Mid-IR stellar interferometer grows in capability
An IR stellar interferometer uses the radio technique of heterodyne detection, but at optical wavelengths; a third telescope has just been added and phase closure achieved.
The past several years have seen a resurgence of the field of optical long-baseline stellar interferometry. The concept was first suggested in 1868 by Armand Fizeau and successfully carried out by Albert Michelson in 1890 and Michelson and Francis Pease in 1921. It was decades later before advances in electronics, interferometry, and laser-distance measurements would contribute to further significant developments in stellar interferometry. Today there are more than a dozen ground-based stellar interferometers, with more being planned, including several space-based interferometers.1
Stationary paraboloid design
The University of California Berkeley Infrared Spatial Interferometer (ISI) is a stellar interferometer designed to make milli-arcsecond-resolution measurements of astronomical sources in the 9- to 12-µm mid-IR range from its site at the Mt. Wilson Observatory, 30 miles north of Pasadena, CA.2 Having operated since 1988 as a two-element interferometer, a third element was recently added to provide three simultaneous baselines and source-imaging information (see Fig.1). Each of the three elements of the array consists of a 1.65-m telescope coupled to an IR heterodyne receiver, both mounted in a standard semi-trailer. The telescopes use a Pfund-type design in which a siderostat reflects starlight onto a stationary paraboloid (see Fig. 2). This design has three major advantages: it is low to the ground and hence stable against vibrations; it is quite open, minimizing local air turbulence; and it allows for mobility of the telescopes for different baseline distances and orientations, providing for varied spatial resolutions. The site allows for baselines ranging from 4 to 70 m in several configurations.
The 1.65-m paraboloid focuses reflected starlight at /3.1 back through a hole in the flat mirror into a Schwarzschild optical system consisting of a pair of spherical front-surface mirrors that change the beam to /89, thus presenting a less-divergent beam to later optics. The beam is then split by a dichroic beamsplitter that reflects the mid-IR (approximately 9 to 13 µm) and transmits the near-IR (approximately 1 to 3 µm). Near-IR radiation is sent to an indium antimonide focal-plane array for guiding and tip-tilt correction. The mid-IR stellar radiation is sent to a second beamsplitter where it is combined with radiation from a carbon dioxide (CO2) laser. This marks the beginning of the heterodyne receiver.
Heterodyne at optical wavelengths
Heterodyne detection for astronomy is a process of mixing a stellar signal and a generated signal (from a local oscillator) onto a nonlinear detector. The detector mixes (multiplies) the two signals to produce a response at frequencies equal to the sum and difference of the input frequencies, while preserving the amplitude and phase information contained in the original signals. The sum-frequency signal is discarded, but the difference-frequency signal presents one of the great advantages of heterodyne detection: this signal (termed the intermediate frequency, or IF) is at a much lower frequency that is easily amplified, distributed, filtered, and so on, using conventional radio and microwave techniques. While heterodyne methods are standard for almost all radio telescopes, the ISI is unique in its use of heterodyne detection at optical wavelengths.
Our local-oscillator (LO) signals are generated by CO2 gas lasers similar to the Freed design.3 They operate with a meter-long semiconfocal cavity having a concave mirror at one end with a radius of curvature of twice the cavity length, and a grating at the other end in a Littrow configuration acting as a plane mirror where the beam waist is located. Oscillation in undesirable modes is prevented by apertures inside the cavity. The gas can be easily changed among a variety of isotopes providing a series of wavelengths between approximately 9 and 12 µm. Within any isotopic series, different laser lines are selected by setting the tilt of the 80-groove/mm diffraction grating. Four invar rods define the nominal cavity length and provide passive frequency stabilization; the gas discharge tube is surrounded by a jacket of circulating, temperature-controlled water. Laser output power is stabilized via a computer-controlled servo loop that can make small adjustments to the cavity length by means of a piezoelectric element. The laser beam is focused by zinc selenide (ZnSe) lenses through a 250-µm pinhole spatial filter, and then combined with the stellar radiation on a beamsplitter. The beams propagate through an antireflection (AR)-coated /1.5 aspheric ZnSe lens and an AR-coated ZnSe window into a liquid-nitrogen Dewar, where they come to focus on the detector.
The heterodyne mixer/detectors currently used by ISI are single-pixel mercury cadmium telluride photodiodes that operate at 77 K capable of responding to frequencies up to approximately 3 GHz; thus, the total (double-sideband) bandwidth of the IF signal from the detector is nearly 6 GHz, or 0.2 cm–1. Because heterodyne down-conversion converts mid-IR stellar radiation of approximately 30 THz to an IF signal of 3 GHz, the IF signals from all three telescopes pass easily through coaxial cables to a central point for further processing.
Heterodyne eases positioning Interference fringes are produced by multiplying these IF signals in a radio-frequency (RF) correlator, which is the electronic analog to interfering optical beams in a direct-detection interferometer. However, each telescope receives stellar radiation along a slightly different path that changes in time with Earth's rotation. To create interference fringes, radiation received by each telescope must remain coherent until correlation, with a precision better than a fraction of the speed of light divided by the usable bandwidth. For a direct-detection interferometer, this could require positioning of optomechanical components with submicron accuracy, but for the ISI this means a positional accuracy at the intermediate frequencies of only about 0.5 cm. Therefore, our delay lines consist of sets of coaxial cables, switched into place in binary coded increments of 2n cm using computer-controlled RF relays.
In addition to geometrical delay compensation, because the IF signals are obtained by heterodyne mixing with separate laser local oscillators, a precise phase relationship must be maintained between the laser local oscillators in each telescope for the detected radiation to remain coherent. To accomplish this, each telescope receives a beam from a fourth, "master" CO2 laser, which it uses as a reference in an electronic phase-locked loop, adjusting its own laser to remain in phase with the master. Because the master laser beams are transmitted through relatively long paths of open air, additional phase errors due to optical pathlength fluctuations must also be corrected. This is accomplished by a novel pathlength-compensation system, in which each telescope returns a portion of the master laser beam back along exactly the same path and combines it with a portion of the original master beam that has been modulated by a small amount in pathlength to serve as a reference.
The round-trip beam and the original beam are mixed and detected in a manner similar to that used for detecting starlight, producing a sinusoidal signal at the frequency of modulation of the reference path. When the two beams are at the null of their interference (that is, in phase) the sinusoidal frequency has zero amplitude. Thus, the amplitude of this detector output is proportional to the pathlength variation of the beam propagating to the telescope. An error signal derived from this output controls the separation of two parallel mirrors between which the master laser beam reflects several times before being transmitted to a telescope, providing control of the beam's pathlength.
The ISI has been used primarily to study the spatial distribution of dust around late-type stars by measuring dust-shell inner radii, temperature, and optical depth. In some cases, diameter measurements of the central star have been obtained to a precision of 1%. Continual monitoring has identified variations in stellar luminosity and also movements and changes in the surrounding dust. The narrowband heterodyne receivers are also well suited for spectral-line research and have been used to measure emission lines of stratospheric ammonia (NH3) produced by the collision of Comet SL9 with Jupiter, as well as absorption-line profiles of CO2 in the Martian atmosphere and of NH3 in carbon stars. In addition, interferometric measurements on spectral lines have been able to locate molecular-formation regions of silane and NH3 around some carbon stars.4
Stellar interference fringes have an amplitude and a phase that provide information about the geometry of the star, but fringe phase measured with a two-element interferometer is too distorted by atmospheric fluctuations to be meaningful. The new third telescope has made it possible to recover much of the previously lost phase information. When measured around a closed triangle of baselines, even random and unknown phase distortions cancel out, leaving essentially the intrinsic phase information contained in the source. From such phase information, determinations regarding the shape and symmetry of sources can be made. On July 9, 2003, the ISI recorded the first-ever measurements of three simultaneous fringes, demonstrating "closure phase" on a stellar source at mid-IR wavelengths.
- C. Freed, IEEE J. Quant.. Elect. QE-4(6) 404 (1968).
- J. D. Monnier et al., Astrophy.s. J. 543, 868 (2000).
DAVID D. S. HALE is an astronomer at the Mount Wilson Observatory, Mount Wilson, CA 91023; e-mail: email@example.com.