Design method of Fourier filter for semiconductor dark field defect detection

Abstract

The invention discloses a design method of a Fourier filter for semiconductor dark field defect detection, and belongs to the field of semiconductor defect detection.The method comprises the steps that the spatial frequency spectrum of a wafer surface pattern is solved based on the Fourier optics principle, a corresponding spatial frequency spectrum shielding needle is designed through spatial frequency spectrum distribution, and the wafer surface pattern is filtered; pattern noise is effectively suppressed, other types of defects on the surface of the wafer are highlighted, and a dark field detection system is assisted to complete defect detection of the wafer with patterns; the method is suitable for patterned wafers with various shapes and sizes, and because spatial spectrum analysis has uniqueness, the wafers with different patterns can be subjected to spectrum analysis and Fourier filter design; in addition, the design requirements of the required modulation wave surface can be met by changing the positions of the filtering needles for band-pass in the filter, the interval between the filtering needles and other parameters, and a certain design freedom degree is achieved.

Description

Technical Field

[0001] The present invention belongs to the field of semiconductor defect detection, and more specifically, relates to a design method of a Fourier filter for semiconductor dark field defect detection. Background Art

[0002] Defects are inevitable in the manufacture of integrated circuits, and different processes will produce different types of defects. With the increasing scale of integrated circuits, the increasing complexity of manufacturing processes, and the decreasing critical dimensions of integrated circuit design, it is a very challenging task to accurately detect defects in integrated circuits. A patterned wafer refers to a wafer with a specific pattern or circuit element structure drawn or produced through a specific process technology. Due to the pattern defects on the wafer and the complex morphology of various materials, defect detection is more complex and challenging than bare wafers. Dark field defect detection technology is an important nondestructive testing technology with the advantages of high detection sensitivity, good directionality, and non-destructiveness to materials. For patterned wafer detection, the main problem faced by the dark field detection method is pattern recognition: the detection system needs to be able to accurately distinguish between the light of the background pattern and the abnormal light caused by the defect. Summary of the invention

[0003] In view of the above defects or improvement needs of the prior art, the present invention provides a design method of a Fourier filter for semiconductor dark field defect detection, which can accurately identify defect signals of patterned wafers.

[0004] To achieve the above object, according to a first aspect of the present invention, a method for designing a Fourier filter for semiconductor dark field defect detection is provided, wherein the Fourier filter is placed between an objective lens and a collection lens group of a scattering dark field detection system, and the distance between the Fourier filter and the objective lens is the focal length f of the objective lens. The method comprises:

[0005] S1, obtain the two-dimensional spatial spectrum intensity F{U(x1,y1)} of the pattern signal on the surface of the wafer to be tested u,v , according to F{U(x1,y1)} u,v , the distance d0 between the objective lens and the surface of the wafer to be measured, the focal length f of the objective lens and the complex amplitude distribution U on the back focal plane of the objective lens f (x, y), converting the frequency point (u, v) of the wafer pattern signal to be measured into the imaging point (x, y) on the rear focal plane of the objective lens;

[0006] S2, removing the imaging points corresponding to the frequency points whose energy is lower than the threshold, and calculating the width and spacing of the main maximum peak and each secondary maximum peak in the horizontal and vertical directions of the frequency spectrum composed of the corresponding frequency points according to the coordinates of the retained imaging points;

[0007] S3, if the wafer pattern to be measured has rotational symmetry, the Fourier filter includes horizontal needles and vertical needles, otherwise the Fourier filter includes horizontal needles or vertical needles;

[0008] Among them, the number of horizontal needles or vertical needles is the sum of the number of main maximum peaks and secondary maximum peaks in the horizontal direction or vertical direction, the width is equal to the width of the main maximum peak and each secondary maximum peak in the horizontal direction or vertical direction, and the spacing between adjacent horizontal needles or vertical needles is equal to the spacing between adjacent peaks in the horizontal direction or vertical direction.

[0009] According to a second aspect of the present invention, a Fourier filter for semiconductor dark field defect detection is provided, wherein the Fourier filter is designed by the method described in the first aspect.

[0010] According to a third aspect of the present invention, there is provided a semiconductor dark field defect inspection system, comprising a scattering type dark field inspection system and a Fourier filter as in the second aspect.

[0011] According to a fourth aspect of the present invention, there is provided an electronic device, comprising: a computer-readable storage medium and a processor;

[0012] The computer-readable storage medium is used to store executable instructions;

[0013] The processor is used to read the executable instructions stored in the computer-readable storage medium and execute the method as described in the first aspect.

[0014] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute the method described in the first aspect.

[0015] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0016] The design method provided by the present invention solves the spatial spectrum of the wafer surface pattern by the Fourier optical principle, designs the Fourier filter by bandpass filtering, places the filter on the rear focal plane of the objective lens of the scattering type dark field detection system, and has a matching relationship with the objective lens, so as to filter out the influence of the wafer surface pattern in the detection imaging process; designs the corresponding spatial spectrum shielding structure by spatial spectrum distribution, filters the wafer surface pattern, effectively suppresses the pattern noise, highlights other types of defects on the wafer surface, and can assist the dark field detection system to complete the defect detection of the wafer with patterns; the method is applicable to patterned wafers of various shapes and sizes, and because the spatial spectrum analysis is unique, the spectrum analysis and Fourier filter design need to be performed separately for wafers with different patterns, and the Fourier filter is selected according to the wafer pattern type during use; in addition, the design requirements of the required modulation wave surface can be achieved by changing the position of the filter needle used for bandpass in the filter, the interval between the filter needles and other parameters, and has a certain degree of design freedom.

[0017] The Fourier filter designed by the design method provided by the present invention can improve the signal-to-noise ratio of the output wafer surface image, effectively filter noise signals, and improve the speed and accuracy of optical detection. The filter designed by the method provided by the present invention can suppress the wafer surface graphic information in the output image and highlight the defect information, solve the impact of the surface pattern on defect detection during the patterned wafer detection process, effectively suppress detection noise, and improve the output signal-to-noise ratio; and the Fourier filter can be designed according to different illumination bands, and a wide range of available materials can be selected; it can realize the control of the wavefront, and is used as a phase filter in optical information processing to filter out the noise signal of the dark field defect detection wafer surface pattern, and accurately identify the defect signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of the placement of the Fourier filter provided in an embodiment of the present invention in a scattering dark field detection system;

[0019] Figure 2 (a) and (b) are schematic diagrams of wafer surface patterns with rotational symmetry and without rotational symmetry, respectively;

[0020] Figure 3 (a) and (b) are schematic diagrams of spatial spectrum intensity on the back focal plane of wafer surface patterns with rotational symmetry and patterns without rotational symmetry, respectively;

[0021] Figure 4 (a) and (b) are schematic diagrams of the binarization of the spatial spectrum intensity of wafer surface patterns with rotational symmetry and without rotational symmetry, respectively.

[0022] Figure 5(a) and (b) are Fourier filter design diagrams for wafer surface patterns with rotational symmetry and without rotational symmetry, respectively;

[0023] Figure 6 (a) and (b) are schematic diagrams of the filter size limitation of the wafer surface pattern with rotational symmetry and the one without rotational symmetry and NA matching, respectively;

[0024] Figure 7 (a) and (b) are schematic diagrams of defect locations of wafer surface patterns with rotational symmetry and without rotational symmetry, respectively;

[0025] Figure 8 (a) and (b) are defect detection effect diagrams of wafer surface patterns with rotational symmetry and without rotational symmetry after being filtered by their respective Fourier filters. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0027] An embodiment of the present invention provides a method for designing a Fourier filter for semiconductor dark field defect detection, wherein the Fourier filter is placed between an objective lens and a collection lens group of a scattering dark field detection system, and the distance between the Fourier filter and the objective lens is the focal length f of the objective lens. The method comprises:

[0028] S1, obtain the two-dimensional spatial spectrum intensity F{U(x1,y1)} of the pattern signal on the surface of the wafer to be tested u,v , according to F{U(x1,y1)} u,v , the distance d0 between the objective lens and the surface of the wafer to be measured, the focal length f of the objective lens and the complex amplitude distribution U on the back focal plane of the objective lens f (x, y), converting the frequency point (u, v) of the wafer pattern signal to be measured into the imaging point (x, y) on the rear focal plane of the objective lens;

[0029] S2, removing the imaging points corresponding to the frequency points whose energy is lower than the threshold, and calculating the width and spacing of the main maximum peak and each secondary maximum peak of the frequency spectrum composed of the corresponding frequency points in the horizontal direction and the vertical direction respectively according to the coordinates of the retained imaging points; it can be understood that, according to the coordinates of the retained imaging points, the width and position of the main maximum peak and each secondary maximum peak of the frequency spectrum composed of the corresponding frequency points in the horizontal direction and the vertical direction respectively can be calculated, and the spacing between adjacent peaks can be calculated according to the position information;

[0030] S3, if the wafer pattern to be measured has rotational symmetry, the Fourier filter includes horizontal needles and vertical needles, otherwise the Fourier filter includes horizontal needles or vertical needles; the horizontal needles and vertical needles are used to filter the wafer surface pattern;

[0031] Among them, the number of horizontal needles or vertical needles is the sum of the number of main maximum peaks and secondary maximum peaks in the horizontal direction or vertical direction, the width is equal to the width of the main maximum peak and each secondary maximum peak in the horizontal direction or vertical direction, and the spacing between adjacent horizontal needles or vertical needles is equal to the spacing between adjacent peaks in the horizontal direction or vertical direction.

[0032] like Figure 1 As shown, the method provided by the present invention places the Fourier filter between the objective lens and the collecting lens group of the scattering dark field detection system, and uses the Fourier transform property of the objective lens to generate the spatial spectrum of the object in the dark field detection system, and filters the information carried by the spatial structure of the wafer surface by measuring and analyzing the spatial spectrum. Therefore, the designed filter position is placed on the Fourier plane (i.e., the back focal plane) of the objective lens in the collecting light path of the dark field detection system.

[0033] The wafer surface pattern is periodic and its spectrum distribution has a certain regularity, while the spectrum of flaws and defects is random. The designed Fourier filter can filter the light signal generated by the wafer surface pattern in the output image (i.e., the image detected by the detector) and only retain the light signal generated by the defect, thereby improving the signal-to-noise ratio of defect signal detection.

[0034] During the detection process, when there are no abnormal defects on the wafer surface, the Fourier filter will filter out the light signal generated by the surface pattern, and the entire image plane will present a dark field; when there are abnormal defects on the wafer surface, the Fourier filter will only filter out the light signal generated by the surface pattern, and the abnormal defects will appear higher in brightness in the dark field, thereby achieving defect detection of patterned wafers.

[0035] The image information on the wafer surface is regarded as the superposition noise of the input. This noise is additive, and the input field distribution can be expressed as

[0036] f(x1,y1)=s(x1,y1)+n(x1,y1) (1)

[0037] Where s(x1, y1) and n(x1, y1) represent the defect signal and noise signal of the point (x1, y1) on the surface of the wafer under test, respectively.

[0038] Performing a two-dimensional Fourier transform on the wafer surface image signal can decompose it into a superposition of several complex plane waves, and then use the Fourier analysis (spectral analysis) method to process it. Performing a Fourier transform on the input field f(x1, y1) can obtain its spectrum F(u, v), then S(u, v) and N(u, v) represent the spectrum of the wafer surface defect signal (the light signal generated by the defective surface) and the wafer surface pattern signal (the light signal generated by the wafer surface pattern when the detection light is incident on the wafer surface):

[0039] F(u,v)=S(u,v)+N(u,v) (2)

[0040] As long as the distribution characteristics of the signal and noise in the frequency domain are different, a filter can be used to effectively suppress noise while transmitting the signal, improve the output signal-to-noise ratio, and filter out the influence of the wafer surface pattern.

[0041] As an example, Figure 2 (a) and (b) are schematic diagrams of wafer surface patterns with rotational symmetry and without rotational symmetry, respectively. The wafer surface patterns are both composed of regular rectangular line segments. Figure 2 The structural size of (a) is: 90nm*90nm. Figure 2 The structural dimensions of (b) are: line width CD=46nm, line length 10*CD, lateral spacing 2*CD, and longitudinal spacing 15*CD.

[0042] Calculate the light field distribution on specific planes one by one along the propagation direction of the light wave: the complex amplitude distribution U(x1,y1) on the plane just behind the wafer under test, the complex amplitude distribution U(x2,y2) on the plane just before the objective lens, the complex amplitude distribution U′(x2,y2) on the plane just behind the objective lens, and the complex amplitude distribution U on the back focal plane of the objective lens. f (x,y).

[0043] The changes in the frequency spectrum of the field distribution when the light wave propagates to the plane just before the objective lens can be calculated based on the angular spectrum theory.

[0044]

[0045] Where H(u,v) is the transfer function describing the Fresnel diffraction effect in the frequency domain

[0046] H(u,v)=exp[-jπλd0(u 2 +v 2 )] (4)

[0047] λ is the wavelength of the detection light, and d0 is the distance between the wafer surface and the objective lens.

[0048]

[0049] Without considering the finite aperture of the objective lens,

[0050]

[0051] According to the Fresnel diffraction formula, it can be calculated that

[0052]

[0053] Substituting into equations (5) and (6), we can obtain

[0054]

[0055] According to the above formula, when the input wafer image is placed on the front focal plane of the objective lens, the complex amplitude distribution on the rear focal plane of the objective lens is proportional to the Fourier transform of the complex amplitude distribution on the wafer surface. The coordinate relationship between the frequency value and the rear focal plane position is:

[0056]

[0057] That is, the point (x f ,y f ) is determined by the spatial frequency of the light wave passing through the wafer surface (f x ,f y ) is determined by the amplitude and phase of the Fourier component. Figure 1 The optical path shown can accurately reflect the Fourier transform of the complex amplitude distribution on the wafer surface on the rear focal plane of the objective lens, and perform spatial filtering operations on the information it carries on this plane.

[0058] In this example, the distance between the objective lens and the wafer surface is d0 = f, and the focal length of the objective lens is f = 20 mm. The frequency domain coordinates (u, v) can be converted to the back focal plane imaging coordinates (x, y) through the conversion relationship of formula (9). The back focal plane imaging result is as follows: Figure 3 As shown in (a) and (b), the strongest energy appears in the frequency domain (u,v) = (0,0), that is, at the image plane (x,y) = (0,0), and the spectrum energy distribution has a periodic repetition pattern.

[0059] There are many design methods for using spatial filters to change the spectrum components on the spectrum plane. Horizontal slits can filter the vertical structure of the image, and vertical slits can filter the horizontal structure of the image. Placing horizontal and vertical slits with a certain spacing on the spectrum plane can filter the spectrum information. The choice of slit direction can be selected according to the energy distribution to select horizontal needle or vertical needle structure, or both directions can be used for filtering.

[0060] Based on the principle of Fourier optics, when there are point defects on the surface of the wafer, the intensity of each point in its spatial spectrum will change. However, the image detail information carried by the high-frequency component can be ignored when detecting the point defect, and the purpose of defect detection can be achieved without all frequency components. Therefore, removing some points that do not appear according to the periodic law and not processing them can not only reduce the design difficulty but also do not affect the detection effect of defects. Based on this, the spectrum energy distribution is analyzed and the appropriate filtering energy threshold is selected. The spatial frequency intensity is binarized and the spatial frequency points above the energy threshold are removed to facilitate the observation of the spectrum distribution. Figure 2 The wafer surface pattern in (a) has rotational symmetry. The spatial frequency energy distribution to be filtered has the same distribution in the horizontal and vertical directions, and both are periodic. When filtering, it is necessary to set up both horizontal and vertical Fourier filters. Figure 2 The spectrum of the wafer pattern in (b) will show different distribution patterns in the horizontal and vertical directions. You can choose to filter the horizontal or vertical structure of the image. In this example, the vertical component shows a regular periodic distribution, so you only need to design vertical filtering (i.e. filtering the horizontal image structure).

[0061] According to different spatial frequency distributions, the filter shapes can be designed separately. Figure 5 As shown in (a) and (b) in Figure 4 The wafer structure (a) is binarized and the Fourier filter structure is designed as follows: Figure 5 As shown in (a), the main maximum and secondary maximum are filtered using a horizontal and vertical multi-needle structure, and the structural dimensions are: the width of the middle needle is 1.4 mm, which is the width of the main maximum peak; the interval between the middle needle and the left and right adjacent needles is 28.9 mm, which is the interval between the main maximum peak and the left and right adjacent secondary maximum peaks. Figure 4 As shown in (a), on the left side of the secondary maximum peak on the left side of the main maximum peak, there are three secondary maximum peaks distributed in sequence at a spacing of 57.8 mm. Therefore, on the left side of the left needle adjacent to the middle needle, three needle structures are distributed at a spacing of 57.8 mm (not shown in the figure), and the widths of these three needles are equal to the widths of the above three secondary maximum peaks in a one-to-one correspondence; similarly, since the right side of the secondary maximum peak on the right side of the main maximum peak also has three secondary maximum peaks distributed in sequence at a spacing of 57.8 mm, therefore, on the right side of the right needle adjacent to the middle needle, three needle structures are distributed at a spacing of 57.8 mm (not shown in the figure).

[0062] because Figure 2 The vertical component of the spectrum of the wafer pattern in (b) presents a regular periodic distribution, so the Fourier filter is only designed to filter in the vertical direction (i.e., filter the horizontal image structure). Accordingly, according to Figure 4The occlusion structure (i.e., needle structure) of the filter is designed based on the binarization result of the wafer structure spectrum in (b). Since the width of the main maximum peak is equal to the width of the secondary maximum peaks adjacent to it on the left and right, and they are distributed periodically, the structure of the Fourier filter is designed to use three needles as a basic unit and repeat periodically. In each group of basic units, the width of each needle is 2.8 mm, and the spacing between adjacent needles is 14.6 mm; the spacing between each group of basic units is 28.9 mm.

[0063] In actual use, it is necessary to repeat periodically according to the numerical aperture NA of the objective lens and the image plane size. When the image plane size is limited, in order to match the filtering effect of the objective lens NA on the spatial frequency, only the spatial frequency components within the limited image plane can be used to set the filter, such as Figure 6 The red areas in (a) and (b) are shown in the figure. The effect of generating random defects and filtering them is shown in the figure. Figure 7 (a), (b), Figure 8 As shown in (a) and (b), it can be seen that the wafer surface pattern information is filtered out in the imaging surface, and only randomly generated defect information exists with strong contrast, which can meet the detection requirements.

[0064] That is, when the image plane size of the dark field detection system is limited, the filter only filters the wafer surface pattern signal within the image plane range of the dark field detection system, the number of the horizontal needles or the vertical needles is the sum of the number of the main maximum peaks and the secondary maximum peaks of the target spectrum plane in the horizontal direction or the vertical direction, the width is equal to the width of the main maximum peaks and the secondary maximum peaks of the target spectrum plane in the horizontal direction or the vertical direction, and the spacing between adjacent horizontal needles or vertical needles is equal to the spacing between adjacent peaks of the target spectrum plane in the horizontal direction or the vertical direction; wherein, the target spectrum plane is a spectrum plane composed of frequency points corresponding to imaging points (imaging points corresponding to frequency points whose energy is lower than the threshold have been removed) within the image plane range of the dark field detection system.

[0065] In summary, the design method provided by the present invention uses the Fourier transform method to perform spatial spectrum analysis on the wafer surface pattern, designs the filter needle position and interval through the spatial spectrum distribution, that is, designs the corresponding spatial spectrum shielding structure, and filters the wafer surface pattern, so that the dark field background is not affected by the noise generated by the surface pattern during defect detection, so as to improve the signal-to-noise ratio and accuracy of the subsequent defect detection process. It has been verified that the Fourier filter can effectively filter the background pattern noise, and can ensure that it eliminates the wafer background pattern without affecting the display of defect information.

[0066] An embodiment of the present invention provides a Fourier filter for semiconductor dark field defect detection. The Fourier filter is designed by using the method described in any of the above embodiments.

[0067] An embodiment of the present invention provides a semiconductor dark field defect detection system, including a scattering dark field detection system and a Fourier filter as described in any of the above embodiments.

[0068] An embodiment of the present invention provides an electronic device, including: a computer-readable storage medium and a processor;

[0069] The computer-readable storage medium is used to store executable instructions;

[0070] The processor is used to read the executable instructions stored in the computer-readable storage medium and execute the method described in any of the above embodiments.

[0071] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute the method described in any of the above embodiments.

[0072] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A design method for a Fourier filter for semiconductor dark field defect detection, characterized in that: The Fourier filter is placed between the objective lens and the collection lens group of the scattering dark field detection system, and the distance between the Fourier filter and the objective lens is the focal length f of the objective lens. The method includes: S1, obtain the two-dimensional spatial spectrum intensity of the pattern signal on the surface of the wafer to be tested according to The distance d0 between the objective lens and the surface of the wafer to be measured, the focal length f of the objective lens, and the complex amplitude distribution U on the back focal plane of the objective lens f (x, y), converting the frequency point (u, v) of the wafer pattern signal to be measured into the imaging point (x, y) on the back focal plane of the objective lens; S2, removing the imaging points corresponding to the frequency points whose energy is lower than the threshold, and calculating the width and spacing of the main maximum peak and each secondary maximum peak in the horizontal and vertical directions of the frequency spectrum composed of the corresponding frequency points according to the coordinates of the retained imaging points; S3, if the wafer pattern to be measured has rotational symmetry, the Fourier filter includes horizontal needles and vertical needles, otherwise the Fourier filter includes horizontal needles or vertical needles; Among them, the number of horizontal needles or vertical needles is the sum of the number of main maximum peaks and secondary maximum peaks in the horizontal direction or vertical direction, the width is equal to the width of the main maximum peak and each secondary maximum peak in the horizontal direction or vertical direction, and the spacing between adjacent horizontal needles or vertical needles is equal to the spacing between adjacent peaks in the horizontal direction or vertical direction.

2. The method according to claim 1, characterized in that Said The distance d0 between the objective lens and the surface of the wafer to be measured, the focal length f of the objective lens, and the complex amplitude distribution U on the back focal plane of the objective lens f The relationship between (x, y) is: Where λ is the wavelength of the detection light.

3. The method according to claim 1 or 2, characterized in that When the image plane size of the dark field detection system is limited, the number of the horizontal needles or vertical needles is the sum of the number of the main maximum peaks and the secondary maximum peaks of the target spectrum surface in the horizontal direction or the vertical direction, the width is equal to the width of the main maximum peak and each secondary maximum peak of the target spectrum surface in the horizontal direction or the vertical direction, and the spacing between adjacent horizontal needles or vertical needles is equal to the spacing between adjacent peaks of the target spectrum surface in the horizontal direction or the vertical direction. The target frequency spectrum surface is a frequency spectrum surface composed of frequency points corresponding to imaging points within the image plane range of the dark field detection system.

4. A Fourier filter for semiconductor dark field defect detection, characterized in that: The Fourier filter is designed by the method according to any one of claims 1 to 3.

5. A semiconductor dark field defect detection system, characterized in that: It comprises a scattering dark field detection system and the Fourier filter as claimed in claim 4.

6. An electronic device, characterized in that: include: A computer readable storage medium and a processor; The computer-readable storage medium is used to store executable instructions; The processor is used to read the executable instructions stored in the computer-readable storage medium and execute the method according to any one of claims 1 to 3.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to execute the method according to any one of claims 1 to 3.

Images