Y. Chauhan, F. Krummenacher and A. M. Ionescu

1
Progress and update in HV-EKV Model
Y. Chauhan, F. Krummenacher and A. M.
Ionescu Ecole Polytechnique Fédérale de Lausanne
(EPFL), Switzerland
http//legwwww.epfl.ch
2
Outline

• HV-EKV
• status and people
• short comment on evaluation feedback
• New validations
• model validation on SOI-LDMOS (AMIS)
• demonstration of quasi-saturation modeling on
  VDMOS (AMIS)
• NEW Modeling of Lateral Non-uniform doping in
  HVMOS (see IEDM06 paper)

3
HV-EKV (1)
EKV Simple, physically based parameters Less
parameters than BSIM HVEKV EKV low voltage
Rdrift charge (Vk)
4
HV-EKV (2) scalable drift resistance
5
HV who _at_ EPFL?
Electronics Laboratory (50 people, 4 profs, 24
PhDs)

• Yogesh Chauhan (100)
• Francois Krummenacher (EKV team, 25)
• Costin Anghel (now with CEA, France)
• Adrian M. Ionescu, prof. (10)
• Michel Declercq, prof. (10)

High Voltage MOS modeling a strategic field but
not our mainstream research à reorganization
needed if HV-EKV adopted by CMC
6
HV-EKV status and feedback

• Status corrected code released to CMC and
  evaluations in progress
• Feedback simplicity, ease of extr, doc
• Accuracy problem in quasi-saturation -
• Accuracy with temperature -
• Relatively slow because of Verilog A
• Compared with sub-circuit models evaluators
  agree its a compact model!

In fact needs appropriate tunning
7
Problem with quasi-saturation?

• Tune avsat quasi-sat for LDMOS and VDMOS!

?
!
8
Model Validation on SOI-LDMOS (1)

• Very accurate on both ID-VG and gm-VG _at_ low VD

9
Model Validation on SOI-LDMOS (2)

• Good accuracy on both ID-VG and gm-VG _at_ high VD

10
Model Validation on SOI-LDMOS (3)

• quasi-saturation very well modeled
• good/acceptable accuracy on gds-VD

11
Quasi-Saturation Modeling on VDMOS

• in contrast with some criticism the model is
  able to accurately describe the quasi-saturation
  region of VDMOS

12
New model extension

• Model of LAteral Non-Uniform Doping in the
  intrinsic MOSFET (new EKV-like model for the low
  voltage transistor)
• Verilog A code not yet available
• Evaluated with Matlab code on both numerical
  simulations and experimental data
• Reported in IEDM 2006

13
Why LAMOS modeling?
14
Modeling of Lateral Non-uniform doping in
HV-MOSFET
Assume
Nonlinear ODE
Drain Current
Not explicit
Total Inversion Charge Density
15
Model Validation on 50V VDMOS
Transfer Characteristics (ID-VG)

• Weak inversion to Strong inversion transition
• Subthreshold slope correctly matched
• Good accuracy

16
gm-VG for VD0.1-0.5V

• Subthreshold slope correctly matched
• Descending slope drift resistance

17
Output Characteristics
Self-Heating
Impact-Ionization

• Linear region correctly modeled by drift
  resistance.
• Self Heating Effect
• Peaks on gds

18
Gate-to-Drain Capacitance CGD vs VG

• Slope Peaks effect of lateral non-uniform
  doping

• Sharp decrease effect of drift region (good
  modeling of drift region or VK must)

19
CGSCGB vs VG

• Peaks effect of lateral non-uniform doping
• Sharp decrease and shift of peaks effect of
  drift region

20
CGG vs VG

• Peaks and shift of peaks little contribution
  from lateral non-uniform doping and greater
  contribution from drift region

21
Conclusion

• General HV-MOS Model including lateral
  non-uniform doping (code will be available in
  March 2007)
• Complex capacitance behavior of high voltage MOS
  was explained using numerical device simulations
• A new model for lateral non-uniformly doped
  devices was presented
• Self-Heating effect included
• Very good performance in DC and transient
  operations
• Model validated on industrial devices VDMOS and
  LDMOS
• Peaks and shift of peaks in capacitances with
  bias correctly modeled

22
Acknowledgements
Christian Maier, Heinisch Holger Robert Bosch,
Germany Andre Baguenier Desormeaux Cadence
Design Systems, France B. Desoete AMI
Semiconductor, Belgium J.-M. Sallesse, A.S.
Roy EPFL, Switzerland
Funded by European Commission project
ROBUSPIC Website- http//www-g.eng.cam.ac.uk/rob
uspic/