Written Plan: Solving Discovered Materials Problem 02 (Interfacial Electrical Properties from Processing History) via Unified Grassmannian Mechanics (G-IEP)


1. Executive Summary & Unified Theoretical Framework (G-IEP)

Modern complementary field-effect transistors (CFETs), gate-all-around nanosheet/nanowire transistors (GAAFETs), wide-bandgap \(\mathrm{SiC}\) power MOSFETs, high-\(\kappa\) \(\mathrm{HfO_2/SiO_2/Si}\) gate stacks, and high-mobility \(\mathrm{Ge} / \mathrm{SiGe}\) channels expose semiconductor interfaces to complex 3D curvatures, crystallographic orientations, and multi-step thermal/ambient recipes (oxidation, cooling, post-oxidation annealing, and forming-gas passivation). Discovered Materials Grand Challenge Problem 02 (“Interfacial Electrical Properties from Processing History — Predict how industrially relevant semiconductor interfaces form, evolve and acquire their electrical properties from the fabrication history: growth, cooling and annealing”) [Discovered Materials et al., 2026] requires a single computationally tractable framework, unchanged in functional form, that maps the starting material, geometry, crystal orientation, composition, and thermal/ambient history to three target property classes:

  1. Layer thicknesses: Oxide thickness \(x_{\mathrm{ox}}(t)\), interlayer thickness \(x_{\mathrm{IL}}(t)\), and consumed substrate thickness \(\Delta x_{\mathrm{sub}}(t) = x_{\mathrm{ox}}(t) / \gamma_{\mathrm{PBR}}\) (where \(\gamma_{\mathrm{PBR}} = 2.27\) is the Pilling–Bedworth molar volume expansion ratio for \(\mathrm{Si} \to \mathrm{SiO_2}\)),

  2. Composition profiles \(C(z)\): Oxidant steady-state flux balance \(F_1 = F_2 = F_3\), suboxide transition width, \(\mathrm{C}\) interstitial / \(\mathrm{CO}\) out-diffusion in \(\mathrm{SiC/SiO_2}\), \(^{18}\mathrm{O}_2\) / \(V_{\mathrm{O}}^{\bullet\bullet}\) exchange in \(\mathrm{HfO_2/SiO_2/Si}\), and \(\mathrm{Ge}\) “snowplow” enrichment pile-up \(C_{\mathrm{Ge}}(z)\) in \(\mathrm{SiGe/oxide}\), and

  3. Electrical quality: Energy-resolved interface-trap density \(D_{\mathrm{it}}(E)\) (\(\mathrm{cm^{-2}\,eV^{-1}}\) across \(E \in [E_v, E_c]\)) and fixed oxide charge \(Q_f\) (\(\mathrm{cm^{-2}}\)),

validated simultaneously across all five benchmark semiconductor interfaces:

  • Interface 1 — Planar \(\mathrm{Si/SiO_2}\) across (100), (511), (110), and (111) orientations, oxidation history, and \(\mathrm{H_2}\) passivation (Vitkavage, Irene & Massoud, J. Appl. Phys. 68, 5262 (1990) [Vitkavage et al., 1990]; Deal & Grove, J. Appl. Phys. 36, 3770 (1965) [Deal and Grove, 1965]),

  • Interface 2 — Curved \(\mathrm{Si}\) Nanowires and Nanosheets (GAAFET / CFET) with viscoelastic stress-retarded self-limiting oxidation (Kao, McVittie, Nix & Saraswat, IEEE Trans. Electron Devices 35, 25 (1988) [Kao et al., 1988]),

  • Interface 3 — \(\mathrm{SiC/SiO_2}\) Si-face (0001) vs. C-face (000\bar{1}) growth kinetics, carbon removal, and \(\mathrm{NO}/\mathrm{H_2}\) passivation (Song et al., J. Appl. Phys. 95, 4953 (2004) [Song et al., 2004]),

  • Interface 4 — \(\mathrm{HfO_2/SiO_2/Si}\) High-\(\kappa\) Gate Stack oxygen exchange, \(\mathrm{SiO_2}\) interlayer regrowth, and trap/dipole evolution (Goncharova et al., Phys. Rev. B 83, 115329 (2011) [Goncharova et al., 2011]), and

  • Interface 5 — \(\mathrm{Ge/GeO_2}\) volatile \(\mathrm{GeO(g)}\) desorption loss and \(\mathrm{SiGe/oxide}\) selective oxidation & \(\mathrm{Ge}\) snowplow enrichment (Wang et al., Appl. Phys. Lett. 111, 052101 (2017) [Wang et al., 2017]; Rabie, Haddara & Carette, J. Appl. Phys. 98, 074904 (2005) [Rabie et al., 2005]).

Our Grassmannian Interfacial Process-to-Electrical Properties (G-IEP) framework synthesizes five verified pillars of the Signals library:

Grassmannian Pillar in Signals

Existing Module & Paper

Role in Problem 02 (Interfacial Electrical Properties from Processing History)

1. Plabic / Causal 3-Flux Reaction–Diffusion Transfer (F₁ = F₂ = F₃)

Signals.Plabic, Signals.QuantumCausalInference (Paper 2, Paper 18)

Models the gas-to-oxide (\(F_1\)), oxide-diffusion (\(F_2\)), and interfacial-reaction (\(F_3\)) series flux network (giepThreeFluxSteadyState), yielding the Deal–Grove–Massoud thickness law (giepDealGroveTimeForThickness), consumed substrate \(\Delta x_{\mathrm{Si}} = x_{\mathrm{ox}}/\gamma_{\mathrm{PBR}}\) (giepConsumedSiliconThickness), and Song \(\mathrm{SiC}\) carbon-retarded flux (giepSiCCarbonRetardedRate).

2. \(\mathrm{Gr}_{\mathbb{R}}(2,4)\) Crystal-Plane Bond Projection & QCI-GXC Fractional-Idempotency Defect Density

Signals.DFT, Signals.Geometry (Paper 22, Paper 23)

Projects the crystal cut (hkl) onto the tetrahedral \(sp^3\) bond basis to determine areal available-bond density \(N_{\mathrm{bond}}(hkl)\), and identifies unpassivated dangling bonds (\(P_{b0}, P_{b1}\)) and suboxide bridges with the Kohn–Sham idempotency defect \(\operatorname{Tr}[P_\sigma(I - P_\sigma)] = \lambda(1-\lambda) > 0\) (twoOrbitalLinearCorrelationEntropy), which is healed by McWeeny purification (mcweenyPurificationPoly) under Brower–Stesmans \(\mathrm{H_2}/\mathrm{NO}\) passivation (giepHydrogenPassivatedDit, giepEnergyResolvedDit).

3. DDF / Splat Viscoelastic Stress Jamming & Eyring Activation Volume

Signals.DDF, Signals.HawkingRadiation (Paper 3, Paper 15, Paper 16)

Models the Pilling–Bedworth volumetric expansion (\(\gamma_{\mathrm{PBR}} = 2.27\)) on curved nanowires/nanosheets of radius \(r_{\mathrm{NW}}\) (giepNanowireStressRetardedRate, Kao et al. 1988), proving self-limiting oxidation slowdown and stress-induced \(D_{\mathrm{it}}, Q_f\) enhancement at small \(r_{\mathrm{NW}}\).

4. Normed Tensor Gaussian Splat Advection–Diffusion Composition Profiles \(C(z)\)

Signals.Coherence, Signals.DDF (Paper 6, Paper 10)

Models the depth-resolved oxidant profile \(C_{\mathrm{O_2}}(z)\), suboxide transition layer, \(\mathrm{HfO_2}\) oxygen-vacancy exchange driving \(\mathrm{SiO_2}\) interlayer regrowth (giepHfO2InterlayerRegrowthSq, Goncharova et al. 2011), \(\mathrm{GeO(g)}\) volatile desorption loss (giepGeOxideNetRate, Wang et al. 2017), and \(\mathrm{SiGe}\) snowplow \(\mathrm{Ge}\) pile-up \(C_{\mathrm{Ge}}(z)\) (giepSiGeSnowplowSurfaceFraction, Rabie et al. 2005).

5. Non-Destructive Inline Optical/Electrical Metrology & Equivalent Oxide Thickness (EOT)

Signals.NonDestructive, Signals.DFT (Paper 23, Paper 25)

Connects interfacial layer regrowth \(x_{\mathrm{IL}}\) and high-\(\kappa\) thickness \(x_{\mathrm{high\text{-}\kappa}}\) to Equivalent Oxide Thickness \(\mathrm{EOT} = x_{\mathrm{IL}} + \frac{\kappa_{\mathrm{SiO_2}}}{\kappa_{\mathrm{HfO_2}}} x_{\mathrm{HfO_2}}\) (giepHighKGateStackEot) and subthreshold swing degradation \(\mathrm{SS}(D_{\mathrm{it}})\) (giepSubthresholdSwingWithDit).


2. Staged Execution Plan Across All Five Interfaces

Phase 1: Formal Verification in Lean 4 (Signals.DFT Section 17 & Signals.Pending Section 35)

  1. Universal 3-Flux Deal–Grove–Massoud & Pilling–Bedworth Laws:

    • Formalize giepThreeFluxSteadyState, giepDealGroveTimeForThickness, giepConsumedSiliconThickness, giepHydrogenPassivatedDit, giepEnergyResolvedDit, and giepDealOxideChargeCoolingTriangle.

  2. Interface 1 — Planar \(\mathrm{Si/SiO_2}\) Orientation & Passivation (Vitkavage, Irene & Massoud, JAP 1990):

    • Formalize the Vitkavage et al. (1990) crystallographic orientation hierarchy (100) < (511) < (110) < (111) for linear oxidation rate \(B/A\), unpassivated/passivated midgap \(D_{\mathrm{it}}\), and fixed oxide charge \(Q_f\), proving \(\mathrm{H_2}\) forming-gas annealing (FGA) strictly reduces \(D_{\mathrm{it}}\) by \(> 10\times\) and that \(D_{\mathrm{it}}(E)\) forms a U-shaped band-gap spectrum with minimum at midgap \(E = E_m\).

  3. Interface 2 — Curved \(\mathrm{Si}\) Nanowires & Nanosheets (Kao et al., IEEE TED 1988):

    • Formalize giepNanowireStressRetardedRate and prove that compressive normal stress \(\sigma_n > 0\) and tensile oxide pressure \(P > 0\) on tighter nanowire radii strictly retard oxidation compared to planar \(\mathrm{Si}\) while elevating unpassivated interfacial bond strain.

  4. Interface 3 — \(\mathrm{SiC/SiO_2}\) Face-Dependent Kinetics & Carbon Removal (Song et al., JAP 2004):

    • Formalize giepSiCCarbonRetardedRate and prove that carbon interstitial accumulation retards oxidation and accounts for the faster oxidation of the (000\bar{1}) C-face relative to the (0001) Si-face, while \(\mathrm{NO}\) annealing passivates carbon-cluster near-interface traps.

  5. Interface 4 — \(\mathrm{HfO_2/SiO_2/Si}\) Gate Stack Interlayer Regrowth & EOT (Goncharova et al., PRB 2011):

    • Formalize giepHfO2InterlayerRegrowthSq and giepHighKGateStackEot, proving monotonic \(\mathrm{SiO_2}\) interlayer regrowth with oxygen diffusion flux and annealing time, and sub-nanometer EOT scaling.

  6. Interface 5 — \(\mathrm{Ge/GeO_2}\) Desorption & \(\mathrm{SiGe/oxide}\) Snowplow Pile-Up (Wang et al., APL 2017; Rabie et al., JAP 2005):

    • Formalize giepGeOxideNetRate and giepSiGeSnowplowSurfaceFraction, proving the transition from net oxide growth (\(v_{\mathrm{growth}} > v_{\mathrm{desorb}}\)) to volatile \(\mathrm{GeO(g)}\) oxide loss (\(v_{\mathrm{growth}} < v_{\mathrm{desorb}}\)) and \(\mathrm{Ge}\) snowplow enrichment at the \(\mathrm{SiGe/oxide}\) boundary (\(z = 0\)).

Phase 2: Rust/WASM & Python Implementation, Per-Interface Error Audit, & MyST Paper 27

  1. Implement all G-IEP functions and unit tests in …/src/signals/numerical_adapter/src/dft_plabic_vrc.rs (test_interfacial_electrical_properties_problem02_check_figures).

  2. Implement InterfacialElectricalProcessEngine in …/sustainablefactory/qci_grassmannian_dft.py and test_interfacial_electrical_properties_problem02_all_five_interfaces in …/tests/test_qci_grassmannian_dft.py, including a complete per-interface error audit comparing G-IEP against classical orientation/stress/carbon-blind Deal–Grove baselines.

  3. Write grassmannian-interfacial-electrical-properties-paper.md (Paper 27), update all cross-repository indices, run the full verification suite (lake build, cargo test, pytest, Markdown math linters, git diff --check), and commit with prefix ENH: .


3. Critique & Five Deep Gap Closures (interfacial_electrical_properties_deep_gap_closure_identities)

  1. Gap 1 (giepDealGroveExplicitThickness, giepMassoudThinOxideRate): Formalizes the explicit forward trajectory \(x_{\mathrm{ox}}(A, B, \tau, t) = \frac{\sqrt{A^2 + 4 B(t + \tau)} - A}{2}\), proving \(x_{\mathrm{ox}}(t)^2 + A x_{\mathrm{ox}}(t) = B(t + \tau)\) and exact right-inversion of giepDealGroveTimeForThickness, together with the Massoud thin-oxide enhancement rate (Vitkavage et al. 1990).

  2. Gap 2 (giepBrowerStesmansEquilibriumCoverage, giepBrowerStesmansEquilibriumDit): Formalizes the Brower–Stesmans competing \(\mathrm{H_2}\) passivation vs. thermal dissociation equilibrium \(\theta_{\mathrm{H}}^{\mathrm{eq}} = \frac{r_{\mathrm{pass}}}{r_{\mathrm{pass}} + k_{\mathrm{dissoc}}}\) (Interface 1).

  3. Gap 3 (giepNanowireCurvatureStress, giepNanowireStressInducedCharge): Formalizes the Kao cylindrical nanowire stress \(\sigma_{\mathrm{norm}} = \eta_{\mathrm{eff}} / r_{\mathrm{Si}}\) and curvature-induced fixed charge \(Q_{f,\mathrm{NW}}(r_{\mathrm{Si}})\) (Interface 2).

  4. Gap 4 (giepSiCCarbonClusterDit, giepHighKFlatbandVoltageShift): Formalizes Song \(\mathrm{SiC}\) carbon-cluster interface traps \((D_{\mathrm{it},0} + \gamma_{\mathrm{C}} r_{\mathrm{C}})(1 - \theta_{\mathrm{NO}})\) (Interface 3) and Goncharova \(\mathrm{HfO_2/SiO_2/Si}\) flatband shift \(\Delta V_{\mathrm{FB}} = V_{\mathrm{dipole}} - q_{\mathrm{fixed}}\cdot\mathrm{EOT}\) (Interface 4).

  5. Gap 5 (giepSiGeSnowplowMassBalance): Proves exact conservation of rejected Germanium mass \(\Delta x_{\mathrm{pile}}\cdot L_{\mathrm{decay}} = x_{\mathrm{Ge},0}\,\Delta x_{\mathrm{sub}}\) in \(\mathrm{SiGe}\) snowplow oxidation (Interface 5b).