Lund University

Mott-ferroelectric heteromaterial for sustainable artificial intelligence

  • Discovery
  • Performance
  • Properties
  • Ferroelectrics
  • Hybrid materials
  • Mott insulator
  • transition metal oxides
Academic project
PhD
Open

Artificial intelligence (AI) has emerged as a key computing technology to drive the transition into a carbon-neutral society but its own environmental impact cannot be ignored. Inspired by the brain, analog computing in-memory (ACiM) promises up to 1000x more energy efficient computation by merging computation and data storage. Here, we will combine Mott insulators and ferroelectric films to design the key building block for a high-performing ACiM system – a ferroelectric tunnel junction memristor, performing well beyond the state-of-the-art as well as being reliable enough for implementation into future FTJ-based ACiM systems.

 

Research question

  • Can we modulate the conductance through a ferroelectric thin film more than 1000x?
  • Can a Mott transition provide a more resilient ferroelectric tunneling junction device than a regular semiconductor could?

 

Sustainability aspects

  • The Mott insulator-ferroelectric heteromaterial can enable computing with drastically (1000x) reduced energy consumption compared to conventional electronics.
  • A non-toxic HfZrO4 ferroelectric replaces conventional heavy metal containing (toxic) ferroelectrics.
  • Ultra-thin films (4-10 nm) mean minimal critical raw material consumption.
researcher photo

Lund University

Mattias Borg

Associate Professor

mattias.borg@eit.lth.se

Explore projects under the WISE program

WISE drives the development of future materials science at the international forefront. The research should lead to the development of sustainable and efficient materials to solve some of today's major challenges, primary sustainability. On this page you can read more about our research projects.

Explore projects