PHYSICAL REVIEW LETTERS 128, 123601 (2022)

Trapped Alkali-Metal Rydberg Qubit

Y. Mei, Y. Li , H. Nguyen , P. R. Berman, and A. Kuzmich*
Department of Physics, University of Michigan, Ann Arbor, Michigan 48109, USA

(Received 10 November 2021; accepted 17 February 2022; published 22 March 2022)

Rydberg interactions of trapped alkali-metal atoms are used widely to facilitate quantum gate
operations in quantum processors and repeaters. In most laboratory realizations using this protocol, the
Rydberg states are repelled by the trapping laser fields, requiring that the fields be turned off during gate
operations. Here we create a quasi-two-level system in a regime of Rydberg excitation blockade for a
mesoscopic Rb ensemble of several hundred atoms confined in a magic-wavelength optical lattice. We
observe many-body Rabi oscillations between the ground and collective Rydberg state. In addition we use
Ramsey interference techniques to obtain the light shifts of both the lower and upper states p
offfiffiffiffithe collective
qubit. Whereas the coupling producing the Rabi oscillations is enhanced by a factor of N , there is no
corresponding enhancement for the light shifts. We derive an effective two-level model which is in good
agreement with our observations. Trapped Rydberg qubits and an effective two-level description are
expected to have broad applicability for studies of quantum simulation and networking using collective
encoding in ensembles of neutral atoms.

DOI: 10.1103/PhysRevLett.128.123601

Quantum technologies offer transformative advances in they are pushed out of the laser field intensity maxima.
storage, processing, and communication of information This problem leads to the necessity of shutting off the trap
compared to established classical approaches. The recipe fields for the duration of quantum operations. The repeated
for combining distant quantum processors into a single process of turning the trap fields on and off heats the atoms
quantum network involves three key ingredients: qubits, and dramatically shortens their lifetime to μs, limiting its
quantum logic for entanglement generation and correction, utility [21,22].
and interaction interfaces [1,2]. Neutral atomic ensembles In a state-insensitive (or “magic-wavelength”) trap pro-
are a strong candidate to serve as a basis for scalable posal of Refs. [23,24], the frequency of an optical lattice is
quantum networks [3–6]. Collective qubits based on atomic tuned to the blue side of an atomic resonance from the
hyperfine ground states can be converted, on demand, into Rydberg level to an intermediate level, creating positive
single photons [7], making them well-suited for scalable polarizability of the Rydberg level. For example, if the
quantum network-type protocols over telecom-wavelength (1012 nm) lattice field is detuned by ≃ð52=nÞ3 GHz from
optical fibers [8,9]. Notably, collective atomic qubit states the jnS1=2 i ↔ j6P3=2 i transition in atomic Rb, the depth of
between ground and Rydberg states can be deterministi- the trapping potentials for the ground- and Rydberg-level
cally created and coherently manipulated in the regime of atoms are approximately equal. This method has been
the excitation blockade [10–18], allowing for dramatically demonstrated by our group [25–27] and by Goldschmidt
faster remote entanglement generation compared with et al. [28]. More recently, trapped single-atom Rydberg
probabilistic approaches [19,20]. qubits have been demonstrated in alkaline-earth atoms [29].
In order to use the Rydberg blockade in quantum Here we report observations of dynamics of a Rydberg
information processing, the (optical) ground-Rydberg qubit encoded in an ensemble of ∼103 atoms that are
atomic coherence must be preserved. Ideally, this is confined in a state-insensitive optical lattice trap. We
achieved if the atoms are confined in a potential UðrÞ that observe coherent driving and Ramsey interference mea-
is identical for the ground and the Rydberg levels. surements of light shifts induced on either the lower or the
However, in a regular far-off-resonance optical dipole trap, upper qubit state, on timescales of order ∼10 μs. The
e.g., based on a 1064 nm laser for Rubidium atoms, the dynamics of the trapped qubit and the light shifts are well
spatially inhomogeneous energy shifts UðrÞ are entirely described by an effective two-level model with possible
different for the ground and Rydberg states. For (trapped) dephasing factors included, i.e., laser phase noise, atom
ground-level atoms, the dynamic polarizability in a (red- number fluctuation, and so on. The experimental setup is
detuned) light field is positive, whereas Rydberg electrons shown in Fig. 1(a). 87Rb atoms are collected in a magneto-
are nearly free, and their polarizability is approximately optical trap (MOT) and are subsequently loaded into a far-
equal to that of a free electron, which is negative. This off resonance (YAG, 1064 nm) cross-dipole trap. The
means that atoms in Rydberg states are antitrapped since longitudinal extent of the atomic cloud is ∼10 μm along

0031-9007=22=128(12)=123601(6) 123601-1 © 2022 American Physical Society

 PHYSICAL REVIEW LETTERS 128, 123601 (2022)

(a) FORT excitation blockade regime, ideally, the atoms will

undergo an oscillation between Q two collective
(Dicke) atomic states, j0i ¼ Ni¼1 jgi i and j1i ¼
pffiffiffiffi P
87Rb MOT ð1= N Þ Ni¼1 eiðk1 þk2 Þ·ri −iðω1 þω2 Þt jg1 …ri …gN i, where N
SILT is the total number of atoms participating in the many-
body blockade. After a controllable delay, T s , following the
SPCMR
excitation pulse, a readout pulse Er (with Rabi frequency
Ωr ) that is resonant with the jri to jpi transition frequency
SPCMT
is applied and leads to phase-matched emission having
(b) (c)
frequency ωe (see the Supplemental Material [30] for
experimental details).
We model our system as an effective two-level system
(d)
1.0 with a closed transition between the states j0i and j1i [30].
0.8 The collective
pffiffiffiffi Rabi frequency of this two-state model is
0.6
ΩN ¼ N Ω, where Ω ¼ ðΩ1 Ω2 =2ΔÞ is the effective two-
photon, single-atom Rabi frequency. With the two-photon
0.4
detuning δ set equal to zero, the population of the many-
γ1 pffiffiffiffi
0.2
body Rydberg state j1i is ρ11 ¼ 12 ð1 − e− 2 T p cos N ΩT p Þ,
0.0
0 10 20 30 40 50 where T p is the pulse duration of a constant amplitude
pulse whose pulse area is equal to that of the actual pulse.
We have incorporated dephasing from relaxation processes
FIG. 1. (a) Experimental setup: an ultracold 87Rb atomic such as that produced by laser frequency noise into our
ensemble is prepared in a one-dimensional state-insensitive model by assuming that the coherence ρ01 decays at rate γ 1 ,
lattice trap (SILT) formed by a 1012 nm retroreflected beam which agrees well with the measured laser linewidth in our
using atoms that have been transferred from a crossed far-off-
system [30]. The intensity of the retrieved signal is propor-
resonance dipole trap (FORT) formed by focused yttrium
aluminum-garnet (YAG) laser beams. Excitation fields E1
tional to this population.
(780 nm) and E2 (480 nm) drive atoms from jgi to jpi and To describe the combined effects of spin-wave dephasing
from jpi to jri, respectively. A retrieval pulse Er leads to phase- and loss of the Rydberg state component, we include
matched emission that is coupled into a pair of single-mode fibers an overall exponential damping of the retrieved signal
and subsequently measured by single-photon counting modules ∼e−αðT s þT p Þ . We account for fluctuations in atomic number
SPCMT and SPCMR . (b) Single atom energy levels for 87Rb: by weighting the retrieved signal with the probability
jgi ¼ j5S1=2 ; F ¼ 2; mF ¼ −2i, jpi ¼ j5P3=2 ; F ¼ 3; mF ¼ −3i, distribution fðk; NÞ to have k atoms in the interaction
and jri ¼ jnS1=2 ; mJ ¼ −1=2i. (c) Timing sequence for the volume when the average number in the volume is N. On
ground-Rydberg spin-wave coherence measurement. (d) Normal- averaging over k for the Poisson distribution fðkÞ ¼
ized signal η as a function of storage time T s for quantum number ðN k e−N =k!Þ, we find that the oscillation amplitude
n ¼ 75. The storage efficiency is normalized to that at 1 μs. Blue pffiffiffi is
and red bands represent temperatures 40% lower and higher than damped by a factor expð−T 2P =τ21 Þ, where τ1 ¼ 2 2=Ω,
the best-fit value, respectively. and that the photoelectric detection probability per trial p1
for the retrieved signal is given by
1 γ1 2 2
pffiffiffiffi
the excitation field direction. Next, the atoms are trans- p1 ≃ ξeð−αðT p þT S ÞÞ ð1 − e− 2 T p e−T p =τ1 cos N ΩT p Þ; ð1Þ
ferred from the crossed dipole trap to a state-insensitive 2
optical lattice. The lattice is formed by a horizontally where ξ is the overall retrieval and detection efficiency.
polarized and retroreflected laser field, which is detuned Intensity fluctuations of the driving fields would also damp
from the j6P3=2 i ↔ j75S1=2 i transition by Δm =2π ≃ the oscillation visibility, but these effects are negligible in
367 MHz—the so-called “magic” value of the detuning our experiment (see Ref. [30] for details for the intensity
for which the potential depths for the ground and the stabilization).
Rydberg atomic states are equal. After the transfer, the The normalized storage efficiency η [26,30] is plotted as
lattice depth is adiabatically lowered. a function of storage period T s for T p ¼ 1 μs and a trap
The atoms are driven by laser fields E1 (780 nm, σ − ) and depth of ≃40 μK in Fig. 1(c), showing that the coherence
E2 (480 nm, σ þ ) from the ground state jgi ¼ j5S1=2 ; F ¼ lifetime for the ground-Rydberg coherence can be extended
2; mF ¼ −2i to the Rydberg state jri ¼ j75S1=2 ; mJ ¼ −1=2i up to ≃20 μs, which is an order of magnitude improvement
with a detuning of Δ=2π ¼ 480 MHz from the intermedi- over that achieved with atomic ensembles in free space [the
ate state jpi ¼ j5P3=2 ; F ¼ 3; mF ¼ −3i, with respective oscillation in the figure is due to motional effects not
Rabi frequencies Ω1 and Ω2 , Fig. 1(b). In the Rydberg included in Eq. (1)] [11–13,22,25]. The longer lifetime

123601-2
 PHYSICAL REVIEW LETTERS 128, 123601 (2022)
(a)
with the atomic density, but all the signals vanish within
5 μs, due to the limitation of τ1 ≈ 4.3 μs. The dashed
curves are theory fits based on Eq. (1). As the YAG power
is raised up from 6 W to 35 W, the fitted value for the
number of atoms N increases from ≃102 to ≃103 . The fit
value for the global dephasing factor γ 1 =2π ≃ 40 kHz is
consistent with the laser linewidth estimated from the
excitation spectra [30]. The best fit for α for all four sets
of data is α=2π ≈ 0.008 MHz, which agrees with the
coherence time of 20 μs shown in Fig. 1(c).
For an independent determination of the atom number
and single-atom Rabi frequency, we measure the atomic
density using absorption of a transmitted probe field, both
with and without a control-EIT (electromagnetic-induced
(b) (c) transparency) field. For example, the measured optical
depth (OD) of ≃3.5 for a YAG power of 20 W that
can be extracted from the transmission curves shown in
Fig. 2(b) corresponds to an atomic density having peak
value of 2.9 × 1011 cm−3 . In Fig. 2(c) we plot the normal-
ized collective Rabi frequency ΩN =Ω as a function of the
number of atoms N a in the interaction volume (see
Ref. [30] for details). We confirm thepcollective
ffiffiffiffi Rabi
FIG. 2. (a) Collective Rabi oscillation as a function of the
pulse duration T p for different numbers of atoms. Red: frequency ΩN is enhanced by a factor N with respect
N ¼ 109ð2Þ; green: N ¼ 326ð3Þ; blue: N ¼ 755ð3Þ; purple: to the single-atom value, by fitting the function ΩN ¼ ΩN ka
N ¼ 930ð4Þ. Here, Ω1 =2π ¼ 9.2 MHz, Ω2 =2π ¼ 10.8 MHz, and finding a best-fit value k ¼ 0.463ð5Þ. The discrepancy
and Ωr =2π ¼ 11.5 MHz. The dashed lines are theoretical results for low atom numbers results from a relatively large error of
using an effective two-state
pffiffiffiffiffi model. The error bars represent one OD fit value when the absorption is small.
standard deviation ( M ) for M photoelectric counting events. We confirm that the Rydberg blockade is fully opera-
(b) Probe transmission (orange squares) and EIT (blue diamonds) tional in our system by measuring the second-order
measurement for N ¼ 755ð9Þ, consistent with an OD ¼ 3.5. autocorrelation function gð2Þ ð0Þ < 0.2 within the time
(c) The enhancement of the collective Rabi frequency ΩN =Ω interval of Rabi oscillations [Fig. 3] and demonstrate the
as a function of number of atoms N a determined by the multiparticle entanglement [41–43] of the W state j1i
absorption measurement. The data are fit with a function ΩN ¼
within the ensemble [30]. To investigate the main contri-
ΩN ka with the best-fit value k ¼ 0.463ð5Þ. The error bars
butions to the damping of the Rabi oscillations, we vary the
represent the standard errors of the respective fits.
Rabi frequency and the atom number in a shallower trap
depth of 40 μK to minimize the effects of atomic thermal
allows us to vary the excitation pulse duration to tens of μs, motion and collisions. In Fig. 3(a), obtained with a lower
instead of varying the excitation field strength as we did Rabi frequency, 9 oscillations occur within a time window
previously [12]. As a result, the light shifts caused by the of 12 μs. Approximately four times higher Rabi frequency,
excitation laser fields can be kept constant over the extent but fewer atoms in the blockade volume are used for data in
of the measurement. The methods we employ to control Fig. 3(b). In this limit, 13 oscillation cycles are damped
laser phase noise and intensity fluctuations, to reduce within ≃6 μs. These results indicate that in the case of
electric field shifts, to minimize atom number variation Fig. 3(a), the dephasing is due mainly to the dephasing
or loss, and to suppress the effects of atomic thermal motion parameter γ 1 term. If this parameter is set equal to zero
are outlined in Ref. [30]. [resulting in the dotted line in Fig. 3(a)], the oscillations
The population of the upper qubit state is measured by damp more slowly than the data, while the atom number
mapping it into a phase-matched retrieved field. To measure fluctuations have little impact on the visibility damping
the dependence of the collective Rabi frequency on density, [resulting in the dash-dotted line in Fig. 3(a)]. In the case of
the associated photoelectric detection probabilities per trial Fig. 3(b), the damping of the oscillations can be ascribed
p1 are recorded as a function of the excitation pulse chiefly to the Poisson distribution of number of atoms, N e ,
duration time T p , while the storage period T s ¼ 1 μs is that limits the visible number of the oscillation periods. If
kept constant. In Fig. 2, the resulting collective Rabi atom number fluctuations are neglected [resulting in the
oscillations are shown for a lattice trap depth of 60 μK dash-dotted line in Fig. 3(b)], the oscillations persist for a
for varying atomic density, controlled by altering the YAG time longer than the experimental observation period.
power. As expected, the oscillation frequency increases However, the theory with γ 1 ¼ 0 [resulting in the dotted

123601-3
 PHYSICAL REVIEW LETTERS 128, 123601 (2022)

(a) (a)

(b)

(b)

(c)

FIG. 3. Collective Rabi oscillation as a function of the pulse

duration for different Rabi frequency, (a) Ω1 =2π ¼ 4.7 MHz,
Ω2 =2π ¼ 5.4 MHz, and N ¼ 797ð7Þ. (b) Ω1 =2π ¼ 9.2 MHz,
Ω2 =2π ¼ 10.8 MHz, and N ¼ 553ð2Þ. The dashed line shows
best fit from theory with dephasing and atom number fluctua- FIG. 4. Ramsey interferometry of the trapped qubit. (a) Sche-
tions. The dotted line shows the simulation without dephasing, matic and timing sequence. (b) Evolution of the j1i state versus
and the dash-dotted line shows the simulation without atom free evolution time T f between the two π=2 pulses of detuning
number fluctuations. The blue hollow circles represent the δs =2π ¼ −0.3 MHz and pulse width 0.45 μs. Dashed line rep-
second-order intensity correlation function at zero delay resents the sinusoidal fit with an exponential
pffiffiffiffiffi decay. The error bars
gð2Þ ð0Þ, which is below 0.2 within the Rabi oscillations, sug- represent one standard deviation ( M ) for M photoelectric
gesting a well-established Rydberg blockade. counting events. (c) The detuning extracted from the sinusoidal
fit versus detuning set to ω1 . The line represents the fitted result of
line in Fig. 3(b)] agrees well with the data. These results δe ¼ kδs þ b with k ¼ 1.022ð6Þ and b ¼ 0.000ð6Þ MHz. The
indicate that a combination of a narrower (e.g., 1 kHz) laser error bar of each point is within the size of the marker.
linewidth and a sub-Poisson atom-number distribution
[44,45] may be able to prolong the lifetime of the collective
Rabi oscillation to tens of microseconds. Notably, with a finite (0.45 μs) duration of the π=2 pulses. We fit our
large intermediate state detuning and a shallow lattice trap results using δe ¼ kδs þ b, with k ¼ 1.022ð6Þ and b ¼
depth, spontaneous decay from the Rydberg level and the 2π × 0.000ð6Þ MHz [Fig. 4(b)].
influence of blackbody radiation can be neglected in our To determine whether or not there are any collective light
system. shifts, we add a dressing field E1 or E2 during the T f
To further investigate the nature of the collective period. With the presence of the dressing field, the
state, we perform Ramsey interferometry. First, we apply oscillation frequency of the Ramsey interferometry will
two off-resonance π=2 pulses separated by a variable free be changed from δe to δe þ ΔE=ℏ, where ΔE is the light
evolution time T f , followed by a phase-matched retrieval of shift produced by the two excitation fields. As is shown in
the j1i state population. Using this protocol, we obtain the Ref. [30], it is expected that ΔE ¼ ℏðΩ21 − Ω22 Þ=ð4ΔÞ, with
oscillation of the retrieved signal as a function of T f shown no enhancement from collective effects. We measure the
in Fig. 4(a). In order to explore different Ramsey scenarios, dependence of ΔE on Ω1;2 by changing the power of
we change the two photon detuning by varying the the dressing field and observing the change in frequency
detuning δs of the ω1 field. We extract the measured of the Ramsey interferometry signal. We observe either an
detuning δe from the oscillation period in the Ramsey increase or a decrease in oscillation frequency as we increase
interferometry by the fitting equation p1 =p1 ð0Þ ¼ Ω1 and Ω2 respectively, as shown in Figs. 5(a) and 5(b).
1
2 ½1 þ exp ð−γT f Þ cos ðδe T f þ ϕÞ. The decay time con- Figure 5(c) shows the light shift induced by different Ω1 and
stant 1=γ for the Ramsey interferometry in Fig. 4(a) is Ω2 as well as theory curves, confirming that there is no
6.89  1.3 μs. The phase offset ϕ=2π ¼ 0.14 is due to the relative collective light shift between the two levels.

123601-4
 PHYSICAL REVIEW LETTERS 128, 123601 (2022)
(a) (c)
teleportation with photonic and atomic qubits, Nat. Phys. 4,
103 (2008).
[7] D. P. Ornelas-Huerta, A. N. Craddock, E. A. Goldschmidt,
A. J. Hachtel, Y. Wang, P. Bienias, A. V. Gorshkov, S. L.
Rolston, and J. V. Porto, On-demand indistinguishable
(b) single photons from an efficient and pure source based
on a Rydberg ensemble, Optica 7, 813 (2020).
[8] A. G. Radnaev, Y. O. Dudin, R. Zhao, H. H. Jen, S. D.
Jenkins, A. Kuzmich, and T. A. B. Kennedy, Quantum
memory with telecom wavelength conversion, Nat. Phys.
6, 894 (2010).
[9] Y. O. Dudin, A. G. Radnaev, R. Zhao, J. Z. Blumoff, T. A. B.
Kennedy, and A. Kuzmich, Entanglement of Light-Shift
FIG. 5. Ramsey interferometry with dressing field light shift. Compensated Atomic Spin Waves with Telecom Light,
(a) Ramsey interferometry with (blue circles) and without (red Phys. Rev. Lett. 105, 260502 (2010).
triangles) dressing field of Ω1 =2π ¼ 3.9 MHz. (b) Ramsey [10] M. Saffman, T. G. Walker, and K. Mølmer, Quantum
interferometry with (blue circles) and without (red triangles) information with Rydberg atoms, Rev. Mod. Phys. 82,
dressing field of Ω2 =2π ¼ 10.8 MHz. The detuning of the π=2 2313 (2010).
pulses is set to δs =2π ¼ −0.3 MHz. (c) Light shift ΔE versus [11] Y. O. Dudin and A. Kuzmich, Strongly interacting
dressing field Rabi frequency Ω1 and Ω2 . Red squares represent Rydberg excitations of a cold atomic gas, Science 336,
varying the Ω1 field and blue circles represent varying the Ω2 887 (2012).
field. Black lines represent the theoretical curve of [12] Y. O. Dudin, L. Li, F. Bariani, and A. Kuzmich, Observation
ΔE ¼ ℏðΩ21 − Ω22 Þ=ð4ΔÞ. of coherent many-body Rabi oscillations, Nat. Phys. 8, 790
(2012).
[13] M. Ebert, A. Gill, M. Gibbons, X. Zhang, M. Saffman, and
In summary, we have demonstrated long-lived Rabi T. G. Walker, Atomic Fock State Preparation Using Rydberg
oscillations and measured light shifts for a collective Blockade, Phys. Rev. Lett. 112, 043602 (2014).
Rydberg qubit held in an optical lattice that is state [14] L. Li and A. Kuzmich, Quantum memory with strong and
insensitive for the ground and Rydberg levels. Whereas controllable Rydberg-level interactions, Nat. Commun. 7,
the coupling 13618 (2016).
pproducing
ffiffiffiffi the Rabi oscillations is enhanced by
a factor of N , there is no corresponding enhancement for [15] W. Xu, A. V. Venkatramani, S. H. Cantú, T. Šumarac, V.
the light shifts. These results provide new evidence that Klüsener, M. D. Lukin, and V. Vuletić, Fast Preparation and
collective Rydberg qubits can be used to create high- Detection of a Rydberg Qubit Using Atomic Ensembles,
Phys. Rev. Lett. 127, 050501 (2021).
fidelity photon-photon gates [10], deterministic single
[16] N. L. R. Spong, Y. Jiao, O. D. W. Hughes, K. J. Weatherill,
photons [7], and multiple qubits [46] for scalable quantum I. Lesanovsky, and C. S. Adams, Collectively Encoded
networking. Rydberg Qubit, Phys. Rev. Lett. 127, 063604 (2021).
The work was supported by the Air Force Office of [17] H. Levine, A. Keesling, A. Omran, H. Bernien, S. Schwartz,
A. S. Zibrov, M. Endres, M. Greiner, V. Vuletic, and
Scientific Research and the National Science Foundation.
M. D. Lukin, High-Fidelity Control and Entanglement of
Y. M. and Y. L. contributed equally to this work. Rydberg-Atom Qubits, Phys. Rev. Lett. 121, 123603
(2018).
[18] I. S. Madjarov, J. P. Covey, A. L. Shaw, J. Choi, A. Kale, A.
Cooper, H. Pichler, V. Schkolnik, J. R. Williams, and
* M. Endres, High-fidelity entanglement and detection of
[email protected]
[1] C. Simon, Towards a global quantum network, Nat. Pho- alkaline-earth Rydberg atoms, Nat. Phys. 16, 857 (2020).
tonics 11, 678 (2017). [19] E. Brion, F. Carlier, V. M. Akulin, and K. Mølmer, Quantum
[2] S. Wehner, D. Elkouss, and R. Hanson, Quantum internet: A repeater with Rydberg-blocked atomic ensembles in fiber-
vision for the road ahead, Science 362, eaam9288 (2018). coupled cavities, Phys. Rev. A 85, 042324 (2012).
[3] D. N. Matsukevich and A. Kuzmich, Quantum state transfer [20] Y. Han, B. He, K. Heshami, C.-Z. Li, and C. Simon,
between matter and light, Science 306, 663 (2004). Quantum repeaters based on Rydberg-blockade coupled
[4] T. Chanelière, D. N. Matsukevich, S. D. Jenkins, S.-Y. Lan, atomic ensembles, Phys. Rev. A 81, 052311 (2010).
T. A. B. Kennedy, and A. Kuzmich, Storage and retrieval of [21] H. Labuhn, D. Barredo, S. Ravets, S. de Léséleuc, T. Macri,
single photons transmitted between remote quantum memo- T. Lahaye, and A. Browaeys, Tunable two-dimensional
ries, Nature (London) 438, 833 (2005). arrays of single Rydberg atoms for realizing quantum Ising
[5] J. K. Thompson, J. Simon, H. Loh, and V. Vuletic, A high- models, Nature (London) 534, 667 (2016).
brightness source of narrowband identical-photon pairs, [22] J. Zeiher, P. Schauß, S. Hild, T. Macrì, I. Bloch,
Science 313, 74 (2006). and C. Gross, Microscopic Characterization of Scalable
[6] Y. A. Chen, S. Chen, Z.-S. Yuan, B. Zhao, C.-S. Chuu, J. Coherent Rydberg Superatoms, Phys. Rev. X 5, 031015
Schmiedmayer, and J.-W. Pan, Memory-built-in quantum (2015).

123601-5
 PHYSICAL REVIEW LETTERS 128, 123601 (2022)

[23] M. S. Safronova, C. J. Williams, and C. W. Clark, Optimiz- 66 (2021), https://meetings.aps.org/Meeting/DAMOP21/

ing the fast Rydberg quantum gate, Phys. Rev. A 67, 040303 Session/N01.38.
(2003). [36] G. M. Stéphan, T. T. Tam, S. Blin, P. Besnard, and M. Têtu,
[24] M. Saffman and T. G. Walker, Analysis of a quantum logic Laser line shape and spectral density of frequency noise,
device based on dipole-dipole interactions of optically Phys. Rev. A 71, 043809 (2005).
trapped Rydberg atoms, Phys. Rev. A 72, 022347 (2005). [37] Y. Mei, Y. Li, H. Nguyen, P. R. Berman, and A. Kuzmich,
[25] L. Li, Y. O. Dudin, and A. Kuzmich, Entanglement between Dynamics of interaction-induced dephasing for collective
light and an optical atomic excitation, Nature (London) 498, Rydberg excitations (unpublished).
466 (2013). [38] Y. O. Dudin, F. Bariani, and A. Kuzmich, Emergence of
[26] J. Lampen, H. Nguyen, L. Li, P. R. Berman, and A. Spatial Spin-Wave Correlations in a Cold Atomic Gas, Phys.
Kuzmich, Long-lived coherence between ground and Ryd- Rev. Lett. 109, 133602 (2012).
berg levels in a magic-wavelength lattice, Phys. Rev. A 98, [39] F. Bariani, Y. O. Dudin, T. A. B. Kennedy, and A. Kuzmich,
033411 (2018). Dephasing of Multiparticle Rydberg Excitations for Fast
[27] H. Nguyen, J. Lampen, P. R. Berman, and A. Kuzmich, Entanglement Generation, Phys. Rev. Lett. 108, 030501
Differential nuclear-spin-dependent light shifts and state (2012).
mixing of Rydberg atoms, Phys. Rev. A 100, 033412 [40] N. Šibalić, J. D. Pritchard, C. S. Adams, and K. J.
(2019). Weatherill, Arc: An open-source library for calculating
[28] E. A. Goldschmidt, D. G. Norris, S. B. Koller, R. Wyllie, properties of alkali Rydberg atoms, Comput. Phys. Com-
mun. 220, 319 (2017).
R. C. Brown, J. V. Porto, U. I. Safronova, and M. S.
[41] F. Fröwis, P. C. Strassmann, A. Tiranov, C. Gut, J.
Safronova, Magic wavelengths for the 5s-18s transition in
Lavoie, N. Brunner, F. Bussières, M. Afzelius, and N.
rubidium, Phys. Rev. A 91, 032518 (2015).
Gisin, Experimental certification of millions of genuinely
[29] J. T. Wilson, S. Saskin, Y. Meng, S. Ma, R. Dilip, A. P.
entangled atoms in a solid, Nat. Commun. 8, 907 (2017).
Burgers, and J. D. Thompson, Trapping Alkaline Earth [42] R. McConnell, H. Zhang, J. Hu, S. Cuk, and V. Vuletic,
Rydberg Atoms Optical Tweezer Arrays, Phys. Rev. Lett. Entanglement with negative Wigner function of almost
128, 033201 (2022). 3,000 atoms heralded by one photon, Nature (London)
[30] See Supplemental Material at http://link.aps.org/supplemental/ 519, 439 (2015).
10.1103/PhysRevLett.128.123601 for more theoretical and [43] F. Haas, J. Volz, R. Gehr, J. Reichel, and J. Estève,
experimental details, which includes Refs. [31–40]. Entangled states of more than 40 atoms in an optical fiber
[31] G. S. Agarwal, Exact Solution for the Influence of Laser cavity, Science 344, 180 (2014).
Temporal Fluctuations on Resonance Fluorescence, Phys. [44] Y. Wang, S. Shevate, T. M. Wintermantel, M. Morgado, G.
Rev. Lett. 37, 1773 (1976). Lochead, and S. Whitlock, Preparation of hundreds of
[32] G. S. Agarwal, Quantum statistical theory of optical-resonance microscopic atomic ensembles in optical tweezer arrays,
phenomena in fluctuating laser fields, Phys. Rev. A 18, 1490 npj Quantum Inf. 6, 54 (2020).
(1978). [45] S. Whitlock, C. F. Ockeloen, and R. J. C. Spreeuw, Sub-
[33] S. de Lèséleuc, D. Barredo, V. Lienhard, A. Browaeys, and Poissonian Atom-Number Fluctuations by Three-Body
T. Lahaye, Analysis of imperfections in the coherent optical Loss in Mesoscopic Ensembles, Phys. Rev. Lett. 104,
excitation of single atoms to Rydberg states, Phys. Rev. A 120402 (2010).
97, 053803 (2018). [46] M. Khazali and K. Mølmer, Fast Multiqubit Gates by
[34] P. Cladé, Ph.D. thesis, Université Paris 6, 2004. Adiabatic Evolution in Interacting Excited-State Manifolds
[35] X. Jiang, M. Friesen, J. Scott, and M. Saffman, The effect of of Rydberg Atoms and Superconducting Circuits, Phys.
laser noise on Rabi oscillation fidelity, Bull. Am. Phys. Soc. Rev. X 10, 021054 (2020).

123601-6