---
title: All-Fluxonium Cross-Resonance Architecture
url: https://www.emergentmind.com/topics/all-fluxonium-cross-resonance-architecture
type: topic
---

# All-Fluxonium Cross-Resonance Architecture

All-fluxonium cross-resonance architecture denotes a class of superconducting quantum-computing designs in which both the control and target qubits are fluxoniums, while the entangling interaction is generated by a microwave drive applied near the target-qubit transition frequency. Across the reported implementations and proposals, the central objective is to obtain a drive-activated conditional $ZX$ interaction together with strongly suppressed static $ZZ$, while retaining fluxonium’s long coherence, large anharmonicity, and operation at the half-flux-quantum sweet spot [2407.15450] [2202.04583] [2201.09374] [2603.17936]. The architecture has been studied in inductively coupled devices that emulate transversely coupled spin-$1/2$ systems, in capacitively coupled devices using selective darkening, and in scalable multipath-coupled processor layouts that explicitly target large detuning bandwidth and fabrication robustness [2407.15450] [2202.04583] [2201.09374].

## 1. Conceptual definition and operating regime

In this architecture, each computational node is a fluxonium qubit: a loop containing a Josephson junction and a large shunt inductance, with computational states defined by the two lowest eigenstates at half-flux bias. Reported computational transition frequencies lie well below those of conventional transmon-based processors. One experimentally characterized inductively coupled pair operated at $f_{01}^A=150\,\text{MHz}$ and $f_{01}^B=230\,\text{MHz}$ [2407.15450], while processor-level proposals place $\omega_{01}/2\pi$ in the range of approximately $0.5$ to $1.2\,\text{GHz}$ [2201.09374] and a selective-darkening study used $\omega_{01}^A/2\pi=0.53\,\text{GHz}$ and $\omega_{01}^B/2\pi=1.02\,\text{GHz}$ [2202.04583].

The cross-resonance mechanism is activated by driving the control qubit at a frequency close to the target qubit’s $|0\rangle\leftrightarrow|1\rangle$ transition. After projection into the computational subspace and transformation to an appropriate rotating frame, the dominant driven interaction takes the form
$$
H_{\mathrm{CR,eff}} = h\,[m\,I\!\otimes X + \mu\,Z\!\otimes X + \delta I\, Z\!\otimes I + \zeta_{ZZ}\, Z\!\otimes Z]
$$
in the scalable multipath treatment [2201.09374], or equivalently
$$
H_{\mathrm{eff}} = \tfrac12 \xi_{ZX}\,\sigma_z^A\!\otimes \sigma_x^B + \tfrac12 \xi_{IX}\,I^A\!\otimes \sigma_x^B + \tfrac14 \xi_{ZZ}\,\sigma_z^A\!\otimes \sigma_z^B + \cdots
$$
in the selective-darkening analysis [2202.04583]. The architectural target is therefore not merely a nonzero entangling rate, but a hierarchy in which the conditional $ZX$ term is appreciable, the always-on $ZZ$ term is minimized, and unwanted single-qubit terms are either suppressed by design or corrected in calibration [2201.09374] [2202.04583].

A recurring feature is operation at the half-flux-quantum sweet spot. The capacitive-only analysis explicitly states that drive-activated cross-resonance gates preserve qubits at their half-flux sweet spot, with minimal dephasing [2603.17936]. This suggests that the architecture is intended to combine the microwave-only tunability of cross-resonance with the low dephasing typically associated with sweet-spot biasing.

## 2. Circuit models and coupling topologies

Three coupling topologies appear in the cited work.

First, the experimentally characterized inductive device consists of two fluxonium qubits, each with shunt inductance $L_{1,2}$, Josephson energy $E_{J1,2}$, and shunt capacitance $C_{1,2}$, sharing a common junction of inductance $L_M$ that produces a mutual inductive coupling $J_L$. The antenna pads simultaneously create weak stray capacitances, producing a capacitive interaction $J_C$ and a spurious lumped $LC$ mode [2407.15450]. In node-flux variables, the qubit Hamiltonians are
$$
H_\alpha \equiv h\!\left(4E_{C,\alpha} n_\alpha^2 + \tfrac12 E_{L,\alpha}\phi_\alpha^2 - E_{J,\alpha}\cos(\phi_\alpha-\phi_{\mathrm{ext}})\right),\qquad \alpha=A,B,
$$
with permanent couplings
$$
H_C = h\,J_C\,n_A n_B,\qquad H_L = h\,J_L\,\phi_A \phi_B.
$$
For that device, the reported parameters were
$J_C \simeq -0.038\,\text{GHz}$,
$J_L \simeq +0.004\,\text{GHz}$,
$E_{J,A}=5.59\,\text{GHz}$,
$E_{L,A}=0.76\,\text{GHz}$,
$E_{C,A}=0.98\,\text{GHz}$,
$E_{J,B}=6.27\,\text{GHz}$,
$E_{L,B}=1.16\,\text{GHz}$,
and
$E_{C,B}=0.99\,\text{GHz}$ [2407.15450].

Second, the selective-darkening proposal studies a purely capacitively coupled pair with
$$
H_0 = H_A^{(0)} + H_B^{(0)} + V_C,\qquad V_C = J_C\,\hat n_A \hat n_B,
$$
and a microwave drive
$$
H_{\mathrm{drive}}(t)=2\hbar f(t)\cos(\omega_d t)\,[\,\hat n_A+\eta \hat n_B\,].
$$
In its representative parameter set, $J_C/h=350\,\text{MHz}$ and the static $ZZ$ rate is $\xi_{ZZ}/2\pi\approx 2\,\text{MHz}$ [2202.04583].

Third, the scalable processor proposal introduces a two-path coupling network designed to combine a sizable exchange interaction with static-$ZZ$ cancellation. In that setting,
$$
H_{\mathrm{coupl}} = h[J_C n_A n_B - J_L \phi_A \phi_B],
$$
with $J_L/2\pi\approx 2\,\text{MHz}$ and $J_C/2\pi\approx 11.5\,\text{MHz}$. The capacitive and inductive contributions are tuned so that $\zeta_{ZZ}\to 0$, while maintaining an effective exchange interaction $J_{\mathrm{eff}}\approx 10\,\text{MHz}$ between $|01\rangle$ and $|10\rangle$ [2201.09374].

The capacitive-only 2026 analysis further isolates a fixed-coupling architecture in which the two-qubit Hamiltonian is
$$
H(t)=H_c(\Omega,t)+H_t+J\,n_c\otimes n_t,
$$
with the control driven as
$$
H_c(\Omega,t)=H_c^0+\Omega \sin(\omega_d t)\,n_c.
$$
There the emphasis is on deriving a simple upper-bound estimate for the conditional interaction strength under strong driving [2603.17936].

## 3. Effective spin mapping and the origin of the cross-resonance interaction

A defining observation of the inductively coupled experiment is that two inductively coupled fluxoniums can behave very closely to two transversely coupled spin-$1/2$ systems [2407.15450]. When the full Hamiltonian is projected onto the computational basis $\{|00\rangle,|01\rangle,|10\rangle,|11\rangle\}$, the effective two-qubit Hamiltonian contains the qubit frequencies, a residual static $ZZ$ term, and a transverse exchange term generated by the inductive coupling through the flux operators:
$$
H_{XX}\simeq g(\sigma_+^A\sigma_-^B+\mathrm{h.c.}) \Rightarrow \tfrac12 g\,\sigma_x^A\sigma_x^B,
$$
with
$$
g = J_L \langle 0|\phi_A|1\rangle \langle 0|\phi_B|1\rangle.
$$
Using flux matrix elements $\phi_{01}^\alpha\approx 0.4$--$0.5$, the reported value was $g/2\pi\approx 0.4\,\text{MHz}$ [2407.15450].

Under microwave driving, the control-target detuning converts this underlying hybridization into a conditional target rotation. In the inductive device, driving qubit $A$ near $f_{01}^B$ and applying a Schrieffer-Wolff expansion for $\Delta=f_{01}^A-f_{01}^B\simeq -80\,\text{MHz}$ yields
$$
H_{\mathrm{CR}}/h \simeq A\,\sigma_z^A\sigma_x^B + B\,\sigma_z^A + C\,\sigma_x^B,
$$
where, to leading order,
$$
A \simeq \xi_B^- (\epsilon_A/\Delta),\qquad
B \simeq \xi_A^+ (\epsilon_A/f_{01}^A),\qquad
C \simeq \xi_B^+ (\epsilon_A/f_{01}^B).
$$
The same source states that choosing $\epsilon_A$ only can suppress $B$ and $C$ and maximize the true $ZX$ term [2407.15450].

The selective-darkening formulation reaches the same general structure from a different route. There, the driven interaction is engineered so that the matrix element for the target transition conditioned on control state $|0\rangle$ vanishes:
$$
\langle 00|H_{\mathrm{drive}}|01\rangle=0
\quad\Rightarrow\quad
\eta_{\mathrm{opt}}=
-\frac{\langle 00|\hat n_A|01\rangle}{\langle 00|\hat n_B|01\rangle}.
$$
For the tabulated device, $\eta_{\mathrm{opt}}\simeq -(0.14)/(0.22)\simeq -0.64$, so $\epsilon_t\approx -0.64\,\epsilon_c$ [2202.04583]. Under this condition, the darkened transition suppresses the unwanted $IX$ component and leaves a native CNOT generator up to local $Z$ rotations [2202.04583].

The strong-drive capacitive analysis gives an additional nonperturbative perspective. It states that the conditional $ZX$ rate saturates near
$$
g_{ZX}^{\max}\simeq J\,|n_1^{10}n_2^{10}|,
$$
which implies a minimum CNOT time
$$
\tau_{\mathrm{CNOT}}^{\min}\simeq \frac{\pi/2}{J\,|n_1^{10}n_2^{10}|}.
$$
The physical interpretation given there is that the strong off-resonant control drive induces a conditional polarization $\Delta p\to 1$, and the target is driven through the weak $J$ coupling by that oscillating polarization [2603.17936]. This suggests a unifying picture: the architecture relies on weak fixed hybridization to encode control-state dependence, then converts that dependence into a target rotation by resonant or near-resonant microwave driving.

## 4. Static $ZZ$ suppression as a central design objective

Suppression of always-on $ZZ$ coupling is a primary architectural criterion, but the cited works achieve it by different mechanisms.

In the inductively coupled experiment, the residual shift is written as
$$
\zeta_{ZZ}^{\mathrm{static}}
=
(E_{00}+E_{11}-E_{01}-E_{10})/h.
$$
The paper attributes its smallness to cancellation among virtual transitions involving noncomputational states. Although static $ZZ$ generally scales through second-order processes involving $|2\rangle$ levels, the relevant flux matrix elements satisfy $\phi_{12}\ll \phi_{01}$, so the contributions nearly cancel; numerically, $\zeta_{ZZ}^{\mathrm{static}}\simeq 2\,\text{kHz}$, confirmed by conditional Ramsey [2407.15450].

In the scalable processor proposal, static $ZZ$ suppression is engineered explicitly through the multipath coupler. A shared-superinductor path contributes one dispersive $ZZ$ channel, while a small capacitive path contributes another with opposite sign. When $J_C$ is tuned appropriately, $\zeta_{ZZ}\equiv \omega_{00}+\omega_{11}-\omega_{01}-\omega_{10}\to 0$ [2201.09374]. The same study reports that over $\pm 10\%$ parameter drift, $|\zeta_{ZZ}|$ remains below $1\,\text{kHz}$ [2201.09374].

By contrast, the selective-darkening work does not eliminate static $ZZ$ at the hardware level in its main numerical example. For $J_C/h=350\,\text{MHz}$, it reports $\xi_{ZZ}/2\pi\approx 2\,\text{MHz}$ [2202.04583]. There, high-fidelity operation is instead obtained by pulse design, selective darkening, and software $Z$ rotations before and after the gate [2202.04583]. A common misconception is therefore that all all-fluxonium cross-resonance schemes inherently exhibit negligible static $ZZ$. The cited literature does not support that blanket claim: negligible $ZZ$ is achieved in some designs by matrix-element structure or multi-path cancellation, whereas capacitively coupled designs may tolerate a larger residual $ZZ$ and compensate at the control level [2407.15450] [2202.04583] [2201.09374].

The 2026 capacitive-only treatment occupies an intermediate position. It states that residual $ZZ$ is $O(J^2)$ in the Floquet-Schrieffer-Wolff treatment, and reports a design point with residual always-on $ZZ\approx 50\,\text{kHz}$ for $J/2\pi\approx 100\,\text{MHz}$ [2603.17936]. This suggests that capacitive-only all-fluxonium cross-resonance need not imply MHz-scale $ZZ$, but the achievable value depends strongly on parameter choice and optimization criterion.

## 5. Gate implementations, pulse prescriptions, and reported performance

The architecture admits several concrete gate constructions.

The selective-darkening protocol realizes a CNOT by simultaneously driving control and target at the target frequency with amplitudes satisfying the darkening condition. A Gaussian-derivative-subtracted envelope is used,
$$
\epsilon_c(t)=A\left\{\exp\!\left[-\frac{(t-t_{\mathrm{gate}}/2)^2}{2\sigma^2}\right]-
\exp\!\left[-\frac{(t_{\mathrm{gate}}/2)^2}{2\sigma^2}\right]\right\},
\qquad
\epsilon_t(t)=\epsilon_c(t)/\eta_{\mathrm{opt}},
$$
with $\sigma=t_{\mathrm{gate}}/4$ and total amplitude chosen so that $\Omega_{\mathrm{CR}} t_{\mathrm{gate}}=\pi$ [2202.04583]. For $t_{\mathrm{gate}}=50\,\text{ns}$, $J_C/h=350\,\text{MHz}$, $T_1^{01}=100\,\mu\text{s}$, and $T_1^{12}=1\,\mu\text{s}$, the reported total fidelity is $F_{\mathrm{tot}}\approx 99.8\%$; improving to $T_1^{01}=500\,\mu\text{s}$ and $T_1^{12}=50\,\mu\text{s}$ pushes the error below $10^{-4}$ [2202.04583]. The same work reports coherent infidelity decreasing from approximately $3\times 10^{-3}$ at $20\,\text{ns}$ to approximately $2\times 10^{-5}$ at $50\,\text{ns}$ [2202.04583].

The scalable processor proposal uses a flat-top control drive with cosine ramps and quotes an analytical small-drive estimate
$$
\mu(\epsilon_d,\Delta)\approx (2\pi J_{\mathrm{eff}}/\Delta)\,\epsilon_d.
$$
For $J_{\mathrm{eff}}/2\pi=11\,\text{MHz}$ and detuning $\Delta/2\pi$ up to $1\,\text{GHz}$, it reports $\mu/2\pi\gtrsim 5\,\text{MHz}$ at small $\Delta$ and $\gtrsim 1\,\text{MHz}$ even at $\Delta=1\,\text{GHz}$ for $\epsilon_d/2\pi=200\,\text{MHz}$ [2201.09374]. Corresponding CNOT gate times are reported as approximately $100\,\text{ns}$ to $300\,\text{ns}$, with simulated coherent errors $\lesssim 10^{-2}$ across $\Delta\in[20\,\text{MHz},1\,\text{GHz}]$ and $\lesssim 10^{-5}$ at small detuning; leakage is reported below $10^{-6}$ for $\tau_g\ge 100\,\text{ns}$ [2201.09374].

The inductively coupled experimental study gives a lower-rate but very low-$ZZ$ operating point. Its summary reports conditional $ZX\simeq 0.4\,\text{MHz}$ for equal drive amplitude $\epsilon$, static $ZZ\simeq 2\,\text{kHz}$, and a cross-resonance gate time $t_{\mathrm{CR}}\simeq 200$--$250\,\text{ns}$, with simulated two-qubit fidelities exceeding $0.99$ and limited by $T_1\approx 5$--$10\,\text{ms}$ of fluxonium [2407.15450].

The strong-drive capacitive-only analysis reports a somewhat different optimization frontier. For half-flux-bias parameter choices with $E_J/E_C\approx 3$--$8$, $E_L/E_J\approx 0.10$--$0.25$, $\omega_c/2\pi\approx 500\,\text{MHz}$, $\omega_t/2\pi\approx 800\,\text{MHz}$, and a residual-$ZZ$ budget $|\mu_{ZZ}|/2\pi\approx 50\,\text{kHz}$, it sets $J/2\pi\approx 100\,\text{MHz}$ and predicts $\tau_{\mathrm{CNOT}}\approx 130$--$230\,\text{ns}$, typically under $200\,\text{ns}$ [2603.17936]. Time-domain simulations with soft-square $\Omega(t)$ confirm $\tau_{\mathrm{CNOT}}\approx 160\,\text{ns}$ at optimum drive, with coherent infidelity $\lesssim 10^{-4}$ [2603.17936].

Taken together, these results show that all-fluxonium cross-resonance admits both low-rate, ultra-low-$ZZ$ regimes and faster, more strongly driven regimes with higher hardware coupling. The trade-off is explicit in the published numbers rather than merely qualitative.

## 6. Readout, control infrastructure, spurious modes, and scalability

Processor-level implementations pair the cross-resonance interaction with a specific readout and control stack. In the scalable proposal, each fluxonium is dispersively coupled to its own $\lambda/4$ coplanar-waveguide resonator with $\omega_R/2\pi=7$--$9\,\text{GHz}$, and four resonators are capacitively connected to a common $50\,\Omega$ readout bus [2201.09374]. The bare resonator linewidth is $\kappa/2\pi\sim 2\,\text{MHz}$, and no Purcell filters are needed because $\omega_{01}\ll \omega_R$ [2201.09374]. For $g/2\pi=100\,\text{MHz}$, the typical dispersive shift is $\chi_{01}/2\pi\ge 1\,\text{MHz}$ across $0$--$1\,\text{GHz}$ qubit frequencies, while the quoted thermal-photon dephasing estimate gives $T_\phi^{\mathrm{th}}>0.5\,\text{ms}$ for $\omega_R\sim 8\,\text{GHz}$ at $T=50\,\text{mK}$ [2201.09374].

Control is supplied by diplexed on-chip lines carrying both DC bias and RF pulses, with a symmetric “hole-in-ground” geometry used to null stray capacitance to ground and neighboring lines [2201.09374]. The same source argues that operating below $1\,\text{GHz}$ pushes microwave crosstalk from wire-bond and package modes below levels seen at $5\,\text{GHz}$, especially with tightly shielded $50\,\Omega$ RF lines [2201.09374].

An experimentally important caveat is the appearance of spurious modes generated by the physical interconnect structure. The inductively coupled device exhibited a bosonic $LC$ mode at $f_{LC}\simeq 3.22\,\text{GHz}$, arising from the coupling inductance together with capacitive links among qubit terminals [2407.15450]. In the more complete Hamiltonian,
$$
H_{\mathrm{full}}
=
H_A+H_B+h(J_C n_A n_B + J_L \phi_A\phi_B)
+h f_{LC} a^\dagger a
-i h[g_A n_A + g_B n_B](a-a^\dagger),
$$
with $g_A\simeq +0.18\,\text{GHz}$ and $g_B\simeq -0.21\,\text{GHz}$ [2407.15450]. Two-tone spectroscopy showed weak anticrossings with the $|1\rangle\leftrightarrow|2\rangle$ lines of both qubits, shifting noncomputational levels by a few MHz and producing extra lines in high-power scans [2407.15450]. The same work states that the mode does not materially affect the cross-resonance gate when far detuned from the $0.15$--$0.23\,\text{GHz}$ computational band, but should be considered carefully in future designs [2407.15450].

Scalability analyses focus strongly on collision statistics and fabrication tolerance. The multipath processor study imposes frequency-allocation constraints including addressability $\Delta_{ij}\ge 20\,\text{MHz}$, two-photon separation $\ge 10\,\text{MHz}$, two-qubit detuning in $[20\,\text{MHz},1\,\text{GHz}]$, and drive-spectator detuning thresholds of $20/10\,\text{MHz}$ [2201.09374]. Under those assumptions, it reports a cross-resonance cell yield $y_{\mathrm{cell}}\approx 0.98$ for $\sigma_f=20\,\text{MHz}$, scaling to a device yield of approximately $60\%$ for $10^4$ qubits at $\sigma_f=42\,\text{MHz}$ [2201.09374]. The capacitive-only 2026 study similarly analyzes control-target, control-spectator, and multiphoton collision windows and reports that zero-collision yield remains above $50\%$ for device sizes up to distance-$21$ provided $\sigma(E_J)/\langle E_J\rangle\lesssim 1$--$2\%$ [2603.17936]. By contrast, the same source states that analogous transmon cross-resonance layouts require $\lesssim 0.3\%$ relative standard deviation for comparable yield [2603.17936].

A final misconception is that all-fluxonium cross-resonance necessarily requires mixed inductive-capacitive coupling. The literature does not support that as a necessity. Inductively dominated devices can realize the desired spin-$1/2$ analogy with nearly absent static $ZZ$ [2407.15450]; capacitive schemes can realize selective-darkening CNOT gates [2202.04583]; multipath coupling can be used to cancel $ZZ$ while keeping exchange large [2201.09374]; and capacitive-only fixed-coupling architectures have been analyzed as viable routes to sub-$200\,\text{ns}$ CNOT gates with residual $ZZ\lesssim 50\,\text{kHz}$ [2603.17936]. The common architectural theme is therefore not a unique coupler topology, but the combination of fluxonium qubits, fixed interqubit coupling, and microwave-activated conditional dynamics.

Source: https://www.emergentmind.com/topics/all-fluxonium-cross-resonance-architecture