1: % Group addresses by affiliation; use superscriptaddress for long
2: % author lists, or if there are many overlapping affiliations.
3: % For Phys. Rev. appearance, change preprint to twocolumn.
4: % Choose pra, prb, prc, prd, pre, prl, prstab, or rmp for journal
5: % Add 'draft' option to mark overfull boxes with black boxes
6: % Add 'showpacs' option to make PACS codes appear
7: % Add 'showkeys' option to make keywords appear
8: %\documentclass[aps,prb,preprint,groupedaddress,showpacs]{revtex4}
9: %\documentclass[aps,prb,preprint,superscriptaddress,showpacs]{revtex4}
10: \documentclass[aps,prl,twocolumn,superscriptaddress,showpacs]{revtex4}
11: \usepackage{graphicx}% Include figure files
12:
13: \begin{document}
14:
15: \title{Phonon Effects on Spin-Charge Separation in One Dimension}
16: \author{Wen-Qiang Ning}
17: \affiliation{Department of Physics, Fudan University, Shanghai
18: 200433, China} \affiliation{Department of Physics and Institute of
19: Theoretical Physics, The Chinese University of Hong Kong, Shatin,
20: Hong Kong, China}
21: \author{Hui Zhao}
22: \affiliation{Department of Physics, Fudan University, Shanghai
23: 200433, China}
24: \author{Chang-Qin Wu }
25: \email[Corresponding author. Email: ] {cqw@fudan.edu.cn}
26: \affiliation{Department of Physics, Fudan University, Shanghai
27: 200433, China}
28: \author{Hai-Qing Lin}
29: \affiliation{Department of Physics and Institute of Theoretical Physics,
30: The Chinese University of Hong Kong, Shatin, Hong Kong, China}
31: \affiliation{Department of Physics, Fudan University, Shanghai 200433,
32: China}
33:
34: \date{\today}
35:
36: \begin{abstract}
37: Phonon effects on spin-charge separation in one dimension are investigated
38: through the calculation of one-electron spectral functions in terms of the
39: recently developed cluster perturbation theory together with an optimized
40: phonon approach. It is found that the retardation effect due to the
41: finiteness of phonon frequency suppresses the spin-charge separation and
42: eventually makes it invisible in the spectral function. By comparing our
43: results with experimental data of TTF-TCNQ, it is observed that the
44: electron-phonon interaction must be taken into account when interpreting
45: the ARPES data.
46: \end{abstract}
47: \pacs{63.20.Kr, 71.27.+a, 71.30.+h}
48:
49: \maketitle
50:
51: It is commonly accepted that most one-dimensional (1D) correlated
52: electronic systems cannot be properly described by the traditional
53: Fermi liquid theory, instead, their behaviors are well predicted
54: by the Luttinger liquid theory
55: \cite{luttinger,lieb,emery,haldane}. One of the key features of
56: the Luttinger liquid is the spin-charge separation: the low-energy
57: excitations are not quasiparticles with charge $e$ and spin 1/2
58: together, rather, they are collective modes of spin and charge
59: excitations separately, called spinons and holons. Since spinon
60: and holon move with different speed, they eventually decouple.
61: Following earlier works \cite{luttinger,lieb,emery,haldane}, many
62: studied spin-charge separation with various theoretical schemes.
63: In particular, one could explore the existence of the spin-charge
64: separation by calculating the spectral function \cite{meden and
65: voit,meden,schrieffer,cpt1,Jeckelmann,mat}, which has direct
66: relation to the angle-resolved photoemission spectroscopy (ARPES).
67: In some recent works, the spinon and holon branches have been
68: observed by ARPES performed on some 1D materials such as
69: $SrCuO_2$.\cite{kim,kobayashi,claessen,sing} Because both
70: electron-electron and electron-phonon interactions exist in many
71: low-dimensional materials, it is important to address the role of
72: these interactions on the spin-charge separation.
73: \cite{meden,tsutsui,kivelson,perfetti,dessau,kmshen}
74:
75: The one-dimensional Holstein-Hubbard model (HHM), which is the simplest
76: model involving both electron-phonon (e-p) and electron-electron (e-e)
77: interactions, has been used extensively to describe some low-dimensional
78: materials. Since electrons in these materials are strongly correlated, the
79: interplay between electron-phonon interaction and Coulomb repulsion should
80: have profound effect on the spin-charge separation, and we expect to
81: observe these effects by investigating the single-particle excitation
82: spectra. The spectral function provides valuable insights into the usually
83: complicated many-body systems, such as high-temperature superconductors,
84: cuprate ladder compounds, and organic conductors. For example, very
85: recently, by using the exact diagonalization method, Fehske \emph{et
86: al.}\cite{fehske} calculated the spectral function of the Holstein-Hubbard
87: model on a finite system and found a Mott-insulator to Peierls-insulator
88: transition at a compatible ratio of the e-e to e-p interactions.
89:
90: Bearing these in mind, we compute the one-electron spectral function of the
91: HHM by applying the recently developed cluster perturbation theory
92: (CPT)\cite{cpt1,cpt2,hohenadler} together with an optimized phonon
93: approach.\cite{zhang,weise,zhao} The spectral function \emph{at full
94: frequency region} with rich satellite structures is obtained in the model
95: of \emph{both} e-e and e-p interactions for the first time. Phonon effects
96: on spin-charge separation are focused in the presence of e-e interactions
97: from weak to strong coupling. It is found that the retardation effect due
98: to the finiteness of phonon frequency does not favor the spin-charge
99: separation. In weak interaction regimes, a peak in the spectral function
100: was observed which is consistent with the existence of a metallic phase as
101: proposed recently by Clay and Hardikar\cite{clay}. Furthermore, it is
102: observed that one must take electron-phonon interaction into account when
103: interpreting the ARPES experimental data in the one dimensional material.
104:
105: The HHM accounts for a tight-binding electron band, on-site
106: Coulomb repulsion between electrons of opposite spin, and coupling
107: of charge carriers to local phonons:
108: \begin{eqnarray}
109: H&=&-t\sum_{i,\sigma}(c_{i,\sigma}^\dag c_{i+1,\sigma}+H.c.)
110: +U\sum_i n_{i\uparrow}n_{i\downarrow}\nonumber\\
111: &&-g\sum_{i,\sigma}(b_i^\dag+b_i)n_{i,\sigma}
112: +\omega_0\sum_ib_i^\dag b_i,
113: \label{Model}
114: \end{eqnarray}
115: where $c_{i,\sigma}^\dagger (c_{i,\sigma})$ creates (annihilates) an
116: electron with $\sigma$ on site $i$, and $b_i^\dag$ and $b_i$ are creation
117: and annihilation operators of the local phonon mode at site $i$,
118: respectively. $t$ is the electron hopping constant between nearest neighbor
119: sites which will be set as the energy unit in our calculations, $\omega_0$
120: is the bare phonon frequency, and $g$ is the electron-phonon coupling
121: constant.
122:
123: For the calculation of the spectral properties within the framework of
124: CPT\cite{cpt1,cpt2}, one divides the lattice into fully equivalent clusters
125: of a finite sites. For each cluster, we calculate the Green's function
126: $G_{i,j}(z)(\equiv G_{i,j}^+(z)+G_{i,j}^-(z))$, with $G_{i,j}^{\pm}(z)$
127: defined as
128: \begin{equation}
129: G_{i,j}^{\pm}(z)=\langle\phi_0|c^{\pm}_i\frac{1}{z\pm
130: (H-E_0)}c_j^{\mp}|\phi_0\rangle,
131: \end{equation}
132: where $c^+_i\equiv c^\dagger_i, c^-_i\equiv c_i$, and $|\phi_0\rangle$
133: being the ground state of the cluster, which is obtained by using the
134: Lanczos exact diagonalization (ED) method within an optimized phonon
135: approach\cite{zhang,weise} under open boundary conditions. Two terms in
136: $G_{i,j}$ corresponding to electron and hole propagation, respectively, can
137: be obtained. The CPT treats the intercluster hopping by a strong-coupling
138: perturbation, i.e., ($t/U$) expansion. The lowest-order CPT approximation
139: to Green's function gives
140: \begin{equation}
141: G_{CPT}(k,z)=\frac{1}{N}\sum_{i,j}e^{-ik(i-j)}\widetilde{G}_{i,j}(Nk,z),
142: \end{equation}
143: where $\widetilde{G}_{i,j}(Q,z)$ is the Green's function of the full system
144: and $N$ is the size of clusters. The spectral function is then
145: $A(k,\omega)=-\frac{1}{\pi}Im[G_{CPT}(k,z)]$, where $z=\omega+i\eta$ with
146: $\eta$ defines the width of peaks in the spectral function. Since the Fermi
147: energy is set to zero, the spectral function has the symmetry of
148: $A(k,\omega)=A(\pi-k,-\omega)$ due to the electron-hole symmetry of the
149: model (1).
150:
151: In the absence of interactions, $A(k,\omega)$ obtained by the CPT
152: method is exact \cite{cpt1,cpt2}, while for interacting models,
153: the accuracy of CPT depends on the size of the cluster and the
154: number of optimal phonon chosen. To test the accuracy of the
155: approach we use, we calculated the spectral function of the
156: Hubbard model with exactly the same parameters as used by Benthien
157: et al\cite{Jeckelmann} and obtained agreeable results. We also
158: calculated the first two spectral momenta and they match exactly
159: to those obtained by White\cite{white}. Furthermore, based on our
160: previous technique analysis\cite{zhao}, system parameters were
161: carefully chosen in this work to ensure that our results mimic
162: thermodynamic limit. Results obtained in this Letter were for
163: $N=6$, $\eta=0.1t$, and three optimized phonons at each site, with
164: relative error $10^{-5}$ for the total energy.\cite{zhao}
165:
166: Three energy scales govern the physics of the HHM: the on-site Coulomb
167: repulsion ($U$), the electron-phonon coupling ($\lambda=2g^2/\omega_0$) and
168: the bare phonon frequency ($\omega_0$). The ground state of the system at
169: half-filling is a Mott-Hubbard insulating (MI) state when $U$ is large, and
170: shows spin-density-wave (SDW) fluctuations. When electron-phonon
171: interaction dominates, the system is in the Peierls insulating state (PI),
172: characterized by the charge-density-wave where both spin gap and change gap
173: are finite, while in the MI state, the spin gap vanishes. Very recently, it
174: was reported that there is a metallic region intermediate between the PI
175: and MI states with superconducting pairing correlation
176: dominates.\cite{clay}
177:
178: In the absence of phonons ($g=0$), Eq. (\ref{Model}) is just the Hubbard
179: model whose physics have been extensively studied and it is well known that
180: in the Hubbard model, spin and charge separate.
181: \cite{schrieffer,cpt1,Jeckelmann,mat} On the other hand, when $U=0$, Eq.
182: (\ref{Model}) is another extensively studied model, the Holstein model
183: (HM). In the strong electron-phonon coupling region, the ground state of
184: the Holstein model at half filling is either a bipolaron insulating (BPI)
185: state in the large $\omega_0$ limit, characterized by the configuration
186: where each site is either empty or doubly occupied because phonons produce
187: an attraction between the electrons, or a traditional band insulating state
188: in the small $\omega_0$ limit.\cite{fehske,hirsch} In the weak coupling
189: region, the Peierls gap is suppressed by the phonon quantum fluctuations
190: and the ground state is at metallic (M) phase.\cite{wu,zhang,zhao} When the
191: phonon degrees of freedom are integrated out, the spin-1/2 Holstein model
192: could be mapped onto the Hubbard model with an effective dynamical
193: attraction $U_{eff}(\omega)=-\lambda/(1-\omega^2/\omega_0^2)$. Here one
194: also expects to observe the spin-charge separation. In the anti-adiabatic
195: limit, i.e., $\omega_0\rightarrow\infty$, the attraction becomes
196: instantaneous and equals to the bipolaron binding energy $\lambda$, which
197: has already been reported long time ago.\cite{hirsch} Obviously, for any
198: finite phonon frequency $\omega_0$, one must consider the retardation
199: effect fully which could not be simply presented by the above
200: $U_{eff}(\omega)$.
201:
202: \begin{figure}
203: \includegraphics[scale=0.35]{fig1.eps}
204: \caption{The obtained spectral function $A(k,\omega)$ of (a) the
205: large-$\omega_0$ Holstein model and (b) the negative-U Hubbard
206: model at half-filling. HO and SP stand for holon and spinon
207: excitations respectively.\label{fig1}}
208: \end{figure}
209:
210: To have a sense of the magnitude of $\omega_0$, we give a comparison of the
211: spectral function of these two models in Figure~1. It is clearly shown that
212: the spectral function of the Holstein model is almost the same as that of
213: the negative-U Hubbard model. Some minor differences due to finiteness of
214: $\omega_0$ are invisible in the figure (in other words, $\omega_0 = 8$ is
215: almost at the antiadibatic limit). This result is not trivial as seen at
216: first glance because it implies the single-particle excitation of the
217: system with a large phonon frequency is similar to that at the
218: antiadiabatic limit, which is consistent with the existence of a quantum
219: metal-insulator phase transition in the Holstein model.\cite{wu,jec,clay}
220: The peaks labelled ``SP'' and ``HO'' in Fig.~1 refer to the spinon and the
221: holon branches, respectively, signaturing the spin-charge separation.
222: Compare to the conventional Luttinger liquids (e.g., the positive-U Hubbard
223: model), the charge velocity ($v_\rho$) is smaller than the spin velocity
224: ($v_\sigma$).
225:
226: %Fig. 2
227: \begin{figure}
228: \includegraphics[scale=0.38]{fig2.eps}
229: \caption{The spectral function $A(0,\omega)$ of the Holstein
230: model. \label{fig2}}
231: \end{figure}
232:
233: Fig.~2 illustrates the retardation effects systematically. Starting from
234: the strong coupling case ($\lambda = 2$), we observe that as we decrease
235: $\omega_0$ from 4 to 2, the incoherent part of the spectra becomes more
236: important. As a consequence, the spectral weights of the spinon and holon
237: excitations are much smaller. This could be regarded as a continuation from
238: Fig. 1: $\omega_0=\infty\rightarrow 8 \rightarrow 4$. When the phonon
239: frequency is further reduced, the spectral weights correspond to phonon
240: excitations become dominant, which is quite different from that in the
241: antiadiabatic regime where the ``spinon'' and ``holon'' excitations are
242: clearly the dominant ones. Therefore, due to the strong mixing of the
243: coherent and incoherent excitations, it is difficult to single out the
244: spinon and holon excitations, instead, one observes an almost flat band
245: dispersion with exponentially small spectral weight. The dominant peaks in
246: the incoherent part of the spectra are related to multiples of the (large)
247: bare phonon frequency broadened by electronic excitations. Such
248: electron-phonon mixed nature of excitations could be seen in the spectra
249: away from the Fermi surface.
250:
251: It is quite natural to expect that the separation of spin and charge
252: excitation will become smaller with the decrease of the electron-phonon
253: coupling strength. This is clearly reflected in the spectral function. As
254: shown in the first row of Fig. 2, when we reduce the electron-phonon
255: coupling $\lambda$ at $\omega_0=4$, the difference between the spinon and
256: the holon excitation at $k=0$ becomes smaller and eventually invisible. As
257: we reduce phonon frequency, retardation comes into play. It is also
258: interesting to observe that there is an excitation split in the weak
259: coupling case (Fig.~2, $\lambda=0.5$, $\omega_0=1$). Such splitting is not
260: due to spin-charge separation. In fact, by carefully comparing this
261: spectral function with that of a spinless Holstein model at corresponding
262: electron-phonon coupling, we found that the splitting is caused by the
263: polaron interaction.
264:
265: %Fig. 3
266: \begin{figure}
267: \includegraphics[scale=0.35]{fig3.eps}
268: \caption{Spectral function $A(k,\omega)$ of the spinful (a) and
269: spinless (b) Holstein model. $\lambda=0.5$ and $\omega_0=1$.
270: \label{fig3}}
271: \end{figure}
272:
273: Figure 3 shows the spectra of up-spin electron for the spin-1/2 Holstein
274: model at half-filling in comparison with the spinless Holstein model. In
275: the weak coupling regime the two spectral are almost the same, indicating
276: that the existence of the down-spin electrons have nearly zero effect on
277: the spectral function of the up-spin electrons. It shows that the
278: phonon-mediated interaction between up- and down-spin electrons is very
279: weak so they are almost decoupled in this case. Thus the splitting of the
280: excitations cannot be attributed to the spin-charge separation, rather, it
281: is due to the interaction among polarons. Notice that there is a small peak
282: labelled as ``A'' in the spectral weight, which is almost dispersionless in
283: small $k$ regime (see Fig.~3) and is suppressed by the on-site repulsion
284: $U$ (see Fig.~4). Since it is appeared in the metallic region intermediate
285: between PI and MI phases, we speculate the peak ``A'' may be related with
286: the electron pairing, as discussed recently by Clay and
287: Hardiker\cite{clay}, although further investigations are definitely
288: deserved.
289:
290: Now we turn the Coulomb interaction $U$ on and discuss its effect
291: on the spectral function in the presence of electron-phonon
292: interaction.
293: \begin{figure}
294: \includegraphics[scale=0.38]{fig4.eps}
295: \caption{Spectral function $A(0,\omega)$ of the Holstein-Hubbard
296: model at half filling. $\omega_0=1$.\label{fig4}}
297: \end{figure}
298: From Figure 4, we also see that, the electron-electron interaction
299: and the electron-phonon interaction have opposite effect on the
300: spin-charge separation, as we intuitively expected. At given
301: electron-electron interaction (for example, $U=4$), increasing the
302: electron-phonon interaction tends to broaden the excitation bands
303: and leave the spin-charge separation invisible. While at fix
304: electron-phonon interaction, the electron-electron interaction
305: increases the separation between the spin and the charge
306: excitations.
307:
308: %Fig. 5
309: \begin{figure}
310: \includegraphics[scale=0.38]{fig5.eps}
311: \caption{Density plot of the spectral function $A(k,\omega)$ for
312: (a) the Holstein-Hubbard, and (b) the Hubbard model. (c) The
313: spectral function $A(0,\omega)$ of corresponding models at
314: different electron-phonon coupling. SH stands for the shadow band.
315: $\omega_0=1.$\label{fig5}}
316: \end{figure}
317:
318: By further increasing the electron-electron interaction, we illustrate the
319: role of electron-phonon coupling in Fig.~5. Comparing Fig.~5(b) with
320: Fig.~5(a), one may still observe the signature of spin-charge separation as
321: $\lambda$ increases, but the holon branch is much more broadened and its
322: spectral weight decreases, while the spinon branch is nearly unchanged.
323: This is clearly shown in Fig.~5(c). Consequently, the so-called shadow
324: band, which originates from the diverging spin fluctuations at
325: $2k_F$,\cite{hass,penc,favand,cpt1} is nearly gone, due to the fact that
326: the shadow band is actually the continuation of the holon band.
327:
328: Finally, we make a comparison on our results to the ARPES experiments of
329: the quasi-one-dimensional organic conductor TTF-TCNQ.\cite{claessen,sing}
330: We notice that in the Fig.~7 of the Ref.\cite{sing}, the charge branch (b)
331: and the spin branch (a) are weakly connected with each other. The result
332: can not be explained by the Hubbard model alone(Fig.~5(b)). As our results
333: suggest, the electron-phonon interaction must be taken into account. From
334: Fig. 5(a), it is clearly seen that the dispersion of the charge branch is
335: weakened in the spectral function and the spinon branch is weakly connected
336: to the holon branch. A similar broadening due to phonon is also observed in
337: one dimensional SrCuO$_2$.\cite{zxshen} These facts indicate that one
338: should expect significant contribution from the electron-phonon interaction
339: to the spectra of these strongly correlated quasi-one-dimensional
340: materials. Of course, our conclusions are based on numerical studies of Eq.
341: (1) so to have detailed analyzes of experiments, one may use models
342: different from Eq. (1) but we believe essential physics remain unchanged.
343:
344: In summary, by applying the CPT together with an optimized phonon approach,
345: we have studied the spectral function of the one-dimensional
346: Holstein-Hubbard model at half filling. A comprehensive picture for the
347: spectral function in the presence of electron-electron interaction and
348: electron-phonon interaction was presented. In particular, we addressed the
349: issue of spin-charge separation and found that the electron-electron
350: interaction competes with the electron-phonon interaction on the
351: spin-charge separation, and the retardation effect due to phonons may
352: diminish the spin-charge separation in the spectral function. We also found
353: polaron splitting and observed a peak that may related to electron pairing
354: in the weak e-p coupling limit.
355:
356: This work was partially supported by the National Natural Science
357: Foundation of China, CUHK 401504, and MOE B06011. The authors are
358: grateful to Prof. R. B. Tao for his generous support and to Prof.
359: D. L. Feng for helpful discussion on the ARPES data. We also thank
360: Profs. H. Chen and H. Zheng for stimulating discussions and Prof.
361: Z.-X. Shen for his preprint.
362:
363: \begin{thebibliography}{99}
364:
365: \bibitem{luttinger} S. Tomonaga, Prog. Theor. Phys. \textbf{5}, 544 (1950);
366: J. M. Luttinger, J. Math. Phys. \textbf{4}, 1154 (1963).
367:
368: \bibitem{lieb} E. H. Lieb and F. Y. Wu, Phys. Rev. Lett.
369: \textbf{20}, 1445 (1968).
370:
371: \bibitem{emery} V. J. Emery,
372: pp. 247-303 in {\it Highly Conducting One-Dimensional Solids}, edited by J.
373: T. Devreese {\it et al.} (Plenum, New York, 1979).
374:
375: \bibitem{haldane} F. D. M. Haldane, Phys. Rev. Lett. \textbf{45}, 1358 (1980).
376:
377: \bibitem{meden and voit} V. Meden and K. Sch\"{o}nhammer, Phys. Rev.
378: B \textbf{46}, 15753 (1992); J. Voit, Phys. Rev. B \textbf{47},
379: 6740 (1993).
380:
381: \bibitem{meden} V. Meden \emph{et al.},
382: Phys. Rev. B \textbf{50}, 11179 (1994).
383:
384: \bibitem{schrieffer} M. G. Zacher \emph{et al.}, Phys. Rev. B \textbf{57}, 6370 (1998).
385:
386: \bibitem{cpt1} D. S\'{e}n\'{e}chal \emph{et al.}, Phys. Rev. Lett. \textbf{84}, 522 (2000).
387:
388: \bibitem{Jeckelmann} H. Benthien \emph{et al.},
389: Phys. Rev. Lett. \textbf{92}, 256401 (2004).
390:
391: \bibitem{mat} H. Matsueda \emph{et al.}, cond-mat 0505670(unpublished).
392:
393: \bibitem{kim} C. Kim \emph{et al.}, Phys. Rev. Lett.
394: \textbf{77}, 4054 (1996); C. Kim \emph{et al.}, Phys. Rev. B
395: \textbf{56}, 15589 (1997).
396:
397: \bibitem{kobayashi} K. Kobayashi \emph{et al.}, Phys. Rev. Lett.
398: \textbf{82}, 803 (1999).
399:
400: \bibitem{claessen} R. Claessen \emph{et al.}, Phys. Rev. Lett. \textbf{88}, 096402
401: (2002).
402:
403: \bibitem{sing} M. Sing \emph{et al.}, Phys. Rev. B \textbf{68}, 125111 (2003).
404:
405: \bibitem{tsutsui} K. Tsutsui \emph{et al.}, Physica C
406: \textbf{392-396}, 199 (2003).
407:
408: \bibitem{kivelson} I. P. Bindloss and S. A. Kivelson,
409: Phys. Rev. B \textbf{71}, 014524 (2005).
410:
411: \bibitem{perfetti} L. Perfetti \emph{et al.}, Phys. Rev. B \textbf{66}, 075107
412: (2002).
413:
414: \bibitem{dessau} D. S. Dessau \emph{et al.}, Phys. Rev. Lett. \textbf{81}, 192
415: (1998).
416:
417: \bibitem{kmshen} K. M. Shen \emph{et al.}, Phys. Rev.
418: Lett. \textbf{93}, 267002 (2004).
419:
420: \bibitem{fehske} H. Fehske \emph{et al.}, Phys. Rev. B \textbf{69}, 165115 (2004).
421:
422: \bibitem{cpt2} D. S\'{e}n\'{e}chal \emph{et al.}, Phys.
423: Rev. B \textbf{66}, 075129 (2002).
424:
425: \bibitem{hohenadler} M. Hohenadler \emph{et al.}, Phys. Rev. B \textbf{68}, 184304 (2003).
426:
427: \bibitem{zhao} H. Zhao \emph{et al.}, Phys.
428: Rev. B \textbf{71}, 115201 (2005).
429:
430: \bibitem{zhang} C. Zhang \emph{et al.}, Phys. Rev. Lett.
431: \textbf{80}, 2661 (1998); Phys. Rev. B \textbf{60}, 14092 (1999).
432:
433: \bibitem{weise} A. Weisse \emph{et al.}, Phys. Rev. B \textbf{62}, R747 (2000).
434:
435: \bibitem{white} S. R. White, Phys. Rev. B \textbf{44}, 4670 (1991).
436:
437: \bibitem{clay} R. T. Clay and R. P. Hardikar,
438: Phys. Rev. Lett. \textbf{95}, 096401 (2005).
439:
440: \bibitem{hirsch} J. E. Hirsch and E. Fradkin,
441: Phys. Rev. B \textbf{27}, 4302 (1983).
442:
443: \bibitem{wu} C. Q. Wu \emph{et al.},
444: Phys. Rev. B \textbf{52}, R15683 (1995).
445:
446: \bibitem{jec} E. Jeckelmann \emph{et al.},
447: Phys. Rev. B \textbf{60}, 7950 (1999)
448:
449: \bibitem{hass} S. Haas and E. Dagotto,
450: Phys. Rev. B \textbf{52}, R14396 (1995).
451:
452: \bibitem{penc} K. Penc \emph{et al.},
453: Phys. Rev. Lett. \textbf{77}, 1390 (1996).
454:
455: \bibitem{favand} J. Favand \emph{et al.}, Phys. Rev. B \textbf{55}, R4859 (1997).
456:
457: \bibitem{zxshen} B. J. Kim \emph{et al.}, preprint.
458:
459: \end{thebibliography}
460:
461: \end{document}
462: