Calculations for Leak Rates of Hermetic Packages

An analysis is presented of the dependence of leak rates on geometry and flow mechanism. A combined-flow equation, valid for viscous, molecular, and diffusional flow is applied to capillary leaks and the results are extended to crack-shaped leaks and orifices. It is shown that for a given leak rate the geometry with the largest viscous contribution to flow is a single long capillary. A distinction is drawn between the rate at which a package leaks during helium leak test (in vacuum) and during use (in air at one atm pressure), and it is shown that the latter may be orders of magnitude smaller than the former because of the viscous contribution to flow during test and during bombing. Two different test procedures are proposed that are designed to prevent viscous flow by avoiding differences in total pressure: one is a one-atm "bath" instead of a "bomb" in radiotracer 85Kr, the other a gas-chromatographic collection of inert halogen-bearing gas that leaks out of a prefilled package. Both are predicted to give test leak rates within a factor of two of the use leak rates. Because of its widespread use and its complexity, the Howl and Mann molecular-flow equation is examined and a number of simplifying approximations are made to allow greater ease in its use. The time rate of change of pressure in a package is calculated using the simplifying approximation that leak rate is proportional only to difference in total or partial pressure, and the concept of maximum allowable dwell time for the "marginal gross leaker" is developed. Based on calculations of the rates of depressurization of a pressurized package, and of the rate of moisture accumulation in an unpressurized one, maximum allowable use leak rates RUfor a package of volume V, much lower than the currently used test leak rates, are proposed: Ru/V 10^-9atm sec^-1for pressurized packages or Ru/V 10^-10for unpressurized ones. It is recognized that for most package volumes these levels are difficult to attain, but it is proposed that\´ less stringent limits fail in their intended purpose of guaranteeing hermeticity for a reasonable package life.