Both human brains and computers depend on memory, or the capacity to retain information in an easily accessible way. The primary distinction is that computer information processing requires the back-and-forth movement of data between a memory unit and a central processing unit (CPU). In contrast, brain information processing involves direct computations of stored data. The von Neumann bottleneck, an inefficient separation, is a factor increasing computer energy costs.
Artificial nanofluidic synapses can store computational memory. Image Credit: EPFL 2024 / Andras Kis
Since the 1970s, researchers have been working on the concept of a memristor (memory resistor), an electronic component that can compute and store data as a synapse does.
Aleksandra Radenovic, of EPFL’s School of Engineering’s Laboratory of Nanoscale Biology (LBEN), has set her sights on something even more ambitious: a functioning nanofluidic memristive device based on ions rather than electrons and their oppositely charged counterparts (holes). Such an approach would more closely resemble the brain’s considerably more energy-efficient, method of processing information.
Memristors have already been used to build electronic neural networks, but our goal is to build a nanofluidic neural network that takes advantage of changes in ion concentrations, similar to living organism.
Aleksandra Radenovic, Professor, EPFL
LBEN postdoctoral researcher Théo Emmerich added, “We have fabricated a new nanofluidic device for memory applications that is significantly more scalable and much more performant than previous attempts. This has enabled us, for the very first time, to connect two such ‘artificial synapses’, paving the way for the design of brain-inspired liquid hardware.”
Nature Electronics published the study.
Just Add Water
Memristors can switch between two conductance states—on and off—by manipulating the applied voltage. Electronic memristors process digital information using electrons and holes, but LBEN’s memristor can utilize a variety of ions. For their investigation, the researchers submerged their device in an electrolyte water solution containing potassium ions, although additional ions such as sodium and calcium can be employed.
“We can tune the memory of our device by changing the ions we use, which affects how it switches from on to off, or how much memory it stores,” Emmerich added.
The device was created on a chip at EPFL’s MicroNanoTechnology core by constructing a nanopore in the center of a silicon nitride membrane. To generate ion nanochannels, the researchers added palladium and graphite layers. As a current travels through the chip, the ions percolate down the channels and converge at the pore, where their pressure forms a blister between the chip’s surface and the graphite.
The blister forces the graphite layer up, making the device more conductive and switching the memory state to ‘on.’. The device ‘remembers’ its prior condition since the graphite layer remains lifted even when no current is applied. A negative voltage returns the layers to contact, restoring the memory to the ‘off’ state.
Ion channels in the brain undergo structural changes inside a synapse, so this also mimics biology.
Yunfei Teng, Ph.D. Student, LBEN
Teng worked on fabricating the devices, which were named highly asymmetric channels (HACs) due to the shape of the ion flow toward the central pores.
Nathan Ronceray, an LBEN PhD student, adds that the team’s real-time monitoring of the HAC’s memory activity is an entirely novel feat in this field.
Because we were dealing with a completely new memory phenomenon, we built a microscope to watch it in action.
Nathan Ronceray, Ph.D. Student, LBEN
The researchers, working with Riccardo Chiesa and Edoardo Lopriore of the Laboratory of Nanoscale Electronics and Structures, led by Andras Kis, were able to link two HACs with an electrode to build an ion-flow logic circuit. This breakthrough marks the first demonstration of digital logic operations using synapse-like ionic devices.
However, the researchers have not done this; their next objective is to connect a network of HACs to water channels to build fully liquid circuits. In addition to offering an in-built cooling mechanism, water would make it easier to construct biocompatible devices with potential uses in brain-computer interfaces or neuromedicine.
Journal Reference:
Emmerich, T., et. al. (2024) Nanofluidic logic with mechano–ionic memristive switches. Nature Electronics. doi:10.1038/s41928-024-01137-9