Pith. sign in

REVIEW 3 major objections 5 minor 65 references

Quantitative electron beam-single atom interactions enabled by sub-20-pm precision targeting

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Atomic lock-on parks a scanning electron beam on a chosen atomic column with sub-20-picometer precision, without prior irradiation of that site, and uses the lock to read single-atom spectra and resolve individual atoms switching bonds.

desk verdict ALO is a genuine in situ STEM targeting advance, and the sub-20 pm precision claim is credible even if not fully pinned down; referee it, but ask for distortion quantification and tighter statistics. read the letter →

arxiv 2506.23255 v1 pith:EPBR37QQ submitted 2025-06-29 cond-mat.mes-hall cond-mat.mtrl-scicond-mat.other

classification cond-mat.mes-hallcond-mat.mtrl-scicond-mat.other PACS 68.37.Ma
keywords atomiclock-onsingle-atompositioningscanningtransmissionelectronmicroscopypicometerprecisionannularscanenergylossspectroscopydynamicsbeam-inducedbistability
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 lock-on (ALO) is a beam-positioning technique that parks a scanning transmission electron microscope beam on a chosen atomic column with sub-20-picometer repeatability, without ever having irradiated that column first. The paper shows this by locking onto Cr columns in 16-layer CrSBr with 18 ± 10 pm precision, and onto Mo, W, and S columns in monolayer dichalcogenides at sub-30 pm. Once locked on, the beam stays on one atom through repeated lock-ons that compensate sample drift, allowing weak single-atom EELS signals to be integrated for seconds and single-atom HAADF intensity to be sampled every 10 microseconds. The upshot is that single-atom events such as sulfur ejection, recapture, and bistable re-bonding become directly observable and countable, which the authors argue opens the way to deterministic electron-beam fabrication.

What carries the argument

The carrying mechanism is the atomic lock-on (ALO) workflow: a three-loop annular scan of about 1 nm radius that crosses atom columns at near-constant velocity while the target interior stays undosed; a threshold-and-cluster step that turns the sparse detector blips into experimental atom positions; and a residual-minimization step that fits a rigid, pre-known lattice (lattice vectors plus rotation) to those points by an optimal translation vector $\vec{\Delta}_{xy}$. Once the sub-lattice is reconstructed, every site in the unit cell, whether atoms, bonds, or voids, is known by a fixed translation from it, so the beam can be positioned without ever observing the target. Radial symmetry and uniform beam speed suppress the non-linear scan distortions that plague raster scans, and the residual minimization absorbs drift at the moment of each lock-on.

What would settle it

Place an ALO lock-on on a well-characterized crystalline sample and measure the offset with an independent calibration, for example the same 1-nm spiral taken with reversed scan rotation, or with the sample rotated 180 degrees, or a reference lattice whose positions are known from a calibrated metrology image, and check whether the 18 ± 10 pm scatter persists as shot-noise-limited or instead reveals a systematic few-picometer distortion floor.

Watch

Extended reading notes

Core claim

The central claim is that a sparse annular scan, executed outside the target region and matched to a known lattice, can determine the absolute position of any lattice site precisely enough to place the beam on it, with measured repeatability of 18 ± 10 pm on a Cr column in 16-layer CrSBr and 27–29 pm on W, Mo, or S columns in monolayers, without a prior raster image and without dosing the target. The paper reports that this precision is achieved in about 100 ms at 20 pA, requires no parent image for periodic crystals, and is actively corrected for drift by re-running the lock-on. With the beam held on one site, the authors resolve, for the first time in STEM, single sulfur atoms switching between bonded and partially bonded states, with the shortest resolvable dwell below 1 ms and sampling limited to 10 µs by the detector.

Load-bearing premise

The reported precision is measured by collecting a tiny 1-nm spiral image after each lock-on and reading the beam offset from it, so the sub-20-pm number depends on the assumption that these verification images themselves carry almost no scan distortion or drift; the paper asserts but does not quantify that residual.

Editorial extensions

If this is right

  • Repeated ALO lock-ons keep the beam on one V dopant atom in monolayer MoS2 for over 10 seconds, compensating drift rates above 1 Å/s, and yield single-shot EELS spectra of the L3,2 edge.
  • HAADF sampling at 100 kHz (10 µs) resolves single sulfur-atom ejection, subsequent recapture, and random telegraph noise between 2S and 1S configurations, interpreted as a beam-driven bistable atomic system.
  • Because the annular scan leaves the target undosed until measurement, ALO is compatible with beam-sensitive monolayers and permits the same site to be revisited many times.
  • Targeting is not limited to atoms: the same precision positions the beam on bonds or voids, and the measured HAADF signal can serve as an endpoint detector for automated atomic manipulation.
  • The authors expect ALO to extend to moderately disordered systems, interfaces, and low-Z materials such as graphene and hexagonal boron nitride with minor algorithmic changes.

Reading between the lines

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

  • If the reference-image distortions are genuinely negligible, the practical ceiling on ALO precision is set by the interaction volume of the probe (~80 pm FWHM) and thermal vibrations, meaning further algorithmic gains reduce positioning scatter but not the physical resolution of the measurement.
  • The 10 µs sampling rate is far faster than the ≤1 ms shortest resolved 2S dwell, so the true occupancy times of the bistable sulfur state are likely even shorter; correlating dwell-time histograms with beam energy and dose rate would give a direct test of the bistability picture.
  • ALO's rigid-lattice assumption is both its strength and its boundary: the sub-20 pm claim applies to periodic crystals, and defects, strain fields, or phase boundaries inside the annular scan would bias the fitted translation unless the algorithm is extended to multiple characteristic arrangements.
  • A natural next experiment is closed-loop fabrication: use the real-time HAADF trace to detect a displacement event and immediately blank or reposition the beam, which the authors suggest and their endpoint-detection argument makes directly implementable.
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.

Referee Report

3 major / 5 minor

Summary. The paper introduces "atomic lock-on" (ALO), a sparse annular-scan beam-positioning technique for STEM that reconstructs the local lattice from a fast, low-dose spiral scan and then positions the beam on a chosen atomic column without first irradiating the target site. The authors report sub-20 pm targeting precision in CrSBr (18 ± 10 pm) and sub-30 pm in MoS2 and WS2, outperforming a DCNN-based approach (>100 pm). They demonstrate repeated single-atom EELS on a V dopant in MoS2 with drift compensation, and fast (10 µs sampling) HAADF time traces in WS2 that they interpret as single-atom dynamics, including S ejection, recapture, and random telegraph noise.

Significance. If the precision claim holds, ALO is a valuable methodological advance for single-atom spectroscopy, site-specific manipulation, and quantitative electron-beam interaction studies. The work is strengthened by a direct experimental precision measurement that is not circular with respect to ALO's internal reconstruction, by systematic grid-search optimization of scan parameters, and by application to multiple materials systems (CrSBr, MoS2, WS2). The low-dose, blind-targeting capability and the ability to repeatedly lock onto a drifting target are demonstrated with concrete experiments, including automated tracking of a single V dopant atom and 10 µs-resolution HAADF monitoring. These strengths make the paper potentially important for the microscopy and quantum-technology communities, provided the load-bearing precision and interpretation issues identified below are resolved.

major comments (3)
  1. [Fig. 2d and SI Data Fig. 1] The central sub-20 pm precision claim is verified using 1 nm FOV spiral HAADF images, and the text states that "the distortions in this small area scan are small" without any quantitative bound on residual distortion, drift, or scan noise. Because ALO is designed precisely to correct scan distortions, the verification pathway must be demonstrated to be free of such distortions at the few-pm level; otherwise the measured offset (18 ± 10 pm) conflates targeting error with imaging distortion. I recommend adding an independent calibration, for example by comparing the spiral image lattice vectors to a distortion-corrected reference (e.g., an orthogonal-scan pair or a larger-field calibration image) and reporting the residual distortion explicitly.
  2. [SI Section 2 (Grid search optimization, Figs. S2-S3)] The simulated ALO precision (~5 pm for CrSBr, ~13 pm for MoS2) is validated against a "ground truth" obtained from HAADF-STEM images via DCNN detection and 2D Gaussian refinement. That ground truth is itself affected by the same scan distortions and drift that ALO claims to correct, so the simulation primarily shows consistency with a distorted reference rather than absolute precision. This is a circularity concern: the simulation cannot independently establish the correction capability. I suggest validating against synthetic images with known atom positions or against a distortion-corrected experimental reference, and reporting the resulting precision separately from the experimental precision.
  3. [Fig. 5 and accompanying text] The single-atom dynamics (S ejection, recapture, RTN, bistability) are inferred from 10 µs HAADF intensity time traces without an independent atomic-scale verification of each event. Intensity steps could in principle arise from small beam displacements, focus changes, or scan instabilities, and the paper does not provide a quantitative model relating HAADF intensity to the proposed atomic configurations beyond the Z-dependence in Fig. 5d. To support the "quantitative electron beam-single atom interactions" claim, the authors should either calibrate the intensity-to-configuration mapping (e.g., through image simulations or simultaneous acquisition) or clearly state the classification ambiguity, and report the noise floor of the time traces relative to the step sizes.
minor comments (5)
  1. [Fig. 5 caption] The text in Fig. 5b contains a typo: "28 ± 18pm ad 27 ± 17pm" should read "28 ± 18 pm and 27 ± 17 pm".
  2. [Methods (Data post processing)] The paper states "We performed no post-processing of collected HAADF-STEM images," but Fig. 5 shows a moving average of 3 ms applied to time traces. This is not an image post-processing step, but the wording should be clarified to avoid confusion.
  3. [References] References [38] and [41] are the same work (Sang et al., Dynamic scan control in STEM: spiral scans); one should be removed or cross-referenced differently.
  4. [Methods (dose calculation)] The dose calculation uses "Ã" and "F W HM" (apparent rendering artifacts for pi and FWHM); the final typeset version should use standard notation. Also, the statement that the target region "receives no dose" (Fig. 1b) is an idealization; while the probe tail at ~1 nm is negligible for a 0.8 Å FWHM probe, a brief quantitative statement of the upper bound on target-area dose would strengthen the low-dose claim.
  5. [General] The paper uses "precision" to describe both the mean offset and its standard deviation (e.g., 18 ± 10 pm and 28 ± 18 pm). Clarify that the reported value is the mean absolute offset and the scatter, and define the metric used for "sub-20 pm" in the abstract and conclusions.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: ALO's sub-20-pm targeting precision is measured by independent post-hoc spiral images, and the paper's self-citations are baselines, not load-bearing premises.

full rationale

The central claim—that ALO positions the beam on a target column with sub-20 pm precision—rests on an external measurement, not on ALO's internal reconstruction. After ALO, a separate 1 nm FOV spiral HAADF-STEM image is recorded centered on the beam position, and the offset of the target Cr column from the image center is fitted from the four nearest-neighbor S/Br columns (main text, "After targeting with either method..."; SI Fig. S1). This offset is therefore a new observable, not a re-statement of the translation vector Δ_xy computed by the ALO lattice fit, so no fitted input is renamed as a prediction. The SI grid search (Figs. S2–S3) validates ALO against atom positions obtained from a HAADF-STEM image via DCNN + 2D Gaussian refinement; this is a consistency check on the reconstruction pipeline, and the paper does not use the resulting ~5 pm simulation value as the experimental precision. The self-citations [33,34] define the DCNN baseline and the V-dopant detector; neither is used to justify the ALO precision, so they are not load-bearing. One legitimacy concern is noted but is not circular: the verification images are small spiral scans, and the paper asserts "The distortions in this small area scan are small and allow us to accurately determine the precision" without quantifying the residual distortion. That is a measurement-validation limitation that belongs under correctness risk, not a reduction of the claimed precision to its own inputs.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the experimental validity of the sparse annular scan and the lattice reconstruction. The key free parameters are the scan geometry, threshold level, and dwell time, all tuned via simulation or operator choice. The main assumptions are periodicity of the sample and the distortion-free nature of the small reference images used to verify precision. No new physical entities are introduced.

free parameters (4)
  • threshold_percentage = 90-120% of DCNN-estimated column intensity distribution
    Threshold for isolating atom-column pixels in the annular scan; chosen by operator and dose, affects which pixels enter the lattice fit (SI Sec 3.1).
  • annular_scan_geometry = N=3 loops, r_out=1 nm, r_in=0.8 nm (CrSBr); r_in=0.7 nm (MoS2)
    Optimal scan shape selected via grid search over 912 shapes; N=1 has failures, N=3 chosen as robust.
  • dwell_time_pixel_density = 100x100 pixels over annulus, 100 ms scan
    Chosen to give <50 pm reconstruction precision 100% of the time in simulations; scan duration 100 ms at 20 pA.
  • HAADF_Z_exponent = Z^1.6-1.7
    Guide-to-eye model for HAADF intensity versus atomic number used in Fig 5d to assign atomic configurations; not used for quantitative cross-sections.
assumptions (5)
  • domain assumption sample material is periodic with known lattice constants and rotation
    Required for constructing the artificial lattice L_j used in reconstruction; stated in Methods/SI 3.2, and the paper notes ALO works 'provided the material is periodic'.
  • domain assumption the 1 nm FOV spiral reference images are free of significant scan distortion
    Used to measure targeting offset by fitting nearest-neighbor columns; the paper asserts 'distortions in this small area scan are small' without quantifying the residual distortion.
  • domain assumption HAADF intensity change maps one-to-one to atomic configuration at the target column
    Basis for interpreting step-like drops as S ejection and intensity levels as 2S vs 1S states (Fig 5d); ignores possible contributions from beam-induced lattice strain, adatoms, or detector noise.
  • domain assumption drift is uniform over the lock-on interval and can be compensated by a single translation vector
    ALO re-centering uses a rigid translation; non-rigid distortions or z-drift are neglected (acknowledged in Discussion).
  • standard math standard statistics and least-squares minimization are valid
    Euclidean distance minimization for residual e (Eqs 1-4) and Gaussian fits for atom positions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantitative electron beam-single atom interactions enabled by sub-20-pm precision targeting." pith.science (2026). https://pith.science/paper/EPBR37QQ

@misc{pith2026250623255,
  author       = {Pith},
  title        = {Pith review of: Quantitative electron beam-single atom interactions enabled by sub-20-pm precision targeting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EPBR37QQ}},
  note         = {Machine review of arXiv:2506.23255}
}
read the original abstract

The ability to probe and control matter at the picometer scale is essential for advancing quantum and energy technologies. Scanning transmission electron microscopy offers powerful capabilities for materials analysis and modification, but sample damage, drift, and scan distortions hinder single atom analysis and deterministic manipulation. Materials analysis and modification via electron-solid interactions could be transformed by precise electron delivery to a specified atomic location, maintaining the beam position despite drift, and minimizing collateral dose. Here we develop a fast, low-dose, sub-20-pm precision electron beam positioning technique, atomic lock-on, (ALO), which offers the ability to position the beam on a specific atomic column without previously irradiating that column. We use this technique to lock onto the same selected atomic location to repeatedly measure its weak electron energy loss signal despite sample drift. Moreover, we quantitatively measure electron beam matter interactions of single atomic events with microsecond time resolution. This enables us to observe single atom dynamics such as atomic bistability in the electron microscope, revealing partially bonded atomic configurations and recapture phenomena. We discuss the prospects for high-precision measurements and deterministic control of matter for quantum technologies using electron microscopy.

Figures

Figures reproduced from arXiv: 2506.23255 by the authors.

Figure 1
Figure 1. FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. summarizes the workflow of ALO. Figure 3a shows the sparse annular scan which consists of three loops and has an outer radius of 1 nm and inner radius of 0.8 nm. This is performed (instead of a full-area raster scan) immediately before targeting a desired site, and can be repeated as needed during extended experiments over time intervals that depend on the stability of the micro￾scope, the sample, and other factors,… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

65 extracted references · 53 canonical work pages

  1. [1]

    E., Dellby, N

    Batson, P. E., Dellby, N. & Krivanek, O. L. Sub- ångstrom resolution using aberration corrected electron optics. Nature 418, 617–620 (2002). URL https: //doi.org/10.1038/nature00972

  2. [2]

    D., Kisielowski, C

    Erni, R., Rossell, M. D., Kisielowski, C. & Dahmen, U. Atomic-resolution imaging with a sub-50-pm electron probe. Physical Review Letters 102 (2009). URL http: //dx.doi.org/10.1103/PhysRevLett.102.096101

  3. [3]

    Zhou, W. et al. Intrinsic structural defects in mono- layer molybdenum disulfide. Nano Letters 13, 2615–2622 (2013). URL http://dx.doi.org/10.1021/nl4007479

  4. [4]

    Because our target is a single, non-periodic feature, (a V dopant atom), we first must collect a single parent image

    Quantifying a single dopant atom is a challenging task [ 19, 39], and therefore ideal for this demonstration. Because our target is a single, non-periodic feature, (a V dopant atom), we first must collect a single parent image. (We emphasize that this step is not necessary if the sample is periodic). We then identify the initial location to be targeted wit...

  5. [5]

    Krivanek, O. L. et al. Vibrational spectroscopy in the electron microscope. Nature 514, 209–212 (2014). URL http://dx.doi.org/10.1038/nature13870

  6. [6]

    Krivanek, O. L. et al. Atom-by-atom structural and chemical analysis by annular dark-field electron mi- croscopy. Nature 464, 571–574 (2010). URL http: //dx.doi.org/10.1038/nature08879

  7. [7]

    & Koshino, M

    Suenaga, K. & Koshino, M. Atom-by-atom spectroscopy at graphene edge. Nature 468, 1088–1090 (2010). URL http://dx.doi.org/10.1038/nature09664

  8. [8]

    Zhou, W. et al. Direct determination of the chemical bonding of individual impurities in graphene. Physical Review Letters 109 (2012). URL http://dx.doi.org/ 10.1103/PhysRevLett.109.206803

Show all 65 references
  1. [9]

    Batson, P. E. Motion of gold atoms on carbon in the aberration-corrected STEM. Microscopy and Microanal- ysis 14, 89–97 (2007). URL https://doi.org/10.1017/ s1431927608080197

  2. [10]

    Susi, T. et al. Silicon–carbon bond inversions driven by 60-kev electrons in graphene. Physical Review Let- ters 113 (2014). URL http://dx.doi.org/10.1103/ PhysRevLett.113.115501

  3. [12]

    Tripathi, M. et al. Electron-beam manipulation of silicon dopants in graphene. Nano Letters 18, 5319–5323 (2018). URL http://dx.doi.org/10.1021/ acs.nanolett.8b02406

  4. [13]

    Dyck, O. et al. Atom-by-atom fabrication with electron beams. Nature Reviews Materials 4, 497–507 (2019). URL http://dx.doi.org/10.1038/ s41578-019-0118-z

  5. [14]

    R.-P., Barbone, M., Aharonovich, I., Atatüre, M

    Montblanch, A. R.-P., Barbone, M., Aharonovich, I., Atatüre, M. & Ferrari, A. C. Layered materials as a platform for quantum technologies. Nature Nanotechnol- ogy 18, 555–571 (2023). URL http://dx.doi.org/10. 1038/s41565-023-01354-x

  6. [15]

    & Malac, M

    Egerton, R., Li, P. & Malac, M. Radiation damage in the TEM and SEM. Micron 35, 399–409 (2004). URL https://doi.org/10.1016/j.micron.2004.02.003

  7. [16]

    Instrumental factors in high-resolution feg stem

    von Harrach, H. Instrumental factors in high-resolution feg stem. Ultramicroscopy 58, 1–5 (1995). URL http: //dx.doi.org/10.1016/0304-3991(94)00172-J

  8. [17]

    & Nellist, P

    Jones, L. & Nellist, P. D. Identifying and correct- ing scan noise and drift in the scanning transmission electron microscope. Microscopy and Microanalysis 19, 1050–1060 (2013). URL http://dx.doi.org/10.1017/ S1431927613001402

  9. [18]

    García de Abajo, F. J. Optical excitations in electron mi- croscopy. Reviews of Modern Physics 82, 209–275 (2010). URL http://dx.doi.org/10.1103/RevModPhys.82.209

  10. [19]

    Susi, T., Meyer, J. C. & Kotakoski, J. Quantifying transmission electron microscopy irradiation effects us- ing two-dimensional materials. Nature Reviews Physics 1, 397–405 (2019). URL http://dx.doi.org/10.1038/ s42254-019-0058-y

  11. [20]

    Ramasse, Q. M. et al. Probing the bonding and elec- tronic structure of single atom dopants in graphene with electron energy loss spectroscopy. Nano Letters 13, 4989–4995 (2013). URL http://dx.doi.org/10.1021/ nl304187e

  12. [21]

    Wang, S. et al. Atomic structure and formation mech- anism of sub-nanometer pores in 2d monolayer mos2. Nanoscale 9, 6417–6426 (2017). URL http://dx.doi. org/10.1039/C7NR01127J

  13. [22]

    Huang, P. Y. et al. Imaging atomic rearrangements in two-dimensional silica glass: Watching silica’s dance. Sci- ence 342, 224–227 (2013). URL http://dx.doi.org/10. 1126/science.1242248

  14. [23]

    O., Huang, Y.-S

    Lin, Y.-C., Dumcenco, D. O., Huang, Y.-S. & Suenaga, K. Atomic mechanism of the semiconducting-to-metallic phase transition in single-layered mos2. Nature Nan- otechnology 9, 391–396 (2014). URL http://dx.doi. org/10.1038/nnano.2014.64

  15. [24]

    Klein, J. et al. Control of structure and spin texture in the van der waals layered magnet crsbr. Nature Commu- nications 13 (2022). URL http://dx.doi.org/10.1038/ s41467-022-32737-8

  16. [25]

    Isaacson, M., Kopf, D., Utlaut, M., Parker, N. W. & Crewe, A. V. Direct observations of atomic diffusion by scanning transmission electron microscopy. Proceed- ings of the National Academy of Sciences 74, 1802–1806 (1977). URL http://dx.doi.org/10.1073/pnas.74.5. 1802

  17. [26]

    Girit, C. O. et al. Graphene at the edge: Stability and dynamics. Science 323, 1705–1708 (2009). URL http: //dx.doi.org/10.1126/science.1166999

  18. [27]

    J., Idrobo, J.-C

    Lee, J., Zhou, W., Pennycook, S. J., Idrobo, J.-C. & Pan- telides, S. T. Direct visualization of reversible dynamics in a si6 cluster embedded in a graphene pore. Nature Communications 4 (2013). URL http://dx.doi.org/ 10.1038/ncomms2671

  19. [28]

    & Meyer, J

    Kotakoski, J., Mangler, C. & Meyer, J. C. Imaging atomic-level random walk of a point defect in graphene. Nature Communications 5 (2014). URL http://dx.doi. org/10.1038/ncomms4991

  20. [29]

    Li, H. et al. Atomic structure and dynamics of single platinum atom interactions with monolayer mos2. ACS Nano 11, 3392–3403 (2017). URL http://dx.doi.org/ 10.1021/acsnano.7b00796

  21. [30]

    Komsa, H.-P. et al. Two-dimensional transition metal dichalcogenides under electron irradiation: De- fect production and doping. Physical Review Let- 11 ters 109 (2012). URL http://dx.doi.org/10.1103/ PhysRevLett.109.035503

  22. [31]

    Meyer, J. C. et al. Accurate measurement of electron beam induced displacement cross sections for single-layer graphene. Physical Review Let- ters 108 (2012). URL http://dx.doi.org/10.1103/ PhysRevLett.108.196102

  23. [32]

    & Krashenin- nikov, A

    Kretschmer, S., Lehnert, T., Kaiser, U. & Krashenin- nikov, A. V. Formation of defects in two-dimensional mos2 in the transmission electron microscope at elec- tron energies below the knock-on threshold: The role of electronic excitations. Nano Letters 20, 2865–2870 (2020). UR...

  24. [33]

    Speckmann, C. et al. Combined electronic excitation and knock-on damage in monolayer mos 2. Physical Re- view B 107 (2023). URL http://dx.doi.org/10.1103/ PhysRevB.107.094112

  25. [34]

    Roccapriore, K. M. et al. Probing electron beam in- duced transformations on a single-defect level via auto- mated scanning transmission electron microscopy. ACS Nano 16, 17116–17127 (2022). URL https://doi.org/ 10.1021/acsnano.2c07451

  26. [35]

    M., Torsi, R., Robinson, J., Kalinin, S

    Roccapriore, K. M., Torsi, R., Robinson, J., Kalinin, S. & Ziatdinov, M. Dynamic stem-eels for single-atom and defect measurement during electron beam trans- formations. Science Advances 10 (2024). URL http: //dx.doi.org/10.1126/sciadv.adn5899

  27. [36]

    & LeBeau, J

    Sang, X. & LeBeau, J. M. Revolving scanning trans- mission electron microscopy: Correcting sample drift distortion without prior knowledge. Ultramicroscopy 138, 28–35 (2014). URL https://doi.org/10.1016/j. ultramic.2013.12.004

  28. [37]

    & Nelson, C

    Ophus, C., Ciston, J. & Nelson, C. T. Correcting non- linear drift distortion of scanning probe and scanning transmission electron microscopies from image pairs with orthogonal scan directions. Ultramicroscopy 162, 1–9 (2016). URL https://doi.org/10.1016/j.ultramic. 2015.12.002

  29. [39]

    Sang, X. et al. Precision controlled atomic resolution scanning transmission electron microscopy using spiral scan pathways. Scientific Reports 7 (2017). URL https: //doi.org/10.1038/srep43585

  30. [40]

    Robertson, A. W. et al. Atomic structure and spec- troscopy of single metal (cr, v) substitutional dopants in monolayer mos2. ACS Nano 10, 10227–10236 (2016). URL http://dx.doi.org/10.1021/acsnano.6b05674

  31. [41]

    Yoshimura, A. et al. Quantum theory of electronic excitation and sputtering by transmission electron mi- croscopy. Nanoscale 15, 1053–1067 (2023). URL http: //dx.doi.org/10.1039/D2NR01018F

  32. [42]

    Sang, X. et al. Dynamic scan control in STEM: spiral scans. Advanced Structural and Chemical Imaging 2 (2016). URL https://doi.org/10.1186/ s40679-016-0020-3

  33. [43]

    G., Dyck, O., Kalinin, S

    Ghosh, A., Sumpter, B. G., Dyck, O., Kalinin, S. V. & Ziatdinov, M. Ensemble learning-iterative training machine learning for uncertainty quantification and au- tomated experiment in atom-resolved microscopy. npj Computational Materials 7 (2021). URL https://doi. org/10.1038/s4...

  34. [44]

    & Brox, T

    Ronneberger, O., Fischer, P. & Brox, T. U-Net: Con- volutional Networks for Biomedical Image Segmentation , 234–241 (Springer International Publishing, 2015). URL http://dx.doi.org/10.1007/978-3-319-24574-4_28 . Supplemental Information: Quantitative electron beam-single atom ...

  35. [45]

    Atomic Lock-On and Deep Convolutional Neural Network (DCNN) Targeting Precision 2

  36. [46]

    Grid search optimization for 16L CrSBr and 1L MoS 2 4

  37. [47]

    Thresholding Condition 6 3.2

    Additional Details of Atomic Lock-On 6 3.1. Thresholding Condition 6 3.2. A Priori Information for Lattice Reconstruction 8 3.3. Relative Targeting 8

  38. [48]

    Automated and dynamic drift compensation. 9

  39. [49]

    Dose-Effectiveness of Atomic Lock-On. 9

  40. [50]

    Atomic Lock-On Precision for 1L WS 2 11

  41. [51]

    12 References 14

    Additional Single-Atom Time Dynamics. 12 References 14

  42. [52]

    Experimental comparison between atomic lock-on and a DCNN

    Atomic Lock-On and Deep Convolutional Neural Network (DCNN) Targeting Precision Figure S1 . Experimental comparison between atomic lock-on and a DCNN. a , HAADF-STEM spiral images (1nm FOV) after ALO targeting of the Cr atom column. Red dots are the four nearest neighbor fitted...

  43. [53]

    Ability to measure the po- sition of an atom - - few-pm [ 1]

  44. [54]

    Ability to place the beam on an atom Accurate <20pm Inaccurate >100pm -

  45. [55]

    Electron dose to atom of interest Atom remains undosed until measurement Atom has been dosed prior to measurement -

  46. [56]

    Image requirement No pre-acquired image nec- essary (works blind) Requires pre-acquired (raster-)scanned image -

  47. [57]

    Distortion and drift compensation Actively compensates for distortion and drift through continuous acceleration in annular scanning Ignores distortion and drift (based on static, pre- acquired image) -

  48. [58]

    Atom position source Positions derived directly from sparse annular (spiral) beam motion, ensuring real- time accuracy Positions determined via DCNN analysis of a (raster- )scanned image -

  49. [59]

    Dosing in region of inter- est (ROI) Minimal, targeted dosing only to essential subregions, minimizing unnecessary exposure All atoms within ROI re- ceive electron dose -

  50. [60]

    Comparison between general ex situ atom position precision and in situ ALO and DCNN based beam positioning

    Capability for Time- Resolved Studies Enables time-resolved stud- ies by preserving the atom of interest, avoiding cumula- tive exposure - Not suitable for tracking single atoms over time due to continuous dosing T able S1. Comparison between general ex situ atom position prec...

  51. [61]

    Our motivation is to determine the annular scan shape with the highest precision and the absence of failures during repeated operations

    Grid search optimization for 16L CrSBr and 1L MoS 2 We perform a grid search optimization to obtain the optimal annular scan shape. Our motivation is to determine the annular scan shape with the highest precision and the absence of failures during repeated operations. Our reas...

  52. [62]

    Thresholding Condition To isolate the pixels associated with the sub-lattice of interest, we threshold the HAADF-STEM annular scan

    Additional Details of Atomic Lock-On 3.1. Thresholding Condition To isolate the pixels associated with the sub-lattice of interest, we threshold the HAADF-STEM annular scan. There are two options to threshold: either using a defined percentage of all pixels based on the expecte...

  53. [63]

    Figure S5

    Automated and dynamic drift compensation. Figure S5 . Automated tracking of a single V dopant atom in 1L WS 2. Repeated ALO on a single V dopant atom in 1L WS 2. ALO compensates a maximum drift rate of 2 Å s−1

  54. [64]

    Besides spectroscopically probing atoms, reconstructing the lattice by ALO allows sub-atomic placement with the ability to target bonds or other high symmetry sites in the crystal

    Dose-Effectiveness of Atomic Lock-On. Besides spectroscopically probing atoms, reconstructing the lattice by ALO allows sub-atomic placement with the ability to target bonds or other high symmetry sites in the crystal. We demonstrate this with 16L 9 Figure S6 . Automated single...

  55. [65]

    ALO precision for targeting W and 2S atom columns on 1L WS 2

    Atomic Lock-On Precision for 1L WS 2 Figure S7. ALO precision for targeting W and 2S atom columns on 1L WS 2. a, b , ALO targeting of the W atom and 2S atom column with 28 ± 18pm ad 27 ± 17pm precision, respectively. c, Comparison to targeting the W atom column using a DCNN. 11

  56. [66]

    Figure S8

    Additional Single-Atom Time Dynamics. Figure S8 . Single-atom dynamics targeting W in 1L WS 2. a, b , Unchanged crystal structure after 1s of targeting W atom column. c, Adatom diffusion with a HAADF intensity in excellent agreement with a W atom. In the final frame after the ex...

  57. [67]

    A. B. Yankovich, B. Berkels, W. Dahmen, P. Binev, S. I. Sanchez, S. A. Bradley, A. Li, I. Szlufarska, and P. M. Voyles, Nature Communications 5 (2014), URL https://doi.org/10.1038/ncomms5155. 14

Pith tools

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