Heat-Blocking Features Imaged by Ultrasound

Tom Adams is a consultant at Sonoscan, Inc., USA

3792
Advertisement

When power is applied, an integrated circuit begins to heat up within its package. It shortly reaches the operating temperature range for which it was designed. Excess heat begins to dissipate along a path designed for the die. If there are no structural anomalies along this path, heat dissipation from the IC will be uneventful.

For a low-power IC packaged in plastic and mounted to a printed wiring board, the dissipation path may be very simple: heat flows downward through the die attach and further dissipates through the metal die paddle. From here it makes its way through the backside mould compound. If the level of heat dissipated is low, heat-blocking anomalies along this path (die attach voids and non-bonds, for example) may have little effect, unless these cover a significant area of the dissipation path. But for higher-power dies, the absence of structural anomalies along the path is more critical and the path may terminate in a metal heat-sink designed to dissipate the higher heat flow.

The task for assemblers is to ensure that every die has a clear heat dissipation path. This means paying attention to die attach voids and surface contamination that can lead to non-bonds. For example, in plastic IC packages gaps may also occur in the bond between the backside of the die paddle and the mould compound. In insulated-gate bipolar transistor (IGBT) modules, even if no voids or other gaps are present, heat dissipation can be impaired by the tilting or warping of one or more interfaces along the dissipation path.

Die failures from imperfect heat dissipation can, to some degree, be anticipated by destructive physical analysis of field failures and of assemblies that have been life tested. These results may point to a process step that needs to be altered, or to a material that needs to be modified. A more direct method for finding structural anomalies is an acoustic micro imaging tool, which non-destructively images and analyses internal structural features, including voids, non-bonds, tilting and warping.

Fig. 1: Ultrasound launched into a sample is reflected not at all by a homogeneous material (left), moderately by interfaces between solids (centre) and almost completely by a solid-to-air interface (right)

Fig. 1: Ultrasound launched into a sample is reflected not at all by a homogeneous material (left), moderately by interfaces between solids (centre) and almost completely by a solid-to-air interface (right)

There are tools available that use an ultrasonic transducer to scan back and forth just above the plastic-encapsulated IC, IGBT module or other sample. The transducer is coupled to the top surface of the sample by a column of water that travels with the transducer; ultrasound at such high frequencies does not travel through the air. While moving, the transducer passes over several thousand x-y locations each second. At each location the transducer launches a pulse of ultrasound, and records the echoes sent back by material interfaces at various depths within the sample.

Ultrasound pulsed into the sample is reflected only by material interfaces. Gates are set by the operator to define the depth range from which echoes will be accepted for imaging. Put a block of homogeneous silicon onto the stage with a gate that includes virtually all of the silicon but neither the top or bottom surface, and the sample will send back no return echo signals at all, as seen at the left of Fig. 1, and the acoustic image will be solid black. If the pulse has been set to have somewhat wider gate that includes both the top and bottom surfaces (interfaces) of the silicon block, these two interfaces will be imaged. In either case, some of the ultrasound will be absorbed and scattered as it passes through the silicon.

The amplitude of any return echo signal depends on the properties of the two materials at a given interface. The amplitude of an echo from the interface between two solids can range from just above a small percentage of the pulse to above 90 per cent, as seen in the centre of Fig. 1. But the amplitude of a return echo from any solid-to-air interface is invariably very nearly 100 per cent, as seen at right in Fig. 1. Even if the gap containing the air or another gas is as thin as a small fraction of a micron, the amplitude will be the same.

The time of flight of the return echo signals can also be recorded and used to measure the distance from a reference point to the interface from which the pulse is being reflected at a given x-y coordinate. In this way a buried contour such as a warped layer can be measured and imaged.

Fig. 2 shows the acoustic image of bonding of a heat-sink. The transducer scanned the area of the heat-sink and collected return echo signals from the gate depth, which encompassed the solder layer bonding the heat-sink to its substrate as well as the two interfaces (solder to heat-sink, and solder to substrate). Gray regions in the acoustic image represent partial reflection of the ultrasonic pulse from one or both of these interfaces.

Advertisement


Enter your email address:

Delivered by FeedBurner

SHARE YOUR THOUGHTS & COMMENTS

Please enter your comment!
Please enter your name here