Stable stratification at the top of the Earth's outer core has been suggested based upon seismic and geomagnetic observations, however, the origin of the layer is still unknown. In this paper we focus on a thermal origin for the layer and conduct a systematic study on the thermal evolution of the core. We develop a new numerical code to model the growth of thermally stable layers beneath the CMB, integrated into a thermodynamic model for the long term evolution of the core. We conduct a systematic study on plausible thermal histories using a range of core properties and, combining thickness and stratification strength constraints, investigate the limits upon the present day structure of the thermal layer. We find that whilst there are a number of scenarios for the history of the CMB heat flow, Q_c, that give rise to thermal stratification, many of them are inconsistent with previously published exponential trends in Q_c from mantle evolution models. Layers formed due to an exponentially decaying Q_c are limited to 250–400 km thick and have maximum present-day Brunt-Väisälä periods, T_BV = 8 − 24 hrs. When entrainment of the lowermost region of the layer is included in our model, the upper limit of the layer size is reduced and can fully inhibit the growth of any layer if our non-dimensional measure of entrainment, E > 0.2. The period T_BV is insensitive to the evolution and so our estimates remain distinct from estimates arising from a chemical origin. Therefore, T_BV should be able to discern between thermal and chemical mechanisms as improved seismic constraints are obtained.

根据地震和地磁观测，已经提出地球外核顶部的稳定分层，但是，该层的起源仍然未知。在本文中，我们专注于该层的热起源，并对核心的热演化进行系统研究。我们开发了一种新的数值代码来模拟 CMB 下热稳定层的生长，并将其集成到用于核心长期演化的热力学模型中。我们使用一系列核心特性对合理的热历史进行了系统研究，并结合厚度和分层强度约束，研究了当前热层结构的限制。我们发现，虽然 CMB 热流的历史有多种情况，但Q _c，导致热分层，其中许多与先前公布的来自地幔演化模型的Q _c指数趋势不一致。由于Q _c指数衰减而形成的层被限制在 250-400 公里厚，并且具有最大的当前布伦特-维萨拉周期，T _BV  = 8 - 24 小时。当我们的模型中包含层的最低区域的夹带时，如果我们的无量纲夹带量E  > 0.2 ，层大小的上限会减小，并且可以完全抑制任何层的生长。周期T _BV对进化不敏感，因此我们的估计与化学来源的估计仍然不同。因此，Ť _BV应该能够为获得改进地震约束热和化学机制之间进行辨别。