The project — Battery Life Cycle

Method

What I take from a paper is not a reference but a number with its circumstances. For every value I record what was measured, its magnitude and units, the material, and — without exception — the conditions: temperature, current, voltage window, cycle number. Beside it goes the verbatim sentence the number stands on, and the page. Where a paper states no conditions, the field reads not stated; an empty field is not allowed, because silence and absent data must stay distinguishable.

Then the number stops being a quotation and becomes part of a calculation. It enters storage with its provenance: which tool produced it, from which record, under which revision of the code, and when. So the database can answer not only how much but from where — and can show where its own calculation disagrees with a published value.

If a value attributed to your paper is recorded here incorrectly, write and say so: it is corrected, and the correction is kept with its date beside the value.

A collection of references answers the question who wrote about this. I answer can these two numbers be put side by side — and often the answer is no, and here is why.

One record, as it is kept

One value taken from a paper, with everything recorded beside it — shown rather than described.

What was measured

Initial discharge capacity of an NMC622 cathode in a solid-state cell

Value

≈147 mAh g⁻¹

Measured on

NMC622 with Super P conductive additive; solid-state cell with a Li₆PS₅Cl electrolyte

Conditions, as stated

20 ± 2 °C · 0.2C, where the paper sets 1C = 160 mAh g⁻¹ · 2.0–3.7 V · cathode loading ≥ 12 mg cm⁻², stated for all cells in the work

Conditions not stated

The cycle number. It reads as the first discharge, but the paper does not say so, and the field is kept as not stated rather than filled in by inference. The loading of this particular cell is not given either — only the figure for the whole work.

Why it cannot be compared blindly

The voltage window is given against a lithium–indium anode, not against lithium metal. The value is sound; putting the window beside a 2.0–3.7 V vs Li/Li⁺ window from another paper would not be. This is the whole point of keeping conditions with the number.

The sentence it rests on

The initial discharge capacity of the NMC622 cell, ∼147 mAh g−1, is somewhat moderate compared to reports involving nanoscale single-crystalline particles — page 2

Source

Bhadra A. et al., “Mitigating Electrochemical Isolation in Ni-Rich Layered Cathodes for Durable Solid-State Batteries”, Advanced Science, 2026 · DOI 10.1002/advs.202518327 · Creative Commons Attribution 4.0

Provenance

Extracted from the article text by a language model in September 2026. The record keeps a fingerprint of itself and of the code that accepted it, so a later change to that code is visible against every record it admitted. This quotation was checked word for word against the article.

Open gaps

Quantities the work still lacks in usable form — the places where there are not enough published numbers to calculate with.

Open-circuit voltage with its measurement conditions

Impedance spectra with the frequency range measured

SEI film growth constants

Lithium plating: onset threshold and irreversible fraction

Loss of active material, separately for anode and cathode

Differential capacity: peak positions and heights

Lithium diffusion coefficients in the electrode

Activation energies of ageing processes

Particle cracking and electrode mechanics

Reference performance tests: procedure and conditions

Of the literature these gaps depend on, roughly one paper in nine is openly available. That is a property of the field rather than a complaint: comparable published values are scarce in the open, and assembling them is most of the work.

Findings so far

Measurement conditions can invert a diagnostic

On one cell after identical ageing, the charge-transfer resistance growth used across the field to judge ageing came out negative at room temperature and strongly positive in the cold — the two measurements differing only in temperature. A value published without its temperature and state of charge cannot be used for comparison, and it is published that way routinely. The underlying figures come from a dataset licensed without derivatives and are therefore not reproduced here.

Disagreement between model and measurement is counted, not assumed away

Verification class	Records	Agreed
Validation	879	280
Validation by agreement	857	846
Solution verification	600	551
Code verification	8	7

The four rows add up to the 2,344 records. “Validation” weighs a result against a measurement; “validation by agreement” weighs it against another published result. The 112 checks of the physical model against an instrument record, of which 16 agreed, are a cut through the same records by the tool used, not a fifth category, and are not added to the total.

I do not claim the model is right. I keep score of where it is not.

A computed value means nothing without the tool's name

Two independent published tools, analysing the same curve, returned peaks differing by an order of magnitude — and on one cell, differing in sign.