摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 92 1 Anion Exchange for the Determination of Arsenic and Selenium by Inductively Coupled Plasma Mass Spectrometry* Jan Goossens Luc Moens and Richard Dams Laboratory of Analytical Chemistry Institute for Nuclear Sciences Ghent University Proeftuinstraat 86 B-9000 Gent Belgium A previously described anion-exchange separation method for the elimination of some of the spectral interferences in inductively coupled plasma mass spectrometry was optimized for the determination of As and Se. Basically samples are applied to a Dowex-1 X8 resin column in nitrate form and CI- is retained while As and Se are eluted with dilute nitric acid and collected in the eluate. The ArCI+ interference on 7 s A ~ + and 77Se+ was eliminated by more than 99.9% and As and Se were determined in sea-water and human urine.A Dowex-1x8 resin column in acetate form however was observed to be preferable for the determination of As and Se in human serum. Keywords Anion exchange; spectral interference; arsenic; selenium From the viewpoint of inductively coupled plasma mass spectrometry (ICP-MS) the elements As and Se have much in common. Besides their similar mass numbers and almost identical first ionization energies (IE) (As 9.8 1 eV; Se 9.75 eV) the determination of both elements is seriously degraded by the occurrence of spectral interferences. These are caused by the formation of multiple species (XU +) from sample constituents (e.g. SO3+) the argon plasma gas (Ar2+) or the combination of both (e.g.ArCl+) during the ionization and/or sampling processes. Owing to the limited resolution of quadrupole mass analysers these polyatoms cannot be separated from As and Se. Considerable work has been and is still being performed on the alleviation of this analytical problem. The introduc- tion of molecular gases into the or the addition of organic solvents to the samples1-2*6 can reduce the XY+ levels to a substantial extent. Alternatives are the introduc- tion of As and Se as hydrides into the ICP,7-11 chromatogra- phic de-salting,12 precipitation of Cl- as AgC1l3 or the coupling of an aerosol desolvation system to the ICP-MS instrument.14 In this paper an anion-exchange separation method" for the elimination of C1 and S interferences was optimized for the separation of As and Se from C1- to allow the determination at mlz=75 (As) and mlz=77 (Se) without interference from ArCl.As discussed in ref. 15 this interference eliminating technique is based on the retention of C1 and S as anions on a strong base anion-exchange resin (Dowex-1x8) in NO3- form while the analytes are eluted with dilute nitric acid and collected in the eluate. However unlike other analytes (e.g. V Cr Cu Zn15 and Mo16) for which the procedure has been used As and Se are present in the sample as anions which makes the separation from C1- more delicate and the re-evaluation of some parameters necessary. In addition the direct application of the proce- dure to complex biological matrices such as human urine and human serum produced some additional problems which are discussed. Experimental Instrumentation The ICP mass spectrometer used is a VG PlasmaQuad (VG Elemental Winsford Cheshire UK) equipped with a Fassel *Presented at the 1993 European Winter Conference on Plasma Spectrochemistry Granada Spain January 10-1 5 1993.Table 1 Operating conditions for the VG Plasma Quad Plasma Radio frequency power ForwardIW ReflectedIW Gas flow rate Plasmdl min-l Nebulized1 min-' Auxiliary/l min-' 1350 < 5 13.5 0.725 0.9 Ion Sampling Nickel sampling cone orifice/mm 1 .o Nickel skimmer cone orifice/mm 0.75 Vacuum Expansion stagelmbar* Intermediate stagelmbar Analyser stage/mbar *I mbar= 100 Pa. 2.3 2.0 x 10-4 4.6 x torch a Gilson Minipuls-2 peristaltic pump (maintaining the sample uptake rate at 1.0 ml min-l) a Meinhard-type Tr-30-A3 concentric glass nebulizer and a double pass Scott-type spray chamber with liquid coolant the tempera- ture of which is controlled at 10 "C with a recirculating refrigeration-heating system.The operating conditions are summarized in Table 1. Reagents and Solutions An Se standard solution (1 g 1-I) was prepared by dissolving Se metal (purity >99.99%) in a limited amount of concen- trated nitric acid. Arsenic (prepared from AsZ03) and Ga standard solutions ( 1 g 1-l) were purchased from Fluka (Buchs Switzerland) and Johnson-Matthey (Royston England) respectively. Further dilution of all standard solutions to the 1 mg 1-l level was accomplished with 0.14 mol 1-1 HN03. For qualitative and semiquantitative experiments use was made of commercial 1 g 1-l of Co (Fluka) and SeIV [Alfa Products (Karlsruhe Germany)] solutions.An SeV1 solution was prepared by dissolving H,SeO in water while an AsV solution was purchased from Merck (Darmstadt Germany). The anion exchanger Dowex- 1 X8 ( 100-200 mesh C1- form) was purchased from Serva (Heidelberg Germany) and the resin was used without preliminary purification. An Sn" solution was obtained by dissolving approxi- mately 3 g of SnClz-2HzO [Union Chemique Belge (UCB,922 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 Leuven Belgium)] analytical-reagent grade] in 3 ml of boiling concentrated hydrochloric acid (10 mol 1-I). After complete dissolution approximately 25 ml of water were added. This solution was boiled for approximately 3 min cooled down and placed in a refrigerator (4 "C) for storage. Immediately before use the solution was diluted 1 0-fold with water.Analytical-reagent grade ammonium acetate was pur- chased from UCB and a 2 rnol l-l solution was prepared by dissolving about 150 g of the salt in 1 1 of water. A 0.17 mol 1-1 acetic acid solution was prepared by 100- fold dilution of concentrated (>99.9%) nitric acid [J. T. Baker Instra-Analyzed (Philipsburg NJ USA)]. High-purity nitric acid ( 14 moll-I) and hydrochloric acid (10 mol 1-I) were obtained by sub-boiling distillation of reagent grade acids from quartz apparatus. Millipore (Milford MA USA) milli-Q water was used throughout. Column Preparation The preparation and regeneration of the resin columns are described in detail in ref. 15.Approximately 50 ml (wet volume) of resin (Cl- form) is transferred to a polyethylene tube (27 mm id.) and is converted to the NO3- form with 300 ml of a 1.4 moll-' HN03 solution and prepared for use by rinsing with 100 ml of 0.014 mol 1-I HN03. Regenera- tion is carried out after each elution. Smaller columns in the CH3C02- form for the treatment of human serum only were prepared by transferring 9 ml (wet volume) of resin to a polyethylene tube (15 mm i.d.) and converting the resin to the CH3C02- form with 150 ml of 2 mol 1-l CH3C02N,. Before use the columns were successively rinsed with 25 ml of 1.7 moll-' CH3C02H and 25 ml of water. Batch Experiments An amount of resin was converted to the NO3- form and rinsed with H20 until the pH of the eluate was >4. The resin was then dried at 105 "C for 48 h.Four series of batch experiments were carried out. Each series consisted of 20 polyethylene bottles containing 1 g of resin 2.5 mi of an aqueous standard solution ( 1 per series) and 15 ml of dilute nitric acid the concentration of which ranged from 0.14 mmol 1-1 to 1.4 mol 1 - I within each series. The following standard solutions were used 1.0 mg 1-1 A P 1.0 mg 1 - I AsV 10.0 mg 1-l SelV and 1 g 1-I C1- (KCl). These solutions were all prepared in pure water and therefore the final concentration of nitric acid in the mobile phase (total volume of 17.5 ml) could be calculated. All bottles were shaken mechanically for 30 min. After precipitation of the resin the supernatant was analysed by ICP-MS and the recovery of the analyte under consideration was calculated.Preparation of Synthetic and Sea-water Samples A synthetic sample was prepared containing 5 g 1-I of NaCl 100 pg 1-l of Cu2+ 100 pg 1-1 of AsV and 500 pg 1-l of SeIv in a 0.03 mol 1-l HNO matrix. Dilute SnCl solution (1 ml) was added to the sample (see under Results and Discussion) and the mixture was stirred for 1 min. The samples were then eluted from the anion-exchange resin column (NO3- form) with 0.03 mol 1-I HN03 at a rate of 3 ml min-I and the eluate was collected in a 100 ml calibrated flask. Eluent was added until the eluate was up to volume. The recoveries of As and Se were determined by A sea-water reference material Community Bureau of Reference (BCR) Certified Reference Material (CRM) 403 Sea-water was separately spiked with As*I1 AsV SeIV and Sevr at the 500 pg I-' level to evaluate the reduction and ICP-MS.ellution of these species in an authentic matrix. Aliquots of 10 ml were acidified to 0.03 rnol 1-1 HN03 reduced eluted and analysed as described for the synthetic samples. Unspiked samples were treated in an analogous way and were regarded as 'blank' solutions. An additional treatment was also applied to some of the samples spiked with SeV*. Concentrated hydrochloric acid (200 pl) was added to 5 ml of the sample. The mixture was heated at 95 "C for 5 min cooled and acidified to 0.03 niol 1-l HN03 after the addition of 20 ml of water. R.eduction elution and analysis were carried out as described. Preparation of Urine Samples A. freeze-dried human urine reference material [National Institute of Standards and Technology (NIST) Standard R.eference Material (SRM) 2670 Low Level Toxic Metals in Hhman Urine] was reconstituted with water.To 10 ml of urine 30 ml of 0.04 mol 1-1 HN03 (final concentration of nitric acid in the sample before elution is 0.03 moll-l) and 1 ml of diluted SnC1 were added while gently shaking the sample solution. Then the sample was applied to the top of the resin column (Dowex-1x8 in the NO3- form) and elution was carried out with 0.03 moll-' HN03 at a rate of 3 ml min-'. The eluate was collected in a 100 ml polyethylene bottle and Ga (100 pg 1-I) was added as an internal standard. Six replicates were thus prepared two of which were spiked with a known amount of As and Se (1 00 pg 1-l) for calibration by single standard addition.Preparation of Serum Samples Approximately 0.5 g of freeze-dried human serum reference rr~ateriall~ (equivalent to 5.5 ml of liquid serum) was reconstituted with water in a Teflon beaker. The freeze- dried human serum was the 'second generation' biological reference material prepared by 'Versieck et a1.I'. This material was collected and stored under rigorously con- trolled conditions in order to avoid contamination so that the concentrations of most trace elements were fairly similar to those expected in normal human serum. A further 500 p1 of 1.7 mol 1-l CH3C02H and 1 ml of diluted SnCl solution were added while gently shaking the sample solution. The sample was then applied to the resin column (IDowex-1 X8 in CH3C02- form) and eluted with 0.17 mol I-' CH3C02H at a rate of 3 ml min-l.The eluate was collected in a 25 ml glass calibrated flask. Eluent was added in small portions until the eluate was up to volume. Four replicates were thus prepared. By pipetting two 10 ml portions into 25 ml glass calibrated flasks each sample (e:luate) was divided into two sub-samples one of which was spiked with known amounts of As and Se (50 pg i - I ) for calibration by single standard addition. Gallium ( 100 pg was added as an internal standard to all solutions which then were diluted to volume with water. Analysis Procedure applied to Sea-water Serum and Urine Siamples Before each experiment the electrostatic lenses and the gas Table 2 VG PlasmaQuad acquisition parameters for the VG Pl.asmaQuad Mass range/u 69-79 No.of channels 512 Dwell time/,us 160 No. of sweeps 500 Total acquisition time/s 40.9JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 923 flow rates were adjusted for maximum 71Ga+ signal intensity. The mass scanning data acquisition mode was used. The acquisition parameters applied are listed in Table 2. Standard additions (single addition) was applied as a calibration method and memory effects were avoided by analysing the solutions in order of increasing concentration i. e. blanks unspiked samples and spiked samples. Each solution was measured seven times and the 75A~+ and 77Se+ signals were normalized to the 71Ga+ signal. Results and Discussion Batch Experiments Several batch experiments were carried out in order to determine the capacity factor kgl of all the species of interest.This factor expresses the extent to which an anion is retained by the resin and from its value it is possible to determine the retention time and the retention volume of a solute under consideration.18 If an anion-exchange stationary phase loaded with NO3- anions is considered then in accordance with Haddad and Jackson,17 kgl of an x-valent anion Bx- is given by where w is the mass (g) of the stationary phase V is the volume (ml) of the mobile phase Q is the ion-exchange capacity of the resin (mmol g-l) [NO3-] is the concentra- tion of nitrate in the eluent and is the selectivity coefficient (ie. the equilibrium constant of the ion- exchange reaction under consideration). Usually concentra- tions are expressed in mol 1-I for the mobile phase and mmol g-l for the stationary phase.Eqn. (1) is of great practical use as it allows the prediction of the value of kgl for each eluent concentration ([NO3-],). This requires KB,NO,- to be known and therefore k;; can be determined experimentally in a batch experiment for a single NO3- concentration. It can be seen that increasing the NO3- concentration results in a lower capacity factor and thus in a lower retention of the analytes. This is caused by an increased competition for the active sites of the resin. However whereas experimental and calculated values for kgl agreed well for some species (e.g. Cl-) poor correlation was obtained for weak acids such as H3As0 and H,SeO,. The reason for this is that when using dilute HN03 as an eluent (instead of for example a buffered NaN03 solution) changing the eluent concentration affects not only the NO3- concentration but also the pH.This in turn influences the dissociation of acidic analytes and thus also their affinity for the resin an effect which has not been taken into account in the derivation of eqn. (1). If the capacity factor kBo is defined therefore as the ratio of the anion concentration in the stationary phase to the sum of the concentration of dissociated and undissociated species in the mobile phase it can be shown that for a monovalent anion where k is the acid dissociation constant of the analyte HB. It has been assumed that HN03 is completely dissoci- 2.0 1.5 1 .o "22 CT 0.5 A 0 -0.5 -1 .o -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0 Log ( [HNOJ; + [HNOJm kn ) km Fig.1 Capacity factors (log kBo) for Dowex-1x8 (NO3- form) as a function of the HN03 eluent concentration [HN03] and the acid dissociation constant (kHB) of the analyte under consideration kHB=10-2*3 for As (H3As04); kHB=10-2*6 for Se (H2Se03); and kHB= lo7 for C1 (HC1) ated in the mobile phase and that [NO,-],=[H+],. It can be seen that if kHB>>[HN03] (e.g. Cl-) which means that the analyte is fully ionized the pH effect disappears and eqn. 2 can be simplified to eqn. (1) (x= I). However if kHB is very small (e.g. H3As03 pkH,=9.2) the ionization of the analyte is completely suppressed and it is not retained by the resin kBo approaches 0. For H3As04 (pkHB=2.3) and H2Se03 (pkH,=2.6) the acid dissociation constant is of the same order of magnitude as the eluent HN03 concentra- tions used in the separation procedure.In Fig. 1 a plot is made of experimentally determined capacity factors (log correlation (r>0.999) for all three species is obtained. Therefore eqn. (2) allows the retention of a monovalent anion over a broad eluent nitric acid concentration range to be evaluted. In Table 3 some experimentally determined capacity factors are listed. It can be seen that completely dissociated species (e.g. C1- and HSe0,-) are strongly retained (high capacity factors) while undissociated species (e.g. H3As03) show no affinity for the resin. Owing to the limited differ- ence in capacity factors it would be most difficult to separate C1- from Sevl. At the lower eluent concentrations Sevl is retained even more strongly than C1- as HSe04- dissociates to SeO,,- (bivalent anions are generally retained more strongly than monovalent ones). It appears from Table 3 that it would be possible to elute AsV and SeIv while C1- is retained using eluent concentra- tions 30.1 mol 1-l HN03.In practice however it was found that this required very low elution rates and large resin columns due to the fact that in authentic samples the C1 content can exceed the As and Se content by a factor of 1 x lo6. Therefore it was decided to add a reductant to the sample prior to elution in order to facilitate the elution of As and Se as explained below. kBo) versus log {([HNO3],'+ k H B [HNO~],,,)/~HB} and good Optimization of a Number of Experimental Parameters A number of experimental parameters of the separation procedure were optimized for the simultaneous separation of As and Se from C1- using a Dowex- 1 X8 resin column in NO3- form and dilute nitric acid as an eluent.Sample preparation A small amount of bivalent Sn was added as a reductant to the samples before elution. The acidification of the sample924 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 Table 3 Capacity factors (kBo) as a function of the eluent HN03 concentration Capacity factor (kBo) Species 0.12 rnol I-' HN03 0.012 rnol 1-l HN03 0.0012 mol I-' HN03 (0.0 1 (0.01 (0.0 1 ASV t O . O 1 0.14 2.6 SeIV (0.01 0.07 2.3 Sevl 0.49 :- 20 > 20 c1- 0.56 :.20 > 20 (addition of nitric acid) was found to be of major impor- tance for the reduction process. Nitric acid concentrations in excess of 0.14 mol 1-l resulted in incomplete elution of Se probably due to the precipitation of metallic SeO as a reddish precipitate which could be observed in the sample vessel and/or on the resin column.However concentrations of nitric acid below 0.014 mol 1-l also yielded low recoveries both for As and Se probably because of incomplete reduction. In practice HN03 was added to all samples up to a final concentration of 0.03 mol 1-l. Eluent concent rat ion Elution of all analytes in a minimum eluate volume requires capacity factors close to zero and therefore concentrations of nitric acid as high as possible [cJ eqn. (2)]. It was found (see Fig. 2) that for concentrations of nitric acid in excess of approximately 0.03 mol l-l all (inorganic) As and Se species (after reduction) are eluted at the void volume of the column (this is in fact the volume of eluent necessary for the elution of unretained solutes).Therefore under these conditions the dilution of the samples due to the separation procedure is limited to an absolute minimurn. Further increase of the eluent nitric acid concentration did not change the situation presented in Fig. 2. However concen- trations of nitric acid in excess of 0.2 mol 1-l may cause partial retention of Se probably because of the reasons mentioned in the section above. It is clear that the use of eluent concentrations in excess of 0.03 mol 1-l HN03 would unnecessarily decrease the retention of C1- [eqn. (2)] and therefore eluent concentra- tions >0.05 mol 1-l were never applied.t % : 0 K As 1 0 20 40 60 80 100 Effluent volume/ml Fig. 2 Simultaneously recorded elution diagrams of Cu As and Se from a synthetic sample (10 ml) eluted with 0.03 rnol I-' HN03 at a flow rate of 3 ml min-l Elution rate At an eluent concentration of 0.03-0.05 mol 1-l HN03 the elution characteristics of As and Se are fairly independent of the elution rate applied. However this parameter strongly affects the retention of C1- and as a general rule it can be stated that the lower the elution rate applied the better the separations obtained. It was observed that at this eluent concentration slight 'bleed off' occurs for C1- if the elution rate exceeds 5 ml min-l. Therefore a 3 ml min-' rate was applied for all real samples. Analysis of Synthetic Samples Increasing amounts of sample were reduced eluted from the resin columns in NO3- form and collected in a 100 ml calibrated flask as described under Sample Preparation. Eluent (0.03 moll-' HN03) was added until the final eluate volume was exactly 100 ml.The maximum sample volume giving quantitative recovery (299%) of all analytes (As Se and Cu) was determined. It was found that for a resin clolumn of the size described under column preparation the sample volume should be <50 ml. Under these conditions (x< 50 ml of sample collected in a 100 ml calibrated flask) the separation method requires a sample dilution by a factor 100/x22. If for a given application the dilution of the sample is critical and only limited amounts of sample are available the amount of resin used can be decreased.However care should be taken that the adsorbable material does not exceed 10% of the total column capacity (1.2 mequiv ml-1 of wet resin). Analysis of Sea-water Samples Spiked sea-water samples were eluted from the resin columns in NO3- form with and without the preliminary reduction (addition of SnI1) and the results are listed in Table 4. 'As could be expected from the batch experiments and eqn. (2) As111 is the only species that is eluted at the void volume of the column without preliminary reduction. 11; was found that when adding SrP the reduction of AsV to A.slI1 is quantitative and independent of the reaction time. The necessity of adding SnClz for the elution of SelV suggests that without Sntl Setv is oxidized to SeVt on the column or that Sn" reduces SelV to H,Se and that this is the only form of Se completely unretained by the resin.The first explanation is dubious since the conversion of Setv to Table 4 Analyte recoveries and conditions for quantitative elution at the void volume Boiling of the Analyte recovery* Species Addition of SnlI solution (Oh) As"' No No l00.5-+ 1.2 AsV Yes No 100.3-tO.8 SeIV Yes No 99.8 k 1.2 Sevl Yes Yes 96.2 f 3.3 *Uncertainties are expressed as standard errors on the mean ( fi! = 3).JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 925 Table 5 Determination of As and Se in human urine and human serum" Sample Human urine Element (m/z value) As (75) Se (77) Human serum As (75) Se (77) Concentration Certified value/ determined* pg 1-I Pg I-' 83 (11) (60)t 39.2 (4.5) 30k8 2.97 (0.38)$ 1.78 f 0.36 2.23 (0.29)g - 1.86 (0.24)j - 124.6 (8.4)§ - 94.8 (6.4)j - 177.8 ( 12.0)$ 95.5 +- 4.6 *Uncertainties are expressed as standard deviations (n = 4).thdicative value. Standard solutions prepared in 0.14 mol 1-I HN03a or 0 17 mol 1-' CH3C02Hb. $External calibration; standard solutions prepared in 0.14 mol 1-' HN03. §External calibration; standard solutions prepared in 0.17 mol I-' CH3C02H. ICalibration by standard addition. Sev* requires a highly oxidizing environment. In addition the batch experiments revealed that there is a distinct difference in the affinity of SeIV and SeV1 for the resin (Table 3). Therefore it is most probable that H,Se (pkHB=3.7) is eluted. The reduction of SeIV to H,Se prior to the elution was never observed to result in significant analyte loss even when the sample solution was vigorously stirred for a period of 15 min probably because H2Se is fairly soluble in water (37.7 g l-l).Ig Although Sevl is not commonly present in biological samplesZo its reduction and elution was studied.Reduction by the addition of SnCl and stirring for 2-3 min yielded low recoveries (-40%). Therefore the solution was heated for several minutes in order to reduce Sevl to SeIv as described by Verlinden,I9 before the addition of SnCl,. Analysis of Serum and Urine Samples In order to investigate the possibility of direct application of the separation method to complex biological materials the determination of As and Se in human urine and human serum was studied.Both matrices contain considerable amounts of Cl (urine 4.4 g 1-l and serum 3.9 g 1-l as certified in the reference materials described under Preparation of Urine Samples and Preparation of Serum Samples ) resulting in ArCl+ interferences at mlz=75 and 77 which far outweigh the 7sA~+ and 77Se+ signals. All other Se isotopes are also interfered with mainly by Arz+ and SO3+ species6 A preliminary digestion of the samples would facilitate the interpretation of the elution characteristics since it is difficult to predict the behaviour of solutes in real organic matrices. In addition the reduction of As and Se species using SnC1 was studied in inorganic matrices but probably cannot be completely extrapolated to urine or serum. However direct application is favourable from the view- point of possible analyte loss and contamination that might occur during a preliminary acid digestion.In addition complete digestion of e.g. human serum requires a substan- tial amount of perchloric acid resulting in column regenera- tion problems.1s Both human urine and human serum contain organoar- senic and organoselenium compounds some of which are still unknown. Generally speaking it is unlikely that these complexes are retained by the Dowex-1x8 resin mainly because of their size configuration and charge (in a 0.03 mol 1' HN03 or 0.17 mol 1 - I CH3COZH solution). The elution characteristics of some As compounds (dimethylar- sinate and arsenobetaine) on a NO3- loaded column were investigated and no retention was observed. No particular problems were encountered with the elution of 4-fold diluted human urine on a Dowex-1x8 resin column in the NO3- form.The C1 content in the eluates was <0.1% of the original samples. Blank solutions (1 0 ml of water) treated in a similar manner to the urine samples did not differ significantly from untreated blank solutions and the count rate at rnlz=75 and 77 obtained from a blank solution could be attributed to background noise only (typically 20 area counts per second). As can be seen from Table 5 there is good agreement between the results obtained and the certified (Se) or indicative (As) value. For both As and Se relative standard deviations (n = 4) of about 12% were established. In the instance of Se poor precision can be attributed to the low Se content in the 10-fold diluted urine samples (about 4 pg 1-l) and the low relative abundance of the 77Se isotope (7.5%).The back- ground signal accounted for approximately 30°/o of the total signal at m/z= 77 obtained from a sample solution (eluate). However at 10-fold dilution of the urine samples no significant suppression or enhancement of the 71Ga+ signal was observed and accurate background correction was possible. For As however there were significant differences in the results obtained from two different urine bottles possibly due to contamination. The application of human serum to a Dowex-1x8 resin column in the NO3- form caused some specific problems with respect to the stability of the serum matrix. In particular when applying high elution rates (>5 ml min-l) some precipitation could be observed in the eluate.Rinsing the column with 100 ml of water before use and decreasing the elution rate to 3 ml min-l seemed to solve the problem and good results both for Se (95.5 pg l-l s=6.4 pg 1 - I ) and As (1.92 pg 1-I s=0.18 pg 1-I) were obtained. However potential protein precipitation in dilute nitric acid remains an uncertain factor and jeopardizes the accuracy of the results. Therefore the potential of a Dowex- 1 X8 resin column in CH3COZ- form and the use of dilute acetic acid as an eluent for the treatment of human serum was evaluated. For this modification a systematic study as described for the NO3- loaded columns was not carried out but experiments showed that chlorine is removed quantitatively from the serum and that for this matrix good results can be obtained both for As and Se when analysing the eluates by ICP-MS and using standard additions as a calibration method (cf.' Table 5).In addition the eluates remain stable for 1 week or longer when stored at 4 "C i.e. no visible precipitation could be observed. The enhancement of the As+ and Se+ signals in the926 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 presence of carbon containing compounds has already been reported6J1 and for this reason it was necessary to use standard additions as a calibration method. As can be seen from Table 5 the use of external calibration with standard solutions in 0.14 mol 1-l HN03 or even in 0.17 mol 1-1 CH3C02H ( i e . the eluent acetic acid concentration) re- sulted in erroneously high results both for As and Se.The extent of the error made by the use of external calibration depended on a number of instrumental parameters of which the nebulizer gas flow rate seemed to exhibit the most importance influence. The relative standard deviations (n=4) of 13 and 7% for As and Se respectively are acceptable taking into account the low analyte content of the eluates. reported that the determination (ICP- MS) of Hg in BCR CRM 422 Cod Muscle by standard additions and stable isotope dilution required an equilibra- tion period between the addition of the spikes and the actual measurements to obtain accurate results. It was postulated that owing to the high first IE of Hg (1 0.44 eV) the ionization efficiency of organically bound Hg would differ significantly from that of inorganic mercury and that an equilibration period provides accurate results since the calibrant (the spike addition) would have time to become bound to an organic ligand and so the forms of mercury would be similar. In human serum both Se (IE=9.75 eV) and As (IE=9.81 eV) are partially if not completely associated with organic complexes.If a similar discrepancy in the ionization efficiency would occur as described for Hg the extent of this effect would probably also depend on the equilibration period and possibly even on the amount of spike added. However eluates that were analysed 2 h after spiking and that were re-analysed 4 d later did not yield significantly different results nor did eluates that were spiked with amounts of As and Se equal to or 100-fold higher than the As and Se contents originally present.Campbell et Conclusions An anion-exchange separation method was optimized for the elimination of interferences by ArCl in the determina- tion of As and Se by ICP-MS. Basically samples are applied to a Dowex-1x8 resin column in NO3- form and C1- is retained while As and Se species are eluted with 0.03 mol 1-1 HN03. Batch experiments were carried out to determine the capacity factors of the solutes of interest and an SnClz reduction in dilute nitric acid was applied to facilitate the elution of AsV and SetV species. When the method was applied to inorganic matrices complete removal of C1 and quantitative analyte recovery at the void volume of the column was obtained. As described elsewhere,15J6 the method eliminates simultane- ously a number of other spectral interferences (generated by S and Br) and therefore allows the determination of Zn Cu V Cr and Mo under similar conditions as those for As and Se.The direct application of the method to complex biologi- cal matrices is favourable from the viewpoint of possible analyte loss and contamination which can occur during a preliminary sample digestion. However generally speaking problems can be encountered with respect to matrix instability and the fact that the reduction and elution characteristics in e.g. human urine and human serum are not necessarily completely similar to those observed for inorganic matrices. The elution of 4-fold diluted urine did not pose specific problems and good results were obtained both for As and Se.When applying human serum to the resin columns in the NO3- form some precipitation in the eluate could be observed particularly when eluting the samples at high elution rates. Therefore a modification of the separation procedure was applied as the serum samples were eluted on a Dowex-1x8 resin in CH3C02- form using dilute acetic acid as an eluent. This modification appeared to be a suitable alternative for the removal of C1- and the determi- nation of As and Se specifically in human serum. The main general disadvantages of the method are the dilution of the samples by a factor 2 2 and the fact that the total salt content of the samples remains the same so that the possible occurrence of matrix effects is not alleviated.The method is however an easily applied and useful ‘clean-up’ procedure for the elimination of spectral interfer- ences particularly the overlap of ArCl on As and Se. Dr. R. Cornelis is acknowledged for her interest and scientific advice particularly with respect to human serum. References 1 Evans E. H. and Ebdon L. J. Anal. At. Spectrom. 1989 4 299. 2 Evans E. H. and Ebdon L. J. Anal. At. Spectrom. 1990 5 425. 3 Lam J. W. and Horlick G. Spectrochim. Acta Part B 1990 45 1313. 4 Branch S. Ebdon L. Ford M. Foulkes M. and O’Neill P. J. Anal. At. Spectrom. 1991 6 151. 5 Beauchemin D. and Craig J. M. Spectrochim. Acta Part B 1990 46 603. 6 Goossens J. Vanhaecke F. Moens L. and Dams R. Anal. Chim. Acta in the press. 7 Janghorbani M. and Ting B. T. G. Anal. Chern. 1989 61 701. 8 Powell M. J. Boomer D. W. and McVicars R. J. Anal. Chem. 1986,58 2864. ‘9 Ting B. T. G. Mooers C. S. and Janghorbani M. Analyst 1989 114 667. 10 Buckley W. T. Budac J. J. Godfrey D. V. and Koenig K. M. Anal. Chem. 1992 54 724. 1 1 Branch S. Corns W. T. Ebdon L. Hill S. and O’Neill P. J. Anal. At. Spectrom. 1991 6 155. 1 2 Lyon T. D. B. Fell G. S. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988 3 601. 13 Lyon T. D. B. Fell G. S. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988 3 265. 1 4 Alves L. C. Wiederin D. R. and Houk R. S. Anal. Chem. 1992,64 1 164. 1 5 Goossens J. and Dams R. J. Anal. At. Spectrom. 1992 7 1167. 16 Vanhaecke F. Goossens J. Dams R. and Vandecasteele C. Tulanta 40 975. I7 Versieck J. Vanballenberghe L. De Kesel A. Hoste J. Wallaeys B. Vandenhaute J. Baeck N. and Sunderman F. W. Anal. Chim. Acta 1988 204 63. L B Haddad P. R. and Jackson P. E. Zon Chromatography Principles and Applications Elsevier Amsterdam 1990 pp. 19 Handbook of Chemistry and Physics eds. Weast R. C. Astle M. J. and Beyer W. H. CRC Press Boca Raton FL 20 Verlinden M. Ph.D. Thesis University of Antwerp 198 1 pp. 21 Allah P. Jaunault L. Mauras Y. Mermet J. M. and Delaporte J. Anal. Chem. 1991 63 1497. 2:2 Campbell M. J. Vermeir G. Dams R. and Quevauviller P. J. Anal. At. Spectrom. 1992 7 617. 1 33- 164. 1983-1984 p. B-134. 44 85-87. Paper 3/01 38381 Received March 9 1993 Accepted April 30 I993

ISSN:0267-9477    

DOI:10.1039/JA9930800921

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 927 Detection Limits Versus Matrix Effects Analysis of Solutions With High Amounts of Dissolved Solids by Flow Injection Inductively Coupled Plasma Mass Spectrometry* Peter Richner Swiss Federal Laboratories for Materials Testing and Research CH-8600 Dubendorf Switzerland Flow injection inductively coupled plasma mass spectrometry (FI-ICP-MS) has been used for the analysis of high-purity nickel (>99.99% m/m). The advantage of analysing solutions with high amounts of dissolved solids (i.e. 3% m/m Ni) offered by FI-ICP-MS compared with conventional solution nebulization with 0.05% Ni solutions results in detection limits in the ng of analyte per g of Ni range because of the small dilution factor. Stable operating conditions can be maintained with the injection of 200 pl sample volumes in a carrier stream of 1% HN03 with a relative standard deviation of less than 2% and 5% drift on 13 consecutive injections in 15 min.For certain elements severe matrix effects caused by the high Ni concentrations were found when the signal depression was compared with 1% HN03 solutions it varied between less than 5% for Rb and 80% for B. This depression is a function of the degree of ionization of the analyte element the smaller the degree of ionization the higher the matrix effect. For quantitative analysis an automated standard additions procedure was applied because it was impossible to compensate for the different matrix effects by internal standards. Keywords Inductively coupled plasma mass spectrometry; flow injection; matrix effects; nickel analysis The demand for the analysis of high-purity materials is growing rapidly because of the wide use of such materials in different industrial applications.Very often these applica- tions require materials with a purity of more than 99.99% m/m. Since in most cases the nature of the impurities is not known before the analysis is carried out only an analytical technique which is able to determine practically all of the elements of the Periodic Table in a very low concentration range can be used to solve this analytical task. Inductively coupled plasma mass spectrometry (ICP-MS) is such a technique. Two different sample introduction techniques are possible either the direct analysis by laser ablation ICP- MS or the analysis of solutions after the dissolution of the sample.Despite the fact that for laser ablation ICP-MS almost no sample preparation is necessary and therefore the possibility of contamination of the sample is smaller the lack of suitable certified reference materials and the poor precision compared with the analysis of solutions make this approach less attractive.' However analysis after dissolution of the samples also has an important disadvantage the total amount of dis- solved solids has to be less than 0.1% m/m otherwise deposition of material will occur on the sample and skimmer cones and this will cause serious drift problems.2 Therefore a large dilution factor from the solid into solution is necessary and this in turn has a negative influence on the detection limits that are achievable.Flow injection (FI) has many applications in the field of atomic spe~trometry,~ ranging from on-line preconcentra- tions4 to automated hydride generati~n,~ elemental specia- tion6 and on-line isotope dilution analy~is.~ Furthermore FI can be used for the analysis of solutions with high amounts of dissolved solids. Since only discrete small sample volumes are introduced into a carrier stream which is fed into the nebulizer the instrument is not exposed to massive amounts of dissolved solids even if very concen- trated solutions are analysed. Flow injection has been applied to the analysis of 1.5-2.0% m/m brine and alumina* and for the determination of T1 Pb and Bi in 0.75% m/m solutions of nickel-based alloy^.^ The determination of Ge Pd and Pt in concentrated phosphoric acid (25% m/v) and ~~ ~ *Presented at the 1993 European Winter Conference on Plasma Spectrochemistry Granada Spain January 10- 15 1993.Table 1 Typical operating conditions R.f. power/W Plasma gas flow rate/l min-l Intermediate gas flow ratell min-l Carrier gas flow ratell min-I Sample uptake rate/ml min-I Dwell time/ms Mass range/u Data acquisition 1000 15 0.8 0.8 1.5 20 6-238 Multichannel analysis 1 point per u (peak hopping) ammonium nitrate solutions (12% m/m) also has been described. l o Because of substantial signal depression by the matrix a standard additions method was used for quantita- tive analysis. This paper describes the advantages and disadvantages of FI-ICP-MS compared with conventional solution nebuliza- tion (CN-ICP-MS) for the analysis of high-purity nickel.The influence of the Ni concentration on the detection limits and the matrix effects was investigated. Furthermore an interpretation of the large differences in the matrix effect for different elements is given. Experimental Instrumentation The work described here was carried out using an Elan 5000 instrument (Perkin-Elmer SCIEX Thornhill Ontario Canada). A cross-flow nebulizer was used in combination with a Scott-type spray chamber. During all of the experi- ments nickel cones were used. The FI device was a FIAS 200 (Perkin-Elmer Norwalk CT USA) with sample loop sizes of 40 200 and 500 pl. All features of the FIAS 200 are controlled by the Elan software. The instrument was optimized for routine multi-element analysis by aspirating a solution containing Mg Rh and Pb at a level of 10 ng ml-I of each. The operating conditions were the same for continuous nebulization and FI and are listed in Table 1.Sample Preparation and Reagents The nitric acid used in this work was re-distilled in quartz from pro anafysi grade nitric acid (Merck Darmstadt,928 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 n-H-7 Sample u P1 Standard U r - r w Carrier JI Waste Fig. 1 Manifold for automated standard additions measurements P1 and P2 pumps; V flow injection valve. The standards are located on the autosampler Germany); de-ionized water from a Milli-Q Plus system (Millipore Bedford MA USA) was used with an indicated outlet resistivity of 18 MQ cm.All vessels were cleaned prior to their use with 10% m/m nitric acid overnight. The nickel chip samples were first etched in 30% m/m nitric acid for 5 min to remove surface contaminants then rinsed with de-ionized water dried and weighed. Dissolu- tion was accomplished with 30% m/m nitric acid in PTFE beakers on a hot-plate at approximately 80 "C. Afterwards the solutions were diluted with de-ionized water to the desired volume. BNRM Ni200 (Analytical Reference Ma- terials International Evergreen CO USA) was used as standard reference material. Recoveries were measured by adding a known amount of analyte to the sample during the dissolution. A 1% m/v HN03 solution was used as a carrier in all FI experiments. In order to minimize the time required for the standard additions measurements an automated method was developed for the on-line addition of the standards as is shown in Fig.1. Three standard solutions with increasing concentrations of the analytes under investigation were placed on the autosampler. Through a second capillary the sample is fed into a T-piece where it comes together with the standard solution and is mixed in a poly(tetrafluor0ethy- lene) capillary (I= 20 cm i.d. = 1 mm) before it enters the FI loop. No differences in the precision compared with the manual addition of the standards were observed. Results and Discussion One of the fundamental parameters in FI is the volume injected. This volume determines the length of the transient signal produced by FI and therefore the amount of time that is available for the measurement of the analyte ions of interest. The signals from the injection of 40 200 and 500 p1 and the continuous aspiration of the same 50 ng ml-I solution are shown in Fig.2. The full width at half maxima (FWHM) of the FI signals vary between 17 s (40 pl) 19 s (200 pl) and 32 s (50 pl). As a result there is in all cases enough time available to perform a multi-element analysis. In association with the analysis of solutions with high amounts of dissolved solids the dispersion D is an important parameter where I, is the intensity for continuous nebulization (CN) measurements; and Z is the peak maximum of the FI ~ignal.~ The dispersion values for the three loops are 8 (40 pl) 1.8 (200 pl) and 1.1 (500 pl). The ability to analyse 8 A 0 30 60 Timeh 90 Fig.2 Signal forms for 50 ng ml-I solution of *OsTl in 1% HN03 tor CN (A) and FI with different loop sizes of B 500; C 200; and I) 40 p1 solutions with high amounts of dissolved solids with FI is based on two different attributes (i) even at the peak rnaximum the concentration of dissolved solids passing the nebulizer is smaller than in CN because of the dispersion; and (ii) during the aspiration of the carrier solution the ICP- Ids system can recover from the temporary overload during the injection of the sample. Therefore less deposition of rnaterial on the cones occurs and the system becomes more stable. In the case of the 500 pl loop this is no longer true. As can be seen from Fig. 2 the Ni concentration is higher than 1% for more than 40 s. Even the short-term stability of the system was poor under those conditions.Hence only the 40 pl and the 200 pl loop were used for the remainder of the work. Detection Limits and Precision The calculation of detection limits (DL) in solutions with high amounts of dissolved solids poses some problems. The aLccepted approach to measuring DL for a solution sample is to calculate the standard deviation (sB) of a suitable blank at the mlz of interest. Ideally this blank would contain all the matrix elements in the sample (ie. Ni) but no analyte. The I>L is then that analyte concentration which produces a s.inga1 of magnitude equal to three times the standard deviation (sB) of the blank. Since a nickel sample with a purity of 100% does not exist the sample with the lowest concentration of the analyte of interest has to be used as a blank.This will give reasonable values for elements that are present in the blank at concentrations below the true DL of the analytical method. On the other hand the procedure will give DLs that are larger than the true DL values for elements that are present at higher concentrations because the sB will be determined by counting statistics. The sB was calculated for FI from five consecutive injections of the blank solution over a time period of 6 min. For CN the blank solution was aspired for 6 min during which five measurements were taken. In both cases no internal standards were used. In Table 2 the DLs for FI with the 40 and 200 pl loops are compared with the results from CN. The FI experiments were made with 3% m/m Ni solutions and CN measure- ments with 0.05% m/m Ni solutions.The estimates for the DLs are certainly too high for elements such as Mg A1 and Zn which are present at the high ng g-I level in the Ni sample used as a blank. On the other hand the DLs for elements such as Li Ga or U should be realistic since their concentrations in the sample were below the DL. Further- more it can be assumed that for elements such as Mg a similar DL could be determined if a suitable blank was available.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 929 Table 2 Comparison of detection limits for FI and CN. All detection limits are given in ng of analyte per g of Ni. Measurements were carried out in 3W m/m Ni solutions for FI and 0.05% m/m for CN FI 40 p1 loop 200 pl loop Element Li Be Mg A1 Mn Zn Ga Rb Sr Ag Cd In T1 Pb Bi U Peak area 12 17 160 350 30 290 8 6 10 10 14 5 1 1 45 1 1 5 Peak height 40 60 170 1600 33 730 16 5 9 31 29 6 1 1 58 1 1 10 Peak area 3 10 100 440 10 210 2 2 1 35 5 3 15 20 12 1 Peak height 3 5 240 2500 13 3 30 10 2 5 46 19 3 16 66 12 2 CN 110 50 780 15000 135 920 85 135 55 95 100 17 25 320 80 17 For the evaluation of transient signals two different approaches are possible peak area or peak height measure- ments.The results in Table 2 show that for most elements lower DLs are achieved when peak area is used. In addition the DLs in the case of the 200 pl loop are in most cases better than for the 40 pl loop. This is mainly due to better sensitivity (Fig. 2). Compared with the 200 pl loop the DLs with CN are approximately one order of magnitude higher.It is important to note that in Table 2 the DLs in the solid are compared (ng of analyte per g of Ni) not the DLs in the analysed solutions. If 1% HNO solutions are used the comparison is clearly in favour of the CN. In this case the DLs for FI are normally considerably higher than in CN because of the lower sensitivity in FI. In the case of the analysis of Ni solutions the dilution factor for FI is 60 times smaller than for CN (33 for FI and 2000 for CN) and this difference makes up for the loss of sensitivity in FI. Drift becomes a serious problem when solutions with high amounts of dissolved solids are analysed. A compari- son of drift and relative standard deviation (RSD) for different solutions over a period of 15 min is shown in Table 3 using 238U as an example.The results for other elements are similar. The same solutions (1 O/o HNO 0.05% Ni-1% HN03 and 3% Ni-3% HNO,) were aspirated continuously for the CN measurements and a measurement was made every 72 s. Even for 0.05% Ni solutions internal standards must be used because of a drift of 9.5% during this time period. When 20*Pb is used as an internal standard both drift and RSD can be improved significantly. How- Table 3 Stability of Fl and CN measurements for different amounts of dissolved solids over a period of 15 min for a 50 ng ml-1 of 238U solution with and without the use of 208Pb as an internal standard (IS) RSD (n= 13) Drift Solution No IS With IS No IS With IS 1% HNO 0.8 0.9 1.9 1.7 0.05% Ni 2.7 1.1 9.5 5.0 3% Ni 17 3.4 45 10.6 1% HNO 1.7 0.9 5.3 3.8 3% Ni 1.4 0.8 4.8 2.5 Continuous nebulization (%)- Flow injection (%)- 0 4 8 12 Ti m e/m i n Fig.3 Thirteen replicate injections of 50 ng ml-I of 238U in 3% Ni-3% HNO solution. The relative standard deviation is 1.4% without standard and 0.8% with 208Pb as internal standard using peak area measurements ever in the case of the 3% Ni solution this method does not work. A drift of 10.6% in 15 min is unacceptable for quantitative analysis. In contrast to the CN results the FI data are very promising. During the 15 min time period 13 injections of 200 jd were made (Fig. 3). The results for a 1% HN03 solution and the 3% Ni solution are practically the same. If zo8Pb is used as an internal standard both drift and RSD become comparable to the results for CN measure- ments with 1% HNO solutions.The values for RSD and drift varied from day to day. This is not surprising given the fact that the blockage of the orifices is a very erratic process. However the same relative behaviour as that given in Table 3 was always found. Matrix Effects While the advantages of FI compared with CN are very clear in the case of the DLs the picture is different for matrix effects. The matrix effect M is calculated according to eqn. (2) sensitivity in Ni solution sensitivity in 1% HNOJ M= (2) The matrix effects for 1% Ni and 3% Ni solutions for 28 elements are shown in Fig. 4 for an injection volume of 200 pl. Because of the scattering of the matrix effects a930 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL.8 1.2 0.9 + P) 0.6 .- L CI s 0.3 I Y 0 Li B Al Cr Cu Ga Sr Ag In La Lu Pt TI Bi Be Mg Ca Mn Z n Rb Nb Cd Ba Nd Ta Au Pb U Fig 4 200 pl s z - .- Fig. 5 Matrix effects for 28 elements in 0 1% Ni and a 3% Ni; loop peak area measurements 2.0 1.6 1.2 0.8 1 .o 0.8 4- 0.6 F X .- L 0.4 H Time/s Matrix effect (B) as a function of the Ni concentration at the nebulizer c ~ ~ ~ ~ ~ (A) for IIICd; 200 pl loop ~ ~ ~ ~ ~ = 3 ~ h compensation by the use of internal standards is impossi- ble. The matrix effect in 1% Ni solutions varies between 1.15 (Sr) and 0.6 (B). Using CN with 0.05% Ni solutions results in smaller matrix effects and the range of scattering is less pronounced (0.75 for B to 1.05 for Rb). However in this case it is also preferable to use a standard additions method for quantitative analysis.The concentration of the matrix element Ni undergoes the same dilution as do the analytes. Since the matrix effect is a function of the Ni concentration the matrix effect itself also changes during the transient signal. For each reading during the analysis the actual concentration of Ni at the nebulizer c ~ ~ ~ ~ ~ can be calculated using the intensity I in a 1% HN03 solution at this point of the signal (3) IF1 cNi,lCP= cNi,sol - ICN where c ~ ~ is the Ni concentration in the solution. The signal profile of cNi,]Cp for the injection of 200 pl of a 3% Ni solution is shown in Fig. 5 together with the matrix effect for I1lCd. At the peak maximum c ~ ~ ~ ~ ~ reaches 1.6% and the sensitivity for Cd is reduced to 30% of that in a I% HN03 solution.By using peak area measurements an average matrix effect is determined for the complete transient signal and therefore the value for Cd shown in Fig. 4 is only 0.55. The differences in the matrix effects for the elements investigated in this work cannot be explained simply as a mass bias. When elements with similar masses such as Li- Be-B Ag-Cd-In or Ta-Pt-Au-T1 (Fig. 4) are compared it becomes clear that the degree of ionization of the analyte has an important influence on the matrix effect. The ionization temperature is considerably changed by the 0 Li B Al Cr Cu Ga Sr Cd Ba Pt TI Bi Be M g Ca M n Z n Rb Ag In Ta Au Pb Fig. 6 Matrix effects (a) in 3% Ni solutions (200 pl loop) and degrees of ionization (0) for 28 elements introduction of high amounts of Ni into the ICP.A theoretical calculation of the ionization temperature is rather complicated. The departure of the ICP from a local thermodynamic equilibrium (LTE) has been discussed extensively in the 1iterature.ll Despite the fact that the absolute values for the ionization temperature and the dlegree of ionization based on LTE calculations are wrong the relative values for different elements are correct. The dlegree of ionization was calculated from the Saha equa- tion:l2 where a is the degree of ionization; n the electron number dLensity ( ~ m - ~ ) ; 2 and 2 are partition functions of the ion and the atom respectively; Ti is the ionization temperature (K); and I is the first ionization energy (ev). The partition fiunctions were calculated according to the method of de Galan et An ionization temperature TI of 6000 K and an electron number density n of 1 x 1014 ~ m - ~ were assumed.Matrix effects and degrees of ionization of different elements are compared in Fig. 6. There is a clear correlation between the two parameters the smaller the degree of ionization the stronger the matrix effect (Li-Be-B Ag-Cd-In Ta-Pt-Au-Tl). This is in accordance with the rlesults found in the literature where larger matrix effects were found for Ge and Pt than for Pd.lo If elements with similar degrees of ionization are compared such as Li Rb In and T1 the smallest matrix effects are found for elements in the middle of the mass range. However it seems that this blehaviour is also a function of the ion lens settings.Quantitative Analysis 'To find out what the nature of the impurities is a slemiquantitative analysis covering the whole mass range has to be performed first. Using a dwell time of 20 ms it is possible to do 12 scans from 6Li to 238U during the injection of 200 pl of sample. This means that for all m/z values only every 4 s is an intensity measured. In order to obtain a reasonable repeatability of the analysis the start of the measurement is triggered by the FI device. Therefore the 12 points measured for each mlz are always located on the srame place on the transient signal. Values for the RSD of less than 5% are achieved routinely at the 50 ng ml-1 level. Once the approximate composition of the impurities has been determined a quantitative analysis for all elements fbund is made using a standard additions technique.In order to check the accuracy of the method standard reference materials (SRMs) with a composition similar to tlhe samples should be analysed. However no SRMs at the purity level under consideration are usually available andJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 93 1 reduction in their degree of ionization. In order to compen- stmdard additions techniques can be applied something which is carried out rather easily because of the high degree for the Of SRM BNRM Ni200 (mean Of sate for this effect which is distinct for each element only three dissolutions and standard deviation) and recoveries in a high- purity sample (99.99%) spiked with 100 ng of analyte per of Ni Element A1 Ti Cr Mn Fe c u Nb BNRM Ni2OO/pg g-I Certified value FI-ICP-MS 300 315+20 340 330-t 15 50 48k5 2000 2050 -t 75 200 195k35 50 51 2 3 10 8+-2 Recovery in high-purit y sample (Yo) 91 92 101 94 92 103 95 - - of automation found in modern instrumentation.It would appear in routine analysis of high-purity materials with FI-ICP-MS the detection limits and the precision of the measurements are mainly determined by blank values in the reagents and the containers used. Despite the fact that direct analysis by laser ablation ICP- MS would not require any sample preparation and there- fore less contamination problems would occur the advan- tages of the calibration procedures available for solution analysis make this approach still more attractive especially in cases such as nickel where the dissolution of the sample is very simple.only a few impurities are certified. Therefore a 99.6% Ni SRM with certified values for seven elements had to be analysed (Table 4). To verify the accuracy of the method recoveries in the sub-pg- I range were determined for several elements in a high-purity sample where the concen- trations of these elements were either below the detection limit of the method or at the sub-pg g-l level (Table 4). Both of these tests show good results. Conclusions The advantages of FI-ICP-MS for the analysis of high- purity materials such as metals or ceramics are demon- strated using nickel as an example. Assuming a purity of 99.99% m/m the total amount of impurities after dissolu- tion of the sample is 50 ng mi-* for CN-ICP-MS (0.05% Ni matrix) and 3000 ng ml-I for FI-ICP-MS (3% Ni matrix).Despite the fact that the DLs for CN-ICP-MS in solutions are better than for FI-ICP-MS the latter method gives much better results in this case because of the smaller dilution factor. While the stability of the ICP-MS system does not pose any problems the matrix effects have to be taken into account when quantitative results are required. The ioniza- tion temperature in the ICP changes considerably due to the introduction of high amounts of Ni into the plasma. Therefore elements with a low degree of ionization such as B Cd or Au undergo a large signal depression due to References 1 Handbook of Inductively Coupled Plasma Mass Spectrometry ed. Jarvis K. E. Gray A. L. and Houk R. S. Blackie Glasgow 1992. 2 McLeod C. W. J. Anal. At. Spectrom. 1987 2 549. 3 Tyson J. Spectrochim. Acta Rev. 1991 14 169. 4 Caroli S. Alimonti A. Petrucci F. and Horvath Zs. Anal. Chim. Acta 1991 248 241. 5 Liisztity A. VicziAn M. Wang X. and Barnes R. M. J. Anal. At. Spectrom. 1989 4 76 1. 6 Vollkopf U. Gunsel A. and Janssen A. At. Spectrosc. 1990 11 135. 7 Thompson J. J. and Houk R. S. Anal. Chem. 1986,58,2541. 8 Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988 3 547. 9 Mochizuki T. Sakashita A. Iwata H. Isibashi Y. and Gunji N. Anal. Sci 1990 6 191. 10 Peng Z. Klinkenberg H. Beeren T. and Van Borm W. Spectrochim Acta Part B. 1991 46 1051. 1 1 Inductively Coupled Plasma Emission Spectroscopy Part II Applications and Fundamentals ed. Boumans P W. J. M. Wiley New York 1987 ch. 1 1. 12 Theory of Spectrochemical Excitation Boumans P. W. J. M. Hilger and Watts London 1966 pp. 156-1 70. 13 de Galan L. Smith R. and Winefordner J. D. Spectrochim. Acta. Part B. 1968 23 521. Paper 3/01 1421 Received February 26 I993 Accepted April 30 1993

ISSN:0267-9477    

DOI:10.1039/JA9930800927

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 933 Abdallah Amin M. 759 Ahmad Rasheed 91 1 Akatsuka Kunihiko 279 Alvarado J b e 253 Andreae Meinrat O. 119 Aoki Hiroyuki 41 5 Arpadjan Sonja 85 Ascanelli Monica 905 Axner Ove 375 Azerado M. A. 279 Bailey Elizabeth H. 551 Barciela-Alonso C. 649 Barciela-Garcia J. 649 Barnes Ramon M. 467 Barshick Christopher M. 875 Bastos M. Lourdes 655 Baxter Malcolm 69 1 Beceiro-Gonzhlez E. 649 Bell Gordon M. 91 1 Benito R. 839 Berman Shier S. 279 Bermejo-Barrera A. 649 Bermejo-Barrera P. 649 Berndt Harald 243 Birchall James D. 91 1 Blades Michael W. 261 Blais Jean-Simon 659 Blanco GonzBlez Elisa 8 15 Bloom Nicolas S. 591 Bloxham Martin J. 499 Boer Peter 61 1 Boer Walther H. 61 1 Boonen Sylvie 71 1 Borer Matthew W. 333 339 Bosi Michele 755 Botto Robert I.51 Boumans Paul 767 Brenner Isaac B. 475 833 Brindle Ian D. 287 Brockman Andreas 397 Brown Garrett N. 21 1 Bucci Gianna 905 Burguera J. L. 229 235 Burguera M. 229 235 Byrne John P. 599 Cai Yong 119 Chmara Carmen 745 Campos Reinaldo C. 247 Caiiada Rudner P. 705 Canals Antonio 109 Can0 Pavdn J. M. 705 Cao Jieshan 379 Carridn Nereida 493 Caruso Joseph A. 427 545 Castillo Juan R. 643 665 Celliers Peter M. 809 Chakrabarti Chuni L. 599 Chakraborti Dipankar 643 Chang Mou-sen 379 Chenery Simon 299 Cheng Jianguo 623 Chen Yalei 379 Chikuma Masahiko 4 15 Chirinos Jose 493 Ciocan Adeline 273 Coba Isabel G. 827 Coedo Aurora G. 827 915 749 843 571 787 737 623 CUMULATIVE AUTHOR INDEX FEBRUARY-SEPTEMBER 1993 Cook Jennifer M. 299 Corns Warren T.71 Cortez Jesus Arroyo 103 Cresser Malcolm S. 269 Crews Helen 691 Cristiano Ana Rita 253 Curtius Adilson J. 247 749 Dams Richard F. J. 433 71 1 781 921 Darke Susan A. 145 Dawson J. B. 51 7 DeGendt S. 859 de la Calle Guntiiias Maria Beatriz 745 del Barrio S. 839 de Oliveira Elisabeth 367 DoEekal Bohumil 637 763 Doidge Peter S. 403 Dorado Teresa 827 Dorfman E. 833 Duckworth Douglas C. Eastgate Alan R. 305 Ebdon Les 71 691 723 El-Defrawy Mohamed M. Eloi Corinne 217 El-Shamy Manal M. 759 Escudero Ester 827 Evans E. Hywel 1 427 Fagioli Francesco 905 Fahal Ibrahim H. 91 1 Fang Zhaolun 577 Fernandez de la Campa M. Fernhndez Shnchez Maria Ferreira Margarida A. 655 Fischer Johann L. 487 Fisher Andrew S. 691 Foner Henry 467 Fransen Rend 61 1 Freedman Philip A. 19 Fry Robert C.305 Galley Paul J. 65 7 15 Gangadharan S. 127 Garcia Alonso Jost Ignacio Garcia de Torres A. 705 Giglio Jeffrey J. 1 Gilchrist Glen F. R. 623 Gill C. G. 261 Gilmutdinov Albert Kh. 387 Gint Maria F. 243 Giovanonne Bruno 673 Gluodenis Thomas J. Jr. Golloch Alfred 397 Gomez M. M. 461 Goodall Phillip 723 Goossens Jan 921 Gower Grant H. 305 Granadillo Victor A. 61 5 Gray Alan L. 899 Grtgoire D. Conrad 599 Hahn Lothar 223 Haigh P. E. 585 Halicz L. 475 Hanna C. P. 585 Hansen Steen Honore 557 Harnly James M. 317 875 759 R. 821 847 Luisa 815 673 843 809 697 Harrison W. W. 859 Hasegawa Ryosuke 48 1 Hernhndez Cbrdoba Manuel Hemandis Vincente 109 Hiddemann Lars 273 Hieftje Gary M. 65 333 Hillamo Risto E. 79 Hill Steve J. 71 499 723 Holclajtner-Antunovic Ivanka D. 349 359 Honda Masatake 453 Huang Chung-Wen 681 Hughes Dianne M.623 Huneke John C. 867 Hutton Robert C. 867 Ince Ahmet T. 899 Inui Syn-ya 595 Israel Yecheskel 467 Ivaldi Juan C. 795 Jakubowski Norbert 88 1 Jarvis Kym E. 25 Jiang Shiuh-Jen 68 1 JimCnez Maria S. 665 JurasoviC Jasna 4 19 Karadjova Irina 85 Kemp Anthony J. 551 Koch Lothar 673 Kojima Isao 11 5 Kondo Shinji 115 Koomans Hein A. 6 1 1 Krivan Viliam 637 Krug Francisco J. 243 Krushevska Antoaneta P. Kujirai Osamu 481 Kumar Sunil Jai 127 Laborda Francisco 643 737 Lachica M. 853 Lam Joseph W. H. 279 Lamoureux Marc M. 599 Larsen Erik H. 557 659 Liang Dong C. 809 Liang Lian 591 Li Anmo 633 Liao Yiping 633 Littlejohn D. 325 Longerich Henry P. 371 439 L6pez Garcia Ignacio 103 Luong Van T. 41 Lu Yi Ming 8 15 Ly Tam 599 Madrid Yolanda 745 Maenhaut Willy 79 Mahalingam T.R. 565 Majidi Vahid 21 7 Marchante Gaybn Juan M. Martines Laura J. 467 Masuda Kimihiko 687 Mathews C. K. 565 Mazzetto G. 89 McAllister Trevor 403 McIntosh S. 585 McLaren James W. 279 McLeod C. W. 461 Mermet Jean-Michel 795 Milella E. 89 Mingorance M. D. 853 Mir J o d M. 643 737 Miyazaki Akira 449 Mizuno Takayuki 595 103 339 715 467 73 1 Moens Luc J. 71 1 921 Mohammad Bashir 325 Morgan C. A. 539 Murillo Miguel 493 Muller-Vogt German 223 Nagai Hisao 453 Navarro Janeth A. 6 15 Nawar Nagwa 759 Niemax Kay 273 Nonn Christine 397 Norberg M. 375 Nbbrega Joaquim A. 243 Ohl Andreas 803 Ohlsson K. E. Anders 41 Ohorodnik S. K. 859 Ohta Kiyohisa 595 Ozaki Elisa Akemi 367 Pakkanen Tuomo A. 79 Perez-Vazquez M. L. 853 Platzner I.19 Poluzzi Vanes 755 Potts Philip J. 293 Prabhu R. Krishna 565 Pretty Jack R. 545 Price W. J. 5 17 Pritzl Gunnar 557 Qi Wenqi 379 Rademeyer Cor J. 487 Radziuk Bernard 409 Ragnarsdottir K. Vala 5 5 1 Raith Angelika 867 Rapsomanikis Spyridon Reimer Paul A. 449 Reis Boaventura F. 243 Ren J. M. 59 Rezende Mario do Carmo Richner Peter 45 927 Roberts Noman B. 91 1 Robertson J. David 2 17 Romero Romer A. 61 5 Ropcke Yiirgen 803 Ruiz Ana I. 109 Salin Eric D. 59 Shnchez Uria J. Enrique 731 Sanz-Medel Alfredo 73 1 815 821 847 915 Sass VPnia A 243 Schmidt Martin 803 Sella Silvia M. 749 Sen Gupta Joy G. 93 Sentimenti E. 89 Sesi Norman N. 65 Shan Xiao-quan 409 Shimamura Tadashi 453 Silva Ivana A. 749 Sjostrom S. 375 Smith B. W. 539 Smith David H. 875 Snook R. D. 517 Soares Elisa M.655 Steers Edward B. M. 309 Stockwell Peter B. 71 723 Story W. Charles 571 Stroh Andreas 35 Stuewer Dietmar 88 1 Stuhne-Sekalec Lidija 445 Sturgeon Ralph E. 41 Styris David L. 21 1 Sugiyama Takehiko 595 Suzuki Tohru 595 119 247 687934 Taddia Marco 755 Takahashi Takako 453 687 Takaku Yuichi 687 Tanner Scott D. 891 Tao Guanhong 577 TeliSman Spomenka 4 19 Templeton Douglas M. 445 Thoby-SchultzendorfT Dominique 673 Thompson K. Clive 723 Thorne Anne P. 309 Tong S. L. 859 Trentini Pier Luigi 905 TripkoviC Mirjana R. 349 359 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1993 VOL. 8 Tserovsky Emil 85 Tyson Julian F. 145 585 Uebbing Jurgen 273 Ure Allan M. 325 Valdts-Hevia y Temprano M. C. 821 847 Valle F. J. 839 Vandecasteele Carlo 433 Vanhaecke Frank 433 Vanhoe Hans 781 Vela Nohora P.787 Venturini Francesco 905 Vereda Alonso E. 843 697 78 I 'Verrept Peter 71 1 'Vijayalakshmi S. 565 'Viiiuales Jorge 737 'Viswanathan K. S. 565 'Voloshin A. V. 387 'Vollkopf Uwe 35 'VyskoEilovA Olga 409 'Walder Andrew J. 19 'Watson John S. 293 Webb Peter. C. 293 Welz Bernhard 409 Wendl Wolfgang 223 Williams John G. 25 Willie Scott N. 41 899 Winefordner J. D. 539 Woo Jin Chun 881 Worsfold Paul J. 499 Wrbbel Kasia 915 Wunderli Samuel 45 Xu Sonny X. 445 Yamada Kei 481 Yang Huacheng 809 Yoffe O. 475 Yu Changbin 809 Zakharov Yu. A. 387 Zanforlini Bernardette Zheng Shaoguang 287 69 1 905

ISSN:0267-9477    

DOI:10.1039/JA9930800933

出版商:RSC    

年代:1993

数据来源: RSC