Dear friends;

I would like to hear some comments and/or real experiences related to the following problem.

A year ago, I validated an Ion-Pair HPLC method using columns prepared from two different batches of bonded silica. New columns obtained now, from the same supplier and packed with the same silica batches, give only 50% of the retention obtained before.

The manufacturer claims that according with his test (reversed phase test mixture) all the columns are equivalent. We have retested the old columns again and they still give retentions close to those originally obtained. Therefore, we have eliminated the possibility of mobile phases, instruments, and reagents being the source of this problem.

Has anyone seen this sort of problem before?, apparently the columns behave different in one test (ion-pair), and are equivalent in the other (reversed-phase). Is anyone aware of this subject being discussed in the literature?.

Thanks

Benjamin

A factor of 2 is a drastic difference, outside of what I would consider to be a possible variation of the bonded phases. We need to explore your story a bit deeper though.

1. You validated a method with "columns prepared from two different batches of bonded silica". How do you know that you got indeed two different preparations of bonded phase? I assume that you used one column to develop your application. The other column, from the other batch, was it brand-new? Or was it used?

2. You obtained columns now "from the same silica batches". What does this mean? They need to be the same batches of bonded phase, not a new bonding on the same silica. If indeed you get different results from a new column obtained with exactly the same packing material from exactly the same bonded phase batch, then the issue is on your side, and we need to do a troubleshooting of your method. If this has not yet been proven to be the case, you should ask the supplier to get you a column of exactly the same batch (=preparation of bonded phase) first. Since you think that you have used two different batches, it still should be possible to get at least one matching column.

You and your supplier have to be absolutely clear on what you understand by "the same batch". Unfortunately, I often find confusion around this.

Uwe;

Thanks for your comments. I thought my description was clear. Let me add some words to the points you make.

1.- The method was indeed validated using columns from two batches of bonded silica. How do I know this?, well the columns were slightly different, and the test data attached to them clearly specified batches A and B. Both were used in the validation. We routinely do this to have some confidence that the columns are reproducible.

2.- After almost one year, we bought columns from the same batches, A and B (not from new bondings on the same silica), and the results are very different (shorter retentions).

3.- The interesting part comes from the fact that in our hands, the original columns employed in the validation (last time used around May 2002)still give adequate results. In my mind this eliminates the possibility that mobile phases or instrumentation are causing the differences observed.

4.- I talk to my supplier just about everyday. There is no confusion as to what adosrbent batches we are talking about it. What we can not agree on is who is doing something wrong or where the problem is.

5.- The supplier claims that their reversed-phase column test gives very much the same results now compared to those about a year ago (once again we are talking about batches A and B). My analysis is something totally different (gradient, ion-pair separation, and I am getting shorter retentions (about 50%).

My interest in hearing comments about this problem is to know if anyone has experienced similar difficulties. Have you seen columns that are equivalent when tested in reversed-phase, and very different when used in other modes?

Benjamin

Benjamin
You have elimated the usual questions. There is no principle reason, why the two types of columns should give you results that are that different. They are from the same batches. Therefore they should give you identical results. Since they do not, we need to check why.

There is the possibility that the first group of columns has undergone changes during the development of the method, while the new columns are brand new. Under normal circumstances, you should have found this in the comparison of the columns from the two batches. Unfortunately, this is not the way things work sometimes, and both columns may have been used columns. Now today, you are getting two brandnew columns et voila - the retention times are different from the older columns, because they are new.
Along the same lines is a thought that is related to the equilibration with the ion-pair reagent. I am sure that you are aware that this process is not instantaneous, especially since you are using a gradient together with ion-pair chromatography.
One thing that you may do is to go back and check the records, what you did with the old columns, either to get them equilibrated or aged. If the old columns were "new" at the time of the validation study, you should be able to duplicate the results by following the exact protocol as in the older set.
I also assume that your manufacturer has not had these two "new" columns sit around in his warehouse for a year, but that they were packed recently.
There is no reason whatsoever for two identical columns to behave differently, if they have been treated the same way.
Keep us posted....

Uwe;

The method has been validated and studied in just about all respects, including column equilibration, robustness, etc. Our problem now, is to explain why the new columns have such reduced retention.

During the development of the method, and the validation (we split this work in two parts), I used at least 6 columns (all new ones) and all behaved very much the same. As I used the column, they gave shorter retentions, which is something you normally observed during extended use in most cases.

The supplier has informed us that the columns are packed on request when we order them, they do not keep them on the shelf. We have always used a column reequilibration time of 15 minutes. The results all looked good.

The experiment I described in my previous message, using one of my old columns (employed in the validation), with the new instruments and mobile phases, and giving adequate retention, clearly points out to the new columns being different.

As to why the differences?, well, I have a possible explanation, but as you can imagine the supplier does not accept it. I think that this packing material and related bonded phase may not be stable. The manufacturer disclosed some of the chemistry and groups involved in this, and there is the possibility that some changes could take place.

Have you ever heard of such problems?, and let me ask you once again, have you had any experiences with columns being equivalent in one mode and different when tested in another?

Thanks;

Benjamin

Hi Benjamin,

We had once a similar problem and our explanation was an extension of what Uwe said in his last reply.

Could you tell us if during the method development in your first columns you have using (much)more hydrophobic ion-pairing reagents? If so maybe you have modified your first columns permenately. A small amount of these hydrophobic ion-pairing reagents maybe are still in your first columns and contributes in the retention of your compounds while you assume that all the retention is thanks your shorter ones. Of course the new columns that didn't have the same treatment they show much less retention.

Now if you test both of your columns in the reversed phase column by using non-ionic hydrophobic compounds you won't see any differents in the performace of your columns. I would suggest that you try to see the retention of your analytes in a mobile phase which do not contain ion-pairing reagents but use as much water you need in order to have some retention (even if this is 100% water with some of formic acid, assuming that you have ionisable cationic analytes).

If my theory is correct, you might see "unjustified" retention of your analytes with your first columns.

As Uwe said, keep us posted...

Kostas

Kostas;

The columns I employed in the validation were always used with the same amount of ion-pair reagent, therefore it is unlikely that they were changed in any way. A common practice in the ion-pair mode is to always flush columns at the end of the day with a clean solvent. This practice should have taken care that nothing was being retained. Only standards were injected in them.

I can not test retention without ion-pair reagents because these compounds are not retained without them. The manufacturer tests the columns with a mixture of non ionic compounds using only ACN/H20 as mobile phase. By this test all columns are very much equivalent, but my analysis says otherwise.

Thanks for your comments;

Benjamin

Benjamin,

The aging of a packing on the shelf in a dry state would be very unusual. At least it would be very unusual for standard reversed-phase packings, which is what I assume you are talking about. Other packings, e.g an amino packing, would be more prone to show changes in the dry state, exposed to air. Contamination of a packing can happen from containers, but again, this is unusual to be a problem in a one-year period.
I would still think that the fact that your supplier gets the same retention after a year rather excludes contamination or aging of the packing at his side. He also packs the columns fresh, so there is no aging of the column in the packed stage (which I know to happen).
There is another possibility which is the opposite of what you are thinking but which could explain all the facts that you gave us until now.
Assume that the supplier is packing the column fresh, and then tests them with the usual acetonitrile-water mixture. Assume that he ships the column with this solvent. Until you use the column, it may usually sit around in your storage room for a while. Then when you start using it, it has aged, for example showing more silanols. All the columns that you had used in your method development have gone through the same aging process, so they give you reproducible results, and you are happy. Now, a year later, you need new columns. You order them, they arrive with overnight delivery, and you use them immediately. You now get short retention, because they had not sat around in your storage room for half a year aging away, happily creating silanols...
Is this a possible scenario? Neither you nor your supplier are really guilty of anything...


I should mention that this may be a possible scenario, but the effects that you are reporting are larger than what I would expect from such an event. We may still need to look at other possibilities...

Uwe;

To the best of our knowledge, all the columns we have purchased, were packed fresh and used or tested immediately by us. This supplier can not afford to have columns packed on his shelf waiting to be sold in the future.

We know somethings about the nature of the bonded phase, it is not the straight C8 or C18 type. It has some polar groups in it.

I have doubts about the correlation between the reversed-phase test of the supplier, and any other user test like ours. Perhaps the fault is at the reversed-phase test not being strict enough, this is to say for example that perhaps a small Tr difference by reversed-phase, could translate into a large difference in retention by some other mechanism (ion-pair).

Thanks for your comments.

Benjamin

Well, if there are funny side groups, strange things can happen. I am giving you my e-mail, so we can continue this discussion off-line, if you so choose to.

Benjamin,

I would try to get a handle on how the columns are different. Like Kostas suggested, I would get out of ion pairing mode and test your columns with a column test mix, not your compound. I would not use the manufacturer's test that is designed to give good numbers. I would use a test mix that is designed to detect differences in columns.

There are a number of test mixes in the literature or you could try NIST SRM 870:
http://patapsco.nist.gov/srmcatalog/certificates/870.pdf

This test mix has the advantage that is has been run on a large number of commercially available columns. If your column is one of these then you might be able to detect which columns are behaving differently, the old ones or the new ones. A test mix might give you information like that the old colums with more retention for your compound have more silanol activity, etc. Good luck