Progress in the development of TES microcalorimeter detectors suitable for neutrino mass measurement
A. Giachero, B. Alpert, D.T. Becker, D.A. Bennett, M. Borghesi, M. De Gerone, M. Faverzani, M. Fedkevych, E. Ferri, G. Gallucci, J.D. Gard, F. Gatti, G.C. Hilton, J.A.B. Mates, A. Nucciotti, G. Pessina, A. Puiu, C.D. Reintsema, D.R. Schmidt, D.S. Swetz, J.N. Ullom, L.R. Vale
11 Progress in the development of TESmicrocalorimeter detectors suitable for neutrinomass measurement
A. Giachero, B. Alpert, D.T. Becker, D.A. Bennett, M. Borghesi, M. De Gerone, M. Faverzani, M. Fedkevych,E. Ferri, G. Gallucci, J.D. Gard, F. Gatti, G.C. Hilton, J.A.B. Mates, A. Nucciotti, G. Pessina, A. Puiu,C.D. Reintsema, D.R. Schmidt, D.S. Swetz, J.N. Ullom, L.R. Vale
Abstract —The HOLMES experiment will perform a precisecalorimetric measurement of the end point of the ElectronCapture (EC) decay spectrum of
Ho in order to extractinformation on neutrino mass with a sensitivity below 2 eV. In itsfinal configuration, HOLMES will deploy 1000 detectors of lowtemperature microcalorimeters with implanted
Ho nuclei. Thebaseline sensors for HOLMES are Mo/Cu TESs (Transition EdgeSensors) on SiN x membrane with gold absorbers. Considering thelarge number of pixels and an event rate of about 300 Hz/pixel, alarge multiplexing factor and a large bandwidth are needed. Tofulfill this requirement, HOLMES will exploit recent advanceson microwave multiplexing. In this contribution we present thestatus of the activities in development, the performances of thedeveloped microwave-multiplexed readout system, and the resultsobtained with the detectors specifically designed for HOLMESin terms of noise, time and energy resolutions. I. I
NTRODUCTION
The experiments on the neutrino flavor oscillation phe-nomenon provide indisputably evidence of the non-vanishingnature of the neutrino mass [1]. Measuring the neutrino massis one of the most urgent issues in particle physics. Severalapproaches are taken in order to assess this parameters, suchas cosmological surveys [2], searches for neutrinoless doublebeta decay [3] and the studies of the end point of beta orelectron capture (EC) decays [4].An interesting approach suitable for a direct determinationof the neutrino mass, is the calorimetric measurement of theenergy released in the EC decay of
Ho, as proposed byDe Rujula and Lusignoli in 1982 [5]. The proximity of the Q -value ( Q EC = 2 . keV [6]) of the decay to the energy ofthe M1 shell enhances the number of events close to the endpoint, where the effect of the non-vanishing neutrino mass is A. Giachero, M. Borghesi, M. Faverzani, E. Ferri and A. Nucciotti are withDipartimento di Fisica, Universit`a di Milano-Bicocca and INFN - Sezione diMilano Bicocca, Milano, I-20126, Italy (e-mail: [email protected]).B. Alpert, D.T. Becker, D.A. Bennett, J.D. Gard, J.A.B. Mates,C.D. Reintsema, D.R. Schmidt, D.S. Swetz, J.N. Ullom and L.R. Vale arewith National Institute of Standards and Technology, Boulder, CO 80305,USA.M. De Gerone, M. Fedkevych and G. Gallucci are with INFN - Sezione diGenova, Genova I-16146 - ItalyF. Gatti, is with Dipartimento di Fisica, Universit`a di Genova and INFN -Sezione di Genova, Genova I-16146 - ItalyG. Pessina is with INFN - Sezione di Milano Bicocca, Milano, I-20126,ItalyA. Puiu is with Gran Sasso Science Institute (GSSI), I-67100 L’Aquila,Italy more easily measurable. The experimental approach suggestedto measure the electrons emitted in the
Ho decay wasfrom the start calorimetric, but yet only in the last decadethe technological improvements in low-temperature detectorsreached the demanding performances needed for measuringthe neutrino mass with sensitivity below ∼ eV.In a calorimetric measurement, the energy released in thedecay process is entirely contained into the detector, exceptfor the fraction taken away by the neutrino. This approacheliminates both the issues connected to the use of an externalsource and the systematic uncertainties arising from decays onexcited final states. The most suitable detectors for this type ofmeasurement are low temperature thermal detectors [7], [8],where all the energy released into an absorber is convertedinto a temperature increase that can be measured by a sensitivethermometer directly coupled to the absorber.The HOLMES experiment [9] aims to perform a calori-metric measurement of the energy released in the electroncapture decay of Ho. The main goal is to reach a neutrinomass statistical sensitivity below 2 eV proving the potentialand the scalability of this technique for a future megapixelexperiment. In order to reach this sensitivity HOLMES willcollect about · events by deploying an array of 1024detectors. The total amount of implanted Ho nuclei willbe about . · , equivalent to 18 µ g for a total activityof 300 dec/s/pixel. The target for the instrumental energyand time resolutions are around 3-5 eV FWHM at 2.6 keVand 20 µ s respectively. Microcalorimeters based on TransitionEdge Sensors (TESs) can easily match these requirements.Considering the large number of pixels and the high eventrate, a large multiplexing factor and a large bandwidth areneeded. To fulfill these requirements, HOLMES will exploitrecent advances on microwave multiplexing ( µ mux) [10].II. TES DETECTORS DESIGN
HOLMES uses TES microcalorimeters coupled to goldabsorbers designed for soft X-ray spectroscopy with fastresponse and high energy resolution. A single TES is a × µ m Mo/Cu bilayer, designed to have a transitiontemperature of T c = 100 mK. A × µ m gold absorberis placed aside the TES to avoid proximity effect between thegold and the sensor itself ( side-car design, Fig. 1, bottom). Theabsorber thickness is µ m in order to provide full containment a r X i v : . [ phy s i c s . i n s - d e t ] J a n Single Pixel Cu/MoTES µ m Absorber . mm Fig. 1. (top) The HOLMES chip featuring 63 TESs each; (bottom) Opticalmicroscope image of the HOLMES Mo/Cu TES with Au absorber placedalongside (side-car design). of the 99.99% (96.73%) of the most energetic electrons(photons) produced in the decay [11]. The entire structureis suspended on 500 nm thick low-stress SiN x membranepreventing phonons to escape into the silicon substrate.The requirement of high pile-up discrimination ability setsstrict constraints on the detector response. At first order, thedecay time is set by the ratio between the thermal capacity C of the absorber and the thermal conductance G towards thesilicon substrate that acts as a constant temperature thermalbath. Since the heat capacity is constrained by the dimensionsof the absorber, for the full containment of the released energy,the only way to decrease the pulse decay time is to increase G . The thermal conductance G is increased by the addition ofa thermal radiating perimeter that increases the conductance inthis 2D geometry [12]. This thermalizing perimeter increasesthe thermal conductance G with negligible increase in the totalheat capacity. With this design G can be tuned from 40 pW/Kup to 1 nW/K. The target thermal conductance for HOLMESis around 600 pW/K, value that guarantees a total pulse timebelow 200 µ s and hence a very short dead time .This pixel configuration ensures an energy resolution betterthan 5 eV at 5.9 keV[13], and an exponential pulse rise time inthe range from 10 to 20 µ s [14]. With this rise time and witha sampling rate of at least 500 kHz, it is possible to obtaina time resolution better than 3 µ s exploiting discriminationalgorithms based on singular value decomposition [15] orWiener filtering [16].The TES pixels are closely packed in a × linearsub-arrays (Fig. 1, top). The design aims to minimize thesignal bandwidth limitations due to stray self-inductance ofthe readout leads; to minimize the signal cross talk due tomutual inductance between readout lines; and to maximizethe geometrical filling for an optimal implantation efficiency.These × linear sub-arrays are developed in two versionsvery similar but with minor differences in order to match adifferent procedures to release the membrane: Silicon KOHanisotropic wet etching and Silicon Deep Reactive Ion Etching(DRIE). In this latter case, the detectors are placed in a morecompact disposition, increasing in this way the geometrical efficiency of the Ho implantation. Each HOLMES sub-array includes 64 TESs and 16 of them will cover all the1024 foreseen pixels.III. TES
DETECTORS FABRICATION
The TES detectors of HOLMES have to undergo twodifferent fabrication steps [17]. The detectors are producedat NIST, where the TES, the copper structure and the firstpart ( µ m thickness) of the absorber are evaporated atop ofthe SiN membrane. All the detectors but the absorbers, arecovered in photoresist and shipped to INFN Genova, wherethe Holmium implantation is made. Holmium is extracted froma metallic target, ionised, accelerated towards the TES arrayand implanted in the uncovered absorbers [18], [19]. Thisimplantation system is composed of an argon penning sputter-based ion source containing Ho, an acceleration section,dipole magnet mass analyzer to select only the isotope ofinterest, a focusing electrostatic triplet, to refocus the beam,and a magnetic XY scanning stage, to defect the beam inthe plane perpendicular to its motion. At the end of theimplanter the last section is a vacuum chamber (hereafterTarget Chamber) that allows a simultaneous Gold evaporationto control the
Ho concentration and to deposit a final Goldlayer to protect the
Ho from oxidizing.The ion implantation system is under commissioning phasein a dedicated laboratory at the INFN Unit of Genoa (Italy).Tests are currently performed using a copper sputter targetinstead of a Ho one, with the goal to study the relationshipbetween the beam current and the ion source parameters. Thefinal sputter target will be a sintered compound composed ofHo (5%) in a metallic mixture of Ti (36%), Ni (41%), and Sn(18%) [18]. This new sputtering target including Ho will betested as soon as the current tests with Copper target will befinished.The Target Chamber has been developed and tested atMilano Bicocca University. The Target Chamber is equippedwith an ion beam assisted sputtering system which willallow a simultaneous Gold deposition to control the
Hoconcentration and compensate the Gold sputtering caused bythe ion implantation, and to deposit the final Gold layer tocomplete the
Ho embedding. The chamber was extensivelytested during the past years, evolving from one ion beammicrowave source to four ones to increase the deposition rateand improve the uniformity and isotropy of the depositedlayer. The uniformity were checked by sputtering Au for ∼ hours on a Si slab × cm with a drilled maskwith × holes on top. The thickness in the center of thecircles were measured with a profilometer finding an averagevalue of d = 865 ± nm (Fig. 2). Considering a measureddeposition rate greater than 50 nm/hours a Gold thickness of1 µ m can be deposited in around 20 hours. After the depositionthe lift-off for the samples is done in hot acetone obtaining a × µ m Gold absorber.The final step of the detector arrays fabrication procedure isthe membrane release. At the Milano-Bicocca University thesilicon anisotropic wet etching using KOH were successfullydeveloped. This is now the baseline for the two-step fabrication
Fig. 2. Uniformity of the sputtered Gold inside the developed TargetChamber. of the first arrays. The design of these sub-arrays is tuned forthe potassium hydroxide (KOH) wet etching and this limitsthe packing of the pixels, because of the sloped side of theopenings in the silicon wafer. This design is not optimalfor the implantation efficiency which could be improved byabout a factor two with a more packed × design. TheDeep Reactive Ion Etching (DRIE) process, which providessteep side openings, will allow this highest pixel density. Thisis currently under study by exploiting an external facility(Trustech, Turin, Italy). After an R&D phase on dummysamples, tests on real arrays will be performed during 2021.Two HOLMES sub-arrays produced with the complete two-step fabrication presented above, though still without the Ho implantation, and with the membrane released by KOHwet etching, were measured during the 2020. The characteri-zation results are presented in the next sections.IV. T
HE MULTIPLEXED READ OUT
The readout for the HOLMES detector array is implementedby exploiting the microwave multiplexing scheme [20]. Eachdetector is coupled to an rf-SQUID (Superconducting Quan-tum Interference Devices), which in turn is inductively coupledto a λ/ resonator [10]. Changes in the TES resistancemodulate the current inside the TES circuit. This currentvariation is then translated into a change in magnetic flux inthe rf-SQUID, which in response varies the resonant frequencyof coupled resonators. By placing in the same multiplexer chipmany microresonators with different resonant frequencies, inthe GHz range, and by coupling all the microresonators to acommon feedline, it is possible to perform the read out of mul-tiple detectors with a single line. The core of the microwavemultiplexing read out is the multiplexer chip. HOLMES usesmultiplexer chip ( µ mux17) developed at NIST and composedby 33 quarter-wave coplanar waveguide covering 500 MHzin the 4-8 GHz frequency range. Each resonator is designedto have a resonance with bandwidth of 2 MHz for reading outdemodulated signals with a sampling frequency up to 500 kS/s.The spacing between adjacent resonances is set to 14 MHz, toassure a negligible cross-talk. The microwave read out and multiplexing system developedfor HOLMES is based on electronics developed for the readoutof microwave kinetic inductance detectors (MKIDs)[21], [22].A comb of signals with frequencies tuned to the resonantfrequencies of the resonators is digitally generated by a digitalprocessing board, then digital-to-analog converted (DAC) atfrequencies of the order of few MHz (base-band, 0-512 MHz)and finally up-converted in the RF band (4-8 GHz) by mixingthem with a GHz signal (local oscillator—LO). At the outputof the resonator chip, these signals are amplified at 4 Kby a low-noise HEMT amplifier, down-converted at roomtemperature by mixing them with the LO, and then acquired byan analog-to-digital converter (ADC). The resulted digitizedwaveforms are finally channelized (i.e. individual channelsignal recovery) and processed by the digital processing boardfor real-time reconstruction of the TES signals. This read outchain covers one multiplexer chip with 32 resonators overa 500 MHz band. One resonator is intentionally left dark toevaluate the SQUID noise performance without the detectorcontribution.A ROACH2 (Reconfigurable OpenArchitecture ComputingHardware) board [23] is used to generate the base-bandcomb signal and to perform the channelization and the digitalprocessing routines for the real-time demodulation. A semi-custom commercial board, designed to meet HOLMES readoutrequirements, is used to perform the up-and down-conversion.This board provides a conversion loss of around 7 dB, valuehigh enough to drive and read out 32 resonators withoutspoiling their performances. The typical RF bandwidth for theHEMT amplifiers selected for HOLMES is 4 GHz in the rangefrom 4 to 8 GHz [24]. This means that a single HEMT canamplify the signals coming from a series 8 multiplexer chipseach one with different frequency bands covering a total of256 pixel per HEMT. To cover the total 1024 pixels expectedfor HOLMES, 4 HEMT amplifiers are needed for a total of32 multiplexer chips, as well as the same number of ROACH2boards+Up/Down conversion boards, since the readout chaindescribed here covers one multiplexer chip.The ROACH2 FPGA firmware used to implement thedigital signal processing is based on the one developed atNIST for X- and gamma-ray spectroscopy [25], [26], withsome modifications needed to handle resonators with largerbandwidth. A preliminary 16-channel version exploiting onlyhalf of the available DACs/ADCs bandwidth is fully workingand it was used in different characterization runs during 2020.The definitive 32-channel version is currently under debuggingphase and it will be used for the next measurements. Thedevelopment of a 64-channel system based on two ROACH2boards and on two remotely programmable semi-commercialup- and down-conversion circuitry is also currently in progress.This setup will be fundamental for the read out of the first mi-crocalorimeter × sub-array with Ho nuclei implanted.V. T
HE DETECTOR PERFORMANCES
Two preliminary × prototype arrays were entirely pro-duced at NIST with the release of the membrane done witha DRIE process. These array presented slight variations in the pixel perimeter/absorber designs. The goal was to studythe different detector responses selecting the best performingsolution for the HOLMES requirements, in terms of energyand time resolutions. The results obtained in these testsallowed to identify the optimal detector. The chosen detectorachieved an energy resolution of (4 . ± . eV on the chlorineK α line, at 2.6 keV, obtained with an exponential rise time anddecay of µ s and µ s, respectively [27].The selected pixel perimeter/absorber were implemented intwo × sub-arrays that were produced in 2019 following forthe first time the entire HOLMES fabrication process presentedin Sec. III, with the exception of the implantation of the Ho.One array presents all the pixels identical to the one selectedin the previous productions while the second one presentsfew differences in the sensor/absorber configuration for furthertuning of the pixel design. At this stage, the release of themembrane was done with a KOH wet etching. These twoarrays have been tested during the 2020 by using the readout system presented in Sec. IV. They were installed in acopper holder designed to host 128 channel (2 sub-arrays and 4multiplexer chips). 8 holders will cover the entire HOLMES inits final configuration (1024 channels). A fluorescence sourcewas employed to test the detectors: this was composed of aprimary Fe source faced several targets obtaining their X-ray characteristic emission lines. The total count rate per pixelwas around 0.5 Hz. During the characterization runs performedin 2020 only two multiplexer chips were mounted inside theholder and only half of the pixels were available for acquisition(32+32). Moreover, only 16 pixels per chip were read out atthe same time due to the limited bandwidth of the used 16-channel version firmware.The obtained results for the array with all the pixelsidentical are here presented. The read out noise resulted inthe (19 − pA / √ Hz range (Fig. 3, top), compatible withthe level obtained in previous measurements [27], [20] andin other applications [25]. Few channel presented an highernoise (around pA / √ Hz) due to problematic resonators andto no optimal rf-SQUID oscillations. The detectors showedFWMH energy resolutions at 5.9 keV within the − keVrange, with a best performing detector having a resolution of (4 . ± . eV (Fig. 3, bottom). The measured pulse risetime resulted around µ s, that despite being higher than forpreviously measured detectors, it still matches the HOLMESrequirements. Since the rise time, at the first order, is set bythe electrical cutoff of L/R , where L is the stray inductanceand R is the resistance of the sensor at the working point,with same L and R this larger value may be probably due tothe different parasitic impedance in the TES biasing circuit.The tested pixels showed also a longer decay time ( µ svs. µ s) with a large spread. Extrapolating the thermalconductance G through the measurement of the IV curves atdifferent temperatures a mean value of 400 pW/K was found.This value resulted lower than the expected one of 600 pW/K,and presented also a dispersion of around ± pW/K. Thecause of this lower G and of its variability is still object ofinvestigation but might be related to the wet etching process.Switching to DRIE might turn out to be the only safe andreliable technique to release the membranes with the grade of Frequency[Hz]10 N o i s e [ p A / √ H z ] Energy [eV]5870 5888 5905 5922 5940 C oun t s / b i n R e s i du a l s √ Hz33.3 pA/ √ Hz FitData
Fig. 3. (top) Example of noise power spectral density measured for 13acquired detectors; (bottom) Separation of the K α and K α of the Mn,obtained with best performing detector. reproducibility required by HOLMES. Array processed withDRIE will be tested and characterized during 2021. In parallel,the wet etching procedure will be refined in order to improvethe uniformity. Lastly, the pulse amplitude estimated at 5.9 keVresulted compatible with the ones obtained in the previousproductions. Since this amplitude depends on the ∆ E/C ratio,where C is the absorber heat capacity, this means that thedeveloped Target Chamber is properly tuned and the depositedthickness is compatible with the expected one.VI. C ONCLUSION
Most of the HOLMES detector production processes arebeing set up, while few of them need some optimizations. Torelease the membrane either a wet (KOH) or a DRIE etchingprocess can be used to remove the silicon underneath themembrane. The wet process is undergoing the final refinementin order to find the most suitable parameters for the detectorsof HOLMES. The target chamber and the gold depositionsystem is functioning properly and it is currently being set-up and the end of the ion implanter beam. To date, a testof the implanter with a target composed of copper has beenperformed. Subsequent tests will be performed with targetsmade of natural holmium and, eventually, of
Ho to producethe proper HOLMES detectors. A first measurement with thefirst × sub-array implanted with Ho, and read out witha 64-channel multiplexing system, is expected to start during2021. A
CKNOWLEDGMENT
This work was supported by the European Research Coun-cil (FP7/2007-2013) under Grant Agreement HOLMES no.
EFERENCES[1] I. Esteban, M. Gonzalez-Garcia, M. Maltoni, T. Schwetz, and A. Zhou,“The fate of hints: updated global analysis of three-flavor neutrinooscillations,”
JHEP , vol. 09, p. 178, 2020.[2] M. Lattanzi and M. Gerbino, “Status of neutrino properties and futureprospects - Cosmological and astrophysical constraints,”
Front. in Phys. ,vol. 5, p. 70, 2018.[3] M. J. Dolinski, A. W. Poon, and W. Rodejohann, “Neutrinoless Double-Beta Decay: Status and Prospects,”
Ann. Rev. Nucl. Part. Sci. , vol. 69,pp. 219–251, 2019.[4] G. Drexlin, V. Hannen, S. Mertens, and C. Weinheimer, “Current directneutrino mass experiments,”
Adv. High Energy Phys. , vol. 2013, p.293986, 2013.[5] A. De Rujula and M. Lusignoli, “Calorimetric Measurements of
HoDecay as Tools to Determine the Electron Neutrino Mass,”
Phys. Lett.B , vol. 118, p. 429, 1982.[6] S. Eliseev et al. , “Direct Measurement of the Mass Difference of
Ho and
Dy Solves the Q -Value Puzzle for the Neutrino MassDetermination,” Phys. Rev. Lett. , vol. 115, no. 6, p. 062501, 2015.[7] S. H. Moseley, J. C. Mather, and D. McCammon, “Thermal detectorsas x-ray spectrometers,”
Journal of Applied Physics , vol. 56, no. 5, pp.1257–1262, 1984.[8] E. Fiorini and T. Niinikoski, “Low Temperature Calorimetry for RareDecays,”
Nucl. Instrum. Meth. A , vol. 224, p. 83, 1984.[9] B. Alpert et al. , “HOLMES - The Electron Capture Decay of
Hoto Measure the Electron Neutrino Mass with sub-eV sensitivity,”
Eur.Phys. J. C , vol. 75, no. 3, p. 112, 2015.[10] O. Noroozian et al. , “High-resolution gamma-ray spectroscopy with amicrowave-multiplexed transition-edge sensor array,”
Appl. Phys. Lett. ,vol. 103, p. 202602, 2013.[11] A. Nucciotti et al. , “Status of the HOLMES Experiment to DirectlyMeasure the Neutrino Mass,”
J. Low Temp. Phys. , vol. 193, no. 5-6, pp.1137–1145, 2018.[12] J. Hays-Wehle, D. Schmidt, J. Ullom, and D. Swetz, “Thermal Con-ductance Engineering for High-Speed TES Microcalorimeters,”
J. LowTemp. Phys. , vol. 184, no. 1-2, pp. 492–497, 2016.[13] A. Puiu et al. , “Transition-Edge Sensors for HOLMES,”
J. Low Temp.Phys. , vol. 199, no. 3-4, pp. 716–722, 2019.[14] ——, “Updates on the Transition-Edge Sensors and Multiplexed Read-out for HOLMES,”
J. Low Temp. Phys. , vol. 193, no. 5-6, pp. 1167–1173, 2018.[15] B. Alpert et al. , “Algorithms for Identification of Nearly-CoincidentEvents in Calorimetric Sensors,”
J. Low Temp. Phys. , vol. 184, no. 1-2,pp. 263–273, 2016.[16] E. Ferri et al. , “Pile-Up Discrimination Algorithms for the HOLMESExperiment,”
J. Low Temp. Phys. , vol. 184, no. 1-2, pp. 405–411, 2016.[17] A. Orlando et al. , “Microfabrication of Transition-Edge Sensor Arrays ofMicrocalorimeters with
Ho for Direct Neutrino Mass Measurementswith HOLMES,”
J. Low Temp. Phys. , vol. 193, no. 5-6, pp. 771–776,2018.[18] M. De Gerone et al. , “
Ho distillation and implantation for theHOLMES experiment,”
Nucl. Instrum. Meth. A , vol. 936, pp. 220–221,2019.[19] G. Gallucci et al. , “Status of the HOLMES experiment to directlymeasure the electron neutrino mass with a calorimetric approach,”
PoS ,vol. NuFact2019, p. 106, 2020.[20] D. Becker et al. , “Working principle and demonstrator of microwave-multiplexing for the HOLMES experiment microcalorimeters,”
JINST ,vol. 14, no. 10, p. P10035, 2019.[21] R. Duan et al. , “An open-source readout for MKIDs,”
Proc. SPIE , vol.7741, pp. 568–577, 2010.[22] S. McHugh et al. , “A readout for large arrays of microwave kineticinductance detectors,”
Review of Scientific Instruments , vol. 83, no. 4,p. 044702, 2012.[23] J. Hickish et al. , “A Decade of Developing Radio-Astronomy In-strumentation using CASPER Open-Source Technology,”
Journal ofAstronomical Instrumentation , vol. 05, no. 04, p. 1641001, 2016.[24]
LNF-LNC4 8C, 4-8 GHz Cryogenic Low Noise Amplifier , Low NoiseFactory, rev. March 2017. [25] J. A. B. Mates et al. , “Simultaneous readout of 128 X-ray and gamma-ray transition-edge microcalorimeters using microwave SQUID multi-plexing,”
Applied Physics Letters , vol. 111, no. 6, p. 062601, 2017.[26] J. D. Gard et al. , “A Scalable Readout for Microwave SQUID Multiplex-ing of Transition-Edge Sensors,”
Journal of Low Temperature Physics ,vol. 193, no. 3, pp. 485–497, Nov 2018.[27] B. Alpert et al. , “High-resolution high-speed microwave-multiplexed lowtemperature microcalorimeters for the HOLMES experiment,”