Robert K. Lowry

Water! Is there a more paradoxical molecule? It’s essential for life, and to manufacturing today’s microdevices. Yet it is a major cause of erratic performance and failure in electronic and micromechanical devices.

In a prior article, we discussed the effect of water vapor on stiction forces which can impact normal function of micro and nano machines. Moisture condensate corrodes metallic structures (aided and abetted by soluble ionics), fogs optics, and as either condensate or adsorbate causes electrical leakage and device instability.

Richard C. Kullberg

It takes very little water to cause these problems. For example, the threshold for electrical leakage across surfaces is only three monolayers of adsorbed water molecules. Very little water is needed in a sealed enclosure for the three monolayers to form. The critical concentration is only one-half volume percent moisture (5000 parts per million by volume)!

Condensation is equally easy to form if you are not careful. The dewpoint temperature at which moisture starts to condense from that concentration is -2ºC, a temperature easily reached in many in-service conditions. As the water concentration increases, the dewpoint increases as well, e.g. +8ºC for a concentration of one volume percent.

Adding in today’s sub cc volume enclosures accommodating many devices, very little water is needed to reach these problem conditions.

Fig. 1. Corroded traces on a PCB.2

The dangers moisture poses are not theoretical. Moisture-related failures of microdevices threatened a recall of the entire US nuclear submarine fleet in the 70’s, affected space missions including a last minute Shuttle launch abort in the 80’s, compromised telecommunication optics in the 90’s, and have caused recalls of sealed medical devices in the 2000’s. Moisture control is absolutely necessary, and it’s an ongoing battle.

Figure 2. Much less moisture than this condensate causes component functional problems.

So where does all this water come from? There are only two sources. It comes either from the materials and processes used to manufacture the parts, or it leaks in through defective hermetic seals.

Mass spectrometry³ is the method for analyzing sealed enclosures for moisture and other volatiles. Mass spectrometry gives a “fingerprint” of volatiles composition which in turn gives clues to sources of moisture. If there was a leak, components of air (oxygen, argon) will be present. If the components of air are negligible, the moisture comes from materials outgassing, for which there is a wide variety of mechanisms.

Table 1 shows mass spectrometric results for typical units that exceed the 5000ppmv threshold discussed above. Unit A appears to have a leak, based on components of air. Unit B has no hint of air; the moisture was desorbed from interior package surfaces that were not adequately baked prior to seal.

Unit C has no hint of air; the methane and ammonia indicate outgassing from polymer die attach material with dicyandiamide curing agent not fully cured prior to seal. Unit D is like C except that the polymer attach material did not contain the curing agent.

Table 1. Internal gas analysis results for several typical packages.

Gas analysis data for 200 units of a variety of package types are summarized in Table 2.

Table 2. Internal gas analysis summary for 200 sealed enclosures.

Fifty nine of 200 units exceeded the moisture threshold. Of these, 8 units (14%) had evidence of air ingress. The remainder (51 units, 86%) showed no evidence of air ingress, just elevated moisture with varying amounts of non-air volatiles. These units are wet because moisture outgassed from package lids, bases, device attachment substances, coatings, and other materials internal to the package.

Interestingly, the 15 “old” units each contained only a few hundred ppmv of moisture and no indication of leakage. Despite skepticism in some circles, it is possible to build parts that are truly hermetic and dry for very long periods of time. All enclosures may have a finite leak rate, but such rates can be so small as to be negligible for practical purposes.

These findings indicate that most enclosures with elevated moisture are wet due to materials outgassing. When developing hermetic sealing, first achieve moisture control by optimizing materials and processes. Only when that is achieved does hermeticity and leak rate testing become important.

Footnotes

^1. SPIE Paper 7206, January, 2009.

^2. Contamination and moisture effects on printed circuit board reliability, http://www.era.co.uk/

^3. Mil Spec 883, Test Method 1018.4.

Robert K. Lowry (321-777-9949, www.electronic-materials.com) is a consultant/materials scientist with 39 years microelectronic industry experience and is also a principal with Arthur Jonath Associates. He has a course available for clients titled “Moisture Measurement and Control for Microelectronics”.

Richard C. Kullberg (719-966-4296, rckullberg@vacuumenergyinc.com, brings 30 years of materials science and microelectronic industry experience to his product and business development work with Vacuum Energy, Inc.