🔥 0 ⚡ 0 XP

REPAIRING THE DAMAGE

Implantation leaves a mess

knocked-out atoms + misplaced dopant

An ion striking the lattice at speed knocks silicon atoms out of place, leaving a damaged, partly amorphous layer. Worse, a freshly implanted dopant atom usually isn't sitting in a lattice site where it can contribute a carrier. Annealing fixes both problems with heat.

Implantation places the dopant; annealing is what makes it actually work.

TWO JOBS, ONE STEP

What the anneal accomplishes

Recrystallization Heat lets damaged silicon regrow into an ordered lattice
Activation Dopant atoms move into substitutional lattice sites, where they donate or accept a carrier
The catch Heat that repairs and activates also lets dopant diffuse, undoing implantation's precision
Every anneal is a race between fixing the crystal and not smearing the profile.

THE SPIKE ANNEAL

Get in and out fast

spike: ~1 s at peak furnace: minutes to hours time → temperature ↑

A furnace anneal holds the wafer warm for minutes to hours, which activates dopant well but lets it diffuse. A spike anneal instead ramps to a high peak temperature and back down in about a second, activating dopant while barely giving it time to move.

High temperature for a very short time beats moderate temperature held a long time.

THE TRADE-OFF

Why speed matters more at small nodes

shorter gate means less room to spread

A junction that can tolerate a little diffusion at a relaxed feature size can't at a tight one: the same nanometers of sideways spread eat a much bigger fraction of a short transistor. That pressure is what pushed anneal times down from minutes to about a second.

As transistors shrink, the anneal has to get faster just to stand still.

UNIT 22 STUDY COMPLETE

♨️

Ready for the Fab Challenge?

You've covered repairing implant damage and activating dopants without letting them spread.