【Semiconductor】🧰 08. SemiDevKit

— An Open Educational Kit Covering Device Physics → BSIM4 → SPICE → Physical Design End-to-End

This article introduces SemiDevKit,
an open educational toolkit that enables end-to-end learning, from:

• semiconductor device physics
• compact modeling (BSIM4)
• SPICE simulation and reliability analysis
• to VLSI physical design using OpenLane

all within a single, coherent framework.

Official website and GitHub repository:

• GitHub Pages (English documentation)
  https://samizo-aitl.github.io/SemiDevKit/
• GitHub Repository
  https://github.com/Samizo-AITL/SemiDevKit

🧭 What Is SemiDevKit?

SemiDevKit is an open-source educational toolkit
that covers the entire workflow listed below:

• Semiconductor device physics (1D Poisson / Drift–Diffusion)
• Compact modeling (BSIM4)
• SPICE analysis (DC / AC / scaling / reliability)
• Reliability evaluation (NBTI / HCI)
• OpenLane-based VLSI physical design flow (RTL → GDSII)

🎯 Target Audience

• Students who want a systematic understanding of semiconductor devices and VLSI
• Researchers and engineers who want to experiment with device modeling, SPICE, and physical design in a lightweight environment
• Educators building teaching materials independent of commercial TCAD/EDA tools

✨ Key Features

🔬 1. Device Physics / TCAD Playground

Under the tcad/ directory,
SemiDevKit provides a TCAD playground centered on:

• 1D Poisson and Drift–Diffusion solvers
• MOSFET Vg–Id and Vd–Id characteristics
• Ferroelectric (FE) P–E modeling using the Landau–Khalatnikov equation

Unlike commercial TCAD tools, which are often massive black boxes,
these implementations are Python-based and traceable directly from equations.

📐 2. Compact Modeling (BSIM4 Suite)

The bsim/ directory contains a comprehensive set of tools for BSIM4.

• Automatic generation of BSIM4 model cards
• Physical parameter extraction, including:
  • Oxide thickness (tox)
  • Substrate doping (Na)
  • Flat-band voltage (Vfb)
  • Mobility (μ0)
  • Channel length (L) and width (W)

This structure explicitly bridges:

Device physics → compact models

⚡ 3. SPICE Analysis (DC / AC / Reliability)

SPICE analysis scripts are also provided under bsim/.

• DC analysis (Vg–Id, Vd–Id)
• AC analysis (Cgg–Vg, etc.)
• Dimensional scaling analysis (L, W sweeps)
• Reliability analysis (NBTI / HCI)

Using Python + ngspice, the flow is fully automated:

Model generation → SPICE execution → visualization

🧱 4. VLSI Physical Design (OpenLane-Lite)

The openlane/ directory includes an education-optimized OpenLane-Lite environment.

• Lightweight flow based on OpenLane 2023
• Docker / WSL2 compatible
• Includes minimal example designs (e.g., spm)
• Full RTL → GDSII execution

This allows users to:

Experience a complete IC physical design flow without commercial tools

🗂 Repository Structure (Overview)

SemiDevKit/
├── tcad/                       # Device physics (TCAD Playground)
│   ├── TCAD_PLAYGROUND
│   └── TCAD_PLAYGROUND_PZT
│
├── bsim/                       # Compact models + SPICE analysis
│   ├── Paramus
│   ├── BSIM4_ANALYZER_DC
│   ├── BSIM4_ANALYZER_CV
│   ├── BSIM4_ANALYZER_DIM
│   └── BSIM4_ANALYZER_RELIABILITY
│
├── openlane/                   # Physical design flow
│   ├── openlane-lite
│   └── openlane-superstable
│
└── docs/                       # Documentation (MathJax enabled)

🚀 Setup and Execution Examples

1️⃣ Clone the Repository

git clone https://github.com/Samizo-AITL/SemiDevKit.git
cd SemiDevKit

2️⃣ Required Environment

• Python 3.10+
• NumPy / SciPy / Matplotlib
• ngspice
• Docker (for OpenLane-Lite)
• WSL2 recommended on Windows

3️⃣ Example: SPICE DC Analysis

cd bsim/BSIM4_ANALYZER_DC/run
python run_vd.py
python run_vg.py

This workflow:

• Generates BSIM4 model cards
• Runs SPICE simulations
• Automatically plots Vd–Id and Vg–Id characteristics

4️⃣ Example: Running OpenLane-Lite

cd openlane/openlane-lite
./docker/run_in_docker.sh

Execution steps:

1. Launch OpenLane 2023 container
2. Load a small example design
3. Run the RTL → GDSII flow
4. Generate GDS output

📘 Documentation (MathJax Enabled)

Theoretical notes under docs/ are published on GitHub Pages.

Covered topics include:

• Device physics (Poisson / Drift–Diffusion)
• Compact model theory (BSIM4)
• SPICE analysis (DC / AC / reliability)
• NBTI / HCI degradation models
• RTL → GDS physical design flow

Here, code and theory are directly linked.

📜 License

📝 Summary

SemiDevKit enables end-to-end learning
from semiconductor device physics to IC physical design
without relying on commercial TCAD or EDA tools.

Its strongest points include:

• Python implementations traceable down to equations
• Full BSIM4-based SPICE analysis
• RTL → GDSII flow using OpenLane-Lite

We recommend starting with:

• bsim/BSIM4_ANALYZER_DC for DC analysis

• openlane/openlane-lite for the physical design flow

• GitHub Pages
  https://samizo-aitl.github.io/SemiDevKit/
• GitHub Repository
  https://github.com/Samizo-AITL/SemiDevKit