An update on SCUBA-2

SCUBA-2 has continued to make good progress over the past year. The highlights have been an extremely successful Proof-of-Concept review for the array technology and a preliminary design review for the main instrument (more details are given below).

On the funding side, the majority of the money needed is now in place following the announcement of the Canadian CFI contribution and approval by PPARC Council. However, following a major re-costing exercise it has become clear that the project is still £1.5m short of the funds needed to complete the instrument (which includes a contingency). Efforts are underway, led by the Director JCMT, to bring extra money into the project before April 2004. Should the shortfall not be realised then the project will be faced with a major de-scope of having only one sub-array at each wavelength - one-quarter of the original focal plane or a loss of a factor of 4 in survey (mapping) speed. Everyone on the project sincerely hopes that this will not be necessary as it represents a huge loss in performance.

In November 2002, a highly successful Proof-of-Concept review was held for the technologies that would be needed to construct a SCUBA-2 sub-array. These included the transition edge pixels themselves, unit cell multiplexers, silicon deep etching, thermal interfaces and bump bonding. Out of the 18 identified technologies, 16 were shown to be under control, with the only uncertainties over the bump bonding - used to hybridise the detector and multiplexer chip, and the deep etching - used to form the pixel and array structure. In particular, optimised single pixels and multiplexers were shown to be capable of meeting the SCUBA-2 requirements in terms of sensitivity, speed of response, power handling and thermal conductance. Figure 1 shows a close-up of a TES pixel and the complex multiplexer unit cell design.

Figure 1: (left) Photograph of a TES test pixel. On the right are the heater leads used to control the operating point, whilst the actual TES leads are on the left. The pixel dimensions (central square) are 1.135mm on a side. The patterning for the legs and slots can also be seen around the periphery of the pixel. (right) The physical layout of the SQUID multiplexer unit cell. The "large" input transformers are seen on the left as well as the summing coil gradiometers on the right. In the centre are the active and dummy SQUIDs (the latter of which is used to reduce crosstalk).

Since the P-o-C review, work has concentrated on understanding and improving the thermal aspects of the bump bond process and refining the deep etch process. New bump-bonded samples are about to undergo thermal tests in Cardiff, following changes in the design that have increased the bump bond density. It is expected that these tests will conclusively demonstrate the required thermal performance. The progress on the deep etching has been spectacular. Figure 2 shows a recent SEM photograph of a section of the array structure.

Figure 2: SEM photograph of a section of a SCUBA-2 sub-array. The white squares are the absorber "bricks" underneath which the TES pixels are deposited. The bricks are suspended on a silicon nitride membrane to control the thermal properties of the pixel. There is a 10 micron wide trench between the bricks and the 50 micron pixel wall (dark area). The deep etching has demonstrated control of the trench can be achieved to the tolerance required.

2. Instrument design

The SCUBA-2 instrument package is being developed in parallel with the array development. This includes the optics, cryostat design, electronics and software. The wide-field optical design has been frozen for some time and the contracts for the 8 re-imaging mirrors have been placed, with the first mirrors expected in March 2004. Figure 3 shows SCUBA-2 positioned at about the level of the current Mezzanine floor on the Left Hand ‘A’ frame. Considerable modifications will be needed to the platform (where SCUBA current resides) before SCUBA-2 can be installed.

Figure 3: View of the instrument on the telescope. The instrument mechanical design is driven by two principal requirements, the large field-of-view of 8 × 8 arcminutes, and a detector operating temperature of 120mK. The large field-of-view results in extremely large mirrors (up to 1.4m across), the last 3 of which must be cooled to temperatures below 10K in order to reduce the thermal background on the arrays. This results in a very large cryostat, the vacuum vessel of which is 2.3m high, 1.7m wide and 2.1m long, with a pumped volume of 5.3m³ – essentially a ‘walk-in’ cryostat!

The cryostat underwent a preliminary design review in March this year and since then work has concentrated on detailing the design of the 1K box (which houses the arrays), the 4K optical bench (cold mirrors), radiation shields and vacuum vessel. Figure 4 shows a part cut-away section of the cryostat. As seen the cryostat is made up of a series of layers. Immediately inside the vacuum vessel is a multi-layer insulation blanket and 60K box that provides radiation shielding for the main optical bench, which is expected to operate at 4K. These are both cooler by a pair of pulse-tube coolers. The 1K box is cooled by the still of a dilution refrigerator. Mounted within the 1K box are the two Focal Plane Units that contain the cold electronics and the 450µm and 850µm arrays. A liquid cryogen-free dilution refrigerator cools the arrays to below 100mK.

Figure 4: Cut-away 3-D CAD drawing of the SCUBA-2 cryostat. The incoming beam envelope can be seen in red on the left-hand side. The beam is re-imaged by three cold mirrors - including one, which is over a metre in diameter and cooled to 4K! The 1K box sits inside the hemispherical section on the right. This box provides a light tight environment and also a magnetic shield for the sensitive SQUID multiplexers. The dilution refrigerator insert can be seen at the top of the cryostat.

The low-temperature thermal design and how the sub-array modules are cooled remains one of the most challenging aspects of the design. Each module is designed as an independent unit, which contains a sub-array of 32 × 40 pixels. Figure 5 shows a drawing of an entire sub-array module.

Figure 5: A SCUBA-2 sub-array module. The gray square is the sub-array, attached via wire bonds, to a ceramic PCB that is in turn supported on copper "balcony" type structure. The niobium flexes are hot-bar soldered to PCBs at each end. The green PCB supports the SQUID series arrays at 1K, housed in shielded cans. Conventional ribbon cables bring the detector signals out of the cryostat to the warm, multi-channel electronics and data acquisition system.