Environmental control for electronic packages and enclosures often involves
understanding and managing moisture concentration. Excessive moisture
combined with processing temperatures can cause damage. High humidity
levels have other potentially harmful effects such as corrosion. Since
the movement of heat and moisture are similar (i.e., diffusive), thermal
engineers may be asked to apply their expertise to this topic. The goal
of this article is to provide an overview of moisture permeation and to
provide some guidance for thermal engineers. Water movement across an
enclosure boundary can occur from both leakage and diffusion. If leakage
is present, it is most likely the dominant transfer mechanism but for
well-sealed systems, moisture diffusion through the boundary materials
is relevant. The similarity of Fick’s diffusion equation (1) to the heat
equation (2) gives an opportunity to use standard thermal tools to solve
moisture concentration problems.
(1)
C = Concentration, D = Moisture Diffusivity
(2)
T = Temperature, a = Thermal Diffusivity
However, applying thermal tools for computing concentration is not entirely
straightforward. While temperature is continuous across material boundaries,
moisture concentration may be discontinuous across material boundaries.
One approach for using thermal modeling tools with multiple materials
is to normalize the concentration of moisture in each material by its
saturation concentration and then treat the normalized fraction of concentration
as the unknown variable and force continuity at the boundaries [1]. Continuing
with the moisture-thermal analogy, the effective conductivity for each
material becomes the moisture diffusivity times the saturation concentration.
While detailed numerical analysis of moisture concentration gradients
is possible (and substantially more complicated than is possible to discuss
in a technical brief article), a frequent concern is simply to determine
the rate of moisture permeation through a material given a vapor pressure
difference on different sides of the material. This is similar to considering
a net heat flow through a material with air on both sides and relating
the heat flow to the air temperature difference (without a distinction
between the convective and conductive components).
The reader is reminded that it is the water vapor pressure difference
that matters for diffusion. Maintaining a positive total pressure (but
dry) condition inside of an enclosure will not prevent moisture from diffusing
in when the outside air contains a water vapor partial pressure higher
than the inside of the enclosure. The permeability of materials to moisture
varies many orders of magnitude as shown in Figure 1. The upper horizontal
scale lists the permeability while the lower horizontal axis lists the
approximate time scale for permeation. The diagonal lines show the effect
of thickness as shown on the left axis. The low permeability of metals
and glasses are the reason they are used as materials of choice for containers
that need to stay near hermetic for long time frames.
Figure 1. Range of permeability for materials [2]
Several engineering disciplines have an interest in understanding and
predicting moisture permeation, notably the food and building material
industries, and have published substantial data. Unfortunately, much of
this data is in various and often confusing forms and may not adequately
describe the test conditions necessary for accurate representation of
the data. ASTM E 96 recommends that measurements of permeance (g/cm2-hr-Pa)
and permeability (g/cmhr- Pa) be both reported [3]. Note that the permeability
as defined by ASTM E 96 (PE96) is expressed in terms of measurable
quantities as shown in equation 3 and is an intensive material property.
Accurate measurement of PE96 requires resolving small mass
changes and can take a long time. For materials with very low permeation
values, uncertainties in mass measurements caused by repeatability and
buoyancy effects can be particularly difficult to minimize.
(3)
A frequent form of data found in literature is in terms of a transfer
rate (WVTR for water vapor transfer rate or MVTR for moisture vapor transfer
rate). Usual units for WVTR are (g/m2-hr) but are sometimes
found with a thickness term in the numerator (gmm/ m2-hr),
which can give some indication of the material thickness used in the test.
Most likely, this data will not be found in SI units but in some combination
of days, area, and possibly even a volumetric gas quantity at standard
temperature and pressure. For this data to be useful, the test conditions
must be reported.
Note that PE96 for a material is equivalent to the diffusivity
times the solubility and is further equivalent to the diffusivity times
the Henry’s law constant when the moisture concentration is relatively
small. Relatively small means that the moisture concentration in the material
will respond linearly to the water vapor partial pressure in contact with
the material. In this form, it is convenient to express the solubility
and Henry’s law constant in mass of water per volume of material because
many materials tested are not homogenous and expressing these terms with
moles is difficult without a well defined molar mass. Because the solubility
of many materials is dependent on temperature and concentration, it is
crucial to know the test conditions that were used to generate the permeance
or water vapor transport data. This would be similar to needing to measure
thermal conductivity as a function of both heat flow and temperature.
As mentioned, permeation data can be found with a variety of units. For
example, the building industry often reports moisture permeance in units
of perms. A perm is defined as one grain of water passing through one
square foot in one hour under the action of a vapor pressure differential
of one inch of mercury. The corresponding unit of permeability is the
perm-inch. Besides the difficulty in manipulating units, the electronics-cooling
engineer is often at a disadvantage because of the need to use data out
of the realm of where it was tested and the test conditions are not well
documented.
After estimating the moisture permeation characteristics of an enclosure,
moisture control methods are usually considered [4]. One method of managing
infiltrated moisture is with the use of desiccants. Figure 2 illustrates
typical desiccant performance curves of loading versus water vapor pressure.
Silica gel is found in many packaging applications and is best suited
for moderate humidity levels. Molecular sieve desiccants are preferred
for packages where a very low humidity level is needed. Dynamic simulation
of a package with internal desiccant and varying environments is difficult
to perform with thermal tools and the interested reader is referred to
[5] for more information.
Figure 2. Typical desiccant performance curves
As a final comment, one reason for the variability in the literature
data for moisture permeation is that, unlike heat flow, it is difficult
to reduce the moisture flow properties for a material to a single coefficient
that is widely applicable. Similar to thin film thermal property variations,
moisture permeation in thin films is also geometry dependent and is further
complicated by the fact that defects contribute to the overall measured
permeation [6].
References
Wong, E. H., Koh, S. W., Lee, K. H., Lim, K.-M., Lim, T. B.; Mai,
Y., “Advances in Vapor Pressure Modeling for Electronic Packaging,”
IEEE Transactions on Advanced Packaging, Vol. 29, pp. 751-759, November
2006.
ASTM E96/E96M-05 Standard Test Methods for Water Vapor Transmission
of Materials, ASTM International, www.astm.org
Tencer, M. and Moss, J., “Humidity Management of Outdoor Electronic
Equipment: Methods, Pitfalls, and Recommendations,” IEEE Transactions
on Components and Packaging Technologies, Vol. 25, pp. 66-72, March
2002.
Myers, D., “Humidity Control Simulation for Electronics Packaging,”
Proceedings of IPACK03, Paper # 35057, 2003.
Graff, G., Willford, R., and Burrows, P., “Mechanisms of Vapor
Permeation through Multilayer Barrier Films: Lag Time Versus Equilibrium
Permeation,” Journal of Applied Physics, Vol. 96, No 4., pp. 1840-1849,
August 2004.
