• PDF / 172,840 Bytes
• 9 Pages / 612 x 792 pts (letter) Page_size
• 17 Downloads / 216 Views

DOWNLOAD

REPORT

---

U4.7.1

Effects of Varying Mean Stress and Stress Amplitude on the Fatigue of Polysilicon

H. Kahn, R. Ballarini*, and A.H. Heuer Department of Materials Science and Engineering, Case Western Reserve University Cleveland, OH 44106 USA *Olin College of Engineering Needham, MA 02492 USA

ABSTRACT

Polycrystalline silicon (polysilicon) single edge-notched fatigue specimens with micrometer-sized dimensions were macromachined and subjected to cyclic loading using an integrated electrostatic actuator. The effects of fatigue were determined by comparing the monotonic bend strength measured after cyclic loading to the monotonic bend strength of specimens that received no cycling. Both strengthening and weakening were observed, depending on the levels of mean stress and fatigue stress amplitude during the cyclic loading. Monotonic loading with similar stress levels prior to bend strength measurements had no effect on measured bend strength. Possible physical mechanisms responsible for this fatigue behavior are discussed.

INTRODUCTION

Polysilicon deposited by low-pressure chemical vapor deposition (LPCVD) is a brittle material at room temperature. However, dynamic fatigue – delayed fracture under applied cyclic stresses – has been reported for micrometer-scale polysilicon specimens [1-5]. Fatigue in polysilicon has been observed for fully reversed (tension/compression) stress cycling [2,3] and for tension/zero stress cycling [4,5]. For both cases, the lifetime under high-cycle fatigue depends only on the number of cycles, not on the total time or the frequency of the test, for testing frequencies ranging from 1 Hz through 40 kHz [4] This implies that dynamic fatigue is

U4.7.2

mechanical in origin, and is not caused by time-dependent environmental effects such as stress corrosion, oxidation, or other chemical reactions. We recently demonstrated that low-cycle fatigue strengths are strongly influenced by the ratio of compressive to tensile stresses in the loading cycle, but not by the ambient (air or vacuum) [1]. This study was achieved by applying a (tensile or compressive) mean stress to a fatigue specimen, and then slowly increasing the amplitude of the cyclic load (the difference between the maximum and the minimum load in the cycle) until fatigue fracture occurred. Therefore, fatigue fracture was obtained at a specific combination of applied mean stress, σm, and fatigue stress amplitude, ∆σ. It is the goal of this paper to perform fatigue tests on polysilicon specimens with independently varied σm and ∆σ in order to assess the effects of these two parameters on fatigue behavior.

EXPERIMENT

The micromachined polysilicon specimen used in this investigation is shown in Fig. 1. It is integrated with an electrostatic comb-drive microactuator whose operation and fabrication have been described previously [1,2]. The devices used in this study were fabricated from 5.7 µm thick LPCVD polysilicon multilayers [6] annealed in nitrogen at 1100°C for one hour to reduce residual stresses to less than 10 MPa. The resulting microstr