Surphysis

Interface qualification, from measured microstructure to verdict

Will this joint, coating or barrier survive its duty cycle?

Surphysis takes a stack, a duty cycle and your own characterisation files, and returns the interface fracture margin, barrier breakthrough time and survival probability, with the measurement that would change the answer.

Run a caseSee the scorecard
Cross-section of a ceramic-to-metal joint with a five-step graded interlayer Al2O3, c-planegraded interlayer5 steps, 3.2 µmCr barrier, 150 nmIN718, LPBFcritical interface: Al2O3 / step 1G / Gc = 0.62 at 1 200 °C

Joints

Ceramic to metal, dissimilar alloys, additive builds

interface fracture margin
Cross-section of a thermal barrier coating on a superalloy substrate gas side, 1 150 °C7YSZ, EB-PVDcolumnar, 180 µmTGO, 4.2 µmNiCoCrAlY bond coatCMSX-4 substratespallation at cycle 1 340 of 1 500

Coatings

Thermal barrier, wear and environmental coatings

cycles to spallation
Cross-section of a chromium hydrogen barrier on pipeline steel, with hydrogen entering along grain boundaries H2, 10 MPa, 80 °CCr barrier, 150 nmgrains 15 to 25 nmX65 pipeline steelgrain-boundary pathbreakthrough at 3.4 a, P(20 a) = 0.02

Barriers

Hydrogen, sour gas and diffusion barriers

breakthrough time
01Hydrogen and ammonia
02Oil and gas
03Power
04Aerospace
05Additive manufacturing
06Electronics

How it works

four steps, one audit trail
  1. 01

    Describe the stack

    Choose materials from the passport library or type them in. Set the duty cycle: temperature history, pressure, environment, number of cycles.

  2. 02

    Attach what you measured

    Accepted formats: XRD as .xy or .xrdml, EBSD as .ctf or .ang, SEM as .tif with a scale bar, permeation as .csv. Each file tightens a prior.

  3. 03

    Watch the engine reason

    Stress, minimax grading, fracture, grain-boundary kinetics and Sobol sensitivity run as numbered rules. Every step is logged with its rule reference.

  4. 04

    Take the verdict to the bench

    Fracture margin, breakthrough time, survival probability, a ranked measure-next list, and the recipe as editable code with an acceptance test plan.

Public scorecard

all blind predictions
42
Blind predictions logged
since 2026-06-01
36 / 42
Within stated error band
86 % hit rate
0.18
Mean absolute log error
dex, life metrics
11
Published failure datasets
4 mechanisms

Every prediction is logged before the published outcome is looked up. Misses stay on the board with the rule that caused them. Datasets: thermal-cycling spallation, braze-joint fracture, hydrogen permeation and interlayer delamination, from the open literature.

Origin

Surphysis began as the modelling stack behind a series of papers on ceramic-to-metal graded interlayers for additive manufacturing. In that work a chromium diffusion barrier that lattice-diffusion theory called safe for a hundred laser pulses was consumed in a single pulse once the measured grain size of 15 nm to 25 nm entered the model.

The platform exists so that every verdict is computed from measured microstructure and a stated set of assumptions rather than from handbook values. Interface toughness by mode mixity, grain-boundary diffusion coefficients and thin-film residual stresses are curated with provenance and uncertainty, and the engine is scored in public against published failure data.