Beyond Moore’s Law: Innovations in solid state physics include ultra-thin 2D materials and more

In the relentless pursuit of energy-efficient computing, new devices being developed at UC Santa Barbara promise improvements in information processing and data storage.

Researchers in the lab of Kaustav Banerjee, Professor of Electrical and Computer Engineering, have published a new paper describing several of these devices in the journal: “Quantum-engineered devices based on 2D materials for next-generation information processing and storage”. Advanced Materials. First author is the recently graduated Arnab Pal.

Each device aims to address the challenges associated with traditional computing in new ways. All four operate at very low voltages and feature low leakage, unlike the traditional metal-oxide-semiconductor field-effect transistors (MOSFETs) found in smartphones, which consume power even when they are off. However, because they rely on processing steps similar to those used to manufacture MOSFETs, the new devices could be manufactured at scale using existing industry-standard semiconductor manufacturing processes.

The most promising of the two information-handling devices, Banerjee says, is the spin-based field-effect transistor, or spin-FET, which harnesses the magnetic moment — or spin — of the electrons that power the device. The materials belong to the group of transition metal dichalcogenides, which are based on transition metals.

Unlike the spin FET, the charge-based field effect transistor, or TFET, works by taking advantage of the quantum mechanical nature of electrons. As a result of a phenomenon known as wavefunction penetration, the electrons are able to tunnel through a thin electrical barrier instead of flowing across it as they do in a conventional transistor. TFETs can also operate at lower voltages, consume less power, and generate less heat. TFETs made from 2D materials perform better due to their thinner and more controllable electronic tunneling barrier, which both improves electron flow and allows the device to work with greater precision.

In order for data to be stored securely, a device’s hard drive must be programmed to retain storage even when the device is powered off. To do this with a regular MOSFET, Pal said, “You have the source (electrons) and the drain (where the electrons are collected) and then the channel between them that controls the flow of those carriers. The channel either acts as a barrier so the electrons can’t pass – the off state – or is very conductive so they can pass, ie the on state. In order to allow flow in a normal n-type device generate, requires the application of a positive gate bias – which attracts the negatively charged electrons from the source through the channel.

Meanwhile, the charge-based floating gate field effect transistor (FG-FET) for information storage works on a similar principle to a MOSFET, but has not just one gate electrode, but two. The additional electrode is called the floating gate. When a MOSFET is unprogrammed, there is no extra charge on the floating gate. However, with each programming operation, a very high voltage is applied to the gate, which pulls many charge carriers (electrons) out of the channel and deposits them where they are trapped. This accumulated negative gate charge makes the device difficult to turn on, thereby programming it to be off.

According to the researchers, there are challenges with this method. One of them, Pal said, “draws a lot of charge to the floating gate, which requires a lot of current.” Banerjee also added: “In order for these electrons to get to the floating gate to program the device, they have to tunnel through a dielectric layer.

A range of charge-based and non-charge-based devices built with 2D materials can enable next-generation information processing and storage technologies.

Another problem is that when many FG-FETs are placed close together – and, Pal said, “we want to put as many on the chip as possible, for example to increase the capacity of a USB stick” – the stored charges build up devices interact with each other and affect the neighboring device. The use of ultra-thin 2D materials minimizes this interaction while increasing control over each device, resulting in high performance even when more devices are more densely concentrated on the chip.

Another approach to information storage is the use of magnetic tunnel junctions (MTJs), which use electron spin to store data. An MTJ consists of two magnetic layers separated by a thin insulating layer; the relative orientation of the magnetic moments of the layers determines the resistance of the device. However, similar to FG-FETs, MTJs face challenges in terms of power consumption, stability, and scalability. Again, ultra-thin 2D materials offer a potential solution by reducing interactions between neighboring MTJs, enabling efficient, high-density data storage.

While devices based on 2D materials can achieve improvements in energy and space efficiency over those made from traditional materials, extraordinary improvements can only be achieved through a change in computer architecture. Enter the radical new architecture known as quantum computing. Based on the quantum superposition of quantum bits, called qubits, quantum computing generates parallel calculations to deliver massive performance gains in terms of speed and efficiency for selected computing tasks.

The unique structural and electromagnetic properties of 2D materials, which allow for efficiency gains for the more conventional charge-based qubits, also enable the efficient design of several other newer types of qubits, referred to as spin, valley, and spin valley qubits.

In a spin qubit, the property of state (on/off) is defined by the electron spin or quantum state of the qubit, which is always either spin-up or spin-down. State is changed by flipping the spin from top to bottom or vice versa to change the state of the qubit and aid in qubit operation.

The Valley qubit works a little differently. Its state is determined by the electron’s momentum and not by its spin. The change in electron momentum is translated into a change in qubit state, helping to realize qubit operation.

Finally, the properties of 2D materials can also help to realize the third type of qubit, the spin valley qubit, whose states are defined by both the momentum and the spin of the electrons. Because spin valley qubits couple these two degrees of freedom, they are believed to be more resilient to decoherence, allowing longer and more complex quantum computations to be performed before quantum entanglement is lost.

“The emerging devices enabled by the unique properties of 2D materials hold the promise of low-power, high-performance computing and storage,” said Pal, “enabling Beyond-Moore integration and spurring new explorations in solid-state physics and its applications.”