The overall designs for planet-finding interferometers have changed substantially since Bracewell first proposed adding a
π phase shift within an astronomical interferometer to create the first nulling interferometer (Woolf and Angel 1998). This Bracewell nuller has a symmetric response on the sky, making it difficult to separate the contribution of a planet from the emission from zodiacal dust orbiting the target star (the exo-zodiacal emission). It is also vulnerable to small drifts in the stray light level or in the gain of the system that can mimic the planet signal. The solution was the Linear Dual Chopped Bracewell design, comprising two single Bracewell nullers that are cross-combined with a relative phase of ±
π/2. By taking the difference in photon outputs of these two phase chop states, the resulting response is anti-symmetric, and is insensitive to the both the symmetric exo-zodi emission and instrumental drifts.
In parallel with the introduction of phase chopping, there was development of higher order nulling configurations. The motivation here is to reduce the impact of stellar leakage on the null depth. Although light from the center of the star can be nulled completely, light from the edges will leak through to some extent. The Bracewell designs have a null that degrades away from the from the optical axis proportional to the angle-squared, leading to relatively high stellar leakage. The Angel Cross combines the light from 4 collectors to give a null that degrades proportional to the angle to the fourth power, reducing the stellar leakage to a negligible level (Angel and Woolf 1997; Beichman, Woolf and Lindensmith 1999). Variations on this design include the Degenerate Angel Cross (DAC) and Generalized Angel Cross (GAC). These basic nulling elements are cross-combined using phase chopping to generate configurations such as the Chopped Degenerate Angel Cross (a linear design based on DACs) and the 6-collector Bow-Tie design that was favored by ESA for a while (Fridlund et al. 2006, and references therein).
These chopped high-order null configurations were thought to be superior in performance to the Linear Dual Chopped Bracewell configuration, but this proved not to be the case. Simulations to predict the number of stars that could be surveyed for planets showed that the Linear DCB could survey approximately twice the number of stars compared to a Bow-Tie with the same total collecting area. The reason is that the Linear DCB is much more efficient at converting planet photons into modulated output signal - it has a higher 'modulation efficiency'. This is offset by the higher stellar leakage, but this only has a significant impact on the bright nearby stars that occupy only a small fraction of the total integration time available.
The architecture evolution from this point has followed two parallel tracks. At ESA, the emphasis was on minimizing the number of spacecraft used. The Diamond and Z-Array are both DCB configurations in which the one spacecraft serves the function of both collector and combiner. In both cases the beams make multiple hops from collector to combiner to balance the path lengths. Another development was the 3-telescope nuller. This is a departure from the DCB, in which the collectors are combined with phases of 0, ±2π/3 and ±4π/3. With 3 spacecraft, the equilateral triangle is the minimal configuration that still supports phase chopping, but the symmetry leads to undesirable imaging properties. It too uses multiple hops to relay the beams from collector to combiner. One of the designs currently favored at ESA is the right-angled Three Telescope Nuller with a dedicated beam combiner spacecraft to alleviate the complexity of the beam relay, and sufficient asymmetry to improve the imaging properties.
|
|
Schematic showing the evolution of the preferred nulling architecture for a TPF-I/Darwin mission. |
|
The focus at NASA was on improving the imaging properties of the array, which is important for separating the contributions from multiple planets, determining the orbit, and discriminating against lumps in the exo-zodi. This led to a rearrangement of the collectors in the Linear DCB to produce the X-Array - a configuration in which the nulling baselines lie along the short side of the rectangle and the imaging baselines (which determine the angular resolution) along the long dimension. The beams are relayed in a single hop from each collector to a central combiner. The decoupling of the nulling and imaging baselines makes the X-Array more flexible than other configurations. This flexibility was subsequently exploited to eliminate 'instability noise' with the 'Stretched X-Array' design. Instability noise - an analog of speckle noise in the coronagraph - arises from fluctuations in the path lengths, pointing, dispersion, etc. of the instrument, and drives the requirement on the null depth down to one part in a million. The long imaging baselines of the Stretched X-Array give the planet signals a unique spectral signature that can be effectively separated from the instability noise, thereby reducing the null depth requirement by a factor of 10. With this approach the required null depth is one part in 100 thousand, not one part in a million. Additionally the longer baselines greatly improve the angular resolution of the instrument.
The configurations above are defined by the relative location of the collectors. Until recently, the combiner spacecraft was always located in the same plane as the collectors, normal to the direction to the target star. ESA then proposed the 'Emma' architecture, in which the combiner is moved out towards the star by about 1 km, and the collectors are reduced to a single spherical mirror. Most of the nulling configurations already described can be implemented in either the classic planar format or the out-of-plane Emma format. The Emma design offers significant advantages which are presently being studied independently by ESA and NASA. Preliminary results of these studies were first reported in the later half of 2006. The appeal of the Emma design is primarily in its simplification of the telescope optics, eliminating the need of any deployables, and also in the design of the sunshields, which become folded into a hard shell, thereby reducing the risk of catastrophic failure. However, the simplification of the telescope optics increases the complexity of the beam combiner, and it currently is not known to what extent this will reduce the overall cost and risk of the mission. What is clear is that there exists obvious agreement in design principles between researchers at NASA and ESA, and the architectures for both TPF-I and Darwin appear to be converging in 2007.
References
Angel, R., and Woolf, N., "An imaging interferometer to study extrasolar planets," Astrophys. J. 475, 373-379 (1997).
Beichman, C. A., Woolf, N. J., and Lindensmith, C. A., TPF: A NASA Origins Program to Search for Habitable Planets, JPL Publication 99-3, Jet Propulsion Laboratory, Pasadena, CA (1999).
http://planetquest.jpl.nasa.gov/TPF/tpf_book/index.cfm
Beichman, C. A., Coulter, D. R., Lindensmith, C. A., and Lawson, P. R., editors, Summary Report on Architecture Studies for the Terrestrial Planet Finder, JPL Publication 02-011, Jet Propulsion Laboratory, Pasadena, CA (2002).
http://planetquest.jpl.nasa.gov/TPF/arc_index.cfm
Fridlund, C. V. M., d'Arcio, L., den Hartog, R., Karlsson, A. 2006, "Status and recent progress of the Darwin mission in the Cosmic Vision program," Advances in Stellar Interferometry, edited by Monnier, J. D, Schöller, M., Danchi, W. C., Proc. SPIE 6268, 62680Q (2006).
Woolf, N. J., and Angel, J. R. P., "Astronomical searches for Earth-like planets and signs of life," Ann. Rev. Astron. Astrophys. 36, 507-538 (1998).