FRACTURE OF PIEZOELECTRIC MEMS STRUCTURES D.F. Bahr, A.L. Olson, M.S. Kennedy, K.R. Morasch, C.D. Richards, and R.F. Richards School of Mechanical and Materials Engineering, Washington State University, Pullman, WA 99164-2920, USA ABSTRACT In thin film systems subject to mechanical strains, such as those experienced in Micro Electro Mechanical Systems (MEMS), failure often occurs via fracture mechanisms. Both through thickness cracking and interfacial delamination leading to failure of the device or layer are possible failure modes. Measuring the stress at which fracture occurs in these thin film systems requires testing methods amicable to both the small scale of the films as well as the complex relationship between the mechanical properties of the film and the substrate. This paper will focus on piezoelectric oxide film systems for MEMS, a piezoelectric ceramic (PZT) on platinum. The behavior of PZT membranes at high deflections and strains for use as power generators will be examined. Experiments have been conducted with a bulge tester to obtain pressure- deflection relationships and residual stresses of the piezoelectric membranes. Comparisons are made between bulge testing to fracture with a nanoindentation method developed for testing fracture of hard coatings on soft substrates. The PZT films are fabricated by solution deposition and heat treating, which leads to significant tensile stresses in the film. The strain at failure in these PZT films, when measured with nanoindentation, is approximately 1.6 GPa, while the applied stress at failure during bulge testing is near 130 MPa. These order of magnitude differences in stresses required for failure between the bulk and the nanoscale tests will be discussed in terms of differences in flaw population as well as the effects of residual stresses. In particular, when the PZT films are first deposited as amorphous films the strength is near 3 GPa when measured via nanoindentation. The crystallization process, required to form the perovskite crystal structure leads to either a change in fracture toughness or a change in the flaw size. Based on the lowering of applied stresses required for fracture within increasing volume sampled, the flaw size is the likely controlling factor in the fracture of these MEMS structures. 1 INTRODUCTION The need for miniaturized power sources for MEMS and microelectronics devices has long been recognized. Among the micro-scale concepts to generate electrical power are fuel cells, static heat engines and dynamic heat engines. Recent work at Washington State University has led to the development of a micro heat engine that incorporates a thin-film piezoelectric membrane generator of lead zirconate titanate (PbZr x Ti 1-x O 3, PZT) (Whalen [1]). PZT is attractive for MEMS applications due to its high piezoelectric and electromechanical coupling coefficients (Xu [2]), and is widely used in sensing and actuating MEMS applications (Polla [3]). The structure of the thin- film piezoelectric membrane generator chosen for the micro engine is a simple two-dimensional sandwich structure similar to that used for pressure transducers and ultrasonic transducers (Baborowski [4]). Two methods of evaluating the mechanical response of this MEMS devices are nanoindentation and bulge testing. Nanoindentation can be used to measure mechanical properties of thin films. However, when thin films that are significantly harder than the underlying materials are strained, they rarely fail via mechanisms of plastic deformation. Instead, complex fracture behavior is the likely failure mechanism. As outlined by Page and Hainsworth [5], there are several possible mechanisms of failure that can be observed by examining the load – depth curve during nanoindentation, leading to their description of using this curve as a “mechanical fingerprint” of a material. Bulge testing has been used in previous studies to examine the elastic and plastic properties of thin films, as well as characterizing the residual stress in the films (Vlassak [6]). Bulge testing can also be used to pressurize membranes to failure. This provides an interesting mechanism to compare a relatively bulk fracture test (bulge testing of membranes on the order of