Integrating storage, memory and processing in one unit is considered to be one of the greatest challenges in computer architecture. This would make computers more energy efficient and faster. University of Groningen physicists have taken a huge step towards this goal by incorporating a niobium doped strontium titanate (SrTiO3) semiconductor with ferromagnetic cobalt. At the interface, this develops a spin-memristor with storage abilities, making way for neuromorphic computing architectures. The results were featured on 22 January in Scientific Reports.
The device created by the physicists, incorporates the memristor effect of semiconductors with a spin-based phenomenon known as tunneling anisotropic magnetoresistance (TAMR) and runs at room temperature. The SrTiO3 semiconductor comprises of a non-volatile variable resistance when interfaced with cobalt: an electric field capable of being used for changing it from low to high resistance and back. This is called the electroresistance effect.
Additionally, when a magnetic field was applied throughout the same interface, in and out of the plane of the cobalt, this presented a tunablity of the TAMR spin voltage by 1.2 mV. This coexistence of both a bigger change in the value of TAMR and electroresistance across the same device at room temperature, has not been established in other material systems.
“This means we can store additional information in a non-volatile way in the memristor, thus creating a very simple and elegant integrated spin-memristor device that operates at room temperature”, explains Professor of Spintronics of Functional Materials Tamalika Banerjee. Attempts for incorporating spin-based storage, computing and memory have been so far been hindered by a complex architecture besides other factors.
The interface between the semiconductor and cobalt is the key to the success of the Banerjee group device. “We have shown that a one-nanometer thick insulating layer of aluminum oxide makes the TAMR effect disappear”, says Banerjee. Engineering the interface took quiet some time. They did so by altering the niobium doping of the semiconductor and thus the prospective landscape at the interface. The same coexistence cannot be realized with silicon as a semiconductor:
You need the heavy atoms in SrTiO3 for the spin orbit coupling at the interface that is responsible for the large TAMR effect at room temperature.
It is possible to use these devices in a brain-like computer architecture. They would behave like the synapses responsible for connecting the neurons. The synapse reacts to an external stimulus, however this response will also rely on the synapse's memory of earlier stimuli.
“We are now considering how to create a bio-inspired computer architecture based on our discovery.” Such a system would move away from the standard Von Neumann architecture. The huge advantage is that it is likely to use less energy, thus generating less heat. “This will be useful for the "Internet of Things", where connecting different devices and networks generates unsustainable amounts of heat.”
The physics of what precisely takes place at the interface of the strontium semiconductor and cobalt is complex and more work will have to be executed in order to understand it. Banerjee: “Once we understand it better, we will be able to improve the performance of the system. We are currently working on that. But it works well as it is, so we are also thinking of building a more complex system with such spin-memristors to test actual algorithms for specific cognition capabilities of the human brain.” Banerjee's device is comparatively simple. Scaling it up to a complete computing architecture is the next huge step.
How to incorporate these devices in a parallel computing architecture that imitates the working of the brain is indeed a question that interests Banerjee. “Our brain is a fantastic computer, in the sense that it can process vast amounts of information in parallel with an energy efficiency that is far superior to that of a supercomputer.” The findings of Banerjee's team could lead to the development of new architectures ideal for brain-inspired computing.