Research
The NEO group focuses in the development of novel approaches to the creation of bio-inspired neuromorphic devices using non-conventional materials for electronics, like ferroelectric oxides.
From the diagnosis of an anomaly in an electrocardiogram to the detection of a collision risk by a car, emerging applications of artificial intelligence require always lower latency. Concurrently, deploying artificial neural networks without compromising our goals for greenhouse gas emission requires the adoption of disruptive, energy-efficient solutions. In this context, neuromorphic systems aim at mimicking the brain in the way it processes information, taking advantage of asynchronous, spiking neural networks as well as event-based, adaptative sensors.
The raison d’être of the NEO group is to discover materials capable of supporting advanced neuromorphic systems with bio-inspired functionalities. Its first ambition is to create ferroelectric materials based on hafnium oxide superlattices, which could support the deployment of artificial neural networks in a competitive and sustainable manner. The second ambition of the consists of leveraging the ferroelectric polarization and the superlattice materials properties to translate into hardware - bio-inspired sensors, smart interfaces and neuromorphic systems - the biological, neurologic functions consolidating learning. To meet these goals, we fabricate devices combining (i) strongly-correlated oxide or textured electrodes & (ii) superlattices composed of ferroelectric and anti-ferroelectric or dielectric layers. They we characterize them electrically and under other stimuli such as light and strain.
On the 'Materials' aspect, we strive to control the texture, the domains size, and the ferroelectric domains dynamics of the ferroelectric materials. On the 'Applications' aspect, we investigate the complex mechanisms for the realisation of bio-inspired synapses and neuromorphic sensors with three objectives: to control (i) topological features in (ii) multiple timescales, and (iii) to convert energy from external stimuli into current spikes.
The technologies developed are expected to have a profound impact on different fields, from event-based vision applications (autonomous cars, surveillance, earth and infrastructure monitoring) to robotics (manufacturing, assistance) and healthcare (electrocardiography, electroencephalography).