ASTRAL · Antiferromagnetic Spin Transport With Relativistic Waves

While photonics has already enabled nearly lossless Tb/s transfer of data using light waves, computing at THz clock rates is the next monumental challenge. Magnonics, which employs spin waves (SWs) instead of light waves, is widely seen as one of the most appealing solutions to this problem, but so far only operates at GHz rate. Interference of SWs enables nearly lossless protocols for logic operations. The large natural nonlinearity of SWs can be used to control their mutual interaction, propagation and manipulation of magnetic bits - altogether facilitating the concepts of transistor and logic-in-memory devices. Since the wavelength of THz SWs is orders of magnitude shorter than that of THz photons, THz SWs offer enviable scalability down to the nanoscale. How to push magnonics into the THz domain and enter the nonlinear regime? How large are the THz nonlinearities? Answering these questions will open up new avenues to scalable technologies for THz and nearly lossless computing.With ASTRAL I want to enter the nonlinear regime of THz magnonics by generating ultrashort large amplitude SW pulses that, similar to femtosecond laser pulses in optics, can zip undisturbed over long distances unlocking the nonlinear regime of interaction between the pulses, other SWs and even macroscopic spin textures. I propose to focus on antiferromagnets, where SW frequencies can easily reach the THz landmark and, similar to light waves in vacuum, follow a linear, so-called relativistic, dispersion relation. Owing to this, a broadband wavepacket of coherent SWs can be compressed to an ultrashort SW pulse - a bunch of few-cycle large-amplitude spin oscillations.To achieve this, ASTRAL will exploit the exclusive ability of light to initiate ultrafast spin dynamics and will attempt to interconvert femtosecond laser pulses into large-amplitude ultrashort SW pulses. Although the idea is fundamental in nature, the ambition is to set the ground for revolutionary new computing technologies.