This project considers the design and fabrication of a modulated microincandescent IR source. To reduce background noise and to improve signal amplification, most infrared applications require a modulated IR output at a fairly high frequency (above 1kHz).
Methods to modulate the output were explored for the microlamp to be used as the signal emitter. We consider specifically the performance specifications of a modulated microlamp for a carbon dioxide-detecting capnometer. We compare three modulation schemes: thermal, resonating filament, and shuttering. Qualitative comparisons favor the shuttered microlamp and it was chosen to be designed, fabricated, and tested.
A nine-mask process was devised to make the shuttered microlamp. Of the nine masks, six were used to fabricate the microlamp and the remaining three used to produce the shutter, at a position above the microlamp. The six-mask microlamp process is modified from the original process which was pioneered by Dr. Carlos Mastrangelo. Aside from layout changes, the new process replaces the potassium hydroxide etch with a nitric acid/hydroflouric acid mixture to remove the silicon from the substrate. It also uses n-type dopants for the fdament. The shutter, suspended from the substrate by folded beams and driven with electrostatic combs, is similar to the resonators first created by Dr. William Tang. Calculations were performed prior to fabrication to assess the shuttered microlamp performance.
Post-fabrication tests and measurements show that device yield is very high, at approximately 98% for the microlamp, Only about 50% of the shutters survived the rinsing process. The lower yield for the shutter is attributed to the fact that most of the failed shutters have lum features that were not strong enough to withstand prolonged rinsing. In contrast, the yield for shutters with 3um features was much higher (at approximately 80-90%).
Visible output radiation, becoming more intense with increasing filament voltage, was clearly observable during rnicrolamp testing, even under normal indoor illumination, To avoid polysilicon recrystallization that would permanently change the filament resistance, the drive voltage must be kept below a "reversible" region that can be found by characterizing the I-V characteristic of the microlamp. Filament lifetime is also strongly dependent on voltage. When the applied voltage is kept below the recrystallization voltage, the microlamp operates for over 4 hours with less than 10% change in its current.
The measured shutter resonant frequencies matched closely those obtained through simulations. The 81 shutters designed have resonant frequencies ranging from 1 to 10 kHz, and the deviations between the measured and the calculated results were no larger than 2%. In contrast, the measured bandwidths (-200Hz), are an order-of-magnitude smaller than the calculated bandwidths (-2kHz). In addition, measured displacements ranges from 1 to 10um and were larger, also by an order-of-magnitude, than simulated expectations. The matches in resonant frequencies, combined with the mismatches in bandwidths and displacements, suggest that the damping effect modeled in simulation may have been overestimated.
The polysilicon shutters were found to absorb most of the visible output radiation. However, an infrared-absorbing metal layer would be needed in order to modulate the lamp's infrared output.
December 31, 1993
Chen, P. Y. (1993). Modulated Infrared Source: Research Project. United States: University of California, Berkeley.