Chromatography Forum

Here goes...

I know little about GC, but I happened to look at a GC-FID chromatogram a co-worker was looking at, and I was confused by the peak shapes I was seeing. She's running a method for Northwest Total Petroleum Hydrocarbons (NWTPH), and for now I'm assuming she's running the method by the book, but I didn't ask her. What confused me was the peak shape I was seeing for her highest aromatic calibration standard. Can anyone give me an idea as to what is happening here, or is this normal for GC-FID?

Total chromatogram of top calibration standard...




First eluting peak...




Last eluting peak...




Any help you can give me would be great, and I'll answer any questions I can abouot the analysis/instrumentation. To me, it seems like column overload, but I'm not sure. I did look at an overlay of all her standards, and instead of each peak increasing in height, it seems like each peak increases in both width and height, with width being the greater proportion as the concentration of standards increases.

My background is in HPLC, so I'm not sure if this effect is normal in GC-FID, or if she's just putting too much material on-column than she should (which seems odd if she's shooting a calibration standard?).

Spot on. Perfect concentration overloading. What column is this run on ? . She needs to dilute the sample, or increase the split ratio by a factor of at least 10.

Peter

Thank you Peter. I'm not sure what column or split ratio she's running, but I told her that's what I thought was happening (overloading). I'll get details when I head back to the lab on Monday. Based on the fact that this is a calibration standard, and all the standards look similar to this one (although this is the extreme case), I would bet she's not using any split ratio at all.

Ok, I went in today to check on some things, and my coworker was there. I think she's running a DB-5, 20 m, 0.18 mm, 0.18 µm column. She was injecting 2 ul splitless (method requirement to run splitless). I had her cut the injection to 1 ul, and we're cutting her top cal standard to 40% of the previous high standard. I also had her revert to the conditions listed in the method (albeit they are for a DB5 30 m 0.32 um column) - 50 degree hold for 2 minutes, then 8 C/min to 320, holding for 10 minutes, with the FID at 290 (I think), and the injection port at 310 (could have swapped temps for det and inj, trying to remember after the fact). I had her shoot her top standard, and peak shape was much better, but still some column overload. Now the issue is separating the first alkane marker (C8) from the solvent peak (hexane). I had her create a few extra methods, varying the ramp (to 12 C/min and 16 C/min) to see if she can get the required separation she needs with a shorter run time, and those same new methods with a 40 C hold for 4 minutes at the beginning to try to separate the C8 peak from the solvent peak.

As per the C8-solvent peak separation - was my suggestion to her helpful? Is a longer isothermal hold the correct solution to separating a low BP constituent from the solvent front? Or did I just make her run longer?

It sounds like you are headin in the right direction. I expect that will see moe effect from using the lower starting temperature. The changes in ramp rate will likely have more effect on the later portion of the run. The fact that your solvent is hexane and you are trying to separate octane makes it a bit tricky. You need the column warm enough to move the solvent, but cool enough to retain the analyte. With the step to a 0.18 ID column, I would even consider going to a 0.5 uL injection (you need a 5 uL syringe for this) or less - simply to eliminate the mass of solvent at the front end. You need to be sure that you can see your low standard, however. I've done this with some petroleum samples on a 0.18 ID column. Injection reproducability was sufficient for the work I was doing. But you need to check it on your application.

The underlying problem is the move from a 0.32 to a 0.18 mm column. How much you can put on before overloading is a cube function of the diameter. On the 0.18 mm column this method really does need split injections. Presumably you moved to a narrower column to get the resolution in a shorter time, but with a splitless injection you lose this advantage because you cannot get the sample onto the column in a narrow enough band to exploit the resolving power of the column.

If you really are stuck with splitless, then make the splitless time as short as possible.

Peter

Thanks for the replies. My coworker knows about as much about GC as I do (she's new, I'm in LC), and she kind of had this work foisted upon her. I don't even think she knew what column she was running until I opened the oven on Saturday. I'll pass along the recommendations as per splitless time/injection amount and column ID/length for future reference. I just hope she got the method working this weekend, as she has a report due today to a client...

Just watch on the short purge off time on the splitless injection. You cann get some discrimination in the inlet if your purge off time is too short - and the reason some people reject the use of a split injection and require splitless is to avoid that discrimination.

If you can get a column with the original dimensions specified in the method - I would reccomend it. The changes you are trying to make would take some time to check to be sure the method gives acceptable results. (Even setting up the method as specifified requires some checking - but there is less risk of failure.)

Indeed. The few changes we made over the weekend (with the addition of a longer hold at 40 C) helped to get the job done. Unfortunately, 'getting the job done' seems to be more important to the powers that be than having analysts who actually understand what they're doing.

Don, good point about the discrimination with short splitless times, I'll wriggle out of it by claiming post hoc that "as short as possible" meant "as short as possible without getting discrimination"

And the "we've got no time to improve the method, the client wants results" while hopping from foot to foot is a worldwide pandemic.

Peter

I am well aware of that policy. It seems to go like this: "We don't have time to get it right the first time - we need to hurry so we can do in again to fix it!"

I am well aware of that policy. It seems to go like this: "We don't have time to get it right the first time - we need to hurry so we can do in again to fix it!"

Yep, it's a variant of my favorite "We just need a number", to which the answer is "Do you want it correct?"

If I had a nickel for every time I explained to my manager the factual limitations to the deadlines being imposed upon me, I would have a s***load of nickels. To compenate, I normally hypothesize the absolute worst-case scenario in terms of when I can get results for a client, then (usually) pleasantly surprise them with earlier-than-expected results.