Multispectral Terahertz Camera
E. N. Grossman 1 , C.R. Dietlein1, M. Leivo2, A. Rautiainen2, and A. Luukanen 2 Abstract We describe a broadband terahertz camera based on a modular 64-element linear array of superconducting hotspot microbolometers. Unlike many superconducting sensor arrays, the readout for this array is performed entirely by uncooled electronics; no SQUIDs or cryogenic HEMTs are employed. The operating principles for the microbolometer and the readout scheme are described and compared with those of similar superconducting sensors. The camera's design has been optimized for standoff detection, at 8 m range, of objects concealed on the person of individuals. Real-time operation is meant to eliminate the need for individuals being screened to remain stationary, or to cooperate in any other way with the screening operation.
1
2.1 Camera optics The camera’s optical design is based on a reflective Schmidt telescope, with a spherical primary and a 6th-order polynomial corrector[7]. However, the corrector’s curvature is sufficiently weak that it can be replaced with a flat mirror without degrading the diffraction-limited resolution below ~750 GHz. A Spherical primary
INTRODUCTION
We have in previous conference presentations [1] described the overall architecture of an ultrawideband terahertz (THz) camera we are developing, and given a number of updates on its progress[2]. It is designed for standoff screening of personnel for threat items concealed beneath clothing. The suitability of imaging in the 100 GHz – 1 THz range for this application is widely recognized [3] and has led to a number of other efforts to develop cameras operating in this range as well [4],[5],[6]. Briefly, our camera is based on a 64-element, linear, focal-plane array of microbolometers, combined with a conical scanner in the aperture plane. This presentation focuses on optical design and performance, which is significantly complicated by the ultrawide fractional bandwidth of the sensor (approx. 3 octaves, nominally 0.2 – 1.8 THz). Section 2 describes the camera optics and the individual detector modules. Section 3 describes measurements of the beam patterns and efficiencies at a reference plane immediately outside the cryostat window. In section 3 and 4, the implications of these measurements are discussed, in terms of NETD and spatial resolution at the target plane. 2
(client) located remotely, over an Ethernet network connection.
SYSTEM OPTICS
The complete assembled system is shown in Fig. 1. Two copies of the system, located at NIST and at VTT, have been built and are currently operating. The system footprint is 1.1(w) x 1.0(d) x 1.8(h), with the acquisition computer (server) located on a separate adjacent cart and the display computer
Downlooking cryostat
Bending mirror Corrector
Conical scanner
64-element array (central 8 modules) Figure 1: (Top) Fully assembled PEATCam system, (lower left) focal plane, (lower right) cryostat assembly, inverted, with IR filters and shields removed. small flat secondary mirror directly outside the cryostat window directs the antenna beams toward the primary; after reflection off the primary, the beams reflect off a large bending mirror and the corrector, before passing through the conical scanner at the entrance aperture. The scanner is in-line and highly
1
Optoelectronics Division, Natl. Inst. Of Standards and Technology, Boulder, CO 80305 e-mail:
[email protected].: +01 303 497 5102, fax: +01 303 497 5746. 2 Millimeter-wave Laboratory of Finland, 3 Tietotie, Espoo, FI e-mail:
[email protected]: ++358 040 736 9717.
978-1-4244-3386-5/09/$25.00 ©2009 IEEE 1002
balanced, allowing it to rotate at the frame rate without significant vibration.
Primary mirror
Cryostat Window and IR filters
2.2 Module optics and construction Blackbody
The construction of the detector module is illustrated in Fig. 2. The use of hyperhemispherical or elliptical substrate lenses (extension lengths of 1.0 to ~1.4 R/n respectively) to couple through a dielectric substrate to a planar antenna is well established [8] and commonly used with ultrawideband antennas in THz time-domain systems. Our hot-spot microbolometers are placed at the feeds of equiangular spiral antennas. Accurate positioning and intimate contact between the chip and substrate lens must be maintained over extensive thermal cycling. This is accomplished with a lithographically patterned, stainless steel flexure that squeezes the lenses against the back of the chip.
1.0R/n
Figure 2. Exploded view (top) and assembled photograph (middle) of 1x8 detector module. Rear lens extension includes 0.50mm chip thickness.
3
OPTICAL MEASUREMENTS
Fig. 3 illustrates the arrangement for optically characterizing the modules. The small secondary mirror was removed, and in its place directly beneath the cryostat window, a blackbody source was scanned. The blackbody was either a 5.1 mm diameter IR cavity blackbody ('T=380 C) or a 12mm diameter liquid-nitrogen cooled ('T=220 C) piece of AN-72 mm-wave absorber. In some cases, low-pass filters consisting of expanded polystyrene (EPS) of various weights (i.e. rolloff frequencies) were placed between the window and the source.
Large bending mirror Y translation
X translation
Fig. 3. Optical measurement setup. 3.1 Encircled energy Planar antenna patterns are generally characterized with a monochromatic source, as we reported in a previous publication on single-pixel modules[9]. However, a broadband blackbody measurement is more directly applicable to the actual use of the camera. A typical 2-D antenna pattern, along with the Gaussian best-fit and the residuals to the fit, are shown in Fig. 4. The deviation from the best-fit Gaussian is very consistent across the 32 antennas measured; it exceeds a Gaussian at the center and the distant wings and falls below the Gaussian at the near wings. This is the deviation expected for a superposition of Gaussian beams whose widths span a significant range. Moreover, the measured FWHM = 10±1.5 degree is consistent with the earlier monochromatic measurements, that fit an inverse-frequency dependence, FWHM= 9 THz-degree/f, if a bandpass centered at 900 GHz is assumed. This is very close to the center of the antenna’s nominal design bandpass of 0.2 – 1.8 THz. Finally, the measured FWHM increased as expected when two lowpass EPS filters were inserted into the beam. For EPS weights of 35 g/L and 96 g/L density, the FWHM increased by approximately 20 % and 40 % respectively. Fig. 4 also displays the encircled energy plot for all 32 measured channels. Distance from the focal plane to the source was 310 mm for these measurements. The secondary mirror subtends an angle of ~22.5o in the narrow (cross-array) direction, corresponding to a radius of 64mm at the measurement plane. As indicated by the marker on the encircled energy plot, this corresponds to a spillover loss below 18% (0.9 dB). (This is a worst case, since the mirror subtends a larger angle in the array direction.) This spillover loss, which is concentrated at the low frequency edge of the band, is relatively small. More significant is the high frequency loss due to obscuration by the secondary of
1003
improvement over the 110 mK of the devices used in our early single-pixel imaging. Most of this improvement is due to lower optical loss in the IR filters and vacuum window. From this NETD, measured outside the window, it is simple to scale the system performance to the target plane, when the conical scanner is used: NETD (t arg et )
/ 2 §¨ NETD ( window) N 1scan ¨
1/ 2
f · ¸ 30 fps ¸¹ ©
1 Șopt
where Nscan=230 is the number of scan angles used in the scanned acquisition, f the frame rate and K the overall optics efficiency. For our best
NbN module
Fig. 5. Measured NETD’s across array. Fig. 4. Typical measured 2D beam pattern (upper left) on a blackbody source, and best-fit Gaussian (upper right). This channel displays a best-fit FWHM of 9.3 degree. Central panels display the fit residuals, highly consistent channel to channel. (Bottom) Encircled energy diagram for all 32 channels. the beams reflected off the primary. This requires a measurement plane situated after reflection off the primary mirror. 3.2 Optical efficiency and misalignment The same data can be integrated over the measurement plane to extract the signal that would be observed from a blackbody source that filled the antenna beam. This can then be combined with a separate noise measurement to yield the optical NETD referred to the measurement plane. It can also be combined with the electrical responsivity (=1/2Vbias) and the nominal design bandwidth of 1.80.2=1.6 THz to yield an optical efficiency. Fig. 5 displays the NETD performance across 4 modules. For this particular dataset, the last module containing NbN devices was incorrectly biased, so their NETDs are spuriously high. The mean NETD on the remaining good devices is 38 mK, a significant
of optics efficiency, 35 %, this yields a final NETD referred to the target of 1.6 K. Fig. 6 displays centroids of the best-fit Gaussians for each channel against the nominal position of the beam, based on the channel’s position in the focal plane. Deviation from nominal location of a beam indicates that beam is emerging from its antenna off-normal. The deviations of modules 1,2, and 4 perpendicular to the array are consistent with a translation error in the position of the flexure that locates the lenses relative to the chip. A 10 mm offset in the measurement plane corresponds to a 35 Pm translation error, which is plausible after cooling the assembled module to 4K. The slope visible in module 3 indicates a rotation error in the flexure alignment. The beam deviation parallel to the array indicates the beams from edge pixels are deflected toward the center of their respective modules. We attribute this to differential thermal contraction between the stainless steel flexure and the silicon chip. These deviations of the antenna beams do not significantly affect performance because the deviations are relatively small, and only affect illumination of the primary, not location at the target. 4
1004
IMAGERY
Finally, fig. 7 shows an image taken by raster Acknowledgments We are grateful to Dr. B. Smith of DHS Science and Technology Directorate (U.S.) and to TEKES (Finland) for support of this program. References [1]
[2]
[3]
[4] Fig. 6 Deviation of antenna beams from nominal (normal) direction along the array. scanning a single 8-channel module across a target. The feature we focus on is best illustrated in the zoomed-in region. The spatial resolution is clearly better (lower) on the subject’s fingers, which lie on the surface, than on the handgun, which is concealed beneath a layer of denim.
[5]
[6]
[7]
[8]
[9]
Fig. 7. 8-channel raster-scanned image
1005
G. Luukanen A., L., Helisto, P., Penttila, J., Seppa, H., Sipola, H., Dietlein, C.D., and Erich N. Grossman, "Passive EuroAmerican Terahertz Camera (PEAT-CAM): Passive Indoor THz Imaging at video Rates for Security Applications," Proc. SPIE, vol. 6548, pp. 6548-08, 2007. H. Luukanen A., P., Lappalainen, P., Leivo, M.,Rautiainen, A., Toivanen, H., Seppa, H., Taylor, Z., Dietlein, C.R., and Grossman, E., "Stand-off Passive THz Imaging at 8meter Standoff distance: Results from a 64channel Real-time Imager," Proc. SPIE, vol. 7309, pp. 73090F-1, 2009. R. Appleby, and Wallace, H.B., "Standoff Detection of Weapons and Contraband in the 100 GHz to 1 THz Region," IEEE J. Anntennas and Propagation, vol. 55, pp. 2944-2956, 2007. D. M. Sheen, Hall, T.E., Severtsen, R.H., McMakin, D.L., Hatchell, B.K., Valdez, P.L.J., "Active Wideband 350 GHz Imaging System for Concealed Weapon Detection," Proc. SPIE, vol. 7309, pp. 73090-I, 2009. K. B. Cooper, Dengler, R.J., Chattopadhyay, G., Schlect, E., Gill, J., Skalare A., Mehdi, I., Siegel, P.H., "A High-Resolution Imaging Radar at 580 GHz," IEEE Microwave and Wireless Components Letters, vol. 18, pp. 64-66, 2008. C. Mann, "A Compact Real Time Passive Terahertz Imager," Proc. SPIE, vol. 6211, pp. 6211-14, 2006. D. Korsch, "Reflective Schmidt Corrector," Applied Optics, vol. 13, pp. 2005-2006, 1974. T. H. Buettgenbach, "An Improved Solution for Integrated Array Optics in Quasi-optical Mm and Submm Receivers," IEEE Trans. Microwave Theory and Tech.. vol. 41, pp. 1750-1761, 1993. C. Dietlein, Chisum, J.D., Luukanen, A., Grossman, E.N., Popovic, Z., "Integrated Microbolometer Antenna Characterization from 95-650 GHz," Intl. Microwave Symposium Digest, pp. 1165-1168, 2007.
