Performance Qualification - Noise and Drift Test

[syx]

Dear Members,

Which is the best condition to check the noise-drift of spectrophotometric detectors (UV-Vis/PDA) in performance qualification test?
- air filled flow cell
- solvent filled flow cell, not flowing
- solvent filled flow cell, flowing

Please give me any explanation.

Best regards,
SYX

[Uwe Neue]

I am not an expert on this, and I have not checked with our experts, so one should treat this as the opinion of an amateur.

If I want to test drift and noise, I would do this in a case where I eliminate other sources of drift and noise. Considering this, I would use the dry cell to do this test.

If I work with a wet cell without flow, the noise may be ok, but depending on the design of the flow cell, a drift might occur as the cell warms up from exposure to the light energy.

If I work with a flowing stream, pulsations of the flow may expand the flow cell and create additional noise. Also, if I work with an automated mixing of the mobile phase, compositional ripples will affect the refractive index of the solvent and create additional noise, if the detector is sensitive to such things.

Actually, after writing this, I realized that the sensitivity of the detector to such miscellaneous events should be part of the evaluation. One can design detectors that are only minimally sensitive to such extraneous items. However, such tests are more useful for an evaluation of different detectors than for a performance qualification of an existing detector.

[andy]

As a worker in a regulated industry, it is important to me to show that my detector works under the conditions which will be found during normal use. Since that is the case, we test the noise and drift in our detectors using a flowing stream. This test is based upon the one found in the ASTM document dealing with UV HPLC Detectors.

[DR]

I would be most interested in knowing what kind of noise I should expect for a given set of running conditions, so I would take noise measurements while running gradients (or isocratically) without having made any injection. Then I would also check for injection contributions to noise by injecting mobile phase blanks.

For drift, I would collect data for at least 20 minutes with no flow, or a very low isocratic flow. The same would generally apply for PDA, except you can look for changes in absorbance maxima in addition to changes to response at a given wavelangth.

Like Uwe, I offer this strictly as opinion as I am not versed in the metrology of detectors.

[Uwe Neue]

Andy, your test under use conditions does not test the detector though. It tests the system. This is good for an operational qualification or performance qualification of the system, but it is not a pure detector test.

I think SYX needs to think about the purpose of his test.

[unmgvar]

I would suggest to perform the test with device flowing, and with a column attached as well.
The priciple of the OQ is to show that the instrument performs adequately when used. Since you always use your UV when the pump is working while you have a column attached to it, then this needs be a part of your procedure.
the procedure itself needs not to take more then 15-20 minutes in itself but you must perform it after the uv as been turned on for at least 1 hour. generally it is performed last.
Best done with water at 250 nm. using a 1ml/flow. if you wish to see if the proportion's valves contribution then run it 50% A 50 % B both channels with water (so not to insert mobile phase drifts).
Both test should be done simultanously. calculated according to ASTM because it is most severe, and more precise then when calculated by hand or according to RMS.

[syx]

Andy, your test under use conditions does not test the detector though. It tests the system. This is good for an operational qualification or performance qualification of the system, but it is not a pure detector test.

I think SYX needs to think about the purpose of his test.

I need it for performance qualitication. Hall and Dolan (2002) determine detector noise and drift in flowing condition (1 ml/min). But, in the manual, it is recommended to use dry cell. I do not know which should I use?
I tend to use Hall and Dolan's method, because it is simple and do not need to dry the cell that may minimize error due to bubbles from incomplete refill.

[Uwe Neue]

OK, this is perfectly fine for your purpose of performance qualification. However, you must realize that such a performance qualification is not a test of the detector per se, but a test of your system, including the pump and the mobile phase. I bet that this is the reason why the detector people suggested to test their detector with an empty cell. There are no contradictions in this, just a question of defining what you want to accomplish.

[syx]

Dear Mr. Neue, thank you for the explanation.
I can get the summary: when we need to check performance of our system we could use the method as recommended by Hall and Dolan (2002)*.

*)Hall and Dolan. Performance Qualification of LC Systems. LCGC North America 20 (9) September 2002. pp. 842-946