Font Size: a A A

Preparation Of Chromium Ditelluride Nanoporous Membranes And Their Structural Evolution Analysis

Posted on:2024-07-02Degree:MasterType:Thesis
Country:ChinaCandidate:Y H SunFull Text:PDF
GTID:2531307052996419Subject:Electronic information
Abstract/Summary:PDF Full Text Request
In recent years,with the advent of graphene,two-dimensional materials with special structures and rich properties have been synthesized successively.Transmission electron microscopy has a high spatial resolution and a rich characterization mode,and it is an important research tool in deeply analyzing the relationship between the performance and the structure of the 2D material.In situ TEM is a technique that enables in situ processing and collects images in real time.However,the in situ TEM image dataset has a small number of samples,single image tone and minimal foreground region.In particular,for the nanostructures of thin layer2 D materials,especially in thin layers,the TEM images are difficult to distinguish with the background.Regular convolutional neural network model is difficult to accurately identify in-situ TEM images,and the manual annotation of foreground regions has low efficiency and poor accuracy,which limits the further exploration of all aspects of two-dimensional materials.Based on the U-Net neural network model,the identification of the two-dimensional material in-situ TEM image data set is conducted based on the preparation of the novel two-dimensional chromium ditelluride nanopore thin film samples.The main research results obtained are as follows:(1)Preparation of chromium ditelluride nanopore film and acquisition of transmission electron microscope images.Chchromium ditelluride film samples were prepared based on micromechanical stripping with the thermal annealing process and the adhesion of chromium ditelluride samples to the tape.The process of making the chromium ditelluride film is simple and the structure can be improved.Prepare chromium ditelluride film sample transferred to the carrier and put into the in situ transmission electron mirror,with the help of in situ transmission electron mirror,using high-energy electron beam bombardment chromium ditelluride film sample surface,form the nanometer scale chromium ditelluride nanopore,finishing image data,get chromium ditelluride nanopore structure in situ transmission electron mirror image data set.(2)U-Net neural network model was used to segment the chromium ditelluride structure.First,the data set was divided into training set,validation set and test set,the U-Net convolutional neural network model was built,the image dataset was preprocessed,the data set was imported into the model for training,and finally the better segmentation results of coincidence degree over 90%(Iou,Intersection over Union)were obtained.According to the segmentation results of the evolution process of the image,it is found that when the irradiation time is increasing.In conclusion,the thin layer PTTE sample was prepared by micromechanical stripping;then in situ TEM was used to obtain the sample in U-Net convolutional neural network model,and the accurate segmentation of foreground and background was achieved,thus further promoting the application of convolutional neural network in the efficient analysis of in situ TEM images.
Keywords/Search Tags:Transmission electron microscopy, Transmission electron microscopy preparation methods, in situ, U-NET, Nanopore
PDF Full Text Request
Related items