CAS OpenIR  > 研究所(批量导入)
Thesis Advisor张香平
Degree Grantor中国科学院研究生院
Place of Conferral北京
Degree Discipline化学工艺
Keyword离子液体 智能算法 性质预测 数据库 量化计算

近年来,离子液体因其特有的低蒸汽压、良好的热稳定性和结构可设计性等优势,在诸多领域展现出了巨大的应用潜力。然而离子液体种类繁多,完全依赖实验方法测量离子液体的性质和筛选合适的离子液体不切实际,迫切需要开发性质预测模型及系统高效的筛选方法,为离子液体工业化应用提供支撑。本论文建立了离子液体纯组分性质、气液相平衡性质预测模型,针对气体分离,综合考虑多种重要性质提出了系统高效的离子液体设计筛选新方法,并采用量化计算对代表性气体分离机理(CO2/CH4、C2H2/C2H4和NH3/CH4)进行了深入研究,研究结果为气体分离工业化应用提供了参考和指导。本论文的主要研究内容及成果如下:(1)基于智能算法的离子液体纯组分性质预测。针对常见的咪唑类离子液体的粘度性质,采用离子片-对应态方法(FC-CS)计算的临界性质等作为输入参数,基于多元线性回归(MLR)及支持向量机(SVM)算法建立了预测模型。建立的模型对45种咪唑类离子液体1079条数据点的平均绝对相对误差(AARD)分别为24.2%和3.95%。为了进一步扩大粘度的预测范围,收集了89种离子液体1502条数据点,采用量化计算得到的Sσ-profiles分子描述符作为输入参数,结合MLR和SVM算法建立了模型,二者的AARD分别为10.68%和6.58%。同时,针对离子液体热容性质,采用Sσ-profiles分子描述作为输入参数结合极限学习机(ELM)算法,分别建立了MLR和ELM预测模型,对46种离子液体2416条数据的AARD分别为2.72%和0.60%。以上结果表明本论文所建立的模型具有良好的预测效果。 (2)H2S在离子液体中溶解度数据库及预测模型。H2S在离子液体中溶解度的数据库及预测模型对于工业气体脱硫新过程开发非常重要。然而目前缺乏相关的数据库及准确的预测模型。本论文收集了2007-2016年1334条H2S溶解度数据建立了数据库。在此基础上以离子片作为输入参数,结合ELM算法建立了预测模型,该模型对27种离子液体1282条数据的AARD为4.12%。进一步采用量化计算得到的Sσ-profiles作为输入参数,结合ELM算法建立了新的预测模型。其中Sσ-profiles采用两种不同的划分形式,得到了两个定量结构性质关系(QSPR)模型,二者的AARD分别为3.73%和3.80%,效果均优于离子片作为输入参数建立的模型,也优于文献中报道的相关模型。(3)建立了考虑多因素的离子液体设计筛选新方法。离子液体在气体分离中的应用已经吸引了工业界和学术界的广泛关注。然而,如何建立高效、系统的设计筛选方法是一个重要的挑战。本研究采用COSMO-RS方法计算获得了包含一万多种离子液体对12种常见工业气体共163020条亨利系数,建立了相应的基础数据库。在此基础上,考虑离子液体对气体的质量和摩尔吸收能力、解吸能力、选择性以及其本身的熔点、粘度、毒性等性质建立了系统高效的设计筛选新方法,提出了吸收分离指数(ASI)新概念,可综合考虑离子液体的吸收和选择能力。以两种典型的气体分离应用为例(CO2/CH4和C2H2/C2H4),验证了上述方法的可靠性。(4)离子液体气体分离机理分析。针对三种典型的气体分离应用(CO2/CH4、C2H2/C2H4和NH3/CH4),分别选取了三种代表性的离子液体[S221][CCN3]、[S222][Ace]和[OHemim][NTf2],采用量化计算的方法对气体分离机理进行了系统深入的研究。结果表明氢键是分离气体的关键,通过AIM分析发现氢键以静电为主导,部分强的氢键具有了共价键的特征,同时发现氢键的键长分别与ρBCP和?2ρBCP存在良好的线性关系。进一步通过SAPT能量分解分析发现静电在离子液体与吸收量大的气体([OHemim][NTf2]-NH3、[S221][CCN3]-CO2和[S222][Ace]-C2H2)之间的总相互吸引能中占主导地位。

Other Abstract

The combination of cations and anions can yield 1018 ionic liquids (ILs) systems, and thus it is experimentally time consuming, expensive and impossible by measuring all possible ILs for a special purpose. Therefore, it is necessary to predict the properties of ILs using computational methods. Recently, ILs have attracted much attention for absorbing and separating gases in both academic and industrial communities. However, a systematic database and screening method is scarce as well as the mechanism study of gas separation is not in depth. In this study, some predictive models for properties of ILs are established firstly, and then a new, systematic, and efficient screening method for ILs selection for absorption and separation of gases subject to important target properties is proposed. Finally, the mechanism study of gas separation is studied using the quantum chemistry calculation method. The main innovative work and results of the dissertation are as follows:(1) Physicochemical properties prediction of pure ionic liquids. Firstly, two models, integrating the fragment contribution-corresponding states (FC-CS) method with multiple linear regression (MLR) and support vector machine (SVM) algorithms, are proposed to predict the viscosity of imidazolium-based ILs. The average absolute relative deviation (AARD) of the entire data set of the MLR and SVM is 24.2% and 3.95%, respectively. Then, two novel QSPR models are developed to predict the viscosity of ILs using the MLR and SVM algorithms based on Sσ-profile molecular descriptors. The AARD of the total data set (89 ILs, 1502 data points) of the MLR and SVM is 10.68% and 6.58%, respectively. Thirdly, In order to estimate the heat capacity of ILs, statistical models have been proposed using the Sσ-profile and MLR and relatively new ELM algorithms. The AARD of the total data set (46 ILs, 2416 data points) of the MLR and ELM is 2.72% and 0.60%, respectively. The above results suggest that the established nonlinear models have better predictive performance.(2) Phase equilibrium study of ionic liquid system (H2S solubility is selected as case study). This study firstly established a comprehensive database on the H2S solubility in ILs, which includes 1334 pieces of data covering the period from 2007 to 2016. Based on the database, a new model is proposed using the ELM intelligence algorithm and number of fragments as input parameters. A total of 1282 pieces of data for 27 ILs is used to build and test the model. The AARD of the entire set of the ELM model is 4.12%. Because the information of the fragments is not very rich, two novel models are developed using the ELM algorithm and Sσ-profile descriptors. The AARD of the two ELM models of the entire data set is 3.73 % and 3.80 %, respectively. These results suggest that proposed ELM models can be useful for the prediction of the H2S solubility in ILs. (3) Systematic screening method for gas separation. In this study, an extensive database of estimated Henry’s law constants of twelve gases in more than ten thousand ILs through the COSMO-RS method, is established (including 163020 data points). Based on the database, a new, systematic, efficient and reliable screening method considering various important properties such as toxicity, bioaccumulation potential, viscosity, melting point, molality and molarity absorption capacity and separation performance, has been proposed. In addition, a new concept, namely, absorption separation index (ASI) is proposed in the above mentioned screening method. Application of the database and the screening method is highlighted through case studies involving two important gases separation problems (CO2 from CH4 and C2H2 from C2H4). The results demonstrate the effectiveness of the screening method together with the database to explore and screen novel ILs meeting specific requirements for absorption and separation of gases. (4) Mechanism study of gas separation. Based on the above mentioned work, three representative ILs ([S221][CCN3], [S222][Ace] and [OHemim][NTf2]) are selected for studying the mechanism of the corresponding gases separation problems (CO2/CH4, C2H2/C2H4 and NH3/CH4) using the quantum chemistry calculation method. The results show that hydrogen bond (HB) is very important for gas separation, and the HB is mainly electrostatic dominant, sometimes even covalent dominat for strong HB. Meanwhile it is found that there may exist linear relationships between the distance of HB and electron density ρBCP/Laplacian value ?2ρBCP. Through symmetry adapted perturbation theory (SAPT) it is found that electrostatic energy plays main role in total attractive energy of ILs and ILs-gases ([OHemim][NTf2]-NH3, [S221][CCN3]-CO2 and [S222][Ace]-C2H2). 

Document Type学位论文
Recommended Citation
GB/T 7714
赵永升. 离子液体性质预测及其在气体分离中的应用[D]. 北京. 中国科学院研究生院,2017.
Files in This Item:
File Name/Size DocType Version Access License
离子液体性质预测及其在气体分离中的应用.(12024KB)学位论文 限制开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[赵永升]'s Articles
Baidu academic
Similar articles in Baidu academic
[赵永升]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[赵永升]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.