what is a sraf
Sub-resolution assist features, or SRAF, are extra shapes added to a photomask that are deliberately too small to print onto the wafer themselves, but which subtly reshape the light diffraction pattern around the real features nearby, making those real features print more accurately.
```flowchart
{
"rows": [
{ "type": "nodes", "items": [
{ "title": "Real feature printed in isolation", "sub": "light diffraction distorts its shape unpredictably", "tone": "red" }
]},
{ "type": "arrow" },
{ "type": "group", "title": "SRAF shapes added nearby on the mask", "items": [
{ "title": "Too small to print themselves", "sub": "but shift the local diffraction pattern favorably", "tone": "green" }
]},
{ "type": "arrow" },
{ "type": "nodes", "items": [
{ "title": "Real feature prints closer to its intended shape", "sub": "improved consistency across different feature spacings", "tone": "blue" }
]}
]
}
```
**SRAF work by exploiting light diffraction rather than fighting it, adding shapes specifically designed to never actually print on the wafer.** Because light diffracts as it passes through a photomask's pattern, how a given feature actually prints depends partly on what other features happen to be nearby it — SRAF are extra shapes sized specifically below the resolution threshold needed to print, added near a real feature purely to favorably influence that diffraction pattern, helping the real feature print more consistently regardless of its specific surrounding layout.
```svg
```
```svg
```
| Aspect | Real feature | SRAF |
|---|---|---|
| Prints on the wafer | Yes | No — deliberately too small |
| Purpose | The actual chip pattern | Improves the real feature's printed accuracy |
| Placement | Determined by chip design | Added algorithmically near real features |
| Removed after design? | N/A | Present only on the mask, never on the chip |
**SRAF placement is generated algorithmically as part of the same optical proximity correction process that also reshapes real feature edges.** Because determining where SRAF will actually help requires simulating how light will diffract around a specific layout, SRAF placement is calculated by the same computational lithography software used for optical proximity correction, working alongside OPC's edge adjustments as part of one broader mask-preparation process aimed at making the printed pattern match the intended design as closely as possible.
**SRAF have become especially important as chip features have shrunk closer to and below the wavelength of the light used to pattern them.** As the size of chip features has shrunk well below the wavelength of the light used in deep-UV lithography, diffraction effects have become proportionally more significant, making techniques like SRAF increasingly necessary just to achieve acceptable pattern fidelity — SRAF usage has grown substantially as chipmakers have pushed toward smaller and smaller process nodes.
**Because SRAF must never accidentally print, mask verification specifically checks that they stay safely below the printing threshold across realistic process variation.** Since an SRAF that unexpectedly prints would introduce an unwanted defect rather than help, mask verification software specifically simulates whether SRAF placement remains safely non-printing across the range of focus and exposure variation a real manufacturing process might experience — this verification step is an essential part of qualifying a mask design before it goes into production.
Read sub-resolution assist features through an invisible-helper lens: these tiny shapes never appear on the finished chip themselves, but by subtly reshaping the light pattern around the features that do matter, they quietly make those real features print closer to their intended shape — assistance that works precisely because it stays too small to be seen.