Page 1 of 1
spiking trace amount
Posted: Mon May 16, 2005 7:47 pm
by MK
i heard about a technique where you can work below LOD:
-spike samples with various amounts of standards
-Plot the response vs. spiked amount
-work out the unknown from the intercept
what is the name of the technique? are people using this? any limitations? where can I found a reference?
thanks
mk
Posted: Mon May 16, 2005 8:05 pm
by DR
"Below LOD"...
where LOD means "Limit of Detection"
eg: You didn't see any.
The only way to lower LOD concentration is by injecting more sample (larger volume, more concentrated prep), maximizing detector sensitivity (minimize AUFS setting, this will add some to your noise, however), or changing to a more sensitive detector (going from UV to MSD, for example), or derivitizing samples with something that a detector is more sensitive to (amino acids treated w/ OPA(?), for example).
These all require at least some linearity testing, if not revalidating the method entirely.
Posted: Mon May 16, 2005 9:10 pm
by tom jupille
MK, what you're describing sounds like "standard addition". It's useful primarily where no matrix blank is available. As DR indicated, however, it will not get you below the LOD.
Google it..
Posted: Tue May 17, 2005 12:01 am
by james little
go to google, and search for standard additions method.
Top two hits look really good, third not so good, many others look promising.
Good luck.
Posted: Tue May 17, 2005 7:34 pm
by MK
ok, I'll try to rephrase:
You have done all the tricks trying to improve sensitivity and you still don't see anything or you still have too low S/N. Can you use standard addition method to determine the concentration of the unknown?
If there is nothing and you spike the sample with different amounts of standard, your curve will pass around origin.
If there is something your intercept is not zero.
thanks
mk
Posted: Tue May 17, 2005 8:42 pm
by Mark Tracy
That is an attempt to get something for nothing, and it does not work. While you gain from multiple measurements of the unknown, you lose it all from the uncertainties of the standards. LOD is by definition, the level below which the signal is indistinguishable from noise. Perhaps you mean the peak is detectable but not quantifiable?
Standard additions might work for peaks below LOQ but above LOD. However, it is easier and cheaper just to make multiple injections of the unknown and average the results.
Basically, there are almost no circumstances where standard additions is useful for HPLC-UV.