A group of physicists from several countries has made a significant impact in the field of brain-inspired computers. A team of theoretical physicists from Utrecht University in the Netherlands and experimental physicists from Sogang University in South Korea have constructed an artificial synapse using a combination of water and salt. This novel computing system utilises fluidic ion channels, similar to the neurons found in human brains. The “neuromorphic” technique holds the potential to be far more energy-efficient compared to conventional computers.
“We have demonstrated, for the first time, that it is possible to create artificial synapses that can process complex information using water and salt, in addition to solid materials,” states Tim Kamsma, the lead author of the study and a doctoral candidate at the Institute for Theoretical Physics and the Mathematical Institute of Utrecht University, in a press release. “We are successfully reproducing the behaviour of neurons using a system that utilises the same medium as the brain.”
The study, published in the journal Proceedings of the National Academy of Sciences, attributes the success to the distinctive structure of the device’s microchannels. Envision a minuscule, conical passage, narrower than the diameter of a human hair, containing a hard crystal lattice composed of charged nanospheres made of silicon dioxide. The interstices among these spheres form an intricate lattice of nanochannels, facilitating the passage of ions, which are electrically charged particles. This process bears resemblance to the transmission of electrical signals among neurons in the brain.
An illustrative depiction of the synapse. The synapse is composed of colloidal spheres connected by nano-channels. The source of this information is Utrecht University.
Ion current rectification is a phenomenon shown by the device when a voltage is placed across the channel, resulting in a preferential conduction of current in one direction due to the difference in ion concentration from one end to the other. However, the most fascinating aspect is that the level of rectification can be modified by adjusting the voltage, enabling the device to operate as a “memristor,” which is essentially a resistor with the ability to retain information.
Scientists devised a theoretical framework to precisely comprehend the mechanism of ion transport in these channels. It was found that the memristor’s “memory” timescale is unexpectedly influenced by the time it takes for ions to diffuse, despite the fact that the process is driven by voltage. By adjusting the channel length during manufacture, the memory retention duration can be precisely controlled.
In order to evaluate the capabilities of their device, the team employed it to address a well-known neuromorphic computing challenge: the recognition of handwritten digits. The user converted pixel pictures of numbers into voltage pulses and inputted them into the fluidic memristor. The researchers were impressed by the device’s ability to convert temporal voltage signals into distinct output current fingerprints for each digit. By inputting these results into a basic neural network, the numbers were accurately classified, achieving a level of performance comparable to other advanced neuromorphic systems.
“According to Kamsma, this indicates the potential to customise channels in order to store and handle information for different lengths of time, similar to the synaptic mechanisms found in our brains.”
The iontronic technique is particularly fascinating since it closely emulates the brain’s inherent computing operations. Neurons communicate by means of ions passing through channels filled with water, and the strength of these connections can change over time, a phenomenon referred to as synaptic plasticity. The fluidic memristor successfully captures the dynamic behaviour that traditional electronic memristors have had difficulty adequately replicating.
In the future, scientists anticipate that this novel gadget could serve as the fundamental component for sophisticated neuromorphic chips. It is possible to connect several channels, similar to the neurons in the brain, in order to enhance the capabilities of computing. Due to nanoscale production, these devices would possess exceptional energy efficiency in contrast to the power-hungry processors currently available.
“According to Kamsma, this is a significant step forward in creating computers that can imitate the communication patterns of the human brain and use the same medium.” “This may eventually lead to the development of computing systems that more accurately mimic the exceptional abilities of the human brain.”
There are still some unresolved issues in the performance and underlying theory of the memristor. However, this work unquestionably signifies a thrilling advancement in the realm of neuromorphic engineering. By adopting the brain’s “wetware” methodology, which involves ions, water, and dynamic connections, we may be on the verge of a groundbreaking revolution in computing. The potential of artificial intelligence in the future may surpass our current expectations, exhibiting a remarkable level of adaptability.
