Narration script

Unit five, defects and impurities. No crystal is perfect. Missing atoms, extra atoms, foreign atoms and misaligned rows all exist. Some impurities are deliberate: dopants, at levels from about ten to the fifteen to ten to the twenty atoms per cubic centimetre. Others, especially transition metals like iron, can destroy a cell's performance at levels as low as ten to the eleven, a few parts per trillion. And oxygen from the quartz crucible, through something called the boron-oxygen defect, explains much of the industry's shift from p-type to n-type wafers.

You will learn to classify defects, distinguish shallow dopant levels from deep trap levels, estimate a carrier lifetime from a trap density, explain the roles of iron, copper, oxygen and carbon, and relate a wafer's resistivity to its doping.

Carrier lifetime, how long a light-generated electron survives before it is lost, is the material quality metric for solar cells, and it is controlled by defects. Contamination control and cleanroom practice show up in twelve of the thirty-one Tesla and SpaceX solar job postings we reviewed. They exist because of this unit.

This schematic shows the point defects in a silicon lattice. A vacancy, where an atom is missing. A phosphorus atom and a boron atom sitting on regular lattice sites, the donor and acceptor dopants. An iron atom squeezed in between the sites, an interstitial. And at the edge, a surface with dangling bonds.

The table sorts defects by dimension. Zero-dimensional point defects, like vacancies, dopants, interstitial iron and the boron-oxygen complex. One-dimensional line defects, the dislocations caused by thermal stress. Two-dimensional defects, such as grain boundaries and surfaces. And three-dimensional ones, such as oxygen precipitates and metal silicide particles.

Now the key electronic distinction. Phosphorus's fifth electron is bound by only about forty-five thousandths of an electron-volt, easily freed by room-temperature thermal energy. Its level sits just below the conduction band: a shallow donor. Boron's level sits just above the valence band: a shallow acceptor. These set conductivity. That is doping. Metals, vacancies and many complexes create deep levels near the middle of the gap. A deep level can capture an electron and then a hole, annihilating the pair. That is Shockley-Read-Hall recombination, and mid-gap levels are the most effective at it.

This diagram places these levels inside silicon's band gap. Phosphorus, just under the conduction band. Boron and gallium, just above the valence band. And interstitial iron, deep in the gap, about zero point three eight electron-volts above the valence band. That deep level is what makes iron so damaging.

A simple formula estimates the lifetime: one over the capture cross-section, times the carriers' thermal velocity, times the trap density. In words, lifetime is the average time before a carrier bumps into a trap and gets caught.

Let's see how little iron it takes, using round illustrative values. With ten to the twelve traps per cubic centimetre, which is only zero point zero two parts per billion, the lifetime is ten microseconds, and electrons diffuse only about one hundred and seventy micrometres. High-efficiency cells want lifetimes in milliseconds, which means trap densities below about ten to the ten. That is astonishing purity.

This chart puts those concentrations on one scale, spanning thirteen orders of magnitude: from five times ten to the twenty-two silicon atoms per cubic centimetre at the top, through oxygen and dopant levels, down to harmful metal levels around ten to the eleven. Some rows are order-of-magnitude estimates, as noted on the chart.

Iron, copper, nickel, chromium and titanium get in from equipment, crucibles, chemicals and handling. They diffuse fast and form deep levels or precipitates. Interstitial iron is much more harmful in p-type silicon than in n-type, which is one reason n-type wafers tolerate contamination better. Cell processes fight back with gettering, which moves metals away from the active region: phosphorus diffusion gettering pulls metals into the heavily doped layer, and hydrogen from the silicon nitride coating passivates what remains.

In Czochralski growth, the melt slowly dissolves the quartz crucible, so wafers contain oxygen around ten to the seventeen to ten to the eighteen per cubic centimetre. Oxygen is a mixed blessing. It strengthens wafers against dislocations. But it forms thermal donors at moderate temperatures, which shift resistivity. It forms precipitates and ring patterns after hot process steps, a known issue for n-type TOPCon wafers. And in boron-doped wafers, it combines with boron into the boron-oxygen defect, which caused light-induced degradation: a few percent of power lost in the first days of sunlight. The industry fixed that by doping p-type wafers with gallium instead, and more fundamentally by moving to phosphorus-doped n-type wafers, which have no boron-oxygen defect at all. Carbon, from graphite furnace parts, mainly matters through its effect on oxygen precipitation.

Dislocations are line defects: an extra half-plane of atoms, or a helical twist. They form when thermal stress during growth or cooling exceeds what the crystal can bear, and they recombine carriers, especially when decorated with metals. When a growing ingot develops dislocations, it loses structure, and the affected part is usually melted back down. That is a key yield metric in crystal growth.

Finally, wafers are specified by resistivity rather than doping concentration, because resistivity is easy to measure with a four-point probe. Roughly, one ohm centimetre corresponds to about five times ten to the fifteen phosphorus atoms for n-type, or about one and a half times ten to the sixteen boron atoms for p-type. P-type needs more dopant because holes are less mobile than electrons.

Misconceptions. Nine-nines purity does not make impurities negligible; specific metals must be far below that level. Not all impurities are bad; dopants are essential, and oxygen even helps mechanically. Defects do not only matter at the surface. And light-induced degradation is not a module defect; it is a wafer material effect, now largely engineered out.

For the job: contamination control, with metal-free tools, clean chemicals, gloves and proper wafer carriers, protects lifetime. Lifetime measurement is standard incoming wafer inspection. And the shift to n-type wafers was a direct response to the boron-oxygen problem.

Try the review questions. Next unit, thermodynamics returns, as phase diagrams and the physics of freezing.