PoP: An EMS perspective on assembly, rework and reliability
package are shown in Figure use in package on package assembly has
8, while the top and bottom emerged—dippable solder pastes. These
of the lower package are materials have a lower metal content than
shown in Figure 9. conventional solder pastes generally used
Three different alloy vari- for screening and also have a smaller mesh
ations were procured for each size. Each of the three material suppliers
package—the top package supplied a dippable solder paste—Supplier
was purchased in SAC105, A recommended a Type VI mesh size,
SAC305 and SAC 405 vari- while Suppliers B and C provided pastes
ants, and the bottom package with Type V mesh size particles. The alloy
Figure 8. Upper PoP package—top (left) and bottom (right).
was purchased in SAC125, for all three dippable pastes was SAC305.
SAC305 and SAC405 vari- A summary of the test cells assembled for
ants. Though packages with the material comparison is shown in Table
lower-silver-content balls 1. All of the components in this group
generally perform better in of cells had SAC105 balls on the upper
drop testing, components package and SAC125 balls on the lower
with higher SAC305 balls are package.
expected to perform better in As package on package components are
accelerated thermal cycling. available with different solder ball alloys,
SAC405-balled components three versions of the package on package
should have even better per- components were studied. The first version
formance in thermal cycling, was the version used in the flux and paste
Figure 9. Lower PoP package - top (left) and bottom (right). particularly given the harsher experiment, with a SAC105 upper package
cycling conditions selected and a SAC125 lower package (these assem-
for this project
6,7
. blies will be referred to as SAC105 for sim-
plicity). The second version had SAC305
Test vehicle balls on both the upper and lower package,
The test vehicle used for this and the third version had SAC405 balls
evaluation was a commercial- on both packages. The build matrix for the
ly available board with space alloy comparison is shown in Table 2. All
for 15 package-on-package of the boards in these cells were built with
placements. The board was the flux or paste from Supplier A.
132 mm x 77 mm in size Two different rework cells were also
and 1.0 mm thick with eight included, one set of ten SAC 105 com-
metal layers. The surface ponents and one set of eight SAC305
finish was a high temperature components. Both were reworked with the
Figure 10. Package on package test vehicle.
OSP. same material.
In an effort to simplify Finally, the effect of underfill was stud-
removal of test vehicles for ied. All underfilled cards had the SAC105
in many cases in the past, components
failure isolation during ATC testing, component sets and were assembled with
that were originally designed for handheld,
surface mount pin headers were soldered the respective suppliers’ dip pastes. Two
portable applications have eventually been
to the test pads on the test vehicle, and different underfill materials were included
designed into devices like servers or tele-
wire adds were routed to modify the test in the matrix (designated as Underfill 1
communications switches, where thermal
vehicle’s pinout to make it suitable for and Underfill 2). For each of the three dip
cycling reliability is a more significant
in-situ monitoring. In order to make most pastes studied, one card was underfilled
concern. In anticipation of that possibil-
efficient use of the standard cabling used on the lower level of joints only using each
ity for package on package devices, it was
in the thermal cycling chamber, two loca- of the two underfill materials. For Paste
decided to perform thermal cycling in
tions on each card were left unpopulated, B, two additional cards (one per underfill
an effort to develop an understanding of
resulting in a maximum sample size per material) were assembled with both the
the failure modes of package on package
card of 13 components. A photograph of upper and lower layer of joints underfilled
devices under thermal stresses, and to
an assembled test vehicle including the pin to determine if this different underfill
develop a process optimized for reliability
headers and wire adds is shown in Figure method has any impact on reliability.
under thermal cycling.
10. Table 3 summarizes the different assembly
variations in the underfill leg of the experi-
components
Test plan ment, and shows the component sample
The package on package components
The main process-related goal for this proj- size for each variation.
selected for this project were 14 mm x 14
ect was to assess various dippable fluxes Following assembly, the yield for all
mm versions. The bottom package has a
and solder pastes for use in assembling the the cells was assessed and the samples were
four-row peripheral array of solder balls
top package of the stack. Three dippable then sent for accelerated thermal cycling
with 353 I/O on 0.5 mm pitch. The top
flux formulations were selected, one from to assess the reliability of the different as-
package has a two-row peripheral array
each of three different manufacturers. sembly variations.
with 152 I/O on 0.65 mm pitch. Photo-
All of the fluxes were no-clean materi-
graphs of the top and bottom of the upper
als. Recently, a new class of materials for
18 – Global SMT & Packaging – March 2009
www.globalsmt.net
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62