We examine the mechanisms by which atmosphere can be eroded by giant impacts onto Earth-like planets with thin atmospheres, using 3D smoothed particle hydrodynamics simulations with sufficient resolution to directly model the fate of low-mass atmospheres. We present a simple scaling law to estimate the fraction lost for any impact angle and speed in this regime. In the canonical Moon-forming impact, only around 10% of the atmosphere would have been lost from the immediate effects of the collision. There is a gradual transition from removing almost none to almost all of the atmosphere for a grazing impact as it becomes more head-on or increases in speed, including complex, non-monotonic behaviour at low impact angles. In contrast, for head-on impacts, a slightly greater speed can suddenly remove much more atmosphere. Our results broadly agree with the application of 1D models of local atmosphere loss to the ground speeds measured directly from our simulations. However, previous analytical models of shock-wave propagation from an idealised point-mass impact significantly underestimate the ground speeds and hence the total erosion. The strong dependence on impact angle and the interplay of multiple non-linear and asymmetrical loss mechanisms highlight the need for 3D simulations in order to make realistic predictions.

We present a new scaling law to predict the loss of atmosphere from planetary collisions for any speed, angle, impactor mass, target mass, and body compositions, in the regime of giant impacts onto broadly terrestrial planets with relatively thin atmospheres. To this end, we examine the erosion caused by a wide range of impacts, using 3D smoothed particle hydrodynamics simulations with sufficiently high resolution to directly model the fate of low-mass atmospheres. Different collision scenarios lead to extremely different behaviours and consequences for the planets. In spite of this complexity, the fraction of lost atmosphere is fitted well by a power law. Scaling is independent of the system mass for a constant impactor mass ratio. We find no evident departure from the trend at the extremes of the parameters explored. The scaling law can readily be incorporated into models of planet formation.

We simulate the hypothesised collision between the proto-Earth and a Mars-sized impactor that created the Moon. Amongst the resulting debris disk in some impacts, we find a self-gravitating clump of material. It is roughly the mass of the Moon, contains $\sim1\%$ iron like the Moon, and has its internal composition resolved for the first time. The clump contains mainly impactor material near its core but becomes increasingly enriched in proto-Earth material near its surface. A graduated composition has recently been measured in the oxygen isotope ratios of Apollo samples, suggesting incomplete mixing between proto-Earth and impactor material that formed the Moon. However, the formation of the Moon-sized clump depends sensitively on the spin of the impactor. To explore this, we develop a fast method to construct models of multi-layered, rotating bodies and their conversion into initial conditions for smoothed particle hydrodynamical (SPH) simulations. We use our publicly available code to calculate density and pressure profiles in hydrostatic equilibrium, then generate configurations of over a billion particles with SPH densities within $1\%$ of the desired values. This algorithm runs in a few minutes on a desktop computer, for $10^7$ particles, and allows direct control over the properties of the spinning body. In comparison, relaxation or spin-up techniques that take hours on a supercomputer before the structure of the rotating body is even known. Collisions that differ only in the impactor's initial spin reveal a wide variety of outcomes: a merger, a grazing hit-and-run, or the creation of an orbiting proto-Moon.

We establish the feasibility of measuring the neutron lifetime via an alternative, space-based class of methods, which use neutrons generated by galactic cosmic ray spallation of planets' surfaces and atmospheres. Free neutrons decay via the weak interaction with a mean lifetime of around 880 s. This lifetime constrains the unitarity of the CKM matrix and is a key parameter for studies of Big-Bang nucleosynthesis. However, current laboratory measurements, using two independent approaches, differ by over 4$\sigma$. Using data acquired in 2007 and 2008 during flybys of Venus and Mercury by NASA's MESSENGER spacecraft, which was not designed to make this measurement, we estimate the neutron lifetime to be $780\pm60_\textrm{stat}\pm70_\textrm{syst}$ s, thereby demonstrating the viability of this new approach.

We perform simulations of giant impacts onto the young Uranus using smoothed particle hydrodynamics (SPH) with over 100 million particles. This 100--1000$\times$ improvement in particle number reveals that simulations with below 10^7 particles fail to converge on even bulk properties like the post-impact rotation period, or on the detailed erosion of the atmosphere. Higher resolutions appear to determine these large-scale results reliably, but even 10^8 particles may not be sufficient to study the detailed composition of the debris -- finding that almost an order of magnitude more rock is ejected beyond the Roche radius than with 10^5 particles. We present two software developments that enable this increase in the feasible number of particles. First, we present an algorithm to place any number of particles in a spherical shell such that they all have an SPH density within 1% of the desired value. Particles in model planets built from these nested shells have a root-mean-squared velocity below 1% of the escape speed, which avoids the need for long precursor simulations to produce relaxed initial conditions. Second, we develop the hydrodynamics code SWIFT for planetary simulations. SWIFT uses task-based parallelism and other modern algorithmic approaches to take full advantage of contemporary supercomputer architectures. Both the particle placement code and SWIFT are publicly released.

We perform a suite of smoothed particle hydrodynamics simulations to investigate in detail the results of a giant impact on the young Uranus. We study the internal structure, rotation rate, and atmospheric retention of the post-impact planet, as well as the composition of material ejected into orbit. Most of the material from the impactor's rocky core falls in to the core of the target. However, for higher angular momentum impacts, significant amounts become embedded anisotropically as lumps in the ice layer. Furthermore, most of the impactor's ice and energy is deposited in a hot, high-entropy shell at a radius of ~3 Earth radii. This could explain Uranus' observed lack of heat flow from the interior and be relevant for understanding its asymmetric magnetic field. We verify the results from the single previous study of lower resolution simulations that an impactor with a mass of at least 2 Earth masses can produce sufficiently rapid rotation in the post-impact Uranus for a range of angular momenta. At least 90% of the atmosphere remains bound to the final planet after the collision, but over half can be ejected beyond the Roche radius by a 2 or 3 Earth mass impactor. This atmospheric erosion peaks for intermediate impactor angular momenta (~3*10^36 kg m^2 s^-1). Rock is more efficiently placed into orbit and made available for satellite formation by 2 Earth mass impactors than 3 Earth mass ones, because it requires tidal disruption that is suppressed by the more massive impactors.

We perform the first tests of various proposed explanations for observed features of the Moon's argon exosphere, including models of: spatially varying surface interactions; a source that reflects the lunar near-surface potassium distribution; and temporally varying cold trap areas. Measurements from the Lunar Atmosphere and Dust Environment Explorer (LADEE) and the Lunar Atmosphere Composition Experiment (LACE) are used to test whether these models can reproduce the data. The spatially varying surface interactions hypothesized in previous work cannot reproduce the persistent argon enhancement observed over the western maria. They also fail to match the observed local time of the near-sunrise peak in argon density, which is the same for the highland and mare regions, and is well reproduced by simple surface interactions with a ubiquitous desorption energy of 28 kJ/mol. A localised source can explain the observations, with a trade-off between an unexpectedly localised source or an unexpectedly brief lifetime of argon atoms in the exosphere. To match the observations, a point-like source requires source and loss rates of ~$1.9\times10^{21}$ atoms/s. A more diffuse source, weighted by the near-surface potassium, requires much higher rates of ~$1.1\times10^{22}$ atoms/s, corresponding to a mean lifetime of just 1.4 lunar days. We do not address the mechanism for producing a localised source, but demonstrate that this appears to be the only model that can reproduce the observations. Large, seasonally varying cold traps could explain the long-term fluctuation in the global argon density observed by LADEE, but not that by LACE.