摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 269 Determination of Rare Earth Elements in Geological Samples by Inductively Coupled Plasma Source Mass Spectrometry Alan R. Date and Dawn Hutchison British Geological Survey, 64 Gray‘s Inn Road, London WCIX 8NG, UK The prototype inductively coupled plasma source mass spectrometer (ICP-MS) operated by the British Geological Survey is described. Preliminary work on its application to the determination of the rare earth elements in geological materials is discussed. Keywords: Inductively coupled plasma source mass spectrometry; rare earth elements; geological materials The determination of the rare earth elements (REE) in geological materials, traditionally of great value in pet- rogenetic studies , is now finding increasing importance in geochemical prospecting.1 The value of REE concentrations in geochemistry is a function of the close chemical similarity of the group, and the gradual change in ionic radius for their cations in octahedral co-ordination. Concentrations of REE are often reported relative to levels in chondritic meteorites, which may be taken as reasonable base-line concentrations for any method developed for their determination. A plot of such chondrite-normalised concentrations against atomic number should produce a smooth graph. The exception is europium, which may exist in the divalent and also the more common trivalent state. Useful information for the geochemist includes the slope of the graph, any changes in slope and the magnitude (and direction) of the europium “anomaly.” Information on the complete REE group may not be essential to define the graph, but is certainly useful.The ideal method for the determination of REE would combine the advantages of high specificity, high sensitivity, excellent detection capability, high precision and accuracy, complete element cover and low cost. Unfortunately, until recently few analytical methods were capable of providing data for the complete REE group with the sensitivity required for geochemical analysis. The situation was alleviated to a certain extent by the isotope dilution mass spectrometry technique2J and neutron activa- tion analy~is,~*5 although both methods are expensive and time consuming and neither provides cover for the complete REE group. The advent of inductively coupled plasma source emission spectrometry (ICP-AES) provided only a partial solution to the problem, but a solution readily adopted by the geochemical c0mmunity.6~7 The more recent development of inductively coupled plasma source mass spectrometry (ICP- MS)s13 may complete the picture, but its application to the determination of REE is still incompletely documented.The determination of REE by ICP-AES is a useful starting point for a consideration of the potential of ICP-MS in this field. The former is limited in two respects. Firstly, the complex emission spectra characteristic of both REE and most geological matrices necessitate some form of separation prior to analysis and correction for mutual interference between REE lines after separation .6,7 Secondly, the detection limits for some REE may be inadequate, even when the separation procedure includes pre-concentration.Detection limits in solution and in the solid (for a l-g sample diluted to a final volume of 5 ml) reported by Crock and Lichte7 are compared in Table 1 with chondrite normalising values.14 Only three elements (La, Eu and Yb) have detection limits at least an order of magnitude better than chondrite values, four (Sm, Gd, Dy and Tm) have detection limits very close to chondrite values and two (Pr and Tb) have detection limits inferior by almost an order of magnitude. Early work on the development and application of ICP-MS showed that REE spectra were simple,l3 suggested that the detection limits were superior to those in ICP-AES15 and demonstrated that the complete REE group could be deter- mined in silicate rocks without pre-concentration or separa- tion from the matrix and using only simple aqueous reference standard solutions.l~*7 Much of that work was carried out on the second UK prototype ICP-MS system operated by the British Geological Survey (BGS) in London, but no details of its performance were given.The instrument is described in this Table 1. REE detection limits by ICP-AES and ICP-MS with chondrite values ICP-AES detection limit7 In solution/ Element ng ml-1 Lanthanum , . . . . . 4.3 Cerium . . . . . . . . 31 Praseodymium . . . . . . 64 Neodymium . . . . . . 22 Samarium . . . . . . . , 30 Europium . . . . . . . . 0.9 Gadolinium . . . . . . 18 Terbium . . . . . . . . 58 Dysprosium .. . . . . 25 Holmium . . . . . . . . 3.6 Erbium . . . . . . . . 8.0 Thulium . . . . . . . . 5.1 Ytterbium . . . , . . . . 2.2 Lutetium . . . . . . . . 1.0 In solid CLg g-’ 0.022 0.155 0.32 0.11 0.15 0.0045 0.090 0.29 0.125 0.018 0.040 0.026 0.01 1 0.005 In solution/ ng ml-1 0.16 0.15 0.16 0.95 1.47 0.48 0.63 0.084 0.34 0.070 0.16 0.048 0.19 0.028 In solid/ 0.032 0.030 0.032 0.19 0.29 0.096 0.13 0.017 0.068 0.014 0.032 0.0096 0.038 0.0056 Pg g-’ ICP-MS detection limit Chondrite valueV CLg g-l 0.33 0.88 0.11 0.60 0.181 0.069 0.249 0.047 0.325 0.070 0.200 0.030 0.200 0.034270 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 work and its initial application to the determination of REE in geological materials is reported. Experimental ICP-MS Instrument The instrument used is the immediate successor to the original UK prototype ICP-MS system described by Gray and Date.13 The system is illustrated schematically in Fig.1 (taken from reference 12) and a list of its main components is given in Table 2. The quadrupole mass spectrometer, including vacuum stages 2 and 3, its control system, the ion detector and its power supply, were supplied by VG Isotopes. The complete system was illustrated in the first brochure describ- ing VG Isotopes' commercial instrument, the PlasmaQuad.18 Sample Dissolution Problems were encountered with running solutions at too high a level of total dissolved solids, and sample fusion techniques were avoided. A two-stage acid dissolution procedure was adopted. After drying overnight at 105 "C, samples (0.1-0.5 g) were weighed into PTFE decomposition vessels, 5 ml of HN03, 5 ml of HC104 and 10 ml of HF (40%) were added and the contents were evaporated to fumes of perchloric acid.The sides of the dissolution vessels were washed down with de-ionised water and the evaporation was repeated to incipient dryness. The residues were taken up in 0.5 ml of HC104 with 10 ml of de-ionised water and the solutions were filtered. The residues and filter-papers were ignited in silica crucibles, 2 ml of HC104 and 2 ml of HF (40%) were added and the contents were returned to the dissolution vessels. The dissolution vessels were sealed and heated at 150 "C overnight. The vessels were cooled and opened, and the evaporation to fumes of perchloric acid was repeated three times.The final residues were taken up in 0.5 ml of HC104, 10 ml of de-ionised water and a few drops of 100-volume Hz02. The solutions were combined with those obtained during the first dissolution stage and made up to 100 ml with de-ionised water to give a final acid concentration of 1%. A reagent blank solution was processed in the same way. Analytical grade reagents were used throughout. Mass analyser control bias sumlies Multiplier Quadrupole I'T-Gl Ll-- acuum stages Ratemeter Multi-channel analyser and display pqr-pzl plotter recorder Table 2. ICP-MS system Inductively coupled plasma . . ICP 2500 with APCS-1 automatic Nebuliser . . . . . . Spraychamber . . . . Vacuumsystem . . . . Mass spectrometer . . Pulse counting chain . . Datasystem . . . . power supply and AMN-2500E automatic matching network.Frequency, crystal controlled, 27.12 MHz. Output power up to 2500 W. Torch, T 1.0. Plasmatherm, Kresson, NJ, USA. Load coil, 3-turn, 0.125-in copper tubing, water-cooled, reverse geometry, laboratory made Fixed geometry, cross-flow, Type 09.790. Jarrell-Ash, Waltham, MA, USA . . . . Sc-1 (Scott type). Plasmatherm . . Stage 1, laboratory made. Stages 2 and 3, VG Isotopes, Winsford, Cheshire. Vacuum pumps: Stage 1: ED 250,250 1 min- 1 . Stage 2: Diffstak 250/2000M, 20001~-~; EDM 12, 244 1 min - 1. Stage 3: Diffstak 100/300M, 3001s-1; EDM6, 108 1 min-1. Edwards High Vacuum, Crawley, Sussex 12-12F, mass range 0-800 a.m.u., resolution >2.5 M, fitted with a Galileo channel electron multi- plier ion detector, Type 4870.VG Isotopes . . Pulse amplifier/discriminator, Type 9302; ratemeter, Type 449. EGG Ortec, Oak Ridge, TN, USA. Scalerltimer, Type 1772. Canberra Instruments, Meriden, CT, USA. NIM to ITL converter. Oakfield Instruments, Eynsham, Oxfordshire plotter, and cassette tape store. Canberra Instruments . . Quadrupole mass filter, Type . . Series 80 MCS, with TTY X - Y I LI- I control Drain --- Fig. 1. Schematic diagram of prototype ICP-MS systemJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 271 Preparation of Reference Standard Solutions Reference standard solutions were prepared from individual REE oxides (Johnson Matthey, Specpure) and from a commercially available REE oxide mix (Johnson Matthey, Spectromel 3). In each instance dissolution was carried out using 50% V/V HN03 plus H202.Analytical grade reagents were used. Stock solutions were prepared at 1000 pg ml-1 in 1% V/V HN03. Instrument calibration was performed with reference standard solutions at 1 pg ml-1 in 1y0 V/V HN03. Instrument Operation Under the terms of the BGS contract with the Commission of the European Communities for the development of ICP-MS, during the period 1983-85 the original prototype system operated by Gray at the University of Surrey was used for further instrument development, while the BGS instrument was operated for a more detailed applications study with its performance restricted to that described by Gray and Date.13 The work reported here was carried out under the conditions shown in Table 3. Sample solutions were introduced to the ICP-MS instru- ment by conventional pneumatic nebulisation, with a solution Table 3.ICP-MS operating conditions Inductively coupled plasma: All-argon plasma Forward power . . . . . . . . . . Reflected power . . . . . . . . Coolant (outer) . . . . . . . . . . Auxiliary(intermediate) . . . . . . Carrier (inner) . . . . . . . . . . Nebuliser pressure . . . . . . . . Solution up-take rate (free aspiration) , . Distance from load coil to sampling aperture . . . . . . . . . . . . Distance from end of torch to sampling aperture . . . . . . . . . . . . Sampling aperture (copper) diameter . . Skimmer aperture (nickel) diameter . . Aperture to skimmer separation . . . . Optimisation . . . . . . . . . . Data acquisition . . . . . . . . . . 1.25 kW <5 w 12.0 1 min-1 0.2 1 min-1 1.21 min-1 28 lb in-2 2.0 ml min-1 10 mm 5 mm 0.5 mm 1.Omm 7 mm Maximum signal, 1 1sIn + Pulse counting, multi- channel scaling (MCS), 1024 MCS channels, 1 ms dwell time, 60 sweeps uptake rate of 2.0 ml min-1.The mass spectrometer is currently limited to two modes of operation, single ion monitoring and mass scanning. After optimisation for maxi- mum signal on llsIn+ using single ion monitoring, the system was operated in the mass scanning mode, covering the range rnlz 135-181. Data acquisition was carried out using pulse counting and multi-channel scaling (MCS). The Canberra Series 80 MCS was set with a data acquisition memory group of 1024 channels, a dwell time per channel of 1 ms and 60 separate sweeps. Under these conditions a complete spectrum was accumulated in just over 1 min.Each sample and standard solution was aspirated for 0.5 min for system equilibration before data acquisition. Results and Discussion General Spectral Characteristics for REE by ICP-MS The spectrum obtained in 1 min for a reference standard solution containing fourteen REE, each at 1 pg ml-1 in 1% V/VHN03, is illustrated in Fig. 2. Monoisotopic REE (Pr, Tb, Ho, Tm) are free from isobaric overlap, whereas others have one (La, Ce, Lu), two (Sm, Eu, Gd, Dy, Er) or three (Nd, Yb) isotopes free from such interference. The most abundant isotope free of isobaric overlap is indicated for each REE, and count rates are shown for 140Ce+ and 175Luf. Under the operating conditions used in this work, principally the use of small (0.5 mm) apertures and with optimisation on the maximum signal for 115In+ (an element with a low first ionisation energy, 5.785 eV, and a high second ionisation energy, 18.87 eV), elements with second ionisation energies lower than the first ionisation energy of the plasma support gas, argon (15.76 eV), will show significant double ionisation. Second ionisation energies for the REE show a general increase with increasing atomic number, and range from 10.55 eV (Pr) to 13.9 eV (Lu).Significant double ionisation is therefore possible, with REE spectra for equi-concentration reference standard solutions showing the pattern illustrated here and in previous work. *3,15-17 With no requirement to determine light elements on which the doubly ionised species of heavy elements will interfere, there is one advantage to this approach.For the chondrite concentration values included in Table 1, there is a ratio between cerium and lutetium of about 26: 1. For one sample included in this work, the ratio approaches 8000 : 1 ! However, whereas the abundance of cerium is much greater than that for lutetium in geological materials, the reverse is true for the signal levels of their most abundant isotopes. In the example illustrated, the signal (313970 counts s-l) l75LU+ 159Tb + 140Ce+(63930 counts s - l ) 4 6 6 7 7 8 7 9 B A - 138 m/z 178 Fig. 2. solution, 1% V/V HN03 each taken in 1 min and covering the range mlz 138-178 ICP-MS spectra for a reference standard solution containing: A, each REE at 1.0 pg ml-1 in 1% V/V HN03 and B, a reagent blank272 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 obtained for the only lutetium isotope free of isobaric overlap (175Luf, 97.4% abundant) is almost 314000 counts s-1, while the only free isotope for cerium (14We+, 88.5% abundant) attains only 64000 counts s-1. In such a 1024-channel spectrum, the integrated count for the channel at the top of each peak (monitored for a total of 60 X 1 ms) may be converted to counts s-1, and corresponds to the rate achieved under single ion monitoring conditions. For data processing, however, the total integral count over the peak is normally used. In this example, each peak occupies approximately 20 channels, representing a total time of 1.2 s. The integral count obtained for each element (using all the available isotopes free of isobaric overlap) is plotted in Fig.3 against the chondritic abundance of each element. This illustrates one of the major advantages of ICP-MS, that the geologically less abundant REE may produce the highest count rates. The second important factor concerning the determination of REE by ICP-MS is also illustrated in Fig. 2. The spectrum obtained for a reagent blank solution (1% V/V HN03) is plotted below the reference standard solution spectrum. The vertical scale is off-set to avoid superimposing the two spectra, and the vertical full-scale deflection is expanded ~ 2 5 6 . The integral counts obtained in the same 20-channel (1.2 s) regions of interest (ROI) are indicated. The mean background count for the 14 ROI is 6.5, equivalent to 5.4 counts s-1, with a standard deviation of 1.9. Similar variation is found for each peak in a series of 14 blank spectra, and it may be possible to define the background using only one blank spectrum.Detection limits in solution (30, blank) are shown in Table 1, converted into detection limits in the solid (assuming 0.5 g in 100 ml, a dilution factor of 200), where they may be compared with the ICP-AES data of Crock and Lichte7 and with chondrite concentrations. 14 With such a dilution factor, only samarium and europium have detection limits above the chondrite values, with ratios to chondrite of only 1.7 and 1.4, respectively. Detection limits for the other REE are satisfac- tory, with lanthanum and cerium better by at least an order of magnitude. REE Calibration Graphs Calibration graphs for eight of the 14 REE, using the most abundant isotope free from isobaric overlap in each instance and covering the range from 0.001 to 10 pg ml-1 for each 10' c c 3 0 - h o) 105 +-I c Y m LL .- 1 04 0.01 0.1 1 .o Chondrite abundance/pg g-1 Fig.3. Relationship between peak integral counts obtained for each REE (taking all the isotopes free from isobaric interference) and their abundances in chondritic meteorites element, are illustrated in Fig. 4. The calibration graphs were prepared from peak integral counts, corrected for background using the average background count from the 14 regions of interest for a reagent blank solution. Bearing in mind the detection limit data presented above (Table l ) , there is strong evidence that the calibration graphs are linear over at least four, and possibly five or six, orders of magnitude.With the calibration graph for lanthanum linear up to a concentration in solution of at least 10 pg ml--l, and that for lutetium linear down to 0.001 pg ml-1 or less, it is possible to consider the simultaneous determination of all REE in one solution, even when the ratio between the most abundant and the least abundant is of the order of 10000: 1. Analysis in this work is carried out using only a two-point calibration, with one reference standard solution containing each REE at 1 pg ml-1 and one reagent blank (normally 1% V/V HN03), and taking only the most abundant isobaric overlap-free isotope in each instance. REE Spectra in Geological Materials The ICP-MS REE detection limits quoted for a sample dilution factor of 200 (0.5 g in 100 ml), compare favourably with ICP-AES detection limits for a dilution factor of only 5 (1 g in 5 ml), and would obviously be far superior to ICP-AES if a similar pre-concentration technique was employed.The advantage of ICP-MS, however, is that such detection limits may be achieved without separation of the REE from the rock matrix. It is fair to state that much of the interest in REE in geological materials is related to silicate rocks, and for this reason they formed the subject of the initial study. The general features of silicate rock ICP-MS spectra are shown in Fig. 5 , the spectrum obtained for the Canadian standard reference syenite SY-3, selected for its relatively high concentrations of REE, and taken at a relatively high solution concentration (nominally 1 g in 100 ml).With the exception of SiOz, lost during the dissolution procedure with HF, all major elements are present in solution. The ICP-MS spectrum, 106 1 175Lu + 169Tm i 165H0+ 166Er+ 105 - 139La+ 153Eu+ 157Gd+ 147Sm- 100 I 1 0.001 0.01 0.1 1 .o 10.0 Element concentration/pg ml-1 Fig. 4. Calibration graphs for eight of the 14 REE, using the most abundant isotope free from isobaric overlap in each instance, covering the range from 0.001 to 10.0 pg ml-1JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 273 however, is very clean, and the relationship between the more abundant light REE and the less abundant heavy REE is obvious. The maximum channel count for the 140Ce+ peak is 63560, equivalent to a rate in excess of 106 counts s-1.The cerium concentration in solution is about 20 pg ml-1. The peaks are more than adequately resolved for analytical purposes, although the minimum channel counts between 139La+, 140Ce+ and 141Pr+ are 29 and 9, equivalent to rates of 483 and 150 counts s-1, respectively. These count rates do not represent the true background, but reflect the relative sizes of 139La+ and 141Pr+ and the fact that abundance sensitivity for quadrupole mass spectrometers is larger on the high-mass side of a peak, i . e . , resolution appears to be better on the high-mass side. The excellent resolution achieved in this example is illustrated in Fig. 6, where the vertical full-scale 135 m/z 181 Fig. 5. ICP-MS spectrum for the Canadian standard reference syenite, SY-3, at 1% in solution (1 gin 100 ml), covering the range mlz 135-181 deflection is changed from 65 X lo3 to 4 X lo3 for the region from 149Sm+ to 181Ta+.The minimum channels counts between the smaller REE peaks and between those for hafnium (the group of small peaks between Lu and Ta) are 0 , l or 2, representing a true background count very close to that achieved for a reagent blank solution. In practice it is found that the general level of non-spectral (random) background is similar for 1% HN03 and 1% HC104. In this example the count rates are shown for 140Ce+ and 1MEr+, and all the REE peaks free from isobaric overlap are identified. Isotope dilution analysis, offering high precision and accuracy, would be possible for Nd, Sm, Eu, Gd, Dy, Er and Yb.Polyatomic Ion Interferences The REE spectra illustrated by Gray and Date13 show small “oxide” peaks for the heavy REE, well separated from the main REE spectra. The fact that such polyatomic ion interferences may be a problem in ICP-MS is now widely appreciated, and has received some attention.19-22 The particular problem of barium and light REE “oxide” inter- ferences on heavy REE has been discussed by Doherty and Vander Voet23 and more recently by Longerich et al. ,24 with reference to the application of a commercial ICP-MS system. Under the operating conditions used in this work, “oxide” levels are fairly low. The relationship between percentage “oxide” and monoxide bond strength, illustrated in Fig. 7, shows a general positive correlation.The most serious potential interference in this context, where only the most abundant isobaric overlap-free isotope is taken in each instance, is 141Pr16O+ on 157Gd+. No correction is made in this work, although it is appreciated that a small degree of interference correction is inherent in the use of a single multi-element reference standard solution for system calibra- tion. La 1 140Ce+ (1 060000 counts s-1) I 166Er+ (47 300 counts s-l) n Srn II Hc 1 FSD 4 x 103 FSD 65 x lo3 I 139 m/ z 181 Fig. 6. ICP-MS spectrum for SY-3, with the vertical full-scale deflection changed from 65 X lo3 to 4 X lo3 for the section from 149Sm+ to 181Ta+. Count rates for 140Ce+ and 166Er+ are calculated from the integral count in the maximum channel in each instance.All isobaric interference-free REE isotopes are identified274 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 Precision and Accuracy Precision and accuracy may be assessed by considering two international standard reference silicate rocks, the Canadian syenite SY-3 and the US Geological Survey (USGS) diabase W-1. Although the SY-3 spectrum shown in Figs. 5 and 6 was obtained for a 1% solution (1 g in 100 ml), this sample contains relatively high levels of REE, and analysis would normally be n c q 0.1 @Ho Sm@ Eu Nd @Lu Trn 0.05 120 140 160 180 200 Oxide bond strength/ kcal rnol-l carried out using a solution at only 0.1%. Data presented in Table 4 for SY-3 at such a dilution factor are compared with previously published values.7.25 In this instance, calibration was performed with a reference standard solution run at each end of the analysis sequence.The precision obtained therefore reflects system stability over a period of more than 10 min. The precision is adequate for these relatively high levels of REE (cerium and lutetium are present in solution at about 2 and 0.008 pg ml-1, respectively) and there is good agreement with the data presented by Crock and Lichte.7 In his 1984 compilation of “usable” values, Abbey25 attached more weight to data obtained by neutron activation analysis, and the agreement here is only adequate. In order to achieve detection capability at chondrite concentration levels, a dilution factor of only 200 (0.5 g in 100 ml) is necessary, Data are therefore presented in Table 5 for REE in the USGS standard reference diabase W-1 at such a dilution.At this level of total dissolved solids, a certain amount of salt condensation occurs on the small (0.5 mm) sampling apertures used in this work. Consequently, the ion signal drifts downwards with time. The data in Table 5 were obtained by running sample and standard solutions alter- nately, and calculating REE concentrations for each sample run using the mean of the bracketing pairs of standards. The inferior precision in this example is a reflection of both the calibration scheme and the lower concentration levels for W-1. The results of applying such a calibration scheme to a series of international standard reference silicate rocks are presented in Table 6. Only a single determination was made in Fig.7. Relationship between “oxide” level and monoxide bond strength for 11 of the 14 REE. “Oxide” ions were not detected for Dy, Er and Yb each instance. There is- good agreement with previously published “usable” and no obvious evidence of “oxide” interference . Table 4. ICP-MS precision and accuracy for REE in SY-3 at 0.1% ICP-MS (this work) Abundance, Range/ Mean/ RSD, Ref. 7/ Ref. 251 Element Ion % Pi! g-l g-l YO CLg g-l Pg g-l Lanthanum . . . . 139La+ 99.9 1199-1310 1250 2.8 1350 1350 Cerium . . . . . . 140Ce+ 88.5 2123-2292 2213 2.6 2290 2200 Praseodymium . . 141pr+ 100 208-227 218 3.1 239 120? Neridvmiiim 146Nd + 17.2 689-746 72 1 2.6 760 800? J -------- I ..,-I Samarium Europium Gadolinium Terbium Dysprosium Holmium Erbium . . T h n > l 4 n q - .. . . . . . . 147Sm+ . . . . 153Eu+ 157Gd+ 15m+ . . . . 163Dy+ . . . . 165H0+ . . . . . . . . 166Er+ 169Tm + . . . . - .- 15.0 52.2 15.7 24.9 33.4 100 100 1 nn --. . 120-137 17.4-20.4 18.4-21.3 123-144 82.0-88.5 i n 0 - 1 3 A 115-132 28.1-34.0 ~~ 129 125 136 18.9 19.8 31.0 85 .O 1 1 R ~~ 4.3 5.2 4.4 5.3 4.2 7.0 2.7 A c ; ~~ 134 121 138 18.3 15.0 29.0 88.0 1 7 < _._ loo? 14? 55? 1 l ? 80? 20? 50? Q7 L11U11U111 . . . . 1111 L V” I”., L k . 7 LA.” 7.V 1b.J U. Ytterbium . . . . 172Yb + 21.9 63 Sk70.0 67.0 3.7 71 .O 65 Lutetium . . . . 175Lu+ 97.4 7.68.9 8.4 4.8 8.6 8? Table 5. ICP-MS precision and accuracy for REE in W-1 at 0.5%. For abundances, see Table 4 ICP-MS (this work) Element Lanthanum . . . . Cerium . . . . . . Praseodymium . . . . Neodymium .. . . Samarium.. . . . . Europium.. . . . . Gadolinium . . . . Terbium . . . . . . Dysprosium . . . . HoImium . . . . . . Erbium . . . . . . Thulium . . . . . . Ytterbium . . . . Rangelpg g-1 Meadpg g-1 RSD, YO Ref. 25/pg g-1 8.40-12.8 10.3 11 9.8? 18.9-24.9 23.5 7.3 23? 2.29-3.18 2.79 8.4 3.4? 9.96-14.4 12.3 13 15 2.20-4.40 3.60 23 3.6? 0.80-1.78 1.26 25 1.1 2.97-4.94 3.55 16 4 0.48-0.98 0.68 26 0.65 3.40-5.29 4.32 12 4 0.66-1.01 0.83 13 0.7? 1.74-2.84 2.27 16 2.4 0.22-0.41 0.32 16 0.3 1.58-2.56 1.98 15 2.1 Lutetium . . . . . . I DLUf u. L4-u. 3Y u.34 14 U.35:rJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 275 REE Mineral Analysis Finally, the limitation on the level of total dissolved solids in solution is irrelevant in the case of REE-rich samples.The BGS standard reference monazite IGS 36 may be considered. Fig. 8 shows the spectrum obtained for IGS 36 at a dilution factor of 10000 (0.1 g in 1 1). Even at this dilution factor, the count rate for 140Cef is in excess of 3 X l o 6 counts s-1. The vertical full-scale deflection is changed from 262 X 103 to 8 X 103 for the section from 151Eu+ to 175Lu+. The ratio between the most abundant and the least abundant REE is clearly very large. In view of the counting losses expected at such a high count rate for cerium, the REE concentration data obtained at such a dilution factor are of interest. Data for REE (expressed as oxides, the normal procedure for REE minerals) in IGS 36 are compared in Table 7 with a previously published compila- tion.27 The duplicate ICP-MS data represent the values obtained using a copper sampling aperture at 0.5 mm and a nickel sampling aperture at 0.7 mm.There is good agreement with previously published values for a concentration range from 23% Ce02 to 59 pg g-1 Tm203. Although the agreement for lutetium, at 29 pg g-1 by ICP-MS, is poor, the four previously published results were obtained by only one laboratory. Counting losses probably account for the slightly low results for cerium and lanthanum, and there is some evidence of a slightly high value for gadolinium, although the value obtained by ICP-MS, uncorrected for any “oxide” interference from praseodymium, is still within the range of previously published values. It is interesting that no single technique was used to obtain values for all REE in the previously published data.ICP-MS is clearly a very powerful method of analysis in the field of REE geochemistry. Conclusion Although ICP-MS clearly has great potential for the determi- nation of REE in geological materials without separation from the matrix, the technique is clearly limited, at least in this particular application, by the inability of the ICP-MS system to tolerate high levels of total dissolved solids. The prototype 1- 14OCe+ (3.28 x lo6 I , I 135 mlz 176 Fig. 8. ICP-MS spectrum for the BGS standard reference monazite IGS 36, at 0.01% in solution (0.1 g in 1 l), covering the range m/z 135 to 176. The rate for “We+ (3.28 x 106 counts s-1) was calculated from the maximum channel count Table 6. ICP-MS data for REE in standard reference silicate rocks (pg g-I).For abundances, see Table 4 SY -2 NIM-G G-2 GSP- 1 BCR-1 Element ICP-MS Ref. 25 ICP-MS Ref. 25 ICP-MS Ref. 25 ICP-MS Ref. 25 ICP-MS Ref. 25 Lanthanum . . . . . . 87 92 Cerium . . . . . . . . 160 160 Praseodymium . . . . . . 16 19? Neodymium . . . . . . 52 58? Samarium . . . . . . . . 5.9 7.2 Europium . . . . . . . . 1.2 1.4 Gadolinium . . . . . . 4.2 5? Terbium . . . . . . . . 0.4 O S ? Dysprosium . . . . . . 2.1 2.3 Holmium . . . . . . . . 0.4 0.4?* Erbium . . . . . . . . 1.5 1.3?* Thulium . , . . . . . . 0.1 0.3?* Ytterbium . . . . . . . . 0.5 0.86 Lutetium . . . . . . . . 0.1 - * Data taken from reference 26. 160 370 46 170 20 12 1.7 1 .o 4.2 0.7 1.8 0.2 1 .o 0.2 195 360 50? * 190? 25? 2.4? 15? * 1.4? 5.7? 3? * - 1.9 0.2? 27 57 31 7.6 7.8 2.0 7.4 0.6 6.7 1.3 4.2 0.5 3.6 0.4 27 63 53 140 7? 17 26? 63 6.5 14 2.0 1.8 6.6?* 16 1 .o 2.2 7? 14 1.2? 3.8 3.5? 12 0.6? 2.0 3.4 16 0.5? 2.5 88 210 - 71? 15? 2.4? 2? 20? 12? 2? 17 3? - - 110 200 21 69 12 15 17 11 11 0.3 2.3 3.6 1.8 1.7 105? 200 - 68? 16? 0.4 11?* 3? 15* 3* lo?* 2? * 14 2? NIM-L ICP-MS Ref.25 240 200? 300 230? 21 - 52 45? 4.6 6? 1.0 l ? 4.0 - 0.6 0.7? 2.9 3?* 0.5 - 2.1 - 0.4 - 2.4 4? 0.5 - Table 7. ICP-MS data for REE oxides in IGS 36 standard reference monazite Reference 27 ICP-MS, Oxide Ion Y O Range, YO n Median, Yo La203 Ce02 Pr,O 11 Sm203 E u ~ O ~ * Tb407 Ho203* Er2O3* Tm203 * Lu203* Nd203 G d 2 0 3 Dy2°3 Yb203* . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . 10.8,10.6 23.0,22.8 2.85,2.79 10.2,9.86 1.68,1.58 331 , 322 1.13,1.09 0.10,0.09 0.35,0.34 418,393 700,690 58,60 250,210 29,29 10.4-13.5 20.6-26.9 2.48-3 .OO 9.80-12.0 1.30-1.88 270-740 0.70-1.24 0.21-0.45 370-430 600-1000 50-80 250-840 0.01-1.05 52-68 24 30 16 20 16 18 16 15 20 6 7 6 13 4 11.9 24.9 10.5 2.80 1.59 0.77 0.11 0.31 350 390 610 65 298 58 * Data in pg g-l.276 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 ICP-MS system used in this work has three limitations: a conventional reversed geometry load coil was used, sampling apertures were normally limited to 0.5 mm and data acquisi- tion was carried out using only fast scanning over one selected mass range. The use of apertures larger than 0.5 mm was alluded to in consideration of the data presented for REE in IGS 36 (Table 7).Applications development work at the BGS is now adopting those improvements in system operation used by Gray at the University of Surrey,28 including the use of larger (1 .O mm) apertures (a change requiring no additional pumping capacity), a load coil ground strap to the torch box and a single pass water-cooled spray chamber. The BGS system will also be improved with the introduction of the VG PlasmaQuad quadrupole control and data processing system. This will speed up data processing and allow the use of peak hopping routines, invaluable for studying the relationship between polyatomic ion interferences and doubly ionised species under the same operating conditions. A detailed investigation of performance with respect to the determination of REE in geological materials for the improved system has been initiated. This work was carried out with the support of the Directorate General for Science and Technology, Commission of the European Communities (Contract No.MSM. 104.UK[H]). The paper is published with the approval of the Director, British Geological Survey (NERC). References Strong, D. F., Can. J . Earth Sci., 1984, 21, 775. Hooker, P. J., O’Nions, R. K., and Pankhurst, R. J., Chem. Geol., 1975, 16, 189. Hanson, G. N., in LaFleur, P. D., Editor, “Accuracy in Trace Analysis: Sampling, Sample Handling, Analysis,” Volume 11, National Bureau of Standards Special Publication No. 422, National Bureau of Standards, Washington, DC, 1976, p. 937. Gordon, G. E., Randle, K., Coles, G. G., Corlis, J. B., Beeson, M. H., and Oxley, S. S . , Geochim. Cosmochim. Acta, 1968,32, 369. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. Haskin, L. A., Wildeman, T. R., and Haskin, M. A., J. Radioanal. Chem., 1968, 1, 337. Walsh, J. N., Buckley, F., and Barker, J., Chem. Geol., 1981, 33, 141. Crock, J. G., and Lichte, F. E., Anal. Chem., 1982, 54, 1329. Houk, R. S., Fassel, V. A., Flesch, G. D., Svec, H. J., Gray, A. L., and Taylor, C. E . , Anal. Chem., 1980, 52, 2283. Date, A. R., and Gray, A. L., Analyst, 1981, 106, 1255. Date, A. R., and Gray, A. L., Spectrochim. Acta, Part B , 1983, 38, 29. Douglas, D. J., Quan, E. S. K., and Smith, R. G., Spectro- chim. Acta, Part B, 1983,38, 39. Date, A. R., and Gray, A. L., Analyst, 1983, 108, 159. Gray, A. L., and Date, A. R., Analyst, 1983, 108, 1033. Hanson, G. N., Annu. Rev. Earth Planet. Sci., 1980, 8, 371. Date, A. R., and Gray, A. L., Spectrochim. Acta, Part B , 1985, 40, 115. Date, A. R., and Gray, .A. L., in “Proceedings of the 31st ASMS Annual Conference on Mass Spectrometry and Allied Topics,” 1983, p. 834. Date, A. R., ZCPZnf. Newsl., 1984, 10, 202. “PlasmaQuad,” Brochure No. 02.511 PJG, VG Isotopes, Winsford, Cheshire, 1983. Gray, A. L., Spectrochim. Acta, Part B, 1986,41, 151. McLeod, C. W., Date, A. R., and Cheung, Y. Y., Spectro- chim. Acta, Part B, 1986, 41, 169. Tan, S. H., and Horlick, G., Appl. Spectrosc., 1986, 40, 445. Date, A. R., Cheung, Y. Y., and Stuart, M. E., Spectrochirn. Acta, Part B, 1987, 42, 3. Doherty, W., and Vander Voet, A,, Can. J . Spectrosc., 1985, 30, 135. Longerich, H. P., Fryer, B. J., Strong, D. F., and Kantipuly, C. J., Spectrochim. Acta, Part B , in the press. Abbey, S . , Geol. Surv. Can. Pap., 1983, 83-15. Abbey, S . , Geol. Surv. Can. Pap., 1980, 80-14. Lister, B., Geostand. Newsl., 1981, 5 , 75. Gray, A. L., personal communication. Paper J6l109 Received November 6th, 1986 Accepted December 19th, I986

ISSN:0267-9477    

DOI:10.1039/JA9870200269

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 277 Determination of Trace Metals in Marine Sediments by Inductively Coupled Plasma Mass Spectrometry James W. McLaren, Diane Beauchemin and Shier S. Berman Division of Chemistry, National Research Council of Canada, Ottawa KIA OR9, Canada The accuracy of analysis of geological materials by inductively coupled plasma mass spectrometry was assessed by the determination of ten trace elements (V, Mn, Co, Ni, Cu, Zn, As, Mo, Cd and Pb) in solutions of the marine sediment reference material BCSS-1. Isobaric interferences by molecular species were minimised by a modification of the dissolution procedure and by a careful choice of plasma operating conditions. Results obtained by external calibration (i.e., the use of reference solutions) showed some evidence of ionisation suppression.Good accuracy was achieved by the method of standard additions for all elements except copper, for which there is an isobaric interference by titanium oxide. Keywords: Inductively coupled plasma mass spectrometry; trace metal determination; marine sediment analysis Current detection limits in dilute nitric acid for inductively coupled plasma mass spectrometry (ICP-MS) are in the range from 0.01 to 0.1 pg 1-1 for most elements.1-3 Whether such remarkable detection power can be achieved in more complex solutions is the subject of much of the current fundamental and applied research in this field. Two problems areas have been identified: isobaric interferences from molecular species arising either from the acids used in sample dissolution4.5 or from the sample itself6.7 and an increasingly severe suppres- sion of ion sensitivity as the total dissolved solids concentra- tion of the sample solutions increases.8 Despite these difficul- ties, a number of impressive demonstrations of the potential of ICP-MS for trace analysis of solutions of geological materials have been reported.sl2 Date and Gray have reported the determination of the rare earth elements in silicate rocks9 and of 17 trace metals in a suite of geological reference materials with a silicate matrix.10 Doherty and Vander Voet,ll as well as Longerich et a1.,12 have also described the determination of the rare earth elements in geological materials.In all of these publications, the assessment of the accuracy of the analyses has been limited somewhat by the fact that, in many instances, the results could be compared only with rather imprecise “recommended” values, or in some instances, values based on a very small number of determinations. The major purpose of this work was to obtain a more rigorous assessment of the accuracy of ICP-MS analyses of geological materials by applying the technique to solutions of a very well characterised reference material, the nearshore marine sediment BCSS-1 .I 3 Reliable values are available for 13 trace elements in BCSS-1, including all of the first row transition metals, for which few ICP-MS analyses of geological materials have been reported. A second purpose was to determine the extent to which ICP-MS might be used to determine additional trace elements for which reliable values have not yet been established.The results presented in this report were obtained by two calibration strategies: external calibration (i.e., the use of reference solutions to establish the calibration) and standard additions. Results obtained by stable isotope dilution techniques are the subject of a separate report. 14 Experimental Instrumentation The inductively coupled plasma mass spectrometer used for this work was the ELAN 250 from SCIEX Division of MDS Health Group Ltd. (Thornhill, Ontario, Canada). The work described here was completed before a modification of the ion optics to the present configuration to improve stability and to reduce suppression of ion sensitivity by concomitant elements.This modification involved the replacement of a set of a.c. rods between the skimmer and the Bessel box lenses with a three-cylinder einzel lens and of a screen positioned imme- diately behind the skimmer, and held at -10 to -50 V, (termed the “ring” electrode) with a 5 mm diameter grounded stop. As described in reference 5, a mass flow controller was used on the aerosol carrier gas line and the solution uptake was controlled by a peristaltic pump. The extended torch provided with the instrument was replaced with a conventional ICP- AES torch and the torch box was positioned as close as possible to the sampler. This resulted in a sampling depth 17 mm from the top of the r.f. load coil. Other details of the hardware and operating conditions are listed in Table 1.Under these conditions, the sampler was positioned approxi- mately 10 mm from the tip of the initial radiation zone15 observed while aspirating a 1000 mg 1-1 yttrium solution. Sample Dissolution The sample dissolution procedure involved only a slight modification of a method developed for the analysis of marine Table 1. ICP-MS hardware and operating conditions Inductively coupled plasma- Plasma Ar . . . . . . . . . . 14 1 min-1 Auxiliary Ar . . . . . . . . . . 2.0 1 min - 1 Nebuliser Ar . . . . . . . . . . 0.85 1 min-1 R.f. power . . . . . . . . . . 1.2 kW Sample flow-rate . . . . . . . . 1.1 ml min-1 Standard ICP-AES torch Meinhard C concentric glass nebuliser Scott-type spray chamber Sampler . . . . . . . . . . . . Nickel, 1,14-mm orifice Skimmer . . . . . . .. . . . . Nickel, 0.89-mm orifice Distance from sampler to load coil . . 17 mm Ring lens . . . . . . . . . . . . -28.2 Photon stop (Bessel box B) . . . . +6.5 Bessel box A lens . . . . . . . . -8.0 Bessel box C lens . . . . . . . . -8.0 Entrance a.c. rod offset . . . . . . -4.0 Mass spectrometer- Ion lens voltages- Operating pressures- Interface region . . . . . . . . 0.9 torr Massspectrometerchamber . . . . 3.5 X 10-5torr NRCC Publication No.: 27337.278 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 sediments by ICP-AES.13 A 0.5-g sample was wet ashed with a mixture of nitric, perchloric and hydrofluoric acids in a sealed PTFE pressure decomposition vessel. The mixture was then transferred into a PTFE beaker for evaporation to dryness. For previous ICP-AES analyses, the residue was re-dissolved in 1 M hydrochloric acid, but for ICP-MS, it was re-dissolved in 50 ml of 0.1 M nitric acid.This procedure works well for most elements of interest, but the dissolution of chromium and titanium in BCSS-1 is known to be incomplete. A further ten-fold dilution of the solutions with 0.1 M nitric acid was performed prior to analysis. Mass Spectra Acquisition In order to choose the most appropriate isotopes for elemental analysis, spectra of the BCSS-1 solutions were examined. All spectra were acquired in the sequential scanning mode; intensity measurements of 0.5 s duration were made at 0.05-u intervals. Analysis with External Calibration Calibration for all the elements of interest was accomplished by running a blank and three reference solutions spanning the appropriate concentration ranges.The reference solutions were prepared by appropriate dilutions of a multi-element stock solution which contained only the elements listed in Table 4. The data acquisition parameters were of course identical for the calibration and subsequent analyses. The resolution was set manually to an intermediate value to assure adequate resolution of the 55Mn peak from the much larger 56Fe peak. Each determination involved three repetitions of a sequential scan in a “peak-hopping” mode through the selected list of isotopes. Three measurements, each of 1-s duration were made at each peak of interest, one at the assumed peak centre and at k0.05 u. Each determination required ca. 2 min. With only two exceptions, the most abundant isotope was used for each element.The mNi isotope was chosen in preference to 58Ni to avoid the 58Fe isobaric interference, and 68Zn was chosen over @Zn and 66Zn in order to avoid the necessity of making a correction for T i 0 interference. The results presented in Table 4 are based on analyses of four solutions of BCSS-1. Prior to running the sediment solutions, intensity data were acquired for the 0.1 M HN03 blank so that these could be subtracted from subsequent readings. After the four solutions had been run, one of the calibrating solutions was re-run to check for drift. It was clear from this check that sensitivity had changed by 10-15% for several of the analytes. No attempt has been made to correct the results of Table 4 for this drift.It was clear, though, that frequent recalibration would be required if a larger number of solutions were to be analysed. Standard Additions Analyses For each standard additions analysis, three solutions were run: an unspiked BCSS-1 solution and two spiked solutions. The amounts of the analytes in the first spike were roughly equal to the amounts present in the unspiked solution; the second spike contained twice these amounts. The spikes were added to 5-ml aliquots of the 1% BCSS-1 solutions, before dilution to 50 ml with 0.1 M nitric acid. Exactly the same data acquisition parameters were used for the standard additions analyses as were used for the analyses with external calibration. Results and Discussion Interpretation of the ICP Mass Spectra The spectra of the BCSS-1 solutions yielded a great deal of information about spectroscopic interferences in ICP-MS .In addition to the expected isobaric interferences, interferences arising from the argon plasma itself, the acids used in sample dissolution and from oxides and hydroxides of elements present in the sample were observed. This led to a modifica- tion of the sample dissolution procedure to minimise inter- ferences from solvent molecular ions and to the establishment of compromise plasma operating conditions to minimise interferences from oxide and hydroxide species. Preliminary investigations on sediment solutions diluted to volume with 1 M HC1 indicated a number of serious isobaric interferences due to chlorine containing molecular species. A comprehensive listing of these species has been published by Tan and Horlick.4 The most serious problems in this work were the overlap of 35Cl16O on 51V and 40Ar3sC1 on 75As.Fortunately it was possible to eliminate both of these difficulties by modifying the dissolution procedure so that the final dilution to volume is made with 0.1 M HN03 rather than 1 M HC1. The effectiveness of this modification is illustrated in Figs. 1 and 2. Fig. 1 shows the spectra in the mass range 50-57 for a 1 M HC1 solution, a 0.1 M HN03 solution and a standard containing 100 pg 1-1 of Cr and V. Fig. 2 shows the spectra in the mass range from 50 to 60 of a BCSS-1 solution, the 0.1 M HN03 blank and a standard containing 50 pg 1-1 of V, Cr, Co and Ni. The only significant peak in the 0.1 M HN03 spectrum is the 40Ar16O peak at mlz 56.Peaks for V, Cr, Co and Ni in the BCSS-1 solution (at mlz 51, 52, 59 and 60, respectively) are free of interference. Also seen in this spectrum are off-scale peaks for 5OTi, 55Mn and the four stable isotopes of Fe at mlz 54,56,57 and 58. Isobaric interferences can also arise from an overlap of an M1160 peak on an M2 peak, where M1 and M2 are two elements whose masses differ by 16. In some instances, the corresponding hydroxide species must also be considered. This problem has been discussed by Vaughan and Horlick6 80000 r I 60000 A .. . 50 52 54 56 m/z Fig. 1. species: A, 0.1 M HN03; B, 1 M HCl; and C, 100 pg 1-1 Cr and V ICP mass spectra illustrating interferences by C10 molecular m/z ICP mass spectra in the mass range 5&60: A, 0.1 M HN03; B, Fig.2. BCSS-1 solution; and C, 100 pg 1-1 V, Cr, Co and NiJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 279 6000 who stressed that the relative proportions of singly and doubly charged ions, oxide and hydroxide species for a particular element are strongly dependent upon the plasma operating conditions, in particular the power and the nebuliser gas flow-rate. Among the more problematic interferences in the present work were the overlaps of T i 0 species on both stable isotopes of Cu at mlz 63 and 65, and on the two most abundant Zn isotopes, 64Zn and 66Zn. The behaviour of Ti and Ti0 species was studied as a function of nebuliser gas flow-rate, with other operating conditions as indicated in Table 1. A 1 mg 1-1 Ti solution in 0.1 M HN03 was used to measure the intensities for the five Ti isotopes and their corresponding oxides at flow-rates ranging from 0.7 to 1.1 1 min-1.The results are illustrated in Fig. 3 and presented in greater detail for the most abundant Ti isotope in Table 2. At 0.7 1 min-1 nebuliser gas flow-rate, Ti0 peaks are not distinguishable above the background, but sensitivity for Ti is more than 100 times lower than that observed at 0.9 1 min-1. At the other extreme, at a flow-rate of 1.1 1 min-1, the Ti0 peaks are larger than the Ti peaks. At flow-rates between these extremes, a reasonable compromise between high sensitivity and low oxides can be made. The minimum value that can be attained for the MO+/M+ intensity ratio appears to be related to the strength of the metal-oxygen bond, as indicated by the data of Table 3.These ratios can be achieved by an appropriate choice of plasma operating conditions, but further reduction does not appear to be possible with the present hardware. Douglas and French17 suggest that levels of metal oxides in the normal analytical zone15 of the ICP ought to be much lower than these ratios would indicate, and therefore that some oxide formation is occurring in the sampling process. Olivares and Houkl8 also concluded that the MO+/M+ ratio is dependent on the ion sampling characteristics of the instrument in use. Interferences by Ti0 species on Cu and Zn at a nebuliser gas flow-rate of 0.9 1 min-1 are illustrated in Figs. 4 and 5. The concentrations of the three elements are similar to those in the BCSS-1 solutions analysed.It is clear from Fig. 4 that interferences by 47Ti16O on 63Cu and 49Ti16O on 65Cu are still significant. Interferences by 48Ti16O on 64Zn and 5OTi16O on MZn are also significant, but fortunately, an interference free isotope, 68Zn, can be used to circumvent this problem. For Cu, no such remedy exists, and there appears to be no alternative to making a software correction in a manner analogous to the way corrections for line overlap interferences are made in ICP-AES. ,'-\ - I ' I I Table 2. Effect of nebuliser gas flow-rate on Ti+ and TiO+ intensities c I v) C ln 4000 3 0 0 2000 Peak intensity/counts s-1 rate/l min-1 48Ti + 48Ti160f intensity ratio 0.7 1 250 < 10 co.01 0.8 40 OOO 360 0.009 0.9 320 000 4 000 0.013 1.0 320 OOO 40 000 0.125 1.1 12 600 80 000 6.3 Nebuliser flow- T i 0 + /Ti + - - u Table 3.Minimum observed MO+/M+ ratios for various metals M-0 bond strength*/ Metal kcal mol - Ca . . . . . . . . 92 f 2 Mo . . . . . . . . 145 f 8 Ti . . . . . . . . 159 f 2 Zr . . . . . . . . 182 k 2 Ce . . . . . . . . 1 W f 3 * Data from reference 16. Minimum observed MO+/M+ ratio 0.001 0.002 0.008 0.016 0.025 Although the oxide ratios are higher for the more refractory oxides (such as those of Ti, Mo and Zr) interferences can also arise from oxides of major constituents of the marine sediments, such as Fe and Ca. The spectra of Fig. 6 illustrate iron oxide overlaps on several Ge isotopes. The peaks at m/z 72,73 and 74 for the BCSS-1 solution are primarily due to the oxides of 56Fe, 57Fe and 58Fe, rather than Ge, the concentra- tion of which in BCSS-1 is ca.1.5 Fg g-1. The 70Ge isotope is overlapped by 54Fe16O (as well as 70Zn), leaving only 76Ge, which is overlapped by the 40Ar36Ar dimer. The determina- tion of Ge in iron-rich geological materials is therefore likely to be very difficult, and was not attempted in this work. Similar difficulties were encountered by Date and Hutchi- son.10 In contrast, the determination of monoisotopic As at mlz 75 appears to be straightforward for BCSS-1. The fact that As can be readily determined in 0.1% solutions of BCSS-1 (i.e., at a concentration of ca. 10 pg 1-l) whereas its determination in 1% solutions was impossible by I 1 r I I . . , 48 52 1 56 m/z . . 3 . . . . . . . ... ... . . I I 60 64 68 Fig.3. ICP mass spectra illustrating the effect of nebuliser gas flow-rate on the relative abundance of titanium oxide s ecies for a 1 mg 1-l Ti solution at flow-rates of: A, 0.7; B, 0.9; and 8, 1.1 1 min-1 v 2000 1 " 61 62 63 64 65 66 m/z Fig. 4. M HN03; B, 5 mg 1-l Ti; and C, 50 pg I-' Cu ICP mass spectra illustrating T i 0 interferences on Cu: A, 0.1 6000 1 i: I 1 U 60 62 64 66 68 70 m/z Fig. 5. M HN03; B, 5 mg 1-1 Ti; and C, 250 pg 1-l Zn ICP mass spectra illustrating Ti0 interferences on Zn: A, 0.1280 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 ICP-AES13 is a good example of the superior detection power of ICP-MS. Another example can be seen in the spectra of Fig. 7, where the spectrum of the BCSS-1 solution shows clearly peaks for four isotopes of Mo (at rnlz 95,97,98 and 100) while the other isotopes (at rnlz 92,94 and 96) are overlapped by Zr isobars at much higher concentrations. Analysis of BCSS-1 Solutions of BCSS-1 were analysed by means of external calibration and by the method of standard additions.Results obtained by the two methods are compared with the accepted values in Table 4. While acceptable accuracy is achieved for As, Cd, Pb and Zn with external calibration, results for Cu and Mo are high and values for V, Mn, Co and Ni are low. The high value for Cu arises from the 47Tl16O interference on 63Cu, for which no correction has been attempted. The low results for V, Mn, Co and Ni may be due to a suppression of ion sensitivity for these elements by the much higher concentrations of Al, Fe, Ca and Mg present in the sample solutions but not in the reference solutions used for calibra- tion.This explanation is supported by the standard additions results for these elements which are in good agreement with the accepted values. In fact good accuracy was achieved by standard additions for all elements except Cu, the value for which remains high, as expected, because of the Ti0 interference. While the low results for V, Mn, Co and Ni by external calibration can possibly be attributed to a suppression of ion sensitivity by the concomitant elements present in the samples but not in the reference solutions, some other explanation must be found for the high result for Mo. The possibility of a spectroscopic interference is ruled out by the fact that an accurate value was achieved by standard additions.Sensitivity and Detection Limits The sensitivities and detection limits achieved at the operating conditions used for these analyses are shown in Table 5. The detection limits are based on the usual definition, i.e., that concentration of the analyte yielding a signal equivalent to three times the standard deviation of the blank signal. The values do not represent the best possible that can be achieved, but rather what is possible under analytical conditions in which oxide interferences must be addressed. It should also be noted that the ion lens modification mentioned at the beginning of the Experimental section, which was primarily intended to improve the instrument's stability, has also resulted in a five-fold improvement in sensitivity.Effects of Concomitant Elements Early experience with ICP-MS has indicated a greater susceptibility to suppression of analyte ion sensitivity by concomitant elements than would be expected either from theoretical predictions or previous ICP-AES experience.19 It also appears that concomitant elements can have an adverse effect on calibration stability. Recent work with more concentrated marine sediment solutions14 indicates that one source of drift is gradual deposition of solid material on the sampler and skimmer. It is also clear, though, that more subtle effects can occur further downstream in the ion optics. The calibration stability of our instrument was later much improved by an ion lens modification which involved no changes in the interface region.While the problem of calibration drift has not been entirely eliminated by the new ion optics, effective compensation for it is now possible by the use of one or more internal standard elements. In addition, the susceptibility of the instrument to suppression of analyte Table 4. Analysis of the marine sediment reference material BCSS-1. Results are in pg g-1; precision expressed as the standard deviation (n = 4) Found External Standard Accepted Element calibration additions value V . . . . . . 7 1 f 3 93 f 16 93 k 5 Mn 220 f 19 229 k 15 . . . . . . 156f8 Co . . . . . . 8.920.2 1 3 2 3 11 f 2 Ni . . . . . . 4 3 f l 57 f 6 55 k 4 Cu . . . . . . 2 4 f 1 29 f 3 1 9 f 3 Zn . . . . . . 1 2 4 f 8 123 f 5 119 f 12 As . . . . . . 14+1 1 2 f 1 11 f 1 M o .. . . . . 3.0k0.1 1.8 f 0.2 (1.9)* Cd . . . . . . 0.26t0.02 0.27k0.03 0.25f0.04 Pb . . . . . . 2 2 k 1 23 k 2 23 k 3 PPhY. * Result obtained by isotope dilution spark source mass spectro- Table 5. Sensitivity and detection limits Sensitivity1 counts s-l Isotope per pg I-' 51v 55Mn 59Co 60Ni 63Cu 68Zn 75As 98Mo 114Cd 208Pb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 109 106 23 49 7.3 9.2 31 33 21 Detection limit (3a)/ 1.18 I-' 0.08 0.04 0.01 0.15 0.07 2.0 0.5 0.1 0.1 0.1 1200t I \ I c 800 - v) 3 0 u + 400 - I I I I I I I I I I I I I I I I I I I I I m/z Fig. 6. ICP mass spectra in the mass range 70-80: A, 0.1 M HN03; B, BCSS-1 solution; and C, 10 pg 1-l As 500 400 7 300 v) 3 w 6 200 100 n "89 91 93 95 97 99 101 m/z Fig.7. ICP mass spectra in the mass range 9&100: A, 0.1 M HNO,; B, BCSS-1 solution; and C, 5 pg 1-l MoJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 281 ion sensitivity by concomitants has been considerably reduced. Both of these performance improvements enhance the accuracy of results obtained by external calibration. Conclusion The detection power of ICP-MS and the relative simplicity of the spectra provide enormous potential for the trace analysis of geological materials. The extent to which this potential will be realised depends largely on the ease with which accurate and stable calibration can be achieved. This work showed that it was possible to obtain accurate results rather easily for nine trace elements (V, Mn, Co, Ni, Zn, As, Mo, Cd and Pb) in the marine sediment reference material BCSS-1 by the method of standard additions.Though some problems of isobaric inter- ferences by molecular species were encountered, most were resolved by a modification of the sample dissolution pro- cedure and a careful choice of plasma operating conditions. While low results obtained for some of the lighter elements by external calibration may be rationalised by an ionisation suppression mechanism such as that suggested by Olivares and Houk,19 it appears that concomitant elements can affect ion sensitivity and calibration stability by other mechanisms which are not clear at present. When a better understanding of these processes is achieved, it will be possible to extend the use of external calibration to a wider variety of applications.References 1. 2. Douglas, D. J., and Houk, R. S . , Prog. Anal. At. Spectrosc., 1985, 8, 1. Houk, R. S . , Anal. Chem., 1986, 58, 97A. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. Gray, A. L., Spectrochim. Acta, Part B, 1986, 41, 151. Tan, S. H., and Horlick, G., Appl. Spectrosc., 1986, 40, 445. McLaren, J. W., Mykytiuk, A. P., Willie, S. N., and Berman, S. S . , Anal. Chem., 1985, 57, 2907. Vaughan, M. A., and Horlick, G., Appl. Spectrosc., 1986,40, 434. McLeod, C. W., Date, A. R., and Cheung, Y. Y., Spectro- chim. Acta, Part B , 1986, 41, 169. Olivares, J. A., and Houk, R. S., Anal. Chem., 1986, 58, 20. Date, A. R . , and Gray, A. L., Spectrochim. Acta, Part B, 1985, 40, 115. Date, A. R., and Hutchison, D., Spectrochim. Acta, Part B, 1986, 41, 175. Doherty, W., and Vander Voet, A., Can. J. Spectrosc., 1985, 30, 135. Longerich, H. P., Fryer, B. J., Strong, D. F., and Kantipuly, C. J., Spectrochim. Acta, Part B , submitted for publication. McLaren, J. W., Berman, S. S . , Boyko, V. J., and Russell, D. S . , Anal. Chem., 1981, 53, 1802. McLaren, J. W., Beauchemin, D., and Berman, S. S . , Anal. Chem., 1987,59,610. Koirtyohann, S. R., Jones, J. S . , Jester, C. P., and Yates, D. A., Spectrochim. Acta, Part B , 1981, 36, 49. Weast, R. C., Editor, “CRC Handbook of Chemistry and Physics,” 65th Edition, CRC Press Inc., Cleveland, OH, 1984, Douglas, D. J., and French, J. B., Spectrochim. Acta, Part B, 1985, 41, 197. Olivares, J. A., and Houk, R. S . , Anal. Chem., 1985,57,2674. Olivares, J. A., and Houk, R. S . , Anal. Chem., 1986, 58, 20. pp. F171-176. Paper JA6/17 Received September 23rd, I986 Accepted October 30th, I986

ISSN:0267-9477    

DOI:10.1039/JA9870200277

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 283 Plasma Potential Measurements for Inductively Coupled Plasma Mass Spectrometry with a Centre-tapped Load Coil R. S. Houk, Jonathan K. Schoer and Jeffrey S. Crain Ames Laboratory - US Department of Energy and Department of Chemistry, Iowa State University, Ames, IA 5001 I , USA Afloating Langmuir probe is used to measure d.c. potentials in an inductively coupled plasma (ICP) while the latter is being sampled for mass spectrometry (MS). With the centre-tapped load coil configuration the plasma potentials are found to be low (+0.5 to -3.5 V) and only moderately dependent on plasma operating parameters. The plasma potential measurements are in basic agreement with previous measurements of ion kinetic energies and mass spectral characteristics for this type of ICP-MS device.Both the numerical values of the plasma potentials and their variation with plasma parameters differ considerably from previous measurements with other load coil configurations that yield a secondary discharge from the plasma to the sampling orifice. Thus, the fundamental processes occurring during ion extraction differ somewhat for the two basic types of ICP-MS device. Keywords: Inductively coupled plasma mass spectrometry; ion sampling processes Inductively coupled plasma mass spectrometry (ICP-MS) has attracted widespread interest in the scientific community because of its analytical figures of merit, such as the excellent powers of detection and the ability to measure isotope ratios.'-3 The ion extraction process is obviously of crucial importance to the analytical performance of an ICP-MS device, and considerable effort has been expended in the fundamental study and improvement of the interface used to sample ions from the plasma.4-11 A key facet of the extraction process is the extent to which the plasma and sampling interface interact electrically, e.g., does an electrical discharge occur between the two? This interaction affects the kinetic energies of the extracted ions and hence the resolution and peak shapes obtainable with the mass analyser. The discharge also influences the abundance of doubly charged ions, metal oxide ions and other species in the observed mass spectra.&*" One way to study the interaction between plasma and sampling interface is to insert a floating Langmuir probe into the ICP to measure the potential of the latter.Naturally, the plasma potential oscillates with the r.f. voltage applied to the load coil; measurements of this r.f. potential have been reported by Douglas and French.10 The interaction of this r.f. potential with the sampling cone creates a d.c. potential in the plasma, and d.c. potential measurements have been described for several load coil geometries that yield a significant discharge effect.9 These load coil geometries are analogous to those used in approximately half the present ICP-MS devices, i.e., the laboratory-made devices and those manufactured by VG Isotopes. The other ICP-MS instruments manufactured by Sciex employ a different load coil configuration (referred to as the centre-tapped geometry, Fig.1) that almost completely eliminates the discharge between the plasma and the sampling interface.l"J1 In this work, measurements of the d.c. plasma potential generated by the centre-tapped load coil are reported and compared with previous measurements of mass spectral characteristics12313 and ion kinetic energies11 for such an ICP-MS device. The behaviour of the two types of ICP-MS devices differs in several respects, and some of these differ- ences are rationalised from the plasma potential measure- ments. Experimental The ICP-MS device1.3,"'.*2 and the Langmuir probeg have been described previously. Particular instrumental com- ponents and standard operating conditions are identified in Table 1, and a diagram showing the spatial positions of the plasma, probe path and sampling orifice is shown in Fig.1. The probe sliced through the centre of the plasma 4 mm directly in front of the orifice. The path of the probe was kept constant relative to the sampling orifice as shown in Fig. 1, except where noted otherwise. The probe path is slightly (ca. 2 mm) closer to the orifice than was the case in reference 9. The observed potentials were measured relative to ground, the sampling orifice being grounded. The aerosol gas flow- rate, forward power and position of the plasma relative to the probe and orifice were varied during the course of the experiments. The values chosen for these parameters corre- sponded approximately to those that were found to be useful for analyses with this particular device.17.18 In the following discussion, the measured potentials are compared with those found previously using other load coil configurations but with considerably lower aerosol gas flow-rates and closer sampling positions.9 These differences reflect the fact that operating conditions for optimum analytical performance differ some- what for the two general types of ICP-MS devices.6,9,11-13 I I 1 I 30 20 10 0 mm Fig.1. Spatial orientation of ICP, probe and ion extraction interface. A-E denote plasma regions traversed successively by the probe284 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 w Forward powerlkW 1.0 Table 1. Instrumental facilities -3.0 $ -4.0 .- 4- 1 .o 1.2 - - 1.4 -3.0 -4.0 1.6 - - 0 ME+ (-0.2) -1.0 -2.0 !-r *B (+0.4) IP Operating conditions, materials or dimensions f low-rate/ I min 1 0 Component Nebuliser power supply- Model UNPS-1, Plasma-Therm, (now RF Plasma Products) .. -4.0 -3.0 I Transducer frequency Forward power 45 W Reflected power <5 W 1.36 MHz (-0.9) w -1.: 2.0 pl?rJ vq 0.5 Nebuliser assembly, spray chamber, desolvation apparatus- Ames Laboratory con~truction~~315 . . . . . . Transducer and desolvation chamber ice water cooled Desolvation temperature 80 “C Solution uptake rate 2.5 ml min- 1 (-1.2) -P + ICP-MS device- ELAN Model 250 with ion optics upgrade (1986), Sciex . . . . Sampling orifice 1.1 mm diameter in Ni cone ICP frequency 27.12 MHz Reflected power S15 W 1 (-1.4) I (+0.3) I (-0.7) ICP torch- Ames Laboratory construction16 -1.0 oPlw- v- ‘ 1.6 Outer tube extended to 40 mm Argon flow-rates: from inner tubes Outer gas 12 I min-1 Auxiliary gas 0.8 I min- -2.0 -3.0 f -4.0 Langmuir probe- Ames Laboratory construction’ Thoriated W rod in quartz Exposed section 2.0 mm long Probe output connected to sleeve x 1 .O mm diameter data acquisition device by coaxial cable ca.1 m long Probe position Fig. 2. Plasma potential profiles obtained during introduction of aqueous aerosol with the sampling orifice on centre and the probe 27 mm from the load coil. Measured values of V, in volts are listed in parentheses. A-E correspond to the ICP regions identified in Fig. 1. Values for forward power and aerosol carrier gas flow-rate are indicated. The base line of each trace is at 0 V and the vertical scale is the same in these and subsequent figures Forward powerikW Data acquisition- Signal averager, Model 1170, Nicolet Instrument .. . . Probe output measured by analogue to digital converter Input impedance 1 MQ Input sensitivity k 2Vfull scale Probe release triggered single sweep of signal averager Potential profiles plotted from memory of signal averager on to X - Y recorder Sweep rate 15 ps per channel over 4096 channels Aerosol flow-rate/ lmin 1 (-1.1) gas w 0 1.4 (-1.0) w -2.0 -3.0 -4.0 (-1.7) ilJ 0.5 A continuous flow ultrasonic nebuliser was used for sample introduction.14J5 The sample was de-ionised, distilled water. Although the aerosol was desolvated prior to its introduction into the ICP, solvent removal was naturally not complete and a substantial amount of water still reached the plasma.For studies with a “dry” ICP, i.e., one without nebulised water, the nebuliser was bypassed and Ar was added directly to the inner tube of the torch so that the water impurity in the Ar supply constituted the only moisture reaching the plasma. (-2.1) I 1.0 (-1.4) (-2.0) 1 Results and Discussion Potential Profiles During Nebulisation of Water Two sets of spatially resolved potential profiles obtained during introduction of nebulised water are shown in Figs. 2 and 3. The various plasma regions sensed by the probe are labelled A-E in Fig. 1; the corresponding features in the potential profiles are also labelled A-E in the upper left profile in Fig. 2. The probe sensed little potential signal until it -3.0 -4.0 Probe position Fig. 3. aqueous aerosol at a probe position 32 mm from the load coil Plasma potential profiles obtained during introduction ofJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 285 -1.0 -2.0 -3.0 -4.0 (-2.2) > c (-2.3) -3.0 - .- -4.0 r - 4- c a 2 0 -3.0 -2.0 :JJ - - -1.0 -4.0 - -3.0 -4.0 Forward powerikW 1.2 1.4 1.6 m 1. (-2.0) I (-1.9) (-2.1) I (-2.1) (-2.3) I (-2.2) Aerosol gas flow-rate/ I min-1 0 0.5 1 .o 1.6 Probe position Fig. 4. aerosol. Probe position was 27 mm from the load coil Plasma potential profiles obtained during introduction of dry reached the outer luminous edge of the plasma (A, Figs. 1 and 2). Upon entering the plasma the measured potential became negative, went through a minimum referred to below as the leading valley (B) and then through a maximum (C).The potential at this central maximum was designated V,. The potential then went through a second valley (D) and rose to approximately zero after the probe left the ICP (E). For a particular set of plasma parameters, the basic shape of the potential profiles was very reproducible for successive swings of the probe. The short-term reproducibility of the V, values was k 0.2 to k 0.3 V at 1.0 kW and an aerosol gas flow-rate of 0 1 min-1. The reproducibility of V, improved to -t 0.1 V at higher power and aerosol gas flow-rate. The potential profiles from the centre-tapped coil were more reproducible and less noisy than those from the other coils investigated previously.9 Potentials were also measured with the probe held stationary for a few seconds at the positions A-E (Fig.1). These potential values were similar to those obtained when the probe swung continuously through the plasma, which indicated that the magnitudes of the measured potentials were not depen- dent on the rate of travel of the probe through the plasma. Potential profiles were also obtained with Ar flowing through the torch and r.f. power applied to the load coil but no plasma present, i.e., the Tesla coil had not been used to initiate a plasma. These profiles showed very little signal (-0.1 to +0.1 V), indicating that the potentials sensed by the probe with the plasma on were related to properties of the plasma and were not merely r.f. pickup from the load coil. Some profiles lacked a well-defined central maximum. In these instances, V, was determined to be the potential at the same position (relative to A) at which central maxima were observed for other profiles.Note that position C was displaced to the right of the geometrical centre of the profile because the probe velocity increased during its travel due to the action of gravity. 9 The most obvious feature of the profiles in Figs. 2 and 3 is that the measured potentials were relatively small (+0.4 to -2.5 V) and either negative or only slightly positive. In contrast, load coils known to cause a significant discharge effect yielded V, values of up to +20 V with sharply defined central maxima.9 It was reasonable that the centre-tapped load coil investigated should have yielded lower V, values because the ion kinetic energies were also lower with the centre-tapped coil. These latter measurements yielded an estimated plasma potential at the sampling orifice of approxi- mately +2 V,11 i.e., a few volts higher than seen by the Langmuir probe.This discrepancy was also expected because the floating probe actually achieved a potential that was somewhat less than the real plasma potential because of cooling in the vicinity of the probe.Y.19 The V, values and ion kinetic energies obtained with the other load coils were strongly dependent on plasma operating parameters.9 As shown in Figs. 2 and 3, the magnitudes of the measured potentials changed only slightly (by <2 V) as plasma parameters were varied with the centre-tapped load coil. This observation was also consistent with previous measurements of ion kinetic energies, which were only slightly dependent on plasma operating conditions for the centre-tapped coil. 11 Some slight but interesting trends in V, and profile shape with changing plasma parameters were still evident in the present work, however.Firstly, compare the V, values and profile shapes obtained at an aerosol gas flow-rate of 0 1 min-l (i.e., no hole punched through the centre of the plasma) to those at 0.5 1 min-1. Injection of the axial channel with Ar containing some water induced a significant decrease in V , but had only a slight effect on the valley potentials. Secondly, as aerosol gas flow-rate increased at constant power, the central maxima generally became less pronounced except at 1 .O k W. Thirdly, V, did not change much as power was increased at constant aerosol gas flow-rate except at the highest flow-rate investigated (1.6 1 min-1).Fourthly, with the axial channel present, V, was more negative when the probe-load coil separation was 32 mm than when it was 27 mm. The trends described above for the centre-tapped load coils were generally in opposite directions to those seen for the other load coil geometries studied previously. Those mass spectral characteristics that depend on plasma potential would also be expected to vary with plasma operating parameters in different manners for the two types of ICP-MS devices. This expectation was fulfilled to some extent by the experimental data. For example, consider the distribution of ionic species generally seen from an analyte element M. With a centre- tapped load coil, the observed ratio M2+/M+ generally decreases and MO+/M+ increases as aerosol gas flow-rate increases.”-13 For other load coil geometries, these ratios generally change in the opposite directions as aerosol gas flow-rate increases.6~7~9 There are similarities in the behaviour of the two types of devices as well, e.g., in both cases a plot of M+ count rate as a function of aerosol gas flow-rate is either hump shaped or monotonically increasing.”11-’3 The different magnitude of the plasma potential and the different depen- dence of plasma potential on operating parameters for the two load coil geometries means that experimental results obtained with one type of ICP-MS instrument may be applicable to the other type but this correlation should not be assumed.Potential Measurements During Introduction of Dry Argon Analogous experiments were performed with dry Ar intro- duced directly into the central channel of the ICP. Typical potential profiles are shown in Fig. 4 for a sampling position 27 mm from the load coil. Profiles with dry Ar at 32 mm are not reproduced here because they had similar shapes and trends as those at 27 mm. In general, the measured potentials were more negative without water present. This difference was noticeable even when the aerosol gas flow-rate was zero or in plasma regions outside the axial channel.9 Only slight changes in V, and286 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 profile shape were seen when the axial channel was injected with dry Ar, i. e. when the aerosol gas flow-rate was changed from 0 to 0.5 1 min-1.Some of the profiles in Fig. 4 were more symmetrical than when water was nebulised. At lower power, the central maxima became slightly less pronounced as aerosol gas flow-rate increased. At higher power there was little change in profile shape as aerosol gas flow-rate changed. The potential valleys were more pronounced and persisted at high aerosol gas flow-rate if water was absent. Again, in the present work V, was generally lower when the probe - load coil separation was 32 mm rather than 27 mm, which is opposite to the behaviour seen with other load coils.9 Additional Observations In some instances (primarily Figs. 2 and 3) the potential profiles were somewhat asymmetrical because the leading valley (B, Fig.2) was at a lower potential than the second valley (D). As shown in Fig. 1, the furthermost left turn of the load coil was connected at its upper end to the coupling box. Thus, the most negative valley in the potential profiles appeared when the probe was in plasma region B, which was adjacent to the load coil section exposed to the applied voltage of largest magnitude. This phenomenon was also observed for the other load coils.9 The potentials described above were all measured with the sampling orifice on the central axis of the ICP. Potential profiles were also obtained with the plasma on and the probe traversing the same path relative to the torch as shown in Fig. 1 but with the sampling orifice removed from the ICP. This was carried out by pulling the plasma assembly as far as possible to the right (Fig.l), by rearranging the interlock system for the sampling interface so that it remained in its “up” position and did not slide down into contact with the ICP, and by moving the probe assembly so that the probe tip still sliced through the plasma. The probe sensed considerably more negative poten- tials (as low as -6 V) with the sampler isolated from the ICP. With the orifice absent the potential profiles also exhibited little or no central maxima. Thus, with the centre-tapped load coil, the measured potentials were influenced by the presence or absence of the sampling orifice, indicating that there was some electrical interaction between the plasma and the sampling interface. The fundamental processes responsible for this interaction have been tentatively described.9>20 For the other, reversed-geometry coils tested previously, the plasma potential changed dramatically (by ca.50 V) when the plasma was moved into contact with the sampling orifice.9 Thus, the extent of the electrical interaction between the plasma and the sampling orifice was much less for the centre-tapped load coil. Conclusion The present work further verifies the validity of the Langmuir probe technique for estimating the potential in an ICP and for studying the interaction between an ICP and a sampling device such as that used for MS. The deduction from previous MS studies that the plasma potential is lower and varies less with plasma operating parameters for the centre-tapped load coil1OJ1 is confirmed in this work.In the future it would be interesting to determine if the plasma potentials with the various load coils depend upon the concentration of concomi- tant elements. This experiment could provide insight into the fundamental reasons for the widely different ionisation interference effects reported by various ICP-MS workers and the generally greater severity of such interferences in ICP-MS relative to ICP emission spectrometry.1321.22 It should also be possible to adapt the probe technique to allow measurement of the potential in the supersonic jet behind the sampling orifice,23 which could yield information about how processes occurring therein are influenced by the potential developed in the plasma. Experiments along these lines are proceeding in our laboratory. This work was supported by the Ames Laboratory - US Department of Energy, Contract No.W-7405-Eng-82 via the Director for Energy Research, Office of Basic Energy Sciences. The authors gratefully acknowledge experience with the Langmuir probe gained in Dr. A. L. Gray’s laboratory. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. References Houk, R. S . , Anal. Chem., 1986, 58, 97A. Gray, A. L., Spectrochim. Acta, Part B , 1985, 40, 1525. Douglas, D. J., and Houk, R. S., Prog. Anal. A t . Spectrosc., 1985, 8, 1. Houk, R. S . , Fassel, V. A., and Svec, H. J., Dyn. Mass Spectrom., 1981, 6, 234. Olivares, J. A., and Houk, R. S . , Appl. Spectrosc., 1985, 39, 1070. Olivares, J. A., and Houk, R. S., Anal. Chem., 1985,57,2674. Gray, A. L., Spectrochim. Acta, Part B , 1986, 41, 151. Gray, A. L., J . Anal. At. Spectrom., 1986, 1 , 247. Gray, A. L., Houk, R. S., and Williams, J. G., J . Anal. At. Spectrom., 1987, 2, 13. Douglas, D. J., and French, J. B., Spectrochim. Acta, Part B , 1986, 41, 197. Fulford, J. E., and Douglas, D. J . , Appl. Spectrosc., 1986,40, 971. Horlick, G., Tan, S. H., Vaughan, M. A., and Rose, C. A., Spectrochim. Acta, Part B , 1985, 40, 1555. Vaughan, M. A., and Horlick, G., Appl. Spectrosc., 1986,40, 434. Olson, K. W., Haas, W. J., Jr., and Fassel, V. A . , Anal. Chem., 1977,49, 632. Bear, B. R., and Fassel, V. A., Spectrochim. Acta, Part B , 1986, 41, 1089. Scott, R. H., Fassel, V. A., Kniseley, R. N., and Nixon, D. E., Anal. Chem., 1974, 46, 75. Palmieri, M. D., Fritz, J. S . , Thompson, J. J., and Houk, R. S., Anal. Chim. Acta, 1986, 184, 187. Thompson, J. J., Ph. D. Dissertation, Iowa State University, 1986. Clements, R. M., and Smy, P. R., J. Phys. D, 1974, 7, 551. Chapman, B., “Glow Discharge Processes,” Wiley, New York, 1980, Chapters 3 and 5. Olivares, J. A., and Houk, R. S., Anal. Chem., 1986, 58,20. Beauchemin, D., McLaren, J. W., and Berman, S. S . , Spectrochim. Acta, Part B, 1987, 42, in the press. Houk, R. S., and Lim, H. B., Anal. Chem., 1986, 58, 3244. Paper JA6l16 Received September 16th, 1986 Accepted November 17th) 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200283

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 287 Determination of Selenium by Graphite Furnace Atomic Absorption Spectrometry Part 1 Interaction Between Selenium and Carbon JiFi D6dina,* Wolfgang Frech,t lngela Lindberg, Erik Lundberg and Anders Cedergren Department of Analytical Chemistry, University of Ume$ S-90187 Ume& Sweden Factors important for the interaction between various types of graphite materials and selenium in the absence of modifier were investigated. By employing radioactive measurements and constant temperature graphite furnace atomic absorption spectrometry it was found that in the presence of hydrogen, selenium compounds could be volatilised during drying from a pyrographite surface. However, stability up to 1773 K could be realised in an argon atmosphere if an activated graphite surface was used.A two-step atomiser with independent control of analyte volatilisation and atomisation made it possible to investigate the reactions between vapourised selenium species and graphite. In uncoated polycrystalline graphite tubes heated to 1173-1573 K 90% of volatilised selenium species were retained as compared with 15% in a pyrolytic graphite coated tube. The equilibrium distribution of condensed as well as gaseous selenium species as a function of temperature and partial pressure of oxygen is given in order to provide support for the experimental results. Keywords: Graphite furnace; graphite surface; atomic absorption spectrometry; selenium stabilisation; distribution of selenium species Much interest has been shown in recent years in the determination of selenium, particularly because of the bio- logical significance of this element.Graphite furnace atomic absorption spectrometry (GFAAS) using Zeeman-effect background correction is now one of the most widely used methods for the determination of selenium in complex matrices, as background correctors with continuum-light sources are not effective in correcting for spectral interfer- ences at the short wavelength used for its determination.1 The main problem with the GFAAS technique arises from the fact that selenium is volatile at low temperatures. To cope with this, a number of stabilising agents such as nickel, copper, platinum, lanthanum and palladium2-5 have been proposed. However, even in the presence of such modifiers losses of selenium have been observed for certain types of samples. For example, in the presence of nickel and sodium chloride no losses were observed up to an ashing temperature of ca.1300 K while the addition of glucose to this matrix resulted in losses at 500 K.6 In an attempt to improve procedures for stabilising sel- enium, this paper reports a study of the role of the graphite surface on the physico-chemical processes that can lead to stabilisation as well as destabilisation of selenium. The first part of the study only deals with the behaviour of selenium in aqueous solutions in order to exclude the effect of the matrix, including various modifiers. Experimental Instrumentation For the atomic absorption measurements, an improved version of the two-step furnace7 was used.The furnace has a T-shaped configuration incorporating a tube with integrated contacts8 and a graphite cup below. (For the determination of activation energies and appearance temperatures, tubes with- out cups were used.) Individual 7.5-kW power supplies, each operated at 15 V, and associated IR-sensitive photodiodes * Present address: Czechoslovak Academy of Sciences, Institute of Nuclear Biology and Radiochemistry, 142 20 Prague 4, Czechoslovakia. t To whom correspondence should be addressed. allowed independent temperature-controlled heating of the tube and cup. Temperature settings referring to the inner surface of the cup and the outer surface of the tube were calibrated with an optical pyrometer (Keller Spezialtechnik Pyro Werk, Model PBO 6AF3) above 1273 K and otherwise with a NiCr - Ni thermocouple.The two-step furnace was installed in a Varian AA-6 spectrometer. The signal damping of the AA-6 read-out module was modified to obtain a fast response time from the electronics.9 In order to study the time dependence of the analytical signal as well as the tempera- tures, a fast on-line data acquisition system was connected to the spectrometer. 10 The sampling interval, i. e., the time resolution between successive measurements, could be varied and was set to 5 ms. After atomisation, corresponding absorbance and temperature values were available from the computer memory with that “time resolution.” After data processing, arbitrary portions of the atomic absorption and temperature signals, as well as log A = f( 7‘) graphs, were displayed on the plotter.Regression was performed on a suitable number of points from the log A = f(T) graph, after which the activation energy was calculated from the slope. The appearance temperature was taken from the point at which the regression line had an absorbance of 0.005. For the radioactivity measurements, a Perkin-Elmer HGA- 74 furnace was used. The y-radiation was measured with a 45 X 50 mm NaI(T1) well-type scintillation detector (Berthold, SZ 44/50-W-N) connected to a high-voltage power supply with integral discriminator (Berthold, LB 2220) and a scalar timer unit (Berthold, BF 2270-1). In these experiments, temperat- ure settings refer to the platform temperature and were calibrated with a NiCr - Ni thermocouple (below 1273 K) and with a PtRh - Pt thermocouple above that temperature.Reagents Selenium(IV) and selenium( Vr) stock aqueous solutions, 1000 pg ml-1. Prepared from selenium dioxide and selenic acid, respectively. Radioactive sodium selenite stock solution. Containing 0.11 mg ml-1 of selenium with an activity of 1.19 mCi ml-I (Radiochemical Centre, Amersham, UK). All gases used were of spectroscopic-reagent grade.288 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 Materials Pyrolytic graphite coated (PGC) tubes (Perkin-Elmer Part No. 081087) with pyrolytic graphite (PG) platforms (Varian Part No. 63-100003-00) were used in the Perkin-Elmer HGA-74. Graphite parts (cups and tubes) for the two-step atomiser were manufactured from RWOl high-density graph- ite (Ringsdorff-Werke GmbH, Bonn-Bad Godesberg, FRG) and subsequently pyrolytically coated by Ringsdorff-Werke. Activated charcoal (Merck, 30-50 mesh ASTM, Part No.3633659) and graphite powder (Ringsdorff-Werke, RW-A, Part No. V1/216/637) were used. Determinations with the Two-step Atomiser Sample aliquots (5 pl) were pipetted manually into the graphite cup through the injection port of the tube. After ashing, as described in Table 1, the tube was heated to a pre-selected atomisation temperature. Cup heating was in- itiated after a time delay selected such that the vapour phase temperature had become constant. Radioactive Measurements Standard 75Se solutions (20 pl) were manually pipetted on to the platform. As the platform had to be removed frequently from the tube, after each ashing, it was only loosely placed in the central part of the tube.After drying and ashing, as described in Table 2, the platform was removed and the amount of 75Se remaining was measured. Care was taken in positioning the graphite parts reproducibly in the detector in order to ensure a similar geometry for all counts. Relative losses of 75Se were compared with the amount of 75Se left after the drying step in argon. Between each injection the platform was cleaned by heating three times to 2650 K for 15 s. High-temperature Equilibrium Calculations The calculations were performed as described earlier.11 The species considered in all calculations were: gaseous Ar, C, CH, CH27 CH37 CH47 co7 co2, C2H, C2H2, C2H4, C2H6, C3H47 C3H6, H7 H2, HCO, H2°, O, 02,O3, OH, Se7 Se2, Se37 SeO, Se02, CSe, CSe2 and H2Se; and condensed C, Se, Se02 and CSe2.Thermodynamic data were taken from references 12-14. Except where stated otherwise, the following input amounts (pmol) were used for the calculations: hydrogen 0.056, argon 4 and analyte 10-5; carbon and oxygen were varied to obtain the partial pressures given in the figures. Results and Discussion Thermodynamic Considerations Fig. 1 shows the equilibrium distribution of various selenium species as a function of temperature and partial pressure of oxygen. The input amount of hydrogen was chosen to correspond to the amount of water left in the tube after a typical drying step. As can be seen, selenium is stable as A A " " C 500 700 900 1100 Tern peratu re/K Fig.1. Distribution of Se species as a function of tem erature for three artial pressures of oxygen (PO,): (a) 10-6 bar; (b? 10-24 bar; and ($ equilibrium condensed selenium dioxide up to at least 500 K for a partial pressure of oxygen higher than 10-6 bar. Assuming that no significant reaction takes place between carbon and oxygen at low temperatures15 this value corresponds to an oxygen impurity in the argon of 1 p.p.m. As the reaction rate between graphite and oxygen is slow, at least up to 1000 K,16 the partial pressure of oxygen in the gas phase will decrease only slightly in the temperature range 50&1000 K. According to the calculations the main selenium species in this range are Se2(g) , SeO(g) and Se02(g). Mass spectrometric analysis of the reaction products formed in a graphite furnace operated under atmospheric pressure, as reported by Styris,l7 showed that these species were the only selenium compounds formed.The fact that the dimer was detected even at 600 K strongly indicates that at this temperature condensed selenium dioxide is reduced at the surface. This indicates that a state close to equilibrium prevails near the surface and that the dimer does not react completely with oxygen, i.e., no equilibrium is established in the gas phase. As can be seen in Fig. 2 the retention of selenium in a PGC tube (equipped with a platform) is strongly dependent on the type of purge gas used. Losses of selenium are obtained as early as the drying step if hydrogen or carbon monoxide are Table 1. Instrumental parameters for the two-step atomiser CUP Tube Stage TemperatureIK Time/s Temperature/K Time/s Drying .. . . . . . . . . 390 80 470 60 . . . . . . . . . . 470 30 Ashing - - Atomisation . . . . . . . . As tube* 8 Varied 11 Atomisationdelay . . . . . . - - - 3 Purge gas . . . . . . . . . . Argon 1000 ml min-1 * If not stated otherwise.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 289 Table 2. Furnace conditions for the Perkin-Elmer HGA-74 used for the radioactivity measurements Stage Ramp time/s Hold time/s TemperatureIK Drying . . . . . . 15 30 420 Ashing . . . . . . 20 40 Varied used. In argon, however, selenium is more efficiently retained by the graphite; the shaded area in Fig. 2 illustrates the variation in the results when measurements were performed on different days.(The relative standard deviation of the individual points measured was, on average, +5%.) It is believed that this is caused by factors such as the conditions of the platform surface and impurities in the purge gas used. The larger losses of selenium observed in the presence of hydrogen or carbon monoxide can be explained by a deactivation of the graphite surface as a result of chemisorption of hydrogen or carbon monoxide on the active carbon sites. It should be observed that in argon more than 10% of the selenium is retained in the graphite tube even at temperatures above 1500 K. Interaction between selenium and the graphite surface cannot be simulated by thermodynamic calculations as no information about such condensed compounds is available in the literature.Interaction Between Selenium and Graphite The results given in Figs. 3 and 4 were obtained by use of a two-step furnace which permits independent control of analyte volatilisation and atomisation (see Experimental). The temperature of the tube can be chosen so that optimum conditions for the formation of selenium atoms are achieved. The appearance temperature for selenium in aqueous solution using a Massmann-type furnace is known to be ca. 1100 K.18 Selenium molecules that are formed in the cup below this temperature will dissociate in the tube and give rise to an absorbance signal. It should be emphasised that in an end-heated Massmann-type furnace such molecules are nor- mally not detected unless thermal dissociation occurs owing to an increase in temperature.Fig. 3 shows that on the surface of a polycrystalline graphite cup where a larger number of active sites should be expected, selenium(1V) is stabilised to a relatively high cup temperature with an appearance temperature of 1550 K. If graphite powder (Ringsdorff-Werke) is added, selenium is retained to an even higher temperature, Tapp = 1870 K. Steady-state studies of selenium retention employing prolonged heating of the cup, over 20 s, showed no losses up to 1173 K. It should be mentioned that when active carbon or reticulated vitreous carbon (glassy carbon with a large surface area) was used instead of graphite powder some seleneium was lost at ca. 673 K. Considering all of these results, we conclude that the structure of carbon determines the extent of stabilisation which takes place.This is in line with the findings of Chung et al. 19 If in a Massmann-type furnace a graphite surface was treated with 1% V/V oxygen in argon at 2273 K selenium(1V) was found to be stabilised up to ashing temperatures of 1773 K, which is in line with the results presented by Droessler and Holcombe.20 Fig. 4 shows that the retention of selenium is much less efficient from a pyrolytic graphite coated surface. The second peak shows that part of the selenium is thermally stable to relatively high temperatures, which is in line with the results presented in Fig. 2. As can also be seen in Fig. 4, the fraction of stabilised selenium is relatively higher for small amounts of selenium, implying that stabilisation takes place at a limited number of active sites on the graphite surface.In separate experiments with the two-step furnace we found that the peak-area sensitivity was independent of the form of selenium for SeIV, SeVI and Se2- - methionine, which means 100 80 8 s g 6o > 40 20 0 I I 1 I 800 1000 1 400 600 Tern pe rat u r e/K Fig. 2. Recovery of 2 ng of 75SeIV in aqueous solution, in the presence of different purge gases, as a function of pre-treatment temperature. The shaded area gives the uncertainty when measure- ments were performed on different days 1 .o 0.8 0.6 e v) 2 0.4 0.2 0 0 2 4 6 8 10 12 Ti me/s Fig. 3. Absorbance versus time profiles for 5 ng of SeIV using the two-step atomiser. A, Cup of polycrystalline graphite; and B, with 3 mg of graphite powder added.The final tube (C) and cup (D) temperatures were 2473 K. The temperature signals are displaced due to different responses of the IR-sensitive diodes 0) 0 0.4 L m e 0.2 2 a D 2473 0) a 7 0 0 2 4 6 8 10 Ti me/s Fig. 4. Absorbance versus time profiles for Se using the two-step atomiser equipped with a yrolytic raphite coated cup. A, 5 ng SeIV, 196 nm (peak area 0.7 A sp; and B, hl ng SeIV, 206 nm (peak area 0.6 A s). Cup heating was initiated after 6 s290 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 Table 3. Fraction of selenium volatilised/atomised and retained at various tube temperatures in PGC and NG tubes. The cup temperature was 2373 K throughout Fraction of selenium atomised in the first firing,* YO Fraction of selenium retained,? % Tube temperature/# NG tube PGC tube NG tube PGC tube Tube unheated$ 1.5 f 0.3 4.2 f 1.2 64 k 1 1 4 t 5 1173 - - 91 r t 2 - 1573 15+1 52 f 3 89 + 2 12 t 7 1723 42 t 3 - 56 f 3 0 1923 110 f 6 - 6 f 2 - 1993 105 t 5 - - 0 * Related to 2373 K using the correction factor ( T/2373)1,95. i Estimated from the peak area of a subsequent firing at a tube and cup temperature of 2373 K following the firing (with injected sample) at the $ A temperature of 773 K is assumed for the diffusion correction.specified tube temperature. that the atomisation efficitxy in this furnace is independent of the temperature at which selenium is volatilised. In agreement with earlier findings using radiotracers ,I8 the signal for Se2- - methionine appeared at very low cup temperatures.However, if graphite powder was added Se2- - methionine was stabilised to temperatures similar to those for SeIV (see Fig. 3). The measurements reported in Table 3 were performed in order to investigate the extent to which selenium is retained in PGC and uncoated polycrystalline graphite (NG) tubes. The amount of selenium retained was established by measuring the absorbance during heating of the tube and cup in the two-step furnace to 2373 K in a second firing. During the first firing only the cup was heated to 2373 K while the tube was kept at the various temperatures shown in the table. The signals obtained during the first firings are also given. It should be noted that an absorption signal is obtained even for unheated tubes, which means that selenium atoms are stable without the cup.Retention is much higher in the uncoated tube, being almost 100% in the temperature range 1173-1573 K. It is interesting that the retention is significantly lower in the unheated tube. The temperatures necessary to prevent retention (above 1573 K for the pyrolytic graphite coated tube and above 1923 K for the uncoated tube) are in reasonable accord with cup temperatures required to volatilise selenium completely from different types of cup surfaces. The following picture emerges from these experiments. Selenium is volatilised, and at least partially atomised, in the cup. In the tube, selenium species interact with the surface in a manner depending on temperature and on the number of active carbon sites. At temperatures below 1923 K relatively firm selenium-carbon bonds are formed on the polycrystalline surface, which are responsible for the retention.The reaction between carbon and selenium species is favoured in the temperature range 1173-1573 K because active carbon sites are created as a result of chemisorbed oxygen removal. The unheated tube (Table 3) is less reactive and hence selenium leaving the furnace is, to a greater extent, in a molecular form. In separate experiments using hydrogen as purge gas no retention above 1273 K was observed. In this instance the main selenium compound present should be hydrogen sele- nide.18 This difference in retention behaviour that is seen for argon (Fig. 4) and hydrogen may also be attributed to the different types of gaseous selenium species present. In argon at 1600 K, Se(g) and Se2(g) are predominant while in hydrogen, H2Se(g) is the main compound, as has been discussed elsewhere.18 From this we conclude that the form of the element and the structure of the graphite surface are the main features of the retention.In separate experiments with 1% V/V hydrogen in argon using the two-step furnace, we obtained three separate peaks corresponding to selenium volatilised at low, medium and high cup temperatures. However, the signal corresponding to Table 4. Activation energies and appearance temperatures for selenium in argon using pyrolytic graphite coated tubes. The tubes had been flushed with different gases prior to sample injection. During sample injection, ashing and atomisation only argon was used; 5 pl of 5 yg ml-l SeIV were used throughout Gas EJkJ mol-1 Temperature/K Peak arealA s Argon . .. . . . 77 + 12 1160 f 20 0.93 Ifr 0.17 Oxygen? . . . . . . 160 1305 0.73 Oxygen$ . . . . . . 317 1025 0.56 Carbon monoxide§ . . 124 L 28 980 f 44 1.27 L 0.38 Hydrogen . . . . 2 0 2 4 (600)* 0.20 * Estimated value. 7 Oxygen present in tube during previous atomisation. 3 Oxygen present prior to sample injection and during drying. § Area increases with number of firings. the formation of selenium at high temperatures shifted with increasing number of determinations towards lower tempera- tures and eventually disappeared. It should also be mentioned that we observed a persistent effect from previous firings in hydrogen, resulting in volatilisation of selenium species at lower temperatures from the cup.These results show that relatively low partial pressures of hydrogen are sufficient to change the surface structure of the graphite so that the number of active sites formed decreases owing to the strong retention of hydrogen at these sites as well as to the precipitation of inactive carbon as a result of the decomposition of hydro- carbons. Table 4 illustrates the effect of treating the PGC tube with different gases prior to sample injection. It can be seen that this treatment can drastically affect peak area, the appearance temperature and the activation energy, although argon is used during drying and atomisation. It should be noted that the results given in the table were generated using side-heated tubes.* As these are heated evenly over their entire length, errors due to condensation and rs-evaporation are avoided.The most drastic effect is caused by hydrogen and these results are in line with those obtained by the radiotracer experiments. The very large differences in the activation energy values shown in the table indicate that the processes leading to the formation of free atoms are critically dependent on the nature of the graphite surface. These results might explain why different results were obtained when the reaction products from graphite furnaces were measured with two different mass spectrometric systems. 17721 Styrisl7 observed SeC2(g), Se2(g) and SeO(g) as the main compounds at low temperatures in vacuum experiments, while Holcombe and Droessler21 obser- ved SeO(g) and Se02(g).JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 291 Conclusions Selenium cannot be determined efficiently by GFAAS unless provision is made to ensure that volatilisation of analyte species takes place at sufficiently high temperatures to facilitate atom formation and minimise interactions with graphite. To achieve this goal, the use of modifiers in combination with, for example, the L’vov platform technique should be considered. The information given in this paper should also be useful for the optimisation of devices used for trapping volatile selenium compounds on heated graphite.22 The authors thank B. Hutsch, Ringsdorff-Werke GmbH, for supplying graphite parts. This work was supported by grants from the Swedish Natural Science Research Council.1. 2. 3. 4. 5. 6. References Welz, B., and Schlemmer, G., J. Anal. At. Spectrom., 1986, 1, 119. Ediger, R. D., At. Absorpt. Newsl., 1975, 14, 127. Welz, B., “Atomic Absorption Spectrometry,” Second Edi- tion, VCH Publishers, Weinheim, Deerfield Beach, FL, 1985. Welz, B., Schlemmer, G., and Vollkopf, U., Spectrochim. Acta, Part B, 1984,39, 501. Schlemmer, G., and Welz, B., Spectrochim. Acta, Part B, 1986, 41, 1157. Dedina, J., Frech, W., Lindberg, I., Lundberg, E., and Cedergren, A., J. Anal. At. Spectrom., submitted for publica- tion (J7/15). 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20 * 21. 22. Lundberg, E., Baxter, D. C., and Frech, W., J. Anal. At. Spectrom., 1986, 1, 105. Frech, W., Baxter, D. C., and Hutsch, B., Anal. Chem., 1986, 58, 1973. Lundberg, E., Chem. Instrum., 1978, 8, 197. Lundberg, E., and Frech, W., Anal. Chem., 1981, 53, 1437. Frech, W., and Cedergren, A., Anal. Chim. Acta, 1976,82,83. Barin, I., and Knacke, O., “Thermochemical Properties of Inorganic Substances,” Springer-Verlag, Berlin, Heidelberg, New York, 1973. JANAF, Thermochemical Tables, Natl. Stand. Ref. Data Ser., Natl. Bur. Stand., 1971, 37. Drowart, J., and Smoes, S., J. Chem. SOC., Faraday Trans., 1977,73, 1755. Huettner, W., and Busche, C., Fresenius Z . Anal. Chem., 1986,323, 674. Laine, N. R., Vastola, F. J., and Walker, P. L., Jr., J. Phys. Chem., 1963, 67, 2030. Styris, D. L., Fresenius Z . Anal. Chem., 1986, 323, 710. Cedergren, A., Lindberg, I., Lundberg, E., Baxter, D. C., and Frech, W., Anal. Chim. Acta, 1986, 180, 373. Chung, C. H., Iwamoto, E., Yamamoto, M., Yamamoto, Y . , and Ikeda, M., Anal. Chem., 1984,56, 829. Droessler, M. S . , and Holcombe, J. A., paper presented at FACSS, Philadelphia, 1984. Holcombe, J. A., and Droessler, M. S., personal communica- tion. Sturgeon, R. E., Willie, N. S., and Berman, S. S . , Fresenius 2. Anal. Chem., 1986, 323, 788. Paper J6l102 Received October 31st, 1986 Accepted February 2nd, I987

ISSN:0267-9477    

DOI:10.1039/JA9870200287

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 293 Comparison of Interferences and Matrix Modifiers in the Determination of Gold by Electrothermal Atomisation Atomic Absorption Spectrometry with Zeeman-eff ect Background Correction Joseph Egila, David Littlejohn and (the late) John M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, Cathedral Street, Glasgow GI IXL, UK Shan Xiao-quan Research Centre for Eco-Environmental Sciences, Academia Sinica, PO Box 934, Beijing, China Interference effects caused by percent. levels of NaCI, MgCI2, CaC12, Mg(C104)2, HN03, HCI and HC104 in the determination of Au by ETA-AAS were compared for tube-wall and platform atomisation. The effectiveness of various matrix modifiers in reducing the interferences was also assessed.The modifiers used were Ni, Pd, Pd - Mo, Cu, or Pd - Mo - Ni for wall atomisation and Ni or Pd - Mo for platform atomisation. All the modifiers increased the char temperature for Au from 800-900 "C to 1200-1300 "C. Of the interferents studied, CaCI2 and NaCl had the greatest suppressive effect on Au atomic absorption signals, at matrix concentrations between 0.1 and 0.5% mN. In contrast, HN03 and HCI caused no interference at levels up to 5% VN. The best general modifiers for Au were Ni and Pd - Mo, with the latter better for wall-atomisation measurements. Although the best Au sensitivity was obtained with Pd, the short-term stability of Au solutions was poorer with this modifier than for Pd plus Mo. For platform atomisation, Ni and Pd - Mo were equally good modifiers, and interferences were less pronounced than for wall atomisation.In general, the best conditions for interference-free analysis of Au-containing samples were addition of Ni or Pd - Mo modifier, platform atomisation and peak-area measurement. Keywords: Interference effect comparison; gold determination; platform and tube-wall atomisation; matrix modification; Zeeman-e ffect background correction Developments in graphite furnace technology have greatly reduced the severity of matrix interferences encountered in the determination of volatile elements by electrothermal atomisation atomic absorption spectrometry (ETA-AAS). For elements such as arsenic, cadmium, lead and selenium, the addition of a matrix modifier to the standard and sample solutions allows the use of a higher than normal ashlchar temperature and more of the matrix can be removed before the atomisation stage, without loss of the analyte. In addition, a combination of rapid heating and platform atomisation ensures that a high vapour temperature exists in the tube when the sample is vapourised during the atomisation stage, which reduces the influence that the residual matrix components exert on analyte atom formation.Application of an efficient background correction routine also ensures that spectral interferences caused by matrix components are adequately compensated for at this stage. For many samples, use of integrated absorbance (peak area) measurements further minimises the effect that matrix components can have on the accuracy of an analysis. Gold is one of the elements likely to benefit most from a combination of the above developments.For aqueous gold solutions, the maximum char temperature prior to loss of the element is 800-900 "C, just below the char temperature of carbonaceous material and the volatilisation temperature of common salts such as sodium chloride and magnesium chloride. Use of a matrix modifier to increase the char temperature by 300-400 "C can therefore be of great benefit as both chemical and spectral interferences are substantially reduced. The determination of gold in samples containing significant concentrations of sodium chloride has often caused difficulties because it exerts a severe chemical interference on the formation of gold atoms1 and a substantial peak in the sodium chloride molecular spectrum occurs close to the most sensitive gold wavelength of 242.8 nm.2 The recommended matrix modifier for gold is nickel,3 and Slavin et aE.4 reported that nickel in combination with platform atomisation and Zeeman-effect background correction reduced the interfer- ence of sodium, potassium and magnesium salts on gold.In general, however, relatively few studies have been conducted to assess the effectiveness of nickel in reducing matrix interferences and no detailed evaluation of alternative modifi- ers for gold has been reported. From the literature on the interference on gold it appears that a number of elements may have a desirable effect on the atomisation of gold and could be considered as possible matrix modifiers.McHugh5 investi- gated the interference effect of a great number of elements on gold and discovered that silver and copper enhanced the recovery. From earlier studies, palladium was known to retard the volatilisation of gold697 and recently this element has been used successfully as a modifier for several elements.8?9 The aims of this study were: ( a ) to compare the interfer- ences caused by various matrices on the determination of gold by ETA-AAS using tube-wall and platform atomisation and ( b ) to evaluate the performance of a number of matrix modifiers in reducing these effects. As the intention was to develop a method for the determination of gold in biological samples, the interferents tested included sodium chloride, magnesium chloride, calcium chloride, magnesium perchlor- ate and mineral acids often used to treat biological samples prior to analysis (hydrochloric acid, nitric acid and perchloric acid).The matrix modifiers investigated were nickel, pallad- ium, molybdenum plus palladium, copper and a combination of molybdenum, palladium and nickel. The results indicated that all the modifiers were able to stabilise gold, although to slightly different maximum char temperatures. A mixture of palladium and molybdenum was found to have some advan- tages over nickel in that it gave improved sensitivity for gold, particularly in platform atomisation, and was more effective in reducing interferences than nickel, except those caused by sodium chloride. As expected, matrix interferences were lower when a combination of platform atomisation, matrix modification and peak-area absorbance measurements were used.294 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 Experimental Equipment A Perkin-Elmer Model Zeeman 5000 atomic absorption spectrometer equipped with an HGA-400 graphite furnace atomiser was used for all experiments. The spectrometer spectral band pass was 0.7 nm and a Perkin-Elmer gold hollow-cathode lamp, operated at 10 mA, was used to obtain peak-height and peak-area absorbance measurements at the Au wavelength of 242.8 nm. A Perkin-Elmer AS-40 auto- sampler was used to deposit 20-p1 volumes of the test solutions in the HGA-400 atomiser, although manual sample introduc- tion was used for some measurements. A Perkin-Elmer Data System 10 was connected to the Perkin-Elmer Zeeman 5000 spectrometer for part of the study.Perkin-Elmer pyrolytic graphite coated tubes were used for wall-atomisation experiments and the platform version of this tube was used in conjunction with a Perkin-Elmer total pyrolytic graphite platform for platform-atomisation studies. The atomiser programmes for tube-wall and platform atom- isation of gold are given in Table 1. Reagents All chemicals used in this study were of AnalaR grade or equivalent and solutions were prepared using distilled water. Analyte solution A 1000 pg ml-1 Au solution was prepared by dissolving 0.2020 g of NaAuC14.2H20 in 100 ml of 1% VlV HN03 and was stored in a calibrated flask in a dark cupboard. Working standard solutions were prepared from this stock reagent by serial dilution with either distilled water or 0.01% VlVHN03.Interferent solutions A 10% mlV NaCl stock solution was prepared by dissolving 10 g of NaCl in 100 ml of distilled water. Interferent solutions in the range 0.054% mlVNaCl were prepared from this stock by dilution with distilled water. A 2.34% mlV MgC12 stock solution was prepared by dissolving 5.00 g of MgC12.6H20 in 100 ml of distilled water. Interferent solutions in the range 0.02-0.5% MgC12 were prepared from this stock by dilution with distilled water. A 2.54% mlV CaC12 stock solution was prepared by dissolving 5.00 g of CaC12.6Hz0 in 100 ml of distilled water. Interferent solutions in the range 0.025-0.5% CaC12 were prepared from this stock by dilution with distilled water. A 4.63% mlV Mg(C104)2 stock solution was prepared by dissolving 5.00 g of Mg(C104)2.H20 in 100 ml of distilled Table 1.HGA-400 atomiser programmes for determination of gold in aqueous solution Stage Parameter Dry Char Atomise Clean (a) Wall atomisation- TemperaturePC . . . . . . 130 800-1300* 2400 2700 Ramp/s 5 5 0 1 30 5 3 Holdis . . . . . . . . . . 30 Internal gas flow-rate/ml min-1 300 300 0 300 Recorder - - x 0 Read . . . . . . . . . . . . . . . . . . . . . . . . . . X - - - ( b ) Platform atomisation- r-rows . . . . . . . . . . 40 30 5 - Internal gas flow-rate/ml min-' 300 300 0 300 Recorder . . . . . . . . - - X 0 Read - Temperature/"C . . . . . . 150 900-1300* 2400 2700 Ramp/s 5 0 1 . . . . . . . . 5 - - X . . . . . . . . . . * Depending on matrix modifier used. water.Interferent solutions in the range 0.1-1.0% m/V were prepared from this stock by dilution with distilled water. Interferent solutions in the range 0.1-5.0% VlV HN03 or HC1 were prepared by dilution of concentrated nitric acid or hydrochloric acid with distilled water. Interferent solutions in the range 0.1-3.0% VlV HC104 are prepared by dilution of concentrated perchloric acid with distilled water. Matrix modifier solutions The concentrations of matrix modifiers included in Au solutions injected into the atomiser were: (a) 1000 pg ml-1 of Ni; ( b ) 200 pg ml-1 of Pd; (c) 1000 pg ml-1 of Mo plus 200 pg ml-1 of Pd; (4 1000 pg ml-1 of Cu; and (e) 1000 pg ml-1 of Mo plus 200 pg ml-1 of Pd plus 1000 pg ml-1 of Ni . Stock solutions of the modifiers were prepared as follows. A 15000 pg ml-1 Ni stock solution was prepared by dissolving 7.4295 g of Ni(N03).6H20 in 100 ml of distilled water.A 1000 pg ml-1 Pd stock solution was prepared by dissolving 0.1669 g of PdC12 in 50 ml of dilute nitric acid with gentle heating. The solution was then transferred into a 100-ml calibrated flask and diluted to the mark with distilled water. A 4000 pg ml-1 Mo stock solution was prepared by dissolving 0.7360 g of (NH4)6M07024.4H20 in 100 ml of distilled water. A 15000 pg ml-1 Cu stock solution was prepared by dissolving 5.7030 g of C U ( N O ~ ) ~ . ~ H ~ O in 100 ml of distilled water. Procedure To assess the effect that each of the above interferents had on the atomisation of Au, a series of solutions were prepared that contained 25 pg 1-1 of Au, a range of concentrations of the interferent and one of the modifiers (a)-(e) given above.As five modifiers were investigated during this study, a total of five sets of solutions were prepared for each interference experiment. The peak-height and peak-area signals obtained for each interferent - modifier combination were compared with the signals obtained from the appropriate solution without a modifier (i.e., prepared in O.0lo/~ V/V HN03). Measurements were obtained for both tube-wall and platform atomisation and Zeeman-effect background correction was used throughout the study. All the interferent and modifier solutions were found to be free of gold contamination, but appropriate blank solution measurements were always made as a precaution during the various interference experiments.Table 2. Maximum char temperatures for gold* in presence of different matrix modifiers-wall atomisation Modifier Absorbance* Char temperaturelac None (0.01% V/VHN03) . . 0.33 k 0.03 800 200 pg ml-1 Pd . . . . . . 0.38 4 0.02 1300 200 pg ml-1 Pd + 1000 pg ml-1 Ni . . . . . . 0.29 4 0.01 1200 1000pgml-1 Mo . . . . 0.35 k 0.01 1200 * 25 pg 1-1 Au; n = 8 (ie., data obtained from eight experiments performed on several days); see Table 1 for full atomiser programme. Table 3. Maximum char temperature for gold* in presence of different matrix modifiers-platform atomisation Modifier Absorbance* Char temperature/"C None (0.01% V/VHN03) . . 0.36 4 0.03 900 1000pgml-1Ni . . . . . . 0.19kO.02 1300 200 pg ml-1 Pd + 1000 pgml-1 Mo . . . . 0.31 4 0.03 1300 * 25 pg 1-1 Au; n = 6 (i.e., data obtained from six experiments performed on several days); see Table 1 for full atomiser programme.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 295 Results and Discussion Effect of Matrix Modifiers on the Maximum Char Tempera- ture for Gold When no matrix modifier was added to an aqueous Au solution, the maximum char temperature that could be used was 800 "C. As Fig. l(a) indicates, this temperature was insufficient to avoid background-correction problems caused by 0.1% mlV NaCl, even when using Zeeman-effect back- ground correction. The base-line disturbance that occurred in the corrected signal for 0.1 YO m/V NaCl at the Au wavelength could cause errors in both peak-height and peak-area measurements when an 800 "C char temperature was used.To remove more of the NaCl matrix, a higher char temperature was required, but as illustrated in Fig. l(b) a significant reduction in the Au signal occurred when a 1200 "C char temperature was used without matrix modification. It has long been known that Ni is an effective modifier for Au and no reduction in both peak height or peak area Au atomic absorption signals was observed up to a char temperature of 1200 "C [Fig. l(c)] when Ni was included in the Au solutions. However, as illustrated in Table 2,1000 pg ml-1 of Ni caused a reduction in the peak-height signal for 25 pg 1-1 of Au obtained with tube-wall atomisation, compared with the signal obtained for the same Au concentration without the Ni modifier. When Pd was used as a matrix modifier for Au a maximum char temperature of 1300 "C was achieved for tube-wall 2.0 I 1 j(dJ nB ' * O t I \ 0 k 0 3.0 0 3.0 0 3.0 Time/s Fig.1. Influence of char temperature on Au AAS and NaCl background signals at 242.8 nm when using Zeeman-effect back- ground correction and an HGA-400 atomiser. (a) 800 "C and no matrix modifier: A, 0.1% mlVNaC1 and B, 100 pg 1-l Au. ( b ) 1200 "C and no matrix modifier: A, 0.1% mlV NaCl and B, 100 pg 1-1 Au. (c) 1200 "C: A, 100 pg 1-1 Au with no matrix modifier and B, 100 pg 1-' Au plus 1000 kg 1-1 Ni as modifier. All signals obtained with wall atomisation at 2400 "C in a pyrolytic graphite coated tube using 20 pl solution volumes atomisation and an enhancement in the peak-height signal for Au was obtained, as indicated in Table 2.The effect of Ni and Pd on peak-area signals for Au with tube-wall atomisation followed the same proportional trend as the peak-height results, with 25 pg 1-1 of Au giving a signal of 0.10 A s in the presence of Ni and 0.13 A s in the presence of Pd. Although a similar enhancement in the Au signal can be achieved with concentrations as low as 25 pg ml-1 of Pd, most experiments were conducted with a concentration of 200 pg ml-1 of Pd to allow a margin of safety when higher concentrations of matrix interferent were present in solution. During the Pd experiments it was observed that the stability of Au solutions containing Pd was reduced. Hence a mixture of Pd and Mo was also considered as a matrix modifier, as this combination enhanced the stability of Au solutions over a longer period.As indicated in Table 3, a combination of 200 pg ml-1 of Pd and 1000 pg ml-1 of Mo allowed char temperatures up to 1200 "C for tube-wall atomisation of Au and gave a similar enhancement in the Au signal to that obtained with Pd alone. A detailed discussion of the stability of Au solutions in the presence of various modifiers is included in a separate paper concerned with the determination of Au in blood.10 The advantages of the Pd - Mo modifier over Ni are even more pronounced in platform atomisation as indicated in Table 3. A similar char temperature can be achieved with both modifiers (i.e., 1300 "C), but better Au sensitivity is achieved with Pd - Mo. Although the results in Table 3 refer only to peak-height measurements for platform atomisation, a similar trend was observed for peak-area measurements.From the results obtained in the char optimisation experi- ments, it was decided to investigate the influence of Ni, Pd and Pd - Mo on the interference effects caused by various reagents on Au. Two other modifiers were also included in the interference study: a combined Pd - Mo - Ni modifier and Cu, which was found to increase the Au char temperature to 1200 "C without significantly influencing the Au sensitivity. Comparison of Interference Effects on Gold Experiments were conducted to compare the degree of interference caused by various chloride salts and mineral acids on Au atomic absorption signals obtained using tube-wall and platform atomisation. The ability of various matrix modifiers in reducing these interference effects was also investigated.In the following discussion the effect of each interferent is considered individually. Mineral acids Nitric, hydrochloric and perchloric acids are often used to decompose Au-containing samples and so the interference caused by various concentrations of each acid was investi- gated. No interference was observed in either peak-height or peak-area measurements for HN03 concentrations in the range 0.1-5.0% V/V with or without the presence of Ni, Pd or Pd - Mo modifiers. Hydrochloric acid concentrations in the range 0.1-5.0% V/V had no influence on the peak-area measurements for Au obtained with either tube-wall or platform atomisation, with or wiEhout the presence of Ni, Pd or Pd - Mo modifiers.For peak height wall atomisation measurements, however, a 5-10% decrease in the Au atomic absorption signal was obtained up to 1.0% V/V HCl in the presence of Ni, Pd or Pd - Mo modifiers. In contrast, a 5-10% increase in the Au signal occurred over the same HC1 concentration range when no modifier was included in the solution. In all instances no further change in the peak-height signals occurred for HC1 concentrations between 1 .O and 5.0% V/V. Similar trends were observed for peak height platform atomisation measurements, although no significant interference of HC1 on Au occurred up to a concentration of296 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 5.0% VlVin the presence of the Pd - Mo modifier. It is known that perchloric acid can cause greater interference effects than other mineral acids fw some elements, but no significant effect was observed on Au for HC104 concentrations in the range 0.1-3.0% V/V provided peak-area measurements were used in either tube-wall or platform atomisation.Peak-height wall atomisation measurements for Au decreased gradually with an increase in perchloric acid concentration from 0 to 1.0% VlV in the absence or presence of the Ni modifier. However, there was no interference over the same perchloric acid concentration range when Pd or Pd - Mo modifiers were included in the solutions. With platform atomisation, peak- height measurements for Au were relatively unchanged over the HC104 concentration range of &3.0% VlV provided either the Ni or Pd - Mo modifier was used.Magnesium perchlorate When perchloric acid is used to decompose samples, the residue often contains significant concentrations of perchlor- ate salts. The interference effect of magnesium perchlorate was therefore investigated. If Ni or no matrix modifier was used, the Au peak-height and peak-area signals obtained with wall atomisation de- creased steadily as the Mg(C104)2 concentration was increased from 0 to 1.0% mlV. For both measurement modes the recovery at 1.0% mlV was only 50%. Use of platform atomisation did not greatly improve the situation when Ni or no matrix modifier was used. In comparison, no significant interference by Mg(C104)2 was observed in the concentration range &1.0% mlV for peak-height or peak-area measure- ments when Pd, Pd - Mo or Pd - Mo - Ni modifiers were used.The Cu modifier was not quite as good as the rest but was still an improvement on Ni at higher Mg(C104)2 concentrations. A summary of the percentage recovery figures obtained for 0.1 and 0.5% mlV Mg(C104)2 is given in Table 4 for wall atomisation and Table 5 for platform atomisation. Sodium chloride When tube-wall and no matrix modification was used the peak-height and peak-area signals for Au decreased rapidly as the NaCl concentration increased from 0 to 1.0% mlV. At 1.0% mlVNaC1, the recovery figures for peak height and peak area were 28 and 42% , respectively. In the presence of Ni, the suppressive effect was greatly reduced and better than 90% recovery was achieved in peak height and peak area for concentrations between 0.5 and 4.0% mlV NaC1.The other modifiers were not as effective in reducing the NaCl interfer- ence on Au, but all were a distinct improvement on no modifier giving recoveries of ca. 80% for most measurements in the NaCl concentration range of 0.5-2.0% mlV. Copper was almost as good a modifier as Ni, especially in the peak-area mode. When platform atomisation was used the recovery of Au was greatly improved over the equivalent tube-wall measure- ments for each modification condition studied. Again, Ni proved the best modifier giving almost quantitative recovery of Au at NaCl concentrations up to 4.0% mlVfor peak-height and peak-area measurements. The Pd - Mo modifier gave only slightly poorer performance and recovery close to 100% was achieved with peak-area measurements up to 2.0% mlVNaC1.A summary of the percentage recovery figures for 0.1 and 0.5% mlV NaCl for various conditions of matrix modification is given in Table 6 for wall atomisation and Table 7 for platform atomisation. In all experiments peak-area measure- ments normally gave better recovery figures than peak-height measurements. Magnesium chloride With no matrix modification and tube-wall atomisation the interference caused by 0.025-0.5% m/V MgC12 was less than that produced by equivalent concentrations of NaCl. Indeed, the recovery was ca. 75% in both peak-height and peak-area modes at a MgC12 concentration of 0.5% mlV. Unlike the NaCl experiments, Ni was the least successful of the modifiers Table 4. Effect of matrix modifier on percentage recovery of gold in presence of Mg(C104)2-wall atomisation (25 pg 1-1 Au, 2 0 4 injections, HGA-400 and Perkin-Elmer Zeeman-5000) Table 6.Effect of matrix modifier on percentage recovery of gold in presence of NaC1-wall atomisation (25 pg 1-1 Au, 2 0 4 injections, HGA-400 and Perkin-Elmer Zeeman-5000) Mg(C10,J2, YO m/V NaCl, Yo m/V 0.1 Modifier HT* A t lOOOpgml-1Ni . . . . . . . . . . 94 94 200pgml-1Pd . . . . . . . . . . 100 94 1000 pg ml-1 Mo + 200 pg ml-1 Pd . . 94 88 lOOOpgml-1Cu . . . . . . . . 86 89 1000 pg ml-1 Mo + 200 pg ml-l Pd + 1000pgml-1Ni . . . . . . . . 103 90 * HT, peak-height absorbance measurements. t A, peak-area absorbance measurements. None (0.01% V/VHN03) . . . . . . 96 92 ~~ 0.5 0.1 0.5 HT* A t 62 62 67 69 95 88 95 85 82 84 100 90 Modifier None (0.01 YO V/V HN03) 1000 pg ml-1 Ni .. . . . . . . . . 200 pg ml-1 Pd . . . . . . . . . . 1000 pg ml-1 Cu . . . . . . . . 1000 pg ml-l Ni . . . . . . . . . . . . . . 1000 pg ml-1 Mo + 200 pg ml-1 Pd 1000 pg ml-1 Mo + 200 pg ml-1 Pd + . . * Result for 0.05% m/V NaCl. ~~ HT A 73 93 96 96 95 93 88 86 97* 96* 96 90 HT A 37 59 90 94 76 86 64 69 86 96 84 85 Table 5. Effect of matrix modifier on percentage recovery of gold in presence of Mg(C104),-platform atomisation (25 pg I-' Au, 2 0 4 injections, HGA-400 and Perkin-Elmer Zeeman-5000) Mg(C104)2, YO m/V 0.1 0.5 Modifier HT A HT A 1000pgml-1Ni . . . . . . . . . . 96 106 75 81 1000 pgml-1 Mo + 200pgml-1Pd , . 103 99 101 96 None(0.01% VIVHNO,) . . . . . . 56 91 41 67 Table 7. Effect of matrix modifier on percentage recovery of gold in presence of NaC1-platform atomisation (25 pg I-' Au, 20-pl injections, HGA-400 and Perkin-Elmer Zeeman-5000) NaCl, YO m/V 0.1 0.5 Modifier HT A HT A None (0.01% V/VHN03) .. . . . . 93 87 * * lOOOp.gml-1Ni . . . . . . . . . . ND-I- ND-I- 114 103 1000 yg ml-1 Mo + 200 pg ml-l Pd . . 97 99 99 98 * Incomplete background correction. t ND, not determined.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 297 Table 8. Effect of matrix modifier on percentage recovery of gold in presence of MgCl,-wall atomisation (25 pg 1-1 Au, 20-pl injections, HGA-400 and Perkin-Elmer Zeeman-5000) MgCl2, YO m/V Table 10. Effect of matrix modifier on percentage recovery of gold in presence of CaCl,-wall atomisation (25 pg 1-1 Au, 2 0 4 injections, HGA-400 and Perkin-Elmer Zeeman-5000) CaCl,, YO m/V 0.1 0.5 0.1 0.5 Modifier HT A None (0.01% V/VHN03) .. . . . . 94 85 1000pgrnlk'Ni . . . . . . . . . . 79 81 200pgml-1Pd . . . . . . . . . . 97 93 1000 pg ml-1 Mo + 200 pg ml-l Pd . . 98 91 1000 pg ml-1 Mo + 200 pg ml-I Pd + lOOOpgml-1Cu . . . . . . . . 105 100 1000 pg ml-1 Ni . . . . . . . . 105 104 HT A 76 73 80 80 88 86 92 84 87 90 99 89 Table 9. Effect of matrix modifier on percentage recovery of gold in presence of MgC1,-platform atomisation (25 pg 1-1 Au, 20-@ injections, HGA-400 and Perkin-Elmer Zeeman-5000) MgC12, YO m/V 0.1 0.5 Modifier HT A HT A 1OOOpgml-lNi . . . . . . . . . . 81 85 84 79 1000pgml-1Mo + 200pgml-lPd . . 98 93 78 74 None(0.01% V/VHN03) . . . . . . 72 87 58 81 investigated. A recovery of ca.80% was achieved with Ni at MgC12 concentrations in the range 0.1-0.5% mlV. Over the same interferent range, recoveries better than 90% are generally achieved for all the other modifiers with the Pd - Mo - Ni combination giving almost quantitative recovery. Surprisingly, with tube-wall atomisation, peak-height recoveries were often better than the equivalent peak-area figures for the MgC12 interferent. Use of platform atomisation did not improve the recovery figures when Ni or no matrix modification was used. Although quantitative recovery of Au was achieved by peak height or peak area with the Pd - Mo modifier up to 0.25% mlVMgC12, the recovery figures at 0.5% mlV with platform atomisation were poorer than the equivalent results obtained with wall atomisation for this modifier.Nevertheless, 80% recovery or better could be achieved with platform atomisation at MgC12 concentrations up to 0.5% mlvwhether matrix modifiers were used or not. A summary of the percentage recovery achieved at 0.1% and 0.5% mlVMgC12 with various matrix modifiers is given in Table 8 for wall atomisation and Table 9 for platform atomisation. Calcium chloride With tube-wall atomisation and no matrix modification, the interference caused by CaC12 was slightly worse than with NaCl and significantly worse than with MgCl2 at equivalent concentrations. As little as 0.025% mlV CaC12 reduced the peak-height recovery for Au to 70%. Again, Ni was not as effective a modifier as the other combinations, particularly at lower CaC12 concentrations. In the range 04.1% mlVCaC12, Pd and Pd - Mo modifiers gave almost quantitative recovery in both peak-height and peak-area measurements with tube-wall atomisation.Both Cu and Pd - Mo - Ni were also more effective than Ni at this CaC12 concentration range with recoveries better than 80%. At higher concentrations of CaC12, the performance of the various modifiers was more similar and close to the situation without a matrix modifier, although Pd was still better with recoveries of 80% in the peak-area mode at 0.5% mlV CaC12. Modifier None (0.01% V/VHN03) . . . . . . 1000 pg ml-1 Ni . . . . . . . . . . 200 pg ml-1 Pd . . . . . . . . . . 1000 pg ml-1 Cu . . . . . . . . 1000 pg ml-1 Ni . . . . . . . . 1000 pg ml-1 Mo + 200 pg ml-1 Pd 1000 pg ml-1 Mo + 200 pg ml-1 Pd + . . * Result for 0.25% m/V CaC1,.~ HT A 47 65 73 78 95 93 95 100 83 86 87 80 HT A 47* 60* 44 53 66 80 66 71 53 62 58 52 Table 11. Effect of matrix modifier on percentage recovery of gold in presence of CaC1,-platform atomisation (25 pg 1-1 Au, 2 0 4 injections, HGA-400 and Perkin-Elmer Zeeman-5000) CaCl,, YO m/V 0.1 0.5 Modifier HT A HT A None(0.01% VlVHNO,) . . . . . . 80 99 77* 88* 1OOOpgml-1Ni . . . . . . . . . . 93 105 88* 101* 1000pgml-lMo+ 200pgml-1Pd . . 106 87 90 81 * Results for 0.25% m/V CaC1,. The use of platform atomisation greatly reduced the interference caused by CaC12 when no matrix modifier was used. In the peak-area mode, recoveries were reduced only slightly to 88% for a CaC12 concentration of 0.5% mlV. Even better recoveries were obtained with matrix modification, with Ni proving to be slightly better than Pd - Mo particularly in the peak-area mode.A summary of the percentage recovery figures obtained for 0.1 and 0.5% mlVCaCl2 with various matrix modifiers is given in Table 10 for wall atomisation and Table 11 for platform atomisation. In general, better recovery was achieved with peak-area measurements, although in platform atomisation there was no significant difference between the recoveries achieved in each experiment with height and area measure- ments. Conclusions The spectral and chemical interferences caused by chloride salts and perchloric acid during the determination of Au by ETA-AAS can be substantially reduced if a char temperature of 1200 "C or higher is used. At this temperature, addition of a matrix modifier to standard and sample solutions is required to prevent pre-volatilisation loss of Au atoms.In this study a comparison of the interference caused by various compounds in tube-wall and platform atomisation of Au has been made, and the effectiveness of a number of matrix modifiers for Au has been assessed. Without matrix modification, recovery of Au was reduced by up to 50% when solutions contained between 0.1 and 0.5% mlV interferent. The suppressive effect on peak-height and peak-area signals of Au was in the order CaC12 2 NaCl> MgC12 > Mg(C104)2. Use of platform atomisation and no matrix modification improved the recovery of Au, particularly in the peak-area mode. However, recoveries were still normally less than 80% at the 0.5% mlvinterferent level, and as little of the matrix was removed during the char stage, background correction problems were often encountered even using the Zeeman-effect system.298 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 Of the various matrix modifiers investigated, Ni, Pd and Pd - Mo were the most useful. For tube-wall measurements, Pd and Pd - Mo were superior to Ni except at NaCl concentrations above 0.1% mlV. The other modifiers studied, Cu and Pd - Mo - Ni also proved useful stabilising agents for Au and were generally of similar performance to Ni. Most of the platform atomisation measurements were obtained using either Ni or Pd - Mo as the matrix modifier. Both agents performed equally well and quantitative recovery was obtained up to 0.5% mlV for all the interferents except MgC12 where the best recovery was ca.80%. In tube-wall atomisation, peak-area recoveries were generally better than peak height, but there was no significant difference between the two measurement modes when platform atomisation was used. As the stability of Au solutions was improved by adding a mixture of Pd and Mo rather than Pd alone, it was concluded that the best matrix modifier for Au was either Ni or Pd - Mo. The best conditions for quantitative recovery with either modifier were a combination of platform atomisation and peak-area measurement. Application of Zeeman-effect back- ground correction also ensured that spectral interferences were adequately compensated for. Both modifiers have been used in a study to develop a direct method for the determina- tion of Au in whole blood and plasma samples obtained from patients receiving chrysotherapy as treatment for rheumatoid arthritis, as described in the accompanying paper.10 The financial support provided by the Institute of Environ- mental Chemistry, Chinese Academy of Sciences and the British Council for the study is greatly appreciated. The authors are grateful to Perkin-Elmer Inc. for substantial support in terms of the instrumentation used in this project including the P-E 5000 Spectrometer (Sabina Slavin, Nor- walk, CT, USA), the AS-40 Autosampler (Bernhard Welz, Bodenseewerk, FRG) and the Model 3600 AS Data System 10 (Paul Mitchell, Beaconsfield, UK). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. References Ward, R. J., Danpure, C. J., and Fyfe, D. A., Clin. Chim. Acta., 1977, 81, 87. Adams, N. J., Kirkbright, G. F., and Rienvatana, P., At, Absorpt. Newsl., 1975, 14, 105. Frech, W., Lundberg, E., and Cedergren, A., Prog. Anal. At. Spectrosc., 1985, 8, 278. Slavin, W., Carnrick, G. R., Manning, D. C., and Pruszkow- ska, E., At. Spectrosc., 1983, 4, 69. McHugh, J. B., At. Spectrosc., 1983,4, 66. Aggett, J., and West, T. S., Anal. Chim. Acta., 1971,55,349. Guerin, B. D., J. S . Afr. Chem. Znst., 1972, 25, 230. Xiao-quan, S., and Kai-Jing, H., Talanta, 1985,32, 23. Xiao-quan, S., Zhe-ming, N., and Zhi-neng, Y., Anal. Chim. Acta., 1985, 171, 269. Xiao-quan, S., Egila, J., Littlejohn, D., and Ottaway, J. M., J. Anal. At. Spectrom., 1987, 2, 299. Paper J61113 Received November 1 Oth, 1986 Accepted December 9th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200293

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 299 Direct Determination of Gold in Whole Blood and Plasma by Electrothermal Atomisation Atomic Absorption Spectrometry Using Zeeman-eff ect Background Correction and Matrix Modifications Shan Xiao-quan Research Centre for Eco-Environmental Sciences, Academia Sinica, PO Box 934, Beijing, China Joseph Egila, David Littlejohn and (the late) John M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, Cathedral Street, Glasgow GI IXL, UK Addition of 1000 pg ml-1 of Ni or 200 pg ml-1 of Pd - 1000 pg ml-1 of Mo matrix modifiers to 80-fold diluted whole blood and 200-fold dilute plasma has allowed the direct determination of Au using aqueous standards. With Pd - Mo, peak-height and peak-area measurements were satisfactory for both tube-wall and platform atomisation (91-108% recovery).With Ni however, peak-area measurements and platform atomisation were required to achieve quantitative recovery of added Au. For the dilution factors applied in this study, the lower limit of measurement was 0.16 pg ml-1 of Au for plasma and 0.064 pg mi-1 of Au for whole blood, based on 95% confidence limits. Solution detection limits in the presence of the Pd - Mo modifier were 0.4 pg 1-1 of Au for aqueous or dilute HN03 solutions and 0.8 pg 1-1 of Au for diluted samples of whole blood or plasma. Storage experiments indicated that glass containers were better than plastic to avoid deposition and loss of Au over periods of up to 80 min. Keywords: Gold in whole blood and plasma; nickel or palladium - molybdenum matrix modification; platform and tube-wall atomisation; Zeeman-effect background correction; electrothermal atomisation atomic absorption spectrometry Gold compounds, such as sodium aurothiomalate and sodium aurothioglucose, are used in the treatment of rheumatoid arthritis and there is good evidence to indicate that they make an effective contribution to therapy.However, the accumula- tion of high concentrations of Au in the blood stream has undesirable effects and so the monitoring of Au levels in the serum of patients undergoing chrysotherapy is important. A number of attempts have been made to determine the concentration of Au in body fluids and related samples by electrothermal atomisation atomic absorption spectrometry (ETA-AAS) ,1-13 but few direct procedures have been repor- ted.In many instances, samples have been digested with acid mixtures and either matrix-matched standards or standard additions used to minimise matrix interferences. In an early ETA-AAS study, it was observed that the matrix components of serum had a suppressive effect on gold signals1 and calibration graphs had to be constructed using standard solutions of Au in serum. Kame1 et aZ.3 compared various procedures for the determination of Au in blood fractions and concluded that direct injection of blood serum into a graphite furnace was the most suitable, provided matrix-matched standards were used. They also applied ETA-AAS to deter- mine the concentration of Au in blood proteins after separation by electrophoresis, gel chromatography and preci- pitation.4 Sharma reported that the use of matrix-matched standards and effective background correction were essential in the determination of Au in digested whole blood, serum and tissue.ll Digestion with H2S04 - HN03 and calibration by standard additions was also considered necessary for the determination of Au in synovial fluid.12 The chemical and spectral interferences encountered when analysing biological samples for Au are a direct consequence of the volatility of the analyte which precludes the use of char or ash temperatures that would remove the bulk of interferent salts and compounds. Various attempts have been made to increase the char/ash temperature of gold in biological analysis.Addition of a dilute mixture of phosphoric acid, ammonium nitrate and a detergent allowed use of a higher char temperature in the determination of Au in serum and matrix interferences were reduced.9 Nickel is a more com- monly applied matrix modifier for Au, but there are relatively few reports of its use in biological analysis.When Ni is added to both standard and sample solutions, a char temperature of up to 1200 "C can be used before loss of Au 0ccurs.1~ Alternatively, interferences can be avoided by separating the Au through solvent extraction,3 or by treating samples and standards with a quaternary ammonium hydroxide reagent in toluene, followed by dilution with isobutyl methyl ketone.10 Matrix interferences are not the only problems encountered in the determination of Au by ETA-AAS.It is well known that the stability of Au solutions is affected by exposure to bright sunlight and by base-exchange reactions and adsorp- tions on glass containers.15 Several papers have dealt with this subject, but often the conclusions reached have been contra- dictory. Chow and Beamishl6 reported that for solutions in the range 2.5-25 pg ml-1 of Au, the concentration of Au remained at 9699% of the original value after 300-d storage. Chow also claimed that a loss of 4% or less occurred over a period of 400 d for solutions containing 12-258 pg ml-1 of Au.17 In contrast, Sighinolfi and Santosls concluded that very dilute aqueous Au solutions (10-20 pg ml-1) were unstable over a period of several hours. The reasons for these contradictory conclusions are unclear, but are probably due to differences in the storage conditions and procedures used for Au determination.To assess the effectiveness of chrysotherapy, it is necessary to conduct a number of biochemical tests. Those involving the determination of Au include the analysis of whole blood, plasma, red cell lysate and clear lysate.13 It is clearly desirable therefore to have a quick and reliable method that is equally suitable for determination of the comparatively high levels of Au in whole blood and plasma and the lower concentrations that exist in lysed material. This study describes such a method based on the direct determination of Au by ETA-AAS using Zeeman-effect background correction and matrix modifica- tion with either tube-wall or platform atomisation.In a previous investigation14 the effectiveness of a number of matrix modifiers for Au was assessed and two reagents, either300 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 Ni or a combination of Pd and Mo, were considered suitable for general application. Both modifiers have been successfully applied in the present study, with the Pd - Mo mixture showing some advantages over Ni. The stability of Au solutions, with and without the presence of matrix modifiers, was also investigated and was found to be better when the solutions were stored in glass containers rather than plastic containers. The methods described are shown to be free of matrix interferences and can be applied to a variety of biological materials. Experimental Equipment A Perkin-Elmer Model Zeeman 5000 atomic absorption spectrometer and HGA-400 graphite furnace atomiser were used for all measurements.The atomiser was equipped with pyrolytic graphite coated tubes for wall atomisation or platform atomisation using a total pyrolytic graphite platform. A 20-pl micro-pipette was used for introduction of sample and standard solutions, and peak-height and peak-area signals were obtained at the Au wavelength of 242.8 nm. The efficiency of the Zeeman-effect background correction system was checked by making equivalent measurements at the Au non-absorbing line of 242.1 nm. The optimum atomiser progammes for tube-wall and platform atomisation of Au are given Table 1. Reagents AnalaR grade reagents were used for the preparation of all solutions.A 1000 pg ml-1 Au stock solution was prepared by dissolving 0.2020 g of NaAuC14.2H20 in 100 ml of 1% V/V HN03. Standard solutions were prepared from this stock by serial dilution with either distilled water or 0.1% V/V HN03. A 1500 pg ml-1 Ni solution was prepared by dissolving 7.4295 g of Ni(N03)2.6H20 in 100 ml of distilled water. The matrix modifier solution prepared from this stock contained 1000 yg ml-1 of Ni. A 1000 pg ml-1 Pd solution was prepared by dissolving 0.1669 g of PdC12 in 50 ml of dilute nitric acid with gentle heating. The solution was then transferred into a 100-ml calibrated flask and diluted to the mark with distilled water. A 4000 pg ml-1 Mo solution was prepared by dissolving 0.7360 g of (NH4)6M07024.4H20 in 100 ml of distilled water. The matrix modifier solution prepared from the Pd and Mo stock solutions contained a mixture of 200 pg ml-1 of Pd and 1000 pg ml-1 of Mo.Table 1. HGA-400 atomiser programme for determination of gold in matrix modifiers, whole blood and plasma solutions. Sample volume, 20 pl; pyrolytic graphite coated graphite tube Stage Parameter Dry (a) Wall atomisation- TemperaturePC . . . . . . 130 Ramp/s . . . . . . . . 5 Holds . . . . . . . . 30 Internal gas flow-rate/ml min-1 . . . . . . . . 300 Recorder - Read - . . . . . . . . . . . . . . . . . . ( b ) Platform atomisation- Temperature/"C . . . . . . 150 Ramp/s . . . . . . . . 5 HoWs . . . . . . . . 40 Internal gas flow-rate/ml min-1 . . . . . . . . 300 Recorder - Read - . . . . . . . . . . . . . . . . . . Char 1200 5 30 300 - - 1300 5 30 300 - - Atomise 2400 0 5 0 X X 2400 0 5 0 X X Clean 2700 1 2 300 0 - 2700 1 3 300 0 - All glassware and plastic ware used in this study were soaked in 50% V/V HN03 overnight and rinsed repeatedly with distilled water before use.Procedures Collection of blood samples and separation of plasma Blood samples were provided by Glasgow Royal Infirmary and collected from patients undergoing treatment with sodium aurothiomalate. The heparinised whole blood samples were centrifuged at 3000 rev min-1 for ca. 10 min and the plasma supernatant removed by suction. All specimens were stored in a refrigerator prior to analysis. Direct determination of gold in whole blood and plasma The concentration of Au in the blood stream of patients undergoing chrysotherapy is often in the range 0.1-3 pg ml-1 of Au.The whole blood and plasma samples were therefore diluted 80-fold and 200-fold, respectively, with distilled water before analysis. A 20-pl volume of the diluted specimen was injected into the graphite furnace atomiser together with the same volume of either 1000 pg ml-1 of Ni modifier, or 200 pg ml-1 of Pd - 1000 yg ml-1 of Mo modifier. Calibration was achieved by injecting 20-pl volumes of aqueous Au solutions with the same volume of the appropriate modifier solution. Peak-height and peak-area gold atomic absorption signals were obtained using the atomiser programmes given in Table 1 for wall and platform atomisation. Digestion of samples Some of the samples were analysed by the direct procedure and a method that involved the pre-digestion of the specimens with acid.A 5-ml volume of diluted whole blood or plasma was transferred into a 25-ml beaker and 2 ml of concentrated nitric acid and 0.1 ml of concentrated perchloric acid added to the sample. The beaker was then placed on a hot-plate and heated until fumes of perchloric acid appeared. The digested sample was then cooled to room temperature and several portions of 0.1% WV nitric acid added to complete the decomposition. The solution was then transferred into a 10-ml calibrated flask and diluted to volume with distilled water. On some occasions, 1 ml of the Pd - Mo matrix modifier solution was added to the diluted whole blood or plasma specimen prior to the addition of the acids and digestion. The modifier was added to prevent loss of Au during the digestion procedure.The digested samples were analysed by injecting 20 1.11 of the final solution into the atomiser followed by an equivalent volume of the Pd - Mo matrix modifier solution. The calibration procedure was as described above for the direct analysis of diluted whole blood and plasma samples. Results and Discussion Stability of Gold Solutions The stability of Au solutions at sub-pg ml-1 concentrations is obviously of great importance when analysing biological samples obtained from patients on chrysotherapy . In this study, it was found that the stability of an aqueous 25 pg 1-1 Au solution, not only depended on the type of container used for storage, but also on the matrix modifier added to the solution. As illustrated in Fig.1, Au solutions that contained 0.01% mlV HN03 or the Ni or Pd - Mo modifier were stable over 75 min if glass containers were used for solution storage. Over a longer storage time, the magnitude of the Au atomic absorption signals started to decrease gradually. Gold solu- tions containing Pd alone as the modifier, were not as stable as those with a mixture of Pd and Mo. It is possible that Au - Pd intermetallic compounds adsorb on the interior walls of glassJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 301 containers more rapidly than Au alone. The addition of Mo seemed to prolong the stability of Au solutions containing Pd possibly because the Mo occupied active sites on the glass surface and reduced the severity of Au - Pd adsorption. When the Au solutions were stored in polyethylene containers, the atomic absorption signals decreased rapidly during a period of 60-80 min, as illustrated in Fig.2. Similar instability problems were encountered when PTFE and 0.4 I I 0.3 fn a a 0.2 1 1 I I 1 0 20 40 60 80 Ti me/mi n Fig. 1. Stability of a 25 pg 1-1 gold solution stored in a glass container at room temperature, under different conditions of matrix modifica- tion: A, 0.01% V/VHN03; B, 200 pg ml-1 Pd plus 1000 pg ml-l Mo; C, 200 pg ml-1 Pd; and D, 1000 pg ml-1 Ni Oa4 I -\ A 0 20 40 60 80 Timeirnin Fig. 2. Stability of a 25 pg ml-1 gold solution stored in a polyethylene container at room temperature under different conditions of matrix modification: A, 0.01% VIV HNO,; B, 200 vg ml-1 Pd plus 1000 pg ml-1 Mo; C, 200 pg ml-1 Pd; and D, 1000 pg ml-1 Ni poly(viny1 chloride) containers were used and none of the matrix modifiers were able to prolong stability.It is not clear why plastic containers should be worse than glassware for storage of dilute Au solutions, but clearly the surface phenomena involved are complex and merit further investiga- tion. In all analytical experiments conducted during the study, acid-washed glass containers were used for storage of standard solutions and diluted whole blood and plasma samples. Recovery of Gold Added to Diluted Whole Blood and Plasma Samples In a previous study14 it was shown that the interferences caused by chloride salts on the determination of Au could be essentially eliminated in tube-wall and platform atomisation by addition of either a Ni or Pd - Mo matrix modifier to standard and sample solutions.The modifiers allowed an increase in the char temperature from 800-900 "C to ca. 1200 "C, which assisted the removal of matrix components prior to the atomisation stage. To check the suitability of both modifiers for the analysis of biological samples, a series of recovery experiments were conducted using a 200-fold diluted plasma sample and an 80-fold diluted whole blood sample received from patients on chrysotherapy . The percentage recovery figures obtained for tube-wall and platform atomisa- tion with no matrix modification, Ni modification and Pd - Mo modification are given in Tables 2, 3 and 4, respectively. The atomiser programmes given in Table 1 were applied with the exception that for no matrix modification, the maximum char temperature was 800 "C.When no matrix modification and tube-wall atomisation was used, recovery of gold added to diluted whole blood and plasma was poor (Table 2). However, if platform atomisation was employed, the recoveries were greatly improved. Close to quantitative recovery was achieved by peak-area measure- ments for whole blood, and both peak-height and peak-area platform results were satisfactory for plasma. When nickel matrix modification and tube-wall atomisation were used (Table 3), quantitative recovery was only achieved by peak-area measurements for plasma. With platform atomisation, recoveries were generally improved and satisfac- tory results were obtained by peak-height and peak-area measurements for whole blood, although peak area was better than peak height for the plasma sample.The recovery experiments confirmed a previous conclusion14 that a combi- nation of platform atomisation and peak-area measurement was essential when using nickel as a matrix modifier for gold. As indicated in Table 4, all the recoveries obtained were close to quantitative (91-108%) when a mixture of Pd and Mo Table 2. Recovery of gold using no matrix modifier Recovery, 70 Au in diluted sample solution/ Au added/ Sample 1-' M - ' Whole blood' . . . . 27 0 10 20 30 40 0 10 20 30 40 * Diluted SO-fold; initial whole blood Au concentration, 2.16 pg ml- t Diluted 200-fold; initial plasma Au concentration, 3.00 v g ml-l. $ ND = not determined. Plasmat . . . . . . 15 Tube wall Platform Peak height 2.0 20.5 22.3 NDS ND ND 55.0 52.0 - - 1 Peak area 4.2 20.0 22.0 ND ND ND 54.0 50.0 - - Peak height 67 87 80 83 112 97 91 95 - - Peak area 61 92 94 90 117 101 92 100 - -302 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 Table 3. Recovery of gold using nickel matrix modifier Recovery, % Au in diluted Tube wall sample solution/ Au added/ Sample 1- pg 1-l Peak height Peak area - - Wholeblood* . . . . 27 0 10 67 85 20 68 87 30 62 89 40 63 88 10 63 110 20 73 113 30 73 100 . . . . . . . . . . 40 60 104 - - Plasmat . . . . . . 15 0 * Diluted 80-fold; initial whole blood Au concentration, 2.16 pg ml-l. ? Diluted 200-fold; initial plasma Au concentration, 3.00 pg ml-l. Platform Peak height 108 113 105 112 85 86 90 85 - - ~ ~~ Peak area 91 115 100 104 93 100 106 93 - - Table 4.Recovery of gold using palladium and molybdenum modifier Recovery, % Au in diluted Tube wall sample solution/ Au added/ Sample pg1-l pg 1-1 Peak height Peak area - - Wholeblood* . . . . 27 0 10 92 100 20 96 93 30 92 91 40 96 97 10 104 90 20 100 100 30 98 94 40 100 96 - - Plasma? . . . . . . 15 0 * Diluted 80-fold; initial whole blood Au concentration, 2.16 pg ml-l. t Diluted 200-fold; initial plasma Au concentration, 3.00 pg ml-l. Platform Peak height Peak area - - 93 92 93 100 91 100 91 108 92 90 99 94 105 104 100 97 - - Table 5. Determination of gold (pg ml-1) in plasma and whole blood by three procedures using wall atomisation. Pd - Mo matrix modifier added to sample and aqueous gold standard solutions in all three procedures (i.e., within atomiser tube addition) Direct procedure Acid digestion* Acid digestion with Pd - Mot Sample Peak height Peak area Peak height Peak area Peak height Peak area Plasma$l .. . . . . 3.80 & 0.10 4.00 k 0.06 2.90 k 0.10 2.90 f 0.10 3.90 t- 0.02 3.90 k 0.06 2 . . . . . . 2.30f0.04 2.30f0.06 1.90 k 0.04 2.10 2 0.04 2.30 k 0.02 2.20 k 0.04 3 . . . . . . 2.55f0.10 2.50f0.08 1.98 k 0.02 2.00 k 0.02 2.55 k 0.06 2.53 t 0.04 Whole bloods 1 . . . . 0.86 f 0.02 0.85 f 0.02 0.80 f 0.01 0.88 k 0.02 0.83 k 0.02 0.81 t 0.02 2 . . . . 0.86-+_0.02 0.88f0.02 0.72 f 0.02 0.80 k 0.02 0.88 k 0.04 0.84 t 0.02 * Diluted samples digested with HC104 - HN03. t Diluted samples digested with HC104 - HN03 containing Pd - Mo as a stabiliser. $ Plasma samples diluted 200-fold prior to analysis.0 Whole blood samples diluted 80-fold prior to analysis. was used as a matrix modifier. There was no significant difference between the peak-height and peak-area recoveries by both wall and platform atomisation for the two samples studied. In this respect, it seems that the Pd - Mo modifier has some advantages over Ni. Determination of Gold in Whole Blood and Plasma To confirm the suitability of the Pd - Mo modifier and wall atomisation for the determination of Au in whole blood and plasma, five samples were analysed by the direct procedure (i.e., dilution with distilled water), and after digestion with HC10, - HN03. The diluted and digested samples were analysed using the procedures given under Experimental (i.e., calibration with aqueous Au standards and matrix modifica- tion with Pd - Mo directly inside the atomiser tube).The concentrations obtained by the direct and digestion proce- dures are given in Table 5 . It is apparent that loss of Au occurred during acid digestion, but this could be prevented if 1 ml of the Pd - Mo modifier was added to the digestion mixture prior to heating. As the digested volume was diluted ten-fold prior to analysis, the total mass of Pd and Mo atomised with Au was only 10% higher than in the other two procedures. It is interesting that loss of Au from whole blood was negligible compared with that experienced for plasma samples when Pd - Mo was not added to the digestion mixture. Possibly some component of whole blood helped prevent loss of volatile Au species, but no clear explanation can be given at this time.As indicated in Table 5 , the concentrations obtained by the direct and modified digestion procedures were in good agreement, confirming the suitability of tube-wall atomisation and matrix modification by Pd - Mo for determination of Au in biological samples. The detection limit (DL) for the directJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL procedure when plasma samples were diluted 200-fold was 0.16 pg ml-1 of Au in both peak-height and peak-area measurement modes (20-1.11 injection volume). The detection limit of measurement when whole blood samples were diluted 80-fold was 0.064 pg ml-1 of Au. The DL values were calculated on the basis of 95% confidence limits (20). The aqueous solution detection limit of Au by wall atomisation was 0.4 pg 1-1 of Au in the presence of the Pd - Mo modifier in both the height and area measurement modes.For diluted whole blood or plasma solutions, the detection limit was 0.8 pg 1-1 of Au, due to a factor of two poorer precision in signal measurement. Conclusion The direct determination of Au in diluted whole blood and plasma samples has been achieved by addition of Ni or Pd - Mo matrix modifiers to sample and aqueous standard solutions. With Pd - Mo, peak-height and peak-area measurements were suitable using tube-wall or platform atomisation, but with Ni, peak area and platform atomisation were necessary to avoid matrix interferences. As the whole blood was diluted 80-fold and the plasma 200-fold, no significant background absorption problems were encountered when using Zeeman-effect cor- rection.Glassware was found to be better than plastic ware for storage of solutions. Even in the presence of modifiers deposition and loss of Au occurred over a relatively short period (60-80 min) when the solutions were stored in plastic containers. The methods developed have been applied to determine the concentration of Au in whole blood, plasma and red and clear lysate obtained from a number of patients receiving chryso- therapy and the biochemical conclusions derived from the study will be described in a future paper. The authors thank the Institute of Environmental Chemistry, Chinese Academy of Sciences and the British Council for financial support of this study. The authors are grateful to 1987, VOL.2 303 Perkin-Elmer Inc. for substantial support in terms of the instrumentation used in this project including the P-E 5000 Spectrometer (Sabina Slavin, Nonvalk, CT, USA), the AS-40 Autosampler (Bernhard Welz, Bodenseewerk, FRG) and the Model 3600 AS Data System 10 (Paul Mitchell, Beaconsfield, UK). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. Results Aggett, J., Anal. Chim. Acta, 1973, 63, 473. Maessen, F. J. M. J., Posma, F. D., and Balke, J . , Anal. Chem., 1974,46, 1445. Kamel, H., Brown, D. H., Ottaway, J. M., and Smith, W. E., Analyst, 1976, 101, 790. Kamel, H., Brown, D . H., Ottaway, J. M., and Smith, W. E., Analyst, 1977, 102, 645. Schattenkirchner, M., and Grobenski, Z., At. Absorpt. Newsl., 1977, 16, 84. Ward, R. J., Danpure, C. J . , and Fyfe, D. A., Clin. Chim. Acta, 1977, 81, 87. Dunckley, J. V., and Staynes, F. A . , Ann. Clin. Biochem., 1977, 14, 53. Kamel, H., Brown, D. H., Ottaway, J. M., Smith, W. E., Cottney, J . , and Lewis, A. J . , Arthritis Rheu., 1978, 21, 441. Wawschinek, O . , Microchim. Acta, 1979, 2, 111. Turkall, R. M., and Bianchine, J. R . , Analyst, 1981,106,1096. Sharma, R. P., Ther. Drug Monit., 1982,4,219. Johnsen, A. C., Wibetoe, G., Langmyhr, F. J . , and Aaseth, J., Anal. Chim. Acta, 1982, 135, 243. Rodgers, A . I. A., Brown, D. H., Smith, W. E., Lewis, D., and Capell, H. A., Anal. Proc., 1982, 19, 87. Egila, J., Littlejohn, D., Ottaway, J. M., and Xiao-quan, S . , 1. Anal. At. Spectrom., 1987, 2, 293. Sighinolfi, G. P., and Santos, A. M., Mikrochim. Acta, 1976,2, 33 * Chow, A., and Beamish, F. E . , Talanta, 1963, 10, 883. Chow, A . , Talanta, 1971, 18, 453. Paper J6l112 Received November loth, 1986 Accepted December 9th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200299

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 305 Direct Determination of Cadmium in Urine by Electrothermal Atomisation Atomic Absorption Spectrometry David J. Halls, Murdoch M. Black* and Gordon S. Fell Trace Metals Unit, Department of Biochemistry, Glasgow Royal Infirmary, Castle Street, Glasgow G4 OSF, UK (the late) John M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, Cathedral Street, Glasgow GI IXL, UK Two direct methods for the measurement of cadmium in urine by electrothermal atomisation atomic absorption spectrometry are described and compared. In the first, standards and samples are diluted 1 + 4 with 6% VNnitric acid and measured with a four-stage furnace programme using a L’vov platform in an uncoated graphite tube.Selective volatilisation is used to help separate the cadmium peak from the background absorbance peak. In the second, standards and samples are diluted 1 + 1 with 1% VNnitric acid and analysed off the wall of an uncoated tube with a short three-stage programme using no ashing stage. Both methods show satisfactory recovery and good correlation with a solvent extraction method confirming that interferences have been successfully overcome by the use of nitric acid as a modifier. The platform method has a cycle time of 129 s while the tube-wall method achieves analysis with a cycle time of only 50 s. Keywords: Cadmium in urine determination; atomic absorption spectrometry; electrothermal atomisation; matrix modifier For monitoring human exposure to the toxic element cad- mium, measurements of both blood and urine cadmium are important; blood levels are considered to reflect recent exposure whereas urine concentrations may give an indication of body burden when exposure is low (environmental expo- sure) or current exposure when exposure is high (industrial exposure).1 The determination of cadmium in urine has proved to be difficult and has received much attention in the literature. Because of the volatility of cadmium, only low ashing temperatures can be used, which have been considered in- adequate to remove salts giving rise to background absorption and to remove matrix interferences, leading to most workers using standard additions for determinations.2-6 These prob- lems can be overcome by separation of the cadmium from the matrix using solvent extraction7-10 or electrochemical separa- tion.11 However, to enable direct determination, Ross and Gonzalezl2 showed that the cadmium peak could be separated from the background by “selective volatilisation.” In this, a low atomisation temperature is chosen so that the cadmium is volatilised before the salts that give rise to background absorption.This idea was adopted by Gardiner et a1.6, who used a tube-wall method and more recently by McAughey and Smith4 who used a L’vov platform. In these methods, the slope of calibration varied from urine to urine requiring standard additions calibration. Although the sample handling necessary for these methods is less than for an extraction method, the necessity for standard additions increases the time of analysis and, in terms of throughput of analyses, offers no advantage over extraction methods.Matrix modifiers have been investigated to increase the ashing temperature. Edigerl3 originally proposed the use of diammonium hydrogen phosphate and this or ammonium dihydrogen phosphate has featured in a number of methods.14-16 Pruszkowska et aZ.14 used a mixture of diammo- nium hydrogen phosphate and nitric acid, which allowed the * Present address: Robens Institute, University of Surrey, Guildford, Surrey GU2 5XH, UK. use of an ashing temperature of 700 “C. Sources of modifier free of cadmium are difficult to find4.14 so that purification by solvent extraction is often necessary, An alternative approach for a matrix modifier is the use of nitric acid,2J?5J7J8 which converts metal halides into nitrates which show much less molecular absorption.Feitsma et al. 18 compared various matrix modifiers and found that 2% mlV nitric acid was the best giving lower background and better sensitivity than the others. Even with matrix modifiers, some workers2J75 find it necessary to use standard additions for calibration. Direct measurement against aqueous standards has been reported by Subramanian et aZ.15 and Pruszkowska et aZ.14 using diammo- nium hydrogen phosphate and nitric acid as a matrix modifier and by Van Diejck and Herber17 using nitric acid. For a routine method, direct measurement against simple aqueous standards is preferable as it allows a greater throughput of analyses. This paper describes two methods allowing direct determination that were developed in our laboratories as a result of two separate projects, one a study on cadmium metabolism through an animal model19 and the other on achieving faster rates of analysis by electrothermal atomisation atomic absorption spectrometry.20.21 Experimental Instrumentation Measurements were made on a system consisting of a Perkin-Elmer 2280 spectrometer, HGA-500 graphite furnace, AS1 autosampler and a Model 56 recorder.Argon was used as the purge gas. Reagents Standards were prepared from a BDH Spectrosol 1000 mg 1-1 solution of cadmium as cadmium nitrate. Nitric acid was of BDH Aristar grade, other chemicals tried as matrix modifiers were of AnalaR grade, ammonium tetramethylenedithiocar- bamate (ammonium pyrrolidinedithiocarbamate, APDC) was of analytical-reagent grade and isobutyl methyl ketone306 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 m ~~~ ~ Table 1. Instrumental conditions for the three methods used General- Lampcurrent . . . . 4mA Slit width . . . . . . 0.7 nm Wavelength . . . . 228.8nm Background correction On Extraction method- Injection volume . . 20 pl Uncoated graphite tube Operating mode . . Peak height Furnace programme: Ramp Hold Step Temperature/"C time/s time/s 1. Dry . . . . 120 1 15 2. Ash . . . . 350 1 10 3. Atomise . . 1900 1 5* 4. Clean . . 2700 1 3 Platform method- Injectionvolume . . 10 pl L'vov pyrolytic platform in uncoated graphite tube Operatingmode . . Peakarea Integrationtime . . 8s Furnace programme: Ramp Hold Step Temperature/"C time/s time/s 1.Dry . . . . 160 4 40 2. Ash . . . . 500 2 15 3. Atomise . . 1200 2 llt 4. Clean . . 2700 1 4 5. Cool . . . . 20 1 20 Wall method- Injection volume . . 20 pl Uncoated graphite tube Operating mode . . Peak height Furnace programme: Ramp Hold Step Temperature/"C time/s time/s 1. Dry . . . . 140 1 7 2. Atomise . . 2400 1 6$ 3. Clean . . 2700 1 5 * -2 s Autozero, -2 s record and 10 ml min-1 internal flow-rate. t 0 s read, 0 s record and 0 ml min-1 internal flow-rate. $30 ml min-1 internal gas flow-rate, -2 s autozero, -2 s record. (4-methylpentan-2-one) of Spectrosol grade. Water was first de-ionised and then distilled. Control of Contamination Samples were stored in 30-ml plastic universal containers (Sterilin, Teddington, UK) which were shown to be free of cadmium; standards were also prepared in these containers.Before use, all autosampler cups and glass test-tubes were filled with 20% V/V nitric acid and allowed to stand for ca. 30 min before washing with distilled water. Only clear plastic tips were used for micropipettes; yellow tips contain a cadmium pigment. The tips were cleaned before use by pipetting 20% V/V nitric acid twice and then distilled water twice. Collection of Urine Specimens Urine collections (24-h specimens) were made in polyethylene bottles. On receipt in the laboratory, an aliquot was placed in a 30-ml plastic universal container and acidified to 1% V/V 8 100 A 4- !! B with nitric acid. For occupational monitoring, random sam- ples are preferable; these can be collected directly in the universal containers.Analytical Procedures Extraction method To 1 ml of urine or standard in a 10 x 75 mm glass tube, 0.5 ml of 10 g 1-1 APDC is added and the solution mixed. Isobutyl methyl ketone (0.5 ml) is then added, the tube capped and shaken with inversion for ca. 30 s. The solvent layer is then separated with a micropipette and transferred into a poly- propylene sample cup. Some samples may need centrifugation for up to 5 min to improve separation between the layers. The extracts are then analysed using the instrumental conditions shown in Table 1. Experience has shown that it is preferable to separate and analyse the extracts without appreciable delay. Platform method Standards and samples are diluted 1 + 4 with 6% VIV nitric acid using the instrumental conditions given in Table 1.Peak-area measurements were taken from the digital display of the spectrometer. Tube-wall method Standards of 0 , 2 , 4 , 6 , 8 and 10 pg 1-1 are prepared in 1% VIV nitric acid. Standards and samples are diluted 1 + 1 with 1% V/V nitric acid and analysed using the conditions given in Table 1. Peak-height measurements were made from the chart recording. Results and Discussion Stability of Samples The importance of acidifying urine samples for storage is seen from Fig. 1. The non-acidified samples either refrigerated at 4 "C or frozen at -20 "C showed decreases in cadmium concentration over the five-week period presumably due to adsorption on the container or on particulate matter in the sample. The same sample acidified to 0.01 M with nitric acid, stored at 4 "C, showed no change in concentration over the same period.L'vov Platform Method Effect of matrix modifiers A number of matrix modifiers were briefly examined for use in the determination of urine cadmium: sodium dihydrogen orthophosphate, EDTA, ammonium nitrate, APDC and nitric acid. With sodium dihydrogen orthophosphate, blank levels of cadmium were a problem necessitating a prior purification by extraction with APDC into isobutyl methyl ketone. Similar problems were reported by Pruszkowska et al.14 and McAughey and Smith.4 Of the matrix modifiersJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 1 1 1 307 I , 1 o.2t \, Table 2. Recovery of cadmium added to urines using the L'vov platform method. Each sample analysed seven times; the mean and standard deviation are reported Urine 1 Parameter Cdaddedpgl-1 .. . . 0 2 4 6 Concentration found, mean f SD/pgl-' .2.1 k 0.2 4.1 k 0.1 6.2 f 0.1 8.3 f 0.2 Recovery,% . . . . - 99+4 1 0 3 f 4 102+4 Urine 2 Cdadded/pgl-l . . . . 0 2 4 6 Concentration found, mean f SD/pg1-1 . . 2.9 k 0.2 4.9 f 0.1 6.9 t- 0.2 9.2 f 0.2 Recovery,% . . . . - 10324 101+6 1 0 5 2 4 tried, nitric acid was found to be the most useful. It very effectively reduced the background absorbance (Fig. 2) and helped in controlling any interference. Moreover, it can be obtained very pure as Aristar grade, giving negligible blank values. Optimisation of procedure Gardiner et a1.6 described a standard additions procedure for urine cadmium which involved selective volatilisation from a standard graphite tube.The intention in the present method was to use the concept of selective volatilisation to volatilise the cadmium at a low atomisation temperature to resolve the cadmium peak from the background absorption caused by the matrix. Nitric acid was used to reduce the background to help this resolution. An atomisation temperature of 1200 "C was chosen. Above this temperature, cadmium and background peaks were not properly resolved. Below this temperature, a sharp decrease in sensitivity occurred due to incomplete volatilisation of cadmium. Measurements were made by peak area, which is more suited to the broader peaks obtained at this lower atomisation temperature. The complete pro- gramme developed as a result of normal optimisation pro- cedures is shown in Table 1.As shown in Fig. 2, the non-specific absorbance of a urine, measured using a neighbouring line, decreases with increasing nitric acid concentration up to 6% V/V at which it levels off. All samples and standards were therefore diluted with 6% V/V nitric acid. Further reduction in matrix concentration was achieved by using a five-fold dilution of the sample. Analytical performance Calibration is linear up to 8 pg 1-1 of cadmium. The sensitivity and detection limit for the method are 0.14 and 0.07 pg 1-1, respectively. 1 2 3 4 5 6 7 Extraction method 1 (b) + r 12 10 5 8 - - U E 3 6 - g 4 - d, 9 - 2 - I 1 1 I 0 2 4 6 8 1 0 1 2 Extraction method Fig. 3. Correlation between results of measurement of Cd in urine in pg 1-l obtained by: (a) platform method; and ( b ) tube-wall method with results obtained by the extraction method. Conditions for the methods are iven in Table 1.Correlation coefficients: (a) 0.964 (n = 15); and (b) 8.997 (n = 21). E uations of lines of best fit using least squares regression: (a y = 0.9% - 0.04; and ( b ) y = 0.97~ - 0.02. Standard deviations of)slope: (a) 0.076; and ( b ) 0.018. Lines shown on graph are those of best fit The accuracy of the method was assessed by standard additions plots, recovery experiments and comparison with the extraction procedure. Slopes of linear calibration graphs for standards, standards added to a urine with low non-specific absorbance and standards added to a urine with high non-specific absorbance were 0.032, 0.030 and 0.034 A s 1 pg-1, respectively.The similarity indicates the absence of any significant matrix effect. Table 2 shows recovery data for additions of differing amounts of cadmium to two urine samples. Good recoveries are obtained confirming the absence of matrix effects. Fig. 3(a) shows the good agreement obtained between this method and the extraction method. Within-batch precision of better than 5% RSD was obtained in the range 2-10 pg 1-1 [Fig 4(a)] and better than 2% RSD between 3 and 8 pg 1-1. Between-batch precision was The results show that using the concept of selective volatilisation and matrix modification with nitric acid on the L'vov platform, cadmium can be determined accurately against simple standards. Standard additions as proposed by other workers2,395 is not necessary.1520% RSD. Tube-wall Method The objective of this method was to achieve accurate analysis with a rapid programme, eliminating as much extraneous time as possible. An uncoated tube was chosen because it had been shown that rapid drying could be achieved with this tube.20 Effect of the ashing stage The ashing stage is conventionally regarded as important to remove compounds which give rise to background absorption and matrix interferences. For the volatile elements, it is unlikely that an ashing temperature can be used that is sufficiently high to have much effect. The effect of ashing308 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 I I I I I 0 2 4 6 8 1 0 Concentration/pg I-' Fig. 4. Within-batch precision as a function of concentration for: A, platform method; and B, the tube-wall method.Each point is calculated as the relative standard deviation of the measurement of 7 aliquots of the sample 0.10 a C m e s (0 Y 3 0.05 n 73 4- 2 0 V 0 I I 1 1 I I 100 200 300 400 500 600 As hing temperatu re/"C 0.3 a 0.2 : ft s I] m U 3 0.1 g Y m m 0 Fig. 5. For the tube-wall method, effect of ashing temperature on peak height (with background correction) for an acidified urine diluted 1 + 1 with 1% nitric acid (A) and on the background absorption for the first and second peak (B and C, respectively). For B and C, a urine with very low Cd concentration was used so that total absorbance was almost entirely background. The same urine, but spiked with 15 pg 1-1 Cd, gave curve A. An ashing step of 5-s ramp time and 10-s hold time was used at the indicated temperature.The rest of the programme is as in Table 1 except that the atomisation temperature was 1900 "C. Absorbance values found with the ash stage omitted are also shown temperature on background absorption and signal height for a urine sample diluted 1 + 1 with 1 YO V/V nitric acid is shown in Fig. 5 . The background signal consists of two temporally separated peaks. Above 350 "C, the cadmium signal (line A) decreases due to loss of cadmium in the ashing stage. At 350 "C, the maximum usable ashing temperature in this instance, the two background peaks (lines B and C) are somewhat reduced (ca. 27%). The ashing stage causes a negligible effect on matrix interferences as shown by only a small difference in signal when the ashing stage is omitted (Fig.5). Because of this and the only minor improvement in background with an ashing stage, it was felt that the ashing stage could be omitted. Background in this instance (with 1% V/V nitric acid as diluent) was below 0.3 absorbance which is within the capabilities of the background-correction system. 0.6 a m 0.4 9 In I] m -0 3 CD Y 2 om 0.2 m 0 0.15, I 0.10 Q) c (11 e 2 a Y m a 0.05 0 c I 1 1 2 Nitric acid concentration, % V N 150 100 8 5 s > a 50 0 Fig. 6. Effect of nitric acid, using the tube-wall method, on the peak absorbance (with background correction) for a 3 pg 1-1 standard solution of Cd (A) and for a urine diluted 1 + 1 with a 6 pg 1-1 Cd standard solution (B). Recovery of added Cd is shown in C, and D shows the background absorbance of the highest peak.Furnace programme used is as in Table 1. Nitric acid concentration shown is the final concentration in the prepared sample Table 3. Effect of ramp time in the atomisation stage on recovery of cadmium added to a urine sample using the tube-wall method. Programme as in Table 1 but with an atomisation temperature of 1900 "C and ramp time as indicated Ramptimek . . 1 2 5 Recovery, YO . . 97 90 78 Effect of nitric acid Fig. 6 shows the effect of nitric acid on the signal for cadmium in urine for a programme without an ashing stage. Nitric acid markedly reduces background (line D) as observed also with the platform method. In this instance, most of the reduction is complete at a final concentration of 1% V/V. As the acid concentration increases, the peak absorbance of a standard solution of cadmium (line A) increases only slightly; the lower absorbance at zero acid concentration is probably due to loss of cadmium by adsorption on the container surface.For cadmium in urine, however (line B), the absorbance decreases with increasing acid concentration leading to a decreased recovery (line C). Optimum recovery (100%) is found at a concentration of 1% VW, which fortunately coincides with the point where background reduction becomes constant (line D). Further work was carried out with a nitric acid concentra- tion of 1% V/V. In the final method, all urines are stabilised by acidifying to 1% V/V with nitric acid and are diluted for analysis 1 + 1 with 1 YO V/V nitric acid. This dilution and the use of some internal gas flow (30 ml min-1) helps reduce background further.Effect of atomisation temperature and ramp rate Table 3 shows that as the ramp time is decreased from 5 to 1 s, the measured recovery of added cadmium improves reaching 97% with a 1-s ramp time, which was used in all further work. With an increase in atomisation temperature, the height of the cadmium peak from a urine sample increased up to a temperature of 2000 "C reaching a plateau extending up to 2400 "C beyond which it started to decrease again. Previous work at an atomisation temperature of 1900 "C had shown thatJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 309 the Royal Infirmary for urine cadmium measurement. This always caused problems whenever a change in operator occurred as it seemed to require a number of attempts for the operator to acquire sufficient expertise to obtain successful results.The tube-wall method which has replaced it has removed this problem. The value of nitric acid as a matrix modifier has again been Table 4. Performance of the tube-wall method Recovery- For 15 urine samples with addition of 5 pg 1-1 cadmium 100 k 5% Range . . . . . . . . . . . . . . 91-110% . . . . . . . . . . . . Detection limit- For measurement of urine (2 x SD) . . Precision- Precision of measurement at 2.9 yg 1-1 Precision of measurement at 7.9 pg 1-1 For entire method: Within-bathprecision . . . . . . Between-batch precision (n = 17) on Lanonorm 1 urine . . . . . . . . Quality control- Measured value for Lanonorm 1 urine (mean + SD) .. . . . . . . Assigned value . . . . . . . . . . Confidence range . . . . . . . . . . 0.13pg1-1 . . 1.7% . . 1.0% . . SeeFig. 4(b) . . 9.9% . . . . 2.91.181-1 . . 2.4-3.4 pgl-1 3.0 L 0.3 pgl-1 a carry-over effect was seen when a clean stage of 3 s at 2700 "C was used. This effect, seen as a progressive decrease in peak height of a sample low in cadmium immediately following a sample high in cadmium, was decreased by increasing the time of the clean stage. It was felt that, in addition, the atomisation temperature should be as high as possible (2400 "C) to reduce the requirements of the clean stage. Although this was so, the extended clean stage (6 s at 2700 "C) improved the precision and so was retained. Analytical performance Table 4 shows that acceptable values for recoveries are obtained.The precision of measurement of a prepared sample is ca. 1?'0 RSD over the range 5-10 yg 1-1 and better than 2% over the range 2-10 yg 1-1. This indicates that despite the short programme, reproducibility is not compromised. Fig. 4(b) shows the within-batch precision of the entire method. Here precision is again better than 2% RSD over the range 2-10 yg 1-1 with precision rapidly becoming poorer at concentra- tions less than 2 yg 1-1. The performance is adequate for measuring occupational exposures which generally result in concentrations in the range 2-10 yg 1-1. Between-batch precision in routine use on a Lanonorm quality-control material has been 9.9% RSD. Fig. 3(b) shows results obtained by the tube-wall direct method compared with the extraction method.Good correla- tion is obtained with a correlation coefficient of 0.997. This confirms the lack of matrix effect with the direct method. Conclusions Both methods give acceptable performance for the monitoring of occupational exposure to cadmium. Accuracy is good and is achieved without the necessity for standard additions. The use of matrix modifiers that require prior purification is avoided. Both give results which agree well with the solvent extraction method but are more straightforward avoiding the sample preparation necessary with the solvent extraction step. Until recently, the solvent extraction method was used routinely in demonstrated. It has previously been shown by u s to be effective in removing interferences in the determination of aluminium in dialysate fluids21 and lead in water.22 The first method illustrates the conventional modern approach to electrothermal atomisation AAS, a platform with a matrix modifier using a four-step furnace programme.The same level of performance is achieved in the second method which dispenses with the platform, uses no ashing stage and uses a fast drying stage. The latter method has a cycle time of 50 s (including 29-s autosampling time) whereas the platform method takes 129 s. This is however not slow for a platform method; the method of Pruszkowska et aZ.14 takes 169 s, for example. It is clear that the platform and ashing stage can be omitted from this method when appropriate. Urine cadmium is therefore another example of a determination that can be made without an ashing stage.This reinforces the statement made earlier20 that it is worthwhile investigating for all determinations whether the ashing stage can be successfully omitted. References 1. Lauwerys, R. R., Buchet, J. P., andRoels, H. Int. Arch Occup. Environ. Health, 1976, 36, 275. 2. Vesterberg, O., and Wrangskogh, K., Clin. Chem., 1983, 29, 477. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22, Mohl, C., Ostapczuk, P., and Stoeppler, M., in Welz, B., Editor, "Fortschritte in der Atomspektrometrischen Spuren- analytik," Volume 1, Verlag, Weinheim, 1984, p. 317. McAughey, J. J., and Smith, N. J., Anal. Chim. Acta, 1984, 156, 129. Dungs, K., and Neidhart, B., Analyst, 1984, 109, 877. Gardiner, P. E., Ottaway, J. M., andFell, G. S., Talanta, 1979, 26, 841. Sperling, K.-R., and Bahr, B., Z . Anal. Chem., 1980,301,29. Sperling, K.-R., and Bahr, B., 2. Anal. Chem., 1980,301,31. Allah, P., and Mauras, Y . , Clin. Chim. Acta, 1979, 91, 41. Stoeppler, M., and Brandt, K., Z . Anal. Chem., 1980, 300, 372. Lund, W., Larsen, B. V., and Gundersen, N., Anal. Chim. Acta, 1976, 81, 319. Ross, R. T., and Gonzalez, J. G., Anal. Chim. Acta, 1974,70, 443. Ediger, R. D., At. Absorpt. Newsl., 1975, 14, 127. Pruszkowska, E., Carnrick, G. R., and Slavin, W., Clin. Chem., 1983, 29,477. Subramanian, K. S., Meranger, J. C., and MacKeen, J. E., Anal. Chem., 1983, 55, 1064. Slavin, W., Manning, D. C., Carnrick, G., and Pruszkowska, E., Spectrochim. Acta, Part B , 1983, 35, 1157. Van Deijck, W., and Herber, R. F. M., Clin. Chim. Acta, 1983, 128, 379. Feitsma, K. G., Franke, J. P., and de Zeeuw, R. A., Analyst, 1984, 109, 789. Black, M. M., Ph. D. Thesis, University of Strathclyde, 1984. Halls, D. J., Analyst, 1984, 109, 1081. Halls, D. J., and Fell, G. S., Analyst, 1985, 110, 243. Sthapit, P. R., Ottaway, J. M., Halls, D. J., and Fell, G. S., Anal. Chim. Acta, 1984, 165, 121. Paper J6l94 Received October 7th, 1986 Accepted November 13th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200305

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 311 Determination of Phosphorus by Graphite Furnace Atomic Absorption Spectrometry Part 3.* Analysis of Biological Reference Materials Adilson J. Curtius,t Gerhard Schlemmer and Bernhard WelzS Department of Applied Research, Bodenseewerk Perkin-Elmer & Co GmbH, 0-7770 Uberlingen, FRG Palladium nitrate mixed with calcium nitrate was found to be the best modifier for the determination of phosphorus in plant and animal tissue samples. The analytical graph is linear only up to 0.06 A s and the analyte addition technique is therefore impractical. The results were calculated from a linear regression equation established using matrix-free references over a limited concentration range after the absence of interferences has been verified.For some samples with high iron content Zeeman-effect background correction was required. Characteristic masses of 3 and 5 ng were found for continuum-source and Zeeman-effect background correction, respectively. The detection limit, however, was better with Zeeman- effect (5 ng) than with continuum-source background correction (8 ng). Keywords: Phosphorus determination; graphite furnace atomic absorption spectrometry; palladium modifier; Zeeman-effect background correction; biological reference materials Graphite furnace atomic absorption spectrometric (GFAAS) determination of phosphorus using the non-resonance line doublet at 213.5 - 213.6 nm was first proposed by L’vov and Khartsyzov.1 The procedure was not widely accepted, however, and it took almost a decade until a detailed study of the analytical parameters for the determination of this element was published.* The problems associated with the determination of phosphorus using GFAAS were carefully investigated by Persson and Frechs who concluded that reproducible results can be obtained only if the heating rate and the final temperature of the furnace as well as the atmosphere inside the graphite tube can be controlled during the course of the determination. The main problems in the determination of this element are that phosphorus tends to form gaseous compounds of high thermostability over a wide temperature range, and that for its determination at the non-resonance lines at 213.5 - 213.6 nm phosphorus must not only be atomised but also excited so that the sensitivity becomes very temperature dependent.The literature on the determination of phosphorus in biological materials using this technique is scarce and shows little consistency. Among the materials that have been analysed are oils and fat~,4~5 plant tissue,6 fish7 and starch samples8 as well as several NBS standard reference material~.6,7>~ The procedures range from those in which no modifier at all is used4 through those with lanthanum579 or nickels as the modifier to those calling for silicon carbide or zirconium carbide coated tubes7 or the addition of zirconium solution to the sample in a zirconium treated graphite tube.6 A number of further modifiers, including palladium, have been listed for phosphoruslo but their applicability to real samples was not investigated.Recently it was shown that the determination of phosphorus without the addition of a modifier is not of practical interest because most of the analyte is lost prior to the atomisation stage.11 The modifiers proposed for the determination of phosphorus were, for the most part, found to be unsatisfactory for routine application because the sensitivity obtained was not sufficiently high or not stable in the short or long term.12 Lanthanum also caused severe corrosion of graphite tubes and platforms when applied at high concentrations. 13 Pallad- ium, which has been shown to be a widely applicable modifier for a variety of analyte elements,l4 provided sufficibnt sensitivity and thermal stability, and gave the best reproduci- bility and long-term stability of phosphorus signals,l2 at least for matrix-free solutions.In this work we investigate the extent to which the experience collected with aqueous reference solutions can be transferred to the determination of phosphorus in plant and animal tissue samples. Also investigated are possible spectral and non-spectral interferences that may occur in biological materials as well as the linearity of the analytical graph for phosphorus. A recommended procedure is given for the routine determination of phosphorus in plant and animal tissue samples. Experimental Instrumentation A Perkin-Elmer Model Zeeman 3030 atomic absorption spectrometer with Zeeman-effect background correction, an HGA-600 graphite furnace and an electrodeless discharge lamp for phosphorus operated at 8 W were used for most of the experiments.Some of the measurements were carried out on a Perkin-Elmer Model 1100 atomic absorption spec- trometer with deuterium-arc background correction, equipped with an HGA-400 graphite furnace. The wavelength was set to 213.6 nm, and all instrumental parameters were chosen according to the manufacturer’s recommendations. 15 The temperature programme used for graphite furnace determination is given in Table 1. Sample injection was done automatically using Perkin-Elmer AS-40 and AS-60 auto- samplers. Pyrolytic graphite coated tubes (Perkin-Elmer Part No. BOO91 504) with L’vov platforms made of pyrolytic graphite (Perkin-Elmer Part No. B 0109 324) were used throughout this work. All data including the time-resolved signals were plotted on a Perkin-Elmer PR-100 printer.Signal evaluation was based exclusively on integrated absorbance values. * For Part 2 of this series, see reference i2. t Present address: Departamento de Quimica, Pontificia Universi- $ To whom correspondence should be addressed. dade Cat6lica do Rio de Janeiro, Rio de Janeiro, Brazil. Chemicals , All chemicals were of analytical-reagent grade or higher purity. The following stock standard solutions were used.312 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 Table 1. Recommended temperature programme for the determination of phosphorus Step number 1 2 3 4 5 6 Time/s Furnace temperaturel'c Ramp Hold 90 1 10 120 15 10 1350 1 30 2650 0 5 2700 1 4 20 1 8 Internal gas flow/ ml min- * Read 300 - 300 - 300 - 300 - 300 - 0 * Phosphorus, 5000 mg 1-1.Prepared from dibasic am- monium phosphate (Merck No. 1207). Lanthanum, 50000 mg 1-1. Prepared from lanthanum nitrate hexahydrate (Merck No. 5326). Palladium, 10 000 mg 1-1. Prepared by dissolving palladium metal powder (Merck No. 12486) in concentrated nitric acid. Calcium, 5000 mg 1-1. Prepared from calcium nitrate tetrahydrate (Merck No. 2121). Boron, 5000 mg 1-1. Titrisol solution (Merck No. 9923). Nitric acid, 65% m/V (Merck No. 456). All solutions were made and diluted with 0.2% V/V nitric acid unless otherwise specified. Samples and Sample Decomposition The following certified standard reference materials were used: milk powder (IAEA A-ll)? bovine liver (NBS No. 1577), oyster tissue (NBS No.1566), spinach (NBS No. 1570), orchard leaves (NBS No. 1571), tomato leaves (NBS No. 1573) and pine needles (NBS No. 1575). Four portions of between 0.2000 and 0.5000 g of each material were weighed and transferred into the PTFE beaker of a Perkin-Elmer Autoclave-3. Five ml of de-ionised water and 5 ml of concentrated nitric acid were added, the autoclave closed, heated to 155 k 5 "C on a hot-plate and kept at this temperature for 90 min. After cooling to room temperature the autoclave was opened, the clear solution transferred quantitatively into a 25-ml calibrated flask and diluted to volume with de-ionised water. A small amount of a white precipitate was found after the decomposition of the spinach sample which, however, was not further investigated. Before the analysis, the milk powder, bovine liver and oyster tissue solutions were further diluted to 250 ml, all other solutions to 100 ml.The plant materials were dried at 85 "C for at least 4 h before weighing, the dry weight of all other samples was determined on the basis of separate portions which were kept at 85 "C for 24 h. The results for these samples were obtained on a wet-weight basis and corrected for the weight loss upon drying. For all determinations, 10 pl of sample or reference solution and 10 pl of modifier solution were pipetted sequentially on to the L'vov platform and analysed using the furnace programme given in Table 1. Results and Discussion Analytical Graph It was found early in our investigations that the analytical graph for phosphorus, even with the blank absorbance subtracted, seemingly did not intersect the integrated absorb- ance axis at zero absorbance but at some positive value. This is shown in Fig.1 for the concentration range 10-100 mg 1-1 (mass range 100-1000 ng) and the palladium and calcium nitrates mixed modifier (20 pg Pd + 5 pg Ca) used for most determinations in this study. The difference in sensitivity is due to the Zeeman-effect background correction used in one 0.5 I i I A A 0.4 0.3 0.2 0.1 0 1 I I I I 0 20 40 60 80 100 Phosphorus concentration/mg 1-1 Fig. 1. Analytical graphs for phosphorus using the palladium and calcium nitrates mixed modifier: A, continuum-source background correction; and B , Zeeman-effect background correction 0.03 v, 0.025 9 8 6 0.02 e 4 0.015 m U 5 0.01 ol - 2 0.005 0 0.4 0.8 1.2 1.6 2.0 0 0:4 0.8 1.2 1.6 2.0 Phosphorus concentration/mg I-' Fig.2. Linear re ression for low phosphorus concentrations using the palladium a n f calcium nitrates mixed modifier. Average and standard deviation for 10 determinations, each: (a) continuum-source background correction; and ( b ) Zeeman-effect background correction of the instruments and the incomplete separation of the a-components from the emission profile. 16 Careful investigation of the concentration range 0.4-2.0 mg 1-1 (mass range 4-20 ng), however, showed that the non-zero integrated absorbance axis intercept is an artifact. The average integrated absorbance and the range for ten determinations each of five reference solutions close to the limit of detection is depicted in Fig.2 for both instruments. The linear regression equations are A = -0.0007 + 0.0100~ ( R = 0.9906) . . (1) for the instrument with Zeeman-effect background correction and A = 0.0014 + 0.0108~ (R = 0.9923) . . (2) for the instrument with continuum-source background correc- tion, respectively. The integrated absorbance intercept calcu- lated for both instruments is thus well within the noise inherent in determinations at this level.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 313 0.20 In < 8 f $ 2 0.10 E 5 0.15 II TI En c 4-l - 0.05 From equations (1) and (2) a characteristic mass of 5 ng (0.0044 A s)-1 can be calculated for the Zeeman-effect instrument and of 3 ng (0.0044 A s)-1 for the continuum- source background corrected instrument.The respective detection limits were 5 ng and 8 ng. The latter detection limit may be improved slightly by reducing the integration time from 5 s to, for example, 3 s. The results, however, show clearly that the lower sensitivity of the Zeeman-effect instru- ment is more than compensated by the substantially better signal to noise ratio brought about by the background correction technique. Using a different graphite tube, the calculation was repeated for the Zeeman-effect instrument based on a wider concentration range of 2-32 mg 1-1 (mass range 20-320 ng). The analytical graph is linear only up to 0.06 A s, correspond- ing to a concentration of 8 mg 1-1 or a mass of 80 ng of P, with the linear regression equation A = -0.0004 + 0.0078~ ( R = 0.9993) . . (3) From this equation a characteristic mass of 5.6 ng (0.0044 A s)-1 can be calculated and the detection limit is 7.7 ng.This means that the analytical graph begins to deviate from Beer’s law at an analyte concentration which is only about ten times the detection limit. This pronounced non-linearity is probably due to the use of a line doublet for the determination of phosphorus, i.e., that more than one transition falls within the spectral band pass of the monochromator. 17 Such a restricted linear range is very much limiting analytically because it makes the analyte additions technique impractical to use. While this technique has some inherent sources of error18 it is nevertheless used frequently in GFAAS to correct for residual interferences. Non-linear analytical graphs, however, prohibit its use for calibration so that other ways have to be found.If the entire concentration range of the previously men- tioned analytical graph of 2-32 mg 1-1 (mass range 20-320 ng) is considered, the linear regression equation is A = 0.0104 + 0.0057~ (R = 0.9907) . . (4) A significant positive integrated absorbance intercept is found and the correlation (R = 0.9907) is unsatisfactory. If, however, a smaller range is chosen, for example the concen- tration range 4-20 mg 1-1 (mass range 40-200 ng), the linear regression equation is A = 0.009 + 0.0063~ ( R = 0.9981) . . (5) - - . / 0 - 0.25 I L I /bl A 0 1 1 I I I I J 5 10 20 30 40 50 Phosphorus concentration/mg 1-1 Fig. 3. Analytical graphs for phosphorus: A, calcium nitrate modifier ( 5 pg Ca); B, palladium nitrate modifier (20 pg Pd); and C, palladium and calcium nitrates mixed modifier (20 pg Pd + 5 pg Ca) This shows that a significant positive integrated absorbance intercept is found due to the flattening of the curve, but the correlation coefficient (R = 0.9981) is quite satisfactory.This means that the deviation from linearity of the analytical graph in this restricted range is small so that linear regression could be used for signal evaluation. In the following work, five phosphorus reference solutions covering the ranges 2-12 mg 1-1 or 4-20 mg 1-1 as required were measured three times each before and after the sample solutions. The average of the six readings was used to calculate the linear regression equation of the analytical graphs which was then used to calculate the phosphorus content in the sample solution.Alternatively, a non-linear calibration could be carried out, fitting the analytical graph to a two or three coefficient equation19 as provided for the instrument used, in order to extend the working range for the determinations. Modifiers and Interferences If the analytical results are to be calculated from the integrated absorbance readings using the linear regression equation obtained from matrix-free reference solutions, all possible interferences have to be under good control. The NBS standard reference material orchard leaves (certified value 0.21 k 0.01% P) was used to investigate the influence of matrix constituents on the determination of phosphorus. It is known that calcium enhances the phosphorus sig- na1.2JOJ2920 This effect is in the first approximation additive to the enhancing effect of modifiers such as pallad- ium, as depicted in Fig, 3.A positive interference can therefore be expected from the calcium contained in biological materials unless it is accounted for in the reference solutions. A phosphorus content of 0.227+0.004% was found in orchard leaves if only palladium nitrate (20 pg Pd) was used as a modifier, whereas a result of 0.207 f 0.006% was obtained with the palladium and calcium nitrates (20 pg Pd + 5 pg Ca) mixed modifier. The mass of calcium added with the modifier solution to the reference and sample solution is five times higher than the highest calcium content found in the samples under investigation. This excess should be sufficient to ensure 0.20 0.15 rn 5 f 0 C m B 0.10 -a c I 0) c - 0.05 0 0.0 O0 OO U 50 000 0 1 10 20 30 40 No.of determinations Fig. 4. Influence of boron on the determination of 200 ng of phosphorus (10 p1, 20 mg 1-1) usin the lanthanum nitrate (10 pg of La) modifier. A sequence of sets ofthree determinations were made with and without the addition of boron. The mass of boron was increased from 0.5 to 50000 ng as indicated. Phosphorus and lanthanum modifier only; and 0 boron added314 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 Table 2. Determination of phosphorus in standard reference materials (SRM) using continuum-source (D2 arc) and Zeeman-effect (Zeeman) background correction, respectively Certified content Phosphorus content found, Yo SRM Fe/pg g-1 P, Yo D2 arc Zeeman Milk powder .. . . . . 3.65 0.910 k 0.102 0.906 0.910 Spinach . . . . . . . . 550 0.55 k 0.02 0.518 0.555 Oystertissue . . . . . . 195 (0.81)* 0.682 0.703 Bovineliver . . . . . . 268 (l.l)* 1.032 1.001 * Not certified. Table 3. Phosphorus content found in standard reference materials (SRM) using the proposed GFAAS procedure Certified Phosphorus content phosphorus found*, SRM content, o/o YO Milk powder . . . . . . 0.910 k 0.102 0.910 k 0.024 (2.6) Bovineliver . . . . . . (1.l)t 1.001 f 0.019 (1.9) Oystertissue . . . . (0.8l)t 0.703 f 0.003 (0.4) Spinach . . . . . . 0.55k0.02 0.555f0.007(1.3) Orchard leaves . . . . 0.21 f 0.01 0.207 f 0.006 (2.9) Tomato leaves . . . . 0.34 k 0.02 0.343 k 0.006 (1.7) Pine needles .. . . . . 0.12 f 0.02 0.123 f 0.010 (8.1) sample decompositions. * Average and standard deviation (YO RSD) of four independent t Not certified. that the calcium content in the original sample becomes insignificant. An influence of the nitric acid concentration on the integrated absorbance signal for phosphorus was also found. If the original decomposition solution, which was diluted to only 25 ml, was analysed directly, an average phosphorus content of 0.182 +_ 0.010% was found for the orchard leaves. This can be compensated only in part if reference solutions are prepared in 20% V/V nitric acid, Such high nitric acid concentrations are not advisable because they would shorten tube life-times and reduce the precision of the measurements. The use of dilute sample solutions, according to the procedure given under Experimental, is therefore recommended.In earlier work21 one of us found an interference from boron in the analysis of algae using lanthanum as the modifier, and it was found necessary to extract boron with a diol into IBMK prior to the determination of phosphorus. The boron content in orchard leaves is 33 pg 8-1 so that an interference could be expected in the analysis of this sample. When lanthanum (10 pg La) was used as a modifier an average phosphorus content of 0.155 k 0.005% was found which is much lower than the certified value. Addition of calcium (5 pg Ca) to the lanthanum modifier resulted in a slightly higher average of 0.170 t- 0.002% which, however, is still substan- tially lower than the certified content.This boron interference was further investigated by carrying out a series of measurements on a phosphorus reference solution with lanthanum modifier in triplicate, alternately with and without the addition of boron. The result of the experiment, using boron concentrations between 0.05 and 5000 mg 1-1 is depicted in Fig. 4. Increasing boron concentra- tions have a progressively depressing effect on the integrated absorbance signal of phosphorus. There is, however, also a substantial influence of the interferent on subsequent determi- nations of phosphorus made with no further addition of boron. The interferent apparently remains on the platform after the atomisation and clean-out stages (this is not surprising considering the low volatility of some boron compounds) , and affects sensitivity for subsequent determinations.No interference from boron on the determination of phosphorus is found if palladium is used as the modifier. This strongly suggests that the mechanisms by which the two modifiers stabilise phosphorus are different as boron is almost certainly left on the platform, independent of the modifier used. A possible explanation for this is that active carbon sites at the graphite surface are directly involved in the stabilisation of phosphorus by lanthanum. If boron reacts preferentially with these active carbon sites, they are no longer available for phosphorus. The palladium modifier, on the contrary, appar- ently does not require participation of active carbon sites to stabilise phosphorus. A similar effect was found in earlier work12 for tantalum carbide coated tubes and platforms where stabilisation of phosphorus by the palladium modifier was more effective than by lanthanum.Iron22 as well as nickel and copper23 have been reported to cause spectral interferences in the determination of phos- phorus if a continuum-source background corrector is used. While the latter two elements are not likely to be present in concentrations that would cause errors in the analysis of biological materials, the same cannot be assumed for iron. Some of the standard reference materials were therefore analysed independently using instruments with continuum- source and Zeeman-effect background correction. The results together with the certified iron contents are summarised in Table 2.There is no significant difference in the results obtained for milk powder and bovine liver, which are also in good agreement with the certificate values. The results obtained using continuum-source background correction are somewhat lower than those using Zeeman-effect background correction for spinach and oyster tissue. For spinach the result obtained using Zeeman-effect background correction agrees well with the certificate, whereas that obtained using the continuum-source technique does not, which may indicate some problems due to spectral interferences. This is supported by the fact that this sample has the highest iron content. The situation is not so clear for the oyster tissue as both values are lower than the recommended, but not certified value, for phosphorus content.Nevertheless it is advisable to use Zeeman-effect background correction for phosphorus deter- mination in samples which may contain high iron concentra- tions in order to avoid possible spectral interferences. Results of the Determination In accordance with the results of the previous experiments all determinations were carried out on sufficiently diluted sample solutions using the palladium and calcium nitrates mixed modifier and an instrument with Zeeman-effect background correction. The phosphorus content was calculated from the linear regression equation obtained for reference solutions free of matrix and containing the same mass of modifier. The results for the standard reference materials investigated are summarised in Table 3 and compared with the certified contents. Good agreement is obtained for all samples except for the oyster tissue, the phosphorus content of which has, however, not been certified.A review of the recent literature on the determination of phosphorus in this sample indicates that the value given by the NBS may be a little high. Contents of 0.76?'0,~ 0.78?'09 and 0.77707 were reported using various AAS procedures, and a value of 0.653% was reported using spectrophotometry.24JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 315 Conclusion Phosphorus can be determined rapidly and accurately with GFAAS. The accuracy of the analytical results, however, depends on the modifier and for some samples also upon the background correction technique used. All interferences have to be under control as the analyte additions technique cannot be applied due to the pronounced non-linearity of the analytical graph.The palladium modifier has now been applied successfully to the determination of more than ten elements. This research work was supported by the Conselho Nacional de Desenvolvimento Cientifico (CNPq), Brazil. 1. 2. 3. 4. 5 . 6. 7. 8. References L’vov, B. V., and Khartsyzov, A. D., Zh. Prikl. Spektr., 1969, 11, 9. Ediger, R. D., Knott, A. R., Peterson, G. E., and Beaty, R. D., At. Absorpt. Newsl., 1978, 17, 28. Person, J. A., and Frech, W., Anal. Chim. Acta, 1980,119,75. PrevBt, A., and Gente-Jauniaux, M., At. Absorpt. Newsl., 1978, 17, 1. Slikkerveer, F. J., Braad, A. A., and Hendrikse, P. W., At. Spectrosc., 1980, 1, 30. Kubota, T., Veda, T., and Okutani, T., Bunseki Kagaku, 1984, 33, 633. Lin, S-W., and Julshamn, K., Anal. Chim. Acta, 1984, 158, 199. Hogen, M. L., Cereal Chem., 1983, 60, 403. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. Langmyhr, F. J., and Dahl, I. M.,Anal. Chim. Acta, 1981,131, 303. Saeed, K., and Thomassen, Y., Anal. Chim. Acta, 1981, 130, 281. Curtius, A. J., Schlemmer, G., and Welz, B., J . Anal. At. Spectrom., 1986, 1, 421. Curtius, A. J., Schlemmer, G., and Welz, B., J . Anal. At. Spectrom., 1987, 2, 115. Welz, B., Curtius, A. J., Schlemmer, G., Ortner, H. M., and Birzer, W., Spectrochim. Acta, Part B, 1986, 41, 1175. Schlemmer, G., and Welz, B., Spectrochim. Acta, Part B , 1986, 41, 1157. “Techniques in Graphite Furnace Atomic Absorption Spec- trometry,” Perkin-Elmer Part No. 0993-8150, Ridgefield, CT, 1985. Fernandez, F. J., Bohler, W., Beaty, M. M., and Barnett, W. B., At. Spectrosc., 1981, 2 , 73. de Galan, L., and Samaey, G. F., Spectrochim. Acta, Part B, 1969,24, 679. Welz, B., Fresenius Z . Anal. Chem., 1986, 325, 95. Barnett, W. B., Spectrochim. Acta, Part B , 1984, 39, 829. Barnett, W. B., Vollmer, J. W., De NUZZO, S. M., At. Absorpt. Newsl.. 1976. 15. 33. Curtius, A. J., unpublished results. Welz, B., Voellkopf, U., and Grobenski, Z . , Anal. Chim. Acta, 1981, 136, 201. Russeva, E., Havezov, I., Spivakov, B. Y., and Shkinev, M. V., Fresenius Z. Anal. Chem., 1983, 315, 499. Pritchard, M. W., Anal. Chim. Acta, 1984, 157, 313. Note-Reference 12 is to Part 2 of this series. Paper J61111 Received November loth, 1986 Accepted December 15th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200311

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 317 Determination of Copper in Biological Microsamples by Direct Solid Sampling Graphite Furnace Atomic Absorption Spectrometry Les Ebdon and E. Hywel Evans Department of Environmental Sciences, Plymouth Polytechnic, Drake Circus, Plymouth PL4 8AA, UK Atomic absorption spectrometry in conjunction with electrothermal atomisation has been employed to determine copper in biological reference materials and real samples using two techniques of solid sampling. Both techniques, namely direct slurry atomisation from the tube wall and direct platform atomisation gave both accurate and precise results for the certified reference materials (orchard and tomato leaves, pepperbush, mussel and bovine liver). Portions of caddis fly larvae and aquatic plants from a copper contaminated stream were analysed and the copper contents determined by these methods.The use of a graphite microboat for direct solid sampling using only microgram samples is proposed as a valuable tool for identifying the location of trace metals in biological tissues. Keywords: Atomic absorption spectrometry; electrothermal atomisation; solid sampling; slurry atomisation; copper in biological samples The analysis of samples by atomic absorption spectrometry (AAS) in conjunction with electrothermal atomisation (ETA), by atomisation directly from the solid state, is attractive on the following grounds: (i) the time-consuming digestion of samples in acid prior to analysis can be omitted; (ii) the risks of contamination and analyte loss involved in (i) are much reduced; (iii) hazards associated with some acid digestions are eliminated; (iv) dissolution of the sample may involve excessive dilution; (v) the selective analysis of micro-amounts of solid samples for traces of analyte is facilitated .Consequently solid sampling ETA-AAS has attracted some interestl-9 but this has been limited because it raises a new set of problems, which stem mainly from the large amounts of matrix typical of solid samples. Similarities between the volatility of analyte and matrix sometimes mean that the matrix cannot be effectively vapourised during the pyrolysis stage without loss of analyte. In turn, a large matrix residue during atomisation can cause chemical interferences and very high background signals. Other limitations include homo- geneity problems, reliability of sample introduction and finding adequate calibration procedures.Therefore the development of solid-sampling techniques has centred mainly on reduction in chemical interferences and background signals caused by the matrix. The selective volatilisation technique has been developed5 to vapourise selectively and remove most of the matrix during the pyrolysis stage at a temperature at which the analyte is not vapourised. Pickford and Ross9 and Langmyhr and Aamodt2 have obtained accurate results for the determination of Ag, Cu, Pb, Mn and Cd, Cu, Pb, Mn, respectively, in NBS bovine liver using solid sampling and the standard additions method of analysis. Likewise, Alder et aZ.3 have studied 13 trace elements in hair with reasonably accurate results for Cu, Pb, Al, Fe, Co, Ni, Cr, Si and Mn, although similar analyses of the relatively volatile elements Zn, Bi and Ag gave low results.Lord et aZ.4 have found that calibration with aqueous standards is accurate for low volatility elements such as Al, Cr and Cu, although not for volatile elements such as Pb and Zn in the analysis of NBS orchard leaves. Results of the analysis of high volatility elements3*4 have proved unsatisfactory due to the similarity in volatility of matrix and analyte. If low pyrolysis temperatures are used to prevent loss of analyte, it leads to high background signals and matrix interferences during atomisation. Conversely, higher pyrolysis temperatures, while eliminating most of the matrix, cause loss of analyte.A major step towards solving this problem has been made by the introduction of the matrix/ analyte modification technique by Ediger et aZ. 10311 Chakrabarti et aZ.5 have used a combination of analyte and matrix modification for the determination of Cd in NBS bovine liver, whilst also utilising the solid-sampling technique, and have achieved good agreement with certificate values. Recently Pb has been successfully determined in various biological CRMs by direct solid sampling using ascorbic acid as a matrix modifier.8 An alternative approach has been utilised by Grobenski et aZ.6 They introduced oxygen during the pyrolysis step in the determination of Rb in NBS spinach and Mn in NBS bovine liver with good results. However, losses of Pb in the analysis of NBS spinach were observed.L’vov12 has suggested placing a graphite platform within the conventional Massmann furnace so that on heating it approxi- mates more closely to a constant-temperature design. Slavin and Manning13 have reported reduced interference on the Pb response in matrices containing chloride, sulphate and phos- phate. The usefulness of platform atomisation in conjunction with the solid-sampling technique has been demonstrated by Chakrabarti et aZ.,5 who have reported higher sensitivity for the relatively volatile elements compared with atomisation from the tube wall, although with similar accuracy. Direct solid sampling would be of value in a number of biological analyses where microsamples could be directly loaded into a graphite furnace for ETA-AAS.There is considerable interest not only in total trace metal levels in botanical and animal samples from polluted environments , but also in identifying the preferred location of trace metals in biological tissues. Accordingly we have developed a method for the direct determination of Cu in biological samples by solid sampling ETA-AAS. The method was validated by the successful analysis of a variety of certified reference materials using both platform and wall atomisation and then applied to dissected tissue samples of caddis fly larvae and grasses. Experimental Apparatus All analyses were performed with an atomic absorption spectrometer (Instrumentation Laboratory Video 12, War- rington, Cheshire, UK) equipped with a controlled-ternpera- ture furnace atomiser (Instrumentation Laboratory IL655).Less sensitive absorption lines as.well as the resonance line for Cu were used due to the variation in sample size for different318 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 techniques. Experimental conditions are shown in Table 1. Commercially available pyrolytically coated graphite cuvettes as supplied by the furnace manufacturer were used, namely round for atomisation from the tube wall and rectangular with solid-sampling microboat for platform atomisation. All weighing was carried out with a microbalance (Cahn 29 Automatic Electrobalance, Cahn Instruments, Cerritos, USA). Reagents and Standards Nitric acid, 70% (AnalaR, BDH Chemicals, Poole, Dorset, UK) was diluted to 0.7% with de-ionised, doubly distilled water and used for all dissolution and washing.A stock solution of Cu, 1000 pg ml-1 (Spectrosol, BDH Chemicals) was diluted as appropriate with HN03 prior to analysis for use as standard solutions. Certified Reference Materials The following CRMs with certified copper content were analysed: NIES No, 1 pepperbush (National Institute of Environmental Studies, Ibaraki, Japan); NIES No. 6 mussel (National Institute of Environmental Studies); NBS 1573 tomato leaves (National Bureau of Standards, Washington DC, USA); NBS 1571 orchard leaves (National Bureau of Standards); and NBS 1577 bovine liver (National Bureau of Standards). Where possible pre-treatment of reference materials was kept to a minimum as they were prepared for slurry atomisation.For NIES pepperbush it was necessary to grind the material until it was of a texture suitable for slurrying. This was achieved without difficulty in a pestle and mortar by manual grinding. This suggests the absence of any critical constraints arising from particle size in this methodology. The reference materials were freeze-dried, suitable amounts accu- rately weighed and slurried in 0.01% Triton X-100 (BDH Chemicals). The slurries were dispersed in an ultrasonic bath for 30 min. Biological Samples Samples of the larvae of the caddis fly Plectrocnemia conspersa (Curtis) were collected in November 1985 and August 1986 from the copper-rich headwaters of Darley Brook, which -~~ ~~ Table 1. Operating conditions for atomic absorption spectrometer Wavelengthhm Parameter 324.7 216.5 244.1 222.6 Bandpasslnm .. . . . . . . 1.0 1.0 1.0 1.0 Lampcurrent/mA . . . . . . 4.2 3.8 4.0 4.0 BackgroundcurrentImA . . . . 2.5 2.5 2.5 2.5 Backgroundcorrection . . . . Smith - Hieftje Integration . . . . . . . . Peak area for 4 s drains water from a disused copper mine on Bodmin Moor, Cornwall, UK. Each larva was placed in a separate cell in a compartmentalised tray filled with water from the site of collection and stored for 48 h at 10°C to remove the con- taminating effects of any ingested material. The larvae were subsequently dissected into head, thorax and abdomen, so that Cu could be determined separately in each of these three regions, then freeze-dried and accurately weighed. Samples of the aquatic plant Juncus bulbosus were collected from the same site, the plants carefully washed and dried and different parts of the plant (namely shoot and root) isolated, ground and slurried to give a 0.004% slurry in 0.008% Triton X-100.Alternate samples were taken for the direct solid sampling microboat techniques and for acid digestion. Appropriate blanks were prepared for all samples. Operating Procedure Optimisation of heating conditions Optimum heating conditions for both tube-wall and platform atomisation were determined for Cu in 0.7% nitric acid solution and the slurried reference materials. Plots of absor- bance as a function of furnace temperature for both pyrolysis and atomisation cycles were established for each material in the usual manner as described elsewhere.14 Tube-wall atomisation Aliquots (5,lO or 20 p1) of the slurries were delivered into the central part of the graphite furnace tube using an adjustable micropipette (Gilson Pipetman P, Anachem, Luton, Bedford- shire, UK).The furnace cycle was run and the integrated (peak area) absorbance noted. Platform atomisation Samples of NIES mussel (1.00-1 S O mg) and NBS bovine liver (1.70-2.70 mg) were weighed accurately into separate graph- ite microboats. For the caddis fly larvae, samples were weighed on to separate graphite microboats and 5 p1 of 0.7% nitric acid were added to each. The samples were carefully crushed with a clean spatula to facilitate entry of the microboats into the furnace. The nitric acid was necessary to wet the brittle samples and hence prevent losses when they were crushed. Each microboat in turn was inserted into the furnace, the cycle run and peak-area absorbance noted.Copper was determined in every sample by direct calibra- tion with conventional acid standards. At least six replicate determinations were made for each of the reference materials. Results and Discussion Optimum Furnace Heating Conditions Optimum furnace heating conditions for Cu in acid solution and the reference materials are shown in Tables 2 and 3. With respect to the analysis of NBS bovine liver and NIES mussel, the heating conditions for Cu in acid solution were used for Table 2. Optimum heating conditions for tube-wall atomisation of copper Stage Matrix Acidsolution . . . . NBSbovineIiver . . . . NIESmussel . . . . NIESpepperbush . . . . NBS tomatoleaves . .NBS orchard leaves . . * Rate of temperature increase, 16 "C s-l. -t Rate of temperature increase, 1150 "C s-1. Drying* TIT tls 150 55 150 55 150 55 100 50 100 50 100 50 Pyrolysis* TIT tls 950 50 1100 50 1000 50 750 50 750 50 750 50 Atomisationt TIT tls 2100 5 2100 5 2100 5 2200 5 2200 5 2200 5 Cleaning T/"C tls 2600 5 2600 5 2600 5 2600 5 2600 5 2600 5JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 319 Table 3. Optimum heating conditions for platform atomisation of copper Stage Drying* Pyrolysis* Atomisation? Cleaning Matrix TPC tls TIT trs TPC tls TIT tls Acidsolution . . . . 150 55 1000 50 2500 5 2600 5 NBSbovineliver . . . . 150 55 1100 50 2500 5 2600 5 NIESmussel . . . . 150 55 1000 50 2500 5 2600 5 * Rate of temperature increase, 16 "C s-1.Rate of temperature increase, 1500 "C s-1. Table 4. Determination of copper by direct slurry atomisation from the furnace tube wall Found Wavelength for Certificate Sample analysislnm valuelpg g-1 Meadpg g-1 RSD, % NBS bovine liver SRM 1577 . . . . 216.5 193 k 10 191 4.9 NIES No. 6 mussel . . . . . . . . 216.5 4.9 k 0.3 5.9 5.5 NIES No. 1 pepperbush . . . . . . 324.7 1 2 5 1 10.6 19.8 NBS tomato leaves SRM 1573 . . . . 324.7 11 -t 1 10.9 7.3 NBSorchardleavesSRM1571 . . . . 324.7 1 2 2 1 12.2 6.6 Table 5. Determination of copper by direct platform atomisation Found Wavelength for Certificate Sample analysishm value/pg g-1 Meadpg g-1 RSD, % NBS bovine liver SRM 1577 . . . . 244.1 193 k 10 192 1.7 NIES No. 6 mussel . . . . . . . . 222.6 4.9 F 0.3 7.7 3.6 r I 200 ri C .: 150 E 8 100 8 O O .4- C 50 a ( a ) 0 0 0 0 0 0 0 400 300 200 600 400 0 0 0 0 o o 0 01 *0° I t I I 0, I I I I I 0 I 01 J o 0.5 0.6 0.7 0.8 0.9 1.0 1.1 0.5 1.0 1.5 2.0 2.5 3.0 0 0 o o 0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Dry mass of headimg Dry mass of thoraxhg Dry mass of abdomen/mg Fig. 1. Variation of copper concentration with dry mass of caddis fly larvae collected in November 1985: (a) head; ( b ) thorax; and (c) abdomen both tube-wall and platform atomisation to minimise the possibility of analyte loss from acid standards during pyrolysis. This was considered to be acceptable as atomisation tempera- tures were identical. In contrast, for the analysis of NIES pepperbush, NBS tomato leaves and NBS orchard leaves, the furnace heating conditions chosen for Cu in the respective matrices were used to avoid analyte loss from the matrices during pyrolysis.The graphite contacts in the furnace used tend to result in slower heating rates (1150 and 1500 "C s-1) but this does ensure no Cu contamination from metal contacts. With these heating rates peak-area measurements are clearly optimal. It was impossible to establish optimum furnace heating conditions for the analysis of the individual caddis fly larvae due to the variation in Cu concentration between samples, hence the heating conditions established for Cu in acid solution and platform atomisation were used. This was considered the best option due to the greater similarity of the caddis fly larvae matrix to those of NBS bovine liver and NIES mussel, rather than to those of the other reference materials. Likewise, the furnace heating conditions established for Cu in the botanical reference materials were used for the analysis of the aquatic plants.Results of Analyses Excellent agreement with certificate values was obtained for the analysis of all the reference materials by direct slurry atomisation from the tube wall, together with acceptable precision in most instances (Table 4). The low degree of precision associated with the analysis of NIES pepperbush may have been due to the difficulty in obtaining a suitably homogeneous slurry. This is a problem which is less likely to occur in samples where the tissue is all taken from one part of the plant. Likewise, excellent agreement with certificate values and acceptable precision was obtained for the analysis of NBS bovine liver and NIES mussel by direct platform atomisation (Table 5).Using the slurry method it is possible to inject much smaller sample loadings accurately, but using direct solid sampling the sample masses were 204-325 times greater, and consequently use was made of less strong absorption lines. With the caddis fly larvae, Cu concentration varied widely between samples, as would be expected in the analysis of individual biological specimens. In an initial study Cu concen- tration was plotted against dry mass for the heads, thoraxes and abdomens, respectively [Fig. l(a), ( b ) and ( c ) ] . The320 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 Table 6. Determination of copper in caddis fly larvae (collected in August 1986) by direct platform atomisation Copper concentratiodmg g-1 ~~ Sample No.Head Thorax Abdomen 1 3.93 1.13 1.69 2 1.66 0.59 0.77 3 <0.08 <0.05 0.38 4 0.11 0.34 0.42 5 <0.07 <0.06 0.10 6 <0.10 ~ 0 . 0 7 0.38 Table 7. Determination of copper in an aquatic plant (Juncus bulbosus) Copper concentratiodmg g-1 ~~ Method Root Shoot Slurry atomisation ETA-AAS . . 3.2 1.2 ETA-AAS . . . . . . . . 3.2 1.1 AciddigestioqFAAS . . . . 2.9 1.1 Solid sampling with microboats, results compared favourably with those of Darlington et al. ,15 who determined Cu in whole acid digested larvae gathered from the same site and found an exponential decrease in Cu concentration with increasing dry mass. The trends exhibited in this study do not follow such a clear relationship; however, this can be explained by the fact that the range of sample size was small compared with the previous work15 and falls on only a small part of the exponential curve.Subsequently, Cu was determined in the heads, thoraxes and abdomens of individual larvae and concentrations in the three body regions compared for each. The sample size was too small to ascertain the significance in the different concentrations found (see Table 6) but generally higher levels were found in the abdomen than in the thorax of individual specimens. This is in agreement with the larger study shown in Fig. l(a)-(c) where the mean Cu concentrations are in the order abdomen > thorax > head. Further work is continuing to investigate these relationships and their biological sig- nificance.The levels of Cu in a typical root and shoot sample of Juncus bulbosus, an aquatic grass, were determined in three ways. A sample (0.2 g) was acid digested in nitric acid and the Cu content determined by conventional flame atomic absorption spectrometry at 324.7 nm. Similar sized samples were taken of shoot and root and slurries prepared (0.004%). These slurries were then analysed by slurry atomisation ETA-AAS as described above. Finally, whole portions of the shoots and roots (ca. 50 pg) were prepared using a microtome and analysed by the direct solid sample microboat ETA-AAS technique described above. The results obtained are given in Table 7. Not unexpectedly higher Cu levels were found in the roots than the shoots. The excellent agreement between the results gives confidence in the solid sampling method which appears to offer an accurate method for the determination of Cu in microsamples of biological tissue.Conclusions The slurry atomisation and direct microboat insertion solid sampling techniques discussed have been shown to be reasonably accurate and precise for the determination of Cu in various certified reference materials. Hence they have been validated to a certain extent for the analysis of real biological samples. The analysis of the caddis fly larvae by direct platform atomisation was an example of this and was qualitatively validated by a comparison with other work. 15 The application of the techniques to both animal and botanical samples shows them to offer a speedy analytical method less prone to contamination or losses than conven- tional dissolution techniques.In addition, the microboat technique offers the possibility of the direct determination of Cu in microgram samples of biological tissues. Such a technique provides a powerful tool for the elucidation of the location of trace metals in such samples of wide environmental and clinical application. We thank Hew6 Tallec for his assistance with the botanical samples, Huw Parry for advice on ETA-AAS and Steven Darlington for his help in collecting samples. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. References Pickford, C. J., and Rossi, G., At. Absorpt. Newsf., 1975, 14, 78. Langmyhr, F. J., and Aamodt, J., Anal. Chim. Acta, 1976,87, 483. Alder, J. F., Samuel, A. J., and West, T. S . , Anal. Chim. Acta, 1976,87, 313. Lord, D. A., McLaren, J. W., and Wheeler, R. C., Anal. Chem., 1977, 49, 257. Chakrabarti, C. L., Wan, C. C., and Li, W. C., Spectrochim. Acta, Part B , 1980,35, 93. Grobenski, Z . , Lehmann, R., Tamm, R., and Welz, B., Mikrochim. Acta, 1982, I, 115. Ebdon, L., and Pearce, W. C . , Analyst, 1982, 107,942. Ebdon, L., and Lechotycki, A., Microchem. J., 1986,34,340. Ebdon, L., and Parry, H. G. M., J . Anal. At. Spectrum., 1987, 2 , 131. Ediger, R. D., Peterson, G. E., and Kerber, J. D., At. Absorpt. Newsl., 1974, 13, 61. Ediger, R. D., At. Absorpt. Newsl., 1975, 14, 127. L’vov, B. V., Spectrochim. Acta, Part B, 1978,33, 153. Slavin, W., and Manning, D. C., Anal. Chem., 1979, 51,261. Ebdon, L., “An Introduction to Atomic Absorption Spectro- scopy,” Heyden, London, 1982, p. 120. Darlington, S. T., Gower, A. M., and Ebdon, L., Environ. Sci. Technol. Lett., 1986, 7, 141. Paper J6f106 Received November 6th, 1986 Accepted December 11 th, 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200317

出版商:RSC    

年代:1987

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 321 Determination of Trace Amounts of Thorium and Uranium in Coal Ash by Inductively Coupled Plasma Atomic Emission Spectrometry after Extraction with 2-Thenoyltrifluoroacetone and Back-extraction with Dilute Nitric Acid Eijiro Kamata, Ryozo Nakashima and Masamichi Furukawa Government Industrial Research Institute, Nagoya, I - 7 Hirate-cho, Kita-ku, Nagoya 462, Japan A method of sample preparation is described that can be used for the determination of trace amounts of thorium and uranium in coal ash by inductively coupled plasma atomic emission spectrometry. After acid digestion of the samples, the elements are separated independently from the matrix by 2-thenoyltrifluoroacetone (lTA) - benzene extraction procedures. Thorium is extracted with 0.25 M TTA at pH 1.2 in the presence of ascorbic acid.Uranium(V1) is extracted with 0.16 M l T A at pH 3.5 in the presence of EDTA. The amount of EDTA to be added to the sample solutions must be properly controlled because an excess reduces the recovery of uranium. Thorium and uranium are finally measured in the back-extract in 2 M nitric acid. The detection limits were 11 pg 1-1 for Th and 29 pg 1-1 for U and the relative standard deviations ranged from 3.0 to 8.6% for Th and 3.9 to 4.3% for U. Keywords : Inductively coupled plasma atomic emission spectrometry; thorium determination; uranium determination; coal ash analysis The determination of thorium and uranium in coal ash has received considerable attention for environmental reasons because of the increasing consumption of coal as a substitute for petroleum. Inductively coupled plasma atomic emission spectrometry (ICP-AES) has been widely adopted as a versatile technique for the determination of trace elements because of its high sensitivity and relative freedom from matrix effects.In ICP-AES, however, problems with spectral interferences and matrix effects occur in the determination of trace amounts of thorium and uranium in complicated matrices such as coal ash and siliceous materials.1-3 Therefore, it is necessary to separate the elements of interest from the matrix. Ion exchangers have been used for the pre-concentra- tion and separation of these elements from the matrix,4-6 and also solvent extraction procedures have often been used for this purpose.7-9 Mahanti and Barnes3 described a method that combined ICP-AES and a poly(dithi0carbamate) chelating resin for the determination of 14 elements, including thorium and uran- ium, in coal and energy-related materials.Coal was ashed prior to acid digestion. After adjustment of the pH, the sample solution was passed through the resin for separation and concentration of the elements of interest. Finally, the resin was digested with hydrogen peroxide and concentrated nitric acid for presentation to the ICP-AES system. In this investigation, a method of sample preparation has been developed that can be used for the determination of trace amounts of thorium and uranium in coal ash samples by ICP-AES. The method is based on extraction with 2-thenoyl- trifluoroacetone (TTA)lo - benzene and back-extraction with dilute nitric acid; in the former the optimum conditions were studied for the separation of the elements from the matrix after acid digestion of the sample.The back-extraction procedure is also indispensable for the accurate determination of trace amounts of these elements by ICP-AES because serious spectral interferences and an increase in background intensity due to the organic solvent are observed on direct introduction of the organic extract into the plasma. Experimental Instrumentation The instrumentation, operating conditions and wavelengths used are listed in Table 1. Reagents All the reagents were of analytical-reagent grade. Glass- distilled water was purified by treatment in a Millipore Milli-Q system and used throughout for dilution purposes.Thorium stock solution, 1000 mg 1-1. Prepared by dissolving 0.238 g of Th(N03)4.4H20 in 100 ml of 2 M nitric acid. The solution was standardised complexometrically by EDTA-Zn back-titration. Urunium(VI) stock sohtion, 1000 mg 1-1. Prepared by dissolving 0.211 g of U02(N03)2.6H20 in 100 ml of 2 M nitric acid. The solution was standardised gravimetrically by the quinolin-8-01 method. 11 Sample Preparation Digestion 12 A finely powdered coal ash sample (0.5 g) was placed in a PTFE beaker (300 ml), 20 ml of hydrofluoric acid (40%) and 5 ml of nitric acid (61%) were added to each sample and the mixture was heated on a hot-plate at 120 "C for 2 h or until it was dry (removal of silica). The residue was cooled and 10 ml of perchloric acid (6244%) and 10 ml of nitric acid were Table 1.Instruments and operating conditions Spectrometer . . R.f. generator Coolantgas . . Plasmagas . . Carrier gas . . Nebuliser . . Data acquisition Integration time Wavelengthhm . . Seiko I & E Model JY 38 P 11,100 cm focal length; 3600 g nm-1 holographic grating; range of optimum efficiency 200-500 nm; entrance and exit slit widths, 35 and 55 nm, respectively Plasma-Therm Type HFP 2500F, 27.12 MHz; incident power, 1300 W; reflected power, 3 W . . Ar, 16lmin-1 . . Ar,3.51min-l . . Ar, 0.2lmin-1 . . . . . . Glass, concentric, with spray chamber; uptake of aqueous solution, 1.3 ml min-1 Microcomputer, Oki Electric Model IF 800 with electronic console (Seiko I & E Model C 20) and chart recorder (Nippon Denshi Kagaku Model U-228) Th 11,401.913; U 11,409.014 .. 5 s . .322 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 added. The beaker was covered with a PTFE watch-glass and the residue was digested at 200 "C for 4 h or until a light brown solution was obtained. The cover was then removed and the solution was evaporated (120 "C) almost to dryness. The slurry thus obtained was warmed with about 30 ml of 2 M nitric acid until it had dissolved. The solution was transferred into a 100-ml calibrated flask, which was filled with the rinsings (2 M nitric acid) from the beaker. Preparation of sample solutions for measurement by ICP-AES Procedure for thorium. An aliquot (usually 10 ml) of the sample solution was placed in a separating funnel (30 ml); 2 ml of 20% ascorbic acid solution were added and the pH of the solution was adjusted to 1.2-1.5 with 4 M ammonium acetate solution.The total volume of the aqueous phase was finally adjusted to 20 ml with water. The thorium was extracted with 10 ml of 0.25 M TTA in benzene with mechanical shaking for 15 min. (Caution-Benzene is highly toxic and appropriate precautions should be taken.) The contents of the funnel were allowed to stand for 1 h for phase separation and the aqueous phase was then discarded. The back-extraction was performed by addition of 10 ml of 2 M nitric acid to the organic phase and mechanically shaking for 15 min. The contents were allowed to stand for 30 min for phase separation. After a small portion of the back-extract had been discarded, the stem of the funnel was washed with water and the droplet remaining in the stem was wiped off with tissue paper. This washing is necessary to prevent contamination of the back-extract by the residual aqueous phase in the first extraction.Finally, the back-extract was transferred into a stoppered test-tube (10 ml). The samples were then ready for presentation to the ICP-AES instrument. A calibration graph was similarly prepared, taking thorium standard solutions through the extraction procedure without addition of the masking agent. Procedure for uranium. An aliquot (usually 10 ml) of the sample solution was placed in a separating funnel (30 ml); a definite volume of 0.1 M EDTA solution, calculated from the content of iron and aluminium in the sample, was added.The amount of EDTA to be added to the aqueous phase is a 1.00 molar ratio for iron and a 0.50 molar ratio for aluminium (see Table 4). The pH of the solution was adjusted to 3.5-4.0 with 4 M ammonium acetate solution. The subsequent procedure is the same as that for thorium except for the use of 0.16 M TTA in benzene as the extractant. Results and Discussion Choice of Analytical Lines The analyte lines were selected on the basis of their peak and background intensities and their freedom from spectral interferences. In our instrument the analyte line for thorium, 401.913 nm, was the most intense and superior to other analyte lines with respect to spectral interferences.13 On the other hand, the analyte line for uranium, 409.014 nm, where the spectral intensity was about 15% lower than that of the U I1 385.958-nm line, was selected in the proposed method because this line showed a better background level.Interferences from Major Elements The effect of major elements on the determination of thorium and uranium(V1) was examined in aqueous solution (2 M nitric acid). The tolerance limits for seven major elements (Al, FeIII, Ca, Mg, K, Na and Ti) in the determination of 0.5 mg 1-1 of thorium and 1 mg 1-1 of uranium(V1) are shown in Table 2. Then, the effect of the matrix of coal fly ash was examined by using a synthetic solution of matrix elements for NBS SRM 1633a coal fly ash (the concentration of each element is given in Table 2); the spectral intensity was reduced to 82% for thorium and 85% for uranium(V1). Therefore, thorium and uranium must be separated from the matrix.TTA Extraction and Removal of Interferences The effect of pH on the extraction of Th - TTA or U - TTA chelate in benzene was investigated. Thorium was extracted with 0.25 M TTA in benzene from an aqueous solution, the pH of which was adjusted by addition of 4 M ammonium acetate solution. The recovery of thorium was 100% above a pH of 1 (Fig. 1). This agreed closely with the results obtained by Hagemann .I4 On the other hand, uranium(V1) was quantitatively extrac- ted with 0.16 M TTA in benzene only in the pH range 3-4.5 (Fig. 1). This is different from the results of Khopkar and De,15 who studied spectrophotometrically the extraction of U - TTA chelate in benzene and reported that the pH range 3.5-8.0 can safely be used for 100% extraction of uranium(V1) from an aqueous solution (5 ml) containing 0.2-1.2 mg of uranium with 0.15 M TTA in benzene.The complete extrac- Table 2. Tolerance limits of major elements for Th (0.5 mg 1-1) and U (1 mg 1-1) in aqueous solution (2 M HN03) Tolerance limit*/mg 1-1 A1 Fe Ca Mg Ti K Na ForTh . . . . . . 20 500 50 200 50 100 100 ForU . . . . . . 500 100 50 500 50 50 50 Synthetic solutiont/mg 1-1 NBS SRM 1633a .700 470 55.5 22.8 40 94 8.5 * The tolerance limits given correspond to the concentration level at which the interference causes an error in the determination of Th and U of more than f 2 % . t The composition is based on the certificate of analysis for NBS 1633a coal fly ash (0.5 g in 100 ml of 2 M HN03). The values are shown for comparison with the tolerance limit of each matrix element.~ ~ ~~ Table 3. Effect of cerium(1V) and uranium(V1) on the determination of thorium Ce or U Th addedpg addedpg 1 (as.)* Ce 10 5 (as.)* Ce 50 5 (ex.)? Ce 50 u 20 U 50 5 (ex.>t U 50 5 (as.>* u 10 5 (as.>* 5 h.>* Recovery of 107 108 100 103 105 106 100 Th, O/o * aq.: aqueous solution (10 ml of 2 M HN03). t ex.: extraction with 0.25 M TTA in benzene (10 ml), followed by back-extraction with 10 ml of 2 M €€NO3. 0 1 I I I I I 1 1 2 3 4 5 6 PH Fig. 1. Effect of pH on the extraction of Th (A) and Uvl (B) with TTA in benzene. P A : 0.25 M for Th; 0.16 M for Urn. Th added, 5 p.5; Urn added, 10 pg. Volume of aqueous phase, 20 ml; volume of organic phase, 10 ml. Back-extract, 10 ml of 2 M HN03JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL.2 323 tion of pg ml-1 levels of uranium in the TTA - benzene extraction system may be limited to the narrower pH range. The TTA extraction of thorium from the digested sample is inhibited by iron(III), which forms a red precipitate; this can be reduced to iron(II), which does not react with TTA in acidic media, with ascorbic acid as reductant. Thorium (5 pg in 20 ml of aqueous solution) was quantitatively extracted with 0.25 M TTA in benzene in the presence of 10 mg of iron(II1) by addition of 2 ml of 20% ascorbic acid solution to the aqueous phase before extraction. In addition, thorium was freed from cerium and uranium, which cause spectral interferences at the analyte line (401.913 nm) of thorium in this extraction procedure (Table 3).Aluminium ion, 10 mg of which were added to the aqueous phase (20 ml), did not interfere in the determination of thorium. Table 4. 7TA extraction of Uw in the presence of various elements 0.1 M EDTA Recovery Element added/ml Molar ratio of UVI,t added/mg* (A) ( B ) Yo Al, 10 mg FeIII, 10 mg 94 0.74 0.20 97 1.48 0.40 98 2.22 0.60 99 2.96 0.80 100 3.70 1.00 99 - - Red ppt. 1.72 0.96 91 1.75 0.98 95 1.79 1 .oo 100 1.83 1.02 98 For Fe$ For Alf - - NBS SRM 1633aQ (Al, 2.14 1.00 0.50 99 7 mg; Fe, 4.7 mg; 2.39 1.00 0.60 99 Ca, 0.56 mg; 2.91 1.00 0.80 99 Ti, 0.4 mg) 4.12 1.20 1.20 95 Mg, 0.23 mg; 3.43 1.00 1.00 98 6.86 2.00 2.00 47 * Amount added to the aqueous phase (20 ml). t 10 pg of UVJ were added to the aqueous phase (20 ml) and extracted with 0.16 M TTA in benzene.Back-extraction was per- formed with 10 ml of 2 M HN03. $ The volume of 0.1 M EDTA solution equivalent to a molar ratio of 1.00 for Fe and A1 in NBS SRM 1633a is 0.84 and 2.59 ml, respectively. § The amount of each major element was calculated from the certificate of analysis for NBS SRM 1633a coal fly ash and based on the use of 10 ml of the digested solution (0.5 g in 100 ml of 2 M HN03). Table 5. Detection limits and background equivalent concentrations Detection Wavelength/ limit */ BEQ/ Element nm Pg I-' mg 1-1 Th . . . . . . 401.913 11 0.66 U . . . . . . 409.014 29 3.50 deviation of the background. * Concentration giving a signal equal to three times the standard i Background level expressed as analyte concentration equivalent.In the 'ITA extraction of uranium(VI), 10 pg of uranium in 20 ml of aqueous solution were determined in the presence of 10 mg of Ca or 1 mg of each of Co, Cu, Pb and Zn. The recovery of uranium was quantitative. On the other hand, in the presence of iron(II1) the extraction of uranium was also reduced owing to the formation of a red precipitate. In addition, it is necessary to take into account the interference from aluminium in the ?TA extraction because of its high content in the sample solution (Table 2). As shown in Table 4, an addition of a 0.4 molar ratio of EDTA solution was sufficient to mask the aluminium, whereas a 1.00 molar ratio of EDTA solution was required for masking the iron(II1). The masking effect of EDTA was confirmed by using a synthetic solution containing matrix elements for NBS SRM 1633a coal fly ash.The volume of EDTA solution to be added to the aqueous phase was calculated by considering only the contents of iron(II1) and aluminium on the basis of the above-mentioned interference study. The EDTA solution was added to the aqueous phase in molar ratios in the ranges 1.00-2.00 for iron(II1) and 0.50-2.00 for aluminium; the recovery of uranium was quantitative up to a 1.00 molar ratio, but it decreased markedly to 47% at a 2.00 molar ratio for both elements. This decrease may be caused by the complexation reaction of the uranyl ion with excess of EDTA.1618 Therefore, the amount of EDTA solution to be added to the sample solution must be properly controlled. It is recommended in the proposed method that the amount of EDTA solution to be added to the sample solution in the aqueous phase before extraction should be a 1.00 molar ratio to the content of iron and a 0.50 molar ratio to that of aluminium (the contents of aluminium and iron in coal ash samples are previously determined by ICP-AES or atomic absorption spectrometry).Quantitative Performance The calibration graph obtained by the proposed method for thorium and uranium was linear for at least two orders of magnitude of concentration above the detection limit. Detec- tion limits and background equivalent concentration (BEC) values for both elements determined in standard solutions with TTA extraction are given in Table 5. These detection limits are much lower than those previously reported3.19 (Th, 50 vg 1-1; U, 50-100 pg 1-1) using ICP-AES. Analytical Results Thorium and uranium were determined in five coal fly ash samples, including NBS SRM 1633a, and a cinder ash sample; the results for the NBS SRM agreed well with the NBS certified values (Table 6) (the contents of A1 and Fe in each sample are listed with reference to the masking for the determination of U).All the results for Th and U given in Table 6 are averages of five replicate determinations; the relative standard deviations ranged from 3.0 to 8.6% for Th and from 3.9 to 4.3% for U. Table 6. Results for the determination of Th and U in coal ash. RSD = relative standard deviation; LOD = limit of detection Ash Th*/pg g-l RSD, YO U*/pg g-l RSD, % Allg g-1 Felg g-1 E . . . . . . . .. . . . 38 5.6 15 4.3 0.150 0.043 D . . . . . . . . . . . . 15 8.6 6 4.0 0.127 0.039 W . . . . . . . . . . . . 24 6.3 8 3.9 0.112 0.021 NBSSRM1633aT . . . . . . 25 3.0 10 3.9 0.14 0.094 . . . . . . . . . . 0.100 0.048 Cinder 16 8.1 LOD - * Mean of five replicate determinations. i NBS certified values: 24.7 k 3 pg g-1 for Th; 10.2 k 0.2 pg g-I for U.324 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY, APRIL 1987, VOL. 2 References Nadkarni, R., Anal. Chem., 1980, 52,929. Floyd, M. A., Fassel, V. A., and D’Silva, A. P., Anal. Chem., 1980, 52, 2168. Mahanti, H. S., and Barnes, R. M., Anal. Chim. Acta, 1983, 149, 395. Korkisch, J., and Dirnirntriadis, D., Talanta, 1973, 20, 1199. Korkisch, J., and Hubner, H., Talanta, 1976, 23, 283. Kiriyarnna, T., and Kuroda, R., Anal. Chim. Actu, 1974, 71, 375. Vernon, F., Kyffin, T. W., and Nyo, K. M., Anal. Chim. Actu, 1976, 87, 491. Motooka, J. M., Mosier, E. L., Sutley, S. J., and Viets, J. G., Appl. Spectrosc., 1979, 33, 456. Bem, H., and Ryan, D. E., Anal. Chim. Acta, 1984,158,119. Poskanzer, A. M., and Foreman, B. M., Jr., J. Znorg. Nucl. Chem., 1961, 16, 323. Claassen, A., and Visser, J., Red. Trav. Chim. Pays-Bas, 1946, 65, 211; Chem. Abstr., 1946,40, 7059l. 12. 13. 14. 15. 16. 17. 18. 19. Kamata, E., Nakashima, R., and Shibata, S . , Bunseki Kagaku, 1984, 33, 173. Winge, R. K., Peterson, V. J., and Fassel, V. A., Appl. Spectrosc., 1979, 33, 206. Hagemann, F., J. Am. Chem. Soc., 1950, 72,768. Khopkar, S. M., and De, A. K., Analyst, 1960, 85, 376. Bri-ntzinger, H., and Hesse, G., 2. Anorg. Allg. Chem., 1942, 249, 113. Hara, R., and West, P. W., Anal. Chim. Actu, 1955, 12, 285. Lassner, E., and Scharf, R., Fresenius 2. Anal. Chem., 1958, 164,398. Page, A. G., Godbole, S. V., Madraswala, K. H., Kulkarni, M. J., Mallapurkar, V. S., and Joshi, B. D., Anal. Lett., 1983, 16, 1005. Paper J6l92 Received September 30th) 1986 Accepted December 15th) 1986

ISSN:0267-9477    

DOI:10.1039/JA9870200321

出版商:RSC    

年代:1987

数据来源: RSC