Computers storing your data all have one thing in common. It’s called CMOS technology: a semiconductor chip that stores and processes information. So far, more computing power has simply implied more and smaller chips. But now that we’re slowly reaching a brick wall when it comes to scaling, engineers have had no choice but to consider alternative concepts to replace CMOS.
Spin waves (SWs) are one such concept, and the SWING project has been aiming to materialise their computing potential. “Our project comes as a response to the limitations of one of the main alternatives to CMOS: optical wave/analogue computing. The latter trades digitalisation for analogue signals and phenomena typical of waves, but it has one major drawback: Miniaturisation is difficult and limited by optical wavelength,” says Riccardo Bertacco, professor of physics at Politecnico Milano and SWING coordinator.
By swapping optical waves for SWs, Bertacco and Marie Skłodowska-Curie fellow Edoardo Albisetti hope to circumvent this problem. As Albisetti points out, “spin waves have a big advantage. They have a wavelength much lower than that of electromagnetic waves, reaching values in the order of tenths of nanometres in the GHz range. This is one order of magnitude lower than optical wavelengths. It allows for the realisation of integrated and CMOS-compatible devices at the submicron scale for wave computing.”
Spin waves through domain walls
SWs are basically propagating disturbances in the alignment of spins in magnetic materials. Besides their inherent advantage, they behave similarly to electromagnetic waves. Their magnetic excitations can be used for computation and memory applications, and Albisetti has already successfully demonstrated a platform using them for analogue computing.
“We’ve had three key achievements,” Albisetti explains. “First, we successfully used a new technique called thermally assisted magnetic scanning probe lithography (tam-SPL) to realise magnonic blocks capable of controlling spin waves. Then, we demonstrated the use of magnetic domain walls (the lines separating two portions of a magnetic film with different uniform magnetisation) as circuits for the propagation and interaction of spin waves. Finally, we tested patterned domain walls of different shapes (linear, convex, concave, etc.) to create our platform for analogue computing.”
Albisetti invented the tam-SPL technique, which is key to the other project achievements, as he spent 6 months of his PhD thesis working with Elisa Riedo at Georgia Tech, United States. As Bertacco underlines: “The Marie Skłodowska-Curie project was designed with the idea of further exploiting this collaboration. When Riedo joined the CUNY Advanced Science Research Centre, we wanted to use its state-of-the-art instrumentation available to further develop tam-SPL. We also aimed to apply it to the proof of concept of new spin wave-based devices for wave computing.”
Eventually, the project’s concept of using domain walls as conduits for the propagation of SWs or as local sources for the generation of wavefronts could be used to build circuits made of such domain walls. These could ultimately act as the equivalent of optical waveguides in integrated optics (resonators, interferometers, etc.), as well as devices for the processing of analogue signals (filters, spectrum analysers, etc.) based on the interference of SW wavefronts.
“Our results open up a range of possibilities which we just started exploring,” Albisetti concludes. “We’ve notably been focusing on two interesting challenges: studying the interaction of spin waves with more complex spin textures and extending the applicability of tam-SPL to different magnetic systems with applications in the field of spintronics.”
Albisetti recently received a European Research Council (ERC) Starting Grant for the B3YOND project which will focus on demonstrating a new nanofabrication concept based on the tam-SPL technique.