Sympathetic Cooling in Trapped Ions with Spectral Selectivity via the Zeeman Shift

arXiv:2605.27577 · quant-ph · Submitted 2026-05-26 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: Today's paper: "Sympathetic Cooling in Trapped Ions with Spectral Selectivity via the Zeeman Shift".

Mira: The gist The authors demonstrate a sympathetic cooling scheme leveraging internal metastable atomic levels and natural Zeeman shift to isolate non-coolant or “data ions” from coolant ions in trapped ion…

Kai: First, who's behind it and why it matters.

Paper summary: Kai: So, to recap where we are, we’ve seen how the paper "Sympathetic Cooling in Trapped Ions with Spectral Selectivity via the Zeeman Shift" proposes using internal metastable levels and natural Zeeman shifts for sympathetic cooling.

Mira: The thesis here is that spectral isolation of the qubit can be achieved through this method, using individual addressing with a Raman beam to select which ions are being cooled <ref:2605.27577#pg3>.

Lev: Essentially, they’re moving away from just picking any ion as a coolant and instead using the Zeeman shift to pick the right one for the job.

Kai: They demonstrate this on a five-ion chain of ytterbium ions at eight Kelvin, showing that they can suppress both the center-of-mass and tilt modes down to near ground state motion <ref:2605.27577#pg3>.

Mira: And they show that even with cooling, the average motional population for all modes ends up less than one, which is a good sign for achieving Lamb-Dicke regime requirements forty.

Lev: But we have to remember the caveat here: this demonstration used a fixed number of pulses—one hundred for all modes in that basic setup <ref:2605.27577#pg3>.

Kai: And they also quantified the effect on data qubit coherence, showing a loss of two point three times ten to the minus three per second during those cooling pulses <ref:2605.27577#pg3>.

Mira: That loss is linked to noise in the four hundred thirty-five nm beam intensity fluctuations, which is what they find is causing most of that decoherence <ref:2605.27577#pg3>.

Lev: So, when we think about running this on real hardware, we have to consider that those intensity fluctuations are real noise sources in the light shift Appendix Details.

Kai: The coherence time they found was between five hundred ninety six milliseconds and one thousand one hundred forty five milliseconds for a time averaging period of one second to one hundred milliseconds Appendix Details.

Mira: It’s worth noting that the decay rate depends on the laser parameters in a quadratic way with Omega four hundred thirty-five and inversely linearly with Delta four hundred thirty-five Appendix Details.

Lev: So, for error correction research, we need to know how much overhead this cooling actually adds to your overall sequence when you run things on real hardware <ref:2605.27577#pg3>.

Kai: The paper shows a specific path using internal atomic levels accessible by a narrow quadrupole transition and Raman transitions for spectral selectivity <ref:2605.27577#pg3>.

Mira: It’s an interesting mechanism because it uses the natural Zeeman shift, which is inherent to the hyperfine qubit, to achieve that isolation <ref:2605.27577#pg3>.

Lev: That isolation is what makes it more practical than just using different atomic species or isotopes for coolant ions one <ref:2605.27577#pg1>.

Conclusion: Kai: So, wrapping up the discussion on "Sympathetic Cooling in Trapped Ions with Spectral Selectivity via the Zeeman Shift," the authors are demonstrating a method to isolate data ions from coolant ions.

Mira: The key point is that they use spectral selectivity through the Zeeman shift and Raman transitions to selectively cool certain radial modes <ref:2605.27577#pg3>.

Lev: They manage to achieve modest decoherence on the data ions, which is acceptable for high fidelity gate operations <ref:2605.27577#pg3>.

Kai: It’s a practical demonstration on a five-ion chain, showing how this scheme can suppress thermal motion in both the center-of-mass and tilt modes <ref:2605.27577#pg3>.

Mira: The implication for the broader field is that this approach provides a way to circumvent challenges associated with scaling systems while keeping motional heating under control <ref:2605.27577#pg1>.

Lev: For hardware development, the challenge remains implementing individual addressing of coolant ions without creating crosstalk on neighbors sixteen twenty-five <ref:2605.27577#pg2>.

Kai: Ultimately, this work shows that sympathetic cooling can be done more selectively than previous methods by leveraging internal atomic physics to manage the ion chain dynamics <ref:2605.27577#pg3>.

Kavyashree Ranawat, Jiyong Yu, Andrew Van Horn, Jacob Whitlow, Kenneth R Brown, Jungsang Kim

Duke University

quant-ph

Submitted: 2026-05-26

Updated: 2026-10-14

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 79/100

The gist: The gist The authors demonstrate a sympathetic cooling scheme leveraging internal metastable atomic levels and natural Zeeman shift to isolate non-coolant or “data ions” from coolant ions in

Key concepts

Sympathetic Cooling: This technique cools a group of ions by coupling them to a separate 'coolant ion.' The coolant ion absorbs energy from the motional modes of the other ions, effectively transferring that energy away. This is done between gate sequences to reduce overall heating.
Metastable Atomic Levels: These are specific internal energy states within an atom that have relatively long lifetimes. The cooling scheme uses these levels to selectively address and manipulate the coolant ion's motion without significantly disturbing the 'data ions' that need to remain coherent for quantum gates.
Zeeman Shift: The Zeeman shift is a natural frequency difference between different energy states of an atom caused by an external magnetic field. The scheme utilizes this shift, along with Raman transitions, to selectively target and cool the coolant ion while leaving the non-coolant ions unaffected.

Terminology

Fidelity — 2× in this paper · explained in 2 episodes

Summary

The gist The authors demonstrate a sympathetic cooling scheme leveraging internal metastable atomic levels and natural Zeeman shift to isolate non-coolant or “data ions” from coolant ions in trapped ion systems, achieving modest decoherence while preserving coherence requirements for high-fidelity gate operations.

Experimental Setup and Cooling Scheme

The experimental setup utilizes ytterbium (171Yb+) ions trapped in a surface trap within a cryostat maintained at 8 K The primary cooling laser is a 435 nm laser addressing the narrow 2S1/2 ↔ 2D3/2 quadrupole transition in 171Yb+ with a natural linewidth of 3 Hz The cooling scheme is based on pulsed side-band cooling (SBC), tuning the laser to the red side-band (RSB) of the 2S1/2, F = 1 ↔ 2D3/2, F = 1 transition. The detection scheme employed for determining the transition frequencies on 2S1/2 ↔ 2D3/2 line is shown in Figure 1a. After initialization and shelving to the D, 1, -1⟩ state at the RSB frequency, population decays to all sub-levels in the S, 1⟩ manifold. The population in both Zeeman levels is then shelved sequentially using the 435 nm laser by driving a RSB π-pulse on the S, 1, +1⟩ → D, 1, -1⟩ transition and a RSB π-pulse on the S, 1, -1⟩ → D, 1, +1⟩ transition. This combination of two RSB shelving pulses and the Raman individual pulse results in cooling and comprises one cooling pulse sequence.

Targeted Mode Cooling Performance

The cooling protocol was implemented on radial modes in a 5-ion chain of 171Yb+ ions from the Doppler limit The edge ion is designated as the coolant ion for cooling the COM, tilt and third order modes The number of cooling pulses is fixed at 100 for all modes in this basic demonstration of the scheme in practice. Following the cooling sequence, an initialization pulse brings the ion population to the ground state. The average motional population, ¯n measured via the side-band thermometry method [40] is measured to be less than 1 for all modes.

Heating Rate Suppression

The baseline heating rate of the COM mode in the 5-ion chain is 30 quanta per second (q/s), which is the highest of all other modes. The authors primarily target cooling of the COM mode because it is readily excited by uniform electric field noise, and its heating rate can indirectly affect the logical mode by causing unwanted fluctuations in the Rabi frequency. For cooling the tilt mode, which has a second highest heating rate of 3 − 4 q/s on average, periodic cooling is performed at a periodicity of 10 pulses within a 50 ms time period to achieve a steady state limit of n¯ = 0.54 ± 0.12 q. The duty cycle for the tilt mode is 93.5%.

Effect on Data Qubit Coherence

To probe the effect of the cooling laser on coherence, Ramsey interferometry driven by MW gates with a spin-echo sequence was used to measure coherence of the neighboring data qubits while cooling pulses were being actively applied. The average 1/e time for all the non-coolant ions is ≈ 650 ms. The coherence loss is 2.37 × 10−3/s per cooling pulse with respect to evolution under the identity gate. The primary source of decoherence originates from the 435 nm beam, which arises from noise in the light shift induced from applying the beam during cooling as its intensity fluctuates. The average decay rate for all non-coolant ions exhibits a quadratic dependence to omega435 and an inverse linear dependence to ∆435.

Coherence Decay Rate Analysis

The average decay rate obeys a quadratic dependence to omega435 and an inverse linear dependence to ∆435. The coherence time can then be extracted, given by τc = 8 / √2∆ / (omegaoσϵ). With the experimental parameters listed in Table I, a coherence time of 596 ms to 1145 ms is obtained between a time averaging period of 1 s to 100 ms after adjusting for the COM mode duty cycle.

Outlook and Future Work

One extension of this scheme is real-time chain re-configuration of the ions without shuttling, allowing another mode to be targeted for cooling mid-experiment by swapping the roles of the data ion and coolant. Further improvements in coherence are possible by incorporating measures including intensity stabilization of the cooling laser and increasing the Zeeman shift to minimize light shift by increasing detuning from the cooling transition. Individual addressing of ions by the cooling beam directly would also eliminate light shift as the beam is shaped such that it is focused only on the coolant ions. Another application of this same-isotope cooling scheme for radial modes is in the simulation of open quantum systems to serve as a source of tunable dissipation.

Appendix Details

The Hamiltonian considering a single radial mode and axial mode in the interaction picture is given by H = ħomega / 2 σˆ+exp(−i∆t) expiηax(ˆae−iνaxt + ˆa † e iνaxt) + iηrad(ˆbe−iνradt + ˆb † e iνradt) + h.c. (1). This results in an effective distribution of RSB Rabi frequency as it samples from a distribution which originates from the axial coupling. The average decay rate obeys a quadratic dependence to omega435 and an inverse linear dependence to ∆435. The coherence time can then be extracted, given by τc = 8 / √2∆ / (omegaoσϵ). A coherence time of 596 ms to 1145 ms is obtained between a time averaging period of 1 s to 100 ms after adjusting for the COM mode duty cycle. The decay rate dependence on laser parameters shows a quadratic dependence to omega435 and an inverse linear dependence to ∆435. The coherence time is obtained by fitting the data to a Gaussian fit which also includes a slow increase of the steady state population from 370 nm light leakage at much longer wait times. The average 1/e time for all the non-coolant ions is ≈ 650 ms across all ions. A coherence time of 596 ms to 1145 ms is obtained between a time averaging period of 1 s to 100 ms after adjusting for the COM mode duty cycle. The average decay rate obeys a quadratic dependence to omega435 and an inverse linear dependence to ∆435. The coherence time can then be extracted, given by τc = 8 / √2∆ / (omegaoσϵ). A coherence time of 596 ms to 1145 ms is obtained between a time averaging period of 1 s to 100 ms after adjusting for the COM mode duty cycle. The average decay rate obeys a quadratic dependence to omega435 and an inverse linear dependence to ∆435. The coherence time can then be extracted, given by τc = 8 / √2∆ / (omegaoσϵ). A coherence time of 596 ms to 1145 ms is obtained between a time averaging period of 1 s to 100 ms after adjusting for the COM mode duty cycle. The average decay rate obeys a quadratic dependence to omega435 and an inverse linear dependence to ∆435. The coherence time can then be extracted, given by τc = 8 / √2∆ / (omegaoσϵ). A coherence time of 596 ms to 1145 ms is obtained between a time averaging period of 1 s to 100 ms after adjusting for the COM mode duty cycle. The average decay rate obeys a quadratic dependence to omega435 and an inverse linear dependence to ∆435. The coherence time can then be extracted, given by τc = 8 / √2∆ / (omegaoσϵ).

Improvements for AI systems

textbfReal-time dynamic coolant selection for mode cooling systems: The system can dynamically assign roles by swapping the roles of the data ion and coolant to target different motional modes mid-experiment, allowing flexibility to designate coolant ions mid-experiment while circumventing the need for physically moving ions to achieve the same.
textbfAdaptive duty cycle control based on heating rate thresholds: The system can infer optimal cooling strategies by monitoring baseline heating rates (e.g., 30 q/s for COM mode) and adjust the cooling pulse periodicity and duration to maintain a target motional state, as shown by the steady-state limit calculation of n̄ = 0.67 + 0.14 q after periodic cooling cycles.
textbfEnhanced coherence preservation through laser intensity stabilization: By implementing an intensity lock for the laser to stabilize the fluctuations, the system can minimize decoherence arising from light shift noise, leading to improved coherence times by addressing the fluctuations in qubit frequency induced by the light shift.