pith. sign in

arxiv: 2605.03457 · v1 · submitted 2026-05-05 · ❄️ cond-mat.mtrl-sci · cond-mat.mes-hall· physics.comp-ph

Energy dissipation at the atomic scale explains how fracture energy depends on crack velocity in silica glass

Marthe Gr{\o}nlie Guren , Sigbj{\o}rn L{\o}land Bore , Fran\c{c}ois Renard , Henrik Andersen Sveinsson This is my paper

Pith reviewed 2026-05-07 16:10 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci cond-mat.mes-hallphysics.comp-ph
keywords silica glassfracture energycrack velocitymolecular dynamicssurface rougheningbrittle materialsatomic scale
0
0 comments X

The pith

Simulations show that fracture energy in silica glass increases with crack velocity due to atomic-scale effects even below the branching threshold.

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

Using molecular dynamics simulations, the paper establishes that the structural fracture energy of silica glass increases by as much as 33 percent as crack velocity rises, and this occurs already at speeds below where the crack starts to branch. The increase splits roughly equally between a higher intrinsic surface energy density required to create the surface and greater actual surface area caused by nanoscale roughening. This means fracture energy cannot be viewed as a fixed property of the material but depends on the dynamics of the fracture process. The work reveals that faster cracks produce a fundamentally different surface at the atomic scale rather than simply more of the same surface.

Core claim

Dynamic fracture in silica glass increases the structural fracture energy by up to 33% with rising crack velocity below the branching threshold. This rise partitions equally between increased intrinsic surface energy density and nanoscale roughening that enlarges the real fracture surface area. The results show that fracture energy is velocity-dependent and that the created surface differs at the nanoscale from static fracture surfaces.

What carries the argument

Molecular dynamics simulations employing a first-principles machine-learned interatomic potential to compute energy dissipation and surface creation at the atomic scale during crack propagation.

If this is right

  • Fracture energy is shown to be a velocity-dependent quantity rather than a constant material property.
  • The contribution from intrinsic surface energy density changes is comparable to that from increased surface area.
  • Dynamic fracture produces a nanoscale surface that is different in character from the surface formed in slow fracture.
  • The effect appears before macroscopic instabilities like branching occur.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • This velocity dependence may require updates to continuum models of brittle fracture to include rate effects at low speeds.
  • Similar atomic-scale mechanisms could be investigated in other amorphous brittle materials.
  • Experimental techniques sensitive to surface energy at the nanoscale could directly test these simulation results.

Load-bearing premise

The machine-learned interatomic potential accurately models the atomic bond breaking and energy dissipation processes that occur during fast fracture in actual silica glass.

What would settle it

An experiment that measures the fracture energy of silica glass at several crack velocities below the branching threshold and finds no increase would falsify the central claim.

Figures

Figures reproduced from arXiv: 2605.03457 by Fran\c{c}ois Renard, Henrik Andersen Sveinsson, Marthe Gr{\o}nlie Guren, Sigbj{\o}rn L{\o}land Bore.

Figure 2
Figure 2. Figure 2: FIG. 2. Post-mortem fracture morphology for four represen view at source ↗
Figure 1
Figure 1. Figure 1: FIG. 1. Simulation setup overview and crack process zone de view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Crack velocities and fracture surface energies. (a) view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Thermal response and reconciliation with fractolumi view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Calculation of Young’s modulus view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. Snapshot of the crack tip where the atoms are colored view at source ↗
read the original abstract

The fracture energy of brittle materials rises with crack velocity, and this effect is typically attributed to surface roughening from path instabilities. Here we show, using molecular dynamics simulations of silica glass with a first-principles machine learned interatomic potential, that the structural fracture energy rises by up to 33 % already below the branching threshold, showing that fracture energy is not a constant material property. This rise in fracture energy is roughly equally partitioned between an increase in the intrinsic surface energy density and nanoscale roughening that increases the real fracture surface area. Results demonstrate that dynamic fracture in silica glass increases the fracture energy not merely by creating more apparent surface, but also by creating a fundamentally different surface at the nanoscale.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript uses molecular dynamics simulations of silica glass with a first-principles machine-learned interatomic potential to demonstrate that structural fracture energy increases by up to 33% with crack velocity below the branching threshold. This rise is reported as roughly equally partitioned between an increase in intrinsic surface energy density and nanoscale roughening that augments the real fracture surface area, implying that fracture energy is velocity-dependent rather than a constant material property and that dynamic fracture produces a fundamentally different nanoscale surface.

Significance. If the simulations accurately capture atomic-scale dissipation, the work supplies a mechanistic explanation for velocity-dependent fracture energy in brittle glasses that goes beyond path instabilities and roughening. The direct computation of energy differences between static and dynamic configurations, with no fitted parameters, is a strength that allows falsifiable comparison to experiment once the potential is validated. This could refine continuum models of dynamic fracture and guide interpretation of experimental measurements that show rising apparent toughness with speed.

major comments (3)
  1. [Methods] Methods section on potential training and validation: the manuscript provides no information on training-set coverage of high-strain-rate bond-rupture configurations or anharmonic dissipation channels. Because the 33% rise and equal partition are obtained exclusively from MD trajectories, absence of such validation leaves open the possibility that systematic bias in the potential rescale or eliminate the reported velocity dependence.
  2. [Results] Results, energy-partition analysis: the operational definition used to separate 'intrinsic surface energy density' from the geometric contribution of increased real area is not stated explicitly. Without a precise, reproducible criterion (e.g., per-atom energy cutoff or surface identification algorithm), the claim that the two contributions are 'roughly equal' cannot be independently verified or falsified.
  3. [Results] Results, system-size and timestep convergence: no data are shown on how the reported 33% increase converges with simulation cell size, timestep, or thermostat parameters. Given that the central claim rests on quantitative energy differences extracted from finite MD runs, lack of convergence tests undermines that the effect is physical rather than numerical.
minor comments (2)
  1. [Abstract] The abstract states the partition is 'roughly equal' while the main text should quantify the fractions with error bars or ranges to allow readers to assess the precision of the equality claim.
  2. [Figures] Figure captions should explicitly define how 'structural fracture energy' is computed from the MD trajectories (e.g., total work minus elastic strain energy) so that the quantity is reproducible from the published data.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the positive assessment of our work and the constructive comments, which have helped us identify areas for improvement. We address each major comment below and will revise the manuscript to enhance clarity, reproducibility, and confidence in the results.

read point-by-point responses

  1. Referee: [Methods] Methods section on potential training and validation: the manuscript provides no information on training-set coverage of high-strain-rate bond-rupture configurations or anharmonic dissipation channels. Because the 33% rise and equal partition are obtained exclusively from MD trajectories, absence of such validation leaves open the possibility that systematic bias in the potential rescale or eliminate the reported velocity dependence.

    Authors: We agree that explicit details on the training-set coverage for high-strain-rate and anharmonic configurations are necessary to fully address potential biases. The machine-learned potential was trained on a dataset that incorporates ab initio molecular dynamics trajectories spanning a range of strain rates and bond-distortion regimes, including those relevant to dynamic fracture. To make this transparent, we will expand the Methods section with a dedicated paragraph describing the composition of the training set, the inclusion of high-strain-rate bond-rupture snapshots, and validation against anharmonic dissipation metrics. This revision will allow readers to evaluate the potential's applicability to the reported velocity dependence. revision: yes

  2. Referee: [Results] Results, energy-partition analysis: the operational definition used to separate 'intrinsic surface energy density' from the geometric contribution of increased real area is not stated explicitly. Without a precise, reproducible criterion (e.g., per-atom energy cutoff or surface identification algorithm), the claim that the two contributions are 'roughly equal' cannot be independently verified or falsified.

    Authors: We acknowledge that the manuscript did not provide an explicit, reproducible operational definition for the energy partition. The intrinsic surface energy density is obtained by summing the excess potential energy of atoms classified as surface atoms (using a coordination-number threshold of less than 3 for silicon and less than 2 for oxygen), while the geometric contribution is calculated from the increase in actual surface area determined by a Delaunay triangulation of surface atoms. We will add a concise methods paragraph in the Results section that states these criteria verbatim, including the exact cutoffs and algorithms, so that the roughly equal partition can be independently reproduced and verified. revision: yes

  3. Referee: [Results] Results, system-size and timestep convergence: no data are shown on how the reported 33% increase converges with simulation cell size, timestep, or thermostat parameters. Given that the central claim rests on quantitative energy differences extracted from finite MD runs, lack of convergence tests undermines that the effect is physical rather than numerical.

    Authors: We performed preliminary convergence tests during the study, confirming that the 33% rise is stable for the chosen parameters, but these tests were omitted from the original submission. We will add a new subsection (or supplementary figure) presenting convergence data for system sizes from 8 nm to 40 nm in the crack-propagation direction, timesteps between 0.25 fs and 1 fs, and two different thermostat damping times. The results show that the velocity-dependent increase remains within 2% across this range, supporting that the effect is physical. This addition will directly address the concern about numerical artifacts. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper's central claims derive from direct MD simulation outputs: energy differences between static and dynamic fracture configurations, real surface area computed from nanoscale topography, and the resulting partition of the observed 33% rise in structural fracture energy. No equations or definitions reduce the velocity-dependent result to a fitted parameter or self-referential input by construction. The machine-learned potential is an external tool whose training is not invoked as a 'prediction' within the fracture analysis itself. This matches the default expectation for simulation-based papers that remain self-contained against their own computed quantities.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The claim rests on the accuracy of the machine-learned potential for dynamic fracture and on the operational definition of structural fracture energy extracted from the simulation trajectories.

axioms (1)
  • domain assumption The machine-learned interatomic potential trained on first-principles data faithfully reproduces silica glass fracture behavior at the atomic scale under dynamic conditions.
    All reported energy values derive from trajectories generated with this potential.

pith-pipeline@v0.9.0 · 5450 in / 1216 out tokens · 65842 ms · 2026-05-07T16:10:34.055069+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages

  1. [1]

    S. M. Wiederhorn, Fracture surface energy of glass, Jour- nal of the American Ceramic Society52, 99 (1969)

    work page 1969

  2. [2]

    Weichert and K

    R. Weichert and K. Sch¨ onert, Heat generation at the tip of a moving crack, Journal of the Mechanics and Physics of Solids26, 151 (1978)

    work page 1978

  3. [3]

    Fineberg, S

    J. Fineberg, S. P. Gross, M. Marder, and H. L. Swinney, Instability in dynamic fracture, Physical Review Letters 67, 457 (1991)

    work page 1991

  4. [4]

    M. J. Buehler and H. Gao, Dynamical fracture instabili- ties due to local hyperelasticity at crack tips, Nature439, 307 (2006)

    work page 2006

  5. [5]

    Sharon, S

    E. Sharon, S. P. Gross, and J. Fineberg, Energy Dissipa- tion in Dynamic Fracture, Physical Review Letters76, 2117 (1996)

    work page 1996

  6. [6]

    Vashishta, R

    P. Vashishta, R. K. Kalia, J. P. Rino, and I. Ebbsj¨ o, Interaction potential for SiO2 : A molecular-dynamics study of structural correlations, Physical Review B41, 12197 (1990)

    work page 1990

  7. [7]

    Rountree, D

    C. Rountree, D. Bonamy, D. Dalmas, S. Prades, R. Kalia, C. Guillot, and E. Bouchaud, Fracture in glass via molec- ular dynamics simulations and atomic force microscopy experiments, Physics and Chemistry of Glasses-European Journal of Glass Science and Technology Part B51, 127 (2010)

    work page 2010

  8. [8]

    M. G. Guren, H. A. Sveinsson, A. Malthe-Sørenssen, and F. Renard, Nanoscale damage production by dynamic tensile rupture inα-quartz, Geophysical Research Let- ters49, e2022GL100468 (2022)

    work page 2022

  9. [9]

    Behler and M

    J. Behler and M. Parrinello, Generalized Neural-Network Representation of High-Dimensional Potential-Energy Surfaces, Physical Review Letters98, 146401 (2007)

    work page 2007

  10. [10]

    J. Zeng, D. Zhang, D. Lu, P. Mo, Z. Li, Y. Chen, M. Rynik, L. Huang, Z. Li, S. Shi, Y. Wang, H. Ye, P. Tuo, J. Yang, Y. Ding, Y. Li, D. Tisi, Q. Zeng, H. Bao, Y. Xia, J. Huang, K. Muraoka, Y. Wang, J. Chang, F. Yuan, S. L. Bore, C. Cai, Y. Lin, B. Wang, J. Xu, J.- X. Zhu, C. Luo, Y. Zhang, R. E. A. Goodall, W. Liang, A. K. Singh, S. Yao, J. Zhang, R. Went...

    work page 2023

  11. [11]

    Batatia, D

    I. Batatia, D. P. Kovacs, G. Simm, C. Ortner, and G. Csanyi, Mace: Higher order equivariant message pass- ing neural networks for fast and accurate force fields, inAdvances in Neural Information Processing Systems, Vol. 35, edited by S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh (Curran Associates, Inc.,

    work page

  12. [12]

    Musaelian, S

    A. Musaelian, S. Batzner, A. Johansson, L. Sun, C. J. Owen, M. Kornbluth, and B. Kozinsky, Learning local equivariant representations for large-scale atomistic dy- namics, Nature Communications14, 579 (2023)

    work page 2023

  13. [13]

    L. C. Erhard, J. Rohrer, K. Albe, and V. L. Deringer, A machine-learned interatomic potential for silica and its relation to empirical models, npj Computational Materi- als8, 90 (2022)

    work page 2022

  14. [14]

    Kobayashi, M

    K. Kobayashi, M. Okumura, H. Nakamura, M. Itakura, M. Machida, S. Urata, and K. Suzuya, Machine learning molecular dynamics reveals the structural origin of the first sharp diffraction peak in high-density silica glasses, 9 Scientific Reports13, 18721 (2023)

    work page 2023

  15. [15]

    I. S. Novikov and A. V. Shapeev, Improving accuracy of interatomic potentials: more physics or more data? a case study of silica, Materials Today Communications 18, 74 (2019)

    work page 2019

  16. [16]

    Balyakin, S

    I. Balyakin, S. Rempel, R. Ryltsev, and A. Rempel, Deep machine learning interatomic potential for liquid silica, Physical Review E102, 052125 (2020)

    work page 2020

  17. [17]

    J. W. Furness, A. D. Kaplan, J. Ning, J. P. Perdew, and J. Sun, Accurate and Numerically Efficient r2 SCAN Meta-Generalized Gradient Approximation, The Journal of Physical Chemistry Letters11, 8208 (2020)

    work page 2020

  18. [18]

    M. J. Buehler, ed.,Atomistic Modeling of Materials Fail- ure(Springer US, Boston, MA, 2008)

    work page 2008

  19. [19]

    C. L. Rountree and W. Feng, Silica and its process zone, International Journal of Applied Glass Science11, 385 (2020)

    work page 2020

  20. [20]

    D. S. Dugdale, Yielding of steel sheets containing slits, Journal of the Mechanics and Physics of Solids8, 100 (1960)

    work page 1960

  21. [21]

    G. I. Barenblatt, The mathematical theory of equilibrium cracks in brittle fracture, Advances in applied mechanics 7, 55 (1962)

    work page 1962

  22. [22]

    L. B. Freund,Dynamic fracture mechanics(Cambridge university press, 1998)

    work page 1998

  23. [23]

    J. W. McCauley, E. Strassburger, P. Patel, B. Paliwal, and K. T. Ramesh, Experimental Observations on Dy- namic Response of Selected Transparent Armor Materi- als, Experimental Mechanics53, 3 (2013)

    work page 2013

  24. [24]

    Fineberg and M

    J. Fineberg and M. Marder, Instability in dynamic frac- ture, Physics Reports313, 1 (1999)

    work page 1999

  25. [25]

    Bouchbinder, T

    E. Bouchbinder, T. Goldman, and J. Fineberg, The dy- namics of rapid fracture: Instabilities, nonlinearities and length scales, Reports on Progress in Physics77, 046501 (2014)

    work page 2014

  26. [26]

    Liu and M

    X. Liu and M. Marder, The energy of a steady-state crack in a strip, Journal of the Mechanics and Physics of Solids 39, 947 (1991)

    work page 1991

  27. [27]

    Stukowski, Computational Analysis Methods in Atomistic Modeling of Crystals, JOM66, 399 (2014)

    A. Stukowski, Computational Analysis Methods in Atomistic Modeling of Crystals, JOM66, 399 (2014)

    work page 2014

  28. [28]

    A. Stukowski, Visualization and analysis of atomistic simulation data with ovito–the open visualization tool, Modelling and simulation in materials science and engi- neering18, 015012 (2009)

    work page 2009

  29. [29]

    J. M. Rimsza, R. E. Jones, and L. J. Criscenti, Crack propagation in silica from reactive classical molecular dy- namics simulations, Journal of the American Ceramic So- ciety101, 1488 (2018)

    work page 2018

  30. [30]

    Pallares, C

    G. Pallares, C. L. Rountree, L. Douillard, F. Charra, and E. Bouchaud, Fractoluminescence characterization of the energy dissipated during fast fracture of glass, Eu- rophysics Letters99, 28003 (2012)

    work page 2012

  31. [31]

    Kadono, K

    T. Kadono, K. Ogawa, K. Shirai, M. Arakawa, K. Kuro- sawa, T. Okamoto, T. Matsui, S. Hasegawa, A. I. Suzuki, and H. Kobayashi, Experimental investigation of visible- light and x-ray emissions during rock and mineral frac- ture: Role of electrons traveling between fracture sur- faces, Minerals12, 778 (2022)

    work page 2022

  32. [32]

    Zhang, S

    Z. Zhang, S. Ispas, and W. Kob, Fracture of silicate glasses: Microcavities and correlations between atomic- level properties, Physical Review Materials6, 085601 (2022)

    work page 2022

  33. [33]

    Sharon and J

    E. Sharon and J. Fineberg, Microbranching instability and the dynamic fracture of brittle materials, Physical Review B54, 7128 (1996)

    work page 1996

  34. [34]

    H. A. Sveinsson, Allegro machine learned interatomic po- tential for silica up to 15000 k, 10.5281/zenodo.17669954 (2025)

    work page doi:10.5281/zenodo.17669954 2025

  35. [35]

    H. M. Cezar, T. Bodenstein, H. A. Sveinsson, M. Ledum, S. Reine, and S. L. Bore, Learning atomic forces from uncertainty-calibrated adversarial attacks, npj Compu- tational Materials11, 200 (2025)

    work page 2025

  36. [36]

    Kresse and J

    G. Kresse and J. Furthm¨ uller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Computational Materials Science 6, 15 (1996)

    work page 1996

  37. [37]

    Kresse and J

    G. Kresse and J. Furthm¨ uller, Efficient iterative schemes forab initiototal-energy calculations using a plane-wave basis set, Physical Review B54, 11169 (1996)

    work page 1996

  38. [38]

    Deschamps, J

    T. Deschamps, J. Margueritat, C. Martinet, A. Mer- met, and B. Champagnon, Elastic Moduli of Perma- nently Densified Silica Glasses, Scientific Reports4, 7193 (2014)

    work page 2014