Analysis of collision shift assessments in ion-based clocks

We consider back-ground gas collision shifts in ion-based clocks. We give both a classical and quantum description of a collision between an ion and a polarizable particle with a simple hard-sphere repulsion. Both descriptions give consistent results, which shows that a collision shift bound is determined by the classical Langevin collision rate reduced by a readily calculated factor describing the decoupling of the clock laser from the ion due to the recoil motion. We also show that the result holds when using a more general Lennard-Jones potential to describe the interaction between the ion and its collision partner. This leads to a simple bound for the collision shift applicable to any single ion clock without resorting to large-scale Monte-Carlo simulations or determination of molecular potential energy curves describing the collision. It also provides a relatively straightforward means to measure the relevant collision rate.

💡 Research Summary

This paper addresses one of the most subtle systematic effects limiting the performance of single‑ion optical clocks: the frequency shift induced by collisions with residual background gas molecules. The authors develop a unified treatment that combines a classical description based on Langevin scattering with a quantum‑mechanical model that incorporates a hard‑sphere repulsion (and later a Lennard‑Jones potential) to represent the short‑range part of the ion‑neutral interaction.

In the classical picture, the collision rate Γ_L is given by the well‑known Langevin formula, which depends only on the ion charge, the polarizability of the neutral species, and the background gas density, but not on the collision energy. For typical ultra‑high‑vacuum conditions (hydrogen at 300 K), Γ_L ≈ 3.6 × 10⁻⁴ s⁻¹ · nPa⁻¹. The authors then examine how a single collision modifies a Ramsey interrogation. Two effects are identified: (i) a sudden phase jump ϕ₁ associated with the change in the internal energy of the clock states during the collision, and (ii) a second‑order Doppler shift (SODS) ϕ₂ arising from the recoil velocity imparted to the ion.

The recoil puts the ion into a coherent motional state, which can be described by the displacement operator D(α). In the interaction picture the laser‑ion coupling acquires a modulation factor that is mathematically equivalent to a Bessel‑function term J₀(k_L v sinθ cosϕ/√2). Averaging over the isotropic distribution of recoil directions yields a “Ramsey suppression factor” (RSF) R = ⟨|β₀|⟩, where β₀ is the Bessel term. For realistic trap frequencies (≈2π × 500 kHz) and a laser propagating at 45° to the principal axes, R is of order Γ T (with T the interrogation time).

Combining the above, the authors derive a compact bound for the fractional clock shift:

 δf/f₀ ≈ (Γ/2πf₀) ⟨|R|⟩ ⟨v⟩,

where ⟨v⟩ is the average recoil speed expressed in dimensionless units. This expression is essentially the “worst‑case collision shift” (WCCS) multiplied by the velocity‑averaged RSF, showing that the dominant reduction of the shift comes from the laser‑ion decoupling caused by recoil.

The quantum treatment starts from an attractive r⁻⁴ potential (the long‑range ion‑induced dipole interaction) together with a hard‑sphere core. Solving the scattering problem yields phase shifts that, when inserted into the Ramsey formalism, reproduce exactly the same RSF‑scaled shift obtained classically. This demonstrates that the classical Langevin rate, together with the recoil‑induced suppression, captures the essential physics even when quantum effects are included.

To test the robustness of the model, the authors replace the hard‑sphere core with a Lennard‑Jones potential, which introduces a more realistic short‑range attraction and repulsion. Numerical evaluation shows that the resulting shift bound changes only marginally, confirming that precise knowledge of the molecular potential energy curves is not required for a conservative estimate.

The paper also discusses the statistical nature of the collision phase ϕ₁. If ϕ₁ is assumed to be uniformly distributed over