Sorption Performance,mechanisms And Machine Learning-based Prediction Of Typical Contaminants By Magnetic Porous Myriophyllum Aquaticum Biochar

With the rapid development of society and the continuous progress of industry,the pollution of water resources by various organic or inorganic pollutants is intensifying.Adsorption method is a simple,economical and efficient method to remove pollutants from water.At present,good results and applications for the removal effect and mechanism of pollutants have been achieved in a relatively simple system.However,in the complex system of the"real"scenario such as industrial park comprehensive sewage,soil leachate wastewater,and landfill leachate,higher performance is required for adsorption materials,including low cost,multi-function,ecological environment friendliness,simultaneous treatment of a variety of pollutants,and easy operation for separation and reuse.The interactions between co-existing pollutants and among co-existing substances,pollutants and adsorbents largely determine the removal of target pollutants and underlying mechanisms,which however is still unclear.Furthermore,the quick and accurate evaluation and prediction of treatment performance in such a complex and multi-parameter system is difficult to achieve.Therefore,in this study,two kinds of magnetic adsorbents（MMa B and MPMa B）derived from waste Myriophyllum aquaticum were synthesized.The adsorption performance of MMa B for polycyclic aromatic hydrocarbons（PAHs）was carried out.The selective adsorption of PAHs and surfactant recovery by MPMa B from soil eluents were investigated.The interferences of dissolved organic matter,ionic strength,p H and other factors were also discussed.The synergistic and competitive adsorption properties,interference factors and mechanisms of MPMa B on Cr（VI）and dye combined pollution were revealed.A model for the removal of pollutants in water by adsorption of porous carbon materials was constructed using machine learning,and the implementation conditions for the removal of pollutants in water by MPMa B were optimized.The paper has obtained some valuable results,specifically as follows:（1）MMa B exhibits good adsorption and magnetic separation performance for PAHs in water,and its saturation magnetization is 13.8 emu·g^-1.The adsorption of MMa B on the seven PAHs is achieved rapidly,and the adsorption equilibrium can be reached in 4-6 h.The fitting partition coefficient K_d for the 7 PAHs are NAP,91.3 L·kg^-1;ACE,563.7 L·kg^-1;FLU,1010.5 L·kg^-1;PHEN,2917.8 L·kg^-1;ANT,2706.6 L·kg^-1;PYR,9046.8 L·kg^-1 and FLUO 15969.0 L·kg^-1.There is a good linear correlation between adsorption coefficient log K_d and PAHs hydrophobicity log K_ow:log K_d=0.8091log K_ow+1.7034（n=7,R²=0.9959）.After 4 consecutive cycles,MMa B has a good cyclic adsorption capacity of 1.93 mg·g^-1.（2）The magnetic porous biochar（MPMa B）prepared by KOH activation,high temperature carbonization and magnetic modification method has rich mesoporous structure,high specific surface area(937.1 m²·g^-1)and good saturation magnetization(14.28 emu·g^-1).The saturated adsorption capacities（Q_m）of MPMa B for PHEN and PYR in aqueous solution are 108.7 mg·g^-1 and 12.22 mg·g^-1,respectively.In solutions containing Tween 80,rhamnolipid or 1:9 rhamnolipid-Tween 80,The Q_m of MPMa B against PHEN and PYR are 144.93 mg·g^-1,185.19 mg·g^-1,185.19 mg·g^-1 and 4.67mg·g^-1,44.05 mg·g^-1,48.31 mg·g^-1,respectively.MPMa B has a high selective removal rate of PHEN and PYR in all eluents.The average removal rate of PAHs is about97.19%±2.61（89.21%-99.86%）,the adsorption loss rate of surfactants ranges from1.46%to 39.74%,and the selective adsorption coefficient is in the range of 64.2-3335.8with a median of 1088.68.The p H（2-8）,Na Cl concentration(0.1-1.0 mol·L^-1)and DOM(50-400 mg·L^-1)of soil leachate had no significant effect on PAHs selective sorption by MPMa B.The average removal rate of PAHs was 95.85%±10.77（56.04%-99.91%）,and the maximum selective adsorption coefficient was 223252.5.（3）MPMa B has good adsorption properties for Cr（VI）and Rhodamine B in univariable system,and Langmuir model fitted saturated adsorption amounts of 175.4mg·g^-1 and 175.4 mg·g^-1,respectively.About 60%of Cr（VI）is reduced by MPMa B to Cr（Ⅲ）,which is less toxic.p H and coexisting ions have no significant effect on MPMa B adsorption of Rh B.The adsorption of Cr（VI）by MPMa B shows a p H-dependent manner.Coexisting anions significantly affect the adsorption of Cr（VI）by MPMa B,and the main mechanism is that the type and concentration of ions change the solution p H value.MPMa B exhibits synergistic and preferential adsorption of Cr（VI）in the presence of Rh B.For example,when coexist Rh B at concentration of 100 mg·L^-1,the sorption efficiency of Cr（VI）by MPMa B increases by up to 118.8%-527.3%.After the pre-adsorption of Rh B,the removal rate of Cr（VI）by MPMa B is still 95.81%.On the contrary,MPMa B can no longer adsorb Rh B（0.84%）after Cr（VI）pre-adsorption.The adsorption mechanism of MPMa B on Cr（VI）mainly includes reduction,surface complexation and electrostatic attraction,while Rh B adsorption is mainly through pore filling,hydrogen bonding andπ-πinteraction.（4）Compared with artificial neural network,decision tree regression and support vector regression models,random forest model（R²=0.906）is more suitable for predicting MPMa B adsorption of Rh B.Decision tree regression model（R²=0.924）is the most suitable for the prediction of Cr（VI）adsorption by MPMa B.By collecting the experimental data in the literature,the four models are used for verification,and the results show that the random forest and decision tree regression model can be better predict the adsorption of pollutants by different adsorbent materials.