the ionic model
The ionic model is the simplest useful cartoon of a ceramic crystal: imagine it built from hard, charged marbles. Every atom has given up or grabbed electrons to become an ion, a small positive marble (the cation) or a larger negative one (the anion), and the crystal is just these marbles stacked so that plus and minus touch, held together by ordinary electric attraction and kept from collapsing by the marbles being incompressible. No shared electrons, no directions, just charge and size.
Two ideas do all the work. First, opposite charges attract and like charges repel, so the ions arrange to put as many anions as possible around each cation, and the reverse, while keeping like ions apart. Second, each ion has a fixed radius, so geometry decides how many neighbours actually fit. Put these together and the model predicts the whole architecture: in MgO each Mg2+ sits touching six O2- neighbours in an octahedron, giving the rock-salt structure, and you can estimate the crystal's binding energy just from the charges and spacings.
The ionic model is astonishingly good for its simplicity: it explains why oxides are hard, high-melting, transparent electrical insulators, and it underpins Pauling's rules and the radius-ratio idea. But it is a deliberate fiction. No bond is 100 percent ionic; even ionic MgO has some electron sharing, and for silicates and nitrides, where covalency is large, the pure charged-sphere picture starts to fail and you must add directional bonding to get the structures right.
Rock salt through the model's eyes: take Na+ (radius about 0.10 nm) and Cl- (about 0.18 nm) as touching spheres of charge +1 and -1, stack them so each is surrounded by six of the other, and you have reproduced table salt's cubic crystal, its cleavage, and roughly its melting point, all without a single shared electron.
A whole crystal rebuilt from charged, incompressible spheres.
Treating ions as spheres with one fixed radius is an approximation twice over: real ions are slightly squishable (polarizable) and their size changes with charge and coordination. The model is a starting point, not the final truth.