High-Resolution Si Etch Process Using Oxide Hard Mask

Obligatory Disclaimer

Process outlines are just that; they are not guaranteed to work perfectly for your specific application and aren’t intended to provide anything more than a useful starting point for engineering your own process. The authors and the MNC in general make no guarantee of anything useful happening if you blindly follow these steps.

Description

All subtractive (etch-based) e-beam lithography processes have to contend with the fact that the most common and easily-used positive e-beam resist (PMMA) is a notoriously bad etch mask. Resists with better etch resistance than PMMA exist (ZEP520, for example) but can be prohibitively expensive. This document outlines a process for etching e-beam-scale structures into a Si substrate by using PMMA to pattern a thin SiOx? hard mask, then using the hard mask to etch the silicon.

Tools Used

• Varian evaporator (optional)
• AJA-2 sputtering system (optional)
• Bay 4 spinner
• Vistec EBPG5000 electron beam lithography tool
• STS plasma etcher
• Oxford ICP etcher
• Wet benches etc.

Process Flow

1. Deposit ~45 nm of SiO?₂/SiO_x on a silicon sample. This can be done in a number of ways, including evaporation, sputtering, and (for best results) thinning a ‘polymonitor’ wafer coated in 100 nm of thermal oxide in 10:1 BOE for 70 sec.
   
   
   1. Denser oxides make better etch masks, so thermal oxide is the way to go here, although sputtered/evaporated SiO?_x can be used if you anneal it in an oxygen-rich environment (the RTA can be used for this). Avoid PECVD low-temperature oxide, as its low density makes it a poor etch mask.
      
      
2. Spin ~100 nm of PMMA resist on your sample. Bake the PMMA at 180C for 5 minutes on a hot plate.
   
   
   1. The A2 concentration of PMMA spun at 2krpm will get you this thickness, although other concentration/spin speed combinations will work.
      
      
   2. It isn’t strictly necessary, but it’s a good idea to thoroughly solvent-clean your wafer (acetone, methanol, isopropanol) and do a dehydration bake at 180C for 1-2 mins prior to spinning. This will minimize film irregularities and maximize adhesion.
      
      
3. Pattern your sample with the Vistec. Critical doses are very geometry-dependent, but for most structures 800 μC/cm2 is a reasonable ballpark dose.
   
   
4. Develop the PMMA in 3:1 IPA:MIBK for 30 sec. Rinse with IPA and dry.
   
   
   1. For improved resolution, cold development (down to -15C) can be used, although this will increase the critical dose correspondingly
      
      
5. Etch the hard mask in the STS using the “pjsoxide” recipe for 120 sec.
   
   
   1. Prior to etching, condition the empty chamber by running the “o2clean” recipe for 5 min., then the “pjsoxide” recipe for 5 min.
      
      
   2. Full recipe parameters are: Pressure = 75mT; Power = 150W; CHF3 = 50sccm; CF4 = 25 sccm; Ar = 50 sccm.
      
      
6. Etch the underlying Si in the Oxford ICP etcher, using the “Si Etch Mo Li – 14” recipe.
   
   
   1. The etch rate here is ~150 nm/min; calculate your etch time accordingly
      
      
   2. Condition the empty chamber with a 5 min. “Si Etch Mo Li – 14” run prior to etching your sample
      
      
   3. Full recipe parameters: Pressure = 8 mT; Fwd. Power = 50W; ICP Power = 400W; Cl2 = 50 sccm; Chiller = 25C
      
      
7. Strip any remaining PMMA by immersing the sample in acetone for ~5 min, then rinse with methanol and isopropanol and dry.
   
   
8. Strip the hard mask using 10:1 BOE for 1 min. Make sure you’ve stripped away all the PMMA before this step!