Observations at near-IR wavelengths (0.9—2.5μm) are crucial for many areas of astronomy. Deep near-IR observations, i.e., those able to reach faint intensity levels with adequate signal-to-noise ratios, allow us to study galaxies at distances of 13 billion light years. These are the first galaxies that formed, when the universe was only 700 million years old. At the other end of the scale, the lowest-mass stars and highest-mass planets also emit (or reflect) most of their light at near-IR wavelengths. However, observations in this spectral regime are very challenging because of the very bright near-IR background. Between 0.9 and 1.8μm, almost all of this background results from de-excitation of atmospheric hydroxyl (OH) molecules at an altitude of ∼90km, giving rise to an extremely bright emission-line spectrum1 (see Figure 1, top panel). In addition, this OH emission is also variable on a timescale of minutes. Subtracting this background from astronomical observations is very inaccurate because of the inherently large Poissonian and systematic noise by which such observations are affected. Solving this difficulty is a long-standing problem in astronomy.
Figure 1.Results of our OH-suppression technology. (top) Model of the night-sky spectrum at near-IR wavelengths (λ), illustrating the strong hydroxyl (OH) emission lines. The J and H passbands are shown by the hatched regions. ph: Photons. (second panel) Zoomed region between 1.44 and 1.7μm (shaded area in the top panel). (third panel) Measured transmission of a fiber-Bragg grating (FBG), showing that 63 OH lines are suppressed from 1.44 to 1.63μm and that the match to the model sky spectrum is excellent. Insets: Typical notch profile and internotch throughput. (bottom) Results of our on-sky demonstration. The blue spectrum shows the night-sky spectrum measured by the control fiber, while the black spectrum shows the sky spectrum after OH suppression by the FBG. All sky lines corresponding to the FBG notches are suppressed.
A solution to this problem is so desirable that NASA, the European Space Agency, and the Canadian Space Agency are jointly planning to launch the IR James Webb Space Telescope to circumvent the IR background. In 2014, this telescope will be placed into an orbit 1.5 million km from Earth, at a cost of ≈$4.5 billion.2 It is also desirable to achieve a ground-based solution. Space-based telescopes have a finite lifetime and are expensive to replace, while ground-based facilities can take advantage of developments in technology (e.g., detectors) and scientific understanding, and allow us to optimize instruments for specific requirements as the need arises.
Three separate approaches to solving this problem have been pursued in the past. Ultra-narrow-band filters have been used to limit observations to a very small wavelength range, uncontaminated by OH emission lines.3 Unfortunately, this restriction also limits the number and nature of objects that can be observed, because only those targets that have a specific, unambiguously identifiable feature within this narrow wavelength range can be identified.
Attempts have been made to mask out the unwanted sky lines by dispersing the incident beam at very high spectral resolution, blocking out the OH emission lines, and then recombining the light.4,5 However, this technique is only partially effective, since dispersion also unavoidably scatters the OH light, such that it can no longer be blocked. This scattering is a seemingly intractable feature of spectrographs.6
In addition, there have been attempts to build holographic filters7 to remove the OH lines before the light enters the spectrograph, but so far these have unacceptably high light losses if one wants to filter out a large number of lines over a wide wavelength range. We have developed (and recently demonstrated) a highly efficient filter imprinted in an optical fiber that suffers from none of these difficulties.
Fiber-Bragg gratings (FBGs) are optical fibers characterized by a periodic variation in their refractive index. As light propagates down the fiber, some of it is reflected at each refractive-index change. The amount of light that is reflected is small, but by tailoring many reflections to be in phase with each other, it is possible to build up strong reflection at a particular wavelength. Such devices are well known in the telecommunications industry, for which they were originally developed. FBGs are highly efficient filters: suppression by a factor of >30dB at the reflected wavelength is easily possible, while maintaining very high throughput (<0.5dB) at other wavelengths. This makes them very attractive for filtering out the OH sky lines.
However, we first had to solve two problems. We need to suppress around 300 OH emission lines between 0.9 and 1.8μm. This could, in principle, be achieved with a series of fibers, tailored separately for each line (wavelength), but such a setup would be extremely inefficient and costly. We achieved a first breakthrough when we demonstrated8 that FBGs could be printed with ∼20 notches, with wavelengths accurately tuned to the OH sky lines. Further developments9 have progressed such that FBGs can now be written with ∼70 notches, covering a wavelength range of 140nm (see Figure 1). In addition, the transmission of the notches is as low as 50dB in some cases, resulting in sky-line suppression by a factor of 105. At the same time, the transmission between the notches is better than 0.3dB, which means that fiber losses amount to less than 10%. Two of these fibers in series could each cover the J and H passbands, windows of high atmospheric transmission—centered at ∼1.1 and 1.6μm—that are routinely used for astronomy (some parts of the IR spectrum cannot be observed from the ground because of atmospheric absorption by water and other molecules).
The second problem that we had to solve was that of fiber aperture. Coupling starlight into a single-moded fiber (10μm core diameter) via a telescope is very difficult, because the wavefront of the light is distorted by turbulence in the atmosphere and the acceptance angle of a narrow fiber is small. Therefore, in astronomy we use much larger fibers with cores of >80μm (although adaptive-optics correction of the wavefront allows use of smaller fibers). However, the latter can support many modes of propagation. The effect of this is to smear out the narrow notches of FBGs into very broad, shallow notches, which makes them useless as filters. This problem has been solved with a ‘photonic lantern.’ This device takes a multi-moded fiber and splits it into the appropriate number of single-moded fibers using a fiber taper.10,11 FBGs can then be printed in the single-moded fibers before a reverse transition into a multi-moded fiber. These devices reproduce exactly the behavior of a single-moded FBG.10
We successfully demonstrated our OH-suppression technology in an experiment at the Anglo-Australian Telescope in December 2008. Two 60μm-core fibers were pointed directly at the night sky through a window in the telescope's dome. Both fibers fed into a photonic lantern equipped with seven single-moded fibers. In one of the photonic lanterns, identical FBGs were printed into each of the seven single-moded fibers, each of which suppressed the 63 brightest OH lines between 1.44 and 1.63μm. In the other lantern, standard single-moded fibers were used as control. The photonic lanterns then fed into a spectrograph with a resolving power of λΔλ≈2400 (λ: wavelength). We obtained exposures of 900s. The bottom panel of Figure 1shows that in the region over which the gratings acted, the OH sky lines were strongly suppressed compared to the control fiber. Outside this region, the sky lines are detected, as for the control fiber, demonstrating high internotch throughput.
We have, thus, successfully demonstrated a novel technology that provides a solution to the long-standing problem of the effect of the bright near-IR background on astronomical observations. Our technology will reduce the background by a factor of ≈40 compared to current values.1 The scientific impact of this has huge potential for many areas of astronomy, especially for studies of the early universe, allowing much deeper observations than currently possible.1
We are now in the process of designing the first instrument to use this technology. The Gemini North OH Suppression Integral-Field-Unit System (GNOSIS) is fully funded by the Australian Research Council and should see ‘first light’ by 2012. In its first implementation, it will feed the IR Imager and Spectrograph-2 (IRIS2)12 on the 3.9m-diameter Anglo-Australian Telescope and will have full OH suppresssion from 1.4 to 1.8μm. Planned future upgrades include extending the wavelength range of OH suppression to 0.9–1.35μm, and relocating the instrument to the 8m-diameter Gemini North telescope in Hawaii, where it will feed the recommissioned Gemini Near Infrared Spectrograph (GNIRS).13 In the future, we envisage fully OH-suppressed spectrographs on the next generation of 30m-diameter telescopes, which should provide even more sensitivity than that offered by the James Webb Space Telescope.
Australian Astronomical Observatory
Simon Ellis is an instrument scientist. He works on OH suppression with FBGs and is the project scientist for the GNOSIS instrument. His research interests also include galaxy formation and evolution.
Sydney Institute for Astronomy, The University of Sydney
Joss Bland-Hawthorn is a Federation Fellow. He heads the astrophotonics group. He has developed many novel solutions for astronomical instrumentation, including Fabry-Perot etalon-tunable filters and nod-and-shuffle sky-subtraction on CCDs. Current developments include OH suppression with FBGs, deployable integral-field units with hexabundle fibers, and integrated photonic spectrographs.
1. S. C. Ellis, J. Bland-Hawthorn, The case for OH suppression at near-infrared wavelengths, Mon. Not. R. Astron. Soc. 386, pp. 47, 2008.
2. J. P. Gardner, Science with the James Webb Space Telescope, The Scientific Requirements for Extremely Large Telescopes, pp. 87-98, 2006.
3. A. Horton, I. Parry, J. Bland-Hawthorn, S. Cianci, D. King, R. McMahon, S. Medlen, DAzLE: the dark ages z (redshift) Lyman-α explorer, Proc. SPIE
5492, pp. 1022, 2004. doi:10.1117/12.551591
4. F. Iwamuro, K. Motohara, T. Maihara, R. Hata, T. Harashima, OHS: OH-airglow suppressor for the Subaru telescope, Proc. Astron. Soc. Jpn 53, pp. 355, 2001.
5. K. Motohara, F. Iwamuro, T. Maihara, S. Oya, H. Tsukamoto, M. Imanishi, H. Terada, M. Goto, CISCO: cooled infrared spectrograph and camera for OHS on the Subaru telescope, Proc. Astron. Soc. Jpn 54, pp. 315, 2002.
6. T. N. Woods, R. T. Wrigley III, G. J. Rottman, R. E. Haring, Scattered-light properties of diffraction gratings, Appl. Opt. 33, pp. 4273, 1994.
7. S. Blais-Ouellette, É. Artigau, F. Havermeyer, K. Matthews, C. Moser, D. Psaltis, G. J. Steckman, Multi-notch holographic filters for atmospheric lines suppression, Proc. SPIE
5494, pp. 554, 2004. doi:10.1117/12.552116
8. J. Bland-Hawthorn, M. Englund, G. Edvell, New approach to atmospheric OH suppression using an aperiodic fibre Bragg grating, Opt. Express 12, pp. 5902, 2004.
9. J. Bland-Hawthorn, A. Buryak, K. Kolossovski, Optimization algorithm for ultrabroadband multichannel aperiodic fiber Bragg grating filters, J. Opt. Soc. Am. 25, pp. 153, 2008.
10. S. G. Leon-Saval, T. A. Birks, J. Bland-Hawthorn, M. Englund, Multimode fiber devices with single-mode performance, Opt. Lett. 30, pp. 19, 2005.
11. D. Noordegraaf, P. M. Skovgaard, M. D. Nielsen, J. Bland-Hawthorn, Efficient multi-mode to single-mode coupling in a photonic lantern, Opt. Express 17, pp. 1988, 2009.
12. C. G. Tinney, S. D. Ryder, S. C. Ellis, V. Churilov, J. Dawson, G. A. Smith, L. Waller, J. D. Whittard, IRIS2: a working infrared multi-object spectrograph and camera, Proc. SPIE
5492, pp. 998, 2004. doi:10.1117/12.550980
13. J. H. Elias, D. Vukobratovich, J. R. Andrew, M. K. Cho, R. W. Cuberly, K. Don, A. Gerzoff, C. F. Harmer, Design of the Gemini near-infrared spectrometer, Proc. SPIE 3354, pp. 555, 1998.