Interface-Controlled Spin-Orbit Torques in Rare-Earth Synthetic Ferrimagnets Probed by Sagnac Magneto-Optics and Harmonic Hall Measurements

In the modern world, the demand for data processing and storage is growing at an exponential rate, driven by the internet and artificial intelligence. This surge places a heavy burden on global energy consumption, with projections suggesting that computing could account for half of the world's electricity use within a decade. To address this, scientists are turning to a field called spintronics, which aims to store and process information using the magnetic orientation of electrons rather than just their electric charge. A key goal in this field is to create memory devices that are non-volatile, meaning they hold their data without needing constant power, while also being fast and energy-efficient. One promising method involves using a phenomenon called spin-orbit torque, where an electric current generates a force that can flip the magnetic direction of a material. However, measuring exactly how strong this force is and understanding how to make it more efficient has been a complex challenge, often requiring indirect electrical measurements that can be difficult to interpret.

A team of researchers has now tackled this challenge by studying a specific type of magnetic material known as a synthetic ferrimagnet, which is built from layers of cobalt and gadolinium. These materials are unique because they combine the useful properties of two different types of magnets, allowing their magnetic strength to be finely tuned. The researchers wanted to understand how the arrangement of these layers and the interfaces between them affect the ability of an electric current to generate the torque needed to switch the magnetic state. To do this, they employed two distinct methods to measure the same physical effect. The first was a standard electrical technique that analyzes voltage changes, while the second was a highly sensitive optical method using a specialized interferometer to directly observe the physical tilting of the magnetization caused by the current. By comparing these two approaches, they were able to confirm that their optical method provides a direct and reliable way to quantify the torque without relying on electrical signals that might be influenced by other factors.

The study revealed that the structure of the material plays a critical role in how effectively the torque is generated. When the researchers varied the thickness of the gadolinium layers and changed the order in which the cobalt and gadolinium were stacked, they found that the efficiency of the torque changed significantly. This indicates that the flow of spin currents and the conversion of angular momentum happen most effectively at the specific interfaces between the rare-earth gadolinium and heavy metals like platinum. In one notable finding, they observed a finite torque in a structure where a platinum layer was placed between the gadolinium and cobalt. Conventional theory might suggest that the contributions from the platinum would cancel each other out in such a setup, but the presence of the gadolinium layer proved to be an active participant in generating the torque, highlighting its importance in the process.

Furthermore, the researchers demonstrated that by engineering the interfaces of these materials, they could achieve a state where the magnetization points perpendicular to the surface of the material, even with a relatively thick cobalt layer of two nanometers. This perpendicular orientation is crucial for creating stable, high-density memory bits. They successfully showed that they could switch the magnetic direction of these structures using electric currents with a density of 8 to 10 million amperes per square centimeter. This achievement confirms that synthetic ferrimagnets based on rare-earth elements are a versatile platform for controlling spin-orbit phenomena. The work establishes a clear path for optimizing these materials to create low-power spintronic devices, offering a potential solution to the rising energy demands of future computing systems.