Characterization of high aspect ratio TiAu TES X-ray microcalorimeters array using the X-IFU Frequency Domain Multiplexing readout
E. Taralli, L. Gottardi, K. Nagayoshi, M. Ridder, S. Visser, P. Khosropanah, H. Akamatsu, J. van der Kuur, M. Bruijn, J.R.Gao
aa r X i v : . [ phy s i c s . i n s - d e t ] F e b Journal of Low Temperature Physics manuscript No. (will be inserted by the editor)
Characterization of high aspect ratio TiAu TES X-raymicrocalorimeters array using the X-IFU Frequency DomainMultiplexing readout
E. Taralli · L. Gottardi · K. Nagayoshi · M.Ridder · S. Visser · P. Khosropanah · H.Akamatsu · J. van der Kuur · M. Bruijn · J.-R.Gao the date of receipt and acceptance should be inserted later
Abstract
We are developing X-ray microcalorimeters as a backup option for the baselinedetectors in the X-IFU instrument on board the ATHENA space mission led by ESA and tobe launched in the early 2030s.5 × Keywords transition-edge sensor, energy resolution, X-IFU, AC bias
E. Taralli · L. Gottardi · K. Nagayoshi · M. Ridder · S. Visser · P. Khosropanah · H. Akamatsu · J. van derKuur · M. Bruijn · J.R. GaoSRON Netherlands Institute for Space ResearchSorbonnelaan 2, Utrecht, 3584 CA, The NetherlandsE-mail: [email protected]. GaoFaculty of Applied Science, Delft University of TechnologyLorentzweg 1, 2628 CJ, Delft, The Netherlands E. Taralli et al.
ATHENA - Advanced Telescope for High-ENergy Astrophysics - is the second large mis-sion of ESA’s Cosmic Vision-programme to study astrophysical phenomena near black holesand galaxy clusters. The X-ray Integral Field Unit (X-IFU) , one of the two instruments onboard, will deliver spatially-resolved high-resolution X-ray spectroscopy over a limited fieldof view. X-IFU requires detectors in an array of nearly 3264 pixels, being sensitive in the0.2–12 keV energy range with 2.5 eV FWHM energy resolution below 7 keV. SRON iscurrently developing a Frequency Domain Multiplexing (FDM) readout system as the base-line and X-ray microcalorimeters arrays based on TiAu transition-edge sensors (TESs) as abackup technology for the US MoAu TES array, currently being developed at NASA God-dard Space Flight Center .In the last few years, a lot of progress has been reported in both the FDM readout and in ourunderstanding of TES physics under AC bias. The latter has been moved forward continu-ously by improving the detector design and performance. For instance, bare TESs have beenfound to be preferable to TESs with normal metal structures. It has also been shown thatAC loss is bias-frequency dependent and depends also on the amount of the normal metalused in the TES. Moreover, bare TESs have shown smaller weak-link effects and the be-haviour of α and β are not affected by kinks in the IV curve. The normal resistance of theTES, together with its saturation power, plays a crucial role in the reduction of the Joseph-son effect which can affect the performance of a TES microcalorimeter under AC bias. TheJosephson effects are non-dissipative so that they do not add noise to the detector, but theyintroduce undesired, bias-dependent nonlinearities, which limit the optimal bias range. Tosimplify the FDM read-out of a large TES array, these effects should be minimised. It wasrecently been shown that the degradation of performance from the frequency multiplexingscheme can be mitigated by the use of TESs with high resistance . It is also important tohave a thicker bilayer to reduce the internal thermal fluctuation noise.All of this new knowledge has led the design and fabrication of the new SRON 5 × is introduced to reduce the internal thermal fluctuation noise. Furthermore,higher aspect ratios increase the normal resistance of the single pixel and thus increase thesaturation power; the absence of normal metal structures reduces the weak-link effect; anda thinner membrane is used to lower the thermal conductance.In this paper we present our experimental results in terms of energy resolution for tens ofTiAu TESs grouped into ten different types of device with aspect ratio (length-to-width)ranging from 1-to-1 up to 6-to-1. They were measured in a single-pixel mode with biasfrequencies between 1 and 5 MHz,using an 18-channel FDM readout system which is a aprototype of the FDM readout system for X-IFU. Fig. 1 shows the design of two different 5 × × width) (from left to right in Fig.1a): 100 ×
30, 120 ×
40, 100 ×
25, 100 ×
100 and 80 × μ m , while the other array, labelledarray R4a, contains very high-aspect ratio devices with the following sizes (from left to rightin Fig. 1b): 100 ×
20, 120 ×
20, 140 ×
30, 80 ×
20 and 80 × μ m . Note that every columnhas the same detector size in order to study the performance of the TES as a function of itle Suppressed Due to Excessive Length 3(a) (b) Fig. 1: 5 × ×
30, 120 ×
40, 100 ×
25, 100 ×
100 and 80 × μ m ( Fig. 1a ). Array labelled R4a with very high-aspect ration devices with the following sizes(from left to right): 100 ×
20, 120 ×
20, 140 ×
30, 80 ×
20 and 80 × μ m ( Fig. 1b ). Eachcolumn has devices with the same aspect ratio.
Red squares identify the TESs that wereconnected and measured. (Color figure online.) R N ( W ) T (K)
Px11-140x30Px5-100x30Px18-100x100
Fig. 2: 4-terminal resistance measurement of pixels with three different aspect ratios:100 ×
100 ( blue dots ), 100 ×
30 ( purple crosses ) and 140 × μ m ( green squares ) measuredwith the same excitation current of 3 μ A. Lines serve no other purpose than to guide the eye.(Color figure online.)
E. Taralli et al. the bias frequency and to have reasonable statistics on pixel with the same design. The redsquares point out the 31 TESs that have been connected and measured. All the TESs havethe same bilayer thickness of Ti (35 nm) and Au (200 nm) and use a 0.5 μ m thick SiN mem-brane. We measured a normal squared resistance R = 25 m Ω /sq and a critical temperatureT c ∼
115 mK. The thermal conductance G at T c ranges from 107 to 172 pW/K and from 61to 104 pW/K for the different aspect-ratio arrays R3b and R4a, respectively.Recently it has been shown that, depending on the quality of the Bismuth in the absorber,a non-Gaussian thermal response can be induced in the energy spectra. In order to identifythe real performance of the pixel design, all the absorbers consist only of Au 2.3 μ m thickand have the same size (240 × μ m ) with a heat capacity C = 1.1 pJ/K at T c . Each ab-sorber has, at its corners, four contacting points to the membrane and two central contactingpoints directly to the sides of the TES. More details on fabrication of the SRON TES arraywill be published in this special issue .Due to the thicker bilayer used in this work, compared to the previous batch , we needto tailor the aspect ratio to obtain the required normal resistance and bias power. Fig. 2shows the 4-terminal resistance measurements as a function of bath temperature measuredfor TESs with different dimensions which are a squared 100 × ×
30 and a 140 × μ m TES.The characterization of the two arrays was performed in two different experimental mea-surement setups, named XFDMLarge for the R3b array and named XFDMProbe for theR4a array. Both setups were placed in the same dilution refrigerator that can provide a bathtemperature of ∼
40 mK. TESs were characterised under AC bias using an existing FDMreadout system (1–5 MHz) in the single-pixel mode configuration. Each TES is connectedin series with an LC resonator on a LC filter chip with a coil inductance L = 2 μ H and a 1:2transformer chip for the XFDMLarge and an L = 1 μ H and a direct connection to the TESs(no transformers) for the XFDMProbe. The TES array chips and the cryogenic componentsof FDM readout were mounted in a low magnetic impurity copper bracket fitted into an Alsuperconducting shield. The bracket of each setup also accommodates a heater, a thermome-ter and a Helmholtz coil. The heater and thermometer are used to regulate the temperaturelocally on the chip. The coil is for applying a uniform magnetic field perpendicular to theTES array to compensate any remnant magnetic field trapped in the experiment setup.
An Fe-55 X-ray source is placed closely above the array in both the setups and illuminatesthe entire TES array at a count rate of ∼ α N , where the Noise Equivalent Power(NEP) scan usually predicted the lowest energy resolution. Fig. 3a and 3b show a graphicalimpression of the global energy resolution measured for arrays R3b and R4a, respectively.About 20 of the pixels show an excellent energy resolution between 2.4 and 2.8 eV, eight ofthem between 2.8 and 3 eV and only three are worse than 3 eV.In Fig. 4 we present four typical spectra, as examples. The best energy resolution obtainedfor array R3b mounted in the XFDMLarge setup is shown in Fig. 4b. It is worth noting howthe energy resolution in both the setups remains excellent when devices with the same aspectratio are biased at different resonant frequency f as shown for instance in Fig. 4c-4d. Thebest energy resolution 2.39 ± × μ m biased at a resonant frequency of 1.6 MHz and between 10- itle Suppressed Due to Excessive Length 5(a) (b) Fig. 3: Overview of the-single pixel energy resolution measured in array R3b array (
Fig. 3a )and in array R4a (
Fig. 3b ). Consider an error bar of ± N . This result is shown in Fig. 4a. At this stage we are not yet able to identify thebest TES design illustrating the ultimate energy resolution. It is also still unclear, whetherthere is a correlation between the energy resolution and the different aspect ratios of theTES. In the ideal scenario, we would expect to find the identical energy resolution along thecolumns of Fig. 3 and possibly a significant variation between them, according to the pixeldesign variation. However, we still have to deal with the different behaviour of the TESat the different bias frequencies, which is reflected in the spread of the energy resolutionalong a single column. If we consider the R3b array (Fig. 3a) and take the mean of theenergy resolution for each column, we get a uniform energy resolution ranging from 2.7 to2.84 eV, although the different squared TESs (last two columns) have worse statistics, i.e.small spread in bias frequencies. In the R4a array these averages are slightly more scattered,giving an energy resolution ranging from 2.55 up to 2.86 eV. This could be due to the broadervariation in the pixel design but also to the poorer statistics for the last two columns of Fig.3b. Of course, with the upcoming measurements on other high aspect-ratio mixed arraysand on the uniform kilo-pixel array, we might define the relationship between the detectorenergy resolution, the pixel design and the bias frequency. What is clear is that seven pixelsout of 16 in R4a (mounted in the XFDMProbe setup) show an energy resolution around2.5 eV, which is a slightly better than that found for array R3b (3 out of 15). This differenceis likely due to the better temperature stability ( ∼ μ K at 55 mK) in the XFDMProbe; theXFDMLarge ( ∼ μ K at 50 mK) is less stable. Furthermore, we notice that square or close-to-square TESs with a lower normal resistance are more affected by the weak-link effect,showing an oscillation in the predicted energy resolution as can be seen in the NEP plottedin Fig. 5a. This means that, although there might be very good energy resolution (Fig. 4b),this happens only for a limited number of bias points through the transition. On the otherhand, high aspect-ratio devices with a higher normal resistance show a flat NEP over a widebias range between 5 and 30% of R N as shown in Fig. 5b.It is important to highlight that these results have been obtained with a T c of 110 mK and forthese arrays this could be a limiting factor. Reducing the T c down to ∼
90 mK should givean improvement in the energy resolution of between 15 and 20%, approaching even closerto the X-IFU requirements (2.5 eV below 7 keV over the whole array).
E. Taralli et al.(a) TES14-140 × μ m , f =1.6 MHz (b) TES24-80 × μ m , f =2.9 MHz(c) TES12-100 × μ m , f =3.8 MHz (d) TES13-100 × μ m , f =4.6 MHz Fig. 4: Small selection of the best spectra at 5.9 keV for TESs mounted in the XFDMLargesetup (from Fig. 4b to 4d) and in the XFDMProbe (Fig. 4a) with bias points between 15-20% of R N . The red solid line is the best fit to the data, and the blue points are the measuredMn-K α emission lines. The lower plots show the residuals of the fit normalised by the errorbars. (Color figure online) (a) (b) Fig. 5: Noise Equivalent Power (NEP) scan for TES24-80 × μ m ( Fig. 5a ) and for TES12-100 × μ m ( Fig. 5b ) as a function of bias point for three different bias frequencies. itle Suppressed Due to Excessive Length 7
New 5 × c of 110 mK, no normal metal structures on the top of the bilayer, andmake use of a thinner SiN membrane (0.5 μ m). We measured an energy resolution between2.4 and 2.8 eV at 5.9 keV on 20 out of the 31 TESs. New TESs with a T c ∼
90 mK shouldimprove the energy resolution further and meet the X-IFU requirements for all the pixels inthe array.Our microcalorimeters represent the best TES microcalorimeters ever reported in Europe.Not only are our detectors able to meet the detector requirements of the X-IFU instrument,but they also potentially offer technology for other future X-ray space missions, fundamen-tal physics experiments, plasma characterization, and material analysis. We are now readyto test a kilo-pixel array in combination with the FDM readout in multi-pixel mode.
Acknowledgements
This work is partly funded by European Space Agency (ESA) and coordinated withother European efforts under ESA CTP contract ITT AO/1-7947/14/NL/BW. It has also received fundingfrom the European Union’s Horizon 2020 Programme under the AHEAD (Activities for the High-EnergyAstrophysics Domain) project with grant agreement number 654215.
References
1. F. Pajot, D. Barret, T. Lam-Trong, J.-W. den Herder et al,
J. Low Temp. Phys. , 901, (2018).2. European Space Agency (ESA), ”ATHENA Mission Summary”, Available:http://sci.esa.int/athena/59896-missionsummary.3. S. J. Smith, J. S. Adams, S. R. Bandler, et. al.,
Proc. SPIE , (2016),https://doi.org/10.1117/12.22317494. H. Akamatsu, L. Gottardi, J. van der Kuur, C.P. de Vries, M. P. Bruijn, J. A. Chervenak, M. Kiviranta,A. J. van den Linden, B. D. Jackson, A. Miniussi, K. Sakai, S. J. Smith and N. A. Wakeham,
Proc. SPIE , (2018), https://doi.org/10.1117/12.23132845. K. Sakai, J. S. Adams, S. R. Bandler, J. A. Chervenak, A. M. Datesman, M. E. Eckart, F. M. Finkbeiner,R. L. Kelley, C. A. Kilbourne, A. R. Miniussi, F. S. Porter, J. S. Sadleir, S. J. Smith, N. A. Wakeham, E.J. Wassell1, W. Yoon, H. Akamatsu, M. P. Bruijn, L. Gottardi, B. D. Jackson, J. van der Kuur, B. J. vanLeeuwen, A. J. van der Linden, H. J. van Weers, M. Kiviranta,
J. Low Temp. Phys. , 356, (2018).6. N. A. Wakeham, J. S. Adams, S. R. Bandler, J. A. Chervenak, A. M. Datesman, M. E. Eckart, F. M.Finkbeiner, R. L. Kelley, C. A. Kilbourne, A. R. Miniussi, F. S. Porter, J. E. Sadleir, K. Sakai, S. J.Smith, E. J. Wassell and W. Yoon,
J. Low Temp. Phys. , 231, (2018).7. E. Taralli, P. Khosropanah, L. Gottardi, K. Nagayoshi, M. L. Ridder, M. P. Bruijn, and J. R. Gao,
AIPAdvances , 045324, (2019), DOI:10.1063/1.5089739.8. L. Gottardi, S. J. Smith, A. Kozorezov, H. Akamatsu, J. van der Kuur, S. R. Bandler, M. P. Bruijn, J. A.Chervenak, J. R. Gao, R. H. den Hartog, B. D. Jackson, P. Khosropanah, A. Miniussi, K. Nagayoshi, M.Ridder, J. Sadleir, K. Sakai and N. Wakeham, J. Low Temp. Phys. , 209, (2018).9. P. Khosropanah, E. Taralli, L. Gottardi, C. P. de Vries, K. Nagayoshi, M. L. Ridder, H. Akamatsu, M. P.Bruijn and J.-R. Gao,
Proc. SPIE , (2018), https://doi.org/10.1117/12.231343910. D. Yan, R. Divan, L. M. Gades, P. Kenesei, T. J. Madden, A. Miceli, J.-S. Park, U. M. Patel, O.Quaranta, H. Sharma, D. A. Bennett, W. B. Doriese, J. W. Fowler, J. D. Gard, J. P. Hays-Wehle, K.M. Morgan, D. R. Schmidt, D. S. Swetz, and Joel N. Ullom,
Appl. Phys. Lett. , 192602, (2017);https://doi.org/10.1063/1.500119811. K. Nagayoshi, M.L. Ridder, M.P. Bruijn, L. Gottardi, E. Taralli, P. Khosropanah, S. Visser and J. R.Gao,
J. Low Temp. Phys. . This Special Issue (2019).12. H. Akamatsu, L. Gottardi, J. van der Kuur, C. P. de Vries, K. Ravensberg, J. S. Adams, S. R. Bandler,M. P. Bruijn, J. A. Chervenak, C. A. Kilbourne, M. Kiviranta, A. J. van der Linden, B. D. Jackson, S. J.Smith,
Proc. SPIE99055S