Medical Device Link.

Originally Published MEM Fall 2004

ELECTRONIC PACKAGING

Nondestructive imaging techniques enable electronic packaging faults to be discovered that routine electrical testing might not expose.

Jack Richtsmeier and Kristopher D. Staller

Manufacturers who need to ensure that their medical electronic systems will deliver high long-term operational reliability must find and eliminate from their products the causes of two broad classes of device-related electrical failures. The first class of failures consists of those that are clear electrical failures. These failures might be caused by an improper solder joint between the component and the board or they might be caused by a chip-level problem. The second class, which may or may not be evident when the product is still in the manufacturer's hands (and which therefore is generally more difficult to control) consists of various packaging anomalies that are capable of causing electrical failure at some point in the life of the electronic system.

The electrical failures in the first class should be caught by electrical testing conducted during or at the end of the assembly process. Although in-circuit testing, functional testing, and other electrical tests are not foolproof, when used efficiently such tests can identify the majority of existing electrical problems and can lead to the discovery of their root causes. The task is made easier by the fact that electrical failures in this class are events that have already occurred. The circumstances causing their occurrence would be unlikely to change in the future, which would not matter because the event would have already caused the device to fail.

Packaging anomalies, on the other hand, are internal mechanical flaws that usually have not resulted in an electrical failure by the time electrical testing takes place. They tend also to change or grow worse over time, with the result that the electrical failure may occur at some unknown future moment.

Internal packaging flaws almost invariably either involve bonding—or the lack of bonding—between materials within the package, or they cause the development of cracks and voids. For example, two solder bumps within a flip chip may bridge during reflow. Not only is there an improper bond in that case (solder to solder), but also, secondarily, the solder will be bonded to the chip face in a location where only underfill material should exist. Internal packaging anomalies are the delaminations, cracks, and voids that can cause future failures and that, occurring in sufficient quantity through faulty production process control, can result in large numbers of field failures.

Some packaging defects—a delamination in a noncritical location would be an example—pose little long-term threat to reliability if they do not grow over time. Still, they are worth discovering because of the light they throw on the effectiveness of process controls.

This article examines some typical packaging anomalies that have been found and characterized by means of three imaging methods: nondestructive microfocus x-ray imaging, acoustic microimaging, and physical sectioning followed by optical imaging. All of the flawed devices discussed were components of medical electronic systems.

Nondestructive Imaging Techniques

Microfocus x-ray and acoustic microimaging technologies involve different energy sources, but both methods are useful for nondestructively imaging internal package features. The roles they play in the imaging of electronic components and assemblies are largely complementary; there are relatively few areas of overlap.

Microfocus x-ray imaging is a through-transmission technique. For this technique, the target sample is placed between a radiation source and a detector. The image that is generated is determined by the absorption properties of the various materials the radiation passes through. These absorption properties, in turn, are determined by the atomic number, density, and thickness of each material. Success in imaging a given internal feature or structure thus depends on adjacent materials having different absorption properties. In some cases, tilting the sample is useful for removing from the image train a feature made of a highly radiation-absorbent material.

The acoustic microimaging technique pulses very-high-frequency or ultra-high-frequency ultrasound into the target sample from a transducer, which also serves as the detector. Acoustic microimaging is thus a reflection-mode technique, although in one mode it is also used as a through-transmission method. As the transducer moves across the target sample, it pulses ultrasound into the sample and collects the return echo signals at a rate of several thousand per second.

Ultrasound waves are reflected from material interfaces, but not from material in bulk. At an interface, how much of the ultrasound pulse is reflected depends on the difference in acoustic impedance between the two interfaced materials. The acoustic impedance of a material is the product of its density and its acoustic velocity.

The highest reflections come from interfaces in which one material is a gas and the other is a solid. Interfaces of this type are characteristic of delaminations, cracks, voids, and other gap-type features in electronic packages. Because gap-type interfaces reflect virtually all of the pulsed ultrasound wave, they are imaged in high contrast. Gap-type defects block ultrasound from reaching features deeper within the sample.

Flip-Chip Solder Bumps

Figure 1. An oblique-angle microfocus x-ray image of flip-chip solder bumps. The arrow points to an oddly shaped bump that is probably not in contact with the bond pad. (click to enlarge)

Figure 1 is a microfocus x-ray image of one row of solder bumps on a flip-chip device. Ceramic chip capacitors and other identifiable components appear in the image as well. The board was tilted to an oblique angle in order to enable the inspection of the solder bumps, which appear dark because of the higher radiation-absorption capacity of the solder.

One bump (indicated by the arrow) is shaped differently than the other bumps. It is very likely an open bump that is not making contact with the pad on the substrate. The area of this bump appears larger from this angle of view, suggesting that the mass of the solder did not wet the pad, but rather spread laterally. The other bumps in the row have a square base indicative of normal contact.

Figure 2. An acoustic image made through a silicon die of a row of flip-chip solder bumps. Bump #5 is disbonded from the board. It appears red because of the very high reflection of ultrasound caused by a gap. (click to enlarge)

Figure 2 is a highly magnified acoustic image of seven solder bumps at the edge of a flip chip. Ultrasound has been pulsed through the acoustically transparent bulk silicon of the die above the chip-to-bump interface. What is of interest is the degree of reflection at the bond between the bond pad and the solder bump. A good bond returns a modest signal, but a bond that contains an anomalous gap returns a far stronger signal because one element of the material interface is a gas. The return echoes have been gated on the interface between the solder bumps and the board, meaning that echoes from that depth only are used in making the acoustic image. The vertical lines are traces on the circuit board under the die.

Most of the solder-bump bonds in Figure 2 have returned the low-amplitude signals that characterize a good bond, in which the material interface that is reflecting ultrasound is the interface between the bond pad and the solder bump. Bump number 6 in the figure has a somewhat brighter image than most and might be considered questionable. But bump number 5 exhibits the very high reflectivity that corresponds to a gap-type defect. Its image indicates that the solder bump is not well bonded to its pad.

In terms of reliability, this interconnect in its present form might pass electrical tests; that is, there may be enough mechanical contact, despite the lack of bonding, to allow conductivity. But gap-type defects are notoriously capable of expanding during product use. If this solder bond has not already failed, it is likely to do so eventually.

Defects in Ceramic Chip Capacitors

Figure 3. An acoustic image of a surface-mounted ceramic chip capacitor. The two gray features within the body of the capacitor are the plates; the red-yellow internal feature is an internal gap-type defect that has high reflectivity to ultrasound. (click to enlarge)

Figure 3 shows an acoustic image of a ceramic chip capacitor surface-mounted to its board. The irregular outline around the capacitor is the solder that bonds it to the board. The multilayer plates within the body of the capacitor are visible as a mildly reflective gray region in the center and right portions of the capacitor. At the left, however, is a red-yellow area of very high reflection indicating the presence of a gap-type internal defect. The extent of the anomalous area suggested that the defect most likely consisted of an internal crack or delamination.

This suspect capacitor was removed from the board after imaging and sectioned to provide an optical view. One advantage of early acoustic imaging is that the acoustic image shows where to section the sample to ensure interception of the defect.

Figure 4. The same ceramic chip capacitor as shown in Figure 3 was removed from its board, sectioned, and viewed optically. The crack extends from the termination and follows the third and fourth plates. (click to enlarge)

The results of the sectioned defective ceramic chip capacitor appear in Figure 4. The anomaly was found to be a crack that extended from the termination and followed the third- and fourth-plate layers from the bottom of the capacitor. Even if a crack of such extent does not grow, it could, in the form shown in the optical image, cause a short between adjacent plates.

Component Adhesion within Implantable Devices

Frequently, adhesives are used to mount batteries, sensors, subassemblies, and components within the sealed housing or within the can of implantable medical devices. The integrity of these adhesives is obviously crucial for the reliability of such devices.

Acoustic microimaging can be used to inspect the bond between the can and the internal parts in these applications and to distinguish between a good bond and a faulty bond (or disbond). A good bond returns a modest echo because the ultrasound wave is being reflected from the interface between two solids. A disbond returns a much stronger echo because the two solids meant to be joined are separated by a gap. Recent research has shown that the ultrasound reflection at a gap is nearly total, even when the gap has a thickness of only 100–1000 Å.

Figure 5. The bonding of internal components and subassemblies to the shell of a medical device, viewed acoustically from the outside surface. The light region at the lower left is not bonded. (click to enlarge)

Figure 5 shows the acoustic image of an implantable device in which the internal parts are mounted with adhesives near the periphery. In the image, the two curved bands of adhesive at the right end of the device and the one curved band at the upper left show the modest acoustic amplitude (the darker reflection) that is characteristic of good bonds. The adhesive band at the lower left, however, appears lighter because of a higher amplitude of reflection. In this region, the internal parts are disbonded. The outline of the curved band indicates that epoxy was dispensed into the can but then delaminated in the middle of the bead.

Conclusion

Bond-related packaging anomalies can compromise the reliability of medical electronic systems. Many such packaging defects do not cause electrical failures that are detectable during production testing, but rather, these defects carry the potential for field failure later. Manufacturers need reliable methods to locate these anomalies before the product reaches the end-user. New nondestructive imaging technologies make this possible. These technologies also provide a means to assess assembly process controls.

Jack Richtsmeier is a business development manager at Sonoscan Inc. (Elk Grove Village, IL). Kristopher D. Staller is a senior failure analysis engineer in the Medtronic Microelectronics Center (Tempe, AZ).

Copyright ©2004 Medical Electronics Manufacturing