contact resistance
**Contact Resistance** is the **electrical resistance at the interface between a metal and a semiconductor** — a critical parasitic that limits transistor on-current and dominates performance in sub-7nm devices where contact dimensions approach atomic scale.
**Origin of Contact Resistance**
- Metal/semiconductor interface forms a Schottky barrier if work functions differ.
- Ohmic contact: Barrier thin enough for quantum tunneling → linear I-V.
- Contact resistivity $\rho_c$ (Ω·cm²): Intrinsic material/process parameter.
- Total contact resistance: $R_c = \rho_c / A_{contact}$ where $A$ = contact area.
**Scaling Problem**
- Transistor on-resistance $R_{on}$ has target ~100Ω·μm.
- Contact area scales as $A \propto L^2$: At 5nm, $A = 25$ nm² = 25×10⁻¹⁴ cm².
- For $R_c = 10Ω·μm$: $\rho_c = R_c \times A = 10 × 25×10⁻¹⁴ = 2.5×10⁻¹⁴ Ω·cm²$ required.
- State-of-art (2024): $\rho_c \approx 5-10×10⁻⁹$ Ω·cm² — orders of magnitude from target.
- Contact resistance now dominates $R_{on}$ at sub-5nm nodes.
**Reducing Contact Resistance**
**High Doping at Interface**:
- Higher active dopant concentration → thinner Schottky barrier → more tunneling.
- Target: > 2×10²¹ cm⁻³ at metal-semiconductor interface.
- Achieved by: In-situ B-doped SiGe S/D epi + laser anneal.
**Silicide Engineering**:
- NiSi: $\rho_c = 10⁻⁸$ Ω·cm² on n⁺Si — adequate for 28nm.
- TiSi2 (C54): $\rho_c = 10⁻⁸$ Ω·cm² — good but rough morphology.
- NiPtSi: Improved thermal stability vs. pure NiSi.
**Alternative Metals**:
- TiSiN, Ti/TiN stack: Better barrier for p+ contacts.
- GeSn alloy contacts: Lower barrier on SiGe.
**Metrology**
- **CTLM (Circular Transmission Line Model)**: Wafer-level $\rho_c$ extraction.
- **Kelvin structure**: 4-point measurement eliminates spreading resistance.
Contact resistance is **the emerging performance bottleneck at sub-5nm nodes** — scaling transistor dimensions without a proportional reduction in $\rho_c$ negates the benefits of gate length reduction and has driven intensive research into novel metal/semiconductor contact schemes.