{"id":"4a63ec37-e6b9-4d55-85d6-6287d8111f2f","arxiv_id":"2607.03465","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.5,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Homogeneous nucleation of stacking-disordered ice is followed by thickening of cubic layers that nucleate defect-free cubic germs, enforcing cubic twinning and eight-branched octahedral dendrites that retain global cubic symmetry even after late-stage hexagonal conversion.","lead":"Deeply supercooled water freezes into stacking-disordered ice that still grows into eight-branched dendrites with global cubic symmetry. Direct cryo-TEM shows how thickening cubic monolayers seed the twinning that locks in that macroscopic order.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"The central kinetic pathway is inferred from post-growth spatial gradients under membrane confinement; residual uncertainty remains whether surface or quasi-2D effects alter the observed cubicity evolution and germ-mediated twinning relative to bulk homogeneous nucleation.","rationale":"The reader correctly isolates the homogeneous-nucleation assumption as the single most load-bearing residual uncertainty. The multi-modal TEM, 3D-ED, and MD evidence for the thickening–germ–twinning sequence is internally consistent and novel; the only soft point is whether that sequence is unperturbed by the carbon-membrane geometry that enables the experiment. The proposed thickness-variation test directly probes that point without requiring new instrumentation. Because the existing temperature, randomness and volume-scaling data already make surface nucleation unlikely, the concern does not rise to a level that would change the ACCEPT verdict; it remains a medium-risk caveat that can be closed by the suggested control.","tokens_in":19214,"tokens_out":601,"duration_ms":25667,"concrete_test":"Map initial cubicity (core HRTEM stacking analysis as in Fig. S7) and secondary/tertiary germ size distributions (Fig. S9) across a statistically powered set of droplets whose central thickness (EFTEM) spans 30–200 nm while keeping diameter fixed; if core cubicity or germ occurrence changes systematically with thickness (or with membrane material when carbon is replaced by graphene/SiN), confinement is modulating the kinetic pathway and the bulk claim weakens.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim (cubic-layer thickening produces defect-free Ic germs that act as facet-registered seeds enforcing successive {111} twinning into eight-branched octahedral dendrites) rests on the premise that the membrane-encapsulated droplets freeze by bulk-like homogeneous nucleation at -40 °C, free of surface templating or confinement bias. Evidence is the matching temperature, site randomness across melt–freeze cycles, and volume (not area) scaling of nucleation probability (Fig. S2). However, the droplets are only 30–150 nm thick (EFTEM, Fig. S1), lens-shaped, and fully sandwiched by amorphous carbon; growth is therefore quasi-two-dimensional. All molecular-resolution data (HRTEM cubicity maps, germ imaging) are acquired after solidification is complete (beam blanked during cooling; Methods). Consequently the radial increase in cubicity (0.35 core \to 0.62 tertiary) and the location of pure-Ic germs at branch junctions could in principle be influenced by membrane–ice interactions or by the thin-film geometry that forces all eight {111} directions into the imaging plane. If either effect is present, the claimed general kinetic pathway from microscopic stacking disorder to macroscopic cubic symmetry would not transfer to unconfined bulk water.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript reports in situ cryogenic TEM imaging of homogeneous ice nucleation and growth from deeply supercooled water (−40 °C) inside membrane-encapsulated microdroplets. Homogeneous nucleation yields stacking-disordered ice (Isd) in which cubic ice initially appears only as isolated monolayers (cubicity ~0.35). Progressive thickening of cubic segments produces nanoscale, defect-free cubic germs on basal planes; these germs act as facet-registered kinetic seeds that enforce successive {111} twinning, reproducibly generating eight-branched dendrites with global octahedral (cubic) symmetry even though each branch remains highly stacking-disordered. At later stages, latent-heat release drives a crossover to pure hexagonal ice while the pre-established macroscopic cubic symmetry is retained. Supporting evidence includes low-dose HRTEM stacking sequences, CBBDF morphology, 3D electron diffraction, SEND+ML phase mapping, coarse-grained MD snapshots, and continuum heat-transfer simulations.","tokens_in":19496,"tokens_out":973,"duration_ms":13333,"significance":"If the claimed kinetic pathway holds, the work supplies a concrete, multi-scale mechanism that converts local stacking disorder into persistent macroscopic cubic symmetry under strong undercooling. This directly addresses a long-standing morphological puzzle in ice and offers a transferable framework for other polytypic systems (SiC, diamond, GaN, etc.) where nonequilibrium growth stabilizes mixed cubic–hexagonal sequences. Strengths include genuine molecular-resolution imaging of a previously inaccessible process, quantitative cubicity evolution, crystallographic confirmation of twinning angles via 3D-ED, and consistent multi-modal support (experiment + MD + continuum modeling). The platform itself is a technical advance for cryo-TEM of liquid-phase nucleation.","major_comments":[{"comment":"Section “Cryo membrane liquid cell and homogeneous nucleation” and Fig. S2: the claim of bulk-like homogeneous nucleation rests on freezing temperature, site randomness across cycles, and volume scaling of nucleation probability. However, EFTEM (Fig. S1) shows the droplets are only 30–150 nm thick and fully sandwiched by amorphous carbon, rendering growth quasi-two-dimensional. All molecular-resolution data (HRTEM cubicity maps, germ imaging) are post-solidification. The manuscript must more explicitly test or discuss whether membrane–ice interactions or the forced coplanarity of the eight {111} directions bias the observed cubicity gradient (0.35\to0.62) and germ-mediated branching relative to unconfined bulk water; otherwise the claimed general kinetic pathway cannot be confidently transferred beyond the confined geometry.","section":null},{"comment":"Fig. 2 and Supplementary Fig. S7: cubicity values (0.35 core, 0.53 secondary, 0.62 tertiary) and cubic-segment thickness distributions are reported without uncertainties, number of independent stacking sequences analyzed, or statistical tests. Because the progressive thickening of cubic layers is presented as the central kinetic mechanism, quantitative error bars and sample sizes are required to establish that the radial increase is significant and reproducible.","section":null}],"minor_comments":[{"comment":"Methods (Low-dose HRTEM): the assertion that the beam was blanked during cooling is important; a quantitative upper bound on residual dose (or a control experiment) would strengthen the claim that crystallization is beam-free.","section":null},{"comment":"Fig. 3E and S10: MD snapshots are qualitative; a brief statement of how the coarse-grained Bond Order Potential was validated against known ice stacking energetics would help readers assess transferability.","section":null},{"comment":"Fig. 1G caption and main text: the classification of primary/secondary/tertiary dendrites is clear in the schematic but could be labeled more explicitly on the CBBDF image itself for non-specialist readers.","section":null},{"comment":"References: a few recent experimental and simulation papers on stacking-disordered ice under confinement or in thin films could be added for completeness, but this is not essential.","section":null}],"recommendation":"minor_revision","confidential_remarks":"The multi-modal data set is unusually strong for a cryo-TEM nucleation study and the central morphological observation (robust eight-branched cubic dendrites from Isd) is compelling. The confinement caveat is real but addressable by discussion and perhaps a short additional control; it does not invalidate the observations themselves. I would not require bulk free-droplet experiments for acceptance. Fit for a high-impact materials/condensed-matter journal is excellent."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This is the first molecular-resolution, in-situ view of homogeneous ice nucleation from liquid water at -40 °C. The core advance is concrete: stacking-disordered ice starts with isolated cubic monolayers (cubicity ~0.35), those layers thicken radially, and defect-free nanoscale Ic germs appear on basal planes and seed successive {111} twinning into eight-branched dendrites that keep global cubic symmetry even after late-stage conversion to Ih. That pathway was missing from the diffraction and simulation literature they cite.\n\nThey do the hard experimental work carefully. Low-dose HRTEM stacking sequences, CBBDF morphology, 3D-ED twinning angles, SEND+ML phase maps, MD snapshots with a published coarse-grained potential, and continuum heat-transfer runs all line up. Freezing temperature, site randomness across melt–freeze cycles, and volume (not area) scaling of nucleation probability (Fig. S2) are the right checks for homogeneous nucleation. Beam blanking during cooling is stated explicitly. The multi-modal consistency is the paper’s real strength.\n\nThe soft spot is real but secondary. Droplets are only 30–150 nm thick and fully membrane-confined, so growth is quasi-2D and all molecular data are post-solidification. The radial cubicity rise and germ locations could in principle feel membrane or thin-film bias. The authors’ bulk-like temperature and volume-scaling arguments mitigate this, but they do not eliminate it; a short discussion of possible confinement effects would help. Cubicity numbers also lack uncertainties. Neither issue collapses the central claim.\n\nThis is for ice physicists, atmospheric chemists, and anyone working on polytypic or stacking-disordered materials under kinetic control. It deserves a serious referee. I would accept after minor clarification of dose and confinement.","headline":"Solid first molecular-resolution look at deep-supercooling ice growth that ties cubic-layer thickening to octahedral dendrites; confinement is a real but secondary caveat.","tokens_in":20157,"tokens_out":454,"would_cite":true,"duration_ms":5802,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Deeply supercooled water freezes into stacking-disordered ice whose cubic layers thicken into germs that seed eight-branched dendrites with global cubic symmetry.","keywords":["Ice crystallization","Stacking disorder","Kinetic growth","Cubic-hexagonal polytypism","In situ cryogenic TEM","Supercooled water","Dendritic morphology"],"falsifier":"Direct observation that nucleation sites repeatedly coincide with membrane-water contact lines, or measurement of cubicity that fails to rise with distance from the core in free (unconfined) supercooled microdroplets of comparable size.","tokens_in":20099,"feed_emoji":"❄️","tokens_out":935,"duration_ms":18417,"temperature":0.7,"pith_summary":"The paper shows how ice forms from water cooled to -40 °C by watching the process live at molecular resolution inside thin membrane-sealed droplets. Nucleation first creates stacking-disordered ice that contains cubic layers only as isolated monolayers. Those cubic layers gradually thicken; once they reach a critical thickness, defect-free nanoscale cubic germs appear on the basal planes and act as registered seeds that force successive twinning. The result is a reproducible eight-branched dendrite whose overall shape has cubic (octahedral) symmetry even though every branch remains highly disordered. Later release of latent heat switches outer growth to pure hexagonal ice while the pre-set cubic outline is retained. The finding matters because it supplies a concrete kinetic route by which atomic-scale disorder is converted into macroscopic order under strong driving force, a process common to many materials made far from equilibrium.","feed_headline":"Disordered ice thickens into cubic germs that seed octahedral dendrites","feed_subtitle":"Kinetic seeds convert atomic stacking faults into global cubic symmetry that survives later hexagonal conversion","key_machinery":"Gradual thickening of cubic ice layers inside the stacking-disordered lattice; once a critical thickness is reached the layers spawn nanoscale defect-free cubic germs that act as facet-registered kinetic seeds for twinning and hierarchical branching.","core_discovery":"Homogeneous nucleation of deeply supercooled water produces stacking-disordered ice of low cubicity in which cubic ice exists only as isolated monolayers. The central kinetic step is the progressive thickening of these cubic layers, which nucleates nanoscale, defect-free cubic ice germs on basal planes. The germs function as facet-registered seeds that enforce cubic twinning and sequentially multiply growth branches, reproducibly generating eight-branched dendrites with global cubic symmetry while each branch stays stacking-disordered. Latent-heat release later drives a crossover to hexagonal ice that preserves the established cubic morphology.","pith_inferences":["The extreme thinness and two-dimensional confinement of the membrane cell may bias basal-plane growth relative to bulk three-dimensional freezing, so free droplets could show fewer or differently angled branches.","If cubic-germ size sets the branching hierarchy, additives that change stacking-fault energy should switch the final number of arms or suppress cubic symmetry entirely.","Retention of cubic outline after conversion to hexagonal ice offers a practical route to non-hexagonal morphologies of the stable ice phase for atmospheric or materials applications.","The observed low initial cubicity contradicts several molecular-dynamics predictions of cubic-rich nuclei, implying that continuum-scale kinetics overtake atomistic preferences after the first few layers."],"forward_implications":["Kinetic pathway engineering can deliberately select macroscopic crystal symmetries by controlling stacking-layer thickness during rapid solidification.","The same thickening-to-germ sequence is expected to operate in other polytypic systems (SiC, diamond, GaN, certain alloys) grown far from equilibrium.","Global crystal habit can preserve a kinetic memory of nucleation even after thermodynamic annealing converts the local structure to the stable phase.","Membrane-encapsulated cryo-TEM now enables molecular-resolution tracking of liquid-to-solid transitions for water and potentially other solvents."],"fun_headline_variants":["Cubic monolayers thicken into germs that seed eight-branched dendrites","Stacking-disordered ice births cubic germs enforcing octahedral symmetry","Thickening cubic layers nucleate germs that multiply ice growth branches","From isolated cubic monolayers to faceted germs seeding cubic dendrites","Kinetic cubic germs turn stacking disorder into retained octahedral form"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"That freezing at -40 °C inside the carbon-membrane liquid cell is true homogeneous nucleation free of surface or beam influence, justified only by the bulk freezing temperature, random nucleation sites across cycles, and volume (not surface) scaling of nucleation probability.","fun_headline_variants_meta":{"raw":{"variants":["Cubic monolayers thicken into germs that seed eight-branched dendrites","Stacking-disordered ice births cubic germs enforcing octahedral symmetry","Thickening cubic layers nucleate germs that multiply ice growth branches","From isolated cubic monolayers to faceted germs seeding cubic dendrites","Kinetic cubic germs turn stacking disorder into retained octahedral form"]},"model":"grok-4.5","effort":"low","cost_usd":0.00458,"raw_usage":{"total_tokens":1380,"prompt_tokens":829,"num_sources_used":0,"completion_tokens":78,"cost_in_usd_ticks":45800000,"prompt_tokens_details":{"text_tokens":829,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":473,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":829,"tokens_out":78,"duration_ms":3895,"temperature":1.0,"reasoning_tokens":473,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-12T02:17:13.362855+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Direct observation that nucleation sites repeatedly coincide with membrane-water contact lines, or measurement of cubicity that fails to rise with distance from the core in free (unconfined) supercooled microdroplets of comparable size.","supporting_citations":[],"review_version":1}