Some times ago there were some opinions about reducing column dimensions obtaining the same resolution and considering this as method adjustment and not modification.
Supposing that the modifications are within the USP guideline, what do we need to do to implement the new column in a GMP laboratory? A method equivalency study/report should be done, and I think that it would be enough for an assay method. But for a related substances method? Shall we not do LOD/LOQ for the impurities?
I would appreciate your comments/opinions, and eventually some bibliography if there is any about this subject.

Kati,
When changing to a smaller column and doing nothing else, you make the causes of band spreading have different relative importance. Mixing at connectors, in tubing, and in detectors may cause your resolution to degrade enough to cause a problem. Run matrix and standard samples to prove that it does not cause a problem, (and document per GMP.) If it does cause a problem, then you may have to change the diameter of the tubing, the injector, the flow-through cell, etc., to reduce the dead volumes. Of course, you still must validate your change and document it.

Showing resolution and LOD remain unchanged offers strong proof to the auditor that you have not made a change that could affect the quality of your product.
Jim

If you decrease the i.d. of the column with a
commensurate decrease in the tubing /connections
volumes you should decrease the LOD/LOQ and very
likely increase the resolution. The detector void
volume will remain the same so the difference
won’t be as great as it could be.

If the above proves to be true it would seem to a
logical person that this constitutes a method
adjustment rather than a method modification i.e.
a change requiring a revalidation. However in the
present climate most labs revalidate rather than
take any chance that a FDA objection will add to
the "time to market".

not possible to increase resolution, only possible to decrease.

Re anon,May 25-and the reason is?

A couple points regarding the comments above. If everything is scaled accordingly (including injection volume and cell volume (but not cell path length)) you won't observe any improvement in LOD/LOQ by going to smaller column formats. Smaller formats only result in improvements in mass detection limits but not concentration detection limits. Improvements in mass detection limit are only of interest in cases of very limited sample size.

Second, I agree with anonymous above that there is no way that miniaturization will result in any improvements in resolution. Chromatographic efficiency is unrelated to scale. The best you can hope for is equivalent resolution. But, of course, as scale diminishes relative extra column effects increase so in most cases resolution actually gets worse, especially for early eluting analytes.

I had thought that resolution was proportional to the square root of the plate number. in that case, since smaller i.d.s generally give larger plates, shouldn't the resolution increase with smaller i.d.? that is assuming that all other parameters are scaled down to make sure that extra column band broadening isn't the limiting source.

Resolution usually increases with smaller i.d. columns with GC but not with HPLC. I believe the reason is the difference between carry gas (GC)and liquid (mobile phase).

Jim,

I guess I shouldn't it been quite so dogmatic in my earlier comment. Perhaps you are referring to open tubular capillary applications. In this case, of course, the rule of thumb you mention is correct: smaller ID results in increasing plates per meter. However, open tubular capillary HPLC is rarely utilized because useful performance requires capillary IDs to be comparable to column packing particle diameters (i.e. useful capillaries must have IDs in the 5-10 micron range!). However, in the case of packed columns the diameter of the packing material particles rather than the diameter of the column defines the chromatographic efficiency. This is true regardless of whether the chromatographic technique is GC or HPLC.

In HPLC columns, reducing column ID requires reduction of flow rates to achive the same linear velocity and pressure. Extra-column effects can increase peak width thus, reduce resolution.
Reducing column length can decrease retention times loosing plates and therefore resolution.
To improve resolution you need to decrease particle size. You can go from 250 mm x 4.6 ID and 10um particles to 100mm x 2.1mm ID and 3 um, reducing run times, solvent consumption and maybe increasing efficiency. You should change connectors to 0.17mm o better 0.12mm. In my experience it is no neccesary to change the UV cell unless you want to use shorter columns (50 mm). You should also reduce sample volumen to inject to 5-10 ul. In my opinion this would requiere full validation of the method.
Marcelo

If we are not thinking about capillary HPLC columns and capillary HPLC systems, the story is very straightforward. For a given linear velocity (flow rate divided by cross section), the column performance does in principle not change. However, in reality, the total separation performance of a smaller diameter column will be inferior due to extra-column bandspreading. Even standard columns on a standard system do not perform as well as they should. Have you ever measured the plate count as a function of retention? It usually increases with increasing retention. However, there is no good reason that the column should so such a thing. The main effect is simply extra-column bandspreading. This effect becomes worse as you decrease the column diameter. You can change the system, to make it less pronounced, but it is a lot of work: reduce the tubing length, diameter, detector flow cell. Some of these things may not even be worth it, since the detector sensitivity decreases with smaller cell volume. So within the world of standard packed columns, you are not likely to get even the same performance when going to a smaller diameter.
In the context of the original question about regulatory guidlines, it is possible to change the column diameter. You should be able to find the information in the USP. I don't have one here at home, so I can't help with this.

Here is the concern I had w decreasing the column
id (e.g. to a 15cm x2.1 mm i.d. from a 15 cmx4.6
mm id) and (assuming) injection of the same sample
mass. The peak concentration will be increased,
the peak height will be increased and therefore
(using peak height measurements) the LOD and LOQ
will be reduced.

I am having trouble with a small id C18 column. I am trying to move a UV method to a MS method. In the UV method I use a C18 column with a diamention of 4.6 x 150 mm, 5 um. In order to achieve a similar column efficiency (30k), I bought a 2.1 x 100mm, 3.5 um C18 column. After a few injections of my sample, the column pressure started to build up, and soon exceeded threshold of max pressure of 5500psi. The system shut down. I tried to clean the column with everything the manufacture had suggested, such as THF, MeCN, and water. Now the column is completely blocked. The column is dead now. Any suggestions on how to unblock the column?

There is a frit at the column inlet. If you remove
and clean it (ultrasound or HNO3) that may help.
During the procedure you'll be compromising the
column integrity so this should be the last
resort.

To B. Buglio: all of this is fine if you do not get a loss in resolution that prevents you from doing this on a smaller diameter column. However, if you have enough sample, you should just inject 5 times more and get to the same spot without the danger of additional headaches.
To Cindy: there is no reason why the smaler diameter column should have died any faster than the 4.6 mm i.d. column. Think about what you did differently, and maybe that will teach us what the cause of the high pressure could have been.

Hi B. Buglio and Uwe,
Thank you for your time and suggestions. The column is dead now. I almost throw it away. I have nothing to lose except time. I will try to take it apart to clean it.

Uwe, that is a good point. I do not want to lose another column. For that column I don't think I did anything different when I tried the samples. My mobile phase was 60/40 water/MeOH. At first the pressure was about 2600 psi when I ran the mobile phase through at 0.2 mL/min. Soon the pressure increased to 3500 psi after a few injections of my samples. After the column pressure went to 3500psi,I changed the mobile phase to 10/90 water/method to clean the column. When I tried to run 60/40 again, the pressure shot up to over 5000 psi and the pump shut down. I could not run 60/40 anymore. Then I back flushed the column with water, ACN and then THF. It did not work. I was never able to run 60/40 water/MeOH anymore.

Here is a question for people with experience using 2.1 mm columns. We use 3 and 4.6 mm columns and have guard frits in our systems before the column. The idea is if there is anything which would plug up the column frit, the guard frit would catch it and it is much easier to clean or replace. Is this something that can be used with 2.1 mm columns or does it tend to introduce too much dead volume? If it can be used, this might help Cindy in the future. Cindy, do you filter your sample solutions before injecting them?

Just for completeness - before working on the
inlet frit, have you positively identified the
column as the problem and not a blockage in the
injector, connecting tubing, or detector? By
disconnecting the above in a logical sequence the
cause of the backpressure can be positively
located.

Russ, I did not filter the samples before injecting. But I was using a 2.1x10 mm guard column in front of that 2.1x100mm column.

B.Buglio, I am pretty sure the problem was that column. When I took away the 2.1x100mm column (the column only) the system only had about 200psi pressure (with the guard column). When I put the column back on the pressure shot up again. That's why I blame on that column.
Thanks guys.

Thanks everybody the answers!
Returning to my initial question, as we are using 250x4.6 mm 5µm columns, I was thinking to reduce only the column length to 15 mm, without reducing id. Is it possible? I mean, if the resolution does not get worst maintaining the same flowrate, I would save time. Or is this not so "simple"? Eventually I could try 15x4.0 (3.9) columns and reduce flowrate...

Other things being equal, cutting length from 250 to 150 mm will decrease your run time by 40% and decrease your resolution by 23%.

Reason: run time is a linear function of column length (60% of the length -> 60% of the time). Resolution is a function of the *square root* of the length (60% of the length -> SQRT(0.6) = 77% of the resolution).