{"id":"4866e727-87e1-4ec6-bb6f-d37a064dcf5d","arxiv_id":"2509.12952","paper_version":5,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Ultrathin HfO2(111) films keep ferroelectric order without electrodes, and their anomalous thickness-dependent lattice expansion comes from negative piezoelectricity plus surface stress.","lead":"This paper explains why ultrathin hafnia films expand as they get thinner, a puzzle in ferroelectric memory research. The explanation combines two effects: a negative piezoelectric response to depolarization fields and surface stress, and it suggests new film orientations could beat the usual size limits.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Surface-stress parameters fitted to a symmetric free-standing HfO2(111) slab may not transfer to the experimental geometry of a film supported on SrTiO3, where the bottom interface and substrate clamping alter the in-plane strain and thickness dependence.","rationale":"The paper's internal checks are solid: the negative piezoelectric coefficient from DFPT matches the direct slope of d111 vs Ed, and the surface-stress model with parameters fitted at four layers reproduces the DFT-computed in-plane strains at other thicknesses. The orientation-dependent hyperferroelectricity argument is supported by direct DFT for PbTiO3(111) and by the LGD model. The weakest link is the bridge from the idealized free-standing, symmetric HfO2 slab to the actual experimental system. The reader correctly identified the fitted surface parameters and graphene screening as the key assumptions. My concern sharpens one of these: the surface-stress model assumes two free surfaces and a free-standing film, whereas the experimental films are supported on SrTiO3, and the substrate imposes a mechanical boundary condition that can change the in-plane equilibrium strain. If the surface stress parameters or the clamping condition differ, the thickness dependence of η_IP(t) would change, directly impacting the quantitative reproduction in Fig. 3b. This is a testable, load-bearing assumption rather than a fatal flaw; the proposed DFT calculation with an explicit substrate would settle it. The verdict remains CONDITIONAL, as the reader recommended: the physics is plausible and internally consistent, but the quantitative comparison to experiment needs the transferability check.","tokens_in":10741,"tokens_out":8281,"duration_ms":99740,"concrete_test":"Perform DFT relaxations of HfO2(111) slabs of 4-9 layers on an explicit SrTiO3(001) substrate (or with a TiN electrode), allowing the in-plane lattice to relax, and extract d111(t) and η_IP(t). Compare η_IP(t) and d111(t) with the surface-stress model Eq. 2 and with the experimental data from Refs. [39,40]. If the supported-film η_IP(t) deviates by more than ~20% from Eq. 2 at the thinnest thickness, or if d111(t) does not reproduce the observed thickness-dependent increase, then the surface-stress transferability assumption is falsified and the quantitative reproduction is coincidental.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim (Fig. 3b) rests on Eq. 3, where the thickness-dependent out-of-plane expansion is driven by the in-plane strain η_IP(t) from the surface-stress model Eq. 2. The parameters σ and S in Eq. 2 are fitted from energy-strain curves of a symmetric free-standing HfO2(111) slab (Fig. S5) that has two free surfaces and no substrate. The experimental data compared in Fig. 3b are for Hf0.5Zr0.5O2 films grown on SrTiO3(001) (Refs. [39,40]). In those films, the bottom interface is HfO2/SrTiO3 rather than a free surface, and the in-plane lattice is constrained by the substrate and by any top electrode. The free-standing slab's two-surface stress is therefore not obviously transferable to a single-interface supported film. Moreover, the model computes the partial-screening curve by 'imposing the same in-plane strain' as the free-standing slab on a HfO2/graphene capacitor, which assumes the substrate does not alter the equilibrium in-plane strain. If the actual interface stress or substrate clamping changes η_IP(t), the predicted d111(t) and the claimed quantitative reproduction would not be robust. The paper does not provide a calculation for a supported film or an interface-aware surface stress, so this transferability assumption is currently unverified.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper addresses the anomalous out-of-plane lattice expansion observed in ultrathin (111)-oriented HfO2 films. Using DFT on HfO2(111) slabs and capacitors with different electrodes, the authors argue that the expansion is not a signature of enhanced ferroelectricity but coexists with suppressed polar displacements. They identify two mechanisms: a negative longitudinal piezoelectric response to the residual depolarization field and a thickness-dependent surface stress. A surface elasticity model with parameters fitted to a DFT slab is used to compute the in-plane strain, and the total out-of-plane strain is decomposed into a Poisson contribution and a depolarization-field contribution. This model is claimed to quantitatively reproduce the experimental d111(t) data for Hf0.5Zr0.5O2 films. The paper also predicts orientation-induced hyperferroelectricity in HfO2(111) and PbTiO3(111) under open-circuit conditions and identifies Ca2N as a near-ideal electrode.","tokens_in":11117,"tokens_out":9905,"duration_ms":125980,"significance":"If the central claim holds, the paper offers a resolution to a long-standing puzzle in hafnia ferroelectrics and proposes a practical design principle: choosing a film orientation with a reduced out-of-plane polarization component can bypass the critical-thickness limit. The internal consistency of the DFT evidence is a notable strength. The DFPT value deff_33 = −1.02 pm/V agrees with the slope extracted from capacitor calculations (−1.28 pm/V), and the surface-elasticity model fitted at one slab thickness reproduces DFT free-standing slab strains at other thicknesses. This shows the model is not merely an interpolation. The LGD treatment provides a transparent mechanism for the orientation dependence. The main caveat, however, is that the quantitative match to experiments rests on unverified transferability assumptions: the fitted surface parameters come from a free-standing symmetric slab, and the partial-screening curve is obtained by imposing that slab's in-plane strain on a different capacitor geometry. The experimental comparison is also made to Hf0.5Zr0.5O2 films on SrTiO3, not to pure HfO2, and no HZO or supported-film calculation is provided.","major_comments":[{"comment":"The central quantitative claim in Fig. 3b rests on the in-plane strain η_IP(t) from Eq. (2), with σ and S fitted from a single four-layer free-standing OCBC HfO2(111) slab (Fig. S5). The model is then validated for other free-standing slab thicknesses (Fig. 3a). However, the partial-screening curve in Fig. 3b is computed by 'imposing the same in-plane strain' from the free-standing slab on a HfO2/graphene capacitor. This is an assumption, not a calculation. The equilibrium in-plane strain of the graphene-capped capacitor is governed by two HfO2/graphene interfaces, not two free surfaces, and no DFT minimization of η_IP(t) for the capacitor geometry is reported. Moreover, the experimental films are supported on SrTiO3(001), where substrate clamping and interfacial stress should alter the in-plane strain. Since the first term of Eq. (3), −ν η_IP(t), is linear in η_IP(t), any transferabilit","section":"Surface elasticity model, Eqs. (2)-(3) and Fig. 3"},{"comment":"The experimental d111 data are for Hf0.5Zr0.5O2 films, while all DFT slabs and the LGD model are for pure HfO2. The partial-screening curve corresponds to a HfO2/graphene capacitor with a specific Ed = −10.4 MV/cm, which is not derived from the actual electrode/interface conditions of the measured films. Without an explicit calculation for Zr-alloyed HfO2, or at least a demonstration that alloying leaves the negative piezoelectric coefficient and the effective surface-stress parameters unchanged, the match in Fig. 3b is suggestive rather than a quantitative validation for the measured system. The manuscript should state this limitation clearly and, ideally, provide bounds on how composition and screening variations shift the predicted curve.","section":"Fig. 3b and comparison to Refs. [39,40]"}],"minor_comments":[{"comment":"The arXiv title reads 'Origin of Anomalous Size Effects', while the main-text title reads 'Origin of Reverse Size Effect'. Please harmonize the terminology.","section":"Title and abstract"},{"comment":"Reference [21] (Lv et al., on perovskite nanocrystals) does not appear related to surface elasticity; the surface-stress model is better supported by Ref. [22]. Please check and correct this citation.","section":"Eq. (1) and reference [21]"},{"comment":"The value C = 4.44 eV/Å^3 is introduced without specifying how it is derived (bulk elastic constant? from the same DFPT calculation?). Please state the source.","section":"Eq. (2), elastic constant C"},{"comment":"The experimental data points from Refs. [39,40] are shown without error bars or film-thickness uncertainty. Adding these would help assess how 'quantitative' the agreement really is.","section":"Fig. 3b"},{"comment":"The generalization to PbTiO3 rests on one DFT slab calculation (Fig. S6). A brief statement about thickness dependence or the implications of the slab termination would strengthen the claim.","section":"Fig. 4 and PbTiO3 slab"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope and the central idea is appealing. The internal DFT consistency is good, but the quantitative match to experiments is currently underpinned by an unverified assumption about the transferability of free-standing slab surface stress to a different capacitor geometry and to HZO films. This is a fixable issue: additional supported-film or capacitor-equilibrium calculations, or a clearly stated and justified model for the experimental geometry, would be needed. I do not see a fatal circularity, since the surface-elasticity model is fitted at one thickness and checked against DFT at other thicknesses. The HfO2-vs-HZO composition mismatch should also be addressed. I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper is worth your time. The core claim is that the anomalous out-of-plane lattice expansion in ultrathin HfO2 is not a sign of enhanced polarization but actually coexists with suppressed polar displacements, and it comes from two separate mechanisms: a negative longitudinal piezoelectric response to the residual depolarization field plus a positive surface stress. That synthesis is genuinely new as far as I know, and the authors back it with internally consistent DFT. The negative d33 story checks out: their computed piezoelectric tensor gives -1.02 pm/V along [111], and the direct slope from the d111-Ed curve is -1.28 pm/V. That agreement is not fitted; it's a real cross-check. The surface elasticity model also reproduces the DFT in-plane strains across thickness with just two parameters, which is reasonable.\n\nThe second new piece is orientation-induced hyperferroelectricity: (111)-oriented HfO2 and PbTiO3 retain polarization under open-circuit conditions because only the out-of-plane component of Ps enters the depolarization penalty. The LGD argument is clean, and they confirm the PbTiO3 prediction with a direct slab calculation. That is a nice generalizable idea and probably the part with the most legs.\n\nNow the soft spots, in proportion. The quantitative reproduction of the experimental d111(t) is the load-bearing claim, and it rests on transferring surface stress parameters sigma and S fitted from a symmetric free-standing HfO2(111) slab to a Hf0.5Zr0.5O2 film on SrTiO3 with a bottom interface and substrate clamping. The stress-test note is right that this transfer is unverified. The paper does not compute a supported film or an interface-aware stress, so the agreement in Fig. 3b could be partly fortuitous. Also, the partial-screening comparison uses HfO2/graphene as a stand-in for the actual screening environment in the experimental HZO films. That is a modeling assumption, not a demonstrated equivalence. The paper should address this with transfer tests—e.g., a slab with a SrTiO3 substrate or at least a discussion of how interface stress would shift the curve. The lack of error bars on the experimental data is a minor issue by comparison.\n\nA smaller concern: the surface elastic parameters are fitted from the same four-layer slab the model is later applied to, so the model is not fully predictive for thickness dependence. But the model does reproduce the DFT strains at other thicknesses, which breaks the circularity enough to keep the central mechanism believable.\n\nWho is this for? Anyone working on hafnia thin films, ferroelectric size effects, or electrode screening. The orientation-hyperferroelectricity idea deserves a serious referee. My verdict: conditional accept, mostly solid, with the transferability of the surface-stress fit as the main thing to fix. I would bring it to a reading group and would cite it once the interface question is addressed.\n\nRecommendation: send it to peer review. The core physics is interesting and mostly well-supported, and the weak spot is fixable rather than fatal.","headline":"A mostly convincing mechanistic explanation for the reverse size effect in hafnia, with a clean orientation-hyperferroelectricity argument; the main soft spot is the experimental comparison resting on fitted surface parameters from a free-standing slab.","tokens_in":11561,"tokens_out":748,"would_cite":true,"duration_ms":11016,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["77.80.-e","77.65.-j"],"model":"deepseek-v4-flash","headline":"Ultrathin hafnia's out-of-plane lattice expansion is a sign of suppressed polarization, not enhanced ferroelectricity.","keywords":["ferroelectric hafnia","reverse size effect","depolarization field","negative piezoelectricity","surface stress","hyperferroelectricity","thin-film orientation","Ca2N electrode"],"falsifier":"Measure, on the same ultrathin hafnia film, both the out-of-plane lattice spacing and the switchable polarization as functions of thickness: if d111 increases while polarization also increases, or if d111 decreases when the depolarization field is screened out by improved electrodes, the paper's central claim would be contradicted.","tokens_in":10666,"feed_emoji":"⚡","tokens_out":3730,"duration_ms":43582,"temperature":0.7,"pith_summary":"The paper argues that the anomalous out-of-plane lattice expansion observed in ultrathin hafnia (HfO2) films is not evidence of stronger ferroelectricity. Instead, the expansion coexists with a suppressed out-of-plane polarization, and it arises from two mechanisms: a negative longitudinal piezoelectric response to the residual depolarization field, and a positive surface stress that grows as the film thins. A model combining these effects quantitatively reproduces the measured thickness dependence of the lattice spacing. The paper further shows that (111)-oriented HfO2 and PbTiO3 films can sustain out-of-plane polarization even under open-circuit conditions, a mechanism it calls orientation-induced hyperferroelectricity, and it identifies the two-dimensional electride Ca2N as a near-ideal screening electrode.","feed_headline":"Ultrathin hafnia's odd expansion marks weaker polarization, not stronger","feed_subtitle":"A two-mechanism model — negative piezoelectricity plus surface stress — reproduces the measured d111 thickness trend","key_machinery":"The argument rests on two quantitative tools. First, a surface elastic model (total energy = bulk strain energy plus surface energy with surface stress σ and surface elastic constant S) yields the thickness-dependent in-plane strain η_IP(t) = -2σ/(Ct+2S), which enters the out-of-plane expansion through the Poisson effect. Second, a phenomenological free-energy model for a ferroelectric under open-circuit conditions adds a depolarization penalty proportional to P_s^2/[1+χ_e]^2, showing that the (111) orientation's reduced out-of-plane polarization component lowers this penalty enough to keep the ferroelectric state stable. The negative piezoelectric coefficient of HfO2, computed as d33^eff =","core_discovery":"For (111)-oriented HfO2 slabs under open-circuit conditions, the out-of-plane polar displacement drops to roughly one-third of its bulk value, yet the out-of-plane interplanar spacing expands by about 0.03 Å. This inverse correlation is the paper's central discovery: the expansion is driven by a negative effective longitudinal piezoelectric coefficient (computed as d33^eff = -1.02 pm/V) responding to the depolarization field, plus a thickness-dependent surface stress described by a surface elastic model with parameters fitted from first-principles slabs. The combined model gives η_OP(t) = -ν η_IP(t) + η_OP(E_d), matching experimental d111 data for Hf0.5Zr0.5O2 films. Separately, a free-energ","pith_inferences":["A direct experimental test could vary the screening environment (e.g., electrode material or capping layer) on identical ultrathin hafnia films; the model predicts that improved screening should decrease d111 while increasing switchable polarization.","The orientation-induced hyperferroelectricity mechanism may apply to other fluorite-structure or non-perovskite ferroelectrics where the polarization vector has a component along the surface normal smaller than the total magnitude.","The surface-stress parameters fitted from a single slab could be checked for transferability by computing them at several thicknesses and seeing whether Eq. 2 still reproduces the DFT strains; such a check would test the model's predictive power.","If Ca2N behaves as calculated, it could serve as a practical electrode for nanoscale ferroelectric memory, potentially avoiding the dead-layer problem that plagues metal electrodes."],"forward_implications":["If the model is correct, the measured increase in d111 with decreasing film thickness is direct evidence of a residual depolarization field and negative piezoelectricity, not of enhanced polarization.","Orientation engineering along (111) provides a general strategy to eliminate the critical thickness limit in conventional perovskite ferroelectrics such as PbTiO3, extending beyond hafnia.","Ultrathin capacitors using Ca2N electrodes should restore near-bulk polarization, avoiding the depolarization-driven suppression seen with weaker screening.","The reverse size effect should vanish in perfectly screened capacitors, where only the surface-stress-induced Poisson expansion remains.","For (001)-oriented hafnia, open-circuit conditions stabilize a nonpolar ground state, affirming the orientation-specific nature of the mechanism."],"fun_headline_variants":["Negative piezoelectricity explains hafnia's odd lattice expansion","Ultrathin hafnia widens as polarization shrinks – model solves puzzle","Two mechanisms, one paradox: why thin hafnia stretches","Hafnia's counterintuitive stretch: negative piezoelectricity plus surface stress"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The quantitative reproduction of the experimental d111 thickness dependence rests on surface-stress parameters fitted from a single four-layer slab and on treating a HfO2/graphene capacitor as representative of the partial screening in the actual Hf0.5Zr0.5O2 films; if those fitted parameters or that screening proxy are not transferable, the quantitative claim is not established.","fun_headline_variants_meta":{"raw":{"variants":["Negative piezoelectricity explains hafnia's odd lattice expansion","Ultrathin hafnia widens as polarization shrinks – model solves puzzle","Two mechanisms, one paradox: why thin hafnia stretches","Hafnia's counterintuitive stretch: negative piezoelectricity plus surface stress"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000679,"raw_usage":{"total_tokens":2958,"prompt_tokens":813,"completion_tokens":2145,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":557,"completion_tokens_details":{"reasoning_tokens":2067}},"tokens_in":557,"tokens_out":2145,"duration_ms":17637,"temperature":1.0,"reasoning_tokens":2067,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T16:28:14.638617+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure, on the same ultrathin hafnia film, both the out-of-plane lattice spacing and the switchable polarization as functions of thickness: if d111 increases while polarization also increases, or if d111 decreases when the depolarization field is screened out by improved electrodes, the paper's central claim would be contradicted.","supporting_citations":[],"review_version":1}