Independent HPLC 91.2% vs COA 98.7% - same vial?

Different columns can absolutely shift area %. What wavelength did they run? 214 vs 220 nm changes integration in my experience.

The 7.3 min unknown is teh part I'd chase. Did the independent lab run a blank and a spiked sample? Solvent artifacts can show up there.

Ask for the full chromatogram, not just the summary PDF.

A gap that large usually isn't just column chemistry. I've seen supplier COAs use a narrow integration window and ignore a broad hump. If the independent lab used a gradient and the main peak splits, purity drops fast. The 3.8% unknown at 7.3 min could be a degradant, a residual solvent, or something the first method missed. You need MS or at least UV spectra to guess. Ask both labs for raw data, method files, system suitability, and blank runs. Also confirm whether one number is area % and the other is mass balance. That distinction alone can explain several points.

I had a 6-point gap once. Turned out the supplier COA was from a different lot. Batch number on the vial matched the PDF?

If the COA shows no unknown peak and the independent run shows 3.8% at 7.3, I'd trust the lab with raw data more than a PDF. But don't panic yet. Method transfer between labs is messy. Ask for system suitability, blank, and spiked sample. If the peak grows after a week in solution, that's degradation. If it stays, maybe it was always there and the first method missed it. Also compare UV spectra if available.

I’m with the raw-data crowd. A PDF alone never convinced me. I’d ask for the method file, injection sequence, and blank overlaid at 214 and 220 nm. If the 7.3 min peak is absent in the blank and grows with time, that tells a clearer story than either COA.

Yep. Batch number first. I’ve seen a COA from a different lot pasted into the same template. Painful.

Diary note: I once chased a 7.3 min ghost for two days. Turned out my mobile phase pH drifted. Not saying that’s your case, but I’d check buffer prep and column equilibration before blaming the supplier.

I’d still request UV spectra from both runs. If the unknown has a different max, it’s likely real. If it matches the main peak, maybe it’s a split or solvent mismatch. MS would settle it, but UV is cheaper.

Spiked sample or it didn’t happen, my friend.

I run a narrow integration window too, but I always report the broad hump separately. If the supplier COA ignored it, that explains the gap without anyone lying. Ask for the full chromatogram and integration events.

Gradient methods can make late peaks look bigger depending on column age. I’d compare void time, pressure trace, and system suitability from both labs. If the independent lab’s tailing factor is off, area % gets weird.

Honestly, I’d trust the lab that shows blanks, spikes, and raw files. Same vial, same lot, then method transfer becomes the suspect. Keep it at 4°C? Does the peak grow after a week? That’s my next journal entry.