Contact resistance concerns current crossing a connection, especially the interface and its access region. Interconnect resistance concerns current traveling along a conductor. A measured path can contain both, so a change in total resistance does not identify which part changed.
Draw the path before naming the resistance
Consider current leaving a transistor source/drain region, entering a contact, and then traveling along a metal line. It encounters semiconductor access resistance, a semiconductor-to-contact interface, the conducting contact body, and later interconnect segments. The exact grouping depends on the measurement and model.
That grouping should be stated explicitly. In some discussions, “contact resistance” means the interface contribution alone; in others, an extracted contact term includes spreading or access effects. Comparing two quoted values without matching definitions can produce a false process conclusion.
Process checkpoint
Understand CT Tungsten CMP in context
Inspect the completed contact level before the process moves into copper wiring.
Process context for “Contact vs Interconnect Resistance: Follow the Current”: 28nm Planar Flow · CONTACT · Step 187
Geometry has different roles at an interface and in a line
For a uniform conductor under simple assumptions, a longer path raises resistance and a larger conducting cross-section lowers it. An interface requires a different description: carrier transmission, interface condition and effective current-transfer area matter. Current can also crowd into part of a contact, so nominal area is not always fully utilized.
| Part of the path | Main conceptual question | Potential source of confusion |
|---|---|---|
| Semiconductor access | How does current reach the connection? | Assigning all access loss to the interface |
| Contact interface | How do carriers cross the boundary? | Treating it as just another length of metal |
| Contact body | Is the conducting path continuous? | Ignoring fill geometry or constriction |
| Metal line | How does current travel along the wiring? | Inferring interface quality from line resistance |
Why changing the metal may not fix the bottleneck
Imagine that the line contribution decreases while the contact interface remains the dominant part of the total path. The overall improvement will be limited by the contribution that did not change. This follows from adding series contributions; it does not require assuming any particular contact material.
The opposite mistake is also possible. A high measured total resistance may arise from a narrowed conductor or an incomplete fill path rather than poor carrier transmission at the semiconductor interface. Both situations motivate looking at the relevant structure, but they lead to different questions.
Compare observations with matched definitions
A useful learning exercise is to vary the length of a conceptual line while holding the same terminal contacts. If resistance changes with line length, that behavior points to a distributed conductor contribution. The remaining term still needs careful interpretation; it can include more than a pure interface resistance.
This is the reasoning behind separating geometry-dependent and connection-related contributions in suitable test structures. It is not a substitute for a calibrated extraction method, and a process SVG cannot provide an electrical resistance measurement.
Self-alignment belongs to another axis of the problem. A contact may tolerate a particular overlap without shorting yet still have an unfavorable electrical connection. Geometric protection and carrier transport must both work; success in one does not certify the other.
Source links
References
Modern Semiconductor Devices for Integrated Circuits - MOS Transistor
Chenming Hu
Modern Semiconductor Devices for Integrated Circuits · Ch6 MOS Transistor