Transmon qubit using Sn as a junction superconductor

Amrita Purkayastha,1, ∗ Amritesh Sharma,1 Param J. Patel,1, 2 An-Hsi Chen,3 Connor P. Dempsey,4
Shreyas Asodekar,1 Subhayan Sinha,1 Maxime Tomasian,1, † Mihir Pendharkar,4, ‡ Christopher
J. Palmstrøm,4, 5, 6 Moı̈ra Hocevar,3 Kun Zuo,7, 8 Michael Hatridge,2 and Sergey M. Frolov∗1, §
1
Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, PA 15260, USA
2
Department of Applied Physics, Yale University, New Haven, CT 06511, USA
3
Univ. Grenoble Alpes, Grenoble INP, CNRS, Institut Néel, 38000 Grenoble, France
4
Department of Electrical and Computer Engineering,
University of California, Santa Barbara, CA 93106
5
California NanoSystems Institute, University of California Santa Barbara, Santa Barbara, CA 93106, USA
arXiv:2508.04007v1 [[Link]-hall] 6 Aug 2025

6
Materials Department, University of California Santa Barbara, Santa Barbara, CA 93106, USA
7
School of Physics, The University of Sydney, Sydney, NSW 2006, Australia
8
ARC Centre of Excellence for Engineered Quantum Systems,
School of Physics, The University of Sydney, Sydney, NSW 2006, Australia
(Dated: August 7, 2025)
Superconductor qubits typically use aluminum-aluminum oxide tunnel junctions to provide the
non-linear inductance. Junctions with semiconductor barriers make it possible to vary the supercon-
ductor material and explore beyond aluminum. We use InAs semiconductor nanowires coated with
thin superconducting shells of β-Sn to realize transmon qubits. By tuning the Josephson energy with
a gate voltage, we adjust the qubit frequency over a range of 3 GHz. The longest energy relaxation
time, T1 = 27 µs, is obtained at the lowest qubit frequencies, while the longest echo dephasing time,
T2 = 1.8 µs, is achieved at higher frequencies. We assess the possible factors limiting coherence
times in these devices and discuss steps to enhance performance through improvements in materials
fabrication and circuit design.

I. INTRODUCTION For instance, moving away from magnetic-flux tuning of

qubits can save chip real estate and streamline circuitry.
Quantum computing has advanced to the stage where Transmon devices based on semiconductor weak links
processors with tens-of-qubits have been demonstrated offer qubit frequency tunability via a gate voltage, pro-
across multiple hardware platforms [1, 2]. Among these viding a faster and more scalable method to tune the
approaches, superconducting qubits have emerged as the qubits. Notable examples include qubits made with Al-
leading architecture [3, 4] having scaled to the hundreds- InAs nanowires [18, 19], Al-InAs two-dimensional elec-
of-qubits regime. However, many challenges remain be- tron gases (2DEGs) [20], Al-InAs selective area growth
fore the threshold for fault-tolerant quantum computa- (SAG) nanowires [21], Al-Si/Ge nanowires [22, 23], pla-
tion can be reached. Leaps forward may come from algo- nar Ge [24] and graphene [25]. A wide range of metals
rithmic [5], architectural [6] and wiring improvements [7]. can be deposited onto a semiconductor weak link [26–30].
Next to these, advances in materials [8, 9] for quantum Nevertheless, aluminum remains the dominant choice
computing stand out because quantum information is ex- in hybrid transmons, imposing the requirement of sub-
tremely sensitive to the microscopic environments both Kelvin temperatures and increasing their susceptibility
within [10] as well as surrounding the qubits[11, 12]. to pair-breaking radiation [31–33].
The centerpiece of superconducting qubits is the We introduce a superconducting transmon qubit based
aluminum-aluminum oxide Josephson junction, a nonlin- on Sn as a junction superconductor, and InAs as the
ear electrical element. This is a 50-year old technology semiconducting weak link. The superconducting gap of
in which oxide is typically grown by oxidizing evaporated β-Sn is approximately three times that of Al [26]. This
aluminum, and this will ultimately constrain decoher- can help alleviate the ultralow operating temperature
ence times [13–15]. Research into improving the perfor- requirements [34]. It may also reduce the impact of
mance of and adding functionality to Josephson junctions quasiparticles on coherence by promoting faster decay
may prove key to scaling up quantum processors [16, 17]. dynamics [35] and enabling their confinement via gap-
engineered structures [36].
InAs is a high-mobility semiconductor [37] that can
facilitate gate-voltage tunable Josephson harmonic con-
∗ Current address: Materials Research Laboratory, University of tent [38], introducing functionality not present in oxide
Illinois at Urbana-Champaign, Urbana, IL 61801
† Current address: Univ. Grenoble Alpes, Grenoble INP, CNRS,
tunnel junctions. Gate tunability of our transmon qubit
Institut Néel, 38000 Grenoble, France is over 3 GHz which offers additional flexibility in multi-
‡ Current address: Department of Materials Science and Engineer- qubit circuits. We attain relaxation times T1 of 27 µs,
ing, Stanford University, Stanford, CA 94305 likely limited by the substrate and qubit cavity design.
§ frolovsm@[Link] Dephasing times T2 are 1.8 µs, pointing at future im-
 2

(a) (b) (c)

50 μm

200 nm

fr
fd

(d) (e)

f02/2
f01

FIG. 1. False-colored SEM image of the NbTiN qubit chip. The readout resonator (blue) is inductively coupled to the
transmission line (purple) and capacitively coupled to the qubit capacitor (green). The gate port is colored red. (b) Schematic
of the junction area showing InAs nanowire (dark brown) with Sn shell (yellow) patched with Al (magenta) to the NbTiN
ciruit (grey). (c) False-colored SEM image of the Sn-InAs nanowire Josephson junction in the red box in (a). (d) Two-tone
spectroscopy showing transmission S21 at fr as a function of qubit drive frequency fd and Vg . Inset: A line cut from (d) at
Vg =1.12 V. (e) Extracted f01 (left) and EJ /Ec (right) as a function of Vg .

provements such as with preventing partial oxidation of GHz for Qubit B. The readout resonators are inductively
Sn and charging energy optimization. coupled to a common transmission line. The qubits are
Identifying superconductor/barrier combinations with designed for a charging energy of Ec /h = e2 /2C ≈ 380
long coherence times opens pathways to explore quan- MHz.
tum computing beyond aluminum. While our metrics do The Josephson element consists of a semiconducting
not overshadow the tunnel junctions at this stage, they InAs nanowire coated with a 15 nm thick shell of β-Sn
show the promise of Sn, and provide a rare example of as shown in Fig. 1(b),(c). The shell covers half of the
alternative junctions with coherence times in the tens of nanowire circumference. Along the nanowire, the shell is
microseconds range. Semiconductor InAs nanowires fa- interrupted to create the junction. The break in the shell
cilitate high transparency barriers to many superconduc- is achieved by shadowing the junction wire with a nearby
tor metals, making them an attractive template for pro- wire that acts as a shadow mask. Using a micromanipula-
totyping materials combinations until a set of the most tor, the nanowire is placed in between the capacitor pads
promising ones is identified. Selective-area grown [39] or of the pre-fabricated NbTiN circuit indicated by the red
top-down etched nanowires [26, 30, 40, 41] can be used box in Fig. 1(a). The nanowire is then connected to the
in the future for scaling up the most promising materials circuit via an Al patch. The schematic of the transmon
platform. circuit with the nanowire Josephson junction is shown in
Fig. 1(c). The full details of device fabrication can be
found in supplementaryS2.
II. DEVICE DESIGN

Scanning electron microscope image of the qubit and

III. QUBIT SPECTROSCOPY
the associated superconducting circuitry is shown in
Fig. 1(a). The transmon is implemented with a symmet-
ric two-pad floating capacitor where a gate voltage line The Josephson energy of superconductor-
is routed to the junction area. The qubit is capacitively semiconductor junctions is determined by the
coupled to a λ/4 coplanar waveguide (CPW) resonator, gate-tunable critical current, and hence Josephson
with a coupling quality factor of Qc ∼ 2000. For the two energy EJ = ϕ2π 0 Ic
, making the qubit frequency
p
qubits shown in the main text, the bare resonator fre- f01 = 8Ec EJ (Vg )/h also gate tunable. To determine
quencies are fr = 8.182 GHz for Qubit A and fr = 7.706 the qubit frequency f01 we perform two-tone spec-
 3

troscopy applying both the resonator frequency fr and t

A
the drive frequency fd as a function of gate voltage as (a) Time (b)
shown in Fig. 1(d).
Qubit A frequency shifts monotonically as a function
of the gate voltage from ≈ 6.5 GHz to ≈ 4.5 GHz over a
voltage range of 0.5 V. In separate gate sweeps, we ob-
serve f01 as low as ≈ 3.5 GHz bringing the total tunable
range to 3 GHz (see S5). From the measured f01 , we es-
timate the critical current for Qubit A to be in the range
Ic ∼ 9 − 35 nA. Extracted f01 and EJ /Ec as function (c) (d)
of Vg are shown in Fig. 1(e). The gate effect is device-
specific since InAs nanowire devices exhibit a spread of
pinch-off voltages and saturation resistances(see supple-
mentary for QubitB ).
At relatively high drive power (Pdrive ), we also observe
the two-photon transition from ground-to-second excited
state, appearing at f02 /2, alongside the f01 transition,
as shown in Fig. 1(d) and highlighted by the linecut in
the inset. This allows us to extract the anharmonicity
α/2h = f02 /2 − f01 , which can be considerably reduced FIG. 2. (a) Rabi oscillations as a function of pulse dura-
from the designed value of charging energy [42]. For tion and drive amplitude at f01 = 6.578 GHz for a side gate
example, in the inset of Fig. 1(d) the anharmonicity is voltage of Vg = 1.38 V. (b) A linecut from (a) at pulse du-
192 MHz and it is generally gate-voltage tunable at each ration σ = 120 ns. (c) Rabi frequency extracted from (a)
as a function of drive. (d) Rabi oscillations as a function of
Vg which also varies as a function of Vg . This is discussed
pulse duration and drive frequency at fixed drive amplitude
further in a manuscript currently under preparation. for Vg = 1.2425 V.

V. COHERENCE TIMES

IV. RABI OSCILLATIONS

We vary the time between the drive pulse and the read-
out pulse to measure the qubit relaxation time T1 . For
Next, we demonstrate coherent control of the qubit by more negative gate voltages, as the qubit frequency de-
inducing Rabi oscillations between its ground and excited creases, T1 increases as shown in Fig. 3(a). The highest
states. This is done by applying a pulse at the frequency T1 = 26.9 ± 0.7 µs is observed close to the lowest studied
f01 followed by a readout pulse at the resonator frequency frequency f01 = 3.494 GHz (inset of Fig. 3(a)).
(schematic in Fig. 2(a)). In Fig. 2(a) we show an example Typical for nanowire junctions, different gate voltage
of Rabi oscillations for f01 = 6.578 GHz. sweeps may result in offsets in measured Josephson en-
ergies, including hysteretic behavior. These properties
A horizontal linecut at σ=120 ns shows the oscillations originate from the semiconductor of the junction. At
in Fig. 2(b). As expected for a driven two-level system, the same time, T1 as a function of f01 is largely gate
the Rabi frequency demonstrates a linear dependence on sweep-independent, as shown in Fig. 3(b). This indi-
the drive amplitude, as shown in Fig. 2(c). Data such cates that T1 is more sensitive to dielectric losses and
as these are used to calibrate the amplitude that cor- Purcell decay through the resonator which depend on
responds to a π pulse on the qubit. We repeat these f01 than to any mesoscopic features of the junction such
measurements to calibrate the π pulses for different gate as the local charge environment which may not show a
voltages. distinct trend based on f01 . We observe a drop in T1
We also study the Rabi oscillations as a function around f01 = 3.885 GHz which repeats in multiple gate
of drive pulse length σ and drive frequency detuning sweeps. We attribute this to coupling to a two-level sys-
δ = fd − f01 , at fixed drive amplitude. These display a tem (TLS).
characteristic chevron-like pattern as shown in Fig. 2(d). We perform Ramsey measurements to determine the
As the detuning increases, the oscillation frequency in- dephasing time T2∗ using a Ramsey sequence. Fig. 3(c)
creases, while the amplitude decreases. Because we ap- shows Ramsey oscillations obtained at Vg = 1.32 V,
plied a higher qubit drive amplitude, we could also ob- where the qubit frequency is f01 = 6.616 GHz. At this
serve Rabi oscillations for the f02 /2 transition. It appear particular gate voltage, we find a beating pattern in the
with a much slower Rabi rate compared to oscillations Ramsey oscillations with two slightly different frequen-
at the qubit frequency f01 because this is a two-photon cies. Fitting the oscillations to two sinusoidal functions
process. with individual exponential decay times, we obtain two
 4

(a) (b)
T1 =26.9 μs

(c) (d) (e)

T*2a =1.4 μs
T*2b =1.3 μs

T2E =1.8 μs

FIG. 3. (a) T1 as a function of side gate voltage Vg for Qubit A. Inset: Relaxation time extracted by fitting an exponential
to the data, T1 = 26.9 ± 0.7 µs at f01 = 3.494 GHz. (b) T1 plotted as a function of qubit frequency f01 for multiple gate
sweeps. (c) Two dephasing times of T2a = 1.4 ± 0.1 µs and T2b = 1.3 ± 0.2 µs are extracted by fitting the decaying oscillations
to double sinusoidal functions with different exponential envelopes. Here, f01 = 6.616 GHz. (d) Echo measurement gives
T2E = 1.8 ± 0.1 µs. (e) T2∗ , T2a and T2b and T2E plotted as a function of f01 . The error bars in (a), (b), and (e) represent the
standard deviations of the fits.

∗ ∗
coherence times T2a = 1.4 µs and T2b = 1.3 µs. We rials properties of the Sn-InAs nanowire junction, the
attribute the beating pattern to a nearby low frequency fabrication details, or by the configuration of the exter-
two-level fluctuator similar to what has been reported in nal circuit. In Fig. 4, we compare the T1 data for two
[22]. qubits fabricated on the same chip. The contour lines,
We also perform echo measurements by applying a π- T1 = Q/2πf01 , for comparison, assume fixed quality fac-
pulse in the middle of the Ramsey sequence to mitigate tors. For Qubit A, the data fall close to a trend consistent
low-frequency noise. We observe a T2E = 1.8 µs, which with the readout resonator quality factors fabricated on
is similar to T2∗ as shown in Fig. 3(d). This indicates the the same NbTiN-Si wafer (Fig. S2). This hints at the
presence of higher-frequency noise in the system. The on-chip losses as the most relevant limiting factor for
relaxation time at this qubit frequency is T1 = 4.12 µs, this qubit. Further optimization could involve varying
thus T2 is not limited by energy relaxation. the substrate, the groundplane superconductor and the
In Fig. 3(e), the measured dephasing time and echo circuit geometry.
times is summarized as a function of the qubit frequency. For Qubit B, the data align closer with the lower con-
T2∗ is extracted by fitting the decaying oscillations to a stant quality factor dependence. This may indicate that
single sinusoidal function. We observe a decrease in T2∗ the details of the nanowire growth and fabrication can
with decreasing qubit frequency f01 . We attribute this constrain T1 for some devices. We note that neither for
behavior to reduced EJ /Ec ratios at these lower frequen- Qubit A nor Qubit B the dependences align perfectly
cies, where the qubit is more sensitive to offset charge with the fixed quality factor lines, a deeper understand-
fluctuations. At f01 = 3.494 GHz, where the highest T1 ing of this would require a more elaborate modeling.
times are obtained, we are unable to measure T2∗ and T2E
due to short timescales. Below we discuss the various factors that may be af-
fecting coherence times. When it comes to the nanowires,
crystalline defects such as stacking faults may still be
present in InAs nanowires [43]. The angle of AlOx cap-
VI. DISCUSSION ping deposition is 60 ◦ away from the angle of Sn deposi-
tion leads to partial capping and to the oxidation of Sn.
A central question in this study is whether the longest The nanowire may land with Sn facing upwards so that
measured coherence times are determined by the mate- the Al patch touches Sn directly, or downwards - in which
 5

identify several pathways to further enhance the qubit

performance through optimization that includes lowering
Qi=106 the charging energy, varying the substrate, ground plane
and patch materials, improvements in the measurement
setup, as well as further optimization of nanowire growth
and fabrication. Replacing aluminum with alternative
superconductors not only enables new possibilities for
Qi=105 designing scalable quantum circuits with enhanced
functionality, but also deepens our understanding of the
materials aspects of the quantum device technology.

VIII. DURATION AND VOLUME OF STUDY

FIG. 4. T1 plotted as a function of qubit frequency f01

for Qubit A and Qubit B. The dotted lines indicate T1 with Qubit A and Qubit B were measured over a period
constant qubit quality factor of 106 and 105 . of four months, during which two cooldowns were per-
formed. The data presented in this article are from the
second cooldown. In total, 20 Sn-InAs nanowire trans-
case the Al patch may be contacting InAs introducing mon devices were measured over the course of two years.
contact resistance. Ion milling may degrade the patch Of the 20 devices measured, all exhibited spectroscopic
area introducing small contact resistance. These factors signatures of the qubit coupled to the resonator. How-
can lead to losses and limit T1 . At the same time, Al ever, only seven devices showed measurable coherence.
within the patch may act as a quasiparticle trap between
two larger gap superconductors, possibly enhancing co-
herence. Additionally, charge noise in the nanowires af- IX. DATA AVAILABILITY
fecting the critical current may be affecting T2 . Operat-
ing the nanowire qubits at a voltage sweet spot, where
Data and code extending beyond what is presented in
∂f /∂Vg → 0, could mitigate this effect. However, we do
the main text and the supplementary are available at
not observe clear voltage sweet spot effects.
10.5281/zenodo.16732149.
Factors related to the external circuit are that the
qubit package box features a copper ground plane below
the chip, which has been reported as limiting T1 through
eddy current dissipation [44]. The dilution refrigerator X. ACKNOWLEDGMENTS
does not include µ-metal shielding, which may lead to
magnetic noise. For T2 times, we point out the relatively Nanowire growth was supported by ANR HY-
high charging energy at lower frequencies, which leads to BRID (ANR-17-PIRE-0001), ANR IMAGIQUE (ANR-
EJ /Ec ∼ 10 − 20 that increases the role of offset charge 42-PRC-0047), IRP HYNATOQ and the Transatlantic
fluctuations [45]. Research Partnership. Sn shell growth was supported
by the NSF Quantum Foundry at UCSB funded via
the Q-AMASE-i program under award DMR-1906325.
VII. CONCLUSIONS Nanowire microscopy characterization was supported by
the U.S. Department of Energy Office of Basic En-
We demonstrate coherent control of transmon qubits ergy Sciences (BES) through grant DE-SC-0019274, and
with Sn as the junction superconductor. The measured transport characterization through the U.S. Depart-
qubits exhibit frequency tunability of order 3 GHz within ment of Energy, Basic Energy Sciences grant DE-SC-
a gate voltage range of < 0.5 V. Coherence time mea- 0022073. Microwave measurements were supported by
surements reveal that the relaxation time T1 scales in- the LPS/ARO nextNEQST program W911NF2210036.
versely with f01 . We measure a maximum relaxation This work made use of the Nanoscale fabrication and
time, T1 ∼ 27 µs at f01 = 3.494 GHz for Qubit A. The Characterization facility (NFCF) at The Gertrude E. and
maximum T2 measured is 1.8 µs at f01 = 6.616 GHz after John M. Petersen Institute of NanoScience and Engi-
an echo pulse. neering (PINSE) as well as of facilities at the Western
As it stands, the Sn-based qubit performs on par Pennsylvania Quantum Information Core (WPQIC) at
with its Al-based hybrid qubit counterparts[19]. We the University of Pittsburgh.
 6

[1] M. Liu, R. Shaydulin, P. Niroula, et al., Certified ran- 82, 124501 (2019).
domness using a trapped-ion quantum processor, Nature [16] S. Kim, H. Terai, T. Yamashita, K. Makise, Y. Uzawa,
640, 343 (2025). G. Fujii, and M. Fujiwara, Enhanced coherence of all-
[2] H. Aghaee Rad, T. Ainsworth, R. N. Alexander, et al., nitride superconducting qubits epitaxially grown on sili-
Scaling and networking a modular photonic quantum con substrate, Communications Materials 2, 98 (2021).
computer, Nature 638, 912 (2025). [17] A. Anferov, K.-H. Lee, F. Zhao, J. Simon, and D. I.
[3] F. Arute, K. Arya, R. Babbush, et al., Quantum Schuster, Improved coherence in optically defined nio-
supremacy using a programmable superconducting pro- bium trilayer-junction qubits, Phys. Rev. Appl. 21,
cessor, Nature 574, 505 (2019). 024047 (2024).
[4] Google Quantum AI and Collaborators, Quantum error [18] T. W. Larsen, K. D. Petersson, F. Kuemmeth, T. S.
correction below the surface code threshold, Nature 638, Jespersen, P. Krogstrup, J. Nygård, and C. M. Marcus,
920 (2025). Semiconducting-nanowire-based superconducting qubit,
[5] Google Quantum AI, Suppressing quantum errors by Physical Review Letters 115, 127001 (2015).
scaling a surface code logical qubit, Nature 614, 676 [19] F. Luthi, T. Stavenga, O. W. Enzing, A. Bruno,
(2023). C. Dickel, N. K. Langford, M. A. Rol, T. S. Jespersen,
[6] H. Putterman, K. Noh, C. T. Hann, et al., Hardware- J. Nygård, P. Krogstrup, and L. DiCarlo, Evolution of
efficient quantum error correction via concatenated Nanowire Transmon Qubits and Their Coherence in a
bosonic qubits, Nature 638, 927 (2025). Magnetic Field, Physical Review Letters 120, 100502
[7] S. Krinner, S. Lazar, A. Remm, A. Di Paolo, C. Hellings, (2018).
T. Walter, S. J. Pauka, A. Vepsäläinen, Y. Salathé, [20] L. Casparis, M. R. Connolly, M. Kjaergaard, N. J. Pear-
J. Heinsoo, P. Kurpiers, P. Magnard, M. Pechal, L. Mon- son, A. Kringhoj, T. W. Larsen, F. Kuemmeth, T. Wang,
dada, V. Lundberg, K. Serniak, M. McEwen, M. Kjaer- C. Thomas, S. Gronin, et al., Superconducting gatemon
gaard, R. E. Lake, T. Frey, S. Ivanov, E. Dumur, A. Blais, qubit based on a proximitized two-dimensional electron
and A. Wallraff, Engineering cryogenic setups for 100- gas, Nature Nanotechnology 13, 915 (2018).
qubit scale superconducting circuit systems, EPJ Quan- [21] A. Hertel et al., Gate-tunable transmon using selective-
tum Technology 6, 2 ((2019)). area-grown superconductor-semiconductor hybrid struc-
[8] A. P. M. Place, L. V. H. Rodgers, P. Mundada, B. M. tures on silicon, Physical Review Applied 18, 034042
Smitham, et al., New material platform for superconduct- (2022).
ing transmon qubits with coherence times exceeding 0.3 [22] H. Zheng et al., Coherent control of a few-channel hole
milliseconds, Nature Communications 12, 1779 ((2021)). type gatemon qubit, Nano Letters 24, 7173 ((2024)).
[9] M. P. Bland, F. Bahrami, J. G. C. Martinez, P. H. Preste- [23] E. Zhuo, Z. Lyu, X. Sun, et al., Hole-type superconduct-
gaard, B. M. Smitham, A. Joshi, E. Hedrick, A. Pakpour- ing gatemon qubit based on Ge/Si core/shell nanowires,
Tabrizi, S. Kumar, A. Jindal, R. D. Chang, A. Yang, npj Quantum Information 9, 51 (2023).
G. Cheng, N. Yao, R. J. Cava, N. P. de Leon, and [24] O. Sagi, A. Crippa, M. Valentini, N. W. Hendrickx, T. A.
A. A. Houck, 2D transmons with lifetimes and coherence de Jong, M. Tagliaferri, A. Sammak, G. Scappucci, and
times exceeding 1 millisecond, arXiv preprint (2025), M. Veldhorst, A gate tunable transmon qubit in planar
arXiv:2503.14798 [quant-ph]. Ge, Nature Communications 15, 6400 (2024).
[10] C. Wang, C. Axline, Y. Y. Gao, T. Brecht, Y. Chu, [25] J. I.-J. Wang et al., Coherent control of a hybrid super-
L. Frunzio, M. H. Devoret, and R. J. Schoelkopf, Sur- conducting circuit made with graphene-based Van der
face participation and dielectric loss in superconducting Waals heterostructures, Nature Nanotechnology 14, 120
qubits, Applied Physics Letters 107, 162601 (2015). (2019).
[11] A. A. Houck, J. A. Schreier, B. R. Johnson, J. M. Chow, [26] M. Pendharkar, B. Zhang, H. Wu, A. Zarassi, S. D. Har-
J. Koch, J. M. Gambetta, D. I. Schuster, L. Frunzio, rington, P. Yu, M. Nahum, T. Nguyen, K. Wickramas-
M. H. Devoret, S. M. Girvin, and R. J. Schoelkopf, Con- inghe, J. S. Lee, and et al., Parity-preserving and mag-
trolling the spontaneous emission of a superconducting netic field–resilient superconductivity in InSb nanowires
transmon qubit, Physical Review Letters 101, 080502 with Sn shells, Science 372, 508 (2021).
(2008). [27] P. Perla, H. A. Fonseka, P. Zellekens, R. Deacon, Y. Han,
[12] S. Huang et al., Microwave package design for supercon- J. Kölzer, T. Mörstedt, B. Bennemann, A. Espiari,
ducting quantum processors, PRX Quantum 2, 020306 K. Ishibashi, et al., Fully in situ nb/inas-nanowire joseph-
(2021). son junctions by selective-area growth and shadow evap-
[13] R. McDermott, Materials origins of decoherence in super- oration, Nanoscale Advances 3, 1413 (2021).
conducting qubits, IEEE Transactions on Applied Super- [28] Y. Chen, D. van Driel, C. Lampadaris, S. A. Khan,
conductivity 19, 2 (2009). K. Alattallah, L. Zeng, E. Olsson, T. Dvir, P. Krogstrup,
[14] J. M. Martinis, K. B. Cooper, R. McDermott, M. Steffen, and Y. Liu, Gate-tunable superconductivity in hybrid
M. Ansmann, K. D. Osborn, K. Cicak, S. Oh, D. P. Pap- insb–pb nanowires, Applied Physics Letters 123, 102601
pas, R. W. Simmonds, and J. M. Martinis, Decoherence (2023).
in josephson qubits from dielectric loss, Physical Review [29] M. Bjergfelt, D. J. Carrad, T. Kanne, M. Aagesen, E. M.
Letters 95, 210503 (2005). Fiordaliso, E. Johnson, B. Shojaei, C. J. Palmstrøm,
[15] C. Müller, J. H. Cole, and J. Lisenfeld, Towards under- P. Krogstrup, T. S. Jespersen, et al., Superconducting
standing two-level systems in amorphous solids: insights vanadium/indium-arsenide hybrid nanowires, Nanotech-
from quantum circuits, Reports on Progress in Physics nology 30, 294005 (2019).
 7

[30] A. Sharma, A.-H. Chen, C. P. Dempsey, A. Purkayastha, [43] H. Shtrikman, R. Popovitz-Biro, A. Kretinin, L. Houben,
M. Pendharkar, S. Tan, C. J. Palmstrøm, S. M. M. Heiblum, M. Bukala, M. Galicka, R. Buczko, and
Frolov, and M. Hocevar, Sn-inas nanowire shadow- P. Kacman, Method for suppression of stacking faults
defined josephson junctions, arXiv preprint (2025), in wurtzite iii–v nanowires, Nano Letters 9, 1506 (2009).
arXiv:2503.13725 [[Link]-sci]. [44] Y. Huang, Y.-H. Huang, H. Wang, Z. Steffen, J. Cripe,
[31] C. H. Liu, D. C. Harrison, S. Patel, C. D. Wilen, O. Raf- F. C. Wellstood, and B. S. Palmer, Identification and mit-
ferty, A. Shearrow, A. Ballard, V. Iaia, J. Ku, B. L. T. igation of conducting package losses for quantum super-
Plourde, and R. McDermott, Quasiparticle poisoning of conducting devices, Applied Physics Letters 123, 044001
superconducting qubits from resonant absorption of pair- (2023).
breaking photons, Phys. Rev. Lett. 132, 017001 (2024). [45] J. Koch, T. M. Yu, J. Gambetta, A. A. Houck, D. I.
[32] K. Serniak, M. Hays, G. de Lange, S. Diamond, Schuster, J. Majer, A. Blais, M. H. Devoret, S. M. Girvin,
S. Shankar, L. D. Burkhart, L. Frunzio, M. Houzet, and R. J. Schoelkopf, Charge-insensitive qubit design de-
and M. H. Devoret, Hot nonequilibrium quasiparticles in rived from the Cooper pair box, Physical Review A 76,
transmon qubits, Physical Review Letters 121, 157701 042319 (2007).
(2018). [46] D. Flanigan, Resonator, [Link]
[33] A. D. Córcoles, J. M. Chow, J. M. Gambetta, C. Rigetti, danielflanigan/resonator (2023).
J. R. Rozen, G. A. Keefe, M. B. Rothwell, M. B. Ketchen, [47] M. Khalil, M. Stoutimore, F. Wellstood, and K. Osborn,
and M. Steffen, Protecting superconducting qubits from An analysis method for asymmetric resonator transmis-
radiation, Applied Physics Letters 99, 181906 (2011). sion applied to superconducting devices, Journal of Ap-
[34] A. Anferov, S. P. Harvey, F. Wan, J. Simon, and D. I. plied Physics 111, 054510 (2012).
Schuster, Superconducting qubits above 20 ghz operating [48] S. Probst, H. F. Song, P. A. Bushev, A. V. Ustinov, and
over 200 mk, PRX Quantum 5, 030347 (2024). M. Weides, Efficient and robust analysis of complex scat-
[35] S. B. Kaplan, C. C. Chi, D. N. Langenberg, J. J. Chang, tering data under noise in microwave resonators, Review
S. Jafarey, and D. J. Scalapino, Quasiparticle and phonon of Scientific Instruments 86, 024706 (2015).
lifetimes in superconductors, Physical Review B 14, 4854 [49] H. Wang, M. Hofheinz, J. Wenner, M. Ansmann, R. C.
(1976). Bialczak, M. Lenander, E. Lucero, M. Neeley, A. D.
[36] M. McEwen, K. C. Miao, J. Atalaya, A. Bilmes, O’Connell, D. Sank, M. Weides, A. N. Cleland, and
A. Crook, J. Bovaird, J. M. Kreikebaum, N. Zo- J. M. Martinis, Improving the coherence time of super-
brist, E. Jeffrey, B. Ying, A. Bengtsson, H.-S. Chang, conducting coplanar resonators, Applied Physics Letters
A. Dunsworth, J. Kelly, Y. Zhang, E. Forati, R. Acharya, 95, 233508 (2009).
J. Iveland, W. Liu, S. Kim, B. Burkett, A. Megrant,
Y. Chen, C. Neill, D. Sank, M. Devoret, and A. Oprem-
cak, Resisting high-energy impact events through gap
engineering in superconducting qubit arrays, Phys. Rev.
Lett. 133, 240601 (2024).
[37] S. Chuang, Q. Gao, R. Kapadia, A. C. Ford, J. Guo,
and A. Javey, Ballistic InAs nanowire transistors, Nano
Letters 13, 555 (2013).
[38] E. Strambini, A. Iorio, O. Durante, R. Citro, C. Sanz-
Fernández, C. Guarcello, I. V. Tokatly, A. Braggio,
M. Rocci, N. Ligato, V. Zannier, L. Sorba, F. S. Berg-
eret, and F. Giazotto, A josephson phase battery, Nature
Nanotechnology 15, 656 (2020).
[39] A. Goswami, S. R. Mudi, et al., Sn/InAs Josephson
Junctions on Selective Area Grown Nanowires with in
Situ Shadowed Superconductor Evaporation, Nano Let-
ters 23, 6860 (2023).
[40] S. A. Khan, S. Martı́-Sánchez, D. Olsteins, C. Lam-
padaris, D. J. Carrad, Y. Liu, J. Quiñones, M. Chiara S.,
T. S. Jespersen, P. Krogstrup, et al., Epitaxially driven
phase selectivity of Sn in hybrid quantum nanowires,
ACS Nano 17, 11794 ((2023)).
[41] S. A. Khan, C. Lampadaris, A. Cui, L. Stampfer, Y. Liu,
S. J. Pauka, M. E. Cachaza, E. M. Fiordaliso, J.-H. Kang,
S. Korneychuk, et al., Highly transparent gatable su-
perconducting shadow junctions, ACS Nano 14, 14605
(2020).
[42] A. Kringhøj, L. Casparis, M. Hell, T. W. Larsen,
F. Kuemmeth, M. Leijnse, K. Flensberg, P. Krogstrup,
J. Nygård, K. D. Petersson, and C. M. Marcus, An-
harmonicity of a superconducting qubit with a few-
mode Josephson junction, Physical Review B 97, 060508
(2018).
 Appendix: Supporting Information and Additional Data for ”Transmon qubit using
Sn as a junction superconductor”
Amrita Purkayastha,1, ∗ Amritesh Sharma,1 Param J. Patel,1, 2 An-Hsi Chen,3 Connor P. Dempsey,4
Shreyas Asodekar,1 Subhayan Sinha,1 Maxime Tomasian,1, † Mihir Pendharkar,4, ‡ Christopher
J. Palmstrøm,4, 5, 6 Moı̈ra Hocevar,3 Kun Zuo,7, 8 Michael Hatridge,2 and Sergey M. Frolov∗1, §
1
Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, PA 15260, USA
2
Department of Applied Physics, Yale University, New Haven, CT 06511, USA
3
Univ. Grenoble Alpes, Grenoble INP, CNRS, Institut Néel, 38000 Grenoble, France
4
Department of Electrical and Computer Engineering,
University of California, Santa Barbara, CA 93106
5
California NanoSystems Institute, University of California Santa Barbara, Santa Barbara, CA 93106, USA
6
Materials Department, University of California Santa Barbara, Santa Barbara, CA 93106, USA
7
School of Physics, The University of Sydney, Sydney, NSW 2006, Australia
8
ARC Centre of Excellence for Engineered Quantum Systems,
School of Physics, The University of Sydney, Sydney, NSW 2006, Australia
(Dated: August 7, 2025)

S1. NANOWIRE GROWTH

InAs nanowires were grown using a gold-assisted vapor-liquid-solid (VLS) mechanism in a molecular beam epitaxy
(MBE) reactor. Gold nanoparticles were deposited on deoxidized InAs (001) substrates, which were thermally prepared
before initiating nanowire growth. The nanowires grow along the [111]B direction, resulting in inclined growth relative
to the substrate surface. On (001) substrates, this leads to nanowires extending in opposing in-plane directions,
producing natural crossing geometries. These geometries allow neighboring nanowires to shadow each other in later
processing steps, enabling selective deposition without the need for etching.
Following nanowire growth, tin shells were deposited at cryogenic temperatures (85 K) in an ultra-high vacuum
(UHV) environment. The native oxide on the nanowires was removed via atomic hydrogen cleaning prior to metal
deposition. A 15 nm thin Sn layer was then evaporated at a shallow angle relative to the substrate normal, allowing
in-situ shadowing to define semiconducting weak links for Josephson junctions. While the sample remained cryogenic,
a 3 nm AlOx capping layer was deposited at normal incidence via electron-beam evaporation. The AlOx coating
prevents Sn oxidation and de-wetting into granular structures; however, due to differing deposition angles, complete
conformal coverage of the Sn shell is not warranted. Samples were subsequently warmed to room temperature under
vacuum.
The nanowires so formed have diameters in the range 50-70 nm. Such etch free junctions can vary in length from
70-120 nm. It was found in transport studies that Sn induces a superconducting gap of order 600 µeV and switching
currents reaching values up to 500 nA. Further details can be found in [30]

S2. DEVICE FABRICATION

The qubit devices are fabricated in multiple steps involving optical lithography and electron beam lithography.
A 120 nm NbTiN film is first sputtered onto a acid (Piranha+BOE) cleaned high resistivity (ρ > 5kΩcm) silicon
wafer at 600◦ C in AJA DC magnetron sputtering chamber. Then using optical lithography, the readout resonators,
transmission line, gate line and qubit capacitors are patterned and etched with ICP-RIE chlorine etching. Fine
markers with Ti/Au are defined with e-beam lithography in between the capacitor pads for positioning the Sn-InAs
nanowires with a micromanipulator needle. Once positioned, SEM images are taken to locate the break in the Sn shell
that defines the Josephson junction. Contacts to the Sn-InAs nanowires are designed using CAD and lithographically

∗ Current address: Materials Research Laboratory, University of ‡ Current address: Department of Materials Science and Engineer-
Illinois at Urbana-Champaign, Urbana, IL 61801
† ing, Stanford University, Stanford, CA 94305
Current address: Univ. Grenoble Alpes, Grenoble INP, CNRS, § frolovsm@[Link]
Institut Néel, 38000 Grenoble, France
 9

defined using e-beam. The connection to the rest of the superconducting circuit is established with aluminum from
an e-beam evaporator after removing 3 nm of the capping AlOx layer with 80 seconds of 250V - 15mA in-situ Argon
milling. After that, the lift-off is done with acetone-IPA.

S3. MEASUREMENT SETUP

10MHz Reference
Clock

Trigger
Marker

Tektronix AWG 5014C

Alazar Tech
f01 fr fr-50MHz ATS9870
CH1 CH2 CH3 CH4
20 dB

20 dB

20 dB

20 dB
I1
S LO R S LO R

Pre-Amplifier
I I2
LO LO
I Q I Q
Qubit drive

Cavity drive
R R
30 dB

VNA
Attenuator
S
Voltage Bias DAC
1 K&L Low Pass
A Filter

5x Amp 2 Eccosorb
Filter
300K
3
Output Lines
Input Lines

C-124 Eccosorb
Gate Lines

Copper Tube Filter

4 VLFX-80 Low Pass

50K Filter

4K Band Pass
H Filter
20 dB

DC Block
800mK

HEMT
H
Amplifier
100mK
20 dB

RT
A
Amplifier

50mK Qubit Drive

B
1 Amplifier

2 3
Dual Junction
Isolator

2
20 dB

4 Signal Generator

2 2 S Splitter

LO
I Q IQ Mixer
R
Device
LO
I IR Mixer
Cu Octobox R

FIG. S1. Schematic of the measurement setup along with fridge wiring.
 10

All measurements discussed in the paper are performed in an Oxford triton (DR200) dilution refrigerator with a
base temperature 20mK. The schematic of the measurement setup is given in Fig. S1. Inside the fridge the input
lines have a total attenuation of 60dB split into 20dB attenuations at different temperature stages as shown in the
schematic. Homemade eccosorb filter and low pass K&L filters are used at the lowest temperature stage before
the input signal is routed onto the device chip. The device chip resides in a copper PCB which is placed in an
copper octobox. The output line has eccosorb filters and dual junction isolator at the MC stage. Then the signal is
amplified with a HEMT amplifier at the 4K stage along with a room temperature microwave amplifier. The voltage
to bias the gates adjacent to the semiconducting weak link of the nanowire transmons was provided by DACs of a
TU Delft IVVI rack, amplified with a 5 V/V battery-driven amplifier. They are also low-pass filtered using Calmont
coaxial cables along with Mini-Circuits VLFX (cutoff 80MHz) and homemade copper tube filters (filled with CR124
eccosorb-cutoff 1GHz-2dB attenuation) at the MCX before arriving at the device chip. For single tone and two tone
measurements, a vector network analyzer (Keysight, E5071C) and a microwave signal generator (Keysight, N5813B)
are used. For the time domain measurements, the qubit control and readout tones are generated using SigCore5511A
and modulated by a gaussian envelope with an arbitrary waveform generator (AWG, Tektronix 5014C). The output
signals are demodulated and detected using AlazarTech ATS9870 digitizer with 1GS/s sampling rate.

S4. LOSS PROBED BY RESONATOR QUALITY FACTOR

To evaluate the relative contributions of NbTiN on high resistivity silicon and the nanowire Josephson junction
to qubit energy loss, we measure the quality factors of readout resonators fabricated under identical conditions but
without the nanowire qubits. We extract the Qi of these λ/4 resonators at single photon levels by functional fits of
the transmission data to the following complex transmission coefficient [46–48]

(Ql /|Qc |)eiϕ

 
iα −i2πf τ
S21 = ae e 1− (S0)
1 + i2Ql (f /fr − 1)

where a accounts for the net attenuation in the fridge lines, α is the global phase shift, τ is the electric delay,
fr is the resonant frequency and Qc = |Qc |e−iϕ is the complex coupling quality factor with ϕ which describes the
asymmetry in response due to impedance mismatches
′
between the transmission line ports. The real valued Ql is the
loaded quality factor Q1l = Q1i + Q1′ where 1/Qc = cos ϕ
|Qc | Fig.S2 shows the plot of the fitted internal quality factors. We
.
c
find that Qi increases with the input power. This behavior is explained by a power-dependent saturation effect[49],
indicating that the primary source of loss in the resonators is due to surface two-level systems in Si-NbTiN.

5.0 7.058 GHz

7.243 GHz
4.5 7.443 GHz
7.759 GHz
4.0 8.075 GHz
8.328 GHz
Qi (×105)

3.5 8.589 GHz

8.866 GHz
3.0
2.5
2.0

100 101 102 103 104 105

Number of Photons

FIG. S2. Power dependence of internal quality factor Qi for eight measured resonators with different resonant frequency fr . All
the resonators show power dependence, with Qi improving as photon number increases. This indicates Qi limited by two-level
systems at low photon numbers.
 11

S5. ADDITIONAL DATA AT LOW QUBIT FREQUENCIES FOR QUBIT A

(a) (b)

FIG. S3. (a) S21 as a function of fd and Vg at low drive power near low qubit frequencies. The linewidth of S21 dips could
be seen broadening as the qubit frequency decreases with Vg . (b) At high drive power, the f02 /2 dip broadens and splits at
certain gate voltages. A possible explanation is charge dispersion at these low qubit frequencies where EJ /Ec ∼ 15. At around
Vg =0.995 V, a state anti-crosses with the qubit at f01 which is a TLS state where we also see a drop in T1

(a) (b)

FIG. S4. (a) Rabi oscillations showing an asymmetric chevron pattern around f01 = 3.481 GHz. (b) Repeated T1 measurement
over time for f01 = 3.481 GHz showing an average T1 = 23.6 ± 0.5 µs. The error bars in (b) represent the standard deviations
of the T1 curve fit.
 12

S6. ADDITIONAL DATA: QUBIT B

Qubit B is another device measured along side Qubit A presented in the main text. Both were designed to have
the same parameters except the readout resonator frequency and went through the same fabrication process.

A. Spectroscopic measurements

(a) (b)

FIG. S5. (a) Variation of resonator frequency fr with gate voltage Vg . Vacuum Rabi splittings due to qubit-resonator
hybridization in gate range 1.8 V < Vg < 3.2 V. (b) Variation of f01 as a function of Vg showing a tunability of ∼ 2 GHz within
a gate range of 0.5 V.
 13

B. Time Domain measurements

(a) (b)
T1 =4.2 μs

(c) (d)

FIG. S6. (a) Rabi oscillations as a function of drive frequency and pulse duration. (b) Decay curve fit to an exponential at
f01 = 4.210 GHz, yielding T1 = 4.2 ± 0.2 µs. (c) T1 as a function of f01 for two different gate sweeps. (d) T2∗ and T2E as a
function of f01 for GateSweep1.