A Quantitative Structure-Property Relationship Study on Reaction Rate Constants for Reductive Debromination of Polybrominated Diphenyl Ethers by Zero-Valent Iron

Abstract

A quantitative structure property relationship (QSPR) study was performed in this work to develop models for predicting reaction rate constants for reductive debromination of polybrominated diphenyl ethers (PBDEs) by zero-valent iron (ZVI). Both multiple linear regression (MLR) and artificial neural network (ANN) methods were employed for QSPR studies based on the experimental kinetic data of the fourteen PBDE congeners. Both the developed MLR and ANN models could give satisfactory prediction abilities, and the performance of the ANN model seems slightly better than that of the MLR model. In addition, energy of lowest unoccupied molecular orbital (ELUMO) and total energy (TE) were found to be the two relatively important variables in the ANN model via the assessment using both the Garson’s algorithm and connection weight approach.

Artificial Neural Network Modeling and Genetic Algorithm Optimization for Cadmium Removal from Aqueous Solutions by Reduced Graphene Oxide-Supported Nanoscale Zero-Valent Iron (nZVI/rGO) Composites

“…Connection weight is to evaluate the variable importance through the calculation that records the input-hidden and hidden-output connection weights between each independent variable and dependent variables across all hidden neurons [55]. The Garson equation is useful to exhibit the relative influence of the independent variables on the dependent variable, which can be described as $P_{av}=\frac{true{\sum }_{b=1}^N\left(\frac{|w_{ab}|}{true{\sum }_{d=1}^M\left|w_{db}\right|}\left|e_{bv}\right|\right)}{true{\sum }_{a=1}^M\left(true{\sum }_{b=1}^N\left(\frac{\left|w_{ab}\right|}{true{\sum }_{d=1}^M\left|w_{db}\right|}\left|e_{bv}\right|\right)\right)}$ where P is the percentage of influence for the input neuron, w represents the weight between input and hidden neuron, e stands for the weight between hidden and output neuron, M is the number of neurons in input layer, N is the number of neurons in hidden layer, and v is the number of output neuron.…”

RM1 Semiempirical Model: Chemistry, Pharmaceutical Research, Molecular Biology and Materials Science

“…Therefore, the use of semiempirical methods shows to be a serious alternative to obtain the optimized geometries for coupling constants calculations, instead to be viewed as just a manner to determine preliminary geometries for the robust QM calculations". 76 The high structural quality was also the reason for Hu et al 77 to employ RM1 as the theoretical method of choice in their investigation of quantitative structure property relationships (QSPR). In their study, RM1 was used to describe models for predicting reaction rate constants for the chemical reactions of reductive debromination of polybrominated diphenyl ethers by using zero-valent iron.…”

“…In their study, RM1 was used to describe models for predicting reaction rate constants for the chemical reactions of reductive debromination of polybrominated diphenyl ethers by using zero-valent iron. Hu et al 77 remarked: "In this study, we utilized a new generation semiempirical RM1 method to obtain more accurate molecular parameters instead of using AM1 method which has been chosen by the previous researchers [6,7,13], since RM1 maintains the mathematical structure and qualities of AM1 while it significantly improves its quantitative accuracy [14]". Spectroscopic data were investigated by using RM1 geometries optimized, such as data of the vibrational infrared and visible absorption spectroscopies.…”

“…Energetic properties of organic compounds can be calculated in both forms: isolated, and in solvent medium. The solvent effects on the RM1 energetic properties of organic compounds 77,100,101,[105][106][107][108]122,123,[125][126][127]133,134 have been calculated, for example, by using the COSMO polarizable continuum model. 135 Another important aspect of RM1 is the easiness to perform the calculations.…”

“…In addition, the RM1 method has also been employed by organic chemists to calculate the energy of the lowest unoccupied molecular orbital (LUMO), and of the highest unoccupied molecular orbital (HOMO), that are associated with the reactivity of organic compounds. 70,77,100,[104][105][106]117,125 Organic compounds, that had their orbital energies (LUMO, and HOMO) calculated by RM1, include: 3-arylideneflavanone; 70 E,Z isomers of chromanone; 70 3-arylflavones; 70 polymethine cyanine dyes in solutions of β-cyclodextrin; 106 trans-1-acetyl-4,5-di-tert-butyl-2-imidazolidinone; 104 a set of (E)-4-aryl-4-oxo-2-butenoic acids; 105 a set of nitroaromatic compounds; 117 species involved in a revised mechanism of Boyland-Sims oxidation; 125 2,6-dibromo-4-[(E)-2-(1-methylquinolinium-4 -y l ) e t h e ny l ] -p h e n o l a t e ; 100 a n d 2 100 In this sense, a QSPR study reported by Hu et al 77 employed the RM1 method to calculate geometry optimizations, energy, and electronic properties of a set of fourteen polybrominated diphenyl ether congeners, as such 2-mono-BDE (BDE-1), and 3-mono-BDE (BDE-2). The descriptors calculated by RM1 were used in statistical analyses based on artificial neural network models.…”