OPERATOR'S GUIDE
EHR DIGITAL TWIN · HOW TO RUN THE MACHINE, WHAT TO WATCH, WHAT IT'S FOR
01 · WHAT THIS IS
A reduced-physics digital twin of a helical-antenna plasma reactor: a cylindrical vessel (radius R, length 4R) in a twisted magnetic field, driven by RF. ~18k macro-electrons are pushed with a Boris integrator through the analytic field; Monte-Carlo collisions with neutral gas scatter, drag, and ionize (15.8 eV threshold); density accumulates on a 64×64×128 grid; the wafer chuck at z=+2 absorbs whatever reaches it. Units are normalized — lengths in radii (u), field in B₀ — so read every number as relative, not engineering-grade.
THE QUESTION UNDER TEST — the antenna launches a helical wave with pitch k_θ/k_z (k_θ = m/r); field lines twist with pitch B_θ/B_z. The match number M = (k_θ/k_z)·(B_z/B_θ) compares them. Hypothesis: RF→plasma coupling peaks when the two helices align, M ≈ 1. This rig lets you bake that in (KERNEL), or force it to emerge — or fail — from geometry (EMERGENT, WAVE).
02 · QUICK START — 60 SECONDS
0In a hurry? PRESETS at the top of the rail jump straight to canonical operating points — matched, detuned, helicon lock, plasma crystal, etch window.
1Stay on SIMULATE. The plasma is live — drag to orbit, wheel to zoom. HOLD freezes the push loop; RESET reseeds.
2Watch MATCH M in the top bar. Drag B_θ/B_z (or ANTENNA k_z) until the marker hits the center tick and M turns green (±5% of 1).
3Watch CPL and n CORE in the LIVE DIAGNOSTICS HUD climb. Now detune M away from 1 and watch the discharge starve.
4Switch COUPLING MODEL to EMERGENT — the M≈1 prior is removed. Does the peak survive on geometry alone?
5Open PLOTS → RUN M SWEEP for the full coupling curve, hands-free.
03 · CONTROL RAIL — LEFT PANEL
RF & ANTENNA
RF POWERHeating drive. Ionization rate and density scale with it — the throttle.
RF FREQInert in KERNEL/EMERGENT. In WAVE it enters the helicon dispersion — density locks to f (see f_RES).
ANTENNA k_zAxial wavenumber of the helical strap. With m it sets antenna pitch k_θ/k_z — the numerator of M.
MODE mAzimuthal mode 1–3. Higher m = tighter helix. In WAVE, deposition follows J₍m₋₁₎(k⊥r)² — m=1 is core-peaked. Verify with the B-DOT array.
MAGNETIC FIELD
SCREW PINCHUniform B_z + azimuthal B_θ ∝ r. Slider sets the wall ratio B_θ/B_z — the denominator of M.
BELTRAMIForce-free field (∇×B = αB). Twist varies with radius, so only one surface can match — watch where deposition lands.
READOUT BOXLive M, coupling %, reflected power (uncoupled RF), field-line pitch L_p at r=0.7R, and the WAVE resonance f_RES.
COUPLING MODEL — THE EPISTEMIC SWITCH
KERNELThe M≈1 hypothesis baked in as a Gaussian prior. A benchmark, not a test.
EMERGENTCoupling computed from local k̂·b̂ alignment + electron transit coherence; deposition at the resonant layer r=√M. No prior — the honest test.
WAVEEMERGENT + helicon eigenmode J₍m₋₁₎ and dispersion: RF FREQ becomes physical and n̄ chases f.
GAS & ACOUSTICS
PRESSURENeutral fill, 0.5–15 mTorr. More gas = more ionization targets but more drag; there's a sweet spot.
ACOUSTIC A/λ/fImpose a neutral-density wave (depth, wavelength, frequency). The plasma should imprint it — check the AXIAL STRUCTURE correlation r in DIAGNOSTICS.
TRAV/STANDTraveling wave sweeps along z; standing wave beats in place at fixed nodes.
WAFER & DUST
CHUCK RF BIASSheath voltage at the wafer, 0–300 V. Sets ion impact energy ⟨E_ion⟩ — trade etch rate against damage.
DUST COUNTInject up to 1000 charged grains. They levitate in the sheath, feel ion drag from the live ∇n, and interact via Yukawa pairs.
q_d · λ_DGrain charge and Debye screening length. Together they set the coupling Γ — past Γ≈170 the cloud freezes into a plasma crystal.
DISPLAY LAYERS
Chips toggle 3D render layers: FIELD LINES, ELECTRONS, VOLUME (density raymarch), ISOSURFACES, ANTENNA, DUST, CHUCK. Turn layers off to declutter or to speed up rendering on slow machines.
04 · SCREENS
SIMULATEThe 3D vessel. Drag to orbit, wheel to zoom. Top-right HUD streams live numbers — click its header to collapse. Bottom-left colorbar maps the density raymarch.
DIAGNOSTICSSix synthetic instruments: Langmuir probes (core/edge density + radial profile), turbulence proxy (δn/n trend + Hann PSD), 8-probe B-dot array (recovers the azimuthal mode m via DFT — compare against drive), synthetic GPI camera (line-integrated n_e), axial structure (n_e(z) vs acoustic drive, correlation r), and the wafer/dust panel (ion flux, etch non-uniformity, dust state + pair correlation g(r)).
PLOTSStrip charts of n̄, R_ec, δn/n, ionization rate. Pick M or f, hit RUN SWEEP: the engine steps the variable, holds each point until statistics settle, and traces the response curve. Completed sweeps are stored (up to 4) and overlay color-coded by coupling model — KERNEL / EMERGENT / WAVE — so you can re-run under a different model and compare curves directly. The trail also logs manual operating points every 1.5 s. EXPORT CSV downloads all stored runs + trail; CLEAR wipes both.
05 · RUN THESE EXPERIMENTS
P1 · DOES M≈1 EMERGE?
COUPLING MODEL → EMERGENT. PLOTS → sweep M. KERNEL would show a peak at 1 by construction — the question is whether EMERGENT reproduces it from k̂·b̂ geometry alone. Compare peak location, width, and asymmetry.
P2 · HELICON DENSITY LOCK
COUPLING MODEL → WAVE. Sweep f. Watch n̄ resonate where drive crosses f_RES, then re-lock as dispersion pulls density along. The f_RES readout in the rail tells you where to expect it.
P3 · GROW A PLASMA CRYSTAL
DUST COUNT ≥ 400, q_d up, pressure low. Watch Γ climb in the HUD; past ~170 the state flips to CRYSTAL and g(r) in DIAGNOSTICS develops sharp lattice peaks. Kick pressure up to melt it again.
P4 · FIND THE ETCH WINDOW
Tune field + bias to minimize ETCH NON-UNIF σ/μ (green < 20%) while keeping ion flux to the chuck high. Matched M with m=1 core deposition tends to flatten the wafer profile — verify it.
06 · READING THE HUD
n̄ / CORE / EDGEMean density, and probes at r<0.3R / r>0.8R (normalized).
R_ecEdge/core ratio. 0.8–1.25 reads green: flat, well-confined profile.
δn/nDensity fluctuation over a 10 s window — the turbulence proxy.
ION RATEIonization events per second (thousands).
⟨E⟩Mean electron energy, eV. Ionization needs the >15.8 eV tail.
τ RESMean electron residence time before loss.
Φ CHUCKIon arrival rate at the wafer.
Γ · κ · aDust coupling, screening ratio, grain spacing. Γ>170 ⇒ crystal.
REDUCED MODEL — analytic fields, fluid ions, normalized units. Trust trends and comparisons between settings, not absolute values. When in doubt, RESET and change one knob at a time.