BSIM4'5 Scalable Rsub Model Evaluation

1
BSIM4.5 Scalable Rsub Model Evaluation

• Tested the following options with data from nine
  RF-NMOS transistors of various gate widths, gate
  lengths and number of fingers. Data was y22 taken
  up to 20GHz on a 65nm CMOS process.
• RBODYMOD2,
• 1-R network Good alignment to the data
• 3-R network Slightly Improved alignment to the
  data
• 5-R network Did not see significant improvement
  on the fitting. Too many parameters. Have a
  feeling that we wont be able t take advantage of
  all these parameters
• Tested the scalability for the 1-R network
• Aligned model to data from transistors with 3
  different L, or W or NF, then interpolated the
  model to a 4th L, or W or NF that is within the
  fitted W, L and NF range.
• Model aligned well to the data for the 4th L or W
  or NF, however we do not understand the physical
  significance to the RBPB scaling parameters
• RBPBYL"-0.28"
• RBPBYW2.0"
• RBPBYNF"1.89"
• Need additional time to further evaluate the
  scalability, especially for the 3-R and 5-R
  network