Micro- and nano-electronic devices based on metal-oxide-metal (MOM) and complementary metal-oxide-semiconductor (CMOS) systems are the basis of modern technology. Over many decades, scaling down of transistor- and memory-device dimensions have led to exponential advances in computing power. This scaling has also led to the replacement of SiO₂ with higher dielectric constant materials, such as HfO₂. However, in spite of decades of perfection, oxides used in CMOS devices, including SiO₂ and HfO₂, are prone to many field-induced reliability issues^[1]. Here, we present the results of multi-scale modelling which link the physical processes responsible for the field induced degradation of a-HfO₂ with the characteristics of time-dependent dielectric breakdown (TDDB), such as temporal evolution of current through the oxide and Weibull plots of TDDB. The model is based on previous work, which uses density functional theory to show that electrons trap into deep, intrinsic states in HfO₂^[2]. The trapping of electrons facilitates the production of vacancies by lowering the activation energy for a vacancy-interstitial pair^[3]. Trapping of electrons at vacancies also facilitates the production of additional vacancies. We calculate a range of parameters associated with these defects and reactions and use this parameterize a device level simulation in the GinestraÒ^[4] code. Effects such as TAT, Fowler-Nordheim tunneling, heat generation and vacancy generation et cetera are all included. This allows us to simulate the time evolution current through a TiN/HfO₂/TiN subjected to electrical stress. The results show that the trap based model agrees well with experiment. We also gain an insight into the formation of the percolation pathway from the spatial distribution of generated vacancies. This also confirms the Joule feedback mechanism, in which hard breakdown is caused by local heating around the percolation path, leading to a catastrophic increase and current and generation of vacancies.