Memory is a crucial function in both computers and human brains. However, the traditional approach in computers, which involves separating memory units and central processing units, is considered inefficient. This separation, known as the von Neumann bottleneck, leads to increased energy consumption. Researchers have been exploring the concept of memristors – electronic components that can both compute and store data, mimicking the brain’s way of processing information.
Ambitious Goals in Nanofluidic Memristive Devices
Aleksandra Radenovic from EPFL’s School of Engineering envisioned a more ambitious approach – a nanofluidic memristive device that relies on ions instead of electrons and their counterparts. This approach closely resembles the brain’s energy-efficient way of information processing. By using ions, similar to those found in living organisms, Radenovic aims to create a nanofluidic neural network that takes advantage of changes in ion concentrations.
The Laboratory of Nanoscale Biology (LBEN) at EPFL, under the leadership of Radenovic, has made significant progress in the development of nanofluidic memristors. By fabricating a new nanofluidic device for memory applications, LBEN researchers have achieved a device that is more scalable and performant than previous attempts. This innovation enables the connection of artificial synapses and paves the way for brain-inspired liquid hardware.
Mechanism of Action
The nanofluidic memristors developed by LBEN can switch between conductance states through the manipulation of an applied voltage. Unlike traditional electronic memristors that use electrons and holes, LBEN’s memristor can leverage various ions like potassium, sodium, and calcium. By immersing the device in an electrolyte water solution, the researchers can control the memory of the device by changing the ions used, affecting how it switches on and off or how much memory it stores.
The nanofluidic device is fabricated on a chip at EPFL’s Center of MicroNanoTechnology, creating a nanopore at the center of a silicon nitride membrane. Palladium and graphite layers are added to create nano-channels for ions. As the ions flow through these channels and converge at the pore, a blister forms between the chip surface and the graphite, leading to a change in conductivity and memory state of the device. The ability to retain memory without a current mirrors biological processes in the brain.
Real-time Observation and Future Plans
LBEN researchers, including Ph.D. students Yunfei Teng and Nathan Ronceray, have observed the memory action of highly asymmetric channels (HACs) in real-time, a significant achievement in the field. Collaborating with other research groups, the team has also demonstrated digital logic operations based on synapse-like ionic devices. The next step involves connecting a network of HACs with water channels to develop fully liquid circuits, with potential applications in brain-computer interfaces and neuromedicine. The research, recently published in Nature Electronics, presents a new frontier in memory technology with nanofluidic memristors.
The development of nanofluidic memristors represents a groundbreaking advancement in memory technology, offering scalability, efficiency, and potential applications in various fields. The fusion of nanotechnology and neuroscience opens doors to innovative and energy-efficient computing solutions, echoing the intricate workings of the human brain.
Leave a Reply