Study On The Role Of Free Radical In The Structural Evalution And Application Of Char

With the depletion of oil resources,the utilization of heavy organic resources(coal,biomass,oil shale and heavy oil,etc)has aroused extensive research interest.In addition to the preparation of tar and gas,solid product coke has been widely used.Solid product char obtained from coal is mainly used in five fields,such as metallurgy,chemical,adsorbent,clean fuel and formed coke.Similarly,coke prepared from coal is not only mainly used in blast furnace ironmaking industry,but also used in casting,chemical,ferroalloy and nonferrous metal smelting.The biochar is mainly used for adsorption,catalysis,chemicals,the production of biofuels and the control of pollutants.Although coke is applied in the above fields,there are still some problems such as overcapacity and narrow utilization scope,which indicates that people have insufficient understanding of coke structure,and then unable to exploit the structural characteristics of coke to open up more potential fields for coke utilization.Besides the properties of raw materials,coke structure is closely related to the pyrolysis process.In order to solve those above problems,it is necessary to deeply understand the relationship between pyrolysis mechanism and coke structure.The pyrolysis of heavy organic resources follows the free radical reaction mechanism,in which the formation of pyrolytic product coke is accompanied by the change of free radical properties.Therefore,associating coke structure from the perspective of pyrolysis mechanism-free radical may open up a new field for coke application.In order to open up a wider application field of coal-char and biochar and make better use of coke,this paper selected four applications,including adsorption by coal-char,adsorption and catalysis by biochar,and metallurgical coke to conduct the following research.A variety of coals and biomass were used as experimental materials.The structural characteristics and free radical properties of coal-char and biochar were comparatively investigated.Based on the strong radical sites newly found in the experiment,construction of single Pd atom or nano Pd cluster sites on biochar surface by the radical sites on biochar was explored.Based on the ultimate analysis,proximate analysis and the radical concentration results of coal blends,the quality of coke(Gray-King Coke Type,GKCT)was forecasted,and the role of multi parameters in predicting GKCT was analyzed by machine learning models.The following main conclusions were obtained:(1)The chars prepared from pyrolysis of HeBi coal(HB)and corncob biomass(CC)at temperatures of 300-750 ℃ all contain 3 types of radicals measurable by electron spin resonance(ESR).The type of radicals that was not reported before strongly adsorbs or reacts with O2 at room temperature(strong radicals).The adsorbed O2 forms C-O and C=O bonds and cannot be removed by N2 purge.Another type of radicals adsorbs O2 which can be removed totally by N2 purge(weak radicals).The third type of radicals locates in the interior of chars which cannot be accessed by O2(enclosed radicals).All chars contain 3 types of radicals determined by ESR,which are strong,weak and enclosed radicals,respectively.Radical concentration(CR)of chars decreases to 0.0 μmol/g when pyrolysis temperature reaches 750℃,and the aromatic structures in the char are completely connected to each other,leaving few isolated electrons.(2)The char formation all shows a two stage behavior,which can be characterized by the char’s yield,the H/C and O/C ratios,the fala determined by 13C nuclear magnetic resonance(13C NMR)that quantifies the methyl group on aromatic rings,the(AD3+AD4)/AG determined by Raman to represent the active sites,and the CR by ESR.And the different results characterized by these parameters are due to the structural differences between HB and CC.That is to say,the structure of HB has larger aromatic rings with fewer side chains than that of CC.The first stage is in 300-600℃,which involves the release of volatiles,the formation and near disappearance of weak and enclosed radicals.In the second stage,600-750℃ involves merging the aromatic structures of the chars through the methylene radicals on aromatic rings,which diminishes the strong radicals.(3)Construction of single Pd atom or nano Pd cluster sites on biochar surface by the radical sites on biochar is explored by the interaction of strong radical sites and Palladium ion(Pd2+)on the surface of biochar.The changes of CR of fresh(oxygen-unexposed)and oxygen-exposed walnut shell derived chars obtained from pyrolysis at 600℃(F-WS-600 and O-WS-600)impregnated Palladium acetate toluene solution can be divided into two stages with 12 h as a demarcation point.The heavy organic matter covered on the surface of biochar will be washed away by toluene solution in the first stage leading to a increase of CR.In the second stage,the Pd2+will interact with the radical sites resulting in a decrease of CR.F-WS-600 impregnated Palladium acetate toluene solution for 24 h will form a small amount of single atom,and the size of Pd will reunite to form nanoatoms after impregnating for 84 h.But Pd cluster can be only formed by O-WS-600 impregnating Palladium acetate toluene solution.(4)CR can correlate mean maxmium vitrinite reflectance of coal well.Multi parameters range(MPR)prediction algorithm developed in this work shows good prediction capability and reasonable results in a chemical reaction viewpoint.The coal blends’ carbon content(C)from the ultimate analysis,CR from the ESR analysis,and ash content(A)from the proximate analysis are the major parameters to predict G-coke.The highest Precisions from single parameter range(SPR)and MPR prediction are 59.1%from C or CR and 81.3%,respectively.The parameter range for G-coke corrsponding to highest Precision from MPR is 82.4-88.7%for C,<11.4%for O,10.1-21.1 μmol/g for CR,7.2-12.7 for A,and<7.2 for M.The machine learning methods,K-Nearest Neighbors(KNN),Linear Discriminant Analysis(LDA)and Support Vector Machine(SVM),are capable to predict G-coke formation with high Precision,Accuracy and Recall when 4-5 parameters are used.The prediction by LDA and SVM are more meaningful in chemical reaction viewpoint than KNN,The prediction performance of Back Propagation Neural Network(BPNN)model on the formation of G-coke is lower than that of other three models and does not agree with the chemical knowledge in the coking process.The parameter CR in SVM plays an important role in predicting GKCT.