The Virtual Nanofab

PreFab simulates photonic chip fabrication with foundry-accurate process models. Predict lithographic effects and process variation before tape-out. Eliminate design-manufacturing iteration loops and accelerate development with fab-aware optimization.

From Design to Fabrication Prediction

Predict how your photonic designs will change during fabrication. PreFab uses real foundry data to model process effects and variation, helping you identify manufacturing risks before tape-out.

See how it works →

Grating device design
Device Design
Grating fabrication prediction results
Fabrication Prediction
Grating SEM view
Chip View (Generated)

Three Lines to Prediction

Load a GDS cell, choose a fabrication model, and predict its geometry. Use the results in your existing Python analysis and simulation.

import prefab as pf

device = pf.read.from_gds("chip.gds", "TOP")
prediction = device.predict(pf.models["SOI"])
Rosette logo

Design with Agents

Rosette pairs a Python API and layout checks with context for coding agents. Review and refine their designs in a GPU-powered editor.

Fabrication-Aware Design

Verify, pre-compensate, and optimize photonic designs for the chip you’ll actually fabricate.

Learn more →

Simulate as Fabricated

Simulate predicted geometries in your optical solver to see how fabrication affects resonances, loss, and performance. Compare against the ideal layout before tape-out.

Ideal
Predicted

Pre-Compensated Designs

PreFab pre-distorts layouts to counter predictable effects such as erosion and dilation. Improve how existing designs fabricate without starting a full redesign.

Target device
Design
Corrected design
Correction
Fabrication outcome
Outcome

Uncertainty, Made Visible

Fuzzy prediction edges highlight potential device-to-device variation. Identify sensitive features and explore how uncertainty could affect optical performance.

Offset grating device design
Design
Predicted fabrication variation
Predicted Variation

Fabrication-Aware Inverse Design

Put differentiable fabrication models inside your optimization loop. Optimize expected post-fab performance, rather than a geometry that only works in simulation.

Correction
Outcome

Python Native

Work with GDS, images, and arrays in the Python tools you already use.

Rapid Iteration

Check fabrication effects during design, not after your first fab run.

Your Process

Custom models use your foundry data to reflect your fabrication process.

Design with Fabrication in Mind

Tell us about your devices, fabrication process, and design goals. We’ll help you find where PreFab fits in your workflow.