In Situ Moisture Diffusion and Swelling Characterization of Molding Compounds in PEMs

Alexander Teverovsky, Ph.D.
QSS Group, Inc./NASA

Moisture diffusion in epoxy molding compounds (MC) is considered as one of the major reliability concerns in PEMs. The rate of moisture ingress to the die surface and/or moisture release from the package depends on the package geometry and on the coefficient of moisture diffusion, D, of the packaging material. Given the values of D at different temperatures, the characteristic time of moisture diffusion, , can be calculated for different environmental conditions during the parts testing and/or field operation. The knowledge of allows for estimation of adequate bake-out regimes and provides a time scale for assessment of a reliability risk related to moisture content in PEMs.

Moisture sorption is known to cause swelling or hygrothermal expansion of MC in humid environments. This effect causes deformation of encapsulating materials and can significantly increase mechanical stresses in plastic packages thus affecting performance and reliability of sensitive microcircuits.

Manufacturers of commercial microcircuits usually do not provide any information on used encapsulating materials. Unfortunately, moisture diffusion and hygrothermal expansion coefficients for MCs are rare to find even in special technical literature in spite of their importance for reliability evaluation of the parts in moisture environments.

Extensive testing and qualification of commercial-off-the-shelf (COTS) parts performed recently in the military and aerospace community showed that performance of PEMs is strongly affected by the type of used molding compound. Lot-to-lot variations in MC and permanent improvement of technology and design of the parts make possible variation of properties of the encapsulating materials even for the same part type. This brings about a need for rapid assessment methods for moisture characterization of materials used in PEMs. The necessity of fast test methods for moisture characterization of MCs is considered as one of the priorities in several currently running projects aimed to reliability evaluation of COTS, e.g. the ROBOCOTS (robust packaging of COTS ICs) program started recently at Rockwell International.

Moisture characteristics of MCs provide valuable data for quality evaluation of the parts operating in moisture environments and for identification of design and technological changes. However, they are also important for reliability evaluation of parts intended for space applications. Based on temperature dependence of moisture diffusion coefficient and existing models of moisture-related degradation mechanisms, it is possible to estimate the risk associated with the exposure of the parts. This exposure consists of moisture environments during the pre-launch period and after the launch, when the system starts operating in space and moisture out-diffusion occurs due to exposure to vacuum.

Similar to hygrothermal expansion of encapsulating materials in humid environments, shrinkage of MC and related changes in mechanical stresses could be expected in dry and/or vacuum conditions. Obviously, the vacuum-related shrinkage of plastic packages can affect performance and reliability of PEMs in space applications.

The purpose of this work is to develop simple and rapid techniques for the estimation of moisture diffusion and hygrothermal expansion characteristics directly on plastic packages of PEMs.

It has been shown that the moisture diffusion coefficient and the activation energy of diffusion could be calculated based on results of thermogravimetry analysis (TGA) measurements which provide changes in weight of a sample as it is heated. The weight measurements are performed on a sample, which had been pre-saturated in moisture environments, while the temperature is linearly increased from 20° C to approximately 170° C. The duration of this test does not exceed a few hours and the necessary calculations can be easily performed in an Excel spreadsheet. It has been shown that in many cases instead of expensive TGA instrumentation, a simple balance with 0.1 mg accuracy and a regular thermal chamber, which could be programmed for a linear temperature increase, fast heating-up and cooling to room temperature, can be successfully used.

The hygrothermal expansion coefficient is calculated based on weight measurements of a plastic package in a free state and after immersion into a low molecular weight perfluoropolyether fluid (a version of hydrostatic weighing). These measurements are performed after saturation of the sample in moisture environments and then repeated after drying (e.g. a high temperature bake) thus allowing estimation of relative changes in mass and volume of the part. It has been shown that the data obtained by this technique are in good agreement with the results of the linear deformation measurements and literature data.

Several types of PEMs and encapsulating materials have been evaluated. The preliminary data show that the diffusion coefficient at 85°C varies from 1x10^-7 to 4x10^-7 sm²/s, the activation energy varies from 0.4 to 0.6 eV, and the volume hygrothermal expansion coefficient is in the range from 0.35 to 1.1. Details of the developed techniques and analysis of moisture characteristics of different encapsulating materials will be published in future IEEE Links issues.

(back to the top)