Available online at www.sciencedirect.com ScienceDirect

Geochimica et Cosmochimica Acta 213 (2017) 593–617 www.elsevier.com/locate/gca

Distinct 238U/235U ratios and REE patterns in plutonic and volcanic angrites: Geochronologic implications and evidence for U isotope fractionation during magmatic processes

Franc¸ois L.H. Tissot a,b,⇑, Nicolas Dauphas a, Timothy L. Grove b

a Origins Laboratory, Department of the Geophysical Sciences and Enrico Fermi Institute, The University of Chicago, 5734 South Ellis Avenue, Chicago, IL, USA b Department of the Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA

Received 28 December 2016; accepted in revised form 28 June 2017; Available online 8 July 2017

Abstract

Angrites are differentiated meteorites that formed between 4 and 11 Myr after Solar System formation, when several short- lived nuclides (e.g., 26Al-26Mg, 53Mn-53Cr, 182Hf-182W) were still alive. As such, angrites are prime anchors to tie the relative chronology inferred from these short-lived radionuclides to the absolute Pb-Pb clock. The discovery of variable U isotopic composition (at the sub-permil level) calls for a revision of Pb-Pb ages calculated using an ‘‘assumed” constant 238U/235U ratio (i.e., Pb-Pb ages published before 2009–2010). In this paper, we report high-precision U isotope measurement for six angrite samples (NWA 4590, NWA 4801, NWA 6291, Angra dos Reis, D’Orbigny, and Sahara 99555) using multi- collector inductively coupled plasma mass-spectrometry and the IRMM-3636 U double-spike. The age corrections range from 0.17 to 1.20 Myr depending on the samples. After correction, concordance between the revised Pb-Pb and Hf-W and Mn- 53 55 Cr ages of plutonic and quenched angrites is good, and the initial ( Mn/ Mn)0 ratio in the Early Solar System (ESS) is recal- culated as being (7 ± 1) 106 at the formation of the Solar System (the error bar incorporates uncertainty in the absolute age of Calcium, Aluminum-rich inclusions – CAIs). An uncertainty remains as to whether the Al-Mg and Pb-Pb systems agree in large part due to uncertainties in the Pb-Pb age of CAIs. A systematic difference is found in the U isotopic compositions of quenched and plutonic angrites of +0.17‰. A difference is also found between the rare earth element (REE) patterns of these two angrite subgroups. The d238U values are consistent with fractionation during magmatic evolution of the angrite parent melt. Stable U isotope fractionation due to a change in the coordination environment of U during incorporation into pyroxene could be responsible for such a fractionation. In this con- text, Pb-Pb ages derived from pyroxenes fraction should be corrected using the U isotope composition measured in the same pyroxene fraction. Ó 2017 Elsevier Ltd. All rights reserved.

Keywords: Angrites; U stable isotopes; Pb-Pb ages; Short-lived chronometers; U stable isotope fractionation

1. INTRODUCTION

⇑ Angrites are differentiated meteorites (achondrites) of Corresponding author at: Department of the Earth, Atmo- basaltic composition. They are of either volcanic (a.k.a. spheric and Planetary Sciences, Massachusetts Institute of Tech- quenched) or plutonic origin and display minimal post- nology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA. crystallization alteration, metamorphism, shock or impact E-mail address: [email protected] (F.L.H. Tissot). http://dx.doi.org/10.1016/j.gca.2017.06.045 0016-7037/Ó 2017 Elsevier Ltd. All rights reserved. 594 F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 brecciation (e.g., Keil, 2012 and references therein). Their assumption of Pb-Pb dating. Because variation in the abso- old U-Pb (e.g., Amelin, 2008a, 2008b) and Pb-Pb ages lute 238U/235U ratio will impact absolute ages, both Pb and (e.g., Wasserburg et al., 1977; Lugmair and Galer, 1992; U isotopic compositions need to be measured to obtain pre- Baker et al., 2005; Connelly et al., 2008) make angrites cise and accurate Pb-Pb ages. The first goal of the present some of the earliest volcanic rocks known in the Solar Sys- paper is to provide high-precision U isotope data on a large tem, with crystallization ages for the oldest samples of 4 array of angrites in order to correct their Pb-Pb ages. These Myr after the formation of Calcium, Aluminum-rich inclu- U-isotope-corrected ages will be used to assess the degree of sions (CAIs, the oldest known solids in the Solar System). concordance between short-lived nuclides systems As such, these achondrites provide insights into early stages (26Al-26Mg, 53Mn-53Cr, 182Hf-182W) and the absolute Pb- of planetary melting and differentiation. They also play an Pb clock. important role as anchors for early solar system (ESS) The second aim of this study is concerned with identify- chronology. Indeed, the quenched angrite specimens have ing whether all angrites have a similar U isotopic composi- estimated cooling rate between 7 and 50 °C/h (Mikouchi tion, and, if not, what is(are) the process(es) responsible for et al., 2000, 2001), which means that it took 18–127 h this variability. Brennecka and Wadhwa (2012) measured a for them to cool from a liquidus temperature of 1190 °C series of angrite samples and suggested that all angrites had (Longhi, 1999) to below the closure temperature of a homogeneous U isotopic composition. They reached this 300 °C relevant to the decay systems discussed in this con- conclusion because they propagated the uncertainties on tribution (Pb: in clinopyroxene Cherniak, 1998, in plagio- the U isotopic composition of the two U double spikes that clase, Cherniak, 1995, in apatite, Cherniak et al., 1991; they used onto the final 238U/235U ratio of each sample. Cr: in olivine, Ito and Ganguly, 2006, in spinel, Posner However, the same double spike (IRMM-3636) was used et al., 2016; Mg: in olivine, Chakraborty, 1997, in plagio- for analysis of all bulk samples, while another spike (in- clase, Muller et al., 2013, in cpx, Van Orman et al., 2014; house ASU spike) was only used to measure a replicate of W: in olivine and model pyroxene, Cherniak and Van D’Orbigny and two mineral separate fractions (D’Orbigny Orman, 2014). Hence, most radiochronometric systems, pyroxenes, AdoR phosphates). Any error on the spike com- and in particular those commonly used in ESS chronology, position will shift the 238U/235U ratios of the samples by a closed simultaneously in quenched angrites (Dodson, 1973). constant value, so that difference in the U isotopic compo- Their mineralogy is also diverse, which makes them amen- sitions of samples corrected by the same double spike are able to dating with various chronometers. Because few sam- better known than one would be led to believe if uncertain- ples meet the requirements of non-disturbance, ties on the spike composition are propagated. When only synchronous isotope closure and phase diversity like the the isotope measurement error is considered, some variabil- quenched angrites, they have become important samples ity in the angrite 238U/235U ratio data set of Brennecka and to cross-calibrate short-lived dating techniques (e.g., Wadhwa (2012) is visible. It is presently unknown whether 26Al-26Mg, 53Mn-53Cr, 182Hf-182W) with absolute dating igneous processes can fractionate U isotopes but the U iso- techniques (i.e., U-Pb, Pb-Pb) (e.g., Nyquist et al., 2009; topic variability of 0.50‰ measured among igneous rocks Dauphas and Chaussidon, 2011; Kleine et al., 2012). As (see compilation in Fig. 6 of Tissot and Dauphas, 2015) the precision of isotope measurement improves, it is possi- raises the possibility that magmatic differentiation on the ble to test at finer levels the concordance of these angrite parent-body could have fractionated U isotopes. chronometers. For instance, and based on a 1.5 Myr dis- Here, we report high-precision U isotopic data for six crepancy between the Al-Mg and Pb-Pb ages of CAIs and angrite samples: NWA 4590, NWA 4801, NWA 6291, quenched angrites, Larsen et al. (2011) proposed that 26Al Angra dos Reis, D’Orbigny, and Sahara 99555. Using this was not uniformly distributed in the Solar System, thus data set and previously published measurements questioning one of the pillars of ESS chronology. To settle (Brennecka and Wadhwa, 2012; Connelly et al., 2012; the vigorous debate around this issue, improvement of the Goldmann et al., 2015), we question the homogeneity of absolute and relative ages of ESS anchors are therefore the U isotope composition of angrite specimens, and discuss critical. the possible processes by which different angrite samples can The most precise absolute ages of ESS objects are acquire different U isotopic compositions. After correcting obtained using the Pb-Pb dating method (Patterson et al., the Pb-Pb ages of angrites, we test their age concordance 1955; Patterson, 1956): a dual chronometer since 238U with short-lived chronometers and discuss the implications decays into 206Pb with a half-life of 4468.3 ± 4.8 Myr while for the distribution of short-lived radionuclides in the ESS. 235U decays into 207Pb with a half-life of 703.81 ± 0.96 Myr (Jaffey et al., 1971). In the late 1970s, to ensure that abso- 2. SAMPLES AND METHODS lute ages could be compared between laboratories, and based on the observed constancy of the 238U/235U ratio in All Teflon labware used in this study (i.e., PFA vials and natural samples, a ‘‘consensus” value of 238U/235U= beakers from Savillex) was pre-cleaned with boiling aqua 137.88 was adopted for use in geochronology (Steiger and regia (3:1 mixture of HCl:HNO3) three times, followed by Jager, 1977). In the past decade, however, a series of studies boiling MQ water. Some reagents and elution cuts were showed that resolvable variations exist in the 238U/235U temporarily stored in polypropylene centrifuge tubes ratio of natural samples (e.g., Stirling et al., 2007; Weyer (Corning). Before use, all centrifuge tubes were rinsed three et al., 2008; Bopp et al., 2009; Brennecka et al., 2010a; times with MQ water and/or leached with 50% (vol) HCl Tissot and Dauphas, 2015), thereby overthrowing a key overnight. F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 595

2.1. Sample selection, preparation and digestion HCl/HNO3 (2:1) on a hot plate for 5 days. The solutions were then dried down and re-dissolved in a low volume of Angrites have relatively low uranium concentrations 11 M HCl before being loaded on U/Teva resin. In the elu- (from 20 to 200 ppb), thus large sample masses are needed tion scheme used by Tang and Dauphas (2012), U and Fe to obtain high-precision U isotopic data. The samples stud- elute together, which is the elution cut that was used for ied here include two volcanic (quenched) angrites U analysis in the present study. As with other samples, (D’Orbigny and Sahara 99555), and four plutonic angrites these solutions were spiked with IRMM-3636, evaporated (NWA 4590, NWA 4801, NWA 6291, and Angra dos Reis). to near dryness, taken back in 3 M HNO3, and placed on Among these, Angra dos Reis (hereafter AdoR) is the only a hot plate at 140 °C overnight to allow for sample/spike observed fall. equilibration. The hand specimens were cut using a Buehler Isomet rock saw. Samples were inspected for signs of alteration 2.2. Uranium purification and mass spectrometry (e.g., veins) and recut or hand broken so as to retain only the least altered areas. Sample pieces were placed in a U separation and purification was done on 2mL car- Teflon beaker and cleaned for 5 min in an ultrasonic bath tridges (l = 2.7cm, £ = 0.8 cm) of U/Teva specific resin, of methanol. Each sample was then crushed manually in following the procedure described previously in Telus an agate mortar dedicated to meteoritic samples that was et al. (2012) and Tissot and Dauphas (2015). The resin triple-cleaned before each sample by powdering fine silica was cleaned with 20 mL of 0.05 M HCl and conditioned grains followed by rinsing with MQ water and methanol. with 6 mL of 3 M HNO3. The samples were loaded onto The sample powder was transferred into a clean Teflon bea- the column in 15–30 mL of 3 M HNO3 and most matrix ele- ker and spiked with the IRMM-3636 spike, which is made ments (except Th and U) were removed in 40 mL of 3 M 233 236 238 235 of 50.45% U and 49.51% U( U/ U = 5.1629 HNO3. The resin was then converted to HCl with 6 mL ± 0.0118; Verbruggen et al., 2008). Based on U concentra- of 11 M HCl and Th was eluted with 20 mL of 5 M HCl tions reported in the literature, enough spike was added to + 0.1 M oxalic acid, followed by 10 mL of 5 M HCl to rinse each sample to obtain a Uspike/Usample ratio of 3%, which off the oxalic acid. The final step was the elution of U in 22– minimizes the consumption of spike, the abundance sensi- 25 mL of 0.05 M HCl. The whole procedure was repeated tivity effect of 238U onto 236U, as well as the contribution two to three times, depending on the mass of sample to the measurement uncertainty from amplifier noise and digested (Table 1), to ensure complete matrix removal. counting statistic on 233U and 236U(Weyer et al., 2008). After chemistry, the U cuts were dried down completely, For NWA 4801, NWA 6291, D’Orbigny and Sahara re-dissolved in concentrated HNO3, before being evapo- 99555, the masses digested ranged from 190 mg to 1.4 g rated to near dryness and taken back in 0.3 M HNO3 for (Table 1). The samples were first treated with two acid isotopic analysis. attacks (6 days and 16 days, respectively) in HF/HNO3/ All U isotope measurements were performed on a HClO4 (3:1: several drops) on a hot plate at 160 °C. ThermoFinnigan Neptune MC-ICPMS upgraded with an Approximately 4 mL of that acid mixture was used per OnTool Booster 150 Jet pump (Pfeiffer) at the Origins Lab- 100 mg of sample. Because undigested particles were still oratory of the University of Chicago. A set of Jet sample visible after the second digestion step, the samples were cen- cones and X-skimmer cones were used, in combination with trifuged in clean centrifuge tubes, and the supernatant was an Aridus II desolvating nebulizer for sample introduction. pipetted out and transferred back into the original Teflon Enhanced signal stability was achieved by placing a spray beaker used for digestion. The residue of each sample was chamber between the Aridus II and the MC-ICPMS. All transferred into a clean 6 mL Teflon beaker and placed in measurements were done in low-resolution mode, using a a Parr Bomb in an oven for 8 days in HF/HNO3 (2:1) at static cup configuration. They involved 50 cycles of 180 °C, to ensure complete dissolution of the refractory 4.194 s integration time each (for more details, see Tissot phases. After the bomb step, the two fractions were recom- and Dauphas, 2015). The measurements were done at U bined and treated with two acid attacks (3 days and 5 days, concentrations between 14 and 30 ppb in 0.5 to 10 mL solu- respectively) in HCl/HNO3 (3:1) on a hot plate at 160 °C. tions, using a 100 ll/min PFA nebulizer. The sensitivity of At this point, the sample was dried completely, re- the instrument was 1 V/ppb on 238U measurement with a 11 dissolved in concentrated HNO3 and put back on a hot 10 ohm resistance amplifier. Baseline and gain calibra- plate for 3 days. The sample was then diluted to 3 M tions were done at least daily. The procedural blank was HNO3, and placed on a hot plate for 3 days. No particle estimated to be 0.06–0.17 ng U (between 0.1 and 0.7% of was visible after this last digestion step. sample uranium). Samples NWA 4590, AdoR and a replicate preparation The isotopic mass fractionation introduced during of NWA 4801 and D’Orbigny (all denoted by a star in chemical separation and mass spectrometry was corrected Table 1) were obtained as solution recovered from a previ- for using the 233U/236U double spike IRMM-3636 and the ous chemistry meant to purify nickel. The details of the data reduction methodology is described in detail in digestion and purification procedure for these samples can Tissot and Dauphas (2015). In brief, the raw signals are be found in Tang and Dauphas (2012). In brief, these sam- corrected for (i) on peak zero, (ii) the 238U tail contribution ples were crushed in an agate mortar, and digested in Teflon onto the 236U, 235U and 234U signals (respectively, 6 6 6 238 beakers using an acid attack in HF/HNO3 (2:1) on a hot 0.6 10 , 0.25 10 and 0.1 10 of the U signal plate at 90 °C for 10 days, followed by an acid attack in intensity), (iii) hydride formation, and (iv) the decay of 9 ...Tso ta./Gohmc tCsohmc ca23(07 593–617 (2017) 213 Acta Cosmochimica et Geochimica / al. et Tissot F.L.H. 596

Table 1 Summary of U isotopic compositions and concentrations of selected angrites. Sample Resin #a Mass (mg) n Blank contrib. Double-spike data reduction Sample-standard bracketing Literature data d238 Ub (‰) d [234 U/ 238 U]c Conc. (ng/g) Usp/Usmp d238 Ub (‰) d [234 U/ 238 U]c Conc. (ng/g)d Ref. Plutonic * NWA 4590 2 389.3 5 0.24% 0.47 ± 0.09/0.18 241.6 ± 2.9 68 ± 3 5.3% 0.37 ± 0.33 238.5 ± 5.3 96 ± 4 (1) * NWA 4801 2 258.8 4 0.28% 0.65 ± 0.10/0.19 73.8 ± 2.4 81 ± 8 2.7% 0.53 ± 0.37 75.1 ± 4.2 94 ± 1 (1) NWA 4801 rep. 3 549.0 7 0.31% 0.50 ± 0.07/0.17 55.2 ± 2.5 87 ± 4 2.8% 0.55 ± 0.10 57.8 ± 3.6 94 ± 1 (1) NWA 4801 average 0.55 ± 0.06/0.17 0.54 ± 0.10 NWA 6291 2 189.4 3 0.33% 0.30 ± 0.12/0.20 472.3 ± 3.3 98 ± 6 4.7% 0.19 ± 0.42 463.9 ± 6.1 156 ± 8 (2) NWA 6291 rep. 3 1390.3 14 0.09% 0.41 ± 0.04/0.16 511.7 ± 3.5 128 ± 1 3.9% 0.45 ± 0.06 507.4 ± 4.6 156 ± 8 (2) NWA 6291 average 0.40 ± 0.04/0.16 0.45 ± 0.06 * Angra dos Reis 2 141.0 6 0.17% 0.51 ± 0.08/0.18 0.0 ± 2.2 238 ± 17 2.3% 0.72 ± 0.30 1.4 ± 4.5 200 ± 10 (1) Volcanic (Quenched) * D’Orbigny 2 213.8 3 0.32% 0.33 ± 0.12/0.20 41.1 ± 2.3 84 ± 6 4.3% +0.07 ± 0.42 41.8 ± 4.6 82 ± 1 (1) D’Orbigny 3 299.9 3 0.70% 0.31 ± 0.14/0.21 64.7 ± 2.6 64 ± 3 6.2% 0.45 ± 0.19 65.8 ± 4.8 82 ± 1 (1) D’Orbigny average 0.32 ± 0.09/0.18 0.36 ± 0.18 Sahara 99555 3 200.0 3 0.71% 0.23 ± 0.14/0.21 21.1 ± 2.4 94 ± 6 4.4% 0.37 ± 0.19 22.1 ± 4.9 106 ± 3 (1) References: (1) Riches et al. (2012);(2)Brennecka and Wadhwa (2012). All measurements were made using: Aridus II + Spray Chamber + Jet Cones. Cup configuration for 233/234/235/236/238U is L1/SEM/H1/H2/H3. Tailing from 238 Uon236 U, 235 U and 234 U was estimated as, respectively, 0.6 ppm, 0.25 ppm and 0.1 ppm. Hydride formation was corrected using the value 238 UH/238 U = 7.3e7 (see Tissot and Dauphas (2015) for more details). * Denotes samples that were elution cuts recovered from a previous chemistry meant to purify Ni. a Successive number of times the sample was purified by column chemistry on U/Teva resin. b Values normalized to CRM-112a. The values are corrected for blank contribution. The effect of this correction on the d238 U value is less than 0.01‰. c Activity ratio [234 U/238 U] of the sample relative to secular equilibrium (in ‰). d [234 U/238 U] = {(234 U/238 U)smp/(234 U/238 U)eq1} * 1000 where (234 U/238 U)eq is the atomic ratio at secular equilibrium and is equal to k238 U/k234 = 5.497e5[Cheng et al. (2013)]. d When no value is available for the uncertainty of the recommended concentration, a value of 5% was assumed (noted in italic). F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 597 the spiked isotopes (233U and 236U) between the date when New Brunswick certificate). These three values cover a the spike was calibrated and the date of the sample mea- range of 0.09‰, and the effect of such a difference on the surement. An additional correction was applied by bracket- final Pb-Pb age of a sample can easily be calculated using ing samples with standard measurements spiked with Eq. (13) (for more details see Tissot and Dauphas, 2015). IRMM-3636 at the same level as the samples. This correc- For a 4.5 Gyr old sample, this 0.09‰ spread in tion is needed in order to obtain reliable absolute isotopic 238U/235U ratio corresponds to a 0.13 Myr shift in absolute ratios. It was observed by Tissot and Dauphas (2015) that Pb-Pb age (Fig. 1), a value equivalent to the uncertainty of the choice of the cup configuration (i.e., which isotope is Pb-Pb ages of some angrite samples. Because in ESS Pb-Pb being measured on the axial faraday cup) and the set of dating, differences in absolute ages is the quantity that mat- cones used to perform the measurement can result in sys- ters, the exact composition of the standard will not matter tematic offsets of up to 0.20‰ in the final 238U/235U ratio as long as all 238U/235U ratios are calculated relative to the measured. Sample-standard bracketing allows one to cor- same standard composition (here 137.837 for CRM-112a). rect for such bias. 3.2. Reporting of uncertainties 3. RESULTS The sources of errors in U-Pb geochronology are of two 3.1. Notations and U standard 238U/235U ratio kinds: those that can systematically bias ages but do not affect relative ages (i.e., age intervals), and the random errors U isotope data are reported as either absolute ratios or that affect both absolute and relative ages. The systematic d238U values relative to the U standard CRM-112a (also errors, which affect all samples in a similar fashion, are asso- named SRM960 or NBL112-a; CRM-145 for the solution ciated with the composition of the U double spike, that of the form): Pb spike or age standard, and the errors on the half-lives of 235U and 238U, while the random uncertainties are specific to d238U ¼½ð238U=235UÞ =ð238U=235UÞ 1103: sample CRM112a each sample and associated with the U and Pb isotopic mea- ð1Þ surements. If random errors should always be considered when comparing d238U values and U-isotope-corrected Pb- Internal errors were smaller or equal to the externalpffiffiffi error and errors are thus calculated as 2 rStandard = n, Pb ages, systematic errors only need to be included in specific where 2 rStandard is the daily external reproducibility of cases. For the sake of clarity, the notation commonly used in repeat measurements of the standard CRM-112a bracketed U-Pb geochronology (e.g., Schoene, 2014) was adopted by itself, and n is the number of sample solution measure- hereafter, and errors will be reported in the form ± x/y/z. ments. A sample, NWA 6291, prepared twice from fresh Table 2 provides the detail of the errors included in x, y powder aliquots gives reproducible results within errors: and z for the different variables discussed in this paper, as d238U=0.30 ± 0.12 vs. 0.41 ± 0.04‰. Similarly, the well as the domain of relevance of each type of error. measurements made on U recovered from elution cuts from Tang and Dauphas (2012) give results in agreement to those 3.3. Decay-constant uncertainties on absolute ages vs. age made on fresh powder aliquots (i.e., d238U=0.65 ± 0.10 intervals vs. 0.50 ± 0.07‰ for NWA 4801 and d238U=0.33 ± 0.12 vs. 0.31 ± 0.14‰ for D’Orbigny), which shows To compare absolute ages obtained from two different that the column procedure of Tang and Dauphas (2012) chronometers (e.g., Pb-Pb and Rb-Sr) decay-constant did not fractionate U isotopes. The [234U/238U] activity ratios are reported as d 4564.5 [234U/238U] relative to secular equilibrium: 4564.0 d½234 =238 ¼½ð234 =238 Þ =ð234 =238 Þ 3; U U U U sample U U eq 1 10 ð Þ 2 4563.5 234 238 where ( U/ U)eq is the atomic ratio at secular equilib- rium and is equal to the ratio of the decay-constants of 238 234 10 6 4563.0 Uand U, k238/k234 = (1.5513 10 )/(2.8220 10 ) = 5.497 105 (Cheng et al., 2013). External errors were

Age of D'Orbigny (Myr) 4562.5 between 1 and 1.5‰ but the internal errors were larger, CRM certificate between 2.2 and 3.5‰. To be conservative, and though this Cond. (2010) Richter (2010) difference is insignificant compared to the size of the 234 238 d[ U/ U] values measured, we report the internal errors 137.75 137.80 137.85 137.90 137.95 in Table 1. 238 235 All 238U/235U absolute ratios used in this study (includ- U/ U ratio of CRM-112a ing literature data) have been calculated (or recalculated) 238 235 Fig. 1. Age of D’Orbigny angrite as a function of the absolute assuming a U/ U ratio of 137.837 for CRM-112a 238U/235U ratio used for the standard CRM-112a. The symbols (Richter et al., 2010). Two other recent estimates of the represent the values recommended by (upward) Richter et al. 238 235 U/ U ratio of CRM-112a exist: 137.844 ± 0.024 (2r, (2010), Condon et al. (2010), and the New Brunswick CRM-112a Condon et al., 2010) and 137.849 ± 0.076 (2r, CRM-112a certificate (2002). 598 F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617

Table 2 Error notation. Variable (unit) 238U/235U ratios, d238U(‰) Uncorrected Pb-Pb age (Myr) U-corrected Pb-Pb ages and Dt age corrections (Myr) and ages intervals (Myr) Error reported as: ±x/y ±x/y/z ±x/y/z x includes errors on: U iso. analysis Pb iso. analysis U iso. analysis Pb iso. analysis Pb spike or age standard compo. y also includes errors on: U spike compo. [±0.16‰ for Pb spike or age standard U spike compo. IRMM-3636, Richter et al. compo. (2008), and ±0.15‰ for in-house double spike used by Brennecka and Wadhwa (2012)] z also includes errors on: 235U decay constant 235U decay constant 238U decay constant 238U decay constant Domain of relevance x: for comparison of U isotope x: for comparison of x: for comparison of ages and age compositions determined using uncorrected ages determined intervals determined using the same U the same double spike. using the same spike/age double spike (and any Pb spike/age standard. standard). y: otherwise. y: for comparison of y: for comparison of ages and age uncorrected ages determined intervals determined using different U using different spikes/age double spikes (and any Pb spike/age standards. standard). z: for comparison with ages z: for comparison with ages and age derived from other intervals derived from other chronometers (e.g., Rb-Sr). chronometers (e.g., Rb-Sr, Al-Mg).

k k errors have to be taken into account. When comparing rel- ðte 238tÞðe 235t 1Þ C ;t ¼ ð Þ 8 k t k t ðk þk Þt 8 ative ages, however, ‘‘the errors of decay-constants almost ðk238e 238 k235e 235 þðk235 k238Þe 235 238 Þ completely cancel and should be excluded” (Amelin, 2006). k t k t The importance of the decay-constant errors of 238U and ðe 235 1Þðe 238 1Þ Cu;t ¼ : ð Þ 235 k t k t ðk þk Þt 9 (mostly) U onto the absolute Pb-Pb age was carefully uðk238e 238 k235e 235 þðk235 k238Þe 235 238 Þ investigated by Ludwig (2000) and is large compared to Note that C2 r2 is equivalent to Eq. (13) below (Eq. 12 the precision of the Pb isotope ratio measurement (for a u;t u 235 238 of Tissot and Dauphas, 2015), which describes the effect of 4.6 Gyr old sample, the U and U decay-constant 238 235 errors are 9.03 and 2.17 Myr, respectively). Writing the the variability of the U/ U ratio on Pb-Pb ages. transcendental Pb-Pb equation as: Considering two samples formed at times t1 and t2, and defining the age interval as Dt21 ¼ t2 t1, the effect of the k ð 235t Þ 235 r ¼ u e 1 ; ð Þ U decay-constant error on the uncertainty of age inter- k 3 ðe 238t 1Þ vals can be calculated as,

207 206 235 238 2 2 2 with r =( Pb/ Pb)radiogenic and u =( U/ U)present, r ¼ðC ;t C ;t Þ r ; ð10Þ Dt21 5 2 5 1 k235 Ludwig (2000) showed that the variance of the absolute 238 age of the sample can be written as: while the effect of the U decay-constant error is given by, 2 k 2 k 2 k 2 k 2 2 2 238t 2 235t 2 238t 2 235t 2 r ¼ðC ;t C ;t Þ r : ð11Þ ðe 1Þ rr þðute Þ rk þðrte Þ rk þðe 1Þ ru Dt21 8 2 8 1 k238 r2 ¼ 235 238 t k k 2 ðuk 235t rk 238tÞ 235e 238e Since the two decay constants are independent variables, ð4Þ the effect of both decay-constant errors on the uncertainty

2 of age intervals is the sum of Eqs. (10) and (11): where the ru term has been added and accounts for the 235 238 2 2 2 2 2 uncertainty on the U/ U ratio. Removing the depen- r ¼ðC ;t C ;t Þ r þðC ;t C ;t Þ r : ð12Þ Dt21 5 2 5 1 k235 8 2 8 1 k238 dence on r by substituting Eq. (3) into Eq. (4) we get: These equations were checked numerically using Monte- r2 ¼ C2 r2 þ C2 r2 þ C2 r2 þ C2 r2; ð5Þ t r;t r 5;t k235 8;t k238 u;t u Carlo simulations and the top, center and bottom panel of Fig. 2 shows the relative errors from U decay-constants, with r =jDt j ‰ 2 Dt21 21 (in , calculated with Eqs. (12), (10) and k 2 ðe 238t 1Þ (11) respectively), as a function of the age of the anchoring Cr;t ¼ ð6Þ k238t k235t ðk235þk238Þt sample (t ), and for various absolute values of Dt . The uðk238e k235e þðk235 k238Þe Þ 1 2 1 k k contribution of the decay-constant errors on Pb-Pb age ð 235tÞð 238t Þ C ¼ te e 1 ð Þ intervals is much smaller than the achievable precision on 5;t k k ðk þk Þ 7 ðk 238t k 235t þðk k Þ 235 238 tÞ 238e 235e 235 238 e the Pb isotope ratio measurement, and varies between F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 599

Fig. 3. Error, in years, on age intervals from 235U and 238U decay- constant errors as a function of the age of the anchoring sample

(t1), and for an absolute value of age interval jDt21j = 5 Myr.

2012), including the decay-constants errors, both on abso- lute ages and age intervals. Because all samples in Table 3 were measured with the same U double spike, the error on the U spike composition should be ignored when com- paring relative ages (bold values of the ±x/y/z for age inter- vals in Table 3). For comparison, the ±x/y/z values are also calculated assuming two different U spikes were used (each having a ±0.16‰ error associated with it), which increases the errors on the age interval by 40% for most samples. This observation emphasizes the need for the use of a unique U double spike to achieve refined resolution in ESS chronol- ogy using the Pb-Pb chronometer.

3.4. Comparison of d238U data with previous works

In the present work, all samples were spiked with IRMM-3636 U double spike (Verbruggen et al., 2008; Richter et al., 2008). Except for three samples from Brennecka and Wadhwa (2012) (D’Orbigny pyroxenes, AdoR phosphates and a replicate of D’Orbigny Bulk) all literature data published so far (Brennecka and Wadhwa, r =jDt j ‰ Fig. 2. Relative errors on age intervals, 2 Dt21 21 (in , 2012; Connelly et al., 2012; Goldmann et al., 2015) also calculated with Eqs. (12), (10) and (11) respectively) due to the 235U used the IRMM-3636 spike. In Fig. 4, the data from this 238 and U decay-constants errors (from Jaffey et al., 1971)asa work are compared to literature data. Since the same spike function of the age of the anchoring sample (t1), and for various was used for all measurements and we are interested in rel- Dt absolute values of age intervals 21. The different panels includes ative ages, the error bars show only the error on the isotopic the decay-constant errors on k and k (top), k only (center), 235 238 235 analysis (i.e., the ‘‘x” in ±x/y reported in Table 4). Our and k only (bottom). 238 measurements are within uncertainty of published values for 4 samples out of 6. Two samples (NWA 4801 and 1.03 and 1.48‰ of the age interval. For samples formed at NWA 6291) are outside of uncertainty from published val- the beginning of the Solar System and for an age interval of ues. The extremely heterogeneous nature of the NWA 4801 5 Myr, relevant to the plutonic/quenched angrite age differ- specimen, described as a ‘‘breccia” by Irving and Kuehner ences as well as the CAIs/quenched angrite age differences, (2007), could explain the difference in U isotope composi- the decay-constant errors only contribute 7000 yrs to the tion between our data (d238U=0.55 ± 0.06‰) and litera- uncertainty on the age interval (Fig. 3). ture data (d238U=0.43 ± 0.07‰, Brennecka and In Table 3 we provide a breakdown of the uncertainties Wadhwa, 2012). For NWA 6291, the disagreement with on U-isotope-corrected Pb-Pb ages and age intervals published value (d238U=0.40 ± 0.04‰, this work, vs. (anchored to D’Orbigny) for the six angrites studied here, d238U=0.49 ± 0.04‰, Brennecka and Wadhwa, 2012) and the two oldest CAI ages reported in the literature is likely the result of minor terrestrial contamination (Amelin et al., 2010; Bouvier et al., 2011; Connelly et al., (10%; see Section 4.1). 0 ...Tso ta./Gohmc tCsohmc ca23(07 593–617 (2017) 213 Acta Cosmochimica et Geochimica / al. et Tissot F.L.H. 600

Table 3 Contributions of systematic and random uncertainties to the total uncertainty on Pb-Pb ages and age intervals. D’Or. Contribution to NWA 4590 Contribution to NWA Contribution to NWA Contribution to AdoR Contribution to Sahara Contribution to Sahara Contribution to Wtd avg Contribution to CAI Contribution to * ** ± x/y/z, in %. ± x/y/z, in %. 4801 ± x/y/z, in %. 6291 ± x/y/z, in %. ± x/y/z, in %. 99555a ± x/y/z, in %. 99555b ± x/y/z, in %. CAIs: SJ-101 ± x/y/z, in %. B4 ± x/y/z, in %. Corrected Age (Myr) 4563.51 4557.76 4556.82 4560.69 4556.45 4564.07 4563.79 4567.30 4567.94 ±x/y/z (Myr) ±0.18/0.29/9.25 ±0.31/0.38/9.24 ±0.17/0.28/9.24 ±0.78/0.81/9.27 ±0.18/0.29/9.24 ±0.43/0.48/9.25 ±0.24/0.33/9.25 ±0.16/0.20/9.25 ±0.21/0.31/9.25 Uncertainty source: ± ± ± ±±±± ± ± Pb iso. analysis 0.12 46.4/17.4/0.02 0.28 81.8/53.1/0.09 0.15 77.1/27.8/0.03 0.78 99.5/91.7/0.71 0.13 54.5/20.4/0.02 0.38 78.7/61.4/0.17 0.14 33.4/17.8/0.02 0.09 33.0/21.3/0.01 0.10 21.4/9.9/0.01 + Pb spike compo. U iso. analysis 0.13 53.6/20.1/0.02 0.13 18.2/11.8/0.02 0.08 22.9/8.3/0.008 0.06 0.5/0.5/0.004 0.12 45.5/17.1/0.02 0.20 21.3/16.6/0.05 0.20 66.6/35.4/0.05 0.13 67.0/43.1/0.02 0.19 78.6/36.3/0.04 U spike composition 0.23 n.a./62.5/0.06 0.23 n.a./35.1/0.06 0.23 n.a./63.9/0.06 0.23 n.a./7.8/0.06 0.23 n.a./62.5/0.06 0.23 n.a./22.0/0.06 0.23 n.a./46.8/0.06 0.12 n.a./35.6/0.02 0.23 n.a./53.8/0.06 235 U decay constant 8.98 n.a./n.a./94.4 8.97 n.a./n.a./94.3 8.97 n.a./n.a./94.4 8.98 n.a./n.a./93.7 8.97 n.a./n.a./94.4 8.98 n.a./n.a./94.2 8.98 n.a./n.a./94.4 8.99 n.a./n.a./94.4 8.99 n.a./n.a./94.4 238 U decay constant 2.17 n.a./n.a./5.5 2.17 n.a./n.a./5.5 2.17 n.a./n.a./5.5 2.17 n.a./n.a./5.5 2.17 n.a./n.a./5.5 2.17 n.a./n.a./5.5 2.17 n.a./n.a./5.5 2.17 n.a./n.a./5.5 2.17 n.a./n.a./5.5

Age rel. to D’Orbigny (Myr) 5.75 6.69 2.82 7.05 +0.57 +0.29 +3.79 +4.43 ±x/y/z (Myr) (recommended) ±0.36/0.36/0.36 ±0.25/0.25/0.25 ±0.80/0.80/0.80 ±0.25/0.25/0.25 ±0.46/0.46/0.46 ±0.30/0.30/0.30 ±0.24/0.24/0.24 ±0.27/0.27/0.27 ±x/y/z (Myr) ±0.36/0.48/0.48 ±0.25/0.40/0.40 ±0.80/0.86/0.86 ±0.25/0.41/0.41 ±0.46/0.56/0.56 ±0.30/0.44/0.44 ±0.24/0.35/0.35 ±0.27/0.42/0.42 (if different U spikes used) Uncertainty source: ± ± ±±±± ± ± Pb iso. analysis + Pb 0.30 73.1/40.3/40.3 0.19 61.3/22.6/22.5 0.79 96.9/83.5/83.5 0.18 50.4/18.9/18.9 0.40 74.0/50.0/50.0 0.18 37.9/17.6/17.6 0.15 40.3/18.7/18.7 0.15 31.7/13.4/13.4 spike compo. U iso. analysis 0.18 26.9/14.8/14.8 0.15 38.7/14.3/14.3 0.14 3.10/2.67/2.67 0.18 49.6/18.6/18.6 0.24 26.0/17.5/17.5 0.24 62.1/28.8/28.8 0.18 59.7/27.6/27.6 0.23 68.3/28.8/28.8 U spike composition 0.32 n.a./44.9/44.9 0.32 n.a./63.2/63.1 0.32 n.a./13.9/13.9 0.32 n.a./62.5/62.4 0.32 n.a./32.5/32.5 0.32 n.a./53.5/53.5 0.26 n.a./53.8/53.7 0.32 n.a./57.8/57.8 235 U&238 U decay constant 0.008 n.a./n.a./0.03 0.009 n.a./n.a./0.05 0.004 n.a./n.a./0.002 0.010 n.a./n.a./0.06 0.0008 n.a./n.a./0.0002 0.0004 n.a./n.a./0.0001 0.005 n.a./n.a./0.02 0.006 n.a./n.a./0.02 * Weigthed average of the age of CAI SJ-101 (Amelin et al., 2010), and three Efremovka CAIs (Connelly et al., 2012), which have indistinguishable ages within uncertainties. ** CAI B4 from NWA 6991 CV3 chondrite (Bouvier et al., 2011). a Pb-Pb age from Amelin (2008b). b Pb-Pb age from Connelly et al. (2008). Table 4 Pb-Pb uncorrected ages, U isotopic composition, age corrections and corrected Pb-Pb ages of angrites. Sample Type Age (Myr) MSWD Materiala Method Ref. 238 U/235 U assumed/used NWA 2999 Plutonic 4561.79 ± n.r./0.42/9.25 Px: residue Pb-Pb isochron§ (1) 137.88 NWA 4590 Plutonic 4557.93 ± n.r./0.28/9.24 0.118 Px: residue Pb-Pb isochron§ (2) 137.789 ± 0.152 NWA 4801 Plutonic 4558.06 ± n.r./0.15/9.23 Px: residue Pb-Pb isochron§ (1) 137.88 NWA 6291 Plutonic 4561.29 ± n.r./0.78/9.27 2.5 WR: last wash + residue Pb-Pb isochron§ (3) 137.84 Angra dos Reis Plutonic 4557.65 ± n.r./0.13/9.23 1.01 Px: residue 207 Pb/206 Pb ages§§ (4) 137.88 D’Orbigny Quenched 4564.42 ± n.r./0.12/9.24 1.18 Px + WR: residue 207 Pb/206 Pb ages§§ (4) 137.88 Sahara 99555 Quenched 4564.86 ± n.r./0.38/9.25 1.5 WR: residue Pb-Pb isochron§ (5) 137.88 Sahara 99555 Quenched 4564.58 ± n.r./0.14/9.24 0.99 WR: some washes + residue Pb-Pb isochron§ (6) 137.88 ...Tso ta./Gohmc tCsohmc ca23(07 9–1 601 593–617 (2017) 213 Acta Cosmochimica et Geochimica / al. et Tissot F.L.H.

Age correction d238 U(‰)b Ref. 238 U/235 Uc Age correction Dt (Myr) Corrected ages (Myr) This work NWA 4590 0.47 ± 0.09/0.18 137.772 ± 0.013/0.025 0.17 ± 0.13/0.26 4557.76 ± 0.31/0.38/9.24 NWA 4801 0.55 ± 0.06/0.17 137.762 ± 0.008/0.023 1.24 ± 0.08/0.24 4556.82 ± 0.17/0.28/9.24 NWA 6291 0.40 ± 0.04/0.16 137.783 ± 0.005/0.022 0.60 ± 0.06/0.23 4560.69 ± 0.78/0.81/9.27 Angra dos Reis 0.51 ± 0.08/0.18 137.766 ± 0.011/0.024 1.20 ± 0.12/0.26 4556.45 ± 0.18/0.29/9.24 D’Orbigny 0.32 ± 0.09/0.18 137.793 ± 0.012/0.025 0.91 ± 0.13/0.26 4563.51 ± 0.18/0.29/9.25 Sahara 99555d 0.23 ± 0.14/0.21 137.805 ± 0.019/0.029 0.79 ± 0.20/0.30 4564.07 ± 0.43/0.48/9.25 Sahara 99555e 4563.79 ± 0.24/0.33/9.25 Previous studies NWA 4590 0.47 ± 0.03/0.16 (7) 137.772 ± 0.004/0.022 0.18 ± 0.04/0.23 4557.75 ± 0.28/0.36/9.24 NWA 4801 0.43 ± 0.07/0.17 (7) 137.778 ± 0.009/0.023 1.07 ± 0.09/0.25 4556.99 ± 0.18/0.29/9.24 NWA 6291 0.49 ± 0.04/0.16 (7) 137.769 ± 0.006/0.022 0.75 ± 0.06/0.24 4560.54 ± 0.78/0.81/9.27 * NWA 6291 Leach 0.27 ± 0.05/0.17 (7) 137.800 ± 0.007/0.023 0.42 ± 0.07/0.24 4560.87 ± 0.78/0.82/9.27 * NWA 6291 Residue 0.57 ± 0.11/0.19 (7) 137.759 ± 0.015/0.026 0.85 ± 0.16/0.28 4560.44 ± 0.80/0.83/9.27 Angra dos Reis 0.48 ± 0.14/0.21 (8) 137.771 ± 0.019/0.029 1.15 ± 0.20/0.30 4556.50 ± 0.24/0.33/9.24 ** AdoR phosphates 0.22 ± 0.24/0.28 (7) 137.806 ± 0.033/0.039 0.78 ± 0.35/0.41 4556.87 ± 0.37/0.43/9.24 D’Orbigny 0.34 ± 0.07/0.16 (7) 137.790 ± 0.010/0.022 0.95 ± 0.11/0.23 4563.47 ± 0.16/0.26/9.24 D’Orbigny 0.33 ± 0.07/0.17 (7) 137.791 ± 0.010/0.024 0.94 ± 0.11/0.25 4563.48 ± 0.16/0.28/9.25 *** D’Orbigny pyroxenes 0.43 ± 0.20/0.25 (7) 137.778 ± 0.028/0.034 1.07 ± 0.29/0.36 4563.35 ± 0.32/0.38/9.25 Sahara 99555d 0.38 ± 0.08/0.18 (9) 137.784 ± 0.011/0.024 1.01 ± 0.12/0.25 4563.85 ± 0.40/0.46/9.25 Sahara 99555e 4563.57 ± 0.18/0.29/9.25 References: (1) Amelin and Irving (2007);(2)Amelin et al. (2011); (3) Bouvier et al. (2011); (4) Amelin (2008a); (5) Amelin (2008b); (6) Connelly et al. (2008); (7) Brennecka and Wadhwa (2012); (8) Goldmann et al. (2015); (9) Connelly et al. (2012). ‘‘n.r.” stands for ‘‘not reported”. § Used when common lead or redistribution of radiogenic Pb after crystallization is a concern. §§ Weighted average of 207 Pb/206 Pb ages, used when only radiogenic Pb is present in the fraction after washing and blank correction. Identical ages (within errors) are obtained by the Pb-Pb isochron and U-Pb-upper concordia intercept methods. * The uncorrected age of the Leach and Residue from NWA 6291 is assumed to be equal that of the bulk NWA 6291 sample. ** The uncorrected age of the phosphates from AdoR is assumed to be equal to that of the bulk AdoR sample. *** The uncorrected age of the pyroxenes from D’Orbigny is assumed to be equal to that of the bulk D’Orbigny sample. a Px = Pyroxene, WR = Whole rock. b Values renormalized to CRM-112a. c Absolute ratios are calculated using the 238 U/235 U ratio of 137.837 for CRM-112a, from Richter et al. (2010). d Pb-Pb age from Amelin (2008b). e Pb-Pb age from Connelly et al. (2008). 602 F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617

3.5. Variability of the U isotope composition of angrites 0.0 -0.1 NWA NWA NWA Considering only the error on the isotopic measurement, 4801 4590 6291 it appears that the U isotopic composition of angrites is -0.2 variable at the 0.10–0.20‰ level and that quenched angrites -0.3 d238 ‰ have on average higher U values ( 0.29 ± 0.07 U (‰) -0.4 weighted average, n = 2, MSWD = 5.04) than plutonic 238 δ Sah. ‰ -0.5 angrites ( 0.46 ± 0.03 weighted average, n = 4, D'Or. MSWD = 1.82) (Figs. 4 and 5 and Table 1). Taking previ- -0.6 ously published data from Brennecka and Wadhwa (2012), AdoR -0.7 Connelly et al. (2012) and Goldmann et al. (2015) the 4554 4557 4560 4563 4566 weighted average d238U of quenched angrites is 0.36 Age (Myr) ± 0.05‰ (n = 2, MSWD = 3.28) compared to 0.47 ± 0.02‰ (n = 4, MSWD = 0.62) for plutonic angrites. Pre- Fig. 6. d238U vs. U-isotope-corrected Pb-Pb ages of selected vious data thus also reveal this dichotomy but this had not angrites from this study and literature: Brennecka and Wadhwa (2012), squares; Connelly et al. (2012), triangle; and Goldmann et al. (2015), diamond. -0.1 been appreciated nor discussed. Our conclusion of a hetero- -0.2 geneous U isotope composition in angrite samples is in con- trast with the findings of Brennecka and Wadhwa (2012), who, by considering both random and systematic errors, -0.3 D'Orbigny concluded that all angrites have similar U isotopic compo- NWA sition. Finally, we observe that there is a broad correlation -0.4 4801 Sahara between the U isotope composition of the samples and their 99555 U-isotope-corrected Pb-Pb ages (Fig. 6). -0.5 AdoR NWA

U (‰), literature data NWA 6291 4. DISCUSSION

238 4590 B&W. 12 δ -0.6 Gold. 15 Conn. 12 Angrites provide a useful set of anchors to tie absolute -0.7 and relative chronometers because (1) their crystallization -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 ages are 4 to 11 Myr younger than the oldest CAIs (which δ238U (‰), this work define the ‘‘time zero” of ESS chronology, 4567– 4568 Myr ago, Amelin et al., 2010; Bouvier and Wadhwa, Fig. 4. Comparison of U isotope composition of selected angrites 2010; Bouvier et al., 2011; Connelly et al., 2012), (2) some measured in this study (x-axis) with literature values (y-axis): of the angrite samples have quenched textures indicative Brennecka and Wadhwa (2012), squares; Connelly et al. (2012), of rapid cooling, and (3) the phases in these samples are triangle; Goldmann et al. (2015), diamond. diverse enough to allow for dating with multiple chronome- ters. In particular, D’Orbigny, which cooled rapidly

0.0 137.837 (Mittlefehldt et al., 2002) and has experienced no thermal Quenched disturbance, is one of the most commonly used anchors -0.1 137.823 Plutonic in ESS chronology (207Pb-206Pb: Amelin, 2008a; -0.2 137.809 26 26 Average SS Al- Mg: Spivak-Birndorf et al., 2009; Schiller et al., -0.3 137.796 2010; 53Mn-53Cr: Glavin et al., 2004; 182Hf-182W: Kleine -0.4 137.782 et al., 2012; and 60Fe-60Ni: Tang and Dauphas, 2012). -0.5 137.768 Because the Pb-Pb age of a sample depends on its U isotope -0.6 137.754 composition, we begin by examining the cause(s) of U iso- -0.7 137.741 tope variations in angrites before evaluating their impact on -0.8 137.727 age determinations. NWA NWA NWA Angra D'Orbigny Sahara 4590 4801 6291 dos Reis 99555 4.1. Distinct 238U/235U ratios of plutonic and volcanic Fig. 5. U isotopic composition of selected angrites. The d238U angrites: a primitive feature values are relative to U standard CRM-112a. The shaded areas represent the weighted average of plutonic (blue) and quenched As can be seen in Fig. 5 and Table 1 and 4, plutonic (red) angrites. The dotted line shows the average Solar System angrites (NWA 4590, NWA 4801, NWA 6291 and AdoR) d238 ‰ value, USS = 0.31 ± 0.04 (Goldmann et al., 2015). have lighter U isotopic composition than quenched angrites Quenched angrites have heavier U isotopic composition than (D’Orbigny and Sahara 99555) (0.46 ± 0.03/0.16‰ vs. plutonic angrites, which has a bearing on the absolute Pb-Pb ages 0.29 ± 0.07/0.17‰, weighted averages, respectively). of the samples (see Table 4). (For interpretation of the references to There are even differences in the U isotopic composition color in this figure legend, the reader is referred to the web version of this article.) of plutonic angrites themselves (e.g., NWA 6291 and F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 603

50 50

5 5 D'Orbigny Sah99555 NWA 1296 AdoR NWA 4801 NWA 4590 Asuka 881371 LEW 86010 NWA 2999 NWA 3164 LEW 87051 NWA 4931 NWA 5167 NWA 6291 0.5 0.5 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Fig. 7. Rare Earth Element pattern in bulk quenched (left) and plutonic (right) angrites. Quenched angrites have essentially flat REE patterns, while plutonic angrites show fractionations consistent with pyroxene, plagioclase and olivine control. Data source, see Table 5.

NWA 4801). To understand these results, we put them in 4590 and NWA 6291 is clearly indicated by their very high perspective using several geochemical tracers. d[234U/238U] values (>200‰). The sample with the highest REEs are refractory and relatively fluid-immobile ele- 234U excess (NWA 6291) has a d238U value between that ments with chemical behavior very similar to one another of other plutonic angrites (at 0.50‰) and the crust of during most chemical processes. A compilation of the the Earth (0.29 ± 0.03‰, Tissot and Dauphas, 2015). REE concentrations in angrites is shown in Fig. 7 and To understand how much the d238U values were affected Table 5 (see table footnote for data source). The data reveal by the open-system behavior recorded in the d[234U/238U] a clear dichotomy in the REE pattern of quenched and plu- values of the samples, we compare our data, obtained on tonic angrites, whereby quenched angrites have chondritic bulk, non pre-cleaned samples, to the results of REE pattern and plutonic angrites show depletion in the Brennecka and Wadhwa (2012) and Goldmann et al. LREE, and for some even a small Eu anomaly. Given the (2015), who pre-cleaned the samples before digestion for typical REE patterns of individual minerals of pyroxene, 5 min in an ultrasonic bath of 0.05 M HCl and 0.5 M plagioclase and olivine measured in angrites (e.g., Floss HCl, respectively, to remove easily dissolvable terrestrial 238 et al., 2003), the fractionated bulk REE patterns of plutonic contaminants. The d U values from this work and the lit- angrites are consistent with variable amount of clinopyrox- erature are indistinguishable for D’Orbigny, AdoR, and ‰ ene and olivine being present as cumulate phases (e.g., Keil, NWA 4590, despite this latter sample having a 250 234 2012 and references therein). In Fig. 8, the d238U value is excess of U. This shows that the alteration processes that 234 238 shown against the La/Sm ratio normalized to CI (a proxy affected the U/ U ratio of the samples did not disturb 238 235 for the flatness of the REE pattern). It is clear that two pop- their U/ U ratio, and indicates that the first order ‰ d238 ulations exist among angrites: 0.20 difference in the U values of quenched and plu- tonic angrites is not the result of terrestrial alteration. d238 (1) Quenched samples, have a chondritic 238U/235U ratio Regarding NWA 4801 the lower value ( U= 0.55 ‰ (d238U=0.29 ± 0.07 compared to 0.31 ± 0.04‰ ± 0.06 ) we obtained compared to Brennecka and d238 ‰ for chondrites, Goldmann et al., 2015) and a flat Wadhwa (2012) ( U= 0.43 ± 0.07 ), is inconsistent REE pattern (La/Sm 1). with presence of a terrestrial component in our sample that (2) Plutonic samples, have a 238U/235U ratio lower than would have been removed by the leaching done in the latter chondritic and fractionated REE patterns (La/ study, and is more likely reflecting the heterogeneous nature ” Sm < 0.8). of this specimen, the only ‘‘breccia angrite (Keil, 2012). Assuming that the difference in d238U values of NWA The non-secular equilibrium values (Table 1) of the 6291 between this work (d238U=0.40 ± 0.04‰) and that d[234U/238U] activity ratio of all angrites finds (between of Brennecka and Wadhwa (2012) (d238U=0.49 +21 to +511‰) indicates that some U mobilization ± 0.04‰) comes from addition of terrestrial U in the sam- occurred in the samples during their residence on Earth ple, means that anywhere from 10 to 73% of the U in the before discovery. Only AdoR (an observed fall) is at secular sample was added during alteration. Because Th is immo- equilibrium (d[234U/238U] = 0.0 ± 0.8‰), indicating no dis- bile during aqueous alteration, the Th/U ratio could pro- turbance of U in this sample in the last 2.5 Myr. For the vide additional insight into the extent of U mobilization other samples, part of the variability in their U isotopic in the samples. Unfortunately, the Th/U ratio was not mea- composition could in principle be due to leaching of U sured in this study, and we have to resort to published data out of the sample or contamination by continental crust acquired on different sample aliquots. U (the contrast in U concentration between Earth’s conti- The Th/U ratio of angrites can be estimated in two main nental crust and angrites is 20:1). In Fig. 9, we plot ways: (1) by Instrumental Neutron Activation Analysis d238U vs. d[234U/238U]. Remobilization of U in NWA (INAA) or on ICPMS instruments (either by laser ablation, 0 ...Tso ta./Gohmc tCsohmc ca23(07 593–617 (2017) 213 Acta Cosmochimica et Geochimica / al. et Tissot F.L.H. 604

Table 5 Compilation of Rare Earth Element (REE), Th and U concentrations in angrites (in ppm), and model Kappa (Th/U weight ratios) in whole rock angrites from Pb isotope studies.

Plutonic angrites Volcanic angrites

Sample LEW LEW NWA NWA NWA NWA NWA NWA NWA NWA NWA NWA AdoR Asuka D’Orbigny D’Orbigny D’Orbigny LEW LEW NWA NWA NWA Sahara Sahara 86010 86010 2999 3164a 4931a 5167a 6291a 4590 4590 4801 4801 4801 881371 87051 87051 1296 1296 1670 99555 99555

Reference M&L90 W&K90 G07 B15 R12 B15 § R12 § R12 R12 § R12 R12 R12 R12 § M&L90 W&K90 R12 B15 B15 R12 B15 Method INAA INAA INAA ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS ICPMS INAA INAA ICPMS ICPMS ICPMS ICPMS ICPMS Mass (mg) 55 190 167 450 50.8 500 50.3 50.1 50.8 50.5 50.3 50.5 65.8 13 27 63.3 360 250 46.1 290 La 3.38 4.7 0.80 ± 0.52 1.39 0.522 ± 0.003 0.48 0.485 3.19 ± 0.04 2.044 3.43 ± 0.01 2.96 ± 0.07 1.848 6.2 ± 0.2 2.50 ± 0.06 3.753 ± 0.002 3.425 3.727 2.32 3.5 3.333 3.40 2.59 3.536 ± 0.005 3.54 Ce 10.1 10.9 2.0 ± 1.0 3.43 1.72 ± 0.01 1.59 1.682 8.8 ± 0.1 6.230 10.68 ± 0.06 9.3 ± 0.1 5.600 18.5 ± 0.3 6.3 ± 0.1 9.22 ± 0.04 8.943 10.255 6.2 9.4 9.097 8.65 6.70 9.01 ± 0.02 9.16 Pr 0.573 0.299 ± 0.006 0.290 0.323 1.35 ± 0.01 1.092 1.75 ± 0.01 1.52 ± 0.03 0.938 3.09 ± 0.04 0.99 ± 0.03 1.331 ± 0.009 1.301 1.695 1.295 1.29 0.97 1.285 ± 0.004 1.34 Nd 8.1 <7 3.10 1.89 ± 0.01 1.71 2.100 7.42 ± 0.09 5.950 10.0 ± 0.1 8.6 ± 0.2 5.180 17.2 ± 0.3 4.9 ± 0.1 6.92 ± 0.01 6.953 8.937 7.079 6.44 4.88 6.70 ± 0.05 6.74 Sm 2.68 2.60 0.67 ± 0.14 1.04 0.69 ± 0.01 0.652 0.765 2.61 ± 0.05 2.212 3.46 ± 0.03 3.02 ± 0.08 1.834 5.9 ± 0.1 1.58 ± 0.06 2.15 ± 0.02 2.169 3.157 1.48 2.01 2.222 2.06 1.56 2.127 ± 0.004 2.16 Eu 0.978 0.97 0.236 ± 0.038 0.408 0.39 ± 0.01 0.290 0.298 1.06 ± 0.03 0.896 1.152 ± 0.008 1.03 ± 0.01 0.602 1.81 ± 0.07 0.63 ± 0.02 0.840 ± 0.007 0.858 0.971 0.536 0.83 0.845 0.780 0.629 0.821 ± 0.008 0.860 Gd 1.40 0.93 ± 0.02 0.956 1.094 3.51 ± 0.07 3.402 4.43 ± 0.04 3.85 ± 0.05 2.688 7.3 ± 0.2 2.09 ± 0.04 2.718 ± 0.004 2.756 4.112 2.880 2.84 2.08 2.70 ± 0.05 2.87 Tb 0.66 0.177 ± 0.032 0.248 0.183 ± 0.004 0.170 0.196 0.70 ± 0.02 0.616 0.844 ± 0.005 0.733 ± 0.009 0.448 1.39 ± 0.05 0.38 ± 0.01 0.508 ± 0.002 0.499 0.770 0.36 0.49 0.527 0.509 0.377 0.505 ± 0.006 0.519 Dy 4.4 1.65 1.26 ± 0.03 1.18 1.333 4.9 ± 0.1 4.312 5.60 ± 0.05 4.90 ± 0.05 2.870 9.2 ± 0.3 2.60 ± 0.08 3.384 ± 0.009 3.479 4.915 3.1 3.685 3.42 2.53 3.40 ± 0.03 3.49 Ho 0.350 0.284 ± 0.007 0.267 0.299 1.14 ± 0.02 0.966 1.22 ± 0.01 1.08 ± 0.01 0.616 2.03 ± 0.08 0.57 ± 0.02 0.760 ± 0.004 0.752 1.058 0.801 0.726 0.538 0.76 ± 0.01 0.748 Er 1.04 0.82 ± 0.02 0.80 0.891 3.32 ± 0.07 3.066 3.388 ± 0.007 2.95 ± 0.01 1.764 5.6 ± 0.2 1.66 ± 0.06 2.15 ± 0.01 2.178 3.141 2.317 2.16 1.61 2.16 ± 0.02 2.23 Tm 0.132 ± 0.002 0.131 0.52 ± 0.01 0.462 0.507 ± 0.001 0.4422 ± 0.0004 0.238 0.83 ± 0.03 0.243 ± 0.009 0.3328 ± 0.0003 0.321 0.433 0.344 0.332 ± 0.002 Yb 2.64 2.61 0.77 ± 0.08 1.02 0.88 ± 0.03 0.87 0.933 3.31 ± 0.07 3.290 3.106 ± 0.001 2.72 ± 0.07 1.596 5.0 ± 0.1 1.60 ± 0.06 2.09 ± 0.02 2.120 3.001 1.52 2.00 2.230 2.06 1.54 2.11 ± 0.01 2.11 Lu 0.386 0.38 0.120 ± 0.010 0.155 0.141 ± 0.003 0.135 0.146 0.50 ± 0.01 0.504 0.460 ± 0.002 0.404 ± 0.006 0.224 0.72 ± 0.01 0.24 ± 0.01 0.31560 ± 0.00001 0.307 0.465 0.239 0.30 0.325 0.296 0.220 0.322 ± 0.002 0.305 Th 0.48 0.40 <0.2 0.030 0.028 ± 0.002 0.028 0.483 0.325 ± 0.009 0.364 0.283 ± 0.004 0.242 ± 0.002 0.182 0.74 ± 0.05 0.33 ± 0.02 0.454 ± 0.009 0.451 0.454 0.22 0.41 0.451 0.43 0.33 0.50 ± 0.02 0.45 U 0.15 0.13 ± 0.14 0.337 0.032 ± 0.002 0.019 0.139 0.096 ± 0.004 0.126 0.094 ± 0.001 0.081 ± 0.001 0.070 0.20 ± 0.01 0.082 ± 0.005 0.082 ± 0.001 0.077 0.081 <0.21 0.067 0.070 0.075 0.106 ± 0.003 0.097 Th/U 2.67 2.80 0.089 0.88 1.47 3.47 3.39 2.89 3.01 2.99 2.60 3.70 4.02 5.54 5.86 5.62 6.73 6.14 4.40 4.72 4.64 ±b 0.13 0.004 0.08 0.07 0.17 0.17 0.14 0.05 0.04 0.13 0.31 0.35 0.13 0.29 0.28 0.34 0.31 0.22 0.23 0.23 Model Kappa Reference L&G92 B11 A 08a A 08b Th/U (wt) 3.84 3.36 3.81 3.84 ± 0.01 0.34 0.17 0.06 References for trace element analysis: B15, Baghdadi et al. (2015); G07, Gellissen et al. (2007); M&L90, Mittlefehldt and Lindstrom (1990); R12, Riches et al. (2012); W&K90, Warren and Kallemeyn (1990). References for Pb isotope data: A 08a, Amelin (2008a); A 08b, Amelin (2008b); B11, Bouvier et al. (2011); L&G92, Lugmair and Galer (1992). a Paired with NWA 2999. b Values in italic assume a 5% uncertainty on the measurement. § Measurements made at ASU on a 5% solution aliquot of the samples from Brennecka and Wadhwa (2012). Measurements were made on Q-ICPMS, with indium internal standard to account for any plasma suppression. Samples were bracketed with geologic standards and clean ICP concentration standards. F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 605

0.0 Plutonic Volcanic D'Or. 8 -0.1 7 -0.2 Sah. 6 -0.3 5 U (‰) -0.4

238 4 δ -0.5 3 -0.6 NWA NWA NWA 6291 4801 AdoR 4590 2 -0.7 1 0.2 0.4 0.6 0.8 1.0 1.2 Th/U from Pb isotopes 0 (La/Sm) norm. to CI 0 1 2 3 4 5 6 7 8 Fig. 8. d238U vs. La/Sm weight ratios normalized to CI of selected Th/U from ICPMS or INAA angrites from this study (circles) and literature: Brennecka and Wadhwa (2012), squares; Connelly et al. (2012), triangle; and Fig. 10. Comparison of the Th/U weight ratio in bulk angrites as Goldmann et al. (2015), diamond. obtained from ICPMS or INAA measurements (x-axis) and Pb isotopes measurements (y-axis, see Table 5 for data source). Circles show samples for which both methods were used on unleached 0 pieces of the same angrite, while shaded areas show the full range Sah99555 D'Orbigny of Th/U ratios obtained by each technique on unleached bulk -0.1 AdoR NWA 4801 samples. The vertical and horizontal grey lines mark the chondritic NWA 4590 NWA 6291 -0.2 value of 3.6 ± 0.1 (Chen et al., 1993). -0.3

U (‰) samples. This value is in line with the lower estimate of -0.4

238 10% of terrestrial contamination in NWA 6291 (derived -0.5 from the difference of d238U value between our measure- -0.6 ment and that of Brennecka and Wadhwa, 2012). Though the evidence discussed above indicates that terrestrial alter- -0.7 ation is not the main driver of U isotope variability for 0 100 200 300 400 500 D’Orbigny, AdoR, NWA 4590 and NWA 4801, we cannot [234U/238U] (‰) rule out that terrestrial contamination affected our mea- Fig. 9. d238U vs. d[234U/238U] values of selected angrites from this surement of NWA 6291. study. In summary, to the potential exception of NWA 6291, there is no evidence that the variable d238U values observed in different angrites is due to terrestrial aqueous alteration, or by solution after digestion with acid attacks), and (2) by and the distinct 238U/235U ratios of plutonic and volcanic measurement of the Pb isotopic composition of a sample, as angrites instead appears to be a primitive feature of these 232 208 238 Th decays into Pb (t1/2 = 14.05 Gyr) and U decays samples. 206 into Pb (t1/2 = 4.468 Gyr), and the ratio of radiogenic 208Pb/206Pb can be used to calculate a model 232Th/238U 4.2. 238U/235U ratios of plutonic angrites: evidence for U ratio in the sample (often called kappa, and noted j). A dif- stable isotope fractionation during magmatic processes ficulty arises in the present case, as the Th/U ratio estimates obtained by both techniques do not agree with each other Three processes could potentially explain the variations (Table 5 and Fig. 10). While the values obtained by in the 238U/235U ratios and/or REE patterns of quenched INAA/ICPMS show variability in the Th/U weight ratios and plutonic angrites: (i) decay of live 247Cm in the ESS, of bulk angrites between 2.6 (NWA 4801, this work) and (ii) alteration, on Earth or on the angrite parent body, 6.7 (NWA 1296, Riches et al., 2012), the model Th/U and (iii) U stable isotope fractionation during magmatic weight ratios of all whole rock angrites calculated from processes. We examine these hypotheses below. Pb isotope analysis are within error of the chondritic value Decay of 247Cm, an extinct radionuclide with a half-life of 3.6 ± 0.1 (Chen et al., 1993). The cause of this discrep- of 15.6 Myr, can produce 235U excesses in ESS materials ancy is unclear, but might be due to counting limitations (Brennecka et al., 2010a; Tissot et al., 2016). If the varia- for INAA and matrix effects during ICPMS analysis, as tions observed in the d238U values of angrites were due to the U and Th concentrations in angrites are quite low the decay of live 247Cm, a correlation should exist between (around 116 ppb for U and 420 ppb for Th). Lead isotopic the d238U values of angrites and their 144Nd/238U ratios analyses are, by nature, more accurate and precise than (Cm has no long-lived isotopes and Nd is used as a proxy). concentration measurements, and we will therefore use In particular, to produce a 0.17‰ effect, given the known those data in the following. The limited variation in this 247Cm/238U ratio of (1.79 ± 0.09) 105 in the ESS (Tissot data set (Th/U wt ratios between 3.36 and 3.84, Table 5) et al., 2016; Tang et al., 2017), the 144Nd/238U atomic indicates that at most 7% of U gain/loss occurred in the ratios of quenched and plutonic angrites would have to 606 F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617

144 238 144 238 differ by 66 (i.e., Nd/ UPlutonic - Nd/ UQuenched = 0 66). The 144Nd/238U ratio for plutonic angrites shows some -0.1 Sahara variability and overlaps with the quenched angrites and the 99555 two groups are indistinguishable (Fig. 11). Decay of 247Cm -0.2 D'Orbigny can only account for an excess of 0.05‰, and is therefore -0.3 238 235

not the main source of variability in the U/ U ratios U (‰) -0.4 of angrites. 238 AdoR Alteration, on Earth or on the angrite parent body, δ -0.5 NWA 238 235 4590 could lead to variability of the U/ U ratios of angrites. -0.6 NWA Except for AdoR, all angrites specimens to date are sample 6291 NWA 4801 finds and have therefore experienced varying degree of -0.7 alteration in hot deserts (NWA 4590, NWA 4801, NWA 0 5 10 15 20 6291, Sahara 99555) or farm fields (D’Orbigny). Hot desert 1/U (1/ppm) samples are known to have experienced more severe chem- Fig. 12. d238U vs. 1/U (1/ppm) of selected angrites from this study ical alteration than cold desert meteorites (Crozaz et al., (circles) and literature: Brennecka and Wadhwa (2012), squares; 2003), which could have affected their U elemental and/or Connelly et al. (2012), triangle; and Goldmann et al. (2015), isotope systematics. Weathering in hot deserts will lead to diamond. The lack of correlation between these two parameters enrichments in LREE, Ba, Sr and/or U in individual miner- indicates that the d238U of the samples is not controlled by als or at the whole rock level (e.g., Barrat et al., 2001; redistribution of U in the samples. Crozaz et al., 2003; Floss et al., 2003), and LREE enrich- ments and Ce anomalies have in fact been reported in some Sahara 99555 minerals and to a much lesser extent in min- samples, unlike what was observed in carbonaceous erals in D’Orbigny (Floss et al., 2003). Several lines of evi- meteorites during thermal metamorphism (Rocholl dence, however, allow us to conclude that alteration, on and Jochum, 1993). Earth or on the angrite parent body, is not the main driver (iii) Sahara 99555 and D’Orbigny have experienced very behind the variable 238U/235U ratios of angrites: different weathering conditions (hot desert vs. farm field), resulting in clearly visible LREE and Ce posi- tive anomalies in some olivine and kirschsteinite (i) Despite U remobilization on Earth, as testified by the grains in Sahara 99555, which are hardly visible in excess 234U in the fall samples (Fig. 9), the dichotomy D’Orbigny (Crozaz et al., 2003; Floss et al., 2003), in d238U values between quenched and plutonic yet, both specimens have similar Ba, Sr, Th and U angrites is clearly visible in samples with the lowest concentrations and identical d238U values. 234U excesses (i.e., the lowest amount of U remobi- (iv) There is no visible LREE enrichment in the samples lization, see Section 4.1). at the bulk level. In fact, the lack of Ce anomalies (ii) There is no correlation between d238U values and U and the low La/Sm ratios (<0.8) of the plutonic content in angrites (Fig. 12), which indicates that angrites (Fig. 8) argue against any significant distur- no significant redistribution of U occurred in the bance of the samples through alteration. (v) Despite their low U content (from 70 to 240 ppb), all 0.1 angrite samples have chondritic Th/U ratios (Fig. 10), NWA 6291 which suggests that no significant U gain/loss occurred 0.0 NWA 4590 during alteration (<10%, see Section 4.1). -0.1 NWA 4801 AdoR (vi) Finally, the sample preparation scheme used by -0.2 D'Orbigny Brennecka and Wadhwa (2012) and Goldmann -0.3 Sah. 99555 et al. (2015) included a 5 min ultrasonic bath cleaning U (‰)

238 -0.4 step in dilute HCl, which was not performed in the δ -0.5 present study. Yet, the three isotopic data sets are in agreement (Fig. 4). -0.6 d238 -0.7 A more likely cause of the U variability and REE 0 10 20 30 40 50 60 70 fractionation in angrites is magmatic differentiation. 144Nd/238U (atomic ratio) Indeed, whereas quenched angrites have essentially flat REE patterns, plutonic angrites show curved patterns with Fig. 11. d238U vs. 144Nd/238U atomic ratio of selected angrites slight Eu anomalies (AdoR, NWA 4801, NWA 4590, LEW from this study (color symbol) and literature (light grey symbols, 86010) or patterns with strong LREE depletions and flat data from Brennecka and Wadhwa (2012), Connelly et al. (2012), HREEs (NWA 2999, NWA 3164, NWA 4931, NWA Goldmann et al. (2015)). The same symbol shape is used for both 5167, NWA 6291), which are likely controlled by pyroxene sets of data. Quenched and plutonic angrites with similar and olivine, respectively (Fig. 7). The REE patterns of 144 238 d238 ‰ Nd/ U ratios show U values differing by 0.15–0.20 , LEW 86010 and NWA 4801 can be well reproduced by indicating that the decay of live 247Cm is not the main source of addition of 25% and 60%, respectively, of clinopyroxene variability in the 238U/235U ratios of angrites. (For interpretation of the references to color in this figure legend, the reader is referred to (see mineral REE pattern in Floss et al., 2003) to a flat the web version of this article.) REE pattern at 15.5 CI (similar to the REE pattern of F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 607 quenched angrites), which is entirely consistent with the the sample. Using the d238U values of the bulk and one cumulus texture of these rocks (e.g., McKay et al., 1988; mineral separate for AdoR and D’Orbigny allows us to esti- Irving and Kuehner, 2007). Similarly, the REE patterns mate the difference of d238U values between clinopyroxenes 238 238 of angrites with LREE depletions, flat HREE tails and and phosphates, Dcpx-phos = d Ucpxd Uphos, in these enrichments of only 5 to 6 CI are consistent with the two specimens. The mass balance goes as: abundance of large olivine xenocrysts (which incorporate d238U ¼ f d238U þð1 f Þd238U ð14Þ very little LREE relative to HREE, see Floss et al., 2003) bulk cpx cpx cpx phos present in these angrites. Because phosphate minerals host where the subscripts ‘‘bulk”, ‘‘cpx” and ‘‘phos” refer to the between 30 and 50% of the LREE, better fits of the REE bulk sample, the clinopyroxenes and the phosphates, patterns can be obtained when the small amount of phos- respectively, and fcpx is the fraction of U in clinopyroxene 238 phates (0.1–0.5%) present in the samples is taken into in the sample. For AdoR, using d Ubulk = 0.51 238 account (e.g., Sanborn, 2012; Baghdadi et al., 2015). Yet, ± 0.08‰, d Uphos = 0.22 ± 0.24‰, and fcpx = 0.90 (see mass balance considerations using the modal abundances [U]cpx, [U]phos and minerals modes detailed above) gives of phosphates and their REE content in NWA 4801 or Dcpx-phos = 0.32 ± 0.28‰. For D’Orbigny, using 238 238 NWA 4590 (Sanborn, 2012) indicate that phosphates host d Ubulk = 0.32 ± 0.09‰, d Ucpx = 0.43 ± 0.20‰, at most 20% of the REE heavier than Nd and that to first and assuming [U]phos D’Orbigny = [U]phos AdoR, we get order, the REE patterns of angrites are in fact controlled fcpx = 0.67, which leads to Dcpx-phos = 0.33 ± 0.68‰,a by pyroxene, plagioclase and olivine abundances. value identical to the one obtained for AdoR. Though phosphates are often considered to be the major Given the oxygen fugacity levels relevant to the angrites carrier of U and Th in angrites (e.g., Baghdadi et al., 2015), (IW+1, Brett et al., 1977; Jurewicz et al., 1993; Mckay we propose that pyroxenes represent another (and some- et al., 1994), U will be present only in the 4+ state in the times more) important host phase of U, and to a lesser angrite mantle (Schreiber and Andrews, 1980; Halse, extent Th, in these samples. For instance, using the phos- 2014), and redox effects are therefore very unlikely to be phate U concentration of 1.79 ppm and the pyroxene U responsible for the observed U isotope variability of concentration of 172 ppb measured in AdoR (Wasserburg angrites. A change in the coordination environment of U et al., 1977) and the modal abundances of phosphates between the silicate liquids and the minerals incorporating and pyroxenes in AdoR of 0.3–0.5% and 46.3%, respec- U (clinopyroxene and phosphates) during the magmatic tively (Mittlefehldt et al., 2002), we estimate that less than evolution of the angrite parent body is therefore the most 10% of the total U (and about 33% of the total Th) of likely cause of the variable 238U/235U ratios of angrites. the sample is hosted in phosphates. AdoR is known for Because phosphates are late crystallizing phases, the frac- its pyroxene cumulate texture and might not be representa- tionation is most likely occurring during the incorporation tive of other angrites, but doing the same calculation for of U into clinopyroxenes. D’Orbigny, using a U concentration in pyroxenes of This hypothesis requires fractionation of U isotopes 92 ppb (Brennecka and Wadhwa, 2012), modal abundances without fractionation of the Th/U ratio in clinopyroxenes. of phosphates and pyroxenes of 0.3–0.5% and 20%, respec- Only limited data is available for crystal-liquid partitioning tively (Mittlefehldt et al., 2002) and assuming that the phos- of Th and U in diopsidic clinopyroxenes and are summa- phates in D’Orbigny have a U concentration similar to rized in Table 6. All available data indicate little to no those in AdoR, shows that phosphates in D’Orbigny only Th/U fractionation as DTh/DU is close to unity both in host 30% of the total U (and 70% of the total Th) of partitioning experiments (Benjamin et al., 1979, 1980;

Table 6 Clinopyroxene-liquid partionning data for Th and U.

Sample D(Th) ± D(U) ± D(Th)/D(U) P T log fO2 Initial P (Cation %) Ref. Synthetic Cpxa 0.0032 0.0004 0.0028 0.0005 1.14 ± 0.25 20 kbar 1375–1390 °C 9 8.4 (1) Synthetic Cpxa 0.0019 0.0002 0.0019 0.0002 1.00 ± 0.15 20 kbar 1375–1390 °C 9 5.3 (1) Synthetic Cpxa 0.002 0.002 1.00 1 atm 1210–1330 °C 9 5.3 (2) Synthetic Cpxa 0.029 0.002 0.018 0.001 1.61 ± 0.14 1 atm 1210–1330 °C 9 0 (1) Juan de Fuca 0.008 0.002 0.004 0.002 2.00 ± 1.12 1 atm 1160–1210 °C 11 0 (3) Takashima 0.016 0.002 0.0146 0.0016 1.10 ± 0.18 1 atm 1160–1312 °C 11 0 (3) Synthetic Cpxb 0.0099 0.0013 0.0089 0.0008 1.11 ± 0.18 1 atm 1273–1312 °C 9.5 0 (3) Synthetic Cpxc 0.0030 0.0006 0.0022 0.0003 1.36 ± 0.33 1 atm 1285–1291 °C 9.76 0 (4) Synthetic Cpxc 0.0014 0.0002 0.0014 0.0002 1.00 ± 0.20 1 atm 1285–1291 °C 9.76 0 (4) Synthetic Cpxc 0.0036 0.0004 0.0035 0.0005 1.03 ± 0.19 1 atm 1285–1291 °C 11 0 (4) AdoR Cpxd 0.89 0.86 1.04 (5) References: (1) Benjamin et al. (1980); (2) Benjamin et al. (1979); (3) Latourrette and Burnett (1992); (4) Lundstrom et al. (1994); (5) Wasserburg et al. (1977). a Typical starting composition of Di50An25Ab25 + 15–25% Ca3(PO4)2 and Di67An33 + 25% Ca3(PO4)2. b Starting composition of Di71An29. c Starting composition (wt%): 52.04% SiO2, 0.97% TiO2, 12.13% Al2O3, 0.21% Cr2O3, 11.14% MgO, 21.11% CaO, 2.40% Na2O. d Values calculated as the ratio of the concentration of the element in pyroxenes over the total concentration of the element of the sample. 608 F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617

Latourrette and Burnett, 1992; Lundstrom et al., 1994) and isotopes during clinopyroxene-liquid partitioning of in AdoR (Wasserburg et al., 1977). The lowest D values, Dcpx/liquid 0.25‰. Sensitivity tests (see Supplementary obtained from partitioning experiments, indicate that Materials) indicate that the uncertainty on this value is at almost no Th and U will be incorporated in pyroxenes, most ±0.10‰ and more likely around ±0.05‰. This which is inconsistent with fractionation of U isotopes by fractionation factor is large but not unreasonable as frac- clinopyroxene. These low estimates are, however, very unli- tionation of U isotopes is caused by both mass-dependent kely to be appropriate for angrites as the mass-balance con- effects, which scale as 1/T2 and are negligible at magmatic siderations discussed above show that pyroxenes can host temperature, and volume dependent effects (i.e., nuclear 50–70% of the total U of the sample. The high DTh and field shift, NFS; Bigeleisen, 1996; Schauble, 2007), which DU estimates in Table 6 (DTh = 0.89 and DU = 0.86) are scale as 1/T and could still be significant at high tempera- calculated as the ratio of U concentration in pyroxenes over ture. The fractionation factor used in Fig. 13 (0.25‰)is the total U concentration in AdoR (Wasserburg et al., of the same order of magnitude as the 0.43‰ equilibrium 1977). Using these upper limit values, we test the hypothesis fractionation expected for U(IV)–U(VI) isotopic exchange of U isotope fractionation during incorporation into pyrox- (including NFS and vibrational effects) at 1200 °C (see enes using a Rayleigh distillation model (Fig. 13). Calcula- Eq. (8) in Fujii et al., 2006). tion of the amount of U removed, noted f, is based on the The value of the fractionation factor Dcpx/liquid = 0.25 amount of crystallization at the time of formation of each ± 0.05‰ derived from the Rayleigh distillation model is sample. For quenched angrites, this value is well con- slightly lower than the Dcpx-phos values estimated by mass- strained in our model (which will be the focus of a subse- balance in AdoR and D’Orbigny, which is consistent with quent publication) and corresponds to about 50% of phosphates forming later than pyroxenes and from a pool crystallization, with only the last 25% crystallizing pyrox- of U depleted in 235U relative to the starting liquid. The enes. For plutonic samples, we used the HREE enrichment higher d238U values in phosphates relative to pyroxenes (Tm or Lu) in clinopyroxenes and the DHREE values rele- may thus be a feature common to all angrites, as also sug- vant to the conditions on the angrite parent body (e.g., gested by the 0.55 ± 0.29 Myr age difference (calculated Mckay et al., 1994) to calculate the HREE concentration assuming a common 238U/235U ratio) between two leachate in the liquid at equilibrium with these pyroxenes. The equi- fractions from NWA 4590, one containing mostly pyroxe- librium concentration in the liquid allows us to constrain nes and the other mostly phosphates (Amelin et al., the amount of crystallization at the time of sample forma- 2011). Though this age offset could be due to different clo- tion, and thus, the amount of U removed. To be conserva- sure temperatures of the two minerals (Amelin et al., 2011), tive, the low and high range of HREE enrichments in it can also be readily explained by a difference of +0.38 pyroxenes of the various angrite specimens are considered, ± 0.20‰ between the d238U values of the two phases, with which translates in sometimes large uncertainties on the heavier U isotopic composition in phosphates than pyroxe- f values of plutonic samples. As can be seen on Fig. 13, nes, a value identical to that obtained from mass balance the d238U data is consistent with a fractionation of U considerations in AdoR and D’Orbigny. Furthermore, in a leaching experiment conducted on NWA 6291 with 0.5 M HNO3, Brennecka and Wadhwa (2012) found that Sahara 99555 NWA NWA the leachate fraction had a lighter U isotopic composition D'Orbigny 4590 4801 AdoR (d238U=0.27 ± 0.05/0.17‰) than the residue 0 238 α Liquid (d U=0.57 ± 0.11/0.19‰). They interpreted this as Px-Liq = 0.99975 Δ = -0.25 ‰ -0.1 Px-Liq evidence that the U released in the leaching step was of ter- Inst. restrial origin as it had d238U value identical to the average Solid U (‰) -0.2 crustal d238U value of 0.29 ± 0.03‰ (Tissot and 235

U/ Dauphas, 2015), but this interpretation might be incorrect. -0.3 238

δ Indeed, the leachate contained about 64% of the total U of

-0.4 Cumul. the sample (Suppl. Material of Brennecka and Wadhwa, Solid 2012), which is too large to be reasonably ascribed to terres- -0.5 trial contamination alone and instead hints to dissolution -0.6 of a U rich phase such as phosphates by the 0.5 M HNO3 0 0.2 0.4 0.6 0.8 1 solution. If that is the case, the leachate fraction contains Fraction of U removed U coming both from terrestrial contamination and phos- phates while the residue contains U coming mostly from Fig. 13. d238U vs. fraction of U removed from the melt. The red, the pyroxenes (which are not digested in 0.5 M HNO3). purple and blue solid lines represent the time-integrated evolution As discussed in Section 4.1, only 10% of the U in NWA of U isotope composition of the liquid, instantaneous solid and 6291 is of extraneous origin, meaning that the 0.30‰ dif- cumulative solid, respectively, during Rayleigh distillation, assum- ference between the d238U of the two fractions is represen- ing a Dcpx-liquid fractionation factor of 0.25‰. The data of selected angrites from this study (circles) and literature is displayed: tative of the pyroxene/phosphate difference, which is in line Brennecka and Wadhwa (2012), squares; Connelly et al. (2012), with the other observations listed above. Finally, support triangle; and Goldmann et al. (2015), diamond. (For interpretation for the hypothesis of magmatic differentiation is found in of the references to color in this figure legend, the reader is referred the bulk chemistry of angrites. In Fig. 14, the d238U values to the web version of this article.) of the angrites are plotted as a function of their modal F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 609

0.0 0.0 NWA NWA NWA NWA -0.1 a) AdoR b) AdoR -0.1 4590 4801 4590 4801 -0.2 -0.2 -0.3 -0.3 U (‰) -0.4 -0.4 U (‰) 238 238 δ -0.5 -0.5 δ S. 99555 S. 99555 -0.6 D'Orbigny D'Orbigny -0.6 -0.7 -0.7 5 15 25 35 45 55 65 0 100 200 300 400 Pyroxene (mode, %) % Pyroxene / % Phosphate

Fig. 14. d238U vs. modal abundance of (a) pyroxene and (b) pyroxene/phosphate ratio of selected angrites. d238U values from this study (circles) and literature: Brennecka and Wadhwa (2012), squares; Connelly et al. (2012), triangle; and Goldmann et al. (2015), diamond. Modal abundances from Mittlefehldt and Lindstrom (1990), Mittlefehldt et al. (2002), Sanborn (2012). The lower d238U values observed in samples with higher pyroxene content (and pyroxene/phosphate ratio) is consistent with U isotope fractionation during incorporation into pyroxene during the magmatic evolution of the angrite parent melt. abundance of pyroxene (panel a) and the ratio of abun- where DU is the difference between the actual and assumed dance pyroxene/phosphate (panel b). Whereas quenched U isotope composition of the sample (in delta notation, ‰), angrites have chondritic d238U values and similar amount and t is the Pb-Pb age of the sample obtained using an ‘‘as- of pyroxenes, plutonic angrites define a trend where the sumed” U isotope composition. After correction, the two samples with the highest pyroxene mode (or pyroxene/ quenched angrites (D’Orbigny and Sahara 99555) have phosphate mode) have the lowest d238U values. The broad essentially indistinguishable ages (Fig. 6), at 4563.51 correlation observed between d238U values and U-isotope- ± 0.18/0.29/9.25 Myr and 4564.07 ± 0.43/0.48/9.25 Myr, corrected Pb-Pb ages in angrites (Fig. 6) thus likely reflects respectively, while the plutonic angrites have ages of the increase in pyroxene content (or pyroxene/phosphate 4557.76 ± 0.31/0.38/9.24 Myr for NWA 4590, 4556.82 ± ratio) of the samples with time, and though no mineral 0.17/0.28/9.24 Myr for NWA 4801, 4560.69 ± 0.78/0.81/9.27 mode data are available for NWA 6291, we predict that Myr for NWA 6291, and 4556.45 ± 0.18/0.29/9.24 Myr the pyroxene/phosphate ratio of this sample is between that for AdoR. If the average U isotopic composition of all of quenched angrites (at 55) and AdoR (at 155). angrites is used instead of the individual U isotopic composition of each sample, as was done in Brennecka 4.3. Implications of the 238U/235U ratio of quenched angrites and Wadhwa (2012), the ages are shifted by 0.17 and 0.29 for ESS chronology Myr for D’Orbigny and Sahara 99555, respectively (Table 7). The correction imparted by the recognition that Quenched angrites are key samples when cross- all angrites do not have the same U isotopic composition calibrating relative (e.g., 26Al-26Mg) and absolute (Pb-Pb) is small but significant. The U-isotope-corrected Pb-Pb ages as they meet the requirements of non-disturbance, syn- ages presented in Table 4, which are all calculated using chronous isotope closure and phase diversity necessary to the individual d238U value of the samples (even for litera- make them reliable chronometric anchors. Our observation ture data) should therefore supersede previously calculated that plutonic and quenched angrites have different U iso- values. topic compositions implies that one cannot use the average In Section 4.2 we propose that a U isotope fractionation 238U/235U ratio of all angrites to correct their Pb-Pb ages as of 0.25 ± 0.05‰ exists between pyroxene and silicate was done previously (Brennecka and Wadhwa, 2012) as this melt. Because all uncorrected Pb-Pb ages are derived from can potentially lead to inaccuracies. Instead, the correction Pb isotope measurements of pyroxenes fractions, age cor- should consider each angrite on a case-by-case basis, or at rections should be calculated using the 238U/235U ratio of the very least, consider separately plutonic and quenched the pyroxenes rather than that of the bulk sample. So far, angrites. only one pyroxene fraction, from D’Orbigny, has been mea- Table 4 presents a summary of the non-corrected Pb-Pb sured (Brennecka and Wadhwa, 2012), which has a d238U ages of the six angrites studied herein, their measured U value 0.09 ± 0.22‰ lower than the bulk sample (the spike isotope ratios (this work and literature data), and their uncertainty is not considered as the same spike was used U-isotope-corrected ages. The age corrections, Dt, are for measurements of both the bulk sample and pyroxene calculated using the following equation (Eq. 12 of Tissot separate) and leads to an age of D’Orbigny of 4563.35 and Dauphas, 2015), ± 0.32/0.38/9.25 Myr. The gain in accuracy is accompanied k k by a loss in precision making this age indistinguishable D ð 238 t Þð 235 t Þ D ¼ U e 1 e 1 t k k ðk þk Þ from our estimate of 4563.51 ± 0.18/0.29/9.25 Myr, calcu- 1000 ðk e 238 t k e 235 t þðk k Þe 235 238 tÞ 238 235 235 238 lated using the bulk 238U/235U. Given the lack of U concen- ð Þ 13 tration data in angrite pyroxenes, we tentatively suggest 610 F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617

Table 7 Effect of the choice of the d238U values used on the Pb-Pb age corrections calculated. Sample d238U(‰)a Age correction Dt (Myr) Difference (Myr) 1: Using the d238U value of each sample, separately NWA 4590 0.47 ± 0.09/0.18 0.17 ± 0.13/0.26 NWA 4801 0.55 ± 0.06/0.17 1.24 ± 0.08/0.24 NWA 6291 0.40 ± 0.04/0.16 0.60 ± 0.06/0.23 AdoR 0.51 ± 0.08/0.18 1.20 ± 0.12/0.26 D’Orbigny 0.32 ± 0.09/0.18 0.91 ± 0.13/0.26 Sahara 99555 0.23 ± 0.14/0.21 0.79 ± 0.20/0.30

238 2: Using the average d U values of plutonic or quenched angrites Dt1–Dt2 NWA 4590 0.46 ± 0.03/0.16 0.16 ± 0.04/0.23 0.02 NWA 4801 0.46 ± 0.03/0.16 1.11 ± 0.04/0.23 0.13 NWA 6291 0.46 ± 0.03/0.16 0.69 ± 0.04/0.23 +0.09 AdoR 0.46 ± 0.03/0.16 1.11 ± 0.04/0.23 0.08 D’Orbigny 0.29 ± 0.07/0.17 0.88 ± 0.11/0.25 0.04 Sahara 99555 0.29 ± 0.07/0.17 0.88 ± 0.11/0.25 +0.09

238 3: Using the average d U value of all angrites Dt1–Dt3 NWA 4590 0.43 ± 0.03/0.16 0.13 ± 0.04/0.23 0.05 NWA 4801 0.43 ± 0.03/0.16 1.08 ± 0.04/0.23 0.16 NWA 6291 0.43 ± 0.03/0.16 0.66 ± 0.04/0.23 +0.06 AdoR 0.43 ± 0.03/0.16 1.08 ± 0.04/0.23 0.11 D’Orbigny 0.43 ± 0.03/0.16 1.08 ± 0.04/0.23 +0.17 Sahara 99555 0.43 ± 0.03/0.16 1.08 ± 0.04/0.23 +0.29 a Values renormalized to CRM-112a.

that an additional correction of 0.19 Myr be applied to reported the Pb and U isotopic compositions of a new CAI ages corrected using the bulk 238U/235U ratios of angrites (B4, from NWA 6991 CV3 chondrite) with canonical 26 27 reported in Table 4. This number assumes that 50% of ( Al/ Al)0 ratio, which gives an age of 4567.94 the U is present in pyroxenes, and uses the fractionation ± 0.21/0.31/9.25 Myr. This value is in agreement with the factor of 0.25‰ determined in Section 4.2 making the : þ0:2 age of 4568 20:4 Myr from CAI 2364-B1 (Bouvier and pyroxenes 0.13‰ lighter than the bulk rock. Wadhwa, 2010), in which the U-isotope composition was not measured but calculated based on the Th/U ratios of 4.4. Concordance of absolute and relative chronologies? the CAI and the broad Th/U vs d238U correlation observed in CAIs (Brennecka et al., 2010b). For comparison, Amelin In Fig. 15 and Table 8, the U-isotope-corrected Pb-Pb et al. (2010) and Connelly et al. (2012) found values of age intervals of angrites anchored to D’Orbigny (Table 3) 4567.18 ± 0.50/0.55/9.26 (Allende CAI SJ-101) and and the oldest known CAIs (Amelin et al., 2010; Bouvier 4567.30 ± 0.16/0.20/9.25 Myr (weithed average of the ages et al., 2011; Connelly et al., 2012), are compared to their rel- of three Efremovka CAIs and SJ-101). The reason for the 182 182 ative ages obtained using the Hf- W (top left panel), discrepancy between these studies is unknown but could 53 53 26 26 Mn- Cr (top right panel) and Al- Mg (bottom panels) stem in part from the chemical leaching procedure used systematics. The inset in each panel shows the residuals. to eliminate common Pb from the measurements. CAIs There are no Mn-Cr ages of CAIs because (i) manganese are rich in U but significant common Pb can be present is volatile and individual minerals in CAIs, which con- in the samples from terrestrial contamination and parent- densed at high-T, only show low Mn/Cr ratios resulting body alteration. To estimate the Pb-Pb age of the CAIs, in imprecise and inaccurate isochrons (see Davis and the most commonly used approach is to leach the sample McKeegan, 2014., and references therein), and (ii) the exis- and to plot the results in a 207Pb/206Pb vs. 204Pb/206Pb dia- 53 tence of nucleosynthetic anomalies in Cr in most CAIs gram. The leached data points define a mixing line between (e.g., Papanastassiou, 1986), further complicates any common Pb and radiogenic Pb at 204Pb/206Pb = 0. Decay 53 chronological interpretation of small Cr variations in of 238U and 235U can implant 206Pb and 207Pb in slightly dif- CAIs. As shown in Fig. 15, there is overall very good agree- ferent crystallographic sites. Such difference, together with ment between the Mn-Cr, Hf-W, and Pb-Pb chronometers, intrinsic differences in chemical bonds between 206Pb, which supports the idea that the parent radionuclides of 207Pb, and surrounding atoms, as well as kinetic effects, these various systems were homogeneously distributed in could induce 207Pb/206Pb fractionation during leaching the ESS. (Zartman et al., 2006; Chen and Papanastassiou, 2008; Comparison between Pb-Pb and Al-Mg ages of CAIs Amelin et al., 2009). To be able to resolve age differences and angrites is complicated by the fact that determinations of better than 1 Myr for 4567 Myr old samples, the of Pb-Pb ages of CAIs do not all agree. Bouvier et al. (2011) 207Pb/206Pb isotopic fractionation imparted by the leaching F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 611

Angrites D'Orbigny CAIs -10 -8 -6 -4 -2 0 2 4 6 -10 -8 -6 -4 -2 0 2 4 6 6 6 182 182 53 53 4 Hf- W SJ-101 Mn- Cr 4 2 B4 2

0 NWA Sahara Sahara 0 4590 99555 NWA 99555 -2 AdoR -2 4801 NWA 1 -4 0.5 -4 ζ 4801 ζ -6 -0.5 0 -6 -8 -1.5 NWA -1 -8

Age rel. to D'Orbigny (Myr) AdoR -8 -4 0 4 4590 -8 -6 -4 -2 0 2 -10 -10 6 6 26 26 SJ-101 26 26 SJ-101 4 Al- Mg Al- Mg 4 Spivak-Birndorf B4 Schiller B4 2 et al. (2009) et al. (2015) 2 0 Sahara Sahara 0 99555 99555 -2 -2 -4 1 1 -4 -6 ζ 0 ζ 0 -6 -8 -1 -1 -8

Age rel. to D'Orbigny (Myr) 0 2 4 0 2 4 -10 -10 -10 -8 -6 -4 -2 0 2 4 6 -10 -8 -6 -4 -2 0 2 4 6 Pb-Pb age interval, Pb-Pb age interval, rel. to D'Orbigny (Myr) rel. to D'Orbigny (Myr)

Fig. 15. Comparison of relative ages from the U-isotope-corrected Pb-Pb system (this work, Tables 3 and 4) with relative ages from the 182Hf-182W system (top left panel), the 53Mn-53Cr system (top right panel), and the 26Al-26Mg system (bottom panels). Insets show residuals. D’Orbigny is used as the anchor. The errors on Pb-Pb relative ages are the ‘‘z” in the ±x/y/z values in Table 3. The CAI point labeled ‘‘SJ- 101” uses the weighted average age of 4567.30 ± 0.16/0.20/9.25 Myr based on CAI SJ-101 (Amelin et al., 2010) and three Efremovka CAIs (Connelly et al., 2012); all of which have the same age within uncertainties and are assumed to be characterized by the initial Solar System 26Al/27Al ratio of (4.96 ± 0.25) 105 (Jacobsen et al., 2008) and 182Hf/180Hf ratio of (10.18 ± 0.43) 105 (Kruijer et al., 2014). The CAI point labeled ‘‘B4” uses the U-isotope corrected Pb-Pb age of 4567.94 ± 0.21/0.31/9.25 Myr (Bouvier et al., 2011), and the 26Al/27Al ratio of (4.90 ± 0.05) 105 (Wadhwa et al., 2014) measured in CAI B4 from the CV3 chondrite NWA 6991. Hf-W and Pb-Pb ages are highly 26 27 concordant. Depending on which ( Al/ Al)0 ratios for D’Orbigny and Sahara 99555 are used (left, Spivak-Birndorf et al., 2009 or right, Schiller et al., 2015), the Al-Mg and Pb-Pb chronometers end up being concordant or discordant. See text for details. procedure should not exceed 0.21‰ (equivalent to an age (2009) are (5.06 ± 0.92) 107 and (5.13 ± 1.90) 107, offset of 0.30 Myr). This is small with regard to the extent respectively. In contrast, Schiller et al. (2015) reports ratios of the stable isotopic fractionations that are observed for of (3.98 ± 0.15) 107 and (3.64 ± 0.18) 107, respec- other isotope systems in low-T aqueous processes (Teng tively, for these samples. When the Al-Mg data from et al., 2016 and references therein), and it is therefore con- Spivak-Birndorf et al. (2009) is considered, the Al-Mg ceivable that the leaching procedure imparts Pb isotopic and Pb-Pb ages of angrites and the oldest CAIs (CAI B4, fractionation at a level that matters for high-precision Pb- Bouvier et al., 2011) agree within uncertainty (Fig. 15, Pb dating of ESS phases. lower left panel, and Table 8), implying uniform distribu- A second problem that prevents one from testing the tion of 26Al in the ESS, while when the Al-Mg data from concordance of the Al-Mg and Pb-Pb systematics is that Schiller et al. (2015) is used, the Al-Mg and Pb-Pb ages of initial 26Al/27Al ratios reported for quenched angrites also angrites and CAIs are discrepant by at least 0.55 do not agree. The initial 26Al/27Al ratios of D’Orbigny ± 0.28 Myr (Pb-Pb age of CAI B4) and at most 1.20 and Sahara 99555 obtained by Spivak-Birndorf et al. ± 0.25 Myr (Pb-Pb age of SJ-101*)(Fig. 15, lower right 1 ...Tso ta./Gohmc tCsohmc ca23(07 593–617 (2017) 213 Acta Cosmochimica et Geochimica / al. et Tissot F.L.H. 612

Table 8 Comparison of Pb-Pb, Hf-W, Mn-Cr and Al-Mg ages of angrites and CAIs.

Sample Pb-Pb Hf-W Mn-Cr Al-Mg (Angrites from Spivak-Birndorf et al. (2009)) Al-Mg (Angrites from Schiller et al. (2015))

a 182 180 53 55 * 26 27 26 27 Dt D’Or. (Myr) ( Hf/ Hf)0 Ref. Dt CAI B4 Dt D’Or. ( Mn/ Mn)0 Ref. Dt D’Or. Dt CAI SJ-101 Dt CAI B4 ( Al/ Al)0 Ref. Dt CAI B4 Dt D’Or. ( Al/ Al)0 Ref. Dt CAI B4 Dt D’Or. (Myr) (Myr) (Myr) (Myr) (Myr) (Myr) (Myr) (Myr) (Myr)

Angrites 5 +0.70 6 +0.67 +0.75 +0.78 NWA 4590 5.75 ± 0.36 (4.63±0.17) 10 (1) 10.120.71 5.58 ± 0.55 (1.01 ± 0.12) 10 (2) 6.290.60 10.080.70 10.720.75 5 +0.69 +0.54 6 +0.39 +0.57 NWA 4801 6.69 ± 0.25 (4.52 ± 0.16) 10 (1) 10.420.70 5.890.53 (1.00 ± 0.07) 10 (3) 6.340.36 10.14 ± 0.52 10.780.58 NWA 6291 2.82 ± 0.80 5 +0.93 +0.83 6 +2.53 +2.55 +2.56 Angra dos Reis 7.05 ± 0.25 (4.02 ±0.24) 10 (1) 11.930.91 7.390.79 (1.10 ± 0.42) 10 (4) 5.831.71 9.621.75 10.261.78 5 +0.60 6 +0.35 +0.42 7 +0.20 7 D’Orbigny 0 (7.15 ±0.17) 10 (1) 4.540.62 0 (3.24 ± 0.04) 10 (5) 0 3.790.37 4.430.45 (5.06 ± 0.92) 10 (7) 4.730.17 0 (3.98 ± 0.15) 10 (9) 4.98 ± 0.04 0 b 5 +0.59 6 +0.31 +0.46 +0.51 7 +0.47 +0.49 7 Sahara 99555 0.57 ± 0.46 (6.87 ±0.15) 10 (1) 5.050.61 0.51 ± 0.41 (3.40 ± 0.19) 10 (6) 0.260.29 3.530.47 4.170.53 (5.13 ± 1.90) 10 (7) 4.720.32 0.010.38 (3.64 ± 0.18) 10 (9) 5.07 ± 0.05 0.09 ± 0.06 Sahara 99555c 0.29 ± 0.30 CAIs 5 § +0.62 6 §§ 5 *** +0.18 5 *** Wtd avg 3.79 ± 0.24 (10.18 ±0.43) 10 0 4.540.60 (6.54 ± 0.44) 10 (4.96 ± 0.25) 10 0.01 ± 0.05 4.740.21 (4.96 ± 0.25) 10 0.01 ± 0.05 4.99 ± 0.06 * CAIs SJ-101 ** 5 § +0.62 6 §§§ 5 +0.17 5 CAI B4 4.43 ± 0.27 (10.18 ±0.43) 10 0 4.540.60 (7.37 ± 0.60) 10 (4.90 ± 0.05) 10 (8) 0 4.730.20 (4.90 ± 0.05) 10 (8) 0 4.98 ± 0.04 References: (1) Kleine et al. (2012); (2) Yin et al. (2009); (3) Shukolyukov et al. (2009); (4) Lugmair and Shukolyukov (1998); (5) Glavin et al. (2004); (6) McKibbin et al. (2015); (7) Spivak-Birndorf et al. (2009); (8) Wadhwa et al. (2014);(9)Schiller et al. (2015). * CAI age of 4567.30 ± 0.16/0.20/9.25 Myr: weigthed average of the age of CAI SJ-101 (Amelin et al., 2010), and three Efremovka CAIs (Connelly et al., 2012), which have indistinguishable ages within uncertainties. ** CAI age of 4567.94 ± 0.21/0.31/9.25 Myr: age of CAI B4 from NWA 6991 CV3 chondrite (Bouvier et al., 2011). *** The canonical 26Al/27 Al ratio from Jacobsen et al. (2008) is assumed to be representative of the CAIs measured by Amelin et al. (2010) and Connelly et al. (2012). a Includes all random and systematic errors, to the exception of the error on the U double-spike composition since all samples were spiked with IRMM-3636. b Pb-Pb age from Amelin (2008b). c Pb-Pb age from Connelly et al. (2008). § The initial 182 Hf/180 Hf ratio from Kruijer et al. (2014) is assumed to be representative of the oldest CAIs. §§ The Solar System initial 53Mn/55Mn ratio is calculated using the 53Mn/55Mn ratio and U-isotope-corrrected Pb-Pb age of D’Orbigny and 4567.30 ± 0.20 Myr as the age of the Solar System (Amelin et al. (2010), Connelly et al. (2012)). §§§ The Solar System initial 53Mn/55Mn ratio is calculated using the 53Mn/55Mn ratio and U-isotope-corrrected Pb-Pb age of D’Orbigny and the age of CAI B4 (4567.94 ± 0.31 Myr) as the age of the Solar System (Bouvier et al. (2011)). Table 9 53 55 613 593–617 (2017) 213 Acta Cosmochimica et Geochimica / al. et Tissot F.L.H. Recent and preferred estimates of the Solar System initial ( Mn/ Mn)0 ratio. 53 55 Study ( Mn/ Mn)0 Method Data used/comments Trinquier et al. (2008) (6.38 ± 1.21) 10 6 Ste. Marguerite Mn-Cr internal isochron Non U-corrected Pb–Pb ages of phosphates in the St. (6.28 ± 0.66) 10 6 Orgueil (C1) Mn-Cr internal isochron Marguerite (4562.7 ± 0.6 Myr, [1]) and CAI E60 (6.53 ± 1.93) 10 6 Bulk carbonaceous, ordinary, and enstatite chondrites (4567.11 ± 0.16 Myr, [2]) from Efremovka isochron 6 53 55 –6 Shukolyukov and Lugmair (2006) (8.5 ± 1.5) 10 Bulk carbonaceous chondrites isochron ( Mn/ Mn)LEW86010 = (1.25 ± 0.07) 10 [3], and Moynier et al. (2007) (8.5 ± 1.2) 10 6 Bulk carbonaceous chondrites isochron, combined non U-corrected Pb–Pb age of LEW86010 (4557.8 with the previous study ± 0.5 Myr, [4]) 6 26 27 5 Nyquist et al. (2009) (9.1 ± 1.7) 10 Mn-Cr and Al-Mg comparison in bulk chondrites and Solar Sysytem initial ( Al/ Al)0 = 5.1 10 [5] achondrites Qin et al. (2010) (5.4 ± 2.4) 10 6 Bulk carbonaceous, ordinary, enstatite and R Non U-corrected Pb–Pb age of LEW86010 (4558.55 chondrites isochron ± 0.15 Myr, [6]) Go¨pel et al. (2015) 6.8 10 6 Using the D’Orbigny/CAI Pb-Pb age difference to U-corrected Pb-Pb age of D’Orbigny (4563.37 ± 0.25 53 55 53 55 6 back calculate the initial Mn/ Mn ratio at Myr, [8]), ( Mn/ Mn)D’Orbigny =3.24 10 [9] TSS = 4567.32 Myr [7] This work If TSS = 4567.30 ± 0.20 Myr Using the D’Orbigny/CAI Pb-Pb age difference to U-corrected Pb-Pb age of D’Orbigny (4563.51 ± 0.25/ 6 53 55 53 55 (6.54 ± 0.44) 10 back calculate the initial Mn/ Mn ratio at TSS 0.29/9.24 Myr, [10]), ( Mn/ Mn)D’Orbigny = (3.24 6 If TSS = 4567.94 ± 0.31 Myr ± 0.04) 10 [9] (7.37 ± 0.60) 10 6 Recommended (7 ± 1) 10 6 References: [1] Go¨pel et al. (1994);[2]Amelin et al. (2006); [3] Lugmair and Shukolyukov (1998); [4] Lugmair and Galer (1992); [5] Lee et al. (1977); [6] Amelin (2008a);[7]Connelly et al. (2012); [8] Brennecka and Wadhwa (2012); [9] Glavin et al. (2004); [10] this work. 614 F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 panel, and Table 8). In the face of the discrepant data composition than plutonic samples (Fig. 5). Present in pre- obtained by these two research groups, one cannot favor vious studies, this variability had been overlooked. one set of data over the other, which is why both are pre- Quenched and plutonic angrites also have distinct REE pat- sented in Fig. 15 and Table 8. We do note, however, that terns: flat for quenched angrites and fractionated and most when using the Al-Mg data from Schiller et al. (2015) the likely controlled by pyroxene, plagioclase and olivine frac- Al-Mg and Pb-Pb ages of the two quenched angrites are tionation for plutonic angrites (Figs. 7 and 8). also in disagreement. Despite the potential for U re-mobilization and fraction- At present, it is impossible to tell whether the Pb-Pb and ation inherent to meteorite finds, the status of quenched Al-Mg systems agree and thus if 26Al was uniformly dis- angrites as reliable anchor for ESS chronology is strength- tributed in the ESS. More work, both on quenched angrites ened by the observations of limited 234U/238U disequilib- and CAIs, will be needed to answer these long standing rium, chondritic REE patterns, Th/U ratios and d238U questions. values in these samples. Using D’Orbigny as the anchor, U-isotope-corrected Pb-Pb of angrites are calculated 53 55 60 56 4.5. Early solar system ( Mn/ Mn)0 and ( Fe/ Fe)0 ratios (Table 4). A careful characterization of the errors on abso- lute ages and age intervals is done (Table 3), including the Even though no reliable Mn-Cr ages have been obtained role of the U decay-constants errors (Figs. 2 and 3). A high thus far in CAIs, the agreement between the U-isotope- degree of concordance is observed between the relative ages corrected Pb-Pb ages and the Mn-Cr ages in angrites obtained from short-lived chronometers (53Mn-53Cr, (Fig. 15, upper right panel) suggests that a reasonable esti- 182Hf-182W) and the revised Pb-Pb ages presented herein 53 55 mate of the initial ( Mn/ Mn)0 in the ESS can be calcu- (Fig. 15 and Table 3 and 8). The improved concordance lated using the Pb-Pb age difference between D’Orbigny between the Mn-Cr and Pb-Pb chronometers allows us to 53 55 and the oldest CAIs. Using the absolute age of D’Orbigny assess the initial ( Mn/ Mn)0 ratio in the ESS as being of 4563.51 ± 0.18/0.29 Myr (this work), the 53Mn/55Mn between (6.54 ± 0.44) 106 and (7.37 ± 0.60) 106 ratio of D’Orbigny of (3.24 ± 0.04) 106 (Glavin et al., (Table 9). 2004), and the age of the Solar System as 4567.94 Conclusions on the concordance of the Al-Mg and Pb- ± 0.31 Myr (CAI B4, Bouvier et al., 2011), the initial Pb systems are hindered by the lack of consensus on the 53 55 6 26 27 ( Mn/ Mn)0 ratio calculated is (7.37 ± 0.60) 10 .If age of the Solar System and the initial ( Al/ Al)0 ratios instead, the younger age of the Solar System of 4567.30 in quenched angrites, which will require further work to ± 0.20 Myr (Amelin et al., 2010; Connelly et al., 2012)is improve the absolute and relative ages of ESS anchors. 53 55 238 used, the initial ( Mn/ Mn)0 ratio calculated is (6.54 The variable d U and REE pattern of quenched and ± 0.44) 106. These two values (within uncertainty of plutonic angrites cannot be readily explained by terrestrial each other) encompass all other recent estimates of the ini- contamination, secondary processes such as aqueous alter- 53 55 247 tial ( Mn/ Mn)0 ratio (see Table 9) and the biggest hin- ation, or decay of live Cm (Fig. 11). Instead, the fraction- drance to properly calculate this value is the uncertain Pb- ation of the REE patterns is consistent with the magmatic Pb age of CAIs. We therefore recommend a conservative evolution of the angrite parent melt, and we propose that 53 55 6 238 value of ( Mn/ Mn)0 =(7±1) 10 until the debate the variability in the d U values of angrites is in fact evi- on the age of the Solar System is settled. Our estimate, based dence for U stable isotope fractionation during magmatic on U-isotope-corrected Pb-Pb ages, should be preferred to processes. We propose that a change in the coordination previous estimates that used non U-isotope-corrected Pb- environment of U during incorporation into pyroxene is Pb ages or didn’t provide uncertainties. responsible for the U isotope fractionation (Figs. 13 and Theoretically, the Pb-Pb ages of angrites and CAIs 14). Investigations of the coordination environment of U could also be compared to their relative ages obtained using in igneous minerals using extended X-ray absorption fine 60 60 the Fe- Ni chronometer (t1/2 = 2.62 Myr). At this writ- structure (EXAFS) spectroscopy will be necessary to con- ing, however, the data set is too small to allow for such a firm this hypothesis. comparison as internal isochrons and initial 60Fe/56Fe The variability of the U isotope composition within dif- ratios have only been determined in two angrite samples ferent minerals of a single sample means that properly cor- (D’Orbigny and Sahara 99555, Quitte et al., 2010; Spivak- rected Pb-Pb ages should use the U only from the fraction Birndorf et al., 2011; Tang and Dauphas, 2012, 2015) and measured for Pb isotopes (most often, the pyroxene there are no data available in CAIs, owing to their low fraction). Fe content and the fact that the Fe carrying phases in CAIs have low Fe/Ni ratios. ACKNOWLEDGMENTS

5. CONCLUSION Constructive criticisms from Greg Brennecka, two anonymous reviewers, and editor Yuri Amelin greatly helped improve the manuscript. This work was funded by NSF (grants CSEDI High-precision uranium isotope composition of four plu- EAR1502591 and Petrology and Geochemistry EAR1444951) tonic angrites (NWA 4590, NWA 4801, NWA 6291, and and NASA (grants LARS NNX17AE86G, EW NNX17AE87G, Angra dos Reis) and two quenched angrites (D’Orbigny and SSW NNX15AJ25G) to ND, NASA grant NNX16AD29G and Sahara 99555) was measured (Fig. 5 and Table 1). to TG, and a Crosby Postdoctoral Fellowship to FT. The Robert 238 235 Some heterogeneity in the U/ U ratios of angrites is A. Pritzker Center for Meteoritics and Polar Studies at the Field observed, with quenched samples having heavier U isotope Museum and P. Heck, the Museum National d’Histoire Naturelle F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 615

(Paris) and the Smithsonian Institute are thanked for providing formation of the first solids in the Solar System, Hawaii, p. meteorite samples. G. Brennecka is thanked for providing REE, 9054. Th and U concentrations data for some angrites. Bouvier A. and Wadhwa M. (2010) The age of the Solar System Authors Contributions: ND and FT designed the research; FT redefined by the oldest Pb-Pb age of a meteoritic inclusion. Nat. performed the research; FT, ND and TG interpreted the data and Geosci. 3, 637–641. wrote the paper. Brennecka G. A., Borg L. E., Hutcheon I. D., Sharp M. A. and Anbar A. D. (2010a) Natural variations in uranium isotope ratios of uranium ore concentrates: understanding the U-238/ 235 APPENDIX A. SUPPLEMENTARY DATA U- fractionation mechanism. Earth Planet Sci. Lett. 291, 228–233. Brennecka G. A., Weyer S., Wadhwa M., Janney P. E., Zipfel J. Supplementary data associated with this article can be and Anbar A. D. (2010b) 238U/235U variations in meteorites: found, in the online version, at http://dx.doi.org/10.1016/ extant 247Cm and implications for Pb-Pb dating. Science 327, j.gca.2017.06.045. 449–451. Brennecka G. A. and Wadhwa M. (2012) Uranium isotope REFERENCES compositions of the basaltic angrite meteorites and the chronological implications for the early Solar System. Proc. Amelin Y. (2006) The prospect of high-precision Pb isotopic dating Natl. Acad. Sci. USA 109, 9299–9303. of meteorites. Meteorit. Planet. Sci. 41, 7–17. Brett R., Huebner J. S. and Sato M. (1977) Measured oxygen Amelin Y. (2008a) U-Pb ages of angrites. Geochim. Cosmochim. fugacities of Angra-Dos-Reis achondrite as a function of Acta 72, 221–232. temperature. Earth Planet Sci. Lett. 35, 363–368. Amelin Y. (elin, 2008 b) The U-Pb systematics of angrite Sahara Chakraborty S. (1997) Rates and mechanisms of Fe-Mg interdif- 99555. Geochim. Cosmochim. Acta 72, 4874–4885. fusion in olivine at 980–1300 °C. J. Geophys. Res. Sol. Ea. 102, Amelin Y. and Irving A. J. (2007) Seven million years of evolution 12317–12331. on the angrite parent body from Pb-isotopic data. In Chronol- Chen J. H. and Papanastassiou D. (2008) The Concordancy of ogy of Meteorites and the Early Solar System. pp. 20–21. Uranium-Lead Ages in Meteorites, Lunar and Planetary Amelin Y., Connelly J., Zartman R. E., Chen J. H., Gopel C. and Science Conference, Houston, p. 1956. Neymark L. A. (2009) Modern U-Pb chronometry of mete- Chen J. H., Wasserburg G. J. and Papanastassiou D. A. (1993) The orites: advancing to higher time resolution reveals new prob- Th and U abundances in chondritic meteorites. Lunar Planet. lems. Geochim. Cosmochim. Acta 73, 5212–5223. Sci. Conf., 277–278. Amelin Y., Kaltenbach A., Iizuka T., Stirling C. H., Ireland T. R., Cheng H., Edwards R. L., Shen C. C., Polyak V. J., Asmerom Y., Petaev M. and Jacobsen S. B. (2010) U-Pb chronology of the Woodhead J., Hellstrom J., Wang Y. J., Kong X. G., Spotl C., Solar System’s oldest solids with variable U-238/U-235. Earth Wang X. F. and Alexander E. C. (2013) Improvements in Th- Planet. Sci. Lett. 300, 343–350. 230 dating, Th-230 and U-234 half-life values, and U-Th Amelin Y., Kaltenbach A. and Stirling, C. H. (2011) The U-Pb isotopic measurements by multi-collector inductively coupled Systematics and Cooling Rate of Plutonic Angrite NWA 4590, plasma mass spectrometry. Earth Planet. Sci. Lett. 371, 82–91. Lunar and Planetary Science Conference, Houston, TX, p. Cherniak D. J. (1995) Diffusion of lead in plagioclase and K- 1682. Feldspar - an investigation using Rutherford backscattering Amelin Y., Wadhwa M. and Lugmair G. (2006) Pb-isotopic dating and resonant nuclear-reaction analysis. Contrib Mineral Petr of meteorites using 202Pb–205Pb double-spike: comparison 120, 358–371. with other high-resolution chronometers., Lunar and Planetary Cherniak D. J. (1998) Pb diffusion in clinopyroxene. Chem. Geol. Science Conference, Houston, TX, p. 1970. 150, 105–117. Baghdadi B., Jambon A. and Barrat J. A. (2015) Metamorphic Cherniak D. J., Lanford W. A. and Ryerson F. J. (1991) Lead angrite Northwest Africa 3164/5167 compared to magmatic diffusion in apatite and zircon using ion-implantation and angrites. Geochim. Cosmochim. Acta 168, 1–21. rutherford backscattering techniques. Geochim. Cosmochim. Baker J., Bizzarro M., Wittig N., Connelly J. and Haack H. (2005) Acta 55, 1663–1673. Early planetesimal melting from an age of 4.5662 Gyr for Cherniak D. J. and Van Orman J. A. (2014) Tungsten diffusion in differentiated meteorites. Nature 436, 1127–1131. olivine. Geochim. Cosmochim. Acta 129, 1–12. Barrat J. A., Blichert-Toft J., Nesbitt R. W. and Keller F. (2001) Condon D. J., McLean N., Noble S. R. and Bowring S. A. (2010) Bulk chemistry of Saharan shergottite Dar al Gani 476. Isotopic composition (U-238/U-235) of some commonly used Meteorit. Planet. Sci. 36, 23–29. uranium reference materials. Geochim. Cosmochim. Acta 74, Benjamin T., Heuser W. R., Burnett D. S. and Seitz M. G. (1980) 7127–7143.

Actinide crystal-liquid partitioning for clinopyroxene and Ca3(- Connelly J. N., Bizzarro M., Krot A. N., Nordlund A., Wielandt Po4)2. Geochim. Cosmochim. Acta 44, 1251–1264. D. and Ivanova M. A. (2012) The absolute chronology and Benjamin T. M., Jones J. H. and Burnett D. S. (1979) Laboratory thermal processing of solids in the solar protoplanetary disk. partitioning studies testing the validity of 244Pu-rare earth Science 338, 651–655. chronology. Lunar Planetary Sci. Conf., 98–100. Connelly J. N., Bizzarro M., Thrane K. and Baker J. A. (2008) The Bigeleisen J. (1996) Nuclear size and shape effects in chemical pb-pb age of angrite SAH99555 revisited. Geochim. Cosmochim. reactions. Isotope chemistry of the heavy elements. J. Am. Acta 72, 4813–4824. Chem. Soc. 118, 3676–3680. Crozaz G., Floss C. and Wadhwa M. (2003) Chemical alteration Bopp C. J., Lundstrom C. C., Johnson T. M. and Glessner J. J. G. and REE mobilization in meteorites from hot and cold deserts. (2009) Variations in U-238/U-235 in uranium ore deposits: Geochim. Cosmochim. Acta 67, 4727–4741. Isotopic signatures of the U reduction process? Geology 37, Dauphas N. and Chaussidon M. (2011) A perspective from extinct 611–614. radionuclides on a young stellar object: the sun and its accretion Bouvier A., Brennecka G. A. and Wadhwa M. (2011) Absolute disk. In: Jeanloz R. and Freeman K. H. (Eds.), Annual Review chronology of the first solids in the Solar System, Workshop on of Earth and Planetary Sciences, vol 39, pp. 351–386. 616 F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617

Davis A. M. and McKeegan K. D. (2014) 1.11 - short-lived Bizzarro (2011) Evidence for magnesium isotope heterogeneity radionuclides and early solar system chronology A2 - Turekian. in the solar protoplanetary disk. Astrophys. J. Lett., 735. In Treatise on Geochemistry (eds. D. Heinrich and K. Hol- Latourrette T. Z. and Burnett D. S. (1992) Experimental-determi- landKarl), second ed. Elsevier, Oxford, pp. 361–395. nation of U-partitioning and Th-partitioning between clinopy- Dodson M. H. (1973) Closure temperature in cooling geochrono- roxene and natural and synthetic basaltic liquid. Earth Planet logical and petrological systems. Contrib. Mineral Petr. 40, 259– Sci. Lett. 110, 227–244. 274. Lee T., Papanastassiou D. A. and Wasserburg G. J. (1977) Al-26 in Floss C., Crozaz G., McKay G., Mikouchi T. and Killgore M. early solar-system - fossil or fuel. Astrophys. J. 211, L107–L110. (2003) Petrogenesis of angrites. Geochim. Cosmochim. Acta 67, Longhi J. (1999) Phase equilibrium constraints on angrite petro- 4775–4789. genesis. Geochim. Cosmochim. Acta 63, 573–585. Fujii Y., Higuchi N., Haruno Y., Nomura M. and Suzuki T. (2006) Ludwig K. R. (2000) Decay constant errors in U-Pb concordia- Temperature dependence of isotope effects in uranium chemical intercept ages. Chem. Geol. 166, 315–318. exchange reactions. J. Nucl. Sci. Technol. 43, 400–406. Lugmair G. W. and Galer S. J. G. (1992) Age and isotopic Gellissen M., Palme H., Korotev R. L. and Irving A.J. (2007). relationships among the angrites Lewis Cliff-86010 and Angra- NWA 2999, A Unique Angrite with a Large Chondritic Dos-Reis. Geochim. Cosmochim. Acta 56, 1673–1694. Component, Lunar and Planetary Science Conference, Hous- Lugmair G. W. and Shukolyukov A. (1998) Early solar system ton, p. 1612. timescales according to Mn-53-Cr-53 systematics. Geochim. Glavin D. P., Kubny A., Jagoutz E. and Lugmair G. W. (2004) Cosmochim. Acta 62, 2863–2886. Mn-Cr isotope systematics of the D’Orbigny angrite. Meteorit. Lundstrom C. C., Shaw H. F., Ryerson F. J., Phinney D. L., Gill J. Planet. Sci. 39, 693–700. B. and Williams Q. (1994) Compositional controls on the Goldmann A., Brennecka G., Noordmann J., Weyer S. and partitioning of U, Th, Ba, Pb, Sr and Zr between clinopyroxene Wadhwa M. (2015) The uranium isotopic composition of the and haplobasaltic melts – implications for uranium series Earth and the Solar System. Geochim. Cosmochim. Acta 148, disequilibria in basalts. Earth Planet Sci. Lett. 128, 407–423. 145–158. Mckay G., Le L., Wagstaff J. and Crozaz G. (1994) Experimental Gopel C., Birck J. L., Galy A., Barrat J. A. and Zanda B. (2015) partitioning of rare-earth elements and strontium – constraints Mn-Cr systematics in primitive meteorites: insights from on petrogenesis and redox conditions during crystallization of mineral separation and partial dissolution. Geochim. Cos- antarctic Angrite Lewis Cliff-86010. Geochim. Cosmochim. Acta mochim. Acta 156, 1–24. 58, 2911–2919. Gopel C., Manhes G. and Allegre C. J. (1994) U-Pb systematics of McKay G., Lindstrom D., Yang S. -R. and Wagstaff J. (1988) phosphates from equilibrated ordinary chondrites. Earth Planet Petrology of Unique Achondrite Lewis Cliff 86010, Lunar and Sci. Lett. 121, 153–171. Planetary Science Conference. Halse H. R. (2014) Using synchrotron Radiation to Determine the McKibbin S. J., Ireland T. R., Amelin Y. and Holden P. (2015) Oxidation State of Uranium in Magmas. Imperial College, Mn-Cr dating of Fe- and Ca-rich olivine from ’quenched’ and London. ’plutonic’ angrite meteorites using Secondary Ion Mass Spec- Irving A. J. and Kuehner S. M. (2007) Plutonic angrite NWA 4801 trometry. Geochim. Cosmochim. Acta 157, 13–27. and model for Angrite parent body consistent with petrological Mikouchi T., McKay G., Le L. and Mittlefehldt D. W. (2000) and chronological constraints, Workshop on Chronology of Preliminary Examination of Sahara 99555: Mineralogy and Meteorites and the Early Solar System, Hawai’i. Experimental Studies of a New Angrite, Lunar and Planetary Ito M. and Ganguly J. (2006) Diffusion kinetics of Cr in olivine and Science Conference, p. 1970. Mn-53-Cr-53 thermochronology of early solar system objects. Mikouchi T., Miyamoto M., McKay G. and Le L. (2001). Cooling Geochim. Cosmochim. Acta 70, 799–809. Rate Estimates of Quenched Angrites: Approaches by Jacobsen B., Yin Q. Z., Moynier F., Amelin Y., Krot A. N., Crystallization Experiments and Cooling Rate Calculations of Nagashima K., Hutcheon I. D. and Palme H. (2008) Al-26-Mg- Olivine Xenocrys, Annual Meteoritical Society Meeting, 26 and Pb-207-Pb-206 systematics of Allende CAIs: Canonical p. A134. solar initial Al-26/Al-27 ratio reinstated. Earth Planet Sci. Lett. Mittlefehldt D. W., Killgore M. and Lee M. T. (2002) Petrology 272, 353–364. and geochemistry of D’Orbigny, geochemistry of Sahara 99555, Jaffey A. H., Flynn K. F., Glendenin L. E., Bentley W. C. and and the origin of angrites. Meteorit. Planet. Sci. 37, 345–369. Essling A. M. (1971). Precision Measurement of Half-Lives and Mittlefehldt D. W. and Lindstrom M. M. (1990) Geochemistry and Specific Activities of U-235 and U-238. Phys Rev C 4, 1889-+. genesis of the angrites. Geochim. Cosmochim. Acta 54, 3209– Jurewicz A. J. G., Mittlefehldt D. W. and Jones J. H. (1993) 3218. Experimental partial melting of the allende (Cv) and murchison Moynier F., Yin Q. Z. and Jacobsen B. (2007) Dating the first stage (Cm) chondrites and the origin of asteroidal basalts. Geochim. of planet formation. Astrophys. J. Lett. 671, L181–L183. Cosmochim. Acta 57, 2123–2139. Muller T., Dohmen R., Becker H. W., ter Heege J. H. and Keil K. (2012) Angrites, a small but diverse suite of ancient, silica- Chakraborty S. (2013) Fe-Mg interdiffusion rates in clinopy- undersaturated volcanic-plutonic mafic meteorites, and the roxene: experimental data and implications for Fe-Mg history of their parent asteroid. Chem. Erde 72, 191–218. exchange geothermometers. Contrib. Mineral Petr. 166, 1563– Kleine T., Hans U., Irving A. J. and Bourdon B. (2012) 1576. Chronology of the angrite parent body and implications for Nyquist L. E., Kleine T., Shih C. Y. and Reese Y. D. (2009) The core formation in protoplanets. Geochim. Cosmochim. Acta 84, distribution of short-lived radioisotopes in the early solar 186–203. system and the chronology of asteroid accretion, differentiation, Kruijer T. S., Kleine T., Fischer-Godde M., Burkhardt C. and and secondary mineralization. Geochim. Cosmochim. Acta 73, Wieler R. (2014) Nucleosynthetic W isotope anomalies and the 5115–5136. Hf-W chronometry of Ca-Al-rich inclusions. Earth Planet Sci. Papanastassiou D. A. (1986) Chromium isotopic anomalies in the Lett. 403, 317–327. allende meteorite. Astrophys. J. 308, L27–L30. Larsen K. K., Trinquier A., Paton C., Schiller M., Wielandt D., Patterson C. (1956) Age of meteorites and the earth. Geochim. Ivanova M. A., Connelly J. N., Nordlund A., Krot A. N. and Cosmochim. Acta 10, 230–237. F.L.H. Tissot et al. / Geochimica et Cosmochimica Acta 213 (2017) 593–617 617

Patterson C., Tilton G. and Inghram M. (1955) Age of the Earth. Stirling C. H., Andersen M. B., Potter E. K. and Halliday A. N. Science 121, 69–75. (2007) Low-temperature isotopic fractionation of uranium. Posner E. S., Ganguly J. and Hervig R. (2016) Diffusion kinetics of Earth Planet Sci. Lett. 264, 208–225. Cr in spinel: Experimental studies and implications for Mn-53- Tang H., Liu M.-C., McKeegan K. D., Tissot F. L. H. and Cr-53 cosmochronology. Geochim. Cosmochim. Acta 175, 20–35. Dauphas N. (2017) In situ isotopic studies of the U-depleted Qin L. P., Alexander C. M. O., Carlson R. W., Horan M. F. and Allende CAI Curious Marie: Pre-accretionary alteration and Yokoyama T. (2010) Contributors to chromium isotope vari- the co-existence of 26Al and 36Cl in the early solar nebula. ation of meteorites. Geochim. Cosmochim. Acta 74, 1122–1145. Geochim. Cosmochim. Acta 207, 1–18. Quitte G., Markowski A., Latkoczy C., Gabriel A. and Pack A. Tang H. L. and Dauphas N. (2012) Abundance, distribution, and (2010) Iron-60 Heterogeneity and Incomplete Isotope Mixing in origin of Fe-60 in the solar protoplanetary disk. Earth Planet the Early Solar System. Astrophys. J. 720, 1215–1224. Sci. Lett. 359, 248–263. Riches A. J. V., Day J. M. D., Walker R. J., Simonetti A., Liu Y., Tang H. L. and Dauphas N. (2015) Low (Fe)-F-60 Abundance in Neal C. R. and Taylor L. A. (2012) Rhenium-osmium isotope Semarkona and Sahara 99555. Astrophys. J., 802. and highly-siderophile-element abundance systematics of Telus M., Dauphas N., Moynier F., Tissot F. L. H., Teng F. Z., angrite meteorites. Earth Planet Sci. Lett. 353, 208–218. Nabelek P. I., Craddock P. R. and Groat L. A. (2012) Iron, Richter S., Alonso-Munoz A., Eykens R., Jacobsson U., Kuehn zinc, magnesium and uranium isotopic fractionation during H., Verbruggen A., Aregbe Y., Wellum R. and Keegan E. continental crust differentiation: The tale from migmatites, (2008) The isotopic composition of natural uranium samples - granitoids, and pegmatites. Geochim. Cosmochim. Acta 97, 247– Measurements using the new n(U-233)/n(U-236) double spike 265. IRMM-3636. Int. J. Mass Spectrom. 269, 145–148. Teng F. Z., Dauphas N. and Watkins J. (2016). Non-Traditional Richter S., Eykens R., Kuhn H., Aregbe Y., Verbruggen A. and Stable Isotopes. Weyer S. (2010) New average values for the n(U-238)/n(U-235) Tissot F. L. H. and Dauphas N. (2015) Uranium isotopic isotope ratios of natural uranium standards. Int. J. Mass compositions of the crust and ocean: age corrections, U budget Spectrom. 295, 94–97. and global extent of modern anoxia. Geochim. Cosmochim. Acta Rocholl A. and Jochum K. P. (1993) Th, U and other trace- 167, 113–143. elements in carbonaceous chondrites - implications for the Tissot F. L. H., Dauphas N. and Grossman L. (2016) Origin of terrestrial and solar-system Th/U ratios. Earth Planet Sci. Lett. uranium isotope variations in early solar nebula condensates. 117, 265–278. Sci Adv. 2. Sanborn M. E. (2012) The Petrogenesis of Angrites and Martian Trinquier A., Birck J. L., Allegre C. J., Gopel C. and Ulfbeck D. Meteorites Inferred from Isotope and Trace Element Systemat- (2008) (53)Mn-(53)Cr systematics of the early Solar System ics. Arizona State University, p. 241. revisited. Geochim. Cosmochim. Acta 72, 5146–5163. Schauble E. A. (2007) Role of nuclear volume in driving equilib- Van Orman J. A., Cherniak D. J. and Kita N. T. (2014) rium stable isotope fractionation of mercury, thallium, and Magnesium diffusion in plagioclase: dependence on composi- other very heavy elements. Geochim. Cosmochim. Acta 71, tion, and implications for thermal resetting of the Al-26-Mg-26 2170–2189. early solar system chronometer. Earth Planet Sci. Lett. 385, 79– Schiller M., Baker J. A. and Bizzarro M. (2010) Al-26-Mg-26 88. dating of asteroidal magmatism in the young Solar System. Verbruggen A., Alonso A., Eykens R., Kehoe F., Kuhn H., Richter Geochim. Cosmochim. Acta 74, 4844–4864. S. and Aregbe Y. (2008). Preparation and certification of Schiller M., Connelly J. N., Glad A. C., Mikouchi T. and Bizzarro IRMM-3636, IRMM-3636a and IRMM-3636b. JRC Scientific M. (2015) Early accretion of protoplanets inferred from a and Technical Reports. reduced inner solar system Al-26 inventory. Earth Planet Sci. Wadhwa M., Kita N. T., Nakashima D., Bullock E. S., MacPher- Lett. 420, 45–54. son G. J. and Bouvier, A. (2014) High precision 26Al 26 Mg Schoene B. (2014) 4.10 - U-Th–Pb Geochronology. In Treatise on isotope systematics for an almost pristine refractory inclusion: Geochemistry, second ed. (ed. H. D. H. K. Turekian). Elsevier, Implications for the absolute age of the Solar System, Lunar Oxford, pp. 341–378. and Planetary Science Conference, Houston, TX, p. 2698. Schreiber H.-D. and Andrews S.-M. (1980) The redox states of Warren P. H. and Kallemeyn G. W. (1990) Geochemistry of the uranium in synthetic basaltic magmas. Lunar Planet. Sci. Conf., LEW87051 Angrite and Other Basaltic Achondrites, Lunar and 1000–1002. Planetary Science Conference. Shukolyukov A. and Lugmair G. W. (2006) Manganese-chromium Wasserburg G. J., Tera F., Papanastassiou D. A. and Huneke J. C. isotope systematics of carbonaceous chondrites. Earth Planet (1977) Isotopic and chemical investigations on Angra-Dos- Sci. Lett. 250, 200–213. Reis. Earth Planet Sci. Lett. 35, 294–316. Shukolyukov A., Lugmair G. W. and Irving A. J. (2009) Mn-Cr Weyer S., Anbar A. D., Gerdes A., Gordon G. W., Algeo T. J. and Isotope Systematics of Angrite Northwest Africa 4801, Lunar Boyle E. A. (2008) Natural fractionation of U-238/U-235. and Planetary Science Conference, Houston, TX, p. 1381. Geochim. Cosmochim. Acta 72, 345–359. Spivak-Birndorf L., Wadhwa M. and Janney P. (2009) Al-26-Mg- Yin Q. Z., Amelin Y. and Jacobsen B. (2009) Project Milestones: 26 systematics in D’Orbigny and Sahara 99555 angrites: Testing Consistent Chronologies Between Extinct 53Mn-53Cr Implications for high-resolution chronology using extinct and Extant U-Pb Systematics in the Early Solar System, Lunar chronometers. Geochim. Cosmochim. Acta 73, 5202–5211. and Planetary Science Conference, p. 2060. Spivak-Birndorf L., Wadhwa M. and Janney P. E. (2011). 60Fe- Zartman R., Jagoutz E. and Bowring, S. (2006) Pb-Pb dating of the 60Ni Chronology of the D’Orbigny Angrite: Implications for D’Orbigny and Asuka 881371 angrites and a second absolute the Initial Solar System Abundance of 60Fe, Lunar and time calibration of the Mn-Cr chronometer, Lunar and Planetary Science Conference, Houston, p. 2281. Planetary Science Conference, p. 1580. Steiger R. H. and Jager E. (1977) Subcommission on geochronol- ogy - convention on use of decay constants in geochronology Associate Editor: Yuri Amelin and cosmochronology. Earth Planet Sci. Lett. 36, 359–362.