Microstructure processing error calculation method based on image processing

Abstract

The application discloses a microstructure processing error calculation method based on image processing, and the micro-nano processing error includes depth error, line width error and overlay error. A white light interferometer is used to detect the step profile of the graphic area near all auxiliary detection mark points, and three-dimensional profile data of corresponding positions are obtained; an image processing method is used to obtain the section line of the three-dimensional profile data along the step section direction, the depth error and the line width error are calculated, then the three-dimensional profile data is projected in the overhead direction, the step edge line segment graph is obtained, and the overlay error is calculated. The application is to improve the existing processing error precision requirement, and provides a microstructure processing error calculation method based on image processing.

Description

Technical Field

[0001] This invention belongs to the field of micro-nano fabrication errors, and specifically relates to a method for calculating microstructure fabrication errors based on image processing. Background Technology

[0002] With the development of optical technology, the integration of optics, mechanics, and electronics has become a trend, requiring optical systems to be integrated, arrayed, and miniaturized. Diffractive optical elements are increasingly widely used. Traditional diffractive optical elements are fabricated using a multi-mask overlay process. Theoretically, a 4-step diffractive optical element can achieve a diffraction efficiency of 81%, and an 8-step diffractive optical element can achieve 95%. However, many processing errors occur during microstructure fabrication, such as depth error, linewidth error, and overlay error. The actual phase structure of the device deviates from the designed distribution, significantly reducing its diffraction efficiency. To better design and fabricate diffractive optical elements, it is necessary to propose requirements for the processing technology based on the design requirements or to estimate the effect of the diffractive optical element based on the existing technology level. Therefore, analyzing the impact of processing errors on diffractive optical elements is very meaningful.

[0003] Depth error: During the etching process, inaccurate etching rate calibration and the decrease in etching rate caused by material deposition during the etching process will cause the etching depth to deviate from the design value. Taking one cycle of a four-step diffraction element as an example, when only the etching depth error exists, a constant phase deviation will be introduced.

[0004] Line width error: When the development time is too long or the exposure is too long, the line width will be greater than the design value. Conversely, insufficient exposure or too short a development time will result in the processed line width value being less than the design value.

[0005] Overlay error: Ideally, the current layer pattern and the previous layer pattern of an integrated circuit are perfectly aligned, meaning the relative displacement is zero. However, in actual manufacturing, there will always be a certain amount of overlay error because, during the alignment and exposure steps, the mask of the current layer pattern cannot maintain a perfect alignment with the mask of the previous layer pattern. Overlay error refers to the relative offset between the current layer pattern and the previous layer pattern caused by the non-ideal alignment of the mask.

[0006] Existing technologies employ optical microscopes to record vernier caliper data, obtaining overlay error results by reading the vernier caliper data. Then, a white light interferometer is used to detect the step contours of the graphic area near all marked points, obtaining etching depth data at the corresponding locations, thus yielding depth and linewidth errors. This invention, through a single detection using a white light interferometer, can simultaneously obtain depth, linewidth, and overlay errors by inputting the data into a program for calculation. Furthermore, setting multiple sampling points removes gross errors, calculates average values, and covers the entire effective range of the tested mirror, resulting in higher reliability. Summary of the Invention

[0007] This invention aims to solve the problem of calculating processing errors introduced during micro- and nano-fabrication. To improve the accuracy requirements of existing processing errors, this invention provides a method for calculating microstructure processing errors based on image processing.

[0008] The technical solution adopted in this invention is: a method for calculating microstructure processing errors based on image processing, wherein the microstructure processing errors include depth error, line width error, and overlay error, characterized by comprising the following steps:

[0009] Step 1: Mark auxiliary detection points on the microstructure being tested;

[0010] Step 2: Use a white light interferometer to detect the contour of the steps in the graphic area near all marked points to obtain the three-dimensional contour data of the corresponding locations;

[0011] Step 3: Take the 3D contour data from Step 2 and cut multiple cross lines along the step tangent direction. Calculate the average of the depth and line width values ​​on these multiple cross lines to obtain the measured depth and line width values ​​at each point.

[0012] Step 4: Compare the measured depth and line width values ​​at each point with the design values ​​to calculate the depth error and line width error at each point, and calculate the average of the depth error and line width error of all points to obtain the depth error and line width error of the microstructure.

[0013] Step 5: Project the 3D contour data from Step 2 onto the top view to obtain the step edge line segment diagram. Fit this line segment diagram and take multiple intercepts along the perpendicular direction of the edge line segments. Obtain the intercept of each intercept in each period. The average intercept of multiple intercepts gives the line width value of the step at that point in each period. The number of steps is 2. n The interferometer measures multiple periods within its range;

[0014] Step 6: Calculate the difference in line width between the current step and the previous step in each cycle. Average the differences to obtain the overlay error in that cycle. The interferometer has multiple cycles in its measurement range. Averaging the overlay errors of multiple cycles will give the overlay error of the marked point.

[0015] Furthermore, the marker point map set in step one can be set to be dense or sparse according to the size of the microstructure being measured and the detection accuracy requirements.

[0016] Furthermore, step five uses only the microstructure processing error of 4 steps for verification. This method is not limited to 4 steps and can be extended to 8, 16 and above.

[0017] Furthermore, in steps three and five, three cross-sections are taken and averaged. To improve accuracy, more cross-sections can be taken.

[0018] Furthermore, in step six, the measurement range of the interferometer is approximately 7 to 8 cycles. Depending on the actual situation, the measurement range of the interferometer can be expanded or reduced.

[0019] The advantages of this invention compared to the prior art are:

[0020] Microstructure fabrication generates various processing errors, such as depth error, linewidth error, and overlay error. Existing inspection techniques use a profilometer to collect data from any period within a given region, measuring the depth and linewidth errors within that period. This invention sets up an N*N measurement area within the effective range of the mirror being tested, and uses image processing to calculate the average value of all periods within the N*N area to obtain the depth and linewidth errors.

[0021] In existing testing techniques, optical microscopes are used to record vernier caliper data, and the overlay error is obtained by reading the vernier caliper data. This invention sets an N*N measurement area within the effective range of the mirror being tested, uses image processing to calculate the average value of all cycles within the N*N area, and calculates the difference in width between steps 1 and 2, and steps 3 and 4 within each cycle. The two values ​​are then averaged to obtain the overlay error within that cycle.

[0022] The present invention has a larger calculation sample and the detection area covers the entire effective range of the tested mirror, resulting in higher reliability. Attached Figure Description

[0023] Figure 1 A diagram illustrating the locations and names of the markers used to assist in detection.

[0024] Figure 2 The data is the three-dimensional contour data detected by a white light interferometer.

[0025] Figure 3 A section line is obtained by cutting the three-dimensional contour data along the step tangent direction.

[0026] Figure 4 This is a diagram of the edge lines of a step in three-dimensional contour data. Detailed Implementation

[0027] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0028] The microstructure fabrication error calculation method based on image processing in this embodiment is implemented according to the following steps:

[0029] Step 1: Make auxiliary testing marks on the mirror being tested, and set the points and names as follows: Figure 1 As shown;

[0030] Step two: Use a white light interferometer to detect the contour of the steps in the graphic area near all marked points, and obtain the three-dimensional contour data of the corresponding locations, such as... Figure 2 As shown;

[0031] Step 3: Extract a section line from the 3D contour data from Step 2 along the step's tangent direction, such as... Figure 3 As shown, this cross section can calculate depth data and line width data;

[0032] Step 4: Compare the measured depth and line width data at each point with the design values ​​to calculate the depth error and line width error at each point, and calculate the average and root mean square of the depth error and line width error at all points.

[0033] Step 5: Project the 3D contour data from Step 2 onto the top view to obtain the step edge line segment diagram, as shown below. Figure 4 As shown, three intercepts are taken along the perpendicular direction of the edge line segment. The average intercept of the three intercepts is the line width value of step 1, 2, 3, and 4 in each period.

[0034] Step 6: Calculate the difference in line width between steps 1 and 2, and between steps 3 and 4 in each cycle. Average the two values ​​to obtain the overlay error in that cycle. The measurement range of the interferometer is approximately 7 to 8 cycles. Averaging the overlay errors of 7 to 8 cycles will give the overlay error of the marked point.

[0035] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for calculating microstructure processing errors based on image processing, wherein the microstructure processing errors include depth error, linewidth error, and overlay error, characterized in that, Includes the following steps: Step 1: Mark auxiliary detection points on the microstructure being tested; Step 2: Use a white light interferometer to detect the contour of the steps in the graphic area near all marked points to obtain the three-dimensional contour data of the corresponding locations; Step 3: Take the 3D contour data from Step 2 and cut multiple cross lines along the step's cross-section. Calculate the average of the depth and line width values ​​of the step on these multiple cross lines to obtain the measured depth and line width values ​​at each point. Step 4: Compare the measured depth and line width values ​​at each point with the design values ​​to calculate the depth error and line width error at each point, and calculate the average of the depth error and line width error of all points to obtain the depth error and line width error of the microstructure. Step 5: Project the 3D contour data from Step 2 onto the top view to obtain the step edge line segment diagram. Fit this line segment diagram and take multiple intercepts along the perpendicular direction of the edge line segments. Obtain the intercept of each intercept in each period. The average intercept of multiple intercepts gives the line width value of the step at that point in each period. The number of steps is 2. n The interferometer measures multiple periods. Step 6: Calculate the difference in line width between the current step and the previous step in each cycle. Average the above differences to obtain the overlay error in that cycle. Average the overlay errors of multiple cycles to obtain the overlay error of that point. Average the overlay errors of all points to obtain the overlay error of the microstructure.

2. The method for calculating microstructure processing errors based on image processing according to claim 1, characterized in that: The marker point map set in step one can be set to be dense or sparse according to the size of the microstructure being measured and the required detection accuracy.

3. The method for calculating microstructure processing errors based on image processing according to claim 1, characterized in that: In both steps three and five, take at least three cut lines.

4. The method for calculating microstructure processing errors based on image processing according to claim 1, characterized in that: The intercept line in step three is the position of the integer bisector line.

5. The method for calculating microstructure processing errors based on image processing according to claim 1, characterized in that: n takes the values ​​4, 8, and 16.

Images