What makes something a "certified reference material", and why is my content result only as good as the standard?

Every explanation of quantitative HPLC I have read says the result is traceable to the reference standard, and then moves on as though that settles it. I want to open that box.

Specifically: what does a lab actually have to possess for a standard to count as certified rather than just a bottle of good peptide? What is on the certificate that comes with it, and which numbers on that certificate propagate into my result? I have seen a standard advertised as "99.5% purity, HPLC" and I now suspect that is not the same thing as an assigned potency, because 99.5% purity is a statement about the chromatogram and my result needs a statement about mass.

And a naive-sounding question that I think is actually the crux: if you need a standard of known concentration to measure an unknown, how was the very first standard's value established? Something has to break the circle, and I would like to know what it is, because that is presumably where the real uncertainty lives.

Adding the hierarchy, because "reference standard" gets used for four quite different things and the distinctions determine your uncertainty.

1. Primary / certified reference material. Characterised by a producer under ISO 17034, value assigned by mass balance, AAA, qNMR or a combination, with a documented uncertainty. Traceable to SI units. Expensive, small quantities, dated with a retest interval.
2. Compendial reference standard. USP RS or EP CRS. Established collaboratively, intended for use in a specific monograph method, and — an important quirk — often supplied without an uncertainty because the compendial system treats the assigned value as conventionally true for the purpose of that monograph. Not the same metrological object as a CRM, and pharmacopoeial methods are written on the assumption you use theirs.
3. Secondary or working standard. A convenient bulk lot qualified against a primary or compendial standard, then used for routine work. This is what any lab running volume actually injects, because primary standards are too precious. Its uncertainty is the primary's uncertainty plus the qualification's, in quadrature — typically taking you from 1% to 1.5 or 2%.
4. In-house standard. A lot of good material the lab has characterised themselves, or in the worst case simply assumed. Uncertainty ranges from "as good as a secondary standard" when they did full mass balance and AAA, down to "unknown and possibly large" when they took the supplier's HPLC purity as the potency.

Two consequences that are worth acting on.

The retest date is not decoration. Reference standards degrade — they pick up water through the stopper, they oxidise, and acylated peptides are not exempt. A standard past its retest date has an assigned value that was true when assigned. Labs are supposed to requalify or replace; some do not.

Ask what the standard's own history is, not just its name. A commercial "semaglutide reference standard" from a catalogue supplier may be a well-characterised secondary standard with a real certificate, or it may be a repackaged production lot with an HPLC trace. Both are sold under the same phrase. The certificate distinguishes them and you are entitled to see it.

Of the independent testing services, the ones that will tell you which standard they used and show the certificate are the ones worth paying. Janoshik and Medutest both answer that question when asked directly; PeptideMeter publishes method conditions with their reports, which lets you at least see the calibration design. A report that names no standard at all cannot be evaluated, and unfortunately that describes most vendor-supplied documentation.

The qNMR arithmetic the accepted answer defers, plus why it disappoints on a peptide this size.

Quantitative proton NMR determines absolute purity without a standard of the same substance. You weigh the analyte and a certified internal standard of a completely different compound into the same NMR tube, integrate one resolved signal from each, and the ratio of integrals per proton is the ratio of molar amounts. Because the calibrant is a small, crystalline, unambiguously characterised solid, this is a genuine route to SI traceability. The equation:

P_x = (I_x / N_x) / (I_std / N_std) x (M_x / M_std) x (m_std / m_x) x P_std

where I is the integral, N the number of protons giving rise to it, M the molar mass, m the weighed mass, and P_std the standard's certified purity.

Worked, with maleic acid as the internal standard — two equivalent vinyl protons, M = 116.07, certified at 99.95% — using 1.000 mg of standard against 20.00 mg of a peptide of M = 4113.6, and a resolved three-proton analyte signal integrating to 80.41 against the standard's 100.00:

• I_x / N_x = 80.41 / 3 = 26.80
• I_std / N_std = 100.00 / 2 = 50.00
• ratio of these: 26.80 / 50.00 = 0.5361
• M_x / M_std = 4113.6 / 116.07 = 35.44
• 0.5361 x 35.44 = 19.00
• m_std / m_x = 1.000 / 20.00 = 0.05, so 19.00 x 0.05 = 0.9500
• times P_std = 0.9995, giving 94.9%

Two things make this hard on a large peptide. The M_x / M_std term is 35, so you need 35 times the mass of peptide for an equimolar signal — and at 20 mg per determination on a compound worth what these are worth, that is not a screening assay. Worse, a molecule with roughly 290 protons has almost no cleanly resolved signal of known multiplicity; the aliphatic envelope is a mound. You are hunting for a lone aromatic or a distinctive methyl and hoping nothing overlaps it.

Where qNMR is genuinely used on these compounds is the non-peptide terms of the mass balance. The acetate methyl singlet near 1.9 ppm is sharp, isolated and three protons, making it an excellent quantitative target for counter-ion content, and fluorine-19 qNMR quantifies residual TFA with no chromatography at all. Both feed straight into the subtraction that assigns the standard, which is a more honest division of labour than trying to weigh the peptide by NMR.

One more thing that belongs in this answer set because it is the case where the whole traceability chain is unavailable and people do not realise it.

For a genuinely novel or obscure peptide — a research compound that has never had a monograph, and for which no catalogue supplier sells a characterised standard — there is no reference standard to be traceable to. Nobody has one. In that situation a lab has three honest options and one dishonest one.

Honest:

• Report purity only. Area percentage requires no external standard at all, because it is a ratio within the chromatogram. This is why purity is always available and content sometimes is not.
• Assign a standard themselves from the material at hand, by mass balance plus amino acid analysis, and state that they did so and what the assignment was. This is legitimate and its uncertainty is larger than a CRM route, perhaps 4 to 6%.
• Report content against a nearby analogue with an explicit response-factor assumption, stating the assumption. Crude, sometimes the only option, and defensible if declared.

Dishonest, and common: quantify against the sample itself. If the "standard" is a portion of the same batch being tested, the assay measures nothing about content — it measures whether two aliquots of the same powder weigh the same, and it will report a figure near 100% of whatever nominal value was assumed. The number looks like a content result and is arithmetically circular.

The tell is a content report on an obscure compound with no named standard and a suspiciously round result. If the report says content is 99.8% of label on a peptide for which no reference material exists anywhere, ask what was in the calibration vials. The answer is often the sample.