Can a Tungsten Substrate Reduce Stress in Polycrystalline Diamond Films?

Growing a polycrystalline diamond film is only the first step. A production process also has to control cracking, residual stress, thickness uniformity, release, warpage, substrate reuse, and cost per usable area.

A paper in the Journal of Synthetic Crystals, titled “Effect of substrate type on stress and crystallinity of grown polycrystalline diamond films,” offers a useful comparison of tungsten (W), silicon (Si), and molybdenum (Mo) substrates in a 2.45 GHz MPCVD system. Its result is promising—but narrower than a production solution.

What the experiment reported

The reported growth conditions were 4,100 W microwave power, 21 kPa pressure, 400 sccm hydrogen, 12 sccm methane, and a growth temperature of approximately 910 °C. The researchers used COMSOL simulation and optical emission spectroscopy to optimize the stage height to 16 mm in an effort to reduce edge effects.

Under the same pretreatment and deposition conditions, the paper reported compressive stress measured by X-ray diffraction of:

Substrate	Reported compressive stress
Tungsten	143.7 ± 15.2 MPa
Silicon	161.3 ± 20.1 MPa
Molybdenum	251.2 ± 34.1 MPa

The team also reported a crack-free polycrystalline diamond film 50.8 mm in diameter and 0.6 mm thick on tungsten. Within this equipment configuration and process window, tungsten produced the lowest measured stress and better crystallinity uniformity among the three substrates.

That is a useful engineering result. It is not yet proof that tungsten solves the substrate-stress problem in general.

The result may belong to a system, not a material alone

The first unresolved question is why tungsten performed better. Thermal expansion, adhesion, carbide formation, and lattice-related explanations may all matter. But the tested substrates also interacted with different support and heat-removal conditions.

The more accurate interpretation may therefore be “tungsten plus its holder and cooling geometry performed better in this system,” rather than “tungsten is intrinsically the best substrate.” Separating those effects would require controlled experiments that vary substrate material and thermal boundary conditions independently.

Deposition success is not release yield

For thick free-standing diamond, successful deposition is not the commercial endpoint. A producer still needs to know:

• how the film is released;
• release yield and crack rate;
• post-release bow and residual stress;
• substrate damage and reuse cycles;
• edge loss and usable area;
• consistency across repeated runs;
• the resulting cost per accepted plate.

The paper, as summarized in the Chinese source, does not provide a systematic release method, release-yield data, substrate-loss data, or reuse economics. It therefore cannot establish that tungsten improves production economics, even if the attached film looks better.

What exactly did the stress measurement capture?

The reported XRD stress was measured while the film remained attached to its substrate. That state includes the constraint imposed by the substrate and thermal history of the coupled structure. It is not automatically the same as residual stress in a released, free-standing diamond plate.

This distinction matters because an attached film can appear manageable and then crack or warp during release. A stronger production study would compare in-situ or attached-film stress, release behavior, and post-release geometry across multiple batches.

How I would use this result

I would treat the study as an engineering comparison that identifies tungsten as a serious candidate for further validation. I would not treat it as a universal mechanism, a qualified release route, or evidence of mass-production readiness.

The next experiments should connect substrate choice to the full process chain: thermal-field design, edge control, deposition consistency, release yield, post-release stress, reuse, and unit economics. Only then can a lower XRD stress value become a manufacturing advantage.

Continue the research

• Why large-area growth is not the same as manufacturing maturity
• How MPCVD growth works
• Contact Wonsen to discuss process evidence