Methodology

Framework Architecture

The Virtual Ion-Optics Testbed progressed through five phases. Production results are driven by VIOT-3 (axisymmetric) and VIOT-4 (3D + DSMC + MD).

Inside the work

Follow the physics across scales.

Concept in motion
ION OPTICSCEX + NEUTRALSSURFACE RESPONSEXe⁺CEXMoTRANSPORT → IMPACT → RESPONSE

Conceptual coupling between ion optics, charge exchange with neutral xenon, and material response.

01

Transport

Particle motion and the electric field evolve together in the plasma solver.

02

Exchange and impact

DSMC represents neutral transport and charge exchange; molecular dynamics resolves an ion impact at the surface.

03

Compare

Campaign runs bring the models together to compare materials and operating conditions.

Phase III

Complete — 70/70 tests

VIOT-3: Axisymmetric

Production framework for the core erosion campaign. Models barrel erosion — axisymmetric, inside aperture wall. Corrects all 15 defects identified in VIOT-2.

Physics implementation

#
Defect in VIOT-2
Correction in VIOT-3
1
CUDA Red–Black SOR Poisson had block-sync errors, non-convergent
SuperLU direct factorization
2
Sputter-yield lookup discarded the angle axis
2D (E, θ) bilinear interpolation
3
Post-CEX velocity set to exactly 0
Maxwellian sampling at 400 K
4
Half-acceleration push (not leapfrog)
True leapfrog + backward half-kick
5
MCC double-counting
Correct null-collision ratio
6
ν_max computed at wrong n_gas
Recomputed from cross-section
7
Erosion ignored macroparticle weight
Y × w_real
8
Particles seeded inside solid cells
Plasma cells only
9
Upstream boundary φ = 0
φ = V_screen
10
No Bohm drift at injection
u_B + thermal spread

Phase IV

Core complete — 2/2 tests

VIOT-4: 3D + DSMC + MD

Extends the framework to 3D to capture pit-and-groove erosion on the downstream face and interstitial web. Integrates DSMC neutral flow and LAMMPS AIREBO crystalline MD for graphene substrate interactions.

Adversarial review: 5 bugs found and fixed

01

Poisson operator sign bug

x/y off-diagonals were +1, z were −1, assembling 6I + Nx + Ny − Nz — not a Laplacian. CG "converged" to the wrong operator. Fixed: all six off-diagonals −1. Verified: residual 1.7e-11.

02

MD projectile velocity off by 10¹⁰

Unit conversion applied ×1e-12 instead of ×1e-2. Projectiles were effectively stationary → ~zero sputter. Fixed; verified 300 eV Xe → 210 Å/ps.

03

Sputter count counted Xe as C

Used total-atom delta, tallying exiting Xe projectiles as sputtered carbon. Fixed: count type-1 (C) atoms only via per-atom type array.

04

Engine reinject teleported all particles

Reseeded every slot each step, discarding mid-flight ions. Erosion was 19 atoms instead of ~1e8. Fixed: refill only inactive slots.

05

DSMC collision not NTC

Fixed p=0.3 with unused σ, no cell-volume/dt scaling. Fixed: proper No-Time-Counter acceptance P = σ·g·dt/V_cell.

Scope

In scope

  • Relative material rankings for barrel erosion and EBS lifetime
  • Atomistic yield derivation (RustBCA + LAMMPS AIREBO MD)
  • Field-emission and thermomechanical risk screening
  • 3D framework construction and verification
  • Fowler–Nordheim breakdown risk (Weibull model)

Explicitly out of scope

  • Absolute wear rates
  • ELT-validated lifetimes
  • Dynamic erosion-coupled mesh evolution
  • In-loop DSMC coupling (pending)
  • Flight qualification of any coating