[scwilson]

I apologize for the long post. I’m of the opinion that less is more, but I was at a loss to include all of the pertinent information in a more succinct manner. It’s the best I could do. I sincerely thank any and all for their help, thoughts, criticisms, and/or advice.

I’m performing a GC-MS analysis on a complex sample. I can safely assume no matrix interferences, but it is a complex sample with many products of interest that are homologous in nature—such as a series of fatty-acid derivatives. My standard contains over a dozen of these derivatives in methylene chloride and is dilute. I can run the standard as is with a 20:1 split with peak abundances no lower than 200-300k and no higher than 1 billion. In other words, it’s not concentrated, so I wouldn’t, typically, consider using it as a “stock solution.” I only have one mL of this standard that comes in an ampule and it’s expensive.

Brass tacks: I’m limited in my standard by both amount and cost, and it’s tricky to work with once opened, since the solvent is volatile. I want to run a 5-point calibration curve, but I don’t think I can work quickly enough to generate a good curve. At absolute best, I could perform a splitless analysis and dilute my standard 1:20 and use that to prepare 5 standards. But the appropriate solvents are all so volatile (methylene chloride, pentane, carbon tet, etc.) that even a 1:20 dilution will be hard to prepare accurately, let alone use of that solution to prepare additional standards.

Hence, the heart of my question follows. What if I prepared a dilution of my standard fresh from the ampule. Say, a 1-to-5 dilution. Then, I inject 1 mic. Run it again but inject 2 mics. Run again but inject 3 mics. And then 4 mics and 5 mics. If we call the 1:5 dilution “soln. A,” my scheme would look like this:

Standard 1: soln. A – 1 mic injection
Standard 2: soln. A – 2 mic injection
Standard 3: soln. A – 3 mic injection
Standard 4: soln. A – 4 mic injection
Standard 5: soln. A – 5 mic injection

In theory, the MS is a mass-dependent detector and not concentration-dependent. Injecting 2 mics of a 5 ppm standard would, in theory, inject the same mass of product as 1 mic of a 10 ppm standard. In theory. In practice, is this okay? I’m aware that there could be confounding issues here. This is REALLY just a test of the linearity of my injector, but, IN THEORY, the calibration curve would be the same as a traditional calibration curve, assuming reproducible injection behavior and no confounding other issues. I also realize that, yes, an MS is mass-dependent, but there could be other concentration-dependent variables that affect the signal. It’s simply the difference between theory and practice.
   
Finally, here’s my ultimate question: for those of you with more extensive GC-MS experience, can I get away with this? I realize it depends a LOT on the nature of my sample and standard, but is this “trick” commonly employed in practice or is it, rather, seen as inadvisable?
   
I don’t need spectacular error bars, but I do need better than “ballpark” results. Lastly, as I’m sure some of you are considering while reading this: reproducibility of injection. The Achilles heel to my plan (or one of them) is that I’d be unable to employ an internal standard to correct for injection variability. After all, if I am constantly changing injection volume, I can’t rely on the signal from an internal standard to remain constant and normalization goes out the window. However, given other complications, no matter how I perform this analysis, I will not be able to employ an internal standard, anyway. I will, however, be able to inject each “standard” (from the outline above) and each sample more than once. I intend to run every solution in triplicate. I also know from extensive use with this GC-MS that its ALS is about as good as it gets. It’s not perfect, but it’s as good as it gets.
   
I’m not shy, and I’m not insecure. If this is a bad idea, please let me know. I’m open to all criticisms. Could this potentially work, or am I simply neglecting too many potential variables that would alter the nature of my calibration curve?

[voidSetup]

In theory I think it is possible but the calculations would be a little bit different.  Usually when doing an analysis one would run the same method for samples and standards, and calculate their response factor as Area/Concentration.  In this case you would have to calculate the actual amount of standard loaded onto the column for each point in the calibration curve (i.e. Area/(Concentration * Injection volume).  Then when you inject your sample you would calculate the amount of analyte present in the volume injected and back calculate to the actual sample calculation.

There are some practical considerations though.  A 5uL injection is higher than I typically use, and depending on the expansion volume of your solvent in the GC inlet, it could potentially backflash into the carrier gas port which could cause carry over.  Agilent used to have a program to calculate solvent expansion volumes for various solvents with different liner models but I think they updated it to iOS and it's not included anymore.

Can you just rinse the contents of the ampule into a volumetric flask using a less volatile solvent (MeOH, DMAc, DMF, etc), and prep your standards from there?

[scwilson]

Hi, Void,

I wanted to thank you for your response and provide an update. It was very considerate of you to share your advice and to lend your experience with me.

Your comments make perfect sense. Re-working and describing my calibration curve as signal (area) vs. mass of analyte is logical and a much wiser way of describing what I'm doing, since I can support that with credible arguments, given the way an MSD works and its underlying theory. I thank you for your insight.

My current project is best described as method development. We're trying to develop a new method to analyze these fatty acid derivatives and using tried-and-true GC-MS as our means of validation. We intend to publish in a relevant analytical chem journal, so I want to make sure I have a leg to stand on when working up my data. I was intending on using the method I previously described to get some preliminary results. From there, we eventually planned on ponying up the cash for enough standard material to do this the "normal" way--i.e. a 5-point calibration curve of signal vs. concentration.

Your final suggestion of just trying a less volatile diluent sort of got me thinking, and since I'll need to eventually do a traditional calibration curve, anyway, I think I'm going to try this now and just nip it in the bud. My analytes tend to shy away from more polar solvents, and we don't have anything on hand like, say, decane, but I'm thinking that maybe MeOH will work if I dilute the standard 1:40 and use 2 mic injections on a splitless inlet. It's only 5 migs of total analyte in the standard (of which there are over 30 individual standardized components), so maybe at such an overall large dilution, I can get the standard to dissolve.

Also, thank you for the advice regarding the injection volumes. I hadn't considered carry-over, and, indeed, we usually use 1 mic injections, so I wasn't really considering the potential problems of larger injections until you'd mentioned it. I've tried the method I initially described above using "test" analytes--i.e. not using the expensive standard but just some cheap aromatics we had laying around--and a plot of signal vs. injection volume for all analytes was straight as an arrow with an R-squared with two 9's or better, so I don't think that the solvent I intended (pentane) expands enough to give carryover at 5 mics, but it's very much worth considering and worrying about. Thank you.

Anyway, I should get things analyzed within the week. I will report back what I find. I hope to run a proper calibration curve AND try my previously-mentioned calibration curve idea. For posterity and the betterment of the forums, I'll report back what I find. I just wanted to update this thread and let you know how much I appreciated your response. Thanks again--I hope I've said that enough.