For Research Use Only. Not for human consumption.
You ordered a research compound. A PDF showed up somewhere on the product page. It says 99%+ purity. You downloaded it, looked at it for four seconds, and moved on.
That is most people. This article is for the ones who want to know what they are actually looking at — because a Certificate of Analysis is either the most useful document in the transaction or a piece of paper that means nothing, depending on who ran the tests and what they measured.
The document is only as good as the lab behind it
Before you read a single number, answer one question: who issued this COA?
There are three tiers in the research peptide market. They are not the same thing.
Self-certified. The vendor ran the analysis — or says they did. Their name is on the document, their equipment made the chart, and nobody outside the building verified it. This is the most common type and the least useful. A seller with money on the line telling you the product meets spec is not independent verification. It is a label with a logo.
Third-party, unverified. An outside lab ran the test, but the PDF lives on the vendor’s server. The vendor decides what gets published. You cannot confirm the file was not edited, cropped, or swapped. Better than self-certification. Still not clean.
Third-party, independently accessible. An ISO-certified lab ran the test, filed the COA under its own accession number on its own platform, and you can pull the report without the vendor in the loop. Scan the QR on the vial, land on the lab’s site, read the report the lab issued. The vendor does not get to touch it after the run.
If the only path to the document runs through the vendor, treat it as self-certified — no matter how official the PDF looks.
What HPLC actually measures
High-Performance Liquid Chromatography is the standard purity assay for research peptides. Here is what the number on the COA actually means — and what it does not.
HPLC pushes a dissolved sample through a packed column under pressure. Compounds interact with the stationary phase at different rates, so they separate and hit the detector at different times. On the analyst’s screen that becomes a chromatogram: spikes on a baseline, like an EKG for your vial. The area under each spike is proportional to how much of that component is in the sample.
Purity by HPLC is the percentage of total peak area belonging to the target compound. 99%+ means at least 99% of the detected signal is your peptide. The rest is impurities, degradation products, synthesis leftovers — whatever else rode along.
HPLC does not tell you what the compound is. A 99% pure sample of the wrong molecule is still 99% pure on paper. Purity and identity are separate questions. That is why HPLC alone is never enough.
Read the chromatogram, not just the headline number. A clean pass is one tall peak, quiet baseline, no rival spikes nudging up beside it. Multiple peaks of similar height are a problem even if the purity math says otherwise — someone may be averaging away what you need to see.
If you only read the percentage and never look at the trace, you did not read the COA. You read the cover sheet.
What mass spectrometry actually confirms
Mass spectrometry answers the identity question HPLC cannot. HPLC tells you how much. MS tells you what.
The instrument ionizes the sample, accelerates the charged fragments through a field, and plots mass-to-charge ratio against signal. The dominant peak is the molecular ion — the intact molecule. For a peptide, that position is the molecular weight set by the amino acid sequence.
If the observed mass matches theoretical mass within instrument tolerance — typically ±1 Da for intact peptides — you have identity confirmation. The molecule in the vial matches the name on the label.
Mass spec does not tell you purity. A vial that is 60% target peptide and 40% junk still shows the correct molecular ion peak. MS confirms identity, not cleanliness. HPLC and MS together are the minimum credible pair. Either one alone leaves a hole big enough to drive a bad batch through.
The COA should list observed mass and theoretical mass side by side. If you get a checkmark and no numbers, you do not have a complete identity read — you have a vendor’s opinion in spreadsheet form.
Identity without the masses printed is like a handshake without a name. Polite. Not documentation.
LAL endotoxin — the test most vendors skip
Purity and identity get the marketing budget. Endotoxin screening gets the footnote. For cell-based work, it may be the test that saves your experiment.
Endotoxins are lipopolysaccharide fragments from gram-negative bacterial membranes. They show up when synthesis or handling goes sloppy. They are also absurdly potent immune activators — nanogram-per-milliliter levels in culture can fire inflammatory signaling in your cells that has nothing to do with the compound you think you are studying.
Your assay lights up NF-κB. Cytokines climb. You spend a week chasing a mechanism that turns out to be bacterial garbage in the vial. That is not a hypothesis problem. That is a contamination problem — and it burns time, reagents, and patience at a rate molasses would find disrespectful.
In cell-based systems, unexpected inflammatory readouts put endotoxin on the differential before anything else. You are not generating data about the peptide. You are generating data about endotoxin. Months of work. Real money. Results you cannot publish, repeat, or explain to your PI without sounding like you skipped step one.
The LAL assay — Limulus Amebocyte Lysate — uses a clotting protein from horseshoe crab blood that detects endotoxin at trace levels. Contact triggers a measurable coagulation cascade. Results show up as Endotoxin Units per milliliter (EU/mL). A passing line means the batch was screened and falls below the threshold for the stated research application.
No LAL on the COA? The compound was not screened for endotoxin. Fine for some uses. For cell culture, you are flying blind on a variable that can void the whole run.
Before you optimize concentrations and chase receptor biology, know whether the vial is clean enough for cells to tell you anything true.
Heavy metals and sterility — the full panel
A serious panel goes past the big three.
Heavy metals by ICP-MS — Inductively Coupled Plasma Mass Spectrometry hunts trace metals from synthesis chemistry: lead, arsenic, cadmium, mercury. Reagents and solvents carry them in at parts-per-billion. ICP-MS is built for that scale in a peptide matrix.
Sterility / microbial conformity — Broader than LAL. LAL targets gram-negative endotoxin specifically. Microbial panels cover gram-positive bacteria, fungi, total aerobic count — the wider contamination picture.
HPLC alone answers one question. A five-panel COA — HPLC purity, mass spec identity, LAL endotoxin, heavy metals, microbial — answers the set. Purity-only paperwork is a document that solved one-fifth of the problem and called it a day.
Ask what was run. What was skipped tells you as much as what was printed.
Reading the document — like someone is sitting next to you
COA in hand, work top to bottom.
Start with the header. Whose name is on it? ISO accreditation listed? Can you reach the lab without going through the vendor’s contact form? If the document only exists as a download on a product page, you are still inside the seller’s story.
Find the lot number. It must match the vial in your freezer exactly — character for character. A COA for Lot B is not documentation for Lot A, no matter how close the numbers look when you are tired.
Scan the test list. HPLC, MS, LAL — what is there, what is missing. Missing LAL on a compound headed for cell culture is not a footnote. It is a gap.
Open the chromatogram. You are looking for one dominant peak and a baseline that behaves. If the trace looks like a city skyline, ask questions before you ask the cells.
Check the masses. Observed versus theoretical. ±1 Da for intact peptides. No numbers, no deal.
Check the date. Stability matters. A two-year-old COA on a compound that degrades in solution tells you what it was at the run — not what it is when you thaw it.
Pass all of that and you have a document worth trusting. Fail one line and you have an open question. In research, open questions get answered before the experiment starts — not after the grant report is due.