1: \documentclass[pra,aps,twocolumn]{revtex4}
2:
3: \usepackage{graphicx}% Include figure files
4: \usepackage{dcolumn}% Align table columns on decimal point
5: \usepackage{bm}% bold math
6:
7: \newcommand{\virg}[1]{``#1'' }%mette fra virgolette
8: \newcommand{\eq}[1]{Eq.~(\ref{#1})}
9: \newcommand{\fig}[1]{Fig.~\ref{#1}}
10: \newcommand{\ket}[1]{|#1\rangle}
11: \newcommand{\bra}[1]{\langle#1|}
12:
13: \begin{document}
14: \title{How the effective boson-boson interaction works in Bose-Fermi mixtures in
15: periodic geometries}
16: \author{G. Mazzarella}
17: \affiliation{Dipartimento di Fisica "G.Galilei", Universit\`{a}
18: degli Studi di Padova, Via F.Marzolo, 8, I-35131 Padova, Italy}
19: \begin{abstract}
20: We study mixtures of spinless bosons and not spin-polarized
21: fermions loaded in two dimensional optical lattices. We approach the problem of
22: the ground state stability within the framework of the linear response theory; by the mean of an
23: iterative procedure, we are able to obtain a relation for the dependence of boson-boson effective interaction on the absolute temperature of the sample.
24: Proceeding from such a formula, we write down analyitical expressions for
25: Supersolid (SS) and Phase Separation (PS) transition temperatures, and plot the phase diagrams.
26: \end{abstract}
27: \maketitle
28: \section{Introduction}
29: The possibility to achieve very low temperatures and the
30: feasibility in the laboratory of implementing optical lattices
31: represent the proper arena where testing the validity of certain
32: condensed-matter theories and observing the manifestations of
33: effects related to quantum mechanical statistics. Within the
34: interesting sinergy between quantum atom optics and condensed
35: matter physics, recent trends are study of the superfluid to
36: Mott-insulator phase transition in bosonic systems, (see
37: \cite{Jaksch} and \cite{Greiner}), and the striving for the
38: realization of a BCS-type condensate in a fermionic system
39: (\cite{Houbiers} and \cite{Heiselberg}). When atomic systems made
40: up of bosons and fermions are considered, a very rich scenario
41: opens up (\cite{Albus}, \cite{Illuminati2}, \cite{Salasnich}). In
42: particular, Bose-Fermi mixtures sympathetically cooled into their
43: quantum degenerate states (\cite{Truscott} and \cite{Roati})
44: exhibit a strong tendency towards Phase Separation (\cite{Molmer}
45: and \cite{Amoruso}), demixing of the bosons and the fermions, and
46: Supersolid (\cite{Meisel}, \cite{Lengua}, \cite{vanotterlo}).
47: Recently, these issues were addressed in \cite{BB}, where spinless
48: bosons and spin-polarized fermions confined in two dimensional
49: ($2$D) periodic geometries were taken into account. As explained
50: in \cite{BB}, the effective interaction between the bosons of the
51: mixture plays a crucial role to the end of determining the
52: boundaries between the various phases of the system.\\
53:
54: The dynamics underlying Phase Separation is driven by a small change
55: ($\delta n_B$) in the bosonic density, which produces a modulation ($\delta n_F$) of the
56: fermionic density, related to $\delta n_B$ by $\delta n_F \sim -U_{BF}
57: N(E_F)\delta n_B$, with $U_{BF}$ the on-site boson-fermion interaction
58: amplitude, and $N(E_F)$ the density of states at
59: the Fermi energy $E_F$ \cite{Mermin}. As consequence of the feedback of the
60: fermionic distortion $\delta n_F$, a shift of the bosonic energy $-U_{BF}^2 N(E_F)(\delta
61: n_B)^2/2$ occurs, thereby inducing an attraction between the bosons with
62: strength $U_{BF}^2 N(E_F)$.
63: Phase Separation emerges when the induced attraction and the intrinsic repulsion
64: between the bosons have the same order of magnitude \cite{BBpra}.\\
65:
66: One of the peculiarities of optical lattices is their periodic
67: arrangement, that is the same of the crystal structure of a solid.
68: The phenomelogical interpretation of particle localization at
69: fixed sites at the base of Mott-insulator mechanism is related
70: just to the crystalline structure . In Supersolid, such a
71: structure is combined with the essence of a superfluid, i.e.
72: stiffnes allowing for particles current to flow without
73: dissipation.\\
74: In general to supersolids are associated two kinds of order, which usually
75: appear in mutual competition. These are the diagonal long-range order
76: (DLRO) associated with the periodic density modulation in a
77: crystal, and the off-diagonal long-range order (ODLRO) associated
78: with the phase order in the condensate \cite{Penrose}. In the strongly
79: interacting case, as in $^{4}$He systems,
80: (\cite{Andreev} and \cite{Leggett}), supersolids have been proposed
81: to exist and analyzed numerically in various model systems
82: describing interacting bosons on a lattice \cite{111213} .\\
83:
84: In this paper, we consider mixtures of spinless bosons and not
85: spin-polarized fermions, so that a non zero $s$-wave interaction
86: between fermions on the same site emerges. The goal of the present
87: work is analyzing the effects of such an interaction on phase
88: diagrams of the system. Approaching the problem within the linear
89: response theory framework \cite{Mermin,Pines,Vignale}, we show
90: that it is possible to calculate in explicit way an effective
91: boson-boson interaction also when the on-site fermion-fermion interaction is taken into account.\\
92:
93: The basic idea relies on sharing tasks between the fermions and the bosons.
94: In particular, the fermions are tuned through a density wave
95: instability establishing crystalline order (DLRO), while the
96: bosons provide the off-diagonal long range order (ODLRO).
97: The interaction between bosons and fermions, and the interaction between the
98: fermions, produce an additional density modulation also in the
99: bosonic density field, hence resulting in a SS phase. To
100: triggering a Density Wave (DW) instability in the fermions, the
101: mixed boson-fermion system is confined to two dimensions and
102: loaded in an optical lattice providing perfect Fermi surface
103: nesting at half-filling \cite{Gruner}.\\
104: The Supersolid transition triggered by the fermions competes with
105: an instability towards Phase Separation in the boson system. Because of the dimensionality
106: and the lattice geometry of our system, the presence of Van Hove
107: singularities (\cite{vanhove} and \cite{Mermin}) -well studied in BCS
108: superconductivity \cite{Scalapino}- strongly enhances the tendency
109: towards Phase Separation and produces new and interesting features
110: in this transition: an arbitrary weak interaction between the
111: bosons and the fermions, and between fermions, is sufficient to
112: drive the Phase Separation at low temperatures. In the following,
113: we investigate the instabilities towards Phase Separation and
114: Density Wave formation. We focus on the weak coupling limit
115: between the bosons and the fermions and between fermions, which
116: excludes a demixing in a repulsive Bose-Fermi system
117: along the lines discussed in \cite{Viverit}.\\
118:
119: The organization of paper is as follows. In the section II, we set
120: the notation, and derive the model Hamiltonian. In the third
121: section, we display the novel result consisting in the analytic
122: formula for the effective boson-boson interaction as a function of
123: the temperature, and discuss the competition between PS and SS; we
124: derive the Phase separation and Density-Wave transition
125: temperatures. In the last section, we write down our conclusions
126: and comment about future perspectives of the topic.
127:
128:
129: \section{The model Hamiltonian}
130: The microscopic Hamiltonian for interacting spinless bosons and
131: not spin-polarized fermionic atoms subject to an optical lattice
132: reads $\hat{H}=\hat{H}_B+\hat{H}_F+\hat{H}_{int}$ (here and in the
133: following, $\alpha=B,F$ and $\sigma={\uparrow,\downarrow}$ denote
134: the atomic species and the spin, respectively) \cite{Albus,
135: Fetter}
136: \begin{eqnarray}
137: \label{fullhamiltonian}
138: &&\hat{H}_{\alpha} =\int d \vec{r} \hat{\Psi}_{\alpha}^{\dagger}\bigg(
139: -\frac{\hbar^{2}}{2
140: m_{\alpha}}\nabla^2+V_{\alpha} \bigg)\hat{\Psi}_{\alpha},\nonumber\\
141: && \hat{H}_{int}=\int d \vec{r} \bigg[\frac{g_{BB}}{2}\hat{\Psi}_{B}^{\dagger}\hat{\Psi}_{B}^{\dagger}
142: \hat{\Psi}_{B}\hat{\Psi}_{B}\nonumber\\
143: &+&g_{BF}(\hat{\Psi}_{B}^{\dagger}\hat{\Psi}_{B}
144: \hat{\Psi}^{\dagger}_{F,\uparrow}\hat{\Psi}_{F,\uparrow}+\hat{\Psi}_{B}^{\dagger}\hat{\Psi}_{B}
145: \hat{\Psi}^{\dagger}_{F,\downarrow}\hat{\Psi}_{F,\downarrow})\nonumber\\
146: &+& \frac{g_{FF}}{2}\sum_{\sigma^{'} \neq \sigma}
147: \hat{\Psi}^{\dagger}_{F,\sigma}
148: \hat{\Psi}^{\dagger}_{F,\sigma^{'}}\hat{\Psi}_{F,\sigma^{'}}\hat{\Psi}_{F,\sigma}\bigg]
149: .\end{eqnarray} Here, we assume a repulsive interaction between
150: the bosons: $g_{BB}=4\pi\hbar^{2}a_{BB}/m_{B}>0$, where $a_{BB}$
151: is associated the $s$-wave scattering length. The boson-fermion
152: interaction strength $g_{BF}=4\pi\hbar^{2}a_{BF}/m_{R}$ is assumed
153: to be the same for both the spin configurations; here, $a_{BF}$ is
154: the boson-fermion $s$-wave scattering length and
155: $m_{R}=m_{B}m_{F}/(m_{B}+m_{F})$ is the reduced mass. Finally,
156: $g_{FF}=2\pi\hbar^{2}a_{FF}/m_{F}$ with $a_{FF}$ the
157: fermion-fermion $s$-wave scattering length. In the following we
158: always assume both $a_{BF}>0$ and $a_{FF}>0$.
159:
160: The optical lattice with wave length $\lambda$ provides an
161: $a=\lambda/2$- periodic potential for the bosons and fermions with
162: $\displaystyle{V_{B,F}(x,y)=V_{B,F}\big(\sin^2 \frac{\pi
163: x}{a}+\sin^2 \frac{\pi y}{a}\big)}$ \cite{Jessen}, where the
164: lattice depth $V_{F}$ is assumed the same for both the spin
165: configurations. In the experiments, the $2$D setup is realized by
166: generating an anisotropic three-dimensional optical potential
167: $\displaystyle{V_{B,F}(x,y)=V_{B,F}\big(\sin^2 \frac{\pi
168: x}{a}+\sin^2 \frac{\pi y}{a}\big)+V^{z}_{B,F}\sin^2 \frac{\pi
169: z}{a_z}}$ with $V^{z}_{B,F} \gg V_{B,F}$; then
170: the interplane hopping is quenched.\\
171: Due to the strong localization around each lattice site
172: $\vec{r}_i$, the annihilation (creation) bosonic and fermionic
173: field operators $\hat{\Psi}_{\alpha}$
174: ($\hat{\Psi}^{\dagger}_{\alpha}$) may be expanded in terms of the
175: Wannier functions $w^{l}_{\alpha}(\vec{r}-\vec{r}_i)$, with $l$
176: the Bloch band index \cite{Jessen}
177: \begin{equation}
178: \label{wannierexpansionB} \hat{\Psi}_{\alpha}(\vec{r}) \, = \,
179: \sum_{i,l} \hat{a}^{l}_{i} w^{l}_{\alpha}(\vec{r}-\vec{r}_i) \; ,
180: \end{equation}
181: where $\hat{a}^{l}_{i}$ is the bosonic ($\hat{b}^{l}_i$) or
182: fermionic ($\hat{c}^{l}_{\sigma,i}$) annihilation operator acting
183: on a particle at the \mbox{$i$th} lattice site and in the $l$th
184: Bloch band. For a strong optical lattice
185: $V_{B,F}>E^{r}_{B,F}=2\hbar^2\pi^2/\lambda^2 m_{B,F}$ the
186: restriction to the lowest Bloch band ($l=0$) is justified
187: \cite{Jaksch}. The translationally invariant lattice version of
188: Hamiltonian (\ref{fullhamiltonian}) reads
189: %such that $L_{x} \equiv L_{\perp}\sqrt{\lambda} \gg a M$, with $M$
190: %denoting the total number of lattice sites. As a consequence, the
191: %local density approximation (LDA) can be applied. Therefore, when
192: %exploiting the expansion \eq{wannierexpansion} to map the full
193: %microscopic Hamiltonian \eq{fullhamiltonian} into its lattice
194: %version, we will discard all terms that are of order
195: %$(aM/L_{x})^2$ or higher. Qualitatively, this means neglecting
196: %those nonlocal effects that are induced by the presence of the
197: %trapping potentials, such as site-dependent hopping terms. The
198: %latter can become important in regions of the lattice very far out
199: %of the central core of the harmonic trap. However, the typical
200: %experimental situations involve only that part of the lattice that
201: %lies well inside the central core of the slowly varying confining
202: %potential \cite{Ferlaino}.
203: \begin{widetext}
204: \begin{equation}
205: \label{latticeh1} \hat{H} \, = \, -J_{B}\sum_{<i,j>}
206: \,\hat{b}^{\dagger}_{i} \hat{b}_{j} \,
207: -J_F\sum_{<i,j>,\sigma} \,\hat{c}^{\dagger}_{i,\sigma}
208: \hat{c}_{j,\sigma}+ \, \frac{U_{BB}}{2}\sum_{i}
209: \hat{n}_{i}(\hat{n}_{i}-1)+ U_{BF}\sum_{i,\sigma}
210: (\hat{n}_{i}\hat{m}_{i,\sigma})+ U_{FF}\sum_{i}
211: (\hat{m}_{i,\uparrow}\hat{m}_{i,\downarrow})+\delta\sum_{i}(\hat{m}_{i,\uparrow}-\hat{m}_{i,\downarrow})
212: \; ,\end{equation}
213: \end{widetext}
214: where we have omitted the band index $l=0$. The symbol $<i,j>$
215: denotes couples of nearest-neighbor lattice sites, and $\delta$
216: the imbalance between spin-up and spin-down fermions;
217: $\hat{n}_{i}=\hat{b}^{\dagger}_{i}\hat{b}_{i}$ and
218: $\hat{m}_{i,\sigma}=\hat{c}^{\dagger}_{i,\sigma}\hat{c}_{i,\sigma}$
219: are the number operators for bosons and fermions with spin
220: $\sigma$ at the $i$th site, respectively. The boson-boson,
221: boson-fermion, and fermion-fermion interaction amplitudes are
222: $U_{BB}=g_{BB}\int d \vec{r}|w_{B}(\vec{r})|^4 $,
223: $U_{BF}=g_{BF}\int d \vec{r}|w_{B}(\vec{r})|^2
224: |w_{F}(\vec{r})|^2$, and $U_{FF}=g_{FF}\int d \vec{r}
225: |w_{F,\uparrow}(\vec{r})|^2 |w_{F,\downarrow}(\vec{r})|^2$,
226: respectively \cite{Jaksch,Albus,BB}.
227:
228: Among the adavantages provided by optical lattices, there is the
229: possibility of tuning the Hamiltonian parameters in such a way to
230: realize different interaction regimes. Here we focus on weak
231: coupling regime, which takes place when $\lambda_{BF}=U_{BF}^2
232: N_0/U_BB<<1$ ($N_0=1/2\pi^2 J_F$) and $t_B=8 J_B/n_{B} U_{BB} >>
233: \lambda_{BF} $, where $n_B$ is the
234: bosonic filling factor. \\
235: The parameters involved in the Hamiltonian (\ref{latticeh1})
236: are related to characteritistic quantities of the optical potential according to
237: \begin{widetext}
238: \begin{eqnarray}
239: \label{amplitudes}
240: J_{B,F}=(4\sqrt{\pi})E^{r}_{B,F}V_{B,F}^{3/4}\exp(-2\sqrt{V_{B,F}});\nonumber\\
241: \frac{U_{BF}}{E_{F}^{r}}=8\sqrt{\pi}\frac{1+m_F/m_B}{1+\sqrt{V_F/V_B}}\frac{a_{BF}}{\lambda\gamma}(V_{F}^z)^{1/4}(V_{F})^{1/2};\nonumber\\
242: \frac{U_{BB}}{E_{B}^{r}}=4\sqrt{2\pi}\frac{a_{BB}}{\lambda\gamma}(V_{B}^z)^{1/4}(V_{B})^{1/2};\nonumber\\
243: \frac{U_{FF}}{E_{F}^{r}}=4\sqrt{2\pi}\frac{a_{FF}}{\lambda\gamma}(V_{F}^z)^{1/4}(V_{F})^{1/2}.\end{eqnarray}
244: \end{widetext}
245: where $\gamma=2 a_z/\lambda$. The hopping and the atom-atom
246: interaction amplitudes (\ref{amplitudes}) are evaluated by
247: extending to our case the calculations performed in
248: \cite{Jaksch,Albus,BB}.\\
249:
250: By exploiting the discrete Fourier transform of $\hat{a}_{i}$ and of its
251: Hermitian conjugate, the Hamiltonian (\ref{latticeh1}) may be written in the
252: momenta space
253: \begin{widetext}
254: \begin{eqnarray}
255: \label{lattice h}
256: \hat{H}=\sum_{\vec{k}}\big[\epsilon_{B,\vec{k}}\hat{b}^{\dagger}_{\vec{k}}\hat{b}_{\vec{k}}+(\epsilon_{\uparrow,\vec{k}}-\delta)
257: \hat{c}_{\uparrow,\vec{k}}^{\dagger}\hat{c}_{\uparrow,\vec{k}}+(\epsilon_{\downarrow,\vec{k}}+\delta)\hat{c}_{\downarrow,\vec{k}}^{\dagger}\hat{c}_{\downarrow,\vec{k}}\big]+\nonumber\\
258: \frac{1}{M^2}\sum_{\vec{k}}\big[\frac{U_{BB}}{2}\hat{n}_{B,\vec{k}}\hat{n}_{B,-\vec{k}}+U_{BF}\hat{n}_{B,\vec{k}}\hat{m}_{F,-\vec{k}}+
259: U_{FF}\hat{m}_{\uparrow,\vec{k}}\hat{m}_{\downarrow,-\vec{k}}]
260: ,\end{eqnarray}
261: \end{widetext} where $M$ is the number of lattice sites in $x$ ($y$) direction. We assume the wave vector $\vec{k}$ be
262: restricted to the first Brillouin zone: $k_x\in [-\pi/a,\pi/a]$,
263: $k_y\in [-\pi/a,\pi/a]$. The density number operators are
264: $\hat{n}_{B,\vec{k}}=\sum_{\vec{p}}\hat{b}_{\vec{k}+\vec{p}}^{\dagger}\hat{b}_{\vec{p}}$,
265: $\hat{m}_{\sigma,\vec{k}}=\sum_{\vec{p}}\hat{c}_{\sigma,\vec{k}+\vec{p}}^{\dagger}\hat{c}_{\sigma,\vec{p}}$
266: \cite{Vignale};
267: the bosonic and fermionic dispersion relations $\epsilon_{B,\vec{k}}$ and
268: $\epsilon_{\sigma,\vec{k}}$ read
269: $\epsilon_{B,\vec{k}}=- 2J_{B}(2-\cos(k_x a)-\cos(k_ya))$ and
270: $\epsilon_{\sigma,\vec{k}}=-2J_{F,\sigma}(\cos(k_x
271: a)-\cos(k_ya))$ \cite{Fetter}. \\
272:
273: \section{Phase separation versus Supersolid}
274: In this section, we want to gain analytical insight into stability of the mixture ground state.
275: As explained in \cite{BB}, to achieve such a goal, the starting point is the
276: derivation of an effective Hamiltonian for the bosons alone, by keeping in mind
277: that fermions enters the description of our system via a modified interaction
278: between the bosons themselves. Within linear response theory \cite{Mermin,Vignale} the boson
279: density $n_{B}(\vec{q})$ (for the sake of simplicity, we
280: refer to $n_{B} (\vec{q}) (m_{\sigma} (\vec{q}))$ as to the
281: induced perturbation) drives the fermionic system trough the
282: following modulation of the density
283: \begin{eqnarray}
284: \label{lrt}
285: <m_{\sigma,\vec{q}}>=\chi_{\sigma}(U_{BF}n_{B,\vec{q}}+U_{FF}<m_{-\sigma,\vec{q}}>).\nonumber\\
286: \end{eqnarray}
287: The function $\chi_{\sigma}$, for the $\sigma$ component of the spin, is
288: the Lindhard function depending upon the absolute temperature $T$
289: and on the wave vector $\vec{q}$ \cite{Mermin,Vignale}:
290: \begin{eqnarray}
291: \label{Lindhard} \chi_{\sigma}(T,\vec{q})=\frac{1}{2}\int
292: \frac{d\vec{k}}{v_0}\frac{f(\epsilon_{\sigma,\vec{k}})-f(\epsilon_{\sigma,\vec{k}+\vec{q}})}{\epsilon_{\sigma,\vec{k}}-\epsilon_{\sigma,\vec{k}+\vec{q}}},
293: \end{eqnarray}
294: where $v_0=(2\pi/a)^{2}$ is the volume of the first Brillouin
295: zone; the integration is performed over this region. The
296: temperature $T$ enters via the Fermi distribution function
297: $\displaystyle{f(\epsilon_{\sigma,\vec{k}})=1/\big((1+\exp(\frac{\epsilon_{\sigma,\vec{k}}-\mu_F}{T})\big)}$,
298: with $\mu_F$ the chemical potential of the fermionic atoms.\\
299: We analyze the behavior of the system for temperatures well below
300: the the superfluid transition temperature $T_{KT}$
301: (Kosterlitz-Thouless). Hence, the mixture is enhanced in a
302: sufficiently low temperature regime so that
303: the fermionic chemical potential $\mu_F$ my be safely identified with the Fermi
304: energy $E_F$.\\
305:
306: Proceeding form the Hamiltonian (\ref{lattice h}), we integrate
307: out the fermionic freedom degrees following the same path as in
308: \cite{BB}. Within the procedure of tracing out the fermions, we
309: treat the spin-up component independently from the other, i.e. we
310: perform a mean-field approximation. We get the Hamiltonian
311: \begin{widetext}
312: \begin{eqnarray}
313: \label{effectivehamiltonian}
314: \hat{H}^{int}_{eff}=\frac{1}{M^2}\big[\sum_{\vec{k}}\frac{U_{BB}}{2}\hat{n}_{B,\vec{k}}\hat{n}_{B,-\vec{k}}+
315: U_{BF}(\hat{n}_{B,\vec{k}}<\hat{m}_{\uparrow,-\vec{k}}>+\hat{n}_{B,\vec{k}}<\hat{m}_{\downarrow,-\vec{k}}>)+
316: U_{FF}<\hat{m}_{\uparrow,\vec{k}}><\hat{m}_{\downarrow,-\vec{k}}>\big],\end{eqnarray}\end{widetext}
317: which describes the effective interaction between the bosons of the mixture.
318: By employing the rules summarized in Eq. (\ref{lrt}) in the Hamiltonian
319: (\ref{effectivehamiltonian}), we obtain
320: \begin{widetext}
321: \begin{eqnarray}
322: \label{recursion}
323: \hat{H}^{int}_{eff}=\frac{1}{M^2}\sum_{\vec{k}}\big[U_{BF}\big(\hat{n}_{B,\vec{k}}(\chi_{\uparrow}U_{BF}\hat{n}_{B,-\vec{k}}+\chi_{\uparrow}U_{FF}<\hat{m}_{\downarrow,-\vec{k}}>\big)+\nonumber\\
324: U_{BF}\big(\hat{n}_{B,\vec{k}}(\chi_{\downarrow}U_{BF}\hat{n}_{B,-\vec{k}}+\chi_{\downarrow}U_{FF}<\hat{m}_{\uparrow,-\vec{k}}>\big)+\nonumber\\
325: U_{FF}\big(\chi_{\uparrow}U_{BF}\hat{n}_{B,\vec{k}}+\chi_{\uparrow}U_{FF}<\hat{m}_{\downarrow,\vec{k}}>)
326: (\chi_{\downarrow}U_{BF}\hat{n}_{B,-\vec{k}}+\chi_{\downarrow}U_{FF}<\hat{m}_{\uparrow,-\vec{k}}>)\big]
327: .\end{eqnarray}
328: \end{widetext}
329: We exploit the rules (\ref{lrt}) in Eq. (\ref{recursion}) in
330: iterative way. Then, the Hamiltonian (\ref{recursion}) can be
331: expressed as an expansion in series of powers of
332: $|\chi_{\sigma}U_{FF}|$; we have verified that this last quantity
333: is much smaller than one. We assume to be in a situation in which
334: the spin-up population is equal to the spin-down one ($\delta
335: =0$), and in which the fermions are in half-filling configuration,
336: $E_F=0$ and $m_{\uparrow}=m_{\downarrow}=1/4$. In such a situation
337: the series associated to the Hamiltonian (\ref{recursion}) may be
338: summed. We employ the fact that when $\delta =0$, is
339: $\epsilon_{\uparrow}=\epsilon_{\downarrow} \equiv \epsilon_F$ and
340: then $\chi_{\uparrow}=\chi_{\downarrow} \equiv \chi$, with
341: \begin{eqnarray}
342: \label{Lindhardbalance} \chi(T,\vec{q})=\frac{1}{2}\int
343: \frac{d\vec{k}}{v_0}\frac{f(\epsilon_{F,\vec{k}})-f(\epsilon_{F,\vec{k}+\vec{q}})}{\epsilon_{F,\vec{k}}-\epsilon_{F,\vec{k}+\vec{q}}},
344: \end{eqnarray}
345: Together with these two last properties, we use the symmetry
346: $\epsilon_{F,\vec{k}}=\epsilon_{F,-\vec{k}}$; then, the effective
347: boson-boson interaction (\ref{effectivehamiltonian}) reads
348: \begin{widetext}
349: \begin{eqnarray}
350: \label{recursionbis}
351: \hat{H}^{int}_{eff}=\frac{1}{2M^2}\sum_{\vec{k}}\bigg[U_{BB}+2\chi(T,\vec{q})U_{BF}^{2}\big(\frac{1}{1-\chi(T,\vec{q})
352: U_{FF}}+\frac{U_{FF}}{(\chi(T,\vec{q})^2
353: U_{FF}-1)^2}\big)\bigg]\hat{n}_{B,\vec{k}}\hat{n}_{B,-\vec{k}}
354: .\end{eqnarray}\end{widetext} The effective boson-boson
355: interaction $U_{eff}(T)$ depends on the temperature according the
356: formula
357: \begin{eqnarray}
358: \label{Ueff}
359: &&U_{eff}(T)=U_{BB}+2\chi(T,\vec{q})U_{BF}^{2}\big(\frac{1}{1-\chi(T,\vec{q})
360: U_{FF}}\nonumber\\
361: &+&\frac{\chi(T,\vec{q})U_{FF}}{(\chi(T,\vec{q})
362: U_{FF}-1)^2}\big).\end{eqnarray} The novelty of the present work
363: relies on Eq. (\ref{Ueff}). Such a formula represents a very
364: useful tool for investigating in analytical way the instability
365: related to the Van Hove singularity and the one associated to the
366: Density Wave. In particular, we are interested in calculating the
367: temperatures of transition to PS and to DW phases.\\
368: To this end, let us focus on fermionic dispersion relation
369: $\epsilon_{F,\vec{k}}=-2 J_{F}[\cos(k_xa)+\cos(k_ya)]$. By
370: analyzing $\epsilon_{F,\vec{k}}$, we realize that the Lindhard
371: function (\ref{Lindhardbalance}) exhibits two logarithmic
372: singularities. These two singularities give rise to instabilities
373: in the system. In particular, the singularity at $\vec{q}=\vec{0}$
374: induces an instability towards Phase Separation. On the other
375: hand, in correspondence of the wave vector
376: $\vec{k}_{DW}=(\pi/a,\pi/a)$ -which joints the saddle points (SP)
377: $\vec{k}_{SP}=(0,\pm \pi/a),(\pm \pi/a,0)$ of the fermionic
378: dispersion relation- the fermionic energy vanishes. The wave
379: vector $\vec{k}_{DW}$ drives a
380: Density Wave, responsible for the Supersolid order. \\
381: Let us focus on the logarithmic Van Hove singularity, and analyze
382: the density states. Due to the balance between the spin-up and the
383: spin-down populations, we have that
384: $N_{\uparrow}(\epsilon)=N_{\downarrow}(\epsilon)\equiv
385: N(\epsilon)$. Within the tight-binding regime, $N(\epsilon)$ reads
386: \cite{BB,Mermin}
387: \begin{eqnarray}
388: \label{densityofstates}
389: N(\epsilon)=\frac{1}{2}N_{0}K[\sqrt{1-\frac{(\epsilon \pm
390: \delta)^2}{16 J_F^2}}]\sim \frac{N_0}{2} \ln|\frac{16
391: J_F}{(\epsilon \pm \delta)}|,\end{eqnarray} with $K[k]$ the
392: complete elliptic integral of the first kind \cite{Abr}.\\
393: We write the response function (\ref{Lindhardbalance}) in the
394: energy space \cite{Mermin}, and analyze it at $\vec{q}=\vec{0}$
395: and in the limit of zero temperature
396: \begin{eqnarray}
397: \label{Lindhardenergy} \chi(T \to 0,0)=\int_{0}^{8J_F} d\epsilon
398: N(\epsilon)_{\delta=0}\partial_{\epsilon}
399: f(\epsilon)_{E_F=0}\nonumber\\
400: \sim -\frac{N_0}{2}\ln(\frac{16 c_1 J_F }{T}),\end{eqnarray} where
401: $8 J_F$ is the bandwidth, and the subscripts $\delta=0$ and
402: $E_F=0$ denote the absence of population imbalances and the
403: half-filled band situation, respectively; $c_1=(2\exp(C))/\pi
404: \approx 1.13$ with $C$ the Eulero constant ($\approx 0.577...$).
405: The coupling between the bosons and the fermions induces an
406: attraction between the bosons, and the effective long distance
407: scattering parameter takes the form
408: \begin{eqnarray}
409: \label{scatteringlongdistance}
410: U_{eff}(T)=U_{BB}+2\chi(T,0)U_{BF}^{2}\big(\frac{1}{1-\chi(T,0)
411: U_{FF}}+\nonumber\\\frac{\chi(T,0)U_{FF}}{(\chi(T,0)
412: U_{FF}-1)^2}\big).
413: \end{eqnarray}
414: The thermodynamic stability of a superfluid condensate at low
415: temperatures requires a positive effective interaction, i.e.
416: $U_{eff}(T)>0$ as commented in \cite{Abrikosov}. The condition
417: $U_{eff}(T_{PS})=0$ defines the critical temperature $T_{PS}$ for
418: Phase Separation
419: \begin{widetext}
420: \begin{eqnarray}
421: \label{PStemperature} T_{PS}=16 c_1 J_{F} \exp[\bigg(\frac{2}{N_0
422: U_{FF}-(1+\sqrt{1-\frac{2 N_0
423: U_{FF}}{\lambda_{BF}}})\lambda_{BF}}\bigg)].\end{eqnarray}\end{widetext}
424: \begin{widetext}
425: Let us focus, now, on the instability in the system triggered by
426: $\vec{k}_{DW}$. By exploiting the symmetry
427: $\epsilon_{F,\vec{q}+\vec{k}_{DW}}=-\epsilon_{F,\vec{q}}$, the
428: energy representation of the response function, in the energy
429: representation, reads
430: \begin{eqnarray}
431: \label{Lindharddensitywave} \chi(T,\vec{k}_{DW})=\int_{0}^{8 J_F}
432: d \epsilon
433: N(\epsilon)\frac{\tanh(\epsilon/2 T)}{-2\epsilon}\nonumber\\
434: \sim-\frac{N_0}{4} \big(\ln(\frac{16 c_1
435: J_F}{T})\big)^{2}.\end{eqnarray} Within Bogoliubov theory, the
436: bosonic quasiparticle spectrum is \cite{Abrikosov}
437: \begin{eqnarray}
438: \label{spectrum}
439: E_{B}(\vec{q})=\sqrt{\epsilon_B(\vec{q})^2+2n_B\epsilon_B(\vec{q})[U_{eff}(T)]}.\end{eqnarray}
440: The induced attraction between the bosons reduces the energy of
441: quasiparticle providing a roton minimum at $\vec{k}_{DW}$, which
442: vanishes at critical temperature $T_{DW}$ given by
443: \begin{eqnarray}
444: \label{DWtemperature} T_{DW}=16 c_1 J_F \exp\bigg[-2
445: \sqrt{\frac{1}{-N_0
446: U_{FF}+\frac{2\lambda_{BF}}{2+t_B}+\frac{2\lambda_{BF}\sqrt{\frac{-N_0
447: (2+t_B)U_{FF}+\lambda_{BF}}{\lambda_{BF}}}}{2+t_B}}}\bigg].\end{eqnarray}\end{widetext}
448: By performing the limit $U_{FF} \to 0$ in Eq.
449: (\ref{PStemperature}) and in Eq. (\ref{DWtemperature}) we are able
450: to reproduce the temperatures signing the Phase Separation and
451: Density Wave calculated in \cite{BB}.\\
452:
453: The fermion-fermion interaction plays a crucial role in
454: determining the order characterizing the ground state of the
455: mixture. The interaction between the fermions influences the
456: behavior of the system via the expression of the effective
457: boson-boson interaction, as summarized in \eq{Ueff}. A non
458: vanishing repulsive fermion-fermion interaction acts in such a way
459: to rising a barrier of potential, which prevents system to achieve
460: Phase Separation. This reflects in a mechanism which advantages
461: the Supersolid order, as shown in the plots reported in Fig. 1
462: ($U_{FF}>0$) and in Fig. 2 ($U_{FF}=0$), that are represented for
463: a mixture made up of bosonic atoms of $^{87}$Rb and of fermionic
464: atoms of $^{40}$Rb in a square optical lattice with $\lambda$
465: equal to $830$nm. From these figures, we see that the PS and SS
466: lines cross in a certain point, characterized by a critical depth.
467: We observe that the PS region of Fig. 1, plotted for $U_{FF}>0$,
468: is smaller than the same region of Fig. 2 obtained with
469: $U_{FF}=0$. The difference between the magnitudes of the two PS zones may be interpreted as
470: a signature of the role of the finite fermion-fermion interaction.
471: The amount of the shift of a PS region with respect to the other might be used as an indirect measure of the fermion-fermion $s$-wave scattering
472: length.\\
473: The role of the temperature is important in determining the kind
474: of order that in the system establishes as well. In fact, when the
475: optical lattice depth $V_{F}$ is greater than its critical value,
476: the greater is the temperature the more the Phase Separation is
477: dominant in the system; below the critical depth, the behavior is
478: reversed. The comparison between the absence of the on-site
479: fermion-fermion interaction and the case in which such an
480: interaction is present, is shown in Fig. 3. Here, we observe that
481: when the fermions of mixture interact with each other, the SS
482: region is wider with respect to $U_{FF}=0$ case. Then, we may
483: deduce that the Fermi-Fermi interaction acts in such a way to make
484: more important the Supersolid order.
485:
486: %\begin{figure}
487: %\centering
488: %\includegraphics[width=8cm]{pdnffbis.eps}
489: %\caption{The points on the grid surface corresponds to Phase Separation, while the points on the contnuous surface
490: %corresponds to Supersolid} \label{int}
491: %\end{figure}.
492:
493: \begin{figure} \centering
494: \includegraphics[width=8cm]{Fdyff.eps}
495: \caption{On the horizontal axis the fermionic optical depth in
496: fermionic recoil energy units; on the vertical axis the
497: temperature normalized to the fermionic recoil energy. The
498: dot-dashed and continuous lines are the PS and DW temperatures,
499: respectively, when $U_{FF} > 0$. The curves are plotted for
500: $a_{FF}=104.8$ a$_0$, with a$_0$ the Bohr radius, $\gamma=3$,
501: $n_B=3/2$, $V_{F}^{z}=20$,$V_{B}/V_{F}=V_{B}^{z}/V_{F}^{z}=7/3$.}
502: \label{globalsecond}
503: \end{figure}
504: \begin{figure}\centering
505: \includegraphics[width=8cm]{Fdnff.eps}
506: \caption{The dashed and dotted lines are the PS and DW
507: temperatures, respectively, when $U_{FF}=0$. Here is, clearly,
508: $a_{FF}=0$, and the other quantities assume the same values as in
509: Fig. 1.} \label{globalfirst}
510: \end{figure}
511: \begin{figure}\centering
512: \includegraphics[width=8cm]{Pd.eps}
513: \caption{The black continuous line represents the separation
514: between SS (the points above the line) and the PS (the points
515: below the line) when $a_{FF}=0$; the dotted line is plotted in
516: with $a_{FF}=104.8$ a$_0$. The other quantities assume the same
517: values as in Fig. 1.} \label{Pd}
518: \end{figure}
519:
520:
521: \section{Conclusions}
522: In this paper, we have investigated the ground state of a
523: Bose-Fermi mixture loaded in a two-dimensional periodic geometry,
524: and made up of spinless bosons and not spin-polarized fermions. We
525: have analyzed the competition between Phase Separation and
526: Supersolid orders. This analysis was carried out within the
527: framework of the linear response theory. By employing an iterative
528: technique, we have obtained an analytical expression for the
529: effective boson-boson interaction as a function of the absolute
530: temperature of the sample. We have performed our study in absence
531: of fermionic population imbalances and in presence of
532: boson-fermion and fermion-fermion repulsive interactions. We have
533: studied the phase diagram of the system by stressing the role of
534: the fermion-fermion interaction in determining the kind of order
535: sustained by the Bose-Fermi mixture Hamiltonian. We have explained
536: the changes that such an interaction introduces in the behavior of
537: the system with respect to the case in which the fermions do not
538: interact on the same lattice site. We have discussed the nice
539: interplay between the critical value of the optical lattice depth
540: for the fermions and the temperature of the sample in
541: establishing the kind of order in the ground state of the mixture.\\
542:
543: The study of such a topic opens up a bunch of very exciting
544: possibilities. In particular, a very interesting relationship
545: could be established between these issues and the ones related to
546: propagation of the zero sound waves both in homogeneous
547: \cite{Pines} and in trapped \cite{Padova} systems of ultracold
548: atoms; for these kind of problems, in fact, the linear response
549: theory is exploited as well.
550:
551: In a forthcoming paper, we wish to accomplish the task of
552: analyzing the case of imbalance between the two fermionic
553: populations and different sign of boson-fermion and
554: fermion-fermion interactions, by employing a suitable modification
555: of the iterative method displayed in the present work.\\
556:
557: This work has been partially supported by Fondazione CARIPARO. Discussions with S.M. Giampaolo are acknowledged.
558:
559: \begin{thebibliography}{99}
560: \bibitem{Jaksch}
561: D. Jaksch {\it et al.} Phys.Rev.Lett. \textbf{81}, 3108 (1998).
562: \bibitem{Greiner}
563: M. Greiner {\it et al.}, Nature \textbf{415} 39 (2002).
564: \bibitem{Albus}
565: A.Albus, F.Illuminati, and J. Eisert, Phys.Rev.A \textbf{68}, 023606 (2003).
566: \bibitem{Illuminati2}
567: F.Illuminati and A.Albus, Phys.Rev.Lett. \textbf{93}, 090406 (2004).
568: \bibitem{Salasnich}
569: L.Salasnich and F.Toigo, Phys.Rev.A \textbf{75}, 013623 (2007).
570: \bibitem{Houbiers}
571: M. Houbiers {\it et al.}, Phys. Rev. A 56, 4864 (1997).
572: \bibitem{Heiselberg}
573: H. Heiselberg {\it et al.}, Phys. Rev. Lett 85, 2418 (2000)
574: \bibitem{Truscott}
575: A. G. Truscott {\it et al.}, Science 291, 2570 (2001); F. Schreck
576: {\it et al.}, Phys. Rev. Lett. 87, 080403 (2001); Z. Hadzibabic
577: {\it et al.}, Phys. Rev. Lett. 88, 160401 (2002).
578: \bibitem{Roati}
579: G. Roati {\it et al.}, Phys. Rev. Lett. 89, 150403 (2002)
580: \bibitem{Molmer}
581: K. M{\o}lmer, Phys. Rev. Lett. 80, 1804 (1998).
582: \bibitem{Amoruso}
583: M. Amoruso {\it et al.}, Eur. Phys. J. D 4, 261 (1998).
584: \bibitem{BB}
585: H.P. B\"{u}chler and G. Blatter, Phys. Rev. Lett. \textbf{91},
586: 130404 (2003).
587: \bibitem{Meisel}
588: M.W. Meisel, Physica, \textbf{178}, 121 (1992)
589: \bibitem{Lengua}
590: G.A. Lengua {\it et al.}, J. Low. Temp. Phys. \textbf{79}, 251 (1990)
591: \bibitem{vanotterlo}
592: A. van Otterlo {\it et al.}, Phys. Rev. B \textbf{52}, 16176 (1995)
593: \bibitem{Mermin}
594: Di Neil W. Ashcroft, N . David Mermin, {\it Solid State Physics},
595: (published by Holt, Rinehart and Winston, 1976).
596: \bibitem{BBpra}
597: H.P. B\"{u}chler and G. Blatter, Phys. Rev. A \textbf{69},
598: 063603 (2003)
599: \bibitem{Pines}
600: David Pines and Philippe Nozieres, {\it Theory of Quantum
601: Liquids}, (The Perseus Books Group, 1970).
602: \bibitem{Vignale}
603: G.F Giuliani and G.Vignale, {\it Quantum Theory of the Electron Liquid}
604: (Cambridge University Press, 2005)
605: \bibitem{Penrose}
606: O. Penrose and L. Onsager, Phys. Rev. 104, 576 (1956).
607: \bibitem{Andreev}
608: A. F. Andreev and I. M. Lifshitz, Sov. Phys. JETP 29, 1107 (1967).
609: \bibitem{Leggett}
610: A. J. Leggett, Phys. Rev. Lett. 25, 1543 (1970).
611: \bibitem{111213}
612: F. H\`{e}rbert {\it et al.}, Phys. Rev. B 65, 014513 (2001); E.
613: Frey and L. Balents, Phys. Rev. B 55, 1050 (1997); K. G\`{o}ral
614: {\it et al.}, Phys. Rev. Lett. 88, 170406 (2002).
615: \bibitem{Gruner}
616: G. Gr\"{u}ner, {\it Density Waves in Solids}, (Perseus Publishing,
617: Cambridge, Massachusetts, 2000).
618: \bibitem{vanhove}
619: L. Van Hove, Phys Rev. \textbf{89}, 1198, (1953).
620: \bibitem{Scalapino}
621: J. E. Hirsch and D. J. Scalapino, Phys. Rev. Lett. 56, 2732
622: (1986).
623: \bibitem{Viverit}
624: L. Viverit {\it et al.}, Phys. Rev. A 61, 053605 (2000).
625: \bibitem{Fetter}
626: A.L. Fetter and J.D. Walecka, {\it Quantum Many Particle Systems}
627: (Westview Press, Boulder, 1998)
628: \bibitem{Jessen}
629: P.S. Jessen and I.H. Deutsch, {\it Optical lattices}
630: Adv. At., Mol., and Opt. Phys. , {\bf 37}, 95 (1996)
631: \bibitem{Abr}
632: M. Abramowitz and I. A. Stegun, {\it Handbook of Mahtem- atical
633: Functions}, (Dover Publications, New York, (1972).
634: \bibitem{Abrikosov}
635: A. A. Abrikosov, L. P. Gorkov, and I. E. Dzyaloshinski, {\it
636: Methods of Quantum Field Theory in Statistical Physics}, (Dover
637: Publications, 180 Varick Street, New-York 10014, 1963).
638: \bibitem{Padova}
639: L. Salasnich, G.Mazzarella, and F.Toigo, in preparation (2008).
640:
641:
642:
643:
644: \end{thebibliography}
645: \end{document}
646: Fig.~\ref{Pd}.
647: