1: %......................................................................

2: % Editor

3: % Physical Review

4: % 1 Research Road

5: % Box 1000

6: % Ridge, NY  11961

7: %

8: % Dear Editor:

9: %

10: %  Please find enclosed a revtex version of the manuscript entitled

11: %

12: % "Electronic Ferroelectricity in the Falicov-Kimball"

13: %

14: %     by: C.D. Batista

15: %

16: % which I submit for publication in the

17: %       PHYSICAL REVIEW LETTERS

18: %

19: %......................................................................

20: %\documentclass[twocolumn,superscriptaddress,showpacs]{revtex4}

21: \documentclass[twocolumn,showpacs]{revtex4}

22: \usepackage{times,xspace}

23: \usepackage{amsbsy,amssymb,amsmath,bm}

24: \usepackage{graphicx,color,epsfig,rotate}

25: \usepackage{fancyheadings}

26: \def\one{{\mathchoice {\rm 1\mskip-4mu l} {\rm 1\mskip-4mu l} {\rm

27: 1\mskip-4.5mu l} {\rm 1\mskip-5mu l}}}

28: \def\bbbc{{\mathchoice {\setbox0=\hbox{$\displaystyle\rm C$}\hbox{\hbox

29: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}

30: {\setbox0=\hbox{$\textstyle\rm C$}\hbox{\hbox

31: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}

32: {\setbox0=\hbox{$\scriptstyle\rm C$}\hbox{\hbox

33: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}

34: {\setbox0=\hbox{$\scriptscriptstyle\rm C$}\hbox{\hbox

35: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}}}

36: \newcommand{\Naturals}{\ensuremath{\mathbb{N}}\xspace}

37: \newcommand{\Integers}{\ensuremath{\mathbb{Z}}\xspace}

38: \newcommand{\Rationals}{\ensuremath{\mathbb{Q}}\xspace}

39: \newcommand{\Reals}{\ensuremath{\mathbb{R}}\xspace}

40: \newcommand{\Complex}{\ensuremath{\mathbb{C}}\xspace}

41: \newcommand{\Zero}{\ensuremath{\mathbb{O}}\xspace}

42: \newcommand{\Punitary}{\ensuremath{\mathbb{I}}\xspace}

43: \newcommand{\UPunitary}{\ensuremath{\mathbb{U}}\xspace}

44: \newcommand{\VPunitary}{\ensuremath{\mathbb{V}}\xspace}

45: \newcommand{\ket}[1]{|{#1}\rangle}

46: \newcommand{\bra}[1]{\langle{#1}|}

47: \newcommand{\kets}[2]{|{#1}\rangle_{{}_{\!\!{#2}}}}

48: \newcommand{\bras}[2]{{}_{{}_{{#2}\!\!}}\langle{#1}|}

49: \newcommand{\slb}[2]{{{#1}^{({#2})}}}

50: \newcommand{\trace}{\mbox{tr}}

51: \renewcommand{\tensor}{\otimes}

52: \newcommand{\vspan}{\mbox{span}}

53: \newcommand{\lowertr}{\mbox{lt}}

54: \newcommand{\diag}{\mbox{dg}}

55: \newcommand{\id}{I}

56: 

57: \newcommand{\ignore}[1]{}

58: \newcommand{\mComment}[1]{}

59: \newcommand{\gComment}[1]{}

60: \newcommand{\jComment}[1]{}

61: \newcommand{\rComment}[1]{}

62: \newcommand{\lComment}[1]{}

63: 

64: \renewcommand{\mComment}[1]{\textcolor{blue}{Manny: #1}}

65: \renewcommand{\gComment}[1]{\textcolor{red}{Gerardo: #1}}

66: \renewcommand{\jComment}[1]{\textcolor{green}{Jim: #1}}

67: \renewcommand{\rComment}[1]{\textcolor{magenta}{Ray: #1}}

68: \renewcommand{\lComment}[1]{\textcolor{purple}{Rolando: #1}}

69: 

70: \pagestyle{fancy}

71: \pagestyle{fancyplain}

72: \footrulewidth 0.4pt

73: \plainheadrulewidth 0.4pt

74: \plainfootrulewidth 0.4pt

75: \lhead{\large LA-UR-00-XXXX}

76: \chead{ \today}

77: \rhead{\sl submitted to Physical Review Letters}

78: \cfoot{\sc\thepage}

79: \lfoot{}

80: \rfoot{}

81: 

82: 

83: \begin{document}

84: \title{Intermediate Coupling Theory of Electronic Ferroelectricity}

85: \author{C. D. Batista$^1$, J. E. Gubernatis$^1$, J. Bon\v ca$^2$, and H. Q. Lin$^3$}

86: \address{$^1$Center for Nonlinear Studies and Theoretical Division\\

87: Los Alamos National Laboratory, Los Alamos, NM 87545\\

88: $^2$ Department of Physics, FMF\\ University of Ljubljana and J. Stefan

89: Institute, Ljubljana, Slovenia\\

90: $^3$ Department of Physics, The Chinese University of Hong Kong, China}

91: 

92: 

93: \date{Received \today }

94: 

95: 

96: 

97: \begin{abstract}

98: We calculate the quantum phase diagram of an extended Falicov-Kimball model

99: for one and two-dimensional systems in the intermediate coupling regime.

100: Even though some features of the phase diagram are obtained analytically,

101: the main results are calculated with a constrained path Monte Carlo

102: technique. We find that this regime is dominated

103: by a Bose-Einstein condensation of excitons with a built-in

104: electric polarization. The inclusion of a finite hybridization between

105: the bands removes the condensate but reinforces the ferroelectricity.

106: \end{abstract}

107: 

108: \pacs{71.27.+a, 71.28.+d, 77.80.-e}

109: 

110: \maketitle

111: 

112: In a recent paper \cite{batista}, one of us demonstrated that a

113: novel ferroelectric state, originally proposed by Portengen {\it et

114: al} \cite{Portengen}, is present in the strong-coupling and

115: mixed-valence regime of an extended Falicov-Kimball model (FKM)

116: \cite{Falicov}. In contrast to the traditional ferroelectric (FE)

117: transitions, which are a subgroup of the structural phase

118: transitions, the new FE state is caused by a purely

119: electronic mechanism of an essential {\it quantum} nature. 

120: Its electronic origin and the strong

121: interplay between the orbital and the spin flavors of the

122: electronic degrees of freedom make this novel state suitable for

123: new technological applications. For example, it opens the

124: possibilities of faster switching ferroelectrics and controlling

125: their optical properties with magnetic fields. Accordingly, it is

126: important to define more clearly the conditions for the

127: experimental observation of this novel state, especially in light

128: of several recent experiments that might have seen it

129: \cite{Moro,Spielman}.

130: 

131: %%%The 

132: Our extended FKM model consists of two dispersive bands 

133: plus an inter-band local Coulomb interaction. In absence of 

134: explicit hybridization between the bands, the half-filled model exhibits a

135: FE state that is a Bose-Einstein (BE) condensate of excitons.  The 

136: electric polarization is the result of a spontaneous

137: phase-coherent hybridization between the two local orbitals with

138: opposite parity under inversion. The FE moment is proportional to

139: the real part of the $U(1)$ order parameter of the condensate.

140: These conclusions were based on a strong coupling

141: treatment of the model. An accompanying perturbative

142: analysis suggested that if hybridization between the bands were

143: included, the BE condensate (BEC) would be replaced by a

144: distinguishing Ising-like FE state \cite{batista}.

145: Because hybridization is natural and the experimental consequences

146: are significant, resolving this point is important and timely.% \cite{czy}.

147: 

148: To resolve this point, we will now step way beyond the original

149: work by moving away from the strong-coupling limit

150: and from the perturbative treatment of the hybridization 

151: and focus on the intermediate coupling regime.

152: We find that ferroelectricity remains robust.

153: In particular, it possesses the same phases as the strong and weak

154: couplings regimes, even though the nature of the condenstate changes

155: continuously from BCS-like in the weak coupling limit 

156: to a BEC of hard-core bosons in the strong couping regime \cite{Nozieres}.  

157: %in the present case  

158: %the condensation of electron-hole

159: %pairs has two different regimes: a BCS-like regime for weak and

160: %intermediate couplings and a BE condensation of hard core bosons

161: %for strong coupling \cite{batista}. The first case is the one most

162: %Thus, we can identiy the phases most

163: %relevant for experiment because most real systems are in the weak

164: %or the intermediate coupling regimes.  

165: We will in fact report the complete

166: one (1D) and two dimensional (2D) quantum phase diagrams of the

167: extended FKM in the intermediate coupling regime. The results are

168: obtained from a scaling analysis of the numerical data produced with the Constrained 

169: Path Monte Carlo (CPMC) technique \cite{zhang1}. When we include a finite 

170: hybridization between the bands, we find as predicted by the

171: perturbation analysis a very large FE (or antiferroelectric)

172: region for the mixed valence regime 

173: which  is no longer a condensate of excitons but a state with an Ising-like symmetry that has 

174: a built-in electric polarization.  Importantly, the intermediate coupling phase diagram 

175: indicates more accessible regions of the ferroelectric phase than in the 

176: strong and weak coupling regimes. 

177: 

178: %In addition,

179: %the transition between the non-mixed and the mixed valence regimes is described

180: %in an exact way.

181: 

182: There are two recent experiments to which our results might apply.

183: Moro {\it et al} observed a FE state in free Nb clusters

184: \cite{Moro}. The experiment shows that cold clusters may attain an

185: anomalous component with very large electric dipole moments. It is

186: clear that in pure Nb the FE moment cannot be produced

187: by a lattice distortion. Therefore, the ferroelectricity must have

188: an {\it electronic origin}. In addition, {\it ab initio}

189: calculations \cite{Grad} in pure Nb indicate that the valence

190: electron has the same amount of $s$ and $p$ character. This

191: observation makes the present theory a candidate to explain

192: the anomalous ferroelectricity of the free Nb clusters.

193: 

194: Another relevant experiment is the phase coherent state

195: observed by Spielman {\it et al} \cite{Spielman}. The two bands are provided by

196: two parallel 2D electron systems interacting through an interlayer

197: Coulomb interaction. There is also an applied magnetic field

198: whose value is such that the Landau level filling factor of each layer is $1/2$.

199: From the results obtained in Ref. \cite{batista} and the theory that we include below,

200: we propose an experiment to observe electronic ferroelectriciy in these bilayer systems.

201: 

202: % Comment on IR-active optical phonons

203: Our extended FKM for spin-less fermions on a $D$-dimensional

204: hypercubic lattice is:

205: \begin{eqnarray}

206: H &=& \sum_{\bf i, \nu} \epsilon_{\nu} n^{\nu}_{\bf i}

207: + \sum_{{\bf i},\eta, \nu,\nu'} t_{\nu \nu'} (f^{\dagger}_{\bf i \nu} f^{\;}_{{\bf i+{\hat e}_{\eta}} \nu'}

208: +  f^{\dagger}_{{\bf i+{\hat e}_{\eta}} \nu'} f^{\;}_{\bf i \nu})

209: \nonumber \\

210: &+& U^{ab} \sum_{\bf i} (n^a_{\bf i}-1/2) (n^b_{\bf i}-1/2),

211: \label{H}

212: \end{eqnarray}

213: where $\nu=\{a,b\}$ is the orbital index, ${\bf \hat e}_{\eta}$ is

214: a vector in the $\eta$ direction ($\eta=\{x_1,x_2,...,x_D\}$), and

215: $n^{\nu}_{\bf i}= f^{\dagger}_{\bf i \nu} f^{\;}_{\bf i \nu}$ is

216: the occupation number of each orbital. Note that $t_{ab}=-t_{ba}$

217: because the two orbitals have opposite parity under spatial

218: inversion.

219: 

220: 

221: This Hamiltonian can be rewritten as an asymmetric Hubbard

222: model if the orbital flavor is represented by a pseudospin variable \cite{batista}:

223: $c^{\dagger}_{\bf i \uparrow}= f^{\dagger}_{{\bf i} a}$, $c^{\;}_{{\bf i} \uparrow} = f^{\;}_{{\bf i} a}$,

224: $c^{\dagger}_{\bf i \downarrow}= f^{\dagger}_{{\bf i} b}$ and

225: $c^{\;}_{{\bf i} \downarrow} = f^{\;}_{{\bf i} b}$.

226: The pseudospin components are given by:

227: $\tau^{\alpha}_{\bf i}= \frac{1}{2} \sum_{\nu,\nu'} f^{\dagger}_{\bf i \nu}

228: {\boldsymbol \sigma}^{\nu \nu'}   f^{\;}_{\bf i \nu'}$, with $\alpha=\{ x, y, z\}$.

229: The new expression for $H$ is an asymmetric Hubbard model with a Zeeman term:

230: \begin{eqnarray}

231: H &=& e_d \sum_{{\bf i}, \sigma} n_{i \sigma} +

232: \sum_{\bf \langle i, j \rangle , \sigma, \sigma'} t_{\sigma \sigma'}

233: (c^{\dagger}_{{\bf i} \sigma} c^{\;}_{{\bf j} \sigma'} +

234: c^{\dagger}_{{\bf j} \sigma'} c^{\;}_{{\bf i} \sigma})

235: \nonumber \\

236: &+& U^{ab} \sum_{\bf i} n_{{\bf i} \uparrow} n_{{\bf i} \downarrow}

237: + B_z \sum_{\bf i} \tau^{z}_i,

238: \label{Hubbard}

239: \end{eqnarray}

240: where $e_d=\frac {1}{2} (\epsilon_a + \epsilon_b)$, $t_{\uparrow\uparrow}= t_{aa}$,

241: $t_{\downarrow\downarrow} = t_{bb}$, $t_{\uparrow \downarrow}=t_{ab}$,

242: $t_{\downarrow \uparrow}=t_{ba}$ and $B_z=\epsilon_a - \epsilon_b$.

243: If there is no hybridization between the bands, there is a  remaining

244: $U(1)$ symmetry generated by the total magnetization along the $z$ axis $M^z=\sum_{\bf i} \tau^{z}_{\bf i}$.

245: In the original language, this $U(1)$ symmetry corresponds to the

246: conservation of the difference between the total number of particles in each band

247: (the bands cannot interchange particles).

248: 

249: We will consider from now on only the half-filled case, i.e., one particle per site.

250: The ground state of $H$ exhibits two possible magnetic orderings

251: in the strong coupling limit: Ising-like antiferromagnetism and $xy$-like ferromagnetism (antiferromagnetism)

252: for $\gamma=-t_{\downarrow \downarrow}/t_{\uparrow \uparrow}>0$ ($\gamma<0$) \cite{batista}.

253: In terms of the original variables, the Ising-like phase corresponds to staggered orbital ordering (SOO)

254: and for $D \geq 2$, the $xy$-like ferromagnetism corresponds to a BEC of excitons with a built-in electric

255: polarization \cite{Portengen}:

256: \begin{equation}

257: {\bf P}= \frac {\boldsymbol{\mu}}{\Omega} \sum_{\bf i} (f^{\dagger}_{{\bf i} a} f^{\;}_{{\bf i} b}+

258: f^{\;}_{{\bf i} a} f^{\dagger}_{{\bf i} b})= \frac {2 \boldsymbol{\mu}}{\Omega} M^{x},

259: \label{pol}

260: \end{equation}

261: where $\boldsymbol{\mu}$ is the inter-band dipole matrix element and $\Omega$

262: is the volume of the system. The electrical polarization is then proportional

263: the real part of the $U(1)$ order parameter of the condensate.

264: 

265: An addtional particle-hole transformation on one of the spin

266: flavors maps Eq.~\ref{Hubbard} into a negative $U$ asymmetric Hubbard model with

267: the Zeeman term replaced by a chemical potential \cite{Fradkin}. Under this

268: transformation, the $xy$-like ferromagnetism is mapped into a

269: superconducting  state. This establishes a formal connection

270: between the BEC of excitons and a superconductor. An

271: eventual condensate of electron-hole pairs in the weak coupling

272: regime is then formally analogous to a BCS-like superconductor.

273: 

274: {\it One dimensional case}. From now on, we will set $\gamma>0$

275: and $t_{\uparrow \uparrow}$ as the unit of energy. 

276: We start by on considering the simplest 

277: case of zero hybridization between the bands: $t_{\uparrow

278: \downarrow}=t_{\downarrow \uparrow}=0$. In Fig.~\ref{fig1}, we

279: present the 1D quantum phase diagram of $H$ as a function of

280: $\gamma$ and $B_z$. Most of the phase diagram was calculated with

281: the unbiased CPMC technique \cite{zhang1}. The diagram shows the same three

282: phases which were obtained in the strong coupling limit

283: \cite{batista}. For $B_z=0$, the ground state is an Ising-like

284: antiferromagnet (SOO). At a critical value of $|B_z|=B^{c1}_z$,

285: there is a transition from the orbitally ordered gaped phase to a

286: critical (gapless) phase with dominating pseudospin-pseudospin

287: correlations in the $xy$ plane. This second phase is also characterized 

288: by a non-zero uniform magnetization along the $z$ axis. When $|B_z|$ reaches a

289: second critical value, $B^{c2}_z$, $M^z$ saturates and a new gap

290: opens. In terms of our original language, this is the transition

291: from a mixed valence to non-mixed valence regime in which one band

292: is full and the other one is empty. This is also a transition from

293: an excitonic to a band insulator.

294: 

295: \begin{figure}[htb]

296: \includegraphics[angle=-90,width=6.5cm,scale=0.6]{fig1new.ps}

297: \vspace{-0.8cm}

298: \caption{(color online) One dimensional quantum phase diagram of the

299: extended FKM $U^{ab} =1$($t_{\uparrow \downarrow}=t_{\downarrow \uparrow}=0$) \cite{note}. 

300: The circles are the calculated

301: points for $B^{c1}_z$ and the full line is an exponential fit. The error bars are smaller

302: than the symbol sizes. The squares are numerical results for $B^{c2}_z$ and the full

303: line is the analytical calculation.}

304: \label{fig1}

305: \end{figure}

306: 

307: The value of $B^{c2}_z$ is exactly obtained by solving the one particle problem of

308: flipping one pseudospin in a fully saturated system. To show that this value is 

309: not altered by a discontinuous change of the magnetization, we also include 

310: the numerical calculation of $B^{c2}_z$ based on the computation of $E_{m}(M^z,B^z=0)$,

311: minimum energy for the subspace with a given magnetization $M^z$. Note that this can be done

312: because $M^z$ is a good quantum number of $H(t_{\uparrow\downarrow}=t_{\downarrow \uparrow}=0)$.

313: $B^{c1}_z$ is the critical value of $B^z$ which is required to change the total magnetization

314: $M^z$ of the ground state, i.e., it is the minimum value of $[E_{m}(M^z, B^z=0)-E_{GS}(B^z=0)]/M^z$,

315: where $E_{GS}$ is the ground state energy.

316: To calculate $B^{c1}_z$ in the thermodynamic limit, we scaled this minimum value for different chain sizes up

317: to $N_s=100$ ($N_s$ is the number of sites). A typical scaling is shown in Fig.~\ref{fig2}a

318: for two different values of $\gamma$. The finite extrapolated value of $B^{c1}_z$ implies a 

319: finite gap $\Delta_I$ for the Ising-like (SOO) phase. The case $\gamma=1$ corresponds to the $SU(2)$

320: critical point for which the pseudospin correlations along the three different axis are the same.

321: For the strong coupling limit \cite{batista}, it is known that the gap of the Ising-like (SOO) phase

322: closes exponentially (Kosterltiz-Thouless transition) when $\gamma$ gets close to one (see Fig.~\ref{fig1}).

323: 

324: The different phases were identified by computing the pseudospin-pseudospin correlation function:

325: \begin{equation}

326: S^{\alpha}({\bf q})= \frac{1}{N_s^2} \sum_{\bf i, j} 

327: (\tau^{\alpha}_{\bf i} \tau^{\alpha}_{\bf j} - \langle \tau^{\alpha}_{\bf i} \rangle 

328: \langle \tau^{\alpha}_{\bf j}\rangle )

329: e^{i{\bf q} \cdot ({\bf r_i-r_j})}

330: \label{str}

331: \end{equation}

332: In Fig.~\ref{fig2}b, we show the scaling of $S^x(0)$ and $S^z(\pi)$ calculated with the CPMC for

333: $B_z=0$ and $\gamma=0.1$. As expected, $S^z(\pi)$ has a finite value in the thermodynamic limit

334: indicating that the ground state has long-range staggered orbital ordering.

335: \begin{figure}[htb]

336: \includegraphics[angle=-90,width=7.5cm,scale=0.8]{fig2new.ps}

337: \vspace{-1.0cm}

338: \caption{(color online) U=1, $t_{\uparrow\downarrow}=t_{\downarrow \uparrow}=0$. 

339: a) Scaling of $B^{c1}_z$ calculated with the CPMC method in chains of

340: length $N_s$. b) Scaling of $S^{x}(0)$ and $S^{z}(\pi)$ for $B_z=0$. 

341: c) Scaling of the energy gap $\Delta_{xy}$ for $M^z/N_s=1/4$. The scaling function is 

342: a second order polynomial expression in $1/N_s$. d) Pseudospin-pseudospin correlation 

343: function for $M^z/N_s=1/4$.}

344: \label{fig2}

345: \end{figure}

346: 

347: In Fig.~\ref{fig2}c, we show the scaling of the energy gap of the $xy$-like phase,

348: $\Delta_{xy}$, for $M^z/N_s=1/4$. 

349: $\Delta_{xy}$ scales to zero in the thermodynamic limit as expected from the critical 

350: nature of this phase. Fig.~\ref{fig2}d  shows that the pseudospin-pseudospin correlation 

351: function (see Eq.~\ref{str}) in the $x$ (or $y$) direction is much larger than the corresponding

352: correlation along the $z$ axis. This shows the $xy$-like nature of this phase. 

353: 

354: {\it Two dimensional case}. The main qualitative difference between the 1D (Fig.\ref{fig1})

355: and the 2D (Fig. \ref{fig3}) quantum phase diagrams of

356: $H(t_{\uparrow \downarrow}=t_{\downarrow \uparrow}=0)$ is that the excitonic condensate is critical 

357: in one dimension, as required by the Mermin-Wagner

358: theorem \cite{Mermin}, and has long range order in two dimensions.

359: In the pseudospin language, this means that the 2D system has a finite in-plane magnetization ${\bf M}^{\perp}$ 

360: (see Fig.~\ref{fig4}). In addition, Eq.~(\ref{pol}) shows that this state has built-in electric polarization 

361: proportional to the $x$ component of ${\bf M}^{\perp}$ or the real part of the complex order parameter of the excitonic

362: condensate \cite{batista}.

363: 

364: \begin{figure}[htb]

365: \includegraphics[angle=-90,width=6.5cm,scale=0.6]{fig3new.ps}

366: \vspace{-0.8cm}

367: \caption{(color online) Two dimensional phase diagram

368: of the extended FKM for $U=2$

369: ($t_{\uparrow \downarrow}=t_{\downarrow \uparrow}=0$).The circles are the calculated

370: points for $B^{c1}_z$ and the full line is a guide to the eye. The error bars are smaller

371: than the symbol sizes. The squares are numerical results for $B^{c2}_z$ and the full

372: line is the analytical calculation.}

373: \label{fig3}

374: \end{figure}

375: 

376: As in the 1D case, $B^{c2}_z$($\gamma$) was computed exactly, and the

377: transition from the SOO to the ferroelectric BEC of

378: excitons is simultaneously a first order valence transition

379: because the relative population between the two bands, $M^z$, 

380: changes discontinuously. Note that in the intermediate coupling

381: regime, the extension of the ferroelectric phase for $\gamma$ not

382: very close to zero is of the order of the bandwidth while in the

383: strong coupling regime the size of this region is proportional to

384: $t_{aa}t_{bb}/U^{ab}$.

385: 

386: We conjecture that the three dimensional (3D) quantum phase diagram of $H$ should be similar to the

387: 2D one since the quantum fluctuations are smaller in the first case. Moreover, we expect the 

388: quantum phase diagram to have the same topology for the weak and intermediate coupling regimes in any 

389: dimension $D \geq 2$. The main expected difference between D=2 and D=3

390: occurs at finite temperatures due to the Mermin-Wagner theorem \cite{Mermin}.

391: The transition temperature associated to the BE condensation  is finite only for the 3D case.

392: The BEC of electron-hole pairs undergoes a  Kosterlitz-Thouless phase transition

393: in a 2D system. 

394: 

395: 

396: From Eq.~(\ref{pol}), we see that in any dimension the observed

397: electronic hybridization is the consequence of a spontaneous phase-coherent hybridization between the

398: two bands. The phase must have a real component in order to produce a finite electric polarization.

399: We can ask what is the physical consequence of a pure imaginary phase 

400: $\phi=\pm \pi/2$ ($\tan\phi=M^x/M^y$), i.e.,

401: a magnetization pointing in the $y$ direction. In this case, instead of ferroelectricity, the phase-coherent state

402: induces an ordering of atomic currents \cite{batista2}. Since the two atomic orbitals have

403: opposite parity, these currents do not produce a net local magnetic dipole but they can

404: have a net quadrupolar magnetic moment. Therefore, the $y$ direction in the pseudospin space

405: corresponds to ferroquadrupolar or antiferroquadrupolar magnetic ordering.

406: 

407: As predicted in \cite{batista}, we will now show that the inclusion of

408: an explicit hybridization between the bands breaks explicitly the

409: $U(1)$ symmetry 

410: which is spontaneously broken in the condensate and removes the degeneracy associated with

411: all the possible orientations of ${\bf M}^{\perp}$. Along with this the

412: natural question to answer is: what is the particular 

413: orientation favored by a non-zero hybridization.

414: 

415: 

416: {\it Inclusion of Hybridization}. In addition to the trivial $U(1)$ symmetry associated

417: with the conservation of the total number of particles, the remaining symmetries of $H$ for non-zero

418: hybridization are the spatial (${\hat I}$) and

419: the temporal (${\hat T}$) inversions. Since the two orbitals have opposite parities,

420: the signs of $t_{\uparrow \downarrow}$ and $t_{\downarrow \uparrow}$ are interchanged

421: under the application of ${\hat I}$. For the pseudospin components we have:

422: ${\hat I}\tau^x_{\bf i}{\hat I}=-\tau^x_{\bf i}$, ${\hat I}\tau^y_{\bf i}{\hat I}=-\tau^y_{\bf i}$

423: and ${\hat I}\tau^z_{\bf i}{\hat I}=\tau^z_{\bf i}$. The temporal inversion (complex conjugation)

424: leaves $\tau^x_{\bf i}$ and $\tau^z_{\bf i}$ invariant and changes the sign of $\tau^y_{\bf i}$.

425: The ferroelectric state occurs when ${\hat I}$ is spontaneously broken. The ordering of atomic

426: currents occurs if both ${\hat I}$ and ${\hat T}$ are spontaneously broken. Note that the

427: product ${\hat I}{\hat T}$ is still a symmetry of this second state.

428: 

429: \begin{figure}[htb]

430: \includegraphics[angle=-90,width=8.0cm,scale=0.6]{fig4new.ps}

431: \vspace{-4.0cm}

432: \caption{(color online) Scaling of $S^{x}(0)$ and $S^{y}(0)$ for a 2D system:

433: $U=4$,

434: $t_{\uparrow \downarrow}=-t_{\downarrow \uparrow}=0.2$ and $B_z=1$ (full lines),

435: $U=2$, $t_{\uparrow \downarrow}=-t_{\downarrow \uparrow}=0$ and $M^z/N_s=1/4$ (dashed line).

436: The scaling function is a second order polynomial expression in $1/N_s$.}

437: \label{fig4}

438: \end{figure}

439: 

440: In Fig.~\ref{fig4}, we plot $S^x(0)$ and $S^y(0)$ as a function of the

441: size of a two-dimensional system.

442: The largest system has $N_s=100$ sites. The scaling clearly shows that the ferroelectric

443: state ($M_x \neq 0$) is stabilized in the thermodynamic limit. This agrees

444: with a perturbative analysis which is only valid in the strong coupling limit \cite{batista}.

445: 

446: The BEC of excitons is thus replaced by an Ising-like

447: ferroelectric state when a finite hybridization is included. This

448: result has important physical consequences since most real

449: systems have an explicit hybridization between two bands. This hybridization is

450: the reason why a BEC of excitons is so hard to observe.

451: The only exception is the bilayer system that we mentioned in the

452: introduction \cite{Spielman}. According to the theories presented

453: here and in Ref. \cite{batista}, this bilayer system must exhibit

454: the phenomenon of electronic ferroelectricity if the two layers

455: are made of bands with opposite parity (like "$s$" and "$p$")

456: instead of being identical. Then, the $d|{\bf P}|/dE$

457: response as a function of the uniform electric field $E$ should

458: show the same resonant behavior observed in the

459: tunneling conductance as a function of the interlayer voltage

460: \cite{Spielman}.

461: 

462: In summary, we derived the quantum 1D and 2D phase diagrams of the extended FKM

463: in the intermediate coupling regime. In the absence of

464: hybridization, we found that the insulating phase

465: obtained at half filling has a transition from a non-mixed valence to

466: a mixed valence regime as a function of the energy difference between the centers of both bands. 

467: At this transition the system changes from a band to an excitonic insulator which is

468: a BEC of electron-hole pairs. Additionally, we showed that this condensate

469: contains two different orderings which are degenerate: {\it ferroelectric order} and {\it chiral order}. 

470: The chiral state is a spontaneous ordering of local atomic currents. Since the two bands

471: have opposite parity under spatial inversion, these currents can only produce a non-zero quadrupolar 

472: magnetic moment. For bands with the same parity, the chiral state can produce orbital magnetism

473: and the ferroelectric state is replaced by quadrupolar electric ordering.

474: If the distance between both bands is further decreased, there is

475: a first order valence transition from the BEC of excitons to a state with

476: staggered orbital ordering.

477: 

478: Addressing an issue most important for physical systems, we also

479: considered the effect of a finite hybridization on these ordered

480: degenerate phases. We

481: showed that the $U(1)$ degeneracy associated with the BEC is

482: lifted by the hybridization, and the degenerate ground state is

483: replaced by an Ising-like FE state 

484: (broken $Z_2$ symmetry).  This result is valid even beyond the

485: perturbative regime $|t_{ab}|<<|t_{aa}|,|t_{bb}|$. It is crucial for

486: applications of the present theory to systems like the free Nb

487: clusters

488: \cite{Moro} which are most likely in the intermediate coupling

489: regime and have a considerable hybridization between the $s$ and

490: the $p$ valence bands.

491: 

492: 

493: %To study the strong interplay between the orbital and the spin

494: %flavors of the electronic degrees of freedom requires adding to

495: %Eq. (\ref{Hubbard}) spin and Hubbard-like interactions for the d

496: %and f orbitals. A quantum Monte Carlo study of this ambitious

497: %Hamiltonian is underway \cite{batista3}. 

498: 

499: We wish to thank M. P. Lilly, J. P. Eisenstein and R. Moro for stimulating discussions.

500: This work was sponsored by the US DOE under contract

501: W-7405-ENG-36. J.B. acknowledges the financial

502: support  of Slovene Ministry of Education, Science and Sports.

503: H.Q. Lin is partly supported by RGC Project CUHK 4037/02P.

504: 

505: \begin{references}

506: 

507: \bibitem{batista}

508: C. D. Batista, Phys. Rev. Lett. {\bf 89}, 166403 (2002).

509: 

510: \bibitem{Portengen}

511: T. Portengen, Th. \"Ostreich and L. J. Sham,

512: Phys. Rev. Lett. {\bf 76}, 3384 (1996); Phys. Rev. B

513: {\bf 54}, 17452 (1996).

514: 

515: \bibitem{Falicov}

516: L. M. Falicov and J. C. Kimball,

517: Phys. Rev. Lett. {\bf 22}, 997 (1969).

518: 

519: \bibitem{Moro}

520: R. Moro, X. Xu, S. Yin and W. A. de Heer, Science {\bf 300}, 1265 (2003).

521: 

522: \bibitem{Spielman}

523: I. B. Spielman, J. P. Eisenstein, L. N. Pfeiffer and K. W. West,

524: Phys. Rev. Lett. {\bf 84}, 5808 (2000).

525: 

526: 

527: \bibitem{Nozieres}

528: P. Nozieres and S. Scmmitt-Rink, J. Low Temp. Phys. {\bf 59}, 195 (1985).

529: 

530: 

531: 

532: %\bibitem{zhang}

533: %S. Zhang, J. Carlson and J. E. Gubernatis, Phys. Rev.  Lett., {\bf 74}, 3652 (1995).

534: 

535: 

536: \bibitem{zhang1} S. Zhang, J. Carlson and J. E. Gubernatis, Phys. Rev.

537: Lett., {\bf 74}, 3652 (1995).

538: 

539: \bibitem{Fradkin} 

540: E. Fradkin, Field Theories of Condensed Matter Physics 

541: (Addison-Wesley, Redwood City, 1991).

542: 

543: 

544: \bibitem{note}

545: The excitonic phase disappears at $\gamma=0$ due to the localization of the 

546: pseudospin down electrons. For D=1 at $\gamma=0$, this phase should undergo a 

547: transition to the most homogeneous array of down electrons or to a phase segregated

548: metallic state if $B_z$ is close to $B^{c2}_z$ and $U \lesssim 1.2$ \cite{far}.

549: A similar behavior is expected for the 2D case \cite{far}. Our results indicate 

550: that these transitions occur for $\gamma < 0.05$ in D=1 and $\gamma < 0.1$

551: in the 2D case.

552: 

553: \bibitem{far}

554: P. Farkas\v sovsky, Eur. Phys. Jour. B {\bf 20}, 209 (2001).

555: 

556: \bibitem{Grad}

557: G. B. Grad {\it et al}, Phys. Rev.  B, {\bf 62}, 12743 (2000).

558: 

559: \bibitem{Mermin}

560: N. D. Mermin and H. Wagner, Phys. Rev. Lett. {\bf 17}, 1133 (1966).

561: 

562: 

563: \bibitem{batista2}

564: C. D. Batista, in preparation.

565: 

566: %\bibitem{batista3} C. D. Batista, J. Bon\v ca, J. E. Gubernatis, and

567: %H.-Q. Lin, in preparation.

568: 

569: \end{references}

570: 

571: 

572: 

573: \end{document}

574: