PC-ART-002 Reference 22 August 2026 8 min read

A certificate is not an independent report

Learning to read a certificate of analysis is the highest-leverage afternoon available to anyone buying in this market, and it starts by throwing away the definition everybody repeats. The common one calls a COA an independent laboratory report verifying identity and purity. It is wrong on the load-bearing word.

A certificate is issued by whoever performed the analysis — ordinarily the manufacturer's own quality unit, or a contract laboratory the manufacturer chose and paid. ICH Q7, adopted by the FDA as guidance, describes it as dated and signed by the quality unit of the manufacturer. Independence is simply not part of the definition.

Third-party certificates are real and worth seeking out. The point is that independence is an optional extra somebody arranged and paid for, not a property the document acquires by being a certificate. And nothing inside the PDF ties it to the vial on your desk: the tested sample and the shipped vial are joined by the seller's word about a batch number, and by nothing else.

Purity is a ratio between peaks

The chemistry here is lovely, and knowing it changes what the number means. Reversed-phase HPLC pushes a mixture down a column packed with hydrophobic particles: greasy molecules cling and travel slowly, soluble ones sprint, and a detector at the far end counts everything that comes off, usually at 214 or 220 nm. Purity is the area of the main peak divided by the total detected peak area.

Which makes it a proportion of what one detector saw at one wavelength. Not a quantity of peptide. Not a mass. And silent about anything that does not absorb, does not elute, or comes off the column at the same moment as the main peak.

Chromatography also cannot name the peak it is measuring. Assigning the dominant peak to the intended compound needs a qualified reference standard with a matching retention time, or an orthogonal method. A vial containing 99% of the wrong peptide produces a 99% certificate.

The number that actually governs a dose is a different one, and it is rarely printed. A lyophilised peptide salt carries counter-ion — trifluoroacetate or acetate — plus bound water. Bachem states that while research-use peptides are typically above 95% pure by HPLC, peptide content of the solid may run from 70 to 90%. A vial labelled by gross mass therefore overstates the peptide delivered by somewhere between 10 and 30%, and chromatographic purity will not tell you which end you are at.

Mass confirms a number, not a sequence

Mass spectrometry is the closest thing on a certificate to a magic trick: it weighs a single molecule. A match tells you the elemental composition is consistent with the intended peptide, which is a genuinely powerful thing to know from a few micrograms of powder.

It also has three blind spots worth memorising. It cannot tell you the residues are in the intended order, because a sequence isomer weighs exactly the same. It cannot separate leucine from isoleucine, which are identical in mass. And it does not see racemisation at any position, though D-amino acid formation is one of the named synthesis-related impurity classes.

The accuracy figure quoted on certificates is often borrowed from a different class of instrument. Sub-5-ppm accuracy is high-resolution territory — Orbitrap and FT-ICR. The MALDI-TOF and single-quadrupole systems typical of research-peptide certificates work at roughly 0.1 to 1 Da. A certificate saying observed mass matches theoretical, with no acceptance window, has not stated a criterion at all.

Which brings us to the impurity that hides best. A deletion peptide missing a single glycine is 57 Da lighter than the target; missing a leucine, 113. Those are enormous, obvious gaps on any instrument. The dangerous one is the deletion that co-elutes with the main peak and never appears as a separate area on the chromatogram.

• Synthesis-related

  Deletion and insertion sequences from missed or doubled couplings, truncated chains, and residual coupling reagents. These are the characteristic impurities of solid-phase synthesis and are catalogued in the peptide impurity literature as a class.

• Degradation-related

  Oxidation at methionine, cysteine, tryptophan and histidine; deamidation of asparagine and glutamine, whose mechanism differs above and below neutral pH; aggregation into dimers and oligomers. The documented consequence of oxidation is usually loss of activity rather than a new one.

• Process residues

  Counter-ion, water, residual solvent and inorganic residue. None of these absorbs where the detector is looking, so none appears in the purity percentage — and together they are the reason peptide content by weight sits below chromatographic purity.

The endotoxin line needs a limit to mean anything

Endotoxin is the line on the page with the sharpest teeth. It is lipopolysaccharide from the cell wall of gram-negative bacteria, recognised by TLR4, and it drives interleukin-1β, interleukin-6 and TNF-α release and produces fever. It also survives sterilisation, which is exactly why a sterility result and an endotoxin result answer two different questions.

A number of endotoxin units on its own is uninterpretable, and this is the trick worth knowing. USP <85> sets the limit as K divided by M — K being 5 EU per kilogram of body weight per hour for parenteral routes, M the maximum dose per kilogram in that hour. The limit is a function of the dose, not of the vial. A result of 10 EU per vial is comfortably inside the limit at one dose and outside it at another, so a certificate printing the result without the dose it was assessed against has printed half a test.

Sterility has a similar gap between what is stated and what is shown. A compendial sterility test examines a sample of containers from a batch and reports that nothing grew under the conditions of the test. It cannot demonstrate that the batch contains no viable organism, and no certificate should be read as claiming so.

There is no purity standard to fail

Here is the part that reframes every number above. The 95% and 98% thresholds quoted across this market are supplier grade bands — commercial product tiers set by vendors against intended application, not floors set by a standards body. No pharmacopoeia and no regulator publishes a minimum purity percentage for research-grade peptides, for the plain reason that most of these compounds have no monograph at all.

Approved pharmaceutical peptides are governed far more tightly, but still not by a total-purity number: the framework is per-impurity thresholds set in each product's own specification. So a 98% claim tells you where a vendor filed a product in its own catalogue, and comparing 98% from one seller with 99% from another compares two conventions rather than two measurements.

One measured figure is worth more than everything else on this page. In a 2024 study of online semaglutide purchases, three of six orders arrived at all. Every one that did was labelled 99% purity. They measured between 7.70% and 14.37%.

Certificate of analysis

A dated, signed record of test results for a batch, issued by whoever performed the analysis. Independence is a separate claim that must be stated and evidenced.

Chromatographic purity

Main peak area divided by total detected peak area, at one detection wavelength. A ratio, not a mass.

Peptide content

The proportion of the solid that is peptide rather than counter-ion, water or salt. Typically 70 to 90% for research-grade material. Determined by amino acid analysis or nitrogen determination, not by HPLC.

Endotoxin limit

K divided by M, where K is 5 EU/kg/hour for parenteral administration and M is the maximum hourly dose per kilogram. A property of the dose, not of the vial.

ISO/IEC 17025

An accreditation of a laboratory's competence for a defined scope of methods. A laboratory can hold it for methods unrelated to the test on the certificate in front of you.

Isobaric

Having the same mass. Leucine and isoleucine are isobaric, so mass spectrometry cannot tell them apart.

Sources and evidence