arXiv subjects

Aaditya Bhat

Publications and source records attributed to Aaditya Bhat.

3 recordsLinked to original sources

arXiv2024

Quantifying Implantation Induced Damage and Point Defects with Multislice Electron Ptychography

Here, we use multislice electron ptychography to quantify damage introduced by ion implantation of Er into 4H-SiC. Comparing reconstructed volumes from experiment (each 2,000 nm$^{3}$) along the implantation direction, the crystal damage is quantified and compared to pristine SiC. Using simulations, we establish that the implantation-induced static displacements limit both Er dopant and silicon vacancy detection. The corresponding damage in the experiment is found to occur up a depth of 100 nm and significantly deeper than expected from implantation simulations, ignoring crystallography. Beyond this depth, we show that silicon vacancies can be identified within the sampled volume and used to measure their local strain. Overall, these results underscore the power of multislice electron ptychography to quantify the impacts of implantation and as a tool to help guide electronic device process optimization.

cond-mat.mtrl-sci↗

arXiv2024

Sensitivity of Multislice Electron Ptychography to Point Defects: A Case Study in SiC

Here, we evaluate multislice electron ptychography as a tool to carry out depth-resolved atomic resolution characterization of point defects, using silicon carbide as a case study. Through multislice electron scattering simulations and multislice ptychographic reconstructions, we investigate the phase contrast arising from individual silicon vacancies, antisite defects, and a wide range of substitutional transition metal dopants (V\textsubscript{Si} to W\textsubscript{Si}) and potential detectability. Simulating defect types, positions, and microscope conditions, we show that isolated point defects can be located within a unit cell along the sample's depth. The influence of electron energy, dose, defocus, and convergence semi-angle is also explored to determine their role in governing defect contrast. These results guide experiments aiming to analyze point defects with multislice electron ptychography.

cond-mat.mtrl-sci↗

arXiv2023

Face Recognition in the age of CLIP & Billion image datasets

CLIP (Contrastive Language-Image Pre-training) models developed by OpenAI have achieved outstanding results on various image recognition and retrieval tasks, displaying strong zero-shot performance. This means that they are able to perform effectively on tasks for which they have not been explicitly trained. Inspired by the success of OpenAI CLIP, a new publicly available dataset called LAION-5B was collected which resulted in the development of open ViT-H/14, ViT-G/14 models that outperform the OpenAI L/14 model. The LAION-5B dataset also released an approximate nearest neighbor index, with a web interface for search & subset creation. In this paper, we evaluate the performance of various CLIP models as zero-shot face recognizers. Our findings show that CLIP models perform well on face recognition tasks, but increasing the size of the CLIP model does not necessarily lead to improved accuracy. Additionally, we investigate the robustness of CLIP models against data poisoning attacks by testing their performance on poisoned data. Through this analysis, we aim to understand the potential consequences and misuse of search engines built using CLIP models, which could potentially function as unintentional face recognition engines.

cs.CV↗