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

Using molecular dynamics simulations with a machine-learned potential, this study reveals that the fracture energy of silica glass increases by up to 33% below the branching threshold due to a combination of rising intrinsic surface energy density and nanoscale roughening, demonstrating that dynamic fracture creates a fundamentally different surface structure rather than merely increasing apparent surface area.

The Gist

Imagine you are trying to snap a piece of thick glass. You might think that the energy required to break it is simply the energy needed to sever the tiny atomic bonds holding it together, like cutting a single strand of spaghetti. However, scientists have long known that breaking glass actually requires much more energy than that simple calculation suggests. It's as if the glass is fighting back, demanding extra effort to break.

For years, researchers believed this "extra cost" was mostly due to the crack getting wobbly and jagged as it sped up, creating a rougher surface area (like tearing a piece of paper into a jagged strip rather than a straight line). But a new study using advanced computer simulations has revealed a more complex story.

Here is what the paper discovered, explained simply:

1. The "Super-Heated" Crack Tip

When a crack moves very fast through glass, the tip of that crack gets incredibly hot. The study found that at high speeds, the atoms right at the tip of the crack reach temperatures of 8,000 Kelvin (hotter than the surface of the sun!).

Think of the crack tip not just as a breaking point, but as a tiny, microscopic blowtorch. This intense heat doesn't just melt the glass; it fundamentally changes the nature of the surface being created.

2. Two Reasons Glass Costs More to Break

The researchers used a super-accurate computer model (like a digital microscope that sees individual atoms) to figure out where all that extra energy goes. They found the "extra cost" is split roughly 50/50 between two things:

• The "Roughness" Factor (Quantity): As the crack speeds up, the surface it leaves behind isn't perfectly smooth. It becomes nano-scopically rough, like a mountain range seen from space. This means the crack actually creates more surface area than it appears to from the outside.
  • Analogy: Imagine tearing a piece of fabric. If you tear it slowly, the edge is straight. If you tear it fast, the edge becomes frayed and jagged. You've used more fabric to make that jagged edge.
• The "Quality" Factor (Energy Density): This is the new discovery. Even if you smoothed out that jagged surface, it would still cost more energy to create than a calm, slow surface. The extreme heat at the crack tip changes the atomic structure of the new surface, making it "higher energy" or more unstable.
  • Analogy: Imagine baking a cake. A slow-cooked cake has a standard texture. But if you blast it with a blowtorch, the outside becomes charred and chemically different. The "charred" surface is fundamentally different and requires more energy to create than the smooth, slow-cooked version.

3. The "Hidden" Roughness

One of the most interesting points is that the "roughness" the computer found is so tiny (at the scale of atoms) that standard tools used by engineers to measure broken glass would miss it completely.

If you looked at a broken piece of glass with a normal microscope, you would see a smooth surface. You would assume all the extra energy went into making the surface "hotter" or more energetic. But this study shows that a significant chunk of that energy actually went into making the surface physically larger and rougher, just at a scale too small for our eyes to see.

4. Fixing the Math on How Fast Cracks Move

The paper also corrected a long-standing formula used to predict how fast a crack moves based on the force applied. The old formula (the "Freund model") was like a map that got a little fuzzy at high speeds. The new study found a better formula (a "square-root relationship") that fits the data perfectly.

This correction is important because it helps explain why previous experiments measuring the heat of breaking glass (using light emitted by the crack, called fractoluminescence) didn't quite match the speed predictions. By using the new formula, the predicted speeds and temperatures finally line up with what the computer simulations showed.

The Bottom Line

Breaking glass isn't just about snapping bonds. When the crack moves fast, it acts like a tiny, super-heated laser that:

1. Makes the surface physically rougher (creating more area).
2. Chemically alters the surface to make it more energetic.

The study proves that the energy required to break glass isn't a fixed number; it changes depending on how fast you break it, and it's driven by both the shape of the break and the extreme heat at the tip.

Technical Summary

Problem Statement
The fracture energy of brittle materials, such as silica glass, is observed to increase with crack velocity, a phenomenon that exceeds the thermodynamic cost of bond rupture. While this velocity dependence is often attributed to surface roughening caused by dynamic instabilities (such as microbranching), the behavior below the branching threshold remains poorly understood. Previous studies on polymers (e.g., PMMA) noted a ~30% increase in fracture energy before branching onset, attributing it to unresolved dissipation processes. A central question persists: does this pre-branching energy increase arise from an increase in the quantity of surface area (via nanoscale roughening invisible to standard measurements) or an increase in the quality (intrinsic energy density) of the created surface due to extreme conditions at the crack tip? Furthermore, empirical interatomic potentials used in previous molecular dynamics (MD) studies of silica have failed to accurately capture high-energy bond-breaking events, leading to underestimated branching speeds and inaccurate process zone descriptions.

Methodology
To address these limitations, the authors employed extensive first-principles molecular dynamics simulations of fused silica glass.

• Interatomic Potential: Instead of empirical potentials, the study utilized a Machine Learned Interatomic Potential (MLIP) based on the Allegro architecture. This model was trained on Density Functional Theory (DFT) data using the r2SCAN functional, covering a vast range of states including molten silica, glassy structures, and mechanical deformation up to 15,000 K.
• Simulation Setup: The system consisted of a 12.3 million-atom silica glass strip (300 × 8.5 × 75 nm³) with a pre-existing notch. Simulations were conducted at 300 K and 1 bar, followed by rapid extensional displacement to induce crack propagation.
• Parameters: Fourteen simulations were run with varying displacement rates, resulting in initial tensile stresses between 3.9 GPa and 7 GPa.
• Analysis: The authors distinguished between "structural fracture energy" (energy stored in the damaged structure after cooling, normalized by projected area) and "intrinsic surface energy density" (energy normalized by the real, rough surface area measured via Delaunay tessellation). They also analyzed crack tip temperatures by averaging kinetic energy over 0.5 ps intervals.

Key Results

1. Velocity-Dependent Fracture Energy: Even below the branching threshold (defined at 0.72 of the Rayleigh wave speed, $v_R$), the structural fracture energy rises by up to 33% as crack velocity increases.
2. Decomposition of Energy Increase: This rise is partitioned roughly equally between two mechanisms:
   • Nanoscale Roughening: A ~16% increase in the real fracture surface area ($A_{real}$) compared to the projected area. This roughening creates a thicker damage zone around the crack but occurs at length scales invisible to standard post-mortem optical analysis.
   • Intrinsic Surface Energy Density: A ~15% increase in the intrinsic surface energy density ($\gamma_s$). This is attributed to the extreme thermal conditions at the crack tip, where temperatures exceed 3,000 K within a process zone of only 3–6 nm.
3. Crack Velocity and Instability: The simulations reveal that crack velocity follows a square-root relationship with the stress intensity factor ($v_c \propto \sqrt{1 - (K_{Ic}/K_I)^2}$), differing from the standard Freund model. Branching occurs at approximately 2.4 km/s (0.72 $v_R$), consistent with experimental observations in fused silica.
4. Process Zone Size: The process zone radius increases with stress intensity, ranging from 3.1 nm for planar cracks to ~6.8 nm for branching cracks. This size aligns with Dugdale–Barenblatt predictions and is smaller than values derived from previous empirical potential simulations (10 nm).
5. Thermal Discrepancy Resolution: The study resolves a discrepancy between simulation temperatures and fractoluminescence-based experimental predictions. By applying the newly identified square-root velocity–stress relationship, the predicted crack tip temperatures for a given velocity are lower, bringing analytical models into closer agreement with the simulation data.

Significance and Claims
The paper claims that dynamic fracture in silica glass increases fracture energy not merely by creating more apparent surface area, but by fundamentally altering the surface at the nanoscale. The authors demonstrate that:

• Fracture energy is not a constant material property; it is velocity-dependent even in the absence of macroscopic branching.
• Standard post-mortem analysis would conflate the contributions of nanoscale roughening and increased intrinsic energy density, interpreting the entire increase as a rise in surface energy density.
• First-principles MD simulations using MLIPs can now achieve quantitative accuracy in modeling dynamic fracture without empirical tuning, successfully reproducing elastic properties, fracture toughness, and branching thresholds that match experimental data.
• The revised velocity–stress relationship provides a more accurate framework for interpreting crack tip temperatures inferred from fractoluminescence experiments.

The study concludes that the pre-branching increase in fracture energy, previously observed but unexplained in materials like PMMA, arises from a combination of nanoscale roughening and elevated intrinsic surface energy density driven by extreme local temperatures.