Anal. Chem. 2003, 75, 6723-6727
Quantitative Elemental Analysis for Rhodium and Palladium in Minerals by Time-of-Flight Resonance Ionization Mass Spectrometry S. S. Dimov,*,† S. L. Chryssoulis,† and R. H. Lipson‡
AMTEL, 100 Collip Circle, UWO Research Park, London, ON, N6G 4X8 Canada, and Department of Chemistry, University of Western Ontario, London, ON N6A 5B7 Canada
Results are presented for the trace analysis of Pd and Rh by time-of-flight-resonance ionization mass spectrometry (TOF-RIMS). The spectrometer, developed at the Advanced Mineral Technology Laboratory (Ontario, Canada), is based on a commercial laser-induced mass analyzer with upgrades that include independent laser ablation and ionization sources and pulsed ion optics to minimize noise caused by primary ion formation. The schemes presented for Rh and particularly for Pd detection are simpler than others reported in the literature. The experimental laser fluences were found to be in reasonable agreement with theoretical estimates. The TOF-RIMS measurements were quantified on the basis of calibration curves derived using reference samples covering 3 orders of magnitude in concentration. Minimum detection limits of ∼15 parts per billion were found for both metals, with a precision of ∼(15%. Samples from sulfide, iron oxide, and silicate minerals were also examined. The results are in excellent agreement with those obtained using dynamics secondary ion mass spectrometry. The mining industry is always seeking analytical techniques to characterize platinum group mineral deposits and to quantify the amount of precious metals (Au, Pd, Rh) in spinels and silicates at parts per billion (ppb) levels. Typically, their main objectives are to establish the maximum attainable recovery by sulfide flotation and to measure losses in the slag phases of roasters. Unfortunately, many of the available quantitative techniques such as electron probe microanalysis1, proton (or particle)-induced X-ray emission1,2, secondary ion mass spectrometry (SIMS3), laser ablation inductively coupled plasma-mass spectrometry4,5, glow discharge mass spectrometry6,7, and time-of-flight laser ionization * Corresponding author. E-mail:
[email protected]. † AMTEL. ‡ University of Western Ontario. (1) Tamana, H.; Criddle, A.; Grime, G.; Vaughan, D.; Stratt, J. Nuc. Instrum. And Methods Phys. Res. 1994, B89, 213-218. (2) Annegarn, H. J.; Eramus, C. S.; Sellschop, J. P. F. Nucl. Instrum. Methods Phys. Res. B 1984, B3, 181-184. (3) Koppenaal, D. Anal. Chem. 1990, 62, 303R-324R. (4) Hinds, M. W.; Kogan, V. K. Spectroscopy 1995, 10, 14-18. (5) Byrne, J. P.; Gre´goire, D. C.; Benyounes, M. E.; Chakrabarti, C. L. Spectrochim. Acta 1997, 52B, 1575-1586. (6) Toland, M. M.; Jarvis, I.; Jarvis, K. E. Chem. Geol. 1995, 124, 21-36. (7) Aldave de las Heras, L.; Bocci, F.; Betti, M.; Actis-Dato, L. O. Fresenius J. Anal. Chem. 2000, 368, 95-102. 10.1021/ac030158c CCC: $25.00 Published on Web 10/31/2003
© 2003 American Chemical Society
mass spectrometry (TOF-LIMS8,9) have limits of detection only in the 0.01-200 parts per million (ppm) range. Time-of-flight resonance ionization spectrometry (TOF-RIMS10-12), however, is an attractive alternative microbeam analytical technique that can meet the needs of this problem, namely, elemental selectivity, a linear response over a large dynamic range and sensitivity in the ppb to parts per trillion (ppt) ranges. A TOF-RIMS analysis takes place in three steps. A sample is first vaporized by laser ablation to form a plume of neutral atoms. The gas-phase atoms are then resonantly excited and ionized using one or more tunable laser sources. Finally, the resultant ions are mass dispersed and detected in a time-of-flight mass spectrometer. No significant matrix effects are observed when separate lasers for ablation and postionization are used.13 Under optimized conditions the ablation process provides a representative sample of the elemental composition of the sample.13 Improved sensitivity is a result of the significant enhancement in atomic cross sections upon resonant excitation, which can lead to as much as a 109-fold increase in ionization efficiency. Thus, only small amounts of analyte are required, and unlike nonresonant methods such as TOF-LIMS, isobaric interferences (same mass but different species) are effectively eliminated. Since the electronic structure of each atom is unique, resonant ionization schemes with specific laser wavelengths can be “designed” for every element of the periodic table. To date, more than 70 elements have been analyzed using RIMS.12,14,15 A TOF-RIMS approach has been used by several laboratories for trace analyses of Au and, to a lesser extent, the platinum group elements (Pd, Pt, Rh). A broad range of terrestrial and extraterrestrial samples has been explored including sulfide deposits, soil, seawater samples, and meteorites.16-21 While detection limits as (8) Southon, M. J.; Witt, M. C.; Harris, A.; Wallach, E. R. Vaccuum 1984, 903908. (9) Simons, D. S. Int. J. Mass Spectrom. Ion Processes 1983, 55, 15-30. (10) Winograd, N.; Baxter, J. P.; Kimock, F. M. Chem. Phys. Lett. 1982, 88, 581-584. (11) Kimock, F. M.; Baxter, J. P.; Pappas, D. L.; Kobrin, P. H.; Winograd, N. Anal. Chem. 1984, 56, 2782-2791. (12) Young, J. P.; Shaw, R. W.; Smith, D. H. Anal. Chem. 1989, 61, 1271A1277A. (13) Dimov, S. S.; Chryssoulis, S. L. Spectrochim. Acta 1998, B53, 399-406. (14) Hurst, G. S.; Payne; M. G.; Kramer, S. D.; Young, J. P. Rev. Mod. Phys. 1979, 51, 767-819. (15) Hurst, G. S.; Payne, M. G. Spectrochim. Acta 1988, 43B, 715-726. (16) Hui, Q.; Chen. D. Y.; Niu, J. G.; Cheng, Y.; Xu, X. Y.; Zhao, W. Z. Inst. Phys. Conf. Ser. 1990, 114, 297-300.
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003 6723
low as 0.1 ppb-0.2 ppt for Au16,17 and 0.02 ppt for Pt and Rh17,20,21 have been reported, those experiments were complicated by the use of many lasers in the RIMS schemes. To the best of our knowledge, there is only one published account on the detection of Pd by TOF-RIMS that involved three tunable dye lasers to excite high-lying atomic Rydberg states, followed by pulse field ionization.22 In previous papers, we described a TOF-RIMS spectrometer developed at the Advanced Mineral Technology Laboratory (AMTEL) that, for various Au samples, was shown to be capable of reproducible minimum detection limits (MDL; 2σ) of ∼3.2 ppb with a precision of ∼(15%.23,24 In this work, we report our findings for samples containing Rh and Pd. Furthermore, our RIMS schemes for Rh and particularly for Pd are much simpler than those previously reported. EXPERIMENTAL SECTION The AMTEL TOF-RIMS system has been described in detail elsewhere.24 Briefly, the apparatus is based on a modified commercial TOF-LIMS instrument, (LIMA-2A, Cambridge Mass Spectrometry, Inc.). The optical system incorporates two pulsed (10 Hz or single-shot operation) infrared Nd:YAG lasers (Quanta Ray, model DCR-2). The output of the first is quadrupled to λ ) 266 nm for laser ablation. The output of the second laser, upgraded with the appropriate optics to provide a Gaussian-shaped beam, can be frequency doubled to λ ) 532 nm, tripled to 355 nm, or quadrupled to λ ) 266 nm. The first two outputs can be used to pump a postionization dye laser (Lumonics, model HD300B) and possibly provide the ionization photon in the second step of the resonance ionization scheme. All three harmonic signals from the second Nd:YAG laser could be generated simultaneously by inserting an additional harmonic generator into the path of the unused portion of the 532-nm output beam. The output of the dye laser could be extended down to wavelengths as short as 198 nm using an Inrad AT-III harmonic generator. In principle, this optical arrangement can be used to devise a RIMS scheme for every element in the periodic table with the exception of He and Ne. The intensities of the dye laser and ionizing radiation at λ ) 266 nm could be continuously adjusted without changing the beam spot sizes, using λ/2 plates placed before a polarizing combiner used to overlap the beams. An optical delay line introduced a variable time delay between the excitation and ionizing beams to maximize the ionization efficiency in the second step of the ionization scheme while minimizing losses due to excited-state spontaneous emission. Wavelength calibration was achieved with an optogalvanic wavemeter.25 (17) Bekov, G. I.; Letokhov, V. S. Inst. Phys. Conf. Ser. 1988, 94, 331-336. (18) Bekov, G. I.; Radyev, V. N.; Letokhov, V. S. Spectrochim. Acta 1988, 43B, 491-499. (19) Bekov, G. I.; Kolpakov, I. V.; Radaev, V. N.; Veselov, V. A. Inst. Phys. Conf. Ser. 1990, 114, 265-270. (20) Bekov, G. I.; Letokhov, V. S. In Optoelectronics for Environmental Science; Martelluci, S., Chester, A. N., Eds.; Plenum: New York, 1990; pp 177183. (21) Asaro, F.; Bekov, G. I.; Belkin, R. S.; Gulevich, V. M.; Khomyakov, N. G.; Kursky, A. N.; Pakhomov, D. Yu. Inst. Phys. Conf. Ser. 1992, 128, 205208. (22) Ishikawa, T. Jpn. J. Appl. Phys. 1993, 32, 4779-4780. (23) Dimov, S. S.; Chryssoulis, S. L.; Lipson R. H. AIP Conf. Proc. 2001, 584, 46-51. (24) Dimov, S. S.; Chryssoulis, S. L.; Lipson, R. H. Rev. Sci. Instrum. 2002, 73, 4295-4306.
6724
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
Figure 1. Simplified energy level diagrams showing the RIMS schemes used for (a) Pd and (b) Rh.
Ions produced by resonant ionization were mass dispersed in a reflectron-type TOF mass spectrometer (mass resolving power m/∆m ∼ 400) and detected with an electron multiplier. The output signals from the detector were logarithmically compressed and amplified for improved dynamic range and precision of digitization. Noise due to the production of primary ions created during the ablation step was eliminated by quickly switching the polarity and potentials of acceleration and ion optics between consecutive ablation and postionization steps. SPECTROSCOPIC AND THEORETICAL CONSIDERATIONS Simplified energy level diagrams showing the RIMS schemes used for Pd and Rh are shown in Figure 1a and b, respectively. Pd and Rh have ionization potentials of 67 242(1) and 60 160.1(4) cm-1, respectively,26 and both elements offer several intermediate levels for resonant excitation and subsequent ionization by either a one-color (ω1 + ω1) or two-color (ω1 + ω2) scheme. For Pd, light at λ ) 276.39 nm (ω1) was used to excite the 4d95p 3P1° (lifetime, τ ∼ 5 ns27) r 4d10 1S0 transition.28 Ionization from this intermediate state is possible using the same wavelength or by using radiation at λ ) 266 nm (ω2). Prior to any experimental work, the laser fluences required for a successful TOF-RIMS detection by either scheme were first estimated theoretically. The power required to saturate the resonant excitation step, Psat, was calculated to be 82.2 W/cm2 (fluence ) 4.1 nJ/mm2) using
Psat ) hc/2τσ21λ
(1)
where σ21 is the absorption cross section. σ21 was calculated from the excited-state lifetime using
σ21 ) λ2/2πτ∆ωD
(2)
where λ is the wavelength of the transition (in cm) and ∆ωD is (25) Dovichi, N. J.; Moore, D. S.; Keller, R. A. Appl. Opt. 1982, 21, 1468-1473. (26) Callender, C. L., Hackett, P. A.; Rayner, D. M. J. Opt. Soc. Am. B 1988, 5, 614-618. (27) Doidge, P. S. Spectrochim. Acta 1995, 50B, 209-263. (28) Moore, C. E. Atomic Energy Levels; National Bureau of Standards Circ. No. 467; U.S. GPO: Washington, DC, 1971; Vol. III.
Figure 2. (a) Pd TOF-RIMS ion signal as a function of the energy of the postionization source used in the (ω1 + ω1) scheme; (b) Pd TOFRIMS ion signal as a function of the energy of the postionization source used in the (ω1 + ω2) scheme. The energy used in the excitation step was 0.3 mJ/pulse.
the transition line width, which for these calculations was chosen to be the Doppler width (in rad s-1) of the atom at its boiling point (3413.15 K for Pd). Bound-continuum photoionization cross sections, σI, for the second step of the resonance ionization scheme are more difficult to establish. However, a reasonable estimate of 9.48 × 10-18 cm2 could be obtained using29
σI )
8 (IP*/E)3 × 10-18 cm2 xIP*/13.6
(3)
where IP* is the ionization potential (eV) from the excited state of the atom and E is the energy (eV) of the ionizing photon. The energy required to saturate the ionizing transition, Esat, was then calculated from30
Esat ) (5.03 × 10-6 σIλ)-1
(4)
where σI is given in units of 10-18 cm2 and λ is in nanometer units. In the Pd (ω1 + ω1) scheme, 0.76 mJ/mm2 is required to saturate the ionization step or 1.52 × 107 W/cm2. The saturating fluence and power are slightly higher for the (ω1 + ω2) scheme at 0.88 mJ/mm2 and 1.76 × 107 W/cm2, respectively. A (ω1 + ω1) scheme is only possible for exciting ground-state Rh atoms using wavelengths of 1 mJ/pulse, while it does for the (ω1 + ω2) scheme (Figure 2b). Figure 3 shows plots of the Pd ion signal as a function of the concentration of Pd in the reference samples for the (ω1 + ω2) scheme. The MDL was determined to be 15 ppb. Using the 266nm laser as the ionization source provided a 2-3 times increase in the TOF-RIMS signal over the one-color scheme. The Rh TOF-RIMS signals were found to saturate at a photoionization energy of 4.5 mJ/pulse when the excitation fluence (0.5 mJ/pulse) saturated the 4d85p 4G11/2° r 4d85s 4F9/2 excitation step. Conversely, the excitation step could be saturated using nanoJoule pulses when the photoionization energy was 5 mJ/ pulse. The experimental detection sensitivity for Rh for a 10 µm ablation spot size was explored for different ion detector voltages. A MDL for Rh of 17 ppb was established using fluences of 0.9 and 5 mJ/mm2 for the excitation and ionization steps, respectively. Increasing the ablation volume analyzed could decrease this limit further. A similar detection sensitivity was attained in experiments carried out with the excitation wavelength tuned to the Rh 4d85p 4F ° r 4d85s 4F 9/2 9/2 transition. (33) All standards were produced by the Royal Canadian Mint (RCM), Ottawa Canada. The Rh concentrations for these standards were provided directly by the RCM.
6726 Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
Table 2. Palladium and Rhodium Content in Mineral Grains Determined by TOF-RIMS and Dynamic SIMS within Their Overlapping Range of Detection Sensitivities Pd grain no.
mineral
concn measd by TOF-RIMS (ppb)
concn measd by dynamic SIMS (ppb)
1 2 3 4 5 6 7 8 9 10
chalcopyrite chalcopyrite pentlandite pentlandite pentlandite pentlandite pentlandite pentlandite pentlandite pentlandite
260 280 1700 2500 2700 3300 3300 5400 5200 6500
250 290 1960 3780 4090 3290 4130 5600 6930 6720
Rh grain no.
mineral
concn measd by TOF-RIMS (ppb)
concn measd by dynamic SIMS (ppb)
1 2 3 4 5 6 7 8 9 10 11 12 13 14
pentlandite pentlandite pentlandite pentlandite pyrrhotite pyrrhotite pyrrhotite pentlandite pentlandite pentlandite pentlandite pentlandite pentlandite pyrrhotite
620 680 690 780 750 970 870 780 820 980 910 910 1000 900
970 1003 890 840 670 1090 930 900 820 970 780 920 900 1000
The reproducibility of the Rh and Pd TOF-RIMS signals was also assessed in this experiment. The Rh data are shown in Figure 4. Ten consecutive sets of TOF-RIMS spectra were taken, and each set was averaged from 50 laser shots. The relative standard deviation of the mean is 8%. Figure 4 also shows the 95% confidence intervals for each particular set and for all sets together (dotted line about the mean line). The Pd experiments yielded similar results.
APPLICATION OF TOF-RIMS TO THE ANALYSES OF TRACE PRECIOUS METAL ELEMENTS IN MINERALS The TOF-RIMS schemes were applied to the trace elemental analysis in sulfide and iron oxide minerals with concentrations of Pd and Rh in the ppm to ppb range. The same sulfide and oxide particles were also analyzed by dynamic SIMS using a Cameca 3f SIMS mass spectrometer for comparison. The estimated Pd and Rh concentrations obtained by the two methods within their overlapping range of detection sensitivity are listed in Table 2. The TOF-RIMS data for Pd and Rh are in good agreement with the data obtained by dynamic SIMS (relative mean deviation, (16%). DISCUSSION AND CONCLUSIONS This study demonstrates that TOF-RIMS has excellent potential for providing quantitative ultratrace analysis of precious metals in minerals. The core instrument used is a commercial laserinduced mass analyzer (LIMA) system, which was upgraded for TOF-RIMS operation with relative ease and whose response is linear over at least 3 orders of magnitude. The schemes used for Rh and particularly Pd detection are simpler than those previously reported. In each case, a (ω1 + ω2) RIMS scheme was found to be superior to a (ω1 + ω1) approach because in the former both the excitation and photoionization (34) Wilson, R. G.; Stevie, F. A.; Chryssoulis, S. L.; Irwin, R. B. J. Vac. Sci. Technol. A 1994, 12, 2415-2419.
steps could be independently optimized. As noted in the introduction, a better detection sensitivity could be obtained (down to ppt levels), for example, by introducing resonant excitation schemes that involve exciting transitions to high-lying Rydberg states, followed by electric field ionization.22 However, the complexity and costs of such an approach were deemed unnecessary because the MDLs obtained in this work are sufficient to address the needs of the mining industry for trace elemental analysis of precious metals at ppb concentration levels. The agreement between the TOF-RIMS and dynamic SIMS measurements is for the most part, excellent. The MDLs for Pd and Rh by SIMS are comparable to our TOF-RIMS results, at 120 and 10 ppb by weight, respectively.34 However, the latter method is deemed more versatile because the pulsed microbeam nature of the laser ablation process minimizes space charge effects that can be a restricting factor for the analysis of insulating materials by dynamic SIMS. For example, in separate experiments carried out for this work, TOF-RIMS analyses of silicate mineral particles were able to establish Pd concentrations in the 80-90 ppb range. ACKNOWLEDGMENT This work was supported in part by the National Research Council of Canada under the Industrial Research Assistance Program (IRAP). Received for review April 21, 2003. Accepted September 9, 2003. AC030158C
Analytical Chemistry, Vol. 75, No. 23, December 1, 2003
6727
