AC/DC characterization of a Ti/Au TES with Au/Bi absorber for X-ray detection
E. Taralli, C. Pobes, P. Khosropanah, L. Fabrega, A. Camon, L. Gottardi, K. Nagayoshi, M. Ridder, 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)
AC/DC characterization of a Ti/Au TES with Au/Bi absorberfor X-ray detection
E. Taralli · . C. Pobes · P. Khosropanah · L. Fabrega · A. Cam´on · L. Gottardi · K.Nagayoshi · M. Ridder · M. Bruijn · J.R. Gao the date of receipt and acceptance should be inserted later
Abstract
Transition-edge sensors (TESs) are used as very sensitive thermometers in mi-crocalorimeters aimed at detection of different wavelengths. In particular, for soft X-rayastrophysics, science goals require very high resolution microcalorimeters which can beachieved with TESs coupled to suitable absorbers. For many applications there is also needfor a high number of pixels which typically requires multiplexing in the readout stage. Fre-quency Domain Multiplexing (FDM) is a common scheme and is the baseline proposed forthe ATHENA mission. FDM requires biasing the TES in AC at MHz frequencies. Recentlythere has been reported degradation in performances under AC with respect to DC bias. Inorder to assess the performances of TESs to be used with FDM, it is thus of great interestto compare the performances of the same device both under AC and DC bias. This requirestwo different measurement setups with different processes for making the characterization.We report in this work the preliminary results of a single pixel characterization performedon a TiAu TES under AC and afterwards under DC bias in different facilities. Extraction ofdynamical parameters and noise performances are compared in both cases as a first stage forfurther AC/DC comparison of these devices.
Keywords
Transition-edge sensor, Complex impedance, AC and DC bias
E. Taralli · P. Khosropanah · L. Gottardi · K. Nagayoshi · M. Ridder · M. Bruijn · J.R. GaoSRON Netherlands Institute for Space ResearchSorbonnelaan 2, Utrecht, 3584 CA, The NetherlandsE-mail: [email protected]. Pobes · A. Cam´onICMA, Zaragoza, SpainE-mail: [email protected]. FabregaICMAB-CSIC, Barcelona, SpainJ.R. GaoFaculty of Applied Science, Delft University of TechnologyLorentzweg 1, 2628 CJ, Delft, The Netherlands E. Taralli et al.
The Netherlands Institute for Space Research (SRON) is currently developing a FrequencyDomain Multiplexing (FDM) readout system as baseline and X-ray TiAu transition-edgesensor (TES) microcalorimeter arrays as a backup technology for the X-ray integral fieldunit (X-IFU) instrument [1] inside the future ATHENA mission [2] led by ESA and to belaunched in 2030s.Our current FDM readout system applies a set of 18 sinusoidal AC carriers (between 1 and5 MHz), which bias the TES detectors at their working points. This is a small version of thebaseline readout for the Flight Model which will consist of channels with 40 pixels.Another technique being developed for the X-ray TES readout as a backup option for X-IFUinstrument is Time-Division Multiplexing (TDM) [3], where the important characteristic isthat the TESs are DC-biased.In order to fully understand the main differences between these two readout schemes andhence the behaviour of the devices involved in large arrays, it is worth to probe and comparetheir functionality and their properties under both DC and AC bias. It is essential to demon-strate that the observed good performance of a single pixel under constant voltage bias aremaintained even when the TES works as a modulator [4].In this paper we present a preliminary comparison by means of IV curves, complex impedancemeasurements and noise spectra of an X-ray TiAu TES microcalorimeter under DC biasperformed at the Institute of Material Science of Aragon (ICMA) and under AC bias at afrequency of 3.5 MHz performed in SRON. × × μ m TiAu (20/50 nm) bi-layer thermometer with three additional normal metal strips, situated on a 1 μ mm thick SiNmembrane. The thermometer has a T c ∼
100 mK, a normal resistance R N = 220 m Ω and athermal conductance to the bath G ∼
150 pW/K. The X-ray absorber, consisting of 3.5 μ m Bion top of 3 μ m Au, has a size of 248 × μ m . It has, at its corners, four contact points tothe membrane and an additional contact point in the centre of the TES. More details on thefabrication of the TES array can be found in Ref. [5].2.2 AC bias and setupThe characterization at SRON has been done in our FDM readout system in single pixelmode, where a TES is biased by a carrier signal with a bias frequency f c between 1 and5 MHz. A high-Q superconducting LC resonator filter chip defines the different bias fre-quencies f c [6]. This chip contains 18 LC-resonator circuits, with resonance frequenciesseparated by nominally 200 kHz. Coil inductance L = 400 nH for all 18. The TES current ispicked up by a two-stage SQUID assembly, consisting of a low-power single SQUID at thebase temperature and a high-power SQUID array at the 2K stage.The TES array chip and the cryogenic components of the FDM readout were mounted in alow magnetic impurity copper bracket fitted into an Al shield and accommodated in a dilu-tion refrigerator with a bath temperature T bath ∼
40 mK. T bath on the bracket can be locally
C/DC characterization of a Ti/Au TES with Au/Bi absorber for X-ray detection 3 tuned by a heater and a thermometer directly connected to the setup itself. A Helmholtz coilplaced on top of the array is used to apply an uniform perpendicular magnetic field on theTES array.2.3 DC bias and setupThe Council for Scientific Research in Spain (CSIC) is also involved in the developmentof TES sensors based on Mo/Au proximity bilayers with Au/Bi absorbers (developed bythe Institute of Material Science of Barcelona ICMAB and the Institute of Material Scienceof Aragon - ICMA). DC characterization at ICMA is performed in an Oxford dilution re-frigerator with a T bath ∼
30 mK. The TES chip is attached to the mixing chamber and theexperimental volume is shielded against external magnetic fields by a mu-metal with a leadlayer inside. The holder hosts a compensating coil to cancel remnant magnetic fields al-though it could not be properly used in these measurements. TES current can be measuredthrough a two-stage low noise SQUID (manufactured at the PTB Institute in Berlin) with 2m Ω shunt resistor.TES is biased by means of a DC current source from the same Magnicon electronics thatcontrols the SQUID and TES polarization. All the measurements were performed in Flux-Locked-Loop (FLL) mode with a feedback resistance of 100 k Ω . We use IV curves, complex impedance and noise measurements in both the setups to char-acterise and compare the behaviour of our TES. In general the complex impedance is themeasurement that needs some additional explanation. All the details about the compleximpedance measurements and related calibration can be found in Ref. [7] for AC and inRef. [8] for DC case. In this section we give some detail on the Markov Chain Monte Carlo(MCMC) fitting method used to fit the complex impedance measurements.A common reason to use the MCMC method is that it would be useful to marginalise oversome parameters and find an estimate of the posterior probability function for others. Forthis purpose we first define the fit function Z
TES = Z inf + (Z inf – Z ) ωτ eff which is de-rived from the one-body model that seems to be enough to describe our TES, based onprevious measurements of such devices [7]. This is our likehood function, where Z andZ inf are the low-frequency and the high-frequency limit of the impedance, respectively and τ eff is the effective time constant of the detector. Afterwards we define their prior probabilitysupposing they are described by normal distribution and their posterior probability that isa conjunction between the prior probability and the likehood function. We began samplingour parameter space using walkers (much more than twice of the number of parameters be-ing varied during the fit) starting from a tiny Gaussian ball around the maximum likelihoodresult obtained from a standard fit with the common method of least squared [9]. The plotin Fig. 1a shows all the one and two dimensional projections of the posterior probabilitydistributions of our parameters. This can quickly demonstrate all of the covariances betweenparameters and shows the standard deviation for each of them. The analysis showed in Fig.1a refer to impedance measurement in AC bias at 31% of the transition with T bath =55 mK.In Fig. 1b there are some of the complex impedance measurement at different bias points inAC and DC case and related fit both at 55 mK (Top) and 75 mK (Bottom). This procedurehas been applied to all the bias points for both the setups. E. Taralli et al.(a) -0.2-0.1 0 0.1 0.2 -0.2 -0.1 0 0.1 0.2 I m a g Z ( w ) ( W ) AC - 55 mK -0.2-0.1 0 0.1 0.2-0.2 -0.1 0 0.1 0.2 I m a g Z ( w ) ( W ) DC - 55 mK -0.2-0.1 0 0.1 0.2 -0.2 -0.1 0 0.1 0.2 I m a g Z ( w ) ( W ) Real Z( w ) ( W ) AC - 75 mK -0.2-0.1 0 0.1 0.2-0.2 -0.1 0 0.1 0.2 I m a g Z ( w ) ( W ) Real Z( w ) ( W ) DC - 75 mK (b)
Fig. 1: The
Top plot shows the statistical relationship among the fit parameters. On the
Bottom complex impedance measurements at different percentages of R N ( Open symbols )with related fits (
Black lines) with a bath temperature of 55 mK ( top graphs ) and 75 mK(
Bottom graphs ) in AC and DC bias respectively. (Color figure online.)
IV curves have been measured at T bath of 55 mK and 75 mK (Fig. 2a and 2c, respectively)and the corresponding extrapolated RT curves are shown in Fig. 2b and 2d. IV curves pointout a small offset between the two curves, probably due to a residual magnetic field that wasnot possible to be canceled in the DC setup.From the fitting parameters of the complex impedance measurements we can derive the sen-sitivity of the TES resistance on the temperature α and on the current β and the loop gain atlow frequency L . We use the following equations: β = Z inf R – 1, L = C τ eff G – 1 and α = L GTP
C/DC characterization of a Ti/Au TES with Au/Bi absorber for X-ray detection 5 I TE S ( m A ) AC-55 mKDC-55 mK R ( W ) AC-55 mKDC-55 mK I TE S ( A ) V TES ( m V) AC-75 mKDC-75 mK R ( W ) T (K)
AC-75 mKDC-75 mK
Fig. 2: IV and extrapolated RT curves of the pixel in AC (
Purple crosses ) and DC case( green squared ) at 55 ( top graphs ) and 75 mK (
Bottom graphs ). (Color figure online).where R is the TES resistance at the specific bias point, G is derived from the P(T) curve, Tis the TES temperature, P is the Joule heat dissipated in the TES at that bias point and thetotal heat capacity C is the sum of the C
ABS (1.18 pJ/K) and the C
TES (0.02 pJ/K) [7].In Fig. 3 are shown the derived parameters. The values of the key parameters α and β in theAC are smaller and the shape seems to be smoothed out compared to the DC. Looking atthe curve of α in the AC, we can see of course the large peak around 68% of the transitionbut only an hint of the second one around 58%. This small peak is a bit more evident inthe shape of β . In the DC those are clearly much more evident and their values are larger asalready said. It is clear that there is a shift between AC and DC and this confirm our previ-ous guess about the presence of a small magnetic field in the DC setup. It has been alreadydemonstrated [10] that any peak of α and β can appear at different R/R N for different ap-plied magnetic fields in TESs with normal metal structures. Moreover, we can remark thatthe behaviour of α and β between AC and DC bias start diverging at lower bias point in thetransition.One can use the parameters obtained from the complex impedance to model the detectornoise. In Fig. 4 the measured noise spectra are shown at 44% and 45% of R N for AC andDC bias, respectively and the results from the model, using the AC bias parameters, areover-plotted. The model noise contributions are: phonon noise, TES Johnson noise, excessJohnson noise and SQUID noise. Those noise sources describe very well the noise observedat frequencies higher than 100 Hz and it looks slightly lower in the DC case. In the frequencyrange where the Johnson noise is dominant there is an excess noise, which is quantified as Mtimes the Johnson noise and introduced by this factor M [11]. Considering the predicted en-ergy resolution 2.355 r b T × C α q nF(1 + 2 β )(1 + M ) for the same bias point of the noisespectra shown in Fig. 4, we obtain 3.3 eV in AC against the 2.8 eV in DC. This discrep-ancy of ∼
10% could be larger when the behaviour of the detector under AC begins to differfrom the DC. This is in line with the assumption that detectors with high saturation powerand high normal resistance, or biased at higher bias points in the transition, show generallysmall or negligible Josephson current under ac bias, mitigating the weak-link behaviour inac-biased detectors.
E. Taralli et al. a AC-55 mKDC-55 mKAC-75 mKDC-75 mK b t e ff ( m s ) R/Rn 0 20 40 0.2 0.4 0.6 0.8 1(d) L oop G a i n R/Rn
Fig. 3: Parameters derived from impedance measurements: α ( a ), β ( b ), τ eff ( c ), and loopgain L ( d ) ( Open symbols ) as a function of the bias point expressed as R/R N . Lines serveno other purpose to guide the eye. (Color figure online.) C u rr e n t N o i s e ( p A / (cid:214) H z ) f (Hz) Exp Noise 44 % of R N - (AC bias)Exp Noise 45 % of R N - (DC bias)Total Noise 44 % of R N with M=1.22 Fig. 4: Noise measurements of TES20 at 44% and 45% of R N for AC and DC bias, respec-tively. Noise contributions: SQUID noise ( yellow dot-dot-dash line ), Johnson noise ( purpledashed line ), Excess Johnson noise ( green dot dash line ) and phonon noise ( blue dot-longdash line ). The cutoff at frequencies above 10 kHz in AC bias is due to the use of a band-passfilter to avoid interference with the neighbouring pixel. (Color figure online.) We have performed a comparison of the performance of a single pixel TES microcalorime-ter under DC and AC bias (f c = 3.5 MHz), by means of IV curves, complex impedance andnoise measurements. The behaviour of the detector under AC begins to differ from the DCat working points lower in the transition. In the AC, values of α and β are lower and theirshape appear to be smoothed out, especially in presence of peaks. A better analysis, includ-ing the error estimation on the parameters obtained from complex impedance data has beenpresented to guarantee a fair comparison.SRON is currently developing high aspect ratio TiAu TES, with a thicker bilayer and with- C/DC characterization of a Ti/Au TES with Au/Bi absorber for X-ray detection 7 out normal metal structures, high normal resistance and high saturation power to accomplishthe goal of having high performance detectors under AC bias[12]. In the future, we are plan-ning for these new pixels extensively AC and DC bias experiments including X-ray energyresolution measurement at different bias points to better understand the interaction betweenthese devices and the voltage bias readout system.
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-Energy As-trophysics Domain) project with grant agreement number 654215. CSIC work financed by the Spanish Min-isterio de Ciencia, Innovaci´on y Universidades-MICINN (projects ESP2016-76683-C3-2-R and RTI2018-096686-B-C22). Personnel from ICMAB acknowledge financial support from MINECO, through the ’SeveroOchoa
Acknowledgements ’ Programme for Centres of Excellence in R&D (SEV- 2015-0496).
References
1. F. Pajot, D. Barret, T. Lam-Trong, J.-W. den Herder et al.,
J. Low Temp. Phys. , 901 (2018). DOI:https://doi.org/10.1007/s10909-018-1904-52. European Space Agency (ESA),”ATHENA Mission Summary,”Available:http://sci.esa.int/athena/59896-mission-summary.3. W. B. Doriese et al.,
J. Low Temp. Phys. , 389 (2016). DOI: https://doi.org/10.1007/s10909-015-1373-z4. L. Gottardi, J. van der Kuur, P. A. J. de Korte, R. Den Hartog, B. Dirks, M. Popescu, H. F. C. Hoevers, M.Bruijn, M. Parra Borderias, and Y. Takei,
AIP Conference Proceedings . DOI: 10.1063/1.3292399.5. P. Khosropanah, E. Taralli, L. Gottardi, K. Nagayoshi, M. Ridder, M. Brujin, and J.-R. Gao,
Proceedingsvolume 9144 SPIE Atronomical Telescopes + Insrumentation, Space Telescopes and Instrumentation2018: Ultraviolet to Gamma Ray , Austin, Texas, USA, 10-15 June 2018, edited by J.-W. A. den Herder,K. Nakazawa, and S. Nikzad.6. L. Gottardi, J. van der Kuur, M. Bruijn, A. van der Linden, M. Kiviranta, H. Akamatsu, R. den Hartog,and K. Ravensberg,
J. Low Temp. Phys.
94, 370, (2018). DOI: https://doi.org/10.1007/s10909-018-2085-y7. E. Taralli, P. Khosropanah, L. Gottardi, K. Nagayoshi, M. L. Ridder, M. P. Bruijn, and J. R. Gao,
AIPAdvances , 045324, (2019), DOI:10.1063/1.50897398. C. Pobes, L. F`abrega, A. Cam´on, N. Casa˜n-Pastor, P. Strichovanec, J. Ses´e, J. Moral-Vico, R. M. J´audenes Calleja, IEEE Trans. Appl. Supercond. , 2101505, (2017), DOI:10.1109/TASC.2016.26373379. D. J. C. MacKay, Information Theory, Inference, and Learning Algorithms.
Cambridge, U.K.: Cam-bridge Univ. Press, 2003.10. S.J. Smith, J. S. Adams, C. N. Bailey, S. R. Bandler, S. E. Busch, J. A.Chervenak, M. E. Eckart, F. M.Finkbeiner, C. A. Kilbourne, R. L. Kelley, S.-J. Lee, J.- P. Porst, F. S. Porter, and J. E. Sadleir,
J. Appl.Phys.