Photoresist surface flatness detection method and system based on model registration

Abstract

The present invention discloses a method for detecting the flatness of a photoresist surface based on model registration, the method comprising: S1: scanning the photoresist surface by using a laser radar, collecting radar data and denoising, and constructing a point cloud model; S2: aligning the direction of the photoresist surface model with the point cloud model; S3: calculating the precise registration parameters of the photoresist surface model and the point cloud model; S4: calculating the deviation of the point cloud model in the point cloud model; S5: evaluating the flatness of the photoresist surface model based on the precise registration parameters obtained in step S3 and the deviation of the point cloud model obtained in step S4. The present invention performs multi-level registration on the photoresist surface model and the point cloud model to improve the registration accuracy; by calculating the deviation of the point cloud model, the flatness detection of the photoresist surface is realized, and the quality of the mask product is improved.

Description

Technical Field

[0001] The invention relates to the technical field of photoresist surface flatness detection, and in particular to a photoresist surface flatness detection method and system based on model registration. Background Art

[0002] Photoresist is widely used in integrated circuit electronics and semiconductor chips. It is a kind of etching-resistant thin film material. The photoresist production process is complicated and has strict process requirements. Photoresist production requires a high surface flatness, and the surface flatness of photoresist directly affects the performance and quality of electronic products. If the surface of the photoresist is not flat, it cannot guarantee that the light can pass through the photoresist to the wafer, and the pattern formed will be deviated and blurred.

[0003] In production, the detection and measurement of the flatness of the photoresist surface is very important. The existing photoresist surface flatness detection is mainly based on the surface position measurement system and the level sensing system to measure the height fluctuation of the photoresist surface to ensure uniform coverage of the photoresist. However, this type of flatness measurement equipment is a precision instrument, and the use cost and maintenance cost of promoting it in large-scale production are high. Summary of the invention

[0004] Laser radar has high measurement accuracy and good robustness, and its production technology is mature and suitable for large-scale promotion and application; the point cloud model established based on laser radar data is highly matched with the target, which provides great convenience for the measurement and processing of the target. In view of this, the technical solution proposed in the present invention is a method for detecting the flatness of the photoresist surface based on model registration, which makes full use of the point cloud model to evaluate the flatness of the photoresist surface. The present invention provides a method for detecting the flatness of the photoresist surface based on model registration, comprising the following steps:

[0005] S1: Use LiDAR to scan the photoresist surface, collect and denoise LiDAR data, and build a point cloud model;

[0006] S2: Align the photoresist surface model with the point cloud model;

[0007] S3: Calculate the precise registration parameters of the photoresist surface model and the point cloud model;

[0008] S4: in the point cloud model, calculating the deviation of the point cloud model;

[0009] S5: Based on the precise registration parameters obtained in step S3 and the deviation of the point cloud model obtained in step S4, the flatness of the photoresist surface model is evaluated.

[0010] Optionally, in step S1, a laser radar is used to scan the surface of the photoresist, radar data is collected and denoised, and a point cloud model is constructed, including:

[0011] S11: Use laser radar to scan the photoresist surface, collect multiple sets of radar data, and perform singular value decomposition on each set of data:

[0012] ;

[0013] in, Represents a set of radar data, and is an orthogonal matrix, It is a diagonal matrix, and the elements on the diagonal are singular values;

[0014] The matrix The elements on the diagonal are arranged in order from largest to smallest, and the value less than the threshold The elements of are set to 0, and the updated diagonal matrix is ​​obtained ; Based on the updated diagonal matrix Get the denoised radar data:

[0015] ;

[0016] in, Represents a set of denoised radar data;

[0017] S12: Use the overlapping areas to stitch together multiple sets of radar data after denoising:

[0018] ;

[0019] in, Respectively represent the target points in the two sets of radar data, Represents the points in the overlapping area of ​​the two sets of radar data, express The distance between express The distance between express The distance between Represents the angle between the planes fitted by the two sets of radar data;

[0020] S13: Draw a three-dimensional model based on the spliced ​​radar data, apply different colors to each surface of the three-dimensional model, and obtain a point cloud model.

[0021] Optionally, in step S2, aligning the photoresist surface model with the point cloud model includes:

[0022] S21: Based on the photoresist object, a proportional geometric model is established, which is the photoresist surface model;

[0023] Select a plane with significant features in the photoresist surface model and construct the normal vector ;

[0024] S22: In the point cloud model, Set the plane fitted by radar data and construct the normal vector ;

[0025] S23: Traverse the normal vector of each plane of the point cloud model:

[0026] ;

[0027] in, Indicates the maximum number of groups of radar data received;

[0028] Find the normal vector to the photoresist surface model Consistent normal vectors enable orientation alignment between the photoresist surface model and the point cloud model.

[0029] Optionally, in step S3, calculating precise registration parameters of the photoresist surface model and the point cloud model includes:

[0030] Select corresponding reference points in the photoresist surface model and point cloud model;

[0031] Calculate the precise registration parameters of the photoresist surface model and the point cloud model.

[0032] For example, based on the alignment of the directions of the photoresist surface model and the point cloud model, the reference surfaces corresponding to the photoresist surface model and the point cloud model are selected; the edge points of the corresponding reference surfaces are selected as the reference points corresponding to the model; in the photoresist surface model, the reference points are not all located in the same plane;

[0033] Optionally, in step S3, calculating precise registration parameters of the photoresist surface model and the point cloud model includes:

[0034] S31: Based on the reference point, calculate the deviation from the photoresist surface model to the point cloud model:

[0035] ;

[0036] in, represents the deviation from the photoresist surface model to the point cloud model, represents the first The coordinates of the reference points, Represents the first The coordinates of the reference points, Represents the rotation matrix from the photoresist surface model to the point cloud model, Represents the translation matrix from the photoresist surface model to the point cloud model;

[0037] S32: Based on minimum deviation Principle, using the least squares method, solve the rotation matrix from the photoresist surface model to the point cloud model and translation matrix ;

[0038] S33: The first Coordinates of the reference points Updated to ;

[0039] Based on the update , and , call step S31 to calculate the deviation from the photoresist surface model to the point cloud model ;

[0040] S34: Repeat steps S32 and S33 until Less than the model deviation threshold, at this time the best rotation matrix from the photoresist surface model to the point cloud model is obtained and the optimal translation matrix ;

[0041] The optimal rotation matrix and the optimal translation matrix Recorded as the precise registration parameters of the photoresist surface model and the point cloud model.

[0042] Optionally, in step S4, in the point cloud model, calculating the deviation of the point cloud model includes:

[0043] S41: In the point cloud model, determining a reference plane for deviation calculation;

[0044] S42: Calculate the average distance between the points on the surface of the point cloud model and the reference plane to represent the deviation of the point cloud model.

[0045] Optionally, in step S5, based on the precise registration parameters obtained in step S3 and the deviation of the point cloud model obtained in step S4, the flatness of the photoresist surface model is evaluated, including:

[0046] S51: The flatness of the photoresist surface model is expressed as:

[0047] ;

[0048] in, Indicates the flatness of the photoresist surface model, Indicates the deviation of the point cloud model; The optimal rotation matrix representing the photoresist surface model to the point cloud model, The optimal translation matrix representing the photoresist surface model to the point cloud model;

[0049] S52: Flatness normalization of photoresist surface model:

[0050] ;

[0051] in, represents the normalized flatness of the photoresist surface model, Represents the diagonal size of the photoresist surface;

[0052] S53: grading the photoresist surface flatness based on the normalized flatness of the photoresist surface model;

[0053] If the normalized flatness is less than 0.3, the photoresist surface flatness is determined to be first-level flatness, and normal production is carried out;

[0054] If the normalized flatness is greater than or equal to 0.3 and less than or equal to 0.6, the photoresist surface flatness is determined to be level 2 flatness and rework is performed;

[0055] If the normalized flatness is greater than 0.6, the photoresist surface flatness is determined to be level three flatness, and production is stopped for maintenance.

[0056] The present invention also provides a photoresist surface flatness detection system based on model registration, comprising:

[0057] Point cloud model building module: collects radar data, removes noise, and builds a point cloud model;

[0058] Model preliminary registration module: align the photoresist surface model with the point cloud model;

[0059] Model precise registration module: calculates precise registration parameters of photoresist surface model and point cloud model;

[0060] Deviation calculation module: in the point cloud model, calculate the deviation of the point cloud model;

[0061] Flatness evaluation module: Evaluates the flatness of the photoresist surface model based on precise registration parameters and the deviation of the point cloud model.

[0062] Beneficial Effects

[0063] The present invention performs multi-level registration on a photoresist surface model and a point cloud model to improve the registration accuracy; by calculating the deviation of the point cloud model, the flatness detection of the photoresist surface is realized, thereby improving the quality of mask products.

[0064] The present invention facilitates the construction of an accurate point cloud model and better restoration of the actual scene through denoising and splicing operations; adopts singular value decomposition to denoise radar data, which is conducive to improving the quality of radar data; splicing multiple groups of radar data based on overlapping areas is conducive to multi-angle scanning of the photoresist surface model and establishment of a three-dimensional model; applying different colors to each plane is conducive to obtaining a more intuitive point cloud model; and utilizing the significant feature surfaces in the photoresist surface model, the direction alignment of the photoresist surface model and the point cloud model can be achieved more quickly.

[0065] The present invention calculates the deviation between models based on reference points, and updates the process through iterative calculation to ensure that the deviation between models is less than a threshold value, and obtains accurate alignment parameters with a small calculation cost within a limited number of steps; the flatness of the photoresist surface model is normalized and graded, and a relatively concise photoresist surface flatness evaluation method is provided, which is conducive to product quality control and maximizes quality assurance and production. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A schematic flow chart of a method for detecting photoresist surface flatness based on model registration according to an embodiment of the present invention. DETAILED DESCRIPTION

[0067] The present invention is further described below in conjunction with the accompanying drawings, but the present invention is not limited in any way. Any changes or substitutions made based on the teachings of the present invention belong to the protection scope of the present invention.

[0068] Embodiment 1:

[0069] A photoresist surface flatness detection method based on model registration, such as Figure 1 As shown, the following steps are included:

[0070] S1: Use LiDAR to scan the photoresist surface, collect and denoise LiDAR data, and build a point cloud model;

[0071] Step S1 includes:

[0072] S11: Use laser radar to scan the photoresist surface, collect multiple sets of radar data, and perform singular value decomposition on each set of data:

[0073] ;

[0074] in, Represents a set of radar data, and is an orthogonal matrix, It is a diagonal matrix, and the elements on the diagonal are singular values;

[0075] The matrix The elements on the diagonal are arranged in order from largest to smallest, and the value less than the threshold The elements of are set to 0, and the updated diagonal matrix is ​​obtained ; Based on the updated diagonal matrix Get the denoised radar data:

[0076] ;

[0077] in, Represents a set of denoised radar data;

[0078] In the embodiment of the present invention, the threshold It is related to the signal-to-noise ratio of radar data. When the useful signal is weak and the noise is large, the threshold The value is small; when the useful signal is strong and the noise is small, the threshold The value is larger; greater than or equal to the threshold The diagonal elements of are related to the target and are less than the threshold The diagonal elements of are related to the noise.

[0079] S12: Use the overlapping areas to stitch together multiple sets of radar data after denoising:

[0080] ;

[0081] in, Respectively represent the target points in the two sets of radar data, The points representing the overlapping areas of the two sets of radar data, express The distance between express The distance between express The distance between Represents the angle between the planes fitted by the two sets of radar data;

[0082] S13: Draw a three-dimensional model based on the spliced ​​radar data, apply different colors to each surface of the three-dimensional model, and obtain a point cloud model.

[0083] S2: Align the photoresist surface model with the point cloud model;

[0084] Step S2 includes:

[0085] S21: Based on the photoresist object, a proportional geometric model is established, which is the photoresist surface model;

[0086] Select a plane with significant features in the photoresist surface model and construct the normal vector ;

[0087] In the embodiment of the present invention, the plane with significant features is selected from the plane and curved surface at the edge, the plane and curved surface at the center, and the plane and curved surface of special parts.

[0088] S22: In the point cloud model, Set the plane fitted by radar data and construct the normal vector ;

[0089] S23: Traverse the normal vector of each plane of the point cloud model:

[0090] ;

[0091] in, Indicates the maximum number of groups of radar data received;

[0092] Find the normal vector to the photoresist surface model Consistent normal vectors enable orientation alignment between the photoresist surface model and the point cloud model.

[0093] S3: Calculate the precise registration parameters of the photoresist surface model and the point cloud model;

[0094] Step S3 includes:

[0095] Select corresponding reference points in the photoresist surface model and point cloud model;

[0096] Calculate the precise registration parameters of the photoresist surface model and the point cloud model.

[0097] In an embodiment of the present invention, based on the alignment of the directions of the photoresist surface model and the point cloud model, reference surfaces corresponding to the photoresist surface model and the point cloud model are selected; edge points of the corresponding reference surfaces are selected as reference points corresponding to the model; in the photoresist surface model, not all reference points are located in the same plane.

[0098] Among them, the precise registration parameters of the photoresist surface model and the point cloud model are calculated, including:

[0099] S31: Based on the reference point, calculate the deviation from the photoresist surface model to the point cloud model:

[0100] ;

[0101] in, represents the deviation from the photoresist surface model to the point cloud model, represents the first The coordinates of the reference points, Represents the first The coordinates of the reference points, Represents the rotation matrix from the photoresist surface model to the point cloud model, Represents the translation matrix from the photoresist surface model to the point cloud model;

[0102] S32: Based on minimum deviation Principle, using the least squares method, solve the rotation matrix from the photoresist surface model to the point cloud model and translation matrix ;

[0103] S33: The first Coordinates of the reference points Updated to ;

[0104] Based on the update , and , call step S31 to calculate the deviation from the photoresist surface model to the point cloud model ;

[0105] S34: Repeat steps S32 and S33 until Less than the model deviation threshold, at this time the best rotation matrix from the photoresist surface model to the point cloud model is obtained and the optimal translation matrix ;

[0106] The optimal rotation matrix and the optimal translation matrix Recorded as the precise registration parameters of the photoresist surface model and the point cloud model.

[0107] S4: in the point cloud model, calculating the deviation of the point cloud model;

[0108] Step S4 includes:

[0109] S41: In the point cloud model, determining a reference plane for deviation calculation;

[0110] In the embodiment of the present invention: the reference plane is required to include the most point cloud model reference points obtained in step S3.

[0111] S42: Calculate the average distance between the points on the surface of the point cloud model and the reference plane to represent the deviation of the point cloud model.

[0112] S5: Based on the precise registration parameters obtained in step S3 and the deviation of the point cloud model obtained in step S4, the flatness of the photoresist surface model is evaluated;

[0113] Step S5 includes:

[0114] S51: The flatness of the photoresist surface model is expressed as:

[0115] ;

[0116] in, Indicates the flatness of the photoresist surface model, Indicates the deviation of the point cloud model; The optimal rotation matrix representing the photoresist surface model to the point cloud model, The optimal translation matrix representing the photoresist surface model to the point cloud model;

[0117] S52: Flatness normalization of photoresist surface model:

[0118] ;

[0119] in, represents the normalized flatness of the photoresist surface model, Represents the diagonal size of the photoresist surface;

[0120] S53: grading the photoresist surface flatness based on the normalized flatness of the photoresist surface model;

[0121] If the normalized flatness is less than 0.3, the photoresist surface flatness is determined to be first-level flatness, and normal production is carried out;

[0122] If the normalized flatness is greater than or equal to 0.3 and less than or equal to 0.6, the photoresist surface flatness is determined to be level 2 flatness and rework is performed;

[0123] If the normalized flatness is greater than 0.6, the photoresist surface flatness is determined to be level three flatness, and production is stopped for maintenance.

[0124] Embodiment 2: The present invention also provides a photoresist surface flatness detection system based on model registration, comprising the following five modules:

[0125] Point cloud model building module: collects radar data, removes noise, and builds a point cloud model;

[0126] Model preliminary registration module: align the photoresist surface model with the point cloud model;

[0127] Model precise registration module: calculates precise registration parameters of photoresist surface model and point cloud model;

[0128] Deviation calculation module: in the point cloud model, calculate the deviation of the point cloud model;

[0129] Flatness evaluation module: Evaluates the flatness of the photoresist surface model based on precise registration parameters and the deviation of the point cloud model.

[0130] It should be noted that the serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. And the terms "including", "comprising" or any other variants thereof in this article are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "including a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0131] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0132] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A photoresist surface flatness detection method based on model registration, It is characterized in that The method comprises: S1: Use LiDAR to scan the photoresist surface, collect and denoise LiDAR data, and build a point cloud model; S2: Align the photoresist surface model with the point cloud model; S3: Calculate the precise registration parameters of the photoresist surface model and the point cloud model; The step S3 comprises: S31: Select corresponding reference points in the photoresist surface model and the point cloud model; based on the reference points, calculate the deviation from the photoresist surface model to the point cloud model: ； in, represents the deviation from the photoresist surface model to the point cloud model, represents the first The coordinates of the reference points, Represents the first The coordinates of the reference points, Represents the rotation matrix from the photoresist surface model to the point cloud model, Represents the translation matrix from the photoresist surface model to the point cloud model; S32: Based on minimum deviation Principle, using the least squares method, solve the rotation matrix from the photoresist surface model to the point cloud model and translation matrix ; S33: The first Coordinates of the reference points Updated to ; Based on the update , and , call step S31 to calculate the deviation from the photoresist surface model to the point cloud model ; S34: Repeat steps S32 and S33 until Less than the model deviation threshold, at this time the best rotation matrix from the photoresist surface model to the point cloud model is obtained and the optimal translation matrix ; The optimal rotation matrix and the optimal translation matrix Recorded as the precise registration parameters of the photoresist surface model and the point cloud model; S4: in the point cloud model, calculating the deviation of the point cloud model; The step S4 comprises: S41: In the point cloud model, determining a reference plane for deviation calculation; S42: Calculate the average distance between the points on the surface of the point cloud model and the reference plane, indicating the deviation of the point cloud model; S5: Based on the precise registration parameters obtained in step S3 and the deviation of the point cloud model obtained in step S4, the flatness of the photoresist surface model is evaluated; The flatness of the photoresist surface model is expressed as: ； in, Indicates the flatness of the photoresist surface model, Indicates the deviation of the point cloud model.

2. The method for detecting the flatness of a photoresist surface based on model registration according to claim 1, It is characterized in that The step S1 comprises: S11: Use laser radar to scan the photoresist surface, collect multiple sets of radar data, and perform singular value decomposition on each set of data: ； in, Represents a set of radar data, and is an orthogonal matrix, It is a diagonal matrix, and the elements on the diagonal are singular values; The matrix The elements on the diagonal are arranged in order from largest to smallest, and the value less than the threshold The elements of are set to 0, and the updated diagonal matrix is ​​obtained ; Based on the updated diagonal matrix Get the denoised radar data: ； in, Represents a set of denoised radar data; S12: Use the overlapping areas to stitch together multiple sets of radar data after denoising: ； in, Respectively represent the target points in the two sets of radar data, The points representing the overlapping areas of the two sets of radar data, express The distance between express The distance between express The distance between Represents the angle between the planes fitted by the two sets of radar data; S13: Draw a three-dimensional model based on the spliced ​​radar data, apply different colors to each surface of the three-dimensional model, and obtain a point cloud model.

3. The method for detecting the flatness of a photoresist surface based on model registration according to claim 2, It is characterized in that The step S2 comprises: S21: Based on the photoresist object, a proportional geometric model is established, which is the photoresist surface model; Select a plane with significant features in the photoresist surface model and construct the normal vector ; S22: In the point cloud model, Set the plane fitted by radar data and construct the normal vector ; S23: Traverse the normal vector of each plane of the point cloud model: ； in, Indicates the maximum number of groups of radar data received; Find the normal vector to the photoresist surface model Consistent normal vectors enable orientation alignment between the photoresist surface model and the point cloud model.

4. The method for detecting the flatness of a photoresist surface based on model registration according to claim 3, It is characterized in that The step S5 comprises: S51: Flatness normalization of photoresist surface model: ； in, represents the normalized flatness of the photoresist surface model, Represents the diagonal size of the photoresist surface; S52: grading the photoresist surface flatness based on the normalized flatness of the photoresist surface model; If the normalized flatness is less than 0.3, the photoresist surface flatness is determined to be first-level flatness, and normal production is carried out; If the normalized flatness is greater than or equal to 0.3 and less than or equal to 0.6, the photoresist surface flatness is determined to be level 2 flatness and rework is performed; If the normalized flatness is greater than 0.6, the photoresist surface flatness is determined to be level three flatness, and production is stopped for maintenance.

5. A photoresist surface flatness detection system based on model registration, It is characterized in that include: Point cloud model building module: collects radar data, removes noise, and builds a point cloud model; Model preliminary registration module: align the photoresist surface model with the point cloud model; Model precise registration module: calculates precise registration parameters of photoresist surface model and point cloud model; Deviation calculation module: in the point cloud model, calculate the deviation of the point cloud model; Flatness evaluation module: evaluates the flatness of the photoresist surface model based on precise registration parameters and the deviation of the point cloud model; To realize a photoresist surface flatness detection method based on model registration as described in any one of claims 1-4.

Images