CN-115793412-B - Non-imaging semiconductor overlay error measurement device and method

Abstract

The invention discloses a non-imaging semiconductor overlay error measurement device and method, which comprise the steps of building a dark field single-pixel imaging light path, carrying out Fourier frequency domain acquisition on the premise of non-imaging based on the single-pixel imaging light path, obtaining two-dimensional cross-correlation coefficient distribution through calculation of the Fourier frequency domain, carrying out contour center sub-pixel positioning by utilizing the two-dimensional cross-correlation coefficient distribution, and calculating an overlay error through difference between sub-pixel positioning of two contours. The invention realizes that the center of the sub-pixel outline of the detection target is directly positioned and the overlay error is calculated only through light intensity measurement on the premise of no imaging, greatly reduces the data volume generated by graphic operation, improves the measurement efficiency, and expands the application of the single-pixel imaging technology in the field of semiconductor overlay error measurement.

Inventors

• GONG XINGLONG
• NI MINGYANG
• DENG HUAXIA

Assignees

• 中国科学技术大学

Dates

Publication Date

20260512

Application Date

20221215

Claims (4)

1. 1. A non-imaging semiconductor overlay error measuring device is characterized by comprising a dual-mode single-pixel imaging light path and a calculating unit, wherein a Fourier frequency domain of a designated area of a detection target is obtained through the dual-mode single-pixel imaging light path in a non-imaging mode, partial frequency point values of the Fourier frequency domain are directly obtained from a one-dimensional electric signal, subsequent calculation can be carried out on the premise of not imaging, pressure and calculated pressure caused by data acquisition due to imaging and post-processing algorithms are effectively avoided, the calculating unit is used for processing the Fourier frequency domain of the designated area to obtain two-dimensional cross-correlation coefficient distribution, the correlation of the two-dimensional cross-correlation coefficient distribution is the largest under the specific displacement condition, the calculating process only involves simple two-dimensional fast Fourier transformation and sequencing, the calculating complexity is extremely low, the overlay error calculating efficiency is effectively improved, the calculating and the communication pressure are reduced, the cross-correlation peak value is subjected to three-time spline interpolation according to the two-dimensional cross-correlation coefficient distribution, the sub-pixel correlation peak value position is averaged, the sub-pixel precision coordinate of the contour center to be measured is obtained, and the three-time spline and interpolation is used, all the correlation peak value position is completely utilized, and the three-time coordinate and the three-dimensional spline is used for calculating the contour center, and the accuracy of the center of the contour to be measured is the accurate, and the error of the center of the semiconductor is calculated, and the error is calculated, and the accurate, and the error is calculated.
2. 2. The non-imaging semiconductor overlay error measurement method is characterized by comprising the following steps of: S1, acquiring a Fourier frequency domain of a designated area of a target to be detected by utilizing a bimodal single-pixel imaging light path and combining a Fourier single-pixel imaging principle under the non-imaging premise, directly acquiring partial frequency point values of the Fourier frequency domain from one-dimensional electric signals, and carrying out subsequent calculation under the premise of not imaging, thereby effectively avoiding the pressure of data acquisition and the pressure of calculation caused by imaging and a post-processing algorithm; S2, obtaining two-dimensional cross-correlation coefficient distribution according to a Fourier frequency domain of the appointed area of the target to be detected, wherein the correlation of the two-dimensional cross-correlation coefficient distribution is maximum under a specific displacement condition, the calculation process only involves simple two-dimensional fast Fourier transform and sequencing, the calculation complexity is extremely low, the calculation efficiency of overlay error is effectively improved, and the calculation and communication pressure is reduced; S3, carrying out cubic spline interpolation on the cross correlation peak value according to the two-dimensional cross correlation coefficient distribution, finding out the position of the sub-pixel correlation peak value, and averaging the position of the sub-pixel peak value to obtain the sub-pixel precision coordinate of the center of the outline of the target to be detected, wherein the three-dimensional spline and the interpolation and the averaging are used, so that all acquired two-dimensional correlation data are completely utilized, and the accuracy is high; and S4, calculating the sub-pixel precision coordinates of the contour center of the target to be measured, namely, calculating the difference between the contour center coordinates of the front layer graph and the current layer graph, and obtaining the semiconductor overlay error.
3. 3. The non-imaging semiconductor overlay error measurement apparatus of claim 1, wherein: the bimodal single pixel imaging light path includes: The system comprises an active single-pixel bright field imaging module, a passive single-pixel dark field imaging module, a light splitter, an active single-pixel bright field imaging module, a passive single-pixel dark field imaging module, a detection target detection module and a light splitter, wherein the light splitter is used for switching between the active single-pixel bright field imaging module and the passive single-pixel dark field imaging module; the imaging light path further comprises a bright field light source, a dark field light source, a spatial light modulator, an imaging lens, a concave reflecting mirror, a beam splitter, a first photoelectric sensor and a second photoelectric sensor, wherein the beam splitter is positioned between the bright field light source and the second photoelectric sensor, forms a 45-degree beam splitting light path with the bright field light path and the dark field light path at a beam splitting interface, and the concave reflecting mirror is used for adjusting the focal plane position and changing the light path direction so as to improve the compactness of the device; in the active bright field single-pixel imaging module, bright field light is emitted from a bright field light source and is incident to a concave reflector through a light splitter, the concave reflector converges and reflects the bright field light to a spatial light modulator, the spatial light modulator modulates the light to obtain a structural light field, the structural light field is projected to the surface of a detection target through an imaging lens, the structural light field reflected by the surface of the detection target reaches a target surface of a first photoelectric sensor, the first photoelectric sensor measures the total intensity of the structural light field and calculates a Fourier frequency point value of a designated position of a bright field Fourier frequency domain through the correlation between the intensity of the structural light field and the structural light field, and the bright field Fourier frequency domain of the detection target is obtained through frequency domain scanning; The passive dark field single-pixel imaging module comprises a dark field light source, a spatial light modulator, a concave reflector, a beam splitter and a second photoelectric sensor, wherein dark field light emits from the dark field light source to carry out dark field illumination on a detection target and diffuse reflection occurs on the surface of the detection target, the diffuse light is imaged on the spatial light modulator through the imaging lens, the spatial light modulator modulates the light to obtain a structural light field, the structural light field is converged and reflected by the concave reflector and then enters the beam splitter beam splitting interface and is reflected to the target surface of the second photoelectric sensor through the beam splitting interface, the second photoelectric sensor measures the total intensity of the structural light field and calculates the Fourier frequency point value of a designated position of a dark field Fourier frequency domain through the correlation between the structural light field intensity and the structural light field, and the dark field Fourier frequency domain of the detection target is obtained through frequency domain scanning.
4. 4. The method for measuring overlay error of a non-imaging semiconductor as recited in claim 2, wherein the bimodal single pixel imaging optical path comprises: The system comprises an active single-pixel bright field imaging module, a passive single-pixel dark field imaging module, a light splitter, an active single-pixel bright field imaging module, a passive single-pixel dark field imaging module, a detection target detection module and a light splitter, wherein the light splitter is used for switching between the active single-pixel bright field imaging module and the passive single-pixel dark field imaging module; the imaging light path further comprises a bright field light source, a dark field light source, a spatial light modulator, an imaging lens, a concave reflecting mirror, a beam splitter, a first photoelectric sensor and a second photoelectric sensor, wherein the beam splitter is positioned between the bright field light source and the second photoelectric sensor, forms a 45-degree beam splitting light path with the bright field light path and the dark field light path at a beam splitting interface, and the concave reflecting mirror is used for adjusting the focal plane position and changing the light path direction so as to improve the compactness of the device; in the active bright field single-pixel imaging module, bright field light is emitted from a bright field light source and is incident to a concave reflector through a light splitter, the concave reflector converges and reflects the bright field light to a spatial light modulator, the spatial light modulator modulates the light to obtain a structural light field, the structural light field is projected to the surface of a detection target through an imaging lens, the structural light field reflected by the surface of the detection target reaches a target surface of a first photoelectric sensor, the first photoelectric sensor measures the total intensity of the structural light field and calculates a Fourier frequency point value of a designated position of a bright field Fourier frequency domain through the correlation between the intensity of the structural light field and the structural light field, and the bright field Fourier frequency domain of the detection target is obtained through frequency domain scanning; The passive dark field single-pixel imaging module comprises a dark field light source, a spatial light modulator, a concave reflector, a beam splitter and a second photoelectric sensor, wherein dark field light emits from the dark field light source to carry out dark field illumination on a detection target and diffuse reflection occurs on the surface of the detection target, the diffuse light is imaged on the spatial light modulator through the imaging lens, the spatial light modulator modulates the light to obtain a structural light field, the structural light field is converged and reflected by the concave reflector and then enters the beam splitter beam splitting interface and is reflected to the target surface of the second photoelectric sensor through the beam splitting interface, the second photoelectric sensor measures the total intensity of the structural light field and calculates the Fourier frequency point value of a designated position of a dark field Fourier frequency domain through the correlation between the structural light field intensity and the structural light field, and the dark field Fourier frequency domain of the detection target is obtained through frequency domain scanning.

Description

Non-imaging semiconductor overlay error measurement device and method Technical Field The invention relates to a non-imaging semiconductor overlay error measuring device and a method, belonging to the field of microelectronic and semiconductor integrated circuit manufacturing. Background Semiconductor chip technology is a pulse for national economic development and information security, and photolithography is a key step in the manufacture of semiconductor integrated circuits. It moves the mask pattern onto the wafer through a series of repeated processes of alignment, exposure and development, etc. And the pattern left on the photoresist after exposure and development (current layer pattern) must be aligned with the existing pattern of the wafer substrate (previous layer pattern). Overlay error (overlay), i.e., the offset of the upper and lower layers of the wafer, must be sufficient to ensure proper connection between the different parts of the device. The overlay error is too large, which is a main cause of short circuit and disconnection of devices, so that the overlay error measurement has important significance and application value for improving the yield of semiconductor production and manufacturing. However, the vast majority of overlay error checking devices are imported, and key technologies are limited by people, so that the benign development of the semiconductor industry in China is seriously threatened. In integrated circuit manufacturing processes, overlay errors are typically determined by detecting the relative position between the current layer pattern and the previous layer pattern of the wafer after lithography using specialized equipment. The overlay error quantitatively describes the deviation of the current pattern relative to the previous pattern in the X and Y directions and the distribution of the deviation on the wafer, and is one of key indexes for checking the quality of the photoetching process. The patterns on the wafer that are used exclusively for calculating overlay errors are overlay marks, which are typically at the edges of the exposure unit. Currently, techniques commonly used in the field of semiconductor surface overlay error measurement technology include overlay error measurement (IBO) based on image recognition technology and overlay error measurement (DBO) based on diffraction. The overlay error measurement based on the image recognition technology mainly comprises the steps of manufacturing specific overlay marks on a current layer and a front layer by using a photoetching process, enabling the two layers of overlay marks to be non-overlapped and easy to calculate center coordinates, performing high-precision imaging on a semiconductor appointed overlay mark area according to a planned route through proper selection of a camera, a lens and a light source, obtaining a high-resolution image of a single measurement object through technologies such as denoising, rotation correction, trimming, picture segmentation and image interpolation amplification of graphics, performing algorithm processing on a double-layer overlay mark image to obtain overlay mark contour center coordinates, and obtaining overlay errors through calculating center coordinate differences of the double-layer overlay marks. However, the technology has the following problems that 1, the processing time is obviously increased for obtaining a high-resolution local image processing flow, the detection efficiency is reduced, 2, huge data flux is generated by using an intelligent algorithm for alignment of overlay marks, huge pressure is generated on a communication server used for production, the high-end server is a weight held by foreign patents and is easy to become a new neck clamping problem, and 3, the problems of insufficient efficiency and overlarge data pressure generated by the prior art are more and more serious along with the reduction of a semiconductor manufacturing process and the increase of the spatial resolution of a camera. The alignment error measurement based on diffraction mainly comprises the steps of manufacturing diffraction grating type photoetching marks on a current layer and a reference layer by using a photoetching process, wherein the two layers of marks are of a periodic structure, illuminating an alignment mark area by using a light source, measuring +1 and-1 diffraction light intensities under illumination, and calculating alignment conditions among different process layers by measuring the light intensity difference by using a good linear relation formed by the light intensity difference and the alignment error value when the alignment error exists. The DBO mark has the advantages of small measurement error and high measurement repeatability compared with the IBO overlay mark. The overlay error measurement based on diffraction has the following problems, and related patent refers to that the related patent is that the company of ASML in the netherlands monopolizes and