Structural Defects in a TO-220

Ultrasound and X-ray work together to image and evaluate harmless anomalies and potential field failures in a TO-220 package

by Tom Adams, consultant, Nordson SONOSCAN

Launched by an ultrasonic transducer, a pulse of ultrasound travels through the plastic package of an electronic component at a speed around 3,000 m/s. If it strikes a material interface, there are two possibilities: a) if the interface is be-tween two solids, a portion of the pulse will be reflected back to the transducer and another portion will cross the interface; or b) if the second material is air, the pulse is almost entirely reflected, and none crosses the interface.

The second possibility is useful because most of the internal structural anoma-lies and future field failures in a component result from cracks, delaminations, voids or other gaps containing air. A few defects that are imaged do not in-volve air - a tilted die, for example, or a missing solder bump, but both types of interfaces can be located and analyzed by ultrasound.

A TO-220 device like the one discussed in this article is designed to be mount-ed by its tab onto a heatsink in order to dissipate large amounts of heat. Inter-nally, wires run from the die, which is mounted on the substrate, upward to the electrical lead posts. The purpose of acoustic micro imaging of this TO-220, carried out by a C-SAM® tool from Nordson SONOSCAN, was to evaluate interfaces for structural defects: the die attach, the bonding of the wires to the die, and the bonding of the wires to the top of the post. A question that could not be resolved by ultrasound was handled by an X-ray tool.

For acoustic imaging, the TO-220 was first flipped over to image the die attach through the metal back side tab, then turned upright to pulse ultrasound through the encapsulate to image first the die face and second, at a higher plane, the top of the post to which the wires were attached.

At each site, the transducer scanned back and forth a few mm above the area of interest, each second launching thousands of ultrasonic pulses and receiving their return echoes. The echo from each x-y scanned location becomes one pixel in the acoustic image of that area.

From each returning echo the following information is gathered: the echo’s acoustic frequency, its amplitude, its polarity, and the depth from which it was reflected. In these images, the colors report amplitude and polarity.

Acoustic image made through the mold compound and showing the die at top and the two posts

An acoustic image of all three elements in the package is shown in Figure 1. The vertical distance between the die and the posts is too great to be within the same focus. Figure 1 is therefore made from parts of two acoustic images, one focused on each of the two depths within the package.

The right lead attached to the die at top has a very small red area, but the wire bond area is essentially intact. The red area at the lower left corner of the die, however, is a void that could grow with repeated thermal cycles and cause wire bonds to break. Whether this void is enough to cause rejection of the part de-pends on the application. It might be suitable in a commercial application, but perhaps not in a military application.

The two wires visible on the die clearly run to the post at the right. There is no indication in this view of wires running to the left post. In this case the wire has a very fine diameter and is too small to be resolved at the low ultrasonic frequency needed to penetrate the mold compound. There is, however, a faintly brighter area on the left post that somewhat resembles a bonded wire.

Die attach, as imaged through the bottom of the package

The back side of the die (Figure 2) was imaged through the metal substrate, to which the die is attached. The variously shaped small red features within the area of the die are voids - red here indicates that there is air in the voids. There are no truly dangerous features such as very large delaminations, or voids close to the corners of the die. The voids cover, in total, about four percent of the die area. The TO-220 package is designed to remove large amounts of heat swiftly, so these relatively small well-spaced voids in the die attach may pose little threat.

The problem area in this component, as can be seen in Figure 1, is the right lead post and the wires attached to its top. The red areas are likely voids, but their exact placement and relative danger must be sorted out. To do so, two different imaging modes were used.

The first was a mode that can, during a single scan, image multiple progres-sively deeper thin slices of the sample being viewed. At each of the thousands or millions of x-y locations in a given image, ultrasound may be reflected from a single interface, or from multiple interfaces at various depths. To avoid depth confusion, each image is limited to echoes arriving within a specific time that matches a vertical extent within the part. For the right lead post, a total of 6 gates was defined and imaged, each 600 microns in vertical extent. The echoes from each gate became a separate acoustic image.

The two wires attached to the right post. Red areas are anomalies

Of the six gates, only gates 3 and 4 revealed details of the two wires and their attachment to the post. In gate 3 (Figure 3). the wire at right is faintly visible. It has no anomalies at this depth except for a small void at its termination. The wire at left has a similar tiny void at its termination, but has a larger void along part of its length.

Same view, but in a slightly deeper gate

Gate 4 is seen in Figure 4. The wire at right is very faintly visible. The small void near its tip is visible at this depth. The white area at right is the top of the post to which the two wires are attached. The post extends to the left at least as far as the leftmost void, although it is hidden by overlying features for most of this distance. The large red feature at center is a void, and other voids are scattered about. Red in a delamination or void identifies an area that reflects nearly all of the arriving ultrasound. Red is at the very bottom (=strongest negative echo) in the color map at left. Where the side of a feature becomes steeper and reflects ultrasound less directly in the direction of the transducer, red changes to black. Even steeper sides reflect yellow, the next color on the map.

To get an enhanced view of the interior of this device, it was also imaged with the C-SAM’s nondestructive cross-sectioning feature called Q-BAM® (Quanti-tative B-scan Analysis Mode). A line was described on the acoustic image of the top of the package through which cross-sectioning would cut through the area of the two wires bonded to the lead post. The resulting acoustic cross-sectional image is the dimensionally accurate equivalent of a physical cross sectioning made by cutting down through the package along the same line.

If the sample were physically sawn open along the same line, the same features would be found in the same locations. A light photograph of the sectioned face would look much like the acoustic cross-section. The concept can also be used in a different manner: before physical sectioning, the sample may be imaged acoustically to spot the location through which the physical section will pro-vide the desired information.

Top portion is a non-destructive cross-section through the horizontal white line below the legend

The white horizontal line just below the center of Figure 5 is the line along which the transducer scanned this component in order to produce the acoustic cross-section in the top portion of the Figure. On its first pass along this line, the transducer accepted echoes only from a defined thin depth at the bottom of the package just above the post. The transducer was then raised very slightly and scanned back along the line. By the time it reached the top of the package it had collected the echoes needed to produce the acoustic cross-section seen in the top section of Figure 5.

In the sectional view, the surface of the package is marked by a horizontal red line. In the bottom half, the cross-sections of five features are shown running from left to right. Each feature lies directly above its location on the thicker white line in the top view below. From left to right, these features are:

1. a large air gap on the surface of the post.

2. the left wire and the void on top of it.

3. the large void to the left of the right wire.

4. largely the interface between the mold compound and the post.

5. a void at the top of the post itself.

There is a sixth, rather small feature, apparently a void, directly above

#1, just below the package surface.

An X-ray beam inserted into the side of the mold compound revealed the four wires

The acoustic images in Figures 1 through 5, however, do not reveal all the wires extending from the die to the posts. The acoustic images revealed two fairly large wires. Smaller diameter wires would be difficult to see because they scatter most ultrasound away from the transducer. In addition, a smaller diam-eter wire might be invisible acoustically at low frequencies. If ultrasound is en-tirely blocked by an air interface, the wires beneath the air gap will not be im-aged.

Subsequently a Dage Quadra™ 7 X-ray system was used to image the TO-220. An X-ray beam is not stopped or even attenuated by an air gap. Imaging was first attempted through the heat sink, but the thickness of the metal was too great. (Ultrasound would have penetrated the metal, but would still have been scattered rather than reflected by the round wires.) to get additional information. The TO-220 was therefore X-rayed from the side, a longer route through the much less attenuating mold compound. The results are seen in Figure 6.

Four wires are visible as vague dark lines marked by numbers in the X-ray im-age. The image colors have been inverted to make the wires more visible. Their precise orientation cannot be discerned, but it seems likely, taking into consideration the very faint features on the left post in Figure 1, that two wires run from the die to each of the two posts.