💎 Processes, Designs, and Miscellaneous

Part I: Heteroepitaxial Crystal Mechanics & Domain-Matching Epitaxy

1.1 Crystallographic Matching and Coordinate Space

Both 4H-SiC and 4H-C inhabit the hexagonal crystal family and share the identical four-layer ABAC stacking sequence (50% hexagonal / 50% cubic), permitting direct basal-plane (0001) heteroepitaxy with no polytype transformation across the interface. The substrate’s own step structure templates the carbon sequence, so only the composition changes along the gradient.

1.2 Domain-Matching Epitaxy (DME) Integral Ratio

Direct pseudomorphic growth yields a nominal misfit of:

Because 17.93% far exceeds the classical Frank–van der Merwe pseudomorphic strain limit (~2%), continuous single-crystal propagation requires Domain-Matching Epitaxy (DME). In the DME framework, integral multiples of lattice planes match across the interface:

The effective residual strain (ε_r) across the 6:5 domain boundary is reduced by an order of magnitude:

This residual strain of 1.507% is low enough to be entirely accommodated within the first 1–2 atomic monolayers via localized, geometrically confined misfit dislocations, permitting dislocation-free bulk single-crystal growth above the interfacial plane.

A longer 11:9 coincidence is also available and is substantially tighter:

The 11:9 registry reduces residual misfit magnitude by a factor of 4.75 and reverses its sign from tensile to compressive, at the cost of a coincidence period nearly twice as long — misfit dislocations are correspondingly sparser but each accommodates more displacement.

4H-SiC Substrate (ABAC...)         Interfacial Gradient Zone              4H-C Hexagonal Diamond (ABAC...)
┌─┬─┬─┬─┬─┐ (5 Substrate Cells)    ┌───┬───┬───┬───┬───┬───┐ (6 Carbon)   ┌─┬─┬─┬─┬─┬─┐ (Pure sp³ C-C, ABAC)
│ │ │ │ │ │                        │   │   │   │   │   │   │              │ │ │ │ │ │ │
┴─┴─┴─┴─┴─┴────────────────────────┴───┴───┴───┴───┴───┴───┴──────────────┴─┴─┴─┴─┴─┴─┴────────────────
[ Si/C = 1.0 ]                 →   [ Atomic Si Depletion Only ]      →   [ Si/C = 0.0 ]
  ABAC stacking  ────────────────── unbroken across the entire gradient ──────────────►  ABAC stacking

1.3 Compositional Grading Without Polytype Conversion

Because substrate and terminus share the ABAC sequence, no stacking transformation is required at any point along the gradient. The vapor-phase stoichiometric precursor ratio (SiH₄ / CH₄) is continuously modulated from 1.0 down to 0.0 under an active atomic hydrogen (H•) or atomic oxygen (O(³P)) flux, which selectively combusts sp² and cubic-seed nucleation centers while sustaining epitaxial step-flow on the (0001) basal plane.

This is the decisive simplification of the four-layer terminus. Growth on a substrate cut slightly off-axis exposes the full ABAC period at the step edges, so each advancing terrace inherits its stacking directly from the layer beneath it — the same step-flow templating that makes 4H-SiC homoepitaxy routine. The 3C twinning channel is never opened, because nothing in the process asks the lattice to change its sequence. Polytype fidelity becomes a matter of maintaining step flow rather than of defeating a thermodynamic preference.

1.4 Polytype Selection: 4H-C vs. 2H-C (Lonsdaleite) Pathways

While both 2H (wurtzite, 2-bilayer ABAB...) and 4H (4-bilayer ABAC...) diamond allotropes are geometrically valid hexagonal carbon frameworks, the 4H-SiC → 4H-C heterojunction is the strictly superior operational pathway for continuous single-crystal manifestation.

Crystallographic symmetry breakdown. Neither route preserves space-group symmetry across the chemical gradient. In both cases the two distinct chemical species in the carbide break inversion symmetry along [0001] (polar P6₃mc), whereas pure sp³ elemental carbon places inversion centres at the C–C bond midpoints (centrosymmetric P6₃/mmc). Stacking is preserved on the four-layer route; symmetry is not, on either.

The advantage is kinetic and thermodynamic. Holding an advancing growth terrace at 1000–1200 °C sits directly adjacent to the ~1400 °C solid-state decomposition boundary of 2H-SiC, risking spontaneous polytype collapse during epitaxy — the two-layer route asks the substrate to survive near its own transformation regime for the entire run. Stacking fault energy also falls as hexagonality rises, so the 100%-hexagonal 2H lattice has no interleaved cubic-like planes to obstruct basal slip; the 50/50 hybridisation of 4H raises γ_SFE and with it τ_crit, the quantity on which the whole envelope of Part III rests. Against those, 4H additionally carries roughly 11% lower DME residual strain, halves the off-axis optical background, and inherits an existing ultra-high-purity wafer supply for which 2H-SiC has no counterpart.

1.5 Precursor Mass Flow Dynamics & Kinetics of the Graded Interface

To cross the heterointerface without triggering three-dimensional island nucleation (Volmer–Weber growth) or plastic yield, the compositional grading depth must satisfy L_g ≥ 24.6 nm. One bilayer in a four-layer polytype is c/4, which contracts from 0.2513 nm in 4H-SiC to 0.2059 nm in 4H-C, so the requirement corresponds to 108 bilayers, not 60 — sixty would deliver only 13.72 nm and drive dε/dz to 1.10 × 10⁶ m⁻¹, well past the microcleavage limit. Deposition runs by high-density MPCVD under closed-loop mass flow control:

                  108-BILAYER MPCVD PRECURSOR RAMP PROFILE
 25 ┌──────────────────────────────────────────────────────────╴ CH₄ 25 sccm
    │                                            ╱╌╌╌╌╌╌╌╌╌╌╌╌╱
 20 │                              ╱╌╌╌╌╌╌╌╌╌╌╌╌╱
sccm│                ╱╌╌╌╌╌╌╌╌╌╌╌╌╱
 10 ┼╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╲                                          SiH₄ 10 → 0
    │                 ╲╌╌╌╌╌╌╌╌╌╌╌╲╌╌╌╌╌╌╌╌╌╌╌╌╲╌╌╌╌╌╌╌╌╌╌╌╌╲
  0 └──────────────────────────────────────────────────────────╴ SiH₄ 0 sccm
    0 (SiC buffer)            54 (interphase)            108 (pure 4H-C)
    ◄───────────────── 49.4-minute continuous ramp ─────────────►

• Chamber pressure: 140 Torr (18.66 kPa) ± 0.1 Torr via downstream throttle valve.
• Substrate temperature: 1120 °C ± 5 °C by dual-wavelength optical pyrometry — inside the 1000–1200 °C step-flow window of §1.4.
• Microwave core power: 4.8 kW at 2.45 GHz, plasma electron density n_e ≥ 2 × 10¹² cm⁻³.
• Carrier gas (H₂): 450 sccm, Pd-purified.
• Kinetic etch flux (O₂): 1.2 sccm (0.27 vol% of carrier), generating in-situ O(³P) to combust sp² graphitic seeds.
• Step-flow growth rate: 0.50 nm/min (8.33 × 10⁻¹² m/s) for 49.4 min, 27.4 s per bilayer. This is ~0.03 μm/hr, roughly two orders below conventional MPCVD rates — the slow ramp is what keeps the composition gradient coherent, not an efficiency target.

Silane depletes linearly, Q_SiH₄(t) = 10.0 · (1 − t / 2963 s) sccm, while methane accelerates concurrently, Q_CH₄(t) = 10.0 + 15.0 · (t / 2963 s) sccm, sustaining supersaturation as the carbide sublattice empties. The resulting grading rate is

which sits below the 6.46 × 10⁵ m⁻¹ microcleavage limit with about 6% margin. Note that the depth column is not linear in k: the bilayer contracts by 18% across the gradient as Si leaves the lattice, so the last bilayers are thinner than the first even at constant deposition time.

Part II: Multi-Scale Hierarchical Tatami Strain Engineering

2.1 Photoelastic Indicatrix Perturbation & Symmetry Breaking

In unstrained cubic diamond (Fd3̄m), the optical indicatrix is spherical (n_x = n_y = n_z = n₀ = 2.420). In hexagonal 4H-SiC and 4H-C, the indicatrix is naturally uniaxial (n_x = n_y = n_o, n_z = n_e along [0001]).

The 4H-C indices follow from the cubic case without new measurement. Both phases are pure sp³ tetrahedral carbon at the same bond length, so the electronic polarizability per atom is transferable: α = 0.837185 ų, extracted from cubic diamond through the Lorentz–Lorenz relation. Applying that α to the ABAC lattice by Ewald dipole summation gives

The method closes on itself: rebuilding cubic diamond as an ABC sequence in a hexagonal cell returns 2.42000 isotropic to five decimals. Intrinsic birefringence scales linearly with hexagonal fraction, so the four-layer terminus carries exactly half the value of a two-layer one.

This intrinsic Δn is thirty times the engineered ceiling derived in Part III, and would overwhelm it entirely — except that it vanishes identically for propagation along [0001], where both transverse indices are n_o. The architecture is therefore an on-axis architecture by necessity, not by preference. Off the c-axis, the natural uniaxial retardance dominates the engineered signal by roughly two orders of magnitude.

Introducing an engineered, non-hydrostatic internal strain tensor (σ_kl) perturbs the optical impermeability tensor via the piezo-optic coefficient matrix (π_ijkl):

In-plane non-symmetric strain components (ε_xx ≠ ε_yy, ε_xy ≠ 0) break the transverse rotational symmetry (C₆ᵥ → C₂ᵥ). This transforms the material from a uniaxial to a triaxial (optically biaxial) indicatrix (n_x ≠ n_y ≠ n_z), splitting the single optic axis into two distinct axes separated by an acute angle (2V):

The slab is treated as traction-free out of plane (σ_zz = 0), so the third principal strain relaxes rather than being clamped at zero:

Only the deviatoric part (e_i − ē) enters the photoelastic perturbation; the hydrostatic component shifts all three indices equally and produces no birefringence. Clamping e₃ to zero would inject a spurious dilatation into the indicatrix.

2.2 The 4-Tier Hierarchical Tatami Architecture

Natural "tatami" strain in gemstones is an uncontrolled defect caused by intersecting plastic slip bands. Engineered hierarchical tatami strain structures coherent elastic stresses across four distinct dimensional tiers:

• Nanoscale (1–20 nm): Coherent Elastic Superlattices. Alternating pure ¹²C and ¹³C isotopic layers (Δa/a₀ ≈ 1.5 × 10⁻⁴) paired with sub-critical substitutional atomic size-mismatch delta-doping (e.g., Boron r_B = 0.88 Å vs Carbon r_C = 0.77 Å; Aluminum vs Nitrogen in 4H-SiC).
• Microscale (0.1–5 μm): Dislocation-Pinned Cross-Hatching. 60°/120° rhombic slip geometry aligned along conjugate basal ⟨112̄0⟩ and pyramidal {101̄1} slip systems.
• Mesoscale (10–50 μm): Sub-Surface Thermal-Kinetic Waveforms. Controlled ±50°C to ±150°C thermal oscillations at 1300 K to interlock microscale retardance zones into macroscopic optical uniformity.
• Macroscale (>100 μm): Global Boundary Stress Envelope. Continuous perimeter composition grading establishing a globally stable biaxial stress field (σ_xx ≠ σ_yy) that prevents edge-crack nucleation.

2.3 Piezoelectric Superlattices & Dopingless 2DEG Networks

Piezoelectricity is available across the carbide-bearing portion of the gradient and not beyond it. 4H-SiC crystallises in P6₃mc and lacks an inversion centre, because inversion would exchange Si for C. Pure 4H-C restores that symmetry — every carbon polytype places inversion centres at its bond midpoints, giving P6₃/mmc — so all odd-rank tensors vanish and d_ijk = 0 at the terminus. Within the graded zone, where Si occupancy is finite, spatial strain divergence (∇·ε) induces built-in piezoelectric fields:

The bound polarization charge density (ρ_bound = −∇·P) alternates sign at the 60°/120° tatami strain band boundaries, inducing intersecting grids of Two-Dimensional Electron Gases (2DEGs) and Two-Dimensional Hole Gases (2DHGs) without chemical impurity ionization scattering. The piezoelectric coefficient decays smoothly to zero as Si occupancy falls, so the 2DEG/2DHG network is confined to the graded envelope and terminates before the pure-carbon volume begins.

In the centrosymmetric terminus the surviving mechanism is flexoelectric rather than piezoelectric. Flexoelectricity is permitted in every crystal class, couples to the strain gradient rather than the strain itself, and scales inversely with the length over which that gradient is developed. The nanoscale tier of §2.2 — coherent elastic superlattices at 1–20 nm — therefore carries the steepest gradients and the strongest response, making the finest tier of the hierarchy the electrically active one in pure 4H-C.

Part III: Mathematically Calculated Physical Frontiers

3.1 Graded Peierls Envelope and the In-Plane Shear Modulus

Maximum in-plane shear stress is governed by C₆₆, not the out-of-plane C₄₄. For principal in-plane strains e₁ and e₂ with γ_max = e₁ − e₂:

The dislocation dissociation envelope scales with the local partial Burgers vector and the effective stacking fault energy:

Slip-system caveat. τ_crit(ξ) above is the basal Peierls limit γ_SFE/b_p, derived for slip on (0001). Under pure in-plane plane stress (σ_zz = τ_xz = τ_yz = 0) the traction on the basal plane vanishes identically, so the basal Schmid factor is zero and this criterion cannot be excited by the loading modelled here. It is retained as a conservative lower-bound proxy. A physically governing in-plane criterion requires the prismatic system {1̄100}⟨112̄0⟩, whose critical resolved shear stress is a separate and considerably larger input.

3.2 Zero-Order Waveplate Retardance Metrics

On-axis retardance follows Γ = Δn · L with Δn = ½ n_o(ξ)³ |p₁₁ − p₁₂| γ_max. Because both n_o and the shear envelope grade with composition, the achievable ceiling differs by almost an order of magnitude between the two ends of the crystal.

The terminus reaches a given retardance in roughly one-sixth the thickness. The gain comes from the shear envelope rather than the optics: n_o falls from 2.650 to 2.410 across the gradient, which alone would reduce Δn by 25%, but γ_max at τ_crit rises 8.8-fold, and that dominates.

Part IV: Continuum Mechanics & Kilometer-Scale Dimensional Constraints

4.1 The Stored Strain Energy Bottleneck & Explosive Griffith Rupture

The binding strain at the substrate follows the corrected in-plane relation τ_max = C₆₆·γ_max, giving γ_max = 82.86 MPa / 205 GPa = 4.042 × 10⁻⁴ at the substrate envelope. Scaling a continuous 1 km³ monolith (V = 10⁹ m³) loaded to that limit:

The critical Griffith flaw size before self-driven dynamic fracture is micron-scale, not nanometre-scale:

Earlier statements of this quantity as ≈10 nm were a units slip of three orders; evaluating the same expression returns tens of microns. The engineering consequence is materially different. Voids below roughly twenty microns are tolerated rather than catastrophic, which places the requirement inside the reach of conventional CVD inclusion control instead of demanding atomically void-free growth. Once a flaw does reach a_c, propagation is still supersonic Mode-I at the Rayleigh speed (v_crack ≈ 0.9 c_R ≈ 10,800 m/s), traversing the 1 km block in 92.6 milliseconds.

At the carbon terminus the envelope opens further: γ_max = 3.548 × 10⁻³ raises u to 6.61 × 10⁶ J/m³ and lowers a_c to ≈ 0.26 μm, so the tolerance for flaws tightens exactly where the retardance ceiling is highest. The two ends of the crystal trade fracture tolerance against optical range.

4.2 Gravitational Self-Weight & Geoid Curvature Limits

Part V: Zero-Gravity Float Zone & Multi-Phase Assembly Physics

5.1 Capillary Float-Zone Dynamics in Microgravity (Bo → 0)

In terrestrial float-zone processing, gravitational hydrostatic slump confines stable melt diameters via the Bond number (Bo = ρ g R² / γ < 1) to R ≤ 15–25 mm. In microgravity (g → 0, Bo → 0), gravitational slump is eliminated, and melt bridge stability is governed strictly by the Rayleigh-Plateau capillary instability limit:

Buoyancy-driven convection vanishes (Grashof number Gr → 0). Thermocapillary Marangoni convection is damped by applying an axial magnetic field (B ≥ 0.5 T) to exert Lorentz drag (F_L = σ_e(v × B) × B), establishing planar, laminar solute crystallization across meter-scale cross-sections.

5.2 Upper Atmospheric Edge Hypersonic Wake-Shield Mechanics

Operating at the mesosphere/thermosphere boundary (~85–200 km) at hypersonic orbital velocities (v ≈ 7.8 km/s) provides two physical mechanisms:

• In-Situ 5 eV Atomic Oxygen (O(³P)) Etching: High-energy translational atomic oxygen flux selectively combusts lower-energy sp² graphitic defect phases into CO/CO₂ gas (ΔH = −360 kJ/mol) while leaving the rigid sp³ wurtzite lattice intact.
• Hypersonic Wake Molecular Pumping: Behind a forward aerodynamic shield, a molecular vacuum shadow achieves pressures of P_wake ≤ 10⁻¹⁴ Torr with infinite volumetric conductance, eliminating impurity backstreaming.

5.3 The Multi-Phase Assembly Sequence

                         MULTI-PHASE HYBRID ASSEMBLY TOPOLOGY

 [ Phase I: Atmospheric Wake ] ──► [ Phase II: Deep-Sea Hydrobaric ] ──► [ Phase III: Core Mimetic ]
  • Hypersonic O(³P) Etch           • Isobaric High Pressure (0.1 GPa)      • Multi-Megabar (4–360 GPa)
  • UHV Infinite Pumping            • Incompressible Fluid Damping          • Dense sp³ Volumetric Clamping
  • Zero-g Capillary Float (Bo→0)   • Supercritical Solute Quench           • Frozen Tatami Strain Locking

In Phase III (Core Mimetic Locking), transferring the crystal to a multi-megabar hydrostatic press increases the critical dislocation glide threshold as a function of confinement pressure:

Hydrostatic pressure raises τ_crit from 82.86 MPa to > 5 GPa, enabling the permanent locking of high-amplitude tatami strain fields (ε > 0.5%) into the 4H-C matrix without risking plastic slip or polytype degradation upon athermal isentropic decompression.