Hints, Tips and Solutions

Q. How can I characterize the parasitic BJT behavior of SOI devices using UTMOST III.

A. UTMOST III SOI module has three routines for characterization of the parasitic BJTs of SOI devices. These routines are "IC/VCE", "Gummel" and "INT_BJT" All routines require to have 5 terminal devices with bulk connection for data collection.

The "Gummel" routine sweeps the voltage for the bulk terminal and keeps a constant voltage for the gate, source, drain and backgate terminals. The measured currents are drain and bulk currents. However during this measurement the bulk and drain diodes can be on and the diode current can be added to the bulk current. This additional diode current prevents the user from characterizing the pure intrinsic bipolar device.

The "INT_BJT" routine was developed to exclude any additional diode currents which may contribute to the intrinsic bipolar behavior. In order to achieve this goal the source terminal bias was swept and bulk and drain terminals were kept at the same potential (typically at ground potential). The gate potential's effect on the parasitic bipolar behavior was also investigated. Therefore the VGS bias stepping function was added to the routine (Figure 1). In order to keep the VGS bias constant the VG bias has to follow the sweeping bias of VS. The constant VGS bias is obtained by using the "offset" (VAR1') function of the DC analyzers HP4145 and HP4155/56. The HP4142 DC analyzer uses the point by point (user mode) measurement technique to keep the constant VGS bias.

Figure 1. The Set measureemnt screen of the INT_BJT routine.

 

The "INT_BJT" routine displays the measured drain and bulk currents versus VSB (source to bulk) voltage (Figure 2). The global and local optimization techniques can be utilized to extract parameters which are used to model the parasitic bipolar behavior in SOI devices. The important model parameters of the parasitic bipolar device in BSIM3v3 partially depleted SOI model are:

ISBJT (BJT injection saturation current)

NBJT (Power coefficient of channel length dependency for bipolar current)

LBJT0 (Reference channel length for bipolar current)

VABJT (Early voltage for bipolar current)

AELY (Channel length dependency of early voltage for bipolar current)

AHLI (High level injection parameter for bipolar current)

Figure 2. Idrain and Ibulk currents versus VSB voltage at three different VGS bias points.

 

Q. Can Silvaco provide a bipolar modeling service using the advanced bipolar models such as MEXTRAM and VBIC?

A. Yes. Both MEXTRAM and VBIC models have been implemented in SmartSpice and UTMOST III. Silvaco characterization lab provides the advanced bipolar modeling service using both models as well as Gummel-Poon and Macro models.

The advanced bipolar models provide many advantages over the Gummel-Poon and Macro models. Some of the major advantages can be listed as:

• Improved early effect modeling
  
  
• Quasi-saturation modeling. (Both Gummel Poon (including level=2 GP model) and macro models can not model this behavior)
  
  
• Parasitic substrate current modeling. (The Gummel-Poon is three terminal model and it doesn't include the substrate node. The Macro models can handle the substrate current but the modeling of the parasitic devices of the Macro model are not accurate)
  
  
• Avalanche multiplication modeling.