Multi-Qubit 3D Transmon Devices

• Multi-qubit 3D transmon devices are superconducting quantum systems that embed transmon qubits in 3D microwave cavities to enhance coherence and control.
• They integrate materials like aluminum and graphene-based hBN stacks using circuit QED principles, electron-beam lithography, and precision cavity engineering.
• Key performance metrics include tunable inter-qubit coupling, high-fidelity gate operations, and predictive electromagnetic modeling for scalable quantum architectures.

Multi-qubit 3D transmon devices constitute a class of superconducting quantum architectures in which multiple transmon qubits, engineered for high coherence and controllable coupling, are embedded within three-dimensional microwave cavities. These platforms leverage circuit QED principles to realize tunable, high-fidelity multi-qubit operations, with implementations spanning conventional Al-based Josephson junctions as well as 2D materials such as graphene. Realizing robust, extensible multi-qubit transmon systems necessitates advancements in device architecture, electromagnetic mode engineering, inter-qubit coupling control, and performance modeling, integrating both circuit-level design and full 3D electromagnetic analysis.

1. Core Device Architectures: Materials and Fabrication

Multi-qubit 3D transmon realizations employ a range of Josephson junction technologies and circuit layout strategies optimized for integration within superconducting cavities. In graphene-based architectures, each junction is constructed as an hBN/graphene/hBN heterostructure, transferred onto intrinsic Si/SiO₂ substrates using polymer-free dry methods to preserve the 2D channel’s integrity. Graphene is encapsulated by hBN layers (~20 nm) on both sides to minimize surface contamination and environmental noise (Chiu et al., 24 Dec 2025). Junction definition utilizes electron-beam lithography (EBL) and ICP–RIE etching, followed by NbTi sputtering (120 nm) for low-impedance, edge-connected superconducting contacts.

Two general qubit forms are realized: flux-tunable SQUID loops (two graphene JJs in parallel, loop area ~20 µm² for high flux sensitivity) and fixed-frequency single-JJ devices. Planar shunt capacitors, formed from large Al/NbTi pads (SQUID pad ≈ 590 × 320 µm²; fixed qubit ≈ 400 × 97 µm²), yield simulated shunt capacitances C ≈ 32–92 fF, dictating the charging energy E_C.

Integration occurs in a high-purity copper 3D cavity, typically supporting the TE₁₀₁ mode at f₀ ≈ 6–6.8 GHz, with dedicated SMA drive and readout ports (Q_ext ≈ 6000–7000 for the drive, Q_ext ≈ 1000 for the overcoupled readout). The dielectric chip is placed at the cavity’s E-field antinode to maximize coupling. Non-magnetic packaging and precise chip placement allow for both maximal electric field interaction and magnetic flux-biasing of SQUID loops.

Other approaches use multi-mode circuit implementations, such as the trimon (Josephson ring modulator with four Al/AlOx/Al junctions in a square, six pad-to-pad capacitors for mode structure), which implements three strongly interacting transmons in a compact footprint. The “dimon” (a two-mode analog) serves as a multi-qubit building block in coupled 3D bus cavities (Roy et al., 2016, Hazra et al., 2019).

2. Hamiltonian Framework and Coupling Regimes

The generic Hamiltonian for a transmon (regime E_J ≫ E_C) coupled to a cavity reads:

$$H = 4E_C(\hat n - n_g)^2 - E_J \cos \hat \phi + \hbar \omega_c a^\dagger a + \hbar g (a + a^\dagger) \hat n$$

where E_C = $e^2/2C_\text{total}$ incorporates all pad, junction, and parasitic capacitances, and E_J for SQUIDs is flux tunable as $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$. The dimensionless coupling $g$ is determined by geometry.

In the dispersive limit ($|Δ| ≡ |ω_q − ω_c| ≫ g$), the effective Hamiltonian yields a qubit-state-dependent cavity frequency shift (dispersive shift):

$\chi = \frac{g^2}{Δ}$

In the resonant regime ($Δ \to 0$), strong hybridization leads to vacuum Rabi splitting:

$$\omega_\pm = \frac{ω_q + ω_c}{2} \pm \frac{1}{2} \sqrt{(ω_q - ω_c)^2 + 4g^2}$$

with resonance splitting $\Omega_R = 2g$.

Multi-mode circuits, such as the trimon, are modeled as three weakly anharmonic oscillators with all-to-all longitudinal ($\sigma_z^i \sigma_z^j$) couplings (Roy et al., 2016):

$e^2/2C_\text{total}$0

where $e^2/2C_\text{total}$1 sets the cross-Kerr interaction.

For more scalable designs, the full cavity-QED Hamiltonian includes bus mode(s), cross-resonance exchange, and explicit longitudinal and transverse interactions between collective qubit modes (Hazra et al., 2019).

3. Spectroscopy, Readout, and Gate Operation

Spectroscopic techniques probe the regimes of qubit-cavity interaction. Two-tone spectroscopy detects $e^2/2C_\text{total}$2 transitions, while measurement of $e^2/2C_\text{total}$3 as a function of flux and drive power reveals vacuum Rabi splittings and dispersive regime physics. In graphene-based devices, observed coupling rates are $e^2/2C_\text{total}$4–112 MHz for SQUIDs and 79 MHz for fixed JJs. Dispersive shifts on the order of $e^2/2C_\text{total}$5 MHz ($e^2/2C_\text{total}$6) support high-fidelity, single-shot readout (Chiu et al., 24 Dec 2025).

In coupled multi-qubit systems, power-dependent measurements reveal multi-stage dispersive shifts, with the cavity resonance shifting sequentially as successive qubits saturate critical photon thresholds:

$e^2/2C_\text{total}$7

In the trimon, always-on longitudinal couplings shift each qubit's transition depending on partner states, directly enabling CNOT gates via single-tone drives and supporting native SWAP operations. Measured Bell-state fidelities reach $e^2/2C_\text{total}$8, with SWAP fidelities of $e^2/2C_\text{total}$9 (Roy et al., 2016).

In 3D bus-cavity architectures with distinct “blocks,” cross-resonance gates are mediated by controlled, microwave-driven exchange between modes; the effective gate strength is tunable via drive amplitude and detuning, achieving $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$0 operations in ~200 ns at fidelities of $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$1 (standard RB) and $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$2 (interleaved, on-gate) (Hazra et al., 2019).

4. Electromagnetic Modeling and Coupling Rate Quantification

Engineering multi-qubit 3D-transmon systems requires predictive modeling of qubit-qubit coupling rates, essential for entanglement speed, gate design, and crosstalk suppression. Field-based macroscopic quantum electrodynamics (QED) formalism expresses the effective exchange rate $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$3 as an explicitly geometry-dependent functional of the electromagnetic dyadic Green’s function connecting qubit locations:

$E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$4

where $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$5 is the charge matrix element for qubit $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$6, $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$7 is its transition frequency, and $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$8 is the transfer impedance between ports $E_J(\Phi) = E_{J,\max}|\cos(\pi\Phi/\Phi_0)|$9 and $g$0 computed via 3D EM simulation (Khan et al., 2024). This approach validates $g$1 against direct numerical diagonalization and experimental data (e.g., $g$2 MHz predicted vs. $g$3 MHz measured in a four-qubit finger-capacitor device).

The formalism extends to multi-path, multi-coupler layouts, supporting predictive crosstalk management and zero-ZZ operating point identification. For complex 3D-cavity-coupled transmon lattices, the method circumvents computational bottlenecks inherent to standard eigenmode solvers, enabling routine design of large-scale multi-qubit devices.

5. Coherence, Crosstalk, and Performance Metrics

For graphene-based qubits, relaxation times $g$4 are observed at $g$548 ns (at $g$6 GHz) and dephasing times $g$7 ns, with $g$8 dominated by low-frequency flux noise amplitude $g$9—substantially larger than for Al-based junctions (Chiu et al., 24 Dec 2025). In trimon-type devices, coherence times reach $|Δ| ≡ |ω_q − ω_c| ≫ g$0 = 20–51 μs, $|Δ| ≡ |ω_q − ω_c| ≫ g$1 = 32–65 μs (Ramsey/Hahn-echo) depending on mode and device (Roy et al., 2016).

Purcell-protected qubits (B, C modes) demonstrate substantially longer $|Δ| ≡ |ω_q − ω_c| ≫ g$2 due to minimal coupling to cavity decay channels. Readout is typically implemented via overcoupled transmission mode and quantum-limited parametric amplification, with $|Δ| ≡ |ω_q − ω_c| ≫ g$3 and $|Δ| ≡ |ω_q − ω_c| ≫ g$4 distinguishable at ~99% fidelity; $|Δ| ≡ |ω_q − ω_c| ≫ g$5 and $|Δ| ≡ |ω_q − ω_c| ≫ g$6 require SWAP-initialization for discrimination where $|Δ| ≡ |ω_q − ω_c| ≫ g$7 is degenerate.

Strategic modeling of the exchange interaction $|Δ| ≡ |ω_q − ω_c| ≫ g$8 directly links device layout and electromagnetic design to crosstalk rates. In multi-coupler topologies, appropriately choosing coupler detunings ($|Δ| ≡ |ω_q − ω_c| ≫ g$9) can realize zero-ZZ interaction points, reducing correlated dephasing (Khan et al., 2024).

6. Scalability and Modular Architectures

The techniques demonstrated for single- and two-qubit 3D transmons generalize to larger arrays. The hBN/graphene/hBN–NbTi edge-contact technology is extensible to linear and two-dimensional qubit arrays, each coupled to distinct or shared 3D cavity modes. Employing multi-cell or multi-mode cavities enables frequency or spatial multiplexing, allowing both individual and collective readout strategies for scalable quantum processors (Chiu et al., 24 Dec 2025).

Multi-modal circuit blocks (trimon/dimon) can be tiled, offering all-to-all longitudinal coupling within each module and controlled exchange interactions between blocks via 3D bus cavities. Control wiring complexity is mitigated through multi-tone sideband modulation from a single local oscillator, scaling efficiently with cluster size and supporting native error-correcting codes and annealing protocols (Roy et al., 2016, Hazra et al., 2019).

Careful layout is required to suppress unwanted direct inter-qubit capacitance, manage mutual inductance among flux-bias lines, and position qubits at distinct cavity field antinodes for selective g_i engineering. The ability to simulate and optimize $\chi = \frac{g^2}{Δ}$0 for arbitrary qubit and cavity configurations using impedance-based field-theoretic methods represents a significant enabling advance for scaling (Khan et al., 2024).

7. Outlook: Integration of 2D Materials and Hybrid Functionality

A salient advantage of 2D-material-based JJs (graphene, encapsulated semiconductors) lies in their gate-tunability, potential for in situ frequency and coupling control, and integration with materials exhibiting topological or semiconducting behavior. This provides a route toward tunable multi-qubit interactions, embedding hybrid quantum systems into the 3D transmon platform. While coherence times in current graphene-based devices are limited by flux noise and interfacial loss, further improvements in materials and electromagnetic design could close the gap with conventional Al-based qubits, advancing the realization of scalable, high-coherence 2D-material quantum processors (Chiu et al., 24 Dec 2025).

In summary, multi-qubit 3D transmon devices encompass a set of scalable superconducting quantum architectures unifying advanced materials science, precision cavity engineering, multi-mode circuit design, and quantitative electromagnetic modeling, enabling programmable, high-fidelity quantum information processing with extensibility to larger, fault-tolerant systems.