The Compact Model Council Minutes

December 12th, 1997

Meeting opened: 8:25am on December 12th, 1997 All handouts were provided to Patti Rusher (EIA). Members absent were AMD, Analog Devices, Avant!, Cadence, Hitachi, Intel, and NEC.  Members present were DEC, HP, IBM, Lucent, Motorola, National Semiconductor, Philips, Rockwell, Siemens, TSMC, and TI.

16 people attended the Compact Model Council meeting. Prospective members attending were SGS-Thompson, National Semiconductor (who also announced their intention to join), and TSMC (a member starting in 1998).

This was the third meeting of the members of the Compact Model Council under the EIA umbrella. After welcoming the attendees, Bhaskar Gadepally (Motorola) and Britt Brooks (TI), reviewed the current membership.

The current members of the CMC are:

Companies that are not yet members, but are pursuing membership are:

Groups monitoring the CMC activities are:

Included here are the Web sites of interest to the CMC members.

• CMC homepage - http://www.eia.org/eig/CMC
• NIST Model Validation homepage - http://ray.eeel.nist.gov/modval.htm
• BSIM3v3 Homepage - http://www-device.EECS.Berkeley.EDU/~bsim3

BSIM3v3 Updates and Action items

Weidong Liu (UCB) handed out a summary of the BSIM3v3 changes for the BSIM3v3.2 release. This allowed Keith Green (TI), Bernd Lemaitre (Siemens), and Joe Watts (IBM) to combine the section on BSIM3v3 updates and the action item list from the previous meeting. Keith began the session by covering the action items and their status. See Keith's handout for details.

Bernd then presented the review of HSPICE/Spectre specific parameters and how they were originally developed. He made several recommendations (see handout for details) that will be introduced into BSIM3v3.2. He also noted that formulation for bias dependence of R_DS was necessary for the proper behaviour to be modeled. Additionally, Bernd presented some information on extracting L_eff, R_D, and R_S from a Siemens paper at ESSDERC. Hiroo Masuda (Hitachi) had also mentioned this topic in a previous message.

During Bernd's presentation, Joe Watts made a proposal for the HDIF and LDIF terms from HSPICE. After intense discussion on the topic, Joe agreed that he would follow up with the vendors (Cadence, Avant!, and Mentor) and try to reach a consensus for how the HDIF and LDIF parameters should be included (Again, please see his handout).

The final portion of this section presented messages from member representatives from Hitachi and NEC. Issues such as code portability and documentation were raised as well as suggestions for improvement of the BSIM3v3 model formulations.

Here is a list of the action items and their status:

1. The UC-Berkeley BSIM3 team will incorporate fixes for the current bug list in the next release of BSIM3v3.

   b3mpar.c:                 "w" is missing in ldvt0w
   b3check.c:           Pd & Ps clamps removed
   b3temp.c:             leff should be leffCV for abulkCVfactor
   b3ld.c and b3acld.c: Bug fixes for NQS model
   b3ld.c:               Zero-bias flat-band voltage Vfbzb is used for
                        capMod=1, 2 and 3; but if (BSIM3version <3.2) the old bias-dependent Vfb is still kept for the consideration of compatibility. b3temp.c: "Vbm is ignored because k1 or k2 is given" is removed, since vbm is still used to limit Vbseff. Clean-ups: Redundant if-clause for BSIM3lprt & BSIM3oxideTrapDensity in b3set.c file and useless One_Third_CoxWL and Two_Third_CoxWL for capMod="2" are removed. 2. Paul Humphries (AD), Mishel Matloubian (Rockwell), and Keith Green (TI) will formulate requirements for an updated Isub model and forward to UCB. Only "alpha1" is introduced into the model: Isub ~ (alpha0 + alpha1 * leff) / leff and verified by Paul's experimental data and model cards. 3. Paul Humphries (AD), Shiuh-Wuu Lee (Intel), Keith Green (TI), Richard Taylor (National), Larry Yu (AMD), Toshiyuki Saito (NEC), and Terenc Masszi (DEC) will formulate requirements for a BSIM3v3-based MOS capacitor model and forward to UCB. More work required. It was agreed that the CMC would provide UCB with specifications for a MOS model that is separate from the BSIM3v3 MOSFET model. 4. Min-Chie Jeng (Cadence) and Bob Daniels (Avant!) will work with UCB to restructure the NQS model in BSIM3v3 to make it easier to transfer to other simulators. Separation of NQS model from DC and QS parts in conductance, equivalent current and matrix evaluations. NQS bug fixing and code clean-up to improve simulation performance. Implementation of NQS model for pole-zero analysis. 5. Weidong Liu (UCB) will document the proposed CAPMOD="3" model. Documentation will be available from Weidong upon request. Available since Nov. 12. Send Weidong Liu e-mail (liuwd@pauli.eecs.berkeley.edu) to obtain a copy. 6. Weidong Liu (UCB) will fix the CGSO- and CGDO-related code bug reported by Joe Watts (IBM). It was agreed there was not a bug. 7. Ping-Chin Yeh (HP) and Joe Watts (IBM) will formulate requirements for modeling bias-dependent inner fringing capacitance and forward More work required. It was agreed this would be considered for inclusion in v3.3, not v3.2. 8. Ping-Chin Yeh (HP) and Joe Watts (IBM) will formulate requirements for adding new parameters to fit C-V "turn on" characteristics and forward to UCB. A turn-on parameter "noff" is introduced into VgsteffCV. if (noff < 0.1) && (noff> 4.0), then a warning message will be output
   due to the poor values.


9. Joe Watts (IBM) and John Krick (TI) will formulate a proposal for adding tox dependence in VTH0, K1, and K2.
   Vth now becomes
    Vth = Vth0ox + k1ox * sqrt(phis - vbseff) - k2ox * vbseff
      + k1ox * (sqrt(1 + nlx / leff) - 1) * sqrt(phis) + deltaVth
   where
    Vth0ox = vth0 - k1 * sqrt(phis)
    k1ox = k1 * tox / toxm;
    k2ox = k2 * tox / toxm;
   if toxm <= 0, then a fatal error message will be generated. 10. Bernd Lemaitre (Siemens) will work with UCB to investigate and fix the bug associated with VFB in BSIM3v3. VFB now is a model parameter with a default value: if (vth0 is given), then VFB will be computed from vth0; otherwise, VFB="-1" for backward compatibility. 11. Keith Green (TI) will work with UCB to remove the source/body and drain/body diode forward-bias current limiting function. A current limiting parameter Ijth is introduced which corresponds to a voltage Vjsm and Vjdm; if... Ijth is specified to be zero, a pure diode model will be triggered; Ijth is given & non-zero, when Vbs>Vjsm (Vbd>Vjdm), the diode
       current will be linearized;
    Ijth not specified, Vjsm=Vjdm=0.5V, backward compatible with v3.1;
    Ijth <0, a fatal error message is given. 12. UCB will add code to BSIM3v3 which will cause a warning to be generated if PSCBE2<="0." If PSCBE2<="0," the user will be warned of the poor value used. 13. UCB will add boundaries to parameters MOIN and ACDE in the developmental CAPMOD="3" code. If (moin < 5.0) or (moin> 25.0), warning messages will be generated.
    If (acde <0.4) or (acde> 1.6), warning messages will be generated.

14. Bernd Lemaitre (Siemens) will investigate the necessity to add HSPICE-specific parameters, e.g., LDIF and HDIF, to the standard BSIM3v3 model and report back to the CMC via e-mail. Needs further discussion on how to calculate S/D diffusion resistance without overlapping the existing parasitic Rds in v3.1.  Joe Watts (IBM) will work with Mentor, Cadence, and Avant! to reach consensus on how the parameters HDIF, LDIF, etc., should best be worked into the BSIM3v3.2 model and report back to the CMC.

These are new items from the Dec. 12 meeting:

15. It was agreed that temperature dependence of junction capacitance would be added to BSIM3v3.2. Joe Watts (IBM) and Dick Klaassen (Philips) have proposals for how to do this. Consensus will need to be reached by the CMC.

16. It was agreed that the inversion charge equation used in the thermal noise model would be made consistent with the charge equations used in the large-signal transient model(s). Details to be worked out with UCB and Keith Green (TI) and reported back to CMC.

BSIM3v3 standardization

Britt Brooks presented a quick synopsis of where BSIM3v3.1 stands as far as standardization is concerned (See his handout for details). He also reviewed the EIA requirements for documentation of a standard.

Membership Matters

The first item in this section was the payment to UCB for the productization effort on BSIM3v3. Chenming Hu's proposal was shown and it was agreed that he had met the requirements based upon the proposal and that the payment of $80K would be made by the EIA to Chenming Hu and UC Berkeley.

Second, the topic of Council leadership was revisited. Bhaskar Gadepally (Motorola) mentioned a possible plan for phasing in new members to serve as co-chairs by the end of 1998. Ideally, they would work with the existing co-chairs starting in July 1998 and begin to take over part of the responsibility and planning in 3Q98.

Third, the FSA hosted a class in November for BSIM3v3. Two Council members, Keith Green (TI) and Min-Chie Jeng (Cadence) were teachers at the course. There were 60 FSA members in attendance and the follow-up reports were excellent.

Fourth, Bhaskar welcomed the new members that were present and announced that there were several other prospective members we could encourage to join the CMC. They are included in the list at the beginning of these minutes. Additionally, we would like to promote the CMC at upcoming meetings and perhaps hold another class on BSIM3v3. It was agreed that IEDM was not the best forum for IC designers, hence other possible avenues would be CICC, ISSCC, and DAC.

Next Meeting Planning

The next meeting was tentatively planned for the week of March 23rd, 1998 in Japan in conjunction with the ICMTS meeting. Dates have not been confirmed yet, but all members should check their calendars for availability. Also, since this would be the first international meeting, travel may need to be booked as early as possible.

Meeting closed: 12:30pm on December 12th, 1997

Here is the list of attendees and their email addresses.

Britt Brooks                    brittb@ti.com                   (972) 997-6037
Texas Instruments

Bhaskar Gadepally      Bhaskar_Gadepally@mesaqm.sps.mot.com    (602) 655-3367
Motorola

Keith Green                     kgreen@msp.sc.ti.com            (972) 480-7645
Texas Instruments

Andre Juge
SGS-Thompson

D.B.M. Klaassen         klaassen@natlab.research.philips.com    +31 40 2744251
Philips

Dr. Bernd Lemaitre       lemaitre@hl.siemens.de                49-89-636-3775
Siemens AG

Boon-Khim Liew               bkliew%td@tsmc.com.tw              (886-3)572-6411
TSMC

Weidong Liu               liuwd@pauli.eecs.berkeley.edu         (510) 642-1010
Univ. of California, Berkeley

Samuel Martin           sam@physics.bell-labs.com               (908) 582-4047
Lucent

Mishel Matloubian      mishel.matloubian@rss.rockwell.com      (714) 221-6641
Rockwell

Stephen Oxx
DEC

Rosana Perez                    rosana@sr.hp.com                (707) 577-5296
Hewlett-Packard

Patti Rusher                    pattir@eia.org                  (703) 907-7545
EIA

Richard Taylor                  Richard.F.Taylor@nsc.com        (408) 721-4884
National Semiconductor Corp.

Joe Watts                       jwatts@btv.ibm.com              (802) 769-1874
IBM

Ping-Chin Yeh                   ypc@hpmozart.corp.hp.com        (415) 857-6198
Hewlett-Packard

Thank you all for your participation and attendance.

This meeting was conducted in accordance with the EIA Legal Guides and EIA Manual of Organization and Procedure.