Developing Transition Edge Sensors for New Space-Based Applications

The goal of the project was to design microcalorimeter technology for a future sounding rocket payload intended to study the diffuse X-ray background in the 0.1-1 keV energy range. The current state of the art for a space-based instrument is the X-ray Quantum Calorimeter (XQC), a silicon thermistor array capable of 6-8 eV FWHM energy resolution below 1 keV. The spectrum of the diffuse background is comprised of many close-packed emission lines from highly ionized common elements (C, N, O, Si, Mg, etc.). To improve our ability to model this emission and understand its source, we need significantly better energy resolution (see Figure 1). Transition-edge sensor microcalorimeters have shown tremendous promise for the next generation of X-ray instruments. A TES detector has been demonstrated in the lab with better than 1 eV FWHM energy resolution at 6 keV (Bandler, 2013). Our group at UW-Madison is part of an MIT-led collaboration currently developing Micro-X, a cryostat designed to carry the first TES microcalorimeters into space. Although the first Micro-X flight will have different scientific goals, with straightforward modifications to the cryostat to enlarge the field of view we could replace the Micro-X TES array with an array optimized to observe the diffuse X-ray background. The diffuse background is a very faint source so the detector array must have a total area of approximately 2 cm2 to observe a significant number of X-ray events in 300 seconds, the typical observing time for a sounding rocket flight. Micro-X will be equipped with 256 readout channels. Therefore, each pixel should be approximately 900 µm x 900 µm. The goal of my NSTRF research was to design a TES detector with 900 µm x 900 µm collecting area and 2 eV FWHM energy resolution at 250 eV. To date, most development effort has been focused on TES devices with 10-100 times smaller surface area that we will need. For example, the 0.9 eV device demonstrated by GSFC was only 65 µm x 65 µm (Bandler, 2013). The TES detectors designed for Micro-X have 590 µm x 590 µm absorbers (very large compared to typical TES pixels) and achieve 2.9 eV FWHM energy resolution (Eckart, 2013). However, these devices still have only 44% of the surface area that our application requires. There are two primary obstacles to scaling TES devices to larger areas. First, the resolution of a detector scales as the square root of the total heat capacity. So, simply enlarging a design without changing materials or operating point will degrade performance. Second, in large devices the time constant for lateral thermal transport can result in pulse shapes that depend on the location of incident X-ray. To prevent degradation of the energy resolution, the time scale on which the device thermalizes much be kept much shorter than the time constant for the detector to return to its equilibrium temperature after an absorption event. To keep the total heat capacity low, we attached a small (140 µm x 140 µm) TES to a 900 µm x 900 µm gold absorber (see Fig. 2). That absorber was to be made as thin as possible while still maintaining good absorption efficiency. Since the primary energy range of interest is 100-500 eV, this means the absorber needs to be only 200 nm thick to maintain near unity quantum efficiency. Modeling showed that a device this size should be capable of 2 eV energy resolution. Although lower heat capacity materials are available, gold was chosen for its excellent thermal transport properties, which would be necessary to avoid position dependent pulse shape degradation in such large, thin absorbers. The detectors were fabricated by the Detector Development Lab at Goddard Space Flight Center. The first wafer was completed in July 2014. Out of concern for potential low yield of the large area thin absorbers, the first devices were given a 500 nm thick gold absorber (rather than 200 nm). Due to the difficulty of reproducing TES bilayer properties between fabrication runs, the detectors operated at 100 mK rather than 70 mK as desired. We measured 6 eV FWHM energy resolution at 1.5 keV, which was expected given the thicker absorbers and warmer operating point. Significantly, we observed no measurable broadening due to position dependent pulse shape at 1.5 keV. GSFC fabricated a second wafer of devices in February, 2015. These detectors had 300 nm thick gold absorbers and operated at 70 mK. As we feared, the absorbers appeared to wrinkle or droop and were, in many cases, making contact with the substrate. Fortunately, we included an alternate pixel design where the absorber is deposited directly onto the nitride membrane that thermally isolates the TES and absorber from the bath. This design has several drawbacks, including lower focal plane filling factor and increased heat loss to the bath - a possible source of position dependent pulse shape variation. A 765 µm x 765 µm x 300 nm device with this design was found to have 3.5 eV FWHM energy resolution at 1.5 keV (Fig. 3). The device had a “baseline” (E = 0 eV) energy resolution of 3 eV, so the increased heat loss in this absorber geometry was significant enough to degrade performance. However, this effect scales linearly with incident Xray energy, so given the amount of degradation observed at 1.5 keV, we expect no significant broadening on the 3 eV baseline resolution for incident energies below 500 eV (in the science energy band). The 3.5 eV resolution we measured for a 765 µm x 765 µm pixel is an encouraging result. We have learned that, even in the case where the absorber sits directly on the supporting membrane and the heat loss is highest, the gold absorber thermalizes quickly enough to prevent degradation of performance, even in very thin absorbers and over a large area. We have also identified a number of areas where further investigation could lead to improved resolution. In all devices, we measured a large excess heat capacity (above calculations made from literature values for the various materials in the detectors). If this could be identified and eliminated, we could improve resolution by approximately 20%. We also found that, despite nominally identical thermometer design, TESs from 900 µm x 900 µm pixels on the first wafer showed significantly higher sensitivity than TESs from the 765 µm x 765 µm pixels on the second wafer. Further testing of pixels from the second wafer should help us pinpoint which aspect of the design is affecting the sensitivity. The 3 eV detectors we have made so far represent a significant improvement over current capabilities, and further testing to better understand the heat capacity of the materials and the sensitivity of the TES can help us make the jump from 3 eV to 2 eV resolution.