🔬 atomic physics

Quantifying the Dual-isotope Advantage for Ytterbium-array Surface Codes using Realistic Noise Models

This paper demonstrates through realistic noise modeling and open-source simulation (DualYbSim) that dual-isotope Ytterbium arrays with in-place ancilla measurement significantly outperform single-isotope surface code architectures in logical error rates, while identifying Rydberg-state decay as the primary bottleneck for future fault-tolerant quantum computing improvements.

Original authors: Fumiyoshi Kobayashi, Toshi Kusano, Nicholas Fazio, Yuma Nakamura

Published 2026-08-31

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Fumiyoshi Kobayashi, Toshi Kusano, Nicholas Fazio, Yuma Nakamura

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

Quantum computers promise to solve problems that are currently impossible for even the most powerful supercomputers, but they are notoriously fragile. The bits of information they use, called qubits, are easily disturbed by heat, vibration, or stray electromagnetic fields, causing them to lose their data. To build a machine that can actually work, scientists must build a system that can detect these errors and fix them faster than the errors occur. This process, known as quantum error correction, requires a vast network of physical qubits working together to protect a smaller number of logical qubits, which hold the actual calculation. The challenge is that the very act of checking for errors introduces new risks and takes time, and if the checking process is too slow or too clumsy, the errors will overwhelm the system before they can be fixed.

Researchers are exploring many different ways to build these machines, but one of the most promising approaches uses neutral atoms trapped in arrays of laser beams. These atoms can be moved around with incredible precision, allowing for flexible connections between qubits. However, a major bottleneck in this technology is the measurement step. To check for errors, the computer must read the state of specific helper atoms, called ancilla qubits, without disturbing the data atoms they are protecting. In many current designs, this requires physically moving the helper atoms to a separate measurement zone or hiding them in a different energy state before reading them. These extra steps take time and introduce new opportunities for errors to creep in, slowing down the entire error-correction cycle.

A team of researchers has now used detailed computer simulations to test a specific design that avoids these extra steps. They focused on a system using two different versions, or isotopes, of the element ytterbium. In this setup, the data qubits are made of one type of ytterbium atom, while the helper qubits are made of a different type. Because these two types of atoms respond to light at slightly different frequencies, the researchers can shine a laser to measure the helper atoms without accidentally disturbing the data atoms. This allows the helper atoms to be measured right where they are sitting, a method known as in-place measurement. The team developed a realistic model of the noise and errors that occur in such a system, including the decay of atoms from high-energy states and the loss of atoms from their traps, and ran millions of simulations to see how well this dual-isotope approach would perform compared to older methods.

The simulations revealed that the dual-isotope design is significantly more effective at protecting quantum information. By measuring the helper atoms in place, the system avoids the time delays and additional errors associated with moving atoms or shuffling them into different energy states. When the researchers compared this approach to single-isotope systems that rely on moving atoms to a measurement zone or hiding them in a "shelf" state, the dual-isotope system consistently produced fewer errors in the final logical data. The advantage was clear across different types of error-correcting codes, with the dual-isotope system achieving the lowest error rates in every scenario tested. This suggests that using two distinct types of atoms to separate the roles of data and measurement is a superior strategy for building a reliable quantum memory.

The study also took a deep dive into exactly where the errors were coming from to identify the most critical targets for future improvement. The analysis showed that the dominant source of failure in all configurations was the decay of atoms from a high-energy Rydberg state, which is used to perform the two-qubit gates necessary for the computer to function. This specific type of decay accounted for between 74 and 80 percent of the scaling of the logical error rate. This finding highlights that while the dual-isotope architecture provides a structural advantage, the overall performance is still heavily limited by the physics of these high-energy states. The researchers found that if the rate of this decay could be reduced by half, the error suppression of the system would improve by a factor of 1.7. Furthermore, they noted that if the decay rate were reduced while simultaneously making small improvements to other types of errors, the benefits would be even greater.

These results provide a clear roadmap for the development of neutral-atom quantum computers. The simulations suggest that the dual-isotope ytterbium architecture is a strong candidate for near-term fault-tolerant quantum computing because it minimizes the overhead of error correction. However, the work also makes it clear that simply changing the architecture is not enough; significant progress must also be made in controlling the atoms during the high-energy operations that cause them to decay. By identifying Rydberg-state decay as the primary bottleneck, the study directs experimental efforts toward improving laser control and gate speeds. The researchers have also made their simulation tools available to the wider scientific community, allowing others to test these noise models and explore further refinements. The path forward involves a dual effort: maintaining the structural benefits of the dual-isotope design while aggressively pursuing the physical improvements needed to tame the decay of the atoms themselves.

Technical Summary: Quantifying the Dual-isotope Advantage for Ytterbium-array Surface Codes using Realistic Noise Models

Problem Statement

Neutral-atom quantum computers, utilizing optical tweezer arrays, offer significant advantages for fault-tolerant quantum computing (FTQC), including long coherence times, high-fidelity gates, and flexible connectivity. However, the performance of quantum error correction (QEC) in these systems is constrained by the time overhead associated with syndrome extraction (SE). In single-isotope architectures, measuring ancilla qubits often requires transporting them to a dedicated measurement zone or employing shelving operations to prevent readout crosstalk. These processes introduce significant latency and additional error sources, potentially limiting the scalability of logical circuits. While dual-species (or dual-isotope) systems have been proposed to mitigate this by enabling in-place measurement through spectral separation, quantitative comparisons of their logical error rates (LER) against single-isotope schemes under realistic noise models have been limited.

Methodology

The authors develop a comprehensive, experimentally motivated noise model for dual-isotope Ytterbium (Yb) systems, specifically utilizing $^{171}\text{Yb}$ and $^{174}\text{Yb}$ .

Noise Modeling and Simulation Tool:
- The authors construct a generalized noise model covering coherent-control, measurement, reset, idling, decay, and transportation channels.
- To enable large-scale QEC simulations, they employ generalized Pauli twirling to map non-Pauli physical noise (including leakage and loss) onto Clifford-simulatable Pauli and erasure channels.
- This model is implemented as DualYbSim, an open-source Python library that wraps the Stim simulator, allowing for efficient Monte Carlo sampling of surface code circuits.
System Configurations:
- Dual-Isotope Architecture: Data qubits are encoded in $^{171}\text{Yb}$ (either ground-state $g$ -qubits or metastable-state $m$ -qubits), while ancilla qubits are encoded in $^{174}\text{Yb}$ optical clock qubits. This configuration allows for in-place measurement of ancillas without transport or shelving, leveraging the spectral separation between isotopes.
- Single-Isotope Architectures: The study compares the dual-isotope approach against single-isotope $^{171}\text{Yb}$ and $^{174}\text{Yb}$ configurations. These single-isotope schemes utilize either shelving (moving qubit states to off-resonant levels) or zoned measurement (physically transporting atoms to a measurement zone) to perform syndrome extraction.
- Syndrome Extraction Variants: The study evaluates both Normal SE (standard ordering) and SWAP SE (exchanging data and ancilla roles to refresh atoms), noting that SWAP SE is incompatible with the dual-isotope in-place measurement scheme.
Codes and Metrics:
- Simulations are performed on both the rotated surface code and the XZZX surface code.
- The primary metric is the logical error rate per round ( $p_L$ ) as a function of code distance ( $d$ ).
- An error budget analysis is conducted by decomposing the inverse error suppression factor ( $\Lambda^{-1}$ ) to identify dominant physical error sources.

Key Contributions

Development of DualYbSim: The creation of a publicly available, open-source simulation tool that incorporates a realistic, generalized Pauli-twirled noise model specifically for dual-isotope Yb systems.
Systematic Comparison: A rigorous numerical comparison of logical error rates across multiple architectural realizations, including dual-isotope in-place measurement versus single-isotope shelving and zoned measurement.
Error Budget Quantification: A detailed breakdown of error contributions, identifying Rydberg-state decay as the dominant limitation across all configurations.

Results

Superiority of Dual-Isotope In-Place Measurement: The simulations demonstrate that the dual-isotope architecture ( $^{171}\text{Yb}$ data + $^{174}\text{Yb}$ ancilla) utilizing in-place measurement achieves the lowest logical error rates among all considered configurations. It outperforms single-isotope schemes that rely on shelving or zoned measurement.
Code Performance: The XZZX surface code generally achieves logical error rates comparable to or lower than the rotated surface code across all configurations.
SWAP SE Limitations: Contrary to some previous expectations, SWAP SE does not outperform Normal SE in the Yb systems studied. The additional gate operations required for SWAP introduce errors that outweigh the benefits of refreshing atoms, particularly when erasure information is not utilized for decoding.
Dominant Error Source: The error budget analysis reveals that Rydberg-state decay is the primary bottleneck, contributing 74–80% of the logical error rate scaling in all configurations.
Sensitivity Analysis: The study shows that the error suppression factor ( $\Lambda$ ) is highly sensitive to the Rydberg decay rate. Reducing the Rydberg decay rate by half could improve the error suppression factor by a factor of 1.7. Further improvements are possible if depolarizing errors are simultaneously reduced.

Significance and Claims

The paper claims that the dual-isotope Yb architecture represents a sensible candidate for performing surface code QEC on current and upcoming hardware. The primary advantage lies in the elimination of the time overhead and associated error accumulation from atom transportation and shelving operations required in single-isotope systems.

The authors emphasize that while the dual-isotope approach offers the best performance among the configurations tested, the ultimate performance is still fundamentally limited by Rydberg decay. Therefore, the most promising pathway to improving FTQC performance in these systems is not merely architectural changes, but experimental improvements in reducing Rydberg decay rates (e.g., via shorter gate times or higher-power pulsed lasers) and reducing depolarizing errors. The work provides a quantitative framework for prioritizing these experimental targets.