Worked example — what the output should sound like¶

Section 2 of 5 — why the battery loses capacity even when nothing breaks.       <!-- ORI-01 -->

Here's a question to hold onto: when a silicon anode loses capacity,            <!-- PRQ-01 -->
what is actually being consumed?
[pause 3s]                                                                     <!-- PAU-01 -->

Two terms first.                                                               <!-- STR-03 -->
The electrolyte is the liquid that carries lithium between the two electrodes.
The S E I — that's the solid electrolyte interphase — is a thin crust that
forms on the negative electrode during the first few charges.
[pause 1s]

Here's how the crust forms. The electrolyte touching the negative electrode
is chemically unstable there, so a little of it decomposes. The decomposition
products stay put and build a layer. Once that layer covers the surface, it
blocks further decomposition. The crust is what stops the reaction that
created it.                                                                    <!-- ELB-01 -->
[pause 1s]

Picture a cast-iron pan that seasons itself: the first heating burns a thin     <!-- IMG-01, VOI-02 -->
layer onto the metal, and that burnt layer is what stops the metal rusting
further. My analogy, not the paper's — and it breaks down in one place, which   <!-- ANA-01 -->
is the whole point of this section: a pan doesn't change size.
[pause 3s]                                                                     <!-- PAU-02 -->

Because a silicon particle does. When lithium goes in, silicon swells — more
than triple its volume. When lithium comes out, it shrinks back.                <!-- NUM-01 -->
[pause 1s]

And that's the core of the argument.                                            <!-- SIG-02 -->
The crust is brittle. The particle underneath it is not staying still.
So the crust cracks, fresh silicon surface is exposed, and new crust forms
on it — which costs more lithium. Every one of those repairs takes lithium
out of circulation permanently.

So: what's being consumed isn't the silicon. It's the lithium inventory,        <!-- PRQ-02 -->
locked up in crust that keeps rebuilding itself. That was the question I
asked at the start.

How much does that cost? About a twelfth of a percent of capacity per cycle.    <!-- NUM-01 -->
That sounds like nothing, until you run it five hundred times: the cell ends
at about 82 percent of where it started. That decline is a real effect,         <!-- NUM-04 -->
not measurement noise.

The evidence is a capacity-versus-cycle-number curve: cycle number along the    <!-- FIG-01 -->
bottom, discharge capacity up the side, running out to five hundred cycles.
The line falls steadily and almost straight, with no sudden drop — which is
what tells you this is gradual lithium loss rather than a single mechanical
failure.                                                                       <!-- FIG-01 (claim) -->
[pause 1s]

Three sentences to close the section.                                          <!-- STR-06 -->
Silicon anodes fade because the protective crust cracks when the particle
swells. Each crack is repaired with lithium that never comes back. The
signature of that mechanism is a straight, gradual decline, not a cliff.
[pause 1s]

One to answer for yourself: if the fade came from the particles cracking       <!-- RET-01, RET-02 -->
apart instead of the crust cracking, what would the curve look like instead?
[pause 5s]
It would show a sharper drop, and it would come at a particular cycle rather
than being spread evenly across all of them. A straight line points at a
process that repeats a little bit every cycle.                                 <!-- RET-04 -->
[pause 2s]