Electron Microscope Aberration Correction Using Ronchigram Transformers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manual tuning of electron-beam lenses in electron microscopes for aberration correction is tedious, time-consuming, and often non-optimal due to drifting processes, necessitating frequent adjustments.
Innovation Solution
An electronic controller employing a transformer with an autoregressive masked language model automatically corrects geometric aberrations in electron-beam optics using Ronchigram acquisition circuitry, guided by a trained alignment process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual tuning of electron-beam lenses is performed, then aberration correction can be achieved, but the process becomes tedious and time-consuming
Solution Approach 1:
The system uses an automated transformer model that independently performs aberration correction without human intervention. The model receives Ronchigram images, processes them through the trained neural network, and automatically determines optimal lens alignment parameters, enabling the system to self-correct aberrations without manual tuning
Solution Approach 2:
The patent replaces the manual mechanical adjustment process with an automated computational system. The transformer model, trained on reference Ronchigrams, substitutes human operators by automatically analyzing diffraction patterns and computing correction parameters through neural network inference
2Reliability
If manual tuning is performed frequently to offset drifting processes, then alignment accuracy is maintained, but productivity decreases
Solution Approach 1:
The system implements continuous feedback by repeatedly acquiring Ronchigram images and processing them through the transformer model. The model compares current Ronchigrams against the trained reference corpus, detects drift conditions, and automatically adjusts alignment parameters to maintain optimal performance without manual intervention
Solution Approach 2:
The automated system enables continuous aberration correction operations without interruption. The transformer model can process Ronchigrams in real-time, maintaining continuous monitoring and adjustment of lens alignment, eliminating the discontinuous nature of manual retuning sessions
3Measurement precision
If manual tuning is performed, then aberration correction is achieved, but the results are often non-optimal
Solution Approach 1:
The transformer model is pre-trained on a comprehensive corpus of reference Ronchigrams that span the alignment parameter space. This preliminary training phase captures optimal correction strategies, enabling the model to rapidly determine best corrections for new Ronchigrams without requiring operators to understand complex alignment procedures
Data Source
AI summary
Disclosed herein are scientific-instrument support systems, as well as related methods, apparatus, computing devices, and computer-readable media. In some embodiments, a support apparatus for a scientific instrument includes an interface device and a processing device. The interface device receives Ronchigrams acquired with the scientific instrument and transmits control signals for the electron-beam optics thereof. The processing device converts a measured Ronchigram into an input token for a transformer, produces an output token based on a tokenized sentence ending with the input token, and determines adjustments to the control signals based on the input and output tokens. The input and output tokens belong to a plurality of tokens representing reference Ronchigrams sampling an alignment parameter space of the electron-beam optics. The transformer implements an autoregressive masked language model trained on a corpus of reference sentences representing paths through the alignment parameter space to a target alignment state of the electron-beam optics.