|
|||
| Joint Astronomy Centre |
|
|
|
| Show document only |
Download alternative PDF file version of this document here
Authors:
In order to maximise the scientific return from the investment in the JCMT by the partner countries, state-of-the-art array instrumentation is required for the telescope. In the continuum bands JCMT has defined the state-of-the-art with the introduction of the SCUBA instrument. This is re-paying handsomely, in science terms, the considerable investment in its development and construction. There is now the opportunity to do the same for spectral line astronomy with an ambitious and well-managed program to deliver array receivers. Therefore, we propose to form a consortium of the JCMT receiver labs and the UKATC to deliver array receivers in the following order and bands:-
We expect the array receivers to increase the current imaging speed of the telescope at the two bands by an order of magnitude. Each array will consist of a single polarisation in a single cryostat. This approach lends itself to upgrades by adding additional cryostats containing another set of pixels in the orthogonal polarisation or an array at another waveband. Thus the optics will allow other bands to be added (e.g. C and E) and natural upgrade paths for B and D in terms of doubling the number of pixels. In addition, the array themselves will be built in a modular fashion, allowing mixer upgrades as the technology matures. Much of the array optics will be cooled to ~ 20 - 30 K to minimise the system temperature. The two-band system will be competed in a time frame of 5 years. This will necessarily involve the JCMT Receiver Laboratories and the ATC working closely together. This will maximise the co-operation between laboratories, maximise the number of common components in the receivers, and minimise the number of different components. The starts for the receivers for B and D band will begin at the same point to maximise the overlap. The approach should minimise cost by using common components and maximise speed. A strong management team will be appointed to deliver this ambitious programme on time and on budget. The scientific case for this project, together with a discussion of the main technical issues, is set forth in the rest of this document. The initial cost estimate for the two arrays with a 16 channels camera in a single polarisation at B band and 8 at D band is US $1.4 M dollars allowing for 2.5 %/annum inflation during the spend lifetime of the project. There is no contingency in this figure.
CONTENTS
The JCMT community has long recognised that provision of powerful heterodyne array receivers will greatly enhance the telescope’s scientific productivity. The spectral line correlator, ACSIS, which will form the backend for these receivers, has already been approved and its design and development are well under way. An immediate start on the frontends is now essential if they are to be delivered on a time-scale which matches that of ACSIS.
In constructing this programme we have taken account of a large number of factors, including: the scientific opportunities and the priorities of the users; the state of development of the technology; the resources expected to be available in the JCMT development fund; and the capabilities of the JCMT receiver development laboratories, taking particular note of the fact that the effort and expertise available is substantial but is distributed amongst a number of groups. It is clear that the programme needs to be ambitious if it is to put the JCMT in the same pre-eminent position in spectral-line observations as it now is in continuum work through SCUBA. Fortunately the work already done on developing key components, such as the mixers, and on issues like how to bring the beam out onto the Nasmyth platform, together with the solid background in all aspects of receiver engineering that has been built up during the development of the existing systems, put us in a strong position to proceed rapidly.
To achieve its goals HARP will require the co-ordinated effort of a number of groups. In particular the plan calls for:
The breakdown of the work packages is as follows:
MRAO B-band mixer development and construction B-band LO B- band cold optics B-band camera integration and test PLL’s for the array programme IF amplifiers for the array programme Electronics modules –including low-level microprocessor control -for the array programme D band camera cold optics design and production/procurement D band LO injection scheme design and production Integration of D band mixers into D band camera and cryostat
HIA Development and construction of D band mixers High-level array control microprocessor and software D band LO procurement LO automation for the array programme
ATC Program management Systems engineering Fore-optics Calibration unit Cryostat design and manufacture Array observing software & integration with OCS End to end testing of completed receivers
Other groups will be involved, mostly through subcontracted work:
CLRC D-band feed-horn manufacture
SRON Junction manufacture under the existing contract with JCMT
NUI Assistance with optics design within the boundaries of the existing contract with JCMT
The scientific need for focal-plane heterodyne array receivers at the JCMT is stronger than ever. With the exception of some molecular line surveys and deep spectroscopy of compact objects, nearly all the heterodyne programmes require mapping of extended sources. Since in the millimetre and submillimetre, most receivers are approaching background-limited performance, only by building multiple detectors can the mapping speed be increased. Without this increase in speed, observations of samples of objects of useful statistical size, and mapping of faint extended objects such as galaxies, will remain beyond the JCMT's capabilities. With arrays, one has the opportunity to increase the mapping speed essentially in proportion to the number of pixels, provided the performance of each detector is not degraded. Almost every millimetre and submillimetre observatory worldwide has developed, or is currently developing, focal-plane array technology. It is interesting to note that it is mainly of the older antennas on average sites that have led this work in an attempt to remain competitive: the FCRAO 14-m will soon have a 32 element array working at 100GHz, the NRAO 12-m has an 8-element array at 230GHz, and the Nobeyama 45-m is testing a 25-element 115-GHz system this winter. Without a similar programme at the JCMT, the telescope will lose the advantages of its large collecting area and superb site which make it the premier submillimetre observatory in the world.
The proposed array programme is highly complementary to the recent science made possible by SCUBA. The JCMT is now routinely used for wide-field continuum studies of molecular clouds and star-forming galaxies. Dust emission is an excellent tracer of column density and mass, and SCUBA is demonstrating that it is the instrument of choice for surveys of cold dust in both distant galaxies and the earliest embedded phases of star formation. However, it must be remembered that the richness of the submillimetre bands is due to the profusion of easily excited molecular lines it contains. SCUBA data provide no information on space densities, velocities, chemical structure or excitation in the gas. Without high-resolution spectroscopy at these wavelengths, we would have very little understanding of the internal structure of molecular clouds and star formation in our own and in external galaxies.
The provision of focal-plane heterodyne arrays that increase current mapping speeds by an order of magnitude or greater will be of enormous value to the JCMT scientific community. Such instruments will allow both wide-field surveys (tens of arcminutes in extent) and deep, narrow surveys (a few arcminutes radius) of regions of molecular gas in our Solar system, in our galaxy and beyond. Operating in the submillimetre range, such programmes will benefit from good angular resolution and be highly complementary to the lower-resolution and lower-excitation millimetre wave work being pursued elsewhere. Such wide-field surveys are also essential scientific preparatory work for the advent of the submillimetre interferometers: these instruments will over the coming decade push angular resolution at these wavelengths to 0.1 arcsec or better. The Smithsonian Submillimetre Array on Mauna Kea will be operating within a few years, perhaps with the JCMT linked in to double its collecting area. The ambitious MMA/LSA interferometer in Chile will have many times the collecting area of the JCMT, but will not be operational before 2007. However, these instruments will not remove the role of the single dishes such as JCMT. Rather, they are primarily designed for small-field mapping at high angular resolution, and although mosaicing will permit wider fields to be imaged, it remains much more efficient to do surveys in the line and continuum using dedicated single dishes with focal plane arrays. Effectively, each pixel in the focal plane acts as a separate antenna. The interesting objects discovered in the single dish surveys - be they shock fronts, outflows, protostars, high-redshift star-forming galaxies, or chemical anomalies - will then form the target lists for interferometric follow up.
In addition to this fundamental advantage in mapping speed on extended objects, arrays offer several secondary advantages. For objects smaller than the array field-of-view, edge pixels can be used to make "off" spectra simultaneously, obviating the need for beam or position switching, so even mapping of compact objects can be speeded up with arrays. Perhaps more importantly, image quality is likely to be greatly improved. Current large-scale single-pixel maps are often beset by serious artefacts, particularly due to calibration variations from point to point, and to pointing errors. An array can greatly reduce the effects of both of these problems: there are obviously no relative pointing errors between the pixels in the array, so provided overlapping submaps are made containing bright features, the pointing drifts between them can be removed (as is done in optical and infrared imaging). In addition, relative calibration of individual pixels is more accurate since data are taken simultaneously through the same atmospheric path.
All of the millimetre and submillimetre windows accessible from Mauna Kea provide a rich and important set of spectral lines for astrophysical study. In general terms, the higher the frequency, the greater densities and temperatures needed to excite the lines, although there are obvious exceptions. The Figure below shows the atmospheric transmission in three different weather bands, the five bands traditionally being named A (centred on 200GHz) through to E (centred on 860GHz).
The A band (230GHz) does have the advantage of its high transmission even in relatively poor Mauna Kea weather; an array at this wavelength would provide a useful workhorse for bad-weather backup. Nonetheless, the competition from the IRAM 30-m telescope at this frequency, which has almost twice the spatial resolution, and from the NRAO 12-m 8-beam system, suggests we should at least in the first instance exploit the unique submillimetre bands available on Mauna Kea which other sites do not have access to. We note that the JCMT was built on Mauna Kea specifically to address the objectives of submillimetre astronomy. It remains the largest submillimetre telescope in the world, so we also achieve the highest angular resolution possible at these wavelengths with a single dish.
Figure 1 Models of the zenith transmission on Mauna Kea, based on Cernicharo’s ATM program. Three different precipitable water vapour values are shown. Fine structure arising from ozone transitions is omitted from this version of the model, but does not grossly affect the transmission properties.
The B window, spanning the 325 to 375GHz range, contains transitions from nearly all the most abundant molecules in interstellar gas. It is also the highest-transmission submillimetre band available. The 345 GHz J=3-2 transition of CO is perhaps the most important, along with its isotopomeric lines near 330GHz. We would achieve about 14 arcsec resolution at this band, which is well matched to the beam of IRAM at the CO 2-1 line, and Nobeyama 45-m at the 1-0 line. It is also the perfect match to the SCUBA 850-micron beam, allowing highly complementary surveys of gas and dust in molecular clouds. The J=3 level of CO lies 33K above ground, and has a critical excitation density of order 104 per cc: these excitation conditions are not extreme, but do mean that this line, in contrast with the 1-0 line, is not excited throughout typical molecular cloud conditions, making it an excellent tracer of molecular gas in the vicinity of regions of star formation, shocks and PDRs. The band also contains important transitions of CS, HCO+, HCN, N2H+ and may other abundant molecules; it allows galaxies to be observed in CO 3-2 up to redshifts of 15,000 Km/s. For these reasons, and the detailed science case below, we identify the B band as the highest scientific priority for array development.
It is worth noting that there is a further atmospheric window just above the B-band, centred at about 405 GHz, which has relatively good transmission. By coincidence there are very few lines of the common molecular species in this window, so it remains relatively low priority. It is however possible that the coverage a an array receiver designed principally for the B-band could be extended into this region without great difficulty, perhaps just through the purchase of an additional local oscillator source.
The C band consists of several rather narrow windows between about 450 and 500 GHz. The Bonn group are nearing completion of a C band array for CSO, named CHAMP, offering 16 elements. One of the prime motivations here is study of the neutral carbon line at 490 GHz. Although this is a very important line, given the lead of the Bonn group it appears to us to be a mistake to try to improve on their efforts with currently available technology, and we suggest deferring C band development until the results from CHAMP are known and technology improves to the point where we can build a significantly faster instrument.
The D window, from roughly 620 to 720 GHz, has similar transmission properties to the C and E bands, but the excitation requirements of the gas are less than at E, and the beam efficiency significantly higher. It also offers greater continuous transmission than C band, which is broken up by numerous water lines. The CO 6-5 line (690 GHz) and its isotopomers (near 660 GHz) are the brightest lines in the band, lying about 115K above ground state. Initial line surveys have shown that the spectra of warm dense regions are extremely rich in this band. We will achieve 7-8 arcsec resolution over this band, which is well matched to the SCUBA beam at 450 micron, and at least as high as that achieved with any single dish in the world. The D-band is clearly the one which offers the best prospects for investigating regions which are qualitatively different from those with which we are already familiar from work at lower frequencies. We thus identify D-band as the second priority for array development.
At E band (roughly 800 to 900 GHz), the JCMT antenna efficiency is currently poor (though is likely to improve), and the excitation requirements of most of the lines are somewhat extreme; it does however include an important neutral carbon line.
As stated above, it is our view that the priority order of first B-band, second D-band emerges clearly from this discussion. We have however considered carefully the merits of choosing the A-band as the second priority. In addition to the advantage of being able to exploit the poorer weather conditions, this would be less challenging technically and might therefore be somewhat cheaper and be completed more quickly. We felt however that the competition from other telescopes at A-band, together with the fact that this choice would give the JCMT two arrays doing rather similar science, whereas the B-band and D-band are much more complementary, made it unattractive at least in the first instance.
Recent weather statistics for Mauna Kea have been compiled by Graeme Watt, and are shown in the Figure below. From this, it is clear that there is a large fraction of the time when B band array would be useable (CSO tau less than 0.12 or so), and that even at D band there is a significant amount of time which would be useable (i.e. when CSO tau less is than 0.07 or so).
Figure 2: Summary of recent atmospheric transmission data, as measured by the 225GHz optical depth measured by the CSO radiometer. Figure courtesy of Graeme Watt.
The scientific case for a B band array is extremely strong, and was detailed in the original proposal to build a B band array for the JCMT. Since then, the urgency for such an array has increased significantly, partly because of array development by the JCMT's competitors, and because of the wide-field images now being generated by SCUBA. Historically, the JCMT has done little "wide-field" astronomy, instead concentrating on spectral surveys and limited mapping projects, exploiting its unique submm capability. Although this approach has been scientifically productive, many of the well-known objects have now been studied by the JCMT in some detail. The ability to perform wide-field surveys in molecular lines, to map cloud kinematics, study chemical gradients, search for molecular outflows, and image interacting galaxies and the Galactic Centre, would transform the JCMT's capabilities, making it a superb submm survey instrument.
The B band is in many ways the prime frequency band for the JCMT. It contains many lines which provide useful probes of moderate density and temperature in molecular clouds, such as CO 3-2, CS 7-6, HCN 4-3, HCO+ 4-3, CH3OH 7-6. A short list of some of the most important lines appears below. These molecules typically trace gas temperatures of 10-100K and densities of 103 to 107 per cc. In very general terms, the spatial extent of the objects studied at B-band will be smaller than those at 3mm (CO 1-0) and 1.3mm (CO 2-1) allowing discrimination of the warmer more excited gas associated with star formation from the cooler, lower density ambient gas that surrounds it. The band also has by far the highest transmission of all the submm bands, and the telescope is very efficient at these wavelengths, with the error beam causing few problems for extended mapping. This is the lowest frequency band at which the JCMT is unique: it is on the best submm site (excepting perhaps the South Pole and Chajnantor) and has the highest angular resolution available at these wavelengths.
Unlike at B-band, very few maps of any extent currently exist at D-band: the low atmospheric transmission and high receiver temperatures in this window make mapping an extremely slow process with existing receivers. An array at these frequencies would allow mapping at these frequencies for the first time.
The combination of B- and D-band arrays provides enhanced scientific capabilities over that provided by a single frequency band. Expanding the frequency coverage to the D-band provides strong constraints that enable astrophysical modelling to better determine the physical parameters of the many types of astronomical objects which can be studied with the JCMT. A list of some of the important lines in the D band appears below, along with their upper energy level values measured in Kelvin. The CO(6-5) transition at 691 GHz (and its isotopomers) acts as a very useful thermometer for warm molecular gas, especially when compared with CO(3-2) at 345 GHz, because the energy of the J=6 level (116K) is substantially above that of the J=3 level (33K). Moreover, the very high critical densities of the high-J lines of such abundant molecular tracers as CO, HCN, CS, etc. enable their D-band transitions to be probes of much denser gas than their B-band lines can address. D-band mapping affords the highest angular resolution available for observing spectral lines of any single-dish submillimetre telescope (7 arcsec). For even higher spatial resolution with telescope arrays such as the SMA, the D-band array on the JCMT will be useful for filling in the low-order spacing; this is a crucial requirement if line ratio experiments and chemical abundance estimates are to be made from SMA data. .
The different critical densities of D- and B-band lines permit more accurate modelling of the physical conditions in cosmic sources - whether they are solar-system objects, molecular clouds, young stellar outflows, circumstellar envelopes, galactic sources, proto-galaxies, etc. Both density and temperature (and their gradients) can be determined better by combining mapping at D-band frequencies with data from B-band (and/or other lower frequencies). This information is important for estimating masses, determining gravitational stability, studying freeze-out of molecules on grains as functions of gas density and temperature, fractionation studies, isotopic abundance determinations, and astrochemistry studies (e.g., ion abundances due to collisional recombination effects). High-critical density lines in D-band, such as those of HCN(8-7) and CS(14-13), are especially important for studying higher density (up to ~ 108 cm-3) regions such as protostellar disks, molecular cloud cores, planetary atmospheres, cometary comae, shocked compressions, cosmic masers, galactic nuclei, outflow boundaries, and the inner parts of circumstellar shells.
Molecular clouds are the sites of star formation, and understanding their structure and evolution, both in the the Solar neighbourhood, the Galactic Centre, and in external galaxies, remains one of the outstanding problems of astrophysics. In the past few years, a great deal of work has been done on large-scale studies of molecular clouds. The FCRAO galactic plane survey of the CO 1-0 line at 50 arcsec resolution is the largest such survey at sub-arcminute resolution; at Nobeyama, 15-arcsec surveys (although at 34 arcsec sampling) of the Galactic Centre and Orion molecular clouds have been recently completed. These large-scale studies have been important in establishing the self-similar structure of clouds over many angular scales, and in locating regions of heating and increased linewidth indicative of shocks and star formation. HARP can make a significant contribution in this area because it has the angular resolution sufficient to resolve the interesting dense cores where stars form, and because the lines it observes typically trace gas which is being disturbed by shocks or gravitational collapse. The FCRAO beam is too crude to resolve dense core structures, and the lines at 3mm are too easily excited to pick out the regions of star formation.
Figure 3 The 850-micron emission mapped with SCUBA from a dark molecular cloud in the Ophiuchus complex is shown in yellow contours, superposed on the digitised sky survey data; this is a quiescent cloud showing no signs of star formation. The B-band HARP array footprint is shown top left. (Data from Visser, Richer and Chandler, 1999).
The HARP arrays will permit rapid surveys of these objects in molecular lines at a resolution equal to that of SCUBA images. Together, these data sets will from a powerful set of diagnostics of the internal state of molecular clouds, and the star formation occurring within them. Due to their low opacities, the isotopic lines of CO, in particular 13CO and C18O, provide excellent tracers of gas column density, while at the same time providing linewidth and velocity measurements which can be used to assess whether clouds are gravitationally bound and/or rotating; such information is essential if the SCUBA observations are to be fully understood. For example, Figure 3 shows a typical dark molecular cloud mapped with SCUBA, in this case showing no evidence of star formation within it; it is 10 arcminutes across, so that obtaining line maps of this object to estimate its virial mass and gas density for comparison with the dust data would be an extremely slow process with a single pixel receiver. The important density tracing lines such as CS, HCO+, and HCN which have moderate opacities allow the detailed structure of cloud cores to be investigated. In particular, it is now known that many such objects appear to be in a state of slow gravitational contraction over very large angular scales. Measurements of the velocity field of the gas, and searches for line profiles characteristic of infall, will be possible with the proposed arrays in a large sample of cores
The main CO line is an excellent tracer of outflowing gas, and the B-band array will for the first time allow surveys of useful statistical size of protostellar outflows, allowing us to study their structure on large angular scales at high angular resolution, and to measure basic physical properties such as their momentum and energy flows. Figure 4 shows a typical bipolar outflow system: even for this young, low-mass object, the flow extends over half a degree in size, demonstrating the need for wide field outflow mapping to determine their physical properties and see how they evolve in time. Outflows may play a key role in injecting turbulent energy into molecular clouds, supporting them against collapse, but few quantitative results exist to support this hypothesis because of the large amounts of observing time currently needed to map large areas of molecular cloud.
We believe that molecular outflows are driven by neutral jets and winds, through J- or C-type shocks. The heating during the shock elevates gas temperatures to several hundred Kelvin. We now know from the ISO mission that typically one per cent of the gas associated with Herbig-Haro jets and outflows is hotter than a hundred Kelvin, but the spatial resolution is insufficient to map the hot gas and so constrain models out outflow acceleration. The very high angular resolution at D-band and the responsiveness of the higher-frequency lines to the presence of warm gas makes this work much more feasible. At the moment, this is extremely difficult with RxW due to its very slow mapping speeds. We already know that extended CO(6-5) emission is seen even in low mass outflows, such as L1527 and TMR-1: both UV heating by photons originating from the inner part of the accretion disks and shock heating where stellar winds or atomic jets collide with dense clumps in the ambient medium appear to be important heating mechanisms, but the details remain unclear due to paucity of good data at these wavelengths.
Figure 4 The CO J =2-1 outflow from the protostar in RNO43, mapped with the JCMT over a period of many nights by Bence, Richer and Padman. Note the very large extent (3 parsec in all) of this flow, even from a relatively young "Class 0" protostar. The B band array field of view is superposed middle left.
Molecular clouds are known to be chemically inhomogeneous. Different rates of gas-phase and dust-surface chemistry at different densities and temperatures, plus the effect of grain-mantle growth causing molecular depletion at high densities and low temperatures, produce strong abundance variations within clouds. Outflow shocks also cause chemical processing of molecular material. The enhanced abundances of methanol and SiO seen in outflows is indicative of shock liberation of molecules from grains. There are two important implications of these recent discoveries. First, they challenge our steady-state chemical models of molecular clouds, forcing us to adopt time-dependent models. Second, from an astrophysical perspective, we must understand the effects of grain depletion and abundance variations if we are to use line emission as a useful estimator of cloud physical parameters. The key to making progress in these areas is to map several molecular lines and the continuum (with SCUBA) in a large sample of different molecular cloud environments. Studies to date have been restricted to a few well known objects. The B- and D-band arrays will allow such studies to be put on a firmer footing by enabling large samples of objects to be observed, and `typical' chemical conditions to be established for various classes of objects. In addition, these studies can be done at high angular resolution – 14 and 8 arcseconds for the B and D band observations. This is a significant increase on previous chemical mapping surveys which have typically had 1 arcminute resolution or worse.
The expanded frequency coverage that is afforded by a D-band array permits observations of multiple transitions of certain molecules, such as CO, CS, HCN and N2H+, thereby permitting more accurate abundance determinations. Chemistry is important not only as a diagnostic tool, but also as a moderator of astrophysical processes. For example, the abundance of deuterated formaldehyde (HDCO) and deuterated ammonia (NH2D) in hot cores is dependent upon the early history of such regions when the gas in these cores was at much lower temperatures and isotopic fractionation occurred. Chemistry also moderates astrophysical processes; the relative abundance of ions such as N2H+ and HCO+ determines whether shocks are J-shocks (strongly dissociative) or C-shocks (weakly dissociative). Also, the cooling rates of pre-protostellar clouds depend upon chemical composition because molecules are the principal coolants in these clouds
CSEs, which are the expanding envelopes of evolved stars, are ejected in both the AGB and post-AGB phases of stellar evolution. They are interesting in their own right and also important as laboratories in which to study physical and chemical molecular processes in a relatively simple geometrical and kinematic context, and as one of the main mechanisms for reprocessing material and returning it back to the ISM. A few CSEs (like W Hya) are tens of arcmin in angular diameter, but 1 arcmin is a more typical angular size. The B- and D-band arrays will be very useful in their study. In the envelope of IRC+10216 molecular species such as C2H and CS form extended shells of various angular sizes from 20 to 45 arcsec. (Other molecular species are present in filled spheres around the central star.) In the case of CRL 2688 the molecular structure appears to be bipolar. Chemical gradients of relatively abundant species, such as the anti-correlation of HCN and CN in the CSE of IRC+10216, could be measured much more easily with HARP than with single-beam receivers. Moreover, the outer parts of these extended envelopes are expected to contain molecular ions such as HCO+, HCS+, etc., due to photochemical processing by interstellar UV photons, but these lines are expected to be weak. The B-band array will make it more practical to search for and perform chemical studies of ions and other rare molecular species in CSEs where results may depend on where in the CSE one happens to observe. Observations with the D-band array will enable much better modelling of density profiles, especially in the warmer inner parts of these envelopes where the molecular densities are higher, the need for better angular resolution is greater, and the models are more easily discriminated from each other.
Since gas is heated to high temperatures in regions of high UV flux, it is the high-J lines of molecules, particularly CO, that best trace PDR regions and constrain heating models. Much of the observational work in this area was pioneered with "Receiver G" on the JCMT several years ago, and work with single-detector receivers continues at the KOSMA 3-m, JCMT and CSO telescopes. It is clear from this work that large amounts of spatially extended, hot molecular gas (temperatures of 100 to 200K or so) exist in these regions, with photoelectrons from dust and shocks dominating the heating processes. Although models of these important transition zones are now relatively sophisticated, few high angular resolution data sets in multiple transitions exist to constrain these models. Given the large ranges of temperatures and densities in these regions, a number of lines which probe different physical conditions is required. The combination of B- and D-band arrays would allow rapid mapping of these molecular cloud interfaces at very high resolution and in moderate-sized samples of sources.
The Galactic Centre is a unique laboratory: its molecular clouds are extended and appear to be permeated by strong magnetic fields, and experience strong rotational shear due to the large central mass concentration. Studies of these clouds, and comparisons with solar neighbourhood clouds, are crucial if we are to understand the workings of nuclei in external galaxies. The evidence for a central black hole of a few million solar masses is now compelling, and SCUBA surveys are underway to map the dust emission and detect polarised flux which indicates the magnetic field geometry. However, line emission maps are of enormous value, as the velocity information is critical in understanding the cloud dynamics in this region. An unbiased, fully-sampled survey of the CO 3-2 line in the entire Central Molecular Zone (approximately 3 degrees by 0.5 degrees in size) would be feasible with the proposed B-band array, and be the ideal follow up to the current large scale survey being performed with SCUBA. At present, such a project would require extremely large amounts of telescope time. It would also be possible to do more limited surveys of parts of the Galactic Centre in optically-thin isotopomer lines of CO and HCN to determine physical conditions in the clouds, and to map the SiO emission produced in the shocks which are perhaps associated with cloud-cloud collisions. Since the cloud temperatures and densities in the galactic centre are on average higher than in the solar neighbourhood, mapping with a D-band array of the most energetic regions would also be extremely valuable to constrain the properties of the densest and hottest gas. These studies are important, not only to understand our own Galactic Centre, but also as a laboratory in which to study material in galactic nuclei, since much extragalactic work carried out so far has involved studying only the nuclei of galaxies (where the emission is brightest).
A heterodyne array will have a dramatic impact on studies of the molecular gas in external galaxies. The B-band array will make it practical for the first time to image large, nearby galaxies with the JCMT. For galaxies such as M51 within about 10 Mpc it will be possible to separate arm and interarm regions and, with the help of the D-band array, to study the effect of spiral arms on the molecular gas (heating due to star formation, increased density due to shocks, etc.). It is likely that observations of the CO(6-5) transition will be required before significant differences can be detected between star-forming and non-star-forming molecular gas in galaxies. (The differences are perhaps starting to show in the CO(3-2) transition, but the changes are very subtle.) So, for example, it would be interesting to compare the arm and inter-arm regions of such classic sources as M51 in some of the high rotational transitions of CO to see if we can see the effects of heating associated with star formation. On the other hand, Local Group galaxies (d < 1 Mpc) are sufficiently close that the JCMT can study the properties of individual giant molecular clouds. In particular, dual-array observations of several transitions and isotopes of CO can constrain both the temperature and the density of these clouds.
There has also been much interest recently in edge-on spiral galaxies, with claims both for and against substantial CO emission high above the plane. Since the lack of short-spacing information may affect the interferometric data while pointing and calibration errors are likely to dominate maps made one point at a time, a high-resolution array like the B-band array would be suited to studying the distribution of molecular gas at high galactic latitudes in edge-on galaxies. CO has been difficult to detect in elliptical galaxies. In some of the cases where it is detected, however, it has been found to be extended (cf. NGC 1275) or distributed asymmetrically relative to the galaxy centre (cf. NGC 185).
By imaging a wider field of view, we will be able to obtain more sensitive as well as more spatially complete measurements of the molecular gas content of early-type galaxies ranging from cooling-flow galaxies to Local Group dwarf spheroidal galaxies. All the galaxies observed in the U. Mass. survey at 115 GHz (roughly 350 in number) are extended enough that observations with the B-band array will result in a substantial increase in mapping efficiency. The B-band array, possibly augmented by the D-band array if the gas is warm enough, will permit detailed observations of a statistically significant sample of galaxies in the isotopomers of CO and high density tracers such as HCO+ and HCN, which can be used to study the molecular gas content, excitation temperature, and density distribution both as a function of galaxy type and as a function of radius within the galaxy. For example, the hypothesis that molecular gas in the nuclei of galaxies with large bulges is significantly denser on average due to the higher pressures in these regions could be tested much more effectively with a larger sample.
Nearby starburst galaxies are important, both because of the unusual physical conditions present in these very active objects and because their cousins, the ultra-luminous galaxies such as IRASF 10214, can be seen out to very large distances. Many starburst galaxies (e.g., M82 and NGC 253) are close enough that they can be profitably mapped with a B-band array. A B-band array would make it feasible to obtain detailed maps of the CO J=3-2 emission in starburst galaxies, and a D-band array would enable the warmer gas of the starburst to be found, mapped, and diagnosed in clouds as dense as 108cm-3. It would also be useful for the study of the molecular gas in Seyfert galaxies, which would help us understand the relationship (if any) between the starburst and the AGN phenomena. Observations of the CO(6-5) line at 691 GHz are particularly important because highly excited CO can be an important coolant for the nuclear molecular gas in Seyferts and AGNs, comparable to the fine-structure lines of [OI] at 63um and [CII] at 158um. The CO(6-5) transition, with a critical density of ~ 106 cm-3 for optically thin lines, requires collisional excitation by gas with kinetic temperatures of order 116K. It is therefore a useful thermometer for the warm, dense gas in the nuclei of active galaxies. It is also important to know if CO is warm in these objects because molecular mass obtained from the integrated CO(1-0) intensity - H2 column density relation is a function of the CO excitation temperature that is assumed.
In general high-redshift sources are extremely weak and not extended to the JCMT beam. Single-dish detections of molecular lines from these objects have always proved extremely difficult in the past, and this likely to remain true. The proposed arrays will however offer a significant increase in sensitivity per pixel: the ability to perform on-array chopping, plus the expected increase in baseline stability and bandwidth available over current receivers, should allow high redshift line detection experiments to be carried out much faster than with current instruments. The obvious observing programmes to be attempted with such an increased sensitivity are the search for CO and CII in quasars, damped Ly-alpha systems and clusters (or at least possible proto-clusters) of galaxies at reasonable redshifts. D-band would give coverage of z ~ 1.7-2.1 for the CII line..
The main impact of B- and D-band arrays on Solar System studies would be on cometary research. The study of comets provides an important link in the chain that leads from proto-planetary disks around YSOs to planet formation in our Solar System (and elsewhere). Comets may even have preserved interstellar matter from the cloud in which the Solar System formed. Hence the study of comets may be capable of providing clues to the physical processes by which interstellar cloud material condensed and accreted to form the Solar Nebula and its planets some 4.5 Gyr ago.
Well-known difficulties in studying large gaseous comae are their time variability and generally weak molecular lines at radio frequencies. Also, cometary ephemerides are always subject to uncertainties due to non-gravitational effects. Positional uncertainties can be worse for newly discovered comets which are often rich sources of parent molecules and interesting radiation-processed ions and radicals. The B- and D-band arrays will undoubtedly aid the study of molecules in comets in a major way since pointing will not be a critical impediment. In an active comet the inner coma is opaque to sunlight, and most of the interesting photo-chemistry occurs at some distance from the nucleus. The angular size of the photochemically active zone is quite variable, depending upon the distance of the comet from the Earth, the rate of outgassing and the gas-to-dust ratio. Because radio lines of cometary molecules are generally weak and variable, observers have been forced to concentrate on the nucleus. Therefore, little is known observationally about the chemical evolution of the gas farther out in the inner coma. An array would allow us to study the spatial, temporal and chemical evolution of the gas in the inner comae of some of the closer, more active comets. In particular, near the nucleus we expect to find parent molecules such as CO, HCN, and CH3OH; farther out we might find CO, CS, and possibly HCO+. HARP would permit rapid mapping and access to more molecular species per observing run, allowing better models of the evolution of comets as they pass through the inner Solar System.
Recently, high-frequency radio observations of comets have begun to make progress in determining the chemical, isomeric (as in HCN and HNC), and isotopic composition of the frozen volatiles that form cometary nuclei. JCMT observations of the unexpected, bright comets C/1996 B2 (Hyakutake) and C/1995 O1 (Hale-Bopp) have been particularly fruitful. Cometary comae are typically warm, with Tk ranging up to 150K. CO and HCN are both prime molecular tracers at submillimetre wavelengths, with line temperatures of order 10K for comets of the ilk of Hyakutake and Hale-Bopp. Hence CO(6-5), HCN(8-7), HNC(7-6) and their more abundant isotopomers in D-band are expected to yield strong lines. Radiative-transfer modelling of their molecular gas will be not only more accurate with the inclusion of higher-J lines but also more realistic if non-spherical models can take account of simultaneous or quasi-simultaneous maps of the cometary emission at both B- and D-bands. Angular sizes of the comae of bright comets are typically of order 1 arcmin at closest approach to Earth, very suitable for the arrays we propose. The factor-of-two increase in resolution that a D-band array would provide is particularly important to obtain a better understanding of the dynamics and chemistry of the inner coma, where the activity fuelled by sublimation of nuclear ice is most intense.
The scientific requirements determine acceptable designs for the proposed arrays. The arrays must offer:
These scientific demands result in the following requirements for the arrays:
Number of pixels: assuming performance no worse than current receivers, 16 elements at B band offers the chance to achieve a factor of 10-20 increase in mapping speed. An upgrade path to 32 elements should be built into the system design in case extra funds become available. At D band, this increase in mapping speed might be achieved by 8 pixels only if the noise per pixel can be reduced significantly compared to current performance, which is well behind the best submm mixers in the world, although an upgrade path to 16 pixels would certainly be of great value if funds and technical limitations permit.
Array configuration: The pixels should be packed as closely as possible consistent with excellent performance and ease of use. A fairly conservative choice would be a pixel spacing of 5 times the Nyquist sampling or 30 arc seconds at B-band and 15 arc seconds at D-band. The small packing gain from hexagonal versus square close packing is marginal given the extra complexity of observing modes and horn and optical design, so square packing is recommended. For very extended objects, with both dimensions much greater than the array field of view, the actual location of the pixels is unimportant. We expect such objects to be mapped in the on the fly mode, and the antenna and/or secondary mirror movements can be used to guarantee full sampling of the image. For 16 elements, a 4 by 4 array configuration appears the simplest approach, and minimises the longest array dimension. For D band, a 3 by 3 array would fulfil the same objective, but since the possibility of 16 pixels being built exists, a 4 by 2 arrangement would make more sense.
Beam rotation: for objects close to or smaller than the array field of view, beam rotation can in some cases optimise the observing efficiency by making sure that no edge pixels miss the source emission. An obvious example is the mapping of an edge-on galaxy using an elongated 4 by 2 array. Beam rotation also simplifies the observing strategy and data reduction for any object, allowing the array axis to be aligned with one of the source axes. Even when performing on-the-fly maps there are only certain orientations of the scan direction with respect to the array which provide good sampling, so that if the region to be mapped is elongated in one direction it will still be advantageous to have a beam rotator. Only for objects which are roughly square or round would there be no advantage in rotating the field. It would nevertheless always be possible to make good maps without a beam rotator providing the antenna is agile enough and the reduction software can efficiently resample the data onto a suitable co-ordinate grid. We therefore consider the provision of a beam rotator desirable but not absolutely essential.
Sensitivity: since the mapping speed varies as the inverse square of the system temperature, it is only worth building an array of a modest number of pixels if the performance of each pixel can be guaranteed to be excellent. The aim must be to achieve the best practicable system temperature per pixel, certainly as low as the current JCMT receivers, and preferably lower.
SSB filtering: even on Mauna Kea, the sky emission in both signal and image sidebands is a significant part of the system temperature. For a completely background limited detector, the termination of the image sideband at zero Kelvin will halve the system temperature compared to a DSB system, making it equivalent to 4 non-SSB pixels. It is thus clear that effective SSB filters which terminate the image sideband at a very low temperature - 15 K or so - will be essential in fulfilling the sensitivity requirements of the arrays. SSB filtering also makes more accurate calibration possible. The ability to tune to DSB for simultaneous observations of lines in opposite sidebands would also be useful in a small number of cases but this is not a primary requirement.
Frequency coverage: as discussed above, B and D bands offer an excellent choice of bands for heterodyne science at the JCMT. We propose that the B band system should cover the full 325 to 375 GHz window, and the D band system cover the 625-710 GHz range. The possibility of extending the range of the B-band array to cover the window from 390 to 420 GHz should be kept in mind.
IF bandwidth: the maximum IF bandwidth possible from the mixers should be available, to allow mapping of very wide spectral lines in galaxies, and for maximum efficiency in spectral line surveys. A target of 2 GHz would give a velocity coverage of 1700km/s at 350GHz, and 920km/s at 650GHz; this would suffice for most programmes, especially at B band. At 650GHz a 50% increase in this quantity would be useful if it can be achieved without major additional cost and complexity.
Calibration: image quality for large-scale images requires both excellent pixel to pixel calibration, and accurate absolute calibration for combining maps from several observing sessions. A scientific target of better than 5% absolute calibration would be sufficient for most programmes, but we should aim for 2% or better.
Baseline stability: Wide bandwidths are only useable for galaxy mapping if the baseline quality is good, with low ripple and slope. The design of the arrays must aim for the best possible baseline quality, in particular by minimising standing wave ripple.
Simultaneous dual-frequency operation: the ability to observe simultaneously at two frequencies - say B and D bands - would make several observing programmes more efficient. However, unlike SCUBA which in good weather is almost as sensitive at 450 micron as 850 micron for optically thin dust emission, the required integration times at D band are likely to be much longer than at B unless the B array is tuned to a fainter, more optically thin line. Consequently, we believe possibility of dual-frequency operation should be investigated and allowed for, but only if the extra complexity and the efficiency losses (due to the dichroic splitter and extra mirrors) are acceptably small.
Polarisation: few lines are polarised at anything more than the one per cent level, and specialised techniques are required to detect this polarised signal. The survey-type projects we envisage HARP undertaking will not require polarisation information, so a single polarisation for each of the arrays will not compromise the scientific returns. If expanded arrays of more pixels were to be built subsequently, they could choose the orthogonal polarisation to enhance the packing of the arrays.
Remote tuning and monitoring capability: it is clear that for efficient operation of a system of this complexity a high level of automation is required. It is expected that it will be possible to set up and optimise the receiver remotely and quickly and that routine monitoring and fault-diagnosis will be possible over the network.
The most important consideration arising from the above specifications is the field of view. The clear diameter required for a 4 by 4 configuration with a 30 arcsec pixel spacing is 170 arcsec. If we wish to allow for a future upgrade to 32 elements at B-band (e.g. in a 6 by 6 configuration with the corner elements missing) or for a 4 by 4 element array at A-band, we should however aim for 210 arcsec. (This assumes a slightly closer packing relative to the beamwidth at A-band than has been assumed at B-band.) We have therefore adopted this as the goal for the optical design. Note that at 3.5 arcmin this is substantially larger than the SCUBA field of view.
The second relates to the need for termination of the image sideband at a very low effective temperature. If the sideband separation is achieved by quasi-optical means, this requires that these optical components are included in the Dewar because otherwise there would be excessive losses and emission in passing out through a window, through a warm SSB filter and back in through another window. The alternative approach of building single-sideband mixers would simplify the optics but is as yet unproven, especially as regards noise performance and rejection ratio. We have therefore adopted the use of cooled optics as the baseline design but propose that the investigation of SSB mixers be pursued as a development programme.
There are a number of key decisions about the nature of the instrument that define its architecture. These include where the instrument is to be located, how the detector arrays and optics are to be cooled and how the signals going to the different arrays are to be divided up. The options are to locate the arrays in the receiver cabin or on the right-hand Nasmyth platform. The Nasmyth offers several important advantages: there is a lot more space there and access is easier; the instruments do not have to tip; the IF cabling to the ACSIS correlator is simpler and will not flex as the telescope moves; the existing receiver would not have to be moved. There is however one serious problem: a way has to be found of bringing the beam out to the platform. This is considerably more difficult than for SCUBA because the path is constricted by the elevation encoder, which has a clear diameter of only 190 mm, and because of the larger field of view required. A number of possibilities were investigated, including bringing the beam across in front of the primary and then through a hole in the surface and finding paths around the encoder through the backing structure, but these all posed difficult problems. Fortunately a scheme has now been found which provides an optical relay from the Cassegrain focus to a convenient location on the Nasmyth platform with the desired field of view and satisfactory imaging performance. This enables us to adopt the Nasmyth platform as the location for the array receivers.
The system has to accommodate at least two arrays of detectors operating at different frequencies and if possible allow for more to be added in the future. One possibility is to build one very large Dewar with a great deal of cooling capacity which could accommodate up to say 4 arrays. To separate the signals going to the different arrays one could use switching mirrors, polarising grids or dichroics, or some combination of these, inside the Dewar. Although this approach might save on some costly items, since only one Dewar and fridge would be needed, it would make the development very difficult and probably compromise performance. For example the windows and IR filters would have to have low loss over the whole submillimetre waveband and for both polarisations. The cooling power required would also have been very large. We have therefore adopted the alternative approach in which there is a separate Dewar for each frequency band and only one polarisation is used in each Dewar. The preliminary layout allows for up to four Dewars to be installed on the Nasmyth platform. This makes the design considerably simpler and makes it practical for work to be carried out on one receiver without interfering with the others. The whole construction and test programme also becomes much more manageable in this scheme. In order to keep down costs and avoid duplicate effort it will however be essential to keep the design of the Dewars identical in all respects except where the differing technical requirements force them to vary.
A choice has then to be made as to how much of the system should be common to all wavebands and how much should be specific. Clearly there are savings to be made if as much as possible of the system is in common. We therefore plan to bring the signal through common relay optics (which will incorporate the beam rotator if one is included) and calibration unit and then to split the beams just before they enter the Dewar. For the initial configuration with one array at B-band and one at D-band a polarising grid can be used to split the beams providing simultaneous operation when required.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
