Viscous Heat Generation in UPLC

[Alex Buske]

Uwe,

I admit that my last post was somewhat provocative. I think with UPLC you can get good resolutions in short time. For people that are limited to <20 min (or <5 min...) cycle time UPLC offers more resolution.

Unmgvar,
Some QC labs have MS, usually they don't. therefore nobody develops QC-methods on LC-MS, because there are no methods, QC-Labs don't buy MS-systems.

Chromoliths in UPLC: I dont see the point. Monolith have low backpreesures, UPLC pumps are designed to work high backpressures. Chromoliths used rather high flow rates and solvent volumes, UPLC instrumtation is designed to work with minimal solvent amounts.

[Uwe Neue]

Alex,

I would go even further. For an isocratic separation, UPLC will outperform (platecount) any standard setup for analysis times faster than 15 minutes. In addition, it will always give you a shorter analysis time for any standard column set-up. This is the advantage of small particles.

For a gradient separation, UPLC will outperform any rational setup with classical particle sizes for run times up to a few hours, provided the UPLC method has been set up properly.

It is very difficult to beat the attraction of very small particles. Of course, you could get more plates if you are willing to wait long enough by using very long column with larger particles. But even under these circumstances, you will get more plates out of an instrument with a higher pressure capability than with a lower pressure capability.

Consequently it is virtually impossible to beat the attraction of the higher pressure capability of UPLC.

Of course, this is more theory and practice, and maybe we should focus on the practice....

[Chris Pohl]

Uwe,

Thus far, I stayed out of this discussion even though it is certainly of interest to us at Dionex, as I consider this discussion to be yours to argue. But since you placed the discussion in a more theoretical vein in this last post, claiming that "UPLC will outperform any rational setup with classical particle sizes for run times up to a few hours, provided the UPLC method has been set up properly". I saw a presentaion by Piotr Gzil at HPLC 2005 (you can find this presentation at http://wwwir.vub.ac.be/chis/TMAS2.htm ) which differs with this conclusion. His general conclusions (for a pressure constrained system, as all commercially available instrumentation is, of necessity, regardless of the actual pressure limit) were:

1). monoliths are superior when more than 50,000 plates are required

2). < 2 micron particles are superior for fast separations

3). classical size particles are preferred for cases were high resolution is more important than speed but which require efficiencies of less than 50,000 plates

[Uwe Neue]

Chris:
Note that I was talking about a gradient separation! I will explain in more detail later how this works.

[Alex Buske]

Chris,

The Gzil-presentation is theoretical, as
-he refers partly to monoliths that are not marketed yet,
-to get resolutions >50000N you would need a total length of monoliths that equals several columns (in that case you could consider to buy an UPLC)
He also points out that with higher pressures the picture changes, but it is not clear for me if the displayed curves are for <2mm PB or conventional material.

Alex

[Uwe Neue]

Chris:

The simplest way to look at column performance versus particle size, time and available pressure is to follow the thought process of Guiochon from his 1975 publications. He determined that the column with the lowest backpressure for a desired combination of plates and analysis time is always operated at the minimum of the van-Deemter curve. This shortens the discussion drastically for isocratic separations. You simply substitute the velocity at the minimum of the van Deemter into the Kozeny-Carman equation, and you get:

N =dp^2/3000 * pressure / (viscosity * Diffusivity)

The only way to get more plates at a limited pressure is to use larger particles. But if you use larger particles, your speed of analysis decreases, i.e. your analysis time must increase.

For gradients the story is a bit more complicated, and where small particles win and where large particles win is a question of the details of the assumed condition.

For a 60 minute and a 120 minute gradient, you will get at 15000 psi the best gradient separation for a 1 hour gradient from 0 to 100% from a 30 cm 1.7 micron column with a peak capacity of about 550, or about 666 for a 2-hour gradient. The best that you can do with a 3 micron particle is to use a 80 cm to 1 m long column for a peak capacity around 370 for a 1-hour gradient and 513 for a 2-hour gradient, also assuming that you will run these things with a UPLC instrument at 15 000 psi.

So this is the rough story on the reality of the compromises with particle sizes.

[Uwe Neue]

For an isocratic separation, the entire story has been published by Guiochon a long time ago. The addition of monoliths is an interesting twist to the game, but one alwasy needs to remember that the perfroamcne is limited due to the pressure limitations of the monolith (vide supra).

[Kostas Petritis]

Speaking of Guiochon, maybe people would be interested at a recent article by Martin and Guiochon on the effects of very high pressure in liquid chromatography J. Chromatogr. A, 1090, 2005, 16-38.

As always a lot of equations, personally I do not see any fundamental problems for the more common solvents/applications. Maybe the take home message of the article is that method development might become more complex because of the pressure dependence on everything.

[Broesen]

Hi Uwe

What about signal to noise. UPLC most be way be on Monolith in terms of mass/volume). Regarding the time, column lenght and particle size: Isn't it always preferable to have the shortest possible run time?

Peter

[unmgvar]

Alex,

It is because an MS system costs at least 3-4 times more then an HPLC system that you barely see them in QC's. also one MS system up to now took at least the bench space of 2 HPLC's if not more. the lack of long term efficiency is one of the major reaosn that MS is not a widelly used alternative. It is more likely to enter QC's due to more increasing regulations regarding the LOD, QOD for cleaning validation methods in the pharma industry then any other reasons.

As stated by Uwe, myself and others, for 4,6 mm i.d. the surface area of a monolith column is that of a 3um column with the dwell volume of a 9mm i.d column. thus the decrease in back pressure. just think how much surface area increment would be gained if you would increase the surface area of such a 4,6 column to the dwell volume we get today for 5um particle size columns? this would by far be better then a <2 um particle size column. and you get to have a "regular" back pressure. so you can still use normal HPLC's, with normal detectors, and normal injection volumes (no need of all that needle in needle stuff for example).

as the technology would get better, you would get to need UPLC instruments for monolith columns as well because of the increase in surface area and the decrease in the dwell volumes. hence my claim that in the future monolith columns would be used in UPLC instruments.
in theory, you can get more surface area with polymerisation then with smaller particle size in the same casing volume, for less back pressure and more dwell volumes. less trouble doing isocratic works (remembrer the peak broadning effect i talked in one of my first posts), easier to reproduce gradients. because of the increased dwell volume it is easier to get out the heat generated by the friction, so you can work with greater i.d's of the column.

too small dwell volumes mean very small capilaries, more clogging (especially in the column), more complex detectors, smaller injection volumes (good for some but not neccessaraly for all), less reproducability again between brands, and increased costs.

Uwe, isn't back pressure the primary problem of UPLC as well. Isn't it better to have the consumable fail due to the pressure then the instrument (which costs more)?

Also, monolith columns do not take their SEC effects from the pore size but from the "sponge" like structure created. this is what gives it that little extra effect of SEC. I certainly don't think that this is the major contribution in the separation power but it is one none the less that already give those types of columns an added value for some type of applications.

Broesen, if i could i would want to have it all all the time
shorter runs, better signal/ noise ratio, better resolutions. but the moment you improve one characteristic you have to get ready to accept some losses somewhere else. the trick is to get the best compromise between your different needs.

[Uwe Neue]

Broesen, sensitivity is for the most part simply an issue of sample dilution on the column. From that standpoint, UPLC is roughly an order of magnitude superior to a standard monolith column. Also, as mentioned above, from the standpoint of column performance (plate count and plates per unit time), monoliths compete at best with 3-micron columns, but they are not even in the same league as the 1.7 micron columns used in UPLC.
And yes, in many cases people today prefer short run times at maximum performance.

[juddc]

It's been a while since I've been in here, but I've just read the UPLC thread and have one semi-subjective data point to add to this conversation. I've recently had the pleasure of visiting Waters HQ for the first time to take their IQ/OQ/PQ course and in the course of the lab portion of the training, got to see UPLC in action for the first time and from what I saw, the puppy works and works well. It is to this chromatographer's mind at least, a sensible evolution of HPLC with the stated goal of producing more data per unit time and volume mobile phase. Not a bad thing, IMHO. I saw chromatograms that were of 1 minute duration that separated more than 7 compounds (I saw 7 majors and a number of minors). From a distance, the chromatogram looked no different from any other, just that the X-axis was very, very short. Peak widths were on the order of 3-4 seconds each. Limitations I see for this are simple: The selection of column chemistries is currently limited and you can't plug any old detector into the beast and expect optimum performance. I expect those limitations to go away as the technology is more broadly adopted. I did note that the cycle time between injections was about 45 seconds, so the "real" run time is about 1:45 for a 1 min run. Still nothing I'd complain about.

For combinatorial chemistry and labs running straightforward analyses, I see this as a fine alternative if high throughput is desired. I would think this would be great for clinical settings. I also think it would be interesting to run one of these with AMDS software and see what could be done to knock down development times. Think of an overnight run with no run exceeding 5 mins. If a column switching valve could be incorporated, a chromatographer could gather a lot of data very quickly and if he or she had done their homework and made some good initial guesses, I'd predict that a workable method wouldn't take too long to materialize.

I should buy a couple of systems and start my own method development factory...

[Uwe Neue]

This week, Agilent also announced a UPLC system, may be not quite UPLC, but at least higher pressure than HPLC. Confirms that the idea of UPLC is correct, doesn't it?

[domenger]

I just join the forum, so I couldn't read in details all comments; we do have now Hypersil GOLD C18 ( and soon aQ and PFP) in 1.9µm particles diameter. Some trials on UPLC show a better separation impedance than other materials in 1.7 µm or 1.8µm. Back pressure is not so high ( oftenly less than 1.9µm columns) and you can also use these columns on traditional HPLC's. Of course you need a good instrument as we offer only at the moment 2.1 mm ID columns. Dead volume, injection and detection are critical of course
They are short too ( from 2 to 5 cm lenght)but we already have 10 cm columns in test.
We have very flat Van Deemter's curves, so that these columns can be used at very high flow rates ( 1ml/min).
Who wants to try these columns?

Thierry

[Yury Zelechonok]

Uwe, of course the idea of UPLC is correct.
I have another idea though. SDUPLC (Supper-Duper-UPLC) with pressure limit 50,000 psi. It is optimized for 1.0 um particles and column 1 mm id. 750!!! peaks capacity in 60 min gradient run… How about that?
Let’s do it…
We always can find enough believers who can spend other people money if we make enough noise. And Agilen will joint the pack; they have their own loyal customers who have a budget and nothing good to buy in the end of the year.