HPLC purity, explained: what a number on a certificate of analysis really means

Reversed-phase high-performance liquid chromatography separates a mixture by hydrophobicity. The sample is pushed through a column packed with silica particles whose surfaces are coated in long hydrocarbon chains — most commonly C18 (octadecyl) or C8 (octyl), which between them account for the great majority of peptide work. A water/acetonitrile gradient, typically sharpened with a small amount of trifluoroacetic or formic acid, flows through continuously. Molecules that are more water-loving spend more time in the moving liquid and come off early; more hydrophobic molecules cling to the hydrocarbon coating and come off later. Peptides elute, to a first approximation, in order of increasing overall hydrophobicity.¹

That "first approximation" carries a real caveat worth stating, because it foreshadows a theme: elution order is modulated by conformation and amphipathicity, not just a residue-by-residue hydrophobicity sum. Two peptides with identical amino-acid composition can resolve differently if they fold or present their hydrophobic faces differently. The column reports on shape as well as content.

Detection is almost always by ultraviolet absorbance, and the wavelength choice is not arbitrary. The peptide bond itself absorbs strongly in the far ultraviolet — below about 220 nm — which is why purity methods are so often run at 214 or 220 nm.¹ At those wavelengths you are detecting the backbone common to every peptide in the sample, which is exactly what you want for a general-purpose purity assay. (Aromatic side chains — tyrosine, phenylalanine, tryptophan — add absorbance up around 250–290 nm, which is useful for other purposes but detects only the subset of molecules that carry those residues.)

So the raw output is a chromatogram: absorbance versus time, a main peak with a cluster of smaller peaks around it. Integrate the area under each peak, divide the main peak's area by the total, multiply by a hundred, and you have your area-percent. This is the conventional primary purity metric for synthetic peptides, and for good reason — it is reproducible, it is sensitive to the process-related and degradation impurities that matter, and it requires no reference standard of the exact peptide in hand.²

Here is the problem the headline number quietly absorbs. Area-percent weights every species by how strongly it absorbs UV at your chosen wavelength, not by how much of it is present. That weighting has two consequences, and both push the reported purity upward relative to the truth.

The first is that some things in the vial do not absorb UV at all, and are therefore invisible to the detector. A synthetic peptide is almost never delivered as pure peptide. It arrives with counter-ions — trifluoroacetate or acetate paired to its basic residues — along with bound water, residual organic solvent, and small amounts of inorganic ash. None of these carry a peptide-bond chromophore. They contribute mass to the vial and nothing to the chromatogram. By area-percent, they simply do not exist.²³

The second is that among the species that do absorb, response is not uniform. UV response depends on chromophore composition, so an impurity with a different absorptivity than the parent peptide is over- or under-counted relative to its actual quantity. And some impurities co-elute — they hide underneath the main peak, unresolved, and get integrated as if they were the intended product.⁴² The detector cannot count what the column has not separated.

Put those together and area-percent is, structurally, a relative measurement: it tells you the intended peptide's share of the UV-absorbing, chromatographically-resolved material. It is not mass-percent and it is not molar-percent. The analytical literature is unanimous on this point, and it is the single most important thing to understand about a purity figure.²³

None of this makes area-percent a bad metric. It makes it a specific metric, one that answers a narrower question than the certificate's framing implies.

It would be easy to wave at this as a theoretical quibble. It is not theoretical, and the cleanest demonstration comes from metrology — the discipline of measuring how much of something is really there, done by national measurement institutes for reference materials where the answer has to be defensible to the milligram.

Consider synthetic glucagon, characterized as a reference material. The manufacturer's stated content was 983.72 mg/g. Liquid-chromatography-with-UV analysis of the main peak put it above 970 mg/g — a healthy, confident-looking purity. The mass-balance determination, which sets out to account for every non-peptide component rather than just integrate a chromatogram, returned 896.36 ± 0.68 mg/g.³ The gap between the chromatographic figure and the rigorous one is on the order of seventy-plus milligrams per gram — roughly seven or eight percent of the mass — and it is composed precisely of the counter-ions, water, and co-absorbing impurities that UV area-percent is built not to see. The authors attribute the overestimate directly to non-absorbing and co-absorbing species that the LC-UV method cannot capture.³

Seven or eight percent is not a rounding error. On a certificate, it is the difference between two products that both say "≥97%" and are not, by mass, the same thing at all.

There is a second, orthogonal limitation that the headline number hides, and it is arguably more important than the mass-balance gap: area-percent tells you how much, never what.

A high purity figure says the main peak is large and clean. It says nothing about whether the main peak is the molecule you ordered. Retention time is suggestive — the right peptide should come off the column at roughly the expected moment — but retention time is not proof, because other molecules can share it. Establishing chemical identity requires orthogonal techniques: mass spectrometry to confirm the mass and, via fragmentation, the sequence; nuclear magnetic resonance; amino-acid analysis; chiral testing to confirm stereochemistry. Purity determination and identity confirmation are different measurements answering different questions, and a certificate that reports one is not reporting the other.²⁵

This is where mass spectrometry earns its place beside HPLC rather than in competition with it. Electrospray LC-MS/MS can confirm a peptide's mass to several decimal places and read its sequence from the fragmentation pattern. High-resolution MS goes further: it can detect and quantify impurities even when they co-elute with the main peak — the exact blind spot that undermines UV area-percent — and confirm amino-acid composition and sequence, all in a single run.⁴ Where the UV detector sees one clean peak, HRMS can resolve the two or three species hiding inside it. This is why serious characterization treats HPLC and MS as a pair: the chromatogram tells you the distribution, the mass spectrum tells you the identity, and neither alone is sufficient.

A note of discipline here, because it matters for how much weight to put on any single MS number: mass-spectrometric quantitation carries its own ionization and response-factor caveats, and specific published validation figures for HRMS impurity quantitation vary by method and should not be treated as universal constants. The reliable, general claim is the one stated above — HRMS resolves co-eluting impurities and confirms sequence — not any particular detection-limit number.

If area-percent is relative, how does anyone arrive at an absolute, defensible purity — the kind a pharmacopeial reference standard requires? The answer is mass balance, and it is worth understanding because it makes explicit everything area-percent leaves implicit.

The mass-balance approach determines purity by identifying and subtracting all non-native species from the total mass, rather than integrating a single peak. In practice this separates into two kinds of accounting. First, the peptide-related impurities — the deletion and truncation sequences, oxidation and deamidation products, and other structurally related species — quantified chromatographically. Second, the water-weight components that never show up on a UV trace: counter-ion content, residual solvents, water, and inorganic ash, each measured by its own appropriate method. Absolute purity is what remains after both are subtracted.²³ USP formalizes this as a two-step value assignment for its peptide reference standards: establish the bulk material's purity by mass balance first, then use that purity to assign the peptide mass content of each vial.⁶

The orthogonal quantitative methods used to pin down absolute content are themselves instructive. Certifying the mass fraction of a peptide reference material is done with techniques like isotope-dilution LC-MS/MS amino-acid analysis (hydrolyze the peptide, quantify the released amino acids against isotope-labeled standards) and quantitative NMR. Both are powerful, and both share a vulnerability that echoes the whole theme of this piece: they are susceptible to interference from structurally related peptide impurities, which have to be measured separately and subtracted out.⁵ Content determination is method-dependent enough that formal inter-laboratory studies — comparing HPLC assay, qNMR, and amino-acid analysis on the same peptide — exist specifically to cross-validate the answer.⁷ There is no single instrument that reads out "true purity." There is a convergence of orthogonal methods that agree.

Because peptides sit at the boundary between small molecules and biologics, formal frameworks exist for how their impurities are profiled, and they supply useful analytical reference points — provided they are read as analytical reference points and nothing more.

The relevant guidance recommends identifying each peptide-related impurity present at or above 0.10% of the drug substance — naming it, characterizing its sequence and structure — and holds any new peptide-related impurity to a limit on the order of 0.5%. It also recommends sensitive, high-resolution orthogonal methods such as UHPLC-HRMS for the impurity work, which is the same orthogonality argument arrived at from the regulatory side.⁸ USP's reference-standard practice, described above, supplies the mass-balance-first methodology.⁶

The framing caveat is essential and non-negotiable for research-use content: these thresholds originate inside a drug-approval context. They are cited here strictly as analytical impurity-reporting reference points — the numbers analysts use to decide what to identify and report — and carry no therapeutic, safety, or human-use meaning whatsoever. A 0.10% identification threshold is a statement about analytical diligence, not about what is safe to put in a body. Anyone extending it to the latter has left chemistry for a claim this article does not make and cannot support.

So what does a careful reader actually do with a COA, given all of the above? Not much arithmetic, but a lot of context-seeking. The purity percentage is necessary and worth having; it is simply not sufficient on its own. The questions that turn a number into information are the obvious ones once you know what the number omits: At what wavelength was it run — 214/220 nm for backbone detection, or something that only sees part of the molecule? Is there an orthogonal mass-spec identity confirmation attached, or only the chromatogram? Is the counter-ion disclosed, since it is part of the mass the percentage silently excludes? Is the method described well enough that the separation could distinguish co-eluting impurities, or could things be hiding under the main peak?

A certificate that answers those questions is doing analytical chemistry. One that prints a single confident percentage and stops is doing marketing with a chromatogram. The difference is not the number — it is everything the number leaves out, made visible.