The crust-core phase transition of neutron stars is quantitatively studied within a unified meta-modelling of the nuclear Equation of State (EoS). The variational equations in the crust are solved within a Compressible Liquid Drop (CLD) approach, with surface parameters consistently optimized for each EoS set on experimental nuclear mass data. When EoS parameters are taken from known Skyrme or RMF functionals, the transition point of those models is nicely reproduced. A model-independent probability distribution of EoS parameters and of the transition density and pressure is determined with a Bayesian analysis, where the prior is given by an uncorrelated distribution of parameters within the present empirical uncertainties, and constraints are applied both from neutron star physics and ab-initio modelling. We show that the characteristics of the transition point are largely independent of the high density properties of the EoS, while ab-initio EoS calculations of neutron and symmetric matter are far more constraining. The most influential parameter for the determination of the transition point governs the surface properties of extremely neutron rich matter, and it is strongly unconstrained. This explains the large dispersion of existing predictions of the transition point. Only if the surface tension is fixed to a reasonable but somewhat arbitrary value, strong correlations with isovector EoS parameters (L sym , K sym and Q sym ) are recovered. Within the present experimental and theoretical uncertainties on those parameters, we estimate the transition density as n t = 0.072 ± 0.011 fm −3 and the transition pressure as P t = 0.339 ± 0.115 MeV fm −3 .

Context. In the cooling process of a non-accreting neutron star, the composition and properties of the crust are thought to be fixed at the finite temperature where nuclear reactions fall out of equilibrium. A lower estimation for this temperature is given by the crystallization temperature, which can be as high as ≈ 7 × 10 9 K in the inner crust, potentially leading to sizeable differences with respect to the simplifying cold-catalyzed matter hypothesis. Aims. We extend the recent work by Fantina et al. (2019) on the outer crust, to the study of the crystallization of the inner crust and the associated composition in the one-component plasma approximation.Methods. The finite temperature variational equations for non-uniform matter in both the liquid and the solid phases are solved using a compressible liquid-drop approach with parameters optimized on four different microscopic models which cover the present uncertainties in nuclear modeling. Results. We consider separately the effect of the different nuclear ingredients with their associated uncertainties, namely the nuclear equation of state, the surface properties in the presence of a uniform gas of dripped neutrons, and the proton shell effects arising from the ion single-particle structure. Our results suggest that the highest source of model dependence comes from the smooth part of the nuclear functional. Conclusions. We show that shell effects play an important role at the lowest densities close to the outer crust, but the most important physical ingredient to be settled for a quantitative prediction of the inner crust properties is the surface tension at extreme isospin values.