The 81st JSAP Autumn Meeting, 2020

Presentation information

Oral presentation

22 Joint Session M "Phonon Engineering" » 22.1 Joint Session M "Phonon Engineering"

[8p-Z09-1~18] 22.1 Joint Session M "Phonon Engineering"

Tue. Sep 8, 2020 1:00 PM - 6:30 PM Z09

Takanobu Watanabe(Waseda Univ.), Takahiro Yamamoto(Tokyo Univ. of Sci.), Takao Mori(NIMS), Koji Miyazaki(Kyushu Inst. of Tech.)

1:45 PM - 2:00 PM

[8p-Z09-3] Semi-analytical method for extrapolating phonon mean free path distribution for thin dielectric films

〇(M2)Boris Yordanov Nedyalkov1, Roman Anufriev1, Jose Ordonez-Miranda2, Masahiro Nomura1 (1.IIS, Tokyo Univ., 2.Institut Prime, CNRS)

Keywords:phonon mean free path, suspended membrane, phononic crystal

Thin suspended films have many applications in photonic, electronic, and phononic systems. One very important application is for the making of phononic crystals (PnCs), used to manipulate the phonon behavior in a given system. PnCs work by creating geometries that enhance phonon boundary scattering. This impacts strongly the phonon mean free paths (MFP) in the film. Until now, there has been no experimental way to determine the phonon MFP distribution in thin films. Thus as a modification of a method proposed by Hao et al. [1], here we create a general method to determine the phonon MFP distribution in any thin film.
There are certain accumulated data in the literature on PnCs, from which we can readily extract the phonon MFP distributions. As such, we apply the model to crystalline Si membranes and poly-crystalline Si0.5Ge0.5 membranes from the PnC studies done by Anufriev et. al [2] and Nomura et. al [3]. Here F represents the cumulative function of the thermal conductivity (kBT). We see from Fig.1 that in both Si and SiGe thin films the average MFP is about ≈120nm. With the majority of the contribution to kBT is between 20-250 nm for Si and 5-250 nm for SiGe. Also, above around 300 nm for Si and 400 nm for SiGe the phonon MFP contribution to kBT becomes negligible. There is some uncertainty in the distribution, which is due to the amount of experimental data points.