Integrated Microbolometer Antenna Characterization from 95-650 GHz

C. Dietlein*tt J. D. Chisum*+ M. D. Ramirez* A. Luukanent E. N. Grossman+ and Z. Popovic'*Department of Electrical and Computer Engineering, University of Colorado at Boulder, Boulder, CO 80309

t MilliLab, VTT Technical Research Centre of Finland, 02044 Espoo, Finland+ Optoelectronics Division, National Institute of Standards and Technology, Boulder, CO 80305

Abstract- Electrical responsivity and radiation patternsare presented for a Nb microbolometer coupled to an ultraw-ideband spiral antenna at 300 K. A hyper-hemispherical sub-strate lens focuses incident radiation through the substrateonto the antenna. The Nb bridge is suspended in air betweenthe feed points of the equiangular spiral antenna with a nom-inal bandwidth of 0.2-1.8 THz. Pattern measurements areperformed at 95GHz, 238 GHz, and 650 GHz. Polarizationat 95 GHz is also explored.

Index Terms- Bolometers, Niobium, lens antennas, sub-millimeter wave antennas, submillimeter wave detectors,submillimeter wave measurements.

I. INTRODUCTION

Ultrawideband operation is important for submillimeter-wave applications: in passive imaging, broad bandwidthmaximizes the incident blackbody signal power fromthe scene; in either passive or active spectroscopy andimaging, target discrimination and recognition are greatlyimproved. Ultrawideband antennas, however, present sig-nificant challenges in terms of coupling power efficientlyinto the detector, particularly in arrays. In this paper, weexplore an approach consisting of a planar spiral antennacoupled through an integrated hyper-hemispherical sub-strate lens (Fig. 1). While hemispherical substrate lens-coupled antennas were initially developed by Rutledge [1],substrate lens-coupled spiral antennas were first investi-gated by Biittgenbach [2]-[3] in connection with SIS het-erodyne receivers, and later extended and generalized [4].

The integrated detector-antenna discussed in this paperis being investigated for application in a scanned, linear fo-cal plane array (FPA), configured in a so-called "fly's eye"geometry, (i.e., separate substrate lenses for each pixel).The system design involves a tradeoff between overallarray size, number of pixels, and operating wavelengthband. System considerations demand more tightly-packedconfigurations than the standard design prescription [2]allows for a particular low-frequency bandwidth limit. Thegoal of the work reported here is to better understand thelow-frequency limits of the lens-coupled spiral antenna.

The important dimensional ratios in the tradeoff be-tween standard design of a broadband self-complimentaryspiral antenna array and more compact systems are thelens radius, the extension length (h+ t in Fig. 1), the outerdiameter of the spiral antenna, and the low-frequency edgeof the design bandwidth. When either the lens radius or the

antenna diameter is too small, low-frequency performanceis degraded. The present antennas have been designedfor 2 mm diameter lenses, which the linear FPA willutilize. Specifically, the ratio of antenna radius to lensradius, 0.19, (Fig. 1) is similar to the value of 0.24used in the original lens-coupled spiral experiments [1].The measurements reported here, however, are on singledevices coupled to 4 mm diameter lenses. Therefore inthis case, the degradation of performance below the designband should reflect limitations due to the antenna size only.

II. BOLOMETRIC DETECTOR CHARACTERIZATION

The antenna-coupled microbolometer discussed in thiswork is an air bridge made of Nb, suspended betweenthe feed points of an Al equiangular spiral antennawith a nominal bandwidth of 0.2-1.8 THz. The bridgeis 24,um x 1 ,um x 20 nm. The Nb and Al layers aredeposited on a high-resistivity (p > 5000 Qcm) Si waferto facilitate antenna pattern testing at 300 K; fabrication ofthe devices is described elsewhere [5]-[6].

Room-temperature characterization of the bridge is per-formed using a current biasing scheme. Fig. 2 (top) is

HK r- .'0mm

4 mm

Fig. 1. Top-view (right) and 90° view (left) of the 5mm chip, thehyper-hemispherical lens, and an accurate layout of the pertinent traceson the chip. The extension length of the hyper-hemisphere is t = 0.3mm, and the thickness of the Si substrate is h = 0.5 mm. In this paper,we define co- and cross-polarized measurements as those in which theincident linearly-polarized field is parallel with and perpendicular to thebolometer axis (and DC bias traces), respectively.

1-4244-0688-9/07/$20.00 C 2007 IEEE 1165Authorized licensed use limited to: UNIVERSITY OF COLORADO. Downloaded on January 14, 2009 at 19:43 from IEEE Xplore. Restrictions apply.

a plot of the measured I-V curve of the antenna-coupledmicrobolometer. The I-V curve is fit to V = I(Ro+/31V),where Ro is the zero-bias resistance of the microbolometerand 3 is the normalized responsivity of the detector inunits of [VlWlmA]. The data range used in the fit to thisequation is highlighted in Fig. 2, and Q and Ro are foundusing a least-squares fit, producing Q = 265 VlWlmA, andRo = 505 Q. These values are typical of the fabricationprocess described in [5]-[6], which involves a XeF2 Sietch beneath the Nb bridge to suspend it in free space.Fig. 2 (bottom) shows the I-V curve plotted in terms ofresistance and power, and the extracted parameters Roand Q plotted in slope-intercept form to verify the fitaccuracy. The apparent large zero-bias resistance is not aproperty of the Nb bridge itself, but is due to a resistance inseries with the Nb bridge, specifically, the Nb-Al interface.Therefore, the extracted value of Q is correct for low biascurrents, as the series Nb-Al resistance is independentand thermally isolated from the Nb bolometric sensingelement. This is confirmed by illuminating the detectorwith a constant irradiance and varying the bias current;the measured response is linear as expected.

III. SUBSTRATE, HEMISPHERICAL LENS, & MOUNTING

The Si wafer is diced into 5 mm square chips; multipledetectors per chip are available but only one is suitablefor antenna pattern measurements with a substrate lenscentered on the chip. Shown in Fig. 1 is a diagram of thechip, substrate lens, and antenna, depicting the importantgeometry and associated dimensions. The spiral antennais 380,um in diameter, and located at the center of thechip, directly below the apex of the lens. The bolometeraxis is collinear with the bias traces extending from thearms of the antenna. It is assumed that the DC bias traceswill affect the low-frequency antenna response, producinga modified loaded dipole pattern.The parameter focused on in prior work [3] is the

dimensionless extension length of the lens, which weparameterize by the quantity n(h + t)/R. For an aplanaticdesign, this is unity. For the optimum efficiency "elliptical"design described in [3], this is larger, typically 1.3-1.6depending on design frequency. Our configuration corre-sponds to n(h + t)/R = 1.41. A metallic mount in theform of a slotted ring captures the substrate lens at its outerdiameter, similar to the configuration described in [3]. Thering geometrically vignettes the incident radiation at anangle of 480 from broadside. The mounted configurationcan be seen in Fig. 3.

IV. ANTENNA PATTERN MEASUREMENTS

Antenna pattern measurements of the lens-antenna-coupled microbolometer are completed at 650 GHz,238 GHz, and 95 GHz. Complete 2D pattern measurementsare taken at each frequency, and a selection of 2D patternsand cuts are shown.

X 10-4

2 ---- ----- C- Measured dataFit toV =IR + IV)

0 0.1 0.2 0.3 O4 0.5DevicevoageM

tir,?

(L3600

180

'- 540

En

0 1 2

Power[x 1l

Fig. 2. (top) I-V data; the highlighted range fit to V = I(R, + 31V).(bottom) I-V data shown in resistance-vs.-power (V/I vs. VI) formwith the extracted parameters Ro and 3 plotted asymptotically.

Fig. 3. Photograph of the assembly utilized for antenna measurements.The Si substrate lens can be seen in the middle of the Al flexure mount;the 5mm chip is hidden beneath the lens.

A. Experimental procedure at 650 GHz

A backward-wave oscillator (BWO) is used as the650 GHz source. Mechanically chopped at approximately170Hz and linearly polarized (LP), 2D patterns are mea-

sured in both orientations of the detector (co- and cross-

polarized) for the fixed LP incident field. Bolometer biascurrent is set to 170 uA. As the irradiance at the plane ofthe detector is low (order of 1 ,uW/mm2) due to low BWOtransmit power, the dynamic range of the measurement isnot greater than 20 dB, even though the measured noiselevel throughout the experiment remained within a factorof two from the known noise level of the preamp utilized(1.6 nV/vHz). A co-polarized 2D pattern at 650 GHz isshown in Fig. 4. The beam is nearly circular, with -3 dBbeamwidths in orthogonal cuts differing by 0.20. The scan

limits in both azimuth and elevation are ±200.

1166

L

---------- ----P+ -----

Measued dataRP PP÷R

nuu I'

Authorized licensed use limited to: UNIVERSITY OF COLORADO. Downloaded on January 14, 2009 at 19:43 from IEEE Xplore. Restrictions apply.

-10 0Azimuth [deg]

Fig. 4. A 2D pattern measured at 650GHz. The incident linearly-polarized electric field lies in the elevation plane, and the bolometer axisis perpendicular to the field. The-3 dB beamwidth in azimuth is 6.9°,and 6.7° in elevation. The first sidelobe levels are approximately -11 dB.

B. Experimental procedure at 238 GHz

A Gunn oscillator [7] is followed by a commercialfrequency doubler to produce a 238 GHz source. The beamis mechanically chopped at 80Hz and the electric fieldis linearly polarized; the bolometer is rotated by 900 tomeasure both linear polarization senses. The irradiance atthe plane of the detector from the LP transmit horn is on

the order of 40,uW/mm2. The bolometer bias current isset at 550 ,uA, and the dynamic range of the measurementis increased to 25 dB due to the greater source power.

Fig. 5 contains a 2D pattern measured in this configuration.The -3 dB beamwidths are increased by a factor of 2.6,which is close to the frequency ratio between 650 GHzand 238 GHz, _2.7. The scan limits in both azimuth andelevation are ±550.

C. Experimental procedure at 95 GHz

An HP 83624B synthesized sweeper is used to drive a

83558A mm-wave source module for W-band measure-

ments. The source is amplitude-modulated at 1.1 kHz, andthe bolometer bias current is set to 200 uA. For thesemeasurements, the bolometer axis is fixed in the elevationscan direction, and the incident electric field polarizationis co-, cross-, and circularly-polarized in both senses.

D. Discussion ofpattern measurements

The one-dimensional cuts shown in Fig. 6 demonstratedecreasing beamwidth with increasing frequency. As ex-

pected, the -3 dB beamwidths decrease nearly monotoni-cally as a function of frequency, narrowing from (meanof all polarizations) 250 at 95 GHz to 70 approachingthe center of the design bandwidth of the spiral antenna,650 GHz. In-band radiation at angles > 480 off broadside,as well as low-frequency radiation, is blocked by the lens

Fig. 5. A 2D pattern measured at 238GHz. The incident linearly-polarized electric field lies in the elevation plane, as does the bolometeraxis. The- 3dB beamwidths are 16.0° and 17.5° in azimuth and eleva-tion, respectively. The first sidelobe levels are approximately -10dB.

mount. In these cases, the measured radiation pattern isno longer a function of only the antenna, but of theentire fixture. It should be noted that 95GHz is wellbelow the designed system operating frequency. Cross-polarized radiation at or near the low edge of the operatingband will not encounter any significant radiating structuresof the antenna or DC traces. On the other hand, co-

polarized radiation below the nominal antenna bandwidthwill encounter the DC bias traces, which tend to act as a

loaded dipole.

95 GHz

e 238 GHz

-20 650 GHz

-5

-10

-20 06

.-2

-45 -30 -15 0 15 30 45Scan Angle [degrees]

Fig. 6. One-dimensional E-plane cuts at 650 GHz, 238 GHz, and95GHz. The electric field is parallel to the bolometer axis, and theazimuth scan is in the same plane.

A summary of the co- and cross-polarized 2D patterndata from 95 GHz to 650 GHz is provided in Table I. Weshow the -3 dB and -10 dB beamwidths for both E- andH-plane cuts. Denoted with (§) are two results at 95 GHzin which an anomalous reflection affected a portion of

1167

a1)

0

a1)a)J

0

2

4

8

-10

12

-14

-16

-18

-2022

G)

wi

'0

0ID

-10CD

0

15E-15 ID

'Eo~~~~~~~~~~~~~~~~~~C

10 20 0Azimuth [deg]

Authorized licensed use limited to: UNIVERSITY OF COLORADO. Downloaded on January 14, 2009 at 19:43 from IEEE Xplore. Restrictions apply.

TABLE I

SUMMARY OF MEASURED BEAMWIDTHS. POLARIZATION & CUT ARE

REFERENCED TO THE INCIDENT LP FIELD AS DEFINED IN FIG. 1.

FrequencyPolarization

-3dB E-plane_-10 dB E-plane-3dB H-plane_-10 dB H-plane

± 95GHzcross co

24.7 23.765.4 85.7§23.0 29.877.4§ 54.3

238GHzcross co

21.1 17.538.2 29.718.3 16.632.3 126.7

650GHzcross co

6.9 6.711.1 11.57.5 7.512.1 110.4

-RHCP-LHCP

-5

-10

-)15

the measured data for positive elevation scan angles (seen

in Fig. 6). Adjusting for the reflection, the B-plane co-

polarized -IO dB beamwidth is reduced to 64.2'; the H-

plane cross-polarized -10 dB beamwidth becomes 75.80.

Gaussicity

The overall coupling of an antenna to a Gaussian beam

can be expressed as

G

f Gc

f(EM)2dS f(E0)2dSS S

where EM is the measured radiation pattern (RP) and

is the fundamental mode Gaussian beam [8]. For

simplification, we assume that the radiating aperture is

square and the beam has constant phase, reducing (1) to

G

fS

~EM(x,y) ~EG(X,y) dS

I Em(x, y) ~2dSf EG(X, y) ~2dSS Ss

for G (0, 1). A two dimensional Gaussian fit EG(x, y)is estimated based on the measured radiation pattern.

The optimal fit was found by minimizing the sum of

square errors between a theoretical Gaussian beam and the

measured data. The mean Gaussicity of the two detector

orientations at 238 GHz is G= 82.80o, and at 650 GHz,

G =86.O0o.

V. DiscusSION

It has been shown that the spiral antenna-coupled

microbolometer has ultrabroadband operation with good

response and Gaussicity within the operating band. Key

characterizing parameters have been summarized distin-

guishing in-band operation from edge-of-band perfor-

mance. Expectations regarding the low-frequency antenna

performance proved correct- at 95 GHz, the antenna is

elliptically polarized, demonstrated by the 3 dB difference

in opposite CP responses in Fig. 7. Elliptical polarization

suggests a transition from the in-band circularly-polarized

spiral antenna pattern toward a linearly-polarized loaded

dipole.

-25L-45 -30 -15 0 15

Scan Angle [degrees]

30 45

Fig. 7. Azimuthal cuts at 95GHz with a circularly-polarized incident

field. A 3-dB difference is noted between RHCP and LHCP response,

indicating elliptical polarization due to out-of-band operation.

ACKNOWLEDGMENTS

The authors acknowledge funding under a NSF In-

terdisciplinary Award, #0501578, and Tekes decision

#40384/05. M. Rami'rez was supported by the NSF AGEP

grant, cooperative agreement #HRD-008655 1. NIST is a

US Government organization, therefore this work is not

subj.ect to copyright.

REFERENCES

[1] D. B. Rutledge, D. P. Neikirk, and D. P. Kasilingam, "Integrated

circuit antennas," in Infrared and Millimeter Waves, vol. 10, pp. 1-

90, K. J. Button, Ed., New York: Academic Press, 1983.[2] T. H. Biittgenbach, R. E. Miller, M. J. Wengler, D. M. Watson, and

T. G. Phillips, "A broad-band low-noise SIS receiver for submillime-

ter astronomy," IEEE Trans. Microwave Theory & Tech., vol. 36,

no. 12, pp. 1720 1726, Dec. 1988.

[3] T. H. Biittgenbach, "An improved solution for integrated array optics

in quasi-optical mm- and sub-mm receivers: the hybrid antenna,"

IEEE Trans. Microwave Theory & Tech., vol. 41, no. 10, pp. 1750-

1761, Oct. 1993.

[4] D. F. Filipovi6, S. S. Gearhart, and G. M. Rebeiz, "Double-slot

antennas on extended hemispherical and elliptical silicon dielectric

lenses," IEEE Trans. Microwave Theory & Tech., vol. 41, no. 10,

pp. 1738 1749, Oct. 1993.

[5] A. Luukanen, E. N. Grossman, A. J. Miller, P. Helistd5, J. S. Penttilii,

H. Sipola, and H. Seppi, "An ultra-low noise superconducting

antenna-coupled microbolometer with a room-temperature read-out,"

IEEE Microwave. & Wireless Comp. Lett., vol. 16, no. 8, pp. 464-

466, Aug. 2006.

[6] A. Luukanen and J. P. Pekola, "A superconducting antenna-coupled

hot-spot microbolometer," Appl. Phys. Lett., vol. 82, pp. 3970 3972,

Jun. 2003.

[7] J. E. Carlstrom, R. L. Plambeck, and D. D. Thornton, "A continu-

ously tunable 65-1 15-GHz Gunn oscillator," IEEE Trans. Microwave

Theory & Tech., vol. 33, no. 7, pp. 610 619, Jul. 1985.

[8] E. N. Grossman, "The coupling of submillimeter corner-cube anten-

nas to Gaussian beams," Infrared Physics, vol. 29, no. 5, pp. 875-

885, 1989.

1168

-20-

U

H

Authorized licensed use limited to: UNIVERSITY OF COLORADO. Downloaded on January 14, 2009 at 19:43 from IEEE Xplore. Restrictions apply.