JP2013145171A - Appearance inspection device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the accuracy of determining the quality of an object to be inspected, such as a semiconductor chip, and shorten a determination time.SOLUTION: Reference image data is created by using a piece of image (destination image) data among image data obtained by actually imaging a plurality of objects (semiconductor chip) to be inspected and image data around the piece of image data to perform averaging processing of the plurality of pieces of image data (actually imaged image data), and the quality of the objects to be inspected is determined by using the reference image data to perform pattern matching processing. Since the influence of an appearance variation can be reduced by such determination processing regardless of the presence of the appearance variation in the plurality of objects to be inspected, the accuracy of quality determination can be improved. In addition, a quality determination time can be shortened because only a small amount of reference image data used for the pattern matching processing is required.

Description

  The present invention relates to an appearance inspection apparatus that images the appearance of an inspection object such as a semiconductor chip and determines the quality of the inspection object from the captured image.

  As an inspection device for inspecting the quality of a semiconductor chip such as an LED chip, an image of a surface to be inspected of the semiconductor chip is taken, and the quality of the semiconductor chip is compared by comparing the image data of the taken semiconductor chip with reference image data. There is an appearance inspection apparatus that determines the above (for example, see Patent Documents 1 and 2). In such an inspection apparatus, the number of good semiconductor chips and the number of defective products are also counted.

  As the quality determination processing of such an appearance inspection apparatus, for example, image data of one semiconductor chip (non-defective image data) is selected from image data obtained by imaging the appearance of a plurality of semiconductor chips, and the selected image is selected. There is a process in which the reference image data is used as the reference image data, and the reference image data is compared with image data of other semiconductor chips (pattern matching process) to determine a non-defective product and a defective product.

  In such a quality determination process, for example, as shown in FIG. 8, when image data A is used as reference image data, image data B and image data C (image data B and C are both non-defective image data) are obtained. Since it is different from the reference image data A, it is determined as a defective product, and when counting the number of non-defective chips, the semiconductor chips of the image data B and C are not counted. There are the following two methods for solving this problem (method for handling the semiconductor chips of the image data B and C as non-defective products).

  (I) The criteria for comparing the image data of the semiconductor chip that is the inspection object and the reference image data are relaxed.

  (Ii) The image data B and C shown in FIG. 8 are also registered (stored) as the reference image data, and if a defect is determined by comparing the inspection target image data with the image data A, the inspection target image data The comparison with the image data B and the comparison with the image data C are performed.

JP 2002-340812 A JP-A-6-337922

  By the way, in the above-described method (i), when there are appearance variations in a plurality of inspection objects, the judgment criteria for pass / fail judgment are relaxed in order to deal with the appearance variations (determination corresponding to the maximum value of appearance variations). The inspection object (semiconductor chip) that should be determined as a defective product may be determined as a non-defective product.

  Further, in the method (ii) described above, it is necessary to register the number of reference image data corresponding to the variation in the appearance of the inspection target, and much labor is required for the registration work. In addition, since the number of comparisons between the image data of the inspection object and the reference image data inevitably increases, the time required for the pass / fail determination becomes longer.

  The present invention has been made in consideration of such circumstances, and in determining the quality of the appearance of an inspection object such as a semiconductor chip, even if there are variations in the appearance of a plurality of inspection objects, the quality determination is performed with high accuracy. An object of the present invention is to provide an appearance inspection apparatus that can perform the determination and shorten the determination time.

  The present invention is premised on an appearance inspection apparatus that determines the quality of an appearance of an inspection object such as a semiconductor chip. In such an appearance inspection apparatus, reference image data is obtained using image data obtained by imaging a plurality of inspection objects. A reference image data creation unit for creating a non-defective product or a defective product by comparing the reference image data created by the reference image data creation unit with each image data of the plurality of inspection objects (pattern matching processing) And a pass / fail determination unit. Then, the reference image data creation unit includes one piece of image data designated by the input unit and image data around the designated image data (designation) among the image data obtained by capturing the plurality of inspection objects. The technical feature is that the reference image data is created by averaging the plurality of image data using peripheral image data close to the image.

  According to the present invention, by using one image (designated image) data of image data obtained by actually capturing a plurality of inspection objects and surrounding image data, the plurality of image data (actually captured images) are used. Since the reference image data is created by averaging the image data), even if there are variations in the appearance of multiple inspection objects (for example, variations in the appearance of semiconductor chips in the wafer), the effect of the variation is reduced. can do. As a result, it is not necessary to loosen the judgment criteria (judgment criteria for the degree of coincidence) in the pattern matching process, and the accuracy of the quality judgment of the inspection object can be increased. In addition, since the number of reference image data used for the pattern matching process is small (for example, only one reference image data is required), the number of comparisons between the image data of the inspection object and the reference image data can be reduced. The determination time can be shortened.

  In the present invention, the image processing apparatus includes an imaging unit (imaging device) that captures images of the inspection target surfaces of a plurality of inspection objects, creates the reference image data using image data captured by the imaging unit, and inspects the inspection object You may make it perform the quality determination of. Further, image data captured by a device other than the appearance inspection apparatus of the present invention (image data obtained by capturing the appearance of a plurality of inspection objects) is loaded into the appearance inspection apparatus of the present invention, and the captured image data is captured. The reference image data may be used to make a pass / fail judgment for the inspection object.

  In the present invention, the number of non-defective products may be counted based on the determination result of the good / bad determination unit. Further, the number of defective products may be counted based on the determination result of the quality determination unit. Further, the number of non-defective products and the number of defective products may be counted based on the determination result of the pass / fail determination unit.

  In the present invention, a reference image data storage unit that stores the reference image data created by the reference data creation unit in association with identification information (lot number, product number, etc.) of the inspection object may be provided. In this way, when the reference image data is stored in association with the identification information of the inspection object, when the quality of the inspection object of the same model number (same product type) as that stored is determined, the above reference image Since the pattern matching process can be performed using the reference image data stored in the data storage unit, the reference image data creation process can be omitted.

  Thereby, for example, when inspecting the inspection object of the same model number continuously (for example, inspecting continuously for each wafer unit), or the inspection object of the reference image data stored in the reference image data storage unit It is possible to reduce the inspection time when, for example, the inspection object of the same model number is inspected after inspecting the inspection object of a different model number.

  According to the present invention, it is possible to accurately determine the quality of the appearance of an inspection object such as a semiconductor chip. In addition, it is possible to shorten the determination time of the appearance quality.

It is a schematic block diagram which shows an example of the external appearance inspection apparatus of this invention. It is a figure which shows typically an example of the test object frame which test | inspects with the external appearance inspection apparatus of FIG. It is a flowchart which shows the content of the external appearance inspection process which the arithmetic processing apparatus of FIG. 1 performs. It is a flowchart which shows an example of a reference | standard image data creation process. It is a figure which shows typically an example of the averaging process of image data. It is a flowchart which shows an example of a quality determination process. It is a figure which shows the example of a display of the external appearance image of a semiconductor chip. It is a figure explaining an example of the quality determination method of the conventional semiconductor chip.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

-Inspection object-
First, in the present embodiment, an expanded semiconductor chip is used as an inspection target.

  Specifically, a dicing tape (stretched film made of synthetic resin) T is attached to the back surface of a wafer on which a large number of semiconductor chips are arranged in a matrix, and the dicing tape is held on the ring frame R in this state. After the wafer is cut (diced) and divided into individual semiconductor chips, the dicing tape is expanded (expanded) so that a gap is formed between the semiconductor chips, and the inspection object frame W (see FIG. 2) A plurality of semiconductor chips S... S arranged in a matrix on the inspection object frame W are used as inspection objects. The plurality of semiconductor chips S..S aligned in the inspection target frame W have some variation in appearance (for example, positions and sizes of patterns such as wiring and electrodes).

  Here, examples of the semiconductor chip S include an LED chip, a power device, a high-frequency device, a photodiode, a laser diode, and a Hall element. Examples of the ring frame R (see FIG. 2) for fixing and holding the dicing tape T include a metal ring frame and a grip ring frame.

-Visual inspection device-
Next, an example of the appearance inspection apparatus 100 will be described with reference to FIG.

  The appearance inspection apparatus 100 in this example includes an imaging device 1, an illumination device 2, an arithmetic processing device 3, an input device 4, a display device 5, a stage 6 on which an inspection target frame W is placed, and the like.

  The imaging device 1 is, for example, a CCD camera, and captures all appearance images (images of a surface to be inspected) of a plurality of semiconductor chips S ·· S (see FIG. 2) placed on the stage 6. Image data captured by the imaging device 1 is taken into the arithmetic processing device 3.

  As for the imaging device 1, for example, a CCD array (image line sensor) in which CCDs are arranged in a one-dimensional manner is scanned in one direction (row direction or column direction) in the arrangement direction of the semiconductor chips S..S. Therefore, the imaging device may be an imaging device that captures the entire appearance image of the semiconductor chip S..S.

  The illuminating device 2 is an illuminating device using, for example, an LED as a light source, and illuminates the entire inspection target frame W placed on the stage 6, that is, all the inspected surfaces of the plurality of semiconductor chips S ·· S.

  The arithmetic processing device 3 is a personal computer, for example, and is connected to an input device 4 and a display device 5 as peripheral devices. The arithmetic processing device 3 includes a control unit 31, an image data capturing unit 32, an image data storage unit 33, a reference image data creation unit 34, a pass / fail determination unit 35, a reference image data storage unit 36, and the like.

  The control unit 31 comprehensively controls various processes for appearance inspection. The image data capturing unit 32, the image data storage unit 33, the reference image data creation unit 34, the pass / fail determination unit 35, and the reference The operation of each functional unit such as the image data storage unit 36 is monitored and controlled.

  The image data capturing unit 32 outputs a command for capturing an appearance image of the semiconductor chip S... S to the imaging device 1, captures the appearance image captured by the imaging device 1 as image data, and records it in the image data storage unit 33. To do.

  The reference image data creation unit 34 reads the image data stored in the image data storage unit 33 and creates reference image data. This reference image data creation process will be described later.

  The pass / fail judgment unit 35 compares the reference image data created by the reference image data creation unit 34 with the image data stored in the image data storage unit 33 (image data for each semiconductor chip S). Then, it is determined whether each semiconductor chip S is a good chip or a defective chip. The pass / fail judgment process in the pass / fail judgment unit 35 will also be described later. The pass / fail determination unit 35 also has a function of counting the number of non-defective chips and the number of defective chips based on the pass / fail determination result of each semiconductor chip S.

  The reference image data storage unit 36 stores the reference image data created by the reference image data creation unit 34 in association with the identification information (product number, lot number, etc.) of the semiconductor chip S (wafer).

  The input device 4 is, for example, a keyboard and a mouse, and various information can be input to the arithmetic processing device 3 by an operator operating the input device 4. In this example, for example, the mouse is displayed on the display screen of the display device 5 with image data (image data obtained by imaging the entire semiconductor chip S..S) and the pointer P (for example, see FIG. 7). By clicking on the mouse with the pointer P set to one of the image data of the plurality of semiconductor chips S..S displayed on the display screen of the display device 5 by the operation of Image data can be specified. Note that the input device 4 may be a tablet, a light pen, a touch panel, or the like.

  The display device 5 is a display device using, for example, a CRT, a liquid crystal display (LCD), a plasma display panel (PDP), and the like, and displays various information such as image data output from the arithmetic processing device 3.

<Appearance inspection process>
Next, an example of the appearance inspection process executed by the arithmetic processing device 3 (control unit 31) in FIG. 1 will be described with reference to the flowchart in FIG.

  First, when processing is started, in step ST <b> 101, an appearance image (an entire image of the semiconductor chips S... S) captured by the imaging device 1 is captured as image data and recorded in the image data storage unit 33.

  Next, in step ST102, a subroutine for reference image data creation processing is executed. This subroutine will be described with reference to FIG.

  When the subroutine of FIG. 4 is started, first, in step ST121, the image data stored in the image data storage unit 33 (the entire image data of the semiconductor chips S..S) is read, and the image data is displayed on the display device 5. On the display screen.

  Next, in step ST122, it is determined whether or not an image displayed on the display screen of the display device 5, that is, an image of any one of the semiconductor chips S. If the determination is negative and the determination result is negative (NO), the process enters a standby state.

  Here, the image data designation process will be described. First, the operator looks at the entire image of the semiconductor chips S... S displayed on the display screen of the display device 5 and is close to a good appearance image among the images of the semiconductor chips S. An image (one image) is extracted. Next, an operation of clicking the mouse with the pointer P on the one image is performed by a mouse operation. When such an image designation is performed by the operator, the determination result in step ST122 is affirmative (YES), and the process proceeds to step ST123. The image data designation process may be performed by other methods.

  In step ST123, image data (also referred to as designated image data) of a designated image (image of one semiconductor chip S) designated by the operator is read from the image data storage unit 33.

  Next, in step ST124, a plurality of image data close to the designated image, which is the peripheral image data of the designated image (see FIG. 5) designated by the operator, is read (the arithmetic processing unit 3 automatically Extract the image and read the image data). Specifically, for example, as shown in FIG. 5, if the image data of the designated image is “D0”, peripheral image data D1 to D8 are read from the image data storage unit 33. As processing for reading out the peripheral image data, for example, after performing binarization processing on each of the image data of the specified image data and the plurality of peripheral image data, image data D1 to D8 close to the specified image data D0 are patterned. The method of extracting by a matching process etc. can be mentioned.

  In step ST125, the specified image data and a plurality of peripheral image data are averaged. Specifically, binarization processing is performed for each of a plurality of peripheral image data read in step ST124. Next, the peripheral image data D1 to D8 after the binarization processing and the specified image data D0 are averaged to generate reference image data shown in FIG. For example, regarding the peripheral image data D1 to D8 and the designated image data D0, the average of the pixel values of the corresponding pixels is set as the reference image data. The reference image data created in this way is stored in the reference image data storage unit 36 in association with the identification information (product number, lot number, etc.) of the semiconductor chip S (wafer).

  As for the above-described peripheral image data, it may be automatically selected (extracted) that is a non-defective product determination based on a normal determination criterion. By using the averaged processing of the peripheral image data D1 to D8 and the specified image data D0 extracted in this way as reference image data, an allowable range for non-defective product spread is widened. An image (appearance image of the semiconductor chip S) that may be determined as a defective product can also be determined as a positive determination (non-defective product determination).

  The reference image data creation process is thus completed, and thereafter, the subroutine of FIG. 4 is terminated and the process returns to the main routine of FIG.

  Returning to the main routine of FIG. 3, in step ST103, a pass / fail judgment process is executed. This subroutine will be described with reference to FIG.

  First, when the process is started, in step ST131, n (positive integer: n = 1, 2, 3,...) Among the plurality of semiconductor chips S..S arranged in the inspection target frame W shown in FIG. The image data of the) th semiconductor chip S is read from the image data storage unit 33, and binarization processing is performed on the read image data. The image data that has been subjected to the binarization processing (hereinafter also referred to as inspection image data) is temporarily stored in a buffer or the like in the arithmetic processing unit 3.

  Next, in step ST132, reference image data is read from the reference image data storage unit 36. In step ST133, pattern matching processing between the reference image data read in step ST132 and the inspection image data is performed to obtain a degree of coincidence. Examples of the evaluation numerical value of the degree of coincidence with the reference image data include a correlation coefficient, the sum of absolute values of differences of corresponding pixels, and the sum of squares of differences of corresponding pixels.

  Next, in step ST134, it is determined whether or not the degree of matching of the pattern matching obtained in step ST133 is greater than or equal to a determination reference value. If the determination result is affirmative (YES), the pattern matching process is performed. It is determined that the semiconductor chip S of the inspection image data that has been subjected to is non-defective (step ST135). On the other hand, when the determination result of step ST134 is negative (NO), it is determined that the semiconductor chip S of the inspection image data subjected to the pattern matching process is a defective product (step ST136). Note that the determination reference value can be set and changed by an input operation of an operator or the like, for example.

  With the above processing, the non-defective / defective product determination processing for one semiconductor chip S is completed. After the quality determination for one semiconductor chip S is completed, the process returns to step ST131 (when the determination result of step ST137 is negative determination (NO)), and the image data of the next semiconductor chip S is subjected to the above steps ST131 to ST136. The process (good / bad determination process) is executed. In this way, the pass / fail judgment is sequentially repeated for each of the plurality of semiconductor chips S... S arranged in the inspection target frame W shown in FIG.

  Then, when the pass / fail determination for all the semiconductor chips S..S is completed (when the determination result of step ST137 is affirmative (YES)), the subroutine of FIG. 6 and the main routine of FIG. To do.

<About processing based on pass / fail judgment results>
The arithmetic processing unit 3 counts the number of non-defective chips and the number of defective chips based on the determination result of the above-described pass / fail determination process. The arithmetic processing unit 3 recognizes, for example, the position information of non-defective chips (chip position information in the wafer) and the position information of defective chips, and shows, for example, the display example of FIG. 7 based on the chip position information. As described above, a process of attaching a bad mark to the defective semiconductor chip S is executed. Then, the inspection results such as the number of non-defective chips, the number of defective chips, the position information of non-defective chips, and the position information of defective chips are sent to an external device, for example, via the output interface of the arithmetic processing unit 3. , And output to an apparatus (map sorter, die bonder, etc.) in a later process.

<Effect>
As described above, according to the present embodiment, one image (designated image) data among the image data obtained by actually capturing the plurality of semiconductor chips S..S as the inspection object, and the surrounding image. Since the reference image data is created by averaging the plurality of image data (image data actually captured) using the data, the appearance of the plurality of semiconductor chips S ·· S arranged in the inspection target frame W Even if there is a variation (the appearance variation of the semiconductor chips S..S in the wafer), the influence of the variation can be reduced. As a result, it is not necessary to loosen the judgment criteria (judgment criteria for the degree of coincidence) in the pattern matching process, and the accuracy of the quality judgment of the inspection object can be increased. In addition, since the pattern matching process is performed using one reference image data, the number of comparisons between the image data of the inspection object and the reference image data can be reduced, and the determination time can be shortened.

  Further, in the present embodiment, the reference image data created by the reference image data creation unit 34 is stored in the reference image data storage unit 36 and the identification information (lot number, product number, etc.) of the semiconductor chip S that is the inspection object. Since the data is stored in association with each other, when the quality determination of the semiconductor chip S having the same model number as that stored is performed, the pattern matching process may be performed using the reference image data stored in the reference image data storage unit 36. It is possible to omit the process of creating the reference image data as described with reference to FIG.

  Thereby, for example, when the semiconductor chip S of the same model number is continuously inspected (for example, continuously inspected for each wafer unit), or the semiconductor chip S of the reference image data stored in the reference image data storage unit 36 is obtained. After inspecting a semiconductor chip having a different model number from that of the test object, it is possible to shorten the inspection time when inspecting the inspection object having the same model number.

  In addition, as a method of reading the reference image data stored in the reference image data storage unit 36, for example, the operator operates the input device 4, and the identification information such as the product number of the semiconductor chip to be inspected is input to the arithmetic processing unit 3. By inputting, a method of reading the reference image data from the reference image data storage unit 36 can be mentioned.

  As another method, for example, a bar code reader is provided in the appearance inspection apparatus 100, and a bar code in which identification information of the semiconductor chip S is recorded is attached to the ring frame R shown in FIG. Then, in the process of transporting the semiconductor chip S to be inspected to the stage 6 in FIG. 1, the barcode attached to the ring frame R or the like is read by the barcode reader, and the identification information is input to the arithmetic processing unit 3. Thus, a method of reading the reference image data from the reference image data storage unit 36 can be mentioned.

-Other embodiments-
In the above embodiment, the quality of the semiconductor chip S is determined using the image data captured by the imaging device 1 included in the appearance inspection apparatus 100. However, the present invention is not limited to this, and the appearance inspection is performed. Image data (image data obtained by imaging a plurality of semiconductor chips) captured by a device other than the device 100 is taken into the arithmetic processing unit 3 from the input device 4 or the like, and the reference image data is used using the fetched image data. May be made to determine whether the appearance of the semiconductor chip is good or bad.

  Further, in the above embodiment, the appearance inspection apparatus 100 includes the imaging device 1, but without such an imaging device 1, the arithmetic processing device 3, the input device 4, and the image display shown in FIG. The appearance inspection device may be configured by the device 5 or the like.

  In the above embodiment, one reference image data is used. However, if a plurality of reference image data is created, the accuracy of the semiconductor chip quality determination can be further increased.

  In the above embodiment, the frame type as shown in FIG. 2 is an inspection target, but the appearance inspection apparatus of the present invention is not limited to this, and a plurality of semiconductor chips are placed on a sheet (film). The present invention can also be applied to the arranged sheet type.

  In the above embodiment, the semiconductor chip is used as the inspection object. However, the present invention is not limited to this, and the appearance inspection apparatus of the present invention can be applied to the appearance inspection of various other inspection objects.

  INDUSTRIAL APPLICABILITY The present invention can be effectively used in an appearance inspection apparatus that images the appearance of an inspection object such as a semiconductor chip and determines the quality of the inspection object from the captured image.

DESCRIPTION OF SYMBOLS 100 Appearance inspection apparatus 1 Imaging apparatus 2 Illumination apparatus 3 Arithmetic processing apparatus 31 Control part 32 Image data acquisition part 33 Image data storage part 34 Reference | standard image data creation part 35 Pass / fail judgment part 36 Reference | standard image data storage part 4 Input device (input part)
5 Display device 6 Stage S / S Semiconductor chip (inspection object)

Claims (5)

An appearance inspection apparatus for determining the quality of an inspection object such as a semiconductor chip,
A reference image data creation unit that creates reference image data using image data obtained by imaging a plurality of inspection objects, reference image data created by the reference image data creation unit, and each image data of the plurality of inspection objects And a pass / fail judgment unit for judging whether the product is a non-defective product or a defective product, and the reference image data creation unit is one of the image data obtained by imaging the plurality of inspection objects, specified by the input unit An appearance inspection apparatus, wherein image data and peripheral image data of the designated image data are used to generate reference image data by averaging the plurality of image data. The appearance inspection apparatus according to claim 1,
An appearance inspection apparatus, comprising: an imaging unit that captures images of a surface to be inspected of the plurality of inspection objects, the reference image data creation unit, and the quality determination unit. The appearance inspection apparatus according to claim 1 or 2,
An appearance inspection apparatus that counts the number of non-defective products based on the determination result of the pass / fail determination unit. In the appearance inspection apparatus according to any one of claims 1 to 3,
An appearance inspection apparatus that counts the number of defective products based on a determination result of the pass / fail determination unit. In the appearance inspection apparatus according to any one of claims 1 to 4,
An appearance inspection apparatus comprising a reference image data storage unit that stores reference image data created by the reference data creation unit in association with identification information of an inspection object.

Images