CN115661136B - Semiconductor defect detection method for silicon carbide material - Google Patents
Semiconductor defect detection method for silicon carbide material Download PDFInfo
- Publication number
- CN115661136B CN115661136B CN202211587516.1A CN202211587516A CN115661136B CN 115661136 B CN115661136 B CN 115661136B CN 202211587516 A CN202211587516 A CN 202211587516A CN 115661136 B CN115661136 B CN 115661136B
- Authority
- CN
- China
- Prior art keywords
- area
- region
- pixel points
- sub
- gray
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Testing Or Measuring Of Semiconductors Or The Like (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention discloses a semiconductor defect detection method for a silicon carbide material, belonging to the technical field of image processing; the method comprises the following steps: acquiring a gray-scale image of the surface of a semiconductor wafer to be detected; dividing the gray scale image into a plurality of areas; acquiring the gray scale range of the abnormal pixel points in each area; screening out a plurality of subregions containing redundancy defects; acquiring the attachment degree of the abnormal pixel points in each sub-area to the area; obtaining the dispersion degree of abnormal pixel points in each sub-area; acquiring the influence degree of noise in each sub-area; acquiring a filtered and denoised image; and carrying out threshold segmentation according to the filtered and denoised image to obtain a defect region. According to the invention, the size of the Gaussian filter window is adaptively adjusted and the image is denoised according to the influence degree of the noise of each sub-region, and the defect region is obtained by performing threshold segmentation according to the denoised image, so that the accuracy of redundancy detection is effectively improved.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a semiconductor defect detection method for a silicon carbide material.
Background
In the manufacturing process of semiconductor wafers by using silicon carbide as a raw material, defects can be generated on the surfaces of the wafers especially by chemical vapor deposition, optical development and chemical mechanical grinding in a series of processes such as single crystal pulling, slicing, lapping, polishing, layer increasing, photoetching, doping, heat treatment, probing, scribing and the like. Among them, the unpatterned wafer and the patterned wafer are the two most common wafer types. The surface of the wafer mainly has defects including redundancy, crystal defects, mechanical damage (scratch pattern), and the like. The redundancy is a common defect type on the surface of the wafer, and mainly comprises nanometer-sized micro particles, micron-sized dust and residues of related processes.
The existing method mainly adopts image processing to detect the defects on the surface of the wafer, generally carries out denoising processing on the wafer image and then carries out defect detection, but because the redundant particles on the surface of the wafer are small, the influence degree of noise on the image is not considered when the wafer image is denoised, so that the edge of the image can be smoothed in the denoising process, or the defects can be filtered as the noise. Meanwhile, in the prior art, a fixed window is adopted to smooth some pixel points during denoising, so that the detail loss of an image is caused. Therefore, when defect detection is carried out, the detection is inaccurate because the characteristics are not obvious and are influenced by image noise.
Disclosure of Invention
In order to solve the defects in the prior art, the invention provides a method for detecting the semiconductor defects of silicon carbide materials, which obtains the influence degrees of the noises of different subregions by the difference of the characteristics of the noises and the defects of redundancy, particularly the difference of the adhesion degree of the defects of the redundancy and the noises to regions and the dispersion degree of the noises, then carries out self-adaptive adjustment on the size of a Gaussian filter window through the influence degree of the noises of each subregion to carry out denoising on the images, and finally carries out threshold segmentation according to the denoised images to obtain the defect regions, thereby effectively improving the accuracy of redundancy detection.
The invention aims to provide a method for detecting defects of a semiconductor made of silicon carbide, which comprises the following steps:
acquiring a gray-scale image of the surface of a semiconductor wafer to be detected; dividing the gray scale map into a plurality of regions; obtaining the chaos degree of pixel points in each region;
acquiring the gray scale range of the normal pixel points in each area according to the gray scale value of the pixel points in each area; acquiring the gray scale range of the abnormal pixel points in each area according to the gray scale range of the normal pixel points in each area;
screening a plurality of sub-regions containing the redundant defect from the plurality of regions according to the chaos degree of the pixel points in each region;
acquiring the attachment degree of the abnormal pixel points in each sub-area to the area according to the proportion of the pixel points corresponding to the gray scale range of the abnormal pixel points in each sub-area in the area and the chaos degree of the pixel points in each sub-area;
acquiring the uniformity of pixel distribution in each sub-area according to the gray value of the pixels in each sub-area and the total number of the pixels;
acquiring the dispersion degree of the abnormal pixel points in each sub-area according to the average distance between the pixel points corresponding to each gray value in the gray range of the abnormal pixel points in each sub-area and the distribution uniformity of the pixel points in each sub-area;
acquiring the influence degree of noise in each sub-area according to the attachment degree of the abnormal pixel points in each sub-area to the area and the dispersion degree of the abnormal pixel points in each sub-area;
acquiring the size of a filtering window of each sub-area according to the influence degree of noise in each sub-area;
performing Gaussian filtering on each corresponding subarea in the gray-scale image of the surface of the semiconductor wafer to be detected according to the size of the filtering window of each subarea to obtain a filtered and de-noised image;
and performing threshold segmentation according to the filtered and denoised image to obtain a defect region.
In an embodiment, the degree of confusion of the pixel points in each region is obtained according to the proportion of the pixel points corresponding to each gray value in each region in the region.
In an embodiment, the gray scale range of the normal pixel point in each region is obtained according to the mean value of the gray scale values in each region and the maximum gray scale value and the minimum gray scale value in the region.
In an embodiment, the lower limit calculation formula of the gray scale range of the normal pixel point in each region is as follows:
in the formula, a represents the lower limit value of the gray scale range of the normal pixel point in each area;
representing the r-th gray value in the region;
is shown as
A grey value;
representing a minimum gray value of the region;
the upper limit value calculation formula of the gray scale range of the normal pixel points in each area is as follows:
in the formula, b represents the upper limit value of the gray scale range of the normal pixel point in each area;
representing the r-th gray value in the region;
is shown as
A gray value;
representing the maximum gray scale value of the region;
that is, the gray scale range of the normal pixel points in each region is (a, b).
In one embodiment, the gray scale range of the abnormal pixel points in each region includes [0, a ] and [ b,255].
In an embodiment, the variance of the gray values of the pixels in each sub-region is used as the uniformity of the distribution of the pixels in each sub-region.
In one embodiment, the plurality of sub-regions containing redundancy defects are selected according to the following steps:
setting a chaos degree threshold value of each area;
when the chaos degree of pixel in every region is greater than the chaos degree threshold value of corresponding region, then this region is the subregion that contains the redundancy defect, sieves out a plurality of subregions that contain the redundancy defect from a plurality of regions in proper order.
In an embodiment, the entropy value of the gray value of the pixel point corresponding to the gray range of the normal pixel point in each region is used as the threshold value of the degree of confusion of the region.
The beneficial effects of the invention are: the invention provides a method for detecting defects of semiconductors made of silicon carbide, which comprises the steps of preliminarily screening out sub-regions containing redundant defects by obtaining the chaos degree of pixels on the surface of a wafer, and analyzing the regions containing the redundant defects to obtain the adhesion degree of abnormal pixel points in each sub-region to the regions and the dispersion degree of the abnormal pixel points in each sub-region; because the characteristics of the noise and the redundant object defect are different, specifically, the difference between the attachment degree of the redundant object defect and the noise to the region and the dispersion degree of the noise is realized, the influence degree of the noise in each sub-region is comprehensively expressed by combining the attachment degree and the dispersion degree, and the larger the influence degree is, the larger the window of the required filter is, the better denoising effect can be obtained. Therefore, the size of a Gaussian filter window is adaptively adjusted according to the influence degree of the noise of each sub-region to denoise the image, and finally, a defect region is obtained by performing threshold segmentation according to the denoised image, so that the accuracy of redundancy detection can be effectively improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic flow chart illustrating the general steps of a method for detecting defects in a silicon carbide semiconductor device according to an embodiment of the present invention;
fig. 2 is a view of a semiconductor wafer containing a redundancy defect.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
The invention mainly aims at detecting the defects of redundant objects on the surface of a semiconductor wafer produced by a silicon carbide material, and because the redundant object particles are tiny, the detection is not accurate because the characteristics are not obvious and the detection is influenced by image noise.
According to the method, the influence degrees of the noises of different sub-regions are obtained through the difference of the characteristics of the noises and the defects of the redundancy, particularly the difference of the adhesion degree of the defects of the redundancy and the noises to the regions and the dispersion degree of the noises, then the size of a Gaussian filter window is adaptively adjusted according to the influence degree of the noises of each sub-region to denoise the images, finally, the threshold segmentation is carried out according to the denoised images to obtain the defect regions, and the accuracy of detecting the redundancy can be effectively improved.
The invention provides a method for detecting defects of a semiconductor made of silicon carbide, which is shown in figure 1 and comprises the following steps:
s1, obtaining a gray-scale image of the surface of a semiconductor wafer to be detected; dividing the gray scale image into a plurality of areas; obtaining the chaos degree of pixel points in each region;
in this embodiment, when detecting the defect on the surface of the wafer, because the wafer is small, the high-resolution camera needs to be used to collect the image when collecting the image, so as to obtain the image with a clear defect. Referring to fig. 2, a schematic view of a wafer containing a redundancy defect can be clearly seen in fig. 2. And carrying out graying processing on the acquired wafer image to acquire a gray image of the surface of the semiconductor wafer to be detected.
It should be noted that, when the embodiment detects the redundant defect on the wafer surface, because noise is introduced due to the influence of ambient light or interference caused by internal components during the process of collecting the image, and the wafer particle is very small, the structural feature of the surface is also not obvious, so that when the defect detection is performed, the noise affects the accuracy of the detection to a great extent, and thus the image needs to be denoised.
In order to improve the detection accuracy, the wafer surface needs to be partitioned, and the size of the original image is assumed to be
In order to obtain an accurate feature distribution, the image is partitioned, since smaller image areas are more accurate when performing the calculations. The method for partitioning is equal-size partitioning, and the image is divided into n areas; each region having a size of
In the present embodiment, 16 is taken based on the historical experience n.
When the redundant object is attached to the surface of the wafer, part of the collected pixels on the surface of the wafer are lost, and the gray value of the pixels in a certain area is changed. The redundancy is characterized by small dispersed points in a bulk shape, and the original structure of the wafer can be shielded on the surface of the wafer, so that the abnormity of local pixels is evaluated according to the disorder degree of the pixels on the surface of the wafer. Therefore, in this embodiment, the degree of disorder of the pixels in each region is obtained according to the ratio of the pixels corresponding to each gray level in each region in the region. The calculation formula of the chaos degree of the pixel points in each region is as follows:
in the formula (I), the compound is shown in the specification,
expressing the chaos degree of pixel points in the c-th area;
indicates the second in the c-th region
The number of pixel points corresponding to each gray value;
representing the number of gray values in the c-th region;
is shown as
The occupation ratio of the pixel points corresponding to the gray values in the whole area;
and expressing the entropy value of the c region to express the overall disorder degree of the region. The more the gradation value distribution, the greater the disorder thereof. It should be noted that, when there is a defect of a redundant object or a noise, the gray level value on the surface of the wafer is changed, so that the pixel distribution is changed, and therefore, the degree of disorder of the gray level value of the pixels on the surface is increased. Therefore, the possibility that defects or noises may occur in a certain area is represented according to the chaos degree of the pixel points, and the local pixel points are evaluated.
S2, screening out a plurality of sub-regions containing redundant object defects;
acquiring the gray scale range of the normal pixel points in each area according to the gray scale value of the pixel points in each area; acquiring the gray scale range of the abnormal pixel points in each region according to the gray scale range of the normal pixel points in each region;
screening a plurality of sub-regions containing redundant object defects from the plurality of regions according to the disorder degree of the pixel points in each region;
it should be noted that, if the abnormality of the wafer surface is expressed according to the degree of disorder of the pixel points, a threshold needs to be selected to determine the degree of disorder. The noise and the redundant defect can change the gray value of the local area, so that the distribution of the gray value of the normal pixel point is concentrated according to the distribution of the normal pixel point in the whole area as a judgment standard, and the pixel points distributed near the gray average value of the whole area are all used as the normal pixel points. In this embodiment, the gray scale range of the normal pixel in each region is obtained according to the mean value of the gray scale values in each region and the maximum gray scale value and the minimum gray scale value in the region. Specifically, the lower limit calculation formula of the gray scale range of the normal pixel points in each region is as follows:
in the formula, a represents the lower limit value of the gray scale range of the normal pixel point in each area;
representing the r-th gray value in the region;
is shown as
A grey value;
representing a minimum gray value of the region;
expressing the mean value of the gray scale and the median value of the minimum gray scale;
the upper limit value calculation formula of the gray scale range of the normal pixel points in each area is as follows:
in the formula, b represents the upper limit value of the gray scale range of the normal pixel point in each area；
Representing the r-th gray value in the area;
is shown as
A grey value;
representing the maximum gray scale value of the region;
expressing the mean value of the gray scale and the median value of the maximum gray scale;
that is, the gray scale range of the normal pixel points in each region is (a, b). Therefore, the gray value of the pixel point in a certain interval represents the gray value of the normal pixel point, and the gray value of the normal pixel point contains most gray values of the normal pixel point.
In this embodiment, the remaining gray scale interval after the elimination of the gray scale range of the normal pixel corresponding to each region is the gray scale range of the abnormal pixel in each region, that is, the gray scale range (a, b) of the normal pixel in each region is eliminated in the whole gray scale range from 0 to 255, and specifically, the gray scale range of the abnormal pixel in each region includes [0, a ] and [ b,255].
In this embodiment, according to the degree of disorder of the pixels in each region, whether the region contains the redundancy defect is determined by using the entropy of the normal pixels as a threshold, and the region containing the redundancy defect is screened out, so that a plurality of sub-regions containing the redundancy defect are screened out according to the following steps:
setting a chaos degree threshold value of each region;
when the chaos degree of the pixel points in each region is larger than the chaos degree threshold of the corresponding region, the region is a sub-region containing the redundant defect, and a plurality of sub-regions containing the redundant defect are sequentially screened out from the plurality of regions; indicating that the region has a redundancy defect or has a large influence degree of noise;
when the chaos degree of the pixel points in each region is smaller than the chaos degree threshold of the corresponding region, no redundant object defect exists in the region, or the influence degree of noise is small.
And taking the entropy value of the gray value of the pixel point corresponding to the gray range of the normal pixel point in each region as the chaos degree threshold of the region.
It should be noted that the gray scale range of the abnormal pixel point in each region obtained in this embodiment also represents the gray scale range of the abnormal pixel point in each sub-region; the acquired chaos degree of the pixel points in each region also represents the chaos degree of the pixel points in each sub-region. And the obtained gray scale range of the normal pixel points in each region also represents the gray scale range of the normal pixel points in each sub-region.
S3, acquiring the attachment degree of the abnormal pixel points in each sub-area to the area;
acquiring the attachment degree of the abnormal pixel points in each sub-area to the area according to the proportion of the pixel points corresponding to the gray scale range of the abnormal pixel points in each sub-area in the area and the chaos degree of the pixel points in each sub-area;
it should be noted that the distribution states of the redundancy and the noise are relatively similar and are all dispersed small particles, so that the noise is regarded as the redundancy in visual observation, which affects the precision of defect detection. Because the redundant defects are distributed on the surface of the wafer in a scattered manner and the redundant objects are adsorbed on the surface of the wafer to a greater extent, part of the redundant objects are in an aggregated state and form clusters. The noise is distributed in a random manner, the loss of the image in a certain area is large due to the large noise in the area, the details of the image are changed, and the adsorption degree of the noise on the surface of the wafer is small. The adsorption degree indicates that the adsorption degree of the wafer surface differs depending on the degree of aggregation of a defect or noise. The redundant object defects are discrete compared with the texture of the wafer, and the distribution of the redundant object defects is irregular, so that the redundant object defects can be distinguished from the texture of the wafer when the disorder degree of the redundant object is calculated. In addition, the distribution of noise is scattered, so that the adhesion to the surface of the wafer is small, the concentration degree of the redundant defect is larger than that of the noise, and the adhesion to the surface of the wafer is large. Therefore, the attachment degree of the abnormal pixel points in the area to the area is calculated, and the noise or the redundant object defect is judged. The calculation formula of the attachment degree of the abnormal pixel point in each sub-area to the area is as follows:
in the formula (I), the compound is shown in the specification,
representing the attachment degree of the abnormal pixel point to the z-th sub-area;
representing the number of pixel points corresponding to the ith gray value in the z-th sub-region;
representing the size of the z-th sub-region;
expressing the number of all gray values corresponding to the gray range of the abnormal pixel points in the z-th sub-region;
b, calculating and obtaining according to the step S2; namely the remaining gray value of the gray value corresponding to the gray range of the normal pixel points in the sub-area.
Expressing the sum of the number of corresponding pixel points in the gray scale range of the abnormal pixel points;
and expressing the proportion of the pixel points corresponding to the gray scale range of the abnormal pixel points in the z-th sub-region in the region.
And representing the chaos degree of the pixel points in the z-th area. It should be noted that, in each sub-region, the degree of attachment of the abnormal pixel point in the sub-region to the region is expressed by multiplying the chaos degree of the whole pixel point by the ratio of the total number of the pixel points in the abnormal interval in the whole sub-region. The larger the proportion is, the larger the number of abnormal pixel points in the sub-area is, and the larger the disorder degree is, the larger the attachment degree of the abnormal pixel points to the area is. If the proportion of the abnormal pixel points is large, but the chaos degree is small, the obtained attachment degree is small, and the fact that the area possibly comprises a large number of self textures is shown.
Therefore, the attachment degree of the abnormal pixel point to the area is expressed according to the occupation ratio of the pixel points in the gray scale range of the abnormal pixel point and the disorder degree of the subarea. The distribution condition of abnormal pixel points with noise points or redundant defect in the sub-area in the area can be represented, and the more the distribution is, the larger the disorder degree is, and the larger the adhesion degree to the area is represented.
S4, obtaining the dispersion degree of the abnormal pixel points in each sub-area;
acquiring the uniformity of pixel point distribution in each sub-area according to the gray value of the pixel points in each sub-area and the total number of the pixel points; the variance of the gray value of the pixel points in each sub-area is used as the uniformity of the distribution of the pixel points in each area;
acquiring the dispersion degree of the abnormal pixel points in each sub-area according to the average distance between the pixel points corresponding to each gray value in the gray range of the abnormal pixel points in each sub-area and the distribution uniformity of the pixel points in each sub-area; it should be noted that, the average distance between the pixels corresponding to each gray value in the gray range of the abnormal pixel in each sub-region refers to an average value of euclidean distances between the pixels corresponding to each gray value in the gray range of the abnormal pixel.
In the present embodiment, the distribution of the abnormal pixels is represented by calculating the degree of attachment of the abnormal pixel points to the region in each sub-region. Calculating the aggregation state of the abnormal pixels in the sub-area, wherein if the aggregation degree is larger, the possible degree of representing noise is smaller, and the distribution of the noise is more uniform; the particles of the redundancy are large, and the aggregation degree is large, so that the proportion of noise and the proportion of the redundancy defects in each sub-area are distinguished by calculating the dispersion degree of abnormal pixel points in the sub-areas; specifically, the dispersion degree of the abnormal pixel points in each sub-area is obtained according to the average distance between the pixel points corresponding to the gray scale range of the abnormal pixel points in each sub-area and the uniformity of the pixel point distribution in each sub-area, and the calculation formula of the dispersion degree of the abnormal pixel points in each sub-area is as follows:
in the formula (I), the compound is shown in the specification,
is shown as
The discrete degree of abnormal pixel points in each sub-region;
denotes the first
The average distance between the pixel points corresponding to the jth gray value in the gray range of the abnormal pixel points in each sub-region;
expressing the number of all gray values corresponding to the gray range of the abnormal pixel points in the z-th sub-region;
b, calculating and obtaining according to the step S2;
denotes the first
The gray value of the kth pixel point in each sub-area;
denotes the first
The average of the gray values in the sub-regions,
denotes the first
The total number of pixel points in each sub-region;
denotes the first
Uniformity of pixel distribution in individual subregions, i.e. first
Representing the variance of the gray value of the pixel points in the sub-regions;
represents a natural constant; therefore, the aggregation degree of the abnormal pixel points is expressed according to the average distance and variance distributed among the abnormal pixel points in each sub-area; the smaller the average distance, the smaller the variance, and the greater the degree of dispersion.
S5, acquiring the influence degree of noise in each sub-area;
acquiring the influence degree of noise in each sub-area according to the attachment degree of the abnormal pixel points in each sub-area to the area and the dispersion degree of the abnormal pixel points in each sub-area;
in this embodiment, the degree of influence of noise in a certain region is comprehensively expressed by calculating the degree of attachment and the degree of dispersion of an abnormal pixel point to a sub-region. The calculation formula of the influence degree of the noise in each subarea is as follows:
in the formula (I), the compound is shown in the specification,
is shown as
The degree of influence of noise in the individual sub-regions;
representing the attachment degree of an abnormal pixel point to the z-th sub-area;
is shown as
The discrete degree of the abnormal pixel points in each sub-region;
represents a natural constant;
and a normalized value indicating the degree of adhesion, wherein a small degree of adhesion in the sub-region indicates a larger degree of influence of noise, and a smaller degree of dispersion indicates a smaller degree of influence of noise.
S6, obtaining the filtered and denoised image, and performing threshold segmentation according to the filtered and denoised image to obtain a defect region;
acquiring the size of a filtering window of each sub-region according to the influence degree of noise in each sub-region;
performing Gaussian filtering on each corresponding subarea in the gray-scale image of the surface of the semiconductor wafer to be detected according to the size of the filtering window of each subarea to obtain a filtered and de-noised image;
and carrying out threshold segmentation according to the filtered and denoised image to obtain a defect region.
In the present embodiment, the degree of influence of noise in each sub-area is calculated; the larger the influence degree is, the larger the window of the required filter is, and the better denoising effect can be obtained. Because the noise on the surface of the wafer is gaussian noise, gaussian filtering denoising is required, and the size of the window is as follows:
in the formula (I), the compound is shown in the specification,
is to show to
The window size of sub-region Gaussian filtering;
represents a natural constant;
denotes the first
The degree of influence of noise in the individual sub-regions;
the rounding function is expressed, and since the degree of influence is a normalized value, multiplication by 10 indicates the enlargement of the window size. For this purpose, the size of a Gaussian filtering window of each sub-region is sequentially obtained; performing Gaussian filtering on each corresponding subarea in the gray-scale image of the surface of the semiconductor wafer to be detected according to the size of the filtering window of each subarea to obtain a filtered and de-noised image;
and performing threshold segmentation according to the filtered and denoised image to obtain a defect region. Therefore, the redundant object defects can be more accurately detected.
The invention provides a method for detecting defects of semiconductors made of silicon carbide, which comprises the steps of preliminarily screening out sub-regions containing redundant defects by obtaining the chaos degree of pixels on the surface of a wafer, and analyzing the regions containing the redundant defects to obtain the adhesion degree of abnormal pixel points in each sub-region to the regions and the dispersion degree of the abnormal pixel points in each sub-region; because the characteristics of the noise and the redundant object defect are different, specifically, the difference between the attachment degree of the redundant object defect and the noise to the region and the dispersion degree of the noise is realized, the influence degree of the noise in each sub-region is comprehensively expressed by combining the attachment degree and the dispersion degree, and the larger the influence degree is, the larger the window of the required filter is, the better denoising effect can be obtained. Therefore, the size of a Gaussian filter window is adjusted in a self-adaptive mode according to the influence degree of noise of each sub-region to denoise an image, and finally threshold segmentation is carried out according to the denoised image to obtain a defect region, so that the accuracy of redundancy detection can be effectively improved.
The present invention is not limited to the above preferred embodiments, and any modifications, equivalent substitutions, improvements, etc. within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (8)
1. A semiconductor defect detection method of a silicon carbide material is characterized by comprising the following steps:
acquiring a gray-scale image of the surface of a semiconductor wafer to be detected; dividing the gray scale map into a plurality of regions; obtaining the chaos degree of pixel points in each region;
acquiring the gray scale range of the normal pixel points in each area according to the gray scale value of the pixel points in each area; acquiring the gray scale range of the abnormal pixel points in each region according to the gray scale range of the normal pixel points in each region;
screening a plurality of sub-regions containing redundant object defects from the plurality of regions according to the disorder degree of the pixel points in each region;
acquiring the attachment degree of the abnormal pixel points in each sub-area to the area according to the proportion of the pixel points corresponding to the gray scale range of the abnormal pixel points in each sub-area in the area and the chaos degree of the pixel points in each sub-area;
acquiring the uniformity of pixel point distribution in each sub-area according to the gray value of the pixel points in each sub-area and the total number of the pixel points;
acquiring the dispersion degree of the abnormal pixel points in each sub-area according to the average distance between the pixel points corresponding to each gray value in the gray range of the abnormal pixel points in each sub-area and the uniformity of the pixel point distribution in each sub-area;
acquiring the influence degree of noise in each sub-area according to the attachment degree of the abnormal pixel points in each sub-area to the area and the dispersion degree of the abnormal pixel points in each sub-area;
acquiring the size of a filtering window of each sub-area according to the influence degree of noise in each sub-area;
performing Gaussian filtering on each corresponding subarea in the gray-scale image of the surface of the semiconductor wafer to be detected according to the size of the filtering window of each subarea to obtain a filtered and de-noised image;
and carrying out threshold segmentation according to the filtered and denoised image to obtain a defect region.
2. The method of claim 1, wherein the degree of disorder of the pixels in each region is obtained based on the percentage of pixels in each region corresponding to each gray level value.
3. The method of claim 1, wherein the gray scale range of the normal pixels in each region is obtained according to the mean gray scale value of each region and the maximum gray scale value and the minimum gray scale value of the region.
4. The method of claim 3, wherein the lower limit of the gray scale range of the normal pixels in each region is calculated as follows:
in the formula, a represents the lower limit value of the gray scale range of the normal pixel points in each area;
representing the r-th gray value in the area;
is shown as
A grey value;
representing a minimum gray value of the region;
the upper limit value calculation formula of the gray scale range of the normal pixel points in each area is as follows:
in the formula, b represents the upper limit value of the gray scale range of the normal pixel points in each area;
representing the r-th gray value in the region;
is shown as
A gray value;
representing the maximum gray value of the region;
that is, the gray scale range of the normal pixel points in each region is (a, b).
5. The method of detecting defects in a silicon carbide semiconductor device as claimed in claim 4, wherein the gray scale range of the abnormal pixel points in each region includes [0, a ] and [ b,255].
6. The method of claim 1, wherein the distribution uniformity of the gray-level values of the pixels in each sub-area is determined as the variance of the gray-level values of the pixels in each sub-area.
7. The method of detecting defects in a silicon carbide semiconductor device according to claim 1, wherein the sub-regions containing redundant defects are selected by:
setting a chaos degree threshold value of each area;
when the chaos degree of pixel in every region is greater than the chaos degree threshold value of corresponding region, then this region is the subregion that contains the redundancy defect, sieves out a plurality of subregions that contain the redundancy defect from a plurality of regions in proper order.
8. The method of claim 7, wherein entropy of gray level values of pixels corresponding to gray level ranges of normal pixels in each region is used as a threshold of disorder level in the region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211587516.1A CN115661136B (en) | 2022-12-12 | 2022-12-12 | Semiconductor defect detection method for silicon carbide material |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211587516.1A CN115661136B (en) | 2022-12-12 | 2022-12-12 | Semiconductor defect detection method for silicon carbide material |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115661136A CN115661136A (en) | 2023-01-31 |
| CN115661136B true CN115661136B (en) | 2023-03-07 |
Family
ID=85018266
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211587516.1A Active CN115661136B (en) | 2022-12-12 | 2022-12-12 | Semiconductor defect detection method for silicon carbide material |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115661136B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116229276B (en) * | 2023-05-05 | 2023-07-28 | 生态环境部华南环境科学研究所(生态环境部生态环境应急研究所) | River entering pollution discharge detection method based on computer vision |
| CN117094998B (en) * | 2023-10-19 | 2024-02-02 | 厘壮信息科技(苏州)有限公司 | Defect detection method in monocrystalline silicon wafer production process |
| CN117173168B (en) * | 2023-11-02 | 2024-02-23 | 深圳福禄山科技有限公司 | OLED display screen quality detection method based on artificial intelligence |
| CN117649412B (en) * | 2024-01-30 | 2024-04-09 | 山东海天七彩建材有限公司 | Aluminum material surface quality detection method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115311292A (en) * | 2022-10-12 | 2022-11-08 | 南通创铭伊诺机械有限公司 | Strip steel surface defect detection method and system based on image processing |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6614924B1 (en) * | 1999-08-02 | 2003-09-02 | Applied Materials, Inc. | Adaptive mask technique for defect inspection |
| JP5498189B2 (en) * | 2010-02-08 | 2014-05-21 | 株式会社日立ハイテクノロジーズ | Defect inspection method and apparatus |
| DE102010061505B4 (en) * | 2010-12-22 | 2012-10-31 | Kla-Tencor Mie Gmbh | Method for inspection and detection of defects on surfaces of disc-shaped objects |
| TWI603074B (en) * | 2015-07-03 | 2017-10-21 | 元智大學 | Optical film defect detection method and system thereof |
| CN117893817A (en) * | 2021-04-30 | 2024-04-16 | 上海众壹云计算科技有限公司 | Wafer defect classification method, device and system thereof, electronic equipment and storage medium |
| CN115018828B (en) * | 2022-08-03 | 2022-10-25 | 深圳市尹泰明电子有限公司 | Defect detection method for electronic component |
| CN115063413B (en) * | 2022-08-04 | 2022-11-11 | 宁波鑫芯微电子科技有限公司 | Feature extraction method for abnormal data of super-large-scale wafer |
| CN115330784B (en) * | 2022-10-13 | 2022-12-23 | 南通金百福纺织品有限公司 | Cloth surface defect detection method |
-
2022
- 2022-12-12 CN CN202211587516.1A patent/CN115661136B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115311292A (en) * | 2022-10-12 | 2022-11-08 | 南通创铭伊诺机械有限公司 | Strip steel surface defect detection method and system based on image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115661136A (en) | 2023-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115661136B (en) | Semiconductor defect detection method for silicon carbide material | |
| JP6312370B2 (en) | System and method for detecting, classifying and quantifying wafer surface features with a wafer geometry metrology tool | |
| WO2022110804A1 (en) | Image noise measurement method and device based on local statistical information | |
| US10360671B2 (en) | Tool health monitoring and matching | |
| Cho et al. | Development of real-time vision-based fabric inspection system | |
| JP6598790B2 (en) | Wafer and lot-based tiering methods that combine customized metrics with global classification methods to monitor process tool status with extreme throughput | |
| CN115330784B (en) | Cloth surface defect detection method | |
| CN115035122B (en) | Artificial intelligence-based integrated circuit wafer surface defect detection method | |
| CN115020267B (en) | Semiconductor surface defect detection method | |
| JP6265608B2 (en) | Advanced site-based nanotopography system and method for wafer surface metrology | |
| CN115775250B (en) | Golden finger circuit board defect rapid detection system based on digital image analysis | |
| WO2010093733A2 (en) | Detecting defects on a wafer | |
| WO2018175669A1 (en) | Dynamic care areas for defect detection | |
| JP2017515097A (en) | Automatic in-line inspection and measurement using projected images | |
| JP2007501944A (en) | Apparatus and method for enabling robust separation between signals of interest and noise | |
| JP2007501944A6 (en) | Apparatus and method for enabling robust separation between signals of interest and noise | |
| CN115880284B (en) | Detection method and system for uniformity of lubricating oil abrasive particles | |
| CN115063413B (en) | Feature extraction method for abnormal data of super-large-scale wafer | |
| CN114677376B (en) | Method and system for testing defects in production of woven belts based on illumination pretreatment | |
| CN115953407B (en) | Semiconductor equipment maintenance system based on computer vision | |
| CN117437219B (en) | Magnetic glue inductance irregular appearance defect detection method based on computer vision | |
| CN116309416A (en) | Visual inspection method for surface of part | |
| CN117649412B (en) | Aluminum material surface quality detection method | |
| KR20110040866A (en) | Critical flaw detecting method | |
| CN117350988A (en) | Method, device, medium and system for detecting scratch defects on wafer surface |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |