No joint field · Bridge A = SPECULATIVE pedagogy

Labels: SOURCED HYPOTHESIS SPECULATIVE OPEN NONE (REAL = same-domain materials or linguistics, shown like SOURCED)

Hub · six paths

Alloy-design literacy for a syntax reader

This site teaches how metallurgists and device people talk about hosts, solutes, dopants, windows, and failure. It does not announce a joint field of doping and theta roles.

Public pitch, from the content map: Alloy-design literacy for syntax/semantics readers, using Bridge A only as a SPECULATIVE count/window teaching aid — not as a joint field.

End point: vocabulary you can use on a phase diagram. Bridge A (concentration windows ↔ valency / saturation / over-arity) is a teaching ladder. It is not identity science.

Start here

New: Triskellion (the Grok design line, labelled speculative; where the ditransitive-verb idea started) · Node-movement view (checked data rows drawn as nested trees, a drawing not a claim).

The corpus reading order names three files before any host. That is the default path for a linguist.

  1. Chem for linguists — front door. Alloy-design literacy without pretending linguistics ≡ metallurgy.
  2. Doping vs alloying vocabulary — when to say dopant, solute, intercalant, or stabilizer.
  3. Mechanism taxonomy — name the family of “how a foreigner acts” before any under / useful / over ladder.

One sitting, still from that reading order: the front door, the vocabulary note, one electronic host (classical Si), one structural host (Al–Mg or Zr nuclear), then one cross-cut (same-element Mg or three clocks).

Six paths

Navajo, separate questions

Path E. Not Bridge A, and not one claim. The typed-participant map asks which participant types a mechanism family can wear. Derivational movement asks where arg1 is. Under the pronominal-argument derivation that was drawn, arg1 merges in Spec,vP and does not move; it is spelled out as the subject exponent inside the verb word. Spec,M as an EPP landing site is a flagged overlay the opened Hale essay rejects. Polar questions (da’ / -ísh) do not extract arguments. Mary ha’át’íí nayiisnii’ keeps the h-word in object position and 3–3 morphology on the verb. No yi/bi tree. The chemistry half is SPECULATIVE pedagogy only. Leaching and evaporation are the wrong parallel.

Lexicalization asks what is listed. The listed unit is the verb base (classifier + stem). The verb theme is the looser non-compositional prefix + base. Beside Grant’s coinage “phrasisized,” the literature’s terms are verb theme, lexicalized base, idiomatic preverb–verb, and incorporated postposition. Those are not English phrasal verbs. Hale: no verbal noun incorporation in Navajo. This page does not merge with movement or with the typed-participant map. The dopant guess stays SPECULATIVE and his.

Causative split is partial, and it is not Japanese lexical vs productive -sase. Hale & Platero 1996 name a simple transitivity alternation (ł only: boil, and fire / shatter / melt / rope from that primary) and a complex causative (ł plus y- plus special causee marking). Some ł is thematic, not derived. New (v0.2): Goertz et al. 2006, working with native-speaker consultants, print Navajo “analytic causatives,” but every one is a reason clause (‘because of the boy, I am talking’), where the cause is the object of the postposition -niinaa or a separate clause, not a causer argument. So in what is opened, a causer becomes an argument of the verb only through the ł morphology inside the word. WALS codes Navajo as having no morphological causative at all, with no source on the datapoint. New (v0.3): the same paper’s positional causatives, checked on the page images, put one stem through three event shapes, ‘he stood up’, ‘I stood him up’, and ‘I hold him standing’, so the ł causative can record a causer who keeps a state, not only one who brings about a change. Not a tree, and not merged with the other Navajo pages.

Phrasal prescription is a separate naming and unit map. The naming split is partly closed: Young 2000’s Theme→Base layering, as quoted, matches Smith, Perkins & Fernald; McDonough’s 1987 “base” report differs. Chee’s “phrasal verb” is a postposition plus a verb. It is not a dopant prescription realized by VPISH in a tense. The HIS frame stays SPECULATIVE. Do not merge it with lexicalization, movement, the causative scout, Zipf, or the VPISH molecule picture.

Inverse ranking is the sourced version of the argument-ranking edge. In Navajo’s yi-/bi- alternation the roles stay fixed, and animacy (with definiteness and topicality) decides which argument is topical. When the patient outranks the agent on animacy, the marked inverse is the good form and the direct form is degraded (Willie & Jelinek 2000). Creamer 1974 supplies the eight-group status ranking (persons down to abstractions) with equal-rank optionality and unequal-rank “not preferred” cells. Some alignments are starred outright. It is not a passive. Only single-wh questions are sourced, so Grant’s multiple-wh test stays open. No chemical correlate is claimed; the page lists what one would need.

Fourth person vs ranking tests whether the obviative “fourth” person is a second instance of that edge. It is not. Willie 1991 (§4.2.7) says the fourth person is definite, human, and “not involved in the animacy hierarchy.” When a lower-ranked agent acts on a human patient, all-third-person Navajo must use the inverse; making the patient fourth person instead keeps the direct form, with the agent still the subject. Willie & Jelinek 2000 add that the fourth cannot enter the direct/inverse alternation and can force a coreference the inverse stars. So the fourth is an escape from the ranking scale, not a flip of the alignment. Roles stay fixed. No chemical correlate; the page only adds one optional point to the search list.

Spinel inversion is the chemistry scout against that five-point list. Same two cations, charges fixed; a charge-independent site-preference scale; inverse preferred when the scales disagree (Fe₃O₄, NiFe₂O₄, CoFe₂O₄). Passes points 2, 3 and 5; partial on 1 and 4; crystal level, not molecule. Candidate, not a correlate. SPECULATIVE correspondence only.

Spinel dopant site preference tests the spinel candidate against the sixth, fourth-person point: a way to take one participant out of the ranked set so the default survives. d⁰, high-spin d⁵ and d¹⁰ ions have no crystal-field site preference, so they are outside that scale. But size, covalency and electrostatics still rank them: Zn, Cd and Mn ferrites stay normal, MgFe₂O₄ and MgGa₂O₄ stay inverse. So the sixth point at most partly passes. Dopants: Ni²⁺ (d⁸) takes the octahedral site against the charge default; Co²⁺ and Mn²⁺ follow size. Price, Price & Burdett sort 98% of spinels without counting d electrons. Crystal level; candidate, not a correlate. SPECULATIVE correspondence only.

TBP apicophilicity ranking edge is the first molecule-level test of the six-point list. Phosphorus with five groups has two kinds of slot, the two “poles” (apical) and the three around the “waist” (equatorial), and chemists rank groups by how much they prefer the poles (apicophilicity: electronegative, small, low π-donating). In a substitution, the incoming and outgoing groups use the poles (Westheimer 1977). When the ranking puts the outgoing group in the wrong slot, the molecule reshuffles first (Berry pseudorotation), and the product records it: retention instead of inversion. A bulky ring can make the rule-breaking arrangement the stable one; cage ligands make some rearrangements impossible. Points 1 and 5 pass on the event reading, 2 passes weakened, 3 and 4 are partial, 6 leans fail. Unlike the Navajo inverse, the reshuffle serves the event roles rather than overriding them. Candidate, not a correlate. SPECULATIVE correspondence only.

EAS directing ranking edge runs the benzene directing rule against the same six-point list, with textbook and primary sources. When a new group replaces a hydrogen on a benzene ring, the groups already there decide where it goes, by their kind (a lone pair or a positively polarized atom next to the ring), not by which atom they are. A separate activation scale decides between two groups that disagree (“the more powerful activating group has the dominant influence, but mixtures of products are often formed”); bulk lowers the ortho share step by step, to zero with tert-butyl on both sides; the spot between two meta groups is made “by some other route”; and an attack on an occupied position can leave no trace in the product. Points 1 and 5 pass, 2 passes weakened, 3 and 4 are partial, 6 leans fail. Unlike Navajo animacy, the scale sits on bystanders and ranks places for one incoming group, not participants for roles. Candidate, not a correlate. SPECULATIVE correspondence only.

Dopant produce vs maintain takes the causative scout’s logged question to metal alloys: do metallurgists, in their own terms, separate a dopant’s job of producing a configuration from its job of keeping one? In part, yes. Inoculant particles and growth-restricting solute make fine grains while the melt freezes; pinning particles, solute drag and segregation hold a grain structure later (“inhibit”, “stabilize”, “preserving the fine-grained structure”); Cottrell & Bilby estimate an atmosphere’s “rate of formation” apart from its “strength of pinning”. The line is drawn by stage and physical form, not by element: scandium does both only as two Al₃Sc populations at two stages, and coarse vs fine particles split the two jobs by size. The nanocrystalline literature splits keeping in two: holding against a driving force that remains (kinetic) vs removing it (thermodynamic). Navajo marks bring-about vs keep on one stem with one causer. Analysis only; no chemical stand-in for a theta role. SPECULATIVE correspondence only.

Dopant DSA / Portevin–Le Chatelier takes up that note’s open question on dynamic strain ageing: inside one deformation, solute atmospheres form, pin dislocations, unlock, and re-age during waiting times at obstacles. The literature keeps formation rate apart from pinning strength; DSA is the diffusional cycle and PLC the macroscopic serrated bands (DSA can run in a wider window than the instability). Al–Mg (AA5083-class) and Fe–C are the opened hosts. Analysis beside Navajo bring-about vs keep; no chemical stand-in for a theta role. SPECULATIVE correspondence only.

Dopant thermodynamic stabilizer: hold or produce takes up another open question from the produce-vs-maintain note: a solute that sits on grain boundaries and lowers their energy toward zero (the “thermodynamic” route) is usually described as stopping grain growth, but is it ever described as making grains finer? In two opened lattice simulations (Hussein & Mishin 2024, 2025), yes: the same solute breaks a single crystal or a large grain into small ones and stops coarsening from the other side, toward one state, but only if the solute can keep up with the moving boundary; the stable state is a balance of grains forming and vanishing. In the one opened experiment (Liu et al. 2024, Ni in TiZrNbHf films), deformation makes the new grains and Ni moves to them and keeps them. The simulation authors say full stabilization has not been shown experimentally. Analysis beside Navajo bring-about vs keep; no chemical stand-in for a theta role. SPECULATIVE correspondence only.

Si donors: slot or element? takes a first published test of a new open question: is a doping outcome set by the kind of site (the slot) rather than by which element fills it? For shallow donors the physics says yes in principle (the effective-mass binding depends only on the host) and it holds in GaAs; in silicon the excited states of P, As, Sb and Bi stay near the slot-only value, but the ground state changes strongly with the element because the elements differ at the core (Saraiva et al. 2014). A shared slot template plus a core-local element correction. This is about the result of doping, not about a process choosing sites. No chemical stand-in for a theta role. SPECULATIVE correspondence only.

Navajo findings vs materials: a scorecard takes the four Navajo findings that survived testing on Navajo Decoded (a role-bearing slot filled by feature match; ranking by general capacity; layered assignment, Grant’s conjecture; viewpoint and causation in the word, telicity not) and sets each beside a sourced materials analog, with verbatim text and a counter-case, graded STRONG, WEAK or NONE. Strongest: the geochemists’ lattice-strain model, where a crystal site has its own ideal radius, stiffness and maximum partition value and an ion enters through its radius and charge (Blundy & Wood 2003; Shimizu et al. 2017); and the failure of a context-free rank, since the “smaller ion first” rule is invalidated and the same two ions swap order when the site size changes. Weak: layered assignment (Wyckoff-position prototypes with the species written last describe a structure but do not show an order of decision; Pauling’s rules 2–5 hold together in 13 % of about 5000 oxides). None found: the speaker-interest tie-break, the moving-object rule, the event-structure layer, telicity. No correlate declared. No chemical stand-in for a theta role. SPECULATIVE correspondence only.

Solute strengthening: host part vs element part is the first metal-host test of the split found for silicon donors and the lattice-strain model. Solute atoms strengthen a metal by getting in the way of dislocations. In the modern elastic theory the host (or, in a high-entropy alloy, the averaged alloy) supplies the lattice constant, elastic constants, line tension and core structure, and each solute supplies only its concentration and its misfit volume (Yin et al. 2020). For Mg, Yasi, Hector & Trinkle 2010 find the silicon-donor shape: a long-range size term set by the host’s dislocation field and an element-specific term at the core, the size term dominant. Strength rises as the square root of concentration (dilute) or the two-thirds power (collective), peaks mid-composition in a concentrated binary (Mo–Ti), and is capped by solubility, which falls as misfit and potency grow. The solute’s number is measured in the host, not carried by the element. The earlier pairing of size and modulus misfit with the lattice-strain site radius and stiffness was mis-mapped: those two belong to the site, and the correct partners are the host’s size and stiffness. No chemical stand-in for a theta role. SPECULATIVE Bridge A shapes only, after the materials.

Misfit volume: does the element carry its size? takes up that page’s open question: is the size number a solute brings (its apparent volume, and the misfit built from it) the element’s own, or set by the host? Two layers. The misfit is the element’s apparent volume minus the alloy average, so it is host-relative by construction: vanadium has nearly the same volume in fcc and bcc alloys, yet is large among fcc partners and small among bcc partners, with a large misfit in both (Yin, Maresca & Curtin 2020). The apparent volume itself carries well within one family of similar metals in one crystal structure: Ni and Cr in NiCoCr match Ni₂Co and Ni₂Cr (Yin et al. 2020), and Vegard’s law with the elements’ own volumes comes within a few percent of first-principles misfits for most elements in two high-entropy alloys. But it is not a constant of the element: vanadium reads 12.20 ų dilute in Ni and 13.8 ų at one-third, cobalt shrinks with magnetism, apparent volumes spread by 20–39 % across unrelated binary hosts (the note’s arithmetic on King’s size factors as restated by Landa et al.), and in 3d alloys the local misfit follows charge transfer rather than size (Oh et al. 2019). This refines the earlier page’s “not a portable feature bundle”. No chemical stand-in for a theta role or a feature bundle. SPECULATIVE Bridge A shapes only, after the materials.

Merge/Agree mid-level settings: any chemistry analog? takes Grant’s typology note on mid-level Merge/Agree settings (Rudin’s [±MFS], Multiple Agree, feature inheritance, Edge Features, phasehood) and asks whether chemistry or materials has a sourced structure of the same shape. One STRONG arrangement-level analog: sympathetic nucleation, where a precipitate crystal already formed becomes the site where the next crystal of the same phase nucleates, “successively”, each separated from its substrate by a boundary, and eutectoid alloy systems do it readily while aging systems (Al-base) rarely do (Aaronson et al. 1995), the shape of [+MFS]. Trans-influence in metal complexes is WEAK for feature inheritance; one centre set by all its ligands is WEAK for Multiple Agree; Edge Features and phases got NONE (the kink-site source could not be downloaded). A later section takes three questions from Grant’s astrophysics handoff: path dependence is STRONG, a retained readable record is STRONG (the Al3(Sc,Zr) core/shell particle), and a boundary that freezes the interior while the surface stays reactive is only WEAK. Arrangement only; no mechanism shared. SPECULATIVE correspondence. Not Bridge A.

Fleischer–Labusch crossover: when does one solute pin alone? takes up the open question left by the two solute-strengthening pages: strength grows as the square root of concentration when each solute is a separate obstacle that a dislocation hangs on (strong pinning; Friedel, Fleischer), and as the two-thirds power when the line is held by chance clumps of many solutes at once (weak, collective pinning; Labusch). Where does one give way to the other? The old zero-temperature estimate put the switch near a solute fraction of 10⁻², which is why strong pinning was assumed for low concentrations. But one solute holds the line with only about 100–200 meV, which heat easily overcomes, so at finite temperature collective pinning takes over far lower: below 10⁻⁴ above 78 K for aluminium and magnesium alloys, in the Varvenne, Leyson, Ghazisaeidi & Curtin 2017 overview’s restatement of Leyson & Curtin 2013 (whose own full text was not opened). “The Friedel model rarely applies for moderate solute/dislocation binding energies”: STRONG. Strong pinning is kept only for solutes that change the dislocation core: WEAK. The square-root law used on the earlier page (Yasi et al.) is, per the overview, a zero-temperature model for ranking solutes, so that page’s reading is corrected. No softening from too much solute inside an fcc solid solution: NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Concentrated alloys: one solute alone or many together, with no host? asks whether that question even arises in high-entropy alloys, where five or so elements sit at about 20 percent each and no element is the host. The leading theory (Varvenne, Luque & Curtin 2016, authors’ pre-print) invents an average metal and treats every real atom as a “solute” in it, so the alloy is “a random solid solution with 100% solute concentration in the average matrix” and strength comes from collective fluctuations: the old dilute question does not arise in that form, STRONG. An independent estimate (Zhai & Zaiser, arXiv 2018), which defines strong and weak pinning by the line’s shape rather than by how high the stress is, puts the Cantor alloy well inside the collective regime, STRONG, and a refractory bcc alloy at the borderline, WEAK. But the question comes back one level down: simulations of a fully random CoCrFeMnNi show individual pinning points, cobalt atoms with chromium-rich surroundings, and only the strongest survive at room temperature (Utt et al. 2022): STRONG as simulated, WEAK as what sets engineering strength, and no opened source says the collective picture is overturned: NONE. Strength is not set by the number of elements. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

bcc concentrated alloys: one by one or collective, when screws matter? takes up the case that page left open: body-centred-cubic refractory alloys, where pure metals are held back not by solutes but by the stiff “screw” kind of dislocation line, which sits in deep troughs of the crystal and moves by throwing out small steps (kinks). The question splits by the kind of line. For the flexible “edge” line the fcc answer carries over: every atom is a “solute” in an average metal, the pinning is collective, the energy barriers are very high (about 1.9–2.5 eV, Maresca & Curtin), and neutron diffraction and electron microscopy find edge lines dominating deformation in NbTaTiV and CrMoNbV (Lee et al. 2021): STRONG; whether edges also control MoNbTaW is contested by simulations that keep its edges 11–20 times faster than its screws (Yin et al. 2021): WEAK. The screw line does not fit the question as asked, since it does not bow around pins like a flexible line: simulations show it locked by discrete cross-kinks and jogs (Chen et al. 2020; Yin et al. 2021), STRONG as simulated, inside a random field that the screw theory treats as collective (read as its abstract only, full text embargoed), WEAK; the two simulations disagree on whether kink trapping or ordinary kink-pair nucleation controls screws, WEAK. No opened source settles the refractory alloy left at the borderline: NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Screw-controlled NbTiZr: the cross-kink plateau and its high-temperature loss follows up the screw half of that page with an experiment. In an alloy of niobium, titanium and zirconium in equal parts (NbTiZr), tension tests and electron microscopy show the stiff screw lines controlling flow: STRONG. Strength falls steeply from 77 K to room temperature, then stays almost flat from 300 K to 873 K while the screw lines carry cross-kinks (knots where kinks on two planes have collided) and debris: STRONG as observed; that the knots are what hold strength flat rests on one alloy and a theory fitted to other data: WEAK. At 1073 K strength drops and the lines are straight, with no knots and no debris: STRONG as observed. So the high-temperature loss of the screw locks, only predicted on the earlier page, is now seen in one alloy, but that missing atoms (vacancies) are what dissolve the knots is not measured: WEAK. The paper says edge lines add little strength here: WEAK, its numbers are in a supplement not opened. Still no opened source splits screw and edge strength for TaNbHfZrTi: NONE; quantum-mechanical calculations show its screw cores change with the local mix of atoms: STRONG as computed (Bhatti et al., arXiv 2024). No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

HfNbTaTiZr: where the screw plateau would be, and what is measured asks the same temperature question of the well-known five-element alloy of hafnium, niobium, tantalum, titanium and zirconium in equal parts (HfNbTaTiZr). No opened source splits its screw and edge strength or images cross-kinks in it: NONE. But two opened data sets give a temperature series. In compression, strength falls from 1670 MPa at 77 K to about 1150 MPa at room temperature, and the authors put the end of the steep cold zone well above room temperature: STRONG as measured (Chen H. et al. 2021). In tension, strength is still falling at 573–673 K: STRONG (Chen S.Y. et al. 2019). At 873 K and 1073 K compression strength is nearly the same (481 vs 461 MPa): STRONG as measured, but that is two points with nothing measured between room temperature and 873 K, so a plateau is WEAK. At 1073 K the alloy already deforms the hot way (it softens while deforming, grains stay rounder, no kink bands), with no strength drop like NbTiZr’s: STRONG as observed. Between room temperature and 673 K the tension curves show dynamic strain ageing, with serrations at 673 K: STRONG as observed; that it is ageing by a dilute solute is the authors’ attribution: WEAK. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Trace oxygen and nitrogen leave the solution: age softening in TiVNbTa takes up that page’s question of which dilute solute is moving, using a different alloy, which it says plainly: no opened source names the solute behind HfNbTaTiZr’s serrations: NONE. In an alloy of titanium, vanadium, niobium and tantalum in equal parts (TiVNbTa), held at 700 °C for up to 40 days, hardness falls about 16%, from 5.05 to 4.22 GPa: STRONG as measured. Over the same time, small titanium-rich particles of a different crystal structure (hcp α-Ti) form, first at grain boundaries and then inside grains, and they carry the alloy’s trace oxygen and nitrogen (atom probe: 18.3 at.% oxygen, 5.88 at.% nitrogen in the particles): STRONG as observed; the oxygen, nitrogen and carbon left in the surrounding crystal halves, from about 1.2 to about 0.6 at.%: STRONG as measured, though the table is in a supplement not opened. The authors blame the softening on the interstitials leaving the solution, not on the particles, which they calculate add only about 5% of the hardness: WEAK, a model estimate. The bulk oxygen measured by gas analysis is about four times lower than the atom-probe value, which the note flags. No opened source yet varies oxygen on purpose: NONE; the candidate is found but not opened. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Oxygen added on purpose: a dose window in refractory high-entropy alloys takes up the question that page ended on: if oxygen is added on purpose, how much does such an alloy take while staying one crystal and getting stronger, and what happens past that? The journal paper that varies oxygen in a TiZrHfNbTa alloy (Iroc et al. 2022) still could not be opened, so the page reads the master’s thesis behind it, said plainly, and the paper’s abstract, which differs from the thesis on density and on oxides at room temperature. In the thesis, an alloy of titanium, zirconium, niobium, hafnium and tantalum (Ti25Zr35Nb20Hf5Ta15) is cast with no added oxygen and with a nominal 2 at.% oxygen, added as titanium oxide powder: the oxygen alloy stays one disordered crystal (bcc): STRONG as observed; how much oxygen it really holds was not measured: WEAK; its hardness rises from 321 to 440 HV, about 37%: STRONG as measured; it yields at 1240 ± 50 MPa in compression and does not break by 35% strain (strain only relative): STRONG as measured; but the oxygen-free alloy’s yield was not measured, so the yield gain from oxygen is NONE in this source. Where the window ends is found in a neighbouring alloy, NbTiZr, with oxygen pushed in from air at 1050 °C (Beaudry et al. 2024): in simulation the metal atoms stay nearly random up to 2 at.% oxygen and then start to group: STRONG as computed; at 6.7 at.% oxygen the metal is still one bcc crystal but holds tiny oxygen- and zirconium-rich patches and is 3.7 times as hard as oxygen-free NbTiZr: STRONG as measured; near a computed 8–10 at.% it splits into two phases, then oxides form: STRONG as computed. Where the alloy stops being ductile is not measured in any opened source: NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Where added oxygen stops helping ductility: a tensile ladder in Ti–Zr–Nb takes up the question that page left open: at what oxygen level does such an alloy stop being able to stretch? No opened source runs that test in TiZrHfNbTa, so the page reads two open papers on related alloys, said plainly. In cast bars of titanium–zirconium–niobium (Ti–30Zr–14Nb) pulled in tension (Jiao et al. 2023), 1 to 3 at.% added oxygen makes the metal both stronger and more stretchable, elongation rising from about 9.5% to about 25.1% at 3 at.%: STRONG as measured (the doses are nominal; the measured contents are in a supplement table not seen). Past 3.5 at.% the bars break in a brittle way: STRONG as measured; at 4 at.% oxygen is found piled up at the grain boundaries, with no oxide formed: STRONG as observed; that this pile-up is the cause is the authors’ reading: WEAK. The ceiling depends on the host: with more niobium it comes sooner, and in a two-phase version with less niobium any oxygen hurts: STRONG as observed. A hafnium-bearing check (Wang et al. 2025) shows the same: one Zr–Hf–Ti–Nb alloy keeps gaining ductility up to 1.8 at.% oxygen, while a less stable one loses most of its ductility at 0.5 at.% through brittle ω particles: STRONG as measured. So the useful window ends well before the dose at which the previous page’s NbTiZr splits into two phases (a comparison across different alloys: HYPOTHESIS). The ceiling in TiZrHfNbTa itself: NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Oxygen in a Ta-bearing five-element alloy: one measured tensile pair opens the 2020 paper that page pointed to, hoping it would repeat the closed 2018 numbers for TiZrHfNb. It does not: NONE. But it pulls a tantalum-bearing alloy (Ta0.5Nb0.5HfZrTi: tantalum and niobium at half the share of hafnium, zirconium and titanium) in tension at two oxygen levels, and it measures the oxygen: 0.377 at.% in the “base” alloy (none added, but not none present) and 2.241 at.% with 2 at.% added (Lei et al. 2020). With the extra oxygen, strength rises from about 810 to about 1138 MPa and stretch before breaking from about 10.8% to about 20%, and the alloy stays one bcc phase: STRONG as measured (one curve per alloy). The share of oxygen in ordered clusters (~18.5%) and their role in the gain are fitted or attributed: WEAK. So in this tantalum-bearing alloy the ductility ceiling lies above about 2.2 at.% oxygen (STRONG as a lower bound), and where it is remains NONE: two levels are not a ladder. The abstract’s “~1400 MPa” with “~20%” joins two different alloys’ numbers; the page flags it. The source shares authors with the Ti–Zr–Nb paper, so the two do not confirm each other independently. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Oxygen in HfNbTaTiZr: measured 0.27 vs 3.24 at.%, stability on ageing goes one oxygen step higher in a tantalum-bearing alloy. The open paper it reads (Poulain et al. 2024) measures the classic equal-shares HfNbTaTiZr at 0.27 and 3.24 at.% oxygen. Held at 900 °C for 1000 hours in clean argon, the low-oxygen alloy stays one bcc phase (STRONG; repeated twice). The high-oxygen alloy breaks down into about 15 vol.% of an oxygen-rich HCP phase and about 35 vol.% of a niobium–tantalum-rich second bcc phase, with HCP covering the grain boundaries (STRONG as observed). One attempt to cold-work the high-oxygen bar by rotary swaging cracked it at about 40% reduction (WEAK: a single remark in the acknowledgments). The paper has no tensile test, so the tensile ceiling is still NONE; the same group’s ageing-plus-tension paper could be read only as its abstract, which has no numbers. Beside the previous page’s alloy (still ductile at 2.24 at.%), an edge between about 2.2 and 3.2 at.% oxygen is a HYPOTHESIS across two alloys and two different tests. Ageing adds a time axis: an amount that is fine as cast may not be fine after a long hold. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

HfNbTaTiZr after hot isostatic pressing: an HCP phase and no tensile stretch takes up the question that page left open: when this alloy loses its single crystal type, does it lose its ability to stretch? The paper that ages it and pulls it in tension (Zhao et al. 2025) is still blocked, so the page reads an open paper from another group (Larianovsky et al. 2017), said plainly. Bars of the classic equal-shares HfNbTaTiZr are cast two ways and then squeezed hot under gas pressure (hot isostatic pressing, 1230 °C) to close pores. As cast they are one bcc crystal type; after pressing a second, zirconium- and hafnium-rich HCP phase is present, spread evenly in one casting and mostly on the grain boundaries in the other: STRONG as observed. Pulled at room temperature, five bars broke at 669–749 MPa with no stretch the machine could detect, cracking between grains: STRONG as measured; squeezed, the same metal yields at 1350–1888 MPa: STRONG as measured. So a two-phase state of this alloy can be fully brittle in tension, and good compression numbers do not promise tensile ductility. The paper does not measure oxygen: NONE; that oxygen made the second phase is a HYPOTHESIS, and that the second phase caused the brittleness is a HYPOTHESIS too (the authors leave the cause open). The long-ageing case stays NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

HfNbTaTiZr at 873 K: nanoscale hcp and a stress rise in one batch (compression only) takes up the question that page left open in the one form an open paper answers it: test one batch of the alloy with and without the second phase. The paper (Yasuda et al. 2025, an independent group) does this in compression, not tension, said plainly. The alloy is first made one bcc crystal type at 1473 K and quenched. Held at 873 K (600 °C), it quickly separates on a scale of about 10 nm into a zirconium- and hafnium-rich bcc and a niobium- and tantalum-rich bcc, and then tiny hcp crystals, under 10 nm, form inside the zirconium–hafnium-rich layers: STRONG as observed. Hardness is already high after 0.2 h at that temperature: STRONG as stated. In the same coarse-grained bars, a test after an 1800 s hold shows a clear rise in the stress needed to squeeze the metal at 873 K, and a test after fast heating and a 120 s hold, too short for the hcp to form, shows almost none: STRONG as a same-batch contrast. So the nanoscale hcp, not the temperature or the separation alone, makes the alloy stronger: STRONG within compression. The test itself changes the specimen while it waits at temperature. Oxygen is not measured: NONE. What the nanoscale hcp does to stretch in tension: NONE in this source; that it lowers it, as the coarse hcp after pressing did, is a HYPOTHESIS (a different state). Read with the earlier pages, the clean alloy is unstable roughly between 500 and 800 °C and stable at 900 °C unless oxygen is present: HYPOTHESIS. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

HfNbTaTiZr between 550 and 1000 °C: where the unstable band ends (two edges, not one) asks where that page’s unstable temperature band ends, using one of the two papers it named (Chen et al. 2019, the authors’ accepted manuscript; an independent group). The other paper (Yasuda et al. 2021) was blocked, so only its abstract is used, said plainly. The alloy is first made one bcc crystal type at 1200 °C and quenched, then held for 96 hours at 550, 700, 900 or 1000 °C in sealed, evacuated quartz tubes. At 1000 °C it stays one bcc type; at 900 °C the bcc X-ray peaks split; at 550 and 700 °C it holds two bcc types and a zirconium- and hafnium-rich hcp phase: STRONG as observed. At 700 °C particles appear on the grain boundaries within 2.5 hours and make up about 41% of the volume after 96 hours: STRONG as stated. With the 2021 abstract (split at 773–1073 K, hcp at 773–973 K: WEAK), the single band becomes two edges: the bcc split reaches somewhere between 900 and 1000 °C, the hcp only to about 700–800 °C. These are brackets between tested temperatures, not measured points. Neither lab measures oxygen: NONE. The 900 °C split in a vacuum tube stands against the clean single bcc another lab found at 900 °C in argon; oxygen picked up in the tube could explain the difference: HYPOTHESIS, not counted against either. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Ti–Zr–Hf–Nb–Ta with Fe micro-alloying: a ductility window set by nano Hf₂Fe at grain boundaries takes up that page’s open question of a real dopant window in this family. The paper (Li et al. 2025, Beijing Institute of Technology) opens as Europe PMC full-text XML; the MDPI PDF was blocked, said plainly. The host is not the equiatomic alloy of the earlier notes: it is Ti₁₅Zr₁₀Hf≈₃₆Nb₁₀Ta₂₅ with Fe at 3.35–3.5 at.%, a true micro-alloying addition rather than a principal element or an interstitial like oxygen. Ageing puts nano Hf₂Fe on the grain boundaries. Across the set, elongation runs from about 19% down to about 2.65% while strength stays between about 955 and 1081 MPa: STRONG as a measured span in this host. A lean Fe dose and a short age that leaves precipitates below what SEM/TEM see sits near 1037 MPa with about 14% elongation: STRONG as measured; that those precipitates exist below the TEM limit is a HYPOTHESIS. Once a continuous grain-boundary network forms, both strength and ductility collapse: STRONG. The same window in equiatomic HfNbTaTiZr: NONE. Oxygen is not measured: NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Ti–Zr–Hf–Nb–Ta–Fe baseline (Liu 2022) opens the HF1 baseline paper that page left as open question 4. Liu et al. 2022 (same BIT group; IOP PDF, CC BY 3.0) names Base and F2.5 (exact at.% not printed). F2.5-ST yield about 971 MPa with about 17.5% ductility, and F2.5-500-30 yield about 1081 MPa after continuous ~25 nm grain-boundary Hf₂Fe, are STRONG as primary prose and match Li’s restated HF1 yield and ST elongation. Aged elongation as a percent is not in Liu’s prose, so Li’s 7.5% stays WEAK. Whether F2.5 is Li’s HF1 (Fe₃.₅) stays OPEN. Liu’s continuous ~25 nm versus Li’s discontinuous ~5 nm for the 500-30 condition is a STRONG text conflict. Oxygen: NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Ti–Zr–Hf–Nb–Ta base printed (Yang 2025, EBM) takes up the question that page left first: what composition did the Fe papers start from? Liu 2022 never prints it. A later paper from the same BIT group, about melting the alloy with an electron beam (Yang et al. 2025; IOP PDF, CC BY 4.0), prints the group’s iron-free Ti–Zr–Hf–Nb–Ta as 15 Ti, 10 Zr, 40 Hf, 10 Nb and 25 Ta atomic percent: STRONG as printed. That is exactly the iron-free end of Li’s formula Ti15Zr10Hf40−xNb10Ta25Fex, so the iron goes in where hafnium comes out: STRONG as arithmetic. That Liu’s “Base” is this alloy is a HYPOTHESIS; no paper says it in words. The code “F2.5” is neither 2.5 at.% nor 2.5 wt% iron if F2.5 is Li’s HF1 (about 1.4 wt%; arithmetic), so what it means stays OPEN. The paper also measures oxygen: 276 ± 12 parts per million by mass in its “Raw material” (about 0.25 at.% by the note’s own conversion: WEAK) and 53–71 after melting: STRONG as tabulated. But it does not say what the raw material was, and electron-beam melting boils off all the titanium, so oxygen in the iron-bearing alloys stays NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Oxygen in a Hf-bearing refractory host takes up the question the iron pages keep leaving open: how much oxygen is in the iron-bearing alloys? No open paper measures it, so this page bounds it instead. A Texas A&M / UC Irvine preprint (Huang et al. 2025; arXiv, CC BY 4.0) melts Nb45Ta25Ti15Hf15 from metals at least 99.95% pure. With no oxygen added, the alloy still holds about 400 parts per million by mass, about 0.3 at.%: STRONG. Above about 0.5 at.% oxygen, hafnium oxide (HfO₂) shows up, and at 1 at.% it sits on the grain boundaries; every button with oxygen added cracked before 10% cold rolling: STRONG as stated. The paper gives its oxygen limit three ways (0.5, 0.5–1, and 0.8–1.0 at.%), and its own computed numbers do not strictly support the 0.8 lower bound (arithmetic), so one number is WEAK; only the band from about 0.5 to 1 at.% is STRONG. An earlier paper from the same circle (Belcher et al. 2025, CC BY-NC) says oxygen that starts in the metal charge is not removed by arc melting unless something reduces it, and that Ti- and Zr-bearing melts pick up more: STRONG as stated. With two other labs’ numbers already in the corpus (0.27 and about 0.25 at.%), about 0.25–0.3 at.% looks like the as-melted scale: HYPOTHESIS. The iron-bearing host’s own oxygen stays NONE, and whether its 40 at.% hafnium raises or lowers the oxide limit is OPEN. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Carbon micro-addition and grain-boundary oxide in a Hf-bearing host follows that page’s open question of which way hafnium moves the oxide limit, and does not answer it, but adds a second variable. A Shanghai preprint (Shen et al. 2025; Research Square, CC BY 4.0) arc-melts Nb44Ta44Hf10W2 with 0 to 1 at.% carbon. Without carbon the alloy holds only 99 parts per million oxygen by mass, about 0.09 at.%: STRONG. Even so, its grain boundaries carry a dense layer of hafnium oxide that grew from the last liquid to freeze, and it breaks along those boundaries at 0.33% elongation: STRONG as observed; the growth path is the authors’ calculation, WEAK as the real path. That is about a fifth of the previous page’s 0.5 at.% oxide edge, so a solid-solubility limit and an as-cast boundary edge look like different numbers: HYPOTHESIS. With 0.2 or 0.4 at.% carbon the alloy stays brittle; from 0.6 at.% it stretches 11.4–13.3%, best at 0.8 at.% (yield 682.2 MPa, tensile 807.0 MPa): STRONG as tabulated. At 0.8 at.% carbon, fine carbides replace the oxide on the boundaries: STRONG as observed. Carbon and oxygen move together in this ladder, so carbon acting alone is OPEN; the iron-bearing host’s oxygen stays NONE. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

Boron inside the grains vs on the boundaries goes back to the iron pages’ open question: is there a small-addition series in the equal-parts five-element alloy itself? One paper turned up with an open licence (Wu et al. 2026, Acta Materialia, CC BY): 30 atom-parts-per-million of boron in equiatomic TiNbZrHfTa goes to the grain boundaries together with zirconium, and the yield strength falls from 920.7 to 869.8 MPa. Its full text was blocked, so this is from the abstract only: WEAK, and it is one step, not a series. The open full text with a readable series is An et al. 2026 (PNAS, CC BY-NC-ND) in a zirconium-free Nb45Ta27Ti13Hf15: 1.5 at.% rhenium raises yield strength from 722 to 866 MPa, and 0.3 at.% boron on top raises it to 968 MPa, while elongation goes from 21.5% to 19.7%: STRONG as tabulated. There, boron was found spread through the grains, not on the boundaries: STRONG at the resolution used, WEAK as “none on boundaries”. Same element, opposite sign, different place: HYPOTHESIS. Oxygen was never measured in An’s alloys (NONE), and no boron-only alloy was made, so whether rhenium and boron cooperate or just add is OPEN. No chemical stand-in for a theta role or a slot. SPECULATIVE Bridge A shapes only, after the materials.

How much oxygen the route puts into equiatomic HfNbTaTiZr takes up the oxygen page’s last open question: how much oxygen does a host with both titanium and zirconium pick up? A Czech–Korean paper (Málek et al. 2019, Materials, CC BY) makes the equal-parts five-element alloy from powders by ten routes and measures the oxygen in each. Blended metal powders, pressed cold, already hold 0.55 wt.% (about 3.9 at.%), and hot isostatic pressing adds more; 42 hours of ball milling ends at 1.07 wt.% (about 7.4 at.%) with hafnium-oxide particles; gas-atomized powder sintered by spark plasma ends at 0.12 wt.% (about 0.9 at.%): STRONG as tabulated. Only that cleanest specimen is a single phase with no pores, and only it survives cold twisting under pressure without cracking: STRONG as observed. That oxygen alone causes the others to crack is a HYPOTHESIS, because porosity and extra phases rise with it. Whether the cleanest one holds any hafnium oxide is WEAK: the text says no, a figure caption hints yes. The source gives hardness, not strength. Nothing here transfers to the melted iron-bearing alloys. Bridge A remarks come after the materials and are SPECULATIVE.

Theta-phenomena tracker is a running list, started after the Morphic Tessellations bot's report of a "crossing flip" in Algonquian inverse voice, of things languages do with theta roles, each with a separate chemistry label. The first new entry is the Algonquian inverse. In Cree, Ojibwe, Passamaquoddy and Blackfoot the inverse never changes who did what to whom. When a third person acts on "me" or "you", it is obligatory and changes only the agreement (Oxford 2023, 2019). When one third person acts on another, it is an optional voice that raises the patient over the agent while keeping the agent, so it is not a passive (Oxford 2023; Bruening 2005). Blackfoot ranks by sentience (Bliss 2005). Navajo yi-/bi-, fourth person, distance as measure, the instrument bee and the causatives are also entered. Chemistry relevance is NONE for most rows. Where a row is marked SPECULATIVE, it is only a sharper test specification for the existing ranking-edge scouts (spinel, TBP, EAS) and the producing-vs-maintaining scout. No chemical stand-in for a theta role.

VPISH, three slots

Path E. VPISH molecule picture is a separate page. It does not merge with movement, lexicalization, the causative scout, or Zipf. SPECULATIVE throughout the correspondence. Ordinary English, as Poole’s 2021 handout draws it, is the clear hop: a lexical subject starts in the verb phrase and moves to the specifier of the tense phrase, still inside the clause. Only that case is likened to an in-solid step, and the only mechanisms are an interstitial hop, a vacancy hop, and segregation that stays in the solid. Navajo on the drawn analysis, Belfast English as that handout reports it, and Welsh as that handout reports an argument, leave the subject in the verb phrase, so they are not given a diffusion or segregation step. Navajo is the edge case, not a failed English hop: nesting, the subject exponent on mode, and theta roles rather than lexical items. Chemistry has no stand-in for a theta role. None is invented. No joint field. Bridge A is a different strand.

Zipf, not Bridge A

Path E. Zipf is ranked frequency, not a joint field, and not Bridge A. REAL in the field that measured it: classic Zipf is ranked word frequency, about 1/r, not exactly −1. Fixed-length n-grams are shallower than −1. Williams et al. 2015 (Scientific Reports 5, 12209, srep12209) put mixed-length phrases near −1 over a long rank range. Their phrase is a random partition, not a phrasal verb. They call morphemes the atoms of language and a mix of words and phrases the molecules. That is a linguistic remark, not a chemical measurement, and it is not the word↔atom / phrase↔molecule cut. Chemistry sources opened do not show Zipf over molecules. Nearby power laws only: Benz et al. substructures and cluster sizes (exponents about 1.5–3); Grzybowski et al. 2005 reaction-network degrees near 2.7 in and 2.1 out; Liu et al. 2026 Zipf-like BPE tokens of SMILES/SELFIES, slopes only described as moderately steeper than −1, no numeric slope in the abstract. Cadeddu et al. match text-fragment frequencies to molecular-fragment frequencies and do not name Zipf. Grant’s phrase↔molecule line stays SPECULATIVE and his. His only Grok-app message on 10 Jun was the Williams URL plus “please help me understand this.” The “meaningful molecules” line was the app’s gloss. His phrasal-verb question is open: he does not know if it is true. His quark / anyon analogy is his, SPECULATIVE, and not a particle-physics result.

Zipf at the frame takes up his phrasal-verb question. In English, the verbs that fill one verb-argument frame (“she ___ across the field”) follow a Zipf curve that is specific to that frame, headed by a generic verb (COME), and speakers carry over the order of semantic roles rather than the phrase structure (Römer, O’Donnell & Ellis 2015; Ellis et al. 2014). In chemistry, the ring-and-linker skeletons (“frameworks”) of 24 million catalogued organic compounds follow a power law, and 143 skeleton shapes cover half of them (Lipkus et al. 2008). The two cumulative slopes are close (−1 for V across n, −1.07 for frameworks), but the objects are opposite: language ranks the fillers of a fixed, role-ordered frame; chemistry ranks the frames after deleting the fillers and the attachment points. Neither side has measured the other’s half yet. No Navajo counts. SPECULATIVE comparison only.

Zipf fillers in chemistry looks for the missing chemistry half. Chemists do count the small groups that hang off a molecule (“substituents”), and the counts are long-tailed: 849,574 different substituents in 3 million commercial molecules, about 64% seen only once (Ertl 2003), and “a typical power law (or ‘long tail’)” in natural products (Ertl 2021 preprint). But no opened source prints a slope, and none counts the fillers position by position on one skeleton, so the like-for-like test is still undone. The nearest object with slots is the benzene ring. There, which positions get filled, ignoring what fills them, is top-heavy: four of twelve patterns make up 76% of benzene rings in approved drugs (Nilova & Stuart). The authors tie this to a textbook rule in which the kind of group already on the ring, not its atom identity, picks the next slot. The gap is narrowed, not closed. No Navajo counts. SPECULATIVE comparison only.

Host gallery

Sixteen REAL pegs. The last column is the SPECULATIVE Bridge A mnemonic, copied from the conclusions addendum. Read a checklist before treating that mnemonic as a mechanism.

HostDiagram / latticeFirst “over” (typical)Bridge A peg (speculative)
Al–Mg AA5083Limited solvus; FCC Al + subst. MgGB β / sensitization (IGC)Solvus overrun → unwanted second phase
Cu–NiIsomorphous complete SSSConductivity / TCR property curvesProperty-curve overrun (no Ni second phase)
Fe–CEutectoid Fe–Fe₃C; interstitial CCementite networks / brittle as-quenched M / temper–RATransformation / morphology overrun
Ni–γ′γ/γ′ solvus; wanted precipitateOveraging / rafting gone wrong / TCPWanted γ′ vs wrong second phase
Mg (host)HCP; AZ/AM/ZK (+ RE)Brittle IMCs; galvanic β; creep; ignition; anisotropy missβ ambivalent (help or hurt); slip-system / safety “over”
Ti (host)HCP α / BCC β; β-transus; Ti-6Al-4Vα-case; O/H embrittlement; ω; β flecks; overagingWanted α morphology vs unwanted ω / α-case / hydride
Oxide electronic (TiO₂ / ZnO)Wide-gap oxide + subst. metal + V_ORecombination centers; μ collapse; secondary oxides; vacancy compensationCarrier / mid-gap window overrun — true dopant language (cleanest electronic-doping Bridge A case)
CNT/graphene intercalationsp² framework + guest in channels/gapsMoisture dedoping; diffusion blur; junction R; no band gap for switchingIntercalant loading / stability window — true intercalant language
GaN:Mg (III–V)Wurtzite GaN + Mg_Ga acceptor (+ H)Self-compensation; PIDs; μ/optical loss; failed activationCarrier window on industrial electronic doping — true dopant
Si:P/As/B (classical)Diamond Si + shallow subst. donors (P, As) / acceptors (B)Impurity band / Mott; clustering; μ collapse; compensationCarrier window on textbook classical electronic doping — true dopant
Ni-rich NMC:Mg/Al/Nb (battery cathode)Layered oxide + dilute Mg/Al/Nb lattice stabilizersInactive capacity; Li-path blocking; secondary phases; thermal still fails if cutoff insaneStructural-stability window — papers say doping; mechanism ≠ Si:P / GaN:Mg
Cantor CoCrFeMnNi / HEAEquiatomic FCC MPEA; five co-principal alloying elementsCryogenic nanotwinning / TWIP; composition or long-anneal phase driftSPECULATIVE, harder peg: multi-component continuous space; alloying ≠ dilute doping
TWIP Fe–Mn (austenitic)FCC γ Fe–Mn–C(–Al–Si); SFE bands (~15–30 wt% Mn)Too-low SFE → TRIP/ε; too-high → no twin; H delayed fractureSPECULATIVE: SFE window as continuous saturation analog — alloying, not dilute doping
Al–Li (aerospace)FCC Al + Li (+Cu…); δ′ Al₃Li / T₁ Al₂CuLiAnisotropy / toughness (planar slip) / SCC / overaging / melt Li lossWanted δ′/T₁ + density/E vs wrong GB/coarse phases — alloying / precipitation, not dilute doping
CoCrMo (biomedical)Co-majority FCC γ (+ ε); ~28Cr–6Mo; F75/F1537; SIMT + M₂₃C₆Brittle carbide networks; Cr depletion; uncontrolled ε; fretting/MACC ionsWear/passivity/fatigue + SFE/SIMT + carbide morphology — alloying, not dilute doping
Zr nuclear (cladding)HCP α service / BCC β process; Zircaloy / Zr–Nb; SPPsNodular/uniform corrosion; hydriding/DHC; LOCA breakaway; SPP fluence evolution; growth/bowNeutron economy + waterside + irradiation anisotropy — alloying / SPP, not dilute doping

Same element, different job: Mg, H, Al, Ni, Mn, and Co. Shared lattice, different duty: HCP hosts.

Framed from the site content map and the conclusions addendum host table. No joint field.