Advancing Memory Devices and Neuromorphic Computing Chips with Protons

A team of researchers led by KAUST has made a breakthrough in the field of memory devices and neuromorphic computing chips. Their discovery revolves around a proton-mediated method that induces multiple phase transitions in ferroelectric materials, opening up possibilities for the development of high-performance, low-power memory devices and brain-inspired computing chips.

Ferroelectric materials, such as indium selenide, possess inherent polarization and can switch polarity when exposed to an electric field. This characteristic makes them attractive for memory technology advancement. The resulting memory devices demonstrate exceptional read/write endurance and write speeds while operating at low voltages. However, their storage capacity has been limited due to current techniques being able to induce only a few ferroelectric phases, with significant challenges in recording these phases, as explained by Xin He, co-leader of the study conducted under the guidance of Fei Xue and Xixiang Zhang.

The team’s innovative approach relies on protonation, achieved by treating indium selenide with protons, to generate numerous ferroelectric phases. To evaluate the material, they incorporated it into a transistor composed of a silicon-supported stacked heterostructure. The heterostructure consisted of an aluminum oxide insulating sheet sandwiched between a platinum layer at the bottom and porous silica at the top. The platinum layer served as electrodes for the applied voltage, while the porous silica acted as an electrolyte, supplying protons to the ferroelectric film.

Through controlled application of voltage, the researchers were able to gradually introduce or remove protons from the ferroelectric film. This process resulted in the reversible formation of multiple ferroelectric phases with varying degrees of protonation, a crucial aspect for implementing memory devices with substantial storage capacity.

The degree of protonation depended on the applied voltage, with higher positive voltages leading to increased protonation, while higher negative voltages significantly reduced protonation levels. The proximity of the film layer to silica also played a role, with protonation reaching maximum values in the bottom layer and decreasing in subsequent layers until it reached a minimum in the top layer.

Interestingly, the proton-induced ferroelectric phases returned to their initial state when the voltage was turned off, a phenomenon attributed to proton diffusion from the material into the silica.

To achieve low-power memory devices, the team focused on creating a film with a smooth and continuous interface with silica, resulting in a device with high proton-injection efficiency operating under 0.4 volts. Reducing the operating voltage posed a significant challenge, but the team discovered that adjusting the proton-injection efficiency over the interface could effectively control the operating voltages.

“Our aim is to develop ferroelectric neuromorphic computing chips that consume less energy and operate at faster speeds,” emphasizes Xue, highlighting the team’s commitment to advancing this field.

Reference: “Proton-mediated reversible switching of metastable ferroelectric phases with low operation voltages” by Xin He, Yinchang Ma, Chenhui Zhang, Aiping Fu, Weijin Hu, Yang Xu, Bin Yu, Kai Liu, Hua Wang, Xixiang Zhang and Fei Xue, 24 May 2023, Science Advances.
DOI: 10.1126/sciadv.adg4561

Frequently Asked Questions (FAQs) about ferroelectric neuromorphic computing chips

More about ferroelectric neuromorphic computing chips

• “Proton-mediated reversible switching of metastable ferroelectric phases with low operation voltages” – Science Advances (DOI: 10.1126/sciadv.adg4561)