1: %\documentclass[aps,showpacs,prb,floatfix,twocolumn]{revtex4}
2: \documentclass[aps,prl,floatfix,epsfig,twocolumn,showpacs,preprintnumbers]{revtex4}
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: \usepackage{amsmath,amssymb,graphicx,bm,epsfig}
5:
6: \newcommand{\eps}{\epsilon}
7: \renewcommand{\a}{\alpha}
8: \renewcommand{\b}{\beta}
9: \newcommand{\vR}{{\mathbf{R}}}
10: \renewcommand{\vr}{{\mathbf{r}}}
11: \newcommand{\vk}{{\mathbf{k}}}
12: \newcommand{\vK}{{\mathbf{K}}}
13: \newcommand{\vq}{{\mathbf{q}}}
14: \newcommand{\vQ}{{\mathbf{Q}}}
15: \newcommand{\vPhi}{{\mathbf{\Phi}}}
16: \newcommand{\vS}{{\mathbf{S}}}
17: \newcommand{\cG}{{\cal G}}
18: \newcommand{\cF}{{\cal F}}
19: \newcommand{\cD}{{\cal D}}
20: \newcommand{\Tr}{\mathrm{Tr}}
21: \newcommand{\npsi}{\underline{\psi}}
22:
23: \begin{document}
24:
25: \title{Avoided Quantum Criticality near Optimally Doped High Temperature Superconductors}
26:
27: \author{Kristjan Haule and Gabriel Kotliar}
28:
29: \affiliation{Department of Physics, Rutgers University, Piscataway, NJ 08854, USA}
30:
31: \begin{abstract}
32: We study the crossover from the underdoped to the overdoped
33: regime in the t-J model.
34: %Cluster Dyamical Mean Field Theory. We
35: %find coherent quasiparticles present in the normal state of both
36: %the underdoped and the overdoped regime. However the character of
37: %these quasiparticles is very different in the two regimes.
38: The underdoped regime is dominated by the superexchange
39: interaction, locking the spins into singlets which weakly
40: perturbe coherent charge carriers. In the overdoped, large carrier
41: concentration regime, the Kondo effect dominates resulting in
42: spin-charge composite quasiparticles which are also coherent.
43: Separating these two Fermi liquid regimes, there is a critical
44: doping where superexchange and Kondo interaction balance each
45: other, bringing the system close to a local quantum critical
46: point near the point of maximal superconducting transition
47: temperature. At this point, particle hole symmetry is dynamically
48: restored and physical quantities such as the optical
49: conductivity, exhibit power law behaviour at intermediate
50: frequencies as observed experimentally. Quantum criticality is
51: avoided by the onset of superconductivity.
52: \end{abstract}
53: \pacs{71.27.+a,71.30.+h}
54: \date{\today}
55: \maketitle
56:
57: The high temperature superconductors have a normal state
58: which is not well described by the standard theory of metals
59: \cite{Lee}. Several anomalies are well established
60: experimentally: a) The normal state close to the doping level
61: with the highest transition temperature is highly incoherent and
62: does not fit the standard Fermi liquid picture, b) at small
63: doping the normal state has a pseudogap, c) the optimally doped
64: regime exhibits anomalous power laws in transport quantities and
65: in the optical conductivity in an intermediate frequency range.
66: This observation has lead to the suggestion that the
67: superconducting dome covers a critical point \cite{Tallon}, d) The
68: superconducting state not only exhibits coherent Bogolubov
69: quasiparticles and charge collective modes, but also a very sharp
70: spin mode or "40 meV resonance". The role of this mode as well as
71: of the putative quantum critical fluctuations in a fundamental
72: theory of cuprate supercondcutivity are important issues.
73:
74: In this letter we present a coherent account of these observations
75: based on a Cluster Dynamical Mean Field Theory analysis of the t-J
76: model. This approach allows us to access the regime of
77: intermediate temperatures in the doping regime between underdoped
78: to overdoped system. In addition, the approach is formulated in
79: terms of pseudoparticles, representing collective excitations of
80: the system in an enlarged Hilbert space. This formulation provides
81: a clear interpretation of the numerical results, providing a
82: transparent physical interpretation.
83:
84:
85:
86: %Our study connects the cuprates with other strongly
87: %correlated systems, for example heavy fermions, for which local
88: %quantum criticality has emerged as an important unifying principle.
89:
90:
91: %Furthermore the finite temperature methodology described here with its
92: %concomitant dramatic reduction of the relevant Hilbert space should be
93: %of general use in other problems.
94:
95:
96:
97: We start from the the t-J model which was proposed by P.W. Anderson as
98: a minimal model to describe the cuprate superconductors
99: \begin{equation}
100: H = -\sum_{ij\sigma} t_{ij}c_{i\sigma}^\dagger c_{j\sigma}+
101: \frac{1}{2}\sum_{ij}\mathbf{S}_i\mathbf{S}_j .
102: \end{equation}
103: It contains two terms; one describes the kinetic energy which
104: delocalizes the holes introduced by doping, and a spin spin
105: interaction.
106: %which causes the spins to lock in singlets. Here
107: A constraint of no double occupancy must be enforced.
108: % at each
109: %moment in time.
110:
111: We used the extended Dynamical Cluster Approximation
112: \cite{Hettler} combined with the slave particle approach
113: \cite{HaulefU} which have been very useful in the theory of
114: strongly correlated electrons. The cluster approach maps the full
115: many body problem, into a a set of local degrees of freedom (in
116: this case a two by two plaquette) which are treated exactly and a
117: bath which is treated self consistently.
118: % \cite{Georges}, allowing
119: %the description of a proper thermodynamical limit including broken
120: %symmetry states.
121:
122: The coupling of the cluster to the medium, which simulates the
123: rest of the lattice, causes the cluster eigenstates to decay in
124: time therefore their spectral functions carry nontrivial
125: frequency dependence and important information about
126: %the
127: %competition of
128: various physical processes
129: %that take place on the
130: %cluster or between the cluster and medium, including
131: such as the RKKY interactions, the Kondo effect and d-vawe
132: superconductivity. To study the dynamics of the cluster states,
133: we diagonalized the cluster, i.e.,
134: $H_{cluster}|m\rangle=E_m|m\rangle$, and assigned to each cluster
135: eigenstate a pseudoparticle $|m\rangle=a_m^\dagger|0\rangle$. The
136: original problem can be exactly expressed in terms of
137: pseudoparticles $a_m$ with the only non-quadratic term of the
138: converted Hamiltonian being the hybridization between the cluster
139: and the medium. This cubic hybridization term is taken into
140: account by the self-consistent resummation of the non-crossing
141: diagrams in the spirit of the well known Non-Crossing
142: approximation to the Anderson impurity model. Since the Hilbert
143: space of the new base of pseudoparticles accesses some
144: non-physical states, one needs to project them out with the
145: constraint expressing the completeness of the cluster eigenbasis.
146: First we calculate the pseudoparticle self energies and the
147: pseudoparticle Greens functions. Finally, physical observables
148: like spectral function or susceptibilities can be calculated by a
149: weighted sum of the convolution of the pseudoparticle spectral
150: functions. The following matrix elements
151: $(F^{\vK\dagger})_{nm}=\langle n|c_\vK^\dagger|m\rangle$ and
152: $(\vS_Q)_{nm}=\langle n|\vS_Q|m\rangle$ give the bare weight of
153: the electron spectra and the spin response in terms of
154: pseudoparticles
155: \begin{eqnarray}
156: G_\vK(i\omega)=-\sum_{nmn'm'}(F^\vK)_{m'n'}C_{m'n'nm}(i\omega)(F^{\vK\dagger})_{nm}\\
157: \chi^{\a\b}_\vQ(i\omega) = \sum_{nmn'm'}(S_\vQ^\alpha)_{m'n'}C_{m'n'nm}(i\omega)(S_{-\vQ}^\beta)_{nm}
158: \end{eqnarray}
159: with
160: \begin{eqnarray}
161: C_{m'n'nm}(i\omega)=T\sum_{i\epsilon}G_{n'n}(i\epsilon)G_{mm'}(i\epsilon-i\omega)
162: \end{eqnarray}
163: Here $G_\vK$, $G_{nn'}$ and $\chi_\vQ$ are electron Green's
164: function, pseudoparticle Green's function and spin
165: susceptibility, respectively. This is the central equation of the
166: approach, relating observables to the pseudoparticle spectral
167: functions plotted in Fig.~\ref{pseudop}. More details of the
168: method can be found in \cite{Haule}.
169: %
170: It has been shown that the cluster approach successfully describes
171: many properties of the high temperature superconductors
172: \cite{Hettler,Haule,Civelli}. In this letter we show that it
173: also unravels the origin of the apparent criticality observed
174: near critical doping, and its elimination below the
175: superconducting state.
176:
177: %We associate with each exact eigenstate of the cluster $|m\rangle$ a
178: %new particle called pseudoparticle.
179: Physically pseudoparticles
180: represent coarsed grained versions of the important many body
181: excitations including fermionic quasiparticles and bosonic collective
182: modes. They have quantum numbers describing their spin, number of
183: particles, (which divided by the cluster size, give the density), and
184: a coarsed grained momentum. In momentum space, the course graining for
185: the smallest cluster divides the Brillouin zone into four different
186: regions, namely an inert patch around $(0,0)$ point, the second patch
187: centered around $(\pi,\pi)$ being highly non-Fermi liquid like at
188: small doping and finally $(\pi,0)$ and $(0,\pi)$ regions which contain
189: the Fermi surface of the model in the underdoped and slightly
190: overdoped regime and are therefore most relevant for thermodynamical
191: and transport properties.
192: %The physical response of the electron is
193: %expressed as a convolution of these pseudoparticles. This gives an
194: %intuitive physical interpretation of experimental observables such as
195: %photoemission spectra, tunnelling spectra, and optical spectra, in a
196: %unified perspective in terms of the spectra of pseudoparticles, and
197: %their evolution as a function of temperature and doping.
198:
199:
200: \begin{figure}
201: \includegraphics[width=0.8\linewidth]{fig1.eps}
202: \caption{
203: The cluster $(\pi,0)$ self-energy at zero frequency as a
204: function of temperature for few doping levels. The inset shows the
205: estimation of the coherent scale in the normal state of the t-J model
206: (black dots) and transition temperature to superconducting state (red
207: dots).
208: }
209: \label{phased}
210: \end{figure}
211:
212:
213: The first indication for an underlying criticality near optimal
214: doping comes from the evaluation of the electron scattering rate
215: obtained from the imaginary part of the electron self-energy as a
216: function of temperature for few different doping levels. As
217: described in Fig.~\ref{phased} both at large and small doping the
218: scattering rate is small as expected for a Fermi liquid.
219: Remarkably it becomes very large in the region near optimal
220: doping when the critical temperature is maximal. The transition
221: to the superconducting state severely reduces the scattering rate
222: eliminating the traces of the underlying critical behaviour,
223: hence the name avoided quantum criticality. A coherence scale,
224: estimated from the scattering rate, is plotted in the inset of
225: Fig.~\ref{phased} and shows it tends to vanish close to the
226: point of maximal superconducting transition temperature.
227:
228:
229: \begin{figure}
230: \includegraphics[width=0.8\linewidth]{powerlaws.eps}
231: \caption{ a) The absolute value of complex optical conductivity
232: $\sigma(\omega)$ is proportional to $\omega^{-2/3}$ in the
233: intermediate frequency region for optimally doped system. b) The
234: absolute value of the $(\pi,\pi)$ spin sussceptibility is proportional
235: to $\omega^{-2}$. c) The angle, calculated as
236: $\arctan(\mathrm{Im}(\sigma)/\mathrm{Re}(\sigma)$) is approximately
237: $\pi/3$ since $\sigma(\omega)\sim(-i\omega)^{-2/3}$. In the legend, N
238: stand for normal state and S for superconducting state.}
239: \label{powerlaws}
240: \end{figure}
241:
242: Additional evidence for the quantum criticality is the emergence
243: of power-law behaviours of the response functions.
244: %In our
245: %calculation, the reduction of the coherence scale is accompanied
246: %by a large number of approximate power laws that emerge in
247: %physical quantities with exponents being fractional numbers.
248: As shown in Fig.~\ref{powerlaws}, the optical conductivity has
249: an approximate power law with an exponent $2/3$. The same
250: powerlaw is realized in the one-electron self-energy while the
251: spin susceptibility is proportional to $\chi_{\pi,\pi}\propto
252: \omega^{-2}$ in the same frequency range. Experimentally, it was
253: found that the optical conductivity is proportional to the
254: $(-i\omega)^{-0.65}$ in the intermediate frequency regime
255: \cite{VanderMarel}.
256:
257:
258:
259: \begin{figure}
260: \includegraphics[width=0.8\linewidth]{40meVpeak.eps}
261: \caption{
262: The dynamical spin susceptibility at $\vq=(\pi,\pi)$ for few
263: different doping levels and three different temperatures:
264: superconducting state, normal state at the transition temperature
265: and at room temperature. The pronounced peak is formed in SC
266: state at $0.16t\approx 48\,\mathrm{meV}$ and a broad peak in normal state is
267: around $100-140\,\mathrm{meV}$.
268: Susceptibility at normal temperture is
269: much smaller and the peak moves to higher frequencies.
270: The resonance is strongest at the optimally
271: doped system. It disappears quickly in the overdoped site and
272: somewhat more slowly in the underdoped side.
273: }
274: \label{susc40}
275: \end{figure}
276:
277:
278:
279: Quantum criticality is avoided when the electrons condense forming
280: d-wave pairs. The electron scattering rate is dramatically
281: reduced (see Fig.~\ref{phased}) and a V-shaped gap opens in the
282: local one-electron density of states.
283: % due to emerging gap around
284: %hot spots.
285: The particle-hole response at $(\pi,\pi)$
286: %(between
287: %hots spots)
288: is severely reduced for frequencies below the superconducting
289: gap and a sharp
290: %excitonic
291: resonance appears in the gap. A sharp resonance peak in the spin
292: susceptibility was
293: %first
294: observed in bilayer cuprates YBCO and Bi2212 and
295: %only recently
296: in
297: singe layer Tl2201.
298: % Experimentaly, the resonance energy
299: It scales with doping like $5 T_c$ and does not depend on
300: temperature.
301: %This is not easily understood in a simple RPA
302: %framework.
303:
304: Our method gives a pronounced peak in the $(\pi,\pi)$ spin response at
305: frequency $0.16t$ in the optimally doped regime, as shown in
306: Fig.~\ref{susc40}. The peak position is temperature independent and
307: taking into account the maximal transition temperature in our model
308: $T_{c-max}\sim 0.035t$ the resonance energy is $4.6 T_c$ in good
309: agreement with experiments.
310:
311: In addition we see a broader peak around $0.35-0.45t$ extending to
312: very high frequencies of order of $t\approx 300\,\mathrm{meV}$
313: which also gains some weight upon condensation and represents an
314: important contribution to the exchange energy difference
315: between superconducting and normal state. This exchange energy
316: gives far the largest contribution to the condensation energy
317: %as
318: %shown in Ref.
319: \cite{Haule}.
320:
321:
322: Our method averages the momentum dependence over $1/4$ of the
323: Brillouin zone centered at $(\pi,\pi)$ therefore it is reasonable
324: to compare our results with the $\vq$ integrated susceptibility
325: from Ref.~\onlinecite{Dai} where both the $35\,\mathrm{meV}$
326: resonant peak as well as broader peak around $75\,\mathrm{meV}$
327: extending up to $220\,\mathrm{meV}$ was observed.
328:
329:
330:
331: \begin{figure}
332: \includegraphics[width=0.9\linewidth]{fig3.eps}
333: \caption{ left: Pseudoparticle spectral functions for the
334: three most important pseudoparticles: ground states for N=4, N=3 and
335: N=2 sectors. Right-top: Sketch of pseudoparticle treshold energies
336: which can be interpreted as the effective many-body levels in normal
337: and superconducting state. Right-bottom: Pseudoparticle occupancies
338: versus doping for most important pseudoparticles. The full lines
339: correspond to the normal state while the dashed lines correspond to
340: the superconducting state. }
341: \label{pseudop}
342: \end{figure}
343:
344:
345: {\it Pseudoparticle interpretation}: These results can be
346: understood in terms of the evolution of the pseudoparticle
347: spectral functions (spectral functions of the eigenstates of the
348: plaquette) depicted in Fig.~\ref{pseudop}. Remarkably, out of the
349: large number of pseudoparticles which are involved in our
350: calculation ($3^4$) only four distinct pseudoparticles are very
351: important, having the largest weight in the ground state and
352: containing more than 95\% of the spectral weight of the low energy
353: correlation functions. Fig.~\ref{pseudop}a shows the evolution of
354: the three most important pseudoparticle spectral functions from
355: the underdoped to the overdoped regime.
356:
357: {\it Underdoped Regime:} At small doping, the singlet state with
358: one particle per site and zero momentum (N=4,S=0,K=0) (half filled
359: singlet) dominates since it has far the lowest treshold energy, is
360: very strongly peaked at the treshold energy, and has the largest
361: occupancy as shown in Fig.~\ref{pseudop}c,. It describes a system
362: locked in a short range singlet state as a consequence of the
363: strong superexchange interaction. The electron spectral function
364: descrbies the process of addition and removal of an electron from
365: the system at frequency $\omega$. It is dominated by convolution
366: of two pseudoparticles with different cluster occupation N and
367: N+1, or N-1 and with integral in convolution taken between zero
368: and $\omega$. The necessary condition for a peak of the
369: one-particle spectral function at the Fermi level is that at
370: least two pseudoparticle spectral functions share common
371: threshold and are strongly peaked at the same threshold. Since
372: the thresholds of the other pseudoparticles are significantly
373: shifted with reference to the half filled singlet, a pseudogap
374: results in the one particle spectra in the underdoped regime.
375: This gap in threshold energies severely limits the possible decay
376: processes of the electron
377: %locked in the singlet state,
378: resulting
379: in a low electronic scattering rate.
380:
381: {\it The local picture of the overdoped regime:} On the overdoped
382: side, all three important pseudoparticles (half-filled singlet,
383: doublet with one hole per plaquette and singlet with two holes
384: per plaquette) have a power law divergence at the same threshold
385: frequency at zero temperature (Fig.~\ref{pseudop}a and
386: Fig.~\ref{pseudop}b) which is a standard signature of the Kondo
387: effect. Hence, the one particle spectral function develops the
388: Kondo-Suhl resonance at the Fermi level since the convolution
389: between the doublet and half-filled singlet (or N=2 singlet)
390: state is large at low frequency. The one-particle spectral
391: function is peaked slightly above the Fermi level. The underdoped
392: regime is distinctively different since only the half-filled
393: singlet state dominates while the doublet has very little
394: spectral weight in the region of the singlet peak.
395:
396: {\it The Transition Region, Normal State:} In the optimal doped
397: regime, the Kondo effect and the superexchange compete giving
398: rise to a regime with very large scattering rate and
399: consequently a small coherence scale. Criticality is the result
400: of the merging of the
401: convergence of the
402: thresholds of the
403: pseudoparticles (See Fig 4) when going from the underdoped to
404: the overdoped regime.
405: %This critical point is of local character
406: %regime dominated by
407: Local criticality has been found in models of heavy fermion
408: %the competition of the Kondo
409: %effect and the RKKY interactions in heavy fermion
410: systems \cite{Si} and frustrated magnets \cite{georges}.
411:
412: Surprisingly this evolution is accompanied by a restoration of
413: particle hole symmetry in the optimally doped regime. As shown in
414: Fig.~\ref{pseudop}b the threshold of the N=2 cluster ground state
415: and N=3 cluster ground state (doublet) merge first resulting in a
416: Kondo-like contribution to the electron spectral function. This
417: contribution is peaked above the Fermi level as usual for the
418: simple Fermi-liquid state in less than half-filled one band
419: model. The half-filled singlet however remains the lowest state
420: in energy and still gives a significant contribution to the
421: electron spectral function. The later contribution is peaked
422: below the Fermi level and keeps a pseudogap-like shape. The
423: result of the two contributions to the electron spectral function
424: is a restoration of the particle-hole symmetry in the density of
425: states both in normal and superconducting state at optimal
426: doping. The particle-hole symmetry of the density of states is
427: yet another hint of the proximity to the quantum critical point
428: since it is well known that only this symmetry allows non trivial
429: criticality in the two impurity Kondo model \cite{varma}.
430:
431:
432: {\it Transition into the superconducting state:} The degeneracy
433: responsible for the strongly incoherent metal with large
434: scattering rate at the Fermi level is lifted by the
435: superconductivity avoiding the quantum critical point.
436: Fig.~\ref{pseudop}a shows that both important singlet
437: pseudoparticles (for N=4 and N=2) develop a very sharp peak at
438: the same treshold frequency and, at the same time, their
439: occupancy increases (see Fig.~\ref{pseudop}c) upon condensation,
440: indicating that electrons are locked into singlets with zero
441: momentum. A gap opens between the singlets and doublets which
442: gives the gap in the one-particle density of states. Because of
443: this gap in the pseudoparticle thresholds, the large imaginary
444: part of the electron self-energy does not persist in the
445: superconducting state (see also Fig.~\ref{phased}). Since the
446: density of states is composed of two almost equally important
447: contributions, i.e., the convolution of the doublet with both
448: singlets (N=4 and N=2), the superconducting gap is almost
449: particle hole symmetric in the optimally doped regime with
450: half-width of the order of 0.12t. When the doping value is
451: changed from its critical value, the asymmetry in the
452: superconducting density of states appears. The magnitude of the
453: assymetry is the same as the asymmetry of the corresponding
454: normal state spectra and comes from the fact that the occupancy
455: and therefore importance of the N=4 half-filled singlet exceeds
456: the importance of the N=2 singlet (see Fig.~\ref{pseudop}c). The
457: spin susceptibility comes almost entirely from the convolution of
458: the half-filled singlet with the half-filled triplet
459: ($N=4,S=1,K=(\pi,\pi)$). The later develops a peak at an energy
460: $0.16t$ upon condensation which causes the resonance in the spin
461: susceptibility.
462:
463:
464: \textit{In conclusion} we showed that at optimal doping, in the t-J
465: model, there is a collapse of coherence scale, accompanied by a a
466: large scattering rate, restoration of particle-hole symmetry, which
467: result in approximate power laws in physical quantities, such as the
468: optical conductivity ($\sigma\propto\omega^{-2/3}$) and spin
469: susceptibility ($\chi_{\pi\pi}\propto \omega^{-2}$). Quantum
470: criticality is avoided when electrons condense into a supercondacting
471: state which results in a dramatic reduction of scattering rate, the
472: formation of an energy gap which in turns leads to a resonance in the
473: spin susceptibility. A transparent physical interpretation of the
474: numerical results was given in terms of pseudoparticles which
475: represent the important many body excitations, fermionic
476: quasiparticles and bosonic collective modes. This real space picture
477: complements the momentum space picture which emerges in other studies,
478: where substantial deformations of the Fermi surface which result in
479: the formation of Fermi arcs at comparable values of the doping
480: \cite{Haule,tdstan}. Finding analytic connection between single
481: particle properties in $k$ space and the collective excitations
482: presented in this study is an interesting direction for future
483: research.
484:
485:
486: \begin{thebibliography}{99}
487: \bibitem{Lee}
488: %P.A. Lee, N. Nagaosa, and X.G. Wen, Rev. Mod. Phys. \textbf{78}, 17 (2006);
489: %M.R. Norman, and C. Pepin, Rep. Prog. Phys. \textbf{66}, 1547 (2003);
490: T. Timusk, and B. Statt, Rep. Prog. Phys. \textbf{62}, 61
491: (1999); M.A. Kastner, R.J. Birgeneau, G. Shirane, and Y. Endoh,
492: Rev. Mod. Phys. \textbf{70}, 897 (1998).
493: \bibitem{Tallon} J.L. Tallon, J.W. Loram, G.V.M. Williams {\it et al.},
494: Phys. Status Solidi B \textbf{215} (1): 531 (1999).
495: \bibitem{Hettler} M.H. Hettler, A.N. Tahvildar-Zadeh, M. Jarrell,
496: T. Pruschke, and H.R. Krishnamurthy, Phys. Rev. B \textbf{58}, R7475 (1998); T. Maier, M. Jarrell,
497: T. Pruschke, and M.H. Hettler,
498: Rev. Mod. Phys. \textbf{77}, 1027 (2005).
499: \bibitem{HaulefU} K. Haule, S. Kirchner, J. Kroha, and P. W\" olfle,
500: Phys. Rev. B \textbf{64}, 155111 (2001).
501: %\bibitem{Georges} A. Georges, G. Kotliar, W. Krauth, M.J. Rozenberg, Rev. Mod. Phys. \textbf{68}, 13 (1996).
502: \bibitem{Haule} K. Haule, G. Kotliar, cond-mat/0601478 (2006).
503: \bibitem{Civelli} M. Civelli, M. Capone, S.S. Kancharla, O. Parcollet,
504: and G. Kotliar, Phys. Rev. Lett. \textbf{95}, 106402 (2005).
505: \bibitem{VanderMarel}
506: A. El Azrak, et. al.
507: % R. Nahoum, N. Bontemps, M. Guilloux-Viry, C.
508: %Thivet, A. Perrin, S. Labdi, Z. Z. Li and H. Raffy,
509: Phys. Rev. B
510: 49, 9846 (1994). D. Van der Marel, \textit{et.al}, Nature
511: \textbf{425}, 271 (2003).
512: \bibitem{Dai} P. Dai, H.A. Mook, S.M. Hayden, G. Aeppli, T.G. Perring, R.D. Hunt,
513: F. Doan, Science, \textbf{284}, 1344 (1999).
514: \bibitem{Si} Q. Si et. al.
515: %S. Rabello, K. Ingersent, J.L. Smith,
516: Nature \textbf{413}, 804 (2001).
517: %\bibitem{Coleman} P. Coleman, A. Schofield, Nature \textbf{433}, 226 (2005).
518: \bibitem{georges} A Georges, R Siddharthan, S Florens Phys. Rev. Lett. 87, 277203 (2001).
519: \bibitem{varma} Jones and C. M. Varma Phys. Rev. Lett. 58, 843–846 (1987).
520: \bibitem{tdstan} T. Stanescu and G. Kotliar, cond-mat/0508302.
521: \end{thebibliography}
522:
523: \end{document}
524: