Materials Science and Engineering A 493 (2008) 299–304 Environmental influence on interface interactions and adhesion of Au/SiO 2 M.S. Kennedy a , N.R. Moody b , D.P. Adams c , M. Clift b , D.F. Bahr a,∗ a Washington State University, Pullman, WA, United States b Sandia National Laboratory Livermore, CA, United States c Sandia National Laboratory Albuquerque, NM, United States Received 9 April 2007; received in revised form 27 August 2007; accepted 24 September 2007 Abstract The mode I interfacial adhesion energy for as-deposited Au/SiO 2 was measured using a stressed overlayer test, and ranged from 0.39 ± 0.09 J/m 2 for spontaneous blisters to 0.37 ± 0.17 J/m 2 for indentation-induced blisters. After these films were heated to 100 ◦ C and 300 ◦ C for 1 h, the interfacial fracture energies increased, to 0.9 J/m 2 and 9.9 J/m 2 , respectively. This was consistent with Au/SiO 2 films aged over an 8-year period, which had a mode I interfacial fracture energy between 1.2 J/m 2 and 1.9 J/m 2 . The blister delamination was monitored over the course of over a year, and exhibited growth after an initial stabilization period. Subsequent testing of delaminations with controlled humidity reproduced this growth mechanism. Changes in interfacial adhesion energies are discussed in light of changes in interfacial chemistry and the exposure of an interfacial crack tip to humidity. © 2007 Elsevier B.V. All rights reserved. Keywords: Adhesion; Humidity; Interface; Interfacial fracture energy 1. Introduction For microelectronic components to be reliable, the delami- nation of films needs to be eliminated over the predicted device usage lifetime. A crack will propagate through an interface and cause delamination when the strain energy of the system exceeds the interfacial fracture energy (adhesion energy) of the interface. To predict conditions under which propagation of a crack will not occur, both the strain energy available in the system and the critical interfacial fracture energy need to be known. Typically, interfacial fracture energies reported in the liter- ature have focused on as-deposited conditions. Depending on the aging mechanisms of the film, however, these initial interfa- cial fracture energies could have little relation to the interfacial fracture energies seen during the life of the interface. Environ- mental changes, such as temperature fluctuations and exposure to humidity, can lead to diffusion of components materials or dif- fusion of environmental species along the interface or through the film. This can lead to less distinct interfaces, chemistry ∗ Corresponding author. Tel.: +1 509 335 8523; fax: +1 509 335 4662. E-mail address: dbahr@wsu.edu (D.F. Bahr). changes at the interface, the development of new phases, and changes in applied stress from thermal expansion mismatch between the film and substrate or the packaging. Studies of metal–oxide systems have tried to define a critical fracture energy for interfacial failure, which reflects the mechan- ical and microstructural contributions of component materials such as the strength and thickness of metal films and crack path locations. Oh et al. noted that in service, the failure of micro- electronic devices from interface fracture occurs more often by subcritical cracking in the bond at stresses below those required for catastrophic failure and that crack growth behavior along the ceramic–metal interfaces is a major issue in the reliability and long-term stability of these interfaces [1]. Deterioration of microelectronic systems is typically slow and these systems can withstand anywhere from thousands to millions of service hours before failure [2]. Recent work has shown that the lifetimes of thin film systems can be shortened with increased thermal cycling [3]. In addition, environmental changes such as humidity are often not constant and can fluctuate over time. In order to test systems, methods to accelerate indi- vidual aspects of the aging process are needed. Once the specific failure mechanisms due to environmental factors are pinpointed and tested, researchers will be able to predict and ultimately 0921-5093/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2007.09.081