烷烃衍生物第一电离能的自相关拓扑研究

文中定义烷烃衍生物分子中非氢原子的染色序数(f_i),它对于非氢原子具有优异的选择性。由f_i建构1阶染色序数自相关拓扑指数(¹F)的定义式为：¹F=［Σ(1+k)^-0.25·(f_i·f_j)^-1］^0.5，它对烷烃衍生物分子具有良好的结构选择性。32种脂肪族胺、醇、醚、硫醇、硫醚类化合物的第一电离能(I_p,eV)与¹F及O，N，S原子的电负性(X_PL)的关系式为：I_p=13.0264-3.8208¹F +0.1773X_PL,R=0.9929。令人满意的27种卤代烷(F，Cl，Br，I)的关系式为：I_p=8.2727-2.6406¹F+1.5168X_PL,R=0.9979。这两个模型较好地揭示了烷烃衍生物第一电离能的递变规律。经Jackknife法检验，具有模型稳健性,并对23种烷烃衍生物I_p估算，显示良好的预测能力。因此，文中为烷烃衍生物第一电离能的预测提供一种有效方法。上述59种烷烃衍生物的I_p与¹F的相关系数为0.9791，明显优于著名的Kier指数(¹X^v)（其R仅为0.4190）。结果表明，所建定量结构性质相关(QSPR)模型的相关性高、稳健性好、预测能力强。

Autocorrelation topological study on the first ionization potentials for aliphatic amines, alcohols, ethers and haloalkanes

An atomic colouring number (f_i) was defined in this paper. It has an excellent selectivity for non-hydrogen atoms in alkane and alkyl derivative molecules. The first order autocorrelation topological index (¹F) of the colouring number was defined as: ¹F =［Σ(1+k)^-0.25·(f_i·f_j)^-1］^0.5, which showed a good structural selectivity for alkyl derivatives. The first ionization potentials (I_p, eV) of 32 compounds of aliphatic amines, alcohols, ethers, thio-alcohols and thio-ethers could be expressed as a function of ¹F and the electro negativity (X_PL) of O, N and S atoms: I_p=13.0264-38208 ¹F+0.1773 X_PL, R=0.9929. A satisfactory equation for 27 compounds of haloankanes was developed as follows: I_p=8.2727-2.6406 ¹F+1.5168X_PL, R =0.9979.These two models elucidate the rule for variation of I_p for different alkyl derivatives. Furthermore, a modified Jackknife test was performed to validate the model robustness, and the values of I_p of 23 compounds of alkyl derivatives are estimated with a good predictability. Therefore, this paper provides an effective method to predicate the first ionization potentials of alkyl derivatives. A correlation coefficient between I_p and ¹Ffor above 59 alkyl derivatives is equal to 0.9791, which is much better than that of the famous Kier's index ¹X^v (R=0.4190). These results demonstrate that the quantitative structure property relationship(QSPR) models have a high correlativity, fine stability and precise predictability.