Viscous Heat Generation in UPLC

[adam]

I've been doing some reading on UPLC, and I'm wondering about something: has anyone experienced any problems with viscous heat generation in the column deteriorating the separation quality?

[Uwe Neue]

Heat is generated in the column. This is the consequence of using such high pressures. However, the smaller the column diameter, the better the heat is transported out of the column and the less of a problem it creates. This is the reason why the UPLC instrument by Waters is layed out for 2 mm and 1 mm columns instead of the 4.6 mm columns used in HPLC. Jorgenson's group uses much higher pressures and therefore capillary columns.

[JI2002]

Is the heat generated by high pressure or high linear vilocity?

[adam]

I think he asked an excellent question: no one wants to answer this one?

[Victor]

If you run a very small particle column (e.g. 1.7um) at the same linear velocity as a small particle column (e.g. 5um) you will generate more pressure in the former and therefore more heat.

However, the optimum linear flow velocity is higher for very small particle columns, so you might want to run them at higher flow velocities. In this case you will generate still more heat, because the pressure required will be still higher.

[adam]

Thanks Victor

But it doesn't really answer the question at hand, which is: does the heat come from the pressure or the linear velocity.

It is an important question. Let's say that the heat comes from the linear velocity (not the pressure): which would be my guess. This would mean that the new high pressure systems could be used with longer columns, with standard packing - let's say 3 um - at the same linear velocity. Without any additional concern about viscous heat. This would allow higher plate counts, but the tradeoff would be longer analysis times.

Of course the high pressure systems can also be used with smaller particle diameters and/or higher flow rates (that's how they are marketed). But, in these cases, the concern about viscous heat exists - and it will be necessary to use narrow bore columns.

Can anyone confirm or dispute my statement?

[Chris Pohl]

The generation of heat when pumping a liquid through a packed column is not caused by either of the parameters, pressure or linear velocity per se. It's better to think of these as process parameters related to the phenomenon. It takes work to pump fluids through a packed column. Some of the energy required to do this work is dissipated (crudely analogous to friction when dealing with work at solid interfaces) in the course of pumping the fluid through the column, generating heat. For a given column geometry the amount of work is proportional to the pressure observed at a given flow rate but it's not that the pressure causes the heat, it's merely indicative of the amount of work necessary to pump the fluid through the bed of particles. Likewise, the amount of work is proportional to the linear velocity employed with a given column and therefore the amount of heat is proportional to the linear velocity but again, linear velocity does not cause heating per se.

[Mark Tracy]

To a first approximation, the rate of energy dissipation (watts) is proportional to (flow rate)*(pressure drop). The approximation is to ignore compressibility of the fluid. The temperature rise is a much more complicated phenomenon involving heat capacities, thermal conductivities and flow rate. To make a long story short, small diameter columns have less heat to dissipate and do it more efficiently.

[unmgvar]

all i am writing on this subject is purely theoretical.

i like to look at it first by going to aerodynamics.

as a plane flyes through the air the friction between him and the air at the points of friction cause a transfer of energy which give rise to an incease in temparature.

the denser the air, the more heat for the same speed. the speedier the plane the more heat for the same density of the air. for exemple in shuttle reentry there is almost no air density but the speed of the shuttle is so great that a hugge amount of heat is created from the friction.

in our case it is the same. it is the friction between the fluids and the stationary phase that cause the heat.

now heat dissiapation as another twist to it. it diffuses better as the delta of temp. is greater, and also faster as the volume of the body from which it disperse is smaller and also the ratio between the area to the volume is smaller. all of these put together bring about the fact that the heat created within the process will in the head of the column heat the fluids better then it diffuses out, only when the fluids will have reach a certain temp. the heat will diffuse faster from the column. this will cause a temp. gradient over the column and fluids. it should very much indeed influence the separation strenght of the column. it should bring about wide peak broadning problems

as the friction created is directly related to the viscosity of the fluids, working at higher temp. will cause less friction and also this heat will diffuse faster out of the column.

i stress again that this is purely theoretical. i have not worked with a UPLC.

i have also seen something else that trouble me about UPLC.
except for one aplication made with the instrument at 1 ml/min that i have read, all the others were at 0.6 and less.
from what i have seen from the Van-Deemter for those column any flow rate less then 2 ml/min, put the effeciency of the column only slightly better then 4.6 i.d.

any thoughts from somebody on the matter? if i am not wrong this also adds to the broadning effect shouldn't it?

[JI2002]

Some good points in the above posts. I raised the question of linear velocity as the heat generating factor in UPLC based on the idea that if a high pressure (e.g. 15,000 psi) is applied to a 1.7 um particle packed column (but without flow), no heat should be generated. Heat generated by running the mobile phase through the stationary bed should be common in HPLC. If it's only a concern for UPLC, the linear velocity of the mobile phase must be high enough for the heat generated to cause any separation problem.

[Uwe Neue]

Adam, you have my book and you can look up the material on page 54.

Under adiabatic conditions, the change in temperature of the solvent is proportional to the pressure, and inversely proportional to the density and the heat capacity of the solvent. This is in essence the pressure limitation of HPLC.

However, in real life, the heat generated by the pressure can be transported out of the column, and thus the temperature change for the solvent is always less than what adiabatic theory predicts.

There is good and bad news about getting the heat out of the column. The good news is that there is less of an increase in temperature. The bad news is that there could be more of a temperature difference between the center of the column and the wall, which in turn would result in lousy performance. Therefore the diameter of UPLC columns is smaller than the diameter of classical HPLC columns, which in turn requires a complete revamping of the instrument (not just the column and the pump). Bottom line is, you end up with a new instrument.

To answer another question: UPLC columns do not generate more plates, they generate more plates in a shorter time. You can do a particular gradient analysis with UPLC in 1 hour that would have taken you 4 to 8 hours with classical HPLC for the same performance. Or you can do an isocratic analysis in 3 minutes that would have taken you 10 to 20 minutes with a classical column.

[HW Mueller]

...... seem to be mentally blocked, again. The pump does the pressurizing, the full pressure difference occurs there, the heating due to the pressure change should occur there. If one were to cool the liquid before it reaches the column then the column should get even colder, especially toward its end, as the liquid returns to atmospheric pressure (assuming no resistance in the post column part).
In this connection, I wonder why I have never seen anything about a flow rate gradient along the column (quite commonly discussed in GC where this is, of course, much more severe).

[Uwe Neue]

No, the heat is not generated at the pump, but during the flow through the column. The pump generates the mechanical energy that is converted to heat in the column.
Flow rate gradients have been used on LC, but they have the disadvantage that the quantitation goes out the window, if the retention window of the peak changes.

[unmgvar]

Uwe
when you say that they genarate more plate in less time, you should actually mean distance. since the time factor is in direct relation of the flow rate used and therefore of distnce in the column.
that is why i asked my question regarding the Van-Deemter plot. according to that plot, the improvement of plate per um count of column for a 1.7 um particles is almost insignificant unless you start working at flows of 1 ml/min, better thou to be at 2 ml/min and not more then 3 ml/min.

still most application i have seen published on the matter are done at flows of 0.5-0.6 ml/min due to the short lenght of column especially, some are done at 1-1.3 ml/ min.
i was wondering if working at those small flow rate was not adding to the peak broadning effect?

as for the temparature effect, can the local effect of the temparature rise cause dommage to the column bead until the heat is transported out of the column? also as the heat is transporte out of the column by the mobile phase it should also mean that the temp. of the column not only differs from in to out but from the inlet side to the outlet side thus mainly adding to peak broadning, just like if you had big temp. difference if you do not use a solvent pre-heater for the solvents before entering the column wivh is at 15-20 degrees more then room temp.
now i am guessing that current columns are too short for those factors to become relevant as the mobile phase exits the column pretty fast, still in theory it means that there is limitation of lenght for UPLC columns. if too long then the effect would increase over distance and hence affect chromatography.
what do you think?

[Uwe Neue]

unmgvar:

you need to look carefully at the applications of UPLC. UPLC does not use 4.6 mm i.d. columns. UPLC uses 2 mm or 1 mm i.d. columns. While the minimum of a van-Deemter plot (= the maximum plate count) may be at 1 mL/min for a 4.6 mm i.d. column, it is at a roughly 5 times lower flow rate for a 2 mm i.d. column. Thus UPLC columns operated on an UPLC instrument are operated in the flow rate range where you do get the advantage of the smaller particle size, i.e. higher performance in a shorter amount of time. The columns available are 5 cm, 10 cm and 15 cm long. The first two types are for standard small-molecule analysis.

Since you are working with a smaller i.d., you are above the minimum of the van-Deemter curve at the commonly used flow rates (0.5 mL/min to 1.5 mL/min) for 2 mm i.d. columns.

The transport of the heat out of the column is the major reason why higher pressure chromatography requires columns with a smaller i.d. A smaller i.d. makes it possible to efficiently remove the heat to a sufficient level that it does not add appreciably to the bandbroadening inside the column. The heat transfer in the column has been studied carefully, before the decisions on the column diameters were make.

In the literature, significantly longer columns and sigificantly higher pressures have been used, of course with a further reduction in column diameter. Look at the publications of Jorgenson! Jorgenson had learned how to deal heat generated inside a device when he developed / invented capillary electrophoresis. He was just applying what he has learned there to chromatography, when he carried out the first experiments in ultra-high-pressure LC.