Ruthenium Process

The ruthenium process deposits Ru as an interconnect liner, seed, or — at the most advanced nodes — the conductor itself, for a reason that has nothing to do with ruthenium being a better bulk conductor than copper. It isn't; copper's bulk resistivity is lower. Ruthenium wins at specific dimensions because of what happens to resistivity as a wire gets narrower than the distance an electron can travel between collisions, and copper and ruthenium respond to that narrowing in opposite ways.

ALD/CVD Ru precursor
Conformal, thin-film compatible
Liner or direct metal
Depends on node and line width
Where Ru gets useddriven by the resistivity size effect
Ru liner/seed
Replaces separate Ta barrier + Cu seed
Ru as the conductor
Below the Cu/Ru resistivity crossover width
Narrowest metal lines
M0/M1 local interconnect at leading nodes
MIM capacitor electrode
Separate high-k compatible use

Bulk resistivity is the wrong number to compare once a wire gets narrower than the electron's mean free path, and that's exactly the regime advanced interconnect now lives in. Every metal's resistivity, measured in a large sample, assumes electrons travel some average distance — the mean free path — before scattering off something and losing momentum. Copper's mean free path is relatively long, around 40 nm; once a copper line's width or thickness shrinks below that distance, electrons start scattering off the wire's own surfaces and grain boundaries far more often than bulk resistivity would predict, and copper's effective resistivity rises sharply — the "resistivity size effect." Ruthenium's mean free path is far shorter, roughly 6–7 nm, so a ruthenium wire is already deep into its own size-effect regime at dimensions where copper is still behaving close to its bulk value — but that also means ruthenium's resistivity doesn't have nearly as much room left to rise as the line narrows further.

The Cu/Ru Resistivity Crossover copper wins at wide lines; ruthenium wins once width drops below its own, much shorter mean free path line width (decreasing →) effective resistivity Cu — long mean free path (~40nm) steep rise below ~15-20nm Ru — short mean free path (~6-7nm) stays gently sloped crossover width below this width, Ru beats Cu neither metal's bulk resistivity changed — the crossover is purely about which metal's mean free path the wire has shrunk past

That crossover point sits roughly in the range of the narrowest local-interconnect and contact-level lines at today's most advanced nodes, which is exactly why ruthenium shows up there first rather than throughout the whole BEOL stack. Wide, upper-level metal lines stay well above the crossover width and keep using copper, where it remains the lower-resistance choice; the thinnest, lowest-level lines — M0 and local interconnect at leading-edge nodes — are narrow enough that ruthenium's gentler size-effect curve wins outright. That's a genuinely new kind of material decision for interconnect: not "which metal has the best properties," but "which metal's intrinsic length scale best matches the specific wire width being built."

Ruthenium's second role, as a liner and seed layer rather than the bulk conductor, solves a separate but related scaling problem: the barrier itself is eating too much of a shrinking wire's cross-section. A conventional copper line needs a Ta/TaN diffusion barrier plus a separate PVD copper seed before electroplating can fill it — two distinct films, each occupying volume that could otherwise carry current, the same volume squeeze covered in the tungsten and cobalt processes. Ruthenium can serve as both an adhesion/barrier-like layer and a direct electroplating seed in one film, deposited conformally by ALD or CVD, which lets a narrow via or line dedicate more of its limited cross-section to actual conductor instead of stacked liner materials.

Outside interconnect entirely, ruthenium also sees use as a metal electrode in metal-insulator-metal (MIM) capacitors, a role that depends on an unrelated property — its compatibility and stability against high-k dielectrics at the deposition and anneal temperatures those capacitors require. That use has nothing to do with the resistivity size effect driving ruthenium's interconnect adoption; it's a separate reason the same element keeps appearing on advanced-node process flowcharts, the same pattern seen with cobalt's three unrelated roles elsewhere in the fab.

Read the ruthenium process as a sign that interconnect scaling has moved past "pick the best bulk conductor" into a regime where the physically relevant length scale — a specific metal's electron mean free path — decides the right material for a specific wire width. Copper's long mean free path made it the obvious winner for decades of progressively narrower lines; ruthenium's short one is exactly what makes it the next winner, right at the dimensions where copper's own advantage finally runs out.

Take ruthenium process further

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