Most of the informations are transmitted either by electric currents or photons. Now a serious contender: spin-wave based devices may just perform some information processing tasks in a lot more efficient and practical way. Spin waves have the potential to make a big hit in future data processing – by combining the best of electronics and photonics in an integral way. Spin-wave-based protocols will shine if complex processing tasks can be made without leaving the spin-wave domain, and construction of devices for non-Boolean computing primitives could be a way toward this goal. The niche for spin-wave based devices is low-power, compact, high-speed signal processing devices, whereas most traditional electronics shows poor performance. A dynamic computing device, one that uses collective excitations of a spin-lattice for information processing is of interest. The use of magnetic materials for high frequency signal processing is not at all new but adding spin-wave interference to the toolbox may open new horizons.
I wish to give a sight on the less-traveled, but, in my opinion, promising route of non-conventional, non-Boolean device structures. The main challenge is the generation of short-wavelength spin waves, which is important in microelectronic applications – the non-localized field distribution of a typical micron-width transmission line couples poorly to submicron-width SWs. Spin-hall effect (SHE)-based manipulation of magnetization and spin-pumping are perhaps the most application-friendly way to generate SWs, especially in magnetic insulators. The dispersion relations (of which material) show that typical frequencies in the 5–100GHz range correspond to wavelengths in the ten nanometers – few micrometers range. This perfectly matches the frequency range and size scale where modern electronic circuits operate. Spin-wave wavelengths go all the way down to the nanometer range, so spin-wave-based devices (unlike photonic structures) can be scaled all the way down to the sizes of end-of-the-roadmap semiconductor devices. Spin waves propagate by two distinct mechanisms: short-wavelength (typically ? <100nm) SWs by the locally strong exchange interactions and long-wavelength SWs by dipolar inter-actions. The propagation length of spin waves for long (something missing here) was considered a problem: due to relatively strong damping and fast decay of the SW amplitude, complex interference patterns were hard to generate in thin-films.
Recently, it became possible to reach SW propagation for at least few-hundred wavelength distances, which is still short compared to optical waves, but sufficient for many device applications. A non-Boolean computing unit avoid the challenges associated with the interconnections of logic blocks through Boolean computing unit. But a non-Boolean computing unit performs a high computing function (i.e. something that is way beyond the function of a simple logic gate) then the input / output bottleneck may be less severe.