JP2002033356A - Wire-bonding post-inspection method and device using the same - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an inspection method, together with a device using it for shortening the time required for inspecting a wire after bonding, with improved inspection precision. SOLUTION: A stitch, which is a joint end part of a wire to a lead as well as its vicinity are imaged, related to a plurality of wires bonded for connection between an electrode pad on a semiconductor chip side and a lead of a lead frame, arranged around a chip. The acquired image data is analyzed for sequential inspection. Here, at analyzing the image data acquired corresponding to the wires, a threshold value of image density for discriminating the stitch is automatically calculated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a wire bonding inspection method for sequentially inspecting a plurality of wires bonded for connection between an electrode pad on a semiconductor chip side and a lead of a lead frame, and an apparatus using the same.

[0002]

2. Description of the Related Art Conventionally, in a semiconductor manufacturing process, after a wire is bonded for connection between an electrode pad provided on a semiconductor chip and a lead of a lead frame arranged around the chip, the appearance of the wire is inspected. It is known. In this inspection, the wire and its surrounding area are usually imaged, and the obtained image data is analyzed. By analyzing the image data, the bonding ends of the wire to the wire, the lead, and the electrode pad (referred to as stitches and balls, respectively) are evaluated for properties such as, for example, position, size, or shape. The quality of the wire bonding state is determined.

<Evaluation of stitch> In the evaluation of stitch,
The position and width of the stitch on the lead are determined, and evaluation is performed based on these values. At the time of this evaluation, first, the stitch and its surrounding area are imaged, and image data is obtained. Then, by analyzing this image data, the position of the stitch is detected and its width is measured.
In the related art, a stitch in image data is identified based on a threshold value of image density in order to determine a position of the stitch. As this threshold, a fixed value obtained from the result of evaluating the test sample under predetermined conditions in advance is used,
Utilizing such a threshold value, stitch positions for all chips and all wires to be inspected are obtained.

In determining an analysis range in image data including a stitch and its surrounding area, a lead included in the image data is detected and referred to. For example,
As shown in FIG. 21, a right corner position 80a and a left corner position 80b on the distal end side of the lead 80 pre-taught, a lead direction D2, a wire direction D3, a stitch 9
Based on the five information of the five reference positions and the threshold value of the fixed image density described above, the edge 80e of the leading end of the lead is detected from both corners 80a and 80b of the leading end of the lead, and along the lead direction D2. The right edge 80c and the left edge 80d of the lead 80 are detected. Thus, the analysis range 85 of the image data is determined along the lead shape.

Conventionally, as shown in FIG. 22, in order to measure the width of the stitch 95, an analysis range 8 of the image data is used.
In step 5, every point below the fixed threshold (so-called black point) is counted for each step, and the count value at the point having the maximum count value is calculated as a width.

<Evaluation of Wire> In the evaluation of the wire 90,
The wire 90 is imaged while the focus is moved, and the wire 90 is identified for each piece of acquired image data. For example, when a line sensor is mounted on a camera, FIG.
Is obtained. Then, on this image data, for example, the evaluation is performed while moving a predetermined range S8 from the left end one pixel at a time.

In the processing in the range S, the wire 90
Is detected, the height of the wire 90 is measured, and the position is detected. In order to detect the focal point, the power of the derivative of the luminance density is calculated in the X direction, the sum thereof is calculated, and the height of the wire 90 from the Y direction coordinate where the value is maximized is calculated. The position of the wire 90 is detected from the center position in the X direction at the coordinate position, and a focal point is derived.

Conventionally, as shown in FIG. 24, a function of correcting the measurement reference position P with respect to the bonding deviation of the semiconductor chip 100 is not set, and the position 93 of the wire 90 at the time of teaching is set as a reference. The displacement amount 91 of the wire 90 is calculated. The lead 80 and the wire 90 drawn by a solid line and a broken line represent those at the time of inspection and at the time of teaching, respectively.

Further, conventionally, when the wire 90 is identified, the same processing is performed without dividing the processing regardless of whether the detection position of the wire 90 is above the chip surface or not. As shown in FIG. 25, for example, in order to inspect all four wires 90 having different heights with respect to the chip surface 100a at one time, the wires 9 are required.
It is necessary to manually adjust the teaching so that the center of the movement range 92 of the focal point of the camera that captures the image 0 is near the reference numeral 94 and perform teaching.

<Correction of magnification at the time of each evaluation of ball, stitch, and wire> Further, conventionally, as shown in FIG. 26A, the in-focus position (Htb, Hts, Ht) of each measurement
w) is taught using a teaching semiconductor chip (hereinafter, chip sample) by adjusting the position of the focal point of the camera in advance, and the height Ht of the chip sample is actually measured and stored. . At the time of inspection, the chip height H of the semiconductor chip 1 to be inspected is measured, and the difference Hofs between the height H and the height Ht of the chip sample is calculated with respect to the teaching position (Htb, Hts, Htw) of each component by the following equation. Correct as follows. As a result, the image is always automatically moved to the focal point (Hb, Hs, Htw) to capture the image and perform each measurement. Hb = Htb + Hofs Hs = Hts + Hofs Hw = Htw + Hofs (where Hofs = H−Ht)

To cope with a change in the distance between the objective lens and the semiconductor chip to be inspected due to the focusing,
Inevitably, the optical magnification changes. In response to this, conventionally, it is known to perform a magnification correction process for deriving a measurement result in consideration of a change in magnification. In this magnification correction processing, the distance per pixel at the focal position Hc, that is, the resolution L and the ratio of the magnification change to the moving distance of the focal position Δk are measured in advance using a predetermined jig. For example, taking the measurement of the diameter (BR) of a ball in ball evaluation with a certain test chip as an example, the resolution L ′ at the chip height H at the time of test is: L ′ = L · (1−Δk · △) h) (△ h = H−Hc) is calculated. The measurement result (image size) is R (pixel)
Then, the diameter becomes BR = RL · [μm]. As described above, magnification correction is performed on the measurement result based on the chip height H at the time of inspection. In addition, FIG.
4 shows an image of a ball at a focal height Hb. Regarding the evaluation of the remaining stitches and the evaluation of the wire, magnification correction is performed in the same manner.

[0012]

However, the above-mentioned prior art has the following problems. <Issues on stitch evaluation> In the conventional stitch evaluation, since the threshold value for identifying the stitch is fixed, the influence of the luminance density change due to the change of the surface state of the inspection object and the light amount adjustment by manual operation at the time of teaching. In the case where the luminance density changes due to a variation in the threshold value, the threshold value is not always appropriate, and when the threshold value is too small (see FIG. 2).
7 (a)), if it is too large (see FIG. 27 (b)), there is a concern that the measurement accuracy may be reduced or erroneous recognition may be caused. (A) and (b) of FIG.
T2 and T3 in the middle represent stitches after binarization, respectively.

In the conventional method for detecting a lead, for example, as shown in FIG. 28, when the shape of the lead 80 is special, not only is it difficult to specify the teaching in the lead direction, but also depending on the teaching position. However, there is a fear that the lead detection may be adversely affected. In the example shown in FIG. 28, the right edge 80a of the lead 80 cannot be detected.

Further, in the conventional method for determining the analysis range 85 in the image data, the range is determined along the shape of the lead 80. Therefore, as shown on the left side of FIG. When the image data is wide or has a special shape, the analysis range 85 in the image data is widened, and there is a possibility that a large processing time is required. FIG.
As shown on the right side of FIG. 9, when a scratch 88 occurs during the lead processing, the scratch 88 is moved to the analysis range 8 in the image data.
5 and could be mistaken for stitch 95.

Further, in the conventional method for measuring the stitch width, as shown in FIG. 30, a point (black point) below the threshold value is simply recognized as a stitch, and the stitch width 95W is defined by a line having the maximum number of black points. Was calculated, the indentation 98
Also, since the number of scratches 99 and the like on the lead are counted, a result different from the actual stitch width may be obtained.

Further, in the conventional stitch evaluation, there is no detection processing of the wire 90 within the range extracted from the image data. Therefore, as shown in FIG. Stitches other than stitches such as 99 are processed as normal stitches, and even stitches in which the neck 90a is cut as shown in FIG. If the value is within the range of the judgment value, the product may be regarded as a good product in spite of the defective product.

Further, in the conventional stitch evaluation, if no stitch exists within the range to be analyzed in the acquired image data, the processing is interrupted there and judged to be defective. For example, if a defect occurs by inspecting a semiconductor chip of a line-connected wire bonding apparatus, a stitch position of a specific wire is intentionally determined for each of the units in order to identify which of the unit's wire bonding apparatuses was manufactured. In some cases, wire bonding is performed on the back side or the front side along the lead direction, and the stitches of those wires are out of the range, and there is a problem that it is determined to be defective.

Further, in the conventional stitch evaluation, when evaluating a plurality of stitches on the same lead, if a plurality of stitches are present in a range to be analyzed in the image data, the stitches are determined. Since they interfere with each other, it has been difficult to individually measure the width and detect the position.
Further, in the conventional stitch evaluation, a plurality of stitches of the overstrike type in which a plurality of stitches are overlapped and bonded on the same lead overlap in a range to be analyzed in the image data. It is difficult to determine the position and position, and it is impossible to judge pass / fail only with such stitches.

<Problems of Conventional Wire Measurement> In the conventional wire measurement method, processing is performed while moving the analysis target range in the acquired image data one pixel at a time.
When the range to be analyzed is wide, there is a problem that processing for obtaining the width and position of the stitch included in the range takes time. In addition, the conventional wire measurement does not have a quantitative threshold for judging the likeness of the wire, and simply measures a portion where the luminance density changes remarkably (the peak sum of the sum of the powers of the differential of the luminance density) is measured as the wire height. For this reason, for example, foreign objects, chip patterns, and the like other than wires may be recognized as wires.

Further, in the conventional wire measurement, the deviation of chip bonding is not taken into consideration, so that the semiconductor chip to be inspected can realize the positional relationship between the lead and the semiconductor chip which have been previously taught. If not, the position of the wire will also be different from the teaching position. In such a case, an error occurs in the resulting positional shift amount, and a non-defective product may be determined as a defective product, or a defective product may be determined as a non-defective product, and the reliability of the inspection is lacking. Further, in the conventional wire evaluation, when a wire is detected above the chip surface, a pattern provided on the chip surface that is captured when the focal point is moved may be erroneously recognized as a wire. Was. In the conventional teaching of wire measurement, as shown in FIG. 25, if the focus position at the time of capturing image data is set to the position of the center C, a wire that does not fit in the measurement image may be generated. In order to cope with this, in order to bring the position as accurate as possible, it is necessary to perform teaching while frequently confirming that all the wires in the field of view have been taken into the imaging range of the image data.

Further, in each of the conventional ball, stitch, and wire measurements, the magnification correction of the evaluation result is performed based on the chip height H. However, as shown in FIG. When the height is low, the tip height H of the ball may be used as a reference. However, in the evaluation of the stitch and the wire, the focal point is different for each visual field (Hs,
Hw), when the magnification correction calculation is performed based on the chip height H, a magnification error corresponding to the difference between the heights (ΔHs, ΔHw) occurs.

The present invention has been made in view of the above technical problems, and an inspection method and an apparatus using the same that can reduce the time required for inspecting a wire after bonding and improve the inspection accuracy are provided. The purpose is to provide.

[0023]

According to a first aspect of the present invention, a plurality of wires bonded for connection between an electrode pad on a semiconductor chip side and a lead of a lead frame arranged around the chip are provided to each of the leads. In a post-wire bonding inspection method in which a stitch forming a bonding end portion of a wire and a peripheral area thereof are imaged, the obtained image data is analyzed, and the inspection is sequentially performed, analysis of the image data obtained corresponding to each of the wires is performed. In this case, a threshold value of the image density for identifying the stitch is automatically calculated.

According to a second aspect of the present invention, in the first aspect, before imaging the stitch and its surrounding area, the stitch is set at an arbitrary position based on a teaching position related to a preset stitch, which is orthogonal to the wire direction. The feature is that a lead is detected within a range that can be set to a value.

Further, according to a third aspect of the present invention, in the second aspect, a range to be analyzed is set based on the lead detection result in the image data including the stitch and its peripheral area. It is characterized by.

Further, in a fourth aspect of the present invention, in any one of the first to third aspects, a range to be analyzed in the image data is referred to along a direction orthogonal to a wire direction. Then, after extracting the edge in the width direction of the stitch, the number of pixels between both edges is sequentially counted along the wire direction, and the line having the maximum number of pixels is calculated as the width of the stitch. It is what it was.

Further, according to a fifth aspect of the present invention, in any one of the first to fourth aspects, a predetermined range is extracted from a range to be analyzed in the image data. In the range, the presence or absence of a wire is determined.

Further, according to a sixth aspect of the present invention, in the fifth aspect, when it is determined that the stitch is defective within a range to be analyzed in the image data, the wire direction is determined. And moving the range to be analyzed in a predetermined direction on the image data along the line, and again determining the presence or absence of the stitch.

Further, a seventh invention of the present application is the method according to any one of the first to sixth inventions, wherein a plurality of stitches present on the same lead are evaluated.

Further, according to an eighth invention of the present application, in any one of the first to seventh inventions described above, when inspecting a plurality of wires over-punched on the same lead, the over-stitched stitch is used. Are characterized by comparing the respective wires with respect to the area of.

Further, according to a ninth invention of the present application, a plurality of wires bonded for connection between an electrode pad on a semiconductor chip side and a lead of a lead frame arranged around the chip are imaged, In the post-wire bonding inspection method of analyzing the acquired image data and sequentially inspecting the wire, the wire is imaged while moving the focal point, the wire is roughly detected from each of the acquired image data, and the temporary position of the wire is obtained. Then, fine detection is carried out after that.

Further, according to a tenth aspect of the present invention, in the ninth aspect of the present invention, a threshold value of an image density in an area which can be arbitrarily set in advance around the focal point is obtained from each of the image data. The wire is identified based on the threshold value.

Further, an eleventh invention of the present application is the invention according to the tenth invention, wherein the image density threshold value is calculated for each visual field.

Further, according to a twelfth aspect of the present invention, in any one of the ninth to eleventh aspects, the image density at the in-focus and out-of-focus is obtained from each of the image data, and the difference between the two is obtained. When the ratio does not reach the predetermined ratio, it is not recognized as a wire.

Further, according to a thirteenth aspect of the present invention, in any one of the ninth to twelfth aspects of the invention, when measuring the wire displacement, the chip bonding displacement for each semiconductor chip is considered. While correcting the measurement reference position.

Further, according to a fourteenth aspect of the present invention, in any one of the ninth to thirteenth aspects, when the position to be inspected is on the chip surface, the inspection range is limited to a predetermined range. It is characterized by doing.

Further, according to a fifteenth aspect of the present invention, in any one of the ninth to fourteenth aspects, a focus movement center position at the time of imaging of the image data is detected after teaching, and the detection is performed within a visual field. Is calculated from the average height of all the wires and automatically set.

Further, according to a sixteenth aspect of the present invention, a plurality of wires bonded for connection between the electrode pads on the semiconductor chip side and the leads of the lead frame arranged around the chip are formed by connecting the plurality of wires to the wire pads and the electrode pads. In a post-wire bonding inspection method in which a ball and a stitch forming a bonding end portion of each wire with respect to a lead and an image are analyzed and acquired image data are sequentially inspected, an inspection chip is used for evaluating the ball, stitch and wire. In consideration of the difference between the result of the tip height measurement performed for each test and the tip height at the time of teaching, the magnification change that occurs when the lens is moved to the focal point for each evaluation is calculated for each evaluation result for each view height. The correction is performed every time.

Further, according to a seventeenth aspect of the present invention, a plurality of wires bonded for connection between an electrode pad on a semiconductor chip side and a lead of a lead frame arranged around the chip are formed by connecting the plurality of wires to the wire and the electrode pad. In a post-wire bonding inspection apparatus for imaging a ball and a stitch forming a bonding end portion of each wire with respect to a lead, analyzing acquired image data, and sequentially inspecting the image, any one of the first to sixteenth aspects of the present invention It is characterized by using the inspection method described in (1).

[0040]

Embodiments of the present invention will be described below with reference to the accompanying drawings. <Evaluation of Stitch> FIG. 1 is an explanatory view schematically showing an inspection apparatus after wire bonding according to the first embodiment of the present invention. The post-wire bonding inspection apparatus 10 captures an image of a wire 11 connecting an electrode pad (not shown) on the semiconductor chip 1 and a lead (see FIG. 2) of a lead frame disposed therearound and a peripheral area thereof. The obtained image data is analyzed to evaluate the stitches, the balls, and the wires themselves, which are the joining ends of the wires 11 to the leads and the electrode pads, and inspect the wires. The inspection apparatus 10 has, as its basic configuration, an image of the wire 11 and its surrounding area, for example, C
An imaging element (hereinafter referred to as a camera) 2 such as a CD; an objective lens 3 disposed to face the semiconductor chip 1; and an imaging lens 4 movable vertically and capable of adjusting the focal length of the camera 2. A half mirror 5 disposed on the optical axis of the objective lens 3 and the imaging lens 4, an epi-illumination 6 for irradiating the semiconductor chip 1 with light that is vertically irradiated from above through the half mirror 5, It has an oblique illumination 7 for irradiating the chip with light obliquely, an image memory 8 for storing the image data obtained by the camera 2, and a control unit 9 for processing and analyzing the stored image data. ing.

<Detection of Threshold> FIG. 2 is an explanatory diagram of a method of calculating a threshold for identifying a stitch included in the image data acquired by the camera 2. When calculating the threshold value, the identification range S1 including the stitch 13 in the image data and orthogonal to the wire direction D1 is to be processed.
The size of the identification range S1 can be set arbitrarily. From the identification range S1, a maximum value and a minimum value of the luminance density are obtained for each step (steps 1 to n) indicated by an arrow within the range, and the luminance data for each step is calculated by the following equation. date (i) = (maximum value−minimum value) × R + minimum value (i: 0 to n−1, R: R can be arbitrarily set in a range of R <0) , An average value of all steps is obtained, and this Mth is used as a threshold for stitch identification.

FIG. 3 is a frequency distribution diagram relating to the luminance density in the identification range S (see FIG. 2). For example, R = 0.
When the threshold value is set to 4, the threshold value Mth is an optimal value with the luminance density at the broken line m in the figure. This optimal threshold Mt
FIG. 4 shows image data of only the identification area S binarized by h. T1 in the figure represents a binarized stitch. As described above, for the range including the stitch in the image data acquired corresponding to each wire, the optimum threshold value is calculated for each stitch 13 of each wire 11 before calculating the stitch width. Thereby, good detection accuracy can always be ensured without being affected by changes in the state of the stitch 13 and its surrounding area, changes in the amount of light, and the like.

FIG. 5 is an explanatory diagram of a method for detecting a lead included in the acquired image data. Lead 20
When detecting stitches, stitch 1
The left edge 20b and the right edge 20c of the lead 20 orthogonal to the wire direction D1 are detected with reference to the position 3 and the average value of the distance between both edges 20b, 20c is calculated as the lead width W1. Further, the center position between both edges 20b and 20c is stored as the center position C1 for lead detection, and the position information is used when determining the range to be analyzed in the image data at the time of stitch evaluation. The size of the range S2 targeted for such read detection can be set arbitrarily. At this time, the teaching is unnecessary because the wire direction D1 is automatically calculated from the relationship with the ball teaching position P1.

Accordingly, it is possible to perform lead detection only within an arbitrarily settable range orthogonal to the wire direction, and to omit teaching of the shape of the lead 20 (for example, both corners at the tip of the lead and the lead direction). Can be. As a result, the teaching position at the time of stitch evaluation
Only the stitch position can be used.

FIG. 6 is a diagram showing a range to be analyzed in the image data at the time of stitch evaluation. The range S3 to be analyzed (hereinafter referred to as an analysis range) is
The center position C of the lead width W1 obtained in the above lead detection processing
1 is determined as the center. The vertical and horizontal sizes of the analysis range S3 can be arbitrarily set within a range not exceeding the lead width W1 in consideration of the width 13W and the length 13Z of the standard stitch 13 to be analyzed. It can be optimized and minimized.

As described above, since the analysis range in the image data is determined at the time of stitch evaluation based on the lead detection result, the analysis range is always determined regardless of the wiring state.
It is possible to keep the direction of the stitch 13 in the image data constant. Further, since the analysis range S3 in the image data can be limitedly set, stitch evaluation can be facilitated and processing can be speeded up. Furthermore, even if there is a processing scratch on the lead tip, it is not affected,
Accurate evaluation can be realized.

In this embodiment, as shown in FIG. 7, the measurement of the stitch width at the time of stitch evaluation is performed in the analysis range S3 in which points equal to or smaller than the threshold (hereinafter referred to as black points) are continuous.
The black point is counted in the horizontal line inside. Then, the maximum value of the count value is calculated as the stitch width 13W. This can avoid erroneously recognizing stitches other than stitches such as indentations 18 and scratches 19.

Further, since the analysis range S3 in the image data processed at the time of stitch evaluation is image data cut out based on the wire direction D1, image data in a fixed direction can be always obtained with any wire. . As a result, as shown in FIG. 7, as long as wire bonding is normally performed, the wire 1
There is one. Therefore, a range S4 having a predetermined size and shape below the stitch 13 is extracted, and the presence or absence of the wire 11 is detected in the extracted range S4. The size of the range S4 can be set arbitrarily. This allows
If the wire 11 is not detected in the extraction range S4, the stitch 13 is determined to be defective, so that it is possible to avoid erroneously recognizing a stitch other than a stitch such as an indentation 18 or a scratch 19 as a stitch.

Further, in this embodiment, as shown in FIG. 8, when the analysis range S3 in the image data is misaligned so as not to include the stitch 13, it is determined that the analysis range S3 is defective. The analysis range S3 is moved along the wire direction D1, and the analysis is retried. Analysis range S3
The amount of movement and the number of retries can be arbitrarily set. As a result, it is possible to evaluate stitches shifted out of the analysis range S3, and it is possible to improve the inspection margin.

Next, the evaluation of a plurality of stitches provided on the same lead will be described with reference to a case where two stitches are provided (so-called double stitch). FIG.
In addition, two stitches 15A provided on the same lead,
15B is shown. In this case, at the time of teaching, the teaching of double stitch is performed in advance by, for example, setting a flag for each of the wires 11A and 11B. Then, after detecting the lead of the wires 11A and 11B on which the double stitch flag is set,
Divide the analysis area.

In this division method, first, it is known that the stitch 15A of one wire 11A is on the left side with respect to the center of the lead based on the teaching position information. The left portion of the detection width LW1 is defined as an analysis range S5 at the time of the stitch evaluation of the wire 11A. Next, regarding the other wire 11B, it is also known that the stitch 15B is on the right side with respect to the lead center based on the teaching position information, so that the right side of the lead detection width LW2 based on the position of the stitch 15B is used. Is defined as an analysis range S6 at the time of stitch evaluation of the wire 11B. In this case, the widths W5 and W6 of the analysis ranges S5 and S6 are determined by a predetermined ratio r (r <1) with respect to the read detection widths LW1 and LW2.
Can be arbitrarily set to LW1 × r and LW, respectively.
2 × r.

The evaluation of a plurality of stitches that are overprinted on the same lead (especially a narrow lead) will be described with respect to double stitches.
FIG. 10 shows two overstitched stitches 11C and 11D determined to be non-defective, and FIG. 11 shows various forms of the two stitches 11C and 11D. In this case, the area (the number of black spots) of the overlapping portion of the two stitches 11C and 11D determined to be non-defective as shown in FIG. 10 is counted and stored as a reference value.

As shown in FIG. 11, when the position of a stitch to be inspected is shifted (left), narrow (center), or one of the stitches is peeled off (right), the number of black spots in the analysis range is determined. Is less than the reference value, it is determined that the width and position are non-defective.If the value is less than the reference value, there is a problem with the width or position, or a defect such as stitch peeling. Is determined to have occurred. This makes it possible to indirectly evaluate a plurality of over-stitched stitches.

<Evaluation of Wire> Next, the evaluation of the wire will be described. In the inspection apparatus 10 after wire bonding shown in FIG. 1, the semiconductor chip 1 is irradiated with the wire 11 at the position where the teaching has been performed in advance by oblique illumination, and the imaging lens is moved (that is, the focal point is shifted downward or downward. For example, when the camera 2 is equipped with a line sensor, image data as shown in FIG. 12 is acquired. In this image data, a thinned area (in this embodiment, areas 21A to 21A every four pixels)
1H) is set. The image data is sequentially referred to from the left end to the right only in these thinned regions, and an average value of the luminance density is obtained for each region.

Further, based on the average value, points over the average value are detected over the entire image data obtained by taking in the wire 11, and the point found first in each of the regions 21A to 21H is determined as a tentative wire. Position 23. Then, the original fine detection processing is performed with reference to the image data in the detection range S7 set such that the temporary wire position 23 is located at the center. The thinning areas 21A to 21H can be set arbitrarily. As described above, the image data acquired while moving the focal point is roughly detected, the temporary wire position 23 is obtained, and then the fine detection, which is the original processing, is performed. It is possible to speed up the inspection.

As shown in FIG. 13, normally, to detect the focal point 25a of the wire 11, the power sum of the differential of the luminance density of the image data within the detection range S7 of the wire 11 is calculated. The wire height 25b is calculated from the Y-direction coordinate 25d at which is the maximum, and the wire position 25c is calculated from the center position in the X-direction at the Y-coordinate position to derive a focal point. In this embodiment, if the sum of the powers of the differentiation of the luminance density at the focal point is not more than a predetermined threshold value 25 g, the wire 11 is not recognized as a wire 11.
A condition for identifying is added. This condition is such that when detecting something other than a wire (for example, a foreign substance attached to a chip or a wire), the value of the power sum of the differentiation of the luminance density becomes extremely low, and on the other hand, a wire consisting of a gold wire is detected. In this case, if the thickness is constant to some extent, the characteristic that the value is substantially constant for any wire regardless of the wire direction is used. The threshold 25g can be set arbitrarily.

Further, as a condition for identifying the “wire likeness”, the value of the sum of the powers of the differentiation of the luminance temperature at the in-focus point is determined as the out-of-focus point (the head position 25e of image data acquisition)
In addition, it is added that the sum of the power of the luminance density at the final position 25f) is n times or more. n can be set arbitrarily. This condition utilizes a characteristic that a wire having a normal height focuses on the vicinity of the center of the image data, and a non-focusing point exists above and below the center.

Further, for example, when two types of cameras (first and second cameras) are used, if there is a gain difference between the two cameras, the brightness value will be different even when the same object is photographed. FIGS. 14A and 14B show images 31 and 32 acquired by the first and second cameras, respectively. When the gain of the first camera is larger than that of the second camera, comparing the image obtained by the first camera with the image obtained by the second camera, the former image is more Bright. Therefore, there is a possibility that a difference may be caused in the value of the power sum of the derivative of the luminance density, which is the threshold value for identifying the likeness of the wire. To cope with this, non-defective samples are measured in advance for each field of view under actual imaging conditions such as the type of camera and the amount of light, and it is determined how much a difference in values occurs. You may make it set.

FIG. 15 is an explanatory diagram showing a method for correcting a chip bonding deviation during wire evaluation. As shown in this figure, in the measurement of the displacement amount of the wire 11, the chip bonding displacement amount 36 is developed at the position of the stitch 13 for each semiconductor chip 1 to be inspected, and the ball is connected to the position of the stitch 13. The straight line 37 is regarded as a normal wire position, and the displacement measurement reference position is corrected to a position 39 in consideration of the displacement amount. The corrected position is used as an inspection reference for the semiconductor chip 1 as a positional shift amount 35 of the wire 11.
Is measured. L in the figure is the distance from the end of the semiconductor chip 1 to the measurement position, and is a value that is previously taught. The lead 20 and the wire 11 drawn by a solid line and a broken line represent those at the time of inspection and at the time of teaching, respectively.

FIGS. 16A and 16B are explanatory diagrams showing a wire evaluation method from above the chip surface. FIG. 17 is a diagram illustrating an image having a focal point of the wire chip and the chip pattern acquired from above the chip surface. Reference numeral 1e in FIG. 16 represents an electrode pad provided on the chip surface, and reference numerals 41 and 42 in FIG. 17 represent a focal point of the wire 11 and a focal point of the chip pattern, respectively. As can be clearly understood from FIG. 16A, when the measurement position of the wire 11 is set above the chip surface, the chip data 1 in addition to the wire 11 is included in the image data at the time of wire evaluation. May be. In this case, the condition that the chip pattern 1c is determined to be the wire 11 is satisfied, and FIG.
The height may be measured from the focal point 42 of the chip pattern 1c shown in FIG. In the present embodiment, in order to deal with this, the focal point existing within the predetermined range 43 on the chip surface side is ignored, and the focal point 4 of the chip pattern 1c is ignored.
2 can be prevented from being recognized as the wire 11. The predetermined range 43 can be set arbitrarily.

FIG. 18 is an explanatory view showing a method for obtaining the image capturing center position at the time of wire evaluation. In this embodiment, after the end of the teaching, first, the actual measurement is performed by using the capture range 46 having twice the size of the capture range 48 that has been pre-taught with the capture center 45 of the image data at the time of teaching as the temporary center. Perform
In this case, even if there is a certain degree of height variation in the field of view between the wire groups 11 to be inspected, it can be included in the captured image range 46. Further, an average value of the heights of all the wires 11 in the measured visual field is obtained, and the height is used as teaching data of the capturing center position 47 of the visual field,
The measurement range is returned to the normal range 48 around the capture center position 47. Further, the chip height at that time is also stored as a reference value.

Next, magnification correction for each visual field height at a focal length according to the height of the semiconductor chip to be inspected will be described. First, the measurement of stitch width is taken up. As shown in FIG. 19A, the resolution Ls at the visual field height Hs of the stitch in consideration of the difference between the height of the semiconductor chip 1 to be inspected and the height at the time of teaching is: Ls = L · (1− Δh) (Δh = Hs−Hc). Here, assuming that the measurement result of the width of the stitch 13 shown in FIG. 19B is Q (pixel), the stitch width SW
Is obtained as follows: SW = Q · Ls [μm]

Next, detection of the wire position will be described. For example, when a line sensor is mounted on a camera, FIG.
The two wires 51 and 52 having different heights shown in FIG.
At the same time, when the visual field height Hw is measured with the center in the Z direction,
When the wires are located at both ends of the visual field, image data as shown in FIG. 20 is obtained. In this embodiment, in the detection of the wire position, the focus position is detected while swinging from the bottom to the top (that is, while increasing the magnification). The image becomes larger.

The wire on the left side with respect to the visual field x direction center 61 moves downward, so that the center line 62 of the image of the wire moves to the left. Toward the center line 63 of the wire image
Goes to the right. At this time, the focal points of the respective wires are 64 and 65, and there is a height difference Δhw between the two. The distances Pw1 and Pw2 from the field of view × direction center 61 of each wire are determined by the resolution (Lw
1, Lw2), and the distance is calculated.
Next, a calculation example will be described. First, the distance Pw1 from the visual field of the left wire to the center in the direction is obtained. Since the left wire almost coincides with the center of the visual field height, the resolution L at that height is obtained.
w1 is as follows: Lw1 = L · (1− △ h) (△ h = Hw1−Hc, Hw1 = Hw) If the number of pixels from the center of the field of view × W1 is W1, the distance Pw1 is Pw1 = Lw1 · W1.

Next, the distance Pw2 from the visual field of the right wire to the center in the direction is determined. Since the right wire is shifted from the center of the visual field height by Δhw, the resolution L at that height is
w2 is Lw2 = L ＝ (1- △ h) (△ h = Hw2-Hc, Hw2 = Hw- △ hw). Then, assuming that the number of pixels from the center of the field of view × W2 is W2, the distance Pw2 is Pw2 = Lw2 · W2. As described above, in the wire measurement, an accurate distance is calculated by calculating the resolution at the magnification at the focal point position for each wire.

The present invention is not limited to the illustrated embodiment, and it goes without saying that various improvements and design changes can be made without departing from the spirit of the present invention.

[0067]

According to the first aspect of the present invention, in the inspection method after wire bonding, an optimum threshold is always calculated without being affected by a change in a state around a stitch and a change in a light amount. Therefore, erroneous recognition can be avoided, and the inspection accuracy can be improved.

According to the second aspect of the present invention, in the inspection method after wire bonding, the teaching process is simplified, and good lead detection is always performed without being affected by the shape of the lead. Can be.

Further, according to the third aspect of the present invention, in the inspection method after wire bonding, the direction of the stitch on the image data is always constant regardless of the wiring state, and the stitch detection range Is limited, useless processing can be omitted, and the processing can be simplified.

Further, according to the invention of claim 4 of the present application, in the inspection method after wire bonding, the stitch width can be accurately measured.

Further, according to the invention of claim 5 of the present application, in the inspection method after wire bonding, it is possible to judge that something other than the wire such as a neck break or an indentation is defective, thereby improving the inspection accuracy. Can be.

Further, according to the invention of claim 6 of the present application, in the inspection method after wire bonding, the stitch is evaluated even when the position of the stitch is intentionally shifted and out of the analysis range. It becomes possible.

Further, according to the invention of claim 7 of the present application, in the inspection method after wire bonding, it is possible to cope with various types of sketches, and good inspection is always possible.

Further, according to the invention of claim 8 of the present application, in the inspection method after wire bonding, even when a plurality of stitches overlap on the same lead, it is impossible to measure the width or detect the position. Stitches can be evaluated.

Further, according to the invention of claim 9 of the present application, in the inspection method after wire bonding, the wire can be inspected at high speed by performing the processing efficiently.

Further, according to the tenth aspect of the present invention, the inspection method after wire bonding has a quantitative threshold value for judging the likeness of a wire, so that the luminance density of a wire other than a wire (gold wire) is determined. Can be identified, thereby preventing erroneous recognition.

Further, according to the invention of claim 11 of the present application, in the inspection method after wire bonding, the inspection accuracy of wire evaluation can be further improved.

Further, according to the twelfth aspect of the present invention, even when different types of imaging means are used in the inspection method after wire bonding, a gain difference between the imaging means and other optical factors are obtained. Since the optimum threshold value is used for the field of view without being affected by the luminance change due to, the wire with higher reliability can be evaluated.

Further, according to the invention of claim 13 of the present application, in the inspection method after wire bonding, it is possible to measure the amount of misalignment of the wire with high accuracy.

Further, according to the invention of claim 14 of the present application, in the inspection method after wire bonding, it is possible to prevent a chip pattern, which appears when the focal point is moved, from being erroneously recognized as a wire. Reliable wire evaluation becomes possible.

Further, according to the invention of claim 15 of the present application, in the inspection method after wire bonding, the teaching processing can be facilitated and optimized.

Further, according to the sixteenth aspect of the present invention, in the inspection method after wire bonding, a change in magnification at a focal length according to the height of a chip to be inspected is determined for each visual field height of each measurement. Because of the correction, a highly accurate measurement result can be obtained.

Further, according to the seventeenth aspect of the present invention, it is possible to reduce the time required for inspecting a wire after bonding and to improve the inspection accuracy in the inspection apparatus after wire bonding. Become.

[Brief description of the drawings]

FIG. 1 is an explanatory view schematically showing an inspection apparatus after wire bonding according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an image including an inspection target area for obtaining a threshold value of an image density from image information at the time of stitch evaluation.

FIG. 3 is a graph showing a frequency distribution of luminance in the inspection target area.

FIG. 4 is a diagram showing an image in which the inspection target area is binarized based on an image density threshold.

FIG. 5 is an explanatory diagram showing a lead detection area in the image information.

FIG. 6 is an explanatory diagram showing a stitch evaluation range in the image information.

FIG. 7 is a diagram illustrating a method of measuring a stitch width at the time of stitch evaluation.

FIG. 8 is an explanatory diagram showing the movement of the evaluation range.

FIG. 9 is an explanatory diagram showing a plurality of stitches provided on one lead.

FIG. 10 is an explanatory diagram showing a plurality of stitches over-punched on one lead.

FIG. 11 is an explanatory diagram showing over-stitched stitches evaluated as defective.

FIG. 12 is an explanatory diagram showing a rough detection method at the time of wire evaluation.

FIG. 13 is an explanatory diagram of a power sum of a derivative of luminance density at the time of wire evaluation.

FIG. 14 is a diagram illustrating a difference in brightness of an image due to a difference in camera gain during wire evaluation.

FIG. 15 is an explanatory diagram showing a chip bonding deviation correction method at the time of wire evaluation.

FIG. 16 is an explanatory diagram showing a processing method on a chip surface during wire evaluation.

FIG. 17 is a diagram showing an image having a focal point of a wire chip and a chip pattern.

FIG. 18 is an explanatory diagram showing a method for obtaining an image capturing center position during wire evaluation.

FIG. 19 is an explanatory diagram illustrating a magnification correction method for a focal position at the time of stitch evaluation.

FIG. 20 is an explanatory diagram illustrating a magnification correction method related to a focal position during wire evaluation.

FIG. 21 is an explanatory diagram of a conventional lead detection method.

FIG. 22 is an explanatory view showing a conventional stitch width measuring method.

FIG. 23 shows an evaluation range in image data at the time of a conventional wire evaluation.

FIG. 24 is an explanatory diagram showing a conventional wire position detection method.

FIG. 25 is a diagram illustrating a conventional image capturing center position adjustment of a wire.

FIG. 26 is an explanatory diagram showing a conventional magnification correction method.

FIG. 27 is an explanatory diagram of a problem caused by a conventional fixed threshold value.

FIG. 28 is an explanatory diagram showing a problem of the conventional lead detection.

FIG. 29 is an explanatory diagram of a problem of a conventional stitch evaluation range.

FIG. 30 is an explanatory diagram of a problem of a conventional stitch evaluation.

FIG. 31 is an explanatory diagram of another problem of the conventional stitch evaluation.

FIG. 32 is an explanatory diagram of a problem of conventional magnification correction.

[Explanation of symbols]

Reference Signs List 1 semiconductor chip, 2 camera, 10 inspection device after wire bonding, 11 wires, 13 stitches, 13W
Stitch width, 20 leads, S3 analysis range

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Masayuki Hiroki 2-6-1 Otemachi, Chiyoda-ku, Tokyo Within Mitsubishi Electric Engineering Co., Ltd. (72) Inventor Katsumi Ohno 2-6-Otemachi, Chiyoda-ku, Tokyo No. 2 Inside Mitsubishi Electric Engineering Co., Ltd. (72) Inventor Koichi Suzuki 2-6-2 Otemachi, Chiyoda-ku, Tokyo Inside Mitsubishi Electric Engineering Co., Ltd. (72) Inventor Masuyuki Yano Otemachi, Chiyoda-ku, Tokyo 2-6-2 Mitsubishi Electric Engineering Co., Ltd. (72) Inventor Shoji Yoshida 2-3-2 Marunouchi, Chiyoda-ku, Tokyo 2-3-2 Mitsubishi Electric Co., Ltd. (72) Inventor Kanehisa Marunouchi 2, Chiyoda-ku, Tokyo F-term (reference) 2-3, Chome 2-3 Mitsubishi Electric Corporation 2G051 AA90 CA04 CA07 EA08 EA11 EA14 EB01 EC02 EC03 ED01 5B057 AA03 BA02 DA03 DA07 DC03 DC16 DC22 5F044 AA01 JJ00

Claims (17)

[Claims] 1. A stitch forming a bonding end of each wire to a lead and a peripheral region thereof, the plurality of wires being bonded for connection between an electrode pad on a semiconductor chip side and a lead of a lead frame disposed around the chip. In the post-bonding inspection method of sequentially capturing and analyzing the acquired image data, in analyzing the image data acquired corresponding to each of the wires, the image density for identifying the stitch is analyzed. An inspection method after wire bonding, wherein a threshold value is automatically calculated. 2. The method according to claim 1, wherein before the imaging of the stitch and its surrounding area, a lead is detected within a range which can be arbitrarily set based on a teaching position related to a predetermined stitch, which is orthogonal to a wire direction. The inspection method after wire bonding according to claim 1. 3. The inspection method after wire bonding according to claim 2, wherein a range to be analyzed is set based on the lead detection result in the image data including the stitch and its peripheral area. 4. A range in the image data to be analyzed is referred to along a direction perpendicular to the wire direction, and an edge in the width direction of the stitch is extracted. The inspection method according to any one of claims 1 to 3, wherein the number is sequentially counted along the wire direction, and the line having the maximum number of pixels is calculated as the width of the stitch. 5. The method according to claim 1, wherein a predetermined range is extracted from a range to be analyzed in the image data, and the presence or absence of a wire is determined in the extracted range. 4. The inspection method after wire bonding according to 1. 6. A range to be analyzed in a predetermined direction on the image data along the wire direction when it is determined that the stitch is defective within the range to be analyzed in the image data. The inspection method according to any one of claims 1 to 5, wherein the presence or absence of a stitch is determined again by moving the wire. 7. The post-wire bonding inspection method according to claim 1, wherein a plurality of stitches existing on the same lead are evaluated. 8. The method according to claim 1, wherein when inspecting a plurality of wires over-punched on the same lead, each wire is compared with respect to an area of the over-stitched stitch. 4. The inspection method after wire bonding according to 1. 9. A plurality of wires bonded for connection between an electrode pad on a semiconductor chip side and a lead of a lead frame arranged around the chip are imaged, and the obtained image data is analyzed. In the inspection method after wire bonding, which sequentially inspects the wire, the wire is imaged while moving the focal point, the wire is coarsely detected from each of the acquired image data, and the fine position is determined after determining the temporary position of the wire. Inspection method after wire bonding. 10. The apparatus according to claim 9, wherein a threshold value of an image density in an area which can be arbitrarily set around a focal point is obtained in advance from each of the image data, and a wire is identified based on the threshold value. Inspection method after wire bonding. 11. The inspection method according to claim 10, wherein the threshold value of the image density is calculated for each visual field. 12. An image density at an in-focus point and an out-of-focus point from each of the image data, and if the difference between the two does not reach a predetermined ratio, the image density is not recognized as a wire. 12. The inspection method after wire bonding according to any one of 11. 13. The measurement reference position according to claim 9, wherein the measurement reference position is corrected in consideration of a chip bonding deviation for each of the semiconductor chips when measuring the wire position deviation amount. 4. The inspection method after wire bonding according to 1. 14. The inspection method after wire bonding according to claim 9, wherein the inspection range is limited to a predetermined range when the position to be inspected is on the chip surface. . 15. The method according to claim 9, wherein a focus movement center position at the time of imaging of the image data is detected after teaching, is calculated from an average height of all wires in a visual field, and is automatically set. 15. The inspection method after wire bonding according to any one of items 14 to 14. 16. A plurality of wires bonded for connection between an electrode pad on the semiconductor chip side and a lead of a lead frame arranged around the chip, and a bonding end of each wire to the wire, the electrode pad and the lead is formed. In a post-wire bonding inspection method for imaging balls and stitches to be formed, analyzing acquired image data, and sequentially inspecting the balls, the results of chip height measurement performed for each inspection chip when evaluating the balls, stitches, and wires are described. Considering the difference with the tip height at the time of teaching, the magnification change that occurs when moving to the focal point of each evaluation is calculated for each evaluation result.
An inspection method after wire bonding, wherein the inspection is performed for each visual field height. 17. A plurality of wires bonded for connection between an electrode pad on the semiconductor chip side and a lead of a lead frame disposed around the chip, and a bonding end of each wire to the wire, the electrode pad, and the lead is formed. A post-wire bonding inspection apparatus that images a ball and a stitch to be formed, analyzes acquired image data, and sequentially inspects the wire data, wherein the inspection method according to any one of claims 1 to 16 is used. Inspection device after bonding.