Improving Manufacturing Quality of Electronic Devices Fabricated Using Pre-Soldering Deformation

Afeef Ahmad, Alejandro Calvo, Salma El-Azab, Hannah FowlerAdvisors: Dr. John Blendell, Dr. Carol HandwerkerIndustrial Sponsor: Dr. Peng Su

Experimental Procedure

Project Background Microstructure

Shear Testing

Sandwiched Board

Discussion

Recommendation

Acknowledgements

The microstructures and shearing data for thedeformed and undeformed solder balls were similar.There were no unusual defects in the intermetallicinterface for the deformed solder balls. Deformingthe solder balls prior to sandwiching should not haveany impact on solderability. It is thus recommendedto deform solder balls prior to sandwiching. Furthertesting is needed to determine exact deformationdistance for specific loads, as well as a bettermethod for measuring the height of the solder balls.

Special thanks to: Travis Dale, Jameson Root, Alyssa Yeager, Alexandra Burch, Raheleh Rahimi, Sukshitha Achar, Tim Vanmeter, and Laura Ripoli

This work is sponsored by Juniper Networks in Sunnyvale, CA

Ball grid array (BGA) circuit boards contain two boards with copper solder pads that are connected through solder balls that provide both electricalconnections and structural. BGA boards warp during reflow due to the different CTE values of components which results in solder ball defects leadingto poor connectivity. The solder balls were deformed prior to reflow at 1 N, 2 N, and 3 N per ball. The boards were reflowed with a blank board placedon top. Cross-sectional analysis using SEM was conducted to identify microstructural changes at the intermetallic Cu-Sn interface. The boards weresheared at constant velocity to failure to observe differences in brittle or ductile failure between deformed and undeformed samples. No signs ofsignificant microstructural changes or defects towards the interface or deformed surface were observed. Mathematical and optical analysis of deformedarea showed no correlation between load and ball displacement. Shear testing showed no difference between fracture surface of deformed andundeformed, suggesting similar mechanical properties.

Ball grid array (BGA) warpage occurs during reflowand is caused by a difference in thermal expansionbetween the epoxy printed circuit board (PCB) andthe solder balls. BGA warpage can cause defects insolder joints, such as voids, cracks, and the head-in-pillow defect. Defects in solder joints cause failurein PCB connections and thus in products.

SAC305 is used in Juniper Networks’ products.SAC305 forms a Cu6Sn5 intermetallic layer with theCopper pad on the circuit board. Understanding themicrostructures and properties of SAC305 is crucialto maintaining proper reflow conditions and formingstrong connections. Solder balls have small sizescales and thus display different mechanicalbehaviors (such as flow stress and yield stress). Tinis also anisotropic, and the various orientations ofgrains impact mechanical properties.

Materials UsedSolder balls of 2 alloys 650 𝜇𝜇m in diameter wereused;● 96.5%Tin - 3%Silver - 0.5%Copper (SAC305)● 95.5%Tin - 4%Silver - 0.5%Copper (SAC405)Two types of 12x12 mm solder pad PCBs were used;● Electroless Nickel Immersion Gold (ENIG)● Organic Solderability Preservative (OSP).

Test Vehicle PreparationEight balls were reflowed at thecorners (two on each) of eachboard using Alpha OM-338paste flux to create the BGAtest samples. Each pair ofsolder balls was either leftundeformed, or deformed at 10,20, or 30 N using a LecoMicroindenter and an Instronload cell.

Deformation AnalysisOptical images of deformedareas were used to calculatedisplacement. The deformedPCBs were “sandwiched” byreflowing another PCB on topof the existing solder. SEMimages were taken of solderballs to determine effect ofdeformation on microstructureand intermetallics.Shear Testing

The sandwiched boards were sheared at a constantvelocity to study structural integrity. A motor pulledone board at 2 𝜇𝜇m/s on one axis, and load vsdisplacement was measured to determine strengthof the solder. Optical images were taken of the shearsurface and categorized as brittle or ductile failure.

MSE 430-440: Materials Processing and Design

Undeformed SAC305 ENIG Deformed SAC305 ENIG at 1.25 N

Undeformed SAC305 OSP Deformed SAC305 OSP at 1.25 N

Deforming the solder balls did not have aquantifiable impact on the solder ball microstructure.There were no unusual defects in the deformedsolder ball interfaces.

Figure 2. Reflowed singleBGA with 8 solder balls.Penny for scale.

Sandwich structures made of deformed balls displaysimilar microstructures and reflow properties. Anyporosity and defects were observed in bothdeformed and undeformed boards. The solderabilityof the ball was not impacted by deformation.

Figure 1: Head-in-pillow defect (1. Bušek, D., et al, 2. Arazna, A.)

Figure 6. Load versus deformed displacement graphs for (a) SAC305 ENIGand (b) SAC305 OSP solder balls. Average displacements are shown inred. No trend can be observed.

(a) (b)

(a) (b)

Figure 5. Micrographs of (a) an undeformed, sandwiched SAC305 ENIGboard and (b) a deformed, sandwiched SAC305 ENIG board. Solder joint (b)was deformed at 1.25 N prior to forming the joint. The micrographs show thatthe undeformed ball and the deformed ball form intermetallics with the topboard.

Figure 7. Optical images of 2 um/s constant velocity sheared sandwichedENIG solder balls showing (a) ductile fracture, (b) brittle fracture and (c)brittle fracture after initial ductile slip.

(a) (b) (c)

Deformed samples showed no signs of abnormal orbrittle failure compared to undeformed samplesduring shearing. This was expected since both hadsimilar microstructures. Furthermore, the deformedsamples were reflowed during sandwiching, relievingthe internal stresses and resetting the solder balls totheir original undeformed state.

Ag3Sn dendrites

Cu6Sn5 Intermetallics

Cu solder pad

The fracture surface images of deformedsandwiched samples after shear testing showed nosignificant brittle failure as compared to theundeformed samples.

Cu solder pad

Ag3Sn dendrites

Cu6Sn5 Intermetallics

Cu solder pad

(Cu,Ni)6Sn5 intermetallic ENIG surface finish

Ag3Sn dendrites

Porosity

(Cu,Ni)6Sn5 intermetallicENIG surface finish

Cu solder pad

Ag3Sn dendrites

Porosity

The stress vs strain plots for deformed andundeformed samples showed no significantdifference or trend. Any variation was likely causedresidual super glue left on test vehicle, tilting thesamples off-axis.

The undeformed and the deformed solder balls hadsimilar microstructures. No unusual defects, such ascracking or delamination at the intermetallic interfacebetween the solder ball and the solder pad wereobserved. This suggests that the force applied to thesolder balls was enough to deform the top of the ball,but not enough to damage the interface. Theseresults are positive for future applications where ahigher number of solder balls will be deformed.

Figure 3. Geometric diagramof how the displacement (X)was calculated based on thedeformed diameter (C) foundoptically, and radius (R).

Solder ball

Solder paste

(a)Figure 4. Stress vs strain curves from shear testing of deformed andundeformed, sandwiched SAC305 (a) OSP and (b) ENIG

(b)

Calculated displacement values, assuming a uniformsphere and an even distribution of displaced mass,showed significant variation, possibly due to initialheight differences. If a large solder ball is adjacent toa smaller solder ball, then the smaller will deformless. A viable method for measuring these heights isnecessary to determine planarity.