ionic model
Take a crystal of ordinary table salt and ask the simplest possible question: why does it hold together? The ionic model gives the simplest possible answer. It pictures the solid as nothing more than a regular three-dimensional array of charged balls — positive ions (cations, like Na+) and negative ions (anions, like Cl-) — held together because opposite charges attract. There are no shared electrons, no directional bonds; just charged spheres stacked so that every plus is surrounded by minuses and every minus by pluses.
More precisely, the model makes two assumptions. First, one atom has handed its outer electrons completely over to another, so that each is a hard sphere of fixed radius carrying a whole-number charge (Na becomes Na+ by losing one electron, Cl becomes Cl- by gaining one). Second, the only force that matters is the electrostatic (Coulomb) attraction and repulsion between these point charges. From just these two ideas you can predict how the ions pack, how strongly the crystal is bound (its lattice energy), and even its melting point — all without any quantum mechanics at all.
The ionic model is a model, and a deliberately extreme one: in reality no bond is ever 100 percent ionic. Electrons are always shared to some degree, so even NaCl has a little covalent character, and small, highly charged cations like Al3+ distort (polarize) nearby anions so much that the ionic picture starts to fail. The wonder is that for compounds of the most electropositive metals with the most electronegative nonmetals — the alkali and alkaline-earth halides and oxides — the simple picture of charged billiard balls works astonishingly well, giving lattice energies that match experiment to within a few percent.
The ionic model predicts the lattice energy of NaCl, treating Na+ and Cl- as point charges of +1 and -1, comes out near 770 kilojoules per mole — and the value measured through a Born-Haber cycle is about 787 kJ/mol. A two-assumption model landing within a few percent of reality is why it remains a cornerstone of inorganic chemistry.
A two-assumption electrostatic model reproduces a measured lattice energy to within a few percent for NaCl.
No bond is purely ionic. The model works best for the most electropositive-plus-electronegative pairs; for small, highly charged cations (Fajans' rules) covalent character grows and the picture must be corrected.