The Vestan cataclysm: Impact-melt clasts in howardites and the bombardment history of 4 Vesta



This article is corrected by:

  1. Errata: Erratum Volume 48, Issue 8, 1529, Article first published online: 9 August 2013



Crystalline impact-melt samples were created in high-temperature environments by relatively large craters and, as such, give additional constraints on the nature of the impacts that created them. This article provides new 40Ar-39Ar ages of impact-melt clasts in howardites and shows that these clasts formed on the HED parent body, 4 Vesta, within the time period 3.3–3.8 Ga. Rather than resulting from an increased number of impacts, however, impact-melted material in howardites may result from unusually high-velocity impacts occurring in the asteroid belt during this period. This scenario is similar to the late heavy bombardment of the Moon, pointing to an unusual dynamical event at this time across the inner solar system. Therefore, impact-melt rocks in howardites uniquely record a Vestan cataclysm.


Our current understanding of the chronology of the inner solar system is based on the Moon's chronology, which suggests that intense bombardment happened early in solar system history, followed by a relatively quiescent period, and increased again in an extraordinary bombardment event (“cataclysm”) in the Earth–Moon system at approximately 3.9 Ga (see reviews in Ryder 1990; Stöffler et al. 2006; Chapman et al. 2007). Dynamical models have suggested that this intense bombardment could have been triggered by the migration of the giant planets, destabilizing Kuiper Belt objects and Main Belt asteroids to form the impactors (Gomes et al. 2005; Tsiganis et al. 2005; Bottke et al. 2012a). If these models are correct, this bombardment episode would have affected the entire inner solar system, and therefore the collisional history within the main asteroid belt should resemble that of the Earth–Moon system, although perhaps modified to reflect the lower interbody velocities within the Main Belt. For instance, ordinary chondrites show crystallization and early impact ages around 4.5 Ga followed by a period of increased bombardment intensity statistically peaking at approximately 3.8 Ga, coincident with the lunar cataclysmic event (Bogard 1995; Swindle et al. 2009; Wittmann et al. 2010).

The Dawn mission showed asteroid 4 Vesta to be an extensively cratered body, ancient in age, with craters in a variety of morphologies and preservation states (Jaumann et al. 2012; Marchi et al. 2012a; Schenk et al. 2012). Tying Vesta's relative crater ages to an absolute impact history can be accomplished through investigations of the HED (howardite, eucrite, diogenite) meteorites. Eucrites are crustal basalts and gabbros, diogenites are orthopyroxenites representing lower crust or upper mantle materials, and howardites are mixed breccias containing both lithologies. Eucrite 53Mn-53Cr systematics show that the HED parent body globally differentiated by approximately 4.56 Ga and fully crystallized soon afterward (Lugmair and Shukolyukov 1998). Much later, many eucrites were brecciated and heated by large impacts into the parent body surface. Disturbance ages in eucrites show that multiple large impacts occurred within 1 Gyr after crystallization, showing a history that largely resembles that of the Moon (see thorough reviews in Bogard and Garrison 1993; Kunz et al. 1995; Bogard 2011).

Dawn images also showed that Vesta is covered with a well-developed regolith that is spectrally similar to howardite meteorites (De Sanctis et al. 2012). Howardites are polymict regolith breccias made up mostly of clasts of eucrites and diogenites, but which also contain clasts formed by melting of the regolith by relatively large, energetic impact events. Impact-melt clast ages from howardites extend our knowledge of the impact history of Vesta, expanding on eucrite disturbance ages and helping give absolute age context to the observed crater-counts produced using Dawn data. This work investigates the composition and age of impact-melt clasts in howardites and examines the importance of impact melt within the record of Vesta's bombardment history.

Meteorite Samples

There are nearly 200 approved meteorites classified as howardites, approximately half from the Antarctic. Many are paired stones; pairing is most easily distinguished during examination in the same year, but may not always be obvious otherwise due to the similarity of the rocks and the paucity of solar-wind and cosmic-ray exposure ages on the large number of individual specimens. This study investigated four howardites in thin section (120 μm thick) from the Antarctic meteorite collection: Elephant Moraine (EET) 87513, Queen Alexandra Range (QUE) 94200, Grosvenor Mountains (GRO) 95574, and QUE 97001. Because all the meteorites appear roughly similar, several different image representations (backscattered electron, optical transmitted light, and element map combinations) are shown in the supplemental data, along with individual clast locations.

EET 87513 is one of a paired group of seven stones collected in the same season. All have a light gray matrix hosting a variety of angular clasts in a groundmass of comminuted pyroxene and plagioclase (Grossman 1994). A pairing relationship may exist between EET 87531, EET 87509 and EET 87513 (Bogard and Garrison 1993). Our section (EET 87513,131) was approximately 75 mm2.

QUE 94200 has a patchy fusion crust and a dull gray interior color with green, black, and white minerals visible. In thin section, it is a microbreccia of pyroxene and plagioclase clasts in a comminuted groundmass of these minerals. Mittlefehldt and Lindstrom (1997, 1998) examined four clasts in this meteorite that they interpreted as impact-melt clasts based on their siderophile-element contents and their bulk compositions similar to the howardite host stone, so it was a promising sample and indeed, the meteorite containing the most identifiable impact-melt clasts (n = 23). Section QUE 94200,25 covered 99 mm2.

GRO 95574 is part of a pairing group with several other howardites found in the 1995 season. All of these rocks have extensive smooth, black fusion crust coverage and very minor oxidation. The interior is gray in color, having a groundmass of comminuted pyroxene and plagioclase grains with a few larger mineral clasts and rare polymineralic clasts up to 2.5 mm across (Grossman 1998). Section GRO 95574,11 is about 68 mm2 and contained many lithic clasts, but few definite impact-melt clasts.

QUE 97001 has a thin black fusion crust covering 50% of the exterior surface and the exposed interior shows a concrete gray matrix with a variety of clasts. In thin section, the rock is a microbreccia of pyroxene with minor plagioclase and a few basaltic clasts (Grossman 1999). QUE 97001 was recovered from the same ice field as QUE 94200 so pairing is a possibility; however, QUE 97001,35 contained a significantly higher proportion of mineral fragments and fewer impact-melt clasts than QUE 94200,25.

Clast Composition and Origin

We characterized individual clasts within these meteorites in 100 μm thick, polished sections (Table 1). Bulk composition was obtained using the JEOL 8200 electron microprobe at the Institute of Meteoritics, University of New Mexico. Mineral compositions were obtained using a 20 nA focused beam and bulk analyses were performed using a 10 μm defocused beam in a grid pattern over the clast. The bulk composition of the clasts was computed using the mean and standard deviation of the analyses, reported in Table 1. The defocused beam analysis (DBA) technique introduces systematic errors when the analysis encompasses coexisting phases of different atomic masses, up to a half a wt% in major element oxides (Warren 1997; Carpenter et al. 2010). This systematic error will be similar among analyses if a similar proportion of minerals occur under the beam, which was not always the case in these analyses. However, the standard deviation among the individual analyses used to compute the bulk composition exceeds the DBA uncertainty (for example, Si varies by up to 1 wt% and Mg by 2–6 wt% within each clast); therefore, the DBA uncertainty was not further considered.

Table 1. Bulk composition of howardite clasts
  1. eu = eucritic; di = diogenitic; im = impact-melt; br = breccia; n.d. = not detected; n.a. = not analyzed.

  2. a

    Number of analyses in mean.

  3. b

    Molar Mg/(Mg + Fe).

GRO 95574,11A18eu4048.0 ± 1.870.30 ± 0.198.58 ± 13.10.21 ± 0.14n.d.24.6 ± 11.30.76 ± 0.359.29 ± 4.557.78 ± 4.680.23 ± 0.360.03 ± 0.02n.a.99.78
GRO 95574,11A1012eu6650.0 ± 2.980.33 ± 0.118.9 ± 2.850.89 ± 0.27n.d.15.6 ± 1.960.49 ± 0.0816.9 ± 5.866.33 ± 1.660.54 ± 0.150.15 ± 0.070.03 ± 0.01100.16
QUE 97001,36B0418im6549.8 ± 0.710.45 ± 0.1310.1 ± 3.701.10 ± 0.450.02 ± 0.0214.5 ± 3.150.47 ± 0.1314.9 ± 4.598.64 ± 4.180.20 ± 0.110.07 ± 0.030.02 ± 0.01100.27
QUE 97001,36B0510eu4548.9 ± 1.330.40 ± 0.3714.5 ± 12.40.56 ± 0.570.02 ± 0.0314.9 ± 10.60.48 ± 0.386.73 ± 4.7712.4 ± 4.210.89 ± 0.820.09 ± 0.05n.d.99.87
QUE 97001,36B0916im5650.2 ± 1.970.52 ± 0.189.91 ± 4.810.53 ± 0.09n.d.16.9 ± 4.890.55 ± 0.1811.8 ± 2.858.94 ± 3.640.72 ± 0.340.09 ± 0.04n.d.100.16
QUE 97001,36B1332im6451.3 ± 1.570.35 ± 0.064.71 ± 2.970.52 ± 0.14n.d.18.7 ± 1.670.62 ± 0.0618.8 ± 2.024.66 ± 1.470.20 ± 0.150.03 ± 0.01n.a.99.89
QUE 97001,36B1627im5348.3 ± 0.980.51 ± 0.1311.3 ± 3.630.64 ± 0.250.03 ± 0.0318.0 ± 3.630.53 ± 0.1311.4 ± 1.898.04 ± 1.540.87 ± 0.330.13 ± 0.04n.a.99.75
EET 87513,131C0213di7349.4 ± 5.340.09 ± 0.073.64 ± 4.991.16 ± 0.63n.d.17.0 ± 4.420.53 ± 0.1025.2 ± 6.602.77 ± 2.930.19 ± 0.290.02 ± 0.02n.d.100.00
EET 87513,131C079im6546.1 ± 2.530.44 ± 0.136.60 ± 1.000.83 ± 0.250.10 ± 0.1819.2 ± 1.940.52 ± 0.0219.5 ± 3.686.02 ± 1.630.60 ± 0.140.07 ± 0.020.06 ± 0.02100.04
EET 87513,131C146eu4547.7 ± 2.240.30 ± 0.2512.5 ± 17.50.24 ± 0.170.06 ± 0.0514.1 ± 11.00.44 ± 0.356.59 ± 5.1017.2 ± 1.660.44 ± 0.520.07 ± 0.03n.a.99.64
QUE 94200,25D0612eu4550.3 ± 6.690.40 ± 0.2115.5 ± 15.30.22 ± 0.150.03 ± 0.0414.7 ± 11.70.48 ± 0.426.65 ± 5.6811.6 ± 5.250.49 ± 0.470.06 ± 0.03n.a.100.43
QUE 94200,25D0819im6549.5 ± 3.840.28 ± 0.084.45 ± 1.431.14 ± 0.420.03 ± 0.0319.7 ± 4.210.57 ± 0.0720.7 ± 2.173.25 ± 1.440.11 ± 0.050.05 ± 0.02n.d.99.78
QUE 94200,25D1111im5457.2 ± 4.090.49 ± 0.2819.5 ± 1.450.10 ± 0.010.02 ± 0.015.58 ± 2.020.18 ± 0.073.73 ± 1.4612.8 ± 1.060.67 ± 0.070.08 ± 0.01n.a.100.35
QUE 94200,25D154im5948.0 ± 0.610.13 ± 0.0715.9 ± 3.240.54 ± 0.080.03 ± 0.0413.4 ± 2.700.42 ± 0.0910.7 ± 2.3410.0 ± 2.250.70 ± 0.210.07 ± 0.03n.d.99.89
QUE 94200,25D164im5949.5 ± 0.380.48 ± 0.0210.4 ± 0.310.55 ± 0.010.02 ± 0.0216.2 ± 0.580.52 ± 0.0213.2 ± 0.488.71 ± 0.300.45 ± 0.020.14 ± 0.01n.a.100.17
QUE 94200,25D175im6951.0 ± 1.070.33 ± 0.164.91 ± 2.480.67 ± 0.080.02 ± 0.0216.9 ± 1.690.50 ± 0.0720.9 ± 3.104.52 ± 2.630.04 ± 0.060.04 ± 0.02n.a.99.83
QUE 94200,25D1915im6749.0 ± 3.140.34 ± 0.214.99 ± 3.230.89 ± 0.24n.d.17.9 ± 4.250.52 ± 0.0620.6 ± 4.385.45 ± 4.070.07 ± 0.040.03 ± 0.020.02 ± 0.0299.81
QUE 94200,25D218im6348.7 ± 3.900.29 ± 0.066.94 ± 5.170.40 ± 0.130.02 ± 0.0319.6 ± 6.000.53 ± 0.1218.5 ± 3.925.32 ± 2.780.10 ± 0.170.05 ± 0.02n.a.100.45
QUE 94200,25D225im6850.7 ± 0.910.61 ± 0.129.17 ± 3.710.38 ± 0.080.02 ± 0.0212.5 ± 1.500.41 ± 0.0614.8 ± 2.2610.8 ± 1.510.48 ± 0.220.09 ± 0.04n.a.99.96

Howardites are polymict rocks composed of mineral grains and lithic clasts. Because they are derived from the regolith of the same body that produced several varieties of ferroan basalts (eucrites), as well as magnesian, cumulate orthopyroxenites (diogenites), howardites contain mostly clasts of these lithologies, along with impact-melt clasts derived from them, plus possible contributions from previously unsampled lithologies and foreign contributions. Impact-melt clasts can be distinguished from primary igneous rocks principally by their compositional makeup and characteristic elemental ratios, which are unable to be derived from normal igneous processes, used in tandem with their characteristic impact-derived textures. We used petrographic texture (Fig. 1), mineral composition (Figs. 2-4), and bulk composition (Table 1) to determine whether lithic clasts were derived from eucritic or diogenitic parent lithologies, or whether the clasts were likely to be impact-derived. The classes of clasts are described in detail, followed by a discussion of the overall compositional range within the howardites.

Figure 1.

Backscattered-electron microphotographs of howardite clasts showing the range of textures. a) A01, a eucritic clast with a subophitic texture consisting of blocky feldspar (gray) and pyroxene (white); b) A10, an olivine-phyric impact-melt clast containing zoned, euhedral olivine in a cryptocrystalline groundmass; c) B04, an impact-melt clast where homogeneous pyroxene lathes (bright) are set in a glassy feldspathic matrix; d) B16, a microporphyritic impact-melt clast with partially digested olivine and pyroxene grains; e) C04, a carbonaceous chondrite clast; f) C07, also a microporphyritic impact-melt clast, but with a higher proportion of relic clasts than in (d); g) D01, acicular pyroxene and plagioclase in an impact-melt clast; h) D08, an olivine-phyric impact-melt clast with large, euhedral olivine in a finely microporphyritic groundmass; i) D11, a symplectic impact-melt clast with a matrix of plagioclase and pyroxene containing partially digested, exsolved pyroxene grains.

Figure 2.

Composition of pyroxene in howardite clasts. Several clasts, particularly eucritic clasts, contain both high-Ca and low-Ca pyroxene (both points are plotted). Eucrite and diogenite pyroxene compositions are from Mittlefehldt et al. (1998).

Figure 3.

Composition of feldspar in howardite clasts. Polymict eucrites and impact-melt clasts can contain unusually high-K feldspar, possibly a remnant of more evolved magmatic products. Eucrite and diogenite feldspar compositions are from Mittlefehldt et al. (1998).

Figure 4.

Composition of olivine in impact-melt clasts. Olivine in polymict eucrites and impact-melt clasts can range to more iron-rich compositions, another signature of more evolved magmatic products. Eucrite and diogenite olivine compositions are from Mittlefehldt et al. (1998) and Shearer et al. (2010).

Eucrite Clasts (A01, A02, A03, A04, B05, B14, B15, C03, C14, D06)

Small clasts of eucrite-derived material are common in all the meteorites. At least one eucritic clast was studied in detail in each meteorite to provide ground-truth for composition and age comparisons. Most clasts consist of blocky plagioclase and pyroxene with a subophitic texture (Fig. 1a), although several have a poikilitic texture where pyroxene encloses blebs of plagioclase. In clasts of both textures, pyroxene grains contain well-developed exsolution lamellae of augite and low-Ca pigeonite, consistent with pyroxene in eucrites. The plagioclase in these clasts is generally homogeneous and An-rich. The clasts' bulk compositions in Fe and Mg plot in the eucrite field (Fig. 5), although they have wide error bars because of the way the bulk composition was achieved through mixed microprobe analyses, where large differences in FeO and MgO between the plagioclase and pyroxene phases causes a large standard deviation among microprobe points.

Figure 5.

MgO versus FeO distinguishes eucrites from diogenites and shows that many howardite clasts (from this study and from Mittlefehldt and Lindstrom 1997) are likely impact-melt rocks incorporating these two main lithologies. The “experimental melt” trend is from experimental melts of St. Severin (Jurewicz et al. 1995), showing the path that more evolved basalts would take from a eucritic source. None of the clasts definitively lie along this trend, further supporting their impact-melt origin. Eucrite and diogenite bulk compositions are from Warren et al. (2009).

Diogenite Clasts (C02)

Only one clast was observed that was obviously derived from a diogenite source. The clast is dominated by euhedral to subrounded mafic minerals with interstitial plagioclase and minor opaques. Pyroxene is unexsolved and has a low-Ca composition (Wo4En72Fs24); olivine grains have faint zoning from core to rim, but cores are uniformly Fo70. The clast's bulk composition lies squarely in the diogenite field in Fe-Mg space (Fig. 5).

Impact-Melt Clasts (A10, A04, B02, B03, B04, B09, B13, B16, C05, C07, C10, D01, D04, D08, D11, D15, D16, D17, D19, D21, D22)

A variety of impact-melt clast textures and compositions were identified, the majority occurring in QUE 97001 and QUE 94200. Most have a fine-grained, microporphyritic groundmass containing anhedral, relic mineral grains (Figs. 1c, 1f, and 1g). The groundmass usually is too fine-grained to directly analyze the plagioclase or mafic portions by electron microprobe. The relic clast population is dominated by plagioclase and orthopyroxene, with minor olivine, chromite, and other minor minerals. Pyroxene grains are typically homogeneous and unexsolved, although some exceptions exist. Plagioclase has a wide range of compositions between typical eucrites and diogenites (An80–95). These clasts have bulk compositions ranging in between the eucrites and diogenites, consistent with their interpretation as impact-melt rocks made up of varying contributions of these two lithologies. Several clasts have symplectic textures of pyroxene and plagioclase. In these clasts, plagioclase can be significantly K-rich.

Three clasts have a distinctive, olivine-phyric texture (A10, D08, D21). A10 (Fig. 1b) has many small (10 μm), euhedral olivine crystals (Fo82) in a glassy to cryptocrystalline feldspathic groundmass (An88Ab8Or4) containing crystallites. The olivine crystals are strongly zoned to Fe-rich compositions in thin rims. This clast also contains several irregular, glassy, more Fe-rich inclusions on which olivine has nucleated. The olivine is much more magnesian than typical diogenitic material. Accessory pyroxene is not equilibrated and spans a wide range of compositions. Clasts D08 and D21 (Fig. 1h) are also olivine-phyric, but are different from clast A10. These clasts have a fine-grained, microporphyritic groundmass that contains much larger, euhedral olivine grains (Fo57). The olivine is not zoned and has a significantly more Fe-rich composition than either clast A10 or diogenites. These clasts also contain minor pyroxene grains, which are unexsolved and have an unequilibrated Ca content (Wo6En62Fs31). The mixture of mineral characteristics, the presence of a glassy plagioclase groundmass, and anhedral included clasts make it probable that all these are impact-melt clasts. It has been suggested (Mittlefehldt and Lindstrom 1997) that clasts of this type may be a special product of early history of the body, with more magnesian diogenitic material making a significant contribution.

Other Clasts

Several clasts were clearly distinct from the main body of the breccia, but appeared to be microbreccias themselves, in which it is not clear whether the individual mineral grains are related to each other. In these clasts, pyroxene grains are homogeneous and unexsolved with either a low-Ca or high-Ca composition. Plagioclase in these clasts has a wide range of compositions (An88–98). We interpret these clasts to be unmelted impact breccias.

Several carbonaceous chondrite clasts were also noted in EET 87513 (Fig. 1e). These have both olivine and pyroxene with mg# = 95–97 and groundmass that gave low analysis totals, presumably due to the abundance of light elements and water. These clasts are similar to those noted by Zolensky et al. (1996) and Pun et al. (1998) and were not studied further.

Compositional Range within Howardites

The clast mineralogy and composition in these howardites are consistent with those of previous studies, particularly in the composition of pyroxene. A few occurrences of high-K feldspar (An80Ab16-17Or 3–4) were found, chiefly in symplectites. K-rich feldspar may be derived from melting of a K-rich mesostasis from evolved basaltic products or incorporation of an exogeneous contribution into an impact-melt clast. Interestingly, olivine grains in the QUE 94200 clasts are consistently more iron-rich (mg# 56–61) than typical diogenites, and several impact-melt clasts have unusually high Fe/Mn ratios compared with cumulate eucrites and diogenites. Clasts C07 and D21 have bulk Fe/Mn ratios of approximately 36, which is higher than the typical ratios of approximately 28–30 in eucrites and diogenites (Mittlefehldt et al. 1998). The Stannern-group eucrites are also marked by high Fe/Mn (33–40; Barrat et al. 2007). It may be that these materials are present in the regolith in small proportions, contributing to the impact-melt target material.

Although the impact-melt clasts in this study appear to be mixtures of eucrites and diogenites, more magnesian primary melts are postulated to have formed on the HED parent body (Stolper 1977), but have not been found in the meteorite collection. Rather than appearing as large clasts, fragments of these basalts may be represented by small pieces, each having an incomplete mineral assemblage, such as has been shown in the polymict ureilites (Cohen et al. 2004; Downes et al. 2008). Characteristics of more evolved magmatic products on the HED parent body may be partially informed by melting experiments on LL chondrites (Jurewicz et al. 1995; Mittlefehldt and Lindstrom 1997). Based on these experiments, primary melts can be distinguished from impact melts of howarditic material principally in the much higher FeO content expected for the primary melts (Fig. 5). The fact that no clasts follow the trend of increasing FeO content, but rather fall in a mixing range between eucrites and diogenites, argues for their origin as impact mixtures of these primary lithologies, a conclusion also drawn by Mittlefehldt and Lindstrom (1997, 1998).

Clast Ages

Clasts were extracted from the howardite thick sections using a Medenbach microcorer. All the samples were irradiated for 350.62 h in the TRIGA reactor at the University of Oregon along with single crystals of PP-20 (Hb3gr) hornblende as the flux monitor, producing a J-factor of 8.87E-2 (±0.3%). Each sample was step-heated at the New Mexico Geochronology Laboratory (New Mexico Bureau of Mines at the New Mexico Institute of Mining and Technology) using a CO2 laser that dwelt on the sample for 30 s per step. Data were corrected for system blanks, decay time, and reactor-induced interferences enabled by simultaneous irradiation of Ca and K salts. The reactor-derived 36Ar/37Ar from Ca is 2.69E-4, which accounted for ≤30% of 36Ar released in the samples. K decay parameters were those recommended by Steiger and Jäger (1977); if using the parameters recommended by Renne et al. (2010) (which are not yet agreed upon by the community, see discussion in Bogard 2011; Schwarz et al. 2011), our reported ages would be systematically too young by approximately 8 Ma. This is well within the uncertainties of our reported data.

Resolution of the relative contributions to nonradiogenic 40Ar correlated with 36Ar from cosmogenic and trapped components was attempted by constructing a “cosmochron” (a plot of 38Ar/36Ar versus 37Ar/36Ar) in the style of Levine et al. (2007), and following Garrison et al. (2000) and Swindle et al. (2009) to assume the minimum released 36Ar/37Ar represented gas release with no trapped Ar, only cosmogenically derived Ar. Samples B16 and C14 had the most gas-release data points so were used as examples. Indeed, after using the degassing step with minimum 36Ar/37Ar to subtract a cosmogenic contribution to 36Ar, these samples showed a better agreement between plateaus and isochrons, with a difference of about 1–5% from the uncorrected to the corrected age. However, for both samples, the 38Ar/36Ar ratio for several degassing steps was higher than the spallation ratio (38Ar/36Ar = 1.5), implying that these samples contain excess 36Ar from a third source, probably derived from reactions on chlorine. In addition, the limited number of heating steps in other clasts lessens confidence that the observed minimum 36Ar/37Ar ratio is a good measure of the cosmogenically derived component in each sample. The minimum 36Ar/37Ar ratio observed in each meteorite as a whole could only be used as a proxy for all clasts in that meteorite if all components had identical cosmic-ray exposure histories, which is a poor assumption.

Bogard and Garrison (2003) actually argue against using 36Ar to correct the 40Ar abundance in their work on eucrites, because in their experience, cosmogenic 36Ar is not generally well-correlated with any radiogenic component or with any trapped Ar, excess 40Ar, or reactor-recoiled 39Ar. The slope of an isochron plot for eucrites, from which the age is derived, is primarily determined by the larger 40Ar/36Ar and 39Ar/36Ar ratios arising from degassing of feldspar at intermediate temperatures. In contrast, the 40Ar/36Ar intercept value of the isochron is primarily determined by high-temperature extractions degassing pyroxene, for which larger releases of cosmogenic 36Ar produce lower 40Ar/36Ar and 39Ar/36Ar ratios. We found this same behavior in our samples where enough heating steps enabled construction of an isochron.

We therefore adopt the approach of Bogard and Garrison (2003) and do not pursue further corrections to 36Ar or 40Ar after corrections for the routine system blanks, decay time, and reactor-induced interferences. Because samples may contain small amounts of nonradiogenic 40Ar, all ages should be considered upper limits. In the cases where sample data did define traditional and/or inverse isochrons (e.g., Jourdan et al. 2011), we report those data for completeness, with the caveat that they may be slightly rotated because of the effects described above. All sample data are given in the supporting information; those with greater than two release steps are shown on traditional plateau diagrams with 2σ uncertainties (Fig. 6) and reported in Table 2, with plateaus defined in the following way: “good” means a plateau is formed in the conventional definition of contiguous steps within 2σ of each other releasing 50% or more of the total gas; “fair” plateaus have steps within 2σ but are not necessarily contiguous; and “poor” plateaus include steps defined as colinear on isochrons but not necessarily in plateau plots.

Table 2. 40Ar-39Ar ages of howardite clasts
MeteoriteClastTypeMicrocore mass (μg)Plateau age ± 2σ (Ma)%39Ar in plateauPlateau qualityaIsochron age ± 2σ (Ma)(40/36)i
  1. eu = eucritic; di = diogenitic; im = impact-melt; br = breccia.

  2. a

    See text for discussion of plateau quality.

QUE 97001,36B05eu1003650 ± 6073Fair3520 ± 12050 ± 50
QUE 97001,36B11br403210 ± 8087Fair3350 ± 6019 ± 7
QUE 97001,36B14eu603320 ± 10070Poor2840 ± 15055 ± 14
QUE 97001,36B16im903550 ± 6072Poor3560 ± 1300 ± 30
EET 87513,131C14eu1703710 ± 3075Good3810 ± 6010 ± 12
EET 87513,131C15im203730 ± 30100Good3630 ± 130100 ± 190
EET 87513,131C17im303740 ± 50100Fair3960 ± 5063 ± 14
QUE 94200,25D08im1203730 ± 3075Good3730 ± 300 ± 20
QUE 94200,25D11im403370 ± 4080Fair2600 ± 60070 ± 70
QUE 94200,25D17im104040 ± 100100Poor4600 ± 200220 ± 100
QUE 94200,25D19im903570 ± 8061Poor3600 ± 1200 ± 15
QUE 94200,25D23im1003640 ± 1249Good3600 ± 30070 ± 150
Figure 6.

Step-release profiles for howardite clasts. The Ca/K ratio and best-fit “plateau” is shown for each sample. The criteria for choosing the “plateau” steps are discussed in the text and Supporting Information.

Samples C14, C15, D08, and D23 (good) have readily apparent plateaus, where >50% of the 39Ar in the sample is released and the steps yield a well-defined age. The D23 gas-release profile suggests two plateaus which differ by 600 Ma; however, isochron representations corroborate the higher temperature (younger) age. In the other samples, the first gas-release step has an anomalously low apparent age. This effect was seen in lunar samples prepared in the same manner (Cohen et al. 2005) and probably corresponds to release of terrestrial air trapped by clinging epoxy, which readily degasses at low temperature.

Samples B05, B11, C17, and D11 (fair) have reasonable plateaus, some after the first degassing step (again, usually having an apparently low age). The sloped age plateaus in samples B5 and C14 strongly suggest diffusive gas loss, so both the plateau age and the isochron ages (with considerable scatter outside individual errors) could well be lower limits. Both the degassing spectrum and isochron plots for sample D11 yield a young apparent age, but the scatter within the sample and the high trapped gas content of the final high-T degassing step may imply that the unusually young apparent age may be unresolved within the limits of the sample data (however, as we shall see later, even this young age falls within the range of all impact-affected HED samples).

Samples B14, B16, D17, and D17 (fair) do not have obvious plateaus in their step-release patterns. In these cases, corroborating evidence from the sample isochrons was brought to bear in choosing steps to add together to form a sample age. The steps included in the “plateaus” scatter around a common line in both the normal and inverse isochrons. In most cases, the isochron-derived age is in reasonable agreement with the plateau-derived age. However, the scatter and unusual shape of these age spectra weaken confidence that their age and true 2σ uncertainty have been constrained. Accordingly, these ages are reported only for completeness and are not considered in the discussion, where the significance of sample ages is assessed. Furthermore, none of the samples from GRO 95574 yielded enough gas to construct a plateau or isochron, so are shown only in the Supporting Information and not further considered in the figures or discussion.

It is worth noting here that assigning a reliable age uncertainty to data sets that include very few extractions, or have a sloped age spectrum suggesting Ar loss, or show considerable scatter in the isochron data, is unlikely to be fully possible. In the samples analyzed here, the reported 2σ uncertainty is derived by adding adjacent heating steps and propagating the uncertainty in the isotopic ratios of each step. This may be mathematically correct, but does not fully account for the scatter among the steps or any trends in that scatter. Therefore, reported uncertainties probably err on the side of being too small in many, if not most, of the samples reported here. However, the absolute magnitude of the uncertainty in each age does not preclude discussion based on the sample ages themselves.


The age distribution among the samples analyzed in this work has a fairly restricted range, between about 3.8 and 3.3 Ga (Fig. 7). The only older sample, D17, has a poor confidence in age and a fairly large uncertainty associated with it. One effect that has been observed in lunar rocks is a possible correlation between older ages and higher amounts of potassium, but that effect is not apparent in our sample set (Fig. 8). Not enough gas was released from these samples to construct an exposure age history, but the generally little trapped 36Ar and 38Ar implies that these clasts did not experience extensive exposure to the solar wind at their parent body surface or to cosmic rays during their transit to Earth.

Figure 7.

a) Histogram of impact-melt sample ages in this study, including all good, fair, and poor plateau ages; b) ideogram including only good (solid gray) and fair (dashed gray) plateau ages, where each sample is represented by a Gaussian curve centered at the sample age, having a width proportional to the sample age uncertainty (2σ), and unit area under the curve, and the sum is represented by the thick black line.

Figure 8.

Sample age versus K2O (wt%) of each sample showing no increase in apparent age with increasing potassium content.

Our new howardite impact-melt ages fall well within the age distribution of all HED impact-reset rocks (Fig. 9), which features a short, intense spike at 4.48 Ga followed by a period of relative quiescence, then a ramping up of impact-reset ages between about 4.1 and 3.4 Ga. Bogard and Garrison (2003) suggested that the earliest ages, largely recorded in cumulate or unbrecciated eucrites, imply excavation of these rocks from depth in a single large impact event approximately 4.48 Gyr ago, which quickly cooled the samples and started the K-Ar chronometer. This event may also have formed a secondary parent body, protecting the cumulate eucrites from further processing and resetting in the Vesta regolith.

Figure 9.

Ideogram of all impact-affected sample ages in eucrites (unbrecciated, brecciated, polymict) and howardites (compiled by Bogard and Garrison 2003 and Bogard 1995; and in Bogard and Garrison 2009; Kunz et al. 1995; Sisodia et al. 2001). The gray line shows the large number of impact-affected samples around 4.48 Ga, a gap between then and around 4.0–4.1 Ga, and an increase in the number of affected samples between then and about 3.4 Ga. The black line includes only ages from impact-melt samples within howardites (Table 3), demonstrating that these products appear to have been created specifically between 4.1 and 3.6 Ga.

After this time, ages in the HED meteorites are consistent with a quiescent period in the Main Belt, followed by increased activity in the period of the late heavy bombardment, between about 4.0 and 3.5 Ga. All of the howardites appear to fall within this age range, as do most brecciated and polymict eucrites (Fig. 9). So late in the history of the solar system, impact is the only plausible mechanism for resetting the ages of these materials. However, this impact history seems at odds with the observed crater distribution on Vesta, which appears to be continuously cratered throughout its history over a range of diameters, rather than in only two distinct episodes (Marchi et al. 2012a).

The lack of young ages may be attributable to the formation of the Vesta-family asteroids, generally considered intermediary parent bodies between Vesta itself and the HED meteoroids. Among the HED meteorites, the howardites have the longest cosmic-ray exposure ages, ranging up to 50 Ma (Welten et al. 1997). Two distinct exposure age clusters appear in the data at around 20–25 Ma and 35–42 Ma, suggesting events that delivered howardites to Earth-crossing orbits, without significantly degassing the whole rocks or the clasts within them. For this to be consistent with delivery from much smaller bodies, these bodies must have been dislodged from Vesta well prior to this time. Furthermore, if the impact-melt clasts formed on Vesta itself, then the youngest clasts provide an upper bound on the time these intermediary parent bodies formed. Estimates of the Vesta family dynamical age (Marzari et al. 1996; Nesvorný et al. 2008) place this event at about 1 Ga, and Dawn results show that the giant Rheasilvia basin, 500 km in diameter, with a crater-counting age of about 1 Ga, may have been the source (Marchi et al. 2012a).

An indication that impact-melt clasts themselves must have formed in impacts directly on the Vesta parent body is that the body must have been large enough to sustain a large impact without being disrupted, and also able to retain impact-reset materials in its surface regolith. This being the case, a continually cratered Vesta should result in a continuous age distribution of impact-reset rocks, which is not what is observed. Therefore, the age distribution within HED meteorites prior to the Vesta-family-forming event appears to only incompletely record the impact history of Vesta, which may be due to the difficulty of resetting rock ages via collisions in the asteroid belt.

Reset of the K-Ar system is a diffusion-controlled process, where a combination of temperature and time is required to lose 40Ar from the crystal lattice sites originally occupied by K cations. Eucrites (and howardites) contain K (and therefore Ar) primarily in pyroxene, plagioclase, and mesostasis. Recent quantitative work on diffusion coefficients provides closure temperatures and diffusion rates in minerals with similar, although not identical, compositions as the howardites (Cassata et al. 2009, 2011; Weirich et al. 2012). Here, I will use these data to illustrate the conditions required to reset the age of these rocks, keeping in mind diffusion from whole-rock meteorites may be different depending on their exact mineralogy and shock state (e.g., Bogard and Hirsch 1980; Bogard 2011). The frequency factor (D0/a2) and activation energy (E) reported for each mineral can be used to determine the diffusivity (D/a2) of Ar across the grain sizes used in the experiments (a = 0.1–10 mm), which is similar to the grain size in eucrites (Mittlefehldt et al. 1998). Diffusive loss can then be calculated as a function of time (t) by setting D/a2 × t appropriately for the observed fractional loss (McDougall and Harrison 1999). Figure 10 shows the time required to diffusively lose 90% of the initial 40Ar through each mineral (D/a2 × t = 0.5, assuming spherical, cylindrical, or cubic geometry). At elevated temperatures over timescales of 0.1–10 yr (such as might occur during formation and cooling of an impact ejecta blanket), significant loss of Ar can occur in feldspar, but pyroxene requires a much longer time to degas, even on the scale of a single grain. This is merely meant to be illustrative; the impact-melt clasts may have experienced multiple episodes of heating, first as an impact melt with rapid cooling, and then within impact ejecta with lower temperatures and slower cooling, leading to a more complex Ar loss profile (suggested by the partial diffusive loss profiles in some samples).

Figure 10.

Time needed as a function of temperature for 90% diffusive loss for argon from feldspar and pyroxene grains (0.1–10 mm grain size). Short of completely melting, residence time in a warm impact ejecta blanket is adequate to reset the 40Ar-39Ar system in feldspar, but unlikely to fully reset pyroxene.

Temperatures in the range that enable significant diffusive loss are relatively high, even up to the melting point, although shocking the mineral before heating it, as might be the case in an impact event, promotes diffusion (Weirich et al. 2012). Whether temperatures near the melting point are achieved in the target rock is a strong function of the impact velocity (Melosh 1989; Pierazzo et al. 1997; Barr and Citron 2011). Whereas the average impact velocity on the Moon is around 11 km s−1, the typical impact velocity between objects in the Main Belt is about 5 km s−1 (Farinella and Davis 1992), which imparts too little energy to raise the temperature of the target material above the closure temperature of typical eucrite minerals (300–400 °C for plagioclase, 600–700 °C for pyroxene) (Marchi et al., personal communication.

The formation of impact-melt clasts, therefore, gives additional constraints on the impact history of Vesta. The samples from this study and others that have characteristics of impact-melted materials (Table 3) have distinct age and compositional characteristics that suggest that they formed in discrete impact events. The range of these events is much narrower than that for all HEDs, between 3.3 and 3.8 Ga (Fig. 9). To create these crystalline impact-melt products on the surface of Vesta, the impacts during this time period must have had velocities much higher than 5 km s−1, the Main Belt average (Marchi et al., personal communication). Only a small fraction of impactors achieve high velocity in a normal distribution, so if impact-melt products were produced by this high-velocity tail of normal collisions in the Main Belt, a small number of impact-melt products would appear throughout Vesta's history. However, impact-melt sample ages do not span the entire time of Vesta's residence in the Main Belt, rather, there are several impact-melt signatures spaced rather closely in time comparatively late in solar system history. This is inconsistent with formation by a normal distribution of impact velocities and points instead to a unique period where high-velocity collisions were more frequent than currently observed.

Table 3. Impact-melt clasts in howardites
MeteoriteClast description40Ar-39Ar age (Ga)Reference
  1. a

    Described in Buchanan and Reid (1991).

  2. b

    Including only “fair” and “good” plateau ages as discussed in text.

BununuGlass intermediate between eucrite and whole-rock4.17 ± 0.05Rajan et al. (1975)
EET 87509,24Clast Qa, vitrophere4.07 ± 0.02Bogard and Garrison (1992)
EET 87509,24Quench-textured clast4.05 ± 0.02Bogard and Garrison (2003)
EET 87509,71Quench-textured clast4.0 ± 0.10Bogard and Garrison (2003)
EET 87509,74Clast Ea, unequilibrated pyroxene and feldspar3.93 ± 0.05Bogard and Garrison (1992)
EET 87509,74Quench-textured clast3.90 ± 0.01Bogard and Garrison (2003)
EET 87513,131Impact-melt clast3.71 ± 0.03This workb
EET 87513,131Impact-melt clast3.73 ± 0.03This work
EET 87513,131Impact-melt clast3.74 ± 0.05This work
EET 87531,21Clast Ja, unequilibrated pyroxene with An-rich feldspar3.83 ± 0.05Bogard and Garrison (1992)
EET 87531,21Moderately recrystallized with inhomogeneous pyroxenes3.81 ± 0.05Bogard and Garrison (2003)
MalvernGlassy clast with abundant angular mineral fragments3.70 ± 0.15Kirsten and Horn (1977)
MalvernGlass intermediate between eucrite and whole-rock3.68 ± 0.05Rajan et al. (1975)
MalvernGlass intermediate between eucrite and whole-rock3.75 ± 0.04Rajan et al. (1975)
QUE 94200,25Impact-melt clast3.37 ± 0.03This work
QUE 94200,25Impact-melt clast3.64 ± 0.01This work
QUE 94200,25Impact-melt clast3.73 ± 0.03This work
QUE 94200,13Clast intermediate between eucrite and whole-rock3.71 ± 0.01Bogard and Garrison (2003)
QUE 97001,36Impact breccia clast3.21 ± 0.08This work
QUE 97001,36Impact-melt clast3.65 ± 0.06This work

Until now, impact-reset ages in the HED meteorites have been interpreted under the umbrella of the canonical lunar cataclysm where an increase in the absolute number of bombarding objects is responsible for creating larger absolute amounts of impact-affected and impact-melted rocks, statistically increasing their chances of being found on Earth and dated. However, even in this scenario, something must have changed in the solar system to abruptly increase the number of collisions within the asteroid belt and therefore on Vesta. Therefore, it is probable that the howardite impact-melt clasts, and probably most of the impacts in this period, are the result not necessarily of an increased number of impacts, but rather result from impacts of higher velocity. Although the new ages presented here are not determined precisely enough to define the temporal limits of the period of unusual bombardment period or of the times of specific events within it, further work is underway to quantify the evolution of this high-velocity population and its heating effects on Vesta and other parent bodies in the Main Belt (Bottke et al. 2012b; Marchi et al., personal communication).

The changeover from a typical Main Belt velocity profile to this regime of increased velocity population at Vesta occurs contemporaneously with a similar transition at the Moon, which Marchi et al. (2012b) used to argue for the onset of the lunar cataclysm, or late heavy bombardment. The source of this change may have been outside the Main Belt, as in the cometary flux of the Nice model (Gomes et al. 2005) or perhaps from the Main Belt itself (Bottke et al. 2012a). Either way, howardite impact-melt clast ages reinforce the notion of a dynamically unusual episode of bombardment in the inner solar system beginning at around 4.0 Ga.


This manuscript benefited from discussions with Tim Swindle and Simone Marchi, and was strengthened by reviews by Don Bogard and Brad Jolliff. I thank the Meteorite Working Group for their wisdom in allocating the howardite samples despite my request for a lunar breccia. This work was supported by the NASA Cosmochemistry program via grant NNX07AI57G to B. A. Cohen and used the NASA Astrophysical Data System Abstract Service.

Editorial Handling

Dr. Carle Pieters