Verdict report

Sapphire to IN718 graded joint, LPBF build

Al2O3 (c) / MoO2 / 5-step Mo-Ni graded interlayer 3.2 µm / Cr 150 nm / NiCr 0.3 µm / IN718 LPBF

C-0412 run 7

2026-09-04

M. Kawwam

engine 0.1, rule set R00 to R24, data 2026.09.05-1

Verdict

Conditional pass

The graded interlayer holds (Phi 0.79, fracture margin 37); the Chromium barrier does not, by diffusion: consumed in 1 pulses against 5 required. Replace or thicken the barrier before the build, or accept one pulse per site.

Duty cycle

LPBF: 5 pulses, 60 W, 80 µs, bed 180 °C, interface peak 1 296 °C. Service: 25 cycles 25 to 380 °C.

Key quantities

Quantity	Value	Rule
Step stress, graded stack	0.63 GPa	R05
Survival criterion Phi	0.79	R06
Interface fracture margin, step 1	G_c / G = 36.7	R18
Seed membrane stress, plastic strain	2.23 GPa, 0.58 %	R08
Barrier breakthrough at pulse peak	8.9 µs	R13
Pulses survived	1.01	R14
P(survive 5 pulses)	0.014	R15
Process window, feasible up to	1 600 K	R23

Assumptions mock table pending assumption registry

Assumption	Why	Effect on verdict	Validation route
Sapphire CTE 7.0 x 1e-6 / K	Yim and Paff 1974, c-axis	Peak stress varies 0.55 to 0.72 GPa over the disputed range	Dilatometry on the actual wafer lot
Grain-boundary diffusion governs Cr consumption	Measured crystallite 15.5 nm; lattice path too slow	Breakthrough moves from about 110 pulses to about 1	TKD map and Cr depth profile after 1 pulse
Ea_gb 120 kJ / mol, range 95 to 150	No direct measurement for sputtered Cr in Ni	Largest single term in the uncertainty, Sobol 0.47	Permeation or diffusion couple at 600 to 900 °C
Plastic shedding in NiCr seed	Membrane stress 2.2 GPa exceeds yield	Reduces transmitted stress to step 5 by 0.4 GPa	Nanoindentation of seed after anneal
Interface toughness 8.5 J / m2 at Al2O3 / Mo	Evans and Dalgleish 1992; Paper 3 wedge tests	Sets the denominator of the fracture margin	Four-point bend on the coupon
1D thermal model, semi-infinite	Spot 80 µm much larger than stack 3.7 µm	Interface peak within 40 K of FE	Pyrometer trace on trial build

Measure next

Rank	Quantity, instrument	S_T	Format
1	D0_gb, permeation or diffusion couple	0.25	.csv t and J
2	Q_gb, permeation or diffusion couple	0.75	.csv t and J
3	grain size d, EBSD	0.09	.ang, .ctf

Acceptance tests mock

No.	Test	Accept when
1	XRD on the Cr layer before build	Crystallite size >= 40 nm or barrier thickness >= 400 nm
2	Single-pulse trial on a witness coupon, Cr depth profile	Cr continuity >= 30 nm after 1 pulse
3	Four-point bend, 5 coupons	Fracture load >= 190 N, failure not at Al2O3 / step 1

Recipe as code mock

# Surphysis recipe, case C-0412 run 7
substrate: Al2O3 c-plane, HCl activated
anchor: MoO2 reactive PVD, 50 nm, O2 stepped 2e-4 to 1e-5 Torr
graded:
  steps: 5
  rule: dalpha_i proportional to 1 / M_i   # R06
  composition_wt_Mo: [85, 70, 54, 38, 20]
  thickness_um: 0.64 each
barrier: Cr, 150 nm   # R21 flags: survives 1.1 pulses
seed: NiCr 80/20, 300 nm
anneal: 450 C, 2 h, vacuum
build: LPBF 60 W, 80 us, spot 80 um, bed 180 C, 1 pulse per site

Rules fired

R00Preferred value selection
R01Biaxial modulus
R02Total CTE mismatch
R03Temperature excursion
R04Unrelieved mismatch stress
R05Step stress, graded stack
R06Survival criterion Phi
R07Minimum step count
R08Membrane stress
R09Plastic strain
R10Grain size from XRD
R11Residual stress from XRD
R12Effective diffusivity
R13Barrier breakthrough time
R14Pulses survived
R15Survival probability
R22Dundurs parameter
R17Phase angle
R16Energy release rate
R18Fracture margin
R24Membrane gate
R23Process window
R19Weakest interface
R20Sobol indices
R21Measure next