Headspace analysis propane, Turbomatrix-Clarus

[MichaelVW]

I'm trying to run headspace on a Turbomatrix 40 and Clarus 590. For this application I just need to know the total FID signal for each standard and sample. I don't care about getting nice chromatography. The standards are propane/nitrogen mixes that are purged into 20ml headspace vials that have been filled halfway with water. The samples will be things like paints, oils, etc which will be added to the vials and allowed to equilateral in the headspace oven I'm testing for how much organic material is in the headspace over the sample.

The GC analytical column is used as a transfer line: it runs up through the injection port and runs through the transfer line sleeve, and connects directly to the headspace analyzer. It's a 30 x 0.32 x 5 Elite-1 column. So the HS should be providing the carrier gas (nitrogen) to the GC. The detector is an FID.

HS settings:
Needle 80C, Transfer 90C, oven 35C, carrier 10 psi
Pressureize 3 min, inject 0.5 min, withdraw 1 min, thermo (equillibrium) 5 min, GC cycle 4 min.
Vial vent is off, high pressure injection is off, column pressure set to 10 psi.

GC Settings:
Inlet gasses off
FID hydrogen/air at 45 and 450 ml/min
Inlet at 150C and oven isothermal at 150C for four minutes.

For method development I have the thermal equilibrium step on the HS set to 1 minute, since it's just propane and I don't imagine it to have much equilibrium with water. For actual calibration and sample analysis the method requires one hour.

My problem is that sometimes I get a real nice propane peak, and sometimes I get basically nothing - just some very small baseline fluctuations. This is my first time using PE equipment and doing headspace, so I'm looking for any suggestions on whether my method settings look reasonable. This method (EPA 40e) is old and nobody seems to use it anymore so I can't find application notes.

[MichaelVW]

I want to add: When i do get a propane peak, the abundance seems to bounce around. Areas on three replicates of a 20% propane sample 35k, 69k, 40k

[Peter Apps]

How do you purge the propane into the headspace vial - cap on/off, through the septum ?

What is the water there for?

[MichaelVW]

Peter,

I'm glad you asked what the water is for! It's what the method tells you to do. https://www.epa.gov/sites/default/files ... od_25e.pdf: "Fill the calibration standard vials halfway (±5 percent) with deionized water."

I've tried adding the gas to the headspace vials two different ways. I haven't noticed a difference.

-Purging the gas into the vial and then quickly capping it (I've tried both screw-threaded and crimp caps/vials). This is what the method describes.

-Purging the gas through a needle, with a second needle purging the septum to allow gas to flow back out of the vial. After doing several vials this way I was seeing less flow into the vial (I tilt the vial so that the gas needle is below the surface of the water to see how much it bubbles). My guess is that the vent needle was getting clogged by bits of septum, but I'm not really sure.

[rb6banjo]

Wow! That's a weird method. Seems like they're jumping through a lot of hoops just to get what is basically a "Total Hydrocarbon" number.

What you really need is a "Total HydroCarbon Analyzer". They use them in the CO2-refining industry because there's a THC spec on CO2. The analyzer is basically a FID where the gas sample is just sent up into the flame. You calibrate the flame with known concentration of gas (50 ppm methane) then anything that burns in the flame gets counted as "THC". You could use multiple standards if you like. No chromatography just vaporized CO2.

For beverage-grade CO2, you need to run the liquid through a vaporizer so that you get an accurate picture of all of the "burnable stuff" that might be dissolved or frozen in the liquid.

I fill vials with gas all of the time. I treat the vial as an exponential diluter. You can solve the linear differential equation (initial value problem) as:

dC/dt = IN-OUT

and get

C = Co*exp(-Ft/V)

where C is the air left in the vial and Co is the air that was in the vial at t=0. This assumes perfect mixing in the vial. Not a bad assumption for gases. Ft = Volume of gas added to the vial (F is the flow rate, t is the time), V = volume of the vial. Put a needle in the septum of the sealed vial so that diluted sample can leave as you add the sample through another needle. Keep the contents of the vial at atmospheric pressure. Use a 60 or 100 mL gas tight syringe for your sample:

C/Co = exp(-100/21) = 0.008 (>99% changeover of the contents of the vial)
C/Co = exp(-60/21) = 0.06 (94% changeover of the contents of the vial)

If you have a different volume syringe for sample transfer, you could always apply the dilution factor. Then, just run the vials. I haven't had trouble with leakage of the vials during pressurization for sampling in my PE2400 headspace sampler.

[MichaelVW]

Wow! That's a weird method. Seems like they're jumping through a lot of hoops just to get what is basically a "Total Hydrocarbon" number.

What you really need is a "Total HydroCarbon Analyzer". They use them in the CO2-refining industry because there's a THC spec on CO2. The analyzer is basically a FID where the gas sample is just sent up into the flame. You calibrate the flame with known concentration of gas (50 ppm methane) then anything that burns in the flame gets counted as "THC". You could use multiple standards if you like. No chromatography just vaporized CO2.

That certainly sounds like it makes sense. The only reason I'm using 25E is that it's called for in some old air emission regulations. I don't know why we have a GC to do it... this problem came with the job!

I tried running a series of vials with just a drop of acetone at the bottom, to eliminate gas sampling variables. I gave it an hour at 35c to equilibrate. Four back to back runs gave area responses of:

374393
550937
506550
221004

[Peter Apps]

What you really need is a six-port gas samping valve instead of a headspacer.

What repeatabilty do you get if you sample twice from the same vial?

[MichaelVW]

What you really need is a six-port gas samping valve instead of a headspacer.

What repeatabilty do you get if you sample twice from the same vial?

I reran the four vials from earlier today. The only other change I made was taking apart the FID and putting it back together - I thought I'd try cleaning it but it looks like it's never been used before.

To recap, originally the area counts for my acetone vials were:

Vial A: 374393
Vial B: 550937
Vial C: 506550
Vial D: 221004

Running the same vials in the same order, I saw

Vial A: 580,000 but also, a second peak later in the run at 3.7 million
Vial B: 566,000 likewise, second peak at the same time as above, 4.3 million
Vial C: 564,000
Vial D: 279,000

Tomorrow I'll try running a single vial multiple times.

[Peter Apps]

Run the single vial repeatability with propane rather than acetone, where evaporation is an added source of variability. Take the samples as soon after one another as possible.

What is the sampling plumbing in the headspacer?, does it have a six-port valve?, if so it may be possible to load the sample loop direct from a syringe if you can reroute some tubing.

[rb6banjo]

Or, seal the vials first and add a known volume of acetone to the vial through the septum. That way, you eliminate the potential of irreproducible evaporation.

[MichaelVW]

Setting the equilibrium time to 5 minutes, I did several runs from the same vial (propane at 20%). Results were:

149223
150656
117077
105762
105901
108992

I'm still trying to understand how the HS works. I don't think there's a trap - I think it just pushes analyte into the transfer line.

[rb6banjo]

I pulled down the manual for the Turbomatrix 40 and you could have a trap or not have the trap. You may have to dive in there and see how it's configured.

For this analysis, I don't see why you'd need the trap. My PE 2400 is just a static headspace sampler. The sample that goes to the column is just the vapor above the sample at equilibrium - or the gas in the vial. There's no concentrating trap in the sampler.

If yours has a trap, you have to worry about breakthrough on that trap. Propane is extremely difficult to trap with most any solid sorbent. The breakthrough volume is going to be very small. This could also be part of your precision problem.

[MichaelVW]

Thanks for looking into that. I don't think there's a trap. The 2nd picture here looks like what I have.

https://americanlaboratorytrading.com/l ... 110-18866/

There are two gas lines that connect to the cylinder with the red insulation, and a connection for the transfer line.

[rb6banjo]

I think you are right. I found photos of the guts of the thing where the trap is present:

https://resources.perkinelmer.com/lab-s ... sGuide.pdf

The trap heater is right next to the control screen on your system. I don't see it in photo #2 from your link.

Well, at least breakthrough of the trap is probably not a source of your precision problem.

What if you just created some type of model system, perhaps fill vials with nothing but natural gas and check the precision on some of analytes in that?

What I do to accomplish this is to open the valve for our house natural gas, put a 21 mL headspace vial over the opening so the gas is diverted inside. As long as the flow is pretty high (like you can faintly hear it coming out of the valve) you're good. Let it flow for about 1 minute. Have your vial-cap with septum installed ready. While the gas is still flowing into the vial, position the cap so that you can quickly snap it down on the opening of the vial when you pull it off of the natural gas source. Crimp it and run. The gas inside the vial is essentially at atmospheric pressure.

Run a few and see what the precision of the methane response is.

Or, you could just seal air in some vials and with a gas-tight syringe, add a known quantity of natural gas to some vials - while venting it with another needle (so that the gas is in the vial at atmospheric pressure). Fill the gas tight syringe by letting the natural gas flow then, stick the needle on your gas tight syringe up into the flow as you aspirate the gas into the syringe. Run a bunch (again, no water) and check your precision.

My system needs to run anyway. I will try triplicate analysis of 25 mL of natural gas to my 21 mL vials and see what I get. That'll be about 69.6% methane in the vial.

[rb6banjo]

My system ran the samples ok and appeared to be collecting the data properly. Something has gone haywire with the data analysis piece of the software. It won't pull up the chromatograms, draw them correctly, and integrate the peaks. I'm not sure what's going on with it. I've never seen this happen before. So, I won't be getting you what I promised yesterday - anytime soon. I'm sorry about that. It was looking good during the acquisition!