Deliverables
Reports and method notes
A report is the product. Everything else in this building exists to produce a document that somebody can still interpret correctly two years later, without having to ask us what the conditions were.
Eight sections, always in this order, however small the job:
- Identification. Client identifiers against ours, dates of intake and of each measurement, the operator for each stage.
- Request. The question as agreed in the order, quoted rather than paraphrased.
- Samples as received. Condition on arrival, masses, voltages, discrepancies against the packing list.
- Protocol as executed. Method code, every parameter, and every deviation with its timestamp and cause.
- Results. Tables first, plots second. Measured values only.
- Uncertainty. The budget for this job, with the calibration dates that back it.
- Observations. What we noticed, separated from what we measured, and labelled as such.
- Data index. File names, checksums, and the column dictionary for each raw file.
Section 7 is deliberately walled off. “Cell 3 lost 20 % at cycle 212” belongs in section 5; “the loss coincided with a chamber excursion and we would repeat this cell” belongs in section 7. Clients who only read one section usually read that one, which is fine, as long as nobody mistakes it for measurement.
Delivery is a single archive: the PDF, the raw time series as CSV, the per-cycle summary as CSV, the method sheet, and calibration certificates where they were requested. One revision round is included — typically to correct an identifier or to add a comparison against your earlier job.
What a results table looks like
| Cycle | Median capacity / mAh | Range / mAh | Retention | Coulombic efficiency |
|---|---|---|---|---|
| 1 | 4512 | 4489–4531 | 100.0 % | 0.9914 |
| 100 | 4448 | 4421–4470 | 98.6 % | 0.9987 |
| 200 | 4395 | 4358–4419 | 97.4 % | 0.9991 |
| 300 | 4344 | 4296–4375 | 96.3 % | 0.9990 |
| 400 | 4290 | 4231–4328 | 95.1 % | 0.9989 |
Three cells per row. Range is the full spread, not a confidence interval. Capacity is as measured at C/2; the C/20 reference figures sit in a separate table of the same section.
Note how the range widens with cycle number, from 42 mAh at the start to 97 mAh by cycle 400. That widening is the normal behaviour of a replicate group and it is the reason a result quoted to four significant figures from one cell should be read with suspicion.
Method notes describe how we do something and what it cost us to find out. They contain no client data and we send them to anybody who asks. Nine have been issued since 2019.
| Note | Subject | Issued |
|---|---|---|
| NG-MN-01 | Formation protocols and their effect on first-cycle efficiency | 2019-11-04 |
| NG-MN-02 | Channel-to-channel repeatability, with the 2026 repeat study appended | 2023-05-22 |
| NG-MN-03 | Stack pressure in pouch holders and how 0.2 MPa was chosen | 2023-09-11 |
| NG-MN-04 | Lifetime matrices: temperature groups, reference tests, reporting | 2024-02-19 |
| NG-MN-05 | Retirement of NG-CYC-03 and NG-EIS-01, with the reasoning | 2024-06-03 |
| NG-MN-06 | Sampling electrolyte from a separator without rinsing artefacts | 2025-01-27 |
| NG-MN-07 | Consistency checks on impedance spectra before fitting | 2025-08-14 |
| NG-MN-08 | Micro-CT settings for wound cells, and the limits of a 9 µm voxel | 2026-04-09 |
| NG-MN-09 | Chamber controller replacement: before-and-after comparison on 18 channels | 2026-09-18 |
Four of our measurements have appeared in work published by clients since 2022, in each case with the method code cited and the laboratory named with permission. We do not list those publications here, because the data in them belongs to the groups who commissioned it and announcing who measured what is their decision, not ours.
- Raw time series
- 5 years, on two separate media, one of them offline
- Report and method sheet
- 10 years
- Photographs from teardown
- 10 years, at full resolution
- Calibration records
- 10 years, independent of any job
- Correspondence about a job
- 3 years
- Physical samples
- 90 days, as described under retention and disposal
Data is released to the client who commissioned it. We will send a second copy to a colleague of theirs on written instruction from the original contact, and we do not otherwise discuss whether a given sample was ever here.
Earlier deletion on request is possible and is confirmed in writing. The exceptions are calibration records, which are not client data, and the method sheet, which we keep because it is the only evidence of what our instruments were doing on a given day.