inter-day reproducibility of GC-MS

[lmh]

I know very little about GC-MS. I have a situation where someone ran a lot of derivatised samples, and then a few days later realised they would have liked to quantify the samples against an external standard calibration, so they derivatised a standard, diluted it to various extents, and injected the dilutions, with the intent of using this curve on the previous samples. Naturally the results are far from perfect, and they're interested in which steps are most likely to be causing the trouble.

As someone who runs LC-MS, my stock answers would have been that what they're doing would give a rough estimate in HPLC-UV because UV absorbance is a physical property that shouldn't vary from day to day, but it would be utterly hopeless in LC-MS because ionisation changes over time. In GC-MS I don't know the answer. How variable is ionisation efficiency from one week to the next? How variable are the injections (if the liner gets dirty, does this affect the amount of the injected material that gets onto the column)? Is it safe to derivatise the samples on one day, and the standards another day, or is derivatisation efficiency variable? There are so many things I don't know! Heeeelp!

[Don_Hilton]

There are many opportunities for failure in the senario you described. While a GC/MS can be a stable sytem, there are no guarantees. Depending on the samples going throug the system and the level of preventative maintanance on the instrument, it may remain stable - or not. Thus, even when running samples with internal standards on an instrument known to be stable, some of us will aways run at least a check standard or two in each batch of samples run on a given instrument - even on the same day.

You have not indicated the method of derivitization. Some derivitization methods work nicely and give essentially complete derivitization. Others - or even the nice methods, but "ugly" analytes - are tempermental and seem to vary day by day. Thus that check sample I mentioned above -- derivatized with the batch of samples being run. And that's even with a "well behaved" derivitization. And there are some compunds that after derivitization change (TMS esters of sugars for instance) from the kenetically favored product to the thermodynamically favored products. And this takes several hours.

Without knowing the compunds involved, the derivitization method used, the tune and cleaning history of the GC/MS, and the techniques used in the lab - you could have results that are fairly reliable. But more than likely, you have something that is rough -- on a good day. But, you can try making some subsequent batches of calibration standards and running them day by day to see if the system appears to be reproducable. But, do expect a step change in results when you tune the MS. Then a calibration does not carry from one day to the next.

[lmh]

Thanks for this. It was indeed sugars. You've given me some useful tips on how I can encourage person to make sure future sample sets are more easily interpreted, and on the limitations with this set. I appreciate it!