Removal of steroid estrogens in drinking water treatment processes

The native estrogenic steroid 17β-estradiol ( E2 ) and it’s metabolites, estrone ( E1 ) and estriol ( E3 ), and the synthetic steroid 17α-ethinyl estradiol（EE2）may distribute to water bodies and impact the eco-system with their estrogenic potency. Most studies focus on the removal efficiency of the sewage treatment plants regarding their emissions; however, there is limited information on the elimination of estrogenic chemicals during drinking water treatment processes.
This research investigated the removal rates of drinking water treatment units by spiking two levels（100 and 500 ng/L）of the four estrogens into raw water. Four processes were simulated in the laboratory to evaluate the removal efficiencies：pre-chlorination, coagulation/sedimentation, rapid filtration, and post-chlorination. The study also reported the concentrations of the four estrogens in the raw water and treated water. Solid-phase extraction and LC/MS/MS with isotope-dilution techniques were utilized to analyze the four chemicals.
20－40％ of E1, E2, EE2 and E3 were removed in the pre-chlorination unit; the coagulation/sedimentation procedure eliminated 17－52％ of the chemicals, and E2 was the highest. The rapid filtration step took out over 95％ of the compounds expect for E3 ( 84－92％). The removal efficacy in post -chlorination process varied widely, which was 17－44％, and E2 was the lowest one. Obviously, the rapid filtration treatment is superior to the other processes in removing the chemicals ( p＜0.0001 ). The whole procedure got rid of over 88％ of the chemicals excluding the E3 ( 64－85％ ).
In terms of the influence of spiked levels, there is no significant difference of E1 and E3 removal among the four processes, but it was significant for E2 in the pre-chlorination and coagulation/sedimentation units ( p＜0.05 ). For E2, better efficiencies were observed at 500 ng/L in the pre-chlorination； however, the removal efficacy of coagulation/ sedimentation unit was higher at 100 ng/L. For EE2, there was only difference in the post-chlorination step ( p＝0.002 ). The elimination rates between these two spiked levels for E3 in the rapid filtration treatment and post-chlorination units and almost reach statisitical significance ( p＝0.053 and 0.059 ). Spiked levels did not influence the elimination performance through the whole procedure except for E3 ( p＝0.002 ).
At 500 ng/L level, the removal efficacy of E3 was much lower than that of E2 in pre-chlorination ( p＝0.031 ), the removal percentage of E3 was also notably lower than those of E1, E2 and EE2 in rapid filtration and through the complete process ( p ≦0.001 ). There was no statistical difference among the removal rates of the four compounds in the coagulation/sedimentation and post-chlorination units. Regarding the 100 ng/L level, the removal efficiencies of the four units for the four estrogenic compounds varied. Nevertheless, the only difference in the removal efficacy for the four compounds was the coagulation/sedimentation, which E2 was taken away much more than EE2 ( p＝0.028 ).
The four steroid estrogens were detected from same raw waters. Among the detected samples, the average levels were 1.57－2.37 ng/L. On the other hand, none of the four compounds was detected in the treated drinking water.