JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 199 1 VOL. 6 605 Minimization of Non-spectroscopic Matrix Interferences for the Determination of Trace Elements in Fusion Samples by Flow Injection Inductively Coupled Plasma Mass Spectrometry Jiansheng Wang E. Hywel Evans and Joseph A. Caruso* Department of Chemistry University of Cincinnati ML 172 Cincinnati OH 45221 USA The efficiency of various ion lens tuning strategies flow injection and the use of an internal standard has been investigated for the determination of trace elements in real samples. Spike recoveries indicated that internal standardization was always preferable if 100% recovery was to be obtained. However good recoveries were also obtained by tuning the ion lenses for a maximum ll5In+ signal in a matrix blank solution containing 100 ng g-l of In while using continuous nebulization sample introduction.Results indicated that tuning the ion lenses in the presence of the matrix yielded improved analyte recoveries when compared with tuning the lenses for a standard solution. Keywords Matrix tuning; flow injection; internal standard; inductively coupled plasma mass spectrometry Inductively coupled plasma mass spectrometry (ICP-MS) has been used for the determination of trace elements in various samples containing high matrix concentrations 1-4 in which severe matrix induced analyte signal suppression or enhancement usually occurs. Various techniques have been applied to reduce non-spectroscopic interferences. These include internal standardizati~n,~-~ isotope dilu- t i ~ n * * ~ standard additions,*-lo hydride generation,' 1*12 ion- exchange de-salting and prec~ncentration*~-~~ and flow inJection.16-19 A recent investigation*O has shown that flow injection (F'I) and re-tuning the ion lenses can be used to compensate for the suppression of analyte signal in the presence of a high concentration of matrix.The flow injection technique is considered to compensate for the matrix effects by lowering the amount of matrix-containing solution to which the nebulizer sampler and plasma are exposed thereby effectively reducing the effects of clogging. Ion-lens tuning in the presence of the matrix (such as synthetic ocean water) also reduced signal suppression in comparison with tuning the ion lenses for a standard solution. In this work the results obtained by comparing contin- uous nebulization flow injection standard and matrix tuning both with and without internal standardization for the determination of trace elements in real samples are presented.By comparing the recovery factors for samples spiked with the elements being studied the reliability of the analytical results has been evaluated. This work is intended to supplement a previous paper.*O Experimental Instrumentation All data were acquired using a commercial ICP-MS instru- ment (VG PlasmaQuad VG Elemental Winsford Chesh- ire UK). The operating conditions used for this work are shown in Table 1 and are typical of those used for routine multi-element analysis. A concentric nebulizer (Meinhard C-2 Precision Glassblowing of Colorado Parker CO USA) and a double-pass Scott-type spray chamber cooled to 6 "C by means of a refrigerated chiller (Neslab Instruments Portsmouth NH USA) were used.The spray chamber was maintained at 6 "C in order to reduce the amount of solvent vapour thereby reducing condensation on the torch elbow and to maintain the spray chamber at a constant tempera- *To whom correspondence should be addressed. Table 1 ICP-MS multi-element operating conditions ICP system- Forward power Reflected power Coolant flow rate Auxiliary flow rate Nebulizer flow rate Sample delivery rate Spray chamber temperature Sampling depth* 1350 W < 5 w 16 1 min-' 1 1 min-* 0.65 I min-' 1 ml min-I 6 "C 12 mm Mass spectrometer- Sampler nickel 0.7 mm orifice Skimmer nickel 1.0 mm orifice First stage pressure 1 .4 ~ lo2 Pa Second stage pressure < l x Pa Third stage pressure X ~ X lo-' Pa *Defined as the distance between the foremost coil of the load coil and the tip of the sampling cone. ture to improve precision. Solution was introduced via a peristaltic pump (Gilson Villiers Le Bel France) at a flow rate of 1 ml min-l into the concentric nebulizer. After each run period the nickel sampler was cleaned. Reagents Samples and Standards The sample stock solutions were obtained from the labora- tories of British Petroleum (Warrensville Research Center Cleveland OH USA). The solid samples were fused with a sodium carbonate flux (EM Science Gibbstown NJ USA) in a platinum crucible in the ratio of 0.1 + 3 (mass samples to mass flux) then dissolved in 1 + 3 HCl one part HC1 (Baker Instra-analyzed Phillipsburg NJ USA) in three parts distilled de-ionized water (DDW 18 Ma Barnstead Newton MA USA) so that the final concentration of dissolved solids was approximately 3% d m .The samples used for the analysis were prepared by diluting the sample stock solutions ten times with DDW with In (100 ng g-l) added as the internal standard. The spiked sample solutions (50 ng g-l of each element) were prepared by adding the appropriate amount of stock solutions containing Be Al V Co Ge As Se Y Rh Ba and Pb at the same time as the internal standard. The matrix blank solution was prepared by combining 7.5 g of sodium carbonate flux with 1 + 3 HCl to a total mass of 250 g and then diluted ten times with DDW so that the606 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1991 VOL.6 final concentration of the matrix blank was 0.3% m/m Na2C03 and 0.9% m/m 12 mol dm-3 HC1. At this time In (100 ng g-l) was also added as the internal standard. Multi-element stock solutions (1 0 pg g-l) of Be Al V Co Y Rh Ba and Pb and Ge As Se Sn and Sb were prepared from 1000 pg ml-l single element stock solutions (certified atomic absorption standards; Fisher Scientific Fair Lawn NJ USA) in 5% nitric acid and 5% hydrochloric acid respectively. Dilute acids were prepared by dilution of concentrated HN03 (laboratory-reagent grade Fisher Scientific) and concentrated HCl (Baker Instra-analyzed) with DDW. Standard solutions were prepared by serial dilution of the multi-element stock solutions in 1% HN03 to concentrations of 5 20 100 and 200 ng g-l.Flow Injection System The flow injection system used for this work has been described in a previous paper.*O The system consisted of a programmable electronic controller (Machine Shop University of Cincinnati Cincinnati OH USA) a flow injection solenoid valve (Model 7 163 Rheodyne Cotati CA USA) a 4-way poly(tetrafluoroethy1ene) rotary valve (Model 5701 Rheodyne) and a 100 pl loop (Valco Instru- ments Houston TX USA). In practice the flow injected samples gave rise to signal peak widths for the analyte of approximately 60 s at half-height which allowed multi- element scans to be performed as the method of data acquisition provided that the start of each scan was timed to begin at the same point for each flow injected sample.This flow injection system allowed the timing of injections and sample volumes to be accurately controlled. Data Acquisition and Calculations All data acquisition was performed in the scanning mode using the software supplied with the instrument. The following data acquisition conditions were used mass range 8-2 13 mlz; channels 2048; sweeps of the mass range 400; and dwell time per channel 80 ps. Calibration and calculations were performed using the instrument software. Procedure Before each batch of samples was analysed the signal was monitored while a matrix solution containing 100 ng g-l of In was aspirated into the system. Initially the signal decreased as matrix was deposited on the cones until the equilibrium was reached when no further obvious decrease was observed.This process took approximately 15 min. After that the system was washed out with 1% HN03 for 5 min. Standard tuning Ion lenses were tuned for a maximum Il5In+ signal without any matrix present using a 100 ng g-l solution of In in 1% HN03. Calibration for all the elements of interest was accomplished by analysing a blank and four standard solutions containing Be Al V Co Ge As Se Sn Sb Y Rh Ba and Pb. Both blank and standards contained 100 ng g-l of In. The sample solutions were analysed in the following order (1) matrix blank containing 0.3% m/m of sodium carbonate and 2.5% HCl with 100 ng g-l of In; (2) sample; (3) sample spiked with 50 ng g-l of the standard; (4) standard blank containing 1% HN03 with 100 ng g-l of In; ( 5 ) 20 ng g-l of standard solution; and (6) wash out with 1 To HN03 for 3 min (continuous nebulization only).This procedure was performed twice first with contin- uous nebulization and then with flow injection. The purpose of steps 4 and 5 was to check the signal drift during the analysis caused by cone clogging. This procedure was repeated for each sample in succession. Each step was repeated three times within a specific procedure for both continuous nebulization and flow injection. The flow injection carrier stream was 1% HN03. Matrix tuning Ion lenses were tuned for a maximum llsIn+ signal with the matrix present using a solution of 100 ng g-l of In in the matrix blank. The blanks standard solutions and samples were analysed as described under Standard tuning. This procedure was again performed twice first with continuous nebulization then with flow injection. Calculation of Recovery Factors Average analyte signals were obtained by calculating the mean of the values found for three repetitions each of the blanks and samples.The sample spike recoveries were calculated by means of the equation Cspiked - &ample Spiked sample recovery factor = 50 where Cspikd is the analyte concentration (in ng g-l) determined in the spiked sample after calibration; csample is the analyte concentration determined in the sample; and 50 represents the concentration (50 ng ml-l) of the spike. The standard recovery factors were calculated as follows Standard recovery factor = - where Csmndard is the concentration of analyte determined in the 20 ng g-l standard solution by calibration and extrapo- lation; and 20 represents the true concentration (20 ng ml-l).20 Semi-quantitative Analysis One of the advantages of ICP-MS is that semi-quantitive analysis can be performed relatively quickly for most of the elements in the Periodic Table. The method is useful for determining the approximate concentration of analyte in unknown samples. The instrumental mass response graph was established by running a solution containing 100 ng g-l of Be Mg Co In Pb and U in 0.3% Na,CO and 2.5% HCl which matched as closely as possible the sample matrix. The elements spiked to the unknown samples were based on information obtained from semi-quantitative analysis. No significant amount of In was found in the samples. Results and Discussion Recoveries for 20 ng g-l Standard Solutions A 20 ng g-l standard solution was run after each sample analysis.The standard recovery factors were used to monitor the signal drift resulting from the deposition of solids on the ICP-MS interface. If no cone or nebulizer blockage occurred during the analysis the recovery factor should be equal to unity. Values of less than unity indicate signal suppression and values of greater than unity indicate signal enhancement. The recovery factor obtained for 20 ng g-l of the Rh standard spike as a function of time is shown in Fig. 1. As can be seen the recovery factor for continuous nebulization (curve A) decreased as more samples were analysed probably owing to a build-up of saltJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1 99 1 VOL.6 607 I 1 1 33 66 99 132 165 198 Ti me/mi n 0.4 ' Fig. 1 Recovery factors for 20 ng g-I of Rh in 1% HN03 as a function of time A continuous nebulization without an internal standard; B flow injection without an internal standard; and C flow injection with an internal standard from the sample matrix on the sampler or skimmer cones. Although the cones had been conditioned beforehand and 1% HNOj was aspirated for 3 min before each sample was introduced cone blockage was still severe for continuous nebulization sample introduction. However for flow inj ec- tion (curve B) the recovery factors were improved to some extent because flow injection effectively reduced the amount of sample solution coming into contact with the ICP-MS interface reducing the possibility of the build-up of solid material.Further improved recoveries were achieved by using flow injection and internal standardiza- tion with In as shown in curve C. Recoveries for 50 ng g-1 Sample Spikes The unknown samples were spiked with 50 ng g-l standards to ascertain which analytical protocol gave the best re- coveries (i.e. those closest to unity). Based on these results the recoveries for trace elements in the samples could be evaluated with greater confidence. Fig. 2(a)-(d) illustrates spike recoveries for various elements covering the mass range in one sample analysed. Almost identical graphs were obtained for the other samples (results not given). The shaded bars represent the spike recoveries obtained by tuning the ion lenses with the standard solution in l0h HN03.The unshaded bars repre- 1.2 1 .o 0.8 0.6 $ 0.4 P o $ 1.2 8 1.0 5 0.2 'c a 0.8 0.6 0.4 0.2 0 Fig. 2 Recovery factors for unknown samples spiked with 50 ng g-* standards. The shaded bars represent the spiked recoveries obtained by standard tuning; the unshaded bars represent the recoveries obtained by matrix tuning. (a) Continuous nebulization without an internal standard; (b) flow injection without an internal standard; (c) continuous nebulization with an internal standard and (d) flow injection with an internal standard sent spike recoveries obtained by tuning the ion lenses in the presence of matrix (0.3Oh Na,C03 and 3% HCl). Each graph represents a different analytical protocol. Fig. 2(a) shows the recoveries obtained using continuous nebulization.As can be seen the recoveries obtained by tuning with the standard are much lower than unity indicating severe signal suppression. However by tuning the ion lenses in the presence of the matrix recoveries were much improved for most elements. In contrast when flow injection was employed [Fig. 2 (b)] no significant difference was observed between recoveries obtained by standard tuning and matrix tuning. Furthermore a comparison between Fig. 2(u) and (b) reveals that in general lower recoveries were obtained using flow injection compared with continuous nebulization when matrix tuning was utilized while the reverse was true when standard tuning was employed. The explanation for this is 2-fold. Two factors affect recovery namely the tuning conditions of the ion lenses and the concentration of sample matrix.For the situation where standard tuning was employed the effect of sample matrix concentration was predominant resulting in better recoveries using flow injection compared with con- tinuous nebulization owing to the effective dilution of the sample by the former technique. For the situation where matrix tuning was employed the effect of the lens tuning conditions was predominant. As all ion lens tuning was performed with a continuous signal (ie. using continuous nebulization) the resulting conditions may not have been optimum for a transient flow injection signal where the matrix was effectively more dilute. Hence better recoveries were obtained using continuous nebulization. Figs. 2(c) and (d) contain an analogous set of plots to Figs.2(u) and (b) except that this time an internal standard was used. For both continuous nebulization and flow injection very little difference was observed regardless of whether ion lenses were tuned in the presence of the matrix or for an aqueous standard. However in general recovery factors much closer to unity were obtained compared with those found without using an internal standard. The best recover- ies were obtained for lo3Rh and 138Ba in all samples. This was thought to be because the internal standard used was llsIn. These three elements are closest in terms of mass compared with 59C0 89Y and *08Pb so that if other factors such as the extent of ionization had a negligible influence this result would be expected and has been observed by other worker^.^*^ From this point of view excellent recover- ies could be obtained if more internal standards were used to cover the whole mass range.Based on the spike recoveries the analytical procedures were arranged in order of increasing recovery in Table 2 so Table 2 Analytical methods used and arranged in ascending order of spike recoveries from top to bottom Method Conditions 1 2 3 4 5(a) 5(6) 5(c) 5(d) Continuous nebulization without internal standard but Flow injection without internal standard but with Flow injection without internal standard but with Continuous nebulization without internal standard Flow injection with internal standard and standard Flow injection with internal standard and matrix Continuous nebulization with internal standard and Continuous nebulization with internal standard and with standard tuning standard tuning matrix tuning but with matrix tuning tuning tuning standard tuning matrix tuning608 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY DECEMBER 1991 VOL.6 Table 3 Results for the determination of 59C0 Io3Rh and 208Pb. All values are concentrations in ng g-l. (Methods as defined in Table 2) S9C0 in sample lo3Rh in sample *O*Pb in sample Method of choice 1 2 3 4 1 2 3 4 1 2 3 4 12.9 3.63 2.00 0.23 13.3 4.24 2.40 0.32 16.6 5.00 2.66 0.38 19.3 6.00 3.46 0.50 19.2 6.21 3.50 0.53 19.7 6.70 3.76 0.59 22.8 6.86 4.12 0.53 20.8 7.90 4.54 0.56 27.5 19.2 10.5 13.9 32.4 25.1 13.8 19.5 39.1 28.1 15.1 20.8 44.3 28.5 16.8 27.2 46.6 36.6 20.0 31.2 46.6 37.4 21.2 32.0 48.2 35.9 21.4 30.6 47.9 37.6 22.2 30.9 1.00 0.34 - 1.29 0.81 - 1.70 0.31 - 1.63 0.15 - 1.90 1.16 - 1.96 0.43 - 1.82 0.68 - 1.68 0.21 - that the method that gave the best recoveries is at the bottom.Methods 5(a)-(d) were grouped together as there were no significant differences among the recoveries ob- tained using these methods. The results obtained for several trace elements in the fusion samples were then arranged in order of analytical method in exactly the same manner and are shown in Table 3. It is evident from Table 3 that in general the analyte concentrations determined in the fusion samples have an apparent increasing trend from top to bottom with methods 5(a)-(d) giving similar results. This trend is true for all of the four samples from low mass 59C0 to high mass 208Pb in the same order as that predicted from Table 2.Various methods of analysis adopted (e.g. matrix tuning flow injection and internal standardization) compensated to various degrees for the matrix effects in the high matrix solutions. The spiked sample recoveries were improved so that the concentrations of trace elements in the samples increased according to corresponding methods of analysis. Hence for example the concentration of Co in sample 1 is most likely to be correct if the results obtained using methods 5(a)-(d) are chosen. Where there is no available internal standard then method 4 would be the best choice as it still yielded good recoveries. The concen- trations of 5 9 C ~ lo3Rh and 208Pb obtained by corresponding methods of analysis from samples 1-4 are also listed in Table 3.In samples 3 and 4 zo8Pb was not found i. e. zo8Pb was not present at a level that was within the detection capabilities of the ICP-MS instrument. Conclusions Various techniques such as matrix tuning flow injection and internal standardization were evaluated by a compari- son of spiked sample recoveries. Based on these results the method of choice for the determination of trace elements in unknown samples should always include internal standard- ization regardless of whether continuous nebulization flow injection standard or matrix tuning is employed. In the absence of a suitable internal standard the next best alternative was found to be continuous nebulization and ion lens tuning in the presence of the matrix. numbered ES 03221 and ES 04908.They also thank the National Institutes of Health Shared Instruments Grants Program for providing the VG PlasmaQuad through grant number S10 RR02714 and BP America for providing partial support. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 References Vaughan M.-A. and Horlick G. J. Anal. At. Spectrom. 1989 4 45. Date A. R. Cheung Y. Y. Stuart M. E. and Jin X.-H. J. Anal. At. Spectrom. 1988 3 653. Hall G. E. M. Park C. J. and Pelchat J. C. J. Anal. At. Spectrom. 1987 2 189. Hutton R. C. Bridenne M. Coffre E. Marot Y. and Simondet F. J. Anal At. Spectrom. 1990 5 463. Vandecasteele C. Nagels M. Vanhoe H. and Dams R. Anal. Chim. Acta 1988 211 91. Thompson J. J. and Houk R. S. Appl. Spectrosc. 1987 41 801. Beauchemin D. McLaren J. W. and Berman S. S. Spectro- chim. Acta Part B 1987 42 467. Garbarino J. R. and Taylor H. E. Anal. Chem. 1987 59 1568. McLaren J. W. Mykytiuk A. P. Willie S. N. and Berman S. S. Anal. Chem. 1985 57 2907. Beauchemin D. McLaren J. W. Mykytiuk A. P. and Berman S . S. Anal. Chem. 1987 59 778. Janghorbani M. and Ting B. T. Anal. Chem. 1989,61 701. Burguera M. and Burguera J. L. Analyst 1986 111 171. Thompson J. J. and Houk R. S. Anal. Chem. 1986,58,2541. Heitkemper D. Creed J. and Caruso J. A. J. Anal. At. Spectrom. 1989 4 279. Beauchemin D. Siu K. W. M. McLaren J. W. and Berman S. S. J. Anal. At. Spectrom. 1989 4 285. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988 3 547. Dean J. R. Ebdon L. Crews H. M. and Massey R. C. J. Anal. At. Spectrom. 1988 3 349. Vickers G. H. Ross B. S. and Hieftje G. M. Appl. Spectrosc. 1989,43 1330. Vaughan M.-A. Horlick G. and Tan S. H. J. Anal. At. Spectrom. 1987 2 765. Wang J. Shen W.-L. Sheppard B. S. Evans E. H. Caruso J. A. and Fricke F. L. J. Anal. At. Spectrom. 1990 5 445. Paper 1/02262H Received May 13th I991 Accepted July 30th 1991 The authors acknowledge the National Institute of Environ- mental Health Sciences for grant support through grants