Search

The traditional method of applying a digital optimal filter to measure X-ray pulses from transition-edge sensor (TES) devices does not achieve the best energy resolution when the signals have a highly non-linear response to energy, or the noise is non-stationary during the pulse. We present an implementation of a method to analyze X-ray data from TESs, which is based upon principal component analysis (PCA). Our method separates the X-ray signal pulse into orthogonal components that have the largest variance. We typically recover pulse height, arrival time, differences in pulse shape, and the variation of pulse height with detector temperature. These components can then be combined to form a representation of pulse energy. An added value of this method is that by reporting information on more descriptive parameters (as opposed to a single number representing energy), we generate a much more complete picture of the pulse received. Here we report on progress in developing this technique for future implementation on X-ray telescopes. We used an 55Fe source to characterize Mo/Au TESs. On the same dataset, the PCA method recovers a spectral resolution that is better by a factor of two than achievable with digital optimal filters.

We are developing kilopixel arrays of TES microcalorimeters to enable high-resolution x-ray imaging spectrometers for future x-ray observatories and laboratory astrophysics experiments. Our current array design was targeted as a prototype for the X-ray Microcalorimeter Spectrometer proposed for the International X-ray Observatory, which calls for a 40×40-pixel core array of 300 micron devices with 2.5 eV energy resolution (at 6 keV). Here we present device characterization of our 32×32 arrays, including x-ray spectral performance of individual pixels within the array. We present our results in light of the understanding that our Mo/Au TESs act as weak superconducting links, causing the TES critical current (I(sub c)) and transition shape to oscillate with applied magnetic field (B). We show I(sub c)(B) measurements and discuss the uniformity of these measurements across the array, as well as implications regarding the uniformity of device noise and response. In addition, we are working to reduce pixel-to-pixel electrical and thermal crosstalk; we present recent test results from an array that has microstrip wiring and an angle-evaporated copper backside heatsinking layer, which provides copper coverage on the four sidewalls of the silicon wells beneath each pixel.