III. Evaluation plan and procedure
3.4 Test sample manufacturing
3.4.2 Quality inspection result
Quality inspection is about inspection, measuring, or testing of product characteristics. To make sure all samples can meet the quality requirement and can release to next evaluation without abnormal phenomenon, we perform tests in production.
1. Appearance inspection
The optical microscope is used to check the outlook quality. All units are inspected and the test result can meet the specification on solder paste print, bonding position
accuracy, topside and backside quality before/ after molding underfill process.
-44-
Fig. 3-20 The appearance inspection
Fig. 3-21 The appearance inspection after molding process
-45-
2. Die shear testing
The die shear testing for six cells has been done to confirm the solder joint strength.
Fifteen units per cell are examined and the result could pass the specification. The defect mode after shear testing is solder ball-neck broken, and this failure phenomenon meets the standard. The test readings are listed on Table 3-4.
The cell 4, cell5 and cell 6 with bare copper LF got better test result, there is no significant test difference from cell 1 to 3 and the PPF treatment lead frame is used in these cells.
unit: Kg Table 3-4 The shearing test result
Cell 1 2 3 4 5 6
1 0.959 1.083 0.967 3.257 3.288 3.213
2 1.185 0.995 0.883 3.033 3.088 3.262
3 0.887 0.954 0.882 3.183 2.672 3.332
4 1.168 1.068 1.083 3.446 3.249 3.397
5 0.852 0.877 0.928 2.541 2.751 3.078
6 0.895 0.898 0.868 3.332 2.825 3.208
7 1.126 1.078 0.892 3.193 3.014 3.574
8 0.994 0.874 1.048 3.224 3.279 3.334
9 0.972 1.056 0.961 3.206 2.905 2.687
10 1.254 0.931 0.898 2.948 3.016 3.275
11 1.065 0.956 1.218 3.152 2.962 3.078
12 0.986 1.185 0.922 2.913 3.303 2.875
13 1.169 0.848 0.972 3.101 3.011 3.371
14 1.193 1.021 1.025 2.763 2.694 2.986
15 1.194 0.908 1.017 3.228 2.948 3.258
max 1.254 1.185 1.218 3.446 3.303 3.574
min 0.852 0.848 0.868 2.541 2.672 2.687
mean 1.060 0.982 0.971 3.101 3.000 3.195
sigma 0.132 0.097 0.095 0.231 0.212 0.224
Spec 0.080 0.080 0.080 0.080 0.080 0.080
Ppk 2.469 3.092 3.122 4.357 4.589 4.643
-46-
Fig. 3-22 Die shearing test analysis
Fig. 3-23 The breaking mode after die shearing test
3. X-ray inspection
The X-ray inspection has been done for six cells to confirm the solder joint
interconnection. Any defect such as chip shift, as shown in Figure 3-24, and internal void
-47-
under bump connection interface should be scraped.
Fig. 3-25shows the X-ray inspection result for normal unit.
Fig. 3-24 The X-ray inspection result for a defect unit
Fig. 3-25 The X-ray inspection result for normal units
4. Scanning Acoustic Microscopy (CSAM) inspection
CSAM inspection is confirmed for six cells to exam the delamination in different interfaces within packages. Figure 3-26 shows the result of normal units and defect units.
All units are inspected and the result could pass the specification. No delamination abnormality is found in these samples.
-48-
Fig. 3-26 CSAM inspection result of normal units and defect units
5. Cross sectioning inspection
After package separation process, we performed cross-section to confirm the
package structure, chip tilt. The unit is grinded from side of package to bonding area and inspected by optical microscope. No abnormal phenomenon was found in the sample.
Fig. 3-27 Cross section of flip chip CSP package
-49-
6. Electrical test
The judgment of reliability test is based on the electrical test results, therefore, all samples are confirmed by electrical test. We use high-accuracy 4-wire measuring system in the test. Table 3-5 shows all initial test readings of resistance. From the test results, the cell 1 to 3 show higher resistance than those of cell 4 to 6 and that is due to the solder plating thickness deviation of package outer-lead. The data is for reference only, because we judge the device by open or short condition.
Unit: mOhm Table 3-5 Initial resistance measurement data
unit Cell-1 Cell-2 Cell-3 Cell-4 Cell-5 Cell-6
1 56 46 54 46 41 48
-50-
Chap 4. Evaluation Result Analysis
In this chapter, we confirm the feasibility assessment of surface mount assembly for flip chip CSP package application and analyze the reliability test result including Weibull distribution, fail point in stress test and the defect mode analysis.
4-1 Pressure cooker test (PCT) failure distribution
Through the resistance measurement method on the package with daisy-chain design, we can know the electrical performance precisely. All units in six DoE Cells are measured after every test point. The test points are 100, 200, 300, 500, 1000 hours. The below table is the failure distribution.
Check points PCT 0hr PCT 100hrs PCT 200hrs PCT 300hrs PCT 500hrs PCT 1000hrs
Cell 1 0 0 0 0 1 3
Table 4-1 PCT test result
We observed the first failure point is at PCT 300 hours for cell 2 and cell 3. Based on the JEDEC standard definition, the pass criteria is 168 hours, so all these units can meet this requirement and have no concern on the reliability performance. For the test results of other cells, we see cell 1, the first failure point is at 500hrs and cell 4, cell 5 and cell 6, the failure point is at PCT 1000hrs.
-51-
4-2 Temperature cycling test (TCT) failure distribution
Following the resistance measurement method as PCT, all units in six DoE Cells are measured after every test point. The test points are 100, 200, 500, 1000, 1500, 2000, 3000 and 3500cycles. The below table is the failure distribution.
Table 4-2 TCT test result
We observed the first failure point is at TCT 200 cycles for cell 1 and cell 2, for cell 3, the failure point is at TCT 500 cycles, for cell 4, cell 5 and cell 6, the first failure point is at TCT 3000 cycles. Based on the definition of JEDEC standard, the pass criteria is 1000cycles, so test units from cell 4, cell 5 and cell 6 can meet this requirement.
Table 4-3 shows the electrical test result, we measured the resistance after every test points and then generated the chart using average data. We observed the electrical performance was stable for cell 4, cell 5 and cell 6. The cell 1, cell2 and cell 3 shows worse resistance after TCT 500C test, and the same lead frame material are used for these three cells.
Cycles MSL 1 TCT 100c TCT 200c TCT 500c TCT 1000c TCT 1500c TCT 2000c TCT 3000c TCT 3500c
Cell 1 0 0 2 3 6 8 12 17
-52-
Table 4-3 Resistance measurement after TCT test
4-3 Weibull distribution
From the reliability test result in the PCT and TCT test, the Weibull distribution can be drew to model the lifetime data. Through the equation (2-4) we discussed in chapter 2.-2-2, we can find the shape parameter (β) and characteristic life time (η).
𝑓𝑓T(t) = βη�t−γη �β−1e−�t−γη �
β
, t ≥ γ (2-4)
A commercial software - Mimitab is used for data calculation and result analysis in this study. Mimitab is distributed by Minitab Inc, and is a powerful statistic and process
management software which is used in statistics-based process improvement methods, like Six-sigma, DoE, reliability and other statistics-based process improvements.
-53-
4-3-1 Weibull distribution for PCT
The experimental result and time interval are fitted into the Minitab software and use the Reliability/ Survival function to analysis these data, the distribution is drew, and then the shape parameter (β) and characteristic life time (η), hazard rate (h) can be estimated.
Fig. 4-1 Weibull distribution for PCT Cell-1
-54-
Fig. 4-2 Weibull distribution for PCT Cell-2
Fig. 4-3 Weibull distribution for PCT Cell-3
-55-
Fig. 4-4 Weibull distribution for PCT Cell-4
Fig. 4-5 Weibull distribution for PCT Cell-5
-56-
Fig. 4-6 Weibull distribution for PCT Cell-6
The shape parameter (β) for all 6 cells are larger than one, it means the distribution that can meet the Weibull normal distribution. The shape parameter and the scale parameter for 6 cells are summarized as following table.
Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6
Shape parameter (β) 2.69 2.26 3.69 2.9 3.05 2.78
Scale parameter (η) 2720 1049 1131 3648 2797 3958
Table 4-4 Weibull distribution for PCT
-57-
4-3-2 Weibull distribution for TCT
Following the same method, we input the failure result and time interval into the Minitab software and use the Reliability/ Survival function to analysis these data, the shape parameter (β) and characteristic life time (η), hazard rate (h) can be estimated.
Fig. 4-7 Weibull distribution for TCT Cell-1
-58-
Fig. 4-8 Weibull distribution for TCT Cell-2
Fig. 4-9 Weibull distribution for TCT Cell-3
-59-
Fig. 4-10 Weibull distribution for TCT Cell-4
Fig. 4-11 Weibull distribution for TCT Cell-5
-60-
Fig. 4-12 Weibull distribution for TCT Cell-6
The shape parameter (β) for all 6 cells larger than one, it means the distribution that can meet the Weibull normal distribution. The shape parameter and the scale parameter for 6 cells are summarized, as shown in Table 4-5.
Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6
Shape parameter (β) 1.65 2.24 2.37 4.82 5.85 4.49
Scale parameter (η) 2905 1768 2678 4551 4233 5036
Table 4-5 Weibull distribution for PCT
-61-
4-4 Life time prediction
For the life time prediction, the key point is the Acceleration factor (AF). If we find the AF, the life time of IC components used in normal environment can be predicted easily.
In chapter 2-2-3, we discussed the relationship about humidity and temperature, the equation is given
AF =LLuse
accl = �RHRHtest
use�n ∙ e�𝑘𝑘∙T𝐸𝐸𝐸𝐸��ttest1 − tuse1 � (2-10)
We also consider below factors:
1. Humidity factor:
a. The max. humidity for normal condition is 60%
b. The humidity on PCT test is 100%
c. Humidity integer constant, default =2.7 2. Temperature factor:
a. The max. temperature for normal operation is 70°C b. The temperature on PCT test is 100°C
And Ea is proposed to 0.7 for solder joint.
Then we substitute these factors into equation (2-10), we can find AF.
( )
( )-62-
From Scale-accelerated failure time models (SAFTs), the relationship equation of AF is as below
AF = LLuse
accl , where Luse = AF ∙ Laccl (3-1)
In Weibull distribution analysis, the eta parameter means the time parameter for the acceleration test.
Given AF = 54.34 , the life prediction is listed as below Table 4-6.
AF Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6
L accl- PCT (h) 2720 1162 1131 3648 2797 3958
L use- PCT (h) 54.34 147805 63143 61459 198232 151989 215078
L use- PCT (year) 16.9 7.2 7.0 22.6 17.4 24.6
Table 4-6 The life time prediction
Finally we combine the reliability test result and life time prediction result into the table as below:
Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6
1st Fail point-PCT (h) 500 300 300 1000 1000 1000
L use- PCT (year) 16.9 7.2 7.0 22.6 17.4 24.6
1st Fail point-TCT (c) 200 200 500 3000 3000 3000
* The base requirement for PCT test is pass 168hrs
* The base requirement for TCT test is pass 1000 cycles
Table 4-7 Combination of life time prediction and reliability test result
-63-
4-5 Failure mode analysis
A IC device is composite by several materials, the joint strengths of these interfaces are different accordingly, in general, the failure happens on the weakest point after stress application. We observed electrical fail after reliability test for some units, and to figure out the failure mode for further improvement, a cross-sectioning and scanning electron microscope (SEM) are performed on failure parts after PCT test.
Fig. 4-13 SEM photos for failure samples of six cells
For cell 1, cell 2 and cell 3, we observe the crack in the interface layer of solder joint from copper pillar bump to lead land. The thickness of this layer seems lower than the other cells and this may be the reason to cause the weak joint strength.
For cell 4 and cell 6, the crack is found in the interface layer from molding underfill to chip surface, we suppose the cause is by delamination. The adhesion strength from copper pillar bump to lead land is robust than the one from underfill to chip surface.
For cell 5, the crack is found in the UBM (Under Bump Metallurgy) layer, we notice the solder paste from copper pillar bump to lead land is too plenty, so the joint strength is
-64-
good and the stress is extended to this layer, we think this defect is related to the solder paste printing process control issue .
4-6 Summary
Here we brief summarize the experiment result from this experiment in this study
1. Cell 1, cell 2 and cell 3 are failed at TCT test due to solder joint crack, the result shows the joint strength is not enough to sustain the stress from temperature cold to hot change.
2. From the result of lifetime prediction, cell 1, cell 4, cell 5 and cell 6 meet the base line - Ten years normal use requirement, even cell 4 and cell 6 can meet twenty years high level requirement.
3. The weak points for cell 1, 2 and 3 are solder-joint layers. From the failure analysis, the bonding layer crack is observed after stress test. The cause is come from the good wetting performance on lead frame surface- PPF to solder paste. The solder paste is bled out after IR reflow process and caused lower thickness of joint layer.
4. Cell 4, cell5 and cell 6 all got good results on die shear strength test, and the result is positive correlation to the TCT test.
5. Compared to the six cells on reliability tests, the solder paste with higher content silver has better test result.
-65-
Chap 5. Summary and Future Work
In this study, we evaluate the surface mount assembly on flip chip CSP package, and we find three cells of total six cells can meet the requirement of the reliability test and product lifetime in normal use condition at least ten years.
The elements for successful cells are pure copper lead frame surface treatment with each one of SAC 305, SAC 300 and SAC405 (cell 4, cell5, cell 6), that is because the solder paste is restricted in the printing land area after IR reflow process, and have a good support on the package.
For the PPF treatment lead frame, the solder paste spreads out of the printing land area and causes lower thickness in the solder joint layer, this is also the reason to explain the lower shearing test compared to other cells.
We found the best result for the reliability performance is the one with Sn 95.5%/ Ag 4%/ Cu 0.5% solder paste. This is verified on pure copper lead frame but for the PPF lead frame, due to there is a noise (solder paste bleed) on the test, so it is difficult to analyze the trend.
Based on the evaluation result, the SMT process is workable on flip chip CSP package, and can reach the reliability target. With this method application, this process requires specific controls for better yield performance.
1. Opening definition on solder paste printing stencil
2. The land opening design on PPF surface should be considered, an obstructive area on inner lead or finger is needed.
3. Copper bump pitch > 80um is preferred for the process due to Chip bonding accuracy requirement (+- 15 um)
-66-
For future work, we plan to apply this method on SiP package with thin thickness and propose to integrate more components in substrate base. Since the thickness of cell phone is from 12mm to 7 mm, it should be a big challenge on package assembly in near future.
-67-
Bibliographies
1. http:// www.wisegeek.com/what-is-flip-chip-technology.htm 2. http:// en.wikipedia.org/wiki/Flip_chip
3. Yunn-Horng Guu, Lung-Sheng Lee & Ming-Zheng Huang, Using Finite Element Method software to enhance teaching about copper pillar bump flip-chip packaging, World Transactions on Engineering and Technology Education, Vol.9, No.3, 2011
4. Ikuo Shohji, Satoshi Tsunoda, Hirohiko Watanabe, Tatsuhiko Asai and Megumi Nagano, Reliability of Solder Joint with Sn–Ag–Cu–Ni–Ge Lead-Free Alloy under Heat Exposure Conditions, Materials Transactions, Vol. 46, No. 12 (2005)
5. Ranjit Pandher, Tom Lawlor, Effect of Silver in common lead-free alloys, Cookson Electronics Assembly Materials, 109 Corporate Blvd., South Plainfield, NJ 07080
6. http:// en.wikipedia.org/wiki/Flip_chip
7. Murthy, D. N. P., Xie, M. and Jiang, R.. Weibull Models, John Wiley, Inc. New York, 4 DEC 2003
8. Thomas P. Ryan, Modern Engineering Statistics: Solutions Manual to Accompany, John Wiley & Sons, Inc. New York, 28 SEP 2007
9. D. Stewart Peck, Comprehensive Model for humidity testing correlation, Proceedings of the IEEE International Reliability Physics Symposium, 24, 44–50., 1986
10. Institute of Electrical and Electronics Engineers (1990) IEEE Standard Computer Dictionary: A Compilation of IEEE Standard Computer Glossaries. New York, NY ISBN 1-55937-079-3
11. Joint Industry Standard IPC/JEDEC J-STD-020D.1, March 2008, page 1 and 6
12. Anand, L., Constitutive Equations for the Rate-Dependent Deformation of Metals at Elevated Temperatures, ASME Journal of Engineering Materials and Technology,
-68-
Vol.104, (1982) pp. 12-17.
13. Changwoon Han and Byeongsuk Song, Development of Life Prediction Model for Lead-free Solder at Chip Resistor, Electronics Packaging Technology Conference, 2006 14. http://www.jedec.org/standards-documents/results/reliability
15. MyoungSu Chae, Eric Ouyang, Strip Warpage Analysis of a Flip Chip Package Considering the Mold Compound Processing Parameters, 2013 Electronic Components and Technology Conference
16. Wen-chin Lin, Determination of the Influential Factors in Manufacturing of Printed Circuit Board, 國立中央大學工業管理研究所碩士論文, 2008
17. R. W. Kay, E. de Gourcuff and M. P. Y. Desmulliez, Stencil Printing Technology for Wafer Level Bumping at Sub-100 Micron Pitch using Pb-Free Alloys, 2005 Electronic Components and Technology Conference
-69-
Appendix 1
MIL-STD/MIL-STD-883E method 2019.8, the die shearing test criteria
-70-
Appendix 2
The environment requirement for electronic component