CN-120274670-B - Longitudinal large-scale three-dimensional morphology measurement method based on non-uniform sparse sampling

CN120274670BCN 120274670 BCN120274670 BCN 120274670BCN-120274670-B

Abstract

The invention discloses a longitudinal large-scale three-dimensional morphology measurement method based on non-uniform sparse sampling, belongs to the field of optical detection, and meets the requirement of millimeter-level longitudinal large-scale microstructure rapid detection. First, the number of interferograms is greatly reduced by non-uniformly spaced sampling on the order of microns. On the basis, the non-uniform sparse sampling interference signal is reconstructed into a uniform interference signal through linear interpolation and sparse reconstruction. And finally, accurately restoring the three-dimensional morphology information of the sample by using a sampling envelope algorithm. The invention improves the interference pattern acquisition efficiency and reduces the interference pattern processing data volume through high-speed acquisition, and the recovery error of the measuring method is less than one percent while effectively improving the measuring speed. The invention is suitable for realizing the three-dimensional shape rapid detection of the longitudinal large-scale microstructure sample in the fields of semiconductor manufacturing, optical processing, mechanical processing and the like.

Inventors

GAO ZHISHAN
Kuang Junhao
GUO ZHENYAN
YUAN QUN
LUO TAO
YU QIQI
LI JINYU

Assignees

南京理工大学

Dates

Publication Date: 20260512
Application Date: 20250331

Claims (5)

1. A longitudinal large-scale three-dimensional morphology measurement method based on non-uniform sparse sampling is characterized by comprising the following steps: Step 1, detecting samples by adopting a Linnik type interference light path and a laser ranging light path together, controlling the two light paths by the same motor to perform synchronous motion, synchronously triggering the two light paths by a motor emission signal, acquiring an interference pattern of the samples by a near infrared interference light path through a CCD camera, calculating position information of the interference pattern by the laser ranging light path, and setting the target surface of the CCD camera to be common A plurality of pixel points; Step 2, determining the motor scanning speed as The camera frame rate is According to at least within the coherence length of the light source Calculating the bandwidth of the light source at the moment according to the requirements of the sampling points and combining a light source coherence length formula, and placing a matched optical filter at a light outlet of the light source; Step 3, sparse sampling is carried out on the CCD camera in scanning movement to obtain a first interference signal intensity matrix The size is Synchronously calculating the distance between corresponding sampling points by using a laser ranging light path to obtain a first position information matrix The size is ; Step 4, for the first position information matrix Performing linear interpolation to obtain a second position information matrix The size is H1 represents the number of sampling points after linear interpolation and is combined with a first interference signal intensity matrix And a second position information matrix After secondary interpolation, a third interference signal intensity matrix is obtained And a third position information matrix The sizes are respectively And ; Step 5, combining the third interference signal intensity matrix And a third position information matrix Deducing a sparse reconstruction formula according to the bandpass sampling theorem and the interference principle to obtain a fourth interference signal intensity matrix And a fourth positional information matrix Wherein the fourth interference signal strength matrix Is of the size of Fourth positional information matrix Is of the size of H2 represents the number of sampling points after the sparse signal is restored; Step 6, combining a fourth interference signal intensity matrix And a fourth positional information matrix According to the sampling envelope theorem, a sampling envelope data matrix is obtained The size is For sampling envelope data matrix Obtaining a morphology data matrix using centroid method The size is 。
2. The method for measuring longitudinal large-scale three-dimensional morphology based on non-uniform sparse sampling according to claim 1, wherein in step 2, the light source coherence length L is expressed as follows: , Wherein, the Is the center wavelength of the light source, Is the bandwidth of the light source.
3. The method for measuring longitudinal large-scale three-dimensional morphology based on non-uniform sparse sampling according to claim 1, wherein step 4, for the first position information matrix Performing linear interpolation to obtain a second position information matrix The size is H1 represents the number of sampling points after linear interpolation and is combined with a first interference signal intensity matrix And a second position information matrix Performing secondary interpolation to obtain a third interference signal intensity matrix And a third position information matrix The method is characterized by comprising the following steps: For the first position information matrix Obtaining a second position information matrix by linear interpolation The size is The calculation formula is as follows: , Wherein, the Is a second position information matrix The interval value of two adjacent elements in the sequence, For the last element value of the first position information matrix Z, Is the first element value of the first location information matrix Z; combining the second position information matrix With a first interference signal intensity matrix Performing secondary interpolation processing, wherein the calculation formula is as follows: , Wherein, the For the ith element value on the pixel points of the x row and the y column in the first interference signal intensity matrix A, For the i-th element value in the first position information matrix Z, The j-th element value in the second position information matrix X; Finally, a third interference signal intensity matrix is obtained The size is The elements are And set up a third position information matrix Equal to the second position information X of the size of 。
4. The non-uniform sparse sampling based longitudinal large scale three dimensional morphology measurement method of claim 1 wherein in step 5, a third interference signal intensity matrix is combined And a third position information matrix Deducing a sparse reconstruction formula according to the bandpass sampling theorem and the interference principle to obtain a fourth interference signal intensity matrix And a fourth positional information matrix Wherein the fourth interference signal strength matrix Is of the size of Fourth positional information matrix Is of the size of The method is characterized by comprising the following steps: the improved sparse signal recovery calculation formula is as follows: , In the abscissa of The range of the values is as follows Ordinate of The range of the values is as follows Set fourth position information matrix The range of the values of the elements is Is common to The value of the individual element(s), I.e. representing the element values in the fourth positional information matrix V; Is an element value in the third positional information matrix Q; min (Q) is the minimum value in the third position information matrix Q, and max (Q) is the maximum value in the third position information matrix Q; Is a second position information matrix The interval value of two adjacent elements.
5. The method for measuring longitudinal large-scale three-dimensional morphology based on non-uniform sparse sampling according to claim 4, wherein in step 6, a fourth interference signal intensity matrix is combined And a fourth positional information matrix According to the sampling envelope theorem, an envelope data matrix is obtained The size is For sampling envelope data matrix Obtaining a morphology data matrix using centroid method The method is characterized by comprising the following steps: The calculation formula of the sampling envelope theorem is as follows: , In the abscissa of The range of the values is as follows Ordinate of The range of the values is as follows Coefficient n is And' "Means that the whole is rounded down, I.e. represents an element in the fourth positional information matrix V; For sampling envelope data matrix Obtaining a morphology data matrix using centroid method 。

Description

Longitudinal large-scale three-dimensional morphology measurement method based on non-uniform sparse sampling Technical Field The invention belongs to the field of optical detection, and particularly relates to a longitudinal large-scale three-dimensional morphology measurement method based on non-uniform sparse sampling. Background With the development of the fields of semiconductor manufacturing, optical processing, mechanical processing and the like, various structures can be produced in batches, so that the white light interferometer is required to be capable of rapidly detecting the three-dimensional profile of the sample surface in batches in the industrial field, the detection efficiency is improved, and the current white light interferometer is challenged. As is well known, scanning white light interferometers are widely used for three-dimensional profile measurement of a sample surface, and the basic principle is that three-dimensional information of the sample is obtained by driving and changing an optical path difference between two arms through piezoelectric ceramics. Although the scanning white light interferometer has the characteristic of high measurement precision, the sampling interval required by the algorithm is required to be the nyquist sampling theorem (eighth wavelength), and uniform sampling is required. For millimeter-scale (large-scale) samples, the traditional white light algorithm needs to take thousands of pictures, and is large in calculation amount and long in sampling time. Meanwhile, the piezoelectric ceramic has a movement stroke of only hundred micrometers, and measurement cannot be realized on a millimeter-sized sample. Aiming at the problem of low detection speed of the existing white light interferometer, related scholars research a sparse sampling strategy, the Ogawa provides a square envelope algorithm in Sampling Theorem for Surface Profiling by White-Light Interferometry, and the calculation cost is saved, but the method is realized by uniformly sampling a micron-sized sample driven by piezoelectric ceramics. For millimeter-sized samples, non-uniform sampling needs to be considered and a higher speed motor with a larger stroke is used. Under the background of driving by a high-speed motor, the invention realizes the rapid three-dimensional morphology measurement of a longitudinal large-scale sample by further expanding the sampling interval and a non-uniform sparse reconstruction method. Disclosure of Invention The invention aims to provide a longitudinal large-scale rapid three-dimensional morphology measurement method based on non-uniform sparse sampling, so as to solve the problem of the detection speed of a wide-field large-scale three-dimensional morphology. The technical scheme for realizing the purpose of the invention is that the longitudinal large-scale rapid three-dimensional morphology measurement method based on non-uniform sparse sampling comprises the following steps: And 1, detecting a sample by combining the Linnik type interference light path and the laser ranging light path. The two sets of light paths are controlled by the same motor to perform synchronous movement. The CCD camera samples the interference pattern in the near infrared interference light path, the laser ranging light path calculates the position information of the interference pattern, both are synchronously triggered by the motor emission signal, and the target surfaces of the CCD camera are shared And a pixel point. Step 2, determining the scanning speed of the motorAnd CCD camera frame rateAccording to at least within the coherence length of the light sourceThe requirement of the sampling points is calculated, so that the bandwidth of the light source at the moment is calculated according to a coherent formula of the light source, and a proper optical filter is replaced in the light source. Step 3, sparse sampling is carried out in the scanning movement according to the CCD camera, and a first interference signal intensity matrix is obtainedThe size isH represents the number of sampling points, and the distance between the corresponding sampling points is synchronously calculated by the laser ranging light path to obtain a first position information matrixThe size is。 Step 4, for the first position information matrixPerforming linear interpolation to obtain a second position information matrixThe size isH1 represents the number of sampling points after linear interpolation, and then the first interference signal intensity matrix A and the second position information matrix are combinedPerforming secondary interpolation to obtain a third interference signal intensity matrixAnd a third position information matrixThe sizes are respectivelyAnd。 Step 5, combining the third interference signal intensity matrixAnd a third position information matrixDeducing a sparse reconstruction formula according to the bandpass sampling theorem and the interference principle to obtain a fourth interf