A point-diffraction common-path interferometric wafer pattern defect detection device and method

Abstract

The application discloses a kind of point diffraction common light path interference wafer pattern defect detection device and method, including light source incident light path, signal collection light path, signal filtering interference light path, wafer displacement system and refrigeration CCD camera, by the precise movement of the wafer to be detected in X direction, interference image in each position is obtained;Each interference image is phase reconstructed, and the phase image related to the surface topography of the wafer to be detected is obtained;The continuous phase image is moved average processing and second-order difference processing, and spliced into a large range image, the processed image is different from reference standard image, and the image abnormal area is obtained;The image of the determined abnormal area is compared with actual wafer size and mapped, and the wafer pattern accurate defect area range is obtained.The application uses the phase information of optical signal to detect wafer pattern defect, improves detection resolution and signal-to-noise ratio and smaller detection scale.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor chip defect detection and morphology detection, and specifically relates to a device and method for detecting wafer pattern defects by point diffraction common-path interference. Background Technology

[0002] Patterned wafers play a crucial role in semiconductor manufacturing. As a key early-stage product, the pattern structure and size on them largely determine the quality and performance of the final chip. With the continuous development of technology, wafer pattern structures are evolving towards smaller and more complex designs. Therefore, comprehensive, efficient, and accurate inspection of patterned wafers after complex processing has become particularly critical.

[0003] Patent document CN 115325963 A, published on November 11, 2022, discloses a wafer surface three-dimensional topography measurement device and method. This device employs a white-light interferometry method based on a Michelson interferometer, utilizing a large-field-of-view objective lens to perform full-area scanning measurement of the patterned wafer. It achieves high-efficiency and high-precision measurement and evaluation of the wafer surface through fast reconstruction and image stitching algorithms. However, because this device uses Michelson interferometry, it requires vertical scanning at each measurement point to satisfy the sampling theorem, thus reducing measurement efficiency. Simultaneously, the increased sampling time amplifies the impact of environmental factors on the measurement results, leading to inaccuracies. Furthermore, the large-field-of-view interferometer objective lens has a low magnification and small numerical aperture, limiting the measurement resolution and failing to meet the requirements of advanced process patterned wafer inspection.

[0004] Patent document CN 115598129 A, published on January 13, 2023, discloses a wafer defect detection system. This system employs a laser defect detection method, simultaneously receiving scattered, fluorescent, reflected, and transmitted optical signals from defects, and introducing polarization during the detection process to improve the signal-to-noise ratio. High-sensitivity full-area defect detection is achieved by scanning the wafer horizontally. However, both laser scattering intensity and imaging resolution are strongly correlated with laser wavelength; therefore, to improve detection resolution and signal-to-noise ratio, a shorter wavelength laser source is required. Summary of the Invention

[0005] To address the aforementioned deficiencies in the existing technology, the present invention aims to provide a point diffraction common-path interference wafer pattern defect detection device and method. This method employs a common-path interference method to detect wafer pattern defects, using non-short-wavelength laser light sources, or even broadband white light sources, to improve detection resolution, signal-to-noise ratio, and smaller detection scale.

[0006] The present invention is achieved through the following technical solution.

[0007] According to one aspect of the present invention, a point diffraction common-path interference wafer pattern defect detection device is provided, comprising:

[0008] The incident light path of the light source includes a non-short wavelength laser light source or a broadband white light light source, a focusing lens, an intermediate lens and a field lens, which are used to shape the light source beam and incident it onto the wafer inspection area.

[0009] The signal collection optical path includes an objective lens, a beam splitter, and a tube lens, used to collect the light signal reflected from the surface of the wafer to be inspected;

[0010] The signal filtering interference optical path, including a transmission diffraction grating, a Fourier lens, and a spatial filter, is used to diffract and separate the reflected light signal from the wafer to be tested and then filter it through a 4f system.

[0011] The wafer displacement system includes X, Y, and Z-axis precision displacement stages for scanning the wafer to be inspected in the horizontal and vertical directions.

[0012] A cooled CCD camera is used to image the two beams of light—the reference beam from the incident light source and the object beam reflected from the wafer—onto the target surface of the CCD camera through beam splitting and interference.

[0013] According to another aspect of the present invention, a method for detecting defects in a wafer pattern using point diffraction co-path interference is provided, comprising:

[0014] Prepare a wafer to be tested, which has a periodic repeating structure and an abnormal defect structure pattern.

[0015] By precisely moving the wafer under test in the X-axis, interference images are obtained at each position;

[0016] Phase reconstruction is performed on each interference image to obtain a phase image related to the surface morphology of the wafer to be inspected;

[0017] Multiple consecutive phase images are processed by second-order difference and moving average, stitched together to form a large-scale image, and time-varying noise of the system is removed to obtain the processed image;

[0018] The processed image is compared with the reference standard image to obtain the abnormal regions in the image;

[0019] By mapping the image of the identified abnormal area to the actual wafer size, the precise defect area range of the wafer pattern can be obtained.

[0020] The present invention, by adopting the above technical solution, has the following beneficial effects:

[0021] 1. The common-path interference defect detection technology used in this invention uses a grating to sample the reflected light from the wafer, avoiding the use of inefficient scanning sampling methods. Only one interference image is needed to collect information about the wafer surface. Compared with vertical scanning white light interferometry, it reduces the influence of environmental factors during the sampling process on the detection results and improves the robustness of the detection.

[0022] 2. The common-path interference defect detection technology used in this invention is more sensitive to the structural height of the wafer surface because it uses the phase information of the optical signal. Compared with the laser scattering defect detection technology, it does not lose detection efficiency, but does not require the use of an ultra-short wavelength laser source. It can achieve a higher signal-to-noise ratio and a smaller detection scale using a visible light source, thus improving the availability of the system. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, do not constitute an undue limitation of the invention. In the drawings:

[0024] Figure 1 This is a schematic diagram of the optical path structure of the common optical path interference wafer pattern defect detection device of the present invention;

[0025] Figure 2 This is a schematic diagram of the spatial filter in the common-path interference wafer pattern defect detection device of the present invention;

[0026] Figure 3 This is a flowchart of the common-path interference wafer pattern defect detection method of the present invention.

[0027] Wherein: 110 - incident light source path, 120 - signal collection path, 130 - signal filtering interference path, 140 - wafer shifting system, 151 - cooled CCD camera;

[0028] 101-Wafer to be inspected, 111-Light source, 112-Focusing lens, 113-Intermediate lens, 114-Field lens, 121-Objective lens, 122-Beam splitter, 123-Tube lens, 131-Transmission diffraction grating, 132-First Fourier lens L1, 133-Spatial filter, 134-Second Fourier lens L2, 141-X-axis precision displacement stage, 142-Y-axis precision displacement stage, 143-Z-axis precision displacement stage. Detailed Implementation

[0029] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention, but are not intended to limit the present invention.

[0030] like Figure 1As shown, this embodiment provides a point diffraction common-path interference wafer pattern defect detection device, including a light source incident light path 110, a signal collection light path 120, a signal filtering interference light path 130, a wafer displacement system 140, and a research-grade cooled CCD camera 151.

[0031] The incident light path 110 shapes the non-short wavelength laser light source or broadband white light source 111 and incident it onto the wafer detection area; the signal collection light path 120 collects the reflected light signal from the wafer surface; the signal filtering interference light path 130 diffracts and splits the reflected light signal from the wafer and filters it through a 4f system, and finally the two beams of light from the incident light path and the reflected light from the wafer are interferometrically imaged onto the target surface of the CCD camera 151.

[0032] In this embodiment, white light is used as the incident light source 111. The light source incident light path 110 adopts a Kohler illumination system, including a focusing lens 112, an intermediate lens 113 and a field lens 114. The incident light source 111 enters the signal collection light path 120 through the light source incident light path 110. Kohler illumination reduces the influence of the light source on the imaging quality, while improving the uniformity of white light illumination and increasing the coherence length of the white light broadband light source.

[0033] The signal collection optical path 120 includes an objective lens 121, a beam splitter 122, and a tube lens 123. The beam splitter 122 reflects the incident light from the light source into the objective lens 121 to illuminate the surface of the wafer 101 to be inspected. At the same time, the objective lens 121 collects the reflected light from the wafer into the signal collection optical path 120. The signal light is transmitted through the beam splitter 122 and imaged on the focal plane of the tube lens 123.

[0034] The signal filtering interference optical path 130 includes a transmission diffraction grating 131, a first Fourier lens L1 132, a spatial filter 133, and a second Fourier lens L2 134. The transmission diffraction grating 131, located on the focal plane of the tube lens 123, diffracts the object light from the wafer surface. The first Fourier lens L1 132 collects the diffracted light and focuses it onto the focal plane of the first Fourier lens L1 132, i.e., the object light spectrum plane. The spatial filter 133, located on this plane, filters the object light diffracted light in the spectrum, allowing the +1st order diffracted light to pass completely as the object light containing wafer surface topography information, while the 0th order diffracted light undergoes low-pass filtering, allowing only the DC component to pass as the reference plane light. The reference light and object light passing through the spatial filter 133 are focused by the second Fourier lens L2 134 onto the target surface of a research-grade cooled CCD camera 151, achieving interference imaging. Figure 2 A schematic diagram of a spatial filter is shown.

[0035] The wafer displacement system 140 includes an X-axis precision displacement stage 141, a Y-axis precision displacement stage 142, and a Z-axis precision displacement stage 143. The precision displacement stages are mainly responsible for scanning the wafer in the horizontal and vertical directions to improve the efficiency and sensitivity of defect detection.

[0036] like Figure 3 As shown, the present invention also provides a method for detecting defects in wafer patterns using point diffraction common-path interference, comprising the following steps:

[0037] Step 301: Prepare a wafer to be inspected, which has a periodic repeating structure and an abnormal defect structure pattern; use X and Y precision displacement stages to precisely move the wafer to be inspected in the horizontal direction with a step size of 0.75μm, and acquire an interferogram at each position.

[0038] In this embodiment of the invention, a precision wafer displacement stage is controlled to drive the wafer to move precisely by a small distance, with each movement being 0.75 μm. At the same time, the aforementioned optical path is used to acquire a common optical path interference image, and the subsequent defect detection results are obtained from this image.

[0039] In this embodiment of the invention, the patterned wafer to be tested is a wafer after pattern photolithography during chip processing. It has a pattern designed during chip design and normal periodic system structural defects and abnormal random structural defects generated during photolithography. In this invention, a robotic arm is used to stably place the patterned wafer to be tested into a fixed position on a precision air-bearing motion platform.

[0040] Step 302: Perform phase reconstruction on each interference image to obtain a phase image related to the wafer surface morphology.

[0041] Specifically, a two-dimensional Fourier transform is used on the interference image to extract the reference light signal and the object light signal; a two-dimensional inverse Fourier transform is used on the object light signal to obtain amplitude and phase information; and a two-dimensional discrete cosine transform and inverse transform are used on the phase information to obtain a phase image related to the surface morphology of the wafer to be tested.

[0042] In this embodiment of the invention, the reference beam is uniformly distributed on the camera plane after being low-pass filtered by a pinhole. It can be written as...

[0043]

[0044] Where A0 and Φ0 are constants, and i is the imaginary unit.

[0045] The signal beam C(x,y) containing sample information becomes, after being modulated by a diffraction grating, a signal beam C(x,y) containing sample information becomes...

[0046] U1(x,y)=C(x,y)e iβx(2) In the formula, β is the spatial frequency, β=2π / A; Λ is the grating period in the image plane; C(x,y) is the reflection amplitude and phase information of the sample.

[0047] The irradiance caused by the interference of the reference beam and the signal beam is:

[0048]

[0049] In the formula, C * (x,y) is the complex conjugate of C(x,y).

[0050] Equation (3) can be used to retrieve sample information C(x,y) using Hilbert or Fourier transform methods. Here, the Fourier transform method is used. In the Fourier domain, equation (3) becomes...

[0051]

[0052] In the formula, For k x and k y Two-dimensional convolution, δ(k) x ,k y ) is a two-dimensional Dirac function, representing a specific point (k) in the frequency domain. x ,k y ), Let C(x,y) be the Fourier transform result, and C... * (-k x ,-k y ) is C * Frequency domain representation of (x,y) C * (-k x ,-k y The Fourier transform result of δ(k) x +β,k y ), δ(k x -β,k y ) are the midpoints of the frequency domain (k x +β,k y ) and point (k) x -β,k y ).

[0053] The second, third, and fourth terms on the right-hand side of formula (4) will create three orders, each containing sample information. The second term is of order 0, but it overlaps with the first term and is not suitable for retrieving sample information. The third term (-1 order) and the fourth term (+1 order) are respectively along k xThe directional frequency shifts are -β and β, so either one can be chosen to retrieve sample information. The +1 order is selected by multiplying equation (4) by a circle function, the radius of which is determined by the cutoff frequency of the microscope objective. After selecting the +1 order, moving it back to the origin, we get:

[0054]

[0055] Now, performing an inverse Fourier transform on equation (5) yields the spatial domain sample information C(x,y). Since the sample is illuminated by a plane wave, if only the reflection from the sample surface is considered, it can be simply written as:

[0056] C(x,y)=|R(x,y)|e iφ(x,y) (6)

[0057] In the formula, R(x,y) is the reflection coefficient, and φ(x,y) is the phase distribution caused by the change in the height h(x,y) of the sample surface.

[0058] In reflection mode, the relationship between φ(x,y) and h(x,y) is as follows:

[0059]

[0060] In the formula, λ is the wavelength of the light source.

[0061] Finally, based on the obtained object light complex field, its arctangent is calculated:

[0062]

[0063] In the formula, Im represents the imaginary part; Re represents the real part; and U(x,y) represents the object light complex field.

[0064] The phase distribution on the sample surface can be solved. According to formula (7), the phase delay can be recovered as the height information of the three-dimensional sample, thereby reconstructing the three-dimensional morphology of the sample surface.

[0065] Step 303: Perform second-order difference processing and moving average processing on multiple continuous phase images, stitch them together into a large-scale image, and remove the time-varying noise of the system.

[0066] Specifically, a second-order difference processing is performed on every three adjacent phase images to obtain a second-order difference image with time-invariant noise eliminated; a moving average processing is applied to all second-order difference images to obtain a panoramic stitched image with time-invariant noise eliminated.

[0067] In this embodiment of the invention, after scanning and sampling with a fixed step size, the mathematical description of the nth sampled image can be written as:

[0068]

[0069] in

[0070]

[0071] In the formula, L(x,y) represents the image generated by the patterned wafer basic structure and various defects, ds is the step size between adjacent image frames, P(x,y) is the point spread function of the imaging system, and N s (x,y) is additive time-invariant spatial noise, N i (x,y) is the additive time-invariant system noise, N t (x,y,t n F is additive spatial noise that changes over time. x and F y These represent the scale ranges of the function along the x and y axes, respectively, where W(x,y) is the camera's scale in the x and y directions, respectively. x and F y The field-of-view window function, S(x,y), is the multiplicative time-invariant noise, i.e., the system response function, used to represent the spatial variation of illumination intensity, illumination path length, or camera response. (Symbols omitted) This represents a two-dimensional convolution.

[0072] Due to sample translation, the image structure shifts by n·ds on x. To completely remove additional time-invariant spatial noise and time-invariant system noise, second-order difference image frames are calculated. The nth frame image can be described as:

[0073]

[0074] The second-order difference process suppresses the high spatial frequency components of the phase signal, while favoring the low spatial frequency components. The high frequency components mainly come from the noise of the underlying structure of the wafer, while the low frequency defect signals are significantly highlighted.

[0075] Additive spatial noise that varies with time at each image capture time t n Because of random spatial variations, the best method for removing temporal noise is averaging. During image stitching, averaging is performed on the overlapping regions to remove the time-varying spatial noise term, resulting in...

[0076]

[0077] in

[0078]

[0079] Where G(x,y) is the panoramic stitched image of the second sequence of different chips, N is the total number of steps, and S′(y) is the average value of S′(x,y) / x.

[0080] The field of view W′(x,y) of the panoramic image is further increased by n·ds in the x direction. Furthermore, since the system response S(x,y) has already been averaged in the x direction, the multiplicative noise is significantly reduced. By using the translation invariance of convolution, equation (12) can be equivalently written as:

[0081]

[0082] in

[0083] K(x,y)=P(x-ds,y)-2P(x,y)+P(x+ds,y)(15)

[0084] L′(x,y) is the image of the stitched structure after slight cropping at both ends in the x direction, and K(x,y) is the second-order difference representation of the point spread function P(x,y) of the imaging system.

[0085] Step 304: After obtaining the stitched image, perform a difference analysis between the acquired image and the reference standard image. Use a threshold algorithm and set the threshold to A. If the difference pixel value is greater than A, mark it as a white abnormal area; if the difference pixel value is less than A, mark it as black.

[0086] Specifically, a normalized cross-correlation method is used to obtain the position with the highest correlation value between the panoramic image and the reference standard image; then, the panoramic image and the reference standard image are differentially processed at the position with the highest correlation value to obtain the image anomaly region.

[0087] In this embodiment of the invention, the correlation value between adjacent wafer images is calculated using normalized cross-correlation:

[0088]

[0089] Among them, T mean I represents the average value of the template image T. mean Let R(x,y) represent the average value of the image to be matched I, (x′,y′) be the coordinates in the template image T, (x,y) be the currently calculated coordinate position in the image to be matched I, and R(x,y) be the cross-correlation value. Image matching is performed at the position with the highest cross-correlation value. First-order difference processing is then performed on the two matched images to obtain...

[0090] D(i,j)=T(i,j)-I(i,j)(17)

[0091] Where D(i,j) is the pixel value of the difference image at position (i,j), T(i,j) is the pixel value of the template image at position (i,j), and I(i,j) is the pixel value of the image to be matched at position (i,j). For a grayscale image R(x,y), setting a threshold A, the threshold segmentation result can be expressed as:

[0092]

[0093] Where G(x,y) is the segmented binary image, and D(x,y) is the pixel value of the difference image at position (x,y).

[0094] Random defects are obtained by threshold binarization of the difference between adjacent wafer images; system defects are obtained by threshold binarization of the difference between the image of the wafer under test and the standard wafer image.

[0095] Step 305: Match the test results of the image of the identified abnormal area with the actual wafer size, so that the processed image and the original wafer image are superimposed and mapped. The location of the abnormal area in the corresponding image is the precise defect location.

[0096] In this embodiment of the invention, the position information of the first image acquired each time within the actual wafer pattern is retained, and this position is used as the reference starting point. The image data in the test results is compared with the actual wafer size data starting from this position to determine the mapping relationship. The corresponding defect information in the test results is then located on the actual wafer size data, achieving relatively accurate defect area localization.

[0097] According to the patterned wafer defect detection method of this application, the light source can be a coherent light source or an incoherent broadband white light source. A coherent light source can reduce the halo effect of a white light source, while a white light source can reduce the speckle noise of a coherent light source. These can be selected and replaced according to different usage environments to improve the signal-to-noise ratio and detection sensitivity. Correspondingly, when using a grating for diffraction, a blazed grating can be used for a coherent light source, while a common rectangular grating can be used for a white light source, thereby ensuring intensity matching between the reference light and the object light, and facilitating spatial arrangement.

[0098] This invention is not limited to the above embodiments. Based on the technical solutions disclosed in this invention, those skilled in the art can make some substitutions and modifications to some of the technical features without creative effort, and all such substitutions and modifications are within the protection scope of this invention.

Claims

1. A method for detecting defects in wafer patterns using point diffraction co-path interference, characterized in that, The apparatus used in the method includes: The light source incident light path (110) is used to shape the light source (111) and incident it onto the wafer detection area; The signal collection optical path (120) is used to collect the reflected light signal from the surface of the wafer to be tested; The signal filtering interference optical path (130) is used to diffract and separate the reflected light signal from the wafer to be tested and filter it through a 4f system. A wafer displacement system (140) is used to scan the wafer to be inspected in both horizontal and vertical directions; A cooled CCD camera (151) is used to image the two beams of the light source incident light path (110) reference light and the object light reflected from the wafer onto the CCD camera target surface by beam splitting interference. The incident light path (110) includes a focusing lens (112), an intermediate lens (113) and a field lens (114). The light source (111) is a non-short wavelength laser light source or a broadband white light light source. It enters the signal collection light path (120) through the focusing lens (112), the intermediate lens (113) and the field lens (114). The point diffraction common-path interference wafer pattern defect detection method includes: Prepare a wafer to be inspected, which has a periodic repeating structure and anomaly defect structure pattern; obtain an interference image at each position by precisely moving the wafer to be inspected in the X direction. Phase reconstruction is performed on each interference image to obtain a phase image related to the surface morphology of the wafer to be inspected; Multiple consecutive phase images are processed by second-order difference and moving average, stitched together to form a large-scale image, and time-varying noise of the system is removed to obtain the processed image; The processed image is compared with the reference standard image to obtain the abnormal regions in the image; By mapping the image of the identified abnormal area to the actual wafer size, the precise defect area range of the wafer pattern can be obtained. Second-order difference processing and moving average processing are performed on multiple consecutive phase images, including: A second-order difference process is performed on every three adjacent phase images to obtain a second-order difference image with time-invariant noise eliminated; A moving average was applied to all second-order difference images to obtain a panoramic stitched image with time-varying noise eliminated.

2. The method for detecting wafer pattern defects using point diffraction common-path interference according to claim 1, characterized in that, The signal collection optical path (120) includes an objective lens (121), a beam splitter (122), and a tube mirror (123). The beam splitter (122) reflects the incident light from the light source into the objective lens (121). The objective lens (121) collects the light reflected from the wafer into the signal light transmission beam splitter (122), and images it onto the focal plane of the tube mirror (123) through the tube mirror (123).

3. The method for detecting wafer pattern defects using point diffraction common-path interference according to claim 1, characterized in that, The signal filtering interference optical path (130) includes a transmission diffraction grating (131), a first Fourier lens L1 (132), a spatial filter (133), and a second Fourier lens L2 (134). The object light transmission diffraction grating (131) located on the focal plane of the tube lens (123) performs diffraction. The first Fourier lens L1 (132) collects the diffracted light and focuses it on the first Fourier lens L1 (132). The spatial filter (133) filters the object light diffracted light. The reference light and object light passing through the spatial filter (133) are focused on the target surface of the cooled CCD camera (151) by the second Fourier lens L2 (134).

4. The method for detecting wafer pattern defects using point diffraction common-path interference according to claim 1, characterized in that, The wafer displacement system (140) includes an X-axis precision displacement stage (141), a Y-axis precision displacement stage (142) and a Z-axis precision displacement stage (143) connected sequentially from bottom to top. The wafer to be inspected (101) is located on the Z-axis precision displacement stage (143).

5. The method for detecting wafer pattern defects using point diffraction common-path interference according to claim 1, characterized in that, Phase reconstruction is performed on each interferometric image to obtain a phase image related to the surface morphology of the wafer under test, including: Two-dimensional Fourier transform was applied to the interferometric image to extract the reference light signal and the object light signal; A two-dimensional inverse Fourier transform is applied to the object optical signal to obtain amplitude and phase information; Two-dimensional discrete cosine transform and inverse transform are used on the phase information to obtain a phase image related to the surface morphology of the wafer to be inspected.

6. The method for detecting wafer pattern defects using point diffraction common-path interference according to claim 1, characterized in that, The processed image is then compared with a reference standard image to obtain the abnormal regions in the image, including: The position with the highest correlation value is obtained by applying a normalized cross-correlation method to the panoramic image and the reference standard image; The panoramic image and the reference standard image are compared at the position with the highest correlation value to obtain the abnormal area of ​​the image.

7. The method for detecting wafer pattern defects using point diffraction common-path interference according to claim 1, characterized in that, The images identifying the anomalous regions are mapped to the actual wafer dimensions, including: The position information of the first image acquired each time on the actual wafer is retained, and this position is used as the reference starting point; By placing the difference image between the panoramic image and the reference standard image at the reference starting point, the abnormal area in the image is mapped onto the actual wafer to obtain the precise range of the defect area.

Images