Master’s student

Implementing a Perceptron in Hardware with Ferroelectric Synapses

Ref. 2021_045


Neural Networks accelerators are based on reprogrammable resistances or synaptic weights. Recently, the discovery of ferroelectricity in hafnium oxide[1] and HfZrO4 (HZO)[2] enabled the demonstration of CMOS compatible ferroelectric synapses. At the Neuromorphic Devices and Systems group – IBM, we demonstrated non-volatile resistive switching (synaptic functionality) in a 4.5 nm thick ferroelectric layer fabricated with a CMOS compatible process.[3]



The goal of this project is two-fold: first, to demonstrate the functionality of a perceptron based on the existing technology of ferroelectric synapses, by reprogramming passive and then active cross-bar arrays. The size will be up to 32x32 synapses. Second, to understand the resistive switching mechanisms and provide guidelines for the fabrication of the next generation of devices. A publication combining the two aspects (array functionality and materials science) is expected at the outcome of the project.



A 32x32 board will be used to characterize the arrays. The first experiment consists in performing a machine-learning experiment (for example the “iris” dataset). The intern will electrically characterize the single devices in the cross-bar array, then quantify the effect of sneak paths and search for the optimal method to perform the synaptic weight update. The 32x32 board will be controlled using Python programming. The conduction mechanisms and the origin of the resistive switching will be explored using temperature dependent measurements on single devices.


Risk Plan Regarding the COVID Situation

The access to the lab is currently not restricted. In case of lock-down, VPN solutions allows remote use of the services provided on-site (software licenses, cloud, access to on-line journals…) and some experiments can be remotely accessed. Simulations represent another research direction: circuit level simulation of a cross-bar array using available experimental data on single devices, for example using NeuroSim framework / Establishment of a Preisach model and SPICE simulations to describe the ferroelectric device, based on available experimental data on HZO capacitors.


Please note that this master's project will unfortunately not be compensated financially.



IBM is committed to diversity at the workplace. With us you will find an open, multicultural environment. Excellent flexible working arrangements enable all genders to strike the desired balance between their professional development and their personal lives.

How to apply

If you are interested in this position, please submit your application below. Please use a few words to present your education, past projects (experimental and programming) and attach your CV and transcripts.


For more information on technical questions please contact
Dr. Laura Begon-Lours ().



[1] T. S. Böscke, J. Müller, D. Bräuhaus, U. Schröder, U. Böttger, Appl. Phys. Lett. 2011, 99, 102903.
[2] J. Müller, T. S. Böscke, D. Bräuhaus, U. Schröder, U. Böttger, J. Sundqvist, P. Kücher, T. Mikolajick, L. Frey, Appl. Phys. Lett. 2011, 99, 112901.
[3] L. Begon-Lours, M. Halter, Y. Popoff, Z. Yu, D. F. Falcone, D. Davila, V. Bragaglia, A. La Porta, D. Jubin, J. Fompeyrine, B. J. Offrein, IEEE J. Electron Devices Soc. 2021, 1.