Research Statement
I research ferroelectric HfO2-based devices and their integration into conventional memory macros as well as neuromorphic CMOS circuits. Together with my colleagues I am realizing circuit building blocks that allow efficient memory storage as well as learning on biological time-scales in hardware. My research interests cover the topics of compact modeling and electrical characterization of semiconductor and memory devices, particularly regarding their integrability and reliability.
Research Areas
Modeling
SPICE/VerilogA phyiscs-based compact models of semiconductor logic and memory devices.
Characterization
Parameter extraction and reliability characterization of semiconductor devices.
Design
Physics-informed design of reliable circuits and memory macros.
Current Projects
Heracles
Heracles is a physics-based compact model for HfO2-based ferroelectric capacitors. It includes thermal models, interface layers and accurately reproduces several device phenomena, such as transient polarization switching and capacitance hysteresis. Heracles consists of a stateful and computationally efficient non-equilibrium thermodynamics description of the device behaviour, allowing for efficient Monte Carlo simulations even in larger analog CMOS circuits. This makes Heracles suitable for Design-Technology Co-Optimization (DTCO) approaches to (analog) compute-in-memory, neuromorphic systems or sensory circuit design.
Olla
Olla is an ASIC project realized in a 180 nm CMOS technology, with the target of being a testbench for emerging memory technolgies, such as FeRAM or ReRAM. For this, Olla has post-processable structures where memory devices can be fabricated in the BEOL of the CMOS wafer. These macros then allow the devices to be used in several macros that are designed specifically for emerging memory devices. Thes include characterization macros, memory read-out macros or in a functional yet minimal, asynchronous spiking neural network.