Mike Peralta

Semiconductor Modeling Using Relxpert and Design of Experiments Mike Peralta

RELIABILITY MODELING USING RELXPERT

Cadence's simulation tool for Reliability Simulation, RelXpertTM, requires models to properly predict such behavior as Hot Carrier Injection, which among other things, causes the threshold voltage, VT, to shift according to ID, VDG, and L (Gate Length). Proper reliability modeling requires sample measurements of VT shifts (or other device parameters such as gm etc.) at known ID, VDG, and L. Since reliability Hot Carrier shifts tend to have significant variation from sample-to-sample, it is imperative to use a few sample measurements at each ID, VDG, and L use conditions. We show how using a 3 factor DOE (Design Of Experiment) with 4 replicates for each factor combination is an excellent approach to distinguishing between actual factor effects and random experimental measurement variation.

In this monograph we cover the following:

1. Replicates are used to distinguish random measurement variation from actual factor effects.

2. DOE's and use conditions such as ID, VDG, and L are used to fully model the use condition range.

3. Core industry standard Hot Carrier model is used.

4. Moderate current levels are used for reasonable time durations for the measurement procedure.

5. ICCAPTM parameter extraction software is used to extract the standard Hot Carrier model parameters.

6. How the RelXpertTM model is scaled to match measured-to-RelXpert-simulator values.

7. Differential Amplifier (Diff Amp) Example to illustrate how Hot Carrier VT shifts can be reduced thereby enhancing circuit reliability.

8. How ID, VDG can also be changed instead of W, L to reduce Hot Carrier VT shifts.