Re–Os geochronology for sulfides and organic-rich sediments

ABSTRACT

Rhenium and osmium are both siderophilic and chalcophilic, exhibiting a strong affinity for organic-rich materials. This makes the Re–Os chronometer a valuable complement to geochronometers based on lithophile elements. In this review, we begin by discussing how the elemental abundances and isotopic compositions impact sample selection, analytical strategy, and data interpretation. We then provide an overview of how ¹⁸⁷Os/¹⁸⁸Os ratios can be used to trace geological processes, followed by a summary of the analytical protocols commonly used in Re–Os geochemistry. We also examine key challenges in isochron dating, including the identification and avoidance of pitfalls such as mixing lines, and inherited initial slopes. We further demonstrate that petrographic and geochemical studies can be very helpful for accurately dating sulfides with contrasting initial ¹⁸⁷Os/¹⁸⁸Os values and/or ages. With state-of-the-art Re–Os dating technique reaching precisions up to 0.05% for molybdenites and 1% for organic-rich sedimentary rocks, it is now possible to resolve the rapid and episodic nature of ore formation, and to investigate the dynamics of environment–life coevolution with unprecedented detail. We conclude this review by outlining future directions for Re–Os geochronology, including developing imaging-guided Re–Os dating techniques for organic-rich sediments, sharpening the in situ Re–Os dating method, and fully integrating the Re–Os geochronometer into the EarthTime initiative.

The Re–Os isotope system holds Earth's memory in deep-time—from the heartbeats of ore formation to the pathways of hydrocarbons, from climate change to life evolution, from hot magmas to shining meteorites. This review explores how Re–Os isotopes trace these processes across Earth's mantle, crust and oceans.

INTRODUCTION

Rhenium (Re) and Osmium (Os) are siderophile (iron-loving) and chalcophile (sulfur-loving) elements that additionally exhibit a strong affinity for organic-rich materials. As such, Re and Os are commonly enriched in sulfide minerals, organic-rich sedimentary rocks, hydrocarbons (e.g. crude oil, bitumen), coal and graphite [1–10]. These characteristics distinguish the Re–Os system from more commonly used radiogenic chronometers, such as Rb–Sr, Lu–Hf, Sm–Nd and U–Th–Pb, which primarily involve lithophile elements hosted in silicate minerals.

Although Os was discovered in 1803 [11], Re was not identified until over a century later, in 1908 [12]. Consistent determination of the ¹⁸⁷Re half-life was only achieved in the 1990s [13]. As one of the most recently developed radiogenic chronometers, the Re–Os system has proved to be a valuable addition to the geochronological toolbox, offering a complementary perspective to traditional methods based primarily on lithophile elements.

The siderophilic nature of Re and Os leads to their preferential partitioning into Earth's core during planetary differentiation (Fig. 1). Approximately 99.86% of Re and 99.94% of Os are sequestered into the core (Table 1), resulting in significant depletion of both elements in the bulk silicate Earth (mantle and crust) [14–17]. The distribution of Re and Os between the mantle and crust is primarily controlled by their contrasting geochemical behaviors during partial melting [18,19]. Rhenium behaves incompatibly, preferentially entering the melt phase, whereas Os is compatible and largely remains in the residual crystalline phases [19]. Over time, this leads to significantly higher Re but lower Os concentrations in the continental crust (∼0.7 ppb Re and ∼0.05 ppb Os; Fig. 1, Table 1) compared to the depleted mantle (∼0.28 ppb Re and ∼3.4 ppb Os). Due to the elevated Re/Os ratio of the crust relative to the mantle (∼14 vs ∼0.08, Fig. 1), the β-decay of ¹⁸⁷Re to ¹⁸⁷Os leads to progressively more radiogenic ¹⁸⁷Os/¹⁸⁸Os ratios in the continental crust over time, making ¹⁸⁷Os/¹⁸⁸Os a powerful tracer for geological processes.

Figure 1.

Distribution of Re and Os in Earth's core, mantle and crust. Nearly all Re and Os (99.86%–99.94%) are sequestered into Earth's core, leaving a Re and Os depleted bulk silicate Earth. During mantle partial melting, incompatible Re preferentially enters into the melt, resulting in a Re-enriched and Os-depleted continental crust (∼0.7 ppb Re and ∼0.05 ppb Os) and an Os-enriched and Re-depleted mantle (∼0.28 ppb Re and ∼3.4 ppb Os). Note that the Re and Os abundances are in ppm levels for the core, but in ppb levels for the mantle and crust. See Table 1 for data.

Table 1.

Re and Os abundances in the Earth's core, mantle and crust, and the isotopic composition of commonly dated materials.

	Re abundance	Os abundance	Mass (tonnes)	Re budget (tonnes)	Os budget (tonnes)	187Re/188Os
Core	∼0.70 ppm [16,17]	∼3.0 ppm [16,17]	1.9 × 1021 [16]	1.4 × 1015	5.8 × 1015
Mantle	∼0.28 ppb [17]	∼3.4 ppb [17]	4.0 × 1021 [16]	1.1 × 1012	1.4 × 1013
Crust	0.40–1.0 ppb [14]	5.0 × 10−2 ppb [15]	3.0 × 1019 [16]	2.1 × 1010	1.5 × 109
Molybdenite	4.2 × 10−4 –1.0 × 104 ppm					∞
Pyrite	1.2 × 10−3 –4.6 × 104 ppb					3.5 × 10−1–2.7 × 106
ORS	3.0 × 10−1–1.4 × 103 ppb					1.1 × 100–6.3 × 103
Crude oil	1.0 × 10−2–1.5 × 103 ppb					1.5 × 100–1.1 × 104
Bitumen	3.0 × 10−2–4.4 × 102 ppb					2.5 × 101–1.8 × 103

Data sources for molybdenite, pyrite, organic-rich sediments (ORS), crude oil and bitumen are provided in Table S1.

The Re–Os chronometer has been successfully applied to a wide range of materials from both the Earth and the solar system, including meteorites [13], sulfide minerals [6,8,9,20–26], organic-rich sedimentary rocks [27–30] and even hydrocarbons [31,32]. These applications have produced a wide range in Re–Os ages, spanning from as old as 4.56 Ga for meteorites [13] to as young as ∼78 years for rheniite [33]. In addition to enhancing our understanding on planetary formation [13], mantle–crust interactions [34,35] and the environment-life coevolution [29,36,37], the Re–Os chronometer has also proved to be essential in providing absolute time constraints for, and identifying the source rocks of metallic ore deposits and hydrocarbons [25,26,38–40].

This review begins with a concise summary of the fundamentals of Re–Os geochemistry and geochronology, then highlights key advances over the past two decades, and concludes with a discussion on potential directions for future research. For more comprehensive accounts of the historical development, analytical challenges and broad applications of the Re–Os chronometer, readers are referred to previous reviews [e.g. 41,42–44].

RE–OS SYSTEMATICS

Re and Os elemental and isotopic composition

Rhenium has two naturally occurring isotopes: ¹⁸⁵Re and ¹⁸⁷Re, with an isotopic abundance of 37.4% and 62.6%, respectively (Fig. 2a). Of these, ¹⁸⁷Re is radiogenic and decays to ¹⁸⁷Os [45]. The general formula for Re–Os geochronology is given in Equation (1):

Figure 2.

Elemental and isotopic composition of Re and Os. (a, b) Isotopic composition of Re and Os in natural samples. The abundance of radiogenic ¹⁸⁷Os varies across samples, with higher values generally found in those with greater Re concentrations and older geological ages. (c) Rhenium concentrations in molybdenite, pyrite, shale, bitumen and crude oil. Molybdenite displays the greatest Re concentrations, ranging from sub-ppm to % levels. In contrast, Re contents in pyrite, organic-rich sediments (e.g. shale), bitumen and crude oil are typically three orders of magnitude lower, ranging from hundreds of ppt to ppm levels. CDF, cumulative distribution function. Data sources and references are provided in the Supplementary data (Table S1; the Re–Os systematics for molybdenite, pyrite, organic-rich sediments, bitumen and crude oil compiled from the literature. The data are presented in Figs 2 and 3. References for the data are also given in Table S1.)

(1)

In Equation (1), t represents the age of the sample, and λ is the decay constant of ¹⁸⁷Re. The terms ¹⁸⁷Re_measured and ¹⁸⁷Os_measured refer to the measured isotopic abundances, while ¹⁸⁷Os_initial denotes the molar abundance of non-radiogenic ¹⁸⁷Os, namely the Os incorporated into the mineral at the time of its formation, commonly referred to as ‘common Os’. Using a decay constant (λ) of (1.666 ± 0.017) × 10⁻¹¹ year⁻¹ [13], the half-life of ¹⁸⁷Re is approximately 41.6 billion years—around nine times the age of the Earth.

The reliable application of any radiogenic chronometer depends on a thorough understanding of the elemental and isotopic composition of the targeted samples (Fig. 2). Given the extremely low concentrations of Re in both the mantle and crust, coupled with the slow decay rate of ¹⁸⁷Re, geochronological studies must focus on minerals or phases that are significantly enriched in Re. To date, the most extensively investigated materials include molybdenite, pyrite, organic-rich sedimentary rocks, bitumen and crude oil [41,42].

Rhenium rarely forms discrete minerals in nature, with the notable exception of rheniite (ReS₂) [33]. Instead, Re is typically concentrated in molybdenite (MoS₂), where its abundance ranges from sub-ppm to % levels (Fig. 2c). The high Re concentrations in molybdenite are generally attributed to the substitution of Re for Mo in the crystal lattice, owing to (i) the similar ionic radii of Re⁴⁺ and Mo⁴⁺ (0.63 and 0.65 Å, respectively) and (ii) the strong partitioning of Re into molybdenite relative to coexisting sulfides (e.g. pyrite, chalcopyrite) and/or silicate minerals. Combined with the typically negligible concentrations of common Os in molybdenites, i.e. ¹⁸⁷Os_initial is generally negligible, molybdenite Re–Os geochronology is widely applied as a high-precision, single-mineral chronometer with high accuracy [46–49].

Another commonly studied mineral in Re-Os geochronology is pyrite [4,22,26,38,50], although other sulfide minerals such as arsenopyrite [23,51,52], bornite [25,53,54], chalcopyrite [55–57], cobaltite [6] and safflorite [24] have also been used. Pyrites typically contain low Re concentrations (e.g. [58]), ranging from a few ppb up to ppm levels (Fig. 2c), while its common Os content is highly variable, from negligible amounts to tens of ppb.

Organic-rich materials, such as organic-rich sedimentary rocks, bitumen and crude oil, contain the lowest Re concentrations among commonly dated samples (Fig. 2c), and typically incorporate initial Os at sub-ppb to ppb levels. Though not exclusively, the presence of elevated Mo concentrations and high total organic carbon in organic-rich sediments may indicate elevated Re abundances [59].

Because of the large variations in Re and Os elemental and isotopic concentrations among these samples (Figs 2 and 3, Table 1), strategies in Re-Os geochronology must match specific sample types. These strategies encompass sample collection, processing, isotopic analysis, data reduction and interpretation—from samples to data, to dates, and ultimately to ages. For example, samples with high Re concentrations require smaller sample sizes to achieve a given precision in isotopic measurements compared to those with lower Re content and/or younger ages. In addition to Re abundance, isotopic composition is also a critical factor, although it is typically unknown prior to analysis. To assist in methodological planning, here we have compiled literature data (Fig. 3a) to provide a framework for guiding Re–Os geochronological studies.

Figure 3.

Sample ¹⁸⁷Re/¹⁸⁸Os ratios govern data visualization and interpretation. (a) The ¹⁸⁷Re/¹⁸⁸Os ratios in pyrite, organic-rich sediments (e.g. shale), bitumen and crude oil typically range from ∼10 to 10⁴. Molybdenite contains high Re and exceptionally low common Os, resulting in extremely high ¹⁸⁷Re/¹⁸⁸Os ratios; thus, they are not shown. Some pyrites also exhibit ¹⁸⁷Re/¹⁸⁸Os ratios much higher than 10⁴ and are referred to as low-level (Re), highly radiogenic (¹⁸⁷Os) samples (LLHR [4]). Here, LLHR samples are defined as those with ¹⁸⁷Re/¹⁸⁸Os > 10⁴. (b) For samples with non-negligible common Os (e.g. ¹⁸⁷Re/¹⁸⁸Os < 10⁴), such as organic-rich sediments, bitumen and crude oil, several co-genetic samples must be analyzed to obtain an isochron in the ¹⁸⁷Re/¹⁸⁸Os vs. ¹⁸⁷Os/¹⁸⁸Os space, as described by Equation (4). (c) For samples with negligible common Os (e.g. ¹⁸⁷Re/¹⁸⁸Os > 10⁴), such as molybdenite and certain pyrites, a model age can be calculated for each sample using Equation (3); Then a weighted average can be calculated for multiple co-genetic samples. Given the imprecise measurement of ¹⁸⁸Os at very low concentrations, the traditional isochron approach is not recommended unless uncertainties are properly propagated with the consideration of the error correlation (ρ) [160,161]. Alternatively, an inverse isochron is recommended by plotting ¹⁸⁷Re/¹⁸⁷Os vs. ¹⁸⁸Os/¹⁸⁷Os [63]. Relevant data and references are provided in Table S1.

The mathematical expression for the age (t) can be derived by rearranging Equation (1):

(2)

Equation (2) contains two unknowns (t and ¹⁸⁷Os_initial), which can be solved using either a single sample or multiple samples, depending on the abundances of ¹⁸⁷Os_initial.

Samples with high ¹⁸⁷Re/¹⁸⁸Os ratios (>10⁴)

If ¹⁸⁷Os_initial is negligible, t can be determined from a single sample using Equation (3):

(3)

Most molybdenites and some pyrites, which have high Re abundances but extremely low common Os content, meet this criterion [4,53], as indicated by their very high ¹⁸⁷Re/¹⁸⁸Os ratios (>10⁴, Fig. 3a).

Dates calculated from Equation (3) are referred to as model ages, based on the assumption that ¹⁸⁷Os_initial is negligible. For samples with ¹⁸⁷Re/¹⁸⁸Os ratios of >10⁴, this assumption introduces an overestimation of <0.09 and <0.05 Ma for samples aged at 100 Ma and 4000 Ma, respectively, which are negligible for an analytical precision of 0.1%. However, this may become a concern for younger samples when high accuracy is required. For example, for samples with ¹⁸⁷Re/¹⁸⁸Os ratios of 10⁴ but ages of 1 Ma and 10 Ma, the model ages will be overestimated by approximately 10% and 1%, respectively.

The assumption that common Os is negligible can be tested analytically using a double spike [60,61]. Alternatively, multiple samples with an identical age (e.g. molybdenites from the same vein) may be analyzed as an independent examination [49]. If samples with lower Re content systematically yield older ages, this provides strong evidence that the assumption of negligible common Os is not satisfied.

When the assumption of negligible common Os is valid, multiple co-genetic samples can be used to calculate a weighted mean age with improved precision (Fig. 3c). Alternatively, an isochron age can be derived from a ¹⁸⁷Re vs ¹⁸⁷Os diagram, which is effectively equivalent to the weighted mean of model ages (Fig. 3c).

Since isotope ratios are measured with a much higher precision and accuracy then isotope abundances, in practice, Equation (1) is divided by a stable isotope of Os, and ¹⁸⁸Os is the typical choice, with a natural isotopic abundance of 13.2%.

(4)

In Equation (4), the uncertainties of the ¹⁸⁷Re/¹⁸⁸Os and ¹⁸⁷Os/¹⁸⁸Os ratios are correlated, as both share a common denominator arising from the ¹⁸⁸Os measurement. However, this approach is not recommended for samples with very high ¹⁸⁷Re/¹⁸⁸Os ratios unless uncertainty propagation is accurately performed with explicit consideration of error correlation [62].

Specifically, because ¹⁸⁸Os concentrations are typically several orders of magnitude lower than those of ¹⁸⁷Re and ¹⁸⁷Os, the uncertainties in the ¹⁸⁷Re/¹⁸⁸Os and ¹⁸⁷Os/¹⁸⁸Os ratios are dominated by the imprecise measurements of ¹⁸⁸Os [4], and are therefore highly correlated (e.g. >0.99). For datasets with very high ¹⁸⁷Re/¹⁸⁸Os ratios, plotting an isochron without accounting for error correlation (i.e. assuming a correlation coefficient of zero) should be avoided, as this will bias the isochron age and lead to overestimated uncertainties. Given the extremely low concentrations of common Os, the isochron's y-intercept is expected to have large uncertainties [53], and using it as a tracer is not recommended.

An improved method for datasets with very high ¹⁸⁷Re/¹⁸⁸Os ratios—and thus highly correlated uncertainties between the ¹⁸⁷Re/¹⁸⁸Os and ¹⁸⁷Os/¹⁸⁸Os ratios—is the inverse isochron approach (Fig. 3c), which involves plotting the data in ¹⁸⁷Re/¹⁸⁷Os vs ¹⁸⁸Os/¹⁸⁷Os space [63]. This method is also recommended for isochron studies with moderate to low levels of correlated uncertainties as it provides a superior visualization of dispersion.

Samples with moderate to low ¹⁸⁷Re/¹⁸⁸Os ratios (<10⁴)

For samples with non-negligible common Os—such as most pyrite, organic-rich sediments, bitumen and crude oil (Fig. 3a)—a model age cannot be obtained using Equation (3). Instead, several co-genetic samples must be analyzed using the isochron approach (Fig. 3b), following Equation (4).

In addition to yielding an age from its slope (e^λt − 1), the isochron approach also provides the isotopic composition for common Os (Os_i = ¹⁸⁷Os/¹⁸⁸Os_initial) from its y-intercept (Fig. 3b). Although the inclusion of common Os introduces an additional dimensional of information, it comes at a cost—limiting the precision of age determinations. This explains why isochron ages are typically less precise than the model ages described above.

OS ISOTOPE AS A TRACER FOR GEOLOGICAL PROCESSES

The preferential enrichment of Re in the continental crust underpins the use of ¹⁸⁷Os/¹⁸⁸Os ratios as tracers of crustal processes (Fig. 4). At present, with an average radiogenic Os_i of approximately 1.4, the continental crust is distinctly different from mantle-derived materials—such as basalts and hydrothermal fluids at middle-ocean ridges—which typically exhibit an Os_i of 0.12. In this regard, The Re–Os isotope system has been shown to be a powerful tool for tracking magma evolution during basaltic eruptions, owing to its exceptional sensitivity to crustal assimilation. Specifically, it can detect the incorporation of crustal materials with high Re/Os ratios and radiogenic ¹⁸⁷Os/¹⁸⁸Os values that often go unnoticed by conventional isotope systems such as Sr–Nd–Pb. For example, Os_i values of 0.188 and 0.132 were used to trace the degree of crustal contamination in the 2021 Fagradalsfjall basaltic eruption in Iceland [64].

Figure 4.

Osmium isotope composition (Os_i, i.e. ¹⁸⁷Os/¹⁸⁸Os_initial) of common geological reservoirs. Basalts derived from the present-day mantle have an average Os_i of ∼0.12, representing the lower limit among geological reservoirs. The modern ocean has an Os_i of ∼1.06, reflecting a balance between more radiogenic continental river runoff (average Os_i ∼ 1.4) and hydrothermal inputs at middle-ocean ridges (average Os_i ∼ 0.12). Cosmic dust and meteorites typically exhibit an Os_i of ∼0.13.

The residence time of Os in the ocean is relatively short, with estimates ranging from 3–4 to 35–50 kyr [65–68], making Os isotopes a sensitive proxy for investigating a range of geological processes, though a short residence time may hinder its application in tracing some global events. An increased flux of mantle-derived materials to the global oceans—either through enhanced weathering of flood basalts or increased hydrothermal input at middle-ocean ridges and large igneous provinces—can be traced using Os_i signatures preserved in organic-rich sediments (Fig. 4), such as during oceanic anoxic events [69–71].

The rate of continental weathering also can be tracked using the Os_i record [72]. For example, postglacial strata younger than 662 Ma exhibiting radiogenic Os_i (∼0.54) have been used to infer intense silicate weathering during the post–Snowball Earth hothouse interval [30]. Radiogenic Os_i values from the Lantian and other Ediacaran shales (>1.0) have been attributed to oxidative weathering of the upper continental crust during the early to middle Ediacaran [36], potentially serving as a trigger for biological evolution and oceanic atmosphere oxygenation [73].

Cosmic inputs exhibit chondrite-like Os_i values (∼0.13) and can be used to indicate the presence of impact events [74,75]. As discussed above, mantle-derived hydrothermal fluid, basalt weathering and cosmic input all contribute to lowering oceanic Os_i (Fig. 4). Therefore, additional geological and geochemical evidence is essential to reliably distinguish these potential sources of Os_i fluctuations [45]. For meteorites, although their Os isotope signatures are often indistinguishable from those of the Earth's mantle, they are highly enriched in platinum-group elements (PGEs)—by factors of up to 1000 relative to mantle rocks and 100 000 compared to average crustal compositions [76]. This unique combination means that even trace amounts of meteoritic material can generate a measurably shift in marine ¹⁸⁷Os/¹⁸⁸Os ratios and PGE concentration. A classic example is the K–Pg boundary, where both a pronounced Ir spike and a drop in Os_i provide compelling evidence of an extraterrestrial impact. Such isotopic fingerprints have been used not only to estimate the magnitude and timing of meteoritic inputs to Earth's surface [77], but also to infer the type [78] and size [79] of the impacting bodies.

ANALYTICAL PROCEDURES

The conventional analytical procedure (Fig. 5) for obtaining Re–Os dates is by isotope dilution. The new in situ approach is not addressed in this section but is discussed separately below. Sample digestion is a critical step in conventional Re–Os geochronology [80]. Isotope dilution requires complete isotope equilibrium between spike and sample, which is straightforward for Re but challenging for Os. This is because Os reacts with oxygen at ambient temperatures to form osmium tetroxide (OsO₄), a volatile and highly toxic compound [81]. To ensure complete Os isotope equilibration between the spike and sample, digestion must occur under highly oxidizing conditions at elevated temperature (e.g., >200 °C) to convert all Os to its highest oxidization state (Os⁸⁺). This necessitates the use of an oxidizing medium, such as inverse aqua regia (2*HNO₃–1*HCl), for sulfides, bitumen and crude oil, in combination with the Carius-tube technique [82]. Due to the unavoidable presence of detrital Re and Os, CrO₃–H₂SO₄, rather than inverse aqua regia, is used for organic-rich sediments to preferentially release Re and Os from the hydrogenous component [1,28].

Figure 5.

A representative analytical workflow for Re–Os geochronology. Re–Os geochronology begins with sample digestion using either HNO₃–HCl (for sulfides, bitumen and crude oil) or CrO₃–H₂SO₄ (for organic-rich sediments) as the digestion medium. The Carius-tube technique is preferred to minimize Os loss during digestion, and an oxidizing environment is essential to ensure complete isotopic equilibration between spike and sample Os. Osmium is separated from the digestion medium by solvent extraction using CHCl₃, followed by back-extraction with HBr and purification by micro-distillation, prior to isotopic analysis by NTIMS. Rhenium is converted to an NaOH medium and separated by solvent extraction using acetone [(CH₃)₂CO]. Further purification of Re is performed using ion exchange before isotopic analysis by NTIMS or MC-ICPMS. The tubes are for demonstration purposes only and are not actual containers used in the experiments.

Solvent extraction (Fig. 5) is the standard method for isolating both elements—using CHCl₃ to extract Os from the digestion medium and to extract Re from an NaOH solution [83–86]. A sparging technique may also be used to extract Os directly by introducing oxygen into Os-bearing solutions [87]. Osmium can be further purified via back-extraction and micro-distillation using HBr [84]. For Re, further purification is carried out using anion exchange chromatography [1,83].

The high ionization potential of Os (8.7 eV) and its low abundance (sub-ppb levels) hinder isotopic composition measurements using traditional thermal ionization mass spectrometry (TIMS) analysis as positive ions, thereby significantly limiting its application as a chronometer. Creaser et al. [88] and Völkening et al. [89] independently demonstrated that Re and Os can be ionized as negatively charged oxides (ReO₄^– and OsO₃^–) using Ba(OH)₂ and/or Ba(NO₃)₂ as an activator. This revolutionary approach led to a multi-fold increase in the ionization efficiency of Re and Os, greatly reducing the sample size required for Re–Os geochronology while improving both precision and accuracy. It represents a major milestone that catalyzed the widespread application of this chronometer over the past 30 years.

CRITICAL CONSIDERATION FOR ACCURATE DATING

Mixing line and the concentration test

A common pitfall in isochron dating is calculating ages from an apparent isochron produced by mixing two end-member components [90]. For example, water–rock interaction between hydrothermal fluids and country rocks can generate alteration assemblages with variable ¹⁸⁷Re/¹⁸⁸Os and ¹⁸⁷Os/¹⁸⁸Os ratios, which may define a linear trend (Fig. 6a). Additionally, analysis of different organic fractions of a single oil may also yield a linear trend.

Figure 6.

Distinguishing between isochron and mixing line. (a) An apparent ‘isochron’ can result from mixing two end-member components. (b, c) This can be identified using the 1/Os vs ¹⁸⁷Os/¹⁸⁸Os (also known as the 1/C test). A linear array in this plot indicates mixing rather than true age information. Note that the 1/C test is not applicable for LLHR samples.

Such linear trends maybe mixing lines without true chronological significance, which can be identified using the concentration test, i.e. by plotting 1/Os versus ¹⁸⁷Os/¹⁸⁸Os, where Os is defined by Equation (5):

(5)

If a correlation is observed from the concentration test (Fig. 6b), it provides strong evidence that the apparent ‘isochron’ may in fact be a mixing line.

However, this test is not universally applicable. For samples with very low common Os content, where the total Os budget is dominated by radiogenic ¹⁸⁷Os—such as molybdenites and pyrites with a ¹⁸⁷Re/¹⁸⁸Os ratio of >10⁴—a negative correlation between 1/Os and ¹⁸⁷Os/¹⁸⁸Os is expected. In such cases, the concentration test becomes invalid, as the observed trends reflect radiogenic ingrowth rather than mixing.

Initial slope

When conducting isochron dating, it is essential that the samples share an identical formation age. However, in practice, samples with slightly different ages are sometimes used to achieve the necessary spread in the ¹⁸⁷Re/¹⁸⁸Os vs ¹⁸⁷Os/¹⁸⁸Os space for linear regression. For example, in Re–Os geochronology of organic-rich sedimentary rocks, sampling along a horizontal profile may yield a dataset with nearly uniform ¹⁸⁷Re/¹⁸⁸Os and ¹⁸⁷Os/¹⁸⁸Os ratios, making it unsuitable for isochron dating. To obtain the required variation, samples are typically collected along a vertical profile, in which case samples from the bottom layer were older and thus accumulated slightly more radiogenic ¹⁸⁷Os prior to the deposition of upper layers. This results in an initial slope in the isochron diagram—either positive or negative depending on the relationship between Os_i and ¹⁸⁷Re/¹⁸⁸Os ratios—leading to either overestimated or underestimated dates. Similarly, variations in Os_i are another concern.

The impact of this initial slope can be significant for younger samples with a slow depositional rate [91,92] or those collected over an extended vertical interval but is generally negligible for older samples sampled over a short vertical range. In practice, the acceptable profile height should be evaluated on a case-by-case basis.

Mixing multisource samples of the same age

In addition to verifying that the samples used for isochron dating formed at the same time, another potential issue arises when mixing samples with different Os_i despite identical ages, which is a common problem when dating mineralization with extensive water–rock interaction [93]. The Re–Os dataset from the Xinqiao massive sulfide deposit in China [38] provides a compelling example for this scenario (Fig. 7).

Figure 7.

Dating pyrites with contrasting elemental and isotopic compositions. (a) Pyrites from the Xinqiao deposit include euhedral pyrites (py1, top panel) and colloform pyrites (py2, middle panel) within the stratabound massive orebody, as well as pyrites occurring in sandstone-hosted stockworks (py3, bottom panel). (b) Pyrites (py1 and py2) from the stratabound massive orebody define a 118 ± 11 Ma Re–Os errorchron (Os_i = 1.50 ± 0.14; MSWD = 29). (c) Pyrites (py3) from the sandstone-hosted stockworks yield a 389 ± 31 Re–Os errorchron (Os_i = −76 ± 45; MSWD = 120). Note that py3 pyrites are characterized as LLHR pyrite, displaying significantly higher ¹⁸⁷Re/¹⁸⁸Os ratios compared to py1 and py2 pyrites, and using the traditional isochron diagram is not recommended. (d) Euhedral pyrites (py1) yield a 136 ± 4 Ma Re–Os isochron (Os_i = 0.79 ± 0.10; MSWD = 2.2). (e) Colloform pyrites (py2) yield a 134 ± 8 Ma Re–Os isochron (Os_i = 1.39 ± 0.08; MSWD = 5.4). (f) Model ages from the stockwork pyrites (py3) vary from ∼380 Ma to 170 Ma and show a positive correlation with Re content, violating the prior of one population. Sulfur isotope data (inserts in d, e and f) demonstrate that these pyrites formed under different conditions. Py, pyrite; Qtz, quartz.

This stratabound deposit features massive pyrite (py1) and colloform pyrite (py2) ores hosted in the Carboniferous limestone, and pyrite stockworks (py3) hosted in the Devonian sandstone. Although these pyrites (Fig. 7a) are genetically linked to the ∼138 Ma diorite [94], they exhibit different ¹⁸⁷Re/¹⁸⁸Os ratios and Os_i values (Fig. 7b and c). Py1 and py2 contain common Os with distinct Os_i values: ∼0.79 for py1 and ∼1.35 for py2 at 134–136 Ma (Fig. 7d and e). Py3 is highly radiogenic, with ¹⁸⁷Re/¹⁸⁸Os ratios ranging from ∼1.4 × 10⁴ to 2.7 × 10⁵ [38,95]. The elemental and isotopic variations in py1, py2 and py3 are consistent with their contrasting morphology (Fig. 7a) and distinct sulfur isotope values (insets in Figs 7d–f). Given these substantial differences in Re–Os isotopic characteristics, careful sample separation is essential for obtaining geologically meaningful ages [38].

Assumptions in isochron dating and mean square weighted deviation

An important but often overlooked assumption in age calculations—whether using the weighted mean or isochron approach—is that the samples constitute a single statistic population [96]. If one or more samples differ significantly from others—e.g. in their uncertainties or elemental/isotopic compositions—it is essential to assess whether they belong to a single population. The Xinqiao study discussed above [38,93,94] is a classical example on this topic.

Anchoring isochrons to plausible initial ¹⁸⁷Os/¹⁸⁸Os ratios may be an option when independent constraints on Os_i are available, but the impact of Os_i on the isochron age must be rigorously evaluated.

Once a date is obtained, its quality is typically evaluated using the probability of goodness of fit (p) or mean squared weighted deviation (MSWD), both of which are a functions of the residual sum of squares (RSS) divided by the degree of freedom, and are inversely correlated (i.e. larger MSWD values correspond to smaller p) [97,98].

The distribution of MWSD is governed by the degree of freedom [97]. For large datasets (i.e. N > 20), the MSWD distribution approximates a normal distribution with an expected mean of 1, but it is skewed to the left for datasets with N < 20. For a dataset with N samples, the acceptable MSWD range at the 95% confidence interval is [1, 1+], where f is the degree of freedom (i.e. f = N − 2 for linear regression).

It is important to note that MWSD should not be used as a simplistic test of date quality [99]. Although a high MSWD (e.g. >1) may suggest the presence of geological dispersion, potentially violating the assumption of isochron dating, it can also result from incomplete propagation of analytical uncertainties [100]. Conversely, a low MSWD (e.g. <1) may reflect an overestimation of analytical uncertainties [100]. One of the most common errors arises from neglecting the correlation between uncertainties in the ¹⁸⁷Re/¹⁸⁸Os and ¹⁸⁷Os/¹⁸⁸Os ratios. Since geochronological studies, except for in situ geochronology discussed below, often use small-N datasets (e.g. N < 20), the common practice of using 1 as a universal threshold is not appropriate.

For the same sampleset, ongoing advancements in analytical techniques will yield more precise measurements; consequently, the isochron will exhibit a lower probability of goodness of fit (p), or higher MSWD in equivalent [99]. We emphasize that high-precision datasets will become increasingly common in the future and are likely to exhibit elevated MSWD values, which should not be viewed unfavorably by editors and reviewers [99]. Similarly, larger datasets may become more prevalent in the future (see the novel prospect of in situ geochronology, discussed below), for which the accepted MSWD threshold is approaching 1.

KEY ADVANCES IN THE PAST DECADE

Molybdenite Re–Os dating with a precision of 0.5‰

Molybdenite is the only ore mineral for Mo, and a common phase in porphyry deposits. Unlike zircon U–Pb or biotite Ar–Ar ages, which require assumptions to link measured dates to ore formation, molybdenite Re–Os dates can be directly linked to ore-forming processes without ambiguity. The enrichment of Re in molybdenite—up to the % level (Fig. 2c)—along with negligible common Os [101], permits high-precision dating using Equation (3) as a single-mineral chronometer.

Previous studies have shown that radiogenic ¹⁸⁷Os may be decoupled from Re within molybdenite crystals, whereby zones with ¹⁸⁷Os gain or loss yield artificially high or low model ages, respectively [102–104], rendering them unsuitable for accurate dating. Within-crystal diffusion of Os, which is enhanced in larger crystals (e.g. >500 μm) and samples with older ages, has been proposed as an explanation [102,103,105]. In practice, fine-grained molybdenite crystals are preferred to avoid this decoupling problem. If only large grains are available, it is critical to make sure that a suitable aliquot is analyzed, and ideally to sample the entire crystal. Grinding a portion of a big crystal into fine powers, as suggested by some studies in the literature, does not solve the decoupling issue. Further discussion on the decoupling between Re and Os is given below in the in situ Re–Os dating section.

The analytical precision of molybdenite Re–Os dating is primarily controlled by the amount of Os loaded onto the filament for measurement [106], which depends on sample size, Re content and sample age. An optimized spike-to-sample ratio is also essential [107], necessitating prior determination of the sample's Re concentration. The exact precision achievable is sample-dependent, and modern molybdenite Re–Os geochronology generally can achieve a precision of 3‰ or better than 0.5‰ (Fig. 8) when only analytical uncertainties are considered, e.g. excluding uncertainties from spike calibration and the decay constant [49]. For molybdenites from the ∼11 Ma Los Pelambres porphyry deposit, using a sample size of ∼20 mg, a precision of 1‰–3‰ (Fig. 8) has been achieved [108], equivalent to a temporal resolution of 10–30 kyr. Because the temporal resolution of radiometric dating is maximized for young samples, a similar precision (1‰–3‰; Fig. 8) for the ∼1 Ma OK Tedi porphyry deposit yields a much higher temporal resolution of 1–3 kyr [47]. For the ∼16 Ma Qulong porphyry deposit, using a sample size of ∼30 mg, a precision of 0.3‰–0.5‰ (Fig. 8) was achieved with a temporal resolution of 6 kyr [49], which is five times better than that from previous studies. Using a smaller sample size of 1–3 mg, an order of magnitude lower than previous studies, a precision of ∼3‰ (Fig. 8) has been achieved for the ∼5 Ma El Teniente porphyry deposit [48], equivalent to a temporal resolution of 15 kyr. A reduced sample size, at the expense of precision, avoids the potential of mixing multi-stage grains and significantly expands the applications of this technique.

Figure 8.

Analytical precision of individual molybdenite Re–Os model ages. Squares represent data from the ∼16 Ma Qulong porphyry deposit, with a reproducibility of ∼0.5‰ using ∼30 mg samples [49]. Circles represent data from the ∼11 Ma Los Pelambres porphyry deposit [108] and the ∼1 Ma OK TEDI porphyry deposit [47] with a reproducibility of ∼2‰ using ∼20 mg samples. Triangles represent data from the ∼5 Ma El Teniente porphyry deposit [48], with a reproducibility of ∼3‰ using 1–3 mg samples.

For Re–Os dates with a temporal resolution of a few kyr, we can now begin to decode the cyclic nature of ore-forming processes in giant porphyry systems, as first demonstrated at the Qulong porphyry deposit [49]. This giant porphyry Cu–Mo deposit has been extensively dated by molybdenite Re–Os geochronology, but with a precision of 10‰–30‰ (Fig. 9a), equivalent to a temporal resolution of 160–500 kyr (see references in [109]). The lower-precision dates were used to argue that an extended longevity (i.e. ∼1.5 Myr) is required to form giant deposits. However, through ultra-high precision (0.5‰) Re–Os dating, new dates with a temporal resolution of 6 kyr instead indicate that the giant Qulong deposit was formed within 266 kyr, a 6-fold shorter duration than previously thought (Fig. 9b). The dates further reveal that the ore-forming process is not continuous but comprises three intermittent pulses, each lasting ∼40–60 kyr (Fig. 9c).

Figure 9.

The timescales and rhythms of porphyry copper deposits revealed by molybdenite Re–Os geochronology. (a) Molybdenite Re–Os dates with an analytical precision of 10‰–30‰ (temporal resolution of 160–500 kyr for a 16 Ma system) were used to suggest a continuous mineralization process for the Tibetan Qulong porphyry Cu–Mo deposit between 16.85 and 15.36 Ma, with a duration of ∼1.5 Myr. (b) Molybdenite Re–Os dating with a precision of ∼0.5‰ (temporal resolution of 6 kyr for the 16 Ma system) demonstrates that the ore-forming process is rapid and episodic. (c) The refined duration of 266 kyr is six times shorter than the previous estimate and consists of three intermittent pulses lasting 40–60 kyr [49].

A rapid, pulsed nature of the ore-forming process in porphyry copper systems represents a paradigm shift, initially defined by ultra-high precision Re–Os dating [110], which has now been consistently demonstrated across magmatic-hydrothermal systems globally [111–116].

The Qulong example presented in Fig. 9 further highlights that when low-precision geochronological datasets are used, a pulsed process with a hiatus cannot be determined. To investigate the pulsed nature of rapid geological processes, we emphasize that a minimum threshold of analytical precision is required for geochronology [117], and a sufficient number of samples is needed [118].

Imaging-guided Re–Os dating of organic-rich sediments

Re–Os geochronology is one of the few methods that can directly and robustly date organic-rich sedimentary rocks in the absence of interbedded volcanic ash layers, and is widely applied to provide age information on climate perturbations and biological evolution [29,36,119–125].

Rhenium and Os are highly mobile during weathering [126–128], post-formation alteration and metamorphism [129,130], resulting in open-system behavior for the Re–Os chronometer in organic-rich sediments; therefore, fresh drill core samples are strongly preferred for geochronological studies [131]. To obtain a high-quality isochron, in addition to meeting the prerequisites—i.e. that samples formed simultaneously, share an identical Os_i and remained a closed isotopic system—samples also need to display a significant spread in ¹⁸⁷Re/¹⁸⁸Os ratios to enable linear regression. This can be achieved by sampling vertically along the drill core, and the height of this vertical profile needs to be sufficiently large; otherwise, the variation in ¹⁸⁷Re/¹⁸⁸Os ratios will be limited. However, if the vertical profile is too extended, it may violate the prerequisite assumptions of isochron dating (i.e. identical age and Os_i).

The sedimentation rates of organic-rich shales are highly variable, with estimates ranging from several centimeters to sub-millimeters per kyr [132–134], corresponding to a thickness of 1–10 m over a 0.1 Ma interval. In this regard, for samples collected from an interval ranging from several tens of centimeters to a few meters, it is generally reasonable to assume a near-synchronous deposition within analytical uncertainties, thereby satisfying the prerequisite of identical formation ages for isochron dating, though compaction should also be considered. However, a broader interval increases the risk of incorporating samples with variable Os_i due to changes in Os flux during sedimentation—a concern particularly relevant to organic-rich sediments near stratigraphic boundaries or recording climate perturbations.

To select the best-preserved drill core for Re–Os geochronology, traditional petrographic examination should be conducted to avoid samples with veins, alteration and weathering. Due to the high mobility of Os, low-temperature hydrothermal alteration and/or metamorphism may disturb the Re–Os system in organic-rich sediments, which is often difficult to detect using traditional petrographic approaches, potentially leading to a failed dating attempt [127]. Modern techniques, such as computed tomography (CT) and X-ray fluorescence (XRF), can be used (Fig. 10). A CT scan can produce a 3D distribution of materials with contrasting densities; for example, pyrites and barite, with higher density can be distinguished from less dense materials, such as silicates, pore space and voids [36,42].

Figure 10.

An imaging-guided approach for shale Re–Os geochronology. CT and XRF scanning can be used to aid in selecting the best-preserved shale intervals for Re–Os geochronology akin to cathodoluminescence imaging used in zircon U–Pb geochronology. The shale core and CT image are from the Lantian biota [36]. The Sr distribution maps obtained by XRF are cores from the Qingjiang biota.

Although a CT scan can provide internal information for shales with high spatial resolution, it is unable to determine the chemical distribution of trace elements, which is the major concern for shales that have experienced low-temperature alteration/metamorphism. In contrast, XRF scanning [36] provides a powerful means to assess potential elemental remobilization using elements such as Sr as a proxy (Fig. 10). Furthermore, Re distributions in shales can be obtained using laser ablation inductively coupled plasma mass spectrometry (LA-ICPMS), permitting sampling of the maximum variation in ¹⁸⁷Re/¹⁸⁸Os ratios within a given organic-rich sediment interval, which is critical for producing high-quality isochrons. We suggest that an imaging approach using both XRF and potentially LA-ICPMS should be applied for future shale Re–Os geochronology, in a way akin to cathodoluminescence imaging used for zircon U–Pb geochronology.

OUTLOOK

In situ Re–Os dating

A recent innovation in mass spectrometry now enables accurate measurement of isotope ratios free from isobaric interferences, leading to the development of in situ beta-decay geochronology [e.g. 135,136]. In this novel approach (Fig. 11), rock blocks or thin sections are ablated by a pulsed laser beam (LA), and the resulting aerosol is atomized and ionized in an inductively coupled plasma (ICP). The ions subsequently travel to a reaction cell, sandwiched between two quadrupole mass analyzers (MS/MS), allowing for accurate, interference-free isotope ratio measurements.

Figure 11.

The principles of in situ Re–Os dating. (a) A schematic diagram of the mass spectrometer with a reaction cell between two mass analyzers (Q1 and Q2), with samples introduced via a laser ablation system. When CH₄ is used in the reaction cell, OsCH₂⁺ is produced and measured at a +14 amu mass shift. When N₂O is used, OsO₄⁺ is produced and measured at a +64 amu mass-shift. (b) Rhenium does not react significantly with either gas, but a mathematical interference correction is required to obtain interference-free Os measurements. The expected interference correction of Re on Os is plotted against sample age for a range of typical Re reaction rates in CH₄, N₂O and N₂O + He reaction gas atmospheres. The superscript ‘+x’ represents the mass shifts for the different reaction gases.

Applied to Re–Os geochronology, the LA-ICP-MS/MS method can effectively separate ¹⁸⁷Os and ¹⁸⁷Re by reacting Os with a gas introduced into the reaction cell of the mass spectrometer (Fig. 11a). This method was pioneered by Hogmalm et al. [137] and refined by Tamblyn et al. [138], demonstrating that Os effectively reacts with CH₄ gas to form OsCH₂⁺ ions, while the equivalent reaction of Re to ReCH₂⁺ is about four orders of magnitude less efficient. However, there is a residual interference that must be accurately corrected by measuring the ¹⁸⁵ReCH₂⁺/¹⁸⁵Re reaction rate in an Os-free material [typically National Institute of Standards and Technology (NIST) glass] and correcting for ¹⁸⁵Re/¹⁸⁷Re mass bias using the natural isotopic composition of Re. This ReCH₂⁺ production rate is typically around 1%–2% [138], which can equate to a significant interference correction on OsCH₂⁺, especially for young samples with low radiogenic Os ingrowth.

It has been demonstrated that Os also reacts with N₂O gas to form OsO₄⁺ [139]. The equivalent production rate of ReO₄⁺ is nearly an order of magnitude lower compared to that of ReCH₂⁺ in the CH₄-based method, leading to reduced interference corrections. Glorie et al. [140] subsequently showed that adding He into the reaction cell further suppresses Re-related interferences (Fig. 11b). However, this can come with the trade-off of lower signal sensitivity, and the most suitable analytical recipe is often a compromise between minimizing interference corrections and maintaining sufficient signal sensitivity.

The in situ method has proved to be very powerful for rapidly obtaining a large quantity of Re–Os dates on molybdenite, i.e. up to 1000 analyses can be performed within a 24-h analytical session. While uncertainties in individual Re/Os ratios are often much larger compared to conventional negative TIMS (NTIMS) measurements (approximately 2%–5% for Re-rich Precambrian molybdenite; >25% for Cenozoic molybdenite), the sample set is often much larger (typically 40–60 dates per sample), resulting in weighted mean dates with reduced uncertainties. For example, the LA-ICP-MS/MS Re–Os uncertainty of 0.6% (n = 37) obtained for the Jinka deposit is only marginally larger than the NTIMS Re–Os uncertainty of 0.5% for the same sample [139]. Similarly, an uncertainty of 0.3% was obtained using LA-ICP-MS/MS [138] for a Re-rich sample (n = 71), which is comparable to an uncertainty of 0.2% by NTIMS (n = 5).

Despite these advances, within-crystal decoupling between Re and Os [102,103,105] remains a potential concern for the accuracy of in situ molybdenite Re–Os dating. X-ray absorption fine structure suggests that Re occurs as Re⁴⁺ in the Mo site of molybdenite, while radiogenic Os occurs as Os³⁺ and Os⁴⁺, and does not form secondary Os phases such as OsS₂ and metallic Os [104]. The smaller ionic radius and lower charge of Os may lead to faster Os diffusion in molybdenite compared to Re [104], as previously proposed [102,103]. Nanoscale secondary ion mass spectrometry (NanoSIMS) imaging was used to argue that ¹⁹²Os is homogeneously distributed in molybdenite and that Re and Os are not decoupled [141]. However, this interpretation is uncertain due to the exceptionally low abundance of common Os (and thus ¹⁹²Os) in molybdenite, and the inability of NanoSIMS to distinguish between ¹⁸⁷Re and ¹⁸⁷Os [142]. High-resolution scanning transmission electron microscopy (TEM) analyses demonstrated that both Re and Os are incorporated into molybdenite through isomorphic substitution for Mo, with their distribution controlled by molybdenite precipitation and subsequent metamorphism and deformation [143]. Re–Os isotope mapping for molybdenite using LA-ICPMS/MS from the Merlin deposit shows no age variation across large crystals, at least within the resolution of the mapping approach [138]. However, this is an extremely Re-rich example and more in situ mapping is required to evaluate the potential decoupling of Os from Re.

The ability to rapidly and accurately date molybdenite in a petrogenetic context [140,144] opens new opportunities for mineral exploration, especially where fast sample throughput is required. Besides molybdenite, Re-rich pyrite, such as the ones from the Peel River deposit and Nick Prospect, also can be dated using this approach [139]. The application of in situ Re–Os geochronology is also being tested on shales [145], though further developments are required.

Inter-laboratory calibration

The development and refinement of Re–Os geochronology, including chemical and mass spectrometric techniques and optimized workflows over the past several decades, have pushed the precision of molybdenite Re–Os model dates to 0.5‰ (Fig. 7) and shale Re–Os isochron ages to 1%. The most important technical breakthroughs include: (ⅰ) the introduction of NTIMS for improved ionization efficiency of Re and Os [88,89]; (ⅱ) the Carius-tube digestion technique for achieving complete Os isotope equilibration between spike and sample under highly oxidized conditions [82]; (ⅲ) the targeted release of hydrogenous components from shales using CrO₃–H₂SO₄ [1]; and (ⅳ) the preparation of Os gravimetric standards using ammonium hexachloroosmate [(NH₄)₂OsCl₆], along with determination of its stoichiometric composition [9,146].

To facilitate the direct comparison of dates from different laboratories collected over a long period of time, uncertainties must be correctly propagated with careful quantification of random and systematic components. When dates from different laboratories are compared, uncertainties from spike calibration must be considered, especially if spikes are calibrated against different gravimetric standards [146]. For molybdenite Re–Os dating, both spike with normal Os isotope composition and spike using enriched natural isotopes are used. While accurately measuring common Os is important, the currently available double spikes—based on enriched natural isotopes [60,61,146], such as ¹⁸⁶Os–¹⁹⁰Os, ¹⁸⁸Os–¹⁹⁰Os or ¹⁹⁰Os–¹⁹²Os—are not ideal compared to double spikes composed of synthetic isotopes, such as the ET2535 spike used in the U–Pb community [147].

High-quality reference materials, including natural samples and synthetic solutions (e.g. age solutions, pre-spiked sample solutions), are also critical for monitoring data quality and enabling inter-laboratory comparison [146,148–152].

An excellent example to monitor and reduce the inter-laboratory uncertainties comes from zircon U–Pb geochronology under the community-driven EarthTime initiative [153,154]. Through distributing a spike calibrated against gravimetric standards that are traceable to the SI unit [147], along with a series of breakthroughs in analytical techniques including chemical abrasion [155], an uncertainty of 0.1% for single-crystal ²⁰⁶Pb/²³⁸U ages was successfully achieved [156], and an ambitious target of 0.01% [157] is now within close reach [158].

Currently, analytical uncertainties for the best molybdenite Re–Os ages are ∼0.5‰, but uncertainties from spike calibration and the decay constant are much larger [13,49]. Without a shared spike and the lack of cross-calibration against other chronometers such as U–Pb, direct comparison of Re–Os dates from different laboratories and with other chronometers must include these uncertainties, which limits applications using the highest temporal resolution available.

We propose that the Re–Os community should be fully integrated into the EarthTime initiative by following the successful model of the U–Pb community. Priority tasks include: (ⅰ) calibrating a shared spike traceable to the SI unit; (ⅱ) developing a data reduction protocol with complete propagation of both random and systematic uncertainties; and (ⅲ) preparing and distributing synthetic solutions with known ages for inter-laboratory evaluation.

Refining the decay constant

The decay constant fundamentally controls the accuracy of radiometric dates, and its uncertainty must be considered when dates from different radiogenic chronometers are compared (i.e. Re–Os and U–Pb). The currently adopted decay constant of ¹⁸⁷Re is calculated from iron meteorites [13]. The accuracy of this meteorite-derived decay constant is linked to: (ⅰ) the assumption that the dated meteorites have identical Pb–Pb and Re–Os ages, and (ⅱ) the spike used for Re–Os isotope determination.

Although an uncertainty of 0.31% is reported by Smoliar et al. [13], this does not include an additional uncertainty of ∼1.2% from the gravimetric standards used for spike calibration. Unless spikes used for Re–Os dating can be traced back to the study by Smoliar et al. [13], an uncertainty of ∼1.2% should be applied. With modern molybdenite Re–Os dating techniques now reaching a precision of <0.5‰, being ∼20 times better than the decay constant uncertainty, reassessing the ¹⁸⁷Re decay constant, and particularly its uncertainty, should be given the highest priority by the Re–Os community.

The accuracy of the decay constant determined by Smoliar et al. [13] has been evaluated using a cross-calibration approach with molybdenites and zircons from a series of Mo-bearing porphyry deposits [159]. In this legacy study, the molybdenite Re–Os dates had a precision of 0.1%–0.3%. Because zircon U–Pb dates were collected before the widespread adoption of chemical abrasion [155], ²⁰⁷Pb/²⁰⁶Pb dates were used, which have a precision of 0.1%–1.8% [159]. Given the uncertainties at the time, assuming molybdenites and zircons were formed at the same time is valid. With more than a 10-fold increase in the analytical precision for modern molybdenite Re–Os geochronology [47–49], and zircon U–Pb geochronology now achieving 0.01% precision and accuracy [162], as well as an improved understanding on the lifetime of porphyry deposits [49], a refined decay constant with substantially improved precision and accuracy is feasible using modern Re–Os and U–Pb dating techniques.

FUNDING

This work was supported by the National Natural Science Foundation of China (42325303). S.G. was supported by an Australian Research Council Future Fellowship (FT210100906).

AUTHOR CONTRIBUTIONS

Yang Li designed the study, compiled the data, and prepared the figures. Yang Li led the writing, with contributions from Stijn Glorie and David Selby.

Conflict of interest statement . None declared.