mattmullaney wrote:
Hi Peter,
My thanks-agreed assuredly on the stability of the FID. To respond without giving away too much, I'm in a situation where I know what the standard I'm using is (I'm using an internal standard, actually BHT), but at this point I've no clue as to what the analytes I'm interested are, other than they are substances that have been extracted from a solid with a solvent. To this point, we've not yet done any other sample preparation other than adding internal standard to these sample extracts.
We're pretty early along as things are going in this process.
I suppose one may call what we're doing at this point "semi-quantitative" at best. We are working toward identifying and eventually having quantities of our analytes prepared to shore up the method...I hope!
The main source of variability is not the QC brackets, but the samples with added IS. Can't recall the variability of the IS in the samples, but I'll post this info later today. And agreed, no reason to slow things down too much with tons of QC injections--perhaps, if I can show that the IS in the QC and the samples are having similar repeatabilities over multiple analyses, then I will feel better about lessening the QC injection frequency in time.
Regards, Sir!
Bracketing standards should be used to detect drift - something that changes over multiple analyses in a consistent direction. Similar to control charts. Random run-to-run variation can put a bracketing standard out of spec when the samples both before and after it have run perfectly. With an IS, drift can only be due to causes which differ for the IS and the analytes - so for example if you use an alkane as an IS for alcohols the alcohol peaks will drift down in relation to the alkane as the inlet cruds up and the column deteriorates. At some stage the calibration based on sharp alcohol peaks from a clean inlet and new column will no longer be valid - a bracketing standard will show that has happened.
So repeatability of the IS has nothing to do with frequency of bracketing standards, because repeatability is (or should be) a measure of random variation, not drift.
A lot of variation in IS peak area is not necessarily a problem as long as you get a correlated variation in analyte peak area - the IS is then doing its job of correcting for whatever is making the peak areas vary.
Peter