Poor reproducibility of GC Peak Area due to varying sample loop pressure

Hello guys,

 

I use an online GC (7820a, with FID) to analyse the reactor exit gas continuously. The reactor is used for catalyst testing for dehydrogenation of Methylcyclohexane (1 mole) to Toluene (1 mole) and Hydrogen (3 mole). The reaction happens at 500C and 1 Bar absolute pressure. As you see, this is a mole changing reaction and depending up on the conversion the volumetric flow rate downstream of the reactor will vary.

 

After the reactor, the flow stream is split into exhaust (6 mm O.D tube with a Fine Dosing Valve to vent) and GC inlet line (3 mm O.D tube with manual Metering Valve). Sample injection in the GC done using a normal 6-port VALCO valve with a sample loop of 50 micro-litres. All flow streams after the reactor is heating to 230C and Valve box is also heating to 230C.

 

The problem I face right now is that, I see variation is peak area and analysis is not reproducible. This is mainly because of the variation in flow rates from the reactor exit to the sample loop in the GC, which in-turn vary the pressure inside the sample loop. According to simple gas law equation, n =p.V/R.T, lower pressure will also reduce the total number of moles injected from the sample loop.

 

Following are the solutions I thought about,

 

1. Shutting off the GC inlet line before Injection using an ON/OFF valve: Shutting off the GC inlet line just couple of seconds before the sample loop injection will allow the sample loop to equilibrate against the atmospheric pressure and attain sample volume injection. But for this to work seamlessly, the ON/OFF valve and GC injection trigger should some how be synchronised (A custom control box which triggers the ON/OFF valve to close and then after few seconds triggers the GC to inject). Seems like not an easy solution

 

2. Adding a 4-port valve before 6-port valve: I thought that the same shut-off of the GC inlet line can be designed by allowing the gas from the GC inlet to first go through a 4-port valve and then to 6-port sampling valve, so that 4-port valve can be turned ON few second before, cutting the sample flow to the sampling loop of the 6-port valve and thus allowing the sample to attain equilibrium, and then the 6-port can turn ON to inject position. After injection, the 4-port valve can be turned OFF again, allowing the sample to flow through the sample loop until it's time for next injection. So, I installed an old 4-port valve into my GC. However, in the EZChrome Elite software, I can only assign this new valve as "sampling valve" or "switching valve" or "OTHER" valve. Since, I cannot assign it as "sampling valve" (because, I already have 6-port valve for the sample injection), I assigned it as a "switching valve". Problem there was that I cannot give certain time (e.g. 0.1 min) for which it has to turn ON and then OFF before the 6-port valve injection. If I simply add the 4-port valve as a "switching valve", it just turns ON and stays ON through out the run and also for the rest of the sequence runs. Assigning it as "OTHER"valve also didn't work.

 

Did anyone of you guys faced the sample problem? is there any other solution for this?

 

Thank you in advance!

 

Kind regards,

Oshin.

  • Hi oshin-sebastian,

    I updated the tags on your post and moved to an application area for better visibility.

    Regards

    James

  • Got a quick solution that is less mechanical. You can add an inline pressure valve on the outlet of the 6 way valve. So the flow will only occur through the line once it reaches a certain pressure in only one direction. Thereby controlling the flow out of the tubing. So when your GC flips the switch so to speak and injects the loop into the inlet it can then be assured it's at a constant representative pressure. Only risk you take is condensation under higher pressure along the line. So you would have to tinker with it to get it right.

     

    What are the area counts you are getting as 50 uL is still big even for a gas injection. The liner maybe getting overloaded so your peak reproducability would be less.

     

    Another though you could do it have a cooler injection loop of say 2 uL. Then as the solvent condenses, this can be the "syringe" for as injection, where it will heat in the inlet back up instantly and give much better chromatographic results similar to a standard GC system.

  • Hi gchaplain,

     

    Yes. You are correct. A simple pressure valve is a solution but I need to heat that pressure valve to 200 C to avoid condensation.

     

    I am not aware of cooler injection loop. If the product condenses, how we gonna take constant measurement from the gas stream?

     

    The best solution I was was to use an "Atmospheric Reference Vent" coupled with a "Stream Switching Unit". The Stream Switching Unit" will periodically divert the gas flow to exhaust and the the "Atmospheric Reference Vent" will allow the sample loop some equilibration time to attain atmospheric pressure before injection. Of course, we have to synchronize this unit when GC injection time.

    https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=2ahUKEwjEpZWone3gAhXRaVAKHeeMDs8QFjAAegQIBxAC&url=ht…

     

    But some other GC has something called "Stop Flow Valve" which stops the flow few seconds before the injection.

     

    Right now, we are trying make a control box which control the opening and closing of electronic pneumatic valve installed in the GC inlet line and then triggers the GC for injection 5-10 seconds after the valve is closed.

     

    Thank for your advices

     

    Cheers,

    Oshin

  • And I am using a liner of volume 870 uL. So I think 50 uL sample loop volume should be fine. right?

    The Area counts that i get is in the range of 100 000 000

  • that is very high for the area count.

    What temperature is the inlet? What usually happens is that the inlet temperature causes an expansion of the vaporised gases that can over fill the liner.

    Agilent has calculators you can download to tell the expansion and if the liner is overloaded:

    The vapour volume calculator can tell you. And if it is indeed overloaded that will definitely affect your reproducability.

    GC Calculators | Agilent

  • I think I never thought that will be problem.

    The valve and sample loop is at 230C and inlet is at 300C. Can 50 uL sample in the loop expand more than 870 uL in the inlet?

     

    Thanks in advance.

  • yes it can.

    P.V=n.R.T.

    Assuming that everything in the loop is a gas;

    If the pressure and moles are the same and the temperature rises by 70 degrees, you are looking at a 70*50 expansion in volume to balance the equation, so the sample volume will increase to 3500 uL. Your liner will be overloaded.

     

    you need an injection loop volume of a maximum of 13.86 uL to reach the liner limit. To be safe, I would reduce it further to a 5 uL loop.

     

    Of course if the pressure is higher in the inlet you can calculate it out by substituting into the equation.

  • Yeah this makes sense.

    I am also thinking if possible to keep the inlet temperature at 230C as well (so same as the sample loop).

    But if nothing works out I will change to a 5 uL sample loop.

     

    Thanks again.

     

    Cheers,

    Oshin

  • Hello,

     

    I am doing analysis of gases using injection valves as well. Area reprducibility, if the 7890 instrument is running well, approx. 1-2 % relative uncertainty must be expected. I don`t think you can expect better, unless very strict controle of gas sample flow and pressure is done.

     

    A 2-way or 3-way closure valve for gas sample stop flow can easily be configured in run time table using a External Event cable. If the gas sample flow are stopped and leveled to athmospheric pressure shortly before injection, you have to take even small changes in room athmospheric pressures into account.

     

    Are you able to check the gas sample flow using a simple ball flowmeter for flow variations shortly after the gas sample valve exit? Only very small changes should be observed.

     

    Best regards -

    Lars, Copenhagen

Was this helpful?