Dean P. Neikirk and Sangwook Han Microelectronics Research Center

1. Design of Infrared Wavelength-Selective Microbolometers Using Planar Multimode Detectors Dean P. Neikirk and Sangwook Han Microelectronics Research Center Department of Electrical and Computer Engineering The University of Texas at Austin Austin, TX 78712 USA SPIE’s Microtechnologies for the New Millennium 15-18 May 2003 Hotel Meliá SevillaSevilla, Spain Proceedings of SPIE Vol. #5836 : Smart Sensors, Actuators, and MEMS SESSION 16, Room: Arenal I Wed. May 18, 12.00 to 13.00: Infrared Sensors 12.20-12.35: Design of infrared wavelength-selective microbolometers using planar multimode detectors, D. P. Neikirk, S. Han, Univ. of Texas/Austin (USA) [5836-60] link to pdf of proceedings paper .

2. Mirror layer thin conductor (absorber) resistive sheet Gap d incident mirror Fabry-Perot Microbolometer Array http://lep694.gsfc.nasa.gov/code693/tdw03/proceedings/docs/session_2/Ngo.pdf Conventional microbolometer infrared focal plane detectors • in an ideal device, the absorber should provide total absorption of the incoming radiation and convert the electromagnetic radiation into heat • to “match” an absorber to free space requires • absorber: e.g., a thin conductor with sheet resistance 377 ohms • mirror placed (odd integer)·l/4 behind absorbing layer • essentially a Fabry-Perot cavity • this is sometimes referred to as “space cloth”

3. LWIR 1 0.8 2 mm gap 2.5 mm gap 0.6 coupling efficiency 0.4 3 mm gap 0.2 0 4 6 8 10 12 14 wavelength (microns) Spectral response of Fabry-Perot microbolometers • is it possible to build “multi-color” IR F-P microbolometer focal plane arrays? • the primary “design variable” is the distance to the mirror 377W mirror the bandwidth of conventional Fabry-Perot microbolometers is too wide to allow easy “color” discrimination in the LWIR wavelength band

4. resistive microbolometer w metal grid mirror g d Single Element inArray a metal grid mirror Planar Multimode Detector Array Alternative: planar multimode detectors • replace standard thin film bolometer with a true antenna coupled microbolometer array • essentially a resistively loaded inductive/capacitive mesh • planar multimode detectors were extensively studied for infrared and millimeter wave detection by Rutledge and Schwarz in the late 1970’s

5. array grid equivalent mirror equivalent resistive microbolometer support layer Gd η0 η0 jBc jBl g a w mirror grid Zin d single pixel d single period Planar Multimode Detectors • grid period a< the shortest wavelength • analysis can be performed using a modification of Eisenhart & Kahn’s waveguide post model

6. grid response depends on array period a, gap g, post width w, distance to mirror d, and sheet resistance RS of microbolometer material grid g a w d 1 * Mag Green Red 7.00 5.33 6.90 * 0.5 5.00 1.00 0.20 power absorption efficiency 4.50 0.96 3.00 2.50 3.61 3.29 377Ω 30Ω 53.5Ω 0 7 8 9 10 11 12 13 14 wavelength [micron] Spectral response of planar multimode grids space cloth all lengths in [micron] • wide range of achievable bandwidths, from broad to narrow

7. mechanical actuation Methods of to achieve varied wavelength selectivity • for fixed grid dimensions, vary the distance between the array and the mirror • three “color” array using three different mirror distances • for fixed grid dimensions, use an actuator to vary the distance between the array and the mirror

8. narrow spectral response allows greater “color” sensitivity Wavelength selectivity varied by changing distance d to the mirror a=6.90μm d=3.29μm g=0.20μm 14 w=3.00μm Rs=53.5Ω 13 12 11 wavelength in [μm] power absorption efficiency 10 9 8 7 2 4 0 1 3 5 6 d in [μm] d=0μm d=6μm

9. ambiguous 1 1 ideal 0 0 1 1 power absorption efficiency power absorption efficiency 0 0 1 1 0 0 11 12 7 9 10 13 8 14 11 12 7 9 10 13 8 14 wavelength in [micron] wavelength in [micron] Selecting spectral response for “color” pixels • if a design produces more than one peak in absorption then the “color” becomes ambiguous pixel 1 pixel 2 “ghost” peak pixel 3

10. Achieving wavelength selectivity by varying d a=6.90 micron g=0.20 micron Good w=3.00 micron d=3.29 micron Rs=53.5Ω 14 13 1 12 11 wavelength in [micron] power absorption efficiency “ghost” peak: bad 10 Good power absorption efficiency 9 8 0 7 7 8 9 10 11 12 13 14 2 4 0 1 3 5 6 wavelength in [micron] d in [micron] d=5.70 micron

11. 1 1 power absorption efficiency power absorption efficiency a=5.07micron d=2.0micron d=1.65micron g=1.43micron d=3.75micron d=2.5micron w=0.74micron d=3.0micron d=5.10micron Rs=21.0Ω Rs=377Ω 0 0 7 8 9 10 11 12 13 14 7 8 9 10 11 12 13 14 wavelength in [micron] wavelength in [micron] Optimized 3-color wavelength selectivity by varying only d optimization performed using a genetic algorithm • design goal: three minimum width spectral peaks centered at 8, 10 and 12 microns Space cloth insufficient spectral selectivity

12. 1 0.9 0.8 0.7 0.6 power absorption efficiency 0.5 0.4 0.3 0.2 0.1 0 7 8 9 10 11 12 13 14 wavelength in [micron] Problem with variable d: complex fabrication processes • requires either • fabrication process with three different sacrificial layer thicknesses • or • mems actuator • in both cases the fabrication would be more complicated than current micromachined microbolometer focal plane array processes • there is still some color ambiguity for the longest wavelength (12 micron) pixel • due to “ghost” peak at 7 microns

13. w g a d Wavelength selectivity by varying lithographically-drawn parameters • potentially simpler process if the sacrificial layer is held fixed for all pixels • sheet resistance of bolometer material also held constant (same material for all pixels) • vary ONLY the lithographically drawn features of the grid • array period a, gap width g, and post widthw

14. 1 0.9 w = 4.57mm 0.8 0.7 0.6 w = 1.30mm w = 2.80mm power absorption efficiency 0.5 0.4 0.3 0.2 0.1 0 7 8 9 10 11 12 13 14 wavelength in [micron] Optimized designs • genetic algorithm used for optimization • design goal: three minimum width spectral peaks centered at 8, 10 and 12 microns • constraints: • d (distance to mirror) is varied, but must be the same under all three pixels • sheet resistance of bolometer material also held constant • vary only grid period a, gap width g, and post widthw • results from design optimization • optimum distance to mirror d = 3.14 mm • optimum RS = 56.6 W • all three pixels share common grid period a and gap width g • a = 6.80 mm • g = 0.20 mm • post width w is critical in determining location of absorption peak

15. Comparison: power absorption efficiencies Fabry-Perot microbolometer variable mirror grid microbolometer grid dimensions varied 1 absorption 0 7 7 7 14 14 14 wavelength in [micron]

16. IR wavelength-selective focal plane arrays • planar multimode detectors exhibit widely tunable spectral response • can tune for much narrower spectral response than conventional Fabry-Perot microbolometers • tuning of wavelength response can be achieved using several methods • for fixed grid dimensions distance to tuning mirror can be used • multiple sacrificial layer thicknesses • mechanical actuation • a wavelength-selective three pixel design, each pixel using different lithographically drawn dimensions with constant mirror separation, shows excellent narrow band response • through the use of planar multimode detectors color vision in the long wavelength band should be achievable