Chromatography Forum

I am creating a standard curve using LC-MS and my standard underwent SPE. My resulting curves showed good linearity and applying a chosen internal standard just messes up the data.

Internal standards compensate for certain types of errors (where the error in the analyte and the error in the IS are correlated), but they exacerbate other errors (where the error in the analyte and the error in the IS are orthogonal).

So the short answer is "yes", if the error levels without the IS are acceptable.

HOWEVER, the fact that the IS makes things worse is a warning flag: unless the IS peak is just so small that it causes difficulties measuring the area, you really need to dig in to find out what's going wrong.

ALSO, if you are in a regulated environment and running a validated/verified method, you cannot just arbitrarily ignore the IS; you must run the method as defined.

Thank you very much for your reply.

We are actually doing quantitative measurements on water matrices.

We will be looking into your suggestions. I am also guessing that the suggested IS to us is not the best analog for our selected compounds.

Is it possible to increase the concentration of the internal standard? That may make it perform with better stability throughout the process.

I am creating a standard curve using LC-MS and my standard underwent SPE. My resulting curves showed good linearity and applying a chosen internal standard just messes up the data.

The quantification via LC-MS without IS is bad idea (it's like roulette). Much better choice is change the IS or optimize its concentration (if you can of course).

Is it possible to increase the concentration of the internal standard? That may make it perform with better stability throughout the process.

That sounds great although, would it be affecting my results should as well should the IS concentration be high?

Tom's answer is spot on. An internal standard is only useful if it is a very good analogue to the target analyte. If the internal standard behaves differently to the target analyte, using it will generally increase errors (randomly). If the internal standard behaves oppositely to the target analyte, it will systematically increase errors. This can happen, for example, when sample prep involves SPE: if the analyte is only just binding to the SPE on loading/washing, and the IS is only just eluting during elution, then a slight mistake in solvent strength can mean that the analyte binds less (lower recovery) while the IS elutes better (higher recovery).

A good IS vastly improves a method. A bad IS messes it up. If you have a well-chosen IS and yet it's making the method worse, then something has gone wrong, and needs diagnosing. The commonest problem is that the IS is just so dilute that its beneficial effects are overwhelmed by large random statistical errors in the peak area of the IS. It is possible to go to the other extreme and have an IS that is over-saturating, and if it's an isotope-labelled version of the analyte, cosuppressing and reducing overall sensitivity.

Another common disaster is if the IS has been integrated badly (again usually because it's too dilute).

Is it possible to increase the concentration of the internal standard? That may make it perform with better stability throughout the process.

That sounds great although, would it be affecting my results should as well should the IS concentration be high?

If it is well separated from the target analyte (and any interference) then increasing the concentration should help, as long as you keep the concentration below a saturation point (not overloading the system). The rule has always been to make it at a concentration near the mid point of the calibration curve, but lately I am finding that if you put it closer to the top of the curve it is more stable. If you get better reproduceability of the internal standard areas then that should help to tell if the target is having problems or not.