What It Means to Validate a Method on Your Matrix

Most analytical problems that reach us are not presented as analytical problems. They arrive as a disagreement: a customer's laboratory and the supplier's laboratory have tested the same batch and reported different numbers, and each is confident.

Usually neither is wrong about what they measured. They measured different things, because the method was never established properly on the material in question.

Start from the question, not the instrument

The first decision is what is actually being asked, because it determines everything downstream.

"How much of the active is present" is an assay question. It needs accuracy near a single concentration and it tolerates a fairly narrow working range. "What else is in there" is an impurity profiling question, and it needs sensitivity three orders of magnitude lower, with the ability to see a peak that nobody predicted. "Does this batch match the last one" is a comparability question, and a fingerprint method can answer it without identifying anything at all.

Only after that does the technique follow:

• HPLC for non-volatile organics, which is most specialty chemical and pharmaceutical intermediate work.
• Gas chromatography for volatiles, residual solvents, and thermally stable compounds that will survive the injector.
• GC-MS and LC-MS when identification matters rather than only quantification, which includes any unknown impurity that has to be explained rather than merely measured.
• Atomic absorption spectrometry for metals, including residual catalyst.
• X-ray diffraction for crystalline phase and polymorph, which no chromatographic method will tell you.

Choosing an instrument before defining the question is how laboratories end up with a beautiful method that answers something nobody asked.

The matrix is the whole problem

A calibration curve built from pure standard in clean solvent describes the behaviour of pure standard in clean solvent. Your sample is not that.

Everything else present, the solvent it came from, the salts, the unreacted starting material, the process aids, changes the measurement. In chromatography it shifts retention and suppresses or enhances response. In spectroscopy it interferes with the wavelength you are reading. In mass spectrometry, ion suppression from co-eluting matrix can halve a response without altering the peak shape enough for anyone to notice.

This is why a method validated on a standard and applied to a real sample can be confidently, reproducibly wrong. It is also why "we use a standard method" is not an answer on its own. The compendial method was validated on someone else's material.

The practical test is recovery: spike a known amount of analyte into the actual matrix, at the concentration you care about, and see how much you get back. If recovery is 82%, you now know something important. If nobody ever ran the experiment, you do not.

What validation actually asks

Five questions, and each one is a specific experiment rather than a statement.

Specificity. Does the method measure the analyte and nothing else? Proved by showing that blank, placebo, known impurities, and degradation products do not interfere at the point where the analyte is read.

Linearity. Does response track concentration across the range you will use? A correlation coefficient alone is a weak answer, because a curve can fit well and still have a poor intercept. The residuals matter.

Accuracy. Does the method return the true value? Established by recovery in the real matrix, at low, middle, and high concentrations, not just at the target.

Precision. Does it return the same value on repeat? Both within a single run, and then again on a different day, by a different analyst, on a different instrument. The second of those is where most methods reveal their weakness.

Limits of detection and quantification. Below what level can the method see the analyte at all, and below what level can it put a defensible number on it. These are different limits and they are routinely confused. Being able to see a peak is not being able to report a figure against it.

Forced degradation is the honest test

The most informative experiment in method development is deliberately damaging the sample.

Stress it with acid, with base, with peroxide, with heat, with light. Then run the method. If the degradation products co-elute with the analyte, the method cannot distinguish a good batch from a decomposed one, and no amount of clean chromatography on fresh material will reveal that.

It is uncomfortable work, because it is designed to break something that appeared to be working. That is precisely why it is worth doing before a method is issued rather than after a disputed batch.

Robustness and the handover

A method leaves the laboratory that wrote it. It will be run by someone who was not there, on an instrument from a different manufacturer, with a column from a different batch.

Robustness testing anticipates that: vary the mobile phase composition slightly, the pH, the column temperature, the flow rate, and see whether the result moves. If a two percent change in organic modifier collapses the separation, the method will fail in transfer, and it is better to know that while it can still be fixed.

System suitability criteria are what carry that knowledge forward. Resolution between the critical pair, tailing factor, plate count, injection repeatability: a set of checks the operator runs before every sequence, so that a system which has drifted refuses to produce data rather than producing bad data quietly.

What a Certificate of Analysis is worth

A CoA is a claim, and its value is entirely determined by the method behind it.

A certificate produced by a validated, matrix-appropriate method with system suitability in place is evidence. A certificate produced by a method that was never challenged is a number with a signature under it. Both look identical on the page, which is why the question worth asking a supplier is not what the assay was, but how it was measured and against what.

Our laboratory runs HPLC, GC, GC-MS, LC-MS, atomic absorption, and XRD, and method of analysis development is a service line in its own right. We hold ISO 9001:2015 certification covering chemical analysis; NABL accreditation has been applied for and is not yet granted, and we say so rather than leaving it ambiguous.

If you have a compound your current method struggles with, or two laboratories disagreeing over one batch, send us the details. Working out which method is answering the right question usually takes a conversation rather than a project.