Interference observation device and interference observation method

Abstract

An interference observation device comprising: a light source that outputs light; an interference optical system that has a reference mirror, that splits the light output from the light source into first light and second light, and that outputs interference light of the first light reflected by an observation target and the second light reflected by the reference mirror; an imaging element that detects the interference light; a processing unit that acquires an interference image on the basis of the detection result by the imaging element and that acquires the degree of disorder of an image of a region of a foreign object adhered to the observation target in the interference image; and an output unit that outputs information relating to the degree of disorder acquired by the processing unit.

Description

Interference observation device and interference observation method

[0001] One aspect of the present invention relates to an interference observation apparatus and an interference observation method.

[0002] An interference observation device is known that includes a light source that outputs light, an interference optical system that splits the light output from the light source into first light and second light and outputs interference light between the first light reflected by an observation object and the second light reflected by a reference mirror, an image sensor that detects the interference light, and a processing unit that acquires an interference image based on the detection result of the image sensor.As an example of this type of technology, Patent Document 1 discloses a cell observation device that is intended to distinguish cell types by acquiring information such as the optical thickness of the cells.

[0003] WO2011 / 132587 JP 08-124982 JP 11-176898

[0004] In the above-described interference observation apparatus, foreign matter may adhere to the object being observed. In such cases, it is desirable to distinguish the foreign matter (i.e., to identify the type of foreign matter). Regarding this issue, for example, Patent Document 2 describes that minute foreign matter adhering to the pattern surface of a semiconductor device is a significant factor in reducing the yield of the semiconductor device. Furthermore, for example, Patent Document 2 describes that in order to prevent the adhesion of such foreign matter, it is important to quickly identify the source of the foreign matter and take measures to prevent contamination, and that for this purpose, analysis of the composition and state of the foreign matter is essential.

[0005] Regarding the composition of foreign matter, for example, Patent Document 3 describes that when a gate oxide film is formed by thermal oxidation in a semiconductor manufacturing process, the presence of metal contamination and organic contamination can cause dielectric breakdown. Patent Document 3 also describes that organic contamination can also cause insulation degradation when attached after the gate oxide film is formed, and also describes a method for detecting and removing organic contamination.

[0006] Therefore, an object of one aspect of the present invention is to provide an interference observation apparatus and an interference observation method that are capable of discriminating foreign matter adhering to an observation object.

[0007] As a result of extensive research, the inventors have found that in an interference image of an object to be observed having foreign matter attached thereto, the degree of image clutter in a foreign matter region varies depending on the type of foreign matter. Therefore, the inventors have found that if the degree of image clutter in a foreign matter region in an interference image can be obtained, it will be possible to discriminate the foreign matter, and have completed one aspect of the present invention.

[0008] That is, an interference observation device according to one aspect of the present invention is [1] "an interference observation device comprising: a light source that outputs light; an interference optical system that has a reference mirror and splits the light output from the light source into first light and second light, and outputs interference light between the first light reflected by an observation object and the second light reflected by the reference mirror; an image pickup element that detects the interference light; a processing unit that acquires an interference image based on a detection result of the image pickup element and acquires a degree of disorder of an image of a foreign matter region adhering to the observation object in the interference image; and an output unit that outputs information about the degree of disorder acquired by the processing unit."

[0009] In this interference observation device, when a foreign substance is attached to the observation object, information about the degree of disorder of the image of the foreign substance region in the interference image can be output. Therefore, based on the output information about the degree of disorder and the above-mentioned knowledge, it becomes possible to discriminate the foreign substance attached to the observation object.

[0010] An interference observation apparatus according to one aspect of the present invention may be [2] "the interference observation apparatus according to [1], which includes a storage unit that stores a foreign matter classification model relating to the relationship between the degree of disorder of foreign matters and the type of the foreign matter, and the processing unit refers to the foreign matter classification model stored in the storage unit to determine the type of the foreign matter from the acquired degree of disorder." In this case, it is possible to discriminate foreign matters adhering to an observation object by referring to the foreign matter classification model.

[0011] The interference observation apparatus according to one aspect of the present invention may be [3] "the interference observation apparatus according to [2], which is equipped with a learning unit that generates a foreign matter classification model relating to the relationship between the degree of disorder of the foreign matter and the type of the foreign matter by machine learning using a plurality of interference images in a case where the foreign matter is attached to the observation object as training data." In this case, it is possible to generate the foreign matter classification model by machine learning.

[0012] An interference observation apparatus according to one aspect of the present invention may be [4] "the interference observation apparatus according to [2] or [3], wherein the foreign matter includes a reflective foreign matter and a transmissive foreign matter." In this case, it becomes possible to distinguish whether a foreign matter attached to an observation object is a reflective foreign matter or a transmissive foreign matter.

[0013] An interference observation apparatus according to one aspect of the present invention may be [5] "the interference observation apparatus according to [4], wherein the reflective foreign matter includes a semiconductor, and the transmissive foreign matter includes an organic matter." In this case, it becomes possible to distinguish whether the foreign matter attached to the observation object is a reflective semiconductor or a transmissive organic matter.

[0014] An interference observation apparatus according to one aspect of the present invention may be [6] "the interference observation apparatus according to any one of [1] to [5], wherein the output unit includes a display unit that displays information about the degree of disorder." In this case, information about the degree of disorder of the image of the foreign substance region in the interference image can be easily grasped.

[0015] An interference observation apparatus according to one aspect of the present invention may be [7] "the interference observation apparatus according to any one of [1] to [6], wherein the interference optical system is configured so that the optical path length of the optical path of the second light is changeable." In this case, by changing the optical path length of the optical path of the second light, it is possible to make foreign matter stand out in the acquired interference image. This makes it possible to easily distinguish foreign matter.

[0016] An interference observation method according to one aspect of the present invention is [8] "an interference observation method comprising: a light output step of outputting light from a light source; an interference light output step of splitting the light output from the light source into first light and second light, and outputting interference light between the first light reflected by an observation object and the second light reflected by a reference mirror; an interference light detection step of detecting the interference light with an image sensor; a processing step of acquiring an interference image based on a detection result of the image sensor, and acquiring a degree of disorder of an image of the foreign matter region adhering to the observation object in the interference image; and an output step of outputting information related to the degree of disorder acquired in the processing step."

[0017] In this interference observation method, when a foreign substance is attached to the object being observed, information regarding the degree of disorder of the image of the foreign substance region in the interference image can be output. Therefore, based on the output information regarding the degree of disorder and the above-mentioned knowledge, it becomes possible to discriminate the foreign substance attached to the object being observed.

[0018] According to one aspect of the present invention, it is possible to provide an interference observation apparatus and an interference observation method that are capable of discriminating between foreign matter adhering to an observation object.

[0019] FIG. 1 is a configuration diagram showing an interference observation apparatus according to an embodiment. FIG. 2 is an enlarged view of an observation object shown in FIG. 1. FIG. 3 is a flowchart showing an example of a method for calculating a disorder index. FIG. 4 is a diagram explaining the relationship between a phase image and a disorder index. FIG. 5(a) is a diagram showing a phase image. FIG. 5(b) is a diagram showing the real part of a complex image corresponding to the phase image of FIG. 5(a). FIG. 5(c) is a diagram showing the imaginary part of a complex image corresponding to the phase image of FIG. 5(a). FIG. 6(a) is a diagram showing the real part of a first smoothed complex image. FIG. 6(b) is a diagram showing the imaginary part of the first smoothed complex image. FIG. 7(a) is a diagram showing the real part of a second smoothed complex image. FIG. 7(b) is a diagram showing the imaginary part of the second smoothed complex image. FIG. 8(a) is a diagram showing the real part of a third smoothed complex image. FIG. 8(b) is a diagram showing the imaginary part of the third smoothed complex image. FIG. 9(a) is a diagram showing a first phase difference image. FIG. 9( b) is a diagram showing a second phase difference image. FIG. 9( c) is a diagram showing a third phase difference image. FIG. 10( a) is a diagram showing an enlarged view of a portion of a phase image including a reflective foreign object. FIG. 10( b) is a diagram showing an enlarged view of a portion of a phase difference image corresponding to the phase image of FIG. 10( a). FIG. 11( a) is a diagram showing an enlarged view of a portion of a phase image including a transparent foreign object. FIG. 11( b) is a diagram showing an enlarged view of a portion of a phase difference image corresponding to the phase image of FIG. 11( a). FIG. 12 is a diagram showing an image of local phase fluctuations in the second phase difference image. FIG. 13 is a diagram explaining a foreign object classification model. FIG. 14 is a diagram showing a reflective foreign object degree table. FIG. 15 is a flowchart showing an example of a method for learning a foreign object classification model. FIG. 16 is a flowchart showing an example of an interference observation method using an interference observation device. FIG. 17 is a diagram explaining a foreign object classification model according to a modified example.

[0020] Hereinafter, embodiments will be described in detail with reference to the drawings. In the following description, the same or equivalent elements will be designated by the same reference numerals, and redundant description will be omitted.

[0021] [Interference Observation Apparatus] As shown in FIG. 1 , the interference observation apparatus 1 includes a light source 2, an interference optical system 3, an image sensor 4, a processing unit 5, and a stage S. The interference observation apparatus 1 is an interference microscope for observing an observation object 8 placed on the stage S using the interference of light. The observation object 8 is, for example, a semiconductor device, but may also be other industrial samples formed from metal, glass, resin, liquid crystal, polymer compounds, etc. The observation object 8 may also be a biological sample such as a cell. The following description will be given by defining an X direction, a Y direction perpendicular to the X direction, and a Z direction perpendicular to both the X and Y directions, as shown in FIG. 1 .

[0022] The interference observation apparatus 1 is configured to be able to observe the inside of an observation object 8. In internal observation, an observation surface R located inside the observation object 8 is observed. In this example, the observation object 8 is a so-called film-coated wafer. As shown in FIGS. 1 and 2 , the observation object 8 includes a device portion 81 constituting a patterned chip, and a film portion 82 which is a passivation film provided on the surface of the device portion 81. The device portion 81 is formed by forming a pattern of functional elements 81b on a wafer 81a. The observation surface R is the surface of the device portion 81.

[0023] The light source 2 outputs incoherent light. The light source 2 may be, for example, a lamp-based light source such as a halogen lamp, a light-emitting diode (LED) light source, a super luminescent diode (SLD) light source, or an amplified spontaneous emission (ASE) light source. In this example, the interference optical system 3 is configured as a Linnik interferometer. The interference optical system 3 includes a lens 11, a beam splitter 12, an objective lens 13, a reference objective lens 14, and a reference mirror 15. The interference optical system 3 is disposed in a housing H together with the light source 2 and constitutes an optical module M. The optical module M is movable along the Z direction by a predetermined actuator 16. The Z direction is parallel to the optical axis of the objective lens 13 and parallel to the direction in which a first light L1 (described later) is incident on the observation object 8.

[0024] The lens 11 collimates the light output from the light source 2. The beam splitter 12 is, for example, a prism having an optical surface 12a, and splits the light collimated by the lens 11 into a first light L1 and a second light L2 at the optical surface 12a. The beam splitter 12 outputs the first light L1 to the objective lens 13 and outputs the second light L2 to the reference objective lens 14. Furthermore, the first light L1 reflected by the observation surface R of the observation object 8 enters the optical surface 12a via the objective lens 13, and the second light L2 reflected by the reference mirror 15 enters the optical surface 12a via the reference objective lens 14. The first light L1 and the second light L2 are combined at the optical surface 12a to become an interference light L3. The interference optical system 3 outputs the interference light L3 to the image sensor 4.

[0025] The objective lens 13 focuses the first light L1 output from the beam splitter 12 onto the observation object 8 placed on the stage S. In addition, the first light L1 reflected by the observation surface R of the observation object 8 is incident on the objective lens 13. The objective lens 13 outputs the incident first light L1 to the beam splitter 12. The reference objective lens 14 guides the second light L2 output from the beam splitter 12 to the reference mirror 15 and focuses the light on the reference mirror 15. In addition, the reference objective lens 14 outputs the second light L2 reflected by the reference mirror 15 to the beam splitter 12. The reference mirror 15 reflects the second light L2 output from the reference objective lens 14 so that it returns to the reference objective lens 14.

[0026] Further disposed within the housing H of the optical module M are a stepping motor 17 for moving the reference objective lens 14, and a stepping motor 18 and a piezoelectric element 19 for moving the reference mirror 15. The stepping motor 17 moves the reference objective lens 14 along the optical axis direction of the second light L2 (e.g., the X direction), which is perpendicular to the Z direction. The stepping motor 18 and the piezoelectric element 19 move the reference mirror 15 along the optical axis direction of the second light L2. In order to obtain interference fringes with a good S / N ratio, it is preferable that the optical path length from the beam splitter 12 to the reference mirror 15 be adjusted to approximately match the optical path length from the beam splitter 12 to the observation surface (reflecting surface) R.

[0027] The response time of the stepping motors 17, 18 is greater than 10 msec, while the response time of the piezoelectric element 19 is less than 1 msec. That is, the response time of the piezoelectric element 19 is shorter than that of the stepping motors 17, 18. The stroke (minimum movement distance) of the stepping motors 17, 18 is several mm, while the stroke of the piezoelectric element 19 is approximately 10 μm. That is, the stroke of the piezoelectric element 19 is shorter than that of the stepping motors 17, 18. The number of times that the stepping motors 17, 18 can be driven over their lifetime is less than 1 million times, while the number of times that the piezoelectric element 19 can be driven over their lifetime is more than 10 billion times. That is, the number of times that the piezoelectric element 19 can be driven over their lifetime is greater than that of the stepping motors 17, 18.

[0028] The imaging element 4 is an image sensor (camera) such as a CCD area image sensor or a CMOS area image sensor. The imaging element 4 detects (images) the interference light L3 output from the interference optical system 3 (beam splitter 12). A lens 41 and a lens barrel 42 are arranged between the imaging element 4 and the interference optical system 3. The lens 41 forms an image of the interference light L3 output from the interference optical system 3 on the imaging surface of the imaging element 4. The lens 41 is housed in the lens barrel 42. The lens barrel 42 is formed, for example, in a cylindrical shape and is fixed to the imaging element 4 so as to surround the imaging surface.

[0029] The processing unit 5 is connected to be able to communicate with each unit of the interference observation device 1, including the light source 2, the interference optical system 3, the image sensor 4, and the stage S, and acquires an interference image based on the detection result of the interference light L3 at the image sensor 4. The interference image includes, for example, a phase image, an amplitude image, a complex image, various images described below, and images equivalent to any of these.

[0030] The processing unit 5 is configured, for example, by a computer C including a processor (CPU) and storage media such as RAM and ROM. The computer C has a memory area (memory unit) 51 that stores various information, an input unit 52 that accepts input of various information, a display unit 53 that displays various information, and a learning unit 54 that generates a foreign object classification model (described later) through machine learning. The input unit 52 is, for example, a device that accepts user input, such as a mouse or keyboard. The display unit 53 is, for example, a display that displays images. The input unit 52 and the display unit 53 may be commonly configured, for example, by a touch panel, in which case they may be configured by a GUI (Graphical User Interface). In this example, the processing unit 5, the memory area 51, the input unit 52, the display unit 53, and the learning unit 54 are configured by a single device, but at least one of them may be configured by a separate device, such as a mobile terminal.

[0031] The stage S is a stage for placing the observation object 8, and is movable along an XY plane perpendicular to the Z direction in which the first light L1 is incident on the observation object 8. This makes it possible in the interference observation apparatus 1 to observe the observation surface R of the observation object 8 (to acquire an interference image) while moving the stage S along the XY plane, that is, while changing the observation position of the observation object 8.

[0032] In the interference observation apparatus 1, first, incoherent light is output from the light source 2 (light output step). The light is collimated by the lens 11 and split into first light L1 and second light L2 by the beam splitter 12. The first light L1 is focused on the observation object 8 by the objective lens 13, reflected at the interface of the observation object 8, and input to the beam splitter 12 via the objective lens 13. The second light L2 is focused on the reference mirror 15 by the reference objective lens 14, reflected by the reference mirror 15, and input to the beam splitter 12 via the reference objective lens 14. The beam splitter 12 combines the input first light L1 and second light L2 to output interference light L3 (interference light output step). The output interference light L3 is detected by the image sensor 4 (interference light detection step). Based on the detection result from the image sensor 4, the processing unit 5 acquires an interference image (processing step).

[0033] In the interference observation apparatus 1, the imaging element 4 captures four interference images, and the processing unit 5 constructs (acquires) one phase image based on the four captured interference images. The piezoelectric element 19 finely moves the reference mirror 15 in synchronization with the timing of capturing the four interference images. This causes the optical path length difference between the first light L1 and the second light L2 to vary between the four interference images. The amount of movement of the reference mirror 15 to capture the four interference images is smaller than the wavelength λ of the light output from the light source 2. In this example, the optical path length difference (phase shift interval) of the second light L2 between the four interference images is λ / 4. When the reference mirror 15 is finely moved so that the optical path length change width is λ / 4, keeping in mind that the reference light (second light L2) travels back and forth through the reference optical path (the optical path length of the optical path of the second light L2), the actual movement step of the reference mirror 15 is λ / 8, which is 66.5 nm when λ = 532 nm. The method for acquiring the interference image is not limited to the above-described example, and various known methods may be used.

[0034] In this embodiment, the processing unit 5 calculates and acquires the degree of disorder in the region of the foreign matter 9 adhering to the observation object 8 in the interference image, and discriminates the foreign matter 9 from the acquired degree of disorder. The foreign matter 9 to be discriminated includes at least a reflective foreign matter 91 that is reflective to the first light L1 and a transparent foreign matter 92 that is transparent to the first light L1. The reflective foreign matter 91 includes, for example, silicon scraps (silicon chips), which are semiconductors. The transparent foreign matter 92 is particularly organic matter, and includes, for example, ethanol stains. The type of foreign matter 9 to be discriminated is not particularly limited, and various types of elements may be used.

[0035] Non-patent document 1: P. Burt and E. Adelson, “The Laplacian Pyramid as a Compact Image Code”, in IEEE Transactions on Communications, vol. 31, no. 4, pp. 532-540, April 1983

[0036] The above-mentioned Non-Patent Document 1 discloses a method for extracting image features by creating multiple smoothed images using different smoothing filter sizes for the original image and then using a difference image between the images with different smoothing filter sizes. However, since the phase in a phase image is folded over between -π and π, there are problems with directly applying a smoothing filter to the phase. An example of a method for smoothing filtering and generating a difference image when phase folding exists is shown below.

[0037] An example of a method for calculating the randomness index of an interference image by the processing unit 5 is described in detail below with reference to the flowchart in FIG. 3. Because phase images are folded between −π and π as shown in FIG. 4, it is difficult to compare acquired phase images when evaluating phase variations. For example, all four phase images in FIG. 4 contain phases between −π and π to the same extent, making it difficult to compare them when evaluating the randomness index of the interference images. Furthermore, when evaluating local phase variations, the evaluation value differs significantly between regions that cross the phase folding and regions that do not. Therefore, the processing unit 5 does not directly calculate the standard deviation of the phase image I (see FIG. 5(a)), but instead calculates and acquires a complex image G0 (see FIGS. 5(b) and 5(c)) corresponding to the phase image I (step S11). Here, the complex image G0 is an image represented by a two-dimensional array of complex numbers whose real part is image G0Re (see FIG. 5(b)) and whose imaginary part is image G0Im (see FIG. 5(c)), and the deflection angle of the complex image G0 corresponds to the phase image I.

[0038] Next, the processing unit 5 smoothes the complex image G0 using a plurality of smoothing distances (i.e., smoothing window sizes) to obtain a plurality of smoothed complex images (step S12). Specifically, as shown in FIGS. 6( a) and 6(b), the processing unit 5 smooths (convolves) the complex image G0 using a 2×2 window size to obtain a first smoothed complex image G1. As shown in FIGS. 7( a) and 7(b), the processing unit 5 smooths the complex image G0 using a 4×4 window size to obtain a second smoothed complex image G2. As shown in FIGS. 8( a) and 8(b), the processing unit 5 smooths the complex image G0 using an 8×8 window size to obtain a third smoothed complex image G3.

[0039] It should be noted that, when smoothing the complex image, smoothing filter processing is performed on each of the real part and the imaginary part. The first to third smoothed complex images G1 to G3 are defined as follows, similarly to the case of the complex image G0. The first smoothed complex image G1 is an image represented by a two-dimensional array of complex numbers whose real part is image G1Re (see FIG. 6(a)) and whose imaginary part is image G1Im (see FIG. 6(b)). The second smoothed complex image G2 is an image represented by a two-dimensional array of complex numbers whose real part is image G2Re (see FIG. 7(a)) and whose imaginary part is image G2Im (see FIG. 7(b)). The third smoothed complex image G3 is an image represented by a two-dimensional array of complex numbers whose real part is image G3Re (see FIG. 8(a)) and whose imaginary part is image G3Im (see FIG. 8(b)).

[0040] Next, the processing unit 5 divides the complex image G0, the first smoothed complex image G1, the second smoothed complex image G2, and the third smoothed complex image G3 by each other and then calculates a polar angle to obtain multiple phase difference images that index the degree of phase fluctuation with respect to a specific spatial frequency (step S13). Specifically, the processing unit 5 divides the complex image G0 by the first smoothed complex image G1 to obtain a polar angle, thereby calculating the phase difference of the first smoothed complex image G1 relative to the complex image G0 (i.e., Arg(complex image G0 / first smoothed complex image G1)). Here, Arg(C) is a function that calculates the polar angle with respect to the complex number C. In this way, a first phase difference image Z1 (see FIG. 9A) is obtained, which represents the phase fluctuation (ΔL=2) between pixels separated by a smoothing distance of 2.

[0041] The processing unit 5 divides the first smoothed complex image G1 by the second smoothed complex image G2 to obtain a polar angle, thereby calculating a phase difference between the first smoothed complex image G1 and the second smoothed complex image G2 (i.e., Arg(first smoothed complex image G1 / second smoothed complex image G2)). This results in a second phase difference image Z2 (see FIG. 9B ) that represents the phase fluctuation (ΔL=4) between pixels that are separated by a smoothing distance of 4. Furthermore, the processing unit 5 divides the second smoothed complex image G2 by the third smoothed complex image G3 to obtain a polar angle, thereby calculating a phase difference between the second smoothed complex image G2 and the third smoothed complex image G3 (i.e., Arg(second smoothed complex image G2 / third smoothed complex image G3)). This results in a third phase difference image Z3 (see FIG. 9C) that represents the phase variation between pixels that are separated by a smoothing distance of 8 (ΔL=8). Note that the phase difference between a complex number C1 of a deflection angle Φ1 and a complex number C2 of a deflection angle Φ2 is generally expressed as Arg(C1 / C2).

[0042] FIG. 10( a) is an enlarged view of a portion of a phase image including a reflective foreign particle 91. FIG. 10( b) is an enlarged view of a portion of a phase difference image corresponding to the phase image of FIG. 10( a). FIG. 10( b) shows a phase difference image in the case where ΔL = 4. The processing unit 5 can obtain the phase difference image of FIG. 10( b) based on the phase image of FIG. 10( a). FIG. 11( a) is an enlarged view of a portion of a phase image including a transparent foreign particle 92. FIG. 11( b) is an enlarged view of a portion of a phase difference image corresponding to the phase image of FIG. 11( a). FIG. 11( b) shows a phase difference image in the case where ΔL = 4. The processing unit 5 can obtain the phase difference image of FIG. 11( b) based on the phase image of FIG. 11( a).

[0043] The processing unit 5 calculates a first phase variation amount, which is a feature amount, for each pixel (each coordinate position) of the first phase difference image Z1. The first phase variation amount is the standard deviation of the first phase difference image Z1. The processing unit 5 calculates a second phase variation amount, which is a feature amount, for each pixel of the second phase difference image Z2. The second phase variation amount is the standard deviation of the second phase difference image Z2. The processing unit 5 calculates a third phase variation amount, which is a feature amount, for each pixel of the third phase difference image Z3. The third phase variation amount is the standard deviation of the third phase difference image Z3.

[0044] The processing unit 5 acquires a randomness index for each pixel based on the first to third phase fluctuation amounts for each pixel (step S14). The randomness index is a feature vector in which the first to third phase fluctuation amounts are arranged. The randomness index is a multidimensional parameter, and in this case, a three-dimensional parameter. The randomness index is acquired for each pixel of the image. Since such a randomness index can be obtained by comparing phase difference images obtained using several smoothing filters, it is possible to classify phase structures with different frequencies.

[0045] The processing unit 5 extracts the foreign particle 9 from the interference image. Specifically, the processing unit 5 extracts the foreign particle 9 using the amount of local variation in the phase difference image as an index. The example shown in FIG. 12 is a diagram illustrating an image of local phase variation in the second phase difference image Z2 where ΔL=4. The processing unit 5 cuts out a region of pixels in the second phase difference image Z2 that shows local phase variation exceeding a threshold value as a region where the foreign particle 9 is present. Note that the process of extracting the foreign particle 9 is not particularly limited, and various known methods can be used, such as using a separately captured bright-field image.

[0046] The processing unit 5 references the foreign matter classification model stored in the memory area 51 and determines the type of foreign matter 9 from the acquired degree of disorder. The foreign matter classification model is a classification standard that serves as a basis for classifying foreign matters 9. The foreign matter classification model corresponds to a classification function that quantifies the "reflective foreign matter degree" or the "transparent foreign matter degree" by substituting the degree of disorder. The reflective foreign matter degree is an index that can be associated with each foreign matter 9 and is a value representing the degree of reflectivity of the foreign matter 9 with respect to the first light L1. The transparent foreign matter degree is an index that can be associated with each foreign matter 9 and is a value representing the degree of transparency of the foreign matter 9 with respect to the first light L1. The foreign matter classification model is obtained by defining a classification surface (boundary surface) in a multidimensional space that can classify multiple types of foreign matter 9. For example, the foreign matter classification model shown in FIG. 13 is a model that classifies three types of foreign matter and is obtained by defining a classification surface MP in a three-dimensional space with the first to third phase variation amounts as axes.

[0047] The display unit 53 displays information relating to the degree of disorder calculated by the processing unit 5. The display unit 53 displays information relating to the type of foreign matter 9 determined by the processing unit 5. The display unit 53 constitutes an output unit that outputs information relating to the degree of disorder acquired by the processing unit 5. For example, the display unit 53 displays a reflectivity foreign matter degree table Tb (see FIG. 14 ) relating to the reflectivity foreign matter degree of each foreign matter 9 as information relating to the degree of disorder.

[0048] The reflectivity foreign particle degree table Tb is generated by the processing unit 5. The reflectivity foreign particle degree table Tb is a table showing the positions of multiple foreign particles 9 in the interference image and the reflectivity foreign particle degree of each of the multiple foreign particles 9. The "area" in the reflectivity foreign particle degree table Tb is the area of ​​the area corresponding to each foreign particle 9 in the interference image. The "volume" in the reflectivity foreign particle degree table Tb is the value obtained by integrating the brightness of the pixels included in the area corresponding to each foreign particle 9 in the interference image over the entire area. For example, the "volume" of an area of ​​2 x 2 pixels with a phase value of 3 [radians] is 2 x 2 x 3 = 12 [radians pixel^2]. The unit may be converted to [radians μm^2] or to [μm^3] using the proportional relationship between the thickness of the foreign particle and the phase value. In the example of the reflective foreign matter degree table Tb shown in Figure 14, a type of foreign matter 9 with a reflective foreign matter degree of 0.9999 can be determined to be a reflective foreign matter 91, and a type of foreign matter 9 with a reflective foreign matter degree of 0.02 can be determined to be a transparent foreign matter 92.

[0049] The learning unit 54 generates a foreign substance classification model by machine learning using as training data a plurality of types of interference images in the case where a foreign substance 9 is attached to the observation target object 8. The learning unit 54 stores the generated foreign substance classification model in the memory area 51. An example of a method for learning the foreign substance classification model will be described in detail below using the flowchart in FIG. 15 .

[0050] First, phase images of multiple types of observation objects 8 having foreign particles 9 attached thereto are acquired as training data (step S21). Step S21 can be realized by performing the step of detecting interference light L3 using the image sensor 4 to acquire phase images for multiple types of observation objects 8 having foreign particles 9 attached thereto. For example, the phase images used as training data include, as annotation information, information about the types of each foreign particle 9 included in the phase images. The phase images used as training data are limited to quantitative phase images and images mathematically equivalent thereto. For example, an image in which the real and imaginary parts of a phase image are stored as complex data can be used to create a corresponding quantitative phase image by arctangent calculation, and therefore can be said to be mathematically equivalent to a quantitative phase image. The phase images used as training data may be any of the following, or a combination thereof: (a) Interference images measured in advance by the manufacturer of the interference observation device 1 (hereinafter simply referred to as the "manufacturer") and stored in a computer on the network. (b) Interference images measured in advance by the user of the interference observation device 1 (hereinafter simply referred to as the "user") and stored in a computer managed by the user. (c) Interference images measured by the user when generating a foreign object classification model.

[0051] Next, the multiple types of phase images acquired in step S21 are stored in the storage area 51 (step S22). The learning unit 54 generates a foreign object classification model through machine learning using the stored multiple types of phase images as training data (step S23). The type and / or configuration of the algorithm used to generate the foreign object classification model is not particularly limited. The type of algorithm may be, for example, a random forest decision tree, a multilayer perceptron (neural network), etc. The configuration of the algorithm may include the number of intermediate layers, the number of units in each layer, the number of branches, the objective variable, the explanatory variable, etc. The type and / or configuration of the algorithm may be generated by a support vector machine or deep learning. If the foreign object classification model is a model for classifying two types of foreign objects, the type and / or configuration of the algorithm may be generated by logistic regression analysis.

[0052] Step S22 may be performed by the manufacturer, in which case the acquired interference images may be stored online or in an image database attached to the device. Step S22 may be performed by the user, in which case the acquired interference images may be stored in a file. Step S23 may be performed by the manufacturer or the user.

[0053] [Operation Example] Next, an interference observation method using the interference observation device 1 will be described with reference to Fig. 16. In the following example, a case will be illustrated in which an observation surface R inside an observation object 8 is observed and a foreign substance 9 is found to be attached to the surface (outer surface) of the film part 82 of the observation object 8.

[0054] First, the image sensor 4 acquires an image of the observation object 8 (step S1). In step S1, the position of the objective lens 13 is adjusted so that the objective lens 13 is focused on the observation surface R of the observation object 8, and an interference image is acquired by the image sensor 4. Specifically, adjustment is made so that the observation surface R and the light receiving surface of the image sensor 4 are optically conjugate by optical imaging by the objective lens 13 and the lens 41. The processing unit 5 executes the processes of steps S11 to S14 and acquires the degree of disorder for each pixel in the interference image (step S2).

[0055] Next, the processing unit 5 extracts the foreign particles 9 in the interference image (step S3). In step S3, the pixels (coordinate positions) where each foreign particle 9 exists in the interference image are extracted using the local fluctuation amount of the phase difference image obtained from the interference image as an index. Next, the processing unit 5 obtains the degree of disorder of the region of the foreign particle 9 in the interference image. The processing unit 5 obtains the degree of disorder of the image of each foreign particle region based on the degree of disorder of each pixel of the interference image obtained in step S2 and the pixels of each foreign particle 9 extracted in step S3. Then, the processing unit 5 uses the foreign particle classification model stored in the memory area 51 to discriminate the foreign particle from the degree of disorder of the image of each foreign particle region (step S4: processing step). The display unit 53 displays a reflective foreign particle degree table Tb as an output of information regarding the degree of disorder of the interference image in the region of each foreign particle 9 (step S5: output step). In step S5, the display unit 53 may further display information regarding the type of each foreign particle 9.

[0056] [Actions and Effects] In an interference image of an observation object 8 having a foreign particle 9 attached thereto, it has been found that the degree of disorder in the region of the foreign particle 9 varies depending on the type of the foreign particle 9. For example, in the case of a reflective foreign particle 91, the foreign particle 9 scatters the first light L1 and the light reflected from the foreign particle 9 overlaps with the light reflected from the observation surface (reflective surface) R, causing the image to resemble a scattered diffraction pattern, whereas it has been found that this is less likely to occur in the case of a transparent foreign particle 92. Therefore, it has been found that if the degree of disorder in the region of the foreign particle 9 in the interference image can be obtained, it becomes possible to accurately distinguish the foreign particle 9, even if it is difficult to distinguish based on the aspect ratio or optical volume alone. In this regard, the interference observation apparatus 1 and the interference observation method output information about the degree of disorder in the region of the foreign particle 9 in the interference image when the foreign particle 9 is attached to the observation object 8. Therefore, according to the interference observation device 1 and the interference observation method, it is possible to distinguish and identify the types of foreign matter 9 adhering to the observation object 8, that is, to discriminate the foreign matter 9, based on information relating to the degree of disorder in the region of the output interference image where the foreign matter 9 is located.

[0057] The interference observation device 1 and the interference observation method include a memory area 51 that stores a foreign matter classification model, and the processing unit 5 determines the type of foreign matter from the acquired degree of disorder by referring to the foreign matter classification model stored in the memory area 51. In this case, it becomes possible to discriminate the foreign matter 9 adhering to the observation object 8 by referring to the foreign matter classification model in the memory area 51.

[0058] The interference observation device 1 and the interference observation method include a learning unit 54 that generates a foreign matter classification model by machine learning using as training data a plurality of interference images in the case where a foreign matter 9 is attached to an observation object 8. In this case, it is possible to generate a foreign matter classification model by machine learning.

[0059] In the interference observation apparatus 1 and the interference observation method, the foreign matter 9 includes a reflective foreign matter and a transmissive foreign matter. In this case, it is possible to distinguish whether the foreign matter 9 is a reflective foreign matter or a transmissive foreign matter. In the interference observation apparatus 1 and the interference observation method, the reflective foreign matter 91 includes silicon, which is a semiconductor, and the transmissive foreign matter 92 includes an organic matter. In this case, it is possible to distinguish whether the foreign matter 9 attached to the observation object 8 is silicon, which has reflectivity, or an organic matter, which has transmissivity.

[0060] The interference observation device 1 and the interference observation method include a display unit 53 that displays information about the degree of disorder. In this case, information about the degree of disorder in the region of the foreign substance 9 in the interference image can be easily grasped.

[0061] In the interference observation apparatus 1 and the interference observation method, in steps S11 to S14 for acquiring the degree of randomness, a phase difference image based on a smoothed image of a complex image is used, thereby making it possible to quantify the amount of local phase variation in an interference image that includes phase folding of 2π or more. Since phase unwrapping is not used in the image processing in steps S11 to S14, there is no influence from various errors resulting from phase unwrapping. Furthermore, regardless of whether the image includes phase folding of 2π or more, it is possible to extract locations with large local phase variations using the same calculation.

[0062] In the interference observation apparatus 1 and the interference observation method, the phase variation (standard deviation) of the phase difference image is used as an example of a multidimensional parameter of the degree of disorder. The phase variation of the phase difference image can be evaluated in a number of ways by changing the size of the smoothing filter, so it is easy to make it multidimensional. Another example of a multidimensional parameter of the degree of disorder may be the contrast difference of the reflection intensity of the interference fringe amplitude in the region of the foreign substance 9 compared to the surrounding area.

[0063] The interference optical system 3 of this embodiment is configured so that the optical path length (reference optical path length) of the optical path of the second light L2 can be changed. In this case, as shown in the following example, by changing the optical path length of the optical path of the second light L2, it is possible to highlight foreign matter 9 adhering to surfaces other than the observation surface R of the object of observation 8 in the acquired interference image, making it possible to easily distinguish the foreign matter 9. That is, the processing unit 5 controls the piezoelectric element 19 to move the reference mirror 15 to the forward position (first position), and the image sensor 4 acquires a first phase image. The processing unit 5 controls the piezoelectric element 19 to move the reference mirror 15 to the rear position (second position), and the image sensor 4 acquires a second phase image. The processing unit 5 performs threshold processing, a known method, on the average image of the first phase image K1 and the second phase image K2 to extract only bright spots above a threshold, thereby acquiring a bright spot-extracted phase image, which is an interference image in which the foreign matter 9 stands out. By acquiring the degree of disorder in the region of the foreign matter 9 in this bright spot-extracted phase image, it is possible to easily distinguish the foreign matter 9. In this case, a stepping motor 18 may be used as a mechanism for moving the reference mirror 15 instead of the piezoelectric element 19. The focal condition of the optical path of the second light L2 may be changed by moving the reference objective lens 14. Instead of extracting bright spots using an average image of the first phase image K1 and the second phase image K2, either the first phase image K1 or the second phase image K2 may be used.

[0064] <Modifications> One aspect of the present invention is not limited to the above embodiment.

[0065] In the above embodiment, the randomness index is a three-dimensional parameter. However, this is not limiting, and the randomness index may be a multidimensional parameter of four or more dimensions. For example, the dimension of the randomness index may be increased arbitrarily by using four or more different smoothing window sizes when acquiring multiple types of smoothed complex images. Furthermore, the randomness index may be reduced to a one- or two-dimensional parameter by using a technique such as principal component analysis. As an example, the randomness index may be a two-dimensional parameter in which the first and second phase fluctuation amounts are arranged side by side. In this case, the foreign object classification model may be a model for classifying three types of foreign objects, as shown in FIG. 17, for example, and may be obtained by defining a classification curve MP2 in a two-dimensional space with the first and second phase fluctuation amounts as axes.

[0066] In the above embodiment, the foreign object classification model generated by the learning unit 54 through machine learning is stored in the storage area 51. However, this is not limiting. The learning unit 54 may not be provided, and a foreign object classification model generated in advance by the manufacturer or user may be stored in the storage area 51. In this case, the hardware used to generate the foreign object classification model may be the user's personal computer or a cloud server. This reduces the amount of data to be read, and speeds up the operation during initial setup.

[0067] In the above embodiment, the reflectivity foreign matter degree table Tb relating to the reflectivity foreign matter degree of each foreign matter 9 is displayed on the display unit 53. However, instead of or in addition to this, information relating to the transmittance foreign matter degree may be displayed on the display unit 53, or information relating to other degrees of disorder may be displayed on the display unit 53. In the above embodiment, the reflectivity foreign matter degree or the transmittance foreign matter degree may be quantified, and then it may be determined whether it is greater or smaller than a threshold value, and the foreign matter 9 may be discriminated based on the determination result.

[0068] In the above embodiment, the display unit 53 is provided as the output unit, but the output unit is not limited to the display unit 53, and may output information related to the degree of disorder acquired by the processing unit 5. For example, the output unit may be a transmission unit that transmits information related to the degree of disorder. For example, the output unit may be a part that outputs information related to the degree of disorder to an element outside or inside the interference observation device 1. In the above embodiment, instead of or in addition to displaying information related to the degree of disorder on the display unit 53, the information may be stored in the memory area 51 or a separate storage medium, or may be transmitted to the outside.

[0069] In the above embodiment, the foreign substance classification model is acquired by reading it from the storage area 51, but it may also be acquired by downloading it from an external source such as a cloud server, or by reading it from a recipe database attached to the apparatus. In the above embodiment, various objects may be used as the observation target 8. For example, the observation target may be a so-called capped device, a so-called bonded wafer, or a single wafer.

[0070] In the above embodiment, the foreign matter 9 is extracted (step S3) after the degree of disorder is acquired (step S2), but the degree of disorder may be acquired after the extraction of the foreign matter 9. In this case, for example, a region in the interference image where the foreign matter 9 exists is extracted by threshold processing or the like, and the degree of disorder of the interference image of the extracted foreign matter 9 is calculated. Incidentally, when the foreign matter 9 is extracted after the degree of disorder of the interference image is acquired, simple processing can be performed on the interference image, which can increase the processing speed.

[0071] In the above embodiment, the focus condition of the optical path of the second light L2 is changed by moving the reference mirror 15. However, instead of or in addition to this, the focus condition of the optical path of the second light L2 may be changed by moving the reference objective lens 14. In this case, the processing unit 5 may control the stepping motor 17 to move the reference objective lens 14 to the forward position, and then construct a first interference image based on the multiple interference images acquired by the image sensor 4. The processing unit 5 may control the stepping motor 17 to move the reference objective lens 14 to the rear position, and then construct a second interference image based on the multiple interference images acquired by the image sensor 4. Then, the processing unit 5 may separate the observed surface information and the non-observed surface information based on the first interference image and the second interference image.

[0072] In the above embodiment, the objective lens 13 is moved in the Z direction by the actuator 16 to adjust the relative position between the objective lens 13 and the observation object 8. However, instead of or in addition to this, the relative position may be adjusted by moving the stage S in the Z direction. In this case, the stage S is configured to be movable in the Z direction in addition to the XY directions. In this case, the actuator 16 may be omitted. The stage S may be movable in any direction that intersects with the direction in which the first light L1 is incident on the observation object 8, and may be movable in a direction inclined with respect to the XY plane, for example. The actuators that drive the reference objective lens 14 and the reference mirror 15 are not limited to the stepping motors 17 and 18, and may be other actuators such as servo motors.

[0073] In the above embodiment, the interference optical system 3 is configured as a Linnik interferometer, but the interference optical system 3 may be configured as a Michelson interferometer or a Mirau interferometer. The reference objective lens 14 may be omitted. In the above embodiment, the number of interference images used to construct a phase image is not particularly limited, and may be one or more.

[0074] The components in the above-described embodiments and modifications are not limited to the materials and shapes described above, and various materials and shapes can be applied. Furthermore, the components in the above-described embodiments and modifications can be arbitrarily applied to the components in other embodiments or modifications.

[0075] 1...Interference observation device, 2...Light source, 3...Interference optical system, 4...Image capture element, 5...Processing unit, 8...Observation object, 9...Foreign matter, 14...Reference objective lens, 15...Reference mirror, 51...Memory area (memory unit), 53...Display unit (output unit), 91...Reflective foreign matter, 92...Transparent foreign matter, L1...First light, L2...Second light, L3...Interference light, R...Observation surface, Tb...Reflective foreign matter degree table (information regarding disorder degree).

Claims

1. An interference observation device comprising: a light source that outputs light; an interference optical system having a reference mirror, which splits the light output from the light source into a first light and a second light, and outputs interference light between the first light reflected by an object to be observed and the second light reflected by the reference mirror; an image sensor that detects the interference light; a processing unit that acquires an interference image based on the detection result of the image sensor, and acquires a degree of disorder of an image of a foreign matter region adhering to the object to be observed in the interference image; and an output unit that outputs information regarding the degree of disorder acquired by the processing unit.

2. An interference observation device as described in claim 1, further comprising a memory unit that stores a foreign object classification model relating to the relationship between the degree of disorder of the foreign object and the type of the foreign object, and the processing unit determines the type of the foreign object from the acquired degree of disorder by referring to the foreign object classification model stored in the memory unit.

3. An interference observation device as described in claim 2, further comprising a learning unit that generates a foreign object classification model relating to the relationship between the degree of disorder of the foreign object and the type of the foreign object by machine learning using multiple types of interference images when the foreign object is attached to the object to be observed as training data.

4. An interference observation device according to claim 2 or 3, wherein the foreign matter includes a reflective foreign matter and a transparent foreign matter.

5. An interference observation device according to claim 4, wherein the reflective foreign matter includes a semiconductor, and the transparent foreign matter includes an organic matter.

6. An interference observation device according to claim 1, wherein said output section includes a display section for displaying information relating to said degree of disorder.

7. An interference observation device according to claim 1, wherein the interference optical system is configured so that the optical path length of the optical path of the second light is changeable.

8. An interference observation method comprising: a light output step of outputting light from a light source; an interference light output step of splitting the light output from the light source into first light and second light, and outputting interference light between the first light reflected by an object to be observed and the second light reflected by a reference mirror; an interference light detection step of detecting the interference light with an imaging element; a processing step of acquiring an interference image based on the detection result of the imaging element and acquiring a degree of disorder of an image of a foreign matter region adhering to the object to be observed in the interference image; and an output step of outputting information regarding the degree of disorder acquired in the processing step.

Images