Worked walkthrough — GaN:Mg p-type electronic doping (industrial III–V)
- 1 Diagram
- 2 Mechanism
- 3 Failure past the window
- 4 Metastability
- 5 SPECULATIVE Bridge A peg
What’s claimed and how well supported
- Labels on this page
- SPECULATIVE ×11
- Strongest label used
- SPECULATIVE
- Main caveat (quoted)
- Locks: Bridge A locked as ◆ SPECULATIVE supporting case only.
Counted automatically from the labels printed on this page. Counts are a reading aid, not a grade of the page.
Terms on this page (11), plain-language definitions
- Dopant
- Intentional dilute impurity meant to tune electronic/optical/catalytic behavior without replacing the host as majority phase.
- Host (matrix)
- Majority element/phase whose lattice dominates the material.
- Solute
- Intentional minority species dissolved in the host (alloying addition).
- Vacancy
- Point defect: a missing atom site in the lattice.
- Valence (chem.)
- Bonding capacity / oxidation-related electron count of an atom in context.
- Dislocation
- Line defect enabling plastic slip at realistic stresses; edge vs screw are the two classic geometries.
- FCC
- Face-centered cubic — many slip systems; Al, Cu, Ni typically ductile.
- Precipitate
- Second-phase particle that forms (often on aging) and blocks dislocations.
- HCP
- Hexagonal close-packed — fewer primary slip systems; Mg, Ti, Zn often anisotropic.
- BCC
- Body-centered cubic — Fe/steel common; plasticity more temperature-sensitive.
- Valency (ling.)
- Arity: how many actants a predicate takes; formal vs semantic valency can diverge.
Definitions quoted from the Chem for linguists primer glossary.
Materials first. The cross-domain peg stays last, and it stays SPECULATIVE.
Date: 2026-09-19
Audience: syntax/semantics readers building alloy-design literacy (Chem for linguists)
Purpose: Industrial semiconductor electronic-doping case from Grant metal-doping-takeaway.txt §5 / §8 — GaN:Mg as the primary p-type dopantIntentional dilute impurity meant to tune electronic/optical/catalytic behavior without replacing the host as majority phase. for GaN LEDs/lasers. Same checklist discipline as oxide electronic (Oxide electronic) and metal hostsMajority element/phase whose lattice dominates the material., but here the host is a mature III–V compound semiconductor, not a research photocatalyst oxide and not SSS.
Locks: Bridge A locked as SPECULATIVE supporting case only. Materials facts first; linguistic parallels only in a flagged box at the end. No joint-field claim. Navajo typed-participant half = light parallel only — not proof of Bridge A count. Dowty / Triskelion / Merge·Agree / kanji not advanced here.
Companion notes: Physics of doping §1.2–1.5 (dopant, substitutional acceptors, GaN:Mg example); Chem for linguists §5; Solubility windows; grant-doping-cast.md (not published here); Grant metal-doping-takeaway.txt §§5–6, 8–9; oxide electronic checklist (contrast: research vs industrial maturity).
- 0. Design brief (problem-driven)
- 1. Chem-for-linguists checklist (§8) — filled (GaN:Mg edition)
- 2. Host and dopant cast (REAL materials)
- 3. Concentration windows — under / useful / compensation / over
- 4. Activation anneal & H passivation (process co-actors)
- 5. “Over” is plural (name which failure)
- 6. Hard contrast — GaN:Mg vs Si:B/P, oxide electronic, metal SSS
- 7. Bridge A add-on checklist (depth note §6) — filled
- 8. Under / useful / over — materials only (no linguistics yet)
- 9. SPECULATIVE Bridge A overlay (supporting only)
- 10. Host contrast card (add GaN:Mg to the peg set)
- 11. What this walkthrough trains you to do next
- 12. Non-claims
- This walkthrough does not revise Bridge A’s lock, invent shared formalism, or announce a product.
- 13. Sources consulted (2026-09-19 note)
0. Design brief (problem-driven)
| Slot | Choice |
|---|---|
| Host | GaN (wurtzite III–V; wide gap ~3.4 eV) |
| Dopant | Mg substituting on Ga sites (Mg_Ga) |
| Named weakness of pure / n-only GaN | Without intentional acceptors, you cannot build p–n junctions, hole-injection LED/laser active regions, or enhancement-mode p-GaN gates — n-type GaN is comparatively easy (Si, O); p-type was historically the bottleneck |
| Intended mechanism | Classical electronic doping — substitutional acceptor → holes as majority carriers (after H depassivation) |
| Service hint | Blue/white LEDs, laser diodes, UV emitters; also p-GaN contacts / E-mode HEMT gates |
Vocabulary lock: here “dopant” is appropriate — intentional dilute impurity for carrier type and density. Same honesty as the oxide electronic note; opposite of Al–Mg SSS / Fe–C / Ni–γ′ (prefer soluteIntentional minority species dissolved in the host (alloying addition). / alloying). Grant takeaway names GaN:Mg explicitly as industrial Mg doping.
1. Chem-for-linguists checklist (§8) — filled (GaN:Mg edition)
1. What is the inherent weakness of the pure / base form?
- Undoped / unintentionally n-type GaN lacks controllable hole density for bipolar optoelectronics.
- Historical barrier: early GaN was hard to make p-type; Mg worked, but as-grown MOCVD material is highly resistive until H is removed.
- Not the weakness: soft yield strength (that was Al, Mg metal hosts). Different design problem → carrier engineering.
2. Which obstacle / tuning mechanism is doing the work?
Primary: substitutional Mg_Ga acceptors → holes when ionized.
Process co-actors (first-class):
1. Hydrogen passivation — MOCVD co-incorporates H → electrically neutral Mg–H complexes; as-grown ρ often ~10⁶ Ω·cm.
2. Activation anneal — RTA / furnace in N₂ (typically ≳600–700 °C; industrial wafer-scale RTA common) dissociates Mg–H and activates p-type (~Ω·cm class when optimized).
3. Native / process donors — nitrogen vacancies (V_N), residual Si/O/C, Mg interstitials — compensate acceptors.
Explicitly not: Labusch SSS; oxygen-vacancyPoint defect: a missing atom site in the lattice. photocatalysis as the headline goal (oxide note); intercalation CT.
3. What trade-offs are accepted?
- Deep-ish Mg acceptor (~135–200 meV thermal class; technique-dependent) → only a fraction of Mg atoms yield free holes at 300 K (~few % to ~10% order, not 100%).
- Higher [Mg] for more holes ↔ self-compensation, pyramidal inversion domains (PIDs), clustering, mobility drop, optical loss.
- Hotter / longer activation ↔ risk of In segregation in InGaN QWs, surface V_N, Mg diffusion, roughness.
4. How critical is processing vs nominal composition?
Extremely. Same SIMS [Mg] can be: passivated Mg–H (dead), activated Mg_Ga (holes), compensated by V_N / donors, or clustered in PIDs / Mg-rich phases — depending on growth ambient (H-rich MOCVD vs MBE), anneal T/atmosphere, and device stack constraints. Characterization (SIMS, Hall, PL, TEM for PIDs) is non-negotiable (Grant takeaway §9).
5. Problem-driven or exploratory?
Problem-driven and industrial. Blue LED / laser industry solved p-GaN with Mg + anneal (Nakamura lineage). Optimization continues (laser activation, contact stacks, E-mode HEMT), but the chassis is mature compound-semiconductor manufacturing, not research-grade oxide photocatalysis trial-and-error.
2. Host and dopant cast (REAL materials)
2.1 GaN host + Mg_Ga acceptor
| Item | Approximate figure / fact | Role |
|---|---|---|
| Host | Wurtzite GaN; E_g ~3.4 eV | Wide-gap III–V optoelectronics chassis |
| Dopant site | Mg on Ga (Mg_Ga) | Substitutional acceptor → holes |
| Acceptor depth | Thermal ionization often cited ~135–160 meV (Hall); some estimates ~190–200 meV; optical/thermodynamic deeper in some DFT | Much deeper than Si:B (~45 meV) → low room-T ionization fraction |
| Useful [Mg] (SIMS) | Often ~1–3×10¹⁹ cm⁻³; peak p near ~2×10¹⁹ cm⁻³ in classic MOCVD Hall series | Industrial LED window (order-of-magnitude) |
| Free holes after activation | Typically peak ~4–6×10¹⁷ cm⁻³ at that Mg window (not equal to [Mg]!) | Incomplete ionization + residual compensation |
| Hole mobility | Often ~10 cm²/V·s class in optimized bulklike layers | Low vs n-GaN; conductivity bottleneck |
| As-grown MOCVD | ρ ~10⁶ Ω·cm (Mg–H passivated) | Requires activation |
| Activated p-GaN | ρ ~1–2 Ω·cm class when well annealed | Device-usable p-type |
Mechanisms to name:
1. Acceptor substitution — Mg_Ga → hole when ionized (valenceBonding capacity / oxidation-related electron count of an atom in context. deficit vs Ga).
2. H passivation / depassivation — Mg–H complexes kill electrical activity until N₂ anneal (or LEEBI historically; laser activation research alternative).
3. Compensation — V_N and related donors; residual impurities (Si, O, C); Mg_I as donor-like compensator in some models.
4. Self-compensation at high [Mg] — doping-driven rise of compensating defects past ~2×10¹⁹ cm⁻³.
5. Structural overrun — PIDs with Mg segregation at boundaries; Mg–N–Mg / Mg₃N₂-like clusters at still higher dose.
Sources (headline): Kaufmann et al. PRB 62, 10867 (2000) p(N_A) peak ~6×10¹⁷ at N_A~2×10¹⁹; Nakamura anneal / Mg–H lineage; Nardo et al. J. Phys. D 2022 laser vs RTA activation (E_a~0.14 eV); arXiv:1607.08353 self-compensation (V_N, Mg_I); JAP 2022 hot-wall MOCVD free-hole / PID review; PMC8308141 heavily Mg-doped p-GaN; Grant takeaway §5 (GaN LEDs/lasers: primary p-type dopant).
2.2 Industrial status vs research oxides
| Aspect | GaN:Mg | Oxide electronic (TiO₂:Fe, research photo) | Al:ZnO TCO |
|---|---|---|---|
| Maturity | Industrial — LEDs, lasers, power/RF GaN fabs | Mostly research / trial-and-error (Grant §6) | Semi-industrial TCO; still process-sensitive |
| Metrology / process control | SIMS, Hall, PL, RTA recipes, SEMI-adjacent fab discipline | Often powder XRD/XPS/EPR campaigns | Hall + optical T; ALD/sputter recipes |
| Primary product property | Hole injection / p-contact / junction | Visible photocatalysis QE | Low ρ + high visible transmittance |
| Grant takeaway | Explicit industrial Mg use-case | Research oxide landscape | Transparent conductor sibling |
3. Concentration windows — under / useful / compensation / over
| Zone | GaN:Mg reading |
|---|---|
| Too little | [Mg] ≲ mid-10¹⁸ cm⁻³ class → p too low for contacts / injection; residual n-type or high resistivity |
| Useful window | [Mg] ~1–3×10¹⁹ cm⁻³ (peak often ~2×10¹⁹); after N₂ activation, p ~10¹⁷–few×10¹⁷ cm⁻³; usable LED/laser p-GaN |
| Compensation-limited | Same [Mg] but unannealed (Mg–H) or donor-rich → Hall p collapses despite SIMS Mg present |
| Self-compensation / structural over | [Mg] ≳ ~2–3×10¹⁹ → p falls; V_N / Mg_I / PIDs / clusters; μ falls monotonically with N_A; near ~10²⁰ holes can vanish |
| Optical / mobility loss | Excess Mg → scattering, defect luminescence, absorption; bad for laser cavities and efficiency |
Anchors (not design specs): Kaufmann 2000 Hall series (p_max≈6×10¹⁷ at N_A≈2×10¹⁹; drop by ~10²⁰); MOCVD power-device digests ([Mg]~2–3×10¹⁹ → p~4–6×10¹⁷); PMC8308141 saturation / PID / cluster notes. See also Physics of doping §2 dilute / useful / over ladder.
4. Activation anneal & H passivation (process co-actors)
| Step | What happens | Industrial note |
|---|---|---|
| MOCVD growth | Abundant H (carrier gas + NH₃) → Mg–H complexes | As-grown layers highly resistive |
| Activation | Anneal in H-free N₂ ≳600–700 °C (often ~650–850 °C / tens of min class; recipe-dependent) | Dissociates Mg–H; H need not fully leave immediately for partial activation |
| Over-anneal risks | Surface V_N, GaN dissociation, Mg diffusion, InGaN QW In segregation, roughness | Optimize by [Mg] and full device stack — not “hottest is best” |
| Alternatives | Historical LEEBI; research laser activation (localized; resistivity comparable to optimized RTA in reports) | RTA remains wafer-scale industrial default |
| Reversible passivation | Re-anneal in NH₃ / H-rich ambient can re-passivate | Process atmosphere is part of the doping story |
Three clocks still apply (Process metastability): equilibrium defect chemistry vs process capture (H incorporation) vs service (contacts, thermal budget of later steps).
5. “Over” is plural (name which failure)
| Overrun mode | Physics | Where it shows up |
|---|---|---|
| Incomplete ionization (baseline, not “over”) | Deep acceptor → p ≪ [Mg] at 300 K | Even in the useful window |
| H passivation | Neutral Mg–H | As-grown MOCVD; failed / skipped anneal |
| Self-compensation | Doping-driven compensating donors (V_N, Mg_I, complexes) | [Mg] past ~2×10¹⁹; p falls while SIMS Mg rises |
| Structural clustering / PIDs | Pyramidal inversion domains; Mg at boundaries; Mg₃N₂-like clusters | Heavy Mg; TEM/SIMS signatures |
| Mobility collapse | Ionized + neutral impurity scattering; defect density | μ decreases with N_A |
| Optical loss / defect luminescence | Mid-gap / Mg-related bands; absorption | Laser / high-efficiency LED pain |
| Process stack damage | Over-anneal In segregation, V_N surface layer | Full LED/LD epi constraint |
Hard contrast vs prior hosts: Al–Mg “over” = GB β; Fe–C = cementite/morphology; oxide electronic = recombination / μ / secondary oxides; CNT intercalation = moisture dedoping. Here “over” is electronic compensation + III–V defect structure on a mature fab chassis.
6. Hard contrast — GaN:Mg vs Si:B/P, oxide electronic, metal SSS
| Case | Prefer saying | Mechanism | Maturity | Window character |
|---|---|---|---|---|
| This file: GaN:Mg | Dopant / p-type doping | Substitutional acceptor + H activation | Industrial LEDs/lasers | ~10¹⁹ cm⁻³ Mg; p ~10¹⁷; self-comp past peak |
| Si:B / Si:P | Dopant (classical) | Shallow acceptor/donor (~45 meV B); ion implant + anneal | Fully standardized (SEMI) | Often 10¹⁵–10¹⁸+ cm⁻³ device windows; high ionization at 300 K |
| Oxide electronic (TiO₂:Fe, AZO) | Dopant | Mid-gap / V_O / TCO donors | Mixed; photo often research | at.% class; recombination / μ collapse |
| Al–Mg SSS | Solute / alloying | DislocationLine defect enabling plastic slip at realistic stresses; edge vs screw are the two classic geometries. pinning | Industrial alloys | wt% Mg; GB β overrun |
| CNT intercalation | Intercalant | Non-covalent CT in channels | Emerging / specialty | Loading vs moisture / damage |
False-friend reminder: GaN:Mg is closer to Si:B (true electronic doping) than to Al–Mg SSS — but Mg is a deeper acceptor, needs H depassivation, and hits self-compensation harder than textbook shallow Si doping. Do not export Si shallow-donor formulas unchanged (Physics of doping open questions).
7. Bridge A add-on checklist (depth note §6) — filled
-
Where is the solubility / activation limit at process conditions?
Useful [Mg] ~1–3×10¹⁹ cm⁻³ with electrical activation after N₂ anneal; past ~2–3×10¹⁹ self-compensation / PIDs dominate. Equilibrium solubility talk is secondary to MOCVD incorporation + defect compensation. -
Intended mechanism?
Electronic doping — p-type holes via Mg_Ga. Name LED/laser / p-contact goal. Do not call Al–Mg SSS “doped.” -
What fails first past the window?
Typically self-compensation (p falls) and μ / optical quality — sometimes PID clustering — not “more Mg = more holes.” -
Metastability?
Yes — H passivation state is process-captured; anneal activates; later H-rich steps can re-passivate; thermal budget fights InGaN QW integrity. Three clocks apply. -
Only then — SPECULATIVE Bridge A mnemonic
See §8. Never invert: do not pick [Mg] from a linguistic arity table.
8. Under / useful / over — materials only (no linguistics yet)
| Zone | GaN:Mg reading |
|---|---|
| Under | Too few acceptors; or passivated Mg–H; p inadequate for injection / contacts |
| Useful window | [Mg] ~1–3×10¹⁹ cm⁻³ + successful N₂ activation → p ~10¹⁷ class; LED/laser p-GaN |
| Over / wrong regime | Self-compensation; PIDs/clusters; μ & optical collapse; over-anneal stack damage |
Honest split: incomplete ionization (deep acceptor) is baseline physics inside the useful window; self-compensation / PID clustering is overrun past it. Do not conflate them.
SPECULATIVE Cross-domain section. Open after the materials.
9. SPECULATIVE Bridge A overlay (supporting only)
Everything in this section is pedagogical analogy. It does not make doping × θ-roles a joint field.
Why this host is a clean industrial electronic-doping Bridge A case: Grant’s takeaway already flags GaN:Mg as industrial Mg doping for LEDs/lasers. Unlike research TiO₂ photocatalyst doping, the materials community’s word is unequivocally dopant, the product is billions of LEDs, and the under/useful/over curve (Hall p vs SIMS [Mg]) is a textbook concentration-window story — with H activation as a process co-actor. Cleaner industrial electronic-doping peg than oxide photocatalysis; complementary to the oxide note’s research electronic-doping peg and to Si:B’s shallow classical peg.
| Materials fact | SPECULATIVE Bridge A reading |
|---|---|
| Useful ~2×10¹⁹ cm⁻³ Mg + anneal → peak holes | Saturated intended electronic configuration for p-GaN devices |
| Undoped / passivated GaN | Under-filling relative to bipolar design intent |
| Excess Mg → self-compensation, PIDs, μ/optical loss | Overrun / failed regime: “extra” participants force compensation or clustering repair that misses the hole density you wanted |
| Continuous [Mg] / p vs discrete lexical arity | Asymmetry: concentration is continuous; verb arity is typically discrete — do not force integer slots onto Hall curves |
| Mg_Ga vs H vs V_N as distinct cast members | Do not smash into one “argument” type |
Earn: helps a linguist remember why SIMS ≠ Hall p, why anneal atmosphere belongs on the doping page, and why “more dopant” can mean fewer carriers.
Collapse: if Mg–H complexes are called θ-roles; if 2×10¹⁹ is treated as an arity integer; if self-compensation is equated with adjunct licensing; if Navajo morphological typing is treated as evidence for this count story; if a joint field is announced.
Navajo typed-participant half — parallel only (not proof)
Navajo morphology saturates typed participants on the verb. Bridge A is the count/windows half. GaN:Mg’s cast (host lattice, Mg_Ga acceptor, hole, H passivator, V_N compensator, anneal process) is a tempting typed-participant mnemonic — keep it parallel, never as evidence that Bridge A’s under/useful/over ladder is proven by Athabaskan morphosyntax.
10. Host contrast card (add GaN:Mg to the peg set)
| Host | Diagram / lattice | First “over” (typical) | Bridge A peg ( SPECULATIVE) |
|---|---|---|---|
| Al–Mg AA5083 | Limited solvus; FCCFace-centered cubic — many slip systems; Al, Cu, Ni typically ductile. Al + subst. Mg | GB β / sensitization | Solvus overrun → unwanted second phase |
| Cu–Ni | Isomorphous complete SSS | Conductivity / TCR curves | Property-curve overrun |
| Fe–C | Eutectoid Fe–Fe₃C; interstitial C | Cementite / martensite morphology | Transformation / morphology overrun |
| Ni–γ′ | γ/γ′ solvus; wanted precipitateSecond-phase particle that forms (often on aging) and blocks dislocations. | Overaging / TCP | Wanted γ′ vs wrong second phase |
| Mg (host) | HCPHexagonal close-packed — fewer primary slip systems; Mg, Ti, Zn often anisotropic.; AZ/AM/ZK | Brittle IMCs; galvanic β; ignition | β ambivalent; safety “over” |
| Ti (host) | HCP α / BCCBody-centered cubic — Fe/steel common; plasticity more temperature-sensitive. β; β-transus | α-case; ω; O/H embrittlement | Wanted α vs unwanted ω / α-case |
| Oxide electronic | Wide-gap oxide + subst. metal + V_O | Recombination; μ collapse; secondary oxides | Carrier / mid-gap window — true dopant (research-leaning) |
| CNT/graphene intercalation | sp² + guest in channels/gaps | Moisture dedoping; junction R | Intercalant loading window |
| GaN:Mg (this file) | III–V wurtzite + Mg_Ga acceptor (+ H) | Self-compensation; PIDs; μ/optical loss; failed activation | Carrier window on industrial electronic doping — true dopant |
11. What this walkthrough trains you to do next
For any new III–V / wide-gap dopant pair:
- Name the goal (LED injection / laser / E-mode gate / contact).
- Name substitutional dopant, passivating species (H), and compensating defects separately.
- Sketch Hall p vs SIMS [dopant] — find the peak, not max incorporated dose.
- Name the activation process and its stack constraints (QW thermal budget).
- Name the first failure past the peak (self-comp vs PID vs μ vs optical).
- Only then, if useful for memory, hang the SPECULATIVE Bridge A count peg.
Contrast vs oxide electronic note: same “dopant” honesty; maturity flip — GaN:Mg is the Grant takeaway’s industrial LED case; TiO₂ photocatalysis remains research-grade for many hosts.
12. Non-claims
- No joint field of doping × θ-roles / valencyArity: how many actants a predicate takes; formal vs semantic valency can diverge..
- Bridge A remains SPECULATIVE pedagogy / probe; lock unrevised.
- Navajo θ-morphology does not prove Bridge A count.
- Numbers herein are order-of-magnitude literature anchors, not fab recipes or LED process specs.
-
This walkthrough does not revise Bridge A’s lock, invent shared formalism, or announce a product.
13. Sources consulted (2026-09-19 note)
| Source | Role |
|---|---|
Grant metal-doping-takeaway.txt §§5–6, 8–9; Physics of doping §1–2; grant-doping-cast.md (not published here) |
Project cast; industrial GaN:Mg callout; acceptor framing |
| Kaufmann et al. Phys. Rev. B 62, 10867 (2000) | Hall p(N_A) peak ~6×10¹⁷ at ~2×10¹⁹; self-compensation |
| Nakamura et al. JJAP / APL activation lineage; Götz et al. APL 1996 | Mg–H passivation; N₂ anneal >600 °C |
| Nardo et al. J. Phys. D 55 (2022) laser vs RTA activation | Industrial RTA context; E_a~0.14 eV |
| Micovic / IOP passivation review lineage (Nanotechnology 2025 digest) | H passivation / reversible NH₃ re-passivation |
| arXiv:1607.08353; JAP 131, 185704 (2022) hot-wall MOCVD | Self-compensation (V_N, Mg_I); PIDs; free-hole limits |
| PMC8308141 heavily Mg-doped p-GaN | Saturation ~2×10¹⁹; clusters / PID notes |
| Hall / admittance Mg ionization energy literature (~135–160 meV class) | Deep vs Si:B (~45 meV) contrast |
| Project-internal: oxide electronic + CNT intercalation + metal host checklists; CONCLUSIONS-v0 + v1-addendum; chem-for-linguists; depth-solubility-windows; depth-process-metastability | Checklist discipline + contrast |
v0 — 2026-09-19 continue-research. Industrial GaN:Mg p-type electronic-doping walkthrough; Bridge A flagged SPECULATIVE. No joint field.
Corpus note: worked-gan-mg-doping-checklist-v0.md. Materials first; Bridge A peg last.