Millisecond anneal (also called ultra-fast anneal) is a thermal processing technique that heats the wafer to very high temperatures (1,000–1,400°C) for extremely short durations (0.1–10 milliseconds) using lasers or flash lamps. This activates dopants with minimal diffusion, enabling the ultra-shallow junctions needed in advanced transistors.

Why Millisecond Anneal?

• In modern transistors, source/drain junctions must be extremely shallow (a few nanometers) to prevent short-channel effects.
• Traditional rapid thermal anneal (RTA, ~1–10 seconds) activates dopants but causes significant thermal diffusion, deepening the junction beyond acceptable limits.
• Millisecond anneal achieves high dopant activation (often >90%) while keeping diffusion to sub-nanometer levels — the wafer simply isn't hot long enough for atoms to move far.

Methods

• Flash Lamp Anneal (FLA): Uses an array of xenon flash lamps to illuminate the entire wafer surface for 0.5–20 ms. The wafer surface heats rapidly while the bulk remains cooler, creating a steep thermal gradient.
• Laser Spike Anneal (LSA): A focused laser beam scans across the wafer, heating a narrow stripe for 0.2–1 ms. The beam dwells briefly on each spot before moving on.
• Pulsed Laser Anneal: Uses pulsed excimer or solid-state lasers for even shorter exposures (microseconds to nanoseconds). Can achieve surface melting and rapid recrystallization.

Temperature-Time Tradeoff

• Conventional RTA: ~1,000°C for 1–10 seconds → good activation, significant diffusion.
• Spike Anneal: ~1,050°C for ~50 ms → better control, moderate diffusion.
• Millisecond Anneal: ~1,200–1,400°C for 0.1–10 ms → excellent activation, minimal diffusion.
• Sub-Millisecond: ~1,300°C+ for microseconds → near-zero diffusion, possible surface melting.

Challenges

• Temperature Non-Uniformity: At these timescales, achieving uniform temperature across the wafer is difficult. Pattern density variations cause local heating differences.
• Thermal Stress: Extreme temperature gradients between the hot surface and cool bulk can cause wafer warpage or even cracking.
• Metrology: Measuring temperature accurately during millisecond-scale heating is extremely challenging.
• Integration: Process windows are very tight — small variations in energy or dwell time significantly affect results.

Millisecond anneal is essential for nodes below 14nm — without it, achieving the abrupt, shallow junctions needed for high-performance FinFET and gate-all-around transistors would be impossible.