JP4050772B1 - Measuring method and measuring device for straightness of fine metal wire - Google Patents

Abstract

Measurement and evaluation of wire bending that causes bonding wire erosion with ease and accuracy.
A sample wire is suspended and photographed from a direction perpendicular to the drooping direction, and the meandering degree of a curve obtained from the photographed image is divided into 2 to 25 times the loop length and evaluated.
The bending of the wire that affects the leaning is grasped as the meandering width from the interval between the parallel lines that sandwich the curve, but the long-period waviness does not affect the leaning even if the meandering width is large, and the radius of curvature that affects the leaning Bend is understood as the meandering width in the above section in relation to the loop length.
In the wire shooting section, the test wire is divided in the length direction in accordance with the accuracy of the device, and the image enlarged in the width direction is processed as binary data and evaluated as a continuous curve.
[Selection] Figure 3

Description

  The present invention relates to a method and an apparatus for measuring the straightness of a thin metal wire that has been drawn for joining and using both ends of a thin wire cut at a certain length, such as wire bonding, and the wire diameter is 0.007 to The present invention relates to a method and an apparatus for measuring straightness of a fine metal wire in a range of 0.1 mm, particularly 0.015 to 0.025 mm.

Conventionally, a thin metal wire that has been drawn is traded by being wound around a spool, and is cut into a predetermined length before use. The fine metal wires handled in such a form include electrical component leads, detector leads, automobile tire bead wires, and connection wires for semiconductor devices, such as bump wires and bonding wires for connecting electronic circuit elements and circuit boards. It is a very common form. Depending on the application, such fine metal wires use metals such as Au, Ag, Pd, Pt, Al, Cu, Fe, Ni, Pb, Sn, and W, or alloys containing these elements as main components. These ingots are continuously drawn from a thick line to a thin line, and heat-treated before, during and after wire drawing as necessary.
The meandering shape of this fine metal wire can be easily measured by suspending and visually observing the fine metal wire under gravity (Japanese Patent Laid-Open No. 11-190604). As a result, the fine metal wire having a larger meander width is less straight, It was said that it was easy to form defective products.

However, in the case of visual observation, the measurement values vary due to individual differences and quantitative evaluation is difficult. Therefore, the applicant draws out a thin metal wire from the tip by a predetermined length, illuminates the target material, images the target material, and evaluates the straightness of the captured image by computer processing. It was developed and proposed earlier (Patent Document 1: Japanese Patent Laid-Open No. 2002-124534). This evaluation device projects the shape of a three-dimensional fine metal wire onto a two-dimensional photographic image, and preferably, since the fine metal wire is a continuous line, the non-photographed portions of the projected image are connected to form a meandering curve. It corrects and recognizes as. When the measurement result obtained by this evaluation apparatus was compared with the measurement result obtained by visual observation, it was possible to significantly suppress variations in straightness measurement values. (See Examples 5, 6, 10-12 and Comparative Examples 1 and 2 in Table 1 on page 5).
Further, in the evaluation method disclosed in Japanese Patent Application Laid-Open No. 2006-86174 (Patent Document 2), the evaluation method of Patent Document 1 states that “when a thin metal wire is suspended, it is deformed into a meandering shape possessed by the thin metal wire due to its own weight. For the reason that it is difficult (patent document 2 third paragraph), “based on the shape of the metal thin wire on the liquid surface of the liquid, floating on the liquid with at least one end of the metal thin wire as a free end. An evaluation method and apparatus for evaluating the straightness of the thin metal wire (Patent Document 2, fifth paragraph) is disclosed.
JP 2002-124534 A JP 2006-86174 A

However, the following problems remain even in the measurement method using these straightness evaluation apparatuses.
First, the case of Patent Document 1 will be described. Since the fine metal wire is thin and deformed three-dimensionally, it is difficult to obtain a photographed image with uniform contrast due to irregular reflection or the like. For this reason, a background portion that cannot be distinguished from the background occurs in the imaging of the fine metal wire, and many errors occur when the meandering curve is created. In particular, as the metal thin wire becomes thinner to a diameter of 0.1 mm or less, it becomes difficult to identify the meandering curve, and more errors occur.
Moreover, even when the image is well-illuminated and the image is taken uniformly, when a meandering curve is created, an error of about ± 2 pixix occurs between the metal thin line and the background portion. For example, when shooting a 30 cm long region using a camera with 4.75 million pixels (vertical 2,500 pix × horizontal 1,900 pix), even if the measurement accuracy of the camera is 120 μm / pix, the maximum error of 4 pxi An error of 480 μm occurred when converted.
In addition, the difference in straightness is small for the continuously drawn thin metal wires, and most of the meandering widths to be judged are compared within 10 pxi. Therefore, the higher the straightness, the larger the size of the meandering. It was difficult to confirm with.

Next, Patent Document 2 will be described. Even if an attempt is made to float a gold wire having a diameter of 18 μm and a length of 350 mm in a liquid holding container filled with pure water as shown in the example of Patent Document 2, the gold wire taken out from the spool is three-dimensional in the first place. It is difficult to float on a two-dimensional horizontal plane. Even if such a gold wire can be floated, the gold wire is attracted to the side wall of the liquid holding container by the surface tension, and it is difficult to keep the gold wire in a predetermined direction and position with respect to the imaging device. It is not practical to apply to production sites. Moreover, since both ends are cut, the stress at the time of cutting appears as the amount of bending at the end of the gold wire (see Patent Document 2, paragraph 21), so whether the amount of bending at the time of cutting is being measured? It is difficult to evaluate whether or not straightness is measured.
As described above, the measurement method and the evaluation method disclosed in Patent Documents 1 and 2 cannot avoid errors that occur in the photographing conditions for measurement.

Moreover, even for a fine metal wire defined in Japanese Patent Application Laid-Open No. 2002-124534 (Patent Document 1), such an operation of feeding from a spool, cutting it to a fixed length, and performing bonding many times is performed. During the process, the product may be deformed into an unexpected shape and become defective. In particular, in the case of a bonding wire used in a semiconductor device, a single semiconductor device with hundreds to thousands of connection locations will be fatal because one failure will make all other connection locations unusable. It was a flaw. Such a defect is called wire bending or leaning and occurs even with a thin metal wire with a small meandering width. Could not be determined.
Specifically, the leaning failure means that a bonding ball is bonded to an Al pad of a semiconductor chip to form a first bond, and then a bonding wire is drawn out by a capillary to draw a loop. This is an abnormal inclination of the bonding wire that occurs in the vicinity. This leaning defect occurs because the distance between adjacent wires is narrowed and too close. In recent years, the pitch interval has been rapidly changed from 100 μm to 35 μm, and the interval between the wires has been narrowed as the fine pitch has progressed. For this reason, when the distance between the wires is wide, a slight inclination is not determined to be a leaning failure, but as a result of the narrowing of the interval, it is increasingly determined that the leaning is defective.
However, in recent years, the degree of integration of integrated circuits has improved, and fine pitches and thinning of wires have progressed simultaneously. The pitch interval was rapidly reduced from 100 μm to 35 μm, and the wire diameter was reduced from 30 μm to 15 μm. As a result, leaning countermeasures have become a very important issue for these ultrafine wires.

The present invention has been made to solve the above-mentioned problems, and a practical and effective method for measuring the straightness of a thin metal wire, such as wire bonding, which is used by joining thin metal wires at a certain length. The purpose is to provide.
The present invention provides an effective straightness measurement method that can distinguish and reliably determine the meandering degree that causes leaning and the meandering degree that does not cause the meandering,
In measuring the degree of meandering of a thin metal wire in three dimensions, the diameter and meandering width of the fine metal wire are extremely small as absolute values. The purpose is to solve the problems that cannot be fully addressed.
It is another object of the present invention to provide a practical measurement device that achieves the above-described problems and is more inexpensive and accurate than conventional measurement devices.

The present invention hangs a thin metal wire , shoots from a direction perpendicular to the hanging direction of the thin metal wire , takes a photographed image as binary data, obtains a virtual center line from the image, and obtains a two-dimensional continuous meander curve. Converted to
Dividing the measurement region of the two-dimensional curve into a plurality of sections of 2 to 25 times the loop length in wire bonding, and evaluating the straightness of the curve for each section of the partitioned two-dimensional curve;
A method for measuring the straightness of a thin metal wire characterized by:
The above-mentioned multiple sections are 5 sections or more, and the meandering width is measured for each section , and the straightness of the curve is evaluated by using only half of the measured values to one-tenth in order of increasing meandering width. Is what
The straightness evaluation of the thin metal wire is performed by the distance between the parallel lines sandwiching the curve of the thin metal wire in the separation section,
When converting the captured image as binary data into a two-dimensional curve, the image is enlarged in the width direction, and the enlarged image is used to reduce errors when converting to a two-dimensional curve. ,
The fine metal wire has a wire diameter of 0.015 to 0.025 mm,
This is a method for measuring the straightness of a fine metal wire that maintains the continuity of the curve and improves the measurement accuracy by overlapping each section of the two-dimensional curve with an adjacent section.

Furthermore, the present invention is a device for measuring the straightness of a wire-bonded fine metal wire for wire bonding,
a) Guide ring for inserting and positioning a fine metal wire hanging from the spool b) Imaging means for photographing in a direction perpendicular to the hanging direction of the fine metal wire, and c) Enlarging the photographed two-dimensional meandering image only in the width direction and divides the two-dimensional curve obtained Searching for the center line in the section 2 to 25 times the loop length in the wire bonding, the parallel lines sandwiching the two-dimensional curves for each section of the divided 2-dimensional curve was An arithmetic processing device for subtracting the distance between the parallel lines,
The apparatus for measuring the straightness of a thin metal wire, characterized in that the illumination of the thin metal wire is a backlight illumination, and the straightness of the thin metal wire is reduced by reducing the influence of irregular reflection of illumination light. It is a measuring device.

In the present invention, by accurately evaluating a bend with a small radius of curvature that causes leaning when used for connection such as wire bonding, a thin metal wire that causes a leaning defect can be surely excluded, and a gentle bending that does not cause a leaning defect is achieved. It is possible to use a thin metal wire having a large radius of curvature.
For this reason, the request | requirement of the higher degree of straightness improvement imposed on the metal wire for wire bondings with remarkable thinning and fine pitch can be responded effectively.
In addition, the high definition required for measurement is ensured, and the burden of imaging equipment and image analysis is reduced, enabling the use of inexpensive imaging equipment and the like, reducing the equipment burden and facilitating their handling. .

In the present invention, on the basis of the loop length in wire bonding, which is the cutting length when using the straightness evaluation of a thin metal wire, a two-dimensional curve obtained from the photographed image is a section of 2 to 25 times the length. This is done by measuring the degree of meandering.
According to the study by the present inventors, meandering that affects leaning as the straightness of a fine metal wire is not a so-called swell or loop with a large curvature but a bending with a small curvature radius.
This relationship is schematically shown in FIG.
When the form of bending of the thin metal wire in the length of the measurement range of the specimen is schematically represented, it is as shown in FIGS. 2 (a) to (d). (a) is an almost ideal form with a small meandering width. (B) is a so-called “swell” -shaped bend with a gentle meandering, (c) is a meandering appearance as a loop having a relatively large curvature, and (d) is a “bump-like” bend with a small curvature. It appears along with these “swells” and “loops”. In general, when the meandering width is measured for these curves, the difference in form does not appear in the measured values in the cases as shown in FIGS.

However, when bonding is performed with a loop length of not more than about 10 mm, such as a bonding wire, these “buzz” and (c) measured between several tens of centimeters are measured. It turns out that meandering that appears in a relatively long section such as a “loop” has little effect on straightness such as leaning.
On the other hand, the “bump-like” bend with a small radius of curvature as shown in (d) has a large change rate of the bend, so it has a significant effect when used in these short sections, especially with bonding wires. If such a bend exists at the joint portion, the meandering width is small, and even if the range is narrow, it appears immediately as a leaning failure, and the influence is great. That is, of these meandering curves, as shown in (d), a small curvature, a large degree of bending, and a bending appearing in a “hump” shape are important in the straightness evaluation.
However, according to the conventional method in which the measurement length is long and the meandering degree is measured from the entire length, the meandering width within the range and the meandering width due to the loop having a large curvature radius correspond to the period or the radius of curvature. Since the length is the maximum, meandering due to these undulations or loops having a large radius of curvature is detected as the maximum meandering width.
Therefore, a bend with a small radius of curvature that has the greatest influence on the leaning characteristics as described above is hidden behind these meanders having a large curvature and is not detected because the meander width is small.
In order to detect meandering with a small radius of curvature that affects leaning, etc., if measurement is performed by dividing into sections with a length corresponding to the curvature to be evaluated, meandering with a larger period or curvature is discarded and erased. Then, the meandering width corresponding to these small curvatures is detected.

FIG. 3 shows a conceptual diagram of a method for measuring the meandering property with respect to the two-dimensional meandering curve based on the photographed image of the specimen, following the meandering form of FIG. In the figure, each two-dimensional curve is divided into sections a1 to a3,..., D1 to d3, and these divided ranges are partially overlapped.
“Waviness” and “loop” are small, and a thin line with good straightness has a small meandering width measured even in a divided range, as shown in (a).
Further, by dividing the meander with a long span as shown in the curves (b) and (c) as shown in the figure, the amplitude of the “swell” and “loop” is eliminated according to the divided length. The The value becomes smaller as the division length is smaller than the period of the “swell” or “loop” or the radius of curvature. On the other hand, as shown in (d), a short meander with a short bending period and a small curvature radius is measured as a larger value as its division length becomes closer to the period or the curvature radius.

  The appropriate length and number of divisions are the diameter of the fine metal wire and the loop length of the bonding (the total length of the wire drawn out from the capillary. In other words, it is planar when the bonded wire is viewed from the vertical direction. Even if the length is the same, if the loop height is increased, the loop length will be longer. The same applies hereinafter.), Depending on how the loop is taken, such as the height of the loop, and the length of the specimen, For a defect, if each of the divided sections has a length in the range of 2 to 25 times the loop length when the thin metal wire is used, the value of the short meander with a small radius of curvature is obtained. It was found that the maximum meandering width was measured. Furthermore, when the length is in the range of 2 to 15 times, it is more preferable because it can be directly compared with practical conditions.

Furthermore, the measurement method of the present invention measures the meandering degree by dividing a two-dimensional curve into a length corresponding to the loop length of wire bonding as a two-dimensional curve obtained by taking a photographed image and binarizing by computer processing. Straightness is evaluated, but the required accuracy can be ensured according to the image processing capability of the imaging equipment by dividing the imaging section of the thin metal wire into a high-definition image necessary for the meandering degree measurement. .
As described above, an enlarged, high-definition image is necessary to capture a small bend in an almost straight line in a photographed image of an ultrathin line with a diameter of several tens of μm or less, which is an object of the present invention. However, it is difficult to obtain a photographic device having the number of pixels suitable for it, and the cost burden is large.
Thus, by capturing images that are appropriately divided and photographed in this way and processing the data, a high-definition image can be obtained, and the burden on these facilities can be greatly reduced.

Next, the photographed image is enlarged in the width direction.
The effect of lens aberration is small in the vicinity of the optical axis and appears along the radial direction, so the meandering width is essentially small, and in the image of the metal thin line taken along the radial direction through the vicinity of the lens optical axis, only the width direction Even if it is enlarged, the effect is small.
Further, by enlarging only in the width direction, the meandering width of the two-dimensional curve is expanded and the measurement becomes easy.
The image data obtained in this way is taken and binarized into a two-dimensional curve.
The ultra-thin line that is the subject of the present invention tends to make the contour of the photographed image unclear due to light diffraction, diffuse reflection, or the like. By increasing the width in the width direction so that the diameter of the wire is greater than or equal to a certain value, it is possible to reduce the error rate when converting to a two-dimensional curve.

The three-dimensional deformation of the fine metal wire is caused by various histories at the time of processing such as wire drawing and heat treatment, and these are intertwined in a complicated manner, so that they are discontinuous and only a part of them can be grasped in a short section. However, when viewed over a certain length, almost all of these deformations appear in various combinations or by changing the direction in the longitudinal direction.
Therefore, it is possible to grasp almost all of the meandering characteristics appearing on the fine metal wires by combining the divided sections of the photographed image to have a length satisfying these conditions.
The length of the specimen satisfying such a meandering condition is specifically determined by the degree of undulation of the thin metal wire wound from the spool, but empirically, the length is determined by the wire diameter. Although it depends, about 30-50cm is sufficient.
These divided sections preferably overlap each other. This is because the respective sections can be interconnected at overlapping portions. An overlapping range of 20 to 70% is practical.
It is preferable that the metal fine wire to be measured has a wire diameter of 0.007 to 0.1 mm, particularly 0.015 to 0.025 mm. This is because when the wire diameter is thicker than 0.1 mm, the wire itself has rigidity, so that the defects of the wire that are present at the time of joining are hidden by the rigidity of the wire. Conversely, if the thickness is too thin than 0.007 mm, it is difficult to handle the extra fine wire.

Hereinafter, the measurement method and measurement apparatus according to the present invention will be described in more detail with reference to the drawings.
FIG. 1 shows an embodiment of the measuring apparatus of the present invention. In FIG. 1, reference numeral 7 denotes a specimen metal fine wire suspended from the measuring length, 12 denotes an imaging device, 13 denotes an image processing device, 14 denotes a monitor, 10 is a windshield housing for preventing the influence of airflow, 11 is an observation window, 3 is a spool wound with a fine metal wire, and 18 is a guide ring.

<Metallic wire support means>
The spool 3 around which the fine metal wire is wound is rotatably supported by the upper portion of the housing, and the fine metal wire is drawn out to a predetermined length and hangs down.
In this state, these fine metal wires with extremely small wire diameters fluctuate due to the influence of the air current, so they hang down in the windshield housing that surrounds the metal wires and take pictures through the observation window. By hanging down through the ring, it is possible to prevent the line from being shaken and to accurately position the position at the center of the field of view of the photographing apparatus.
The guide ring is preferably coated with chromium oxide, titanium nitride, tungsten carbide or the like in order to improve the slipperiness of the fine metal wire, and is formed in a two- to three-fold coil shape.

<Lighting means>
As the imaging condition, in order to capture a captured image of a fine metal wire clearly, a plane such as a black board with a high contrast is arranged in the background, and an image is taken as a two-dimensional image on an imaging surface parallel to the plane. At this time, it is necessary to directly illuminate the thin metal wires from multiple directions so as not to cause an intermittent image due to shadows or reflected light.
In the conventional illumination method, since the reflected light from the thin metal wire 7 illuminated by the light source is observed, the diffuseness of the fine metal wire 7 varies due to irregular reflection of the fine metal wire 7, and the image of the entire thin metal wire 7 is captured uniformly. It was difficult.
On the other hand, when the backlight illumination is taken as a background, only the fine metal wire 7 is observed as a dark silhouette because the fine metal wire 7 directly blocks light. Therefore, even if the thin metal wire 7 has a small curvature, the thin metal wire 7 is observed to be almost uniformly dark because there is no diffuse reflection although it is affected by the diffracted light of the backlight light.

<Imaging device>
A commercially available digital camera can be used as the two-dimensional imaging device 12. Examples of digital cameras include “3D Digital Finescope VC4500” manufactured by OMRON Corporation, “Digital HD Microscope VH-7000” manufactured by Keyence Corporation, “Penguin 600CL” manufactured by Pixela, and “5.0RTV manufactured by Micro Publisher”. Or “Eos” series (Eos Kiss) manufactured by Canan. Since the distal end portion of the thin metal wire 7 is deformed by cutting, it is preferable to photograph the fine metal wire at a position spaced apart from the distal end as necessary.

  As described above, the range to be measured is obtained by dividing the entire thin metal wire into a plurality of sections according to the definition of the imaging device. By making the divided sections overlap each other, the sections can be interconnected at the overlapping portions to have continuity. An overlapping range of 20 to 70% is practical.

<Image processing device>
The image taken by the imaging device 12 is enlarged by the image processing device 13 while maintaining the magnification in the longitudinal direction of the fine metal wires 7, preferably 3 to 30 times, for example, 10 times. The range and magnification of the fine metal wire 7 can be enlarged by appropriately selecting the fine metal wire drawn to, for example, 300 mm to 1000 mm within an arbitrary range of 5 mm to 50 mm while observing the monitor device 14.

  Perform binarization of the enlarged image. In the binarization method, the discontinuous image appearing due to light diffraction or irregular reflection as well as the contrast between the thin metal wire and the background is omitted, and the wire diameter is 7 μm. It was possible to measure even a fine metal wire. In addition, when the binarized data cannot be obtained continuously because the boundary between the fine metal wire and the background is not clear, the blank portion is connected with a straight line or a regression curve to minimize the error in binarization. A technique for obtaining these continuous curves is widely known as a so-called digital image data processing method, and commercially available software may be adopted as appropriate, so that the description of the technique is omitted here.

  Next, a virtual center curve is obtained from the binarized image by computer processing. Even with binary data where the boundary between the outer periphery of the metal thin line and the background is not clear, the image of the metal thin line with the magnification in the horizontal direction is expressed by a thick curve, so the influence of the outer periphery (such as A high-precision and smooth central curve could be obtained without any discontinuity. This curve is a virtual meandering curve. On the other hand, when the magnification in the horizontal direction is not enlarged, the influence of the discontinuity of the outer peripheral portion appears greatly in a virtual meandering curve, and thus an error in binarization cannot be removed. For this reason, an expensive lens with a deep focal depth and expensive software capable of complicated analysis calculation are required.

  From the obtained virtual meander curve, the degree of meandering can be measured using parameters such as the expanded curvature and number of meanders, the expanded width of the meander and the expanded meander width, and the like. It was possible to contrast the defect of the fine metal wire connected by actual wire bonding, for example, the defect caused by the straightness of the fine metal wire such as fine wire bending or leaning in the connection wire for the semiconductor device.

The metal thin wire 7 to be measured has a length of about 100 cm and a wire diameter of 0.015 to 0.025 mm. The appropriate length to be measured varies depending on the wire diameter. However, in order to accurately measure the above-described three-dimensional change, it is preferable to set the length as described above when the wire diameter is large. When the wire diameter is small, the aspect ratio with respect to the wire diameter is as large as about 3 × 10 ^{4 to} 3 × 10 ⁵ even if the measurement length of the thin metal wire 7 is about 30 cm, and the above meandering characteristics can be sufficiently grasped. be able to.

FIG. 4 is a schematic diagram for obtaining a meandering width for each section obtained by taking a photographed image of a drawn metal thin line into a two-dimensional curve and dividing the two-dimensional curve into predetermined measurement section lengths.
These separation sections C1 to C16 are overlapped in order to maintain continuity and improve measurement accuracy.
Various methods for measuring these meandering widths have been proposed. For example, the method shown in FIGS. 5A and 5B is generally used.
(A) First, a line connecting the upper and lower ends of the wire at the delimiter position of the delimiter is drawn as a center line, two straight lines parallel to the center line are drawn, a meandering curve is sandwiched, and the distance between the two is obtained.
(b) Alternatively, the center line in the section may be obtained from the meandering curve of the section by the least square method, and two parallel lines sandwiching the meandering curve may be drawn in the same manner.
Next, these operations are performed for each section, and those values are compared to obtain the maximum value.

An Au wire bonding wire having a wire diameter of 25 μm and a purity of 99.99% by mass or more was used as a sample wire, the measurement range was 30 cm in total length, and the two-dimensional curve obtained as described above had a meandering measurement range of 0.5 cm, When the relationship between the meandering degree and the leaning (the loop shape has a loop length of 2.56 mm) measured for each of these separation sections was obtained by dividing the data into 2 cm, 5 cm, and 15 cm, the data shown in Table 1 was obtained. .
Here, for the meandering degree, a center line is obtained by the least square method within each section, and two parallel lines are drawn so that the width of two straight lines parallel to the center line and sandwiching the curve is minimized. The interval between the parallel lines and the meandering width were defined as the meandering degree. The unit is mm.

＊（株）新川製のボンダー（UTC−４００）を用い、ピッチ間隔５３μｍで、 １,５００回のワイヤボンディングしたとき、隣接ワイヤが接触したワイヤ の本数を数えた。 In the meandering measurement of the entire two-dimensional curve, in order to avoid errors due to the cutting of the wire, the measurement area was fixed at 30 cm from the lower part of the wire at a distance of 10 cm, and the overlapping section was 50%. The meandering degree is obtained by determining an arbitrary section length within the measurement region and rounding the meandering widths of the plurality of obtained sections by an average value or the like. However, if the delimiter section is set short, the total number of data increases, and if there are several large meanders even though the number is small, the average value processing makes it inconspicuous. Therefore, based on the following calculation formula, the meandering degree was determined from the total data by an average value using only 1/3 in descending order of the meandering width. In addition to this, a plurality of methods for rounding the meandering degree are conceivable, such as taking into account variations, and may be appropriately adopted as necessary.
That is, for the measurement area (cm): L and the separation interval (cm): K,
If the overlapping length is x, the overlap number in the measurement region L is n−1.
L = nk− (n−1) x where 50% is taken as the overlap length,
L = nk− (n−1) 0.5k = k / 2 (n + 1),
Solving this, as the number of sections in the measurement area,
Total number of data (pieces): n = (L / K) + ((L-K) / K)
If the top 1/3 is adopted from these data, the result is as follows.
Number of valid data (pieces): {(L / K) + ((L-K) / K)} / 3 (pieces) in descending order of meandering width
(Example) Therefore, according to the above example, from L = 30cm, K = 0.5cm,
Total number of data (pieces): = (30 / 0.5) + ((30-0.5) /0.5) = 119 (pieces)
Number of valid data: 40 in descending order of meandering width.
In this example, this measurement was repeated 10 times, and five values from the higher meandering degree were adopted and the average values are shown in Table 1.

* Using a bonder (UTC-400) manufactured by Shinkawa Co., Ltd., the number of wires in contact with adjacent wires was counted when wire bonding was performed 1,500 times at a pitch interval of 53 μm.

The relationship of Table 1 is displayed as a graph in FIGS. 6-1 and 6-2 as the degree of meandering and the number of leaning contacts / line for each segment length.
That is, there is a certain relationship between the degree of meandering and the number of leaning, and the meandering measurement section interval is 0.5, 2, 5, and 15 cm. It can be seen that there is a strong correlation with the leaning failure in the vicinity of the loop length measurement section of 2.56 mm. Table 2 shows the results of obtaining the correlation coefficient in order to quantify this.

That is, it was confirmed that there is a correlation between the meandering degree measurement section length and the leaning, and that the tendency of correlation changes by changing the section.
Here, the correlation coefficient (r) is, where the leaning number is x and the meandering degree is y.
r = Sxy / (Sx · Sy) ^1/2
It is represented by
However, Sx = Σxi ² − (Σxi) ² / n,
Sv = Σyi ² − (Σvi) ² / n,
Sxy = .SIGMA.xivi-(. SIGMA.xi) (. SIGMA.yi) / n.

In addition, the total number of data in the 0.5 cm section was 119 as described above, and 29, 11 and 3 at 2 cm, 5 cm and 15 cm, respectively.
From this result, when the correlation function is 0.7 or more, ０．６, 0.6 to 0.7 is indicated by ○, 0.5 to 0.6 is indicated by △, and less than 0.5 is indicated by-, the separation interval of 2 cm is indicated by ◎. The relationship between the degree of meandering and the leaning is best expressed, and it can be seen that the 0.5 cm and 5 cm separation sections become ◯ and the correlation is good.

From Table 2, a strong correlation was observed between the meandering degree and the leaning when the length of the meandering section was 2 cm when the wire diameter was 25 μm and the loop length was 2.67 mm. On the other hand, since the wire diameter and loop length affect meandering and leaning, the optimum meandering section is expected to change depending on the wire diameter and loop length.
As the wire diameter increases, the radius of curvature of the wire meander increases, so the average period of the wire meander increases. Therefore, it is better to set the length of the meandering section longer. Further, when the loop length is shortened, the influence of the meander having a short period on the leaning is increased. For this reason, it is better to set the length of the meandering degree section shorter. In order to clarify these relationships, as shown in Table 3, evaluation was performed using 32 spools of samples each having a combination of the wire diameter and the loop length and changing to 4 levels.
First, the meandering measurement range was fixed at 30 cm for each of the four levels of 32 spools, and the range of the obtained quadratic curve was divided into 0.5 cm, 2 cm, 5 cm, 15 cm, and 30 cm, and according to each division section The degree of meandering was measured.
Next, leaning was measured for 32 spools of 4 levels each. Finally, the correlation between the degree of meandering and leaning was obtained, and the results are summarized in Table 3 based on the following criteria.
In addition, it was set as correlation coefficient: (double-circle)> = 0.7, 0.7> (circle) ≧ 0.6, 0.6> (triangle | delta)> 0.4, 0.4> =-.

From the above results, the leaning does not depend on the wire diameter of the bonding wire of 15 μm to 25 μm, and as the meandering degree increases, the leaning defect increases. However, the degree of leaning defect depends on the length of the wire used, that is, the loop length. It can be seen that the quality of the leaning is judged most appropriately and accurately by evaluating the meandering degree in the section between 2 and 25 times the loop length, and the measurement accuracy is improved.
Therefore, before using a thin metal wire that has been drawn as a bonding wire, a captured image of the thin metal wire is taken and evaluated as a two-dimensional curve. By determining the meandering degree for each section and evaluating the straightness, it is possible to easily and quickly evaluate the goodness / badness of the straightness appropriately.

It is a lineblock diagram showing typically the whole measuring device of the present invention. Four types of representative meander curves of drawn metal wires are shown. The relationship between the meandering curve and the degree of meandering of a thin metal wire is shown. The method for determining the meandering width of a fine metal wire in the present invention will be described. It is explanatory drawing which shows the meandering measuring method of a metal fine wire. The figure which shows the correlation with the meandering degree and leaning defect in meandering measurement area 0.5cm (A) and 2cm (B). The figure which shows the correlation with the meandering degree and leaning defect in meandering measurement area 5cm (A) and 15cm (B).

Explanation of symbols

DESCRIPTION OF SYMBOLS 3 ... Metal wire spool, 7 ... Metal wire, 10 ... Windshield housing, 11 ... Measuring window, 12 ... Imaging device, 13 ... Image processing device, 14 ... Monitor, 18 ... Guide ring

Claims (8)

A metal thin line is suspended and photographed from a direction perpendicular to the direction in which the metal thin line is suspended, a captured image is taken as binary data, a virtual center line is obtained from the image and converted into a two-dimensional continuous meandering curve,
The measurement region of the two-dimensional curve is divided into a plurality of sections of 2 to 25 times the loop length in wire bonding, and the straightness of the curve is evaluated for each section of the divided two-dimensional curve, To measure straightness of thin metal wires. A metal thin line is suspended and photographed from a direction perpendicular to the direction in which the metal thin line is suspended, a captured image is taken as binary data, a virtual center line is obtained from the image and converted into a two-dimensional continuous meandering curve,
The measurement area of the two-dimensional curve is divided into five or more sections in a section 2 to 25 times the loop length in wire bonding,
Measuring the meandering width for each section , and evaluating the straightness of the curve by selecting and using the obtained measured values in order of magnitude from half to one- tenth of the number of sections. To measure straightness of thin metal wires. 3. The method for measuring the straightness of a thin metal wire according to claim 1, wherein the straightness of the thin metal wire is evaluated by a distance between parallel lines sandwiching a curve of the thin metal wire in the section. In converting a two-dimensional curve and processing said photographic image as binary data, and enlargement processing image in the width direction, according to claim 1, which comprises using the expanded image To measure straightness of thin metal wires. Method of measuring the straightness of the fine metal wire according to claim 1, wherein said metal thin wire is a wire diameter of 0.015～0.025Mm. 6. The method for measuring the straightness of a thin metal wire according to claim 1 , wherein each section dividing each two-dimensional curve overlaps with an adjacent section. A device for measuring the straightness of a fine wire for wire bonding that has been drawn,
a) Guide ring for inserting and positioning a thin metal wire suspended from the spool b) Imaging means for photographing in a direction perpendicular to the direction of the metal fine wire, and c) Enlarging the photographed two-dimensional meander image only in the width direction Then, the two-dimensional curve obtained by obtaining the center line is divided into sections of 2 to 25 times the loop length in wire bonding,
An arithmetic processing unit for drawing a parallel line sandwiching the two-dimensional curve for each section of the divided two-dimensional curve and obtaining a distance between the parallel lines ;
An apparatus for measuring the straightness of a thin metal wire, comprising: 8. The apparatus for measuring straightness of a fine metal wire according to claim 7, wherein the illumination with respect to the fine metal wire is backlight illumination.

Images