Query regarding linearity

[Serpico]

Hi all,

I have a quick query regarding linearity:

The observation:

A compounds response has been shown to be linear over a large range (even up to a 20ul injection of a 3mg/ml solution) when the UV detector (using default settings) is set to the UV maxima (260nm).

When repeating the above at 220nm (on the slope of the UV response for the main component) the linearity is shown only to be acceptable over 1/4 of the above range. Above this concentration the response flattens out significantly.

Whlst I know that it's advisable not to work on UV slopes I'd like to better understand the factors that cause the above observation. Particulalry around why its linear at lower concentrations and what actually causes the response to flatten out.

Many thanks in advance,

Serp

[tom jupille]

I would have expected the opposite: that you would would more likely to exceed the detector's linear range at λmax. That said, I'll make a wild guess that your mobile phase has a substantially higher absorbance at 220 than it does at 260, and that background absorbance is "consuming" some of you linear range.

[HW Mueller]

Or, could it be that the 220 nm is on an "upslope" of another maximum? That is, the absorbance (of the analyte) at 220 nm is much higer than at 260 nm?

[danko]

Hi Serpico,

There was a very similar topic recently, but we might as well continue here.
Here are some general thought on this matter:

1. When measuring at lamda max. the change in absorbance for a given concentration change is grater compared to measurements performed at any other wavelength i.e. on a slope. This means grater sensitivity and most importantly accuracy.

2. The relative effect of the mobile phase absorption is smaller.


3. The measurements are not affected to the same degree by small errors in the wavelength setting.

I’ve generated (some years ago) data demonstrating all this graphically, but I can’t find it now – douh
So, if you have a spectrometer at hand and a half an hour to spend you might like to perform the experiment yourself. Just prepare 3 – 4 solutions with increasing/decreasing concentrations of the same compound/analyte and make scans so that the whole absorbance spectrum for this particular compound is covered (i.e. the whole peak). Then overlay these 3 – 4 scans and you’ll see immediately why all this is true.
If you choose to gather the information mentioned above it would be very kind of you if you shared it with this forum for future reference.

Best Regards

P.S. Did I understand it correctly that the peaks measured at 260 nm were higher than when measured at 220 nm - given the concentration and injection volumes were the same?

[Serpico]

Apologies, clearly I left out some important facts.

I should have mentioned that this compound has a very poor chromophore.

260nm is a very small λmax with the UV response shooting up after about 230nm. 220nm is the λmax of the precursors and impurities, which is why I had hoped to use that for a combined assay and impurities method. So peaks are larger at 220nm than 260nm.

But I should also point out that the response at 220nm isn't huge and the peaks are easily on scale at <1Au.

I'm not an expert on detectors - to my mind I can put more concentrate solutions through the system and still get linearity at a different wavelength (260nm) so I'm assuming its not a concentration effect?

Equally, with other compounds I've seen larger peaks produce linear curves so surely that can't be the reason either?

I should also point out that the mobile phase is just ACN:Water gradient (both with 0.03% TFA).

Many thanks for your replies so far. Let me know if you need any more information.

[HW Mueller]

Do I understand this right? You are not comparing the linearity of the two wavelengths of a single, highly pure compund, but of a whole bevy of gunk? Do you know how much of each you have? Are you sure some don´t precipitate, etc., out at the higher concentrations? If everything clearly stays in solution (how would you knw?), your statements up to now would tend me to go with Tom on this: Your TFA gives enough absorbance at 220 nm such that you are blacking out the light beam with the higher mixture concentrations.

[Serpico]

The purity of the sample was established as >99.5%w/w. So I don't believe this to be the issue.

I believe the sample is stable in solution at very high concentrations - as was shown by the linearity achieved at 260nm.

Interesting regarding the TFA - I will investigate the linearity without TFA present. But suspect there is something more gong on though.

[danko]

Serpico,

0.03 % TFA will barely absorb light at 220 nm – or far from any critical level anyway.
Actually I believe it is now even more important than before to perform the spectrophotometric test I suggested in a previous post (with and without TFA in the solvent) in order to clarify the source of the error.
If everything (linearity) works fine there, than you have a chromatographic related issue.

Best Regards

[Serpico]

Thanks for all the replies. I'll look into this more tomorrow.

One thought - Any chance the dynamic range of the HPLC UV detector decreases with decreasing wavelength?

I've looked back at some other methods I've developed and seen peak area's and peak heights in excess of what I'm currently working at but still showing good linearity. However, they're all at 260nm or higher. At 220nm it currently appears that I need to be working at a significantly lower response than if I was working at 260+nm.

[HW Mueller]

A suggestion: Do the linearity experiment in the best water you can obtain (leave out the ACN as well as the TFA), degas very well and tell us the relative absorbances at the two wavelengths.

[Serpico]

Unfortunately, we only have a Xe lamp in our stand alone UV spectrometer. I understand this can cause trouble at low wavelength.

I'm also aware that the Au should be significantly lower in an ideal world.

Observations from the below data:

On the stand alone UV spec using 224nm to 300nm all data points are linear over the range 0.1mg/ml to 3mg/ml even when over 2 Au.

At 224nm and 268nm the Au result are approximately the same.

Anything less than 224nm is non linear, however, the response does shooting up and then platteau into noise above 2 Au (Xe lamp?).

When the same solutions are run by HPLC they are linear for 260nm (lambda max) and linear at 268nm (the steepest part of that peak). However, the linearity is worse than in the stand alone UV.

Very poor linearity was seen at 224nm even though acceptable linearity was seen in the stand alone UV.

The peak areas are significantly larger at 224nm than at 268nm even though the Au were ~ equivalent by stand alone UV.

I have achieved linearity on a different compound when peak areas and peak heights were significantly larger.

Conclusion?:

Outside dynamic range of detector - But why have I achieved linearity on other compounds at higher peak areas/heights?

Solutions too concentrate? Is this deviation due to the close proximity of compound molecules effecting the response? If so, why is this only apparent at low wavelength and why is it seen in the HPLC data before the stand alone UV data?

A result of working on a slope? - If so why does the steep slope at 268nm produce acceptable linearity?

Something else?

Many thanks for those that have taken the time to respond.,

Serp

---------------------------------------------------------------------------

Data from the stand alone UV:

(apologies for the poor formatting)

Conc (mg/ml) Au @224nm
0.1 0.066
0.2 0.137
0.4 0.27
0.6 0.419
0.8 0.563
1 0.714
2 1.451
3 2.154

Conc (mg/ml) Au @260nm
0.1 0.127
0.2 0.259
0.4 0.523
0.6 0.797
0.8 1.063
1 1.335
2 2.601
3 3.388

Conc (mg/ml) Au @268nm
0.2 0.155
0.4 0.315
0.6 0.483
0.8 0.645
1 0.811
2 1.601
3 2.275

Data from the HPLC analysis:

Conc (mg/ml) Area @224nm On steep slope
0.1 290672
0.2 518897
0.4 856338
0.6 1112575
0.8 1323408
1 1517143
2 2270910
3 2886038

Conc (mg/ml) Area @260nm lambda max
0.1 53565
0.2 103636
0.4 203253
0.6 302728
0.8 400790
1 499787
2 982055
3 1456471

Conc (mg/ml) Area @268nm On steep slope
0.1 37514
0.2 74402
0.4 146508
0.6 217557
0.8 286748
1 356107
2 682885
3 992664

HPLC linearity on another compound @270nm

Conc (mg/ml) Area
0.099728 574443
0.199456 1142334
0.299184 1706905
0.49864 2837442
0.598368 3396438
0.62955 3521730
0.7577 4226784
1.01884 5643595

[HW Mueller]

Don´t you have a published spectrum of your compound? Do I understand this correctly that both machines showed very strong absorption at 220 nm and below? What were your solvents in the different apparati? Did these "other compounds" show linearity at 220 nm or was it 260 nm?

[Serpico]

I can't upload from my work PC (will try from home at some point) but for now it basically looks something like the one below:



However, the first peak is larger and at 205nm and the seond peak is at 260nm.

Using the standalone UV spectrometer the AU at 220nm is no greater than that at 260nm. However, by HPLC the peak areas are significantly greater at 220nm than at 260nm.

I didn't look at low wavelength with the other comnpound.

I used the same solutions for both the stand alone UV spectrometer and the HPLC analysis.

[HW Mueller]

The picture doesn´t show here.
OK, you are doing the HPLC detector in flow mode.
Possibly your slit width is a lot wider in the HPLC detector than in the spectrometer, so that you are getting a lot of the higher absorption toward 205 nm. But: I still don´t know how much of the higher absorption at low wavelength is due to the compound and how much due to the solvent. So I don´t see how we can distinguish (and whether you need to) between Tom´s and my first suggestion.