When the Wires Inside a Battery Start to Fray: Modelling Conductivity Loss Under Vibration

Picture an old string of festive lights. Each bulb works, but one loose contact makes the current search for another path, and a few more loose contacts leave the whole string dim. The electrode of a lithium-ion cell behaves in a similar way. Its particles are held together by a carbon-binder network that carries electrons, and vibration slowly loosens that network. A recent vibration study captures this fraying with a single damage variable that lets the model track how the electrode’s electrical backbone weakens over time.

The network that carries the electrons

Active material particles do not conduct electrons well enough on their own. Carbon additives and polymer binder form a connected web, the carbon-binder domain (CBD), which links every particle to the current collector. If enough links break, particles become electrically isolated. They may still hold lithium, but the cell can no longer use it.

Repeated mechanical loading attacks exactly this web. Binder fatigues, contacts slip, and conductive pathways are lost one by one.

One number for the health of the network

In a healthy electrode, effective electronic conductivity depends on the intrinsic conductivity of the material, the solid volume fraction, and a Bruggeman exponent describing how tortuous the solid pathways are. The authors multiply this by a degradation factor:

σ_eff = σ₀ · ε_s^b · (1 − D)^m

Here D is the damage variable, running from 0 (fully intact network) to 1 (complete electronic disconnection). The exponent m, typically between 1 and 4, controls how sharply conductivity collapses as damage grows. It plays the role of a percolation sensitivity: the more the electrode depends on a few critical pathways, the larger m should be.

A quick illustration, using my own numbers rather than the paper’s, shows why the exponent matters. With m = 2, a damage level of 0.2 leaves 64% of the original conductivity, while a damage level of 0.5 leaves only 25%. Damage that looks moderate on paper can already cut conduction severely.

How damage grows

Damage is not imposed. It evolves according to the mechanical energy the electrode absorbs. The model accumulates a cumulative vibration energy density, E_v, and compares it with a critical energy density, E_c:

∂D/∂t = k_d · (E_v / E_c)^n · (1 − D)

Three features of this law are worth noticing:

  • Energy-driven: stronger or longer vibration pumps in more energy, so damage accelerates.
  • Self-limiting: the (1 − D) factor slows growth as the network approaches complete failure, so D approaches 1 gradually rather than overshooting.
  • Tunable: the exponent n decides whether damage builds steadily or only becomes serious once the energy passes a threshold.

Integrating over the charge-discharge history gives a compact closed form, D(t) = 1 − exp[−k_d ∫(E_v/E_c)^n dt], which rises smoothly from zero and saturates toward one.

Where the damage shows up

Once conductivity falls, the consequences cascade through the cell:

  1. Current is forced through fewer pathways, raising local current density.
  2. Local overpotential increases, so polarisation grows.
  3. Ohmic losses rise, which means more heat for the same current.
  4. The discharge curve shifts downward and reaches cutoff voltage earlier.

This is one of the mechanisms the study uses to explain the earlier voltage drop seen under vibration, and it links directly to the temperature rise discussed later in the paper.

Honest limits of the approach

The authors are open about what this term is and is not:

  • It is phenomenological. Direct measurements of vibration-induced damage in the CBD are scarce, so the rate constant k_d was estimated through physically constrained sensitivity analysis rather than fitted to dedicated experiments.
  • It is lumped. D is a spatially averaged quantity representing the overall state of the conductive network, not a field resolved across the electrode. Local hotspots and uneven damage are not captured.
  • It is not the whole story. Damage works together with tortuosity growth, contact resistance, and SEI formation, and the combined effect is what the model reports.

Because of these choices, the values of D should be read as an indicator of trend and relative severity, not as a measured physical property.

Why a simple variable is useful

A single scalar might look crude next to a full microstructure simulation, but it has a real advantage: it ties mechanical loading to electrical performance with very little computational cost. Engineers can test how different vibration profiles change conductivity loss, compare binder or electrode designs by adjusting the damage parameters, and see quickly which conditions push the cell toward early failure.

The takeaway for design is intuitive. If vibration slowly cuts the wires that feed the particles, then stronger binder adhesion, more robust conductive networks, and well-damped mounting should all slow the fraying, and a model like this gives a way to test those ideas before building hardware.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top