Column Calibation

[wrighty]

Hi all,

I am currently working on an Agilent 6890N GC and am in the middle of working on the column. There is a question on how accurate the current column length. When the column was installed it was not calibrated and as bits have been trimmed off we have no record of what the true length is.

We have decided to calibrate the column ourselves and have come across a couple of problems. The column in question is a DB23 produced by J&W scientific and was originally calibrated using methane. We tried to use butane but due to the bigger size of butane compared to methane the data was void.

Could anybody suggest a way to calibrate this column? We would try methane but sourcing it in the low quantities that we require is prooving hard.

Thanks in advance for your help

Steve

[DR]

Calibrating a column? No good can come from this.
There should be suitability requirements for any assay you run on any column that are specific to the analytes being examined. Once resolution between a critical pair falls off, or tailing goes up beyond a predetermined limit or a plate number drops (or when bleed gets excessive), then it's time for a new column. While it is probably a good idea to log a column's use, abuse and trims, calibrating it using an arbitrary analyte set is really of very little use.

In the case of LC columns, I have seen some columns get retired because they could no longer meet suitability for a specific assay, but they still worked perfectly well for another (frequently less demanding) assay. This means that you could possibly be inclined to throw out a perfectly good column just because it couldn't retain methane long enough (or even worse, insist on using a junk column because methane is the one thing it could retain).

[AICMM]

Wrighty,

MS or FID?

Best regards.

[wrighty]

its an FID detector.

i realise that this will be a pain of job, its just so we can rely on the flow and velocity of gas through the column for method development.

cheers

steve

[JI2002]

The right way to find out the true length of a column is by measuring the diameter of the loop D and counting the loops of the column N. Then

column length L = pi * D * N

I never heard of column calibration. By runing methane or butane, you can find out dead time to, but without knowing the average linear velocity, how do you know what's the true column length? The only way to calculate the average velocity is to know the column length. It's a catch 22 situation.

The technique people use to calculate average linear velocity(or flow rate) is to inject methane. Average linear velocity = L/to

[DR]

If (big if) you can get an accurate carrier gas vol/time value (have a very good flow meter or a cheap one & some time?), measuring time for an unretained (CH4) gas shot would allow a decent length calc. assuming your column's ID is close to the nominal 0.25 or 0.53 or whatever (it should be pretty close)...

[LocoLobo]

wrighty,

I always calculate column length the way JI2002 says. I think that the 6890 uses the column dimensions you input to calculate the correct pressure to obtain the flow you input (if in constant flow mode).

So after several clippings count the number of loops and calculate the Length. On the other hand if your chromatograms meet spec that might be overkill.

[Bruce Hamilton]

As above, I count loops and measure diameter to determine length.

The easiest source of methane is some natural gas from anywhere that has piped gas. It's usually about 70+% methane. Just ignore later peaks ( up to C6 ).

Another way would be to get a cow, as they belch around 100 - 350 litres of methane each day, mainly at dawn and dusk. There would be additional benefits, such as milk.

Bruce Hamilton

[chemikas]

Bruce

There is one problem with the source - you need 100 kg of grass every day (even if you won't need methane)

[wrighty]

thanks for all the advice.

the cow idea was a first idea but i couldnt find any volunteers to collect samples lol.

Seriously though, i'm going to give the formula a go.

i'll post an update when ive got some results.

cheers all

steve[/quote]

[gcguy]

I work with a guy that could supply all the methane you need, all he needs is 8 pints of Guinness

[lmb]

Dear wrighty,

To keep the same retention times and resolutions of all peak pairs after trimmin a column or replacing it with a new one (hopefully from the same manufacturer), you do not need to worry about changes in a column length or average gas velocity. What you need to do is to make sure that hold-up time, tM, (also known as dead time, void time, etc.) is always the same.

There are many way to restore tM after making the changes. For example, you can do this:

1. Choose any good peak in a chromatogram and use it as a reference (it might be a slightly retained peak like methane or butane, but it does not have to be slightly retained one).

2. Keep a record of retention time (RT) of that reference peak. (If you did not do it before trimming or replacing a column, try to find that RT from previous chromatograms of the same sample).

3. If, after trimming or replacing a column, RT of the reference peak changed, adjust column pressure (or flow) to restore that RT to its previous value. You might need to do several iterations because there is no good general formula for pressure correction in all cases. The only certain thing is that the pressure change should always be in the same direction with RT change (for example, if RT became shorter, reduce the pressure). Although there is no simple formula for all cases, there are recommendations for special cases.

3a. If analysis is isothermal or if the reference peak is slightly retained, %-change in pressure across the column should be the same as %-change in RT (whether a column was trimmed or replaced).

3b. If a column was trimmed, measure the length, of the peace that was cut, find its approximate relative value in relation to approximate column length. Relative reduction in pressure across the column should be twice the relative reduction in column length. This works for isothermal and temperature-programmed conditions.

3c. The most uncertain case is replacing a column in a temperature-programmed analysis and the reference peak is not slightly retained. Try to make a relative change in pressure across the column that is five times larger than relative change in RT. This might or might not reduce the number of iterations to restore RT.

Notice that, in all special cases, you deal not with pressure reading on GC instrument (gauge pressure), but with the pressure across the column. For example, in case of GC-MS (vacuum at column outlet) pressure across the column is gauge pressure + 1 atm.

Finally, as far as I know, Agilent sells RTL (retention time locking) Software essentially based on theses procedures.

Good luck,
lmb