[jlong]

Our lab has recently acquired an ICP-OES and there has been a great debate on how often to run the standard curves for each element being analyzed.  While we are a small lab there is a great disagreeance in what would be an acceptable length of time to use the standard curves... ideas include; Every time we analyze.  Every time we introduce a new Ar tank.  Every 3 months.  Every 6 months.  Once a year (that is how long our standards are good for).  Keep in mind; we use an Ar tank that will allow us to do 100 samples per tank at our current settings, we are switching out Ar tanks about every 3-4 weeks, we use 5 point standard curves for each element, a 1 pt. check is done each time a different element is analyzed, we do an internal instrument Ar calibration at all designated wavelengths each time the instrument is initialized. 

Currently we run new standard curves for each new Ar tank (usually includes 7 to 8 different elements). 

So my question is simply how often would you run the standard curves to maintain consistent reliable results?  P.S. time and cost do matter

Thanks in advance.     

[Train]

Our practice is to generate a new calibration curve with every set of samples and run a 1 pt. check at least at the end - sometimes more often depending on how many samples we're running.  If you don't want to go with the most conservative approach then you should have data to demonstrate why it's ok.  Do you have any data showing that a sample quantitated with an old calibration curve gives the same result as if it were quantitated with a fresh calibration curve over the range of your method?

After all, it's easy to have an opinion. 

[jlong]

@Train

I like the idea of obtaining data for the same samples on older curves and newer curves to see how they correlate.  Ultimately the goal is to obtain the most accurate, consistent results w/o squandering resources.  At this point it seems to be a fine line to find.

Thanks for the input.