Quantum Device Screening via Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for screening quantum devices are inefficient and time-consuming, often relying on electrical testing at later stages or subjective visual inspections, which fail to identify inconsistencies in quantum devices such as Josephson junctions that may not use semiconductor materials, leading to reduced yield and increased costs due to malfunctions like shorts and opens.
Innovation Solution
A computer-based system employing a feature machine learning model and classification neural network to analyze images of quantum devices during fabrication, identifying features and characteristics that indicate whether the devices will function as desired, allowing for timely rework and improving yield by detecting issues like particle presence or misalignment before final assembly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If electrical testing is used to screen quantum devices, then functional performance can be verified, but the screening process becomes time-consuming and occurs at later stages
Solution Approach 1:
The patent applies preliminary action by performing optical inspection of quantum devices during the fabrication process before electrical testing is conducted. The system captures images of quantum devices at intermediate fabrication stages and uses machine learning models to identify potential defects such as particle contamination, lithographic misalignment, and dimensional variations. This early detection enables timely rework or adjustment of fabrication parameters, preventing defective devices from proceeding to later stages and reducing overall screening time while maintaining functional performance verification.
2Loss of time
If visual inspection is used to screen quantum devices, then screening can be performed earlier, but the method is subjective and fails to identify all inconsistencies
Solution Approach 1:
The patent replaces the subjective mechanical visual inspection system with an automated optical inspection system using machine learning. The system captures high-resolution images of quantum devices and applies trained machine learning models that objectively analyze features such as particle contamination, lithographic misalignment, lateral dimension variations, and purity variations. This substitution eliminates human subjectivity and improves measurement precision while maintaining early screening capability during fabrication processes.
Solution Approach 2:
The patent creates a digital copy of the quantum device through optical imaging, allowing the machine learning system to analyze the copied image data without physically handling or disturbing the actual device. The machine learning models process the visual information from the copied images to identify defects and predict functional performance, replacing direct human visual inspection with automated analysis of digital replicas.
3Device complexity
If traditional screening methods are used, then manufacturing process is simpler, but yield is reduced due to undetected defects
Solution Approach 1:
The patent implements preliminary action by integrating optical inspection and machine learning analysis into the fabrication process at intermediate stages. The system identifies defects such as particle contamination, lithographic misalignment, and dimensional variations early in the manufacturing process, enabling timely corrective actions. This prevents defective quantum devices from proceeding through subsequent fabrication steps, thereby reducing waste and improving manufacturing yield without requiring complete redesign of the manufacturing process.
4Device complexity
If screening is performed at later stages, then fewer process steps are needed, but rework opportunities are lost
Solution Approach 1:
The patent applies preliminary action by performing optical inspection and defect identification during intermediate fabrication stages rather than at final testing stages. The machine learning system analyzes images of quantum devices to detect defects such as particle contamination, lithographic misalignment, and dimensional variations while the devices are still in the fabrication process. This early detection provides timely feedback that enables rework or adjustment of fabrication parameters before subsequent process steps are completed, maximizing repair opportunities and reducing waste.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method, apparatus, system, and computer program product (822) for screening quantum devices (134, 206). A computer system (216) sends features (222) extracted from an image of a quantum device into a classification neural network (144, 302). The classification neural network (144, 302) is configured to identify a set of characteristics (140, 224) for the quantum device from a group of mutually exclusive characteristics (226) based on the features (222) identified in the image of the quantum device and output the set of characteristics (140, 224) identified for the quantum device. The computer system (216) receives the set of characteristics (140, 224) identified by the classification neural network (144, 302) for the quantum devices (134, 206). The set of characteristics (140, 224) indicates whether the quantum devices (134, 206) will function as desired.