Fuel atomization from a micromachined ultrasonic droplet generator: Visualization, scaling, and modeling

Existing battery technologies have become a major obstacle to advances in the performance of portable energy-intensive devices primarily due to a limited lifetime between charge cycles.^1,2 Fuel-cell-based energy sources are a viable alternative due to the high energy density of liquid fuels and the potential for high efficiency power generation. The focus of recent work has been the development of two types of fuel cells for portable applications, hydrogen-based fuel cells with external fuel reformation, i.e., conversion to hydrogen, and direct-methanol fuel cells that oxidize methanol directly at the cell anode.^1,3 Regardless of whether internal or external fuel reformation is used, power-efficient atomization of liquid fuels ranging from methanol to higher hydrocarbons and diesel to kerosene and logistic fuels, e.g., JP-8, is an essential processing step for conversion of a fuel from liquid to gas phase. We present the experimental characterization and theoretical modeling of the fluid mechanics underlying the operation of a micromachined ultrasonic atomizer. This droplet generator utilizes fluid cavity resonances in the 0.5 to 3 MHz range along with acoustic wave focusing for low power atomization of liquids for fuel processing. The device comprises a fuel reservoir located between a bulk ceramic piezoelectric transducer for ultrasound generation and a silicon micromachined array of liquid horn structures as the ejection nozzles. The array size can be scaled to meet flow rate requirements for any application because a single piezoelectric actuator drives ejection from multiple nozzles. The atomizer is particularly well-matched to fuel processing applications because it is capable of highly controlled atomization of a variety of liquid fuels at low flow rates. This low-flow-rate requirement intrinsic to small-scale, portable power applications is especially challenging since one cannot rely on the conventional jet-instability-based atomization approach. Further, the planar configuration of the nozzle array is suited to integration with the planar design of fuel cells. Experimentally-validated finite element analysis (FEA) simulations of the acoustic response of the device are used to estimate the fraction of the electrical input power to the piezoelectric transducer that is imparted to the ejected fluid. Results of this efficiency analysis indicate that it is not optimal to design the ejector such that a cavity resonance (corresponding to acoustic wave focusing at the tips of the pyramidally-shaped nozzles and thus fluid ejection) coincides with the longitudinal resonance of the piezoelectric transducer. It also appears that the efficiency of the device increases with decreasing frequency. Atomization of methanol and kerosene from 5 to 25 μm diameter orifices is demonstrated at multiple frequencies between 0.5 and 3 MHz. In addition, high-resolution visualization of the ejection process is performed to investigate whether or not the proposed atomizer is capable of operating in either the discrete-droplet or continuous-jetting mode (see Figure 1). The results of the visualization experiments provide a basic understanding of the physics governing the ejection process and allow for the establishment of simple scaling laws that prescribe the mode of ejection; however, it is likely that the phenomena that dictate the mode of ejection (i.e., discretedroplet vs. continuous-jet) do not occur within the field of view of the camera. Further, the most important features that determine the initial interface evolution occur within the nozzle orifice itself. A detailed computational fluid dynamics (CFD) analysis of the interface evolution during droplet/jet ejection yields additional insight into the physics of the ejection process and provides further validation of the scaling laws. Figure 2 provides examples of simulation of both discrete-droplet and continuous-jet mode ejection.