JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 445 Quantitative Analysis of Trace Elements in Carbonates Using Laser Ablation Inductively Coupled Plasma Mass Spectrometry William T. Perkins Ronald Fuge and Nicholas J. G. Pearce Geochemistry and Hydrology Research Group institute of Earth Studies University College of Wales Aberystwyth UK Laser ablation inductively coupled plasma mass spectrometry has been applied to the analysis of carbonate materials. Multi-element synthetic standards prepared both as pressed powders and fused glass discs were used for calibration. The elements Mg Mn Sr Ba and Pb were added to the pressed powder standards and these elements together with U were added to the fused glass standards. Calibration graphs which are linear over at least three orders of magnitude were produced using both types of standard but the fused glass discs gave better precision.The accuracy of the technique was evaluated using reference materials. Acceptable values were obtained using the pressed powders [e.g. BCS 393 (limestone) certified values of 905 ppm (Mg) 77 ppm (Mn) and 160 ppm (Sr)] but better accuracy was achieved with fused glass discs [e.g. BCS 393; 957 ppm (Mg) 79.6 ppm (Mn) and 167 ppm (Sr)]. The technique is applied to the analysis of carbonate shell material and demonstrates its potential in environmental monitoring. Keywords Laser ablation inductively coupled plasma mass spectrometry; quantitative analysis; carbonates; trace element; environmental monitoring The use of laser ablation (LA) was first proposed in the early 1960s following the publication of work on laser action in ruby.The subsequent development of laser microanalysis has been reviewed by Moenke-Blankenburg.2 The tech- nique which has been applied to atomic emission spectro- metry (AES) and inductively coupled plasma (ICP) AES,2-s was first coupled with ICP mass spectrometry (ICP-MS) by Gray,6 who used a JK Type 2000 ruby laser and demon- strated the applicability of this technique to geological materials both for trace element determination and isotope ratio measurements. Arrowsmith7 reported the use of a Nd:YAG laser with ICP-MS for the analysis of microprobe reference materials and Cu standards More recently van Heuzen* described procedures for quantitative analysis using both fused glass and pressed powder materials.In this latter study the importance of matrix matching of samples and standards was emphasized. The laser ablation system has the potential to perform spatial analysis with the laser spot size being to a certain extent controlled by the power output of the laser; typical craters are about 100 pm in diameter. The aim of this work was to demonstrate the capability of LA-ICP-MS as a quantitative analytical tool for the analysis of trace components in carbonate materials covering a compositional range of CaC03-MgC03. A comparison was made between pressed powder samples and fused glass discs in terms of accuracy and precision. The merits of different methods of internal standardization are discussed. Experimental Instrumentation A VG Instruments PQII+ ICP mass spectrometer and VG LaserLab were used during this work.The VG LaserLab is based on a Spectron Laser Systems 500 mJ Nd:YAG laser operating at 1064 nm. This may be run in either fixed Q- or Q-switched mode. The output of the laser at 10 Hz is rated at 500 mJ in fixed Q- and 250 mJ in the Q-switched mode. It is however possible to run the laser at repetition rates greater than 10 Hz and in some applications a repetition rate of 15 Hz was employed. A standard VG sample chamber was used throughout this study. The ablated material was transferred from the LaserLab to the PQ1I-t using 2 m of 4 mm i.d. poly(viny1 chloride) (PVC) tubing. This configuration is similar to that used by Gray.6 The general operating conditions of the ICP mass spectrometer are given in Table 1.Pressed Powder Standards The first attempt at standardization was based on pressed powder discs. Specpure (Johnson Matthey) carbonates or oxides of the desired constituents were accurately weighed and mixed with a known mass (1 0 g) of the CaC03 matrix to produce a series of standards. Specpure In203 powder was also added as an internal standard. The powder mixture was transferred into glass jars which contained synthetic leucite [K(AlSi,O,)] balls. The jars were closed and the mixtures shaken in a laboratory mixer-mill for 5 min. The powder was then mixed with a 10% m/v poly(viny1 alcohol) binder and pressed at 25 tons into 30 mm diameter discs. As this work was concerned with the analysis of shell material which contains a significant proportion of organic matte^,^,^^ it was decided to use such material as the basis for a set of pressed powder standards.It was considered that using such material would overcome any ablation effects in the shells caused by the presence of the organic matter. For this reason the first series of standards produced were based on crushed shell material and contained added components (Mn Sr Ba and Pb) up to 10 000 p g g - l . The ablation of each standard was performed in fixed Q- mode with a focused laser. The system was set at 750 V giving a laser output of approximately 200 mJ. A 2 x 10 raster pattern with single laser shots at each position and a repetition rate of 1.33 Hz was used. A repetition rate of more than 1 Hz should produce an almost constant signal Table 1 Operating conditions for ICP-MS Forward power Gas flow rate Cool gas Auxiliary Carrier gas Reflected power No.of sweeps Mass range Dwell time 1250 W 12.75 1 min-l 0.5 1 min-' 1.00 1 min-l ow 400 22-246 rnlz 160 p s Institute of Earth Studies No. 161.446 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 I VOL. 6 (cf. Gray6). In this configuration the amount of material removed by each set of 20 shots averaged 720 pg with a standard deviation of 83 taken over five determinations. Each standard was analysed five times and the raw data were collected using a peak .width of 0.7 u with valley integration. The data were reported as counts per second for the given area [normally quoted as area counts per second (ACPS)]. The data were processed using Microsoft Excel which is a spreadsheet package. Calibration graphs for Mn Sr Ba and Pb are presented in Fig.1. These graphs were produced using In as an internal standard. In general linear calibration graphs were pro- duced over this wide concentration range although the degree of scatter shown by the 20 error bars is consider- able. Careful observation of the pressed discs revealed variation in the grain size of the carbonate matrix despite every effort to ensure homogeneity. This variation could account for much of the scatter observed in the calibration graphs. In an attempt to overcome this problem a second series of pressed powder standards were produced. The second series of standards were prepared using AnalaR CaCO (Merck) as the matrix material. This has the advantage over the shell matrix of being fine-grained and homogeneous.These standards were analysed in the focused fixed Q-mode and in a further attempt to over- come small scale inhomogeneity in the de-focused Q- switched mode. The standards were produced with a range of concentrations up to 5000 pg g-l. As with the first series of standards this set was analysed five times and the data collected as ACPS. The data were processed using both the added In internal standard and 44Ca as a true internal standard. Calibration graphs were produced using a least- squares regression and the correlation coefficients for Mn Sr Ba and Pb are presented in Table 2. The data produced good correlations by both methods of internal standardiza- tion although the 44Ca isotope produced marginally better data overall.This might again relate to the homogeneity of the standard since the In,03 is added as a powder and there is a problem when trying to mix thoroughly small additions of powders prior to pressing. As a result of this test subsequent analyses made use of the 44Ca isotope for internal standardization. E 12 ;; (c) E" 1 0 - v) al * C * 6 - 0 ru 4 - Y 2 - - 5 8 - L cc E al a O A 0 2500 5000 7500 10000 Concentration of element added/pg g-' Fig. 1 Calibration graphs for the first series of pressed powder standards using crushed shell material as the base ( a ) 55Mn; (b) %r; (c) 137Ba; and (d) 207Pb. Indium is used as an internal standard. Plots show the best-fit linear regression line with ? 20 error bars Table 2 Correlation coefficients for pressed powder CaC0,-based standards using both fixed Q- and de-focused Q-switched lasers with In and 44Ca as internal standards Correlation coefficient Fixed Q-mode Q-switched mode Element In 44Ca In 44Ca 0.9805 Mn 0.9355 0.9851 0.9509 0.989 1 Sr 0.9963 0.9763 0.9982 Ba 0.9355 0.9809 0.9906 0.9999 Pb 0.9726 0.9123 0.9473 0.9799 Fused Glass Standards The work on pressed powder standards suggests that there are problems in trying to produce homogeneous mixtures of powders If the mixtures were fused into a glass disc such as those commonly produced for X-ray fluorescence analysis then a solid solution should result and the standard would then be homogeneous.In this study three matrix compositions were chosen and a series of standards produced for each.The three matrix compositions were CaCO,; (Mg,Ca)CO,; and MgC03 (calcite/aragonite; dolomite; and magnesite respectively). AnalaR CaCO and Specpure MgO were used to give the correct Ca and Mg concentrations. These formed the matrix for the CaC0 and MgC0 standards respectively. The (Mg,Ca)CO standard was produced using a mixture of these two in order to produce a Ca Mg ratio close to that of natural d01omite.l~ The flux used was lithium tetraborate LiZB407 (Johnson Matthey Spectroflux loo) with a flux to sample ratio of 5 1 for the calcite and dolomite and 10 1 for the magnesite (2.5 g Li2B407:G.5 g matrix and 5.0 g Li2B407:0.5 g matrix respectively). Solutions of trace elements Mg Mn Sr Ba Pb and U were added in small volumes (1 000 pg ml-I or diluted Aldrich standard solu- t ions) using a gravimetrically calibrated Labsystems elec- tronic Finnpipette.The matrix flux and additions were mixed in a Pt-Au crucible and dried at 110 "C. The Fig. 2 Scanning electron microscope image of a typical ablation crater produced by ablation of a fused glass disc. The central crater is approximately 250 pm in diameter and is surrounded by a raised area 500 pm in diameter. The outer zone shows deposits of material ejected from the ablation crater this zone is approxi- - - '4 mately 1.5 mm in diameterJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 447 mixtures were then fused for 30 min over a Meker burner. The resulting melts were allowed to cool in the crucibles and following a period of annealing were removed and stored prior to analysis.The system must be operated in the focused Q-switched mode in order for the laser to couple with a glass. It was found that the efficiency of ablation was much poorer than with the pressed powder materials and it was necessary to use a rapid repetition rate (15 Hz) in order to achieve a usable signal. Because of the homogeneity of the glass material it was not necessary to raster the sample surface and the data were collected from a single spot. The laser power was set at approximately 125 mJ. The sample surface was subjected to pre-ablation for a period of 20 s before the spectra were obtained. The craters produced during Q- switched laser ablation of these glasses were examined using a scanning electron microscope (SEM) as shown in Fig.2. The SEM image shows a crater of approximately 250 pm diameter similar to those reported by van Heuzen.* The ablated hole is surrounded by a raised area of smooth material extending 250 pm beyond the crater. This area is interpreted as being either melted during the ablation period or deposited by sputter of molten glass from the crater. Outside the smooth zone is a region characterized by fragmented material extending over a diameter of more than 1.5 mm. This is almost certainly produced by the deposition of both fragmented and molten material. Results Reference Materials Three reference materials [Geological Survey of Japan (GSJ) JLS-1 Limestone Institute of Geophysical and Geochemical Prospection (IGGE) Ministry of Geology China GSR-6 Limestone and British Chemical Standards (BCS) Certified Reference Material (CRM) 393 Limestone] were analysed as pressed powders against a calibration graph produced from the CaCO blank and a 500 pg g-l multi-element standard.The results are tabulated in Table 3. There is general agreement between the values deter- mined in this work and the recommended and proposed values although the accuracy varies from 8% at best to 102% at worst. If this inaccuracy results from the inhomo- Table 3 Comparison of LA-ICP-MS data with recommended or proposed values for certified reference materials; calibration and analysis using pressed powders Sample BCS 393 (Limestone)- Mg* Mn* Sr* Ba Mg Mn Sr GSJ JLS- 1 (Limestone)- IGGE GSR-6 (Limestone)- m * Mn* Sr* Ba* Pb* Recommended/ LA-ICP-MS proposed (PPm) (PPm) 905 77 160 53 3 739 15 296 31 298 46 5 913 120 18.3 1043/730? 24/44? 1 1 1/123? 49/82? 5 472 11 26 1 51 146 394 604 191 37 * Recommended value all others are proposed values. t Duplicate determination.geneous distribution of the small amount of powder added to the matrix then it was considered that fusing the mixture into a glass disc would be one way to overcome the problem. However for a comparative study of the distribu- tion of elements in carbonate shell material these pressed powder standards were adequate and in any event represent an improvement on the qualitative data previ- ously attained. Shell Walls Two specimens of Arctica islandica a shallow-burrowing marine bivalve living in sands and muds were examined as a test of the calibration graphs produced.One specimen was collected from the beach at Borth just north of Aber- ystwyth. The sample was a single valve washed up after a winter storm and the age of the specimen was unknown. The second sample set were collected as live specimens by the Scottish Universities Marine Biological Station Mill- port Isle of Cumbrai. Analyses of the trace element composition from the inner to the outer shell were taken from two of the Scottish specimens and compared with the profile of the Borth specimen. The results for typical distributions of Mg Sr and Pb are presented in Fig. 3. The shells show parallel trends for Mg and Sr although the abundance of the trace elements is different between the two sample localities. This variation is a reflection of the Sr:Ca and Mg:Ca ratios since 44Ca is used as the internal standard.These ratios have been shown to be dependent on salinity and temperature at the time the organism secreted the shell although the ratios may be subject to change following death and burial of the shell material.'' The Pb levels in the specimen from Borth are elevated when compared with the Scottish samples. The rivers draining into Cardigan Bay are known to be polluted with Pb from the Pb mining activities in the area during the last century. This pollution is reflected in the distribution of heavy metals in Cardigan Bay.12 It seems probable therefore that Pb contamination of the marine environment is reflected in 400 300 200 - '0 100 0 0 $ 3000 f - 0 3 2000 C 0 .- 5 1000 C C I - E 0' I I I I 0 0 1 2 3 4 Distance/m m Fig.3 Profiles across three specimens of Arctica isfandica from the inner to the outer shell margin for the elements (a) Mg (b) Sr and (c) Pb. The sample from Borth (A) shows significantly higher Pb levels than the Millport samples (B and C)448 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1991 VOL. 6 the chemistry of the shells of marine organisms. This is the subject of further research at Aberystwyth. Fused Glass Standards Glass discs of six reference materials (GSJ JLS-1 Lime- stone JDo-1 Dolomite; GSR-6; BCS CRM 393 Limestone 368 Dolomite and 389 High Purity Magnesite) were also produced. Four analyses were obtained from different points on the disc surfaces and the data collected as ACPS. These data were transferred to a spreadsheet and the calibration graphs and calculations performed off-line.Minor isotopes of the major elements were used as internal standards (44Ca for the calcite and dolomite and 24Mg for the dolomite and magnesite). By using either Ca or Mg as the internal standard the effects of volatile loss can be overcome since the trace element to major element ratio will remain constant. The correlation coefficients for the calibration graphs generated by least-squares linear regres- sion are presented in Table 4. Good linear calibration graphs (i. e. with a least-squares regression correlation coefficient of better than 0.95) are produced using the glass discs although the slope for a given element varies in the three matrices because of the different concentrations of the major element internal standard present.When the true concentration of the internal standard is used (z.e. ACPS divided by the mass fraction of the element present) the calibration graphs give similar slopes. The slope values obtained after this correlation are presented in Table 5. There is no apparent matrix effect throughout the range of compositions analysed although this might in part be a function of the dilution by the flux (5:l for the calcite and dolomite 10 1 for the magnesite). The fused glass standards produced from the reference materials were analysed using the synthetic standards to calibrate the instrument. The values obtained during this study are presented in Table 6. These data are in agreement with the recommended values the exceptions being the Sr data for GSR-6 after correction for the CaC0 concentration and BCS CRM 368.It is not obvious why these data are poor when the other Sr values are in good agreement. In general however the data give an accuracy of better than k 10% and in many instances better than ?5%. This is excellent especially when the range of values concerned is from 15 ppm of Mn in JLS-1 to 3 I 298 ppm of Mg in GSR-6. These data are similar to the values reported by van Heuzens although this work deals with a carbonate rather than a silicate matrix. Comparison of Calibration Methods This work has demonstrated that linear calibration graphs can be produced using pressed powder standards based both on crushed shell material and AnalaR CaCO,. Calibra- tions based on crushed shell materials gave relatively large errors (Fig.1) which have been attributed to a combination of inhomogeneous grain size in the matrix and an uneven distribution of the powder additions. When these standards are compared with a series of standards based on AnalaR CaC0 there is no significant improvement in the precision for Sr [Fig. 4(a)-(c)]. However when the calibration data obtained using added In as an internal standard are compared with the data obtained using a minor Ca isotope as internal standard there is an improvement in the precision of the results. This supports the conclusion that the problem lies with the mixing of powder additions in a powder matrix. When the laser is de-focused in the Q- switched mode so that a larger area of the specimen is ablated the calibration graphs for the shell- and CaCO based standards are almost parallel the offset being caused by the higher level of Sr in the shell material.The level of precision is comparable to that obtained in the focused fixed Q-mode despite the larger area sampled. These data illustrate a fundamental problem in producing standards as mixtures of powders which are homogeneous. In Fig. 4(d) a calibration graph for Sr in the fused glass standards is presented. This demonstrates the improvement in precision gained when fused glass standards (true solid solutions) are used to calibrate the instrument. Table 4 Table of correlation coefficients for the fused glass standards internal standards in parentheses Correlation coefficient Limestone Dolomite Dolomite Magnesite (“Ca) (44Ca) (24Mg) (24Mg) Element - - - Mg 0.9980 Mn 0.9999 0.9992 0.9996 0.9854 Sr 0.9997 0.99 13 0.9744 0.98 14 Ba 0.9988 0.993 1 0.9860 0.9807 Pb 0.9974 0.9874 0.984 1 0.9467 U 0.9993 0.9969 0.9945 0.9997 Table 5 Values for the slope of least-squares regression analysis of the fused glass standards following correction for the mass fraction of the major element internal standard; internal standards in parentheses Slopelpg - Limestone Dolomite Dolomite Magnesite (“Ca) (“Ca) (24Mg) (24Mg) Element Mn 1 .4 2 6 ~ 1 . 6 0 8 ~ 1 . 4 7 6 ~ 1 . 4 3 4 ~ Sr 3.599 x 10-6 3.437 x 2.929 x 2.875 x Ba 3 . 4 2 7 ~ 2 . 9 1 3 ~ 2 . 8 1 6 ~ lop7 4.781 x low7 Pb 3 . 5 8 4 ~ 4.931 x 4 . 5 3 4 ~ 5.081 x U 8.948 x 9.144 x 8.460 x lo-’ 7.252 xJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL.6 449 '/ ~~ ~ _ _ _ _ ~ ~ _ _ _ _ Table 6 Comparison of LA-ICP-MS data with recommended or proposed values for certified reference materials; calibration and analysis using fused glass discs Sample BCS CRM 393 (Limestone)- Mg* Mn* Sr* Ba Mg Mn Sr GSJ JLS- 1 (Limestone)- IGGE GSR-6 (Limestone)- Mg* Mn* Sr* Ba* IGGE GSR-6t- Mg* Mn* Sr* Ba* Mn* Sr Pb Mn Sr Mn* BCS-CRIM 368 (Dolomite)- GSJ JDO- 1 (Dolomite)- BCS-CRM 389 (Magnesite)- Recommended proposed LA-ICP-MS (PPm) (PPm) 905 957 160 167 77 79.6 53 66.5 3 739 3 867 296 324 15 19.6 31 298 39 598 465 690 913 849 120 192 31 298 34417 465 439 913 540 120 122 465 39 1 67 117 61 59.8 46 44. I 119 138 62 51 *Recommended value all others are proposed values. 7 IGGE GSR-6 is a slightly dolomitic limestone (MgO = 5.19%) and the values for this standard were recalculated using the recommended value for Ca.2.5 2.0 1.5 C - g 1.0 0.5 0 1 .o 0.8 8 0.6 ,,f) 0.4 0.2 0 s L 1.2 0.8 0.4 0 0 1000 2000 300( 0.03 0.02 0.01 0 1 0 1000 2000 3000 Additions of Sr/pg g-' Fig. 4 Comparison of calibration graphs for Sr produced using different standards and ablation conditions. Solid lines represent best-fit linear regression lines for the AnalaR CaCO based standards and broken lines join the blank and 2500 pg g-I crushed shell-based standards. Graphs illustrate (a) curves obtained in focused fixed Q-mode using added llsIn as the internal standard; (b) curves obtained in focused fixed Q-mode using 44Ca as the internal standard; (c) curves obtained in de-focused Q-switch mode using 44Ca as internal standard; and (d) calibration graph for the fused glass standards using 44Ca as internal standard Conclusion Laser ablation ICP-MS is capable of producing quantitative analytical data for a carbonate matrix providing the standards are matrix matched.Pressed powder standards give reasonable calibration graphs when a true internal standard is used but there are problems in producing homogeneous standards by the mixing of powders. One of the advantages of the laser system when com- pared with sample dissolution is the ability to obtain spatial information about element distribution. This aspect has been demonstrated for the bivalve Arctica islandica and illustrates the potential of the technique as a pol- lution monitor this being an area of active research at Aberystwyth.The production of fused glass standards has the potential to obtain accurate determinations of trace constituents in carbonate materials. Given the dilution of the flux to sample ratio used there is no apparent matrix effect for the range of compositions studied in this work. This is in agreement with the work of van Heuzen8 although the accuracy demonstrated here is somewhat better than that quoted by van Heuzen. Since the fused glass discs are solid solutions there is no requirement to raster the surface in order to achieve analytical precision. Thus only a small 1 2 3 4 5 6 7 8 9 10 11 12 13 portion of the surface needs to be used in an analysis and these durable discs can be stored for repeated use. References Maiman T. H. Nature (London) 1960 187 493. Moenke-Blankenburg L. Laser Microanalysis Chemical Ana- lysis Wiley New York 1989 vol. 105. Thompson M. Goulter J. E. and Sieper F. Analyst 1981 106 32. Carr J. W. and Horlick G. Spectrochim. Acta Part B 1982 37 1. Ishizuka T. and Uwamino Y. Spectrochim. Acta Part B 1983 38 51 9. Gray A. L. Analyst 1985 110 551. Arrowsmith P. Anal. Chem. 1987 59 1437. van Heuzen A. A. Ph.D. Thesis University of Amsterdam 1990. Taylor J. D. Kennedy W. J. and Hall A. Bull. Brit. Mus. (Nut. Hist.) Zool. 1969 Suppl. 3. Taylor J. D. Kennedy W. J. and Hall A. Bull. Brit. Mus. (Nat. Hist.) Zool. 1 973 22 253. Brand U. and Morrison J. O. Geosci. Can. 1987 4 85. Abdullah M. I. Royle L. G. and Morris A. W. Nature (London) 1972 253 158. Deer W. A. Howie R. A. and Zussman J. An Introduction to the Rock-forming Minerals Longman London 1975 part 5 p. 474. Paper I /OO 7098 Received February 14th 1991 Accepted April 25th 1991