Bizarre saw-toothed response for air and CO2 on a TCD

[pi3832]

I'm running a 1% Air, 1% Carbon Dioxide balance Helium standard on a GC-2010. The response has been highly variable, with a relative standard deviation in the 10% range.

This is, to say the least, not acceptable. So, today I ran the standard repeatedly, using a stopwatch to ensure I made a run every 12 minutes (+/- 30 seconds). When I plotted the area counts for each run, I got this very unexpected graph:



What in the world could cause an instrument to do that?!

Details:

• Helium carrier
  6-port Valco injection valve
  1 ml sample loop
  4:1 split
  Supelco, 80/100 Porapak N, 2m x 1/8" column
  Temperature program:
  • 120 C for 3 min
    20 C/min to 160 C
    160 C for 2.5 min
    -20 C/min 120 C
    120 C for 0.5 min
  TCD at 200 C with 80 mA current

Any ideas are more than welcome.

TIA.

[pi3832]

Oh, and an example chromatogram can be seen HERE.

[chromatographer1]

Mr. Freeman,

I interpret the data you presented as a consistent variable splitting of the sample.

Preliminary trouble-shooting questions:

Do you have an EPC?

If so do you see a constant pressure being mantained? or does it change during the runs.

OR, are you using a constant pressure regulator?

Do you see a constant flow rate through the column during the runs?

Once I know the answers to the above two questions I will continue.

best wishes,

Rod

[Bruce Hamilton]

Have you performed the same, or similar, analyses on the instrument before, with acceptable precision?.

My first thought would be the sample loop is being filled to variable pressures - are you using a back pressure ( regulator, or 1-2 cm water ), or just ambient pressure?.

A 10% variation in split pressures may also be possible. Can you directly plumb the gas sampling valve to the column ( bypass the injector ) and reduce sample loop volume to 0.2 ml?

Another trick to try is to run isothermally at 120C, as both peaks come off well before the gradient takes effect. I would perform injections every 4 minutes over a period, and then every 8 or 12 mins, and see if the trend is duplicated, or is related to the number of injections. If there are some late eluters, they can be baked of every few runs.

Please keep having fun,

Bruce Hamilton

[pi3832]

Do you have an EPC?

If so do you see a constant pressure being mantained? or does it change during the runs.

OR, are you using a constant pressure regulator?

Do you see a constant flow rate through the column during the runs?

It's a Shimadzu 2010 GC. It lets me either maintain a constant pressure or a constant linear velocity. Both are electronically controlled. I'm currently using linear velocity control.

[AICMM]

pi3832,

I am not familiar with the Shimadzu gc so a couple of questions. First, how is the reference gas flow controlled? Does it rely on the same EPC as the carrier gas. If so, any errors in the EPC should get compensated for by errors on the reference side, right? Second, it is every fifth injection (every hour or so) which I don't think is how you would see an EPC fail but I don't know for sure.... Third, are you using synthetic air or compressed air? If compressed air, where does the water come off?

I agree with Bruce, I would try 120 isothermal for 4 minutes for an hour and then bake at 180 or so for 20 minutes then try again. This is how one of my customers does his hydrogen analysis on a ShinCarb column in order to eliminate accumulated water on column.

By the way, did you ever resolve your H2S and mercaptans analysis needs?

Best regards.

[chromatographer1]

Mike at AICMM has hit the nail on the head.

The problem I suppose it to be will be evident by the cyclic changes in the pressure required to maintain the constant flow through the column.

I suspect that you have some heavy component of whatever you are sampling that is dewing in the column. This can be water or something else in your sample unknown to us.

As the plugs of this heavy elute from the column the backpressure of the column changes, and this affects the split ratio and this affects the area of your peaks.

You should put a back flush column into your analytical train or have an elevated temperature run to clean off your porous polymer column before injecting your next sample. Since I suspect this is a lab instrument and not a control analyzer the latter would be the easiest solution.

The best solution is to inject directly without using a split and then your air and CO2 amounts will be reproducible. Since your heavy peak lies down well as evidenced by your chromatogram that is an option. I would prefer to use a backflush column, however.

best wishes,

Rod

[Ron]

It's a little difficult to believe that this problem is caused by accumulation of material in the column that bakes out once an hour. You inject 5 times in one hour, the backpressure increases for four injections, then goes back to the starting pressure doesn't seem logical. If there was an increase in backpressure with each injection there should be an increase in retention time, but the area would not decrease unless there was a leak in the system. If there is an increase in retention time that correlates with the decrease in area then the first thing I would do is to check for leaks in the injection port.

I am a little confused by the 4:1 split ratio. The Shimadzu packed injection port does not support split injections.

Have you contacted Shimadzu technical support about this problem?