Reliability Analysis of Bumping Schemes under Chip Package Interaction
Sri Ramakanth Kappaganthu
1
, Aditya Karmarkar
1
, Xiaopeng Xu
2
,
Karim El Sayed
2
, Ibrahim Avci
2
, Vikas Chawla
2
, Bikash Mishra
2
,
Andrey Kucherov
2
, Weixing Zhou
2
, Mark Johnson
2
, Pratheep Balasingam
2
1
Synopsys (India) Private Limited, Hyderabad, Andhra Pradesh 500032, India
2
Synopsys, Inc., Mountain View, CA 94043, U.S.A.
ABSTRACT
Reliability analysis for three bumping configurations is
performed under typical chip package interaction. A
sequential submodeling technique is employed to capture
stress evolution during entire package assembly process.
Mechanical stresses are assessed in various regions around
bumps to determine the optimal bumping scheme with the
minimal reliability risk. Underfill material property impact
on package reliability is also examined. This study
provides important guidelines to design robust bumping
configurations with fine-tuned material properties.
INTRODUCTION
Packaging and assembly operations for advanced
electronics involve high temperature steps that introduce
thermomechanical stresses in the package due to material
property differences. Chip package interaction (CPI)
becomes a significant cause of reliability concern due to
these residual stresses [1], [2]. Previous studies have
revealed that the stress concentrations caused by CPI can
lead to low-k dielectric cracking, interlayer delamination
and bump failures [3], [4]. Chip package interaction has
been observed to strongly depend on the bumping scheme
that includes the conventional bumping configuration with
spherical solder bumps, and new bumping configurations
with copper pillar bumps [5], [6]. Hence, in order to
design a robust electronics package, it is important to
analyze the reliability of different bumping schemes under
chip package interaction.
In this paper, reliability analysis is performed for three
different bump configurations under typical chip package
interaction. These bumping configurations consist of
spherical solder bump, copper pillar with a contact pad
and copper pillar landing directly on the substrate trace. A
sequential submodeling technique is used to examine the
mechanical stress evolution at each package assembly step
[7]. Underfill material properties are varied to examine
their impact on the mechanical stress distribution around
bumps. An advanced FEM based 3D TCAD simulator is
employed to perform the sequential package assembly
simulation [8].
SIMULATION METHODOLOGY
A. Finite Element Model
An advanced electronics package is a large structure
that consists of multiple elements at various length scales.
These scales represent a difference of about five orders of
magnitude from the package to backend level. To
effectively simulate the package assembly flow at these
length scales, a multi-level multi-scale submodeling
simulation approach is used. A two stage submodeling
simulation that consists of a global model and a submodel
is used in this paper. Fig. 1 (A) shows the submodeling
simulation routine used here. The global model is
approximated as a thin 3D diagonal strip extracted from a
package structure as shown in Fig. 1 (B). The submodel
simulates the bump and the surrounding region with
back-end passivation layers. Fig. 1 (C) shows various
geometries and dimensions in the global model.
Fig. 1: (A) Submodeling routine; (B) Diagonal strip model; (C) Chip
package stack near die corner.
Fig. 2: (A) Solder bump; (B) Cu pillar; (C) Cu pillar on trace; (D) Far
back end layers; (E) Averaging volumes in bump and solder joint; (F)
Averaging volume in ELK
Fig. 2 shows the three different bumping
configurations analyzed in the submodel, along with the
back-end layers. Fig. 2 (A) shows a bump structure that
uses a spherical solder bump, Fig. 2 (B) shows a bump
structure with a copper pillar bump and Fig. 2 (C) shows a
bump structure where the copper pillar bump lands
directly on the substrate trace. Fig. 2 (D) shows a close-up
view of the far back end layers indicated by the dotted
rectangle in Fig. 2 (B). The far back end layers are the
same in all the bumping schemes. Black boxes in Fig. 2
(E) indicate the locations where stresses are averaged in
3
978-1-4799-5018-8/14/$31.00 ©2014 IEEE 151