📘 BSIM4 Analyzer Reliability

Links

Language	GitHub Pages 🌐	GitHub 💻
🇺🇸 English

Unified NMOS HCI / PMOS NBTI Reliability Analysis Framework
(NGSPICE + Python)

📌 Overview

BSIM4 Analyzer Reliability is an integrated framework for analyzing semiconductor aging mechanisms:

NMOS Hot-Carrier Injection (HCI)
PMOS Negative Bias Temperature Instability (NBTI)

The framework combines:

SPICE-based measurement at t = 0
Python-based degradation models for t > 0
to produce a hybrid, reproducible reliability analysis workflow.

Key Features

✔ Automated VG–ID sweep & DC current extraction
✔ Computes four degradation quantities:
ΔVtg, ΔVtc, ΔIdlin, ΔIdsat
✔ Full support for both NMOS HCI and PMOS NBTI
✔ Structured directory for reproducible results
✔ Suitable for academic research & corporate reliability evaluation
✔ Parameterized degradation models for extensible aging studies

📁 Directory Structure

bsim4_analyzer_reliability/
├── models/
│    ├── nmos130.sp
│    └── pmos130.sp
│
├── templates/
│    ├── template_hci_nmos.cir
│    ├── template_nmos_vgid.cir
│    ├── template_nbti_pmos.cir
│    ├── template_pmos_vgid.cir
│
├── run/
│    ├── run_hci_nmos.py
│    └── run_nbti_pmos.py
│
├── plot/
│    ├── plot_hci_nmos.py
│    └── plot_nbti_pmos.py
│
└── results/
     ├── 90nm/
     └── 130nm/
          ├── hci_nmos/
          ├── hci_nmos_vgid/
          ├── nbti_pmos/
          └── nbti_pmos_vgid/

🔧 Installation & Requirements

■ Python Packages

numpy
pandas
matplotlib

■ NGSPICE

Verified with ngspice 33–42
Ensure ngspice is available in your system PATH

🔬 Reliability Analysis Workflow

The same conceptual flow applies to both NMOS and PMOS.

t = 0
 ├─ VG–ID sweep
 │     ├→ extract Vtg0 (gmmax method)
 │     └→ extract Vtc0 (constant-current method)
 ├─ DC extraction
 │     └→ Idlin0, Idsat0

t > 0  (Python degradation model)
 ├─ ΔVth(t)
 ├─ ΔId_rel(t)
 └─ reconstruct Vtg1, Vtc1, Idlin1, Idsat1

→ Save CSV results
→ Generate degradation plots
→ VG–ID overlay plots

🟥 NMOS HCI (Hot-Carrier Injection)

▶ Stress Bias Conditions

| Node   | Voltage      |
|--------|--------------|
| Drain  | High Vd (e.g., 1.2 V) |
| Gate   | Stress voltage |
| Source | 0 V |
| Bulk   | 0 V |

▶ Extracted Quantities

ΔIdlin
ΔIdsat
ΔVtg (gmmax-based Vth shift)
ΔVtc (constant-current Vth shift)

▶ Run HCI Simulation

python run/run_hci_nmos.py

🟦 PMOS NBTI (Negative BTI)

▶ Stress Bias Conditions

| Node   | Voltage |
|--------|---------|
| Source | +1.2 V |
| Bulk   | +1.2 V |
| Drain  | 0 V |
| Gate   | 0 → −1.2 V |

▶ VG–ID Output Specification (4 columns, fixed)

Vg
Vs
Vgs = V(g) − V(s)
Id = abs(i(Vd))

Python side:

Vgs = arr[:, 2]
Id  = arr[:, 3]

▶ NBTI Degradation Model (PMOS)

■ Threshold Voltage Shift

dVth = A_vth * (t ** p_vth)
Vtg1 = Vtg0 - dVth
Vtc1 = Vtc0 - dVth

■ Drive Current Degradation

dIdrel = -A_id * (t ** p_id)
Idlin1 = Idlin0 * (1 + dIdrel)
Idsat1 = Idsat0 * (1 + dIdrel)

▶ Run NBTI Simulation

python run/run_nbti_pmos.py

📊 Output Files

The framework creates structured result sets for each node (90nm / 130nm).

■ NMOS HCI

results/<node>/hci_nmos/
    dIdlin_vs_time.png
    dIdsat_vs_time.png
    dVtg_vs_time.png
    dVtc_vs_time.png
    hci_summary.csv

■ PMOS NBTI

results/<node>/nbti_pmos/
    dIdlin_vs_time.png
    dIdsat_vs_time.png
    dVtg_vs_time.png
    dVtc_vs_time.png
    nbti_pmos_summary.csv

■ VG–ID Overlays

<device>_vgid/
    vgid_all_linear.png
    vgid_all_log.png

🧠 Internal Degradation Models

■ NMOS HCI

ΔVth = A_vth * t^p
ΔId  = -A_id * t^p

■ PMOS NBTI

Vtg(t), Vtc(t) = Vth0 − ΔVth
Id(t) = Id(0) * (1 + ΔId)

🧩 Extensible Architecture

Easily extendable to:

NMOS PBTI (Positive BTI)
Arrhenius temperature acceleration
Duty-cycle / AC stress
BSIM4 parameter aging injection
Packaging as a Python library:
```
pip install aging-model
```

📎 Reliability Analysis — Reference Figures

■ NMOS HCI : Vg–Id Degradation (Linear Scale)

■ NMOS HCI : ΔVtg vs Stress Time (gmmax Method)

■ PMOS NBTI : Vg–Id Degradation (Linear Scale)

■ PMOS NBTI : ΔVtg vs Stress Time

📄 Hybrid License

Item	License	Description
Source Code	MIT License	Free to use, modify, redistribute
Text Materials	CC BY 4.0	Attribution required
Figures / Plots / Generated Data	CC BY-NC 4.0	Non-commercial use only
External References	Original license applies	Cite properly

👤 Author

📌 Item	Details
Name	Shinichi Samizo
GitHub