Modeling Optimizes MOEMS Production
Miniaturized optoelectronic systems are an attractive choice for applications from telecommunications to inertial sensing. The setup costs for manufacturing micro-optoelectromechanical systems (MOEMS) are high, however, and economies of scale quickly evaporate if multiple prototyping runs are necessary. Now researchers have demonstrated a process for modeling the performance of MOEMS using commercially available software.
Researchers have constructed a miniature infrared spectrometer using commercially available modeling software. The technique promises to ease time and cost requirements for the production of micro-optoelectromechanical systems. Courtesy of VTT Electronics.
The use of analytical modeling and optoelectronic simulation to predict system performance can eliminate manufacturing steps and minimize production costs. But a shortage of modeling software has led some researchers to develop their own custom tools, adding time and cost to production.
"The lack of capable simulation tools is a real problem to a system integrator working with MOEMS," said Kimmo Keränen, a research scientist with VTT Electronics in Oulu, Finland, and a member of the project team. He and colleagues at VTT and at Infotech Oulu research center modeled, built and tested a miniaturized IR spectrometer to verify the simulation process. The spectrometer consists of three micromachined devices in silicon: an electrically modulated IR emitter, an electrically tunable Fabry-Perot interferometer and a photodetector.
The researchers generated analytical and ray-trace models of the design using Mathcad from MathSoft Engineering & Education Inc. of Cambridge, Mass., and ASAP Pro 6.0 from Breault Research Organization Inc. of Tucson, Ariz., respectively. "The analytical modeling gives us the basic understanding of the critical physical processes and capabilities of the components and modules," Keränen explained. "The achieved accuracy in the ray-trace simulations depends on how well the models ... reflect reality."
They modeled three parameters to optimize the system's performance: throughput from source to detector, optical crosstalk between source and detector, and angular distribution of the light incident on the interferometer. The analytical model predicted that 2 percent of the total radiant power would be incident on the detector, while the ray-trace model predicted only 0.4 percent, partly attributable to its more accurate estimate of spherical aberration in the reflector. The ray-trace model also predicted that crosstalk would be 12,000 times smaller than the maximum signal. Both models indicated that an aperture was needed to control the angle of incidence at the interferometer.
The researchers integrated the detector and source monolithically on a 0.5 x 5 x 15-mm silicon substrate. The 3.3 x 3.3 x 0.5-mm interferometer, bonded to the substrate above the photodetector, had a typical transmission of 65 percent and a bandpass of 70 nm, measured full width half maximum. A preamplifier integrated circuit also was bonded to the substrate between the source and detector.
In operation, the 1 x 1-mm source emitted as a near-blackbody at 1000 K. Radiation from the emitter reflected off a curved mirror above the assembly and returned to the surface through the interferometer. The researchers measured the output of the prototype and found the value to be 83 percent of that predicted by the ray-trace model, sufficient to eliminate steps in the manufacturing process and to smooth the way for additional development of micromachined silicon MOEMS.
Although the researchers intended the development and production of the miniaturized IR spectrometer to validate the technology and methodology, the product is ready to be refined if industrial applications are found.
MORE FROM PHOTONICS MEDIA