Skip to content

Spectral mapping of thermal transport across SiC-water interfaces

Summary

NEMD simulations of 3C–SiC–water interfaces are combined with spectral mapping: interfacial heat flux is decomposed using a spectral heat flux built from solid–liquid force–velocity correlations, and phonon density of states (DOS) from velocity autocorrelations for interface vs bulk regions. The analysis ties mode-resolved contributions to TBC, crystallographic plane, termination, and wetting.

Motivation in the article is that thermal boundary conductance at semiconductor–water interfaces controls heat removal in power electronics and related technologies, yet macroscopic TBC metrics alone obscure which phonon polarizations and frequencies carry heat across chemically distinct terminations and wetting states.

Methods

  • MD setup: SPC/E water; SHAKE; PPPM electrostatics; MEAM for SiC; LJ Si–O and C–O cross-interactions with ε\(_{Si–O}\) varied by plane and wetting (σ\(_{Si–O}\) fixed at 2.63 Å; C–O parameters as in the companion ACS study).
  • NEMD: Two SiC slabs, ~10 nm slab length, transverse areas 2.62×2.62 nm\(^2\) (100) and 2.78×2.67 nm\(^2\) (111); 6 nm water; periodic x,y; fixed outer z layers; 1.5 nm heat add/remove regions; LAMMPS + VMD visualization.
  • Protocol: Minimization; NVT 300 K for 1 ns (Nosé–Hoover thermostat, 0.1 ps damping time constant); NVE 1 ns; 5 ns heating; production 5 ns with KE, coordinates, and velocities sampled every 25 ps (spectral post-processing uses stored forces/velocities at that cadence; see article for the integration timestep in fs).
  • Pressure / barostat: N/A — NEMD applies fixed heat flux in slab regions rather than a global NPT target; bulk liquid is not barostatted to a stated GPa value in the protocol summarized here (constant-volume segments only).
  • Spectral mapping: Spectral heat flux \(q(\omega)\) from real part of Fourier transform of F\(_{ij}\)·v\(_i\) correlations (solid→liquid forces); DOS \(\propto\) FT of VACF for interface (~3 Å interfacial slice) vs bulk solid atoms; DOS overlap related to TBC and liquid structure.
  • Pressure: NEMD cells are treated at effectively constant volume in the summarized protocol; N/A — NPT barostat / N/A — imposed hydrostatic pressure targets beyond maintaining bulk-like water density between slabs.

Findings

  • DOS at the interface shifts with termination, plane, and wetting; low-frequency modes dominate TBC for C- and Si-terminated (100) and C-terminated (111); Si-terminated (111) shows more high-frequency contribution.
  • Out-of-plane modes contribute strongly to interfacial heat flux; in-plane flux composition is smaller, especially on (111), linked to bonding strength and liquid layering.
  • TBC vs DOS overlap aligns with interfacial liquid structure; interfacial bonding strength alone does not fully predict thermal transport across the studied SiC–water interfaces.
  • Mechanisms / comparisons / sensitivity / limitations / PDF: Mode-resolved heat flux mechanisms are compared to macroscopic TBC from the same NEMD setups, showing how phonon frequency content shifts with wetting and termination. Sampling (25 ps) and classical potentials are limitations for quantitative experiment matching; see discussion in the Int. J. Heat Mass Transfer PDF (pdf_path).

Limitations

Spectral analysis depends on force decomposition, sampling interval (25 ps), and classical potentials; results are for non-reactive SiC–water models.

Relevance to group

Complements 2018gonzalez-valle-acs-thermal-transport-2 with mode-resolved analysis of the same physical system class.

Citations and evidence anchors

  • DOI: 10.1016/j.ijheatmasstransfer.2018.11.101.