Hybrid magnetic nanostructures for nanoscale microwave circuits

Speaker: Prof. Dirk Grundler
Affiliation: Laboratory of Nanoscale Magnetic Materials and Magnonics, Institute of Materials, School of Engineering, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland
Link to MS Teams meeting
Meeting ID: 336 577 067 327, Passcode: BY2Bif
Download Teams | Join on the web

Spin waves (magnons) are collective spin excitations in magnetically ordered materials. In ferro- and ferrimagnetic thin films their typical eigenfrequencies range from roughly 1 to 100 GHz. Thereby, they reside in a frequency regime which is of utmost importance for today’s information and computation technologies. Indeed, spin-wave based computing gains increasing interest [1]. Recently, a nanoscale neural network was proposed which made use of interfering spin waves (magnons) in yttrium iron garnet (YIG) below an array of ferromagnetic nanomagnets [2]. However, the assumed magnon steering and multi-directional interferometry of coherently scattered short-wave magnons had not been experimentally verified. The so-called extinction (on/off) ratio which is relevant for binary 1/0 output operations was not known either. Hence, experimental evidence is urgently needed to substantiate the prospects of unconventional computing schemes reaching from neural networks to in-memory computation.

In the lecture we discuss how to excite and detect on-chip magnon signals with a wavelength down to 50 nm at frequencies of about 25 GHz consistent with the 5G frequency band. Note that the magnon wavelength l is several orders of magnitude smaller than the wavelength of the exciting electromagnetic wave and special transducers have been developed. We will present transducers based on ferromagnetic nanostructures integrated on top of ferrimagnetic YIG. Based on such hybrid structures, we explored the extinction ratios by conducting interference experiments for magnons with l = 69 nm (154 nm). They showed unprecedentedly high values of 26 (±8) dB [31 (±2) dB] [3], obtained over macroscopic propagation lengths of 350 x l. The results are extremely encouraging in view of nanoscale neural networks as proposed in [2]. Our work is supported by SNSF via grant 197360.

References
[1] Chumak A.V. et al. (2022). IEEE Trans. Magn. 58, 1–72.
[2] Papp A., Porod W., Csaba G. (2021). Nat. Commun. 12, 6422.
[3] Watanabe S. et al. (2023). Adv. Mater. 35, 2301087.

Chairmen: Jarosław W. Kłos

Google Calendar

iCal Export