Wednesday, May 23, 2012 | Steve Martell, Sonoscan Inc.
Electronic components that go into military and aerospace applications are often inspected more critically--and in some instances far more critically--than components destined for consumer or even industrial applications. A plastic encapsulated integrated circuit (IC) may be acceptable for use in a low-risk consumer product, such as a game controller, if it has no internal structural defect larger than a given size, while a similar IC to be mounted in a higher-risk military or aerospace system, such as a fly-by-wire controller, may be permitted to have only tiny anomalies or none at all.
Time-Domain Imaging (TDI)
Internal anomalies and defects are typically found by acoustic microscopes because ultrasound is both nondestructive and highly sensitive to material continuity and interfaces. An acoustic microscope employs a scanning transducer that pulses ultrasound into an x-y location and receives the return echoes from the depth of interest a few microseconds later. As it moves, the transducer performs its pulse-echo function thousands of times a second. Since only echoes that arrive within a time window matching the depth of interest are used to make the image, this is known as time-domain imaging. It is the most frequently used of several available modes of acoustic microscope imaging.
The highest amplitude echoes come from the interface between a solid and a gap, and generally indicate a defect such as a non-bond, void or crack. Solid-to-solid interfaces produce mid-level echoes, while the lack of an interface (i.e., a homogeneous material) produces no echo. In the traditional acoustic image, gap-defects are displayed as bright white, bonded solid-to-solid interfaces are shades of gray, and homogeneous materials are black.
After their solder bumps have been bonded to a substrate, flip chips are conventionally underfilled with a particle-filled polymer. This leaves the upper surface of the silicon available to be scanned by the ultrasonic transducer--a real advantage because ultrasound travels through silicon with little loss and can produce high-resolution acoustic images of the underfill layer, including the solder bump connections. After acoustic imaging, the flip chips that have passed acoustic inspection criteria may be overmolded, (as for example in a BGA package configuration) or they may be left without overmold.
The new packaging technique, known as molded underfill (MUF), performs underfilling and overmolding simultaneously. It has several advantages for assemblers of mil/aero systems. Overall, MUF is less costly because the package is formed in one step rather than two. Vacuum-assisted transfer molding pulls the molding compound fluid under the chip. As a result there is no fillet surrounding the chip, so chips can be more closely spaced on the board. The MUF material itself contains a higher percentage of particles, and those particles are considerably smaller to enable them to fill the tiny spaces under the die. The resulting cured underfill/overmold provides better coefficient of thermal expansion (CTE) matching with the chip and with the substrate. MUF is increasingly common in applications such as cell phones where a row of flip chips may be mounted on a strip-style substrate.
But because underfill and overmolding happen at the same time, MUF provides no opportunity to acoustically inspect the critical underfill environment through the bare silicon. Ultrasound must therefore be pulsed through the overmolding. As the transducer is scanning, it sends a pulse through the fluid (typically water) that couples the transducer to the sample. The pulse must then travel through the overmolding, through the silicon, and into the underfill layer from which it will send back the echoes that will be converted into acoustic data and images.