Tutorial # 68
Dan Baldwin, PhD, Paul Houston, and Brian Lewis
November 2006
An extended abstract of the paper “Effects of Plasma Pretreatment on Flip Chip and CSP substrate-level assembly yield and reliability” by Dan Baldwin, PhD, Paul Houston, and Brian Lewis, presented at SMTAI 2006.
A comprehensive study of the influence of plasma pretreatment on assembly yield and reliability for solder flip chip in package and for CSP assembly to SIP modules shows that plasma pretreatment can lead to significant differences in assembly yield, underfill coverage and thermo-mechanical reliability.
After pre-assembly plasma treatment of chips, substrates, and packages, the devices were assembled with eutectic solder and no-clean flux. Both the flip chip and CSP device assemblies were then underfilled with fast flow, snap cure underfill adhesive. Baseline assemblies without plasma pretreatment were similarly prepared for comparison testing. Various plasma pretreatment techniques and two substrate surface finishes were analyzed.
ASSEMBLY RESULTS
Visual inspection of the plasma-treated flip chip and CSP assemblies revealed higher, more uniform fillet shapes, especially in the corners of the components, compared with the untreated devices. Figure 1 is a is an edge micrograph of a typical untreated flip chip assembly, showing a lower underfill fillet formation at the corners. Figure 2 shows a typical pre-treated flip chip assembly.
Figure 1. Untreated flip chip assembly.
Figure 2. Plasma-treated flip chip assembly.
Figures 3 and 4 are a pair of similar photos of untreated and plasma-treated CSP assemblies, showing higher,larger fillets for the plasma-treated device.
Figure 3. Untreated CSP assembly.
Figure 4. Plasma-treated CSP assembly.
Underfill flow times for the plasma-treated flip chip samples were 12% to 20% faster than for the untreated samples. The flow times for the plasma-treated CSPs were 55% faster than the flow times for the untreated samples.
Scanning acoustic microscopy (CSAM) of the underfilled flip chip devices showed no significant differences between the plasma-treated and untreated samples in underfill uniformity and void formation during processing.
No statistical difference in interconnect yield was found between the plasma treated and non-plasma treated flip chip assemblies on gold (ENIG) surface finished boards. However, plasma-treated flip chip assemblies to boards with organic (OSP) coatings showed a lower solder interconnect yield than with boards not subject to plasma treatment.
This is likely caused by a negative interaction between the Argon plasma and the organic surface protectant over the Cu pads, suggesting that plasma treatment of OSP finished boards should be avoided. CSP assembly yields showed no significant difference between plasma-treated or untreated boards.
RELIABILITY RESULTS
A detailed reliability screening was performed on the assemblies to assess the impact of plasma pretreatment on long term reliability of the assemblies. Plasma-treated flip chip assemblies on gold-finish substrates showed a 70% improvement in reliability results over untreated samples.
In contrast, plasma-treated flip chip assemblies onto OSP finish substrates had 43% lower reliability compared with the untreated assemblies. For CSP components, the reliability results were generally the same for treated and untreated assemblies.
The decreased reliability of flip chips on OSP-finished boards is likely caused by a negative interaction between the Argon plasma and the organic surface protectant, as described above.
SUMMARY
- Plasma treatment significantly improved underfill fillet uniformity and underfill flow time for both flip chip and CSP.
- Plasma treatment had no effect on assembly yield for flip chip on gold finish boards.
- Plasma treatment of flip chip assemblies on OSP finish boards reduced both assembly yield and reliability, and is not recommended.
- Plasma treated flip chip assemblies on gold finish boards showed a 70% reliability improvement over untreated samples.
- CSP assembly yields and component reliabilities showed no significant difference between plasma-treated and untreated samples.
FOR MORE INFORMATION
The full 10-page paper, including experimental details, further photographs, and full results may be found in the proceedings of SMTA International, September 2006, available from SMTA.
Author Contacts
Daniel Baldwin, Ph.D.
Advanced Assembly Process Technology – AdAPT Laboratory
The George W. Woodruff School of Mechanical Engineering
Georgia Institute of Technology, Atlanta, GA 30332-0405
Paul Houston and Brian Lewis
Engent, Inc. – Enabling Next Generation Technologies
Norcross, Georgia