Mitigating the Thermal Bottleneck in Polycrystalline Diamond Films by Gradient ICP Etching of the Nucleation Layer
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Materials |
| Published | 2026-02-15 |
| Authors | Yuhan Lv, Lei Zhao, Xiangbing Wang, Zhanpeng Sheng, Rongchen Zhang |
| DOI | 10.3390/ma19040759 |
Selective ICP etching removed the defect-rich nucleation layer from freestanding polycrystalline diamond. The reported in-plane thermal conductivity rose from 1549.9097 to 1792.0250 W·m⁻¹·K⁻¹ after 5 h. The net gain was about 15.4%, although the film became 112.8 µm thick and the etched surface roughness exceeded the as-grown value.
Film preparation
Section titled “Film preparation”Polycrystalline diamond was deposited on 2-inch (100) single-crystal silicon wafers in a 6 kW MPCVD system. The growth used 1 vol% methane, a substrate temperature of 850 to 900 °C, 4500 W microwave power, and 25 kPa chamber pressure. The growth produced a 145.2 µm film. After hydrofluoric acid removed the silicon, laser cutting produced freestanding discs with a diameter of 25.4 mm. Before deposition, the wafers were ultrasonically cleaned in acetone and anhydrous ethanol for 5 min each. A 100 nm diamond powder abrasion for 10 min preceded ultrasonic seeding in a diamond suspension for 10 min. A final 2 min acetone rinse removed residues before drying and loading.
Controlled ICP removal
Section titled “Controlled ICP removal”Freestanding discs were etched in Ar, O2, and SF6 at 20 mTorr. The flow rates were 50, 50, and 25 sccm, with 600 W ICP power and 120 W RF bias. Cumulative etching lasted 1 h, 3 h, or 5 h at an average rate of about 6.8 µm h⁻¹. In the reported protocol, the gradient came from cumulative removal depth while gas flows and plasma settings remained fixed. Thickness measurements before and after each step controlled the removed depth. The film thickness fell from 145.2 µm to about 112.8 µm by the final etched condition. The reported thickness map gave a maximum etch depth of 33.2 µm and a minimum of 31.9 µm. The small depth variation indicates good in-plane uniformity and rules out lateral thickness variation as the main source of the thermal trend. The authors assign argon sputtering to C-C bond breaking, oxygen oxidation to volatile CO and CO2, and fluorine chemistry to C-F species. No direct in situ measurement of plasma composition is reported, so the reaction assignment remains a process interpretation.
Thermal measurement
Section titled “Thermal measurement”Laser flash analysis measured in-plane diffusivity by recording rear-face temperature rise after a short front-face laser pulse. The instrument was a NETZSCH LFA 467. Each surface received a uniform graphite coating to improve laser absorption and infrared signal generation. Added coating mass was held within 0.01 to 0.02 g on every specimen. Measurements were performed at 25 °C with three points per sample. The Cowan model with pulse correction was used to extract diffusivity from each temperature curve. All diffusivity uncertainties were below 1%, and the average curve goodness-of-fit exceeded R² = 0.85. The authors calculated conductivity using literature heat capacity and density values of 0.515 and 3.52, respectively. The results text sometimes labels α with conductivity units, although the methods define α as thermal diffusivity. The reported conductivity values are used here because the abstract and conclusions identify them as thermal conductivity. The dataset contained four samples, comprising one as-deposited specimen and three etched specimens.
Defect removal in the data
Section titled “Defect removal in the data”Scanning electron microscopy showed dense cracklike defects and irregular protrusions on the as-grown nucleation surface. After 1 h, micron-scale polygonal grains were exposed, but a connected network of concave pits remained. After 3 h, the pits had fragmented into isolated defects and grain boundaries were clearer. After 5 h, the pits were essentially eliminated, and no prominent cracks or protrusions appeared in the observed surface. The authors associate these changes with fewer grain boundaries, less sp² carbon, and fewer microcrack sites.
Laser profilometry quantified a nonmonotonic roughness response. Areal roughness increased from 71.77 nm as grown to 258.93 nm after 1 h. It then fell to 222.30 nm after 3 h and rose to 252.97 nm after 5 h. The authors associate the later increase with exposure of underlying layers having larger grains or rougher microstructures.
XRD peaks at 2θ of about 44.0°, 75.3°, and 91.5° indexed to diamond (111), (220), and (311). At 3 h, the peaks sharpened and the (220) reflection became dominant. At 5 h, diffraction intensities reached their maximum and the background signal decreased. Raman spectra retained the diamond feature near 1332 cm⁻¹, while the sp²-related D and G bands near 1350 and 1580 cm⁻¹ decreased markedly. The diamond peak width changed nonmonotonically, which the authors interpret as a changing near-surface defect and stress state. Together, these observations support lower defect density and reduced non-diamond carbon after etching.
Thermal response and reference range
Section titled “Thermal response and reference range”The largest gain occurred during the first 3 h. Conductivity rose 6.8% from 0 to 1 h and another 7.7% from 1 to 3 h. It increased only 0.5% from 3 to 5 h, showing saturation under the tested conditions. The authors interpret this plateau as the thermal response approaching the effective conduction limit of the columnar growth layer.
| ICP etching time | Reported in-plane thermal conductivity, W·m⁻¹·K⁻¹ | Reported change |
|---|---|---|
| 0 h | 1549.9097 | As-grown reference |
| 1 h | 1656.1743 | 6.8% above the as-grown condition |
| 3 h | 1783.3763 | 7.7% above the 1 h condition |
| 5 h | 1792.0250 | 0.5% above the 3 h condition and 15.4% above as-grown |
Each etched specimen was thinner than the as-grown film, so the comparison did not hold thickness constant. The final value falls within the paper’s stated range for well-crystallized polycrystalline diamond, typically about 1500 to 1800 W·m⁻¹·K⁻¹. The as-grown film sat near the lower end of that range. The paper gives a theoretical value of about 2200 W·m⁻¹·K⁻¹ for single-crystal diamond, so the etched polycrystalline film did not reach that intrinsic reference. Despite increased roughness, the authors conclude that subsurface interfacial defects dominated the as-grown thermal resistance.
Limits and device implications
Section titled “Limits and device implications”The evidence supports a film-level process change, not a completed device integration. No bonded interface or packaged heat-spreader assembly was tested. The authors state that long-term effects on interfacial thermal resistance require targeted evaluation in packaged devices. They also call for a model linking etch depth, interfacial roughness, and conductivity. The authors suggest fluorine or chlorine gas mixtures to improve removal while limiting roughness, but they did not test those chemistries. Interfacial modification, bias-enhanced nucleation, and seeded growth on engineered templates are proposed for suppressing the layer during growth. The reported diffusivity uncertainty does not quantify process variation across growth or etching runs. For engineering use, the data imply measuring conductivity on the finished film rather than selecting etch time alone. The result applies to freestanding polycrystalline diamond grown and etched under the reported conditions. The introduction cites GaN HEMTs approaching 10 W·mm⁻¹, where junction temperature affects performance, reliability, and lifetime. The data support screening nucleation-layer removal for high-power devices. Device-level improvement remains unvalidated.
Materials and support
Section titled “Materials and support”For a freestanding polycrystalline diamond heat spreader, the paper requires MPCVD-grown diamond with a controlled nucleation layer and a defined ICP removal depth. The diamond materials catalog lists 6CCVD polycrystalline MPCVD-grown plates and wafers up to 125 mm, with thicknesses from 0.1 to 500 micrometers. Custom etch and interface requirements can be specified through the linked engineering support page.
View Original Abstract
A defect-rich nucleation layer near the substrate is widely regarded as a key thermal bottleneck in thick polycrystalline diamond films. Here, we quantitatively evaluate this effect by progressively removing the nucleation layer via depth-controlled inductively coupled plasma (ICP) etching and measuring the thermal conductivity. The thermal conductivity increases from 1549.9097 W·m−1·K−1 (as-grown) to 1656.1743 W·m−1·K−1 (1 h), 1783.3763 W·m−1·K−1 (3 h), and 1792.0250 W·m−1·K−1 after 5 h of etching, consistent with the reduction of defects and non-diamond carbon revealed by X-ray diffraction (XRD) and Raman analyses. These results provide a quantitative, depth-resolved validation of the nucleation-layer thermal resistance and establish an effective post-growth route to enhance thermal transport in thick polycrystalline diamond films.
Advanced Thermal Simulation Service Diamond Heat Spreader Performance Analysis
View Specs →
Electronic Grade Intrinsic Diamond Epitaxial Wafer H Terminated Superior 100 Mosfet Ready Substrate
View Specs →
Diamond Heat Spreader With Copper Coated Metalized Surface
View Specs →Original source
Section titled “Original source”References
Section titled “References”- 2018 - Effective Thermal and Mechanical Properties of Polycrystalline Diamond Films [Crossref]
- 2019 - Fabrication and Performance of CVD Diamond Cutting Tool in Micro Milling of Oxygen-Free Copper [Crossref]
- 2021 - Electron emission from H-terminated diamond enhanced by polypyrrole grafting [Crossref]
- 2019 - Conductivity of Boron-Doped Diamond at High Electrical Field [Crossref]
- 2012 - High Temperature Application of Diamond Power Device [Crossref]
- 2019 - A Diamond Made Microchannel Heat Sink for High-Density Heat Flux Dissipation [Crossref]
- 2008 - Thermal conductivity of nitrogenated ultrananocrystalline diamond films on silicon [Crossref]
- 2021 - Thermal Conductivity of Nano- and Micro-Crystalline Diamond Films Studied by Photothermal Excitation of Cantilever Structures [Crossref]
- 2018 - Nanoscale Thermal Transport: Theoretical Method and Application [Crossref]