Nickel Process

The nickel process forms nickel monosilicide (NiSi) contacts — nickel's successor to the cobalt silicide covered elsewhere — and it inherits a problem genuinely different from anything TiSi2 or CoSi2 ever had. Those earlier silicides needed help *reaching* their best low-resistance phase; nickel silicide's best phase forms easily and then has to be stopped from *continuing past it* into a worse one.

Ni (or NiPt alloy) sputter
Thin film over exposed Si
Low-temperature RTA
~300-500°C — far below CoSi2's anneal
Phase control is the whole challengeNiSi is good; NiSi2 is not
NiSi forms first
Low resistance, desired phase
Risk: further anneal → NiSi2
Higher resistance, rough, can pipe through junction
Selective wet strip
Removes unreacted Ni, not NiSi
Stop before NiSi2 forms
Tight thermal budget control

Nickel silicide's main advantage over cobalt silicide is that it forms at a meaningfully lower temperature and consumes less silicon doing it, which matters enormously once junctions get too shallow to spare much silicon at all. NiSi forms in the 300–500°C range, well below CoSi2's required anneal temperature, and consumes less of the underlying silicon per unit of silicide thickness than cobalt silicide does. Both properties matter directly for ultra-shallow source/drain junctions: a lower thermal budget reduces unwanted dopant diffusion elsewhere in the device during the silicide anneal, and consuming less silicon leaves more of a shallow junction intact rather than silicided straight through it — exactly the margin advanced nodes with increasingly shallow junctions needed.

The problem nickel introduces instead is that NiSi is not the most thermodynamically stable phase available, and pushing the anneal too far converts it into something worse. Nickel and silicon can form several different compounds depending on temperature and nickel-to-silicon ratio; NiSi — the phase that forms first and gives the lowest resistance — is a metastable intermediate, not the end state the system wants to settle into. Push the thermal budget higher, whether from the silicide anneal itself or from later high-temperature steps in the flow, and NiSi can transform further into nickel disilicide (NiSi2): a phase with higher resistivity, a rougher, less uniform interface, and a tendency to grow unevenly downward into the silicon in narrow, finger-like intrusions called "piping" — intrusions that can reach deep enough to short through an ultra-shallow junction entirely.

NiSi's Opposite Problem: Stopping Before the Next Phase unlike TiSi2, the good phase forms easily — the risk is letting it keep going NiSi — anneal stopped correctly shallow junction (intact) NiSi — smooth, low-R uniform interface, junction undisturbed NiSi2 — over-annealed shallow junction NiSi2 — rough, higher-R piping intrudes toward/through the junction the process target is a thermal budget ceiling, not a floor — the opposite emphasis from silicides that struggle to nucleate their good phase at all

Alloying the nickel film with a small platinum fraction before deposition is the standard fix, and it works by making the desired phase more stable rather than by changing the anneal recipe. A NiPt alloy — typically a few percent platinum by composition — raises the temperature at which NiSi starts transforming into NiSi2, widening the thermal process window between "silicide has fully formed" and "silicide has started degrading." Platinum also measurably suppresses agglomeration, a related failure mode where the silicide film breaks up into isolated islands rather than staying continuous, which independently raises contact resistance even without piping. The platinum addition doesn't eliminate the phase-transition risk — it buys margin against it, the same way copper addition buys margin against aluminum electromigration without eliminating the underlying mechanism.

This metastable-phase story is a structurally different engineering problem from every other silicide covered in this glossary, and it's worth noticing the pattern across all of them. TiSi2 needed nucleation help to reach its best phase (C49 to C54) in narrow lines; CoSi2 largely avoided that narrow-line sensitivity by not having an equivalently fragile phase transition in the first place; NiSi flips the challenge entirely — reaching the good phase is easy, and the engineering effort goes into stopping the reaction before it proceeds to a worse one.

Read the nickel process as the silicide story's mirror image: where titanium's and cobalt's silicide processes are built around helping a reaction happen, nickel's is built around stopping one — NiSi is the lowest-resistance phase available at the lowest thermal budget in the silicide family, but it's a metastable waypoint rather than a destination, and platinum alloying exists specifically to widen the margin between arriving at that waypoint and overshooting it into a silicide that's worse than the one it replaced.

Take nickel process further

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