- © 2016 Mineralogical Society of America
The highly siderophile elements (HSE) include the fifth-period transition metals ruthenium (Ru), rhodium (Rh), palladium (Pd), and the sixth-period transition metals rhenium (Re), osmium (Os), iridium (Ir), platinum (Pt) and gold (Au). In addition to being iron-loving, these elements are also resistant to oxidation, have high melting temperatures and are important as industrial catalysts. HSE abundances in geologic materials vary significantly, ranging from ~1 mg/g in ore materials down to a few pg/g in basalts (Table 1). These elements comprise two long-lived radiometric decay schemes: 187Re decays to 187Os, and 190Pt decays to 186Os.
The HSE have been targeted to address a wide variety of geochemical and cosmochemical questions. Early work suggested HSE concentrations can constrain Hadean mantle evolution (Chou 1978) and showed the geochronologic potential of the Re/Os isotope system (Herr and Merz 1955; Herr et al. 1961; Markey et al. 1998). More recent applications combine the Re–Os decay system with abundance data for the HSE to investigate the evolution of the planets and the moon (Day et al. 2010, 2016, this volume), the terrestrial mantle (Rehkämper et al. 1997; Aulbach et al. 2016, this volume; Harvey et al. 2016, this volume; Luguet and Reisberg 2016, this volume), impact craters (Koeberl and Shirey 1993), geochemistry and geochronology of ore formation (Markey et al. 1998; Barnes and Ripley 2016, this volume), tektites (Koeberl and Shirey 1993), as well as the formation and evolution of the continental curst (e.g., Peucker-Ehrenbrink and Jahn 2001). Non-mass dependent isotope variations in Re, Os, Ru, Pt and Pd are also present in some meteorites and lunar samples and arise from nucleosynthesis and cosmogenic radiation (Yokoyama and Walker 2016, this volume).
The chemical characteristics of the HSE that make these metals so useful in industry also dictate the methods by which they can be separated for chemical analysis. The siderophile behavior of these metals has long made fire assay the preferred technique, not only for production, but also for concentration determinations. The extremely low abundances of the HSE in natural samples and the diversity of applications and matrices has led to a variety of measurement procedures for their quantification. This chapter provides an overview of existing measurement procedures for bulk samples, including digestion procedures, purification protocols, and mass spectrometric analysis, as well a discussion of reference materials. Laser ablation ICP-MS is used to study the spatial distribution of HSE on the μm scale, not for bulk sample analysis. The preparation of sulfides for LA-ICP-MS and the merits and pitfalls associated with the technique are discussed separately in Harvey et al. (2016, this volume).
DATA QUALITY CONSIDERATIONS FOR THE HSE
Sample heterogeneity and reproducibility
The terminology used to describe the quality of measurement results is defined in the appendix. Mining applications typically analyze samples with HSE contents of 0.1–2 μg/g and use 10–100 g sample test portions; geochemical applications typically use sample sizes of 0.5–3 g. With these smaller test portion sizes, heterogeneous distribution of mineral hosts for HSE in samples can compromise both intermediate measurement repeatability and measurement reproducibility (Meisel et al. 2001a). This so-called “nugget effect” arises because HSE in terrestrial rocks are often hosted not in the major silicate minerals but instead in trace minerals, including sulfides and alloys, that may not be homogeneously distributed throughout a finely ground sample (Harvey et al. 2016, this volume; Lorand and Luguet 2016, this volume; O’Driscoll and González-Jiménez 2016, this volume).
Sample heterogeneity is expected to increase with decreasing test portion size, which can be expressed by the Ingamells sampling constant (Ingamells 1974). If high variance in the replicated analyses of HSE results only from sample heterogeneity and not from some other analytical issue, then the sample sizes used for analysis were too small. Whether HSE are homogeneously distributed at the preferred sample size ideally should be evaluated for each matrix, taking into account the possibility of other contributors to uncertainty such as incomplete dissolution, contamination or spectral interferences. Poor repeatability often can remain undetected, because the sample preparation is time consuming and sample amounts may be so limited that intermediate measurement repeatability remains unknown. As a consequence, a single analysis of a sample makes the implicit, but possibly inaccurate, assumption that it provides accurate data.
Knowledge of the geochemical distribution of HSE in minerals and metals can assist in the interpretation of variations in replicate analyses. Distributions of HSE in minerals and rocks are discussed more fully for peridotites in Aulbach et al. (2016, this volume), Becker and Dale (2016, this volume), Luguet and Reisberg (2016, this volume); for basalts in Gannoun et al. (2016, this volume); for sulfides in Harvey et al. (2016, this volume); for platinum group minerals in O’Driscoll and González-Jiménez (2016, this volume); and for magmatic ore deposits in Barnes and Ripley (2016, this volume). For example, when 1–3 g sample sizes of fertile peridotite are digested, measured Os abundances might show high variance, while Re concentrations are more or less constant. This result would be expected if Os were hosted in trace alloys or other HSE-rich minerals and Re hosted in silicates and base metal sulfides. Alternatively, Meisel et al. (2001a) showed that replicate analyses of Re in a komatiite showed scatter while the Os concentration data remained constant and the Re–Os isotope systematics provided an isochron of meaningful age. In this case, heterogeneous distribution of Re-rich molybdenite (MoS2) probably caused the variation in Re and 187Os abundance while the bulk of the non-radiogenic Os was homogeneously distributed. In this instance, sample heterogeneity hampered the determination of the “true” HSE contents; nevertheless, correct age information may still be obtained if both parent and daughter isotopes are completely liberated from replicate samples portions (Ishikawa et al. 2014).
A comparison of Pt data for two reference materials, UB-N (a lherzolite) and MUH-1 (a harzburgite) shows how different matrices manifest the nugget effect (Fig. 1). Platinum concentrations in UB-N show a relatively small variation, in contrast to MUH-1 which shows marked larger variation and skews towards higher concentrations. Harzburgites, in contrast to lherzolites, typically are sulfur-poor, and the major hosts of HSE are alloys and HSE sulfides, such as laurite (Ru, Ir, Os)S2. These HSE-rich phases can be difficult to completely dissolve in aqua regia (or other mixtures of HCl and HNO3), as discussed further in a later section. These contrasting datasets suggest that, in addition to incomplete recovery of HSE during dissolution, representative subsampling of the powdered sample is hampered by the nugget effect. These points demonstrate that validating a particular measurement procedure requires the use of matrix matched reference materials.
Isotope dilution determinations
The isotope dilution method provides concentration data having the smallest measurement uncertainty and is critical for determination of the parent/daughter ratio for use with radiogenic isotope systems. In this technique, a known amount of isotopically enriched “spike” of the element of interest is added to carefully weighed sample powder, along with dissolution acids. Complete equilibrium between the spike and the sample must be attained during dissolution, after which the element of interest is chemically purified. The resulting isotopic composition of the element is measured precisely by mass spectrometry, then used to calculate the concentration of the element in the sample. The isotope dilution equation is derived in Heumann (1988). As long as spike and sample equilibration is attained, complete recovery of the element during chemical purification is not required to achieve accurate concentration data. In geologic samples and in acidic solutions, the HSE typically are stable in multiple valence states and molecular forms, which makes attainment of equilibrium between spike and sample a challenge. Chemical purification of some of these analytes, in addition, often yields less than perfect recoveries, particularly for Pt, Pd, Ir, and Ru. Both of these points are further discussed below.
Which isotope of a given element is chosen for the spike depends on a combination of qualities: its relative artificial enrichment, its abundance in natural samples, the absence of mass interferences, and its commercial availability. All modern isotope dilution studies use 190Os, 185Re and 191Ir; nearly all use 99Ru. None of the isotopes of Pt and Pd provide an ideal spike; as a result, different laboratories utilize 194Pt, 196Pt and 198Pt; and 105Pd, 106Pd, and 108Pd and 110Pd. Examples of spikes used is various labs are described in (Pearson and Woodland 2000; Meisel et al. 2001b; Puchtel et al. 2007; Fischer-Gödde et al. 2010). For ease of use and better precision of concentration ratios among the HSE, a mixed, calibrated spike containing all or some of the HSE usually is prepared. Rhodium and Au are monoisotopic and thus cannot be analyzed by isotope dilution. High quality concentration data can still be obtained, however, by combining high and known recovery of those elements after chemical purification and internal and external standardization (Meisel et al. 2003a; Qi et al. 2004; Fischer-Gödde et al. 2010; Savard et al. 2010). These standardizations yield uncertainties that are typically two to three times larger than those for elements measured by isotope dilution.
Digestion of geologic samples
Two dissolution procedures dominate the scientific literature for the determination of bulk HSE abundances and isotopic compositions of geologic samples: fire assay and acid digestion.
For the bulk analysis of ores and other samples with high abundances of HSE, Ni sulfide fire assay with Te co-precipitation of the HSE is widely used, primarily in the mining and exploration industries,. In the fire assay technique, the sample powder (up to 100 g) is fused at around 1000 °C and the HSE are concentrated in a NiS bead (Savard et al. 2010). Use of large quantities of Ni metal and alkali salts (Na2CO3, Na2O2, Li2B4O7) during fusion, relative to the test portion size, gives high blank levels for the HSE. It is not possible to analyze Re or Au by NiS fire assay as they are partially volatilized during processing (e.g., Savard et al. 2010). Whether the HSE are quantitatively recovered during fire assay is debated (e.g., Puchtel and Humayun 2005; Savard et al. 2010). This main advantage of this technique is in its ability to digest up to 100 g of sample which can minimize the effects of sample heterogeneity.
Acid digestion at elevated temperatures and pressure is more frequently used for dissolution of samples having normal crustal or mantle abundances. Because HSE have high electrochemical redox potentials, most recent work utilizes the high oxidative power of aqua regia (3:1 ratio by volume of concentrated HCl and HNO3) or mixtures of HCl and HNO3 either in a volume ratio of 5:2 or 5:3 (sometimes described by “inverse” or “reverse” aqua regia) for dissolution. Acid digestion of HSE-bearing minerals and alloys in geologic samples, in particular those rich in Rh, Ru, Os and Ir, cannot be achieved at the boiling point of aqua regia under atmospheric conditions. The highly resistant metals and alloys are best attacked in a closed system at elevated temperatures (> 200 °C) under high pressure conditions (> 50 bar) in sealed borosilicate or quartz glass containers. For isotope dilution analyses, a mixed isotopic spike is added to the sample powder along with the dissolution acid. Both high temperature and closed system dissolutions are necessary for full isotopic equilibration between sample powder and the isotope spike (Shirey and Walker 1995).
One technique uses Carius tubes, a borosilicate or quartz glass tube with a narrow neck for easy sealing with a torch. Carius tubes were originally developed for elemental analysis (C, H, O, N, S, P) of organic substances (Carius 1860, 1865), described for digestion of platiniferous materials (Gordon 1943; Gordon et al. 1944), then applied to Re-Os isotopic analysis (Shirey and Walker 1995). This technique does not involve expensive instrumentation, but requires some skill with a torch to produce leak-proof seals. Most studies use borosilicate Carius tubes, but modified cleaning techniques for borosilicate and quartz Carius tubes can be used to better assure lower blanks for Pt (Day and Walker 2016). Up to 3 g of finely ground sample powder is reacted in a Carius tube in reverse aqua regia. Each Carius tube is placed in a steel protection vessel prior to heating in an oven at 220–345 °C for up to several days. Use of temperatures above about 270 °C usually requires an air-tight protection vessel and counterpressure supplied by dry ice (e.g., Becker et al. 2006). After heating, the Carius tubes are cooled to room temperature then chilled in a mixture of dry ice and alcohol, or in ice, to minimize overpressure during opening. Tubes are opened behind a protective shield.
The other method of acid digestion uses a high pressure asher (HPA, Anton Paar, Graz) which employs reusable quartz glass containers to maintain high temperature and high pressure under controlled conditions. Up to 5 g of sample powder can be treated in reverse aqua regia (HNO3 and HCl volume ratio 5:2) in the HPA vessels at 300–320 °C maintained for several hours. However, the manufacturer recommends not to use the HPA for prolonged periods above 280 °C when HCl is present. Some level of user skill is required to assemble the quartz tubes so that a leak-proof seal is produced. The main drawbacks to this dissolution technique are the high initial expense of the HPA and maintenance expenses caused by corrosion of the autoclave and tubing in the HPA by HCl fumes. This system, however, can yield especially low blanks for HSE (Meisel et al. 2001b).
Low-temperature acid attack utilizing concentrated HBr and HF in closed Teflon®PFA developed by the Paris group (Birck et al. 1997) shows severe underestimation for Os results for whole rock analysis as shown by Meisel et al. (2003b) but can serve as low blank alternative for individual base metal sulfide and aggregates of silicate grains.
Systematic studies comparing results using different digestion methods and conditions have yielded contradictory results. For example, Meisel et al. (2003b) found that for a serpentinized peridotite reference material, UB-N, temperatures between 230 to 240 °C, attained in Carius tubes, were insufficient to completely digest its HSE-bearing spinels and HSE alloys; the highest yields of HSE were obtained at temperatures of ≥ 300 °C in the HPA. By contrast, Puchtel and Humayun (2005), Harvey et al. (2010) and Ishikawa et al. (2014) found little difference between HPA results for UB-N and those obtained by dissolution in Carius tubes at 230 °C. Other rock types yield different comparative results. For example, Becker et al. (2006) found that spinel grains from fresh peridotites remained undigested at 220–240 °C in Carius tubes but were completely dissolved or disaggregated at 345 °C. The latter protocol utilized counter pressure to achieve the highest temperature in Carius tubes. The contrast in these results with those for UB-N suggests that “the process of serpentinization may be helpful in pre-digesting peridotites” (Becker et al. 2006).
At least two studies indicate that some basalts may benefit from treatment with HF after aqua regia dissolution. Dale et al. (2012) showed that de-silicification with HF, after dissolution using aqua regia in a HPA, more efficiently extracted Ir, Ru Pt, Pd and Re. Basalt reference material TDB-1 was used by Ishikawa et al. (2014) to study the effectiveness of different digestion techniques. Improved results in TDB-1 were obtained by additional treatment with hydrofluoric acid after Carius tube processing, a result confirmed by the Na2O2 sintering technique (Bokhari and Meisel 2014) (Fig. 2). In TDB-1, prolonged heating in aqua regia at elevated temperatures also completely releases Re from the silicate minerals, without additional treatment in HF, but does not completely release Ru (Figs. 2 and 3). Like TDB-1, the use of HF during dissolution appears essential to fully release all Ru in some basalts (Fig. 3). These results suggest that some proportion of HSE in some basalts is hosted in phases not accessed by aqua regia. On the other hand Day et al. (2015) showed that the use of HF can alter measured Re/Os and Pt/Os ratios in some high-MgO basalts and increase blanks. Thus, the effectiveness of the digestion technique of choice needs to be validated by applying different techniques to the samples of interest and by the use matrix matched reference materials (see later).
Chemical separation of HSE
For samples from which more than about 100 ng of HSE are obtained, interferences and matrix effects become less important during ICP-MS analysis. In this case direct ICP-MS measurement may be possible without further chemical isolation of these elements. For lower abundances and for more interference-free analysis, preconcentration of the HSE is necessary. Isolation of Os is usually by solvent extraction from aqua regia into CCl4 or CHCl3 and back extraction into HBr (Cohen and Waters 1996). In this process, Os, oxdized as OsO4, is extracted from aqua regia through liquid-liquid extraction into the solvent by shaking. This extraction is repeated for a total of 3 times (as each step extracts about 60–70 %), with the Os-containing solvent removed to a clean beaker after each extraction. Osmium is then recovered from the solvent by addition of concentrated HBr which is shaken and emulsified. During this step, Os is reduced and partitioned into HBr. After removing and discarding the solvent, the HBr solution containing the Os is dried. Alternatively, OsO4 can be distilled directly from the solution after acid digestion (Morgan and Walker 1989; Nägler and Frei 1997). Further cleanup of Os is accomplished by microdistillation (Roy-Barman 1993; Birck et al. 1997). The Os is transferred to the cap of a 5 mL Teflon®PFA conical beaker, covered with in a solution of CrO3 in 6 mol/L H2SO4. A small drop of HBr (20 μL) is placed in conical tip of the beaker. The beaker is capped in an upside down position and heated to 80 °C for 2–3 h. During the microdistillation, the Os is oxided by the Cr solution, volatilized and trapped in the HBr. The solution of Os in HBr can then be dried to 1–2 μL, and is ready for mass spectrometry.
For purification of the remaining HSE, each laboratory appears to have developed its own unique procedure. Two general approaches are used: separations using anion exchange resin (e.g., Rehkämper and Halliday 1997; Pearson and Woodland 2000; Meisel et al. 2001b; Horan et al. 2003; Chu et al. 2015) or using cation exchange resin (e.g., Meisel et al. 2003a; Fischer-Gödde et al. 2010). The HSE typically form anion complexes in HCl (most HSE) or HNO3 (Re). Use of anion exchange resin, typically AG1-X8, and high molarity acids allows separation of HSE into groups, such as Re + Ru and Pt + Ir, that are especially well-suited for analysis by magnetic sector ICP-MS. Potential isobaric interferences on Ru from chromium oxides can be minimized by treatment of the sample with H2O2 prior to anion exchange chromatography (e.g., Dale et al. 2012). The main drawback to separation of the HSE + Re by anion exchange is that some HSE, particularly Pd, may have relatively low recoveries.
In the other approach, the HSE are little adsorbed by cation exchange resin and therefore elute quickly as a group in weak acid, while the major and most trace elements are retained on the resin. This technique can be applied off-line in batches or on-line by coupling the column directly to the ICP-MS. In both cases near-quantitative recovery of the HSE is obtained, and therefore can be modified for analysis of unspiked monoisotopic Rh and Au (Meisel et al. 2003a; Qi et al. 2004; Fischer-Gödde et al. 2010). Interferences from Cd and Zr must be carefully monitored or separately removed. For cosmochemical applications in which the isotope compositions of the HSE are to be measured to ɛ-level (1 part in 10,000) or better precision, additional purifications steps are necessary (Yokoyama and Walker 2016, this volume, and references therein).
Quantification by mass spectrometry
Osmium is usually measured as OsO3− by negative thermal ionization mass spectrometry (N-TIMS) on Pt-filament material using Ba(OH)2 as an electron emitter (Creaser et al. 1991; Völkening et al. 1991). Alternatively, OsO4 may be sparged from a solution directly into an ICP-MS (Hassler et al. 2000).
Quantification of the other HSE is by ICP-MS, either single collector analysis using a magnetic sector field (ICP-SFMS) or quadrupole (ICP-QMS) mass spectrometer (e.g., Peucker-Ehrenbrink and Jahn 2001; Meisel et al. 2003a; Wang and Becker 2014) or by magnetic sector multicollector ICP-MS (e.g., Puchtel et al. 2007). The ICP-SFMS and ICP-QMS allow the possibility of measuring all of the HSE in a single aliquot. Magnetic sector mass spectrometers such as the single collector ICP-SFMS “ELEMENT” and MC-ICP-MS (e.g., Nu Plasma and Thermo Neptune) offer higher sensitivity analysis. For isotope dilution analysis, the high mass range elements, i.e., Ir, Os, Re, Pt, and Au are nearly interference-free. Lower mass HSE (Ru, Rh, and Pd) have spectral interference either from molecules produced by the argon plasma, doubly charged species, or molecular species of elements inadequately removed during chemical purification (e.g., Meisel et al. 2001b; Puchtel and Humayun 2005; Fischer-Gödde and Becker 2012). Higher precision measurement results of the full isotope compositions of the HSE for nucleosynthetic or cosmogenic applications require more stringent monitoring of interferences are discussed in the references cited in Yokoyama and Walker (2016, this volume).
The need for lower analytical blanks is driven by the interest in the HSE distribution in geologic samples having HSE levels at sub-ng/g levels and also for the study of individual grains. The HSE belong to the least abundant group of elements in the Earth’s crust, but may be elevated through anthropogenic influence on the environment, for example by the use of automobile catalytic converters. Even in a dust-free environment, contamination occurs from mineral acids, alkaline flux, resins, metals, and glassware used during sample preparation. The relatively low concentrations of HSE in most meteorites and terrestrial rocks require the use of highest-purity reagents. Sub-pg/mL blank levels of HSE in mineral acids usually can be achieved by subboiling distillation.
Total procedural blanks for solvent extraction of Os of less than_a few pg are typical, with an 187Os/188Os ratio of 0.15–0.20 (e.g., Puchtel et al. 2007; Dale et al. 2012; Wang and Becker 2014). A major contributor this blank is Os in HNO3 that is not removed by subboiling distillation. Further purification of HNO3 can be achieved by careful treatment with H2O2 (R. Creaser, personal communication). Procedural blanks levels for Re are similar to those for Os, while those for Pt, Pd, and Ru may be higher (e.g., Puchtel et al. 2007; Dale et al. 2012; Wang and Becker 2014). Chromatographic resin can be a major contributor to these blanks; anion resin should be cleaned using high molarity HNO3 and/or HCl before use. For samples having low abundances of HSE, the contribution of the procedural blank to their measurement uncertainties cannot be neglected. Figure 4 shows how the procedural blank and its own uncertainty is propagated into the resulting uncertainty for sample Os concentrations at various levels. For details see Moser et al. (2003). The uncertainty in Os concentrations at the lowest levels can be improved not only by a decrease in the absolute amount of the procedural blank, but also by reduction in the variance of the average of the blank determinations.
Some borosilicate Carius tubes have been found to have high-Pt blanks; more aggressive cleaning procedures (e.g., Puchtel et al. 2008) or substitution of quartz Carius tubes (e.g., Day et al. 2010) can help. The lowest and most reproducible blanks for the HSE can often be obtained by dissolution using the HPA. In the analysis of HSE for Apollo lunar samples, Day et al. (2007) pointed out that Pt was the element that showed the most discrepancy between HPA and Carius tube dissolutions, possibly resulting from underestimation of Pt-blank variability for the latter method. They concluded that Pt-blank measurements for the HPA were highly reproducible.
REFERENCE MATERIALS FOR HSE ANALYSIS
Reference materials (RM) play a key role in method development and for quality control in routine analyses. Further, they can serve as calibration standards and can be used to establish a traceability chain (BIPM et al. 2008). Reference materials should be homogenous, meaning that the contribution of sample heterogeneity must be very small compared to the overall measurement uncertainty. The homogeneity of a RM can only be guaranteed for sample sizes that are larger than some minimum sample size. Such a minimum should be given in a certificate of analysis for a reference material. HSE abundance and isotopic data for RM ideally should have small uncertainties in order to better resolve results produced by different measurement procedures or in different laboratories. This goal is currently only achievable with relatively easy- to-digest and relatively homogeneous materials such as fertile peridotites.
Certified reference materials produced following ISO guides are widely used in commercial gold, silver, palladium, and platinum analysis. Examples include WMG-1 (mineralized gabbro, NRCan), SARM-7 (Merensky Reef platinum ore, SA Bureau of Standards). In contrast, few well-characterized and matrix-matched reference materials exist for samples having non-commercial levels of HSE. We briefly discuss some of these low-level HSE RM used in the geochemistry community.
Mafic reference materials include TDB-1 (basalt) and WGB-1 (gabbro) which have been certified for Pd, Pt, and Au concentrations (NRCan 1994a). These materials were characterized during method validation studies and are now among the best studied non-mineralized, low abundance RM with a silicate matrix (Enzweiler et al. 1995; Plessen and Erzinger 1998; Meisel et al. 2001b; Bédard and Barnes 2002; Qi et al. 2003; Meisel and Moser 2004a; Qi et al. 2004; Boulyga and Heumann 2005; König et al. 2012; Li et al. 2013).
Icelandic basalt BIR-1, a coarse-grained olivine tholeiite provided by the USGS, has proven to be homogeneous for HSE even for test portion sizes of less than 2 g (Ishikawa et al. 2014) The USGS RM basalts BHVO-1, BHVO-2 and BCR-2 have also been characterized for HSE concentrations (Meisel and Moser 2004a; Li et al. 2013; Chu et al. 2015) but fewer systematic studies of digestion techniques are available.
An initiative to develop a komatiite sample, KAL-1 from a single Alexo lava flow, was unsuccessful, as only a small amount of material was produced and differences in the HSE distribution between batches became apparent (J. Carignan, CRPG, personal communication). Instead, an existing komatiite RM named OKUM collected from Abitibi, Canada, by the Ontario Geological Survey is being characterized (Wang and Becker 2014). More than 100 kg of material were specially prepared for the certification of major and trace elements by the International Association of Geoanalysts (IAG) following ISO Guidelines, and is being distributed to laboratories for development as a RM for HSE.
Lherzolites are especially useful as RM for HSE. The bulk of the HSE budget of fertile peridotites, i.e., sulfur-rich lherzolites, typically is hosted in base metal sulfides. These sulfides are easily dissolved during acid digestion of the peridotites. Lherzolite RM such as UB-N and GP13 have been well characterized for HSE contents, as well as for their 187Os/188Os compositions (Pearson and Woodland 2000; Meisel et al. 2003b, 2004; Pearson et al. 2004; Day et al. 2012). GP13, a fertile lherzolite from the Beni Bousera massif, Morocco, was initially developed as in-house RM but is now no longer available. UB-N, a serpentinized, fertile lherzolite, is still available from the SARM at the CRPG but only in coarse grained batches that are further pulverized to powder on demand. Differences in Cr content and in HSE contents are present in different batches, probably as a result of variations in the spinel abundances (J. Carignan, CRPG, personal communication; H. Becker, FU Berlin, personal communication).
Satisfactory RM for harzburgites are lacking. A harzburgite RM, MUH-1, from the Preg Quarry, Kraubath, Austria, was prepared for the certification of major and trace elements, and is being distributed for interlaboratory comparisons of HSE and Os isotopic compositions. This sample, however, appears to be subject both to incomplete recovery of HSE during dissolution and the nugget effect at 5 g sample size, as discussed in an earlier section.
Table 2 gives an overview of RM that have been used for method validation and quality control. Several of these RM were not developed for the intended use as HSE RM. As such more RM that have been analyzed for HSE content but were not included in this table as too few data have been published.
Terminology and definitions
A reported result without quantitative indication of its quality is of limited use, and makes comparison among samples difficult and evaluation of results obtained by different methods or in different labs complex. A common language for scientific measurements has been defined in the VIM3 (BIPM et al. 2008). Because some of these terms are little used in the geologic community, here we provide definitions and recommendations to avoid misunderstandings of metrological concepts and make data easier to compare. Italicized text are taken from VIM3 (BIPM et al. 2008) unless otherwise noted.
“Quantity to be measured”. For example the measurand can be defined as: (a) The mass fraction of Pt in a 100 mg test portion; (b) The mass fraction of Pt in the laboratory sample; (c) The mass fraction of Pt in particular geological formation. “Analyte”, by contrast, refers to the name of a substance or compound.
“The closeness of agreement between indications or measured quantity values obtained by replicate measurements on the same or similar objects under specified conditions, usually expressed numerically by measures of imprecision, such as standard deviation, variance or coefficient of variation.” While the term precision is correctly used within the earth science community, the conditions under which the measurement precision was obtained also should be reported. These conditions can be, for example, repeatability conditions of measurement, intermediate precision conditions of measurement or reproducibility conditions of measurement (see ISO 5725-3:1994 and vide infra ). The most common kinds of precision are listed below and shown in Appendix Figure 1.
Measurement precision obtained using the same operators, same measuring system, same operating conditions and same location, and replicate measurements on the same or similar objects over a short period of time. This term is preferred over “internal reproducibility” or “internal precision”.
Intermediate measurement precision
This term includes the same measurement procedure, same location and replicate measurements on the same or similar objects over an extended period of time. It may include other changes, including new calibrations, calibrators, operators and measuring systems, and should specify which conditions are changed and unchanged, to the extent practical. This term is preferred over “external reproducibility” or “external precision”. Appendix Figure 1a and b show intermediate measurement precisions obtained in two laboratories for the Ir content of the komatiite powdered reference material OKUM.
Measurement precision under reproducibility conditions that includes different locations, operators, measuring systems, procedures and replicate measurements on the same or similar objects. This precision can usually only be approximated, and will be the largest all types of precision. It is thus the opposite of measurement repeatability. Information on the measurement conditions that changed and remained unchanged, to the extent practical, should be provided. Appendix Figure 1c shows the reproducibility precision for Ir in OKUM, as measured in 7 laboratories.
This term refers to the “closeness of agreement between a measured quantity value and a true quantity value of a measurand.” Measurement accuracy is not a numerical quantity, but a measurement is said to be more accurate when it offers a smaller measurement error.
Quantification and detection limit
The use of the term “detection limit” is not preferred, as its usage is inconsistent and provides little information on the actual uncertainty of a given measurement. More useful information is given by the “quantification limit” which has no rigorous definition, but is generally defined as the lowest limit above which the measurement uncertainty of the result is fit for purpose.
This term refers to “the dispersion of the values that could reasonably be attributed to the measurand” (BIPM et al. 1993). Measurement uncertainty (Appendix Fig. 2a and b), quantified by standard deviations, comprises the statistical distribution not only of a series of measurements, but also includes other components that arise from probability distributions based on experience or other information (BIPM et al. 1993). A best estimate for the value of the measurand, therefore, includes a measurement uncertainty that incorporates all components of uncertainty that contribute to the dispersion, including systematic effects, for example those associated with corrections and reference standards, (BIPM et al. 1993).
An older but less general definition of measurement uncertainty is “an estimate characterizing the range of values within which the true value of a measurand lies” (BIPM et al. 1984, definition 3.09). This definition is not inconsistent with the latest VIM3 definition but it makes clear that the measurement uncertainty is actually a range of values large enough to encompass the true, but unknown, value (Appendix Fig. 2c). Thus, information on precision and accuracy is not sufficient as it does not provide a range. More information on the correct use of “measurand”, “error” and “uncertainty” is given in Annex D of the Guide in estimation uncertainty (BIPM et al. 2008).
This term refers to a measured quantity value minus a reference quantity value (Appendix Fig. 2b), and is only applicable in the rare case when the reference quantity value has a negligible measurement uncertainty.