prefab 0.4.1

Machine learning based prediction of photonic device fabrication

PreFab

PreFab leverages deep learning to model fabrication-induced structural variations in integrated photonic devices. Through this virtual nanofabrication environment, we uncover valuable insights into nanofabrication processes and enhance device design accuracy.

Prediction

PreFab accurately predicts process-induced structural alterations such as corner rounding, washing away of small lines and islands, and filling of narrow holes in planar photonic devices. This enables designers to quickly prototype expected performance and rectify designs prior to nanofabrication.

Correction

PreFab automates corrections to device designs, ensuring the fabricated outcome aligns with the original design. This results in reduced structural variation and performance disparity from simulation to experiment.

Models

PreFab accommodates unique predictor and corrector models for each photonic foundry, regularly updated based on recent fabrication data. Current models include (see full list on docs/models.md):

Foundry	Process	Latest Version	Latest Dataset	Model Name	Model Tag	Status
ANT	NanoSOI	v5 (Jun 3 2023)	d4 (Apr 12 2023)	ANT_NanoSOI	v5-d4	Beta
ANT	SiN (Lower Edge)	v5 (Jun 3 2023)	d0 (Jun 1 2023)	ANT_SiN	v5-d0-lower	Alpha
ANT	SiN (Upper Edge)	v5 (Jun 3 2023)	d0 (Jun 1 2023)	ANT_SiN	v5-d0-upper	Alpha

New models and foundries are regularly added. Usage may change. For additional foundry and process models, feel free to contact us or raise an issue.

Installation

Local

Install PreFab via pip:

pip install prefab

Or clone the repository and install in development mode:

git clone https://github.com/PreFab-Photonics/PreFab.git
cd PreFab
pip install -e .

Online

Use PreFab online through GitHub Codespaces:

Getting Started

Visit /examples for usage notebooks.

Performance and Usage

PreFab models are served via a serverless cloud platform. Please note:

• 🐢 CPU inferencing may result in slower performance. Future updates will introduce GPU inferencing.
• 🥶 The first prediction may take longer due to cold start server loading. Subsequent predictions will be faster.
• 😊 Be considerate of usage. Start small and limit usage during the initial stages. Thank you!

License

This project is licensed under the LGPL-2.1 license. © 2023 PreFab Photonics.