EUV was never going to be single patterning

[Fred Chen]

The focus for minimal LER is different for different pitches and patterns. A big contribution is from "fading" which originates from the light propagating in the EUV mask being different for different angles.

 

[hskuo]

The focus for minimal LER is different for different pitches and patterns. A big contribution is from "fading" which originates from the light propagating in the EUV mask being different for different angles.

If you can not measure it correctly, then you can not control/adjust the process properly. Currently, CD/LER/LWR and LCDU are reported from CDSEM(rarely from AFM, OCD will be different approach). Result from thin resist(<20nm) is impacted by CDSEM image quality severely. CDSEM noises might contribute more than 100% on true data from the wafer(process). It used to be nuisance in mature nodes like 14nm and above, but it becomes critical in advanced nodes. For details, you might refer to the paper reported by imec in SPIE conference: https://www.spiedigitallibrary.org/...sist-for-high-NA-EUVL/10.1117/12.2614046.full

 

[Fred Chen]

If you can not measure it correctly, then you can not control/adjust the process properly. Currently, CD/LER/LWR and LCDU are reported from CDSEM(rarely from AFM, OCD will be different approach). Result from thin resist(<20nm) is impacted by CDSEM image quality severely. CDSEM noises might contribute more than 100% on true data from the wafer(process). It used to be nuisance in mature nodes like 14nm and above, but it becomes critical in advanced nodes. For details, you might refer to the paper reported by imec in SPIE conference: https://www.spiedigitallibrary.org/...sist-for-high-NA-EUVL/10.1117/12.2614046.full

Fractilia FAME 300 relies on SEM images, is there already noise control applied?

 

[Tanj]

It used to be nuisance in mature nodes like 14nm and above, but it becomes critical in advanced nodes. For details, you might refer to the paper reported by imec in SPIE conference:

There is also this one: https://doi.org/10.1117/1.JMM.21.2.021207

It is impractical to use AFM for metrology in production, but for research work it looks increasingly necessary to use AFM to get accurate results.

 

[hskuo]

Fractilia FAME 300 relies on SEM images, is there already noise control applied?

They analyze and subtract SEM noises and provide on-wafer results which should be independent to SEM recipe setting and SEM tools which do measurement.

 

[hskuo]

There is also this one: https://doi.org/10.1117/1.JMM.21.2.021207

It is impractical to use AFM for metrology in production, but for research work it looks increasingly necessary to use AFM to get accurate results.

AFM could be complementary to CDSEM and OCD. There is hybrid metrology methodology to be used in advanced nodes been discussed in the past 5-10 years. In the thin resist CDSEM challenge study, AFM was used to validate PR thickness and profile if applicable.