Modelling 4.5 billion years of Earth’s thermal evolution: Insights from core-mantle coupling, lithospheric viscosity, grain-size-dependent rheology, and surface boundary conditions

• Earth’s thermal evolution over 4.5 Ga modeled using a pseudo-spectral approach. • Lithospheric viscosity of ∼10 22 Pa.s predicts petrological mantle cooling trends. • Predicted inner-core growth onset at ∼2.0–1.5 Ga aligns with independent estimates. • Plate-like surface boundary condition yields the most realistic thermal evolution. • Grain-growth-controlled diffusion viscosity tunes mantle viscosity structure without altering the pattern of mantle cooling. We investigate Earth’s evolution through thermally coupled core-mantle models spanning 4.5 billion years. These models employ a spherical pseudo-spectral approach to solve the conservation equations for mass, momentum, and energy within a compressible, self-gravitating mantle. The methodology incorporates time-dependent treatments for core-mantle coupling, dislocation and diffusion creep mechanisms, internal heating, and thermal conductivity. Using 3-D numerical simulations, we evaluate the sensitivity of mantle cooling, viscosity structure, and inner-core growth to variations in lithospheric viscosity, diffusion viscosity, mechanical surface boundary conditions, and initial core-mantle boundary and core liquidus temperatures. Results underscore the central role of lithospheric viscosity, particularly near an effective value of ∼10 22 Pa · s , in producing a mantle cooling pattern consistent with petrological constraints, characterized by net warming prior to ∼3 billion years ago (Ga) followed by long-term cooling, as predicted by low-Urey-ratio thermal evolution models. Notably, one model with lithospheric viscosity allowed to vary between 10 18 and 10 24 Pa · s exhibits nonlinear rheological feedbacks that trigger an early-stage thermal rebound. This behavior results from a relatively abrupt increase in lithospheric viscosity which redirects the mantle onto a sustained warming trajectory that departs from the expected monotonic cooling. This example also demonstrates how nonlinear parameter interactions can produce non-monotonic thermal evolution. However, lithospheric viscosity alone cannot fully account for present-day observations of heat flux, inner-core radius, and depth-dependent viscosity profiles. We find that varying the activation enthalpy ratio for grain-growth-controlled diffusion viscosity modifies the radial viscosity structure while leaving the overall cooling pattern intact. Furthermore, surface boundary conditions permitting viscous coupling between rigid surface plates and underlying mantle flow – specifically in our plate-like (PL) model – yield the most acceptable mantle cooling rates and dynamic evolution. This PL configuration also facilitates more realistic coupling between surface kinematics and internal convection, allowing plate velocities to emerge from the flow dynamics rather than being imposed. The PL model exhibits patterns which are similar to independently estimated present-day mantle viscosity profiles, including features such as the lithosphere-asthenosphere gradient and the viscosity jump at the 660 km discontinuity. The PL model also exhibits persistent large-scale lateral temperature anomalies, consistent with previous billion-year convection studies, and illustrates how plate-like surface coupling promotes the emergence and maintenance of hemispheric-scale heterogeneity. Findings confirm that initial core-mantle boundary temperature and liquidus temperature at the inner-core boundary significantly influence inner-core growth rates. To isolate the effects of the initial thermal state of the core, we adopt a simplified initialization in which the liquidus temperature is set equal to the core-mantle boundary (CMB) temperature at time zero – this condition is imposed only at initialization. For the PL model, initializing both the core-mantle boundary temperature and the inner-core liquidus temperature at 5600 K optimizes predictions of present-day inner-core radius and suggests an inner-core onset around 2.0 ‒ 1.5 Ga, aligning with previous independent estimates. This study emphasizes that a robust modeling of Earth’s core-mantle thermal and dynamic history requires careful calibration of lithospheric viscosity and grain-size-sensitive mantle viscosity, surface boundary dynamics, and initial temperatures at the CMB and core liquidus. All model predictions are empirically anchored and evaluated against a wide array of observational constraints. These findings provide a framework for refining constraints on the timing of plate tectonics onset and the coupled evolution of mantle and core, while enabling systematic evaluations of system parameters, initial and surface boundary conditions, and their interactions throughout Earth’s history.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.