We have studied the thermal and magnetic field evolution of planet Mercury with a core of Fe–Si alloy to assess whether an Fe–Si core matches its present-day partially molten state, Mercury's magnetic field strength, and the observed ancient crustal magnetization. The main advantages of an Fe–Si core, opposed to a previously assumed Fe–S core, are that a Si-bearing core is consistent with the highly reduced nature of Mercury and that no compositional convection is generated upon core solidification, in agreement with magnetic field indications of a stable layer at the top of Mercury's core. This study also present the first implementation of a conductive temperature profile in the core where heat fluxes are sub-adiabatic in a global thermal evolution model. We show that heat migrates from the deep core to the outer part of the core as soon as heat fluxes at the outer core become sub-adiabatic. As a result, the deep core cools throughout Mercury's evolution independent of the temperature evolution at the core-mantle boundary, causing an early start of inner core solidification and magnetic field generation. The conductive layer at the outer core suppresses the rate of core growth after temperature differences between the deep and shallow core are relaxed, such that a magnetic field can be generated until the present. Also, the outer core and mantle operate at higher temperatures than previously thought, which prolongs mantle melting and mantle convection. The results indicate that S is not a necessary ingredient of Mercury's core, bringing bulk compositional models of Mercury more in line with reduced meteorite analogues.