Carbon as a micro-addition that takes oxygen off the grain boundaries: a 0–1 at.% C ladder in arc-melted Nb44Ta44Hf10W2 with a tensile threshold between 0.4 and 0.6 at.% C, and grain-boundary HfO₂ at only 0.09 at.% overall oxygen (Shen et al. 2025 preprint) — materials scout,
What’s claimed and how well supported
- Labels on this page
- STRONG ×22 HYPOTHESIS ×7 WEAK ×6 SPECULATIVE ×6 OPEN ×7 NONE ×16
- Strongest label used
- STRONG
- Main caveat (quoted)
- The Bridge A section comes after the materials and is ◆ SPECULATIVE.
Counted automatically from the labels printed on this page. Counts are a reading aid, not a grade of the page.
Terms on this page (7), plain-language definitions
- Host (matrix)
- Majority element/phase whose lattice dominates the material.
- Dopant
- Intentional dilute impurity meant to tune electronic/optical/catalytic behavior without replacing the host as majority phase.
- Solid solution
- Solute atoms dissolved into the host lattice (substitutional or interstitial) without a separate precipitate phase.
- FCC
- Face-centered cubic — many slip systems; Al, Cu, Ni typically ductile.
- Precipitate
- Second-phase particle that forms (often on aging) and blocks dislocations.
- 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.
Date: 2026-11-11 PT (weekday research push, evening)
Status: Materials scout. The metallurgy is same-domain description from the one source opened below (REAL inside physical metallurgy). Grades, arithmetic and cross-source comparisons are my analysis. The Bridge A section comes after the materials and is SPECULATIVE. Not a correlate. Not a joint field.
Audience: Grant (syntax/semantics; less chemistry). End goal: designing new metal alloys. Takes up open question OQ2 ("Which way does Hf push the ceiling?") of Oxygen in a Hf-bearing refractory host: ~0.3 at.% as melted, HfO₂ above ~0.5 at.% (Nb45Ta25Ti15Hf15) §5, in part and sideways: it does not say which way Hf pushes the solubility ceiling, but it shows, in an open source with measured oxygen, that a Hf-bearing refractory hostMajority element/phase whose lattice dominates the material. can put HfO₂ on its grain boundaries at an overall oxygen far below last night's ~0.5–1 at.% band, because the oxide forms from the last liquid. Side-light on OQ1 of Ti–Zr–Hf–Nb–Ta with Fe micro-alloying: a ductility window set by nano Hf₂Fe at grain boundaries ("Fe (or other micro-addition) ladder"): a second sourced micro-addition tension window, in a different host; equiatomic HfNbTaTiZr transfer stays NONE.
Source, stated plainly: Shen, Wang, Wang, He, Yang, Shu, Xiao & Sun (author order of the manuscript; the Research Square record lists Da Shu first, and the supplement omits Xiao), "Transformation-aggravated oxide embrittlement mechanism and carbon microalloying ductilisation strategy in refractory complex concentrated alloys", Research Square preprint, posted 12 Dec 2025, doi:10.21203/rs.3.rs-7842583/v1, CC BY 4.0, not peer-reviewed as posted; no journal version of record listed on the preprint page tonight. Shanghai Jiao Tong University with one AECC Beijing Institute of Aeronautical Materials author: independent of the BIT group (L4/L5/Y5) and of the Texas A&M / UC Irvine circle (H/B).
Vocabulary lock kept: Nb, Ta, Hf, W are principal elements (W at 2 at.% is a minor principal element by the authors' design, not a dopantIntentional dilute impurity meant to tune electronic/optical/catalytic behavior without replacing the host as majority phase.). Carbon (0.2–1.0 at.%) is the micro-addition — the authors' word is "microalloying" — added as TaC. Oxygen is an interstitial impurity here, not added on purpose. Fe is not in the source.
Does not: claim that an atom, an impurity, a carbide, an oxide or a phase is an argument, operator, slot, feature bundle or theta role; supply a chemical stand-in for a theta role; invent numbers or quotes; read numbers off figures (Figs. 1–5, Extended Data Figs. 1–8 and Supplementary Figs. are not read; measured C and the O of the 0.2C/0.4C/0.6C/1C alloys are figure-only and stay NONE); transfer this host's numbers to the BIT host or to equiatomic HfNbTaTiZr as facts; revise Bridge A.
Why this note. Last night's note bounded oxygen in a Hf-bearing host (Nb45Ta25Ti15Hf15: ~0.3 at.% as-melted; HfO₂ above ~0.5 at.%) and left open which way more Hf moves the oxide edge (OQ2). OpenAlex is_oa searches tonight again found no equiatomic HfNbTaTiZr micro-addition ladder with open full text (§1.2). The one open paper that pairs (i) a measured overall oxygen, (ii) a Hf-bearing refractory bcc host with grain-boundary HfO₂, and (iii) a true micro-addition ladder with tensile numbers in text form is Shen et al. 2025. It does not answer OQ2's direction. It does add a variable last night's note did not have: where the oxygen goes during solidification.
The case, in plain words. A Shanghai team arc-melts Nb44Ta44Hf10W2 (no Ti, no Zr) and five carbon steps up to 1 at.%. Without carbon the alloy holds only 99 mass ppm oxygen (0.09 at.%) — about a third of the corpus's usual as-melted scale — yet it breaks in tension along its grain boundaries before it yields, and those boundaries are coated with a dense layer of hafnium oxide that grew like frost from the last bit of liquid. With 0.2 or 0.4 at.% carbon nothing improves (and the oxygen even goes up, because the TaC pellets carry oxygen). From 0.6 at.% carbon up, the alloys stretch 11–13% in tension: the oxygen drops (to 59 ppm at 0.8 at.% C), the grain-boundary oxide is gone, and fine carbides that can hold some oxygen sit on the boundaries instead.
What is new (one sentence). In arc-melted Nb44−xTa44−xHf10W2C2x (C = 0, 0.2, 0.4, 0.6, 0.8, 1.0 at.%; Shen et al. 2025, CC BY preprint), the carbon-free alloy holds 99 wppm ≈ 0.09 at.% O ( STRONG as measured; my 0.088) yet carries a dense grain-boundary m-HfO₂ layer and fails intergranularly at 0.33% total elongation ( STRONG as observed / tabulated), which the authors trace to HfO₂ growing from the last liquid (~80% of O left in the residual liquid by their equilibrium calculation; STRONG as computed, WEAK as the actual path) — so in this host the as-cast grain-boundary oxide appears at about a fifth of last night's ~0.5 at.% XRD edge (arithmetic), and an XRD-based oxide onset is at most an upper bound on the first boundary oxide ( HYPOTHESIS, two hosts, two groups); carbon gives a sharp tensile threshold between 0.4 at.% (0.41%) and 0.6 at.% (12.7%) with a peak at 0.8 at.% (YS 682.2 MPa, UTS 807.0 MPa, 13.3%; STRONG as tabulated, no scatter printed), but carbon dose and overall oxygen move together (O higher than 0C at ≤ 0.4 at.% C, "drastically" lower at ≥ 0.6; STRONG as stated, values NONE in text), so carbon-alone vs oxygen-drop is not separated ( OPEN); which way Hf pushes the solubility ceiling stays OPEN, BIT-host oxygen stays NONE, and equiatomic transfer stays NONE.
1. Sources
1.1 Opened this run
| Key | Source | How read | Used for |
|---|---|---|---|
| S | Shen, Wang, Wang, He, Yang, Shu, Xiao & Sun, Research Square preprint rs-7842583/v1 (posted 2025-12-12), CC BY 4.0 (licence line on the PDF cover page) | Research Square PDF HTTP 200 (3.25 MB, 40 pp.; PDF p. 1 is the Research Square cover; manuscript page n = PDF page n + 1) → lit/_pdfs/rcca-carbon-microalloy-shu2025/shu2025.pdf + shu2025.txt (pdftotext -layout) + shu2025.raw.txt; preprint page HTML rs-page.html; openalex.json (W-id in file; oa_status green, cc-by) |
Composition ladder, charge, melting, test method (PDF pp. 26–27); oxygen numbers (PDF pp. 12–13, 35); fracture and GB HfO₂ (PDF pp. 8–11); carbon ductilisation (PDF pp. 12–13); main-text elongation range (PDF p. 6) |
| S-supp | Supplementary information Supplementaryv5.docx linked from the preprint page (assets-eu.researchsquare.com; HTTP 200) |
docx XML → shu2025-supp.txt; EndNote field codes stripped → shu2025-supp.clean.txt (text only; no figure read) |
Supplementary Table 1 (tension, all six alloys), Supp. Tables 4 and 6 (TEM-EDS of GB HfO₂ and GB carbide), Supp. Notes 2, 3, 4, 6, 7 (ingot zones, intrinsic ductility, HfO₂ growth from the last liquid, deoxidation, scavenging) |
| (cross) | H (Huang et al. 2025, arXiv:2512.08080), B (Belcher et al. 2025), Y5 (Yang 2025), L5 (Li 2025) — opened on earlier nights | Prior notes / captures; not re-fetched | Last night's ~0.3 at.% floor and ~0.5–1 at.% HfO₂ band; BIT host formula |
| (calc) | My arithmetic, standard atomic masses | lit/_pdfs/rcca-carbon-microalloy-shu2025/arithmetic_carbon.py → arithmetic_carbon.out |
Ladder at.% ↔ wppm C; O wppm ↔ at.% check; Table 1 differences/ratios; TEM-EDS sums; VEC check; Hf/O ratios; scale comparison with H |
Weakness of the copy, stated plainly. A preprint (not peer-reviewed as posted). Every figure is an image; the measured carbon of each alloy (Extended Data Fig. 5b) and the measured oxygen of the 0.2C, 0.4C, 0.6C and 1C alloys (Extended Data Fig. 5c) are figure-only and are NONE here. Supplementary Table 1 prints one value per cell with no scatter, although "all the tensile tests were repeated at least 3 times" (PDF p. 27); I take the cells as the authors' representative values, not means with known spread. TEM-EDS compositions are "semi-quantitative" in the authors' words, and they warn of "the exaggerated value of O concentration commonly seen in TEM-EDS" (PDF p. 13).
What the source does not report, stated plainly. No Ti, no Zr, no Fe; no heat treatment after casting (all tension is as-cast, cut from the in-situ-annealed mid-height of the ingot; Supp. Note 2); no oxygen-added series (oxygen is whatever the charge and melter left); no nitrogen value in text; no equiatomic HfNbTaTiZr.
Independence, stated plainly. S cites B (Belcher et al.) in Supp. Note 1 as a source on where oxygen comes from, and cites the Nb45Ta25Ti15Hf15 creep paper of the H circle (main-text ref. 11); it shares no author with H, B, L4, L5 or Y5. Agreement with H/B on feedstock oxygen is therefore a second group, not continuity.
1.2 Not opened / searched without a hit (do not treat as read)
- OpenAlex
is_oasearches tonight (HfNbTaTiZr / TiZrHfNbTa withaddition,doped,doping,microalloying, nitrogen, carbon, boron, silicon, aluminum;TiZrNbHfTa minor): no micro-addition ladder in equiatomic HfNbTaTiZr with open full text. Fe-note OQ1 / Liu-note OQ4 unchanged. - Wang, Tang, Lei et al. 2021, "Achieving high strength and ductility in nitrogen-doped refractory high-entropy alloys", Mater. Des. 213: 110356 (CC BY; (TiZrNbTa)100−xNx, x = 0–0.9): OpenAlex/Unpaywall list only the ScienceDirect PDF; HTTP 403 to curl (browser user-agent), the fetch tool timed out, a headless-browser load returned the ScienceDirect challenge page; Europe PMC 0 hits; CORE has the record without full text. Not opened; not routed around. Abstract-level lead only (an equiatomic quaternary sibling with a nitrogen ladder); nothing from it is used.
- Yasuda et al. 2021 body (Mater. Sci. Eng. A 809: 140983): ScienceDirect HTTP 403 again; fetch tool timed out. Not routed around (unstable-band OQ3/OQ4 unchanged).
- Anber et al. 2023 (SSRN 4418820; Al additions to HfNbTaTiZr, oxidation): SSRN HTTP 403; not opened.
- S's own references (Wang et al. 2022 Mater. Today 54: 83 on GB engineering against oxygen embrittlement; the Ta-W-Hf literature behind the Hf+W attribution, Supp. ref. 125): not opened; used only as S restates them.
2. Sourced materials
2.1 Alloys, charge, route, tests (S Methods)
- Ladder: "RCCA ingots with nominal compositions of Nb44-xTa44-xHf10W2C2x (x = 0, 0.1, 0.2, 0.3, 0.4, 0.5, in at.%) were prepared by non-consumable arc melting in water-cooled copper crucibles, referred to as 0C, 0.2C, 0.4C, 0.6C, 0.8C and 1C alloys." (PDF p. 26)
- Charge: "Feedstocks are Nb (99.9 wt.%), Ta (99.9 wt.%), Hf (99.9 wt.%), W (99.9 wt.%) and TaC (99.9 wt.%)." Ingots "flipped and remelted at least 10 times". (PDF p. 26)
- Why Hf: "A relatively high Hf addition is to largely enhance solid solutionSolute atoms dissolved into the host lattice (substitutional or interstitial) without a separate precipitate phase. strengthening" and "to ensure the formation of favorable oxygen-soluble FCCFace-centered cubic — many slip systems; Al, Cu, Ni typically ductile.-MC type carbides" (PDF p. 26).
- Chamber (Supp. Note 6): the arc melter "has lower levels of vacuum (2~3×10-3 Pa) and is backfilled with Ar".
- Tension: dog-bone, gauge 2 × 1 × 8 mm, extensometer, 1×10⁻³ s⁻¹, "repeated at least 3 times" (PDF p. 27).
- Oxygen and carbon: "Quantitative measurements of the overall oxygen and carbon concentrations were conducted using an O/N/H analyser (NCS, ONH3500) and a carbon/sulphur analyser", "based on the average of 3 samples taken from different ingot positions" (PDF p. 27).
- Specimen location (Supp. Note 4): "the tensile samples were sectioned from the mid-height of the ingot where grains had been fully in-situ homogenized". The ingot is "highly heterogeneous" from bottom to top (Supp. Note 2).
2.2 Oxygen, as printed
- Extended Data Fig. 4 caption: "Oxygen content followed the experimentally measured 99 wppm (0.09 at.%) for 0C alloy and 59 wppm (0.054 at.%) for 0.8C alloy." (PDF p. 35)
- Main text: "After 0.8 at.% C microalloying, the overall O content in the RCCA ingot drops from 0.09 at.% to 0.054 at.%" (PDF p. 12).
- Supp. Note 6, the intermediate steps in words only: "For C addition ≤ 0.4 at.%, the O content is even higher than 0C alloy, which can be explained by the high oxygen content inherently present in the commercial TaC pellets"; "For C addition ≥ 0.6 at.%, the O content drastically decreases, which corresponds well with the gained tensile ductility at RT".
- Supp. Note 4 on purity: "the total oxygen content is only 99 wppm (around the critical value between commercial purity and high purity)".
- Supp. Note 6, mechanism: the carbon–oxygen reaction to CO "is effective only when [C]inmelt is abundant", and "the actual C content for the alloys in the range of 0.4C~0.6C increased more slowly than nominal" (figure only).
2.3 Tension (S-supp Table 1; main text)
| Alloy | YS (MPa) | UTS (MPa) | Total elong. (%) | Uniform elong. (%) |
|---|---|---|---|---|
| 0C | / | 487.9 | 0.33 | / |
| 0.2C | / | 308.3 | 0.11 | / |
| 0.4C | / | 674.7 | 0.41 | / |
| 0.6C | 653.3 | 743.1 | 12.7 | 9.9 |
| 0.8C | 682.2 | 807.0 | 13.3 | 10.1 |
| 1C | 602.7 | 718.9 | 11.4 | 8.1 |
(slash = as printed; the main text says 0C, 0.2C and 0.4C "fracture before (or at most, upon) macroscopic" yielding, PDF pp. 5–6, and the ductile set has "11.4% ~ 13.3% overall elongation", PDF p. 6.)
2.4 Why the carbon-free alloy breaks (S)
- Mode: "0C alloy fractures in a typical intergranular mode in RT tension" (PDF p. 8); the grains themselves are ductile: "0C alloy is experimentally determined to be intrinsically ductile within grains" (PDF p. 8).
- What is on the boundaries: "A dense layer of monoclinic m-HfO2 oxides with geometrically-regular dendritic morphology is found on the fractured GB surfaces" (PDF p. 8). Size "50 nm ~ 1 μm" (PDF p. 11). TEM-EDS inside the oxide (Supp. Table 4): Nb 4.74, Ta 2.3, Hf 32.94, W 0.49, O 59.53 at.% (semi-quantitative).
- Where it comes from: "HfO2 was most likely to precipitateSecond-phase particle that forms (often on aging) and blocks dislocations. from the remaining liquids at the final stage of solidification" (Supp. Note 4); "The oxygen in the completely solidified BCCBody-centered cubic — Fe/steel common; plasticity more temperature-sensitive. phase only accounted for ~20% of total oxygen content (Extended Data Fig. 4c), implying ~80% O in the retained liquids at the final stage of solidification" (Supp. Note 4; equilibrium CALPHAD).
- Why so much oxide at so little oxygen, in the authors' attribution: "In literature, this is attributed to the coexistence of Hf and W." (Supp. Note 4; the Ta-W-Hf sources are not opened.)
- Why it is brittle: the oxide forms tetragonal and transforms to monoclinic on cooling ("During cooling around 1900℃, t-HfO2 transforms completely into m-HfO2", PDF p. 8), which, the authors argue, spoils the oxide–metal interface; plus cracking inside the oxide and strain localisation in thin bcc channels between oxide arms (PDF pp. 9–11).
- XRD: the Extended Data Fig. 1 caption reports "the major phase as single BCC, as consistent with the XRD result in Supplementary Fig. 1" (PDF p. 32); Supp. Note 4 calls the GB oxide "HfO2 dendrites in themselves, exceeding the detection limit of XRD" (wording ambiguous; I read it as beyond what XRD detects, a WEAK reading).
2.5 What carbon changes (S)
- "Oxygen segregates at the GBs in 0C alloy, whilst in 0.8C alloy, carbon segregates at the GBs and subgrain boundaries (SGBs) without obvious oxygen segregation." (PDF p. 5)
- GB oxygen by ToF-SIMS/AES: "a sharp enrichment (266% ~ 370%) of O at high and low angle GBs in 0C alloy"; in 0.8C "a slight (111% ~ 113%) O enrichment at GBs and SGBs" (PDF p. 12).
- Carbides: FCC-MC at GBs, "discontinuous", "finer (20 nm ~ 100 nm) than m-HfO2 (50 nm ~ 1 μm)" (PDF p. 13); TEM-EDS inside a GB carbide (Supp. Table 6): Nb 9.34, Ta 22.30, Hf 32.37, W 1.58, O 10.60, C 23.81 at.% (semi-quantitative). The authors cite "~4.4 at.% for HfC" as an oxygen capacity of such carbides (PDF p. 13; their ref. 37, not opened).
- Matrix oxygen after carbon: "This O concentration in the BCC matrix (< 0.054 at%) scarcely leads to atomic-form-oxygen embrittlement" (PDF p. 13).
- Order of events (Supp. Note 7): oxygen scavenging by FCC-MC "is only possible if FCC-MC precipitated in the liquids ahead of potential oxides"; once HfO₂ has formed, "it was difficult to “re-scavenge” the oxygen in solid state".
- Summary sentence: "the ductilisation after C microalloying is attributed to the full elimination of GB HfO2 and the avoidance of new potential embrittling factors." (PDF p. 13)
3. Grades and analysis
| Claim | Grade | Basis |
|---|---|---|
| 0C holds 99 wppm ≈ 0.09 at.% O; 0.8C holds 59 wppm ≈ 0.054 at.% | STRONG as measured (ONH analyser, 3 positions); my conversions 0.088 / 0.052 at.% agree to the printed rounding | S PDF pp. 12, 35; arithmetic_carbon.out [2] |
| O at 0.2C/0.4C higher than 0C; "drastically" lower at ≥ 0.6C | STRONG as stated (direction); values NONE (figure only) | S-supp Note 6 |
| Tensile table: brittle at C ≤ 0.4 at.% (0.11–0.41% total), ductile at C ≥ 0.6 (11.4–13.3%), peak at 0.8C (YS 682.2, UTS 807.0, 13.3%) | STRONG as tabulated (preprint; no scatter printed; as-cast; mid-height specimens) | S-supp Table 1; arithmetic_carbon.out [3] |
| Lower edge of the carbon window lies between 0.4 and 0.6 at.% nominal C | STRONG as a bracket (×31 jump in total elongation; arithmetic); location inside the bracket NONE | same |
| Upper edge of the window | NONE: 1C (1.0 at.%) still 11.4%, with lower YS than 0.8C; no step above 1 at.% | same |
| UTS non-monotonic in the brittle set (487.9 → 308.3 → 674.7 MPa) | STRONG as tabulated; meaning NONE (brittle breaks; scatter unknown) | same |
| 0C fails along GBs coated with dendritic m-HfO₂; grains intrinsically ductile | STRONG as observed (fractography, EBSD, TEM; figures not read, prose relied on) | S PDF p. 8; Supp. Note 3 |
| GB HfO₂ grows from the last liquid; ~80% of O in residual liquid | STRONG as the authors' equilibrium CALPHAD output; WEAK as the real non-equilibrium path (the authors say their Scheil run gave "erroneous results", Supp. Note 7) | Supp. Notes 4, 7 |
| Over-proliferation of HfO₂ at 99 wppm due to Hf + W together | STRONG as the authors' attribution; WEAK as mechanism (literature analogy; not tested in S; sources not opened) | Supp. Note 4 |
| t→m transformation of HfO₂ as the aggravating step | STRONG as the authors' interpretation of their EBSD/TEM; HYPOTHESIS as the controlling cause (not isolated by experiment) | S PDF pp. 8–11 |
| Carbon removes GB HfO₂ and puts discontinuous FCC-MC on GBs (0.8C) | STRONG as observed for 0.8C only; GB phases of 0.2C/0.4C/0.6C/1C NONE | S PDF pp. 12–13 |
| Carbon's effect = carbon alone (vs simply less oxygen) | OPEN: in this ladder carbon dose and total O co-vary (Supp. Note 6); no fixed-O carbon series and no fixed-C oxygen series | analysis |
| TEM-EDS oxide O/Hf = 1.81 (HfO₂ = 2); carbide (C+O)/metal = 0.52 (MC = 1) | arithmetic on semi-quantitative cells; not used as stoichiometry | arithmetic_carbon.out [4] |
| VEC of 0C = 4.92 | arithmetic agrees with S's print | arithmetic_carbon.out [5] |
Bearing on OQ2 of the oxygen note — "Which way does Hf push the ceiling?" (analysis). Not answered; reframed with a sourced second variable. Last night's ceiling (H) is an equilibrium-style quantity: how much oxygen a solid bcc host holds before HfO₂ is favourable, detected by XRD (above ~0.5 at.%) and by STEM at 1 at.%. S shows a different route to the same oxide: in an as-cast Hf-bearing host, oxygen that the first-freezing bcc rejects is pushed into the last liquid together with Hf (Hf segregates interdendritically; Supp. Note 2), and HfO₂ forms there — at an overall oxygen of 0.09 at.%, about 0.18 of H's ~0.5 at.% XRD edge and 0.30 of H's own no-O-added baseline (arithmetic [6]). The Hf/O atom ratio in S's oxide-bearing 0C (~111) is higher than in H's oxide-free 0.3 at.% button (~50) (arithmetic [6]), so a simple Hf-to-oxygen ratio does not predict where GB HfO₂ appears ( STRONG as arithmetic on two sourced states; the reason — segregation path, W, melter, cooling — OPEN). What this changes for OQ2: the question splits in two — (a) the solid-solubility ceiling (H's quantity; Hf direction still OPEN), and (b) the solidification edge, where segregation can put HfO₂ on boundaries far below (a) ( HYPOTHESIS as a general rule; STRONG only for S's host and route). XRD-based onsets like H's are then upper bounds on the first boundary oxide ( HYPOTHESIS; S's GB oxide is reported with a single-bcc XRD/phase map, WEAK reading of "exceeding the detection limit of XRD"). H's 0.5 at.% button showed no GB Hf/O enrichment by STEM, so in H's host the solidification edge did not bite at 0.5 at.% ( STRONG as stated in H) — the two hosts differ (Ti 15 vs 0; W 0 vs 2; Nb/Ta ratio; melter; remelts 5 vs ≥ 10), so this is a contrast, not a contradiction.
Bearing on the BIT host (analysis). BIT's Ti15Zr10Hf40Nb10Ta25 has no W (S's attributed amplifier), much more Hf (40 vs 10 at.%), and Ti + Zr. Whether its last liquid is Hf- and O-rich like S's is NONE (no segregation data for the BIT host in the corpus). The design reading is only this ( HYPOTHESIS): an as-cast or homogenized BIT specimen can be checked for GB oxide directly (fractography/TEM), not inferred from a bulk oxygen number, because S shows the bulk number alone (0.09 at.%) does not exclude a GB oxide layer. Measured oxygen in the BIT host remains NONE.
Bearing on the micro-addition line (Fe-note OQ1, Liu-note OQ4) (analysis). S is the corpus's second sourced micro-addition ladder with tension in a Hf-bearing refractory host, after Li 2025's Fe in the BIT host. Shapes in common, stated as materials: a minor addition (Fe ~3.5 at.%; C ≤ 1 at.%) whose effect runs through grain boundaries (Hf₂Fe on GBs in L4/L5; FCC-MC replacing HfO₂ on GBs in S), and a ductility that switches over a narrow dose range (Fe: ageing clock in one host; C: between 0.4 and 0.6 at.% as-cast). Differences: S has no ageing; L5 has no oxygen; S's carbon changes the oxygen too. No transfer to equiatomic HfNbTaTiZr ( NONE); whether carbon at ~0.6–1 at.% would do in a Ti/Zr-rich 40 at.% Hf host what it does here is HYPOTHESIS at best (S's authors chose Hf partly to make FCC-MC, PDF p. 26).
Counting rule. Only contradicting data count. S is one group, one preprint; its supplement and main text agree (main-text "11.4% ~ 13.3%" = min/max of Supp. Table 1's ductile set; arithmetic [3]). No figure digitized; the 0.2C/0.4C/0.6C/1C oxygen and every measured carbon stay NONE. S vs H is a contrast between two independent groups and two hosts, not a correction of either.
SPECULATIVE Cross-domain section. Open after the materials.
4. Bridge A — SPECULATIVE (after the materials)
Bridge A (locked): doping concentration windows and over-doping ↔ valencyArity: how many actants a predicate takes; formal vs semantic valency can diverge., argument saturation and over-arity. Shapes only:
- A window with a step edge. Below ~0.5 at.% C nothing changes; from 0.6 at.% the host stretches. Shape: an addition can act as a threshold, not a slope. Shape only.
- The second addition rehouses the first impurity. Carbon does not make the metal hold more oxygen in the matrix; it changes which structure the oxygen ends up in at the boundaries (carbide instead of oxide), and when that happens (before the oxide, in the liquid). Shape: what an extra element does can be to redirect where an unwanted guest lands, with timing mattering. Shape only; no carbide or oxide is an argument or a slot.
- The zero setting is not zero, again. The 0C alloy is not oxygen-free and its oxide is not where the bulk number suggests. Shape: the empty setting of a window can already be occupied, locally. Shape only.
- Still continuous. at.%, wppm and MPa are continuous amounts; the lock "continuous wt% ≠ discrete arity" holds; nothing here counts slots.
Bridge A pressure: none that forces revision. Bridge A unrevised. No chemical stand-in for a theta role, slot or feature bundle; not a correlate; not a joint field. Merge/Agree not involved. No Navajo remark tonight.
5. Open questions
- Carbon vs oxygen, separated. A fixed-oxygen carbon ladder (or a fixed-carbon oxygen ladder) in S's host: does carbon at 0.6–1 at.% help when overall O is held at 99 wppm, or is the window mostly the oxygen drop? Recurs, not answered, in Boron inside the grains vs on the boundaries: a dilute Re + B ladder in Nb45Ta27Ti13Hf15, beside a blocked equiatomic boron lead (2026-11-12 evening): An et al. 2026's +102 MPa step from 0.3 at.% B in Nb45Ta27Ti13Hf15 Re1.5 comes with no measured oxygen in any alloy, so B-alone vs B-plus-impurity is OPEN there too.
- Where in 0.4–0.6 at.% C is the edge, and is there an upper edge? Steps at 0.5 and above 1.0 at.%; tensile scatter for each cell.
- The intermediate numbers. Extended Data Fig. 5b/c (measured C and O for all six alloys) in a text form or source data ("Source data are provided with this paper" — not in the preprint package captured tonight).
- Solidification edge in a Ti/Zr-bearing host. Does an as-cast Hf-bearing host with Ti and Zr and without W (H's or BIT's) put HfO₂ on boundaries below its XRD edge? Fractography/TEM at ~0.3 at.% O would test the §3 HYPOTHESIS.
- Hf + W attribution. Is the over-proliferation really from W pushing Hf to the last liquid? A W-free control of S's host would test it.
- Carried: OQ1, OQ2 (direction stays OPEN), OQ3–OQ5 of Oxygen in a Hf-bearing refractory host: ~0.3 at.% as melted, HfO₂ above ~0.5 at.% (Nb45Ta25Ti15Hf15); Nov 9 OQ1–OQ5 (Ti–Zr–Hf–Nb–Ta base printed (Yang 2025, EBM): nominal Ti15Zr10Hf40Nb10Ta25, the Fe-free end of the Fe family); Liu-note OQ1–OQ5; Fe-note OQ1 (no open equiatomic micro-addition ladder found again tonight; Wang et al. 2021 N-doped TiZrNbTa blocked), OQ2, OQ3, OQ5; OQ1–OQ4, OQ6 of HfNbTaTiZr between 550 and 1000 °C: where the unstable band ends (two edges, not one) (Yasuda 2021 body blocked again); prior carried stack listed there.
6. Separation
SOURCED (inside metallurgy)
S: Nb44−xTa44−xHf10W2C2x, x = 0–0.5 (0C…1C); 99.9 wt.% Nb, Ta, Hf, W and TaC; non-consumable arc melting, ≥ 10 remelts, 2–3 × 10⁻³ Pa then Ar; in-situ annealing; tension 1×10⁻³ s⁻¹, ≥ 3 repeats, mid-height specimens; ONH and C/S analysers, 3 positions; 0C 99 wppm (0.09 at.%) O, 0.8C 59 wppm (0.054 at.%); O at ≤ 0.4C above 0C, ≥ 0.6C drastically lower (words); Supp. Table 1 tension; 0C intergranular, grains intrinsically ductile; dense dendritic GB m-HfO₂ 50 nm–1 μm; HfO₂ from last liquid, ~80% O in residual liquid (equilibrium CALPHAD); Hf + W attribution; t→m at ~1900 °C; GB O enrichment 266–370% (0C) vs 111–113% (0.8C); discontinuous GB FCC-MC 20–100 nm; Supp. Tables 4 and 6 TEM-EDS; matrix O < 0.054 at.%; carbide-before-oxide order; authors' summary of ductilisation.
ANALYSIS (mine, beside the sources)
The grades; wppm ↔ at.% checks; Table 1 differences and the ×31 jump; TEM-EDS ratio sums; VEC check; the Hf/O comparison with H; the split of OQ2 into a solubility ceiling and a solidification edge (HYPOTHESIS as a general rule); XRD onset as an upper bound (HYPOTHESIS); the carbon/oxygen confound (OPEN); BIT design reading (HYPOTHESIS); counting.
SPECULATIVE
§4 (Bridge A shapes).
Not claimed
No chemical correlate. No chemical stand-in for a theta role, a feature bundle or a slot. Not a joint field. Bridge A unrevised. No number read off a figure. No oxygen value for the BIT host. No transfer of S's window to equiatomic HfNbTaTiZr or to the BIT host as a fact. Nothing from the not-opened list. The dropped roles-not-words theory is not treated as open.
7. Verdict
Does more Hf raise or lower the oxide edge? Still open — but the edge has two parts. In an independent, CC BY preprint (Shen et al. 2025), an arc-melted Nb44Ta44Hf10W2 with only 0.09 at.% overall oxygen already carries a dense grain-boundary HfO₂ layer grown from the last liquid and breaks at 0.33% elongation ( STRONG as measured/observed; growth path STRONG as computed, WEAK as real), about a fifth of last night's ~0.5 at.% XRD edge in a Ti-bearing host — so a bulk solubility ceiling and a solidification edge are different numbers ( HYPOTHESIS as a rule). A carbon micro-addition gives a sharp tensile window (brittle ≤ 0.4 at.% C; 11.4–13.3% at 0.6–1.0; peak 0.8 at.% C at 682/807 MPa; STRONG as tabulated), with GB carbide replacing GB oxide at 0.8 at.% C ( STRONG as observed) — but carbon and oxygen move together, so carbon-alone is OPEN. BIT-host oxygen NONE; equiatomic micro-addition ladder still no open source. Bridge A unrevised; shapes only, after the materials.
v0 — 2026-11-11 PT, evening. Pointers added: corpus-reading-order row; CONCLUSIONS-v1-addendum paragraph; site-content-map (inventory, slug, near-ready, Path E page row); back-pointers at nbtatihf-oxygen-solubility-pickup §5 OQ2 and tizrhfnbta-fe-ductility-window §5 OQ1 (+ footers). Quotes machine-checked against the captured text (lit/_pdfs/rcca-carbon-microalloy-shu2025/check_quotes_carbon.py); arithmetic in lit/_pdfs/rcca-carbon-microalloy-shu2025/arithmetic_carbon.py. Speculation flagged. Bridge A lock unrevised. No joint field. Back-pointer 2026-11-12: §5 OQ1 confound recurs in Boron inside the grains vs on the boundaries: a dilute Re + B ladder in Nb45Ta27Ti13Hf15, beside a blocked equiatomic boron lead.
Corpus note: nbtahfw-carbon-oxide-window-v0.md. The metallurgy is same-domain description from the one source the note lists as opened: Shen, Wang, Wang, He, Yang, Shu, Xiao & Sun 2025 (Research Square preprint rs-7842583/v1, CC BY 4.0; not peer-reviewed as posted), the PDF read through its text layer, with the supplementary Word file read as text for its tables. Shanghai Jiao Tong University; independent of the Beijing Institute of Technology group and of the Texas A&M / UC Irvine circle. Figures were not digitized, so the measured carbon of every alloy and the oxygen of the 0.2, 0.4, 0.6 and 1.0 at.% C alloys are not given. Quotes are verbatim and machine-checked; conversions (mass ppm to at.%) and table checks are the note's own arithmetic in a script beside the capture. The host is Nb44Ta44Hf10W2 with carbon added (no Ti, no Zr, no Fe); its window is not transferred to the Beijing Institute of Technology host or to equiatomic HfNbTaTiZr as a fact, and no oxygen level is given for the Fe-micro-alloyed host. In this ladder carbon dose and overall oxygen move together, so carbon acting alone is open. The grades (STRONG / WEAK / NONE / OPEN / HYPOTHESIS) and the comparisons between sources are analysis, not a declared correlate. The Bridge A section comes after the materials and is speculative. No chemical stand-in for a theta role or a slot. Bridge A unrevised. Not a joint field.