Xray DROIDS
Superconducting tunnel junction (STJ) technology offers tremendous potential for the next generation of spaceborne astronomical Xray missions. STJ's have intrinsic energyresolution and timetagging capability and are sensitive across a broad energy range. One problem, however, is that each device in an array needs to be individually biassed and readout, necessitating at least one electrical connection per pixel. This requirement hampers the development of largearea arrays.
DROIDS, Distributed ReadOut Imaging Devices, are one possible solution to this problem in both the Xray and optical regimes. In a DROID, photons are absorbed in a single, largearea crystal (or a smaller polycrystalline layer), at the corners or edges of which are situated individual STJ's. Timecoincident event measurement allows the energy and position of an incoming photon to be reconstructed. The DROID therefore behaves similarly to a conventional largearea array of STJ's, at the possible expense of some energy resolution and the requirement to sustain a low count rate. The Detectors and Optics group has been investigating the viability of 1 and 2dimensional DROIDS as prototypical astronomical Xray detectors, with the emphasis on spatial and energy resolution. The group's work on the feasibility of such devices as optical/UV detectors is summarised here, while matrix readout, another approach to the largearea format problem, is covered here.
Experimental setup: Image ReconstructionThe 1 and 2dimensional droids consist (see crosssection below) of an epitaxial Ta absorber with a thickness of 100 nm and an RRR of ~48, deposited on a sapphire substrate. The STJ's are symmetrical, with 100 nm thick Ta layers and 55 nm thick Al trapping layers. The base electrode is common with the absorber. To prevent oxidation, the whole detector is covered with a 480 nm thic SiOx layer. The STJ's have individual Nb connections to their top electrodes and a common Nb return contact via the absorber, intended to avoid quasiparticle losses through outdiffusion into the electrical connections. The junctions have a normal resistance of ~4.2 microOhms per square centimeter, with a gap energy of 0.42 meV.
Three 1dimensional and two 2dimensional DROIDS were tested, and details of the absorber and STJ sizes of these individual devices may be found in Hartog et al (2000). In the first series of experiments the DROIDS were illuminated with a ^{55}Fe source (producing emission features at 5.9 and 6.5 keV), with the devices contained in a ³He cryostat at a base temperature of 320 mK. The second and third series of tests were performed in collaboration with the PhysikalischTechnische Bundesanstalt (PTB) in Berlin, using the PTB SX700 beamline at the BESSY I synchroton facility and the PTB FCM beamline at BESSY II. BESSY I offers photons in the range ~40 eV to 1.95 keV, while BESSY II runs from 1.7 to 11 keV. Details of subgap currents and bias voltages in these tests may be found in Hartog et al (2000) , together with details about the readout modes of the STJ's.
Image ReconstructionAn incoming photon hitting a 2D DROID somewhere in the absorber will produce coincident signals in the four corner STJ's. How do we reconstruct the (x,y) coordinates of the event from the coincident charge signals? In order to do so, we need to define a model that predicts the charge signals as a function of (x,y), and then invert this model to derive event positions for our list of measured coincident charge signals. The model has to take into account qp diffusion, losses and tunneling, as well as the characteristics of the pulseprocessing electronics. It turns out to be quite easy to construct such a model based on numerical finitedifference technqiues, but in order for the model to supply useful spatial resolution, exceedingly large computational resources are needed. Further details are found in Hartog et al (2000), but as an example of a reconstructed image, we show the illumination of a DROID with a Fraunhofer diffraction pattern (see below):
Energy ResolutionA number of factors currently complicate the determination of the energy resolution of 2D DROIDS, so we here restrict our discussion to onedimensional devices. The analysis of data from 1D DROIDS is clearly simpler, but there are still several complications. It is found that when the signal from one STJ is at maximum, the signal from the other is still ~ 60% of its maximum value, indicating that quasiparticles are hardly trapped in our STJ's. Another complicating factor is that the shape of the current pulses generated in the STJ's depends on absorption position. This means that a single shaping filter cannot be optimized for all the different pulse shapes, requiring us to correct for the differences in charge output. The steps involved are outlined in the figures below.
The energy resolution of a DROID is determined by three essentially Poissonian processes, their contribution to the resolution scaling with the square root of the photon energy:
where the three terms F,G and H in the above equation are:
There are several other statistical contributions which only apply to DROIDS which (unlike ours) operate in the strongtrapping regime. In addition, there are also electronic noise and bias instability contributions to the overall resolving power degradation. The first term may be subtracted from the total noise by reference to test pulser observations. Finally there are contributions from variations of quasiparticle losses with absorber position, both along the width and length of the absorber. Details on the modelling of the energyresolution may be found in Hartog et al (2000). The table below displays the best measured energy resolutions at several energies against the expected contributions from the various mechanisms discussed above, with the various discrepancies (and likely explanations) again discussed in Hartog et al.
Requirements for Largeformat DetectorsIf DROIDS are to be flown on future space missions, they will need to be increased in size such that they have absorber dimensions of several mm. The tunnel noise is the main limiting factor; increasing the length of the absorber while keeping the STJ's at the same size decreases the barrier area and hence tunnel probability, and thus raises the tunnel noise. Scaling up the STJ's increases their capacitive noise. Moving to lower energygap materials such as Hafnium improves the quasiparticle statistics, but the time resolution must be improved by a factor of 10 before it reaches the < 5 microsecond level required by a mission such as XEUS. This may be achieved by decreasing the impurity content of the absorber (see Hartog et al (2000) for more details).
Last Update: 06 September 2013
