The practical implementation of wave ICs faces major hurdles. First, waves often lack data storage mechanisms, contrary to the simple act of storing charges in capacitive or floating-gate memory elements in electronic ICs. Second, device miniaturization is limited by the wave nature itself, as the wavelength sets a lower bound on the footprint. Third, on-chip signal routing and interconnections -- straightforward with metal wiring in charge-based ICs -- are complex in wave-based systems. The first two challenges are best met by spin waves (SWs) -- collective spin oscillations in magnetic materials. SWs offer built-in memory functionality as data can be stored in the magnetization direction, and their short (potentially nanometre-scale) wavelengths hold promise for ICs with dense device integration. These two features are unique to SWs and unmatched by optical or acoustic waves. However, efficient waveguiding for low-loss, long-distance SW routing within an IC remains underdeveloped, especially compared with mature optical counterparts capable of guiding light in optical fibres or on-chip waveguides.

Writing in Nature Materials, Jannis Bensmann and collaborators report a novel fabrication technique using focused ion beam (FIB) implantation to define low-loss SW waveguides in the magnetic insulator yttrium iron garnet (YIG), achieving propagation lengths exceeding 100 μm in sub-micrometre structures -- nearly doubling previous benchmarks. Transporting information carriers -- charge, photon, spin -- often relies on 'index-contrast' guiding, which refers to spatial variation in a relevant material parameter governing transport (Fig. 1). In charge-based interconnects (for example, copper wires), a contrast in electrical conductivity (σ) guides current (Fig. 1a). In optical fibres (Fig. 1b), a contrast in dielectric permittivity (ε) guides photons. Similarly, SWs can be guided by engineering a contrast in magnetic susceptibility (χ) (Fig. 1c). Conventionally, such contrast in SW systems is created by etching away material, but lithographic patterning and etching -- developed for standard complementary metal-oxide-semiconductors -- often damages YIG sidewalls and reduces SW propagation length. The key innovation by Bensmann et al. is to avoid etching entirely. Instead, they implant Si ions into regions flanking the waveguide, amorphizing the YIG and reducing its magnetic susceptibility. This forms a lateral 'index contrast' that confines SWs within the unimplanted core (Fig. 1c), analogous to optical confinement in fibres (Fig. 1b). While the underlying physics differs -- energy reduction via strong demagnetization for SWs versus total internal reflection for photons -- in both, the wave localizes in the region with higher wave index. Electron microscopy confirms that the FIB approach yields smoother sidewalls than etching, thus preserving YIG's intrinsically long SW propagation length.