{"id":"0c04afc6-ba90-4ac5-8802-35293777b822","arxiv_id":"2507.18592","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"The InP(111) surface termination alone selects whether MBE-grown MnTe forms the hexagonal NiAs (altermagnetic) or cubic ZnS (non-centrosymmetric) polymorph.","lead":"Growing manganese telluride on indium phosphide wafers whose surface ends in either indium or phosphorus atoms yields two different crystal forms of the material, one of them a candidate 'altermagnet' and the other a wide-gap non-centrosymmetric semiconductor. The work shows that a single substrate type, with only its top atomic layer changed, can be used to program which functional phase forms.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The DFT rationalization rests on an interface structure (Mn-terminated MnTe on P-terminated InP) that contradicts the STEM-EELS observation of a mixed In/Te double layer, so the small 5 meV/Å2 preference for ZnS may be an artifact; the experimental phase-selectivity result is not undermined.","rationale":"The reader's weakest assumption identifies essentially the same issue: the DFT interface-energy comparison uses restricted, idealized terminations and an imposed in-plane lattice match, so the small energy differences may not represent the physical interfaces. My independent reading adds two supporting observations that make the concern more concrete. First, XPS (Figure 2) shows that after the Te-anneal step both InP(111)A and InP(111)B are covered by a terminating Te layer, so the 'In-terminated' and 'P-terminated' surfaces used in the DFT models are not the actual surfaces immediately before MnTe deposition. Second, STEM-EELS (Figure 4d) resolves a mixed In/Te double layer at the ZnS/InP interface, which is structurally different from the model's Mn-termination on P-terminated InP. These facts mean the 5 meV/Å2 preference assigned to the ZnS/P-terminated combination is computed for an interface that the experiments do not show. The experimental core of the paper is robust across multiple independent probes, so the verdict should remain CONDITIONAL: accept the phase-selectivity observation while withholding full endorsement of the proposed interfacial-energy mechanism until the model is reconciled with the measured interface. I therefore recommend no change to the reader's verdict, and I partially agree with the reader's weakest-assumption analysis.","tokens_in":12143,"tokens_out":4019,"duration_ms":49237,"concrete_test":"Recompute the interface energy density Eint (Eq. 1) for ZnS-MnTe on P-terminated InP(111)B using the experimentally observed interface structure from Figure 4(b,d)—an In-rich lower layer and a Te-rich upper layer with direct Te-P bonding—rather than the idealized Mn-terminated interface used in Figure 6. If the ZnS preference over NiAs disappears or reverses when the realistic mixed interface is modeled, then the proposed interfacial-energy mechanism is not established, and the experimental phase selectivity must stand without that explanation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central experimental claim—that InP(111)A yields NiAs-MnTe and InP(111)B yields ZnS-MnTe—is strongly supported by XRD, RHEED, HAADF-STEM, EELS, APT, and XPS. The load-bearing weakness is in the theoretical mechanism that the paper uses to explain this selectivity. In the DFT section, the authors state that they considered only Mn-terminated MnTe on P-terminated InP and Te-terminated MnTe on In-terminated InP, explicitly to avoid artificial charge accumulation. However, STEM-EELS in Figure 4(b,d) shows that the real ZnS-MnTe/InP(111)B interface is a mixed double layer: a lower In-rich layer (with some Te) and an upper Te-rich layer (with some In), with direct Te-P bonding. This is not the Mn-on-P interface used in the calculation. Moreover, XPS in Figure 2 shows that both substrates are Te-terminated prior to growth, so neither surface presents the bare InP termination assumed in the model. The computed interface-energy advantage of ZnS on P-terminated InP is only 5 meV/Å2, and the advantage of NiAs on In-terminated InP is only 1 meV/Å2, which is within the typical numerical uncertainty of such DFT slab calculations. Because the calculated preference is small and is evaluated for an interface geometry that differs from the one observed, the theoretical rationalization is not secured. The paper itself acknowledges that the exact mechanism remains unresolved. This does not invalidate the experimental demonstration, but it means the claim that interfacial energetics explain the phase selection is conditional on an unverified interface model.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports epitaxial growth of two MnTe polymorphs on InP(111) substrates by molecular beam epitaxy, with the In-terminated (111)A surface yielding hexagonal NiAs-MnTe and the P-terminated (111)B surface yielding cubic ZnS-MnTe. The films are characterized as phase-pure and of high crystalline quality by XRD, RHEED, HAADF-STEM, EELS, XPS, and APT, and the interfaces are reported to be abrupt with minimal chemical intermixing. The authors propose that the surface termination triggers phase selection at the interface, and they support this with DFT interface-energy calculations that show a preference for NiAs on In-terminated InP and for ZnS on P-terminated InP.","tokens_in":12370,"tokens_out":3846,"duration_ms":41429,"significance":"If the termination-controlled phase selection is correct, the work provides a simple and integrable route to grow both the altermagnetic NiAs-MnTe and the non-centrosymmetric ZnS-MnTe on a single lattice-matched semiconductor platform, which is of substantial interest for spintronics and multiferroic applications. The experimental evidence is strong and multi-technique: XRD and RHEED establish phase purity and epitaxial quality, HAADF-STEM and EELS resolve the interfacial structure, and APT rules out chemical intermixing as a stabilization mechanism. The DFT calculation is a valuable attempt to provide a mechanism, but as presented it has limitations that prevent it from fully securing the interfacial-energetics explanation.","major_comments":[{"comment":"The DFT interface models do not correspond to the experimentally observed interfaces. XPS in Figure 2 shows that both substrate terminations are Te-terminated prior to growth, and STEM-EELS in Figure 4(b,d) shows that the ZnS-MnTe/InP(111)B interface consists of a mixed In/Te double layer with direct Te-P bonding, rather than the Mn-terminated MnTe on P-terminated InP used in the calculation. The computed 5 meV/Ų preference for ZnS on P-terminated InP is therefore evaluated for a geometry that is not the physical one, so the theoretical rationalization is not secured.","section":"Theoretical modeling and Figure 4"},{"comment":"The computed interface-energy differences are very small: 1 meV/Ų for NiAs on In-terminated InP and 5 meV/Ų for ZnS on P-terminated InP. These values are within the typical numerical uncertainty of DFT slab calculations, particularly given the Hubbard U treatment of Mn 3d states and the constrained in-plane lattice. The manuscript does not report convergence tests with respect to slab thickness, k-point sampling, or U value, nor does it provide error estimates; the claim that interfacial energetics 'favor' one polymorph over the other is therefore quantitatively under-supported.","section":"Eq. (1) and Figure 6"},{"comment":"The in-plane lattice parameters of both polymorphs are constrained to match InP, which imposes a large strain on ZnS-MnTe (approximately 8.3% mismatch relative to its unstrained lattice parameter of 6.4 Å). The strain contribution to the interface energy is not separated or discussed, and the calculation does not consider the possibility of strain relaxation at or near the interface. Without this analysis, it is unclear whether the calculated ZnS preference is a physical effect or an artifact of the imposed strain.","section":"Theoretical modeling"},{"comment":"The Discussion states that 'the exact mechanism remains unresolved,' yet the Conclusion says that the DFT calculations 'confirm' that NiAs-MnTe forms a more favorable interface on In-terminated InP and ZnS-MnTe on P-terminated surfaces. Given the interface-model mismatch and small energy differences noted above, 'confirm' overstates the level of support. The theoretical mechanism should be presented as a hypothesis consistent with, but not conclusively established by, the calculations.","section":"Discussion and Conclusion"}],"minor_comments":[{"comment":"The word 'stochiometric' is misspelled and should be 'stoichiometric'.","section":"Introduction"},{"comment":"The phrase 'III-V ZnS structures' appears to be an error: ZnS-MnTe is a II-VI compound, not a III-V one.","section":"Conclusion"},{"comment":"The University of Missouri affiliation lists 'Colombia, MO'; the correct spelling is 'Columbia, MO'.","section":"Affiliation"},{"comment":"The schematic in Figure 3(a) and the RHEED sequence in Figure 3(b) show steps (i), (ii), and (v), but steps (iii) and (iv) are not described in the text or visible in the figure; this should be clarified.","section":"Figure 3"}],"recommendation":"major_revision","confidential_remarks":"The experimental core of this paper is convincing and likely of high interest to the mtrl-sci community. The main weakness is in the DFT rationalization, which is undermined by the interface-model mismatch and the lack of numerical uncertainty analysis. This is fixable by reframing the DFT as a hypothesis rather than confirmation, or by performing additional calculations on the actual observed interface structures. I do not see grounds for rejection, but the theory section needs substantial revision before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main result is real. In-terminated InP(111)A gives phase-pure NiAs-MnTe and P-terminated InP(111)B gives phase-pure ZnS-MnTe, with the phase locked in from the first layers. The evidence is about as good as epitaxial thin-film work gets: XRD, RHEED, HAADF-STEM, EELS, APT, and XPS all point the same way, and the growth recipe is documented carefully, including the Te pre-treatment. The finding is new—previous InP growth work did not control termination—and it gives the altermagnetism and multiferroics communities a reliable platform for two functional polymorphs on one substrate.\n\nThe soft spot is the DFT rationalization. The interface-energy differences are small, 1 to 5 meV per square angstrom, which is within the typical noise of slab calculations. More seriously, the model assumes bare In- and P-terminated InP surfaces, but the XPS data show both surfaces are Te-terminated before growth, and the STEM-EELS of the ZnS interface shows a mixed In/Te double layer rather than the Mn-on-P interface used in the calculation. So the calculated preference for ZnS on the P-terminated surface may not describe the physically realized interface. The paper itself concedes the exact mechanism is unresolved, yet the DFT section still presents the termination-dependent interface energies as the explanation. That framing needs adjustment.\n\nNone of this undermines the experimental phase-selection claim. The phase purity and interface abruptness are established by multiple independent probes, and the theoretical argument is supplementary, not load-bearing. The discussion honestly floats Pauling's rules and kinetic limitations as alternative contributors. What a referee should request is a more careful presentation of the DFT as provisional, maybe with additional interface terminations and error bars, not a re-do of the growth.\n\nA minor complaint: data are only \"available from the corresponding author upon reasonable request.\" For a synthesis paper, that is weaker than a public repository, but not disqualifying.\n\nThis paper is for MBE practitioners and anyone working on MnTe or altermagnetic candidate materials. I would cite it if I were in that space, and it deserves a serious referee. My recommendation: send to peer review, treat the experimental core as strong, and push for a more cautious theoretical interpretation.","headline":"Solid experimental demonstration of termination-controlled polymorph selection; the DFT rationalization is suggestive but not load-bearing.","tokens_in":13129,"tokens_out":1531,"would_cite":true,"duration_ms":19875,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The chemical termination of an InP(111) surface selects which MnTe polymorph grows under identical MBE conditions.","keywords":["MnTe","molecular beam epitaxy","surface termination","polymorph selectivity","altermagnetism","zinc blende","NiAs structure","InP(111)"],"falsifier":"Grow MnTe on InP(111) faces whose outermost monolayer is deliberately flipped (an In layer on the P face, or a P layer on the In face) under otherwise identical conditions: if the phase follows the intentional top layer rather than the bulk face, the termination-template claim is supported; if it does not, the claim fails. As a second test, recompute the interface energies including the mixed In/Te double-layer structure seen by STEM-EELS for the ZnS interface; losing the 5 meV/Ų preference for ZnS on P-terminated InP would remove the DFT support for interfacial stabilization.","tokens_in":11839,"feed_emoji":"🧲","tokens_out":7887,"duration_ms":74742,"temperature":0.7,"pith_summary":"Molecular beam epitaxy of MnTe on InP(111) is shown to produce either of two functional crystal phases depending on one detail of the substrate: the chemical identity of its outermost atomic layer. An In-terminated (111)A surface nucleates the hexagonal NiAs phase, which is the altermagnetic antiferromagnet with a room-temperature Néel transition, while a P-terminated (111)B surface nucleates the cubic ZnS phase, a wide-gap, non-centrosymmetric semiconductor. The paper argues, on the basis of microscopy, diffraction, spectroscopy, and density-functional interface energies, that the phase is decided at the interface and then propagates through the film. This matters because it turns polymorph selection into a programmable growth parameter, giving clean platform films for testing intrinsic altermagnetism and for exploring magnetism-charge coupling on a standard semiconductor substrate.","feed_headline":"Flip one atomic layer, grow a different MnTe crystal","feed_subtitle":"InP(111)A grows altermagnetic NiAs-MnTe; the B face grows wide-gap ZnS-MnTe under identical MBE conditions.","key_machinery":"The central mechanism is the substrate surface termination acting as an epitaxial template: the outermost atomic plane of InP(111), either In for the (111)A face or P for the (111)B face, sets the coordination and interfacial energy for the first MnTe layers. The argument is carried by the interface energy density $E_{\\mathrm{int}} = (E - E_{\\mathrm{MnTe}} - E_{\\mathrm{InP}})/A$ computed with spin-polarized density-functional theory with a Hubbard $U$ on Mn $3d$ states for MnTe slabs lattice-matched to InP, by in-situ RHEED showing that the in-plane lattice parameter is set from the first layer, and by Pauling's-rule analysis placing MnTe on the tetrahedral/octahedral borderline (cation/anion radius ratio $\\approx 0.414$).","core_discovery":"Under identical growth conditions, the MnTe phase that forms on InP(111) is determined by the substrate surface termination: the In-terminated (111)A face yields phase-pure hexagonal NiAs-MnTe, and the P-terminated (111)B face yields phase-pure cubic ZnS-MnTe. The films show atomically abrupt, chemically clean interfaces with no In or P uptake, as seen by HAADF-STEM, STEM-EELS, and atom-probe tomography, and their out-of-plane and in-plane lattice parameters match the respective bulk polymorphs. First-principles calculations with a Hubbard U find that NiAs-MnTe is the bulk ground state but ZnS is close in energy, and that the computed interface energy density favors NiAs on the In-terminated interface ($-11$ meV/Ų vs $-10$ meV/Ų) and ZnS on the P-terminated interface ($-37$ meV/Ų vs $-32$ meV/Ų), matching the observed selectivity. The exact microscopic mechanism is not yet resolved, but the MnTe cation/anion radius ratio sits near the ZnS/NiAs coordination border, so subtle interfacial chemistry may tip the balance.","pith_inferences":["A direct test of the template picture would be to cap a P-terminated surface with one In monolayer and an In-terminated surface with one P monolayer; if the phase follows the added monolayer, surface termination alone is the active selector.","Because the computed ZnS-on-P advantage (5 meV/Ų) is small and the actual interface shows a mixed In/Te double layer, including that measured interface structure in the DFT could shift the energy ordering, indicating whether the selectivity is truly interfacial energetics or a kinetic nucleation effect.","The same termination-as-selector principle might transfer to other lattice-matched compound semiconductors and other borderline-coordination tellurides, offering a general polymorph-by-design epitaxy strategy."],"forward_implications":["A single InP substrate platform, with only its (111) termination changed, yields either altermagnetic NiAs-MnTe or wide-gap non-centrosymmetric ZnS-MnTe, making phase selection a growth knob rather than a lucky outcome.","Phase-pure NiAs-MnTe films provide a cleaner test bed for intrinsic altermagnetism than bulk crystals, which are inevitably Mn-rich or contain MnTe$_2$ intergrowths.","Phase-pure ZnS-MnTe films on a standard III-V substrate open a route to studying antiferromagnet/ferroelectric coupling and possible multiferroic behavior.","The near-complete selectivity (rare ZnS intergrowths in NiAs films, none in ZnS films) indicates the interfacial template is robust under the reported growth window."],"supporting_citations":[{"why":"Supplies the MnTe phase diagram establishing NiAs as the stable bulk phase and ZnS as a metastable polymorph the growth must select against.","marker":"[1]"},{"why":"Defines the altermagnetic class and motivates why phase-pure NiAs-MnTe is a target material.","marker":"[5]"},{"why":"Identifies NiAs-MnTe as a prototypical altermagnet, the property that makes the In-terminated growth result significant.","marker":"[6]"},{"why":"Provides the modified oxide-strip and MBE growth recipe on InP(111) that this work adapts and whose phase selectivity it establishes.","marker":"[11]"},{"why":"Gives the NiAs-MnTe lattice parameters used to identify the film phase and its near lattice match to InP.","marker":"[13]"},{"why":"Reports the first epitaxial ZnS-MnTe films, establishing the precedent that the metastable phase can be stabilized by a substrate.","marker":"[21]"},{"why":"Provides the ZnS-MnTe cubic lattice parameter and magnetic ordering used to index and interpret the P-terminated films.","marker":"[28]"},{"why":"Implements the density-functional total-energy and relaxation calculations used for the interface energy comparison.","marker":"[42]"},{"why":"Implements the density-functional total-energy and relaxation calculations used for the interface energy comparison.","marker":"[43]"}],"fun_headline_variants":["One atomic layer decides MnTe crystal phase on InP","Switch InP surface termination, switch MnTe polymorph","InP(111) face dictates altermagnetic or wide-gap MnTe","MnTe phase on InP: it's all about the last atomic layer","Tuning one surface atom layer programs MnTe polymorph"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The finding hinges on the assumption that the physically realized interfaces match the calculated ones: Te-terminated MnTe on In-terminated InP and Mn-terminated MnTe on P-terminated InP, even though the actual ZnS interface shows a mixed In/Te double layer and sits at an 8.3% unstrained lattice mismatch.","fun_headline_variants_meta":{"raw":{"variants":["One atomic layer decides MnTe crystal phase on InP","Switch InP surface termination, switch MnTe polymorph","InP(111) face dictates altermagnetic or wide-gap MnTe","MnTe phase on InP: it's all about the last atomic layer","Tuning one surface atom layer programs MnTe polymorph"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000261,"raw_usage":{"total_tokens":1651,"prompt_tokens":1059,"completion_tokens":592,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":675,"completion_tokens_details":{"reasoning_tokens":504}},"tokens_in":675,"tokens_out":592,"duration_ms":6574,"temperature":1.0,"reasoning_tokens":504,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:31:54.595756+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Grow MnTe on InP(111) faces whose outermost monolayer is deliberately flipped (an In layer on the P face, or a P layer on the In face) under otherwise identical conditions: if the phase follows the intentional top layer rather than the bulk face, the termination-template claim is supported; if it does not, the claim fails. As a second test, recompute the interface energies including the mixed In/Te double-layer structure seen by STEM-EELS for the ZnS interface; losing the 5 meV/Ų preference for ZnS on P-terminated InP would remove the DFT support for interfacial stabilization.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the MnTe phase diagram establishing NiAs as the stable bulk phase and ZnS as a metastable polymorph the growth must select against."},{"cited_title":"Šmejkal, J","cited_arxiv_id":null,"evidence_quote":"Defines the altermagnetic class and motivates why phase-pure NiAs-MnTe is a target material."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Identifies NiAs-MnTe as a prototypical altermagnet, the property that makes the In-terminated growth result significant."},{"cited_title":"Chilcote, A","cited_arxiv_id":null,"evidence_quote":"Provides the modified oxide-strip and MBE growth recipe on InP(111) that this work adapts and whose phase selectivity it establishes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the NiAs-MnTe lattice parameters used to identify the film phase and its near lattice match to InP."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports the first epitaxial ZnS-MnTe films, establishing the precedent that the metastable phase can be stabilized by a substrate."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the ZnS-MnTe cubic lattice parameter and magnetic ordering used to index and interpret the P-terminated films."},{"cited_title":"Kresse, J","cited_arxiv_id":null,"evidence_quote":"Implements the density-functional total-energy and relaxation calculations used for the interface energy comparison."}],"review_version":1}