Please help to clarify the different baseline of two labs

[yzhang]

Dear all,
Our lab run a HPLC method same as co-validation lab. But we get very different blank chromatograms.
There is an obvious peak in blank run of covalidation lab but our lab do not have. Although this peak is far away from intented analyzed peaks, covalidation lab want to clarify what is this peak.
It is very strange that we have no any hints to clarify. We did all investigates we can do, we used same bottles of mobile phase and same colunm, two HPLCs of our lab presented better blank run, but four HPLCs all had this strange peaks.
Now we have no any idea to find what factor could lead to this difference, so anyone can help to give me more suggetion?
Sorry I don't know how to post images. Best regards. Ying

[mattmullaney]

Hi Ying,

What is the separation like--gradient, isocratic, ion-pairing...were the steps to prepare all four HPLCs the same prior to their use in co-validation? Do the HPLCs involved all employ seal and/or needle wash solutions?

Guess what I'm asking for are more details...anything you can think of that could be related to generation of peaks, including the source of water for the separations and purity of buffer salts/eluent additives that are used in the method.

[yzhang]

Hi Ying,

What is the separation like--gradient, isocratic, ion-pairing...were the steps to prepare all four HPLCs the same prior to their use in co-validation? Do the HPLCs involved all employ seal and/or needle wash solutions?

Guess what I'm asking for are more details...anything you can think of that could be related to generation of peaks, including the source of water for the separations and purity of buffer salts/eluent additives that are used in the method.

Dear Mattullaney,
This method uses a gradient with mobile phase A 0.04M KH2PO4 PH=2.6, adjusted with H3PO4, Mobile phase 100% Methanol.
Grandienti 0-5min 80%A: 20%B
5-10min 80%------40%A, 20%-------60%B,
10-20min hold at 40%:60%
20.1min from 40%:60% back to 20%:80% and keep 20%:80% to 25min
my intented peaks presented at 7min, 13min and 14min, the unknown peak was at about 24min.
I am guessing it is related with gradient because every run has it. But I can not explain why there is much lower absorbance in my own lab.
Thank you for reply. Waiting for more from you.
Ying

[mattmullaney]

Hi Ying,

I think you may kick yourself...to me this sounds a Lot like a "gradient recovery hump" as the 40%A:60%B is quickly returned to 80%A:20%B at 20.1 minutes...would take about four minutes to reach the detector on your LCs. The peak-like feature is akin to a gradient artifact--if you have a photodiode array detector available you may confirm this, or plot the ratio of the absorbance of two separate wavelengths before, during and after the "peak" at 24 minutes. The ratio(s) should be roughly constant over time...only the refractive index of the eluent has changed.

The good thing is that this feature doesn't interfere with what you want to look at, and the better thing is that the explanation I give works for all four LCs. No contamination or instrumental problems, only a rapid change in refractive index at the point i time the gradient flip occurred in your method plus about four minutes--the time it takes for the new eluent composition (the 80:20 comp from the 40:60 comp) to reach the detector.

Please confirm my thoughts...I feel you have no worries at all on this count, though. Best of Luck to You!

[yzhang]

Hi Ying,

I think you may kick yourself...to me this sounds a Lot like a "gradient recovery hump" as the 40%A:60%B is quickly returned to 80%A:20%B at 20.1 minutes...would take about four minutes to reach the detector on your LCs. The peak-like feature is akin to a gradient artifact--if you have a photodiode array detector available you may confirm this, or plot the ratio of the absorbance of two separate wavelengths before, during and after the "peak" at 24 minutes. The ratio(s) should be roughly constant over time...only the refractive index of the eluent has changed.

The good thing is that this feature doesn't interfere with what you want to look at, and the better thing is that the explanation I give works for all four LCs. No contamination or instrumental problems, only a rapid change in refractive index at the point i time the gradient flip occurred in your method plus about four minutes--the time it takes for the new eluent composition (the 80:20 comp from the 40:60 comp) to reach the detector.

Please confirm my thoughts...I feel you have no worries at all on this count, though. Best of Luck to You!

Hi, Dear mattmullaney,
Thank you very much for so detailed knowledge. I am trying to convince covalidation lab to accept this peak, however, it doesn't interfere the analysis.
Another story, the reason why this covalidation lab care about this peak so much is that sometime they failed to reply challege from an audit to clarify an unkown peak.
Best regards to you. Ying

[HPLCaddict]

Another story, the reason why this covalidation lab care about this peak so much is that sometime they failed to reply challege from an audit to clarify an unkown peak.
Best regards to you. Ying

With gradient methods, it might be wise to exclude the "late" part of the chromatogram (i.e. the reequillibration part) explicitely from integration via the SOP for this method. No need anymore to discuss with auditors why system peaks in this part are not inetgrated .
Of course this exclusion should reflect the actual gradient run. Peaks occuring during the "separation part" of the gradient should always be integrated. Even when they are system/ghost/dead/whateveryoucallthem peaks.

[mattmullaney]

HPLCaddict is quite correct...at the moment I'm early in a Method Development, and the wash-out and re-equilibration are being "left in" my chromatogram traces. Tomorrow afternoon, if all continues to go well, they will disappear with a written justification as to why.

Nothing HPLCaddict says is lost on me. Always great to know that Someone is Looking Over Your Shoulder...that way it is So Much Less Lonely.

What else so they say...if it was All Fun, It Would Be Called Play, Not Work...

[HPLCaddict]

Too much flattery . But I might give that right back to you, Matt. Keep on with your valuable contributions!

[mattmullaney]

Hey HPLCaddict--was actually referring to Compliance concerning "looking over shoulders," they are Good At That. Bless them all!

However, my thanks for your kind opinion--I've had help in my past and feel obliged to help others in return if I am able. Your posts are also quite good in my own small opinion...

[yzhang]

Hi, Dear mattmullaney and HPLCaddict,
I want to add a question, is it posibile differenet detectors lead this peak? I checked with my colleague and knew that all detectors in covalidation lab are UV detectors and those of our lab all are DAD. If DAD is better when meet the sharp gradient change than traditional UV detector?
Regards. Ying

[mattmullaney]

Hi YZhang,

Interesting Question...in my experience it is true that some detectors (UV and DAD/PDA) are more susceptible to changes in RI than others (another indicator of this "phenomenon" is the mid-gradient "hump"), but I've not seen any clear definitive evidence that DAD/PDA are always less susceptible to RI changes than any given UV detector. It seems true that the newer UV and DAD/PDA detectors (past four years or so) are noticeably more tolerant to RI changes than their older fellow DAD/PDA and UV detectors.

I'm tired...I think what I just wrote made sense, I'm frankly not sure. If it doesn't, please let me know. Don't know HPLCaddict's take on this question, I'm interested in what he has to say, though.

[LCbob]

Hi

Different detectors do give differing refractive effects. In fact a few years ago Agilent subtly changed the VWD flowcell, the reason ( I was told ) was to reduce the RI effects. So...even 2 detectors from the same manufacturer may give differing results.

Have fun

[HPLCaddict]

AFAIK the "gradient recovery peak", if it's mainly related to refractive index change, will heavily depend on the alignment of the flow-cell in the optical path as this determines the amount of stray light. So it's perfectly reasonable that different detectors give different sorts of "gradient recovery peaks".
Concerning UV vs. PDA I've actually also observed differences in "gradient recovery peaks" among these two, but not generally. It seems to depend on the method. All else being the same (especially no reference wavelength set on the PDA) I'd suspect this might be due to the general difference concerning the optical path in UV and PDA detectors. AFAIK in UV detectors, the light is diffracted before the flow cell so that only light of the analytical wavelength passes. In PDA detectors, obviously, all the light passes through the flow cell and diffraction is accomplished afterwards. This might lead to significantly different amounts of stray light and thus generally different refractive index behaviour. Just my thoughts.
Nonetheless, I wouldn't deem a PDA better or worse than UV detector just because of this. It's a gradient hump. Dot. You don't have to evaluate it, no matter how it looks like .