{"id":"d57da84f-1d3b-41a8-a7ab-6d6f6618d991","arxiv_id":"2509.07194","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Operando electron microscopy reveals a four-stage field-driven pathway in PbZrO3 from antiferroelectric through a monoclinic ferroelectric intermediate to rhombohedral and tetragonal ferroelectric states, with a dead layer near the substrate.","lead":"Using electron microscopy with an electric field applied to a working capacitor, researchers watched lead zirconate crystals switch from a non-polar antiferroelectric to a polar ferroelectric state atom by atom. The switch passes through hidden intermediate phases, and the film's behavior varies sharply with depth near the substrate.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"FE_M phase identification is the load-bearing weak point: the monoclinic intermediate is inferred from qualitative 2D-DFT superlattice spots and averaged motifs, without quantitative image simulation or 3D diffraction, so a heterophase AFE_O+FE_R mixture could mimic it and collapse the four-stage pa","rationale":"The reader's weakest_assumption correctly identifies the FE_M identification as the load-bearing point. My independent reading of the manuscript converges on the same issue: the phase sequence AFE_O→FE_M→FE_R→FE_T is the central novel claim, and every subsequent interpretive step (low-energy pathway, residual AFE-like order, relative energy ranking from the transition front) relies on FE_M being a distinct phase. The main text's Fig. 2b and the Fourier-filtered images in Fig. 5e-f present only qualitative support; the actual DFT pattern is not shown, no reflection indices are given, and the comparison to refs 11 and 15 is by analogy. Since the specimen is imaged in 2D projection, the projected Fourier transform cannot distinguish a homogeneous FE_M phase from a finely intermixed AFE_O/FE_R state. The paper does include real strengths: same-field-of-view tracking across a full bias cycle, consistency of the AFE_O, FE_R, and FE_T motifs with known structures, and an explicit lower-dose control (Fig. S4/Video S1, though the SI is not provided in the preprint). These support the broad observation of field-induced AFE-to-FE transformation and the identification of the endpoint phases. However, they do not resolve the uniqueness of the FE_M assignment. The proposed multislice simulation test is specific and would settle whether a homogeneous FE_M model is required by the data or whether the observed images are equally explained by phase coexistence. If the latter, the central claim should be downgraded to a two-phase coexistence observation. This is exactly the condition attached to the reader's CONDITIONAL verdict, so I recommend no change in verdict.","tokens_in":10121,"tokens_out":5582,"duration_ms":58838,"concrete_test":"Perform quantitative multislice image simulation of the experimental STEM parameters (acceleration voltage, Cs, aperture, defocus, thickness) for three structural models: (1) a homogeneous FE_M model with MPb-type ↑↓↑↓ Pb displacements and MO-type a0a0c+ octahedral rotations; (2) a two-phase AFE_O+FE_R mixture within the same projected region; (3) a strained/canted AFE_O variant. Compare simulated ADF and dDPC images and their Fourier transforms to the experimental FE_M motif (Fig. 2b) using normalized cross-correlation or RMS difference. If model (1) is not the best fit, or if model (2) is statistically indistinguishable, the four-stage pathway is not established; the same comparison should be run for the dynamic front images (Fig. 5c-d).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The decisive step in the paper's central claim is the identification of FE_M as a distinct intermediate phase (Fig. 2b, main text 'Atomic-Scale Phase Transition Pathway'). The evidence is (i) 'monoclinic-type superlattice reflections' in the discrete Fourier transform of a 2D ADF/dDPC image, described qualitatively but not indexed or shown; (ii) an averaged motif showing checkerboard octahedral rotations and a residual canted antiparallel displacement pattern; and (iii) analogy to refs 11 and 15. None of these uniquely determines a homogeneous FE_M structure. In a projection through a thin foil, the same Fourier spots and averaged motif could be produced by a two-phase mixture of AFE_O and FE_R domains within the field of view, or by a strained AFE variant whose canted displacements generate the checkerboard appearance. Because the four-stage pathway (AFE_O→FE_M→FE_R→FE_T), the 'stepwise transformation mechanism', and the relative-energy comparison at the phase front all depend on FE_M being a real thermodynamic phase rather than a projection artifact of coexistence, this is the load-bearing assumption. The paper provides no quantitative image simulation matching the experimental ADF/dDPC images to a FE_M model, and no 3D diffraction/tilt-series verification of the monoclinic reflections.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports operando scanning transmission electron microscopy (STEM) biasing experiments on 100-nm-thick PbZrO3 (PZO) thin-film capacitors, tracking the antiferroelectric-to-ferroelectric (AFE-to-FE) transition at atomic resolution. The authors identify a four-stage pathway, AFE_O → FE_M → FE_R → FE_T, as the applied field increases, with FE_M being a monoclinic-like intermediate phase exhibiting residual antipolar lead displacements and octahedral rotations. They further observe depth-dependent heterogeneity, including a 'dead layer' near the substrate, and a dynamic transition front at which AFE and FE phases coexist. Based on the front position and its fluctuations under bias, they infer the relative energies of the intermediate phases and the presence of a large internal field arising from substrate clamping and dislocation gradients. The central claim is that the transition proceeds via genuine, field-stabilized intermediate phases that are intrinsic to the switching process, not artifacts of electron-beam irradiation or two-phase coexistence.","tokens_in":10494,"tokens_out":2946,"duration_ms":37562,"significance":"If the four-stage pathway is correct, the work provides a long-sought direct atomic-scale view of the AFE-to-FE transition under device-relevant bias, resolving whether intermediate phases are intrinsic to field-driven switching. The experimental strengths are substantial: the same-field-of-view tracking across a full bias cycle, the explicit electron-dose checks, the correlation of NBED with macroscopic polarization, and the careful distinction of reversible dynamic behavior from irreversible beam damage. These features make the paper a strong candidate for a high-impact experimental contribution. However, the identification of the pivotal FE_M phase rests on qualitative interpretation of a few superlattice reflections in 2D projection and on analogy to prior work in doped PZT and ultra-thin PZO; the energy-ordering and internal-field claims also involve a degree of circularity. The significance of the findings is therefore conditional on strengthening these structural and inferential points.","major_comments":[{"comment":"The identification of FE_M as a distinct monoclinic phase is load-bearing for the central four-stage pathway, but the evidence is not definitive. The text cites 'monoclinic-type superlattice reflections in its discrete Fourier transform' without indexing them or showing the transform, and the averaged motif is consistent with several structural possibilities. In a 2D projection through a thin foil, the same Fourier spots and checkerboard rotation pattern could be produced by a two-phase mixture of AFE_O and FE_R domains within the field of view, or by a strained AFE variant with canted displacements. To support the claim that FE_M is a homogeneous thermodynamic phase, the authors should provide (i) indexed reflections with a clear assignment to a structural model, (ii) quantitative image simulation of ADF/dDPC images from the proposed FE_M structure matched to experiment, and/or (iii) 3D","section":"Probing Dynamic Phase Equilibria, Fig. 5e–g"},{"comment":"The claim that FE_M has lower relative energy than FE_R (main text: 'FE_M exhibits a lower relative energy when compared to FE_R') is inferred from the spatial gradient of octahedral rotations and from the transition-front behavior. However, the same front-position data are used to infer the internal-field profile (Fig. 5e–g), and then the front behavior is attributed to that inferred field. This creates a circularity: the internal field is an ad-hoc construct, and the energy ranking is read off from the very data used to define the field. An independent constraint—for example, a quantitative model of how the front position depends on applied field and an assumed internal-field profile, or direct calorimetric/structural evidence of energy differences—is needed to break the circularity. As written, the relative-energy comparison is suggestive but not uniquely determined.","section":"Probing Dynamic Phase Equilibria, 'beam-induced charging and heating'"},{"comment":"The manuscript acknowledges that 'beam-induced charging and heating effects can transiently disturb the phase boundary' and that the transition-front width increases at higher electron dose (Supplementary Fig. S4). Since the dynamic coexistence and front fluctuations are used to infer phase stability and energy ordering, the demonstration that these effects are small and non-directional is essential. A lower-dose series is a useful check, but it does not quantify the perturbation magnitude or prove that the inferred energy ordering is unchanged in the zero-dose limit. The authors should either provide a dose-series extrapolation showing that the front-width and phase-occupancy statistics converge, or explicitly discuss the uncertainty this introduces into the energy-ordering claim. As it stands, the beam-perturbation channel weakens the strength of the 'relative energies compared as a fu","section":"Fig. 4e–g and Depth-Dependent Film Response"},{"comment":"The 'dead layer' interpretation at the bottom electrode is based on reduced 1/4{110} superlattice intensity and lattice-parameter variations that remain pinned under bias. The authors rule out an electric-field gradient or depolarization field based on symmetric electrode configuration, but this is an indirect argument. A more direct test would be to measure the local field profile (e.g., via Kikuchi lines or position-dependent lattice strain) or to compare with phase-field simulations that include the proposed clamping/dislocation internal field. As written, the internal-field mechanism is plausible but not uniquely established; other mechanisms, such as oxygen vacancy accumulation near the substrate, could produce similar depth-dependent response. This does not invalidate the central pathway, but it is a secondary claim that should be softened or further supported.","section":"Minor comments"}],"minor_comments":[{"comment":"There are repeated typos: 'field field' and 'electric electric field' in the 'Atomic-Scale Phase Transition Pathway' section. Please proofread.","section":null},{"comment":"The caption states 'mean-subtracted displacements for the FEM motif is shown in the inset in (b), with a saturation range of [0, 2] pm.' The units and the choice of saturation range should be clarified; also, the inset is difficult to discern in the figure as provided.","section":null},{"comment":"The details of the dDPC imaging and displacement/rotation extraction are referred to the Supplementary Information but not described in the main text. For a general readership, at least a brief explanation of how octahedral rotations are measured from dDPC images and how errors are estimated should be included in the main text or a Methods section.","section":null},{"comment":"Refs. 11 and 15 are central to the FE_M identification, but they are not cited with specific structural parameters that would allow a quantitative comparison. Please provide the relevant details (space group, lattice parameters, reflection indices) in the text or in a table.","section":null}],"recommendation":"major_revision","confidential_remarks":"The paper is experimentally impressive and the central observation of a field-driven multi-step transition is likely to be of broad interest. However, the identification of FE_M as a distinct thermodynamic phase is the linchpin of the four-stage pathway, and the current evidence is not conclusive. The 'internal field' narrative also carries a risk of circular reasoning. I would recommend major revision, asking for stronger structural evidence (image simulation or 3D diffraction) and a more transparent treatment of the internal-field inference. If the authors can supply these, the paper would be a strong candidate for acceptance in a high-impact venue."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is the first controlled-field operando STEM study I know that tracks the same field of view through a complete bias cycle in PbZrO3, and the main story—AFE order collapses gradually and the film goes through a rhombohedral then tetragonal ferroelectric state—is credible and well supported by the atomic-scale motifs, the NBED superlattice data, and the lattice parameter trends. Second, the load-bearing claim that a true monoclinic ferroelectric phase (FE_M) appears between AFE and FE_R is not proven. The evidence is a handful of superlattice reflections in a 2D DFT of a projected image, plus averaged displacement and rotation motifs that match a model borrowed from doped PZT and ultra-thin PZO. That is a reasonable hypothesis, but a two-phase mixture of AFE_O and FE_R in the projection could produce the same checkerboard appearance and the same extra spots. There is no tilt series, no 3D diffraction, and no quantitative image simulation tying the experimental images to a FE_M model. Without that, calling it a distinct phase and using it to build a 'stepwise transformation mechanism' is overreach.\n\nWhat the paper does well beyond that: the dose-control experiments rule out beam-induced switching as the origin of the main sequence; the depth-dependent 'dead layer' near the substrate is a useful observation, and the dynamic phase front at –6 V, with reversible oscillations rather than directed growth, is a nice counterpoint to earlier irradiation studies. The interpretation in terms of competing internal and external fields is plausible, but the relative energy ranking of FE_M vs FE_R is inferred from the same front-position data that is used to estimate the internal field, so it is somewhat circular. The Data Availability statement is just 'on reasonable request' — for a paper this dependent on image analysis, that is a real weakness; the SI is not included with the preprint either.\n\nOverall, the experimental core is likely right, and the FE_M identification is a fixable weakness rather than a fatal flaw. The paper deserves serious referee time, but I would not accept the four-stage pathway as established until the FE_M phase is corroborated by diffraction or simulation and the energy comparison gets independent support.","headline":"A careful operando STEM study of PZO switching that gets the big picture right, but the claimed monoclinic intermediate phase needs stronger evidence than 2D projections and imported models.","tokens_in":10926,"tokens_out":2850,"would_cite":true,"duration_ms":33139,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["77.80.-e","68.37.Ma"],"model":"deepseek-v4-flash","headline":"Field-driven switching in PbZrO3 passes through a transient monoclinic ferroelectric phase before reaching the tetragonal ferroelectric state, and the pathway is modulated by a near-substrate dead layer.","keywords":["antiferroelectric-ferroelectric transition","PbZrO3","operando STEM","field-induced phase transition","monoclinic intermediate phase","dead layer","phase coexistence","hysteresis"],"falsifier":"Collect three-dimensional reciprocal-space data (for example, precession-assisted or tomographic nano-beam electron diffraction) at the applied bias where FE_M is claimed, and test whether the full reflection set matches the monoclinic FE_M model rather than a strained AFE_O variant or a superposition of AFE_O and FE_R domains. If the 1/4{110}-type superlattice reflections and checkerboard rotation pattern are reproduced without the monoclinic phase, the four-stage pathway claim collapses.","tokens_in":2035,"feed_emoji":"⚡","tokens_out":2535,"duration_ms":96338,"temperature":0.7,"pith_summary":"Antiferroelectric lead zirconate (PbZrO3) is a leading candidate for energy-storage capacitors because an applied voltage can switch it between a non-polar, oppositely-ordered dipole state and a polar ferroelectric state, but the atomic steps of that switch have been inferred rather than seen. This paper reports live atomic-resolution electron microscopy of a working thin-film capacitor that resolves the switch directly: as bias rises, the antiparallel lead displacements and oxygen-octahedra rotations of the ground state are suppressed together, and the film passes through a transient monoclinic ferroelectric phase before settling into a rhombohedral, then tetragonal, ferroelectric state. The authors argue this four-stage AFE_O -> FE_M -> FE_R -> FE_T pathway is intrinsic to field-driven switching, not a beam artifact, and that local energetics, such as substrate clamping, dislocations, and a near-substrate \"dead layer,\" modulate where and when each phase forms. If correct, it gives device designers an atomistic blueprint for tuning the hysteresis loop and the energy-release profile of antiferroelectric capacitors. The paper also shows a dynamic phase front where antiferroelectric and ferroelectric phases interconvert reversibly, letting the relative energies of the intermediate phases be ranked under operating conditions.","feed_headline":"PbZrO3 switching proceeds through four atomic phases","feed_subtitle":"Live atomic-resolution imaging under applied bias identifies the transient bridging phase that decides the hysteresis loop.","key_machinery":"The identifying machinery is the coupled pair of order parameters of the AFE ground state, antiparallel Pb displacements (up-up-down-down along <110>_PC) and antiferrodistortive oxygen octahedral rotations, tracked unit-cell-by-unit-cell in simultaneously acquired ADF and dDPC images. The intermediate phase is recognized by its checkerboard rotation pattern and monoclinic-type superlattice reflections (MPb/MO-type modulations), the same signatures used at ferroelectric/antiferroelectric boundaries in PZT. The operando biasing geometry, a focused-ion-beam-prepared capacitor on a MEMS chip with the electric field applied along the film normal, is what makes the pathway observable under device-","core_discovery":"Under an applied electric field, PbZrO3 does not jump directly from its antiferroelectric ground state to a single ferroelectric state. The authors show that the two structural order parameters that define the ground state, the up-up-down-down antiparallel lead displacements and the antiferrodistortive oxygen octahedral rotations, are suppressed together, and their competition with polar distortions creates a monoclinic bridging ferroelectric (FE_M) that retains residual antiparallel cation canting and checkerboard octahedral rotations. Beyond that, the film becomes rhombohedral (FE_R), then tetragonal (FE_T) as the field continues to align dipoles out of plane. This sequence is established","pith_inferences":["If the FE_M assignment from projected reflections is confirmed by three-dimensional diffraction or quantitative image simulation, the same live-bias approach could be extended to other antiferroelectrics to test whether a monoclinic bridge is a general feature of AFE-to-FE switching.","The dead-layer mechanism suggests a defect-engineering lever: patterning or modifying dislocations near the substrate could shift the internal field profile and turn the dead layer into a tunable switching region.","Because the transition front moves only 30-40 nm for a 400 kV/cm bias step, tracking the front position versus bias can serve as a local probe of the internal electric-field profile, connecting microstructure to macroscopic hysteresis.","The observed reversible beam-induced fluctuations at the front imply that stochastic switching signals in operating devices may carry a fingerprint of the local energy landscape, an avenue for using fluctuation statistics to infer phase stability."],"forward_implications":["If the four-state pathway is correct, thermodynamic and phase-field models of PZO switching must include the monoclinic FE_M state as a metastable or saddle-point phase on the AFE-to-FE route, not just a two-phase coexistence.","The FE_M phase retains AFE-like antiparallel canting and octahedral rotations, so its checkerboard and superlattice signatures can be used to detect incipient switching in other antiferroelectrics before full polarization reversal.","The depth-dependent dead layer means the effective switchable volume of a PZO capacitor is smaller than the nominal film thickness; device metrics such as stored energy density should be normalized to the active region.","Coercivity and remanent polarization in these films can arise from sub-lattice polarization shifts and local structural distortions, not only from domain-wall motion or pinning.","Reversible phase interconversion at the transition front gives an experimental handle to rank intermediate-phase energetics: near the front, FE_M sits lower in energy than FE_R."],"supporting_citations":[{"why":"Provides the experimentally determined AFE ground-state structure of PbZrO3 with antiparallel lead displacements, the zero-bias reference for the observed motifs.","marker":"[25]"},{"why":"Defines the atomic structure including oxygen octahedral rotations used for the AFE crystal model in the comparison.","marker":"[26]"},{"why":"Supplies the structural model of the intermediate monoclinic phase with MPb-type antiparallel lead displacements and MO-type octahedral rotations observed at PZT phase boundaries.","marker":"[11]"},{"why":"Reports similar intermediate states in ultra-thin PZO films near the critical thickness for ferroelectricity, the prior evidence used to cross-identify the FE_M reflections.","marker":"[15]"},{"why":"Documents a transient cycloidal polarization phase in PZO under electron-beam irradiation, the baseline the present operando measurement must distinguish from beam-induced artifacts.","marker":"[16]"},{"why":"Shows how substrate clamping modulates electromechanical responses in antiferroelectric thin films, the basis for the depth-dependent internal-field and dead-layer argument.","marker":"[9]"},{"why":"Demonstrates strain-mediated competing ferrielectric and antiferroelectric phases in PbZrO3, supporting the interpretation of translation boundaries as intermediate states.","marker":"[31]"},{"why":"Provides the exit-wave power-cepstrum analysis method used to map lattice parameters and superlattice intensities from nano-beam electron diffraction data.","marker":"[41]"},{"why":"Phase-field simulations of polar boundaries as nucleation sites for AFE-to-FE transitions, the analogue used to interpret the dynamic transition front.","marker":"[14]"}],"fun_headline_variants":["PbZrO3 phase transition in four steps seen live","Operando imaging reveals PbZrO3's multi-step switching","PbZrO3's field-driven four-phase switch","Real-time microscopy tracks PbZrO3's phase transition path"],"cache_read_input_tokens":12672,"weakest_assumption_plain":"The load-bearing premise is that the intermediate FE_M phase is a distinct phase identified from a handful of superlattice reflections in two-dimensional projected images and from averaged displacement and rotation motifs, with its structural model imported from prior studies of doped and ultra-thin PZT/PZO; if those reflections instead come from a strained antiferroelectric variant or from overlapping AFE_O and FE_R domains, the four-stage pathway reduces to a two-phase coex","fun_headline_variants_meta":{"raw":{"variants":["PbZrO3 phase transition in four steps seen live","Operando imaging reveals PbZrO3's multi-step switching","PbZrO3's field-driven four-phase switch","Real-time microscopy tracks PbZrO3's phase transition path"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001219,"raw_usage":{"total_tokens":4880,"prompt_tokens":800,"completion_tokens":4080,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":544,"completion_tokens_details":{"reasoning_tokens":4012}},"tokens_in":544,"tokens_out":4080,"duration_ms":31787,"temperature":1.0,"reasoning_tokens":4012,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T22:40:29.321886+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Collect three-dimensional reciprocal-space data (for example, precession-assisted or tomographic nano-beam electron diffraction) at the applied bias where FE_M is claimed, and test whether the full reflection set matches the monoclinic FE_M model rather than a strained AFE_O variant or a superposition of AFE_O and FE_R domains. If the 1/4{110}-type superlattice reflections and checkerboard rotation pattern are reproduced without the monoclinic phase, the four-stage pathway claim collapses.","supporting_citations":[{"cited_title":"& Hoshino, S","cited_arxiv_id":null,"evidence_quote":"Provides the experimentally determined AFE ground-state structure of PbZrO3 with antiparallel lead displacements, the zero-bias reference for the observed motifs."},{"cited_title":"& Egami, T","cited_arxiv_id":null,"evidence_quote":"Defines the atomic structure including oxygen octahedral rotations used for the AFE crystal model in the comparison."},{"cited_title":"Commun.13, 1390 (2022)","cited_arxiv_id":null,"evidence_quote":"Supplies the structural model of the intermediate monoclinic phase with MPb-type antiparallel lead displacements and MO-type octahedral rotations observed at PZT phase boundaries."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports similar intermediate states in ultra-thin PZO films near the critical thickness for ferroelectricity, the prior evidence used to cross-identify the FE_M reflections."},{"cited_title":"Mater .32, e1907208 (2020)","cited_arxiv_id":null,"evidence_quote":"Documents a transient cycloidal polarization phase in PZO under electron-beam irradiation, the baseline the present operando measurement must distinguish from beam-induced artifacts."},{"cited_title":"Mater .23, 944–950 (2024)","cited_arxiv_id":null,"evidence_quote":"Shows how substrate clamping modulates electromechanical responses in antiferroelectric thin films, the basis for the depth-dependent internal-field and dead-layer argument."},{"cited_title":"Commun.15, 3438 (2024)","cited_arxiv_id":null,"evidence_quote":"Demonstrates strain-mediated competing ferrielectric and antiferroelectric phases in PbZrO3, supporting the interpretation of translation boundaries as intermediate states."},{"cited_title":"Ultramicroscopy214, 112994 (2020)","cited_arxiv_id":null,"evidence_quote":"Provides the exit-wave power-cepstrum analysis method used to map lattice parameters and superlattice intensities from nano-beam electron diffraction data."},{"cited_title":"& Huang, H","cited_arxiv_id":null,"evidence_quote":"Phase-field simulations of polar boundaries as nucleation sites for AFE-to-FE transitions, the analogue used to interpret the dynamic transition front."}],"review_version":1}