sign in sign up
<<
Home , Einstein coefficients, Quantum clock, Time dilation, Time in physics

Quantum dilation in atomic spectra

Piotr T. Grochowski ,1, ∗ Alexander R. H. Smith ,2, 3, † Andrzej Dragan ,4, 5, ‡ and Kacper Dębski 4, § 1Center for Theoretical , Polish Academy of , Aleja Lotników 32/46, 02-668 Warsaw, Poland 2Department of Physics, Saint Anselm College, Manchester, New Hampshire 03102, USA 3Department of Physics and Astronomy, Dartmouth College, Hanover, New Hampshire 03755, USA 4Institute of , University of Warsaw, Pasteura 5, 02-093 Warsaw, Poland 5Centre for Quantum Technologies, National University of Singapore, 3 Drive 2, 117543 Singapore, Singapore (Dated: March 9, 2021) Quantum occurs when a moves in a superposition of relativistic momentum wave packets. The lifetime of an excited hydrogen-like atom can be used as a clock, which we use to demonstrate how quantum time dilation manifests in a process. The resulting emission rate differs when compared to the emission rate of an atom prepared in a mixture of momentum wave packets at order v2/c2. This effect is accompanied by a quantum correction to the Doppler shift due to the coherence between momentum wave packets. This quantum Doppler shift affects the shape at order v/c. However, its effect on the decay rate is suppressed when compared to the effect of quantum time dilation. We argue that spectroscopic experiments offer a technologically feasible platform to explore the effects of quantum time dilation.

I. INTRODUCTION The purpose of the work is to provide evi- dence in support of the conjecture that quantum time The quintessential feature of is dilation is universal. We consider the lifetime of an ex- the superposition principle. When combined with rel- cited hydrogen-like atom as a clock [37] and demonstrate ativistic effects, this principle gives rise to a number of that when such an atom moves in a coherent superposi- exciting and novel phenomena [1–28]. In particular, it tion of momenta its lifetime experiences the same quan- is natural to ask whether there is a quantum contribu- tum time dilation as the considered in [36]. This tion to the time dilation observed by a clock moving yields a spectroscopic signature of a clock experiencing in a superposition of relativistic . This question a superposition of proper , alternative to in- has been examined in several contexts: a modified twin- terferometry proposals that aim to observe a decrease in paradox in which one twin is placed in a superposition interference visibility [31–34]. of [29]; an analogue twin-paradox scenario in su- Spectroscopic signatures of classical time dilation have perconducting circuits [30]; interferometry experiments been observed for atoms moving as speeds as low as in which a clock experiences a superposition of proper 10 m/s [38]. Nonclassical effects in emission spectroscopy times [31–34]; and sequential boosts of quantum clocks due to the coherent spreading of the atomic center-of- mimicking a twin-like scenario have been shown to lead were first studied in the early 1990s to nonclassical effects in trap atomic clocks [35]. [39–42], and the effect of center-of-mass superposition Recently, a probabilistic formulation of relativistic was recently investigated in a field model [43]. In time dilation observed by quantum clocks was devel- the present work, we show that the exact quantum time oped [36]. It was shown that a clock moving in a lo- dilation effect described in [36] is observed in the sponta- calized momentum observes on average clas- neous decay rate of an atom moving in a coherent super- sical time dilation in accordance with . position of relativistic momenta. This observation moti- However, the time dilation observed by a clock moving in vates a new class of spectroscopic that are a coherent superposition of two momentum wave packets sensitive to relativistic effects due to quantum coherence. experiences quantum correction compared to a classical In addition, a novel correction to the classical Doppler arXiv:2006.10084v3 [quant-ph] 7 Mar 2021 clock moving in a probabilistic mixture of the same two shift is shown to modify the shape of the atomic emission wave packets. This quantum time dilation effect was es- spectrum. This correction is present for light emitted in tablished for an idealized model of a clock, and it thus the direction of the atom’s . On the other hand, remains open as to whether quantum time dilation is uni- if the spectrum is measured through perpendic- versal, analogous to the way in which classical time dila- ular to motion, effects that are first-order in momentum tion affects all clocks in the same way. vanish and give way to -order, relativistic correc- tions. This is clearly seen from the angular distributions of coming from moving atoms which we present. These distributions show the directions of emission that are the most affected by motion and suggest the optimal ∗ [email protected] † [email protected] way to measure the quantum time dilation. ‡ [email protected] Furthermore, we analyze potential experimental sce- § [email protected] narios in which both the quantum Doppler and the quantum time dilation effects can be measured. As of 2 now, spectroscopic experiments have been able to ob- is associated with the classical time dilation of a clock serve classical time dilation in atoms moving as slowly moving in a statistical mixture of momenta p¯1 and p¯1 as 10 m/s [38]. Basing on that and the subsequent ad- with probabilities cos2 θ and sin2 θ, and vances, we argue that state-of-the-art techniques involv- ing atomic ion clocks and large momentum transfer se- h 2 2 2 i tups can reach necessary parameter regimes. cos φ sin 2θ (p¯2 − p¯1) − 2 p¯2 − p¯1 cos 2θ γ−1 ≡ , (5) Q 2  (p¯2−p¯1)  8m2c2 cos φ sin 2θ + e 4∆2 II. QUANTUM TIME DILATION quantifies corrections to classical time dilation resulting As a model of a , consider a relativistic from coherence across the momentum wave packets car- particle with an internal degree of freedom, described by rying the internal clock. Equation (3) can be thought the Hamiltonian of as the generalization of the classical time dilation for- mula that takes into account the possibility of the clock q −1 Hˆ = pˆ2c2 + Mˆ 2c4, (1) moving in nonclassical states of motion and nonzero γQ leads to quantum time dilation effects. where pˆ is the particle’s momentum and Mˆ ≡ m + The above considerations were based on an ideal clock 2 model in which the of the clock was asso- Hˆclock/c is the so called mass , which is a com- bination of the particle’s rest mass m and the dynamical ciated with an operator that was canonically conjugate 2 to the clock Hamiltonian. It is thus not clear whether mass Hˆclock/c stemming from the energy of the internal quantum time dilation is universal, affecting all clocks in degree of freedom governed by the Hamiltonian Hˆclock. This internal degree of freedom can serve as a clock that the same way analogous to its classical counter part, and tracks the particles proper time as measured by a time the answer must ultimately come from experiment. In what follows we present evidence that supports the Tclock that transforms covariantly with respect 1 conjecture that quantum time dilation between clocks to the group generated by Hˆclock; such covariant observ- ables are common in parameter estimation tasks and in moving in superpositions of inertial trajectories is univer- this case gives the best estimate of the proper time ex- sal. We consider an entirely different clock model based perienced by the particle. on the lifetime of an excited atom and demonstrate that Consider the particle to be prepared in a superposition such a clock observes quantum time dilation in accor- of momentum wave packets, which up to normalization dance with Eq. (5). This brings quantum time dilation is taken to be effects closer to experiment by demonstrating that they manifest in a realistic clock model based on spontaneous iφ |ψi ∼ cos θ |ϕp¯1 i + e sin θ |ϕp¯2 i , (2) emission, the mechanism by which atomic clocks operate. π where θ ∈ [0, 2 ), φ ∈ [0, π), and hp|ϕp¯i i = 2 2 √ e−(p−p¯i) /2∆ /π1/4 ∆ with ∆ being the spread of the III. SPECTROSCOPY OF MOVING ATOMS wave packet in momentum . Let the clock be char- acterized by the Hilbert space L2(R), with the Hamil- Consider a two-level atom of mass m and suppose that tonian equal to the momentum operator, Hˆclock = its |gi and |ei are separated ˆ cPclock, and the covariant time observable satisfies by an energy difference }Ω in the atom’s rest frame. The ˆ ˆ ˆ [Tclock, Hclock] = i}. The average time read by the clock of the atom and electromagnetic fields, E and ˆ hTˆclocki when the clock of an observer relative to which B, are described by the Hamiltonian the Hamiltonian in Eq. (1) generates an reads Hˆ = Hˆ + Hˆ + Hˆ , (6) the time t can be shown to be equal to [36] atom field af where the free Hamiltonian of the atom to leading rel- ˆ  −1 −1 hTclocki = γC + γQ t, (3) ativistic order in the atom’s center-of-mass momentum pˆ/mc (e.g. [47]) is where to leading relativistic order pˆ2 1 pˆ4  1 pˆ2  Hˆ = − + Ω 1 − |eihe| , (7) p¯2 cos2 θ + p¯2 sin2 θ − ∆2/2 atom 3 2 } 2 2 γ−1 ≡ 1 − 1 2 , (4) 2m 8 m c 2 m c C 2m2c2 and the electromagnetic field Hamiltonian is Hˆfield = P † k,ξ }ωkaˆk,ξaˆk,ξ, which is a mode sum over the wave vector k and polarization index ξ with the corresponding 1 In more detail, the time observable Tclock is defined as eigenfrequencies ω = kc and annihilation operators aˆ . a positive operator valued measure with effect operators k k,ξ Eˆ(t) that satisfy the covariance condition Eˆ(t + t0) = The atom is coupled to the electromagnetic field through 0 ˆ 0 ˆ e−it Hclock Eˆ(t)eit Hclock [44–46]. We make the further assump- the interaction Hamiltonian [48–52] tion that this time observable is sharp and can thus be associated 1 h     i with a self-adjoint operator Tˆ . Hˆ = −dˆ· Eˆ⊥ − pˆ · Bˆ × dˆ + Bˆ × dˆ · pˆ , (8) clock af 2m 3 where the first is the usual dipole interaction, the atom is much smaller than both its rest energy and dˆ = d (|gihe| + |eihg|) the dipole operator in the lab , which ensures that a first order expansion frame, and the second term is the so-called Röntgen term in both }Ω/mc2 and }Ω/hpˆ2/2mi is valid. that accounts for the Lorentz-transformed electromag- Suppose an atom begins in its excited state with netic field felt by the moving atom.2 It is important to center-of-mass wave function ψ(p) and the electromag- R note that all the operators entering the Hamiltonian (6) netic field in the vacuum, |Ψ(0)i = dp ψ(p) |e, p, 0i. At are expressed in the laboratory frame.3 The electromag- a later time t, the composite evolves to the state netic fields appearing in (8) are given by Z |Ψ(t)i = dp α (p, t) |e, p, 0i r ˆ⊥ X }ωk ik·rˆ E (r) = −i k,ξaˆk,ξe + H.c., (9) X Z 20V k,ξ + dp βk,ξ (p, t) |g, p − }k, 1k,ξi , (13) r k,ξ X } ik·rˆ Bˆ(r) = i (k × k,ξ)a ˆk,ξe + H.c., 20V ωk which has been expanded in the energy eigenstates k,ξ |e, p, 0i and |g, p − k, 1 i, associated respectively with (10) } k,ξ the energies where 0 is the and V the quanti- p2 p4  1 p2  zation volume, while  is the polarization vector per- ωe(p) = − + Ω 1 − , (14) k,ξ } 2m 8m3c2 } 2 m2c2 pendicular to the wave vector k. By invoking the rotat- 2 4 ing wave approximation [49, 51], the interaction Hamil- (p − k) (p − k) ω (p, k) = } − } + ω . (15) tonian (8) assumes the form } g 2m 8m3c2 } k r The time-dependent coefficients in |Ψ(t)i can be ob- ˆ X }ωk ik·rˆ Haf = −i} gˆk,ξe |eihg| aˆk,ξ + H.c., (11) tained by solving the associated Schrödinger equation 20V k,ξ via a as commonly utilized in Wigner- Weisskopf theory [62]. Using a single pole approxima- where the coupling ‘constant’ depends on the atom’s mo- tion [40] one finds mentum −iω (p)t − Γ(p) t   α (p, t) = e e e 2 ψ (p) , (16) 1 }k gˆk,ξ = k,ξ · d + pˆ − · [(k × k,ξ) × d] , (12) r mωk 2 }ωk βk,ξ (p, t) = gk,ξ(p)ψ (p) 20V and is an operator itself with eigenvalues gk,ξ. Note that −iω (p)t − Γ(p) t −iω (p,k)t ˆ e e e 2 − e g Haf has an explicit dependence on the center-of-mass po- × , (17) ˆ Γ(p) sition operator r, which is treated as a quantum degree i [ωe(p) − ωg(p, k)] + 2 of freedom. This Hamiltonian couples the internal en- ergy levels of the atom to the center-of-mass degree of where Γ(p) is the total transition rate of the spontaneous freedom, and as a consequence causes a recoil of the de- decay of the atom moving with momentum p: caying atom. The second term of Eq. (12) is a direct X ωk 2 result of the Röntgen term in Eq. (8). Γ(p) = 2 3 gk,ξ(p)δ [ωe(p) − ωg(p, k)] . (18) 8π ~0c Let us take a to consider the energy scales k,ξ characterizing the situation just described, namely the Then, the total transition rate in the long-time limit is energy of the atom’s internal degree of freedom }Ω, rest 2 2 Z Z energy mc , and kinetic energy hpˆ /2mi. In what follows d X 2 2 Γ= lim dp |βk,ξ (p, t)| = dp |ψ (p)| Γ(p), we will consider regimes in which the internal energy of t→∞ dt k,ξ (19) where the results of [55, 56] are recovered when ψ (p) is 2 This term has been known since the 19th [53], however a momentum eigenstate. As detailed in the AppendixB, it was omitted in the early works on light-matter interactions. the angular distribution of the emitted radiation is ob- One of its first incorporations dates back to works of Babiker tained by omitting the angular integration in Eq. (19), in the 1980s [54], yet it wasn’t until the early 2000s that this yielding term was rigorously shown to be necessary to retrieve agreement   with special relativity [49, 55, 56]. Nowadays, it is routinely Γ(Θ, Φ) 3 ~Ω utilized in studies of atom-field interactions involving moving = Ξ0(Θ, Φ) 1 − 2 bodies [51, 57–60]. The Röntgen term is a consequence of the Γ0 2 mc vectorial nature of the electromagnetic field, and thus would not 1 Z + Ξ (Θ, Φ) dp p|ψ(p)|2 appear in an analogous scalar field model [43, 61]. mc 1 3 For example, the dipole moment is connected to its rest value d0 0 0 Z by a d = d0 − d ·v v+ d ·v v/p1 − v2/c2, 1 2 2 v2 v2 + Ξ (Θ, Φ) dp p |ψ(p)| , (20) where v is the of the moving atom. 2m2c2 2 4

FIG. 1. a) Angular dipole distribution of emitted photons from a decaying atom at rest with respect to the center of mass momentum p and dipole momentum d. Motional corrections to this angular distribution b) linear and c) quadratic in the atom’s center of mass momentum. Magnitudes are represented by the distance to the origin and color (red positive, blue negative)

2 where Θ and Φ are the azimuthal and polar angles of k 3~Ω p 2 2 3~Ω p¯ Γ = Γ0(1 − 2mc2 ) 1 − v /c ≈ Γ0(1 − 2mc2 − 2m2c2 ), Ω3d02 vector relative to p, respectively, Γ0 = 3 is the to- which agrees with Eq. (24) and ensures consistency with 3π0 c tal decay rate of a standing atom ignoring} recoil effects special relativity. (i.e., ~Ω  mc2), and we have assumed that the atom Additionally, from Eqs. (16) one can extract the shape moves only along the z axis, perpendicular to the dipole of an emission line, which can be straightforwardly trans- moment vector d. As such, ψ(p) has to be understood formed to the absorption spectrum through the Einstein as a marginal distribution of a full center-of-mass wave coefficients. The probability that an atom emits a 2 R 2 function, |ψ(p)| = dpxdpy |ψ(p)| , where momentum with momentum }k is distributions in the x and y directions are well localized Z X 2 around the z axis. Ξ0(Θ, Φ) is the standard angular dis- P (k) = lim dp |βk,ξ (p, t)| . (25) t→∞ tribution of dipole radiation ξ

3 2 2  Again, assuming that the atom has a large mass m, moves Ξ0(Θ, Φ) = 1 − sin Θ cos Φ , (21) 8π along the z axis and dipole moment points perpendicu- larly to motion, we arrive at the following characteristic while of a transition line for photons emitted along the direc- 3 tion of motion (see AppendixB): Ξ (Θ, Φ) = cos Θ 1 − 2 sin2 Θ cos2 Φ , (22) 1 4π 3 3 Ξ (Θ, Φ) = 6 cos 2Θ + 5 cos2 Φ (cos 4Θ − cos 2Θ) , Pk(ω) = × 2 16π 8π (23) Z 1 + 3 p  Γ /2π dp |ψ (p)|2 mc 0 , (26) 2 Γ2 ω − Ω 1 + p  + 0 1 + 2 p  are first and second order corrections in p/mc to the mc 4 mc dipole distribution appearing due to the motion of the and perpendicular to both the dipole moment and to the atom [49]. These motional corrections to the angular direction of motion: distribution of radiation are universal as they manifest unless the atom is at rest and their shape is independent 3 P (ω) = × of the momentum wave function ψ(p) (see Fig.1). ⊥ 8π Integration over Θ and Φ recovers the familiar for-  3 p2  Z 1 − 2 m2c2 Γ0/2π mula [49, 51, 55, 56] dp |ψ (p)|2 . h  2 i2 2  2  1 p Γ0 p  3 Ω 1 Z  ω − Ω 1 − 2 m2c2 + 4 1 − m2c2 Γ = Γ 1 − ~ − dp p2|ψ(p)|2 . (24) 0 2mc2 2m2c2 (27)

If the atom were to move along a classical trajectory with Note that both Pk(ω) and P⊥(ω) have been expanded momentum p¯, corresponding |ψ(p)|2 = δ(p−p¯), the tran- up to the leading relativistic order and are proportional sition transition rate Γ is related to transition rate in to the center-of-mass momentum distribution |ψ(p)|2 in- 3~Ω  the atom’s rest frame Γ0 1 − 2mc2 via a , tegrated against a Lorenz distribution. When observed 5

FIG. 2. The difference in total emission rate between a superposition and a classical mixture of two momentum wave packets of an atom as a function of the wave packets’ momentum difference and their relative phase and weight: a) equal weighted superposition of momentum wave packets, θ = π/4, b) Relative phase fixed at φ = 0, c) Relative phase fixed at φ = π. The red line marks the maximum value of the effect for a given relative phase or a relative weight, while the red circles signify maximum and minimum values across the whole plot. Nonzero value for a finite momentum difference signifies the of quantum time dilation. In each of the panels, the momentum spread of each of wave packets is ∆ = 0.01mc and the sum of their average momenta is equal to p¯1 +p ¯2 = 0.05mc. in the direction of motion, the transition line is Doppler of momentum wave packets. For simplicity, we consider shifted, as the Lorentz distribution is shifted linearly in p¯1 and p¯2 to be collinear. By evaluating (19) we arrive momentum by an amount Ω → Ω(1 + p/mc). On the at the following relative difference of total emission rates other hand, light emitted/absorbed perpendicular to the between these two cases: motion is not affected by this Doppler shift and relativis- Γ − Γ 1 Z   tic corrections are dominant, shifting the center of the sup cl = dp p2 |ψ (p)|2 − P (p) Γ 2m2c2 sup cl Lorentz distribution by an amount Ω → Ω(1−p2/2m2c2), 0 −1 which is quadratically in momentum. = γQ , (31) Each of the quantities of interest — the angular distri- which is equal to the quantum correction to the classical bution of radiation, the total decay rate, and the shape of time dilation contribution given in Eq. (5) and derived the emission line — are routinely measured in various ex- in [36]. This is a surprising result because the clock model periments [63]. As we have shown how these considered here, based on the spontaneous decay of an depend on the center-of-mass momentum distribution, atom, observes the same quantum time dilation effect we are now equipped to show how nonclassical center- as the quantum clock considered in [36] and described of-mass motion in such experimental scenarios can be in Sec.II. This observation supports the conjecture that utilized as a direct probe of quantum time dilation. quantum time dilation for constant is universal, affecting all clocks in the same manner. −1 IV. SPECTROSCOPIC SIGNATURES OF The difference γQ in transition rate between a co- QUANTUM TIME DILATION herent superposition and classical mixture of momentum wave packets can be of positive or negative sign, depend- ing on the relative phase between two wave packets φ and First, we will compare the transition rate Γ between their relative weight θ (see Fig.2). For instance, for an atoms in coherent superpositions and incoherent classi- equally weighted superposition it is seen that Eq. (31) cal mixtures of localized momentum wave packets. Anal- does not depend on the sum of the wave packets’ mo- ogously to the quantum clock model described in Sec.II, menta; it is positive for a relative phase smaller than an atom is considered to be either in a superposition (2) φ = π/2 and becomes negative for a larger value. In with a momentum distribution given by this case, the structure of the quantum contribution ex- iφ  ψsup(p) = N cos θ hp|ϕp¯1 i + e sin θ hp|ϕp¯2 i , (28) hibits a distinctive peak for a given relative phase φ. If φ = 0, this peak occurs at a finite momentum difference, where p¯2 − p¯1 ≈ 2∆; however, if the relative phase is φ = π, the   2 −1/2 √ − (p¯1−p¯2) position of the peak shifts towards p¯2 − p¯1 ≈ 0. N = π∆ 1 + cos φ sin 2θ e 4∆2 , (29) This behavior can be understood by analyzing the structure of the wave packets in momentum space — or in a classical mixture such that when the wave packets almost fully overlap, their relative 2 2 2 2 Pcl(p) = cos θ |hp|ϕp¯1 i| + sin θ |hp|ϕp¯2 i| (30) phase plays a crucial role. If the separation in momentum 6 space between the wave packets vanishes, then there is no quantum correction is proportional to δQ, it will van- distinction between the coherent superposition and inco- ish if the atom’s center-of-mass is prepared in an equally herent classical mixture, as two wave packets are iden- weighted superposition of momentum wave packets. On tical. As the phase approaches π, the real part of the the other hand, the amount of photons emitted perpen- center-of-mass wave function goes to 0, pronouncing the dicular to both the dipole moment and the direction of imaginary part which is an antisymmetric function. This motion is not affected linearly in atom’s momentum. Ad- is in stark contrast to the classical mixture for which the ditionally, as shown in Fig.1(c), the second order con- density stays single peaked. tribution is the largest in this direction, suggesting that Surprisingly, an equally weighted superposition is not photon detection in this direction is optimal for measur- optimal for maximizing the effect of quantum time dila- ing quantum time dilation. tion, as it saturates at −∆2/2m2c2 for a relative phase Similar to the total emission rate and the angular dis- equal to φ = π. The global maximum is also achieved for tribution, the shape of a transition line is also affected by φ = π, however for a√ slightly unbalanced superposition, the nonclassicality of the center-of-mass state as shown θ ≈ π/4±(¯p2 −p¯1)/2 2∆. If one considers wave packets’ in Eqs. (26) and (27). That is, the of with average momenta much larger than their spreads, an atom in a coherent momentum superposition is dis- (¯p1 +p ¯2)/∆  1, then this maximum√ becomes propor- tinct from that of an incoherent classical mixture. As 2 2 tional to the sum of the momenta, ± 2∆|p¯1 +p ¯2|/4m c . suggested by analysis of the angular distribution of ra- This indicates that the effect of quantum coherence on diation, we will focus on two cases: photons emitted the emission rate increases as the average momenta of parallel and perpendicular to the atom’s motion. Ex- the wave packets increases. perimentally, both scenarios can be realized by emission −1 and , with the latter producing Note that the quantum correction γQ to the time di- lation observed by the atom’s decay rate is second or- an absorption line shape that can be derived from the der in the atom’s average momentum, see Eqs. (5) and emission shape via the Einstein coefficients. To keep the (31), analogous to the a classical time dilation contribu- discussion simpler, we will discuss only the emission line. −1 First, photons emitted in the direction of motion are tion governed by γC . However, linear effects, such as a Doppler shift, can also be affected by momentum coher- affected by the classical Doppler effect, shifting the center ence. Such effects can be characterized by the difference of the transition line linearly in p/mc, contrary to rela- in the first moments of the momentum distributions as- tivistic effects, that cause shifts quadratic in p/mc. Anal- sociated with a coherent superposition and incoherent ogously to quantum time dilation, the correction coming classical mixture: from momentum coherence to the Doppler shift can be dubbed a quantum Doppler shift. It is important to note Z 1  2  that this effect only modifies the shape of the emission δQ ≡ dp p |ψsup(p)| − Pcl(p) mc spectrum, not the total emission rate, which is affected cos φ sin 4θ (¯p − p¯ ) by quantum time dilation. The quantum Doppler effect = 2 1 . (32) 2  (p ¯2−p¯1)  smooths the contrast between two transition rate peaks 4mc cos φ sin 2θ + e 4∆2 associated with two different Doppler shifted emission lines; see Fig.3(a)-(b). The difference between the quan- tum and classical Doppler shift is most pronounced in The behavior of δ is qualitatively different than γ−1. Q Q between the emission peaks, which may suggest that the Most notably, δ vanishes if the superposition of sym- Q postselection of the final momentum of the atom may fur- metric wave packets is equally weighted. On the other ther enhance the effect. A quantitative analysis describ- hand, as it is linear in momentum, it is easier to mea- ing how the quantum Doppler effect modifies emission sure as the absolute magnitude of this effect is necessar- line shape is provided in the AppendicesA andB. ily larger than the second order quantum time dilation In the case of an emission perpendicular to the direc- effect characterized by γ−1. The detailed analysis of δ Q Q tion of motion, the classical Doppler shift is not present; is provided in the AppendixA. that is, corrections linear in p/mc are not present. The If one considers an angular distribution of emitted pho- center of the transition line is shifted quadratically in tons from the decaying atom, (20), the difference between p/mc, heralding the onset of relativistic effects. Rela- coherent and incoherent cases is given by tivistic corrections of this magnitude can be measured Γ (Θ, Φ)− Γ (Θ, Φ) in state-of-the-art experiments [38] and are affected by sup cl = Ξ (Θ, Φ) δ + Ξ (Θ, Φ) γ−1. a momentum coherence analogously to the quantum Γ 1 Q 2 Q 0 Doppler shift; see Fig.3(c)-(d). (33)

Note that the angular distribution of the radiation can be affected linearly in momentum and analysis of Fig.1(b) V. EXPERIMENTAL CONSIDERATIONS shows that the contribution stemming from the momen- tum coherence is the most pronounced for photons emit- Since the initial prediction of wave-particle duality by ted in the direction of atom’s motion. However, as this de Broglie, atomic interferometry has witnessed tremen- 7

5 dous conceptual and technological progress [66]. With 4 the advent of large momentum beamsplitters [67–70], su- 3 perpositions of atomic beams travelling along distinct 2 trajectories have been realized, leading to quantum-based 1 alternatives to classical gravimeters, gradiometers, and 0 accelerometers [34, 71–75]. In these settings, the usual 20 40 60 80 strategy is to suppress radiation losses as they disrupt 5 phase relations between arms of an interferometer [70]. 4 In contrast, our proposal is to test the nonclassical- 3 ity of center-of-mass motion through spectroscopic mea- 2 surements. As shown above, a coherent superposition 1 of relativistic momenta affects the spontaneous emission 0 rate beyond classical time dilation effects, thus offering a 20 40 60 80 spectroscopic signature of quantum time dilation. More- 5 over, spectroscopic methods offer a plethora of schemes 4 that might witness similar quantum-relativistic effects — 3 among others, stimulated absorption and emission spec- 2 troscopy, and techniques involving Mössbauer effect and 1 Rydberg states [66]. In particular atomic clocks provide 0 -150 -100 -50 0 a natural test bed for a relativistic theory due to their unparalleled accuracy [64, 76, 77]. Such clocks have been 5 4 used to observe classical time dilation at velocities as low 3 as several meters per second [38]. 2 To go beyond and measure quantum time dilation, one 1 has to deal with experimental challenges that involve an 0 interplay between different time scales. The lifetime of -150 -100 -50 0 the excited atom has to be long enough to allow for the creation of sufficient momentum separation and to pre- cisely excite the atomic beam. However, the lifetime can- FIG. 3. Emission line shape P(ω) of the spontaneous decay of not be longer than the coherence time of the center-of- an atom that is initially prepared in a coherent superposition mass superposition. Fortunately, due to advanced meth- (Psup) and in a classical mixture (Pcl) of two momentum wave ods of phase imprinting in atomic [78, 79], ini- −8 packets sharply peaked at different momenta, p¯1 = 2·10 mc tial states maximizing the quantum contribution might −8 and p¯2 = 4 · 10 mc (velocities achievable for ion clocks [64] be engineered. or momentum cat states [65]). The emission line is measured Specifically, there are promising experimental se- parallel, P (ω) or perpendicular, P (ω) to the motion of the k ⊥ tups offering access to the accuracy needed to ob- wave packets and is normalized to the maximum probabil- serve quantum-relativistic effects. Among others, quan- ity for a single stationary wave packet in a given case, Pmax. In the former case, the dominant shift of the transition peak tum clocks based on have recently comes from the Doppler shift, while for the latter case—from achieved precision going beyond leading relativistic cor- the time dilation. Note that transverse emission in suppressed rections [38, 64, 77]. In such setups, an aluminium ion compared to parallel emission. Panels a) and b) are calculated is confined to a quadrupole trap acting with an effective 9 2 1 for a broad transition, Ω/Γ0 ≈ 1.5·10 (e.g. hydrogen P − S harmonic potential and is prepared close to zero-point transition), while panels c) and d) are associated with the ex- motion energy by advanced cooling techniques. The ion 17 1 3 tremely narrow Ω/Γ0 ≈ 1.5 · 10 (e.g. aluminium S0 − P 0 is perturbed triggering oscillatory motion in a coherent- transition). It showcases the fact that quantum relativistic ef- state-like fashion. Spectroscopic allows for fects can be probed even for broad transitions, if the Doppler resolving the resulting frequency shift due to the ion’s shift is affected. If the spread of the momentum wave packets motion below 10−18Ω, which is far below the leading rel- is much smaller than their separation, ∆  |p¯2 − p¯1|, coher- −15 ent and incoherent cases are almost indistinguishable, with ativistic correction of 10 Ω [64]. two sharp shifted peaks clearly visible (panels (a) and (c), To observe quantum time dilation. a coherent momen- ∆/mc = 6 · 10−9). Note the broadening of structures due to a tum superposition of such an ion must be prepared, a mo- finite spread of momentum (i.e., a homogeneous Doppler ef- mentum Schrödinger cat state. This on its own is a state- fect). If the momentum spread becomes larger and the over- of-the-art task, however recent advances have reported lap of the two wave packets increases (panels (b) and (d), tremendous progress in this direction [65, 71, 80, 81]. −9 ∆/mc = 8 · 10 ), interference effects become visible, man- For example, ytterbium ions have been prepared in meso- ifesting direct confirmation of quantum relativistic effects in scopic superpositions of motional states [65]. the atomic spectrum. Creation of such an ion that exhibits a narrow transi- tion line is required for observation of the quantum time dilation effect in order to resolve the associated frequency 8

∆ p2−p1 shift which is second order in the average momenta of mc mc Γ0, it is still too far away to be measured in these the wave packets. Mean velocity of a trapped ion easily experiments. However, as the interest in large momen- resolvable for an ion clock in a laboratory, 5 m/s, cor- tum transfer grows with a potential use in low energy responds to a coherent state |αi with |α| ≈ 12. State- studies of quantum gravitational effects, the engineering of-the-art separation between coherent states can go up of a quantum time dilation experiment might be pos- to |α| ≈ 24 [65], showing that a coherent superposition sible in the near . Such an experiment could be of momenta can be achieved within spectroscopic resolu- achieved by either a larger momentum spread of a single tion. wave packet or a larger momentum separation between Generally speaking, in experiments a momentum wave packets. transition line is measured. The difference in transi- Similar to the superposition of momentum wave pack- tion line shape exhibited by a coherent superposition and ets considered above, quantum effects manifest for atoms incoherent classical mixture of momentum wave pack- in spatial superpositions [40–43, 86]. From an experi- ets, as shown in Figs.3(c)-(d), would be confirmation mental point of view, such studies have been helpful in of quantum time dilation. The difference between these analyzing phase coherence in Bose-Einstein condensates two cases is most pronounced for a frequency that corre- interacting with light [86]. These systems provide an ex- sponds to the average of mean momenta of the superim- tremely clean environment to study atomic systems with posed wave packets. The upshift of the transition proba- possibility of fine tuning of interactions and spatial ge- bility due to momentum coherence at this specific point ometry. As such, they might accommodate experiments be as much as 40% if the parameters of the superposition with coherent superposition of momentum wave packets are optimized, while also not being far away from reso- (e.g. non-equilibrium dynamics in double-well trap [87]). nance as depicted in Fig.3(d). Changes of this magnitude Moreover, in contrast to ion clocks, experiments involv- are routinely measured in state-of-the-art experiment in- ing large momentum transfer or trapped ultracold gases volving ion clocks [38, 64]. have still not achieved the necessary velocities to be sen- In such an experimental setup the balance between the sitive to relativistic motion of particles. However, such ability to create a superposition of momentum wave pack- experiments have the advantage that they utilize large ets and the precision of a given ion clock is crucial in order ensembles of atoms, which results in a stronger signal to observe a signature of quantum time dilation. Other that should scale proportionally to the particle number. obstacles exist, such as excess motion, secular motion, the quadratic Zeeman effect, deviations from harmonic trap- ping etc. [77], which are usually well resolved in atomic VI. CONCLUSIONS AND OUTLOOK clock experiments. Nonetheless, additional work needs to be done to characterize these effects in presence of We have proposed a spectroscopic signature of quan- relativistic momentum coherence to deduce the optimal tum time dilation that manifests in the spontaneous experimental scenario. emission rate (lifetime) of an excited atom moving in On the other hand, experiments involving large mo- a superposition of relativistic momentum wave packets. mentum transfer between between light and atomic We have shown that the total transition rate is strongly beams might also provide a viable alternative for a mea- affected by momentum coherence in the center-of-mass surement of quantum time dilation [68, 70, 82–84], where state of the atom. Furthermore, the quantum contribu- limitations due to excited state decay would work as tion to the time dilation observed by an excited atom can an advantage. As they are not yet realized for narrow be either positive or negative, depending on the relative transitions at the level of atomic clocks, such an up- phase between the superposed momentum states and is grade is widely sought as a key ingredient for gravita- within reach of the existing experimental setups [38, 67– tional wave [83] and dark matter detectors [85]. Advances 70]. We observed that the quantum contribution to the in this direction could make possible a measurement of time dilation of the atomic lifetime in Eq. (31) was equal quantum time dilation in such a setup, as narrower tran- to the quantum time dilation observed on average by the sitions would enhance the spectroscopic precision. ideal clock considered in [36, 88]. These two clocks are The momentum separation currently achieved in these built on very different mechanisms (spontaneous decay experiments is around 140 }k for the strontium transition and a particle on a line), so this result indicates that 1 3 S0- P1, where k is the magnitude of the wave vector of quantum time dilation is universal, affecting all clocks in the incident light [70]. Such a momentum separation the same way. corresponds to a 1 m/s velocity difference between two The effects of quantum time dilation on atomic spectra clouds of atoms. Larger momentum transfer is expected complement the growing literature on relativistic clock in the future, with experimental proposals promising up interferometry, which also probes the effects of suppo- to 1000 }k [70]. The widths of these momentum wave sitions of clocks experiencing different proper times due packets are relatively small for the detection of quan- to both special or general relativistic effects [16, 18, 31– tum time dilation, with a rms Doppler width of 25 kHz, 34, 75]. In contrast, we propose a spectroscopic signature corresponding to a velocity width of 0.02 m/s [70]. As of proper time superpositions that can probe quantum the maximal value of quantum time dilation scales like theory and relativity in the regime in which coherence 9 across relativistic momentum wave packets plays a role. In case of Gaussian wave packets considered above, K1 We have also characterized a quantum Doppler effect and K2 take the explicit form: that occurs when the center-of-mass of an atom is cos φ sin 4θ (¯p − p¯ ) K = 2 1 = δ , in a superposition of momentum wave packets. The 1 2 Q  (p ¯2−p¯1)  effect is present in the shape of the emission spectrum, 4mc cos φ sin 2θ + e 4∆2 affecting its structure by smoothing the contrast between h i distinctive Doppler-shifted peaks. 2 2 2 cos φ sin 2θ (¯p2 − p¯1) − 2 p¯2 − p¯1 cos 2θ K = = γ−1. 2 2 Q  (p ¯2−p¯1)  Note added.—Following the initial posting of this ar- 8m2c2 cos φ sin 2θ + e 4∆2 ticle a related preprint on delocalized center-of-mass atomic wave functions and the light-matter interaction (A4) appeared [61]. Analysis of K2 is presented in Fig.4 and of K1 in Fig.5. Note that K1 vanishes for an equally weighted super- position θ = π/4. Let us show that this feature is a common feature of all symmetric wave packets. Let ϕ(p) ACKNOWLEDGMENTS be a normalized wave packet symmetric with respect to p = 0. Then, we can write an equally weighted, coherent We would like to thank Kazimierz Rzążewski for superposition of two momentum wave packets as fruitful discussions and pointing out [40], and Mehdi N  iφ  Ahmadi for his valuable contributions to the ideas ϕsup(p) = √ ϕ(p − p1) + e ϕ(p − p2) , (A5) developed here. This work was supported by the 2 Natural Sciences and Engineering Research Council 2  R −1 with N = 1 + cos φ dp ϕ(p − p1)ϕ(p − p2) . For of Canada and the Dartmouth Society of Fellows. the corresponding classical mixture, the momentum dis- This article has been supported by the Polish Na- tribution takes the form tional Agency for Academic Exchange under Grant No. 1 PPI/PZA/2019/1/00094/U/00001. P = ϕ2(p − p ) + ϕ2(p − p ) . (A6) cl 2 1 2 Then, by an explicit evaluation, one finds that Z 2mc 2 h 2 i Appendix A: Momentum wave packets and K1 = dp p |ϕsup (p)| − Pcl (p) signatures of coherence N 2 N 2 h Z  = − cos φ dp ϕ(p − p1)ϕ(p − p2) Here, explicit forms of momentum wave packets from Z  the main text are presented. It is assumed that the atom 2 2  in consideration moves along the z-direction with its mo- × dp p ϕ (p − p1) + ϕ (p − p2) mentum distribution in perpendicular directions well lo- Z i calized around px = py = 0. Moreover, we consider − 2 dp p ϕ(p − p1)ϕ(p − p2) . (A7) the atom to be either in a coherent superposition of two Gaussian wave packets: The term in the third row of Eq. (A7) equals p1 + p2, be- cause ϕ(p−p1,2) are normalized and well localized around  (p−p¯ )2 (p−p¯ )2  − 1 iφ − 2 p1,2. By substituting p → p−(p1 +p2)/2 and utilizing the ψsup(p) = N cos θe 2∆2 + e sin θe 2∆2 , (A1) fact that expression ϕ(p−p0)ϕ(p+p0) is an even function of p, one finds that √ 2 2 Z where N = [ π∆(1 + cos φ sin 2θe−(¯p1−p¯2) /4∆ )]−1/2, or 2 dp p ϕ(p − p1)ϕ(p − p2) in an incoherent mixture described by the momentum distribution: Z = (p1 + p2) dp ϕ(p − p1)ϕ(p − p2). (A8)

 (p−p¯ )2 (p−p¯ )2  1 2 − 1 2 − 2 P = √ cos θe ∆2 + sin θe ∆2 . (A2) Substituting these two results into Eq. (A7) it is seen cl π∆ that that K1 = 0. The only necessary condition is for ϕ(p) to be an even function with respect to p = 0. The difference between coherent superpositions and clas- sical mixtures of momentum wave packets can be char- acterized by the difference in moments associated with Appendix B: Derivation of the emission rate and their respective momenta distributions: spectrum shape Z 1 j h 2 i Let us focus on the angular distribution of the emitted Kj ≡ dp p |ψsup (p)| − Pcl (p) . (A3) j!mjcj radiation. The angular distribution can be obtained by 10

FIG. 4. Fig.2 from the main text; shown in Appendix for a purpose of comparison with Fig.5. The difference in total emission rates between a superposition and a classical mixture of two momentum wave packets of an atom as a function of the wave packets’ momentum difference and their relative phase and weight. The red line marks the maximum value of the effect for a given relative phase or relative weight, while the red circles signify maximum and minimum values across the given subplot. a) equal weighted superposition of momentum wave packets, θ = π/4. b) Relative phase fixed at φ = 0. The red line p is not continuous at the point θ = π/4, and it sharply ends at (¯p2 −p¯1)/∆ = 2 1 + W0(1/e) ≈ 2.261, where W0 is the principal branch of the Lambert W function. c) Relative phase fixed at φ = π. It can be seen that two extrema exist for small values of π 1 p¯2−p¯1 p 2 2 (¯p1 − p¯2)/∆ and θ ≈ π/4. One can show that these extrema are placed at θ = − (1 ± 1 + 2(¯p2 +p ¯1) /∆ ) and their 4 4 p¯2+¯p1 2 p 2 2 2 2 corresponding values are ±∆ ( 1 + 2(¯p1 +p ¯2) /∆ − 1)/4m c . Nonzero value for a finite momentum difference signifies the phenomenon of quantum time dilation. In each of the panels, the momentum spread of each of wave packets is ∆ = 0.01mc and the sum of their average momenta is equal to p¯1 +p ¯2 = 0.05mc.

FIG. 5. The difference in first moments K1 = δQ of the momentum distributions associated with a superposition and a classical mixture of two momentum wave packets as a function of the wave packets’ momentum difference and their relative phase and weight, which quantifies the difference in magnitude of angular distribution of emission: a) unequal weighted superposition of momentum wave packets, θ = π/8, b) Relative phase fixed at φ = 0, c) Relative phase fixed at φ = π. The red line marks the maximum value of the effect for a given relative phase or a relative weight, while the red circles signify maximum and minimum values across the whole plot. Nonzero values for a finite momentum difference signifies the phenomenon of quantum time dilation. In each of the panels, the momentum spread of each of the wave packets is ∆ = 0.01mc and the sum of their average momenta is equal to p¯1 +p ¯2 = 0.05mc. omitting angular integration in Eq. (19). Note that such Ref. [51] for a derivation) an approach does not explicitly utilize the photon distri- Z bution that comes from an integration over the proba- d X 2 Γ= lim dp |βk,ξ (p, t)| bilities |β |2. However, as shown in Ref. [49] this ap- t→∞ dt k,ξ k,ξ proach is consistent with the special relativity, reproduc- Z ω ing dipole pattern of radiation in the comoving frame. 2 X k 2 = 2π dp |ψ (p)| 3 gk,ξ(p) 2 0(2πc) Writing explicitly Eq. (19) gives (see Eqs. (9)-(11) in k,ξ ~  1  × δ Ω − ω + k · (p − k/2) . (B1) k m ~ 11

By taking a continuous limit of the summation over k, Note we are now working in the continuous limit and so the sum over ωk has been replaced with an integral X Z Z → dκ dω ω2 sin θ, (B2) over ω. k Under the assumption that the atom is heavy the cou- 2 pling constant gk,ξ(p) can be expanded to first order in where κ = kc = (sin Θ cos Φ, sin Θ sin Φ, cos Θ) and 2 2 2 2 ω ~Ω/mc and to second order in p /m c : R R π R 2π dκ = 0 dΘ 0 dΦ, we can omit the integral over di- rection κ to get the angular distribution: 2 2 gk,ξ(p) ≈ (d · k,ξ) Z Z ∞   π 2 X 3 2 }ωkκ + (d · k,ξ) p − · [(κ × k,ξ) × d] Γ(Θ, Φ) = 3 dp |ψ (p)| dω ω sin Θ 0(2πc) mc 2c ~ ξ 0 1  1  + (p · [(κ ×  ) × d])2 . (B4) × g2 (p) δ Ω − ω + k · (p − k/2) . m2c2 k,ξ k,ξ m ~ (B3) Making use of the vector equalities:

X 2 2 2 (d · k,ξ) = d − (d · κ) , (B5) ξ X 2 (d · k,ξ) A · [(κ × k,ξ) × d] = (d · κ)(A · d) − d (A · κ) , (B6) ξ X 2 2 2 2 2 (A · [(κ × k,ξ) × d]) = d (A · κ) + A (d · κ) − 2 (A · κ)(d · κ)(A · d) , (B7) ξ

2 } with A = mc p − mc k, it follows that where

"     3 2 2  }ω X 2 2h 2 }ωk 2 η(ω) ≡ ω sin Θ 1 − sin Θ cos Φ 1 + g (p) ≈ d κ 1 + − p · κ + 2 k,ξ ⊥ mc2 mc ⊥ mc ξ 2 # 1 2 1 i 2p p 2 2 2  (p · κ)2 − (p · κ) κ p + p2κ2 , (B8) − cos Θ + cos Θ sin Φ + cos Φ , m2c2 m2c2 k k m2c2 k mc m2c2 p ω2 λ(ω) ≡ Ω − ω + ω cos Θ − ~ , where ⊥ and k indicate projections perpendicular and mc 2mc2  2  parallel to the vector d. p p ~Ω ω0 ≡ Ω 1 + cos Θ + 2 2 cos 2Θ − 2 . We now go back to the angular distribution, in which mc 2m c 2mc we have to compute the following integral: (B11)

Z ∞   The denominator can be expanded up to first order in X 1 2 2 2 2 dω ω3 sin Θ g2 (p) δ Ω − ω + k · (p − k/2) . ~Ω/mc and up to the second order in p /m c yielding k,ξ m ~ ξ 0 (B9) 1 p p2 Ω Again, supposing that the atom moves in the z-direction 2 ~ 0 ≈ 1 + cos Θ + 2 2 cos Θ − 2 , (B12) with its momentum distribution in the perpendicular di- |λ (ω0)| mc m c mc rections given by delta functions centered at px = py = 0. Thus we can consider p = (0, 0, p). We will also suppose and finally that d = (d, 0, 0) points in a direction perpendicular to p. Then, Eq. (B9) simplifies to   η(ω0) 8π h 3 ~Ω 0 ≈ sin Θ Ξ0(Θ, Φ) 1 − 2 |λ (ω0)| 3 2 2mc Z ∞ 2 2 2 η(ω0) p p i d dω η(ω)δ (λ(ω)) = d 0 , (B10) + Ξ1(Θ, Φ) + Ξ2(Θ, Φ) , (B13) 0 |λ (ω0)| mc 2m2c2 12

FIG. 6. A comparison between shapes of the emission spectrum Pk(ω) associated with a coherent superposition (Psup) and an incoherent classical mixture (Pcl) of momentum wave packets a) The transition line for an atom initially prepared in a superposition of two momentum wave packets as a function of emitted photon’s frequency and difference between wave packets’ momenta. The two-peak structure stemming from two distinct Doppler shifts is clearly visible. b) Absolute difference of emission probabilities between a superposition and a classical mixture of momentum wave packets as a function of the frequency of the emitted photon and difference between wave packets’ momenta. The difference is most pronounced in regimes where wave packets overlap. c) Relative difference of emission probabilities between a superposition and a classical mixture of momentum wave packets as a function of the frequency of emitted photon and difference between the wave packets’ momenta. It can be seen that the quantum contribution is largest in between the two transition peaks. This suggests that a postselection of final measured states of center-of-mass motion may increase the general visibility of the quantum Doppler effect.

where If ψ(p) is a wave packet well localized at p0, then

3 2 2   2  Ξ0(Θ, Φ) ≡ 1 − sin Θ cos Φ , (B14) 3~Ω p0 8π Γ = Γ0 1 − 2 − 2 2 . (B20) 3 2mc 2m c Ξ (Θ, Φ) ≡ cos Θ 1 − 2 sin2 Θ cos2 Φ , (B15) 1 4π One can also immediately see that the quantum time 3 Ξ (Θ, Φ) ≡ 6 cos 2Θ + 5 cos2 Φ (cos 4Θ − cos 2Θ). dilation manifests in the total transition rate: 2 16π (B16) Γ − Γ sup cl = γ−1. (B21) Γ Q Substituting this expression into Eq. (B3) yields the an- 0 gular distribution The first term in Eq. (B17) corresponds to the distribu-   tion of dipole radiation for an atom at rest which when Γ(Θ, Φ) 3 ~Ω = Ξ (Θ, Φ) 1 − integrated over Θ and Φ gives the transition rate Γ . Γ 0 2 mc2 0 0 The second term is a correction linear in p that asso- Z 1 2 ciated with a Doppler shift. This term vanishes when + Ξ1(Θ, Φ) dp p|ψ(p)| mc integrated over Θ and Φ, ensuring consistency with spe- 1 Z cial relativity as the total transition rate is Γ = Γ /γ(p), + Ξ (Θ, Φ) dp p2|ψ(p)|2. (B17) 0 2m2c2 2 which when expanded in p is seen to have zero contribu- tion linear in momentum. The difference between coherent and incoherent cases is However, terms linear in momentum modify the an- then given by gular distribution of radiation, manifesting as a pattern distinctively different than that of the dipole radiation Γsup(Θ, Φ)− Γcl(Θ, Φ) −1 distribution. The magnitude of this quantum correction = Ξ1(Θ, Φ) δQ + Ξ2(Θ, Φ) γQ . Γ0 depends on K1 (i.e. δQ), which is an explicit function of (B18) the parameters characterizing the atomic wave functions and surprisingly vanishes if the atom moves in an equally Integrating Γ(Θ, Φ) over Θ and Φ yields the total tran- weighted superposition. sition rate Also of interest is how momentum coherence affects the  Z  shape of the atomic emission line, which can be probed 3~Ω 1 2 2 Γ = Γ0 1 − − dp p |ψ(p)| . (B19) through spectroscopic methods. For a plane wave char- 2mc2 2m2c2 acterized by a wave vector k, the probability of emission 13 is given by On the other hand, in the case of photons measured par- allel to the direction of motion, one obtains Z X 2 P (k) = lim dp |βk,ξ (p, t)| , (B22) t→∞ ξ which utilizing Eq. (B11) can be cast into form

Z 2 2 3Γ0 2 X gk,ξ(p)/d P (k) = dp |ψ (p)| . 16π2 λ2(ω) + Γ2(p)/4 ξ 3 Z P (ω) = dp |ψ (p)|2 (B23) k 8π 1 + 3 p  Γ /2π By expanding λ(ω), g2 (p)/d2 and Γ2(p) up to second mc 0 k,ξ × 2 2 . (B25)  p  Γ0 p  order in momentum, under the assumption that the emis- ω − Ω 1 + mc + 4 1 + 2 mc sion line is measured perpendicular to the direction of motion, one finds 3 Z P (ω) = dp |ψ (p)|2 ⊥ 8π  3 p2  1 − 2 m2c2 Γ0/2π × . h  2 i2 2  2  1 p Γ0 p In Fig.6 we compare the parallel emission spectrum ω − Ω 1 − 2 m2c2 + 4 1 − m2c2 Pk(ω) for coherent superposition and incoherent classi- (B24) cal mixtures of momentum wave packets.

[1] L. Viola and R. Onofrio, Testing the tanglement between Two Massive Particles is Sufficient through freely falling quantum objects, Phys. Rev. D 55, Evidence of Quantum Effects in , Phys. Rev. Lett. 455 (1997). 119, 240402 (2017). [2] A. Peres and D. R. Terno, and [12] R. Pierini, K. Turzyński, and A. Dragan, Can a charged relativity theory, Rev. Mod. Phys. 76, 93 (2004). decaying particle serve as an ideal clock in the presence [3] G. Chiribella, G. M. D’Ariano, P. Perinotti, and B. Val- of a magnetic field?, Phys. Rev. D 97, 045006 (2018). iron, Quantum computations without definite causal [13] M. Zych and Č. Brukner, Quantum formulation of structure, Phys. Rev. A 88, 022318 (2013). the Einstein equivalence principle, Nat. Phys. 14, 1027 [4] K. Lorek, J. Louko, and A. Dragan, Ideal clocks—a con- (2018). venient fiction, Class. Quantum Grav. 32, 175003 (2015). [14] P. A. Höhn and A. Vanrietvelde, How to switch be- [5] T. Kovachy, P. Asenbaum, C. Overstreet, C. A. Donnelly, tween relational quantum clocks, arXiv:1810.04153 [gr- S. M. Dickerson, A. Sugarbaker, J. M. Hogan, and M. A. qc] (2018). Kasevich, at the half- scale, [15] C. Anastopoulos and B. L. Hu, Equivalence principle Nature 528, 530 (2015). for quantum systems: Dephasing and phase shift of [6] I. Pikovski, M. Zych, F. Costa, and Č. Brukner, Univer- free-falling particles, Class. Quantum Grav. 35, 035011 sal decoherence due to gravitational time dilation, Nat. (2018). Phys. 11, 668 (2015). [16] M. Zych, Ł. Rudnicki, and I. Pikovski, Gravitational [7] E. C. Ruiz, F. Giacomini, and Č. Brukner, Entanglement mass of composite systems, Phys. Rev. D 99, 104029 of quantum clocks through gravity, PNAS 114, E2303 (2019). (2017). [17] A. R. H. Smith and M. Ahmadi, Quantizing time: Inter- [8] M. P. E. Lock and I. Fuentes, Relativistic Quantum acting clocks and systems, Quantum 3, 160 (2019). Clocks, in , Tutorials, Schools, and [18] S. Khandelwal, M. P. Lock, and M. P. Woods, Universal Workshops in the Mathematical Sciences, edited by quantum modifications to general relativistic time dila- R. Renner and S. Stupar (Springer International Pub- tion in delocalised clocks, Quantum 4, 309 (2020). lishing, Cham, 2017) pp. 51–68. [19] M. P. E. Lock and I. Fuentes, Quantum and classical [9] A. Bassi, A. Großardt, and H. Ulbricht, Gravitational effects in a light-clock falling in Schwarzschild geometry, Decoherence, Class. Quantum Grav. 34, 193002 (2017). Class. Quantum Grav. 36, 175007 (2019). [10] S. Bose, A. Mazumdar, G. W. Morley, H. Ulbricht, [20] P. A. Höhn, Switching Internal Times and a New Per- M. Toroš, M. Paternostro, A. A. Geraci, P. F. Barker, spective on the ‘Wave Function of the ’, Universe M. S. Kim, and G. Milburn, Spin Entanglement Wit- 5, 116 (2019). ness for , Phys. Rev. Lett. 119, 240401 [21] M. Zych, F. Costa, I. Pikovski, and Č. Brukner, Bell’s (2017). theorem for temporal order, Nat. Commun. 10, 3772 [11] C. Marletto and V. Vedral, Gravitationally Induced En- (2019). 14

[22] A. Dragan and A. Ekert, Quantum , of a photon: Wave-packet structures and atom-photon New J. Phys. 22, 033038 (2020). entanglement, Phys. Rev. A 72, 032110 (2005). [23] F. Giacomini, E. Castro-Ruiz, and Č. Brukner, Quantum [43] N. Stritzelberger and A. Kempf, Coherent delocalization mechanics and the covariance of physical laws in quantum in the light-matter interaction, Phys. Rev. D 101, 036007 reference frames, Nat. Commun. 10, 494 (2019). (2020). [24] E. Castro-Ruiz, F. Giacomini, A. Belenchia, and [44] A. S. Holevo, Probabilistic and Statistical Aspects of Č. Brukner, Quantum clocks and the temporal localis- Quantum Theory, Statistics and Probability, Vol. 1 ability of events in the presence of gravitating quantum (North-Holland, Amsterdam, 1982). systems, Nat. Commun. 11, 2672 (2020). [45] S. L. Braunstein, C. M. Caves, and G. J. Milburn, Gen- [25] P. A. Höhn, A. R. H. Smith, and M. P. E. Lock, The Trin- eralized Uncertainty Relations: Theory, Examples, and ity of Relational , arXiv:1912.00033 Lorentz Invariance, Annals of Physics 247, 135 (1996). [gr-qc, math-ph, quant-ph] (2019). [46] P. Busch, M. Grabowski, and P. J. Lahti, Time observ- [26] L. J. Henderson, A. Belenchia, E. Castro-Ruiz, C. Bu- ables in quantum theory, Physics Letters A 191, 357 droni, M. Zych, i. c. v. Brukner, and R. B. Mann, Quan- (1994). tum temporal superposition: The case of quantum field [47] M. Zych, Quantum Systems under Gravitational Time theory, Phys. Rev. Lett. 125, 131602 (2020). Dilation, Springer Theses (Springer International Pub- [27] L. C. Barbado, E. Castro-Ruiz, L. Apadula, and lishing, 2017). Č. Brukner, Unruh effect for detectors in superposition [48] M. Wilkens, Spurious velocity dependence of free-space of , Phys. Rev. D 102, 045002 (2020). spontaneous emission, Phys. Rev. A 47, 671 (1993). [28] J. Foo, S. Onoe, and M. Zych, Unruh-deWitt detectors [49] M. Wilkens, Significance of Röntgen current in quantum in quantum superpositions of trajectories, Phys. Rev. D optics: Spontaneous emission of moving atoms, Phys. 102, 085013 (2020). Rev. A 49, 570 (1994). [29] V. Vedral and F. Morikoshi, Schrödinger’s Cat Meets [50] S. M. Barnett and M. Sonnleitner, The vacuum friction Einstein’s Twins: A Superposition of Different Clock paradox and related puzzles, Contemporary Physics 59, Times, Int. J. Theor. Phys. 47, 2126 (2008). 145 (2018). [30] J. Lindkvist, C. Sabín, I. Fuentes, A. Dragan, I.-M. [51] M. Sonnleitner, N. Trautmann, and S. M. Barnett, Will Svensson, P. Delsing, and G. Johansson, Twin para- a decaying atom feel a friction force?, Phys. Rev. Lett. dox with macroscopic clocks in superconducting circuits, 118, 053601 (2017). Phys. Rev. A 90, 052113 (2014). [52] M. Sonnleitner and S. M. Barnett, Mass-energy and [31] M. Zych, F. Costa, I. Pikovski, and Č. Brukner, Quantum anomalous friction in , Phys. Rev. A 98, interferometric visibility as a witness of general relativis- 042106 (2018). tic proper time, Nat. Commun. 2, 505 (2011). [53] W. C. Röntgen, Ueber die durch Bewegung eines im [32] P. A. Bushev, J. H. Cole, D. Sholokhov, N. Kukharchyk, homogenen electrischen Felde befindlichen Dielectricums and M. Zych, Single electron relativistic clock interfer- hervorgerufene electrodynamische Kraft, Ann. Phys. ometer, New J. Phys. 18, 093050 (2016). Chem. 35, 264 (1888). [33] S. Loriani, A. Friedrich, C. Ufrecht, F. D. Pumpo, [54] M. Babiker, Theory of momentum changes in atomic S. Kleinert, S. Abend, N. Gaaloul, C. Meiners, C. Schu- translation due to irradiation with resonant light,J. bert, D. Tell, É. Wodey, M. Zych, W. Ertmer, A. Roura, Phys. B At. Mol. Phys. 17, 4877 (1984). D. Schlippert, W. P. Schleich, E. M. Rasel, and E. Giese, [55] L. G. Boussiakou, C. R. Bennett, and M. Babiker, Quan- Interference of clocks: A quantum , Sci. tum theory of spontaneous emission by real moving Adv. 5, eaax8966 (2019). atoms, Phys. Rev. Lett. 89, 123001 (2002). [34] A. Roura, Gravitational in Quantum-Clock In- [56] J. D. Cresser and S. M. Barnett, The rate of spontaneous terferometry, Phys. Rev. X 10, 021014 (2020). decay of a moving atom, J. Phys. B: At. Mol. Opt. Phys. [35] A. J. Paige, A. D. K. Plato, and M. S. Kim, Classical and 36, 1755 (2003). Nonclassical Time Dilation for Quantum Clocks, Phys. [57] R. Matloob, Decay rate of an excited atom in a moving Rev. Lett. 124, 160602 (2020). medium, Phys. Rev. Lett. 94, 050404 (2005). [36] A. R. H. Smith and M. Ahmadi, Quantum clocks observe [58] M. Sonnleitner and S. M. Barnett, Mass-energy and classical and quantum time dilation, Nat. Commun. 11, anomalous friction in quantum optics, Phys. Rev. A 98, 5360 (2020). 042106 (2018). [37] A. Peres, Measurement of time by quantum clocks, Am. [59] P. K. Schwartz and D. Giulini, Post-Newtonian Hamilto- J. Phys. 48, 552 (1980). nian description of an atom in a weak gravitational field, [38] C. W. Chou, D. B. Hume, T. Rosenband, and D. J. Phys. Rev. A 100, 052116 (2019). Wineland, Optical Clocks and Relativity, Science 329, [60] F. Shafieiyan, E. Amooghorban, and A. Mahdifar, Spon- 1630 (2010). taneous emission of a moving atom in the presence of [39] Z. Białynicka-Birula, P. Meystre, E. Schumacher, and magnetodielectric material: A relativistic approach, Opt. M. Wilkens, Velocity-dependent spontaneous emission, Commun. 426, 63 (2018). Opt. Commun. 85, 315 (1991). [61] R. Lopp and E. Martín-Martínez, Quantum delocaliza- [40] K. Rzążewski and W. Żakowicz, Spontaneous emission tion, gauge, and quantum optics: Light-matter interac- from an extended wavepacket, J. Phys. B At. Mol. Opt. tion in relativistic quantum information, Phys. Rev. A Phys. 25, L319 (1992). 103, 013703 (2021). [41] O. Steuernagel and H. Paul, Decoherence from sponta- [62] V. Weisskopf and E. Wigner, Berechnung der natürlichen neous emission, Phys. Rev. A 52, R905 (1995). Linienbreite auf Grund der Diracschen Lichttheorie, Z. [42] M. V. Fedorov, M. A. Efremov, A. E. Kazakov, K. W. Phys. 63, 54 (1930). Chan, C. K. Law, and J. H. Eberly, Spontaneous emission [63] B. Welz, Atomic absorption spectrometry (Wiley-VCH, 15

Weinheim New York, 1999). N. D. Lemke, K. Beloy, M. Pizzocaro, C. W. Oates, and [64] S. M. Brewer, J. S. Chen, A. M. Hankin, E. R. Clements, A. D. Ludlow, An atomic clock with 10−18 instability, C. W. Chou, D. J. Wineland, D. B. Hume, and D. R. Science 341, 1215 (2013). Leibrandt, 27Al+ Quantum-Logic Clock with a System- [77] A. D. Ludlow, M. M. Boyd, J. Ye, E. Peik, and P. O. atic Uncertainty below 10−18, Phys. Rev. Lett. 123, Schmidt, Optical atomic clocks, Rev. Mod. Phys. 87, 637 033201 (2019). (2015). [65] K. G. Johnson, J. D. Wong-Campos, B. Neyenhuis, [78] Ł. Dobrek, M. Gajda, M. Lewenstein, K. Sengstock, J. Mizrahi, and C. Monroe, Ultrafast creation of large G. Birkl, and W. Ertmer, Optical generation of vortices Schrödinger cat states of an atom, Nat. Commun. 8, 1 in trapped Bose-Einstein condensates, Phys. Rev. A 60, (2017). R3381 (1999). [66] A. D. Cronin, J. Schmiedmayer, and D. E. Pritchard, Op- [79] J. Denschlag, J. E. Simsarian, D. L. Feder, C. W. Clark, tics and interferometry with atoms and molecules, Rev. L. A. Collins, J. Cubizolles, L. Deng, E. W. Hagley, Mod. Phys. 81, 1051 (2009). K. Helmerson, W. P. Reinhardt, S. L. Rolston, B. I. [67] P. Cladé, S. Guellati-Khélifa, F. Nez, and F. Biraben, Schneider, and W. D. Phillips, Generating Solitons by Large Momentum Beam Splitter Using Bloch Oscilla- Phase Engineering of a Bose-Einstein Condensate, Sci- tions, Phys. Rev. Lett. 102, 240402 (2009). ence 287, 97 (2000). [68] S.-w. Chiow, T. Kovachy, H.-C. Chien, and M. A. Kase- [80] H. Y. Lo, D. Kienzler, L. De Clercq, M. Marinelli, V. Neg- vich, 102~k large area atom interferometers, Phys. Rev. nevitsky, B. C. Keitch, and J. P. Home, Spin-motion en- Lett. 107, 130403 (2011). tanglement and state diagnosis with squeezed oscillator [69] G. D. McDonald, C. C. N. Kuhn, S. Bennetts, J. E. Debs, wavepackets, Nature 521, 336 (2015). K. S. Hardman, M. Johnsson, J. D. Close, and N. P. [81] T. Kovachy, P. Asenbaum, C. Overstreet, C. A. Donnelly, Robins, 80~k momentum separation with bloch oscilla- S. M. Dickerson, A. Sugarbaker, J. M. Hogan, and M. A. tions in an optically guided atom interferometer, Phys. Kasevich, Quantum superposition at the half-metre scale, Rev. A 88, 053620 (2013). Nature 528, 530 (2015). [70] J. Rudolph, T. Wilkason, M. Nantel, H. Swan, C. M. Hol- [82] J. M. McGuirk, M. J. Snadden, and M. A. Kasevich, land, Y. Jiang, B. E. Garber, S. P. Carman, and J. M. Large area light-pulse atom interferometry, Phys. Rev. Hogan, Large momentum transfer clock atom interferom- Lett. 85, 4498 (2000). etry on the 689 nm intercombination line of strontium, [83] P. W. Graham, J. M. Hogan, M. A. Kasevich, and S. Ra- Phys. Rev. Lett. 124, 083604 (2020). jendran, New method for detection [71] M. Kasevich and S. Chu, Atomic interferometry using with atomic sensors, Phys. Rev. Lett. 110, 171102 (2013). stimulated raman transitions, Phys. Rev. Lett. 67, 181 [84] B. Plotkin-Swing, D. Gochnauer, K. E. McAlpine, E. S. (1991). Cooper, A. O. Jamison, and S. Gupta, Three-Path Atom [72] Z.-K. Hu, B.-L. , X.-C. Duan, M.-K. Zhou, L.-L. Interferometry with Large Momentum Separation, Phys. Chen, S. Zhan, Q.-Z. Zhang, and J. Luo, Demonstration Rev. Lett. 121, 133201 (2018). of an ultrahigh-sensitivity atom-interferometry absolute [85] C. J. Kennedy, E. Oelker, J. M. Robinson, T. Bothwell, gravimeter, Phys. Rev. A 88, 043610 (2013). D. Kedar, W. R. Milner, G. E. Marti, A. Derevianko, and [73] I. Dutta, D. Savoie, B. Fang, B. Venon, C. L. Gar- J. Ye, Precision meets cosmology: Improved rido Alzar, R. Geiger, and A. Landragin, Continuous constraints on ultralight dark matter from atom-cavity cold-atom inertial sensor with 1 nrad/sec sta- frequency comparisons, Phys. Rev. Lett. 125, 201302 bility, Phys. Rev. Lett. 116, 183003 (2016). (2020). [74] S. Abend, M. Gebbe, M. Gersemann, H. Ahlers, [86] M. Saba, T. A. Pasquini, C. Sanner, Y. Shin, W. Ket- H. Müntinga, E. Giese, N. Gaaloul, C. Schubert, C. Läm- terle, and D. E. Pritchard, Light Scattering to Determine merzahl, W. Ertmer, W. P. Schleich, and E. M. Rasel, the Relative Phase of Two Bose-Einstein Condensates, Atom-chip fountain gravimeter, Phys. Rev. Lett. 117, Science 307, 1945 (2005). 203003 (2016). [87] N. R. Thomas, A. C. Wilson, and C. J. Foot, Double-well [75] R. Geiger, A. Landragin, S. Merlet, and F. P. D. San- magnetic trap for Bose-Einstein condensates, Phys. Rev. tos, High-accuracy inertial measurements with cold-atom A 65, 063406 (2002). sensors, AVS Quantum Sci. 2, 024702 (2020). [88] A. R. H. Smith, Quantum time dilation: A new test of [76] N. Hinkley, A. Sherman, N. B. Phillips, M. Schioppo, relativistic quantum theory, arXiv:2004.10810 [quant-ph] (2020).