FuSe2 Topic 3: Strain and Temperature Ex-Situ Processing of Ferroelectric Oxides (STEP FOx) for BEOL Performance

Purdue University, West Lafayette IN

Investigators

Abstract

Nontechnical Description The demand for energy needed to store and process data is growing at an unsustainable rate. Data centers alone consumed over one percent of all global electricity use in 2022 and are projected to double their consumption in the near future. Much of this energy is not even used for doing actual computation. It is instead spent simply moving data to, and from, memory. To overcome this problem, dense, monolithic memory solutions built into, or on top of, the computing logic are needed. Ferroelectric hafnia-based compounds provide a potential solution. Ferroelectrics have a switchable electric polarization with potential for use in energy efficient devices. Critically, hafnia-based compounds can be integrated into modern logic devices while maintaining their memory-enabling ferroelectric properties. Unfortunately, ferroelectric hafnium oxide devices do not yet meet required endurance targets when fabricated in realistic geometries under the required processing conditions. This project addresses this challenge with a co-design framework that links materials science, advanced thermal and mechanical characterization, and machine learning with memory element design. Thus, investigators will maximize ferroelectric performance and endurance of ferroelectric hafnia in geometries typical of modern memory devices. This project addresses the multi-faceted reality of modern semiconductor systems which requires a multidisciplinary workforce. The project will provide research, internship, and touring opportunities built organically from our project’s significant “non-electrical” thermo-mechanical component. These opportunities, combined with direct messaging to students outside electrical and computer engineering, enhance semiconductor recruiting from under-represented, but vital, backgrounds. Technical Description The project’s technical objective centers on developing the material processes and resulting devices that maximize ferroelectric performance and endurance of trench capacitors analogous to those used in dynamic random access memory (DRAM) within the process envelope of the back end of line (BEOL). Realizing this objective will require addressing fundamental questions on how to cultivate the ferroelectric phase, defect, and endurance properties under these constraints. To answer these questions, the project will leverage post-synthesis, nanophotonic enhanced laser anneals with in-depth defect, phase, and strain characterization. Two major technical outcomes will result. First, the project will demonstrate non-planar, hafnia-based devices exhibiting greater than 10^15 switching endurance under BEOL conditions. Second, a suite of thermomechanical tools for imaging phase, strain, and defects in scaled layers (<50 nm) will be developed. The characterization suite’s development will occur in tandem with first-principles analysis to enable a machine-learning driven “process finder”. In total, the project will have impact beyond ferroelectric memory devices: The project is not creating a single process for a particular material; it’s enabling a paradigm for “in place” material development extensible to other dielectric systems. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.