why 300mm wafers
Modern fabs process silicon in the shape of 300-millimeter round wafers, not because that size is fundamentally special, but because it is the point where a bigger circle stopped paying for itself: a wafer's usable area grows with the square of its diameter, so a larger wafer squeezes far more chips out of every expensive process cycle, right up until handling and equipment costs start climbing just as fast as the yield gains.
```flowchart
{
"rows": [
{ "type": "nodes", "items": [
{ "title": "Bigger wafer diameter", "sub": "150mm → 200mm → 300mm over decades", "tone": "neutral" }
]},
{ "type": "arrow" },
{ "type": "group", "title": "The economics that drove each jump", "items": [
{ "title": "Area grows with the square", "sub": "far more die per wafer, same process cost", "tone": "green" },
{ "title": "Fixed cost per wafer pass", "sub": "one litho, one etch step covers the whole wafer", "tone": "green" }
]},
{ "type": "arrow" },
{ "type": "nodes", "items": [
{ "title": "Handling and flatness get harder", "sub": "bigger wafers sag, warp, and cost more to move", "tone": "orange" },
{ "title": "New size wins only if savings beat new costs", "sub": "why the industry hasn't moved to 450mm", "tone": "orange" }
]}
]
}
```
**The core economic driver is that die count scales with area, not diameter.** Going from a 200-millimeter to a 300-millimeter wafer more than doubles the diameter's ratio, but because area scales with the diameter squared, it roughly delivers over twice the usable surface — meaning more than twice as many chips come out of every single pass through an expensive lithography scanner, etch chamber, or deposition tool. Since most of a fab's cost is per-wafer-pass rather than per-chip, packing more die onto each wafer directly lowers the cost of every individual chip.
```svg
```
**Every jump in wafer size has followed the same pattern: it only happens once the economics clearly favor it.** The industry moved from 150 to 200 millimeters in the 1990s and from 200 to 300 millimeters in the early 2000s, and each transition required retooling an entire generation of fab equipment — new wafer handling robots, new chucks, new furnaces — all sized for the new diameter. A proposed jump to 450 millimeters was explored for years but ultimately stalled, because the wafer handling, flatness control, and equipment redesign costs at that size grew large enough to erase much of the expected per-chip savings.
**Bigger wafers also fight against their own added handling difficulty as they scale up.** A larger, thinner disc of silicon sags and warps more under its own weight and under the heat of process steps, so equipment has to hold it flatter and support it more carefully to keep every point on the wafer in focus for lithography and uniform for etch and deposition. Every increase in diameter has to out-earn this growing handling burden, which is exactly the calculation that made 450 millimeters a jump too far, at least so far.
| Wafer size | Approximate usable area | Era of adoption |
|---|---|---|
| 150mm | ~17,700 mm² | 1980s-1990s |
| 200mm | ~31,400 mm² | 1990s-2000s, still common for mature nodes |
| 300mm | ~70,700 mm² | Early 2000s onward, today's standard |
| 450mm | ~159,000 mm² (proposed) | Explored, stalled on handling and retooling cost |
**Wafer size also interacts directly with lithography, since a scanner's optics have to hold focus and overlay accuracy across the entire usable field.** A larger wafer does not simply mean scaling up the same lithography step — it means the scanner's stage, focus system, and metrology all have to maintain nanometer-level precision across a physically bigger area, which is part of why every wafer-size transition has required a parallel generation of new lithography and inspection tools rather than just bigger chucks.
Read wafer size through a per-pass-economics lens: a fab does not care how large a single wafer is for its own sake, it cares how many finished chips come out of every expensive lithography and etch cycle — and 300 millimeters is simply the diameter where that math has worked out best against the cost of building and running equipment able to handle it.