Correlations are given between ESCA chemical shifts and partial atomic changes calculated by the modified Sanderson method. Elements included are boron, carbon, silicon, germanium, tin, nitrogen, phosphorus, arsenic, oxygen, sulfur, selenium, fluorine, chlorine and bromine. Successful correlations were obtained for most elements; the best success was for situations involving a constant number of σ bonds. The method was least successful for nitrogen, oxygen, chlorine and bromine. The modified Sanderson method performed comparably to CNDO and Jolly electronegativity calculations in most cases. Correction for molecular potential is automatically achieved using the modified Sanderson method. When the data for all elements are considered the MS model seems to work reasonably well. Correlations for B, C, Ge, F, N, As and Se seem to be as good as most methods applied to these elements. The MS method works particularly well for Si, Sn, P and S. It is inapplicable to O, Cl and Br. For compounds of elements which exist in more than one formal oxidation state, it is necessary to obtain a separate plot of E B, versus q i for each oxidation state. Once this is done, the calculated and experimental binding energies show good correlation. The slopes and intercepts given in Table 3 may be used for this purpose. The reason for separate correlation lines in the case of multiple oxidation states may be understood when one considers the method by which Sanderson originally calculated his atomic E values. Sanderson's original E values were calculated as the ratio of the electron density of an element stripped of all its valence electrons to the electron density of the corresponding inert gas. By considering only the inert shell configuration, Sanderson's E values are valid for filled electron shell oxidation states. However, no allowance for lone electron pairs is made. Thus, charges calculated by Sanderson's method for elements bound in less than their maximum oxidation state may not be correct relative to other oxidation states of the same element. Attempts to correct for this neglect of lone pairs were made during the course of this work by assigning E values for an electron pair or by calculating E values for non-inert shell valence configurations met with failure. The low slopes of the binding energy-charge correlations for oxygen and the halogens may also be explained by the method Sanderson used to calculate his atomic E values. Sanderson assumed that all elements shared all valence electrons in bonding. This is not the case for the halogens or oxygen which retain essentially inert lone electron pairs. Since the effect of these lone pairs is not specifically dealt with, the charges calculated by Sanderson's method and the MS method are larger than they should be in the case of elements which retain nonbonded lone pairs.Throughout these studies, it has been noted that the MS method needs no correction for molecular potential. This may be explained by the manner in which group E values are calculated. Since the electronegativities of the atoms comprising a group are “sequentially” equalized from the terminal atom of a group inward to the site of the central atom, through bond inductive effects are accounted for. This procedure reduces a multiatom group to a form in which it may be treated as a single atom attached to the atom of interest. In this form, a molecule is reduced to Pauling's “nearest neighbor” picture. Thomas 19 has found that Pauling's “nearest neighbor” approximation works well for correlating ESCA data of small (e.g. methane-line) molecules without additional correction for molecular potential. Thus, the MS method has reduced a molecule to a form where the “nearest neighbor” approximation is applicable. The MS method works well for covalent molecules. As was pointed out earlier, the method fails for ionic solids primarily because of anomalies introduced by the Madelung potential. It is speculated that the MS method will work well for isolated ionic molecules in the gas-phase.
Read full abstract