optimum helium linear velocity

[ece]

What is the optimum helium linear velocity? I have seen on the Van Deetmer curve that the optimum helium linear velocity is 20 cm/sec., yet I have seen others on the forum suggest 35 cm/sec.

Why is that? Is this because of back-pressure from the FID and carrier gases that requires a higher carrier gas flow rate to compensate? If so, how was this determined?

[Don_Hilton]

The optimum linear velocity is about 22 cm/sec when using helium. But you will notice the shape of the van Deemter curve. And, the height equivalent theoretical plate (HETP) is about 0.3 mm in an 25 m x .25 mm x .25 u OV-101 column, measuring the value on heptadecane at 175 degrees C. (From High Resolution Gas Chromatography by K. J. Hyver and P. Sandra, third edition). The HETP does increase as you increase carrier velocity. At 35 cm/sec. you may be up to 0.31 or 0.32 mm. And you have considerably shortened your GC run. So, you have a lot of forgiveness if your flow is on the high side.

Secondly notice the number of conditions stated for the measurement of the van Deemter curve. This is isothermal and a specific compound (so it has a specific k value -- a function of temperature and the compound). Because k changes from compound to compound (which gives us separation) and with temperature and because HETP is a function of k, we don’t expect the curves to be identical from compound to compound and temperature to temperature. And when we run a temperature ramp – temperature is changing (thus also k) through the run.

I’ve seen the van Deemter curve plotted for several different compounds. The minimum seems to fall in the range from 20 to 23 cm/sec. For most separations running a bit fast does not degrade the chromatogram enough to notice. I’ve encountered this only once – and the peak in question was so small that it was barely distinguishable by mass spec and not quantifiable. So with a faster gas flow - and the peak gone, did I really lose anything?

[ece]

Ok....I don't think I understand the reasoning behind why 35 cm/sec is still the conventional, because the Van deetmer optimum is still 20-22 cm/sec. I understand that different compounds will have different k values and that this curve utilizes a standard, but no one has yet explained still why 35 cm/sec is standard for helium when people likely separate different components.

[Don_Hilton]

Short answer: You get no practical difference in column efficiency and you get your work done faster.

[mbicking]

First, the van Deemter curves will change somewhat with column dimensions (diameter, film thickness) and retention (k). All of these parameters are in the expanded van Deemter/Golay equation.

The value of 20 - 25 cm/sec reflects "typical" operating conditions. As noted, you need to pay careful attention to what assumptions are being made.

Regarding the common practice of operating at 35 cm/sec: if you calculate both N and tR for a range of flows, you will find that the value of N/tR (the number of plates generated per second) increases at flows faster than the minimum on the curve. At some point (usually about 1.5 - 2X the optimum value), this plot flattens out, and there is no benefit at faster flows. This point on the N/tR plot is called the "Optimum Practical Gas Velocity" (OPGV). Many methods are set here because it represents the combined benefit of both efficiency and time. There is nothing magical about it otherwise. If you need the absolute highest efficiency, and time is not important, then work at flows representing the minimum of the curve.

[ece]

so the general consensus is that separation efficiency doesn't change significantly but faster flow rates mean faster analyses.

thank you very much everyone who answered!

[lmb]

Dear ece,

I presume that, speaking of linear velocity, you mean parameter that is measured as

(linear velocity) = (column length) / (hold-up time)

Let also be certain that we are speaking of capillary columns, and that optimal linear velocity is the one that minimizes the column plate height (H), and, therefore, maximizes its plate number (N).

Optimal linear velocity of helium in 25m-0.25mm column with FID is about 35 cm/sec. Using any velocity between 25 cm/sec and 50 cm/sec will have only a minor effect on plate number and on peak resolutions. However, 50 cm/sec significantly reduces analysis time compared to the time at 35 cm/sec.

If 25m-0.25mm column with FID and helium as a carrier gas is all that you need to know, stop here.


Following is a bigger picture.

Linear velocity is not a good measure of optimal flow of carrier gas because it is a VERY COMPLEX function of
a) column length
b) column diameter
c) outlet pressure
and this function is different for different carrier gases (helium, hydrogen, nitrogen, etc.)

It is not true that helium velocity of 35 cm/sec is even close to optimal for all column dimensions and for all outlet pressures.

Examples:

1. Optimal linear velocity of helium in the same 25m-0.25mm column, but with MS as a detector (outlet at vacuum) is about 47 cm/sec.

2. Optimal linear velocity of helium in 1m-0.1mm column, with MS as a detector (typical secondary column and detector in GCxGC) is about 235 cm/sec.

When it comes to column optimization, carrier gas flow rate (measured in mL/min at 1 atm and 25 ºC regardless of actual column temperature and outlet pressure) is much better choice than the gas linear velocity. For any carrier gas, optimal flow rate is
a) proportional to column diameter
b) does not depend on column length
c) does not depend on outlet pressure

Agilent recommends the following default flow rates for 6890 GC and newer products:

Recommended Flow Rate = f*ID

where ID is a column internal diameter, and f is gas-dependent parameter that takes the following values:
f_hydrogen = 10 mL/min/mm
f_helium = 8 mL/min/mm
f_nitrogen = 2.5 mL/min/mm
f_argon = 2.2 mL/min/mm

Example: Recommended flow rate of helium in 0.25 mm column is
(8 mL/min/mm)*0.25 mm = 2 mL/min

Recommended flow rates are 40% higher than flow rates that minimize the column plate height. These flow rates provide the shortest analysis time for any efficiency that you can obtain (by changing column length) in a column of given ID.

You can also check my previous posting on similar topics. The most relevant are posted in:
viewtopic.php?t=6468&highlight=
viewtopic.php?t=5486&highlight=

Hope you find this useful.

[Peter Apps]

OK, we have been here before (I wonder why the people who keep posting this question never look in the archives ). I try not to take up time arguing with people who know more than I do.

There are some nuts and bolts practicalities to consider.

If the volume flow rate is set to optimal at any given temperature and is maintained at that setting throughout a temperature programme (by whatever means) it will not be optimal at any other temperature.

Similarly for (average) flow velocity.

If the inlet (column head) pressure is set to give the optimum flow (of either kind) at a given temperature, the changes in flow due to temperature changes compensate for changes in optimum flow (something I learned from lmb's earlier posts ).

Doesn't this take us back to the good old days of pressure regulated inlets, with constant pressure during a run , before the fancy programmable EPCs that maintain a constant flow of one sort or the other ?

Does this mean that GCs can now become simpler and less costly

Peter

[lmb]

Dear Peter Apps,

This is a delicate one. Let me try.

Limited data suggest that optimal (volumetric) flow rate (Fopt) is proportional to T^(-0.6) where T is column temperature (Fopt drops in proportion with T in 0.6th power).

Therefore, in constant flow mode, the ratio, F/Fopt, of actual flow rate (F) to Fopt increases in proportion with T^0.6. For example, if F was optimal at 50 ºC (323 kelvin) then, at 250 ºC (523 kelvin), F = Fopt(523/323)^0.6 = 1.34Fopt (at 250 ºC, F is 34% higher than Fopt).

In constant pressure mode, actual flow rate (F) drops in proportion with T^(-1.7). Therefore, the ratio F/Fopt drops in proportion with T^(-1.1). For example, if F was optimal at 50 ºC (323 kelvin) then, at 250 ºC (523 kelvin), F = 0.59Fopt (at 250 ºC, F is 41% lower than Fopt).

One can conclude that, in a temperature-programmed analysis optimized at the beginning of temperature program, whether it is the constant flow or the constant pressure mode, there is a mismatch between optimal and actual flow rate at the end of the run. How significant is the mismatch? It follows from Golay formula for column plate height that factor of two departure of F from Fopt reduces the column plate number by 20% and, therefore, reduces peak separation by 10%. This might be noticeable, but generally it is not a big deal. And we are talking here about a rare factor of two mismatches which can be avoided if necessary (for example, one can choose such constant pressure that flow rate is optimal in the middle of temperature range of temperature-programmed run).

One should also keep in mind that, at any given time, many components of a sample can simultaneously travel through the column. The flow rate that is optimal for one component might not be optimal for all others. Optimal flow rate is simply some reasonable compromise between many conflicting requirements.

All in all, departure of flow rate from optimal one during a temperature program is not a real practical concern. (Some vendors offer constant average velocity mode of operating temperature-programmed GC. I was never able to find out what is it for, and even what does the constant average velocity mean). In my view, the whole issue of the need to faithfully maintaining exact optimal conditions during temperature-programmed analysis is highly exaggerated. A practical need in it almost never exists.

So why not always use the simplest constant pressure mode?

I know of at least two reasons for using constant flow mode.

First, detectors prefer constant flow mode.

Second, constant flow mode can be faster without loosing total peak capacity. Switching from constant pressure to constant flow always reduces analysis time. However, it usually comes at the expense of lower peak capacity. (It is not a great achievement to reduce analysis time at the expense of separation. Let me do no separation, and I will accomplish it in zero time). Nevertheless, I think (no experimental confirmation) that using constant flow mode with proper temperature programming can significantly reduce analysis time (maybe up to 50%) without loosing peak capacity.

And there is question about the very need for the electronic pressure control. Being already spoiled, I like it. But this is a subjective opinion.

Does this answer your questions? You can find more info on these topics in these sources:

1. Grob, K., "Increasing GC Separation Efficiency by Electronic Pressure Control?", J. High Resolut. Chromatogr. 1994, 17, 556.
2. Blumberg, L. M., Berger, T. A., Klee, M. S., "Constant Flow versus Constant Pressure in a Temperature-Programmed Gas Chromatography", J. High Resolut. Chromatogr. 1995, 18, 378-380.
3. Blumberg, L. M., Wilson, W. H., Klee, M. S., "Evaluation of Column Performance in Constant Pressure and Constant Flow Capillary Gas Chromatography ", J. Chromatogr. A 1999, 842/1-2, 15-28.

[lmb]

And, you might add, by sharpening the peaks at the end of the run, constant flow mode mode reduces (improves) detection limits. Why I keep forgetting it? lmb

[Peter Apps]

Hi lmb

Great stuff, thanks.

It seems that we need EPC to be able to programme inlet pressures to maintain optimum (compromise) flows throughout a temperature programmed run.

Nonetheless I often get the feeling that complexity is added to instrumentation simply because it can be - and that the added features are then pushed as being a necessity for efficient lab performance - constant linear flow programming being one such.

Another feature that strikes me as odd is the advent of back-pressure regulation, which coincided more or less with the appearance of EPC. The benefit of constant column flow rates in the face of leaks seems a relatively minor one in the face of the extra cost and complexity and vulnerability to contamination of the back-pressure system, but as far as I know there are no GCs on the market now with a pressure regulator upstream of the inlet and a simple robust heatable needle valve to regulate the split flow. With EPC this could easily be pressure programmed to provide optimum volume flows. Do you have any thoughts on this.

Peter

[Peter Apps]

Hi lmb

Great stuff, thanks.

It seems that we need EPC to be able to programme inlet pressures to maintain optimum (compromise) flows throughout a temperature programmed run.

Nonetheless I often get the feeling that complexity is added to instrumentation simply because it can be - and that the added features are then pushed as being a necessity for efficient lab performance - constant linear flow programming being one such.

Another feature that strikes me as odd is the advent of back-pressure regulation, which coincided more or less with the appearance of EPC. The benefit of constant column flow rates in the face of leaks seems a relatively minor one in the face of the extra cost and complexity and vulnerability to contamination of the back-pressure system, but as far as I know there are no GCs on the market now with a pressure regulator upstream of the inlet and a simple robust heatable needle valve to regulate the split flow. With EPC this could easily be pressure programmed to provide optimum volume flows. Do you have any thoughts on this.

Peter

[lmb]

Thanks. I have to be out for some time.