Double Transmon Coupler (DTC) Overview

• Double Transmon Coupler (DTC) is a flux-tunable superconducting architecture that employs two coupler transmons to mediate tunable two-qubit interactions with minimal residual coupling.
• Its design leverages quantum interference to achieve a robust zero-coupling state across a broad range of qubit detunings, effectively reducing crosstalk and control-line complexity.
• Experimental demonstrations show gate times as short as 18–48 ns with fidelities exceeding 99.85%, underscoring its potential for scalable, high-coherence quantum processors.

The Double-Transmon Coupler (DTC) is a flux-tunable superconducting circuit architecture that mediates high-fidelity, fast, and fully tunable two-qubit gates between fixed-frequency transmon qubits, with vanishing residual interaction even for large qubit detunings. The DTC leverages quantum interference between two parallel coupler transmons connected by a small Josephson junction in a shared loop, enabling both transverse (XY) and diagonal (ZZ) interaction strengths to be precisely controlled via a single external flux. Its intrinsic zero-coupling state, broad operating window, and robust decoupling properties make it a leading candidate for scalable, high-coherence quantum processors, addressing longstanding challenges in frequency crowding, crosstalk, and control-line complexity in multiqubit systems (Goto, 2022, Kubo et al., 2022, Kubo et al., 2024, Li et al., 4 Mar 2025, Cai et al., 4 Nov 2025).

1. Circuit Architecture and Hamiltonian

The DTC circuit topology consists of two fixed-frequency computational transmons (Q₁, Q₂) and two fixed-frequency coupler transmons (C₃, C₄ or nodes 3, 4), all capacitively coupled, with C₃ and C₄ connected in series via a weak third Josephson junction (J₅) to form a superconducting loop (Goto, 2022). External magnetic flux Φ_ex threads this loop and modulates the Josephson energy of J₅, providing tunability (see Table 1).

Node	Role	Typical Parameters
Q₁, Q₂	Data qubits	ω₁/2π ≈ 4–7.7 GHz, α₁,₂ ≈ –200 to –300 MHz
C₃, C₄	Coupler modes	ω₃,₄/2π ≈ 7–10.2 GHz
J₅	Weak junction	E_{J5} ≪ E_{J3,4}
Capacitances	Coupling	C₁₃, C₂₄ ≈ 5–6 fF, C₃₄ ≈ 1–5 fF

The quantum Hamiltonian in the node basis is: $$H = 4e^2\,\mathbf{n}^T C^{-1} \mathbf{n} - \sum_{i=1}^4 E_{J,i} \cos \varphi_i - E_{J5} \cos(\varphi_4 - \varphi_3 - \Theta_{\mathrm{ex}})$$ where the capacitance matrix C encodes all inter-node coupling, and Θ_ex = Φ_ex/φ₀ is the reduced flux (Goto, 2022, Li et al., 4 Mar 2025).

After quantization and transformation to the transmon basis, the low-energy effective Hamiltonian contains all relevant qubit, coupler, and interaction terms, including flux-tunable coupler normal modes (labelled p and m) (Li et al., 2024).

2. Coupling Mechanisms and Decoupling Regimes

The DTC enables flux-controlled, variable, and even vanishing effective qubit–qubit exchange coupling via virtual excitation of the coupler modes: $$J_{\mathrm{eff}}(\Phi) \approx \tfrac{1}{2}g_{1p}g_{2p}\left( \tfrac{1}{\Delta_{1p}} + \tfrac{1}{\Delta_{2p}} \right) - \tfrac{1}{2}g_{1m}g_{2m}\left( \tfrac{1}{\Delta_{1m}} + \tfrac{1}{\Delta_{2m}} \right)$$ where the indices p and m correspond to the symmetric and antisymmetric normal modes of the coupler, and Δ_{ik} = ω_i – ω_k (Li et al., 2024).

A key feature is that at a unique bias (Θ_ex^idle), destructive interference in the two virtual coupler paths enforces $J_{\mathrm{eff}} = 0$, simultaneously suppressing transverse and longitudinal (ZZ) coupling between the qubits, independent of their detuning (up to several hundred MHz or more) (Goto, 2022, Kubo et al., 2024). This idle point does not require adjustment of qubit frequencies and can be achieved even for highly detuned qubits—a marked contrast to single-transmon coupler schemes, where ZZ cancellation is only possible in the straddling regime.

The DTC's residual ZZ coupling at idle is suppressed to below 1–10 kHz, with on/off coupling ratios exceeding 10³–10⁴, and the zero-coupling point exhibits remarkable flatness across a wide band of qubit detunings (Kubo et al., 2022, Kubo et al., 2024, Li et al., 4 Mar 2025).

3. Gate Protocols and Performance

Two primary gate schemes are realized with the DTC: (i) DC flux-pulsed CZ gates and (ii) parametric (AC flux or microwave-driven) iSWAP-type gates (Kubo et al., 2022, Li et al., 2024, Li et al., 4 Mar 2025).

CZ Gate: An adiabatic DC flux pulse moves the system from idle (J_eff = 0, ζ_ZZ ≈ 0) to a bias maximizing ZZ, accumulates a conditional π-phase, and returns. Optimal pulse shapes (Slepian, fast adiabatic, or reinforcement-learning-optimized), durations of 18–100 ns, and average fidelities of 99.85–99.99% have been demonstrated, with leakage below 0.01–0.03% (Li et al., 2024, Li et al., 4 Mar 2025).

Parametric iSWAP: An AC flux modulation at frequency |ω_1 – ω_2| resonates the |10⟩ ↔ |01⟩ transition, effecting iSWAP or √iSWAP rotations. Gate times are as short as 24 ns with fidelities >99.99% (Kubo et al., 2022).

Error budgets show incoherent errors (e.g., due to coupler-mode induced qubit relaxation) dominate for shorter gates, scaling linearly with gate duration and with the magnitude of the activated ZZ coupling (Li et al., 4 Mar 2025). The best compromise between speed and leakage suppression is reached around 40–80 ns for CZ, with total infidelity ≲0.1%.

4. Integration in Multiqubit Systems and Scalability

The DTC supports scalable layouts by providing robust suppression of both nearest- and next-nearest-neighbor (NNN) residual couplings in 1D and 2D lattices. In three-qubit arrays coupled via two DTCs, residual ZZ is suppressed to ∼1 kHz for all NNs and NNNs, even with large detunings, and both individual and simultaneous π/2 pulses as well as CZ gates achieve fidelities >99.99% (Kubo et al., 2024).

Compared to single-transmon coupler (STC) schemes, DTC architectures yield up to 50× smaller residual couplings and 100× lower crosstalk-mediated infidelity during parallel gate operations. The zero-coupling point's insensitivity to fabrication imperfections and local frequency variation, coupled with the lack of any requirement for frequency crowding or qubit retuning, supports large-scale deployment (Kubo et al., 2024, Cai et al., 4 Nov 2025).

Control-line multiplexing is enabled by the broad, flat “zero-ZZ” operating point: multiple DTCs can be biased through shared flux lines with local offsets, reducing routing complexity from O(N) (one per coupler) to O(√N) in large 2D arrays. Experimentally, this decreases the number of flex lines for a 100-qubit grid by up to one order of magnitude, with Bell and GHZ state fidelities of 99% and 96%, respectively, in multiplexed configurations (Cai et al., 4 Nov 2025).

5. Engineering Variants and Practical Implementations

Variants such as the capacitively shunted DTC (CSDTC) further optimize idling by shifting the zero-coupling point to zero external flux, eliminating DC bias currents and providing robustness to remnant flux even with imperfect magnetic shielding (Li et al., 4 Mar 2025). With carefully chosen C₃₄, the zero-coupling condition is achieved at Φ_ex=0 and remains flat across ±0.2 Φ₀ flux offsets; single- and two-qubit fidelities above 99.97% and 99.85% are retained over this range.

Physical implementations use lithographically defined α-Ta or Al/AlOx/Al junctions on Si, with Slepian or net-zero pulses for dynamic noise cancellation; reinforcement learning shapes optimize pulse profiles for minimal leakage and depolarization (Li et al., 2024, Li et al., 4 Mar 2025). Error budgets indicate that the dominant gate infidelity is set by coupled-mode induced depolarization, which can be offset with echo techniques and broadband pre-distorted pulses.

6. Comparison with Alternative Coupling Schemes

The DTC fundamentally differs from single-transmon couplers by providing a uniquely adjustable “zero-coupling” point decoupled from qubit detuning and device asymmetry. While STCs require operating in the straddling regime and large direct capacitance for ZZ cancellation—becoming ineffective at large detuning—the DTC’s interference mechanism guarantees cancellation of both transverse and diagonal couplings independent of detuning (Goto, 2022, Kubo et al., 2024).

Additionally, DTCs achieve on/off coupling ratios >10³–10⁴, fast gates (18–48 ns vs. typical 40–150 ns for STC systems), and resilience to frequency crowding and crosstalk, which are principal impediments to surface-code scale-up (Kubo et al., 2022, Li et al., 2024).

7. Novel Functionalities and Emerging Applications

DTCs are not limited to gate mediation. In networked quantum systems, they serve as compact, tunable interfaces between otherwise isolated transmons and transmission lines for single-microwave-photon emission, capture, and detection, supporting rapid reset, high-efficiency metrology, and quantum networking primitives. Conversion between source and detector roles is dynamically configurable by drive parameters, with reset and capture/decode times ≈ 2 μs, end-to-end photon detection efficiencies >95%, and frequency tunability >300 MHz (Campbell et al., 16 Jan 2026).

Potential further applications include implementation of scalable modular quantum error correction codes, surface-code architectures with O(√N) control wiring, and distributed quantum interfacing for clock networks or remote entanglement protocols.

---

References

• (Goto, 2022, Kubo et al., 2022, Kubo et al., 2024, Li et al., 2024, Li et al., 4 Mar 2025, Cai et al., 4 Nov 2025, Campbell et al., 16 Jan 2026)