1: \documentclass[prl,twocolumn]{revtex4}
2: %\documentclass[prl,preprint]{revtex4}
3: \usepackage{amsfonts}
4: \usepackage{amsmath}
5: \usepackage{amssymb}
6: \usepackage{graphicx}
7:
8: %========================================================================
9: % MACROS
10: \newcommand{\xh}{\mathbf{x}}
11: \newcommand{\yh}{\mathbf{y}}
12: \newcommand{\rv}{\mathbf{r}}
13: \newcommand{\Qv}{\mathbf{Q}}
14: \newcommand{\Tr}{{\rm Tr}}
15: \newcommand{\R}{\rangle}
16: \renewcommand{\L}{\langle}
17: \newcommand{\e}{\mathrm{e}}
18: \newcommand{\dxx}{$d_{x^2-y^2}$}
19: \newcommand{\dxy}{$d_{xy}$}
20:
21: %=======================================================================
22:
23:
24: \begin{document}
25: %*
26: \title{Antiferromagnetism and superconductivity in layered organic conductors: Variational cluster approach}
27: \author{P.~Sahebsara and D.~S\'en\'echal}
28: \affiliation{D\'{e}partement de physique and Regroupement qu\'{e}b\'{e}cois sur les mat\'{e}riaux de pointe, Universit\'{e} de Sherbrooke, Sherbrooke, Qu\'{e}bec, Canada, J1K 2R1}
29: \date{April 2006}
30:
31: \begin{abstract}
32: The $\kappa$-(ET)$_2$X layered conductors (where ET stands for BEDT-TTF) are studied within the dimer model as a function of the diagonal hopping $t^\prime$ and Hubbard repulsion $U$.
33: Antiferromagnetism and $d$-wave superconductivity are investigated at zero temperature using variational cluster perturbation theory (V-CPT).
34: For large $U$, N\'eel antiferromagnetism exists for $t' < t'_{c2}$, with $t'_{c2}\sim 0.9$.
35: For fixed $t'$, as $U$ is decreased (or pressure increased), a $d_{x^2-y^2}$ superconducting phase appears.
36: When $U$ is decreased further, the a $d_{xy}$ order takes over.
37: There is a critical value of $t'_{c1}\sim 0.8$ of $t'$ beyond which the AF and dSC phases are separated by Mott disordered phase.
38: \end{abstract}
39:
40: \pacs{71.27.+a,74.70.Kn, 71.10.Fd,78.20.Bh}
41: \maketitle
42: %=======================================================================
43:
44: The proximity of antiferromagnetism (AF) and d-wave superconductivity (dSC) is a central and universal feature of high-temperature superconductors, and leads naturally to the hypothesis that the mechanisms behind the two phases are intimately related.
45: This proximity is also observed in the layered organic conductor $\kappa$-(ET)$_2$\-Cu\-[N(CN)$_2$]\-Cl, an antiferromagnet that transits to a superconducting phase upon applying pressure\cite{Lefebvre:2000} (here ET stands for BEDT-TTF).
46: Other compounds of the same family, $\kappa$-(ET)$_2$\-Cu\-(NCS)$_2$ and $\kappa$-(ET)$_2$\-Cu\-[N(CN)$_2$]\-Br, are superconductors with a critical temperature near 10K at ambient pressure.
47: However, another member of this family, $\kappa$-(ET)$_2$\-Cu$_2$\-(CN)$_3$, displays no sign of AF order, but becomes superconducting upon applying pressure\cite{Shimizu:2003, Kurosaki:2005}.
48: The character of the superconductivity in these compounds is still controversial.
49: While many experiments indicate that the SC gap has nodes (presumably $d$-wave), others are interpreted as favoring a nodeless gap.
50: The literature on the subject is rich, and we refer to a recent review article\cite{Powell:2006} for references.
51:
52: The interplay of AF and dSC orders, common to both high-$T_c$ and $\kappa$-ET materials, cannot be fortuitous and must be a robust feature that can be captured in a simple model of these strongly correlated systems.
53: $\kappa$-ET compounds consist of orthogonally aligned ET dimers that form conducting layers sandwiched between insulating polymerized anion layers.
54: The simplest theoretical description of these complex compounds is the so-called dimer Hubbard model\cite{Kino:1996, McKenzie:1998} (Fig.~\ref{fig_dimer}A) in which a single bonding orbital is considered on each dimer, occupied by one electron on average, with the Hamiltonian
55: \begin{eqnarray}
56: H=t\sum_{\langle \rv\rv'\rangle,\sigma}c^\dagger_{\rv\sigma}c_{\rv'\sigma} +
57: t' \sum_{[\rv\rv'],\sigma} c^\dagger_{\rv\sigma}c_{\rv'\sigma} + U \sum_\rv n_{\rv\uparrow} n_{\rv\downarrow}
58: \end{eqnarray}
59: where $c_{\rv\sigma}$ ($c^\dagger_{\rv\sigma}$) creates an electron (hole) at dimer site $\rv$ on a square lattice with spin projection $\sigma$, and $n_{\rv\sigma}$=$c^\dagger_{\rv\sigma}c_{\rv\sigma}$ is the hole number operator. $\langle \rv\rv' \rangle$ ($[\rv\rv']$) indicates nearest- (next-nearest)-neighbor bonds.
60: As the ratio $t'/t$ grows from 0 towards 1, N\'eel AF is increasingly frustrated.
61: The values of $t'$ for the Br, Cl, NCS and CN$_3$ compounds are thought to be roughly $0.5-0.65$, $0.75$, $0.75-0.85$ and 1 respectively~\cite{McKenzie:1998,Shimizu:2003}.
62: The local Coulomb repulsion $U$ is hard to calculate from first principles, but is estimated to be of the same order of magnitude as the band-width\cite{Powell:2006}.
63: We assume that applying pressure has the effect of increasing $t$ and $t'$ in proportion, without affecting $U$; thus, for fixed $t$, we refer to $t/U$ as the pressure\cite{Kino:1996, McKenzie:1998}.
64: Different anions correspond also to different chemical pressures, so that, at ambient pressure, the various members of the family of compounds correspond to different points on the $(U,t')$ plane.
65: Henceforth we will set $t$ to 1, thereby measuring $t'$ and $U$ in units of $t$.
66:
67: %-----------------------------------------------------------------------
68: %: FIG:fig_dimer
69: \begin{figure}[tbp]
70: \centerline{\includegraphics[width=8.5cm]{fig1}}
71: \caption{(Color online) (A) : Schematic view of the hopping terms in the dimer model. V-CPT uses a tiling of clusters such as the 4-site cluster drawn here (dashed lines). (B) : Qualitative phase diagram inferred from our calculations (not to scale). See text for details.
72: }
73: \label{fig_dimer}
74: \vglue-4mm
75: \end{figure}
76: %-----------------------------------------------------------------------
77:
78: The dimer model has been studied with a variety of methods.
79: For instance, The fluctuation exchange approximation (FLEX)\cite{flex_REF} predicts the existence of d-wave superconductivity based on diverging susceptibilities.
80: Quantum Monte Carlo calculations\cite{Kuroki:1999} find enhanced dSC correlations.
81: Ground state wave-functions obtained from variational methods\cite{variational_GS} predict a transition between a dSC and a spin liquid phase, akin to a Mott transition, in the highly frustrated case.
82: The spin liquid phase was also predicted by variational methods, and through a gauge theory of the Hubbard model\cite{Motrunich:2005}.
83: Variational calculations based on the dimer model plus additional exchange terms reveal a transition between dSC and an AF insulator\cite{RVB}.
84: Cellular Dynamical Mean Field Theory (CDMFT)\cite{Parcollet:2004} has also been used to reveal the Mott transition in that case, without treating the broken symmetry phases.
85:
86: We report results of a zero-temperature study of the dimer model using Variational Cluster Perturbation Theory (V-CPT)\cite{Potthoff:2003}.
87: This method captures short-range correlations exactly, treats broken symmetries with a rigorous dynamical variational principle, and provides dynamical information (the spectral function).
88: It has been recently applied with success to the broken symmetry phases in models of high-$T_c$ superconductors\cite{Senechal:2005, Aichhorn:2005}: transitions between AF and dSC states were found as a function of doping, both for hole- and electron-doped materials, at roughly the correct doping levels.
89: Let us summarize our results (see Fig.~\ref{fig_dimer}B): Beyond the normal (disordered) state, the following states have been investigated: an AF (N\'eel) state, a dSC state with \dxx\ symmetry and an extended $s$-wave state which, for reasons explained below, we will call \dxy.
90: For all values of $t'$ we find a \dxx\ phase at low $U$.
91: For $t'$ smaller than a critical value $t'_{c1}$, we find a dSC to AF transition driven by $U$.
92: For intermediate $t'$ ($t'_{c1}<t'<t'_{c2}$), the AF and dSC phases are separated by a Mott insulator phase (MI), with no AF order (spin liquid phase).
93: For $t'>t'_{c2}$, the AF phase disappears, leaving only the dSC and MI phases.
94: We find that $t'_{c1}\sim 0.8$ and $t'_{c2}\sim 0.9$.
95: For $t'>0.25-0.3$, we find a \dxy\ phase at sufficiently low $U$.
96: The possibility of other magnetic orders, in particular of a $120^\circ$ order at $t'=t$, has not been explored in this work, and therefore we make no claim that the MI phase corresponds to a spin liquid \cite{Kurosaki:2005} everywhere.
97:
98: %-----------------------------------------------------------------------
99: %: Variational cluster perturbation theory
100:
101: \paragraph{Variational Cluster Perturbation Theory.}
102:
103: V-CPT is an extension of Cluster Perturbation Theory (CPT) \cite{Senechal:2000} based on Potthoff's self-energy-functional approach (SFA)\cite{Potthoff:2003, Dahnken:2004}.
104: In the SFA one defines a functional $\Omega_{\bf t}[\Sigma]$ of the self-energy that is stationary at the physical self-energy (here ${\bf t}$ stands for the matrix $t_{ij}$ of one-body terms).
105: This functional can be evaluated exactly for a self-energy $\Sigma'$ that is the exact self-energy of a Hamiltonian $H'$ that differs from the original Hamiltonian only through its one-body terms (i.e., $H$ and $H'$ share the same interaction part).
106: In V-CPT, the lattice of the model is tiled into identical clusters (Fig.~\ref{fig_dimer}), and $H'$ is the restriction of the dimer model to a finite cluster, to which one adds various Weiss fields that allow for broken symmetries.
107: The Green function of the Hamiltonian $H'$ is then calculated exactly (by the Lanczos method) and the self-energy functional may then be exactly expressed as\cite{Senechal:2000}
108: \begin{equation}
109: \Omega_{\bf t}({\bf t}')=\Omega'\kern-0.1em - \kern-0.1em\int_C \frac{d\omega}{2\pi}\sum_{\mathbf{K}}\ln\det\left(
110: 1 \kern-0.1em + \kern-0.1em (G_0^{-1}\kern-0.2em -G_0'{}^{-1})G'\right)
111: \end{equation}%
112: where $G'$ is the exact Green function of $H'$, $G_0$ and $G_0'$ are the non-interacting Green functions of $H$ and $H'$ respectively, and $\Omega'$ is the exact grand potential of $H'$.
113: A trace over frequencies and wave-vectors of the reduced Brillouin zone is carried, and the Green functions carry discrete indices related to spin and sites within a cluster.
114: The functional $\Omega_{\bf t}[\Sigma]$ has become a function $\Omega_{\bf t}({\bf t}')$ of the parameters of the cluster Hamiltonian $H'$.
115: The task is then to find a stationary point of that function, and the exact self-energy of $H'$ at that stationary point, denoted $\Sigma^*$, is adopted as an approximate self-energy for the original model $H$.
116:
117: %-----------------------------------------------------------------------
118: %: FIG:fig_OP1
119: \begin{figure}[tbp]
120: \centerline{\includegraphics[width=8.5cm]{fig2}}
121: \caption{(Color online) Pressure ($t/U$) dependence of the antiferromagnetic (blue triangles), \dxx\ (red circles, scaled by 2) and \dxy (green squares, scaled by 5) order parameter, obtained with V-CPT on $2\times2$ clusters, for various values of $t'/t$.
122: The vertical lines indicate the transition points, separating the various phases.
123: }
124: \label{fig_OP1}
125: \vglue-5mm
126: \end{figure}
127: %-----------------------------------------------------------------------
128:
129: We insist that the only approximation used is the restriction of the space of self-energies to the set of exact self-energies of a family of cluster Hamiltonians, with the same interaction part as the original Hamiltonian.
130: In particular, short-range correlation effects are treated exactly through the exact cluster Green function.
131: In order to study antiferromagnetism and superconductivity, one includes in $H'$ the following ``Weiss terms'':
132: \begin{eqnarray}
133: H^{\prime}_{AF}&=&M\sum_{\rv}(-1)^{\sigma} e^{i\Qv\cdot\rv} n_{\rv\sigma} \\
134: H^{\prime}_{SC}&=&\sum_{\rv,\rv'}\left(\Delta _{\rv\rv'} c_{\rv\uparrow }c_{\rv'\downarrow}+\mathrm{H.c.}\right)
135: \end{eqnarray}
136: where $\Qv=(\pi,\pi)$ is the N\'eel wave vector.
137: The \dxx\ state is probed by letting $\Delta _{\rv\rv'}=\Delta$ if $\rv'=\rv\pm\xh$ and $\Delta _{\rv\rv'}= -\Delta$ if $\rv'=\rv\pm\yh$.
138: These Weiss fields allow for the physics of long-range order to seep through the cluster self-energy.
139: But this is not mean-field theory: the interaction term is never factorized, and the Weiss fields are not the same as the corresponding order parameters.
140: The latter can be calculated from the Green function associated with the solution: $G^*{}^{-1} = G_0 - \Sigma^*$.
141: The density of electrons $n$ is also computed from the Green function $G^*$, as its trace: $n = \Tr\, G^*$.
142:
143: %-----------------------------------------------------------------------
144: %: Gap symmetry
145: \paragraph{Gap symmetry.}
146: The point group of the dimer model is $C_{2v}$, consisting of a rotation of $\pi$ about the $z$ axis, and reflections accross the $\xh+\yh$ and $\xh-\yh$ lines.
147: This group admits two possible gap symmetries for singlet superconductivity: the \dxx\ state defined above ($A_2$ representation), and the isotropic state ($s$-wave, or $ A_1$ representation).
148: See Ref.~\onlinecite{Powell:2006} for a more detailed group-theoretical analysis.
149: We also included in our study a SC Weiss fields belonging to the $A_1$ representation:
150: \begin{eqnarray}
151: \Delta_{\rv,\rv+\xh+\yh} = \Delta_{\rv,\rv-\xh-\yh} &=& \phantom{-}\Delta_1 \\
152: \Delta_{\rv,\rv+\xh-\yh} = \Delta_{\rv,\rv-\xh+\yh} &=& -\Delta_2
153: \end{eqnarray}
154: (note that the on-site Coulomb repulsion excludes the possibility $\Delta_{\rv,\rv}\ne0$).
155: If $\Delta_1=\Delta_2$, these two possibilities form together what is customarily called the \dxy\ state, but the relative values of $\Delta_1$ and $\Delta_2$ are not constrained by the symmetry of the model.
156: We will nevertheless refer to this state as \dxy.
157: Even though this state belongs to the $A_1$ representation, the corresponding gap function will display nodes (roughly along the $x$ and $y$ axes), even though these nodes may not be robust -- e.g. with respect to impurities -- on group-theoretical grounds \cite{Powell:2006}.
158: Recent STM observations\cite{Ichimura:2006} are interpreted as favoring this state.
159:
160: In this work, five variational parameters were used: the Weiss fields $M$, $\Delta$, $\Delta_1$ and $\Delta_2$, as well as the chemical potential $\mu'$ on the cluster, the latter in order to ensure thermodynamic consistency\cite{Aichhorn:2005}.
161: The cluster chemical potential $\mu'$, which is part of $H'$, must not be confused with the actual chemical potential $\mu$, which is part of $H$ and controls the density of electrons on the lattice.
162: The Nambu formalism was used in order to treat anomalous averages.
163: Time-reversal violating states such as $d+is$ (e.g. a complex mixture of the $A_1$ and $A_2$ representations) were not investigated, because of our use of real-number quantum mechanics in numerical computations.
164: %-----------------------------------------------------------------------
165: %: FIG:fig_OP2
166: \begin{figure}[tbp]
167: \centerline{\includegraphics[width=8.5cm]{fig3}}
168: \caption{(Color online) Same as Fig.~\protect\ref{fig_OP1}, this time for $2\times4$ clusters.
169: }
170: \label{fig_OP2}
171: \vglue-5mm
172: \end{figure}
173: %-----------------------------------------------------------------------
174:
175:
176: %-----------------------------------------------------------------------
177: %: Results
178: \paragraph{Results.}
179: Fig.~\ref{fig_OP1} shows the order parameters as a function of pressure $t/U$, for different values of $t'$, using a 4-site cluster ($2\times2$).
180: Fig.~\ref{fig_OP2} shows the same type of data obtained on an 8-site cluster ($2\times4$).
181: The blue triangles were obtained by treating only $M$ and $\mu'$ as variational parameters, i.e. by forbidding a SC solution.
182: The red circles were obtained by treating only $\Delta$ and $\mu'$ as variational parameters, i.e. by forbidding an AF solution.
183: The green squares were obtained by treating $\Delta_1$, $\Delta_2$ and $\mu'$ as variational parameters.
184: No mixed solutions were found by performing a 3-parameter variation of $M$, $\Delta$ and $\mu'$.
185: Normal state solutions, where the only variational parameter is $\mu'$, were also obtained.
186: For each value of $U$ and $t'$, the lattice chemical potential $\mu$ was adjusted in order for the electron density $n$ to be unity (half-filling) within 0.1\%.
187: Then a comparison of the ground state energy per site $\Omega+\mu n$ was made in order to decide which of the four solutions (AF, \dxx, \dxy\ or normal) was preferred (only the lowest-energy solutions are shown here).
188: On Figs \ref{fig_OP1} and \ref{fig_OP2}, vertical lines separate the different phases thus determined, with the SC phases having the lowest energy at high pressure, the AF phase at low pressure, with an intercalated normal phase -- in fact a paramagnetic Mott insulator -- appearing for $t'$ greater than some critical value $t'_{c1}$ that is close to 0.8.
189: The AF phase disappears altogether for $t'$ greater than some critical value $t'_{c2}$ of about 0.9.
190:
191: The transitions between the different ordered states (AF, \dxx\ and \dxy) are of the first order, whereas the transitions from AF to normal look continuous.
192: The results obtained on the 4-site and 8-site clusters are in overall agreement on the existence of all four phases and their relative location.
193: The critical value $U_c(t')$ for the SC to AF/MI transition increases slightly with $t'$, but stays between 5 and 6 for both cluster sizes, at least for $t'\geq 0.75$.
194: The maximum magnitude of the \dxx\ order parameter does not vary much with $t'$.
195: By contrast, the critical $U$ for the MI to AF transition increases sharply with $t'$, as the saturation value of the AF order parameter drops.
196: Differences between the two cluster sizes lie mostly in the extent of the \dxy\ phase, which is more important in the 8-site than on the 4-site cluster.
197: This is reflected in the ``foggy'' boundary between the two SC phases that we display on the schematic phase diagram of Fig.~\ref{fig_dimer}.
198: Note that the cluster sizes used are too small to perform any kind of finite-size analysis, and we cannot say that the results are converged.
199:
200: The \dxy\ phase only appears at sufficiently high $t'$ (it was not found at $t'=0.25$) and low $U$: it seems favored by the dispersion, but at intermediate coupling the tide turns towards a \dxx\ order.
201: In this V-CPT calculation, there is no evidence of a lower critical $U$: the SC phase appears to extend all the way to $U=0$ and no normal metallic phase is found.
202:
203: %-----------------------------------------------------------------------
204: %: FIG:fig_edc
205: \begin{figure}[tbp]
206: \centerline{\includegraphics[width=7cm]{fig4}}
207: \caption{
208: Energy distribution curves (i.e. spectral function) for three sample solutions with $t'=0.8t$.
209: (A): an antiferromagnetic solution at $U=9t$.
210: The SDW dispersion curves for the same values of the parameters ($U$, $\mu$, $t'$ and $t$) and a mean field $h=0.4t$ are also drawn.
211: (B): the same, for a Mott insulator solution ($U=7t$).
212: What looked like the AF gap in (A) is now a Mott gap.
213: (C): the same, for a \dxx\ solution ($U=5t$).
214: Notice the $d$-wave gap maximum along the $(0,0)-(0,\pi)$ segment.
215: In all cases, a Lorentzian broadening of $0.15t$ has been used.
216: }
217: \label{fig_edc}
218: \vglue-4mm
219: \end{figure}
220: %-----------------------------------------------------------------------
221:
222: Fig.~\ref{fig_edc} shows the spectral function for sample solutions for three different values of $U$ corresponding to (A) the AF phase, (B) the MI phase and (C) the \dxx\ phase at $t'=0.8t$, on an 8-site cluster.
223: In (A) the mean-field AF dispersion is also plotted (full lines), with a mean-field $h=0.4t$, a value chosen so as to best match the V-CPT dispersions (the actual V-CPT order parameter in that case is $\L M\R = 0.497$).
224: Note that the mean-field dispersion roughly matches the V-CPT spectral function, at least where the weight is important.
225: In (B), the spectral function displays a Mott gap across all wave-vectors.
226: It is very similar to the AF spectral function (A), except that it does not curve back to follow the AF mean-field dispersion near $(0,\pi)$ and $(\pi/2,\pi/2)$.
227: In (C), we notice the \dxx\ gap, maximum along the $(0,0)-(0,\pi)$ segment, and vanishing along the $(-\pi,\pi)-(0,0)$ segment.
228: Except for that dSC gap, the spectral function is that of a metal, i.e. there is no other gap in the spectrum.
229: Unfortunately, photoemission (ARPES) experiments, which would offer a probe on the negative-frequency part of the spectral function, have not, to our knowledge, been performed on these compounds.
230:
231: To conclude, the dimer model of layered organic conductors captures the essential features of this family of compounds: AF and SC phases separated by a first order transition, except at higher frustration levels where a MI phase (possibly spin liquid) appears.
232: We found that the superconducting gap symmetry changes from the $A_1$ (i.e., \dxy) to the $A_2$ (i.e., \dxx) representation of the point group as $U$ increases, which leads to the possibility of a close interplay between those two states as a function of pressure in these compounds.
233: Of course an exciting development would be the synthesis of a new member of the family displaying N\'eel order at ambient pressure, with a value of $t'$ in the 0.8-0.9 range.
234: Then this compound would be predicted to go through a spin liquid, and then through a dSC phase upon applying pressure.
235:
236: %-----------------------------------------------------------------------
237:
238: \begin{acknowledgments}
239: We are indebted to A.-M. Tremblay, C.~Bourbonnais, B.~Kyung and D. Fournier for useful discussions.
240: This work was supported by NSERC (Canada).
241: Computations were performed on the Dell clusters of the R\'eseau Qu\'eb\'ecois de Calcul de haute performance (RQCHP).
242: \end{acknowledgments}
243:
244:
245: %=======================================================================
246: %: BIBLIO
247: \begin{thebibliography}{99}
248:
249: \bibitem{Lefebvre:2000}
250: S. Lefebvre {\it et al.}, Phys. Rev. Lett. {\bf 85}, 5420 (2000).
251:
252: \bibitem{Shimizu:2003}
253: Y. Shimizu, et al. Phys. Rev. Lett. {\bf 91}, 107001 (2003).
254:
255: \bibitem{Kurosaki:2005}
256: Y. Kurosaki {\it et. al.}, Phys. Rev. Lett. {\bf 95}, 177001 (2005).
257:
258: \bibitem{Powell:2006}
259: B.J. Powell and R.H. McKenzie, cond-mat/0607078.
260:
261: \bibitem{Kino:1996}
262: H. Kino and H. Fukuyama, J. Phys. Soc. Jpn. 65 (1996) 2158.
263:
264: \bibitem{McKenzie:1998}
265: R.H. McKenzie, Comm. Cond. Mat. Phys. {\bf 18} 309 (1998).
266:
267: \bibitem{flex_REF}
268: J. Schmalian, Phys. Rev. Lett. {\bf 81}, 4232 (1998);
269: H. Kino and H. Kontani, J. Phys. Soc. Jpn {\bf 67}, 3691 (1998).
270: H. Kondo and T. Moriya, J. Phys. Soc. Jpn {\bf 67}, 3695 (1998).
271: K. Kuroki {\it et al.}, Phys. Rev. B {\bf 65}, 100516 (2002).
272:
273: \bibitem{Kuroki:1999}
274: K. Kuroki and H. Aoki, Phys. Rev. B {\bf 60}, 3060 (1999).
275:
276: \bibitem{variational_GS}
277: J. Liu, J. Schmalian, and N.~Trivedi, Phys. Rev. Lett. {\bf 94}, 127003 (2005).
278: T. Watanabe {\it et al.}, cond-mat/0602098.
279:
280: \bibitem{Motrunich:2005}
281: O.I. Motrunich, Phys. Rev. B{\bf 72}, 045105 (2005); S.-S. Lee and P.A. Lee, Phys. Rev. Lett. {\bf 95}, 036403 (2005).
282:
283: \bibitem{RVB}
284: B.J. Powell and R.H. McKenzie, Phys. Rev. Lett. {\bf 94}, 047004 (2005);
285: J. Y. Gan {\it et al.}, Phys. Rev. Lett. {\bf 94}, 067005 (2005).
286:
287: \bibitem{Parcollet:2004}
288: O. Parcollet, G. Biroli and G. Kotliar, Phys. Rev. Lett. {\bf 92}, 226402 (2004).
289:
290: \bibitem{Potthoff:2003}
291: M. Potthoff, Eur. Phys. J. B \textbf{32}, 429 (2003).
292:
293: \bibitem{Senechal:2005} D.~S\'en\'echal, P.-L. Lavertu, M.-A. Marois and A.-M.~S. Tremblay, Phys. Rev. Lett. \textbf{94}, 156404 (2005).
294:
295: \bibitem{Aichhorn:2005}
296: M. Aichhorn, E. Arrigoni, M. Potthoff and W. Hanke, cond-mat/0511460.
297:
298: \bibitem{Senechal:2000}
299: D.~S\'en\'echal \textit{et al.}, Phys. Rev. Lett. \textbf{84}, 522 (2000);
300: Phys. Rev. B \textbf{66}, 075129 (2002).
301:
302: \bibitem{Dahnken:2004}
303: C.~ Dahnken et al., Phys. Rev. B \textbf{70}, 245110 (2004).
304:
305: \bibitem{Ichimura:2006}
306: K.~Ichimura and K.~Nomura, J. Phys. Soc. Jpn {\bf 75}, 051012 (2006).
307:
308:
309: \end{thebibliography}
310: \end{document}
311: %=======================================================================
312: