Scalable Information Processing Architectures with reliable molecular components

The recent discovery of a new molecular material, dynamic molecular switches (DMS), provides a fundamentally important step towards new AI hardware for brain-like In-Memory computing. DMS are able to adapt and change their conductivity in response to electrical stimuli, enabling history-dependent electronic behaviour and mimicking self-learning (or neuromorphic behaviour) needed for In-Memory computing. However, several fundamental challenges must be solved before DMS can be used in practical electronic devices. In particular, it is necessary to understand how these molecular materials behave when incorporated into extremely thin polymer films that can be integrated into electronic circuits.

The SPARC project will investigate how ultrathin DMS films (10-15 nm) can be fabricated and integrated with metal electrodes without compromising their switching and data storage characteristics. The project will study how film thickness influences the switching mechanism, how stable these films remain during metal deposition, and how the interface between the polymer and the metal electrodes affects the electrical behaviour of the devices. This fundamental understanding will form the bases for designing new grid-like architectures, known as crossbar arrays, that can both process and store information. Such systems could form the basis of future energy-efficient hardware for artificial intelligence and data processing.