Structure-Property Relationship Study Of Surfactant Systems And Its Applications

Surfactants have a characteristic molecular structure, an amphipathic structure, consisting of hydrophilic and hydrophobic group. They are easy to be absorbed on the interface of two-phase and make their surface tension decreased significantly. These surface actives or properties are closely related to their unique molecular structure. Therefore, in order to have a good grasp of the molecular mechanism of surface activity and provide useful structure-based molecule synthesis information even applications, it is indeed extremely meaningful to illustrate the relationship between surfactant function and structure, especially from the level of molecular and electronic structure.A philosophy law is that material’s composition and structure defines the property and then defines the usage, the usage area reflects the property. The codes of physical, chemical and biological properties involve in composition and structure. There are two ways to accomplish the research of surfactants:experimental technology and theoretical calculation. Quantitative Structure-Activity Relationship/Quantitative Structure-Property Relationship (QSAR/QSPR) is the right method based on statistic modeling and theoretical calculation. Statistic simulation methodology could assist us to extract the useful information from large quantities of data and used to explain/predict the related activities, while theoretical calculation method could provide accurate analysis of physical properties, biological active, surface active, interaction, charge distribution and so on from atomic-level. In this dissertation, the application scope of QSPR and quantum chemistry were extended to discuss cloud point (CP)-structure relationship of nonionc surfactants, analysis the critical micelle concentration (cmc)-structure relationship by quantum chemistry calculation, statistical modeling and analysising the cmc behave of Gemini surfactants. According to the best model, a lead compound was desgined and synthesized. The cmc value was0.83mmol/L compared with the predict one0.89mmol/L. This experimental result confirmed the forcast ability of the best moldel.Many machine learning methods including multi-linear regression (MLR), partial least squares regression (PLS), support vector machine (SVM), least-squares support vector machine (LSSVM), Gaussian process (GP) and random forest (RF) coupled with Heuristic or Genetic algorithm (GA) variable selection. Critical cross criterion, leave one out (LOO) and leave N out (LNO), was used to show the statistics result in difference models. Joint with sussessfully used topologieal and constitutional descriptors the single point energy of the nonionic surfactants was calculated by density function theory (DFT), the most polpulator quantum chemistry method to model. These details were shown as following:(1) Discussed the CP of nonionic surfactants by QSPR approach. Nonionic surfactants are used widely in detergents, textile, pharmacy and enhanced oil recovery. More recently, a new application area is cloud point extraction (CPE) technology. CPE is a separation and pre-concentration procedure that has been extensively applied for trace metal determination in agreement with the principles of "green chemistry’ compared to organic solvent extractions. In view of that, we herein present a non-linear statistical modeling with Gaussian process to meet the gap. In present study, we first develop a linear model for a panel of62structures then extend to82structures including3Gemini molecular by various descriptors of hydrophilic or hydrophobic domain selected from heuristic algorithm, respectively. We demonstrate that (ⅰ) using the descrptors calculated from hydrophilic group can give compareable model even better than descriptors derived from hydrophobic group. This means hydrophilic group give important effect to CP and quite difference with traditional views,(ⅱ) The best moldeling was build by GP, a nonlinear model with many convariance functions, had better fitting and forcast ability than MLR and PLS, SVM and LSSVM means more sutibale to solve the CP-structure relationship of nonionic surfactants. And the best modeling was made by descriptors derived from hydrophilic group r2and rpre2was0.922and0.962, respectively.(ⅲ) Gemini nonionic surfactants had lower CP than related single tail ones.(2) The cmc of nonionic surfactants was studied by quantum chemistry method. Gaussian03software was used to perform quantum-chemical calculation and generated nine descriptors, such as ΔE,ΔG,ΔH, EHOMO and ELUMO, dipole moment D and related branch componment D-x, D-y and D-z. In this procedure, full geometry optimizations were performed at the B3LYP6-31g basis set in the following single-point calculations so as to check the stability of the results obtained at the different levels of DFT. Two topological descriptors, c-KHO and c-AIC2, calculated by CODESSA program were used to moldel, too. Stepwise regression was then used to create linear statistics models with experimentally determined critical micelle concention of nonionic surfactants. We demonstrate that MLR method could describe the relationship between nonionic surfactants’ structure and critical micelle concention very well with the most important factors as following:c-KHO>c-AIC2> RNNO>EHOMO>ELUMO. It was found that there were multicollinearity relationship among descriptors through further discuss. The linear equation was build after the PCA and the best four parameters model was shown as following:log cmc=3.509+0.671X1-0.593X2-0.372X3+0.061X4, R2=0.994, F=18.032, P=0.(3) A systematic statistics study and its application of Gemini surfactants’ cmc. Descriptors were generated by CODESSA, too. Investigation on the critical micelle concentration of23Gemini sulfonated/sulfated anionic surfactants and120Gemini cationic surfactants were performed coupled with variable selection by GA. The results were imparted with the following remarks:(ⅰ) Gemini had lower cmc and slower slope ratio than corresponding single tail surfactants. The best MLR model of sulfonated/sulfated surfactants shown log cmc=-4.462-0.016PNSA-1+7.844ABIC1+1.705XY, R2=0.9142、F=67.45、s2=0.2080.(ⅱ) The best MLR model of120Gemini cationic surfactants was log cmc=(3.483±0.3217)-(0.323±0.001) KH1-(1.948±0.132) NP-(0.306±0.07) BI-(0.06±0.02) HASA-2r2=0.900(R2=0.88RMSP=0.39with heuristic variable-selection.(ⅲ) Among these five models (GA-GP, GA-PLS, GA-SVM, GA-LSSVM and GA-RF), the GP model gave the best results by GA selection. Topological descriptors were very important factors which could affect the critical micelle concentration.(ⅳ) In addition, guided by the best model, a Gemini was designed and synthesized by1-bromotetradecane and N,N,N-tetramethylethylenediamine in ethanol solution followed with recrystallization in ethanol-ethyl acetate and vacuum drying. The product was characterized by IR, elenmental analysis, MS,’H NMR and13C NMR. A promising candidate, C14-2-C14Gemini cationic surfactant that possesses particularly high cmc potency as0.83mmol/L at25℃, about one fifth compared with tetradecyltri-methyl ammonium bromide means five times activity ability, was finded by conductimetric method. This experimentally measured value is in agreement with the model-predicted0.89mmol/L fairly well.