摘要:

Quantitative Analysis for Impurities in Uranium by Laser Ablation Inductively Coupled Plasma Mass Spectrometry: Improvements in the Experimental Setup† CHRISTOPHE LELOUPa , PIERRE MARTY*a , DIDIER DALL’AVAa AND MARCEL PERDEREAUb aL aboratoire de Chimie Analytique, Commissariat a` l’Energie Atomique, Centre de Valduc, 21120 IS sur T ille, France bL aboratoire de Recherche sur la Re�activite� des Solides, Universite� de Bourgogne, 21000 Dijon, France With the aim of reducing the volume of radioactive wastes to limit the interference problems and improve the sensitivity.It was necessary to study in detail and optimise the operation produced during the use of elemental analysis methods, combining LA with ICP-MS in a glove box has been studied. of the analytical unit, from the laser to the spectrometer, with a view to improving the reproducibility and repeatability in The ablation of solid uranium samples was achieved with an XeCl excimer laser (308 nm), using an ablation cell developed order to obtain a precise quantitative analysis.The success of this method depends on influencing the in this laboratory. After spatial filtering, the pulse-to-pulse repeatability (RSD) of the laser beam intensity was 1.2%. The principal parameters, in particular those which control the LA and aerosol transportation mechanisms. This paper presents, precision (RSD) achieved for the determination of impurities (Si, Cr, Fe, Co, Cu, Mo, Sn and Pb) contained in the uranium in a concise manner, the technological choices which have been made to adapt this installation to a glove box and the matrix was 1–3% for most elements, after optimising the ion lens settings of the spectrometer.This precision can be influence of essential parameters, such as the stability of the laser beam and the adjustment of the spectrometer. Finally, a improved for elements subjected to interferences, such as Fe and Si, by modifying the atmosphere in the glove box study on the influence of the atmosphere surrounding the plasma shows a way of improving the quantitative analysis of surrounding the plasma.The detection limits for the impurities studied were less than 0.1 mg g-1. elements subjected to interferences and which are dicult to distinguish such as Fe or Si. Keywords: L aser ablation; inductively coupled plasma mass spectrometry; glove box; uranium; quantitative analysis; excimer laser EXPERIMENTAL Instrumentation The production of high-purity actinide metals or actinide During the setting up of a coupled LA-ICP-MS method in a oxides is an important factor in the nuclear industry.1 During glove box, the main objective is to integrate the minimum the production process, the determination of impurities and amount of equipment into the glove box.The principle of the alloying elements in radioactive or toxic materials is often apparatus adapted to a glove box is illustrated schematically achieved using plasma source techniques,2–4 either ICP-AES in Fig. 1. or ICP-MS. In these two cases, the conventional method for introducing a sample into the plasma is the nebulization of a solution. Therefore prior to analysis the sample must be L aser dissolved. A comparison of the performances of the dierent types of Apart from the resulting analytical disadvantages (e.g., conlaser6 suitable for ablation for analytical purposes led to the tamination by reagents and loss of volatile elements), this selection of a UV laser.The mass ablated by a UV laser is methodology is often time consuming, especially when the larger than for an IR laser under similar operating conditions, work has to be carried out in a glove box. In addition, it the laser produced plasma is more reproducible, the spatial generates wastes which must be treated or stored. It is for resolution is better and the risk of selective vaporization is these reasons that a coupled LA-ICP-AES method has already been implemented in this laboratory.The integration of this system into a glove box has revealed its analytical potential and limits. On the one hand, the expected limitations with heavy element matrices such as uranium or plutonium (i.e., spectral interferences due to the numerous lines in ICP-AES spectra), and on other hand, the limitations of the measuring apparatus in terms of reproducibility (up to 15%) and detection limits (up to 50 mg g-1 for most elements) have been demonstrated. 5 These performances are not compatible with the quality control of nuclear products where a precision of a few % for minor and trace elements is required.The use of a combined LA-ICP-MS system was considered † Presented at the 1997 European Winter Conference on Plasma Fig. 1 Outline of a coupled LA-ICP-MS installed in a glove box. Spectrochemistry, Gent, Belgium, January 12–17, 1997. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (945–950) 945Table 1 Characteristics of the laser employed reflection treatment. Simple geometric optical principles should be used to estimate approximately the pit size: Wavelength 308 nm Energy 30–200 mJ Pit size=wdiaphragm× d2 d1 Frequency 0–100 Hz Pulse duration 17 ns Beam size 23×11 mm2 where d1 is the distance from the diaphragm to focusing lens Divergence 3×1 mrad 2, d2 the distance from focusing lens 2 to the sample. With such a long focal length, large pits are obtained (0.4 mm), limiting the possibility of spatially resolved analysis.However, minimized.7 Moreover, a UV laser probe for LA-ICP-MS this has the advantage of providing a sucient depth of field provides superior analytical performance.8 (in the mm range) to allow for the problems of demagnification The laser used is a pulsed XeCl (308 nm) excimer laser during ablation.11 Moreover, large pits are required to avoid (QUESTEK Model 2340). The principle performance characproblems with the heterogeneity of the sample.12 teristics are summarised in Table 1.The pulse-to-pulse energy is stabilized using an optical–electronic feedback circuit which monitors and regulates the energy of each pulse in real time. Description of the ablation cell The heart of a system to generate an aerosol by LA is located Spatial filtering at the level of the ablation cell. This was developed entirely in this laboratory (Fig. 3). Its principal features were based on Throughout its path, the laser beam is subjected to fluctuations installations described in the literature.6,13 However, some in intensity.They originate from the scattering of light waves specific characteristics such as a multiposition carriage have by imperfections on the optical surfaces, impurities in the gas been implemented in order to simplify the glove box handling. mixture or airborne particles. The ablation cell consists of two main parts. These phenomena are partly responsible for the poor repeatability (RSD=4–6%) in the pulse-to-pulse energy delivered Mechanical section. The mobile mechanical section, in aluby the excimer laser that was used for this study,9,10 in spite minium, allows for the positioning of the sample.Samples of of the real time regulation. variable size (maximum diameter 20 mm, thickness 10 mm) are Being of a dierent nature, the random fluctuations of the located in a reproducible manner in a sample holder. A mobile beam can be separated at the focal plane of a lens.By placing carriage holding five sample holders fits, also in a reproducible a screen pierced by a hole in the axis of the beam, the scattered manner, underneath the glass cell. The major part of the argon radiation can be blocked while the ‘clean’ part of the beam flow (Arc) arrives from underneath the sample. This mechanical passes through. The greater the focal length of the lens and system provides excellent reproducibility in positioning, ensurthe smaller the aperture, the better the selectivity will be.ing that experiments can be conducted with a constant focusing However, the diameter of the hole should be selected so as to lens to sample distance and with minimum handling. This give the minimum feasible energy loss. Also, if the screen has assembly, consisting of the carriage and sample holder, cao absorb too much of the laser beam energy, this may lead displaced by an intermediary of two micrometric translation to its progressive destruction by ablation. stages (UT100, Micro-Controle, Evry, France).In the apparatus used (Fig. 2), lens F1 is plano-convex, with Glass section. The other section in glass is fixed and ensures a focal length of 1000 mm. The aperture consists of a thin steel the ablated matter is carried away towards the plasma torch. or alumina plate (thickness 1 mm) pierced by a 2 mm diameter The connection between mechanical and glass sections is hole.The beam, thus filtered, is directed towards the optical facilitated by rubber bellows. The available volume is approxi- apparatus incorporated into the glove box. mately 400 cm3. This glass cell supports a fused silica window (UV quality), with an anti-reflection treatment. Another por- Optical apparatus tion of the argon flow (Arv) is directed along the surface of this window. Most of the components (optical, ablation cell and translation All gas flows are controlled with mass flowmeters.The stages) are fixed on a ‘breadboard’ and are incorporated into transfer of the argon which carries the aerosol to the torch, is a stainless steel glove box. ensured by a 2.5 m long polyethylene flexible tubing (id= The laser beam, described in the previous paragraph, passes 4 mm). A three-way valve allows purging of the cell while the through an optical window to enter the glove box and is sample is changed without stopping the plasma.This purge deflected by a mirror before being focused on the sample. In avoids air entrainment in the plasma, which would damage this way the laser spot is the image of the aperture responsible the torch. for the spatial filtering. The lens (F2=150 mm) used to focus the beam is of fused silica, with an anti-reflection treatment. All the optical components (mirror and windows) in the apparatus have an anti- Fig. 3 Schematic diagram of the ablation cell. Fig. 2 Principal of spatial filtering. 946 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Table 3 Influence of the tuning of lenses on the ICP-MS signal for Inductively coupled plasma mass spectrometer laser ablated particles [RSD(%), five replicates]. Sample is for The inductively coupled plasma mass spectrometer used was ‘eglantine’ an instrument supplied by VG, Elemental–Fisons (Winsford, After tuning with After tuning with Cheshire, UK) (Model PQ2+, 1991), in a nuclear version.Isotope wet plasma dry plasma Nuclear versions have already been described in the literature. 2–4 The electronic section was completely removed from 27Al 22 000 (8.7) 285 000 (5.1) the rest of the instrument. Only the torch box, the expansion 58Ni 10 000 (5.9) 117 400 (3.7) chamber and the slide valve are contained within the stainless 90Zr 12 100 (6.2) 120 900 (2.0) 234U 60 000 (7.2) 245 000 (1.5) steel glove box. High eciency filters (HEPA) isolate the fluid circuits (water, argon and pumping arrangement).A supplementary pump (S-option), with an adjustable flow, allows, on the one hand, the loss in pumping eciency caused by the filters to be compensated for, and on the other hand, a point, the following experiment was carried out. After allowing lower pressure in the expansion chamber thus improving the a suitable time (about 30 min) for the apparatus to warm up, analytical performances.14 the sampling position and the voltages applied to the lenses were optimised using a 100 mg l-1 multi-element solution (SPEX, ICPMS2, Longjumeau, France).Normally, such an RESULTS AND DISCUSSION optimisation produces signals of the order of 200 000 counts Because of the performance that can be attained with ICP-MS, per s-1 for the entire m/z range, and more precisely for isotopes accurate quantitative analysis can now be achieved. The such as 27Al, 115In or 238U. ICP-MS instruments satisfy well-established criteria defining With these adjustments, measurements of the signal proshort and medium term stabilities, especially for the introducduced by an aerosol created by LA of a solid uranium sample tion of samples by liquid means1 (of the order of a few %).were made (Table 3). Laser ablation, when used as a source for the introduction of The same data acquisition was then repeated after optimising the sample, should therefore satisfy the same stability requirethe sample cone position, the carrier gas flow and the voltages ments, at least in the short term.This is why it is important applied to the lenses, using a dry aerosol produced by LA of to improve the stability of the processes taking place during the same sample. The measurements showed that the resultant LA, and first of all to the signal delivered by the laser. Spatial gain in sensitivity reaches a factor of ten along with an filtering is an initial solution to this problem. improved stability. The craters obtained using a filtered laser beam are circular and of 0.4 mm diameter.Their size is only dependent on the Analytical performance optical arrangement. Taking into account the value of the energy delivered by the laser, the power density is 5 GW cm-2. The analytical performance for LA-ICP-MS was established One direct result of processing the beam is improved definition, using uranium reference materials (Se� rie des Floralies, and the reproducibility of the observed craters, implying a CETAMA, Marcoule, France), the impurity contents of which homogeneous distribution of the power density.12 are listed in Table 4.Another result is the improvement obtained in the stability The composition of the principal uranium isotopes in these of the energy that is delivered. Measurements (Table 2) were standards is 99.3% for 238U, 0.7% for 235U and 0.055% for made using a pyroelectric joulemeter (ED-200, Gentec, Palo 234U. The 234U signal was monitored in order to evaluate the Alto, CA, USA), in two positions.The first one is made 30 cm stability and reproducibility of the measurements on the uranafter lens F1, before the spatial filter, and the second one just ium reference materials. The measurements taken (using the after lens F2. operating conditions given in Table 5) gave the results indicated The energy loss suered during the processing of the beam in Table 6. The signal for 234U, given by ane�mone (the purest is 70%. The measurements reveal a factor of five gain in the material ), was used for normalization.pulse-to-pulse energy repeatability, between the two positions. With an unprocessed laser beam the intensity of the matrix This contributes to an improvement in the stability of the signal from one sample to the next varied by a maximum of power density delivered and ablated mass at each pulse, and 7%. The typical relative variation after five measurements, finally, to the stability of the analytical signal produced by the lasting 30 s each, was around 5%.With the benefit of spatial spectrometer. filtering, the variation in the recorded signal intensity from one sample to the next was less than 3%. The repeatability was improved considerably, reaching 1–2%, values normally Adjustment of ICP-MS parameters achieved in a liquid environment. The use of LA as a source for introducing the sample into the The signal for 234U will also be used as an internal standard plasma imposes modifications to the characteristics of the in all follow up work.plasma (temperature, energy, etc.). The change from a wet plasma, where the introduction of the sample is achieved in Calibration curves the form of an aerosol produced from a liquid, to a dry plasma requires adjustments to the instrumental settings. The sample The calibration curves were established using the five uranium cone position in the plasma, the carrier gas flow and the lens standards (Fig. 4). Owing to the choice of scale, points for Al voltage are the most important. In order to illustrate this (825 mg g-1) and Fe (400 and 480 mg g-1) are not indicated on the curves, but were included in the calculation. All the correlation coecients were better than 0.99, including those Table 2 Influence of spatial filtering on the laser energy (1000 shots for Fe and Si, both of which aicult to determine in a per 5 Hz liquid environment. Parameter Before filtering After filtering The comparison (Table 7) between the performance achieved with an unprocessed beam and a filtered beam illustrates that Delivered energy/mJ 100 30 the sensitivity can be increased by up to a factor of four.At RSD(%) 5.7 1.2 the same time, the repeatability reaches 2–3%, for concen- Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 947Table 4 Concentration (mg g-1) of impurities in the uranium reference materials Reference material B Al Si Cr Mn Fe Ni Co Cu Mo Ag Sn Pb Ane�mone <0.2 15 10 3.5 3.5 25 7 <1 2 <1 <0.5 2 <1 Capucine 0.55 40 22 32 25 140 190 4 55 20 0.5 4 5 Eglantine 0.25 120 13 15 4 38 50 15 7 50 1.2 20 15 Geranium 2.15 210 230 115 120 480 320 55 100 210 10 50 23 Ulmaire 0.8 825 195 50 52 400 85 25 54 215 – 200 – Table 5 Operating parameters Excimer laser— Energy/mJ 100 Repetition rate/Hz 5 (fixed point) ICP-MS— Rf/reflected power/W 1350/<5 Gas flow rate/l min-1 Outer 14.00 Intermediate 0.78 Aerosol carrier 0.750 Total ablation cell 1.0 Arc 0.6 Arv 0.4 Ion optics settings (potentiometer turns)— Dry Wet 450000 400000 350000 300000 250000 200000 150000 100000 50000 0 50 100 150 200 250 Concentration/mg g–1 Intensity (counts s–1) r = 0.996 r = 1 Al Fe Cu Mo (480 mg–1/g) (825 mg–1/g) r = 0.996 r = 0.992 (480 µg g–1) (825 µg g–1) plasma plasma Fig. 4 Calibration of some impurities in uranium. Extractor 1.9 1.9 Collector 7.0 6.3 trations less than 25 mg g-1. Limits of detection are improved L1 6.7 7.1 L2 5.7 6.2 in the same manner as the sensitivity.L3 6.6 5.2 Limits of detection are given by the following classical L4 2.1 2.4 equation: Collector type Pulse counting LOD= 3sbl×C Ic-Ibl Dwell time 10.24 ms Acquisition time 30 s Runs 5 where C is the concentration of the analyte, sbl the standard deviation on the blank (argon flow alone), Ic the count rate for the analyte and Ibl the count rate for the blank. Table 6 Influence of spatial filtering on the matrix signal, for 234U This result is identical to, or even better than, that which is normally observed in a liquid environment.Without spatial filtering With spatial filtering The measured repeatabilities for Fe (4.0%) and Si (7.0%), in spite of their significant concentrations, show the limitations Reference Signal Signal material (arbitrary units) RSD (%) (arbitrary units) RSD (%) of this type of quantitative analysis in producing an accurate determination for these elements. Ane�mone 100 5.0 100 1.6 Capucine 105 3.5 101 1.5 Eglantine 99 4.5 102 2.1 Influence of the atmosphere surrounding the plasma Ge�ranium 97 4.3 100 2.3 Ulmaire 102 6.2 99 2.0 In a liquid environment, quantitative analysis for elements such as Fe or Si is rendered very dicult, or even impossible, Table 7 Analytical performance with and without spatial filtering for trace measurements. The blank used for LOD calculations is the signal of the argon flow alone Sensitivity/ Limit of detection (3s)/ Repeatability counts per s-1 ppm mg g-1 [RSD (%), n=5] Concentration/ Isotope mg g-1 Without With Without With Without With 27Al 15 800 2018 0.03 0.02 26.0 3.4 28Si 10 175 426 0.61 0.36 7.5 7.0 52Cr 3.5 856 2044 0.03 0.02 3.1 3.3 55Mn 3.5 1060 2577 0.01 0.01 8.3 2.9 56Fe 25 806 2220 0.03 0.02 16.0 4.0 59Co 4 870 2224 0.02 0.01 6.1 2.9 58Ni 7 412 957 0.06 0.04 7.0 1.6 63Cu 2 370 829 0.04 0.03 8.6 3.5 98Mo 20 276 555 0.03 0.01 3.1 2.0 107Ag 0.5 780 2217 0.01 0.01 11.3 4.7 120Sn 4 550 1376 0.03 0.01 11.0 3.8 208Pb 5 2741 2545 0.01 0.01 15.0 2.9 948 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12owing to interferences from oxygen (40Ar16O+/56Fe+) and to oxygen or nitrogen, and in order to choose the neutral gas, the atmosphere in the glove box containing the ICP was nitrogen (14N2+/28Si+) originating from dilute nitric acid solutions. These same interferences exist with a dry aerosol, but to modified. Two series of trials were carried out by replacing the air in the glove box, with argon or nitrogen.a lesser degree. They originate from the traces of nitrogen, oxygen and carbon dioxide present in the argon, or in the To estimate the time for a set fraction of gas in the atmosphere to be replaced, the following relationship was sample, but also in the air surrounding the plasma and that which is drawn in by it. Some workers have developed accessor- used:16 ies to reduce these interferences15 and the resulting instabilities.The atmosphere in a glove box, containing oxidisable actin- Pi Pf =eDt/V ides such as uranium or plutonium, has to be free from oxygen, for safety reasons. Inert gases, such as argon or nitrogen are used. The flow of neutral gas varies between 0 (static mode) where Pi and Pf represent the initial and final amounts of the gas in the atmosphere, V (m3) the volume in the glove box, D and 3 renewals of the glove box atmosphere per hour (45 l min-1, dynamic mode).(m3 h-1) the ventilation rate and t (h) the time. In order to decrease the experiment time, the inward flow rate of gas into To show the influence of the air on the interferences related Fig. 5 Influence of a change in a glove box atmosphere (air–argon) on the blank signal. A, 15N; B, 23Na; C, 30Si/NO; D, 54Fe/ArN/Cr; E, 17O; F, 28Si/N2; G, 44Ca/N2O/CO2; and H, 56Fe/ArO. Fig. 6 Influence of a change in a glove box atmosphere (air–nitrogen) on the blank signal.A, 15N; B, 23Na; C, 30Si/NO; D, 54Fe/ArN/Cr; E, 17O; F, 28Si/N2; G, 44Ca/N2O/CO2; and H, 56Fe/ArO. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 949the glove box was fixed at 100 l min-1, whereas the volume of plutonium containing materials, where the advantage in the reduced volume of liquid radioactive wastes required, in par- the glove box was approximately 0.9 m3. Therefore it took 20 min to replace 90% of the initial atmosphere and 42 min ticular, will be appreciated.Moreover, the analysis time is markedly reduced, which will help minimise the exposure time for 99%. The blank signal (argon alone with no sample) was moni- to radiation hazards during the analysis of samples with potentially high radioactivity. tored during the periods corresponding to the change in atmosphere. Most of the peaks aected by the elements nitrogen and oxygen were monitored, as well as the sodium signal, We particularly thank the Laboratoire de Spectroscopie Laser always present in a non-negligible amount during the blank Analytique at CEA-Saclay, directed by P.Mauchien, for help trials. with setting up the spatial filtering. This work was supported On changing to an argon atmosphere (Fig. 5), all the signals by the Conseil Re�gional de Bourgogne. of the species subjected to interferences decreased when the presence of oxygen and nitrogen surrounding the plasma was REFERENCES reduced, by 40% for NO+ and 15% for ArO+.The sodium signal (m/z=23), not subject to interference, remained stable. 1 Crain, J. S., and Gallimore, D. L., J. Anal. At. Spectrom., 1992, The signals returned to their initial values on changing back 7, 605. 2 Pilon, F., Lorthioir, S., Birolleau, J.-C., and Lafontan, S., J. Anal. to an air atmosphere. At. Spectrom., 1996, 11, 759. The influence of changing the atmosphere from air to 3 Garcia Alonso, J. I., Thoby-Schultzendor, D., Giovanonne, B., nitrogen is illustrated in Fig. 6.The signals influenced by the Koch, L., Wiesmann, H., J. Anal. At. Spectrom., 1993, 8, 673. presence of oxygen decreased with its disappearance from the 4 Kopajtic, Z., Ro� llin, S., Wernli, B., Hochstrasser, C., atmosphere, whereas those dependent on nitrogen increased Ledergerber, G., and Jurcek, P., J. Anal. At. Spectrom., 1995, as the atmosphere became rich in nitrogen. The signal for 10, 947. 5 Masseau, S., Dall’ava, D., and Bergey, C., Analusis, 1994, 22, 445.m/z=44 increased, which shows that the interference from 6 Chartier, F, Ph.D. Thesis, Universite� Claude Bernard, Lyon, 1991. N2O is more important than that from CO2. 7 Geertsen, C., Briand, A., Chartier, F., Lacour, J.-L., Mauchien, P., During this acquisition a general tendency als Sjo� stro�m, S., and Mermet, J. M., J. Anal. At. Spectrom., 1994, 9, 17. to decrease was noticed, which is probably related to 8 Jeries, T. E., Perkins, W. T., and Pearce, N.J. G., Analyst, 1995, modifications in the characteristics of the plasma (energy, etc.). 120, 1365. These two examples illustrate that the composition of the 9 Chan, W. T., and Russo, R. E., Spectrochim. Acta. Part B, 1991, 46, 1471. atmosphere surrounding the plasma plays an important role 10 Pang, H., Wiederin, D. R., Houk, R. S., Yeung, E. S., Anal. Chem., in signals that are aected by the usual interferences. The 1991, 63, 390. reduction in the disturbances to the signal, arising from the 11 Arrowsmith, P., Anal. Chem., 1987, 59, 1437. various gaseous components in the atmosphere, should lead 12 Chale�ard, C., Mauchien, P., Andre, N., Uebbing, J., Lacour, J. L., to an improvement in the detection limits and in the stability and Geertsen, C., J. Anal. At. Spectrom., 1997, 12, 183. of elements such as Fe and Si.17 13 Arrowsmith, P., and Hughes, S. K., Appl. Spectrosc., 1988, 42, 1231. 14 Chiappini, R., Taillade, J.-M., Bre�bion, S., J. Anal. At. Spectrom., 1996, 11, 497. CONCLUSION 15 Longerich, paper presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January The results obtained confirm the potential of combining LA 12–17, 1997, paper number P 1–6. with ICP-MS, with a view to its application to accurate 16 Villermaux, J., Ge�nie De L a Re�action Chimique, T echnique et quantitative analysis for impurities in nuclear materials. At Documentation, Lavoisier, France, 1982. 17 Leloup, C., and Marty, P., unpublished data. present, the available linear calibrations, sensitivities, detection limits and stabilities allow analysis by LA to be achieved under conditions which are just as good as those found in a liquid Paper 7/01450C Received March 3, 1997 environment. the actual performance of the analytical installation presented here encourages the pursuit of studies on Accepted June 13, 1997 950 Journal of Analytical Atomic Spectrometry, September 1

ISSN:0267-9477    

DOI:10.1039/a701450c

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

X-ray Photoelectron Spectroscopic and Atomic Force Microscopic Studies of Pyrolytically Coated Graphite and Highly Oriented Pyrolytic Graphite Used for Electrothermal Vaporization† GA� BOR GALBA� CS*a, JA� NOS SNEIDERb , ALBERT OSZKO� c , FRANK VANHAECKEd AND LUC MOENSd aDepartment of Inorganic and Analytical Chemistry, Jo� zsef Attila University, H-6720 Szeged, Do�m te� r 7., Hungary bResearch Group for L aserphysics, Jo� zsef Attila University, H-6720 Szeged, Do�m te� r 9., Hungary cDepartment of Solid State- and Radiochemistry, Jo� zsef Attila University, H-6720 Szeged, Do�m te� r 7., Hungary dL aboratory of Analytical Chemistry, Ghent University, Institute for Nuclear Sciences, Proeftuinstraat 86, B-9000 Ghent, Belgium The interaction between solid or liquid samples on the one injection hole area and was explained by graphite redeposition.Habicht et al.12 used atomic force microscopy (AFM) for the hand and pyrolytically coated graphite or highly oriented pyrolytic graphite (HOPG) sample holders on the other hand first time to study the change of topography of PCG tube surfaces at the sub-micrometre level.A continuous roughening during electrothermal vaporization was studied. For the characterisation of the micrometer scale topographical changes of the surface was observed with the number of firings done and the size of dominant protrusions shifted towards 500 nm. occurring on these graphite surfaces as a result of solid sample evaporation, atomic force microscopy (AFM) was used.The Most recently, Vandervoort et al.13 applied scanning tunneling microscopy (STM) to elucidate the sub-micrometre defect struc- migration of Cd(NO3)2 and Na2HAsO4 deposited as solutions on the surface of the HOPG was studied by depth resolved tures on the graphite substrates used in ETAAS. The interaction of graphite surfaces in oxidation, carbide formation, etc. chemical X-ray photoelectron spectroscopy ( XPS) using argon ion sputter etching.It was found that the investigated compounds processes with the sample materials was surveyed by Huettner and Busche14 in general, and by Holcombe and Droessler15 migrate into the graphite to a depth of at least 1–1.5 mm. XPS data suggest that the migration involves either the hydrated specifically investigating the disturbance of Pb signals in ETAAS upon the exposure of graphite to oxygen. Both papers identified metal ions or the molecules. the lattice defects and graphite plane edges, typical of the Keywords: Atomic force microscopy; depth profiling; graphite materials used in furnaces, as the primary active sites electrothermal vaporization; highly oriented pyrolytic graphite; for surface chemical reactions.Volynsky10 recently reviewed the pyrolytically coated graphite; solid sample; X-ray chemical processes in graphite furnaces from the viewpoint of photoelectron spectroscopy catalysis. He pointed out that the presence of platinum metal compounds or transition metal carbides accelerates the reduction of inorganic compounds and the thermal destruction The analytical performance of atomic spectrometric techniques of organo-elemental compounds in the graphite furnace.Eloi applying electrothermal vaporization (ETV) or electrothermal et al.16 used Rutherford backscattering spectrometry to study atomisation (ETA) methods is well known to be influenced by the interaction of metals with a phosphate matrix modifier and the physical and chemical properties of the furnace materials observed the migration of the metals into PCG to a depth of used.Apart from the numerous publications on electrographite at least 3 mm. Jackson et al.11 studied the migration phenomenon (EG) uncoated and coated with pyrolytic graphite (PCG), for silver, cadmium and copper on highly oriented pyrolytic many investigations have been made on the use of other graphite (HOPG) flats and PCG platforms.In the above furnace materials, like glassy carbon (GC),1–5 totally pyrolytic mentioned investigations, as well as in other works in the graphite (TPG),5–8 graphite coated with metals5,9 and graphite literature, liquid sample evaporation/atomisation were studied modified with refractory carbides.5,10 on graphite surfaces. In solid sample graphite furnace analysis, The eect of the tube and/or sample holder properties in one can expect a more intensive interaction between the sample graphite furnaces on the analytical signal is caused by processes and the sample holder due to the complexity and relatively high such as direct reaction with sample components (carbide amount of the sample matrix present.This is supported by formation, intercalation, oxidation of graphite, etc.), sample previous observations17 made in connection with the quick diusion into the graphite and topological changes of the degradation of PCG sample holders used for the direct analysis surface. These processes have been most frequently investigated of certain solid samples.for ETA atomic absorption applications, and found to be In the current study, AFM and X-ray photoelectron spec- influencing the shape of signal profiles, appearance temperatroscopy (XPS) are applied to the investigation of the topograph- ture, sensitivity, memory eects, etc.3,4,6,11 ical and compositional changes occurring on PCG and HOPG The changes in morphology of PCG tube surfaces during surfaces as a result of interaction with solid samples.XPS depth ETAAS determinations were extensively documented by Ortner profiling is applied to the study of the migration of Cd and As and coworkers5,7,8 using scanning electron microscopy (SEM). deposited on the surface as solutions of their salts into HOPG. The PCG surface of both new and used tubes was described as ‘richly structured with nodules of typically 30–100 mm in height, displaying an onion-like structure when broken open’.The EXPERIMENTAL nodule formation on used tubes was observed especially in the Materials and Instrumentation Cadmium and As solutions were freshly prepared by diluting † Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997. a commercially available 1000 mg l-1 Cd stock solution and Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (951–955) 951dissolving Na2HAsO4·7H2O in diluted Suprapur quality nitric acid to give 300 mg l-1 solutions in a 0.15 M HNO3 matrix (all chemicals were obtained from E. Merck, Darmstadt, Germany).BCR CRM reference materials 142 Light Sandy Soil, 281 Rye Grass, 277 Estuarine Sediment and 146 Sewage Sludge of Industrial Origin were used as model solid samples (Community Bureau of Reference, Geel, Belgium). HOPG platforms of ZYH grade were obtained from Advanced Ceramics Corp. (Cleveland, OH, USA) in the size of 12×12×2 mm and were cut in half to give a size of approximately 12×6×2 mm compatible with the ETV furnace.The platforms were always newly cleaved, by freshly peeling o the surface layers using adhesive tape. Pyrolytically coated graphite sample boats were obtained from Ringsdorwerke Fig. 1 Plane corrected AFM image of the bottom area of a PCG GmbH. (Bonn, Germany) and were applied without any further sample boat after five firings. treatment. An SM-30 ETV furnace (Gru�n Analytische Mess-Systeme The impact of the evaporation of two solid samples of GmbH., Ehringshausen, Germany) with high purity argon as dierent origin (BCR CRM 142 Light Sandy Soil and BCR carrier gas was used in those experiments where high tempera- CRM 281 Rye Grass) on PCG boats was studied.Light Sandy ture sample treatment was needed. The temperature of the Soil was one of the materials for which the reaction with furnace was monitored pyrometrically (PY 20, Gru�n Optik, sample boats, resulting in significant sample boat erosion, was Wetzlar, Germany).The atomic force microscope was a formerly observed.17 In that work, it was established that the Topometrix Explorer 2000 (Topometrix Corp., Santa Clara, reaction occurred at vaporization temperatures 1550–1600 °C CA, USA) equipped with an atmospheric Si-nitride scanner. and above, so for this curren XPS measurements were made using a Kratos XSAM 800 1800 °C as the vaporization temperature.(Kratos Analytical, Manchester, Great Britain) with a Mg K a Fig. 2 depicts the typical topography that could be observed radiation source for depth profiling and a monochromatized after five firings of the Light Sandy Soil sample. The erosion, Al Ka radiation source for surface analysis. which could be detected with the naked eye, caused the topography to change dramatically. The highly eroded surface Procedure exhibits island-like protrusions that are 10–20 mm in length and several mm in height.These protrusions show some surface For the purpose of topographical experiments, a new empty substructure and are located on the side or bottom of large PCG sample boat was subjected to five firings in the ETV valley-like formations. The high corrugation of the surface was furnace at about 1800 °C. This boat served as topographical evidenced by the fact that scanning with the AFM instrument reference. PCG boats of the same batch were taken and heated was very dicult (the standard pyramidal tip used cannot (conventional heating programme, #1800 °C vaporization cope with slopes 45° and has about 14 mm height working temperature) with 0.5–1 mg of dierent solid samples.After range). Signs of the high corrugation are the unnaturally removing the remaining sample with a blast of air, the boats smooth regions and parallel line groups on the upper left and were loaded and fired again. This procedure was repeated five lower right area of Fig. 2, which appear on AFM images when times, in order to average the impact of individual sample the cantilever not the tip touches the surface. These findings loads. HOPG platforms were similarly used, with a vaporizsuggest that the surface height changes are at least 10–15 mm. ation temperature #1600 °C. After this treatment procedure, This means that the erosion caused by Light Sandy Soil can the sample holders were examined using AFM, performing easily reach typical values #50 mm thickness of pyrolytical many scans over the exposed areas.In the migration studies, coatings after 10–20 firings. We observed a similar degree of a 5 ml drop of As or Cd solution containing 1.5 mg analyte was erosion with river sediment reference material too. placed on a visually defect-free surface of HOPG platforms. There is no doubt that the described formations are the The platforms were then placed in a laboratory vacuum oven result of a chemical reaction between the graphite and a major and dried at 120 °C in a high purity argon atmosphere.Depth component of the sample, which is ground to micrometre sized profiling XPS measurements were performed by sputter etching particles. Depending on the nature of the surface reactions with a beam of 5 keV, 20 mA Ar+ ions rastered over an area of #10 mm2 on the HOPG and incident at an angle of 45° to the surface normal. After each bombarding session, the sample was repositioned into the measuring position and relevant parts of the XPS spectra for As, Cd, C and O were recorded.Sputter etching was continued until significant analyte signals could be detected. RESULTS AND DISCUSSION Topographical Studies PCG surfaces are known to be relatively rough and to contain protrusions of a size of several micrometres.5,7,8 Fig. 1 is an AFM picture taken of the bottom area of a new PCG sample holder boat after five firings (this plane-corrected picture, as well as other AFM images in this paper, represents many taken on dierent spots of the exposed areas).The surface Fig. 2 Plane corrected AFM image of the bottom area of a PCG contains irregularly shaped protrusions as well as relatively sample boat after five firings loaded with BCR CRM 142 Light Sandy flat sloping regions, but the overall vertical spread is quite Soil. The unnaturally smooth areas and groups of parallel lines are scanning artefacts, which reflect the high corrugation of the surface.uniform; 3–4 mm. 952 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12processes based on the small spread of AFM images. At the same time, cleaved HOPG flats, which are made totally of pyrolytic graphite and comprise graphite layers in a highly stacked and oriented manner, provide a well defined, almost atomically flat surface plane for such observations.13,19 These flats were also used for the evaporation of solid samples (BCR CRM 277 Estuarine Sediment and BCR CRM 146 Sewage Sludge of Industrial Origin), which were expected to behave as aggressively as Light Sandy Soil based on the above terms of composition and thermal decomposition.18 From Fig. 4 it can be seen that the Sewage Sludge impact only caused small, terraced erosion to the HOPG surface. Therefore, the 2–2.5 mm high formations, of highly protruding shape, are depositions. This is also supported by the XPS spectra, which showed the presence of SiO2 and TiO2 on the surface (Ti is present in a Fig. 3 Plane corrected AFM image of the bottom area of a PCG sample boat after five firings loaded with BCR CRM 281 Rye Grass. considerable, #19 mg g-1, amount in Sewage Sludge), while on the other hand, metal carbides could not be detected. Similarly a small extent of erosion and very lightly bonded occurring, the formations can be mainly graphite, if the depositions, which could be easily removed by soft handling oxidation of the graphite is dominating, or refractory com- (e.g., laboratory wipes), were also observed when Estuarine pounds when carbides, metal oxides, etc.are formed upon Sediment was heated on HOPG. The low anity of HOPG heating the sample. Based on the significant mass loss caused for sample materials correlates with the more perfectly graphitby the erosion to the sample boat,17 the oxidation of the ized condition and the high planar crystallite size of the surface, graphite is the main process in our work.This reaction can be which mean that fewer active sites (CMC bonds, edge carbons, initiated by (i ) the metal oxide content of the sample (e.g., fragments) are present to initiate surface reactions.20 Light Sandy Soil contains 0.28% iron oxide and 68.2% silicate) or (ii ) the CO2 released during the thermal decomposition of this sample as seen18 by thermogravimetry coupled with mass Migration Studies spectrometry (TG–MS) at about 1300–1400 °C, since this Migration of metals into pyrolytic graphite layers was recently temperature is high enough14 to allow the oxidation of carbon studied by Jackson et al.11 and Eloi et al.16 Determined by the by CO2. techniques applied (ETAAS and Rutherford backscattering The BCR CRM 281 plant sample behaved quite dierently, spectroscopy, respectively), these works could only focus on as shown in Fig. 3. It is apparent that the decomposition and the behaviour of metals in the migration process and almost evaporation of Rye Grass covered the surface with a large nothing could be established about the fate of the counter ions.number of small particles with a diameter of 0.5–1 mm and a XPS is a surface analytical technique, which is sensitive for height of 100–200 nm (the nature of AFM causes the particles to appear on the image as cones rather than spherical objects). These particles are believed to be almost exclusively carbon (soot), because of the highly reductive environment that is provided by the pyrolysis of plant samples.Apart from these small particles, the graphite surface morphology appears to be intact; major contours and the vertical spread (3–5 mm) have not changed significantly. The reaction paths, formerly suggested to be responsible for the erosion, conform with these findings; the CO2 (or H2O, CO) from the pyrolysis of plant samples is produced at much lower temperatures than those necessary to cause oxidation of the graphite,14 and the metal oxide and silicate content of Rye Grass is 10–100 times lower than that of Light Sandy Soil.From the richly structured surface of PCG, it is not easy to distinguish between negative (intrusion) and positive (protrusion) surface formations caused by erosion and deposition Fig. 5 XPS signals of (a) cadmium 3d3/2 and 3d5/2, and (b) oxygen 1s Fig. 4 Plane corrected AFM image of the exposed area of an HOPG flat after five firings loaded with BCR CRM 146 Sewage Sludge of for Cd(NO3)2 on HOPG graphite flat in the depths of: A, surface; B, 0.067 mm; C, 0.201 mm; D, 0.402 mm; E, 1.050 mm.Industrial Origin. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 953both light and heavy elements, and is also capable of depth it can explain why the migration depths were found to be similar for both compounds. resolved analyses by using sputter etching.21 In this latter instance an energetic Ar+ ion beam focused on the sample It is apparent from Fig. 5 and Fig. 6 that the metal and oxygen signals decay in the same way with the depth. The surface is used to etch o the surface layers, literally atom by atom. This technique was applied in studying the migration of finding that oxygen always accompanies the metals suggests that the migrating species are either hydrated metal ions Cd(NO3)2 and Na2HAsO4 into HOPG flats. Considering the nature of the sputtering process and the nanometre range (hydrated Cd+ and AsO4-) or maybe the whole molecule.In the case of arsenic this is also strongly supported by the fact sampling depth of XPS, PCG platforms were not tested owing to their relatively high surface roughness. The etching rate that the As5O quantity ratio in each depth was found to agree well with the 0.25 value of arsenate (mean and standard under the conditions described in the experimental section was calculated to be 0.2 nm s-1 based on the experimental results deviation from more than ten measurements: 0.27±0.04) and that the full width at half maximum (FWHM) of these peaks of Eklund et al.22 obtained for very similar conditions.Fig. 5(a) and Fig. 6(a) show the observed changes of metal did not change. On the other hand, the Cd5O ratio for cadmium nitrate was measured to be 0.53±0.17 (mean and signal peaks with depth. It is apparent that the metal signals could be traced into the HOPG to about 1 mm depth. However, standard deviation).The ion beam-induced reduction of metal oxide samples by preferential oxygen sputtering is a known the maximum migration depth can be estimated to be somewhat more than that for two reasons: the high vacuum side eect of the depth profiling XPS technique.21,24,25 This eect is described as more pronounced for heavy metals than (10-9 mbar) in the XPS sample chamber, which imposes a driving force on the migrating material towards the sides and for lighter ones. It is probable therefore, that a similar process occurred for cadmium nitrate, as also suggested by the increase top surface of HOPG and, therefore, also towards the etched measuring spot.When the HOPG remained in the vacuum of the FWHM of the oxygen peak profiles with the depth (with longer ion beam impact) (Fig. 6). Another feature shown chamber overnight (it took two days to record the curves of the figure because of the very slow sputtering speed) this eect in Fig. 5 and Fig. 6 is the 0.5–1.5 eV energy shift in peak positions towards lower binding energies with increasing depth. was probably observed: the metal signals normalised to carbon were found to be significantly higher than on the preceding This phenomenon is explained in the XPS literature by the final state quantum eect.26 evening. The other possible factor in action is the elevated temperature of the sample caused by the impact of the argon When trying to give an explanation for our observations, it has to be stated that migration through intact graphite planes, ion beam,21 which also works against the migration into the graphite (the eect of temperature on cadmium nitrate due to the 246 pm CMC bond length, is highly improbable, if not impossible.As HOPG is known to have a very low migration was recently also investigated by Majidi et al.23). Taking these two eects into account, the total migration permeability and to trap gases to a very low extent,20 it seems probable that the migration starts in solution form (before the depth can be estimated to be 1–1.5 mm.The concentration of As and Cd solutions was identical and the time the solutions drying of the solution) through imperfections (cracks, steps, etc.) occasionally present on the surface. However the mechan- spent on the surface before complete drying was similar. As these two factors should clearly influence the diusion process, ism is not completely clear yet; this initial process could be driven by an action similar to capillary action, as proposed by Jackson et al.11 After the drying, the further spread of the material can be imagined as mainly driven by the entropy, which should favour the a–b direction of the graphite and interplane lattice defects.The first author wishes to thank OMFB under No. 89805/96 for financial support and Yi Hu for his help in the topographical studies. REFERENCES 1 de Galan, L., de Loos-Vollebregt, M. T. C., and Oosterling, R.A. M., Analyst, 1983, 108, 138. 2 de Loos-Vollebregt, M. T. C., de Galan, L., van Uelen, J. W. M., Slavin, W., and Manning, D. C., Spectrochim. Acta, Part B, 1983, 38, 799. 3 Lynch, S., Sturgeon, R. E., Luong, Van T., and Littlejohn, D., J. Anal. At. Spectrom., 1990, 5, 311. 4 Schlemmer, G., and Welz, B., Fresenius’ J. Anal. Chem., 1986, 323, 703. 5 Ortner, H. M., Birzer, W., Welz, B., Schlemmer, G., Curtius, J. A., Wegscheider, W., and Sychra, V., Fresenius’ J.Anal. Chem., 1986, 323, 681. 6 Brown, A. A., and Lee, M., Fresenius’ J. Anal. Chem., 1986, 323, 697. 7 Welz, B., Schlemmer, G., and Ortner, H. M., Spectrochim. Acta, Part B, 1986, 41, 567. 8 Ortner, H. M., and Wilhartitz, P., Mikrochim. Acta, 1991, II, 177. 9 Marawi, I., Olson, L. K., Wang, J., and Caruso, J. A., J. Anal. At. Spectrom., 1995, 10, 7. 10 Volynsky, A. B., Spectrochim. Acta, Part B, 1996, 51, 1573. 11 Jackson, G. J., Fonseca, R. W., and Holcombe, J. A., Spectrochim.Acta, Part B, 1995, 50, 1837. 12 Habicht, J., Prohaska, Th., Friedbacher, G., Grasserbauer, M., Fig. 6 XPS signals of (a) arsenic 3p1/2 and 3p3/2, and (b) oxygen 1s and Ortner, O. M., Spectrochim. Acta, Part B, 1995, 50, 713. 13 Vandervoort, K. G., Butcher, D. J., Brittain, C. T., and Lewis, for Na2HAsO4 on HOPG graphite flat in the depths of: A, surface; B, 0.067 mm; C, 0.201 mm; D, 0.402 mm; E, 0.871 mm. B. B., Appl. Spectrosc., 1996, 50, 928. 954 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 1214 Huettner, W., and Busche, C., Fresenius’ J. Anal. Chem., 1986, 22 Eklund, E. A., Snyder, E. J., and Williams, R. S., Surf. Sci., 1993, 323, 674. 285, 157. 15 Holcombe, J. A., and Droessler, M. S., Fresenius’ J. Anal. Chem., 23 Majidi, V., Smith, R. G., Bossio, R. E., Pogue, R. T., and 1986, 323, 689. McMahon, M. W., Spectrochim. Acta, Part B, 1996, 51, 941. 16 Eloi, C., Robertson, J. D., and Majidi, V., J. Anal. At. 24 Berto� ti, I., To� th, A., Mohai, M., Kelly, R., and Marletta, G., T hin Spectrom.,1993, 8, 217. Solid Films, 1994, 241, 211. 17 Galba� cs, G., Vanhaecke, F., Moens, L., and Dams, R., Microchem. 25 Berto� ti, I., Kelly, R., Mohai, M., and To� th, A., Surf. Interface J., 1996, 54, 272. Anal., 1992, 19, 291. 18 Galba� cs, G., PhD. T hesis, Jo� zsef Attila University, Szeged, 1997. 26 Mason, M. G., Phys. Rev. B., 1983, 27, 748. 19 Reiss, G., Vancea, J., Wittman, H., Zweck, J., and Homan, H., J. Appl. Phys., 1990, 67, 1156. Paper 7/01979C 20 Ashida, K., Kanamori, K., and Watanabe, K., J. Vac. Sci. T echnol., ReceivedMarch 21, 1997 1988, A6, 2232. Accepted June 16, 1997 21 Homann, S., in Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, eds., Briggs, D., and Seah, M. P., Wiley, 1983, ch. 4, pp. 141–179. Journal of Analytical Atomic Spectrometry, September 199

ISSN:0267-9477    

DOI:10.1039/a701979c

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Determination of Metals in Airborne Particles by Electrothermal Vaporization Inductively Coupled Plasma Mass Spectrometry After Accumulation by Electrostatic Precipitation† BEAT A. BITTERLI*, HERVE� COUSIN AND BALAZS MAGYAR Inorganic Chemistry L aboratory, ETH Zu�rich, Universita� tstr. 6, 8092 Zu�rich, Switzerland A method to determine metals in airborne particles by with ETAAS.11–13 Recently, a technique has been proposed in which the suspended particles are collected in a graphite cup electrothermal vaporization ICP-MS after sampling by electrostatic precipitation into a graphite tube is described.A filled with graphite felt which is subsequently analysed by either ETV–ICP-MS or LEAFS.14 Earlier, the same group mobile sampler was used to collect samples. Calibration was performed with dried aerosol produced by the nebulization of a proposed sampling by drawing the air through the wall of a porous graphite tube, so that the particles are collected on the standard solution.For Cr, Fe, Mn, Cu, Zn, Sr, Cd, Sb, Ba and Pb, the absolute detection limits were in the picogram inner surface.15 Subsequently the tube acted as an electrothermal atomizer when applying FANES. range. The high sensitivity allowed the monitoring of several metals in ambient air with a time resolution of hours. The In this work, airborne particles were sampled by a mobile sampler using electrostatic precipitation into a graphite tube.16 precision was 10–15%.The addition of 5 mg g-1 of Na (as NaNO3) to the standard solutions improved the signal A graphite furnace was modified to be coupled to an ICP-MS system and to permit calibration by electrostatic precipitation intensities and reduced the curvature of the calibration curves. The influence of CHF3 was investigated but subsequently of a dried aerosol produced by nebulization of a standard solution. This calibration method was chosen because, in AAS, abandoned owing to a general increase not only of the signals but also of the background.the sensitivities of the calibration curves for liquid sample introduction are not the same as when calibrating using an Keywords: Electrostatic precipitation; electrothermal electrostatically precipitated aerosol.13,17 vaporization; inductively coupled plasma mass spectrometry; airborne particles EXPERIMENTAL The determination of metal concentrations in airborne particu- Instrumentation late matter permits the monitoring of anthropogenic pollution.A VG PlasmaQuad II+ (Fisons Instruments, Winsford, UK) The collection of suspended particles is usually performed by ICP-MS system was coupled to a modified Pye Unicam filter methods. For the elemental analysis of airborne particu- (Cambridge, UK) SP-9 graphite furnace. In order to obtain a lates, dierent methods have been used such as FAAS,1 electrodry aerosol, a Mistral aerosol drier (Fisons Instruments) was thermal (graphite furnace) AAS,2 XRF,3,4 PIXE,4,5 ICP-MS,6,7 used.The voltage for the electrostatic precipitation was sup- ICP-OES8 and instrumental neutron activation analysis.5 plied by an LHV 5 POS (Applied Kilovolts, Portslade, Sussex, Since filters normally have to be digested or extracted, there UK). The experimental conditions for all devices are given is often the problem of high background values due to the in Table 1. filter material.8 Therefore, sample amounts should be relatively For calibration, nitrogen at 1 l min-1 was supplied to the high (about 0.1 mg when Teflon filters are used)6 so that the aerosol drier since the conditions for electrostatic precipitation collection times used are usually in the range of days (for lowin nitrogen are similar to those in air but dierent to those in volume samplers).The digestion is a time-consuming step with argon. An additional nitrogen gas flow at 1 l min-1 joined the the possibility of the loss of some elements6 or the introduction primary nitrogen flow behind the aerosol drier to provide a of contamination.total gas flow rate of 2 l min-1, corresponding to the accumu- For short-term concentration monitoring (i.e., hours or lation of suspended particles in a real air sample (same gas minutes), multi-element methods with very low detection limits velocity in the precipitator). are needed. If the sampling is done into a graphite tube, ICP-MS with sample introduction by electrothermal vaporization (ETV–ICP-MS) fulfills these requirements.Owing to the Reagents transient signals in ETV, the multi-element capability of Working standard solutions were prepared from commercial ICP-MS cannot be fully exploited, but about 10 elements can single-element stock standard solutions [Ba, Fe, Zn, be measured per sample.9 The absolute detection limits of 1000 mg l-1 (Fluka, Buchs, Switzerland); Cd, Cu, Cr, Mn, Na, ETV–ICP-MS are in the picogram or sub-picogram range for Ni, Pb, Sn, Sr, Titrisol 1.000 g l-1 (Merck, Darmstadt, many elements.10 Germany); Sb, 1000 mg l-1 (Merck); and In, 1000 ppm Collection of the particles by electrostatically precipitating (Johnson Matthey, Royston, Hertfordshire, UK)].Dilution them into a graphite tube or on to a graphite rod eliminates was carried out with water (18 MV cm-1) obtained from a sample preparation procedures and thus minimizes the risk of Milli-Q water purification system (Millipore, Bedford, MA, contamination. This technique has been used in combination USA).Concentrated HNO3 of Suprapur grade (Merck) was used to provide a 0.1 M HNO3 matrix for all solutions. † Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997. Concentrations were chosen to give intensities similar to those Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (957–961) 957Table 1 Experimental parameters ICP-MS— Rf power 1350 W Outer argon flow rate 13 l min-1 Intermediate argon flow rate 0.8 l min-1 Carrier argon flow rate 1 l min-1 T ime-resolved acquisition parameters — Acquisition mode Peak jumping Dwell time 10240 ms Quadrupole settling time 1000 ms Number of isotopes determined per firing 10–15 Graphite furnace (all temperatures as given by the control unit; values in parentheses for liquid sample introduction) — 1st drying step 350 °C for 2 s (100 °C for 10 s) 2nd drying step, ramp 3 s (20 s) 475 °C for 5 s (200 °C for 1 s) Ashing step 700 °C for 10 s (700 °C for 5 s) Vaporization step 2500 °C for 3 s Cooling step 0 °C for 3 s Cleaning step 2500 °C for 3 s Inner diameter of graphite tube 5 mm Tube length from graphite furnace to torch 1.05 m Aerosol drier — Carrier gas flow rate (Ar, N2) 1 l min-1 Liquid uptake 1 ml min-1 Heating/cooling temperatures ~155 °C/-3 °C Electrostatic precipitation voltage — Calibration +2000 V Sampling +2500V Air sampling — Pump rate 2 l min-1 found in real samples.The concentrations of a typical standard sampling orifice (1.5 m above ground) was pointed downwards in order to minimize eects from changes in the wind direction. solution are given in Table 2. After the sampling procedure, the graphite tubes were stored in small plastic bags until they were measured. Prior to Sampling sampling, the graphite tubes were cleaned by heating and after cooling they were stored in new plastic bags. The amount of A specially designed mobile sampler was developed to meet sample collected on the discharge electrode was found to be the needs posed by the method (Fig. 1). Samples were usually less than 10%. Therefore, only the graphite tube was measured. collected for 1 h in front of the chemistry building. The Mathematical modelling (according to Brockmann)18 of the sampling eciency showed that the cut-o diameter (the Table 2 Typical mean concentrations in a synthetic standard solution (ng g-1) particle diameter where the sampling eciency of fine particles has dropped to 50%) is about 10 mm at wind speeds below Cr Fe Mn Ni Cu Zn Sr Cd Sb Ba Pb 1 m s-1 if the sampler inlet is pointed downward.The absolute 30 300 15 40 20 40 15 0.3 3.5 40 25 sample mass could not be determined with the available equipment since the amount of sample collected afterccumulation was estimated to be only 1–3 mg, assuming19 a fine dust concentration of 15 mg m-3 (a graphite tube weighs about 1 g).Apparatus for Calibration The graphite furnace was modified to be coupled to the ICP-MS system and to permit the insertion of the discharge electrode for the calibration (Fig. 2), while providing the opportunity to replace the graphite tube easily. Analytical Procedure Before measurements, the ICP-MS parameters (i.e., the ion lenses) were optimized for 115In using a dried aerosol produced by a Mistral aerosol drier. Subsequently the Ar carrier gas (1 l min-1) was switched to the graphite furnace, which was Fig. 1 Sampling head of the mobile sampler. The main body (1) consists of a polypropylene cylinder (5 cm long) with a compartment coupled to the torch. A sample tube could be inserted and (3 cm long) into which the graphite tube (2) fits. An aluminium plate measured. After two background measurements (empty tube), (3) with an orifice (1 cm long) ensures electrical contact between the the discharge electrode was inserted and two further graphite tube and ground (4) and holds the tube in place.The graphite background readings were acquired. tube can be easily changed by turning the plate aside after unscrewing By means of electrical valves, the graphite furnace was fed the left screw (5) and loosening the second screw (6). A pump (7) for with N2 gas loaded with the dried aerosol during accumulation sucking air is connected from one side, permitting the back wall to hold the discharge electrode tight (8).and with Ar to transport the sample from the furnace to the 958 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12standard solution. A first attempt by adding 10 mg g-1 of NH4NO3 (this compound is found in atmospheric aerosols19 and does not lead to interferences with the observed elements) to the standard solution did not show any changes. This compound is probably vaporized during the ashing step ( b.p. 210 °C). Since Na+ also occurs in large amounts in atmospheric aerosols in dierent compounds, 5 mg g-1 of Na+ (as NaNO3) was added to the standard solutions. The eect of the Na addition can be seen in Fig. 3 for Pb. All elements showed a general signal enhancement of about 50% which can Fig. 2 The graphite tube (1) is held in place by the (original) copper be attributed to a carrier eect.20 Furthermore, the variances clamps (2), which provide the electrical current for heating. At each decreased and the calibration curves became straighter side of the graphite tube a quartz tube (3) that is slightly narrowed at one end is inserted about 2 mm.The quartz tubes are held by in the low concentration range. A further increase in Na polypropylene adapters (4a, 4b) which are screwed to an aluminium concentration did not improve the signal any further. frame (5) and allow the tubes to be retracted a short way so that the The eect of mixing 0.1–0.2% v/v of CHF3 to the carrier graphite tube can be replaced. A spring mechanism (6) imposes gas flow was also examined since it had been shown to increase a slight pressure of the quartz tube against the graphite tube.A the signals for several elements.21 All elements examined polypropylene tube (7) at the right-hand adapter (4b) contains a showed an increase in signal intensity, but the background moveable Teflon cylinder (8). This cylinder holds the discharge electrode (9) (a tapered glassy carbon rod of 2 mm diameter) and allows also increased and many new interferences (due to molecular the electrode to be withdrawn when the graphite tube is replaced.The F and C compounds) were found. Additionally, cone wear was sample introduction hole in the graphite tube is closed by a pneumati- accelerated, so this method was not pursued further. A signifi- cally operated graphite cone (10). The left-hand adapter provides an cant enhancement of the signal-to-background ratio was found additional tube connection (cool gas) for optional splitting of the for Sr and Ba (approximately three and five times better, carrier gas.respectively). However, these elements are not the most important ones in air pollution, so the addition of CHF3 is not plasma. The accumulation was started by applying+2000 V recommended here. to the discharge electrode. After a defined time (30–120 s), the voltage was switched o and an Ar gas flow flushed the graphite tube for 20 s in the same direction as the N2 gas had Precision flowed before.At this time the graphite furnace heating cycle As shown in Table 3, the RSDs of repeated measurements are and the ICP time-resolved acquisition were started simulin the range 10–15%, averaging around 11% (n=12). In taneously. The calibration steps were repeated with either environmental analysis this order of magnitude is usually increasing accumulation time or with solutions containing acceptable. increasing metal concentrations (shortens the analysis time).After two background readings (empty graphite tube) and six calibration readings, the graphite tube was replaced and the Detection Limits whole procedure was repeated with the next sample. The The limits of detection (LOD) are given in Table 4 and are the calibration had to be repeated with each graphite tube since means of 10–15 values calculated from individual calibration they diered in sensitivity. curves (five graphite tubes with two or three calibration The signals were evaluated by integrating all peaks over the measurements) as 3 sblank/slope. Therefore, these LODs can same time interval of 2.5 s.The sample measurement was be reached regularly in practical work. The absolute detection corrected by subtraction of the mean of the first two backlimits for Pb and Cd are about the same as when measured ground measurements without the electrode, the calibration by ETAAS. measurements by subtraction of the two background measure- In Table 4, the detection limits are also given for a 1 h ments with the electrode.This method was chosen because the accumulation of ambient air. The high sensitivity of ETV– electrode imposes both a peak distortion due to sample condensation (on the electrode) and an increased background signal. In order to give the results in terms of concentration instead of signal intensity, the exact amount of aerosol delivered to the graphite furnace must be known. This was done by accumulating a standard solution of 2 mg g-1 of In for 15 min. The tube was then extracted for 2 h with 10 ml of 1 M HNO3 in a polyethylene bottle on an orbital shaker.Since the aerosol nebulization showed variances from day to day (the nebulization eciency was 0.5–2%), this extraction was repeated every day when real samples were measured. RESULTS AND DISCUSSION Modifier The calibration curves of all elements examined (Table 2) showed a non-linear response in the low concentration range (accumulation time shorter than 30 s).This eect is mainly due to the increased nucleation with increasing sample mass and therefore leads to less condensation of sample vapour on Fig. 3 Calibration curves for the electrostatic accumulation of Pb the tube walls.20 Since in real air samples the total sample (0.84 mg g-1). Top, without the addition of a Na modifier, and bottom, mass is much higher than by accumulation of a laboratory with 5 mg g-1 of Na. The bottom curve shows better linearity at low concentrations and less variances.produced aerosol, it is justified to add some element to the Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 959Table 3 Relative standard deviations for the peak areas for several isotopes from 12 consecutive accumulations (30 s each) Concentration Peak area RSD in solution (mean)/ (n=12) Isotope ng g-1 counts s-1 (%) 52Cr 9.9 87904 11.7 54Fe 805 594594 9.8 57Fe 805 414601 9.6 55Mn 19.2 225727 9.7 63Cu 14.4 50319 10.4 65Cu 14.4 172858 10.3 64Zn 97.2 24956 13.5 66Zn 97.2 105582 9.4 Fig. 4 Relative signal intensities of five consecutive blank firings 114Cd 2.2 10053 12.5 relative to a 30 s accumulation of the memory test solution (concen- 121Sb 1.6 18648 13.0 trations are given in the text). 123Sb 1.6 15569 13.9 206Pb 61.3 935240 12.7 208Pb 61.3 1886698 13.3 of the quartz tubes or of the carbon rod closing the sample introduction hole. Table 4 Limits of detection for several isotopes.The second row gives Accuracy the absolute LOD and the third gives the LOD for a 1 h accumulation (120 l of air) In order to test the method for its accuracy, the NIST standard reference material SRM 1648 (Urban Particulate Matter) was Limit of detection measured by introducing 10 ml of a 0.1M HNO3 suspension containing 95 mg g-1 of the standard (the suspension was kept For 1 h Element Absolute/ng accumulation/ng m-3 in an ultrasonic bath before the measurement).It should be noted that the sampling procedure is not identical with that 52Cr 0.16 1.3 for real samples or for a laboratory-generated aerosol. The 54Fe 0.74 6.1 55Mn 0.06 0.5 reference material would have to be digested first, resulting in 58Ni/60Ni 3.70/4.55 30.9/37.9 a loss of elements. Furthermore, by the introduction of the 63Cu/65Cu 0.26/0.27 2.1/2.3 suspended reference material a matrix comparable to real 64Zn 0.31 2.6 airborne particulate matter can be assumed.The reference 88Sr 0.09 0.7 material could not be inserted into the graphite tube as a solid 114Cd 0.02 0.2 sample since about 1 mg of sample would be required. 121Sb 0.01 0.07 138Ba 0.17 1.4 In Table 5 the concentrations determined for several 208Pb 0.01 0.09 elements are given. For Fe, Pb and Sb the results are in fairly good agreement with the certified values. Both Cu and Zn are overestimated by 50%, possibly owing to contamination from ICP-MS to Pb allows this element to be measured after an the electrodes. Because of the interference of 40Ar12C+, the accumulation time of only a few minutes.The high detection confidence limit for Cr is high (the concentration of Cr was limit for Ni is due to high background signals resulting from too low to allow the evaluation of 53Cr); Ni and Cd could not cone wear. be evaluated. Analysis of Real Samples Memory Eects Fig. 4 shows the magnitude of the memory eect. With an Fig. 5 shows the results of five consecutive accumulations of 1 h each in front of the chemistry building.Only signals accumulation time of 30 s (memory test solution containing 205 ng g-1 Fe, 36 ng g-1 Pb, 42 ng g-1 Mn, 38 ng g-1 Ba, (shaded bars) exceeding the calculated detection limits (white bars) are shown. The error bars depict an estimation of the 36 ng g-1 Pb, 31 ng g-1 Cu, 23 ng g-1 Zn, 19 ng g-1 Cr, 4 ng g-1 Cd, and 2 ng g-1 Sb) no significant memory eect error based on the standard deviation of two background measurements and the slope of the calibration (usually five was observed. With longer accumulation times a memory eect can be observed, mainly due to increased sample loading.points). Owing to dierences in tube sensitivities and background variations, the errors and the detection limits vary However, normally short accumulation times are sucient if the air has been sampled for 1 h. Chromium shows a sudden greatly (only two measurements were used for the estimate of the detection limit).The errors (calculated according to Miller drop after three blanks, probably due to a change in the heating conditions of the graphite tube (a lower release of and Miller)22 seem to be overestimated, since the concentrations are often fairly constant (e.g., Fe, Pb). By increasing carbon decreases the signal of 40Ar12C). The high background signal of 57Fe can be explained by polyatomic interference of the number of calibration measurements the error would become smaller, but the analysis time would increase. 40ArOH, which indicates a slight leakage due to wearing out Table 5 Analysis of SRM 1648. The determined values are shown with their 95% confidence limits (three replicates). Values in parentheses are not certified and are given for information only Determined Certified Determined/ Certified/ Element (% m/m) (% m/m) Element mg g-1 mg g-1 Fe 4.06±0.6 3.91±0.10 Cu 939±77 609±27 Pb 0.674±0.017 0.655±0.008 Cr 333±650 403±12 Zn 0.698±0.074 0.476±0.014 Sb 45.0±5 (45) Mn 0.12±0.01 (0.086) 960 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12ICP-MS can be used for measuring trace elements in air without any sample preparation. A mobile sampler can be used since the error produced when the sample amount deposited on the discharge electrode is neglected is within the precision and therefore is not significant. This method was used for monitoring concentration fluctuations of up to 12 elements simultaneously in the range of minutes or hours, depending on their concentration.For example, Pb could be detected after only a few minutes of accumulation. The measurement errors are relatively high since every sample can be measured only once with this sampler. The analytical procedure can be automated in order to improve the sample throughput. The field of application of this method would range from the workplace to environmental pollution monitoring (fixed and mobile stations). The authors thank the Stiftung zur Fo� rderung Schweizerischer Volkswirtschaft durch wissenschaftliche Forschung for financial support.REFERENCES 1 Pastuszka, J., and Hlawiczka, S., Atmos. Environ., Part B, 1993, 27, 39. 2 Low, P. S., and Hsu, G. J., Fresenius’ J. Anal. Chem., 1990, 337, 299. 3 Nicholson, K. W., and Branson, J. R., Atmos. Environ. Part B, 1993, 27, 265. 4 Eltayeb, M. A. H., Xhoer, C. F., Van Espen, P. J., Van Grieken, R. E., and Maenhaut, W., Atmos.Environ. Part B, 1993, 27, 67. 5 Maenhaut, W., Cornille, P., Pacyna, J. M., and Vitols, V., Atmos. Environ., 1989, 23, 2551. 6 Wang, C. F., Chen, W. H., Yang, M. H., and Chiang, P. C., 10.03 10.03 1.2 0.8 0.4 0 0 0.04 0.08 0.12 0 1.0 2.0 10.03 11.0612.09 13.1214.15 10.03 11.0612.09 13.1214.15 10.03 11.0612.09 13.1214.15 Starting time/h 121Sb 52Cr 114Cd [Metal]/ng m–3 3.0 Analyst, 1995, 120, 1681. Fig. 5 Concentrations (shaded bars) of various metals found in 7 Katoh, T., Akiyama, M., Ohtsuka, H., Nakamura, S., ambient air in front of the chemistry building for five consecutive Haraguchi, K., and Akatsuka, K., Fresenius’ J.Anal. Chem., 1995, hours (July 17, 1996). The estimated detection limit for each calibration 352, 577. is also shown (white bars). 8 Sugimae, A., and Barnes, R. M., Anal. Chem., 1986, 58, 785. 9 Berryman, N. G., and Probst, T. U., Fresenius’ J. Anal. Chem., In real air samples, Fe, Pb, Mn and Sb could always be 1996, 355, 783. 10 Sturgeon, R. E., Willie, S. N., Zheng, J., Kudo, A., and Gre�goire, detected and Ba and Sr nearly always. The high errors for Ba D. C., J. Anal. At. Spectrom., 1993, 8, 1053. can be explained by a high background signal of the empty 11 Torsi, G., and Bergamini, G., Ann. Chim., 1989, 79, 45. tube that appears after the sample signal, but is not completely 12 Sneddon, J., Anal. Chim. Acta, 1991, 245, 203. separated. This second peak is mainly due to contamination 13 Magyar, B., Cousin, H., and Yu, N., in 5. Colloquium or condensation on the discharge electrode, since it was greatly Atomspektrometrische Spurenanalytik, ed.Welz, B., Bodenseewerk reduced when measured without the electrode. For the quanti- Perkin-Elmer, U� berlingen, 1989, pp. 257–265. 14 Tilch, J., Lu�dke, C., and Homann, E., Fresenius’ J. Anal. Chem., tative measurement of Cr longer accumulation times are needed 1996, 355, 913. (high background due to the molecular interference of 15 Lu� dke, C., Homann, E., and Skole, J., J.Anal. At. Spectrom., 40Ar12C+). 1994, 9, 685. Investigations of calibration by pipetting liquid standard 16 Bitterli, B. A., PhD. T hesis, Federal Institute of Technology, solutions into the graphite tube led to the conclusion that the Zu� rich, 1997. calibration curves show the same sensitivity as when the 17 Torsi, G., and Palmisano, F., J. Anal. At. Spectrom., 1987, 2, 51. 18 Brockmann, J. E., in Aerosol Measurement: Principles, T echniques, standard solutions are precipitated electrostatically. This is not and Applications, ed. Willeke, K., and Baron, P. A., Van Nostrand the case when the electrostatic precipitation is used in combi- Reinhold, New York, 1993, ch. 6. naAAS, where the sensitivities of the precipitated 19 Hidy, G. M., Atmospheric Sulfur and Nitrogen Oxides: Eastern solutions are distinctly lower.16,17 This contradiction can be North American Source–Receptor Relationships, Academic Press, explained by the dierent sample type (dried aerosol versus San Diego, 1994, ch. 3. pp. 56–57. liquid drop) that leads to a dierent behaviour when atomized. 20 Ka�ntor, T., Spectrochim. Acta Part B, 1988, 43, 1299. 21 Wanner, B., Richner, P., and Magyar, B., Spectrochim. Acta Part This does not aect ICP-MS since the sample in the graphite B, 1996, 51, 817. tube only has to be vaporized and not atomized. 22 Miller, J. C., and Miller, J. N., Statistics for Analytical Chemistry, Owing to the nature of the present set-up, a lower precision Ellis Horwood, Chichester, 2nd edn., 1988, ch. 5. for liquid standard calibration was observed, which is the reason for calibrating with a dried aerosol. Paper 7/01451A Received March 3, 1997 Accepted April 23, 1997 CONCLUSIONS The collection of atmospheric particles by electrostatic precipitation in a graphite tube with subsequent analysis by ETV– Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 9

ISSN:0267-9477    

DOI:10.1039/a701451a

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Characterization of National Food Agency Shrimp and Plaice Reference Materials for Trace Elements and Arsenic Species by Atomic and Mass Spectrometric Techniques† ERIK H. LARSEN*a , GITTE A. PEDERSENa AND JAMES W. McLARENb aNational Food Agency of Denmark, Institute of Food Chemistry and Nutrition, 19 Mørkhøj Bygade, DK-2860 Søborg, Denmark. E-mail: EHL @L ST .MIN.DK bNational Research Council of Canada, Institute for NationalMeasurement Standards,Montreal Road, Ottawa, Ontario, Canada K1A OR9 The National Food Agency (NFA) of Denmark has produced the unknown samples and the reference material used, there is a risk that the results of the unknowns may be wrong.1 and characterized NFA Plaice and NFA Shrimp reference Therefore, analytical chemists need reference materials with a materials (RMs) for the control of the accuracy of trace wide range of matrices with certified contents of trace elements element and arsenic species determinations in similar seafood which closely match those of the unknown samples.With the samples. The physical preparation of the materials included increasing interest in elemental speciation there is an urgent dissection, drying, milling and sieving to collect the fraction of need for reference materials characterized for their content of particles less than 150 mm in size. In this fraction the trace elemental species of relevance to problems within, e.g., the elements were homogeneously distributed using a 400 mg toxicological, nutritional or environmental sciences.2 sample intake for analysis.The total trace element According to the International Organization for Standard- concentrations were determined by graphite furnace and cold ization the following definitions are given for reference mate- vapour atomic absorption spectrometry, inductively coupled rials.3 (i ) Reference Material (RM): A material or substance, plasma mass spectrometry (ICP-MS) and isotope dilution one or more properties of which are suciently well established ICP-MS.The contents of arsenobetaine and the to be used for the calibration of an apparatus, the assessment tetramethylarsonium ion were determined by cation exchange of a measurement method, or for assigning values to materials. high performance liquid chromatography (HPLC) coupled (ii ) Certified Reference Material (CRM): A reference material, with ICP-MS, or coupled with ion-spray (IS) tandem mass one or more of whose property values are certified by spectrometry (MS/MS) for qualitative verification. Based on a a technically valid procedure, accompanied by or traceable rigorous statistical analysis of the analytical data using the to a certificate or other documentation which is issued by a DANREF software, it was decided to assign certified values certifying body.for mercury, cadmium and arsenic in the NFA Shrimp, and When describing the preparation and characterization of mercury, selenium and arsenic in the NFA Plaice.Indicative RMs and CRMs, their selection, physical pretreatment, testing values were given for selenium, lead and chromium in the of homogeneity and stability as well as the assignment of NFA Shrimp, and arsenobetaine and the tetramethylarsonium certified properties are of interest to the end-user. The ion in both RMs. It is recommended that the certified mean Standards, Measurement and Testing Programme of The value and the standard deviation of the distribution of means European Commission has established a set of guidelines for are used to construct Shewhart control charts (x-charts) in the preparation of CRMs.4 However, the mathematical–statisti- order to evaluate the accuracy of a single as well as multiple cal methods of analysis and their use have not been described determinations of a certified value in the RMs.in any detail. Supplementary to these guidelines, a useful Keywords: Reference materials; trace elements; arsenic statistical software package has therefore been designed5 to species; atomic and mass spectrometry; certification aid the chemist in performing the statistical tests necessary prior to deciding on the assignment of certified properties (values) to an RM or CRM.In analytical chemistry the quality of a measurement is often The National Food Agency (NFA) of Denmark is a national described by its accuracy and precision. It is customary to reference laboratory for chemical contaminants in food and assess the accuracy of a result for a trace element in a biological the control and supervision of the quality of chemical measuresample by analysing an appropriate reference material with a ments in food is within the mandate of the NFA.Therefore matrix composition and concentration of the analyte which the two new marine RMs were developed to support the closely matches that of the sample. Reference materials may control of the accuracy of trace element analyses in marine additionally be used to monitor long-term changes in the food samples which are carried out to control the maximum performance of the analytical method used.In laboratory limits for lead, cadmium and mercury laid down in Danish practice, this may be illustrated by plotting results obtained national legislation,6 or which are carried out in the Danish for a certified component in a reference material in a suitable Monitoring System for Food.7 Additionally, there was a need control chart, or by demonstrating the equivalence of results for the assignment of values for arsenic species in the candidate obtained for independent methods of analysis.Even if the RMs because in internationally available CRMs such values laboratory finds a value in agreement with the certified prop- are not presently available. erty it should realize that, owing to dierences in composition, The analyst’s demand for a diverse selection of biological sample intake or trace element concentration level between RMs is not presently met in all cases by CRMs available from internationally recognized certifying bodies.Therefore there is a need for the production of additional RMs by other † Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997. organizations or institutes. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (963–968) 963This paper describes how the production, analytical–chemical characterization of the trace element and arsenic species content, statistical data evaluation and the certification were carried out for NFA Plaice and NFA Shrimp, two new seafood RMs.Furthermore, a procedure for the evaluation of the accuracy of a single as well as multiple results for a certified property is suggested. MATERIALS AND METHODS The two candidate marine biological materials were submitted to a series of physical treatments and chemical analyses according to the flow sheet given in Fig. 1. Origin and Identity of the Materials Fig. 2 Particle size distribution by mass (%) of the shrimp bulk Fresh specimens of plaice (Pleuronectes platessa), which orig- material. inated from the North Sea, were obtained from harbours on the west coast of Jutland, Denmark. After arrival in the NFA’s material passed the same screen. The coarse fractions of the regional laboratory in Aalborg, the fillets (i.e., muscle tissue) materials consisted of shell, leg and bone fragments which were dissected, taking care that bones or bone fragments were were not visible before sifting the materials.Following the not included. The fillets were stored frozen until used for analysis of the particle size distribution, the materials were further processing. passed through a mechanically operated large-scale 150 mm Dry powdered North Atlantic cold water shrimp was mesh stainless-steel screen. This screen, which was closest in obtained from a commercial source (Danish Freeze Drying mesh size to 125 mm, was selected to ensure that when using Ltd., Kirke Hyllinge, Denmark).The origin of the material the RMs, sub-samples of a few hundred mg could be taken was the North Atlantic, but the animal species was unknown. representatively,8 and at the same time to ensure a high yield At the time of acquiring, the material was stored in air-tight of the materials after sieving.The amount of each material laminate bags filled with nitrogen. which passed the sieve was collected directly in polyethylene bags to minimize contamination from the equipment. After sieving, the materials were placed in plastic containers and Physical Processing, Homogenization and Bottling mixed mechanically for 4 h. The frozen plaice fillets were divided into several sub-batches Prior to placing the materials in the selected brown glass which were freeze dried followed by milling using agate balls vials, they were cleaned by soaking in nitric acid and water in a rotating ceramic cylinder (Haltenwanger, Berlin, (1+7) for two days.The vials were then rinsed in water Germany). The shrimp powder, which was already in the form produced in a Millipore (Milford, MA, USA) Super Q apparof a dry powder, was not treated any further at this stage. The atus and finally air-dried. Following this cleaning procedure, milled sub-batches of the plaice and the bulk of the shrimp the possible migration of cadmium, chromium and nickel from material were combined in a baker’s dough mixer made from the vials including the plastic lids was tested by extraction for stainless steel.Following the milling and mixing processes, 24 h at room temperature into nitric acid and water (1+4). each of the materials was submitted to an analysis of the The measurements which were carried out by Zeeman-ETAAS particle size distribution using a mechanical sifting tower showed no increase above the blank value level of the three equipped with a stack of 10 screens of gradually decreasing elements in the acid–water mixture.The vials were then used mesh sizes. The distribution by mass of each fraction obtained as containers for the candidate RMs. Approximately 20 g of (Fig. 2) indicated that 80% of the shrimp bulk material passed the material were weighed into the vials which were closed by the 125 mm mesh screen.For the plaice, 93% of the bulk screw-top plastic lids. A membrane seal, which had to be torn o prior to the first time of use, ensured that the RMs were delivered unchanged in composition to the end-user. Homogeneity Testing The homogeneity of the distribution of trace elements in the candidate RMs was evaluated by randomly selecting 20 containers of each candidate RM for the estimation of inter-bottle variation which was compared with the within-bottle variation (n=20) using an F-test. The sample intake for the chemical analyses was 400 mg.Atomic absorption spectrometry (AAS) provided suciently good analytical precision to be used as the method of chemical analysis for this test. The results of the homogeneity testing are given in Table 1. Analytical Methods and Quality Control Atomic absorption spectrometry The candidate reference materials were wet ashed (0.4–0.6 g dry mass) in polytetrafluoroethylene (Teflon)-lined highpressure steel bombs, Model DAE II (Berghof GmbH, Fig. 1 Flow sheet showing the individual steps of the characterization of the NFA Plaice and NFA Shrimp RMs. Tu�bingen, Germany) at 160 °C for 4 h using 4 ml of sub-boiling 964 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Table 1 Homogeneity testing using the within-bottle variance (s2w) oven at 10% of full power for 5 min, at 20% power for 5 min and between-bottle variance (s2B) for nickel, chromium and mercury and at 25% power for 20 min.in the NFA Plaice and NFA Shrimp Reference Materials Variance/(mg kg-1)2×104 Arsenic species analysis The arsenic species were liberated from the candidate reference Sample s2w s2B F*= max {s2w,s2B} min {s2w,s2B} materials using four repeated ultrasound-assisted extractions Shrimp (n=20) into a methanol–chloroform-water system as described in detail Ni 2.56 4.84 1.89 elsewhere.11 The polar arsenic species were dissolved in the Cr 4.84 3.61 1.34 methanol–water phase whereas a minor part of the total Hg 0.12 0.20 1.69 arsenic, which was fat-soluble, was dissolved in the chloroform Plaice (n=20) phase.The arsenic species in the methanol–water phase were Ni 4.84 9.00 1.86 separated by anion exchange or cation exchange high perform- Cr 4.84 3.24 1.49 Hg 2.59 2.49 1.04 ance liquid chromatography (HPLC)11,17,18 using on-line detection of arsenic by ICP-MS at m/z 75.18,19 The identity of * Critical value for F19,19 (p=0.05) is 2.53.the separated cationic arsenic-containing molecular species was verified by coupling the cation exchange HPLC separation systems with ion spray (IS) tandem mass spectrometry distilled nitric acid. After dilution with water to 20 ml the (MS/MS) for detection of the molecular ions and the associated concentration of lead, cadmium, nickel, chromium, selenium species-characteristic collision-induced fragments.2,20 and arsenic was determined by Zeeman-ETAAS as detailed elsewhere,9–11 using platform atomization and an appropriate chemical modifier according to the manufacturer’s recommen- Quality control dations.12 Mercury was determined in the wet ashed residue Contamination of the samples, which was assessed by pro- by AAS using an alkaline solution of 3% sodium tetrahydrobocedural blanks, was kept to a minimum by using sub-boiling rate in water for reduction of mercury to the free metal in a distilled acids, purification of chemicals by solvent extraction MHS-20 (Perkin-Elmer, Norwalk, CT, USA) batch-type autoand by cleaning all utensils (pipette tips, autosampler vials mated reduction system.The elemental mercury was swept by etc.) in 2% v/v nitric acid. The manipulations of the samples argon from the reactor vessel to a quartz cell electrically heated during and after the wet ashing were kept to a minimum by at 200 °C positioned in the light path of the AAS instrument. taking the wet-ashed residue to volume with water in the tared The background-corrected (D2) absorbance was read at Teflon liners of the pressure bombs.10 In order to assess the 254 nm using a hollow cathode lamp as the light source.13 accuracy of the analytical work, one sub-sample of the following CRMs was analysed for every 10 analyses of the candidate Inductively coupled plasma mass spectrometry RMs.The NIES No. 6 Mussel (National Institute of Environmental Sciences, Tsukuba, Japan), and the NIST Inductively coupled plasma mass spectrometry (ICP-MS) Oyster Tissue 1566a (National Institute for Science and employing an ELAN 5000 instrument (SCIEX Perkin-Elmer, Technology, Gaithersburg, MD, USA) were used to assess the Concorde, Canada) in the quantitative multi-element mode accuracy of the analyses of the NFA Shrimp, and for the NFA was used for the determination of lead (208Pb), cadmium Plaice the dogfish tissue DORM-1 by NRCC (National (114Cd), nickel (58Ni or 60Ni), chromium (53Cr), arsenic (75As), Research Council of Canada, Ottawa, Canada) was used.If selenium (82Se) and mercury (202Hg) with the same sample the analytical results obtained for the CRMs were not in pretreatment as for the AAS methods. A radiofrequency power agreement with the certified values, the results obtained in the of 1000 W was used to sustain the argon plasma and the same run for the candidate RMs were not used. analyte signal intensities were measured using three readings of 35 ms dwell time per mass each and 40 sweeps per reading. To ensure high signal-to-noise ratios of the analytes and short Statistical Analysis wash-out times between samples, a cyclonic low-dead volume Five laboratories participated in the analytical work and four spray chamber was used in combination with a Meinhardor five sets of results were obtained for the total concentration type concentric glass nebulizer.The optimum gas flows and of each element and reference material (Table 2).Each set of other instrumental settings were as given elsewhere.14 To results comprised 4–10 individual values which were based on prevent memory eects of mercury arising from adsorption sub-sampling from randomly selected vials. The statistical occurring in the sample introduction system of the ICP-MS analysis of the sets of results was carried out using the instrument, 200 ng ml-1 of gold as the chride was added to DANREF software.5 The statistical analysis included tests for the analyte solutions as a matrix modifier,15 and calibration normality of distribution of mean values (Komolgrov-Smirnov- was carried out according to the method of standard additions.Lillifors test), outlying variances (Cochran’s test), outlying Isotope dilution (ID) ICP-MS was employed for the analysis mean values (Grubb’s test), homogeneity of variances of lead using 207Pb and 208Pb, cadmium using 111Cd and 114Cd, (Bartlett’s test) and a one-sided analysis of variance (ANOVA).chromium using 53Cr and 52Cr, nickel using 61Ni and 60Ni, selenium using 82Se and 78Se and mercury using 201Hg and 200Hg as spike and reference isotopes, respectively. The meas- RESULTS AND DISCUSSION ured signal intensities recorded at the above mentioned masses Preparation of the Candidate Reference Materials were corrected for isobaric overlaps, i.e., 114Sn on 114Cd and 82Kr on 82Se whereas other procedural details were as discussed The physical preparation of the candidate RMs involved several steps of fractionation and mixing aimed at obtaining in detail elsewhere.16 The spike isotopes were added prior to the ashing of 0.1 g (dry mass) of the biological materials which homogeneous biological powders.To test a possible inhomogeneity in distribution of the trace elements in the candidate was the maximum routinely used sample intake in the Teflon liners of the microwave-heated Teflon pressure vessels (CEM RMs, nickel and chromium were used as indicators of contamination from the steel equipment used.Mercury was selected Corporation, Matthews, NC, USA). The mineralization proceeded after addition of 2 ml nitric acid and 1 ml hydrofluoric for the test because of the potential risk of redistribution of this volatile element in the bulk of the materials during the acid. After closing, the bombs were heated using a microwave Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 965Table 2 Trace element contents in the NFA Shrimp and NFA Plaice RMs; concentration values are given as the mean±one SD Trace element concentration/mg kg-1 dry mass Laboratory 1 Laboratory 2 Laboratory 3 Laboratory 4 Laboratory 5 ZETAAS* ZETAAS* ZETAAS* ICP-MS ID–ICP-MS Element n=10 n=10 n=10 n=5 n=4 NFA Shrimp Mercury 0.132±0.005 0.125±0.004 0.140±0.009 0.140±0.012 0.142±0.015 Cadmium [0.072]†±[0.005]† 0.075±0.003 0.075±0.002 0.071±0.003 0.075±0.002 Selenium 0.977±0.023 0.957±0.015 1.26±0.082 1.17±0.06 n.a.Lead n.a.‡ 0.074±0.019 0.096±0.017 0.057±0.015 0.068±0.020 Chromium 0.157±0.020 0.127±0.026 0.142±0.040 0.476±0.046 [0.706]†±[0.082]† Arsenic 41.5±1.5 40.5±2.3 42.9±0.9 44.0±1.7 n.a. NFA Plaice Mercury 0.218±0.012 0.206±0.009 0.206±0.011 0.213±0.017 0.214±0.007 Selenium 2.98±0.03 3.28±0.14 3.39±0.15 3.60±0.15 2.93±0.04 Arsenic 41.9±2.2 41.0±0.6 46.0±1.1 44.0±1.7 n.a. * CVAAS was used for the mercury analyses. † Data not included because of outlying variance.‡ n.a.=not analysed. Table 3 Intercomparison statistics for NFA Shrimp RM Statistical property Mercury Cadmium Selenium Lead Chromium Arsenic Number of accepted sets of results 5 4 4 4 4 4 Number of accepted replicates 39 29 35 26 35 34 Normality of distribution of mean values (Y/N) Y Y Y Y Y Y Outlying variances None 1 None None 1 None Outlying mean values None None None None None None Homogeneity of variances (Y/N) N Y N Y Y Y Mean value/mg kg-1 0.136 0.073 1.09 0.074 0.226 42.2 SD of the distribution of lab.means/mg kg-1 0.007 0.002 0.15 0.018 0.172 1.6 95% confidence limits of the mean value/mg kg-1 0.127–0.145 0.070–0.077 0.85–1.33 0.046–0.102 Not given 39.8–44.7 Within-laboratory SD/mg kg-1 0.008 0.0026 0.051 0.018 0.033 1.69 Between-laboratory SD/mg kg-1 0.006 0.0016 0.16 0.016 0.137 1.31 Are all laboratory means equal (Y/N) Y N N N N N physical processing, or following microbial activity. The results Table 3 and Table 4. For each element and RM, the sets of results were tested for outlying variances and means.If an in Table 1 show, however, that the between-bottles variance was not significantly larger than the within-bottles variance. outlier was detected, which was the case for one set of results for cadmium and one for chromium in the NFA Shrimp, these Consequently, the three elements were not inhomogeneously distributed at the 400 mg sample intake level used. sets of data were not used for further statistical evaluations.Homogeneity of variances of the sets of data is a basic In contrast to other available CRMs, from which the fatsoluble matter has been removed by extraction using e.g., requirement which should be met prior to conducting a one-sided analysis of variance. Fortunately, the latter statis- acetone, the natural fatty constituents of the candidate RMs were not removed. Therefore, the analyst is confronted with tical analysis is quite robust against deviations from this the same potential problems and sources of error when analysing the RMs as when analysing the same trace elements or Table 4 Intercomparison statistics for NFA Plaice RM species in real samples. Mercury Selenium Arsenic Statistical property Contents of Total Trace Elements Number of accepted sets of 5 5 4 results Results are given in Table 2 for six elements in the NFA Number of accepted 39 39 35 Shrimp and for three elements in NFA Plaice RMs.Before replicates calculating the mean and standard deviation values reported Normality of distribution of Y Y Y in Table 2, the individual results were subjected to an outlier mean values (Y/N) test (Grubb’s test) by which a few values were rejected.Trace Outlying variances None None None Outlying mean values None None None elements which were below approximately five times the limit Homogeneity of variances Y N N of detection10 in concentration were not considered for (Y/N) certification because of the associated high relative standard Mean value/mg kg-1 0.211 3.24 43.2 deviation of the measurements.Because of the low concen- SD of the distribution of 0.005 0.28 2.3 tration level of cadmium, lead, nickel and chromium in the lab. means/mg kg-1 NFA Plaice in addition to the polyatomic interferences encoun- 95% confidence limits of 0.205–0.218 2.89–3.58 39.6–46.8 tered when using the ICP-MS-based methods (e.g., CaO or Fe the mean value/mg kg-1 on Ni) no data have been given for these elements.Within-laboratory 0.012 0.11 1.5 SD/mg kg-1 Between-laboratory 0.004 0.25 2.4 Statistical Analysis SD/mg kg-1 All laboratory means equal Y Y Y The results of the statistical analyses are given for the NFA (Y/N) Shrimp and NFA Plaice Reference Materials, respectively, in 966 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12requirement. Since dierent methods of chemical analysis The ANOVA shows that the laboratory means are not equal, and considerable dierences between the individual laboratory cannot be expected to produce identical variances, the statistical evaluation was continued disregarding the fact that, in means are seen (Table 2).A tentative rejection of the dataset with the highest average value did not improve the conclusion some cases, the variances were not homogeneous.4 Following this the mean of the means was calculated or, if all statistical of the ANOVA. In spite of the fact that no systematic error could be identified, the relatively large spread of the laboratory tests for all sets of data for a given element and RM were evaluated positively, the mean was calculated as the mean of mean values led to assignment only of an indicative value based on all four datasets for selenium in the NFA Shrimp all individual results.4,5 The standard deviation of the distribution of means or of the distribution of all individual results RM.For similar reasons, only an indicative value has been given for lead in the NFA Shrimp.is also given, and was used to calculate the 95% confidence limits of the mean value. The ANOVA test provided information on the within- and between-laboratories standard Concentration of Arsenic Species deviation. If these standard deviations were significantly dierent, one or more of the laboratory means were not equal to The concentrations of arsenobetaine (AsB) and tetramethylarthe rest. sonium ion (TMAs) given as the mean and standard deviation values of five replicate analyses were 41.3±4.0 mg g-1 and 0.120±0.013 mg g-1, respectively, in the NFA Plaice, and Assignment of Certified or Indicative Values for Total Trace 37.2±1.6 mg g-1 and 0.030±0.003 mg g-1, respectively, in Elements the NFA Shrimp (all concentrations calculated as arsenic).However, since only one laboratory carried out quantitative If all laboratory mean values were not equal, each set of data analysis using HPLC–ICP-MS and one laboratory used for a given element was carefully evaluated with respect to its HPLC–IS-MS–MS for qualitative verification, only indicative accuracy.Potential interferences and systematic errors were values could be assigned for AsB and TMAs. The accuracy of scrutinized in order to find the sources of the deviation. the results obtained for AsB was substantiated by determi- In the case of the results for chromium in the NFA Shrimp, nation of this arsenic species in the NRCC DORM-1, which a potential positive interference for the ICP-MS methods was analysed in parallel with the NFA Shrimp and NFA Plaice caused by the ArC+ polyatomic ion was identified.The RMs. The result obtained for AsB in the DORM-1, which was ID–ICP-MS results had already been disregarded on the 14.1±1.7 mg g-1 (as arsenic; mean±1 SD), did not deviate grounds of an outlying variance (Table 2) and the ArC+ significantly (p=0.01) from a reported,21 but non-certified interference caused the set of results based on ICP-MS to be value of 15.7±0.8 mg g-1.Furthermore, the extraction of rejected too. Furthermore, for the ICP-MS method, the result arsenic from the two candidate RMs was 99.4% complete as for the chromium analysis in the NIST Oyster CRM was evaluated by an analysis of total arsenic remaining after higher than the certified chromium value assigned, which extraction. Finally, the good accuracy of the speciation results provided an additional reason for the rejection of the ICP-MS was illustrated by comparing the total arsenic content with results.The remaining three sets of data which comprised only the sum of the determined concentrations of the two arsenic one method of analysis (Zeeman-ETAAS) did not fulfil the species plus the concentrations of other species or fractions general requirement that at least four sets of data produced determined earlier in the two RMs.11 This sum amounts to by at least two independent methods of analysis should be 39.1±1.7 mg g-1 for the NFA Shrimp, which is close to but used for certification.4 Therefore only the assignment of an slightly lower than the total arsenic value of 42.2±1.6 mg g-1 indicative value for chromium was possible.(Table 3). This balance, however, did not include all arsenic A second case is represented by cadmium in the NFA species known to be present in the NFA Shrimp (e.g., dimethyl- Shrimp where one set of data had already been rejected because arsinate).For the NFA Plaice the sum of the individual of outlying variance (Table 2). The one-sided ANOVA showed arsenic species and other arsenic fractions11 amounts to that the remaining four means were not equal. It was, however, 42.5±4.0 mg g-1 (calculated as arsenic), which coincides with not possible to trace any systematic error and it was decided the total arsenic content of 43.2±2.3 mg g-1 (Table 4). to assign a certified value based on the four sets of results, realizing that a small systematic error was included in the certified value.Notes on the Use of the Shrimp RM and Plaice RM The third case to be discussed is arsenic, also in the NFA Shrimp RM. Again the ANOVA showed that not all laboratory The certified values for total trace elements and indicative values for AsB and TMAs in the NFA Shrimp and NFA Plaice means were equal. None of the datasets could be rejected on the grounds of outlying variance, and no systematic errors RMs are given in Table 5 as the mean value±the half-width of the 95% confidence interval of the mean value.Other could be identified for any of the four sets of results. The accuracy of all sets of results was substantiated by the accurate standard deviations which may be of relevance to the practical use and evaluation of results obtained for the RMs are also analysis of arsenic in the DORM-1 CRM by all laboratories. Therefore all four sets of results which comprised two indepen- given in Tables 3 and 4.For every ten unknown samples which are analysed in dent methods of analysis were accepted, but the assigned certified value contained a minor systematic error. parallel by the same method of analysis it is recommended that additionally one determination of an RM of similar The fourth and final case is selenium in the NFA Shrimp. Table 5 Certified and indicative values (mg kg-1 dry mass) for trace elements and arsenic species in NFA Shrimp and NFA Plaice RMs; values are the mean±the half-width of the 95% confidence interval of the mean RM Hg Cd Se Pb Cr As AsB TMAs Certified/mg kg-1 NFA Shrimp 0.136±0.009 0.073±0.004 42.2±2.5 NFA Plaice 0.211±0.007 3.24±0.35 43.2±3.6 Indicative/mg kg-1 NFA Shrimp 1.1 0.07 0.14 37.2±1.0 0.030±0.002 NFA Plaice 41.3±2.5 0.120±0.008 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 967composition is included. This enables the analyst to evaluate in Copenhagen.Finally, we thank Brad Methven, National Research Council of Canada for skilful technical support for the accuracy of the results obtained for the unknowns by the results obtained for the RM. The analyst therefore needs to the ID–ICP-MS work and Dr. Peter Holm of the Institute of Pharmacy, Copenhagen for assistance with the equipment for evaluate if the single determination of the RM diers from the certified value by an amount greater than that expected due handling the large bulks of the powdered biological materials.to chance only. If such a deviation is identified it is likely that the similar measurement of the unknown samples is non- REFERENCES satisfactory. To evaluate the accuracy of a single determination, 1 Quevauviller, P., J. Anal. At. Spectrom., 1996, 11, 1225. it is recommended to use Shewhart control charts for individ- 2 Larsen, E. H., Spectrochim. Acta, Part B, 1997 (in the press). uals.22 The upper and lower control limits are constructed 3 Certification of Reference Materials—General and Statistical from the certified mean value and the standard deviations of Principles, ISO/IEC Guide 35–1989, International Organization the distribution of laboratory means which have been given in for Standardization, Geneva, Switzerland, 1985. Table 3 and Table 4 for the NFA Shrimp and NFA Plaice, 4 European Commission, Directorate General XII: Science, Research and Development, Standards, Measurement and Testing respectively.Any single measurement of the RMs must not Programme. Guidelines for the production and certification of BCR exceed the mean±3 SDs. reference materials. Doc. BCR/48/93, Brussels, Belgium, 1994. After a number of repeated measurements of a certified 5 Kristiansen, J., Christensen, J. M., Lillemark, L., Linde, S. A., property have been plotted in the Shewhart chart, it becomes Merry, J., Nyeland, B., and Petersen, O., Fresenius’ J. Anal. possible also to reveal systematic variations in the performance Chem., 1995, 352, 157.of the analytical method by applying a set of tests for assignable 6 Order on Maximum L imits for Certain Metals in Food, Ministry of Food, Agriculture and Fisheries, Order No. 447, September 5, causes of deviation.22 This enables the analyst to identify 1985, Copenhagen, Denmark. otherwise ‘hidden’ variation such as drift patterns and to seek 7 Food Monitoring, 1988–1992, National Food Agency of Denmark.its causes and finally correct them. Publication No. 232, ISBN 87-601-5787-9, 1995. 8 Maier, E. A., T rends Anal. Chem., 1991, 10, 340. 9 Larsen, E. H., and Ekelund, J., Analyst, 1989, 114, 915. CONCLUSIONS 10 Larsen, E. H., and Rasmussen, L., Z. L ebensm. Unters. Forsch., 1991, 192, 136. By selecting a limited number of experienced laboratories for 11 Larsen, E. H., Pritzl, G., and Hansen, S. H., J. Anal. At. Spectrom., the analytical work and by applying a strict protocol for the 1993, 8, 1075.statistical data analysis it is possible for an organization, which 12 Analytical Methods for Furnace Atomic Absorption Spectrometry, Perkin-Elmer Corporation, CT, USA, Publication B332, 1985. is not considered a certifying body, to produce two reference 13 Pedersen, G. A., Mortensen, G. K., and Larsen, E. H., Fd. Addit. materials with certified and indicative values for a number of Contamin., 1994, 11, 351. trace elements and arsenic species. During the course of the 14 Larsen, E. H., and Ludwigsen, M. B., J. Anal. At. Spectrom., 1997, analytical chemical characterization of these materials the use 12, 435. of CRMs was valuable for the control of the accuracy of the 15 Stu�rup, S., and Bu� chert, A., SAC 95, An International Symposium measurements. The trace elements, for which certified values on Analytical Chemistry, Poster A 1.4, University of Hull, England, July 1995. have been given, are of practical importance in Danish food 16 McLaren, J. W., Beauchemin, D., and Berman, S. S., Anal. Chem., control and research, and the materials are suciently well- 1987, 59, 610. characterized to be used in the place of certified reference 17 Larsen, E. H., Fresenius’ J. Anal. Chem., 1995, 352, 582. materials. Additionally, the indicative values given for AsB 18 Larsen, E. H., and Stu�rup, S., J. Anal. At. Spectrom., 1994, 9, 1099. and TMAs will be useful to the analyst who wants to begin 19 Hansen, S. H., Larsen, E. H., Pritzl, G., and Cornett, C., J. Anal. arsenic speciation analysis or to control the accuracy of such At. Spectrom., 1992, 7, 629. 20 Corr, J. J., and Larsen, E. H., J. Anal. At. Spectrom., 1996, 11, 1215. work already in progress. 21 Beauchemin, D., Bednas, M. E., Berman, S. S., McLaren, J. W., Siu, K. W. M., and Sturgeon, R. E., Anal. Chem., 1988, 60, 2209. We thank Jette Boyer, Marianne Hansen, Lisbet Pihlkjær, 22 Shewhart Control Charts, International Standard ISO 8258, 1991, Merete Ludwigsen and Birgitte Koch Herbst of the National International Organization for Standardization, Geneva, Food Agency for skilful technical assistance; Gerda Krog Switzerland. Mortensen of the Regional Laboratory in Aalborg for supervising the preparation and for the homogeneity testing as well as Paper 7/01530E the associated Zeeman-ETAAS work; Lis Rasmussen for super- Received March 4, 1997 Accepted May 19, 1997 vising the Zeeman-ETAAS work at the Regional Laboratory 968 Journal of Analytical Atomic Spectrometry, September 1997, V

ISSN:0267-9477    

DOI:10.1039/a701530e

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Determination of Uranium to Calcium and Strontium to Calcium Ratios in Corals by Inductively Coupled Plasma Mass Spectrometry† FLORENCE LE CORNEC* AND THIERRY CORREGE PAL EO-OCEAN Program, L aboratoire des Formations Superficielles, ORSTOM, 32 Avenue Henri Varagnat, 93143 Bondy Cedex, France A method has been developed to determine Ca (about 40% instrument to measure Ca, Sr and U precisely in the same aliquot. The main advantage of ICP-MS over TIMS is the m/m in corals), Sr (about 1–2%) and U (3–5 mg g-1) in the same aliquots.All the samples were spiked with 45Sc, 89Y and large sample output attainable with ICP-MS (between 50 and 100 samples per day). The present work also demonstrates 236U. None of these isotopes are present in significant amounts in corals. The Ca and Sr were determined using two dierent that the external isotope dilution method recently developed by Lea and Martin4 is not a viable option to measure Ca and approaches. Firstly, 45Sc and 89Y were used as spikes to measure 43Ca and 86Sr, respectively, by the isotope dilution Sr with good precision.technique (external isotope dilution). It was established that the behaviour of the various elements in the plasma varied EXPERIMENTAL from day to day and cannot be modelled with good accuracy (i.e., better than 1%). Therefore, a second approach was Instrumentation developed using 45Sc and 89Y as internal standards. This The ICP-MS instrument used in this study is a VARIAN method yielded extremely good reproducibility within a run UltraMass (Varian, Les Ulis, France).Samples are introduced and also on a day to day basis, providing the ICP-MS via a V-groove nebulizer and an inert Sturman–Masters spray instrument is properly tuned. Values for %RSD of 0.2–0.3% chamber. The sample introduction assembly is always at for Ca and Sr measurements were routinely obtained, as were constant room temperature because it is mounted outside the calibration curves with correlation coecients better than torch box.Thus, any possible drift eect caused by reflected 0.9998. The U was determined by classic isotope dilution. heat from the plasma compartment is eliminated. Ions are Analyses on a standard solution of coral powder over 5 d detected by a discrete dynode electron multiplier (DDEM) yielded %RSDs of 0.25% for Sr5Ca and 0.5% for U5Ca. detector. All parameters, such as xyz positioning of the plasma Accuracy was assessed using the certified reference material relative to the sampler cone, rf power, ion lens voltages and CCH-1 and a coral standard (NC20), which were both all plasma gas flows (Table 1), are computer-controlled.analysed by ICP-MS and ICP-AES. Keywords: Inductively coupled plasma mass spectrometry; Reagents calcium carbonate; palaeothermometry; calcium; strontium; uranium High purity de-ionised water (18.2 MV cm) was obtained from a Millipore Milli-Q185 Plus unit (Millipore, Villiers-le-bel, France).All working standard solutions were prepared by Corals secrete a skeleton of aragonite (CaCO3) in which some appropriate dilution of mono-elemental 1000 mg ml-1 certified trace elements such as Sr, U and Mg appear to be incorporated stock solutions (Spex, Longjumeau, France) in 2% nitric acid as a function of water temperature.1–3 Thus, determination of (Suprapur 65% nitric acid, Merck, Darmstadt, Germany). The Sr5Ca, U5Ca and Mg5Ca ratios in fossil corals can provide 236U uranium spike was purchased from Amersham (Les Ulis, information on past sea surface temperatures (SST).This France). To validate the method, a sedimentary rock reference information is crucial to provide climate modellers with quantimaterial, CCH-1 (Geology, Petrology and Geochemistry tative data and for a better understanding of climatic change through time. However, these three tracers have very contrasting concentrations in corals (about 8000 mg g-1 for Sr, Table 1 Acquisition parameters and plasma operating conditions for Ca, Sr and U 1100 mg g-1 for Mg and 3 mg g-1 for U).Their rates of change with respect to temperature are also dierent, about 0.9% Method Ca, Sr U per °C for Sr5Ca, 3.5% per °C for Mg5Ca and 4% per °C for Acquisition parameter — U5Ca. Magnesium is fairly abundant in corals, and shows Scan mode Peak hopping variations which can be easily quantified with ICP-AES. Spacing 0.1 Strontium and U are dierent, because Sr is abundant but Points per peak 1 varies little whereas U is just the opposite.As a consequence, Dwell time/ms 5000 10 000 all the pioneering work on Sr5Ca and U5Ca palaeothermo- Scans per replicate 100 250 metry has been carried out so far with thermal ionisation mass Number of replicates 3 Detector voltage/V 2200 spectrometry (TIMS), which gives good precision (<0.1%) but is expensive and slow. Shen and Dunbar4 have developed a Plasma parameter — fast method to measure U by isotope dilution ICP-MS but Ca Plasma gas flow/l min-1 16 18 Intermediate gas flow/l min-1 1.15 1.10 determinations were made separately by FAAS.The present Aerosol carrier gas flow/l min-1 0.78 0.82 work proposes a new method developed on an ICP–MS Sampling depth/mm 5 6 Power/kW 1.33 1.35 † Presented at the 1997 European Winter Conference on Plasma Sample uptake rate/ml min-1 1 Spectrochemistry, Gent, Belgium, January 12–17, 1997. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (969–973) 969Laboratory, Lie`ge University, Belgium), and an in-house coral applied to the first lens had a great influence on the stability of the Ca and Sr count rates, and therefore this particular lens standard labelled NC20 were used. All standards and samples were prepared by weighing reagents with a precision of 0.1 mg was fine-tuned each day. and stored in pre-cleaned polypropylene or Teflon bottles. Sr5Ca Measurement Sample Preparation To measure Ca and Sr in corals, two approaches were tried, Samples consist of calcium carbonate powder (typically less each involving the addition to the samples of the monothan 1 mg) drilled from a coral slab.This powder is dissolved isotopic elements 45Sc and 89Y. Neither Sc nor Y are present in 6 ml of 2% nitric acid (Suprapur 65% nitric acid, Merck) in significant amounts in corals. Calcium and Sc are very close which contains all the trace elements necessary to the analysis.in mass and first ionisation energy and show a similar behav- On average, samples contain 70 mg g-1 of Ca, 1.5 mg g-1 of Sr iour with time in the plasma. The same is true of Sr and Y. and 0.5 ng g-1 of U. The ‘external isotope dilution’ analytical method proposed by Lea and Martin was tried first,5 then, the present method was developed which is a slight variation of the internal standard Development of the Method method. The purpose of this work was to measure, with the same instrument and on the same aliquots, Ca, Sr and U.However, ‘External isotope dilution’ when developing the method, two major problems had to be dealt with: (i) Ca (about 40% m/m in corals) and Sr (about In their article,5 Lea and Martin claimed that 45Sc and 89Y 1–2%) are major elements in corals, whereas U is present at can be used to quantify Ca and Sr, respectively, by isotope the trace level (about 3–5 mg g-1); and (ii) these three elements dilution. This method was tested, and although the following have very dierent mass to charge ratios (m/z) and therefore a discussion will focus on Ca and Sc, the same conclusions apply compromise in the optimisation of instrumental parameters, to Sr and Y.especially ion lenses, had to be found. The ICP-MS operating To perform isotope dilution on Ca with another element conditions were selected in order to obtain the best sensitivity such as Sc, the behaviour in the plasma of the dierent isotopes for U coupled with the best precision for Ca and Sr.However, involved had to be investigated in order to try to answer some although the method for Ca and Sr gave relatively good key questions. (i) Do Sc and Ca ionise to the same extent in precision for U, it was decided to develop a separate method the plasma? (ii) If there is a dierence in ionisation, can it be specifically for this element. The U method allows the %RSD quantified? (iii) Is this dierence constant or is it dependent to be lowered by a factor of 2–3 for this element compared on factors such as the concentration of Ca? with the previous method.Pure Ca (between 30 and 90 mg g-1), Sr (between 1 and 2 mg g-1) and Sc–Y (100 ng g-1) standards were run alternately with mixed Ca–Sc–Sr–Y standards with similar concentrations, Choice of Isotopes on four dierent days during a two week period. The count The most abundant isotope of Ca, 40Ca, has an isobaric rates per ng g-1 for 43Ca, 48Ca, 45Sc, 86Sr and 89Y were interference with 40Ar and therefore cannot be measured using calculated and compared.Several important facts can be noted. ICP-MS. In any case, because of the high concentration of Ca 1. Calcium does not interfere with the 45Sc signal. The Sc that has to be quantified (70 mg g-1 on average), even 44Ca counts are identical in ultrapure HNO3, i.e., blanks and in and 42Ca are still too abundant (2.086 and 0.647% of the total pure Ca standards.Ca, respectively). Therefore, 43Ca and 48Ca which have an 2. The dierence in counts per second (cps) per ng g-1 abundance of 0.135 and 0.187%, respectively, were selected. between 43Ca and 45Sc expressed in % over 43Ca as a function There is a small interference of doubly charged 86Sr on 43Ca, of Ca concentration is shown in Fig. 1. The same treatment therefore the operating conditions of the instrument were was applied to the pairs 48Ca–45Sc and 86Sr–89Y, in Fig. 1(b) adjusted to minimise this.Each day, the count rate for 86Sr2+ and (c), respectively. Ideally, if Sc is to be used for isotope was measured and the count rate for 43Ca eventually corrected. dilution of Ca, the percentage dierence in cps per ng g-1 No evidence was found for a 43Ca-43Ca interference on 86Sr. between 43Ca and 45Sc should be independent of the Ca Results given by the two measured isotopes of Ca were concentration. This is clearly not the case. The higher the compared for accuracy and reproducibility. concentration of Ca, the lower the dierence calculated between For quantification of Sr, 88Sr could not be used because the 43Ca and 45Sc, and also between 48Ca and 45Sc.The present count rates were well over the saturation threshold of the experiment also indicates that for the same standard, large detector, hence 86Sr was chosen (with a natural abundance fluctuations of this dierence occur on a day to day basis, of 9.86%). which makes it dicult, if not impossible, to model.The most abundant isotope for U was measured i.e., 238U. The relationship linking Ca concentration to the dierence in count rates between 43Ca and 45Sc appears to be linear. The Optimisation of ICP-MS Instrument equation of the daily regression line is given and demonstrates that the intercepts are extremely variable and that a dierence In order to obtain the best results for Ca, Sr and U, two methods were developed, one to measure Ca and Sr, and one of up to 50% in the slope of the line can be expected from day to day. In conclusion, these data clearly indicate that to determine U.The acquisition parameters and plasma operating conditions used for each method are given in Table 1. neither Sc nor Y can be used with confidence as external spikes to conduct isotope dilution measurements on Ca and Sr. To improve the reproducibility of the U measurements, a longer dwell time and a greater number of scans per replicate were set.The ion optic conditions are not listed because they Internal standard method vary slightly on a day to day basis. Ion lens voltages were very dierent between the two methods, especially for the second The internal standard method is recommended to correct for sample matrix eects and for short and long term fluctuations and third lenses. For U, they were set to obtain the best sensitivity for high m/z values to the detriment of low m/z in signal. For Ca and Sr measurements, Sc and Y are ideally suited because of their above mentioned characteristics.values. On this instrument, it was found that the voltage 970 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Fig. 2 (a) Mass ratios versus Ca concentration for four multi-element standards. The regression lines and their equations are also shown. (b) 86Sr589Y versus Sr concentration for four multi-element standards. The regression line and its equation are also shown.curves were calculated each day. The output of this method is 20 analyses per hour with a dilute nitric acid wash of 1 min 30 s in the fast sample uptake rate between each sample. Calcium and Sr concentrations in coral samples were then Fig. 1 (a) Dierence in cps per ng g-1 between 43Ca and 45Sc derived from the equation of the regression lines [see Fig. 2(a) expressed in % over 43Ca as a function of Ca concentration; (b) dier- and (b)], and the Sr5Ca molar ratio in the sample calculated.ence in cps per ng g-1 between 48Ca and 45Sc expressed in % over 48Ca as a function of Ca concentration; and (c) dierence in cps per ng g-1 between 86Sr and 89Y expressed in% over 86Sr as a function of Sr concentration. U5Ca Measurement Uranium was measured by classic isotope dilution after However, in classical internal standardisation, small dierences addition of a 236U spike to the 2% nitric acid used to dissolve in the concentration of Sc and Y between the standard solutions the coral samples.The 235U5238U of a solution of known and the samples, owing to dierences in preparation, are not isotopic abundance, NBL SRM U500 (New Brunswick accounted for. Furthermore, mass bias, which can vary substan- Laboratories, USA), was measured each ten samples to assess tially in the mass range of Ca of interest, cannot be corrected. the mass bias. This mass discrimination eect was always less The method developed eliminates these two problems and than 1% per m/z value and the variation after 3 h did not yields highly accurate and reproducible Ca and Sr measure- exceed 0.5%.The RSDs for 238U5236U calculated on three ments. Scandium and Y were added to the four standard stock successive replicates in coral samples at a level of 0.3 ng g-1 solutions and to the samples at a concentration of 100.0 ng g-1. in 238U were around 1%. The U content in the sample was This value was chosen because it yields target ratios for obtained directly from 238U5236U corrected by mass bias and 43Ca545Sc, 48Ca545Sc and 86Sr589Y close to unity.These ratios multiplied by the 236U concentration and factor of natural were corrected for an instantaneous mass bias eect (measured abundance of 238U. Then the U5Ca molar ratio was calculated with 43Ca548Ca for Ca and 86Sr587Sr for Sr), and normalised from the Ca concentration found previously. The output of to a true internal standard concentration of 100.0 ng g-1.Four this method is around 25 samples per hour and no calibration multi-element standards were run and calibration lines linking is necessary. Ca concentration to 43Ca545Sc and 48Ca545Sc [Fig. 2(a)] and Sr concentration to 86Sr589Y [Fig. 2(b)] were calculated. The correlation coecients (R) were always higher than 0.9998 for RESULTS AND DISCUSSION Ca and Sr. As many parameters, such as configuration and properties of the cones, properties of the lenses, eciency of The ICP-MS method described herein can be compared with other techniques of quantification such as TIMS or ICP-AES, sample introduction system and plasma conditions, can change the sensitivity and the rate of interfering species, calibration in terms of sample output and accuracy. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 971Table 2 Reproducibility over 5 d, on coral standard NC20 A. Ca, Sr and Sr:Ca— Sr543Ca/ Sr548Ca/ Date 43Ca/mg g-1 48Ca/mg g-1 86Sr/mg g-1 mmol mol-1 mmol mol-1 17/09/96 74.22 74.20 1.469 0.009051 0.009054 74.36 74.34 1.466 0.009015 0.009018 74.16 74.14 1.463 0.009026 0.009028 18/09/96 74.43 74.08 1.462 0.008988 0.009030 74.24 73.90 1.467 0.009036 0.009078 74.60 74.26 1.468 0.008998 0.009040 24/09/96 74.21 74.51 1.466 0.009039 0.009003 74.14 74.43 1.468 0.009059 0.009022 74.12 74.41 1.468 0.009058 0.009022 26/09/96 73.91 73.92 1.465 0.009066 0.009065 74.19 74.21 1.465 0.009031 0.009030 74.26 74.28 1.472 0.009067 0.009066 9/10/96 74.22 74.46 1.472 0.009074 0.009044 74.13 74.37 1.467 0.009052 0.009022 74.14 74.39 1.467 0.009052 0.009022 Mean 74.22 74.26 1.467 0.009041 0.009036 s 0.16 0.19 0.003 2.5×10-5 2.1×10-5 RSD (%) 0.21 0.25 0.19 0.28 0.23 t-test value* 0.6046 Student’s 2.0484 t-value (a=5%) B.Ca, U and U5Ca— Date Ca/mg g-1 U/ng g-1 U5Ca/mmol mol-1 17/09/96 74.23 0.5054 1.155 18/09/96 74.08 0.5072 1.161 24/09/96 74.45 0.5091 1.160 26/09/96 74.13 0.5109 1.169 09/10/96 74.41 0.5090 1.160 Mean 74.26 0.5083 1.161 s 0.16 0.0021 0.005 RSD (%) 0.22 0.41 0.43 * The t-test indicates that results from 43Ca and 48Ca are not significantly dierent at the 5% risk level.Sample Consumption and Output therefore extremely important to obtain a good day to day reproducibility. With the TIMS technique, a very small amount of sample is The repeatability of the ICP-MS method was assessed by sucient for Sr5Ca determination: between 20 and 50 mg of measuring a standard solution for a 3 h period.The %RSD material.1,2,6 For the determination of U, samples weighing for 43Ca545Sc, 48Ca545Sc and 86Sr589Y were all less than 0.3% around 2 mg are generally used2 because of the low concen- (Fig. 3) while the RSDs of the counts per second for 43Ca, tration of U and the chemical treatment necessary prior 48Ca and 86Sr for the same period were 4.0, 3.8 and 3.9%, to analysis. respectively. For ICP-MS determination, the three elements of interest To evaluate the reproducibility of the ICP-MS method, the can be measured by dissolving 1 mg of coral in about 6 ml of same coral (i.e., NC20) was analysed on 5 d over a 3 week 2% nitric acid.Since no special treatment is needed after the period (Table 2A and B). Each day, Sr, Ca and U were coral powder has been dissolved, preparation time is much determined three times. The Sr5Ca values were calculated from shorter than for TIMS. calibration equations for 43Ca and 48Ca, respectively.The A second advantage of ICP-MS is the large sample output dierences between the two results are not significant in comparison with TIMS. On average, 50–80 samples can be (Table 2A) and are well below the precision of calibration analysed daily for Sr, Ca and U in the manual mode, and up to 100 if an autosampler is available. On the contrary, TIMS is very slow, and in the best case, a Sr–Ca–U determination takes at least 2 h. Therefore ICP-MS will be cheaper and is a particularly interesting technique to generate series of sea surface temperatures over long periods of time.Short and Long Term Precision The short (repeatability) and long term stability (reproducibility) over several days are two essential factors for work on corals for two reasons. Firstly, the rate of change of Sr5Ca per °C is small (about 0.9%), and the yearly variations in sea surface temperatures usually do not exceed 5–6 °C. Secondly, series over long periods of time are analysed which sometimes Fig. 3 Repeatability over 3 h for 43Ca548Ca, 43Ca545Sc, 48Ca545Sc, 86Sr587Sr and 86Sr589Y ratios. represent several hundred samples on the same coral, and it is 972 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Table 3 Comparison of Ca and Sr determinations by ICP-MS and CONCLUSION ICP-AES for reference carbonate CCH-1 and for coral standard NC20 A rapid method has been developed to determine precisely Ca, Parameter Certified value ICP-MS ICP-AES Sr and U in calcium carbonates.Although this method was used for palaeothermometry on aragonitic corals, it can be CCH-1 — equally applied to studies of calcitic organisms, such as ostra- Ca (%) 37.25 37.01 37.74 Number of analyses 54 3 cods or foraminifers. Furthermore, preliminary results indicate NC20 — that other elements of interest to palaeoceanography (Mg and Ca/mg g-1 83.69 84.18 Ba in particular) can be determined at the same time by isotope Sr/mg g-1 1.654 1.653 dilution, simply by introducing the appropriate spikes in the Sr5Ca/mmol mol-1 0.0090394 0.0089825 nitric acid prior to dissolving the coral powder.Number of analyses 9 3 t-test value* (on Sr5Ca) 1.381 Student’s t-value 2.228 This research is part of the ‘PALEO-OCEAN’ programme of (a=5%) ORSTOM. We thank Jacques Re�cy for initiating this programme, Guy Cabioch (ORSTOM, Villefranche sur mer) for * The t-test indicates that results from ICP-MS and ICP-AES are collecting coral NC20, Francis Sondag (ORSTOM, Bondy) for not significantly dierent at the 5% risk level.help with the ICP-MS, and Warren Beck (AMS Laboratory, University of Arizona) for critical comments. curves. The precisions on U5Ca are lower than on Sr5Ca. However, as the amplitude of the U5Ca variations as a function of SST is higher, the precision on SST will be the same or REFERENCES better. In the literature, the short and long term precision 1 Beck, J. W., Edwards, R. L., Ito, E., Taylor, F. W., Recy, J., reported for TIMS methods is generally better than 0.1% for Rougerie, F., Joannot, P., and Henin, C., Science, 1992, 257, 644. less than eight measurements.1,2 2 Min, G. R., Edwards, R. L., Taylor, F. W., Recy, J., Gallup, C. D., The precision of the method developed here for ICP-MS is and Beck, J. W., Geochim. Cosmochim. Acta, 1995, 59, 2025. 3 Mitsugushi, T., Matsumoto, E., Abe, O., Uchida, T., and Isdale, certainly good enough for palaeothermometry work on coral. P. J., Science, 1996, 274, 961. Coupled with the high sample output attainable with ICP-MS, 4 Shen, G. T., and Dunbar, R. B., Geochim. Cosmochim. Acta, 1995, it indicates that this instrument has the capability to replace 59, 2009. TIMS as the best technique for generating long term records 5 Lea, D. W., and Martin, P. A., Geochim. Cosmochim. Acta, 1996, of palaeotemperatures. 60, 3143. 6 McCulloch, M. T., Gagan, M. K., Mortimer, G. E., Chivas, A. R., and Isdale, P. J., Geochim. Cosmochim. Acta, 1994, 58, 2747. Accuracy Paper 7/01691C The accuracy of the ICP-MS method was assessed in various ReceivedMarch 11, 1997 ways. Coral standard NC20 was analysed with ICP-AES to Accepted June 3, 1997 check for Ca and Sr concentrations (Table 3). The reference carbonate material CCH-1 was also analysed. In both cases, the values generated by ICP-MS were consistent with those obtained with ICP-AES or other techniques. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 9

ISSN:0267-9477    

DOI:10.1039/a701691c

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Digestion Methods for Advanced Ceramic Materials and Subsequent Determination of Silicon and Boron by Inductively Coupled Plasma Atomic Emission Spectrometry† S. MANNa , D. GEILENBERGa , J. A. C. BROEKAERTa AND M. JANSENb aFachbereich Chemie, Universita� t Dortmund, D-44221 Dortmund, Germany bInstitut fu� r Anorganische Chemie, Rheinische Friedrich-W ilhelms-Universita�t Bonn, Gerhard-Domagk-Straße 1, D-53121 Bonn, Germany Various wet chemical digestion procedures, such as pressure with outstanding ceramic properties.The stoichiometry of one of these novel ceramic materials (Fig. 1) where the main digestion, microwave-assisted pressure digestion and components are Si, B, N and C12–14 was determined in the decomposition via alkali fusion, were applied to dierent present work. ceramic powders including Si3N4, BN and a new advanced As there are no adequate certified reference materials avail- ceramic material. The last powder was prepared from novel able for the calibration of analytical methods for the charac- precursor compounds by pyrolysis of inorganic polymers, and terization of this new class of compounds, no direct methods consists of Si, B, N and C.The precision obtained in the could be used and so it was decided to use wet-chemical determination of the main components Si and B by sequential decomposition with subsequent determination of the main ICP-AES varied from 0.6 to 3% (m/m) for Si and from 0.2 to components Si and B by sequential ICP-AES.The main 2% (m/m) for B depending on the digestion method and the components N and C as well as the oxygen impurities were ceramic material. Special attention was also given to the determined by carrier gas heat extraction.15–17 The results of recovery of Si and B, which is in most cases 100% within the the dierent digestion methods and the sum of the components standard deviation, and to the time consumption for the of the samples will be discussed.dierent digestion methods, which varied between 10 min and 23 h. It will be shown how the stoichiometry of the novel EXPERIMENTAL SiMBMNMC ceramic SiBN2.35C0.78 can be determined reliably by ICP-AES analysis for Si and B. The C, N and O The decomposition of the ceramic materials investigated, even were evaluated by carrier gas heat extraction. in the case of powders, was found to be dicult because of Keywords: Advanced ceramic material; determination of stoichiometry; pressure digestion; microwave-assisted pressure digestion; fusion with flux; inductively coupled plasma atomic emission spectrometry; carrier gas heat extraction The demands for high purity and homogeneous advanced ceramic materials have increased continuously in the course of their history.1,2 Many eorts have been made to improve their chemical, thermal, electrical, optical and mechanical properties for applications such as coatings, binders and fibres for reinforced materials. In a complex production process composite ceramics are prepared from mixtures of powders such as Si3N4, SiC and BN together with auxiliary or filler materials with the aim of accomplishing several specific properties. 3–8 The partly extreme properties of the final ceramic depend decisively on the stoichiometry and on the purity of the basic powder materials. However, these materials already contain impurities and co-milling of the binary compounds leads to heterogeneous distribution of impurities, which impairs the mechanical and physical properties of the final ceramic.To overcome these problems new processing routes have been developed. In view of the development of ceramics with tailor-made properties, preparation by pyrolysis of inorganic polymers9–11 oers new perspectives. The basic materials are organometallic compounds, which can easily be purified. By thermal or chemical cross-linking of these monomerics, inorganic polymers are formed. These polymers are pyrolysed in an ammonia or nitrogen stream at 1000–1400 °C into amorphous materials Fig. 1 Synthesis of a silicoboron carbonitride via the polymer † Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997. route.12–14 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (975–979) 975the high resistance of these advanced ceramics to thermal It was found that BN, Si3N4 and SiMBMNMC could be dissolved completely as with the PMD system (Table 2).The and chemical attack.1,2 For the determination of the main components Si and B in the novel SiMBMNMC ceramic by experimental parameters used with the Multiwave system are also included in Table 2. ICP-AES, several combined wet-chemical procedures were used. The conditions for the decomposition procedures were optimised with respect to the recovery of B in commercial BN, Decomposition by Alkali Fusion and to the recovery of Si in Si3N4 .Another technique which is well known for the decomposition of ceramic powders is alkali fusion.33–37 Although the determi- Pressure Digestion With Acids nation of trace elements in ceramic powders subsequent to fusion is limited, as a result of the high blank values stemming Conventional pressurised digestion with acid mixtures18–23 was from the flux, determination of the main components is performed in the DAB III pressure system (Berghof, Eningen, possible. Germany).The digestion vessels were made of PTFE and had For the fusion procedure 50 mg of BN, Si3N4 and a volume of 250 ml. Because of the thermal properties of SiMBMNMC were fused in a mixture containing 0.5 g of PTFE, the temperature should not exceed 250 °C. Up to four Na2CO3 and 0.5 g of K2CO3 . The fusion was performed in a pressure vessels can be heated simultaneously in the aluminium platinum crucible and the mixture was heated for 10 min with heating block. a Bunsen flame.The cooled melt of BN was dissolved in 5 ml The acid mixtures and the experimental parameters, which of HNO3 . Silicon containing samples form white residues of were found to be optimal for the digestion of BN, Si3N4 and silica gel which are insoluble in HNO3 . These residues can be the SiMBMNMC ceramic, are summarized in Table 1. dissolved in a mixture of 3 ml of HNO3 and 2 ml of HF. Microwave-assisted Pressure Digestion With Acids Carrier Gas Heat Extraction In addition to the conventional pressure digestion with convec- In the determination of C, N and O in ceramic powders by tive heating microwave-assisted pressure digestion24–29 was this method, new problems including inadequate accuracy and also used.Matusiewicz and co-workers26,27 and Tanimoto and the need for very high temperatures were encountered.4,15,38 Fukumura29 have shown that ceramic powders such as Al2O3 , As O is omnipresent there have been numerous studies con- AlN, BN, SiC and Si3N4 can be dissolved by microwavesidering the dierent states of chemical bonding between O assisted pressure digestion with decomposition times ranging and, for example, Si3N415–17 where O can be bound to the from 0.1 to 7 h.Decomposition time depends on the acids particle surface or dissolved in the matrix. However, with used, the temperature, the pressure, the surface area of the appropriate flux additives and temperatures above 2000 °C the sample material, its grain size and sometimes also on the ceramic samples are decomposed completely.purity of the powders.27 The determination of N and O was performed with a Leco (Kirchheim, Germany) TC-436 analyser. About 10 mg of PMD system sample were weighed in an Sn capsule and about 0.9 g of Ni powder and 0.35 g of granulated Sn were added. The com- The pressurised microwave digestion (PMD) system30,31 pressed capsule, which serves as the metallic flux, was dropped (Anton Paar GmbH, Graz, Austria), used in this work, was a into a preheated graphite crucible and burned at about 2500 °C pressure controlled system.The microwave power supply is in an atmosphere of He. Carbon monoxide was formed and interrupted when the pressure reaches a preset limit. The PFA this reaction product was oxidised to CO2 and determined by (perfluoroalkoxy ethylene) vessels used have a volume of 50 ml orption. Nitrogen was detected in a heat conductivity and a pressure limit of 35×105 Pa.An exhaust and cooling cell.15,16 unit (ECU) is incorporated in the system to minimise the time The total carbon content in the SiMBMNMC ceramic required for cooling and to protect the oven from corrosive was determined with the aid of a Leco C-200 analyser. fumes. Approximately 40 mg of the sample were mixed with 1 g of The acid mixtures and the conditions for the digestion of Cu and 0.75–0.8 g of Fe powder, dropped into a Mullit crucible the samples are given in Table 2. and heated to about 2000 °C in a stream of O2.The CO2 formed was detected by IR absorption.15,16 For calibration a Multiwave system certified WC sample was analysed using the same method. The other microwave-assisted pressure digestion system used was the Multiwave32 (Perkin-Elmer, U� berlingen, ICP-AES Germany/Anton Paar GmbH, Graz, Austria). It also employs The Si and B in the digestion solutions were determined by pressure control, however, in addition the temperature of each ICP-AES using a JY 24 sequential spectrometer (Instruments vessel can be measured by an IR temperature sensor.The SA, Metuchen, NJ, USA). The instrumental and operating TFM (tetrafluoromethaxil) vessels used have a volume of parameters used are listed in Table 3. 100 ml and a pressure limit of 35×105 Pa. With this system The digestion solutions, independent of the digestion up to six samples can be decomposed simultaneously. method, were first diluted to 100 ml with deionized water and 1 ml of these solutions were diluted once more to 100 ml, ready Table 1 Acid mixtures and working conditions for the digestion of for analysis.It is not necessary to complex HF to protect the BN, Si3N4 and the SiMBMNMC ceramic by conventional pressure digestion (DAB III) cross-flow nebulizer, which is made of PTFE, and the quartz spray chamber because of the overall dilution. The concen- Sample Reagents/ml Temper- Digestion tration of HF in the analysed solutions is about 0.02%. At Sample amount/ ature time/ this concentration, the spray chamber is not damaged by HF.(powder) mg HF HNO3 HCl °C min For the determination of Si, calibration was carried out by BN 50 2 4 6 180 17 standard additions to minimise matrix eects. The calibration Si3N4 30–50 2 4 6 180 15 graph was obtained by using a blank solution and four sample SiMBMNMC 50 2 4 6 180 15–20 solutions with standard additions of between 0 and 3 mg ml-1 976 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Table 2 Acid mixtures and working conditions for the digestion of BN, Si3N4 and the SiMBMNMC ceramic by microwave-assisted pressure digestion (PMD, Multiwave) Total time Sample Reagents/ml including Sample amount/ Digestion cooling-down (powder) mg HF HNO3 H2SO4 time/min time/min PMD BN 20 2 2 2 30 40 Si3N4 20 2–3 2 2 3×50* 210 SiMBMNMC 20 2–3 2 2 2×50* 140 Multiwave BN, Si3N4, SiMBMNMC 30 3 2 2 18† 33 * Cooling-down period of 20 min after each 50 min heating without opening the vessels.† Power ramp 500–1000 W within 3 min; decomposition period 1000 W, 15 min. Table 3 Instrumentation and operating parameters sample, irreproducible amounts of Si were found, which also corresponds with the findings of Kaiser et al.40 This indicates ICP-AES Sequential ICP-AES spectrometer JY 24 that Si present as SiF4 is adsorbed by PTFE. However, after (Instruments SA) optimisation of the reaction mixture the results of the determi- Generator 900 W at 40.68 Hz nation of Si in Si3N4 agree with the specified values within the Monochromator 0.64 m Czerny–Turner mounting (2400 lines mm-1) standard deviation indicated (Table 4).Outer gas 16 l min-1 Ar For the microwave assisted pressure digestions the acid Aerosol carrier gas 1.2 l min-1 Ar mixtures again have to be optimised. Indeed, when using acid Intermediate gas 0.3 l min-1 Ar mixtures with low boiling points, such as HF, HCl and HNO3 , Nebuliser Cross-flow (Instruments SA) the pressure limit of 35×105 Pa for both systems is reached Peristaltic pump Perimax 12 and 16 (Spetec, Erding, Germany) at temperatures which are not high enough to dissolve the 1.3 l min-1 ceramic materials.To increase the maximum temperature H2SO4 was used instead of HCl. Despite the volatile SiF4 that is formed during the decomposition, no losses of Si were of Si. The analytical lines Si I 251.611 nm and Si I 212.412 nm observed. were used for the determination of Si.As it is known that fluoropolymer materials can only be It appeared that it was not necessary to use standard heated to temperatures up to 250 °C the digestion times were additions for the determination of B. The calibration for B limited to a maximum of 50 min in one procedure so as to was carried out using a blank solution and three standard prevent the PFA and TFM vessels from overheating. solutions with concentrations of B of between 0.5 and With the PMD system, the decomposition times generally 2 mg ml-1, which also contained the reagents used for the could be reduced from almost 1 d with the convectively heated digestion.The analytical lines B I 249.678 nm and B I DAB III system to 0.5–3 h (Table 2). However, Si3N4 , which 249.773 nm were used. has the lowest solubility of all materials studied, is not dissolved As the samples did not contain detectable amounts of Fe, completely even after three digestion procedures of 50 min spectral interferences of the selected B and Si lines with Fe each (Table 2).Although the concentrations of Si determined lines were not found. agree with the specification of the producer within the standard deviation, the low concentration of 57% found and the high Reagents standard deviation of 3% reflect the problems of digesting Si3N4 in the PMD system. The results obtained using the The test samples used were BN (hexagonal-BN, 99% Aldrich, Multiwave system agree with the specification of the producer Milwaukee, WI, USA) and a-Si3N4 (‘predominantly a-phase’, within a lower standard deviation (Table 4).The results for Sigma, St. Louis, MO, USA). The acids used were HF (38%, the fusion with Na2CO3–K2CO3 are in agreement with the pro analysi, J. T. Baker, Phillipsburgh, NJ, USA), HNO3 (67%, specification of the producer (Table 4) and the decomposition puriss, Riedel-de Hae�n, Hannover, Germany), HCl (37%, times may be very short.However, the precision for the puriss, Riedel-de Hae�n) and H2SO4 (95–97%, pro analysi, determination of Si is not very good (Table 4). The sum of the Merck, Darmstadt, Germany). The standards included silicon components is calculated considering the propagation of errors standard solution (acidic, 1000 mg l-1 Si, Merck) and Titrisol obtained with the individual standard deviations for each boron standard solution (5.000±0.005 g l-1 B, Merck). element. With the results of the determinations by ICP-AES RESULTS AND DISCUSSION Table 4 Comparison of dierent digestion methods for the determi- Docekal et al.39 used an acid mixture of 4 ml of HF, 4 ml of nation of Si in Si3N4 by ICP-AES (specification of producer 60.06% Si) HNO3 and 4 ml of H2SO4 for the decomposition of silicon carbide by pressure digestion.In the present work with this Digestion system Silicon (%) Sum of components* (%) mixture it was found that losses of Si of up to 20% occur with DAB III 60±1 101±2 the DAB III pressure digestion system at temperatures of (n=10) RSD 1.7% 240 °C.These observations agree with those of Kaiser et al.40 PMD 57±3 98±5 (n=12) RSD 5.3% and these eects could be related to the volatility of SiF4. To Multiwave 59±1 100±2 minimise these losses the temperature was reduced to below (n=8) RSD 1.7% 200 °C and the composition of the acid mixture modified Salt fusion 60±2 101±3 (Table 1). When using an excess of HCl as compared with HF, (n=8) RSD 3.3% the losses of SiF4 can be reduced, because less volatile SiCl4 will then be formed.When performing blank digestion pro- * Determination of N and O by carrier gas heat extraction: 38.92±0.06% N; and 1.smn;0.01% O. cedures in the PTFE vessels, after the decomposition of a Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 977(Si ) and carrier gas heat extraction (N and O) a sum of 100% the Multiwave and fusion with Na2CO3–K2CO3 agree within the standard deviation; however, the concentration found after is obtained within the standard deviation (Table 4).The concentration of B in BN found after digestion of the digestion with the DAB III system is lower (Table 6). This can be explained by losses of SiF4 as described above. sample in the DAB III system (Table 5) did not agree with the specification of the producer; the recovery was 108% and The results for B in the case of the PMD, the Multiwave and fusion with Na2CO3–K2CO3 agree within the standard the relative standard deviation high.There were no significant blank values found for B, and the standard solutions were deviation (Table 6). The high concentration found by conventional pressure digestion with the DAB III system is similar prepared with the same acid concentrations as the sample solutions. Hence this result remains unexplained. to the BN. Because of the uncertainty of the results for Si and B in the The concentration of B in BN found after digestion with the Multiwave system corresponds to a recovery of 108% but with case of the DAB III system, it was decided not to consider these values for the calculation of the stoichiometry of the a relatively high degree of precision.No noticeable blank values were found, so this overestimate also remains SiMBMNMC ceramic. The results of the present work as well as those obtained by unexplained. The results obtained by fusion with Na2CO3–K2CO3 are in some other laboratories are listed in Table 7.The molar ratio of Si to B is of particular interest. In the starting materials for very good agreement with the specification of the producer. The sum of the components for all methods is 100% within the synthesis of the SiMBMNMC ceramic (Fig. 1) the molar ratio of Si to B is 151. Since no losses of Si and B were the standard deviation. The precision of 5–7% is exceptional and results from the low precision of the determination of N observed during the synthesis, the molar ratio in the final ceramic material should also be 151.The stoichiometry of the by carrier gas heat extraction (Table 5). In the same manner as for BN and Si3N4, the new new ceramic material was calculated from all results with the exception of the concentrations obtained with the DAB III SiMBMNMC ceramic was dissolved using the methods described above. The results for Si in the case of the PMD, system and it was found to be SiBN2.35C0.78.Table 5 Comparison of dierent digestion methods for the determination of B in BN by ICP-AES (specification of producer 42.5% B) CONCLUSION Digestion system Boron (%) Sum of components* (%) The determination of the main components of ceramic materials is dicult because of the high resistance of these materials DAB III 46±2 103±7 to chemical digestion. (n=11) RSD 4.3% PMD 43.4±0.6 101±5 For wet-chemical procedures, digestion with acids at high (n=9) RSD 1.4% pressure enables the use of very pure reagents, such as obtained Multiwave 46.0±0.4 103±6 by sub-boiling distillation.With digestions in a closed system (n=6) RSD 0.9% losses of volatile compounds and contaminations from the Salt fusion 42.3±0.8 100±6 environment can be minimized. The high temperatures which (n=9) RSD 1.8% can be reached at high pressure are necessary to dissolve * Determination of N and O by carrier gas heat extraction: highly resistant materials such as ceramic powders.However, 56.9±0.6% N; and 0.57±0.03% O. the lengthy time required for pressure digestion methods that employ convective heating is not acceptable for routine work. Table 6 Concentration of Si and B in a SiMBMNMC-ceramic as Microwave-assisted pressure digestion procedures have been determined by ICP-AES after decomposition by dierent digestion shown to be fast and reliable alternatives to conventional methods pressure digestion. They have the same advantages of using pure reagents in a closed system, but in addition allow a Sum of components* remarkable gain in sample preparation throughput.Digestion system Silicon (%) Boron (%) (%) Decomposition by alkali fusion has advantages of low time DAB III 31.8±0.8 13.6±0.2 95±7 consumption and of low costs. On the other hand, working in (n=10) RSD 2.6% RSD 1.5% an open system may lead to analyte losses and to contami- PMD 32±1 12.7±0.3 95±7 (n=11) RSD 3.1% RSD 2.4% nations from the environment. Further, the solid fusion Multiwave 33.4±0.6 13.1±0.2 97±7 reagents are not readily obtained in a suciently pure form, (n=7) RSD 1.8% RSD 1.5% which is problematic for trace determinations.Additionally, Salt fusion 33±2 13.4±0.3 97±9 analyses of the solutions by ICP-AES produces matrix eects (n=12) RSD 6.0% RSD 2.2% owing to high salt loading. According to the present work, microwave-assisted pressure * Determination of N, O and C by carrier gas heat extraction: 39±1% N; 0.59±0.03% O; and 11.1±0.3% C.digestion shows the most promise for the digestion of ceramic Table 7 Summary of all analysis results for the SiMBMNMC ceramic Molar ratio Method/laboratory Si (%) B (%) Si5B 32±1 12.7±0.3 151.02 Microwave-assisted pressure digestion, PMD (HNO3–HF–H2SO4) Microwave-assisted pressure digestion, multiwave 33.4±0.6 13.1±0.2 151.02 (HNO3–HF–H2SO4) Fusion with NaKCO3 33±2 13.4±0.3 151.06 Commercial analysis 35.7/35.8* 12.6/12.7† 150.92 Krupp Hoesch Stahl AG 33.23‡ 12.77§ 151 Mean value 33±2 12.9±0.5 1.0051.00 * Fusion with Na2CO3–H3BO3 , ICP-AES.† Pressure digestion with HNO3–HF, ICP-AES. ‡ Fusion with Na2B4O7 , gravimetric determination as SiO2 . § Fusion with Na2CO3–K2CO3–KNO3 ; titration of the boric acid with NaOH. 978 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 1217 Adelhelm, C., and Hirschfeld, D., Fresenius’ J. Anal. Chem., 1992, materials. Since the application of this method in the field of 342, 125. ceramic materials is fairly recent, further improvements should 18 Graule, T., von Bohlen, A., Broekaert, J.A. C., Grallath, E., be possible, including the use of simultaneous ICP-AES with Klockenka�mper, R., Tscho� pel, P., and To� lg, G., Fresenius’ Z. internal standardization to provide enhanced precision for the Anal. Chem., 1989, 335, 637. major elements.41 19 Graule, T., Tscho� pel, P., Broekaert, J. A. C., and To� lg, G., Ceram. Forum Int., 1991, 68, 5. 20 Jackwerth, E., and Gomiscek, S., Pure Appl.Chem., 1984, 56, 479. The authors thank the Anton Paar GmbH, Graz, and the 21 White, R. T., Jr., J. Assoc. O. Anal. Chem., 1989, 72, 387. Bodenseewerk Perkin-Elmer GmbH, U� berlingen, for supplying 22 Knapp, G., Mikrochim. Acta, 1991, II, 445. 23 Buresch, O., Ho� nle, W., Haid, U., and v. Schnering, H. G., the microwave-assisted digestion systems. S. M. thanks the Fresenius’ Z. Anal. Chem., 1987, 328, 82. Deutsche Forschungsgemeinschaft (DFG) for the grant of a 24 Nakane, K., Uwamino, Y., Morikawa, H., Tsuge, A., Iida, Y., and postdoctoral research fellowship.Ishizuka, T., Bunseki Kagaku, 1995, 44, 319. 25 Introduction to Microwave Sample Preparation. T heory and Practice, ed. Kingston, H. M., and Jassie, L. B., American Chemical Society, Washington, DC, 1988. REFERENCES 26 Matusiewicz, H., and Sturgeon, R. E., Prog. Anal. Spectrosc., 1 Broekaert, J. A. C., Graule, T., Jenett, H., To� lg, G., and Tscho� pel, 1989, 12, 21. 27 Matusiewicz, H., Mikrochim. Acta, 1993, 111, 71. P., Fresenius’ Z. Anal. Chem. 1989, 332, 825. 28 Tata� r, E., Varga, I., and Za�ray, G., Mikrochim. Acta, 1993, 111, 45. 2 Broekaert, J. A. C., and To� lg, G., Mikrochim. Acta, 1990, II, 173. 29 Tanimoto, M., and Fukumura, H., Bunseki Kagaku, 1996, 45, 357. 3 Mazdiyasni, K. S., and Ruh, R., J4, 415. 30 Panholzer, F., L aborPraxis, 1994, 10, 32. 4 Lee, J. D., Moeller, H. H., Petrak, D. R., and Berneburg, P.L., 31 Knapp, G., Maichin, B., and Panholzer, F., Colloquium Am. Ceram. Soc. Bull., 1984, 63, 422. Atomspektrometrische Spurenanalytik, 1991, 6, 571. 5 Bellosi, A., Guicciardi, S., and Tampieri, A., J. Eur. Ceram. Soc., 32 Kainrath, P., Kettisch, P., Schalk, A., and Zischka, M., 1992, 9, 83. L aborPraxis, 1995, 11, 34. 6 Sawaguchi, A., Toda, K., and Niihara, K., J. Am. Ceram. Soc., 33 Spex Handbook of Sample Preparation and Handling, ed. Obenauf, 1991, 74, 1142.R. H., Spex Industries, Edison, NJ, 1985. 7 Momeya, K., and Matsui, M., inMaterials Science and T echnology, 34 Ishizuka, T., Uwamino, Y., Tsuge, A., and Kamiyanagi, T., Anal. ed. Swain, M. V., VCH, Weinheim, 1994, vol. 11, pp. 517–565. Chim. Acta, 1984, 161, 285. 8 Wo� rner, W., Kaiser, G., and Schubert, H., Mikrochim. Acta, 1993, 35 Homeier, E. H., Kot, R. J., Bauer, L. J., and Genualdi, J. T., 110, 173. J. Anal. At. Spectrom., 1988, 3, 829. 9 Winter, G., Verbeek, W., and Mansmann, M., (Bayer AG), DE 36 Foner, H. A., Analyst, 1984, 109, 1469. 3892583, 1975. 37 Aufschlußmethoden der anorganischen und organischen Chemie, ed. 10 Yajima, S., Hayashi, J., and Omori, M., Chem. L ett., 1975, 931. Bock, R., Verlag Chemie, Weinheim, 1st edn., 1972. 11 Seyferth, D., and Wiseman, G. H., Polym. Prep., 1984, 25, 10. 38 Garten, R. P. H., J. Chin. Chem. Soc. (T aipei), 1994, 41, 259. 12 Baldus, H. P., Wagner, O., and Jansen, M., Mater. Res Soc. Symp. 39 Docekal, B., Broekaert, J. A. C., Graule, T., Tscho� pel, P., and Proc., 1992, 271, 821. To� lg, G., Fresenius’ J. Anal. Chem., 1992, 342, 113. 13 Baldus, H. P., Passing, G., Sporn, D., and Thierauf, A., in High- 40 Kaiser, G., Meyer, A., Friess, M., Riedel, R., Harris, M., Jacob, T emperature Ceramic-Matrix Composites II: Manufacturing and E., and To� lg, G., Fresenius’ J. Anal. Chem., 1995, 352, 318. Materials Development, ed. Evans, A. G., and Naslain, R., Ceramic 41 Wu� stkamp, D., Kucharkowski, R., and Broekaert, J. A. C, Trans. 58, American Ceramic Society, Westerville, OH, 1995, Fresenius’ J. Anal. Chem., 1996, 355, 281. pp. 75–84. 14 Jansen, M., and Baldus, H. P., Angew. Chem., 1997, 109, 338. Paper 7/01446E 15 Grallath, E., in Gase in Metallen, ed. Hirschfeld, D., DGMReceived March 3, 1997 Informationsgesellschaft, Oberursel/D, 1984, pp. 1–26. 16 Grallath, E., and Ortner, H. M., Talanta, 1978, 25, 195. Accepted May 15, 1997 Journal of Analytical Atomic Spectrometry, September 1997, V

ISSN:0267-9477    

DOI:10.1039/a701446e

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Addition of Small Amounts of Helium and Hydrogen to the Working Gases in Slurry Nebulization Inductively Coupled Plasma Atomic Emission Spectrometry for the Analysis of Ceramic Powders† P. HEITLAND AND J. A. C. BROEKAERT* Universita�t Dortmund, Fachbereich Chemie, D-44221 Dortmund, Germany Small amounts of hydrogen and helium were added to the with aqueous solutions is possible when (i) the stability and homogeneity of the slurry is guaranteed, (ii) the analyte internal gas flow of a 1.5 kW inductively coupled argon plasma to investigate the capabilities of the resulting mixed gas transport eciency of the slurry is identical with that of the solution and (iii ) the atomization eciencies of the analyte in plasmas with respect to the particle evaporation in slurry nebulization inductively coupled plasma atomic emission the slurry and in the solution are similar.Owing to the high melting and decomposition temperatures spectrometry for the analysis of refractory ceramic powders.The addition of up to 15% v/v of hydrogen to the internal gas of refractory ceramic powders the evaporation of the slurry particles has been recognized as a possible limiting factor in results in an increase in the Al and Si line intensities for slurries of Al2O3 and SiC powders of up to 20–100%, after the analysis of ceramic powders. To estimate the evaporation eciency of refractory particles models have been devel- optimization of all parameters. The corresponding increases in the recoveries were found to be of the order of 9–14% for oped.21,27,28 Barnes and Schleicher27 have calculated that it is possible to vaporize Al2O3 particles of a size between 1 and Al2O3 and SiC powders with sizes of 0.3–18 and 0.3–12 mm, i.e., from 48 to 62% and from 43 to 52%, respectively. 10 mm in the first 15–20 mm of the plasma discharge. Chen and Pfender28 have calculated a time of 41.5 ms for evaporation Additions of helium, however, were only 50% as ecient. These eects refer to a more ecient evaporation of the of Al2O3 particles with a diameter of 50 mm in a thermal plasma with a temperature of 12 000 K. Raeymaekers et al.22 powders in the plasma.Axially resolved measurements of analyte line intensities for both slurries and solutions also calculated the evaporation eciencies for Al2O3, SiO2 and ZrO2 powders and found that a 50% evaporation is obtained reflect the influence of the amounts of added hydrogen or helium on the plasma geometry.for Al2O3 powders with a particle size of 18 mm, for SiO2 at 29 mm and for ZrO2 at 7 mm. Keywords: Ceramic powders; slurry nebulization; inductively The eect of mixed gas plasmas such as argon–oxygen, coupled plasma atomic emission spectrometry; hydrogen and argon–nitrogen or argon–hydrogen on slurry nebulization helium addition ICP-AES has been investigated by several workers.29–31 In this work the addition of hydrogen and helium to the working gases of an ICP was studied to improve the evaporation Al2O3 and SiC ceramic powders are of great importance for the production of advanced ceramics and a well-defined purity eciency for slurries of ceramic powders in the ICP as it is known that helium and hydrogen increase the gas kinetic of these powders is very often required.1 Slurry nebulization inductively coupled plasma atomic emission spectrometry temperature in an ICP.It has been reported that the addition of 16.7% v/v of helium to the central gas of an argon ICP (ICP-AES) has been widely shown to be a powerful approach for the direct analysis of these ceramic powders.The technique results in a temperature enhancement of 1500 K and that an addition of the same concentration of hydrogen results in a was introduced by Fuller et al.2 and Ebdon and Cave.3 Over the years it has been used for the analysis of dierent types of temperature enhancement of 2000 K.32 Both enhancements were attributed to the higher thermal conductivities of hydro- samples such as coal,4–6 clay,7 kaolin8,9 or geological samples.10,11 For the analysis of important ceramic powders gen and helium as compared with argon.As the higher thermal conductivity leads to an improvement of the energy transfer the technique has been investigated for Al2O3,12–15 SiC,12,14,16 Si3N414 and ZrO2.12,14,17,18 The main advantage of slurry from the ring plasma towards the injected aerosol,33 it was decided to investigate whether this eect was of any benefit in atomization is the minimal sample preparation required, as a result of which the duration of the analysis is decreased and slurry nebulization ICP-AES for the analysis of Al2O3 and SiC ceramic powders.risks of contamination, losses of volatile elements, adsorption of analytes on surfaces and incomplete dissolution during the sample decomposition step are circumvented. EXPERIMENTAL The capabilities and limitations of slurry nebulization depend on the particle size distribution of the suspended ICP Operating Conditions powder.The particle size influences the stability and homogen- A free running FS 4 generator (Linn, Hirschbach, Germany), eity of the slurry as well as the aerosol transport eciency in a modified Greenfield burner and a microcomputer-controlled the spray chamber and the evaporation eciency for the solid 0.4 m Czerny–Turner monochromator (PE 4000, Perkin- particles in the plasma. The eects of stability or homogeneity Elmer, U� berlingen, Germany) were used for the ICP-AES of slurries19,20 and transport phenomena of slurries21–26 have measurements.been investigated in detail. It was found that direct calibration To the working gas flows two additional gas flows could be added at the exit of the spray chamber of the G. M. K. † Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997. nebulizer (Labtest, Ratingen, Germany).One of these Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (981–986) 981additional gases was argon, the other was hydrogen or helium. for 10 min. The slurry was then pumped through a cuvette, where the laser light (He-Ne laser; wavelength, 633 nm) is In this way the total flow of the additional gases could be held scattered by the particles. The precision of the determination constant and for our experiments it was 0.15 l min-1.Hence, of the average diameter of the powders is 4%. a variation of the hydrogen content in the internal gas could be achieved at a constant nebulizer gas flow (0.65 l min-1). The advantage of the modified Greenfield burner is that both RESULTS AND DISCUSSION the aerosol generation eciency and the total internal gas flow Particle size distributions for the Al2O3 slurry were obtained (nebulizer gas plus total amount of additional gases) can be by laser light scattering and found to extend from 0.5 to 18 mm held constant.All gases added to the ICP were controlled by (Table 2). It could be observed that 95% of the Al2O3 particles gas flow rotameters separately, which were calibrated for the in the slurry are smaller than 10 mm and that the average respective gases. diameter is 3.1 mm. The SiC slurry contains particles with sizes The parameters optimized with respect to maximum net between 0.5 and 12 mm and the average diameter is 4.8 mm.In intensities of Al and Si lines after background subtraction and the SiC slurry, 95% of the particles are smaller than 8 mm. further operating conditions are listed in Table 1. Measurements of the number particle size distributions by electron probe microanalysis gave results which agreed well Preparation of the Solutions and the Slurries with those of laser light scattering. The Al2O3 particles were found to be angular or spherical and the SiC particles always For the preparation of the solutions, Titrisol standards (Merck, had an angular and sharp-edged geometry.Both Al2O3 and Darmstadt, Germany) with an analyte concentration of 1 g l-1 SiC slurries are stable, when they are prepared in an ultrasonic were diluted. Slurries were prepared from two powders, Al2O3- bath and stirred vigorously during the sample aspiration. The VAW and SiC-F 1200, which are typical refractory ceramic dard deviations of the line intensities for the matrix element materials (Table 2).In order to achieve analyte concentrations (Al or Si) in the slurries are between 0.9 and 2.1% over a of 0.5 g l-1, the powders were weighed and transferred quantimeasurement period of at least 30 min. It was also demontatively into a calibrated flask (50 ml ). After shaking vigorously strated that the intensities of matrix lines for Al and Si in the slurry was agitated in an ultrasonic bath (Sonorex TT 3, solutions and slurries at Al or Si concentrations of 0.5 g l-1 Bandelin Electronics, Berlin, Germany) with a frequency of are within the linear calibration range. 35 kHz and a power of 450 W for 10 min. It was found that The use of slurry nebulization ICP-AES for the analysis of slurries prepared in this way are stable for at least 30 min. ceramic powders requires calibration with aqueous solutions During the sample aspiration the slurries were stirred continubecause certified reference materials are usually not available. ously with a magnetic stirrer (Janke and Kunkel, Staufen, In order to investigate whether hydrogen and helium addition Germany; frequency 500 rev min-1).to the argon plasma can improve the evaporation of solid particles, which is a prerogative for calibration with aqueous solutions, the recovery of the matrix element in the slurry was Particle Size Distribution of the Powders determined at dierent helium or hydrogen contents in the The particle size distributions of the powders were determined nebulizer gas.The recovery was calculated from the line by laser light scattering with the aid of a Malvern 2600 Particle intensity ratio of the matrix element in the slurry and solution. Sizer. The instrument could be used for particle sizes between 0.5 and 188 mm. For the determination of the particle size Addition of Hydrogen to the Internal Gas distribution, 100 mg of the powder were dispersed in 300 ml of water and the slurry was homogenized by stirring and The addition of hydrogen to the internal gas of the ICP was found to be possible up to a hydrogen content of 15% v/v.ultrasonic vibration (frequency, 40 kHz; power, 0.5 W cm-3) Table 1 Instrumental parameters and ICP operating conditions FS 4 (Linn), free-running, frequency: 40.68 MHz, forward power: 1.5 kW, reflected power: <50 W Rf generator Torch According to Greenfield, internal tube: 2 mm, intermediate tube: 18 mm, outer tube: 25 mm Nebulizer G.M.K. without impact bead (Labtest) Outer gas flow/l min-1 28 Intermediate gas flow/l min-1 2 Nebulizer gas flow/l min-1 0.65 Additional gas flow/l min-1 0–0.15 Total internal gas flow/l min-1 0.8 Sample uptake rate/ml min-1 2.3 Spectrometer Czerny–Turner monochromator (PE 4000, Perkin-Elmer), microcomputer-controlled, focal length: 408 mm, UV range grating: 2880 lines mm-1, visible range grating: 1440 lines mm-1, reciprocal linear dispersion: 0.65 nm mm-1 in the UV and 1.3 nm mm-1 in the visible Observation height/mm 10 Analytical lines/nm Al I 396.152 Si I 288.158 Table 2 Particle size distributions for the Al2O3 and SiC powders Laser light scattering Electron probe microanalysis Mean diameter/ Size fraction/ Mean diameter/ Size fraction/ Ceramics Type mm mm mm mm Al2O3 VAW 3.1 0.5–18 2.8 0.3–16 (VAW, Bonn, Germany) SiC F 1200 4.8 0.5–12 4.3 0.3–12 (ESK, Kempten, Germany) 982 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12The use of higher additions of hydrogen leads to an unstable plasma and at contents higher than 20% v/v plasma extinction may occur.At hydrogen contents higher than 12% v/v a correction of the generator power was found to be required and, therefore, such a high addition of hydrogen has to be avoided for analyses. To investigate the eect of adding hydrogen to the nebulizer gas, the axial intensity profiles of the Al and Si line intensities for solutions and slurries were measured. On variation of the hydrogen content in the internal gas, no appreciable blank for Si, as could be released from the torch, was measured.Hydrogen addition to the internal gas with respect to the axial intensity profiles was found to cause an overall increase in the Al and Si line intensities at all observation heights (Figs. 1 and 2). This increase was proportional to the hydrogen content at all heights. Over all observation heights the maximum enhancement in line intensity for slurries is 20% for Al and 100% for Si.A higher line intensity enhancement is obtained for the Si line which has a higher excitation energy (for Si I 288.158 nm, Eexc=5.08 eV and for Al I 396.152 nm, Eexc=3.14 eV). This finding is in agreement with the results of Schramel et al.,34 who also found that, in the presence of hydrogen, the degree of intensity enhancement is more significant for lines with high excitation energies. Schramel et al.34 have reported improvements in the limits of detection for several elements due to increased intensities and a more stable plasma, expressed by a Fig. 2 Axial intensity profiles for the Si I 288.158 nm line for solutions reduced standard deviation of the background intensity.In (a) and for slurries of SiC (b) at dierent hydrogen contents in the our investigations the standard deviations of the background internal gas: A, 0; B, 5; C, 10; and D, 15% v/v. Sample concentrations: 0.714 g l-1 SiC (0.5 g l-1 Si) for the slurry and 0.5 g l-1 Si for intensity and of the Al and Si line intensities for solutions were the solution. 0.5–1.5%, and they did not decrease on adding hydrogen to the argon plasma. It was also found that, at higher hydrogen contents, the maximum intensity tends to shift to lower obser- matrix line intensities decrease to values below those for lower hydrogen contents at higher viewing heights. Both observations vation heights (Figs. 1 and 2). The intensity maxima for the Al and Si lines shift from 10 mm (curve A) to 8 mm (curve D) can be understood in terms of a reduction of the plasma volume when hydrogen is added to the internal gas.This for solutions [Figs. 1(a) and 2(a)] and from 12 mm (curve A) to 8 mm (curve D) for slurries [Figs. 1(b) and 2(b)]. volume contraction of the plasma can be observed visually for hydrogen contents in the internal gas of 5% v/v. Another Furthermore, when adding 15% v/v hydrogen (curve D) the reason for the shift of the zone with the maximum line intensity to lower observation heights on adding hydrogen to the internal gas may lie in a faster vaporization of the particles contained in the slurries, which would also result in an earlier excitation.In Fig. 1 it can be seen that the line intensity maxima for slurries are obtained at higher observation heights than those for the solutions. The highest intensity for Al for the pure argon plasma is obtained at an observation height of 9 mm for the solution [Fig. 1(a) curve A] and at 12 mm for the slurry [Fig. 1(b) curve A]. For Si the maximum line intensity is obtained at an observation height of 10 mm for the solution [Fig. 2(a) curve A] and at 12 mm for the slurry [Fig. 1(b) curve A]. This shift of the maximum line intensity for slurries to higher observation heights could be explained by a slower evaporation of the particles of the slurry in the ICP. Hence, a longer residence time of the particles of the slurry in the plasma could improve the evaporation of refractory powders.The residence time in the ICP can be increased by a reduction of the internal gas flow, by widening the diameter of the injector gas tube and by increasing the plasma length. The eect of the hydrogen content in the internal argon gas flow on the recoveries of Al and Si as a function of the observation height is demonstrated in Fig. 3. As a result of transport losses of larger particles in the nebulization process the recoveries of Al and Si are not quantitative.It has been shown by several workers that transport eects are the most important interference eects in slurry nebulization.21–26,31 At Fig. 1 Axial intensity profiles for the Al I 396.152 nm line for solutions the optimized observation height of 10 mm, the recovery for (a) and for slurries of Al2O3 (b) at dierent hydrogen contents in the Al continuously increased from 48 to 62% and for Si from 43 internal gas: A, 0; B, 5; C, 10; and D, 15% v/v.Sample concentrations: to 52%. Because all operating parameters were held constant, 0.944 g l-1 Al2O3 (0.5 g l-1 Al) for the slurry and 0.5 g l-1 Al for the solution. the improved recoveries are likely to be the result of a more Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 983on the one hand and to calibration with aqueous solutions in slurry analysis on the other may be required. For the analysis of SiC powders this has been demonstrated to be easily possible with the aid of a simplex procedure.14 Addition of Helium to the Internal Gas As compared with argon, helium has a higher ionization potential, thermal conductivity, heat capacity and electrical resistivity. Because the ionization potential of helium (24.5 eV) is higher than that of argon (15.8 eV), the addition of helium to the working gases in ICP-AES might not only increase the heat uptake but also possibly directly the excitation in the plasma.The higher thermal conductivity of helium could lead to a faster dissipation of heat to the tubes of the torch and it is known that, with increasing amounts of helium, special torch configurations are required.35 There is evidence that the addition of helium to the argon plasma can increase the gaskinetic temperature in an ICP.32 To investigate the eect of helium addition on the evaporation of refractory materials the experiments described above with hydrogen were repeated.On adding helium up to a content of 15% v/v to the internal gas no instability of the ICP is observed and the reflected power of the ICP generator remains unchanged.At each helium addition to the internal gas, no appreciable blank for Si from the torch could be measured. In contrast to the addition of hydrogen to the internal gas, no change in the shape or decrease in the volume of the plasma could be observed. To discuss the eect of helium in the internal gas in more detail, the axial intensity profiles of Al and Si for both solutions and slurries were measured as a function of the helium content (Figs. 4 and 5). It could be concluded that (i) the line intensities for Al and Si are enhanced on adding helium to the carrier Fig. 3 Eect of the observation height and the hydrogen content in gas. As compared with a pure argon plasma the addition of the internal gas on the recoveries for Al (a) and Si (b). Recovery: ratio helium up to a content of 15% v/v leads to an enhancement of intensities for the Al I 396.152 nm line and the Si I 288.158 nm line, of the line intensities for Al and Si of 13 and 46%, respectively, respectively, for slurries and solutions.Sample concentrations: 0.944 g l-1 Al2O3 (0.5 g l-1 Al) or 0.714 g l-1 SiC (0.5 g l-1 Si) for the slurries for solutions. This enhancement is lower than that for hydrogen and 0.5 g l-1 Al and 0.5 g l-1 Si for the solutions. ecient evaporation of the slurry particles in the hydrogen– argon plasma as compared with the pure argon plasma.For Al, the recovery increases at higher values of the observation height and of the hydrogen content in the internal gas. The improvement in the recoveries at lower observation heights (i.e., 4–6 mm), however, is more significant than at higher locations in the plasma (i.e., 18–22 mm). As described by several workers,30,32 the improved ability of the hydrogen containing argon plasma to vaporize aerosol particles correlates with higher temperatures in the hydrogen containing argon ICP compared with the pure argon ICP.Ebdon and Goodall30 have reported a rotational temperature enhancement of 1700 K on adding 5% v/v hydrogen to the injector gas. Sesi et al.32 have added up to 16.7% hydrogen to the carrier gas and have shown that the gas-kinetic temperature then increases by 2000 K. They explained the higher temperatures by an increase in the energy transfer from the torus to the central channel as a result of the higher thermal conductivity of hydrogen as compared with argon.The eect of an increase in particle evaporation seems to be more significant at lower observation heights because at higher regions in the plasma the evaporation is completed to a great extent as a result of the long residence time of the particles in the ICP. It was found that the highest recoveries (62% for Al2O3 and 56% for SiC) are reached at viewing heights of 18–22 mm (Fig. 3). This region in the plasma, however, is too high for obtaining the Fig. 4 Axial intensity profiles for the Al I 396.152 nm line for solutions highest power of detection. Hence, an optimization of the (a) and for slurries of Al2O3 (b) at dierent helium contents in the plasma parameters with regard to the recovery can lead to internal gas: A, 0; B, 5; C, 10; and D, 15% v/v. Sample concentrations: higher detection limits. Therefore, a compromise optimization 0.944 g l-1 Al2O3 (0.5 g l-1 Al) for the slurry and 0.5 g l-1 Al for the solution.of the plasma parameters with respect to the power of detection 984 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Fig. 5 Axial intensity profiles for the Si I 288.158 nm line for solutions (a) and for slurries of SiC (b) at dierent helium contents in the internal gas: A, 0; B, 5; C, 10; and D, 15% v/v. Sample concentrations: 0.714 g l-1 SiC (0.5 g l-1 Si) for the slurry and 0.5 g l-1 Si for the solution. Fig. 6 Eect of the observation height and the helium content in the addition. (ii ) For helium addition the intensity maximum shifts internal gas on the recoveries of Al (a) and Si (b). Recovery: ratio of to lower positions in the plasma as for the pure argon ICP. intensities for the Al I 396.152 nm line and the Si 288.158 nm line, respectively, for slurries and solutions. Sample concentrations: 0.944 g Compared with hydrogen addition this shift is lower. The l-1 Al2O3 (0.5 g l-1 Al) or 0.714 g l-1 SiC (0.5 g l-1 Si) for the slurries intensity profiles for an addition of helium of 10% v/v (Figs. 4 and 0.5 g l-1 Al and 0.5 g l-1 Si for the solutions. and 5, curve C) and 15% v/v (Figs. 4 and 5, curve D), respectively, do not fall below the profile for the pure argon ICP (Figs. 4 and 5, curve A) at higher observation heights. (iii ) The maximum intensities for the solutions are found at phology were observed when hydrogen or helium was added lower observation heights for the slurries.In the pure argon to the internal gas of the ICP. ICP (0% v/v helium) there is a shift of the maximum intensity from 9 mm for a solution to 12 mm for a slurry for Al (Fig. 4) and from 9 mm for a solution to 11–12 mm for a slurry for CONCLUSIONS Si (Fig. 5). The eect of the addition of helium to the internal gas on For slurry nebulization ICP-AES it has been shown that limitations in the evaporation of the particles can be decreased the recoveries at dierent observation heights is demonstrated in Fig. 6. For an observation height of 10 mm, a 5–6% increase by adding hydrogen or helium to the argon plasma. Both helium and hydrogen have a greater thermal conductivity in recovery was found for both matrix elements, Al and Si. Consequently, in contrast to hydrogen addition, the recoveries compared with argon, suggesting a higher energy transfer from the plasma to the aerosol channel. Compared with helium, the for helium addition are hardly improved, as the maximum standard deviation of the recovery is 3.5%.An increase in addition of hydrogen to the internal gas of an argon ICP results in a higher signal and recovery enhancements. This both the helium content and the observation height results in an increase in the recoveries of Al from 40 to 54% [Fig. 6(a)] could be shown by axial intensity profiles for matrix element lines in solutions and slurries and by examining the eect of and for Si from 33 to 43% [Fig. 6(b)]. As discussed for hydrogen addition, the increased recoveries at higher helium hydrogen addition on the recoveries of matrix elements for slurries of ceramic powders at dierent observation heights in contents might be due to a better particle evaporation for slurries as helium has a higher thermal conductivity. However, the plasma. The addition of hydrogen to the internal gas leads to increased recoveries of the matrix elements and to line the increased recoveries at higher viewing positions are partly also due to a longer residence time of the particles in the intensity enhancements but its use is limited because the plasma volume decreases significantly and high contents of plasma.In order to investigate the evaporation behaviour of the hydrogen finally lead to plasma instability. A similar helium addition of 15% v/v produces a lower enhancement of the Al2O3 and SiC particles in the nebulized slurries, the exhaust aerosols were collected on Nuclepore filters placed 40 cm recoveries and line intensities and no plasma instabilities are observed.above the induction coil22,31 and electron probe micrographs of the particles were recorded. The particles sampled in the For further determinations of trace impurities in refractory ceramic powders the use of the matrix element as an intrinsic plasma exhaust had a spherical structure and diameters between 5 and 300 nm. These particles were well molten or internal standard is a way of correcting transportation and atomization interferences which has been described previously8 formed by condensation.In comparison with the pure argon plasma, no significant dierences in particle size and mor- and should be investigated in each particular case. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 98520 Farin� as, J. C., Moreno, R., and Mermet, J.-M., J. Anal. At. REFERENCES Spectrom., 1994, 9, 841. 1 Broekaert, J. A. C., Graule, T., Jenett, H., To� lg, G., and 21 Goodall, P., Foulkes, M.E., and Ebdon, L., Spectrochim. Acta, Part B, 1993, 48, 1563. Tscho� pel, P., Fresenius’ Z. Anal. Chem., 1989, 322, 825. 22 Raeymaekers, B., Graule, T., Broekaert, J. A. C., Adams, F., and 2 Fuller, C. W., Hutton, R. C., and Preston, B., Analyst, 1981, Tscho� pel, P., Spectrochim. Acta, Part B, 1988, 43, 923. 106, 913. 23 Saba, C. S., Rhine, W. E., and Eisentraut, K. J., Anal. Chem., 3 Ebdon, L., and Cave, M. R., Analyst, 1982, 107, 172. 1981, 53, 1099. 4 Mohamed, N., McCurdy, D. L., Wichman, M. D., Fry, R. C., and 24 McCurdy, D. L., Wichmann, M. D., and Fry, R. C., Appl. O’Reilly, J. E., Appl. Spectrosc., 1985, 39, 979. Spectrosc., 1985, 39, 984. 5 Ebdon, L., and Wilkinson, J. R., J. Anal. At. Spectrom., 1987, 2, 325. 25 Van Borm, W. A. H., Broekaert, J. A. C., Klockenka�mper, R., 6 Ebdon, L., Foulkes, M. E., Parry, H. G. M., and Tye, C. T., Tscho� pel, P., and Adams, F. C., Spectrochim. Acta, Part B, 1991, J.Anal. At. Spectrom., 1988, 3, 753. 46, 1033. 7 Laird, D. A., Dowdy, R. H., and Munter, R. C., J. Anal. At. 26 Ebdon, L., Foulkes, M. E., and Hill, S., J. Anal. At. Spectrom., Spectrom., 1990, 5, 515. 1990, 5, 67. 8 Ebdon, L., and Collier, A. R., J. Anal. At. Spectrom., 1988, 3, 557. 27 Barnes, R. M., and Schleicher, R. G., Spectrochim. Acta, Part B, 9 Ebdon, L., and Collier, A. R., Spectrochim. Acta, Part B, 1988, 1974, 30, 109. 43, 355. 28 Chen, X., and Pfender, E., Plasma Chem.Plasma Process., 1982, 10 Halicz, L., and Brenner, I. B., Spectrochim. Acta, Part B, 1987, 2, 185. 42, 207. 29 Verbeek, A. A., and Brenner, I. B., J. Anal. At. Spectrom., 1989, 11 Darke, S. A., Long, S. E., Pickford, C. J., and Tyson, J. F., 4, 23. Fresenius’ J. Anal. Chem., 1990, 337, 284. 30 Ebdon, L., and Goodall, P., J. Anal. At. Spectrom., 1992, 7, 1111. 12 Broekaert, J. A. C., Lathen, C., Brandt, R., Pilger, C., Pollmann, D., 31 Xhoer, C., Lathen, C., Van Borm, W., Broekaert, J. A. C., Tscho� pel, P., and To� lg, G., Fresenius’ J. Anal. Chem., 1994, 349, 20. Jacob, W., and Van Grieken, R., Spectrochim. Acta, Part B, 1992, 47, 155. 13 Graule, T., von Bohlen, A., Broekaert, J. A. C., Grallath, E., 32 Sesi, N. N., MacKenzie, A., Shanks, K. E., Yang, P., and Hieftje, Klockenka�mper, R., Tscho� pel, P., and To� lg, G., Fresenius’ Z. G. M., Spectrochim. Acta, Part B, 1994, 49, 1259. Anal. Chem., 1989, 335, 637. 33 Murillo, M., and Mermet, J. M., Spectrochim. Acta, Part B, 1989, 14 Broekaert, J. A. C., and To� lg, G., Mikrochim. Acta, 1990, II, 173. 44, 359. 15 Pollmann, D., Pilger, C., Hergenro� der, R., Leis, F., Tscho� pel, P., 34 Schramel, P., Fischer, R., Wolf, A., and Hasse, S., Fresenius’ Z. and Broekaert, J. A. C., Spectrochim. Acta, Part B, 1994, 49, 683. Anal. Chem., 1984, 319, 229. 16 Docekal, B., Broekaert, J. A. C., Graule, T., Tscho� pel, P., and 35 Abdalla, H., and Mermet, J. M., Spectrochim. Acta, Part B, 1982, To� lg, G., Fresenius’ J. Anal. Chem., 1992, 342, 113. 37, 391. 17 Lobinski, R., Broekaert, J. A. C., Tscho� pel, P., and To� lg, G., Fresenius’ J. Anal. Chem., 1992, 342, 569. Paper 7/00464H 18 Lobinski, R., Van Borm, W., Broekaert, J. A. C., Tscho� pel, P., Received January 21, 1997 and To� lg, G., Fresenius’ J. Anal. Chem., 1992, 342, 563. 19 Farin� as, J. C., ICP Inf. Newsl., 1993, 11, 691. Accepted April 11, 1997 986 Journal of Analytical At

ISSN:0267-9477    

DOI:10.1039/a700464h

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Determination of Arsenic by Inductively Coupled Plasma Mass Spectrometry in Mung Bean Seedlings for use as a Bio-indicator of Arsenic Contamination† KRISTEL VAN DEN BROECK*a , CARLO VANDECASTEELEa AND JAN M. C. GEUNSb aDepartment of Chemical Engineering, University of L euven, de Croylaan 46, 3001 Heverlee, Belgium bL aboratory of Plant Physiology, Kard.Mercierlaan 92, 3001 Heverlee, Belgium Mung bean seedlings can be used as a bio-indicator of As plant parts to determine the critical toxic threshold level above which the normal plant synthesis becomes inhibited.To contamination. Endogenous As concentrations in the bioindicator were determined by ICP-MS. Matrix eects and measure the endogenous concentrations, ICP-MS was the chosen analytical technique. polyatomic interferences originating from the plant matrix and their correction procedures were studied. Experiments were Organic matrices such as plant material may present several problems for measurements using ICP-MS.The plant material carried out with arsenate and arsenite additions of 0.5 up to 50 mm of As to the growth medium. The arsenate uptake has to be dissolved by an acid digestion procedure before analysis. Incomplete digestion of the organic material (carbon) increased with increasing arsenate concentrations in the growth medium, the highest accumulation occurred in the roots. may cause matrix eects. Possible occurrence of chloride can cause interferences of ArCl on the As signal.These analytical Arsenite seemed to be far more toxic than arsenate ( lethal dosage at <10 mm). Phosphate was shown to lower the toxic problems and the procedures used to overcome these problems will be discussed. eects of arsenate but had no eect on the arsenite uptake. As an illustration, the endogenous arsenate concentration was correlated with the polyamine synthesis. It was shown that EXPERIMENTAL above 2.2 mg g-1 of As in the dry mass the seedlings were in a ‘stress’ situation.Growing Plant Material Keywords: Bio-indicator; As contamination; polyatomic Mung bean seeds (V igna radiata L ., Wilczek) were chosen interferences; matrix eects; inductively coupled plasma mass because they are very sensitive to dierent sources of pollution. spectrometry; biological material They also oer the advantage of growing very rapidly, so that the eect of pollution can be evaluated in a short period of time (6 d). A test always started with 100 seeds which were In the present paper a study is made of a bio-indicator used germinated on wet filter paper in Petri dishes (20 cm diameter) to evaluate the degree of pollution of water and soil.A bioat 25 °C in the dark. After 3 d, seedlings with a root length of indicator can be defined as an organism [plants, animals (even about 30 mm were selected to grow further on a special growth men can act as a bio-indicator), or bacteria] that reacts upon medium called ‘Hoagland solution’, which contained all the a high pollution level by changes in the metabolic system and essential nutrients.The composition of this growth medium is preferably with visual changes. To be selected as a biogiven in Table 1. Additions of 0.5 up to 50 mM of As (arsenate indicator, the organism has to fulfil a series of conditions such or arsenite) were done to the ‘Hoagland solution’ to simulate as reproducibility of the reaction, specificity of the reaction to an As contamination.A test with a constant As concentration a certain pollutant and sucient sensitivity. The normal con- (10 mM) and increasing additions of phosphate (0 up to 500 mM dition of the organism must, of course, be known exactly. Na2HPO4) was also set up. The seedlings were planted on the Plants are well suited for this purpose because they are already Hoagland solution by pulling the roots through 2 mm holes connected to a certain environment and are usually available in a 3 mm thick polystyrene disc. The discs were floated on in large amounts (so that statistical measurements are possible). 250 ml of solution in 500 ml polyethylene beakers, that were The use of a bio-indicator has some advantages compared kept under a 16 h fluorescent light regime at 25 °C. In this way with physico-chemical methods, such as lower cost. The most only the roots of the seedlings were in direct contact with important dierence is the fact that the bio-indicator gives the solution. direct information on the biologically available fraction of the pollutant and its impact on the environment.In this work the Table 1 Composition of the ‘Hoagland solution’ eect of one toxic component, As, was studied but the overall purpose of this research was to develop a test system that is Component Concentration useful for a wide range of heavy metals. The plant material H3BO3 1.430 mg l-1 used was Mung bean seedlings. MnCl2·4H2O 0.9 mg l-1 To evaluate the eect of the pollutant on the bio-indicator CuSO4·5H2O 0.04 mg l-1 several parameters such as fatty acids, growth, pigments, sterols H2MoO4·H2O 0.045 mg l-1 and polyamines were measured.These parameters had to be ZnSO4·H2O 0.11 mg l-1 related to the exogenous As concentration, i.e., the concen- KNO3 102 g l-1 tration of As in the growth medium, and to the endogenous Ca(NO3)2·4H2O 83.6 g l-1 NH4H2PO4 23 g l-1 As concentration, i.e., the concentration of As in the dierent MgSO4·7H2O 49gl-1 FeSO4·7H2O 0.43 mg l-1 † Presented at the 1997 European Winter Conference on Plasma Na2EDTA 0.30 mg l-1 Spectrochemistry, Gent, Belgium, January 12–17, 1997.Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (987–991) 987Lyophilisation and Dissolution of the Plant Material Because it is impossible to simulate the plant matrix, control samples (appropriate parts of seedlings grown on a growth The length of the roots, hypocotyls and epicotyls were measmedium without As) were used as the sample matrix.Two ured under dierent conditions to evaluate the growth as a additions of a diluted As standard solution (1 g l-1 stock function of the arsenate concentration in the growth medium. solution, Merck), giving 10 and 100 mg l-1 in the standard, The seedlings were then cut into 6 dierent parts, namely the were made to the control sample and were used as calibration root, upper half of the hypocotyl, lower half of the hypocotyl, solutions.The control sample itself (without additions of As) epicotyl, leaves and cotyledons. The dierent parts of a seedling acted as a blank solution for the standards. A blank containing are shown in Fig. 1. These parts were then freeze-dried, milled the same amount of nitric acid as used during the dissolution and stored in a desiccator before further analysis. Lyophilised process was subtracted from all samples. The samples and and milled plant material gave, for one test, dry samples of standards were transfered into 10 ml sample cuvettes and 200–400 mg (depending on the part of the plant).After accurate 100 ml of indium (final concentration of 100 mg l-1) and 100 ml determination of the mass, these samples were then transferred of nitric acid (65%, Suprapur, Merck) were added. This method into a 20 ml Pyrex beaker and 2 ml of nitric acid (Merck, was used to determine the endogenous As concentrations in Suprapur, 65%) were added.The beakers were then covered the seedlings grown on solutions with increasing arsenite and with a watch glass and gently heated on a heating plate for arsenate contaminations as well as for the test with phosphate. about 1 h. When the solution became colourless, indicating The possible occurrence of Cl in the plant matrix can cause that most of the organic material had been digested, it was polyatomic interference of 40Ar35Cl on 75As. Because As is tranferred into a 50 ml volumetric flask.Two ml of nitric acid mono-isotopic, it is not possible to choose another isotope to were added to the flasks and the volume was adjusted to the overcome this interference. The only suitable alternative is mark with milli-Q water (Millipore). mathematical correction or matrix separation. The mathematical correction for 40Ar35Cl used in this work is based upon the fact that the ratio of the signal for 40Ar35Cl to that for ICP-MS Measurements, Matrix Eects and Spectral 40Ar37Cl will be equal to the ratio of the abundance of the two Interferences isotopes 35Cl and 37Cl, which is 75.77524.23.Measuring the The ICP-MS instrument used in this work was a PlasmaQuad signal for 40Ar37Cl at m/z 77 (77Se) makes it possible to calculate PQ2+ (VG Elemental) equipped with a conventional the signal for 40Ar35Cl present at m/z 75 (75As). Correction for Meinhard nebuliser for sample introduction. The instrument the possible occurrence of selenium in the plant matrix will be was optimised for As measurements.To correct for instrumenbased on the 82Se signal. The following equation was therefore tal drift during analysis, indium was added to all samples and used in this work: standards as an internal standard. 75As=75signal-3.065[77signal-(0.8484 82signal )] When analysing real samples with ICP-MS, matrix eects play an important role.1 In the case of organic matrices The mathematical correction was applied to all results containing large amounts of carbon, e.g., plant material, a mentioned in this work.special eect is observed for elements such as As with high ionisation energies (9.82 eV). The carbon content in the plant material can cause signal enhancement. To correct for this RESULTS AND DISCUSSION problem standard additions or matrix matching can be applied. Matrix Eect on Arsenic Originating From Carbon In the first method, standard additions involves the addition of known concentrations (usually four) of As to the sample.To illustrate the eect of carbon on the As signal, an experiment was set up with increasing additions of CH3OH to a standard The calibration curve obtained when plotting the element signal as a function of the added concentration is used to solution containing 10 mg l-1 of As and In. The results for this experiment are given in Fig. 2 and show clearly that high C calculate the concentration of As in the sample. This method was used to determine the As concentration in the ‘Hoagland concentrations cause signal enhancement. It appears also from Fig. 2 that In as an internal standard does not allow correction solution’ and in the seeds. The second correction method, matrix matching is based upon the fact that standards and for the eect of C because it behaves totally dierent to As. The stronger enhancement of As is probably due to the blank solutions are made up in the same matrix as the sample.high ionisation energy of As (9.82 eV) compared with In (5.8 eV), which results in As being a hard-to-ionise element under standard ICP-MS running conditions. The C present in CH3OH will probably give a more energetic plasma and will make ionisation more ecient.2 Of course the C concentration Fig. 1 Dierent parts of the Mung bean seedlings. Fig. 2 As and In signal as a function of the CH3OH content. 988 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12in the sample solution will be well below 1%, but even then a that increasing arsenate concentrations inhibit the growth process of the seedlings, the greatest eect occurring on the significant matrix eect may occur.roots. The number of side roots as a function of increasing arsenate concentrations were counted as well and also showed ArCl Interference on Arsenic a decrease. At concentrations above 50 mM of arsenate in the growth medium, the occurrence of side roots was rare and if To illustrate the interference of ArCl on As, increasing amounts side roots do occur, they were coloured black.of hydrochloric acid (Merck, suprapur, 32%) were added to a blank solution, containing only ultra pure water (milli-Q, Millipore). The apparent As concentration as a function of the Arsenic Uptake (Endogenous Arsenic Concentration) as a Cl concentration in the sample is shown in Table 2. The Cl Function of the Exogenous AsV Concentration content in plant material is approximately 0.14% of dry mass, After dissolution of the lyophilised plant material, the endogen- which corresponds to 5 up to 10 mg l-1 of Cl in the solutions ous concentration of As in the dierent plant parts was after dissolution and dilution, which is high enough to give a measured with ICP-MS.As mentioned before, mathematical considerable interference on As. correction procedures to eliminate the interference of ArCl on As and the use of a control sample as the ‘sample matrix’ to Determination of Arsenic in the Growth Medium and Seeds make up the calibration solutions were necessary.The results for the endogenous As concentration in the root as a function The impurities of As in the ‘Hoagland solution’ and seeds were of the exogenous arsenate concentration, both for ‘undiluted’ determined to explain the possible occurrence of As in the and ten-fold diluted samples are shown in Table 3. The two control sample (seedlings grown on an arsenate-free ‘Hoagland sets of results are in good agreement.For comparison the solution’). As mentioned before, standard additions to overresults for ‘undiluted’ samples without correction for ArCl come the matrix eects and mathematical correction to elimininterference and using aqueous standards for calibration are ate the ArCl interference were applied. The ‘Hoagland solution’ also given. These results are all significantly higher than when contained 0.77 mg l-1 of As and the seeds 0.36 mg g-1.The appropriate correction and calibration methods are applied, rather high concentration of As in the seeds and cotyledons the relative dierence is however higher for low concentrations, could be explained by the use of disinfectants for the seeds. probably due to the ArCl interference. The distribution of As in the dierent plant parts as a Growth as a Function of Exogenous and Endogenous Arsenate function of increasing arsenate concentrations in the growth Concentration solution is shown in Fig. 4(a). It can be concluded here that most of the As is accumulated in the roots, which is probably The seedlings were grown on a ‘Hoagland solution’ containing due to the fact that the roots are in direct contact with the increasing concentrations of arsenate (0.5, 1, 5, 10, 20 and contaminant. 50 mM). Before lyophilisation and cutting the seedlings into the Plotting the length increase of the roots as a function of the six dierent parts, the lengths of the root, hypocotyl and inverse of the endogenous concentration allowed the critical epicotyl were measured.The lengths of these parts as a function toxic threshold to be determined, defined as the crossing point of the increasing arsenate concentration in the growth medium of the best fitting curve through these points (excluding the are shown in Fig. 3. From these results it can be concluded points where no changes occurred compared with the control sample) with the dashed horizontal (no eect) curve, above Table 2 Apparent As concentration as a function of increasing Cl which growth became inhibited [Fig. 4(b)]. For the roots this concentrations threshold value is about 1.23 mg g-1 of As in the dry mass. Cl concentration/g l-1 As concentration/mg l-1 0.0035 0.29 Arsenic Uptake (Endogenous Arsenic Concentration) as a 0.007 0.33 Function of the Exogenous AsIII Concentration 0.0175 0.99 0.035 2.6 The plants were totally damaged after exposure to a ‘Hoagland 0.07 5.7 solution’, with AsIII additions in the same concentration range 0.175 13.9 of the arsenate additions, which proves that AsIII is eectively 0.35 26.2 more toxic than AsV. 0.7 54 1.75 122 3.5 231 Influence of Increasing Phosphate Concentrations on the Arsenic 7 386.74 (Arsenate) Uptake 17.5 837 It has been shown that As exists as arsenolipids and As phospholipids in marine algae. Because of its chemical similarity, As can replace phosphorus, particulary in phosphate deficient waters and can be accumulated by algae.3 Experiments with increasing additions of phosphate (0, 50, 250 and 500 mM) to a phosphate-free ‘Hoagland solution’ which contained a constant arsenate concentration of 10 mM, were carried out to find a possible explanation for the uptake mechanism for arsenate in the Mung bean seedlings.It is shown clearly in Fig. 5, which gives the endogenous As concentration in the roots as a function of the NaH2PO4 concentration in the growth medium, that phosphate inhibits the arsenate uptake, which means that phosphate lowers the toxic eect of arsenate.The endogenous concentration of As was determined using the same correction procedures as mentioned above and is Fig. 3 Growth of seedlings as a function of the arsenate concentration in the growth medium. only shown for the roots because these parts of the seedlings Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 989Table 3 Results for the endogenous As concentration (with/without correction) in the root as a function of increasing arsenate concentrations in the growth medium Endogenous As concentration/mg g-1 of As in dry mass 1+9 dilution Undiluted Undiluted As concentration in medium/mM with correction* with correction* no correction 0.5 1.318±0.014 1.1010±0.0074 1.93±0.013 1 2.261±0.017 2.2020±0.0101 3.069±0.014 5 4.734±0.016 4.590±0.017 5.64±0.021 10 33.5±1.8 35.290±0.103 37.60±0.11 20 60.85±0.15 63.59±0.34 66.86±0.36 50 149.24±0.61 151.04±0.56 157.93±0.58 * Use of ‘control sample’ for the calibration solutions and mathematical correction for the ArCl interference.Fig. 6a Polyamine content as a function of the arsenate concentration in the growth medium; and (b) polyamine content in the roots as a Fig. 4 (a) Uptake of arsenate in the dierent plant parts as a function function of the inverse of the endogenous As concentration. of the arsenate concentration in the growth medium; and (b) growth of the Mung bean roots as a function of the inverse of the endogenous As concentration.Polyamine Content as a Function of the Increasing Arsenate Concentrations in the Growth Medium Three polyamines, namely putrescine, spermine and spermidine, were measured as a function of increasing arsenate concentrations in the growth medium (the same conditions as mentioned above). It is shown in Fig. 6(a) shows that the putrescine content in the roots increased from about 400 nmol g-1 dry mass in the controls up to 2300 nmol g-1 dry mass, whereas the spermine and spermidine concentrations did not change significantly. Putrescine accumulation, especially in cereals, has been shown to occur in response to a large variety of stress factors.4,5 Under the influence of heavy metals, plants undergo an accelerated ageing process.As a result of this, the putrescine content in the plant will increase, probably to inhibit the ageing process.Plotting the root putrescine content as a function of the inverse of the endogenous As concentration Fig. 5 As uptake (expressed as mg of As per g of dry mass as well as mg of As per root) as a function of the NaH2PO4 concentration in allowed the threshold value above which putrescine increase the growth medium. started to be calculated [Fig. 6(b)]. The crossing of the best fitting curve through the measuring points, excluding the points without changes compared with the control sample, are more aected than the others; similar conclusions can however be made for the other parts.Measurements of the with the dashed horizontal (no eect) curve gives a value of 0.46, which corresponds to an endogenous As concentration mass and length of the seedlings also showed that higher phosphate concentrations in the growth solution stimulate the of 2.2 mg g-1 dry mass. This value is rather close to the value found for growth inhibition in the roots (1.23 mg g-1 dry mass).growth of the seedlings and lower the toxic eects of arsenate. 990 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12At a certain concentration (>20 mM, the putrescine synthesis synthesis as a function of the endogenous As concentration it can be concluded that above 2.2 mg of As per g of dry mass, did not increase further, indicating that at this concentration level the whole plant metabolic system was inhibited. The the plant appears to be in a ‘stress’ situation. In the near future speciation studies will be carried out to results for the polyamine content in the other plant parts were similar but not as pronounced (in relation to the endogenous give a possible explanation for the uptake mechanism of As and possible transformations or detoxification reactions in the As concentration in these plant parts).plant. Hydride generation with cold trapping as well as HPLC coupled on-line with ICP-MS will be used for this purpose. CONCLUSION The authors thank Liesbeth Smet and An Hacour for their ICP-MS is a well suited analytical technique to determine the collaboration in this research. endogenous concentration of As in plant material when taking into account all the necessary corrections. From the results REFERENCES obtained it can be concluded that the endogenous concen- 1 Vandecasteele, C., Vanhoe H., and Dams R., J. Anal. At. Spectrom., tration increases and the growth is inhibited with increasing 1993, 8, 781. concentrations of arsenate in the growth medium, with the 2 Larsen, E. H., and Stu�rup, S., J. Anal. At. Spectrom., 1994, 9, 1099. highest accumulation of arsenate in the roots. Growing the 3 Morrison, G. M. P., Batley, G. E., and Florence, T. M., Chem. bio-indicator on a growth medium with 10 mM of arsenate Brit., 1989, 25, 791. gives an endogenous As concentration of about 35 mg of As 4 Galston, A. W., and Sawhney R. K., Plant Physiology, 1990, 94, 406. per g of dry mass in the roots while at 10 mM of arsenite the 5 Smith, T. A., Annu. Rev. Plant Physiol., 1985, 36, 117. bio-indicator was totally damaged, which indicates that AsIII is far more toxic than AsV. Experiments with phosphate Paper 7/01610G additions to the growth medium showed that phosphate lowers Received March 7, 1997 Accepted May 2, 1997 the toxic eect of arsenate. From the results of the putrescine Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 9

ISSN:0267-9477    

DOI:10.1039/a701610g

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

The Plasma Emission Detector—A Suitable Detector for Speciation and Sum Parameter Analysis† B. ROSENKRANZ*, C. B. BREER, W. BUSCHER, J. BETTMER AND K. CAMMANN Institut fu� r Chemo- und Biosensorik e.V.,Mendelstraße 7, D-48149 Mu�nster, Germany The technical arrangement and the fundamental principles of a Speciation of Organomercury Compounds plasma emission detector (PED) are presented. Based on Information on speciation is vital in ecotoxicological studies oscillating interference filters, the PED is able to identify of the natural production of organic forms of elements in the dierent spectral positions, which allow the measurement of environment, of the fate of organometallic anthropogenic emission line intensities together with the accompanying pollutants (degradation, partition in biota and long-range background.To demonstrate the range of possible applications, transportation) and of transformation of organoelements via coupling of the PED with two plasma sources (MIP and biotic processes in biogeochemical cycles.8,9 Speciation governs CMP) and the analytical principles [adsorbable organohalogen the bioavailability, metabolism and toxicity of trace metals compounds (AOX) and Hg speciation] are described.which are of interest in the agrochemicals industry and in Furthermore, the analytical performance is demonstrated by clinical toxicology. Concerns over the contamination of aquatic statistical treatment of multi-element calibrations and by the edible resources by mercury and methylmercury formed in situ determination of methylmercury (MeHg) in CRMs 463 and date back to the Minamata accident.10 464 from the European Commission’s Standard, Measurement The aim of the work presented in this paper was the coupand Testing Programme.ling of capillary GC with an MIP detection system for the speciation of mercury. Keywords: Organomercury compounds; sum parameter; adsorbable organohalogen compounds; extractable organohalogen compounds; atomic emission spectrometry; EXPERIMENTAL microwave-induced plasma; capacitively coupled microwave Reagents and Standard Solutions plasma; gas chromatography; hyphenated technique Acetic acid (100% pro analysi), methanol ( pro analysi ), sodium hydroxide ( pro analysi ), activated charcoal (for AOX determi- Hyphenated techniques are important and powerful tools for nation), p-chloro-phenol and nitric acid, (65% pro analysi ) the determination of organometallic and organohalogen comwere obtained from Merck (Darmstadt, Germany); sodium pounds (OHC).In the area of atomic emission spectrometry tetraphenylborate (STPB) and 4-bromo-phenol were obtained in particular the combination of a separation system such as from Fluka (Buchs, Switzerland); methylmercury chloride gas chromatography or a cryofocusing unit with element ( pro analysi ) was obtained from Alfa Ventron (Karlsruhe, selective detectors have often been applied.1,2 In the present Germany).Clean water was produced with a Seralpur PRO paper the combination of a laboratory-developed plasma emis- 90 CN system (Ransbach, Germany). Liquid nitrogen sion detector (PED)3–5 with two dierent types of plasma (>99.999%) was obtained from Westfalen AG (Mu� nster, sources for the determination of organomercury compounds Germany) and p-fluoro-phenol p.S. from Riedel-de Haen and OHC compounds is presented. An atmospheric pressure (Hannover, Germany).The CRMs 463 and 464 (tuna fish helium microwave-induced plasma (MIP) and a capacitively samples) were obtained from the Standard, Measurement and coupled microwave plasma (CMP) were thus combined with Testing Programme (Brussels, Belgium). the PED to show the flexible operating possibilities of this detector. Technical Set-up and Fundamental Principles of the PED Determination of OHC Sum Parameters The separation of the wavelength of interest is realized by oscillating narrow banded interference filters. Depending on To control the pollution of water and soils by OHC several the angle between the radiation and the surface of the filter, analytical methods have been developed.Sum parameters the maximum the transmission is shifted is described by the allow a quick determination, avoiding the determination of following equation:11 single substances, which requires more sophisticated techniques. The most important sum parameters of the OHC are l(h)=l S1- Ane n*B2 sin2h (1) adsorbable (AOX) and extractable (EOX) organohalide compounds. 6,7 These methods have the separation from the inorwhere l(h is the maximum transmission dependent on the ganic compounds and the mineralization of organic substances angle (h) between the radiation and the filter surface; l is to hydrogen halides (HX) in common. The usual (commercial) maximum transmission for h=0; ne is the refractive index of way to determine these HX compounds is microcoulometry, air; and n* is the refractive index of the filter.wherein all halides are considered to be chlorine. The use of Observing the intensity of transmitted radiation during an plasma emission spectrometry oers the advantage of element oscillation cycle leads to the measurement of dierent spectral selective information. positions (see Fig. 1). The optical signals are transformed into current by photomultipliers and later processed by lockin- amplifiers. In this way, ecient background correction is † Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997.achieved without using expensive mono- and polychromators. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (993–996) 993Fig. 1 Fundamental principle of the PED. Description of the System for AOX and EOX Parameters for operating the analyser and the CMP for AOX are shown in Table 1. The construction of the analyser is shown in Fig. 2.12 It consists of three major components according to the three steps in the analysis process. Depending on the method (EOX or AOX) Description of the System for Mercury Speciation an easy change of the sample port is required. The complete system for mercury speciation is shown in Fig. 3. In step 1, the analytes, either enriched in an organic solvent There are three main technical components in the instrument. (in the case of EOX) or adsorbed onto activated charcoal, can For extraction of the hydrophobic species (methylmercury be mineralized in a newly developed combustion unit.Using chloride, for example, was converted into methylphenylmercury two independent streams of oxygen, the sample is burnt by derivatization with STPB at pH 4.75)13 solid phase micro- completely, without any residue (soot). A new micro-condensor extraction (SPME) was used successfully with a polydimethyl- (length 50 mm, diameter 12 mm) utilizes the idea of the siloxan (PDMS) phase.14 The commonly used ethylation was Vigreaux distillation column and so guarantees complete not used because of the high reactivity of the reagent itself, in separation of carbon dioxide.The removal of carbon dioxide order to ensure ecient action of the PDMS coating. In the is necessary as it disturbs the plasma. next step the absorbed analytes were thermally desorbed in During step 2 the micro condensor is heated to 100 °C. With the GC injector at a temperature of 250 °C.After separation a helium flow the steam is lead through a drying unit, filled of the dierent organometallic compounds in the gas chromato- with concentrated sulfuric acid (volume 30 ml), to remove the graph, the analytes were excited in the atmospheric pressure water, which also disturbs detection. The HX are frozen out helium MIP and the emission lines were sent to the PED via in a cryofocusing unit (length 100 mm) cooled down to fibre optics.Further technical data of the GC and the MIP as -196 °C by liquid nitrogen. well as for the extraction procedure are given in Table 2. Step 3 consists of heating the cryofocusing unit to 140 °C using the stopped-flow method and carrying the now gaseous HX into a CMP where all substances are atomized and excited to emit their elemental characteristic radiation. The opto- Table 1 Technical data for the determination of AOX elic treatment of this radiation is achieved with the PED.Step 1 Combustion— Capillary gas Oxygen, 300 ml min-1 Casing gas Oxygen, 30 ml min-1 Time 11 min Step 2 Cryofocusing— Gas flow Helium, 100 ml min-1 Time 6 min Step 3 Excitation in the plasma— CMP Constructed by AHF, Tu� bingen, Germany Microwave generator MWG 600 CRL by AHF Aerosol carrier gas Helium, 30 ml min-1 Intermediate plasma gas Helium, 1.5 l min-1 Outer gas Helium, 1.32 l min-1 Forward power 320 W Reflected power 38 W Detection of plasma radiation— Maximum transmission Fig. 2 Element selective AOX/EOX analyser (here with EOX sample of interference filter F 685.50 nm, Cl 479.55 nm, Br 478.8 nm port). 1, EOX injector with electric motor and syringe; 2, oven Halfwidth 0.3 nm (all filters) (1000 °C); 3, combustion unit with PE net; 4, oxygen supply; 5, helium Angle (radiation, filter) 4° supply; 6, 9 and 11, three-way valve; 7, micro-condenser; 8, drying Amplitude of oscillation ±5° unit; and 10, cryofocusing unit. 994 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Fig. 4 Calibration of analyser for AOX. Fig. 3 System for the speciation analysis with SPME, GC and MIP. 1, SPME; 2, MIP; 3, microwave generator; 4, gas chromatograph; 5, Table 3 Linear regression for the calibration of dierent OHC injector; 6, optical lens; 7, fibre optics; and 8, quartz take. Parameters for linear regression AOF AOCl AOBr Sensitivity/mm mg-1 9.14 7.66 10.2 RESULTS AND DISCUSSION sy/mm 13.5 10.6 6.8 AOX/EOX sx0/mg 1.48 1.38 0.7 sx0(%) 3.0 4.1 2.0 For an initial investigation of the analytical performance Correlation coecient, r 0.9999 0.9998 0.9999 of the newly developed analysis system, a calibration with Limit of detection/mg 1.5 1.1 1.2 p-chloro-phenol, p-fluoro-phenol and p-bromo-phenol was performed.15 points to improve the analytical performance.Moreover, the The results are shown in Fig. 4 and in Table 3. eect of high organic contents in samples with a small AOX Four real waste water samples were obtained from a comcontent should be examined.mercial analytical laboratory (E.Wessling, Mu� nster, Germany). These samples, P1–P4, were each investigated three times with the new AOX analyser. Samples were treated according to Speciation ref. 7. AOF, AOCl and AOBr were assumed to constitute the AOX sum. The mean values were later compared with the To demonstrate the capabilities of the system two certified reference materials (CRM 463 and CRM 464) were investi- results from Wessling.The results are shown in Table 4. Comparison of the present data and reference values proved gated. For the preparation of the sample, the material was treated with 10% (m/m) sodium hydroxide solution to remove to be in very good accordance, even considering the exclusion of AOI. Future research will include investigation into recover- the methylmercury. In the next step the ionic mercury compound was alkylated using STPB at pH 5 (adjusted with acetic ies of each single step in the process in order to eliminate weak Table 2 Technical data for the separation of the mercury compounds GC parameters— Gas chromatograph Hewlett-Packard (Avondale, PA, USA) 5890 A Column Ultra-1, 25 m×0.32 mm i.d.×0.52 mm film thickness Column head pressure 100 kPa helium Injection port temperature 250 °C Injection Splitless Purge value, o/on 0–2/2–7 min Oven programme— Initial temperature 60 °C Initial time/final time 1/2 min Ramp rate 35 °C/min Final temperature 245 °C Interface parameter— Transfer line Ultra-1 Transfer temperature 250 °C Extraction parameters for SPME— Fibre Dimethylsiloxan, 100 mm (Supelco, Deisenhofen, Germany) Extraction temperature 40 °C (kept in an ultrasonic bath) Time of extraction 10 min Detection parameter— MIP Modified with Beenaker TM010 cavity (AHF) Generator GMW 24–301 DR (AHF) Microwave power 75 W (reflected power 3 W) Helium gas flow 135 ml min-1 Central wavelength 253.65 nm Angle of oscillation 4 ° at 20 Hz Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 995Table 4 Element selective determination in four real samples. Results ture.16 Regression data (standard additions method) for the in mg l-1 two CRMs are given in Table 5. This shows that sensitivity is strongly dependent on the matrix being analysed, which should Sample be recognized for every type of sample, such as hair samples, No. AOF AOCl AOBr AOX AOXmean AOX (ref.)* as has been demonstrated in earlier studies.17 P 1 <LOD 120.7 11.3 125.7 120±10 110±? <LOD 115.9 5.36 118.3 <LOD 122.6 5.8 125.2 CONCLUSIONS P2 <LOD 233.6 24.4 244.4 240±31 240±? <LOD 196.6 19.6 205.3 The PED is of simple design, but is a powerful device for <LOD 254.1 26.0 265.6 detection and background correction in analytical atomic P3 <LOD 50.6 <LOD 50.6 50±7 66±? spectrometry. It shows high reproducibility and selectivity and <LOD 45.2 <LOD 45.2 is suitable for detecting elements even in complex matrices.<LOD 53.9 <LOD 53.9 The AOX/EOX-CMP-PED analysis system oers the P 4 <LOD 529 <LOD 529 530±55 430±? advantage of a single and independent trace determination of <LOD 570 <LOD 570 <LOD 498 <LOD 498 the OHCs that together build up the AOX or EOX sum, plus the determination of extractable organofluorine compounds, * No data of uncertainty available. which is impossible with commercial analysers.The application of a PED for the determination of organomercury compounds has been presented. The combination of SPME, GC and the PED led to a powerful and sensitive system for speciation analysis. The time of analysis could be reduced to 15 min, including the whole extraction procedure. With the powerful background correction of the PED in combination with GC separation, analysis of even complex real samples, such as fish can be carried out. The authors thank the Deutsche Forschungsgemeinschaft DFG for financial support of this work and the Ministerium fu� r Wissenschaft und Forschung des Landes Nordrhein-Westfalen for the support of the Institut fu� r Chemo- und Biosensorik, Mu� nster e.V.(ICB). REFERENCES 1 Weber, H., and Puk, R., Appl. Organomet. Chem., 1994, 8, 293. 2 Lobinski, R., Analusis, 1994, 22, 37. 3 Cammann, K., and Mu� ller, H., Fresenius’ Z. Anal. Chem., 1988, 331, 336. 4 Mu� ller, H., and Cammann, K., J. Anal. At. Spectrom., 1988, 3, 907.Fig. 5 Chromatograms of the CRMs 463 and 464. 5 Element-Specific Chromatographic Detection by Atomic Emission Spectroscopy, ed. Uden, P. C., ACS Symposium Series 479, Table 5 Standard additions treatment of CRMs 463 and 464 ( linear American Chemical Society, Washington, DC, 1992. regression) 6 Deutsche Norm DIN 38409, Section H, Part 8, 1984. 7 Deutsche Norm DIN 38409, Section H, Part 14, 1985. Parameters of linear regression CRM 463 CRM 464 8 Krull, I. S., T race Metal Analysis and Speciation, Elsevier, Amsterdam, 1991.Sensitivity/mm mg-1 7.25 4.98 9 Harrison, R. M., and Rapsomanikis, S., Environmental Analysis sy/mm 5.25 3.50 Using Chromatography Interface with Atomic Spectroscopy, sx0/mg 0.73 0.71 ed. Ellis Horwood, Chichester, 1989. sx0(%) 1.15 2.88 10 Metals and their Compounds in the Environment, ed. Merian, E., Correlation coecient, r 0.998 0.998 VCH Verlagsgesellschaft, Weinheim, 1991. 11 Naumann, H., and Schro� der, G., Bauelemente der Optik: T aschenbuch der technischen Optik, Hanse-Verlag, Munich, 6th acid).In Fig. 5 a typical chromatogram of this sample is edn., 1992. shown. The retention time for methylphenylmercury (PMM) 12 Deutsches Patent 4309045, 1997. is 4.65 min (±0.02 min). The first peak is for hydrophilic 13 Mena, M. L., McLeod, C. W., Jones, P., Withers, A., Minganti, compounds resulting from the sample preparation. The chrom- V., Capelli, R., and Quevauviller, Ph., Fresenius’ J. Anal. Chem., 1995, 351, 456. atogram shows no interferences from other compounds, 14 Pawliszyn, P., Louch, D., and Motlagh, S., Anal. Chem., 1992, although it must be mentioned that high concentrations of 64, 1187. tetraphenylborate genyl, which may result in a 15 International Union of Pure and Applied Chemistry (IUPAC) peak at 5.05 min, but nevertheless the selectivity of Hg against Recommendations, Pure Appl. Chem., 1995, 67, 1699. C was 1510 000. This can be explained by insucient blocking 16 Quevauviller, Ph., Appl. Organomet. Chem., 1994, 8, 715. of the carbon emission line at 247.86 nm (central wavelength 17 Bettmer, J., Bradter, M., Buscher, W., Erber, D., Rieping, D., and Cammann, K., Appl. Organomet. Chem., 1995, 9, 541. for mercury was 253.65±0.5 nm). The detected values for methylmercury were 2.92±0.17 mg g-1 for CRM 463 and Paper 7/01553D 5.17±0.49 mg g-1 for CRM 464 (certified values were Received March 5, 1997 3.04±0.16 and 5.50±0.18 mg g-1 as MeHg). The analytical Accepted May 23, 1997 methods used for certification were as described in the litera- 996 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12

ISSN:0267-9477    

DOI:10.1039/a701553d

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

Quantitative Analysis of Additives in Solid Zinc Alloys by Laser-induced Plasma Spectrometry† LOUIS ST-ONGE, MOHAMAD SABSABI, AND PAOLO CIELO National Research Council Canada, Industrial Materials Institute, 75 deMortagne Blvd., Boucherville, Que�bec, Canada, J4B 6Y 4 In this paper the use of laser-induced plasma spectrometry has been shown to occur in reduced-pressure laser-induced plasmas from brass (Cu–Zn) samples, the vapour pressures of (LIPS) for the quantitative analysis of Al, Cu, Fe, Pb and Sn components in solid zinc alloys is evaluated. Laser-induced Cu and Zn being dierent by several orders of magnitude.6 As a result, when using internal standardization, a longer time is plasmas are characterized using spectroscopic diagnostic techniques that yield the excitation temperature and electron needed after the laser pulse to attain a constant line intensity ratio (e.g., IZn5ICu).6 However, such a phenomenon has never density.Optimal experimental conditions for analysis are evaluated, including time gating parameters and distance from been observed in the denser plasmas obtained at atmospheric pressure, such as those studied here.focusing lens to target where it is found that the focus of the laser beam should be positioned behind the target in order to Analysis of aluminium impurities in zinc alloys using LIPS has been reported recently by Kim et al.7 A Nd5YAG laser prevent secondary air plasmas from forming in front of the target.Calibration curves are produced for several analytical (1064 nm) was used to initiate the plasma and the light detection was resolved spatially and temporally. Dierent gas lines, and the analytical performance of the technique is assessed. While low detection limits (<60 ppm) are found, the environments (air and argon at dierent pressures) were investigated. They found that the ratio of the Al I 308.22 nm line to precision of measurement could be improved.the Zn I 307.59 nm line is the best for building calibration Keywords: L aser-induced plasma spectrometry; zinc; atomic curves, as long as a time delay of at least 13 ms is imposed emission spectrometry; plasma characterization prior to recording the plasma emission. Furthermore, they concluded that an argon environment (at 100 Torr) yields the best sensitivity and lowest fluctuations for aluminium analysis. Laser-induced plasma spectrometry (LIPS) is being more and more recognized as a powerful tool for field-based elemental Similarly, we are concerned in this paper with the analysis of solid zinc alloys using LIPS, but put more emphasis on the analysis of solids, liquids, gases and aerosols (see recent review papers1,2).In LIPS, a focused and pulsed laser beam vaporizes characterization of the plasma (produced in air at atmospheric pressure), while seeking to find the most suitable lines and the a small amount of the target object and a transient plasma is formed.The light given o by the plasma is spectrally resolved best conditions for quantitative analysis of the elements Al, Cu, Fe, Pb and Sn. and the atomic line spectrum is subsequently analysed and calibrated in order to determine elemental concentrations in the sample. The analysis of metals in the solid and molten EXPERIMENTAL states has attracted particular attention. In the steel industry the galvanization process, wherein a zinc coating is applied to Instrumentation steel surfaces, requires strict monitoring.During this process, The instrumentation, shown in Fig. 1, has been described in it is necessary to control the amount of certain minor metal previous papers from this laboratory.8,9 In the experiments components of the molten zinc pool into which the steel sheet reported here, a Q-switched Nd5YAG laser operating at is immersed. For instance, the aluminium and iron concen- 1064 nm (Surelite I-10, Continuum, Santa Clara, CA, USA) trations should fall within the ranges 0.1–0.2% m/m and produced pulses of 6 ns duration at a maximum repetition rate 0.05–0.06% m/m, respectively.The LIP formed on the molten of 10 Hz. The quasi-gaussian beam was focused on (or near) zinc surface together with the use of atomic emission specthe solid target by a 515 mm focal length lens. A large focal trometry for probing the plasma would provide a most direct length is chosen in this case to provide a large depth of focus.and rapid means of controlling these amounts on line. The We estimate that the focused spot has a diameter of roughly present study of solid zinc alloys is seen as a first step towards 0.5 mm and that the depth of focus is in the order of 100 mm. the analysis of molten zinc. It is known that the amount of metal ablated from pure solid samples by laser radiation is larger for metals of low melting point.3 Having a low melting point T m and boiling point T b (693 and 1181 K, respectively4), zinc is more easily ablated than aluminium (T m=933 K, T b=2600 K) and iron (T m=1808 K, T b=3273 K).In alloys, although laser ablation itself can be considered stoicheiometric, element-selective vaporization of the solid after the laser pulse is possible through a plasma–material interaction.5 It is also possible that the discrepancies in T m and T b between zinc and other metals could result, during analysis of zinc alloys, in element-selective atomization of particles and droplets present in the plasma, up to the time when atomization can be considered complete.This † Presented at the 1997 European Winter Conference on Plasma Fig. 1 Schematic drawing of the experimental arrangement. Spectrochemistry, Gent, Belgium, January 12–17, 1997. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (997–1004) 997The laser energy was monitored with a pyroelectric joulemeter and so were discarded in the analysis that followed.A temperature was then calculated for each of the next 20 laser plasmas,† (ED-200, Gentec Inc., Sainte-Foy, Quebec, Canada) and was adjusted by using a combination of half-wave plate and and an average temperature was computed from these. The average temperature is considered to be one measurement. Such polarizing cube to attenuate the beam. The spectra emitted were analysed by means of a 0.66 m a procedure was followed at five dierent positions on the sample.For given experimental conditions, five measurements spectrometer (Model 207, McPherson, Acton, MA, USA) with a 2400 grooves per mm grating, thus providing a linear disper- of temperature were thus obtained and the average and standard deviation (s) of these five measurements were used in sion at the outlet port of 0.62 nm mm-1. The latter was fitted with a time-gated optical multichannel analyser (OMA Model plotting the results. IRY-700 S/B, Princeton Instruments, Trenton, NJ, USA) with 700 intensified elements (of dimensions 25 mm×2.5 mm) which Electron density allowed simultaneous light detection over a spectral range approximately 10 nm wide.The spectrometer entrance slit was The electron density is deduced from the Stark broadening of imaged (and centred) at 551 magnification onto the plasma. the Al II 281.62 nm line, as in ref. 8 for aluminium alloys. For an entrance slit 25 mm wide for example, and neglecting However, a diculty arises in the present case from the fact the lens aberrations, this configuration allowed the sampling that Al concentrations in the plasma are much lower (Al is of a 125 mm×12.5 mm section integrated over a slice of plasma only a minor component of the zinc alloy).Consequently, the perpendicular to the target surface. The slit width was in fact Al II line is small and is superimposed on the wing of the very chosen between 20 and 50 mm, depending on the type of intense and wide neighbouring Zn I line at 280.09 nm, as shown experiment, thus enabling a control of the luminosity.in Fig. 2. In determining the line width, the sloping background signal from the wing of the Zn I line thus needs to be subtracted carefully. Also, the signal is accumulated over 20 laser shots Plasma Diagnostics in order to increase the signal5noise ratio. A zinc sample containing 0.99% m/m Al was used for these measurements. Excitation temperature Sincour present experimental conditions are similar to The plasma temperature under the given experimental conthose in ref. 8, we also assume here that the only important ditions was determined by constructing a Boltzmann plot contributions to the line broadening are the instrumental and using the area under eight neutral iron lines found between the Stark broadening.In both cases, a Lorentzian line shape 370 and 377 nm, as in ref. 8. Assuming that local thermois assumed, and thus the full width at half maximum (FWHM) dynamic equilibrium (LTE) conditions prevail (see later), the attributable to the Stark eect is the dierence between the slope of a linear regression through the Boltzmann plot can observed FWHM and the instrumental FWHM, which is be linked directly to the temperature.The spectroscopic conapproximately 0.03 nm. The electron density is then given by stants for these lines are given in ref. 8. Note that their upper the relation level energies are centred around two values (approximately 27 000 and 34 000 cm-1, see ref. 8). Also, no zinc lines interfered (FWHM)Stark=2w ne 1016, (1) with the iron lines in the 370–377 nm range. The measurements were performed using a zinc alloy reference sample containing where ne is the electron density (in cm-3), and w is the electron 0.67% m/m Fe, positioned at the focus of the laser beam. impact parameter. We assume w=0.00212 nm for the Temperatures were obtained for single laser plasmas, i.e., Al II 281.62 nm line, as determined experimentally by Colo�n without accumulation of light over several laser pulses.Indeed, et al. for ne=1016 cm-3 and a temperature of 10 500 K.10 even though the latter scheme increases the signal5noise ratio, we found that, in doing so, information on individual shot variation is lost, and a systematic error in the calculated RESULTS AND DISCUSSION temperature is introduced. Consequently, at a given position Time-resolved Plasma Characterization on the sample, the emissions from 40 consecutive laser plasmas were recorded individually.Among these, the first 20 pulses Excitation temperature served to clean the surface and to reduce surface roughness, Fig. 3 shows the plasma temperature as a function of time after the laser pulse, for laser energies of 75 and 150 mJ. The time delays and corresponding gate widths (in parentheses) were 0.5(1), 0.75(1), 1(1), 2(2), 5(2), 10(5), 20(5), 30(10), 40(20), and 50(20) ms. The error bars depict the variability over five measurements (as defined previously).Systematic errors could also be present, and thus the temperatures are assumed to be known with an uncertainty of ±6%, coming from the uncertainty on the transition probabilities for the lines used in the Boltzmann plots.8 As observed previously for aluminium8 and copper9 targets, the plasma cools rapidly during the first 10 ms, after which the temperature decreases more slowly, remaining above 5000 K for 40 ms.Moreover, absolute temperature values are similar to those obtained with aluminium and copper targets. It is remarkable that the temperature increases only slightly with a two-fold increase in laser energy (a somewhat similar behaviour was observed with an aluminium sample8). It seems that the temperature in the periphery of the expanding plasma, from Fig. 2 Time-resolved spectrum around the Al II 281.62 nm line. The † The linear regression coecient, r, of the Boltzmann plot was -0.95 on average.Boltzmann plots with |r|<0.6 were discarded. width of this line is proportional to the electron density. 998 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12density values are practically independent of the laser energy. This contrasts with the line intensities, which usually show a strong dependence on laser energy. It can be assumed that an increase in laser energy leads to a larger amount of ablated matter and also a larger energy absorption in the plasma.This will probably cause an increase in plasma volume and in the total number of emitting atoms. The line intensity thus increases, without the need for a higher electron density or temperature. Furthermore, a larger volume of plasma is sampled, since in our configuration the light is integrated along the direction perpendicular to the target. Finally, knowing the electron density and the plasma temperature, we can determine whether the LTE assumption is valid.Using McWhirter’s criterion,8,13 we find that the electron densities of Fig. 4 are high enough to establish LTE conditions (through collisions), at least until 3 ms. Other measurements (not shown here) using the Stark broadening of a neutral iron line at 538.3 nm seem to indicate that the electron density Fig. 3 Time-resolved plasma temperature for two dierent laser decreases past 3 ms at the same slow rate as between 1 and energies. 3 ms. LTE can thus be expected to prevail for much of the plasma lifetime (a few tens of ms), and as a result we can rely on Boltzmann’s law to express line intensities and line intensity which the neutral iron emission mostly originates,11,12 is practiratios as a simple function of temperature.cally independent of laser energy. The latter parameter rather influences the ablated mass and the plasma volume (and not plasma density as will be seen next). Choice of Optimal Conditions for Analysis Choice of analytical lines Electron density In LIPS, the choice of spectral region and specific analytical As was shown in Fig. 2, the Al II 281.62 nm line width decreases lines depends on several factors, including detector sensitivity, with time after the laser pulse, indicating a rapidly decreasing spectral interference, transition probabilities and possibility of electron density. The time-resolved electron density, calculated self-absorption in the case of resonant lines.Moreover, in from these line widths (as described above), is shown in Fig. 4 analytical atomic spectrometry, it is usually advantageous to for laser energies of 75 and 150 mJ. The time delays and carry out an internal standardization by comparing the analytcorresponding gate widths (in parentheses) were in this case ical line intensity with that of the major component of the 0.05(0.05), 0.1(0.1), 0.2(0.1), 0.3(0.1), 0.4(0.1), 0.5(0.1), 0.6(0.2), sample. In the case of LIPS, such an intensity ratio compensates 0.8(0.2), 1(0.5), 2(0.5), and 3(1) ms.The measurements could for shot-to-shot variations in the amount of ablated matter not be performed at larger delays because of a too faint ionic and in the excitation characteristics. When the upper level of aluminium line emission. Also, no error bars are given in this the transition has a similar energy for both lines, the line case since only one measurement is performed per time delay intensity ratio is all the more insensitive to temperature (one measurement being derived from the accumulated specvariations, since the latter then aect atomic excitation to the trum of 20 laser shots).The electron density can in fact be same degree for both atomic species. We have used internal assumed to lie within ±15% of the values given on the plot, standardization throughout our study. An added requirement due to uncertainties on the width measurement, the instrumento the choice of lines is thus that both analytical line and tal broadening deconvolution and the electron impact reference line can be found in the same spectral window (10 nm parameter.wide in the present case). The electron density decreases with time even more rapidly Based on the above considerations, we have chosen two than does the temperature (c.f., Fig. 3), especially going from distinct spectral windows (centred at 284 and 305 nm) that 0 to 1 ms. However, similarly to the temperature, the electron include emissions from Al, Cu, Fe, Pb and Sn, as well as reference lines from zinc.It can be noted that by using a less resolving grating (600 grooves mm-1 instead of 2400 grooves mm-1), a single spectral window could be made to encompass all these lines, but of course to the detriment of the spectral resolution. The dierent lines we have included in our investigation are listed in Table 1. In this table, the velength of the line is denoted by l, the energy levels and statistical weights are denoted by E and g, with subscripts 1 and 2 for the lower and upper level of the transition, respectively, while A21 is the transition probability.With regard to the benefit discussed above of choosing lines with similar upper level energies (E2), it is clear from Table 1 that, for the 2nd spectral window, the Zn line at 307.59 nm is a better choice of internal standard. Fig. 5(a) and (b) show typical spectra corresponding to the spectral windows selected.These spectra were obtained with a gate delay and width both of 10 ms, and a laser energy of 150 mJ. Lines used for quantitative analysis by LIPS, which appear in Table 1, are given with their wavelength while other Fig. 4 Time-resolved electron density for two dierent laser energies. lines are simply identified as an element. A semi-logarithmic Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 999Table 1 Analytical lines used for LIPS analysis of zinc alloys.The To our knowledge, there are no zinc lines at these positions. spectroscopic data were taken from refs. 14 and 15, as well as from We can only suppose that there were either aluminium impuritrefs. 16 and 17 for zinc lines ies in their ‘blank’ zinc sample, or iron impurities leading to possibly interfering lines at 308.37 and 309.16 nm.We observed Element l/nm E1/eV E2/eV A21/108 s-1 g1 g2 with a zinc sample containing no aluminium but 0.67% m/m 1st spectral window: iron that the iron lines indeed appeared (observation took Zn 280.09 4.08 8.50 — 5 7 place after a time delay of 10 ms and with a gate width of Cu 282.44 1.39 5.78 0.078 6 6 10 ms), but these lines did not produce a net signal at the Al Pb 283.31 0.00 4.38 0.58 1 3 Sn 284.00 0.42 4.79 1.7 5 5 line positions; interference is avoided because the Fe and Al Sn 286.33 0.00 4.33 0.54 1 3 lines are separated by more than 0.1 nm. Pb 287.33 1.32 5.63 0.37 5 5 Note finally that for both spectral windows, a background value chosen in a line-free region was subtracted from line 2nd spectral window: intensity maxima, and these corrected (net) intensity values Fe 302.05 0.09 4.19 — 5 5 were then used in the analysis. 302.06 0.00 4.11 — 9 9 Sn 303.41 0.21 4.30 2.0 3 1 Zn 307.21 4.08 8.11 — 5 3 Gating parameters Zn 307.59 0.00 4.03 3.29×10-4 1 3 Al 308.22 0.00 4.02 0.63 2 4 With laser-induced plasmas, a time delay is generally required Al 309.27 0.01 4.02 0.74 4 6 to gate o the early part of the signal in order to avoid the 309.28 0.01 4.02 0.12 4 4 intense initial continuum emission and improve the line resolution (poor at first because of Stark-broadening by the dense plasma).This allows a larger dynamic range for the detection of neutral line emission, which is usually the most useful in quantitative analysis by LIPS.The choice of the gate delay and width is crucial in the optimization of operating conditions for such analysis. Fig. 6 shows the time-resolved intensity of Al and Fe lines ratioed to a zinc line. For these measurements, the gate width is chosen equal to the gate delay (to ensure a sucient time resolution). Assuming the plasma to be satisfying LTE conditions, and according to the Boltzmann law, these ratios should be independent from the variation of temperature (shown in Fig. 3). Taking into account the uncertainty of the measurements, Fig. 6 shows that in fact the intensity ratios are approximately constant only from about 5 ms onwards.Not only can it be stated with more certainty that LTE prevails after this delay, but it may also be supposed that atomization of matter (possibly element-selective) is then practically complete. Also, in the first microseconds, the gas density is very high and self-absorption of resonant lines is likely. Therefore, because of a very low transition probability for the Zn line compared with Al (and possibly Fe), the Al5Zn and Fe5Zn ratios may increase with time as the gas becomes optically Fig. 5 Portions of spectra selected for the analysis of zinc alloys: (a) semi-logarithmic plot of the 1st spectral window, (b) linear plot of the 2nd spectral window. The sample used contained 0.99, 0.26, 0.04, 0.63 and 0.06% m/m of Al, Cu, Fe, Pb and Sn, respectively. The detection gate delay and width were both of 10 ms for these spectra, and the laser energy was 150 mJ. scaling was chosen for Fig. 5(a) in order to clearly show all the analytical lines, which are much less intense than the zinc line.In fact, since the zinc line at 280.09 nm sometimes saturated the detector, we chose to read the intensity on its left wing (at 280.05 nm) instead of reading the intensity at the maximum. This intensity value was then used as internal standard for the 1st spectral window. Note that the MgII lines are both smaller by nearly a factor of 2 than the Zn line, while the Mn line is smaller by an order of magnitude, so that these Fig. 6 Time-resolved line intensity ratios of Al and Fe over Zn, for a lines in fact do not interfere significantly with the Zn line. Kim zinc alloy containing 0.99% m/m of Al, and 0.04% m/m of Fe. For et al.7 discuss a possible interference from small zinc lines with each condition, the average and s of five measurements is plotted, a the aluminium lines at 308.22 and 309.27 nm; these interfering measurement being itself the average (not accumulation) of intensity lines show up in the spectrum from a ‘pure’ zinc sample.They ratios for 20 consecutive laser shots at one position on the target add that these lines disappear after about 13 ms, which gives a (following 20 preparation shots). The laser energy was 150 mJ, and the gate width was equal to the gate delay. lower limit for the time delay required prior to observation. 1000 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12thin. In any case, it is clear that quantitative analysis can be principal plasma.The line intensities were maximum when conducted only after a certain delay. The conditions chosen dl–t=47 cm (no secondary plasma). Also, it was found from for analysis were a 10 ms gate delay and a 10 ms gate width. simple line intensity ratios and from actual excitation tempera- Because of the limited lifetime of the plasma, a larger gate ture measurements that the plasma temperature is highest for width is not justified as the signal becomes too low to this dl–t value, although the dependence on dl–t was smaller.substantially improve the signal5noise ratio. For quantitative analysis, large line intensities are desirable since large signal5noise ratios will yield lower detection limits. We thus have chosen a dl–t of 47 cm (focus position 5 cm L ens-to-target distance and laser energy behind the target) as the best one for analytical purposes, at As will be seen here, positioning the focus position of the laser the same time ensuring a better shot-to-shot reproducibility of beam exactly at the target surface is not necessarily the best laser ablation because sporadic air breakdown is avoided.The situation for analysis. This parameter needs to be explored. laser energy chosen was 150 mJ. Combined with a larger spot When the focus position of the laser beam is moved from diameter (due to our choice of dl–t=47 cm instead of 52 cm), 5 cm in front of the target to 5 cm behind the target (by this yields a laser irradiance at the target which is intermediate moving the focusing lens, not the target), there is only a small between that oered by a 75 mJ and a 150 mJ laser pulse with change in the spot diameter; the long focal length of the lens dl–t=52 cm (the latter conditions corresponding to those (52 cm) leads to a large depth of focus.For example, the explored in our plasma characterization experiments). diameter of the crater found after forty consecutive 150 mJ laser shots at one position, estimated using an optical microscope, varies symmetrically from slightly more than 600 mm when the focus is at the target surface (lens-to-target distance dl–t equal Number of preparation shots to 52 cm), to slightly less than 800 mm when the focus is either 5 cm in front of the target (dl–t=57 cm) or 5 cm behind the We have already alluded to the need for a certain number of target (dl–t=47 cm).There is therefore only a moderate vari- preparation laser shots prior to conducting a given measureation (about 50%) of laser irradiance at the target surface ment.This precautionary step is often carried out in LIPS (power per unit area) when varying dl–t over 10 cm. studies (see ref. 8, and references therein). It is intended to Contrary to what would be expected from this symmetrical ensure that the measurement performed at the surface of the variation of laser irradiance, we observed an asymmetrical target material is representative of its bulk composition. dependence of the plasma parameters on dl–t, similar to that Preparation shots will namely eliminate possible layers of described recently by Multari et al.in their study of LIPS chemically modified material (e.g., oxide), impurities at the sampling geometry.12 First of all, we found that the intensity surface, or remove surface roughness that can modify some of the popping sound accompanying the laser plasma did not characteristics of the ablation process (surface reflectivity, vary symmetrically around a dl–t of 52 cm.Rather, it was thermal conductivity, etc.). maximum for dl–t=47 cm and minimum for dl–t=57 cm, vary- For some of the experiments already reported here, as much ing monotonically between the two positions. At the same as 20 preparation shots were carried out. This was motivated time, it was observed that small air plasmas formed sporadi- by the fact that rough portions of the zinc alloy samples were cally in front of the target and that the extent of these air used, portions that contained craters from previous laser shots.breakdown occurrences (intensity, number, and distance over In contrast, the experiments to produce calibration curves, which they appeared) also varied monotonically with dl–t. In reported in the next section, were conducted on fresh (not fact, when this distance was 47 cm, corresponding to the most previously ablated) portions of the certified materials.It was intense popping sound, there was practically no secondary at first intended to carry out five preparation shots at a given plasma formation in air, indicating that such plasma formation position, followed by 20 additional shots which were to be otherwise prevented part of the laser energy from reaching the used for the analysis (25 shots in all ). However, as Fig. 7 target. This also showed that the probability of breakdown reveals, five preparation shots were sometimes not quite over a given distance in front of the target (e.g., 10 cm) increased sucient to obtain a measurement representative of the bulk.with the power density of the laser beam in that volume. This Rather, it was determined to separate the 25 shots at a given density is lowest when the laser is focused 5 cm behind the position in 9 preparation shots and 16 shots used for the target (dl–t=47 cm) and largest when it is focused 5 cm in front analysis.of the target (dl–t=57 cm). It thus appears that the main parameter determining the amount of power available for ablation is not the laser irradiance at the target surface but the power density in front of the target, in this case through an inverse proportionality. The laser power density in the vicinity of the target has to be low in order to prevent secondary plasma formation. We tried to determine if there were other parameters aecting the formation of secondary plasmas.We found no evidence that the type of material (zinc alloy, pure zinc, or aluminium alloy), the position or angle of incidence on the target, or the laser repetition rate (varied from 1 to 10 Hz) had any eect. On the other hand, but predictably, there was an eect of the laser energy. The latter globally aected the intensity and spatial extent of secondary plasma formation. However, it did not change the qualitative dependence on lens-to-target distance.It can thus be concluded that air breakdown randomly occurred on dust particles present in the atmosphere, without Fig. 7 Line intensity ratio of Al5Zn as a function of the laser shot an evident link to matter ejected from the target. number. These are 25 successive shots at one position on a zinc alloy The lens-to-target distance, apart from aecting the popping sample containing 0.10% m/m of aluminium. The detection gate delay and width were both of 10 ms, and the laser energy was 150 mJ.sound intensity, also aected the line intensities from the Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 1001Fig. 8 Calibration curves for dierent additives in zinc alloys, using dierent analytical lines: (a) iron, (b) aluminium, (c) copper, (d) lead, and (e) tin. There are error bars for all points although they may be too short to be distinguishable. Quantitative Analysis operating conditions suitable for analysis were detailed in the preceding section. The sampling procedure is as follows.At a Calibration curves given position on the target, 25 laser plasmas are produced. The first nine are preparation shots and the remaining 16 are The composition of the standard reference zinc alloys used for building calibration curves is given in Table 2, while the used in the analysis. For each of these plasmas, net line intensities at given wavelengths are obtained by subtracting a background value (as explained above) from the peak intensity Table 2 Aluminium, copper, iron, lead and tin concentrations of the line.In each spectral window, ratios are calculated (% m/m) in the standard reference zinc alloy samples used in the between the analytical lines and the corresponding zinc refer- present work ence line. These ratios are then averaged (by groups of 16) Sample Al Cu Fe Pb Sn and as such represent a measurement. For each sample, measurements are performed at five positions spaced by 3 mm 41XZ5 0.10 0.019 0.026 0.026 0.021 along a line.The average and s of these five measurements are B.S.SP 3B 0.11 (<0.0005) 0.038 0.10 0.002 BS 743 0.23 0.15 0.24 0.50 0.051 then plotted for samples of dierent concentrations (see 41X 0336 Zn 3 0.60 0.33 0.16 0.02 0.12 Table 2) to form the calibration plots. 41X 0336 Zn 2 0.99 0.26 0.04 0.63 0.06 Fig. 8(a) shows two calibration curves for the determination CZn 4/2 — 0.39 0.67 1.97 0.96 of iron concentration in zinc alloys, using the Fe 302.06 nm 1002 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12Table 3 Detection limits for aluminium, copper, iron, lead and tin in analytical line and two dierent zinc reference lines. As already zinc alloy samples, obtained using dierent analytical lines stated, the Zn 307.59 nm line is a better choice of reference line since its upper level energy is very close to that of the analytical Element Wavelength/nm Detection limit/ppm lines in that spectral window (contrary to the Zn 307.21 nm Al 308.22 16 line), thus reducing the possibility of temperature fluctuations Al 309.27 9 aecting the ratios.For this reason, ratios using the Cu 282.44 544 Zn 307.21 nm line are not shown in the remainder of the Fe 302.06 22 calibration plots. Still, it can be seen in Fig. 8(a) that it gives Pb 283.31 54 Pb 287.33 405 in this case results equivalent to those obtained with the Sn 284.00 33 Zn 307.59 nm line.Sn 286.33 54 The shape of both iron calibration curves indicates that the Sn 303.41 34 iron emission is self-absorbed in the plasma when iron concentration is suciently large; this is the case at least for the concentration of 0.67% m/m. Indeed, the Fe 302.06 line is resonant, i.e., the lower level of the transition is the ground state 324.75 and 327.40 nm, would give detection limits in the order of the atom (see Table 1). For concentrations of 0.16% m/m of a few ppm. and lower, the curve is quasi-linear and is suitable for determining iron concentrations in the range typical of the galvanization process (0.05–0.06% m/m).Precision of measurement A similar phenomenon of self-absorption is seen in Fig. 8(b) The half length of the error bar associated to a particular line which shows calibration curves for aluminium, built up using intensity ratio in Fig. 8(a)–(e) represents the s of five measure- two dierent aluminium lines (with the Zn 307.59 nm reference ments performed on a given sample (see above our definition line); both aluminium lines are resonant.The highest sensitivity of measurement). Although reasonable values of relative s (i.e., slope of the calibration curve) is provided by the (RSD) were found in some cases (RSD#3–5%), they were on Al 307.59 nm line. The curves are also in this case applicable average 13% for this set of experiments, which is quite high. to the control of the galvanization process (Al concentration This by no means represents the lowest precision achievable in the range 0.1–0.2% m/m).with the LIPS technique. The day-to-day reproducibility of The calibration curve for copper, shown in Fig. 8(c), provides our measurements was not tested, and therefore systematic an example of a linear calibration curve; the Cu 282.44 nm errors may have escaped our attention. The present study was line is non-resonant. The unusually large error bar for the focused on identifying suitable analytical lines and operating 0.33% m/m concentration is seemingly due to a large copper conditions rather than on seeking to obtain the best possible inhomogeneity in that particular sample, at least between precision.We examine below how the RSD could be improved. the five positions sampled during that experiment. The shot-to-shot reproducibility of the laser ablation process Fig. 8(d) shows the two behaviours related to self-absorption, and of line intensities depends on surface conditions and on for the case of lead.The calibration curve obtained for the the degree of sample homogeneity, among other parameters.3,19 Pb 283.31 nm line, which is resonant, shows a saturation Performing an internal standardization and allowing for prep- behaviour at large lead concentrations because of selfaration shots eliminates for the most part the influence of absorption, while that obtained from the Pb 287.33 nm nonsurface conditions. A lack of sample homogeneity however can resonant line, is quasi-linear.For trace analysis, resonant lines still manifest itself, especially since the laser ablates only a such as the Pb 283.31 nm line should be chosen since they are small amount of material (digging through only fractions of a in general more intense. However, if one needs to determine mm per shot). Although the large scatter observed here cannot concentrations over a wide range, a more linear calibration be entirely associated to sample inhomogeneities, a somewhat curve, provided by a non-resonant ( less intense) line such as larger number of measurements on a given sample would help the Pb 287.33 nm line, is indicated; the slope (sensitivity) of improve precision.In the present case, the five measurements such a curve remains large at high concentrations. (125 laser shots in all ) take only 12.5 s to complete when using Finally, Fig. 8(e) shows dierent calibration curves for tin, a laser repetition rate of 10 Hz.It follows that increasing the coming from the two spectral windows. Note that the calinumber of measurements would lead to a better representa- bration curve for the Sn 303.41 nm line does not go through tion of the bulk composition and a smaller RSD, without the origin. As can be seen in Fig. 5( b), this tin line is superimcompromising the usefulness of the LIPS technique for rapid posed on the left wing of a zinc line, which explains the nonmonitoring of industrial processes.zero intercept. This needs to be carefully taken into account when determining trace amounts (as well as the detection limit) The authors acknowledge the skilled technical support of using this line. Rene� He�on. Detection limits REFERENCES The calculation of detection limits is based on the 3s-IUPAC 1 Majidi, V., and Joseph, M. R., Crit. Rev. Anal. Chem., 1992, 23, 143. definition, s being the s of the noise (coming from all sources) 2 Radziemski, L.J., Microchem. J., 1994, 50, 218. at the position of the analytical line. Any background- 3 Geertsen, C., Briand, A., Chartier, F., Lacour, J.-L., Mauchien, subtracted signal which is at least three times s can be P., Sjo� stro�m, S., and Mermet, J.-M., J. Anal. At. Spectrom., 1994, attributed to the presence of a trace with a 99.86% confidence 9, 17. level. In order to determine s, we follow guidelines given by 4 T he Merck Index, 12th Ed., Merck & Co., Inc., Whitehouse Station, NJ, 1996. To�ro�k et al.18 The detection limits, listed in Table 3, are 5 Russo, R. E., Appl. Spectrosc., 1995, 49, 14A. calculated by converting the 3s intensity value to concentration 6 Uebbing, J., Brust, J., Sdorra, W., Leis, F., and Niemax, K., Appl. values, using the slope at low concentration of the net line Spectrosc., 1991, 45, 1419. intensity calibration curves. Limits lower than about 60 ppm 7 Kim, D. E., Yoo, K. J., Park, H. P., Oh, K. J., and Kim, D. W., are achieved for all elements except copper, for which a Appl. Spectrosc., 1997, 51, 22. 544 ppm 3s-limit is found for the line used.We should mention 8 Sabsabi, M., and Cielo, P., Appl. Spectrosc., 1995, 49, 499. 9 Sabsabi, M., and Cielo, P., J. Anal. At. Spectrom., 1995, 10, 643. that lines of copper in other spectral regions, namely those at Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 100310 Colo� n, C., Hatem, G., Verdugo, E., Ruiz, P., and Campos, J., 16 Hetzler, C. W., Boreman, R. W., and Burns, K., Phys. Rev., 1935, J. Appl. Phys., 1993, 73, 4752. 48, 656. 11 Sabsabi, M., Cielo, P., Boily, S., and Chaker, M., SPIE, Proc. 17 Johansson, I., and Contreras, R., Arkiv fo�r Fysik, 1967, 37, 513. Opt. Meth. Chem. Process Control, 1993, 2069, 191. 18 To� ro� k, T., Mika, J., and Gegus, E., Emission Spectrochemical 12 Multari, R. A., Foster, L. E., Cremers, D. A., and Ferris, M. J., Analysis, Adam Hilger, Bristol, 1978. Appl. Spectrosc., 1996, 50, 1483. 19 Davies, C. M., Telle, H. H., Montgomery, D. J., and Corbett, 13 McWhirter, R. W. P., in Plasma Diagnostic T echniques, ed. R. E., Spectrochim. Acta, Part B, 1995, 50, 1059. Huddlestone, R. H., and Leonard, S. L., Academic Press, New York, 1965, ch. 5. Paper 7/03102E 14 Reader, J., Corliss, C. H., Wiese, W. L., and Martin, G. A., Received May 6, 1997 Wavelengths and T ransition Probabilities for Atoms and Atomic Accepted June 17, 1997 Ions, NSRDS-NBS 68, US Government Printing Oce, Washington, 1980. 15 Striganov, A. R., and Sventitskii, N. S., T ables of Spectral L ines of Neutral and Ionized Atoms, IFI/Plenum, New York, 1968. 1004 Journal of Analytical Atomic Spectrometry, Sep

ISSN:0267-9477    

DOI:10.1039/a703102e

出版商:RSC    

年代:1997

数据来源: RSC