 |
| Robert K. Lowry
|
Leaks, what they are, how to measure them, how to prevent them, their impacts on hermeticity, and if they even matter at all, is a real kettle of fish.
In fact even well intentioned discussions on the topic often descend into theological arguments that would make medieval clerics debating the number of angels who can dance on the head of a pin proud!
Since we are not theologians, but hermeticians,
 |
| Richard C. Kullberg
|
let's start with one thing we can all agree on. MIL-STD-883 Test Method 1014 is based on fundamental leak rate equations and sets certain guidelines the industry works to. For example,
- Gross Leak is characterized by viscous gas flow and is defined as 10-1 to 10-5 atm-cc/sec.
- Transitional flow physics occurs in a sort of "gray" area between gross and fine leak ranging from typically 10-5 to 10-6 atm-cc/sec.
- Fine Leak is characterized by molecular gas flow and is defined as 10-6 to 10-8 atm-cc/sec.
Recent discussion in the industry has emphasized an extended fine leak regime from 10-8 to below 10-12 atm-cc/sec in an effort to achieve more robust moisture control. But what do these very small numbers really mean inside your hermetically sealed package? Let’s draw a picture and see what it tells us:
We started our picture with a simple plot of how long it will take a given volume to leak all the way to atmospheric pressure at a given leak rate. Our range of volumes was chosen to cover traditional package sizes as well as very small MEMS packages.
Next we marked out the range of normal life times. This leads to some interesting insights. 10-8 is just fine for normal packages and if you get it all is well. But if you are in a small MEMS package that depends on maintaining a level of vacuum you need to worry about even the smallest flows of gas into the package, which could compromise vacuum.
Basically you can’t have leaks at all, the effective rate must be zero! Of course, even if you achieve this leak nirvana, pressure increases from other mass flows like materials outgassing will compromise the headspace.
What about using more sensitive leak testing techniques as has been recently proposed? Even if you can measure into the extended fine leak (10-8 to 10-12 atm-cc/sec) you must distinguish leaks from other mass flows in the package.
Only internal gas signatures from package gas analysis can differentiate between a leak, materials outgassing or other internal sources that compromise the headspace. We have addressed this issue in depth in an SPIE paper earlier this year. (Examining internal gas compositions of a variety of microcircuit package types & ages with a focus on sources of internal moisture, Proc. of SPIE Vol. 7206 720606-1)
Bottom line?
- Leak testing is important, but not the whole answer, especially when you work with smaller volumes.
- Practical hermeticity requirements (i.e. leak rates) should take into account expected product lifetime, types of components inside the hermetic enclosure, enclosure pressure, and internal volume.
- Leak testing itself is a secondary line of defense against moisture and other gas related problems from materials.
- Control to classical fine leak rates is good enough if M&P and outgassing are under control, especially for the larger volume enclosures.
- No leaks are acceptable in tiny volumes and mastering M&P and outgassing issues are critical for success.
Welcome to the head of the pin. To begin our effort to unravel the leak issue and put it in terms that you can engineer against, a subsequent article will take a closer look at the unpredictability of leak physics of our old friends oxygen, argon, moisture and helium.
See previous articles written for this series in Semiconductor Packaging News
It's Still Way To Much Water posted on October 27, 2009
Way Too Much Water posted on August 26, 2009
Removing Water posted on July 23
Where Does Water Come From? posted on June 15, 2009
STICTION posted on May 13, 2009
