TWI877038B - Depth image generation method and depth image generation system for semiconductor device inspection - Google Patents

Abstract

A depth image generation method and a depth image generation system for semiconductor device inspection are provided. The depth image generation method includes the following steps. A plurality of cross-sections of an object to be measured is captured by infrared light to form a plurality of initial cross-section images corresponding to the cross-sections, wherein the cross-sections are spaced apart from each other along a first direction. Adjacent images of the initial cross-section images are compared to generate a plurality of comparison images, and adjacent images of the comparison images are added to form a plurality of resolution-enhanced images. Adjacent images of the resolution-enhanced images are compared sequentially and the pixel with the highest grayscale value at each position of the resolution-enhanced image is selected to form a depth image corresponding to the object to be measured.

Description

Depth image generation method and depth image generation system for semiconductor device detection

The present invention relates to detection technology, and more particularly to a depth image generation method and a depth image generation system for semiconductor device detection.

In order to ensure the quality of semiconductor devices, various inspection methods and inspection systems are applied to semiconductor devices. For example, semiconductor devices such as wafers and chips may produce defects such as chipping during the manufacturing process. To this end, automatic optical inspection equipment or inspection systems can be used to confirm the location or size of these cracks. However, when the cracks occur inside the semiconductor device (for example, hidden chipping), the detection of the cracks becomes more difficult. In other words, although existing inspection methods and inspection systems have gradually met their established uses, they do not meet the requirements in all aspects. Therefore, there are still some problems to be overcome regarding the inspection methods and inspection systems used for semiconductor devices.

In some embodiments, a depth image generation method for semiconductor device detection is provided. The depth image generation method includes: photographing a plurality of cross-sections of an object to be tested with infrared light to form a plurality of initial cross-section images corresponding to the cross-sections, wherein the cross-sections are spaced apart from each other along a first direction; comparing adjacent images in the initial cross-section images to generate a plurality of comparison images, and superimposing adjacent images in the comparison images to form a plurality of resolution-enhanced images; and sequentially comparing adjacent images of the resolution-enhanced images, and selecting pixels with the highest grayscale value at each position in the resolution-enhanced images to form a depth image corresponding to the object to be tested.

In some embodiments, a depth image generation system for semiconductor device detection is provided. The depth image generation system includes a light source, an image capture module, and a processing module. The light source provides infrared light to illuminate the object to be tested. The image capture module uses infrared light to photograph a plurality of cross-sections of the object to be tested to form a plurality of initial cross-section images corresponding to the cross-sections, wherein the cross-sections are spaced apart from each other along a first direction. The processing module compares adjacent images in the initial cross-sectional image to generate a plurality of comparison images, and superimposes adjacent images in the comparison image to form a plurality of resolution-enhanced images. The processing module also sequentially compares adjacent images of the resolution-enhanced image, and selects pixels with the highest grayscale value at each position in the resolution-enhanced image to form a depth image corresponding to the object to be measured.

The depth image generation method and depth image generation system disclosed in the present invention for semiconductor device detection can construct a perspective three-dimensional image (i.e., a depth image) of a semiconductor device to perform non-contact detection of the semiconductor device through the depth image, and the depth image generation method and the depth image generation system can be applied to a variety of semiconductor devices and have a high degree of application. In order to make the features and advantages of the present invention more obvious and easy to understand, various embodiments are specifically cited below, and are described in detail with the accompanying drawings as follows.

The following disclosure provides many different embodiments or examples for implementing the provided device. Specific examples of each component and its configuration are described below to simplify the embodiments of the present disclosure, but of course are not intended to limit the present disclosure. For example, if the description refers to a first component formed on a second component, it may include an embodiment in which the first component and the second component are directly in contact, and it may also include an embodiment in which an additional component is formed between the first component and the second component so that the first component and the second component are not in direct contact. In addition, the present disclosure may repeat component symbols and/or characters in different embodiments or examples. Such repetition is for simplicity and clarity, and is not used to indicate the relationship between the different embodiments and/or examples discussed.

In some embodiments of the present disclosure, terms such as "disposed", "connected" and similar terms, unless otherwise specifically defined, may refer to two components being in direct contact, or may refer to two components not being in direct contact, wherein an additional component is located between the two structures. Terms related to disposition and connection may also include situations where both structures are movable, or both structures are fixed.

In addition, the terms "first", "second" and similar terms mentioned in this specification or patent application are used to name different components or distinguish different embodiments or scopes, and are not used to limit the upper or lower limit of the number of components, nor are they used to limit the manufacturing order or setting order of the components.

In this article, the terms "approximate", "about", and "substantially" generally mean within 10%, within 5%, within 3%, within 2%, within 1%, or within 0.5% of a given value or range. The quantities given here are approximate quantities, that is, in the absence of specific description of "about", "approximately", and "substantially", the meanings of "about", "approximately", and "substantially" can still be implied. The term "ranging from a first value to a second value" means that the range includes the first value, the second value, and other values therebetween. Furthermore, there may be a certain error between any two values or directions used for comparison. If the first value is equal to the second value, it implies that there may be an error of about 10%, or within 5%, or within 3%, or within 2%, or within 1%, or within 0.5% between the first value and the second value. If the first direction is perpendicular to the second direction, the angle between the first direction and the second direction may be between 80 degrees and 100 degrees. If the first direction is parallel to the second direction, the angle between the first direction and the second direction may be between 0 degrees and 10 degrees.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art. It is understood that these terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning consistent with the background or context of the relevant technology and the present disclosure, and should not be interpreted in an idealized or overly formal manner unless specifically defined in the embodiments of the present disclosure.

It should be understood that for the sake of clarity, some elements of the device are omitted in the drawings and only some elements are schematically shown. In some embodiments, additional components may be added to the device described below. In other embodiments, some components of the device described below may be replaced or omitted. It should be understood that in some embodiments, additional operating steps may be provided before, during and/or after the method of forming the device. In some embodiments, some of the operating steps described may be replaced or omitted, and the order of some of the operating steps described is interchangeable.

The present disclosure provides a depth image generation method and a depth image generation system for semiconductor device detection, which obtains multiple cross-sectional images of an object to be tested (e.g., semiconductor devices such as wafers and chips) by infrared light, and merges the multiple cross-sectional images into a depth image by image processing (e.g., image processing such as comparison, superposition, and subtraction described below) to effectively determine the specific location or size of defects such as cracks. In this way, subsequent processing of defective semiconductor devices can be avoided, thereby avoiding unnecessary waste. In some embodiments, the object to be tested may be or may include a wafer, a chip, a combination thereof, or other suitable semiconductor devices, but the present disclosure is not limited thereto.

Next, some embodiments of the depth image generation system of the present disclosure will be described with reference to FIG. 1 and FIG. 2 to make the present disclosure clearer and easier to understand. FIG. 1 is a block diagram showing a depth image generation system for semiconductor device detection according to some embodiments of the present disclosure. FIG. 2 is a schematic diagram showing a depth image generation method for semiconductor device detection at different stages according to some embodiments of the present disclosure.

As shown in FIG. 1 , the depth image generation system 1 is used to photograph the object to be tested 2 (i.e., the various semiconductor devices described above) to form a depth image of the object to be tested 2. The depth image generation system 1 includes a light source 10, an image capture module 12, and a processing module 14, and the image capture module 12 is electrically connected to the processing module 14.

As shown in FIG. 1 , a light source 10 provides infrared light to illuminate the object to be tested 2. In some embodiments, the wavelength of the infrared light provided by the light source 10 may be between 760 nanometers (nm) and 1 millimeter (mm). For example, the wavelength of the infrared light provided by the light source 10 may be 760nm, 800nm, 1µm, 10µm, 100µm, 500µm, 1mm, any value or any range between the above values, but the present disclosure is not limited thereto. In some embodiments, the wavelength of the infrared light may be determined according to the type of the object to be tested 2. For example, the light source 10 may emit infrared light with a longer wavelength to improve the penetration of the object to be tested 2, or the light source 10 may emit a plurality of infrared lights with different wavelengths to improve the penetration of different parts of the object to be tested 2.

Referring to FIG. 2 , in some embodiments, the light source 10 may illuminate the object 2 to be tested along the normal direction of the object 2 (e.g., the first direction DR1). For example, the light source 10 may emit infrared light L1 along the first direction DR1 to illuminate the object 2 to be tested, but the present disclosure is not limited thereto. In other embodiments, the light source 10 may also illuminate the object 2 to be tested along a second direction DR2 different from the first direction DR1. For example, the light source 10 may emit infrared light L2 along the second direction DR2 to illuminate the object 2 to be tested. In some embodiments, the first direction DR1 and the second direction DR2 may have an angle between 0 degrees and 180 degrees, such as 3 degrees, 10 degrees, 30 degrees, 45 degrees, 90 degrees, 120 degrees, 150 degrees, 175 degrees, any value or any range between the above values, but the present disclosure is not limited thereto. By making the light source 10 provide infrared light in different directions, the image capture module 12 can obtain different initial cross-sectional images, thereby providing different depth images for subsequent recognition.

In some embodiments, the light source 10 may be a backlight, a ring light, a parallel light, a diffuse light, a coaxial light set, a combination thereof, or other suitable light sources, but the present disclosure is not limited thereto. In some embodiments, the efficiency of image capture may be improved by providing a plurality of light sources 10. In some embodiments, a light switching controller connected to different light sources 10 (e.g., having different types, different wavelengths, or different illumination angles) may be provided. By switching the light mode to highlight the contrast between features, it is beneficial to obtain different initial cross-sectional images, thereby providing different depth images for subsequent recognition.

As shown in FIG. 1 and FIG. 2, the image capture module 12 uses infrared light to photograph a plurality of cross-sections of the object to be tested 2 to form a plurality of initial cross-section images ICS corresponding to the cross-sections, wherein these cross-sections are spaced apart from each other along the first direction DR1. Specifically, the image capture module 12 uses the high penetration of infrared light to the object to be tested 2 (semiconductor device) to capture the images of the plurality of cross-sections of the object to be tested 2 at specific intervals. In this way, the image of the internal structure of the object to be tested 2 (i.e., the initial cross-section image ICS) can be obtained without destroying the object to be tested 2. In some embodiments, the image capture module 12 sequentially captures the initial cross-sectional image ICS from the bottommost cross-sectional surface of the object 2 to the topmost cross-sectional surface, as shown in FIG. 2. However, the present disclosure is not limited thereto. In some embodiments, the image capture module 12 sequentially captures the initial cross-sectional image ICS from the topmost cross-sectional surface of the object 2 to the bottommost cross-sectional surface.

In some embodiments, the bottom cross-section may be the bottom surface of the object 2, but the present disclosure is not limited thereto. In some embodiments, the top cross-section may be the top surface of the object 2, but the present disclosure is not limited thereto. In other words, a specific cross-section (which may or may not include the bottom and top surfaces of the object 2) may be selected according to actual needs to capture the required initial cross-section image ICS.

In some embodiments, the image capture module 12 may include an optical lens and a photosensitive element coupled to the optical lens. For example, the optical lens may be or may include a telecentric lens, which can make the captured image not affected by lens parallax within a certain physical distance and at the same time obtain a wide depth of field effect. Alternatively, the optical lens may also be a general lens, a wide-angle lens, a telephoto lens, a combination thereof, or other suitable lenses, but the present disclosure is not limited thereto. For example, the photosensitive element may be a charge-coupled device or a complementary metal-oxide-semiconductor (CMOS), a combination thereof, or other suitable photosensitive elements, but the present disclosure is not limited thereto.

It is worth mentioning that the shapes, relative positions and sizes of the object 2 to be tested and the multiple cross-sections shown in FIG. 1 and FIG. 2 are only for illustration and are not intended to limit the present disclosure. In some embodiments, the shape of the object 2 to be tested may be or may include: a plate, such as a circular plate, an elliptical plate, a triangular plate, a rectangular plate or a polygonal plate; a cylinder, such as a cylinder, an elliptical cylinder, a triangular cylinder, a rectangular cylinder or a polygonal cylinder; an irregular block; a combination of the above or other suitable shapes, but the present disclosure is not limited thereto. In some embodiments, the spacing between each group of adjacent cross-sections may be the same or different. For example, the images of multiple cross-sections may be captured at a fixed spacing. Alternatively, the image of the cross section may be captured at a smaller interval at a position where the cracking is likely to occur, and the image of the cross section may be captured at a larger interval at a position where the cracking is not likely to occur. In some embodiments, the projection areas of the various cross sections are substantially the same to avoid difficulty in subsequent image processing. However, the present disclosure is not limited thereto. In some embodiments, when the object to be measured 2 has a structure of unequal widths such as wide at the top and narrow at the bottom, narrow at the top and wide at the bottom, or funnel-shaped, only a portion of the cross section may be captured so that the projection areas corresponding to the various initial cross section images ICS are substantially the same.

As shown in FIG. 1 and FIG. 2, the processing module 14 is used to compare adjacent images in the initial cross-sectional image ICS to generate a plurality of comparison images corresponding to the initial cross-sectional image ICS. Specifically, the current initial cross-sectional image ICS (corresponding to a certain cross-sectional surface) can be compared with the previous initial cross-sectional image ICS (corresponding to the previous cross-sectional surface) to obtain a front comparison image (e.g., the front comparison image FCn in FIG. 8), and/or the current initial cross-sectional image ICS (corresponding to a certain cross-sectional surface) can be compared with the next initial cross-sectional image ICS (corresponding to the next cross-sectional surface) to obtain a rear comparison image (e.g., the rear comparison image RCn in FIG. 8).

In the present disclosure, a “pre-comparison image” refers to an image that presents the difference between an initial cross-section image corresponding to a current cross-section and an initial cross-section image corresponding to a previous cross-section (for example, the difference may be a grayscale change value of a pixel, and it is an absolute value), and a “post-comparison image” refers to an image that presents the difference between an initial cross-section image corresponding to a current cross-section and an initial cross-section image corresponding to a subsequent cross-section.

In addition, the processing module 14 is also used to superimpose the comparison images to form a plurality of resolution-enhanced images (e.g., the resolution-enhanced image RE in FIG. 10 ). These resolution-enhanced images can be used to form a depth image corresponding to the object under test 2 (e.g., the depth image DI in FIG. 12 ). In some embodiments, the obtained depth image can be used to detect defects in semiconductor devices, such as cracks, but the present disclosure is not limited thereto. The specific steps of image processing for comparing, superimposing, and forming depth images can be referred to the generation method described later.

In some embodiments, the processing module 14 may include processing and storage components such as a processor, a computer-readable medium, and a memory to execute a computer program to implement the functions described above. Examples of the processor may include a central processing unit (CPU), a multi-core CPU, a graphics processing unit (GPU), etc., but the present disclosure is not limited thereto. Examples of the computer-readable medium may include a compact disc read-only memory (CD-ROM), a hard disk drive, an erasable programable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc., but the present disclosure is not limited thereto. Examples of memory may include dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, etc., but the present disclosure is not limited thereto.

It is worth mentioning that the term "computer program" used herein refers to an application program stored in a computer-readable medium that can be read into a memory for processing by a processor. In some embodiments, the application program can be written in any combination of one or more programming languages. Programming languages include object-oriented programming languages such as Java, Smalltalk, C++, or similar languages, and traditional procedural programming languages such as C programming language or similar programming languages.

In the above, possible configurations of the depth image generation system 1 have been described, and the functions between these configurations have been described. In the following, some embodiments of the present disclosure will be further described, using the depth image generation method executed by the depth image generation system 1 as described above. It is worth mentioning that in some embodiments, the present disclosure can also use devices with similar or identical functions to execute the depth image generation method, and is not limited to using only the devices described above.

Refer to Figures 3 to 12. Figures 3 to 5 are flowcharts of a depth image generation method for semiconductor device detection according to some embodiments of the present disclosure. Figures 6 to 12 are schematic diagrams of a depth image generation method for semiconductor device detection at different stages according to some embodiments of the present disclosure.

As shown in FIG. 2 and FIG. 3, in step S10, a plurality of cross-sections of the object to be tested 2 are photographed with infrared light to form a plurality of initial cross-section images ICS corresponding to the cross-sections, wherein the cross-sections are spaced apart from each other along the first direction DR1. As shown in FIG. 2, in some embodiments, the infrared light can illuminate the object to be tested 2 in the first direction DR1, the second direction DR2, a combination thereof, or other directions, but the present disclosure is not limited thereto.

As shown in FIG. 6 , in some embodiments, the initial cross-sectional image ICS includes the first initial cross-sectional image ICS1 to the (n+1)th initial cross-sectional image ICS(n+1), and n is a positive integer greater than or equal to 1. For example, when n=1, the initial cross-sectional image ICS includes the first initial cross-sectional image ICS1 and the second initial cross-sectional image ICS2. Alternatively, when n=2, the initial cross-sectional image ICS includes the first initial cross-sectional image ICS1, the second initial cross-sectional image ICS2, and the third initial cross-sectional image ICS3.

As shown in FIG. 3 , in step S12, adjacent images in the initial cross-section image ICS are compared to generate a plurality of comparison images, and adjacent images in the comparison images are superimposed to form a plurality of resolution enhanced images RE. In some embodiments, the image processing performed by a device such as the processing module 14 may further include sub-steps S140 to S148, which are used to obtain comparison images (e.g., a front comparison image and a rear comparison image) of each cross-section, and the obtained front comparison image and rear comparison image can be used to form the resolution enhanced image RE.

As shown in FIG. 4 and FIG. 7 , in sub-step S120, the first initial cross-sectional image ICS1 and the second initial cross-sectional image ICS2 are subtracted to form the first rear comparison image RC1 (which can be regarded as the first resolution enhanced image RE1 in the resolution enhanced image RE). Specifically, in this sub-step, the first initial cross-sectional image ICS1 is the initial cross-sectional image corresponding to the bottommost cross-sectional image. Therefore, in the absence of the previous initial cross-sectional image, the first initial cross-sectional image ICS1 will only be subtracted from the next initial cross-sectional image (i.e., the second initial cross-sectional image ICS2). In this case, the obtained first post-comparison image RC1 can be directly regarded as the first resolution enhanced image RE1 in the resolution enhanced images RE.

As shown in FIGS. 4 and 8 , in sub-step S122 , the nth initial cross-section image ICSn is subtracted from the (n-1)th initial cross-section image ICS(n-1) to form the nth front comparison image FCn.

As shown in FIG. 4 and FIG. 8 , in sub-step S124 , the nth initial cross-section image ICSn is subtracted from the (n+1)th initial cross-section image ICS(n+1) to form the nth post-comparison image RCn.

As shown in FIG. 4 and FIG. 8, in sub-step S126, the nth front comparison image FCn and the nth rear comparison image RCn are superimposed (add) to form the nth resolution enhanced image REn in the resolution enhanced image RE. By comparing the current initial cross-section image with the previous initial cross-section image and the next initial cross-section image, and superimposing the comparison results together, a high-resolution resolution enhanced image RE corresponding to the current cross-section can be obtained.

As shown in FIG. 4 and FIG. 9, in sub-step S128, the (n+1)th initial cross-sectional image ICS(n+1) is subtracted from the nth initial cross-sectional image ICSn to form the (n+1)th front comparison image FC(n+1) (which can be regarded as the (n+1)th resolution enhanced image RE(n+1) in the resolution enhanced image RE). In this sub-step, the (n+1)th initial cross-sectional image ICS(n+1) is the initial cross-sectional image corresponding to the topmost cross-sectional image. Therefore, in the absence of a subsequent initial cross-sectional image, the (n+1)th initial cross-sectional image ICS(n+1) is only subtracted from the previous initial cross-sectional image (i.e., the nth initial cross-sectional image ICSn). In this case, the obtained (n+1)th previous comparison image FC(n+1) can be directly regarded as the (n+1)th resolution enhanced image RE(n+1).

As shown in FIG. 3 , following step S12, in step S14, image processing is performed on the resolution enhanced image RE to form a depth image corresponding to the object to be measured 2, wherein the image processing includes: comparing adjacent images of the resolution enhanced image RE, and selecting pixels with the highest grayscale value at each position in the resolution enhanced image RE to form a depth image DI corresponding to the object to be measured 2. By selecting pixels with higher grayscale values at each position in the resolution enhanced image RE, a depth image DI with a sense of three-dimensionality can be formed. In some embodiments, the above-mentioned image processing performed by a device such as the processing module 14 may further include sub-steps S140 to S146.

As shown in FIG. 5 and FIG. 10, in sub-step S140, the first resolution enhanced image RE1 to the (n+1)th resolution enhanced image RE(n+1) are compared sequentially to form a forward depth image PDI. In the present disclosure, "sequential comparison" in sub-step S140 refers to comparing the first resolution enhanced image RE1 with the second resolution enhanced image RE2 to the (n+1)th resolution enhanced image RE(n+1) in sequence at the corresponding pixels at each position in the image, and selecting pixels with higher grayscale values. Since the comparison is made from the bottom cross-section to the top cross-section, it is called a forward depth image. In some embodiments, the layer number of the pixel with the highest grayscale value can be recorded according to the order of appearance. For example, once the gray level of the pixel in the xth layer is determined to be the highest value 255, even if a pixel with gray level 255 appears again in a subsequent comparison process (eg, comparison with the x+yth layer), the layer with the highest pixel is only determined to be the xth layer.

As shown in FIG. 5 , following sub-step S140, in sub-step S142, the forward depth image is adjusted by the preset valve value to form a depth image. In the present disclosure, the “adjustment” in sub-step S142 refers to deleting pixels in the unclear area according to the preset valve value. For example, the unclear area can be defined as: the pixels in a certain area of all cross-sections do not have a clear maximum grayscale value (for example, the same black noise (dots) or white noise (dots) are presented in any cross-section image). For example, when the grayscale value of a pixel is 10, and the grayscale values of the corresponding pixels in other layers (for example, the 2nd layer and the 13th layer) are all 10, then the position can be called an unclear area. In some embodiments, the inizum residual method may be used to delete the unclear region, but the present disclosure is not limited thereto. Alternatively, the Otsu algorithm may be used to delete the unclear region.

As shown in FIG. 5 and FIG. 11, following sub-step S140, in some embodiments, sub-step S144 and sub-step S146 may replace sub-step 142 to form a depth image. In sub-step S144, the (n+1)th resolution enhanced image RE(n+1) is sequentially compared to the first resolution enhanced image RE1 to form a reverse depth image NDI. Similarly, in the present disclosure, "sequentially comparing" in sub-step S144 refers to comparing the corresponding pixels at each position in the (n+1)th resolution enhanced image RE(n+1) with the nth resolution enhanced image REn to the first resolution enhanced image RE1 in sequence, and selecting pixels with higher grayscale values. Since the comparison is made from the topmost cross section toward the bottommost cross section, it is called a reverse depth image.

As shown in FIG. 5 and FIG. 12, following sub-step S144, in sub-step S146, the forward depth image PDI and the reverse depth image NDI are compared to form a depth image DI. In the present disclosure, the "comparison" in sub-step S146 refers to adjusting the unclear areas of the forward depth image PDI and the reverse depth image NDI to obtain a clear and three-dimensional depth image DI. As mentioned above, since the layer number of the pixel with the highest grayscale value may be recorded according to the order of appearance, when the layer numbers of the pixel with the highest grayscale value appearing in the forward depth image PDI and the reverse depth image NDI are different (for example, one is in the xth layer and the other is in the x+yth layer), this area (or this pixel) can be identified as an unclear area.

Referring to FIGS. 13A to 13F, they respectively show the initial cross-sectional image, the comparison image, the forward depth image, the reverse depth image, the comparison image between the forward depth image and the reverse depth image, and the depth image of the object to be tested according to some embodiments (or application examples) of the present disclosure. As shown in FIG. 13A, this is an initial cross-sectional image of a certain layer (e.g., the horizontal layer where the circuit layer is located) of the object to be tested photographed by infrared light. As shown in FIG. 13B, this is a comparison image of a certain layer (e.g., the horizontal layer where the circuit layer is located) of the object to be tested photographed by infrared light. As shown in FIG. 13C, this is a forward depth image superimposed from the comparison image of the bottom layer all the way up by processing the film group. As shown in FIG. 13D, this is a reverse depth image that is superimposed from the topmost comparison image all the way down by the processing module. As shown in FIG. 13E, after comparing the forward depth image and the reverse depth image, an unclear area can be confirmed, such as the black and white noise area UA in the figure. As shown in FIG. 13F, after removing the unclear area, a depth image of the object to be tested can be obtained. In some embodiments, the defects of the object to be tested can be confirmed by the depth image shown in FIG. 13F.

In summary, the present disclosure provides a depth image generation method and a depth image generation system for semiconductor device inspection. After using infrared light to capture the initial cross-sectional image of the object to be tested, the present disclosure improves the resolution of the initial cross-sectional image by a first image processing, and forms a three-dimensional depth image by a second image processing. In this way, the defects in the object to be tested (such as internal hidden chipping defects) can be confirmed without destroying the object to be tested through the formed depth image, so as to avoid subsequent processing of defective semiconductor devices, thereby causing unnecessary waste.

Several embodiments are summarized above so that those skilled in the art can better understand the perspective of the embodiments disclosed herein. Those skilled in the art should understand that other processes and structures can be designed or modified based on the embodiments disclosed herein to achieve the same purpose and/or advantages as the embodiments introduced herein. Those skilled in the art should also understand that such equivalent processes and structures do not deviate from the spirit and scope of the disclosure and can be variously changed, substituted and replaced without violating the spirit and scope of the disclosure.

1: Depth image generation system 10: Light source 12: Image acquisition module 14: Processing module 2: Object to be measured DR1: First direction DR2: Second direction FCn: nth front comparison image FC(n+1): (n+1)th front comparison image ICS: Initial cross-section image ICS1: 1st initial cross-section image ICS2: 2nd initial cross-section image ICS3: 3rd initial cross-section image ICSn: nth initial cross-section image ICS(n-1): (n-1)th initial cross-section image ICS(n+1): (n+1)th initial cross-section image L1: Infrared light L2: Infrared light NDI: Reverse depth image PDI: Forward depth image RC1: 1st rear comparison image RCn: Comparison image after the nth time RE: Resolution enhanced image RE1: 1st resolution enhanced image REn: nth resolution enhanced image RE(n+1): (n+1)th resolution enhanced image S10-S14: Steps S120-S128: Sub-steps S140-S144: Sub-steps UA: Area

The following detailed description in conjunction with the attached drawings will provide a better understanding of the viewpoints of the disclosed embodiments. It is worth noting that, according to standard industry practices, some features may not be drawn in proportion. In fact, the sizes of different features may be increased or decreased for clarity. FIG. 1 is a block diagram of a depth image generation system for semiconductor device detection according to some embodiments of the present disclosure. FIG. 2 is a schematic diagram of a depth image generation method for semiconductor device detection at different stages according to some embodiments of the present disclosure. FIG. 3 is a flow chart of a depth image generation method for semiconductor device detection according to some embodiments of the present disclosure. FIG. 4 is a flow chart of a depth image generation method for semiconductor device detection according to some embodiments of the present disclosure. FIG. 5 is a flow chart showing a depth image generation method for semiconductor device detection according to some embodiments of the present disclosure. FIG. 6 to FIG. 12 are schematic diagrams showing the depth image generation method for semiconductor device detection at different stages according to some embodiments of the present disclosure. FIG. 13A shows an initial cross-sectional image of the object to be tested according to some embodiments of the present disclosure. FIG. 13B shows a comparison image of the object to be tested according to some embodiments of the present disclosure. FIG. 13C shows a forward depth image of the object to be tested according to some embodiments of the present disclosure. FIG. 13D shows a reverse depth image of the object to be tested according to some embodiments of the present disclosure. FIG. 13E shows a comparison diagram between the forward depth image and the reverse depth image of the object to be tested according to some embodiments of the present disclosure. Figure 13F shows the depth image of the object to be tested according to some embodiments of the present disclosure.

S10-S14: Steps

Claims (16)

A method for generating a depth image for semiconductor device detection comprises: Photographing a plurality of cross-sections of an object to be tested with infrared light to form a plurality of initial cross-section images corresponding to the cross-sections, wherein the cross-sections are spaced from each other along a first direction; Comparing adjacent images in the initial cross-section images by a processing module to generate a plurality of comparison images, and superimposing adjacent images in the comparison images by the processing module to form a plurality of resolution-enhanced images; and Comparing adjacent images of the resolution-enhanced images sequentially by the processing module, and selecting pixels with the highest grayscale value at each position in the resolution-enhanced images to form a depth image corresponding to the object to be tested. The depth image generation method for semiconductor device detection as described in claim 1, wherein the step of sequentially comparing adjacent images of the resolution-enhanced images further includes: Along the first direction, the processing module sequentially compares adjacent images of the resolution-enhanced images to form a forward depth image. The depth image generation method for semiconductor device detection as described in claim 2, wherein the step of sequentially comparing adjacent images of the resolution-enhanced images further includes: Adjusting the forward depth image with a preset valve value by the processing module to form the depth image. The depth image generation method for semiconductor device detection as described in claim 2, wherein the step of sequentially comparing adjacent images of the resolution-enhanced images further includes: Along a second direction relative to the first direction, the processing module sequentially compares adjacent images of the resolution-enhanced images to form a reverse depth image. The depth image generation method for semiconductor device detection as described in claim 4, wherein the step of sequentially comparing adjacent images of the resolution enhanced images further includes: Comparing the forward depth image and the reverse depth image by the processing module to form the depth image. A depth image generation method for semiconductor device detection as described in claim 1, wherein the initial cross-sectional images include a first initial cross-sectional image to an (n+1)th initial cross-sectional image, wherein n is a positive integer greater than or equal to 1, and the step of comparing adjacent images in the initial cross-sectional images includes: The processing module performs subtraction on the first initial cross-sectional image and a second initial cross-sectional image to form a first comparison image among the comparison images; The processing module performs subtraction on an nth initial cross-sectional image and an (n-1)th initial cross-sectional image to form a front comparison image; The processing module performs subtraction on the nth initial cross-section image and the (n+1)th initial cross-section image to form a rear comparison image; The processing module performs superposition on the front comparison image and the rear comparison image to form an nth comparison image among the comparison images; and The processing module performs subtraction on the (n+1)th initial cross-section image and the nth initial cross-section image to form an (n+1)th comparison image among the comparison images. The depth image generation method for semiconductor device detection as described in claim 1, wherein the step of photographing multiple cross-sections of the object to be tested with infrared light includes: Providing a light source, and the light source illuminates the object to be tested by coaxial light or side light. The depth image generation method for semiconductor device inspection as described in claim 1 further includes: After forming the depth image, using the depth image to confirm the defects of the object to be tested by the processing module. A depth image generation system for semiconductor device detection includes: a light source, providing an infrared light to illuminate an object to be tested; an image capture module, photographing a plurality of cross-sections of the object to be tested by the infrared light to form a plurality of initial cross-section images corresponding to the cross-sections, wherein the cross-sections are spaced apart from each other along a first direction; and A processing module compares adjacent images in the initial cross-sectional images to generate a plurality of comparison images, and superimposes adjacent images in the comparison images to form a plurality of resolution-enhanced images. The processing module also sequentially compares adjacent images of the resolution-enhanced images, and selects pixels with the highest grayscale value at each position in the resolution-enhanced images to form a depth image corresponding to the object to be tested. A depth image generation system for semiconductor device detection as described in claim 9, wherein the processing module is used to: Sequentially compare adjacent images of the resolution-enhanced images along the first direction to form a forward depth image. A depth image generation system for semiconductor device detection as described in claim 10, wherein the processing module is used to: Adjust the forward depth image by a preset valve value to form the depth image. A depth image generation system for semiconductor device detection as described in claim 10, wherein the processing module is used to: Sequentially compare adjacent images of the resolution-enhanced images along a second direction relative to the first direction to form a reverse depth image. A depth image generation system for semiconductor device detection as described in claim 12, wherein the processing module is used to: Compare the forward depth image and the reverse depth image to form the depth image. A depth image generation system for semiconductor device detection as described in claim 9, wherein the initial cross-sectional images include a first initial cross-sectional image to an (n+1)th initial cross-sectional image, and n is a positive integer greater than or equal to 1, wherein the processing module is used to: Subtract the first initial cross-sectional image from a second initial cross-sectional image to form a first comparison image among the comparison images; Subtract an nth initial cross-sectional image from an (n-1)th initial cross-sectional image to form a front comparison image; Subtract the nth initial cross-sectional image from the (n+1)th initial cross-sectional image to form a rear comparison image; Superimposing the front comparison image and the rear comparison image to form an nth comparison image among the comparison images; and Subtracting the (n+1)th initial cross-section image from the nth initial cross-section image to form an (n+1)th comparison image among the comparison images. A depth image generation system for semiconductor device detection as described in claim 9, wherein the light source illuminates the object to be detected by coaxial light or side light. A depth image generation system for semiconductor device inspection as described in claim 9, wherein the processing module confirms defects of the object to be inspected through the depth image.