A MICRO GAS CHROMATOGRAPH WITH INTEGRATED BI-DIRECTIONAL PUMP FOR QUANTITATIVE ANALYSES Yutao Qin and Yogesh B. Gianchandani Center for Wireless Integrated MicroSensing and Systems (WIMS 2 ) University of Michigan, Ann Arbor, MI 48109, USA ABSTRACT This paper describes a micro gas chromatography (μGC) system that comprises a Knudsen pump with bi- directional capability (KP2), a two-stage preconcentrator- focuser (PCF2) and a separation column. In this valveless system, the bi-directionality of the pump allows flow reversal in the multi-stage preconcentrator. The KP2, PCF2, and separation column are arranged in a 4.3 cm 3 stack, and used with a commercial flame ionization detector. In preliminary experiments, the μGC system demonstrated quantitative separation of benzene, toluene, and xylene (BTX) in ambient room air. INTRODUCTION Typical μGC systems include several components: the preconcentrator, which accumulates vapor analytes and provides vapor injection; the column, which separates the vapor analytes; the gas detector, which quantifies the eluents (retention peaks) from the column; and the gas pump, which provides flow [1-3]. In the simplest architectures, the preconcentrator, column, detector, and pump are connected in series, and operated with unidirectional flow [4-5]. In more complex architectures, valves are used to create opposite flow directions (bi-directional flow) in the preconcentrator during sampling and separation; examples appear in [6-8]. While bi-directional flow is potentially beneficial for a single-stage preconcentrator, it is essential for a multi-stage (or multi-bed) one. A multi-stage preconcentrator has different sorbents packed in its stages that are connected in series (Fig. 1). Weaker stages – those packed with lower surface-area sorbents – are located upstream in the sampling flow and are intended to trap vapor analytes with lower volatility. Stronger stages – those packed with higher surface-area sorbents – are located downstream in the sampling flow, and are intended to trap vapor analytes with higher volatility that pass through weaker stages [9]. The vapor sample must flow from the weaker stages to the stronger stages to permit upstream capture of low volatility analytes during the sampling phase. In the analytical separation phase, the flow is reversed, allowing the low volatility sample to enter the separation column together with the high volatility sample that was captured deeper within the preconcentrator. In conventional systems, the flow reversal is accomplished by valves [6-8]. This effort explores a valveless μGC architecture, which is comprised of a bi-directional Knudsen pump (KP2), a two-stage preconcentrator (PCF2), and a separation column. During vapor sampling, vapor analytes enter the μGC system through the separation column and settle into the PCF2 (Fig. 1). During analytical separation, the flow is reversed by the KP2; the sampled vapor analytes are then thermally desorbed from the PCF2 and separated in the column. A commercial flame ionization detector (FID) is used as the detector for accurate quantification of the bi- directional operation. In the future, a micro gas detector can be integrated for realizing a complete, bi-directional μGC system. DESIGN Bi-Directional Knudsen Pump (KP2) Knudsen pumps operate by thermal transpiration [10]. High reliability and simple configuration make these pumps attractive for integration with µGC systems [5, 11]. One example of a Knudsen pump implementation utilizes nanoporous mixed cellulose ester (MCE) membrane(s) (thickness ≈105 μm, pore diameter ≈25 nm, porosity ≈70%, Millipore, MA) [12]. In the presence of a temperature gradient, thermal transpiration flow is generated from the cold side to the hot side of the membranes. A previously reported bi-directional Knudsen pump used thermoelectric elements to provide reversible temperature gradients and gas flow [13]. In this effort, however, the bi-directional Knudsen pump is implemented simply by integrating resistive heaters on both sides of the MCE membranes. During operation, one of the sides is heated while the other is cooled (by a heat sink or natural convection), providing a temperature gradient. The KP2 consists of four glass dice (Die 1a, 1b, 2a, and 2b, thickness =500 μm) sandwiching a stack of four MCE membranes (Fig. 2). Side-A of the KP2 consists of Die 1a Fig. 1: Concept of multi-stage preconcentrator and the bi- directional operation of the μGC system in this effort. 978-1-4799-3509-3/14/$31.00 ©2014 IEEE 294 MEMS 2014, San Francisco, CA, USA, January 26 - 30, 2014