Effect of methane concentration on the growth kinetics, microstructure, and crystalline quality of thick MPCVD-grown polycrystalline diamond
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Carbon letters |
| Published | 2026-08-09 |
| Authors | Mohammed Bin Afif, Amine Mesbahi, A. S. Altakhov, Sebastian Farag, Daniel Choi |
| DOI | 10.1007/s42823-026-01123-5 |
At 890 °C, 6% CH₄ produced the highest measured growth rate, about 2.8 μm h⁻¹, while 4% CH₄ produced the best crystalline quality. The study grew freestanding polycrystalline diamond, or PCD, films approximately 190 to 325 μm thick. Raman FWHM reached 3.17 cm⁻¹ at 4% CH₄ and rose to 5.86 cm⁻¹ at 8% CH₄. The authors therefore identify 890 °C with 4% to 6% CH₄ as the practical range for balancing deposition rate and crystalline quality.
Growth rate peaks at moderate methane
Section titled “Growth rate peaks at moderate methane”At 4% CH₄, raising substrate temperature from 840 °C to 890 °C increased the growth rate from about 0.8 μm h⁻¹ to more than 2 μm h⁻¹. At 890 °C, moving from 4% to 6% CH₄ raised the rate to about 2.8 μm h⁻¹. Increasing methane to 8% lowered it to about 2.4 μm h⁻¹. Although 6% delivered the fastest growth, Raman quality did not improve.
The authors interpret the lower rate at 8% as entry into a carbon-rich regime. Their proposed mechanism links excess carbon-containing radicals to secondary nucleation and amorphous or sp²-bonded carbon. They also cite a lower atomic-hydrogen-to-hydrocarbon ratio at fixed total flow, which slows hydrogen abstraction and etching of non-diamond carbon. The authors present this as a mechanistic interpretation. The rate change itself was measured from film thickness divided by deposition time.
Controlled deposition and freestanding preparation
Section titled “Controlled deposition and freestanding preparation”Growth used a 6 kW, 2.45 GHz MPCVD reactor operated at 4 kW. Each run used a single-side-polished Si (100) substrate measuring 10 × 10 mm² and 0.5 mm thick. Seeding used fully deagglomerated detonation nanodiamond with 5 to 10 nm particles in a 1.25 mg mL⁻¹ DMSO-isopropanol suspension. Substrates were ultrasonated for 15 min, rinsed in isopropanol, and dried in clean air.
Chamber pressure was 80 Torr, total H₂/CH₄ flow was 1060 sccm, and microwave power was 4 kW. Plasma distance and holder geometry were fixed. A two-color pyrometer measured substrate temperature through the quartz window at about 45° incidence. The study used growth periods extending to 100 h.
After growth, the silicon was removed in aqueous 20 wt% KOH at 90 °C for 5 h under continuous stirring. The resulting freestanding PCD films were measured on their growth surfaces.
Raman quality and residual stress
Section titled “Raman quality and residual stress”Raman spectra were collected mainly near the center of the growth surface. A 633 nm laser, 10% neutral density filter, and 10× objective were used with an 1800 lines mm⁻¹ grating over 800 to 2900 cm⁻¹. A 40-point map covered about 350 × 350 μm². A Lorentzian fit yielded diamond peak position and FWHM.
At 4% CH₄, raising the temperature from 840 °C to 890 °C lowered mean FWHM from 4.38 to 3.17 cm⁻¹. Its FWHM interquartile range also fell from 1.73 to 0.61 cm⁻¹. At 890 °C, FWHM rose to 4.65 cm⁻¹ at 6% CH₄ and 5.86 cm⁻¹ at 8%. The interquartile range was 1.42 cm⁻¹ for S3 and 1.87 cm⁻¹ for S4. This broadening indicates lower crystalline quality and greater spatial heterogeneity.
The I₁₄₅₀/I₁₃₃₂ ratio tracks trans-polyacetylene and grain-boundary sp² carbon. It fell from 0.199 in S1 to 0.138 in S2 at 4% CH₄. S3 retained 0.132 at 6% CH₄, while S4 reached 0.263 at 8%. The I₂₂₅₀/I₁₃₃₂ ratio tracked SiV⁻ defect luminescence. It fell from 0.236 in S1 to 0.125 in S2, then remained 0.127 in S3. No distinct graphitic D or G bands were observed, which the authors interpret as minimal graphitization.
Mean Raman peak positions were 1332.11, 1332.56, 1332.60, and 1332.65 cm⁻¹ for S1 through S4. Using σ = −0.567(ν − ν₀), the corresponding compressive stresses were approximately −62, −317, −342, and −368 MPa. Because these films had lost their silicon support, the authors attribute the residual stress mainly to growth-related mechanisms. Temperature governed the mean stress, whereas methane concentration affected its spatial distribution. The peak-position interquartile range grew from 0.24 cm⁻¹ in S1 to 0.50 cm⁻¹ in S4.
The paper notes that single-crystal diamond typically has FWHM below 3 cm⁻¹. PCD is broader because grain boundaries and defects broaden the diamond Raman response.
Texture, grain size, and roughness
Section titled “Texture, grain size, and roughness”XRD used Cu Kα radiation at 45 kV and 40 mA over 5° to 100°. Every sample had its strongest diamond reflection at the (220) plane. The table summarizes the growth rate, Raman width, texture ratio, and grain size.
| Growth condition | Growth rate, μm h⁻¹ | Mean FWHM, cm⁻¹ | I₁₁₁/I₂₂₀ | Mean grain size, μm |
|---|---|---|---|---|
| S1, 840 °C, 4% CH₄ | about 0.8 | 4.38 | 0.04300 | 50.871 ± 3.274 |
| S2, 890 °C, 4% CH₄ | more than 2 | 3.17 | 0.03300 | 53.435 ± 0.826 |
| S3, 890 °C, 6% CH₄ | about 2.8 | 4.65 | 0.00280 | 62.331 ± 2.904 |
| S4, 890 °C, 8% CH₄ | about 2.4 | 5.86 | 0.095 | 51.214 ± 1.966 |
S3 had the strongest (220) texture. Its I₁₁₁/I₂₂₀ ratio was 0.00280, while I₃₁₁/I₂₂₀ was 0.00036. At 8% CH₄, those ratios rose to 0.095 and 0.00139. The authors therefore associate 8% CH₄ with partial texture loss and more competing crystallite orientations.
Laser microscopy used 100× images over roughly 129 × 129 μm². Grain size was determined by the ASTM E112 line-intercept method. The S3 grain coarsened to 62.331 ± 2.904 μm, while S4 fell to 51.214 ± 1.966 μm. The authors suggest competitive growth, secondary nucleation, or defect incorporation at 8%.
Root mean square roughness Sq was 2.964 μm for S1, 6.156 μm for S2, 4.444 μm for S3, and 5.608 μm for S4. S2 was rougher despite a grain size close to S1 because its facets had greater peak-to-valley height. S3 had larger grains but a smoother surface envelope. The paper concludes that faceting and height dispersion govern Sq more strongly than lateral grain size.
Why the methane optimum differs from prior work
Section titled “Why the methane optimum differs from prior work”Tang et al. grew 165 to 430 μm freestanding films at 4 kW and 105 to 110 Torr. Their methane study also changed substrate size, thickness, preparation, pressure, holder temperature, and holder geometry. Those concurrent changes made it difficult to isolate the methane effect.
Li et al. reported near-linear rate growth to about 10% CH₄ at 100 Torr. Their rate reached about 5 μm h⁻¹ at 10%, but the film was predominantly nanocrystalline graphite. Four-hour growth limited their maximum thickness to 20 to 30 μm. Chen et al. reported 8 to 12 μm h⁻¹ at 250 Torr, but two-hour growth produced films no thicker than about 20 μm.
The present study operated at 80 Torr. The authors state that pressure changes plasma chemistry and the balance between carbon incorporation and hydrogen-assisted etching. The methane concentration for peak rate therefore depends on reactor pressure. Rates from different pressures should not be treated as a single universal curve.
Selection and unresolved limits
Section titled “Selection and unresolved limits”For material selection within this reactor, S2 at 890 °C and 4% CH₄ favors Raman quality, stress uniformity, and strong (220) texture. Six percent is the throughput option because it reached about 2.8 μm h⁻¹ and produced the largest grains. Its FWHM was 4.65 cm⁻¹ and its compressive stress was about −342 MPa. Eight percent is unfavorable for quality because FWHM reached 5.86 cm⁻¹, the sp² ratio reached 0.263, and (220) texture weakened.
Four growth conditions were tested on 10 × 10 mm² substrates in one 80 Torr process. The Raman map covered only about 350 × 350 μm² near the film center. XRD was acquired from the growth surface. The authors expect secondary nucleation to continue through the thickness but do not report cross-sectional texture measurements.
Optical transmission and scattering were not measured, and post-growth polishing was left for future work. For optical windows, Raman quality alone therefore does not establish optical performance. For thermal-spreader designs, the measured compressive stress should also be included in mounting and thermal design.
Materials and support
Section titled “Materials and support”The paper needs freestanding MPCVD-grown PCD for optical-window and heat-spreader applications, with optical transmission, scattering, and polishing left for future study. 6CCVD’s diamond materials catalog lists MPCVD-grown polycrystalline diamond plates and wafers up to 125 mm, with thickness from 0.1 to 500 micrometers, which covers the paper’s reported film range. Process translation for those requirements can be discussed through engineering support.
View Original Abstract
In diamond growth using microwave plasma chemical vapor deposition (MPCVD) process, the methane fraction controls the balance between carbon supply and hydrogen etching, while substrate temperature governs surface diffusion and facet‑dependent growth. Yet for polycrystalline diamond (PCD) films used as optical windows and heat spreaders, the combined influence of these two parameters is rarely quantified across multiple, cross‑validated metrics. Here we isolate the role of CH 4 at fixed temperature by growing PCD on identically prepared Si (100) wafers with all parameters held constant except the methane fraction (4 to 8% at 890 °C). In addition, at a fixed CH 4 concentration of 4%, the growth temperature was varied from 840 °C to 890 °C. Freestanding films with thicknesses of ~ 190-325 μm were obtained, enabling correlation of growth kinetics with texture, grain coarsening, and defect content. Raising temperature from 840 °C to 890 °C at 4% CH 4 more than doubled the growth rate and strengthened a (220) preferred orientation. At 890 °C, increasing CH 4 from 4% to 6% maximized the growth rate (~ 2.8 μm h − 1 ) but increased sp² signatures (Raman I 1450 /I 1332 ), the SiV-related emission intensity ratio (I 2250 /I 1332 ) and broadened the diamond FWHM; at 8% CH 4 the lattice disorder and non‑diamond carbon rose sharply (FWHM ≈ 5.86 cm − 1 ) with partial loss of (220) texture. These results define a practical window characterized by high temperature with moderate CH 4 where thick PCD attains a useful trade‑off between deposition rate and crystalline quality for device‑relevant components.
Advanced Thermal Simulation Service Diamond Heat Spreader Performance Analysis
View Specs →
Diamond Heat Spreader With Copper Coated Metalized Surface
View Specs →