Incoming submm radiation produces a small amount of heat in the absorber. The TES detects this temperature rise and translates it into a signal current which is then measured by a SQUID ammeter. The TES is voltage biased on the normal-superconducting transition. The resistance of the TES
has a very steep dependence on temperature in the transition region resulting in a very sensitive detector.
Figure 2. Detector hybridized to mux
Each pixel is connected to a SQUID via an input transformer to form a two-dimensional array of SQUIDs that make up an array of 32 columns x 40 rows. Address currents are applied sequentially to turn on each row, one at a time. The output is inductively coupled to a summing coil, common to all SQUIDs in a column. The output is tracked to the molybdenum wire bonding pads at the edge of the detector where it is routed through to the 1K pcb which supports the SQUID Series Array Amplifiers (SSA).
The detector is wire bonded onto a ceramic pcb which matches the thermal contraction of the detectors. The tracking is silver-plated as aluminum wire bonding is required for the molybdenum pads. Niobium flex cables provide thermal isolation and carry the signals to the 1K pcb which supports four magnetically shielded cans, each of which contain eight SSAs.
To manage the high-density of connections, a method commonly found in mobile phone technology was used. The flexes are attached to the ceramic pcb and the 1K pcb using an anisotropically-conductive adhesive. After applying a combination of heat and pressure to the adhesive, conducting particles within the adhesive are aligned in the z-axis therefore conducting vertically but not horizontally.
As the SQUIDs used for multiplexing and amplification are extremely sensitive magnetometers they must be well shielded from magnetic fields. Shielding is incorporated, but material having any magnetic properties is excluded from the subassembly. At these temperatures (<10K) nickel, which is used in gold-plating for example, becomes magnetic.
To cool the array to ~120 mK a good thermal contact has to be made to the back of the MUX wafer, while at the same time relieving stresses caused by differential contraction that could shatter the wafer. The solution adopted on the SCUBA-2 project is to use a beryllium copper 'hairbrush' (see Figure 3). There is one tine under each pixel and a few under the second-stage SQUIDs. The wafer is attached to the hairbrush using Stycast 1266. A controlled amount of epoxy is deposited onto each individual tine using a commercial liquid deposition system then the wafer lowered onto the hairbrush. The hairbrush is then bolted to the copper cold finger that leads to the mixing chamber of the dilution fridge . A completed subassembly, lying flat, is shown in Figure 4.
The output of 1K pcb is routed to the room temperature readout electronics through a woven cable constructed from superconducting niobium titanium wires weaved with a nomex fibre.
Figure 5 shows the SCUBA-2 focal plane unit which consists of 4 detector subassemblies butted together, to fill most of the 60 arcmin2 field-of-view of the JCMT. There will be two focal plane units to cover the 850 ^m and 450 ^m wavelengths.
Was this article helpful?