Cross-Kerr Inter-Qubit Interaction

• Cross-Kerr interaction is a Hamiltonian term defining a ZZ coupling that enables high-fidelity, conditional-phase (CZ) gates without population transfer.
• It is implemented using flux-tunable couplers and engineered nonlinear circuit elements in superconducting and photonic systems to achieve precise, fast gate tuning.
• Advanced techniques like squeezing-enhanced protocols boost the effective interaction strength, promising scalable, low-error quantum information processing.

A cross-Kerr inter-qubit interaction refers to a Hamiltonian term of the form $\chi_{zz}\,\sigma_z^{(1)}\sigma_z^{(2)}$ (or in bosonic language, $\chi\,a^\dagger a\,b^\dagger b$), where $\chi_{zz}$ parameterizes the strength of a longitudinal two-qubit or inter-mode coupling. In the context of gate-based quantum information processing, the cross-Kerr interaction enables fast, high-fidelity entangling gates—especially conditional-phase (CZ) gates—without requiring population transfer between computational states or relying on resonant drive-induced hybridization with higher levels. Cross-Kerr or $ZZ$-type couplings can be implemented via dispersive Schrieffer–Wolff processes, engineered couplers, or direct nonlinear elements, and are widely deployed in both superconducting and photonic quantum architectures (Collodo et al., 2020).

1. Fundamental Hamiltonian Origin and Derivation

The canonical cross-Kerr interaction arises when two quantum modes (typically qubits or resonators) are coupled through either a nonlinear circuit element (e.g., a Josephson junction-based coupler) or via virtual transitions mediated by auxiliary states. The general effective Hamiltonian for two coupled qubits (or modes) is

$$H_{\mathrm{eff}}/\hbar = \sum_{i=1,2} (\omega_i'/2)\sigma_z^{(i)} + \chi_{zz}\,\sigma_z^{(1)}\sigma_z^{(2)},$$

where $\omega_i'$ are the Lamb-shifted qubit frequencies and the cross-Kerr rate $\chi_{zz}$ is derived by adiabatically eliminating (Schrieffer–Wolff transformation) higher-energy or coupler degrees of freedom in the large-detuning regime. For two transmons coupled via a flux-tunable coupler (Collodo et al., 2020),

$$\chi_{zz} = \frac{2J^2\,\alpha_q}{\Delta(\Delta+\alpha_q)}$$

where $J$ is the total transverse coupling ($J \simeq g_{1c}g_{2c}/\Delta + g_{12}$), $\Delta$ is the detuning between the qubits and the coupler, and $\alpha_q$ is the transmon anharmonicity.

Alternatively, in the bosonic mode language relevant for photonic implementations, the cross-Kerr Hamiltonian takes the form $H_{\mathrm{int}} = \chi\,a^\dagger a\,b^\dagger b$, inducing a photon-number–dependent frequency shift (Zhang et al., 2016, Brod et al., 2016).

2. Circuit Architectures and Physical Implementations

In superconducting quantum circuits, the cross-Kerr interaction is commonly realized with two fixed-frequency transmons coupled via a flux-tunable transmon coupler, which mediates the $ZZ$ interaction by its frequency detuning. The full system Lagrangian includes quadratic and nonlinear Josephson terms, with coupling strengths set by shunt capacitances and coupler Josephson energies. By tuning the external flux through the coupler, the effective $\chi_{zz}$ can be modulated smoothly from strong positive values (up to tens of MHz) down to near-zero (residual $|\chi_{zz}|/2\pi \lesssim 0.06$ MHz) (Collodo et al., 2020).

Advanced schemes, e.g. "quarton" couplers, employ a coupler potential engineered to eliminate linear coupling (zero $\phi^2$ term) and leave only quartic (pure cross-Kerr) nonlinearity, enabling gigahertz-scale $\chi$ values without populating higher transmon levels and allowing self-Kerr cancellation for "linearized" qubits or resonators (Ye et al., 2020).

Photonic qubits (dual-rail, time-bin encodings) can leverage cross-Kerr interaction via nonlinear optical materials, multi-level (e.g., N-type) circuit QED molecules, or site-resolved atomic ensembles, yielding Hamiltonians $H_{\rm int} = \chi\,a^\dagger a\,b^\dagger b$ with parametric tunability (Zhang et al., 2016, Brod et al., 2016).

3. Gate Protocols, Tunability, and Performance

The cross-Kerr interaction is directly mapped to entangling gates by exposing the qubits/modes to a nonzero $\chi_{zz}(t)$ for a controlled duration. The acquired conditional phase is

$$\varphi = \int_0^\tau \chi_{zz}(t) \, dt \approx \tau\chi_{zz}$$

(for flat-top pulses), so a full CZ gate ($\varphi = \pi$) is implemented in time $\tau = \pi / \chi_{zz}$. This scheme is used to realize 38 ns CZ gates with fidelity $97.9\%$ and leakage $\sim0.14\%$, with on/off $\chi_{zz}$ tuning ratio $>10^3$ (Collodo et al., 2020).

An analogous approach in cat-qubit architectures allows a fast, high-fidelity $R_{ZZ}(-\pi/2)$ gate by dynamically lifting engineered level degeneracies to transiently enable $\chi_{zz}$; gate fidelities exceeding $99.9\%$ and sub-25 ns gate times are feasible in idealized conditions (Aoki et al., 2024).

Photon-photon CZ gates in optical or microwave circuits can operate in continuous (distributed) or discrete (site-chain) geometries. For instance, a chain of $N$ cross-Kerr sites with counter-propagating photons and $\chi/\gamma \gtrsim 1$ yields a controlled phase shift of $\pi$ in the infinite-chain and strong-coupling limit, with infidelity scaling as $O(1/N^2)$ (Brod et al., 2016).

4. Theoretical Enhancements and Scaling Strategies

Fundamental limitations of the bare cross-Kerr interaction—typically weak in standard media—can be overcome by interleaving modulated squeezing operations with Kerr evolution. Protocols employing alternating single-mode (or two-mode) squeezing exploit the identity

$$\lim_{N\to\infty} \Bigl( S_{0}^\dagger U_{\Delta t} S_{0} \, S_{\pi}^\dagger U_{\Delta t} S_{\pi} \Bigr)^N = \exp\bigl[-i\cosh(2r) H_0 t\bigr]$$

achieving an effective enhancement $\chi_{\mathrm{eff}} = \cosh^2(2r)\chi$ in two-mode schemes. Gate times $T_{\mathrm{gate}}^{(\pi)} = \pi/[\chi\cosh^2(2r)]$ thus decrease exponentially with squeezing strength $r$. Loss and Trotter errors can be made subdominant with realistic $r\sim 10$–$20$ dB, allowing deterministic photonic CZ gates with errors $<1\%$ (Tiwari et al., 2024).

These squeezing-amplified strategies are applicable to both optical (nanophotonic waveguides, hollow-core fibers) and microwave platforms, subject to constraints on attainable squeezing, losses, and available cross-Kerr rates.

5. Error Channels, Crosstalk, and Suppression Mechanisms

Key error channels include:

• Leakage to non-computational (higher) levels, which is minimized in cross-Kerr (non-swap) gates compared to transverse-coupled (iSWAP, sideband, frequency-tuned) schemes (Collodo et al., 2020).
• Residual $ZZ$ coupling at "idle" (off) bias, which can induce undesired multi-qubit phases and crosstalk during single-qubit gates. Engineering four-fold level degeneracy or employing echo-like suppression schemes reduces residual $\chi_{zz}$ to sub-kHz scales, enabling high on/off ratios ($>10^3$) (Aoki et al., 2024).
• Photon loss and Trotterization errors in photonic protocols, which are exponentially suppressed with squeezing and scaling with $1/N$, respectively (Tiwari et al., 2024).

Engineering couplers with minimal residual linear interactions (e.g., quarton), precise flux control, and predistorted flux pulses further suppress frequency excursions and crosstalk (Collodo et al., 2020, Ye et al., 2020).

6. Comparison to Alternative Inter-Qubit Couplings

Traditional two-qubit gates in superconducting circuits include:

• Transverse ($\sigma_x \sigma_x$) mediated gates (iSWAP, sideband), which require population transfer between computational and non-computational states and are susceptible to leakage and timing constraints.
• Frequency-tunable CZ gates, where qubits are brought near resonance, but are sensitive to flux noise and associated decoherence.
• Parametric gates, where periodic modulation can drive spurious $ZZ$ couplings and additional crosstalk.

The cross-Kerr ($ZZ$) approach offers strict population conservation, fast on/off switching, continuous phase tunability via a single control parameter, and compatibility with fixed/fabricated detuned qubits (Collodo et al., 2020, Aoki et al., 2024). Extensions to any circuit QED platform supporting nonlinear tunable couplers are straightforward.

7. Scalability and Implications for Quantum Information Processing

Strong, tunable cross-Kerr inter-qubit interactions are instrumental for scalable gate-based quantum processors, especially where high on/off ratios and low crosstalk are required, such as in surface-code or bias-noise-protected cat-qubit arrays. Architectures based on engineered $\chi_{zz}$ enable both fast entangling gates and active suppression of residual coupling during idle or single-qubit operations (Aoki et al., 2024).

Cross-Kerr couplings also facilitate high-fidelity, passive all-optical CZ gates, quantum nondemolition (QND) photon detection, continuous-variable error-correcting codes, and advanced bosonic logic gate protocols (Zhang et al., 2016, Ye et al., 2020, Brod et al., 2016).

A plausible implication is that further advances in coupler design, flux/pulse control, and squeezing technology will drive both superconducting and photonic platforms toward deterministic, low-error, scalable quantum information processing utilizing the cross-Kerr inter-qubit interaction.