Board-level shear, bend, drop and thermal cycling reliability of lead-free chip scale packages with partial underfill: a low-cost alternative to full underfill

Full capillary flow underfill (FCFU) has been proven to be effective in improving the board-level mechanical reliability of lead-free (LF) area array packages (AAPs). However, the FCFU may have negative effects on the thermal cycling reliability of AAPs depending on the material properties of underfills, including coefficients of thermal expansion, glass transition temperature, modulus, and adhesion strength. In addition, the increased cost, time-consuming processes, and poor reworkability caused by the application of FCFU have also hindered the widespread use of the board-level underfills. In order to address these challenges, a partial capillary flow underfill (PCFU) or corner-only underfill approach was developed. However, data are scarce for board-level solder joint reliability of LF AAPs with PCFU, especially for portable electronics applications. In this paper, the overall reliability of LF chip scale packages (CSPs) with FCFU and PCFU was comparatively studied using the AAP-to-board interconnection shear test, monotonic 3-point bending test, vertical free drop test, and thermal cycling test. One set of non-underfilled CSP assemblies was tested as the control. The test results indicated that the mechanical performance of underfilled CSPs was significantly enhanced compared to the CSPs without underfill, especially for drop reliability. However, the characteristic life values of CSPs with FCFU and PCFU during the thermal cycling test were reduced by 15% and 8%, respectively. The improvement in overall boardlevel solder joint reliability of LF CSPs provided by the PCFU was comparable to that of the FCFU. Hence, partial underfill can be used as a good alternative to full underfill. Failure analysis demonstrated that the dominant failure mode was PCB pad cratering in shear and bending test, and the brittle fracture at the CSP intermetallic compound/solder interface was dominant for all the test groups under drop loading conditions. In contrast, the failure mechan- sms of the underfilled and control boards were different during the thermal cycling test: PCB pad cratering and bulk solder fatigue crack were found in the CSPs with and without underfill, respectively.