Transmon Molecule Circuit Design

• Transmon molecule circuits are composite superconducting architectures that combine multiple transmon-like modes to create engineered nonlinear elements.
• They leverage strong internal couplings, such as ZZ and cross-Kerr interactions, to enable efficient qubit operations and optimized readout schemes.
• Hybrid implementations use microscopic weak links like carbon nanotubes and quantum dots to explore non-perturbative Hamiltonian regimes and design trade-offs.

A transmon molecule circuit is a superconducting-circuit architecture in which the transmon paradigm is extended from a single weakly anharmonic mode to a composite object with engineered internal structure. In current arXiv literature, the expression is used for several closely related cases: coupled transmons sharing a cavity mode, compact multimode devices such as the trimon, qubit–ancilla composites engineered for non-perturbative cross-Kerr readout, and transmons whose Josephson element is itself a molecular or hybrid weak link, including carbon nanotubes, quantum dots, and van der Waals superconductors (Maurya et al., 4 Mar 2026, Riechert et al., 3 Mar 2025, Dassonneville et al., 2022, Blumenthal et al., 27 Jan 2026). This suggests that the term denotes a family of circuit motifs rather than a single standardized topology.

1. Scope and terminology

Across the cited literature, the defining idea is the replacement of the isolated transmon junction-plus-shunt picture by a structured nonlinear element or by several strongly interacting transmon-like modes. In one usage, a transmon molecule is a pair or trio of transmon-like modes inside one superconducting object, with strong internal $ZZ$ or cross-Kerr couplings (Roy et al., 2016, Maurya et al., 4 Mar 2026). In another, it is a transmon-style qubit whose nonlinear element is a single molecule or a microscopic conductor, such as a carbon nanotube or a quantum-dot Josephson junction (Riechert et al., 3 Mar 2025, Bargerbos et al., 2022). In a third, it is a readout-optimized bimodal circuit whose qubit and ancilla modes hybridize with a cavity to produce polaritonic meters (Dassonneville et al., 2022, Mori et al., 4 Jul 2025).

Usage in the literature	Defining feature	Representative work
Coupled-transmon module	Two transmon qutrits coupled via a single cavity mode	(Ye et al., 2018)
Compact multimode circuit	One device with several transmon-like modes and strong internal $ZZ$ coupling	(Roy et al., 2016, Maurya et al., 4 Mar 2026)
Readout-oriented molecule	Qubit mode + ancilla mode + cavity/polariton structure with non-perturbative cross-Kerr	(Dassonneville et al., 2022, Mori et al., 4 Jul 2025, Mori et al., 5 Sep 2025)
Molecular or hybrid weak link	Carbon nanotube, quantum dot, or vdW superconductor as the Josephson element	(Riechert et al., 3 Mar 2025, Bargerbos et al., 2022, Blumenthal et al., 27 Jan 2026)
Complementary transmon species	Inductively shunted transmon used to tailor $ZZ$, sidebands, or coupling parity	(Richer et al., 2017, Fasciati et al., 2024)

A common ambiguity follows directly from this breadth: “molecule” may refer either to the internal normal modes of one compact superconducting circuit or to the microscopic object inserted into the Josephson element. Both usages appear explicitly in the literature above.

2. Hamiltonian framework

At the circuit level, most realizations remain anchored in the standard transmon Hamiltonian

$$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$

with the transmon regime defined by

$\frac{E_J}{E_C} \gg 1,$

and with approximate lowest transition frequency and anharmonicity

$$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$

In gatemons and related hybrid devices, the central generalization is that $E_J$ becomes a function of a control parameter such as gate voltage $V_g$, i.e. $E_J(V_g)$, because the weak link hosts Andreev bound states or quantum-dot states rather than a conventional tunnel barrier (Riechert et al., 3 Mar 2025).

Flux-tunable hybrid transmons preserve the same circuit-QED description while replacing the Josephson element. A representative model is the dressed transmon–cavity Hamiltonian

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

used for a flux-tunable transmon whose nonlinear inductive element is an Al/AlO$ZZ$0/4Hb-TaS$ZZ$1 Josephson junction (Blumenthal et al., 27 Jan 2026).

For intrinsically multimode “molecular” circuits, the effective Hamiltonian is often written directly in number operators. In the planar trimon, the low-energy description is

$ZZ$2

so that the internal interaction is purely longitudinal in the qubit subspace and appears as strong all-to-all $ZZ$3 coupling (Maurya et al., 4 Mar 2026).

A further generalization appears in inductively shunted designs, where flux can be used to choose between pure transverse and pure longitudinal coupling to an embedded harmonic mode. This provides a building block in which one coupling type can be used for readout and the other for qubit–qubit interaction (Richer et al., 2017).

3. Coupled and multimode transmon realizations

The simplest transmon molecule in the coupled-device sense is a pair of transmon qutrits coupled by a single cavity or transmission-line resonator. In such a system, the cavity couples the $ZZ$4 transition, classical pulses address $ZZ$5 or $ZZ$6, and resonant control prepares states of the form

$ZZ$7

Using realistic parameters, the total operation time was estimated as $ZZ$8 ns, and for $ZZ$9 $ZZ$0s and $ZZ$1 $ZZ$2s the fidelity exceeds $ZZ$3 (Ye et al., 2018).

A more compact interpretation is the trimon: a single superconducting element with three transmon-like modes hosted by a ring of four identical Josephson junctions and four large capacitor pads. In the original implementation, the device realizes three qubits with all-to-all longitudinal coupling, with extracted pairwise couplings $ZZ$4 MHz, $ZZ$5 MHz, and $ZZ$6 MHz, while two of the three modes are protected against Purcell decay (Roy et al., 2016). The planar trimon extends this to a planar geometry and emphasizes Hamiltonian control: the device features three transmon-like modes with strong all-to-all $ZZ$7 coupling, implements all 16 two-qubit Pauli operators in the two-qubit space through multi-tone driving, and supports qudit operation with up to 8 states (Maurya et al., 4 Mar 2026).

The multimode interpretation has two important consequences. First, conditional transition-frequency splittings $ZZ$8 of order a few hundred MHz make the internal transitions spectrally resolvable, so conditional rotations, Raman $ZZ$9, and $$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$0 gates become native primitives rather than perturbative effective interactions. Second, the same device can be viewed either as a three-qubit register or as a single qudit whose computational manifold is built entirely from $$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$1 and $$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$2 of each mode, avoiding high-lying transmon levels and their usual decoherence penalties (Maurya et al., 4 Mar 2026).

4. Readout-oriented transmon molecules

A distinct branch of the literature uses the transmon molecule as a readout device. In one implementation, two nominally identical transmons are coupled so that their normal modes behave as a protected qubit mode and a weakly anharmonic ancilla mode. The ancilla couples strongly to a 3D cavity, whereas the qubit couples to the ancilla through a non-perturbative longitudinal interaction. The resulting Hamiltonian is

$$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$3

and cavity–ancilla hybridization produces nonlinear polaritonic meters (Dassonneville et al., 2022). In the mesoscopic bifurcation regime, this architecture achieved a single-shot fidelity of $$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$4 with an integration time of $$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$5 ns and no requirement for an external quantum-limited amplifier (Dassonneville et al., 2022).

High-power operation of the same basic architecture pushes the readout mode deep into the many-photon regime. A readout fidelity of $$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$6 with 89 photons utilizing a parametric amplifier was reported, and at this elevated photon number the QND nature remained high at $$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$7. Even with up to 300 photons, the QNDness was only reduced by a few percent, with a theoretical critical value $$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$8 photons for the sample’s parameters (Mori et al., 4 Jul 2025).

A more recent analysis isolates the symmetry responsible for this robustness. After eliminating the ancilla-like mode, the effective two-mode Hamiltonian becomes

$$\hat H_\text{transmon} = 4E_C(\hat n-n_g)^2 - E_J\cos\hat\varphi,$$9

which at zero flux reduces to a parity-conserving cos$\frac{E_J}{E_C} \gg 1,$0-coupling (Mori et al., 5 Sep 2025). Experimentally, the system was free of MIST up to high powers, with more than 300 photons in the readout mode. The MIST could be controllably turned on by breaking the parity symmetry of the coupling using flux-tuning, producing identifiable $\frac{E_J}{E_C} \gg 1,$1 and $\frac{E_J}{E_C} \gg 1,$2 pathways in multi-state single-shot readout up to the fifth excited state (Mori et al., 5 Sep 2025).

5. Hybrid Josephson elements and molecular weak links

Another major meaning of transmon molecule circuit replaces the conventional Al/AlO$\frac{E_J}{E_C} \gg 1,$3/Al tunnel junction by a microscopic or unconventional weak link. The carbon-nanotube gatemon is the clearest literal example: a single-wall carbon nanotube, operating as a quantum-dot Josephson junction, serves as the nonlinear element of a transmon-style qubit in a circuit-QED architecture (Riechert et al., 3 Mar 2025). In that device, measured $\frac{E_J}{E_C} \gg 1,$4 values were $\frac{E_J}{E_C} \gg 1,$5 MHz and $\frac{E_J}{E_C} \gg 1,$6 MHz for two devices, $\frac{E_J}{E_C} \gg 1,$7 ranged from a few $\frac{E_J}{E_C} \gg 1,$8 MHz up to $\frac{E_J}{E_C} \gg 1,$9 GHz, qubit tunability exceeded 4 GHz, and the best coherence values reached $$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$0 ns and $$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$1 ns at one operating point, with $$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$2 ranging up to $$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$3 ns and $$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$4 up to $$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$5 ns across gate points (Riechert et al., 3 Mar 2025). The same work links sharp jumps in $$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$6 to $$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$7–$$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$8 quantum phase transitions in an interacting quantum-dot Josephson junction (Riechert et al., 3 Mar 2025).

A related architecture uses a reference Josephson junction in parallel with a gate-defined quantum-dot Josephson junction in an InAs/Al nanowire, forming a DC SQUID that shunts the transmon island (Bargerbos et al., 2022). There the Josephson potential is parity dependent,

$$hf_{01} \simeq \sqrt{8E_JE_C} - E_C,\qquad \alpha \simeq -\frac{E_C}{h}.$$9

so the transmon becomes a probe of singlet–doublet, or $E_J$0–$E_J$1, physics. The reported phase diagrams in plunger gate, tunnel coupling, flux, and magnetic field agree with a single-impurity Anderson model with superconducting leads, and deep in the singlet or doublet phase the ground-state parity lifetimes reach the millisecond regime (Bargerbos et al., 2022).

Van der Waals superconductors extend the same logic to unconventional condensates. A flux-tunable transmon whose nonlinear inductive element is an Al/AlO$E_J$2/4Hb-TaS$E_J$3 Josephson junction shows a SQUID-like flux-dependent spectrum that is quantitatively reproduced by a standard dressed transmon–cavity Hamiltonian (Blumenthal et al., 27 Jan 2026). Across measured devices, $E_J$4 ranged from $E_J$5 to $E_J$6s, Ramsey measurements indicated dephasing faster than the $E_J$7 ns time resolution, and the inferred $E_J$8 product was anomalously high by a factor $E_J$9 relative to the Ambegaokar–Baratoff expectation based on room-temperature resistance and known gaps (Blumenthal et al., 27 Jan 2026). No distinct material-specific subgap modes were resolved in the present geometry, but the work establishes a practical route to integrating 4Hb-TaS$V_g$0 into coherent quantum circuits (Blumenthal et al., 27 Jan 2026).

6. Architectural trade-offs and design directions

Several recent designs treat the transmon molecule as a way to shape interactions rather than merely to introduce new materials. An inductively shunted transmon can be engineered so that flux chooses between pure transverse and pure longitudinal coupling to an embedded harmonic mode, providing a cell in which one interaction can be used for readout and the other for qubit–qubit coupling (Richer et al., 2017). A different two-species architecture couples an inductively shunted transmon to a conventional transmon and exploits opposite-sign anharmonicities to suppress static $V_g$1. In that system, the measured $V_g$2 falls below $V_g$3 kHz near $V_g$4, while away from the sweet spot fast first-order sideband transitions enable a $V_g$5 ns CZ gate with fidelity $V_g$6 (Fasciati et al., 2024).

Distributed-mode versions of the same idea appear when a transmon is embedded in a finite-length transmission line. There the coupling factor is position dependent,

$V_g$7

and entanglement between the first transmission-line mode and the transmon is maximized around the Josephson-junction location, whereas the second mode remains weakly entangled (Salmanogli, 2021). This adds a spatial degree of freedom to transmon-molecule design: coupling is set not only by $V_g$8, $V_g$9, and detuning, but also by placement within the standing-wave structure.

Taken together, these works indicate a recurring set of trade-offs. Stronger hybridization, larger cross-Kerr, or added microscopic structure enlarge the accessible Hilbert space and the available control primitives, but they also introduce frequency crowding, flux sensitivity, dielectric participation, quasiparticle channels, and nontrivial calibration problems. Multimode devices such as the trimon show that strong internal $E_J(V_g)$0 can be made a resource rather than a parasitic effect (Maurya et al., 4 Mar 2026). Readout-oriented molecules show that a deliberately engineered nonlinear coupling can outperform the standard linear transverse readout Hamiltonian in the high-power regime (Mori et al., 4 Jul 2025). Hybrid-junction implementations show that microscopic fermionic physics can be embedded in familiar transmon Hamiltonians, but often with substantially shorter coherence and with spectroscopic parameter extraction replacing simple room-temperature resistance heuristics (Blumenthal et al., 27 Jan 2026, Bargerbos et al., 2022).

In that sense, the transmon molecule circuit is best understood as a research program within superconducting-circuit design: use composite Josephson elements, multiple internal modes, or microscopic weak links to extend the transmon beyond the single-junction, single-mode template while retaining enough circuit-level regularity that standard circuit-QED methods remain applicable.