{"id":"03308f24-2f09-449b-bebc-846b8041d2b0","arxiv_id":"2412.02600","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"Layered metastable polymorphs of MgZrN2 and MgHfN2 were synthesized by topochemical ion exchange from Li2MN2 precursors, with optical absorption onset near 2.0 eV.","lead":"Scientists used a lithium-containing nitride crystal as a template and swapped in magnesium ions to make a never-before-made layered form of MgZrN2. The same method worked for MgHfN2, suggesting a general route to new nitride semiconductors.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Cation ordering in the 47 wt% R3m product is assumed rather than refined; the unindexed 64° peak and possible off-stoichiometry leave the structural and optical assignment underdetermined.","rationale":"The reader's weakest assumption—that diffraction uniquely identifies the product as cation-ordered R3m MgZrN2—is exactly the load-bearing point. The reported Rietveld analysis fixes occupancies and never challenges the ordering model; adding Fm3m improves the fit but does not validate the R3m cation arrangement. The unindexed 64° peak and the authors' own caveat about Mg loss or oxide incorporation mean the phase composition is also uncertain. Because Mg and Zr differ strongly in scattering power, a simple occupancy refinement is a decisive and inexpensive test. If the cation ordering is wrong, the claim of a new layered polymorph collapses and the optical comparison to R3m calculations is moot. The reader already conditioned the verdict on addressing this, so my read does not move the verdict; it sharpens the specific check needed. I do not see an independent, more severe internal inconsistency: the in situ diffraction does show the precursor disappearing and a new phase growing, and the DFT thermodynamics are separate from the structural assignment. Therefore UNCHANGED is appropriate.","tokens_in":20294,"tokens_out":8843,"duration_ms":93357,"concrete_test":"Re-refine the washed 10-min MgZrN2 PXRD (Figure 1) with the two cation sites in R3m allowed independent Mg/Zr occupancies (and, if warranted, N/O occupancy on the anion site), and also fit an antisite-disorder model where the two sites exchange fractional Mg and Zr. If the refined occupancies deviate by more than ~2σ from the fully ordered Mg1/Zr1 arrangement, or if a partially disordered model achieves the same Rwp with no unindexed peak, then the cation-ordered R3m assignment is not uniquely supported and the 2.0 eV optical onset cannot be cleanly attributed to that phase.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing assumption is that the R3m component in the washed MgZrN2 product is stoichiometric, cation-ordered α-NaFeO2-type MgZrN2. The Rietveld model was built by substituting Mg and Zr into α-NaFeO2 and refining only lattice parameters, Lorentzian size broadening, and Biso (Methods); Mg/Zr occupancies were fixed at the fully ordered 1:1 arrangement. The reported Rwp improvement from 5.749% to 4.336% comes from adding the Fm3m phase, not from testing the ordering model. The fit still excludes an unindexed peak near 64° 2θ, and the authors concede that Mg1−xZr1+xN2−yOy or Mg loss cannot be ruled out from XRD. Since Mg (Z=12) and Zr (Z=40) have strong X-ray scattering contrast, a free refinement of occupancies or an antisite-disorder model could discriminate order from disorder; without it, the 'cation-ordered R3m MgZrN2' identification—and the 2.0 eV onset attributed to it—rests on an assumed model rather than a demonstrated one.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the synthesis of previously unreported layered polymorphs of MgZrN2 and MgHfN2 (α-NaFeO2-type, space group R3m) via topochemical ion exchange from Li2ZrN2 and Li2HfN2 with MgBr2 or MgCl2. In situ synchrotron powder X-ray diffraction shows that the layered ZrN6 framework of the Li2ZrN2 precursor is retained during the exchange, and ex situ Rietveld analysis of the washed product assigns a 47(1) wt% R3m phase alongside a 53(1) wt% Fm3m rocksalt phase. Diffuse-reflectance UV-vis spectroscopy gives an absorption onset near 2.0 eV, consistent with the calculated bandgap of the R3m polymorph. Attempts to extend the reaction to Fe, Cu, and Zn analogs yield decomposition products (A + ZrN + 1/6 N2), which the authors rationalize with temperature-dependent free-energy calculations. The work concludes that Li-M-N ternaries are viable precursors for ion-exchange synthesis of metastable ternary nitrides.","tokens_in":20525,"tokens_out":2972,"duration_ms":32640,"significance":"If the structural assignment is correct, this is a valuable advance in ternary nitride synthesis: it demonstrates a topochemical route to a cation-ordered polymorph that is metastable with respect to the known Fm3m disorder and the calculated I41/amd ground state, and it provides an experimental realization of a computationally predicted semiconductor with a plausible 2.0 eV optical onset. The computational thermodynamics and band-structure results appear independent of the experimental synthesis, with no free parameters tuned to match the measured onset or reaction outcomes. The in situ diffraction study of the reaction pathway is a particular strength. However, the central claim of 'cation-ordered R3m MgZrN2' rests on a Rietveld model in which the Mg/Zr occupancies are fixed at the fully ordered values, and the product is a two-phase mixture with an unindexed impurity; the optical assignment inherits this uncertainty.","major_comments":[{"comment":"The Rietveld model for the R3m phase is built by substituting Mg and Zr into α-NaFeO2 and refining only lattice parameters, Lorentzian size broadening, and Biso, with Mg/Zr occupancies fixed at the fully ordered 1:1 arrangement. Because Mg (Z=12) and Zr (Z=40) have very different X-ray scattering powers, a free refinement of site occupancies or an explicit antisite-disorder model could distinguish cation order from disorder. Without such a test, the 'cation-ordered' identification is assumed rather than demonstrated, and the subsequent optical assignment inherits this assumption. The authors should either refine the occupancies (or report limits on antisite disorder) or temper the ordering claim accordingly.","section":"Results and Discussion, Figure 1; Methods, 'Analysis of diffraction data'"},{"comment":"The product shown in Figure 1 is a 47(1)/53(1) wt% mixture of R3m and Fm3m MgZrN2 with an unindexed impurity peak near 64° 2θ that is excluded from the fit. The UV-vis absorption onset near 2.0 eV in Figure 3 is measured on this mixture, yet the paper attributes the onset to the R3m polymorph. Since the known Fm3m phase absorbs near 1.8 eV and the impurity is uncharacterized, the connection between the measured 2.0 eV onset and the R3m phase needs additional support, for example a spectrum of a sample with a substantially higher R3m fraction or a compositional/structural analysis of the impurity.","section":"Results and Discussion, Figure 1 and Figure 3"},{"comment":"The MgHfN2 assignment in Figure 7 uses the same fixed-occupancy R3m model and reports only 31(1) wt% R3m with 66(1) wt% Fm3m and 2(0) wt% residual Li2HfN2; the claim that the material is 'cation-ordered' is therefore not independently established. The authors also explicitly state on page 5 that off-stoichiometry (e.g., Mg1−xZr1+xN2−yOy) 'cannot be accurately refined from these XRD data.' This admitted limitation is load-bearing for the central claim of stoichiometric, ordered, semiconducting MgZrN2 and should be addressed explicitly, for example by refining metal occupancies or by complementary characterization.","section":"Results and Discussion, Figure 7; main-text statement on off-stoichiometry"}],"minor_comments":[{"comment":"The text contains typographical errors and an incomplete sentence: 'Sequential Rietveld refinements were conducted onri in situ SXPRD datasets' should read 'conducted on the in situ SPXRD datasets,' and the phrase 'We therefore modeled preferred orie' is cut off before the description of the preferred-orientation model.","section":"Methods, 'Analysis of diffraction data'"},{"comment":"The x-axis label in Figures 1 and 7 appears as '2  , CuK )', missing the theta symbol; it should read '2θ (Cu Kα)'.","section":"Figure 1 and Figure 7 axis labels"},{"comment":"The axis label 'T emperature (°C)' in Figure S11 contains a stray space; it should read 'Temperature (°C)'.","section":"Supporting Information, Figure S11"},{"comment":"The Colorado State University affiliation contains a typo: 'F ort Collins' should be 'Fort Collins'.","section":"Affiliations"}],"recommendation":"major_revision","confidential_remarks":"The core synthesis and computational story is compelling, but the structural identification of the R3m phase is the load-bearing element, and the current XRD analysis does not conclusively establish cation ordering. The authors have the tools to fix this: a free refinement of Mg/Zr occupancies, an antisite-disorder model, or neutron diffraction would greatly strengthen the claim. If the fixed-occupancy model is retained, the paper should be reframed as reporting a 'layered rocksalt-type' phase without asserting full cation order. I do not see grounds for rejection, but the revision needs to address the ordering evidence directly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Chris, you should know this paper is a solid synthesis-first contribution. The genuinely new thing is that Li2ZrN2 undergoes topochemical ion exchange with MgBr2 to give the R3m (alpha-NaFeO2-type) polymorph of MgZrN2, and the same route works for MgHfN2 from Li2HfN2. Both are previously unreported, and they stand in contrast to the disordered rocksalt obtained by sputtering and metathesis and to the I41/amd ground state from calculations. The in situ synchrotron diffraction showing the reaction pathway and the LiBr/Li2MgBr4 intermediates is a real asset, as is the clean computational story: MgZrN2 is only +42 meV/f.u. above ground state, while Fe/Cu/Zn analogues decompose to A + ZrN + N2, which matches the observed products. The 2.0 eV onset is consistent with the calculated R3m gap, and the GW comparison is honest since the calculation was not tuned to the measurement.\n\nThe soft spots are real but not fatal. The product is a 47/53 wt% mixture of R3m and Fm3m phases, there is an unindexed peak near 64 deg 2theta, and the authors concede that Mg loss or oxide incorporation cannot be ruled out from XRD. The stress-test note is on target: the Rietveld model fixes Mg/Zr occupancies at the fully ordered 1:1 arrangement and refines only lattice parameters, size broadening, and Biso. The Rwp improvement from adding Fm3m does not test the ordering model. Given the strong scattering contrast between Mg and Zr, a free refinement of occupancies or an antisite-disorder model would discriminate order from disorder, and I would want that before calling the product unambiguously cation-ordered. The optical measurement is also on a phase mixture, so the 2.0 eV onset is attributed to the R3m component rather than proven for it. These are addressable concerns: phase-purified or compositionally characterized samples, a free occupancy refinement, and ideally a replicate optical measurement on a richer sample would firm up the central claim.\n\nCitation practice looks fine; the prior works cited are background results, and the self-citation to Zn3WN4 is legitimate. The paper is written clearly and does not oversell; the authors themselves flag the impurity and off-stoichiometry caveats. The claim that this opens a generalizable route from the 63 Li-M-N ternaries is plausible but supported mainly by two examples, which the paper acknowledges.\n\nWho is this for: the nitride synthesis community and anyone doing metastable-phase discovery. It deserves a serious referee. I would send it out, with the ordering refinement as the main requested change. My own verdict would be conditional: the synthesis is likely correct, but the structural and optical assignment needs one more pass of evidence.","headline":"A credible ion-exchange route to two new layered ternary nitrides, with the main caveat being that the cation-ordered R3m product is a 47 wt% phase mixture whose ordering is assumed rather than refined.","tokens_in":21095,"tokens_out":937,"would_cite":true,"duration_ms":11059,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A topochemical Li-for-Mg swap in Li2ZrN2 produces a metastable layered polymorph of MgZrN2, and the same reaction yields layered MgHfN2; both absorb near 2.0 eV.","keywords":["ion exchange","ternary nitrides","topochemical synthesis","metastable polymorphs","MgZrN2","MgHfN2","layered semiconductors","powder X-ray diffraction"],"falsifier":"A concrete test: prepare a phase-pure or near-phase-pure $R\\overline{3}m$ sample (for instance, by lowering the reaction temperature or shortening the dwell) and measure its diffuse reflectance; if the absorption onset is not near 2.0 eV, or if a neutron-diffraction or resonant X-ray measurement shows Mg and Zr are not ordered on the distinct cation sites of the $R\\overline{3}m$ model, the central claim fails. A simpler check would be to see whether the unindexed peak near 64° 2θ can be assigned to an impurity that accounts for the optical onset.","tokens_in":20114,"feed_emoji":"🧪","tokens_out":9475,"duration_ms":83065,"temperature":0.7,"pith_summary":"This paper reports a moderate-temperature ion exchange route that converts the layered lithium nitride Li$_2$ZrN$_2$ into a previously unknown layered polymorph of MgZrN$_2$ with space group $R\\overline{3}m$ and the $\\alpha$-NaFeO$_2$ structure, and extends the same reaction to make layered MgHfN$_2$. The authors argue the exchange is topochemical: the octahedral [ZrN$_6$] layers of the precursor survive into the product, yielding a metastable cation-ordered arrangement rather than the calculated $I4_1/amd$ ground state or the disordered rocksalt phase obtained in earlier syntheses. Optical diffuse reflectance shows an absorption onset near 2.0 eV for the MgZrN$_2$ product, consistent with calculated bandgaps for the $R\\overline{3}m$ polymorph and slightly higher than the 1.8 eV onset reported for disordered rocksalt MgZrN$_2$. The same exchange fails for Fe, Cu, and Zn analogues, which decompose into metal, ZrN, and N$_2$; temperature-dependent free energy calculations attribute that failure to the greater metastability of those phases, and the authors use the success with Mg to argue that the roughly 63 known Li-M-N compounds are promising precursors for new ternary nitrides.","feed_headline":"Ion exchange yields two new layered nitride semiconductors","feed_subtitle":"A topochemical Li-for-Mg swap preserves octahedral nitride layers, giving MgZrN2 and MgHfN2 with ~2.0 eV absorption.","key_machinery":"The load-bearing object is the layered Li$_2$ZrN$_2$ precursor, whose octahedral [ZrN$_6$] layers act as a structural template for the product. During the exchange, Li$^+$ leaves and Mg$^{2+}$ enters while the Zr-N layers are preserved; the anion sublattice shifts from hexagonal close packing in the precursor to cubic close packing in the product, while the cation sublattice stays cubic close packed. The reaction pathway is followed with in situ synchrotron powder X-ray diffraction and sequential Rietveld refinement, and the thermodynamic competition facing each candidate AZrN$_2$ phase is assessed with DFT formation enthalpies plus a machine-learned vibrational-entropy descriptor that gives temperature-dependent Gibbs free energies.","core_discovery":"On the paper's own terms, the central discovery is that Li$^+$ and Mg$^{2+}$ undergo topochemical ion exchange in the reaction Li$_2$ZrN$_2$ + MgBr$_2$, preserving the layers of octahedral [ZrN$_6$] from the $P\\overline{3}m1$ Li$_2$ZrN$_2$ precursor and yielding a metastable $R\\overline{3}m$ ($\\alpha$-NaFeO$_2$-type) polymorph of MgZrN$_2$ rather than the $I4_1/amd$ ground state predicted by DFT. The same procedure on Li$_2$HfN$_2$ gives layered MgHfN$_2$. The $R\\overline{3}m$ phase is 42 meV per formula unit above the calculated ground state and, in the as-made powder from a 10-minute 650 °C reaction, coexists with a cation-disordered $Fm\\overline{3}m$ rocksalt phase in a 47/53 wt% ratio; unindexed impurity scattering near 64° 2θ remains outside the fit. UV-vis diffuse reflectance on this un-optimized powder shows an absorption onset near 2.0 eV, matching the GW-calculated absorption of the $R\\overline{3}m$ polymorph, and in situ synchrotron powder X-ray diffraction shows Li$_2$ZrN$_2$ being consumed as MgZrN$_2$ grows above roughly 500 °C, supporting the topochemical pathway.","pith_inferences":["The paper leaves implicit that the measured 2.0 eV onset comes from a powder that is only 47 wt% $R\\overline{3}m$; a testable extension is that a higher-$R\\overline{3}m$ preparation should show a sharper, slightly blue-shifted onset, while any sub-gap absorption should shrink.","The Mg-versus-Fe/Cu/Zn stability window suggests that other divalent cations with formation enthalpies closer to Mg's, such as Ca or Sr, may be the next viable targets for layered AMN$_2$ nitrides by the same exchange.","A natural next experiment is the one the paper only flags: exchanging Li$_2$WN$_4$ or LiWN$_2$, whose WN$_6$ units are tetrahedral or trigonal prismatic rather than octahedral, which would test whether the topochemical template survives for non-octahedral nitride frameworks."],"forward_implications":["Cation ordering shifts the optical response: $R\\overline{3}m$ MgZrN$_2$ shows an absorption onset near 2.0 eV, versus 1.8 eV for the cation-disordered $Fm\\overline{3}m$ thin films, so controlling structure directly controls the gap.","The exchange is generalizable within the Li-M-N family: the same chemistry converts Li$_2$HfN$_2$ to layered MgHfN$_2$, and the roughly 63 known Li-M-N phases are candidate precursors for other A-M-N ternaries.","The layered products are metastable by design: DFT places $R\\overline{3}m$ MgZrN$_2$ at +42 meV/formula unit above the $I4_1/amd$ ground state, and longer heating converts more of the sample to the disordered rocksalt phase.","The Fe, Cu, and Zn analogues are not accessible by this route at the temperatures required, because temperature-dependent free energies favor A + ZrN + $\\tfrac{1}{2}$N$_2$; the calculations explain both the Mg success and the others' decomposition."],"supporting_citations":[{"why":"Shows the heterovalent Li+ to Zn2+ ion exchange route to Zn3WN4, the direct methodological precedent for swapping two Li+ for one divalent cation.","marker":"[23]"},{"why":"Provides the computational map of ternary nitrides that predicts cation-ordered structures such as I41/amd MgZrN2, framing the synthesis target.","marker":"[8]"},{"why":"Reports the disordered rocksalt MgZrN2 thin films with a 1.8 eV absorption onset that the new R3m polymorph is compared against.","marker":"[10]"},{"why":"Documents prior bulk metathesis synthesis of rocksalt MgZrN2, the earlier bulk result this work contrasts with.","marker":"[12]"},{"why":"Argues that the α-NaFeO2 (R3m) and γ-LiFeO2 cation orderings are nearly equal in energy, supporting the feasibility of the R3m polymorph.","marker":"[28]"},{"why":"Supplies the machine-learned descriptor for temperature-dependent Gibbs energies used to explain why Fe, Cu, and Zn analogues decompose.","marker":"[32]"},{"why":"Demonstrates ion exchange synthesis of metastable CuNbN2, a precedent for accessing metastable ternary nitrides by this route.","marker":"[18]"},{"why":"Demonstrates ion exchange synthesis of metastable CuTaN2 delafossite, another precedent for the method's reach into metastable phases.","marker":"[19]"},{"why":"Gives the synthesis of Li2ZrN2 and Li2HfN2 precursors that the exchange reactions start from.","marker":"[38]"}],"fun_headline_variants":["Ion exchange unlocks layered MgZrN2 and MgHfN2","New metastable nitride semiconductors from Li swap","Topochemical ion exchange builds layered nitride pairs","Layered MgZrN2 and MgHfN2 via lithium exchange","Ion exchange makes two new layered nitride semiconductors"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the diffraction data uniquely identify the product as cation-ordered $R\\overline{3}m$ MgZrN$_2$ with the modeled Mg/Zr ordering, even though the sample is a 47/53 wt% mixture with the disordered $Fm\\overline{3}m$ phase, has an unindexed peak near 64° 2θ, and could, by the authors' own caveat, contain Mg loss or oxide incorporation that XRD alone cannot rule out.","fun_headline_variants_meta":{"raw":{"variants":["Ion exchange unlocks layered MgZrN2 and MgHfN2","New metastable nitride semiconductors from Li swap","Topochemical ion exchange builds layered nitride pairs","Layered MgZrN2 and MgHfN2 via lithium exchange","Ion exchange makes two new layered nitride semiconductors"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000205,"raw_usage":{"total_tokens":1526,"prompt_tokens":1211,"completion_tokens":315,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":827,"completion_tokens_details":{"reasoning_tokens":234}},"tokens_in":827,"tokens_out":315,"duration_ms":3503,"temperature":1.0,"reasoning_tokens":234,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T23:16:12.729882+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete test: prepare a phase-pure or near-phase-pure $R\\overline{3}m$ sample (for instance, by lowering the reaction temperature or shortening the dwell) and measure its diffuse reflectance; if the absorption onset is not near 2.0 eV, or if a neutron-diffraction or resonant X-ray measurement shows Mg and Zr are not ordered on the distinct cation sites of the $R\\overline{3}m$ model, the central claim fails. A simpler check would be to see whether the unindexed peak near 64° 2θ can be assigned to an impurity that accounts for the optical onset.","supporting_citations":[{"cited_title":"L.; O'Donnell, S.; Huang, K.; Klein, R","cited_arxiv_id":null,"evidence_quote":"Shows the heterovalent Li+ to Zn2+ ion exchange route to Zn3WN4, the direct methodological precedent for swapping two Li+ for one divalent cation."},{"cited_title":"J.; Arca, E.; Bauers, S","cited_arxiv_id":null,"evidence_quote":"Provides the computational map of ternary nitrides that predicts cation-ordered structures such as I41/amd MgZrN2, framing the synthesis target."},{"cited_title":"R.; Holder, A.; Sun, W.; Melamed, C","cited_arxiv_id":null,"evidence_quote":"Reports the disordered rocksalt MgZrN2 thin films with a 1.8 eV absorption onset that the new R3m polymorph is compared against."},{"cited_title":"L.; Fallon, M","cited_arxiv_id":null,"evidence_quote":"Documents prior bulk metathesis synthesis of rocksalt MgZrN2, the earlier bulk result this work contrasts with."},{"cited_title":"A.; Armiento, R.; Alling, B.; Eklund, P","cited_arxiv_id":null,"evidence_quote":"Argues that the α-NaFeO2 (R3m) and γ-LiFeO2 cation orderings are nearly equal in energy, supporting the feasibility of the R3m polymorph."},{"cited_title":"J.; Millican, S","cited_arxiv_id":null,"evidence_quote":"Supplies the machine-learned descriptor for temperature-dependent Gibbs energies used to explain why Fe, Cu, and Zn analogues decompose."},{"cited_title":"J.; Zhang, X.; Vidal, J.; Cui, Z.; Lany, S.; Yang, M.; DiSalvo, F","cited_arxiv_id":null,"evidence_quote":"Demonstrates ion exchange synthesis of metastable CuNbN2, a precedent for accessing metastable ternary nitrides by this route."},{"cited_title":"S.; DiSalvo, F","cited_arxiv_id":null,"evidence_quote":"Demonstrates ion exchange synthesis of metastable CuTaN2 delafossite, another precedent for the method's reach into metastable phases."},{"cited_title":"G.; Alexander, I","cited_arxiv_id":null,"evidence_quote":"Gives the synthesis of Li2ZrN2 and Li2HfN2 precursors that the exchange reactions start from."}],"review_version":1}