Improving computer architecture with biomimicry

Research, coordinated by the Department of Mechanical and Aerospace Engineering at Sapienza University of Rome, developed a new nanofluidic device modelled on the functioning of the brain's ion channels, which are essential for the propagation and processing of electrical signals, with low energy consumption. The project, funded by the European Research Council (ERC), was published in Nature Communications

The brain is particularly energy-efficient because it handles information by storing it in the same elements that process it. In contrast, for computers, the way calculations are performed currently requires storing and processing information in different parts of the computer, making the process not very energy-efficient. The construction of neuromorphic devices, i.e. inspired by components of the human brain, may represent a major step forward in artificial intelligence applications.

A study coordinated by Alberto Giacomello of the Department of Mechanical and Aerospace Engineering of Sapienza University of Rome, in collaboration with the Groningen Biomolecular Sciences and Biotechnology Institute (The Netherlands), the Sichuan University and Collaborative Innovation Centre (China) and the University of Rome Tor Vergata, takes inspiration from ion channels that control the passage of ions in the brain and are essential for the propagation and processing of electrical signals in the nervous system. This work is part of the HyGate project on hydrophobic gating funded by the European Research Council (ERC).

The work, published in the journal Nature Communications, was based on the engineering of a particular transmembrane channel - a nanopore - that, by exploiting hydrophobicity and the formation of nanobubbles, can reproduce the electrical behaviour of natural ion channels, while functioning through simpler processes.

It was already known that hydrophobic nanopores, which are typically non-conductive, can become conductive when a voltage is applied (electrowetting). This new study develops a quantitative theory for this phenomenon, showing, for the first time, that under certain conditions, it can be used to store information. The theory was tested by experimentally realising the single 'memristive' device and implementing it in demonstrative neuromorphic applications.

'We were interested,' says Alberto Giacomello, 'in designing a voltage-controlled nano-switch, i.e. one in which the change in conduction properties is due to hydrophobic gating, a mechanism by which nanometer-sized bubbles, called nanobubbles, block the passage of ions through channels. We then exploited the phenomenon of electrowetting to cause the hydrophobic nanopore to fill up and thus make it conductive in a controlled manner'.

The individual device was designed from a particular biological nanopore that could be modified and bioengineered according to the criteria suggested by the theory and simulations carried out in this study. This device was realised through collaboration with experimental partners who work with these channels and were able to study their voltage response.

The research thus demonstrated a new way of constructing ionic memristive devices, based on hydrophobic gating, helping to build a more concrete basis for studying the effect of voltage on conduction in hydrophobic nanopores.

Furthermore, new neuromorphic architectures could enhance current algorithms and artificial intelligence systems, making them more energy-efficient and sustainable. These include iontronics, which, using ions instead of electrons as conducting elements, opens up new perspectives also in the medical field.



Hydrophobically gated memristive nanopores for neuromorphic applications - Gonçalo Paulo, Ke Sun, Giovanni di Muccio, Alberto Gubbiotti, Blasco Morozzo della Rocca, Jia Geng, Giovanni Maglia, Mauro Chinappi, Alberto Giacomello - Nature Communications (2023). DOI 10.1038/s41467-023-44019-y


Further Information 

Alberto Giacomello
Department of Mechanical and Aerospace Engineering

Monday, 18 December 2023

© Sapienza Università di Roma - Piazzale Aldo Moro 5, 00185 Roma - (+39) 06 49911 - CF 80209930587 PI 02133771002