Viscous Heat Generation in UPLC

[Alex Buske]

Andreas,

Temperature can play an important role and the extrems are seldomly used. That might be because column thermostats are usually restricted to something like 20-60°C. I just imagine the backpressure at -20°C.

Alex

[Newman768]

As I wrote before, I have quite often the effect that with increasing temp seperation isn't enhancing.
With uplc I am limited to a flow rate of less than 0.6 ml per min when using the long column (100 mm) - and only ACN as solvent, no MeOH. I am forced to use in most cases the 100 mm columns so MeOH is only possible at higher temp - but in such cases the seperation is lost.
That limites the use of the uplc system to a small area of seperation methods and also limited the use of the uplc to 1/4th or 1/5th of possible flow rate.
For the majority of methods running on the uplc system I use 30°C - every once in a while I also test higer temps, normally with no seperation.

So it is a logical step to think about the use of other columns then Acquity columns. One idea is the use of monolithic material which MAY be in the future able to fit the seperation of 3 or 2 µm silica material. With the lower pressure the next logical step is thinking about temperate - I guess there are quite some methods where seperation egts possible at lower temp.

I don't have tested 2 µm silica columns with the uplc but I expect at least while determinating impurities to get better seperation than with the Acquity columns - imho most of the time selectivity in those cases is more important than theoretical plates.

After all, I am using the uplc whereever possible, but in the moment I see some restrictions, most of them resulting in the miniaturizing from 4.6 mm ID to 2.1 mm (nothing new for us, we knew that from ms-analysies), the restrictions of pressure (as mentioned above) and the general problems of high speed lc (what is not restricted to the waters uplc system).
Some problems will be solved in the near future, some problems demands a rethinking how to solve samples and some problems may be solved with technics which does not exist now.

[Victor]

Andreas,

What is the optimum flow rate for your compounds on the UPLC system?

Most people run 4.6mm 5um columns at around 1ml/min which not too far from the optimum flow. The equivalent flow rate on a 2.1mm column is a factor of 4.79 less (about 0.21 mls/min). The optimum flow on a 1.7um column is likely to be higher than on a 5um column of the same dimensions, admittedly. But I guess you should be able to acheive the optimum flow at less than 0.6 ml/min. How fast do you want to go?

I guess also you mean acetonitrile-water when you say "only acetonitrile". Otherwise I share you rconcern.

[adam]

Uwe

Again, I must respectfully disagree. I hate to do it because I know you are very knowledgable on these issues - and you have been very helpful to me in the past. But I do think monoliths have been improved - such that there performance is better that what you indicate. Below are a few references that, I think, will help to make this point (please note: it seems that "monolith people" are mostly chemical engineers, and they speak a somewhat different language than us chromatographers). The first two are the grandfather papers for the 'equivalent particle diameter' concept.

Thanks Adam

U. Tallarek, F.C. Leinweber, and A. Seidel-Morgenstern, Chem. Eng. Technol., 25, 1177 (2002).
- In the conclusion they indicate an equivalent particle diameter for pressure calculations of around 9.5 um and an equivalent particle diameter of around 1.0 um for performance - or van Deemter - issues.

F.C. Leinweber, D. Lubda, K. Cabrera, and U. Tallarek, Anal. Chem., 74, 2470 (2002).
- Table 5 indicates an equivalent particle diameter for pressure calculations of around 15 um and an equivalent particle diameter of around 3.0 um (on average) for performance - or van Deemter - issues.

F.C. Leinweber, U. Tallarek, J. Chromatogr. A, 1006, 207 (2003).
- Table 6 indicates an equivalent particle diameter for pressure calculations of around 11 um and an equivalent particle diameter of 1.0 um (on average) for performance - or van Deemter - issues.

[Newman768]

to Victor:
you're right, I am running water and acetonitril in gradient mode.
Ideal flow rate for the 2.1mm ID and 1.7 um particle column begins at 0.4 ml and ends up at around 1 ml per minute (data taken from the waters seminar paper).

to Adam:
you started the thread by asking about observing trouble with the high pressure on the uplc.
The high pressure has some big effect to the temperature within the column (so it seems to me), but that is no negative effect.
When I start initial conditions pressure shoots above high pressure limits (14500 psi - this pressure is higher than normal), but if I increase flow over some time (1 to 2 minutes) from 75% to 100% of initial conditions, pressure won't go above limit (meens: stays normal).
I have observed pressure "drops" of up to 2000 psi.
Within a sequence the pressure is on "normal" values.

I think that this is realted to the friction coming from the higher operating pressure. The friction heats up the column and with higher temp pressure drops. Starting form zero flow to stable pressure conditions I need about 10 minutes.

[Uwe Neue]

Ulrich is referring to different aspects of the Chromoliths than what I am referring to. At least, we agree on the backpressure. I think that you see this as well. So we are OK with that?

Ulrich is referring to the steepness of the curves obtained with monoliths as representing 1 micron particles. This may be correct, I have not checked on this. However, he is not taking into account the equivalent of the column packing term in packed columns. This is what I am referring to, when I say that a Chromolith barely makes the performance of a 5 micron column. This is what really counts, not how steep the curve is.

I have seen over and over again in the past that people have claimed for a particular type of device that the change in performance with flow rate was much superior. I have always commented on cases like this, that this perception arises because the performance was so bad that it barely could get worse when you increase the flow.

This is not quite the case for monolith, but to claim the performance of a 1 micron particle is definitely not appropriate, when the minimum of the van Deemter curve is at 13 micron, as shown by Kele and Guiochon.

[Uwe Neue]

Andreas
If you are close to the minimum of the van Deemter curve, an increase in temperature does not gain you anything in separation performance. From what you are describing, you may be there.

[HW Mueller]

Hm, I would have guessed that temperature shifts the position of the entire van Deemter curve so that if one is at optimum conditions it would still show a temp effect?
Also, if I remember correctly, UPLC gives an advantage over HPLC of around a factor of ~2, so maybe a factor of ~3 in comparison to monolithics? Worth all the $$$?

I still think that manufacturers, etc., should go more into affinity chromatography, with weak affinities (so that one does not get the typical "all or nothing" phenomenon of present affinity chrom.).

[Uwe Neue]

Hans:
If you have a messy chromatogram, and you can suddenly see things that you have never seen before, I think it is worth it.
If your analysis used to take 15 minutes, and you can get the same resolving power now in less than 5 minutes, I think it is worth it.

Of course, it all depends on what you need.

[HW Mueller]

Uwe,
I can agree if the other things mentioned would get equal or even a bit more effort.

[Alex Buske]

Uwe
You wrote:
"If you have a messy chromatogram, and you can suddenly see things that you have never seen before, I think it is worth it. "
I doubt that one. With a 25 cm 4, 3.5 or 3 µm column you can get 20000 - 25000 plates. Is that possible with UPLC? I have a collegue who even used two 25cm 3µm columns in series. Would that be possible in UPLC?
Having 50000 or so plates would be worth it. Even at higher costs.

[tom jupille]

Uwe
With a 25 cm 4, 3.5 or 3 µm column you can get 20000 - 25000 plates. Is that possible with UPLC? I have a collegue who even used two 25cm 3µm columns in series. Would that be possible in UPLC?
Having 50000 or so plates would be worth it. Even at higher costs.

Depends on how you calculate costs. I fired up my old copy of DryLab and plugged in some typical values. Bear in mind that the pressure and plate values here are optimistic, because they assume perfectly monodisperse packings. This assumes about 60% ACN/water at 24 degrees, and a 4.6-mm id column.

25-cm, 5-micron
flow: 4 mL/min
N = 14,000
Rs = 1.7 (for the example file I used)
ΔP = 3,000 psi
Run time = 7 minutes

Double the column length and drop the particle size:
50-cm, 3-micron
flow = 4 mL/min
N = 58,000
Rs = 3.4 (i.e., roughly double)
ΔP = 16,000 psi
Run time = 14 minutes

Drop the flow rate to get the pressure back down to the original level:
50-cm, 3-micron
flow = 0.75 mL/min
N = 64,000
Rs = 3.6
ΔP = 3,000 psi
Run time = 74 minutes

That's a bit more than twice the resolution at a cost of about a 10X increase in run time.

Basically, we are pressure-limited. There are currently two evolving approaches to improving the situation:
- get more pressure (UPLC)
- change the tradeoff between performance and pressure (monoliths).
Both will require new instrumentation.

[Uwe Neue]

Alex: a standard 10 cm UPLC column under standard testing conditions yields 22 000 plates. The key thing though is that it gets you to this high performance in a shorter time.

We have application examples showing much improved resolution for long gradients of complex samples, commonly of biological origin such as metabolic profiles or peptide analysis. With a reasonable run time (30 to 60 minutes) you can now see things that you could not see before with larger particles in the same time frame. Of course, you could run a 4 hour gradient to get to this also with the larger particles, but there is a limit of what people are willing to do and how much time one wants to spend on a particular analysis.

The same is actually also true for real-life isocratic analyses.

[unmgvar]

Uwe,

>2 particle size columns can't be SEC colunms, they don't have the geometry for that. monolith columns on the other hand have it even thou it is not their main "separtion power". >2 particle size do the work throu absorption only. i do not mean that they cannot separate proteins and peptides (they do it with absorption) but, in comparison monolith column do it better i believe.

If detector's rate are not of an issue then why do Agilent and Waters introducing a new 80Hz detector? the detection issue is critical for both small particle size and monolith columns. as you decrease peak width, increase flow rates or both at the same time, you need to increase detection power. hence the need for faster detectors. i wonder thou the S/N they get now. Saw a similarity check done by agilent in one of their advertisement. i would never have aproved such a check to one of my analysts. they must have used all the "tricks" in the book to make this one stick.

Anyway, as Tom Jupille reminded us, the "battle" is basically between monolith columns to >2u particle size columns. smaller particle size is just more of the same. Monolith are far more innovative (doesn't make them better of course).

in the long run it is should be by far easier to polymerise a greater surface area in the same column i.d. then to have to press it.
monolith columns also provide us with a greater internal dwell volume and therefore less back pressure. exactly the problems that >2 particle size column have.

and with all that said i will not get started on HPTLC. which is also a very good solution for all of our troubles. especially in saving time and solvent consuption, and cost per batch. we are just to much in love with our HPLC's to even look that way.

Newman768 ,
you must know of course that at those flow rates according to the Van-Deemter plot your >2 particle size column performs barely better then a 3u column. you need higher flow rates in order to get better results. i am guessing that your column is too short and that your product gets out too soon. but then the longer the column the higher the pressure.

Alex Buske,
you wrote: Having 50000 or so plates would be worth it. Even at higher costs.
then why don't all QC labs have a few MS in them?
Costs per batch are very much a factor for opting a certain technology. >2 particle size instrument increase the costs in general even if you can show that in some points they do save money. the overall picture is costlier.
Monolith column also have a big drawback with the huge increase in solvent consuption.

and still i hold my bet. in the near future, people will be connecting monolith column to UPLC instruemnts.

[Uwe Neue]

unmgvar: While nobody at this moment is going for this, I do not see a fundamental problem why one should not be ably to make SEC columns with a particle size under 2 micron. The pores of SEC columns are typically 300 to 1000 A. Monoliths on the other hand take their advatage from a larger interstitial space than packed beds. Therefore there is little room left to create the pore space needed for SEC.

With respect to detectors: the mere fact that detectors with a higher speed capability have been introduced should tell you that detector speed is an important aspect of UPLC. The people that make detectors do not go out of their way to create higher-speed detectors, if there is not a good reason for doing so.

There is less material in a monolith than in a standard column. Therefore a monolith will always loose in terms of retention compared to a packed bed, if the comparison is made on an equal footing.

However, it is not worth comparing monoliths to UPLC since the current monolith technology neither provides the same separation power nor the same separation speed nor the same sensitivity as UPLC.