Skip to main content
TOMPW Service logo
TOMPW
← Back to Knowledge Hub

Case Study: 12.24 × 15.24 mm Die on 200 mm Places 127 Dies — What the Map Says

A real 200 mm wafer map for a 12.24 × 15.24 mm die with a 240 µm scribe lane places 127 gross dies against a closed-form estimate of 136. Why notch clearance and reticle placement account for the difference, and how a multi-product dicing plan changes the picture.

September 26, 2026Updated September 26, 20267 min readBy Henry
Case StudyMPWYieldTape-OutPackaging

This is a real wafer map, not a textbook example. The layout file describes a 200 mm wafer carrying a product with a 12.24 mm × 15.24 mm die and a 240 µm scribe lane. The map places 127 gross dies per wafer.

The value of a real map is that it is a calibration point. Formulas for dies per wafer are approximations, and this one shows exactly how far off they can be on a large die.

The layout parameters

The map file records the following for the product, identified by the code 0EXH:

  • Wafer diameter 200 mm, edge exclusion 3.5 mm, notch clearance 6 mm at the 6 o'clock position
  • Die size 12.24 mm × 15.24 mm (186.5 mm²)
  • Shot size 24.48 mm × 30.48 mm with a 2 × 2 reticle matrix
  • Scribe (street) width 240 µm in both directions
  • Gross dies per wafer: 127

What the standard estimate gives

The common closed-form estimate — effective wafer area divided by die pitch area, minus a perimeter correction — returns 136 for exactly these inputs. The real map places 127. That is about 7% fewer.

The gap has two named sources. The formula does not model notch clearance: 6 mm of unusable material at one point on the wafer edge removes dies that the circular effective-area calculation still counts. And it does not model reticle placement: a 2 × 2 reticle with a matrix shift places dies on a lattice whose interaction with the wafer circle is not the same as the ideal lattice the formula assumes.

On a small die the error is invisible in practice — a few dies out of several thousand. On a die this size, where each die is 186.5 mm² and represents a meaningful share of wafer cost, 7% is a real number in a quotation.

The dicing plan made it harder, not easier

The delivery plan for this wafer was not a single product. It called for seven different chip designs, fifty dies of each, cut from the wafer as blocks with the package shells handled separately. That kind of multi-product split changes the arithmetic again: the constraint is no longer how many dies fit, but whether the specific dies each product needs are in the usable region at all.

It also changes the conversation with the assembly line. Wafer grinding to a 12 mil target, dicing, and then holding package shells separate from dies are all normal requests, but each one is a parameter the assembly house needs in writing before the wafer arrives.

What to take from this

Treat a closed-form dies-per-wafer figure as an upper bound for large dies. For a 12 mm-class die on 200 mm, expect the real number to land 5–10% below the formula. If the number feeds a cost model or a quotation, either apply that margin or get the actual map — and the actual map is available earlier than most teams ask for it, because the foundry generates it at layout release.

Outcome

The dicing, grinding and packaging plan was executed against the real map rather than an estimate. The seven chip designs were cut, ground to the 12 mil target and supplied with their package shells as specified, and the customer's die-count and package requirements were met.

Article FAQ

Quick answers tied to this topic

Scribe width is set by the process and the dicing method, and it is wider on some mature nodes and for some dicing approaches. It matters directly: scribe width adds to the die pitch, so a 240 µm lane on a 12 mm die costs roughly 2% of the dies per wafer compared with a 100 µm lane. It is worth confirming with the fab rather than assuming a default.

It is one of the inputs, and on a 200 mm wafer it removes a ring about 3.5 mm wide from the usable area. For a large die, a ring of that width cannot host a complete die row, so its effect is larger than its area suggests. Notch clearance is a separate subtraction on top of it, and that is the one closed-form estimates usually omit entirely.

It is a realistic one. Dividing usable wafer area by die area gives around 157 as a theoretical ceiling with no perimeter loss at all; the map's 127 reflects the perimeter correction, the notch and the reticle lattice together. If a quotation assumes a number closer to the ceiling, the cost per die will be optimistic by a similar margin.

Related Tools

Plan this with a free calculator

Related Reading

Where this topic leads next

One deeper read on the same stage of the program, and one on the next decision you will hit after it.