Abstract
The extended BoseHubbard model captures the essential properties of a wide variety of physical systems including ultracold atoms and molecules in optical lattices, Josephson junction arrays, and certain narrow band superconductors. It exhibits a rich phase diagram including a supersolid phase where a lattice solid coexists with a superfluid. We use quantum Monte Carlo to study the supersolid part of the phase diagram of the extended BoseHubbard model on the simple cubic lattice. We add disorder to the extended BoseHubbard model and find that the maximum critical temperature for the supersolid phase tends to be suppressed by disorder. But we also find a narrow parameter window in which the supersolid critical temperature is enhanced by disorder. Our results show that supersolids survive a moderate amount of spatial disorder and thermal fluctuations in the simple cubic lattice.
Similar content being viewed by others
Introduction
A superfluid and a solid can, in principle, coexist in the same place at the same time. This unique state of matter, a supersolid, has attracted numerous research efforts since it was proposed more than 45 years ago in the context of superfluid ^{4}He^{1,2}. ^{4}He experiments suggesting possible observation of a supersolid have remained controversial^{3,4}.
Lattice models of supersolids (the extended BoseHubbard model^{5} in particular) were used to study the critical properties of the supersolid phase of ^{4}He^{6,7}. But subsequent work showed that these lattice models capture the essential properties of many other physical systems, including ultracold atoms and molecules in optical lattices^{8,9,10,11,12,13}, Josephson junction arrays^{14,15}, and narrowband superconductors^{16}. The latter connection can be made rigorous via a direct mapping between local Cooper pairs and bosons. The supersolid of bosons in this case maps to coexisting superconducting and charge density wave order which has been of interest in a variety of compounds, e.g., BaBiO_{3} doped with K or Pb^{17}.
Experiments with ultracold atoms offer an excellent opportunity to observe supersolids. These systems are clean and parameters can be tuned to maximize the strength of a supersolid if one is predicted to exist. Different routes to observing a lattice supersolid have been explored^{8,9,10,11,12,13}. The key to supersolidity is coexisting order which is, in turn, derived from coexisting diagonal and offdiagonal longrange order in the density matrix. Along these lines measurements in three dimensions (3D) consistent with coexisting order have been made with ultracold atoms in cavities mediating longrange interactions^{18,19,20} and synthetic spinorbit interactions^{21} therefore offering new evidence for supersolids in a controlled and pristine environment.
Work on lattice models suggests that supersolids should be rather delicate and therefore difficult to observe when quantum fluctuations are more pronounced (particularly in lower dimensions). In two dimensions (2D) it is now known that the lattice supersolid competes with phase separation. A meanfield argument^{22} shows that the formation of domain walls favors phase separation because (for low coordination number) the domain wall intrinsic to the phase separated state gains in kinetic energy. But on lattices with higher coordination number, e.g., the triangular lattice, quantum Monte Carlo (QMC) calculations show^{23} that phase separation is suppressed and the lattice supersolid state gains in energy.
Furthermore, QMC results show that lattice supersolids in 2D are also highly susceptible to disorder. Results on the square lattice^{24} show that spatial disorder destroys the solid itself leaving no chance for the supersolid. This sensitivity stems from an ImryMatype mechanism^{24,25} implying that the solid is unstable in the presence of arbitrarily weak disorder in less than three dimensions (3D).
In 3D we expect a solid to be robust against disorder because the ImryMa mechanism is avoided^{26}. Furthermore, high coordination numbers have been shown to suppress phase separation on the simple cubic lattice. QMC results^{27,28} (in the absence of disorder) report a strong supersolid and no phase separation. 3D lattice models therefore seem to be the best arena to study supersolid behavior.
Study of the extended Bose Hubbard model in 3D has become more pressing because of recent work^{13} that has successfully demonstrated placement of bosonic chromium atoms in a cubic optical lattice. The atoms have a magnetic dipole moment. When polarized these moments induce long range interactions. A theoryexperiment comparison^{13} shows that the extended BoseHubbard model quantitatively captures the physics of this system thus paving the way for the possibility of a direct observation of a lattice supersolid as predicted by an extended BoseHubbard model.
We use QMC to study the stability of the lattice supersolid in the simple cubic lattice. We study the phase diagram of the extended BoseHubbard model. Our primary results are summarized in Fig. 1 where the phase diagram sketches critical temperature versus lattice hopping energy. We include disorder in our study to examine the stability of the supersolid. We find that disorder lowers the critical temperature but the supersolid still survives moderately strong disorder.
We also find that disorder enhances the supersolid critical temperature^{26} within a narrow parameter window. For low hopping the critical temperature of the solid component of the supersolid remains robust against disorder while the superfluid critical temperature is actually enhanced. We use systematic finitesize scaling to show that increasing disorder increases the critical temperature of the supersolid. Our results therefore indicate that supersolids in the extended BoseHubbard model are stable (and even enhanced) in the simple cubic lattice for a moderate amount of disorder. Our result is consistent with the disorder effect in dirty superconductors, where one finds that Anderson’s Theorem^{29} is generally observed in weakly interacting regimes, while, for strong interaction, the disorder enhances the superconducting transition temperature due to broadening of the conduction band^{30}.
Results
Model
We study a tightbinding model of repulsive softcore bosons hopping in a simple cubic lattice of side lengths L with onsite disorder:
where \({a}_{i}({a}_{i}^{\dagger })\) is the boson annihilation (creation) operator at site i, \({n}_{i}={a}_{i}^{\dagger }{a}_{i}\) is the particle number operator, t is the hopping integral, U is the onsite repulsion, and V is the nearestneighbor repulsion. Here μ_{ i } = μ−ε_{ i }, where μ is the average chemical potential of the system, and the uniformly distributed random number ε_{ i } ∈ [−Δ, Δ] is the onsite disorder potential. We use periodic boundary conditions. In contrast to the hardcore boson model, here multiple bosons can occupy the same lattice site in our softcore boson model, allowing an increase of onsite interaction energy. In the limit of U → ∞, the softcore boson model reduces to the hardcore boson model. Hereafter we will use U as the energy unit and set the Boltzmann constant k_{ B } = 1.
For V = 0 and Δ = 0 the model reduces to the well known BoseHubbard model^{31}. At zero temperature there exist two competing phases, an incompressible Mott insulator at low hopping and a superfluid (SF) at large hopping that spontaneously breaks the continuous U(1) gauge symmetry of the model (the phase invariance of the bosonic operators). At fixed μ these phases are separated by a quantum critical point at a critical t.
Including a nearestneighbor repulsion, V > 0, leads to additional phases. For large V the bosons tend to sit at every other site to form a charge density wave, a solid (S), which spontaneously breaks the \({{\mathbb{Z}}}_{2}\) sublattice symmetry. When the hopping and interaction terms are comparable a supersolid forms which derives from dual spontaneous symmetry breaking of both the U(1) gauge symmetry and the \({{\mathbb{Z}}}_{2}\) sublattice symmetry throughout the entire sample. The result is simultaneous superfluid and solid order, a supersolid (SS). To study a regime consistent with spatially decaying interactions and a strong supersolid we choose zV = U = 1^{22,27,28,32}, where z = 6 is the lattice coordination number.
Meanfield analyses of the disorderfree extended BoseHubbard model^{11,14,26} show that the supersolid sits between the solid and the superfluid in the phase diagram. Figure 2 shows the zero temperature mean field phase diagram in the dilute (low μ) regime with Δ = 0. In the following, we select a specific average chemical potential, μ = 0.7, unless otherwise stated. The horizontal line indicates that increasing t while keeping μ = 0.7 allows us to transverse three of the phases discussed so far, i.e., S → SS → SF. This choice also keeps the density at or below one. By adding disorder to phases lying along the horizontal line, we can obtain other intriguing phases, such as, the Bose glass (BG)^{31}. We also identify a compressible regime which maintains the character of a solid (long range order in the density). We call this regime a disordered solid (DS) phase and assume it plays the role of the Griffiths phase which must^{33,34} lie between the incompressible solid and compressible supersolid phases. This assignment follows from a low t mapping to disordered phases of the Spin1 BlumeCapel model (See Supplemental Material).
To focus primarily on the interplay of the supersolid with disorder we focus our study on weak disorder. We exclude higher disorder here because proper order parameters for finite size scaling analyses to determine the BG and DS phase boundaries at finite temperatures are unknown. Since there has been some recent progress in constructively identifying the BG phase using local condensate fraction^{35} at finite temperatures we think future studies of strong disorder phases should be able map out the phase boundaries.
Definition of Order Parameters
Each of the states discussed as low temperature phases of Eq. (1) correspond to unique combinations of order parameters. At high temperatures the normal phase (N) is defined by the absence of order (either local or nonlocal). Whereas low temperature regimes tend to show order in either the diagonal or offdiagonal parts of the singleparticle density matrix (or both as in the supersolid phase). This section lists the phases we find and the corresponding order parameters.
Solid order is defined by longrange oscillations in the densitydensity correlation function (diagonal longrange order in the density matrix) or, equivalently, peaks in the static structure factor at wavevector, Q:
that indicate a spontaneous breaking of the sublattice symmetry. N_{ s } = L^{3} is the number of sites. For the large values of V considered here an oscillation of the density between sublattices is favored, i.e., Q = (π, π, π) on the simple cubic lattice.
The solid phase we discuss here is incompressible. The compressibility defines how easily the particle number fluctuates in the system, and is given by:
Where T is the temperature and the average particle density is given by:
with:
The last equality in Eq. (2) shows that the compressibility is intrinsically nonlocal because it relates to density fluctuations across the entire system, \(\langle {\hat{N}}^{2}\rangle \).
The superfluid density describes the system’s response to external perturbations, such as translation or rotation. It is characterized by offdiagonal longrange order in the density matrix even in the presence of interactions. In the pathintegral QMC formalism the superfluid density is given by^{36}:
where the squared winding number is \({W}^{2}={W}_{x}^{2}+{W}_{y}^{2}+{W}_{z}^{2}\) and \({W}_{i}\) is the winding number in the i th direction with i = x, y, or z. We find that the above order parameters adequately characterize the lowtemperature phases of Eq. (1) at weak disorder.
Figure 3 summarizes the order parameters and the phases we discuss. As we vary T, t, and Δ, we find the following phases: normal, solid, superfluid, and supersolid. We also find evidence for a disordered solid in finitesize simulations. The absence of order at high T signals the normal phase. The system forms a ρ = 1/2 solid when it has longrange diagonal order, S_{ π } > 0, while maintaining incompressibility, κ = 0. Superfluid order is described by ρ_{ s } > 0 and κ > 0. To obtain supersolid order, the system needs to have coexisting solid and superfluid orders, i.e., S_{ π } > 0, ρ_{ s } > 0, and κ > 0. The disordered solid arises in the presence of disorder. Defects lead to domains with gapless edges that leave the system compressible, i.e., S_{ π } > 0 but with κ > 0. The Boseglass phase occurs for large disorder strengths. It has only local superfluid order (no offdiagonal longrange order). It is compressible but exists only at low T.
Quantum Monte Carlo Evaluation of Order Parameters
This section summarizes QMC calculations of order parameters as a function of parameters in Eq. (1). Parameter sweeps are used to qualitatively identify regions of the phase diagram with (and without) disorder. These parameter sweeps are then used to find phase boundaries using finitesize scaling.
To qualitatively locate phases on the T vs. t phase diagram of Eq. (1) we scan t as well as disorder Δ. We choose four disorder strengths: Δ = 0.0, 0.1, 0.3, and 0.5. To obtain temperature dependence we also sample the following set of temperatures: T = 0.025, 0.05, 0.1, 0.125, 167, 0.2, 0.25, and 0.5. We first do these simulations at L = 10 for our qualitative estimate. Figure 4 plots the order parameters of the model as a function of temperature for several different t. The top panels plot the superfluid density. We can compare all four top panels to see that for large t the disorder does not suppress the superfluid density much. We can understand this effect using the mapping to the attractive FermiHubbard model (See Supplementary Material) where the superfluid corresponds to an swave superconductor. The robustness of the superfluid found here then follows from the Anderson’s theorem^{29} for the robustness of swave superconductivity to disorder.
The middle panels in Fig. 4 plot the compressibility. Here we see that the finite size of the system keeps κ > 0 for all but the lowest t and t with Δ = 0. Using finitesize scaling we find that the solid phase is incompressible in the thermodynamic limit.
The bottom panels in Fig. 4 plot the structure factor. Here we see that at large t and/or T, the structure factor vanishes. This indicates that we have either the superfluid or normal phase. But for low t and low T the structure factor increases to reveal a supersolid and, for very low t, a pure solid. As disorder is increased to Δ = 0.1 the pure solid gives way to what appears to be a compressible phase. Here the distinct T dependence of the compressibility indicates a distinction from the pure solid. We tentatively assign this regime to be the disordered solid phase.
At Δ = 0.1, we find that the supersolid phase at t = 0.03 is destroyed. The system has ρ_{ s } = 0, κ > 0 and S_{ π } > 0, satisfying the definition of the disordered solid phase. By increasing t to t = 0.04, we recover the supersolid phase, which persists until t = 0.06. For larger t the system enters the superfluid phase. Δ = 0.3 shows a similar set of transitions. However, the critical temperature for Boseglass to normal phase transition is undetermined in our study, since we have not found a suitable scaling relation to describe the transition.
For Δ = 0.5 and small t (t = 0.02), we have κ > 0, ρ_{ s } = 0, and S_{ π } ~ 0, which is the Boseglass phase at low T. Our result is consistent with the existence of Boseglass phase predicted by theorem of inclusions^{37}. As we increase the hopping to t = 0.03, the system turns into the superfluid at low temperatures. The superfluid phase persists as we further increase t values.
Note that the above rough determination of phase boundaries for L = 10 will change with system size. The critical points deduced from Fig. 4 are only approximate. Precise determination of critical points can be achieved through finitesize scalings to be discussed below. However, from the numerical simulations at L = 10, we already see the rich phase diagram contained in the disordered extended BoseHubbard model. Numerical simulations at L = 10 also serve as a rough guide to phase transitions, which will suggest parameters for a precise finitesizing scaling analysis.
FiniteSize Scaling
To map out the finite temperature phase boundaries, we used QMC data to carry out finitesize scaling analyses for the order parameters. We found two distinct universality classes governing transitions: Ising and 3D XY. The Ising universality class applies to the longrange charge order/disorder transition while the 3D XY universality class applies to superfluid/nonsuperfluid transitions. In this section we discuss the methods we used to identify the transition points using finitesize scaling relations.
Since the longrange charge order to disorder transition belongs to the Ising universality class, the structure factor obeys the following scaling relation^{38}:
where \(\mathop{t}\limits^{ \sim }=(T{T}_{c})/{T}_{c}\) is the reduced temperature that measures the dimensionless distance from T to the critical temperature T_{ c }, β = 0.3265(3), v = 0.6301(4), a_{1} is a nonuniversal metric factor, and \(\mathop{S}\limits^{ \sim }\) is a scaling function. From Eq. (3) we see that if we plot \({L}^{\beta /\nu }{S}_{\pi }\) vs. T for different lattice sizes, different curves will intersect at \(T={T}_{c}\). Two example scaling figures are shown in the upper two panels of Fig. 5 for \({\rm{\Delta }}=0.1\).
On the other hand, the superfluid to nonsuperfluid transition belongs to the 3D XY universality class, and the superfluid density scaling satisfies the following scaling relation^{39}:
where \({\mathop{\rho }\limits^{ \sim }}_{s}\) is a scaling function, \(d=3\) is system dimension, and \({a}_{2}\) is a nonuniversal metric factor.
In 3D, we can plot \(L{\rho }_{s}\) vs. T for different lattice sizes. Different curves again intersect at \(T={T}_{c}\) for the transition. The lower two panels of Fig. 5 show example finitesize scaling analyses of \({\rho }_{s}\) for disorder strength \({\rm{\Delta }}=0.1\). We have checked that the \(L=610\) data are sufficient to give accurate critical points by including larger system sizes (\(L\le 20\)) for select parameters.
We also perform scaling analysis to locate the quantum critical point t_{ c } for superfluid density as we vary t. The superfluid density satisfies the following scaling relation^{38,40}:
where \(\alpha =2dz\), \(\delta =t{t}_{c}\) measures the distance to the critical point, z is the dynamical exponent, which is predicted to be z = d^{41}, a is a nonuniversal metric number, and the function f is universal. For our cubic lattice, we have α = −4. Figure 6 shows results from QMC simulations, where we keep \({L}^{3}/T=0.03125\). Hence, for \(L=\mathrm{4,6,}\) and 8, simulations are carried out at \({T}^{1}=\mathrm{2,6.75,16}\), respectively, for various t values around the critical point. Using these scaling relations we are able to locate phase transition lines to construct a phase diagram for the supersolid. Note that it is also possible to determine the critical temperatures using data collapse technique^{42} with the above scaling relations. However, one would need to perform more QMC simulations with a much finer temperature grid for various lattice sizes.
Phase Diagrams
This section culminates the results and methods presented in previous sections to construct QMC phase diagrams of Eq. (1). Finitesize scaling of the superfluid stiffness and the structure factor are used to find finite temperature critical points for the solid, supersolid, and superfluid phases. Finitesize scaling is also used to get the quantum critical points as a function of t. We find that disorder tends to suppress the supersolid critical temperature over much (but not all) of the phase diagram. Our central finding is that the supersolid is present at intermediate hopping even in the presence of disorder.
The \({\rm{\Delta }}=0\) panel in Fig. 7 plots the phase diagram of Eq. (1) in the absence of disorder as determined by QMC. Here squares and circles plot the critical temperature determined by finitesize scaling of the structure factor and the stiffness, respectively. We see that the solid and superfluid dominate at small and large hopping, respectively. The supersolid is found at intermediate hoppings.
The vertical dashed lines in Fig. 7 indicate an expected phase boundary. Our conclusions here are based on finitesize data without extrapolation. For example, for increasing system size drives the critical temperature to zero for \(t\mathop{ > }\limits_{ \tilde {}}0.0525\). Here we were not able to resolve the critical temperature uniquely given our method because the phase boundary is nearly vertical here.
The remaining panels in Fig. 7 plot the same as the top panel but in the presence of disorder. We find that increasing spatial disorder tends to lower the maximum critical temperature of the solid phase. Here the T_{ c } of the solid order tends to be more sensitive to disorder than the superfluid. It is therefore the lowering of T_{ c } of the solid that suppresses the supersolid behavior.
Griffiths effects should be particularly important in the thermodynamic limit near phase boundaries separating incompressible and compressible phases^{33,34}. The solid and supersolid are incompressible and compressible, respectively. Our phase diagrams omit the quantum Griffiths phase which, according to the theorem of inclusions^{33}, must separate these two phases. We tentatively assign the intermediate quantum Griffiths regime to be a disordered solid (in analogy to the BoseGlass in the ordinary BoseHubbard model^{33,34}) based on our preliminary finitesize results (Fig. 4). We have not been able to use finitesize scaling to identify the \({T}_{c}\) for the disordered solid. We therefore label the solid phase in the presence of disorder in Fig. 7 as S/DS to allow for the disordered solid phase between the solid and supersolid phases.
Disorder Enhanced Supersolids
The addition of disorder can, counterintuitively, enhance supersolidity in a narrow parameter window of the phase diagram. We first consider the impact of disorder on the solid component of the supersolid. For large hopping, \(t\mathop{ > }\limits_{ \tilde {}}0.035\), the disorder suppresses the \({T}_{c}\) of the solid because the disorder destroys translational invariance required by the solid. But for low t, the solid is more robust and weak disorder does not significantly impact \({T}_{c}\) of the solid. There is therefore a narrow regime (we find it to be near \(t\approx 0.03\)) where the \({T}_{c}\) of the solid component of the supersolid is not significantly impacted by disorder.
The superfluid component of the supersolid, on the other hand, can be increased by disorder. Previous work looking at the ordinary BoseHubbard model (\(V=0\)) found that the \({T}_{c}\) of the superfluid can be increased by disorder^{26,34,40,43}. The mechanism required disorder to create pathways for the superfluid to percolate across the entire sample. The pathways enlarged the phase space for superfluidity, and therefore \({T}_{c}\).
The combined effects of a stable solid with enhanced superfluidity leads to an enhanced \({T}_{c}\) for the supersolid with disorder. To see this in QMC we use finitesize scaling to show that disorder can increase the critical temperature of the supersolid phase in the thermodynamic limit in a narrow parameter window. We extract critical temperatures for the supersolid for various disorder strengths (\({\rm{\Delta }}=\mathrm{0,\; 0.1,\; 0.2,}\) and \(0.3\)). Finitesize scaling is performed for lattice sizes \(L=\mathrm{6,\; 8,\; 10}\). Figure 8 shows an enhancement of the critical temperature for the supersolid phase from \({T}_{c} \sim 0.02\) to \({T}_{c} \sim 0.06\) as we increase the disorder strength from \({\rm{\Delta }}=0.0\) to \({\rm{\Delta }}=0.3\) at \(t=0.033\). An approximate 3fold increase of the supersolid critical temperature is achieved by increasing disorder in a narrow window of t. Figure 8 is consistent with previous results^{26} but carries the calculation into the thermodynamic limit with an explicit calculation of \({T}_{c}\). \({T}_{c}\) drops quickly for larger disorder strengths.
The disorder enhanced supersolid can also be understood in a meanfield percolation picture^{26}. Consider the pure solid phase near the solidsupersolid phase boundary in the absence of disorder. The gap in the solid phase prevents density fluctuations and therefore suppresses intersite tunneling needed for concomitant superfluidity. The addition of site disorder allows tunneling between sites with sufficiently strong disorder. If the collection of bonds allowing tunneling percolates across the sample, then a superfluid forms. In this way the superfluid has been found to be triggered by the addition of disorder in BoseHubbard models^{33,34,43,44,45}. But here the background solid remains intact leading to a supersolid that has been triggered by the addition of disorder.
Discussion
We have used quantum Monte Carlo to study the extended BoseHubbard model with disorder on the simple cubic lattice. We have computed the finite temperature phase diagram at fixed chemical potential. We find that disorder lowers the maximum critical temperature of the supersolid. But our results show that disorder and thermal fluctuations still allow the supersolid phase, in contrast to lower dimensions where disorder suppresses the supersolid^{25}. We have also found that in a narrow parameter regime, the critical temperature of the supersolid is actually enhanced by disorder where the disorder opens percolating pathways to strengthen superfluidity. Overall, our results show that in 3D the supersolid is more robust than in lower dimensions^{22}.
Methods
We solve Eq. (1) using a numerically exact QMC method: the Stochastic Series Expansion representation with directed loop updates^{46,47}. Various physical quantities, either diagonal or offdiagonal, can be calculated according to the path integral formulation of the QMC simulations. Our results are converged with respect to truncation of the boson number, the number of QMC steps, and the number of disorder profiles. Our estimates of order parameters are therefore exact to within Monte Carlo error. We have also checked that our implementation of the Stochastic Series Expansion algorithm produces the same results as the ALPS implementation^{48}.
Disorder averaging is a key part of the numerical procedure. We perform several runs over distinct disorder profiles to ensure proper averaging. To ensure convergent disorder averages, we typically run 1000 QMC simulations with different disorder realizations for each set of parameters. We then plot histograms for the resulting measurement of various physical quantities.
We find three types of distributions in our disorder averaging. The most common distribution is a single Gaussian peak without any “fat” tails in the distribution curve. This type of distribution signifies a unique phase for the parameter set. A Gaussian distribution offers fast convergence with respect to the number of disorder realizations.
We also find doublepeaked Gaussian distributions at low \(T\) and large systems, \(L\ge 10\). Our QMC simulations usually end up in one of the two phases depending on the initial configuration. In this case, numerical data are sorted according to the two phases and separate averages need to be done, one for each phase. We choose the phase with the lowest free energy. It is worth noting that twopeak distribution does not necessarily imply the coexistence of two phases. Instead we believe that the twopeak structure is due to trapping in a free energy local minimum (the small peak in the distribution, usually less than 5% of the disorder samples). Updates are then unable to find a path to the free energy global minimum (the large peak in the distribution). We have checked our calculations against ALPS code^{48}, and found that ALPS exhibits the same trapping.
The third type of distribution is a single Gaussian peak but with a “fat” tail^{40}. This happens in the Boseglass phase, where our order parameters do not assume a definite value. In this case physical quantities will have a slow convergence rate with respect to disorder configurations^{40}.
The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.
Change history
14 March 2018
A correction to this article has been published and is linked from the HTML and PDF versions of this paper. The error has not been fixed in the paper.
References
Chester, G. V. Speculations on BoseEinstein condensation and quantum crystals. Phys. Rev. A 2, 256 (1970).
Boninsegni, M. & Prokofev, N. V. Colloquium: Supersolids: What and where are they? Rev. Mod. Phys. 84, 759 (2012).
Kim, E. & Chan, M. Probable observation of a supersolid helium phase. Nature 427, 225 (2004).
Kim, D. Y. & Chan, M. H. W. Absence of supersolidity in solid helium in porous vycor glass. Phys. Rev. Lett. 109, 155301 (2012).
Matsubara, T. & Matsuda, H. A lattice model of liquid helium, i. Prog. of Theor. Phys. 16, 569 (1956).
Matsuda, H. & Tsuneto, T. Offdiagonal longrange order in solids. Prog. Theor. Phys. Suppl. 46, 411 (1970).
Liu, K.S. & Fisher, M. E. Quantum lattice gas and the existence of a supersolid. J. Low Temp. Phys. 10, 655 (1972).
Goral, K., Santos, L. & Lewenstein, M. Quantum phases of dipolar bosons in optical lattices. Phys. Rev. Lett. 88, 170406 (2002).
Buchler, H. & Blatter, G. Supersolid versus phase separation in atomic BoseFermi mixtures. Phys. Rev. Lett. 91, 130404 (2003).
Scarola, V. W. & Das Sarma, S. Quantum phases of the extended BoseHubbard Hamiltonian: Possibility of a supersolid state of cold atoms in optical lattices. Phys. Rev. Lett. 95, 033003 (2005).
Scarola, V. W., Demler, E. & Das Sarma, S. Searching for a supersolid in coldatom optical lattices. Phys. Rev. A 73, 051601 (2006).
Bloch, I., Dalibard, J. & Zwerger, W. Manybody physics with ultracold gases. Rev. Mod. Phys. 80, 885 (2008).
Baier, S. et al. Extended BoseHubbard models with ultracold magnetic atoms. Science 352, 201 (2016).
van Otterlo, A. et al. Quantum phase transitions of interacting bosons and the supersolid phase. Phys. Rev. B 52, 16176 (1995).
Roddick, E. & Stroud, D. Supersolid phases in underdamped Josephson arrays: Quantum Monte Carlo simulations. Phys. Rev. B 51, 8672 (1995).
Micnas, R., Ranninger, J. & Robaszkiewicz, S. Superconductivity in narrowband systems with local nonretarded attractive interactions. Rev. Mod. Phys. 62, 113 (1990).
Blanton, S. H. et al. Infrared study of Ba _{1−x} K _{ x } BiO _{3} from chargedensitywave insulator to superconductor. Phys. Rev. B 47, 996–1001 (1993).
Baumann, K., Guerlin, C., Brennecke, F. & Esslinger, T. Dicke quantum phase transition with a superfluid gas in an optical cavity. Nature 464, 1301–1306 (2010).
Mottl, R. et al. Rotontype mode softening in a quantum gas with cavitymediated longrange interactions. Science 336, 1570–1573 (2012).
Leonard, J., Morales, A., Zupancic, P., Esslinger, T. & Donner, T. Supersolid formation in a quantum gas breaking continuous translational symmetry. Nature 543, 87 (2017).
Li, J.R. et al. A tripe phase with supersolid properties in spinorbitcoupled boseeinstein condensates. Nature 543, 7643 (2017).
Sengupta, P., Pryadko, L. P., Alet, F., Troyer, M. & Schmid, G. Supersolids versus phase separation in twodimensional lattice bosons. Phys. Rev. Lett. 94, 207202 (2005).
Wessel, S. & Troyer, M. Supersolid hardcore bosons on the triangular lattice. Phys. Rev. Lett. 95, 127205 (2005).
Bernardet, K., Batrouni, G. & Troyer, M. Destruction of diagonal and offdiagonal longrange order by disorder in twodimensional hardcore boson systems. Phys. Rev. B 66, 054520 (2002).
Imry, Y. & Ma, S. Randomfield instability of ordered state of continuous symmetry. Phys. Rev. Lett. 35, 1399 (1975).
Kemburi, B. M. & Scarola, V. W. Percolationenhanced supersolids in the extended BoseHubbard model. Phys. Rev. B 85, 020501 (2012).
Yamamoto, K., Todo, S. & Miyashita, S. Successive phase transitions at finite temperatures toward the supersolid state in a threedimensional extended BoseHubbard model. Phys. Rev. B 79, 094503 (2009).
Ohgoe, T., Suzuki, T. & Kawashima, N. Commensurate supersolid of threedimensional lattice bosons. Phys. Rev. Lett. 108, 185302 (2012).
Anderson, P. Theory of dirty superconductors. J. Phys. Chem. Solids 11, 26 (1959).
Kuchinskii, E. Z., Kuleeva, N. A. & Sadovskii, M. V. Attractive hubbard model with disorder and the generalized anderson theorem. Journal of Experimental and Theoretical Physics 120, 1055 (2015).
Fisher, M. P. A., Weichman, P. B., Grinstein, G. & Fisher, D. S. Boson localization and the superfluidinsulator transition. Phys. Rev. B 40, 546 (1989).
Ohgoe, T., Suzuki, T. & Kawashima, N. Groundstate phase diagram of the twodimensional extended bosehubbard model. Phys. Rev. B 86, 054520 (2012).
Pollet, L., Prokof’ev, N. V., Svistunov, B. V. & Troyer, M. Absence of a direct superfluid to Mott insulator transition in disordered Bose systems. Phys. Rev. Lett. 103, 140402 (2009).
Gurarie, V., Pollet, L., Prokof’ev, N. V., Svistunov, B. V. & Troyer, M. Phase diagram of the disordered BoseHubbard model. Phys. Rev. B 80, 214519 (2009).
Meldgin, C. et al. Probing the bose glass–superfluid transition using quantum quenches of disorder. Nature Physics 12, 646 (2016).
Pollock, E. L. & Ceperley, D. M. Pathintegral computation of superfluid densities. Phys. Rev. B 36, 8343 (1987).
Pollet, L., Prokof’ev, N., Svistunov, B. V. & Troyer, M. Absence of a direct superfluid to Mott insulator transition in disordered Bose systems. Phys. Rev. Lett. 103, 140402 (2009).
Privman, V. & Fisher, M. E. Universal critical amplitudes in finitesize scaling. Phys. Rev. B 30, 322 (1984).
Cha, M.C., Fisher, M. P. A., Girvin, S. M., Wallin, M. & Young, A. P. Universal conductivity of twodimensional films at the superconductorinsulator transition. Phys. Rev. B 44, 6883 (1991).
Lin, F., Sorensen, E. S. & Ceperley, D. M. Superfluidinsulator transition in the disordered twodimensional Bosehubbard model. Phys. Rev. B 84, 094507 (2011).
Fisher, M. P. A. Quantum phase transitions in disordered twodimensional superconductors. Phys. Rev. Lett. 65, 923–926 (1990).
Binder, K. & Landau, D. P. Phase diagrams and critical behavior in ising square lattices with nearest and nextnearestneighbor interactions. Phys. Rev. B 21, 1941 (1980).
Krauth, W., Trivedi, N. & Ceperley, D. Superfluidinsulator transition in disordered boson systems. Phys. Rev. Lett. 67, 2307–2310 (1991).
Sheshadri, K., Krishnamurthy, H. R., Pandit, R. & Ramakrishnan, T. V. Percolationenhanced localization in the disordered bosonic Hubbard model. Phys. Rev. Lett. 75, 4075 (1995).
Dang, L., Boninsegni, M. & Pollet, L. Disorderinduced superfluidity. Phys. Rev. B 79, 214529 (2009).
Sandvik, A. W. Stochastic series expansion method with operatorloop update. Phys. Rev. B 59, R14157 (1999).
Syljuasen, O. & Sandvik, A. Quantum Monte Carlo with directed loops. Phys. Rev. E 66, 046701 (2002).
Bauer, B. et al. The ALPS project release 2.0: open source software for strongly correlated systems. J. Stat. Mech: Theory Exp. 5, P05001 (2011).
Acknowledgements
V.W.S. acknowledges support from AFOSR (FA95501510445) and ARO (W911NF1610182). TAM acknowledges support from the Center for Nanophase Materials Sciences, which is a DOE Office of Science User Facility.
Author information
Authors and Affiliations
Contributions
F.L. performed numerical simulations. All authors analyzed the results and reviewed the manuscript.
Corresponding author
Ethics declarations
Competing Interests
The authors declare that they have no competing interests.
Additional information
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
A correction to this article is available online at https://doi.org/10.1038/s41598018230024.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Lin, F., Maier, T.A. & Scarola, V.W. Disordered Supersolids in the Extended BoseHubbard Model. Sci Rep 7, 12752 (2017). https://doi.org/10.1038/s41598017130409
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598017130409
This article is cited by

Localisation of weakly interacting bosons in two dimensions: disorder vs lattice geometry effects
Scientific Reports (2019)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.