Samples were derivatized and analyzed with a gas
chromatograph (GC) coupled to a quadrupole mass spectrometer
following a previously described method (Yu et al.,
2006; Yu, 2007), with the exceptions that the injection standard
was unlabeled 4,4-di-tert-butylbiphenyl (Sigma-Aldrich-
Fluka) and the gas chromatograph column was an Rtx-5 GC
column (Restek, Bellefonte, PA) with 30 m length 0.32 mm
i.d. 1 mm phase thickness. Calibration curves ranged from
0 to 980 mg/L for each PPCP to account for the 35-fold increase
in concentration caused by extracting 7 mL aqueous samples
into 200 mL aliquots of cyclohexane. The raw PPCP peaks were
normalized to the peak area of the injection standard, and the
fraction remaining for each PPCP was calculated by dividing
the mean concentration of triplicate effluent samples by the
mean concentration of triplicate influent samples. The standard
deviation for each mean concentration was estimated by
averaging the standard deviations for each of the triplicate
samples obtained from the calibration curve (Peters et al.,
1974). These average standard deviations were then used to
estimate the error of the quotient (mean effluent concentration
divided by mean influent concentration). Confidence
intervals (95%) were determined using this standard deviation
of the quotient with the Student’s t distribution.