Growth evolution and stress distribution in heteroepitaxial diamond with laser-patterned templates
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Functional Diamond |
| Published | 2026-05-08 |
| Authors | Pengfei Qu, Peng Jin, HJ Xu, Zhaorun Dong, Z L Wang |
| DOI | 10.1080/26941112.2026.2670053 |
Laser-grooved Ir/YSZ/Si (001) templates produced a crack-free, approximately 1 mm freestanding single-crystal diamond layer after 200 h of MPCVD growth. Raman mapping gave an estimated residual compressive stress of −0.21 ± 0.18 GPa over a 1 mm × 1 mm area. At about 50 µm, the patterned film stayed intact, while an unpatterned film grown under identical conditions showed extensive surface cracking. The authors link this stability to polycrystalline channels that later become compliant stress-relief zones.
Why stress limits heteroepitaxial thickness
Section titled “Why stress limits heteroepitaxial thickness”Diamond growth on Ir carries an approximately 7.1% lattice mismatch, which produces strain and misfit dislocations near the interface. The paper also identifies growth temperature and non-uniform plasma density as sources of stress variation across a substrate. Prior work cited by the paper found thermal stress of several gigapascals after cooling from growth temperatures of 800 to 1000°C. Film bending, cracks, and delamination become more likely as the diamond layer thickens. The authors selected Ir/YSZ/Si because silicon’s thermal expansion coefficient is closer to diamond’s than that of many oxide substrates.
Template fabrication and damage control
Section titled “Template fabrication and damage control”The authors used 2-inch Ir/YSZ/Si (001) composite substrates with a nominal silicon miscut below 0.1°. The process sequence and its reported conditions were:
| Step | Reported conditions | Result or purpose |
|---|---|---|
| Substrate stack | 120 nm Ir film on a 100 nm YSZ buffer over Si (001) | This formed the composite template. |
| Bias-enhanced nucleation | 2.45 GHz; 40 min; −300 V bias; 750°C; 1.5 kW; 25 Torr; 5% CH₄/H₂; base pressure below 5 × 10⁻⁵ Pa before ignition | This step performed bias-enhanced nucleation. |
| Seed growth | 5 kW; 50 Torr; CH₄/H₂ flow at 6/300 sccm; approximately 900°C | An approximately 500 nm seed layer was deposited. |
| Laser patterning | 355 nm continuous-wave UV; 100 mW; 25 µm spot; 4 mm s⁻¹; effective fluence above 100 J cm⁻² | This formed 50 µm grooves around 500 µm × 500 µm squares with edges along ⟨110⟩. |
| Acid cleaning | Boiling H₂SO₄/HNO₃ at 3:1 by volume | This removed most graphitic residues. |
SEM showed that ablation left cracked YSZ fragments and Ir-rich droplets in the groove. A heat-affected zone extended about 15 µm from the groove edge. Raman spectra confirmed laser-induced damage. The intact seed gave a diamond peak at 1332.0 cm⁻¹, while the groove edge developed a graphite G band near 1580 cm⁻¹. Its center showed no carbon Raman signal and exposed the silicon F₂g peak near 520 cm⁻¹. Boiling mixed acid reduced the G band, although a faint residual signal remained at the groove edge.
Groove filling and closure
Section titled “Groove filling and closure”At 2 h, fine polycrystalline crystallites filled the grooves while adjacent seed regions retained smooth (001) epitaxial growth. The groove contained dense grains after 10 h. At 18 and 26 h, those grains enlarged as the opening narrowed. By 34 h, only large grains remained near the center, and the epitaxial regions had extended over most of the recess. The groove was nearly closed at 42 h and fully closed at 50 h. With further deposition, the surface evolved toward a step-flow morphology.
The authors interpret the early polycrystal as a consequence of laser damage and residual Ir/YSZ material. These features disrupt the epitaxial template and lower the nucleation barrier for randomly oriented grains. The high grain-boundary density reduces the vertical growth rate in the polycrystalline regions, while neighboring (001) epitaxial regions advance and bury them. The surface grooves therefore disappear, but their buried polycrystalline channels remain as mechanically compliant zones that accommodate local strain during cooling.
Stress and crystal quality at 50 µm
Section titled “Stress and crystal quality at 50 µm”At about 50 µm, XRD showed the same phase constituents and diffraction peak positions for patterned and unpatterned films. Laser patterning did not change the diamond/Ir/YSZ/Si orientation relationship. The diamond (004) rocking-curve FWHM was 0.22° for the patterned film and 0.14° for the unpatterned film. The wider curve indicates more mosaic spread, although the paper says the film retained well-defined epitaxial alignment.
Raman mapping covered a 1 mm × 1 mm area at the sample center with a 20 µm step. The patterned sample had an average peak position of 1332.3 ± 0.3 cm⁻¹, corresponding to estimated compressive stress of −0.19 ± 0.14 GPa. The unpatterned sample gave 1334.6 ± 0.3 cm⁻¹ and −1.61 ± 0.22 GPa. These estimates assumed biaxial in-plane stress and used 1332.0 cm⁻¹ as the stress-free peak and −0.61 GPa cm⁻¹ as the stress coefficient. Average linewidths were 5.1 ± 0.8 cm⁻¹ and 4.8 ± 0.3 cm⁻¹, respectively. Local stress and linewidth changes stayed near the grooves instead of extending across the epitaxial domains.
One millimeter freestanding layer
Section titled “One millimeter freestanding layer”The patterned film was extended for 200 h at a typical rate of about 5 µm h⁻¹, with 0.02 sccm N₂ added to the feed gas. The resulting layer was approximately 1000 µm thick. It detached during cooling without cracking and became a freestanding single-crystal diamond. Its surface had a clear step-flow mode, and the laser pattern was no longer visible. A dominant diamond (004) XRD peak confirmed the (001) epitaxial relationship. Rocking-curve FWHM values were 280 arcsec for (004) and 331 arcsec for (311). The authors compare these values with literature for large-area, millimeter-thick freestanding diamond on Ir/YSZ/Si, while noting that they do not reach the lowest reported mosaic spreads.
Central Raman mapping gave a peak distribution centered at 1332.3 ± 0.3 cm⁻¹ and estimated compressive stress of −0.21 ± 0.18 GPa. Most linewidth values were below 3.0 cm⁻¹, with an average of 2.6 ± 0.4 cm⁻¹. A central spectrum reached a minimum FWHM of 2.0 cm⁻¹ after Voigt fitting. Broadening remained in regions overlying buried polycrystalline channels, indicating local lattice distortion at former coalescence boundaries. No graphite-related Raman features were found. A weak feature near 1415 cm⁻¹ was assigned to NV⁰ centers introduced by the intentional nitrogen addition.
Engineering implications and remaining limits
Section titled “Engineering implications and remaining limits”The measurements support using a patterned seed layer to localize strain near grooves and preserve a continuous single-crystal surface. This is most relevant to large-area plates where thermal mismatch limits attainable thickness. The tradeoffs are specific. Groove formation caused a 15 µm heat-affected zone, partial graphitization, YSZ fragments, and Ir-rich droplets. Patterning also increased the (004) mosaic spread at 50 µm and left buried polycrystalline channels with localized lattice distortion.
The 1 mm layer was not fully transparent. The authors identify residual Ir/YSZ on the backside as the likely main cause. Nitrogen-related color centers and scattering by defects and buried polycrystalline regions also reduced transmission. The channel mechanism is supported by time-resolved morphology, Raman maps, and crack behavior rather than a direct through-thickness stress measurement. The paper reports no dislocation density for the 1 mm layer.
The authors also note that finer polycrystalline grains could provide greater stress relaxation. However, the required growth conditions may conflict with stable epitaxial growth around the grooves. Device selection should therefore weigh low residual stress and thickness against mosaic spread, buried defects, transparency, and a 200 h extended deposition.
Materials and support
Section titled “Materials and support”The paper needs a freestanding, (001)-oriented MPCVD single-crystal diamond plate, about 2 inches wide and 1 mm thick, for large-area substrate fabrication. 6CCVD lists MPCVD single-crystal diamond plates and wafers up to 125 mm in the diamond materials catalog, but its stated single-crystal thickness range of 0.1 to 500 micrometers does not cover the paper’s 1 mm layer. The engineering support link is the relevant contact point for custom fabrication support when the required thickness exceeds the standard range.
View Original Abstract
The heteroepitaxial growth of large-area thick single-crystal diamond is strongly limited by stress accumulation and crack formation during prolonged deposition. Laser-patterned templates have been demonstrated as an effective approach for obtaining crack-free freestanding diamond layers. In this work, the influence of laser grooves on the growth behavior of diamond films grown on Ir/YSZ/Si (001) substrates by microwave plasma chemical vapor deposition (MPCVD) is systematically investigated. Laser grooves introduced into a thin diamond seed layer generate a spatially non-uniform growth template. During the early growth stage, polycrystalline diamond forms inside the grooves, whereas the surrounding regions maintain stable (001)-oriented epitaxial growth. As deposition proceeds, the faster vertical growth of single-crystal diamond progressively overgrows the groove regions, leading to gradual groove filling and closure, ultimately forming a laterally continuous epitaxial diamond film. The buried grooves with polycrystalline diamond act as mechanically compliant zones that facilitate stress redistribution during thick-film growth, thereby suppressing crack formation and enabling the fabrication of millimeter-scale freestanding diamond layers with high crystalline quality. This work clarifies how laser-patterned templates regulate growth evolution and stress accommodation during diamond heteroepitaxy, providing insights for the controlled growth of large-area thick single-crystal diamond films on highly mismatched substrates.
Electronic Grade Intrinsic Diamond Epitaxial Wafer H Terminated Superior 100 Mosfet Ready Substrate
View Specs →
Single Crystal Cvd Diamond Plate
View Specs →
Boron Doped Single Crystal Diamond Plate
View Specs →References we cover
Section titled “References we cover”- High‐Temperature and High‐Electron Mobility Metal‐Oxide‐Semiconductor Field‐Effect Transistors Based on N‐Type Diamond
- Stress engineering of high-quality single crystal diamond by heteroepitaxial lateral overgrowth