Sorry, I don't agree with the other replies.
The "mismatch" in the calibration curve is cause by the choice of your fit.
Likewise if you choose calculated intercept y=ax+b), the negative bias will be higher as when you choose linear thru zero (y=ax).
According to ISO17025 your lower allowed error is 10% on the lowest and highest calibration curve. (Sorry for my English style, but I hope you understand what I mean).
If it is not on your instrument software, you can use excel to calculate these errors including for instance the 95% error bars.
Good luck!
- See more at: http://www.scientistsolutions.com/forum/analytical-chemistry-environmental-analysis-icp-icp-ms/question-about-aasicp-calibration-curve#sthash.ZXy4mCYv.dpuf