Asymmetric Energy Landscapes Control Diffusion in Glasses

This study establishes that asymmetric energy landscapes in glasses drive large macroscopic diffusion activation energies through dominant back-and-forth correlated atomic motions, rather than high local rearrangement barriers, providing a quantitative framework that links atomic-scale dynamics to macroscopic transport across various disordered materials.

The Gist

The Big Mystery: Why is Glass "Stuck"?

Imagine you are trying to walk through a crowded, chaotic dance floor (this is glass). In a crystal (like a diamond or salt), the dancers are lined up in perfect rows. If you want to move, you just step into the empty spot next to you. It's predictable and easy to calculate.

But in glass, the dancers are jumbled up randomly. You might think that because the crowd is messy, it would be harder to find a spot to move, or that the "energy" required to push through the crowd is huge.

The Puzzle: Scientists have known for a long time that atoms in glass do move locally. They wiggle, swap places, and rearrange themselves. These little moves require very little energy (like a gentle nudge). However, when we measure how fast atoms actually travel across the whole glass (diffusion), it is incredibly slow. It acts like the atoms are stuck in molasses, requiring a massive amount of energy to move even a tiny distance.

The Question: If the atoms can easily wiggle locally, why is the overall movement so incredibly slow? Where does that huge "energy cost" come from?

The Solution: The "Back-and-Forth" Trap

The authors of this paper solved the mystery by breaking the movement of atoms into two parts: The Steps and The Memory.

1. The Steps (Random Walk)

Imagine you are blindfolded and taking steps. If you just take steps in random directions, you will eventually get somewhere. This is the "Random Walk."

• In the paper: The scientists calculated how fast atoms would move if they just took their local wiggles and kept going in a straight line without looking back. They found that these local wiggles are actually quite easy and fast. The energy needed for just these steps is low.

2. The Memory (Correlation)

Now, imagine you are walking, but every time you take a step forward, you are magically forced to take a step backward immediately after. You are doing a lot of work, moving your legs, but you aren't going anywhere. You are just dancing in place.

• In the paper: This is the "Correlation." The atoms in glass are constantly rearranging, but they have a bad habit of undoing what they just did. They jump forward, hit a wall, and immediately jump back to where they started.

The Big Discovery: The reason glass diffuses so slowly isn't because the atoms can't move; it's because they keep reversing their moves. The "energy cost" of diffusion in glass is mostly the energy needed to break this cycle of going forward and immediately coming back.

The "Asymmetric Landscape" Analogy

Why do the atoms keep going back? The paper explains this using the shape of the "energy landscape."

Imagine a hiker trying to cross a mountain range.

• In a Crystal: The mountains are like perfect, symmetrical steps. Going up is the same as coming down.
• In Glass: The landscape is a jagged, messy canyon.
  • To jump forward (from one spot to the next), the atom has to climb a steep, high hill.
  • But once it gets to the top and falls into the new spot, the path back to the old spot is a gentle, easy slope.

Because the path back is so much easier than the path forward, the atom is statistically likely to slide back down the easy slope rather than climb the steep hill again. It gets "trapped" in a loop of going up the hard way and sliding down the easy way, canceling out its progress.

What About Cooling Rates?

The paper also looked at how fast the glass was made.

• Fast Cooling (Quenching): Like freezing water instantly. The atoms get stuck in a very messy, jumbled state. The "hills" and "valleys" are very uneven. The atoms get stuck in deep loops, reversing their moves constantly.
• Slow Cooling: Like letting water freeze slowly. The atoms have time to settle into slightly better positions. The "hills" become less steep, and the "back-and-forth" trap is less effective.

The Result: The slower you cool the glass, the more "stuck" the atoms become, not because they can't move, but because they are better at undoing their own moves.

The Surface Surprise

Finally, the team looked at the surface of the glass (the outside edge).

• Old Theory: People thought atoms on the surface move faster because the "walls" holding them back are lower (like standing on a cliff edge is easier than being in a deep valley).
• New Theory: The paper found that the local "walls" are actually only slightly lower. The real reason surface atoms move faster is that they don't bounce back as much. The surface is less crowded and less "jagged," so when an atom jumps, it's less likely to get forced back to its starting spot. It gets to keep its forward progress.

Summary: The Takeaway

1. Glass isn't slow because atoms are lazy. They are actually very active, constantly wiggling and rearranging.
2. Glass is slow because atoms are indecisive. They spend most of their time jumping forward and immediately jumping back, canceling out their progress.
3. The "Energy Cost" is a trap. The huge energy we measure for diffusion isn't the cost of moving; it's the cost of escaping the tendency to return to where you started.
4. It's about structure, not chemistry. This happens because of the messy shape of the glass, not because of what the glass is made of. Even a glass made of just one type of atom behaves this way.

In a nutshell: Diffusion in glass isn't a struggle to climb a mountain; it's a struggle to stop walking in circles.

Technical Summary

1. Problem Statement

Diffusion in crystalline solids is well-understood through defect-mediated atomic hops, where the macroscopic diffusion coefficient ($D$) is directly linked to the migration barrier ($E_m$) via Transition State Theory. However, a comparable quantitative framework is missing for glasses (amorphous solids).

• The Paradox: Experiments and simulations show that diffusion in glasses follows an Arrhenius temperature dependence with large activation energies ($E_D \approx 1\text{--}3$ eV). Conversely, local atomic rearrangements (known as $\beta$-relaxations) involve much smaller energy barriers ($\approx 0.3\text{--}0.5$ eV).
• The Question: Why is the macroscopic activation energy so large if the local barriers are small? Is it determined solely by local barriers, or do other features of the disordered structure (e.g., correlated motion, defect-like excitations) play a dominant role?

2. Methodology

The authors utilized Molecular Dynamics (MD) simulations on several model glass systems to decompose the diffusion process.

• Systems Studied:
  • Primary system: Binary metallic glass Cu$_{50}$Zr$_{50}$.
  • Validation systems: Ni$_{80}$P$_{20}$ (metallic), SiO$_2$ (oxide), and a single-component Lennard-Jones (LJ) glass.
  • Surface diffusion analysis on Cu$_{50}$Zr$_{50}$.
• Decomposition Framework: The authors introduced a quantitative decomposition of the diffusion coefficient:
  $$D = D_{RW} \times f$$
  • $D_{RW}$ (Random-Walk Contribution): Represents the intrinsic propensity for local rearrangements. It is calculated by reconstructing a "random-walk" trajectory from MD data. In this reconstruction, atomic displacements exceeding a threshold ($d = 1$ Å) are treated as barrier-crossing events, but their directions are randomized to remove memory effects.
  • $f$ (Correlation Factor): Defined as $f = D / D_{RW}$. It measures the efficiency of rearrangements in producing net displacement. Values $f \ll 1$ indicate strong temporal correlations (back-and-forth motion).
• Energy Landscape Analysis:
  • Nudged Elastic Band (NEB): Used to map the distribution of local rearrangement barriers ($E_{act}$) for events identified in MD.
  • Activation-Relaxation Technique (ART): Used to sample forward ($E_{fw}$) and reverse ($E_{rev}$) activation barriers to analyze the asymmetry of the potential energy landscape (PEL).
• Cooling Rate Variation: Simulations were performed with cooling rates ranging from $10^{12}$ K/s (MD typical) down to $5 \times 10^8$ K/s to extrapolate behavior toward experimentally relevant conditions.

3. Key Results

A. Decomposition of Activation Energy

The total diffusion activation energy ($E_D$) is the sum of the random-walk barrier ($E_{RW}$) and the correlation barrier ($E_f$):
$$E_D = E_{RW} + E_f$$

• Findings in Cu$_{50}$Zr$_{50}$:
  • $E_D \approx 1.22$ eV.
  • $E_{RW} \approx 0.43$ eV (consistent with the low local barriers observed in $\beta$-relaxations).
  • $E_f \approx 0.79$ eV.
• Conclusion: The correlation barrier accounts for ~65% of the total activation energy. The large macroscopic barrier is not due to high local barriers, but due to the strong tendency of atoms to reverse their motion.

B. Origin of Correlations: Asymmetric Energy Landscapes

• Mechanism: In disordered systems, neighboring energy minima are connected by asymmetric barriers. The forward barrier ($E_{fw}$) is often significantly higher than the reverse barrier ($E_{rev}$).
• Consequence: Atoms that successfully cross a high forward barrier often encounter a low reverse barrier, causing them to "retrace their steps" rather than escape to a new region. This creates a "memory effect" where motion is reversible and cancels out, suppressing net diffusion.
• Generality: This mechanism is structural, not chemical. It was observed in SiO$_2$ and single-component LJ glasses, proving that chemical complexity is not required for this effect.

C. Cooling Rate Dependence

• Slower cooling rates (closer to experimental conditions) lead to more relaxed glasses with higher forward barriers but similar reverse barriers.
• This increases the asymmetry of the energy landscape.
• Result: As cooling rate decreases, $E_f$ increases significantly, while $E_{RW}$ remains nearly constant.
• Extrapolation: For experimentally relevant cooling rates ($10^1\text{--}10^5$ K/s), the correlation factor $f$ is predicted to be extremely small ($\approx 10^{-7}$), meaning correlations dominate diffusion even more strongly than in MD simulations.

D. Surface Diffusion

• Surface diffusion is faster than bulk diffusion with lower activation energy.
• Mechanism: The reduction in activation energy at the surface is not primarily due to lower local rearrangement barriers ($E_{RW,s} \approx 0.35$ eV vs. bulk $0.43$ eV).
• Primary Driver: It is driven by a reduction in the correlation barrier ($E_{f,s} \approx 0.44$ eV vs. bulk $0.79$ eV). The altered surface topology reduces the tendency for atoms to reverse their steps, making surface motion more irreversible.

4. Significance and Impact

• Resolves the Energy Paradox: The paper provides a unified explanation for why glasses exhibit large diffusion activation energies despite low local rearrangement barriers. The "cost" of diffusion is the energy required to break the reversibility of atomic motion, not just to overcome local barriers.
• New Paradigm for Glass Kinetics: It shifts the focus from local barrier heights to the topology of the energy landscape (specifically forward-reverse asymmetry) as the controlling factor for transport.
• Predictive Framework: The decomposition $D = D_{RW} \times f$ offers a predictive basis for understanding aging, stress relaxation, and crystallization in glasses.
• Materials Design Implications:
  • Fast Ion Conductors: Designing glasses with reduced forward-reverse asymmetry could enhance diffusion.
  • Radiation Tolerance: High-entropy metallic glasses with strong correlations might suppress diffusion, aiding in defect recombination and self-healing.
  • Ultrastable Glasses: The mechanism explains why surface diffusion is enhanced, facilitating the formation of ultrastable glasses via physical vapor deposition.

In summary, the authors establish that asymmetric energy landscapes in disordered materials drive strong, temperature-dependent correlations that suppress net diffusion, fundamentally altering our understanding of transport in amorphous solids.