【作者】 李世刚；

【导师】 刘辉；

【作者基本信息】 北京化工大学 ， 化学工程， 2004， 硕士

【摘要】 本文主要以择型催化为应用背景,开展了两方面工作:一部分工作是分子筛晶内扩散的宏观热力学模型研究,并提出预测晶内扩散的活化能、预报扩散系数的新方法;另一部分工作是采用分子模拟方法,开展分子筛晶内吸附的Monte Carlo模拟研究工作。首先针对烃类在分子筛催化剂内的扩散进行研究。收集了近年来发表的烃类扩散质在ZSM-5分子筛中的大量扩散活化能和扩散系数实验数据,基于此对广为引用的Xiao和Wei的GT扩散系数模型进行了评估和分析。结果表明,该模型的扩散系数预测值只与C3-C6等简单烷烃、苯和环己烷的扩散系数实验值在相同数量级,而对于结构较复杂的烃类,预测与实测相差几个数量级。建立了基于基团贡献法的分子筛晶内扩散活化能以及扩散系数的计算模型。通过与GT模型计算值以及文献报道的大量在ZSM-5分子筛中的扩散活化能和扩散系数实验数据的比较表明,基团贡献法比GT模型适用面广、精确度较高。特别是可以弥补GT模型不适用较复杂扩散质的缺陷。首次报道将晶内扩散活化能与扩散质的标准沸点相关,得到线性相关系数都接近或大于0.9,表明可以用与分子间作用力相关的标准沸点作为扩散质分子与分子筛孔道间相互作用大小的量度。接下来是从目前常用的分子模拟方法去模拟分子在分子筛里面的吸附行为,在文章的第三章主要介绍分子模拟原理和方法,同时用简单分子的相平衡对分子模拟的可靠性进行了检验。由于在分子模拟中要涉及到随机抽样,其随机数发生器的优劣直接会影响到模拟效果。在第四章中针对目前随机数发生器存在的缺陷,开发了一套新的随机数生成的新方法——超素数法,同时给出了相关证明和推导。在此基础上,又开发了两种更为优良的方法:优选乘子算法和超素数长周期算法,并且从统计性能方面进行了比较,其效果明显优于传统算法。在文章的最后,用GCMC方法模拟了N2在5A分子筛中的吸附情况,并且和文献报道的试验值进行了比较,得到了较为理想的结果,另外还通过截面的吸附情况分析了在5A分子筛里面的吸附能力分布。然而在石油化工中关心的基本上是长链烷烃以及芳香烃的反应,同时ZSM-5具有优良的分子选择特性,因此成为众多石油化工厂反应的催化剂。然而对于有强烈内部作用的长链分子,常用<WP=6>的GCMC方法模拟可能效率不是很好,因为链状分子的其中一个原子被取代对于整个构型的变化没有太大影响会导致统计性能很差。并且插入的成功的概率几乎为零,长链分子很容易和体系中其它分子重叠导致很高的相互作用势能。针对长链分子在模拟过程中难以插入的问题,引入了偏倚Monte Carlo方法,详尽的介绍其原理及具体实施过程,用此方法模拟了戊烷在ZSM-5中的吸附情况,得到了较为理想的结果。更多还原

【Abstract】 The objectives of this paper are related to the modeling of intra-crystalline adsorption and diffusion in zeolite with its potential applications to the shape selective catalysis. To this end, a new thermodynamic model is developed to predict diffusivities of hydrocarbons in zeolites, and molecular simulations are also carried out to predict adsorption isotherms of hydrocarbons with long chains in zeolites.In the first part of this paper, first, the well-known Xiao and Wei’s GT model for prediction of diffusivity of hydrocarbons in zeolites was critically evaluated based on a database set of the diffusional activation energy and the diffusional coefficients reported in the literature. It was demonstrated that while the GT model predicts quantitatively the diffusional coefficients of C3-C6 hydrocarbons, it yields large errors of several orders of magnitude in calculating the diffusional coefficients of hydrocarbons with complicated molecular structures. Secondly, a new model for prediction of activation energy and diffusivity of hydrocarbons in zeolites was developed based on the concept of group contribution method. Compared with the calculated diffusivity of GT model and experimental data, the present group-contribution-based model gives the same order of magnitude diffusivities of alkenes, while the predictions for most aromatic hydrocarbons, cyclohexane and it’s derivatives are in the same order of magnitude with the largest difference being 1～2 orders of magnitude. Finally, based upon an analysis of the physical meaning of the diffusional activation energy, the activation energy was found, for the first time, to be correlated with boiling points of the guest molecules. In most cases investigated in this paper, the correlation coefficient of diffusional activation energy of linear alkenes is over 0.90.In the second part of the paper, absorption in zeolites has been researched by using molecular simulation. The third chapter introduces basic theory and technique of molecular stimulation, and makes an example about phase-equilibrium of simple molecule to test the dependability of molecular stimulation. In addition, random sample is related to in this method, and stand or fall of its property influence the result of simulation directly. So addressing a method for generating quasi-random numbers may be timely and useful. To this end, in the fourth chapter a novel method has been presented to overcome the limitation existed in the conventional method. Based on a <WP=8>special property of prime numbers, two property-choiceness methods have been developed for generating quasi-random numbers, including optimum multiplicator method and long period method. Statistical results show that the new methods have advantages of satisfactory statistic properties over the multiplicative or mixed congruential methods. In the last part of the thesis, the grand canonical Monte Carlo (GCMC) method has been used to investigate the phenomenon of absorption for nitrogen in 5A zeolite. Satisfactory results has been obtained by comparing values of the simulation with those of the literature. At the same time, distributing of the adsorbent capability is discussed by investigating the absorption of nitrogen on some sections of 5A zeolite. Generally speaking, it is the linear and branched alkanes or benzene and its homologues that are interested in petrochemical reactions. And as one kind of catalyst for shape-selective reactions, ZSM-5 plays an important role in the chemical and the petroleum industries for its special apertures having the size with the adsorbent molecular. While, for long chain molecules with strong intra-molecular potentials GCMC algorithm will not be efficient because a displacement of a single atom will not change the conformation of molecule very much, then this will lead to poor sampling statistics. It is worse that the probability for adding a molecule in system successfully is almost equals zero because the molecule stand a good chance to overlap with other molecules having existed in the system and causes p更多还原