Half-life and initial Solar System abundance of 146Sm determined from the oldest andesitic meteorite.

The formation and differentiation of planetary bodies are dated using radioactive decay systems, including the short-lived 146Sm-142Nd (T½ = 103 or 68 Ma) and long-lived 147Sm-143Nd (T½ = 106 Ga) radiogenic pairs that provide relative and absolute ages, respectively. However, the initial abundance and half-life of the extinct radioactive isotope 146Sm are still debated, weakening the interpretation of 146Sm-142Nd systematics obtained for early planetary processes. Here, we apply the short-lived 26Al-26Mg, 146Sm-142Nd, and long-lived 147Sm-143Sm chronometers to the oldest known andesitic meteorite, Erg Chech 002 (EC 002), to constrain the Solar System initial abundance of 146Sm. The 26Al-26Mg mineral isochron of EC 002 provides a tightly constrained initial δ26Mg* of −0.009 ± 0.005 ‰ and (26Al/27Al)0 of (8.89 ± 0.09) × 10−6. This initial abundance of 26Al is the highest measured so far in an achondrite and corresponds to a crystallization age of 1.80 ± 0.01 Myr after Solar System formation. The 146Sm-142Nd mineral isochron returns an initial 146Sm/144Sm ratio of 0.00830 ± 0.00032. By combining the Al-Mg crystallization age and initial 146Sm/144Sm ratio of EC 002 with values for refractory inclusions, achondrites, and lunar samples, the best-fit half-life for 146Sm is 102 ± 9 Ma, corresponding to the physically measured value of 103 ± 5 Myr, rather than the latest and lower revised value of 68 ± 7 Ma. Using a half-life of 103 Ma for 146Sm, the 146Sm/144Sm abundance of EC 002 translates into an initial Solar System 146Sm/144Sm ratio of 0.00840 ± 0.00032, which represents the most reliable and precise estimate to date and makes EC 002 an ideal anchor for the 146Sm-142Nd clock.

The timescales of formation and differentiation of the early Earth (i.e., the Hadean period) and the Moon as well as terrestrial planets such as Mars are essential to generally understand the origin and evolution of planetary bodies (1). Various isotopic chronometers (e.g., U-Pb, Sm-Nd, Lu-Hf, Hf-W, and I-Pu-Xe) provide broad constraints on early differentiation processes such as core segregation, mantle stratification, and atmospheric and crust formation, and the extinct radionuclides are powerful tracers for the first million years of the Solar System (2–4). Among these systems, the rare-earth elements Sm and Nd play a unique role in geochemistry because they combine both short- (146Sm-142Nd), and long-lived (147Sm-143Nd) alpha decay systems. The short-lived 146Sm-142Nd decay system was active during the Hadean Era (>4 Ga). Therefore, measurements of 142Nd/144Nd were used to calculate model ages for the Earth’s mantle-crust differentiation (5, 6) and to determine the age of crystallization of the martian and lunar magma oceans (7–9). However, the significance of the ages obtained using this method is limited by the disputed estimates of the half-life and the initial Solar System abundance of 146Sm.

The 146Sm isotope is a pure p-process radionuclide that is possibly produced by the supernova of Type Ia events in the early Solar System (10). There are only four determinations of the half-life of 146Sm by particle counting experiments reported, with errors that show a range of values from 68 to 103 Ma (11–14). The latest determination at 68 ± 7 (1) Ma (14) is about a third shorter than the previously used value of 103 ± 5 Ma (12, 13), which casts further doubts on using 146Sm-142Nd systematics to obtain accurate timescales of early planetary differentiation processes. Each counting experiment may have possible systematic errors, making a consensus difficult to reach (15). For example, the different half-lives proposed for 146Sm result in differences of up to 100 Myr for magmatic events that occurred toward the end of the Hadean eon (4). Furthermore, the large variation of the 146Sm half-life weakens the potential for back-calculating the 146Sm/144Sm ratios of the initial Solar System using achondrites such as eucrites and angrites based on their absolute ages and 146Sm abundance (14, 16). Nowadays, the widely used Solar System initial 146Sm/144Sm ratio (0.00828 0.00044), on which all calculated ages are anchored, is currently based on data obtained on mineral separates from calcium-aluminum–rich refractory inclusions (CAIs) from carbonaceous (CC) chondrites (17). However, the Allende Al3S4 CAI, on which the initial 146Sm/144Sm was estimated, has not been dated by Al-Mg or Pb-Pb systems. Besides, it has a much younger Rb-Sr age of 4247 ± 110 Ma (17) than the reported Pb-Pb absolute ages of different refractory solids (18–20) and thus might have been modified by secondary processes and may not represent the earliest reservoir of the Solar System. Additionally, the radiogenic contribution to 142Nd might be compromised by nucleosynthetic anomalies (21, 22). In particular, the different mineral fractions analyzed in Allende Al3S4 CAI exhibit large and variable 148Nd and 150Nd anomalies (17), questioning the accuracy of the measured 142Nd/144Nd ratios and the validity of the deduced initial 146Sm/144Sm.

Erg Chech 002 (EC 002) is a unique silica-rich achondrite recognized as the oldest primordial crust sample discovered to date, with an estimated crystallization age of 2.255 ± 0.013 Myr after CAI formation obtained using in situ secondary ion mass spectrometry (SIMS) 26Al-26Mg measurements (23). The abundance of coarse crystals of pyroxene and plagioclase in EC 002 makes it possible to separate minerals with fractionated Sm/Nd ratios. Given its old reported age, large variations in 142Nd abundances are expected. Altogether, these characteristics make EC 002 a suitable igneous sample to estimate the 146Sm half-life and to establish the initial 146Sm/144Sm Solar System ratio by comparing its isotopic systematics with those of other asteroidal and planetary objects. Furthermore, the measurement of mass-independent Nd stable isotope ratios may be used to identify whether the EC 002 parent body is of CC or noncarbonaceous (NC) chondritic source (22, 24–26) and therefore helps to obtain more insights into the origin of EC 002.

In this study, we report the Sm and Nd isotopic compositions of mineral fractions, leachates, and bulk samples of EC 002, as well as a high-precision internal 26Al-26Mg isochron obtained on the same mineral fractions. A robust crystallization age for EC 002 is constrained by 26Al-26Mg and 147Sm-143Nd mineral isochrons. Literature 146Sm/144Sm data of CAIs and asteroidal and planetary bodies are then combined with the 146Sm/144Sm value of EC 002 to obtain the best fit value for the half-life of 146Sm. Finally, the 146Sm-142Nd isochron of EC 002 is used to deduce a precise and best estimate for the initial 146Sm/144Sm ratio of the Solar System.

Results

The Sm and Nd isotopic data and 144Sm/144Nd and 147Sm/144Nd ratios are reported in . Neutron capture effects caused by galactic cosmic rays are significant for this meteorite as evidenced by the Sm isotopic composition of the bulk rock (ε149Sm = −5.81 ± 0.05, ε150Sm = 10.46 ± 0.06). Therefore, all the measured Nd isotopic ratios were corrected for neutron capture effects using the method of Borg et al. (9) and the measured Sm isotope composition of EC 002. The corrections on µ142Nd, µ143Nd, µ145Nd, µ148Nd (µxNd = {(xNd/144Nd)sample/(xNd/144Nd)standards − 1} 106 in parts per million [ppm], with x = 142, 143, 145, or 148) are 2.5, 5.4, 3.9, and 0.7 ppm, respectively. After the neutron capture corrections, the two measured bulk-rock fractions provide an average µ145Nd and µ148Nd of 8.9 2.5 (2SD) and 12.0 3.6 (2SD) ppm, respectively. The ranges of 147Sm/144Nd and 143Nd/144Nd ratios of 15 EC 002 fractions are 0.1282 to 0.2349 and 0.510733 to 0.513784, respectively. The 147Sm-143Nd isochron (, mean square of weighted deviation [MSWD] = 130) returns an initial 143Nd/144Nd ratio of 0.50679 0.00018. An initial 143Nd value of 0.43 ± 0.83 is calculated using the chondritic uniform reservoir (CHUR) parameters of Bouvier et al. (27) and the method referred by Fletcher and Rosman (28). We note that the separation “D < 2.9 L (leachate of the mineral fraction D < 2.9)” falls off the correlation line for 143Nd/144Nd and 142Nd/144Nd ratios as defined by the seven other fractions (), which might be due to its low Nd content; therefore, we excluded it from the rest of the discussion for the 146Sm-142Nd system. The range of 142Nd/144Nd ratios of seven fractions of EC 002 was 1.141726 to 1.141880, and all 144Sm/144Nd and 142Nd/144Nd ratios defined a regression line (Fig. 1, MSWD = 0.99) with an initial 142Nd/144Nd ratio of 1.141479 0.000013 (142Nd0 = −17 11 ppm, compared to the terrestrial standard JNdi-1) and a 146Sm/144Sm ratio of 0.00830 0.00032.

Fig. 1.

146Sm-142Nd mineral and bulk-rock isochron defined by seven mineral, leachate, and bulk fractions of the EC 002. The slope corresponds to (146Sm/144Sm)0 = 0.00830 ± 0.00032 (2) and the intercept of 1.141479 ± 0.000013 (2) as initial 142Nd/144Nd (142Nd0 = −17 11 ppm, compared to the terrestrial standard JNdi-1). L, leachate. The detailed annotations of the fraction names can be found in .

The Mg isotopic data and 26Al/24Mg ratio are reported in for six separated fractions and one bulk sample. The 27Al/24Mg ratios ranged from 0.18 to 83.81, and the radiogenic 26Mg excesses (26Mg* = 26Mg-25Mg/, where xMg = Ln[(xMg/24Mg)sample/(xMg/24Mg)standards] 1000, x = 25 or 26; = 0.521) ranges from 0.009 ± 0.011 to 5.257 ± 0.015 ‰. The linear correction between 26Mg* and 27Al/24Mg (Fig. 2) defines an initial 26Mg/24Mg ratio (26Mg*0) of −0.009 0.005 ‰ (2) and a slope corresponding to a (26Al/27Al)0 of (8.89 0.09) 10−6. Using the half-life of 26Al of 0.705 Myr (29) and the (26Mg/24Mg)standard = 0.13932 (30), the (26Al/27Al)0 ratio of EC 002 corresponded to an age of 1.80 0.01 Myr after CAIs formation using the initial 26Al/27Al ratio defined by CAIs [5.23 10−5 (18)].

Fig. 2.

26Al-26Mg mineral and bulk-rock isochron defined by plagioclase, pyroxene, and bulk-rock fractions of the EC 002 meteorite (red line). The slope and the intercept of the isochron provide an initial (26Al/27Al)0 ratio of (8.89 ± 0.09) × 10−6 (2) and 26Mg*0 = −0.009 0.005 ‰ (2), respectively. When anchored to the “canonical” initial Solar System 26Al/27Al ratio of 5.23 10−5 (18), we obtained an age of 1.80 ± 0.01 Myr. The 26Al-26Mg isochron defined by in situ SIMS measurements is plotted for comparison (blue line). The slightly steeper slope from this study corresponds to an ∼450,000-y-older crystallization age for EC 002 compared to the age reported in Barrat et al. (23).

Discussion

EC 002 is the oldest andesitic achondrite recognized to date based on an initial 26Al/27Al ratio of (5.72 0.07) 10−6 determined from ion microprobe analyses of crystals of plagioclase and pyroxene exposed in a polished section of EC 002 (23). The isochron obtained from multicollector inductively coupled plasma mass spectrometer (MC-ICP-MS) analyses of separated minerals (pyroxenes and plagioclases) yielded a slightly steeperslope corresponding to an initial 26Al/27Al ratio of (8.89 0.09)10−6 (Fig. 2). This gives an age ∼450,000 y older than that of Barrat et al. (23). Furthermore, the 26Mg*0 (initial 26Mg/24Mg ratio) calculated here (−0.009 0.005 ‰) is slightly lower than in Barrat et al. (23) (0.065 0.080 ‰), while consistent within uncertainty but more precise. The small discrepancy between these two datasets cannot result from systematic analytical errors. To reconcile the two datasets, the error should be ∼120 ‰ on the 26Mg* measured by ion probe on plagioclase with 27Al/24Mg 5,000, or alternatively, the error on the 27Al/24Mg ratio should be larger than 1,500 with 26Mg* 200 ‰. Most likely, the two isochrons date two different events. This might be possible because of the much faster diffusion of Mg in plagioclase than in pyroxene [∼5 orders of magnitude faster; e.g., (31, 32)]. Thus, the core of the plagioclase crystal analyzed by ion microprobe is likely not a closed system for Mg diffusion, contrary to the plagioclase-rich fractions analyzed by MC-ICP-MS (see for detailed discussion). In this hypothesis, the MC-ICP-MS 26Al age would be closer to the crystallization age and will be used in the following discussion.

As a whole, we confirmed here that EC 002 is the oldest magmatic achondrite of the Solar System and that the small age difference based on two 26Al/27Al ratio estimates had no consequences for the discussion of the 146Sm-142Nd systematics. In addition, the initial distribution of 26Al is debated to be either homogeneous with a “canonical” value of ∼5 10−5 estimated from CAIs (18, 33) or to be heterogeneous with an initial ratio of ∼1 to 2 10−5 for the forming region of some groups of chondrules and the angrite parent body (34, 35). Using the two potential initial 26Al/27Al ratios, the 26Al/27Al ratio of EC 002 translates into an age of 1.80 0.01 Myr [anchored to a value of 5.23 10−5 (18)] or of 0.43 0.01 Myr [anchored to a reduced value of 1.36 10−5 (34)] after the formation of the Solar System [T = 4,567.30 0.16 Ma (20)]. Therefore, the absolute crystallization age of EC 002 could be considered to be between 4,565.5 and 4,566.9 Ma. However, such a small age range (<1.5 Myr) can only cause an ∼1% variation on the 146Sm initial abundance and half-life (much less than the error) and is thus insignificant for this study. Therefore, a crystallization age of 1.80 0.01 Myr after CAIs, obtained using an initial 26Al/27Al ratio of 5.23 10−5 as Barrat et al. (23) did, is used in the rest of the discussion.

The fundamental dichotomy between NC and CC chondrite groups, which supposedly reflect a difference between inner and outer Solar System materials, has been established based on elementary and isotopic anomalies of Cr, Ti, Mo, and other elements (36–39). Recently it has been found that chondrites and achondrites from NC and CC Reservoirs have different Nd isotopic compositions with a small but resolvable Nd isotopic “gap” (22, 25, 26). EC 002 displays a µ145Nd and µ148Nd nucleosynthetic composition of 8.9 2.5 (2SD) and 12.0 3.6 (2SD) ppm, respectively. The µ148Nd value of EC 002 is distinctly lower than any CC meteorite (Fig. 3) and falls within the NC region as other differentiated planetesimals such as angrite and eucrite parent bodies. This suggests that the parent body of EC 002 is likely derived from the NC meteorite reservoir. This conclusion is consistent with the fact that the negative Tm anomaly of the bulk EC 002 sample is similar to NC chondrites and achondrites (23).

Fig. 3.

µ145Nd-µ148Nd compositions of EC 002 compared with the mean compositions of chondrite groups (22, 24, 25, 40), achondrite groups (22, 26, 40, 41), lunar samples (3, 9, 42–44), and martian meteorites (26, 40, 45–48). Compiled µ145Nd and µ148Nd of meteorite groups and planetary bodies are from Frossard et al. (26). CC chondrites and achondrites (NWA 6704 and Tafassasset) have generally higher 148Nd compared to inner Solar System and NC materials (26). EC 002 falls within the range of the NC Reservoir and also within the group of achondrite parent bodies (the outlier NWA 5363 has large errors) formed within 1.5 Myr after CAIs (26).

The 15 mineral and bulk fractions of EC 002 define a 147Sm-143Nd age of 4,521 152 Ma, which is consistent despite its large uncertainty with its corresponding Al-Mg age. The initial 143Nd value of 0.43 ± 0.83 suggests that EC 002 is derived from a chondritic source. The slope and the intercept of 144Sm/144Nd and 142Nd/144Nd ratios defined by the seven mineral fractions and bulk samples correspond to an initial 146Sm/144Sm and 142Nd/144Nd of 0.00830 0.00032 and 1.141479 0.000013, respectively (Fig. 1). The precise Al-Mg age and 146Sm/144Sm ratio of the EC 002 parent body can be used to constrain the initial Solar System 146Sm/144Sm ratio, which, however, is strongly affected by two very different half-lives used in the scientific community [68 and 103 Ma (15)]. On the basis of the same set of data from achondrites (eucrite, mesosiderite, and angrite), the initial Solar System 146Sm/144Sm values were estimated as 0.0094 ± 0.0005 (14) or 0.0085 ± 0.0007 (49) using the 146Sm half-life of 68 and 103 Ma, respectively. Moreover, the large differences of the 146Sm half-life estimate not only hinder the reliability of the initials but also impede the potential of 146Sm-142Nd systematics for early Solar System chronology. Therefore, we provide an independent method using the coupled fit lines of crystallization ages from other robust dating systems (147Sm-143Nd, Pu-Xe, Pb-Pb, and 26Al-26Mg; see ) and 146Sm/144Sm ratios of the meteorites, as well as the evolution trends of 146Sm/144Sm back-calculated from EC 002 to estimate the value of the 146Sm half-life. The temporal evolution line of the Ln(146Sm/144Sm) ratio (Fig. 4) is modeled with the slope (−, where represents decay constant) corresponding to either a 146Sm half-life of 103 Ma or 68 Ma and a certain point constrained by EC 002 (Age = 1.80 Ma and Ln(146Sm/144Sm) = −4.79). In Fig. 4, the Ln(146Sm/144Sm) ratios and ages of CAIs and achondrites are plotted together. All the points, especially the young lunar samples, fall closer to the evolution trend based on the 146Sm half-life of 103 Ma (12, 13) compared to that of 68 Ma (14) (Fig. 4). Using all the scatters, we further defined the best fit curve and its confidence interval based on regression fitting. For the regression calculation, we converted the 146Sm/144Sm ratios of CAIs, achondrites, and lunar samples to Ln(146Sm/144Sm), which we assumed to be linearly correlated with their ages [taking the Solar System age as T = 4,567.3 Ma (20)]. The linear regression of the values and errors of Ln(146Sm/144Sm) and ages gives a slope of −0.00680 0.00061 and an intercept of −4.80 0.03 (Ln(146Sm/144Sm) initial) using the IsoplotR Model 3 (50). The slope translates to the best fit line passing through all the data points (Fig. 4) fit to a half-life of 102 9 Ma (95% confidence interval), providing robust support for the experimentally determined 146Sm half-life value of 103 Ma, which is 35 Myr longer than the most recent determination (14). The combination of different radioactive systematics in CAIs, asteroidal, and planetary samples of various origins provides consistent results and an independent constraint on the half-life. It is important to stress the significance of the lunar samples (17) in the determination of the correct half-life, as the very ancient samples such as EC 002 actually work as an end-member for a more precise constraint in this approach (Fig. 4). The 35-Myr difference in the 146Sm half-life affects the age of the early mantle differentiation and the age of differentiation of the lunar magma ocean (e.g., 9, 51) by over 100 Myr (52). Using the 103-Ma half-life confirmed here eliminates a large source of uncertainty of these age determinations.

Fig. 4.

Temporal evolution trend of the Ln(146Sm/144Sm) ratios using Ln(146Sm/144Sm) and age of EC 002 with a 146Sm half-life of 68 Ma (blue line) and of 103 Ma (red line). For comparison, data for CAIs, eucrites, angrites, mesosiderite, and lunar samples () are reported. Using the linear correlation between Ln(146Sm/144Sm) and corresponding ages of these meteorites, the IsoplotR Model3 (50) provides the slope and intercept with respective errors for the best fit line (gray solid line) and its 95% confidence interval (95% C.I., gray dashed lines), corresponding to a fitted 146Sm half-life value and Solar System initial 146Sm/144Sm ratio of 102 9 Ma and 0.0082 0.0003, respectively.

By combining the 26Al-26Mg crystallization age of EC 002 of 1.80 0.01 Myr after CAI formation with the 146Sm/144Sm abundance recorded by the same mineral fractions, we deduced an initial 146Sm/144Sm for the Solar System of 0.00840 0.00032 and the temporal evolution curve of 146Sm abundance () using the 146Sm half-life of 103 Ma (12, 13). This initial Solar System 146Sm/144Sm value is consistent within error with previous estimates such as the ones determined from eucrites of 0.0084 0.0005 (T = 103 Ma) (49), a combination of eucrites and angrites of 0.0085 0.0005 (T = 103 Ma) (16), and Allende CAI Al3S4 of 0.00828 0.00044 (17), albeit with better precision. Moreover, our 146Sm/144Sm value corresponds to (146Sm/144Sm)initial of 0.0082 ± 0.0003 of the fitted line (Fig. 4), which is compiled using the most reliable coexisting ages and 146Sm-142Nd data of various extraterrestrial samples. The consistency of the initial 146Sm abundance estimated from different meteorite materials supports the previous suggestions (16, 49) that the distribution of 146Sm was homogeneous in the early Solar System, at least for the CAIs and NC-forming regions. The initial Solar System 146Sm/144Sm abundance of 0.00840 0.00032 deduced from EC 002 is therefore the most reliable and precise estimate to use for 146Sm-142Nd chronology.

In conclusion, we obtained 26Al-26Mg and 146,147Sm-142,143Nd mineral internal isochrons for the EC 002 ungrouped achondrite meteorite. The Al-Mg systematics confirmed that EC 002 is the oldest andesitic meteorite with a formation age at 1.80 0.01 Myr after the Solar System formation [when anchored to an initial 26Al/27Al ratio of 5.23 10−5 (18)]. Stable Nd isotopic anomalies indicate that the parent body of EC 002 formed within the NC Reservoir, in agreement with the negative Tm anomaly displayed by the whole rock (23). By combining chronological records with the 146Sm-142Nd systematics of extraterrestrial samples formed within the first 200 Myr of Solar System history [especially lunar samples formed at 4.30 to 4.36 Ga (9, 51)], we confirmed that the most consistent half-life of 146Sm is 103 Ma. Using this half-life, the formation age of EC 002, and the 146Sm/144Sm mineral isochron slope, we deduced a 146Sm/144Sm initial Solar System ratio of 0.00840 0.00032. In association with its large recovered mass (∼32 kg), abundant coarse mineral grains, and trace element–enriched composition, EC 002 represents the best anchor for 146Sm-142Nd systematics in planetary materials and possibly for other short-lived radioactive decay systems.

Materials and Methods

A mass of ∼4 g of EC 002 was crushed in an agate mortar dedicated to meteorite work, from which several mineral and bulk-rock fractions were separated. Thirteen of them (seven mineral fractions and six leachates) in addition to two bulk-rock powders were analyzed for Sm and Nd isotope compositions. We measured unspiked isotope compositions for stable Sm, Nd, and radiogenic Nd isotopic ratios and used the isotope dilution method for determining precisely the 147Sm/144Nd of all the fractions [following the splitting method detailed in Bouvier and Boyet (21)]. Among these samples, eight fractions with high Nd contents were analyzed for Nd isotopic compositions independently once again. Six mineral fractions and one bulk rock were analyzed for their Al-Mg isotopic compositions. The mineral fractions were separated using a combination of hand magnet, magnetic separator, and heavy liquids of the density of 2.9 and 3.3 (see ). In addition, three widely available terrestrial samples (United States Geological Survey, basalt BHVO-2, BCR-2, and andesite AGV-2) were also analyzed to monitor data quality. All the fractions were passed through columns with ion exchange and specific resins in several steps to discard matrix and purify solutions for Nd and Mg (see details in ). Neodymium and Sm isotope compositions were analyzed at the Laboratoire Magmas et Volcans, and detailed methods can be found in the . The Mg isotopic composition and 27Al/24Mg ratio were analyzed at the Institut de Physique du Globe de Paris (IPGP) following the method described in the . Both 146Sm-142Nd, 147Sm-143Nd and 26Al-26Mg internal isochrons were calculated using IsoplotR, Model1 (50).