Water, Water, Everywhere

In my last column, we spent a significant amount of time outlining the finer details of hermeticity. When it is needed, how to define it, and how to measure it. And to be quite honest, that discussion could have been continued for several more columns (and maybe one day it will). But, I would be remiss if I did not follow up that discussion with some cold, hard truth: the problem with hermetic devices is not always their hermeticity.

A substantial amount of effort and technology has been put forth on ensuring hermeticity; that is keeping the ‘bad air’ out. But, what if there was not enough ‘good air’ in the package to begin with? Increasingly, the root-cause investigations at DfR Solutions have found that it is not the leakers that are killing the hermetic devices; it is that the devices were not put together correctly in the first place. When this happens, the ‘air’ in the device is or becomes ‘bad’ without any leakage. This begs the obvious questions: What ‘air’ is typically in hermetic devices and when does it become ‘bad’ (I’m probably not being ‘grammatically correct’ with so many quotation marks)?

One of the things to note is that the purest environment and possibly the one least likely to do harm, a vacuum, is almost never used in a hermetic device. The absence of this option is driven by a number of factors, but it all comes down to Aristotle and ‘Horror Vacui’ (Nature Abhors a Vacuum). This behavior makes maintaining a vacuum in a welding / brazing / soldering environment almost impossible. In addition, even if you were successful maintaining a vacuum in a hermetic device, it will increase any kind of outgassing from materials within the hermetic device. And two, it will accelerate the input of any air from the outside through a leak.

With vacuum a non-starter, most hermetic device manufacturers backfill their devices with inert gases to atmospheric pressure. Options include nitrogen, argon, helium, some mixture of the three, or clean dry air. The rationale for the specific mixture is not always clear to the end-user (that is, YOU the reader), but is typically driven by cost and the manufacturing process.

Clean dry air is the preferred fill for hermetic devices with glass frit, as the absence of oxygen in the fill gas will reduce the glass seal (reduce means take away the oxygen atoms, which stops the frit from being a glass).

Hermetic devices that are welded, brazed, or soldered tend to not be big fans of the oxygen found in dry air, because they would tend to oxidize and reduce wetting behavior, so these devices use inert gases. The most popular choice tends to be 100% nitrogen (N2) because it is inert and cheap (its cost is not much more than dry air, which is primarily made up of nitrogen in the first place).

Next most common is a mixture is either nitrogen or argon and a MAXIMUM of 20% helium. The use of helium is primarily advantageous in that a helium bomb is not required to check leak rates. Helium is a problem in that it costs more money than argon or nitrogen and has a much lower breakdown voltage and can become conductive at higher percentages. Why 20% is a magic number is not quite clear, but seems to relate back to some work done at Bell Labs (shout out to my colleague John McNulty). Argon vs. Nitrogen? From my limited vantage point, it can seem to be arbitrary as on more than one occasion, we have seen similar products (e.g., implantable devices) with similar packaging (titanium), but different fill gases.

Now that we have defined the ‘good air’ (there are those pesky quotations again), when does ‘good air’ go bad? There are primarily two types of ‘bad’ air: moisture and reducing gases (e.g., hydrogen). Moisture is a well-known problem. It is so well known that is the only gas specified in both relevant military standards (MIL-STD-750, Test Methods for Semiconductor Devices – Method 1018.4 Internal Gas Analysis and MIL-STD-883, Test Method for Microcircuits – Method 1018.6 Internal Gas Analysis). Both standards state 5000 ppmv as the default limit for moisture content in hermetic devices.

First interesting observation: The two standards actually treat 5000 ppmv slightly differently. MIL-STD-750 states moisture content 5000 ppmv or greater as failure. MIL-STD-883 assigns 5000 ppmv as a specification limit. Terminology? Actual difference? Who knows. Second interesting observation: MIL-STD-750 also defines 10000 ppmv or greater of oxygen as failure, but makes no mention of hydrogen.

The rationale behind the 5000 ppmv limit is to prevent condensation before water freezes. Therefore, if the hermetic device does experience sub-zero temperatures, any water present in the device will precipitate on exposed surfaces in the form of non-conductive, non-corrosive ice. This approach would seem to take care of moisture content and for the most part every hermetic device manufacturer follows these limits. But, reality always has a way of being a little more complex.

The small complexity is that the actual content that will prevent condensation is 6000 ppmv, not 5000 ppm. However, we engineers like round numbers and a little bit of margin in our designs. The bigger complexity is applications that do not see temperatures below 0C. Desktop computers and implantable medical electronics are just two of many applications that were not designed and are not expected to ever see temperatures below 0C. Medical implants, as one example, are expected to operate at 37C. As long as they are not biased during shipping (which could see temperatures below 0C), these hermetic devices could theoretically have 40,000 ppmv moisture and still never see condensation.

Reducing gases, primarily hydrogen, are also a major cause of ‘bad air’ in hermetic devices. These gases, if present in high enough quantities (typically between 0.5% to 5%), can reduce oxides in the device to metal and water. Reducing oxides to metal can over time lead to changes in resistance (in the case of thin film resistors) or other electrical properties (such as capacitance in ceramic capacitors made from BaTiO3). And the other artifact of this reaction, moisture? Well, we’ve covered that pretty thoroughly already.

Remember, hermeticity and good air are the guaranteed paths for success in hermetic devices. And now that we have covered exactly what that means (for both terms), we can focus on the 999 other things that keep us electronic engineers up at night.