Neuronale Graphnetzwerke zur Ausnutzung Räumlicher und Semantischer Beziehungen in Metrologiedaten

Details

Veröffentlichungs-Nr.

EP4752630

Anmeldetag

2. Dezember 2024

Veröffentlichung

3. Juni 2026

Rechtsraum

EP

IPC

G03F7/00

Abstract

A method of using a graph neural network to process metrology data of a patterned portion of a substrate patterned using a lithographic apparatus, to generate characterization information characterizing the patterned portion of the substrate. The method comprises obtaining metrology data, defining a graph of a graph neural network, defining feature data for each node of the graph based on the metrology data, and at least once generating, using the graph neural network and for each node, corresponding message data for each node associated with the node and updating, for each node, the feature data for the node based on the corresponding one or more message data, to form updated feature data. The method further comprises generating the characterization information from the updated feature data.

Anmelder

Firma

ASML Netherlands B.V.

Land

🇳🇱 Niederlande