Semiconductor fabrication process

Abstract

This invention provides a semiconductor manufacturing process, comprising using a computer system to divide a wafer area into multiple exposure areas and distribute multiple observation points in each exposure area (shot), identifying a portion of boundary areas from all exposure areas, calculating the number of observation points in each boundary area, and eliminating boundary areas with fewer than 3 observation points, counting the observation points in the remaining boundary areas, and deleting a portion of the observation points until the total number of observation points meets a preset total and the observation points are uniformly distributed, and performing an alignment measurement step on the remaining observation points to generate an offset vector map.

Description

Technical Field

[0001] This invention relates to the field of semiconductor manufacturing processes, and in particular to a method for improving the accuracy of semiconductor overlay manufacturing processes. Background Technology

[0002] In semiconductor manufacturing processes, stacked structures are frequently formed, which involves creating multiple stacked material layers on the same substrate or material layer. After one material layer is formed, before the next material layer is formed, an overlay process is needed to align it with the previous material layer and place it in the correct position. The higher the precision of the overlay process, the more accurately the semiconductor material layers can be formed, preventing damage to the material layers and ensuring the overall yield of the semiconductor device. Summary of the Invention

[0003] This invention provides a semiconductor manufacturing process, comprising using a computer system to divide a wafer area into multiple exposure areas and distribute multiple observation points in each exposure area (shot), identifying a portion of boundary areas from all exposure areas, calculating the number of observation points in each boundary area, and eliminating boundary areas with fewer than 3 observation points, counting the observation points in the remaining boundary areas, and deleting a portion of the observation points until the total number of observation points meets a preset total and the observation points are uniformly distributed, and performing an alignment measurement step on the remaining observation points to generate an offset vector map.

[0004] The key feature of this invention is that, in existing technologies, observation points in boundary regions are deliberately omitted from the calculation of their offsets, resulting in significant errors in vector offset calculations for boundary regions and reducing the yield of dies in these areas. Unlike existing technologies, this invention considers observation points in boundary regions in addition to the complete region. Through a selection process, these observation points are evenly distributed, improving reliability. Therefore, when generating the offset vector map, the accuracy of the offset vectors within the boundary regions can be effectively improved, leading to a significant increase in the yield of dies in these areas. Attached Figure Description

[0005] Figure 1 This is a schematic diagram illustrating the definition of complete and boundary regions on a chip. Figure 2 The flowchart for improving the accuracy of the overlapping step in this invention; Figure 3 This is a schematic diagram showing the numbering of different locations within an exposure area; Figure 4 This is a schematic diagram illustrating an example of listing the remaining observation points on a histogram.

[0006] Explanation of main component symbols

[0007] 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12: Location

[0008] A, B: Circled parts

[0009] D: Cutting channel

[0010] P: Observation point

[0011] R: Effective radius

[0012] S: Area

[0013] S1: Complete Region

[0014] S2: Boundary Area

[0015] S01, S02, S03, S04, S05: Steps

[0016] W: Chip Detailed Implementation

[0017] To enable those skilled in the art to further understand the present invention, preferred embodiments of the present invention are described below, and the composition and desired effects of the present invention are explained in detail with reference to the accompanying drawings.

[0018] For ease of explanation, the accompanying drawings are merely illustrative to facilitate understanding of the invention, and their detailed proportions can be adjusted according to design requirements. The vertical relationships between relative elements in the drawings described herein should be understood by those skilled in the art to refer to the relative positions of objects; therefore, all can be flipped to present the same components, and this should all fall within the scope of this specification, as stated herein.

[0019] Overlay fabrication is a commonly used technique in the semiconductor industry. Its principle involves distributing multiple observation points on a wafer, then measuring the differences in position between these observation points in a computer simulation system and their actual location on the wafer. These differences are then used to create an offset vector map. The offset vector map displays the difference between the expected and actual positions of each observation point on the wafer. The pattern shape or position can then be corrected using the offset vector map, allowing the pattern to be formed in the desired location (e.g., precisely stacked on another material layer).

[0020] The applicant found that the current overlay process still has some shortcomings. Figure 1 Draw the definitions of the complete region and the boundary region on a chip. For example... Figure 1As shown, a wafer W is provided. Wafer W can be diced into multiple dies in subsequent steps. For simplicity, the diced dies are not shown in the diagram. Multiple exposure areas S (shots) are defined on wafer W, where the size and arrangement of each exposure area S can be defined by the user. The exposure areas S defined here can also be called shots or fields. The size of each exposure area S depends on the unit area that the light source can illuminate during a single exposure by the exposure machine. That is, during the exposure process, the exposure machine sequentially performs one exposure operation in each exposure area S in a step-by-step manner. Typically, each exposure area S contains multiple dies, and there are dicing channels between each die. Figure 1 For the sake of simplicity, the blank film and cutting path within each exposure area S are not shown in the accompanying diagram. However, it is understood that the size or arrangement of the exposure area S, the blank film, and the cutting path can be adjusted according to actual needs.

[0021] Within the aforementioned exposure area S, a complete region S1 and a boundary region S2 are further defined. In this embodiment, the complete region S1 is defined as follows: the entire extent of the complete region S1 completely overlaps with the wafer W; that is, all dies contained within the complete region S1 are within the wafer W's area. The boundary region S2 is defined as follows: only a portion of the boundary region S2 overlaps with the wafer W, while the remaining portion lies outside the wafer W's area. Another method for determining the boundary region S2 is to set the effective radius of the wafer W as R, and draw a circle from the center of the wafer W with the effective radius R. The regions S passing through the effective radius R are then the boundary regions S2.

[0022] The effective radius R here varies depending on the size of the wafer W. For example, a 12-inch wafer has a diameter of approximately 300 millimeters (mm), so the effective radius R is approximately 145 mm. The effective radius R is defined as the area on the wafer W where dies can be formed. Because the area at the outermost boundary of the wafer W where dies cannot be formed needs to be deducted, the effective radius R will be slightly smaller than the actual wafer radius (i.e., 300 / 2 mm). However, it is worth noting that the above-mentioned effective radius R can be adjusted according to the actual manufacturing process or wafer size, and this invention is not limited to this.

[0023] Additionally, multiple observation points P are distributed on the wafer W. The positions of observation points P can be defined by the user, but the positions of the observation points in each exposure area S will be consistent, and the observation points P will be located on the dicing path. In some embodiments, the observation points P can also be regarded as overlay marks in the overlay step, used to observe changes such as offset, rotation, or scaling before and after overlay.

[0024] In existing overlay processes, if observation point P is located in boundary region S2, since boundary region S2 is adjacent to the boundary of wafer W, some observation points are outside the wafer W and cannot be measured. Furthermore, the measurable observation points P within boundary region S2 may be unevenly distributed, significantly impacting the accuracy of the measurement data. In other words, measuring observation points in boundary region S2 is prone to error. The existing overlay process addresses this by deliberately ignoring observation points P in boundary region S2 and only calculating the offset of observation points P in the remaining region (e.g., the complete region S1). However, this is equivalent to abandoning the calculation of the offset of observation points P in boundary region S2, thus significantly reducing the yield of dies located within boundary region S2.

[0025] This invention provides a method for improving the accuracy of the overlay process, particularly a method for filtering observation points within a boundary region S2. The method provided by this invention can effectively filter observation points within the boundary region S2, ensuring that the filtered observation points P within S2 are evenly distributed, thus improving the accuracy of the overlay fabrication process within the boundary region S2. Please refer to... Figure 2 The flowchart illustrates the present invention's method for improving the accuracy of the overlapping steps. For example... Figure 2 As shown in step S01, a computer system is used to divide the area of ​​a wafer W into multiple exposure regions S, and multiple observation points P are defined in each exposure region S. Next, as shown in step S02, a portion of the boundary regions S2 is identified from all the exposure regions S. The method for determining the boundary region S2 is the same as described above and will not be repeated here. Furthermore, other regions besides the boundary region S2, such as the complete region S1, are calculated using the same method as existing overlay techniques; this part will not be elaborated upon in this invention. The feature of this invention is that, in addition to the complete region S1, it also considers the offset or rotation of the observation points P within the boundary region S2. Steps S01 and S02 described here are as described above. Figure 1 As mentioned in the relevant paragraphs, I will not repeat them here.

[0026] Next, a list of all observation points needs to be compiled. Please refer to the following: Figure 3 , Figure 3 Draw the numbering of different locations within a region, such as... Figure 3 As shown, multiple locations can be defined by the user within an exposure area S (e.g., boundary area S2), where each location corresponds to the observation point P within the exposure area S. For example, in this embodiment, a total of 12 locations are defined, namely... Figure 3 The locations 1 to 12 are all situated on the boundary of the exposure area S or on the cutting path D. In other words, Figure 1 Each exposure area S in the diagram also contains 12 observation points, located at the corresponding locations 1 to 12. It is worth noting that these 12 locations are defined by the user; more or fewer locations can be defined as needed, and the locations do not necessarily need to be aligned with each other. Figure 3 same( Figure 3 (The illustration shown is merely one embodiment of the present invention). After defining the locations 1 to 12, each observation point P can be listed in a table, as shown in Table 1 below:

[0027]

[0028] Table 1

[0029] As shown in Table 1 above, the boundary region numbers from 1 to 52 represent different boundary regions S2. The number of boundary regions can be adjusted according to the user-defined regions. Locations 1 to 12 represent observation points P at the corresponding locations within the boundary region S2. The numbers in the table represent the shortest distance from observation point P to the center of the wafer W, in millimeters (mm). For example, taking boundary region number 1 as an example, there are 4 observation points located in this boundary region: observation point located at location 3 with a shortest distance of 134.1172 mm from the center of the wafer W; observation point located at location 6 with a shortest distance of 139.3561 mm from the center of the wafer W; observation point located at location 7 with a shortest distance of 143.7654 mm from the center of the wafer W; and observation point located at location 11 with a shortest distance of 143.3301 mm from the center of the wafer W. As for the blank parts in the rest of the table, it means that the location of the exposure area S may be outside the range of the chip W, so the observation point P cannot be measured.

[0030] Next, as shown in step S03, boundary regions with more than 3 measurable observation points are selected. Refer to the rightmost column of Table 1, which represents the statistical count of measurable observation points P in each boundary region 1-52. It can be observed that some boundary regions have fewer than 3 observation points (e.g., boundary regions numbered 12, 19, 23, 30, 33, and 40). According to the applicant's experimental results, if the number of observation points in a boundary region is too small, it is easier to reduce the calculation accuracy of subsequent overlay steps. Therefore, in this invention, boundary regions with fewer than 3 observation points are removed to improve the reliability of the calculation. That is, after step S03, a portion of boundary regions S2 and observation points P are removed, and the remaining boundary regions all contain at least 3 or more observation points P. The results after removal are shown in Table 2 below:

[0031]

[0032] Table 2

[0033] Then as Figure 4 As shown, Figure 4 The following is an example of listing the remaining observation points in a histogram. The vertical axis of the histogram represents the number of observation points, and the horizontal axis represents the radius units. Here, the radius unit is defined as follows: the user first sets a preset total number N, and then the size of each radius unit is: [(the straight-line distance from the farthest observation point to the center of the wafer to the center) - (the straight-line distance from the nearest observation point to the center of the wafer to the center)] / preset total number N. For a practical example, please refer to Table 2. The observation point farthest from the center (i.e., the largest value in the table) is 144.9594 mm, and the observation point closest to the center (i.e., the closest value in the table) is 123.0128 mm. If the preset total number N is set to 120 (N in this invention can be adjusted according to needs and is not necessarily 120; N=120 here is only an example), then... Figure 4 The horizontal axis has N radius units, and the size of each radius unit is (144.9594-123.0128) / 120. That is to say, Figure 4 The horizontal axis of the histogram should have 120 radius cells, and each radius cell may contain a corresponding number of observation points. However, for the sake of brevity in the accompanying drawings, not all 120 radius cells are shown in this embodiment. The remaining observation points from Table 2 are listed below. Figure 4 The histogram can be obtained from... Figure 4 Different numbers of observation points were observed under different radius units.

[0034] Next, as in step S04: count the remaining observation points in the boundary region and delete some of them until the total number of observation points meets the preset total and the observation points are evenly distributed. For details, please refer to [link / reference needed]. Figure 4 The purpose of this invention is to filter these observation points, delete redundant ones, and ensure that the remaining observation points are evenly distributed. This improves the accuracy of subsequent overlay steps. In other words, the number of observation points listed in Table 2 is too large and not necessarily evenly distributed. Therefore, one of the objectives of this invention is to delete densely distributed observation points, reduce the total number of observation points, and ensure that the remaining observation points are evenly distributed.

[0035] In order to achieve the above objectives, Figure 4 In the histogram, the observation points that are most numerous within the same radius cell are deleted first, such as... Figure 4The circled portion A can be deleted first to avoid having too many observation points within the same radius cell. Next, when there are no observation points within a certain radius cell... Figure 4 If the vertical axis of the histogram is 0, and there are two or more observation points within a neighboring radius cell, an observation point can be moved from or assigned to a radius cell with a vertical axis value of 0 (e.g., ...). Figure 4 (As shown in circle B). Furthermore, when deleting observation points, computer calculations can be used to ensure that the number of observation points at each location remains as consistent as possible. Repeat the above steps of deleting and relocating (assigning) observation points until the final number of observation points reaches the preset total number N (120 in this embodiment, but not limited to this). At this point, the remaining observation points should satisfy several conditions: (1) each boundary region has more than 3 observation points; (2) the number of observation points at each location 1 to 12 is approximately consistent; and (3) each radius unit contains one observation point.

[0036] Finally, as in step S05, an alignment measurement step is performed on the remaining observation points to generate an offset vector map. It is worth noting that the observation points here include those located within the complete region S1 and the remaining observation points after filtering within the boundary region S2. The offset vector map formed through the alignment measurement step can be applied to the overlay step to correct the parameters or positions of the subsequently generated overlay pattern. The technical content regarding the generation of the offset vector map is prior art in this field and will not be elaborated upon here.

[0037] Based on the above description and accompanying drawings, the present invention provides a semiconductor manufacturing process, comprising using a computer system to divide a wafer area into multiple exposure areas (shots), defining multiple observation points P in each exposure area S, identifying partial boundary areas S2 in each exposure area S, calculating the number of observation points P in each boundary area S2, and removing some boundary areas with fewer than 3 observation points P, counting the remaining observation points P in the remaining boundary areas S2, and deleting some observation points P until the total number of observation points P meets a preset total number N, and the observation points P are uniformly distributed, and performing an alignment measurement step on the remaining observation points P to generate an offset vector map.

[0038] In some embodiments of the present invention, the method for finding the boundary region S2 in each exposure region S includes defining a center of a circle within the wafer W range and an effective radius value R, and drawing a circle based on the center and the effective radius, wherein the region passing through the circle in each exposure region S is defined as the boundary region S2 (e.g., Figure 1 (As shown).

[0039] In some embodiments of the present invention, each exposure region S contains a plurality of complete dies, and each boundary region S2 contains at least one complete die.

[0040] In some embodiments of the present invention, the method for counting the remaining observation points P in the remaining boundary region S2 includes counting the remaining observation points P in a table based on the straight-line distance of each observation point P from the center of the circle (i.e., Figure 4 (histogram).

[0041] In some embodiments of the present invention, the table is a histogram, and the preset total number is defined as N, wherein the horizontal axis of the table is N radius units, and the vertical axis of the table is the number of observation points.

[0042] In some embodiments of the present invention, the straight-line distance from the observation point farthest from the center of the circle among the remaining observation points is defined as X1, the straight-line distance from the observation point closest to the center of the circle among the remaining observation points is defined as X2, and each radius unit is: (X1-X2) / N.

[0043] In some embodiments of the present invention, the method for deleting a portion of observation points P includes: based on a table (histogram), prioritizing the deletion of observation points P with the largest vertical axis values.

[0044] In some embodiments of the present invention, it is further included in the horizontal axis of the table that if a first radius cell contains 0 observation points and a second radius cell adjacent to it contains more than 2 observation points, then an observation point is assigned from the observation points of the second radius cell to the first radius cell (i.e., the aforementioned observation point shifting step B).

[0045] In some embodiments of the present invention, each boundary region is defined with multiple location locations (e.g., Figure 3 The locations are 1 to 12, and also include the locations to which the remaining observation points P belong.

[0046] In some embodiments of the present invention, after deleting a portion of the observation points, the observation points located at each location are also evenly distributed.

[0047] In summary, the key feature of this invention is that, in existing technologies, observation points in boundary regions are deliberately omitted from calculation, resulting in significant errors in vector offset calculations for boundary regions and reducing the yield of dies in these areas. This invention differs from existing technologies by considering observation points in boundary regions in addition to the complete region. Through a selection process, these observation points are evenly distributed, improving reliability. Therefore, when generating the offset vector map, the accuracy of the offset vectors within the boundary regions can be effectively improved, leading to a significant increase in the yield of dies in these areas.

[0048] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made in accordance with the claims of the present invention should be included within the scope of the present invention.

Claims

1. A semiconductor manufacturing process, comprising: Using a computer system, the wafer area is divided into multiple exposure zones (shots), and multiple observation points are defined in each exposure zone; Identify the boundary regions within each of the exposed areas; Calculate the number of observation points in each boundary region, and remove boundary regions with fewer than 3 observation points; The remaining observation points in these boundary regions are counted, and densely distributed observation points are deleted to reduce the total number of observation points and ensure that the total number of observation points meets the preset total. Furthermore, the remaining observation points in these boundary regions are evenly distributed by shifting or redistributing observation points. Alignment measurement steps are performed on the remaining observation points to generate an offset vector map.

2. The semiconductor fabrication process of claim 1, wherein the method for identifying the boundary regions in each of the exposed regions comprises: Define the center of the circle within the wafer range and the effective radius value, and draw the circle based on the center and the effective radius; Within each of these exposure areas, those exposure areas passing through the circle are defined as the boundary areas.

3. The semiconductor fabrication process of claim 2, wherein each of the exposure regions contains a plurality of complete dies, and each of the boundary regions contains at least one complete die.

4. The semiconductor fabrication process of claim 2, wherein the method for statistically analyzing the remaining observation points in the boundary regions comprises: The remaining observation points are statistically analyzed in a table based on their straight-line distance from the center of the circle.

5. The semiconductor manufacturing process of claim 4, wherein the table is a histogram, and the preset total number is defined as N, wherein the horizontal axis of the table is N radius units, and the vertical axis of the table is the number of observation points.

6. The semiconductor fabrication process of claim 5, wherein the straight-line distance from the observation point farthest from the center of the circle among the remaining observation points is defined as X1, the straight-line distance from the observation point closest to the center of the circle among the remaining observation points is defined as X2, and each radius unit is: (X1-X2) / N.

7. The semiconductor fabrication process of claim 6, wherein the method for deleting densely distributed observation points comprises: Based on the table, the observation points with the largest vertical axis values ​​should be deleted first.

8. The semiconductor manufacturing process of claim 7, further comprising: In the horizontal axis of the table, if the number of observation points contained in the first radius cell is 0, and the second radius cell adjacent to it contains more than 2 observation points, then one observation point is assigned from the observation points in the second radius cell to the first radius cell.

9. The semiconductor fabrication process of claim 1, wherein each of the boundary regions defines a plurality of locations, and further includes statistically analyzing the locations to which the remaining observation points belong.

10. The semiconductor manufacturing process of claim 9, wherein after removing densely distributed observation points, the observation points at each of the respective locations are also evenly distributed.

Images