The Martian surface composition appears mainly mafic but recent observations have revealed the presence of differentiated rocks, only in the Highlands. Here, we demonstrate that differentiated melts can form during the construction of thick crustal regions on Mars by fractional crystallisation of a mafic protolith, without plate tectonics. On a stagnant-lid planet, regions of thicker crusts contain more heat-producing elements and are associated to thinner lithospheres and to higher mantle melt fractions. This induces larger crustal extraction rates where the crust is thicker. This positive feedback mechanism is favoured at large wavelengths and can explain the formation of the Martian dichotomy. We further develop an asymmetric parameterised thermal evolution model accounting for crustal extraction, where the well-mixed convective mantle is topped by two lithospheres (North/South) characterised by specific thermal and crustal structures. We use this model in a Bayesian inversion to investigate the conditions that allow crustal temperatures to be maintained above the basalt solidus during crustal growth, resulting in the formation of evolved melts. Among the thermal evolution models matching constraints on the structure of the Martian crust and mantle provided by the InSight NASA mission, a non-negligible fraction allows partial melting and differentiation of the crust in the south, which can occur very early (<100 Myr) as well as during the Hesperian ; partial melting in the north appears unlikely. Although crustal differentiation may occur on a hemispheric scale on Mars, its vertical extent is limited to less than a third of the crustal thickness.

We report observations of Rayleigh waves that orbit around Mars up to three times following the S1222a marsquake. Averaging these signals, we find the largest amplitude signals at 30 s and 85 s central period, propagating with distinctly different group velocities of 2.9 km/s and 3.8 km/s, respectively. The group velocities constraining the average crustal thickness beneath the great circle path rule out the majority of previous crustal models of Mars that have a >200 kg/m3 density contrast across the dichotomy. We find that the thickness of the martian crust is 42-56 km on average, and thus thicker than the crusts of the Earth and Moon. Together with thermal evolution models, a thick martian crust suggests that the crust must contain 50-70% of the total heat production to explain present-day local melt zones in the interior of Mars.

Recent estimates of the crustal thickness of Mars show a bimodal result of either ∼20 km or ∼40 km beneath the InSight lander. We propose an approach based on random matrix theory applied to receiver functions to further constrain the subsurface structure. Assuming a spiked covariance model for our data, we first use the phase transition properties of the singular value spectrum of random matrices to detect coherent arrivals in the waveforms. Examples from terrestrial data show how the method works in different scenarios. We identify three new converted arrivals in the InSight data, including the second multiply reflected phase from a deeper third interface. We then use this information to jointly invert receiver functions with the absolute S-wave velocity information in the polarization of body waves. Results show a crustal thickness of 43±5 km beneath the lander with two mid-crustal interfaces at depths of 8.5±1.5 km and 22±3 km.