Gate Pitch
Gate pitch is the center-to-center distance between adjacent transistor gates, and contacted poly pitch (CPP) is that same distance measured to include the contact that connects each gate's source and drain to the metal layers above — together they're the real density number engineers use now that process node names no longer describe anything physical.
Gate pitch and CPP solve the problem that a process node name like "5nm" doesn't tell anyone how densely transistors are actually packed, which real engineering decisions require. A node name is a marketing-era label with no fixed physical meaning across foundries, but CPP is a direct, measurable distance on the actual layout — it's gate pitch plus whatever extra space the source/drain contact needs to land without shorting to the gate, and because every logic cell on a process is built from repeating rows at that same pitch, CPP directly sets how many standard cells, and therefore how much logic, fits into a given area.
Smaller CPP translates almost directly into more logic per square millimeter, which is why it's one of the figures of merit foundries actually compete on. If CPP shrinks by a given percentage, the number of transistor rows that fit into the same die area grows by roughly the inverse of that shrink, assuming everything else about the layout stays constant — that direct, nearly mechanical relationship between CPP and density is exactly why CPP, not the marketing node name, is the number process-technology papers and foundry roadmaps actually report when comparing generations.
| Metric | What it measures | Why it matters |
|---|---|---|
| Gate pitch | Center-to-center spacing of transistor gates | The base repeat unit of a standard cell row |
| Contacted poly pitch (CPP) | Gate pitch plus source/drain contact clearance | The real minimum repeat distance; sets logic density directly |
| Metal pitch | Minimum spacing of interconnect wiring | Often the true density-limiting dimension alongside CPP |
| Transistor density | Transistors per square millimeter | The combined result of CPP, metal pitch, and cell architecture |
CPP alone doesn't tell the whole density story, which is why it's always quoted alongside metal pitch rather than by itself. A process can have an aggressively small CPP but still lose overall density if its metal pitch — the minimum spacing between interconnect wires — doesn't shrink proportionally, because the wiring layers above the transistors have their own, independent limits on how tightly signals can be routed; the combination of CPP and metal pitch, not either one alone, is what ultimately determines how many usable transistors a given process can pack into a square millimeter of silicon.
Shrinking CPP has gotten harder precisely because it now runs into physical and electrical limits that didn't matter at looser pitches. As CPP shrinks, the contact between the gate and the metal above it has to shrink too, which raises its resistance and starts to work against the very performance gains that a tighter pitch was supposed to deliver — balancing CPP reduction against that resistance penalty, along with gate-to-contact leakage risk at tighter spacing, is one of the real engineering tradeoffs behind every new process generation's pitch numbers.
Read gate pitch and CPP as the density number that replaced the node name: where "7nm" or "5nm" is a foundry-specific marketing label with no fixed meaning across companies, CPP is a measurable physical distance on the actual layout, directly and almost mechanically tied to how many transistors fit into a given area — which is exactly why it's the number serious process comparisons actually use.