Kinetex columns/extra-column volume/instrument design

[bisnettrj2]

Journal of Chromatography A. Volume 1217, Issue 18, 30 April 2010, Pages 3000-3012; "Achieving the full performance of highly efficient columns by optimizing conventional benchmark high-performance liquid chromatography instruments". Fabrice Grittia, Carl A. Sanchezb, Tivadar Farkasb and Georges Guiochon

Above is an interesting article regarding the Kinetex family of columns by Phenomenex (specifically, the C18 versions in a 2.1*50mm and 4.6*50 mm/100 mm columns). The authors concluded that the optimization of the systems running these columns could allow the use of these Kinetex columns with efficiencies similar to UPLCs in a standard Agilent 1100 system.

My questions, to the really smart guys out there:

1. The math went over my head with a really loud 'whoosing' sound. For those that get it, is it sound? Any complaints?

2. Any problems with the methodology? The experiments?

3. The closing statement of the paper is one I found intriguing:

"It is becoming necessary to design and build instruments that provide lower extra-column band broadening contributions than the newest high-pressure LC systems, lower by one order of magnitude. Without these further improvements, analysts will not be able to take full advantage of the next generations of high-performance columns."

In light of the statement, are there any objections? Does the statement hold water for 'regular' sub-2-micron columns? Will there be a time when the columns are theoretically much more efficient than the instruments can realize, due to design issues? Are there any ideas on how to reduce band spreading/extra-column volume in new versions of instruments, like a comlete re-thinking of the way instruments are arranged/plumbed?

Thanks in advance for your consideration.

[tom jupille]

I like the use of Douglas Adams' "whooshing sound" line in this context!

The math is indeed complex, but the "bottom line" makes sense:
- what we see in terms of peak width (peak volume, actually) is the combined effect of what happens in the column and what happens in the rest of the system
- in a well-implemented system, the extra-column contribution is negligible compared to what happens in the column (I'll call that the "intra-column contribution" to band broadening).
- as the columns get smaller, we have to decrease the extra-column volume in proportion.

As a "back of the envelope" estimate, a peak eluting at k'=3 from a 50 x 2.1mm column with 10,000 plates will have a baseline width of about 12 microliters. If you have 5 microliters of extra-column band broadening, that will increase your peak width (i.e. drop your resolution) by 10%.

Now, let's look at an Agilent 1100 or Waters Alliance-class instrument. A standard UV detector cell is about 8 microliters, and standard 0.010" id tubing has about half a microliter of volume per centimeter of length (which means that if you have 6 inches of tubing -- call it about 15 cm -- between injector and column plus between column and detector, you've got another 7 or 8 microliters. If you're injecting 10 microliters, that brings the extra-column total up to about 25 microliters. So yes, we'd like to push that down by an order of magnitude or so!

I suspect that a lot of people lose sight of the fact that this is not due to the use of small particles per se, they are just an enabling technology that let's us get a lot of plates out of a short column (and in a short time). The narrowness of the column helps sensitivity, but I suspect the real reason for going down in diameter is to help with heat dissipation (all the energy that get's pumped in to keep the solvent flowing against a high back pressure has to go somewhere!).

[bisnettrj2]

Thank you for your response, Tom.

Personally, I'm interested in using the Kinetex columns (but probaby not the Halo columns, at least not until Agilent and AMT finalize their lawsuit...) in my analysis, so this paper gives me the ammo to retrofit one of my 1100's to a low-volume setup so I can better utilize these columns. I have a 4.6x150 C18 Kinetex that I tried to use, but the extra-column volume on my system was too great to utilize the column to its true potential. And me asking my boss to give me 2k to buy upgrades to a system that *might* help me utilize the column and thus decrease my analysis time was a bit of a stretch without some backup.

I think most will agree that anything that happens post-column (including detection in a UV-cell) is going to have a negative effect on the separation which occurred in the column, and is thus counteractive and should be eliminated. That being said, it seems (at least to me, in my very limited experience) that Agilent is stuck on it's current system design - modular stack, top-to-bottom orientation. Since post-column volume is going to, inevitably, become more and more important as column technology advances toward smaller particles and smaller column diameters, are there any ideas out there towards decreasing the extra-column volume (and pre-column sample dilution and gradient delay volume) to as little as possible? Maybe re-thinking column design to somehow eliminate the distance between injector-column and column-detector? Perhaps re-designing the system to eliminate not only extra-column volumes, but also as many user-generated (e.g. - potentially bad) connections as possible? Or is there an apparent limiting factor in the technology currently available in terms of instrumentation that will limit the analytical chemist in their abilty to use the newest columns?

Oh, and Tom - thanks for all the fish!

[Bruce Hamilton]

It depends how standard you want your system to be. Certainly the 13 uL volume of the standard 1100 DAD flowcell is not really amenable to Kinetex.

But the next size down ( 5 uL ) also has a path length of only about 60% of the standard cell. If you go even smaller ( 0.50 uL ), you can recover the path length ( 10.0 mm versus 9.8 mm for Std cell ).

However, your system should also be optimised for the low flows that the cell may restrict you to. The high pressure cell ( 1.7 uL ) might become a superior option, if not too expensive. Agilent certainly offer a lot of information on the available options.

It would be easy to use the 0.12 mm ID lines between column and detector, and also run lines from the cell holder downwards, and also bypassing the connection holder, saving another 100 mm of line.

Not sure whether it's possible to get a single unbroken line from cell to column, possibly for permanent installation. Injection issues shouldn't be too bad, because careful sample solvent selection may help concentrate sample at head of column.

However, if you require a system that's going to be flexible, you may have to compromise. Ultimately, in a production environment, it's usually better to purchase the low volume options from the manufacturer, and possibly a new instrument matching your needs.

With the reduced price of acetonitrile back to more normal levels, the economic rationale isn't so compelling to beancounters.

Bruce Hamilton

[HW Mueller]

Havn´t these problems been solved by manufacturers of capillary electrophoresis and those of "lab on a chip´" apparati?

[Bruce Hamilton]

My understanding was that the issue was retrofitting an Agilent 1100 to handle Kinetex columns, not to build a new instrument.

I've tried some of my suggested steps on my 1100 with a 150 x 4.6 Kinetex column. I currrently don't need to reduce my instrument's flexibility by using lower volume cells, but others might.

The issue of going further outside the design envelope causes extra costs, eg higher data acquisition rates, low volume injections, higher system pressures, etc. etc.

However, at the end of the day, in a regulated environment, the superior option may e to purchase a more appropriate instrument - but that's a significant cost.

[bisnettrj2]

A little bit of both, actually. I need to get ahold of Agilent and price out the appropriate MWD/DAD cell that is a good compromise of volume and path length, and the tubing I can handle. Oh, and I wish the cost of acetonitrile was a real driver in my lab's purchasing decisions. We only use MeCN for extraction of my samples, and I don't account for a whole lot of the lab's total business.

I am interested, however, if anyone has heard plans of any (or sketched any on the back of a napkin) further system redesigns to further reduce extra-column volume in analytical instrumentation? CE and the 'chip-lab' are pretty specialized and not as robust or sensitive, from what I understand (but I don't have any experience with them, so I could be wrong - it's happened before and will again). Since part of the paper talked about the ability of instrument manufacturer's to ignore post-column volume while particle sizes were large, but now that particle sizes are trending dramatically smaller, are there any whispers out there about radical system or column hardware redesigns to eliminate this ECV? I mean, if you can design some sort of system to have a ZDV connector from injector to column and column to detector, then the problems with using small particle columns (and transferring a method developed on such columns to larger particles) should, I think, become much more managable. As the paper points out, with further column advances, there will be greater apparent losses in efficiency even with the low-volume UPLCs on the market now. My question is more of a thought-exercise on the future of analytical HPLCs, so if you have any ideas, I think it would be interesting to hear them. This is the Water Cooler, after all!

[tom jupille]

Actually, that kind of reduction is done fairly frequently in labs that do high-throughput Lc-MS/MS (mainly bioanalytical work). The column (typically something like 30 x 2.1 mm) is butted right up against the ionization source and connected by the minimum length (a cm or so) of 0.003" id tubing. The injector was removed from the autosampler and mounted on a bracket immediately in front of the column, again with a minimal length of 0.003" tubing, and a pillow heater was placed around the whole thing to maintain temperature. That adds extra tubing from the autosampler assembly to the injector valve, so some sample gets wasted.

All of that said, I'd be surprised if you had to go to that much effort for a 150 x 4.6 mm column. The problem is not small particles, it's the narrow (in volume terms) peaks. A 150 x 4.6 mm has about 15 x the volume of a 50 x 2.1 mm, and shouldn't be *that* sensitive to extra-column volume.

It's a bit of work, but you *can* actually measure the "true" plate number of the column along with the extra-column contribution to band broadening. What you have to do is to run a set of at least three "well behaved" compounds (i.e., no tailing problems; something like alkyl phenones, or benzene/toluene/xylene). Tweak the mobile phase to get a reasonable spread of retention and then measure the peak width as a function of retention volume (VR).

Peak width can be expressed in terms of σ units (the square root of the variance; baseline width of a Gaussian peak is 4σ). If we plot σ^2 (the variance) versus VR^2, the intercept should equal σec^2 (the extra column variance; take the square root of that to find the extra-column contribution to σ); the slope should be the reciprocal of the column plate number, N.

This is only an approximation, because it assumes that all peaks have the same plate number (which they don't), but it give as reasonable estimate.

[bisnettrj2]

Tom, that setup you described is almost exactly what I had sketched on the back of an envelope. Re-inventing the wheel is fun...

Part of the problem I've run into is the lack of ability to run the column at a fast enough flow rate to get to the theoretical ideal flow rate for these columns using methanol:water as my mobile phase. Were I running ACN, backpressure wouldn't be as big an issue. I think if I were able to get my hands on a modified 1200 SL or an Acquity H-Class, I could have fun. I'll keep plugging away, though, when I get the time. Thanks for the input!

[Uwe Neue]

Let me throw a small (?) wrench in your enterprise. The pressure required to reach the minimum of the van Deemter curve is a function of the particle size, the MW of the analyte that you choose, and its retention factor, but it is largely independent of the solvent composition. So it does not matter whether you are using acetonitrile or methanol.

[bisnettrj2]

Uwe, I'm not sure I understand what you mean by your statement that it doesn't matter if I'm running acetonitrile or methanol. I didn't think pressure had anything to do with the van Deemter minimum, that it was essentially a function of flow rate (linear velocity) for a particular column, as measured with relatively small analytes. So, unless the optimum velocity shifts with different mobile phase systems (does it?), then to reach the same optimum velocity with a methanol:water mobile phase as with an acetonitrile:water mobile phase will take more pressure, for the most part, given the higher viscosity of methanol:water than acetonitrile:water (depending on your mobile phase composition, that is, but maybe that's what you meant?). Given that I have pressure limitations on my system, I can't achieve a mobile phase velocity high enough to operate the column within the recommended flow rates for optimum efficiency by Phenomenex for the Kinetex columns at the mobile phase proportions I'm using to achieve optimum separation, so I have to run a little slower.

[Uwe Neue]

It is actually very simple. The van Deemter optimum depends (at equal k', and constant particle size) exclusively on the diffusion coefficient of the analyte. The diffusion coefficient is inversely proportional to the viscosity of the mobile phase. The backpressure is proportional to the viscosity of the mobile phase. So the backpressure that you need to get to the minimum of the van Deemter does not depend on the viscosity any more - it cancels out.
You can look at the algebra in J. Sep. Sci. 30 (2007), 1158-1166.
This is a very useful rule of thumb: the backpressure that you need to run a column is independent of the mobile phase composition, but depends of course on the the MW of the analyte (peptides have lower diffusion coefficient and therefore require less pressure to reach the van Deemter minimum), a bit on the retention factor, and a bit on the temperature (you need higher pressures at higher temperature (!!!)).
Interesting, isn't it?

[bisnettrj2]

I'll take a look at the article sometime today - nothing like some light reading of the J. Chrom during lunch. BTW, regarding the statement I put in bold at the beginning of this thread, do you, Uwe, feel there will come a time where analytical instruments will need to be overhauled to remove more extra-column volume than has already been removed in the current UPLC-type instruments, in order to keep up with the advances in column technology? Or has the field of liquid chromatography hit the apex of development in the analytical LC field, and we'll be looking at another decade before the next big thing (UUPLC - ultra-ultra-performance LC?) hits the market, where we'll be using sub 1-um columns and 1 minute runtimes with 50 nL injections? In any case, thanks for the input, and I'll take a wrench in my enterprises from you anytime.

[Bryan Evans]

The problem with using particles < 2um is it becomes very challenging to pack long (high resolution) columns.
In addition, pressure becomes way to high for today's instruments. It's not practical.

[Uwe Neue]

Bryan, I understand that your company has not yet figured out how to make sub-2-micron particles, and that this is the reason that you must insist that there is no advantage to these sub-2-micron things.

Here is the problem: in a comparison between the Halo particles and Waters sub-2-miron particles in a kinetic plot (were the column length is varied), it turns out that the cross-over point where the Halo particles match the performance of the sub-2-micron particles is around 200 minutes, which is at analysis times exceeding 3 hours. For anything faster, the 1.7 micron columns outperformed the larger particles.