I split one lyophilized sample into three vials and sent them to two independent labs, plus ran it on a university HPLC. Lab A: 98.6% area, Lab B: 96.8%, in-house: 98.1%. Same week, same batch, all C18, 214 nm. The COA I was shown says 99.2% but only gives a peak table, no chromatogram.
I'm trying to figure out if that 1.8% spread is method drift or real degradation. Anyone seen a gap this wide on a split sample?
1.8% is a lot for a split sample. What mobile phase?
TFA vs formic acn shift area. Ask what both labs used.
Both at 214 nm? If one ran 220, area shifts.
I had a peptide standard that gave 98.4% at one lab and 97.1% at another. The lower lab integrated a shoulder at 12.4 min that the other missed. Once both used the same baseline settings, the gap closed to 0.4%. Your 96.8 vs 98.6 could easily be integration, not degradation.
I would ask for the raw chromatogram, the integration table with start and end times, and the blank injection. Also check whether either lab used a different wavelength or bandwidth.
The number on a COA is not magic; it is just what one analyst drew with one method on one day.
That COA with no chromatogram is the bigger red flag to me. Peak tables alone are weak.
What's the water content? Hygroscopic samples can drop a percent or two after a week.
Water content wouldn't explain a 1.8% area gap if both labs ran the same vial. Method transfer is my suspect.
Method transfer is likely. Different C18 lots and gradient delay can move that 12.4 min shoulder into the main peak.
Yep. I won’t sign off on a COA without the chromatogram and integration table. Peak table alone is just a rumor with decimals.
We had a method transfer once where a different C18 lot split the main peak. Same sample, same vial, 1.2% swing. Raw data settled it.
Ask for the blank. Always ask for the blank.
I keep a notebook of every lab’s baseline settings. Sounds obsessive, but it saved me an argument over 0.9% last month.
Water content can move results, sure, but only if sampling and vial prep differed. Same vial rules that out. I'd still ask Karl Fischer.
At 214 vs 220, peak areas can shift unevenly. I learned that after a two-lab comparison went sideways. Bandwidth matters too.
Integration settings are like handwriting. Everyone thinks theirs is normal. Overlay the chromatograms and the disagreement shows up.
I’d request raw .raw/.d files, injection sequence, and column lot. If they won’t share, that tells you more than the COA.
Login to reply to this thread.
Login / Sign up