standby for kenna photo
Found February 1972
33° 54' N., 103° 33.2' W.

A single stone of 10.9 kg was found in Roosevelt County, New Mexico by Ivan "Skip" Wilson on his ranch. It is composed mainly of olivine and the calcium-poor pyroxene, pigeonite, in a black carbon-rich matrix containing graphite, lonsdaleite, and diamond. Melt pockets and secondary veins contain grains of Ca-rich augite, andesine, K-feldspar, diopside, and chromite.

An initial classification scheme was proposed by Berkley et al. (1980) in which the main group ureilites were subdivided into three distinct compositional subgroups based on the FeO/MgO ratio in olivine, expressed as mol% Fo (=Mg#):

  1. high FeO/MgO; <80 mol% Fo (e.g., Kenna [Fo~79], Novo Urei, Goalpara, Haverö)
  2. intermediate FeO/MgO; 80-90 mol% Fo (e.g., Dingo Pup Donga, Hajmah (a), Dyalpur)
  3. low FeO/MgO; >90 mol% Fo (e.g., Y-74659, LEW 85440, Almahata Sitta #051)

Subsequent studies of a larger sample set indicates that these FeO/MgO ratios might actually represent a continuum. More recently, Goodrich et al. (2002, 2004) have proposed a model to describe the petrogenesis of this meteorite group; some details from their model are included here. Their model recognizes three mineralogical types of monomict/unbrecciated ureilites:

  1. The most numerous type, comprising ~90% of all ureilites, has an olivine–pigeonite composition. These ureilites are augite-depleted, partial melt residues produced through low degrees (<15–30%) of fractional melting at temperatures reaching ~1250°C and occurring over a range of depths, while undergoing various degrees of reduction through smelting. Both the basaltic melts and the molten metal were extracted through veins and dikes, likely propelled off of the planetesimal by the pressure of smelting-derived CO+CO2 gas. Alternatively, an oblique collision with a smaller asteroid could have resulted in the removal of the basaltic mantle, with the mixing of olivines in the regolith occurring through subsequent impact gardening (Downes et al., 2008). This ureilite type has FeO/MgO ratios in the range of Fo~76–87, and it features Δ17O values of –0.5 to –1.0.
  2. Of the remaining ~10% of ureilites, an olivine–orthopyroxene composition has been found in a few samples (e.g., LEW 85440 and pairings, Y-74659, Y-791538, EET 87517, MET 01085, and EET 96262 and pairings). These ureilites are residues formed from rare, fractionated, late-stage melt pockets in close association with olivine–pigeonite ureilitic residues at temperatures reaching ~1275°C. Smelting/reduction likely occurred both at the source and at lower pressures as the melt ascended through shallower depths; some pigeonite may still remain. This ureilite type is highly magnesian with FeO/MgO ratios in the range of Fo~86–92, and it features Δ17O values of –2.0 to –2.5.
  3. The remainder of the monomict/unbrecciated ureilites have an olivine–augite (augite-bearing) composition, and only a small number of examples have been described thus far (e.g., ALH 82130 and pairings, EET 96293 and pairings, FRO 90054 and pairings, HaH 064, Hughes 009, LEW 88774, META78008, and Y-74130. In this ureilite type, pigeonite has been replaced by augite (~5–30%), while orthopyroxene also is typically present (~29–56%), usually having poikilitic textures. A poikilitic or bimodal texture is present in many of these ureilites, exhibiting large orthopyroxene oikocrysts that formed from a late melt phase enclosing olivine–augite textural zones; however, a few have a typical ureilite texture. Members of this type are probably cumulates or paracumulates that formed from melts originating at great depths, crystallizing after ascent to shallower depths and experiencing some degree of smelting/reduction. They span a broad range of ureilite FeO/MgO ratios of Fo76–95.

Paradoxically, the pressure-controlled smelting model shows that ureilites in our collections are biased in sampling only a thin layer at intermediate depth within the pre-impact parent body, which exhibits a spike at Fo76–81 (Warren, 2012). A model of gasless anatexis (magmatism) is consistent with high internal pressures, suggesting a relatively large pre-impact parent body of at least 400 km in diameter. Efforts continue in ernest to constrain the numerous parameters of the ureilite petrogenetic history by employing numerical modeling (e.g., Michel et al., 2013 #1300) including the following:

Kenna is a member of Berkley's least reduced subgroup I, being FeO-rich with a high fayalite content (Fo~79). It is proposed that members of this subgroup formed early at the deepest layers, and are associated with the highest C content. It also belongs to Goodrich's subgroup 1, having a composition consisting primarily of olivine and pigeonite. Goodrich et al. (2002; 2004) described a stratification of the ureilite precursor material consistent with the existence of both magnesian (shallower) and ferroan (deeper) ureilites. This stratification is thought to have been originally established through pre-igneous aqueous flows allowing for the heterogeneous distribution of unfractionated O-isotopes, as well as the concentration of other mineral elements such as Ca.

The precursor material of ureilites had a bulk composition similar to that of carbonaceous chondrites, including O-isotopic compositions having strongly negative Δ17O plotting along the carbonaceous chondrite anhydrous mineral (CCAM) trend line. In addition, it contains a relatively high C content (6–7 wt%). In particular, characteristics very similar to precursor ureilite material are found among the CV-group chondrites, such as bulk FeO/MgO ratios equivilant to Fo~62. However, noteworthy differences exist between the ureilite precursor material and CV-like material; e.g., the former has a higher, superchondritic Ca/Al ratio (~2.5 × CI) compared to CV-like material, and depending on the model used, the ureilite precursor material may have had an increased Si/Mg ratio comparable to that found in ordinary chondrites as well as a very low alkali content (Goodrich et al., 2007). Modelling to determine the precursor end-member components from which the ureilite parent body was formed was pursued by Rai et al. (2015). They searched for combinations of components which when combined would exhibit the elemental ratios and oxygen isotope signature of known ureilites. They found that a combination of three end-member components comprising Fe-rich and Fe-poor chondrules plus either CI-type or CM-type chondritic material best matched the range of values found in ureilites.

Carbonaceous chondrites are too FeO-rich and SiO2-poor to represent ureilite precursor material, and ureilite feldspathic glass is not a consistent product of such an origin (Warren, 2011). Moreover, ε48Ca is significantly different between ureilites and carbonaceous chondrites (Chen et al., 2011). In the same way, a plot of ε50Ti vs. ε54Cr and Δ17O vs. ε54Cr finds ureilites actually cluster far from, and in an orthogonal direction from, the carbonaceous chondrite trend. Moreover, ureilites have lower ε62Ni and far lower ε50Ti and ε54Cr than any known carbonaceous chondrite. Based on isotopic and compositional parameters, it was concluded that an extension of the carbonaceous chondrite trend to incorporate the ureilites was not supported, and a predominantly noncarbonaceous chondritic precursor for ureilites is recognized as more reasonable (Warren, 2011). Based on REE correlation trends and their abundances in ureilite augites, Huang et al. (2009) concluded that ureilites were formed through partial melting of precursor material that is consistent with depleted spinel peridotite with a 5% clinopyroxene component.

In calculating the required Ca/Al ratio from which a precursor magma would produce pigeonite-type residues, it was determined that this ratio was substantially higher than chondritic levels (2.5–3 × CI), and therefore, a parent magma depleted in plagioclase would be a requirement; e.g., segregation during melting and loss through explosive volcanism. Alternatively, the Ca could have been concentrated during aqueous alteration and dehydration phases to form heterogeneous regions. Still, another process described as fractional smelting has been invoked by Singletary and Grove (2006) to account for the high Ca/Al ratios in ureilites (ave. 4.2 × CI, but as high as 14.5 × CI). They suggest that an olivine+melt+carbon+metal source underwent decompression smelting, perhaps promoted by cracks, which lead to an enrichment in Ca compared to Al as temperatures cooled below ~1240°C. At the same time, this process resulted in an enrichment of chromite along with high Mg# silicate residues.

An alternative model for the existence of superchondritic Ca/Al ratios in ureilites was proposed by Goodrich et al. (2007). They determined that pigeonite-bearing residues with the highest measured Mg# could be produced with a Ca/Al ratio of ~2.5 × CI. They proposed a disequilibrium model in which melt was rapidly extracted through a complex vein and dike network faster than it could be replenished, on a time scale of weeks to a year—a process they refer to as fractional melt extraction. This rapid-to-immediate extraction process only allows for a very low diffusivity of O-isotopes, Ca, REE, and highly siderophile trace elements from the silicate residue to the melt phase, resulting in the preservation of heterogeneity and certain chondritic ratios within the thermally processed ureilites—features which had been historically considered paradoxical (Wilson and Goodrich, 2012). This model reasonably addresses the concerns propounded by Mittlefehldt et al. (2005) as to why there was an increase in the abundance of siderophile elements like Ni and Ir among the most magnesian ureilites at the same time that the smelting-produced Fe-metal host phase was removed.

In their study of Re–Os isotope systematics of ureilites, Rankenburg et al. (2007) showed that a similar Os-isotopic distribution exists between ureilites and carbonaceous chondrites, particularly the CM, CV, CO, and CR groups, consistent with their formation in a common nebular region. The constant Os-isotopic ratios in ureilites demonstrate that there was a lack of fractionation of Re from Os on the ureilite PB, similar to what is indicated in carbonaceous chondrites. They also found that a metal component must have remained in the ureilite PB during its melt phase in order to stabilize Re against its extraction through the melt. However, in a rapid fractional melt extraction model, molten metal is also thought to be removed prior to any significant diffusion of highly siderophile elements from the silicate residue. Paradoxically, the theory that the carbonaceous chondrite parent asteroids did not undergo fractionation or differentiation runs contrary to the model of Bunch et al. (2005), who utilized O-isotopic data to reconstruct the CV and CR parent bodies (see the Allende page for details).

The disequilibrium model of fractional melt extraction does provide a means for the extraction of large quantities of smelting-produced metal. Petrogenetic constraints based on ureilite mineralogy and a CV-like precursor require a loss of ~15–24 wt% FeO through either core migration or explosive volcanism of reduced molten metal. Rapidly formed CO+CO2 gas bubbles could have assisted in the ascent of S-rich liquid metal upwards to the surface and into space. Singletary and Grove (2006) proposed that the high pressures generated from smelting-produced CO+CO2 gas may have had no escape pathway and promoted a disruption of the ureilite parent body. An alternative disruption mechanism, proposed by Downes et al (2008), invokes an oblique impact between the proto-ureilite body and a smaller planetesimal resulting in the loss of its outer layers coupled with asteroid-wide pressure-release reduction processes.

Studies of the REE contents of Kenna are most consistent with an incremental batch melting process. However, in their analysis of the conditions that prevailed during the partial melting phase, including a calculation of the predicted Sm content in ureilites, Warren et al. (2006) became convinced that a continuous partial melting process is the best fit to the data, and that a silicate melt porosity of ~5–10% was probably established during the partial melt phase. This permitted the loss of a sulfur melt component (and other siderophile fractionations) into the core, while sustaining the explosive volcanism of a significant Al-bearing basaltic component.

Rankenburg et al. (2008) showed that the Fe deficiency that exists in ureilites, compared to their probable carbonaceous chondrite-like precursor, is proportional to the amount of S presumed to be initially present, which together would be utilized to form an FeS total metallic melt composition. They also suggested a theory for how the highly siderophile element depletion of ureilites could have been established, which they propose occurred through the grain-scale mixing of two separate components, each having different refractory HSE (highly siderophile element) and HSE abundances, during late brecciation processes. An early (0.5 m.y. after CAIs) formation of the ureilite parent body would result in a maximum degree of partial melting of ~30% (Wilson and Goodrich, 2012). An early and rapid phase of silicate melt extraction would result in the removal of radiogenic 26Al, thus preventing further melting of the UPB while retaining its heterogeneous nature. In the course of their studies, Wilson and Goodrich (2012) reconsidered the commonly accepted magma ocean model and rejected it in favor of a model in which melt accumulated in massive intrusions at the base of the lithosphere.

The following model for the thermal and melt extraction history of the UPB is rendered from Fractional melting and smelting on the ureilite parent body, Goodrich et al. (2007), and from Thermal Evolution and Physics of Melt Extraction on the Ureilte Parent Body, Wilson et al. (2008):

Evidence shows that the ureilite parent asteroid must have accreted from an ~80:20 mixture of silicates and ice early in Solar System history ~0.55 m.y. after CAI formation, and that it experienced rapid melting primarily from the heat of 26Al decay. Production of an abundant volume of melt was initiated by ~1 m.y. after CAI formation as temperatures reached 1050°C, and this melt production continued by progressively declining degrees over the next ~4 m.y., at which time an overall degree of melting of 30% was achieved.

Their model demonstrates that during the early stage of radiogenic heating, as the degree of melting reached 0.15%, the melt was rapidly buoyed upwards through an interconnected network of veins and fractures, these thought to be largely created during the previous hydration and dehydration phases on the asteroid. The size of the veins constituting this hierarchical network ranged from 1 mm at grain boundaries at the melt initiation depths, to sizes of over 10 km as the majority of the melt was efficiently lost into space through dikes during episodic explosive volcanism. The speed of the rising melt was bolstered by high volumes of smelting-produced CO+CO2 gas to easily exceed the escape velocity. The entire journey of this melt through the network from grain boundary until extraction from the UPB is calculated to have taken a month at its peak, or only slightly longer thereafter. However, ~15–25% of the total melt volume is expected to have been retained within an extensive asteroid-wide system of perhaps five sill-like intrusions located ~7 km beneath the cold, unmelted outer crust, each reservoir extending ~4.7 km × ~144 km. It was shown that a large percentage of this sill component (~76%) likely represents melt from deeper, gas-poor, more ferroan source regions. Such a source region would be consistent with a scenario for the origin of the feldspathic clasts found in polymict ureilites, given their formation ~1 m.y. after CAIs and their accumulation into a near-surface intrusion, where they underwent slow cooling and closure over the next ~4 m.y.

Consistent with the large sampling of ureilites studied, it was calculated that these rocks represent strata located at depths of between ~10 km and ~50 km, corresponding to pressures of 3 MPa (most magnesian) and 10 MPa (most ferroan), respectively, on an asteroid ~100 km in radius (arguably the radius was between ~20 km and ~475 km). Pressures at its center are estimated to have been ~15.2–26.6 MPa. Melting and smelting are initiated at different times depending on depth. At the shallowest levels where pressures are lowest, ~3 MPa, melting and smelting occur simultaneously as temperatures reach ~1060°C, producing pigeonite residues corresponding to an increase in Mg# from ~62 to 86. This is followed by orthopyroxene production resulting in further progressive increases in Mg# up to ~91, at which point melting and smelting institute pigeonite production again. At increasing depths, smelting begins progressively later than the onset of melting. At the greatest depths inferred for ureilites, corresponding to ~10 MPa of pressure, smelting begins after a temperature of 1200°C is reached, at a point when 21% partial melting has occurred. The production of pigeonite residues is initiated as augite is successively removed through the melt phase. Complete removal of augite occurs at ~30% melting, having attained a molar Fo content of ~76; orthopyroxene production begins thereafter. If ureilite samples should be found from even greater depths, corresponding to pressures greater than ~12.5 MPa, it is predicted that they will be composed entirely of augite and that no smelting will have occurred. The metallic iron that was produced as an end product of the smelting process is believed to have settled into a central core with a radius of ~41 km.

A scenario proposed by Goodrich (2004, 2007) calls for impact disruption of a >200 km diameter proto-ureilite asteroid and its reassembly into one or more daughter objects from which the ureilites were ultimately derived. In support of this model is the non-regolith breccia FRO 93008, which contains adjacent fragments of all three "Goodrich-model" lithologies, consistent with the reassembly of a disrupted body. Evidence supporting the early disruption and reassembly of a single ureilite parent body was presented by Downes et al. (2008).

It has been generally accepted that the microscopic diamonds and lonsdaleite found in Kenna and other ureilites were formed by the medium-level impact-shock forces that liberated this meteorite from its parent body. The correlation that exists between the carbon and O-isotopic compositions among the ureilite groups implies that the carbon was indigenous to the source rocks and was not introduced later through impact-melt injection. Examinations of the graphite/diamond relationships found in Kenna and other ureilites using x-ray diffraction techniques have revealed the presence of compressed graphite. Compressed graphite has been known to occur experimentally under high pressures and temperatures as part of the phase transition to diamond. Furthermore, graphite can be converted to diamond at much lower pressures (above 5.5 GPa) at high temperatures (1400°C) when in the presence of a molten metal transformation catalyst. Consistent with this idea is the fact that kamacite is found in association with all diamond phases but in none of the graphite phases. This research provides solid evidence for the high-pressure catalytic conversion of graphite to diamond in ureilites resulting from an impact event.

In their study of primary and polycrystalline secondary graphite phases present in the diamond-bearing, main group ureilite UAE 001, Hezel et al. (2008) concluded that the evidence favors the formation of diamonds in this and other ureilites through shock forces of varying degree and duration. They argued that a prolonged shock duration could be the cause of the smaller FWHM (full width at half maximum) values that are observed in a subset of the Raman bands of meteoritic diamond. They found that the small FWHM values reflect the conversion to a well-ordered state, and are only coincidentally similar to the FWHM values obtained for diamonds formed through a chemical vapor deposition (CVD) process. Moreover, they found that the large abundance of secondary graphite coexisting with diamond was inconsistent with CVD. In a separate study, Guillou et al (2009) concluded that a quick shock event is the most reasonable formation process given their observation of diamond inclusions within aligned precursor graphitic structures, a feature inconsistent with a diamond graphitization origin.

In the least-shocked, diamond-free ureilite, ALH 78019, the absence of primordial noble gases in graphite, along with a heavy N-isotopic signature in graphite, was found to be inconsistent with the theory that graphite was a precursor to nanodiamond formed by in situ shock conversion processes (Rai, et al., 2002). Utilizing the ureilite NWA 4742, Guillou et al (2009) studied this paradox in which graphite precursor material is depleted in noble gases, while the nanodiamonds into which it was transformed are noble gas-rich. Their investigation led to a proposal that a mixture of two diamond populations is present; i.e., an early population of unknown origin which contains noble gases, and a later population that was formed by shocked graphite depleted in noble gases. They further suggest that the presence of a noble gas-containing graphitic phase surrounding some nanodiamonds could be the result of back-transformation of the early population of diamonds under conditions of slow cooling following a late shock event.

To differentiate between the two competing scenarios for diamond formation on the ureilite parent body, i.e., impact shock vs. chemical vapor deposition (CVD), Nagashima et al. (2012) utilized micro-Raman spectroscopy to study of carbonaceous material in a number of ureilite samples. The resulting spectral data obtained for the major parameters for diamond (peak position, band intensities, and full width at half maximum [FWHM]) were a better match to diamond produced under CVD rather than shock pressure. Moreover, they demonstrated that there was no correlation of the diamond:graphite ratio to the shock level, and found the noble gas and N-isotopic compositions of graphite, amorphous carbon, and diamond to be in accordance with the CVD model, but not with the shock model. Their results suggest a scenario of chemical deposition of graphite, amorphous carbon, and diamond directly onto high-temperature condensates in the primitive solar nebula, with the formation of each phase being associated with specific variations in CH4:H2 ratios commensurate with temperature and pressure changes. The migration of carbonaceous material to silicate grain contacts, as well as the occurrence of compressed graphite in conjunction with diamond, was the result of later shock events on the ureilite parent body.

All ureilites show evidence of having been rapidly cooled (~2–6°C/hr) through the range of ~1100°C to 650°C, and show signs of a sudden pressure drop that initiated smelting reactions between silicates and graphite to produce reduction rims on olivine and pyroxene grains. This extremely rapid cooling and pressure drop led Herrin et al. (2010) to suggest that the catastrophic disruption of the single ureilite parent body resulted in a cloud of innumerable, small, 2nd-generation bodies having sizes of tens of meters or smaller, which were subsequently reassembled. The common presence of these P–T characteristics in diverse types of ureilites supports the view that they all originated on a common UPB (Warren, 2012). Polymict ureilites contain components with a broad range of Mg# and reduced rims normally found in monomict clasts. This is consistent with an origin on a newly assembled parent body. In another example, the olivine compositions within a single thin section of polymict ureilite EET 87720 was found to span the entire range of olivine compositions recorded for unbrecciated ureilites, and the Mg# distribution is nearly identical to that of unbrecciated ureilites, factors which demonstrate a common origin for all ureilites (Downes and Mittlefehldt, 2006). Furthermore, the wide variety of xenolithic clasts from numerous unrelated parent bodies found within the Almahata Sitta ureilite fall supports the scenario of a collisionally-disrupted and reassembled ureilite parent body.

Four stages of reduction are now recognized to represent the continuum of the reduction process that occurred on the ureilite parent body. A reduction grade sequence from R1 to R4 reflecting an increased conversion of graphite to metal has been demonstrated by investigators from Northern Arizona University (J. Wittke and T. Bunch) and Kingsborough Community College (C. Goodrich), and they have developed the following classification parameter:

  R1 R2 R3 R4
Graphite/metal (vol%) >10 10–1 1 0
Rim thickness of reduced olivine <15 µm 15–50 µm <50 vol% of olivine >50 vol% of olivine
Degree of hardness soft medium very hard extreme
Diamonds none few irregular distribution abundant

Based on textural studies, a number of distinct types of ureilite metal have been identified by Goodrich et al. (2009). Primary metal exists along grain boundaries and in association with C, P, and S in melt spherules within low-Ca pyroxene (mostly pigeonite), both sites probably existing in a complementary solid–liquid arrangement. Secondary metal occurs as a result of reduction processes in olivine, as inclusions in graphite, and as sub-micron-sized trails produced by shock mobilization. Reduction metal and graphite inclusion metal have been identified in Kenna; however, grain boundary metal in Kenna has been completely oxidized, and no Fe–C–S–P spherules have yet been identified.

Chronometers such as Sm–Nd reveal an early differentiation age for ureilites of ~4.56 b.y., while the Mn–Cr chronometer applied to feldspathic clasts gives a similar absolute age. High precision Mg-isotopic analyses by Baker and Bizarro (2005) reveal even older ages comparable to that of CAI formation. The short-lived Hf–W chronometer has been utilized in the determination of a very early differentiation stage for ureilites. Budde et al. (2014) utilized the W-isotopic system to constrain the onset of melting and the extraction of both a silicate and a metallic melt. Corresponding to the lowest calculated initial ε182W values obtained for their ureilite samples (~ –3.25), the onset of melting occurred within the first ~2 m.y. of solar system history. Furthermore, thermal evolution modeling based on an 26Al heat source demonstrated that accretion of the UPB occurred within the first ~1 m.y. With regard to the higher calculated initial ε182W values (up to ~ –2.8), it can be inferred that distinct metal–silicate segregation events occurred, or alternatively, that a temporal variation existed between extractions of metallic and silicate melts. A study of the Mn–Cr isotopic systematics by Shukolykov and Lugmair (2006) revealed that the Cr isotopes in Kenna equilibrated after all of the 53Mn had decayed, and therefore Kenna formed later than some of the other ureilites. An isotopic disturbance is observed for Kenna 4.1 b.y. ago, and it has a CRE age of 23 m.y.

Rare opal-A has been identified in the polymict ureilite EET 83309, which was likely formed through hydration of amorphous silica in the parent body regolith when water was introduced (Downes et al., 2016). The specimen of Kenna shown above is a 3.5 g partial slice. The photo below is an excellent petrographic thin section micrograph of Kenna, shown courtesy of Peter Marmet.

standby for kenna ts photo
click on photo for a magnified view
Photo courtesy of Peter Marmet

Ureilites are finally figured out! >>click here