THF quality in HPLC analysis

[kkovats]

Some times ago I posted a topic about problems with C18 columns when used for a LC method with CH3CN : THF : H2O = 21 : 13 : 66 % (v/v), 40ºC, 1.3 ml/min, 150 x 4.6mm C18 columns (different branches, as Inertsil ODS3, Symmetry, Luna), 240nm. The sample was a known steroid that we had been analysing for years with two other diferent methods, one CH3CN : H2O and onether with THF : H2O, these two methods never giving us problems. (The idea was to develop one single method selective for all impurities.)

The problem with the columns was that after 1-2 days of use theylost their efficiency and the peaks began to split. It happened with 4-5 columns, independently of using or not guard column, of filtering or not the solvents.

We made a reclamation to the column's supplier, they took back the columns and analysed them. The beginning of the columns were very contaminated, causing the peaks splitting.

Several people answered to my first topic and made questions and/or suggestions, but nothing explained the phenomenon.

One day we tried to dissolve the samples in a mixture of CH3CN : H2O instead of the mobile phase (CH3CN : THF : H2O) and beginning with that day the problem dissapeared. Although could not scientifically explained why the problem dissapeared, we decided to validate the method.

We made a complete validation during 3-4 weeks, working day and night in two equipments, and the two columns worked very well, maintaining the relative resolution of the method's critical peak pair between 2.4 - 2.7. During the validation, for column robustness we tried several other columns (different batches of the same supplier and different suppliers). everything went OK.

Than suddenly, from one day to the other, the problem appeared again. the same two columns that made the whole validation work and other two new columns damaged in 2 days. It seemed that changing the dissolution mixture was not the solution after all...

at this point I made an investigation of all mobile phases we had used and concluded that all the work that was OK, was made with one single lot of THF. Before we began tu use this lot and after it finished, the method did not work, the columns got damaged after a few injections.

we experimented a stabilized Aldrich THF, too, but had no luck.

we determined the peroxide content in the "good" lote (Merck), in the Aldrich lote and in another Merck THf lote. The last two bottles were opened at the time of the determination, they had about 0.003 - 0.004% peroxids. The "good" lote, that had been opened two months ago, had 0.037% peroxids. So it was not the presence of the peroxides that caused the problem.

The THF used was from Merck. We asked their help to try to understand what was going on, but their answer was "It seems to be a combination of HPLC-specific conditions and the solvent quality", so it did not help much.

So I am asking now everybody's help. Has somebody had a similar experience? Do you have any idea of what can be going on, how could we solve the problem?

Thanks,

Kati

[HW Mueller]

Did you ever check what washing of the columns (reverse if allowed) did after you got split peaks?

[Albany-12303]

"It seems to be a combination of HPLC-specific conditions and the solvent quality" - technical support way of saying 'I Dunno"

I worked years ago on a corticosteroid (flunisolide) and I found that using THF (10% as I recall) was the only way that I could 'pull out' one of the known impurities from under the active. But I had no problems.

Just this year though, I had similar issue with a C18 column (splitting and poor efficiency after a couple days). I found that the source of the problem was the presence of hydroxypropylmethylcellulose (HPLC) in my matrix (and filtering the sample did not help - the stuff went through the membrane).

I am guessing that you have HPMC, ethylcellulose or carbomer (carbopol) in your matrix (let me know if I am wrong)

I solved the problem by preparing a more conc sample in 100% ACN. This preciptated the HPLC, which sank to the bottom of the flask. I then filtered a portion of the upper layer (once precipitated the HPMC is filterable) and then diluted an aliquot of this portion to my desired conc with water.

If you cannot change your sample prep, then I suggest the use of a guard-column. I tried this with my original sample prep and it also worked for me.

Hydroxypropyl methylcellulose
ethyl cellulose
carbomer (carbopol)

All these give the same problem.

[Albany-12303]

I would like to comment on HW Muller's post.

In my case, washing was to no avail, - the crap stayed in the column.

[kkovats]

Albany,

Our product is an API with more than 99% purity (area percentage), so I don't think we have Hydroxypropyl methylcellulose, ethyl cellulose or
carbomer (carbopol) in the sample matrix. We could have something in the THF, but I don't know what and why one lot did not have and the others had...

HW Mueller,

Yes, we tried to reverse and wash out the column, following the supplier's methodology (I can't remember well but something like 100% water, than 100% MeOH or CH3CN, 100% THF and than back to water). A couple of time resulted - for half a day or one day, other times did not resulted. It was before we had the "good" THF.

Later, once after a column began to split the peaks with one of the different lots of THF, we reversed it and used it with no problem with a rest of the "good" THF batch. So it do seems that the problem is the THF...

Thank you both the answers!

[Mark Tracy]

I think what happens to the column is the bad solvent clogs the spaces between particles at the head of the column, forming a plug. If you keep using the column, all the excess pressure is concentrated at the top of the column where it compresses the bed underneath, leading to an irreversible headspace. By catching the problem early and backflushing with clean solvent, you forestall the irreversible damage.

[kkovats]

My question is: why it does not happen with a specific THF batch? because it does not happen with it...

[Mark Tracy]

This is pure, unfounded speculation: a bit of polymer dissolved in the THF then forms a gelatinous precipitate when you mix the mobile phase. By the way, are you filtering your mobile phase? If so, what are the filters made of?

[leadazide]

"I solved the problem by preparing a more conc sample in 100% ACN. This preciptated the HPLC, which sank to the bottom of the flask.

Wow... Pretty strong stuff to precipitate a HPLC!

Just kinding.. I just couldn't help commenting it when I read this.

Meassuring the peroxide value in a bottle that has been open for 2 months might not give the correct value of peroxides in the fresh THF. Try ordering a new THF from the same lot (if available) and test the fresh THF.

[HW Mueller]

Kati, could you ever repeat the good results with the "good lote" at a later time with either a new column or one which already produced the split peaks?
If not, my guess would be that there is an "idiot relationship" between the THF and your results. I would keep looking for some action, methodology, etc. that could have introduced some dirt.
Also you might be more successful with a more powerful washing solvent, THF, CH2Cl2, hexane? It might even be interesting to see what happens if you wash a bad (split peaks) column and a new one with a "bad lote" THF.
Did you ever replace the frit and a bit of the stat. phase of a bad column?

[Albany-12303]

OK, so my Carbomer, HPMC hypothisis was wrong

Can you use a guard column?

[kkovats]

HW Mueller,

Yes, we repeated the good results. After the validation work, when we damaged several columns and then I made the investigation about the THF batches and concluded that the validation was made with only one batch, I looked for that specific batch in another laboratory (we have three labs in our factory) and found an opened bottle with some rest of THF. We used that THF with success for two days.

Then we tried the Aldrich THF and the same columns that worked well for two days with the good THF, splitted the peaks on the 4th injection (each injection of 50 min.). Then we reversed the same column and returned to the good THF (our last mililiters...) and the column worked again without any problem.

As for the washing with a more powerful solvent, when we followed the supplier's washing methodology it contained or CH2Cl2, or hexane or both. After washing in the reverse sence, the column worked well (normal sence) for some injections, then failed again.

The problem is that with these "bad THF" batches, the columns hardly work for some hours, we can not perform a complete analysis. So we can not define that after each analysis the column should be washed with a specific methodology.

What do you mean by "It might even be interesting to see what happens if you wash a bad (split peaks) column and a new one with a "bad lote" THF".

We have never replaced the frit because in the acse of the columns we used it is not possible to open the column and replace the frit.


Albany,

Yes, we could use guard-columns, although before the validation (before the good THF batch) we had damaged several of them, too, till we ran out of stock. It happened to be right when the problem dissapeared (when the good THF batch appeared), and since we had urgency in the validation, we made the complete validation without guard columns (only a robustness analysis with guard column). After the validation we did not use guard columns.


Leadazide,

We measured the peroxid in the two months old solvent because it was the rest of the "good THF". The point is, that even with much higher peroxide content than in the "bad THF" batches, the "good THF" worked well.

I contacted the supplier but they answered that they were out of stock of that specific batch.


Mark,

As I wrote above, we have not been able to catch the problem early enough...

As for the filtration - we always use HPLC grade solvents and usually do not filter them. But before the validation (before the good THF), we tried to filter all solvents and the problem maintained. then when we did not have problems with the good THF, we stopped filtering the solvents.

We use solvent compatible filters for solvents and water compatible ones for water. I can not tell at the moment what are they made of because today we had holiday and I am writing from home. Tomorrow I can come back with this information.

(I hope I answered all questions...)

I know that you can think " they do not use guard columns, they do not filter the solvents, what do they want?" - and you are partly right. But we had the same methodology with the good THF and had no problems.

The next step we can do is to use guard column, filter the solvents and see if the column survives at least a complete analysis. If it does, we can wash the column with strong solvents and try a second analysis...

Thank you all again!

[ym3142]

the following is sure:
1) the splitting was caused by the garbage accumulated in the column;
2) THF does polymerize;

the following is not sure:
1) did one of the good injection(any one) leave some garbage in the column? SO it was a bad run;
2) how many bad injection will cause a bad column which splits peaks

to be continued.

[HW Mueller]

What I meant above is what happens when you wash a "good" column with 50-100mL (or more) "bad" THF, is the column still "good" afterwards? And if you take a "bad" column and pass "bad" THF through it does it get better or worse?
That brings out another question: If you really have "bad" THF, can you clean it up by passing it through a column (maybe even a "bad" one)?

(If manufacturers now resort to proprietary opening of columns maybe we have to saw them off and put different fittings on.)

[JM]

Hmm!!! Interesting problem

Could you pass your bad THF ( or better mobile phase using bad THF) through a C18 cartridge or through a simple bed of C18 prep grade( may be 10-20 micron) material? discard the initial few ML and use the rest. check the condition of bed if any change ? ppt?residue?

one more question ? did you use bad and good THF on same HPLC system or different?

Dont give up, there must be a reason for this.

JM