Catch up on our latest webinar for users of electron microscopy in the semicon industry. In this webinar we covered the latest in subsampling & image processing, and saw how CNR-IMM used SenseAI to investigate the Projected Electrical Properties of 2-dimensional Electron Gas using 4-D STEM. Questions and answers from the webinar are also summarised below the video. Thank you for watching!
Watch: Superfast, Low-dose Electron Microscopy in the Semicon Industry
Q&A Summary from the live webinar
What is the SNR increase of our method compared to low-pass Gaussian filtering?
AR: No loss of high frequency information as seen with G. filtering. Dictionary retains high and low freq. components. We’ve demonstrated a two-fold increase in resolution using super resolution methods.
NB: Filtering is post acquisition. Filtering a damaged image just gives a filtered damaged image. Using subsampling, we avoid the initial damage to increase the SNR and acquire data faster.
How does the method affect the accuracy of quantitative EDS maps, when quantification is done using the Cliff-Lorimer method where the error will be >10% of the SNR.
NB: As our method provides with a equivalently higher SNR acquisition, quantification should be approximately the same with lower quality fully sampled data, when damage or speed are a concern. Subsampling/inpainting provides a way of computationally increasing the effective SNR.
What are semicon use cases for low-dose, superfast techniques?
GN: Depends on the specimen. For industry specimens, beam sensitive materials require very low dose. 4D STEM is useful for these samples but the datasets are very large and acquisition times are very long. These techniques allow us to acquire high quality data using 4D STEM while being compressed enough to be easily handled. Data being mobile is important to keep costs down. Same information in less memory is important.
What speed increase do you provide?
DN: We’ve demonstration a 100x speed increase and 400x compression rate for 4D STEM acquisitions. For general applications, we see 10x speed and compression rates across the board.
For any questions relating to the webinar or our low dose, superfast imaging please email info@senseai.vision.
