502
P. Vianco and M. Neilsen
through their respective T g values, then the temperature cycle used in accelerated
aging should not pass through T g .
Test Vehicle Construction
The second complication that arises in the thermal cycling of stacked packages
concerns the daisy chain loops that are used to monitor for electrical opens per Ref.
[24]. These components have a combination of first-level interconnections—one for
PoP and two for PoPoP—and second-level interconnections, which are between the
bottom package and the PCB. All signals must eventually pass through the secondlevel interconnections, which would “control” the occurrence of failures. It is for
this reason that the temperature cycling of stacked packages is also referred to as
board level reliability (BLR) evaluations because of the de facto emphasis is placed
on second-level interconnections. In the event that one of the first-level solder joints
fails prior to the second-level interconnections in that loop, that fact can only be
learned, later on, by destructive analysis
1 .
Materials Set for Computational Modeling
Numerous empirical studies have examined the reliability of PoP components under
temperature cycling [28, 30]. The challenge faced with understanding and, more
importantly, making use of those data is the multiple variables that can affect the
reliability. The PoP and PoPoP parameters are:
• Molding compounds: material type and thickness;
• Substrate: laminate material and Cu features;
• Solder joints: alloy composition, pitch, and ball size; and
• Surface finishes: compositions, which can affect solder joint fatigue.
There are also the non-PoP-specific parameters:
• Underfill: material type and geometry (some or all gaps);
• Encapsulants and conformal coatings: material properties and geometry;
• PCB; laminate, thickness, and Cu loading;
• Service environments: time-temperature profile(s); and
• Test environment: selected temperature limits, ramp rates, and hold times,
that together must correlate to service environments through the appropriate
acceleration factor.
Even in the event that all of the temperature-dependent CTE, elastic modulus, and
Poisson’s ratio parameters are available, an underlying assumption is that they do
not change significantly over time due to any intrinsic aging phenomenon.
1 Based on the statistical aspects of reliability, the failure of a package does not discern whether the
open occurred in the first-level or second-level interconnections.
P. Vianco and M. Neilsen
through their respective T g values, then the temperature cycle used in accelerated
aging should not pass through T g .
Test Vehicle Construction
The second complication that arises in the thermal cycling of stacked packages
concerns the daisy chain loops that are used to monitor for electrical opens per Ref.
[24]. These components have a combination of first-level interconnections—one for
PoP and two for PoPoP—and second-level interconnections, which are between the
bottom package and the PCB. All signals must eventually pass through the secondlevel interconnections, which would “control” the occurrence of failures. It is for
this reason that the temperature cycling of stacked packages is also referred to as
board level reliability (BLR) evaluations because of the de facto emphasis is placed
on second-level interconnections. In the event that one of the first-level solder joints
fails prior to the second-level interconnections in that loop, that fact can only be
learned, later on, by destructive analysis
1 .
Materials Set for Computational Modeling
Numerous empirical studies have examined the reliability of PoP components under
temperature cycling [28, 30]. The challenge faced with understanding and, more
importantly, making use of those data is the multiple variables that can affect the
reliability. The PoP and PoPoP parameters are:
• Molding compounds: material type and thickness;
• Substrate: laminate material and Cu features;
• Solder joints: alloy composition, pitch, and ball size; and
• Surface finishes: compositions, which can affect solder joint fatigue.
There are also the non-PoP-specific parameters:
• Underfill: material type and geometry (some or all gaps);
• Encapsulants and conformal coatings: material properties and geometry;
• PCB; laminate, thickness, and Cu loading;
• Service environments: time-temperature profile(s); and
• Test environment: selected temperature limits, ramp rates, and hold times,
that together must correlate to service environments through the appropriate
acceleration factor.
Even in the event that all of the temperature-dependent CTE, elastic modulus, and
Poisson’s ratio parameters are available, an underlying assumption is that they do
not change significantly over time due to any intrinsic aging phenomenon.
1 Based on the statistical aspects of reliability, the failure of a package does not discern whether the
open occurred in the first-level or second-level interconnections.