Unveiling Supersolid Order via Vortex Trajectory Correlations

Abstract

The task of experimentally investigating the inherently dual properties of a supersolid, a simultaneous superfluid and solid, has become more critical following the recent experimental evidence for supersolids in dipolar Bose-Einstein condensates (BECs) of ${}^{164}\text{Dy}$. We introduce a supersolid order parameter that uses vortex-vortex trajectory correlations to simultaneously reveal the periodic density of the underlying solid and superfluidity in a single measure. We propose experiments using existing technology to optically create and image trajectories of vortex dipoles in dipolar BECs. We numerically test our observable and find that vortex-vortex correlations reveal the supersolid lattice structure while distinguishing it from superfluidity. Our method sets the stage for experiments to use vortex trajectory correlations to investigate fundamental properties of supersolids arising from their dynamics and phase transitions.

The coexistence of a superfluid and solid at the same time and place defines a supersolid [1, 2, 3, 4]. The superfluid component spontaneously breaks the $U(1)$ gauge symmetry of the many-body wavefunction with signatures such as quantized vortices and irrotational flow. Whereas the solid component spontaneously breaks translational symmetry with signatures that include persistent spatial oscillations in particle-particle pair correlations. Combined observation of such overlap** order parameters signals a supersolid, Fig.  1(a).

Definitive signatures of supersolids obtained from globally averaged observables are challenging to interpret. For example, torsional oscillation experiments were proposed to reveal irrotational flow of supersolid ${}^{4}\text{He}$ [5]. Experiments observing expected behavior [6] were brought into question [7, 8] by the prospect of solid dislocations and defects [9, 4] containing superfluid components that mimic irrotational flow expected in a supersolid.

The search for informative experimental probes of supersolids has become more pressing with the advent of engineered supersolidity [10] of ultracold bosonic atoms [11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27], particularly in highly magnetic atoms such as ${}^{164}\text{Dy}$. The tunable dipole-dipole interaction between ${}^{164}\text{Dy}$ atoms allows strong interaction to drive the formation of periodic high-density crystalline structures (quantum droplets) [28, 29, 30, 31] that spontaneously break translational invariance [27]. This solid gives way to a supersolid as the strength of dipolar interaction is lowered to become comparable to kinetic energy. Experiments identifying and probing dysprosium supersolids [15, 19, 17, 16, 18, 21, 20, 22, 23, 24, 25, 26] rely on close support from theory [32, 33, 34, 35, 36, 37, 38, 39, 40].

Figure 1: (a) Schematic phase diagram where weak particle-particle interaction (compared to kinetic energy) favors a superfluid and strong interaction favors a solid. Overlap** order parameters define a supersolid. (b) The same as (a) but here a non-zero vortex-vortex trajectory correlation function offers a distinct supersolid order parameter. (c) Proposal to create a vortex dipole amid quantum droplets defining a quasi-two dimensional supersolid in a dipolar BEC using a pair of Laguerre-Gaussian beams.

Quantized vortices offer probes of superfluid order and therefore supersolids [41, 42, 43, 44, 45, 46]. Several methods have been used to create vortices in atomic BECs. Rotation by stirring [47, 48, 49] was recently used to observe vortices in rotating supersolid ${}^{164}\text{Dy}$ [45]. Other methods include quenching [50, 51, 52, 53, 54], optical phase-engineering and holographic methods [55, 56, 57, 58], and Laguerre-Gaussian (LG) beams [59, 60, 61, 62, 63, 64, 65, 66]. Ref. 57 holographically created a vortex dipole (a pair of vortices with opposite angular momentum) [67] in a BEC with no net angular momentum thus avoiding net rotation. Furthermore, vortex dipoles move as quasiparticles making them excellent candidates for recent experimental advances observing vortex trajectories in BECs [68, 69, 70, 71, 72, 73], including the use of non-destructive in situ trajectory imaging [74]. The creation of vortex dipoles and the direct imaging of their trajectories is therefore within reach of near-term experiments in supersolid ${}^{164}\text{Dy}$.

We propose vortex-vortex correlations in trajectory imaging as quantitative probes of the fundamental duality inherent to supersolids, Fig.  1(b). Our central observation is that quantized vortices necessitate a superfluid component but also move along density gradient contours [75, 76, 77] to reveal the underlying lattice structure of a supersolid (akin to pair-correlations of the constituent bosons). Fig.  1(c) depicts a proposed experiment wherein a pair of LG beams creates vortex dipoles in a supersolid. We numerically test the efficacy of our setup as a route to extract vortex-vortex trajectory correlations experimentally. We consider uniform [78] and harmonic traps. We find that our correlation function distinguishes the supersolid from the superfluid and, in the uniform trap, reveals the supersolid lattice structure.

Our proposal for a unique supersolid order parameter and its experimental implementation establishes a method for identifying supersolids, measuring lattice properties, tracking phase transitions, and monitoring supersolid lattice dynamics.

Model.— We consider $N$ ¹⁶⁴Dy atoms trapped in an external potential $V_{\text{ext}}(\mathbf{r})$ while interacting via contact and dipolar interaction potentials [79, 80, 81]:

$$V_{\rm c}(\mathbf{r})=g\delta(\mathbf{r})\mbox{\quad and\quad}V_{\rm dd}(\mathbf{r})=\frac{\mu_{0}\mu^{2}}{4\pi}\frac{1-3\cos^{2}\theta}{r^{3}},$$ (1)

respectively. The contact interaction strength $g$ is related to the $s$-wave scattering length $a_{s}$ by $g=4\pi\hbar^{2}a_{s}/m$ where $m$ is the mass of the atom, $\mu_{0}$ is the vacuum permeability, and $\mu$ is the magnetic moment of the atom. The dipolar interaction strength varies with the angle $\theta$ between the relative position vector $\mathbf{r}$ and the polarization direction ($z$-axis) of the dipole.

We model the dynamics of the entire system with the extended Gross-Pitaevskii equation (eGPE) [28, 82]:

$$\mathrm{i}\hbar\partialderivative{t}\psi(\mathbf{r})=\Big{[}- \frac{\hbar^{2}}{2m}\laplacian+V_{\text{ext}}(\mathbf{r})+g\absolutevalue{\psi(\mathbf{r})}^{2}+\gamma\absolutevalue{\psi(\mathbf{r})}^{3}$$

$$+\int\differential{\mathbf{r^{\prime}}}V_{\rm dd}(\mathbf{r}-\mathbf{r^{\prime}})\absolutevalue{\psi(\mathbf{r^{\prime}})}^{2}\Big{]}\psi(\mathbf{r}).$$ (2)

Here, $\psi(\mathbf{r})$ is the system wavefunction normalized with $\int\differential{\mathbf{r}}\absolutevalue{\psi}^{2}=N$. The Lee-Huang-Yang [83, 84, 85] correction term contains $\gamma(\epsilon_{\rm dd})=(128\sqrt{\pi}\hbar^{2}a_{s}^{5/2}/3m)(1+\frac{3}{2}\epsilon_{\rm dd}^{2})$ [86, 87], which depends on the ratio of the dipolar length scale to $a_{s}$, $\epsilon_{\rm dd}=a_{\rm dd}/a_{s}$, where $a_{\rm dd}=\mu_{0}\mu^{2}m/(12\pi\hbar^{2})$. $\epsilon_{\rm dd}$ can be tuned using a Feshbach resonance in ¹⁶⁴Dy to vary $a_{s}$ [88, 27] thus allowing exploration of phase transitions [82]. For ¹⁶⁴Dy, the dipole moment $\mu{=}9.93\mu_{B}$, sets a dipolar length scale $a_{\rm dd}{=}131a_{0}$ where $\mu_{B}$ and $a_{0}$ are the Bohr magneton and Bohr radius, respectively. We choose $a_{s}=94a_{0}(\epsilon_{\rm dd}=1.39)$ for and $110a_{0}(\epsilon_{\rm dd}=1.19)$ to achieve the supersolid and superfluid phases, respectively [16].

To study vortex dynamics in superfluid and supersolid phases, we assume two different external trap** potentials: (i) uniform in the $xy$ plane and harmonic along $z$-axis, $V_{\rm ext}{=}m\omega_{z}^{2}z^{2}/2$ and (ii) an oblate 3D harmonic trap, $V_{\rm ext}{=}m\omega^{2}(x^{2}+y^{2}+\lambda^{2}z^{2})/2$ where trap** anisotropy $\lambda$ is the ratio of trap** frequency along the $z$-axis to that in the $xy$ plane. In both cases, we induce vortex dipoles using the phase-imprinting method [89, 90] by multiplying the wavefunction $\psi(\mathbf{r})$ with a phase factor of $\exp[\mathrm{i}l(y-y_{0})/(x-x_{0})]$. Here $l\hbar$, with $l=\pm 1$, is the vortex angular momentum and $(x_{0},y_{0})$ is the vortex initial position.

Vortex motion in a uniform trap.— We first consider a uniform planar trap to ignore the effect of density variation due to trap** confinement. We solve the eGPE  2 using the split-step Crank-Nicolson method [91] and study the vortex dynamics in the system by tracking the vortex trajectories over time. In the uniform trap, $N{=}1.3\times 10^{6}$ atoms are trapped along the $z$-axis with $\omega_{z}{=}2\pi\times 150$ Hz, and hard walls in the $xy$ plane. The simulation box size is $L_{x}{=}L_{y}{=}2L_{z}{=}32.8\mu$m and the grid size is $(256\times 256\times 128)$.

Figure 2: Groundstate density profile in the $z=0$ plane and vortex motion in a uniform trap in (a) supersolid and (b) superfluid phases. $\rm v_{\it{i}}^{+}$ and $\rm v_{\it{i}}^{-}$ are the initial position of the $l=+1$ and $l=-1$ vortex of $i$-th vortex dipole. The red (green) line represents the trajectory of a $l=+1$ ($l=-1$) vortex. The black arrows in (b) represent the direction of motion of vortex dipoles in the superfluid phase. In (b), the simulations of vortex dipoles $(\rm v_{1}^{+},v_{1}^{-})$ and $(\rm v_{2}^{+},v_{2}^{-})$ are done separately and the background density is the product of two groundstates to identify the vortices.

By solving the eGPE, we obtain the superfluid and supersolid ground state for different $\epsilon_{\rm dd}$. In a uniform trap the superfluid phase has a flat density profile whereas the supersolid phase shows a density modulation. The supersolid crystalline structure of quantum droplets contains a low density background, Fig.  1(c).

Vortices follow qualitatively distinct trajectories in the supersolid and superfluid phases. To see this, we dynamically imprint a vortex dipole. The vortex dipole will have a linear momentum in the superfluid phase which is inversely proportional to the separation $d$ of $l=+1$ and $l=-1$ vortices [92, 93]. Both vortices therefore move in straight lines parallel to each other with a constant velocity in the superfluid. In contrast, in the supersolid phase, the vortex trajectories depend on non-uniform and periodic density patterns. In the vortex’s frame of reference, density changes with time and space leading to a non-uniform force on the vortices. Due to this non-uniform force, vortices move in non-linear contours outlining the supersolid density gradient. In contrast to the superfluid phase, the vortex dipole is not stable in the supersolid phase and eventually annihilates when the $l=+1$ and $l=-1$ vortices approach each other. We test these qualitative expectations quantitatively using eGPE.

Figure  2 shows the vortex trajectories in the superfluid and supersolid phases. We choose a vortex dipole $(\rm v_{1}^{+},v_{1}^{-})$ in both phases as shown in Figs.  2(a) and  2(b). The $l=+1$ and $l=-1$ vortices follow non-uniform trajectories and eventually annihilate in the supersolid phase. While in the superfluid phase, they move in straight lines parallel to each other. The vortex dipole motion in the superfluid phase is stable, and they do not annihilate.

We verify that the velocity of the vortices is inversely proportional to the separation. In the superfluid phase we prepare a well separated $l=\pm 1$ vortex pair $(\rm v_{1}^{+},v_{1}^{-})$. Fig.  2(b) shows that the distances traveled by this vortex dipole are small. In Fig.  2(b), we also show motion of a closely spaced vortex dipole $(\rm v_{2}^{+},v_{2}^{-})$ in the superfluid phase (this is studied separately from $(\rm v_{1}^{+},v_{1}^{-})$, not shown in Fig.  2(a) for clarity of the figure). These vortices traveled a longer distance (full paths are not shown) compared to the vortex dipole $(\rm v_{1}^{+},v_{1}^{-})$.

Correlation of vortex trajectories.— We now turn to our central proposal. Given distinct behavior of vortex motion in each phase we seek to extract quantitative information from the vortex trajectories. We define a vortex-vortex trajectory correlation function:

$$C(\mathbf{r})=\expectationvalue{s(\mathbf{r}_{0})s(\mathbf{r}_{0}+\mathbf{r})},$$ (3)

where $s(\mathbf{r})$ is 1 if a vortex of any $l$ passes through $\mathbf{r}$ and 0 otherwise. In the following calculations we choose to compute $C(\mathbf{r})$ using trajectories of opposite $l$ with no loss of generality. Note that $C(\mathbf{r})$ vanishes in the quantum droplet phase (because there is no phase coherence for vortices to form) and should show only trivial structure in the superfluid phase because the vortices move in straight lines. But in the supersolid phase, measurements of $C(\mathbf{r})$ will probe the underlying crystal structure of the supersolid. $C(\mathbf{r})$ can therefore be used as a supersolid order parameter. Importantly, the initial configuration of vortices must be chosen to break the underlying translational symmetry of the supersolid phase, otherwise straight-line trajectories will arise [94]. Straight trajectories will not allow measurements to distinguish between superfluid and supersolid states. In the following, we choose the initial positions of the vortex dipoles on the opposite arcs of a circle in the simulations to avoid biased choices in probing the quantum droplet square lattice in the supersolid phase.

We test the utility of $C(\mathbf{r})$ using eGPE simulations designed to replicate repeated measurements in a uniform trap. Figs.  3(a) and  3(b) show nine separate simulations of the resulting motion of the vortex dipoles $[(\rm v_{1}^{+},v_{1}^{-}),\cdots,(\rm v_{9}^{+},v_{9}^{-})]$ in both superfluid and supersolid phases. For all the vortex dipoles, the $l=+1$ and $l=-1$ vortex separation $d\approx 13{\mu\rm m}$ is the same as shown by the initial position of the $l=+1$ ($l=-1$) vortex by red (green) dots in Figs.  3(a) and  3(b). We locate the vortex position at different instants of time and find the trajectories of the vortices as time evolves. We find different trajectories for different sets of initial positions of the vortices. By merging all of the trajectories we obtain a combined 2D trajectory on the $xy$ plane shown in Figs.  3(a) and  3(b).

Figure 3: Combined vortex trajectories of nine vortex dipoles $[(\rm v_{1}^{+},v_{1}^{-}),\cdots,(\rm v_{9}^{+},v_{9}^{-})]$ from separate simulations in (a) supersolid and (b) superfluid phases. The red (green) dots represent the initial position of the $l=+1$ ($l=-1$) vortex of the vortex dipoles. The grey lines represent the vortex trajectories. The correlation function $C(\mathbf{r})$ of the vortex trajectories in (c) superfluid and (d) supersolid phases. (c) and (d) show the correlation function $C(\mathbf{r})$ of the vortex trajectories in the supersolid and superfluid phases, respectively. In (c), apart from the primary peak at $\mathbf{r}=0$, there are the secondary peaks at $\mathbf{r}\neq 0$ denoted by yellow arrows for $C(0,y)$. In contrast, in (d), the correlation function $C(\mathbf{r})$ shows only the primary peak at $\mathbf{r}=0$. All simulation parameters are the same as in Fig.  2.

We find that even with a limited number of trajectories, $C(\mathbf{r})$ reveals aspects of the underlying lattice structure of the supersolid. Fig.  3(c) shows that the correlation function $C(\mathbf{r})$, apart from the primary trivial peak at $\mathbf{r}=0$, shows secondary peaks at $\mathbf{r}\neq 0$ due to the non-uniform and periodic density patterns in the supersolid phase. Whereas in the superfluid phase, $C(\mathbf{r})$ shows only the primary peak at $\mathbf{r}=0$ due to the straight-line motion of the vortices. Thus, the correlation function $C(\mathbf{r})$ of vortex trajectories is a useful tool to distinguish the superfluid and supersolid phases.

Use of $C(\mathbf{r})$ to probe a supersolid requires the following conditions to be met: (i) As mentioned above, initial positions must break the supersolid crystal symmetry. If initial vortex positions are chosen, for example, to be random, a sufficient number of trajectories must be imaged to resolve auxiliary peaks. The limited number of runs in Fig.  3 were chosen near this threshold. (ii) Vortices are, ideally, to be used as non-invasive probes that leave the supersolid intact. We assume an LG beam spot size ($\sim 1\mu\text{m}$) below the supersolid lattice spacing ($\sim 3\mu\text{m}$ in our simulations) and with initial inter-vortex spacing high enough to avoid high vortex speeds. (iii) We assume the use of a uniform trap (or a sufficiently large number of atoms) to allow resolution of translational symmetry breaking. The supplementary material [94] shows data relaxing condition (i) in order to map the supersolid lattice whereas the following simulations relax conditions (ii) and (iii).

Vortex motion in a harmonic trap.— We now turn to small system sizes and harmonic trap** relevant for ongoing experiments. Tight trap** will limit the range of the solid and can even lead to spurious supersolid signatures [95]. We choose $N{=}80000$ atoms to be trapped in an oblate-shaped 3D harmonic trap with trap** frequencies $(\omega,\omega_{z}){=}2\pi\times(45,133)$ Hz [34], where $\omega$ is the transverse trap** frequency. The simulation box size is $L_{x}{=}L_{y}{=}L_{z}{=}30\mu$m.

We studied the motion of vortex dipoles in both superfluid ($a_{s}=110a_{0}$) and supersolid ($a_{s}=94a_{0}$) regimes. Without any vortices, the density profile of the superfluid is Gaussian, whereas the supersolid regime shows a single quantum droplet in the center of the trap with a low-density superfluid background [16, 18]. In both phases, we prepare ground states with a vortex dipole $(\rm v_{1}^{+},v_{1}^{-})$ separated by a distance $d\approx 6\mu\rm m$, and we observe the motion of the vortices by tracking the vortex trajectories over time.

Figure 4: (a) Snapshots of density profiles in the $z=0$ plane at different times in the harmonic trap after the vortex dipole is imprinted in the supersolid phase. The red (green) circled “+”(“-”) marker represents the $l=+1$ ($l=-1$) vortex. (b) and (c) show the $l=+1$ (red dots) and $l=-1$ (green dots) vortex trajectories in the supersolid and superfluid regimes, respectively.

Figure  4(a) shows density profiles in the supersolid regime after the vortex dipole is imprinted. At $t=0$ the creation of the vortices causes the single droplet at the center to split into multiple droplets. Over time, these droplets continuously deform and form at different locations due to the motion of the vortices. Here the dynamics explores a variety of metastable small supersolid configurations. Once the $l=\pm 1$ vortices annihilate each other, the small supersolid reaches an equilibrium state with three droplets, as we see at $t=700\,\rm ms$ in  4(a). The trajectories of the vortices are affected by the density modulation in the supersolid regime, moving in non-linear paths before annihilation [Fig.  4(b)]. In contrast, the vortices in the superfluid regime do not annihilate. They keep moving in an elliptical path with a constant speed, as shown in Fig.  4(c) [96, 97]. We therefore see that vortex dipole trajectories in the supersolid and superfluid regimes show qualitatively distinct behavior even in small systems with tight trap**.

Discussion.— We introduced and numerically tested a supersolid order parameter defined by correlations in vortex trajectories. Our tests used a limited number of trajectories to replicate sampling for near term experiments. Larger numbers of samples allow resolution of complete supersolid lattice structures [94]. Our correlation function can also be used to study important properties of supersolids, including lattice dynamics due to supersolid phonons and phase transitions. Possible applications to different systems include exploration of broken translational symmetry in thin films of superfluid ${}^{4}\text{He}$ [98, 99, 100] and Fulde-Ferrell-Larkin-Ovchinnikov [101, 102] superconductors.