ON September 28, the Polar Satellite Launch Vehicle (PSLV), the workhorse launch vehicle of the Indian Space Research Organisation (ISRO), will deliver, in its PSLV-XL configuration, a dedicated astronomy satellite, Astrosat, weighing about 1.5 tonne, in a 650-km-high near- equatorial orbit with a 6° inclination. Nearly 20 years in the making from the day the idea of such a satellite was put forward (see box), and about 15 years since the idea was given a concrete shape, the final realisation of what promises to be a true astronomical observatory in the sky may appear to have been unduly delayed, but Astrosat will still be unique in its concept and is expected to make a significant and niche contribution to the important field of X-ray astronomy and the study of the X-ray universe.
Astrosat is a truly multi-institutional project, including collaborations with foreign institutions and agencies. The institutions involved in the mission include the Tata Institute of Fundamental Research (TIFR), Mumbai; the Indian Institute of Astrophysics (IIA), Bengaluru; the Raman Research Institute (RRI), Bengaluru; the Inter-University Centre for Astronomy and Astrophysics (IUCAA), Pune; the ISRO Satellite Centre (ISAC); the Laboratory for Electro-Optics Systems (LEOS), Bengaluru; the Vikram Sarabhai Space Centre (VSSC); the ISRO Inertial Systems Unit (IISU), Thiruvananthapuram; the Canadian Space Agency (CSA); and the University of Leicester, the United Kingdom.
In the mid-1990s it was conceived as just a broadband X-ray astronomy satellite, covering both soft and hard X-ray regions, which itself would have made it state-of-the-art given the nature of other contemporary X-ray satellites. But its final design has evolved to extend the satellite’s reach to higher wavelengths by including the near ultraviolet (NUV), the far ultraviolet (FUV) and the visible as well (Figure 1). And that is where its uniqueness lies, which renders it a potential world-class observatory. More importantly, the performance characteristics of the slew of instruments and detectors that the satellite carries—in terms of angular, energy and timing resolutions—are competitive with, and in some respects better than, not just the currently operating X-ray satellites but some of the planned ones as well. Such a multi-wavelength platform would enable simultaneous observations on a given celestial object and the associated astrophysical phenomena across different wavelengths to gain a total perspective of the dynamics involved.
Astrosat cannot definitely match the excellent performance characteristics of the big (11-12 tonne class) X-ray observatories like NASA’s Chandra X-Ray Observatory (CXO) or the European Space Agency’s XMM-Newton in terms of angular and energy resolution. However, most of the other X-ray satellite missions, except XMM-Newton and Swift, have limited wavelength coverage. While Swift is mainly a gamma-ray mission with an X-ray telescope as well, Astrosat’s uniqueness comes from its simultaneous observations over a broad wavelength band, very high resolution UV observations, and high resolution timing studies by one of its instruments which none of these observatories has. And there is no other planned mission in the near future that will cover the entire X-ray spectral band from 0.3 kiloelectronvolt (keV) to 100 keV and UV bands from 130 nm to 300 nm.
“In a modest way, in the 1.5 tonne class of X-ray satellites, Astrosat is unique,” said Suryanarayana Sarma of ISAC, the current project director of Astrosat. “To correlate observations from the existing satellites over broad wavelength regions to get a complete picture has been a real challenge. It requires coordination of data taken with different instruments at different places and at different times. It is like the elephant story. With Astrosat’s multi-wavelength capabilities, we are looking at the elephant more comprehensively,” he said.
“Right at the initial stage of conceptualisation, it was decided that if at all we are doing something [in X-ray astronomy] we must do multi-wavelength,” said Koteswara Rao, former project director of Astrosat. “All instruments and their specifications were defined with this sole objective” (Figure 2).
Astrosat carries five experimental payloads: (i) three Large Area X-Ray Proportional Counters (LAXPCs) for X-ray timing studies, which, together, have the largest area proportional counters ever, and with an unprecedented 10 microsecond timing resolution and high photon counting rate; (ii) the Soft X-Ray Telescope (SXT), which has imaging capability from 10 keV right down to 0.3 keV; (iii) two Ultraviolet Imaging Telescopes (UVITs), one for visible and NUV and the other for FUV, with very high angular resolution of 1.8 arc-seconds in UV; (iv) hard X-ray new technology imaging detector called Cadmium-Zinc-Telluride Imager (CZTI); and, (v) the Scanning Sky Monitor (SSM) to monitor and detect bright objects, particularly transients, up to 10 keV X-ray energy.
The most important aspect of the satellite from an observational perspective is that they are all co-aligned. That is, the Fields of View (FOVs) of all the instruments will be looking at the same object within a narrow specified window. For instance, the three LAXPCs themselves will have to be aligned within 5 arc-sec, pointed out Suryanarayana Sarma. The pointing accuracy of the satellite is 0.03 degrees in all the axes and the pointing stability is 5x10 deg/sec. This is how the simultaneous observation over multiple wavelengths from a common platform is achieved. With the suite of the above instruments performing in a coordinated manner, Astrosat will be able to address a host of outstanding questions in X-ray and UV astronomy.
Complex thermal design “Thermally, this satellite is very challenging,” said Suryanarayana Sarma. “Because you have to point to one source and then move on to another and then to another, the constant reorientation of the satellite keeps altering the thermal condition of the satellite due to the changing solar radiation and earth’s albedo, depending on the pointing direction. So, the thermal design of the satellite becomes complex. Initially, we had thought of adding a multilayer insulation, an optical solar reflector, etc. for solving the problem. But, finally, when we tested it in thermovac, we realised that these were not needed and we saved about 30-40 kg in thermal material for the main satellite frame. The payloads themselves, which account for 780 kg, were able to meet the mass specifications. This resulted in the overall satellite mass reduction from 1,560 kg to 1,515 kg,” he said.
Notably, the PSLV, for which a polar launch is the norm, is being used to deliver Astrosat in an equatorial orbit. For astronomical observations, an equatorial orbit is ideal because sources in both the northern and the southern sky can be observed. Actually, according to Koteswara Rao, when the configuration was being optimised, the committee had contemplated two options: a Geostationary Satellite Launch Vehicle (GSLV) launch or a PSLV launch. But, given the original estimated satellite mass of about 1,560 kg, it would have been too small for a GSLV launch.
The orbit has been defined on the basis of the following considerations: The satellite should have a reasonably long life of at least five years without requiring orbit manoeuvres. The higher the orbit, the more its lifetime. The solar cells in the panel should not deteriorate over the lifetime. “Above 500 km altitude is better because there will not be any drag on the satellite,” pointed out Suryanarayana Sarma. “In this mission, I do not see anything that limits the life actually. It should be able to survive even for 10 years,” he said. “With the originally estimated mass, we were getting a 600-km orbit with about 8° inclination. Now, with the reduced satellite mass, we are comfortable with 650-km orbit and 6 inclination,” pointed out Koteswara Rao.
With an equatorial low-earth orbit (LEO), one is, however, faced with the problem of the South Atlantic Anomaly (SAA). The SAA refers to an area near the equator region above South America where the inner surface of the van Allen radiation belt—the doughnut shaped region around the earth which traps highly energetic charged particles from the solar wind—dips down to an altitude of 200-800 km as against 1,200-1,300 km in the north. These high-energy charged particles (energies greater than 10 MeV) can affect the on-board electronic system and cause glitches in the astronomical data. That is why an inclined orbit at 6° was chosen. At this angle the satellite path only tangentially grazes the SAA without actually entering it (Figure 3). Also, at less than 6° inclination, the elevation will not be appropriate for the satellite visibility from the ground support station at Bangalore. At least a clear 3-4 minutes of the pass of the satellite during every orbit is needed to download the data stored on-board, pointed out Suryanarayana Sarma.
“The delay [of about 6 years] is essentially because all the instruments are complex, some of which are state-of-the-art, and have been technologically challenging,” said Suryanarayana Sarma. “Of the five instruments, we did not have any previous experience with four. We have built them from scratch. We had estimated that we may be able to complete it in four to five years, but we really could not do it because the concept had to be first converted into a model, then into an engineering model, its performance evaluated and then its design fine-tuned and finally a space-worthy instrument built which had to go through all kinds of rigorous tests. A couple of them failed at different stages, which we had to redo, leading to delays,” he said. For instance, an imported component of the detector assembly in the UVIT failed in the vibration tests as late as January 2014 and nearly a year and a half was lost in rectifying it. Even with the proven LAXPC, problems with its hardware surfaced in 2012 when the instrument was being tested.
“Also, we had very small teams of skilled people. Such missions abroad have much larger teams and they all would have had some prior experience in similar missions. Even at the TIFR, where three of the five instruments were built, each team had only three to four engineers. Smaller teams, building from scratch, with much of the learning going on in parallel with building the instrument is the cause of the delay. So we ended up underestimating the time. But if we had to build a similar instrument again, we will not take so much time,” Suryanarayana Sarma said.
For instance, the Rossi X-ray Timing Explorer (RXTE), which carried a much smaller area proportional counter array detector compared to Astrosat’s LAXPC, and no other instrument, took 10 years from concept to realisation. According to a leading scientist of Pennsylvania State University, which was responsible for making, in collaboration with others in the United Kingdom and Italy, the currently operating Swift X-ray satellite, there were about 100 people involved in making its instruments and it took about five years to build the satellite. Swift has three instruments compared to five of Astrosat, and none of them was actually built from scratch. Much of the electronics had also been outsourced.
The UVITs and the SXT are all large telescopes; about three metres long. According to Suryanarayana Sarma, in terms of satellite structure, the major challenge was how to place and align the telescopes, and how to mount them, with each one having a clear FOV and each one needing to be mounted with a specified accuracy so that all are co-aligned. “We needed to do some kind of spacecraft interface and holding brackets and then we had to use special techniques for alignment and make measurements to ensure that it will be sustained even after vibrations and qualifying that these levels were not much disturbed.” According to Koteswara Rao, during the early configuration studies, one of the issues was whether to mount the telescopes outside or inside. “But luckily there is a large space on the central deck of the IRS bus structure which is very stable and which could accommodate two telescopes inside. Though three telescopes for the three channels would have been ideal, there is not enough space in an IRS bus to accommodate three,” he said.
Special Challenges Each of the instruments presented special challenges. Consider, for example, the LAXPC, which is an established system with a smaller size having been used earlier in the Indian X-ray Astronomy Experiment (IXAE) in 1996 (See box). “The big challenge in LAXPC,” said P.C. Agrawal, the former TIFR scientist who built the detector, “was that the detector is 15 cm deep and it is filled with xenon gas at about 2 atmosphere pressure, unlike RXTE which had xenon at 1 atm pressure. The inside has to withstand the 1 atm pressure differential and also it is 1.2 m long. So the cavity was made by milling from a single forged aluminium alloy block, which had to be procured. All the milling work for the three detectors was done at the TIFR workshop on two CNC milling machines operating day and night and all the LAXPC parts too were fabricated there. Each LAXPC had to be within 130 kg. Making the collimator housing, which is 45 cm long, was also a big challenge because here too it has to withstand the pressure differential. Made of 50 micrometre thick tin, with copper layer on it, the collimator design was unique.”
“At some point of time we realised that even the walls over such a large area can contaminate the gas over the five-year lifetime,” said Koteswara Rao. Xenon is very sensitive and even 10 ppm (parts per million) contamination can degrade the resolution. So it was decided to purify it on-board itself using adsorbers. “Each LAXPC has a separate purifier built by the IISU. It is a 0.5 kg bellow-based compressor mounted on the sides by which the gas will be recycled periodically. The contamination build-up is slow and each purification cycle takes about an hour or so. Qualification of these took some more time. We did one purification cycle on ground itself and saw the improvement,” Koteswara Rao said.
Fabricating an X-ray telescope is extremely difficult because of the complex nature of X-ray optics. Because the refractive index for X-rays is less than one, the normal reflecting or refracting optics of visible light does not work. However, X-rays can be reflected at grazing angles, from 10 arc-mins to 2°, from certain surfaces like nickel, gold, platinum and iridium arranged in a certain geometry consisting of co-axial and confocal shells of paraboloid and hyperboloid mirrors (Figure 4). X-rays are first internally reflected off the paraboloidal shells on to the hyperboloidal shells from where they are reflected and focussed to a point. However, at grazing angles, the light collecting area becomes small. This is, however, increased by nesting arrangements of these mirror shells.
Astrosat’s SXT uses shells of conical mirrors approximating paraboloidal and hyperboloidal shapes and the telescope is made of 40 complete shells of such mirrors assembled quadrant-wise (a total of 320 mirrors). “Initially,” according to Koteswara Rao, “use of glass or plastic with appropriate coating to make these shells was thought of but that would have made the telescope very heavy. K.P. Singh had gone to Japan and had studied the method used for the Suzaku X-ray telescope.” Suzaku used gold-coated thin (0.2 mm) aluminum foils pressed into appropriate shapes. While there is a considerable saving in weight, only arc-min angular resolution is achievable. “But the real challenge was to make a large number of such mirrors with only a small unit in TIFR,” pointed out Koteswara Rao. “Besides the space-qualified telescope that is mounted on the satellite, a flight spare telescope and the engineering model of one quadrant had to be made and flight-tested, which meant about 700 mirrors! This really took a lot of time.”
The imaging focal plane assembly in SXT is a CCD array-based detector cooled to –80° C. The CCDs used were made by a British company, E2V, for XMM-Newton. Since the company made only customised CCDs, buying just three for SXT’s use was estimated at Rs.20-25 crore. But, fortunately, a win-win collaboration with the University of Leicester, which was involved in the making of soft X-ray focal plane CCD array for the Swift telescope, could be worked out. The Leicester group had a few leftover CCDs after the Swift work, which were used for the SXT assembly. The CCD signal processing electronics was, however, done by K.P. Singh’s team at TIFR. “For the first time in the world, this was done using Field Programmable Gate Arrays [FPGA] electronics,” said Singh. The required on-board cooling is achieved in two stages: a passive cooling by a radiator plate to which heat is migrated via heat pipes. This cools the CCD array down to -40° C. Then there is an active thermoelectric cooler, which cools it down further to –80° C. Also, since SXT is a long telescope, thermal control could be difficult if the telescope had a normal metal housing. So a CFRP structure was built for SXT by the VSSC.
The CZTI too is a state-of-the-art detector. The Europeans and Americans have flown them in space. But space-qualified CCDs are not readily available. Commercially available ones are usually used for dental X-rays. Large arrays were required for the imager, and detectors meant for medical applications were procured from an Israeli company. “The suppliers had done some improvement in their quality, but still we had to qualify a large number to weed out those that were not of not space-quality,” said Koteswara Rao. According to him, the original plan was to cool the devices down to –20° C but the detector quality was found to be good enough even at 0° C, meeting the specifications. Originally, when it was planned to cool the devices down to –20° C and some detectors were tested, interestingly, the yield was found to be poor. “As we cool down, of course, the thermal noise goes down. But engineering noise begins to dominate,” Koteswara Rao explained.
At zero degrees itself, the performance was found to be optimum. The on-board cooling to 0° C is achieved passively by migrating the heat away by heat pipes to a radiator plate which radiates it away into space. Unlike the SXT CCDs, here there is no need for active cooling with a thermoelectric cooler because it is only 0° C.
“When the developmental work on instruments for LAXPC began, CZTI was not part of the Astrosat payloads,” said P.C. Agrawal. “A.R. Rao [of TIFR] said that he wanted to develop a new technology detector based on the emerging imaging technology based on cadmium-zinc-telluride [CZT] CCDs. I wholeheartedly supported it. The proposal for CZTI was also presented to the configuration committee [see box]. The original proposal had four LAXPCs. To accommodate CZTI, I dropped one LAXPC. If it had been included, the area would have been 30 per cent more,” Agrawal said. The final configuration of Astrosat is based on an IRS bus with four instruments accommodated in the central deck to point at sources in a co-aligned way. “K. Thyagarajan, who was the then director for the IRS mission, played a very crucial role in evolving the Astrosat mission,” said Agrawal.
In the original proposal for the UVIT, there was no visible channel, according to Koteswara Rao. “The discussion with regard to the configuration of the UVIT was mainly on whether there should be a single telescope or two telescopes, one for FUV and another for NUV,” said Koteswara Rao. “Subsequently, what happened was that we enhanced the performance specification of the UVIT to a two arc-sec angular resolution. But with that resolution, even a minute jitter on the spacecraft can spoil the image. Then it was decided to include a visible telescope to monitor the jitter and correct the image by appropriate integration. Actually there is no great science expected from the visible telescope because there are many instruments looking at that region. The main purpose of the visible channel even today is to correct for the spacecraft jitter,” said Koteswara Rao.
There are two UVITs catering to the three channels, with one for FUV and the other for NUV and the visible channel together. “But it is a little complex design,” said Koteswara Rao. “The second telescope has a dichroic filter inside to split the incoming photon beam into two channels. So the two telescopes are not identical in the focal plane. While the CCD arrays are the same, the other components of the assembly, namely the Photomultiplier Tube [PMT] and Micro-channel Plate [MCP], are different. Besides the fabrication of the mirrors themselves, the real technology is in the detector.”
LEOS developed the super-polished mirrors for UVIT, which involved developing new technologies for FUV coatings. According to Koteswara Rao, the specification was 60 per cent reflection of the incident light. But finally what was achieved was 76 per cent reflection. Also, there is a class difference between the surface finish of the visible channel and FUV. “A surface finish of 1.2-1.3 nm surface finish was targeted for the UV mirrors by super-polishing, which we never did for remote sensing payloads. Also, to avoid any stray light scattering effects which could degrade the UV mirrors, a special high absorption (0.98) inorganic coating was developed by ISAC with which the inside baffle is fully coated,” said Koteswara Rao.
According to P.C. Agrawal, the tubes in the housings of the UVIT are made of Invar, the iron-nickel alloy material with low thermal expansion, and were made by the Indo-Russian company BrahMos Aerospace. “There is also an interface cone which is made from titanium, which we could not make in India. This had to be got from the United States. Forging of a single big piece of titanium was required for this purpose,” he said.
One of the complexities involved in the detector system is the critical separation between the PMTs and the MCP located below. Across the two, a high voltage of 6,000 volts has to be applied. If the separation is too much, the image resolution becomes poorer. The two cannot be too near to each other also because of the risk of the two, with high voltage across them, coming into contact with each other. This is exactly what happened during a vibration test and the problem was resolved by the British company Photek, which had provided the basic design for the complex detector assembly, by working closely with ISRO. “While MCP is available commercially, the technology challenge was in adjusting the gap appropriately,” pointed out Koteswara Rao.
The other critical aspect in the design of the UVIT detector assembly is a mismatch in the format of the photon detector and the CCD array. While the PMT system is circular, the CCD array, which is made by the Canadian Space Agency, is square or rectangular. Photek solved the problem by using a precisely tapering optical fibre bundle, which was then glued to the CCD (Figure 5). Sorting out these two problems in the UVIT itself caused a delay of more than three years, said Koteswara Rao.
Having conquered the complex technological challenges on the ground, and having successfully developed the above suite of state-of-the-art instruments, it is to be now hoped that Astrosat would deliver good science given its strength. As mentioned earlier, there is no other mission with this kind of multi-wavelength capability, which even today, despite many upcoming missions, remains the unique strength of Astrosat.
For the first time, you will have the full wavelength coverage from UV to soft X-ray to hard X-ray up to 100 keV. Also, for timing studies, like millisecond variability and Quasi-Periodic Oscillations (QPOs) in black holes, Astrosat’s LAXPC will be important because only a large collecting area can give large counting rates. So, in all likelihood, Astrosat will deliver.
As Koteswara Rao put it: “There still remains a gap in multi-wavelength observatories. We were thinking that by 2012-13, new missions may come up to fill that gap, and that we would be outdated. But surprisingly, that did not happen. Even today, Astrosat has got much relevance. And when our scientists go abroad, they see a lot of interest in the mission. People are waiting in the wings to use its data.”