A method for automatic determination of silicon wafer particle stacking based on algorithms

Abstract

This invention relates to an algorithm-based method for automatically determining the particle stacking pattern of silicon wafers, belonging to the field of silicon wafer inspection technology. It uses an SPX machine to test and stack 12-inch silicon wafers, and includes the following steps: Step 1: Divide the wafer into four quadrants (first, second, third, and fourth) with the center as the origin. Step 2: Using the center as the origin and radii of 35mm, 100mm, and 142mm respectively, obtain ZONE1, ZONE2, and ZONE3 regions. Step 3: If the total defect count (ZONE1 + ZONE2 + ZONE3) is greater than 1000, it is considered NG; less than 15, it is considered OK; and between 15 and 1000, proceed to the next step. Step 4: Compare the defect counts pairwise in each of the four quadrants. If a ratio > 6, it is considered OK; otherwise, proceed to the next step. Step 5: Classify the three regions according to the defect count and relative density. Step 6: If none of the above conditions are met, it is considered OK. This method has the advantages of convenient operation and good judgment stability. It takes into account the special pattern characteristics of COP silicon wafers while eliminating subjective interference.

Description

Technical Field

[0001] This invention relates to the field of silicon wafer inspection technology, and more specifically to a method for automatically determining the particle stacking pattern of silicon wafers based on an algorithm. Background Technology

[0002] When testing 12-inch silicon wafers, wafers with COP (Chip-on-Particle) inclusions will display a special pattern in the test pattern, such as a central cluster or a ring-shaped edge. Previously, the identification of these COP-containing silicon wafers could only be done by controlling the total number of particles and having personnel visually inspect a limited number of samples based on the COP silicon wafer's special pattern.

[0003] First, the scheme of controlling the total number of particles does not take into account the special pattern of COP silicon wafers themselves. If the total number is set too high, it is easy to miss COP silicon wafers. If the total number is set too low, it is easy to over-identify non-COP silicon wafers as COP silicon wafers.

[0004] Secondly, the visual inspection of COP silicon wafer special pattern limit samples is greatly influenced by the subjective judgment of the inspectors, and different inspectors may make different judgments. Summary of the Invention

[0005] This invention addresses the shortcomings of existing technologies by providing an algorithm-based method for automatically determining silicon wafer particle overlay patterns. This method offers advantages such as ease of operation and high stability. It considers the special pattern characteristics of COP silicon wafers while eliminating subjective interference.

[0006] The above-mentioned technical problems of the present invention are mainly solved by the following technical solutions:

[0007] A method for automatically determining silicon wafer particle stacking patterns based on an algorithm, using an SPX machine to test and stack 12-inch silicon wafers, includes the following steps:

[0008] Step 1: Using the center of the silicon wafer as the origin, divide it into the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant.

[0009] Step 2: Using the center of the silicon wafer as the origin, and radii of 35mm, 100mm, and 142mm, respectively, we obtain the ZONE1, ZONE2, and ZONE3 regions.

[0010] Step 3: If the number of defects in ONE1+ZONE2+ZONE3 is greater than 1000, it is considered NG; if it is less than 15, it is considered OK; if it is between 15 and 1000, proceed to the next step.

[0011] Step 4: Compare the number of defects in each of the four quadrants pairwise. If the ratio is greater than 6, it is considered OK; otherwise, proceed to the next step.

[0012] Step 5: Classify the three regions according to the number and relative density of defects.

[0013] Step 6: If none of the above conditions are met, the result is OK.

[0014] As a preferred option, ZONE1 is a green circular area, ZONE2 is a red circular area, and ZONE3 is a yellow circular area; the area ratio of the three areas is calculated to obtain ZONE1:ZONE2:ZONE3=1 : 7.16 : 8.3.

[0015] As a preferred method, the defects that need to be considered are screened out through the Klarf file after particle testing, the number of defects in each region is calculated in detail, and the relative density between each region is obtained by using the particle number / area ratio.

[0016] As a preferred option, ZONE1 has the highest relative density, which is twice or more than the relative density of ZONE2. At the same time, if the number of defects in ZONE1 is greater than 10, it is judged as NG and the mode is central clustering.

[0017] As a preferred option, ZONE3 has the highest relative density, which is 3.2 times or more than that of ZONE2. At the same time, if the number of defects in ZONE3 is greater than 60, it is judged as NG and the pattern is edge clustering.

[0018] The present invention can achieve the following effects:

[0019] This invention provides an algorithm-based method for automatically determining the particle stacking pattern of silicon wafers. Compared with existing technologies, it has the advantages of convenient operation and good judgment stability. It takes into account the special pattern characteristics of COP silicon wafers while eliminating the interference of subjective factors. Detailed Implementation

[0020] The technical solution of the invention will be further described in detail below through examples.

[0021] Example: A method for automatic determination of silicon wafer particle stacking patterns based on an algorithm, using an SPX machine to test and stack 12-inch silicon wafers, includes the following steps:

[0022] Step 1: Using the center of the silicon wafer as the origin, divide it into the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant.

[0023] Step 2: Using the center of the silicon wafer as the origin, and radii of 35mm, 100mm, and 142mm, respectively, we obtain the ZONE1, ZONE2, and ZONE3 regions.

[0024] ZONE1 is a green circular area, ZONE2 is a red circular area, and ZONE3 is a yellow circular area; the area ratio of the three areas is calculated to be ZONE1:ZONE2:ZONE3 = 1 : 7.16 : 8.3.

[0025] After particle testing, the Klarf file is used to filter out the defects that need to be considered. The number of defects in each region is calculated in detail, and the relative density between each region is obtained by using the particle number / area ratio.

[0026] Step 3: If the number of defects in ONE1+ZONE2+ZONE3 is greater than 1000, it is considered NG; if it is less than 15, it is considered OK; if it is between 15 and 1000, proceed to the next step.

[0027] Step 4: Compare the number of defects in each of the four quadrants pairwise. If the ratio is greater than 6, it is considered OK; otherwise, proceed to the next step.

[0028] Step 5: Classify the three regions according to the number and relative density of defects.

[0029] ZONE1 has the highest relative density, which is twice or more than that of ZONE2. At the same time, ZONE1 has more than 10 defects, so it is judged as NG and the pattern is central clustering.

[0030] ZONE3 has the highest relative density, which is 3.2 times or more than that of ZONE2. At the same time, ZONE3 has more than 60 defects, so it is judged as NG and the pattern is edge clustering.

[0031] Step 6: If none of the above conditions are met, the result is OK.

[0032] The overlay images were automatically judged and manually checked twice. A total of 933 images were processed, with 1 image missed and 29 images over-judged, resulting in a miss rate of 0.11%, an over-judgment rate of 3.12%, and an accuracy rate of 96.77%. Except for silicon wafers with both serious processing abnormalities and COP that could not be correctly identified, all others could be classified.

[0033] In summary, this algorithm-based method for automatically determining silicon wafer particle stacking patterns offers advantages such as ease of operation and high stability. It takes into account the special pattern characteristics of COP silicon wafers while eliminating the interference of subjective factors.

[0034] The above description is only a specific embodiment of the present invention, but the structural features of the present invention are not limited thereto. Any changes or modifications made by those skilled in the art within the scope of the present invention are covered by the patent scope of the present invention.

Claims

1. A method for automatically determining silicon wafer particle stacking patterns based on an algorithm, characterized in that: The following steps are involved in testing and overlaying 12-inch silicon wafers using the SPX equipment: Step 1: Using the center of the silicon wafer as the origin, divide the area into four quadrants: the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant. Step 2: Using the center of the silicon wafer as the origin, and radii of 35mm, 100mm, and 142mm, respectively, we obtain the ZONE1, ZONE2, and ZONE3 regions. ZONE1 is a green circular area, ZONE2 is a red circular area, and ZONE3 is a yellow circular area; the area ratio of the three areas is calculated to be ZONE1:ZONE2:ZONE3 = 1 : 7.16 : 8.3; After particle testing, the Klarf file is used to filter out the defects that need to be considered. The number of defects in each region is calculated in detail, and the relative density between each region is obtained by using the particle number / area ratio. Step 3: If the number of defects in ZONE1+ZONE2+ZONE3 is greater than 1000, it is considered NG; if it is less than 15, it is considered OK; if it is between 15 and 1000, proceed to the next step. Step 4: Compare the number of defects in each of the four quadrants pairwise. If there is a ratio > 6, it is considered OK; otherwise, proceed to the next step. Step 5: Classify the three regions according to the number and relative density of defects; ZONE1 has the highest relative density, which is twice or more than the relative density of ZONE2. At the same time, ZONE1 has more than 10 defects, so it is judged as NG and the pattern is central clustering. ZONE3 has the highest relative density, which is 3.2 times or more than that of ZONE2. At the same time, ZONE3 has more than 60 defects, so it is judged as NG and the pattern is edge clustering. Step 6: If none of the above conditions are met, the result is OK.