Crystal Lattice

A crystal lattice is the regular, repeating three-dimensional arrangement of atoms in a crystalline solid, described by a small repeating block called the unit cell that, copied by translation in all directions, fills the whole crystal. The periodicity is the whole point: because every cell looks the same, an electron moving through the solid sees a repeating pattern of attraction, and that repetition is what gives rise to the allowed and forbidden energy bands that distinguish metals, semiconductors and insulators. Without a lattice there would be no clean band theory, and without single-crystal silicon there would be no integrated circuits as they are made today.

Silicon and germanium both crystallize in the diamond cubic structure, in which each atom bonds to four neighbors at the corners of a tetrahedron. The conventional cubic cell contains eight atoms: atoms at the corners and face centers, plus four more tucked inside at a quarter of the way along the body diagonals. The cell edge, the lattice constant, is about 0.543 nanometres for silicon and about 0.566 for germanium, so neighboring silicon atoms sit only about 0.235 nanometres apart and a cubic centimetre holds on the order of 5 × 10^22 atoms. Each bond is a shared pair of valence electrons, and the open, tetrahedral geometry is why the structure is much less densely packed than a typical metal and why silicon cleaves along particular planes. Gallium arsenide uses the closely related zincblende arrangement, with gallium and arsenic alternating on the two interleaved sublattices.

The lattice was proved by X-rays before anyone could see an atom. In 1912 Max von Laue proposed that crystals might diffract X-rays, and his collaborators Walter Friedrich and Paul Knipping confirmed it with a pattern of spots from a copper sulfate crystal. In 1912-13 the father and son William Henry Bragg and William Lawrence Bragg turned this into a measuring tool: waves scattering from parallel atomic planes reinforce only when the path difference is a whole number of wavelengths, which is Bragg's law, n λ = 2 d sin θ. With it they worked out structures of simple crystals including diamond, and the Braggs shared the 1915 Nobel Prize in Physics for it. The method is still how wafer makers verify crystal orientation, and X-ray diffraction remains routine in checking the epitaxial films used in modern chips.

Real crystals are never perfect, and the imperfections are as important as the order. A vacancy is a missing atom, an interstitial is an extra atom squeezed between lattice sites, and a substitutional impurity such as phosphorus or boron replaces a silicon atom, which is how doping works. A dislocation is a line defect where an extra half-plane of atoms ends inside the crystal, and a grain boundary is where two crystals of different orientation meet. Dislocations and boundaries scatter carriers and leak current, which is why a crystal that is acceptable for a window pane is useless for a processor. Methods for growing silicon with essentially no dislocations were developed in the late 1950s, and modern ingots are among the most perfect large objects made by humans.

Single crystals matter because periodicity must be unbroken over distances far larger than a transistor. Gordon Teal and John Little at Bell Labs adapted Czochralski's pulling method around 1950 to grow single-crystal germanium, and the same approach, later applied to silicon, made reproducible devices possible. In a polycrystal, grain boundaries trap charge and cut mobility, so circuit-grade chips use one crystal, while polycrystalline silicon survives in cheaper roles such as transistor gate electrodes and many solar panels. The lattice constant also governs how different crystals can be stacked: silicon-germanium alloy layers are about 4 percent larger in lattice spacing than silicon and, grown thin enough, are squeezed to match, producing strained channels that raise carrier mobility in processors from the 2000s onward.

A common misconception is that the diagrams of balls and sticks show how atoms really look; they are a schematic, since atoms vibrate about their sites, bonds are clouds of shared electron density, and above absolute zero the lattice is always shaking with quanta called phonons that scatter moving carriers. Another is that a crystal of one element has only one structure: carbon forms diamond or graphite, and tin forms a metallic and a diamond-cubic variant, so the arrangement matters as much as the atoms.

Read the crystal lattice as the repeating stage on which electrons perform: its perfect periodicity creates the energy bands, its small flaws supply dopants and defects, and its unbroken extent across a whole wafer is what turns a lump of purified sand into the substrate of modern computing.