Crazy Idea

[Triple Quad]

I have two crazy ideas that I probably should keep to myself. I need to shake them off. Please tell me what is wrong with my thought experiments. These cannot be right. Point me to the flaws.

Even though I think these are crazy ideas, surely someone has tried them before I even thought of them. Both ideas have a common theme.

First idea: For a multipurpose LC setup that is extremely versatile, what if a guy was to use a five column system instead of one column. Five columns that were each 25mm long with the first column being strongly polar and the last column being strongly non-polar. Then the separations question would no longer need to consider which column. The only necessary questions would be what are the solubilities of the analytes and how polar non-polar are the analytes. The most significant drawback would be that every run would have longer run times due to the system volume.

Second Idea: When separations are tough to resolve, I have the crazy notion in my head that instead of one long column (150-250mm) two or three short columns (25-50mm) in series might resolve the analysis just as well if not better. Especially if the three column were different. If the first column were polar, then the second less polar, then the third very non-polar. My reasoning is that with two close eluters being slowed down in the column, the slightly faster one would leave the first column gain separation as is travels faster to the next column through the connector tubing. Maybe the better idea is to have 10x10mm columns for ultimate resolution, or 100x5mm columns. LOL! Carry that on out and we simply have one long column. Almost. Long columns increase the number of interactions between analyte and packing material, and create space or separation based only on interation with packing. It seems plausible that my adding an element of space to further enhance the separations, then we could have a two dimensional LC column separation.

Perhaps one column could be used with 3 packing zones or gradient packings.

The two elements are very closely related, but column packing has the effect of slowing the progress off all the analytes that pass though. The analytes are impeded by the size of the packing and the amount of the packing plus the attractive forces between analyte and packing. If we can create within the column the ability for the analytes that impeded less and give them the opportunity to run, then the separation is enhanced.

All comments and critiques are welcome. I wish I had the resources and time to experiment with these ideas.

Have a great day, my fellow separations chemists.


But I know that if we carry that on out we could think of a column as many small columns or plates.

[mbulentcakar]

Hi,
Yo may find this link helpful, which I remember Mark Tracy gave in this forum, so I added to favorites:
http://www.poplc.de/index_en.html
Good luck,

[Triple Quad]

Hi,
Yo may find this link helpful, which I remember Mark Tracy gave in this forum, so I added to favorites:
http://www.poplc.de/index_en.html
Good luck,

Thank you very much for that link. My idea isn't so crazy after all! That is really cool. I had never heard of Stationary Phase Optimized Liquid Chromatography before
...

Technically this is
realized by using a segmented column system

Very cool.

[HW Mueller]

Just because it´s on the market doesn´t mean it isn´t crazy.

[gcguy]

I wonder if anybody makes columns with varying polarity along the length of the coulmn, so a single column would offer the flexibility Triple Quad is talking about?

GCguy

[Triple Quad]

Just because it´s on the market doesn´t mean it isn´t crazy.

I didn't say because it's on the market therefoe it's not crazy...being on the market gives it some credibility. It still takes a bit of crazy to swim against the current and stick one's head out far enough to think differently about LC.

If you keep doing things the way everyone else does, then you'll keep getting the same results that everyone else gets.

From the following example:



There is the example of attempting separate 8 different triazine pesticides on four different LC columns. There isn't one of the four that works to separate all 8 pesticides.

But, using the segmented column with 4 different stationary phases, produced much better results. Not perfect but better resolution.

It looks to me that there is some real potential for the segmented column.

[Mark Tracy]

I did something like that with carbonyl-DNPH derivatives where one way used a 250mm C18 column and the other used coupled 150mm C18 and PolarAdvantage columns. Look at http://www1.dionex.com/en-us/lp49255.html for the two chromatograms. I have even coupled 75mm C18 and PolarAdvantage II with good results, but I don't have a published chromatogram. Also, we have tried EPA-8330 explosives with coupled columns and the results were pretty good.

After seeing this work, we developed a blended stationary phase especially for EPA-8330. We lot-select the two phases and blend them to get the right separation. This is the Explosives E2 column.

Column coupling of dissimilar stationary phases reduces the plate count; the more dissimilar, the greater the loss. Coupling can succeed if the improvement in selectivity exceeds the loss of efficiency.

[Victor]

Mark- you say that coupling dissimilar phases reduces the plate count. Coupling of any two columns will reduce the plate count because there must be a small dead space introduced between the 2 columns. Having said this, I have managed to couple columns containing the same phase with virtually no loss of plate count by minimising this volume. Why should coupling dissimilar phases be any different?

There are two arguments in this thread. One, as given by Mark is along the lines that one can get a phase of intermediate selectivity by linking phases of two different selectivites. This seems to be the principle of the POPL approach and apparently works fine. However, the notion that using connecting tubing will improve the separation (first post) is odd, since any such tubing will cuase peak braodening and loss of column efficiency. There are further problems with mobile phase conditions not being correct for the second column if correct for the first column. And there may be problems of the second column destroying some of the separation from the first column. Many of these problems are overcome by using 2 dimensional LC which is somewhat different

[Mark Tracy]

Victor,
First consider an experiment like what you did with identical C18 columns, isocratic conditions and a neutral probe like toluene. Both columns contribute to the retention and band-spreading equally. Now consider an extreme case: column 1 is silica, column 2 is C18. The toluene will be unretained on column 1, but you will still get diffusional band-spreading. Column 2 will provide all the retention, and will also contribute to band-spreading. At best you will get half the retention of the first experiment, and the same band-spreading. Your plate count will be about 25% as good. Fortunately this is the worst case, and one you are unlikely to ever implement for a real application.

[Uwe Neue]

Oh boy, this may be the start of a looong sequence of posts...

Mark, I got a question: if I have two columns of sufficiently different retentivity, I will get lower plates than the sum of the plate counts of both columns when I couple them. I can calculate this, no problem. But what is happening if I blend the two materials? I have never made the experiment (except for GPC, but this is different), and I can argue with myself about what the outcome should be. At this moment, the part of my brain that argues that the plate count should be as good as for an unblended column appears to have the upper hand, but the other part of my brain has not yet given up...

Did you test this? You must have...

[Alex Buske]

Hallo,

Triple Quad, would you like to have all 5 columns in series? Thats impractical because of back pressure and runtimes (and the band spreading mentioned above).
The guy from POPLC told us that coupling two columns isn't so much a problem. But with more the 3 columns efficieny loss causes roblems.

Columns with varying polarity along the column length have been mentioned. I think the could be interesting for special applications, but i definitely would prefer homogenous columns: if a separation was ok on a full length homogenous phase column chances are that it doesnt work on a segmented polarity coulm. And with homogenous column there are still so many method development possibilities.

Blending of phases should work. A small company from Berlin or Leuna offered a method development kit containg 3 different phases. Based on some scouting runs they would offer custom-blended columns (in a kind of prisma-approach. Thats again for special separations and not necessarily universal.
Alex

[HW Mueller]

This pople stuff has been discussed recently?
My brain would tell me that blending would dilute the effects of the differnt phases. Thus, I go with Victor on this: The more efficient and elegant way is multi-step (multi-dimensional). I can also see that one can get some specialty seps that way, but if one thinks that through to the end, one should mix all the available phases so as to separate everything and all on one column.....crazy in a bad sense?

[Triple Quad]

Hallo,

Triple Quad, would you like to have all 5 columns in series? Thats impractical because of back pressure and runtimes (and the band spreading mentioned above).
The guy from POPLC told us that coupling two columns isn't so much a problem. But with more the 3 columns efficieny loss causes roblems.

Columns with varying polarity along the column length have been mentioned. I think the could be interesting for special applications, but i definitely would prefer homogenous columns: if a separation was ok on a full length homogenous phase column chances are that it doesnt work on a segmented polarity coulm. And with homogenous column there are still so many method development possibilities.

Blending of phases should work. A small company from Berlin or Leuna offered a method development kit containg 3 different phases. Based on some scouting runs they would offer custom-blended columns (in a kind of prisma-approach. Thats again for special separations and not necessarily universal.
Alex

I agree that 5 separate columns is probably impractical. Perhaps two or three different packings would be enough. But, as a thought experiment, the point was to utilize more than one packing material that consists of a gradient polarity. But, thinking about the using individual short columns (25-50mm) be it 3 or 5, if each was sufficiently long enough to effect separation but not so long that pressure would become a problem. I had reasoned that band spreading would not be too much of a problem thusly; the analyte velocity through the interconnecting tubing would be fast relative to the velocity through the next column. I haven't done the calculations but that is my assumption.

Then, we still probably wouldn't be able to use an isocratic mobile phase. Although, perhaps, we could use a standard gradient where starting with the a high aqueous/high polarity mobile phase that would be appropriate for the first column segment, then ending on high organic/low polarity solvent which would be appropriate for the last segment.

This would probably be better as one long column and would be effective for resolving compounds that are very similar or compounds in the sample that are very different.

Nevertheless, by the excellent responses to my post, this is helping me to think more like a chromatographer should think. It's been a few years since I sat in class and discussed LC. I never thought that I would actually become a chromatographer. Actually, I fought becoming a chromatogrpher, now that I am in a position that requires it. It's actually been fun.

One column with the correct mobile phase and packing material is sufficient for most separations. The first project I was working on last spring really opened my eyes. We were using a 50mm column and an isocratic mobile phase at 75% methanol and 25% water. Our analytes were coming off at ~6 and 7 minutes. Changing the mobile phase to 65% changed the retention times to more than 30 minutes.

[PHOBIUS]

yes yes, the same story like this:

http://www.zlodejina.cz/kradez.php?id=11008

[Triple Quad]

yes yes, the same story like this:

http://www.zlodejina.cz/kradez.php?id=11008

I sure wish I could read Czech.