Skip to content

RF bias-enhanced heteroepitaxial nucleation of (100) diamond on (100) Ir/a-sapphire

FieldValue
JournalJournal of Applied Physics
Published2026-03-02
AuthorsYonhua Tzeng, Jui-Cheng Chien, I.-Lun Chiu, Bo-Han Wu, Lung-Hsin Hsu
DOI10.1063/5.0311777

RF-BEN produced (100)-oriented heteroepitaxial diamond on (100) Ir over a-plane sapphire by 2.45 GHz MPCVD. The optimized sample 7 reached a diamond (004) rocking-curve FWHM of 1.27°. XRD and pole-figure measurements confirmed the preferred (100) orientation and in-plane epitaxial alignment.

Conventional DC-BEN requires a conductive path from the plasma to the substrate holder. On Ir/a-sapphire, bias current must cross the iridium edge, so resistance depends on film thickness, lateral path length, and edge conductivity. A two-inch sapphire wafer with a 200 nm iridium coating had resistance of 3.6 Ω at room temperature and 7.4 Ω at 300 °C. The estimated resistance at 900 °C was four times the room-temperature value. Path resistance makes bombardment nonuniform, and the authors expected large wafers to develop tens of volts of drop beyond the 10 to 20 V window. RF bias crosses the dielectric as a capacitive response and produces a time-average negative bias relative to the plasma. The authors propose that this removes the need for a current path through the insulating substrate.

Buffer preparation and nucleation conditions

Section titled “Buffer preparation and nucleation conditions”

The authors used two-inch a-plane single-side-polished sapphire, 430 ± 15 μm thick. Acetone, ethanol, isopropanol, and deionized water each received a 5-minute ultrasonic bath. After cleaning, they sputtered a (100) iridium film at 900 °C, 30 W, and 7.6 mTorr. X-ray diffraction showed only iridium (100) and sapphire peaks in the scan. The iridium (004) rocking-curve FWHM was 0.158°, supporting the reported single-crystal quality of the buffer. Each Ir/a-sapphire piece was diced to 7 × 7 mm² and re-cleaned before nucleation.

StageConditions
Pre-bias carbonCH₄ exposure for 15 min before RF bias
RF-BEN2.45 GHz, 21 Torr, 4%, 8%, or 10% CH₄ in hydrogen, 750 to 800 °C, 950 to 1050 W microwave, 40 to 120 W RF, 30 to 50 min
Texture growth45 Torr, 850 to 900 °C, 1300 W microwave, 50 to 70 min, with 0.5%, 1%, or 1.5% CH₄ in hydrogen

Sample 7 used 120 W RF bias for 50 minutes with 8% methane during nucleation. It then received 70 minutes of texture growth in 0.5% CH₄ diluted by hydrogen.

The authors model plasma, sheath, sapphire, and DC-blocking capacitances in the bias circuit. The substrate and holder remain negative relative to the plasma because electrons move faster than positive ions. At 120 W, nucleation changed the sheath capacitance and reduced the measured negative substrate-to-ground DC bias with time. The authors identify this decrease as an indication of diamond nucleation, unlike the rising DC current used in DC-BEN. At 40 W and 60 W, SEM showed amorphous carbon and some polycrystalline diamond nuclei after 30 minutes. They attributed this morphology to insufficient ion energy and activation of the iridium surface. At 100 W and 120 W, trenches and furrows aligned along [110] appeared instead. The aligned features indicated sufficient ion energy to activate the (100) surface. Studies cited by the authors place the initial iridium nuclei at a few nanometers inside disordered carbon domains. Most nuclei were not resolved as grains by SEM during biasing and grew only after the bias ended.

At 4% CH₄, the surface showed trenches and furrows without domain structures. At 8%, sparse branch-like fractal domains appeared over the trench network. At 10%, domain coverage increased, but amorphous or graphitic agglomerates also appeared. The authors attributed the agglomerates to carbon oversupply that exceeded surface reconstruction and hydrogen etching. Optical emission spectroscopy showed stronger C2-related emission as methane concentration increased. Excess C2 was proposed as a cause of overnucleation or graphitic phases, but that mechanism was not verified. During texture growth, hydrogen-rich 0.5% CH₄ etched poorly attached or misoriented nuclei. Under 0.5% methane, the surface showed isolated pyramids bounded by four {111} facet families. The corresponding pole figure had fourfold symmetry. At 1% and 1.5% methane, secondary nucleation covered {111} facets and the pole figures became disordered. The proposed two-step strategy first establishes strong {100} texture, then accelerates {111} facet coalescence. The authors did not measure the relative growth-rate parameter α, so feed-gas calculations served only as reference values.

Measured film structure and residual stress

Section titled “Measured film structure and residual stress”

Raman spectroscopy used 532 nm excitation, and XRD used Cu Kα radiation.

MeasurementResultCondition
XRDDiamond (400) peak at 119.52°Scan at 88° tilt
XRCFWHM of 1.27°Diamond (004), sample 7
Pole figureFour peaks at 0°, 90°, 180°, and 270°{111} reflection at 2θ = 43.9°
Raman3.3 cm⁻¹ shift and −1.8711 GPa compressive stressDensely merged center grains
Raman1333.7 cm⁻¹ peak and −0.9639 GPa compressive stressPartly merged edge grains
Raman0.5 cm⁻¹ shift and −0.2835 GPa compressive stressDiscrete diamond grain
EDS99.99 at.% carbon and 0.01% iridiumSample 7 mapped area

Only the diamond (400) peak appeared in the 88° scan, supporting (100) heteroepitaxy. Pole-figure symmetry also showed no twins, according to the authors. The center contained dense grains with partial coalescence, but local delamination appeared after cooling. The edge showed more voids and less complete merging. Stress was lowest in the discrete grain because edges, surfaces, and boundaries allowed stress redistribution. The authors interpret the rocking-curve width as evidence of small disorientations and low-angle boundaries despite macroscopic alignment.

The authors reported a nucleation density of 10⁸ to 10⁹ cm⁻², comparable with their DC-BEN data. Nucleation also depended on microwave power density and gas pressure. Schreck et al. produced 90 mm freestanding wafers with a 0.064° (004) rocking-curve FWHM. Kim et al. reported 0.027° on 2-inch diamond/Ir/YSZ. The current rocking-curve width is therefore much broader than either reference. The comparison is not identical because the current film remained attached to Ir/a-sapphire. Previous RF-BEN studies produced random nuclei on insulating substrates, while iridium heteroepitaxy had not been reported. Sample 7 demonstrated feasibility rather than a finished freestanding wafer. The tested pieces were 7 × 7 mm², so large-area performance was not demonstrated. Center and edge morphology differed, and some grains remained isolated with small voids. Further thick-film growth and analysis were still under study. Uniform microwave plasma and RF bias over a large substrate remain unresolved. The authors suggest larger wafers may become possible, but they did not demonstrate a larger-diameter film.

Diamond’s single-crystal thermal conductivity of 2200 W m⁻¹ K⁻¹ motivates its use in high-heat-flux electronic packages. For material selection, the demonstrated structure is a (100)-textured heteroepitaxial film on a composite Ir/a-sapphire substrate. A coalesced (100) film could address that application only if it has few voids, limited delamination, and manageable compressive stress. Device engineers would still need to qualify orientation spread, grain-boundary structure, and thermal-cycle reliability. The work does not demonstrate electrical performance in a diamond device. RF-BEN should therefore be treated as a nucleation route that still requires controlled coalescence and stress management.

6CCVD offers single-crystal diamond plates and wafers up to 125 mm grown by MPCVD in its diamond materials catalog. The paper’s proposed large-area heteroepitaxial wafer route needs (100)-oriented diamond nucleated on a (100) Ir buffer over a-plane sapphire. That process-specific requirement should be reviewed through engineering support, because the paper did not produce a freestanding wafer.

View Original Abstract

We report radio-frequency (RF) bias-enhanced nucleation (RF-BEN) of heteroepitaxial (100) diamond on (100) Iridium (Ir)/a-sapphire by 2.45 GHz microwave plasma chemical vapor deposition (MPCVD) in 8% CH4 diluted by hydrogen. The electrically insulating sapphire substrate acts as a capacitor of low impedance between the iridium film and the substrate holder and allows RF biasing current to flow through it. Because of a higher electron mobility than positive ions, a time-average negative DC bias voltage is induced on iridium with respect to the plasma leading to the nucleation of heteroepitaxial diamonds. Subsequent diamond growth in 0.5% CH4 diluted by hydrogen produces (100) diamond films. X-ray diffraction (XRD) at sample tilt = 88° confirms the diamond (400) reflection with a rocking-curve full width half maximum of ≈ 1.27°. XRD {111} pole-figure measurements confirm reduced mosaic spread and the absence of twins. The RF-BEN technique is promising for producing heterogeneous diamond wafers on iridium-coated large-area electrical insulating substrates.

  1. 1991 - Properties and Applications of Diamond
  2. 1971 - Diamond growth rates and physical properties of laboratory-made diamond [Crossref]
  3. 2023 - Cooling future system-on-chips with diamond inter-tiers [Crossref]
  4. 2018 - Recent advances in diamond power semiconductor devices [Crossref]
  5. 2019 - Diamond power devices: State of the art, modelling, figures of merit and future perspective [Crossref]
  6. 2001 - High-frequency performance of diamond field-effect transistor [Crossref]
  7. 1942 - Energy relations of the surface of solids. I. Surface energy of the diamond [Crossref]
  8. 1991 - Calculation of surface energies for low index planes of diamond [Crossref]
  9. 2021 - Nucleation of diamond films on heterogeneous substrates: A review [Crossref]
  10. 2006 - Seeding, growth and characterization of nanocrystalline diamond films on various substrates [Crossref]