REVIEW 3 major objections 3 minor 2 references
Standardizing Force Reconstruction in Dynamic Atomic Force Microscopy
T0 review · 3 major / 3 minor · reviewed 2026-08-07 · deepseek-v4-flash
Pith's one-line read In dynamic AFM, the force you measure depends on the formula you pick; this paper unifies the field's methods in one open-source tool and identifies hybrid reconstructions as the most faithful.
desk verdict A competent review of AFM force-reconstruction methods whose central claim—an open-source benchmarking package—is undermined by the fact that the code is only available on request, and the supporting comparison is qualitative and self-referential. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the general integral equation linking the measured frequency shift (FM mode) or amplitude and phase (AM mode) to the tip–sample force, together with the stabilization device introduced to solve it. That device is the hybrid reconstruction: the small-amplitude term in the perturbative expansion, which is the main source of numerical instability in steep repulsive force gradients, is replaced by a first-order Bessel-function approximation (Section 2.2.3). Around this mathematical core, the paper also builds an open-source MATLAB/Simulink application, with numerical tolerances set to $10^{-15}$, that filters, interpolates, and integrates amplitude, phase, and frequency-shift data and then computes the force via finite differencing of the reconstructed potential energy.
What would settle it
Generate synthetic frequency-shift and amplitude/phase data from a known interaction potential with realistic white noise at several signal-to-noise ratios, run all six reconstructions, and compute RMS deviation from the true force across the full distance range; the hybrid method wins the comparison only if its error is lowest across most of that grid.
Extended reading notes
Core claim
The central claim is that force reconstruction from dynamic AFM observables is an ill-posed inverse problem, and the choice of formulation dominates the recovered force curve. The paper demonstrates, under ideal noise-free conditions on a single interaction model, that frequency- and amplitude-based matrix and Laplace methods either reproduce only a fraction of the force or develop oscillations near closest approach, while the hybrid formulation—which replaces the small-amplitude term of the perturbative expansion with a first-order Bessel-function approximation—tracks the true force with minimal deviation. On this basis it asserts that hybrid methods more accurately capture the true force profile and provides the first publicly available software tool that unifies all major reconstruction methods for side-by-side benchmarking.
Load-bearing premise
The performance ranking rests on a single noise-free interaction model with no quantitative error metrics; if real data, with noise, amplitude variation, or a different force law, reorders the methods, the headline comparison loses its force.
Editorial extensions
If this is right
- A lab using the provided software can compute force–distance curves from the same raw data by every major method, making method choice a reproducible parameter rather than a hidden assumption.
- The comparison indicates that small-amplitude experiments are better served by the hybrid formulation than by the polynomial or Laplace approaches.
- Community adoption of a standardized benchmark would allow quantitative force values from different AFM groups to be compared directly, something the paper says is currently not possible.
- Future method development can be judged against a common reference implementation rather than against bespoke validation data.
Reading between the lines
- If the hybrid family's advantage holds when noise, amplitude drift, and force-model mismatch are added, one would expect it to become the default for small-amplitude FM and AM experiments; the paper itself does not yet provide those noise-benchmarked results.
- A testable extension is to run all six algorithms on synthetic data with controlled signal-to-noise ratio and a known interaction law, then report RMS error curves across amplitude and distance—this would turn the qualitative Figure 2 comparison into a quantitative ranking.
- The same software architecture could be extended to dissipative forces and multichannel data (simultaneous amplitude, phase, and frequency shift), which the Discussion identifies as an open need; nothing in the current comparison addresses energy-loss channels.
- Because the paper's benchmark uses one assumed force model, the comparative ranking may not transfer to heterogeneous samples; a multi-model benchmark (e.g., van der Waals plus short-range repulsion, with and without adhesion hysteresis) would reveal which methods are model-robust.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript reviews the main mathematical methods for reconstructing tip–sample interaction forces in dynamic AFM, covering both frequency-modulation (FM) and amplitude-modulation (AM) modes. It presents derivations in a lengthy Supplementary Information, summarizes method assumptions and computation times in Table 1, and shows a side-by-side reconstruction comparison in Figure 2. The paper's stated central contribution is an "open-source software package" that unifies all major reconstruction methods to enable standardized benchmarking, and it concludes that hybrid methods—particularly the authors' hybrid formulation from Ref. 27—most accurately recover the true force profile.
Significance. If the software package were publicly available and the benchmark were quantitative and independently validated, this would be a genuinely useful contribution to dynamic AFM force spectroscopy, where methodological fragmentation and inconsistent validation are real problems. The literature review and the tabulated comparison of assumptions could also serve as a useful reference. As submitted, however, the central reproducibility claim is not verifiable because the code is not actually open, and the only performance comparison is a single idealized, noise-free, qualitative figure that favors a method developed by the corresponding author. The significance is therefore conditional on substantial revision and provision of the promised artifact and evidence.
major comments (3)
- [Data and Code Availability; Methods; Section 2.2.3] The paper's central claim, repeated in the Abstract, Section 2.2.3, and Conclusions, is that the authors "developed an open-source software package" for benchmarking force reconstruction methods. This claim is contradicted by the Data and Code Availability section, which states that the MATLAB codes "can be accessed from the authors" and that data are available "upon reasonable request." No repository URL, DOI, license, or version identifier is provided anywhere in the manuscript, and the Methods statement that the software interface is "available as supplementary material" is not accompanied by any actual supplementary file in the submitted text. A package that must be requested from the authors is not open source by any standard definition, and without a public, runnable artifact the reproducibility and standardization claims cannot be independently checked. This is a load-bearing issue and must be fixed before the paper can support its stated contribution.
- [Figure 2 and caption] The benchmark that is supposed to support the standardization claim is qualitative. The Figure 2 caption explicitly states that all reconstructions were "computed under ideal, noise-free conditions," and the comparison uses one assumed interaction force model; no quantitative error metric such as root-mean-square error, no confidence intervals, and no sensitivity analysis over noise level, oscillation amplitude, or force-law family are reported. Despite this, the caption asserts that the hybrid methods "more accurately capture the true force profile with minimal deviation." That assertion is not supported by the evidence shown. For a paper whose contribution is benchmarking and standardization, the comparison needs quantitative accuracy metrics, multiple synthetic and experimental datasets, and clearly specified parameters of the input force model, amplitude, and discretization.
- [Table 1; Section 2.2.3] The claimed superiority of the hybrid method is evaluated against the authors' own prior formulation (Ref. 27, by the corresponding author), without any independent implementation or external validation. This would be acceptable if the benchmark were objective and quantitative, but as presented the comparison is self-referential: the hybrid method is both the recommended approach and the basis for the qualitative comparison. The Discussion itself concedes that noise, amplitude variation, and deviations from the assumed force model degrade accuracy, which further undercuts the generality of the conclusion that hybrid methods are most accurate. An independent benchmark, or at minimum a blinded comparison with code from other groups, is needed to support the central claim.
minor comments (3)
- [Table 1] The computation times in Table 1 (e.g., "A few hours") are reported without specifying hardware, matrix size, Chebyshev order, or data length; please provide these conditions so that the times are meaningful.
- [Figure 2] The panels in Figure 2 lack visible axis labels and units, making it difficult to assess the force scale and distance range; please add labeled axes and state the parameters of the model force used for the ground truth.
- [Data and Code Availability] The phrase "open-source" should not be used until a repository with a software license, version tag, and usage instructions is provided; until then, the availability statement should match what is actually delivered.
Circularity Check
No significant circularity: the reconstruction derivations are self-contained, and the qualitative benchmark favoring the authors' hybrid method is not a definitional reduction.
full rationale
The paper's mathematical content is a systematic derivation and comparison of established force-reconstruction formulas. Each method is derived from physical input (Hamilton's principle, perturbation theory, Laplace transforms), and the inversion equations are not defined in terms of the target force: for example, the SI derives an integral kernel relating Δf to the tip-sample force and then inverts that fixed kernel. No equation sets a predicted force equal to a fitted parameter or to the data-generation model by construction. The hybrid method is taken from Ref. 27 as prior work and evaluated in Fig. 2; the claim that the hybrid methods 'more accurately capture the true force profile' is a benchmark outcome, not a derivation, and the benchmark is explicitly synthetic and noise-free, so it may be weak evidence but it is not circular. No uniqueness theorem is imported from the authors, and no ansatz is smuggled in via citation: the Bessel replacement is attributed to prior work and implemented for comparison. The availability statement ('The suite of MATLAB codes developed for this study can be accessed from the authors') conflicts with the 'open-source' label, and the lack of a repository, error metrics, and experimental validation are serious reproducibility concerns, but those are correctness/openness defects rather than circularity. Under the rule that circularity requires a specific equation-to-equation or fit-to-prediction reduction, none is exhibited, so the score is 0.
Assumptions & free parameters
free parameters (1)
- Padé approximant coefficient a =
1/8
assumptions (4)
- standard math Classical mechanics framework (Hamilton's principle, canonical transformations) applies to the cantilever problem.
- domain assumption Cantilever motion is harmonic with constant amplitude A.
- domain assumption First-order perturbation theory for the frequency shift is valid.
- domain assumption The tip-sample interaction force can be treated as conservative in the main derivations.
Cite this review
Pith. "Pith review of Standardizing Force Reconstruction in Dynamic Atomic Force Microscopy." pith.science (2026). https://pith.science/paper/4BJN2DVI
@misc{pith2026250606467,
author = {Pith},
title = {Pith review of: Standardizing Force Reconstruction in Dynamic Atomic Force Microscopy},
year = {2026},
howpublished = {\url{https://pith.science/paper/4BJN2DVI}},
note = {Machine review of arXiv:2506.06467}
}
read the original abstract
Atomic force microscopy (AFM) enables high-resolution imaging and quantitative force measurement, which is critical for understanding nanoscale mechanical, chemical, and biological interactions. In dynamic AFM modes, however, interaction forces are not directly measured; they must be mathematically reconstructed from observables such as amplitude, phase, or frequency shift. Many reconstruction techniques have been proposed over the last two decades, but they rely on different assumptions and have been applied inconsistently, limiting reproducibility and cross-study comparison. Here, we systematically evaluate major force reconstruction methods in both frequency- and amplitude-modulation AFM, detailing their theoretical foundations, performance regimes, and sources of error. To support benchmarking and reproducibility, we introduce an open-source software package that unifies all widely used methods, enabling side-by-side comparisons across different formulations. This work represents a critical step toward achieving consistent and interpretable AFM force spectroscopy, thereby supporting the more reliable application of AFM in fields ranging from materials science to biophysics.
Figures
Reference graph
Works this paper leans on
-
[1]
S18 Calculation of the Df using perturbation theory .................................................................................... S18 2.2 Analytical calculation of the Df for tip-sample interaction forces with the approximation of inverse power laws .......................................................................................................
work page 1997
-
[2]
"! "" " # $%&' ( ' ) ( &*+' ( ' ) ( ! !! !!
and t0 = 1/4fd, the equation for A can be simplified. 4.63 4.64 4.65 If g is small then Qeff » Q0 4.66 4.67 !!!! ""! "" " # $%&' ( ' ) ( &*+' ( ' ) ( ! !! !! "# AA %& ' ! # % & ' ! #A( A +! = "# " #!+ $ ! + ! $ +%& % &'( ' ( ! !! " !" ! #$%& ' & ( ' )*%& ' + ,-#& ' & ( ' ! ! ! ! " #A % ! &'" " #A % ! &" ! + !" + #= !+ " ! !!! ! "" " " ###$ % & ! !! "## A%...
work page 1997
Reviewed August 7, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.