JPS62103548A - Defect inspecting device for lead frame - Google Patents

Abstract

PURPOSE:To enable accurate automatic detection of a defect of a lead frame on a single process, by arranging first and second lighting means, an image pickup means, a signal processing means and the like. CONSTITUTION:A first lighting means 40 irradiates a surface to be inspected about surface condition of a lead frame 29 with light of the optical axis vertical thereto and an image by a reflected light from a frame 29 is formed on an image receiving surface of a linear image sensor 9 with a photographing lens 10. At the same time, a second lighting means 50 irradiates the back of the frame 29 with light vertical thereto while aligning the optical axis of the means 40 to form an image crated by a transmission light of the frame 29 overlapping an image created by the reflected light on a light receiving surface of the sensor 9. The sensor 9 reads an optical image on the image receiving surface and outputs an electrical signal proportional to the quantity of light to be fed to signal processing section 80. Then, an output signal of the sensor 9 is binary-coded by various thresholds by binary coding circuits 16 and 17 to obtain a binary-coded image signal of the image by the transmission light through the frame 29 and a binary-corded image signal separated from a rough-surfaced by etching. Based on the signal, a judgement processing section 19 detects a defect.

Description

Detailed Description of the Invention (Field of Industrial Application) The present invention relates to a defect inspection device for lead frames, which are components of semiconductor devices such as ICs and LSIs.
More specifically, the present invention relates to an apparatus for automatically inspecting defects such as holes and protrusions between leads and bits and chips on the surface that occur in a lead frame manufactured by an etching method.

(Background of the invention) A lead frame is made by etching to a thickness of 100 mm to 100 mm.
A 300 μm metal plate is processed into the shape shown in the second figure, and the tips of the island 1 and the inner lead 2 are partially plated with metal such as gold or @. Attach the chip (die bonding), and then coat the inner lead tip and the electrode of the semiconductor chip with gold.

Semiconductor devices are manufactured by connecting them with thin metal wires such as aluminum (wire bonding).

In Figure 2, 3 is the dam and 4 is the outer reed.

5 represents an outer frame, and 6 represents a partial plating area.

The above is an explanation of lead frames, but with lead frames made using the etching method, during the manufacturing process, scratches and attached foreign matter on the original plate for pattern formation may be transferred, and the metal plate that is the base material of the lead frame may be coated. Foreign matter, bubbles, pinholes, etc. in the resist may cause
Defects as shown in 26 may occur, and defective products with such defects may be mixed into the product. Since the use of such defective products may cause defects in the completed semiconductor device or decrease in reliability, it is essential to inspect the lead frame after manufacturing it and eliminate defective products.

(Prior art and its problems) Conventionally, such tests have been carried out with the naked eye or using a microscope, but it requires a lot of manpower to test a large number of products, and it is difficult to perform sensory tests. Therefore, there were problems with test accuracy and reliability.

In order to solve such problems, for example, protrusions 22 and shorts 21 that occur on the leads 30 as shown in FIG. Regarding the method of inspecting the concave defect 26, the defect is detected by using an ITV and comparing the video signal obtained by photographing the lead frame with transmitted light illumination with the signal obtained by similarly photographing the reference pattern or the adjacent pattern. For defects that occur on the surface, such as pits 25 and chips 23.24, methods have been proposed in which defects are detected by performing similar processing using reflected light, but each method is different. Although the objective can be achieved, different illumination methods must be used for each inspection item, and there are also problems such as there are cases where something that does not cause a defect, such as a minute chip 23, is detected and judged as a defect. A) It was difficult to put it into practical use.

(Object of the Invention) The object of the present invention is to solve the above-mentioned problems and provide a lead frame defect inspection device that can accurately and automatically detect various defects in lead frames in one process. be.

(Summary of the Invention) To achieve this object, the present invention irradiates light perpendicularly to the surface of the lead frame whose surface condition is to be inspected.
a first illumination means configured to obtain an image by the specularly reflected light, and irradiating light perpendicularly to the opposite surface of the lead frame so as to coincide with the optical axis of the first illumination means;
A second illumination means configured to obtain an image by the transmitted light, and a position where it can receive specularly reflected light and transmitted light of the lead frame that is simultaneously irradiated with light by the first and second illumination means. It is characterized by comprising an installed imaging means and a signal processing section that processes the image signal obtained from the imaging means and detects defects in the IJ-deframe.

(Example) The present invention will be described in detail below based on Examples.

First, FIG. 1 shows an example of the configuration of a defect inspection apparatus according to the present invention. Lens 11. Half mirror 13
The lead frame 2 is illuminated by the first illumination means 40 consisting of
The lead frame 2 is irradiated with light so that the optical axis is perpendicular to the surface of the lead frame 9 whose surface condition is to be inspected.
At the same time, the light source 48 forms an image by the reflected light from the light source 48
A second illumination means 50 consisting of a lens 41 irradiates light perpendicularly to the back surface of the lead frame 29 and coincides with the optical axis of the first illumination means 40, and an image of the lead frame 29 by the transmitted light is formed. The linear image sensor 9
The image is superimposed on the image of the reflected light on the image receiving surface. The linear image sensor 9 reads an optical image on its image receiving surface and outputs an electric signal proportional to the light l, and the output signal of the sensor 9 is sent to a signal processing section 80.

The signal processing section 80 basically includes two binarization circuits 16.
17 and a determination processing section 19. First, the output signal of the sensor 9 is binarized by the binarization circuits 16 and 17 using different threshold values, and the lead frame 29
Binarized image signal of image by transmitted light and bit 2 of Fig. 3
A binary image signal is obtained by separating only the roughened portion due to etching, as shown in Fig. 5, and a defect is detected by the determination processing section 19 based on the information of both image signals. 18 is a memory for storing an image signal obtained by imaging a reference pattern or an adjacent pattern for reference when performing defect detection by comparing image signals; 12 is driven by motors 14, 14 and a stage controller 15; The XY stage 20 moves the lead frame 29 along a predetermined scanning path, and 20 is a control unit that controls the entire inspection apparatus.

Next, the operation of detecting defects in a lead frame will be explained using the above embodiment. First, FIG. 4 explains a method of obtaining an image signal of transmitted light and an image signal of only the rough surface portion generated on the surface of the lead frame from the signal of the linear image sensor 9 using the configuration shown in FIG. Image sensor 9
The light source 8 and the light source 480 are arranged so that the level T of the signal generated by the transmitted light on the image receiving surface of the lead frame exceeds the level R of the signal generated by the light reflected from the defect-free portion of the surface of the lead frame.
When the strength is set, a straight line A-A passing through the chip 24 and the bit 25, which are surface defects on the lead 30 shown in FIG.
The signal change when scanning the top is as shown in Figure 4(b), and the image is divided into two by setting the threshold Ts to T)Ts)R.
When converted into a value, it becomes an image signal based on transmitted light, and when converted into a binary value by setting the threshold value Rs such that Rs < R, it becomes a fourth image signal.
Figure jal chipping 24. An image signal is obtained in which only the rough surface portion (defect portion) generated on the surface of the lead 300 such as the bit 25 is extracted. Although the threshold values Ts and Rs are shown as curves in FIG. 4 (bl), if the signal level T and changes within the general photographic field of view are sufficiently small, Ts and Rs can be processed with constant values.

Furthermore, if the illumination means, image pickup means, etc. are configured as shown in FIG. 5, the irradiation light from the light source 8 and the irradiation light from the light source 48 are converted into light having polarization planes perpendicular to each other by the polarizing plates 31 and 51, and the lead frame is 29 at the same time, and the transmitted light and reflected light are guided to two linear image sensors 9 and 49 by an Is-fumirror 53, and each sensor 9,
The incident light of 49 is controlled by polarizing plates 32 and 52 so that each linear image sensor receives an image of only transmitted light and an image of only reflected light, and after binarizing each image signal,
In either case, the dark portion can be detected as a defect such as a chip 24 or a pit 25 on the surface of the lead 30, and an image signal based on transmitted light can also be obtained.

Furthermore, if the illumination means and the imaging means are configured as shown in FIG.
1, the light is converted into light having polarization planes perpendicular to each other, and then irradiated onto the lead frame 29 from both sides by half mirrors 13 and 63, and transmitted light and reflected light seen from each of the linear image sensors 9 and 49 provided on both sides. Since the light intensity ratio of 2 to 3 can be adjusted by the polarizing plate 32.52, the signals of each linear image sensor can be binarized by the method explained in FIG. An image signal is obtained in which only the signal and the rough surface portion (defect portion) generated on the surface of the lead frame 29 are extracted, and both sides of the lead frame can be inspected at the same time.

Next, a method for detecting a defect in the determination processing section 19 based on the image signal of the transmitted light and the image signal of the rough surface portion obtained by the above method will be explained.

First, Figure 3 Nyoto 21. Detection of defects with changes in contour shape as shown in the protrusion 22 and concave defect 26 is performed by detecting the image signal of the binarized transmitted light obtained by the above method in advance by detecting the corresponding portion of the reference pattern or adjacent pattern. is imaged in a similar manner and stored in the memory 18 and compared with a certain image signal, or by assembling a plurality of the imaging means and the illumination means, each imaging means is used to create a lead frame in which a plurality of unit patterns are repeatedly arranged. Compare the image signals of transmitted light separated from the image signals obtained by photographing corresponding parts of different unit patterns, and calculate the size of the discrepant part due to quantization error of the double-sided image signal being compared. If it exceeds the limit, it can be detected as a defect. A method that excludes quantization errors and detects mismatched areas larger than a predetermined size is a method often used in pattern feature extraction using binary image signals.Pixel data within a predetermined size area For example, when the comparison results of the two image signals to be compared are sequentially input to a group of shift registers arranged two-dimensionally as shown in FIG. The data in the area W is shifted along with the input, and an effect equivalent to scanning the entire image while shifting the area of size W one pixel at a time with respect to the input image signal is obtained. If a judgment logic is set for the element, it is possible to detect only the part that matches the logic.For example, for the element Wi, j in this area W, Wt
+, WL2. If you layer W2.1 and W2.2 and detect it as a defect when all of the data indicates a mismatch between the two images, the mismatch will only be the part that includes the size of 2 x 2 pixels. can be detected, and as a result, it is possible to exclude mismatched portions smaller than 2×2 pixels that occur on the pattern contour due to quantization errors. In addition, if the size of the defect to be detected is larger than 2 x 2 pixels, the logic within the area W is set to a larger value such as 3 x 3° or 4 x 4, and when all of the data shows a mismatch, it is detected as a defect. By doing so, it is possible to exclude not only quantization errors but also minute defects that would not result in a lead frame being defective.

However, in defect detection using the comparison method, in order to exclude quantization errors as described above, the detection limit is a defect with a size where the mismatch area is 2 x 2 pixels in the comparison result of image signals, but in reality, this detection limit is For defects that can be detected 100% by this method, it is also necessary to take into account fluctuation factors in the binary image signal and comparison results, such as the resolution of the imaging system, noise contained in the video signal, and the positional relationship between the defect and the pixel. Result, 3×3 to 4×4
This results in a defect with a size equivalent to a pixel. For example, if the size of one pixel is 203m x 20μm, the defect for which the detection rate is 100% is 60μm x 60μm to 808m x 80μm.
Therefore, defects smaller than this size may be overlooked.

Therefore, when protruding defects occur relative to adjacent leads and each has a height of 50 μm as shown in FIG. There may be a possibility that defects are not detected because 100 chips are not detected, but for example, in the inner lead area of a lead frame, parts where the lead spacing is less than a predetermined value are considered defects, so these defects are also detected reliably. There is a need. In order to detect such defects, in the comparative method, it is possible to improve the detection ability by setting the pixels small, but this method has the drawback that the inspection speed decreases when the pixels are made small. On the other hand, if we focus on the width of the pattern corresponding to each lead in the image signal generated by transmitted light and detect the portion whose width is less than the minimum allowable width as a defect, the result will be as shown in Fig. 322. It is possible to reliably detect even highly fatal defects, which are detected in less than 100 chips, which occur relative to adjacent leads, without reducing the inspection speed.

For example, the image signal from the transmitted light is input to a group of shift registers in FIG. 7 (al), and elements as shown in FIG. The data corresponds between the two elements in the circumferential array, such as ``J1. Town and W?, S, W2.2 and Wa, a'', and both of the two elements corresponding to ``1. If it is detected as a defect when the lead spacing corresponds to the diameter of the circumferential array, it is possible to detect as a defect the part that is smaller than the lead spacing depending on the diameter of the circumferential array. According to the detection logic, a defect like the one shown in Fig. 321 cannot be determined to be a defect because a part with a small lead interval does not occur.Therefore, a defect accompanied by a change in the outline shape of the lead frame can be detected by detecting an image signal using transmitted light. By detecting both the parts where there is a mismatch as a result of the comparison and the parts where the lead spacing is less than a predetermined value as defects, all defects such as the short 21, protrusion 22 and concave defects 26 shown in Fig. 3 can be detected efficiently. It can be detected well.

Next, a method for detecting defects occurring on the lead surface as shown in FIG. 3, 24 and 25V will be explained. The above surface defects are caused by pinholes or missing parts of the resist film formed on the surface of the lead frame base material during the manufacturing process using the etching method. A recess is formed by etching, and the surface becomes rough. Therefore, when the lead frame is simultaneously irradiated with light and imaged by the first and second illumination means as shown in the first diagram, only the rough surface becomes dark due to the light scattering effect, so the video signal level will decrease. As shown in Fig. 4 (bl), when binarized using the l value Rs, an image signal is obtained in which only the rough surface is separated from the gaps between the leads and from the smooth lead surface. However, the roughness separated by this method Figure 3 2 on the surface
In addition to the defects that should be detected such as 4.25, there are also minor chips near the contour of the lead frame that are acceptable and do not cause functional problems, as shown in 23, and defects such as 2 due to the taper of the cross-sectional shape of the lead.
Since edges such as 7.28 are included, these parts that do not need to be detected are excluded and chipped 24. Only defects such as bit 25 need to be detected.

In order to detect defects by excluding unnecessary parts from the detected rough surface area, a predetermined image signal is detected from the pattern edge of the transmitted light image signal between the transmitted light image signal and the image signal of the rough surface area whose mutual positional relationship is known. It is sufficient to determine and detect a rough surface area detected at a distance greater than 1 as a defect. For example, the binarized transmitted light image and the image of the rough surface area may be separated by shifting as shown in FIG. 7(a). Input to a group of registers, and in each region W, for the transmitted light image signal, the elements set on the inner circumference of the determination elements shown in FIG. Each element corresponding to the center is set as a determination element, and all of the determination elements set in the area W of the transmitted light image signal have corresponding data on the lead,
Furthermore, if a defect is detected when the element set for the image signal of the rough surface becomes data corresponding to the rough surface, then in the rough surface near the edge of the lead, any of the circumferentially arrayed elements is detected between the leads. Since it becomes corresponding data, it does not conform to the above logic and can be excluded. Moreover, it is a binary image signal.
It is possible to detect even the minute defects that form pixels.
In particular, it has the advantage of being able to save money without reducing the pixel size when detecting defects in areas with strict defect standards, such as the wire bonding portion at the tip of the inner lead.

The above method shows a method for detecting defects that are the same size or smaller than chips that occur near contours, but defects that exceed the size of chips that do not require detection can be detected by detecting defects on rough surfaces. For example, by setting determination elements of a desired size such as 3×3 pixels, 4×4 pixels, etc. in the area W for the rough surface portion, and when all of the determination elements correspond to the rough surface portion, the defect can be detected. In addition, in areas where the rough surface area is provided as a normal pattern, the images of the rough surface area are compared and the mismatched area is detected as a defect, and if necessary, the areas near the contour are excluded using the above method. and can be inspected.

(Effects of the Invention) As explained above, according to the present invention, it is possible to detect contour defects, surface defects, etc. that occur in lead frames manufactured by the etching method in one step, and to automatically exclude defects that do not become defects. Since the method can be inspected quickly, it is possible to obtain effects such as greatly improved inspection accuracy, reliability, and efficiency compared to conventional methods.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing an embodiment of a lead frame defect inspection apparatus according to the present invention, Fig. 2 is a plan view showing an example of a lead frame which is an object to be inspected, and Fig. 3 is a block diagram showing an example of a lead frame defect inspection device according to the present invention. Examples of defects are shown in Figure 3 (al.
3 is a plan view, BL is a cross-sectional view taken along the line X-X in FIG. Figure 4 (al is a plan view of the lead,
FIG. 4(b) is a waveform diagram of an image signal when scanning along A-A in FIG. 4(a), and FIGS. 5 and 6 are illuminating means and imaging means of the inspection apparatus according to the present invention. 7 and 8 are explanatory diagrams for explaining a method of detecting defects from a binary image signal. ■・・・Island 2・・・Inner lead 3・・・Dam 4・・・Outer lead 5・・・Outer frame 8,48・
...Light source 9.49...Linear image sensor 10, 70...Photographing lens 11, 41...Lens 12...XY stage 13...Half mirror 14...・・・・・・
Motor 15... Stage control section 16, 17... Binarization circuit 18... Memory 19... Judgment processing section 20... Control section 29...・Lead frame 30...
- Leads 31, 51.32.52...Polarizing plate 40...First illumination means, 50...Second
Illumination means 80...Signal processing unit Patent applicant Dainippon Printing Co., Ltd. Agent Atsumi Konishi No. 1111 Figure 2 Figure 3 (a) (b) Figure 4 (a) (b) Figure 5 Figure 6

Claims (7)

[Claims] (1) In a lead frame defect inspection system that optically detects the shape and surface condition of a lead frame, the surface condition of the lead frame is irradiated perpendicularly to the surface to be inspected. A first illumination means configured to obtain an image by reflected light, and irradiating light perpendicularly to the opposite surface of the lead frame and coinciding with the optical axis of the first illumination means, and transmitting the light. a second illumination means configured to obtain an image by light, and the first and second illumination means
an imaging means installed at a position capable of receiving specularly reflected light and transmitted light of the lead frame simultaneously irradiated with light by the illumination means; and detecting defects in the lead frame by processing image signals obtained from the imaging means. 1. A lead frame defect inspection device comprising: a signal processing section for detecting defects in a lead frame; (2) The lead frame defect inspection apparatus according to claim 1, wherein the imaging means receives specularly reflected light and transmitted light at the same time. (3) The light intensity of the transmitted light on the image receiving surface of the image pickup means is made larger than the light intensity of the specularly reflected light of the smooth surface of the lead frame. 2. The lead frame defect inspection device according to item 2. (4) The signal processing unit includes at least two binarization circuits and one determination processing unit, and converts the image signal output from the photographing means that receives specularly reflected light and transmitted light simultaneously into the two binarization circuits. The circuit binarizes each with different level thresholds,
The method of claim 1 is characterized in that an image signal of transmitted light and an image signal of only a rough surface portion generated on the surface of a lead frame are separated, and defects are detected in the determination processing section based on the two image signals. A lead frame defect inspection device according to scope 1. (5) A plurality of sets of the first illumination means, second illumination means, and imaging means are provided, and each imaging means photographs corresponding portions of different unit patterns of a lead frame formed by repeatedly arranging a large number of unit patterns. 2. The lead frame defect inspection apparatus according to claim 1, wherein defects are detected by comparing image signals obtained by the above-described methods. (6) The first illumination means and the second illumination means are each provided with a polarizing plate, and the polarizing plates cause the light emitted from each of the illumination means to have planes of polarization perpendicular to each other; Further, two of the photographing means are provided, and the photographing means is further provided with a polarizing plate, and the polarizing plate separates the transmitted light and the specularly reflected light and guides them to separate imaging means. A lead frame defect inspection device according to claim 1. (7) One set of the imaging means and the first illumination means is provided on opposite sides of the lead frame, and the light emitted from each of the first illumination means faces each other. The lead frame defect inspection apparatus according to claim 1, wherein the first illumination means also functions as a second illumination means by acting as transmitted light. .