{"id":"2d6863e3-9890-44dc-9971-fa775de39b6f","arxiv_id":"2507.00393","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Atomically ordered HfO2/ZrO2 superlattices sustain ferroelectricity to 100 nm, more than ten times thicker than conventional hafnia films.","lead":"Epitaxial (HfO2)n/(ZrO2)n superlattices retain ferroelectric switching up to 100 nm thickness, far beyond the usual 10 nm limit of hafnia-based films. The authors attribute the stability to interfacial energetics and report a low coercive field of about 0.85 MV/cm with endurance above 10^9 cycles.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 100 nm ferroelectricity claim lacks leakage-excluded electrical evidence: PUND and frequency-dependent P-E are shown only for 20 nm, so the thick-film loops could be leakage-dominated.","rationale":"The reader's weakest-assumption concern about a '7 nm critical thickness versus 100 nm film' appears to conflate the per-layer critical thickness with the total film thickness. In the n=3 superlattice, each HfO2 and ZrO2 layer is only a few unit cells thick, well below the computed 7 nm threshold, and the stack contains many alternating interfaces that can repeatedly suppress m-phase formation. The theoretical mechanism is therefore not internally inconsistent with a 100-nm total film, though a full-stack calculation would strengthen it. After setting that concern aside, the most load-bearing risk to the central claim is the absence of leakage-excluded electrical evidence at 100 nm. The paper's main electrical controls—PUND, frequency dependence, and temperature dependence—are demonstrated on the 20-nm superlattice, and the endurance test is on a 6-nm film. The 100-nm P-E loops in Fig. 4d are accompanied by I-E current peaks, but those peaks are not compared to the leakage current at the same fields, and no PUND data are shown for that thickness. Because the coercive field of 0.85 MV/cm requires roughly 8.5 V across a 100-nm film, leakage and charge-injection artifacts are a real risk in hafnia-based capacitors. This concern is directly testable, and the proposed PUND/frequency/leakage check on the 100-nm sample would settle whether the headline thickness record is genuine. If the measurement confirms true switching, the central claim stands; if not, the paper should be revised to restrict the ferroelectric-stability claim to thinner films or to add leakage corrections. The reader's overall conditional verdict remains appropriate, so no verdict change is recommended.","tokens_in":13906,"tokens_out":8794,"duration_ms":112635,"concrete_test":"Perform positive-up-negative-down (PUND) measurements and frequency-dependent P-E measurements on the 100-nm n=3 superlattice using the same Pt/LSMO electrode stack, and record leakage current as a function of field near ±0.85 MV/cm. If the PUND switched polarization is well above the noise level (e.g., >2–3 μC/cm²) and the P-E loops remain essentially unchanged from 1 to 10 kHz, the 100-nm ferroelectricity claim is supported. If PUND shows negligible switched charge or the apparent polarization collapses with increasing frequency, the thick-film loops are leakage-dominated and the headline claim would be unsubstantiated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that optimized-period superlattices remain ferroelectric at 100 nm rests on P-E hysteresis loops and I-E switching-current peaks in Fig. 4d. For a ferroelectric capacitor, thick films measured at high voltages can produce apparent polarization from leakage and charge injection that mimics ferroelectric switching, especially when the coercive field is as low as 0.85 MV/cm and the applied voltage is about 8.5 V. The paper provides PUND measurements (Supplementary Fig. S6) and frequency-stability data (Fig. 3f) only for the 20-nm n=3 superlattice, not for the 100-nm film. Without a leakage-excluded measurement on the thick film, the 'robust ferroelectricity at 100 nm' claim is not fully secured. Note that the reader's identified weakness about a 7-nm critical thickness versus a 100-nm film is likely a misreading: the computed critical thickness refers to an individual ZrO2 layer, and the n=3 superlattice repeats thin layers (each below 7 nm) 55 times, which is consistent with the proposed interfacial stabilization mechanism. The load-bearing uncertainty is therefore experimental: whether the 100-nm hysteresis loops represent true ferroelectric switching rather than non-ferroelectric artifacts.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports epitaxial (HfO2)n/(ZrO2)n superlattices grown on LSMO-buffered SrTiO3 by pulsed laser deposition, with n = 2, 3, 6, 11, 16 and total thicknesses from 4 to 100 nm. Structural characterization (XRD, XRR, AFM, XAS/XLD, and atomic-resolution STEM/EDXS) indicates well-defined interfaces and a dominant polar orthorhombic phase, with n = 3 the optimal periodicity. Electrical measurements on ~20-nm films show ferroelectric P-E loops, I-E switching peaks, butterfly permittivity loops, PUND results, low leakage, and frequency/temperature stability; a 6-nm film is reported to survive 10^9 fatigue cycles. For thicker films, P-E loops and I-E peaks are reported up to 100 nm, with coercive field as low as ~0.85 MV/cm. DFT calculations examine kinetic barriers and o-m interface formation energies, giving a critical thickness of about 7 nm for an individual ZrO2 layer. The authors argue that kinetic barriers and interfacial energy stabilize the polar phase over a wide thickness range.","tokens_in":14072,"tokens_out":6468,"duration_ms":65247,"significance":"If fully substantiated, the results would significantly extend the thickness window of ferroelectric HfO2-based films, which is of technological relevance for memories and low-power electronics. The structural characterization is extensive and internally consistent, and the DFT calculation is first-principles and not fitted to the measured polarization, so there is no circularity in the theory-experiment comparison. However, the central thickness claim at 100 nm currently lacks leakage-excluded electrical evidence, and the theoretical mechanism as written does not directly bridge the per-layer critical thickness to the total film thickness. Both points are addressable with additional measurements and clearer modeling, and I would not reject the manuscript on the basis of the existing data alone.","major_comments":[{"comment":"The 100-nm ferroelectricity claim rests on the P-E hysteresis loops and I-E switching-current peaks in Fig. 4d, but PUND and frequency-dependent P-E measurements are shown only for the 20-nm n = 3 film (Supplementary Fig. S6 and Fig. 3f). Because thick films measured at high voltages can exhibit apparent polarization from leakage or charge injection, please provide leakage-excluded measurements for the 10-100 nm films (for example PUND, frequency-dependent P-E, or transient current analysis), or explicitly qualify the claim as loop-based evidence.","section":"Thickness variation ferroelectric properties of the superlattices (Fig. 4d)"},{"comment":"The calculated critical thickness of about 7.01 nm refers to an individual ZrO2 layer, whereas the experimental claim concerns a 100-nm total film assembled from 55 periods. The manuscript moves from the per-layer value to the conclusion that the superlattice can maintain the polar phase in thicker dimensions without a quantitative argument for the whole stack. I do not read the 7 nm number as a contradiction of 100-nm ferroelectricity, because each sublayer is below 7 nm, but the text should explicitly distinguish these two length scales and explain (or model) how repeated interfaces preserve the polar phase across the full 100-nm thickness.","section":"The origin and stability of ferroelectricity in the superlattices (critical-thickness paragraph)"},{"comment":"The abstract attributes excellent fatigue resistance exceeding 10^9 switching cycles to the optimized-period superlattices in the same sentence as the 100-nm thickness range, but the endurance data in Fig. 4c are for a 6-nm superlattice. Please state which thickness each headline claim refers to, and if no fatigue data exist for the 100-nm film, adjust the wording or add the corresponding measurement.","section":"Abstract and Figure 4c"}],"minor_comments":[{"comment":"The Pt electrode diameter is given as 12.5 µm², which mixes diameter and area; the quantity should be a length (e.g., 12.5 µm) or an area such as 12.5 µm² for the electrode area.","section":"Methods (electrode fabrication)"},{"comment":"The sentence 'the upper limit of periodicity for retaining ferroelectricity in the system is appears to be n = 16' contains a grammatical error ('is appears'); it should read 'appears to be n = 16'.","section":"Results, periodicity dependence"},{"comment":"The phrase 'maintain stable ferroelectricity from up to 100 nm' is awkward and should be 'at thicknesses up to 100 nm'.","section":"Abstract and Figure 4d"},{"comment":"The labels identifying the five thicknesses in the P-E loops of Fig. 4d are difficult to distinguish in the printed figure; using distinct colors with a clear legend would help.","section":"Figure 4d"}],"recommendation":"major_revision","confidential_remarks":"For the editor: the primary revision should focus on obtaining or clearly labeling leakage-excluded electrical evidence for the thick films and on clarifying that the DFT critical thickness is per sublayer. The other comments are presentation-level. If the authors can supply those data and revise the modeling discussion, the manuscript would be publishable in a high-impact venue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a paper worth taking seriously. The experimental picture is unusually complete—epitaxial (111) films, atomically sharp interfaces, no interdiffusion beyond one monolayer, and ferroelectric hysteresis in a 4 nm film. The n=3 superlattice keeps a clear polar phase at 100 nm, with a low coercive field (~0.85 MV/cm) and 10^9-cycle endurance at 6 nm. That is new. Earlier hafnia/zirconia superlattices were polycrystalline or ultrathin, and epitaxial hafnia usually loses polarization around 20 nm.\n\nThe weak spot is not the DFT mechanism, though you could read it that way. The paper's computed critical thickness of ~7 nm is for an individual ZrO2 layer, not for the whole stack. In the n=3 superlattice, each HfO2/ZrO2 period is only about 1–2 nm thick, repeated 55 times, so every layer stays below the threshold. The mechanism—interfacial o-m energy plus kinetic barriers suppresses m-phase nucleation—is internally consistent with the thickness range observed, and the calculations are first-principles, not fitted to the polarization.\n\nThe real soft spot is electrical. The 100 nm ferroelectricity claim rests on P-E loops and switching-current peaks at applied fields around 8.5 V. PUND and frequency-stability measurements are shown only for the 20 nm film. At those voltages, leakage or charge injection can produce loop shapes that look ferroelectric, even if the I-E curves show maxima. The authors do report low leakage for the superlattice, but only at 20 nm. That gap matters. It is fixable—add PUND and frequency-dependent loops on the 100 nm film—but without it, the headline claim is not fully secured.\n\nMinor issues: the categorical statement that no epitaxial HfO2-based superlattices have been reported before needs verification; the \"most exceptional ferroelectric properties\" phrase in the results is too strong; and Figure 4e has no error bars even though the methods say each sample was measured multiple times.\n\nBottom line: if the 100 nm data survive a leakage check, this is a significant advance. The materials science is careful, the literature is engaged, and the mechanism makes sense once you map the critical thickness onto the individual layers. Send it to referees.","headline":"Epitaxial HfO2/ZrO2 superlattices show credible ferroelectricity to 100 nm, but the thick-film data need a leakage check before the headline claim is fully secure.","tokens_in":14731,"tokens_out":3367,"would_cite":true,"duration_ms":35929,"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":"Epitaxial HfO2/ZrO2 superlattices preserve ferroelectricity from 4 to 100 nm, far beyond the usual hafnia limit.","keywords":["ferroelectric hafnia","HfO2/ZrO2 superlattice","polar orthorhombic phase","ferroelectric stability","epitaxial thin film","coercive field","fatigue endurance","first-principles calculations"],"falsifier":"Measure the monoclinic-phase fraction by X-ray diffraction or cross-sectional TEM as a function of total thickness from 10 to 100 nm for the $n=3$ superlattice; if it rises steeply beyond the computed roughly 7 nm critical thickness, the proposed stabilization mechanism is contradicted.","tokens_in":13662,"feed_emoji":"⚡","tokens_out":14639,"duration_ms":141609,"temperature":0.7,"pith_summary":"Epitaxial $(\\mathrm{HfO}_2)_n/(\\mathrm{ZrO}_2)_n$ superlattices with a short repeat period are reported to keep ferroelectricity across an unusually wide thickness range, from 4 nm to 100 nm, while conventional $\\mathrm{Hf}_{0.5}\\mathrm{Zr}_{0.5}\\mathrm{O}_2$ solid-solution films lose it beyond roughly 10 nm. The best period ($n=3$) shows a remnant polarization around $15\\ \\mu\\mathrm{C}/\\mathrm{cm}^2$, a coercive field that drops to about $0.85$ MV/cm in thick films, and endurance beyond $10^9$ switching cycles. First-principles calculations attribute the stability to a kinetic barrier: when ZrO$_2$ is constrained by the ferroelectric orthorhombic lattice of neighboring HfO$_2$, the barrier for the transition to the stable non-ferroelectric monoclinic phase rises by about $273$ meV/f.u., while the barrier to the ferroelectric phase is nearly unchanged. The paper also computes an orthorhombic/monoclinic interface formation energy that is higher for the HfO$_2$-ZrO$_2$ interface than for an HZO solid-solution interface, giving a critical thickness of about 7 nm and, by extension, the ability of the polar phase to dominate at much larger thickness. If correct, this would extend the usable thickness window of hafnia-based ferroelectrics by an order of magnitude and lower the operating field for memory devices.","feed_headline":"Ferroelectricity survives to 100 nm in HfO2/ZrO2 superlattices","feed_subtitle":"Ordered HfO2/ZrO2 layers keep their polarization through a billion cycles and reach a coercive field near 0.85 MV/cm.","key_machinery":"The central object is the $(\\mathrm{HfO}_2)_n/(\\mathrm{ZrO}_2)_n$ superlattice grown epitaxially on LSMO-buffered SrTiO$_3$, with the period $n$ counted in unit cells and $n=3$ as the optimal value. The argument rests on two first-principles quantities: the kinetic energy barrier of the tetragonal-to-monoclinic transition in ZrO$_2$, which increases by about $273$ meV/f.u. when the ZrO$_2$ layer is clamped by the polar HfO$_2$ lattice while the barrier to the orthorhombic phase barely changes, and the orthorhombic/monoclinic interface formation energy, computed as $70.4$ meV/Å$^2$ for the HfO$_2$-ZrO$_2$ interface and $64.6$ meV/Å$^2$ for a comparable HZO interface. Comparing the interface energy with the bulk monoclinic-orthorhombic energy difference yields a critical thickness of about $7$ nm for the ferroelectric ZrO$_2$ layer; the paper uses this comparison to argue that the elemental discontinuity at the superlattice interfaces suppresses monoclinic-phase growth more effectively than a solid solution.","core_discovery":"The central claim is that a chemically sharp, epitaxial HfO$_2$/ZrO$_2$ superlattice suppresses the non-polar monoclinic phase through two computed mechanisms: the kinetic barrier for the tetragonal-to-monoclinic transition in ZrO$_2$ rises substantially when ZrO$_2$ is constrained by the polar lattice of HfO$_2$, while the barrier to the polar orthorhombic phase stays low, and the orthorhombic/monoclinic interface formation energy is high enough (70.4 meV/Å$^2$ for the HfO$_2$-ZrO$_2$ interface, versus 64.6 meV/Å$^2$ for a comparable HZO interface) to make monoclinic nucleation expensive during growth. The paper reports that the polar orthorhombic phase remains dominant up to 100 nm total thickness, with a polar-phase fraction above 80% at all measured thicknesses, and that the $n=3$ period gives the best ferroelectric response: remnant polarization near 15 $\\mu$C/cm$^2$, stable switching over $10^9$ cycles, and a coercive field as low as $\\sim0.85$ MV/cm in thick films. This is the paper's evidence that ordered superlattices, built from two paraelectric bulk oxides, offer a defect-free route to stable ferroelectricity in hafnia-based materials beyond the conventional thickness window.","pith_inferences":["A direct test of the mechanism would be to grow the same superlattice on substrates with different lattice mismatches: if kinetic clamping by the HfO$_2$ lattice is the dominant stabilizer, the polar phase should persist with only weak strain dependence; if strain is dominant, the thickness window should collapse when the substrate constraint is removed.","The critical-thickness argument is computed for a thin (111) supercell; extending the first-principles calculations to much thicker supercells could show whether the o-m interface energy genuinely controls phase selection across the full 100 nm or whether strain relaxation and defect kinetics dominate in the upper layers.","The same elemental-discontinuity design could be transferred to other oxide pairs, such as HfO$_2$/CeO$_2$ or doped variants; testing whether a higher interface formation energy further widens the ferroelectric window would isolate the proposed control parameter.","The low coercive field at 100 nm, if reproducible in polycrystalline films on metal electrodes, would be the practical payoff for memory applications more than the epitaxial geometry itself."],"forward_implications":["If the stabilization mechanism is correct, hafnia-based ferroelectrics can be made tens of nanometres thick while keeping a low coercive field, which relaxes the voltage and leakage constraints of memory cells.","The reported endurance of more than $10^9$ cycles, with leakage current nearly unchanged, points to interfaces that block charged-defect migration and could yield more reliable ferroelectric memories.","Switching that is nearly independent of frequency and temperature is consistent with nucleation-limited switching, a property useful for high-temperature and neuromorphic electronics.","The $n=3$ design rule gives a recipe for maximizing the polar phase in fluorite-oxide superlattices and suggests that other paraelectric oxide pairs could be engineered the same way.","Two paraelectric bulk oxides can be combined into a ferroelectric without doping, offering a defect-free route to stabilize HfO$_2$ for CMOS-compatible devices."],"supporting_citations":[{"why":"Supplies the strain/symmetry-constraint mechanism for stabilizing epitaxial hafnia and the critical-thickness concept the superlattice argument extends.","marker":"[22]"},{"why":"Reports the earlier HfO2-ZrO2 superlattice gate stack that motivates the superlattice approach and gives a comparison for ferroelectric performance.","marker":"[28]"},{"why":"Establishes the O K-edge XAS/XLD signature used to identify the polar orthorhombic phase and provides the ultrathin-film baseline.","marker":"[38]"},{"why":"Documents the thickness limit and temperature behavior of doped HfO2 films that the superlattices are claimed to surpass.","marker":"[18]"},{"why":"Shows epitaxial strain stabilizing a ferroelectric phase in Hf0.5Zr0.5O2, the comparison point for coercive field and phase confirmation.","marker":"[10]"},{"why":"Provides prior HfO2-ZrO2 superlattice endurance data that the 10^9-cycle result extends and supports the interface-defect-blocking explanation.","marker":"[44]"}],"fun_headline_variants":["HfO2/ZrO2 superlattices sustain ferroelectricity to 100 nm","Epitaxial design stabilizes HfO2 ferroelectricity without doping","Billion-cycle endurance from HfO2/ZrO2 superlattice ferroelectrics","Interface engineering boosts HfO2/ZrO2 ferroelectric stability"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that computer-modeled energy barriers and interface energies, calculated for a few atomic layers, determine which phase forms in the top 90 nm of a 100 nm film grown at 600 °C, even though the computed critical thickness is about 7 nm.","fun_headline_variants_meta":{"raw":{"variants":["HfO2/ZrO2 superlattices sustain ferroelectricity to 100 nm","Epitaxial design stabilizes HfO2 ferroelectricity without doping","Billion-cycle endurance from HfO2/ZrO2 superlattice ferroelectrics","Interface engineering boosts HfO2/ZrO2 ferroelectric stability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000341,"raw_usage":{"total_tokens":1948,"prompt_tokens":1085,"completion_tokens":863,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":701,"completion_tokens_details":{"reasoning_tokens":771}},"tokens_in":701,"tokens_out":863,"duration_ms":8495,"temperature":1.0,"reasoning_tokens":771,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T21:18:03.852829+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the monoclinic-phase fraction by X-ray diffraction or cross-sectional TEM as a function of total thickness from 10 to 100 nm for the $n=3$ superlattice; if it rises steeply beyond the computed roughly 7 nm critical thickness, the proposed stabilization mechanism is contradicted.","supporting_citations":[{"cited_title":"& Liu, S","cited_arxiv_id":null,"evidence_quote":"Supplies the strain/symmetry-constraint mechanism for stabilizing epitaxial hafnia and the critical-thickness concept the superlattice argument extends."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports the earlier HfO2-ZrO2 superlattice gate stack that motivates the superlattice approach and gives a comparison for ferroelectric performance."},{"cited_title":"Enhanced ferroelectricity in ultrathin films grown directly on silicon","cited_arxiv_id":null,"evidence_quote":"Establishes the O K-edge XAS/XLD signature used to identify the polar orthorhombic phase and provides the ultrathin-film baseline."},{"cited_title":"Intrinsic ferroelectricity in Y-doped HfO2 thin films","cited_arxiv_id":null,"evidence_quote":"Documents the thickness limit and temperature behavior of doped HfO2 films that the superlattices are claimed to surpass."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows epitaxial strain stabilizing a ferroelectric phase in Hf0.5Zr0.5O2, the comparison point for coercive field and phase confirmation."},{"cited_title":"HfO2-ZrO2 Superlattice Ferroelectric Capacitor With Improved Endurance Performance and Higher Fatigue Recovery Capability","cited_arxiv_id":null,"evidence_quote":"Provides prior HfO2-ZrO2 superlattice endurance data that the 10^9-cycle result extends and supports the interface-defect-blocking explanation."}],"review_version":1}