Preliminary results of trace element contents in mantle derived specimens from Kakanui, New Zealand, of the Jarosewich's Microbeam Reference Samples Collection: The first step of a new program
Over the last two decades, microbeam instrumentation has been the subject of a great technological development—ever more sophisticated analytical capabilities including lower detection limits and better mass- and spectral resolution. Nevertheless, the accuracy of microananalysis is limited because of the lack of reliable reference materials for trace elements (<0.1 wt. %) with a similar composition to the unknown. The program started at the Department of Mineral Sciences, NMNH, Smithsonian Institution (Logan, 2009) has the long-term objective of establishing working values for the trace element contents of the 59 specimens (including silicates, oxides, carbonates, REE orthoposphates, and volcanic and synthetic glasses) from the Jarosewich Microbeam Reference Samples Collection. During the 1970s and 1980s Eugene Jarosewich characterized these specimens for major and minor elements by wet chemistry to be used as RM for microanalysis, Jarosewich et al., (1980). Currently, more than 600 reference samples from this collection are yearly requested by laboratories all over the world. Suitable homogeneous crystals for reference samples can only grow in an environment with constant P-T conditions for a long time such as the mantle. However, during the ascent of mantle nodules to the surface of the earth, their original composition could be affected by metasomatism. Therefore, this program focuses on a rigorous analysis of the degree and scale of homogeneity of trace element contents in mantle derived specimens, as they are the most likely to satisfy the requirements to become RMs. Because the most frequently requested specimens of this collection are from the well know locality Kakanui, New Zealand, a statistical analysis was performed based on 640 electron microprobe analyses (P, Sc, V, Cr, Ni and Zn) on the tschermakitic augite (NMNH 122142), kaersutitic hornblende (NMNH 143965) and a pyrope garnet,(NMNH 143968) in traverses across three randomly selected grains. In general, the t-test and the Fisher test (variance analysis) indicate that similar means and compositional ranges should be expected in different grains from the sampled master vials. Further micronanalytical investigation of the P distribution in the hornblende is necessary. Correlations (Cr vs. V, Ca vs. V, and Cr vs. Ti, among others) show trends that appear to represent pristine fractionations. The EPMA means are in good agreement with XRF analysis of bulk samples (1.6 g), except for Zn in the Kakanui augite. The results obtained for the Ni and V contents indicate that the Kakanui suite covers a fairly broad compositional range. The slightly high t- and low F- values obtained for Cr in the pyrope and the augite indicate a mild zonation that appears to represent a pristine fractionation. These results indicate that the Kakanui suite has not been affected by substantial post-depositional metasomatism and are therefore; good candidates for further characterization by LA-ICP-MS and secondary ion mass spectrometry in an inter-laboratory comparison effort.
Jarosewich, E., Nelen, J.A. and Norberg, J.A. (1980) Reference samples for electron microprobe analysis. Geostandards Newsletters 4, 43-47. Logan, M. Amelia V. (2009) Trace element contents in the Smithsonian Reference Materials Collection: A program to continue Eugene Jarosewich's legacy. Goldschmidt Conference Abstracts 2009, A788. Mason, B. (1968) Eclogitic xenoliths from a volcanic breccia at Kakanui, New Zealand. Contr. Mineral. and Petrol., 19, 316-27.
Improving EPMA Standards and Standards Measurement by Updating Correction Parameters, Refining Measurement Procedures, and Accounting for Surface Coatings (Invited)
A modern, well-tuned electron microprobe is capable of measuring relative x-ray intensities with a precision of 0.1-0.2%, as determined by replicate measurements of the same sample and standard. Ideally, the measurement of a sample and standard quite close in composition should produce an answer whose accuracy approaches the measurement precision. Yet such levels of accuracy are exceedingly difficult to achieve, even within a factor of five of this. EPMA results, when samples and standards differ considerably in composition (or when the compositions of standard are imprecisely known), are commonly considered to have an analytical uncertainty of 3-5%. Many EPMA applications require better than that. The EPMA group at the Geophysical Laboratory has been using a common set of silicate standards for >20 years. Many of these standards were synthesized at the laboratory and are used by numerous other geological EPMA labs. We have two mounts of these standards and are in the process of measuring them vs. each other and from one mount to the other to determine their self-consistency and the degree to which we can decrease our analytical uncertainties. We are using multiple correction procedures (employing the CITZAF and Monte Carlo correction programs) and multiple accelerating potential (MAP) analysis to test the effects of the correction parameters and surface coatings on the measurements. Results so far indicate that a number of the standards differ to minor extents from their nominal compositions, that there are reproducible internal inconsistencies in calculated compositions depending on the correction algorithms employed, and that differences in carbon coat (and surface oxidation) thickness both within and between standard mounts significantly degrade the results. Re-estimation of standard composition (or replacement by other materials), upgrading of outdated correction parameters (particularly mean ionization potentials, ionization cross section estimates, and mass absorption coefficients), and utilization of layered specimen corrections to account for surface oxidation and carbon (or Au, Pt, Cr, Al…) coating thickness can reduce the analytical uncertainties by more than a factor of three. We will describe the methodology we employ, show results to date, and propose improved estimates of the composition of several of our widely used synthetic silicate standards.
Advances in Quantitative Analyses and Reference Materials Related to Laser Ablation ICP-MS: A Look at Methods and New Directions
The role of laser ablation ICP-MS (LA-ICP-MS) continues to expand both in geological sciences and other fields. As the technique continues to gain popularity, so too does the need for good reference materials and methods development and validation. Matrix matched reference materials (RMs) are required for calibration and quality control of LA-ICP-MS analyses. New advances in technology such as <200nm lasers and femtosecond lasers have reduced the dependence on matrix matching to some degree, but general matrix matching is still preferred. Much work has revolved around the available RMs such as the NIST 61x silicate glasses and several series of basaltic composition glasses such as the USGS natural basaltic glasses BCR-2g and synthetic basaltic glasses, the GS series (e.g. GSD-1g). While many quantitative hurdles have been recognized by analogous techniques such as EPMA and SIMS, some of these hurdles have not been fully addressed or validated for some cases of LA-ICP-MS. Trace element mapping by LA-ICP-MS is rapidly becoming more widespread for samples. Here relative differences in raw signal can be easily and rapidly obtained. However as too often is the case the magnitude of the relative differences in raw intensity are a function of different ablation yields, sample density or other factors. Methods of quantification for trace element mapping will be presented. The USGS has been developing microanalytical RMs intended for LA-ICP-MS for several years. The widely popular basaltic rock powders BCR-2, BIR-1 and BHVO-2 have all been successfully converted to homogeneous glasses suitable for LA-ICP-MS and have been in use by many workers. The newer synthetic basaltic glass GS series consists of 4 glasses of basaltic composition artificially doped at nominal concentrations of almost of trace elements at 400, 40, 4 and < 1 ppm. Additional developments in non-silcate or basaltic materials include the previously released MASS-1 Cu, Fe, Zn sulfide calibration RM (Wilson et al. 2002), the MACS-1 and MACS-3 Ca carbonate RMs and a prototype Ca phosphate RM. Other work in-house currently includes testing of additional sulfide materials (Fe and Ni sulfides) and a gypsum material. Data for several matrices and RMs will be presented using multiple laser wavelengths. For new methods development regarding quantitative analyses, we have developed several new methods for quantitative trace element mapping in a variety of mineral, biomineral and materials applications. Rapid trace element mapping in bones (Koenig et al. 2009) is not only quantitative for trace elements but provides data that would be difficult to obtain as quickly or accurately by EPMA or other techniques. A method has been developed for rapid mapping of trace elements in building materials and other complex rock materials using a modification of the sum to 100% method presented by others (e.g. Leach and Heftje, 2001). This paper will outline new methods of integrating imaging and analytical data from EPMA, SEM, Raman and other techniques that improve the utility, accuracy and overall science of the subsequent LA-ICP-MS. Additional new directions for quantitative analyses of fluid inclusions, tissues, minerals and biological samples will be discussed.
Sm-Nd isotopic compositions of LREE minerals for use as reference materials for in situ analysis by LA-MC-ICPMS
Recent work has demonstrated the possibility of obtaining both accurate and precise in situ Sm-Nd isotopic data in light rare earth enriched (LREE) accessory minerals including apatite, titanite, and monazite, using laser ablation-multicollector-inductively coupled plasma mass spectrometry (LA-MC-ICPMS). A distinct advantage of using LA-MC-ICPMS is that Sm-Nd isotopic data from these minerals can be determined in sub-grain domains potentially avoiding problems of isotopic mixing from inherited or xenocrystic components and allowing both valuable tracer isotope and geochronologic data to be obtained. However, a number of analytical obstacles complicate accurate Sm-Nd determination by LA-MC-ICPMS including mass bias corrections, the 144Sm isobaric interference on 144Nd, and potential offset (ca. 20-40 ppm) from thermal ionization mass spectrometry (TIMS) determination of similar materials. Thus, in order to verify Sm-Nd isotopic determination from unknowns, matrix-matched quality control standards (i.e., reference materials) must be developed to test the data handling protocol. This talk will present new Sm-Nd isotopic data determined by both TIMS as well as LA-MC-ICPMS of a number of natural potential reference minerals including Durango apatite, Fish Canyon titanite, Daibosatsu allanite, Trebilcock monazite, as well as a monazite from the Doi Inthanon core complex in northern Thailand. Our preliminary LA-MC-ICPMS results demonstrate that Durango apatite, Fish Canyon titanite, and Thailand monazite show both inter- and intra-grain homogeneity at current levels of precision (ca. 0.3-0.5 εNd) and close agreement with our TIMS data.
A new correction scheme and standards for the analysis of oxygen isotopes in garnet by ion microprobe (Invited)
Improvements in technique and instrumentation for analysis of oxygen isotopes by ion microprobe have dramatically increased analytical precision, creating the capability and need for better standardization. Accurate ion microprobe analysis of oxygen isotope ratios is possible only if appropriate standards are employed to correct for instrumental bias. In minerals with solid solutions, a component of the bias depends on the cation chemistry of the analyzed mineral; because garnets have a wide variety of solid solutions, a broad range of standards is required. Although the first δ18O(Gt) analyses by ion microprobe were made in Ca-rich garnets with variable Fe3+/Al ratios, at present the majority of published garnet standards are Ca-poor, and current Ca-rich standards yield conflicting results. Here we examine 13 existing garnet standards that span the compositional range of pyrope, almandine, grossular and spessartine, and introduce 14 new standards with variable Ca content including 6 standards along the grossular-andradite join. Bias due to cation composition in garnet is found to correlate with grossular content in pyralspite garnets and with andradite in ugrandite garnets. Instrumental bias is correlated with molar volume in garnets of all compositions in this study, however, there is substantial scatter about this linear relationship, particularly among grossular-rich standards. Although this correlation can be used as a correction scheme for bias, a more accurate method based on a 2nd-order polynomial relationship between X(grossular) in pyralspite or X(andradite) in ugrandite and bias is proposed. This correction reproduces instrumental bias in all but one of the 27 standards to within ±0.4‰. Thus accuracy approaches the spot-to-spot reproducibility of analyses (±0.3‰, 2 S.D.) of the homogeneous master garnet standard UWG-2. The new correction scheme successfully reproduces laser fluorination analyses along a traverse of a polymetamorphic, zoned skarn garnet from the Adirondack Mountains. Ion microprobe analyses resolve a gradient in δ18O of 2.1‰ over 16 µm. Assuming a 50 My period of post-skarn regional metamorphism, this step in δ18O constrains the diffusion coefficient of oxygen in garnet to < 10-26m2s-1 at 750C and low fH2O during granulite facies annealing.
Influence of Bulk Chemical Composition on Relative Sensitivity Factors for 55Mn/52Cr by SIMS: Implications for the 53Mn-53Cr Chronometer
The 53Mn-53Cr systematics of meteorite samples provide an important high resolution chronometer for early solar system events. Accurate determination of the initial abundance of 53Mn (τ 1/2=3.7 Ma) by secondary ion mass spectrometry (SIMS) is dependent on properly correcting for differing ion yields between Mn and Cr by use of a relative sensitivity factor (RSF). Ideal standards for SIMS analysis should be compositionally and structurally similar to the sample of interest. However, previously published Mn-Cr studies rely on few standards (e.g., San Carlos olivine, NIST 610 glass) despite significant variations in chemical composition. We investigate a potential correlation between RSF and bulk chemical composition by determining RSFs for 55Mn/52Cr in 11 silicate glass and mineral standards (San Carlos olivine, Mainz glasses KL2-G, ML3B-G, StHs6/80-G, GOR128-G, BM90/21-G, and T1-G, NIST 610 glass, and three LLNL pyroxene-composition glasses). All standards were measured on the Cameca ims-3f ion microprobe at LLNL, and a subset were also measured on the Cameca ims-1270 ion microprobe at the Geological Survey of Japan. The standards cover a range of bulk chemical compositions with SiO2 contents of 40-71 wt.%, FeO contents of 0.05-20 wt.% and Mn/Cr ratios between 0.4 and 58. We obtained RSF values ranging from 0.83 to 1.15. The data obtained on the ims-1270 ion microprobe are within ~10% of the RSF values obtained on the ims-3f ion microprobe, and the RSF determined for San Carlos olivine (0.86) is in good agreement with previously published data. The typical approach to calculating an RSF from multiple standard measurements involves making a linear fit to measured 55Mn/52Cr versus true 55Mn/52Cr. This approach may be satisfactory for materials of similar composition, but fails when compositions vary significantly. This is best illustrated by the ~30% change in RSF we see between glasses with similar Mn/Cr ratios but variable Fe and Na content. We are developing an approach that uses multivariate analysis to evaluate the importance of different chemical components in controlling the RSF and predict the RSF of unknowns when standards of appropriate composition are not available. Our analysis suggests that Fe, Si, and Na are key compositional factors in these silicate standards. The RSF is positively correlated with Fe and Si and negatively correlated with Na. Work is currently underway to extend this analysis to a wider range of chemical compositions and to evaluate the variability of RSF on measurements obtained by NanoSIMS.
A Synthetic Calcite Standard for Determination of Relative Mn/Cr Sensitivity Factor
Primitive chondrites which suffered from aqueous alteration often contain carbonates such as calcites, dolomites and breunnerites. 53Mn-53Cr decay system (half-life: 3.7 Myr) is applicable to dating their precipitation and many authors have measured their Mn-Cr ages using the SIMS. However, the relative Mn/Cr sensitivity factor: RSF (measured Mn+/Cr+ ratio divided by the true ratio) is not well established due to the absence of suitable standards, therefore the ages have systematic uncertainties. We prepared a synthetic Mn and Cr bearing calcite to evaluate the Mn/Cr RSF. Here we report the technical details of preparation for the standard and its Mn/Cr RSF. We also measured the Mn/Cr RSF of San Carlos olivine which is often assumed to be the same as that of a carbonate, and compared it with that of our synthetic calcite. The Cr-bearing calcite was produced in a N2 filled closed system by the reaction Ca2+ + CO32- = CaCO3 in an aqueous solution. The reaction proceeded by continuous addition of ammonium carbonate vapor to the solution. The crystal size of the calcite was ~300 μm. A small amount of hydrazine was added to the solution in order to keep chromous ion from oxidation. Mn and Cr concentrations in the calcite grains were determined by the SEM-EDS. A weak, defocused beam was used due to prevent electron beam damage. In a spherical grain, radial zoning of Mn and Cr concentrations occur and they decrease towards the periphery. At the center of the grains, Mn and Cr concentrations are ~0.5 atomic % and the values of the Mn/Cr ratios are relatively constant. The Mn/Cr RSF was determined with the CAMECA NanoSIMS 50 at Ocean Research Institute of Univ. of Tokyo. A primary O- beam of ~1 nA and 5 μm diameter was used. 43Ca+, 52Cr+, 53Cr+ and 55Mn+ ions were analyzed in a combined peak-jumping/multi-detection mode. The total measurement time was typically ~20 minutes. The measurements were started after presputtering of 15 minutes. The mass resolution power was ~3500, sufficient to resolve all relevant isobaric interferences. Several points near the center of calcite grains with constant Mn and Cr concentrations were measured. As for San Carlos olivine, a primary beam was rastered over 20 x 20 μm2 areas. Signals of all ions decreased with time during the measurements. The 55Mn/52Cr RSF also decreased with time. Initially it was ~0.7-0.8, and approached to a nearly constant value of ~0.5-0.6 after ~20 minutes. 53Cr+/52Cr+ isotopic ratio was 0.1105 ± 0.0002 (1σ), not dependent on time and space. This value corresponds to ~-26 permil fractionation from the reference value. On the other hand, the RSF of San Carlos olivine was found to be ~0.9, consistent with previous studies. The RSF of our calcite is significantly lower than the values used in previous studies and the RSF of San Carlos olivine measured in this study. This implies that Mn-Cr ages for carbonates obtained in previous studies may have systematic biases (~3 Myr).
U-Pb Homogeneity of Duluth Gabbro Baddeleyite from Microgram to Nanogram Scales
Baddeleyite has significant potential for U-Pb geochronology of mafic rocks, but due to small crystal sizes it can be exceedingly difficult to extract by conventional mineral separation techniques. We therefore developed in-situ dating of baddeleyite crystals with lateral dimensions between 5 and 20 μm (micro-baddeleyite) in polished petrographic thin sections using a CAMECA ims 1270 ion microprobe, and tested the homogeneity of a baddeleyite standard from Duluth gabbro complex over a wide range of grain sizes. Large (100 - 200 μm in diameter) baddeleyite crystals were separated from sample FC4-b from the Duluth gabbro complex and individually analyzed by isotope dilution thermal ionization mass spectrometry (ID-TIMS). Three FC-4b baddeleyite analyses overlap within error with a weighted mean 207Pb/206Pb date of 1099.6±1.5 Ma that closely agrees with published Duluth gabbro zircon dates. The weighted mean ID-TIMS 206Pb/238U date for FC4-b baddeleyite crystal separates (1096.8±0.3 Ma) is slightly younger than those for zircon. Large FC4-b baddeleyite crystals were also mounted along with pieces of polished thin-sections containing micro-baddeleyite and analyzed by ion microprobe using oxygen flooding to enhance sensitivity for positively charged Pb ions by a factor of ten. Ion microprobe 207Pb/206Pb ages for micro-baddeleyite (average 1096.9±2.6 Ma; MSWD = 1.2; n = 27) agree with the ID-TIMS age. With U-Pb relative sensitivities calibrated on FC4-b crystal separates, the weighted average 206Pb/238U micro-baddeleyite date is 1113±11 Ma (MSWD = 2.6; n = 27). This demonstrates that ion microprobe U-Pb baddeleyite analyses are unbiased by crystallographic orientation or grain size, and that 207Pb/206Pb and 206Pb/238U dates for Precambrian micro-baddeleyite are accurate and precise to within <0.3% and <2% relative uncertainty, respectively. For Phanerozoic samples, we anticipate similar 206Pb/238U age uncertainty if radiogenic yields are high. This opens new possibilities for dating silica-undersaturated rocks for which little alternatives exist.