MaoyuChen
Hi Maoyu,
The problem here seems to be your preferences for beam seconds. The problematic samples have low U238, but you've based beam seconds on U238 (since it is the index channel) with a cutoff of 1000.
Without valid beam seconds, the down-hole fractionation correction will not work as expected. In this case, you could choose TotalBeam as the index with a cutoff of 1000000. Then those same samples look like this:
Now the downhole correction is fine and none of the 13NQ37 DC ratios are < 0:
All the best,