Automate semiconductor process control with software
Amfibian connects model development, Digital Twin validation and production operation in a single flow. It reduces process variability and automates the tuning and operating work engineers used to repeat by hand.
Stop rebuilding process control from scratch.
Amfibian comes with the capabilities you need for modeling, validation, deployment, and production operations. Configure it for your processes and equipment, validate in a Digital Twin, and start operating.
Each change in process or equipment means repeating much of the same development and validation work.
Start with a ready-built product, then reuse proven methods for your next products and equipment.
Equipment keeps getting more capable. Process control is still tuned by hand.
A model is only the beginning.
It has to keep running in production.
Build the process relationship and
control model.
Use process context and data to configure the models needed for VM, APC and Profile Control.
Validate before putting it on the line.
Check expected outcomes and constraints in a Digital Twin.
Run validated control in production.
Connect production data and apply control under the customer's approval rules.
Reuse the validated method
for the next product or tool.
Reuse process context, validation criteria and operating methods for the next product or tool.
After deployment, your engineers should be able to run it themselves.
“Do we have to call the vendor back every time we add a new tool or process?”
Customer engineers can use validated model structures and settings for new products, tools and processes.
“Can we diagnose and respond to problems ourselves?”
Engineers can review model state, key drivers and operating history to narrow down issues and respond on site.
“Where do our production data and models stay?”
Production data, models and operating settings can be managed in the customer's on-premises environment.
Reduce process variability and repetitive tuning together.
Lower process variability
Feed prediction and control results into the next run to reduce variation around the target.
Less repetitive tuning
Handle state checks and condition adjustments in the system to reduce repeated work.
Faster rollout to new products and tools
Use validated methods as the starting point for new products and tools.
Results from prior production operation by the team
* Based on the team's prior semiconductor production operating experience.
Built for manufacturing processes that need precise control.
Semiconductor Manufacturing
Manage variation across deposition, etch, CMP, thermal and other complex wafer processes.
Power & Compound Semiconductors
Control process outcomes and tool-to-tool and wafer-to-wafer variation across SiC, GaN and other material systems.
Photonics & MEMS
Predict and control uniformity across epitaxy, thin films, etch and microstructure processes.
Solar & Display
Manage film properties and spatial variation in large-area and high-throughput production.
Go deeper into process control.
What is the difference between
a fab that uses AI
and a fab that runs on AI?
What else is required to keep a model running in production after it has already shown good performance in a PoC?
Read the article →Questions customers ask
before deployment.
Does AI change process conditions immediately?
Recommended values and expected outcomes are reviewed first, then applied through the customer's own approval process. The customer also decides the scope of automation.
Do we have to replace our MES, FDC or APC?
No. Existing manufacturing systems stay as they are. Amfibian connects the model development, validation and operation features you need to the data and operating environment you already have.
Do our data or models need to stay connected to an external service?
Amfibian can be deployed on-premises so production data, models and operating settings remain in the customer's environment.
Let's see how far your current process
can be automated.
We will review the operating method, equipment conditions, applicable control scope and validation approach.