⚛️ quantum physics

Parametric two-qubit gates via Landau-Zener interference

This paper proposes and experimentally demonstrates a fast, tunable two-qubit gate based on Landau-Zener interference that operates on both modular chiplet and monolithic transmon architectures, offering a unique solution for multiplexing control and interconnecting superconducting quantum processors.

Original authors: Simon Geisert, Albert Hertel, Soeren Ihssen, Zhongyi Jiang, Paul Kugler, Nicolas Zapata, Nicolas Gosling, Ameya Nambisan, Yuan Gao, Asier Galicia, Jéferson R. Guimarães, Yorgo Haddad, Marc Neis, Harsh

Published 2026-09-15

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Simon Geisert, Albert Hertel, Soeren Ihssen, Zhongyi Jiang, Paul Kugler, Nicolas Zapata, Nicolas Gosling, Ameya Nambisan, Yuan Gao, Asier Galicia, Jéferson R. Guimarães, Yorgo Haddad, Marc Neis, Harsh Bhardwaj, Dmitriy A. Volkov, Juan Cereijo, Marcello Guardascione, Yebin Liu, Markus Jerger, Pavel Bushev, Frank Wilhelm-Mauch, Wolfgang Wernsdorfer, Shai Machnes, Mohammad Ansari, Rami Barends, Ioan M. Pop

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

To build a computer that can solve problems beyond the reach of any machine today, scientists are turning to the strange rules of quantum mechanics. They are trying to construct processors using tiny circuits that act like atoms, holding information in states that can exist in multiple possibilities at once. These circuits, known as superconducting qubits, are the leading candidates for building such a machine. However, for these computers to work, the individual qubits must be able to talk to one another, swapping information to perform calculations. This conversation happens through a two-qubit gate, a mechanism that links two qubits together just long enough to interact and then lets them go. The challenge is that these interactions must be incredibly fast and precise, yet they must not happen when they are not wanted, or the delicate quantum information will be lost to the surrounding noise. For years, the standard way to manage this has been to use a fixed, passive link that is turned on and off by microwaves, or by slowly shifting the energy of the qubits. While effective, these methods often struggle with speed or require complex, rigid control schemes that limit how many qubits can be connected in a single processor.

A team of researchers has now demonstrated a new way to make these qubits talk, one that is faster and more flexible than previous methods. They built a system where three qubits are arranged in a line: two on the ends that hold the data, and one in the middle that acts as a bridge. Instead of using microwaves to trigger a connection, they rhythmically push the middle qubit back and forth in energy, forcing it to repeatedly cross paths with the two data qubits. As the middle qubit sweeps past the others, it creates a quantum interference pattern, much like how light waves can cancel each other out or amplify each other when they meet. By carefully timing these crossings, the researchers found they could force the information to jump from one end qubit to the other, leaving the middle bridge exactly as it started. This process, which relies on a phenomenon known as Landau-Zener transitions, allows them to tune the speed and timing of the interaction across a wide range of frequencies, from very slow to hundreds of millions of cycles per second.

The researchers tested this idea on two different types of superconducting hardware. The first was a modular design where the qubits were built on separate chips and connected together, while the second was a single, solid piece of material with all the components on one surface. In both cases, they successfully moved quantum information from one data qubit to the other. When they tuned the system to the right rhythm, the data qubits swapped their states completely, while the middle bridge returned to its original state, effectively acting as a perfect messenger. The team showed that they could achieve this swap at almost any frequency they chose, simply by adjusting the strength and timing of the push on the middle qubit. This flexibility is a significant departure from older methods, which are often locked into a narrow range of frequencies. In their experiments, the researchers were able to complete these swaps in just a few hundred nanoseconds, a speed that approaches the theoretical limit set by how strongly the qubits are connected.

One of the most striking aspects of this work is that the data qubits themselves never need to be touched by a control signal. They can sit quietly in their most stable positions, safe from electrical noise, while the middle qubit does all the work of shuttling information. This is a crucial advantage for scaling up quantum computers, as it means the delicate data qubits do not need to be constantly bombarded with control pulses that could disturb their state. The researchers also found that by stopping the process halfway through, they could create entangled pairs of qubits, a special state where two particles become linked in a way that defies classical physics. This capability suggests that the method could be used not just for swapping information, but for generating the complex quantum states needed for advanced calculations.

The team measured the accuracy of these swaps and found them to be highly reliable, with the information transferring correctly about 98 percent of the time in their fastest tests. They noted that this number is currently limited by how long the qubits can hold their state before natural environmental noise causes them to lose coherence. However, their simulations suggest that if the hardware were improved to hold states longer, the accuracy could rise to nearly 99.9 percent. This level of performance, combined with the ability to tune the gate speed and frequency on the fly, offers a promising path forward for connecting large numbers of qubits in future quantum processors. By proving that this interference-based method works on two very different hardware platforms, the researchers have shown that it is a robust and versatile tool for the next generation of quantum computing.

Technical Summary: Parametric Two-Qubit Gates via Landau-Zener Interference

Problem Statement
Current superconducting quantum processors rely on two-qubit gates that must operate significantly faster than qubit coherence times to meet error correction thresholds. Standard approaches often utilize dispersive couplers, which mediate interactions via a fixed-frequency or flux-tunable mode. However, operating in the dispersive regime to suppress state leakage limits gate speeds to well below the vacuum Rabi rate. Furthermore, parametric modulation in these systems is typically constrained to narrow transition-frequency windows defined by the qubit-coupler level structure. There is a need for a gate mechanism that utilizes the full coupling strength, operates at speeds comparable to the coupling rate, and offers tunability across a broad frequency range without requiring modulation of the data qubits themselves.

Methodology
The authors propose and demonstrate a two-qubit gate mechanism based on the quantum interference of consecutive Landau-Zener (LZ) transitions. The system consists of three capacitively coupled superconducting qubits: two data qubits ( $q_1, q_2$ ) and a central, flux-tunable coupler qubit ( $c$ ).

Theoretical Framework: The gate operates by periodically modulating the coupler's frequency ( $\epsilon(t)$ ) such that it repeatedly crosses the resonance frequency of the data qubits. This creates a sequence of LZ transitions between the "bright" states (symmetric superpositions of the data qubits and the coupler) and the "dark" states (antisymmetric combinations decoupled from the coupler).
Interferometric Model: The process is modeled using the adiabatic impulse model, analogous to a time-domain array of Mach-Zehnder interferometers. Each crossing of the coupler through the data qubit frequency acts as a beam splitter with transmission ( $T$ ) and reflection ( $R$ ) amplitudes determined by the LZ formula. The system evolves through free phase accumulation intervals between crossings.
Gate Condition: By tuning the drive amplitude ( $\epsilon_d$ ), offset ( $\epsilon_0$ ), and frequency ( $\omega_d$ ), the authors satisfy a constructive interference condition ( $\phi_+ = n\pi$ ). Under this condition, the coupler returns to its initial state while the population of the data qubits is fully inverted.
Experimental Platforms: The mechanism was validated on two distinct architectures:
1. A modular chiplet architecture using coupled generalized flux qubits.
2. A monolithic transmon architecture with flux-tunable couplers.

Key Contributions and Results

Broadband Tunability: The gate can be activated at arbitrary drive frequencies ranging from baseband up to hundreds of MHz. This is achieved because the constructive interference condition can be satisfied by adjusting the drive amplitude and offset for any given frequency, enabling in situ tuning across a continuous interval.
Gate Speed: The population swap frequency ( $\Omega$ ) scales with the coupling strength ( $g$ ). In the non-adiabatic regime, the gate speed approaches the limit of $\sqrt{2}g$ , significantly faster than dispersive gates. Experimental results on the monolithic transmon platform (with $g/2\pi \approx 300$ MHz) demonstrated gate times on the order of hundreds of nanoseconds.
Data Qubit Isolation: A distinct feature of this mechanism is that the data qubits remain idle at their flux-insensitive "sweet spots." Only the coupler is modulated, eliminating the need for direct flux or microwave driving of the data qubits during the gate operation.
Experimental Demonstration:
- Population Transfer: The authors demonstrated coherent population transfer from one data qubit to another via the coupler, observing characteristic interferometric patterns (hyperbolic lines of constructive interference) in parameter sweeps of drive strength and frequency.
- Fidelity Benchmarking: On the monolithic transmon device, the authors benchmarked the population swap fidelity. By optimizing drive parameters for different cycle counts ( $m=1$ and $m=8$ ), they achieved a swap fidelity of $F \approx 0.98$ . This fidelity is currently limited by the system's decoherence time ( $T_2^{echo} \approx 2.2$ $\mu$ s).
- Scalability: The results suggest that with improved coherence times (two orders of magnitude larger), fidelities approaching 99.9% are theoretically achievable.

Significance
The paper establishes the Landau-Zener interference gate as a unique tool for superconducting quantum hardware. It bridges the gap between baseband and parametric control, offering a mechanism that combines the speed of resonant Rabi oscillations with the versatility of parametric gates. The ability to tune the control frequency across a wide continuous interval facilitates frequency-multiplexed control of couplers, which is critical for scaling to large arrays of interconnected qubits. Furthermore, the mechanism's reliance on the coupler's active participation, rather than treating it as a passive mediator, allows for the generation of entangled states (such as Bell states) and offers a pathway for interconnecting modular chiplet architectures. The authors note that while the mechanism requires high-fidelity control and measurement of the coupler state to restrict the gate to specific unitaries (like $G_0$ ), the trade-off provides a distinct advantage in gate speed and tunability compared to conventional dispersive approaches.

More like this