The ferroelectric to paraelectric phase transition in LiTaO$_3$ and in pure as well as Mg doped LiNbO$_3$ is investigated theoretically by atomistic calculations in the framework of the density functional theory, as well as experimentally by calorimetry and electrical conductivity measurements. First principles models within the stochastic self-consistent harmonic approximation (SSCHA) allow to consider anharmonic effects and thus to obtain a realistic estimate of the Curie temperature $T_C$ of both ferroelectrics. \textit{Ab initio} molecular dynamics (AIMD) calculations performed on large supercells confirm the Curie temperatures estimated with the SSCHA approach. Moreover, they also suggest that the structural phase transition is a continuous process beginning at temperatures well below $T_C$. According to AIMD, significant ionic displacements occurr already at temperatures of about 100\,K and 300\,K below $T_C$ in LiTaO$_3$ and LiNbO$_3$, respectively. To asses whether and how far the ionic displacements affect the materials properties, the AIMD results are compared with measurements of the electrical conductivity and of the heat capacity across the phase transition. Our first principles calculations moreover show that Mg ions, a frequently employed dopant, raise the Curie temperature in LiNbO$_3$.
