Page 245 - Six Sigma Advanced Tools for Black Belts and Master Black Belts

OTE/SPH
OTE/SPH
August 31, 2006
3:4
Char Count= 0
JWBK119-15
230 Development of A Moisture Soak Model For Surface Mounted Devices
gives nearly parallel lines. This justifies the use of a common scale parameter, b, for
different experimental conditions.
In the following, we shall use the loglogistic distribution for further analysis and
discussion. However, it should be borne in mind that the same principle can be ex-
tended to other distributions.

15.3.3 Combined analysis with the acceleration model
In order to estimate the equivalent moisture soak time for conditions other than those
tested, we need to establish an acceleration model that best fits the moisture gain data
under various relative humidity (RH) and temperature (T) combinations. From the
reliability engineering literature, two models are commonly used for describing the
relation between ‘lifetime’ and the two stress variables (T and RH): the Peck model
and the generalized Eyring model. These models express the ‘nominal’ product life
as some function of the T and RH. The exact forms of these models are given on pp.
13
98--102 of Nelson’s book. In the case of moisture absorption, since the time to reach
saturation is akin to the time to failure, the logarithm of the “nominal” moisture soak
time is equivalent to the location parameter a. In particular, for the Peck model, we
have
E −n

L = A exp RH , (15.3)
kT
where L is the lifetime, Ais a constant, E is the activation energy and k is the Boltzmann
constant. (These also apply to subsequent equations.) In the current context, with α i
(i = 0, 1, 2) as the coefficients in a linear model, we have
1
a = α 0 + α 1 + α 2 ln(RH). (15.4)
kT
For the generalized Eyring model, we have

A E C
L = exp exp RH B + , (15.5)
T kT kT

where B and C are constant. In the current context, this becomes
1 RH
a + ln(T) = α 0 + α 1 + α 2 RH + α 3 . (15.6)
kT T
13
A variant of the Eyring model that has been used by Intel (see p. 102 of Nelson )is
E

L = A exp [exp(−B.RH)]. (15.7)
kT
This gives

1
a = α 0 + α 1 + α 2 RH. (15.8)
kT