JP2008096605A - Alignment microscope, method for detecting mark and method for manufacturing stacked semiconductor device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve such a problem that it is necessary to layer many wafers where a predetermined circuit is formed in order to manufacture a stacked semiconductor device, and it is important to accurately detect the position of an alignment mark in order to align the wafers to be layered when layering many wafers, and such an effect becomes an important element in the yield of manufacture, and especially, it is difficult to accurately and rapidly detect the alignment mark, because the alignment mark must be detected through the wafer when layering a next wafer on a wafer layered body where the wafers have been already layered to form layer structure. <P>SOLUTION: AF optical systems of both an active type AF optical system and a passive type AF optical system are attached to a microscope for detecting a mark, and the position of a surface opposite from a mark forming surface is detected by the active type AF optical system, and by using the result of the above detection, the mark forming surface is moved to the action range of the passive type AF optical system. At such a time, infrared light is for the passive type AF optical system. Thus, the mark is rapidly and accurately detected to secure positional accuracy in sticking. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an apparatus for manufacturing a stacked semiconductor device, and more particularly to a microscope having an AF mechanism for highly accurate alignment between stacked substrates.

    In recent years, portable electronic devices such as mobile phones, notebook computers, portable audio devices, and digital cameras have made remarkable progress. Along with this, in addition to improving the performance of the chip itself as well as the performance of the chip itself, improvements in the chip mounting technology are also required, especially from the viewpoint of reducing the chip mounting area and driving the semiconductor device at high speed. There is a need for improved packaging technology.

  In order to reduce the chip mounting area, stacking chips is used to increase the amount of mounted chips without increasing the mounting area, thereby reducing the effective mounting area. For example, JP-A-2001-257307, 2002-050735, and JP-A-2000-349228 disclose such a technique. The first one is based on a wire bond system in which a chip and a chip or a chip and a mounting substrate are connected by a wire. The second one is based on a flip chip method in which a chip and a chip or a chip and a mounting substrate are connected via a micro bump provided on the back surface of the chip. In the third method, the chip and the chip or the chip and the mounting substrate are connected by using both the wire bond method and the flip chip method.

  In order to increase the driving speed of a semiconductor device, a method realized by reducing the thickness of the chip and using a through electrode is effective. For example, Japanese Patent Laid-Open No. 2000-208702 shows an example of mounting with a thickness of micron.

  In the wire bond method, a wire is stretched around the semiconductor bare chip. For this reason, a large occupied area larger than the occupied area of the semiconductor bare chip itself is required, and it takes time because the wires are stretched one by one. On the other hand, in the flip chip method, since the connection is made by the micro bump formed on the back surface of the semiconductor bare chip, the area for connection is not particularly required, and the area necessary for mounting the semiconductor bare chip is the semiconductor bare chip. It can be almost equal to its own area. Further, since the connection surface can have all the bumps necessary for connection, connection with the wiring board can be performed in a lump. Therefore, the flip-chip method is the most suitable method for minimizing the area occupied by mounting the semiconductor bare chip to achieve high-density mounting, reducing the size of the electronic device and shortening the construction period.

  In addition to the improvement of the chip-mounting substrate and the connection method between the chip and the chip, as a means for reducing the manufacturing cost, a rewiring layer or Connection bumps are formed, and in some cases, sealing with resin is performed. A semiconductor device in which processing at the wafer level is effective is a semiconductor device having a high manufacturing yield and a small number of pins, and has many advantages especially in memory production. (NIKKEI MICRODEVICE February 2000, page 56 and NIKKEI ELECTRONICS 2003.9.1 P.127).

  On the other hand, development of a manufacturing apparatus for manufacturing a semiconductor device at the wafer level is also eagerly performed. For example, the literature introduces an apparatus for aligning and bonding wafers to be bonded. (P. Lindner et al .: 2002 Electronic Component and Technology Conference P.1439). In addition, a similar technique is disclosed in Japanese Patent Laid-Open No. 9-148207.

  By the way, when manufacturing a laminated type three-dimensional semiconductor chip (semiconductor device) by wafer lamination / electrode bonding at the wafer level as described above, alignment between wafers to be bonded is important, and alignment on the wafer to be bonded is performed. It is required to detect a mark quickly with high accuracy. Conventionally proposed such mark detection methods include the following methods.

Patent Document 1: A two-field recognizing unit is inserted between wafers facing each other in close proximity, and the focus of the mark is determined based on the degree of blurring of the mark image when the recognizing unit is moved in the optical axis direction. (Actually, this calculates the distance to the mark)
Patent Document 2: The first mark on the substrate is aligned with the reference position of the infrared camera, then the elements are opposed to each other, and the distance between the substrate and the element surface is measured to adjust the element surface to be within the depth of focus of the infrared camera. Then, the second mark and the first mark on the element are clearly observed.

  Patent Document 3: When observing the bump bonding surface, the height of the non-bonding surface of the element is measured with a laser displacement meter, and the object surface of the infrared camera is aligned with the surface whose height is adjusted by the thickness of the element.

  In a lithography process in manufacturing a semiconductor device, an autofocus (AF) function is used as a function related to mark detection using a microscope. This function is a function for quickly obtaining a clear image (referred to as an in-focus image or an in-focus image) of the alignment mark.

  Specifically, this function detects the position (focus state) of the detection target mark in the microscope optical system, and based on the position information (focus information), the formation surface of the detection target mark is accurately determined by the object surface of the microscope. Or the object surface of the microscope is aligned with the mark forming surface, and the target mark can be clearly observed. An optical system incorporated for realizing this mechanism is called an AF optical system. Regarding the AF function, the function of simply detecting the position of the observation object on the optical axis is sometimes referred to as the AF function, and in the present invention, this function is meant in a broad sense. This AF mechanism is used for, for example, an operation of aligning the object surface of the projection optical system that projects the pattern on the mask onto the wafer surface and the wafer surface.

There are two types of optical systems incorporated in the AF mechanism.
The first is what is called the active method. In order to measure the position of the observation target surface (mark formation surface) with respect to the observation optical system of the microscope, this type of optical system irradiates the observation target surface with a light beam in addition to the observation optical system and light source of the microscope. An optical system and a light source for projecting an optical system or a slit (index), and an optical system and an optical sensor for detecting a reflected light beam or a slit image reflected from the observation target surface are included. Then, the coordinates of the reflection point of the light beam on the surface to be observed, the start point and direction of the traveling vector of the reflected light beam reflected, or the conjugate position coordinates of the slit image are detected. That is, when the position of the observation target surface changes on the optical axis of the observation optical system, the state of the reflected light beam changes or the positional relationship between the slit image and the observation target surface changes to detect the slit image. This utilizes the fact that the re-imaging position of the lens changes. (For example, refer to JP-A-6-13282 and JP-A-2002-40322). One of the active methods is triangulation.

  The second is called a passive system. In this method, a mark formed on the observation target surface is directly observed by an observation optical system. At this time, the mark and the AF optical system are relatively moved to measure the characteristics of the mark image, particularly the degree of blurring of the image (spatial frequency characteristics, contrast) as a function of the position in the AF optical system. A mark position having a predetermined value is obtained. When measuring the blur characteristic, the image position with the least blur is detected. In general, the image position having the highest spatial frequency in the spatial frequency characteristics of the image and the image position having the highest contrast of the image are detected. (For example, see Patent Document 1, Japanese Patent Laid-Open No. 5-21318, Japanese Patent Laid-Open No. 2003-218137).

Furthermore, as a method for measuring the position of the observation surface (the position of the mark surface) without using an optical system, a method using a capacitance sensor is known. (For example, refer to JP 2001-332595 A)
JP 2001-308597 A JP 2000-243762 A Japanese Patent No. 3395721

  As described above, when two substrates on which semiconductor devices are formed are bonded to each other, alignment between the substrates is important, and various methods have been proposed. As a result of examining the mark detection methods described in Patent Documents 1 to 3, the inventor of the present application has found the following problems. In the method described in Patent Document 1, a two-field observation device is inserted between substrates to be bonded to observe marks formed on the respective substrates, and the relative positional relationship between the marks can be measured. This is an excellent method. However, the two-field observation apparatus has a size (thickness), and a lower limit is imposed on the lower limit of the distance between the two substrates held close to each other. For this reason, if the two-field observation device is removed after the relative positional relationship between the two marks is measured and the substrate is overlaid, the relative positional relationship between the marks changes due to the movement of the substrate, and the paste is applied with a predetermined accuracy. I could not match. Although it is conceivable to reduce the thickness of the two-field observation apparatus in order to improve the bonding position accuracy, the distortion of the observation image increases and the mark position error increases. Next, in the method described in Patent Document 2, the first substrate is brought into focus, the distance between the first substrate and the second substrate is measured, and the distance is a predetermined value (depth of focus). The two marks are simultaneously and clearly observed by adjusting so as to be within the range. This method is effective when the numerical aperture of the observation device (optical system) is small and the depth of focus is large, but when observing the mark at a high magnification, the two marks cannot be detected simultaneously. . Further, when the first substrate is placed at the in-focus position with respect to the observation optical system, the first substrate is placed at a predetermined position. At this time, the depth of focus of the observation optical system is small (a high numerical aperture). It is difficult to observe even the mark on the first substrate. Further, the method described in Patent Document 3 obtains the height of the chip back surface (position on the optical axis of the microscope) from the surface height of the chip and the thickness of the chip, and is formed on the back surface of the chip based on this information. The mark is observed by capturing the image once. This method is effective when the observation accuracy is not so high, but when high-precision mark observation accuracy is required, a clear mark image necessary for high-precision mark detection can be obtained with only one adjustment. There wasn't. Further, since the laser displacement meter and the infrared microscope are separate, it is necessary to always calibrate the distance between the laser displacement meter and the infrared microscope.

  As described above, since the mark cannot be detected with high accuracy by the conventional method, an error occurs in the alignment between the substrates to be bonded, and as a result, it becomes difficult to manufacture the bonded three-dimensional stacked semiconductor device at the wafer level. It was.

  The present invention has been made to solve the above-described problems, and has an object to quickly and accurately detect the position of the alignment mark formed on the back surface of the wafer. Another object of the present invention is to provide an alignment apparatus for manufacturing a bonded three-dimensional semiconductor device at a wafer level with a high yield, and a method for manufacturing the semiconductor device.

The means of the present invention for solving the above-described problems is as follows.
A microscope having an autofocus function,
An active AF optical system and a passive AF optical system sharing an objective lens;
A control unit that outputs focus state information;
Have
This is a microscope using an infrared light source as a light source of the passive AF optical system.

  If the microscope has an active AF optical system and a passive AF optical system that share the objective lens as in the present invention, the height measurement of the wafer surface (opposite surface to be bonded) is active. It is possible to perform alignment by performing focusing using the AF system and then focusing using the passive AF system in consideration of the thickness of the wafer, and a positional offset amount between the two AF systems. Therefore, the position detection accuracy of the mark (on the bonding surface) is improved.

  Furthermore, if the alignment between the stacked wafers is performed when manufacturing the stacked three-dimensional semiconductor device by the alignment method using the microscope, the wafer position and the height can be easily measured with high accuracy. This is facilitated and a reduction in manufacturing yield is prevented.

As described above, the mark formed on the back surface of the wafer is detected using a microscope equipped with an active optical system and a passive AF optical system in which the objective lens is shared, thereby obtaining a three-dimensional coordinate value of the mark. It becomes possible to measure easily with high accuracy. In addition, since the detection system has only one main part of the optical system, the detection apparatus is effective from the viewpoint of manufacturing cost.

An embodiment of a microscope and a mark detection method of the present invention will be described with reference to FIGS.
The microscope 10 includes, as the observation optical system 100 , a light source 116, an objective lens 112 (a lens system made up of a plurality of lenses, constituting an epi-illumination system), and an eyepiece system 190 (a lens system made up of a plurality of lenses). ) As a main component, and has an active AF optical system and a passive AF optical system for focusing. In the figure, reference numeral 110 denotes an illumination system and detection system of an active AF optical system, and reference numeral 120 denotes a detection system of a passive AF optical system.

  The active AF optical system is mainly composed of an illumination system, an objective lens 112, and a detection system. The illumination system includes a light source 121, a slit 122, a condenser lens 124, and a light beam limiting mask 130, and the detection system is an imaging lens. 126 and a first two-dimensional sensor 128. Here, the function of these components is described. (In order to limit the explanation to the essential part, the description of the polarizing optical element, the dichroic optical element, etc. used in the beam splitter is omitted.)

  The slit 122 is illuminated by the light beam emitted from the light source 121, and the light beam 152 emitted from the slit 122 can pass only the light beam 154 on one side of the optical axis by the light beam limiting mask 130, and further passes through the beam splitter 162. Although a passive AF optical system will be described later, a beam splitter 142 is disposed in the passive AF optical system, and the optical axis of the active AF optical system is bent so as to coincide with the optical axis of the passive AF optical system. Playing a role. The light beam 154 that has passed through the beam splitter 162 is reflected by the beam splitter 142 and condensed by the objective lens 112, and an image of the slit 122 is formed. The image of the slit 122 is formed by the condenser lens 124 and the objective lens 112. The position where the slit 122 is formed is the object surface of the microscope (more precisely, the object surface of the observation optical system, and the surface on which the object to be observed is in focus). ), The position of the slit 122 is adjusted.

FIG. 1 shows a case where a slit image is formed on the wafer surface 101 of the wafer, which is the object to be observed, and the wafer surface 101 is in focus when observed with the observation optical system 100 of the microscope. The light beam 154 is reflected by the wafer surface 101 and takes an optical path that is symmetrical to the optical path at the time of incidence relative to the optical axis. The reflected light 155 is reflected by the beam splitter 142, further reflected by the beam splitter 162, and condensed on the first two-dimensional detector 128 by the imaging lens 126. Although FIG. 1 shows a state in which a slit image is formed on the wafer surface 101 and an image conjugate with the slit image is formed on the two-dimensional detector 128, FIG. The positional relationship (focusing state, degree of focus) will be described.

  FIG. 2A shows a case where an image P of the slit is formed on the wafer surface (the same as the case of FIG. 1). In this case, the reflected light 155 travels along the optical path as if it came from the image point P. However, since the light beam is restricted to one side of the optical axis by the light beam limiting mask, the region through which the incident light beam passes and the region through which the reflected light beam pass do not overlap. FIG. 2B shows a case in which the slit image P is formed in front of the wafer surface 101 (on the optical system side), and the reflected light beam 155 follows an optical path that is emitted from the virtual image P ′ inside the wafer. move on. FIG. 2C shows a case where the slit image P is formed behind the wafer surface 101 (the side away from the optical system), and the reflected light beam 155 appears from the virtual image P ′ in front of the wafer surface 101. Follow the light path. In these cases, FIG. 3 shows what kind of difference occurs in the output in the detection system.

  FIG. 3 is a principle diagram of the position detection system of the slit image in the active AF optical system, showing the imaging lens 126, the first two-dimensional sensor 128, and the light rays related to the re-imaging of the slit image. . The positional relationship between the imaging lens 126 and the sensor 128 is such that when a slit image is formed on the wafer surface 101 as shown in FIG. 2A, a conjugate image Fa via the imaging lens 126 is the first detector 128. It is designed to be formed on top. However, as shown in FIG. 2B, when it is considered that the reflected light beam 155 is emitted from the virtual image P ′ formed farther than the wafer surface 101, the image conjugate with the virtual image P ′ is the point Fb in FIG. Formed. On the contrary, when the virtual image P ′ is formed near the wafer surface 101 as shown in FIG. 2C, an image conjugate with the virtual image P ′ is formed at the Fc point in FIG.

  By the way, as described above, the light flux that forms an image on Fa, Fb, and Fc passes through only one side of the optical axis by the light flux limiting mask 130. Accordingly, when an image is formed on Fa, the first detector 128 has an output on the optical axis portion of the imaging lens 127, but when the image is formed on Fb, the first detector 128 has an output. If an image is formed on Fc, the output is made to the region 302 on the first detector 128. Therefore, if the relationship between the output region and the position of the conjugate image with P ′ is measured and calibrated in advance, the control unit 182 (FIG. 1) generates a conjugate image based on the output from the first detector 128. The position is obtained, and the positional relationship (focus state) between the wafer surface 101 and the slit image P is determined and output.

  The positional relationship between the wafer surface 101 and the slit image can be obtained by scanning the wafer surface 101 on the optical axis of the objective lens 112 in addition to the method using the previously calibrated output of the detector 128 as described above. It is also possible to adopt a method of arranging the wafer surface at a position where the output of the first detector 128 is output from the optical axis equivalent portion of the imaging lens 126.

  Next, the passive AF optical system will be described. The main members of the passive AF optical system are an objective lens 112, an AF eyepiece 114, an illumination system 116, a focus detection beam splitter 149, and a second two-dimensional sensor 129 with reference to FIG. The illumination system 116 is connected to the objective lens 112 via a beam splitter 147. As described above, in the passive AF method, the characteristics of the image of the mark formed on the wafer surface (for example, the amount of blur of the image such as spatial frequency characteristics) are measured at several points on the optical axis. It is. In this embodiment, measurement is performed using a focus detection beam splitter 147. This will be described below. When the image formed on the eyepiece 190 of the observation optical system is correctly focused, the two light beams divided by the beam splitter 149 have the same characteristics (for example, image blur) on the second two-dimensional sensor 129. Thus, the positional relationship between the beam splitter 149 and the second two-dimensional sensor 129 is adjusted.

As an example, adjustment is performed so that the spread of the point images is the same or the contrast of the images has the same value. (See Patent Document 3). FIG. 5A shows such a state. Consider a case in which the position I of the image is formed farther than the in-focus position (that is, on the side away from the beam splitter 149) as shown in FIG. The two output values of the second two-dimensional sensor 129 are different from each other, indicating that the image of the observation optical system 100 is not correctly focused. For example, in FIG. 5B, the light beam forming the image I becomes more blurred than the light beam forming the image I ′, and the light beam forming the image I shows a larger blur on the sensor. Therefore, if the blur output of the two images and the deviation between the observation surface of the objective lens and the mark image are calibrated, the positional relationship between the mark and the observation surface of the objective lens can be measured by one observation.

  As another method of the passive AF, as the mark is moved on the optical axis of the microscope, the mark image detected by the detector 129 is observed for blur, for example, the spatial frequency distribution is detected by Fourier transform to detect the mark image. This is a method for acquiring information on the focus state and detecting the image position including the highest order term. In principle, the beam splitter 194 is unnecessary when this method is employed.

  In addition, although it is a light source, the light source which emits infrared rays is used for the light source 116 of a passive system optical system so that a mark can be observed through a wafer. This infrared light source also serves as the light source of the microscope. However, when light in another wavelength region is used for the microscope, a light source for emitting the light is separately prepared, and a beam splitter is provided as the light source 116 in FIG. It can be built in.

  The components of the active and passive optical systems and their operations are considered to be understood as described above. Here, a mark using the microscope of the present invention equipped with these active and passive AF optical systems is used. A detection method will be described.

  First, the positional relationship between the wafer surface and the slit image is measured by the active AF optical system, that is, the position of the surface opposite to the mark forming surface is measured as described. FIG. 4 shows a mark detection procedure after the positional relationship (focus state) between the wafer surface 101 (the back surface of the mark forming surface) and the slit image is measured. Assume that the slit image P is formed at a position separated by a distance d from the wafer surface 101 of the wafer having a thickness t as shown in FIG. In order to focus the mark 412 formed on the opposite surface of the wafer more accurately by the passive AF optical system, the wafer is t + d by a stage mechanism (not shown) according to information on the focus state output from the control unit 182. The mark 412 is moved relative to the microscope (in the state indicated by the broken line in the figure) so as to enter the operating range of the passive AF optical system. In this state, the passive AF optical system is operated to observe the mark in a more focused state. With this passive AF, a clearer mark image can be obtained, and the mark position can be measured with a predetermined position accuracy. In this case, if the relative movement amount t + d is small, the operation range of the passive AF system may be entered as it is. Further, in FIG. 4A, the active AF optical system observation surface (the surface conjugate with the detector surface) and the passive AF optical system observation surface are designed to be the same surface. The observation object surface of the system AF optical system and the observation object surface of the active AF optical system may not be the same surface.

  As shown in FIG. 4B, if the position of the observation surface 463 of the passive AF optical system is shifted so as to be away from the optical path length by the thickness t of the wafer (the focus is formed on the mark formation surface of the wafer by the microscope). When the wafer surface 101 coincides with the observation surface 461 of the active AF optical system, the imaging light beam 420 of the passive AF optical system when matched is indicated by a broken line in FIG. There is no need to move the Even if they do not match, the amount of movement of the wafer for focusing is reduced. By moving the wafer using this measured value, it becomes possible to detect the mark quickly and with high accuracy.

  Next, a manufacturing method for manufacturing a three-dimensional semiconductor using the microscope of the present invention will be described. Please refer to FIG. FIG. 6 is a flowchart showing a manufacturing method of a stacked three-dimensional semiconductor device to which the present invention is applied. The manufacturing method includes steps S1, S2, S3, S4, and S5. Each process will be briefly described.

S1: Wafer preparation step of preparing a predetermined number of wafers on which a plurality of semiconductor devices are formed /
Reference is made to FIG. A wafer 511 on which circuit elements 513 are formed by reducing and projecting a circuit pattern on a mask on a resist-coated wafer using a normal semiconductor exposure apparatus, developing the resist, and performing etching or thermal diffusion treatment of impurities. Get.

S2: Alignment process for measuring positional relationship between wafers to be stacked /
As shown in FIG. 7B, the wafer 211 is held in the holder 210. A plurality of alignment marks 222 are formed on the wafer 211, and a plurality of reference marks 231 are formed on the holder 210. FIG. 8 shows a method for aligning two wafers (one wafer is a wafer stack on which the wafers are already stacked).

  As shown in FIG. 8A, the microscope 230 is moved relative to the alignment mark 222 on the first wafer 211 and the reference mark 231 on the first holder 210 to observe the individual marks. The positional relationship of the alignment mark 222 with respect to is grasped. Since there are a plurality of reference marks and alignment marks (for example, two reference marks and 10 alignment marks), the position of the wafer 211 relative to the holder 210 is determined. Next, the positional relationship between the wafer stack (hereinafter also referred to as a second wafer) 213 and the second holder 212 is examined.

  As shown in FIG. 8B, the microscope 230 is moved relative to the alignment mark 224 of the second wafer 213 and the reference mark 233 on the second holder 212 to observe individual marks. The positional relationship of the alignment mark 224 is grasped. Since there are a plurality of reference marks and alignment marks (for example, two reference marks and 10 alignment marks), the position of the second wafer 213 with respect to the holder 212 is determined. Here, the method for detecting the alignment mark of the second wafer 213 will be described in more detail. As shown in FIG. 9, the bonded second wafer 213 is held by the wafer holder 212 in a state where the alignment mark 412 of the outermost wafer 801 is formed on the bonded surface. The electrodes on the wafer are bonded to each other. Reference numeral 890 indicates this state. When performing a focusing operation (using an AF mechanism) to obtain a sharp image (focused image) of the alignment mark 412, there are the following problems. When the alignment mark 224 passing through the wafer is detected using an active AF optical system using infrared light, the alignment mark is formed in multiple stages because the second wafer is a laminate, and the operating range is Whether the target mark is observed in a wide active optical system is unclear. Needless to say, an active optical system using visible light cannot be observed through the wafer. Conversely, when focusing is performed using a passive optical system, the operating range is narrowed, and the AF function may not operate correctly. Therefore, the microscope of the present invention is used.

  First, as shown in FIG. 9A, the outer surface 803 of the outermost wafer is aligned by the active AF optical system 110 of the microscope of the present invention. Next, as shown in FIG. 9B, the alignment mark 412 is detected by the passive AF optical system 120 having a narrow operation range as the AF function but high position detection accuracy. For this purpose, the height (position on the optical axis) of the wafer stack is adjusted in consideration of the thickness of the wafer 801 so that the alignment mark 412 falls within the operating range of the passive AF optical system 120. Thereby, the mark formed on the bonding surface of the wafer is observed in a preferable focus state, and the position thereof is grasped with a predetermined accuracy.

  Returning to FIG. 8, the description of the manufacturing method of the stacked semiconductor device will be continued. Referring to FIG. 8C, the wafer 211 whose positional relationship is determined with respect to the holder 210 as described above and the second wafer 213 whose positional relationship is also determined with respect to the holder 212 are faced and held close to each other. The positional relationship between the reference marks 231 and 233 on the holder is observed with the microscope 230. When the positional relationship between the two reference marks is determined by observation, the positional relationship between the two wafers is determined. If the positional relationship between the two wafers is not a predetermined relationship, the position of one wafer is adjusted by a positioning device (not shown) so that the predetermined positional relationship is obtained. Reference numeral 910 denotes a window provided in the holder for mark observation.

S3: Wafer superimposition process for superimposing wafers whose positional relationship has been measured /
When the alignment of the two adjacent wafers is completed, the two wafers are superimposed as shown in FIG. 7C by a wafer vertical movement mechanism (not shown). After the contact, in order to maintain the superimposed positional relationship, the holders are temporarily fixed (for example, a clamp mechanism) or temporarily fixed by an adhesive having a weak bonding force (not shown). The temporarily fixed holder and wafer stack 561 are transported to the next step by a robot arm (not shown).

S4: Electrode joining process for joining the connection electrodes on the superimposed wafers /
Reference is made to FIG. The aligned and temporarily fixed wafer stack 561 is mounted with a pressurizing / heating device. The parallelism adjustment of the upper pressurizer 551, the lower pressurizer 553, and the wafer laminate 561 is performed. When the adjustment is completed, the wafer laminate 561 is pressurized by the two pressurizers 551 and 553. At the same time, heating is performed by the heaters 541 and 543 incorporated in the holder according to the sequence determined. By applying a predetermined pressure for a predetermined time, the electrodes (metal bump and pad, metal bump and metal bump) on the wafer are bonded. At this time, in some cases, the resin may be sealed between the wafers and heated.
This alignment process, wafer overlay process, and electrode bonding process are repeated as many times as the number of wafers to be laminated. In some cases, a sealing resin may be encapsulated between the laminated wafers or a process of thinning the laminated wafers by grinding, polishing, or etching after lamination bonding.

S5: A dicing process for separating individual semiconductor devices from a predetermined number of stacked wafers. Wafers stacked and bonded at the wafer level are cut according to a dicing line and separated as chips. For example, cutting is performed according to the broken line in FIG. In general, a dicing saw method in which cutting is performed using a dicing blade, a method in which the wafer surface is melted and split by a laser beam, and a method in which a cutting line is drawn by a diamond cutter are used. However, a dicing saw method is preferable as a method for separating the wafer stack into chips.

  From the above description, it is considered that the microscope, the mark detection method, and the manufacturing method of the stacked semiconductor device of the present invention have been understood.

  Increasing the density and driving speed of semiconductor devices is an inevitable demand in the industry, and the use of the present invention for that purpose is therefore inevitable in the industry.

It is a figure which shows the whole image of the microscope of this invention. The positional relationship between the slit image and the wafer surface and the relationship between the reflected light beams are shown. It is a figure which shows the relationship between the incident light beam to the sensor in an active system AF optical system, and the blur of an image. It is a figure which shows operation after active system AF operation | movement. It is a figure which shows the relationship between image blurring and detector output in passive AF. The manufacturing process of the laminated semiconductor device to which the microscope of the present invention is applied will be described. It is illustration of the manufacturing method of the laminated semiconductor device of this invention. The alignment process at the time of wafer lamination | stacking is shown. A method for detecting alignment marks of stacked wafers will be described.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Observation optical system 101 ... Wafer surface 112 ... Objective lens 114 ... AF eyepiece lens 116 ... Illumination system 121 ... Light source 122 ... Slit 124 ... Condenser lens 126 ... Imaging lenses 126, 128 ... First two-dimensional sensor 129 ... Second two-dimensional sensor 130 ... Light flux limiting masks 162, 142 ... Beam splitter 149 ... Focus detection Beam splitter 182 ... Control unit 190 ... Eyepiece system 190
210, 212 ... Wafer holder 213 ... Wafer stack 222, 224 ... Alignment mark 230 ... Microscope 231, 233 ... Reference mark 412 ... Alignment mark

Claims (3)

A microscope having an autofocus function,
An active AF optical system and a passive AF optical system sharing an objective lens;
A control unit that outputs focus state information;
Have
A microscope characterized in that a light source of the passive AF optical system is an infrared light source. A mark detection method for detecting an alignment mark formed on a wafer surface from the back surface,
A mark detection method, comprising: detecting an alignment mark using the microscope according to claim 1. A method of manufacturing a three-dimensional semiconductor device by laminating wafers on which a plurality of semiconductor devices are formed,
Preparing a predetermined number of wafers on which a plurality of semiconductor devices are formed, a wafer preparation step,
An alignment process for measuring the positional relationship between wafers to be laminated,
A wafer superposition process, which superimposes wafers whose positional relationship has been measured,
An electrode joining process for joining the connection electrodes on the stacked wafers;
A dicing process for separating individual semiconductor devices from a predetermined number of stacked wafers;
The method for manufacturing a stacked three-dimensional semiconductor device, wherein the microscope according to claim 1 is used in the alignment step.

Images