Greg Blackman discovers that building larger ground-based telescopes is pointless without adaptive optics
How do you see further into space, observe the beginnings of the Universe, or pick out planets orbiting distant stars in which the host star burns a billion times brighter than the planetary system? Part of the answer is simply to build larger telescopes that gather more light – and the largest telescopes in the world, in the eight to 10m class, are certainly employed in these investigations. There are also extremely large telescopes under development, with apertures of more than 20m diameter. These include the European Extremely Large Telescope (E-ELT), the Giant Magellan Telescope (GMT), both planned for construction in Chile, and the Thirty Meter Telescope (TMT) to be built in Hawaii.
Building a larger telescope isn’t the whole story though; the ability to gather more light from the ground is useless without a way to overcome atmospheric turbulence that would otherwise limit the resolution of the instrument. Therefore, most large telescopes use adaptive optics (AO)to improve their resolving power. ‘Without adaptive optics, once the telescope becomes more than about a fifth of a metre in diameter, the resolution of the images doesn’t get any better, the instrument just collects more light,’ says Glen Herriot, systems engineer working on The Narrow-Field Infrared Adaptive Optics System (NFIRAOS) at the National Research Council of Canada. ‘This is because the atmospheric turbulence blurs the image.’
NFIRAOS, pronounced ‘nefarious’, is being designed and built by the National Research Council of Canada for the planned Thirty Meter Telescope (TMT) scheduled for completion in 2019. The AO system is currently in the preliminary design stage.
NFIRAOS will take the light gathered by the TMT’s primary mirror, correct for atmospheric turbulence and some of the imperfections of the telescope itself, and feed the corrected light to three instruments on the telescope. The AO system will deliver a nearly flat wavefront, (191nm rms wavefront error over the 17 arcsecond field of view of the camera fed by NFIRAOS).
The maximum resolution of a telescope is determined by the diffraction limit of the instrument. Space-based telescopes, which don’t have the atmosphere to contend with, get close to this. NFIRAOS is designed to shepherd roughly half the light into the image area that would be produced by a perfect telescope – a Strehl ratio of about 1/2 in the near infrared.
‘Using AO to get close to the diffraction limit of the telescope means you win two ways,’ says Herriot. ‘Not only do you collect more light, but you also shepherd the light into a smaller spot whose area is proportional to the diffraction limited diameter of the spot. More light is collected and the sensitivity is increased, meaning greater scientific productivity, which increases with the telescope diameter to the fourth power.’
The optics in NFIRAOS operate at -30°C to reduce their thermal background. NFIRAOS therefore achieves 2.5 times the sensitivity of an uncooled AO system for observations in the K-band (2.0-2.4µm). This is a window in the atmosphere with low water vapour and OH light blockage and a prime observing band. ‘If we didn’t cool the system, a time exposure would take around 2.5 times as long,’ notes Herriot. ‘By cooling it we get tremendous observing efficiency.’
Large telescopes tend to conduct work with adaptive optics in the infrared, because the wavelengths are longer and the same atmospheric perturbation in relative terms is smaller than in the visible. The correction is more accurate in the infrared for a set number of deformable mirror actuators. ‘You can approach the diffraction limit in the near infrared at an affordable cost, which you couldn’t hope to do in the visible with today’s technology,’ states Herriot. ‘In general, astronomers would rather have really good performance in the infrared compared to lower performance in the visible wavelength.’
An adaptive optics system is typically comprised of a wavefront sensor, measuring the incoming light and determining the profile of the wavefront. A deformable mirror then alters its actuators according to the data received from the wavefront sensor to change its shape and flatten the wavefront.
Either a natural star is used to measure atmospheric turbulence, or, if observations are made in an area of the sky with no bright stars, an artificial laser guide star can be utilised. NFIRAOS will have six laser wavefront sensors and three wavefront sensors using natural guide stars to measure tip, tilt, image motion and defocus.
Laser guide stars use laser light to excite a layer of sodium ions at around 90km in altitude, created from micro-meteorites burning up in the atmosphere. The resulting artificial star allows the AO system to assess and correct for atmospheric turbulence. ‘You need to make a lot of higher order wavefront measurements in order to assess the turbulence in the atmosphere and you need to do it quickly,’ says Herriot. ‘There just aren’t enough bright stars providing enough photons to make such detailed measurements in millisecond timescales.’ He adds that calculating image motion and defocus does need a natural guide star.
Data from the wavefront sensors, all working in slightly different directions, is used for tomography and to calculate the 3D volume of turbulence in the telescope’s field of view. The system then calculates in real time the best shape for the NFIRAOS’ two deformable mirrors (DM) to correct the wavefront over the field of view. NFIRAOS is unique in the scale of the problem, Herriot says. It will have roughly five times the number of actuators than the AO system for the Hawaii-based 8m Gemini telescope. And the computing problem to take the information from the wavefront sensors and calculate how to shape the DM increases dramatically with the number of actuators.
Discovering new planets
NFIRAOS and the TMT have a long process ahead until they actually start making scientific observations. At the Subaru Telescope, an 8.2m telescope in Hawaii operated by the National Observatory of Japan, Olivier Guyon and his team are developing extreme adaptive optics to use the telescope to discover and chart planets around other stars.
‘Contrast is vital for these observations,’ explains Guyon. ‘To identify a planet the size of Earth around these stars, the contrast ratio would have to be at least one billion to one. That’s not possible from the ground with an 8m telescope. Therefore, we’re looking for planets around the size of Jupiter. Depending on the observation wavelength and the planet’s age and mass, these planets can be from 1,000 to 10 million times fainter than the star.’
The AO system being developed provides around one million to one contrast ratio – it’s able to see objects one million times fainter than the adjacent star. It uses a deformable mirror from Boston Micromachines. ‘The system is extreme in the way it’s designed to clean up the image, but this is only done over a small region of the sky,’ says Guyon. ‘We don’t need big deformable mirrors, but we do need them to be very precise with a lot of actuators. The MEMS technology from Boston Micromachines is well suited for these observations.’
Boston Micromachines’ MEMS deformable mirrors have no hysteresis – so, when a voltage is applied to a channel, it results in the same deflection every time. This compares to piezoelectric devices, in which the positioning of the actuators is dependent on their previous position.
‘The big innovation recently has been increases in actuator counts,’ comments Michael Feinburg, director of product marketing at Boston Micromachines. ‘The more control points you have, the more precisely you can compensate for atmospheric aberrations,’ he says, adding that devices have also become faster.
The AO system developed for the Subaru Telescope measures atmospheric turbulence around 1,000 to 2,000 times per second to achieve the high level of correction required. The system also employs a custom-designed coronagraph to remove the diffraction rings from around a star in the image. ‘Even if you have a perfect wavefront generated by adaptive optics, the star will have diffraction rings around it,’ explains Guyon. ‘Those rings mask the light from planets, so we have to build an optical system that removes them.’
The system has recently been tested on the Subaru Telescope for four nights for troubleshooting and to demonstrate it works. The team will begin scientific observations later this year, mostly looking at well known stars relatively close to Earth.
It doesn’t make sense to build a large telescope without adaptive optics, because the atmosphere would limit it to the resolution of a small telescope. Smaller telescopes, though, tend not to use AO. A team from Caltech Optical Observatories in San Diego County, led by Christoph Baranec, has developed an adaptive optics system, called Robo-AO, to increase the power of mid-sized telescopes.
Astronomers at Caltech have access to the Keck telescopes, two 10m machines that are on the leading edge of large telescopes that exist today, as well as the Palomar observatory with the 5m Hale telescope, which for almost 50 years was the largest telescope in the world. There are also several other smaller telescopes – 60-, 48-, 24-, and 18-inch instruments – which are used for other cutting-edge research. The 48- and 60-inch telescopes have recently been used to discover the supernova PTF11kly as part of the Palomar Transient Factory (PTF) project.
A lot of these mid-sized telescopes in the 1-3m range are not utilised as much compared to the large telescopes like Keck, where there is a lot of competition for time, notes Baranec. ‘The idea for Robo-AO was that, if these smaller telescopes can be equipped with adaptive optics, they can be made more powerful,’ he says. ‘Even a small, AO-equipped telescope has a higher resolution than a seeing-limited larger telescope. In addition, by robotising the AO system and the telescope [the 60-inch telescope is robotic and can run autonomously] you can start to do things that you can’t do with the large telescopes – that is, large observational programmes that wouldn’t be practical to do on Keck, for example, because of the restraints on time.
‘By putting an AO-system on a 60-inch telescope we’d have an enormous amount of time and be able to look at tens of thousands, or even hundreds of thousands, of targets. It would be possible to do really large targeted surveys that you’d never be able to do with Keck or with Hubble or with any of the other space telescopes.’
The team is trying to make it cheap, at least relative to the cost of a telescope. Robo-AO costs roughly US $0.5m. A typical 1.5m telescope might be in the $3-5m range, so adding $0.5m on top of that to greatly increase its capability is, for most, Baranec feels, worth it.
In addition, because the AO system is robotic, it can monitor specific targets over time. ‘We’re finding out now that the universe doesn’t really stand still. That’s one of the hot topics in astronomy right now,’ Baranec says. ‘At the moment, telescopes just take a snapshot, which doesn’t give any idea about how entities are changing. Astronomy is starting to get into the realm where we actually need to monitor things and see how they’re changing, and what’s interesting about their dynamics.’
Robo-AO uses a lot of the same components that the larger AO telescope systems have, but they’re all off-the-shelf components to fit small telescopes. It uses a Rayleigh laser guide star, as opposed to a sodium laser to reduce cost. The Rayleigh beacon based on a UV laser can only really be used with small telescopes.
The other advantage with equipping smaller telescopes with AO is that observations can be made in the visible spectrum, as opposed to the infrared, to which most of the larger telescopes’ AO systems are constrained.
The atmospheric aberrations scale with wavelength. ‘It’s very difficult, almost impossible, to do visible light correction on large telescopes right now,’ states Baranec. ‘It’s not that it’s physically impossible, it’s just that the technology isn’t quite there yet.’
However, for small telescopes, because they are measuring light travelling through less atmosphere, the AO system doesn’t have to work as hard and there is more tolerance in the system. ‘The nice thing is you can actually get visible light AO-corrected images, which in of itself is unique to smaller telescopes,’ Baranec continues, adding that it expands the spectrum and gives astronomers different ways to look at things, which makes these small telescope AO systems very scientifically compelling.
Robo-AO was developed over the last year and installed in April 2011, with the system fully operational in the middle of August 2011. To date it has captured images of the Blue Snowball Nebula NGC 7662 and a section of the Hercules Globular Cluster M13 showing masses of stars. With the AO system, the images are sharp enough to see some of the features in the nebula and determine individual stars in the M13 cluster with greater clarity.
‘We have a system that works and can get these beautiful images. We’re now trying to turn this into a fully robotic system to make the observations automatically,’ says Baranec. ‘This is something that no AO system does at the moment. Making it automated will increase the efficiency and we’ll be able to carry out an order of magnitude more observations per night than doing it manually with a team of scientists.’
The team also want to encourage other groups to build their own systems and hopes in the future to make public the system’s software and designs so that others can take advantage of it to further science.