What sets a wire's resistance and capacitance?

Short Answer

Resistance grows with length and falls with width and thickness. Capacitance depends on how much surface faces other metal and how close it is: long, tall, closely spaced wires have the most. For routing decisions the practical rules are simple: wider wires lower resistance, more spacing lowers coupling, and fewer vias lower resistance further.

Technical Explanation

• Resistance is R = ρ·l/(w·h). Since thickness h is fixed per layer, foundries give it as sheet resistance ρ/h, and resistance becomes sheet resistance times the number of squares l/w.
• Upper layers are usually thicker, so they have lower sheet resistance. Long nets and clocks are pushed to higher layers.
• Width and thickness vary with manufacturing. Etch changes width and polishing changes thickness, so the same wire has different resistance in different RC corners.
• Sidewall capacitance Cs = ε·L·h/s grows with the length two wires run together and with metal height, and falls as spacing grows. Doubling spacing roughly halves it.
• Area capacitance to the layers above and below grows with width, so a wider wire picks up more of it even as its resistance drops. So widening is not a free improvement.
• Vias add series resistance, often more than the wire segments around them on short nets. Via count belongs in any resistance estimate.
• Temperature raises metal resistance, which is one reason hot corners are slower on long wires.

Formula Or Decision Rule

• Resistance: R = ρ · l / (w · h) = (ρ / h) · (l / w), where ρ/h is the sheet resistance in Ω per square.
• Sidewall coupling capacitance: Cs = ε · L · h / s, where L is the length the two wires run side by side and s is their spacing.
• Area capacitance: Cp = ε · L · W / d, where d is the dielectric thickness to the layer below.

What To Check

• Per-net resistance and capacitance from report_nets -physical or the extracted parasitics.
• Layer usage for long critical nets, since upper layers are less resistive.
• Nets with long parallel runs next to a single neighbour, which carry high coupling.
• Via count on critical nets.

Healthy, Suspicious & Hard-stop Results

• Healthy (illustrative): Long critical nets mostly on upper layers, with modest via counts and no long parallel runs.
• Suspicious (illustrative): A critical net with most of its length on M2 or M3, or with one neighbour for hundreds of microns.
• Hard stop: A net whose wire delay alone exceeds the stage budget, which needs a buffer or layer change rather than tuning.

What The Interviewer Is Testing

• Whether you can reason about the direction of each change: width, spacing, length, layer.
• Do you know wider wires trade resistance for capacitance?
• A strong answer shows you include vias and temperature in resistance.

Follow-up Question & Model Response

"Why do long wires get worse faster than their length suggests?"

Candidate Model Response: Resistance and capacitance both grow with length, and wire delay roughly follows their product, so it grows with the square of length. Doubling a wire's length roughly quadruples its intrinsic wire delay. So long nets are split with repeaters: two half-length segments with a buffer between them can be faster than one long wire, even after adding the buffer's own delay.