Documente Academic
Documente Profesional
Documente Cultură
The moisture sensitivity problem for plastic surface mount Experimental Process Flow
packages is widely accepted in the industry as a significant The experimental process flow is shown in Figure 5. Initial
reliability issue and is well documented in the literature [3- C-SAM analysis was performed so that units with “as-
5]. Plastic IC packages readily absorb moisture from received” delamination could be identified. This sorting
humidity in the air. As moisture diffuses into the package, it process serves to separate otherwise convoluted variables
begins to collect and condense at internal material that significantly affect the results of moisture/reflow
interfaces. When packages are reflow soldered to printed testing.
circuit boards, moisture present at the internal package
interfaces vaporizes. The magnitude of the vapor pressure After initial C-SAM analysis, the devices were baked
that develops in the package during reflow increases with completely dry at 125°C. Weight loss measurements were
peak reflow temperature. The stresses that develop within performed to determine the required bake time for each
the package as a result of vapor pressurization can cause package type in the study. A package was considered dry
interfacial material delamination if the adhesion strength of when no significant change in the measured weight (less
the interface is exceeded. Furthermore, if the stresses that than 0.002%) was observed over a 24 hour period. As in the
develop exceed the fracture strength of the plastic, package case of sorting units with “as received” delamination, this
cracking may occur. In the worst case scenario, “popcorn” method of baking the test samples completely dry was
cracks, or rapid crack propagation during reflow, may occur. chosen over the more conventional method of a fixed
Delamination and package cracking pose serious reliability duration bake to ensure a separation of otherwise
concerns in that catastrophic failures may result due to lifted convoluted variables.
ball or sheared wedge bonds. Long term reliability may also
be jeopardized by the possible ingress of ionic contaminants Although not the subject of this work, setting the appropriate bake
along exposed cracks thereby increasing the likelihood of times to ensure that product is completely dry when sealed in
moisture barrier bags prior to shipment may be very important
device failure by corrosion [5].
when higher peak reflow temperatures are used. Bake times are
particularly important in packages with complex interconnect
The objective of this study is to characterize the effects of schemes. Another potentially significant factor at high reflow
increased reflow temperatures on the moisture sensitivity temperatures may be very low levels of moisture absorbed while
performance of plastic surface mount packages. Several product is stored in moisture barrier bags. We suspect that success
different package families were represented in this study in high volume production may depend upon the proper
including thick and thin leadframe-based packages, PBGAs anticipation, understanding, and control of these factors.
(2 and 4 metal layer substrates), and LFBGAs (low profile
fine pitch BGAs). Moisture/reflow testing was performed at Moisture/reflow analyses were performed at various
peak reflow temperatures of 250°C and 260°C. C-SAM was moisture sensitivity levels (depending upon package type)
used to judge package integrity before and after and peak reflow temperatures of 250°C and 260°C.
moisture/reflow testing and cross-sectional failure analysis Packages were also tested at the current IPC/JEDEC
was performed to verify the location of the failing interface. moisture sensitivity level and peak reflow temperature as a
control for the reflow experiments. Moisture pre-
EXPERIMENTAL TEST METHODOLOGY conditioning was performed according to IPC/JEDEC J-
Moisture/reflow analyses were performed on five different STD-020A [6] which is shown in Figure 3. Where
package types as shown in Figure 4. Each device was applicable (i.e., levels 2a-5a), the accelerated 60°C/60%RH
chosen to represent a specific package family. The test conditions were used. After moisture pre-conditioning,
moisture/reflow performance of each family is unique due devices were subjected to three passes in a convection
to differences in assembly materials and/or package reflow oven. Lead-free reflow profiles with peak
thickness and was thus characterized independently. temperatures of 250°C and 260°C were created for each
Packages with large chip-to-package ratios were typically package type per the NEMI [7] recommended conditions
selected to represent near worst case sensitivity to that are shown in Figure 6. The reflow profiles that were
moisture/reflow testing at lead-free reflow temperatures. In used to test at conventional eutectic tin-lead reflow
temperatures were established per J-STD-020A.
C-SAM was used to judge package integrity after Preheat
150°C (+/-25°C) for a total of 60-
moisture/reflow testing was completed. For a given package 120 seconds
Time >50°C 3.5 – 6.0 minutes
type, initial moisture/reflow evaluations were generally Time >217°C 60-150 seconds
performed at MSL-4/260°C conditions. If the test result was Ramp rate >217°C 3°C/sec maximum
a “pass”, an evaluation was performed on a new sample at Time within 5°C of peak 10-20 seconds
the next higher moisture level. Alternatively, if the result of Peak temperature range
260°C (-5/+0°C)
the test was a “fail”, an evaluation was performed on a new (or 250°C, 240°C, etc.)
Ramp down rate 6°C/sec maximum
sample at the next lower moisture level. Using this
approach, moisture/reflow testing was performed until a Figure 6. NEMI lead-free reflow profile requirements.
passing and failing moisture level was identified for each
package type at both 250°C and 260°C. Once completed, Chip surface delamination is classified into two categories:
cross-sectional failure analysis was performed on failed
packages to confirm the location of the failing interface. 1. Failure: >5% of the chip surface area is delaminated.
Experience has shown that this amount of delamination
may lead to ball bond shear or chip passivation
cracking.
Initial C-SAM analysis to
identify units with as- 2. Marginal Pass: >1% but <5% of the chip surface area is
received delamination delaminated. Ball bond shear and chip passivation
cracking are unlikely. While not considered a
failure, marginal performance is a concern in high
volume manufacturing and additional studies are
Bake dry at 125°C recommended.
(verify with weight
loss measurements) For thin leadframe packages (LQFP), bottom paddle
delamination greater than 50% of the pad area is considered
a failure because it may lead to belly bulge (swelling of the
package underside) and subsequent board attach problems.
Pre-condition at various
moisture sensitivity levels For laminate-based packages (PBGA, LFBGA), through-
mode C-SAM must be performed to detect “popcorn” type
failures and delamination at any of the internal package
interfaces. With the exception of die-attach delamination
(delamination between the chip and die-pad area of the
Convection oven reflow substrate), any measurable amount of delamination detected
at 250°C or 260°C by C-SAM may lead to cracking in the wedge bonds,
(3 passes) substrate vias, or traces and is considered a failure. Die
attach delamination greater than 10% of the die-pad area
may also lead to belly bulge and is therefore considered a
failure.
C-SAM to assess The pass, marginal pass, and fail criteria are summarized in
Pass package integrity Figure 7.
Type of Applies
Fail Marginal Pass Fail
Delamination to
1% - 5% chip >5% of chip
Chip surface All
surface area surface area
Cross-section to verify Bottom of >50% of
LQFP n/a
location of failing interface Die-Pad die-pad area
Bottom of Die: Bottom of Die:
Laminate 1% -10% by area >10% by area
Figure 5. Experimental process flow used for Through mode Only
And Or
moisture/reflow analysis. delamination (PBGA &
LFBGA) No Detectable Any Detectable
Delamination Delamination
Evaluation Criteria Mold-compound cracks Criteria per IPC/JEDEC J-STD-020A
After moisture/reflow testing, results are reported as pass,
Figure 7. Evaluation criteria for judging moisture/reflow
marginal pass, or fail. If there are any failures, the number
response.
of failing units is reported along with the failure mode.
RESULTS through the mold compound towards the bottom of the
Weight Loss Analysis package but does extend to the external package boundary.
The time required to bake devices dry at 125°C is dependent
on the package type and the ambient exposure history of the Peak Reflow Temperature
samples. A standard 24 hour bake was sufficient for the 2- MSL 225°C 250°C 260°C
Pass Fail Fail
layer PBGA samples as shown in Figure 8. However, the 4- 2a (9 units) (9/9 internal mold- (8/8 internal mold-
layer PBGA package required a minimum bake time of 96 compound cracks) compound cracks)*
hours at 125°C to completely dry the internal package 3
Pass Pass
interfaces (see Figure 9). As previously mentioned, the (9 units) (9 units)
intricacies of completely drying packages with complex Pass
4
(9 units)
interconnect schemes may be critical to high-volume lead- * one unit was lost during testing
free manufacturing and will be the subject of future work.
Figure 10. Moisture/reflow results for thin leadframe
packages (144LQFP).
Percent Weight Change vs. Bake Time
at 125°C (2-layer PBGA)
0.015
Leadframe paddle edge
0.012
% Weight Change
0.006
0.003
0
0 5 10 15 20 25
Bake Time (hours)
Bottom package edge
Figure 8. Weight loss data for the 2-layer 272PBGA
package.
Percent Weight Change vs. Bake Time Figure 11. Cross-section of 144LQFP package after level
at 125°C (4-layer PBGA) 2a/260°C indicating crack in mold compound.
0.1
0.08
QFPs
% Weight Change
REFERENCES
Chip
[1] 4th draft of the European Commission (EC) directive,
Crack in substrate “Waste from Electronic and Electrical Equipment”
(WEEE), 2000.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the Assembly and
Process Development Group in Singapore for assembling
the packages for this study. The authors also wish to
acknowledge Susan Geiger for performing the reflow testing
and Jonathan Scharwz for performing C-SAM analyses.
Finally, the authors wish to acknowledge Ray Nika and