STAR

SOFIA Terahertz Array Receiver

Introduction

STAR will be one of the second-generation heterodyne instruments for SOFIA (Stratospheric Observatory For Infrared Astronomy). For observations of submillimeter and far-infrared radiation SOFIA is an excellent platform since at an altitude of 12.5 - 13.7 km (41000 - 45000 ft) water vapor absorption becomes small to negligible as can be seen in Fig. 1, especially in the wavelength range of 30 - 1000 microns which is blocked for any ground based observations (for a description of the heterodyne detection technique see here).

Fig. 1: Atmospheric transmission for good ground based sites and a stratorpheric platform

For the detection of the 158 micron (1.9 THz) fine structure line of singly ionized carbon (CII) we are developing an array receiver with a 4x4 beam pattern on the sky. This line contributes to cooling of the interstellar medium. Since in most cases this line is optically thin it serves as an important diagnostic of the energy balance of gaseous matter.

To date, only a handful of ground based heterodyne array receivers above 200 GHz are operational in the world:

HERA: 9x 230 GHz,
CHAMP: 16x 490 GHz,
PoleStar: 4x 810 GHz,
DesertStar: 7x 345 GHz,
SMART: 8x (460-490 GHZ + 810-880 GHz),

The STAR instrument will contain a 4x4 element heterodyne mixer array for the frequency range from 1.6 to 1.9 THz (187 to 158 microns). Its main scientific goal is large scale mapping of the 158 micron fine structure transition of singly ionized carbon. The design frequency range covers this line out to moderate red shifts and also allows to observe a variety of other spectral lines.

The 16 element detector array will consist of two interleaved subarrays of waveguide mixer blocks with diffusion cooled niobium hot electron bolometers (HEB) mixers. Local oscillator power will most likely be provided by a frequency multiplied phase locked backward wave oscillator (BWO). For the efficient distribution of the local oscillator beam among the array elements we use newly developed Fourier gratings. Four four-channel acousto optical spectrometers (AOS) will be used to spectrally analyze the receiver outputs.

Ionized and neutral carbon are found to be abundant in photon dominated regions (PDRs), where far-ultra-violet photons from the average interstellar radiation field and/or from massive young stars dissociate molecules such as CO, and ionize carbon. While neutral carbon, C, samples the transition interface between ionized and neutral/molecular layers, C+ is intrinsic to the regions of high UV field. As a large fraction of the interstellar medium is in a photon dominated state, efficient and fast large scale mapping of C+ is very important. This can best be accomplished in its fine structure transition in the far infrared range.

Along this path of technological development, and due to it being an astrophysically important complement, we have developed a dual frequency 4x2 array receiver SMART for the two fine structure transitions of neutral carbon at 490 and 810 GHz, which is installed the KOSMA observatory on the Gornergrat, Zermatt, Switzerland, and is operational since September 2001. As a second step, we will develop a 1.9 THz channel for SOFIA's first generation receiver system GREAT (German Receiver at Terahertz Frequencies GREAT), which will be a modularly built dual frequency instrument for up to 4.7 THz. This will serve also as a test platform for the further development of components for STAR.

Given a SOFIA beam FWHM of 15 arcsec at 1.9 THz, a 4 x 4 array with a beam separation of about square root of 2 FWHM is well suited to image nearby and weakly red-shifted galaxies very efficiently. With a typical diameter of a few 10000 light years, galaxies will appear at the size of the STAR footprint (60 arcsec square) at a distance of up to 100 Megaparsec. Smaller galaxies can be observed self-chopping over the array footprint.

As the baseline option we plan to use a frequency-tripled backward wave oscillator (BWO) as a local oscillator (LO), covering red-shifts up to -200 GHz, i.e. up to where several broad and partially obscuring residual absorption features of the atmosphere cut in.

We nevertheless aim at a maximum frequency range to be covered by the LO, as a wealth of other species is expected to have features in this spectral region. This includes some of the lowest rotational transitions of several metal hydrides, puckering vibrations of some PAHs and bending vibrations of larger linear carbon clusters. KOSMA is investing efforts in laboratory spectroscopy to enlarge the transition frequency database necessary for the first space detection of these species.

As a heterodyne instrument, STAR is a perfect complement to the spectral imaging photoconductor array FIFI-LS from the MPE group FIFI. Given state of the art detectors, the per-pixel and per-spectral-resolution-element sensitivity is about identical at a resolution comparable to typical galactic line widths. FIFI-LS has higher sensitivity in detecting integrated line fluxes out to more distant galaxies. STAR has the unique capability of fully resolving the line profiles, thus allowing the identification of individual kinematic and dynamic components.

Description of the STAR receiver concept

Fig. 2: Grating structure (left) and sub-array beam pattern (right). Two of these beam patterns will be interleaved to form a 4x4 beam pattern.

The field of view will consist of two subarrays, as shown in Fig. subarrays, split by polarization, following the concept used with the MPIfR 490 GHz array receiver CHAMP. The sub-arrays will have an approximate beam spacing of 2 FWHM, so that the complete spacing will be close to 2^-0.5 FWHM. Given the present state of development, the detectors will probably be diffusion cooled Niobium hot electron bolometers (HEB) in a LHe cooled dewar. The 16 receiver outputs will be analyzed by 4 four-channel acousto-optical spectrometers (Array-AOS).

Local oscillator sources for these high frequencies still require development. The LO for STAR may likely be a frequency-tripled 633 GHz backward wave oscillator, because it starts already with a relatively high power of > 5mW at a high fundamental frequency. We decided to design the STAR receiver with warm optics, because the increase of system noise due to the warm mirrors will not be critical at an expected HEB receiver noise temperature of approx. 2000 K . In Fig. 3 we show an overview of the optics. Another option for the generation of LO power is using a Quantum Cascade Laser (QCL) which promises plenty of power at 1.9 THz. Still quite some amount of development is to be undertaken to achive that, though.

Fig. 3: Overview of the STAR optics concept. In the right part the incoming telescope beam is reimaged in a Gaussian telescope.

Optics setup: Signal and LO beam path

The nearly parallel telescope signal beams 1.9 THz first go through a beam rotator, consisting of three flat mirrors in a K-mirror configuration, located prior to the focal plane of the telescope. A Martin-Puplett diplexer is used for each polarisation separately to superimpose LO and signal beams. Polarizations are split by a wire-grid in front of the diplexers and not merged thereafter but led to the two interleaved focal plane units.

As local oscillator we employ a frequency-tripled BWO which is distributed to two perpendicularly polarized subarrays of 8 beams. The single LO beam is dispersed into 8 by newly developed collimating Fourier gratings.

Mixer block "STAR" arrangement

Due to absorption and reflection losses we do not want to use an array of individual lenses as collimators in front of the mixer horn antennas, but will use off-axis mirrors instead. However, for an array with more than two rows of detectors, it is not straight forward to find an overall arrangement of mixers and mirrors. Additionally, we have to account for the restriction, that the size of the individual mixer blocks has to be at least on the order of 10 mm square.

Fig 4: Facet mirror element at the focal plane. The size is about 10 cm left to right.

An option for a 3-dimensional arrangement of mirrors and mixers is shown in Fig. 4. Its advantage is the independence from the actual size of the mixer blocks. The six outer beams are reflected outwards in a "STAR"-like shape under 90° and the two inner beams under 70°. The angles of the individual elliptical mirrors are mutually arranged in such a way that the edges of neighboring mirrors are flush respectively, so that they can be CNC-milled from one piece with no aperture loss at their edges. This condition leads to an odd polarisation angle for four of the mixers (see figure), which is adjusted to 45° by the overall facette arrangement. Additionally, the holding devices for the horn antennas can be CNC-fabricated together with this facette mirror out of one block. We expect that in this way the relative positions of the horns are precise to 10-20 microns and that the resulting beams will then be positioned on the sky precisely enough (e.g. 0.2 FWHM).

With the above mentioned overall optics magnification of about 1, the total optics cross section near the mixers will have a diameter of only about 33 mm. This offers the possibility to put the dewar window directly in front of the "STAR" mixer arrangement which reduces the dewar volume.

A typical beam waist of a 2 THz feed horn will be 0.3 mm. A distance of d approx. 14.5mm between horn and collimator mirror results by demanding that the beam widens up from w₀ approx. 0.3 mm to a radius of w approx. 2.3 mm on the mirror. As mentioned above, we have a beam spacing of about 3.25*w₀ approx. 8 mm. We see that under these assumptions we can indeed extract the two middle beams through the outer ones without vignetting the outer beam with the horn for the inner one.

Detector, Local Oscillator and Backends

The detector elements will be diffusion-cooled Niobium Hot Electron Bolometers, currently under development at KOSMA. In addition to their low-noise performance at terahertz frequencies, this type of HEB mixers has the potential of an IF bandwidth beyond 8 GHz.

The local oscillator power demand of an HEB unit will probably be on the order of 50nW, which results in about 1*10^-6 W of LO power for the whole array. Up to now the baseline configuration is to employ a frequency-tripled BWO, but there are also all-solid-state LO chains in the offing the promise to supply sufficient power to pump 16 HEBs at a time.

The most promising option though is a 633GHz backward wave oscillator (BWO) (ISTOK Company, Russia) with 70 GHz tunability and a power of more than 5mW. We hope that this is enough to produce 2*10^-6 W of power out of a frequency tripler. Frequency multiplied, phase locked BWOs are extensively used for high resolution THz lab spectroscopy in our institute. To have a compact BWO suitable for airborne purposes, the water-cooled high-power consuming electro-magnet of a typical high frequency BWO is replaced by a permanent magnet, which will provide a sufficient magnetic field for BWO operation up to 700 GHz (ca. 1.25 Tesla).

Another other option, a Quantum Cascade Laser, is pursued in parallel since it promises sufficient power, too, and a certain range of tunability. For tuning, an external cavity, a PLL-type frequency locking circuit, probably strong superconducting magnet coils with minimised stray field for use aboard SOFIA and further safety features are under development.

Four four-channel acousto-optical spectrometer (AOS) (Fig. 6) will be used to spectrally analyze the receiver outputs. To shift the output bandpass of the mixer elements to the required input bandpass of the AOS at 2.1 GHz +/- 0.5 GHz and to provide calibration signals for the backends, IF processors need to be included. For the KOSMA array-receiver we have developed a miniaturized plug-in version of such devices (Fig. 7), from which we will develop the IF processors for STAR.

Fig 6: A schematic of the array-AOS optics.

Fig. 7: Miniaturized single channel IF processor plug-in module.The module consists of two frequency mixing stages, a step attenuator to set the output power level, a detector for power monitoring, a zero switch to measure the AOS dark current and a directional coupler to inject a comb signal for the frequency calibration of the AOS. It also includes all the filtering and amplification required, including a 25 dBm power amplifier at the output.

Fig 5: A 350 GHz backward wave oscillator integrated into a permanent magnet. The dimension left to right is approximately 28 cm. For STAR we will purchase a BWO for 633 GHz.

Contact

References:

2005
1. Armin Wagner-Gentner, K.Jacobs, U. U. Graf, and D. Rabanus, 2005. Low Loss THz Window. Proc. of the 16th Int. Symp. on Space THz Technol.
2. David Rabanus, Christophe Granet, Axel Murk and Thomas Tils, 2005, Measurement of properties of a smooth-walled spline-profile feed horn around 840~GHz. IRPT, submitted
3. T. Lüthi, D. Rabanus, U. U. Graf, C. Granet and A. Murk, 2005. A new multibeam receiver for KOSMA with scalable fully reflective focal plane array optics, 2005, submitted to ISST.
4. T. Tils and A. Murk and D. Rabanus and C.E. Honingh and K. Jacobs, 2005, High performance smooth-walled horns for THz waveguide applications. ISSTT 16, submitted
2004
5. Martin Philipp, U. U. Graf, F. Lewen, D. Rabanus, A. Wagner-Gentner, P. Müller, J. Stutzki 2004 Compact 1.6-1.9 THz local oscillator as stand-alone unit for GREAT. Proc. of the 15th Int. Symp. on Space THz Technol. 15:248-254
6. David Rabanus, Urs U. Graf, Martin Philipp, Jürgen Stutzki and Armin Wagner-Gentner, 2004, Cryogenic design of KOSMA's SOFIA Terahertz Array Receiver (STAR). SPIE 5498:473-480
2002
7. D. Rabanus, U. U. Graf, S. Heyminck, S. Stanko, and J. Stutzki, 2002. Techniques for Heterodyne Array Receivers. Proc. Far-IR, SubMM & MM Detector Technology Workshop NASA/CP-211408, 179-182.
8. S. Heyminck, U. U. Graf, E. A. Michael, S. Stanko, D. Rabanus, C. E. Honingh, K. Jacobs, R. Schieder, and J. Stutzki, 2002. Heterodyne Array Receiver Development at KOSMA. Proc. 2nd Workshop on New Concepts for Far-IR Submillimeter Space Astronomy 2.
9. S. Stanko, U. U. Graf, and S. Heyminck, 2002. Automatic Tuning of SMART, KOSMA's 490/810 GHz Array Receiver. Proc. of the 13th International Symposium on Space THz Technology 13.
2001
10. S. Heyminck and U. U. Graf, , 2001. Array Receiver LO Unit using Collimating Fourier-Gratings. Proc. 12th International Symposium on Space Terahertz Technology 12.
11. U. U. Graf und S. Heyminck, 2001. Fourier Gratings as Submillimeter Beam Splitters. IEEE Transactions AP 49, 542.
12. U. U. Graf, S.Heyminck, E. A. Michael, and S. Stanko, , 2001. KOSMA's 490/810 GHz Array Receiver. Proc. of the 12th International Symposium on Space Terahertz Technology 12.
2000
13. Michael, E.A., Graf, Urs U., Honingh, C. E., Jacobs, K., Lewen, F., Schieder, Rudolf T., and Stutzki, Juergen, 2000. SOFIA terahertz array receiver (STAR). Proc. SPIE Vol. 4014, p. 109-115, Airborne Telescope Systems, Ramsey K. Melugin; Hans-Peter Roeser; Eds. 4014, 109-115.
14. R. Güsten, P. Hartogh, H.-W. Hübers, U. Graf, K. Jacobs, H.-P. Röser, F. Schäfer, R. Schieder, R. Stark, J. Stutzki, P. v. d. Wal, and A. Wunsch, 2000, GREAT - The First-Generation German Heterodyne Receiver for SOFIA. Proc SPIE: Airborne Telescope Systems 4014.
15. S. Heyminck und U. U. Graf, 2000, Reflection Gratings as THz Local Oscillator Multiplexers. Proc. SPIE: Airborne Telescope Systems 4014.
16. U. U. Graf und S. Heyminck, , 1999. Fourier Gratings as THz Local Oscillator Multiplexers. Proc. IEEE Seventh International Conference on Terahertz Electronics
1997
17. F. Lewen, E.Michael, R. Gendriesch, J. Stutzki and G. Winnewisser, 1997. Terahertz Laser Sideband spectroscopy with Backward Wave Oscillators. Journal of Molecular Spectroscopy 183, 207-209.