Memristors, a term which combines the words memory and resistor, refer to the class of innovative electrical circuitry which support resistors due to their charge-recalling capabilities and non-volatile properties.
Now, researchers from the University of Michigan (U-M) are developing a memristor which is able to mimic the behaviors of synapses.
They join the company of other scientists developing memristors or artificial intelligence systems modeled around replicating neural or brain function.
Synaptic Gaps Offer a Model
To achieve their results, the team employed a two-dimensional layering technique with the very promising semiconductor known as molybdenum disulfide. They then introduced the step of arranging lithium ions between the gaps existing between the layers.
The lithium ions are easy to rearrange within the layer by sliding them with an electric field. This changes the size of the regions that conduct electricity little by little and thereby controls the device’s conductance.
“Because we change the ‘bulk’ properties of the film, the conductance change is much more gradual and much more controllable,” U-M Electrical and Computer Engineering Professor and study senior author Wei Lu said.
The result of the method is "facilitating controlled ion migration and efficient ionic coupling among devices". The relatively simplified and scale-down device offered the team a way to get around the issue of relying on overly complicated circuitry to achieve the same results.
This innovation in efficiency is at the heart of the team's collective success in the study. As Lu explains:
"Neuroscientists have argued that competition and cooperation behaviors among synapses are very important. Our new memristive devices allow us to implement a faithful model of these behaviors in a solid-state system."
Linking Memristors to Create a 'Synaptic Network'
Another area in which the team's work has an impact is related to synaptic cooperation. Synapses in the human body naturally strengthen and weaken over time, and in the process release proteins, known as plasticity-related proteins.
As the work is based on mimicking several aspects of synaptic behavior, this offered a new area of investigation for the researchers.
To test this theory within their own context, the team constructed a network of memristors which included four devices and found that although the signal strength varied, that ions--like proteins--could successfully be shared among the devices, an improvement in efficiency.
Evolving the Research
The next steps for the team include expanding the scope of the research to discover applications in the area of neuromorphic computing, an exciting research area with major developments in this year alone, both in the areas of supercomputers and superconducting switches.
This study represents a perfect coupling of neuroscience concepts and electrical engineering, with the theories of one directly informing the other. As the memristor's capabilities continue to evolve, so too will the important links between these two disciplines.
Details about the study appear in a paper, titled "Ionic modulation and ionic coupling effects in MoS2 devices for neuromorphic computing", which was published on December 17th in the Nature Materials journal.