A CMOS-based Microelectrode Array for Information Processing with Natural Neurons S. Hafizovic, F. Heer, U. Frey, T. Ugniwenko, A. Blau, C. Ziegler and A. Hierlemann Abstract— We report on a complementary-metal-oxide- semiconductor-based system that is capable of bidirectionally communicating (stimulation and recording) with electrogenic cells such as neurons or cardiomyocytes. It is is targeted at investigating electrical signal propagation within cellular networks in vitro. Experiments including the stimulation of neurons with two different spatio-temporal patterns and the recording of the triggered spiking activity have been carried out. The neuronal response patterns have been successfully clas- sified (83% correct classifications) with respect to the different stimulation patterns. It will be demonstrated that information processing using natural neuronal networks may be possible. I. I NTRODUCTION Microelectrode arrays (MEA) [1], [2] are used to extracellu- larly measure the induced voltage on an electrode upon the occurrence of action potentials in electrogenic cells [3]. The obtained electrode signal amplitudes typically range between tens of μV and 1 mV, depending on the cell type. With increasing cell-electrode distance, or in the case of less direct cell-electrode contact, signal amplitudes may become very small [4], [5]. The CMOS-based microelectrode array presented here is capable of bidirectionally communicating (stimulation and recording) with electrogenic cells, such as neurons or car- diomyocytes. Cell recordings and stimulations have been repeatedly and reliably performed, with the signal amplitudes of neurons being up to hundred times smaller than those of cardiomyocytes that have been published previously [6]. II. DESIGN The overall system consists of three major subunits: The core component is a 6.5-by-6.5 mm 2 CMOS (complementary metal oxide semiconductor) chip (Fig. 1) fabricated in indus- trial CMOS technology [2]. The cells are cultured directly on top of the chip. The chip features 128 bidirectional electrodes, each equipped with dedicated analog filters and amplification stages and a stimulation buffer, sampled at 20 kHz with 8 bit resolution. Additionally, temperature sensors, a digital- to-analog converter for stimulation (voltage stimulation, 8 bit, 60 kHz), and a digital interface for data transmission Funding has been generously provided by the European "Information Society Technologies" "Future and Emerging Technologies" program, and the Swiss Bundesamt fuer Bildung und Wissenschaft (BBW), contract number IST-2000-26463. S. Hafizovic, F. Heer, U. Frey and A. Hierlemann are with the Physical Electronics Laboratory, ETH Zurich, Zurich, Switzerland {sadik.hafizovic, fheer, hierlema}@phys.ethz.ch T. Ugniwenko, A. Blau and C. Ziegler are with the Department of Physics, University of Kaiserslautern, Kaiserslautern, Germany {ugniwenko, blau, cz}@physik.uni-kl.de are integrated [7]. The second component is a reconfigurable logic device, which provides chip control, event detection, data buffering and an USB interface capable of processing the 2.56 MSamples/s delivered by the CMOS chip. The third component includes the software that is running on a standard PC and performs data capturing, processing, and visualization (Fig. 2). The advantages of the presented microsystem over state- of-the-art microelectrode arrays [8] include the capability to produce any arbitrary spatio-temporal stimulation pattern, and the availability of a fast electrode reset function that enables to record neuronal signals on a stimulated electrode only 5 ms after stimulation while continuously recording DAC CMOS-Al contacts In-house Platinum Pt-Electrode 15 μm diameter Repeating electrode unit 250 μm 2 Temperature sensor Digital control 16 ADC 8x16 electrode array at 250 μm pitch Chip size: 6.5 x 6.5 mm Fig. 1. Micrograph of the microelectrode array chip and close-up of one electrode repeating unit. Left: The chip features an 8-by-16 electrode array in the center, and 16 A/D-converters and the digital block at the right. Right: Close-up of the 128-times-repeated circuitry unit. Proceedings of the 3rd International IEEE EMBS Conference on Neural Engineering Kohala Coast, Hawaii, USA, May 2-5, 2007 SaE1.4 1-4244-0792-3/07/$20.00©2007 IEEE. 692