Thermoelectric MEMS Coolers C. Hilbert, R. Nelson, J. Reed, B. Lunceford, A. Somadder, K. Hu Microelectronics and Computer Technology Corporation (MCC) 3500 West Balcones Center Drive, Austin, Texas 78759 hilbert@,mcc.com U. Ghoshal IBM Austin Research Laboratory, Austin, TX 78758 ghoshal@us.ibm.com Abstract To date the advantages of solid-state coolers such as high reliability, low-mechanical noise, and localized temperature control have often been negated by their inefficiency and the frequent need for multiple stages to achieve the desired temperature difference. While recent research on novel materials and low-dimensional structures raises the hope for improved performance, additional innovations are required to make solid-state cooling competitive. MCC is developing a novel implementation and operational paradigm for solid- state coolers based on transient operation of thermoelectric (TE) coolers and micro-electro-mechanical switch (MEMS) technology. Application of a short transient pulse on top of a steady state bias to a correctly designed TE cooler results in a temporary additional temperature drop. MEMS switches can exploit this effect by providing a thermal switch between the cold end of the TE cooler and the device to be cooled, only connecting them during repeated transient pulses. The objective of this contract is to demonstrate the feasibility of the TE MEMS cooler concept by fabricating a prototype and achieving a cold temperature lower than the one achievable by steady-state operation of a thermoelectric cooler. In this presentation we will describe the general concept and report on the progress made to date. Introduction Detector/sensor performance and signal processing benefit significantly from reduced temperature operation. Solid-state coolers provide this benefit while offering the advantages of high reliability, low mechanical noise, and localized cooling. Furthermore, thermoelectric cooling enables the system designer to tightly control the device temperature to a specific value or range over varying power dissipation levels by using active sensing and feedback. In detector and sensor applications such as the thermal management of charge-coupled devices (CCDs), infrared detectors and imaging sensors, low-noise amplifiers and temperature stabilization of laser diodes and optoelectronic circuits, improved thermoelectric coolers will provide operating efficiencies and achieve lower temperatures with a single thermoelectric stage. In these “spot cooling” applications the maximum temperature differential that can be achieved is frequently one of the most important parameters. Military systems such as electronic warfare and radar will benefit from more advanced and reliable coolers in a variety of instruments. In digital processing increasing circuit performance by continuing the CMOS device scaling (“Moore’s law”) is becoming increasingly cost prohibitive. Improved solid-state cooling methods will enable higher power dissipation at the device level for a given junction temperature or lower operating temperatures at a given power level. To date however, the advantages of solid-state coolers have often been negated by their inefficiency (low coefficient of performance) and the frequent need for multiple stages to achieve the desired temperature difference. While recent research on novel materials and low-dimensional structures raises the hope for improved performance, additional innovations are required to make solid-state cooling competitive. This effort investigates a novel mode for thermoelectric coolers in order to increase their performance. The proposed approach combines pulsed operation of thermoelectric (TE) coolers, and micro-electro-mechanical switch (MEMS) technology. The thermoelectric MEMS cooler concept was first proposed by Dr. Uttam Ghoshal of IBM [l]. Technical Concept The cooling performance of thermoelectric materials is determined by the dimensionless figure of merit ZT defined by the relation [2]: where S is the Seebeck coefficient, (J is the electrical conductivity, and h is the thermal conductivity. This figure of merit directly impacts both the minimum temperature that can be achieved at the cold end of the thermoelectric junction and the efficiency (i.e. coefficient of performance) of the cooler. Unfortunately materials with a large Seebeck coefficient, high electrical conductivity and low thermal conductivity are difficult to find. Among traditional thermoelectric materials doped semiconductors based on bismuth telluride have one of the best performances with the figure of merit ZT close to I at room temperature. The drive to increase ZT has been hampered by the difficulty of increasing the electrical conductivity (J without increasing the thermal conductivity of the material, or reducing the thermal conductivity without affecting the electrical conductivity, all while maintaining a high Seebeck coefficient. Z=S201h (1) 0-7803-5451-6/00/$10.00 02000 IEEE 117 18th International Conference on Thermoelectrics (1 999)