EFFECT OF SUBSTRATE THICKNESS ON ANCHOR DAMPING IN MEMS DEVICES Gabrielle D. Vukasin 1 , Veronica K. Sanchez 1 , Christopher P. Cameron 1 , Hyun-Keun Kwon 1 , Janna Rodriguez 2 , Ian B. Flader 3 , Yunhan Chen 4 and Thomas W. Kenny 1 1 The Kenny Group, Stanford University, USA, 2 Intel, Santa Clara, CA, USA 3 InvenSense, San Jose, CA, USA and 4 Apple Inc., Cupertino, CA, USA ABSTRACT We present unexpected results showing that thinning the bottom substrate of a resonant MEMS tuning fork resonator decreases anchor damping. We also present findings that the tuning fork experiences more anchor damping when mounting the die with silver paste. This is important for wearable devices where minimizing the volumetric footprint of sensors is paramount. KEYWORDS Resonant MEMS, anchor damping, wearable devices, quality factor. INTRODUCTION MEMS are used as accelerometers, gyroscopes, timing references, etc. for many applications [1]. One increasingly popular application is wearable devices because MEMS have the advantage of smaller footprints and volumes than their macro-sized alternatives [2]. Size is important to wearable devices because their aim is to monitor data metrics of a person while being as minimally invasive and unencumbering to the wearer as possible. This means being as small, light-weight, and least power consuming as possible. One way to adapt a MEMS device for use in wearable technology is to decrease the volume of the device. How does this affect the performance of the device? We study the quality factor of MEMS resonators in order to determine the effect of decreasing the volume on device performance. The quality factor is inversely proportional to the total damping of the resonator. The reciprocal sum of the quality factor due to each damping mechanism is the total quality factor, Q: ଵ ொ =∑ ଵ ொ ೑೎೟೚ೝೞ = ଵ ொ ಸೞ + ଵ ொ ೅ಶವ + ଵ ொ ಲ೙೎೓೚ೝ + ଵ ொ ಲೖ೓೔೐೥೐ೝ + ଵ ொ ೀ೟೓೐ೝೞ (1) This builds on previous work in the community on measuring sources of damping in MEMS resonators [3-5]. Measurements of the quality factor over the temperature range of 80K to 300K are compared with temperature profiles of thermoelastic dissipation (TED), air damping and Akhiezer damping [6]. The double-ended tuning fork (DETF), operated in the mode in Figure 1c, is a device that has two clamped-clamped beams. The expected QTED is calculated [7]  ா஽ = ఘ௖ ೛ ாఈ మ  ೚ ଵା(ఠఛ ೙ ) మ ఠఛ ೙ (2) where E is the Young’s modulus, ߙis the coefficient of thermal expansion (CTE),  ௢ is the average temperature,  is the resonant frequency of the beams, and  ௡ is the thermal time constant. Using this knowledge of  ா஽ and ruling out other damping mechanisms, we can measure anchor damping. EXPERIMENTAL METHOD These devices are sealed using a wafer-scale encapsulation process (0.1 to 1 Torr) [8]. The DETFs are tested in a pressure chamber that allows flowing liquid nitrogen to cool an individual die down to 80K. To measure Q, ringdown measurements of the DETFs are taken as they warmup to 300K over 4-5 hours. We measured Q(T) for DETFs anchored to thin and thick (Figures 1a,b) bottom substrates. After fabrication, part of the wafer was thinned by 420um using a DISCO backgrinder. When performing these experiments, the dies are either silver pasted to the chip carrier, “pasted,” or suspended on four wirebonds used to operate the DETF electrostatically, “floating,” as seen in Figure 2. Previous studies have shown that pasting the die increases anchor damping [3]. Figure. 1: a) Cross-section of thick die, b) cross-section of thin die, with oxide layers in blue (3 um thick), c) flexural mode of double-ended tuningfork, d) SEM of DETF. Figure 2: Cartoons of pasted and floated dies with wirebonds on a gold chip carrier. 978-1-5386-8104-6/19/$31.00 ©2019 IEEE 1843 M3P.093 Transducers 2019 - EUROSENSORS XXXIII Berlin, GERMANY, 23-27 June 2019