CN-121981953-A - Automatic film thickness measuring method, system and medium based on TEM image

CN121981953ACN 121981953 ACN121981953 ACN 121981953ACN-121981953-A

Abstract

The invention discloses a film thickness automatic measurement method, a measurement system and a medium based on a TEM image, wherein the method comprises the following steps of image preprocessing, contour recognition and rotation correction; selecting a region, obtaining a cut rectangular image, analyzing signals, extracting an average gray level curve of the film growth direction, judging the upper and lower boundaries of each film and a film interlayer diffusion interface based on gray level distribution characteristics and threshold segmentation, calculating thickness, calculating and outputting the thickness of each film and each diffusion layer according to the determined boundary position, and visualizing to generate a film thickness distribution line graph and a film thickness labeling image. The invention is based on taking MRC format image output by film cross section by transmission electron microscope, self-adapting level correcting picture by interactive software with interface, automatically calculating and outputting thickness of each film and diffusion layer between films, avoiding repeated coordinate recording and repeated operation of difference value operation in traditional measuring process, avoiding deviation of threshold value selection of each period in manual measurement, obviously reducing measuring time of multilayer film thickness, improving degree of automation of measurement and consistency and repeatability of measuring result.

Inventors

GAO KEJIA
ZHU MEIPING
LI JINGPING
SHAO JIANDA

Assignees

中国科学院上海光学精密机械研究所

Dates

Publication Date: 20260505
Application Date: 20251225

Claims (8)

1. An automatic film thickness measuring method based on a TEM image, which is characterized by comprising the following steps: Step S1, image preprocessing and self-adaptive level correction: the TEM original image data of the film is read, and contrast enhancement and denoising treatment are carried out to enhance the detectability of the boundary of the film layer; Identifying each film layer region in the TEM image by adopting a contour extraction method based on minimum area and length-width ratio screening; calculating the principal axis direction angle of each film contour through principal component analysis, and taking an average value after removing extreme values to obtain the average inclination angle of the whole image; performing rotation correction on the original image according to the average inclination angle to automatically realize horizontal alignment of all film boundaries; step S2, intelligent region clipping: Providing a rectangle region selection mechanism combining automatic and manual operation on the horizontally corrected image, wherein the automatic cutting adopts a maximum axis pair Ji Najie rectangle algorithm to ensure that the cutting region completely covers all effective film layer structures; Step S3, sub-pixel precision boundary judgment and diffusion layer quantization: Calculating pixel gray average values line by line for the rectangular area to generate a one-dimensional gray change curve; Smoothing the curve by adopting Gaussian filtering, and detecting all local peaks and troughs by a peak searching algorithm; sub-pixel interpolation is carried out on the curve; Carrying out normalization processing on the curve, and mapping the curve to a [0,1] interval, wherein the wave peak value is 1, and the wave trough value is 0; Searching from the positions of each wave crest and each wave trough to two sides based on a preset threshold value, and determining a point closest to the threshold value as a sub-pixel level locating point of the upper boundary and the lower boundary of the film layer, wherein the sub-pixel level locating point comprises a wave crest left threshold value point, a wave crest right threshold value point, a wave trough left threshold value point and a wave trough right threshold value point; Step S4, calculating the thickness of the integrated multilayer structure and the diffusion layer: (1) Calculating the pixel thickness of each light-colored film layer, wherein the pixel thickness is equal to the right side threshold point abscissa of the corresponding wave crest minus the left side threshold point abscissa; (2) Calculating the pixel thickness of each dark color film layer, wherein the pixel thickness is equal to the right side threshold value point abscissa of the corresponding trough minus the left side threshold value point abscissa; (3) Calculating a diffusion layer pixel thickness between adjacent film layers, comprising: trough-crest diffusion layer thickness is equal to left side threshold point abscissa of crest minus right side threshold point abscissa of previous trough; the thickness of the wave crest-wave trough diffusion layer is equal to the left side threshold point abscissa of the wave trough minus the right side threshold point abscissa of the previous wave crest; All pixel thicknesses are converted to nanoscale physical thicknesses in combination with the unit pixel actual size extracted from the TEM image file header. Step S5, multi-level interactive visualization: Generating and outputting structured data containing the actual thickness of all the film layers and the diffusion layers; the thickness distribution of each layer is displayed in a line diagram mode, the horizontal axis is a film layer serial number, and the vertical axis is a thickness value; and (2) directly superposing and labeling the calculated actual thickness values of all layers on the rectangular area image selected in the step (S2) to form a thickness labeling image, and simultaneously providing a step-by-step visual interface, wherein clicking a button can pop up related images, wherein the related images comprise an original image and a contour recognition result, a horizontal correction and cutting area, a gray level curve and a normalization curve thereof and a visual verification of boundary positioning points.
2. The method according to claim 1, wherein in step S1, a contour extraction method based on morphological screening is adopted, and by setting a minimum pixel area threshold, a minimum aspect ratio threshold and a maximum roundness threshold, interference contours of non-film regions are eliminated, so that accuracy of contour extraction is ensured.
3. The method according to claim 1, wherein the peak-to-valley filtering mechanism based on the relaxation threshold is adopted in the peak-finding algorithm in the step S3, and by setting the minimum peak height and minimum peak-to-valley distance parameters, the false peak-to-valley detection caused by noise is effectively suppressed, and the robustness of boundary determination is ensured.
4. The method for automatically measuring the film thickness based on the TEM image according to claim 1, wherein the normalization process in step S3 adopts a piecewise linear interpolation method, wherein the original gray scale value is linearly mapped to a [0,1] interval between adjacent peaks and troughs, the trough position is mapped to 0, and the peak position is mapped to 1, so that the normalization process of the gray scale curve is realized.
5. The method according to claim 1, wherein the threshold searching in step S3 is performed by a two-way nearest neighbor algorithm, wherein starting from each peak or trough position, searching in the left and right directions along the gray curve, and finding the point of the first gray value closest to the preset threshold as the boundary point, so as to achieve accurate definition of the boundary of the diffusion layer.
6. The method according to claim 1, wherein the automatic cropping algorithm in step S2 further comprises a validity verification mechanism for automatically determining whether the cropping area contains an effective film structure by detecting whether a complete peak-trough periodic sequence exists in the cropping area, and prompting reselection or adjustment of parameters if the condition is not satisfied.
7. An automated film thickness measurement system based on TEM images, comprising: The image preprocessing module is used for executing the functions of image enhancement, contour extraction and self-adaptive horizontal correction; The intelligent clipping module is used for executing automatic and manual region selection functions and comprises a validity verification mechanism; the boundary analysis module is used for executing sub-pixel level boundary judgment and diffusion layer quantization functions, and comprises Gaussian smoothing, sub-pixel interpolation, normalization processing and a two-way threshold search algorithm; the thickness calculation module is used for executing layering thickness calculation and unit conversion functions; and the visualization and interaction module is used for executing multi-level visualization display.
8. A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the TEM image based film thickness automatic measurement method as claimed in any one of claims 1 to 7.

Description

Automatic film thickness measuring method, system and medium based on TEM image Technical Field The invention belongs to the technical field of film measurement and image analysis, and particularly relates to a film thickness automatic measurement method, a film thickness automatic measurement system and a film thickness automatic measurement medium based on a transmission electron microscope (Transmission Electron Microscope, TEM) image, which are suitable for measuring film thickness based on a film picture shot by the transmission electron microscope, and can be widely applied to the fields of film coating technology, measurement technology, optical instruments and the like. Background The film material is widely applied to the fields of optical devices, electronic devices and energy sources, and the thickness precision of the film material directly influences the performance and stability of the devices. For example, in the field of optical coating, precise control of film thickness determines optical characteristics such as reflectivity, transmittance, center wavelength, and bandwidth, and in semiconductor and new energy materials, precise control of film thickness affects characteristics such as conductivity, insulation, and device lifetime. Therefore, the thickness of the film can be measured efficiently and accurately, and effective guidance can be provided for the optimization of the film preparation process, so that the preparation of the high-performance film is realized. The existing film thickness measuring method mainly comprises cross section transmission electron microscope measurement, ellipsometer measurement, X-ray measurement and the like. Ellipsometers and X-ray methods generally rely on complex optical models or diffraction models, are sensitive to thin film material characteristics, have high requirements on sample preparation and testing conditions, are sensitive to film structures, and have limited application range. TEM images can directly provide the cross-sectional morphology of thin films, and are an important means of studying multilayer film structures. However, the conventional TEM measurement process relies on a large number of repeated manual operations, and the problems of a large number of repeated operations, low efficiency, strong subjectivity, inconsistent results, difficulty in reproduction of measurement results, non-uniformity in threshold selection of each period and the like exist in that the film thickness is manually calculated one by naked eyes to identify and record coordinates. With the development of thin film devices to multilayering, nanocrystallization and functional integration, the structure of the thin film layer is more and more complex, and higher requirements are put on the automation and the efficiency of thickness measurement. The existing manual or semi-automatic measurement method can not meet the requirement of mass reproducible film thickness analysis. Therefore, developing a measurement method capable of automatically processing a TEM image, intelligently identifying a film layer and an interface, and rapidly outputting information of each layer thickness and diffusion layer is a technical problem to be solved in the field. Disclosure of Invention The invention aims to overcome the defects of the prior art and provides a film thickness automatic measurement method, a film thickness automatic measurement system and a film thickness automatic measurement medium based on a TEM image. The invention automatically calculates and outputs the thickness of each layer of film by interactive software with an interface based on the MRC format image output by shooting the cross section of the film by using a transmission electron microscope, thereby avoiding repeated operation in the traditional TEM measuring process, obviously reducing the measuring time of the thickness of the multilayer film and ensuring that the measurement has reproducibility. The technical scheme of the invention is as follows: The automatic film thickness measuring method based on the TEM image is characterized by comprising the following steps of: Step S1, image preprocessing and self-adaptive level correction: the TEM original image data of the film is read, and contrast enhancement and denoising treatment are carried out to enhance the detectability of the boundary of the film layer; Identifying each film layer region in the TEM image by adopting a contour extraction method based on minimum area and length-width ratio screening; calculating the principal axis direction angle of each film contour through principal component analysis, and taking an average value after removing extreme values to obtain the average inclination angle of the whole image; performing rotation correction on the original image according to the average inclination angle to automatically realize horizontal alignment of all film boundaries; step S2, intelligent region clipping: Providing a rectangle region s