validation of solvents

[offroad]

Dear friends,

How are you?

I am trying to figure out a way of making sure HPLC grade solvents are really pure (99.999%). Friends of mine have reported problems with this. So... I would like to know whether you could suggest a method to verify the purity of solvents.

is it by making a calibration curve? If so, how would I do it? How would I define the max and min points?

Thank you.

[Peter Apps]

If you can find an HPLC solvent that is this pure you can split it up into 1 ml vials and sell it at a huge profit as a reference material.

Seriously, demonstrating purity at this level is hugely difficult, and just opening the bottle introduces another lot of impurities. HPLC solvents do not have to be pure - they just have to be free of impurities that interfere with the analysis, either by interfering with the detector signal due to the analyte, or by interfering with the separation. If there are no interfering peaks, and the separation is working, the solvent is suitable for purpose, which is all you can wish for, and all that you need.

Peter

[offroad]

Thank you Peter for your clarification. I certainly have a lot to learn.

But... in this particular case, how do we make sure the seller is not selling us a low quality solvent?

I have heard of people who have returned boxes and boxes of HPLC grade solvents because they did not meet the requirements.

Would there be another simpler way of measuring this without running analytes? would a calibration curve work?

Thank you.

[JGK]

If you want to analyse your solvent for impurities then I suspect you'll need more than one method. It's unlikely that a single method will pick up every possible inpurity.

As for running a calibration curve what will you use? how do you know the calibration STDs are pure?

you need to be able to compare your test sample against something.

[grzesiek]

you can start by asking the manufacturer how they analyse the solvents

[offroad]

you need to be able to compare your test sample against something.

this may be funny and/or stupid, but... could I compare them against the impurities themselves?

For example:
I got these two samples of the same solvent from diferent providers, I injected them and marked sample A (quantitatively and qualitatively) as 1 and sample B as 2 (relative concentration). Of course the calibration curve shall be a straight line, R2 = 1.000.

Here comes the question:

If both samples are more or less 99.99% pure, it means there are impurities. If I inject a unknown sample C (still same solvent, of course, but a third provider), I might get 1.5 (concentration) meaning it is between 1 and 2, maybe higher or not.

Would this work?

Thank you for your time.

[HW Mueller]

One runs a spectrum (in detector) of the solvent, compares it to the literature and a spectrum of water (you also should know what it looks like). We found that fluorescence detectors are particularily good for this: if there is a fluorescence, the manufacturer goofed (common HPLC solvents).

[Ron]

One thing often overlooked is the purity of the water and any modifiers used in the aqueous portion of the mobile phase. I have gone into more than one site that was reporting issues with solvent purity with known good solvent and water, and in most cases the problems did not go away when I switched the solvent, it went away after I switched over to my water instead of the laboratory supply. I suspect that if people return boxes and boxes of solvents from a reputable supplier the problem is not in the solvent, but something else used in the analysis.

[bisnettrj2]

My question to you is: How would you know your solvents are contaminated? Unless you are doing really sensitive LC/MS work, and the contaminants are either present at a high background m/z level, or you have some sort of persistent m/z artifact showing up at a specific retention time, then the contaminants should travel with the solvent at the speed of the solvent and at the concentration of the contaminant through the column for the entirety of the use of that particular solvent bottle or lot. If neither the first or second example hold true for you, then how would you know (and why would you care) in the first place?

[grzesiek]

"then the contaminants should travel with the solvent at the speed of the solvent" - how do you know?? this statement can not be considered to be always true

[bisnettrj2]

grzesiek, you're right. As soon as I read your reply, I thought of the Gradient Composition Test for checking the gradient profile of pumps. If you've got a UV-absorbing (or some other contaminant) in your mobile phase, your baseline will be affected by the contaminant, even if it's 'evenly distributed' in the organic solvent. I had a longer original post, but my laptop battery died duing the composition, so I posted a short thought in haste. I think the original poster has been given some good ideas, especially by HWMueller and yourself, in how to determine if his organic solvents are contributing to problems in his analysis. I think what I was trying to say was that if the solvent is fit for your particular application, then use it, unless you have an issue like Peter Apps described.

[grzesiek]

surely nice disscussion

I have never had problems with fresh organic solvents for RP, definietly it is also in the manufacturer hands to sell only high quality solvents, cos as a customer I wouldn't buy again from the manufacturer who sells rubbish

"I have heard of people who have returned boxes and boxes of HPLC grade solvents because they did not meet the requirements. " - I for example have not heard about such cases

[rhaefe]

we did once. Case of HPLC grade methanol. Very noticeable mercaptane smell.
Company claimed that all was good and faxed us a GC FID chromatogram showing one HUGE methanol peak. Of course all other peaks were indistinguishable from the base line. And that was "proof" that the methanol was good and clean and HPLC grade. Anyway, we did not buy solvents from them ever again.

regards

[skunked_once]

I too have experienced a bad batch of organic solvent, in this case GC grade hexane that had some type of impurity that maxed out our electron capture detector. The solvent manufacturer was very helpful and sent us out replacements bottles without problem. We found out that the whole lot was contaminated with some volatile component but did not follow up on isolating the contamination. Interestingly enough, if the hexane was put on the rotary evaporator, the collected hexane did not contain the contamination.

[AICMM]

So your solvent is 99.99% pure. That means, at most, you have 100 ppm total impurities. Assume just 4 extra impurities and your down to around 25 ppm each which is getting near the limit of FID's. Assuming they even respond on an FID (not CS2 or COS or H2S....) Assuming they elute before or after the solvent peak because, of course, you used splitless for the sensitivity instead of split for the nice peak shape. Better to use an ECD which has ppb sensitivity, except that it only has that for some compounds and not really others. So now you are supporting at least two detectors....

Or, you boil the solvent down for assay, in which case how clean was your concentration step or did you boil away the contamination....

So it really goes on a case by case basis. If doing pesticides or PCB's for example, boiling down the solvent for testing will tell you if you have interferences in the range you are interested in.

As per Peter's comments, you can see the challenges mounting up quickly. Same thing applies in UHP gases (four 9's, five 9's or six 9's.)


Best regards.