Impurity

An impurity is any atom in a crystal that is not part of the intended host lattice, whether it arrived by accident or was added on purpose. In semiconductor work the word covers two opposite stories: unwanted foreign atoms such as sodium, iron, copper and carbon that ruin a device, and deliberately introduced atoms such as phosphorus and boron that make the device possible at all. What separates the two is not chemistry but control, meaning whether the species, the amount and the location are chosen or merely suffered.

Purity in semiconductors is measured in parts per billion because the electrical effect of a foreign atom is out of all proportion to its number. A cubic centimeter of silicon holds roughly 5 × 10²² atoms, so one part per billion is still about 5 × 10¹³ foreign atoms in that cubic centimeter, and the thermally generated carriers in pure silicon at room temperature number only around 10¹⁰ per cubic centimeter. An impurity level that would be invisible to a chemist can therefore outnumber the intrinsic carriers by thousands of times and dictate the conductivity. Electronic-grade polysilicon is commonly described with purity of nine to eleven nines, meaning that fewer than one atom in a billion is anything other than silicon, a standard far beyond almost any other industrial material.

The Bell Labs materials work of the late 1930s to the early 1950s is where impurity changed from a nuisance into a design variable. In 1939 and 1940 Russell Ohl, working on silicon for radio detectors, found a sample that produced a voltage when light fell on it, and the explanation was that the crystal had frozen with different impurity content on its two sides, creating what we now call a pn junction. Jack Scaff and Henry Theuerer then connected the two behaviors of silicon to which trace elements it contained, which is the origin of the labels n-type and p-type. After the 1947 transistor, the demand reversed: William Pfann invented zone refining at Bell Labs in the early 1950s, sweeping a molten zone along a bar so that impurities concentrate at one end, and it pushed germanium to impurity levels near one part in ten billion. Gordon Teal and John Little grew single crystals with controlled additions, which is how impurities became something to be dosed rather than merely endured.

1939-1940: Ohl's silicon sample
light-sensitive barrier traced to impurity differences
Scaff and Theuerer
link n-type and p-type behavior to trace elements
1947-1948: transistor demands pure crystals
polycrystalline, dirty germanium limits devices
Bell Labs purification and growth
Zone refining (Pfann, early 1950s)
molten zone sweeps impurities to one end
Pulled single crystals (Teal and Little, 1948-50)
controlled additions while growing
Impurity becomes a design variable
remove the unwanted, dose the wanted

Contamination in a fabrication plant is a catalogue of specific elements, each with its own way of doing damage. Sodium is the classic case: a mobile positive ion in the gate oxide of a MOS transistor, it drifts under the electric field and shifts the threshold voltage, which plagued early MOS devices in the 1960s until cleaner processing and phosphorus-containing glass layers that trap sodium were adopted. Iron, copper, nickel and gold sit in silicon as deep-level traps that capture carriers and shorten minority-carrier lifetime, leaking current across junctions and degrading image sensors with bright pixels. Metals also diffuse remarkably fast, so a stray trace from a wrench, a chemical bath or a furnace tube can travel through the whole wafer during a heat cycle.

Gettering is the strategy of not only keeping contaminants out but also giving them somewhere harmless to go. Metal atoms are mobile at process temperatures, so engineers create sinks away from the active transistor layer: damage or oxygen precipitates in the bulk of the wafer, heavily doped or polysilicon layers on the back, or phosphorus-rich regions that bind metals strongly. During heating the unwanted atoms migrate to those sinks and stay there on cooling, leaving the thin device layer at the surface cleaner than the starting wafer. The same idea underlies the cleanroom itself, where air filtration, ultra-pure water and high-purity gases serve a single goal of keeping unplanned atoms off the silicon.

Doping is best understood as impurity under strict management, which is why it needs its own entries but still shares a root with contamination. A phosphorus atom and an iron atom are both foreign to silicon, but phosphorus sits on a lattice site at a chosen depth and concentration, typically between 10¹⁵ and 10²⁰ atoms per cubic centimeter, while iron appears where it is not wanted and acts as a trap. Gold was even used on purpose in some older switching diodes to shorten lifetime, a reminder that a given atom is neither good nor bad in itself. The common misconception is that purer is always better; a perfectly pure crystal is nearly an insulator at room temperature and useless for circuits, and what the industry wants is purity in the background with precise impurity in the foreground. Modern chips with feature sizes of a few nanometers have so few atoms in each transistor channel that a single misplaced dopant or one stray metal atom can shift a device, which is why contamination control remains central to yield.

Read impurity as a statement about control rather than about dirt: the same foreign atom is a defect when it appears by chance and a design feature when it is placed on purpose.

Take impurity further

Ask the copilot about this term, or have our engineers assess it against your process.