{"id":"20d01a48-2f9f-4eca-a28e-72b1b3d8406c","arxiv_id":"1908.03257","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Yttrium preferentially segregates to the highest-expansion atomic site (E3) inside YSZ grain boundary dislocation cores, as shown by combined atomic-resolution strain mapping and EELS.","lead":"Atomic-scale microscopy shows that yttrium atoms segregate to the single most stretched atomic column in the dislocation cores of a yttria-stabilized zirconia grain boundary. This pinpoints the structural origin of grain-boundary ionic blocking, a key obstacle in solid oxide fuel cells and batteries.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The measured E3 Y 'doubling' depends on on-axis EELS channeling that is acknowledged but not controlled; local strain/occupancy changes at the core can modulate Y/Zr L-edge ratios and mimic preferential segregation. A multislice EELS simulation with the DFT core model would settle it.","rationale":"The reader's conditional verdict already identifies both 2D strain and on-axis channeling as weak assumptions. I prioritize the channeling issue because it attacks the compositional measurement itself: if the E3 map is a beam-propagation artifact, the paper's central correlation disappears even if the 3D strain ranking were exactly as claimed. The paper's own text concedes the on-axis condition is 'not optimal' and that prior on/off-axis comparisons showed 'a difference of a few percent,' but the claimed E3 enhancement is roughly 100% relative to bulk, and it occurs at the most distorted site where channeling changes should be largest. The authors provide no channeling simulation, no exit-wave or multislice analysis, and no independent column-resolved chemical measurement. The DFT calculation cited for confirmation comes from the authors' prior work and is used to interpret, not to test, the measured site. The other concerns—single specimen, small statistics, 2D strain projection—are real but already reflected in a CONDITIONAL verdict; the channeling issue is the one that would most directly invalidate the strongest claim. The proposed multislice test is standard, feasible, and decisive because the atomic model already exists. If the simulation validates the interpretation, conditional acceptance stands; if not, the central claim is unverified. Thus the reader's CONDITIONAL verdict should remain unchanged, with the channeling simulation as the key condition.","tokens_in":7333,"tokens_out":6065,"duration_ms":70309,"concrete_test":"Run multislice frozen-phonon EELS simulations (e.g., abTEM or Prismatic) using the DFT-relaxed 33° [001] YSZ bicrystal model from ref. 16 (Fig. 4d) at 200 kV, with the same probe convergence, EELS collection angle, 1 eV/channel dispersion, and an equivalent quantification pipeline (power-law background, PCA, relative concentrations). Simulate three Y distributions: (i) no segregation (uniform bulk 9 mol%), (ii) Y preferentially substituted at E3 to match the claimed ~2x bulk enrichment, and (iii) uniform Y enhancement over all core columns. Compare the simulated apparent Y concentration at E3, E1, E2, C1, and bulk. If configuration (i) or (iii) produces an apparent E3 'doubling' relative to bulk, the experimental site-specific segregation claim is not established; if only configuration (ii) reproduces the observed E3 contrast, the channeling concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing condition for the central claim is that the E3 Y signal reflects a real increase in Y occupancy, not an on-axis channeling artifact. The paper states in §3.3 that 'it is necessary to acquire the EEL spectrum images on-axis conditions,' while admitting 'our quantitative values can be slightly affected by the not optimal conditions but the overall trends found should be robust.' That concession is insufficient for the quantitative headline: the plotted E3 value is roughly twice the bulk Y concentration (5.7% to about 11%), far beyond the 'few percent' offset seen in the authors' previous on/off-axis comparison, and that comparison was for average compositions, not for a strongly distorted dislocation core. In the core, local strain and partial column occupancy change channeling of the 200 keV probe along the beam direction, so the Y/Zr L2,3 intensity ratio at the E3 pixel can be modulated by local structure even if the actual Y fraction is uniform. The 'pixel-by-pixel' quantification in Figure 3(c) only makes the arithmetic independent of neighbors; the underlying inelastic signal is still generated by a probe broadened by propagation through the strained core. Thus the central observation—Y at E3 doubling the bulk while E1/E2/C1 do not—could in principle be a channeling contrast variation rather than preferential segregation. A 3D strain reordering (the reader's first concern) would weaken the interpretive link to 'highest expansive strain,' but would leave the segregation observation intact; a channeling artifact would remove the central evidence entirely. This makes the EELS channeling assumption the more load-bearing issue.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports an atomic-resolution scanning transmission electron microscopy study of a 33° [001] tilt grain boundary in a 9 mol% Y2O3-doped ZrO2 bicrystal. The authors use peak pairs analysis of HAADF images to map the local mean dilatation and identify three expansive sites (E1, E2, E3) and one compressive site (C1) per dislocation core. Electron energy-loss spectrum imaging with pixel-by-pixel quantification shows that the yttrium relative concentration at the E3 site is approximately twice the bulk value, while the other core sites remain near bulk composition. Density-functional-theory calculations from the authors' earlier work (ref 16) are cited to support the assignment of E3 as the preferential Y segregation site. The paper concludes that Y segregates to the site of highest expansive strain, which is relevant to oxygen-vacancy localization and grain-boundary ionic transport in YSZ.","tokens_in":7628,"tokens_out":8969,"duration_ms":85614,"significance":"If the central claim holds, the paper provides a direct atomic-scale correlation between local strain and dopant segregation at a specific dislocation core, which is valuable for understanding space-charge and strain effects at YSZ grain boundaries. The experimental strengths are the site-resolved composition analysis with error bars over four dislocation cores, the simultaneous strain and composition mapping, and the explicit use of an independent DFT prediction as an interpretive anchor. The main limitations are the reliance on on-axis EELS quantification in a locally distorted region and the two-dimensional projection of the strain field; both are acknowledged in the text but are not controlled for quantitatively.","major_comments":[{"comment":"The central quantitative claim that the Y content at E3 doubles the bulk value is based on EELS quantification performed on-axis. The manuscript states that on-axis conditions are necessary to compare specific sites and admits that 'quantitative values can be slightly affected by the not optimal conditions but the overall trends found should be robust.' However, the reported effect is a doubling (5.7% to about 11%), far beyond the few-percent offset seen in the authors' previous on/off-axis comparison (ref 16), and that comparison was for average compositions, not for a single strongly distorted dislocation core. At the core, partial column occupancy and local strain can alter channeling of the 200 keV probe and the Y/Zr L-edge ionization cross-sections, so the pixel-by-pixel relative concentrations may not reflect true compositional changes. No multislice EELS simulation using the DFT core structure is provided to rule out a channeling artifact. Given that the identification of E3 as the segregation site is the main result, this is a load-bearing gap. Please provide a multislice simulation or an equivalent control, or explicitly restrict the conclusion to a qualitative trend.","section":"§3.3, ¶3; §3.4; Fig. 4(c)"},{"comment":"The strain values used to rank the expansive sites are derived from a two-dimensional projected HAADF image. The manuscript acknowledges that 'the current results reflect a two-dimensional measurement of a three-dimensional structure and the possible strain distortions along the beam direction are not considered.' The correlation between Y segregation and the 'position under higher expansive strain' therefore depends on the 2D ranking of E3. If out-of-plane strain were to reorder the expansive sequence, the claimed correlation would be weakened. Since the DFT calculations in ref 16 provide a 3D structural model, a comparison between the projected 2D strain and the 3D computed strain field at the same sites would substantiate the ranking. In the absence of such a comparison, the strain-segregation correlation is only as robust as the 2D approximation. Please add this comparison or strengthen the caveat accordingly.","section":"§3.2; Fig. 2(d) and Fig. 4(c)"}],"minor_comments":[{"comment":"The word 'demining' should be 'undermining' in the sentence 'ionic blocking processes at grain boundaries constitute one of the main obstacles demining their performance.'","section":"Introduction (p. 2)"},{"comment":"The title 'Porf. S. J. Pennycook' should be 'Prof. S. J. Pennycook.'","section":"Author line"},{"comment":"The phrase 'relative miss-orientation' should be 'relative misorientation.'","section":"§3.2"},{"comment":"The phrase 'the higher expansive strain' would be more precise as 'the highest expansive strain,' since the text refers to the maximum among the three expansive sites.","section":"§3.4"},{"comment":"The text in §3.2 describes the expansive and compressive sites as marked with 'white and black dots,' while the Figure 2 caption refers to 'red and blue arrows'; please reconcile the inconsistency.","section":"Figure 2"},{"comment":"The caption should state explicitly that the error bars are the standard deviation over the four dislocation cores in the spectrum image region, as mentioned in the text.","section":"Figure 4(c)"}],"recommendation":"major_revision","confidential_remarks":"The paper leans on the authors' prior publication (ref 16) for the DFT segregation energies. This is acceptable as an independent anchor, but the manuscript should make clear which parts of the conclusion are new experimental evidence and which are adopted from the prior DFT. The channeling concern is the main technical barrier; a multislice EELS simulation is standard practice for atomic-column quantification at defects and would substantially strengthen the claim. The 2D strain limitation is a secondary but real issue that could be addressed by comparing against the DFT strain field."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new result here is site-specific: the same group had shown Y segregation to YSZ GB dislocation cores, but this paper identifies which atomic column takes it up — the E3 site under the largest expansive strain, where Y content reads double bulk. That is a genuine advance, and the experimental backbone is decent. The strain and EELS maps are independently measured, the composition averages come from four cores with error bars, and the DFT from their earlier paper is used as corroboration, not fitted.\n\nI read the stress-test note as landing on the right spot. The paper is honest about its limitations, but the channeling worry is not minor. The text says the on-axis values may be 'slightly affected' and that the overall trends should be robust. Yet the E3 'doubling' is far beyond the few-percent offset they saw in their own earlier on/off-axis comparison, and that comparison was on average compositions, not on a strained dislocation core. In the core, local strain and partial column occupancy can change how the 200 keV probe propagates and how the Y/Zr L2,3 ratio is generated. A multislice EELS simulation with the DFT core model would settle it. Without that, the central quantitative claim carries a real alternative explanation.\n\nI do not share the reader's circularity concern. Quoting one's own prior DFT segregation energy is self-citation, not a fitted circular step; the measurement does not depend on it.\n\nSecondary cautions: one bicrystal, one boundary orientation, and a 2D strain measurement. The last point is explicitly stated in the paper, and if out-of-plane strain changes the ranking, the correlation with 'highest expansive strain' weakens but the basic segregation observation would survive. The conclusion calling this 'the main mechanism' of strain relaxation goes beyond the data; that phrasing should be tempered.\n\nThis paper deserves a serious referee. The site-specific experimental observation is likely to hold up, but the quantitative E3 value needs either an off-axis control or a channeling simulation, plus bulk error bands shown on the same graph. For a reading group on atomic-scale interfacial characterization, it is worth the time. I would cite it in a review of grain-boundary segregation in fluorite oxides.","headline":"Site-specific Y segregation at a single YSZ grain-boundary column is a real advance; the load-bearing caveat is on-axis EELS channeling, which needs a control or simulation before the 'doubling' claim is taken as quantitative.","tokens_in":8193,"tokens_out":1655,"would_cite":true,"duration_ms":17815,"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":"Yttrium segregation at a YSZ grain boundary is not spread across the interface: it concentrates at the single atomic column under the highest expansive strain, roughly doubling the bulk concentration at that site.","keywords":["yttria stabilized zirconia","grain boundary segregation","yttrium","dislocation core","strain mapping","electron energy loss spectroscopy","scanning transmission electron microscopy","density functional theory"],"falsifier":"Acquire an atomic-resolution tilt series of the same 33° [001] YSZ bicrystal and reconstruct the three-dimensional strain tensor; if the column with the largest true volumetric dilatation is not the column where Y content doubles, the strain–segregation correlation fails. A channeling-corrected EELS simulation of the same boundary could also settle the matter: if it shows the apparent Y enrichment at E3 is an artifact of electron-beam propagation, the site assignment is refuted.","tokens_in":7172,"feed_emoji":"🔬","tokens_out":6329,"duration_ms":58895,"temperature":0.7,"pith_summary":"This paper aims to show where, at atomic scale, yttrium dopants sit when they segregate to a grain boundary in yttria-stabilized zirconia (YSZ), and why they sit there. Using atomic-resolution electron microscopy, the authors measure both the local strain and the chemical composition of the same atomic columns in the boundary's dislocation cores. They find that yttrium does not enrich the whole boundary uniformly: its concentration roughly doubles only at the single atomic column under the highest expansive strain, while neighboring expansive and compressive columns stay near bulk composition. Density functional theory supports this site-specific picture, identifying yttrium substitution there as the strain-relief mechanism and as a way to localize oxygen vacancies. If correct, this means grain-boundary chemistry and ionic transport in YSZ should be understood column by column, not as an averaged boundary layer.","feed_headline":"Yttrium doubles at one strained grain-boundary column","feed_subtitle":"In YSZ, Y concentrates only at the column under +23.8% expansion, making boundary chemistry a one-column effect.","key_machinery":"The load-bearing machinery is the pairing of an atomic-column-resolved strain map with an atomic-column-resolved composition map on the same dislocation cores. Peak-pairs analysis of a high-angle annular dark-field image produces a mean-dilatation map ($d_{xy}=\\varepsilon_{xx}+\\varepsilon_{yy}$) that sorts the core columns into expansive sites E1, E2, E3 and compressive site C1, with E3 carrying the largest expansion (+23.8%). EELS quantification of the Y and Zr L2,3 edges at those same columns supplies the composition per site. Density functional theory closes the argument by giving the mechanism: a Y atom substituting Zr at E3 has a segregation energy of 2.9 eV and lowers the formation energy of oxygen vacancies, so the most expansive column becomes both the Y-rich and vacancy-rich site, and the strain is released there.","core_discovery":"On a 33° [001] tilt grain boundary in a 9 mol% yttria YSZ bicrystal, the boundary's dislocation cores contain three expansive atomic columns (mean dilatations +15.3%, +10.7%, +23.8%) and one compressive column (−2.7%). Site-by-site EELS quantification shows that the yttrium concentration at the most expansive column, E3, is roughly double the bulk Y content, while the other expansive columns and the compressive column stay near bulk values within error. Density functional theory calculations agree: Y substitutes Zr preferentially at E3 with a segregation energy of 2.9 eV, even though that site has the larger coordination number, and the substitution lowers the oxygen-vacancy formation energy there. The conclusion is that Y segregation at this boundary is not a uniform boundary enrichment but is localized to one atomic column, and that this localization is the strain-relaxation mechanism that also confines oxygen vacancies to the dislocation core.","pith_inferences":["Editorial inference: the same strain–composition correlation should be testable in other fluorite and perovskite ion conductors; if it holds, the site-selection rule for oversized dopants is simply to occupy the column with the maximum expansive dilatation.","Editorial inference: a three-dimensional strain reconstruction, such as atomic-resolution electron tomography on the same boundary, would decide whether E3 remains the volume-maximizing column when out-of-plane distortions are included; if it does, the strain argument becomes quantitative rather than projected.","Editorial inference: because the Y-rich column also anchors oxygen vacancies, locally manipulating strain through epitaxy, pressure, or boundary geometry might shift the segregation column and thereby tune the grain-boundary ionic resistance.","Editorial inference: the reported site-resolved doubling suggests that continuum models using a single boundary-excess value lose information; a discrete-site segregation-energy landscape may be needed to predict transport blocking."],"forward_implications":["The grain-boundary dopant excess in YSZ is concentrated at one atomic column per dislocation core, so boundary chemistry is site-specific rather than uniform.","Yttrium segregation to E3 explains how strain is relieved at the core and why oxygen vacancies are confined there, linking atomic structure to ionic-conductivity blocking.","Dopants with larger ionic radius than Zr should be expected to populate the most expansive boundary columns, giving a predictive rule for other stabilizing oxides.","The near-bulk composition at E1, E2, and C1 means that only the maximum-expansion site matters for the dominant segregation, simplifying models of boundary charge and transport.","The DFT segregation energy of 2.9 eV at E3 provides a quantitative benchmark for atomistic simulations of YSZ grain boundaries."],"supporting_citations":[{"why":"Prior study that established significant Y segregation to the dislocation cores and computed the 2.9 eV segregation energy used here.","marker":"[16]"},{"why":"Supplies the peak-pairs analysis method that produces the strain maps identifying the expansive and compressive columns.","marker":"[17]"},{"why":"Establishes that dislocation cores generate inhomogeneous compressive and expansive strain fields, the framework for interpreting the strain map.","marker":"[24]"},{"why":"States the general rule that larger ions occupy expansive strain sites, motivating why Y should favor the most expansive column.","marker":"[25]"},{"why":"Accounts for probe spreading and ionization delocalization, justifying the choice of Y and Zr L2,3 edges for site-resolved analysis.","marker":"[30,31]"},{"why":"Earlier observations of compositional changes with dislocation-core periodicity at grain boundaries support the measured Y, Zr, and O variations.","marker":"[32,33]"},{"why":"Provides the mechanism by which Y lowers oxygen-vacancy formation energy and releases strain at the core.","marker":"[42]"}],"fun_headline_variants":["Single atomic column holds all Yttrium segregation in YSZ","Yttrium's one-column rule at YSZ grain boundaries","Strain-relief: Yttrium piles into one YSZ dislocation core","Yttrium doubles only at the most expanded YSZ boundary column"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The ranking of E3 as the most expansive site comes from a two-dimensional strain map of a three-dimensional structure, and the site-to-site Y trend rests on on-axis EELS quantification despite channeling; if either gives way, the claimed correlation may not hold.","fun_headline_variants_meta":{"raw":{"variants":["Single atomic column holds all Yttrium segregation in YSZ","Yttrium's one-column rule at YSZ grain boundaries","Strain-relief: Yttrium piles into one YSZ dislocation core","Yttrium doubles only at the most expanded YSZ boundary column"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000915,"raw_usage":{"total_tokens":3899,"prompt_tokens":889,"completion_tokens":3010,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":505,"completion_tokens_details":{"reasoning_tokens":2935}},"tokens_in":505,"tokens_out":3010,"duration_ms":24753,"temperature":1.0,"reasoning_tokens":2935,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:18:54.870360+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Acquire an atomic-resolution tilt series of the same 33° [001] YSZ bicrystal and reconstruct the three-dimensional strain tensor; if the column with the largest true volumetric dilatation is not the column where Y content doubles, the strain–segregation correlation fails. A channeling-corrected EELS simulation of the same boundary could also settle the matter: if it shows the apparent Y enrichment at E3 is an artifact of electron-beam propagation, the site assignment is refuted.","supporting_citations":[],"review_version":1}