Skip to content

Interface Phonon Transmission Behaviors and Their Resultant Influence on Thermal Transport in Cu/Diamond Composites

FieldValue
JournalMaterials Genome Engineering Advances
Published2026-04-28
AuthorsQinghe Zheng, William Yi. Wang, L. Bolzoni, Fei Yang
DOI10.1002/mgea.70069
Citations1

A 50 nm Cr₃C₂ interlayer yielded 83.2 MW/(m²·K) in the model, close to 86 MW/(m²·K) reported experimentally. With a 100 nm interlayer and 65 μm diamond particles, WC gave the largest predicted rise in composite thermal conductivity.

Carbide interlayers address the interface bottleneck

Section titled “Carbide interlayers address the interface bottleneck”

The calculated room-temperature thermal boundary conductance for bare Cu/diamond is 43.4 MW/(m²·K). Copper supplies mechanical properties, while diamond supports a wider phonon spectrum for heat transport. A carbide layer can bridge that spectral gap.

The authors represented a fully reacted interlayer as Cu/carbide/diamond. They also modeled element/carbide/diamond to represent incomplete reactions and residual elements on the copper side. The examined carbides were TiC, ZrC, Cr₃C₂, B₄C, and WC. Their corresponding elements were Ti, Zr, Cr, B, and W.

Each interface and the interlayer were placed in a series thermal resistance model. Thermal boundary conductance describes one interface, while interfacial thermal conductance describes the combined multilayer stack.

First-principles phonons determine transmission

Section titled “First-principles phonons determine transmission”

The authors used density functional theory and density functional perturbation theory with VASP and phonopy. The PAW-PBE calculations used a 500 eV cutoff, with forces below 0.01 eV/Å during relaxation. The phonon Brillouin zone was sampled with a 30 × 30 × 30 q-mesh.

The diffuse mismatch model assumes elastic, diffuse, memoryless scattering with frequency conservation. Landauer integration produced thermal boundary conductance from 0 to 1000 K and spectral conductance at 300 K. A differential effective medium model then predicted composite thermal conductivity against diamond volume fraction.

This is a computational study rather than a new specimen measurement. Experimental interface and composite results were used to validate the calculated trends.

Spectral overlap alone does not rank interfaces

Section titled “Spectral overlap alone does not rank interfaces”

Diamond has acoustic modes exceeding 20 THz, while Cu’s maximum is about 7.3 THz. Carbide dispersions lie between them. Average group velocities are mostly below about 4 km/s for Cu, about 4 to 10 km/s for carbides, and above 5 km/s for diamond.

The calculated interface conductance at 300 K shows why a single interface ranking is unsafe.

CarbideCu/carbide TBC, MW/(m²·K)Element/carbide TBC, MW/(m²·K)Carbide/diamond TBC, MW/(m²·K)Carbide conductivity, W/(m·K)
TiC98.32Ti/TiC: 73.71TiC/diamond: 307.6917
ZrC167.71Zr/ZrC: 80.26ZrC/diamond: 217.2521
Cr₃C₂183.97Cr/Cr₃C₂: 304.25Cr₃C₂/diamond: 253.3319
B₄C76.91B/B₄C: 271.18B₄C/diamond: 347.0267
WC243.09W/WC: 184.78WC/diamond: 154.13120

The calculations identify four coupled controls on conductance. These are phonon group velocity, phonon density-of-states overlap, transmission coefficient, and effective participating frequency range.

WC gives the highest Cu/carbide conductance, 243.09, and the lowest carbide/diamond conductance, 154.13. B₄C gives the reverse ranking: 76.91 and 347.02. A large phonon overlap ratio does not guarantee high conductance. Cu/ZrC has limited group velocity and modest transmission, while W/WC is constrained by the W cutoff. The low WC/diamond value is attributed to WC’s large phonon band gap.

Optical modes and branch conversion matter

Section titled “Optical modes and branch conversion matter”

At Cu/carbide interfaces, conductance rises steeply at low temperature and saturates above 300 K. The transport window is then dominated by low-frequency, fully populated phonons. Carbide/diamond conductance continues rising beyond 300 K because high-frequency modes join the transport spectrum.

Optical phonons contribute nothing at Cu/carbide because Cu has no modes above its 7.3 THz cutoff. W/WC also has zero optical contribution because WC optical branches lie far above W’s acoustic band. For the remaining element/carbide interfaces, optical phonons supply about 50% to 90% of conductance at 300 K.

The free branch conversion model permits complete acoustic mode mixing. The no branch conversion model permits only same-polarization transmission. At 300 K, the paper reports FBC-to-NBC differences of 9 MW/(m²·K) for Cu/WC and about 50 MW/(m²·K) for its Ti/diamond example.

Neglecting optical phonons tends to underestimate conductance. Assuming complete branch conversion can overestimate it when mode mixing is limited.

Interlayer thickness changes the preferred carbide

Section titled “Interlayer thickness changes the preferred carbide”

For thicknesses below about 150 nm, most carbide interlayers increase overall interfacial thermal conductance. Beyond about 164 nm, TiC, ZrC, and Cr₃C₂ cause marked reductions because their intrinsic conductivities are only 17, 21, and 19 W/(m·K).

B₄C and WC retain more stable conductance at large thickness. Their intrinsic conductivities are 67 and 120 W/(m·K). When thickness can be held below 100 nm, WC, ZrC, and Cr₃C₂ give higher conductance than B₄C and TiC.

Residual B or Cr can add a phonon coupling channel between Cu and carbide. No corresponding benefit was found for residual Ti, Zr, or W.

The series model omits non-equilibrium spectral evolution and incomplete conversion in the ultra-thin limit. Its predictions should not be extrapolated to extremely small thicknesses.

The differential effective medium model used Cu and diamond conductivities of 398 and 1800 W/(m·K). It also used a 65 μm diamond particle and a uniform 100 nm carbide layer.

Every carbide produced a composite thermal conductivity above the bare Cu/diamond case. Conductivity rose monotonically with diamond volume fraction, and WC gave the largest predicted gain.

Two interface predictions were checked against experiments. Cu/100 nm TiC/diamond gives 51.8 MW/(m²·K), compared with 76 MW/(m²·K) experimentally. Cu/50 nm Cr₃C₂/diamond gives 83.2 MW/(m²·K), compared with 86 MW/(m²·K) experimentally.

The calculations assume perfectly bonded interfaces, smooth and clean surfaces, and continuous carbide layers. The authors describe both predictions as good agreement, although the TiC value is below its experimental result. Defects, roughness, disorder, and interfacial mixing should lower real interface conductance.

Interface design must account for the full stack

Section titled “Interface design must account for the full stack”

The results rule out choosing a carbide from one favorable interface value. WC performs best on the Cu side but worst on the diamond side. B₄C shows the opposite pattern.

WC gives the largest overall gain under the model’s 100 nm interlayer condition. If thickness control is poor, B₄C or WC better retains conductance at larger thicknesses. Residual B or Cr may provide an extra coupling route, while residual Ti, Zr, and W should be minimized.

Fabrication tests should specify the carbide phase, interlayer thickness, residual element, and diamond fraction. Each variable changes the predicted composite thermal conductivity. The paper supports carbide interlayers as a route to higher Cu/diamond conductivity, but its interface values remain upper-bound estimates.

For the heat-spreading application, the relevant material family is MPCVD-grown diamond used as reinforcement or a thermal spreading element. The diamond materials catalog lists single-crystal and polycrystalline diamond in plates and wafers up to 125 mm. Interface-specific carbide layers and multilayer builds should be defined through 6CCVD’s engineering support service.

View Original Abstract

ABSTRACT In high‐power electronic packaging, Cu/diamond composites are attractive heat‐spreader candidates owing to diamond’s ultrahigh thermal conductivity. However, their practical thermal performance is often limited by the intrinsically low thermal boundary conductance (TBC) at Cu/diamond interfaces, which can be improved by introducing a manufacturable carbide interlayer. To elucidate the phonon‐mediated interfacial thermal transport mechanisms affected by interlayer structures, the present work performs a systematic analysis based on density functional theory (DFT) and the diffuse mismatch model (DMM). The results reveal that TBC is jointly determined by phonon group velocity, PDOS overlap, transmission probability, and the effective participating frequency range, providing qualitative criteria for selecting carbide interlayers. The analysis further indicates that neglecting optical-phonon contributions and restricting phonon branch conversion can substantially underestimate TBC. The interfacial phonon transport is highly sensitive to interlayer thickness and composition: When thickness control is challenging, high‐thermal-conductivity carbides (B 4 C and WC) better retain interfacial thermal conductance (ITC) at large thickness. Residual B or Cr can provide an additional phonon‐coupling pathway and further improve ITC. This work deepens the understanding of phonon‐mediated interfacial thermal transport and provides mechanistic guidance for interlayer design in Cu/diamond heat spreaders.