Photocatalytic (UV-A/TiO₂) degradation of 17¦Á-ethynylestradiol in environmental matrices: Experimental studies and artificial neural network modeling

详细信息    查看全文

• 作者：Zacharias Frontistis ; Vasileia M. Daskalaki ; Evroula Hapeshi ; Catherine Drosou ; Despo Fatta-Kassinos ; Nikolaos P. Xekoukoulotakis ; Dionissios Mantzavinos
• 关键词：ANN ; Endocrine disrupting compounds ; Kinetics ; Synergy ; Transformation products ; Ultrasound
• 刊名：Journal of Photochemistry and Photobiology A: Chemistry
• 出版年：2012
• 出版时间：15 July, 2012
• 年：2012
• 卷：240
• 期：Complete
• 页码：33-41
• 全文大小：1041 K

文摘

The efficiency of heterogeneous photocatalysis to degrade 17¦Á-ethynylestradiol (EE2), a synthetic estrogen hormone, in environmentally relevant samples was investigated. In most cases, UV-A radiation at a photon flux of 2.81 ¡Á 10^?4 einstein/min was provided by a 9 W lamp and experiments were conducted at various concentrations of Aeroxide P25 TiO₂ (50-1000 mg/L), EE2 concentrations (50-900 ¦Ìg/L) and water matrices (from ultrapure water to secondary treated wastewater). Some runs were performed at photon fluxes between 6.4 ¡Á 10^?7 and 3.7 ¡Á 10^?4 einstein/min to study the effect of intensity on degradation. Changes in estrogen concentration were followed by high performance liquid chromatography. EE2 degradation, which follows first order kinetics, increases with (i) increasing catalyst loading up to a threshold value beyond which it remains unaffected; (ii) increasing photon flux and (iii) decreasing matrix complexity, i.e. the organic and inorganic constituents of wastewater retard degradation. This may be overcome coupling photocatalysis with ultrasound radiation at 80 kHz and 41 W/L power density; the combined sonophotocatalytic process acts synergistically toward EE2 degradation. Several transformation products were identified by means of UPLC-MS/MS and a reaction network for the photocatalytic degradation of EE2 is suggested.

An artificial neural network comprising five input variables (reaction time, TiO₂ and EE2 concentration, organic content and conductivity of the water matrix), thirteen neurons and an output variable (EE2 conversion) was optimized, tested and validated for EE2 degradation. The network, based on tangent sigmoid and linear transfer functions for the hidden and input/output layers, respectively, and the Levenberg-Marquardt back propagation training algorithm, can successfully predict EE2 degradation.