【作者】 孙晓岩；

【导师】 陈标华；

【作者基本信息】 北京化工大学 ， 化学工艺， 2008， 博士

【摘要】 本文以苯与丙烯烷基化反应的MCM-22分子筛催化剂为重点研究对象,首先利用计算机模拟技术,较为深入地研究了苯与丙烯在不同织构特性分子筛MCM-22、β和ZSM-5上的行为特性后,利用MCM-22分子筛相对优越的特点,着重对其作为催化剂的微观和宏观织构特性设计及其性能控制进行了较为系统的研究,从不同层面上对其结构进行优化,调整催化剂本征性能,并结合人工智能技术构建了关联催化剂结构性能、反应工艺条件和催化反应性能的人工神经网络模型。利用分子模拟中巨正则蒙特卡罗和分子动力学两种经典的计算方法,对MCM-22,β和ZSM-5分子筛上苯与丙烯分子的吸附和扩散行为进行模拟计算,借此研究比较不同织构特性分子筛中的不同孔道内分子行为的差异,对三种分子筛用于催化烷基化反应不同的结果进行分析和预测。结果表明,苯和丙烯分子同时在分子筛上吸附时,存在着竞争吸附行为,其在不同的孔道内扩散速率差异很大。苯与丙烯在分子筛上发生反应时,由于β分子筛具有较大的孔道尺寸,其产物异丙苯可以较快的发生扩散,减少了异丙苯与丙烯发生深度烷基化反应的机会,可以提高目标产物异丙苯的选择性,但其过强的酸性也使其失活较快。ZSM-5分子筛具有的10MR孔道更小,更加不利于反应的发生。MCM-22分子筛具有更大的12MR超笼,但由于笼间由10MR窗口相连,苯分子在其中扩散较困难,而且也不利于苯与丙烯反应的发生,所以反应主要发生在层间半圆环口袋的酸性位。如果能够使更多的酸性位暴露在表层,可以使MCM-22分子筛的催化性能大大提高。采用动态水热合成法,通过添加不同用量有机辅助助剂丙三醇的方式合成了一系列用于苯与丙烯烷基化反应的MCM-22分子筛催化剂,在表征分析和实验评价的基础上,详细考察了丙三醇的加入对分子筛物理织构特性(比表面积、孔容和表面形貌等)及其催化性能的影响。结果表明,适量有机辅助助剂丙三醇的添加,可以起到分散模板剂,提高模板剂溶解性的作用,更好发挥其模板导向作用,提高分子筛的结晶度。丙三醇的加入还可以合成较大孔容以及较高比表面积的MCM-22分子筛,比表面积可以提高至489 m~2/g,孔容可以增加到0.55 cm~3/g。所获得的分子筛样品在焙烧过程中,由于丙三醇存在较多的醇羟基可以改变层间的T-OH之间的相互作用,使层与层的T-OH之间不易脱水形成T-O-T键,使焙烧后的分子筛具有较小的晶体尺寸以及较薄的层厚度,使更多的半圆环口袋暴露在外层表面,即使较多的酸性位暴露在外层表面,从而提高其催化活性;丙三醇的加入使分子筛层间距加大,晶粒分散,在催化苯与丙烯烷基化反应过程中,目的产物异丙苯较容易扩散,这就减少了异丙苯与丙烯进一步深度烷基化生成多异丙苯的几率,使异丙苯的选择性可以提高5%以上。分子筛催化剂的宏观特性调控主要是对粉体分子筛加工成工业粒级催化剂过程中的工艺条件进行研究,基于正交实验设计系统考察了挤条成型过程中各类助剂,如粘合剂、胶溶剂、扩孔剂和水粉比等对分子筛催化剂物理化学特性和催化性能的影响。结果表明,成型过程中,适宜的助剂加入,可以调节催化剂的本征性能,包括比表面积、孔径分布和酸性。SB粉主要调节催化剂颗粒强度,PEG20000主要调变孔径分布以及孔容,硝酸对分子筛可以产生骨架脱铝效应,调节骨架铝含量,进而可以调节催化剂的酸性。这样可以调控催化剂的本征性能使之与烷基化反应过程以及扩散过程达到相互匹配,达到有机结合。以异丙苯选择性为目标函数确定的最佳工艺条件为SB粉用量25%,硝酸用量20%,PEG20000用量10%,水粉比0.9。采用微观和宏观织构特性联合调控技术所制备的催化剂,其整体性能得到较好的提高,在相对苛刻的评价条件下,丙烯转化率提高大约2%,目标产物异丙苯选择性提高约3%。在实验完成了分子筛催化剂微观和宏观层次织构特性调控后,通过构建人工神经网络模型来关联催化剂织构特性,反应条件和催化性能的关系,更好的服务于催化剂设计。构建的网络模型表明,无论单输出还是多输出网络,预测值和实验值两者之间的平均相对误差较小并且具有较高的相关系数,说明所建立的BP神经网络模型可以较准确的预测苯与丙烯烷基化反应性能,得到较好的预测结果。然而,单输出网络由于其针对性较强,较多输出网络具有更加精确的预测能力。该神经网络不仅具备了预测苯与丙烯烷基化反应性能的功能,为实际生产提供理论指导,而且该网络模型的开发还可以反向应用于催化剂的设计与开发,确定适用于反应的催化剂织构特性和反应条件,增强实际催化剂制备的目的性,减少催化剂开发过程中繁琐的实验过程。更多还原

【Abstract】 Zeolite MCM-22 had been used as the catalyst for alkylation of benzene and propylene.The dissertation used the cyber-simulation technology to study the action of benzene and propylene in the zeolite MCM-22,βand ZSM-5 with different textural,providing the theories of benzene alkylation with propylene on zeolites.Then the technology of catalyst design engineering to optimize the configuration of the MCM-22 zeolite catalyst from micromechanism to macromechanism,improving the performance of the catalyst.Moreover,the BP neural network model had been established to predict the reaction results according to the catalyst properties and reaction condition.The diffusion and adsorption behaviors of benzene and propylene in the zeolite MCM-22,βand ZSM-5 had been studied by Molecular Dynamics（MD） and Grand Canonical Monte Carlo（GCMC） simulations. The diffusion coefficients of benzene and propylene in the MFI,MWW and BEA zeolites were calculated by simulating the mean-square displacements（MSD）.The competitive adsorption was taken place when benzene and propylene adsorbed on zeolites.In the alkylation reaction, because of the large channels inβ,cumene molecule could diffuse quickly;the cumene and propylene could react and create multipropylbenzene not easily,so the selectivity of cumene was higher than MCM-22 and ZSM-5.MCM-22 has supercages which sizes are larger than the 12MR channels ofβ,but the locations of the 10MR windows in the supercages are perpendicular to the direction of the motion,which could hinder the alkylation reaction to occur.So,for MCM-22,the alkylation happened on the cavities at the top and bottom surface,which had the advantages to improve the catalytic performance for the benzene alkylation with propylene.The sizes of channels in ZSM-5 were smaller than the others which were the disadvantages for alkylation.We used the auxiliary chemicals—glycerin under hydrothermal conditions to obtain the MCM-22 zeolite with various morphologies, including the specific areas and the platelets size.The alkylation of benzene with propylene under liquid-phase reaction conditions catalyzed by the prepared MCM-22 zeolites was also carried out to investigate the catalytic performances of the modified zeolites.A conceivable formation mechanism and the reason of catalytic performance improvement in the presence of glycerin as an auxiliary chemical had been suggested that adding appropriate amount of glycerin when MCM-22 synthesis,not only could increase the solubility of hydrophobes and improve the crystallinity, but also could increase the difficulty of forming T-O-T bonding.During calcinations,glycerin molecules enhanced the hindrance of two double-layers and led them to form T-O-T uneasily,and make the platelets thinner and smaller.On these zeolites,cumene could not be further alkylated to form di-isopropylbenzene or tri- iso- propylbenzene easily and the selectivity of cumene could be improved about 5%.The catalyst macroscopical properties were controlled through molding progresss.The influences on the textural properties and catalytic performance were studied by adding different amount of SB powder, HNO₃,PEG20000 and water.The results showed that in the molding progress,adding the suitable amount auxiliary materials could adjust the properties of catalyst,including the surface areas,pore size and acidity. So the properties and alkylation reaction-diffusion were matching each other,which could improve the performance of catalyst.Taking the selectivity of cumene as the objective variable,the optimum technology was followed:15%SB powder:20%HNO3:10%PEG20000:90%H2O. Base on the optimization of microscopical and macroscopical configuration,the conversion of propylene was improve about 2%,and the selectivity of cumene was improved about 3%.Because the alkylation process was very complex and the model of mechanism was established difficultly,The BP neural network was used to predict the results of benzene alkylation with propylene.The average relative deviations of determination showed a good correlation between estimated and experimental data sets.There were high correlations between experimental and estimated data curves,which were another proof that the high performance of ANN for estimation of the product distributions of alkylation reaction.Moreover,the network could be applied to direct the catalysis design,confirming the optimized configurations and reaction conditions and improving the efficiency of experiments on catalysis research.更多还原