Automated in situ trace element analysis of silicate materials by laser ablation inductively coupled plasma mass spectrometry

Transcription

1 Fresenius J Anal Chem (2000) 368 : Springer-Verlag 2000 SPECIAL ISSUE PAPER Zhongxing Chen Dante Canil Henry P. Longerich Automated in situ trace element analysis of silicate materials by laser ablation inductively coupled plasma mass spectrometry Received: 1 March 2000 / Revised: 14 June 2000 / Accepted: 16 June 2000 Abstract This paper describes the automated in situ trace element analysis of solid materials by laser ablation (LA) inductively coupled plasma mass spectrometry (ICP-MS). A compact computer-controlled solid state Nd:YAG Merchantek TM EO UV laser ablation (LA) system has been coupled with the high sensitivity VG PQII S ICP-MS. A two-directional communication was interfaced in-house between the ICP-MS and the LA via serial RS-232 port. Each LA-ICP-MS analysis at a defined point includes a 60 s pre-ablation delay, a 60 s ablation, and a 90 s flush delay. The execution of each defined time setting by LA was corresponding to the ICP-MS data acquisition allowing samples to be run in automated cycle sequences like solution auto-sampler ICP-MS analysis. Each analytical cycle consists of four standards, one control reference material, and 15 samples, and requires about 70 min. Data produced by Time Resolved Analysis (TRA) from ICP-MS were later reduced off-line by in-house written software. Twenty-two trace elements from four reference materials (NIST SRM 613, and fused glass chips of BCR-2, SY-4, and G-2) were determined by the automated LA-ICP-MS method. NIST SRM 610 or NIST SRM 613 was used as an external calibration standard, and Ca as an internal standard to correct for drift, differences in transport efficiency and sampling yield. Except for Zr and Hf in G-2, relative standard deviations for all other elements Zhongxing Chen ( ) D. Canil School of Earth and Ocean Sciences, University of Victoria, PO Box 3055, Victoria, BC Canada V8W 3P6 Zhongxing.Chen@usm.edu H. P. Longerich Department of Earth Sciences and Centre for Earth Resources Research, Memorial University of Newfoundland, St. John s, Newfoundland, Canada A1B 3X5 Present address: Zhongxing Chen, Department of Marine Science, University of Southern Mississippi, Stennis Space Center, MS 39529, USA are less than 10%. Results compare well with the data reported from literature with average limits of detection from 1 ng g 1 to 455 ng g 1 and less than 100 ng g 1 for most trace elements. Introduction Laser ablation (LA) inductively coupled plasma inductively coupled plasma mass spectrometry (ICP-MS) has been widely used as a powerful analytical technique for solid micro sampling analyses in geological [1 10], biological [11, 12], environmental [13 15], nuclear [16 18], and metallurgical [19] applications. It is a rapid and moderate cost technique for analytical chemists. The technique combines the advantages of the high sensitivity multi-element capability by ICP-MS and in situ micro solid sampling by LA. It is particularly attractive to scientists who want to study dissolution-resistant solid materials or study spatial distributions of trace elements in a micro scale area on the sample surface. However, due to complexity of data acquisition and reduction of transit signals, the analysis so far has been limited to manual operation requiring an operator to be present at all times, which is time consuming. In this paper, a method was described for the automated in situ trace element analysis of solid materials by LA-ICP-MS. The automated method was demonstrated by determination of 22 trace elements from reference materials (NIST SRM 613, and fused glass chips of BCR-2, SY-4, and G-2). Experimental Instrumentation A VG Elemental PQ II S high sensitivity ICP-MS (Winsford, Cheshire, England) was used. The instrument was described previously [10]. The software has been upgraded to the OS/2 based PQ version A two-directional communication was interfaced inhouse between the ICP-MS and the LA via serial RS-232 (i.e., COM port). With the new PQ software, the execution of each userdefined time setting by LA was synchronized with the ICP-MS

2 74 Fig. 1 Signal versus time, and time settings of LA and ICP- MS for each automated LA- ICP-MS analysis. Each analysis includes a 60 s pre-ablation delay for background, a 60 s ablation for sample signal, and a 90 s sample delay for flush. Data were acquired for 120 s by ICP-MS using Time Resolved Analysis (TRA) data acquisition (Fig. 1). Similar to a non-automated LA-ICP-MS, optimization of the plasma and mass spectrometer conditions in the automated LA-ICP-MS was accomplished using NIST SRM 613 glass nominally containing approximately 50 µg g 1 of trace elements. For each analytical run, the nebulizer gas flow rate, torch position and lens setting were adjusted to optimize signal intensity while ablating NIST SRM 613 with a spot size of approximate 50 µm and laser beam energy of less than 3 mj. Typical ICP-MS operating conditions for the automated LA-ICP-MS was given in Table 1. The LA used in this work was a Merchantek EO UV laser system (Carlsbad, California, USA). It is a compact computer-controlled solid state Nd:YAG laser whose output was frequency quadrupled to 266 nm with maximum energy of 4 mj. The LA software has been upgraded to UV. Additional pre-ablation delay feature required by the authors was added by Merchantek to configure the individual analysis shown in Fig. 1. The LA software was run in remote mode controlled by the PQ ICP-MS software during the automated LA-ICP-MS analysis. The laser was operated in a gated Q-Switch mode for optimum stability, i.e., the flash lamp was fired at a frequency of 20 Hz and laser output at computer-controlled pulse rates between 1 and 20 Hz. A beamsplitter directs approximately 2% of the laser beam to a built-in energy meter and approximately the other 98% of the beam to the objective lens. This arrangement allows an operator to continuously monitor the laser power before the final objective lens during ablation. Standards and samples NIST SRM 610 or NIST SRM 613 synthetic silicate reference material was used as a calibration standard. The NIST SRM glass button was analyzed as received. Major and trace element concentrations of the two standards, determined by solution nebulization ICP-OES and ICP-MS, and LA-ICP-MS, were given in Chen et al. [4] and Pearce et al. [20]. Three geological reference materials: Canada Centre for Energy and Mineral Technology (CANMET) SY-4 ( Diorite Gneiss ), and United State Geological Survey (USGS) reference materials BCR-2 (Basalt) and G-2 (Granite) were analyzed by the automated LA-ICP-MS. Approximately 4 to 5 g of rock powder was weighed into a 20 ml Pt crucible, and melted and homogenized in a muffle furnace (Lindberg Blue M) at 1550 C in an air atmosphere for 12 to 15 h. No Fe loss occurs from the sample to the Pt crucible in an Table 1 LA-ICP-MS operating conditions Inductively coupled plasma Plasma gas argon Forward power (W) 1350 Reflected power(w) < 5 Gas flow Plasma gas flow rate (L min 1 ) 14 Auxiliary gas flow rate (L min 1 ) 0.94 Inner gas flow rate (L min 1 ) ~ 1.20 (see text) Interface Sampling distance (load coil to 16 sample aperature (mm) Sampling aperature nickel, 1.0 mm diameter Skimmer aperature nickel, 0.7 mm diameter Ion lense settings Extraction lens (V) 365 Collector lens (V) 77.9 L1 lens (V) +2.8 L2 lens (V) 10.7 L3 lens (V) +4.8 L4 lens (V) 29.9 Pole Bias (V) 0 Data acquisition parameters Measurement mode peak jumping Dwell time (ms) Data acquisition time (s) 110 Points per peak 1 LA operating parameters Laser mode Q-switch Wave length 266 nm Flash lamp frequency 20 Hz Laser output frequency 10 Hz Laser energy before objective lens < 3 mj Spot size 3 (~ 50 µm)

3 air atmosphere [21]. The crucible was removed from the furnace and the melt was quickly quenched to a glass in water. Cm-sized chips of the glass were removed from the crucible, mounted in epoxy and polished with SiC powder under distilled water. The resultant glass chips were examined using a petrographic microscope and were found to be free of any crystal phases. Data acquisition All the analyses were carried out in single spots with a spot size of approximate 50 µm. Sample spots were selected under the microscope and their positions (X, Y, and Z) stored in the computer in a sequences order before the analyses. After both instruments were optimized and a analytical sequence was set up, all analyses were performed automatically following the analytical sequence from the first to last analysis without operator assistance. Data acquisition was modified from Longerich et al. [2]. Elements and their mass (m/z) selected for analysis are listed in Table 2. Samples were run in the following automated cycle sequences: standard, standard, QC reference material, sample 1,, sample 15, standard, standard, QC sample, sample 16,, sample 30, standard, standard etc. Each analysis consisted of a 60 s pre ablation delay for the background (measured), a 60 s lasering for signal (measured), and a 90 s flush delay with no measurement (Fig. 1). The entire cycle of 20 analyses required 70 min. Data were acquired in the peak-jumping mode with a dwell time of ms. Background levels for each element were obtained by acquiring data for a gas blank for 60 s prior to laser sampling. Count rates were measured and exported to comma-delimited ASCII files by PQ Version 4.36 Time Resolved Analysis (TRA) software. Data reduction and calibration All subsequent data manipulations were accomplished off-line using in-house written software. Data collected from each analytical cycle were reduced individually by the following procedure. Step 1: import data from the ASCII files to a spreadsheet. Step 2: generate signal vs. time graphs. Step 3: select time intervals for the average background count rates and average signal count rates (Fig.1). Step 4: input concentrations of internal standard and calculate sample concentrations. Step 5: check quality control (including background, sensitivity, drift, limits of detection, precision and accuracy). Step 6: save result files. Step 7: go to the next cycle or exit data reduction program. Calibration and quantification of the analysis utilize both external and internal calibrations. Drift, matrix effects, changes in laser sampling yield, and sample transport efficiency during the analysis were corrected for by using naturally occurring internal standards. Therefore, the accurate known concentration of at least one analytical suitable major element (e.g., Ca) homogeneous in the sample is required. The criteria for selecting internal standards have been discussed by Fryer et al. [22]. Fractionation of elements in laser ablation sampling and transport is an important parameter in the internal standard selection1 [10]. An element whose fractionation is similar to those of analytes to be measured was chosen as internal standard. In this study, Ca was used as an internal standard. Results and discussion To demonstrate the precision and accuracy of automated LA-ICP-MS analysis, homogeneous materials are required due to the high spatial resolution micro scale sampling. NIST SRM 610 and NIST SRM 613 are very well documented synthetic silicate reference materials and are sufficiently homogeneous to be used in many microanalytical applications [4, 20, 23]. In the initial work, NIST SRM 610 was used as an external calibration standard 75 and Ca as an internal standard. Sixteen replicates were analyzed in a cycle for over a period of 70 min. Results of this experiment are presented in Table 2. The RSDs were less than 5% for all the trace elements measured. Accuracy expressed as the relative difference between the LA-ICP-MS analyses and accepted literature values were less than 4% for most trace elements and 12.3%, 7.4%, 11.7% and 6.5% for V, Zr, Nb, and Gd respectively (Fig. 2). To further demonstrate the method, fused glass chips of three geological reference materials (SY-4, BCR-2, and G-2) were analyzed by the automated LA-ICP-MS. NIST SRM 613 was used as an external calibration standard and Ca as an internal standard. The results are listed in Table 2. Twenty-two elements (Table 2) in four geological reference materials were determined by the automated LA-ICP-MS. Gallium and Ge were included in the initial automated LA-ICP-MS analysis. However, due to much significantly high fractionation of these two elements relative to Ca [4], it is unsuitable to use Ca as an internal standard to calibrate these two elements. Gallium and Ge were not commented on further. Except for Zr and Hf, all elements in the three reference materials have RSDs less than 10%. RSDs of Zr and Hf in G-2 are 33.7% and 30.1% respectively (Table 2). The high RSDs of Zr and Hf in G-2 are due to the sample heterogeneity since the two elements are concentrated in refractory zircon which is very enriched in G 2 (granite) and not homogeneously distributed in the samples. The relative difference between the automated LA-ICP-MS analyses and the accepted literature values is generally less than 20% (Fig. 2). Mean chondrite-normalized data are shown in Fig. 3. The low values of Y and Tm in the literature are suspect as Y and Tm are mono-isotopic elements and there is a lack of isotope dilution mass spectrometry data for the two elements. Y data from X-ray fluorescence spectrometry are significantly interfered by Rb [26]. The limit of detection (LOD) for the LA-ICP-MS analyses which was well documented by Longerich et al. [2] is defined as three times the standard deviation of the Ar gas blank: LOD = 3σ (1/N a + 1/N b ) Where σ is the standard deviation, and N a and N b are the number of slices in the ablation time that are integrated for background and signal intensity, respectively. Since the amount of material ablated and transported in LA-ICP-MS was different from one sample to another, LOD for each sample analysis was calculated individually. Spot size and laser power are two important parameters affecting the limits of detection. The average LODs from each individual analysis of the four reference materials are shown in Table 2 for a spot size of 50 µm and laser energy of less than 3 mj per pulse. The LOD ranges from 1 ng g 1 to 455 ng g 1, and for most of the trace elements the LOD is less than 100 ng g 1. The advantages of the automated LA-ICP-MS analysis was demonstrated in the method. Compared to the non-automated LA-ICP-MS analysis, the automated LA-ICP-MS

4 76 Table 2 Trace element composition of synthetic silicate glass NIST 613, and geological reference materials BCR-2, SY-4, and G-2 determined by automated LA-ICP-MS Element V Sr Y Zr Nb La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Hf Th U Measured mass (u) NIST 613 Average (n = 16) (µg/g) RSD (n = 16) 3.8% 1.1% 2.2% 2.5% 1.2% 1.9% 1.9% 1.9% 2.1% 2.4% 2.1% 2.5% 2.8% 2.8% 2.6% 2.8% 2.5% 2.7% 2.6% 2.1% 2.3% 4.4% Literature value [4] (µg/g) Average detection limit (ng/g) BCR-2 Average (n = 5) (µg/g) RSD (n = 5) 2.4% 7.2% 7.2% 6.1% 4.3% 4.4% 3.7% 4.5% 5.8% 5.5% 3.5% 3.6% 6.5% 4.8% 7.2% 7.8% 9.4% 7.5% 8.6% 6.5% 2.6% 4.2% Literature value [24] (µg/g) N/A N/A 1.33 N/A Average detection limit (ng/g) SY-4 Average (n = 6) (µg/g) RSD (n = 6) 4.9% 1.4% 1.8% 4.9% 2.6% 1.6% 3.0% 2.8% 1.7% 2.6% 4.2% 1.3% 3.0% 0.7% 1.4% 1.9% 1.4% 2.3% 2.0% 5.4% 5.5% 2.6% Literature value [25] (µg/g) Average detection limit (ng/g) G-2 Average (n= 4) (µg/g) RSD (n = 4) 3.4% 0.5% 2.7% 33.7% 2.5% 1.2% 1.2% 1.5% 1.2% 0.5% 0.9% 6.5% 2.6% 2.0% 5.1% 4.2% 9.5% 11.8% 11.9% 30.1% 2.2% 2.8% Literature value [26] (µg/g) Average detection limit (ng/g)

5 77 Fig. 2 Accuracy expressed as relative difference between results obtained in this work, using automated LA-ICP-MS, and literature values for NIST SRM 613, BCR-2, SY-4, and G-2 Fig. 3 Mean chondrite-normalized data for geological reference materials for BCR-2, SY- 4, and G-2. The literature value for Tm, indicated by an arrow, is suspect, BCR-2 (This work, n = 5); SY-4 (This work, n = 6); G-2 (This work, n = 4); BCR-2 (Ref [24]); SY-4 (Ref [25]); G-2 (Ref [26]) analysis reduced the time for the analyses of one sequence cycle (20 analyses) from approximately 120 to 70 min with the sample spot positions pre-selected. With the instrument full-automation for data acquisition, an operator was not necessary to be present during the analyses, which significantly saved the laboratory labor. Unlike non-automated analysis, all the analysis have the same time intervals for blank and signal peak acquisition, which significantly facilitate the off-line data reduction. However, unlike the non-automated LA-ICP-MS analysis, adding a sample during the automated LA-ICP-MS analysis which may sometimes be required for in situ microsampling analysis, cannot be accomplished without breaking an automated analytical sequence. Conclusions A method has been developed for the automated LA-ICP- MS analysis and subsequent off-line data reduction by inhouse written software. With the instrument automation, an operator is not necessary to be present during the analyses. The feasibility of the automated method was demonstrated by determination of 22 trace elements from several reference materials (NIST SRM 613, and fused glass chips of BCR-2, SY-4, and G-2). NIST SRM 610 or NIST SRM 613 was used as a calibration standard and Ca as an internal standard. Good precision and accuracy and low limits of detection were obtained.

6 78 Acknowledgements The PQ II S ICP-MS and Merchantek TM EO LA systems were purchased through an NSERC major installation grant (Grant # EQM ) to J.K.B. Bishop, D.C. Canil, and K.M. Gillis. The authors thank S. Jackson for his comments and acknowledge the technical assistance of J. Zhao of the Thermal Instrument Canada, S. Oltrogge and S. Pinkham of Merchantek TM EO and, P. Stow of Isomass Scientific. We thank two anonymous reviewers for helpful comments which improved the manuscript. References 1. Jackson SE, Longerich HP, Dunning GR, Fryer BJ (1992) Can Mineral 30: Longerich HP, Jackson SE, Gunther D (1996) J Anal At Spectrom 11: Moissette A, Shepherd TJ, Chenery SR (1996) J Anal At Spectrom 11: Chen Z, Doherty W, Gregoire DC (1997) J Anal At Spectrom 12: Gunther D, Audetat A, Frischknecht R, Heinrich C (1998) J Anal At Spectrom 13: Hoskin PWO, Wysoczanski R (1998) J Anal At Spectrom 13: Shibuya EK, Sarkis JES, Enzweiler J, Jorge APS, Figueiredo AM (1998) J Anal At Spectrom, 13: Jeffries T, Jackson SE, Longerich HP (1998) J Anal At Spectrom 13: Norman M, Griffin WL, Pearson NJ, Garcia MO, O Reilly SY (1998) J Anal At Spectrom 13: Chen Z (1999) J Anal At Spectrom 14: Wang S, Brown R, Gray DJ (1994) Appl Spectros 48: Evans RD, Outridge PM, Richner P (1994) J Anal At Spectrom 9: Price GD, Pearce NJ (1997) Marine Pollution Bulletin 34: Garbe-Schonberg CD, Reimann C, Pavlov VA (1997) Environ Geol 32: Tanaka S, Yasushi N, Sato N, Fukasawa T, Santosa SJ, Yamanaka K, Ootoshi T (1998) J Anal At Spectrom 13: Crain JS, Gallimore DL (1992) J Anal At Spectrom 7: Gastel M, Becker JS, Kuppers G, Dietze HJ (1997) Spectrochim Acta Part B 52: Leloup C, Marty P, Dallava D, Perdereau M (1997) J Anal At Spectrom 12: Raith A, Hutton RC, Abell ID, Crighton J (1995) J Anal At Spectrom 10: Pearce NJG, Perkins WT, Westgate JA, Gorton MP, Jackson SJ, Neal CR, Chenery SP (1997) Geostand Newsl 21: Canil D (1991) Earth Planetary Sci Lett 106: Fryer BJ, Jackson SE, Longerich HP (1995) Can Mineral 33: Kane S (1998) Geostand Newsl 22: Wilson SA (1998) USGS, Open file report 25. Bowman WS (1995) Geostand Newsl 19: Govindarraju K (1994) Geostand Newsl 18: Longerich HP (1995) X-Ray Spectrom 24: