The Virtual Nanofab
PreFab simulates photonic chip fabrication with foundry-accurate process models. Predict lithographic effects, etch bias, and process variation before tape-out. Eliminate design-manufacturing iteration loops and accelerate development with fab-aware optimization.
From Design to Fabrication Prediction
Most photonic designs fail on the first fab run. PreFab predicts how your design will actually look after manufacturing, capturing lithography, etching, and process variation from real foundry data.
Device Design
Fabrication Prediction
Chip View (Generated)
Three Lines to Prediction
Integrate PreFab into your existing Python workflow. Load your design, select your fab model, and get manufacturing predictions in seconds.
import prefab as pf
device = pf.read.from_gds("your_device.gds")
prediction = device.predict(model="your_fab")
Or Design Visually
Rosette is our browser-based photonic layout editor with built-in virtual nanofabrication. Design circuits visually and predict manufacturing outcomes without writing code.
Fabrication-Aware Design
Beyond prediction, PreFab integrates manufacturing constraints directly into your design workflow. Our differentiable models enable fabrication-aware optimization, from verifying optical performance with realistic geometries to inverse design that optimizes for post-fab outcomes.
Optical Simulation with Predicted Geometry
Use PreFab's predicted post-fab geometry in your optical solver to simulate realistic device performance. Capture fab-induced effects like broadened resonances, wavelength shifts, and loss before tape-out.
Ideal
Predicted
Pre-Compensated Designs
Automatically adjust your input geometry to compensate for known fabrication biases. PreFab's corrector model pre-distorts designs so the manufactured structure matches your target specification.
Design
Correction
Outcome
Certain Uncertainty
Our data-driven approach captures random device-to-device variations inherent in any manufacturing process. The fuzzy edges in our predictions highlight regions of uncertainty, giving you insight into potential yield and robustness before fabrication.
Design
Predicted Certainty
Beyond the Surface
PreFab predicts etch-induced effects like angled sidewalls and depth-dependent profiles. Capture the full 3D geometry of your fabricated structures for more accurate optical simulations.
Design
Predicted Sidewall
Inverse Design and Optimization
Optimize for post-fabrication performance using PreFab's differentiable models. Run gradient-based inverse design workflows that account for manufacturing constraints from the start, designing topologies for real-world outcomes.
Correction
Outcome
Python Native
Import PreFab directly into your existing Python workflows. Works seamlessly with most photonic design tools and simulation frameworks.
Lightning Fast
Get predictions in seconds for individual devices and minutes for full chip layouts. Fast enough to fit into iterative design loops.
Any Fab
Our approach adapts to most fabrication processes. Talk to us about developing models tailored to your process and technology.
Design with Manufacturing Reality
See how your photonic devices will actually fabricate.
Schedule a demo to explore PreFab.