Dead Time Measurements of Custom-Packed Columns

[dkreller]

Hi Chromatography Forum,

For some geochemical work that I am doing, I am packing my own columns (stainless, 100x4.6 mm) with sorbents that consist of different coatings (metal oxides, and clays) that are deposited on quartz sand. The quartz sand has a relatively narrow size distribution centered around 300 microns. Huge, by chromatographic standards, I know. But I am not doing yer usual chromatography either.

Someone took a look at some of my experimental data and expressed concern that what I was seeing (systematic band broadening over a series of injections) was somehow due to inhomogeneity in the columns.

So I made a bunch of columns and measured the column's dead times by NaNO3 injections. The average of 8 columns was 0.9 min with a RSD of 3.5% I understand that this variability is of course a lot greater than commercially-available columns that are packed with finer particles, but to me this doesn't seem too bad. However, I thought I'd put this out there for you to respond too. Is this variability horrible, considering my packing material, and how I pack it, i.e. dry, by pouring sand into the columns?

Also, I am really a bit confused by this person's (a reviewer, to be exact) assertion that my band broadening results could be associated with column packing variability. Basically, in experiments (which are carried out with single columns) we see that the chromatographic peak from our analyte/adsorbate gets systematically broader. I think, and wrote in our manuscript, that the bands get broader because there are weak interactions between our adsorbate and the surface which has been modified by the adsorption of our adsorbate/analyte that weren't there prior to the adsorption.

I can see that there is some variability between columns, but I can't see how someone could assert that all this band broadening, which is very systematic over a series of a large number of injections, is meaningless and could be easily chalked up to packing variability... How could there be packing variability within a single column?

Thanks for your consideration,

David

[tom jupille]

First off, 3.5% RSD on dead time for 8 dry-packed columns filled with sand is not bad at all.

Second, it's not unusual for packed beds to settle and develop void spaces in use, with the result that peaks do broaden over time; so the reviewer's questions are not out of line.

One way to check: since you're using nitrate as your void volume marker, you (presumably) know that it has no chemical interaction with your packing. Therefore, if the band broadening is due to chemical factors (secondary interactions), the width of the nitrate peak should remain constant. On the other hand, if the band broadening is due to physical factors (bed settling / packing problems), then it should affect the nitrate peak as well as your analyte peaks.

[dkreller]

When I was doing the nitrate elution dead time experiments I was seeing some systematic drift in the dead time measurements for single columns-for example, the first DT would be 0.910, the next 0.906, then 0.901, 0.899 etc. So I now appreciate your comment - about how DTs for packed beds can increase over use. If I have rigid 300 um particles, how long do you think it would take the column to settle? Actually I think I will do an experiment to get at this information - just keep injecting nitrate over a 24 hr period - the timescale of my experiments - and see how much DT drift band broadening there is.

In my data there was clear broadening under one set of conditions that wasn't seen under the other conditions. So I think I am understanding this better now. I will post again once I have had a chance to do more work on this and overlay nitrate chromatograms.

Thanks again,

David

[Uwe Neue]

What you are saying NOW is getting strange. When you have a column that is closed, the total volume does not change, and the amount of stuff that is inside the column does not change. Then how can the retention volume of your "unretained" peak get smaller, if there is no volume change in the column? The retention change of the NaNO3 must change for a different reason, or your peak shape is changing and the tip of the peak is migrating due to this phenomenon...

[dkreller]

I wonder, did you read Mr. Jupille's post before making this comment?

- ....it's not unusual for packed beds to settle and develop void spaces in use, -

-DK

[HW Mueller]

dkreller, I don´t see that Tom said anything about a change in rt due to the void. He seemed to have talked only about broadening. One might expect a change in rt if the settling produced a lot of fines which went right through the frit, out of the column.

[dkreller]

Sir,

The way I read it, Tom's first post directly mentioned indicated that there can be changes in void volume as a column 'settles'.

Second, it's not unusual for packed beds to settle and develop void spaces in use, with the result that peaks do broaden over time...

[/quote]

I can imagine that as the column gets used, that its packing efficiency increases.

[Peter Apps]

Unless you are losing solid particles (by dissolution or washout of fines) the total void volume cannot change (as Uwe points out). What happens as a column settles is that you get a large void at the inlet end of the bed, and on average the rest of the particles are packed more tightly together.

Peter

[dkreller]

Gentlemen,

OK, it's me that was not really fully reading the comment that peaks do broaden over time. I apologize. Please bear with me.

I can show you the type of band broadening that I felt was real and was related to adsorbate-adsorbent interactions on an experimental column; take a look at panel C of the figure here http://chemistry1.che.georgiasouthern.e ... dening.gif and compare those chromatograms to those in A, B, and D. Does it look like that kind of broadening could be due to column settling?

Does anyone know of a standard reference to a well established book or article that discusses the phemenonon of peak broadening due to column settling?

Thanks for your time and consideration,

David

[mardexis]

Hi David,

I could imagine that the broadening you showed in Frame C could be due to a void in the case that there were two main channels of flow. Typically, a large void will just cause a large tailing peak.

I would suggest that you measure the first moment or center of mass of the peaks. If the broadening is due to a void volume or other inhomogeneity of packing then the center of mass should remain the same. If the broadening is due to interaction between your probe molecule and the surface then you ought to see a small increase in the center of mass of the peak.

In order to make your data easier to interpret and demonstrate visually there are uniform glass beads available in the small micron size region that might be used. I can point you towards a source if you are interested.

Good luck,
Marc

[Uwe Neue]

What are the differences between the different chromatograms that you are overlaying?
I am seeing a shift in the pattern. For a column that has a void, the pattern would be rather independent of the injected amount...

[dkreller]

Hi,

These are repeated injections of a solution of an organic substance (my adsorbate) onto a substrate (the adsorbent) to which the organic substance is known to bind in a Langmuir - like interaction. There is a limit in the amt adsorbed. The chromatograms are from successive injections of the adsorbate onto the adsorbent-packed column. All the injections are identical in injection volume and all are from the same stock solution. My interpretation of this data is that as the adsorbate builds up on the surfaces of the sorbent particles there are new weak interactions between the adsorbate molecules in solution and the adsorbent-coated surface. There weren't such weak interactions prior to adsorption, i.e. between the adsorbate molecules and the uncoated surface.

I appreciate this discussion, I think I am finally understanding this better now. Does this sound correct, in summary? - If the void volume of a column gradually increased over the course of an experiment then it woud lead to a gradual broadening of the peaks - but not a change in the shape of peaks. -

Thanks again,
David

[Uwe Neue]

Hmm - I am slowly understanding what you are doing. In which direction is this moving? Do you get the larger peak area at the beginning or at the end of the study?

[dkreller]

Hi,

The peaks are systematically growing in area over the course of the experiment, so the biggest peaks are from the last injections. My interpretation is that we detect the fraction of the injected quantity of adsorbate that elutes and is not sorbed. Across injections the surface sites get increasingly consumed, and there is less and less adsorption, i.e. more and more eluting and detected. Finally, if the adsorption is Langmuir - like there is 100% recovery and the peak area hits a plateau.

DK

[Uwe Neue]

OK, here is my interpretation: you get adsorption, and the more you inject, the more you saturate the surface until at some point when the surface is saturated, you get a consistent band shape. I think that the difference between the pattern in graph C compared to the other graphs could be due to a lousy bed structure or some other non-uniformities. However, even if it is a lousy bed, this has nothing to do with the observation that you get an increasing signal with subsequent injections.

If you would have a packed bed with a much smaller particle size, you would see nothing at all until you have saturated the entire surface. You could also run the columns with a much slower flow rate to avoid seeing the breakthrough, or to at least reduce it.