Defect migration and phase transformations in 2D iron chloride inside bilayer graphene

Arkady V. Krasheninnikov; Chi-Chun Cheng; Haiming Sun; Hiroki Ago; Kazu Suenaga; Mahdi Ghorbani-Asl; Po-Wen Chiu; Qiunan Liu; Silvan Kretschmer; Yung-Chang Lin

REVIEW 2 major objections 4 minor 4 references

Defect migration and phase transformations in 2D iron chloride inside bilayer graphene

T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper argues that electron irradiation transforms FeCl3 into FeCl2 inside bilayer graphene by selectively sputtering chlorine, and reports a new Fe5Cl18 phase stabilized in the confined gap.

desk verdict A solid STEM-DFT study of defect migration and e-beam-driven FeCl3-to-FeCl2 transformation in bilayer graphene, but the new Fe5Cl18 phase claim rests on an unexplained 18% lattice rescaling and is not yet convincing. read the letter →

arxiv 2507.05665 v1 pith:TC45NPDN submitted 2025-07-08 cond-mat.mtrl-sci cond-mat.mes-hallphysics.chem-ph

Qiunan Liu , Haiming Sun , Yung-Chang Lin , Mahdi Ghorbani-Asl , Silvan Kretschmer , Chi-Chun Cheng , Po-Wen Chiu , Hiroki Ago

show 2 more authors

Arkady V. Krasheninnikov Kazu Suenaga

This is my paper

classification cond-mat.mtrl-scicond-mat.mes-hallphysics.chem-ph

keywords ironchlorideintercalationbilayergrapheneSTEMdefectmigrationphasetransformationelectron-beamirradiationFe5Cl18densityfunctionaltheory

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

This paper watches individual atoms move inside iron chloride layers sealed between two sheets of graphene, and it explains how electron irradiation converts one iron chloride phase into another. The central claim is that the transformation happens because chlorine atoms are knocked out of FeCl3 about three times more easily than out of FeCl2, so the beam pushes the material toward the more radiation-resistant FeCl2 structure. The authors identify three defect types that carry the motion: iron vacancies in FeCl2, iron interstitials in FeCl3, and iron adatoms that form dumbbells, and they calculate migration barriers of 1.4 and 1.8 electron volts for the first two. They also report a previously unknown crystalline phase, Fe5Cl18, that crystallizes in the confined space and then converts to FeCl2 under continued exposure. If the picture is right, the space between graphene sheets acts as a nanoscale reaction cell in which electron beams can pattern magnetic FeCl2 islands at selected locations.

What carries the argument

The load-bearing mechanism is the electron-beam displacement cross-section, calculated with a standard relativistic scattering formula, which gives how likely a 60 kV electron is to knock an atom out of its site. The paper shows chlorine in FeCl3 is three times more susceptible, about 210 barns, than chlorine in FeCl2, about 70 barns, making chlorine sputtering the selective switch that changes stoichiometry. The second piece is the set of nudged-elastic-band migration barriers, 1.4 eV for an iron vacancy exchanging with a nearest neighbor in FeCl2 and 1.8 eV for an iron interstitial filling a vacancy site in FeCl3, which give the defect-level vocabulary for how iron atoms actually move. Together these numbers turn images of individual moving atoms into an atomistic mechanism: chlorine leaves, iron refills, and the lattice densifies from FeCl3 to FeCl2.

What would settle it

An observation that would settle it: acquire a high signal-to-noise ADF image of the proposed Fe5Cl18 region and quantify how well simulated images of Fe5Cl18 and of a strained or beam-damaged FeCl3 supercell of the same stoichiometry match; if the FeCl3-based model matches equally well, the new phase is not established. A complementary calculation with the graphene layers explicitly included would show whether the 18 percent rescaling is physically reproducible without hand-scaling.

Watch

Extended reading notes

Core claim

The paper establishes that under a 60-kilovolt scanning transmission electron beam, FeCl3 intercalated in bilayer graphene transforms into FeCl2 not by adding iron but by losing chlorine and letting existing iron atoms rearrange. The driving force is an asymmetry in displacement cross-sections: at 60 kV a chlorine atom in FeCl3 has a displacement cross-section of about 210 barns versus 70 barns in FeCl2, so irradiation selectively depletes chlorine from FeCl3 and makes the denser FeCl2 lattice radiation-resistant. Single iron vacancies in FeCl2 migrate by exchanging with nearest-neighbor iron atoms over a 1.4 eV barrier; iron interstitials in FeCl3 move to fill nearby vacant iron sites over a 1.8 eV barrier; iron adatoms sit on iron sites in dumbbell configurations and only hop between iron sites, not into the lattice. The paper also reports a new crystalline Fe5Cl18 phase observed in the confined gap and shows it can crystallize from amorphous iron chloride under the beam and then evolve into FeCl2. The same encapsulation space allows electron-beam-induced growth of iron oxychloride at domain edges when oxygen is locally available.

Load-bearing premise

The new Fe5Cl18 phase stands on the assumption that the 18 percent stretching of the zero-temperature, graphene-free theoretical model to match the measured image is a real physical effect of confinement, pressure, or temperature, and not a sign that the image was misread as a new phase.

Editorial extensions

If this is right

Electron irradiation can write FeCl2 domains inside FeCl3 only in the scanned area, providing a way to pattern two-dimensional iron chloride islands at selected locations.
Because chlorine displacement is three times more likely in FeCl3 than in FeCl2, continued irradiation drives the material toward FeCl2, the more radiation-resistant phase, and not back toward FeCl3.
Fe vacancies in FeCl2 move by nearest-neighbor exchange but cannot grow into FeCl3 domains, so the FeCl2-to-FeCl3 direction is not observed under the beam.
The newly observed Fe5Cl18 phase crystallizes from amorphous iron chloride under electron irradiation and then converts to FeCl2 as chlorine is progressively displaced, showing a beam-driven reaction sequence inside the graphene gap.
The defect-modified electronic structure, with spin-polarized states near the Fermi level and enhanced charge transfer from graphene, means beam-written phase patterns could also be electronic or magnetic patterns.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the paper, this displacement-cross-section asymmetry suggests a selection rule for other encapsulated metal chlorides: electron irradiation will tend to convert a chloride into the lower-chlorine, more radiation-resistant endmember, regardless of which phase is thermodynamically favored in bulk.
The 18 percent model rescaling is an implicit warning that lattice parameters computed for free-standing, zero-temperature crystals can differ substantially inside a graphene sandwich; a follow-up calculation with the graphene layers explicitly included would test whether pressure and charge transfer account for the full mismatch.
Because the migration barriers are known, the electron beam could in principle steer single iron vacancies or interstitials along designed tracks, turning the graphene-sealed layer into an atom-scale writing surface.

Share X Bluesky LinkedIn Reddit HN

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

The paper delivers something genuinely useful: a direct atomic-scale movie of how Fe vacancies, interstitials, and adatoms move in 2D iron chlorides, and a coherent mechanism for how the electron beam converts FeCl3 to FeCl2. The DFT barriers (1.4 eV for vacancy migration, 1.8 eV for interstitial migration) are plausible and compare naturally with the observations, and the displacement cross-section argument for chlorine sputtering is a nice piece of quantitative reasoning. The patterning of FeCl2 domains inside FeCl3 by beam irradiation is a believable, well-supported result. For these parts, the paper deserves a serious referee.

The soft spot is the claimed Fe5Cl18 phase. The structural model is calculated without graphene, at zero temperature, and then scaled up by 18% to match the experimental ADF image. That is a large correction, and the physical justifications offered (anisotropic pressure, charge transfer, finite temperature) are hand-waving rather than demonstrated. No calculation shows that graphene encapsulation or charge transfer actually produces such an expansion, and no stability check (phonons, MD, or relaxation with the graphene sheets included) is presented. The supporting EELS ratios have no error bars, and little raw data or coordinates are deposited. So the new phase is not established. That said, the defect migration and phase transformation results are largely independent of this identification, so they can stand even if the Fe5Cl18 claim is weakened.

My overall take is conditional acceptance with the Fe5Cl18 part treated as tentative. A referee should ask the authors to either pin down the structure with calculations that include the graphene environment and quantitative comparison, or to present it as an unresolved observation. The migration and transformation story is solid enough to publish regardless.

Referee Report

2 major / 4 minor

Summary. The manuscript reports atomic-resolution STEM observations of iron chloride layers intercalated between bilayer graphene sheets, identifying Fe vacancies in FeCl2 and Fe interstitials/adatoms in FeCl3, tracking their electron-beam-driven migration, and documenting an irradiation-induced transformation from FeCl3 to FeCl2. The authors supplement the imaging with DFT calculations of migration barriers (1.4 eV for a Fe vacancy in FeCl2, 1.8 eV for a Fe interstitial in FeCl3), electron-beam displacement cross-sections, and electronic-structure calculations for FeClx. They also report a previously unreported crystalline phase assigned as Fe5Cl18, with a DFT model matched to ADF images after applying an isotropic rescaling of approximately 18%. The defect-migration and phase-transformation observations are direct and largely independent of the new-phase assignment, whereas the Fe5Cl18 identification is the principal weak point of the paper.

Significance. The defect dynamics and the FeCl3-to-FeCl2 transformation are significant because they connect atomic-scale defect processes to phase stability in a confined 2D ionic crystal, and the displacement cross-section analysis provides a quantitative, falsifiable framework for the irradiation-driven transformation. The DFT migration barriers are genuine predictions compared with observations, so those claims are not circular. The Fe5Cl18 phase, if confirmed, would be an interesting confined phase, but the currently presented evidence, based on an 18% rescaling of a graphene-free, zero-temperature model, is not yet convincing. The paper's overall significance is therefore conditional on strengthening or tempering the new-phase claim.

major comments (2)

[Results and Discussion, 'Unexpected iron chloride and iron oxychloride nanostructures' (Fig. 5c)] The identification of Fe5Cl18 rests on an approximately 18% isotropic rescaling of a graphene-free, zero-temperature DFT model to align with the experimental ADF dimensions. An 18% linear expansion corresponds to roughly a 39% increase in in-plane area, which is far beyond typical PBE+vdW lattice errors or room-temperature thermal expansion estimates. The proposed physical origins (anisotropic interlayer pressure, charge transfer, finite temperature) are not demonstrated for this material: no encapsulated relaxation of Fe5Cl18 between graphene layers is shown, no stress-strain calculation is presented, and no sensitivity analysis of the rescaling is given. The statement that the model 'is scaled up by approximately 18%' therefore leaves the structural assignment unsupported. Please provide independent confirmation, for example by relaxing the proposed structure inside a graphene bilayer model, or by quantitative image simulation with an explicitly justified lattice parameter, or by clearly presenting the Fe:Cl ratio with uncertainties from the EELS data in Fig. S12.
[Results and Discussion, 'Unexpected iron chloride and iron oxychloride nanostructures'] The manuscript asserts that the interlayer space of bilayer graphene 'stabilizes novel 2D materials' and, by implication, stabilizes the proposed Fe5Cl18 phase, but no thermodynamic or dynamical stability analysis of Fe5Cl18 is provided. No phonon calculation, ab initio molecular dynamics, or formation-energy comparison against competing phases such as FeCl3, FeCl2, or FeClO is presented. Without such analysis, the claim that Fe5Cl18 is a stable or metastable confined phase, rather than a beam-damaged or distorted form of a known iron chloride, is not established. The authors should either add stability calculations or explicitly label Fe5Cl18 as a tentative structural assignment.

minor comments (4)

[Migration behavior of multiple Fe vacancies in FeCl2] The sentence 'Overall, the energy of the FeCl2 phase (per structural unit) plus half the energy of isolated Cl2 molecule is higher than that of FeCl2 by 0.5 eV' is internally inconsistent as written; it likely should refer to FeCl3 plus half a Cl2 molecule relative to FeCl2. Please correct this, as it is used to support the thermodynamic direction of the phase transformation.
[Phase transformation from FeCl3 to FeCl2] The phrase 'release of chlorin molecules' should be 'chlorine molecules'. In addition, the text would benefit from a clearer distinction between the thermodynamic driving force and the irradiation-induced non-equilibrium pathway when discussing why the FeCl3-to-FeCl2 transformation occurs.
[Author contributions / Methods] The author contributions state that 'MD calculations' were performed, but the Methods section describes only DFT, NEB, and dispersion-corrected calculations. Either add a description of any molecular dynamics simulations that were actually used, or remove the mention of MD.
[Data availability] For a newly reported crystalline phase, the atomic coordinates of the proposed Fe5Cl18 model and the raw EELS spectra supporting the stoichiometry should be deposited in a public repository rather than being available only 'upon request', so that the structural assignment can be independently evaluated.

Circularity Check

1 steps flagged · score 4.0 of 10

The Fe5Cl18 structural assignment partially reduces to an 18% ad hoc rescaling of the DFT model; the defect-migration and phase-transformation results are independent.

fitted input called prediction [Results and Discussion, 'Unexpected iron chloride and iron oxychloride nanostructures' section, near Fig. 5c]
"We note that the proposed atomic model (calculated without graphene and at zero temperature) is scaled up by approximately 18% to align with the experimental dimensions."

The DFT model is rescaled by about 18% so that its lattice matches the experimental ADF image, and that same image is then used as support for the Fe5Cl18 structural assignment. The lattice match is therefore not an independent confirmation: the scale factor is a free parameter fitted to the very data the model is used to interpret. The paper offers only qualitative post hoc rationalizations (interlayer pressure, charge transfer, finite temperature) for the mismatch, with no separate calculation of the rescaled structure. EELS Fe:Cl ratios are independent for stoichiometry, but they do not determine the atomic arrangement, so the previously unreported Fe5Cl18 phase is identified by a model whose agreement with the image is enforced by construction.

full rationale

Most of the paper's derivation chain is self-contained. The migration barriers (1.4 eV for Fe vacancies in FeCl2, 1.8 eV for Fe interstitials in FeCl3), divacancy energetics, displacement cross-sections (210 barn vs 70 barn for Cl at 60 kV), and electronic-structure results are standalone DFT calculations compared with imaging observations or with prior independent calculations, so they are not circular. The FeCl3-to-FeCl2 transformation is supported by direct sequential ADF images and by calculated sputtering cross-sections, giving independent experimental and theoretical legs. The only partially circular element is the Fe5Cl18 phase identification: the DFT model is scaled by ~18% to align with the experimental ADF dimensions, yet the scaled model is then presented as the atomic model for that same image. The paper honestly discloses the rescaling, and the EELS-determined Fe:Cl ratio provides some independent stoichiometric evidence, so the circularity is partial rather than total. The invoked physical mechanisms for the 18% expansion are not quantitatively demonstrated, which weakens the new-phase claim, but that is a correctness risk rather than an additional circular step. Overall score 4: one significant fitted-rescaling element affecting the central novelty of the new phase, while the defect-dynamics and transformation results retain independent content.

Assumptions & free parameters 1 free parameters · 4 assumptions · 1 invented entities

The central claims rest on direct STEM observations plus DFT calculations. No truly free parameters appear in the migration-barrier calculations. The single explicit fitted quantity is the 18% rescaling of the Fe5Cl18 model, which is a load-bearing adjustment for the new phase claim. The displacement cross-section calculation and the PBE-D3 energetics are domain assumptions that the authors do not validate against experiment or higher-level theory.

free parameters (1)

Fe5Cl18 lattice rescaling = ~1.18 (18% increase)
DFT model scaled to match experimental ADF dimensions; the paper attributes the mismatch to pressure, charge transfer, or temperature without quantifying these effects.

assumptions (4)

domain assumption DFT-PBE with DFT-D3 dispersion gives reliable relative energies and migration barriers for FeClx monolayers.
Used to compute the 1.4 eV vacancy barrier, 1.8 eV interstitial barrier, divacancy binding energies, and phase energetics; no benchmark against higher-level methods is provided.
domain assumption McKinley-Feshbach displacement cross-section formalism, applied to isolated Cl and Fe atoms, correctly predicts relative sputtering rates from FeCl3 and FeCl2 at 60 kV.
This is the basis for the proposed Cl-loss-driven FeCl3 to FeCl2 mechanism; it neglects recombination, cascades, and the confined bilayer environment.
ad hoc to paper An isotropic ~18% expansion of the DFT Fe5Cl18 lattice can be caused by graphene pressure, charge transfer, and finite temperature without changing stoichiometry.
Invoked to match the model to ADF images in Figure 5c; no independent quantification of the 18% strain is given.
domain assumption ADF intensity variations can be assigned to Fe vacancies, adatoms, and interstitials through comparison with simulated STEM images.
The experimental classification of defect types relies on matching intensity profiles to simulations, which is done visually rather than with reported statistics.

invented entities (1)

Fe5Cl18 phase
purpose: Explains an observed ADF lattice and EELS Fe-to-Cl ratio that do not match FeCl2 or FeCl3; proposed as a new confined 2D iron chloride phase.
The phase is inferred from the same images and EELS spectra used to build the model; the 18% rescaling is a fitted adjustment, and no independent predicted observable (unique diffraction pattern, vibrational spectrum, or formation energy) is supplied to verify it outside this dataset.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Defect migration and phase transformations in 2D iron chloride inside bilayer graphene." pith.science (2026). https://pith.science/paper/TC45NPDN

@misc{pith2026250705665,
  author       = {Pith},
  title        = {Pith review of: Defect migration and phase transformations in 2D iron chloride inside bilayer graphene},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TC45NPDN}},
  note         = {Machine review of arXiv:2507.05665}
}

read the original abstract

The intercalation of metal chlorides, and particularly iron chlorides, into graphitic carbon structures has recently received lots of attention, as it can not only protect this two-dimensional (2D) magnetic system from the effects of the environment, but also substantially alter the magnetic, electronic, and optical properties of both intercalant and host material. At the same time, the intercalation can result in the formation of structural defects, or defects can appear under external stimuli, which can affect materials performance. These aspects have received so far little attention in the dedicated experiments. In this study, we investigate the behavior of atomic-scale defects in iron chlorides intercalated into bilayer graphene (BLG) by using scanning transmission electron microscopy (STEM) and first-principles calculations. We observe transformations between the FeCl2 and FeCl3 phases and elucidate the role of defects in the transformations. Specifically, three types of defects are identified: Fe vacancies in FeCl2 domains, Fe adatoms and interstitials in FeCl3 domains, each exhibiting distinct dynamic behaviors. We also observed a crystalline phase with an unusual stoichiometry of Fe5Cl18 which has not been reported before. Our findings not only advance the understanding of intercalation mechanism of 2D materials but also highlight the profound impact of atomic-scale defects on their properties and potential technological applications.

Figures

Figures reproduced from arXiv: 2507.05665 by the authors.

**Figure 2.** Migration behavior of Fe vacancies and Fe interstitials. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

**Figure 4.** Migration of Fe interstitials in FeCl3 and phase transformation. (a) Sequential ADF images reveal the structural transformation from FeCl3 to FeCl2. The FeCl2 domains, indicated by blue shading, grow from the borders of the imaging area where the electron dose accumulates. (b) Displacement cross-section threshold for Cl and Fe atoms, calculated using the McKinley-Feshbach formalism, showing the susceptibility of Cl … view at source ↗

Figures from the paper (1 more)

**Figure 5.** Figure 5: Structural dynamics of Fe5Cl18 and FeClO in BLG. (a) ADF image of Fe5Cl18 intercalated in BLG. (b,c) IFFT image and atomic model of the Fe5Cl18. (d-f) Sequential ADF images depicting the crystallization process of Fe5Cl18 initiated by ebeam irradiation. (g-i) ADF imag…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

4 extracted references · 4 canonical work pages

[2]

Fe vacancies

These groups of Fe vacancies can migrate collectively over longer distances as compared to isolated Fe vacancies in FeCl2 lattice. However, the migration of Fe vacancy groups does not expand the FeCl3 domain size sufficiently to induce a phase transformation from FeCl 2 to FeCl3 under electron beam scanning (see Movie S5). On the contrary, we observed tha...

work page
[3]

(e) The energy profile and intermediate atomic configurations of a Fe vacancy migrating in the FeCl 2 lattice indicate a migration barrier of 1.4 eV

(d) A schematic model showing the migration trajectory of a single Fe interstitial. (e) The energy profile and intermediate atomic configurations of a Fe vacancy migrating in the FeCl 2 lattice indicate a migration barrier of 1.4 eV . (f) The energy profile and intermediate atomic configurations of a Fe interstitial migrating in FeCl 3 lattice indicate a ...

work page
[4]

(17) Storm, A.; Köster, J.; Ghorbani -Asl, M.; Kretschmer, S.; Gorelik, T

Science 2015, 349 (6248), 625-628. (17) Storm, A.; Köster, J.; Ghorbani -Asl, M.; Kretschmer, S.; Gorelik, T. E.; Krasheninnikov, A. V .; Kaiser, U. Structural Transformations in Few -Layer MnPSe 3 Stimulated by Thermal Annealing and Electron Irradiation. J. Phys. Chem. C 2023, 127 (51), 24713-24723. (18) Lin, J.; Pantelides, S. T.; Zhou, W. Vacancy- Indu...

work page 2015
[5]

Science of 2.5 Dimensional Materials: Paradigm Shift of Materials Science Toward Future Social Innovation

Appl. Phys. Lett. 2015, 106 (19), 192404. (33) Ceyhan, E.; Yagmurcukardes, M.; Peeters, F. M.; Sahin, H. Electronic and Magnetic Properties of Single-Layer FeCl 2 with Defects. Phys. Rev. B 2021, 103 (1), 014106. (34) Dai, J.; Yadav, S.; Paulus, B. Electronic and Magnetic Properties of FeCl 3 Intercalated Bilayer Graphene. In C, 2023; V ol. 9. (35) Kim, N...

work page 2015

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [2]

Fe vacancies

These groups of Fe vacancies can migrate collectively over longer distances as compared to isolated Fe vacancies in FeCl2 lattice. However, the migration of Fe vacancy groups does not expand the FeCl3 domain size sufficiently to induce a phase transformation from FeCl 2 to FeCl3 under electron beam scanning (see Movie S5). On the contrary, we observed tha...

work page

[2] [3]

(e) The energy profile and intermediate atomic configurations of a Fe vacancy migrating in the FeCl 2 lattice indicate a migration barrier of 1.4 eV

(d) A schematic model showing the migration trajectory of a single Fe interstitial. (e) The energy profile and intermediate atomic configurations of a Fe vacancy migrating in the FeCl 2 lattice indicate a migration barrier of 1.4 eV . (f) The energy profile and intermediate atomic configurations of a Fe interstitial migrating in FeCl 3 lattice indicate a ...

work page

[3] [4]

(17) Storm, A.; Köster, J.; Ghorbani -Asl, M.; Kretschmer, S.; Gorelik, T

Science 2015, 349 (6248), 625-628. (17) Storm, A.; Köster, J.; Ghorbani -Asl, M.; Kretschmer, S.; Gorelik, T. E.; Krasheninnikov, A. V .; Kaiser, U. Structural Transformations in Few -Layer MnPSe 3 Stimulated by Thermal Annealing and Electron Irradiation. J. Phys. Chem. C 2023, 127 (51), 24713-24723. (18) Lin, J.; Pantelides, S. T.; Zhou, W. Vacancy- Indu...

work page 2015

[4] [5]

Science of 2.5 Dimensional Materials: Paradigm Shift of Materials Science Toward Future Social Innovation

Appl. Phys. Lett. 2015, 106 (19), 192404. (33) Ceyhan, E.; Yagmurcukardes, M.; Peeters, F. M.; Sahin, H. Electronic and Magnetic Properties of Single-Layer FeCl 2 with Defects. Phys. Rev. B 2021, 103 (1), 014106. (34) Dai, J.; Yadav, S.; Paulus, B. Electronic and Magnetic Properties of FeCl 3 Intercalated Bilayer Graphene. In C, 2023; V ol. 9. (35) Kim, N...

work page 2015