betaFeedback

WorksPhysics and Astronomy

Lithographic integration of TES microcalorimeters with SQUID multiplexer circuits for large format spectrometers

Abstract

Arrays of hundreds or thousands of low temperature detectors have been deployed for many experiments, both bolometers for long wavelength applications and calorimeters for shorter wavelength applications. One challenge that is common to many of these arrays is the efficient use of focal plane area to achieve a large fill fraction of absorbers coupled to detectors. We are developing an integrated fabrication of soft X-ray transition edge sensors (TES) and microwave SQUID multiplexers (μμMUX) with the goal of maximizing the fill fraction of the focal plane area on a scale of many thousand pixel detectors. We will utilize lithographically defined high density interconnects to circumvent limitations in existing solutions that use wirebonds or flip-chip bonds. Here we report the first demonstration of combining TES and μμMUX processes into a single TES-System-on-a-Chip (TES-SoC) fabrication on a silicon wafer. The μμMUX SQUIDs and TES electrothermal feedback circuits are microfabricated first and protected with passivating SiO2_2, then the TES devices and TES-to-SQUID interconnects are fabricated, and finally the protective layer is removed before the fabrication of the microwave resonators. We show that the microwave SQUIDs are functional and have reasonable yield, and that we are able to read out the transition temperature of the connected TESs using those SQUIDs.

Introduction

X-ray spectrometers that utilize transition edge sensor (TES) microcalorimeter arrays offer an ideal combination of exceptional energy resolution and quantum efficiency. These spectrometers are well positioned to revolutionize multiple science applications including nondestructive nanoscale tomography [1], testing quantum electrodynamics using exotic atoms [2], ultrafast photochemistry [3], and soft X-ray spectroscopy at X-ray light sources [4]. Towards this goal, TES spectrometers are now deployed to various synchrotron beamlines including the National Synchrotron Light Source (I and II), the Advanced Photon Source, the Stanford Synchrotron Radiation Lightsource, SPring-8, and BESSY II.

NIST’s largest scale X-ray spectrometer utilizes a collection of microsnout submodules [1] [5], with each microsnout containing an array of 250 TES microcalorimeter detectors.

Each microsnout also contains four interface (IF) chips with bias and filtering circuits and four μ\muMUX chips [6]. In order to maximize the fill-factor for active detector area, the IF and readout electronics chips are mounted ninety degrees out of the detector plane. The TES detectors are connected to their individual readout channels via around-the-corner wire bonds. The signal from the μ\muMUX chips on multiple microsnouts can be read by one pair of input-output microwave cables.

While μ\muMUX technology has already enabled the current generation of compact and energy-efficient X-ray spectrometers with ∼1000\sim1000 pixels, further improvement in TES pixel density is desirable for the next generation of spectrometers. One prime application that will benefit from the enhanced TES pixel density is carbon catalysis using Resonant Inelastic X-ray Scattering (RIXS) where currently deployed conventional spectrometers require collection times of tens of minutes to an hour. To reduce RIXS collection time to a few seconds, we are developing high-resolution X-ray spectrometers with ∼10,000\sim10{,}000 TES pixels. Here we present an integrated fabrication of microwave SQUID multiplexers and soft x-ray TESs on the same wafer that provides a path towards high pixel density in these spectrometers.

As mentioned, our current X-ray spectrometer technology utilizes wirebonds to interconnect TES to their μ\muMUX based readout channels. The TES-to-SQUID wirebond pad pitch is 275 μ\mum and sets the upper limit on the achievable pixel density on each microsnout. In addition, interconnecting 10,000 TES pixels to their readout electronics would require the challenging completion of over 40,000 wirebonds. To address these limitations, we have developed a new TES-System-on-Chip (TES-SoC) architecture where TES detectors are integrated with their μ\muMUX readout electronics on a single monolithic wafer platform. We utilized lithographically defined TES-to-SQUID interconnects to eliminate the upper bound on the TES pixel density that is conventionally set by the wirebond pad pitch. The final design of our 10,000 pixel X-ray spectrometer will leverage TESs with overhanging absorbers to maximize photon collecting area [7] [8] [9]. We will combine TES-SoC with our previously demonstrated SOI-Flex technology [10] to fabricate TES-to-SQUID interconnects with the bias and readout components ninety degrees out of the detector focal plane. In this transaction, we will discuss our first demonstration of TES-SoC fabricated on bulk silicon where we studied effects of integration of TES detectors with μ\muMUX read-out electronics.

In the following sections, we will describe the microfabrication methods for the TES-SoC including in-process wafer level

Authors thank John Biesecker, Sandra Diez, and Shannon M. Duff of National Institute of Standards and Technology, Boulder, CO, USA for useful discussions.

Robinjeet Singh, Avirup Roy, Daniel Becker, Joel N. Ullom, and Joel C. Weber are with the University of Colorado, Boulder, CO, USA and the Quantum Sensors Division of the National Institute of Standards and Technology, Boulder, CO, USA. Johnathan D. Gard is with the University of Colorado, Boulder, CO, USA.

Mark W. Keller, John A. B. Mates, Kelsey M. Morgan, Galen C. O’Neil, Nathan J. Ortiz, Daniel R. Schmidt, Daniel S. Swetz, Leila R. Vale, and Michael Vissers are with the Quantum Sensors Division of the National Institute of Standards and Technology, Boulder, CO, USA. diagnostics that aid in detecting any drift in the critical current (IcI_{\mathrm{c}}) of the μ\muMUX circuits and tuning the superconducting critical temperature (TcT_{\mathrm{c}}) of the TES detectors. We will then present measurements of a complete TES-SoC integrated chip, including resonator quality factor, bandwidth, and response to magnetic flux.

Fabrication Details

Our μ\muMUX technology is based on niobium, aluminum, aluminum oxide, and niobium (Nb-Al/Alx_xOy_y-Nb) Superconductor-Insulator-Superconductor (SIS) tunnel Josephson junctions (JJ) [11]. We typically fabricate these μ\muMUX devices on a 76.2 mm intrinsic undoped Si wafer grown with a float-zone (FZ) process. We first grow a very thin layer of thermal SiO2_2 on top of the silicon wafer that acts as an etch stop for subsequent Nb processing.

We typically fabricate molybdenum-gold (MoAu) bilayer based TES microcalorimeter detectors [12] on a Si wafer with a 500 nanometer thick layer of super-low stress low-pressure-chemical-vapor-deposition (LPCVD) silicon nitride (SiNx_x). The Si handle underneath the TES electronics is selectively etched to form a SiNx_x membrane. Suspending the TES detectors on a SiNx_x membrane provides the thermal isolation required to set the time constant for the detector response.

The need for two different thicknesses of varied dielectric materials for TES and μ\muMUX devices, respectively, poses a challenge for integrated TES-SoC fabrication. To circumvent this challenge, we utilize a silicon on insulator (SOI) wafer platform for TES-SoC. As discussed for SOI-Flex in our previous work [10], we chose a 4 μ\mum thick silicon device layer and a thin (200 nm) buried oxide layer. The relatively thick device layer minimizes additional loss to the μ\muMUX electronics due to the buried oxide layer. A 25 nm thick layer of thermal oxide is still grown on the SOI wafer to match the existing μ\muMUX process. To thermally isolate the TES detectors in TES-SoC on SOI, our final design will suspend each detector on a membrane made from the silicon device layer instead of SiNx_x. The Si handle is selectively etched underneath the desired area to release the Si membrane using buried oxide as an etch stop. Further details of the silicon membrane design are discussed in A. Roy et al. work being published concurrently [13].

For the initial demonstration described in this transaction, we fabricated a TES-SoC on a bulk intrinsic silicon with 32 TES detectors lithographically interconnected to their bias and μ\muMUX readout components. The TES detectors being directly on thick silicon are not thermally suspended. With this demonstration, we tested the TES-SoC fabrication process, ensured proper working of all components at cryogenic temperatures, and demonstrated the first lithographic integration of TES detectors with μ\muMUX readout.

Readout Fabrication

The complete fabrication flow for TES-SoC on bulk silicon is depicted in Figure 1. We utilize both numeric and descriptive layer IDs. We start with a 76.2 mm (3 inch) diameter FZ silicon wafer with 25 nm of thermal SiO2_2. In L1, we deposit a Nb-Al/Alx_xOy_y-Nb SIS tri-layer on the whole wafer using sputter deposition and diffusive oxidation techniques. We first deposit 200 nm of niobium as base-electrode (BE) followed by ∼\sim 15 nm of aluminum (Al). The Al thickness is selected to minimize any pinhole defects that may lead to shorted junctions. Once BE and Al are deposited, the wafer is transferred without breaking vacuum into an oxidation chamber to form an insulating Alx_xOy_y layer. We adjust the depth of oxidation and hence the critical current density (JcJ_c) of our SIS junctions using Eq. (1).

Figure 1. Layer schematic (not to scale) showing the microfabrication steps for TES-SoC on bulk Si. First, Nb-Al/Alx_xOy_y-Nb SIS trilayer is grown as L1. L2-L4 are etch steps for Nb counter-electrode (CE), aluminum (Al), and Nb base-electrode (BE) layers of SIS trilayers, respectively. Pd0.53_{0.53}Au0.47_{0.47} shunt resistors are deposited as L5 using liftoff process. L6 and L7 include deposition of passivating PECVD SiO2_2 (I1) and etching into I1 to open vias. Subsequent L8 layer is the deposition and etching of Nb (W1) that forms interconnects and termination wiring circuits using I1 vias. For layers L9 and L10, first I1 is removed in areas where microwave resonators, TES detectors, and TES-to-SQUID interconnects will be fabricated, then another layer of passivating SiO2_2 (I2) is deposited to protect μ\muMUX electronics. In L11, I2 is removed from areas where TES detectors and TES-to-SQUID interconnects will be fabricated, then a thin layer of molybdenum (Mo) is sputter-deposited, patterned, and etched using wet chemical bath. L12 is e-beam liftoff deposition of gold (Au) completing our MoAu bilayer TES detectors. L13 and L14 include RIE etching of vias into I2 and deposition of liftoff Nb to form TES-to-SQUID wiring interconnects. As final steps (L15 and L16), I2 and BE niobium are etched to form microwave resonators. The wafer is then diced into individual dies (step not shown in this schematic)

Jc≈ηE−0.51,E=P×t,(1)J_c \approx\eta E^{-0.51}, \\ E = P \times t, \tag*{(1)}

where the pre-factor η\eta is specific to our oxidation chamber and the oxidation exposure, EE, is the product of the oxygen pressure, PP, and the oxidation duration, tt. For μ\muMUX devices, PP is set to 1200 Pa (9 Torr) and we achieve Jc≈0.8J_c \approx0.8 μ\muA/μ\mum2^2 with a duration of 312 min. This corresponds to an estimated thickness of Alx_xOy_y between 1 nm and 2 nm, calculated using WKB approximation for a rectangular-barrier tunneling model [14].

After the oxidation of Al, again without breaking the vacuum, we move the wafer back into the process chamber for deposition of 120 nm of niobium counter-electrode (CE), completing our SIS tri-layer. We want the room temperature residual stress of the SIS tri-layer to be overall compressive to compensate for the change as metal layers contract at cryogenic temperatures [15]. We deposit Al at a set chamber pressure of 0.67 Pascal and adjust the chamber pressure for the niobium BE and CE layers to yield the desired film stress.

After L1, we lithographically define the majority of our μ\muMUX circuits that were first detailed in [16]. These circuits consist of dissipationless gradiometric rf-SQUIDs, each connected to a quarter-wave microwave resonator, and filter circuits for the flux ramp and TES bias. The output of the microwave resonators is capacitively coupled to a common coplanar microwave feedline, with the coupling capacitance set by changing dimensions of the interdigi­tated pattern. In L2, we pattern and etch CE using step and repeat i-line optical lithography and a sloped etch using O2_2:SF6_6 in a parallel plate reactive-ion-etch (RIE) system. The designed area of CE squares is 6.25 μ\mum2^2 targeting the desired IcI_c of 5 μ\muA for our JJs. We use an oxygen plasma ash and solvent-based clean after each etch step to prepare the wafer for subsequent processing.

In L3, we pattern aluminum pillars that are etched using tetramethylammonium hydroxide (TMAH) based wet chemistry. The Al features are made wider than CE to avoid any short circuit of CE to BE due to aluminum undercut. L4 involves sloped etching of BE (BES) to define the JJ base-electrode, slotted washers for SQUID inductors, microwave feedline, and input filter circuits. For TES-SoC, BES also defines the interdigi­tated circuits for the TES bias shunt resistors and the Nyquist inductor filters. We use O2_2:CF4_4 based RIE chemistry that ensures a sloped etch profile to facilitate conformal coverage of the subsequent metal layers. We note that the coplanar waveguides defining the quarter-wave microwave resonators are designed to have vertical sidewalls and are etched as a separate processing step.

In subsequent L5, we deposit 135 nm of Pd0.53_{0.53}Au0.47_{0.47} as R1 to shunt the inductive input filters for the μ\muMUX SQUIDs and bias resistors for the TES detectors. We use liftoff photolithography and e-beam evaporation techniques. Thin layers of titanium are used before and after R1 to promote adhesion. Excess metal is removed using organic solvents and ultrasonic agitation. After R1, we deposit a capping layer of 350 nm thick plasma enhanced chemical vapor deposition (PECVD) SiO2_2 (II) as L6 to protect the SIS metals from the top niobium wiring (W1). For L7, we pattern and etch via holes into I1 (using O2_2:CHF3_3 RIE chemistry) to selectively interconnect underlying SIS layers using the W1 wiring layer, which consists of 400 nm of sputtered niobium. W1 is patterned and etched using O2_2:CF4_4 based RIE chemistry to complete L8. W1 layer forms CE-to-BE interconnects for our JJ, grounding straps over the microwave feedline to prevent waveguide slot-mode excitations, and complete input flux ramp and TES bias lines [16]. W1 also forms the inductor washers for the TES IF circuits.

JJ and SQUID Characterization

To monitor the microfabrication processes, we employ various in-process wafer-level diagnostics on test chips that are

Optical micrograph of a TES-SoC die showing MoAu bilayer based TES detectors, IF electronics, and μMUX readout integrated onto a single monolithic wafer platform

caption: Optical micrograph of a TES-SoC die showing MoAu bilayer based TES detectors, IF electronics, and μ\muMUX readout integrated onto a single monolithic wafer platform. The chip dimension are 10 mm ×\times 20 mm. TES detectors are fabricated directly on bulk silicon and are not suspended. (a), (b), (c), (d), and (e) are higher magnification optical micrographs of the TES detectors, TES-to-SQUID liftoff niobium interconnects, Nyquist inductors, shunt resistors, and the μ\muMUX resonators, respectively.

strategically distributed throughout the wafer. As a diagnostic for our μ\muMUX JJs, we track the room temperature resistance RnR_n of 200 test JJs in series and utilize the Ambegaokar-Baratoff approximation [17] to estimate IcI_c from RnR_n,

IcRn≈πΔ2e,(2)I_cR_n \approx\frac{\pi\Delta}{2e}, \tag*{(2)}

where ee is the elementary charge and Δ\Delta is the superconducting energy gap calculated from BCS theory. As shown in our previous work [18], the RnR_n measurements provide a reliable in-fab diagnostic to ensure our JJ performance is within the design threshold. For post-fabrication, cryogenic testing of our rf-SQUIDs, we track the quantity λ\lambda defined as the ratio of self-inductance (LsL_s) and Josephson-inductance (LJL_J) within the SQUID. For dissipationless rf-SQUIDs to respond monotonically to the applied flux, the λ\lambda is desired to be ∼0.33±20%\sim0.33 \pm20\%. The LsL_s of our μ\muMUX SQUIDs is designed to be 22.4 pH, setting the target per unit area RnR_n to be (68.8±14) Ω/μm2(68.8 \pm14)\ \Omega/\mu\mathrm{m}^2. The uncertainty accounts for wafer-to-wafer microfabrication variations.

In TES-SoC, we tracked the performance of test JJ arrays distributed at the center and corners of the wafer. Based on these measurements, we calculated the average RnR_n of TES-SoC JJs with area 6.25 μm2\mu\mathrm{m}^2 to be ∼362 Ω\sim362\ \Omega. This equates to per unit area room temperature resistance of ∼58 Ω/μm2\sim58\ \Omega/\mu\mathrm{m}^2, well within uncertainty due to variations in microfabrication.

TES Detectors Fabrication

Once the room temperature diagnostic of our JJ is confirmed to be within specification, we proceed to L9 where we etch the I1 at sites where TES MoAu bilayers, TES-to-SQUID interconnects, and microwave resonators will be formed. In L10, to protect the μ\muMUX and IF circuits from degradation during MoAu processing, we deposit 400 nm of PECVD SiO2_2 (I2). We then proceed to microfabricate MoAu bilayer TES detectors as described in [12]. To define TES detectors, we follow the hybrid additive-subtractive TES process where 45 nm of molybdenum is sputter-deposited, patterned, and then etched using a phosphoric acid and acetic acid-based commercial etchant solution. In L12, a 411 nm thick layer of gold is deposited on top of molybdenum features via e-beam evaporation and liftoff process after a 1 min Ar RF plasma clean to improve MoAu interface transparency. The whole wafer is then baked for 20 min in an ambient environment at 150 °C to anneal the MoAu bilayer and stabilize TES TcT_c and RnR_n from additional thermal processing. In order to accurately tune the target TcT_c of the fabricated TES, we perform wafer-level cryogenic TcT_c measurements as described in work by J. Weber et al. published concurrently with this transaction [19]. Additional gold is added or removed based on our TcT_c diagnostics to obtain the final target TcT_c. A thinner layer of 250 nm gold with a 2 nm Ti adhesion layer is then deposited as a separate e-beam evaporation and lift-off process to define the absorbers for our TES detectors. A final ambient environment anneal at 150 °C for 20 min is completed to ensure good thermal contact between all Au layers.

We note that in our current demonstration of TES-SoC, we removed both I2 and the thin thermal oxide from TES detector sites. MoAu bilayers were then directly deposited on top of the silicon substrate. We observed across-wafer TcT_c non-uniformities and broad superconducting-normal transitions in our TES detectors. We attribute these negative characteristics to molybdenum-silicide formation. In future designs, we will fabricate TES on top of I2 SiO2_2.

TES-to-SQUID Interconnects

After the MoAu bilayer process, in L13, we etch via holes into the I2 layer to expose W1 for TES-to-SQUID niobium interconnects defined by L14 where 500 nm of sputtered Nb is deposited by lift-off process. During this step, the detectors and other components are protected by photoresist, whereas doing this step with a niobium etch process would require an additional passivating SiO2_2 layer that would complicate TES-SoC integration.

Microwave Resonator Fabrication

Once the MoAu bilayers, gold absorbers, and TES-to-SQUID interconnects are fabricated, we check the RnR_n of the test structures to track any unexpected changes due to additional processing. In L15, we pattern and etch the I2 layer to expose BE niobium at sites for microwave resonators that are patterned and etched in L16 using SF6_6 chemistry.

TES-SoC Die Release

In the current demonstration of TES-SoC on bulk Si, individual dies are released using mechanical dicing. Fig. 2 shows a micrograph of the diced TES-SoC die. The TES detectors, located on bulk silicon with no thermal isolation, were not expected to be suitable for X-ray spectrum or pulse measurements. The final design of TES-SoC on SOI will be released using deep reactive ion etch (DRIE) where the Si device layer will be first patterned and etched to define the chip boundary. We will then selectively etch the Si handle from the backside using DRIE to simultaneously suspend TES detectors on Si membranes and release the individual dies.

Results

After fabrication and singulation of individual TES-SoC on bulk Si dies, we carried out cryogenic testing to confirm the final TcT_\mathrm{c} of the fabricated TES detectors, response of the rf-SQUIDs to the input flux, and the quality factors of the microwave resonators.

Since the TES detectors were fabricated directly on the silicon and were not suspended on thin SiN or Si membranes, their time constant was too fast and current too large to be read out well with these μ\muMUX devices. We measured an average transition temperature of (141±3)(141 \pm3) mK for several TES-SoC detectors by varying the bath temperature in small increments.

The IcI_\mathrm{c} of the TES-to-SQUID sputtered niobium was tracked on a separate test wafer and was measured to be much higher than the IcI_\mathrm{c} of the molybdenum for wiring traces of similar dimensions. We measured IcI_\mathrm{c} greater than 80 mA for liftoff niobium wiring traces of varying lengths and widths. The response of the TES detectors and their transition temperatures were measured on the integrated TES-SoC to confirm that TES-to-SQUID interconnects were functioning as designed.

We measured the key design parameters of the μ\muMUX devices and found that they matched the design values well and had a relatively high yield of above 96%. Figure 3 shows shifts in the peak-to-peak frequencies of μ\muMUX resonators as a function of applied magnetic flux. We observed an average of 0.43 MHz peak-to-peak frequency shifts in the resonators. We calculate the average mutual inductance between the input circuits and the rf-SQUIDs (MinM_\mathrm{in}) and λ\lambda by fitting the measurements shown in Figure 3 to theoretical models. MinM_\mathrm{in} for TES-SoC μ\muMUX circuits is calculated to be ∼9.2\sim9.2 pH.

Resonator frequency shifts as a function of applied magnetic flux

Figure 3. Data showing shifts in the resonant frequencies of the 32 resonators on TES-SoC, as a function of applied magnetic flux. An x-axis offset is applied to match the phase between all channels. The average value of Min∼9.2M_\mathrm{in} \sim9.2 pH is calculated by fitting the frequency shifts to theoretical models.

The TES-SoC resonators were designed to have 1 MHz bandwidth and 45 MHz spacing between adjacent resonators. Figure 4 shows histograms of (a) measured frequency spacings and (b) resonator bandwidths. Measured values match the target values with variations that are either by design or are within our fabrication uncertainty budget.

Figure 4. Histograms showing (a) the frequency spacings between resonators, (b) the bandwidth of resonators for the μ\muMUX channels on the TES-SoC.

We calculated λ\lambda based on modulation depth of the resonator frequencies due to applied magnetic flux. Figure 5 is a histogram with average λ\lambda of ∼0.25\sim0.25 for all the measured μ\muMUX channels. We note that although the λ\lambda value is within design tolerance, it may have been slightly suppressed due to the microwave probe power being too high.

Figure 5. Histogram showing λ\lambda spread of all the μ\muMUX channels on the TES-SoC.

One crucial aspect in designing the μ\muMUX die form factor is to select the die dimensions that will avoid interference from resonant standing waves or so-called box modes in the silicon substrate. The resonant frequencies of these box modes in a rectangular waveguide can be calculated using Eq. (3) [20].

fmnp=c2ϵrμr(nl)2+(mw)2+(ph)2(3)f_{mnp} = \frac{c}{2\sqrt{\epsilon_r\mu_r}}\sqrt{\left(\frac{n}{l}\right)^2+\left(\frac{m}{w}\right)^2+\left(\frac{p}{h}\right)^2} \tag*{(3)}

where m,n,pm,n,p are the order of TEM modes, l,w,hl,w,h are the length, width, and height of the rectangular waveguide, cc is the speed of light in vacuum, and ϵr\epsilon_r and μr\mu_r are the relative permittivity and TEM mode permeability of the waveguide material, respectively.

For a traditional standalone μ\muMUX chip, lateral dimensions of 4 mm ×\times 20 mm are selected to place the box-mode frequencies outside the (4 to 8) GHz range of interest. The TES-SoC chip described in this work is larger, 10 mm ×\times 20 mm, and we expect increased loss of the microwave resonators due to interference with multiple box-modes. Figure 6 shows the measured internal quality factor (QiQ_i) of resonators on the 10 mm ×\times 20 mm TES-SoC chip which is expected to have box mode degradation. In the same figure, resonators on a 4 mm ×\times 20 mm test chip, fabricated simultaneously on the TES-SoC wafer, are shown to have much larger QiQ_i due to reduced interference from the box-modes. Threshold QiQ_i for 300 kHz, 1 MHz, and 2 MHz resonators are also shown in Figure 6; internal quality factors lower than the threshold will cause poor noise performance. We note that even with box mode degradation, 1 MHz μ\muMUX threshold was surpassed by above 80% of the resonators on 10 mm ×\times 20 mm TES-SoC. Above 90% of the TES-SoC resonators surpassed 2 MHz threshold.

Figure 6. Internal Quality factor (QiQ_i) of the TES-SoC resonators measured on two chips with different dimensions. The test TES-SoC chip is 4 mm ×\times 20 mm similar to a standalone μ\muMUX die. The TES-SoC with fully integrated TES detectors, IF and μ\muMUX circuits is 10 mm ×\times 20 mm. We attribute degraded QiQ_i of resonators on the fully integrated TES-SoC to interference from the standing-wave box modes in the silicon substrate. Threshold traces show the minimum QiQ_i for each bandwidth of the μ\muMUX resonators.

Figure 7 shows process dependence of QiQ_i in our standard μ\muMUX fabrication. Four different test wafers were fabricated starting with a 200 nm thick sputtered Nb on an oxidized intrinsic Si wafer. Microwave resonators were etched in the Nb film after each of the progressive processing steps: BE deposition, BE deposition + R1 liftoff, BE deposition + R1 liftoff + I1 deposition and etch, and BE deposition + R1 liftoff + I1 deposition and etch + W1 deposition and etch, respectively. Sixteen chips from four processed wafers were measured to study the degradations in QiQ_i of the resonators. All of the resonators on 4 mm ×\times 20 mm test TES-SoC shown in Figure 6 have QiQ_i comparable to the standard μ\muMUX resonators in Figure 7 and are at least an order of magnitude larger than the threshold for 1 MHz resonators. We did not observe any evidence of additional degradation in QiQ_i due to added microfabrication steps for TES-SoC processing.

Figure 7. QiQ_i of bare microwave resonators after different amount of processing. The wafer with only a single layer of resonators had much higher internal quality factor than any of the other wafers. While additional processing degrade the QiQ_i of our microwave resonators, the exact mechanism causing these degradations is still a topic of active research. Reduced QiQ_i of resonators at ∼\sim 7.2 GHz is possibly due to interference with cavity modes of the chip enclosure. Threshold traces are drawn for a qualitative comparison to QiQ_i of TES-SoC resonators.

In future designs of larger form-factor TES-SoC, we will implement grounding structures within the chip to effectively reduce the dimensions of the microwave cavity and minimize box-mode losses. Based on our observations and planned future design improvements, we estimate that the TES-SoC process will produce high yield μ\muMUX resonators with various bandwidths including 1 MHz targeted for soft x-ray TES readout. This is a topic of ongoing research and is key to extending the number of μ\muMUX channels on each TES-SoC die by orders of magnitude for a 10,000 TES pixel spectrometer.

Further statistics are needed to confidently pinpoint the particular microfabrication process causing the major reduction in the QiQ_i and is a topic of ongoing investigation.

Conclusion

We have demonstrated the integration of TES microcalorimeter detectors with μ\muMUX readout components using lithographically defined niobium wiring interconnects. To our knowledge, this is the first demonstration of such integration for superconducting low-temperature detectors. We note that in addition to enabling a new generation of high TES pixel density X-ray spectrometers, TES-SoC technology is useful for a variety of applications including integrated TES based Gamma-ray detectors and large array CMB telescopes. Our TES-SoC technology minimizes the need for chip-level packaging traditionally required to interconnect TES detectors to their readout electronics through TES-to-SQUID wirebonding or flip-chip bonding. The TES-SoC process requires a large number of fabrication steps, but may still be cost-effective for many applications. The lithographically defined wiring interconnect in our TES-SoC technology does not require any additional thermal processing and hence the integration process is not expected to degrade the performance of our TES detectors. The yield of the μ\muMUX devices was above threshold and good.

References

  1. [1]P. Szypryt, D. A. Bennett, W. J. Boone, A. L. Dagel, G. Dalton, W. B. Doriese, M. Durkin, J. W. Fowler, E. J. Garboczi, J. D. Gard, G. C. Hilton, J. Imrek, E. S. Jimenez, V. Y. Kotsubo, K. Larson, Z. H. Levine, J. A. B. Mates, D. McArthur, K. M. Morgan, N. Nakamura, G. C. O’Neil, N. J. Ortiz, C. G. Pappas, C. D. Reintsema, D. R. Schmidt, D. S. Swetz, K. R. Thompson, J. N. Ullom, C. Walker, J. C. Weber, A. L. Wessels, and J. W. Wheeler, “Design of a 3000-pixel transition-edge sensor x-ray spectrometer for microcircuit tomography,” IEEE Transactions on Applied Superconductivity, vol. 31, no. 5, pp. 1–5, 2021.DOI
  2. [2]Okumura, T., Azuma, T., Bennett, D. A., Chiu, I., Doriese, W. B., Durkin, M. S., Fowler, J. W., Gard, J. D., & al., H. et . (2023). Proof-of-Principle Experiment for Testing Strong-Field Quantum Electrodynamics with Exotic Atoms: High Precision X-Ray Spectroscopy of Muonic Neon. Phys. Rev. Lett., 130, 173001. https://doi.org/10.1103/PhysRevLett.130.173001
  3. [3]J. W. Fowler, L. Miaja-Avila, G. C. O’Neil, J. N. Ullom, H. Whitelock, and D. S. Swetz, “The potential of microcalorimeter X-ray spectrometers for measurement of relative fluorescence-line intensities,” Radiation Physics and Chemistry, vol. 202, p. 110487, 2023. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0969806X22005254DOI
  4. [4]S.-J. Lee, C. J. Titus, R. Alonso Mori, M. L. Baker, D. A. Bennett et al., “Soft x-ray spectroscopy with transition-edge sensors at stanford synchrotron radiation lightsource beamline 10-1,” Review of Scientific Instruments, vol. 90, no. 11, p. 113101, 11 2019. [Online]. Available: https://doi.org/10.1063/1.5119155
  5. [5]Syzpryt, P., Nakamura, N., Becker, D. T., & Bennett, D. A. et . al . (2023). A Tabletop X-Ray Tomography Instrument for Nanometer-Scale Imaging: Demonstration of the 1,000-Element Transition-Edge Sensor Subarray. IEEE Transactions on Applied Superconductivity, 33(5), 1–5.
  6. [6]J. A. B. Mates, D. T. Becker, D. A. Bennett, B. J. Dober, J. D. Gard, J. P. Hays-Wehle, J. W. Fowler, G. C. Hilton, C. D. Reintsema, D. R. Schmidt, D. S. Swetz, L. R. Vale, and J. N. Ullom, “Simultaneous readout of 128 x-ray and gamma-ray transition-edge microcalorimeters using microwave squid multiplexing,” Applied Physics Letters, vol. 111, no. 6, p. 062601, 08 2017. [Online]. Available: https://doi.org/10.1063/1.4986222
  7. [7]Kilbourne, C. A., Bandler, S. R., Brown, A. D., Chervenak, J. A., Figueroa-Feliciano, E. M., Finkbeiner, N. I., Lomoto, R. L., Kelley, F. S., Porter, F. S., Saab, J., Sadleir, J., & White, J. (2006). High-density arrays of x-ray microcalorimeters for Constellation-X. Space Telescopes and Instrumentation II: Ultraviolet to Gamma Ray, 6266, 626621. https://doi.org/10.1117/12.672419
  8. [8]J. P. Hays-Wehle, P. J. Lowell, D. R. Schmidt, G. C. Hilton, D. A. Rudman, D. S. Swetz, and J. N. Ullom, “An overhanging absorber for tes x-ray focal planes,” IEEE Transactions on Applied Superconductivity, vol. 27, no. 4, pp. 1–4, 2017.DOI
  9. [9]E. Wassell, J. Adams, S. Bandler, J. A. Chervenak, R. Cumbee, F. Finkbeiner, J. Fuhrman, S. Hull, R. Kelley, C. Kilbourne, J. Mateo, H. Muramatsu, F. Porter, A. Rani, K. Sakai, S. Smith, N. Wakeham, and S. Yoon, “Microcalorimeter Absorber Optimization for ATHENA and LEM,” Journal of Low Temperature Physics, vol. 216, no. 1, pp. 417–426, Jul. 2024. [Online]. Available: https://doi.org/10.1007/s10909-024-03173-3
  10. [10]G. O’Neil, D. Swetz, R. Doriese, D. Schmidt, L. Vale, J. Weber, R. Singh, M. Keller, M. Vissers, K. Morgan et al., “Flexible Superconducting Wiring for Integration with Low-Temperature Detector and Readout Fabrication,” Journal of low temperature physics, vol. 217, no. 3, pp. 426–433, 2024.DOI
  11. [13]Roy, A., Singh, R., Weber, J. C., Doriese, W. B., Gard, J., Keller, M. W., Mates, J. A. B., Morgan, K. M., Ortiz, N. J., Swetz, D. S., Schmidt, D. R., Ullom, J. N., Jahrman, E. P., Allison, T. C., Jammuch, S., Titus, C. J., Jaye, C., Fischer, A., & O’Neil, G. C. (2026). Characterization of silicon-membrane tes microcalorimeters for large-format x-ray spectrometers with integrated microwave squid readout. IEEE Transactions on Applied Superconductivity.
  12. [14]J. G. Simmons, “Generalized formula for the electric tunnel effect between similar electrodes separated by a thin insulating film,” Journal of Applied Physics, vol. 36, no. 6, pp. 1793–1803, 06 1963. [Online]. Available: https://doi.org/10.1063/1.1702682
  13. [15]Imamura, R., Shiota, T., & Hasuo, S. (2011). Fabrication of High Quality Nb/AlOx-Al/Nb Josephson Junctions: I-Sputtered Nb Films for Junction Electronics. IEEE Transactions on Applied Superconductivity, 2(1).
  14. [16]J. A. B. Mates, “The microwave SQUID multiplexer,” Ph.D. thesis, University of Colorado, 2011.
  15. [17]V. Ambegaokar and A. Baratoff, “Tunneling between superconductors,” Phys. Rev. Lett., vol. 10, pp. 486–489, Jun 1963. [Online]. Available: https://link.aps.org/doi/10.1103/PhysRevLett.10.486DOI
  16. [18]Jones, D., Singh, R., Austermann, J., Beall, J. A., Daniel, D., Duff, S. M., Dutcher, D., Groh, J., Hubmayr, B. R., Johnson, R. M., Link, T., Lucas, T. J., Mates, J. A. B., Staggs, J., Ullom, L., Vale, L., Van Lanen, M., Vissers, M., & Wang, Y. (2024). Qualification of Microwave SQUID Multiplexer Chips for Simons Observatory. Journal of Low Temperature Physics, 216(1), 50–56. https://doi.org/10.1007/s10909-024-03102-4
  17. [19]Ji, W. J. C., Nobles, E. E., Singh, R., Roy, A., Morgan, K. M., Ortiz, N. J., Van Lanen, J. V., Bae, K. S., Duff, S. M., Swetz, D. S., Doriese, W. B., Schmidt, D. R., & Ullom, J. N. (2026). Hardware low-temperature measurements of transition-edge sensors. IEEE Transactions on Applied Superconductivity.
  18. [20]R. E. Collin, Foundations for Microwave Engineering. Wiley-IEEE Press, 2001.DOI

Paper details

vlnc:W_qpvc5m5m
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents