Peak Purity

[umoshplc]

Hi Everyone
I have a peak purity issue. In one of my stressed samples the purity is not passing. though there is not any co elution or an impurity underlying that peak, still my purity angle is greater than purity threshold. Any reasons why this is happening like that.

[HPLCCONSULT]

Oh my there are hundreds of reasons why.

First, how do you know that you do not have any co-eluters or real impurities (sounds like a dumb question, but it is not) ? Are the peak purity settings in the method locked down just like the method ? Can you run a "pure" sample to verify that the system still considers a "pure" sample pure ? Are the detector settings (reference wavelength OFF in all cases), flow cell parameters all the same ? Are the lamp checks good and is the system operating at 100% ? Is the column and chromatography OK ? Retention times stable (not shifting)?

Many times the purity level of a 100% pure sample can indeed show up as impure, based on the settings input and values obtained. Some reasons for this would include baseline noise, sample signal level low, reference set points bad or inappropriate. (I hope you have at least two solid reference points chosen) and all of those detailed settings which define what a "pure" sample is to the system. *We teach an eight hour class on this one topic alone ! It can get very complicated.

[umoshplc]

Thanks for your reply. What are reference points. Also please clarify me that what are the starting wavelength and ending wave length(ranges) we have to choose in finding the purity of a peak.

[HPLCCONSULT]

When you create the Peak Purity settings for a specific peak/sample you need to provide a long list of parameters to the system (*DO NOT use the default values supplied by the manufacturer as they were not selected for your sample).

One of the parameters is to assign reference points (Rt) within the chromatogram where you have a clean baseline at ALL spectra (these are saved as part of the peak purity settings, usually within the method). Usually you select one point somewhere before your peak of interest and one somewhere after your peak of interest. This establishes in real time, a reference signal which can be used and averaged to then evaluate your peak's spectra against. If you are running a gradient (where the spectra change over time because of the changing solvent composition) this is really important. *Some systems also provide an "auto" reference. I do not recommend this setting as it is usually does not select a very good reference point within your run. Besides, it you are going to use the feature, then you really should be the one selecting these points as they are very important.

You also specify windows (much like you do in a calibration), which wavelengths to include/exclude and thresholds just to name a few.

Please, if you are not sure how to use the feature (and based on the classes we teach, 99.5% of chromatographers are not), then get some training and/or help. It is not something we can go over in a few minutes. Failure to understand and use diode array based peak purity correctly will result in data which is inaccurate at best.

[umoshplc]

Hi HPLCConsultant
Thanks for your advice and valuable information. If we are looking a peak at, let's say 280 nm, what is the range of spectrum we can choose for purity. Is that ok to choose + or - 5 nm of the processing method wave length(ie 275 nm to 285 nm) or we have to choose what ever suggested in the pda instrument method(190-350 nm). If I choose 275 to 285 nm my peak purity is passing, if I choose 190-350 peak purity is failing, I know there is nothing coeluting along with the peak, how can I justify if I use narrow window in this case. I appreciate your valuble time and advice.

[HPLCCONSULT]

Based on your response you do not understand how to use this feature. Never use "instrument defaults" to make measurements with. Peak Purity is a very complex function and can not be covered in forum such as this. Peak Purity parameters can be easily set up to provide inaccurate data. I will repeat what I said before.

"Please, if you are not sure how to use the feature (and based on the classes we teach, 99.5% of chromatographers are not), then get some training and/or help. It is not something we can go over in a few minutes. Failure to understand and use diode array based" peak purity correctly will result in data which is inaccurate at best.

[Alfred88]

Dear HPLCCONSULT:
As you stated 99% of people don't know how to do it. So, you have lots of potential customers!
I am interested in your service too. You have office in West coast (USA)?

Alfred

[HW Mueller]

We have been through this many times. Purity via UV can be a gamble. The most obvious case that comes to my mind on when it can not work at all is if several compounds have the same Uv spectrum.

[Rob Burgess]

If your suspected impurity compound does have a different UV spectrum than your main peak, then you may just have a chance at proving "peak purity". However, running at default 190 -350 nm is sure to fail - look at the apectrum at 190 nm and I can guarantee it will be ultra noisy - better to go for something like 220-350. Even better still for more confidenec on peak purity would be to use LC-MS!

[tom jupille]

To go back to the original post:

though there is not any co elution or an impurity underlying that peak,

How do you know? It's a stressed sample, after all.

[HPLCCONSULT]

Peak purity by UV/VIS at best is a null test. If something has a very different spectra then your sample does, then you may detect it. If it does not, then you have no idea if there is an impurity or not. *Most importantly, my statement assumes the user knows how to set up all of the parameters in the Peak Purity software correctly. IMHO: This is rare indeed. ** DO NOT JUST CHOOSE Wavelength windows for purity determination at random. They must be selected based on your sample which we (the chrom forum members) know nothing about.

Anyone out there that thinks that MS alone can provide peak purity, then they are forgetting that MS has one main problem... not everything forms ions that you can detect. Again, based on the MS Settings selected you can generate all kinds of results. So again, just because you do not see an impurity, does not mean it is not there.

As a scientist, I would prefer to use multiple (2 or 3) techniques to determine purity within "reason". Such techniques would include: HPLC, MS, IR, NMR to name a few.

[Alex Buske]

This week in our lab:
Our "formulation scientists" were playing with different presevatives in a granules formulation. In one batch we found 130 % of one preservative (lets call it P2). We repeated the analysis, reviewed the batch record .... no results. When I checked the identity spectra the PPI was around 970 in both standard and sample. When I checked the spectra there was a slight difference at our detection WL. When I subtracted the standard from the sample spectrum I got a distinct difference spectrum with a maximum at our detection wavelength. That was exactly the spectrum of another preservative (P1).
Result of checking the log of the granulator:
They made some batches granules with P1, cleaned it and made for the beloved analysts a placebo batch containing P1 (doesn't make much sense to make a placebo with out all that junk that is in the stability testing batches, but even less sense is to have a placebo with API in).
As we are detecting P2 at one of the valleys in the UV spectrum and P1 has at this WL its maximium (around 4 AU at concentrations employed) traces of P1 are vastly overdetected.

Basic line: the Peak purity index of standard was 973 and the peak purity index of the sample was 970, so no difference, however 30% impurity at the detection WL, coeluting.
I have to measure the preservatives at their absorption minima to get the in the same scale as the API that is two magnitudes lower in concentration.

Thats on peak purity testing. Monday I will spend on the OOS procedure.

Alex

PS; Right now i 'll try my new toy http://www.google.de/imgres?imgurl=http ... swbQs6iIDg
PPS: this computer hasn't a english spellchecker