for 60 s. The parameters of the ICP-AES measurements were set as described previ~usly.'~ Results and Discussion Reduction of the Memory Effect The memory effect for boron was controlled after injecting a 100 ng 8-l boron solution for 2min (Fig. 1). Using 1.5 mol I-' HNO as wash-solution and rinsing for 2 min boron wash-out was still incomplete. When rinsing with a slightly acidified (0.02 moll-' HNO,) 2 mg 8-l NaF solution for 60 s and then injecting the blank solution the original blank level was found. We assume that borate which tends to adsorb to the sampling system surface is complexed by the high excess of fluoride to stable boron trifluoride and is washed out.By rinsing the sampling system with acidified NaF after every injection the wash-out time is reduced by at least a factor of 2. Principal Studies on Non-spectroscopic and Spectroscopic Interferences The material applied was chosen to include a variety of different biological matrices. Hay was selected as material with a medium boron content whereas beef kidney and especially wheat flour are very low in boron. The majority of the matrix r I v) cJ7 .cI 1 x 1 0 6 I i A o r B 1.5 rnol I-' 1.5 rnol I-' HNO \ I 3 -. > .I- v) C g 1 ~ 1 0 4 .- c - I 0 1 2 3 4 5 Ti me/m i n Fig. 1 Study of the memory effect after injection of a 100 ng g-l boron standard solution for 2 min with the wash solutions A 1.5 mol 1-' HNO (broken line) or B acidified 2 mg g-' NaF solution (solid line) for 1 min followed by the blank solution (1.5 mol I-' HNO,) elements in biological materials are of a low atomic mass (e.g.Na K Ca P and S). In the digestion solution of the flour sample salt concentrations of less than 200 pg g-' occur while in the solutions of hay and kidney ~ 2 0 0 0 pg g-' are present. As the matrix effect is dependent on the absolute amount of matrix elements rather than on the molar ratio of matrix element to analyte,I2 simple dilution with water could solve all matrix problems. The boron content however only allows dilution of the hay sample solution. According to Gregoire4 a loss in boron sensitivity occurs above a total salt concentration of 2000 pg g-'. Thus interferences from the matrix components are negligible when measuring the flour sample or the diluted hay sample but should be observed when analysing samples of the kidney type.The major source of matrix elements in the sample solutions however results from the HNO added in the acid digestion step. No problems are experienced in HNO solutions of 0.2mol I-' or less because of the high ionization energies of nitrogen and oxygen.16 Essentially identical background spec- tra were found with distilled water and 1 mol I-' HNo3.I3 Vanhaecke et found that the signals from light analytes showed a high suppression in 0.5mol I-' H2S04 matrix compared with a 0.14mol 1-' HNO matrix. However this effect can be eliminated by the addition of an appropriate internal standard.In our study after digestion and dilution to 10 g the concentration of HNO is between 1 and 3 mol 1-'. Thus the influence of the HNO concentration on the boron signal was checked. Measurements were performed with 50 ng g-' boron solutions in a 0.2 1 3 and 5 mol 1-l HNO,. With both internal standards the measured B standard ratio relates logarithmically with the acid concentration but with different trends. With Sc the B Sc signal ratio using 3 mol 1-l HNO is 3 and 7% lower compared with the ratios using 1 and 0.2 mol 1-' HNO respectively. With Be the corresponding B Be ratio is 6 and 15% higher. This trend can be explained by instrumental mass-discrimination. However the mass difference between Sc (m/z 45) and B is clearly higher than between Be (m/z 9) and B.Instrumental mass discrimination would also imply a variation in the ''B:'OB intensity ratio with increasing acid concentration which was not observed. While the error occurring from this shift is negligible with the B Sc ratio systematic errors can result with the B Be ratio if the acid concentrations of sample and standard solutions are not adjusted. The calibration solutions can significantly influ- ence the accuracy of the determination especially at acid concentrations of less than 1 moll-'; between 1 and 3 moll-' the influence is smaller. Thus all standard solutions blanks and sample dilutions must have acid concentrations close toJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1251 the HNO concentration in the sample solutions. In this study all solutions were prepared in 1.5 mol 1-' HN03.Spectral interferences of ''C on the "B peak were checked by scanning the mass range and controlling the resolution. When measuring in the normal resolution mode (0.8 u) a high interference from 12C occurred with all materials indicating a significant amount of organic matter remaining in solution after digestion. In the higher resolution mode (0.6 u) the "B peak of the hay sample was resolved completely. However with the kidney and flour sample only a resolution of 75 and 55% respectively was reached. Here an enhancement of the "B signal was likely and must be checked by comparing the results obtained by the loB and the "B signal. Determinations by the Standard Additions and External Calibration Modes The use of an internal standard with a mass close to the analyte is recommended in the ICP-MS 1iterat~re.l~ The measurement of the standard isotope must be free of spectral interferences.In addition a low natural occurrence of the standard isotope in the samples must be ensured. The suit- ability of Be and Sc as internal standards in the EC and SA was checked by comparing the analysed boron contents deter- mined by the two calibration modes. The ICP-AES technique was used as a reference method. The results of both internal standards in SA and with Be in EC are in good agreement with each other and the ICP-AES results (Table 1). The measured boron contents are all at least within two standard deviations. A similar precision was found with the different calibration modes for all matrices.The boron contents deter- mined in the EC mode with Sc as internal standard were significantly lower for all materials. With the hay sample the standard deviation was about three times higher than that of the determinations with Be. In SA the choice of the internal standard does not affect the determination. This confirms that matrix effects and mass discrimination play a minor role in SA and the internal standard here is only added to control plasma fluctuation. Precisions of both EC with Be as internal standard and SA are comparable and range from 2.5% in the hay sample to 13% in the wheat sample due to the decreasing boron concentration. As spectral interferences are not corrected for in either the EC or the SA method an interference of 12C on the "B signal is possible and has to be checked. The boron contents found by evaluating the "B or the loB signal were compared. For the hay sample no difference was found between the "B and the 'OB determination (a factor of 1.01+0.017).In contrast significantly higher boron determinations resulted from the "B signal with factors of 1.07_+0.014 and 1.07_+0.055 respect- ively when analysing the kidney and flour samples. In the samples of low boron content the dissolved 12C appeared to interfere with the "B signal even in the higher resolution mode and the use of the loB isotope is recommended for EC and SA determinations. Principal Studies on the Isotope Dilution Technique When measuring boron isotopes a high instrumental mass bias occurs both from instrumental parameters such as ion- lens voltages and from mass discrimination from concomitant element^.^ The mass bias can be determined by measuring the ratio of a certified solution and an instrumental discrimination factor can be evaluated.The ID equation as adapted from Faure17 has to be corrected by this discrimination factor j where cs and c are the concentrations of the sample and the spike in ng g-' respectively ms and msp the corresponding initial masses in the mixture in g WN and Wsp the relative atomic masses of the normal element and the spike Ab the theoretical abundances of the "B or 'OB isotopes in the sample solution or the spike and R the measured l1B:loB ratio of the mixture. Gregoire4 examined a matrix-induced mass discrimination effect at sodium concentrations of more than 2000 pg g-'.As discussed above this problem can affect the kidney sample. In any case the ratio of the biological sample has to be controlled considering the mass discrimination factor and compared with the natural ratio. Error evaluations showed that three factors contributed most to the precision of the ID determinations. (i) The error of the sample solution depends strongly on the error in the spike concentration which can be determined by the reverse ID technique with the certified standard solution NIST SRM 951. The statistical error can be decreased by multiple measure- ments at different ratios. (ii) The error of the measured ratio contributes significantly to the error of the determined concen- tration. An optimal range of the mixed ratio is found with all bi-isotopic elements using a bi-isotopic spike as described by Heumann.l* The optimum measuring range of the "B:loB ratio was calculated with the addition of the mass discrimi- nation factor f where efis the error multiplication factor and Rs and RSp are the ratios of the pure sample and the spike respectively.The optimum range with the lowest error multiplication factor (efc 1.5) was found at measured "B IOB ratios between 0.41 and 1.4. It should be noted that the error multiplication factor is squared in the calculation of the error. (iii) If the ratio of the pure sample deviates from the natural ratio this ratio must enter the calculation. In this case the error of the ratio of the Table I parentheses Boron determination in hay beef kidney and wheat flour by different calibration modes and methods; number of replicates given in Boron concentration & SD of the mean/lg g-' Method (internal standard) Standard additions (Be)*.? Standard additions (Sc)**t External calibration (Be)* External calibration (Sc)* Isotope dilution ICP-AESS Hay 5.38f0.12 5.46 k 0.16 5.87k0.19 (5) 4.67f0.55 (4) 5.16f0.12 (4) 5.3k0.2 Beef kidney (freeze-dried) 1.02 0.07 1.02 _+ 0.08 0.94 f 0.07 (4) 0.88 k 0.07 (3) 1.01 f0.07 ( 6 ) 1.1 20.2 Wheat flour 0.205 k 0.026 0.197 _+ 0.005 0.164+0.023 (6) 0.21 1 & 0.022 ( 6 ) 0.221 f 0.020 (7) 0.23 _+ 0.02 * Values determined with the loB isotope.7 Two independent series each with a zero additions sample and two additions. $ Mean of the two boron lines at 249.773 and 249.678 nm.1252 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 pure sample is an additional important factor in the accuracy of the determination. Using the reverse-ID technique other factors such as the errors in the discrimination factor and in the ratio of the spike play a minor role as both parameters are neutralized and become negligible. It must be emphasized that these errors can only be neglected when the reverse-ID technique is applied. Determinations by Isotope Dilution In ID the blank is of even greater importance than in EC or SA as the blank from both isotopes can seriously influence the ratio. The error was minimized by measuring the blank after each determination and subtracting these succeeding blank counts from the previous sample determination. Extreme variations of the ratio within two succeeding samples still caused systematic deviations and should be avoided.By measuring the solutions with certified ''B 'B ratios (NIST SRM 951 and the Hempel spike) a mass bias of 20% was found. Thus a discrimination factor of 1.2 was used for all calculations of the isotope ratio. Taking this discrimination factor into account the natural ratio of 4.1 was found in the hay sample. As expected from the SA and EC measurements due to the "C interference higher ratios were found in the kidney and wheat samples. To achieve accurate measurements these ratios of the pure sample that are subject to interference have to be determined precisely. In the case of the kidney and wheat samples the abundances appearing (but corrected for by the discrimination factor here 1.2) were inserted into the calculation of the concentration.Using the spike concentrations obtained with reverse-ID a mass discrimination factor of 1.2 and the observed isotope abundances of the pure sample boron contents as shown in Table 1 were found. Taking ICP-AES as the reference method a high accuracy was found for all materials. A precision of 2-10% resulted while the measured ratios varied between 0.5 and 1.3. Analysis of Biological SRMs To check the accuracy of the ICP-MS measurements the boron contents of four biological SRMs were determined with EC (using Be as internal standard) and with ID (Table2). With the plant materials the boron contents determined with both calibration modes are within one standard deviation of the literature va1~es.l~ The high boron contents of these plant materials must however be considered.They require a high dilution of the digested solutions thereby automatically elimin- ating possible interferences. With Rice Flour a boron value of el pg g-' is given in the reference literature." A more precise determination of 0.57 k0.03 pg g- ' was found in an earlier study" using ICP- AES. For the Bovine Liver sample the boron values determined with either calibration mode deviated from the reference value," where a boron content of 2.940.8 pg g-' is given. However our values correspond with the determination by Anderson et aL2' who used neutron capture prompt-? acti- vation and found 0.40f0.12 pg g-'. They also agree with the ICP-AES value from the earlier study,15 where the difference from ref.19 was already noted. Thus we conclude that for all biological samples accurate values could be obtained with the different ICP-MS calibration modes. Limits of Detection and Determination The limits of detection and determination were calculated from the standard deviation of the blanks (3s and lOs respectively) of more than ten measurements. With the EC method limits of 0.8 and 2.7 ng g-' respectively were found. To determine the standard deviation of the blank in ID similar amounts of spike solution were added to the blank solution and to the real samples and the blank values were evaluated. Limits of detection and determination of 0.9 and 3.1 ng g-' respectively were obtained. Conclusions Memory effects in the determination of boron by ICP-MS were significantly reduced by rinsing the sampling system for 60s with acidified NaF solution.In the ID mode the boron isotope ratio of the blank has to be determined and subtracted after each sample measurement to ensure precise determinations. When analysing biological samples of low boron content a spectral interference from ''C with "B can occur even in the higher resolution mode. Thus with EC and SA evaluation of the 'B rather than the more abundant "B isotope is advised to avoid a possible spectral overlap. In ID the influence of the ''C isotope on the "B signal should be monitored. Matrix effects caused by concomitant elements in external calibration can be controlled by inserting Be as an internal standard.High HNO concentrations cause a too high B Be ratio. Thus similar acid concentrations should be assured for the samples standards and blanks. Precise and accurate determinations of boron by ICP-MS in biological materials can be achieved with all calibration modes tested in the present study. Solutions with boron contents 2 3 ng g-' can be determined corresponding to 0.06 pg g-' in the original biological material if 0.5 g sample material is digested. The EC is the simplest mode as rapid determinations are possible and small amounts of sample solution are required. Accurate results were found with Be as internal standard. The SA method can be applied but is inferior to EC with respect to time and sample consumption for preparation and measure- ment of at least three solutions per determination.In ID the instrumental mass-bias on the boron isotopes is large and Table 2 Boron determination in NIST biological reference material; number of replicates given in parentheses ~~ ~ Boron concentration + SD of the mean/p.g g- Method External calibration* Isotope dilution Literature value1 ICP- AES§ Tomato Leaves Pine Needles SRM 1573 SRM 1575 Rice Flour SRM 1568 Bovine Liver SRM 1577 36.8 & 0.5 (3) 37.6k0.5 (3) 33+4 (3) 35+1 0.35 k 0.08 17.6k0.4 (3) 0.57 & 0.03 (4 j 0.57 (0.09)t (2) 17.5 ? 0.9 (3) 0.59 & 0.02 (4) 0.51 (0.07)t (2) 17+2 (3) < 1 2.9k0.8 (4) 16f 1 0.57 f0.03 ~~~ ~~ * Values determined with the loB isotope. # Difference of two determinations. $ See ref. 19. 0 Earlier study,15 mean of the two boron lines.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL.9 1253 must be corrected if accurate results are to be obtained. The advantage of the ID technique for fast and accurate measure- ment is out-weighed by various supplementary measurements that have to be performed. These relatively extensive additional procedures ensure that the ID mode is not first choice for the determination of boron by ICP-MS. The measurements were carried out at the Swiss Federal Office of Public Health (BAG Division of Food Science) which also provided financial support for this work (FE 316-90-300). We thank M. Haldimann who has contributed by many discussions and practical help H.R. von Gunten for a critical review and the referees for constructive criticism. References 1 Mazi.P. Ann. Inst. Pasteur 1919 33 139. 2 Gupta U. C. Jame Y. W. Campbell C. A. Leyshon A. J. Nicholaichuk W. Can. J. Soil Sci. 1985 65 381. 3 Nielsen F. H. Nutrition Today 1992 3 6. 4 Gregoire D. C. Anal. Chem. 1987 59 2479. 5 Gregoire D. C. J. Anal. At. Spectrom. 1990 5 623. 6 Smith F. G. Wiederin D. R. Houk R. S. Egan C. B. and and Serfass R. E. Anal. Chim. Acta 1991 248 229. Vanhaecke F. Vanhoe H. Vandecasteele C. and Dams R. in Applications of Plasma Source-Mass Spectrometry eds. Holland G. and Eaton A. N. Royal Society of Chemistry 1991 p. 149. 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Russ G. P. and Bazan J. M. Spectrochim. Acta Part B 1987 42 49. Gregoire D. C. Spectrochim. Acta Part B 1987 42 895. Ward N. I. Abou-Shakra F. R. and Durrant S. F. Biol. Trace Element Res. 1990 26 177. Durrant S. F. Trends Anal. Chem. 1992 11 68. Evans E. H. and Giglio J. J. J. Anal. At. Spectrom. 1993 8 1. Tan S. H. and Horlick G. Appl. Spectrosc. 1986 40 445. Vanhaecke F. Vanhoe H. Dams R. and Vandecasteele C. Talanta 1992 39 737. Evans S. and Krahenbiihl U. Fresenius’ J . Anal. Chem. 1994 349 454. Gray A. L. Spectrochim. Acta Part B 1986 41 151. Faure G. Principles of Isotope Geology Wiley New York 2nd edn. 1986 p. 62. Heumann K. G. in Inorganic Mass Spectrometry eds. Adams F. Gijbels R. van Grieken R. Wiley New York 1988 p. 309. Gladney E. S. O’Malley B. T. Roelandts I. and Gills T. E. 1987 Compilation of elemental concentration data for NISTclinical biological geological and environmental Standard Reference Materials National Bureau of Standards Special Publication no 260-111. Anderson D. L. Cunningham W. C. and Mackey E. A. Fresenius’ J. Anal. Chem. 1990 338 554. Paper 4102995 J Received May 19 1994 Accepted July 14 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901249

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1255 Study of the Atomization of Boron in Electrothermal Atomic Absorption Spectrometry and Hollow Cathode Furnace Atomic Non-thermal Excitation Spectrometry Guy A. Wiltshire David T Bolland and David Littlejohn* Department of Pure and Applied Chemistry University of Strathclyde Cathedral Street Glasgow UK G 7 7XL Coating a total pyrolytic graphite tube with tungsten carbide or lanthanum carbide increased the optimum pyrolysis temperature of B from 850 to >22OO"C. Addition of a calcium-magnesium modifier to the boron solutions increased the pyrolysis temperature to 1200 "C whereas a titanium-ascorbic acid modifier had no significant effect. The lowest characteristic mass of B 0.8 ng was obtained with the calcium-magnesium modifier.The low temperature loss of boron without a modifier was investigated using dynamic secondary- ion mass spectrometry which confirmed that vaporization of boron species occurs above 900 "C. Boron atomic emission signals were obtained at < 800 "C by hollow cathode furnace atomic non-thermal extraction spectrometry (HC-FANES) with a 30 Torr helium plasma. Molecular dissociation and boron atom excitation apparently occurred through a one-step collisional process. The detection limit of B by He HC-FANES under non-optimum conditions was 71 pg (no modifier). The study suggests that atomization of B in electrothermal atomic absorption spectrometry (ETAAS) is likely to occur through molecular dissociation rather than solely by sublimation of B.The modifiers probably prevent low temperature dissociative desorption of B203 occurring at active carbon sites and so increased the optimum pyrolysis temperature to >850 "C. The poor detection limit of B in ETAAS is due to the inefficient thermal dissociation of the B-containing species (probably oxides and carbides) produced by dissociative desorption of B203. Also once formed B atoms apparently undergo a series of condensation-vaporization steps which cause a persistant plateau in the tail of the AAS signal and result in severe memory effects. Keywords Electrothermal atomization; boron; chemical modifiers; secondary-ion mass spectrometry; hollow cathode furnace atomic non-thermal excitation Spectrometry Few papers concerned with the determination of B by electro- thermal atomic absorption spectrometry (ETAAS) fail to men- tion the comparatively poor sensitivity of the method and the difficulties caused by memory effects. Jiang and Yao' reported that the sensitivity for B was so poor that it was difficult to obtain a signal for lOpgml-' B with atomization in an uncoated graphite tube.Many workers have attempted to improve the characteristic mass of B in ETAAS by addition of a chemical modifier or alteration of the graphite tube surface. Manning et aL2 added CaC1 as a modifier and suggested that the gain in sensitivity obtained was due to formation of more volatile calcium boride in preference to refractory boron carbide. Szydlowski3 evaluated other Group 2 elements and reported that addition of 1000 p.g ml-' Ba(OH)2 provided the greatest sensitivity improvement.Szydlowski3 concluded that the action of the modifier cation involved competition with B for formation of carbides in addition to the formation of the metal boride. Van der Geugten4 preferred addition of a Ca-Mg modifier as memory effects were less than for addition of Ba and a higher pyrolysis temperature could be used than with e.g. Ca alone. Jiang et a1.' also favoured the use of a Ca-Mg mixture. Some workers have noted that the form of the alkaline earth modifier has a small but noticeable effect on the improve- ment in the sensitivity of B obtained. Szydlowski3 reported that Ba(OH) was more effective than BaCl whereas Botelho at d6 noted that Ca(N03)2 was not as effective as CaCl in improving the characteristic mass of B.Other modifiers besides the Group 2 elements have been used to enhance the sensitivity of B. Goyal et aL7 proposed a combination of 0.25% m/v ascorbic acid and 10 pg ml-' Ti and Barnett et d8 used Ni for the determination of B in National Institute of Standards and Technology (NIST) stan- dard reference material (SRM) Orchard Leaves and Tomato Leaves. Some workers have improved the sensitivity of B by adding solutions which form chemically and thermally stable carbides on the surface of the tube prior to introduction of the boron- containing solutions. Refractory carbide coatings appear to cover defects in the pyrolytic graphite layer and minimize formation of analyte carbide^.^ Jiang and Yao' found that pre- treatment of graphite tubes with solutions containing La W or Ta improved the characteristic mass of B.Treatment with Ta was preferred as it gave a better tube lifetime and improved the precision of the B AAS signals. Luguera et a1.l' reported that pre-treatment of the graphite tube with Zr and addition of 1000pgml-' Ni to the B-containing solutions reduced memory effects and improved the characteristic mass of B to 88 pg. It has been suggested" that the formation of tungsten carbide or zirconium carbide linings on a graphite tube can lower the free oxygen content of the gas phase during atomiz- ation because of the catalytic effect of the refractory metal carbides on the oxidation of carbon leading to preferential formation of oxides or oxycarbides. Such a reduction in the partical pressure of oxygen could have beneficial effects on the atomization of elements such as B which form thermally stable oxides.Only a few papers have discussed in any detail the mechan- ism of B atom formation and the effect of the modifier on the reactions involved. Jiang et ~ 1 . ' ~ proposed that B203 is reduced on the tube surface by carbon at temperatures greater than 1600"C to produce B(sl which is vaporized at temperatures greater than 2300°C to form B atoms. This was supported by the observation that the B atom appearance temperature of 2295 "C closely resembled the melting point of boron (2300 "C) although sublimation of boron does not occur until 2550 "C.' Goyal e t aL7 calculated the activation energy for B atom formation to be 93 +2 kcal mol-' which is in good agreement with the literature value of 97 -+_ 3 kcal mol-' for sublimation of solid boron.It was also suggested that B,O could react with carbon to produce B,C,, which was identified by XRD.12 Jiang et also used XRD to investigate the modifying effect of Sr on atomization of B. They proposed that SrO reacts with B203 and B4C to give SrB which can be formed at tempera- tures greater than 1600°C and was identified by XRD albeit at boron concentrations 1000-fold greater than levels normally considered in ET-AAS. The presence of Sr did not alter the1256 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 final step in the atomization mechanism as SrB decomposed to B(sl at temperatures between 1600 and 2300 "C with volatil- ization to form B atoms above 2300°C.However as the sensitivity of the B AAS signals was only improved 10-fold by addition of Sr it must be concluded that the reactions of B2O3 are not dramatically affected by the presence of the modifier. Indeed with the exception of the study by Luguera et a!.," most modifiers proposed for B only give a 3 to 10-fold improvement in the characteristic mass value. In this study the impact of various modifiers on the sensi- tivity of B has been assessed. Dynamic SIMS has been used to analyse boron-containing deposits on total pyrolytic graph- ite platforms heated to different temperatures. The modifying effect of Ca-Mg has also been investigated to establish if it performs a similar role to that of Sr.12 Of particular interest is the temperature at which most of the boron is volatilized from the graphite surface.Various w ~ r k e r s ~ ~ ' ~ ' ~ have sug- gested that the formation of refractory boron carbide is the main reason for the comparatively poor detection limit of B by ETAAS but this has not been substantiated and so is also considered in this study. Experimental Boron Atomic Absorption Spectrometry Instrumentation A Unicam PU 9400 atomic absorption spectrometer PU 9390 graphite furnace atomizer and PU 9380 autosampler were used. The PU 9390 atomizer was equipped with either total pyrolytic graphite (TPG) or pyrolytic graphite coated electro- graphite tubes for AAS measurements with wall atomization. A PU 9385 furnace autoprobe was attached for probe atomiz- ation studies. In this case a slotted TPG tube was fitted to the PU 9390 atomizer and sample vaporization was from the head of a 'ridge' pyrolytic graphite coated graphite probe inserted through the slot and into the pre-heated tube.13 The PU 9400 atomic absorption spectrometer was operated with a bandpass of 0.5 nm and atomic absorption measurements were made at 249.77 nm.The PU 9390 atomizer programme is given in Table 1 which indicates the ranges of pyrolysis and atomization temperatures studied. The PU 9400 atomic absorption spectrometer was equipped with D background correction which was used for all measurements. Reagents Solutions containing 1000 pg ml-' B were prepared by dissolv- ing 0.8826 g of sodium tetraborate (AnalaR grade Merck Poole UK) in 100ml distilled water. The chemical modifiers were either added to the B-containing solutions and injected into the atomizer tube or were applied to obtain a uniform coating on all of the tube prior to use.The calcium-magnesium mixed modifier solution4 was added to give concentrations of 100pgml-' Ca and 200 pg ml-' Mg in the B solution. The modifier was prepared by dilution of SpectrosoL (Merck) solutions of the nitrate salts. The titanium-ascorbic acid mixed modifier was added to give concentrations of 10 pg ml-' Ti and 0.25% m/v ascor- bic acid as specified by Goyal et ~ 1 . ~ A stock solution of the Table 1 PU 9390 atomizer programme for B Ramp rate/ Ar gas flow/ Step TemperaturerC "C s - l Hold time/s ml min-' Dry 125 10 40 200 Pyrolysis 300-2500 100 30 200 Atomize 2400-2850 FP* 3 0 Clean 2800 FP* 2 200 * FP=full power (>2OOO"C s-').modifier containing 100 pg ml-' Ti and 2.5% m/v ascorbic acid was prepared from AnalaR titanyl potassium oxalate and ascorbic acid (Merck). The mixed modifier solutions were normally added to the B-containing solutions prior to injection into the PU 9390 atomizer. Alternatively separate aliquots of the modifier and analyte-containing solutions could be added directly into the tube by the autosampler. Some TPG tubes were pre-treated to obtain a refractory metal carbide coating as reported by Fritzsche et ~ 1 . ' ~ and Runnels et u1." The tubes were soaked for 18 h in a solution of 7.8% m/v sodium tungstate (4.3% m/v W) prepared by dissolving 7.8 g of AnalaR sodium tungstate (Merck) in 100 ml distilled water. The tubes were then dried for 1 h at 100°C and conditioned in the PU 9390 atomizer by heating to 2200 "C for 5 s to form the tungsten carbide ~ 0 a t i n g .l ~ A coating of lanthanum carbide" was prepared by injecting 600 pg La into a TPG tube. The La was added as 20pl of a 3% m/v La solution prepared by dissolution of 8.185 g lanthanum nitrate in 100 ml distilled water. The tube was heated to 100 "C to dry the solution and then to 1950 "C for 10 s. The lanthanum carbide coating was not as stable as that of tungsten carbide so it was enforced by injecting 30 pg La along with the B-containing solution at the start of each atomization cycle. Dynamic SIMS Analysis of Graphite Surfaces A VG SIMSLAB Mark 111 spectrometer was used to analyse the surface of pyrolytic graphite platforms on to which B solutions had been deposited dried and pyrolysed.The SIMSLAB spectrometer was evacuated to 10-9Torr and a surface area of 400 x 400 pm2 was irradiated by a 10 keV Ga ion beam of 1 pm diameter. A scanning electron microscope incorporated into the system enabled the surface to be inspected. Specimens were prepared by depositing a B-containing solution on to a TPG platform located in the normal manner in the TPG tube of a PU 9390 atomizer. A number of platforms was used to investigate the effect of the pyrolysis temperature on deposits containing 160 ng B or 160 ng B plus 2 pg Ca and 4 pg Mg (ie. the Ca-Mg modifier). Normally 20p1 of the appropriate solution was dried at 140 "C for 60 s. The tube temperature was then raised to pyrolyse the deposit at the required temperature with a heating rate of 100°C s-' and a hold time of 30 s.The pyrolysis temperatures investigated were 200 500 900 1200 and 2000°C for B with and without the modifier and 1600°C for B with the Ca-Mg modifier only. The mass of B used for the secondary-ion mass spectrometry (SIMS) analysis was similar to the levels measured by conven- tional ETAAS. The SIMS analyses were also performed to verify the X-ray diffraction (XRD) work of Jiang et a l l 2 concerning the effects of a Sr modifier on the atomiz,ation of B. In this case the platforms contained 12 pg B and 20 pg Sr (the mass of B used by Jiang et a l l 2 ) or 200 ng B and 20 pg Sr (a mass of B similar to that used for measurements by conventional ETAAS) and a 1600°C pyrolysis temperature was used. Before obtaining SIMS spectra care was taken to ensure that a representative part of the deposit on the platform was selected for analysis.Furnace Atomic Non-thermal Excitation Spectrometry (FANES) A hollow cathode (HC)-FANES atomizer (Carl Zeiss Jena Germany) was operated in conjunction with a Spectrometrics SMI 111 echelle spectrometer. An oscillating quartz refractor plate was positioned behind the entrance slit of the spec- trometer to provide automatic background correction of emis- sion signals by wavelength modulation. Details of the atomizer and spectrometer are described el~ewhere.'~.'~ The FANES atomizer was fitted with a pyrolytic graphite coated electro-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL 9 1257 graphite tube which also acted as the cathode in conjunction with a pin anode positioned outside one end of the tube.Sample solutions (20 p1) were injected manually into the tube and dried at atmospheric pressure. The atomizer was then sealed evacuated and a pressure of 25 or 30 Torr He established under gas flow conditions. A He HC discharge was initiated at 88 mA during the pyrolysis step and the tube heated to a temperature in the range 650-1900 "C to pulse-vaporize the B-containing deposit into the plasma. For most measurements 1360 "C was used (Table 2). Transient atomic emission signals were measured at the B wavelength of 249.77 nm. No modifiers were used in the FANES experiments. Results Effects of Tube Type and Chemical Modifiers on Atomization of B The pyrolysis and atomization temperature curves of B for wall atomization in TPG tube are shown in Fig.1. The temperature programme given in Table 1 was used. When the pyrolysis temperature was varied the atomization temperature was set at 2700°C and when the atomization temperature was varied the pyrolysis temperature was set at 850 "C. At pyrolysis temperatures above 850"C there was a decrease in the peak height and integrated atomic absorption signals of B suggest- ing either loss of volatile B species or conversion into a more refractory B compound. The optimum pyrolysis temperature was similar to that reported by Botelho et aL6 but lower than the value of 1200°C indicated by Jiang et a1.l' The increase in the sensitivity of B with increasing atomization temperature Table 2 HC-FANES temperature programme for B Temperature/ Ramp rate/ Hold time/ Helium pressure/ Step "C "C s-l S Torr Dry 100 10 33 760 Pyrolysis 170-450 50 15 25 or 30" Atomize 1360 2000 5 25 or 30 Clean 2130 2000 2 25 or 30 * Discharge initiated at 88 mA during the pyrolysis step.0.6 I (a) 1 0.5 t 0.4 t - . . E 0 ' +? s I I I I I I I I 400 500 600 700 800 900 1000 1100 1200 Char temperaturePC 0.6 0.5 - 0.4 0.3 0.2 0.1 0 2300 2400 2500 2600 2700 2800 2900 ( b ) B Atomization temperaturePC Fig. 1 (a) Pyrolysis temperature and (b) atomization temperature plots for 140 ng B (no modifier see Table 1) A peak height and B integrated absorbance illustrated in Fig. 1 is similar to the results reported by other worker^.^^^*^^ The pyrolysis-atomization characteristics of B are unusual in that there is a large difference between the optimum pyrolysis temperature and the atom appearance temperature.A typical B atomic absorption signal obtained for wall atomization of 160 ng B in a TPG tube is given in Fig. 2(u). The atomizer programme in Table 1 was used with pyrolysis and atomization temperatures of 850 and 2750 "C respectively. Fig. 3 shows the pyrolysis and atomization curves of 300 ng B for probe atomization in a slotted TPG tube in the PU 9390 atomizer. The atomizer programme given in Table 1 was used with the exception that a dry temperature of 150°C was required to evaporate the lop1 droplet injected on to the pyrolytic graphite coated graphite probe. When the pyrolysis temperature was varied the pyrolysis temperature was set at 1150°C. The results shown in Fig.3 are similar to those obtained for wall atomization with the exception that the optimum pyrolysis temperature for probe atomization is higher than that of wall atomization by about 300°C. This does not indicate greater stability of the B-containing species with probe atomization but is owing to the fact that the probe is heated indirectly. It is quite common to select a higher tube tempera- ture at the pyrolysis step when the sample is deposited on a probe rather than on the tube wall.13 At a tube temperature of 115O"C it is likely that the probe temperature is around 800-900°C. An example of a B atomic absorption signal obtained for probe atomization of 200 ng B is given in Figure 2(b). Char and atomization temperatures of 1150 and 2750 "C respectively were used.Comparison of the signals given in Fig. 2(u) and (b) indicates that the peak obtained by probe atomization is sharper and has an earlier appearance time. These features are explained by the faster heating rate of the probe compared with the tube wall (leading edge) and the greater diffusional loss of the atoms in the slotted TPG tube at a higher temperature (trailing edge). Both signals exhibit a plateau on the tail of the peak which persists for many seconds after the peak maximum. The persistence of B constitutes a severe memory effect in analytical studies. Frequent blank atomizations and a 'first peak rejection' procedure are required for accurate and precise analysis. Interestingly when 300 ng B was atomized from a probe in a slotted tube at 2750°C and the probe heated a second time to the same temperature in another tube comparison of the absorbance signals indicated that 86% of the B was vaporized initially.The characteristic mass values of B obtained by wall and probe atomization were the same 3 ng. Fig.4 shows the pyrolysis temperature plots of B for wall atomization in a TPG tube when (i) the tube was impregnated with W solution to form a tungsten carbide coating (ii) a La solution was injected onto the tube wall and heated to form a lanthanum carbide coating (iii) a Ti-ascorbic acid modifier was added with the B-containing solutions and (iu) a Ca-Mg modifier was added with the B-containing solutions. In each case the atomizer conditions for wall atomization in a TPG tube without chemical modification were used.When the tube surface was treated to form either a tungsten or lanthanum carbide coating the thermal stability of the B-containing species was greatly improved. A decrease in the B pyrolysis curve did not occur until temperatures in excess of 2200°C for tungsten carbide and there was only a slight decrease in the B signal from 1600-2000 "C with the lanthanum carbide coating. The addition of Ti and ascorbic acid to the B-containing solutions improved slightly the sensitivity of B but reduced the optimum pyrolysis temperature However with the Ca-Mg modifier a pyrolysis temperature of 1200°C could be used before the B signal decreased; there was also an improvement in sensitivity. The characteristic mass values of B obtained with the modifiers are given in Table 3.Typical AAS signals of B (with the modifiers) are given in Fig. 2(c)-(f). The mass of B injected was 100 ng for WC 120 ng1258 0.32 0.24 0.16 0.08 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 ( e) - - - 0.32 0.24 0.16 0.08 Background P I - . . Background 0 0.6 1.2 1.8 2.4 0.04 I 0.80 0.60 0.40 0.20 Background 0 0.6 1.2 1.8 2.4 Time/s Fig. 2 Atomic absorption signals for (a) 160 ng B; no modifier; (b) 200 ng B no modifier probe atomization; (c) 100 ng B WC coating; ( d ) 120 ng B Lac coating; (e) 120 ng B Ti-ascorbic acid; and (f) 160 ng B Ca-Mg (all at 2750°C) 0.5 1 0.3 0.2 0.1 - a c " 0 - OJ 450 550 650 750 850 950 1050 1150 1250 1350 1450 a 0.6 e $ a [ ( b ) 1 Char temperaturePC 0.5 0.4 0.3 0.2 0.1 0 '' B 1 2500 2600 2700 2800 2900 Atom i za t i on t e m pe rat u r e/"C Fig- 3 (a) Pyrolysis temperature and (b) atomization temperature plots for 300 ng B probe atomization (no modifier); A peak height; and B integrated absorbance for L a c and Ti-ascorbic acid and 160 ng for the Ca-Mg modifier.In each case the atomization temperature was 2750°C and the pyrolysis temperatures were the optima indi- cated in Table 3. Each of the signals exhibit similar character- istics. There is a rapid increase in the signal from the peak start time of about 0.6 s to a peak maximum which occurs at 0.7 or 0.8 s. The signal then decays to form a slowly decreasing plateau which is 40-60% of the peak height absorbance. The Ca-Mg modifier was selected for further investigation on the basis that it gave a reasonable improvement in the pyrolysis temperature and the characteristic mass of B and it was more convenient to use than the formation of carbide coatings.Dynamic SIMS Analysis of Graphite Surfaces The pyrolysis-atomization curves of B (Fig. 1) suggest that either loss of volatile B species or conversion of B into a more refractory form occurs at pyrolysis temperatures above 850 "C. To investigate the reason for the decrease in the B signal aliquots of B-containing solutions were dried on a TPG platform pyrolysed at a temperature in the range 200-2000 "C and the residue analysed by dynamic SIMS with a VG SIMSLAB Mark I11 spectrometer. The mass spectra of the deposits with and without the Ca-Mg modifier are illustrated in Fig. 5 with the relevant peaks marked to identify the species corresponding to that particular atomic mass number.At 200 and 500°C (Fig. 5 ) B peaks were present in both sets of spectra at mass numbers 10 and 11. The B peaks were observed in the spectra for platforms heated to 900 and 1200°C whenJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 0.4 0.3 0.2 0.1 1259 ( C) - - - - 1 1 1 1 1 1 1 1 1 1 > 0 I I L t I - 1.2 0.6 0.5 - 1.0 0.4 - (b) ( a ) - 0.3 - 0.2 - .q v.a g o O n 2 1 O . ' 0 200 600 1000 1400 1800 2200 2600 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 0 I 1 1 1 1 0' I I L t I 0.5 0.4 0.3 0.2 0.1 Fig. 4 Pyrolysis temperature plots for B with different modifiers (a) 200 ng B. WC coating; (b) 100 ng B LaC coating; (c) 120 ng B Ti-ascorbic acid; and (d) 80 ng B Ca-Mg (atomization temperature of 2750°C for all) A peak height and B integrated absorbance Table 3 Optimum pyrolysis temperatures and characteristic mass values of B with different chemical modifiers using a TPG tube Modifier Optimum pyrolysis temperature/"C Characteristic(b' mass/ng Peak height plateau absorbance* None Lac Coating WC Coating Ti-Ascorbic acid Ca-Mg 8 50 2200 2100 600 1200 3 .O 2.7 2.5 1.3 0.8 3.3 2.2 2.4 2.3 2.5 * For signals giving about the same peak height absorbance.I x i o 4 1 x 1 0 ~ C (a) - 1 XI02 r lv) lo $ 1 v) + I xi04 I xi03 1 x 102 I0 1 5.0 9.0 13.0 17.0 21.0 25.0 29.0 33.0 37.0 41.0 45.0 m/z Fig. 5 Dynamic SIMS analysis of 160 ng B on a TPG platform for a pyrolysis temperature of 500 "C; (a) no modifier and (b) 2 pg Ca plus 4 Mg I x lo3 1 x 102 I0 1 5.0 9.0 13.0 17.0 21.0 25.0 29.0 33.0 37.0 41.0 45.0 m/z Fig.6 Dynamic SIMS analysis of 160 ng B on a TPG platform for a pyrolysis temperature of (a) 1200 "C no modifier and (b) 1600 "C and 2 pg Ca plus 4 pg Mg the Ca-Mg modifier was present but not in the absence of the modifier [see Fig. 6(u) for 1200"C]. At 1600"C B could just be identified in the mass spectrum of the deposit containing the Ca-Mg modifier [Fig. 6(b)] but by 2000°C (Fig. 7) no B was detected. The SIMS measurements confirm the results of the pyrolysis-atomization curves obtained by ETAAS and1260 JOURNAL OF ANALYT1CA:L ATOMIC SPECTROMETRY NOVEMBER 1994 1x104 v) v) E 1x102 E3 10 C 0 0 3 1 ,- (a) - BlIc 5.0 9.0 13.0 17.0 21.0 25.0 29.0 33.0 37.0 41.0 45.0 m/z Fig. 7 Dynamic SIMS analysis of 160 ng B on a TPG platform for a pyrolysis temperature of 2000 "C and 2 pg Ca plus 4 pg Mg indicate that B is lost mainly as a volatile species at tempera- tures above 800-900 "C in the absence of a modifier.Addition of Ca-Mg to the B-containing solutions stabilizes the B to at least 1200°C. Jiang et a l l 2 studied the mechanism of Sr as a modifier for B by XRD and suggested that volatilization losses of B,03 and production of refractory B4C were decreased by formation of SrB which was detected on the surface of a platform heated to 1600 and 1800 "C. The SIMS analyses of deposits of 12 pg B and 20 pg Sr (as used by Jiang et all2) and 200 ng B and 20 pg Sr are shown in Fig. 8 for pyrolysis at 1600 "C. The SIMS results confirm that Sr is an efficient stabilizing modifier for B.HC-FANES of Boron Fig. 9 illustrates an atomic emission signal for 20 ng B (as Na2B407) recorded at 249.77 nm. The He discharge current was 88 mA and the analyte was vaporized at 1360°C (see Table2 for other details). A plot of the B atomic emission intensity against discharge current is given in Fig. 10. A linear dependence of intensity on current is demonstrated suggesting a one-step collisional process is responsible for excitation in the He discharge." This is not unusual for HC-FANES when the analyte is introduced into the plasma as an atom. However at the atomizer temperature used (1360 "C) thermal atomiz- ation of B is not possible. This leads to the conclusion that B-containing molecules are vaporized from the tube surface I xio3 1 x lo2 7 10 v) w 1 3 '5.0 9.0 13.0 17.0 21.0 25.0 29.0 33.0 37.0 41.0 45.0 m Ga g 1x10~ C ( b ) i ~ j 1x105 - cn I x10' I x lo3 1 x 102 10 1 0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 m/z Fig.8 Dynamic SIMS analysis of (a) 12 pg B and (b) 200 ng B on a TPG platform with 20 pg Sr; pyrolysis temperature 1600 "C I2O0 1 1000 v) C 4- .- 800 2 4- .- 600 Y a 400 Q v) 22 ; 200 m o C 0 .- -200 L ' I I I I I I 'OL. 9 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 Tim& Fig. 9 HC-FANES signal of 20 ng B (as Na,B,07) at 249.77 nm 30 Torr He; 88 mA pyrolysis temperature 230 "C; atomization temperature 1360 "C 6o 3 30 40 2ol 10 0 10 20 30 40 50 60 70 80 Discharge current/mA Fig. 10 Plot of atomic emission intensity (0 peak height and 0 integrated) of 4 ng B (as Na,B,O,) against discharge current for HC-FANES 249.77 nm; 30 Torr He; pyrolysis temperature 230 "C; atomization temperature 1360 "C and these molecules are dissociated in the plasma to produce excited B atoms in a single step. Fig.11 shows that for B as H3B03 the atomic emission intensity decreased at pyrolysis temperatures above 225 "C at 30Torr He indicating loss of B species from the atomizer tube. Okamoto et all9 also reported that B as H3B03 was lost from a tungsten boat at low temperatures during the dry step when ETV-ICP-AES was used for determination of B. A detection limit of 71 pg B was obtained by HC-FANES (for B as Na,B,07) using peak height measurements and the conditions given in Table2. This is a large improvement on the detection limit of 8 ng (no modifier) obtained by ETAAS with a pyrolytic graphite coated graphite tube but poorer than the value of 2 pg reported by Riby and Harnly2' for HA-FANES. The atomizer used2' had a static gas system rather than the continuous flow arrangement of HC-FANES which reduces the residence time of the atoms in the discharge and hence impairs the detection limit.The measurements obtained by HC-FANES confirmed that low temperature vaporization of B species occurs from deposits of Na2B407 or H,B03.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1261 140 $ 120 : 100 2 C 0 8 0 - 60- .- m I? 40- 20 I6O 5 I4O - - - - A -I 120 v) - 100 C - 8 0 2 -60 $ I! -40 $ -0 0 - '-1 0 1 I I I I I 10 150 200 250 300 350 400 450 Pyrolysis temperaturePC Fig. 11 Pyrolysis temperature plot for 40 ng B (as H3BO3) by HC-FANES; 249.77 nm; 25 Torr He; 88 mA; and atomization tempera- ture 1360°C Discussion Without a modifier the pyrolysis temperature plot for B (as Na2B407) shows a decrease in the AAS signal above 850°C (Fig.1). The SIMS results confirmed that above 900°C a substantial amount of B was vaporized from a TPG platform. Any remaining B in the form of oxide carbide or whatever was present at levels lower than the detection limit of B by SIMS which is equivalent to <10 ng B injected on to the platform from the spectra in Fig. 5. Also when H3B03 was used to prepare solutions for HC-FANES a significant reduction in the B atomic emission signal occurred when pyrolysis temperatures above 225 "C were used at 25 Torr He. Treatment of the graphite tube with tungsten or lanthanum solutions prevented the volatilization of B species at tempera- tures in excess of 2000 "C.Addition of calcium and magnesium nitrates as modifiers also extended the maximum pyrolysis temperature of B but only to 1200°C. It is possible that these modifiers act in different ways. The presence of different types of carbon sites on the surface of a graphite tube has been postulated by Prell et ~ 1 . ~ ' The formation of tungsten or lanthanum carbide coatings could block the carbon sites at which low temperature vaporization of B species occurs. The B would then be retained until temperatures were in excess of the boiling point of B203 (1860"C).22 However the results in Fig. 4 also support the hypothesis that vaporization of B203 may be prevented by reaction with the tungsten or lanthanum carbide coating on the tube surface as AAS signals of B were obtained up to pyrolysis temperatures of 2500°C for WC and at least 2200°C for Lac.With lanthanum modification the sensitivity of the B signals improved at least two-fold when the pyrolysis temperature was 2 1200 "C which suggests that the form of the modifier on the tube surface either influences its ability to prevent loss of B species or enhances the atomization efficiency of B. With Ca-Mg as modifier it is possible that a calcium and/or magnesium borate was formed which was stable up to at least 1200°C. When the borate decomposed B203 was probably formed which vaporized at the active carbon sites referred to above. Alternatively the calcium and magnesium oxides pro- duced by decomposition of the nitrate modifiers or the borates if formed could have blocked the active sites until vaporization or dissociative desorption of MgO and CaO occurred." Vaporization of B203 would then be possible. The SIMS results indicated that loss of B at pyrolysis temperatures above 1200 "C coincided with vaporization of Mg and Ca.By 2000 "C no B was detected on the TPG platform by SIMS and most of the Mg and a significant fraction of Ca had also been vaporized. The pyrolysis temperature plot for B in the presence of the Ca-Mg modifier shows a similar trend to that obtained with lanthanum modification. At 1100-1200 "C the B sensi- tivity is two-fold better than at e.g. 600"C which confirms that thermal alteration of the Ca-Mg modifier is necessary.Although the W La and Ca-Mg modifiers prevented low temperature vaporization of B203 the characteristic mass of B was only improved two- or three-fold which implies a minor effect on the atomization efficiency. The results and conclusions of t h s study are different in many respects to those of Jiang et al.'*12 who used XRD to investigate the atomization of B. Although solutions of B were prepared from H3B03 low temperature loss of B was not reported by these workers and a decrease in the B AAS signal did not occur until pyrolysis temperatures greater than 1200"C.12 The main conclusion of the XRD measurements which were not interpreted quantitatively was that B203 existed on the graphite platform at char temperatures up to 1800 "C.Jiang et a1.12 postulated that at temperatures greater than 1627"C Bz03 was reduced by C(s) to give B(s) on the surface of the graphite tube or platform. Atomization of B then occurred at 22300°C. They also suggested that B203 could react with carbon to form B4C a compound they claim to have identified by XRD after pyrolysis at 1600"C but not at lower temperatures despite the fact that AGO values for the reaction were negative over the range 1000-3000 K.12 Unfortunately the SIMS measurements reported here do not support these proposals. Detectable levels of B (in any form) were not observed at 2 900 "C which implies that at best only a small amount of B203 is converted to BqC(s) or B(s) by reaction with carbon. Most of the oxide is vaporized from the graphite surface at a temperature that depends on whether or not a modifier is used and on the nature of the modifier.also used XRD to study the effect of strontium nitrate as a modifier for B. The presence of 1.5 mg ml-1 Sr in the B solutions only extended the maximum pyrolysis tempera- ture by 200"C but gave about ten-fold improvement in the sensitivity of the B AAS signals. From the XRD measurements Jiang et al." proposed that SrB was formed as an intermediate in the production of B(s). The B sensitivity was improved because formation of SrB prevented the formation of B4C. The XRD data implied that SrB was formed above 1400"C but decomposed [to B(s)] by 2000 "C. The atomization of B(s) then occurred as before at 22300°C. The results obtained in this study (Fig.8) confirmed that Sr is an effective modifier at the level of B used by Jiang et (12 pg) and at a level of B more normally determined by ETAAS (200 ng). As Sr(NO& will decompose on heating to SrO and SrO acts similarly to CaO and MgO in a graphite furnace,21 it is likely that the alkaline earth oxides have a similar modifying action. Indeed a characteristic mass of 0.4 ng B can be derived from the results of Jiang et al.," which is only two-fold better than the value given in Table 3 for B with the Ca-Mg modifier. No evidence for the production of SrB MgB or CaB was obtained by SIMS although complete fragmentation of these compounds may have been caused by the ion beam. In contrast significant levels of the alkaline earth oxides were always indicated by the SIMS spectra.Information about the atomization of B can be obtained from the AAS signals. The examples shown in Fig. 2 all exhibit similar characteristics. This suggests that the modifiers do not change the atomization process significantly irrespective of their influence on the low temperature vaporization of B203. The signals are characterized by a sharp rise to the peak maximum and a decay of the signal to a plateau which is about half the peak height absorbance except with probe atomization when the plateau is a quarter of the peak height absorbance. The shape of the signals suggest that two processes are involved in atom formation. The initial production of B atoms is rapid and could be indicative of the sublimation of B(s) as proposed by Jiang et a2.I' Alternatively B atoms may be formed by thermal dissociation of B-containing molecules.This is unlikely to be direct dissociation of B20,(g) but may Jiang et1262 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 involve dissociation of molecules formed by dissociative desorption of B203 on the graphite surface. The peak height absorbance of B is strongly dependent on the atomization temperature increasing almost linearly over the temperature range 2400-2850°C (see Figs. 1 and 3). Vaporization in a probe atomizer apparently enhances the atomization efficiency of B giving a characteristic mass equivalent to that obtained by wall atomization in an unslotted tube. Normally analytes that form atoms by direct vaporization of the element from the tube surface show about four-fold poorer sensitivity by probe atomi~ation.~~ In contrast elements which form atoms by thermal dissociation of molecules in the gas phase exhibit better sensitivity as the atomization temperature is increased and the characteristic mass becomes equivalent to or better than that obtained by wall atomi~ation.'~ Given these obser- vations it is difficult to discount an atomization mechanism which involves vapour-phase dissociation of B-containing mol- ecules produced by decomposition of Bz03 at specific sites on the graphite surface.However although vaporization of the B species is relatively efficient (as indicated by the probe and SIMS studies) atomization is not. Also once formed the B atoms have a high affinity for the graphite tube and re-condense initiating a series of vaporization-condensation steps which cause the plateau in the signal and account for the memory effects encountered in the determination of B.A similar mechanism has been proposed for atomization of Dy which also exhibits severe memory effects in ETAAS.24 The various modifiers examined in this study do not influence the B atomization sequence as their main action is in preventing low temperature vaporization of B203 (as discussed above). It is possible that the procedure used by Luguera et a1.l' to treat the graphite tube with Zr gave a more complete carbide coating than achieved with W in this study. Consequently the Zr coating prevented significant reaction of the B atoms with the tube surface reducing the memory effect problem.No AAS signals of B were presented by Luguera et a1.l' and a pyrolysis temperature study was not reported. However the character- istic mass obtained for B with the Zr-treated tube and addition of Ni as a modifier to the B solutions was ten-fold lower than that reported here for modification with Ca-Mg. Luguera et a1.l' suggested that formation of nickel boride reduced the tendency of B to react with graphite. However it is more likely that it was low temperature vaporization of B203 that was prevented by Ni. If catalytic dissociation of B203 occurs at active sites on the graphite surface it may be possible to detect the production of B-containing molecules (oxides or carbides) in the gas phase by real-time mass spectrometry as used by PreU et a1.21p25 It seems that one of the best techniques for trace determi- nation of B is FANES where B atom formation is enhanced by the non-thermal collisional processes in the He discharge. As sensitive B atomic emission signals can be obtained at an atomizer tube temperature as low as 80O0C it is clear that the thermal mechanisms responsible for atom formation in ETAAS are not important in FANES.Molecular dissociation and atom excitation can apparently be achieved in a one-step process probably through collision with high energy electrons,'* giving a B detection limit 10-100 times better than that of ETAAS. Clearly further study of the excitation processes in the He discharge of FANES is merited. Although not applied in this study modifiers such as Ca-Mg will be useful in FANES to minimize low temperature losses of B species and allow pyrolysis at temperatures 2 1000 "C.Another technique which should prove useful for B is ET-ICP-MS. The authors thank E. Maydell for assistance with the production of SIMS spectra and acknowledge the award of a CASE studentship by the SERC to D.T.B. and the award of an SAC Research Studentship by the Analytical Chemistry Trust Fund of the Analytical Division of the Royal Society of Chemistry to G.A.W. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 References Jiang Y. and Yao J. Yankuang Ceshi 1988 7 323. Manning D. C. Fernandez J. J. and Peterson G. E. Paper presented at 3rd Annual FACSS Meeting Philadelphia USA 1976. Szydlowski F. J. Anal. Chim. Acta 1979 106 121. Van der Geugten R. P. Fresenius' 2. Anal. Chem. 1981,306 13. Jiang Y. Yao J. and Huang B. Fenxi Huaxue 1989 17 456. Botelho G. M. A. Campos R. C. and Curtius A. J. paper presented at 5th BNASS Meeting Loughborough University of Technology UK 1990. Goyal N. Dhobale A. R. Patel B. M. and Sastry M. D. Anal. Chim. Acta 1986 182 225. Barnett N. W. Ebdon L. Evans E. H. and Ollivier P. Anal. Proc. 1988 25 233. Runnels J. H. Merryfield R. and Fisher H. B. Anal. Chem. 1975 47 1258. Luguera M. Madrid Y. and Camara C. J. Anal. At. Spectrom. 1991 6 669. Volynsky A. B. and Sedykh E. M. J. Anal. At. Spectrom. 1989 4 71. Jiang Y. Yao J. and Huang B. Acta Chim. Sin. (Engl. Ed.) 1989 5. 437. Corr; S P. and Littlejohn D. J. Anal. At. Spectrom. 1988,3 125. Fritzsche H. Wegscheider W. Knapp G. and Ortner H. M. Talanta 1990 26 219. Runnels J. H. Merryfield R. and Fisher H. B. Anal. Chem. 1975 47 1258. Littlejohn D. Carroll J. Quinn A. M. Ottaway J. M. and Falk H. Fresenius' 2. Anal. Chem. 1986 323 762. Baxter D. C. Nichol R. Littlejohn D. Ludke Ch. Skole J. and Hoffmann E. J. Anal. At. Spectrom. 1992 7 727. Falk H. Hoffmann E. and Ludke Ch. Prog. Anal. At. Spectrosc. 1988 11 417. Okamoto Y. Sugawa K. and Kumamaru T. J. Anal. At. Spectrom. 1994 9 89. Riby P. G. and Harnly J. M. J. Anal. At. Spectrom. 1993,8,945. Prell L. J. Styris D. L. and Redfield D. A. J. Anal. At. Spectrom. 1991 6 25. Handbook of Chemistry and Physics 63rd edn. ed. Weast R. C. and Astle M. J. CRC Press Boca Raton Fl. USA 1982. Ajayi 0. O. Ansari T. M. and Littlejohn D. J. Anal. At. Spectrom. 1992 7 689. Chaudhry M. M. Littlejohn D. and Whitley J. E. J. Anal. At. Spectrom. 1992 7 29. Prell L. J. Styris D. L. and Redfield D. A. J. Anal. At. Spectrom. 1990 5 231. Pa per 4/03 9591 Received June 30 1994 Accepted July 22 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901255

出版商:RSC    

年代:1994

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1263 Determination of Boron by Electrothermal Atomic Absorption Spectrometry Testing Different Modifiers Atomization Surfaces and Potential lnterferents Gloria M a A. Botelho lnstituto de Projetos Especiais Centro Tecnologico do Exercito 23020-4 70 Rio de Janeiro Brazil Adilson J a Curtius Departamento de Quimica da Universidade Federal de Santa Catarina 88040-9 70 Florianopolis SC Brazil Reinaldo C. Campos" Departamento de Quimica Pontificia Universidade Catolica Rio de Janeiro 22453-900 Rio de Janeiro RJ Brazil The determination of B by electrothermal atomic absorption spectrometry (ETAAS) was investigated. Calcium (as chloride and nitrate) Ni (as chloride and nitrate) Mg Y and La (all as chloride) were tested as modifiers.Considering pyrolysis temperature and sensitivity Ca as chloride showed the best performance. Mineral acids and common salts led to severe interference effects. A long tail in the absorption pulse was always observed and a blank had to be performed between consecutive measurements in order to avoid memory effects. Boron was successfully determined by ETAAS in two National Institute of Standards and Technology (NIST) standard reference material (SRM) digests NIST SRM 1570 Spinach Leaves and NET SRM 1571 Orchard Leaves after a separation step using the standard additions technique for calibration. Atomization from surfaces other than the pyrolytic graphite coated graphite tube wall did not lead to any improvement in the analytical performance. Keywords Boron; electrothermal atomic absorption spectrometry; modifiers; atomization surfaces; in t erferen ts Boron cannot be determined sensitively by atomic absorption spectrometry (AAS) and its determination is subject to several interferences. Consequently a separation and pre- concentration step is usually necessary involving solvent extraction as used by Horta and Curtius' and reviewed by Welz,2 or volatilization of B as methylborate3 or fl~oride.~ Boron determination by flame AAS can then be performed.In the electrothermal technique chemical modifiers must be used Ca,' Ca and Mg,6 Ba Mg Sr7 and even Ti8 and La9 have been proposed. In this work different modifiers are compared considering not only the sensitivities and pyrolysis tempera- tures achieved but also their tolerance to interferences.Ex per imen t a1 Reagents and solutions All analytical solutions of B were prepared by convenient dilutions of a 5000 pg ml-l stock solution (Merck No. 19810) with de-ionized water. The modifiers were prepared from lo00 pg ml-I stock solutions as chloride (Ca Mg Ba and Ni) or nitrate (Ca and Ni). The Pd modifier solution was obtained from the metal powder as described by Schlemmer and Welz." All other solutions were prepared by convenient dissolution of analytical reagent grade salts or oxides made up to a known volume with de-ionized water. Instrumentation Most measurements were performed in a Zeeman 3030 atomic absorption spectrometer equipped with an HGA 600 graphite furnace and an AS 60 autosampler all from Perkin-Elmer.Some measurements were performed in a Model 1100 atomic absorption spectrometer equipped with an HGA 300 graphite furnace and an AS 40 autosampler also from Perkin-Elmer. Argon 99.6% was used as the shielding gas. A B Intensitron hollow cathode lamp (Perkin-Elmer) was used as the primary source. Pyrolytic graphite coated graphite tubes (Perkin-Elmer No. B0109322) and in some measurements a totally pyrolytic graphite platform (Perkin-Elmer No. B0109324) inserted in a coated graphite tube or electrographite tube (Perkin-Elmer No. B0070691) were used. Pyrolytic graphite coated graphite tubes treated with tungsten according to the procedure of Zatka," as modified by Monteiro and Curtius," were also used. Integrated absorbance was measured. All other instru- mental parameters were those recommended by the manufac- turer.The absorbance values shown in this work are the mean of at least three measurements from which the respective blank was subtracted. Table 1 shows the typical temperature pro- gramme followed. The glassware was washed with detergent rinsed with tap water immersed in 10% v/v HN03 and copiously rinsed with de-ionized water before use. Sample pretreatment The samples two reference materials were dried weighed and ashed at 550 "C for 3 h.I3 The ashes were dissolved with 2 ml of 1 moll-' H,S04 at 60 "C for 10 min and made up to 10 ml with de-ionized water. Considering the interferences observed as will be discussed later a separation step was performed by Table 1 Temperature programme for B using Ca (10 pg as CaCl,) as modifier Temperature/ Ramp/ Hold/ Gas flow rate/ Step "C S S ml min-' 1 90 1 10 300 2 120 10 10 300 3 1350 10 10 300 4* 2650 0 5 0 5 2650 1 10 300 6 20 1 5 300 * To whom correspondence must be addressed.* Read in this step.1264 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 percolating this solution through a 3 ml plastic column (diam- eter 0.6 cm) containing a cationic resin (Dowex 50) at a rate of 0.6 ml min-'. After the addition of 3 drops of glycerine (to enhance the stability of H,BO,) the eluate was evaporated to 2 ml and made up again to 10 ml with de-ionized water. Results and Discussion Modifiers Fig. 1 shows pyrolysis and atomization temperature curves for B using different modifiers. The respective absorption pulses under the optimized conditions are shown in Fig. 2.In all cases 50 ng (10 pl) of boron and 10 pg (10 pl) of modifier were used. For comparison the pyrolysis and atomization tempera- ture curves for B alone are also shown. In this case 500 ng of B were used. If no modifier is used [Fig. 1 (a)] B is lost at 800 "C not in its elemental form but as a compound probably B203.5 When Ca is used as modifier [Fig. l(b)] 1400°C can be taken as the indicated pyrolysis temperature but the sensitivity depends on the counter-ion as confirmed in Table 2. However the absorp- tion pulses [Fig. 2(a) (b)] using the different forms of Ca show the same pattern. A long tail was observed not only for Ca but for all the other modifiers tested (Fig. 2). Hence in order to avoid memory effects a blank always had to be fired between 0.600 0.400 0.200 (b) 0 1000 2000 3000 0 1000 2000 3000 v) 0.400 .0 0.300 0 n 0.200 a) + 2 g 0.100 .w - 0.400 Id) 0 1000 2000 3000 400 800 1200 1600 0.300 (el 0.400 - 0.200 0.200 - 0.100 0.100 - 0 400 800 1200 1600 2000 0 I 1 I f ) 1000 2000 3000 TemperaturePC Fig. 1 Pyrolysis ( 0 A ) and atomization (.,A) temperature curves for B without (a) and with different modifiers; (b) Ca 10 pg as CaC1 (A) or as Ca(NO,) (B); (c) Ni lOpg as NiC1 (A) or as Ni(NO,) (B); (d) Mg 10 pg as MgCl,; (e) Y 10 pg as YCI and (f) La 10 pg as La(NO,),. Pyrolysis temperature for the atomization curves 1350 "C except when Ni (900 "C) or no modifier (800 "C) was used. Atomization temperature for the pyrolysis curves 2650 "C each measurement cycle. Although this procedure reduces the tube lifetime it was essential in order to ensure reproducibility.When Ni was tested as a modifier a significantly lower pyrolysis temperature and sensitivity were achieved [Figure l(c) and Table 2). In contrast to Ca the behaviour of the two forms tested (chloride and nitrate) was about the same the only difference being the slightly shorter appearance time for Ni(N03)2. Magnesium and Y (both as chloride) and La (as nitrate) were also tested as modifiers as shown in Fig. l(d) (e) and (f). The respective absorption pulses under the optimized con- ditions are shown in Fig. 2 0 (g) and (h) respectively. Barium (10 pg as chloride or nitrate) and Pd (10 pg as nitrate) were also tested as modifiers but they led to a much lower sensitivity.The optimum pyrolysis temperatures and the characteristic masses found for the different modifiers used are summarized in Table 2. In spite of the good performance of Y and La (as indicated by their thermal stabilization capacity and enhance- ment of sensitivity) they showed the strongest memory effects. Hence the following studies were performed using only Ca or Ni both as chloride. Mass of modifier Fig. 3 shows that masses of Ca between 10 and 40 pg produced about the same sensitivity. However the background absorbance increased with the mass of modifier and 10 pg was confirmed as the mass to be used. The same general trend was observed using Ni as modifier (Fig. 3). Interferences Interferences were studied by adding increasing masses of Mg Ni Fe Na and Mn as chlorides to 50ng of By using Ca (Fig.4) or Ni (Fig. 5) as modifier. The effects of P (as Na2HP04) and of mineral acids are also shown in these figures. Among the interferents tested Na and Mg showed the least pronounced effects for both modifiers. However Mn P Fe and the mineral acids caused an intense decrease in the response for B. Except for the acids [Figs. 4(c) and 5(c)] a general trend is observed in the interference curves after an initial drop in sensitivity the curve moves back up to the original sensitivity value and finally drops again to lower values. As a possible explanation one can suppose that at the left part of the curve the interferent somehow disturbs the thermal stabilization ability of the modifier. After a certain mass the interferent itself behaves as the modifier.Finally the excess of interferent depresses the absorbance signal as gener- ally happens when an excess of modifier is used (Fig. 3). To support this explanation it is observed that when Ca is the interferent using Ni as modifier [Fig. 5(a)] the sensitivity for masses of Ca of around 1Opg is higher than that observed when no Ca is present and as is shown in Table2 the sensitivity using Ca as modifier is actually higher than that reached using Ni. Accordingly when Ca is the modifier and Ni the interferent the sensitivity after the initial drop does not increase back to the original value but to a somewhat lower one. Determination of B in certified reference materials Boron was determined in two reference materials National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1570 Spinach Leaves and NIST SRM 1571 Orchard Leaves.Boron was determined in the final solution of the digestion and separation procedures described under Experimental using 10 pg of Ca (as chloride) as modifier. Both the analytical curve in 0.2 moll-' H2S04 and the stan- dard additions method were used. Only the later led to a good agreement between found and certified values. The results are shown in Table 3. It was not possible to perform standardJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1265 .*... ... ........ .... ...... . .... 1 0 * * *.. L 0 a 0 5.0 Tim e/s Fig. 2 Absorption pulses for B in the presence of different modifiers (a) Ca as CaCl,; (b) Ca as Ca(N03),; (c) Ni as NiC1,; ( d ) Ni as Ni(NO,),; (e) Mg as MgCI,; (f) Y as YCI,; ( g ) La as La(NO,),. Atomization temperature 2650°C; pyrolysis temperature 1350"C except for Ni; boron mass 50 ng; and modifier mass 10 pg.Wall atomization pyrolytic graphite coated graphite tubes Table 2 Optimum pyrolysis temperatures and characteristic masses for B using different modifiers Pyrolysis temperature/ Modifier "C CaCI 1350 Ca(N03 12 1350 MgC12 1400 Y Cl 1350 12 1400 800 NiC1 900 - Characteristic mass*/ Pg 450 530 680 670 1000 610 6100 * Characteristic mass is the mass of analyte giving an integrated absorbance of 0.0044 s. additions directly in the solution resulting from the ashes dissolution. The separation step was insufficient to eliminate all the interferences but it was required to avoid too much reduction of the sensitivity.Detection limits The detection limits were derived from a blank solution (n = 10 k = 3) using analytical curves prepared in de-ionized water and in 0.2% v/v HNO,; the values were 560 and 1990 pg respectively. Atomization surface In order to try to improve the sensitivity of the determination of B by ETAAS different atomization surfaces were tested using Ca or Ni as modifier. The inner surface of a pyrolytic graphite coated graphite tube was coated with a tungsten carbide or with tantalum Boron was also atomized from a totally pyrolytic graphite platform inserted in a pyrolytic graphite coated graphite tube. The results 0.3 ul 0 . $ 0.2 0 a a I I I I I 1 0 20 40 60 80 100 Modifier mass/pg Fig. 3 Dependence of the integrated absorbance of 50 ng of B on the modifier mass modifiers (a) Ca as CaCl,; and (A) Ni as NU2.Pyrolysis temperatures 1350 "C (Ca) or 900 "C (Ni); and atomization temperature 2650 "C. Wall atomization pyrolytic graphite coated graphite tube obtained are displayed in Table 4. No enhancement in sensi- tivity was observed in agreement with the suggestion of Slavin' that B is lost as an oxide (and not as a carbide). Boron (50 ng) could not be detected using tantalum foil even when using argon of higher purity (99.997%) as recommended in the literature." However it must be pointed out that in the present study the tantalum foil edges were not welded but were mechanically fixed. Boron (50 ng) also could not be detected by using an uncoated electrographite tube as shown in Table4; this could be due to the formation of gaseous analyte oxide by reaction with CO or COZ.l6 According to Huettner and Busche,17 oxygen is chemisorbed in the graphite surface by forming surface oxides of graphite at temperatures1266 cc) -0 + L 0 C 4- JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 t 0.L 0.02 2 200 I n te rfe re nt mass/pg 0.5 0.4 0.3 0.2 0.1 0 2 4 6 8 10 [Acidl/mol I-' Fig. 4 Dependence of the integrated absorbance of 50 ng of B on the presence of increasing masses of some salts and mineral acids using Ca ( 10 pg as CaCl,) as modifier (a) Fe as FeC13 (0); Mg as MgC1 (A) and P as Na2HP04 ( x); (b) Ni (O) Na (A) and Mn (x) as chloride; (c) H2S04 (O) HCl (A) and HN03 (x). Pyrolysis temperature 1350 "C; and atomization temperature 2650 "C. Wall atomization pyrolytic graphite coated graphite tubes 0.45 0.40 0.35 0.30 0.25 0.20 0.15 t - 0 0.02 2 200 lnterferent mass/pg 'Table 3 Concentrations of B in two NIST SRMs using Ca (10 pg as CaC1,) as modifier Concentration of B/pg g-' Found Certificate NIST SRM 1570 Spinach Leaves 29+3 (30) 1571 Orchard Leaves 33+3 33+3 'Table 4 Characteristic masses for B atomized from different surfaces using different modifiers Ca and Ni as chloride Tube PGt PG EGS EG PG PG PG-Wg PG-W PG-TaT[ PG-Ta Atomization Wall Wall Wall Wall Platform Platform Wall Wall Wall Wall Modifier/ Characteristic mass*/ 10 Pg Ca 449 Ni 61 1 Ca No signal Ni No signal Ca 580 Ni 730 Ca 1100 Ni 2200 Ca No signal Ni No signal *Characteristic mass is the mass of analyte giving an integrated -f PG pyrolytic graphite coated graphite tube.1 EG electrographite tube. 0 PG-W PG treated with tungsten. 7 PG-Ta PG lined with Ta foil. ,absorbance of 0.0044 s. of 400-500 "C and desorption of oxygen at 900-1000 "C leads to the formation of CO and CO,. Since an uncoated electro- graphite tube has a more active surface the formation of CO and COz should be favoured in this tube in comparison to a coated tube. Conclusions Among the elements tested Ca as CaC1 showed the best performance as modifier for the determination of B by ETAAS. The counterion chloride or nitrate had no influence on the optimum pyrolysis temperature for this modifier but a strong influence on the sensitivity was verified. The sensitivity enhancement observed with all modifiers tested shows that the modifying action for B does not return to a thermal stabiliz- ation effect.The determination of B by ETAAS is very prone to inter- ferences. All substances tested including the mineral acids influenced the response of B. Such interferences demanded a separation step and calibration by the standard additions method in the analysis of real samples. Owing to memory effects a blank had to be measured after each sample or standard. The determination of B by ETAAS is actually less sensitive and more prone to interferences than other methods such as inductively coupled plasma atomic emission spectrometry but if ETAAS is the method available the use of Ca as modifier a separation step and the standard additions technique yield accurate results.I I I I 1 0 2 4 6 8 10 [Acidl/mol I-' Fig. 5 Dependence of the integrated absorbance of 50 ng of B on the presence of increasing masses of some salts and mineral acids using Ni (10 pg as NiC12) as modifier (a) Mg (0) Ca (A) and Fe (x) as chloride; (b) P (O) as NaH2P04 and Na (A) and Mn ( x) as chlorides; (c) H,SO (O) HCl (A) and HNO ( x) Pyrolysis temperature 900 "C; and atomization temperature 2650 "C. Wall atomization pyrolytic graphite coated graphite tubes References 1 Horta A. M. T. and Curtius A. J. Anal. Chim. Acta 1978,96,207. 2 Welz B. Atomic Absorption Spectrometry Second edition VCH Publishers Weinheim 1985 275-276. 3 Castillo J. R. Mir J. M. Bendicho C. and Martinez C. At. Spectrosc. 1985 6 152. 4 Chapman J. F. and Dale L. S. Anal. Chim. Acta 1977 89 363. 5 Slavin W. Graphite Furnace AAS A Source Book. Perkin-Elmer Norwalk USA 1984.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY NOVEMBER 1994 VOL. 9 1267 6 7 8 9 10 11 12 13 Van der Geugten R. P. Fresenius' 2. Anal. Chem. 1981 306 13. Szydlowski F. J. Anal. Chim. Acta 1979 106 121. Goyal N. Dhobale A. R. Patel B. M. and Sastry M. D. Anal. Chim. Acta 1986 182 225. Barnett N. W. Ebdon L. Evans E. H. and Ollivier P. Anal. Proc. 1988 25 233. Schlemmer G. and Welz B. Spe0:rochim. Acta Part B 1986 41 1157. Zatka V. J. Anal. Chem. 1978 50 538. Monteiro M. I. C. and Curtius A. J. J. Anal. At. Spectrom. in the press. Ogner G. Analyst 1980 105 916. 14 L'vov B. V. and Pelieva L. A. Can. J. Spectrosc. 1978 23 1. 15 L'vov B. V. Nikolaev V. G. and Norman E. A. Zh. Anal. Khim. 1988 63 46. 16 Ohlsson K. E. A. Ph.D. Thesis University of Umeb 1993 p. 32. 17 Huettner W. and Busche C. Fresenius' 2. Anal. Chem. 1986 323 674. Paper #/00301 B Received January 18 1994 Accepted June 15 1994

ISSN:0267-9477    

DOI:10.1039/JA9940901263

出版商:RSC    

年代:1994

数据来源: RSC