Silicon stores more lithium. It also needs room.

Charge silicon and graphite side by side. Then give silicon internal void space and explore why repeated swelling can damage a particle.

Teaching model
Silicon and graphite particle radii follow cube roots of their volume ratios; optional void shell and illustrative damageVolume changes faster than radius.Thin inner ring: starting particle sizeSilicon: 4.11× volumeGraphite: 1.10× volumeSolid particle: no internal space for expansionCycle 0: illustrative capacityLittle damageCracks, crust and capacity bar are qualitative, not battery-life predictions.

Enlarge for bigger labels, then swipe sideways →

Silicon volume4.11 ×
Graphite volume1.10 ×
Material capacity ratio11.3 ×
Full-charge void needed75.7 %

At 100% charge, silicon occupies 4.11× its starting volume and graphite 1.10×. Silicon's radius is 1.60×, not 4.11×. Switch on internal void to see how empty space can accommodate growth. The cycle display shows an invented trend, not a lifetime forecast.

What this model assumes

Volume ratios are V/V₀ = 1 + 3.11 × charge fraction for silicon and 1 + 0.10 × charge fraction for graphite; linear variation with charge is a teaching assumption. Circular cross-sections represent spherical particles, so radius scales with the cube root of volume. The 311% expansion and 4,200 mAh/g endpoint refer to the Li₂₂Si₅ convention; they are not a claim about a commercial cell. Graphite is compared with 372 mAh/g. Void fraction is void / (starting silicon + void); enough free volume at full charge requires 3.11/4.11 = 75.7%. This geometric condition alone cannot guarantee an intact shell. For illustration only, unaccommodated expansion gives strain = max(0, 3.11 − void/(1−void))/3.11 at each full-charge cycle; damage = 1−exp(−0.09 × cycles × strain), and the capacity bar = 1−0.6 × damage. Those arbitrary coefficients and visual cracks/crust are not fitted data. A zero-damage drawing does not establish zero real degradation. Particle size, phase, additives, electrolyte, pressure, transport and anisotropy are omitted.

Space for expansion comes at a cost: less active material in the same starting volume.

The ideal Li₂₂Si₅ endpoint gives silicon a theoretical material capacity of about 4,200 mAh/g, compared with 372 mAh/g for graphite: about 11.3 times as much per gram of active material. That is not an 11.3-fold increase in a commercial battery’s capacity. Room-temperature lithiation can end in different phases.

This example uses a 311% silicon volume increase at full lithiation, so its final volume is 4.11 times its initial volume. Graphite is shown with a 10% increase, or 1.10 times. For spherical particles, the radius changes by the cube root of the volume ratio. A fourfold volume change does not mean a fourfold diameter.

If a rigid outer envelope starts with silicon plus empty space, fully accommodating a 311% increase requires void/(silicon + void) = 3.11/4.11, or 75.7%. That is a geometric lower bound under this idealization. It does not prove that a real shell survives or that all its void space is accessible.

Repeated expansion and contraction can fracture particles, disrupt electrical contact and expose fresh surface. Electrolyte decomposition can then rebuild the solid-electrolyte interphase, consuming lithium. The cracks, crust and capacity bar here show an invented trend, not measured cycle life. Porous structures can help, but they also change packing, transport and surface area.

What must be true for internal void space to accommodate a fourfold silicon volume without expanding the outer envelope?

Taking a real process
beyond the lab?

Jon and I are building RARE Labs toward an autonomous pilot plant builder using AI and robotics. Join the waitlist for updates, and tell us what you’re trying to scale.

Join the RARE Labs waitlist ↗

Go a little deeper.