Process Examples
Diffusion
- Example (1D) Script for 1D diffusion of an implant profile.
Grid Testing
Implant boron at 10keV and 1e15/cm2 and anneal for 1 minute at 1100C. Use a uniform grid spacing down to 0.5um or 0.05, 0.02, 0.01, 0.005, and 0.001um.
How does the profile vary (think about why), and how does the total diffusion time (cpu time) vary?
Etch, Deposit 2D Grid
Start with a 2D silicon slab 1um wide and 1um depp. Deposit and oxide layer 0.2um thick. Etch an opening from 0.4um to 0.6um in the top of the structure. What is the maximum energy of a boron implant (1.0e15/cm2) that can be blocked ? (meaning a surface doping of boron < 1.0e16/cm3)
Model Variation
Use a 0.01um grid down to 0.5um, and then increase the grid spacing to 10um at 100um deep. Use a 0.01um oxide cap on the top of the silicon. Implant phosphorus at 40keV and 1.0e14/cm2. Diffuse with the Fermi model and plot. Then set the interstitials (Int) equal to the phosphorus. And run the react model. Do both anneals for 1min at 900C.
Be sure to turn on both Int and Vac solutions using: pdbSetSwitch Si I Diffmodel Numeric pdbSetSwitch Si V Diffmodel Numeric solution name = Int solve !damp !negative solution name =Vac solve !damp !negative
Also use: Math diffuse dim=1 sblock scale bcgs row
Save the result from the react model –we’ll use it again the in the next exercise.
Dt Extraction
We often use flops to do data analysis – how much diffusion enhancement was there? Use the result from the react model as a “measured” result. Implant the phosphorus again and diffuse it using the Fermi model. How long is the diffusion required to match the “measured” profile?
Model Development
Build a simple model for carbon diffusion in silicon. Use a constant diffusivity: Dc = 1.0 * exp(-3.5 / kT)
Set the initial carbon profile to: Cc = 1.0e18 * exp( (x-0.1)*(x-0.1) / 0.001 ) + 1.0e15
Diffuse the profile for 60min. at 900, 30min. at 1000, and 10min. at 1100C. Does it agree with what you expect from a simple estimated hand calculation? Modify the model so the diffusivity it proportional to carbon concentration / 1e16 . Run the same anneals – what happens qualitatively?
See the Solution at Model Development Solution