[Read the paper for free here!]
Planetary histories are forever marked - and perhaps defined - by collisions. Starting from the interaction of cosmic dust, asteroids and planets are built by collisions and, subsequently, they may be severely perturbed or even destroyed by them. It is well established that asteroid collisions radically changed the course of life on Earth (see: dinosaurs). However, the rate of collisional activity in our Solar System is in fact quite poorly understood. This situation is far from ideal, as the collisional evolution of the Solar System encompasses important events like:
- The formation of the Moon, in a cataclysmic collision
- The migration of the giant planets, which rearranged the orbital mechanics of the Solar System
- The possible supply of limiting elements for life to planetary surfaces, during episodes of enhanced bombardment of the inner Solar System that coincided with Earth's 'Prebiotic Era'
These proposed events are expected to have directly or indirectly affected rocky objects throughout the inner Solar System: determining whether by debris from the moon-forming collision directly colliding with asteroids, orbital scattering in the wake of giant planet migration, or collisional cascades in the early evolution of the Main Belt. In all cases, collisions should have written a record of dynamical activity into rocks throughout our Solar System.
But where exactly should we look for evidence of these collisions?
Craters populate planetary surfaces. However, planets are actually rather terrible at recording collisional activity. At least in their youth, rocky planets are active worlds; in the case of Earth and Mars having rivers and mountain building and volcanism. Over time, they erase the evidence for early collisional activity via continuous resurfacing. In contrast, primitive asteroids have very long memories. After the first few tens of million years of Solar System history, as asteroids become nice and cold, there is essentially nothing happening to an asteroid except for collisions.
In order to access the asteroidal archive of collisions, what we need to do is measure the ages of impact induced craters. We can obtain a (somewhat) random sample of asteroidal craters using meteorites which contain evidence of having experienced high temperatures and pressures during impact metamorphism. One of the most useful minerals in meteorites is apatite: a calcium phosphate mineral of some ubiquity.
Uranium is compatible in the apatite crystal structure, Lead is not. This allows us to assume that all of the Lead eventually found in an apatite grain has been generated by radioactive U decay, for which the relevant half-lives are known. By measuring Uranium/Lead isotope ratios in an apatite grain, an estimate can be obtained of how long the mineral grain has spent below the closure temperature for Lead diffusion – above which the radiogenic clock for the mineral would remain perpetually at zero.
If a grain has remained a closed system since it first formed, we will obtain what is a called a concordant data point, lying on the Concordia line. The Concordia line is the solution to the age equations of U decay. A closed system grain will move smoothly along this line, to a point that we observe at whatever time the grain has found itself under the steely gaze of a mass spectrometer! However, nature is fickle. Our closed system may be subject to an episode of high temperatures, above the closure temperature for Lead diffusion. Conditions such as these may occur in the wake of a collision. Whilst temperatures remain sufficiently high, Lead will escape from the crystal and our isotope compositions will evolve, moving along a straight line towards the origin – data points lying in this region are called discordant (Figure 1).
Figure 1: General form of a concordia diagram for calculating Uranium-Lead ages. The age of formation of a hypothetical U-bearing phase is obtained with a concordant data point (in blue), along with the timing of a recent thermal disturbance being recorded by discordant data points (in red) from several grains (e.g., of different sizes).
Because diffusion rates from crystals may vary depending on local temperature reached, or the grain size, the degree of Lead-loss, or discordance, will vary (Figure 1). By plotting a linear regression through these newly scattered points, we can recover information: both their original placement on the concordia line, called the upper intercept, and find the lower intercept of the Discordant data regression line with the concordia line, revealing the time at which the Lead-loss most recently took place. The upper intercept age may reflect the time at which Lead was last zero within the grains – possibly the time of formation, or some earlier high temperature event that flushed the system completely. In this way, we can see how extremely different ages may be recovered by a single suite of phosphates, and which may reasonably span even the 4.5 billion years of Solar System history witnessed by asteroids. Understanding which estimates of collision age are meaningful is critical if we wish to trace dynamical events with the meteorite record.
And yet, nobody can agree! A good example is the Chelyabinsk meteorite, which fell over Russia in 2013. Chelyabinsk is a highly shocked LL5 type ordinary chondrite meteorite, preserving 3 distinct lithologies: a light lithology (equivalent to asteroidal host rock with some evidence of having been 'shocked' during an impact), a dark lithology (with a higher percentage of impact-induced melt), and a melt lithology (which is exactly what it says on the tin). Macro-textural relationships clearly show that all three lithologies formed from the same parental material, and that they formed simultaneously. Each lithology reached different peak temperatures before thermally equilibrating with one another, but clearly failed to texturally equilibrate.
Figure 2: Rock textures in Chelyabinsk. Panels a-d: evolution of Chelyabinsk breccia (a), from (b) initial formation during shock-melting, brecciation, and shock darkening of host rock material, through (c) solidification into light, dark, and melt lithologies, and (d) subsequent minor disturbances, such as the propagation of fracture networks. Pink symbols represent host-rock phosphate minerals, which are only found in the light and dark lithologies. Photograph used is of Chelyabinsk specimen NHMV-O707; Credit Ludovic Ferriere-NHM Vienna, Austria.
Previous work established two candidate ages for a major collision using the apatite U-Lead method. Some would argue that the lower intercept is good evidence of recent Lead-loss, and that a major reheating event occurred at this time. However, as mentioned earlier, the upper intercept age need not reflect formation: this age may also record complete resetting in an earlier and even more intensive collisional reheating episode. Indeed, for Chelyabinsk, the upper intercept age obtained is younger than we would expect for ‘normal’ asteroidal phosphates, which should have gone ‘cold’ about 50 Myr earlier. This young upper intercept age could suggest that Chelyabinsk phosphates were reheated at some earlier time, becoming completely reset. At this point we have two apparent collisional events, and we want to know whether they are both meaningful. Clearly, without confidence in what individual ages mean, we have a problem for our grand vision of using meteorite ages to trace dynamics!
We wondered if we might help resolve this quandary with the aid of mineral micro-textures, which can provide an alternative line of evidence for the superposition of shock events and their associated temperature conditions over time. We carried out a detailed study of isotopic ages and micro-textures of phosphates in Chelyabinsk, to see if we could come up with a simple interpretation of the impact history.
The micro-textures of phosphates in Chelyabinsk fall into two major classes. There are high temperature micro-textures, formed simultaneously with the light-dark-melt macro-texture of the meteorite. These include crystal-plastic deformation of individual crystals, where strain is accommodated across stacked slip planes throughout the crystal over the duration of an impact event. There are also recrystallised grains, where strain is totally absent. Here, parental grains have reorganised into individual smaller grains with clear grain boundaries and random orientations of their respective crystal lattices – this tends to take a long time, but can be sped up at high temperature. Indeed, you only find such textures in the dark lithology, which was subject to more intensive heating than the light lithology. The formation of these textures goes along with high pressure-temperature conditions - and probable Lead-loss.
Overprinting all of these high-temperature micro-textures are fractures developed during brittle failure. Crystal domains with defects such as these may be more susceptible to Lead-loss than those without. However, fractured domains are often avoided during Uranium-Lead analyses in favour of ‘pristine’ areas. But this misses a crucial point: what if the Uranium-Lead discordance observed in Chelyabinsk – and other meteoritic phosphates - is simply derived from measuring damaged domains, despite people’s best efforts to avoid them?
The next step, then, is to construct some hypotheses to test competing interpretations of phosphate Uranium-Lead ages in Chelyabinsk (Figure 3):
Figure 3: end-member expectations for Uranium-Lead isotope compositions of Chelyabinsk phosphates.
- Option 1) If recent Lead-loss occurred during the formation of the light, dark, and melt lithologies, we would expect a correlation between lithology and Lead-loss. Dark lithology grains will have reached higher temperatures for longer, and should thus have lost more Lead. That is: we will observe a macro-scale correlation between Lead-loss and phosphate setting.
- Option 2) If recent Lead-loss occurred during a more minor recent event, and the three lithologies formed a long time ago, then there should be no such correlation of Lead-loss with lithology. We may instead expect Lead-loss to correlate within each lithology with overprinting textural features, such as fractures, which could have damaged the phosphate crystal lattice.
We tested each scenario with Secondary Ionisation mass spectrometry analyses, giving us spot compositions that we can tie to individual grain textures. Figure 4 shows the key results. Once again you are looking at spot data plotted on a concordia diagram, with light blue markers indicating the center of ellipsoidal errors. I have also plotted previously measured data from one of those earlier studies on Chelyabinsk phosphate Uranium-Lead ages, with brown triangle markers. On the left hand plot, the data are colour coded by lithology: purple for dark lithology, orange for light lithology. On the right hand plot, the data are colour coded by crystal domain, light purple are damaged domains (spots overlapping fractures) and red is pristine domains (no visible defects in proximity). Below each plot are graphs of the cumulative distribution of 207Pb/235U compositions, allowing us to visualize how similar these data groupings are. Shaded in green is a reasonable range of graph space that corresponds - given the analytical errors we tend to get - to concordant data.
Figure 4: Uranium-Lead isotope systematics of the Chelyabinsk chondrite - reproduced from Walton et al., 2022.
There are several key observations to make of these data. The first is that our data define a much wider range of compositions, falling well along the regression line and almost reaching the lower intercept itself. That is to say, we have observed grains with much more extensive Lead-loss than previous studies! That is an important finding.
For one, it allows us to much more tightly constrain the lower intercept age, as we are no longer extrapolating far from where we have the most data to get the lower intercept. We can also clearly see that discordance in Chelyabinsk is mainly related to overprinting fractures a.k.a damaged domains, as we speculated about earlier.
What all of that comes down to is a rejection of the first hypothesis: there is no macro-scale correlation of lithology with phosphate Uranium-Lead ages. Meanwhile, there is a strong correlation between damage-related micro-textures and Uranium-Lead isotope composition. We can therefore infer that plastic deformation and recrystallisation of phosphates and the major collision that created the light-dark-melt macro-texture of Chelyabinsk occurred at the time indicated by the upper intercept – 4.47 Ga, and late Lead-loss took place during a more recent event that overprinted those high energy textures with fractures.
Taking all of the data together to calculate a best estimate regression age, we get an upper intercept age of 4473 million years and a lower intercept age of 0 Ma. This is supportive of a history where the Chelyabinsk parent asteroid experienced major collisional reheating at ~ 4473 Ma, and then a recent fragmentation event that damaged phosphate grains, caused Lead-loss, and sent Chelyabinsk on its way to Earth.
Finally, let us go back the big picture and ask: with our new confidence in what meteorite phosphate Uranium-Lead ages probably mean, can we bolster the idea of tracing dynamical events with meteorites?
Figure 5 shows a compilation of meteorite Uranium-Lead ages. On the left hand plot, we have lower intercept ages, and on the right hand side we have upper intercept ages. Our new ages for Chelyabinsk are indicated with dark markers in the LL portion (it is an LL chondrite). You can see how dating damaged crystal domains hugely revised our estimate of the lower intercept age. You can also see how the Chelyabinsk upper intercept age falls squarely in a cluster of similar ages, which span about 4450-4480 Ma.
Figure 5: Uranium-Lead age record of chondritic meteorites - reproduced from Walton et al., 2022.
From our new data, we can have more confidence that upper intercept ages younger than the end of parent body heating on asteroids reliably record intense collisional reheating, as they do in Chelyabinsk. The clustering of ages at 4450-4480 Ma, in meteorites from various different parent bodies, is indicative of a major dynamical event that generated numerous energetic collisions in this time frame – this may be either late instability or Moon-formation. Meanwhile, we must be cautious of lower intercept ages that were obtained from data that lie far from the lower intercept, and/or did not take into account grain micro-textures when placing spots for analysis. However, the lower intercept record may still be geologically meaningful, especially if texturally guided analyses become the norm.
The next big question is to figure out what the ancient phosphate Uranium-Lead age cluster actually means. Is this a sampling artefact? If not, what event exactly has been recorded? Using shocked meteorites as our Rosetta Stone for reading Solar System collisional history, I am optimistic that one day soon we might just find out.
This content is a summary of Walton et al., 2022. Ancient and recent collisions revealed by phosphate minerals in the Chelyabinsk meteorite. DOI : 10.1038/s43247-022-00373-1. Numerous talented researchers based in the UK, USA, Italy, Austria, and China contributed to this work.