The Use of Quantum Geometry in Increasing Data Storage Speed and Reducing Energy Costs

Thanks to the emergence of machine learning and artificial intelligence techniques, the world is radically changing with innovative applications, like real-time image processing, autonomous vehicles, Internet of Things, and big data analytics in the healthcare sector.

It has been predicted that in 2020, the global data volume will reach 44 Zettabytes, and that this volume will continue to increase beyond the present capacity of storage and computing devices. By 2030, the consumption of related electricity will also increase as much as 15 times, consuming 8% of the global energy demand. Hence, there is an urgent need to reduce energy consumption and boost the speed of data storage technology.

Now Berkeley scientists, in association with Professor Aaron Lindenberg’s research team from Stanford University, developed a new method for storing data. The Berkeley team was headed by Professor Xiang Zhang, President of The University of Hong Kong (HKU) when he was in Berkeley.

The researchers created odd-numbered layers to slide laterally in relation to even-numbered layers in tungsten ditelluride, which has a thickness of just 3 nm. These atomic layers are arranged such that they represent 1 and 0 for data storage.

The team creatively used quantum geometry—that is, Berry curvature—to read out the information. Hence, a material platform like this works perfectly for memory, with separate “read” and “write” operation. By using the new data storage technique, the energy consumption can be reduced more than 100 times when compared to the conventional technique.

The study is a conceptual advancement for non-volatile storage types and could lead to a technological revolution.

The team demonstrated for the first time that 2D semi-metals, apart from the conventional silicon material, can be used for both data storage and reading. The study was published in the new issue of the Nature Physics journal.

When compared to the prevalent non-volatile (NVW) memory, the novel material platform is predicted to reduce energy cost by three orders of magnitude and boost storage speed by two orders of magnitude. Moreover, the platform can significantly enable the realization of emerging neural network computing and in-memory computing.

The study was motivated by a research work performed by Professor Zhang’s group on “Structural phase transition of single-layer MoTe₂ driven by electrostatic doping,” which was published in the Nature journal in 2017; and also by Lindenberg Lab’s research on “Use of light to control the switch of material properties in topological materials,” which was also published in the same journal in 2019.

Earlier, scientists had observed that in the case of 2D material tungsten ditelluride, especially when the material exists in a topological state, the so-called “Weyl nodes” can be produced by the unique arrangement of atoms in these layers. These Weyl nodes will have special electronic characteristics, like zero resistance conduction.

Such points are believed to have wormhole-like properties, in which electrons move rapidly between reverse surfaces of the material.

In an earlier experiment, the team noted that terahertz radiation pulse can be used to modify the structure of the material, thus rapidly swapping between the non-topological and topological states of the material. This approach effectively turns off the zero-resistance state and subsequently turns it on again.

Zhang’s research team has demonstrated that in the case of 2D materials, their atomic-level thickness considerably decreases the screening effect of the electric field, and its structure is easily influenced by the electric field or electron concentration. Hence, topological materials at 2D limit can enable scientists to turn optical manipulation into electrical control, potentially leading to the development of electronic devices.

In the latest study, the team arranged three atomic layers of tungsten ditelluride metal layers, similar to a nanoscale deck of cards. When they applied a vertical electric field or injected an insignificant amount of carriers into the stack, it made each odd-numbered layer to slide laterally in relation to the even-numbered layers both above and below it.

Through the equivalent electrical and optical characterizations, the researchers noted that this slip is irreversible until another electrical excitation activates the layers to reorganize. Moreover, to read the information and data preserved between such moving atomic layers, the team applied the considerably large “Berry curvature” in the semi-metallic material.

Such a quantum property is similar to a magnetic field, which can guide the propagation of electrons and lead to nonlinear Hall effect. Through this nonlinear Hall effect, the researchers can read the arrangement of the atomic layer without affecting the stacking.

With the help of this quantum property, metal polarization states and different stacks can be suitably differentiated. This latest finding solves the long-standing reading problem in ferroelectric metal caused by their weak polarization.

This discovery also makes ferroelectric metals fascinating in terms of rudimentary physical exploration and also shows that such kinds of materials can have application prospects similar to traditional ferroelectric insulators and semiconductors. To alter the stacking orders, the Van der Waals bond can simply be broken.

Hence, at the theoretical level, the energy consumption is two orders of magnitude lower when compared to the energy consumed by breaking the covalent bond in conventional phase change materials. This also offers a novel platform for developing storage devices that are more energy-efficient and helps move towards a smart and sustainable future.

Source: https://www.hku.hk/