Adaptive, Cricket-Inspired Artificial Hair Sensor Arrays R. K. Jaganatharaja, N. Izadi, J. Floris, T. S. J. Lammerink, R. J. Wiegerink and G. J. M. Krijnen Abstract—In this work, we present a model for our biomimetic, artificial hair sensors, to analyze the sensitivity dependence on their structural and geometrical parameters. Based on this model, feasible design improvements to achieve an increased sensitivity are discussed and a figure of merit to evaluate sensor performance is defined. Also, we discuss the results of a novel approach to implement adaptive sensor arrays through DC-biasing based on the electrostatic spring-softening effect. Experimental results show a clear theoretical accordance and tunability of system’s resonance frequency, providing opportunities for frequency focusing and selective sensitivity. Index Terms—adaptivity, artificial hair sensors, biomimetics, spring-softening I. INTRODUCTION Crickets have, quite often, been a subject of common interest to biologists and engineers. They have evolved with a pair of special flow-sensitive appendices called cerci with numerous mechano-receptive filiform hairs of different lengths, distri- buted on the surface. Fig. 1. Mechano-receptive hairs found on cerci of crickets. [SEM image courtesy by Jerome Casas, IRBI, Université de Tours.] Manuscript received October 1, 2007. This work was done as a part of Bio- EARS (funded by STW/NWO) and CILIA projects. The Customized Intelligent Life-Inspired Arrays (CILIA) project is funded by the Future and Emergent Technologies arm of the IST Program. All the authors are with the Transducer Science and Technology group, MESA+ and IMPACT Research Institutes, University of Twente, Postbus 217, 7500 AE Enschede The Netherlands (phone: 053-489-4438; fax: 053- 489-3343; email: r.kottumakulal@ewi.utwente.nl). These filiform hairs are extremely sensitive to acoustic signals, down to thermal noise levels [1], enabling them to identify and escape from approaching predators. Each filiform hair has a mechano-sensitive neuron at its base which fires a neuro-signal whenever there is a flow-induced deflection on the hair, see Fig. 1. Inspired by crickets and making use of technical advancements in MEMS techniques, SU-8 based artificial hair sensor arrays were successfully implemented recently [2]. Ways to improve the sensitivity of these artificial hair sensor arrays have been demonstrated; increasing the hair length and arranging the sensors on an artificial cercus-like platform that can be assembled to facilitate 3D-flow sensing, see Fig. 2. Fig. 2. SEM images of the realized biomimetic hair sensor arrays, arranged on the artificial cerci-like substrate. In this work, we present a model for our artificial hair sensor to further study and optimize the effects of structural and geometrical parameters on the sensitivity and a figure of merit for our sensors is defined based on this model. Parallel to this, we present a new approach to develop adaptive hair sensors which are tunable with respect to the best frequency and sensitivity [3]. II. PRINCIPLE OF OPERATION The artificial hair sensor is based on a differential capacitive sensing technique. Drag-torque due to the air flow, picked-up by the SU-8 hair, results in a membrane tilt. This flow-induced tilting of silicon (rich) nitride (SiRN) membrane with electrodes on top, causes a change in the sensor capacitance, with respect to the bottom electrode (bulk silicon), see Fig. 3. The capacitance change is read-out using charge amplifiers and synchronous detection [2]. 589