GC-FID LVI settings

[Rndirk]

Hi,

I was asked to develop an LVI method (10µL) for the GC-FID analysis of mineral oils to get rid of an evaporation step in the sample prep. The routine method is a split method with 2µL injection.

Easy, was my first thought, this will only take me a couple of hours max since I have experience with developing LVI methods (on an Agilent GCMS..). But I was wrong.

My main problem is that I can not seem to get rid of some stuff that comes after the solvent (hexane) peak in my LVI run, which makes my first analyte peak (decane: C10) invisible.

I started by injecting 1µL in splitless mode, so I could compare the signals with the 10µL LVI method. In splitless mode, I couldn't get rid of the large hexane tail which made my C10 invisible (see chromatogram, green). Next I decided to go for LVI since I could vent off most hexane that way. The problem here is that, just after the main solvent peak there is this weird signal (white chromatogram, circled in red) I just can't get rid of.

I've tested different things but I hope someone can discover an obvious error in my method, or just tell me the origin of the signal that i want to get rid of... Is it even hexane or maybe some effect of the split to splitless switch on the FID?

GC: Thermo Trace 1300 with PTV injector
Column: Rxi-5HT 15m x 0.32mm x 0.1µm
Liner: 2.0 mm ID Baffled Inlet Liner
Flow: 1ml/min, Helium
Injection: 10µL
Solvent: Hexane
Injection speed: 1µL / sec

Other settings are shown in the screenshot below the chromatograms comparing 1µL splitless injection (green) and 10µL LVI injection (white) of an alkane mix in hexane.





EDIT: if you want to take a look at the settings, it's easier to read if you open it in another tab. I've tried to resize it but that didn't work.

[Peter Apps]

Large volume injections are prey to a host of problems. First; your splitless time for the 1ul splitless injections is probably too long, reduce it to 30 s to get the hexane tail back down to baseline.

With a simple baffled liner at 50C and 80 ml/min gas flow you might be getting liquid flowing to the bottom of the inlet. Put a little bit of glass wool in to see if it helps. With 80 ml/min gas flow at 50 C, 10 ul of hexane is gone very quickly. I would reduce the gas flow rate. The wavy bit of the signal from the LVI looks to me like hexane cooking off some surface as the inlet temperature fluctuates at the top of the very rapid ramp, it might go away if you stop liquid getting to the bottom of the inlet.

Which peak is the C10 ?

Peter

[Rndirk]

The C10 peak is invisible in both of the chromatograms. Perhaps it's the one touching the red line in the green, splitless one.

The sample is mixture of even alkanes (10ppm C10-C40), where the 3 peaks rather close together are C16 isomers, then C14, and the first one which is slightly visible in the white one C12.

I've done some more tests in the meantime and you are correct. What reduces the wavy bit of signal the most, and makes my C10 peak visible, is reducing the splitless time. Reducing the vent flow increases the C10 signal, and now i realize that I was actually venting off a lot of low boilers as well, which is also obvious from the C12 signal in the white chromatogram.

I thought that, as long as i still have a decent amount of hexane left (which i can actually see in FID) i wouldn't be venting off C10 with a boiling point of 150°C. But my sample is likely all over my liner, and some spots are completely dried down while others are not.

I'm on the right track now i think. I will certainly try the glass wool and/or maybe lowering the injection speed to avoid hexane hitting the bottom. I have yet to calculate the discrimination with a baseline correction, it has to meet certain criteria in this analysis: the C40-C20 over C20-10 ratio

Thanks a lot !!

[Peter Apps]

But my sample is likely all over my liner, and some spots are completely dried down while others are not.

That is exactly what is happening - to get the best out of LVI and the various solvent effects you have to have a way to hold the sample film in place while it evaporates only from its upstream edge. This works well in the right kind of porous packing, and you can get octane in hexane onto a flat baseline, so decane would be easy.

Peter

[Rndirk]

I solved the separation between the solvent peak and C10 based on your comments, which was my original problem.

But now I'm having troubles getting the discrimination to meet the criteria. The C20-40 / C10-20 area ratio is just below the acceptable range (this is measured on a reference oil).

In other words, I'm detecting relatively less high boilers than I should be. It was more pronounced in LVI, so i decided to just go for splitless 1ul injection and solve it that way, with the following settings:

Liner: http://www.restek.com/catalog/view/9397
Inlet temperature: 350 °C
Splitless time: 0.6 min
Flow: 1.5 ml/min
Surge time: 0.1 min (pressure 2x normal pressure)
Oven: 1min 50°C, 25°C/min to 350°C, 1min

It's a straight liner now, which works better than the original baffled one i was using (tried with and without adding glass wool) in terms of getting a nice baseline and less discrimination.

To get more high boilers the logical way should be increasing the inlet temperature and splitless time. But both are already pretty high, and increasing the splitless time will increase the hexane tail.

Any other ideas?

[James_Ball]

With the LVI parameters it looks like you are injecting 10ul at 1ul/sec, so that will take 10 sec to do the injection. If the hold time on the solvent venting is 0.1 min then you are only holding it in vent mode for 6 seconds, so you have another 4ul being injected after you start putting everything on column(if I am reading those settings correctly). I had a similar problem and had to increase the vent time so that it was just longer than the time it takes to inject the sample, and that solved a lot of my problems at injection.

[Rndirk]

The settings were confusing to me at first, but it's not an issue: there's an 'injection phase' of 0.1min and an 'evaporation phase' of 0.3min with the same temperature and vent flow. They're basically the same kind of phases so it's 0.4min of venting at 50°C.

It's better optimized right now compared to the screenshot, but it's hard to get everything right: not venting off C10 and keeping just enough hexane, and getting a nice baseline before C10..

It's the splitless method that bothers me right now (my previous post). Increasing the surge time/flow, and the inlet temperature gives me exact the same out-of-range discrimination (in favor of low boilers), leading me to believe there's something wrong with the reference oil itself...

If i'd make a solution of the same volume of C10 and C40 of around 500ppm myself, for example, should I be getting the same area for both in FID provided everything reaches the detector?

[James_Ball]

I see it now, on the times.

For splitless I have had to put a small amount of glass wool at the bottom of the inlet to get better response for high boilers, especially PAH compounds. Most of the time the needle will shoot the solvent straight down so if it hits the bottom it may not evaporate the high boiling analytes fast enough but the glass wool gives them more surface area to draw heat from.

Have you tried cold splitless yet? Injecting 1ul in splitless mode with a fast plunger speed and holding the inlet below the hexane boiling point for about 0.02 minutes, then fast ramping to 350. This lets the syringe transfer the solvent to the inlet without the problem of needle discrimination then flash evaporate while still in splitless mode to transfer all of the sample to the column.

You may also want to use a retention gap to get better results with hexane recondensation/solvent focusing. I found adding one helped my peak shapes quite a bit when using cold injection techinques.

[Peter Apps]

Simpler is always better, so drop the pressure pulse/surge thing. Its only likely effect will be to transfer extra low boilers while the high boilers are still evaporating relatively slowly and so do not get pulsed onto the column.

Glass wool or a frit to give the high boilers somewhere to sit while they evaporate will help things along.

Peter

[Rndirk]

Have you tried cold splitless yet? Injecting 1ul in splitless mode with a fast plunger speed and holding the inlet below the hexane boiling point for about 0.02 minutes, then fast ramping to 350. This lets the syringe transfer the solvent to the inlet without the problem of needle discrimination then flash evaporate while still in splitless mode to transfer all of the sample to the column.

Thanks for this tip. I tried this just like you suggested, and now i'm discriminating the other way around - above the acceptable range of the ratio C20-C40/C20-10

So in summary:

Hot splitless: Discrimination of high boilers
Cold splitless: Discrimination of low boilers

I could decrease the target inlet temperature of the cold splitless method, but that would be cheating, no?

[Hornet]

I solved the separation between the solvent peak and C10 based on your comments, which was my original problem.

But now I'm having troubles getting the discrimination to meet the criteria. The C20-40 / C10-20 area ratio is just below the acceptable range (this is measured on a reference oil).

In other words, I'm detecting relatively less high boilers than I should be. It was more pronounced in LVI, so i decided to just go for splitless 1ul injection and solve it that way, with the following settings:

Liner: http://www.restek.com/catalog/view/9397
Inlet temperature: 350 °C
Splitless time: 0.6 min
Flow: 1.5 ml/min
Surge time: 0.1 min (pressure 2x normal pressure)
Oven: 1min 50°C, 25°C/min to 350°C, 1min

It's a straight liner now, which works better than the original baffled one i was using (tried with and without adding glass wool) in terms of getting a nice baseline and less discrimination.

To get more high boilers the logical way should be increasing the inlet temperature and splitless time. But both are already pretty high, and increasing the splitless time will increase the hexane tail.

Any other ideas?

Try a run with the initial oven temp at 60°C, 1 min splitless time.
I run this method with a single gooseneck liner, with wool.