In the world of Cosmology and Astronomy, SCUBA  was doubtlessly one of the most important instruments ever built. SCUBA has served the scientific community well, it seems however Astronomers have outgrown her.
SCUBA is mounted on what is the world’s largest submillimetre telescope at the Joint Astronomy Centre on the island of Mauna Kea, Hawaii, move over SCUBA, SCUBA 2
SCUBA-2 (SCUBA – Submillimetre Common User Bolometer Array) camera is housed at the James Clerk Maxwell Telescope – JCMT – on the island of Mauna Kea, Hawaii, is the world’s largest submillimetre camera. Submillimetre refers not to the physical size of the new camera itself, but to the submillimetre waveband between the far-infrared and microwave wavebands that the telescope observes. Being far more sensitive and powerful than its predecessor, SCUBA-2 will be able to map areas of the sky faster than ever before and provide information about the early life of stars, planets and galaxies.
Submillimetre observations provide information about the early life of stars by allowing astronomers to examine molecular clouds and dark cloud cores that are usually obscured by the very dust and gas cloud that collapsed under its own gravity to form the star.
“When you look up at the stars, you only see the light they are emitting in the visible part of the spectrum. Many galaxies, including our own Milky Way, contain huge amounts of cold dust that absorbs visible light and these dusty regions just look black when seen through an optical telescope. The absorbed energy is then re-radiated by the dust at longer, submillimeter, wavelengths”, explains Professor Gary Davis, Director of the JCMT.
As well as being the world’s largest submillimeter telescope, Professor Gary Davis says the interior of SCUBA-2 is, “colder than anything in the Universe that we know of!” To allow it to detect the extremely low energy radiation in the submillimeter waveband emitted by the very cold material associated with the earliest evolutionary stages of galaxies, stars and planets, the detectors inside SCUBA-2 have to be cooled to just 0.1 degree above absolute zero (-273.5°C or -460.3°F).
SCUBA-2 is a replacement for SCUBA, which was retired from service in 2005. Containing more than 10,000 superconducting sensors, the 4.5-ton SCUBA-2 measures 9.8 ft (3 m) high, 7.9 ft (2.4 m) wide and 8.5 ft (2.6 m) deep. With four sub arrays of 1,280 pixels each, the submillimeter camera boasts 5,120 pixels and provides a much larger field of view and sky-background limited sensitivity and will be able to map large areas of the sky up to 1,000 times faster than its predecessor.
“With SCUBA, it typically took 20 nights to image an area about the size of the full Moon. SCUBA-2 will be able to cover the same area in a couple of hours and go much deeper, allowing us to detect faint objects that have never been seen before,” said Professor Wayne Holland of the UKATC, and the SCUBA-2 Project Scientist.
SCUBA-2 will allow astronomers to map sites of star formation within our Milky Way galaxy, as well as planet formation around nearby stars. Along with surveying our galactic neighbors, it will also be able to look deep into space to examine faint, faraway galaxies and help researchers understand how galaxies evolved since the Big Bang. Its speed will also allow for speedier identification of targets for further high-resolution examination by other telescopes, such as the Atacama Large Millimeter/submillimetre Array that started operations earlier this year.
SCUBA-2 was constructed as part of an international collaboration between the UKATC in Edinburgh, four universities from Canada and Britain (British Columbia, Cardiff, Edinburgh and Waterloo), the US National Institute of Standards and Technology (NIST), and the Joint Astronomy Centre, which operates the James Clerk Maxwell Telescope.
In science there is always a nudge that gets new developments across the line, that nudge is often unseen. In the case of SCUBA 2 that nudge – we suspect – came in the form of the following:
SCUBA-2, the next generation, wide-field submillimetre camera for the JCMT
Ian Robson and Wayne Holland
Joint Astronomy Centre,660 N. A’Ohoku Pl., Hilo, HI96720, USA
UKATC, Royal Observatory, Blackford Hill, Edinburgh, Eh9 3HJ, UK
Following on from the enormous success of SCUBA on the JCMT in totally revolutionizing submillimetre continuum astronomy, its successor, SCUBA-2 is now well into the development phase. SCUBA-2 will be a simultaneous dual-waveband, wide-field imager, having an unvignetted field-of-view of 64 arcminutes square, limited in sensitivity by the sky background alone, reaching the confusion limit in around an hour. SCUBA-2 will utilize an array of superconducting TES devices built by NIST and theUniversityofEdinburgh. The array will be directly illuminated and the pixels will be half the diffraction spot diameter, requiring 25,600 pixels at 450 microns and 6,400 at 850 microns. This architecture will allow full diffraction resolution imaging without the need to jiggle the secondary mirror. SCUBA-2 passed its Conceptual Design Review in 1999 and the detector architecture downselect in May 2000, and has been enthusiastically supported by the JCMT user community, the JCMT Board
and PPARC. The project is led by the prime contractors the UKATC, and, like SCUBA, involves close collaboration with QMW.
SCUBA  is undoubtedly one of the most important instruments ever built for astronomy. As well as being the first large-scale array for submillimetre astronomy, it is a true facility instrument. SCUBA is mounted on the world’s largest submillimetre telescope, which is well supported by scientific and technical staff; it has a dedicated suite of data reduction software making data analysis and hence publication of the results readily achievable. As such, SCUBA has opened up the submillimetre, perhaps the last unexplored window on the Universe. Three areas of science stand out as truly revolutionary: galaxy evolution in the early Universe [e.g. 2,3,4]; dust disks around main-sequence stars [e.g. 5,6]; large-scale survey programmes addressing star formation [7,8].
In 1998 the JCMT Board set up an International Review to address the future scientific direction and competitiveness of the facility. In preparation for that review, one of us (IR) looked at the competition SCUBA would be facing from other facilities and also the science that would be needed to be done. One thing was clear, a larger and faster array was the obvious next development. Although SCUBA continues to be upgraded to improve its performance and reliability, there are natural limitations beyond which it is more cost effective to build a new instrument rather than continue upgrading. Again, one of us (IR) issued a challenge to see if it was possible to construct a much larger array: one that would fill the field-of-view of the JCMT; that would have no moving parts (and hence would be simpler to build and operate than SCUBA); would operate at helium-three temperatures (as opposed to 100mK for SCUBA); would use state-of-the-art (purchased) arrays and would be, in effect, the first CCD-type of submillimetre imager, potentially allowing stare-mode of observing without the AC chopping/nodding of previous devices. Two of us (WH & WD) responded to the challenge and came up with a pre-proposal plan which eventually developed into what will be presented here as SCUBA-2.
2 Why we need a SCUBA-2
Many key scientific programs require large areas to be imaged in order to provide unbiased surveys of sources and source structure. SCUBA has a limited field-of-view (fov) of only 2.3 arcminutes, which each of the two arrays simultaneously samples. This immediately poses limitations on the key science programs that can be undertaken in reasonable times, even on the JCMT with its fully flexible scheduling mode of operation and on the excellent site ofMauna Kea. Thus, the main reason for a new SCUBA is the requirement for a much larger field-of-view. The unvignetted fov of the JCMT at Nasmyth focus is a circle of diameter just over 11 arcminutes. Although the optical design challenges are severe, it is a key goal that a wide-field replacement for SCUBA to use a minimum of an 8 by 8 arcminute patch of the unvignetted focal plane. This is the maximum size of square array that will fit the field of view without having obscured pixels. Furthermore, the throughput of SCUBA is less than the design goal and so there is scope for improving the throughput and hence raw pixel sensitivity, so that a SCUBA-2 would be truly photon-noise limited by the background sky. As the JCMT moves into a new era in which it will operate as part of the Smithsonian Submillimetre Array (to give subarcsecond imaging) and has a suite of heterodyne cameras, the time available for continuum imaging will be reduced. Hence, being able to map and image faster is a clear goal for facility productivity. Such a device would also keep the JCMT at the forefront of submillimetre astronomy, a key factor for the funding agencies of theUK,Canadaand theNetherlands.
However, achieving such an ambitious goal is only possible if the arrays can be procured, and this was the key factor in the early phase of the project. The fact that such arrays have now been demonstrated to work in the lab, and can be constructed is the breakthrough that changes SCUBA-2 from an interesting design study to a feasible instrument.
3 The Scientific Goals
The user community was polled with respect to their interest in a SCUBA-2 opportunity and the response was extremely encouraging, in fact amazingly so. A huge range of science goals were identified, from solar system to cosmology, and these are listed on the scientific case for SCUBA-2 which is presented on the SCUBA-2 web-pages of the JCMT. In terms of operational goals, the key requirements are: the ability to reach the confusion limit in an hour or two enabling very deep imaging to be carried out; the ability to map, to a reasonable depth, an area of several square degrees in a few hours enabling large-scale mapping to be undertaken; to have dual waveband capability for spectral index determination of dust properties (and in the context of this meeting – redshifts). Finally, SCUBA-2 must be sensitive to point-source photometry of known objects enabling the spectral energy distributions to be determined. This latter goal implies that mapping speed cannot be traded for a loss of pixel sensitivity.
The goals were then refined to a specification of a mapping speed of at least 100 times that of the current (upgraded) SCUBA, a pixel sensitivity at least 50% better than the upgraded SCUBA and simultaneous operation at two wavelengths. DC operation was preferred with no sky-chopping. High image fidelity and map dynamic range were also important goals, all of which can be summarized by “faster, deeper, better”. This will allow SCUBA-2 to undertake very deep, but large-scale extragalactic surveys (crucial for the topic of this meeting) and very large-scale, high-resolution Galactic surveys. Table 1 gives an indication of examples of SCUBA-2 performance assuming the baseline design parameters.
Table 1. Examples of SCUBA-2 performance
Integration time (hours)
SCUBA SCUBA-2Point-source photometry to 5-s flux limit of 2 mJy
Map of the Hubble Deep Field (N) to noise of 0.5 mJy
Galactic plane survey (20´2o) to noise level of 30 mJy
Survey 5o diameter molecular cloud to noise of 10mJy
Deep extragalactic survey of a 1 deg2 area to noise level of 0.5mJy
4 Baseline specification
SCUBA-2 will consist of two arrays operating simultaneously at currently preferred wavelengths of 850 and 450 microns and they will be diffraction limited at each wavelength. The field-of-view will be a minimum of 8×8 arcminutes, the detectors will be background limited from the sky and the pixels will be DC coupled. This field-of-view, of some 64 arcminutes2, compares extremely well with other facilities
Figure 1. The detection rates as a function of 5s depth for a variety of instruments. The lines stop at the left-hand side due to confusion limits, and on the right where there is only one source in the sky. Note the highly competitive nature of SCUBA-2 for the deepest surveys, where the smaller spaceborne telescopes lose out to confusion. Diagram courtesy of Andrew Blain (see later this volume).
and coupled with the pixel sensitivity means that SCUBA-2 is very competitive with any instrument currently proposed (see Fig. 1). Indeed, SCUBA-2 also turns out to be extremely complementary to many other planned facilities, specifically the space and airborne observatories and in addition,ALMA. This latter point is particularly important as althoughALMAreaches a very deep level, and has a much lower confusion limit, its instantaneous field-of-view is extremely small and so having deep surveys to act as finding sources forALMAfollow-up is a key role for SCUBA-2. This is also the case for FIRST, albeit for different reasons.
Not to be forgotten the importance of imaging polarimetry has not been neglected and this will be possible with SCUBA-2, in the same manner as currently undertaken by SCUBA.
5 Instrument design
The heart of SCUBA-2 is the detector arrays. These will use the next generation detector technology to enable large format arrays to be realized. The detecting element will be a transition edge sensor (TES); these will be voltage biased and superconducting. The key point about a superconducting TES is that in the transition region between the superconducting and normal states, the resistance is an extremely steep function of the temperature. Hence such a device, coupled to an antenna to absorb photons is an extremely sensitive temperature sensor. Any change in the temperature (due to absorption of photons) of a TES film held at a constant voltage bias, results in a change of current which can be amplified and measured. Another key property of voltage biased TESs is that they are self-biasing. This is extremely important in ensuring that an array can be biased using a single bias supply despite small variations in the critical temperatures of the superconducting-normal bi-layers of each device during manufacture on the wafers. Self-biasing is achieved because the bias power is a source of heating for the device and as the temperature of the film is reduced below the critical temperature, the resistance falls sharply and the joule heating increases. This is a negative feedback loop in which an initial increase in temperature and hence resistance due to an absorption of photons is rapidly opposed by a decrease in the bias current and the Joule heating and a return to the original device temperature. Equilibrium is rapidly achieved when the heating and cooling are matched – these devices are true power meters. The extremely sharp transition region also means that this electro-thermal feedback effect can significantly speed up the device compared with the physical time constant for thermal relaxation not assisted by the feedback effect.
Transition edge sensors do not come without risk and complications, however. TESs require magnetic shielding, there is a question over their 1/f performance (which might limit the DC coupling requirement), there is a question about how robust they will be under repeated thermal cycling to helium temperatures, and finally, there is the question about their dynamic range when exposed to an uncertain power loading from photon illumination.
In all large array devices for the submillimetre, where the detectors are at helium temperatures, a key practical issue is the electrical leadouts from the array to the outside of the cryostat. Indeed, this is the limiting factor the size of arrays unless multiplexing is used. Multiplexing is an integral part of the SCUBA-2 novelty, and amplification and multiplexing is enabled through SQUIDS . The detector device architecture that was eventually selected is similar to that which has been in development for the SPIRE detectors for the FIRST space mission  . The construction of the arrays is very complex and is not covered here suffice to say that the array is in two parts. The top part is the micro-machined bolometer array, this is bump-bonded with indium to the lower SQUID multiplexer and interconnect chip. Needless to say, producing such a challenging, bonded component was one of the key factors in determining the detector architecture (see next section). The detector resides at 100mK, requiring additional technology such as a dilution refrigerator. Ribbon cables take the signals to the next stage at 0.8K where SQUID arrays perform amplification before the final amplification stage at 300K.
5.2 Detector architecture
The above description refers to the fully-filled, 0.5Fl architecture that came out of the detector downselect review on May 4th 2000. The basic choices for a large format array boil down to either the traditional format of 2 Fl spaced pixels fed by feedhorns (such as SCUBA), or, a filled array with the pixels spaced by half the diameter of the diffraction disk size (0.5 Fl) and illuminated by direct radiation (as in a CCD for example). There are pros and cons to using each of these along with attendant risks, and so a detailed risk analysis and cost-benefit analysis was undertaken leading up to the downselect. It is immediately obvious that the filled array is more susceptible to stray light and so great care must be taken to define the illumination from the telescope by a (very) cold stop. In turn, this brings the benefit of producing a top-hat illumination of the telescope by the detector. Further disadvantages of a filled array are that it has a lower speed for detecting known point sources, it has greater vulnerability to stray radio-frequency signals, 12 times more detectors are required for a given field-of-view, and the photon noise-equivalent-power for each detector must be lower by a factor of around two because of the reduced background power per pixel. On the other hand, the advantages are large: it produces a much higher speed for mapping; it provides instantaneous sampling of the image without jiggling of the telescope secondary mirror; it has a slightly narrower beam on the sky due to the illumination function.
The advantages were agreed to outweigh the disadvantages and as the risks were not dissimilar for both architectures the bare-fully-filled-array was selected. As noted above, as well as the implications of operate temperature and bump-bonding, the number of pixels needed is immense: 6,400 at 850 microns and 25,600 at 450 microns. The arrays will be constructed in sub-arrays and butted together (with no gaps) in the focal plane.
5.3 Optical arrangement
SCUBA-2 will be fed by the tertiary mirror of the JCMT, through a set of relay optics to the Nasmyth platform. In fact the optics for SCUBA-2 are far from trivial due to the aberrations in the focal plane of the telescope, the large fov and the fact that in the simplest mechanical and cryogenic arrangement, the optics on the antenna are optically sheared as the telescope elevation changes during observing To make calibration and re-gridding data from Nasmyth coordinates to RA/DEC easy, good image quality and low field distortion are required. Key aspects of the internal system of optics will be the need for high quality baffling and control of stray light as noted above.
The Conceptual Design Review was held in September 1999 and the outcome was presented to the JCMT Advisory Panel and Board in November, where it was enthusiastically received and endorsed as the highest priority scientific project for the JCMT. Funds were provided for the detector architecture selection phase, which was completed on May 4th 2000. In the meantime, the UK Particle Physics and Astronomy Research Council (PPARC) approved proof-of-concept funding for a prototype array. The JCMT Board will return to address the funding of SCUBA-2 at the forward look meeting in November.
The prime contractor for the construction of SCUBA-2 is the United Kingdom Astronomical Technology Centre (UKATC) at the Royal Observatory Edinburgh, with collaboration fromQueenMaryWestfieldCollege(London). The detectors will be provided under contract by the National Institute of Standards and Technology (NIST) atBoulder,Coloradoin collaboration with theUniversityofEdinburgh Micromachining Laboratory. The project scientists has just been appointed, Dr Wayne Holland (JCMT) and the project is now underway. Delivery of this exciting and world-beating instrument is expected to the JCMT in late 2005.
1. Holland,W.S. et al.. SCUBA: a common-user submillimetre camera operating
on the James Clerk Maxwell Telescope. MNRAS, 303 (1999) pp. 659-672.
2. Smail,I., Ivison,R.J. and Blain,A.W. A Deep Submillimeter Survey of Lensing
Clusters: A New Window on Galaxy Formation and Evolution. Ap.J. 490,
(1997) pp. L5-L8
3. Hughes,D.H. et al. High-redshift star formation in the Hubble Deep Field
revealed by a submillimetre wavelength survey. Nature 394 (1998) pp241-247.
4. Barger,A.J. et al. Submillimetre-wavelength detection of dusty star-forming
galaxies at high redshift. Nature 394 (1998) pp. 248-251.
5. Holland,W.S. et al. Submillimetre images of dusty debris around nearby stars.
Nature 392 (1998) pp. 788-791.
6. Greaves,J. et al. A Dust Ring around e Eridani: Analog to the Young Solar
System. Ap.J. 506 (1998) pp. L133-L138.
7. Johnston,D. and Bally,J. Submillimeter Wavelength Imaging of the Integral-
shaped Filament in Orion. Ap.J. 510 (1999) pp. L49-L54.
8. Pierce-Price,D. et al. A SCUBA submillimetre survey of the Galactic Centre
ApJ., (2000) submitted.
9. Chervenak,J.A. et al., Superconductor multiplexer for arrays of transition edge
sensors. Appl.Phys.Lett. 74 (1999), 26-
10. Griffin,M.J. et al. Spire-a bolometer instrument for FIRST. Proc. SPIE 3357
(T.Phillips, ed.) (1998) pp. 404-413.
11. Bock,J.J. et al. Silicon nitride micromesh bolometer arrays for SPIRE. Ibid.