High-fidelity transport of trapped-ion qubits in a multilayer array

We investigate a scalable trapped-ion array architecture for quantum control of many particles, extending techniques to access multiple dimensions. Our experimental study focuses on automated loading and shuttling of single ${\mathrm{Mg}}^{+}$ ions within our platform that contains up to 13 trapping sites in a three-dimensional arrangement. When shuttling ions from a loading hub to multiple sites, we achieve a success rate of over 0.99 999 and demonstrate coherence preservation and coherent manipulation of superposition states of a hyperfine qubit during the process. Our findings suggest promising applications of these techniques in future large-scale quantum architectures.

Figure 1

Illustration of our trap-array architecture. In the background, we show a schematic of the two radio-frequency and 30 control electrodes used for the generation and fine control of our trapping landscape. Overall, we create 13 discrete trapping sites, marked by disks in the overlay. We depict calculated pseudopotential equipotential surfaces for 0.5 meV (red, high opacity), mutually separating seven regions with single or multiple sites, and 4 meV (red, low opacity with black mesh). To effectively open and close inter-regional ion transport channels, we switch between dedicated control potentials. We highlight four sites that define a unit cell of a $\mathrm{2D}+$ lattice in the shape of a triangular pyramid (dashed lines). The inset shows a false-color fluorescence image of four single ions at ${\mathrm{T}}_{\text{0}}$, ${\mathrm{T}}_{\text{1}}$, ${\mathrm{T}}_{\text{2}}$, and ${\mathrm{T}}_{\text{H}}$, respectively. Ions are observed at an ${16}^{\circ }$ angle relative to the surface normal, and their positions are adjusted by local control fields.

Figure 2

Automated loading of single ions at the hub for subsequent redistribution within the unit cell. Inset: A single loading attempt is comprised of the following sequence: (1) Ablation (red, 5–10 pulses), photoionization (yellow), and cooling lasers (blue), as well as control potentials, are engaged for loading in ${\mathrm{T}}_{\text{H}}$; (2) cooling laser remains on during a triggered snapshot of our camera, and the image is analyzed for the presence of a single ion; and (3) a successfully loaded ion is kept Doppler cooled. Typically, we use about 5–10 laser ablation pulses in a single loading attempt. The histogram shows the accumulated 973 attempts of $\approx 180$ days of data taking. The accumulated failure probability (right scale and red disks) decreases below $10\%$ after five loading attempts.

Figure 3

Quality of shuttling single ions from ${\mathrm{T}}_{\text{H}}$ to ${\mathrm{T}}_{\text{1}}$ and back (round trip). We enable transport by playback of wave forms between fine-tuned control potential configurations ${\mathrm{\Phi }}_{\text{H}}$ and ${\mathrm{\Phi }}_{\text{1}}$ within 0.1 ms. During reference experiments, we fix the control potential configuration to ${\mathrm{\Phi }}_{\text{H}}$. In (a), we illustrate transport (squares) and reference (disks) sequences that initialize (I) and detect (D) ${}^{24}\mathrm{Mg}^{+}$ at ${\mathrm{T}}_{\text{H}}$. Here, we execute 11 000 repetitions of shuttling without ion loss. In (b), we show the corresponding photon count histograms (reference: disks joined by black line) and find a fluorescence change of $-2.1\left(6\right)\%$ due to shuttling. We summarize the relocation results to all sites in Table 1, and find that mode heating due to shuttling is $<300$ quanta per round trip.

Figure 4

Coherent manipulation of qubit states in ${}^{25}\mathrm{Mg}^{+}$ with multiple relocations. (a) In three single-ion sequences, we initialize (I) the $|\downarrow \rangle \text{−}|\uparrow \rangle$ superposition and probe the coherence with analysis pulses before detection (D) in ${\mathrm{T}}_{\text{H}}$. In the first and second sequences, we probe single relocation round trips to ${\mathrm{T}}_{\text{1}}$ (disks) and apply optional local AC Stark shifts (squares). In the third sequence, we study the effect of multiple subsequent relocation round trips to all sites (diamonds). In (b), we compare the resulting $P_{\downarrow }$ for all sequences as a function of the analysis phase. Model fits result in the anticipated phase shifts and the probed coherence amplitudes are consistent with the background effects.