The harder coating was not the more durable coating.
Nine depositions looked like a recipe-tuning problem. Together, they exposed a material constraint: the energy that hardened a-C also reduced thickness and adhesion.
Nine comparable a-C depositions spanning two bias regimes, gas-flow and pulse settings.
- 01Reconstructed the process
Converted ion-probe traces into surface ion energy.
- 02Calculated the relationships
Calculated how hardness, thickness, stress and adhesion changed across all nine runs.
- 03Challenged the explanation
Tested the stress–thickness model and 0.15 GPa rule against observed peeling.
- 04Turned evidence into a decision
Separated measured, interpreted and proposed evidence; the result pointed to a layered coating.
What the panel revealed
Higher ion energy increased hardness while thickness and adhesion fell. Residual stress alone did not distinguish held coatings from peeled ones.
-
01
Hardness came with a cost
Across nine runs, hardness reached 13.6 GPa as thickness and adhesion fell (r = −0.93 and −0.88).
-
02
The stress screen missed failures
Three of six peeled coatings were below 0.15 GPa. Stress alone would have passed failed films.
-
03
One layer carried three jobs
One deposition-energy setting controlled all three properties. No tested recipe met every target.
Separate the three jobs across a graded coating.
A single a-C film could not deliver all three outcomes. Assign adhesion, thickness and hardness to separate layers.
Best measured single-layer results
| Candidate | Bias | Ar flow | Thickness | Hardness | Stress | Lc2 / Lc3 | Wear |
|---|---|---|---|---|---|---|---|
| Stress / Lc2 margin | 80 V | 120 sccm | 1190 nm | 7.1 GPa | 0.077 GPa | 43 / 59 N | 168 |
| Wear / Lc3 | 80 V | 90 sccm | 1200 nm | 7.8 GPa | 0.133 GPa | 25 / 62 N | 79 |