openuv 1.0.1

Open Source EUV Lithography Simulator — from plasma source to CD metrology

OpEnUV — Open Source Extreme Ultraviolet Lithography Simulator

From plasma source to CD metrology — full-stack EUV lithography simulation on your laptop.

⚛️ First-principles physics	CXRO atomic scattering → TMM reflectivity → RCWA mask diffraction → Hopkins imaging → Dill ABC resist
🚀 GPU-native	PyTorch autograd throughout — differentiable from mask geometry to CD
🧪 Validated	455 unit tests, cross-checked against CXRO database, independent literature, and Grok verification
📦 Zero commercial dependencies	Apache 2.0 — fork, modify, deploy freely

---

What makes OpEnUV different?

OpEnUV is the only open-source tool that models the complete EUV lithography pipeline from photon to CD:

Module	Method	OpEnUV	IMD	GD-Calc	OpenLithoHub
Material constants	CXRO/Henke f₁,f₂	✅	✅	❌	❌
Multilayer mirror	Transfer-Matrix (S-matrix)	✅	✅	❌	❌
Mask diffraction	RCWA 1D/2D + S-matrix cascade	✅	❌	✅	❌
Aerial image	Hopkins/Abbe + pupil + source	✅	❌	❌	❌
High-NA optics	Anamorphic 4×/8×, Zernike	✅	❌	❌	❌
Resist chemistry	Dill ABC + PEB + development	✅	❌	❌	❌
Plasma source	LPP Sn-droplet spectral model	✅	❌	❌	❌
Stochastics	Shot noise, LER/LWR	✅	❌	❌	✅
CD metrology	Process window, Bossung	✅	❌	❌	❌
Optimization	Differentiable OPC/ILT bridge	✅	❌	❌	✅
Web API	REST + dashboard	✅	❌	❌	❌

IMD is the gold standard for multilayer reflectivity. GD-Calc solves rigorous coupled-wave analysis. OpenLithoHub benchmarks OPC quality. OpEnUV is the only tool that connects all the physics — from plasma spectrum through mask diffraction to developed resist profile.

---

Validated test calculations

All results independently verified against the CXRO/Henke database and literature values. Full details in testberechnungen.md.

Test	Result	Method
Mo optical constants (91.84 eV)	n = 0.92335, k = 0.00647	CXRO database interpolation
Si optical constants (91.84 eV)	n = 0.99900, k = 0.00183	CXRO database interpolation
Mo/Si multilayer 50 BL @ 6°	R = 64.7% (ideal) / 60.6% (σ=0.5 nm)	S-matrix TMM + Névot-Croce
Wavelength: 13.5 nm	—	E = hc/λ = 91.84 eV
Illumination: NA 0.33	Conventional, σ = 0.8	Dipole / annular / quasar
Aerial image NILS	4.36 @ 64 nm pitch, 32 nm CD	Hopkins formulation
Process window	CD vs. dose × focus	Bossung plot

"The calculations are solid and physically sound." — Grok (independent review, 2026-07-09)

---

Quick start

pip install openuv                    # install from PyPI
pip install -e ".[dev]"               # or from source (development)

# End-to-end simulation
euv simulate --period=64 --cd=32 --dose=20

# Process window (dose-focus Bossung plot)
euv process-window --period=64 --cd=32

# Web dashboard
euv serve
# → http://localhost:8000

# Query material database
euv materials Mo --energy=91.84

# Performance benchmark
euv bench

Or from Python:

from euv.pipeline import SimulationConfig, run_simulation

cfg = SimulationConfig(
    period_nm=64, line_width_nm=32,
    dose_mj_cm2=20, na=0.33, sigma=0.8,
    resist_model="full_chem",
)
result = run_simulation(cfg)
print(f"CD = {result.cd_nm:.1f} nm")
print(f"NILS = {result.nils_value:.3f}")

---

Architecture

src/euv/
├── source/         LPP Sn-plasma emission model (spectrum + dose)
├── materials/      CXRO atomic scattering factors f₁,f₂ (Z = 1–92)
├── optics/         Multilayer TMM (S-matrix, Névot-Croce, grading)
├── mask3d/         RCWA 1D/2D Fourier Modal Method (stable S-matrix)
├── aerial/         Abbe/Hopkins imaging, pupil, source shapes
├── resist/         Dill ABC exposure, PEB, development, shot noise
├── io/             CLI, GDSII, rasterization
├── metro/          CD metrology, process window, SEM rendering
├── opc/            Differentiable OpenILT bridge
├── accel/          GPU acceleration + VRAM management
├── etch/           Etch bias model
├── calibrate/      Wafer calibration (scipy optimisation)
├── api/            FastAPI REST server + web dashboard
└── pipeline.py     End-to-end simulation orchestration

┌────────┐   ┌────────┐   ┌────────┐   ┌────────┐   ┌────────┐   ┌────────┐
│ SOURCE │ → │  MASK  │ → │OPTICS │ → │ AERIAL │ → │ RESIST │ → │ METRO  │
│ LPP Sn │   │ RCWA   │   │ TMM   │   │Hopkins│   │Dill ABC│   │ CD/PW  │
└────────┘   └────────┘   └────────┘   └────────┘   └────────┘   └────────┘

---

CLI reference

Command	Description
euv simulate	Full end-to-end simulation (CLI args or YAML/JSON config)
euv process-window	Dose-focus Bossung grid → DoF + EL
euv make-mask	Generate line/space test mask (GDSII)
euv materials	List elements / query n + ik at any photon energy
euv serve	Launch the REST API + web dashboard
euv bench	Performance benchmark
euv info	System + module overview

---

Tutorials

Notebook	Description
basic_simulation.ipynb	Run your first EUV simulation
materials.ipynb	Query CXRO optical constants
process_window.ipynb	Compute a Bossung plot

---

Project status

Milestone	Status
Project scaffold & CXRO materials	✅
Multilayer optics (S-matrix TMM)	✅
Mask 3D (RCWA 1D + 2D)	✅
Aerial image (Abbe/Hopkins)	✅
High-NA imaging (anamorphic)	✅
End-to-end pipeline	✅
LPP source model	✅
Resist chemistry (Dill ABC + PEB)	✅
Stochastic effects (shot noise, LER/LWR)	✅
CD metrology & process window	✅
Inverse lithography (OpenILT bridge)	✅
GPU acceleration	✅
REST API + web UI	✅
Tutorials & documentation	✅
Docker deployment	✅
CI/CD pipeline	✅
Test count	455 / 455 passing
License	Apache 2.0
PyPI	🔜
v1.0 release	🔜

---

Benchmark: small system performance

Config	Grid	Orders	Time (CPU)	Time (GPU)
Small	256×256	21	~0.1 s	—
Medium	512×512	31	~0.3 s	—
Standard	1024×1024	51	~1.2 s	~0.1 s

---

Documentation

Full Sphinx documentation at docs/:

• Overview
• Installation
• Architecture
• Quickstart
• API reference
• Contributing

---

License

Apache 2.0 — see LICENSE.

---

How to contribute

We welcome contributions! See CONTRIBUTING.md for guidelines.

• 🐛 Open an issue for bugs
• 💡 Start a discussion for features
• 🔀 Submit pull requests on the main branch
• 📝 Improve documentation or add tutorials
• 🎓 If you use OpEnUV in research, cite it! (citation coming with v1.0)

---

Support

PayPal: [e-mail] — send via PayPal Friends & Family

Development supported by GitHub Sponsors and PayPal donations. OpEnUV is open source — contributions welcome. ❤️