This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY 1
Reliability of Copper Through-Package
Vias in Bare Glass Interposers
Kaya Demir, Andac Armutlulu, Venky Sundaram, P. Markondeya Raj, and Rao R. Tummala, Fellow, IEEE
Abstract— Thermomechanical reliability of copper-plated
through-package vias (TPVs) in ultrathin bare glass interposers
was investigated through modeling, design, fabrication, reliability
characterization, and failure analysis. Finite-element models
were developed to analyze stress and strain distribution in
TPV structures, and to obtain design guidelines for reliable
TPVs. In order to experimentally validate the predictions of
simulations, bare glass substrates of 100 μm thickness with vias
of 30 μm diameter at 120 μm pitch were metallized using Ti/Cu
sputtering, followed by patterning and electroplating. Cu TPV
daisy chains were fabricated and subjected to thermal cycling
test between -55 °C and 125 °C to assess their thermomechanical
reliability. Detailed cross-sectional analysis was also carried out
by scanning electron microscope imaging of TPV cross sections.
No electrical failures were detected in the Cu TPV chains. Failure
analysis revealed copper delamination and crack formation in
glass. The experimental reliability results are consistent with the
thermomechanical models. Design and process recommendations
are provided based on the modeling and experimental results.
Index Terms— Failure mechanism analysis, finite-element
modeling (FEM), glass interposer, reliability, through-package
vias (TPVs).
I. I NTRODUCTION
T
HE increasing demand for memory-to-logic bandwidth
for both mobile and high-performance applications has
been driving continual evolution in electronics packaging
from traditional 2-D multichip packages to 3-D packages [1].
To achieve such vertical integration at low cost, low power,
and high performance, 2.5-D and 3-D interposers, as illustrated
in Fig. 1, have been proposed as a compelling alternative
to 3-D IC stacks with through-silicon vias (TSVs) in logic
and memory [2]. These ultrathin interposers can facilitate
short and high-density vertical interconnections between logic
and memory to achieve high bandwidth. However, the inter-
poser material is required to have ideal electrical, thermal,
mechanical properties, and low-cost manufacturing processes
for ultrasmall through-package via (TPV) formation.
As an interposer substrate material, organic laminates are
commonly preferred because of their low cost and compati-
bility with existing manufacturing infrastructure [2]. However,
the primary disadvantage of organic interposers arises from
their limited fine-pitch I/O capability due to three reasons:
1) dimensional instability; 2) high CTE; and 3) high warpage
Manuscript received October 3, 2016; revised January 26, 2017; accepted
March 4, 2017. Recommended for publication by Associate Editor
H.-C. Cheng upon evaluation of reviewers’ comments. (Corresponding
author: Kaya Demir.)
K. Demir, V. Sundaram, P. M. Raj, and R. R. Tummala are with the
Packaging Research Center, Georgia Institute of Technology, Atlanta, GA
30332 USA (e-mail: kaya@gatech.edu).
A. Armutlulu is with ETH Zurich, 8092 Zurich, Switzerland.
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TCPMT.2017.2691407
Fig. 1. Cross-section schematic of proposed (a) 2.5-D and (b) 3-D interposer
packages.
during assembly [3]. To mitigate these shortcomings of organic
interposers, silicon interposers are being developed and fab-
ricated with standard back-end-of-line wafer processes to
achieve required wiring and I/O density [4]–[6]. However, dis-
advantages of silicon interposers are: 1) expensive fabrication
processes, and 2) high electrical loss in RF domain due to the
semiconducting silicon.
To address the disadvantages of organic and silicon inter-
posers, glass is proposed as an alternative substrate mate-
rial due to its: 1) superior mechanical properties of glass
compared to organic laminates, and 2) lower material cost
and electrical loss compared to silicon [7]. Previous studies
demonstrated superior electrical performance of RF passives,
integrated RF modules, and high-bandwidth 2.5-D interposers
with large-area panel-based glass interposers offering cost
benefits [8]–[11].
In spite of these advances, reliability of copper-plated TPV
in glass faces two key challenges: 1) brittleness of glass, whose
strength further reduces in contact with water [12], [13] and
by formation of defects during handling and via formation,
and 2) high coefficient of thermal expansion (CTE) mismatch
between glass (3 ppm/K) and copper (17 ppm/K), leading to
thermomechanical stresses that can cause crack propagation in
glass, interfacial delamination of copper and glass, and plastic
deformation in copper. Therefore, there is a compelling need to
study TPV reliability to develop design guidelines to achieve
reliable TPV structures.
Polymer lamination on glass is being pursued as a
solution to address challenges with glass handling and
metallization [14]. Low-modulus polymer has at least five
purposes: 1) control glass damage during handling; 2) inhibit
moisture contact to glass; 3) filling ultrasmall defects on
glass surfaces to increase its strength; 4) act as a buffer layer
2156-3950 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.