A Picoliter-Volume Mixer for Microfluidic Analytical Systems Bing He, ² Brian J. Burke, Xiang Zhang, Roujian Zhang, and Fred E. Regnier* Department of Chemistry, Purdue University, Lafayette, Indiana 47907 Mixing confluent liquid streams is an important, but difficult operation in microfluidic systems. This paper reports the construction and characterization of a 1 0 0 - pL mixer for liquids transported by electroosmotic flow. Mixing was achieved in a microfabricated device with multiple intersecting channels of varying lengths and a bimodal width distribution. All channels running parallel to the direction of flow were 5 μm in width whereas larger 27-μm-width channels ran back and forth through the parallel channel network at a 4 5 ° angle. The channel network composing the mixer was ∼10 μm deep. It was observed that little mixing of the confluent solvent streams occurred in the 1 0 0 -μm-wide, 300-μm-long mixer inlet channel where mixing would be achieved almost exclu- sively by diffusion. In contrast, after passage through the channel network in the ∼200-μm-length static mixer bed, mixing was complete as determined by confocal micros- copy and CCD detection. Theoretical simulations were also performed in an attempt to describe the extent of mixing in microfabricated systems. Microfluidic systems of 50-100- μm channel width are now being described that require mixing as part of an analytical protocol. 1-4 Liquid streams generally enter these systems laterally along one or both sides of a central channel, frequently at different points. Channels of this width are perhaps too large for rapid diffusive mixing and too small to allow installation of a dynamic mechanical mixer. At issue is the rate and degree to which mixing occurs in the micromachined systems noted above. Some type of static mixer 5 capable of substantial lateral transport would seem to be a better alternative than either purely diffusive or mechanical mixing. Because the volume of current microfluidic systems is generally in the range of 1-10 nL/ cm and it would be desirable to effect mixing within 0.1-1 mm of transport distance along a channel, mixing would have to be achieved in a volume of hundreds of picoliters. The question is how to build a static mixer of this volume with a high degree of lateral transport. Static mixing is frequently accomplished in liquid chromato- graphic packed beds. The primary mechanism for mixing in this case results from splitting the stream of liquid moving through the bed into a large number of microstreams traveling between particles at slightly different velocities. 6,7 These velocity differences result from heterogeneity in channel dimensions and mix by a mechanism referred to as eddy diffusion in the chromatography literature. 6 It is also known from the chromatography literature that mixing or dispersion in packed beds is further enhanced by poor mass transfer between these streams and stagnant pools of liquid within porous particles in the bed. The problem with packed-bed static mixers is that they are far more effective in promoting longitudinal than lateral mixing. As a consequence, their mixing efficiency is poor and large volumes of liquid are required for mixing. The mixing require- ments for microanalytical devices 8-12 are far smaller than can be met with these packed beds. One solution to this problem is to merge two liquids to be mixed in many microchannels simulta- neously. 9 Although the streams laminate at the point of confluence, mixing by lateral diffusion occurs with much higher efficiency than in packed beds. Mixers ranging down to 600 nL have been produced in this way. 9 However, capillary electrochromatography columns (CEC) are now being reported that range down to 15 nL in volume. 13 Solvent gradient formation with columns of this size requires a mixer of a few hundred picoliters or less. To date, a microfabricated mixer has not been reported that can effectively mix volumes needed for these miniaturized CEC systems. It was the objective of this work to design, fabricate, and test an electroosmotically driven mixer that was at least a 1000-fold smaller than previously reported mixers. Efficacy of the resulting mixer and the extent to which mixing occurs by diffusion alone in microchannel systems was evaluated. MATERIALS AND METHODS Materials. Photolithography masks, SL-4006-2C-AR3-AZ1350, and 3-in. quartz wafers, QZ-3W40-225-UP, were purchased from † Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320-1799. (1) Jacobson, S. C.; Hergenroder, R.; Moore, A. W., Jr.; Ramsey, J. M. Anal. Chem. 1994 , 6, 4127. (2) Seiler, K.; Fan, Z. H.; Fluri, K.; Harrison, D. J. Anal. Chem. 1994 , 66, 3485. (3) Hadd, A. G.; Raymond, D. E.; Halliwell, J. W.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1997 , 69, 3407. (4) Fan, Z. H.; Harrison, D. J. Anal. Chem. 1994 , 66, 177. (5) The term static mixing is obviously both a misnomer and a contradiction but will be used here because of its broad usage in the literature. Actually, any system in which there is convective transport is in a dynamic state of flux. (6) Farkas, T.; Guiochon, G. Anal. Chem. 1997 , 69, 4592. (7) Terry, S. C.; Jerman, J. H.; Angell, J. B. IEEE Trans. Electron Dev. 1979 , ED-26 ( 12) , 1880. (8) Branebjerg, J.; Birgit, F.; Gravesen, P. Proceedings of the μ-TAS ′94 Workshop; Kluwer: London, 1994. (9) Bessoth, F.; deMello, A.; Manz, A. Anal. Commun. 1999 , 36, 213-215. (10) Havenkamp, V.; Ehrfeld, W.; Gebauer, K.; Hessel, V.; Lowe, H.; Richter, T.; Wille, C. Fresenius J. Anal. Chem. 1999 , 364, 617-624. (11) Haswell, S.; Skelton, V. Trends Anal. Chem. 2000 , 19, 389-395. (12) Ameel, T.; Papautsky, I.; Warrington, R.; Wegent, R. J. Propul. Power 2000 , 16, 577-582. (13) He, B.; Tait, N.; Regnier, F. E. Anal. Chem. 1998 , 70, 3790-3797. Anal. Chem. 2001, 73, 1942-1947 1942 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001 10.1021/ac000850x CCC: $20.00 © 2001 American Chemical Society Published on Web 03/28/2001