Numerical modeling of in-plane thermal conductivity measurement methods based on a suspended membrane setup

TL;DR

This study uses numerical modeling to evaluate and improve ac methods for measuring in-plane thermal conductivity of thin films on suspended membranes, proposing a new data reduction technique that enhances accuracy and broadens applicability.

Contribution

It introduces a new data reduction methodology based on phase shift and amplitude analysis, improving measurement robustness and expanding the valid frequency range for in-plane thermal conductivity assessment.

Findings

The frequency window narrows with higher in-plane diffusivity using the MPS method.

The new methodology reduces sensitivity to diffusivity variations, allowing broader applicability.

The proposed method achieves less than 5% uncertainty for typical thin film conductivities.

Abstract

A numerical modeling study based on 3D finite element method (FEM) simulation and 1D analytical solutions has been carried out to evaluate the capabilities of two ac methods for measuring in-plane thermal conductivity of thin film deposited on the back of a suspended SiNx membrane setup. Two parallel metal strips are present on the top of the dielectric membrane. One strip (S1) serves as both heater and thermometer, while another one (S2) acts as thermometer only. For a modified phase shift (MPS) method, it is crucial to extract the in-plane thermal diffusivity from the phase shift of the temperature oscillation on S2. It was found that the frequency window for carrying out the data fitting became narrower as the in-plane thermal diffusivity of the composite membrane (${\alpha_{\parallel ,C}}$) increased, primarily due to the failure of the semi-infinite width assumption in the low frequency region. To ensure the validity of the method, the upper limit of ${\alpha_{\parallel ,C}}$ should not exceed ~1.8$ \times $10-5 m2 s-1 for the specific membrane dimension under consideration (1$\times $1 mm2). On the other hand, inspired by a modified Angstrom method proposed by Zhu recently, we suggest a new data reduction methodology which takes advantage of the phase shift on both S1 and S2 as well as the amplitude on S1. Based on the simulation results, it is expected that the non-ideality associated with the three observables may be at least partially cancelled out.Therefore, the frequency window selection for carrying out the data fitting is not sensitive to the magnitude of ${\alpha_{\parallel ,C}}$. For typical specimen films whose in-plane thermal conductivity ranges from 0.84 W m-1 K-1 to 50 W m-1 K-1, the method proposed here yields a theoretical measurement uncertainty of less than 5%.