Superconducting Flux-Tunable Resonator

• Superconducting flux-tunable resonators are microwave resonators whose eigenfrequency, coupling, and dissipation are controlled by magnetic flux via elements like SQUIDs and kinetic-inductance loops.
• Their architectures range from coplanar-waveguide resonators with flux-dependent boundary conditions to designs that separate storage and control or use junction-free kinetic inductance for continuous tuning.
• These devices enable precise dynamic reconfiguration for applications such as on-demand microwave storage, tunable dissipation, qubit control, and high-speed flux sensing.

Searching arXiv for recent and foundational papers on superconducting flux-tunable resonators. A superconducting flux-tunable resonator (tRes) is a superconducting microwave resonator whose eigenfrequency, external coupling, or effective dissipation is controlled by magnetic flux. In the literature, this control is realized through dc SQUIDs, rf SQUIDs, nanoSQUIDs, or flux-coupled kinetic-inductance loops embedded in, terminating, or indirectly coupled to a resonant structure. The resulting devices are used not only for frequency tuning, but also for tunable linewidth engineering, on-demand storage and release of microwave fields, parametric modulation, qubit control, and dispersive flux sensing at millikelvin temperatures (Pierre et al., 2014, Paradkar et al., 28 Dec 2025, Wang et al., 2024, Li et al., 2023, Stack et al., 3 Jun 2026).

1. Architectural classes

The most common tRes architecture is a coplanar-waveguide resonator whose boundary condition is made flux dependent by a SQUID termination. Quarter-wave and half-wave implementations both appear in the literature. Quarter-wave devices include dc-SQUID-terminated CPW resonators used for efficient on-chip and flip-chip modulation (Paradkar et al., 28 Dec 2025), RF-SQUID-terminated quarter-wave notch resonators used for magnetometry (Stack et al., 3 Jun 2026), and a monolithic Nb nanoSQUID-embedded quarter-wave resonator fabricated with a neon focused-ion beam (Potter et al., 2024). Half-wave devices include doubly tunable resonators with a dc SQUID at each end (Svensson et al., 2017) and half-wave resonators whose low-quality-factor auxiliary branch is made tunable by a dc SQUID (Partanen et al., 2017).

A second architectural class separates storage from control. In “Storage and on-demand release of microwaves using superconducting resonators with tunable coupling,” a fixed-frequency $\lambda/4$ storage resonator is connected to a transmission line only through a flux-tunable $\lambda/2$ coupling resonator containing a SQUID (Pierre et al., 2014). In this arrangement, the “tunable resonator” is not the memory element itself but the auxiliary coupler that controls the storage mode’s effective external decay rate.

A third class is junction-free or junction-light flux tuning through kinetic inductance. One implementation uses antiparallel dc currents in dedicated NbN ground wires to generate a localized out-of-plane magnetic field at the CPW center conductor, thereby changing the kinetic inductance and shifting the resonance frequency (Wang et al., 2024). Another uses a closed superconducting LC loop placed beside a current-biased feedline; Meissner screening currents alter the kinetic inductance of narrow Al inductors and produce continuous tuning, upconversion, and three-wave mixing without Josephson junctions in the resonator body (Li et al., 2023).

Closely related devices also use a tunable resonator as a network element rather than as a standalone mode shifter. Examples include rf-SQUID-mediated sign-reversible coupling between two transmission-line resonators (Wulschner et al., 2015), a resistor-terminated tunable dissipator used as a heat sink for a high-$Q$ mode (Partanen et al., 2017), and a dc-SQUID LC resonator used to generate a longitudinal photon-number-dependent shift of a flux-qubit transition (Toida et al., 2020). These works show that “tRes” in practice denotes a family of flux-controlled resonant boundary and coupling elements, not a single fixed topology.

2. Flux transduction and circuit descriptions

For dc-SQUID-based tRes devices operated in the small-signal regime, the central mechanism is flux-dependent Josephson inductance. Several papers use the symmetric-SQUID relation

$$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$

so tuning follows from how this inductance enters the resonator boundary condition or effective modal inductance (Foxen et al., 2018, Bengtsson et al., 2018, Svensson et al., 2017, Partanen et al., 2017). In quarter-wave terminations, the SQUID acts as an inductive shunt to ground and modifies the standing-wave condition. One formulation gives

$\tan(k l)=\frac{L_l}{k L_S(\Phi)},$

with an accurate approximation near the fundamental

$$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$

which makes explicit that flux tuning is boundary-condition engineering rather than a mere perturbative frequency trim (Paradkar et al., 28 Dec 2025).

When the tunable element is an RF SQUID, the small-signal inductance is commonly written as

$$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$

with $\beta_L=(2\pi I_c L_g)/\Phi_0<1$ in nonhysteretic operation (Stack et al., 3 Jun 2026). The same nonhysteretic criterion also underlies rf-SQUID-mediated couplers, where the effective inductance can change sign as $\cos(2\pi\Phi/\Phi_0)$ changes sign, enabling sign-reversible inter-resonator coupling (Wulschner et al., 2015).

In two-mode tRes systems, flux control often enters through a coupled-oscillator Hamiltonian. The storage-and-release device is described in the linear regime by

$$H=\hbar\omega_1 a^\dagger a+\hbar\omega_2(\Phi)b^\dagger b+\hbar g(a^\dagger b+a b^\dagger),$$

where the tunable mode $\lambda/2$0 controls how a storage mode $\lambda/2$1 hybridizes with, and leaks into, a transmission line (Pierre et al., 2014). The same coupled-mode logic reappears in tunable dissipators, where a lossy tunable resonator broadens a high-$\lambda/2$2 mode when the two are brought into resonance (Partanen et al., 2017).

Junction-free implementations obey a different transduction law. In the flux-coupled kinetic-inductance device, the loop screening current $\lambda/2$3 modifies a nonlinear kinetic inductance $\lambda/2$4, giving a quadratic detuning law $\lambda/2$5 around a chosen fluxoid sector (Li et al., 2023). In the local-magnetic-field NbN device, the field-induced kinetic inductance obeys $\lambda/2$6, so the fractional shift follows $\lambda/2$7 (Wang et al., 2024).

3. Tunable coupling, storage, and dissipative engineering

A defining feature of many tRes implementations is that flux can control not only $\lambda/2$8 but also the effective coupling of a protected mode to its environment. The clearest demonstration is the two-resonator storage device of Sandberg and co-workers, where the lifetime of the storage resonator is tuned by more than three orders of magnitude: a field can be stored for $\lambda/2$9 when the coupling resonator is tuned off resonance, and released in $Q$0 when the coupling resonator is tuned on resonance (Pierre et al., 2014). In the large-detuning limit, the storage mode effectively decouples from the line approximately as $Q$1, while near zero detuning the dynamics become non-exponential because coherent inter-cavity exchange competes with external leakage.

Flux-tunable resonators are also used as engineered heat sinks. In the resistor-terminated two-resonator device of Govenius and collaborators, a low-$Q$2 tunable resonator is coupled to a fixed-frequency high-$Q$3 resonator so that photons are rapidly dissipated when the two come into resonance (Partanen et al., 2017). The reported loaded quality factor is tunable from above $Q$4 down to a few thousand at $Q$5, which corresponds to switching the protected mode between a storage-like regime and a rapid reset regime. This usage places the tRes in the broader context of tunable dissipation engineering for resonator and qubit initialization.

A related but distinct role is tunable inter-resonator coupling. The rf-SQUID-mediated coupler of Baust and co-workers realizes a flux-dependent mutual inductance between two $Q$6 transmission-line resonators, with an extracted coupling range from $Q$7 to $Q$8 (Wulschner et al., 2015). At two flux biases, the mediated coupling cancels the residual direct coupling, and the cross-transmission is suppressed by about $Q$9 on resonance. This demonstrates that a tRes-adjacent flux element can act as a switchable and sign-reversible microwave interaction rather than merely as a tunable frequency shifter.

These experiments establish a recurrent principle: a tRes can be designed to tune a mode’s decay rate, its hybridization with other modes, or the sign and magnitude of an intercavity interaction. The significance is methodological as much as functional. In these devices, flux bias is used to reconfigure the network Hamiltonian and the system’s open-system boundary conditions in situ, on demand, and often over orders of magnitude.

4. Parametric modulation and nonlinear operation

Because SQUID inductance is both flux dependent and nonlinear, a tRes is naturally suited to parametric modulation. In the doubly tunable half-wave resonator with a dc SQUID at each end, each boundary can tune the fundamental frequency by approximately $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$0, and two phase-locked pumps applied near twice the fundamental generate photons above a threshold pump amplitude (Svensson et al., 2017). The relative pump phase controls the effective modulation of the two boundaries: “breathing mode” operation at $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$1 lowers the threshold, whereas “vibrating mode” operation at $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$2 suppresses oscillation. The paper also reports deviations from the ideal boundary-modulation theory and attributes them to parasitic couplings that drive SQUID currents directly rather than producing pure local flux modulation. This is one of the clearest documented controversies in the tRes literature, and it has direct layout implications.

Nondegenerate parametric oscillation has been studied in a multimode $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$3 resonator terminated by a dc SQUID, with a pump near the sum of two mode frequencies (Bengtsson et al., 2018). At $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$4, the relevant modes are at $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$5 and $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$6, with total damping rates $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$7 and $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$8. The instability threshold is $$L_J(\Phi) \approx \frac{\Phi_0}{4\pi I_c \left|\cos\!\left(\pi \Phi/\Phi_0\right)\right|},$$9, and the measured oscillation amplitudes and frequency shifts agree quantitatively with the Kerr-nonlinear two-mode model. The free-running oscillation shows strong phase diffusion and broad linewidths, while injection of a weak on-resonance tone reduces the linewidth by more than three orders of magnitude and drives the $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$0 point below the $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$1 resolution bandwidth.

Junction-free kinetic-inductance tRes devices also support parametric processes. The flux-coupled Al loop resonator demonstrates continuous tuning within $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$2 around $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$3, sideband generation under $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$4 modulation, and three-wave mixing under a strong pump near $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$5 (Li et al., 2023). The reported idler rises to approximately $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$6 above the noise floor as pump power is increased, and both nondegenerate and degenerate parametric amplification reach approximately $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$7 gain. The physics differs from the SQUID case in detail, but the operational result is similar: a flux-controlled resonant element simultaneously serves as a tunable mode and a nonlinear mixer.

The broader significance is that the tRes concept unifies static reconfiguration and time-dependent Hamiltonian engineering. The same flux channel that repositions a resonance can, under sufficiently fast or periodic modulation, implement amplification, frequency conversion, sideband generation, or oscillation thresholds.

5. Sensing, transduction, and hybrid integration

Several works use the tRes primarily as a flux transducer. In the capacitively shunted SQUID transducer for high-speed flux sampling, the tRes is a compact on-chip LC resonator probed in reflection so that fast flux pulses are converted to a microwave phase waveform (Foxen et al., 2018). The device is intentionally broadband, with a reported transducer response bandwidth of $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$8, maximum gain of $\tan(k l)=\frac{L_l}{k L_S(\Phi)},$9, flux sensitivity of approximately $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$0, and step-settling resolution better than $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$1. Rather than storing photons, this tRes is used to qualify cryogenic flux lines and package-induced settling in tunable-qubit hardware.

A closely related sensing direction is direct magnetometry. The RF-SQUID quarter-wave tRes reported in 2026 is read out in transmission as a high-contrast notch with near-unity signal-to-background ratio and a measured full width at half maximum of $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$2 at $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$3 (Stack et al., 3 Jun 2026). Two measurement modalities are demonstrated: flux-dependent frequency-modulation spectroscopy and fixed-frequency iso-frequency scans. The reported mutual inductances are $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$4 for the control coil and $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$5 for the target coil. The device is explicitly presented as a low-loss, potentially quantum non-demolition magnetometer operating at millikelvin temperature with MHz magnetic sampling rates.

High-responsivity frequency tuning has also been pursued as an enabling technology for multiplexing and local flux delivery. The flip-chip and on-chip FTR study reports more than $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$6 of tunability, flux responsivities up to approximately $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$7, and flux-transfer efficiencies up to $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$8 for the flip-chip coil and $$f_r(\Phi)\approx \frac{f_0}{1+\gamma(\Phi)}, \qquad \gamma(\Phi)=\frac{L_S(\Phi)}{lL_l},$$9 for the on-chip coil (Paradkar et al., 28 Dec 2025). These devices use large-loop asymmetric SQUIDs to suppress branch switching while maintaining large screening inductance and efficient microamp-scale control. The same paper notes that $$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$0 decreases away from zero flux bias as responsivity grows, consistent with flux-noise-induced broadening.

For hybrid quantum systems, material choice becomes central. The Nb nanoSQUID-embedded resonator operated at $$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$1 retains $$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$2 at $$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$3 and shows a tuning range of $$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$4 with a discontinuous jump at $$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$5 (Potter et al., 2024). The paper identifies dielectric TLS noise, rather than intrinsic SQUID flux noise, as the dominant noise source under current operating conditions, and argues that larger tunability, in the $$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$6–$$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$7 range, would substantially improve flux sensitivity for coupling to small spin clusters.

The tRes can also mediate strong longitudinal interactions with qubits. In the flux-qubit experiment of 2020, a dc-SQUID-embedded lumped resonator around $$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$8–$$L_T(\Phi)=\frac{L_g}{1+\beta_L \cos(2\pi \Phi/\Phi_0)},$$9 is inductively coupled to a flux qubit, yielding a photon-number interaction

$\beta_L=(2\pi I_c L_g)/\Phi_0<1$0

with $\beta_L=(2\pi I_c L_g)/\Phi_0<1$1 per photon (Toida et al., 2020). By increasing the resonator photon number, the qubit transition is tuned by approximately $\beta_L=(2\pi I_c L_g)/\Phi_0<1$2. This is not a dispersive Stark shift in the usual large-detuning sense, but a flux-mediated longitudinal control channel whose sign depends on the qubit flux bias.

6. Limitations, misconceptions, and design trade-offs

A common misconception is that a tRes is necessarily a SQUID-terminated resonator whose only function is to shift its resonance frequency. The published record is broader. Flux tuning can act on an auxiliary coupling cavity rather than on the storage mode itself (Pierre et al., 2014); on a dissipative resonator that functions as a heat sink (Partanen et al., 2017); on an rf-SQUID mutual inductance that reverses the sign of cavity-cavity coupling (Wulschner et al., 2015); or on a junction-free kinetic-inductance loop (Wang et al., 2024, Li et al., 2023). Frequency tuning is the most visible effect, but linewidth tuning, release-rate control, and dispersive transduction are equally central uses.

A second misconception is that stronger flux responsivity is unambiguously beneficial. Several papers document the corresponding penalties. In broadband flux transducers, operation too close to $\beta_L=(2\pi I_c L_g)/\Phi_0<1$3 increases gain but drives the SQUID toward nonlinearity and loss of small-signal validity (Foxen et al., 2018). In large-loop FTRs, increasing geometric inductance raises flux-transfer efficiency but also increases the screening parameter and susceptibility to hysteresis; asymmetric junctions with $\beta_L=(2\pi I_c L_g)/\Phi_0<1$4 are used specifically to suppress branch switching (Paradkar et al., 28 Dec 2025). In Nb nanoSQUID devices, $\beta_L=(2\pi I_c L_g)/\Phi_0<1$5 produces hysteretic tuning with discontinuous branch jumps (Potter et al., 2024). In RF-SQUID magnetometry and coupling applications, maintaining $\beta_L=(2\pi I_c L_g)/\Phi_0<1$6 is explicitly required for nonhysteretic operation (Stack et al., 3 Jun 2026, Wulschner et al., 2015).

A third recurring issue is parasitic coupling. The doubly tunable resonator study attributes unexpected threshold behavior and phase dependence to parasitic current pumping caused by a large low-impedance loop in the layout (Svensson et al., 2017). The 2024 local-field NbN resonator finds that antiparallel bias currents provide smooth tuning, whereas parallel currents mainly introduce Joule heating and a drop in $\beta_L=(2\pi I_c L_g)/\Phi_0<1$7 (Wang et al., 2024). The on-chip-coil FTRs of 2025 retain comparable flux-transfer efficiency to flip-chip devices, but suffer lower $\beta_L=(2\pi I_c L_g)/\Phi_0<1$8 because coil proximity increases parasitic capacitive coupling to the CPW (Paradkar et al., 28 Dec 2025). These results indicate that flux routing, return-current geometry, and unwanted microwave current paths are often as important as the nominal inductive model.

Finally, the literature shows a persistent design trade-off between dynamic range, loss, speed, and material constraints. SQUID-based devices generally offer larger tunability and stronger nonlinearity, but they are more sensitive to flux noise, hysteresis, and junction nonidealities. Junction-free kinetic-inductance devices offer higher current handling and simpler fabrication, but typically require milliamp-scale currents or larger local fields for comparable tuning (Wang et al., 2024, Li et al., 2023). This suggests that the choice of tRes architecture is application specific: fast capture-and-release protocols favor auxiliary-coupler designs (Pierre et al., 2014); reset and initialization favor tunable dissipators (Partanen et al., 2017); GHz-band flux metrology favors low-$\beta_L=(2\pi I_c L_g)/\Phi_0<1$9 transducers (Foxen et al., 2018, Stack et al., 3 Jun 2026); and hybrid spin interfaces motivate field-resilient Nb nanoSQUID implementations (Potter et al., 2024).

In aggregate, the superconducting flux-tunable resonator is best understood not as a single device but as a design paradigm: flux is used to reconfigure the inductive boundary conditions, dissipation channels, or modal interactions of superconducting microwave circuits with sufficient speed and dynamic range to support storage, routing, reset, amplification, and sensing in circuit-QED-scale hardware.