I run Boron with 10 other analytes on ICP-OES. As a commercial lab we run a large amount of samples. I have noticed that all the other analytes in the control (QC) remains relatively and accurate through the run +/- 10%. On the other hand B recovery is accurate 90% at the beginning of the run and by 60 samples, it drops by 50%. By the end of a long run, it is usually off by 90%. Any ideas on how to tackle this.