The differential inelastic energy loss $Q$ has been determined for keV atomic collisions between ${\mathrm{Na}}^{+}$, ${\mathrm{Ne}}^{+}$, ${\mathrm{Ne}}^{++}$, ${\mathrm{F}}^{+}$, ${\mathrm{O}}^{+}$, and ${\mathrm{N}}^{+}$ projectiles and Ne, ${\mathrm{N}}_{2}$, and N${\mathrm{H}}_{3}$ targets for scattering angles between 2\ifmmode^\circ\else\textdegree\fi{} and 15\ifmmode^\circ\else\textdegree\fi{}. The data, which have been obtained under single-collision conditions, determine the probability of producing a $K$ vacancy and the corresponding $K$-excitation energy for specified charge states of the projectile before and after the collision. The production of a $K$ vacancy has been shown to take place preferentially in the low-$Z$ collision partner with a probability depending on the collision's distance of closest approach and the projectile velocity. The data suggest that rotational coupling between the $2p\ensuremath{\sigma}$ molecular orbital (MO) and the $2p\ensuremath{\pi}$ MO produces the $K$ vacancy. The $K$-excitation probability is shown to depend strongly on the outer-shell ($2p$), precollisional configuration of the high-$Z$ collision partner. For ${\mathrm{N}}^{+}$-N${\mathrm{H}}_{3}$, ${\mathrm{Ne}}^{+}$-Ne, ${\mathrm{Ne}}^{++}$-Ne, and ${\mathrm{Na}}^{+}$-Ne collisions, the maximum $K$-excitation probabilities are 35%, (10-15)%, 20%, and \ensuremath{\le} 1%, respectively. These observations are in accord with the Fano-Lichten model. For ${\mathrm{N}}^{+}$ and ${\mathrm{O}}^{+}$ projectiles incident on ${\mathrm{N}}_{2}$ and N${\mathrm{H}}_{3}$ targets, triple-peaked $Q$ spectra occur, indicating a fair probability of the production of two $K$ vacancies. For asymmetric collision systems such as ${\mathrm{O}}^{+}$-N${\mathrm{H}}_{3}$, the two $K$ vacancies are produced preferentially in the low-$Z$ collision partner.