Pseudo-purity at low concentration

[pancaker]

I'm creating a standard curve for a compound that is ~90% pure in its stock formulation.

At high concentration (2-5 mg/ml), the purity is constant around 91% as determined by area percent. However, as I analyze lower and lower concentrations, the calculated purity increases:

1.0 mg/ml, 91.5%
0.5 mg/ml, 92.9%
0.25 mg/ml, 93.3%
0.125 mg/ml, 94.1%

Small peaks just become nonexistent at these lower concentrations and are eliminated from analysis.

How do I account for this in the rest of my analysis? Should all samples be concentrated >2.0 mg/mL? Should I create a calibration curve of sorts to adjust the purity back to the correct values?

Any help would be appreciated.

[Gerhard Kratz]

When operating around the detection limit it is recommended to add a certain amount of standard compound. That is one possibility.

[Peter Apps]

Hi Gerhard

How does spiking the sample with the contaminant improve matters ?

Generally

Surely a quantitative analysis is only valid if all the components to be quantitated are within their calibration ranges - as pancaker dilutes his samples the impurities drop off the lower end of their calibration (peak area is no longer proportional to quantity), they cannot be quantitated / detected and so purity is progressively overestimated as dilution increases, exactly as seen on the data.

Problem is that with more concentrated samples where the impurities can be quantitated the response to the active might be too high for the linear range of the calibration.

Peter

[HPLCaddict]

What are you actually trying to accomplish? Do you want to get an estimate of that compound's purity (i.e. you are aiming at the impurities) or do you want to determine that compound's amount in unknown samples (i.e. you need the compound's purity only for the calibration)?

[lmh]

what it really means is that you cannot estimate the purity of something if the sample is diluted below the LOD of the least-abundant-relevant-contaminant, and you cannot be completely certain of your result if the least-abundant-relevant-contaminant is present at less than its LOQ.

I don't think there is a solution to the problem: you just have to work at above that level.

Not quite what's going on here, but Purification by Dilution is sadly a very common trap encountered by biological people: keep putting the sample through more steps of purification until there is so little left that only one peak is popping up above the base-line, and then declare it pure.

[pancaker]

My goal here is to determine the purity of samples after separation by HPLC. The data in the original post is from a standard, prior to purification. I was creating a standard curve to relate peak area to concentration when I noticed the purity issue.

It's fairly trivial to concentrate my samples, just something I had never considered doing. Looks like I will need to do that in the future. Thanks everyone.