Peaks Out of Order

[GCHanf]

Been running the same method for years and the peaks have always come out on the chromatogram in a specific order. Recently, when we received new (specialy made) columns (C18 3u) the peaks are coming out in a different order. Does anyone know of any reason this might happen.

I am thinking the columns are bad, but the system suitability is passing.

[danko]

Funny, that the system suitability accept-criteria do not include elution order, or at least retention time reproducibility. I’m just curious as to what does indicate that the system suitability test passes?
And now to your question: The specially made column does not demonstrate the same selectivity, as the one you used to utilize before. A C18 column is not just a C18 column. There are additional properties that contribute to the specific selectivity and other parameters.
What was so special about that column anyway – particle size, column dimensions, pore size?

Best Regards

[GCHanf]

Funny, that the system suitability accept-criteria do not include elution order, or at least retention time reproducibility. I’m just curious as to what does indicate that the system suitability test passes?

It does, but if they integrate the peaks individually and then do the math it passes (sort of I guess).

And now to your question: The specially made column does not demonstrate the same selectivity, as the one you used to utilize before. A C18 column is not just a C18 column. There are additional properties that contribute to the specific selectivity and other parameters.
What was so special about that column anyway – particle size, column dimensions, pore size?

Yeah, it's a short column 80mm x 4.1mm 3 micron 120 angstrom (Swaged Hardware)

Best Regards

[Bruce Hamilton]

I am thinking the columns are bad, but the system suitability is passing.

Finally, a rational system suitability test!. If the separation works, and you can identify the peaks, it's all go...

The problem you have is that you haven't specified the column appropriately. That means you can get differing separations with new columns.

You need to define the column and stationary phase - you can do that by specifying a supplier's well-characterised market product that you know performs the separation ( eg Waters ubondapak, Xbridge; Agilent Eclipse XDB C18; Phenomenex Luna; etc etc.) , and/or specifying in detail performance parameters for the specific separation you wish to use the column for.

It's best to use packings from manufacturers who can show their product is consistent over many years. You don't have to buy the latest and greatest packing, just one that works and will be around for along time.

Please keep having fun,

Bruce Hamilton

[GCHanf]

Where I totally understand what you're saying, the method we are running was validated and filed with that column, which puts me in a bit of a spot. We've contacted the manufacturer of the column but are getting a little stonewalled, but they are supposedly "looking into it". We've been running this method for like 8 years and never had this problem before.

[Bruce Hamilton]

If you filed with the regulator quoting that supplier and column, and you are purchasing against the same specification, your best option is to ask both your quality group and the supplier's quality group what the outcome of their last audits for that product were. There's no point in barking when you've got dogs paid to bark.

I'm assuming that you have previously ordered replacements that have performed the same up until now.

If the supplier can show your specification is still complied with ( eg your purchasing specification just said C18 rather than a product with a well- defined specification sheet ). then you probably will have to revalidate, and also explain to your quality group how such a critical component did not have the appropriate purchasing specification.

Please keep having fun,

Bruce Hamilton

[lmh]

Any manufacturer worth their salt should be using retention time of a set of known test analytes as part of their quality control, irrespective of whether the customer asks for it.

HPLC is entirely dependent on retention time. It is the only assurance you have that a chemical is the right thing. If your columns do not have reproducible retention times, you have no business selling them. They may be useful as pea-shooters if you take the filling out, and could make useful paperweights or attractive executive desk ornaments, but that is just about all they could do.

So if your manufacturer doesn't take this issue seriously, make sure that they know you will Tell the World just how bad they are.

[HW Mueller]

Imh, did you ever hear of the fact that HPLC has to be caibrated ever so often? Even the same column doesn´t deliver the same performance throughout its lifetime.

[Bruce Hamilton]

Any manufacturer worth their salt should be using retention time of a set of known test analytes as part of their quality control, irrespective of whether the customer asks for it.
...
So if your manufacturer doesn't take this issue seriously, make sure that they know you will Tell the World just how bad they are.

Last time I looked at column certificates for columns that I've purchased, only Agilent actually listed the acceptance limits on the column certificate, the other vendors just listed the results. I'm supposed to trust them that the product passed.

[Edit]

My comment above covers columns that I've recently purchased - in the case of Waters, it was an Atlantis T3 from 2007, but Uwe advises that other Waters columns, such as their Symmetry series, do still specify the acceptance limits.

So my generalising by manufacturer may not be valid, it may depend on the product ranges of each manufacturer. Sorry for any confusion, I obviously often purchase on price.

[ End edit ]

The test compounds used by manufacturers also may not reflect the application you wish to use the column for. That's why you are expected to define the column in the User Requirement Specification, perhaps even including a performance test that is relevant to your application.

If you are in a regulated industry, your quality groups are expected to have a regular regime of supplier audits, and the best solution for a recalcitrant supplier is to set the quality dogs onto them. Being stricken from an approved supplier listing can be very harmful to any business.

A better indication of the relationship between columns from different manufacturers may be found in the freely-available online USP and PQRI databases, although I prefer the freely-available downloadable Waters reversed Phase Column Selectivity Chart software.

Please keep having fun,

Bruce Hamilton

[lmh]

HW Mueller, of course I run calibration curves with every sequence I run.

But if I buy two identical columns from one manufacturer (even if there are a few months between purchases) and I find that the peaks actually elute in a different order, I think the manufacturer has a very serious QA problem! We're not talking minor variations here.

Every column I have ever bought has come with some sort of test chromatogram. I assume peak shape/width and retention time are the parameters the manufacturer is controlling.

[ntruong]

I run into the same scenario of peak elution order as I develop and validate a method. I bought total 5 columns, all coming from different batch. Two of them give me the same elution order, the rest is not. What could be wrong? Why are the retention of the compound varied with each column?

[HW Mueller]

The original question by GCHanf talks about a specially made column, etc., which I read to say that the material was not identical to the old column.
Peak sequence changes are, in some cases, easily produced with one and the same column when you change the pH, so much for sloppy pH adjustment. (Judging from my experience one has to pretty much make a mistake, not just being sloppy, to produce such changes via organic modifyer).
So, unless one got a different generation column, etc., the cause is most likely ones own doing.

[ntruong]

Hi, HW Mueller,
I used same mobile phase on the there columns and they all came out different. It is more likely column dependence, in my case.

[Uwe Neue]

I grew up with HPLC, and I have seen all aspects of it. A bonding procedure is much more sophisticated than most people think, and in order to get reproducible results, the manufacturer needs to have control over all aspects of the particle synthesis process, from the particle itself to the final bonding. Furthermore, one must have meaningful quality control tests. Well designed QC tests tell the manufacturer how to do a good job in controlling the synthesis process.

We (at Waters) have gone through this learning process a long time ago (25 years ago). This learning experience enabled us to design highly reproducible processes, as well as novel processes. Our Symmetry product line is a prime example of what can be achieved with good process control and high-quality QC tests. Not everybody has such an experience...

It should be said though that an equal amount of care is expected from the chromatographer. Only well designed mobile phases / chromatographic conditions will give us the high reproducibility that we all want.

[lmh]

Sorry all, I think I'm misunderstanding. From gchanf's posting...

"Where I totally understand what you're saying, the method we are running was validated and filed with that column, which puts me in a bit of a spot. We've contacted the manufacturer of the column but are getting a little stonewalled, but they are supposedly "looking into it". We've been running this method for like 8 years and never had this problem before."

...I assumed this was a custom-made column identical to those that they've been having custom made for 8 years. Of course if it's a different C18, the peak order may change.

I still feel strongly that if you use a new batch of columns the same as you've bought previously, the peak order should be the same. I take Uwe Neue's point that this is far from easy for a manufacturer to achieve, but I've certainly never seen significant variation of RT between C18 columns bought from Phenomenex, and I don't for a moment doubt what Uwe says about Waters columns.