Inter-element fractionation and correction in laser ablation inductively coupled plasma mass spectrometry Zhongxing Chen{ School of Earth and Ocean Sciences, University of Victoria, P.O. Box 3055, Victoria, BC, Canada V8W 3P6 Received 26th April 1999, Accepted 6th September 1999 Inter-element fractionation in laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) analysis is one of the major challenges for using the technique for in situ trace element determination and isotopic ratio measurement of geological, environmental and biological solid samples.Attempts have been made to reduce inter-element fractionation in LA-ICP-MS analysis. However, this fractionation cannot be eliminated. The mechanism of the fractionation in LA-ICP-MS analysis is not very well understood. This study investigated the inter-element fractionation of seven elements (Ca, V, Zn, Ga, Sr, La and Nd) in three different sample matrices (NIST 613, BCR-2 and SY-4) using a UV 266 nm laser.The study showed that the inter-element fractionation depends on the sample matrices and varies with time. The inter-element fractionation behaviour of V, Zn and Ga in the synthetic silicate glass NIST 613 is different from that in the quenched glass of fused silicate rocks (BCR-2 and SY-4). Relative to Ca, V, Zn and Ga show less fractionation in NIST 613 but larger fractionation in BCR-2 and SY-4. The relative internal standard normalized element intensity (RISNEI) is not linear with time for a laser ablation period of 210 s.Therefore, data acquisition using prolonged laser ablation without a matrix match will not improve the precision and accuracy for elements whose fractionation behavior is different from that of the internal standard element. The RISNEI versus time relationship for the Ærst 100 s laser ablation can be treated as linear to simplify the data calculation. In this paper, the internal standard normalized fractionation factor (ISNFF) is deÆned as the sum of the second half average RISNEI and the difference between the second and Ærst half average RISNEI, divided by the second half RISNEI of data acquisition, for the analyte concentration calculation.The ISNFF was applied for the correction of the data reduction in LA-ICP-MS analysis. The data accuracy for these seven elements is generally improved, particularly for an element whose calibration standard normalized ISNFF is signiÆcantly greater or less than 1 (e.g., Zn and Ga in this study).Good accuracy can be obtained for elements without ISNFF correction and matrix matches only if the calibration standard normalized ISNFF of the elements is close to 1. Since laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) was Ærst introduced by Gray1 in 1985, it has been widely used as a powerful analytical technique in various disciplines2±7 for in situ solid micro-sampling analyses. For LA-ICP-MS analysis, a combination of external and internal standardization is most commonly used for calibration and quantiÆcation.Good precision and accuracy have been reported if an element whose fractionation is similar to those of the analytes was chosen as an internal standard.2,3,8±10 However, if the fractionation of the analyte is signiÆcantly different from that of the internal standard element, poor accuracy is obtained. The analytically suitable and naturally occurring major elements with known concentrations (e.g., Ca and Si in silicate rocks and minerals) are commonly used as an internal standard for in situ multi-element LA-ICP-MS analysis.The selection of these elements is very limited. It is difÆcult to choose an internal standard which is capable of standardizing elements with different chemical characteristics (e.g., chalcophile, alkali and lithophile elements). Inter-element fractionation in LA-ICP-MS analysis has been studied by several investigators recently.9±13 However, the mechanism of inter-elemental fractionation in LA-ICP-MS analysis is still not very well understood.Every effort has been made to minimize inter-element fractionation. The local temperature has been reported to be an important factor in the fractionation and methods were adopted to reduce it.9,11 Active focusing of the laser beam also reduced the interelement fractionation in LA-ICP-MS analysis.9,11 Laser ablation using a 266 nm UV laser produced less inter-elemental fractionation than using 1064 and 532 nm lasers.9±11 More recently, Jeffries et al.13 reported that laser ablation using a frequency quintupled Nd:YAG laser (213 nm) signiÆcantly decreased the inter-element fractionation.Despite all these efforts, inter-element fractionation in LAICP- MS cannot be eliminated. In this work, inter-element fractionation in three different matrices (NIST 613, BCR-2 and G-2) during laser ablation at 266 nm was investigated.Interelement fractionation in LA-ICP-MS analysis depends on the sample matrix and varies with time during the laser ablation and transport. A fractionation factor was deÆned and applied to the correction of the LA-ICP-MS data reduction. Data accuracy for elements whose fractionation is different from that of an internal standard element was signiÆcantly improved after inter-element fractionation was corrected. Experimental Instrumentation A VG Elemental (Winsford, Cheshire, UK) PQ II S high sensitivity ICP-MS instrument was used.The software was upgraded to the OS/2 based PQ version 4.36. A two-directional communication is interfaced between the ICP-MS and the LA systems via a serial RS-232 (i.e., COM port). With the new PQ version software, the execution of each deÆned time setting by {Present address: Department of Marine Science, University of Southern Mississippi, Stennis Space Centre, MS 39529, USA. E-mail: zhongxing.chen@usm.edu; Tel: z1 228 688 1180; Fax: z1 228 688 1121.J. Anal. At. Spectrom., 1999, 14, 1823±1828 1823 This journal is # The Royal Society of Chemistry 1999LA corresponds to the ICP-MS data acquisition, allowing samples to be run in automated cycle sequences. Optimization of the plasma and mass spectrometer conditions was accomplished using NIST 613 glass containing about 50 mg g21 of trace elements. For each analytical run, the nebulizer gas Øow rate, torch position and lens setting were adjusted to optimize the signal intensity while ablating NIST 613 with a spot size of approximately 50 mm and a laser beam energy of less than 3 mJ, so that (1) sensitivity of 42Caw7.0 counts s21 mg21, (2) sensitivity of 139Law4200 counts s21 mg21 and (3) sensitivity of 232Thw3250 counts s21 mg21 and 232Th16O/232Thv0.3%.Typical ICP-MS operating conditions are given in Table 1. The LA used in this work was a Merchantek (Carlsbad, CA, USA) EO UV laser system. It is a compact computercontrolled solid state Nd:YAG laser whose output is frequency quadrupled to 266 nm with a maximum energy of 4 mJ.The LA software was upgraded to 2.12.0 UV. The laser was operated in a gated Q-switch mode for optimum stability, i.e., the Øash lamp was Æred at a frequency of 20 Hz and the laser output at computer-controlled pulse rates between 1 and 20 Hz. The UV laser beam is split by a beamsplitter that allows approximately 2% of the laser beam to go to a built-in energy meter and the other 98% of the beam to the objective lens, allowing an operator to monitor continuously the laser power before the Ænal objective lens during the ablation.Typical LA operating conditions are given in Table 2. Standards and samples NIST 613 synthetic silicate reference material was used as a test and calibration standard. The NIST 613 glass button was analysed as received. Major and trace element concentrations of the reference material, determined by solution nebulization ICP-OES and ICP-MS, were given by Chen et al.3 Two geological reference materials, Canada Centre for Energy and Mineral Technology (CANMET) SY-4 (diorite gneiss rock) and United State Geological Survey (USGS) BCR-2 (basalt) were analysed by LA-ICP-MS.Approximately 4±5 g of rock powder were weighed into a 20 cm3 Pt crucible and heated in a mufØe furnace (Lindberg Blue M) at 1550 �C in an air atmosphere for 12±15 h for melting and sample homogeneity.No Fe loss occurred from the sample to the Pt crucible in an air atmosphere. The crucible was removed from the furnace and the rock melt was quickly quenched to a glass in water. Several centimetre-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 and calibration After the instruments had been optimised and a procedure set up, both LA and ICP-MS were run automatically following the analytical procedure from the Ærst to last analysis.Experiments were carried out in the following cycle sequences: NIST 613, BCR-2, SY-4, NIST 613, º, NIST 613, BCR-2, SY-4, NIST 613. Each analysis consists of a 100 s pre-ablation delay for the background (data collected), 210 s lasering for signal (data collected) and a 90 s sample delay for Øush (no data collected).Seven elements (42Ca, 51V, 66Zn, 69Ga, 88Sr, 139La and 146Nd) which range from the highest (Zn) to least (Nd) fractionation relative to Ca were investigated for their fractionation behaviour and correction procedure. Data were acquired in the peak-jumping mode with a dwell time of 10.24 ms. Background levels for each element were obtained by acquiring data for a gas blank for 100 s prior to laser sampling. Sample data were corrected from 210 s ablation and data acquisition.Count rates were collected and exported as CSV (comma delimited values) Æles by PQ Version 4.36 Time Resolved Analysis (TRA) software. All subsequent data manipulations were later accomplished off-line using a commercial spreadsheet program by manual operation and by in-house written software. All sample data reported were background subtracted. The Ærst few data after initial laser ablation were not included in the data reduction to avoid potential surface contamination.Calibration and quantiÆcation of the analysis utilized both external and internal calibrations. NIST 613 was used as an external calibration standard and Ca as an internal standard. Results and discussion Matrix effect The relative internal standard (e.g., Ca in this study) normalized element intensity (RISNEI) is deÆned here as RISNEI~ÖIElement=IInternal standardÜt =ÖIElement=IInternal standardÜInitial ablation Fig. 1. shows the RISNEI of Zn, Ga, V and La in three different matrices (NIST 613, BCR-2 and SY-4).The data shown in the Ægures are averages of every nine readings from TRA data acquisition to smooth the curve. The RISNEI versus time relationship of Sr and Nd in the three matrices is similar to that of La, and is not shown in the Ægure. Inter-element fractionation of the elements depends on the sample matrices. The interelement fractionation behaviour of the elements in the synthetic silicate glass NIST 613 is different from that in the the fused silicate rocks (BCR-2 and SY-4).Relative to Ca, V, Zn and Ga show less fractionation in NIST 613 but greater fractionation in BCR-2 and SY-4. The fractionation of an element in SY-4 is higher than that of the same element in BCR-2. Although the fractionation of the same element is different in three sample matrices (Fig. 1), the RISNEI of all elements in the same sample matrix displays a similar trend, e.g., the RISNEI of all the Table 1 ICP-MS operating conditions Inductively coupled plasma– Plasma gas Argon Forward power 1350W ReØected power v5W Gas Øow rates– Plasma gas Øow rate 14 dm3 min21 Auxiliary gas Øow rate 0.94 dm3 min21 Inner gas Øow rate y1.20 dm3 min21 (see text) Interface– Sampling distance (load coil to sample aperture) 16 mm Sampling aperture Nickel, 1.0 mm diameter Skimmer aperture Nickel, 0.7 mm diameter Ion lens settings– Extraction lens 2365 V Collector lens 277.9 V L1 lens z2.8 V L2 lens 210.7 V L3 lens z4.8 V L4 lens 229.9 V Pole bias 0 V Data acquisition parameters– Measurement mode Peak jumping Dwell time 10.24 ms Data acquisition time 310 s Points per peak 1 Table 2 LA operating condition Laser mode Q-switch Wavelength 266 nm Flash lamp frequency 20 Hz Laser output frequency 10 Hz Laser energy before objective lens v3 mJ Spot size 3 (y50 mm) 1824 J.Anal. At. Spectrom., 1999, 14, 1823±1828elements in SY-4 increases with time after the initial ablation and transport, reaches its maximum at approximately 275 s, and then decreases with time.The RISNEI is not linear with time for a period of 210 s during laser ablation and transport. Therefore, data acquisition using prolonged laser ablation without a matrix match will not improve the precision and accuracy for elements whose fractionation behaviour is different from that of the internal standard. Variations with time Fig. 2 shows the RISNEI of Zn and Ga in SY-4 and BCR-2 of the Ærst and last analyses.The last analysis was repeated using the same LA-ICP-MS operating conditions approximately 2 h after the Ærst analysis. Fig. 2 indicates that the inter-element fractionation varies with time. Physical conditions during the laser ablation and transport may cause this variation. The oxygen content in the sample chamber signiÆcantly affects the inter-element fractionation during the laser ablation and transport and the mechanism is still under investigation. It was observed that different laser beam focusing may also affect the fractionation.9,11 The local temperature of the sample is an important factor in fractionation.11 Despite the RISNEI changes for the two different analyses, interestingly, all the elements in the same sample matrix displayed the same RISNEI versus time trend described above (Fig. 2).Internal standard normalized fractionation factor (ISNFF) Fig. 3 shows the relative Ca normalized Ga and Zn intensity in SY-4 versus time for the Ærst 100 s laser ablation.A very good linear relationship was observed. To simplify the calculation here, it is assumed that the RISNEI is linear with time for the Ærst 100 s during the laser ablation and transport. This is reasonable for our routine LA-ICP-MS analysis which uses 60 s for data acquisition. Fig. 4 is a schematic diagram shown the RISNEI versus time. AB is the RISNEI when the laser is initially Æred, CD the RISNEI when the laser is off, EF the average RISNEI of the data acquisition, GH the average RISNEI of the Ærst half of the data acquisition and IJ an average RISNEI of the second half of the data acquisition. The internal standard normalized fractionation factor (ISNFF) is deÆned here as the sum of the second half average RISNEI and the difference between the second and Ærst half average RISNEI, divided by the second half RISNEI: ISNFF~âIJzÖIJ{GHÜä=IJ~Ö2IJ{GHÜ=IJ~2{GH=IJ The ISNFF calculated varies with the time interval selected for data reduction.AnaverageISNFFof individual elements for the Ærst 100 s laser ablation in three different matrices is presented in Table 3.Relative to Ca, six elementsshow different fractionation behaviour in three sample matrices during the LA-ICP-MS analysis. ISNFF ranges from 0.965 for La to 1.259 for Zn in NIST 613, from 0.991 for La to 1.332 for Zn in BCR-2 and from 0.992 forLato 1.432 forZnin SY-4. The fractionation behaviour of the elements in fused glass chips of two geological silicate reference materials (BCR-2 and SY-4) is different from those in synthetic silicate glass standard. The relationship between ISNFF and the major components (SiO2 and Al2O3) in three sample matrices is shown in Fig. 5. The ISNFF increases and decreases with Al2O3 and SiO2 contents in the sample matrices, respectively. The relationship between the fractionation and elemental properties such as Æeld strength and melting temperature has been investigated previously,9 but no signiÆcant correlation was observed.A linear relationship between the ISNFF and the sum of Ærst and second ionizati enthalpies is shown in Fig. 6, indicating that inter-element fractionation during laser ablation dependsonionization energy. This suggests that a low ionization energy element was ablated easier than that of high ionization energy, which lead to the progressive enrichment of high ionization-energy elements in the ablation Fig. 1 Relative Ca normalized element intensity of (a) Zn, (b) Ga, (c) V and (d) La in NIST 613, BCR-2 and SY-4 in LA-ICP-MS analysis.J. Anal. At. Spectrom., 1999, 14, 1823±1828 1825pit. The mechanism of inter-element fractionation in LA-ICPMSis not very well understood, and will be investigated further. Inter-element fractionation correction The ISNFF for individual elements in NIST 613 standard and samples was calculated using the equation described above. Fig. 2 Relative Ca normalized element intensity of (a) Zn and (b) Ga in SY-4 and of (c) Zn and (d) Ga in BCR-2 for the Ærst and last analyses.The last analysis was repeated using the same LA-ICP-MS operating conditions approximately 2 h after the Ærst analysis. Fig. 3 Relative Ca normalized element intensity of (a) Zn and (b) Ga versus time, showing linearity for the Ærst 100 s laser ablation and transport of LA-ICP-MS. Fig. 4 Simplied diagram of relative internal standard normalized element intensity (RISNEI) versus time.AB is the RISNEI when the laser is initially Æred, CD the RISNEI when the laser is off, EF the average RISNEI of the data acquisition, GH the average RISNEI of the Ærst half of the data acquisition and IJ the average RISNEI of the second half of the data acquisition. The internal standard normalized fractionation factor (ISNFF) is deÆned here as the sum of the second half average RISNEI and the difference between the second and Ærst half average RISNEI, divided by second half RISNEI of data acquisition. 1826 J. Anal. At. Spectrom., 1999, 14, 1823±1828The ISNFF of the element in the samples was then normalized to that of the same element in standard, and applied to the Ænal elemental concentration calculation described previously:15 CÖSample analyte; correctedÜ~CÖSample analyteÜ |ISNFFÖStandard analyteÜ=ISNFFÖSample analyteÜ The results with and without inter-element fractionation correction are present in Table 4. The accuracy expressed as the relative difference between this work and literature value is shown in Fig. 7. The data accuracy by LA-ICP-MS with ISNFF correction developed in this study (using the equation described above) is signiÆcantly improved, particularly for Zn and Ga. The relative difference for Zn is reduced from 53.5 to 2.6% for BCR-2 and from 81.5 to 7.2% for SY-4, and that for Ga from 16.3 to 23.0% for BCR-2 and from 30.9 to 25.4% for SY-4. Since Gray1 performed the Ærst LA-ICP-MS analysis in 1985, good precision and accuracy by LA-ICP-MS have been obtained without inter-element fractionation correction and matrix matches if an element whose fractionation is similar to those of the analytes was chosen as an internal standard.2,3,9,10 This can be very clearly explained in Table 3.Relative to Ca, the elements Sr, La and Nd have ISNFFs very close to 1 in both the external calibration standard (NIST 613) and samples (BCR-2 and SY-4). The results calculated using NIST 613 as an external standard and Ca as an internal standard are not very different between the data with and without ISNFF correction (Table 4).However, if the calibration standard normalized ISNFF of the elements (e.g., Zn and Ga in this study) is signiÆcantly greater or less than 1, an error is introduced into the calculated results of LA-ICP-MS analysis if the procedure used here is not adopted. Conclusions Inter-element fractionation of seven elements (Ca, V, Zn, Ga, Sr, La and Nd) in three different sample matrices (NIST 613, BCR-2 and SY-4) during LA-ICP-MS analysis using a UV 266 nm laser was investigated.The fractionation depends on the sample matrices and varies with time. A linear relationship between the ISNFF and the ionization enthalpy indicates that Table 3 ISSF of elements for the Ærst 100 s laser ablation in three different matrices Element 51V 66Zn 69Ga 88Sr 139La 146Nd NIST 613 (average, n~5) 1.057 1.259 1.142 1.012 0.965 0.974 RSD (%) 1.8 3.0 2.3 0.3 1.6 1.3 BCR-2 (average, n~5) 1.109 1.332 1.199 0.991 0.983 0.985 RSD (%) 1.4 1.6 2.1 2.0 2.5 3.0 SY-4 (average, n~5) 1.131 1.432 1.286 0.992 0.998 0.980 RSD (%) 2.1 3.0 3.0 1.3 1.1 1.7 Fig. 5 Relationship between the ISNFF of V, Zn and Ga and major components [(a)SiO2 and (b) Al2O3] in three different sample matrices. Fig. 6 Relationship between the ISNFF of V, Zn and Ga and the sum of Ærst and second ionization enthalpies in three sample matrices: (a) NIST 613, (b) BCR-2 and (c) SY-4.Ionization enthalpy values from ref. 14. J. Anal. At. Spectrom., 1999, 14, 1823±1828 1827inter-element fractionation during laser ablation depends on ionization energy. In this paper, the internal standard normalized fractionation factor (ISNFF) is deÆned as the sum of the second half average relative internal standard (e.g., Ca in this study) normalized element intensity (RISNEI) and the difference between the second and Ærst half average RISNEI, divided by second half RISNEI of data acquisition.A procedure was developed to correct inter-element fractionation. The ISNFF was applied to the LA-ICP-MS data reduction to correct the inter-element fractionation during laser ablation and transport. Data accuracy is generally improved, particularly for an element whose ISNFF normalized to the calibration standard is signiÆcantly greater or less than 1. Good accuracy can be obtained without ISNFF correction and matrix matches only if the calibration standard normalized ISNFF is close to 1.Further work is needed to investigate the inter-element fractionation of more elements in various matrices, e.g., Pb, Th and U in zircons. The isotopic ratio measurement by LA-ICP-MS in zircons has been used for geochronology.18,19 However, owing to the inter-element fractionation, this measurement has been limited to the determination of only Pb207/Pb206 ratios. With the interelement fractionation correction, it will be possible to measure more useful U/Pb and Th/Pb ratios.The PQ II S ICP-MS and Merchantek EO LA system were purchased through an NSERC major installation grant (Grant No. EQM0184535) to J. K. B. Bishop, D. Canil and K. M. Gillis. The author thanks H. Longerich and D. Canil for their useful comments and suggestions in improving the manuscript. References 1 A. L. Gray, Analyst, 1985, 110, 551. 2 S. E. Jackson, H. P. Longerich, G. R. Dunning and B. J.Fryer, Can. Mineral., 1992, 30, 1049. 3 Z. Chen, W. Doherty and D. C. Gregoire, J. Anal. At. Spectrom., 1997, 12, 653. 4 R. D. Evans, P. M. Outridge and P. Richner, J. Anal. At. Spectrom., 1994, 9, 985. 5 S. Tanaka, N. Yasushi, N. Sato, T. Fukasawa, S. J. Santosa, K. Yamanaka and T. Ootoshi, J. Anal. At. Spectrom., 1998, 13, 135. 6 C. Leloup, P. Marty, D. Dallava and M. Perdereau, J. Anal. At. Spectrom., 1997, 12, 945. 7 A. Raith, R. C. Hutton, I. D. Abell and J. Crighton, J.Anal. At. Spectrom., 1995, 10, 591. 8 B. J. Fryer, S. E. Jackson and H. P. Longerich, Can. Mineral., 1995, 33, 303. 9 T. E. Jeffries, N. J. Pearce, W. T. Perkins and A. Raith, Anal. Commun., 1996, 33, 35. 10 H. P. Longerich, D. Gunther and S. E. Jackson, Fresenius' J. Anal. Chem., 1996, 355, 538. 11 D. Figg and M. S. Kahr, Appl. Spectrosc., 1997, 51, 1185. 12 P. M. Outridge, W. Doherty and D. C. Gregoire, Spectrochim. Acta, Part B, 1997, 52, 2093. 13 T. Jeffries, S. E. Jackson and H. P. Longerich, J. Anal. At. Spectrom., 1998, 13, 935. 14 A. M. James and M. P. Lord, in Macmillan's Chemical and Physical Data, Macmillan, London, 1992. 15 H. P. Longerich, S. E. Jackson and D. Gunther, J. Anal. At. Spectrom., 1996, 11, 899. 16 S. A. Wilson, United States Geological Survey, Open File Report, in the press. 17 W. S. Bowman, Geostand. Newsl., 1995, 19, 101. 18 B. J. Fryer, S. E. Jackson and H. P. Longerich, Chem. Geol., 1993, 109, 1. 19 R. Feng, N. Machado and J. Ludden, Geochim. Cosmochim. Acta, 1993, 57, 3479. Paper 9/03272J Table 4 Concentrations determined by LA-ICP-MS before and after the ISNFF correction, and literature values for reference materials BCR-2 and SY-4 Element 51V 66Zn 69Ga 88Sr 139La 146Nd Before ISNFF correction– BCR-2 (average, n~5)/mg g21 431 195 27 312 24.2 27.2 RSD (%) 1.1 8.8 3.8 2.2 3.9 4.8 SY-4 (average, n~5)/mg g21 6.18 169 45.8 1099 54.9 53.8 RSD (%) 2.0 7.0 3.4 1.6 1.6 1.7 After ISNFF correction– BCR-2 (average, n~5)/mg g21 399 130 22.3 322 23.6 26.7 RSD (%) 2.8 12.7 5.0 1.7 7.6 8.4 SY-4 (average, n~5)/mg g21 5.67 100 33.1 1130 52.6 52.7 RSD (%) 0.1 11.1 13.6 9.1 3.4 3.4 Literature value– BCR-216/mg g21 416 127 23 346 25 28 SY-417/mg g21 8.0 93 35 1191 58 57 Fig. 7 Accuracy expressed as the relative difference between results obtained in this work (A) before and (B) after ISNFF correction, and literature values for (a) BCR-2 (ref. 16) and (b) SY-4 (ref. 17). 1828 J. Anal. At. Spectrom., 1999, 14, 1823±1828