Unraveling atomic-resolution valence electron energy-loss spectroscopic imaging in a single-crystal CaNb2O6

Guo-Jiun Shu; Hsin-An Chen; I-Ching Lin; Sz-Chian Liou; Vladimir P. Oleshko; Xiang-Lin Huang; Yin-Ping Lan

REVIEW 2 major objections 6 minor 2 references

Unraveling atomic-resolution valence electron energy-loss spectroscopic imaging in a single-crystal CaNb2O6

T0 review · 2 major / 6 minor · reviewed 2026-07-13 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Atomic-resolution valence EELS maps plasmons and plexcitons in CaNb2O6, with contrast that flips or follows Z-contrast depending on collection geometry.

desk verdict Solid multi-geometry demonstration that atomic-resolution VEELS maps are achievable on a complex oxide, with clear on-axis contrast reversal and off-axis Z-contrast recovery; the elastic-contrast interpretation is supported by controls even without full inelastic multislice. read the letter →

arxiv 2607.09466 v1 pith:QKMHXUNG submitted 2026-07-10 cond-mat.mtrl-sci

Sz-Chian Liou , Xiang-Lin Huang , Vladimir P. Oleshko , I-Ching Lin , Yin-Ping Lan , Hsin-An Chen , Guo-Jiun Shu This is my paper · ORCID

classification cond-mat.mtrl-sci

keywords atomic-resolutionVEELSplexcitonvolumeplasmonCaNb2O6Cs-STEM-EELSelasticcontrastpreservationweak-beamEELShybrid-pixeldetector

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Atomic-resolution imaging of valence electronic excitations has long been blocked by the delocalized nature of inelastic scattering at low energy loss. This paper shows that modern Cs-corrected STEM paired with a hybrid-pixel direct detector can still produce energy- and atom-resolved maps of volume plasmons, plexcitons, and near-gap interband transitions in single-crystal CaNb2O6. On-axis collection yields zero-loss and valence images whose contrast is reversed relative to HAADF (atomic columns dark), which the authors attribute to preserved elastic contrast; increasing the collection angle or switching to a weak-beam (off-axis) geometry recovers ordinary Z-contrast. The same data set also shows that the 3.2–3.5 eV interband intensity is localized on the [NbO6] octahedra. The practical claim is that low-loss STEM-EELS can now be used to visualize bonding and electronic structure at individual atomic columns, interstitial sites, and defects in complex oxides.

What carries the argument

On-axis versus weak-beam (off-axis) collection geometry, combined with systematic variation of the EELS collection semi-angle relative to the probe convergence angle, which controls whether elastic contrast is preserved or suppressed in the low-loss images.

What would settle it

A full multislice inelastic simulation of the same spectrum-imaging data set that predicts the observed on-axis contrast reversal and its disappearance under weak-beam conditions; if the simulated inelastic maps lack atomic contrast while experiment shows it, the elastic-preservation account fails.

Watch

Extended reading notes

Core claim

In CaNb2O6, Cs-STEM-EELS with hybrid-pixel detection produces atom-resolved maps of volume plasmons (~6 eV and ~15 eV) and plexcitons (~7.3 eV). On-axis images reverse HAADF contrast; weak-beam images restore Z-contrast. The 3.2–3.5 eV interband maps are dominated by the [NbO6] octahedra.

Load-bearing premise

The atomic contrast seen in on-axis zero-loss and valence maps is assumed to come mainly from preserved elastic scattering rather than residual inelastic delocalization or zero-loss-peak tails.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

This is a careful experimental paper that actually delivers atomic-column VEELS maps on CaNb2O6 under both on-axis and weak-beam geometries. That is the new result. Prior atomic VEELS/plasmon work exists for Si, graphene (mostly EFTEM), and a few oxide heterostructures; what they add is a systematic collection-angle and on/off-axis study on a technologically relevant complex oxide, plus site-specific intensity at the 3.2–3.5 eV interband feature tied to [NbO6].

They do the work properly. Aloof spectra, Kramers–Kronig dielectric functions, DFT PDOS, multislice HAADF/BF/CBED simulations, β varied relative to α, IBF elastic correction, and weak-beam Kikuchi-band collection are all there. Spectral assignments (VPs near 6 and 15 eV, the ~7.3 eV feature as SEP/plexciton, CR at interstitial sites) rest on standard criteria and are cross-checked. The contrast systematics are the strongest part: on-axis ZLP/VEELS reverse relative to HAADF (HABF-like), larger β cleans the interference, IBF leaves residual delocalized VP intensity, and off-axis recovers Z-contrast while still giving usable VEELS. That multi-geometry package makes the experimental claim robust.

The soft spot the reader flagged is real but not load-bearing. They lean on elastic-contrast preservation as the origin of the atomic contrast without quantitative inelastic multislice of the full SI datasets. That would tighten the mechanism. It is not required for the demonstration itself, because the experimental controls already isolate the same conclusion. Data are not public (cost/feasibility note), and the instrument is specialized, so independent reproduction will be limited. The “plexciton” label for the bulk 7.3 eV feature is a bit of an extension of the surface-exciton language; minor. Free parameters (bandgap intercept, integration windows) are ordinary for the field.

This is for electron microscopists and oxide electronic-structure people who care about low-loss mapping beyond core-loss elemental maps. It deserves a serious referee. I would accept it for peer review and would cite the geometry systematics and the [NbO6] interband maps if I were working in the area.

Referee Report

2 major / 6 minor

Summary. The manuscript reports atomic-resolution valence EELS (VEELS) imaging of single-crystal CaNb2O6 using Cs-corrected STEM with a hybrid-pixel direct electron detector. Above the ~3.8 eV bandgap, volume plasmons (~6 eV, ~15 eV) and a ~7.3 eV feature assigned as a plexciton (mixed plasmon–exciton) are identified via aloof-beam spectra, Kramers–Krönig dielectric functions, and DFT PDOS. On-axis ZLP and VEELS maps reverse contrast relative to HAADF (HABF-like), while weak-beam (off-axis) collection recovers Z-contrast; systematic variation of collection angle β relative to convergence α and IBF elastic correction are used to address elastic-contrast preservation. Interband maps at 3.2–3.5 eV are linked to [NbO6] octahedra, with a possible Čerenkov contribution at interstitial sites. The central claim is that Cs-STEM-EELS can deliver energy- and atom-resolved low-loss contrast for structure, bonding, and electronic properties of complex oxides.

Significance. Atomic-resolution VEELS has remained difficult because of inelastic delocalization and ZLP tails; prior STEM results on graphene were largely delocalized, while EFTEM and limited STEM work on Si and oxide heterostructures showed mixed outcomes. This work supplies a multi-geometry experimental demonstration (on-axis vs weak-beam, β vs α series, IBF correction) on a technologically relevant mixed-cation oxide, with independent spectral assignment from dielectric analysis and DFT. The hybrid-pixel detector and CEFID spectrometer enable usable S/N at atomic columns. If the contrast mechanisms hold, the approach is a practical route to site-specific low-loss mapping of bonding and defects in complex crystals, complementary to core-loss and phonon mapping. Strengths include thorough experimental controls, multislice HAADF/BF/CBED simulations, and transparent reporting of intensity trends and acquisition parameters.

major comments (2)

The ~7.3 eV feature is assigned as a plexciton (and surface exciton polariton) on the basis of ε2 > ε1 > 0, surface-like energy shifts in aloof spectra, and literature on plasmon–exciton hybrids. For bulk CaNb2O6 this assignment is nonstandard; the manuscript should either strengthen it with a quantitative comparison (e.g., oscillator strength, dispersion, or comparison to optical data) or soften the language to “mixed interband/plasmon character” so that the central imaging claim does not rest on a contested label.
Elastic-contrast preservation is invoked as the dominant origin of atomic contrast in on-axis ZLP/VEELS (Figs. 3–5 and discussion in §III). The experimental controls (β series, IBF correction, weak-beam reversal) are strong, but the paper lacks quantitative inelastic multislice simulations of the full SI datasets that would separate residual delocalization, ZLP tails, and elastic interference. Adding such simulations (or a clear statement of why they are beyond scope) would make the mechanistic interpretation load-bearing rather than qualitative.

minor comments (6)

Energy-integration windows for SI maps (e.g., 7 ± 1 eV, 15 ± 1 eV, 3.25 ± 0.25 eV) should be stated once in Methods and justified relative to peak widths and energy resolution (0.7 eV).
Fig. 4: experimental vs simulated BF images differ in O/interstitial contrast at β = 20 mrad; a short note on residual aberrations or TDS treatment would help readers assess the comparison.
The optical bandgap linear-fit procedure is cited to prior work; a one-sentence description of the intercept method and uncertainty (already given as ±0.05 eV) would make the manuscript self-contained.
Typographical consistency: “plexcitons” vs “plexciton excitations”; “weak-beam (off-axis)” is used repeatedly—define once and use a single term thereafter.
Table 1 (refinement) is referenced but the numerical content is not fully reproduced in the provided text; ensure lattice parameters, Wyckoff sites, and DW factors are complete and match the HAADF simulations.
Data-availability statement is restrictive; if raw SI cubes or reduced spectra can be shared on request, a repository DOI would strengthen reproducibility.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: experimental VEELS imaging with independent dielectric/DFT assignments and multi-geometry controls; self-citations are methodological only.

full rationale

The paper is a predominantly experimental STEM-EELS study of CaNb2O6. Spectral features (VPs near 6/15 eV, plexciton near 7.3 eV, interband 3.2–3.5 eV, Eg ≈ 3.8 eV) are assigned via standard Kramers–Krönig analysis of measured VEELS (ε1 = 0 criterion, Im(−1/ε) maxima) plus independent DFT-PDOS (VASP/GGA-PBESOL), not by fitting parameters that are then re-labeled as predictions. Atomic-resolution contrast claims rest on direct SI datasets under controlled on-axis β/α ratios, IBF elastic correction, and weak-beam/off-axis geometries (Figs. 3–6), which reverse or recover Z-contrast as measured; these are empirical observations, not derivations that reduce to inputs by construction. Self-citations ([8], [15], [39]) supply prior methodological details or related materials work by overlapping authors but are not load-bearing uniqueness theorems or ansätze that force the present results. No self-definitional loops, fitted-input-as-prediction steps, or renaming of known results appear. Minor self-citation of methods yields the residual score of 1; the central experimental claims remain independently supported by the new data.

Assumptions & free parameters 2 free parameters · 5 assumptions · 1 invented entities

The central claims rest on standard EELS dielectric theory, DFT electronic-structure assumptions, multislice elastic scattering, and the experimental geometry of the CEFID spectrometer. No free parameters are fitted to force the atomic-resolution conclusion; the few numerical thresholds (bandgap linear fit, energy windows) are conventional. Invented entities are limited to the interpretive label “plexciton” for an observed mixed feature already known in the plasmonics literature.

free parameters (2)

optical bandgap linear-fit intercept = 3.8 eV
Eg ≈ 3.8 ± 0.05 eV obtained by linear fitting of the onset; the precise intercept depends on the chosen energy window and background model.
energy-integration windows for SI maps = ±1 eV or ±0.25 eV
Windows such as 0±1 eV, 7±1 eV, 15±1 eV, 3.25±0.25 eV are chosen by hand to isolate spectral features; small shifts alter map contrast.

assumptions (5)

domain assumption Volume-plasmon condition ε1=0 with positive slope and small ε2; surface-exciton-polariton criterion ε2>ε1>0
Standard macroscopic dielectric interpretation of EELS (Egerton, Raether) used throughout §III to assign 6 eV, 7.3 eV and 15 eV features.
domain assumption Inelastic delocalization length L50 ≈ 0.5 λ / θE^{3/4}
Classical formula invoked to argue that atomic resolution should be impossible yet is observed, motivating the elastic-contrast discussion.
domain assumption GGA-PBEsol + PAW DFT yields reliable partial densities of states for assignment of O-2p/Nb-4d hybridizations
Used in §II–III and Fig. 2(d) to link spectral features to [NbO6] octahedra; standard but approximate for band gaps and excitonic effects.
domain assumption Multislice elastic scattering (xHREM) with tabulated Debye–Waller factors adequately simulates HAADF/BF/CBED contrast
Simulations in Figs. 1, 4, 6 underpin the claim that observed contrast matches elastic channeling.
domain assumption Weak-beam (off-axis) collection of high-angle Kikuchi bands localizes the signal to TDS on heavy-atom columns
Taken from prior phonon-mapping literature and used to explain recovery of Z-contrast in Fig. 6.

invented entities (1)

plexciton assignment for the ~7.3 eV feature in bulk CaNb2O6
purpose: To interpret the mixed surface/volume character of the 5–7 eV spectral region as strongly correlated plasmon–exciton hybrid modes.
The term is borrowed from nanoparticle plasmonics; independent evidence inside the paper is the surface-excitation shift and ε2>ε1>0 criterion, but no external spectroscopic confirmation (e.g., optical or momentum-resolved) is supplied for this crystal.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Unraveling atomic-resolution valence electron energy-loss spectroscopic imaging in a single-crystal CaNb2O6." pith.science (2026). https://pith.science/paper/QKMHXUNG

@misc{pith2026260709466,
  author       = {Pith},
  title        = {Pith review of: Unraveling atomic-resolution valence electron energy-loss spectroscopic imaging in a single-crystal CaNb2O6},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QKMHXUNG}},
  note         = {Machine review of arXiv:2607.09466}
}

read the original abstract

Despite advancements in electron optics and spectrometer design over the past twenty years, atomic-resolution valence-electron energy-loss spectroscopy imaging remains challenging due to the delocalization of inelastic electron scattering. In this study, we used an energy-filtered spectrometer equipped with a hybrid-pixel direct electron detector and spherical aberration-corrected scanning transmission electron microscopy to analyze many-electron excitations and interband transitions in a single-crystal calcium niobate, CaNb2O6, with spatial resolution ranging from the nanometers to the atomic scale. In the low-loss region above the bandgap at about 3.8 eV, we observed volume plasmons, around 6 eV and 15 eV energy loss, as well as a mix of strongly correlated plasmons and excitons, known as plexcitons, at approximately 7.3 eV energy loss. Additionally, we employed an on-axis EELS setup for atomic-resolution zero-loss peak (ZLP) imaging and visualized energy- and atom-resolved images of plexcitons and VPs, which showed contrast reversal relative to high-angle annular dark-field images. To investigate elastic contrast preservation, we also analyzed the effect of the collection angle and minimized its influence to produce delocalized VP images. In fact, the ZLP and VEELS images obtained using the weak-beam setup demonstrate that, in both cases, the contrast resembles Z-contrast. Moreover, we found that [NbO6] octahedra directly contributed to the lateral maps of interband transitions in the range from 3.2 eV to 3.5 eV energy loss. These findings demonstrate that Cs-STEM-EELS, which examines atomic-scale contrast associated with low-energy losses, can be a powerful tool for visualizing the structure, bonding, and electronic properties of complex crystalline nanostructures, including individual atomic sites, interstitial sites, and point defects.

Figures

Figures reproduced from arXiv: 2607.09466 by the authors.

**Figure 5.** FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p021_5.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

2 extracted references

[1]

F. S. Hage, D. M. Kepaptsoglou, Q. M. Ramasse, and L. J. Allen, Phonon spectroscopy at atomic resolution, Phys. Rev. Lett. 122, 016103 (2019). [13] Z. L. Wang, Lattice imaging using plasmon energy-loss electrons in an energy-filtered transmission electron microscope, Ultramicroscopy 67, 105-111 (1997). [14] C. B. Boothroyd, Why don’t high-resolution simul...

2019
[2]

Berechnung der Energieverluste schneller Elektronen in dünnen Schichten mit Retardierung,

X. Fan, J. Yi, B. Deng, C. Zhou, Z. Zhang, J. Yu, W. Li, C. Li, G. Wu, X. Zhou, T. Sun, Y. Zhu, J. Zhou, J. Xia, Z. Wang, K. Lai, Z. Peng, D. Li, A. Pan, Y. Zhou, 2D edge-seeded heteroepitaxy of ultrathin high-κ dielectric CaNb2O6 for 2D field-effect transistors, Nat. Commun. 16, 2585 (2025). [28] Y. Liang, L. Wang, S. Wu, J. Wu, J. Zhu, J. Qin, X. Fan, Z...

2025

Pith tools

Reviewed July 13, 2026 · model on record in the stance chip above.

[1] [1]

F. S. Hage, D. M. Kepaptsoglou, Q. M. Ramasse, and L. J. Allen, Phonon spectroscopy at atomic resolution, Phys. Rev. Lett. 122, 016103 (2019). [13] Z. L. Wang, Lattice imaging using plasmon energy-loss electrons in an energy-filtered transmission electron microscope, Ultramicroscopy 67, 105-111 (1997). [14] C. B. Boothroyd, Why don’t high-resolution simul...

2019

[2] [2]

Berechnung der Energieverluste schneller Elektronen in dünnen Schichten mit Retardierung,

X. Fan, J. Yi, B. Deng, C. Zhou, Z. Zhang, J. Yu, W. Li, C. Li, G. Wu, X. Zhou, T. Sun, Y. Zhu, J. Zhou, J. Xia, Z. Wang, K. Lai, Z. Peng, D. Li, A. Pan, Y. Zhou, 2D edge-seeded heteroepitaxy of ultrathin high-κ dielectric CaNb2O6 for 2D field-effect transistors, Nat. Commun. 16, 2585 (2025). [28] Y. Liang, L. Wang, S. Wu, J. Wu, J. Zhu, J. Qin, X. Fan, Z...

2025