LAUSR

In the Bisphenol A measurement challenge [1], Prof. Phenole proposed to her class to develop a high-precision analytical procedure for the determination of bisphenol A in water. The students promptly embraced the challenge and developed an isotope dilution gas chromatography mass spectrometry (ID GC-MS) method using the isotopically labeled C5-BPA internal standard [2]. The students did all due diligence to set up the method but, during the quality control check, they found a − 8.0% bias at the low-end of the calibration curve. When looking for a possible explanation, the students came across two interesting studies: in 1976 Pickup and McPherson showed that the isotope dilution mass spectrometry (IDMS) calibration curve can be nonlinear [3], and in 2016 Raposo noted that the coefficient of determination (R2) is not at all suitable to establish whether a calibration curve is linear or not [4]. With this information at hand, the students decided to spend some time investigating the linearity of their calibration curve. As expected, the graph of the residuals showed a trend typical of a nonlinear response (Fig. 1) [5]. The students with a theoretical background were reassured by this finding and were confident that this was a data analysis problem with no additional experiments needed. They decided to use a nonlinear mathematical model for the calibration curve. At first, they used a generic quadratic function of the