How analytical testing works
Two techniques carry almost every certificate you will encounter. They answer different questions, and neither answers the one most readers assume it does.
Chromatography: how many, and in what proportion
Reversed-phase high-performance liquid chromatography (RP-HPLC) pushes a dissolved sample through a packed column under pressure. Components travel at different speeds depending on how strongly they interact with the column packing, so they leave it at different times. A detector at the outlet records what passes.
The output is a chromatogram: a baseline with peaks. Each peak is a component. The time at
which it appears is its retention time. The area under it, expressed as a
percentage of the total area of all peaks, is the figure quoted as purity — properly written
as 98.4 % area.
The critical qualifier: area percent is a proportion of what the detector saw. A UV detector at 220 nm sees compounds that absorb at 220 nm. Anything that does not absorb there — many salts, water, some solvents — is invisible to it and is therefore excluded from the denominator. It is not counted as an impurity because it is not counted at all.
Why the method line matters
The same sample can legitimately produce different purity figures under different conditions. A shallower gradient separates closely-eluting impurities that a steep one merges into the main peak, lowering the reported purity — by resolving reality more finely, not by degrading the sample. A different wavelength changes which components are visible. A different column chemistry changes the elution order entirely.
This is why a bare number is unusable. A complete method line names at minimum:
- The technique and column chemistry —
RP-HPLC, C18 - The detection wavelength —
220 nmor214 nmfor peptide bonds - The mobile phase and gradient —
0.1% TFA / MeCN, 5→65% over 30 min
Mass spectrometry: what the thing actually is
Chromatography separates. It does not identify. A peak at 14.2 minutes is a peak at 14.2 minutes; the instrument has no opinion about its identity. Two entirely different compounds can co-elute at the same retention time.
Mass spectrometry answers the other question. The sample is ionised and the ions are sorted by mass-to-charge ratio. The result is a mass spectrum, from which the molecular weight is read. Compared against the theoretical mass computed from the declared structure, it either matches within instrument tolerance or it does not.
Purity without identity is close to meaningless. It states that the contents are uniform, not that they are the declared substance. A report showing 99% area purity and no mass confirmation is compatible with a vial that is 99% one wrong thing.
LC-MS: the two in series
Coupling the two — liquid chromatography feeding directly into a mass spectrometer — is the combination worth looking for. Components are separated first, then each is identified as it elutes. That is what allows a report to state both that the material is 98.4% one component and that this component has the expected mass.
What sits outside both techniques
Neither method measures the following. Each requires its own assay, its own sample, and appears on the certificate only if it was specifically commissioned.
| Property | Method | Unit |
|---|---|---|
| Water content | Karl Fischer titration | % w/w |
| Net peptide content | Amino acid analysis or nitrogen determination | % w/w |
| Counterion load | Ion chromatography | % w/w |
| Bacterial endotoxin | LAL (kinetic chromogenic or gel-clot) | EU/mg |
| Sterility | USP <71> membrane filtration or direct inoculation | pass / fail |
| Heavy metals | ICP-MS | ppm |
| Residual solvents | Headspace GC | ppm |
The sample is not the batch
Every figure on every certificate describes the vial that reached the laboratory. Extending it to the rest of the batch is an inference, and its strength depends entirely on how the sample was drawn — a step that is documented far less often than the analysis itself.
The questions that determine whether the inference holds:
- Who selected the sample — the laboratory, or the party commissioning the test?
- Was it drawn at random from finished units, or supplied as a dedicated specimen?
- Does the certificate carry a lot identifier that also appears on the unit itself?
- Do the manufacturing, sampling and reporting dates form a coherent sequence?
A report that is silent on all four is still a valid analysis. It is simply an analysis of one vial, and it supports no claim beyond that vial.