Understanding and Controlling the Energy Landscape of Hf_0.5Zr_0.5O₂

As the thickness of field effect transistors (FETs) reaches the atomic limit, attention has shifted towards increasing the functionality of FETs to preserve Moore’s law. Enhanced FET functionality can be achieved by modifying the gate oxide, and Hf_0.5Zr_0.5O₂ (HZO) has proven to be incredibly effective to this end.

However, the energy landscape of HZO is a complex triple well potential, which features an energy barrier for the non-polar to polar transition, and stable ferroelectric and antiferroelectric structures that are energetically similar. The coexistence of these two structures leads to the equivalent oxide thickness (EOT) of a gate oxide stack with HZO being lower than the interfacial SiO₂ thickness, which usually determines the lower limit of the EOT. This is indicative of a higher capacitance and, consequently, a more effective FET.

This talk will explore the energy landscape of HZO, which is fundamental to its effectiveness in FETs. Multi-scale data will be used to elucidate this landscape, from machine learning and Landau potentials fitted to density functional theory calculations, to the results of phase-field simulations. These data will be employed to explain the effects of oxygen defects and strain on the HZO energy landscape, and ascertain how such conditions can be used to control it.