⚛️ quantum physics

Influence of fast and slow laser phase noise on the fidelity of the Mølmer-Sørensen trapped-ion gate

This paper presents a comprehensive theoretical analysis of how fast and slow laser phase noise components distinctively impact the fidelity of the Mølmer-Sørensen gate in trapped-ion quantum computing, deriving analytical expressions for average gate error and validating them against existing numerical simulations.

Original authors: Nikita Semenin, Ksenia Khabarova, Nikolay Kolachevsky

Published 2026-09-24

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Nikita Semenin, Ksenia Khabarova, Nikolay Kolachevsky

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 today's machines, scientists are turning to a technology that uses the strange rules of the very small. These quantum computers rely on tiny particles, often atoms held in place by invisible electric fields, to store and process information. The power of such a machine comes from its ability to link these particles together in a delicate state of connection, allowing them to work as a single unit. However, this connection is fragile. To make the computer useful, scientists must perform operations, or "gates," that manipulate these particles with extreme precision. If the operation is even slightly off, the information is lost. One of the most promising ways to perform these operations uses lasers to nudge the atoms, but the lasers themselves are not perfect. They carry a subtle, constant jitter in their timing, known as phase noise, which can disrupt the delicate work of the quantum computer.

A team of researchers has now mapped out exactly how this laser jitter ruins the performance of a specific, widely used quantum operation called the Mølmer-Sørensen gate. This gate is a favorite among scientists working with trapped ions because it is robust against temperature changes and has already achieved some of the highest success rates in the field. The researchers focused on the two main ways the laser's imperfections cause errors. They identified that the noise acts differently depending on how fast it fluctuates. There is "fast" noise, which ripples at speeds similar to the vibration of the atoms themselves, and "slow" noise, which drifts over timescales comparable to the duration of the gate operation. By developing a new theoretical framework, the team calculated how these two types of noise interact with the gate, creating a detailed guide that predicts the error rate based on the specific characteristics of the laser light.

The study reveals that the impact of the laser noise depends heavily on the speed of the fluctuations relative to the gate's timing. The fast noise, which occurs at frequencies near the natural vibration of the trapped atoms, creates errors by interfering with the primary interaction between the laser and the atoms. The researchers found that this type of noise is most damaging when its frequency matches the specific tuning of the laser used to drive the gate. They derived a mathematical description, known as a filter function, which acts like a sieve, showing exactly which frequencies of noise are allowed to pass through and cause damage. This filter function peaks sharply at the frequency of the laser's detuning, meaning that even a small amount of noise at that specific frequency can significantly lower the quality of the gate.

In contrast, the slow noise operates on a much longer timescale, roughly the inverse of the time it takes to complete the gate. This type of noise is caused by factors like thermal shifts or mechanical vibrations in the equipment that holds the laser steady. Because these fluctuations are slow, they do not shake the atoms violently but instead cause the laser to drift slightly off its intended frequency. The researchers showed that this drift leads to errors because the laser components, which are meant to be perfectly balanced, become slightly unbalanced as they approach the resonance of the atoms. They found that for this slow noise, the error is directly proportional to the width of the laser's frequency spread and the duration of the gate. If the laser is not stable enough, or if the gate takes too long, the error accumulates.

A key finding of the work is that the specific arrangement of the laser beams does not change the fundamental outcome for fast noise. Whether the laser beams travel in the same direction or in opposite directions, the total error caused by fast noise remains the same, provided the laser power is consistent. This simplifies the engineering requirements for building these quantum computers, as it means researchers do not need to worry about one beam geometry being inherently superior to the other regarding this specific type of noise. However, for slow noise, the situation is more nuanced. The researchers distinguished between the errors seen in a theoretical "rotating frame," which is a mathematical tool used to simplify calculations, and the errors seen in the actual laboratory frame where the experiment takes place. They determined that for a single gate or a circuit using different, uncorrelated laser sources, the laboratory frame calculation is the correct one to use. But if a long sequence of gates is performed using the same laser source, the errors can be treated differently, and the rotating frame calculation becomes more appropriate.

The team validated their theoretical predictions by comparing them with previous computer simulations and found that their new formulas matched the existing data perfectly. This agreement gives them confidence that their analytical expressions are accurate and can be used as a reliable rule of thumb for designing future quantum systems. They used their findings to calculate the strict requirements a laser must meet to achieve the high fidelity needed for error correction, a threshold where quantum computers begin to offer a real advantage over classical ones. To reach a gate error rate of one in a thousand, which is the target for practical quantum computing, the laser must have a phase noise level below a specific threshold at the detuning frequency and a linewidth, or frequency spread, of less than two hertz. Additionally, the central frequency of the laser must not drift by more than 65 hertz.

These requirements are demanding but achievable with current technology, though they push the limits of what is possible. Achieving such stability requires lasers that are locked to extremely stable references, such as high-quality optical cavities, and often necessitates the use of solid-state lasers rather than simpler diode lasers to minimize inherent noise. The researchers also noted that for very low-frequency noise, which behaves differently than the white noise assumed in their simplified estimates, the error depends on the total duration of the experiment. This means that for long, uninterrupted sequences of operations, the laser must maintain its stability over time, not just in an instant. By providing these clear, analytical tools, the study offers a roadmap for engineers and physicists to build better lasers and design more robust quantum gates, moving the field closer to the realization of powerful, useful quantum computers.

Technical Summary: Influence of Fast and Slow Laser Phase Noise on the Fidelity of the Mølmer-Sørensen Trapped-Ion Gate

Problem Statement
High-fidelity two-qubit entangling gates are a prerequisite for scalable quantum computing. While the Mølmer-Sørensen (MS) gate is a standard choice for trapped-ion systems due to its resilience to ion temperature, its performance is limited by the spectral impurity of the driving laser field. Specifically, laser phase noise induces errors through mechanisms operating on distinct timescales. Previous literature has addressed pure dephasing or specific beam geometries but has lacked a comprehensive analytical treatment of intrinsic laser phase noise affecting the MS gate across both "fast" (high-frequency) and "slow" (low-frequency) spectral ranges. This work addresses the gap by deriving analytical expressions for gate errors caused by these noise components, moving beyond purely numerical simulations.

Methodology
The authors perform a theoretical analysis of the MS gate Hamiltonian under the influence of laser phase noise. The study covers two common laser beam geometries:

Phase-sensitive configuration: Co-propagating bichromatic beams.
Phase-insensitive configuration: Counter-propagating bichromatic beams.

The methodology involves:

Hamiltonian Derivation: The authors derive the MS gate Hamiltonians for both geometries in the Lamb-Dicke regime, explicitly separating the ideal evolution from noise terms arising from fluctuations in the laser source phase ( $\delta\phi(t)$ ).
Noise Classification: The noise is categorized into two spectral ranges based on the introduction's analysis:
- Fast Noise: Frequencies near the motional mode spectrum (hundreds of kHz to MHz), primarily driven by servo loop bumps and relaxation oscillations. This noise interacts with the carrier transition.
- Slow Noise: Frequencies on the order of the inverse gate time (kHz range), driven by thermal/acoustic effects and $1/f$ flicker noise. This noise interacts with the first motional sidebands.
Filter Function Formalism: Using first-order perturbation theory (Dyson series), the authors derive the average gate error as the convolution of the laser noise Power Spectral Density (PSD) and a gate-specific "filter function."
- For fast noise, the error is expressed in terms of the phase noise PSD ( $S_\phi$ ).
- For slow noise, the analysis is conducted in both a rotating frame and the laboratory frame to account for the frame-dependent nature of the noise correction. The error is expressed in terms of the frequency noise PSD ( $S_\omega$ ).
Analytical and Numerical Validation: The authors derive closed-form expressions for filter functions in simplified cases (e.g., constant Rabi frequency) and validate their analytical results against previously published numerical simulations (specifically referencing [24]).

Key Contributions

Analytical Filter Functions: The paper provides the first analytical expressions for the filter functions governing both fast and slow noise regimes for both phase-sensitive and phase-insensitive beam geometries. These functions depend solely on gate parameters (Rabi frequency, detuning, gate time, Lamb-Dicke parameters).
Unified Error Framework: The work demonstrates that the average gate error is the convolution of the noise PSD and the filter function. It establishes that for fast noise, both beam geometries yield identical error estimates under the assumption of uniform laser power.
Frame-Dependent Analysis for Slow Noise: The authors clarify the distinction between rotating and laboratory frame analyses for slow noise. They show that while the filter functions differ, the resulting error estimates are of the same order of magnitude. They provide specific use cases for when each frame's filter function is appropriate (e.g., multi-gate circuits with a single source vs. single gates or uncorrelated sources).
Simplified Error Estimates: The paper derives simplified rules of thumb for gate error estimation. For fast noise, the error scales with the phase noise PSD at the detuning frequency and the square of the Rabi frequency. For slow noise (assuming white frequency noise), the error scales linearly with the laser linewidth ( $\Delta\omega$ ) and the gate time ( $\tau$ ).

Results

Fast Noise: The filter functions for both configurations are peaked at the detuning frequency ( $\mu$ ) with a width approximately equal to the inverse gate time ( $1/\tau$ ). The analytical derivation confirms that the gate error for both geometries is given by $\epsilon \approx \frac{4}{5} S_\phi(\mu/2\pi) \Omega^2 \tau$ (for a random starting state), which aligns with previous numerical findings when adjusted for the specific starting state used in those simulations.
Slow Noise: The filter functions are prominent at frequencies up to $\sim 1/\tau$ . The analysis reveals that for slowly varying frequency noise (white noise approximation), the error in the rotating frame is $\epsilon^{(rot)}_{slow} \approx \frac{2}{5} \Delta\omega \tau$ , while the laboratory frame error has an upper bound of $\epsilon^{(lab)}_{slow} \leq \frac{8}{5} \Delta\omega \tau$ .
Low-Frequency Noise: The paper addresses the divergence issues associated with $1/f$ noise at very low frequencies by introducing a cutoff based on the measurement time and the transition from $1/f$ to white noise. It provides an estimate for errors induced by low-frequency components based on the mean squared frequency deviation.
Experimental Requirements: Based on a target gate error of $10^{-3}$ $1 0^{- 3}$ (a threshold for error correction advantage) and standard parameters ( $\tau \approx 167 \mu s$ $τ \approx 167 μ s$ , $\Omega \approx 2\pi \cdot 20$ $Ω \approx 2 π \cdot 20$ kHz), the authors calculate the following requirements for the laser system:
- Phase noise PSD at the detuning: $S_\phi < -93$ dBc/Hz.
- Laser linewidth: $\Delta\omega < 2$ Hz.
- Central frequency deviation: $\delta\omega < 65$ Hz.

Significance
The paper claims that its primary significance lies in providing a rigorous theoretical foundation for understanding laser phase noise in MS gates, complementing existing numerical studies with analytical tools. By deriving explicit filter functions, the work enables researchers to calculate gate error requirements directly from laser spectral purity specifications without relying solely on time-consuming numerical simulations. The authors note that while the derived requirements (e.g., sub-Hz linewidths and specific phase noise floors) are demanding, they are achievable with current solid-state laser technologies and auxiliary techniques like filtering cavities or injection locking. The work serves as a guide for optimizing laser systems to meet the fidelity thresholds necessary for large-scale trapped-ion quantum computing.

More like this