GC-MS SENSITIVITY/split vs. splitless mode

[reecai06]

Hey all,
I am currently in the process of developing a method to quantify ergosterol (converting to fungal biomass) in leaf litter, performed with an Agilent 7890A gas chromatograph interfaced to an Agilent 5975C inert mass-selective detector, containing an HP-5MS (30m x 0.250 mm I.D., 0.25 μm film thickness) capillary column. Standards and samples were injected in splitless mode at a volume of 1 μL with a starting oven temperature of 100° C. However, I have a set sub-samples that were sent to another lab in Mississippi and analyzed using HPLC and the comparative results indicate that the sensitivity of the GC-MS drops after reaching over approx 100 ng/uL. My faculty sponsors are ecologists and are not analytical chemists and they have suggested that I re-run my samples in split mode (1:10). I do not see the benefit of this. I feel like my samples would have even lower abundances than they already do (~500), therefore becoming un-quantifiable. I have few resources to talk with about this and I was hoping someone out there could help! Would my ergosterol concentrations be lower than they already are if I turn on split mode? please help!! Thank you so much!!!


-Caitlin Reece

[Don_Hilton]

100 ng on column seems to be a lot of material for quantification with a mass spectrometer - and has the potential for reaching into a range of non-linear response. So, the first question is: does your calibration curve bracket the results obtained from the samples? If not, go back and extend the calibration set.

Are you using MassHunter software? I ask because abundance values for well shaped peaks appear to be numerially lower than I would expect from other software. (I've only seen data shown by another; I've not had hands on to see what can/can not/should not be done wihh the software.) The size of the number should not be an issue as long as you can get an adequate number of significant figures for the low level standards.

With a 1:10 split, you are putting approximately 1/11 the quantity of material into the column than would have gone in a splitless injection - which could move you back into an area of linearity. Take note that you must use a calibration curve generated the same way as you run the samples, so you must run the curve with the same split ratio.

[reecai06]

I do have a calibration curve and it does bracket my samples and I have an r-squared value of 0.998. I am using ChemStation software and it didn't occur to me that I would have to re-run the standards in split mode and recreate a calib curve until now. The real problem I am having, however, is that I am attempting to validate this method trying to perform a method detection limit and it just is not working. The reason why using split mode was suggested is because I've attempted to do a method detection limit and processed MDL samples exactly like my leaf samples but with a standard and it just isn't being quantified and my sponsors wanted better comparative statistics to the HPLC method, pretty much busy work. My internal standard (7-dehydrocholesterol) qualifier ions are not being detected and I suspect there is peak overlap. I checked the purity of my compound with a replicated melting point test that proved to be impure. So i ordered a 7-DHC and I'm still having the same problem with my internal standard not being quantified. This research has been a year in the making and I have limited time to figure this out before I graduate in June so any help is welcome.

[nishanth]

There is a definite reason why ergosterol is usually analyzed with an LC than using a GC because with GC it requires derivatization to increase volatility, and sylilation of ergosterol are notoriously hard (require 60C for more than an hr, and still will have some partial derivatives). So if you are interested in just ergosterol and no other sterols from your samples LC is the most convenient option. There are couple of good papers that uses a SPE for purification for LC, but in a fungal abuntant litters you can directly inject the saponified solution to LC after filtration with no other pre-treatments .

This brings to the main question. Did you sylilate your samples for GC? Usually you need to try different conditions and I have found that 60C for 60 minutes is somewhat acceptable. Probably something is going wrong with your derivatization, everything except for the derivatization should work like any normal analysis.

Make sure that the derivatizing reaction contain no protic organic solvent like any alcohols. Pyridine or tolune is the best for ergosterol. Use m/z 363, 337,& 468 as qualifier ion (sylilated molecular ion & fragment ions).
Hope this helps
NT

[Don_Hilton]

One other issue comes to mind - effect of matrix on peak size. I am guessing that you get good sized peaks from leaf, but small peaks from standards prepared like leaf?

If so, you need to determine limit of detction and preform calibration in matrix. This would be because variouis extraneous compounds in the matrix cover active sites in the GC inlet when a sample from leaf is injected, and the extraneous compunds are not present in a standard mix made up without matrix compounds. Given taht you are using a different compund as the internal standard (as opposed to an isotope labeled internal standard) the effect can be different between the analyte and the internal standard.

This is not necessarily busy work. If you have matrix enhancement going on, particularly if it differs depending on sample history, you may have some weakness in your data.

[Peter Apps]

Hi Caitlin

You say that the "sensitivity of the GC-MS drops after reaching over approx 100 ng/uL" - does this mean that the HPLC results for samples with more than 100 ng/ul were higher than the GC-MS results ? but that the HPLC and GC-MS results agreed at concentrations lower than this ?

An are the concentrations that you give the concentrations in the prepared sample extract that is actually injected into the instrument ?

Given that LC-MS (if that is what the Mississippi lab is using) is even more susceptible to interference from contaminants than is GC-MS, how do you and your sponsors know that the HPLC results are correct ? If they did not use HPLC-MS what did they use (fluorescense ?), and has their method been validated ?

Peter

[reecai06]

@ nishanth-I realized that ergosterol is usually measured with LC about 3 weeks ago during a lab meeting because I hadn't come across a paper involving GC-MS, ergosterol, AND leaf samples. I am derivatizing my compunds with BSTFA, but the original method that I am optimizing said 60C for 30min. All of my standards and samples have derivatized fully because the internal standard had the same peak size (area) for each sample and standard. I wish that my campus had an LC instrument, but we do not. I've found other papers measuring ergosterol using GC-MS & SPE but it was for mould in building materials which usually had much smaller ergosterol concentrations. But it could be the length of time derivatizing, if I happen to come across more time I shall attempt lengthen the time. Thank you for your help

@Don_Hilton-I've determined that my limit of detection is approximately 0.7 ng/uL of ergosterol and from the papers and professors they say to make a sample with standard that is around 1-5 times larger than suspected detection limit, which I used 3.5 ng/uL. I'm assuming that because the leaf samples do have other compounds besides ergosterol and 7-DHC, the data is being skewed. And after doing this research, I know I do have weakness in my data, I didn't even do a sample clean-up which I feel is necessary when dealing with so many organic compounds in leaves. Thank you!

@Peter Apps- I'm sorry I've gotten my units confused!! I'm sorry, after I analyzed on GC-MS, I received ng/uL I converted them. The units I am actually reporting are ug/g (ergosterol/dry leaf mass) to compare to HPLC which were also given to me in ug/g. So, yes, the HPLC results from samples more than 100 ug/g were higher than GC-MS results, and the HPLC & GC-MS results agreed below 100 ug/g. I have been trying to find the method in which the HPLC samples were ran, however, the Mississippi lab is in the process of having that paper reviewed/published and I do not have access to it. My sponsors tried getting into contact with the lab about methods but it has been difficult so I may just have to ask myself. Thank yoU!

[Don_Hilton]

One thought that crosses my mind - do you have sufficient MSTFA to obtain complete derivitization for the higher level samples? It may be worth adding a bit more MSTFA to aouple of samples and seeing if the analyte/internal standard area ratio changes.

Did you make your calibration curve in matrix? This can help support good quantitation. This is best done with matrix that is analyte free, but if that is not possible, matrix with very low concentration analyte can be used, with calcuation of the background analtye as part of the process. Also, matrix extract diluted with solvent can be helpful - the matrix effect may be present with the concentration of background analyte reduced by dilution.

And by the way, weaknesses in data are not a show stopper as long as the extent of the weakness is known and they do not comprimize the conclusions of the work.

[Peter Apps]

Hi Caitlin

The simplest explanation for your results being low at high concentration is that you are going above the linear range of the analysis. What is the concentration of ergosterol derivative in the solution that you inject to the GC-MS when you have 100 ug/g in the sample ? Your calibration curve spanning the sample concentrations, and being approximately linear suggests that the departure from linearity is not happening in the GC-MS, which leaves something in the sampe prep.

By how much are the GC-MS results lower than the HPLC results ?, and does the discrepancy get larger as the concentration rises ?

Peter

[mckrause]

Hi Caitlin,

You state that you are using a 5975, and that your detection limit for ergosterol is 0.7 ng/uL. The '75 is unbelievably sensitive - if you brought me that data I'd say that you were off by a factor of 1000. That, to me, would indicate two things: 1. your derivatization is not working and 2. as Peter points out, you're probably working in a range that is close to saturating the detector. If you are going to derivatize, you absolutely, positively must go through a relatively vigorous separation and purification process to isolate the sterols prior to derivatization. If you do not do this, you will get data all over the map. There are injection techniques that you can use to analyze sterols without derivatization, but most of them take a strong background in chromatography, which from your comments I would say is lacking in your case. The literature is replete with cleanup techniques for sterol analysis. Look at the veterinary science literature, and check the FDA methods, to get a reasonable cleanup procedure that will allow you to adequately separate the sterols prior to your derivatization step.