Simulation software speeds microfluidics development
BY RON MANGINELL, SANDIA NATIONAL LABORATORIES
Physical testing will tell you whether your MEMS or NEMS design works-but only after a time-consuming process of prototype creation. If the design does not work as expected, which is typical initially, it can be difficult to determine why. The small size of the devices makes it nearly impossible to instrument them, and without quantitative information engineers must rely upon instinct and guesswork to optimize design. That is, unless they use simulation software.
Simulation is an important tool for diagnosing phenomena such as the microfluid flow, which is critical not only in the manufacture of nearly every MEMS, but also a key to the operation of many MEMS, including inkjet printheads and lab-on-a-chip devices. MEMS typically involve multiple, coupled physics effects such as actuation, pumping, and fluid and solid motion; and computational fluid dynamics (CFD) offers multiphysics simulation, including coupled solutions of fluid flow, structures, electrostatics, AC/DC conduction, electromagnetics, lead zirconate titanate (PZT), free surface flows with surface tension, and magneto-hydrodynamics (MHD) to address them.
The preconcentrator is patterned with tiny adsorbent pillars.
Sandia Corp., a Lockheed Martin company, used simulation to develop μChemLab, a handheld system designed for use by first responders to detect toxic agents. The first stage of the system is a preconcentrator (see figure, above) that samples and collects analytes from an inlet gas stream (“loading inlet”) and ejects them, on command, onto a separation stage. The planar preconcentrator consists of a thin silicon nitride membrane supporting a metal film heating element patterned with tiny pillars. The membrane is coated with a templated porous sol-gel to selectively and reversibly absorb analytes of interest while allowing interferents to pass by.
In the collection phase, a gas stream containing target analytes flows through the preconcentrator chamber and adsorbs to the film. Then the adsorbent surfaces are heated so that the collected analytes desorb in a narrow, concentrated chemical pulse. Using a sample collection time of 30 to 60 seconds causes a 100-fold concentration enhancement in the desorbed pulse over the inlet stream.
Engineers designing the preconcentrator were challenged to collect as much of the analyte as possible on the adsorbent surfaces in the least possible time and provide a sharp peak of analyte flux for detection by the sensor. Design parameters included the geometry of the preconcentrator, the mass flow rates during the adsorption and desorption phases, the carrier gases during adsorption and desorption phases, adsorption and desorption kinetics, and the desorption temperature profile.
Running the software simulation on the original design (top) showed poor wetting of pillars. Rearranging the pillars (bottom) resulted in a greatly improved flow profile with improved wetting.
Sandia worked with consultants at ESI Group (www.esi-group.com) to simulate fluid flow and chemical reactions through the preconcentrator with ESI’s CFD-ACE+ software. The CFD simulation showed that in the original design, flow rushes along the edges of the preconcentrator, avoiding the pillars. To force the gas to flow through the pillars, engineers rearranged them and blocked off the sides. The figures on p. 26, bottom, illustrate how re-running the analysis on the new geometry showed a greatly improved flow profile with much more wetting of the adsorbent pillars.
Modeling adsorption and desorption
Once the engineers had determined the optimal flow geometry, they performed four simulations to measure the surface reactions involved in adsorption and desorption of the analyte. Two flow rates were used for the adsorption phase, while two gases and two flow rates were used for the desorption phase.
To gain a deeper understanding of the adsorption and desorption processes, the team placed point probes on the pillars in the front, middle, and back rows of the simulated product. With this, the software was able to predict analyte site fraction, the amount of adsorbed analyte relative to the total analyte input, versus adsorption time.
Preconcentration is often critical for microsensor systems because it can increase the concentration of many analytes above their limit of detection and remove some interferants. Yet preconcentrators are difficult to design using conventional methods. CFD can overcome these challenges by providing accurate performance predictions. In this case, CFD helped to optimize the design from a flow standpoint and then provided information that was used to improve both the adsorption and desorption phases.
Ron Manginell is principal member of the technical staff at Sandia National Laboratories. You can reach him at firstname.lastname@example.org, tel: (505) 845-8223.