linear range... not linear

[lupetto]

dear community,
in our lab we found that the dynamic range for a sample detected with DAD was not linear. for higher concentration the response factor of the detector is higher than for lower concentrations (therefore the not linearity trend is kind of not usual). initial (highest) concentration of the sample was 1gr/L (10 mu injected) and the lowest was 0.05% of the initial concentration.
I personally think that we have an issue with the column recovery, and some sample gets stuck in the column.
what do you think? am i missing something?

[Consumer Products Guy]

initial (highest) concentration of the sample was 1gr/L (10 mu injected)

If I understand your post correctly:

1 gram of active per liter is EXTREMELY concentrated, sounds like you're seriously overloading your DAD. We typically use API in about 1 x 10-4 grams per liter concentrations.

[lupetto]

let me get it straight: your API sample is 1x10-4 gr/L which also means 1x10-4 milligrams per milliliter?!

1gr/L (or 1mgr/mL) is a pretty standard concentration of the sample for 4.6 mm columns. is perhaps a little bit high (sometime 0.1-0.2 mg/mL is better), but not EXTREMELY high at all.

[HW Mueller]

"Overloading" the UV detector and overloading the column are two completely different things. One should be able to "overload" the DAD at much lower concentration (and maybe even at ideal concentrations for the column) if the absorption coefficient is very high.
lupetto, could it be that your integratin parameters are set such that they loose part of the peaks of low concentration?
"Overloading" the UV would tend to give lower upper values.

[lupetto]

that's right, an overloaded DAD would give lower upper values, but in our case it's the opposite, that's why i'm a bit puzzled.

poor integration is indeed a possible explanation, but unfortunately i don't think that is the reason this time.

[Consumer Products Guy]

initial (highest) concentration of the sample was 1gr/L (10 mu injected)

If I understand your post correctly:

1 gram of active per liter is EXTREMELY concentrated, sounds like you're seriously overloading your DAD. We typically use API in about 1 x 10-4 grams per liter concentrations.

Mis-typed: meant 1 x 10-4 grams/ml to the 1 x 10-5 grams/ml range is typical for us. Your 1 gram/liter (1 x 10-3 grams/ml) is pretty concentrated.

[AdrianF]

Could you post some chromatograms and some data so we can look at the non linearity

[lupetto]

consumer guy, with the concentration range you mentioned in most of cases we would not be able to detect impurities down to 0.05%. but of course this depends on the sample we need to detect.
anyway, for the APIs (and ohter stuff) we're working with, the response is linear even for the concentrations we're using, at least in most of the cases.

[danko]

Hi Lupetto,

If I were you I’d remove the column and connect the capillary tube directly to the detector. Then I’d inject some representative volumes (e.g. 5, 10. 15 etc. μL) from the same vial in order to eliminate the possibility of concentration errors. If the same tendency is observed, I’d clean the flow cell thoroughly and examine the linearity again.

Best Regards

[lupetto]

0.0012532 1929
0.0012532 1724
0.0025064 4259
0.0025064 4617
0.006266 16439
0.006266 16265
0.012532 49840
0.012532 48395
0.025064 137200
0.025064 135535
0.050128 358944
0.050128 360274

here we go. column one is the API concentration (mgr/mL) and column 2 is the peak area. if the peak height is used instead of area the trend is the same.

thanks to all for the suggestion you gave so far.

[AdrianF]

I noticed you have used doubling dilutions which is not advisable.

I suggest you make up a stock standard and make up concentations of 0.01,0.02 --- 0.05 with even spacing and see what happens.

[Bruce Hamilton]

Have you checked the peak spectrum to ensure that you are reading on the top of a single peak, and have suitable sampling rate and spectral band width parameters?. It could also be caused by either reading on the shoulder, or shoulder of another peak.

I'm assuming that all the dilutions and parameters work for other sample analyses, so the most likely cause is either a quantum loss of the analyte at each injection (eg binding to column, precipitation, injector malfunction, etc ), or a detection issue, which includes ensuring the slope sensitivity catches the correct beginning of the peak, as a low slope sensitivity would start part way up the peak.

I assume the system is working OK, so look for unique features of the assay, such as ensuring the sample is dissolved in the initial mobile phase, and investigate what happens when you inject 2 - 4 times the volume of the low concentrations, does the area double, and how does that area correlate with smaller injectionsof the highest standards?.

Please keep having fun,

Bruce Hamilton

[ym3142]

signal to noise ratio is too low.

at the low end, the difference between the two inj is more than 10%( bad precision), while at the high end, only 0.3%(much better precision), indicating the method does not have enough sensitivity.

Hope this helps,

Good luck

[lupetto]

thanks for the suggestions folks. i really appreciate your help.