摘要
利用量子化学计算方法和基团贡献法,采集了61种聚酰亚胺分子结构模型单元的12种量子化学结构参数,并通过通径分析法筛选出5种影响该聚合物介电常数的主要因素;在此基础上,基于多元线性回归(MLR)和人工神经网络(ANN)方法构建了2种定量构效关系研究模型(QSPR),分析了模型的稳定性及预测能力.计算结果揭示了5种结构参数与材料介电常数之间的内在关系——含氟量的自然律e^(-F%)、偶极距μ、溶度参数δ与介电常数之间存在正相关关系,而最负原子净电荷q^-、侧基长度L则与介电常数则存在着负相关关系. MLR-QSPR模型具备较好的物理意义,ANN-QSPR则具有较好的精度,实验数据证明2种模型的平均误差均低于10%.依据MLR-QSPR模型设计了5种不同含氟量的聚酰亚胺链节结构,结果显示含氟量的增加有利于降低材料的介电常数值,但当含氟量达到一定程度后,介电常数趋于稳定,与文献报道实验结果相一致;当含氟量为34%时(k-3),材料的介电常数最低,为2.02.
Using the quantum chemical calculation method and the group contribution method, 12 kinds of quantum chemical structure parameters of 61 polyimide molecular structure model units were collected. In order to simplify the calculation process, a segment of the polymer chain was selected and saturated with methyl group, which was used as the model of the polyimide. Through the path analysis, 5 main factors that affect the dielectric constant of the polyimide films were further screened. On this basis, the multiple linear regression (MLR) and artificial neural network (ANN) methods were constructed. Two quantitative structure-property relationship model (QSPR) were built, and the stability and prediction ability of the models were analyzed. The results revealed the intrinsic relationship between the 5 structural parameters and the dielectric constant, i.e., the fluorine content e^-F%, the dipole pitch μ, and the solubility parameter δ of polyimide are positively correlated with the dielectric constant, while the most negative atomic net charge q^- and the side length L are negatively correlated with the dielectric constant. MLR-QSPR model has better physical significance and the ANN-QSPR has better accuracy. The accuracy of the models is validated by combining four structures: 6FDA-TriPMPDA, 6FDA-TriPMMDA, 6FDATPCF3PDA and 6FDA-TPCF3MDA in our lab. The experimental data show that the average error of the two models is lower than 10% under 1 kHz test condition. Five different fluorine-containing polyimide chain structures were designed according to MLR-QSPR model. The results show that the increase in fluorine content is beneficial to reduce the dielectric constant of the material, but when the fluorine content reaches a certain level, the dielectric constant tends to be stable, which are consistent with the experimental results reported in the literature. When the fluorine content is 34%(k-3), the material possesses the lowest dielectric constant of 2.02. Based on the results of this study, it is believed that QSPR has a good application prospect and theoretical significance in designing new polyimide materials and predicting its properties.
作者
范振国
陈文欣
魏世洋
刘腾
刘四委
池振国
张艺
许家瑞
Zhen-guo Fan;Wen-xin Chen;Shi-yang Wei;Teng Liu;Si-wei Liu;Zhen-guo Chi;Yi Zhang;Jia-rui Xu(Laboratory of Polymeric Composite and Functional Materials, Guangdong Laboratory of High-Performance Polymer Composites, Guangdong Engineering Technology Research Center for High-performance Organic and Polymer Photoelectric Functional Films, State Key Laboratory of Optoelectronic Materials and Technologies, School of Chemistry and Chemical Engineering, Sun Yat-sen University, Guangzhou 510275)
出处
《高分子学报》
SCIE
CAS
CSCD
北大核心
2019年第2期179-188,I0005,共11页
Acta Polymerica Sinica
基金
国家重点基础研究计划(973计划
项目号2014CB643605)
国家自然科学基金(基金号 51873239
51373204)
广东省"特支计划"科技创新领军人才项目(项目号 2016TX03C295)
广东省前沿与关键技术创新专项(项目号 2015B090915003
2015B090913003)
中国博士后科学基金(基金号 2017M612801)
高等学校基本科研业务费(项目号 161gzd08)资助项目
关键词
聚酰亚胺
介电常数
定量构效关系
多元线性回归
人工神经网络
Polyimide
Low dielectric constant
Quantitative structure-property relationship
Multiple linear regression
Artificial neural network
