This study aimed to examine the recently proposed idea that the ionic contribution to atomic bonds is essential in determining the charge state of sputtered atoms. Use was made of negative secondary ion yields reported by Wilson for a large number of elements implanted in silicon and then sputter profiled by Cs bombardment. The derived normalized ion yields (or fractions) P vary by 6 orders of magnitude, but the expected exponential dependence on the electron affinity EA is evident only vaguely. Remarkably, a correlation of similar quality is observed if the data are presented as a function of the ionization potential IP. With IP being the dominant (if not sole) contributor to the electronegativity π, one is led to assume that P depends on the sum π + EA. About 72% of the "nonsaturated" ion yields are in accordance with a dependence of the form P ∝ exp[(π + EA)/ε], with ε ≃ 0.2 eV, provided the appropriate value of π is selected from the electronegativity tables of Pauling (read in eV), Mulliken or Allen. However, each of the three sources contributes only about one-third to the favorable electronegativity data. This unsatisfactory situation initiated the idea to derive the "true" electronegativity πSIMS from the measured ion yields P(π + EA), verified for 48 elements. Significant negative deviations of πSIMS from a smooth increase with increasing atomic number are evident for elements with special outer-shell electron configurations such as (n-1)dg-1ns1 or (n-1)d 10ns2np1. The results strongly support the new model of secondary ion formation and provide means for refining electronegativity data.