cobalt interconnect

**Cobalt Interconnect** — using cobalt instead of copper for the narrowest local metal layers, addressing the resistivity scaling challenge where copper's advantage disappears at very small wire widths. **The Problem with Copper at Small Widths** - Bulk Cu resistivity: 1.7 μΩ·cm - But at <15nm wire width: Effective resistivity rises to 5–10+ μΩ·cm due to: - Electron scattering at grain boundaries (grains are small in narrow wires) - Surface scattering at wire sidewalls - Barrier liner (TaN/Ta) occupies 30–40% of wire cross-section **Why Cobalt?** - No barrier needed (Co doesn't diffuse into dielectric like Cu) - Better gap fill in narrow trenches (CVD cobalt flows into small features) - Resistivity comparable to Cu at very narrow widths (barrier-free advantage) - Better electromigration resistance **Current Usage** - Intel: Cobalt for M0/M1 (finest pitch local wires) since 10nm node - TSMC: Cobalt contacts (not yet for wires) - Cu remains dominant for wider intermediate and global wires **Future Metal Candidates** - **Ruthenium (Ru)**: Even shorter mean free path than Co → potentially better at <10nm width. Barrierless. Active research - **Molybdenum (Mo)**: Very low resistivity at narrow widths. Intel exploring for future nodes **The shift from copper** to alternative metals at the finest pitches is inevitable — the physics of electron scattering at nanoscale dimensions demands it.