Problems with THF? (contd.)

[kkovats]

This post is the continuation of two former topics (24 Mar. 2006, "Mobile phase with acetonitrile, tetrahydrofurane and water", and 25 jul "THF quality in HPLC analysis".

Basically the problem is that a newly developed method that uses Inertsil ODS3 column, CH3CN : THF : H2O = 21 : 13 : 66 (v/v/v) mobile phase, 242nm, 40ºC does not work, since the peaks get splitted after a few injections.

The strange thing is that after the development we validated the method without any problem (about 3-4 weeks' day and night work), but after the validation the peak splitting appeared. After an investigation on the solvent batches, it seemed that the whole validation was performed using one single THf batch, and when that THF batch ended, the problem appeared.

At the end of July / beginning of August I received on the Chromforum some ideas to experiment, but at that time had no possibility to do it because of holiday and other projects.

Now I have some new information on the topic that would like to share and would like to get your oppinion / help on the issue.

1. The THf used during the development and validation was from Merck. We now tried THF from Labscane, Lancaster, aldrich and Chromsolv, sooner or later all of them resulting in the same peak splitting phenomena (sometimes after 15-20 injections, sometimes after 5 injections or less).
During these experiments we also noticed that sometimes the peak profile get worst with each injection, then suddenly from one injection to the next one get "normal". This "normality" only maintained for 1-2 injection, getting worst again and at the end splitting the peak.

2. We filtered THF (from Aldrich) through a C18 column and prepared mobile phase with this "filtrated" THF.
The result was that the peaks were splitted right in the first injection. We confirmed this fact with three different columns.

3. We distiiled THF (from Aldrich) and prepared mobile phase with the distilled THF. Sooner or lated the peaks got splitted again. (two tentatives, first one after 6 injections, second after 12-15.)

The first tentative was conducted with the column used for the THF filtration.

4. We set a precolumn between the pump and the injector and tried our luck. Unfortunately the peaks splitted again after 12 injections.

5. Since for all the above mentioned experiments we injected only resolution mixture, another experiment was to perform several blank injections between the resolution injections, to see if the problem came from the sample or from the mobile phase.

We injected once the resolution, three times the blank (CH3CN : H2O) and then the resolution again. The peaks were splitted.

6. As I explained in my previous post, this new method was developed to substitute two other methods that are currenlty used for the analysis of the respective product. One of the current methods is a CH3CN : H2O gradient and the other is a THF : H2O = 25 : 75 isocratic method.

Traditionally the isocratic method have not given problems, but we tried to perform it to see if the problem is the mixture of the three solvents or it is present when the mixture of only two solvents (THf and water) is used.
using Chromsolv THf we were able to perform 10 injections without splitting the peak (we did not perform more injections because of the week end), then the same column using THF from Merck splitted the peaks in 15 injections.

I have to note that all columns used were old columns that previously had been used for other projects, but considering the type of the problem, we obviously did not want to destroy new columns.

These were the experiments that we have made, unfortunately without solving the problem, therefore I am looking forward to receiving your opinion. If anybody is interested, I can send chromatograms or overlays by email.

Thanks in forward,

Kati


P.S. It seems very strange that all over the world only we have this problem, only we are not able to perform a CH3CN / THf / H2O analysis. What is happening here???

[gbalock]

Kati,
I'm not sure that your problem is the THF. In peak splitting, there are a number of things that can cause a problem. The one question I have, is your sample solvent stronger thatn you MP. This can sometimes cause a peak splitting, especially in early eluting peaks.

[kkovats]

Gbalock,

Thanks for the quick replay!

The dissolution solvent is CH3CN : H2O = 20 : 80, I tried not to use a stronger dissolution solvent than the mobile phase.

The main peak elutes at 20 min, and when splitting occurs, it occurs with every peak.

The peak splitting gets worst with each injection. First the peak gets lower and wider, than it splits in two and if we continue the injections, it can split in 6-8 peaks...

Thanks,

Kati

[HW Mueller]

Kati, how about answering some of the questions you were asked before?

[danko]

Hi Kati,

I’m far from sure, but I’d like to make a suggestion: It could be column voids, causing the problem you describe. You should try a brand new column and see whether the problem disappears. I don’t think you’ll destroy the column conducting this test.
If you don’t have a new column available, try reversing one of the columns you find “the best performingâ€

[gbalock]

Kati,
You don't say what the analyte is, but I'm guessing it's not ionic in nature, based on the non-buffered MP. If your analyte is ionic and you have an unbuffered mobile phase, perhaps you have ionized and non-ionized species that you are seperating. Especially if the unbuffered MP is near the pKa of the analyte. I've seen separations where the MP buffer pH is near the pKa of one component and split the peak.

THF does form peroxides, so is it possible that you used stabilized THF foir your validation and have since switched to unstabilized THF for your other analysis, or vice versa?

Another thought, is there a possibility that you are seperating isomers and in fact what is happening, your chromatography is actually "improving"?

These are just a few thoughts.

[DR]

Stabilized THF is pretty much out for UV work as the stabilizers absorb (lots).

Ionization would be worse as a function of peroxide concentration if the above idea is on track. Do you have peroxide strips & pre/post distillation numbers for the THF? Did you distill it w/ some anh. sodium sulfate (I think that's what I used to use for an assay where any peroxide would foul the results), leaving 5-10% behind?

[cristobal]

Kati:

As I developed a related substances analysis of an steroid with a mobile phase that a remember to be very similar to the one you posted (sorry I am at home, no lab records available), maybe my experience helps.
The method was a variation from the one published for Norgestimate in the USP.
When we selected the HPLC tetrahidrofuran we finally decided for the HPLC grade Fisher sells (Optima Cat Number T 427-1 in 1 liter amber glass bottles).
The solvent worked well and we have no troubles neither in the validation nor in our QC unit (until now).
The column is a Phenomenex Luna C18(2) 150x4.6

Regards

[MEH]

Decreasing the injection volume can eliminate splitting related to the sample solvent.

[kkovats]

Danko,

It could be a column void, but the void is made by the mobile phase. Or it could be dirt - caused again by the mobile phase. This void or dirt gets bigger and bigger with the quantity of the mobile phase that passes through the column (with the number of injections).

We have tried reversing the column, it worked for a few injections, but then the peak splitting appeared.

Gbalock, DR,

The analyte is a non-ionic steroid, it has been studied by MS, NMR, etc. and it has no isomers.

As for peroxides, the last flasks of the single batch that worked during the validation had 10 times more peroxides than other batches that did not work.

When trying several THF manufacturers, we tried stabilized and non stabilized THF and saw no difference (in what concerns peak splitting).

Cristobal,

Thanks for the THf reference, we will try to buy it as soon as possible.

HW Mueller,

I thought that I answered all old questions. I made a list with all ideas that received in July - August and only returned to the chromforum when tried all the ideas.

I forgot to mention that the column that we used to "filter" the THF was an already used column. Before filtration it splitted the peaks, after filtration it mainatined the peaks for 9 injections, but we had to stop the injections because pressure over limit.

please remember me if i missed any suggested idea.

Thank you all,

Kati

P.S. Yesterday somebody told me that simply the 13% THF is too high THF percentage. Do you have this experience?

[HW Mueller]

I get dizzy when going in circles, but Ok, you now partially answered one more question that I vaguely remember. That answer would indicate that THF has nothing to do with your problem. There was also the question about the filtered THF which I don´t remember exactly (try it on another new? old? column?). Probably not necessary to check, you are getting some crud which stands in what we used to call (as students) an "idiot relationship", a coincidence of two unrelated events. (We had a few more testy expressions for this type of thing, but enough for today).
In cases like this it would be nice if a link was given to the original chain.

[tom jupille]

http://www.sepsci.com/chromforum/viewtopic.php?t=3680
http://www.sepsci.com/chromforum/viewtopic.php?t=4341

[Bruce Hamilton]

OK, it's a bit deja vue, but might help clarify my thoughts. I asked the following earlier, but don't recall your answers...

[ Begin extract ]
Does increasing the column temperature by 10C change the split peak behaviour of good or bad solvent?.

If your earlier methods worked fine, what were the sample solvents and mobile phase systems, and if they included THF, does using a bad batch also kill those analyses?.

Have you tried diluting the sample and injecting more?.

If you perform multiple injections of blank sample solvent, after a couple of samples, then repeat those samples, is there any indication that the quantity of bad solvent will change the peak splitting onset.

My recommendations would be to premix the mobile phase, allow it to stand for an hour and then filter. I'm dubious that it's the THF quality ( but it's possible ). I think something else is marginal, and you may have to modify your method systematically to find it. I'd start with sample solvent/concentration, mobile phase composition, and column temperature, as split peaks can often result from one of those.

[ End extract ]

Your later information reinforces my view that it's not the THF, and something else is marginal. As well as the above, I would like to know if injection of smaller volumes or equal volumes of more dilute solutions still causes failures.

As the columns are preused columns, have you requalified the columns using the manufacturers' test procedures and confirmed they still comply?. If not, I would perform that test before using a column, and again after peaks have split. and presuure increased.

As the pressure increases on bad runs, I'm inclined to believe the solvent mixture is somehow involved in producing the insoluble material. assuming that the pumps and mixers are behaving correctly, I would be looking closely at the water, or even ingress of atmospheric components, such as CO2.

I can't see why the column brand should affect splitting or backpressure, so I'd narrow the investigation down to one brand of each solvent and column and vary parameters of those.

Have you tried the effect of using a dilute buffer ( eg 0.02M ) rather than water?. I'd also try filtering the mobile phase mixture.

I would ensure that the pumps are behaving correctly, especially the mixing valves.

If you get deseperate, you could pump the solvent from several runs through a 4mm 0.22 um Millex filter in place of the column, and check to see what appears, then I would inject the samples through the same system and look at the filter..

Bruce Hamilton

[HW Mueller]

If I understand this correctlly:

"I forgot to mention that the column that we used to "filter" the THF was an already used column. Before filtration it splitted the peaks, after filtration it mainatined the peaks for 9 injections, but we had to stop the injections because pressure over limit. ",

than you got nine unsplit (normal or good) peaks after using the column as a filter for "bad" THF, we need to discuss THF as possible cause no further. If the blockage occurred after the nine injections than you know what causes the blockage.
If the column still had a high backpressure after "filtering" THF through it than the THF is perhaps not the proper claeaning solvent, or your column is irreparably gunked up.
That this splitting and flow resistance happened with "bad" THF seems to be a "idiot relationship".

[kkovats]

Thank you Tom, for the links.

Bruce,

It really seems that I missed some ideas. But I have some answers to you, too.

One of the earlier methods is water acetonitrile gradient, the other is water THF isocratic. The first one works well, the second one sometimes gives the same problem; not so quickly as the new method; sometimes it is possible to perform a complete analysis (about one day) without splitting the peaks, sometimes the column does not resist.

Multiple blank injections between sample injections did not help.

I still have to try injectind more diluted sample, rising the temperature (although the column manufacturer told that with higher temperatures the permitted THF quantity decreases) and to filter the mobile phase one hour after the preparation.

I will get back as soon as I have the answers.

I did not perform the requalification of the old columns, but in the first injection they usually give "normal" profile for the new method's resolution mixture.

I have not tried either using buffer solution instead of water. Since the compound is not ionic, I developed the method without buffers. However, it could be another thing to try.

Thanks,

Kati