PhD - Leuven | More than two weeks ago
The microelectronics industry is facing a roadblock on the Moore’s law of miniaturization, hence there is a need to employ new physical principles, materials and devices to keep improving the energy efficiency and performance of logic/memory technologies and go beyond simple downscaling. On that front, several emerging memory technologies, based on ferroelectricity/magnetism gain traction. The proposed SCM (Storage Class Memory) is positioned between fast/volatile DRAM and the slow/non-volatile FLASH. One of the possible technologies to use as memory cell would be ferroelectric switching of polarization in a FEFET transistor. Other newly emerging fields, like quantum computers would take advantage of mixed memory-logic devices, built with multiferroic materials.
To help improve the existing technologies or build the medium-to-long-term vision for the future microelectronics devices, there is a need to understand theoretical limitations, properties, and interactions of different ferroelectric / multiferroic materials that build-up complex memory or logic devices. However, these types of materials require deeper understanding in terms of defects, interaction with other materials that they are in contact with (effective work functions, for example), morphology change in time (phase transformations or ferroelectric switching) of complex polycrystalline films, multiferroic interactions, etc.
In this Ph.D. project, we aim at solving this issue by taking advantage of first-principles and/or multi-scale simulations, which are excellent for understanding thin films, (metallic electrode) interfaces, electronic/ thermal properties, and the effect of defects on those properties on ferroelectric/multiferroic materials. As such, they can be used to understand and select the best materials for a specific device at hand. For multi-scale insight into the polycrystalline/polydomain ferroelectric switching dynamics in different types of device shapes and sizes, a Machine-Learning approach can be used to derive neuromorphic ML atomic potentials for faster simulations to gain understanding at larger size/time-scales. Building fundamental insights on the materials interaction and evolution in time at different scales will help developing the much-needed theoretical understanding of materials performance and reliability and constitutes the skeleton of this PhD project.
During this project, the PhD. student will be performing state-of-the-art ab initio calculations. Carrying-on this correctly and efficiently requires a proper understanding of the theoretical concepts on which the methods are based and on their implementation in computer code to be executed on super computers. The type of simulations to perform also requires an understanding of the relevant chemistry. The understanding and the necessary skills will be trained at imec.
Eligibility criteria: Master’s degree in physics or chemistry (focusing on theoretical aspects). Due to the complexity and the high amount of individual calculations, an efficient and robust automation and data processing infrastructure is essential. We continuously develop and improve such an infrastructure for all our calculations, written in python. Good knowledge of this language is a plus. A strong motivation, good knowledge of solid-state physics or quantum chemistry and UNIX/LINUX are a plus. Excellent writing / oral communication skills are desired.
Required background: Engineering Technology, Engineering Science, Chemistry, Physics
Type of work: 80% modeling/simulation, 10% literature, 10% reporting
Supervisor: Michel Houssa
Co-supervisor: Sergiu Clima
Daily advisor: Sergiu Clima
The reference code for this position is 2023-009. Mention this reference code on your application form.