The limits of conventional electronics
The digitisation of our world – and the relentless downsizing of electronic circuits – has pushed conventional electronics to its limits. In a few decades, the number of transistors on a single chip has grown from a thousand on the earliest integrated circuits to more than two billion. Sizes smaller than those in use today are becoming difficult and expensive to realise, while the relative gains are also decreasing.
At the same time, the search is on for the kind of energy-efficient computational power that can meet the requirements of highly challenging applications: from simulating complex pharmaceutical compounds and quantum systems to machine learning or recognising complex patterns, for instance in facial recognition.
These developments combined are driving us to rethink electronics. New paradigms are needed in electronic devices and architectures.
Facial recognition calls for unconventional electronics
Recognising complex patterns, for example in facial recognition, is a typical example of a potential application that has pushed conventional electronics to its limits. In order to deliver the computational power needed for the highly complex task of recognising individual facial features, we need reconfigurable, neuromorphic materials that can be developed with controllable and dynamically adjustable electric or magnetic properties capable of simulating simulate neurons and synapses. In all of this the human brain is both our inspiration and our model for coming up with new ways of designing powerful, energy-efficient computers.
Proof of principle for an unconventional nanoscale network
Companies, currently are using conventional computers to simulate neural network computing. Our ambition is to realise physical nanomaterial networks that are much more obvious candidates for performing such computations efficiently. In 2015, MESA+ researchers provided proof of principle for an unconventional nanoscale network carrying out logic operations – an important step towards that breakthrough.
The main research directions
Here are two key directions we are pursuing:
- Neuromorphic computing
Neuromorphic computing involves the creation of powerful, energy-efficient computers, based on highly interconnected networks inspired by the (amazingly powerful) human brain.
- New electronic materials
Conventional electronics are based on silicon. In the search for new paradigms and architectures, we are searching for, and experimenting with, numerous exciting alternatives. These include:
- complex oxides
- 2D materials
- piezoelectric materials