Combining the methods reveals not only the anomaly, but the extent of 3-D distortion.

Plastic-encapsulated microcircuit (PEM) warping is of interest to manufacturers for several reasons. First, component warping may create internal stresses that, if eventually released, may result in electrical failure. For example, warping eventually may cause the silicon die to crack.

Second, warping may cause internal delaminations or cracks within the PEM. In this case, the stresses already have been at least partly relieved by the formation of a delamination or crack. Electrical failure might be immediate if the anomaly breaks a wire or a solder bump. But delaminations or cracks that slowly expand or promote corrosion may cause electrical failure in the future.

In some components, surface topography irregularities are not the result of warping, but are caused by upward pressure from internal anomalies such as voids.

All these scenarios make it worthwhile to collect data on both the surface topography of the component and its internal features, including possible anomalies. A new (patent pending) development in acoustic micro imaging makes it possible to map component surface topography at the same time that internal features are imaged. Ultrasound pulsed from the scanning transducer of an acoustic micro imaging system is reflected from material interfaces, but not from homogeneous bulk materials. The first reflection is from the interface between the water couplant and top surface of the PEM. The time-of-flight of this reflection is used only to measure distance, and as the transducer scans thousands or millions of x-y points, the data for mapping the surface topography are collected.

A portion of the ultrasonic pulse travels deeper into the PEM, where it is reflected from deeper material interfaces; these reflections are used to image internal features. From each of the x-y points, signal amplitude data are collected. The amplitude determines the brightness of a pixel in the acoustic image.

In operation, this method collects both surface topography and internal feature data several thousand times a second. The result is two images: one showing the internal images, and the other showing the surface topography. The user is therefore immediately able to relate these two sets of data. A PEM might, for example, have no internal anomalies, but might be so warped – as measured by the distance between its highest and lowest elevations – that it will be rejected because of the high likelihood of eventual electrical failure caused by the internal stress load. Or a PEM might have no significant warping, but may show internal anomalies such as cracks, voids or delaminations. Other possible scenarios are useful. For example, the surface may show distortion, but the internal acoustic image may reveal no anomalies. In this case, the irregular surface topography might mean anomalies are present in the PEM, but at a depth lower than the depth from which return echoes were selected to make the acoustic image.

Simultaneous surface mapping and internal imaging is useful with many component types. Some are obvious candidates for this method: BGAs may be bowed; flip chips may be domed. But larger items such as multilayer printed wiring boards can also display surface warping and internal damage.

Figure 1 is an at-depth acoustic image of the interior of a plastic BGA. To make this image, return echo signals from the depth of interest – the interface between the mold compound and the top of the silicon die – were used. The high-amplitude white circle at the die center is a void (arrow), a small bubble of air trapped in the cured epoxy. Other voids are present at the die edges. Voids and delaminations between the die surface and mold compound are unacceptable because they may expand as a result of normal thermal cycling and break a wire bond.


Figure 2 is the topographic map of the surface of the same BGA, made by using the time-of-flight information (but not the amplitude or polarity) of the echoes from the top surface of the part. The color bar at left indicates changes in elevation: The lowest points are white, magenta or blue, and the highest points are green. No points on the surface of this BGA reach the yellow-red-brown region at the top of the color bar.


Away from the periphery, the lowest region on the surface of the BGA is the dark blue region near the part center. The conspicuous pale blue dot near the center is a higher point on the surface where the epoxy has been pushed up by the void between the die and epoxy. There are additional voids at the die edges, but they are too thin to have pushed up the surface, or – more likely – are indistinct in this image because they are at a distance from the center where the BGA is beginning to bow upward.

Once the map of surface topography has been made, differences in elevation can be measured. The pale blue surface “spike” caused by the delamination at the center of the die is 0.0008" (0.020 mm) higher than the dark blue region just southeast of it. The extent of bow in the whole BGA package can be seen by comparing the lowest point (dark blue) with the highest point (pale green near southeast corner): The difference is 0.0017" (0.043 mm). The “spike” is therefore about half as high as the highest points on the upwardly bowed edges of the BGA.

Differences in elevation can be seen more easily by using the data to make a 3-D image that permits controlled exaggeration of the differences in elevation. The 3-D topographic image also can be rotated and tilted to provide a desired perspective.

Figure 3 is the 3-D image of the plastic BGA package. The overall bowing of the package is very conspicuous in this image, and the surface “spike” caused by the void at the epoxy-die interface is clearly visible, and is enlarged in the inset. The color map at right shows the highest region is the red-brown region at the right end of the BGA.


Figure 4 is the optical image and Figure 5 the surface topography of a printed circuit board measuring 1.989" by 3.038". The wide range of colors shows the board is warped, most likely in response to heat. When heat is sufficient to warp a board, it may also cause internal delaminations among the board’s innerlayers. Depending on the board structure and thickness, internal board delaminations may be visible in depth-of-interest acoustic images.


In this surface topography image, white areas are the components whose tops are outside of the level being imaged. The highest regions (shortest time-of-flight) are the red-to-black areas along the left edge. The lowest regions are the magenta portions along the top edge. The difference in elevation between the highest and lowest points is approximately 0.020". Differences in elevation are more pronounced in the 3-D image (Figure 6).


The new combination of the two methods – acoustic surface flatness measurement and acoustic imaging of internal features – creates a direct link between these two sources of data. By itself, measuring surface topography may reveal a location that needs in-depth imaging for a potential anomaly. Made simultaneously, the two methods reveal not only the anomaly, but the extent of 3-D distortion the anomaly has caused.

Tom Adams is a consultant at Sonoscan Inc. (sonoscan.com).

• Prev
• Next