Quick Acoustic Methods for Part Verification

Acoustic microscopes traditionally look for delaminations and similar internal defects in components. Nowadays, they are being used as quick material characterization tools to verify incoming parts as a matter of supply chain management — a type of verification that is critical to any assembler, but especially to fabless companies.

The manufacturer of an IC generally makes the assumption that the packaging process coming from suppliers or their own factories meets specifications. In cases where this assumption carries some risk, engineers may use acoustic microscopes to investigate incoming plastic-packaged ICs for structural defects, primarily delaminations, cracks and the like. Where the economic impact of such defects is fairly modest, a low percentage of incoming parts may be imaged acoustically; where the potential economic impact is very high (as with high-end microprocessors), all incoming parts may be subject to high-resolution acoustic imaging inspection.

Used for the screening of incoming parts, acoustic imaging is essentially a kind of reliability or quality control function, and has become increasingly common during the past decade or so. More recently, however, engineers have quietly begun using acoustic microscopes to identify reliability-threatening features quite different from the conventional gap-type structural defects (Table 1).

Focusing on molding compounds

Many of the concerns that engineers have center on the molding compound and, in particular, on whether the molding compound has been changed since the last delivery of the part. One of the functions of an acoustic microscope is the measurement of the acoustic impedance of a material. In the case of a molding compound, the quick measurement of the molding compound's acoustic impedance gives the engineer a strong, though not absolutely conclusive, indication whether the molding compound has been changed or not.

The acoustic impedance of a material is found by multiplying two values: the material's density and acoustic velocity. Acoustic impedance is stated in units known as MegaRayls and, for particle-filled epoxy molding compounds, the acoustic impedance is usually between 3 and 7 MegaRayls.

While scanning to make an image, the ultrasonic transducer of an acoustic microscope usually interrogates thousands of X-Y coordinates. But for determining the acoustic impedance of the molding compound, only a single X-Y coordinate is needed — the equivalent of a single pixel in an acoustic image. The molding compound, though, is a mixture of epoxy and filler particles, and its acoustic impedance is influenced by the distribution of the filler particles. Often, particle distribution is not homogeneous, which can lead to differential stresses in the part, and an accurate acoustic impedance distribution can be obtained by taking measurements at several locations across the part and averaging them.

If the engineer obtains the anticipated value of 4.5 MegaRayls from the incoming parts, he can be almost certain that the molding compound has not been changed. Since there is some natural expected variation in acoustic impedance values of molding compounds, a reading of 4.4-4.6 MegaRayls would still indicate an unchanged molding compound. There is one small caution. In theory, it is possible for two different molding compounds to have acoustic velocities and densities whose products are both equal to 4.5. The engineer can determine the acoustic velocity separately by measuring the time required for a pulse to travel through a given thickness of the molding compound. Dividing the known acoustic impedance by the acoustic velocity gives the density. Knowing the acoustic velocity and density separately can help to separate two molding compounds that happen to have very similar acoustic impedance values.

Measuring the attenuation coefficient

A second molding compound characteristic that the engineer can examine is the attenuation coefficient — the degree to which a particular molding compound attenuates or absorbs the ultrasonic pulse passing through it. Attenuation is measured in decibels per millimeter (dB/mm). If the molding compound were, for example, 2 in. thick, all of the ultrasound would probably be absorbed before the pulse reached the bottom of the sample.

The attenuation coefficient will vary from one molding compound to another, and is independent of the density, acoustic velocity and acoustic impedance. To some extent, the attenuation of a particular molding compound is determined by the distribution of filler particles and voids and the presence of low-stress additives.

When an ultrasonic pulse is introduced into a part under investigation, the pulse travels downward through the molding compound, losing some of its energy because of scattering by the filler particles and absorption by the epoxy. It then strikes a material interface, which might be the interface between the molding compound and die face.

The problem in measuring attenuation is to separate interface reflection losses from attenuation losses caused by passage through the molding compound. One way to solve this problem is by grinding away a known percentage of the thickness of the molding compound over part of an internal interface. If the molding compound above the die face is 1 mm thick, for example, 0.5 mm of the molding might be ground away above part of the die face. Pulses are then inserted to measure the ultrasonic attenuation at both thicknesses. Since the loss caused by reflection is the same at both locations, the actual attenuation of the molding compound, in dB/mm, can easily be calculated.

The degree of attenuation in a given material varies from one acoustic frequency to another. Thus, an engineer may measure the attenuation of a given incoming part's molding compound (without grinding) using three different ultrasonic transducers, such as 30, 50 and 100 MHz. The three values will create a distinctive attenuation profile for that molding compound and provide another quick method for determining independently whether the molding compound is the same as in previous lots.

A recent advance can make use of multiple transducers unnecessary in measuring attenuation. An ultrasonic pulse from, for example, a 30 MHz transducer actually includes a range of frequencies on both sides of 30 MHz. By using fast Fourier transforms (FFTs), this pulse can be decomposed into numerous (15, 20 or more) individual frequencies, each of which will display its own attenuation characteristic. The resulting profile, since it contains many more individual values, is even more distinctive than the attenuation profile obtained by using multiple transducers. Profiles of three molding compounds are shown in the Figure.

Measuring the attenuation of ultrasound across a range of frequencies gives a profile curve that can distinguish a molding compound from other molding compounds.

The following technique is often used in order to maximize the value of attenuation data. Suppose that the sample is considered as a block of homogeneous material. First, an echo is collected from the top surface of the sample — actually, the interface between the sample and coupling fluid — and decomposed into individual frequencies by FFT. Suppose that the echo has been decomposed into 20 individual frequencies; the engineer will see 20 attenuation coefficients, and will notice that attenuation is more pronounced at the higher frequencies in the array.

Next, an echo is bounced off of the back surface of the sample, and this echo is likewise decomposed into individual frequencies. When the two sets of attenuation coefficients are compared, the differences between them show the variation of attenuation with ultrasonic frequency and amount to a very distinctive attenuation "fingerprint" of the molding compound.

There are limitations to the characterizations that can be made — at least quickly — with acoustic microscopes. For example, an acoustic microscope cannot generally determine whether the finish on the lead frame within a component package contains lead. The finish itself is much too thin to permit this characterization (although a delamination far thinner than the finish is imaged in sharp contrast).

The techniques described here have been developed in order to provide a fast way to ensure that the key material characteristics of incoming parts remain unchanged. For most incoming parts, these techniques involve nothing more than inserting a pulse into the part and letting software analyze the echo. Together with acoustic imaging to visualize internal features and look for internal defects, these techniques can create confidence that product reliability is being maintained.