Single Atoms with 6000-Second Trapping Lifetimes in Optical-Tweezer Arrays at Cryogenic Temperatures

We report on the trapping of single $\mathrm{Rb}$ atoms in tunable arrays of optical tweezers in a cryogenic environment at approximately 4 K. We describe the design and construction of the experimental apparatus, based on a custom-made UHV-compatible closed-cycle cryostat with optical access. We demonstrate the trapping of single atoms in cryogenic arrays of optical tweezers, with lifetimes up to 6000 s, despite the fact that the vacuum system has not been baked out. These results open the way to large arrays of single atoms with extended coherence, for applications in large-scale quantum simulation of many-body systems and, more generally, in quantum science and technology.

Figure 1

The experimental setup. (a) A schematic rendering of the entire apparatus, comprising the atomic source and the cryostat. (b) A longitudinal cross section of the cryostat, showing the pulse-tube refrigerator with its two stages at 30 K and 4 K, together with the corresponding thermal shields to which they are connected through vibration-decoupling copper braids. (c) An enlarged cross section of the lens-holder piece.

Figure 2

The fluorescence decay of the MOT. An exponential decay fit at long times (white dashed line) gives a $1/e$ decay time of about 140 s. The inset shows the MOT cloud (arrow) facing one of the aspheric lenses.

Figure 3

(a) The survival probability of a single atom as a function of the time it is held in optical tweezers, without PGC (red), with continuous PGC (green), and with a 15-ms pulse of PGC every 10 s (blue). The exponential-decay fits (solid lines) give $1/e$ decay times of about 162 s for continuous PGC and of 335 s for pulsed PGC. The inset shows an averaged fluorescence image of the $9\times 9$ tweezer array, with a spacing of $10\mu \mathrm{m}$ between adjacent microtraps. (b) Lifetime measurement for pulsed PGC (blue) after improvement of the vacuum [28, 29], note the change in the horizontal scale. The 1/e decay time is now 6050 s (solid line). For comparison, the no-PGC and continuous-PGC curves from panel (a) are plotted again.

Figure 4

(a) Examples of fluorescence images of a $6\times 6$ array, showing, between successive frames, either the loss of an atom accompanied by the loading of a previously empty trap (left) or the “simultaneous” loss of two atoms (right). (b) Experimental fluorescence traces, showing the evolution of the array occupancy as a function of time, with correlated loss-and-recapture events highlighted in white and red, respectively. (c) The result of a Monte Carlo simulation of the simple model discussed in the text, showing the same qualitative behavior as the experimental traces in (b).