Autonomous multi-ion optical clock with on-chip integrated photonic light delivery

TL;DR

This work demonstrates a fully autonomous, chip-scale optical clock built from an ensemble of four $^{171}$Yb$^{+}$ ions in a multi-site trap with all light delivered on-chip via integrated photonics. The system integrates automated ion loading, shuttling, and interleaved clock interrogation across multiple sites, achieving a short-term fractional frequency instability of $${3.14(5)}\times 10^{-14}/\sqrt{\tau}$$ and approaching the quantum projection noise limit for the given ion number. Site uniformity measurements show negligible first-order clock-shift differences between sites (below $1.75$ Hz), supporting scalability to larger arrays. The results establish a practical, manufacturable architecture for portable, multi-ion quantum sensors and quantum computing platforms, with clear pathways to improvement through vacuum upgrades, reduced heating, and advanced cooling techniques.

Abstract

Integrated photonics in trapped-ion systems are critical for the realization of applications such as portable optical atomic clocks and scalable quantum computers. However, system-level integration of all required functionalities remains a key challenge. In this work, we demonstrate an autonomously operating optical clock having a short-term frequency instability of $3.14(5)\times 10^{-14} / \sqrtτ$ using an ensemble of four $^{171}\textrm{Yb}^{+}$ ions trapped in a multi-site surface-electrode trap at room temperature. All clock operations are performed with light delivered via on-chip waveguides. We showcase the system's resilience through sustained, autonomous operation featuring automated ion shuttling and reloading to mitigate ion loss during interleaved clock measurements. This work paves the way beyond component-level functionality to establish a viable and robust architecture for the next generation of portable, multi-ion quantum sensors and computers.

Figures (6)

• Figure 1: Schematic image for the integrated multi-ion trap(a) Illustration of trapped ions above the surface of the integrated waveguide chip, with the loading site indicated by the star-burst and the clock sites indicated by clock icons. The trapping regions where the ions are addressed consist of waveguides on two different embedded layers (reds and blues), DC electrodes (green) and RF electrodes (teal) both on the top layer. (b) Camera view of trapped ions. As a check, ions are aligned to a camera prior to fine alignment with a multi-channel PMT array. Ions are imaged from the side to reduce background photons from the light scattering out of the waveguides and off the chip surface.
• Figure 2: Multi-ion clock protocol(a) Pulse sequence used in the interleaved measurement. The sequence involves (i) cooling immediately followed by state preparation (ii). Afterwards, the sequence is handed off to one of two independent second-order integrators. They apply their followed frequency plus (in the R case) or minus (in the L case) half the transition's spectroscopic linewidth, obtained prior to running the clock. Operation (iii) indicates the read out period of the ion. The green vertical lines indicate when the algorithm updates the center frequency of each integrator. Requests for new ions are also sent at this point if they occur at the reporting period. (b) Cartoon of the refill procedure that occurs as the clock is running. Ions are loaded at a loading site (red, dashed vertical line) and shuttled to clock sites (purple, dashed vertical lines). The star-burst indicates the loading of a new ion. When multiple ions are lost simultaneously, the farthest sites from the loading site are preferentially filled first.
• Figure 3: Allan deviation. Single-integrator fractional frequency instability. The fit line of 3.14(5)$\times 10^{-14} / \sqrt{\tau}$ is plotted in gray. The expected quantum projection noise of 4 ions and a $T_{\text{cycle}}$ of four side interrogations is plotted in orange.
• Figure 4: Uniformity across multiple sites.(a) Representative Rabi flops. Ion positions in the waveguide beams are qualitatively tuned using DC electrodes to achieve Rabi flops with similar features. Fit parameters according to equation (\ref{['eq:dephaseFit']}) are listed in Table \ref{['tab:rabiFit']}. (b) First-order frequency shifts for single-ion data. Each mean is consistent with zero.
• Figure 5: A sample of the DC potential at the ions' height above the chip. Here ions are shuffled from sites 1 and 2 to sites 2 and 3, respectively, to fill a vacancy at site 3. Ions at the loading site and site 4 remain stationary and have minimal perturbations to their respective DC axial trapping potentials. The minimums containing ions are labeled according to the site they originated from in "stationary" case. The colors indicate the step to which they belong. Vertical dashed-lines mark the regions addressable by different waveguides. The loading region marked with a red vertical dashed-line. The clock sites used in this experiment are marked with purple vertical dashed-lines.